cpet/openbsd-ports
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
21c4e79445
openbsd-ports/infrastructure/mk/bsd.port.mk

2820 lines
81 KiB
Makefile
Raw Normal View History

switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
#-*- mode: Makefile; tab-width: 4; -*-
Cosmetic patch: don't append DIST_SUBDIR if it's not defined. Set vim filetype, as vi doesn't care.
2000-02-02 10:28:17 -05:00
# ex:ts=4 sw=4 filetype=make:
all MULTI_PACKAGES are new, stop testing -main.
2006-11-26 12:36:07 -05:00
# $OpenBSD: bsd.port.mk,v 1.836 2006/11/26 17:36:07 espie Exp $
change email address of maintainer to ports-admin@openbsd.org
1999-05-10 17:35:25 -04:00
# $FreeBSD: bsd.port.mk,v 1.264 1996/12/25 02:27:44 imp Exp $
# $NetBSD: bsd.port.mk,v 1.62 1998/04/09 12:47:02 hubertf Exp $
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
#
# bsd.port.mk - 940820 Jordan K. Hubbard.
# This file is in the public domain.
Documentation clean-up: group deprecated variables, clarify MAINTAINER comment. List bsd.port.mk(5) as the future source of documentation.
2000-09-13 10:06:55 -04:00
# Each port has a MAINTAINER, which is the email address(es) of the person(s)
# to contact if you have questions/suggestions about that specific port.
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
# To obtain that address, just type
zap show VARNAME= usage.
2003-07-12 08:51:19 -04:00
# make show=MAINTAINER
Documentation clean-up: group deprecated variables, clarify MAINTAINER comment. List bsd.port.mk(5) as the future source of documentation.
2000-09-13 10:06:55 -04:00
# in the specific port's directory.
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
#
Documentation clean-up: group deprecated variables, clarify MAINTAINER comment. List bsd.port.mk(5) as the future source of documentation.
2000-09-13 10:06:55 -04:00
# The ports@openbsd.org address is the `default' MAINTAINER (the generic
# OpenBSD ports mailing-list).
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
# Enquiries as to the bsd.port.mk framework should usually be directed
Documentation clean-up: group deprecated variables, clarify MAINTAINER comment. List bsd.port.mk(5) as the future source of documentation.
2000-09-13 10:06:55 -04:00
# to ports@openbsd.org.
USE_X11 should mark ports that need X11, but don't say anything about where they install. E.g., DON'T TOUCH USE_X11 in existing ports for now, we'll see how we go about them.
2000-03-26 16:23:20 -05:00
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
Due to the way make is working, flavors passed as `make FLAVOR=a' don't work, as they propagate down subdirs. Use `FLAVOR=a make' (sh) or `env FLAVOR=a make' (csh) instead. Error out with a useful error message if old usage is encountered.
2001-03-28 10:01:01 -05:00
.if ${.MAKEFLAGS:MFLAVOR=*}
Replace .BEGIN lines with explicit ERRORS framework.
2001-07-17 12:11:45 -04:00
ERRORS+= "Fatal: Use 'env FLAVOR=${FLAVOR} ${MAKE}' instead."
Due to the way make is working, flavors passed as `make FLAVOR=a' don't work, as they propagate down subdirs. Use `FLAVOR=a make' (sh) or `env FLAVOR=a make' (csh) instead. Error out with a useful error message if old usage is encountered.
2001-03-28 10:01:01 -05:00
.endif
Enforce correct use of SUBPACKAGE.
2001-09-29 13:36:02 -04:00
.if ${.MAKEFLAGS:MSUBPACKAGE=*}
ERRORS+= "Fatal: Use 'env SUBPACKAGE=${SUBPACKAGE} ${MAKE}' instead."
.endif
Due to the way make is working, flavors passed as `make FLAVOR=a' don't work, as they propagate down subdirs. Use `FLAVOR=a make' (sh) or `env FLAVOR=a make' (csh) instead. Error out with a useful error message if old usage is encountered.
2001-03-28 10:01:01 -05:00
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
.for v in PKGREPOSITORY PKGREPOSITORYBASE CDROM_PACKAGES FTP_PACKAGES SED_PLIST
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. if defined($v)
ERRORS+= "Fatal: Variable $v is obsolete, use PACKAGE_REPOSITORY instead."
. endif
.endfor
zap stuff documented elsewhere.
2003-07-14 10:02:18 -04:00
# The definitive source of documentation to this file's user-visible parts
# is bsd.port.mk(5).
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
#
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
# Any variable or target starting with an underscore (e.g., _DEPEND_ECHO)
# is internal to bsd.port.mk, not part of the user's API, and liable to
# change without notice.
Remove documentation that is now in bsd.port.mk(5)
2000-12-01 11:31:11 -05:00
#
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
#
# Variables to change if you want a special behavior:
#
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
# DEPENDS_TARGET - The target to execute when a port is calling a
# dependency (default: "install").
Remove documentation that is now in bsd.port.mk(5)
2000-12-01 11:31:11 -05:00
#
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
# Default sequence for "all" is: fetch checksum extract patch configure build
#
# Please read the comments in the targets section below, you
update comment to match reality after recent removal of support for using pre-*/post-* scripts
2000-01-26 19:09:44 -05:00
# should be able to use the pre-* or post-* targets (which
# are available for every stage except checksum) or provide
# an overriding do-* target to do pretty much anything you
# want.
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
#
Add intermediate `distpatch' target. This is done in a tricky way, so as not to disturb the usual sequence of targets. Namely, you still have fetch, extract, patch, configure, build, install, package, but patch does invoke a subtarget distpatch when needed, and you can fetch, extract, distpatch, patch, configure, build, install, package for the same results. Mostly useful for porters: provide an intermediate point to work on patches, and make it easy to override distpatch/OpenBSD patches separately.
2000-01-30 10:19:40 -05:00
# For historical reasons, the distpatch target is a subtarget of patch.
# If you provide a do-patch, you MUST call distpatch explicitly.
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
# The sequence of hooks actually run is:
Add intermediate `distpatch' target. This is done in a tricky way, so as not to disturb the usual sequence of targets. Namely, you still have fetch, extract, patch, configure, build, install, package, but patch does invoke a subtarget distpatch when needed, and you can fetch, extract, distpatch, patch, configure, build, install, package for the same results. Mostly useful for porters: provide an intermediate point to work on patches, and make it easy to override distpatch/OpenBSD patches separately.
2000-01-30 10:19:40 -05:00
#
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
# pre-patch `real distpatch' post-distpatch `real patch' post-patch
aesthetical tweaks. Oh, and uninstall doesn't need to remove the package cookies, obviously.
2001-04-11 12:06:07 -04:00
#
move variables that only users can tweak, or that are constants, to the front of bsd.port.mk. This allows Makefile.inc and modules to use them more easily, and also cleans up things slightly. Also kill .include ${PORTSDIR}/../Makefile.inc (???)
2004-01-06 10:02:57 -05:00
# User settings
TRUST_PACKAGES?=No
try fetching missing packages from PKG_PATH, activated by setting FETCH_PACKAGES set PKG_PATH consistently when calling pkg_add idea, help and ok espie
2005-11-04 04:34:50 -05:00
FETCH_PACKAGES?=No
move variables that only users can tweak, or that are constants, to the front of bsd.port.mk. This allows Makefile.inc and modules to use them more easily, and also cleans up things slightly. Also kill .include ${PORTSDIR}/../Makefile.inc (???)
2004-01-06 10:02:57 -05:00
CLEANDEPENDS?=No
USE_SYSTRACE?= No
BULK?=No
RECURSIVE_FETCH_LIST?= Yes
always define WRKOBJDIR, FAKEOBJDIR (simpler logic). change tests to !empty instead of defined. Move BULK_TARGETS up with other user settings.
2004-02-07 17:18:49 -05:00
WRKOBJDIR?=
FAKEOBJDIR?=
BULK_TARGETS?=
generic BULK_DO fragment that can do anything, invoked right before clean. can be used to run targets that may fail without impacting further stuff, or to collect stats on anything. For instance BULK_DO= mkdir -p ${PORTSDIR}/config/${FULLPKGPATH}; \ cp -f ${WRKBUILD}/config.{log,status} ${PORTSDIR}/config/${FULLPKGPATH} || true; \ cd ${PACKAGE_REPOSITORY}/${MACHINE_ARCH}/all && perl ${PORTSDIR}/infrastructure/package/check-modes ${PKGNAMES} || true
2006-10-23 10:33:01 -04:00
BULK_DO?=
industrial-strength updates: - separate UPDATE_COOKIES_DIR, unless set to empty. Makes for easier management. - FORCE_UPDATE variable, that makes certain packages are updated when a dependency is run, and that forces the replacement of packages that are already there. Still uses the update cookie, so that this update occurs just once unless you wipe out yur UPDATE_COOKIES_DIR (which justifies putting update cookies in a separate directory).
2004-11-15 08:52:03 -05:00
FORCE_UPDATE?=No
introspection mechanism: dump-vars
2006-07-09 07:10:16 -04:00
# All variables relevant to the port's description
let LIB_DEPENDS and WANTLIB depend on the SUBPACKAGE. Build a special list of _BUILDLIB_DEPENDS, so that separate subpackages can depend on each other without making an infinite loop...
2006-11-20 05:49:22 -05:00
_ALL_VARIABLES?=HOMEPAGE DISTNAME \
introspection mechanism: dump-vars
2006-07-09 07:10:16 -04:00
BUILD_DEPENDS RUN_DEPENDS REGRESS_DEPENDS USE_GMAKE MODULES FLAVORS \
index PKG_ARCH, PERMIT_PACKAGE_FTP, PERMIT_PACKAGE_CDROM on SUBPACKAGE name.
2006-11-13 09:14:57 -05:00
NO_BUILD NO_REGRESS SHARED_ONLY ONLY_FOR_ARCHS IS_INTERACTIVE \
allows PREFIX and CATEGORIES to be subpackage dependent
2006-11-20 08:50:16 -05:00
BROKEN MULTI_PACKAGES PSEUDO_FLAVORS \
introspection mechanism: dump-vars
2006-07-09 07:10:16 -04:00
REGRESS_IS_INTERACTIVE DISTFILES DIST_SUBDIR \
PERMIT_DISTFILES_CDROM PERMIT_DISTFILES_FTP \
let LIB_DEPENDS and WANTLIB depend on the SUBPACKAGE. Build a special list of _BUILDLIB_DEPENDS, so that separate subpackages can depend on each other without making an infinite loop...
2006-11-20 05:49:22 -05:00
CONFIGURE_STYLE USE_LIBTOOL SEPARATE_BUILD \
introspection mechanism: dump-vars
2006-07-09 07:10:16 -04:00
SHARED_LIBS USE_MOTIF PACKAGES MASTER_SITES \
MASTER_SITES0 MASTER_SITES1 MASTER_SITES2 MASTER_SITES3 MASTER_SITES4 \
MASTER_SITES5 MASTER_SITES6 MASTER_SITES7 MASTER_SITES8 MASTER_SITES9 \
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
MAINTAINER SUBPACKAGE PACKAGING SUPDISTFILES \
+CONFIGURE_ARGS
2006-07-10 06:52:08 -04:00
AUTOCONF_VERSION AUTOMAKE_VERSION CONFIGURE_ARGS
zap a few non relevant lines, make _ALL_VARIABLES* internal. Use TMPDIR in pkgpath.mk
2006-11-01 07:41:34 -05:00
# and stuff needing to be MULTI_PACKAGE'd
add an extra level of indirection for dependency lists: _xxx_DEPLIST. allows us to get rid of some NO_DEPENDS tests later, also allows for RUN_DEPENDS to become indexed on SUBPACKAGE. LIB_DEPENDS and WANTLIB are going to need more surgery, since they need to be checked during build and packaging...
2006-11-19 07:32:53 -05:00
_ALL_VARIABLES_INDEXED?=COMMENT FULLPKGNAME PKGNAME PKG_ARCH \
allows PREFIX and CATEGORIES to be subpackage dependent
2006-11-20 08:50:16 -05:00
PERMIT_PACKAGE_FTP PERMIT_PACKAGE_CDROM RUN_DEPENDS LIB_DEPENDS WANTLIB \
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
CATEGORIES DESCR
move variables that only users can tweak, or that are constants, to the front of bsd.port.mk. This allows Makefile.inc and modules to use them more easily, and also cleans up things slightly. Also kill .include ${PORTSDIR}/../Makefile.inc (???)
2004-01-06 10:02:57 -05:00
# special purpose user settings
PATCH_CHECK_ONLY?=No
REFETCH?=false
# Constants used by the ports tree
ARCH!= uname -m
OPSYS= OpenBSD
OPSYS_VER= ${OSREV}
add humppa64 to lp64_archs
2005-04-01 10:55:36 -05:00
LP64_ARCHS=alpha amd64 hppa64 sparc64 mips64
add sh to NO_SHARED_ARCHS for now
2006-11-19 14:52:03 -05:00
NO_SHARED_ARCHS=m88k sh vax
move variables that only users can tweak, or that are constants, to the front of bsd.port.mk. This allows Makefile.inc and modules to use them more easily, and also cleans up things slightly. Also kill .include ${PORTSDIR}/../Makefile.inc (???)
2004-01-06 10:02:57 -05:00
Always define NO_SHARED_LIBS.
2004-08-02 09:01:52 -04:00
# Set NO_SHARED_LIBS for those machines that don't support shared libraries.
move variables that only users can tweak, or that are constants, to the front of bsd.port.mk. This allows Makefile.inc and modules to use them more easily, and also cleans up things slightly. Also kill .include ${PORTSDIR}/../Makefile.inc (???)
2004-01-06 10:02:57 -05:00
.for _m in ${MACHINE_ARCH}
. if !empty(NO_SHARED_ARCHS:M${_m})
Always define NO_SHARED_LIBS.
2004-08-02 09:01:52 -04:00
NO_SHARED_LIBS?= Yes
move variables that only users can tweak, or that are constants, to the front of bsd.port.mk. This allows Makefile.inc and modules to use them more easily, and also cleans up things slightly. Also kill .include ${PORTSDIR}/../Makefile.inc (???)
2004-01-06 10:02:57 -05:00
. endif
.endfor
Always define NO_SHARED_LIBS.
2004-08-02 09:01:52 -04:00
NO_SHARED_LIBS?= No
move variables that only users can tweak, or that are constants, to the front of bsd.port.mk. This allows Makefile.inc and modules to use them more easily, and also cleans up things slightly. Also kill .include ${PORTSDIR}/../Makefile.inc (???)
2004-01-06 10:02:57 -05:00
# Global path locations.
PORTSDIR?= /usr/ports
LOCALBASE?= /usr/local
X11BASE?= /usr/X11R6
DISTDIR?= ${PORTSDIR}/distfiles
BULK_COOKIES_DIR?= ${PORTSDIR}/bulk/${MACHINE_ARCH}
industrial-strength updates: - separate UPDATE_COOKIES_DIR, unless set to empty. Makes for easier management. - FORCE_UPDATE variable, that makes certain packages are updated when a dependency is run, and that forces the replacement of packages that are already there. Still uses the update cookie, so that this update occurs just once unless you wipe out yur UPDATE_COOKIES_DIR (which justifies putting update cookies in a separate directory).
2004-11-15 08:52:03 -05:00
UPDATE_COOKIES_DIR?= ${PORTSDIR}/update/${MACHINE_ARCH}
move variables that only users can tweak, or that are constants, to the front of bsd.port.mk. This allows Makefile.inc and modules to use them more easily, and also cleans up things slightly. Also kill .include ${PORTSDIR}/../Makefile.inc (???)
2004-01-06 10:02:57 -05:00
TEMPLATES?= ${PORTSDIR}/infrastructure/templates
honour TMPDIR and PKG_TMPDIR in systrace policies prodded by jolan@
2004-05-31 08:27:07 -04:00
TMPDIR?= /tmp
- new read-only variable, PKGNAMES. - plist repository under PLIST_DB (optional). - print-package-signatures shows what's used for signatures. - make fetch shows full url you can copy/paste, simpler to look for typos that way.
2005-09-04 18:32:37 -04:00
PLIST_DB?=
move variables that only users can tweak, or that are constants, to the front of bsd.port.mk. This allows Makefile.inc and modules to use them more easily, and also cleans up things slightly. Also kill .include ${PORTSDIR}/../Makefile.inc (???)
2004-01-06 10:02:57 -05:00
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
PACKAGE_REPOSITORY?=${PORTSDIR}/packages
move variables that only users can tweak, or that are constants, to the front of bsd.port.mk. This allows Makefile.inc and modules to use them more easily, and also cleans up things slightly. Also kill .include ${PORTSDIR}/../Makefile.inc (???)
2004-01-06 10:02:57 -05:00
move up pkgpath.mk, so that PKGPATH is available. Use it to define per-PKGPATH user-settings: BULK, WRKOBJDIR, ...
2004-02-07 17:34:02 -05:00
# local path locations
.include "${PORTSDIR}/infrastructure/mk/pkgpath.mk"
WRKOBJDIR_${PKGPATH}?= ${WRKOBJDIR}
FAKEOBJDIR_${PKGPATH}?= ${FAKEOBJDIR}
BULK_${PKGPATH}?= ${BULK}
BULK_TARGETS_${PKGPATH}?= ${BULK_TARGETS}
generic BULK_DO fragment that can do anything, invoked right before clean. can be used to run targets that may fail without impacting further stuff, or to collect stats on anything. For instance BULK_DO= mkdir -p ${PORTSDIR}/config/${FULLPKGPATH}; \ cp -f ${WRKBUILD}/config.{log,status} ${PORTSDIR}/config/${FULLPKGPATH} || true; \ cd ${PACKAGE_REPOSITORY}/${MACHINE_ARCH}/all && perl ${PORTSDIR}/infrastructure/package/check-modes ${PKGNAMES} || true
2006-10-23 10:33:01 -04:00
BULK_DO_${PKGPATH}?= ${BULK_DO}
move up pkgpath.mk, so that PKGPATH is available. Use it to define per-PKGPATH user-settings: BULK, WRKOBJDIR, ...
2004-02-07 17:34:02 -05:00
CLEANDEPENDS_${PKGPATH}?= ${CLEANDEPENDS}
move variables that only users can tweak, or that are constants, to the front of bsd.port.mk. This allows Makefile.inc and modules to use them more easily, and also cleans up things slightly. Also kill .include ${PORTSDIR}/../Makefile.inc (???)
2004-01-06 10:02:57 -05:00
# Commands and command settings.
PKG_DBDIR?= /var/db/pkg
FETCH_CMD?= /usr/bin/ftp -V -m
PKG_TMPDIR?= /var/tmp
PKG_CMD?= /usr/sbin/pkg_create
PKG_DELETE?= /usr/sbin/pkg_delete
# remount those mount points ro before fake.
# XXX tends to panic the OS
PROTECT_MOUNT_POINTS?=
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
Need to include Makefile.inc sooner.
2000-04-09 11:24:13 -04:00
.if exists(${.CURDIR}/../Makefile.inc)
.include "${.CURDIR}/../Makefile.inc"
.endif
More stringent PERMIT_* checking: if any info is missing, assume the maintainer made a mistake, and turn everything to No. Record it as _BAD_LICENSING for describe purpose. Yep, having this as ERRORS is obnoxious. Then, we can rely on PERMIT_* always being defined. Rewrite ftp-packages/cdrom-packages to use more current rules. Fixes naddy's issue of ignoring IGNORE. Use a .for loop, since those rules are basically identical. No longer use them to try and record port logging, as we have a much better script to do that. `Enhance' bulk-packages to do ftp-packages/cdrom-packages linking as well, so that one single sweep of the tree should be enough for everything.
2001-09-30 08:26:57 -04:00
.if !defined(PERMIT_PACKAGE_CDROM) || !defined(PERMIT_PACKAGE_FTP) || \
!defined(PERMIT_DISTFILES_CDROM) || !defined(PERMIT_DISTFILES_FTP)
ERRORS+="The licensing info for ${FULLPKGNAME} is incomplete."
ERRORS+="Please notify the OpenBSD port maintainer:"
ERRORS+=" ${MAINTAINER}"
_BAD_LICENSING=Yes
PERMIT_PACKAGE_CDROM = No
PERMIT_DISTFILES_CDROM = No
PERMIT_PACKAGE_FTP = No
PERMIT_DISTFILES_FTP = No
.endif
introspection mechanism: dump-vars
2006-07-09 07:10:16 -04:00
.if defined(verbose-show)
.MAIN: verbose-show
.elif defined(show)
Change WRKDIR name, to flatten things somewhat. work-* is too long, because we want to include the PKGNAME in the dir name (so that PKGNAME changes are less likely to work with old work directories) Change the way clean works, so that you can say make clean='dist depends packages' and have it do what you think it should do. Common fragment to _create_wrkobjdir. Put some info into PACKAGING (name of subpackage being built)
2001-08-16 10:49:31 -04:00
.MAIN: show
.elif defined(clean)
.MAIN: clean
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
.elif defined(_internal-clean)
clean=${_internal-clean}
.MAIN: _internal-clean
Change WRKDIR name, to flatten things somewhat. work-* is too long, because we want to include the PKGNAME in the dir name (so that PKGNAME changes are less likely to work with old work directories) Change the way clean works, so that you can say make clean='dist depends packages' and have it do what you think it should do. Common fragment to _create_wrkobjdir. Put some info into PACKAGING (name of subpackage being built)
2001-08-16 10:49:31 -04:00
.else
.MAIN: all
.endif
Turn FAKE=Yes on by default
2000-05-30 10:51:59 -04:00
FAKE?=Yes
`Fake' infrastructure: if a port sets FAKE=Yes, this means it can be `pre'-installed elsewhere, with DESTDIR set to WRKINST (=work/fake-${ARCH} by default). the infrastructure takes care of pre-install/do-install/post-install targets, assuming those install stuff under PREFIX. To help ports to cope with DESTDIR, you can set FAKE_FLAGS and FAKE_TARGET (used for the fake installation). Ports with FAKE=Yes are the way to go: they can be packaged directly without a real installation, and the installation proceeds from the package, thus forcing porters to check the package.
2000-03-03 16:24:50 -05:00
get thru _clean to handle clean, since it's set in stone. Add work if only depends is set. problem found by naddy@
2003-08-01 04:07:30 -04:00
# need to go through an extra var because clean is set in stone,
# on the cmdline.
_clean=${clean}
.if empty(_clean) || ${_clean:L} == "depends"
_clean+=work
.endif
move up pkgpath.mk, so that PKGPATH is available. Use it to define per-PKGPATH user-settings: BULK, WRKOBJDIR, ...
2004-02-07 17:34:02 -05:00
.if ${CLEANDEPENDS_${PKGPATH}:L} == "yes"
get thru _clean to handle clean, since it's set in stone. Add work if only depends is set. problem found by naddy@
2003-08-01 04:07:30 -04:00
_clean+=depends
Change WRKDIR name, to flatten things somewhat. work-* is too long, because we want to include the PKGNAME in the dir name (so that PKGNAME changes are less likely to work with old work directories) Change the way clean works, so that you can say make clean='dist depends packages' and have it do what you think it should do. Common fragment to _create_wrkobjdir. Put some info into PACKAGING (name of subpackage being built)
2001-08-16 10:49:31 -04:00
.endif
get thru _clean to handle clean, since it's set in stone. Add work if only depends is set. problem found by naddy@
2003-08-01 04:07:30 -04:00
.if ${_clean:L:Mwork}
_clean+=fake
Change WRKDIR name, to flatten things somewhat. work-* is too long, because we want to include the PKGNAME in the dir name (so that PKGNAME changes are less likely to work with old work directories) Change the way clean works, so that you can say make clean='dist depends packages' and have it do what you think it should do. Common fragment to _create_wrkobjdir. Put some info into PACKAGING (name of subpackage being built)
2001-08-16 10:49:31 -04:00
.endif
get thru _clean to handle clean, since it's set in stone. Add work if only depends is set. problem found by naddy@
2003-08-01 04:07:30 -04:00
.if ${_clean:L:Mforce}
_clean+=-f
Change WRKDIR name, to flatten things somewhat. work-* is too long, because we want to include the PKGNAME in the dir name (so that PKGNAME changes are less likely to work with old work directories) Change the way clean works, so that you can say make clean='dist depends packages' and have it do what you think it should do. Common fragment to _create_wrkobjdir. Put some info into PACKAGING (name of subpackage being built)
2001-08-16 10:49:31 -04:00
.endif
verify the spelling of clean options.
2004-12-19 07:20:12 -05:00
# check that clean is clean
_okay_words=depends work fake -f flavors dist install sub packages package \
readmes is yet another target that should recurse.
2006-09-19 16:10:40 -04:00
readmes bulk force
verify the spelling of clean options.
2004-12-19 07:20:12 -05:00
.for _w in ${_clean:L}
. if !${_okay_words:M${_w}}
ERRORS+="Fatal: unknown clean command: ${_w}"
. endif
.endfor
Change WRKDIR name, to flatten things somewhat. work-* is too long, because we want to include the PKGNAME in the dir name (so that PKGNAME changes are less likely to work with old work directories) Change the way clean works, so that you can say make clean='dist depends packages' and have it do what you think it should do. Common fragment to _create_wrkobjdir. Put some info into PACKAGING (name of subpackage being built)
2001-08-16 10:49:31 -04:00
make now can lowercase/uppercase variables. Use it. NOTE: YOU MUST HAVE A CURRENT MAKE FOR THIS FILE TO WORK !!!
1999-09-26 06:47:30 -04:00
NOMANCOMPRESS?= Yes
Accept WRKOBJDIR for NFS mounted ports tree and local objs. DEF_UMASK is different on OpenBSD. Was this OK to commit, imp?
1997-09-21 06:58:41 -04:00
DEF_UMASK?= 022
Merge changes from FreeBSD
1996-08-23 07:37:41 -04:00
Kill opsys cruft.
1999-11-28 09:12:24 -05:00
.if exists(${.CURDIR}/Makefile.${ARCH})
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
.include "${.CURDIR}/Makefile.${ARCH}"
use .elif
2004-01-06 11:27:11 -05:00
.elif exists(${.CURDIR}/Makefile.${MACHINE_ARCH})
Make MACHINE_ARCH available for PLIST. Allow for MACHINE_ARCH specific files or patches, if no such files for ARCH found. Put compiled packages in MACHINE_ARCH directory, instead of ARCH. All of this to improve support for compound architectures, mainly m68k-based architectures at this time. While there, mention mvme88k does not have shared libraries, to be sure this isn't forgotten when mvme88k gets back in shape. ok'ed by espie@
2001-05-05 19:42:03 -04:00
.include "${.CURDIR}/Makefile.${MACHINE_ARCH}"
.endif
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
move modules reading ways higher, after checking it doesn't affect a bulk build.
2006-07-11 02:34:32 -04:00
# MODULES support
# reserved name spaces: for module=NAME, modname*, _modname* variables and
# targets.
# These variables must be defined before modules
CONFIGURE_STYLE?=
enable config.x11.site if USE_X11 is set. ok espie@
2006-11-20 18:18:25 -05:00
USE_X11?= No
move error handling later for greater flexibility. Move default NO_* after including local Makefiles.
2001-10-03 04:41:16 -04:00
NO_DEPENDS?= No
NO_BUILD?= No
basic regression testing framework.
2001-10-03 04:53:18 -04:00
NO_REGRESS?= No
move modules reading ways higher, after checking it doesn't affect a bulk build.
2006-07-11 02:34:32 -04:00
INSTALL_TARGET?= install
.if ${CONFIGURE_STYLE:L:Mautomake} || ${CONFIGURE_STYLE:L:Mautoconf} || ${CONFIGURE_STYLE:L:Mautoupdate}
. if !${CONFIGURE_STYLE:L:Mgnu}
CONFIGURE_STYLE+=gnu
. endif
.endif
.for _i in perl gnu imake
. if ${CONFIGURE_STYLE:L:M${_i}}
MODULES+=${_i}
. endif
.endfor
.if defined(MODULES)
_MODULES_DONE=
. include "${PORTSDIR}/infrastructure/mk/modules.port.mk"
.endif
###
### Variable setup that can happen after modules
###
move PKG_ARCH after MODULES
2006-11-12 05:52:59 -05:00
.if ${MACHINE_ARCH} != ${ARCH}
PKG_ARCH?=${MACHINE_ARCH},${ARCH}
.else
PKG_ARCH?=${MACHINE_ARCH}
.endif
Move SHARED_ONLY default definition down so that it can be inherited from Makefile.inc/MODULES.
2004-08-03 15:30:25 -04:00
SHARED_ONLY?= No
Do not check if SEPARATE_BUILD is defined but check what is its value (defaults to No) help & blessing from espie@, ok naddy@
2005-04-20 21:44:50 -04:00
SEPARATE_BUILD?= No
Move SHARED_ONLY default definition down so that it can be inherited from Makefile.inc/MODULES.
2004-08-03 15:30:25 -04:00
always define DIST_SUBDIR, simplifies logic.
2004-02-07 17:36:13 -05:00
DIST_SUBDIR?=
move error handling later for greater flexibility. Move default NO_* after including local Makefiles.
2001-10-03 04:41:16 -04:00
always define DIST_SUBDIR, simplifies logic.
2004-02-07 17:36:13 -05:00
.if !empty(DIST_SUBDIR)
Rename _DISTDIR to FULLDISTDIR, make it a supported variable (so that I can clean up the few uses of underscore variables I've done... which were wrong; underscore variables are plain unsupported interface which may change)
1999-09-22 06:16:58 -04:00
FULLDISTDIR?= ${DISTDIR}/${DIST_SUBDIR}
Cosmetic patch: don't append DIST_SUBDIR if it's not defined. Set vim filetype, as vi doesn't care.
2000-02-02 10:28:17 -05:00
.else
FULLDISTDIR?= ${DISTDIR}
.endif
ftp-packages, cdrom-packages targets. These are the targets I use to build packages and log them.
2000-04-09 08:04:13 -04:00
Kill opsys cruft.
1999-11-28 09:12:24 -05:00
.if exists(${.CURDIR}/patches.${ARCH})
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
PATCHDIR?= ${.CURDIR}/patches.${ARCH}
.if .if .else .elif . if --> .else . else .endif . endif .endif
2004-02-07 17:14:21 -05:00
.elif exists(${.CURDIR}/patches.${MACHINE_ARCH})
Make MACHINE_ARCH available for PLIST. Allow for MACHINE_ARCH specific files or patches, if no such files for ARCH found. Put compiled packages in MACHINE_ARCH directory, instead of ARCH. All of this to improve support for compound architectures, mainly m68k-based architectures at this time. While there, mention mvme88k does not have shared libraries, to be sure this isn't forgotten when mvme88k gets back in shape. ok'ed by espie@
2001-05-05 19:42:03 -04:00
PATCHDIR?= ${.CURDIR}/patches.${MACHINE_ARCH}
.else
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
PATCHDIR?= ${.CURDIR}/patches
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
.endif
new functionalities: - PATCH_LIST for conditional patches - addsum for non destructive md5 checksum updates
1998-11-18 23:20:09 -05:00
PATCH_LIST?= patch-*
Kill opsys cruft.
1999-11-28 09:12:24 -05:00
.if exists(${.CURDIR}/files.${ARCH})
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
FILESDIR?= ${.CURDIR}/files.${ARCH}
.if .if .else .elif . if --> .else . else .endif . endif .endif
2004-02-07 17:14:21 -05:00
.elif exists(${.CURDIR}/files.${MACHINE_ARCH})
Make MACHINE_ARCH available for PLIST. Allow for MACHINE_ARCH specific files or patches, if no such files for ARCH found. Put compiled packages in MACHINE_ARCH directory, instead of ARCH. All of this to improve support for compound architectures, mainly m68k-based architectures at this time. While there, mention mvme88k does not have shared libraries, to be sure this isn't forgotten when mvme88k gets back in shape. ok'ed by espie@
2001-05-05 19:42:03 -04:00
FILESDIR?= ${.CURDIR}/files.${MACHINE_ARCH}
.else
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
FILESDIR?= ${.CURDIR}/files
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
.endif
Kill opsys cruft.
1999-11-28 09:12:24 -05:00
.if exists(${.CURDIR}/pkg.${ARCH})
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
PKGDIR?= ${.CURDIR}/pkg.${ARCH}
.if .if .else .elif . if --> .else . else .endif . endif .endif
2004-02-07 17:14:21 -05:00
.elif exists(${.CURDIR}/pkg.${MACHINE_ARCH})
Make MACHINE_ARCH available for PLIST. Allow for MACHINE_ARCH specific files or patches, if no such files for ARCH found. Put compiled packages in MACHINE_ARCH directory, instead of ARCH. All of this to improve support for compound architectures, mainly m68k-based architectures at this time. While there, mention mvme88k does not have shared libraries, to be sure this isn't forgotten when mvme88k gets back in shape. ok'ed by espie@
2001-05-05 19:42:03 -04:00
PKGDIR?= ${.CURDIR}/pkg.${MACHINE_ARCH}
.else
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
PKGDIR?= ${.CURDIR}/pkg
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
.endif
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
PREFIX?= ${LOCALBASE}
Always make TRUEPREFIX/DESTDIR available
2001-01-08 17:10:36 -05:00
TRUEPREFIX?= ${PREFIX}
Knob DESTDIRNAME.
2001-07-18 12:19:47 -04:00
DESTDIRNAME?= DESTDIR
Always make TRUEPREFIX/DESTDIR available
2001-01-08 17:10:36 -05:00
DESTDIR?= ${WRKINST}
tell gnu configure packages to stuff their files under /etc.
1999-03-16 18:35:37 -05:00
add LIBTOOL to MAKE_FLAGS when USE_LIBTOOL is defined.
2001-08-26 17:42:18 -04:00
MAKE_FLAGS?=
LIBTOOL_FLAGS?=
.if !defined(FAKE_FLAGS)
FAKE_FLAGS=${DESTDIRNAME}=${WRKINST}
.endif
tell gnu configure packages to stuff their files under /etc.
1999-03-16 18:35:37 -05:00
# where configuration files should go
SYSCONFDIR?= /etc
USE_LIBTOOL, USE_GMAKE, USE_X11 should really be Yes/No variables.
2002-04-09 09:52:58 -04:00
USE_GMAKE?= No
.if ${USE_GMAKE:L} == "yes"
revert
2002-12-08 06:04:47 -05:00
BUILD_DEPENDS+= ::devel/gmake
sync with NetBSD where possible. Fix WRKOBJDIR. Add deinstall target. Make uninstall an alias for deinstall.
1998-09-07 18:33:19 -04:00
MAKE_PROGRAM= ${GMAKE}
.else
MAKE_PROGRAM= ${MAKE}
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
.endif
move gnu dreck (autoconf, automake...) to gnu.port.mk
2002-03-15 20:09:23 -05:00
USE_LIBTOOL, USE_GMAKE, USE_X11 should really be Yes/No variables.
2002-04-09 09:52:58 -04:00
USE_LIBTOOL?=No
set LIBname_LTVERSION=-version-info <major>:<minor>:0 in MAKE_ENV/MAKE_FLAGS/FAKE_FLAGS if USE_LIBTOOL=Yes.
2005-12-29 07:48:04 -05:00
_lt_libs=
USE_LIBTOOL, USE_GMAKE, USE_X11 should really be Yes/No variables.
2002-04-09 09:52:58 -04:00
.if ${USE_LIBTOOL:L} == "yes"
Add DEPBASE for dependencies. Add USE_FAKE_LIB to turn on FAKE=LIB for ports that have been converted. Tweak buil/lib depends and LD_LIBRARY_PATH in FAKE=lib case.
2005-04-17 06:10:07 -04:00
LIBTOOL?= ${DEPBASE}/bin/libtool
Update dependencies (kill first part when needed). Update japanese/README... 2.4 is looong past.
2001-09-19 12:03:09 -04:00
BUILD_DEPENDS+= ::devel/libtool
set LIBname_LTVERSION=-version-info <major>:<minor>:0 in MAKE_ENV/MAKE_FLAGS/FAKE_FLAGS if USE_LIBTOOL=Yes.
2005-12-29 07:48:04 -05:00
CONFIGURE_ENV+= LIBTOOL="${LIBTOOL} ${LIBTOOL_FLAGS}" ${_lt_libs}
MAKE_ENV+= LIBTOOL="${LIBTOOL} ${LIBTOOL_FLAGS}" ${_lt_libs}
MAKE_FLAGS+= LIBTOOL="${LIBTOOL} ${LIBTOOL_FLAGS}" ${_lt_libs}
FAKE_FLAGS+= LIBTOOL="${LIBTOOL} ${LIBTOOL_FLAGS}" ${_lt_libs}
USE_LIBTOOL from NetBSD
1999-05-04 14:12:24 -04:00
.endif
move SHARED_LIBS_LOG into WRKBUILD
2006-10-21 07:50:48 -04:00
MAKE_FLAGS+= SHARED_LIBS_LOG=${WRKBUILD}/shared_libs.log
sync with NetBSD where possible. Fix WRKOBJDIR. Add deinstall target. Make uninstall an alias for deinstall.
1998-09-07 18:33:19 -04:00
all MULTI_PACKAGES are new, stop testing -main.
2006-11-26 12:36:07 -05:00
.if !defined(MULTI_PACKAGES)
MULTI_PACKAGES framework. This is somewhat orthogonal to FLAVORS. Principle: one port may build several packages in one go. For instance, egcs will build all compilers, and package stuff as base, C++, Fortran, etc. This simplifies some japanese ports greatly, e.g., one Canna port that builds libs/server/app packages. How to use: just set MULTI_PACKAGES=-ext1 -ext2... in the Makefile. Then make package will invoke itself recursively with SUBPACKAGE set to -ext1, -ext2. The SUBPACKAGE will fetch package info as PLIST-ext1, COMMENT-ext1, DESCR-ext1, etc. The port Makefile itself can test the value of SUBPACKAGE if things differ. Some problems: - there still is a main package (with PLIST, DESCR, etc), and it's built last, so you can't add RUN_DEPENDS=main_package to subpackages, you have to encode them explicitly in the PLIST for now. - other ports can't depend on subpackages yet. This is usually not a problem, it's just a question of choosing the `right' main package (e.g., for Canna, it will be canna-lib). - PLIST.sed recognize @FLAVOR@, which is extended to the current flavor, so that a subpackage may depend on the corresponding main package in explicit ways. describe is also invoked recursively, so that all subpackages appear in the INDEX.
2000-04-17 16:12:03 -04:00
SUBPACKAGE?=
start of multi-packages simplification: if MULTI_PACKAGES contains -main, consider it to be the main package, and do some equal treatment for each package. (to do: add subpackage where needed to WANTLIB, RUN_DEPENDS, LIB_DEPENDS and act on them accordingly). Also define _DONE_FILES for _fetch-makefile to avoid blocking if it's not invoked from a higher level target.
2006-11-17 19:15:33 -05:00
PKGNAMES=${FULLPKGNAME}
_FMN=${PKGPATH}/${FULLPKGNAME}
.else
SUBPACKAGE?=-main
_FMN=
.endif
dir -> dir[,-multi][,-flavor...] notation for dependencies. Finally !
2001-03-28 09:33:24 -05:00
FLAVOR?=
basic regression testing framework.
2001-10-03 04:53:18 -04:00
FLAVORS?=
PSEUDO_FLAVORS support from/ok espie
2002-08-07 11:48:19 -04:00
PSEUDO_FLAVORS?=
FLAVORS+=${PSEUDO_FLAVORS}
basic regression testing framework.
2001-10-03 04:53:18 -04:00
.if !empty(FLAVORS:L:Mregress) && empty(FLAVOR:L:Mregress)
NO_REGRESS= Yes
.endif
MULTI_PACKAGES framework. This is somewhat orthogonal to FLAVORS. Principle: one port may build several packages in one go. For instance, egcs will build all compilers, and package stuff as base, C++, Fortran, etc. This simplifies some japanese ports greatly, e.g., one Canna port that builds libs/server/app packages. How to use: just set MULTI_PACKAGES=-ext1 -ext2... in the Makefile. Then make package will invoke itself recursively with SUBPACKAGE set to -ext1, -ext2. The SUBPACKAGE will fetch package info as PLIST-ext1, COMMENT-ext1, DESCR-ext1, etc. The port Makefile itself can test the value of SUBPACKAGE if things differ. Some problems: - there still is a main package (with PLIST, DESCR, etc), and it's built last, so you can't add RUN_DEPENDS=main_package to subpackages, you have to encode them explicitly in the PLIST for now. - other ports can't depend on subpackages yet. This is usually not a problem, it's just a question of choosing the `right' main package (e.g., for Canna, it will be canna-lib). - PLIST.sed recognize @FLAVOR@, which is extended to the current flavor, so that a subpackage may depend on the corresponding main package in explicit ways. describe is also invoked recursively, so that all subpackages appear in the INDEX.
2000-04-17 16:12:03 -04:00
Misc. small tweaks. - let USE_MOTIF be sensible: any, openmotif, lesstif. any triggers a lesstif flavor. - show=name as a shortcut. - package-dir-depends to check that packages are okay. - error message for non-existent dirs.
2001-04-18 10:43:55 -04:00
USE_MOTIF?=No
.if ${USE_MOTIF:L} != "no"
. if ${USE_MOTIF:L} == "lesstif"
convert a few libspecs to new style. Note: no package bump, those are only BUILD dependencies, the pkg_create code will create the correct wantlib specs.
2006-08-01 06:50:19 -04:00
LIB_DEPENDS+= Xm.>=1::x11/lesstif
Misc. small tweaks. - let USE_MOTIF be sensible: any, openmotif, lesstif. any triggers a lesstif flavor. - show=name as a shortcut. - package-dir-depends to check that packages are okay. - error message for non-existent dirs.
2001-04-18 10:43:55 -04:00
. elif ${USE_MOTIF:L} == "openmotif"
convert a few libspecs to new style. Note: no package bump, those are only BUILD dependencies, the pkg_create code will create the correct wantlib specs.
2006-08-01 06:50:19 -04:00
LIB_DEPENDS+= Xm.>=2::x11/openmotif
Misc. small tweaks. - let USE_MOTIF be sensible: any, openmotif, lesstif. any triggers a lesstif flavor. - show=name as a shortcut. - package-dir-depends to check that packages are okay. - error message for non-existent dirs.
2001-04-18 10:43:55 -04:00
. elif ${USE_MOTIF:L} == "any" || ${USE_MOTIF:L} == "yes"
FLAVORS+=lesstif
can't set both motif and lesstif flavors at once.
2002-03-17 20:55:40 -05:00
. if ${FLAVOR:L:Mlesstif} && ${FLAVOR:L:Mmotif}
ERRORS+="Fatal: choose motif or lesstif, not both."
. endif
Misc. small tweaks. - let USE_MOTIF be sensible: any, openmotif, lesstif. any triggers a lesstif flavor. - show=name as a shortcut. - package-dir-depends to check that packages are okay. - error message for non-existent dirs.
2001-04-18 10:43:55 -04:00
. if ${FLAVOR:L:Mlesstif}
convert a few libspecs to new style. Note: no package bump, those are only BUILD dependencies, the pkg_create code will create the correct wantlib specs.
2006-08-01 06:50:19 -04:00
LIB_DEPENDS+= Xm.>=1::x11/lesstif
Misc. small tweaks. - let USE_MOTIF be sensible: any, openmotif, lesstif. any triggers a lesstif flavor. - show=name as a shortcut. - package-dir-depends to check that packages are okay. - error message for non-existent dirs.
2001-04-18 10:43:55 -04:00
. else
convert a few libspecs to new style. Note: no package bump, those are only BUILD dependencies, the pkg_create code will create the correct wantlib specs.
2006-08-01 06:50:19 -04:00
LIB_DEPENDS+= Xm.>=2::x11/openmotif
Misc. small tweaks. - let USE_MOTIF be sensible: any, openmotif, lesstif. any triggers a lesstif flavor. - show=name as a shortcut. - package-dir-depends to check that packages are okay. - error message for non-existent dirs.
2001-04-18 10:43:55 -04:00
. endif
. else
Replace .BEGIN lines with explicit ERRORS framework.
2001-07-17 12:11:45 -04:00
ERRORS+= "Fatal: Unknown USE_MOTIF=${USE_MOTIF} settings."
Misc. small tweaks. - let USE_MOTIF be sensible: any, openmotif, lesstif. any triggers a lesstif flavor. - show=name as a shortcut. - package-dir-depends to check that packages are okay. - error message for non-existent dirs.
2001-04-18 10:43:55 -04:00
. endif
Add DEPBASE for dependencies. Add USE_FAKE_LIB to turn on FAKE=LIB for ports that have been converted. Tweak buil/lib depends and LD_LIBRARY_PATH in FAKE=lib case.
2005-04-17 06:10:07 -04:00
MOTIFLIB=-L${DEPBASE}/lib -lXm
Misc. small tweaks. - let USE_MOTIF be sensible: any, openmotif, lesstif. any triggers a lesstif flavor. - show=name as a shortcut. - package-dir-depends to check that packages are okay. - error message for non-existent dirs.
2001-04-18 10:43:55 -04:00
.endif
- Fix FULLPKGPATH again. I missed it in naddy's buggy bug-fix. - save subdir into package, that's what is useful, since the name is already there. - check that SUBPACKAGE is a member of MULTI_PACKAGES. Make make SUBDIR=misc/fileutils,-ls work without error.
2001-04-17 12:45:14 -04:00
.if !empty(SUBPACKAGE)
. for _i in ${SUBPACKAGE}
. if !defined(MULTI_PACKAGES) || empty(MULTI_PACKAGES:M${_i})
Replace .BEGIN lines with explicit ERRORS framework.
2001-07-17 12:11:45 -04:00
ERRORS+= "Fatal: Subpackage ${SUBPACKAGE} does not exist."
- Fix FULLPKGPATH again. I missed it in naddy's buggy bug-fix. - save subdir into package, that's what is useful, since the name is already there. - check that SUBPACKAGE is a member of MULTI_PACKAGES. Make make SUBDIR=misc/fileutils,-ls work without error.
2001-04-17 12:45:14 -04:00
. endif
. endfor
.endif
fix size test finally (|| > |) check that MULTI_PACKAGES do begin with -.
2006-03-24 14:28:13 -05:00
.if defined(MULTI_PACKAGES) && !empty(MULTI_PACKAGES:N-*)
ERRORS+= "Fatal: SUBPACKAGES should always beging with -: ${MULTI_PACKAGES:N-*}."
.endif
- Fix FULLPKGPATH again. I missed it in naddy's buggy bug-fix. - save subdir into package, that's what is useful, since the name is already there. - check that SUBPACKAGE is a member of MULTI_PACKAGES. Make make SUBDIR=misc/fileutils,-ls work without error.
2001-04-17 12:45:14 -04:00
Move PKGFILE definition up, so that it can be used in dependencies.
2001-07-18 14:23:04 -04:00
# Build FLAVOR_EXT, checking that no flavors are misspelled
FLAVOR_EXT:=
define BUILD_PKGPATH/BASE_PKGPATH, to document later
2006-11-25 13:31:23 -05:00
BASE_PKGPATH:=${PKGPATH}
PSEUDO_FLAVORS support from/ok espie
2002-08-07 11:48:19 -04:00
# _FLAVOR_EXT2 is used internally for working directories.
# It encodes flavors and pseudo-flavors.
_FLAVOR_EXT2:=
define BUILD_PKGPATH/BASE_PKGPATH, to document later
2006-11-25 13:31:23 -05:00
BUILD_PKGPATH:=${PKGPATH}
Move the whole definition of FLAVOR_EXT up, so that it can be used directly. Tighten the flavor check, so that ports without flavors error out if flavors are specified.
2001-07-20 09:13:47 -04:00
# (applies only to PLIST for now)
basic regression testing framework.
2001-10-03 04:53:18 -04:00
.if !empty(FLAVORS)
Move the whole definition of FLAVOR_EXT up, so that it can be used directly. Tighten the flavor check, so that ports without flavors error out if flavors are specified.
2001-07-20 09:13:47 -04:00
. for _i in ${FLAVORS:L}
. if empty(FLAVOR:L:M${_i})
remove all SED_PLIST work, since pkg_create handles it now. defines appropriate PKG_ARG arguments instead. Move message to -M ${MESSAGE}.
2004-09-15 14:57:31 -04:00
PKG_ARGS+=-D${_i}=0
Move the whole definition of FLAVOR_EXT up, so that it can be used directly. Tighten the flavor check, so that ports without flavors error out if flavors are specified.
2001-07-20 09:13:47 -04:00
. else
PSEUDO_FLAVORS support from/ok espie
2002-08-07 11:48:19 -04:00
_FLAVOR_EXT2:=${_FLAVOR_EXT2}-${_i}
define BUILD_PKGPATH/BASE_PKGPATH, to document later
2006-11-25 13:31:23 -05:00
BUILD_PKGPATH:=${BUILD_PKGPATH},${_i}
PSEUDO_FLAVORS support from/ok espie
2002-08-07 11:48:19 -04:00
. if empty(PSEUDO_FLAVORS:L:M${_i})
Move the whole definition of FLAVOR_EXT up, so that it can be used directly. Tighten the flavor check, so that ports without flavors error out if flavors are specified.
2001-07-20 09:13:47 -04:00
FLAVOR_EXT:=${FLAVOR_EXT}-${_i}
define BUILD_PKGPATH/BASE_PKGPATH, to document later
2006-11-25 13:31:23 -05:00
BASE_PKGPATH:=${BASE_PKGPATH},${_i}
PSEUDO_FLAVORS support from/ok espie
2002-08-07 11:48:19 -04:00
. endif
remove all SED_PLIST work, since pkg_create handles it now. defines appropriate PKG_ARG arguments instead. Move message to -M ${MESSAGE}.
2004-09-15 14:57:31 -04:00
PKG_ARGS+=-D${_i}=1
Move the whole definition of FLAVOR_EXT up, so that it can be used directly. Tighten the flavor check, so that ports without flavors error out if flavors are specified.
2001-07-20 09:13:47 -04:00
. endif
. endfor
.endif
remove all SED_PLIST work, since pkg_create handles it now. defines appropriate PKG_ARG arguments instead. Move message to -M ${MESSAGE}.
2004-09-15 14:57:31 -04:00
.if ${NO_SHARED_LIBS:L} == "yes"
PKG_ARGS+=-DSHARED_LIBS=0
.else
PKG_ARGS+=-DSHARED_LIBS=1
.endif
explicitly forbid flavors starting with [0-9].
2004-09-14 19:06:02 -04:00
.if !empty(FLAVORS:M[0-9]*)
ERRORS+="Fatal: flavor should never start with a digit"
.endif
remove all SED_PLIST work, since pkg_create handles it now. defines appropriate PKG_ARG arguments instead. Move message to -M ${MESSAGE}.
2004-09-15 14:57:31 -04:00
Move the whole definition of FLAVOR_EXT up, so that it can be used directly. Tighten the flavor check, so that ports without flavors error out if flavors are specified.
2001-07-20 09:13:47 -04:00
.if !empty(FLAVOR)
basic regression testing framework.
2001-10-03 04:53:18 -04:00
. if !empty(FLAVORS)
Move PKGFILE definition up, so that it can be used in dependencies.
2001-07-18 14:23:04 -04:00
. for _i in ${FLAVOR:L}
. if empty(FLAVORS:L:M${_i})
ERRORS+= "Fatal: Unknown flavor: ${_i}"
ERRORS+= " (Possible flavors are: ${FLAVORS})."
. endif
. endfor
Move the whole definition of FLAVOR_EXT up, so that it can be used directly. Tighten the flavor check, so that ports without flavors error out if flavors are specified.
2001-07-20 09:13:47 -04:00
. else
ERRORS+= "Fatal: no flavors for this port."
Move PKGFILE definition up, so that it can be used in dependencies.
2001-07-18 14:23:04 -04:00
. endif
.endif
PKG_SUFX?= .tgz
Lots of internal changes. - define _FULLPKGNAME* and _PKGFILE*. Use _PKGFILE* as package cookie instead of package_done. - use ${PKG_DBDIR}/${_FULLPKGNAME${SUBPACKAGE}/+CONTENTS as install cookie. - BIN_PACKAGES variable. If set to yes, remove coupling between PACKAGE_COOKIE and dependencies, so that binary packages exist independently of working directories. - mark packages as precious. - change all refs to PKGFILE to _PKGFILE${SUBPACKAGE}. - add bulk-packages target. Note that BIN_PACKAGES and TRUST_PACKAGES are off by default. This interface is likely to change. The name bulk-packages will probably change as well. We will probably get rid of the old PKGFILE/FULLPKGNAME (taking SUBPACKAGE into account) and use the new PKGFILE/FULLPKGNAME PKGFILE-foo/FULLPKGNAME-foo instead (with explicit subpackage information). This is less confusing and simpler.
2001-07-30 10:13:16 -04:00
Change PKGNAME and FULLPKGNAME semantics slightly. Since those are mostly infrastructure-side, making SUBPACKAGE transparent only adds to the confusion, so don't interpolate SUBPACKAGE into FULLPKGNAME, hiding it into _FULLPKGNAME, but rather let FULLPKGNAME tag the main package, and use FULLPKGNAME-foo for subpackage -foo. This will simplify some other stuff...
2001-08-12 07:33:10 -04:00
PKGNAME?=${DISTNAME}
FULLPKGNAME?=${PKGNAME}${FLAVOR_EXT}
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
_MASTER?=
Strip spaces from CFLAGS everywhere. Pass off -a to pkg_add if used to solve dependencies.
2004-09-18 09:45:23 -04:00
_SOLVING_DEP?=No
Lots of internal changes. - define _FULLPKGNAME* and _PKGFILE*. Use _PKGFILE* as package cookie instead of package_done. - use ${PKG_DBDIR}/${_FULLPKGNAME${SUBPACKAGE}/+CONTENTS as install cookie. - BIN_PACKAGES variable. If set to yes, remove coupling between PACKAGE_COOKIE and dependencies, so that binary packages exist independently of working directories. - mark packages as precious. - change all refs to PKGFILE to _PKGFILE${SUBPACKAGE}. - add bulk-packages target. Note that BIN_PACKAGES and TRUST_PACKAGES are off by default. This interface is likely to change. The name bulk-packages will probably change as well. We will probably get rid of the old PKGFILE/FULLPKGNAME (taking SUBPACKAGE into account) and use the new PKGFILE/FULLPKGNAME PKGFILE-foo/FULLPKGNAME-foo instead (with explicit subpackage information). This is less confusing and simpler.
2001-07-30 10:13:16 -04:00
.if defined(MULTI_PACKAGES)
. for _s in ${MULTI_PACKAGES}
Change PKGNAME and FULLPKGNAME semantics slightly. Since those are mostly infrastructure-side, making SUBPACKAGE transparent only adds to the confusion, so don't interpolate SUBPACKAGE into FULLPKGNAME, hiding it into _FULLPKGNAME, but rather let FULLPKGNAME tag the main package, and use FULLPKGNAME-foo for subpackage -foo. This will simplify some other stuff...
2001-08-12 07:33:10 -04:00
. if !defined(FULLPKGNAME${_s})
. if defined(PKGNAME${_s})
FULLPKGNAME${_s} = ${PKGNAME${_s}}${FLAVOR_EXT}
. else
FULLPKGNAME${_s} = ${PKGNAME}${_s}${FLAVOR_EXT}
. endif
Lots of internal changes. - define _FULLPKGNAME* and _PKGFILE*. Use _PKGFILE* as package cookie instead of package_done. - use ${PKG_DBDIR}/${_FULLPKGNAME${SUBPACKAGE}/+CONTENTS as install cookie. - BIN_PACKAGES variable. If set to yes, remove coupling between PACKAGE_COOKIE and dependencies, so that binary packages exist independently of working directories. - mark packages as precious. - change all refs to PKGFILE to _PKGFILE${SUBPACKAGE}. - add bulk-packages target. Note that BIN_PACKAGES and TRUST_PACKAGES are off by default. This interface is likely to change. The name bulk-packages will probably change as well. We will probably get rid of the old PKGFILE/FULLPKGNAME (taking SUBPACKAGE into account) and use the new PKGFILE/FULLPKGNAME PKGFILE-foo/FULLPKGNAME-foo instead (with explicit subpackage information). This is less confusing and simpler.
2001-07-30 10:13:16 -04:00
. endif
. endfor
.endif
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
_SYSTRACE_COOKIE= ${WRKDIR}/systrace.policy
Use cookies for dependencies as well, and for the creation of a workdir. This can speed up some iterations, as a dependency won't be checked more than once. This also allows for make extract NO_DEPENDS=Yes, then later to have the dependencies pulled in at configure stage (even though this might not work all the time, as some dependencies may be needed earlier but then you'll certainly notice).
2001-09-05 05:13:18 -04:00
_WRKDIR_COOKIE= ${WRKDIR}/.extract_started
Make cookie names internal to bsd.port.mk Don't allow the user to override these. It's a major can of worms anyway.
1999-12-08 12:11:09 -05:00
_EXTRACT_COOKIE= ${WRKDIR}/.extract_done
_PATCH_COOKIE= ${WRKDIR}/.patch_done
Add intermediate `distpatch' target. This is done in a tricky way, so as not to disturb the usual sequence of targets. Namely, you still have fetch, extract, patch, configure, build, install, package, but patch does invoke a subtarget distpatch when needed, and you can fetch, extract, distpatch, patch, configure, build, install, package for the same results. Mostly useful for porters: provide an intermediate point to work on patches, and make it easy to override distpatch/OpenBSD patches separately.
2000-01-30 10:19:40 -05:00
_DISTPATCH_COOKIE= ${WRKDIR}/.distpatch_done
_PREPATCH_COOKIE= ${WRKDIR}/.prepatch_done
Change PKGNAME and FULLPKGNAME semantics slightly. Since those are mostly infrastructure-side, making SUBPACKAGE transparent only adds to the confusion, so don't interpolate SUBPACKAGE into FULLPKGNAME, hiding it into _FULLPKGNAME, but rather let FULLPKGNAME tag the main package, and use FULLPKGNAME-foo for subpackage -foo. This will simplify some other stuff...
2001-08-12 07:33:10 -04:00
_INSTALL_COOKIE= ${PKG_DBDIR}/${FULLPKGNAME${SUBPACKAGE}}/+CONTENTS
kill bulk-packages. Use `make package BULK=Yes' instead. The main advantage is that workdirs are cleaned up as soon as package is built (or installed), not later.
2002-04-17 11:58:48 -04:00
_BULK_COOKIE= ${BULK_COOKIES_DIR}/${FULLPKGNAME}
MULTI_PACKAGES framework. This is somewhat orthogonal to FLAVORS. Principle: one port may build several packages in one go. For instance, egcs will build all compilers, and package stuff as base, C++, Fortran, etc. This simplifies some japanese ports greatly, e.g., one Canna port that builds libs/server/app packages. How to use: just set MULTI_PACKAGES=-ext1 -ext2... in the Makefile. Then make package will invoke itself recursively with SUBPACKAGE set to -ext1, -ext2. The SUBPACKAGE will fetch package info as PLIST-ext1, COMMENT-ext1, DESCR-ext1, etc. The port Makefile itself can test the value of SUBPACKAGE if things differ. Some problems: - there still is a main package (with PLIST, DESCR, etc), and it's built last, so you can't add RUN_DEPENDS=main_package to subpackages, you have to encode them explicitly in the PLIST for now. - other ports can't depend on subpackages yet. This is usually not a problem, it's just a question of choosing the `right' main package (e.g., for Canna, it will be canna-lib). - PLIST.sed recognize @FLAVOR@, which is extended to the current flavor, so that a subpackage may depend on the corresponding main package in explicit ways. describe is also invoked recursively, so that all subpackages appear in the INDEX.
2000-04-17 16:12:03 -04:00
_FAKE_COOKIE= ${WRKINST}/.fake_done
`Fake' infrastructure: if a port sets FAKE=Yes, this means it can be `pre'-installed elsewhere, with DESTDIR set to WRKINST (=work/fake-${ARCH} by default). the infrastructure takes care of pre-install/do-install/post-install targets, assuming those install stuff under PREFIX. To help ports to cope with DESTDIR, you can set FAKE_FLAGS and FAKE_TARGET (used for the fake installation). Ports with FAKE=Yes are the way to go: they can be packaged directly without a real installation, and the installation proceeds from the package, thus forcing porters to check the package.
2000-03-03 16:24:50 -05:00
_INSTALL_PRE_COOKIE=${WRKINST}/.install_started
industrial-strength updates: - separate UPDATE_COOKIES_DIR, unless set to empty. Makes for easier management. - FORCE_UPDATE variable, that makes certain packages are updated when a dependency is run, and that forces the replacement of packages that are already there. Still uses the update cookie, so that this update occurs just once unless you wipe out yur UPDATE_COOKIES_DIR (which justifies putting update cookies in a separate directory).
2004-11-15 08:52:03 -05:00
.if !empty(UPDATE_COOKIES_DIR)
_UPDATE_COOKIE= ${UPDATE_COOKIES_DIR}/${FULLPKGNAME${SUBPACKAGE}}
.else
rename _UPDATE_COOKIE to always be defined (wilfried) and to have a name changing when the FULLPKGNAME changes.
2004-11-10 18:40:42 -05:00
_UPDATE_COOKIE= ${WRKDIR}/.update_${FULLPKGNAME${SUBPACKAGE}}
industrial-strength updates: - separate UPDATE_COOKIES_DIR, unless set to empty. Makes for easier management. - FORCE_UPDATE variable, that makes certain packages are updated when a dependency is run, and that forces the replacement of packages that are already there. Still uses the update cookie, so that this update occurs just once unless you wipe out yur UPDATE_COOKIES_DIR (which justifies putting update cookies in a separate directory).
2004-11-15 08:52:03 -05:00
.endif
Do not check if SEPARATE_BUILD is defined but check what is its value (defaults to No) help & blessing from espie@, ok naddy@
2005-04-20 21:44:50 -04:00
.if ${SEPARATE_BUILD:L} != "no"
Make cookie names internal to bsd.port.mk Don't allow the user to override these. It's a major can of worms anyway.
1999-12-08 12:11:09 -05:00
_CONFIGURE_COOKIE= ${WRKBUILD}/.configure_done
_BUILD_COOKIE= ${WRKBUILD}/.build_done
basic regression testing framework.
2001-10-03 04:53:18 -04:00
_REGRESS_COOKIE= ${WRKBUILD}/.regress_done
* make the _PORT_USE macro more explicit and less magic in its use of cookies, so that cookie names are no longer hard-coded. * new SEPARATE_BUILD feature: for ports that can build outside their source tree, use build-${ARCH} to build, and move configure/build/install cookie down inside build-${ARCH}. Those few people who use NFS to mount ports tree around different architectures may find this helps.
1999-02-28 18:23:47 -05:00
.else
Make cookie names internal to bsd.port.mk Don't allow the user to override these. It's a major can of worms anyway.
1999-12-08 12:11:09 -05:00
_CONFIGURE_COOKIE= ${WRKDIR}/.configure_done
_BUILD_COOKIE= ${WRKDIR}/.build_done
basic regression testing framework.
2001-10-03 04:53:18 -04:00
_REGRESS_COOKIE= ${WRKDIR}/.regress_done
* make the _PORT_USE macro more explicit and less magic in its use of cookies, so that cookie names are no longer hard-coded. * new SEPARATE_BUILD feature: for ports that can build outside their source tree, use build-${ARCH} to build, and move configure/build/install cookie down inside build-${ARCH}. Those few people who use NFS to mount ports tree around different architectures may find this helps.
1999-02-28 18:23:47 -05:00
.endif
MULTI_PACKAGES framework. This is somewhat orthogonal to FLAVORS. Principle: one port may build several packages in one go. For instance, egcs will build all compilers, and package stuff as base, C++, Fortran, etc. This simplifies some japanese ports greatly, e.g., one Canna port that builds libs/server/app packages. How to use: just set MULTI_PACKAGES=-ext1 -ext2... in the Makefile. Then make package will invoke itself recursively with SUBPACKAGE set to -ext1, -ext2. The SUBPACKAGE will fetch package info as PLIST-ext1, COMMENT-ext1, DESCR-ext1, etc. The port Makefile itself can test the value of SUBPACKAGE if things differ. Some problems: - there still is a main package (with PLIST, DESCR, etc), and it's built last, so you can't add RUN_DEPENDS=main_package to subpackages, you have to encode them explicitly in the PLIST for now. - other ports can't depend on subpackages yet. This is usually not a problem, it's just a question of choosing the `right' main package (e.g., for Canna, it will be canna-lib). - PLIST.sed recognize @FLAVOR@, which is extended to the current flavor, so that a subpackage may depend on the corresponding main package in explicit ways. describe is also invoked recursively, so that all subpackages appear in the INDEX.
2000-04-17 16:12:03 -04:00
Make cookie names internal to bsd.port.mk Don't allow the user to override these. It's a major can of worms anyway.
1999-12-08 12:11:09 -05:00
_ALL_COOKIES=${_EXTRACT_COOKIE} ${_PATCH_COOKIE} ${_CONFIGURE_COOKIE} \
basic regression testing framework.
2001-10-03 04:53:18 -04:00
${_INSTALL_PRE_COOKIE} ${_BUILD_COOKIE} ${_REGRESS_COOKIE} \
add _SYSTRACE_COOKIE to _ALL_COOKIES ok espie@
2004-06-22 16:05:46 -04:00
${_SYSTRACE_COOKIE} ${_PACKAGE_COOKIES} \
Use cookies for dependencies as well, and for the creation of a workdir. This can speed up some iterations, as a dependency won't be checked more than once. This also allows for make extract NO_DEPENDS=Yes, then later to have the dependencies pulled in at configure stage (even though this might not work all the time, as some dependencies may be needed earlier but then you'll certainly notice).
2001-09-05 05:13:18 -04:00
${_DISTPATCH_COOKIE} ${_PREPATCH_COOKIE} ${_FAKE_COOKIE} \
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
${_WRKDIR_COOKIE} ${_DEPBUILD_COOKIES} \
rename some internal variables: _DEPlib_COOKIES -> _DEPLIB_COOKIES. simplify some .if defined(A) . for i in $A ... . endfor .endfor into .for i in $A ... . endfor .endfor since empty loops work just fine.
2006-11-19 07:05:10 -05:00
${_DEPRUN_COOKIES} ${_DEPREGRESS_COOKIES} ${_UPDATE_COOKIE} \
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
${_DEPBUILDLIB_COOKIES} ${_DEPRUNLIB_COOKIES} \
typo
2006-11-20 07:04:40 -05:00
${_DEPBUILDWANTLIB_COOKIE} ${_DEPRUNWANTLIB_COOKIE} ${_DEPLIBSPECS_COOKIES}
Make cookie names internal to bsd.port.mk Don't allow the user to override these. It's a major can of worms anyway.
1999-12-08 12:11:09 -05:00
touch -f ? what's that ?
2005-11-05 18:41:36 -05:00
_MAKE_COOKIE=touch
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
# Miscellaneous overridable commands:
GMAKE?= gmake
Replace md5 checksum with a choice of sha1, rmd160, md5. Upward compatible.
1999-03-03 13:18:46 -05:00
transform files/md5 into distinfo magically. I would credit FreeBSD or NetBSD for the idea if I could remember which one came up with it first. :)
2001-11-11 07:59:50 -05:00
CHECKSUM_FILE?= ${.CURDIR}/distinfo
Replace md5 checksum with a choice of sha1, rmd160, md5. Upward compatible.
1999-03-03 13:18:46 -05:00
# Don't touch !!! Used for generating checksums.
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
_CIPHERS= sha1 rmd160 md5
Replace md5 checksum with a choice of sha1, rmd160, md5. Upward compatible.
1999-03-03 13:18:46 -05:00
# This is the one you can override
Introduce _MAKE_COOKIE, rename CIPHERS to _CIPHERS (and change it). Deprecate a whole slew of unneeded variables.
1999-12-03 09:57:12 -05:00
PREFERRED_CIPHERS?= ${_CIPHERS}
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
use both DEPBASE and LOCALBASE in PORTSPATH since we don't fake everything yet.
2005-04-17 14:31:58 -04:00
PORTPATH?= ${WRKDIR}/bin:/usr/bin:/bin:/usr/sbin:/sbin:${DEPBASE}/bin:${LOCALBASE}/bin:${X11BASE}/bin
- revert change now that we have patch -C - fix PATCH_LIST so that is works - add USE_AUTOCONF - fix PATH env logic plus some documentation...
1998-11-24 20:08:35 -05:00
pass COPTS in CFLAGS for those ports that do not override CFLAGS; NOTE: Some (most?) ports override CFLAGS. This includes all ports that use imake. Therefore this change is not as useful as it should be.
1999-03-10 18:22:19 -05:00
# Add any COPTS to CFLAGS. Note: programs that use imake do not
# use CFLAGS! Also, many (most?) ports hard code CFLAGS, ignoring
# what we pass in.
Don't append COPTS if it's not defined. This avoids passing CFLAGS="-O2 -g " with extra unwanted spaces off to submakes.
2001-07-11 08:57:20 -04:00
.if defined(COPTS)
pass COPTS in CFLAGS for those ports that do not override CFLAGS; NOTE: Some (most?) ports override CFLAGS. This includes all ports that use imake. Therefore this change is not as useful as it should be.
1999-03-10 18:22:19 -05:00
CFLAGS+= ${COPTS}
Don't append COPTS if it's not defined. This avoids passing CFLAGS="-O2 -g " with extra unwanted spaces off to submakes.
2001-07-11 08:57:20 -04:00
.endif
Synch with src (CDIAGFLAGS, CXXOPTS, WARNINGS)
2001-07-18 10:52:27 -04:00
.if defined(CXXOPTS)
CXXFLAGS+= ${CXXOPTS}
.endif
.if defined(WARNINGS) && ${WARNINGS:L} == "yes"
. if defined(CDIAGFLAGS)
CFLAGS+= ${CDIAGFLAGS}
. endif
. if defined(CXXDIAGFLAGS)
CXXFLAGS+= ${CXXDIAGFLAGS}
. endif
.endif
pass COPTS in CFLAGS for those ports that do not override CFLAGS; NOTE: Some (most?) ports override CFLAGS. This includes all ports that use imake. Therefore this change is not as useful as it should be.
1999-03-10 18:22:19 -05:00
MAKEFILE-> MAKE_FILE
2000-02-22 12:08:31 -05:00
MAKE_FILE?= Makefile
Trip ports that mistakenly try to write into $HOME.
2002-03-02 08:38:25 -05:00
PORTHOME?= /${PKGNAME}_writes_to_HOME
Extend CONFIGURE_STYLE to deal with GNU_CONFIGURE, HAS_CONFIGURE, DESTDIR tweaks for fake, old-style gnu-configure (no sysconf), imake, and autoconf...
2000-06-10 11:27:54 -04:00
MAKE_ENV+= PATH='${PORTPATH}' PREFIX='${PREFIX}' \
Add DEPBASE for dependencies. Add USE_FAKE_LIB to turn on FAKE=LIB for ports that have been converted. Tweak buil/lib depends and LD_LIBRARY_PATH in FAKE=lib case.
2005-04-17 06:10:07 -04:00
LOCALBASE='${LOCALBASE}' DEPBASE='${DEPBASE}' X11BASE='${X11BASE}' \
remove trailing spaces from CFLAGS/CXXFLAGS, to please autoconf...
2004-09-14 19:07:20 -04:00
MOTIFLIB='${MOTIFLIB}' CFLAGS='${CFLAGS:C/ *$//}' \
Trip ports that mistakenly try to write into $HOME.
2002-03-02 08:38:25 -05:00
TRUEPREFIX='${PREFIX}' ${DESTDIRNAME}='' \
HOME='${PORTHOME}'
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
.bak.orig looks nicer than bak.orig
2000-03-03 15:41:11 -05:00
DISTORIG?= .bak.orig
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
PATCH?= /usr/bin/patch
implement $PATCHORIG, a variable with the ability to change from .orig to another suffix for 'update-patches' and 'patch', since there are some distfiles that have foo and foo.orig already; ok espie@
2001-08-24 10:43:28 -04:00
PATCHORIG?= .orig
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
PATCH_STRIP?= -p0
PATCH_DIST_STRIP?= -p0
Clean PATCH_DEBUG handling
2000-09-16 09:54:56 -04:00
PATCH_DEBUG?=No
.if ${PATCH_DEBUG:L} != "no"
patch -b is now -z
2003-07-24 22:17:51 -04:00
PATCH_ARGS?= -d ${WRKDIST} -z ${PATCHORIG} -E ${PATCH_STRIP}
PATCH_DIST_ARGS?= -z ${DISTORIG} -d ${WRKDIST} -E ${PATCH_DIST_STRIP}
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
.else
patch -b is now -z
2003-07-24 22:17:51 -04:00
PATCH_ARGS?= -d ${WRKDIST} -z ${PATCHORIG} --forward --quiet -E ${PATCH_STRIP}
PATCH_DIST_ARGS?= -z ${DISTORIG} -d ${WRKDIST} --forward --quiet -E ${PATCH_DIST_STRIP}
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
.endif
Typo.
2000-09-23 18:03:21 -04:00
.if ${PATCH_CHECK_ONLY:L} == "yes"
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
PATCH_ARGS+= -C
PATCH_DIST_ARGS+= -C
.endif
add USE_ZIP for distfiles that use zip (.zip) for its packaging instead of tar and add USE_BZIP2 for distfiles that use bzip2 (.tar.bz2) for compression instead of gzip.
1999-05-14 00:38:36 -04:00
TAR?= /bin/tar
UNZIP?= unzip
BZIP2?= bzip2
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
Support an ONLY_FOR_ARCHS var, that prohibits doing anything for a certain port except for certain architectures. If OBJMACHINE is set set WRKDIR to work.${MACHINE}. Fix a shell construct which pdksh interprets differently than ash as it was written. Make bsd.own.mk be auto-included in make(1) invocations below this one.
1997-01-11 06:58:11 -05:00
MAKE_ENV+= EXTRA_SYS_MK_INCLUDES="<bsd.own.mk>"
sync with NetBSD where possible. Fix WRKOBJDIR. Add deinstall target. Make uninstall an alias for deinstall.
1998-09-07 18:33:19 -04:00
Use flavor in WRKINST/WRKBUILD name. Optimize space: if SEPARATE_BUILD contains `flavored', one WRKSRC for all flavors is enough.
2000-04-03 13:32:42 -04:00
move up pkgpath.mk, so that PKGPATH is available. Use it to define per-PKGPATH user-settings: BULK, WRKOBJDIR, ...
2004-02-07 17:34:02 -05:00
.if !empty(FAKEOBJDIR_${PKGPATH})
WRKINST?= ${FAKEOBJDIR_${PKGPATH}}/${PKGNAME}${_FLAVOR_EXT2}
Add FAKEOBJDIR support
2003-08-21 16:22:45 -04:00
.else
WRKINST?= ${WRKDIR}/fake-${ARCH}${_FLAVOR_EXT2}
.endif
move up pkgpath.mk, so that PKGPATH is available. Use it to define per-PKGPATH user-settings: BULK, WRKOBJDIR, ...
2004-02-07 17:34:02 -05:00
.if !empty(WRKOBJDIR_${PKGPATH})
Do not check if SEPARATE_BUILD is defined but check what is its value (defaults to No) help & blessing from espie@, ok naddy@
2005-04-20 21:44:50 -04:00
. if ${SEPARATE_BUILD:L:Mflavored}
move up pkgpath.mk, so that PKGPATH is available. Use it to define per-PKGPATH user-settings: BULK, WRKOBJDIR, ...
2004-02-07 17:34:02 -05:00
WRKDIR?= ${WRKOBJDIR_${PKGPATH}}/${PKGNAME}
Make WRKOBJDIR be a basis for WRKDIR when defined. Do not create a symlink from CURDIR. Based on a suggestion of Hakan, tested and approved by various ports people.
2003-01-14 13:18:23 -05:00
. else
move up pkgpath.mk, so that PKGPATH is available. Use it to define per-PKGPATH user-settings: BULK, WRKOBJDIR, ...
2004-02-07 17:34:02 -05:00
WRKDIR?= ${WRKOBJDIR_${PKGPATH}}/${PKGNAME}${_FLAVOR_EXT2}
Make WRKOBJDIR be a basis for WRKDIR when defined. Do not create a symlink from CURDIR. Based on a suggestion of Hakan, tested and approved by various ports people.
2003-01-14 13:18:23 -05:00
. endif
sync with NetBSD where possible. Fix WRKOBJDIR. Add deinstall target. Make uninstall an alias for deinstall.
1998-09-07 18:33:19 -04:00
.else
Do not check if SEPARATE_BUILD is defined but check what is its value (defaults to No) help & blessing from espie@, ok naddy@
2005-04-20 21:44:50 -04:00
. if ${SEPARATE_BUILD:L:Mflavored}
Change WRKDIR name, to flatten things somewhat. work-* is too long, because we want to include the PKGNAME in the dir name (so that PKGNAME changes are less likely to work with old work directories) Change the way clean works, so that you can say make clean='dist depends packages' and have it do what you think it should do. Common fragment to _create_wrkobjdir. Put some info into PACKAGING (name of subpackage being built)
2001-08-16 10:49:31 -04:00
WRKDIR?= ${.CURDIR}/w-${PKGNAME}
Use flavor in WRKINST/WRKBUILD name. Optimize space: if SEPARATE_BUILD contains `flavored', one WRKSRC for all flavors is enough.
2000-04-03 13:32:42 -04:00
. else
PSEUDO_FLAVORS support from/ok espie
2002-08-07 11:48:19 -04:00
WRKDIR?= ${.CURDIR}/w-${PKGNAME}${_FLAVOR_EXT2}
Use flavor in WRKINST/WRKBUILD name. Optimize space: if SEPARATE_BUILD contains `flavored', one WRKSRC for all flavors is enough.
2000-04-03 13:32:42 -04:00
. endif
sync with NetBSD where possible. Fix WRKOBJDIR. Add deinstall target. Make uninstall an alias for deinstall.
1998-09-07 18:33:19 -04:00
.endif
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
Introduce WRKDIST: place where stuff unpacks. May be different from where source lives. And from where build happens. NO_WRKSUBDIR=Yes is equivalent to WRKDIST=${WRKDIR}, and not particularly smaller, deprecated it. Remove CLEANDISTORIG, use patch -b instead. Note that DISTORIG ends in .orig to let rm *.orig still work...
2000-02-08 19:23:26 -05:00
WRKDIST?= ${WRKDIR}/${DISTNAME}
sync with NetBSD where possible. Fix WRKOBJDIR. Add deinstall target. Make uninstall an alias for deinstall.
1998-09-07 18:33:19 -04:00
and of course, nobody spotted the obvious typo in the patch I gave on ports...
2000-02-08 19:54:33 -05:00
WRKSRC?= ${WRKDIST}
Introduce WRKDIST: place where stuff unpacks. May be different from where source lives. And from where build happens. NO_WRKSUBDIR=Yes is equivalent to WRKDIST=${WRKDIR}, and not particularly smaller, deprecated it. Remove CLEANDISTORIG, use patch -b instead. Note that DISTORIG ends in .orig to let rm *.orig still work...
2000-02-08 19:23:26 -05:00
Do not check if SEPARATE_BUILD is defined but check what is its value (defaults to No) help & blessing from espie@, ok naddy@
2005-04-20 21:44:50 -04:00
.if ${SEPARATE_BUILD:L} != "no"
PSEUDO_FLAVORS support from/ok espie
2002-08-07 11:48:19 -04:00
WRKBUILD?= ${WRKDIR}/build-${MACHINE_ARCH}${_FLAVOR_EXT2}
Change WRKDIR name, to flatten things somewhat. work-* is too long, because we want to include the PKGNAME in the dir name (so that PKGNAME changes are less likely to work with old work directories) Change the way clean works, so that you can say make clean='dist depends packages' and have it do what you think it should do. Common fragment to _create_wrkobjdir. Put some info into PACKAGING (name of subpackage being built)
2001-08-16 10:49:31 -04:00
WRKPKG?= ${WRKBUILD}/pkg
* make the _PORT_USE macro more explicit and less magic in its use of cookies, so that cookie names are no longer hard-coded. * new SEPARATE_BUILD feature: for ports that can build outside their source tree, use build-${ARCH} to build, and move configure/build/install cookie down inside build-${ARCH}. Those few people who use NFS to mount ports tree around different architectures may find this helps.
1999-02-28 18:23:47 -05:00
.else
Provisions for building in a separate directory tah sources
1999-02-03 12:53:13 -05:00
WRKBUILD?= ${WRKSRC}
Change WRKDIR name, to flatten things somewhat. work-* is too long, because we want to include the PKGNAME in the dir name (so that PKGNAME changes are less likely to work with old work directories) Change the way clean works, so that you can say make clean='dist depends packages' and have it do what you think it should do. Common fragment to _create_wrkobjdir. Put some info into PACKAGING (name of subpackage being built)
2001-08-16 10:49:31 -04:00
WRKPKG?= ${WRKDIR}/pkg
* make the _PORT_USE macro more explicit and less magic in its use of cookies, so that cookie names are no longer hard-coded. * new SEPARATE_BUILD feature: for ports that can build outside their source tree, use build-${ARCH} to build, and move configure/build/install cookie down inside build-${ARCH}. Those few people who use NFS to mount ports tree around different architectures may find this helps.
1999-02-28 18:23:47 -05:00
.endif
WRKCONF?=${WRKBUILD}
2004-09-15 14:58:49 -04:00
WRKCONF?= ${WRKBUILD}
Provisions for building in a separate directory tah sources
1999-02-03 12:53:13 -05:00
Cut out some parts of bsd.port.mk into separate modules. This is work-in-progress. Supplemental hooks will be added when they become necessary. The goal is to permit separate subsystems to tweak the package building process slightly. This should help e.g., python or qt2, for instance. Documentation will come when it's tested more thoroughly...
2001-08-25 07:23:46 -04:00
ALL_TARGET?= all
FAKE_TARGET ?= ${INSTALL_TARGET}
typo
2001-10-04 18:20:38 -04:00
REGRESS_TARGET ?= regress
Seems like we can't avoid REGRESS_ENV/REGRESS_FLAGS. Some testsuites use HOME to test their getenv function. Which seems to be reasonable. But we definitely don't want to allow HOME into MAKE_ENV by default.
2001-10-07 06:50:47 -04:00
REGRESS_FLAGS ?= ${MAKE_FLAGS}
basic regression testing framework.
2001-10-03 04:53:18 -04:00
Lots of internal changes. - define _FULLPKGNAME* and _PKGFILE*. Use _PKGFILE* as package cookie instead of package_done. - use ${PKG_DBDIR}/${_FULLPKGNAME${SUBPACKAGE}/+CONTENTS as install cookie. - BIN_PACKAGES variable. If set to yes, remove coupling between PACKAGE_COOKIE and dependencies, so that binary packages exist independently of working directories. - mark packages as precious. - change all refs to PKGFILE to _PKGFILE${SUBPACKAGE}. - add bulk-packages target. Note that BIN_PACKAGES and TRUST_PACKAGES are off by default. This interface is likely to change. The name bulk-packages will probably change as well. We will probably get rid of the old PKGFILE/FULLPKGNAME (taking SUBPACKAGE into account) and use the new PKGFILE/FULLPKGNAME PKGFILE-foo/FULLPKGNAME-foo instead (with explicit subpackage information). This is less confusing and simpler.
2001-07-30 10:13:16 -04:00
_PACKAGE_COOKIE_DEPS=${_FAKE_COOKIE}
.for _s in ${MULTI_PACKAGES}
- new read-only variable, PKGNAMES. - plist repository under PLIST_DB (optional). - print-package-signatures shows what's used for signatures. - make fetch shows full url you can copy/paste, simpler to look for typos that way.
2005-09-04 18:32:37 -04:00
PKGNAMES += ${FULLPKGNAME${_s}}
revise the way package multi-builds are handled. - move FLAVOR_EXT up so that WRK* are correct when they're needed for dependencies. - build a list of _PACKAGE_COOKIES, one for each subpackage. - let package go through an indirection: package: ${_PACKAGE_COOKIES} _PACKAGE_COOKIE_${SUB}: SUBPACKAGE=${SUB} ${MAKE} _package so that all packages are `flat', and subpackages can depend on each other freely.
2001-04-02 06:35:51 -04:00
.endfor
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
.for _s in ${MULTI_PACKAGES}
. for _v in PKG_ARCH PERMIT_PACKAGE_FTP PERMIT_PACKAGE_CDROM \
allows PREFIX and CATEGORIES to be subpackage dependent
2006-11-20 08:50:16 -05:00
RUN_DEPENDS WANTLIB LIB_DEPENDS PREFIX CATEGORIES
index PKG_ARCH, PERMIT_PACKAGE_FTP, PERMIT_PACKAGE_CDROM on SUBPACKAGE name.
2006-11-13 09:14:57 -05:00
${_v}${_s} ?= ${${_v}}
. endfor
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
.endfor
.for _s in ${MULTI_PACKAGES}
. for _v in MESSAGE UNMESSAGE DESCR PLIST
. if defined(${_v})
${_v}${_s} ?= ${${_v}}
. endif
. endfor
.endfor
index PKG_ARCH, PERMIT_PACKAGE_FTP, PERMIT_PACKAGE_CDROM on SUBPACKAGE name.
2006-11-13 09:14:57 -05:00
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
_PACKAGE_LINKS=
_PKGFILE= ${FULLPKGNAME${SUBPACKAGE}}${PKG_SUFX}
put no-arch into a variable, so that people can override it.
2006-10-18 06:51:39 -04:00
NO_ARCH?= no-arch
allows one to short-circuit PKG_ARCH=* in the build process by setting NO_ARCH to MACHINE_ARCH/all
2006-11-17 12:16:16 -05:00
.if ${PKG_ARCH${SUBPACKAGE}} == "*" && ${NO_ARCH} != ${MACHINE_ARCH}/all
put no-arch into a variable, so that people can override it.
2006-10-18 06:51:39 -04:00
_PACKAGE_COOKIE= ${PACKAGE_REPOSITORY}/${NO_ARCH}/${_PKGFILE}
_PACKAGE_LINKS+= ${MACHINE_ARCH}/all ${NO_ARCH}
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
_PACKAGE_COOKIES+= ${PACKAGE_REPOSITORY}/${MACHINE_ARCH}/all/${_PKGFILE}
.else
_PACKAGE_COOKIE= ${PACKAGE_REPOSITORY}/${MACHINE_ARCH}/all/${_PKGFILE}
.endif
_PKG_REPO= ${PACKAGE_REPOSITORY}/${MACHINE_ARCH}/all/
fix FETCH_PACKAGES: create a real target (packages/arch/cache/pkgname.tgz) and link to it from elsewhere. Fixes the no-arch case, and makes it ways easier to find out what was retrieved through the cache.
2006-10-02 05:37:14 -04:00
_CACHE_REPO= ${PACKAGE_REPOSITORY}/${MACHINE_ARCH}/cache/
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
PKGFILE=${_PKG_REPO}${_PKGFILE}
_PACKAGE_COOKIES += ${_PACKAGE_COOKIE}
index PKG_ARCH, PERMIT_PACKAGE_FTP, PERMIT_PACKAGE_CDROM on SUBPACKAGE name.
2006-11-13 09:14:57 -05:00
.if ${PERMIT_PACKAGE_FTP${SUBPACKAGE}:L} == "yes"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
_PACKAGE_COOKIES += ${PACKAGE_REPOSITORY}/${MACHINE_ARCH}/ftp/${_PKGFILE}
_PACKAGE_LINKS += ${MACHINE_ARCH}/ftp ${MACHINE_ARCH}/all
.endif
index PKG_ARCH, PERMIT_PACKAGE_FTP, PERMIT_PACKAGE_CDROM on SUBPACKAGE name.
2006-11-13 09:14:57 -05:00
.if ${PERMIT_PACKAGE_CDROM${SUBPACKAGE}:L} == "yes"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
_PACKAGE_COOKIES += ${PACKAGE_REPOSITORY}/${MACHINE_ARCH}/cdrom/${_PKGFILE}
_PACKAGE_LINKS += ${MACHINE_ARCH}/cdrom ${MACHINE_ARCH}/all
.endif
FULLPKGPATH is PKGPATH plus flavor/multipackage modifiers. Use it where we can, including WRKOBJDIR `real' names. NOTE: if you are using WRKOBJDIR, *all* flavored builds directories will change names, from work-no_x11 to work,no_x11
2001-04-08 12:55:17 -04:00
.if empty(SUBPACKAGE)
oops
2005-09-17 10:47:40 -04:00
FULLPKGPATH=${PKGPATH}${FLAVOR_EXT:S/-/,/g}
FULLPKGPATH is PKGPATH plus flavor/multipackage modifiers. Use it where we can, including WRKOBJDIR `real' names. NOTE: if you are using WRKOBJDIR, *all* flavored builds directories will change names, from work-no_x11 to work,no_x11
2001-04-08 12:55:17 -04:00
.else
oops
2005-09-17 10:47:40 -04:00
FULLPKGPATH=${PKGPATH},${SUBPACKAGE}${FLAVOR_EXT:S/-/,/g}
FULLPKGPATH is PKGPATH plus flavor/multipackage modifiers. Use it where we can, including WRKOBJDIR `real' names. NOTE: if you are using WRKOBJDIR, *all* flavored builds directories will change names, from work-no_x11 to work,no_x11
2001-04-08 12:55:17 -04:00
.endif
Merge changes from FreeBSD
1996-08-23 07:37:41 -04:00
# A few aliases for *-install targets
INSTALL_PROGRAM= \
COPY -> INSTALL_COPY and STRIP -> INSTALL_STRIP changes. For some reason the previous commit did not grab these.
1997-04-27 17:38:33 -04:00
${INSTALL} ${INSTALL_COPY} ${INSTALL_STRIP} -o ${BINOWN} -g ${BINGRP} -m ${BINMODE}
Merge changes from FreeBSD
1996-08-23 07:37:41 -04:00
INSTALL_SCRIPT= \
COPY -> INSTALL_COPY and STRIP -> INSTALL_STRIP changes. For some reason the previous commit did not grab these.
1997-04-27 17:38:33 -04:00
${INSTALL} ${INSTALL_COPY} -o ${BINOWN} -g ${BINGRP} -m ${BINMODE}
Merge changes from FreeBSD
1996-08-23 07:37:41 -04:00
INSTALL_DATA= \
COPY -> INSTALL_COPY and STRIP -> INSTALL_STRIP changes. For some reason the previous commit did not grab these.
1997-04-27 17:38:33 -04:00
${INSTALL} ${INSTALL_COPY} -o ${SHAREOWN} -g ${SHAREGRP} -m ${SHAREMODE}
Merge changes from FreeBSD
1996-08-23 07:37:41 -04:00
INSTALL_MAN= \
COPY -> INSTALL_COPY and STRIP -> INSTALL_STRIP changes. For some reason the previous commit did not grab these.
1997-04-27 17:38:33 -04:00
${INSTALL} ${INSTALL_COPY} -o ${MANOWN} -g ${MANGRP} -m ${MANMODE}
Use DIRMODE to install dirs, avoid collisions with mtree (keep a duplicate of DIRMODE until 2.7)
2000-02-08 19:26:32 -05:00
sync with NetBSD where possible. Fix WRKOBJDIR. Add deinstall target. Make uninstall an alias for deinstall.
1998-09-07 18:33:19 -04:00
INSTALL_PROGRAM_DIR= \
Use DIRMODE to install dirs, avoid collisions with mtree (keep a duplicate of DIRMODE until 2.7)
2000-02-08 19:26:32 -05:00
${INSTALL} -d -o ${BINOWN} -g ${BINGRP} -m ${DIRMODE}
sync with NetBSD where possible. Fix WRKOBJDIR. Add deinstall target. Make uninstall an alias for deinstall.
1998-09-07 18:33:19 -04:00
INSTALL_SCRIPT_DIR= \
${INSTALL_PROGRAM_DIR}
INSTALL_DATA_DIR= \
Use DIRMODE to install dirs, avoid collisions with mtree (keep a duplicate of DIRMODE until 2.7)
2000-02-08 19:26:32 -05:00
${INSTALL} -d -o ${SHAREOWN} -g ${SHAREGRP} -m ${DIRMODE}
sync with NetBSD where possible. Fix WRKOBJDIR. Add deinstall target. Make uninstall an alias for deinstall.
1998-09-07 18:33:19 -04:00
INSTALL_MAN_DIR= \
Use DIRMODE to install dirs, avoid collisions with mtree (keep a duplicate of DIRMODE until 2.7)
2000-02-08 19:26:32 -05:00
${INSTALL} -d -o ${MANOWN} -g ${MANGRP} -m ${DIRMODE}
Merge changes from FreeBSD
1996-08-23 07:37:41 -04:00
INSTALL_MACROS does not need to be visible.
2000-09-17 12:26:57 -04:00
_INSTALL_MACROS= BSD_INSTALL_PROGRAM="${INSTALL_PROGRAM}" \
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
BSD_INSTALL_SCRIPT="${INSTALL_SCRIPT}" \
BSD_INSTALL_DATA="${INSTALL_DATA}" \
sync with NetBSD where possible. Fix WRKOBJDIR. Add deinstall target. Make uninstall an alias for deinstall.
1998-09-07 18:33:19 -04:00
BSD_INSTALL_MAN="${INSTALL_MAN}" \
BSD_INSTALL_PROGRAM_DIR="${INSTALL_PROGRAM_DIR}" \
BSD_INSTALL_SCRIPT_DIR="${INSTALL_SCRIPT_DIR}" \
BSD_INSTALL_DATA_DIR="${INSTALL_DATA_DIR}" \
BSD_INSTALL_MAN_DIR="${INSTALL_MAN_DIR}"
INSTALL_MACROS does not need to be visible.
2000-09-17 12:26:57 -04:00
MAKE_ENV+= ${_INSTALL_MACROS}
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
# setup systrace variables
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
NO_SYSTRACE?= No
.if ${USE_SYSTRACE:L} == "yes" && ${NO_SYSTRACE:L} == "no"
let systrace log to stderr
2004-01-11 10:04:01 -05:00
_SYSTRACE_CMD?= /bin/systrace -e -i -a -f ${_SYSTRACE_COOKIE}
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
.else
_SYSTRACE_CMD=
.endif
use TRUEPREFIX locally sort SYSTRACE_SUBST_VARS while here
2004-11-19 17:12:28 -05:00
SYSTRACE_FILTER?= ${PORTSDIR}/infrastructure/db/systrace.filter
Unconditionally use do-extract and create that target in case it doesn't exist with the old, regular EXTRACT_CASES stuff. This allows for systrace protected extracts. espie@ ok
2003-08-11 14:42:07 -04:00
_SYSTRACE_POLICIES+= /bin/sh /usr/bin/env /usr/bin/make \
protect patch targets with systrace as well no objection from espie, looks good steven
2006-06-18 06:10:04 -04:00
/usr/bin/patch ${DEPBASE}/bin/gmake
use TRUEPREFIX locally sort SYSTRACE_SUBST_VARS while here
2004-11-19 17:12:28 -05:00
SYSTRACE_SUBST_VARS+= DISTDIR PKG_TMPDIR PORTSDIR TMPDIR WRKDIR
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
.for _v in ${SYSTRACE_SUBST_VARS}
_SYSTRACE_SED_SUBST+=-e 's,$${${_v}},${${_v}},g'
.endfor
basic framework to have control over shared libs numbers: SHARED_LIBS enumerates the libraries like so: SHARED_LIBS= foo 1.0 \ zop 2.5 \ tag 3.0 This just defines LIBfoo_VERSION=1.0, LIBzop_VERSION=2.5, LIBtag_VERSION=3.0 and adds these to SUBST_VAR, further tweaks must be done, usually at the MAKE_FLAGS/MAKE_ENV level. For gnu ports that use automake, the supplementary MODGNU_SHARED_LIBS variable can be used like this: MODGNU_SHARED_LIBS= foo '-no-undefined' and it will define libfoo_la_LD_FLAGS=--version-info 1:0:0 -no-undefined Some gnu ports have weird variable names for libraries. For instance, the libORBit-2 is set using libORBit_2. LIBORbit-2_ALIAS=ORBit_2 will make sure the correct name is used. much feedback from bernd@
2005-12-23 07:41:37 -05:00
SHARED_LIBS?=
.for _n _v in ${SHARED_LIBS}
LIB${_n}_VERSION=${_v}
SUBST_VARS+=LIB${_n}_VERSION
set LIBname_LTVERSION=-version-info <major>:<minor>:0 in MAKE_ENV/MAKE_FLAGS/FAKE_FLAGS if USE_LIBTOOL=Yes.
2005-12-29 07:48:04 -05:00
_lt_libs+=LIB${_n}_LTVERSION='-version-info ${_v:S/./:/}:0'
prepare to pass stuff over to libtool.
2006-01-05 14:33:17 -05:00
_lt_libs+=lib${_n:S/+/_/g:S/-/_/g:S/./_/g}_ltversion=${_v}
basic framework to have control over shared libs numbers: SHARED_LIBS enumerates the libraries like so: SHARED_LIBS= foo 1.0 \ zop 2.5 \ tag 3.0 This just defines LIBfoo_VERSION=1.0, LIBzop_VERSION=2.5, LIBtag_VERSION=3.0 and adds these to SUBST_VAR, further tweaks must be done, usually at the MAKE_FLAGS/MAKE_ENV level. For gnu ports that use automake, the supplementary MODGNU_SHARED_LIBS variable can be used like this: MODGNU_SHARED_LIBS= foo '-no-undefined' and it will define libfoo_la_LD_FLAGS=--version-info 1:0:0 -no-undefined Some gnu ports have weird variable names for libraries. For instance, the libORBit-2 is set using libORBit_2. LIBORbit-2_ALIAS=ORBit_2 will make sure the correct name is used. much feedback from bernd@
2005-12-23 07:41:37 -05:00
.endfor
Generate PLIST, MESSAGE, DESCR from templates systematically. - the PLIST, MESSAGE, DESCR variables refer to the sources. - files are generated under ${WRKBUILD}, always. - SUBST_VARS refer to the variables being substituted. ARCH, HOMEPAGE, PREFIX are always substituted. Use $\{VAR} to escape substitution. Plus ${FLAVORS} -> -flavor1-flavor2 for PLIST. - the %%FRAG%% and %%!FRAG%% notations are always used for PLIST. - SED_PLIST can be augmented if needed. Plus some compatibility kludges with the current situation.
2000-06-09 12:26:54 -04:00
# Create the generic variable substitution list, from subst vars
add BASE_PKGPATH to SUBST_VARS
2006-11-25 14:47:53 -05:00
SUBST_VARS+=MACHINE_ARCH ARCH HOMEPAGE PREFIX SYSCONFDIR FLAVOR_EXT \
MAINTAINER BASE_PKGPATH
almost completely new make-plist. The new version uses the pkg_* infrastructure to read existing plists and produced new plists. Lots and lots of improvements: - copy most stuff over from original lists correctly. For instance, attach @exec/@unexec/@sample to other files and copy them in the right location. - generic post-treatment of @commnent, using stringize(), so that most things can be commented out and will stay commented out. - parse all packing-lists using relevant PREFIXes. Grab PFRAG.xx if %xx% is seen. - walk through all of fake, and distpatch files according to PREFIXes if nothing else works. - identifies man, libraries, info, handling symlink correctly.
2004-08-05 19:43:45 -04:00
_tmpvars=
Generate PLIST, MESSAGE, DESCR from templates systematically. - the PLIST, MESSAGE, DESCR variables refer to the sources. - files are generated under ${WRKBUILD}, always. - SUBST_VARS refer to the variables being substituted. ARCH, HOMEPAGE, PREFIX are always substituted. Use $\{VAR} to escape substitution. Plus ${FLAVORS} -> -flavor1-flavor2 for PLIST. - the %%FRAG%% and %%!FRAG%% notations are always used for PLIST. - SED_PLIST can be augmented if needed. Plus some compatibility kludges with the current situation.
2000-06-09 12:26:54 -04:00
_SED_SUBST=sed
Strip spaces from CFLAGS everywhere. Pass off -a to pkg_add if used to solve dependencies.
2004-09-18 09:45:23 -04:00
better naming, discussed with nikolay
2005-11-05 06:16:17 -05:00
_PKG_ADD_AUTO?=
Strip spaces from CFLAGS everywhere. Pass off -a to pkg_add if used to solve dependencies.
2004-09-18 09:45:23 -04:00
.if ${_SOLVING_DEP:L} == "yes"
better naming, discussed with nikolay
2005-11-05 06:16:17 -05:00
_PKG_ADD_AUTO+=-a
Strip spaces from CFLAGS everywhere. Pass off -a to pkg_add if used to solve dependencies.
2004-09-18 09:45:23 -04:00
.endif
Generate PLIST, MESSAGE, DESCR from templates systematically. - the PLIST, MESSAGE, DESCR variables refer to the sources. - files are generated under ${WRKBUILD}, always. - SUBST_VARS refer to the variables being substituted. ARCH, HOMEPAGE, PREFIX are always substituted. Use $\{VAR} to escape substitution. Plus ${FLAVORS} -> -flavor1-flavor2 for PLIST. - the %%FRAG%% and %%!FRAG%% notations are always used for PLIST. - SED_PLIST can be augmented if needed. Plus some compatibility kludges with the current situation.
2000-06-09 12:26:54 -04:00
.for _v in ${SUBST_VARS}
use '|' rather than ',' for sed substitution delimiter since the comma clashed with multiple names into MAINTAINER (separated with comma), now that it is part of SUBST_VARS. '|' should be safer wrt all other variables there; ok espie
2003-07-11 12:46:20 -04:00
_SED_SUBST+=-e 's|$${${_v}}|${${_v}}|g'
remove all SED_PLIST work, since pkg_create handles it now. defines appropriate PKG_ARG arguments instead. Move message to -M ${MESSAGE}.
2004-09-15 14:57:31 -04:00
PKG_ARGS+=-D${_v}='${${_v}}'
almost completely new make-plist. The new version uses the pkg_* infrastructure to read existing plists and produced new plists. Lots and lots of improvements: - copy most stuff over from original lists correctly. For instance, attach @exec/@unexec/@sample to other files and copy them in the right location. - generic post-treatment of @commnent, using stringize(), so that most things can be commented out and will stay commented out. - parse all packing-lists using relevant PREFIXes. Grab PFRAG.xx if %xx% is seen. - walk through all of fake, and distpatch files according to PREFIXes if nothing else works. - identifies man, libraries, info, handling symlink correctly.
2004-08-05 19:43:45 -04:00
_tmpvars += ${_v}='${${_v}}'
Generate PLIST, MESSAGE, DESCR from templates systematically. - the PLIST, MESSAGE, DESCR variables refer to the sources. - files are generated under ${WRKBUILD}, always. - SUBST_VARS refer to the variables being substituted. ARCH, HOMEPAGE, PREFIX are always substituted. Use $\{VAR} to escape substitution. Plus ${FLAVORS} -> -flavor1-flavor2 for PLIST. - the %%FRAG%% and %%!FRAG%% notations are always used for PLIST. - SED_PLIST can be augmented if needed. Plus some compatibility kludges with the current situation.
2000-06-09 12:26:54 -04:00
.endfor
remove all SED_PLIST work, since pkg_create handles it now. defines appropriate PKG_ARG arguments instead. Move message to -M ${MESSAGE}.
2004-09-15 14:57:31 -04:00
PKG_ARGS+=-DFLAVORS='${FLAVOR_EXT}'
only pass FULLPKGPATH, PERMIT_PACKAGE_CDROM, PERMIT_PACKAGE_FTP to pkg_create, and quote the later properly...
2004-11-15 21:08:59 -05:00
PKG_ARGS+=-DFULLPKGPATH=${FULLPKGPATH}
index PKG_ARCH, PERMIT_PACKAGE_FTP, PERMIT_PACKAGE_CDROM on SUBPACKAGE name.
2006-11-13 09:14:57 -05:00
PKG_ARGS+=-DPERMIT_PACKAGE_CDROM=${PERMIT_PACKAGE_CDROM${SUBPACKAGE}:Q}
PKG_ARGS+=-DPERMIT_PACKAGE_FTP=${PERMIT_PACKAGE_FTP${SUBPACKAGE}:Q}
almost completely new make-plist. The new version uses the pkg_* infrastructure to read existing plists and produced new plists. Lots and lots of improvements: - copy most stuff over from original lists correctly. For instance, attach @exec/@unexec/@sample to other files and copy them in the right location. - generic post-treatment of @commnent, using stringize(), so that most things can be commented out and will stay commented out. - parse all packing-lists using relevant PREFIXes. Grab PFRAG.xx if %xx% is seen. - walk through all of fake, and distpatch files according to PREFIXes if nothing else works. - identifies man, libraries, info, handling symlink correctly.
2004-08-05 19:43:45 -04:00
_tmpvars += FLAVORS='${FLAVOR_EXT}'
remove all SED_PLIST work, since pkg_create handles it now. defines appropriate PKG_ARG arguments instead. Move message to -M ${MESSAGE}.
2004-09-15 14:57:31 -04:00
_SED_SUBST+=-e 's,$${FLAVORS},${FLAVOR_EXT},g' -e 's,$$\\,$$,g'
Flavored ports
2000-03-29 10:59:50 -05:00
Generate PLIST, MESSAGE, DESCR from templates systematically. - the PLIST, MESSAGE, DESCR variables refer to the sources. - files are generated under ${WRKBUILD}, always. - SUBST_VARS refer to the variables being substituted. ARCH, HOMEPAGE, PREFIX are always substituted. Use $\{VAR} to escape substitution. Plus ${FLAVORS} -> -flavor1-flavor2 for PLIST. - the %%FRAG%% and %%!FRAG%% notations are always used for PLIST. - SED_PLIST can be augmented if needed. Plus some compatibility kludges with the current situation.
2000-06-09 12:26:54 -04:00
# find out the most appropriate PLIST source
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
.if !defined(PLIST${SUBPACKAGE})
. if exists(${PKGDIR}/PLIST${SUBPACKAGE}${FLAVOR_EXT}.${ARCH})
PLIST${SUBPACKAGE}= ${PKGDIR}/PLIST${SUBPACKAGE}${FLAVOR_EXT}.${ARCH}
. elif exists(${PKGDIR}/PLIST${SUBPACKAGE}${FLAVOR_EXT}.${MACHINE_ARCH})
PLIST${SUBPACKAGE}= ${PKGDIR}/PLIST${SUBPACKAGE}${FLAVOR_EXT}.${MACHINE_ARCH}
. elif ${NO_SHARED_LIBS:L} == "yes" && exists(${PKGDIR}/PLIST${SUBPACKAGE}${FLAVOR_EXT}.noshared)
PLIST${SUBPACKAGE}= ${PKGDIR}/PLIST${SUBPACKAGE}${FLAVOR_EXT}.noshared
. elif exists(${PKGDIR}/PLIST${SUBPACKAGE}${FLAVOR_EXT})
PLIST${SUBPACKAGE}= ${PKGDIR}/PLIST${SUBPACKAGE}${FLAVOR_EXT}
. elif exists(${PKGDIR}/PLIST${SUBPACKAGE}.${ARCH})
PLIST${SUBPACKAGE}= ${PKGDIR}/PLIST${SUBPACKAGE}.${ARCH}
. elif exists(${PKGDIR}/PLIST${SUBPACKAGE}.${MACHINE_ARCH})
PLIST${SUBPACKAGE}= ${PKGDIR}/PLIST${SUBPACKAGE}.${MACHINE_ARCH}
. elif ${NO_SHARED_LIBS:L} == "yes" && exists(${PKGDIR}/PLIST${SUBPACKAGE}.noshared)
PLIST${SUBPACKAGE}= ${PKGDIR}/PLIST${SUBPACKAGE}.noshared
. else
PLIST${SUBPACKAGE}= ${PKGDIR}/PLIST${SUBPACKAGE}
Always define NO_SHARED_LIBS.
2004-08-02 09:01:52 -04:00
. endif
Make MACHINE_ARCH available for PLIST. Allow for MACHINE_ARCH specific files or patches, if no such files for ARCH found. Put compiled packages in MACHINE_ARCH directory, instead of ARCH. All of this to improve support for compound architectures, mainly m68k-based architectures at this time. While there, mention mvme88k does not have shared libraries, to be sure this isn't forgotten when mvme88k gets back in shape. ok'ed by espie@
2001-05-05 19:42:03 -04:00
.endif
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
Generate PLIST, MESSAGE, DESCR from templates systematically. - the PLIST, MESSAGE, DESCR variables refer to the sources. - files are generated under ${WRKBUILD}, always. - SUBST_VARS refer to the variables being substituted. ARCH, HOMEPAGE, PREFIX are always substituted. Use $\{VAR} to escape substitution. Plus ${FLAVORS} -> -flavor1-flavor2 for PLIST. - the %%FRAG%% and %%!FRAG%% notations are always used for PLIST. - SED_PLIST can be augmented if needed. Plus some compatibility kludges with the current situation.
2000-06-09 12:26:54 -04:00
# Likewise for DESCR/MESSAGE/COMMENT
_FEXT -> FLAVOR_EXT, make this visible FULLPKGNAME as a knob for ex ${PKGNAME}-${_FEXT}, so that python and others can tweak it directly, without needing post bsd.port.mk tweaks.
2001-03-21 19:22:36 -05:00
.if defined(COMMENT${SUBPACKAGE}${FLAVOR_EXT})
kill some old stuff: - all COMMENTs live in Makefiles now. - NEW_DEPENDS is always on. - 2.8 warning is woefully out of date (rohee@)
2001-05-23 09:28:14 -04:00
_COMMENT=${COMMENT${SUBPACKAGE}${FLAVOR_EXT}}
Allow for COMMENT-foo= inside the Makefile. NetBSD did this a while ago, this is a natural idea. It just took me some time to figure out how to do that in a MULTI_PACKAGE context, while maintaining compatibility with existing stuff.
2001-03-16 09:56:41 -05:00
.elif defined(COMMENT${SUBPACKAGE})
kill some old stuff: - all COMMENTs live in Makefiles now. - NEW_DEPENDS is always on. - 2.8 warning is woefully out of date (rohee@)
2001-05-23 09:28:14 -04:00
_COMMENT=${COMMENT${SUBPACKAGE}}
Flavored ports
2000-03-29 10:59:50 -05:00
.endif
Generate PLIST, MESSAGE, DESCR from templates systematically. - the PLIST, MESSAGE, DESCR variables refer to the sources. - files are generated under ${WRKBUILD}, always. - SUBST_VARS refer to the variables being substituted. ARCH, HOMEPAGE, PREFIX are always substituted. Use $\{VAR} to escape substitution. Plus ${FLAVORS} -> -flavor1-flavor2 for PLIST. - the %%FRAG%% and %%!FRAG%% notations are always used for PLIST. - SED_PLIST can be augmented if needed. Plus some compatibility kludges with the current situation.
2000-06-09 12:26:54 -04:00
.if exists(${PKGDIR}/MESSAGE${SUBPACKAGE})
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
MESSAGE${SUBPACKAGE}?= ${PKGDIR}/MESSAGE${SUBPACKAGE}
Generate PLIST, MESSAGE, DESCR from templates systematically. - the PLIST, MESSAGE, DESCR variables refer to the sources. - files are generated under ${WRKBUILD}, always. - SUBST_VARS refer to the variables being substituted. ARCH, HOMEPAGE, PREFIX are always substituted. Use $\{VAR} to escape substitution. Plus ${FLAVORS} -> -flavor1-flavor2 for PLIST. - the %%FRAG%% and %%!FRAG%% notations are always used for PLIST. - SED_PLIST can be augmented if needed. Plus some compatibility kludges with the current situation.
2000-06-09 12:26:54 -04:00
.endif
UNMESSAGE
2004-10-11 06:32:04 -04:00
.if exists(${PKGDIR}/UNMESSAGE${SUBPACKAGE})
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
UNMESSAGE${SUBPACKAGE}?= ${PKGDIR}/UNMESSAGE${SUBPACKAGE}
UNMESSAGE
2004-10-11 06:32:04 -04:00
.endif
Generate PLIST, MESSAGE, DESCR from templates systematically. - the PLIST, MESSAGE, DESCR variables refer to the sources. - files are generated under ${WRKBUILD}, always. - SUBST_VARS refer to the variables being substituted. ARCH, HOMEPAGE, PREFIX are always substituted. Use $\{VAR} to escape substitution. Plus ${FLAVORS} -> -flavor1-flavor2 for PLIST. - the %%FRAG%% and %%!FRAG%% notations are always used for PLIST. - SED_PLIST can be augmented if needed. Plus some compatibility kludges with the current situation.
2000-06-09 12:26:54 -04:00
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
DESCR${SUBPACKAGE}?= ${PKGDIR}/DESCR${SUBPACKAGE}
Flavored ports
2000-03-29 10:59:50 -05:00
Create an intermediate mtree.spec that is used for fake installations. Use of VAR_SUBSTS will solve a few nasty problems.
2001-03-06 13:59:38 -05:00
MTREE_FILE?=
MTREE_FILE+=${PORTSDIR}/infrastructure/db/fake.mtree
Generate PLIST, MESSAGE, DESCR from templates systematically. - the PLIST, MESSAGE, DESCR variables refer to the sources. - files are generated under ${WRKBUILD}, always. - SUBST_VARS refer to the variables being substituted. ARCH, HOMEPAGE, PREFIX are always substituted. Use $\{VAR} to escape substitution. Plus ${FLAVORS} -> -flavor1-flavor2 for PLIST. - the %%FRAG%% and %%!FRAG%% notations are always used for PLIST. - SED_PLIST can be augmented if needed. Plus some compatibility kludges with the current situation.
2000-06-09 12:26:54 -04:00
# Fill out package command, and package dependencies
switch to using @wantlib and @depends, say byebye to libdepends and newdepends. (this will break your machine if you don't have reasonably current pkg_add)
2004-11-27 04:59:35 -05:00
_PKG_PREREQ= ${WRKPKG}/DESCR${SUBPACKAGE} ${WRKPKG}/COMMENT${SUBPACKAGE}
remove -v in pkg_create(1) since pkg_create is going to become much more verbose.
2005-06-25 06:40:50 -04:00
PKG_ARGS+= -c '${WRKPKG}/COMMENT${SUBPACKAGE}' -d ${WRKPKG}/DESCR${SUBPACKAGE}
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
PKG_ARGS+=-f ${PLIST${SUBPACKAGE}} -p ${PREFIX${SUBPACKAGE}}
remove all SED_PLIST work, since pkg_create handles it now. defines appropriate PKG_ARG arguments instead. Move message to -M ${MESSAGE}.
2004-09-15 14:57:31 -04:00
.if exists(${PKGDIR}/INSTALL${SUBPACKAGE})
PKG_ARGS+= -i ${PKGDIR}/INSTALL${SUBPACKAGE}
.endif
.if exists(${PKGDIR}/DEINSTALL${SUBPACKAGE})
PKG_ARGS+= -k ${PKGDIR}/DEINSTALL${SUBPACKAGE}
.endif
.if exists(${PKGDIR}/REQ${SUBPACKAGE})
PKG_ARGS+= -r ${PKGDIR}/REQ${SUBPACKAGE}
.endif
- new read-only variable, PKGNAMES. - plist repository under PLIST_DB (optional). - print-package-signatures shows what's used for signatures. - make fetch shows full url you can copy/paste, simpler to look for typos that way.
2005-09-04 18:32:37 -04:00
.if exists(${PKGDIR}/MODULE${SUBPACKAGE}.pm)
PKG_ARGS+= -m ${PKGDIR}/MODULE${SUBPACKAGE}.pm
.endif
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
.if defined(MESSAGE${SUBPACKAGE})
PKG_ARGS+= -M ${MESSAGE${SUBPACKAGE}}
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
.endif
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
.if defined(UNMESSAGE${SUBPACKAGE})
PKG_ARGS+= -U ${UNMESSAGE${SUBPACKAGE}}
UNMESSAGE
2004-10-11 06:32:04 -04:00
.endif
prefer pkg_create -B
2004-08-10 09:48:21 -04:00
PKG_ARGS+= -B ${WRKINST}
index PKG_ARCH, PERMIT_PACKAGE_FTP, PERMIT_PACKAGE_CDROM on SUBPACKAGE name.
2006-11-13 09:14:57 -05:00
PKG_ARGS+=-A'${PKG_ARCH${SUBPACKAGE}}'
use -L if necessary.
2004-01-28 17:15:21 -05:00
.if ${LOCALBASE} != "/usr/local"
PKG_ARGS+=-L${LOCALBASE}
.endif
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
.if !defined(_COMMENT)
ERRORS+="Fatal: Missing comment."
.endif
`Fake' infrastructure: if a port sets FAKE=Yes, this means it can be `pre'-installed elsewhere, with DESTDIR set to WRKINST (=work/fake-${ARCH} by default). the infrastructure takes care of pre-install/do-install/post-install targets, assuming those install stuff under PREFIX. To help ports to cope with DESTDIR, you can set FAKE_FLAGS and FAKE_TARGET (used for the fake installation). Ports with FAKE=Yes are the way to go: they can be packaged directly without a real installation, and the installation proceeds from the package, thus forcing porters to check the package.
2000-03-03 16:24:50 -05:00
- update MASTER_SITE_XCONTRIB and MASTER_SITE_GNU - add new variables CHMOD and CHOWN, set to full pathnames of those commands.
1999-08-22 19:44:51 -04:00
CHMOD?= /bin/chmod
CHOWN?= /usr/sbin/chown
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
GUNZIP_CMD?= /usr/bin/gunzip -f
Merge our chagnes with FreeBSD's latest changes. We should now have identical bsd.port.mk files again. This file has successfully compiled many of the archivers, gnu make, and emacs (althought the emacs deltas have not been merged back into the FreeBSD tree yet since I don't have it working yet). The FreeBSD Id line may be slightly off since I generated it by hand to match the commit I just made there.
1996-12-25 15:10:09 -05:00
GZCAT?= /usr/bin/gzcat
GZIP?= -9
GZIP_CMD?= /usr/bin/gzip -nf ${GZIP}
new functionalities: - PATCH_LIST for conditional patches - addsum for non destructive md5 checksum updates
1998-11-18 23:20:09 -05:00
M4?= /usr/bin/m4
add new variable STRIP, set to full pathname of strip command
1999-09-02 17:57:32 -04:00
STRIP?= /usr/bin/strip
- revert change now that we have patch -C - fix PATCH_LIST so that is works - add USE_AUTOCONF - fix PATH env logic plus some documentation...
1998-11-24 20:08:35 -05:00
Pass YACC to configure script. Ensure reproducability of builds for autoconf ports: those would tend to use bison by default otherwise.
1999-12-08 12:00:15 -05:00
# Autoconf scripts MAY tend to use bison by default otherwise
YACC?=yacc
- revert change now that we have patch -C - fix PATCH_LIST so that is works - add USE_AUTOCONF - fix PATH env logic plus some documentation...
1998-11-24 20:08:35 -05:00
# XXX ${SETENV} is needed in front of var=value lists whenever the next
# command is expanded from a variable, as this could be a shell construct
Use setenv -i, for predictability.
2000-05-16 14:14:29 -04:00
SETENV?= /usr/bin/env -i
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
SH?= /bin/sh
Introduce _MAKE_COOKIE, rename CIPHERS to _CIPHERS (and change it). Deprecate a whole slew of unneeded variables.
1999-12-03 09:57:12 -05:00
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
# Used to print all the '===>' style prompts - override this to turn them off.
Use echo directly, except where ECHO_MSG makes sense. Clean up error conditions somewhat.
1999-11-22 18:44:01 -05:00
ECHO_MSG?= echo
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
Add multiple MASTER_SITES/PATCH_SITES
1999-09-26 06:50:07 -04:00
# basic master sites configuration
Add ftp.openbsd.org as a MASTER_SITE_BACKUP before ftp.freebsd.org; If MASTER_SITE_OPENBSD is defined only ftp.openbsd.org will be used; Re-do the way COMES_WITH was implemented to allow overide on the command line by defining NO_IGNORE.
1998-04-04 23:20:38 -05:00
Use db/network.conf only if it exists, defaults to the template otherwise.
2000-01-27 10:32:15 -05:00
.if exists(${PORTSDIR}/infrastructure/db/network.conf)
ftp sites configuration now live in network.conf
1999-09-26 06:48:47 -04:00
.include "${PORTSDIR}/infrastructure/db/network.conf"
Use db/network.conf only if it exists, defaults to the template otherwise.
2000-01-27 10:32:15 -05:00
.else
.include "${PORTSDIR}/infrastructure/templates/network.conf.template"
.endif
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
Add multiple MASTER_SITES/PATCH_SITES
1999-09-26 06:50:07 -04:00
# Empty declarations to avoid "variable XXX is recursive" errors
MASTER_SITES?=
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
# I guess we're in the master distribution business! :) As we gain mirror
# sites for distfiles, add them to this list.
.if !defined(MASTER_SITE_OVERRIDE)
say goodbye to MASTER_SITE_SUBDIR
2002-05-15 14:23:21 -04:00
MASTER_SITES:= ${MASTER_SITES} ${MASTER_SITE_BACKUP}
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
.else
say goodbye to MASTER_SITE_SUBDIR
2002-05-15 14:23:21 -04:00
MASTER_SITES:= ${MASTER_SITE_OVERRIDE} ${MASTER_SITES}
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
.endif
Finish cleansing PATCH_SITES out.
2000-02-04 06:15:16 -05:00
# _SITE_SELECTOR chooses the value of sites based on select.
_SITE_SELECTOR=case $$select in
Wrap MASTER_SITESn/PATCH_SITESn into a .for loop. Don't bother computing MASTER_SITESn/PATCH_SITESn if they're not set in the first place...
1999-12-21 16:43:37 -05:00
.for _I in 0 1 2 3 4 5 6 7 8 9
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. if defined(MASTER_SITES${_I})
. if !defined(MASTER_SITE_OVERRIDE)
say goodbye to MASTER_SITE_SUBDIR
2002-05-15 14:23:21 -04:00
MASTER_SITES${_I}:= ${MASTER_SITES${_I}} ${MASTER_SITE_BACKUP}
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. else
say goodbye to MASTER_SITE_SUBDIR
2002-05-15 14:23:21 -04:00
MASTER_SITES${_I}:= ${MASTER_SITE_OVERRIDE} ${MASTER_SITES${_I}}
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. endif
Finish cleansing PATCH_SITES out.
2000-02-04 06:15:16 -05:00
_SITE_SELECTOR+=*:${_I}) sites="${MASTER_SITES${_I}}";;
make sure the site selector has an entry for each :<digit>, ensuring incorrect entries will be spotted.
2003-08-11 16:07:59 -04:00
. else
_SITE_SELECTOR+=*:${_I}) echo >&2 "Error: MASTER_SITES${_I} not defined";;
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. endif
Wrap MASTER_SITESn/PATCH_SITESn into a .for loop. Don't bother computing MASTER_SITESn/PATCH_SITESn if they're not set in the first place...
1999-12-21 16:43:37 -05:00
.endfor
Finish cleansing PATCH_SITES out.
2000-02-04 06:15:16 -05:00
_SITE_SELECTOR+=*) sites="${MASTER_SITES}";; esac
Wrap MASTER_SITESn/PATCH_SITESn into a .for loop. Don't bother computing MASTER_SITESn/PATCH_SITESn if they're not set in the first place...
1999-12-21 16:43:37 -05:00
change CDROM_SITE to default to empty; ok espie
2004-09-19 05:00:15 -04:00
# OpenBSD code to handle ports distfiles on a CDROM.
existing fetch from cdrom was freebsd specific and would not work on OpenBSD; Add code to copy or link from cdrom that works with OpenBSD.
1998-04-06 17:46:00 -04:00
#
change CDROM_SITE to default to empty; ok espie
2004-09-19 05:00:15 -04:00
#CDROM_SITE?= /cdrom/distfiles/${DIST_SUBDIR}
CDROM_SITE?=
existing fetch from cdrom was freebsd specific and would not work on OpenBSD; Add code to copy or link from cdrom that works with OpenBSD.
1998-04-06 17:46:00 -04:00
slight semantic change in CDROM_SITE.
2004-01-28 17:13:40 -05:00
.if !empty(CDROM_SITE)
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. if defined(FETCH_SYMLINK_DISTFILES)
Make _CDROM_OVERRIDE work again
2000-07-20 22:38:33 -04:00
_CDROM_OVERRIDE=if ln -s ${CDROM_SITE}/$$f .; then exit 0; fi
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. else
Make _CDROM_OVERRIDE work again
2000-07-20 22:38:33 -04:00
_CDROM_OVERRIDE=if cp -f ${CDROM_SITE}/$$f .; then exit 0; fi
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. endif
Clean do-fetch target: it can depend on the files it's trying to fetch. Plays some make trickery to get all details correct. Numerous benefits: - make automatically handles badly fetched files, - possible extension to special-case *each* file fetch, - can be easily parallelized. Similar rules for fetch-list soon coming. Also simplify CDROM handling while we're at it, no need for 10 knobs when one suffices.
1999-12-23 19:27:58 -05:00
.else
_CDROM_OVERRIDE=:
.endif
Deprecate old EXTRACT_* variables. Use some weird code to allow backward compatibility. Rely on distfiles suffixes and a case statement (with matching variable EXTRACT_CASES) to allow distinct extraction rules for multiple distfiles.
2000-09-11 14:06:24 -04:00
EXTRACT_SUFX?= .tar.gz
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
DISTFILES?= ${DISTNAME}${EXTRACT_SUFX}
Finish cleansing PATCH_SITES out.
2000-02-04 06:15:16 -05:00
Defining ALLFILES in every case is much better...
2000-02-04 07:55:33 -05:00
_EVERYTHING=${DISTFILES}
_DISTFILES= ${DISTFILES:C/:[0-9]$//}
ALLFILES= ${_DISTFILES}
Finish cleansing PATCH_SITES out.
2000-02-04 06:15:16 -05:00
.if defined(PATCHFILES)
Defining ALLFILES in every case is much better...
2000-02-04 07:55:33 -05:00
_PATCHFILES=${PATCHFILES:C/:[0-9]$//}
Finish cleansing PATCH_SITES out.
2000-02-04 06:15:16 -05:00
_EVERYTHING+=${PATCHFILES}
Defining ALLFILES in every case is much better...
2000-02-04 07:55:33 -05:00
ALLFILES+= ${_PATCHFILES}
Finish cleansing PATCH_SITES out.
2000-02-04 06:15:16 -05:00
.endif
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
.if make(makesum) || make(addsum) || defined(__FETCH_ALL)
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. if defined(SUPDISTFILES)
Finish cleansing PATCH_SITES out.
2000-02-04 06:15:16 -05:00
_EVERYTHING+= ${SUPDISTFILES}
ALLFILES+= ${SUPDISTFILES:C/:[0-9]$//}
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. endif
Add SUPDISTFILES/SUPPATCHFILES functionality: files that are not always needed for all builds, but that must be fetched for maintainance purposes (makesum, mirror...) Not to use for intl/usa programs, probably...
1999-04-20 14:04:27 -04:00
.endif
Merge our chagnes with FreeBSD's latest changes. We should now have identical bsd.port.mk files again. This file has successfully compiled many of the archivers, gnu make, and emacs (althought the emacs deltas have not been merged back into the FreeBSD tree yet since I don't have it working yet). The FreeBSD Id line may be slightly off since I generated it by hand to match the commit I just made there.
1996-12-25 15:10:09 -05:00
kill tweaks that make no sense: NO_DESCRIBE, NO_PACKAGE, FETCH_BEFORE_ARGS, FETCH_AFTER_ARGS, NO_WARNINGS (is this used ?), VARNAME, FORCE_PACKAGE. Make CKSUMFILES invisible from users. Make sure repackage and reinstall go through normal clean targets. Kill pre-clean, pre-distclean, pre-repackage. Don't allow customization of distclean.
2003-07-14 09:33:04 -04:00
__CKSUMFILES=
Compute CKSUMFILES in two steps, use that to remove duplicates from ALLFILES as well...
2000-02-22 09:07:09 -05:00
# First, remove duplicates
.for _file in ${ALLFILES}
kill tweaks that make no sense: NO_DESCRIBE, NO_PACKAGE, FETCH_BEFORE_ARGS, FETCH_AFTER_ARGS, NO_WARNINGS (is this used ?), VARNAME, FORCE_PACKAGE. Make CKSUMFILES invisible from users. Make sure repackage and reinstall go through normal clean targets. Kill pre-clean, pre-distclean, pre-repackage. Don't allow customization of distclean.
2003-07-14 09:33:04 -04:00
. if empty(__CKSUMFILES:M${_file})
__CKSUMFILES+=${_file}
Compute CKSUMFILES in a better way, to avoid duplicates. Simplifies SUPDISTFILES choice: put everything that might ever be fetched there, don't worry about duplicates.
2000-02-04 15:33:42 -05:00
. endif
.endfor
kill tweaks that make no sense: NO_DESCRIBE, NO_PACKAGE, FETCH_BEFORE_ARGS, FETCH_AFTER_ARGS, NO_WARNINGS (is this used ?), VARNAME, FORCE_PACKAGE. Make CKSUMFILES invisible from users. Make sure repackage and reinstall go through normal clean targets. Kill pre-clean, pre-distclean, pre-repackage. Don't allow customization of distclean.
2003-07-14 09:33:04 -04:00
ALLFILES:=${__CKSUMFILES}
Compute CKSUMFILES in two steps, use that to remove duplicates from ALLFILES as well...
2000-02-22 09:07:09 -05:00
.if defined(IGNOREFILES)
byebye IGNOREFILES Okay naddy@
2004-07-11 16:44:33 -04:00
ERRORS+= "Fatal: don't use IGNOREFILES"
Compute CKSUMFILES in two steps, use that to remove duplicates from ALLFILES as well...
2000-02-22 09:07:09 -05:00
.endif
Merge our chagnes with FreeBSD's latest changes. We should now have identical bsd.port.mk files again. This file has successfully compiled many of the archivers, gnu make, and emacs (althought the emacs deltas have not been merged back into the FreeBSD tree yet since I don't have it working yet). The FreeBSD Id line may be slightly off since I generated it by hand to match the commit I just made there.
1996-12-25 15:10:09 -05:00
# List of all files, with ${DIST_SUBDIR} in front. Used for checksum.
always define DIST_SUBDIR, simplifies logic.
2004-02-07 17:36:13 -05:00
.if !empty(DIST_SUBDIR)
kill tweaks that make no sense: NO_DESCRIBE, NO_PACKAGE, FETCH_BEFORE_ARGS, FETCH_AFTER_ARGS, NO_WARNINGS (is this used ?), VARNAME, FORCE_PACKAGE. Make CKSUMFILES invisible from users. Make sure repackage and reinstall go through normal clean targets. Kill pre-clean, pre-distclean, pre-repackage. Don't allow customization of distclean.
2003-07-14 09:33:04 -04:00
_CKSUMFILES= ${__CKSUMFILES:S/^/${DIST_SUBDIR}\//}
Merge our chagnes with FreeBSD's latest changes. We should now have identical bsd.port.mk files again. This file has successfully compiled many of the archivers, gnu make, and emacs (althought the emacs deltas have not been merged back into the FreeBSD tree yet since I don't have it working yet). The FreeBSD Id line may be slightly off since I generated it by hand to match the commit I just made there.
1996-12-25 15:10:09 -05:00
.else
kill tweaks that make no sense: NO_DESCRIBE, NO_PACKAGE, FETCH_BEFORE_ARGS, FETCH_AFTER_ARGS, NO_WARNINGS (is this used ?), VARNAME, FORCE_PACKAGE. Make CKSUMFILES invisible from users. Make sure repackage and reinstall go through normal clean targets. Kill pre-clean, pre-distclean, pre-repackage. Don't allow customization of distclean.
2003-07-14 09:33:04 -04:00
_CKSUMFILES= ${__CKSUMFILES}
Merge our chagnes with FreeBSD's latest changes. We should now have identical bsd.port.mk files again. This file has successfully compiled many of the archivers, gnu make, and emacs (althought the emacs deltas have not been merged back into the FreeBSD tree yet since I don't have it working yet). The FreeBSD Id line may be slightly off since I generated it by hand to match the commit I just made there.
1996-12-25 15:10:09 -05:00
.endif
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
# This is what is actually going to be extracted, and is overridable
# by user.
Add multiple MASTER_SITES/PATCH_SITES
1999-09-26 06:50:07 -04:00
EXTRACT_ONLY?= ${_DISTFILES}
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
Deprecate old EXTRACT_* variables. Use some weird code to allow backward compatibility. Rely on distfiles suffixes and a case statement (with matching variable EXTRACT_CASES) to allow distinct extraction rules for multiple distfiles.
2000-09-11 14:06:24 -04:00
# okay, time for some guess work
indentation
2000-12-23 07:27:17 -05:00
.if !empty(EXTRACT_ONLY:M*.zip)
drop old compatibility cruft and make USE_ZIP/USE_BZIP2 internal variables
2001-08-18 19:16:57 -04:00
_USE_ZIP?= Yes
indentation
2000-12-23 07:27:17 -05:00
.endif
Add support for PATCH_CASES, modelled on EXTRACT_CASES.
2001-06-03 10:42:07 -04:00
.if !empty(EXTRACT_ONLY:M*.tar.bz2) || (defined(PATCHFILES) && !empty(_PATCHFILES:M*.bz2))
drop old compatibility cruft and make USE_ZIP/USE_BZIP2 internal variables
2001-08-18 19:16:57 -04:00
_USE_BZIP2?= Yes
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
.endif
drop old compatibility cruft and make USE_ZIP/USE_BZIP2 internal variables
2001-08-18 19:16:57 -04:00
_USE_ZIP?= No
_USE_BZIP2?= No
Deprecate old EXTRACT_* variables. Use some weird code to allow backward compatibility. Rely on distfiles suffixes and a case statement (with matching variable EXTRACT_CASES) to allow distinct extraction rules for multiple distfiles.
2000-09-11 14:06:24 -04:00
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
EXTRACT_CASES?=
Deprecate old EXTRACT_* variables. Use some weird code to allow backward compatibility. Rely on distfiles suffixes and a case statement (with matching variable EXTRACT_CASES) to allow distinct extraction rules for multiple distfiles.
2000-09-11 14:06:24 -04:00
'perl -p' does not seem to behave as I expected; use -n instead. Fixes a problem with the perl shar filter noticed by pval@
2002-05-19 14:51:21 -04:00
_PERL_FIX_SHAR?=perl -ne 'print if $$s || ($$s = m:^\#(\!\s*/bin/sh\s*| This is a shell archive):)'
Fix EXTRACT_CASES for shell archives (handles leading cruft, too!) ok espie@
2001-07-30 10:45:26 -04:00
Deprecate old EXTRACT_* variables. Use some weird code to allow backward compatibility. Rely on distfiles suffixes and a case statement (with matching variable EXTRACT_CASES) to allow distinct extraction rules for multiple distfiles.
2000-09-11 14:06:24 -04:00
# XXX note that we DON'T set EXTRACT_SUFX.
drop old compatibility cruft and make USE_ZIP/USE_BZIP2 internal variables
2001-08-18 19:16:57 -04:00
.if ${_USE_ZIP:L} != "no"
don't depend on the current version of unzip/bzip2. Any should do.
2001-10-07 06:47:01 -04:00
BUILD_DEPENDS+= :unzip-*:archivers/unzip
pass -o to unzip so it doesn't prompt when overwriting a file ok espie@
2006-02-06 12:09:08 -05:00
EXTRACT_CASES+= *.zip) ${UNZIP} -oq ${FULLDISTDIR}/$$archive -d ${WRKDIR};;
Remove compatibility scaffolding for EXTRACT_CMD, EXTRACT_BEFORE_ARGS, EXTRACT_AFTER_ARGS
2000-12-16 10:44:35 -05:00
.endif
drop old compatibility cruft and make USE_ZIP/USE_BZIP2 internal variables
2001-08-18 19:16:57 -04:00
.if ${_USE_BZIP2:L} != "no"
don't depend on the current version of unzip/bzip2. Any should do.
2001-10-07 06:47:01 -04:00
BUILD_DEPENDS+= :bzip2-*:archivers/bzip2
Deprecate old EXTRACT_* variables. Use some weird code to allow backward compatibility. Rely on distfiles suffixes and a case statement (with matching variable EXTRACT_CASES) to allow distinct extraction rules for multiple distfiles.
2000-09-11 14:06:24 -04:00
EXTRACT_CASES+= *.tar.bz2) ${BZIP2} -dc ${FULLDISTDIR}/$$archive | ${TAR} xf -;;
Remove compatibility scaffolding for EXTRACT_CMD, EXTRACT_BEFORE_ARGS, EXTRACT_AFTER_ARGS
2000-12-16 10:44:35 -05:00
.endif
Deprecate old EXTRACT_* variables. Use some weird code to allow backward compatibility. Rely on distfiles suffixes and a case statement (with matching variable EXTRACT_CASES) to allow distinct extraction rules for multiple distfiles.
2000-09-11 14:06:24 -04:00
EXTRACT_CASES+= *.tar) ${TAR} xf ${FULLDISTDIR}/$$archive;;
Replace _SED_FIX_SHAR with _PERL_FIX_SHAR which can deal with shar archives that don't start with "#!/bin/sh"; espie@ OK
2002-05-08 14:31:49 -04:00
EXTRACT_CASES+= *.shar.gz|*.shar.Z|*.sh.gz|*.sh.Z) ${GZIP_CMD} -dc ${FULLDISTDIR}/$$archive | ${_PERL_FIX_SHAR} | /bin/sh;;
EXTRACT_CASES+= *.shar | *.sh) ${_PERL_FIX_SHAR} ${FULLDISTDIR}/$$archive | /bin/sh;;
take care of *.tar.gz before extracting any *.gz
2001-06-22 03:21:24 -04:00
EXTRACT_CASES+= *.tar.gz) ${GZIP_CMD} -dc ${FULLDISTDIR}/$$archive | ${TAR} xf -;;
EXTRACT_CASES+= *.gz) ${GZIP_CMD} -dc ${FULLDISTDIR}/$$archive >`basename $$archive .gz`;;
Deprecate old EXTRACT_* variables. Use some weird code to allow backward compatibility. Rely on distfiles suffixes and a case statement (with matching variable EXTRACT_CASES) to allow distinct extraction rules for multiple distfiles.
2000-09-11 14:06:24 -04:00
EXTRACT_CASES+= *) ${GZIP_CMD} -dc ${FULLDISTDIR}/$$archive | ${TAR} xf -;;
Add support for PATCH_CASES, modelled on EXTRACT_CASES.
2001-06-03 10:42:07 -04:00
PATCH_CASES?=
drop old compatibility cruft and make USE_ZIP/USE_BZIP2 internal variables
2001-08-18 19:16:57 -04:00
.if ${_USE_BZIP2:L} != "no"
Add support for PATCH_CASES, modelled on EXTRACT_CASES.
2001-06-03 10:42:07 -04:00
PATCH_CASES+= *.bz2) ${BZIP2} -dc $$patchfile | ${PATCH} ${PATCH_DIST_ARGS};;
.endif
PATCH_CASES+= *.Z|*.gz) ${GZCAT} $$patchfile | ${PATCH} ${PATCH_DIST_ARGS};;
PATCH_CASES+= *) ${PATCH} ${PATCH_DIST_ARGS} < $$patchfile;;
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
# Documentation
More verbose default maintainer
2001-10-24 07:43:06 -04:00
MAINTAINER?= The OpenBSD ports mailing-list <ports@openbsd.org>
Merge our chagnes with FreeBSD's latest changes. We should now have identical bsd.port.mk files again. This file has successfully compiled many of the archivers, gnu make, and emacs (althought the emacs deltas have not been merged back into the FreeBSD tree yet since I don't have it working yet). The FreeBSD Id line may be slightly off since I generated it by hand to match the commit I just made there.
1996-12-25 15:10:09 -05:00
.if !defined(CATEGORIES)
Replace .BEGIN lines with explicit ERRORS framework.
2001-07-17 12:11:45 -04:00
ERRORS+= "Fatal: CATEGORIES is mandatory."
Merge our chagnes with FreeBSD's latest changes. We should now have identical bsd.port.mk files again. This file has successfully compiled many of the archivers, gnu make, and emacs (althought the emacs deltas have not been merged back into the FreeBSD tree yet since I don't have it working yet). The FreeBSD Id line may be slightly off since I generated it by hand to match the commit I just made there.
1996-12-25 15:10:09 -05:00
.endif
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
Add standard method for perl MakeMaker configuration
2000-06-02 07:42:53 -04:00
Remove . from the path passed to configure; adjust the configure script path appropriate for SEPARATE_BUILD; this seems to work for all ports, some ports had probs with earlier changes
1999-03-02 23:16:03 -05:00
CONFIGURE_SCRIPT?= configure
Handles the case where CONFIGURE_SCRIPT is an absolute path gracefully (to deal with a common shared Cygnus configure) Move MODGNU_configure definition to gnu.port.mk. Should have done that on creation. Typo in brad's hack (to be superseded shortly anyways)
2001-09-16 10:56:42 -04:00
.if ${CONFIGURE_SCRIPT:M/*}
_CONFIGURE_SCRIPT=${CONFIGURE_SCRIPT}
Remove . from the path passed to configure; adjust the configure script path appropriate for SEPARATE_BUILD; this seems to work for all ports, some ports had probs with earlier changes
1999-03-02 23:16:03 -05:00
.else
Do not check if SEPARATE_BUILD is defined but check what is its value (defaults to No) help & blessing from espie@, ok naddy@
2005-04-20 21:44:50 -04:00
. if ${SEPARATE_BUILD:L} != "no"
Handles the case where CONFIGURE_SCRIPT is an absolute path gracefully (to deal with a common shared Cygnus configure) Move MODGNU_configure definition to gnu.port.mk. Should have done that on creation. Typo in brad's hack (to be superseded shortly anyways)
2001-09-16 10:56:42 -04:00
_CONFIGURE_SCRIPT=${WRKSRC}/${CONFIGURE_SCRIPT}
. else
Remove . from the path passed to configure; adjust the configure script path appropriate for SEPARATE_BUILD; this seems to work for all ports, some ports had probs with earlier changes
1999-03-02 23:16:03 -05:00
_CONFIGURE_SCRIPT=./${CONFIGURE_SCRIPT}
Handles the case where CONFIGURE_SCRIPT is an absolute path gracefully (to deal with a common shared Cygnus configure) Move MODGNU_configure definition to gnu.port.mk. Should have done that on creation. Typo in brad's hack (to be superseded shortly anyways)
2001-09-16 10:56:42 -04:00
. endif
Remove . from the path passed to configure; adjust the configure script path appropriate for SEPARATE_BUILD; this seems to work for all ports, some ports had probs with earlier changes
1999-03-02 23:16:03 -05:00
.endif
Handles the case where CONFIGURE_SCRIPT is an absolute path gracefully (to deal with a common shared Cygnus configure) Move MODGNU_configure definition to gnu.port.mk. Should have done that on creation. Typo in brad's hack (to be superseded shortly anyways)
2001-09-16 10:56:42 -04:00
Remove . from the path passed to configure; adjust the configure script path appropriate for SEPARATE_BUILD; this seems to work for all ports, some ports had probs with earlier changes
1999-03-02 23:16:03 -05:00
CONFIGURE_ENV+= PATH=${PORTPATH}
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
Always define NO_SHARED_LIBS.
2004-08-02 09:01:52 -04:00
.if ${NO_SHARED_LIBS:L} == "yes"
Note that NO_SHARED_LIBS can not be tested until after bsd.port.mk is included; define CONFIGURE_SHARED that is set to either --enable-shared or --disable-shared depending upon the architecture.
1999-03-01 14:44:18 -05:00
CONFIGURE_SHARED?= --disable-shared
.else
CONFIGURE_SHARED?= --enable-shared
.endif
FETCH_MANUALLY okay naddy
2003-07-25 08:46:26 -04:00
FETCH_MANUALLY?=No
.if ${FETCH_MANUALLY:L} != "no"
_ALLFILES_PRESENT=Yes
Fix a typo that caused ports marked with FETCH_MANUALLY always to be skipped for BATCH=Yes. Reported by Kurt Miller <truk@optonline.net>.
2003-08-14 11:29:20 -04:00
. for _F in ${ALLFILES:S@^@${FULLDISTDIR}/@}
FETCH_MANUALLY okay naddy
2003-07-25 08:46:26 -04:00
. if !exists(${_F})
_ALLFILES_PRESENT=No
. endif
. endfor
. if ${_ALLFILES_PRESENT:L} == "no"
IS_INTERACTIVE=Yes
. endif
.endif
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
################################################################
# Many ways to disable a port.
#
# If we're in BATCH mode and the port is interactive, or we're
# in interactive mode and the port is non-interactive, skip all
# the important targets. The reason we have two modes is that
# one might want to leave a build in BATCH mode running
# overnight, then come back in the morning and do _only_ the
# interactive ones that required your intervention.
#
# Ignore ports that can't be resold if building for a CDROM.
#
# Don't build a port if it's broken.
Add ftp.openbsd.org as a MASTER_SITE_BACKUP before ftp.freebsd.org; If MASTER_SITE_OPENBSD is defined only ftp.openbsd.org will be used; Re-do the way COMES_WITH was implemented to allow overide on the command line by defining NO_IGNORE.
1998-04-04 23:20:38 -05:00
#
# Don't build a port if it comes with the base system.
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
################################################################
avoid ignoring ports while running a `describe' target. triggered by setting DESCRIBE_TARGET=Yes in bsd.port.subdir.mk
2006-11-13 08:55:10 -05:00
.if !defined(NO_IGNORE) && !defined(DESCRIBE_TARGET)
REGRESS_IS_INTERACTIVE This makes things a bit less symetric, but it can't really be avoided since regress is out of the normal build chain.
2001-10-07 07:30:30 -04:00
. if (defined(REGRESS_IS_INTERACTIVE) && defined(BATCH))
_IGNORE_REGRESS= "has interactive tests"
fix REGRESS_IS_INTERACTIVE logic; ok espie@
2002-02-28 19:49:03 -05:00
. elif (!defined(REGRESS_IS_INTERACTIVE) && defined(INTERACTIVE))
REGRESS_IS_INTERACTIVE This makes things a bit less symetric, but it can't really be avoided since regress is out of the normal build chain.
2001-10-07 07:30:30 -04:00
_IGNORE_REGRESS= "does not have interactive tests"
. endif
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. if (defined(IS_INTERACTIVE) && defined(BATCH))
Merge our chagnes with FreeBSD's latest changes. We should now have identical bsd.port.mk files again. This file has successfully compiled many of the archivers, gnu make, and emacs (althought the emacs deltas have not been merged back into the FreeBSD tree yet since I don't have it working yet). The FreeBSD Id line may be slightly off since I generated it by hand to match the commit I just made there.
1996-12-25 15:10:09 -05:00
IGNORE= "is an interactive port"
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. elif (!defined(IS_INTERACTIVE) && defined(INTERACTIVE))
Merge our chagnes with FreeBSD's latest changes. We should now have identical bsd.port.mk files again. This file has successfully compiled many of the archivers, gnu make, and emacs (althought the emacs deltas have not been merged back into the FreeBSD tree yet since I don't have it working yet). The FreeBSD Id line may be slightly off since I generated it by hand to match the commit I just made there.
1996-12-25 15:10:09 -05:00
IGNORE= "is not an interactive port"
USE_LIBTOOL, USE_GMAKE, USE_X11 should really be Yes/No variables.
2002-04-09 09:52:58 -04:00
. elif ${USE_X11:L} == "yes" && !exists(${X11BASE})
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
IGNORE= "uses X11, but ${X11BASE} not found"
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. elif defined(ONLY_FOR_ARCHS)
. for __ARCH in ${MACHINE_ARCH} ${ARCH}
. if !empty(ONLY_FOR_ARCHS:M${__ARCH})
Kill a few extraneous for loops.
1999-12-03 12:37:33 -05:00
_ARCH_OK=1
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. endif
. endfor
. if !defined(_ARCH_OK)
. if ${MACHINE_ARCH} == "${ARCH}"
sync with NetBSD where possible. Fix WRKOBJDIR. Add deinstall target. Make uninstall an alias for deinstall.
1998-09-07 18:33:19 -04:00
IGNORE= "is only for ${ONLY_FOR_ARCHS}, not ${MACHINE_ARCH}"
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. else
Test ${ONLY_FOR_ARCHS} against both ${ARCH} and ${MACHINE_ARCH}.
1998-11-17 01:39:25 -05:00
IGNORE= "is only for ${ONLY_FOR_ARCHS}, not ${MACHINE_ARCH} \(${ARCH}\)"
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. endif
. endif
Add NOT_FOR_ARCHS
2001-03-28 05:25:38 -05:00
. elif defined(NOT_FOR_ARCHS)
. for __ARCH in ${MACHINE_ARCH} ${ARCH}
. if !empty(NOT_FOR_ARCHS:M${__ARCH})
IGNORE= "is not for ${NOT_FOR_ARCHS}"
. endif
. endfor
Introduce SHARED_ONLY variable, as a more specific semantic way to say port works only with shared libraries. Mostly useful for `update-plist' which will recurse differently if LIB_DEPENDS are only BUILD_DEPENDS or if they becomes RUN_DEPENDS. okay pvalchev@
2004-08-03 04:04:02 -04:00
. elif ${SHARED_ONLY:L} == "yes" && ${NO_SHARED_LIBS:L} == "yes"
IGNORE="requires shared libraries"
Fix bug in IGNORE logic.
2000-04-09 07:34:01 -04:00
. endif
avoid ignoring ports while running a `describe' target. triggered by setting DESCRIBE_TARGET=Yes in bsd.port.subdir.mk
2006-11-13 08:55:10 -05:00
.endif # NO_IGNORE
.if !defined(NO_IGNORE)
. if defined(BROKEN)
IGNORE= "is marked as broken: ${BROKEN}"
. elif defined(COMES_WITH)
Change PKGNAME and FULLPKGNAME semantics slightly. Since those are mostly infrastructure-side, making SUBPACKAGE transparent only adds to the confusion, so don't interpolate SUBPACKAGE into FULLPKGNAME, hiding it into _FULLPKGNAME, but rather let FULLPKGNAME tag the main package, and use FULLPKGNAME-foo for subpackage -foo. This will simplify some other stuff...
2001-08-12 07:33:10 -04:00
IGNORE= "-- ${FULLPKGNAME${SUBPACKAGE}:C/-[0-9].*//g} comes with ${OPSYS} as of release ${COMES_WITH}"
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. endif
avoid ignoring ports while running a `describe' target. triggered by setting DESCRIBE_TARGET=Yes in bsd.port.subdir.mk
2006-11-13 08:55:10 -05:00
.endif
Changes tests of IGNORE/NO_IGNORE so that `main' targets (checksum, fetch, extract...) are always defined. This removes the possibility of erroneously overriding them. Dependency fix: - move the NO_xxx test to the _xxx_COOKIE target, so that each cookie is built at exactly one point in the Makefile. - separate phony targets from the real thing: `top' dummy targets (e.g., build) depend on the corresponding cookie, cookies depend on each others and each cookie does trigger the real targett (unless NO_xxx) before creating the cookie. This does repair parallel makes, which were completely broken. Might induce problems into odd-balls ports, but I couldn't get anyone to comment on this patch after a week on ports@ and tech@. Maybe, if something breaks, I'll finally get some comments...
1999-12-19 18:48:36 -05:00
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
.if !defined(DEPENDS_TARGET)
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. if make(reinstall)
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
DEPENDS_TARGET= reinstall
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. else
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
DEPENDS_TARGET= install
make handles indented tests nicely. So, we indent each test/loop level two spaces to let the structure of bsd.port.mk be more visible. Large indentations show that something blatantly unobvious/obfuscated is going on. Design decision made after consulting with brad@. We agree it's much clearer that way.
2000-01-13 12:40:20 -05:00
. endif
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
.endif
Move dependency checking above targets, so that it can be changed to generate cookies for each dependency.
2001-03-28 06:48:12 -05:00
################################################################
# Dependency checking
################################################################
Some comments
2001-04-02 07:45:11 -04:00
# Various dependency styles
Move dependency checking above targets, so that it can be changed to generate cookies for each dependency.
2001-03-28 06:48:12 -05:00
Always define NO_SHARED_LIBS.
2004-08-02 09:01:52 -04:00
.if ${NO_SHARED_LIBS:L} == "yes"
fix typo. sigh. this code is not used that much.
2002-09-11 15:35:21 -04:00
_noshared=-noshared
TRUST_PACKAGES: use actual installed packages to check dependencies and avoid calling pkg_add. Solve some of art's problems.
2001-04-20 12:25:24 -04:00
.else
Use perl lib depends checker, that parses the full gamut of libspecs.
2001-11-17 05:39:19 -05:00
_noshared=
Replace the lib dependency check with a more mechanical check, that no longer relies on ldconfig, but rather on the path stored inside LIB_DEPENDS. Introduce a lib_resolve_fragment that can be used to resolve the libraries from a given list (e.g., listlibs). To be used in the next patch. Note that this check is currently more or less equivalent to the old check, apart from the fact that: - libs can't trust packages, - lib-depends always resort to non-shared libraries, even on shared architectures, if the shared check doesn't work.
2001-11-12 09:24:06 -05:00
.endif
_libresolve_fragment = \
case "$$d" in \
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
*/*) shdir="${LOCALBASE}/$${d%/*}";; \
*) shdir="${LOCALBASE}/lib";; \
Replace the lib dependency check with a more mechanical check, that no longer relies on ldconfig, but rather on the path stored inside LIB_DEPENDS. Introduce a lib_resolve_fragment that can be used to resolve the libraries from a given list (e.g., listlibs). To be used in the next patch. Note that this check is currently more or less equivalent to the old check, apart from the fact that: - libs can't trust packages, - lib-depends always resort to non-shared libraries, even on shared architectures, if the shared check doesn't work.
2001-11-12 09:24:06 -05:00
esac; \
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
check=`eval $$listlibs 2>/dev/null| LOCALBASE=${LOCALBASE} X11BASE=${X11BASE} perl \
Use perl lib depends checker, that parses the full gamut of libspecs.
2001-11-17 05:39:19 -05:00
${PORTSDIR}/infrastructure/build/resolve-lib ${_noshared} $$d` \
* move the dependency computation out of PKG_ARGS, so that we can check for errors (and get rid of _NODEPS since it's not needed any more) * set check to Failed in *libresolve_fragment, so that we can test for it and report properly. guys here have no objection...
2005-11-02 15:28:11 -05:00
|| check=Failed
Replace the lib dependency check with a more mechanical check, that no longer relies on ldconfig, but rather on the path stored inside LIB_DEPENDS. Introduce a lib_resolve_fragment that can be used to resolve the libraries from a given list (e.g., listlibs). To be used in the next patch. Note that this check is currently more or less equivalent to the old check, apart from the fact that: - libs can't trust packages, - lib-depends always resort to non-shared libraries, even on shared architectures, if the shared check doesn't work.
2001-11-12 09:24:06 -05:00
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
_syslibresolve_fragment = \
case "$$d" in \
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
/*) shdir="$${d%/*}/";; \
*/*) shdir="${DEPBASE}/$${d%/*}";; \
*) shdir="${DEPBASE}/lib"; listlibs="$$listlibs /usr/lib/lib* ${X11BASE}/lib/lib*";; \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
esac; \
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
check=`eval $$listlibs 2>/dev/null| LOCALBASE=${LOCALBASE} X11BASE=${X11BASE} perl \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
${PORTSDIR}/infrastructure/build/resolve-lib ${_noshared} $$d` \
* move the dependency computation out of PKG_ARGS, so that we can check for errors (and get rid of _NODEPS since it's not needed any more) * set check to Failed in *libresolve_fragment, so that we can test for it and report properly. guys here have no objection...
2005-11-02 15:28:11 -05:00
|| check=Failed
Replace the lib dependency check with a more mechanical check, that no longer relies on ldconfig, but rather on the path stored inside LIB_DEPENDS. Introduce a lib_resolve_fragment that can be used to resolve the libraries from a given list (e.g., listlibs). To be used in the next patch. Note that this check is currently more or less equivalent to the old check, apart from the fact that: - libs can't trust packages, - lib-depends always resort to non-shared libraries, even on shared architectures, if the shared check doesn't work.
2001-11-12 09:24:06 -05:00
fiddle a bit with LD_LIBRARY_PATH: allow mixed fake/non-fake. and create a _set_ld_library_path internal, so that bzip2 can be installed only as a dependency.
2005-04-23 11:24:49 -04:00
PORT_LD_LIBRARY_PATH=${LOCALBASE}/lib:${X11BASE}/lib:/usr
_set_ld_library_path=:
Add DEPBASE for dependencies. Add USE_FAKE_LIB to turn on FAKE=LIB for ports that have been converted. Tweak buil/lib depends and LD_LIBRARY_PATH in FAKE=lib case.
2005-04-17 06:10:07 -04:00
DEPBASE=${LOCALBASE}
DEPDIR=
allow us to specialize DEPENDS_TARGET depending on whether this is a LIB/RUN/BUILD/REGRESS_DEPENDS.
2004-10-23 06:02:51 -04:00
oops, do not unset SUBPACKAGE Move BULK_COOKIE for package to the redirector, so that we do not clean in subpackage cases, but instead wait for all the packages to be built. Also, introduce FORCE_UPDATE=hard: with signatures, -F installed is most often not necessary...
2006-10-02 13:26:34 -04:00
.if ${FORCE_UPDATE:L} == "yes" || ${FORCE_UPDATE:L} == "hard"
more fixes for new issues: let the update cookie depend on the `full' package cookie, so that an update will get all subpackages built. And a fix for an old issue: let make update proceed to update ALL subpackages, as seems the most natural. Create a subupdate target if one specifically wants to update a single package (and use it as dependency in the FORCE_UPDATE case).
2006-09-27 06:09:34 -04:00
_force_update_fragment=eval $$toset ${MAKE} subupdate
oops, do not unset SUBPACKAGE Move BULK_COOKIE for package to the redirector, so that we do not clean in subpackage cases, but instead wait for all the packages to be built. Also, introduce FORCE_UPDATE=hard: with signatures, -F installed is most often not necessary...
2006-10-02 13:26:34 -04:00
_PKG_ADD_FORCE=-F update -F updatedepends -r
. if ${FORCE_UPDATE:L} == "hard"
_PKG_ADD_FORCE+= -F installed
. endif
industrial-strength updates: - separate UPDATE_COOKIES_DIR, unless set to empty. Makes for easier management. - FORCE_UPDATE variable, that makes certain packages are updated when a dependency is run, and that forces the replacement of packages that are already there. Still uses the update cookie, so that this update occurs just once unless you wipe out yur UPDATE_COOKIES_DIR (which justifies putting update cookies in a separate directory).
2004-11-15 08:52:03 -05:00
.else
_force_update_fragment=:
better naming, discussed with nikolay
2005-11-05 06:16:17 -05:00
_PKG_ADD_FORCE=
industrial-strength updates: - separate UPDATE_COOKIES_DIR, unless set to empty. Makes for easier management. - FORCE_UPDATE variable, that makes certain packages are updated when a dependency is run, and that forces the replacement of packages that are already there. Still uses the update cookie, so that this update occurs just once unless you wipe out yur UPDATE_COOKIES_DIR (which justifies putting update cookies in a separate directory).
2004-11-15 08:52:03 -05:00
.endif
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
_FULL_PACKAGE_NAME?=No
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
_BUILDLIB_DEPENDS= ${LIB_DEPENDS}
_BUILDWANTLIB= ${WANTLIB}
let LIB_DEPENDS and WANTLIB depend on the SUBPACKAGE. Build a special list of _BUILDLIB_DEPENDS, so that separate subpackages can depend on each other without making an infinite loop...
2006-11-20 05:49:22 -05:00
# strip inter-multi-packages dependencies during building
.for _path in ${PKGPATH:S,^mystuff/,,}
. for _s in ${MULTI_PACKAGES}
_BUILDLIB_DEPENDS+= ${LIB_DEPENDS${_s}:N*\:${_path}:N*\:${_path},*}
_BUILDWANTLIB+= ${WANTLIB${_s}}
. endfor
.endfor
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
add an extra level of indirection for dependency lists: _xxx_DEPLIST. allows us to get rid of some NO_DEPENDS tests later, also allows for RUN_DEPENDS to become indexed on SUBPACKAGE. LIB_DEPENDS and WANTLIB are going to need more surgery, since they need to be checked during build and packaging...
2006-11-19 07:32:53 -05:00
.if ${NO_DEPENDS:L} == "no"
Disallow "nasm::devel/nasm" pkgspecs for BUILD, RUN and REGRESS_DEPENDS again. All _DEPENDS should be correct now. 'go for it' espie@
2006-11-21 05:32:51 -05:00
_BUILD_DEPLIST= ${BUILD_DEPENDS:S/^://}
_RUN_DEPLIST= ${RUN_DEPENDS${SUBPACKAGE}:S/^://}
_REGRESS_DEPLIST= ${REGRESS_DEPENDS:S/^://}
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
_BUILDLIB_DEPLIST= ${_BUILDLIB_DEPENDS:C/^[^:]*://}
let LIB_DEPENDS and WANTLIB depend on the SUBPACKAGE. Build a special list of _BUILDLIB_DEPENDS, so that separate subpackages can depend on each other without making an infinite loop...
2006-11-20 05:49:22 -05:00
_RUNLIB_DEPLIST= ${LIB_DEPENDS${SUBPACKAGE}:C/^[^:]*://}
add an extra level of indirection for dependency lists: _xxx_DEPLIST. allows us to get rid of some NO_DEPENDS tests later, also allows for RUN_DEPENDS to become indexed on SUBPACKAGE. LIB_DEPENDS and WANTLIB are going to need more surgery, since they need to be checked during build and packaging...
2006-11-19 07:32:53 -05:00
.endif
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
_DEPLIST= ${_BUILD_DEPLIST} ${_RUN_DEPLIST} ${_REGRESS_DEPLIST} ${_BUILDLIB_DEPLIST} ${_RUNLIB_DEPLIST}
add an extra level of indirection for dependency lists: _xxx_DEPLIST. allows us to get rid of some NO_DEPENDS tests later, also allows for RUN_DEPENDS to become indexed on SUBPACKAGE. LIB_DEPENDS and WANTLIB are going to need more surgery, since they need to be checked during build and packaging...
2006-11-19 07:32:53 -05:00
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
.for _DEP in BUILD RUN BUILDLIB RUNLIB REGRESS
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
_DEP${_DEP}_COOKIES=
add an extra level of indirection for dependency lists: _xxx_DEPLIST. allows us to get rid of some NO_DEPENDS tests later, also allows for RUN_DEPENDS to become indexed on SUBPACKAGE. LIB_DEPENDS and WANTLIB are going to need more surgery, since they need to be checked during build and packaging...
2006-11-19 07:32:53 -05:00
. for _i in ${_${_DEP}_DEPLIST}
simplify dep cookies names. Since we no longer use the dep part, let it vanish completely. Also, stop converting . to -, as it's useless and makes for less readable names.
2006-11-19 12:48:14 -05:00
_DEP${_DEP}_COOKIES+=${WRKDIR}/.dep${_i:C,[|:/<=>*],-,g}
add an extra level of indirection for dependency lists: _xxx_DEPLIST. allows us to get rid of some NO_DEPENDS tests later, also allows for RUN_DEPENDS to become indexed on SUBPACKAGE. LIB_DEPENDS and WANTLIB are going to need more surgery, since they need to be checked during build and packaging...
2006-11-19 07:32:53 -05:00
. endfor
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
.endfor
# Normal user-mode targets are PHONY targets, e.g., don't create the
# corresponding file. However, there is nothing phony about the cookie.
MODSIMPLE_configure= \
WRKCONF?=${WRKBUILD}
2004-09-15 14:58:49 -04:00
cd ${WRKCONF} && ${_SYSTRACE_CMD} ${SETENV} \
Strip spaces from CFLAGS everywhere. Pass off -a to pkg_add if used to solve dependencies.
2004-09-18 09:45:23 -04:00
CC="${CC}" ac_cv_path_CC="${CC}" CFLAGS="${CFLAGS:C/ *$//}" \
CXX="${CXX}" ac_cv_path_CXX="${CXX}" CXXFLAGS="${CXXFLAGS:C/ *$//}" \
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
INSTALL="/usr/bin/install -c -o ${BINOWN} -g ${BINGRP}" \
ac_given_INSTALL="/usr/bin/install -c -o ${BINOWN} -g ${BINGRP}" \
INSTALL_PROGRAM="${INSTALL_PROGRAM}" INSTALL_MAN="${INSTALL_MAN}" \
INSTALL_SCRIPT="${INSTALL_SCRIPT}" INSTALL_DATA="${INSTALL_DATA}" \
YACC="${YACC}" \
${CONFIGURE_ENV} ${_CONFIGURE_SCRIPT} ${CONFIGURE_ARGS}
VMEM_WARNING?= No
_FAKE_SETUP=TRUEPREFIX=${PREFIX} PREFIX=${WRKINST}${PREFIX} ${DESTDIRNAME}=${WRKINST}
_CLEANDEPENDS?=Yes
# mirroring utilities
always define DIST_SUBDIR, simplifies logic.
2004-02-07 17:36:13 -05:00
.if !empty(DIST_SUBDIR)
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
_ALLFILES=${ALLFILES:S/^/${DIST_SUBDIR}\//}
.else
_ALLFILES=${ALLFILES}
.endif
.if defined(MULTI_PACKAGES)
. for _S in ${MULTI_PACKAGES}
_FMN+= ${PKGPATH}/${FULLPKGNAME${_S}}
. endfor
.endif
# Internal variables, used by dependencies targets
# Only keep pkg:dir spec
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
_BUILD_DEP2= ${BUILD_DEPENDS:C/^[^:]*:([^:]*:[^:]*).*$/\1/}
_BUILD_DEP3= ${BUILD_DEPENDS:C/^[^:]*:([^:]*:[^:]*).*$/\1/}
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
_RUN_DEP2= ${RUN_DEPENDS${SUBPACKAGE}:C/^[^:]*:([^:]*:[^:]*).*$/\1/}
_RUN_DEP3= ${RUN_DEPENDS${SUBPACKAGE}:C/^[^:]*:([^:]*:[^:]*).*$/\1/}
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
_REGRESS_DEP2= ${REGRESS_DEPENDS:C/^[^:]*:([^:]*:[^:]*).*$/\1/}
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
.if ${NO_SHARED_LIBS:L} != "yes"
let LIB_DEPENDS and WANTLIB depend on the SUBPACKAGE. Build a special list of _BUILDLIB_DEPENDS, so that separate subpackages can depend on each other without making an infinite loop...
2006-11-20 05:49:22 -05:00
_RUN_DEP2+= ${LIB_DEPENDS${SUBPACKAGE}:C/^[^:]*:([^:]*:[^:]*).*$/\1/}
_LIB_DEP3= ${LIB_DEPENDS${SUBPACKAGE}:C/^[^:]*:([^:]*:[^:]*).*$/\1/}
_DEPRUNLIBS= ${LIB_DEPENDS${SUBPACKAGE}:C/:.*//:S/,/ /g}
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
.else
fix NO_SHARED_LIBS case, found out by bernd@
2006-11-20 06:15:39 -05:00
_LIB_DEP3=
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
_DEPRUNLIBS=
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
.endif
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
_BUILD_DEP2+= ${_BUILDLIB_DEPENDS:C/^[^:]*:([^:]*:[^:]*).*$/\1/}
_DEPBUILDLIBS= ${_BUILDLIB_DEPENDS:C/:.*//:S/,/ /g}
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
_DEPBUILDLIBS+= ${_BUILDWANTLIB}
let LIB_DEPENDS and WANTLIB depend on the SUBPACKAGE. Build a special list of _BUILDLIB_DEPENDS, so that separate subpackages can depend on each other without making an infinite loop...
2006-11-20 05:49:22 -05:00
_DEPRUNLIBS+= ${WANTLIB${SUBPACKAGE}}
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
.if ${NO_DEPENDS:L} == "no"
. for i in ${_DEPBUILDLIBS:C,[|:/<=>*],-,g}
_DEPBUILDLIBSPECS_COOKIES+=${WRKDIR}/.spec-$i
better semantics for library checking: just check the pkgspec for recursive building, and then do one big check of all specs in LIB_DEPENDS and wantlib. This should be faster and more accurate.
2005-10-09 09:31:50 -04:00
. endfor
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
_DEPBUILDWANTLIB_COOKIE=${WRKDIR}/.buildwantlibs
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
. for i in ${_DEPRUNLIBS:C,[|:/<=>*],-,g}
_DEPRUNLIBSPECS_COOKIES+=${WRKDIR}/.spec-$i
. endfor
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
_DEPRUNWANTLIB_COOKIE=${WRKDIR}/.runwantlibs${SUBPACKAGE}
better semantics for library checking: just check the pkgspec for recursive building, and then do one big check of all specs in LIB_DEPENDS and wantlib. This should be faster and more accurate.
2005-10-09 09:31:50 -04:00
.else
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
_DEPBUILDWANTLIB_COOKIE=
_DEPRUNWANTLIB_COOKIE=
better semantics for library checking: just check the pkgspec for recursive building, and then do one big check of all specs in LIB_DEPENDS and wantlib. This should be faster and more accurate.
2005-10-09 09:31:50 -04:00
.endif
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
_DEPLIBSPECS_COOKIES=${_DEPBUILDLIBSPECS_COOKIES} ${_DEPRUNLIBSPECS_COOKIES}
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
_BUILD_DEP= ${_BUILD_DEP2:C/[^:]*://}
_RUN_DEP= ${_RUN_DEP2:C/[^:]*://}
_REGRESS_DEP= ${_REGRESS_DEP2:C/[^:]*://}
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
README_NAME?= ${TEMPLATES}/README.port
move the automake inter-dependencies to another file. checked by naddy@ on a full build. idea okay'ed by naddy and pvalchev.
2004-06-06 07:49:08 -04:00
REORDER_DEPENDENCIES?=
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
# Lock infrastructure:
simplify locks
2005-09-18 08:20:00 -04:00
# nothing happens unless LOCKDIR is defined to a non-empty value
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
simplify locks
2005-09-18 08:20:00 -04:00
LOCKDIR?=
LOCK_CMD?= perl ${PORTSDIR}/infrastructure/build/dolock
UNLOCK_CMD?=rm -f
more generic protection against multiple locking: instead of one _MASTER_LOCK, we keep a list of _LOCKS_HELD by one port builder in an env variable, and we don't relock stuff that's already locked. This allows us to, e.g., have fake depend on regress without issue (even though, internally, it's more efficient to use the _internal-* targets to avoid testing locks).
2006-09-17 15:08:59 -04:00
_LOCKS_HELD?=
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
LOCK_VERBOSE?=No
simplify locks
2005-09-18 08:20:00 -04:00
.if !empty(LOCKDIR)
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
. if ${LOCK_VERBOSE:L} == "yes"
_LOCK=echo "Locking $$lock from $@"; ${LOCK_CMD} ${LOCKDIR}/$$lock.lock
_UNLOCK=echo "Unlocking $$lock from $@"; ${UNLOCK_CMD} ${LOCKDIR}/$$lock.lock
. else
_LOCK=${LOCK_CMD} ${LOCKDIR}/$$lock.lock
_UNLOCK=${UNLOCK_CMD} ${LOCKDIR}/$$lock.lock
. endif
Do not check if SEPARATE_BUILD is defined but check what is its value (defaults to No) help & blessing from espie@, ok naddy@
2005-04-20 21:44:50 -04:00
. if ${SEPARATE_BUILD:L:Mflavored}
fix postgresql lock issue: in recursive locking, don't relock the same port. okay pvalchev@
2004-08-11 18:25:23 -04:00
_LOCKNAME=${PKGNAME}
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
. else
fix postgresql lock issue: in recursive locking, don't relock the same port. okay pvalchev@
2004-08-11 18:25:23 -04:00
_LOCKNAME=${FULLPKGNAME}
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
. endif
more generic protection against multiple locking: instead of one _MASTER_LOCK, we keep a list of _LOCKS_HELD by one port builder in an env variable, and we don't relock stuff that's already locked. This allows us to, e.g., have fake depend on regress without issue (even though, internally, it's more efficient to use the _internal-* targets to avoid testing locks).
2006-09-17 15:08:59 -04:00
. for _i in ${_LOCKNAME}
. if empty(_LOCKS_HELD:M${_i})
fix postgresql lock issue: in recursive locking, don't relock the same port. okay pvalchev@
2004-08-11 18:25:23 -04:00
_DO_LOCK=\
more generic protection against multiple locking: instead of one _MASTER_LOCK, we keep a list of _LOCKS_HELD by one port builder in an env variable, and we don't relock stuff that's already locked. This allows us to, e.g., have fake depend on regress without issue (even though, internally, it's more efficient to use the _internal-* targets to avoid testing locks).
2006-09-17 15:08:59 -04:00
lock=${_LOCKNAME}; \
_LOCKS_HELD="${_LOCKS_HELD} ${_LOCKNAME}"; export _LOCKS_HELD; \
${_SIMPLE_LOCK}
. endif
. endfor
_SIMPLE_LOCK= \
${_LOCK}; trap '${_UNLOCK}' 0 1 2 3 13 15
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
.endif
more generic protection against multiple locking: instead of one _MASTER_LOCK, we keep a list of _LOCKS_HELD by one port builder in an env variable, and we don't relock stuff that's already locked. This allows us to, e.g., have fake depend on regress without issue (even though, internally, it's more efficient to use the _internal-* targets to avoid testing locks).
2006-09-17 15:08:59 -04:00
_SIMPLE_LOCK?=:
_DO_LOCK?=:
let fetch check the size of the retrieved file, if this is recorded in distinfo. This allows ftp to skip files which return 404 error files and other such nonsense... (note that if you want to really grab a file with the wrong size, you can just say CHECKSUM_FILE=/dev/null) with input from by xsa, sturm, ian..
2005-01-04 15:44:45 -05:00
_size_fragment=wc -c $$file 2>/dev/null| awk '{print "SIZE (" $$2 ") = " $$1}'
introduce new macros to make things more readable. Kill really old stuff: LD_CONFIG, that's no longer used.
2005-11-05 06:11:17 -05:00
# commands used all the time
_lines2list= tr '\012' '\040' | sed -e 's, $$,,'
_zap_last_line= sed -e '$$d'
_sort_dependencies=tsort -r|${_zap_last_line}
_version2default= sed -e 's,-[0-9].*,-*,'
_grab_libs_from_plist= sed -n -e '/^@lib /{ s///; p; }' \
-e '/^@file .*\/lib\/lib.*\.a$$/{ s/^@file //; p; }'
move comment accordingly
2005-11-27 07:31:18 -05:00
###
### end of variable setup. Only targets now
###
fix FETCH_PACKAGES: create a real target (packages/arch/cache/pkgname.tgz) and link to it from elsewhere. Fixes the no-arch case, and makes it ways easier to find out what was retrieved through the cache.
2006-10-02 05:37:14 -04:00
${_CACHE_REPO}/${_PKGFILE}:
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
@mkdir -p ${@D}
fix FETCH_PACKAGES: create a real target (packages/arch/cache/pkgname.tgz) and link to it from elsewhere. Fixes the no-arch case, and makes it ways easier to find out what was retrieved through the cache.
2006-10-02 05:37:14 -04:00
@${ECHO_MSG} -n "===> Looking for ${_PKGFILE} in \$$PKG_PATH - "
put no-arch into a variable, so that people can override it.
2006-10-18 06:51:39 -04:00
@if ${SETENV} PKG_CACHE=${_CACHE_REPO} PKG_PATH=${_CACHE_REPO}:${_PKG_REPO}:${PACKAGE_REPOSITORY}/${NO_ARCH}/:${PKG_PATH} PKG_TMPDIR=${PKG_TMPDIR} pkg_add -n -q ${_PKG_ADD_FORCE} ${_PKGFILE} >/dev/null 2>&1; then \
fix FETCH_PACKAGES: create a real target (packages/arch/cache/pkgname.tgz) and link to it from elsewhere. Fixes the no-arch case, and makes it ways easier to find out what was retrieved through the cache.
2006-10-02 05:37:14 -04:00
${ECHO_MSG} "found"; \
exit 0; \
else \
${ECHO_MSG} "not found"; \
exit 1; \
fi
try fetching missing packages from PKG_PATH, activated by setting FETCH_PACKAGES set PKG_PATH consistently when calling pkg_add idea, help and ok espie
2005-11-04 04:34:50 -05:00
${_PACKAGE_COOKIE}:
protect against further bugs, refuse to build the _PACKAGE_COOKIE if we end up there with PACKAGING not set. We do not try to `fix it' ourselves, because PACKAGING not set may have some non obvious issues, like PKG_ARCH or PERMIT_PACKAGE_* not set correctly at all.
2006-10-16 06:39:40 -04:00
.if !defined(PACKAGING)
@echo "Error: can't make $@ without PACKAGING set"; exit 1
.endif
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
@mkdir -p ${@D}
fix FETCH_PACKAGES: create a real target (packages/arch/cache/pkgname.tgz) and link to it from elsewhere. Fixes the no-arch case, and makes it ways easier to find out what was retrieved through the cache.
2006-10-02 05:37:14 -04:00
.if ${FETCH_PACKAGES:L} == "yes" && !defined(_TRIED_FETCHING)
@f=${_CACHE_REPO}/${_PKGFILE}; \
cd ${.CURDIR} && ${MAKE} $$f && \
{ ln $$f $@ 2>/dev/null || cp -p $$f $@ ; } || \
cd ${.CURDIR} && ${MAKE} _TRIED_FETCHING=Yes $@
.else
oops, do not unset SUBPACKAGE Move BULK_COOKIE for package to the redirector, so that we do not clean in subpackage cases, but instead wait for all the packages to be built. Also, introduce FORCE_UPDATE=hard: with signatures, -F installed is most often not necessary...
2006-10-02 13:26:34 -04:00
@cd ${.CURDIR} && unset PACKAGING || true && exec ${MAKE} ${_PACKAGE_COOKIE_DEPS} ${_PKG_PREREQ}
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. if target(pre-package)
@cd ${.CURDIR} && exec ${MAKE} pre-package
try fetching missing packages from PKG_PATH, activated by setting FETCH_PACKAGES set PKG_PATH consistently when calling pkg_add idea, help and ok espie
2005-11-04 04:34:50 -05:00
. endif
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. if target(do-package)
@cd ${.CURDIR} && exec ${MAKE} do-package
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
. else
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
# What PACKAGE normally does:
@${ECHO_MSG} "===> Building package for ${FULLPKGNAME${SUBPACKAGE}}"
@${ECHO_MSG} "Create ${_PACKAGE_COOKIE}"
@cd ${.CURDIR} && \
deps=`${MAKE} _print-package-args` && \
if ${SUDO} ${PKG_CMD} `echo "$$deps"|sort -u` ${PKG_ARGS} ${_PACKAGE_COOKIE}; then \
mode=`id -u`:`id -g`; ${SUDO} ${CHOWN} $${mode} ${_PACKAGE_COOKIE}; \
if ${_register_plist}; then \
exit 0; \
fi; \
fi && \
${SUDO} ${MAKE} _internal-clean=package && \
exit 1
# End of PACKAGE.
try fetching missing packages from PKG_PATH, activated by setting FETCH_PACKAGES set PKG_PATH consistently when calling pkg_add idea, help and ok espie
2005-11-04 04:34:50 -05:00
. endif
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. if target(post-package)
@cd ${.CURDIR} && exec ${MAKE} post-package
. endif
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
@rm -f ${_BULK_COOKIE} ${_UPDATE_COOKIE}
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
.endif
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
.PRECIOUS: ${_PACKAGE_COOKIES} ${_INSTALL_COOKIE}
missing dependency
2004-06-22 12:09:05 -04:00
${_SYSTRACE_COOKIE}: ${_WRKDIR_COOKIE}
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
@rm -f $@
.for _i in ${_SYSTRACE_POLICIES}
@echo "Policy: ${_i}, Emulation: native" >> $@
allow for port specific additional systrace filter rules in ${.CURDIR}/systrace.filter ok espie@
2004-01-18 02:52:49 -05:00
@if [ -f ${.CURDIR}/systrace.filter ]; then \
substitute variables in a port's systrace.filter file as well
2006-06-16 14:48:58 -04:00
sed ${_SYSTRACE_SED_SUBST} ${.CURDIR}/systrace.filter >> $@; \
allow for port specific additional systrace filter rules in ${.CURDIR}/systrace.filter ok espie@
2004-01-18 02:52:49 -05:00
fi
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
@sed ${_SYSTRACE_SED_SUBST} ${SYSTRACE_FILTER} >> $@
.endfor
allow for additional port specific systrace policies, needed to make jdk-linux and netscape build with systrace ok espie and others
2004-05-01 10:27:07 -04:00
@if [ -f ${.CURDIR}/systrace.policy ]; then \
sed ${_SYSTRACE_SED_SUBST} ${.CURDIR}/systrace.policy >> $@; \
fi
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
# create the packing stuff from source
${WRKPKG}/COMMENT${SUBPACKAGE}:
@echo ${_COMMENT} >$@
let PLIST, DESCR, MESSAGE, UNMESSAGE be subpackage-dependent. careful: they do not get defined to a default value unless the `main' variable is defined, because they get set later otherwise... remove the .if defined(MULTI_PACKAGES) guards as they don't serve any purpose. Simplify the PLIST tests to make them more uniform. Move SED_PLIST to the list of obsolete variables instead of giving it special treatment.
2006-11-23 19:10:00 -05:00
${WRKPKG}/DESCR${SUBPACKAGE}: ${DESCR${SUBPACKAGE}}
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
@${_SED_SUBST} <$? >$@.tmp && mv -f $@.tmp $@
@echo "\nMaintainer: ${MAINTAINER}" >>$@
.if defined(HOMEPAGE)
@fgrep -q '$${HOMEPAGE}' $? || echo "\nWWW: ${HOMEPAGE}" >>$@
.endif
makesum: fetch-all
.if !defined(NO_CHECKSUM)
@rm -f ${CHECKSUM_FILE}
@cd ${DISTDIR} && \
for cipher in ${_CIPHERS}; do \
$$cipher ${_CKSUMFILES} >> ${CHECKSUM_FILE}; \
done
let fetch check the size of the retrieved file, if this is recorded in distinfo. This allows ftp to skip files which return 404 error files and other such nonsense... (note that if you want to really grab a file with the wrong size, you can just say CHECKSUM_FILE=/dev/null) with input from by xsa, sturm, ian..
2005-01-04 15:44:45 -05:00
@cd ${DISTDIR} && \
for file in ${_CKSUMFILES}; do \
${_size_fragment} >> ${CHECKSUM_FILE}; \
done
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
@sort -u -o ${CHECKSUM_FILE} ${CHECKSUM_FILE}
.endif
addsum: fetch-all
.if !defined(NO_CHECKSUM)
@touch ${CHECKSUM_FILE}
@cd ${DISTDIR} && \
for cipher in ${_CIPHERS}; do \
$$cipher ${_CKSUMFILES} >> ${CHECKSUM_FILE}; \
done
let fetch check the size of the retrieved file, if this is recorded in distinfo. This allows ftp to skip files which return 404 error files and other such nonsense... (note that if you want to really grab a file with the wrong size, you can just say CHECKSUM_FILE=/dev/null) with input from by xsa, sturm, ian..
2005-01-04 15:44:45 -05:00
@cd ${DISTDIR} && \
for file in ${_CKSUMFILES}; do \
${_size_fragment} >> ${CHECKSUM_FILE}; \
done
reorder the bottom part of bsd.port.mk so that all variable tweaks are done before any target is evaluated. This makes sure a whole series of bugs can't happen. Tests and comments by naddy@ and sturm@
2004-01-10 19:49:01 -05:00
@sort -u -o ${CHECKSUM_FILE} ${CHECKSUM_FILE}
@if [ `sed -e 's/\=.*$$//' ${CHECKSUM_FILE} | uniq -d | wc -l` -ne 0 ]; then \
echo "Inconsistent checksum in ${CHECKSUM_FILE}"; \
exit 1; \
else \
${ECHO_MSG} "${CHECKSUM_FILE} updated okay, don't forget to remove cruft"; \
fi
.endif
################################################################
# Dependency checking
################################################################
Move dependency checking above targets, so that it can be changed to generate cookies for each dependency.
2001-03-28 06:48:12 -05:00
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
_internal-depends: _internal-lib-depends _internal-build-depends \
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
_internal-run-depends _internal-buildwantlib-depends \
_internal-runwantlib-depends _internal-regress-depends
Move dependency checking above targets, so that it can be changed to generate cookies for each dependency.
2001-03-28 06:48:12 -05:00
Some comments
2001-04-02 07:45:11 -04:00
# and the rules for the actual dependencies
Change PKGNAME and FULLPKGNAME semantics slightly. Since those are mostly infrastructure-side, making SUBPACKAGE transparent only adds to the confusion, so don't interpolate SUBPACKAGE into FULLPKGNAME, hiding it into _FULLPKGNAME, but rather let FULLPKGNAME tag the main package, and use FULLPKGNAME-foo for subpackage -foo. This will simplify some other stuff...
2001-08-12 07:33:10 -04:00
_print-packagename:
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
.if ${_FULL_PACKAGE_NAME:L} == "yes"
@echo '${PKGPATH}/${FULLPKGNAME${SUBPACKAGE}}'
.else
Change PKGNAME and FULLPKGNAME semantics slightly. Since those are mostly infrastructure-side, making SUBPACKAGE transparent only adds to the confusion, so don't interpolate SUBPACKAGE into FULLPKGNAME, hiding it into _FULLPKGNAME, but rather let FULLPKGNAME tag the main package, and use FULLPKGNAME-foo for subpackage -foo. This will simplify some other stuff...
2001-08-12 07:33:10 -04:00
@echo ${FULLPKGNAME${SUBPACKAGE}}
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
.endif
Change PKGNAME and FULLPKGNAME semantics slightly. Since those are mostly infrastructure-side, making SUBPACKAGE transparent only adds to the confusion, so don't interpolate SUBPACKAGE into FULLPKGNAME, hiding it into _FULLPKGNAME, but rather let FULLPKGNAME tag the main package, and use FULLPKGNAME-foo for subpackage -foo. This will simplify some other stuff...
2001-08-12 07:33:10 -04:00
Now that all dependencies are evaluated in the same way, build the cookies in one single loop, based on _DEPLIST. Extract dependencies off the loop.
2006-11-19 13:07:38 -05:00
.for _i in ${_DEPLIST}
. if !target(${WRKDIR}/.dep${_i:C,[|:/<=>*],-,g})
simplify dep cookies names. Since we no longer use the dep part, let it vanish completely. Also, stop converting . to -, as it's useless and makes for less readable names.
2006-11-19 12:48:14 -05:00
${WRKDIR}/.dep${_i:C,[|:/<=>*],-,g}: ${_WRKDIR_COOKIE}
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
@unset PACKAGING DEPENDS_TARGET _MASTER WRKDIR|| true; \
Proper quoting in build depends too. Discovered by avsm@. Slightly more intricate, because we can't quote each value indiscriminately, simplest way is to move the for loop from shell to make.
2001-04-12 16:35:59 -04:00
echo '${_i}'|{ \
simplify dep cookies names. Since we no longer use the dep part, let it vanish completely. Also, stop converting . to -, as it's useless and makes for less readable names.
2006-11-19 12:48:14 -05:00
IFS=:; read pkg subdir target; \
Now that all dependencies are evaluated in the same way, build the cookies in one single loop, based on _DEPLIST. Extract dependencies off the loop.
2006-11-19 13:07:38 -05:00
extra_msg="(DEPENDS ${_i})"; \
rephrase dependency evaluation without $$dep. Also, more error checking, check specifically for which dependency targets are allowed.
2006-11-19 13:20:01 -05:00
${_flavor_fragment}; defaulted=false; checkinstall=true; \
remove confusing FAKE=lib scaffolding that's not finished, and that won't work in that way in any case...
2006-11-19 13:01:44 -05:00
case "X$$target" in X) target=${DEPENDS_TARGET};; esac; \
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
case "X$$target" in \
Xinstall|Xreinstall) early_exit=false;; \
let fake not touch the WRKDIR, since it's an `install' type target.
2006-08-07 04:57:18 -04:00
Xpackage|Xfake) early_exit=true;; \
rephrase dependency evaluation without $$dep. Also, more error checking, check specifically for which dependency targets are allowed.
2006-11-19 13:20:01 -05:00
Xpatch|Xconfigure) \
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
early_exit=true; mkdir -p ${WRKDIR}/$$dir; \
set _MASTER_LOCK in the correct location. fix postgresql locking *for good*. okay pvalchev@
2004-08-12 15:02:45 -04:00
toset="$$toset _MASTER='[${FULLPKGNAME${SUBPACKAGE}}]${_MASTER}' WRKDIR=${WRKDIR}/$$dir"; \
rephrase dependency evaluation without $$dep. Also, more error checking, check specifically for which dependency targets are allowed.
2006-11-19 13:20:01 -05:00
checkinstall=false;; \
Xextract) \
${ECHO_MSG} "===> Error: bad dependency ${_i}"; \
exit 1;; \
*) \
${ECHO_MSG} "===> Error: don't know how to depend on $$target"; \
exit 1;; \
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
esac; \
Strip spaces from CFLAGS everywhere. Pass off -a to pkg_add if used to solve dependencies.
2004-09-18 09:45:23 -04:00
toset="$$toset _SOLVING_DEP=Yes"; \
fed up of merge errors in dependency makefiles, output a decent error message if cd dir && make print-pkgname does not work.
2006-06-04 10:56:10 -04:00
case "X$$pkg" in X) \
if pkg=`eval $$toset ${MAKE} _print-packagename`; \
then \
defaulted=true; \
also defaults to stem-* for build depends, thus simplifying code a bit...
2006-11-19 12:40:35 -05:00
pkg=`echo $$pkg|${_version2default}`; \
fed up of merge errors in dependency makefiles, output a decent error message if cd dir && make print-pkgname does not work.
2006-06-04 10:56:10 -04:00
else \
${ECHO_MSG} "===> Error in evaluating dependency ${_i}"; \
exit 1; \
fi;; esac; \
Proper quoting in build depends too. Discovered by avsm@. Slightly more intricate, because we can't quote each value indiscriminately, simplest way is to move the for loop from shell to make.
2001-04-12 16:35:59 -04:00
for abort in false false true; do \
if $$abort; then \
${ECHO_MSG} "Dependency check failed"; \
exit 1; \
fi; \
found=false; \
Enforce /nonexistent for non-package/install/reinstall dep.
2001-09-30 07:27:05 -04:00
what=$$pkg; \
rephrase dependency evaluation without $$dep. Also, more error checking, check specifically for which dependency targets are allowed.
2006-11-19 13:20:01 -05:00
if $$checkinstall; then \
industrial-strength updates: - separate UPDATE_COOKIES_DIR, unless set to empty. Makes for easier management. - FORCE_UPDATE variable, that makes certain packages are updated when a dependency is run, and that forces the replacement of packages that are already there. Still uses the update cookie, so that this update occurs just once unless you wipe out yur UPDATE_COOKIES_DIR (which justifies putting update cookies in a separate directory).
2004-11-15 08:52:03 -05:00
$$early_exit || ${_force_update_fragment}; \
also defaults to stem-* for build depends, thus simplifying code a bit...
2006-11-19 12:40:35 -05:00
if pkg_info -q -e "$$pkg"; then \
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
${ECHO_MSG} "===> ${FULLPKGNAME${SUBPACKAGE}}${_MASTER} depends on: $$what - found"; \
Move dependency checking above targets, so that it can be changed to generate cookies for each dependency.
2001-03-28 06:48:12 -05:00
break; \
Proper quoting in build depends too. Discovered by avsm@. Slightly more intricate, because we can't quote each value indiscriminately, simplest way is to move the for loop from shell to make.
2001-04-12 16:35:59 -04:00
else \
LIB_DEPENDS tweaks. - allow for several libraries, separated by commas. - try to heed dependency most closely. Notably, make sure the spec stops at a sensible place. This may break some ill-formed dependencies.
2001-09-05 05:10:34 -04:00
: $${msg:= not found}; \
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
${ECHO_MSG} "===> ${FULLPKGNAME${SUBPACKAGE}}${_MASTER} depends on: $$what -$$msg"; \
rephrase dependency evaluation without $$dep. Also, more error checking, check specifically for which dependency targets are allowed.
2006-11-19 13:20:01 -05:00
fi; \
fi; \
show what is actually checked (the pkg, no longer the file)
2001-09-19 11:16:39 -04:00
${ECHO_MSG} "===> Verifying $$target for $$what in $$dir"; \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
if eval $$toset ${MAKE} $$target; then \
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
${ECHO_MSG} "===> Returning to build of ${FULLPKGNAME${SUBPACKAGE}}${_MASTER}"; \
Proper quoting in build depends too. Discovered by avsm@. Slightly more intricate, because we can't quote each value indiscriminately, simplest way is to move the for loop from shell to make.
2001-04-12 16:35:59 -04:00
else \
exit 1; \
fi; \
Give dependencies a fighting change, spell early_exit correctly.
2001-10-10 20:01:27 -04:00
if $$early_exit; then \
Proper quoting in build depends too. Discovered by avsm@. Slightly more intricate, because we can't quote each value indiscriminately, simplest way is to move the for loop from shell to make.
2001-04-12 16:35:59 -04:00
break; \
fi; \
done; \
}
Side-effect of BULK=Yes: WRKDIR gets wiped out before the depend cookie is written. So, we simply recreate WRKDIR manually. This is a kluge, but it should do the trick. From espie@ ok pvalchev@ espie@
2004-08-13 19:28:40 -04:00
@mkdir -p ${WRKDIR} ${WRKDIR}/bin
Use cookies for dependencies as well, and for the creation of a workdir. This can speed up some iterations, as a dependency won't be checked more than once. This also allows for make extract NO_DEPENDS=Yes, then later to have the dependencies pulled in at configure stage (even though this might not work all the time, as some dependencies may be needed earlier but then you'll certainly notice).
2001-09-05 05:13:18 -04:00
@${_MAKE_COOKIE} $@
Now that all dependencies are evaluated in the same way, build the cookies in one single loop, based on _DEPLIST. Extract dependencies off the loop.
2006-11-19 13:07:38 -05:00
. endif
Move dependency checking above targets, so that it can be changed to generate cookies for each dependency.
2001-03-28 06:48:12 -05:00
.endfor
Now that all dependencies are evaluated in the same way, build the cookies in one single loop, based on _DEPLIST. Extract dependencies off the loop.
2006-11-19 13:07:38 -05:00
_internal-build-depends: ${_DEPBUILD_COOKIES}
_internal-run-depends: ${_DEPRUN_COOKIES}
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
_internal-lib-depends: ${_DEPBUILDLIB_COOKIES}
Now that all dependencies are evaluated in the same way, build the cookies in one single loop, based on _DEPLIST. Extract dependencies off the loop.
2006-11-19 13:07:38 -05:00
_internal-regress-depends: ${_DEPREGRESS_COOKIES}
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
_internal-buildlib-depends: ${_DEPBUILDLIB_COOKIES}
_internal-runlib-depends: ${_DEPRUNLIB_COOKIES}
Move dependency checking above targets, so that it can be changed to generate cookies for each dependency.
2001-03-28 06:48:12 -05:00
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
. if !empty(_DEPLIBSPECS_COOKIES)
${_DEPLIBSPECS_COOKIES}: ${_WRKDIR_COOKIE}
need an extra indirection: the complete wantlib cookie can easily grow over 256 characters, so we create an extra file for each wantlib, and depend on it. That way the full _DEPlibs_COOKIE is regenerated (and retested) each time WANTLIB changes.
2005-10-09 20:29:38 -04:00
@${_MAKE_COOKIE} $@
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
.endif
need an extra indirection: the complete wantlib cookie can easily grow over 256 characters, so we create an extra file for each wantlib, and depend on it. That way the full _DEPlibs_COOKIE is regenerated (and retested) each time WANTLIB changes.
2005-10-09 20:29:38 -04:00
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
.for _m in BUILD RUN
. if !empty(_DEP${_m}WANTLIB_COOKIE)
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
${_DEP${_m}WANTLIB_COOKIE}: ${_DEP${_m}LIBSPECS_COOKIES} ${_DEP${_m}LIB_COOKIES} ${_DEPBUILD_COOKIES} ${_WRKDIR_COOKIE}
@${ECHO_MSG} "===> Verifying specs: ${_DEP${_m}LIBS}"
better semantics for library checking: just check the pkgspec for recursive building, and then do one big check of all specs in LIB_DEPENDS and wantlib. This should be faster and more accurate.
2005-10-09 09:31:50 -04:00
@listlibs="echo ${LOCALBASE}/lib/lib* /usr/lib/lib* ${X11BASE}/lib/lib*"; \
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
for d in ${_DEP${_m}LIBS:S/>/\>/g}; do \
better semantics for library checking: just check the pkgspec for recursive building, and then do one big check of all specs in LIB_DEPENDS and wantlib. This should be faster and more accurate.
2005-10-09 09:31:50 -04:00
case "$$d" in \
/*) listlibs="$$listlibs $${d%/*}/lib*";; \
*/*) listlibs="$$listlibs ${DEPBASE}/$${d%/*}/lib*";; \
esac; \
done; \
if found=`eval $$listlibs 2>/dev/null| \
LOCALBASE=${LOCALBASE} X11BASE=${X11BASE} perl \
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
${PORTSDIR}/infrastructure/build/resolve-lib ${_noshared} ${_DEP${_m}LIBS:S/>/\>/g}`; then \
output cosmetics ok espie@
2005-10-10 12:07:33 -04:00
line="===> found"; \
better semantics for library checking: just check the pkgspec for recursive building, and then do one big check of all specs in LIB_DEPENDS and wantlib. This should be faster and more accurate.
2005-10-09 09:31:50 -04:00
for k in $$found; do line="$$line $$k"; done; \
${ECHO_MSG} "$$line"; \
else \
echo 1>&2 "Fatal error"; \
exit 1; \
fi
@${_MAKE_COOKIE} $@
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
. endif
better semantics for library checking: just check the pkgspec for recursive building, and then do one big check of all specs in LIB_DEPENDS and wantlib. This should be faster and more accurate.
2005-10-09 09:31:50 -04:00
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
_internal-${_m:L}wantlib-depends: ${_DEP${_m}WANTLIB_COOKIE}
.endfor
better semantics for library checking: just check the pkgspec for recursive building, and then do one big check of all specs in LIB_DEPENDS and wantlib. This should be faster and more accurate.
2005-10-09 09:31:50 -04:00
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
.if defined(IGNORE) && !defined(NO_IGNORE)
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
_internal-fetch _internal-checksum _internal-extract _internal-patch \
_internal-configure _internal-all _internal-build _internal-install \
_internal-regress _internal-uninstall _internal-deinstall _internal-fake \
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
_internal-update _internal-plist _internal-package _internal-install-all \
more fixes for new issues: let the update cookie depend on the `full' package cookie, so that an update will get all subpackages built. And a fix for an old issue: let make update proceed to update ALL subpackages, as seems the most natural. Create a subupdate target if one specifically wants to update a single package (and use it as dependency in the FORCE_UPDATE case).
2006-09-27 06:09:34 -04:00
_internal-update-plist update-patches _internal-subupdate \
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
dump-vars describe _internal-subpackage sublib-depends-check \
_internal-manpages-check:
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
. if !defined(IGNORE_SILENT)
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
@${ECHO_MSG} "===> ${FULLPKGNAME${SUBPACKAGE}}${_MASTER} ${IGNORE}."
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
. endif
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
.else
flag for a.out
2005-01-31 05:12:24 -05:00
. if ${ELF_TOOLCHAIN:L} == "no"
_NEWLIB_DEPENDS_FLAGS=-o
. else
_NEWLIB_DEPENDS_FLAGS=
. endif
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
sublib-depends-check:
@cd ${.CURDIR} && exec ${MAKE} subpackage
flag for a.out
2005-01-31 05:12:24 -05:00
@perl ${PORTSDIR}/infrastructure/package/check-newlib-depends \
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
${_NEWLIB_DEPENDS_FLAGS} -d ${_PKG_REPO} ${_PACKAGE_COOKIE}
First cut at a lib-depends checker: this version does use WRKINST instead of the actual package, so it WILL get things wrong in some multi-packages case.
2001-10-24 07:57:34 -04:00
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
_internal-manpages-check: ${_FAKE_COOKIE}
Add a target to check that makewhatis will build correct index entries.
2002-04-02 10:55:55 -05:00
@cd ${WRKINST}${TRUEPREFIX}/man && \
${SUDO} /usr/libexec/makewhatis -p . && \
cat whatis.db
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
# Most standard port targets create a cookie to avoid being re-run.
#
# fetch is an exception, as it uses the files it fetches as `cookies',
# and it's run by checksum, so in essence it's a sub-goal of extract,
# in normal use.
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
#
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
# Besides, fetch can't create cookies, as it does not have WRKDIR available
# in the first place.
#
# IMPORTANT: pre-fetch/do-fetch/post-fetch MUST be designed so that they
# can be run several times in a row.
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
_internal-fetch:
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
# See ports/infrastructure/templates/Makefile.template
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
@${ECHO_MSG} "===> Checking files for ${FULLPKGNAME}${_MASTER}"
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
. if target(pre-fetch)
@cd ${.CURDIR} && exec ${MAKE} pre-fetch
. endif
. if target(do-fetch)
@cd ${.CURDIR} && exec ${MAKE} do-fetch
. else
# What FETCH normally does:
. if !empty(ALLFILES)
@cd ${.CURDIR} && exec ${MAKE} ${ALLFILES:S@^@${FULLDISTDIR}/@}
. endif
# End of FETCH
. endif
. if target(post-fetch)
@cd ${.CURDIR} && exec ${MAKE} post-fetch
. endif
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
_internal-checksum: _internal-fetch
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
. if ! defined(NO_CHECKSUM)
transform files/md5 into distinfo magically. I would credit FreeBSD or NetBSD for the idea if I could remember which one came up with it first. :)
2001-11-11 07:59:50 -05:00
@checksum_file=${CHECKSUM_FILE}; \
don't throw an error in fetch, when no distinfo file exists but error out in _internal-checksum, when there should be one ok espie
2006-10-18 12:53:04 -04:00
if [ -z "${DISTFILES}" ]; then \
${ECHO_MSG} ">> No distfiles."; \
elif [ ! -f $$checksum_file ]; then \
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
${ECHO_MSG} ">> No checksum file."; \
don't throw an error in fetch, when no distinfo file exists but error out in _internal-checksum, when there should be one ok espie
2006-10-18 12:53:04 -04:00
exit 1; \
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
else \
cd ${DISTDIR}; OK=true; list=''; \
for file in ${_CKSUMFILES}; do \
for cipher in ${PREFERRED_CIPHERS}; do \
transform files/md5 into distinfo magically. I would credit FreeBSD or NetBSD for the idea if I could remember which one came up with it first. :)
2001-11-11 07:59:50 -05:00
set -- `grep -i "^$$cipher ($$file)" $$checksum_file` && break || \
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
${ECHO_MSG} ">> No $$cipher checksum recorded for $$file."; \
done; \
case "$$4" in \
"") \
${ECHO_MSG} ">> No checksum recorded for $$file."; \
OK=false;; \
"IGNORE") \
byebye IGNOREFILES Okay naddy@
2004-07-11 16:44:33 -04:00
echo ">> Error: checksum for $$file is set to IGNORE in distinfo"; \
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
OK=false;; \
*) \
CKSUM=`$$cipher < $$file`; \
case "$$CKSUM" in \
"$$4") \
${ECHO_MSG} ">> Checksum OK for $$file. ($$cipher)";; \
*) \
echo ">> Checksum mismatch for $$file. ($$cipher)"; \
list="$$list $$file $$cipher $$4"; \
OK=false;; \
esac;; \
esac; \
done; \
set --; \
if ! $$OK; then \
if ${REFETCH}; then \
pull a few more targets out of user-visible space. Sprinkle a few more cd ${.CURDIR} && exec move distclean into convenience targets area.
2003-07-30 15:59:48 -04:00
cd ${.CURDIR} && ${MAKE} _refetch _PROBLEMS="$$list"; \
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
else \
transform files/md5 into distinfo magically. I would credit FreeBSD or NetBSD for the idea if I could remember which one came up with it first. :)
2001-11-11 07:59:50 -05:00
echo "Make sure the Makefile and checksum file ($$checksum_file)"; \
Stop advising users to ignore wrong checksums, instead suggesting from the distfiles/ directories on the OpenBSD FTP servers where a good copy is normally carried. From Peter Hessler, ok espie lebel
2003-03-02 12:54:27 -05:00
echo "are up to date. If you want to fetch a good copy of this"; \
echo "file from the OpenBSD main archive, type"; \
echo "\"make REFETCH=true [other args]\"."; \
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
exit 1; \
fi; \
fi ; \
fi
. endif
pull a few more targets out of user-visible space. Sprinkle a few more cd ${.CURDIR} && exec move distclean into convenience targets area.
2003-07-30 15:59:48 -04:00
_refetch:
PROBLEMS->_PROBLEMS, this shouldn't be user-visible.
2003-07-09 07:16:21 -04:00
. for file cipher value in ${_PROBLEMS}
pull a few more targets out of user-visible space. Sprinkle a few more cd ${.CURDIR} && exec move distclean into convenience targets area.
2003-07-30 15:59:48 -04:00
@rm ${DISTDIR}/${file}
@cd ${.CURDIR} && exec ${MAKE} ${DISTDIR}/${file} \
MASTER_SITE_OVERRIDE="ftp://ftp.openbsd.org/pub/OpenBSD/distfiles/${cipher}/${value}/"
Use new make feature.
2001-08-16 11:13:17 -04:00
. endfor
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
cd ${.CURDIR} && exec ${MAKE} _internal-checksum REFETCH=false
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
Move dependency checking above targets, so that it can be changed to generate cookies for each dependency.
2001-03-28 06:48:12 -05:00
# The cookie's recipe hold the real rule for each of those targets.
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
_internal-extract: ${_EXTRACT_COOKIE}
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
_internal-patch: ${_DEPBUILD_COOKIES} ${_DEPBUILDLIB_COOKIES} ${_DEPBUILDWANTLIB_COOKIE} \
Use cookies for dependencies as well, and for the creation of a workdir. This can speed up some iterations, as a dependency won't be checked more than once. This also allows for make extract NO_DEPENDS=Yes, then later to have the dependencies pulled in at configure stage (even though this might not work all the time, as some dependencies may be needed earlier but then you'll certainly notice).
2001-09-05 05:13:18 -04:00
${_PATCH_COOKIE}
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
_internal-distpatch: ${_DEPBUILD_COOKIES} ${_DEPBUILDLIB_COOKIES} ${_DEPBUILDWANTLIB_COOKIE} \
Use cookies for dependencies as well, and for the creation of a workdir. This can speed up some iterations, as a dependency won't be checked more than once. This also allows for make extract NO_DEPENDS=Yes, then later to have the dependencies pulled in at configure stage (even though this might not work all the time, as some dependencies may be needed earlier but then you'll certainly notice).
2001-09-05 05:13:18 -04:00
${_DISTPATCH_COOKIE}
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
_internal-configure: ${_DEPBUILD_COOKIES} ${_DEPBUILDLIB_COOKIES} ${_DEPBUILDWANTLIB_COOKIE} \
Use cookies for dependencies as well, and for the creation of a workdir. This can speed up some iterations, as a dependency won't be checked more than once. This also allows for make extract NO_DEPENDS=Yes, then later to have the dependencies pulled in at configure stage (even though this might not work all the time, as some dependencies may be needed earlier but then you'll certainly notice).
2001-09-05 05:13:18 -04:00
${_CONFIGURE_COOKIE}
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
_internal-build _internal-all: ${_DEPBUILD_COOKIES} ${_DEPBUILDLIB_COOKIES} \
${_DEPBUILDWANTLIB_COOKIE} ${_BUILD_COOKIE}
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
_internal-install: ${_INSTALL_COOKIE}
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
_internal-fake: ${_FAKE_COOKIE}
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
_internal-subupdate: ${_UPDATE_COOKIE}
kill bulk-packages. Use `make package BULK=Yes' instead. The main advantage is that workdirs are cleaned up as soon as package is built (or installed), not later.
2002-04-17 11:58:48 -04:00
REGRESS_IS_INTERACTIVE This makes things a bit less symetric, but it can't really be avoided since regress is out of the normal build chain.
2001-10-07 07:30:30 -04:00
. if defined(_IGNORE_REGRESS)
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
_internal-regress:
REGRESS_IS_INTERACTIVE This makes things a bit less symetric, but it can't really be avoided since regress is out of the normal build chain.
2001-10-07 07:30:30 -04:00
. if !defined(IGNORE_SILENT)
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
@${ECHO_MSG} "===> ${FULLPKGNAME${SUBPACKAGE}}${_MASTER} ${_IGNORE_REGRESS}."
REGRESS_IS_INTERACTIVE This makes things a bit less symetric, but it can't really be avoided since regress is out of the normal build chain.
2001-10-07 07:30:30 -04:00
. endif
. else
rename some internal variables: _DEPlib_COOKIES -> _DEPLIB_COOKIES. simplify some .if defined(A) . for i in $A ... . endfor .endfor into .for i in $A ... . endfor .endfor since empty loops work just fine.
2006-11-19 07:05:10 -05:00
_internal-regress: ${_BUILD_COOKIE} ${_DEPREGRESS_COOKIES} ${_REGRESS_COOKIE}
REGRESS_IS_INTERACTIVE This makes things a bit less symetric, but it can't really be avoided since regress is out of the normal build chain.
2001-10-07 07:30:30 -04:00
. endif
move targets around so they get handled gracefully in BROKEN case.
2005-03-09 10:40:37 -05:00
# packing list utilities. This generates a packing list from a recently
# installed port. Not perfect, but pretty close. The generated file
# will have to have some tweaks done by hand.
# Note: add @comment PACKAGE(arch=${MACHINE_ARCH}, opsys=${OPSYS}, vers=${OPSYS_VER})
# when port is installed or package created.
#
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.if ${SHARED_ONLY:L} == "yes"
move targets around so they get handled gracefully in BROKEN case.
2005-03-09 10:40:37 -05:00
_do_libs_too=
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.else
move targets around so they get handled gracefully in BROKEN case.
2005-03-09 10:40:37 -05:00
_do_libs_too=NO_SHARED_LIBS=Yes
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.endif
move targets around so they get handled gracefully in BROKEN case.
2005-03-09 10:40:37 -05:00
_extra_prefixes=
.if defined(MULTI_PACKAGES)
. for _s in ${MULTI_PACKAGES}
allows PREFIX and CATEGORIES to be subpackage dependent
2006-11-20 08:50:16 -05:00
_extra_prefixes+=PREFIX${_s}=`cd ${.CURDIR} && SUBPACKAGE=${_s} PACKAGING=${_s} ${MAKE} show=PREFIX${_s}`
move targets around so they get handled gracefully in BROKEN case.
2005-03-09 10:40:37 -05:00
. endfor
.endif
remove the dependency of make-plist on dependent packages: instead, go directly use `make print-plist' to find these lists in the ports tree.
2006-11-09 03:16:22 -05:00
_internal-plist _internal-update-plist: _internal-fake
move targets around so they get handled gracefully in BROKEN case.
2005-03-09 10:40:37 -05:00
@${ECHO_MSG} "===> Updating plist for ${FULLPKGNAME}${_MASTER}"
@mkdir -p ${PKGDIR}
@DESTDIR=${WRKINST} PREFIX=${WRKINST}${PREFIX} \
TRUEPREFIX=${TRUEPREFIX} \
INSTALL_PRE_COOKIE=${_INSTALL_PRE_COOKIE} \
remove the dependency of make-plist on dependent packages: instead, go directly use `make print-plist' to find these lists in the ports tree.
2006-11-09 03:16:22 -05:00
DEPPATHS="`${MAKE} run-dir-depends ${_do_libs_too}|${_sort_dependencies}`" \
MAKE="${MAKE}" \
PORTSDIR=${PORTSDIR} \
repair update-plist
2006-11-24 11:49:37 -05:00
PLIST=${PLIST${SUBPACKAGE}} \
move targets around so they get handled gracefully in BROKEN case.
2005-03-09 10:40:37 -05:00
PFRAG=${PKGDIR}/PFRAG \
FLAVORS='${FLAVORS}' MULTI_PACKAGES='${MULTI_PACKAGES}' \
OKAY_FILES='${_FAKE_COOKIE} ${_INSTALL_PRE_COOKIE}' \
SHARED_ONLY="${SHARED_ONLY}" \
OWNER=`id -u` \
GROUP=`id -g` \
${SUDO} perl ${PORTSDIR}/infrastructure/install/make-plist \
${_extra_prefixes} ${_tmpvars}
update-patches:
@toedit=`WRKDIST=${WRKDIST} PATCHDIR=${PATCHDIR} \
PATCH_LIST='${PATCH_LIST}' DIFF_ARGS='${DIFF_ARGS}' \
DISTORIG=${DISTORIG} PATCHORIG=${PATCHORIG} \
/bin/sh ${PORTSDIR}/infrastructure/build/update-patches`; \
case $$toedit in "");; \
*) read i?'edit patches: '; \
cd ${PATCHDIR} && $${VISUAL:-$${EDITOR:-/usr/bin/vi}} $$toedit;; esac
Repair IGNORE. I forgot to move this around when the rest needed rearranging. Found by naddy@.
2001-03-29 14:01:05 -05:00
.endif # IGNORECMD
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
# Top-level targets redirect to the real _internal-target, along with locking
# if locking exists.
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
.for _t in extract patch distpatch configure build all install fake \
subupdate fetch checksum regress depends lib-depends build-depends \
run-depends regress-depends clean manpages-check \
plist update-plist update package install-all
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. if defined(_LOCK)
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
${_t}:
@${_DO_LOCK}; cd ${.CURDIR} && ${MAKE} _internal-${_t}
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. else
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
${_t}: _internal-${_t}
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. endif
.endfor
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
subpackage:
@${_DO_LOCK}; cd ${.CURDIR} && PACKAGING='${SUBPACKAGE}' ${MAKE} _internal-subpackage
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
# Redirectors for top-level targets involving subpackages
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
.for _t _r in _internal-package-only _internal-subpackage \
describe subdescribe lib-depends-check sublib-depends-check \
dump-vars subdump-vars _internal-install-all _internal-install \
shorten dependency names, so that regress and run can share. add print-plist-all, for better introspection (to be used by pkg_mklocatedb).
2006-11-11 11:10:36 -05:00
print-plist-all print-plist \
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
readmes _readme _internal-update _internal-subupdate
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
${_t}:
. if defined(MULTI_PACKAGES)
. for _s in ${MULTI_PACKAGES}
Top-level redirectors which recurse for multi-packages should display where they are. This does unconfuse pkg_mklocatedb
2006-11-11 11:32:12 -05:00
@${ECHO_MSG} "===> ${PKGPATH}${FLAVOR_EXT:S/-/,/g},${_s}"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
@cd ${.CURDIR} && SUBPACKAGE='${_s}' PACKAGING='${_s}' exec ${MAKE} ${_r}
. endfor
all MULTI_PACKAGES are new, stop testing -main.
2006-11-26 12:36:07 -05:00
. else
@cd ${.CURDIR} && SUBPACKAGE='' PACKAGING='' exec ${MAKE} ${_r}
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. endif
.endfor
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
_internal-package:
.if defined(MULTI_PACKAGES)
. for _s in ${MULTI_PACKAGES}
@cd ${.CURDIR} && SUBPACKAGE='${_s}' PACKAGING='${_s}' exec ${MAKE} _internal-subpackage
. endfor
all MULTI_PACKAGES are new, stop testing -main.
2006-11-26 12:36:07 -05:00
.else
@cd ${.CURDIR} && SUBPACKAGE='' PACKAGING='' exec ${MAKE} _internal-subpackage
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
.endif
.if ${BULK_${PKGPATH}:L} == "yes"
@cd ${.CURDIR} && exec ${MAKE} ${_BULK_COOKIE}
.endif
can not depend directly on PACKAGE_COOKIES. Found out by nikolay, fun to debug...
2006-10-21 08:10:35 -04:00
${_BULK_COOKIE}:
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
@cd ${.CURDIR} && exec ${MAKE} _internal-package-only
kill bulk-packages. Use `make package BULK=Yes' instead. The main advantage is that workdirs are cleaned up as soon as package is built (or installed), not later.
2002-04-17 11:58:48 -04:00
@mkdir -p ${BULK_COOKIES_DIR}
move up pkgpath.mk, so that PKGPATH is available. Use it to define per-PKGPATH user-settings: BULK, WRKOBJDIR, ...
2004-02-07 17:34:02 -05:00
.for _i in ${BULK_TARGETS_${PKGPATH}}
kill bulk-packages. Use `make package BULK=Yes' instead. The main advantage is that workdirs are cleaned up as soon as package is built (or installed), not later.
2002-04-17 11:58:48 -04:00
@${ECHO_MSG} "===> Running ${_i}"
pull a few more targets out of user-visible space. Sprinkle a few more cd ${.CURDIR} && exec move distclean into convenience targets area.
2003-07-30 15:59:48 -04:00
@cd ${.CURDIR} && exec ${MAKE} ${_i} ${BULK_FLAGS}
kill bulk-packages. Use `make package BULK=Yes' instead. The main advantage is that workdirs are cleaned up as soon as package is built (or installed), not later.
2002-04-17 11:58:48 -04:00
.endfor
generic BULK_DO fragment that can do anything, invoked right before clean. can be used to run targets that may fail without impacting further stuff, or to collect stats on anything. For instance BULK_DO= mkdir -p ${PORTSDIR}/config/${FULLPKGPATH}; \ cp -f ${WRKBUILD}/config.{log,status} ${PORTSDIR}/config/${FULLPKGPATH} || true; \ cd ${PACKAGE_REPOSITORY}/${MACHINE_ARCH}/all && perl ${PORTSDIR}/infrastructure/package/check-modes ${PKGNAMES} || true
2006-10-23 10:33:01 -04:00
.if !empty(BULK_DO_${PKGPATH})
@${BULK_DO_${PKGPATH}}
.endif
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
@cd ${.CURDIR} && exec ${SUDO} ${MAKE} _internal-clean
kill bulk-packages. Use `make package BULK=Yes' instead. The main advantage is that workdirs are cleaned up as soon as package is built (or installed), not later.
2002-04-17 11:58:48 -04:00
@${_MAKE_COOKIE} $@
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
# The real targets. Note that some parts always get run, some parts can be
# disabled, and there are hooks to override behavior.
Use cookies for dependencies as well, and for the creation of a workdir. This can speed up some iterations, as a dependency won't be checked more than once. This also allows for make extract NO_DEPENDS=Yes, then later to have the dependencies pulled in at configure stage (even though this might not work all the time, as some dependencies may be needed earlier but then you'll certainly notice).
2001-09-05 05:13:18 -04:00
${_WRKDIR_COOKIE}:
@rm -rf ${WRKDIR}
Add DEPBASE for dependencies. Add USE_FAKE_LIB to turn on FAKE=LIB for ports that have been converted. Tweak buil/lib depends and LD_LIBRARY_PATH in FAKE=lib case.
2005-04-17 06:10:07 -04:00
@mkdir -p ${WRKDIR} ${WRKDIR}/bin ${DEPDIR}
Use cookies for dependencies as well, and for the creation of a workdir. This can speed up some iterations, as a dependency won't be checked more than once. This also allows for make extract NO_DEPENDS=Yes, then later to have the dependencies pulled in at configure stage (even though this might not work all the time, as some dependencies may be needed earlier but then you'll certainly notice).
2001-09-05 05:13:18 -04:00
@${_MAKE_COOKIE} $@
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
${_EXTRACT_COOKIE}: ${_WRKDIR_COOKIE} ${_SYSTRACE_COOKIE}
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
@cd ${.CURDIR} && exec ${MAKE} _internal-checksum _internal-build-depends _internal-buildlib-depends _internal-buildwantlib-depends
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
@${ECHO_MSG} "===> Extracting for ${FULLPKGNAME}${_MASTER}"
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
.if target(pre-extract)
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} pre-extract
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
.endif
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} do-extract
Unconditionally use do-extract and create that target in case it doesn't exist with the old, regular EXTRACT_CASES stuff. This allows for systrace protected extracts. espie@ ok
2003-08-11 14:42:07 -04:00
.if target(post-extract)
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} post-extract
.endif
@${_MAKE_COOKIE} $@
.if !target(do-extract)
do-extract:
Kill _PORT_USE macro. This trims down the number of sub-makes the port system runs quite a bit, and makes some other fun stuff possible. THIS KILLS THE pre-*/post-* SCRIPT STUFF The functionality is unneeded, as it can be done with normal pre-*/post-* targets, but it must be kept in mind when porting Free/NetBSD ports.
2000-01-26 18:11:08 -05:00
# What EXTRACT normally does:
Deprecate old EXTRACT_* variables. Use some weird code to allow backward compatibility. Rely on distfiles suffixes and a case statement (with matching variable EXTRACT_CASES) to allow distinct extraction rules for multiple distfiles.
2000-09-11 14:06:24 -04:00
@PATH=${PORTPATH}; set -e; cd ${WRKDIR}; \
fiddle a bit with LD_LIBRARY_PATH: allow mixed fake/non-fake. and create a _set_ld_library_path internal, so that bzip2 can be installed only as a dependency.
2005-04-23 11:24:49 -04:00
${_set_ld_library_path}; \
Deprecate old EXTRACT_* variables. Use some weird code to allow backward compatibility. Rely on distfiles suffixes and a case statement (with matching variable EXTRACT_CASES) to allow distinct extraction rules for multiple distfiles.
2000-09-11 14:06:24 -04:00
for archive in ${EXTRACT_ONLY}; do \
case $$archive in \
${EXTRACT_CASES} \
esac; \
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
done
Kill _PORT_USE macro. This trims down the number of sub-makes the port system runs quite a bit, and makes some other fun stuff possible. THIS KILLS THE pre-*/post-* SCRIPT STUFF The functionality is unneeded, as it can be done with normal pre-*/post-* targets, but it must be kept in mind when porting Free/NetBSD ports.
2000-01-26 18:11:08 -05:00
# End of EXTRACT
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
.endif
Kill _PORT_USE macro. This trims down the number of sub-makes the port system runs quite a bit, and makes some other fun stuff possible. THIS KILLS THE pre-*/post-* SCRIPT STUFF The functionality is unneeded, as it can be done with normal pre-*/post-* targets, but it must be kept in mind when porting Free/NetBSD ports.
2000-01-26 18:11:08 -05:00
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
# Both distpatch and patch invoke pre-patch, if it's defined.
# Hence it needs special treatment (a specific cookie).
Kill _PORT_USE macro. This trims down the number of sub-makes the port system runs quite a bit, and makes some other fun stuff possible. THIS KILLS THE pre-*/post-* SCRIPT STUFF The functionality is unneeded, as it can be done with normal pre-*/post-* targets, but it must be kept in mind when porting Free/NetBSD ports.
2000-01-26 18:11:08 -05:00
.if target(pre-patch)
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
${_PREPATCH_COOKIE}:
protect patch targets with systrace as well no objection from espie, looks good steven
2006-06-18 06:10:04 -04:00
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} pre-patch
Don't change the semantics of the patch target if BATCH is defined, this is stupid (BATCH is only there to skip IS_INTERACTIVE ports). Make PATCH_CHECK_ONLY a Yes/No switch. Let it be a slightly better no-op, e.g., skip creating the cookies in checkpatch operation.
2000-09-23 08:36:39 -04:00
. if ${PATCH_CHECK_ONLY:L} != "yes"
More regularity, use $@ to regenerate target cookies
2001-03-28 06:32:29 -05:00
@${_MAKE_COOKIE} $@
Don't change the semantics of the patch target if BATCH is defined, this is stupid (BATCH is only there to skip IS_INTERACTIVE ports). Make PATCH_CHECK_ONLY a Yes/No switch. Let it be a slightly better no-op, e.g., skip creating the cookies in checkpatch operation.
2000-09-23 08:36:39 -04:00
. endif
Kill _PORT_USE macro. This trims down the number of sub-makes the port system runs quite a bit, and makes some other fun stuff possible. THIS KILLS THE pre-*/post-* SCRIPT STUFF The functionality is unneeded, as it can be done with normal pre-*/post-* targets, but it must be kept in mind when porting Free/NetBSD ports.
2000-01-26 18:11:08 -05:00
.endif
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
# The real distpatch
${_DISTPATCH_COOKIE}: ${_EXTRACT_COOKIE}
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
.if target(pre-patch)
Use exec to strip a large number of intermediate shells that don't serve any useful purpose except for eating up PCB. Not sure whether this is a ksh bug or not, but this is at least a fine work-around.
2000-05-17 08:14:53 -04:00
@cd ${.CURDIR} && exec ${MAKE} ${_PREPATCH_COOKIE}
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
.endif
protect patch targets with systrace as well no objection from espie, looks good steven
2006-06-18 06:10:04 -04:00
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} do-distpatch
.if target(post-distpatch)
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} post-distpatch
.endif
.if ${PATCH_CHECK_ONLY:L} != "yes"
@${_MAKE_COOKIE} $@
.endif
.if !target(do-distpatch)
do-distpatch:
Add intermediate `distpatch' target. This is done in a tricky way, so as not to disturb the usual sequence of targets. Namely, you still have fetch, extract, patch, configure, build, install, package, but patch does invoke a subtarget distpatch when needed, and you can fetch, extract, distpatch, patch, configure, build, install, package for the same results. Mostly useful for porters: provide an intermediate point to work on patches, and make it easy to override distpatch/OpenBSD patches separately.
2000-01-30 10:19:40 -05:00
# What DISTPATCH normally does
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
. if defined(_PATCHFILES)
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
@${ECHO_MSG} "===> Applying distribution patches for ${FULLPKGNAME}${_MASTER}"
Rename _DISTDIR to FULLDISTDIR, make it a supported variable (so that I can clean up the few uses of underscore variables I've done... which were wrong; underscore variables are plain unsupported interface which may change)
1999-09-22 06:16:58 -04:00
@cd ${FULLDISTDIR}; \
Add support for PATCH_CASES, modelled on EXTRACT_CASES.
2001-06-03 10:42:07 -04:00
for patchfile in ${_PATCHFILES}; do \
thinko
2000-09-21 22:20:30 -04:00
case "${PATCH_DEBUG:L}" in \
Clean PATCH_DEBUG handling
2000-09-16 09:54:56 -04:00
no) ;; \
Add support for PATCH_CASES, modelled on EXTRACT_CASES.
2001-06-03 10:42:07 -04:00
*) ${ECHO_MSG} "===> Applying distribution patch $$patchfile" ;; \
Optimizations: - if cd construct >/dev/null - prefer case to if [ ... ] - remove unneeded subshells
1999-07-28 08:56:15 -04:00
esac; \
Add support for PATCH_CASES, modelled on EXTRACT_CASES.
2001-06-03 10:42:07 -04:00
case $$patchfile in \
${PATCH_CASES} \
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
esac; \
Optimizations: - if cd construct >/dev/null - prefer case to if [ ... ] - remove unneeded subshells
1999-07-28 08:56:15 -04:00
done
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
. endif
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
# End of DISTPATCH.
.endif
Add intermediate `distpatch' target. This is done in a tricky way, so as not to disturb the usual sequence of targets. Namely, you still have fetch, extract, patch, configure, build, install, package, but patch does invoke a subtarget distpatch when needed, and you can fetch, extract, distpatch, patch, configure, build, install, package for the same results. Mostly useful for porters: provide an intermediate point to work on patches, and make it easy to override distpatch/OpenBSD patches separately.
2000-01-30 10:19:40 -05:00
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
# The real patch
${_PATCH_COOKIE}: ${_EXTRACT_COOKIE}
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
@${ECHO_MSG} "===> Patching for ${FULLPKGNAME}${_MASTER}"
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
.if target(pre-patch)
Use exec to strip a large number of intermediate shells that don't serve any useful purpose except for eating up PCB. Not sure whether this is a ksh bug or not, but this is at least a fine work-around.
2000-05-17 08:14:53 -04:00
@cd ${.CURDIR} && exec ${MAKE} ${_PREPATCH_COOKIE}
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
.endif
.if target(do-patch)
protect patch targets with systrace as well no objection from espie, looks good steven
2006-06-18 06:10:04 -04:00
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} do-patch
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
.else
Add intermediate `distpatch' target. This is done in a tricky way, so as not to disturb the usual sequence of targets. Namely, you still have fetch, extract, patch, configure, build, install, package, but patch does invoke a subtarget distpatch when needed, and you can fetch, extract, distpatch, patch, configure, build, install, package for the same results. Mostly useful for porters: provide an intermediate point to work on patches, and make it easy to override distpatch/OpenBSD patches separately.
2000-01-30 10:19:40 -05:00
# What PATCH normally does:
# XXX test for efficiency, don't bother with distpatch if it's not needed
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
. if target(do-distpatch) || target(post-distpatch) || defined(PATCHFILES)
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
@cd ${.CURDIR} && exec ${MAKE} _internal-distpatch
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
. endif
fix PATCH_LIST absolute path handling in case of missing PATCHDIR ok espie@
2001-10-04 18:43:45 -04:00
@if cd ${PATCHDIR} 2>/dev/null || [ x"${PATCH_LIST:M/*}" != x"" ]; then \
Use error=true/false instead of error=0/1 (consistency)
2001-03-28 06:52:20 -05:00
error=false; \
- revert change now that we have patch -C - fix PATCH_LIST so that is works - add USE_AUTOCONF - fix PATH env logic plus some documentation...
1998-11-24 20:08:35 -05:00
for i in ${PATCH_LIST}; do \
new functionalities: - PATCH_LIST for conditional patches - addsum for non destructive md5 checksum updates
1998-11-18 23:20:09 -05:00
case $$i in \
*.orig|*.rej|*~) \
${ECHO_MSG} "===> Ignoring patchfile $$i" ; \
;; \
*) \
if [ -e $$i ]; then \
Clean PATCH_DEBUG handling
2000-09-16 09:54:56 -04:00
case "${PATCH_DEBUG:L}" in \
no) ;; \
*) ${ECHO_MSG} "===> Applying ${OPSYS} patch $$i" ;; \
Another if -> case conversion
1999-07-29 11:41:47 -04:00
esac; \
protect patch targets with systrace as well no objection from espie, looks good steven
2006-06-18 06:10:04 -04:00
${_SYSTRACE_CMD} ${PATCH} ${PATCH_ARGS} < $$i || \
Let do-patch be more useful in case of failure. - says which patch fails, - go on applying other patches before failing globally. This helps for quickly updating ports.
1999-07-27 18:12:26 -04:00
{ echo "***> $$i did not apply cleanly"; \
Use error=true/false instead of error=0/1 (consistency)
2001-03-28 06:52:20 -05:00
error=true; }\
new functionalities: - PATCH_LIST for conditional patches - addsum for non destructive md5 checksum updates
1998-11-18 23:20:09 -05:00
else \
Let PATCH_LIST handle absolute paths. make sure update-patches can handle lists (naddy@).
2001-09-27 06:34:19 -04:00
if [ $$i != "patch-*" ]; then \
echo "===> Can't find patch matching $$i"; \
error=true; \
new functionalities: - PATCH_LIST for conditional patches - addsum for non destructive md5 checksum updates
1998-11-18 23:20:09 -05:00
fi; \
fi; \
;; \
esac; \
Let do-patch be more useful in case of failure. - says which patch fails, - go on applying other patches before failing globally. This helps for quickly updating ports.
1999-07-27 18:12:26 -04:00
done;\
Use error=true/false instead of error=0/1 (consistency)
2001-03-28 06:52:20 -05:00
if $$error; then exit 1; fi; \
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
fi
Kill _PORT_USE macro. This trims down the number of sub-makes the port system runs quite a bit, and makes some other fun stuff possible. THIS KILLS THE pre-*/post-* SCRIPT STUFF The functionality is unneeded, as it can be done with normal pre-*/post-* targets, but it must be kept in mind when porting Free/NetBSD ports.
2000-01-26 18:11:08 -05:00
# End of PATCH.
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
.endif
.if target(post-patch)
protect patch targets with systrace as well no objection from espie, looks good steven
2006-06-18 06:10:04 -04:00
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} post-patch
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
.endif
move gnu dreck (autoconf, automake...) to gnu.port.mk
2002-03-15 20:09:23 -05:00
.for _m in ${MODULES}
. if defined(MOD${_m:U}_post-patch)
@${MOD${_m:U}_post-patch}
Don't change the semantics of the patch target if BATCH is defined, this is stupid (BATCH is only there to skip IS_INTERACTIVE ports). Make PATCH_CHECK_ONLY a Yes/No switch. Let it be a slightly better no-op, e.g., skip creating the cookies in checkpatch operation.
2000-09-23 08:36:39 -04:00
. endif
move gnu dreck (autoconf, automake...) to gnu.port.mk
2002-03-15 20:09:23 -05:00
.endfor
move the automake inter-dependencies to another file. checked by naddy@ on a full build. idea okay'ed by naddy and pvalchev.
2004-06-06 07:49:08 -04:00
.if !empty(REORDER_DEPENDENCIES)
REORDER_DEPENDENCY missing a silent @.
2004-07-18 18:45:14 -04:00
@sed -e '/^#/d' ${REORDER_DEPENDENCIES} | \
move the automake inter-dependencies to another file. checked by naddy@ on a full build. idea okay'ed by naddy and pvalchev.
2004-06-06 07:49:08 -04:00
tsort -r|while read f; do \
cd ${WRKSRC}; \
case $$f in \
/*) \
find . -name $${f#/} -print| while read i; \
do echo "Touching $$i"; touch $$i; done \
;; \
*) \
if test -e $$f ; then \
echo "Touching $$f"; touch $$f; \
fi \
;; \
esac; done
.endif
move gnu dreck (autoconf, automake...) to gnu.port.mk
2002-03-15 20:09:23 -05:00
.if ${PATCH_CHECK_ONLY:L} != "yes"
More regularity, use $@ to regenerate target cookies
2001-03-28 06:32:29 -05:00
@${_MAKE_COOKIE} $@
Don't change the semantics of the patch target if BATCH is defined, this is stupid (BATCH is only there to skip IS_INTERACTIVE ports). Make PATCH_CHECK_ONLY a Yes/No switch. Let it be a slightly better no-op, e.g., skip creating the cookies in checkpatch operation.
2000-09-23 08:36:39 -04:00
.endif
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
# The real configure
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
${_CONFIGURE_COOKIE}: ${_PATCH_COOKIE}
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
@${ECHO_MSG} "===> Configuring for ${FULLPKGNAME}${_MASTER}"
Create all packing information under a subdir of WRKDIR (WRKPKG). WRKPKG should seldom be changed, unless it conflicts with the port's actual naming conventions (default is ${WRKBUILD}/pkg). Substitute stuff in INSTALL/DEINSTALL script as well. The only drawback is that those LOOK like shell variables, but are actually simple sed substitution. Preconise a style such as prefix=${PREFIX} (then use ${prefix}) to avoid confusion.
2000-06-16 19:53:40 -04:00
@mkdir -p ${WRKBUILD} ${WRKPKG}
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
.if target(pre-configure)
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} pre-configure
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
.endif
.if target(do-configure)
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} do-configure
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
.else
Kill _PORT_USE macro. This trims down the number of sub-makes the port system runs quite a bit, and makes some other fun stuff possible. THIS KILLS THE pre-*/post-* SCRIPT STUFF The functionality is unneeded, as it can be done with normal pre-*/post-* targets, but it must be kept in mind when porting Free/NetBSD ports.
2000-01-26 18:11:08 -05:00
# What CONFIGURE normally does
Enforce the order of MOD*_configure fragments to match CONFIGURE_STYLE. Create pseudo gnu/simple fragments to make the situation more regular. Found by aaron@.
2001-08-28 04:09:31 -04:00
. for _c in ${CONFIGURE_STYLE:L}
. if defined(MOD${_c:U}_configure)
@${MOD${_c:U}_configure}
Cut out some parts of bsd.port.mk into separate modules. This is work-in-progress. Supplemental hooks will be added when they become necessary. The goal is to permit separate subsystems to tweak the package building process slightly. This should help e.g., python or qt2, for instance. Documentation will come when it's tested more thoroughly...
2001-08-25 07:23:46 -04:00
. endif
. endfor
Remove NO_EXTRACT/NO_PATCH/NO_CONFIGURE
2000-06-16 19:10:41 -04:00
# End of CONFIGURE.
.endif
.if target(post-configure)
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} post-configure
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
.endif
More regularity, use $@ to regenerate target cookies
2001-03-28 06:32:29 -05:00
@${_MAKE_COOKIE} $@
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
# The real build
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
${_BUILD_COOKIE}: ${_CONFIGURE_COOKIE}
Trim documented parts. Give default values to NO_DEPENDS, NO_BUILD
2001-03-28 06:43:16 -05:00
.if ${NO_BUILD:L} == "no"
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
@${ECHO_MSG} "===> Building for ${FULLPKGNAME}${_MASTER}"
o Instead of copying the "virtual memory exhausted" warning for each individual port, introduce a new variable for bsd.port.mk, VMEM_WARNING. By setting this variable to `Yes', the user is given a warning about the virtual memory requirements just before the pre-build stage. o Also give some examples for different shells for what to do about it, and mention login.conf(5). bash, for example, has a confusing -v flag to ulimit. o Document this in bsd.port.mk.5 espie@ ok
2001-11-11 08:57:32 -05:00
.if ${VMEM_WARNING:L} == "yes"
@echo ""; \
echo "*** WARNING: you may see an error such as"; \
echo "*** virtual memory exhausted"; \
echo "*** when building this package. If you do you must increase"; \
echo "*** your limits. See the man page for your shell and look"; \
echo "*** for the 'limit' or 'ulimit' command. You may also want to"; \
echo "*** see the login.conf(5) manual page."; \
echo "*** Some examples are: "; \
mention tcsh
2002-03-10 01:02:16 -05:00
echo "*** csh(1) and tcsh(1): limit datasize <kbytes of memory>"; \
o Instead of copying the "virtual memory exhausted" warning for each individual port, introduce a new variable for bsd.port.mk, VMEM_WARNING. By setting this variable to `Yes', the user is given a warning about the virtual memory requirements just before the pre-build stage. o Also give some examples for different shells for what to do about it, and mention login.conf(5). bash, for example, has a confusing -v flag to ulimit. o Document this in bsd.port.mk.5 espie@ ok
2001-11-11 08:57:32 -05:00
echo "*** ksh(1), zsh(1) and bash(1): ulimit -d <kbytes of memory>"; \
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
echo ""
o Instead of copying the "virtual memory exhausted" warning for each individual port, introduce a new variable for bsd.port.mk, VMEM_WARNING. By setting this variable to `Yes', the user is given a warning about the virtual memory requirements just before the pre-build stage. o Also give some examples for different shells for what to do about it, and mention login.conf(5). bash, for example, has a confusing -v flag to ulimit. o Document this in bsd.port.mk.5 espie@ ok
2001-11-11 08:57:32 -05:00
.endif
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
. if target(pre-build)
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} pre-build
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
. endif
. if target(do-build)
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} do-build
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
. else
Kill _PORT_USE macro. This trims down the number of sub-makes the port system runs quite a bit, and makes some other fun stuff possible. THIS KILLS THE pre-*/post-* SCRIPT STUFF The functionality is unneeded, as it can be done with normal pre-*/post-* targets, but it must be kept in mind when porting Free/NetBSD ports.
2000-01-26 18:11:08 -05:00
# What BUILD normally does:
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${WRKBUILD} && exec ${_SYSTRACE_CMD} ${SETENV} ${MAKE_ENV} ${MAKE_PROGRAM} ${MAKE_FLAGS} -f ${MAKE_FILE} ${ALL_TARGET}
Kill _PORT_USE macro. This trims down the number of sub-makes the port system runs quite a bit, and makes some other fun stuff possible. THIS KILLS THE pre-*/post-* SCRIPT STUFF The functionality is unneeded, as it can be done with normal pre-*/post-* targets, but it must be kept in mind when porting Free/NetBSD ports.
2000-01-26 18:11:08 -05:00
# End of BUILD
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
. endif
. if target(post-build)
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${.CURDIR} && exec ${_SYSTRACE_CMD} ${MAKE} post-build
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
. endif
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
.endif
More regularity, use $@ to regenerate target cookies
2001-03-28 06:32:29 -05:00
@${_MAKE_COOKIE} $@
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
basic regression testing framework.
2001-10-03 04:53:18 -04:00
${_REGRESS_COOKIE}: ${_BUILD_COOKIE}
.if ${NO_REGRESS:L} == "no"
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
@${ECHO_MSG} "===> Regression check for ${FULLPKGNAME}${_MASTER}"
basic regression testing framework.
2001-10-03 04:53:18 -04:00
. if target(pre-regress)
@cd ${.CURDIR} && exec ${MAKE} pre-regress
. endif
. if target(do-regress)
@cd ${.CURDIR} && exec ${MAKE} do-regress
. else
# What REGRESS normally does:
In retrospect, just allow REGRESS_FLAGS. This would be redundant otherwise
2001-10-07 06:53:43 -04:00
@cd ${WRKBUILD} && exec ${SETENV} ${MAKE_ENV} ${MAKE_PROGRAM} ${REGRESS_FLAGS} -f ${MAKE_FILE} ${REGRESS_TARGET}
basic regression testing framework.
2001-10-03 04:53:18 -04:00
# End of REGRESS
. endif
. if target(post-regress)
@cd ${.CURDIR} && exec ${MAKE} post-regress
. endif
.else
@echo 1>&2 "No regression check for ${FULLPKGNAME}"
.endif
@${_MAKE_COOKIE} $@
no need to generate an intermediate mtree.spec
2006-11-22 07:40:50 -05:00
${_FAKE_COOKIE}: ${_BUILD_COOKIE}
let build of dependencies that are not 'install' or 'package' happen inside a subdirectory of the working directory, instead of elsewhere. Allows packages BULK=Yes to work correctly, and increase reproducibility of builds. tested by naddy@ through several incarnations. Ports that use patch/configure/build dependencies must be adjusted for this. Pleasingly enough, the adjusted version is often simpler. The change uses an internal _MASTER variable to recall the list of ports built during dependencies, for less confusion.
2002-04-24 17:29:26 -04:00
@${ECHO_MSG} "===> Faking installation for ${FULLPKGNAME}${_MASTER}"
zap uses of ${SH}
2006-11-09 03:29:47 -05:00
@if [ x`${SUDO} /bin/sh -c umask` != x${DEF_UMASK} ]; then \
echo >&2 "Error: your umask is \"`/bin/sh -c umask`"\".; \
`Fake' infrastructure: if a port sets FAKE=Yes, this means it can be `pre'-installed elsewhere, with DESTDIR set to WRKINST (=work/fake-${ARCH} by default). the infrastructure takes care of pre-install/do-install/post-install targets, assuming those install stuff under PREFIX. To help ports to cope with DESTDIR, you can set FAKE_FLAGS and FAKE_TARGET (used for the fake installation). Ports with FAKE=Yes are the way to go: they can be packaged directly without a real installation, and the installation proceeds from the package, thus forcing porters to check the package.
2000-03-03 16:24:50 -05:00
exit 1; \
fi
Support ${SUDO}
2000-05-16 13:06:46 -04:00
@${SUDO} install -d -m 755 -o root -g wheel ${WRKINST}
no need to generate an intermediate mtree.spec
2006-11-22 07:40:50 -05:00
@${_SED_SUBST} ${MTREE_FILE}| \
${SUDO} /usr/sbin/mtree -U -e -d -n -p ${WRKINST} >/dev/null
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.for _p in ${PROTECT_MOUNT_POINTS}
Add PROTECT_MOUNT_POINTS, to help package builders spot ports that write outside of the fake area.
2002-04-09 19:55:11 -04:00
@${SUDO} mount -u -r ${_p}
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.endfor
.for _m in ${MODULES}
. if defined(MOD${_m:U}_pre-fake)
Cut out some parts of bsd.port.mk into separate modules. This is work-in-progress. Supplemental hooks will be added when they become necessary. The goal is to permit separate subsystems to tweak the package building process slightly. This should help e.g., python or qt2, for instance. Documentation will come when it's tested more thoroughly...
2001-08-25 07:23:46 -04:00
@${MOD${_m:U}_pre-fake}
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
. endif
.endfor
Add standard method for perl MakeMaker configuration
2000-06-02 07:42:53 -04:00
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.if target(pre-fake)
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${.CURDIR} && exec ${SUDO} ${_SYSTRACE_CMD} ${MAKE} pre-fake ${_FAKE_SETUP}
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.endif
Support ${SUDO}
2000-05-16 13:06:46 -04:00
@${SUDO} ${_MAKE_COOKIE} ${_INSTALL_PRE_COOKIE}
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.if target(pre-install)
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${.CURDIR} && exec ${SUDO} ${_SYSTRACE_CMD} ${MAKE} pre-install ${_FAKE_SETUP}
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.endif
.if target(do-install)
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${.CURDIR} && exec ${SUDO} ${_SYSTRACE_CMD} ${MAKE} do-install ${_FAKE_SETUP}
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.else
`Fake' infrastructure: if a port sets FAKE=Yes, this means it can be `pre'-installed elsewhere, with DESTDIR set to WRKINST (=work/fake-${ARCH} by default). the infrastructure takes care of pre-install/do-install/post-install targets, assuming those install stuff under PREFIX. To help ports to cope with DESTDIR, you can set FAKE_FLAGS and FAKE_TARGET (used for the fake installation). Ports with FAKE=Yes are the way to go: they can be packaged directly without a real installation, and the installation proceeds from the package, thus forcing porters to check the package.
2000-03-03 16:24:50 -05:00
# What FAKE normally does:
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${WRKBUILD} && exec ${SUDO} ${_SYSTRACE_CMD} ${SETENV} ${MAKE_ENV} ${_FAKE_SETUP} ${MAKE_PROGRAM} ${FAKE_FLAGS} -f ${MAKE_FILE} ${FAKE_TARGET}
`Fake' infrastructure: if a port sets FAKE=Yes, this means it can be `pre'-installed elsewhere, with DESTDIR set to WRKINST (=work/fake-${ARCH} by default). the infrastructure takes care of pre-install/do-install/post-install targets, assuming those install stuff under PREFIX. To help ports to cope with DESTDIR, you can set FAKE_FLAGS and FAKE_TARGET (used for the fake installation). Ports with FAKE=Yes are the way to go: they can be packaged directly without a real installation, and the installation proceeds from the package, thus forcing porters to check the package.
2000-03-03 16:24:50 -05:00
# End of FAKE.
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.endif
.if target(post-install)
this makes our build infrastructure systrace aware original idea from jsyn@, discussed and first tests at c2k3 Warning! - this commit is different from all patches sent around, please remove them before updating - due to a few bugs in systrace this is currently not ready for the casual porter and several ports will fail to build, you've been warned The idea of this patch is to help a porter when developing a new port. With systrace the configure, build and fake stages are not allowed to open network connections or write outside some well defined directories. This way misbehaving programs will be noticed due to logfile entries in /var/log/messages and the port can be fixed. There is generally no need for endusers to use this, as the checksum ensures that ports in the future will behave the same as they did when porting. :) To activate systrace'd port building, set USE_SYSTRACE=Yes (e.g. in /etc/mk.conf) tested by some people, ok espie@
2003-07-28 13:17:04 -04:00
@cd ${.CURDIR} && exec ${SUDO} ${_SYSTRACE_CMD} ${MAKE} post-install ${_FAKE_SETUP}
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.endif
.for _p in ${PROTECT_MOUNT_POINTS}
Add PROTECT_MOUNT_POINTS, to help package builders spot ports that write outside of the fake area.
2002-04-09 19:55:11 -04:00
@${SUDO} mount -u -w ${_p}
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.endfor
More regularity, use $@ to regenerate target cookies
2001-03-28 06:32:29 -05:00
@${SUDO} ${_MAKE_COOKIE} $@
`Fake' infrastructure: if a port sets FAKE=Yes, this means it can be `pre'-installed elsewhere, with DESTDIR set to WRKINST (=work/fake-${ARCH} by default). the infrastructure takes care of pre-install/do-install/post-install targets, assuming those install stuff under PREFIX. To help ports to cope with DESTDIR, you can set FAKE_FLAGS and FAKE_TARGET (used for the fake installation). Ports with FAKE=Yes are the way to go: they can be packaged directly without a real installation, and the installation proceeds from the package, thus forcing porters to check the package.
2000-03-03 16:24:50 -05:00
revise the way package multi-builds are handled. - move FLAVOR_EXT up so that WRK* are correct when they're needed for dependencies. - build a list of _PACKAGE_COOKIES, one for each subpackage. - let package go through an indirection: package: ${_PACKAGE_COOKIES} _PACKAGE_COOKIE_${SUB}: SUBPACKAGE=${SUB} ${MAKE} _package so that all packages are `flat', and subpackages can depend on each other freely.
2001-04-02 06:35:51 -04:00
# The real install
quick-fix for make install, as reported by stevens@ and naddy@. Need more thinking about...
2006-10-12 04:45:56 -04:00
${_INSTALL_COOKIE}:
Fix install, as noticed by fgs: just do a partial depend on the package cookies, and also insist on running make package, which is necessary to make sure all MULTI_PACKAGES get built. Repair make clean=packages: make sure all packages get named using the same scheme as other multi-packages targets (that way, arch-indep subpackages get cleaned properly)
2006-09-23 05:40:06 -04:00
@cd ${.CURDIR} && exec ${MAKE} package
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
@cd ${.CURDIR} && DEPENDS_TARGET=install PACKAGING='${SUBPACKAGE}' exec ${MAKE} _internal-run-depends _internal-runlib-depends _internal-runwantlib-depends
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
@${ECHO_MSG} "===> Installing ${FULLPKGNAME${SUBPACKAGE}} from ${_PKG_REPO}"
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.for _m in ${MODULES}
. if defined(MOD${_m:U}_pre_install)
Cut out some parts of bsd.port.mk into separate modules. This is work-in-progress. Supplemental hooks will be added when they become necessary. The goal is to permit separate subsystems to tweak the package building process slightly. This should help e.g., python or qt2, for instance. Documentation will come when it's tested more thoroughly...
2001-08-25 07:23:46 -04:00
@${MOD${_m:U}_pre_install}
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
. endif
.endfor
.if ${TRUST_PACKAGES:L} == "yes"
use pkg_info -q -e instead of pkg dependencies check now that it does what it should. Use the exact same rules for checking LIB_DEPENDS that will later be used to build the package itself, e.g., add the pkg dependency check.
2004-11-15 06:49:05 -05:00
@if pkg_info -q -e ${FULLPKGNAME${SUBPACKAGE}}; then \
Change PKGNAME and FULLPKGNAME semantics slightly. Since those are mostly infrastructure-side, making SUBPACKAGE transparent only adds to the confusion, so don't interpolate SUBPACKAGE into FULLPKGNAME, hiding it into _FULLPKGNAME, but rather let FULLPKGNAME tag the main package, and use FULLPKGNAME-foo for subpackage -foo. This will simplify some other stuff...
2001-08-12 07:33:10 -04:00
echo "Package ${FULLPKGNAME${SUBPACKAGE}} is already installed"; \
TRUST_PACKAGES: use actual installed packages to check dependencies and avoid calling pkg_add. Solve some of art's problems.
2001-04-20 12:25:24 -04:00
else \
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
${SUDO} ${SETENV} PKG_PATH=${_PKG_REPO} PKG_TMPDIR=${PKG_TMPDIR} pkg_add ${_PKG_ADD_AUTO} ${PKGFILE}; \
TRUST_PACKAGES: use actual installed packages to check dependencies and avoid calling pkg_add. Solve some of art's problems.
2001-04-20 12:25:24 -04:00
fi
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
.else
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
@${SUDO} ${SETENV} PKG_PATH=${_PKG_REPO} PKG_TMPDIR=${PKG_TMPDIR} pkg_add ${_PKG_ADD_AUTO} ${PKGFILE}
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
.endif
say goodbye for FAKE=no, for real.
2006-10-15 15:29:06 -04:00
@-${SUDO} ${_MAKE_COOKIE} $@
`Fake' infrastructure: if a port sets FAKE=Yes, this means it can be `pre'-installed elsewhere, with DESTDIR set to WRKINST (=work/fake-${ARCH} by default). the infrastructure takes care of pre-install/do-install/post-install targets, assuming those install stuff under PREFIX. To help ports to cope with DESTDIR, you can set FAKE_FLAGS and FAKE_TARGET (used for the fake installation). Ports with FAKE=Yes are the way to go: they can be packaged directly without a real installation, and the installation proceeds from the package, thus forcing porters to check the package.
2000-03-03 16:24:50 -05:00
fix update: now that we no longer recurse at the cookie level, we have to recurse higher.
2006-09-19 07:25:17 -04:00
${_UPDATE_COOKIE}:
more fixes for new issues: let the update cookie depend on the `full' package cookie, so that an update will get all subpackages built. And a fix for an old issue: let make update proceed to update ALL subpackages, as seems the most natural. Create a subupdate target if one specifically wants to update a single package (and use it as dependency in the FORCE_UPDATE case).
2006-09-27 06:09:34 -04:00
@cd ${.CURDIR} && exec ${MAKE} _internal-package
can't depend on WRKDIR unconditionally, throws update off when UPDATE_COOKIES_DIR is set.
2004-11-17 06:09:44 -05:00
.if empty(UPDATE_COOKIES_DIR)
@exec ${MAKE} ${WRKDIR}
.else
industrial-strength updates: - separate UPDATE_COOKIES_DIR, unless set to empty. Makes for easier management. - FORCE_UPDATE variable, that makes certain packages are updated when a dependency is run, and that forces the replacement of packages that are already there. Still uses the update cookie, so that this update occurs just once unless you wipe out yur UPDATE_COOKIES_DIR (which justifies putting update cookies in a separate directory).
2004-11-15 08:52:03 -05:00
@mkdir -p ${UPDATE_COOKIES_DIR}
.endif
Add a simple update target, that relies on pkg_info -e pkgpath to know if there is something to update...
2004-11-10 05:16:41 -05:00
@${ECHO_MSG} "===> Updating for ${FULLPKGNAME${SUBPACKAGE}}"
@a=`pkg_info -e ${FULLPKGPATH} 2>/dev/null || true`; \
case $$a in \
Finally rephrase what's going on.
2005-03-09 10:21:07 -05:00
'') ${ECHO_MSG} "Not installed, no update";; \
Add a simple update target, that relies on pkg_info -e pkgpath to know if there is something to update...
2004-11-10 05:16:41 -05:00
*) ${ECHO_MSG} "Upgrading from $$a"; \
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
${SUDO} ${SETENV} PKG_PATH=${_PKG_REPO} PKG_TMPDIR=${PKG_TMPDIR} pkg_add ${_PKG_ADD_AUTO} -r ${_PKG_ADD_FORCE} ${PKGFILE};; \
Add a simple update target, that relies on pkg_info -e pkgpath to know if there is something to update...
2004-11-10 05:16:41 -05:00
esac
@${_MAKE_COOKIE} $@
revise the way package multi-builds are handled. - move FLAVOR_EXT up so that WRK* are correct when they're needed for dependencies. - build a list of _PACKAGE_COOKIES, one for each subpackage. - let package go through an indirection: package: ${_PACKAGE_COOKIES} _PACKAGE_COOKIE_${SUB}: SUBPACKAGE=${SUB} ${MAKE} _package so that all packages are `flat', and subpackages can depend on each other freely.
2001-04-02 06:35:51 -04:00
# The real package
MULTI_PACKAGES framework. This is somewhat orthogonal to FLAVORS. Principle: one port may build several packages in one go. For instance, egcs will build all compilers, and package stuff as base, C++, Fortran, etc. This simplifies some japanese ports greatly, e.g., one Canna port that builds libs/server/app packages. How to use: just set MULTI_PACKAGES=-ext1 -ext2... in the Makefile. Then make package will invoke itself recursively with SUBPACKAGE set to -ext1, -ext2. The SUBPACKAGE will fetch package info as PLIST-ext1, COMMENT-ext1, DESCR-ext1, etc. The port Makefile itself can test the value of SUBPACKAGE if things differ. Some problems: - there still is a main package (with PLIST, DESCR, etc), and it's built last, so you can't add RUN_DEPENDS=main_package to subpackages, you have to encode them explicitly in the PLIST for now. - other ports can't depend on subpackages yet. This is usually not a problem, it's just a question of choosing the `right' main package (e.g., for Canna, it will be canna-lib). - PLIST.sed recognize @FLAVOR@, which is extended to the current flavor, so that a subpackage may depend on the corresponding main package in explicit ways. describe is also invoked recursively, so that all subpackages appear in the INDEX.
2000-04-17 16:12:03 -04:00
clean-up register-plist logic, so that package is uniformously deleted if a problem occurs.
2005-09-05 08:43:07 -04:00
.if empty(PLIST_DB)
_register_plist=:
.else
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
_register_plist=perl ${PORTSDIR}/infrastructure/package/register-plist ${PLIST_DB} ${_PACKAGE_COOKIE}
clean-up register-plist logic, so that package is uniformously deleted if a problem occurs.
2005-09-05 08:43:07 -04:00
.endif
tweak _print-package-args a bit. Add internal NODEPS to turn off computing dependencies. Write code similar to _print-package-args that is strictly ports based, and hence really walks the wantlibs, and needs the pkg_create stuff to get at the real ports contents. Recode print-package-signature to refer to the ports tree contents exclusively (pkg_info -S for the package). Thx bernd@ for trying that out. This should make for a slow, but accurate, out-of-date printer.
2005-09-24 15:46:56 -04:00
print-plist:
also mention -n along -q on `print-plist' commands. I've started documenting it, and having -q do things on its own is quirky...
2006-10-15 05:20:53 -04:00
@${PKG_CMD} -n -q ${PKG_ARGS} ${_PACKAGE_COOKIE}
tweak _print-package-args a bit. Add internal NODEPS to turn off computing dependencies. Write code similar to _print-package-args that is strictly ports based, and hence really walks the wantlibs, and needs the pkg_create stuff to get at the real ports contents. Recode print-package-signature to refer to the ports tree contents exclusively (pkg_info -S for the package). Thx bernd@ for trying that out. This should make for a slow, but accurate, out-of-date printer.
2005-09-24 15:46:56 -04:00
print-plist-contents:
-Q implies -q, anyways.
2006-10-15 05:22:49 -04:00
@${PKG_CMD} -n -Q ${PKG_ARGS} ${_PACKAGE_COOKIE}
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
_internal-subpackage: ${_PACKAGE_COOKIES}
tweak _print-package-args a bit. Add internal NODEPS to turn off computing dependencies. Write code similar to _print-package-args that is strictly ports based, and hence really walks the wantlibs, and needs the pkg_create stuff to get at the real ports contents. Recode print-package-signature to refer to the ports tree contents exclusively (pkg_info -S for the package). Thx bernd@ for trying that out. This should make for a slow, but accurate, out-of-date printer.
2005-09-24 15:46:56 -04:00
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
fetch-all:
kill tweaks that make no sense: NO_DESCRIBE, NO_PACKAGE, FETCH_BEFORE_ARGS, FETCH_AFTER_ARGS, NO_WARNINGS (is this used ?), VARNAME, FORCE_PACKAGE. Make CKSUMFILES invisible from users. Make sure repackage and reinstall go through normal clean targets. Kill pre-clean, pre-distclean, pre-repackage. Don't allow customization of distclean.
2003-07-14 09:33:04 -04:00
@cd ${.CURDIR} && exec ${MAKE} __FETCH_ALL=Yes __ARCH_OK=Yes NO_IGNORE=Yes fetch
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
# Separate target for each file fetch will retrieve
Finish cleansing PATCH_SITES out.
2000-02-04 06:15:16 -05:00
.for _F in ${ALLFILES:S@^@${FULLDISTDIR}/@}
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
${_F}:
FETCH_MANUALLY okay naddy
2003-07-25 08:46:26 -04:00
. if ${FETCH_MANUALLY:L} != "no"
. for _M in ${FETCH_MANUALLY}
@echo "*** ${_M}"
. endfor
@exit 1
. else
more generic protection against multiple locking: instead of one _MASTER_LOCK, we keep a list of _LOCKS_HELD by one port builder in an env variable, and we don't relock stuff that's already locked. This allows us to, e.g., have fake depend on regress without issue (even though, internally, it's more efficient to use the _internal-* targets to avoid testing locks).
2006-09-17 15:08:59 -04:00
@lock=${_F:T}.dist; ${_SIMPLE_LOCK}; mkdir -p ${_F:H}; \
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
cd ${_F:H}; \
Finish cleansing PATCH_SITES out.
2000-02-04 06:15:16 -05:00
select=${_EVERYTHING:M*${_F:S@^${FULLDISTDIR}/@@}\:[0-9]}; \
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
f=${_F:S@^${FULLDISTDIR}/@@}; \
${ECHO_MSG} ">> $$f doesn't seem to exist on this system."; \
${_CDROM_OVERRIDE}; \
Finish cleansing PATCH_SITES out.
2000-02-04 06:15:16 -05:00
${_SITE_SELECTOR}; \
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
for site in $$sites; do \
- new read-only variable, PKGNAMES. - plist repository under PLIST_DB (optional). - print-package-signatures shows what's used for signatures. - make fetch shows full url you can copy/paste, simpler to look for typos that way.
2005-09-04 18:32:37 -04:00
${ECHO_MSG} ">> Fetch $${site}$$f."; \
kill tweaks that make no sense: NO_DESCRIBE, NO_PACKAGE, FETCH_BEFORE_ARGS, FETCH_AFTER_ARGS, NO_WARNINGS (is this used ?), VARNAME, FORCE_PACKAGE. Make CKSUMFILES invisible from users. Make sure repackage and reinstall go through normal clean targets. Kill pre-clean, pre-distclean, pre-repackage. Don't allow customization of distclean.
2003-07-14 09:33:04 -04:00
if ${FETCH_CMD} $${site}$$f; then \
let fetch check the size of the retrieved file, if this is recorded in distinfo. This allows ftp to skip files which return 404 error files and other such nonsense... (note that if you want to really grab a file with the wrong size, you can just say CHECKSUM_FILE=/dev/null) with input from by xsa, sturm, ian..
2005-01-04 15:44:45 -05:00
file=${_F:S@^${DISTDIR}/@@}; \
switch to right directory before checking.
2005-01-06 14:30:34 -05:00
ck=`cd ${DISTDIR} && ${_size_fragment}`; \
don't throw an error in fetch, when no distinfo file exists but error out in _internal-checksum, when there should be one ok espie
2006-10-18 12:53:04 -04:00
if [ ! -f ${CHECKSUM_FILE} ]; then \
${ECHO_MSG} ">> Checksum file does not exist"; \
exit 0; \
elif grep -q "^$$ck\$$" ${CHECKSUM_FILE}; then \
let fetch check the size of the retrieved file, if this is recorded in distinfo. This allows ftp to skip files which return 404 error files and other such nonsense... (note that if you want to really grab a file with the wrong size, you can just say CHECKSUM_FILE=/dev/null) with input from by xsa, sturm, ian..
2005-01-04 15:44:45 -05:00
${ECHO_MSG} ">> Size matches for ${_F}"; \
exit 0; \
else \
if grep -q "SIZE ($$file)" ${CHECKSUM_FILE}; then \
${ECHO_MSG} ">> Size does not match for ${_F}"; \
fix size test finally (|| > |) check that MULTI_PACKAGES do begin with -.
2006-03-24 14:28:13 -05:00
test `{ wc -c "$$file" 2>/dev/null || echo 0 ; }| awk '{print $$1}'` -lt 30000 && rm -f $$file; \
let fetch check the size of the retrieved file, if this is recorded in distinfo. This allows ftp to skip files which return 404 error files and other such nonsense... (note that if you want to really grab a file with the wrong size, you can just say CHECKSUM_FILE=/dev/null) with input from by xsa, sturm, ian..
2005-01-04 15:44:45 -05:00
else \
${ECHO_MSG} ">> No size recorded for ${_F}"; \
exit 0; \
fi; \
fi; \
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
fi; \
done; exit 1
FETCH_MANUALLY okay naddy
2003-07-25 08:46:26 -04:00
. endif
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
.endfor
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
.for _l _o in ${_PACKAGE_LINKS}
${PACKAGE_REPOSITORY}/${_l}/${_PKGFILE}: ${PACKAGE_REPOSITORY}/${_o}/${_PKGFILE}
@echo "Link to $@"
@mkdir -p ${@D}
@rm -f $@
@ln $? $@ 2>/dev/null || \
cp -p $? $@
simplify the creation of links in the package repository. ok naddy@, pval@
2003-07-18 14:34:26 -04:00
.endfor
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
# Cleaning up
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
_internal-clean:
get thru _clean to handle clean, since it's set in stone. Add work if only depends is set. problem found by naddy@
2003-08-01 04:07:30 -04:00
.if ${_clean:L:Mdepends} && ${_CLEANDEPENDS:L} == "yes"
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
@PACKAGING='${SUBPACKAGE}' ${MAKE} all-dir-depends|tsort -r|${_zap_last_line}|while read subdir; do \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
${_flavor_fragment}; \
more generic protection against multiple locking: instead of one _MASTER_LOCK, we keep a list of _LOCKS_HELD by one port builder in an env variable, and we don't relock stuff that's already locked. This allows us to, e.g., have fake depend on regress without issue (even though, internally, it's more efficient to use the _internal-* targets to avoid testing locks).
2006-09-17 15:08:59 -04:00
eval $$toset ${MAKE} _CLEANDEPENDS=No clean; \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
done
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
.endif
Change PKGNAME and FULLPKGNAME semantics slightly. Since those are mostly infrastructure-side, making SUBPACKAGE transparent only adds to the confusion, so don't interpolate SUBPACKAGE into FULLPKGNAME, hiding it into _FULLPKGNAME, but rather let FULLPKGNAME tag the main package, and use FULLPKGNAME-foo for subpackage -foo. This will simplify some other stuff...
2001-08-12 07:33:10 -04:00
@${ECHO_MSG} "===> Cleaning for ${FULLPKGNAME${SUBPACKAGE}}"
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
.if ${_clean:L:Mfake}
Display `cleaning for' at the right point. Somehow, this got swapped.
2000-06-07 11:46:15 -04:00
@if cd ${WRKINST} 2>/dev/null; then ${SUDO} rm -rf ${WRKINST}; fi
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
.endif
.if ${_clean:L:Mwork}
. if ${_clean:L:Mflavors}
Change WRKDIR name, to flatten things somewhat. work-* is too long, because we want to include the PKGNAME in the dir name (so that PKGNAME changes are less likely to work with old work directories) Change the way clean works, so that you can say make clean='dist depends packages' and have it do what you think it should do. Common fragment to _create_wrkobjdir. Put some info into PACKAGING (name of subpackage being built)
2001-08-16 10:49:31 -04:00
@for i in ${.CURDIR}/w-*; do \
if [ -L $$i ]; then ${SUDO} rm -rf `readlink $$i`; fi; \
${SUDO} rm -rf $$i; \
done
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
. else
Remove the real-* targets, do things directly at the cookie level. More submakes die. Remove NO_WRKDIR `feature' Fix `make clean' to do the right thing even if WRKOBJDIR is not defined at that point. It's doubtful package-noinstall is useful, but I left it in for now, with slightly altered semantics.
2000-02-04 06:09:33 -05:00
@if [ -L ${WRKDIR} ]; then rm -rf `readlink ${WRKDIR}`; fi
@rm -rf ${WRKDIR}
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
. endif
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
.endif
.if ${_clean:L:Mdist}
Fix clean/dist
2001-08-24 10:39:36 -04:00
@${ECHO_MSG} "===> Dist cleaning for ${FULLPKGNAME${SUBPACKAGE}}"
@if cd ${FULLDISTDIR} 2>/dev/null; then \
zap old test that's no longer needed since 2001.
2005-11-05 18:54:53 -05:00
rm -f ${_DISTFILES} ${_PATCHFILES}; \
Fix clean/dist
2001-08-24 10:39:36 -04:00
fi
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
. if !empty(DIST_SUBDIR)
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
-@rmdir ${FULLDISTDIR}
Change WRKDIR name, to flatten things somewhat. work-* is too long, because we want to include the PKGNAME in the dir name (so that PKGNAME changes are less likely to work with old work directories) Change the way clean works, so that you can say make clean='dist depends packages' and have it do what you think it should do. Common fragment to _create_wrkobjdir. Put some info into PACKAGING (name of subpackage being built)
2001-08-16 10:49:31 -04:00
. endif
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
.endif
.if ${_clean:L:Minstall}
. if ${_clean:L:Msub}
. for _s in ${MULTI_PACKAGES}
remove "-f" from calls to pkg_delete, fixes "make uninstall" ok espie@
2004-01-28 15:16:14 -05:00
-${SUDO} ${PKG_DELETE} ${FULLPKGNAME${_s}}
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
. endfor
. else
remove "-f" from calls to pkg_delete, fixes "make uninstall" ok espie@
2004-01-28 15:16:14 -05:00
-${SUDO} ${PKG_DELETE} ${FULLPKGNAME${SUBPACKAGE}}
Change WRKDIR name, to flatten things somewhat. work-* is too long, because we want to include the PKGNAME in the dir name (so that PKGNAME changes are less likely to work with old work directories) Change the way clean works, so that you can say make clean='dist depends packages' and have it do what you think it should do. Common fragment to _create_wrkobjdir. Put some info into PACKAGING (name of subpackage being built)
2001-08-16 10:49:31 -04:00
. endif
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
.endif
.if ${_clean:L:Mpackages} || ${_clean:L:Mpackage} && ${_clean:L:Msub}
. if defined(MULTI_PACKAGES)
. for _s in ${MULTI_PACKAGES}
Fix install, as noticed by fgs: just do a partial depend on the package cookies, and also insist on running make package, which is necessary to make sure all MULTI_PACKAGES get built. Repair make clean=packages: make sure all packages get named using the same scheme as other multi-packages targets (that way, arch-indep subpackages get cleaned properly)
2006-09-23 05:40:06 -04:00
@cd ${.CURDIR} && PACKAGING='${_s}' SUBPACKAGE='${_s}' exec ${MAKE} clean=package
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
. endfor
all MULTI_PACKAGES are new, stop testing -main.
2006-11-26 12:36:07 -05:00
. else
@cd ${.CURDIR} && PACKAGING='' SUBPACKAGE='' exec ${MAKE} clean=package
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
. endif
.elif ${_clean:L:Mpackage}
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
rm -f ${_PACKAGE_COOKIES}
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
.endif
.if ${_clean:L:Mreadmes}
all MULTI_PACKAGES are new, stop testing -main.
2006-11-26 12:36:07 -05:00
. if defined(MULTI_PACKAGES)
. for _s in ${MULTI_PACKAGES}
rm -f ${.CURDIR}/${FULLPKGNAME${_s}}.html
. endfor
. else
kill old readme (useless distinction). Let make clean=readmes work.
2003-08-04 09:25:36 -04:00
rm -f ${.CURDIR}/${FULLPKGNAME}.html
start of multi-packages simplification: if MULTI_PACKAGES contains -main, consider it to be the main package, and do some equal treatment for each package. (to do: add subpackage where needed to WANTLIB, RUN_DEPENDS, LIB_DEPENDS and act on them accordingly). Also define _DONE_FILES for _fetch-makefile to avoid blocking if it's not invoked from a higher level target.
2006-11-17 19:15:33 -05:00
. endif
tweak the depends clean recursion, so that it does not recurse into the main directory... since pseudo-flavors are not entered as pkgpath, this is necessary to ensure the top-level directory gets cleaned up correctly... Problem reported by Moritz Grimm.
2005-11-11 06:31:53 -05:00
.endif
.if ${_clean:L:Mbulk}
kill bulk-packages. Use `make package BULK=Yes' instead. The main advantage is that workdirs are cleaned up as soon as package is built (or installed), not later.
2002-04-17 11:58:48 -04:00
rm -f ${_BULK_COOKIE}
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
.endif
Basic framework for fetching distfiles for mirrors.
2000-03-03 12:51:37 -05:00
# mirroring utilities
fetch-makefile:
move everything to _fetch-makefile, so that __FETCH_ALL can put all file names where it should. noticed by jolan@, okay sturm@
2004-07-12 04:45:32 -04:00
@exec ${MAKE} __FETCH_ALL=Yes __ARCH_OK=Yes NO_IGNORE=Yes _fetch-makefile
_fetch-makefile:
skip ports marked COMES_WITH on mirror-maker runs; ok espie@
2001-06-30 15:06:09 -04:00
.if !defined(COMES_WITH)
Add fetch-makefile to bsd.port.subdir.mk, mirror-maker to main Makefile. Tweak fetch-makefile to take PERMIT_* into account, giving us the level of control needed.
2000-03-04 13:16:02 -05:00
@echo -n "all"
More stringent PERMIT_* checking: if any info is missing, assume the maintainer made a mistake, and turn everything to No. Record it as _BAD_LICENSING for describe purpose. Yep, having this as ERRORS is obnoxious. Then, we can rely on PERMIT_* always being defined. Rewrite ftp-packages/cdrom-packages to use more current rules. Fixes naddy's issue of ignoring IGNORE. Use a .for loop, since those rules are basically identical. No longer use them to try and record port logging, as we have a much better script to do that. `Enhance' bulk-packages to do ftp-packages/cdrom-packages linking as well, so that one single sweep of the tree should be enough for everything.
2001-09-30 08:26:57 -04:00
. if ${PERMIT_DISTFILES_FTP:L} == "yes"
Add fetch-makefile to bsd.port.subdir.mk, mirror-maker to main Makefile. Tweak fetch-makefile to take PERMIT_* into account, giving us the level of control needed.
2000-03-04 13:16:02 -05:00
@echo -n " ftp"
skip ports marked COMES_WITH on mirror-maker runs; ok espie@
2001-06-30 15:06:09 -04:00
. endif
More stringent PERMIT_* checking: if any info is missing, assume the maintainer made a mistake, and turn everything to No. Record it as _BAD_LICENSING for describe purpose. Yep, having this as ERRORS is obnoxious. Then, we can rely on PERMIT_* always being defined. Rewrite ftp-packages/cdrom-packages to use more current rules. Fixes naddy's issue of ignoring IGNORE. Use a .for loop, since those rules are basically identical. No longer use them to try and record port logging, as we have a much better script to do that. `Enhance' bulk-packages to do ftp-packages/cdrom-packages linking as well, so that one single sweep of the tree should be enough for everything.
2001-09-30 08:26:57 -04:00
. if ${PERMIT_DISTFILES_CDROM:L} == "yes"
Add fetch-makefile to bsd.port.subdir.mk, mirror-maker to main Makefile. Tweak fetch-makefile to take PERMIT_* into account, giving us the level of control needed.
2000-03-04 13:16:02 -05:00
@echo -n " cdrom"
skip ports marked COMES_WITH on mirror-maker runs; ok espie@
2001-06-30 15:06:09 -04:00
. endif
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
@echo ":: ${_FMN}"
# write generic package dependencies
@echo ".PHONY: ${_FMN}"
. if ${RECURSIVE_FETCH_LIST:L} == "yes"
filter out the current PKGPATH in fetch_makefile, prevents infinite recursion.
2006-11-22 03:20:20 -05:00
@echo "${_FMN}: ${_ALLFILES} "`_FULL_PACKAGE_NAME=Yes ${MAKE} full-all-depends|fgrep -v ${PKGPATH}/`
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
. else
@echo "${_FMN}: ${_ALLFILES}"
. endif
skip ports marked COMES_WITH on mirror-maker runs; ok espie@
2001-06-30 15:06:09 -04:00
.endif
If DISTFILES + PATCHFILES = ALLFILES is empty, no fetch is needed...
2000-03-05 11:33:38 -05:00
.if !empty(ALLFILES)
. for _F in ${_ALLFILES}
start of multi-packages simplification: if MULTI_PACKAGES contains -main, consider it to be the main package, and do some equal treatment for each package. (to do: add subpackage where needed to WANTLIB, RUN_DEPENDS, LIB_DEPENDS and act on them accordingly). Also define _DONE_FILES for _fetch-makefile to avoid blocking if it's not invoked from a higher level target.
2006-11-17 19:15:33 -05:00
@: $${_DONE_FILES:=/dev/null}; if ! fgrep -q "|${_F}|" $${_DONE_FILES}; then \
Check for duplicates in mirror-maker entries, build Makefiles that won't heap tons of warnings with gnu-make (and that will be slightly smaller as well).
2003-12-15 12:56:40 -05:00
echo "|${_F}|" >>$${_DONE_FILES}; \
Use somewhat dirty trick. Repairs mirror-maker for multiple MASTER_SITES. Noticed by naddy@
2003-12-23 19:08:48 -05:00
${MAKE} _fetch-onefile _file=${_F}; \
Check for duplicates in mirror-maker entries, build Makefiles that won't heap tons of warnings with gnu-make (and that will be slightly smaller as well).
2003-12-15 12:56:40 -05:00
fi
. endfor
.endif
@echo
_fetch-onefile:
Use somewhat dirty trick. Repairs mirror-maker for multiple MASTER_SITES. Noticed by naddy@
2003-12-23 19:08:48 -05:00
# XXX loop so that M${_F} will work
.for _F in ${_file}
Basic framework for fetching distfiles for mirrors.
2000-03-03 12:51:37 -05:00
@echo '${_F}: $$F'
@echo -n '\t@MAINTAINER="${MAINTAINER}" '
always define DIST_SUBDIR, simplifies logic.
2004-02-07 17:36:13 -05:00
. if !empty(DIST_SUBDIR)
Basic framework for fetching distfiles for mirrors.
2000-03-03 12:51:37 -05:00
@echo -n 'DIST_SUBDIR="${DIST_SUBDIR}" '
Use somewhat dirty trick. Repairs mirror-maker for multiple MASTER_SITES. Noticed by naddy@
2003-12-23 19:08:48 -05:00
. endif
Basic framework for fetching distfiles for mirrors.
2000-03-03 12:51:37 -05:00
@echo '\\'
@select='${_EVERYTHING:M*${_F:S@^${DIST_SUBDIR}/@@}\:[0-9]}'; \
${_SITE_SELECTOR}; \
echo "\t SITES=\"$$sites\" \\"
Use somewhat dirty trick. Repairs mirror-maker for multiple MASTER_SITES. Noticed by naddy@
2003-12-23 19:08:48 -05:00
. if ${FETCH_MANUALLY:L} != "no"
don't try to fetch distfiles of ports marked FETCH_MANUALLY ok naddy@
2003-08-28 12:19:00 -04:00
@echo '\t FETCH_MANUALLY="Yes" \\'
Use somewhat dirty trick. Repairs mirror-maker for multiple MASTER_SITES. Noticed by naddy@
2003-12-23 19:08:48 -05:00
. endif
. if !defined(NO_CHECKSUM) && !empty(_CKSUMFILES:M${_F})
transform files/md5 into distinfo magically. I would credit FreeBSD or NetBSD for the idea if I could remember which one came up with it first. :)
2001-11-11 07:59:50 -05:00
@checksum_file=${CHECKSUM_FILE}; \
typo
2001-11-13 07:52:22 -05:00
if [ ! -f $$checksum_file ]; then \
use double quotes to permit expansion of shell variable in message
2004-12-29 09:30:53 -05:00
echo >&2 "Missing checksum file: $$checksum_file"; \
Basic framework for fetching distfiles for mirrors.
2000-03-03 12:51:37 -05:00
echo '\t ERROR="no checksum file" \\'; \
else \
for c in ${PREFERRED_CIPHERS}; do \
transform files/md5 into distinfo magically. I would credit FreeBSD or NetBSD for the idea if I could remember which one came up with it first. :)
2001-11-11 07:59:50 -05:00
if set -- `grep -i "^$$c (${_F})" $$checksum_file`; then break; fi; \
Basic framework for fetching distfiles for mirrors.
2000-03-03 12:51:37 -05:00
done; \
case "$$4" in \
"") \
echo >&2 "No checksum recorded for ${_F}."; \
echo '\t ERROR="no checksum" \\';; \
"IGNORE") \
transform files/md5 into distinfo magically. I would credit FreeBSD or NetBSD for the idea if I could remember which one came up with it first. :)
2001-11-11 07:59:50 -05:00
echo >&2 "Checksum for ${_F} is IGNORE in $$checksum_file"; \
Basic framework for fetching distfiles for mirrors.
2000-03-03 12:51:37 -05:00
echo >&2 'but file is not in $${IGNORE_FILES}'; \
echo '\t ERROR="IGNORE inconsistent" \\';; \
*) \
echo "\t CIPHER=\"$$c\" CKSUM=\"$$4\" \\";; \
esac; \
fi
Use somewhat dirty trick. Repairs mirror-maker for multiple MASTER_SITES. Noticed by naddy@
2003-12-23 19:08:48 -05:00
. endif
Check for duplicates in mirror-maker entries, build Makefiles that won't heap tons of warnings with gnu-make (and that will be slightly smaller as well).
2003-12-15 12:56:40 -05:00
@echo '\t $${EXEC} $${FETCH} "$$@"'
Use somewhat dirty trick. Repairs mirror-maker for multiple MASTER_SITES. Noticed by naddy@
2003-12-23 19:08:48 -05:00
.endfor
sync with NetBSD where possible. Fix WRKOBJDIR. Add deinstall target. Make uninstall an alias for deinstall.
1998-09-07 18:33:19 -04:00
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
# This target generates an index entry suitable for aggregation into
# a large index. Format is:
#
# distribution-name|port-path|installation-prefix|comment| \
match the comment for the describe target with the actual code; ok espie@
2001-11-22 11:02:26 -05:00
# description-file|maintainer|categories|lib-deps|build-deps|run-deps| \
# for-arch|package-cdrom|package-ftp|distfiles-cdrom|distfiles-ftp
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
#
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
subdescribe:
Change PKGNAME and FULLPKGNAME semantics slightly. Since those are mostly infrastructure-side, making SUBPACKAGE transparent only adds to the confusion, so don't interpolate SUBPACKAGE into FULLPKGNAME, hiding it into _FULLPKGNAME, but rather let FULLPKGNAME tag the main package, and use FULLPKGNAME-foo for subpackage -foo. This will simplify some other stuff...
2001-08-12 07:33:10 -04:00
@echo -n "${FULLPKGNAME${SUBPACKAGE}}|${FULLPKGPATH}|"
allows PREFIX and CATEGORIES to be subpackage dependent
2006-11-20 08:50:16 -05:00
.if ${PREFIX${SUBPACKAGE}} == ${LOCALBASE}
Don't record ${PREFIX} in index if it's equal to ${LOCALBASE}
2000-03-28 04:27:03 -05:00
@echo -n "|"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
.else
allows PREFIX and CATEGORIES to be subpackage dependent
2006-11-20 08:50:16 -05:00
@echo -n "${PREFIX${SUBPACKAGE}}|"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
.endif
Fix quoting
2001-05-23 11:48:08 -04:00
@echo -n ${_COMMENT}"|"; \
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
if [ -f ${DESCR} ]; then \
Remove PORTSDIR from describe:, in portsubdir, and in descr. (noticed by turan, quick and dirty implementation until I clean the DIRPRFX mess)
2000-03-26 20:52:17 -05:00
echo -n "${DESCR:S,^${PORTSDIR}/,,}|"; \
Initial import of "ports" mk-files from FreeBSD, probably needs work
1996-06-03 18:47:10 -04:00
else \
Use echo directly, except where ECHO_MSG makes sense. Clean up error conditions somewhat.
1999-11-22 18:44:01 -05:00
echo -n "/dev/null|"; \
Merge 1.264 and 1.20 from FreeBSD's bsd.port.mk and bsd.port.subdir.mk respectively
1997-09-09 11:11:28 -04:00
fi; \
allows PREFIX and CATEGORIES to be subpackage dependent
2006-11-20 08:50:16 -05:00
echo -n "${MAINTAINER}|${CATEGORIES${SUBPACKAGE}}|"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
.for _d in LIB BUILD RUN
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
. if !empty(_${_d}_DEP3)
Use perl lib depends checker, that parses the full gamut of libspecs.
2001-11-17 05:39:19 -05:00
@cd ${.CURDIR} && _FINAL_ECHO=: _INITIAL_ECHO=: exec ${MAKE} ${_d:L}-depends-list
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. endif
Use echo directly, except where ECHO_MSG makes sense. Clean up error conditions somewhat.
1999-11-22 18:44:01 -05:00
@echo -n "|"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
.endfor
Add NOT_FOR_ARCHS
2001-03-28 05:25:38 -05:00
@case "${ONLY_FOR_ARCHS}" in \
"") case "${NOT_FOR_ARCHS}" in \
"") echo -n "any|";; \
*) echo -n "!${NOT_FOR_ARCHS}|";; \
esac;; \
*) echo -n "${ONLY_FOR_ARCHS}|";; \
esac
licensing info in make describe. make print-licenses target
2000-02-21 17:09:57 -05:00
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
.if defined(_BAD_LICENSING)
More stringent PERMIT_* checking: if any info is missing, assume the maintainer made a mistake, and turn everything to No. Record it as _BAD_LICENSING for describe purpose. Yep, having this as ERRORS is obnoxious. Then, we can rely on PERMIT_* always being defined. Rewrite ftp-packages/cdrom-packages to use more current rules. Fixes naddy's issue of ignoring IGNORE. Use a .for loop, since those rules are basically identical. No longer use them to try and record port logging, as we have a much better script to do that. `Enhance' bulk-packages to do ftp-packages/cdrom-packages linking as well, so that one single sweep of the tree should be enough for everything.
2001-09-30 08:26:57 -04:00
@echo "?|?|?|?"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
.else
index PKG_ARCH, PERMIT_PACKAGE_FTP, PERMIT_PACKAGE_CDROM on SUBPACKAGE name.
2006-11-13 09:14:57 -05:00
. if ${PERMIT_PACKAGE_CDROM${SUBPACKAGE}:L} == "yes"
Compute more of INDEX within make, without external test calls. Re-indent rules correctly. Only the make tests are indented, the rule itself still starts after one tab.
2000-02-21 18:53:52 -05:00
@echo -n "y|"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. else
Compute more of INDEX within make, without external test calls. Re-indent rules correctly. Only the make tests are indented, the rule itself still starts after one tab.
2000-02-21 18:53:52 -05:00
@echo -n "n|"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. endif
index PKG_ARCH, PERMIT_PACKAGE_FTP, PERMIT_PACKAGE_CDROM on SUBPACKAGE name.
2006-11-13 09:14:57 -05:00
. if ${PERMIT_PACKAGE_FTP${SUBPACKAGE}:L} == "yes"
Compute more of INDEX within make, without external test calls. Re-indent rules correctly. Only the make tests are indented, the rule itself still starts after one tab.
2000-02-21 18:53:52 -05:00
@echo -n "y|"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. else
Compute more of INDEX within make, without external test calls. Re-indent rules correctly. Only the make tests are indented, the rule itself still starts after one tab.
2000-02-21 18:53:52 -05:00
@echo -n "n|"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. endif
. if ${PERMIT_DISTFILES_CDROM:L} == "yes"
Compute more of INDEX within make, without external test calls. Re-indent rules correctly. Only the make tests are indented, the rule itself still starts after one tab.
2000-02-21 18:53:52 -05:00
@echo -n "y|"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. else
Compute more of INDEX within make, without external test calls. Re-indent rules correctly. Only the make tests are indented, the rule itself still starts after one tab.
2000-02-21 18:53:52 -05:00
@echo -n "n|"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. endif
. if ${PERMIT_DISTFILES_FTP:L} == "yes"
remove trailing -n in make describe pretty INDEX
2000-03-04 09:40:21 -05:00
@echo "y"
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
. else
remove trailing -n in make describe pretty INDEX
2000-03-04 09:40:21 -05:00
@echo "n"
kill tweaks that make no sense: NO_DESCRIBE, NO_PACKAGE, FETCH_BEFORE_ARGS, FETCH_AFTER_ARGS, NO_WARNINGS (is this used ?), VARNAME, FORCE_PACKAGE. Make CKSUMFILES invisible from users. Make sure repackage and reinstall go through normal clean targets. Kill pre-clean, pre-distclean, pre-repackage. Don't allow customization of distclean.
2003-07-14 09:33:04 -04:00
. endif
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
.endif
readmes is yet another target that should recurse.
2006-09-19 16:10:40 -04:00
_readme:
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
@cd ${.CURDIR} && exec ${MAKE} README_NAME=${README_NAME} ${FULLPKGNAME${SUBPACKAGE}}.html
More correct check for describe target... .e.g., if we have special depends target, shortcuts can't be taken. This involves moving those default target around so that they appear AFTER describe. Note that these tests mean ANYTHING adding to *_DEPENDS *MUST* come before the targets that test these for bsd.port.mk to work...
1999-09-30 13:28:31 -04:00
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
${FULLPKGNAME${SUBPACKAGE}}.html:
minor fixes to readme
2003-08-01 05:07:06 -04:00
@echo ${_COMMENT} | ${HTMLIFY} >$@.tmp-comment
Change PKGNAME and FULLPKGNAME semantics slightly. Since those are mostly infrastructure-side, making SUBPACKAGE transparent only adds to the confusion, so don't interpolate SUBPACKAGE into FULLPKGNAME, hiding it into _FULLPKGNAME, but rather let FULLPKGNAME tag the main package, and use FULLPKGNAME-foo for subpackage -foo. This will simplify some other stuff...
2001-08-12 07:33:10 -04:00
@echo ${FULLPKGNAME${SUBPACKAGE}} | ${HTMLIFY} > $@.tmp3
Add HOMEPAGE to README.html
2000-02-15 13:19:12 -05:00
.if defined(HOMEPAGE)
@echo 'See <a href="${HOMEPAGE}">${HOMEPAGE}</a> for details.' >$@.tmp4
.else
@echo "" >$@.tmp4
.endif
put _flavor_fragment into a separate file, that's included from both bsd.port.mk and bsd.port.subdir.mk. Make sure flavors are properly separated. Change the way bsd.port.subdir.mk works, slightly: always set full directories. - SKIPDIR with full paths will work. - all dirs are displayed as full package specs. Generate complete html files.
2003-08-02 05:53:27 -04:00
.if defined(MULTI_PACKAGES)
. if empty(SUBPACKAGE)
@echo "<h2>Subpackages</h2>" >$@.tmp-subpackages
@echo "<ul>" >>$@.tmp-subpackages
. for _S in ${MULTI_PACKAGES}
@name=`SUBPACKAGE=${_S} ${MAKE} _print-packagename _FULL_PACKAGE_NAME=No`; \
echo "<li><a href=\"$$name.html\">$$name</a>" >>$@.tmp-subpackages
. endfor
@echo "</ul>" >>$@.tmp-subpackages
. else
@name=`unset SUBPACKAGE; ${MAKE} _print-packagename _FULL_PACKAGE_NAME=No`; \
echo "<h2>Subpackage of <a href=\"$$name.html\">$$name</a></h2>" >$@.tmp-subpackages
. endif
.else
@>$@.tmp-subpackages
.endif
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
.for _I in build run
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
. if !empty(_${_I:U}_DEP)
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
@cd ${.CURDIR} && ${MAKE} full-${_I}-depends _FULL_PACKAGE_NAME=Yes| \
while read n; do \
- don't display `creating README.html for...', the subdir macros already swamp us with information. - perform the same optimization for README.html, namely don't invoke depends-list and package-depends if they're known not to be needed.
1999-09-30 17:07:09 -04:00
j=`dirname $$n|${HTMLIFY}`; k=`basename $$n|${HTMLIFY}`; \
put _flavor_fragment into a separate file, that's included from both bsd.port.mk and bsd.port.subdir.mk. Make sure flavors are properly separated. Change the way bsd.port.subdir.mk works, slightly: always set full directories. - SKIPDIR with full paths will work. - all dirs are displayed as full package specs. Generate complete html files.
2003-08-02 05:53:27 -04:00
echo "<li><a href=\"${PKGDEPTH}$$j/$$k.html\">$$k</a>"; \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
done >$@.tmp-${_I}
. else
minor fixes to readme
2003-08-01 05:07:06 -04:00
@echo "<li>none" >$@.tmp-${_I}
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
. endif
- don't display `creating README.html for...', the subdir macros already swamp us with information. - perform the same optimization for README.html, namely don't invoke depends-list and package-depends if they're known not to be needed.
1999-09-30 17:07:09 -04:00
.endfor
Introduce _MAKE_COOKIE, rename CIPHERS to _CIPHERS (and change it). Deprecate a whole slew of unneeded variables.
1999-12-03 09:57:12 -05:00
@cat ${README_NAME} | \
minor fixes to readme
2003-08-01 05:07:06 -04:00
sed -e 's|%%PORT%%|'"`echo ${FULLPKGPATH} | ${HTMLIFY}`"'|g' \
Simplify README.html, reuse last pattern instead of re-specifying it.
2000-02-28 13:13:18 -05:00
-e '/%%PKG%%/r$@.tmp3' -e '//d' \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
-e '/%%COMMENT%%/r$@.tmp-comment' -e '//d' \
minor fixes to readme
2003-08-01 05:07:06 -04:00
-e '/%%DESCR%%/r${PKGDIR}/DESCR${SUBPACKAGE}' -e '//d' \
Simplify README.html, reuse last pattern instead of re-specifying it.
2000-02-28 13:13:18 -05:00
-e '/%%HOMEPAGE%%/r$@.tmp4' -e '//d' \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
-e '/%%BUILD_DEPENDS%%/r$@.tmp-build' -e '//d' \
-e '/%%RUN_DEPENDS%%/r$@.tmp-run' -e '//d' \
put _flavor_fragment into a separate file, that's included from both bsd.port.mk and bsd.port.subdir.mk. Make sure flavors are properly separated. Change the way bsd.port.subdir.mk works, slightly: always set full directories. - SKIPDIR with full paths will work. - all dirs are displayed as full package specs. Generate complete html files.
2003-08-02 05:53:27 -04:00
-e '/%%SUBPACKAGES%%/r$@.tmp-subpackages' -e '//d' \
- don't display `creating README.html for...', the subdir macros already swamp us with information. - perform the same optimization for README.html, namely don't invoke depends-list and package-depends if they're known not to be needed.
1999-09-30 17:07:09 -04:00
>> $@
@rm -f $@.tmp*
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
print-build-depends:
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
.if !empty(_BUILD_DEP)
Use echo directly, except where ECHO_MSG makes sense. Clean up error conditions somewhat.
1999-11-22 18:44:01 -05:00
@echo -n 'This port requires package(s) "'
introduce new macros to make things more readable. Kill really old stuff: LD_CONFIG, that's no longer used.
2005-11-05 06:11:17 -05:00
@${MAKE} full-build-depends| ${_lines2list}
Use echo directly, except where ECHO_MSG makes sense. Clean up error conditions somewhat.
1999-11-22 18:44:01 -05:00
@echo '" to build.'
print-depends-list and print-package-depends need similar checks, so they must appear before the default depends targets are generated.
1999-09-30 13:34:22 -04:00
.endif
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
print-run-depends:
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
.if !empty(_RUN_DEP)
Use echo directly, except where ECHO_MSG makes sense. Clean up error conditions somewhat.
1999-11-22 18:44:01 -05:00
@echo -n 'This port requires package(s) "'
introduce new macros to make things more readable. Kill really old stuff: LD_CONFIG, that's no longer used.
2005-11-05 06:11:17 -05:00
@${MAKE} full-run-depends| ${_lines2list}
Use echo directly, except where ECHO_MSG makes sense. Clean up error conditions somewhat.
1999-11-22 18:44:01 -05:00
@echo '" to run.'
print-depends-list and print-package-depends need similar checks, so they must appear before the default depends targets are generated.
1999-09-30 13:34:22 -04:00
.endif
add regress-dir-depends/full-regress-depends, similar mechanism to run-depends, only the basic test changes.
2006-11-05 10:46:51 -05:00
# full-build-depends, full-all-depends, full-run-depends full-regress-depends
.for _i in build all run regress
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
full-${_i}-depends:
introduce new macros to make things more readable. Kill really old stuff: LD_CONFIG, that's no longer used.
2005-11-05 06:11:17 -05:00
@${MAKE} ${_i}-dir-depends|${_sort_dependencies}|while read subdir; do \
Move the `check and change directory' part to flavor_fragment, as it's invariably duplicated right after otherwise. This makes the code cleaner, and makes sure the check happens all the time.
2001-11-12 09:14:46 -05:00
${_flavor_fragment}; \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
eval $$toset ${MAKE} _print-packagename ; \
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
done
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
.endfor
Fix recursive dependency computation.
1999-11-15 13:37:58 -05:00
advisory license-check: warns if a package is apparently built from bad sources. Will returns some false positives.
2003-08-11 16:10:41 -04:00
license-check:
index PKG_ARCH, PERMIT_PACKAGE_FTP, PERMIT_PACKAGE_CDROM on SUBPACKAGE name.
2006-11-13 09:14:57 -05:00
.if ${PERMIT_PACKAGE_CDROM${SUBPACKAGE}:L} == "yes" || ${PERMIT_PACKAGE_FTP${SUBPACKAGE}:L} == "yes"
introduce new macros to make things more readable. Kill really old stuff: LD_CONFIG, that's no longer used.
2005-11-05 06:11:17 -05:00
@${MAKE} all-dir-depends|${_sort_dependencies}|while read subdir; do \
advisory license-check: warns if a package is apparently built from bad sources. Will returns some false positives.
2003-08-11 16:10:41 -04:00
${_flavor_fragment}; \
index PKG_ARCH, PERMIT_PACKAGE_FTP, PERMIT_PACKAGE_CDROM on SUBPACKAGE name.
2006-11-13 09:14:57 -05:00
_MASTER_PERMIT_CDROM=${PERMIT_PACKAGE_CDROM${SUBPACKAGE}:Q}; \
_MASTER_PERMIT_FTP=${PERMIT_PACKAGE_FTP${SUBPACKAGE}:Q}; \
advisory license-check: warns if a package is apparently built from bad sources. Will returns some false positives.
2003-08-11 16:10:41 -04:00
export _MASTER_PERMIT_CDROM _MASTER_PERMIT_FTP; \
eval $$toset ${MAKE} _license-check; \
done
.endif
_license-check:
.for _i in FTP CDROM
. if defined(_MASTER_PERMIT_${_i}) && ${_MASTER_PERMIT_${_i}:L} == "yes" && ${PERMIT_PACKAGE_${_i}:L} != "yes"
@echo >&2 "Warning: dependency ${PKGPATH} is not allowed for ${_i}"
. endif
.endfor
comment to /targets
2005-11-05 18:39:51 -05:00
# run-depends-list, build-depends-list, lib-depends-list
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
.for _i in RUN BUILD LIB
${_i:L}-depends-list:
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
. if !empty(_${_i}_DEP3)
dir -> dir[,-multi][,-flavor...] notation for dependencies. Finally !
2001-03-28 09:33:24 -05:00
@unset FLAVOR SUBPACKAGE || true; \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
: $${_INITIAL_ECHO:='echo -n "This port requires \""'}; \
: $${_ECHO='echo -n'}; \
: $${_FINAL_ECHO:='echo "\" for ${_i:L}."'}; space=''; \
eval $${_INITIAL_ECHO}; \
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
for spec in `echo '${_${_i}_DEP3}' \
Fix recursive dependency computation.
1999-11-15 13:37:58 -05:00
| tr '\040' '\012' | sort -u`; do \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
$${_ECHO} "$$space$${spec}"; \
space=' '; \
done; eval $${_FINAL_ECHO}
. endif
.endfor
More correct check for describe target... .e.g., if we have special depends target, shortcuts can't be taken. This involves moving those default target around so that they appear AFTER describe. Note that these tests mean ANYTHING adding to *_DEPENDS *MUST* come before the targets that test these for bsd.port.mk to work...
1999-09-30 13:28:31 -04:00
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
# recursive depend targets
Fix recursive dependency computation.
1999-11-15 13:37:58 -05:00
- new read-only variable, PKGNAMES. - plist repository under PLIST_DB (optional). - print-package-signatures shows what's used for signatures. - make fetch shows full url you can copy/paste, simpler to look for typos that way.
2005-09-04 18:32:37 -04:00
print-package-signature:
@echo -n ${FULLPKGNAME${SUBPACKAGE}}
distinguish between _BUILDLIB_DEPENDS and LIB_DEPENDS, and _BUILDWANTLIB and WANTLIB. Get the rules for the LIBSPECS_COOKIES out of the loop that dictates their usage. Introduce *DEP3 variables, to keep *-depends-list compatible with old stuff. Use _BUILDLIB_DEPENDS and _BUILDWANTLIB everywhere this makes sense.
2006-11-20 05:36:13 -05:00
.if !empty(_DEPRUNLIBS)
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
@cd ${.CURDIR} && PACKAGING='${SUBPACKAGE}' LIST_LIBS=`${MAKE} _list-port-libs` ${MAKE} _print-package-signature-lib _print-package-signature-run| \
sort -u| \
while read i; do echo -n ",$$i"; done
.else
@cd ${.CURDIR} && PACKAGING='${SUBPACKAGE}' ${MAKE} _print-package-signature-run | \
tweak _print-package-args a bit. Add internal NODEPS to turn off computing dependencies. Write code similar to _print-package-args that is strictly ports based, and hence really walks the wantlibs, and needs the pkg_create stuff to get at the real ports contents. Recode print-package-signature to refer to the ports tree contents exclusively (pkg_info -S for the package). Thx bernd@ for trying that out. This should make for a slow, but accurate, out-of-date printer.
2005-09-24 15:46:56 -04:00
sort -u| \
- new read-only variable, PKGNAMES. - plist repository under PLIST_DB (optional). - print-package-signatures shows what's used for signatures. - make fetch shows full url you can copy/paste, simpler to look for typos that way.
2005-09-04 18:32:37 -04:00
while read i; do echo -n ",$$i"; done
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
.endif
- new read-only variable, PKGNAMES. - plist repository under PLIST_DB (optional). - print-package-signatures shows what's used for signatures. - make fetch shows full url you can copy/paste, simpler to look for typos that way.
2005-09-04 18:32:37 -04:00
@echo
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
_print-package-args:
also use RUN_DEPENDS${SUBPACKAGE} in the simple parts.
2006-11-19 07:37:55 -05:00
.for _i in ${RUN_DEPENDS${SUBPACKAGE}}
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
@echo '${_i}' |{ \
IFS=:; read dep pkg subdir target; \
${_flavor_fragment}; \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
default=`eval $$toset ${MAKE} _print-packagename`; \
introduce new macros to make things more readable. Kill really old stuff: LD_CONFIG, that's no longer used.
2005-11-05 06:11:17 -05:00
case "X$$pkg" in X) pkg=`echo $$default|${_version2default}`;; esac; \
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
echo "-P $$subdir:$$pkg:$$default"; \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
}
.endfor
.if ${NO_SHARED_LIBS:L} != "yes"
let LIB_DEPENDS and WANTLIB depend on the SUBPACKAGE. Build a special list of _BUILDLIB_DEPENDS, so that separate subpackages can depend on each other without making an infinite loop...
2006-11-20 05:49:22 -05:00
. for _i in ${LIB_DEPENDS${SUBPACKAGE}}
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
@echo '${_i}'|{ \
IFS=:; read dep pkg subdir target; \
${_flavor_fragment}; \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
libspecs='';comma=''; \
default=`eval $$toset ${MAKE} _print-packagename`; \
introduce new macros to make things more readable. Kill really old stuff: LD_CONFIG, that's no longer used.
2005-11-05 06:11:17 -05:00
case "X$$pkg" in X) pkg=`echo $$default|${_version2default}`;; esac; \
still need to look in the default package for multi-package cases. use -K that we now have...
2004-11-27 07:36:14 -05:00
if pkg_info -q -e $$pkg; then \
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
listlibs='echo ${DEPDIR}$$shdir/lib*'; \
special case: if PKGPATH is the LIB_DEPENDS base, then it means we're building a subpackage, and then we definitely want the `current' stuff, not the stuff that's already installed. Allows shared libs updates in MULTI_PACKAGES to proceed gracefully without having to uninstall stuff. (we do not do this stuff in general, because in other cases, builds are `staggered', e.g., a port is built against the existing base, not the stuff in other WRKDIRS)
2006-10-18 07:09:30 -04:00
case $$dir in ${PKGPATH}) \
listlibs="$$toset ${MAKE} print-plist-contents|${_grab_libs_from_plist}; $$listlibs";; \
esac; \
still need to look in the default package for multi-package cases. use -K that we now have...
2004-11-27 07:36:14 -05:00
else \
introduce new macros to make things more readable. Kill really old stuff: LD_CONFIG, that's no longer used.
2005-11-05 06:11:17 -05:00
listlibs="$$toset ${MAKE} print-plist-contents|${_grab_libs_from_plist}"; \
still need to look in the default package for multi-package cases. use -K that we now have...
2004-11-27 07:36:14 -05:00
fi; \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
IFS=,; for d in $$dep; do \
${_libresolve_fragment}; \
case "$$check" in \
*.a) continue;; \
* move the dependency computation out of PKG_ARGS, so that we can check for errors (and get rid of _NODEPS since it's not needed any more) * set check to Failed in *libresolve_fragment, so that we can test for it and report properly. guys here have no objection...
2005-11-02 15:28:11 -05:00
Failed) \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
echo 1>&2 "Can't resolve libspec $$d"; \
exit 1;; \
*) \
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
echo "-W $$check";; \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
esac; \
done; \
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
echo "-P $$subdir:$$pkg:$$default"; \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
}
. endfor
let LIB_DEPENDS and WANTLIB depend on the SUBPACKAGE. Build a special list of _BUILDLIB_DEPENDS, so that separate subpackages can depend on each other without making an infinite loop...
2006-11-20 05:49:22 -05:00
. for _i in ${WANTLIB${SUBPACKAGE}}
add missing quote, fix WANTLIB fuck-up
2006-08-04 19:29:47 -04:00
@d='${_i}'; listlibs='echo $$shdir/lib*'; \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
${_syslibresolve_fragment}; \
case "$$check" in \
overeager code cloning. noticed by matthieu@/mbalmer@
2005-10-07 17:08:16 -04:00
*.a) ;; \
* move the dependency computation out of PKG_ARGS, so that we can check for errors (and get rid of _NODEPS since it's not needed any more) * set check to Failed in *libresolve_fragment, so that we can test for it and report properly. guys here have no objection...
2005-11-02 15:28:11 -05:00
Failed) \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
echo 1>&2 "Can't resolve libspec $$d"; \
exit 1;; \
*) \
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
echo "-W $$check";; \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
esac
. endfor
.endif
tweak _print-package-args a bit. Add internal NODEPS to turn off computing dependencies. Write code similar to _print-package-args that is strictly ports based, and hence really walks the wantlibs, and needs the pkg_create stuff to get at the real ports contents. Recode print-package-signature to refer to the ports tree contents exclusively (pkg_info -S for the package). Thx bernd@ for trying that out. This should make for a slow, but accurate, out-of-date printer.
2005-09-24 15:46:56 -04:00
_list-port-libs:
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
.if defined(_PORT_LIBS_CACHE) && defined(_DEPENDS_CACHE) && defined(_DEPENDS_FILE)
@if ! fgrep -q -e "r|${FULLPKGPATH}|" -e "a|${FULLPKGPATH}" $${_DEPENDS_FILE}; then \
${MAKE} run-dir-depends >>${_DEPENDS_CACHE}; \
fi
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
@perl ${PORTSDIR}/infrastructure/build/extract-dependencies ${FULLPKGPATH} <${_DEPENDS_CACHE}|while read subdir; do \
fulldir=${_PORT_LIBS_CACHE}/$$subdir; \
don't bother building default package, just ask print-plist-contents for its contents. avoid dirname in print-package-signature, use sh built-ins.
2005-10-10 15:20:00 -04:00
if test -f $$fulldir; then \
cat $$fulldir; \
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
else \
don't bother building default package, just ask print-plist-contents for its contents. avoid dirname in print-package-signature, use sh built-ins.
2005-10-10 15:20:00 -04:00
mkdir -p $${fulldir%/*}; \
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
${_flavor_fragment}; \
* move the dependency computation out of PKG_ARGS, so that we can check for errors (and get rid of _NODEPS since it's not needed any more) * set check to Failed in *libresolve_fragment, so that we can test for it and report properly. guys here have no objection...
2005-11-02 15:28:11 -05:00
eval $$toset ${MAKE} print-plist-contents | \
introduce new macros to make things more readable. Kill really old stuff: LD_CONFIG, that's no longer used.
2005-11-05 06:11:17 -05:00
${_grab_libs_from_plist}|tee $$fulldir; \
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
fi; \
done
.else
introduce new macros to make things more readable. Kill really old stuff: LD_CONFIG, that's no longer used.
2005-11-05 06:11:17 -05:00
@${MAKE} run-dir-depends|${_sort_dependencies}|while read subdir; do \
tweak _print-package-args a bit. Add internal NODEPS to turn off computing dependencies. Write code similar to _print-package-args that is strictly ports based, and hence really walks the wantlibs, and needs the pkg_create stuff to get at the real ports contents. Recode print-package-signature to refer to the ports tree contents exclusively (pkg_info -S for the package). Thx bernd@ for trying that out. This should make for a slow, but accurate, out-of-date printer.
2005-09-24 15:46:56 -04:00
${_flavor_fragment}; \
* move the dependency computation out of PKG_ARGS, so that we can check for errors (and get rid of _NODEPS since it's not needed any more) * set check to Failed in *libresolve_fragment, so that we can test for it and report properly. guys here have no objection...
2005-11-02 15:28:11 -05:00
eval $$toset ${MAKE} print-plist-contents ; \
introduce new macros to make things more readable. Kill really old stuff: LD_CONFIG, that's no longer used.
2005-11-05 06:11:17 -05:00
done | ${_grab_libs_from_plist}
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
.endif
@echo /usr/lib/lib* ${X11BASE}/lib/lib*
tweak _print-package-args a bit. Add internal NODEPS to turn off computing dependencies. Write code similar to _print-package-args that is strictly ports based, and hence really walks the wantlibs, and needs the pkg_create stuff to get at the real ports contents. Recode print-package-signature to refer to the ports tree contents exclusively (pkg_info -S for the package). Thx bernd@ for trying that out. This should make for a slow, but accurate, out-of-date printer.
2005-09-24 15:46:56 -04:00
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
_print-package-signature-run:
also use RUN_DEPENDS${SUBPACKAGE} in the simple parts.
2006-11-19 07:37:55 -05:00
.for _i in ${RUN_DEPENDS${SUBPACKAGE}}
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
@echo '${_i}' |{ \
IFS=:; read dep pkg subdir target; \
${_flavor_fragment}; \
tweak _print-package-args a bit. Add internal NODEPS to turn off computing dependencies. Write code similar to _print-package-args that is strictly ports based, and hence really walks the wantlibs, and needs the pkg_create stuff to get at the real ports contents. Recode print-package-signature to refer to the ports tree contents exclusively (pkg_info -S for the package). Thx bernd@ for trying that out. This should make for a slow, but accurate, out-of-date printer.
2005-09-24 15:46:56 -04:00
default=`eval $$toset ${MAKE} _print-packagename`; \
echo "$$default"; \
}
.endfor
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
_print-package-signature-lib:
forgot one DEPRUNLIBS
2006-11-20 05:38:31 -05:00
@echo $$LIST_LIBS| LOCALBASE=${LOCALBASE} X11BASE=${X11BASE} perl ${PORTSDIR}/infrastructure/build/resolve-lib ${_DEPRUNLIBS:S/>/\>/g}
let LIB_DEPENDS and WANTLIB depend on the SUBPACKAGE. Build a special list of _BUILDLIB_DEPENDS, so that separate subpackages can depend on each other without making an infinite loop...
2006-11-20 05:49:22 -05:00
.for _i in ${LIB_DEPENDS${SUBPACKAGE}}
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
@echo '${_i}' |{ \
IFS=:; read dep pkg subdir target; \
${_flavor_fragment}; \
tweak _print-package-args a bit. Add internal NODEPS to turn off computing dependencies. Write code similar to _print-package-args that is strictly ports based, and hence really walks the wantlibs, and needs the pkg_create stuff to get at the real ports contents. Recode print-package-signature to refer to the ports tree contents exclusively (pkg_info -S for the package). Thx bernd@ for trying that out. This should make for a slow, but accurate, out-of-date printer.
2005-09-24 15:46:56 -04:00
default=`eval $$toset ${MAKE} _print-packagename`; \
echo "$$default"; \
}
tweak the way library dependencies are resolved to speed them up. Now, resolve-lib can take a big list of libraries with full paths, and it can solve a big list of spec at once. Basically, we move most of the parsing of spec paths into resolve-lib. Since print-package-signature does build a full list of libs, let's solve it all at once, instead of invoking a costly perl script repeatedly. Add some caching possibilities for out-of-date. Specifically: - store libraries for each package under the directory _PORT_LIBS_CACHE - use the dependency cache _DEPENDS_FILE to avoid recreating dependency chains, add a new file _DEPENDS_CACHE that will accumulate all dependencies, and extract these with a simple script extract-dependencies. Use echo to build libraries lists instead of ls, that's a bit simpler... Some more clean-up will happen: it's probably simpler to parse libspecs at once, extract the libraries needed and go fetch the corresponding libraries just once.
2005-10-09 08:01:22 -04:00
.endfor
tweak _print-package-args a bit. Add internal NODEPS to turn off computing dependencies. Write code similar to _print-package-args that is strictly ports based, and hence really walks the wantlibs, and needs the pkg_create stuff to get at the real ports contents. Recode print-package-signature to refer to the ports tree contents exclusively (pkg_info -S for the package). Thx bernd@ for trying that out. This should make for a slow, but accurate, out-of-date printer.
2005-09-24 15:46:56 -04:00
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
# recursively build a list of dirs for package running, ready for tsort
_recurse-run-dir-depends:
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
.for _dir in ${_RUN_DEP}
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
@echo "$$self ${_dir}"; \
tag cache entries with b(uild), r(un), a(ll). Fixes a rare logic error, where recursive build depends would not trigger all depends because an existing build depends was taken instead.
2004-07-24 09:53:12 -04:00
if ! fgrep -q -e "r|${_dir}|" -e "a|${_dir}|" $${_DEPENDS_FILE}; then \
echo "r|${_dir}|" >> $${_DEPENDS_FILE}; \
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
subdir=${_dir}; ${_flavor_fragment}; \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
toset="$$toset self=\"${_dir}\""; \
if ! eval $$toset ${MAKE} _recurse-run-dir-depends; then \
Move the `check and change directory' part to flavor_fragment, as it's invariably duplicated right after otherwise. This makes the code cleaner, and makes sure the check happens all the time.
2001-11-12 09:14:46 -05:00
echo 1>&2 "*** Problem checking deps in \"$$dir\"."; \
New Features for NEW_DEPENDS: dir-depends: list of SUBDIRS this port depends on (with flavors, subpackages), to use with tsort. build-depends-list, run-depends-list: list of pkgspec:dir needed for this port, no recursion. To use for INDEX.
2001-04-04 04:17:31 -04:00
exit 1; \
fi; \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
fi
.endfor
run-dir-depends:
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
.if !empty(_RUN_DEP)
Use trap to remove _DEPENDS_FILE, so that it does not stay around. Since all the code is now in one chunk, put it into _depfile_fragment.
2003-08-04 10:00:46 -04:00
@${_depfile_fragment}; \
tag cache entries with b(uild), r(un), a(ll). Fixes a rare logic error, where recursive build depends would not trigger all depends because an existing build depends was taken instead.
2004-07-24 09:53:12 -04:00
if ! fgrep -q -e "r|${FULLPKGPATH}|" -e "a|${FULLPKGPATH}" $${_DEPENDS_FILE}; then \
echo "r|${FULLPKGPATH}|" >>$${_DEPENDS_FILE}; \
set PACKAGING in RUN_DEPENDS, to avoid picking up bogus subpackages dependencies. Tests by sturm@, noticed by pval@
2003-09-28 06:57:01 -04:00
self=${FULLPKGPATH} PACKAGING='${SUBPACKAGE}' ${MAKE} _recurse-run-dir-depends; \
move _depfile_fragment to pkgpath.mk so that it can be reused more widely. Fix the logic so that the file is only removed where it is set. Add some checks to *-dir-depends for more `global situations'.
2003-08-04 10:37:10 -04:00
fi
New Features for NEW_DEPENDS: dir-depends: list of SUBDIRS this port depends on (with flavors, subpackages), to use with tsort. build-depends-list, run-depends-list: list of pkgspec:dir needed for this port, no recursion. To use for INDEX.
2001-04-04 04:17:31 -04:00
.else
FULLPKGPATH is PKGPATH plus flavor/multipackage modifiers. Use it where we can, including WRKOBJDIR `real' names. NOTE: if you are using WRKOBJDIR, *all* flavored builds directories will change names, from work-no_x11 to work,no_x11
2001-04-08 12:55:17 -04:00
@echo "${FULLPKGPATH} ${FULLPKGPATH}"
New Features for NEW_DEPENDS: dir-depends: list of SUBDIRS this port depends on (with flavors, subpackages), to use with tsort. build-depends-list, run-depends-list: list of pkgspec:dir needed for this port, no recursion. To use for INDEX.
2001-04-04 04:17:31 -04:00
.endif
add regress-dir-depends/full-regress-depends, similar mechanism to run-depends, only the basic test changes.
2006-11-05 10:46:51 -05:00
# recursively build a list of dirs for package regression, ready for tsort
_recurse-regress-dir-depends:
.for _dir in ${_REGRESS_DEP}
@echo "$$self ${_dir}"; \
if ! fgrep -q -e "R|${_dir}|" $${_DEPENDS_FILE}; then \
echo "R|${_dir}|" >> $${_DEPENDS_FILE}; \
subdir=${_dir}; ${_flavor_fragment}; \
toset="$$toset self=\"${_dir}\""; \
if ! eval $$toset ${MAKE} _recurse-run-dir-depends; then \
echo 1>&2 "*** Problem checking deps in \"$$dir\"."; \
exit 1; \
fi; \
fi
.endfor
regress-dir-depends:
.if !empty(_REGRESS_DEP)
@${_depfile_fragment}; \
if ! fgrep -q -e "R|${FULLPKGPATH}|" $${_DEPENDS_FILE}; then \
echo "R|${FULLPKGPATH}|" >>$${_DEPENDS_FILE}; \
self=${FULLPKGPATH} PACKAGING='${SUBPACKAGE}' ${MAKE} _recurse-regress-dir-depends; \
fi
.else
@echo "${FULLPKGPATH} ${FULLPKGPATH}"
.endif
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
# recursively build a list of dirs for package building, ready for tsort
# second and further stages need _RUN_DEP.
_recurse-all-dir-depends:
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
.for _dir in ${_BUILD_DEP} ${_RUN_DEP}
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
@echo "$$self ${_dir}"; \
tag cache entries with b(uild), r(un), a(ll). Fixes a rare logic error, where recursive build depends would not trigger all depends because an existing build depends was taken instead.
2004-07-24 09:53:12 -04:00
if ! fgrep -q "a|${_dir}|" $${_DEPENDS_FILE}; then \
echo "a|${_dir}|" >> $${_DEPENDS_FILE}; \
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
subdir=${_dir}; ${_flavor_fragment}; \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
toset="$$toset self=\"${_dir}\""; \
if ! eval $$toset ${MAKE} _recurse-all-dir-depends; then \
Move the `check and change directory' part to flavor_fragment, as it's invariably duplicated right after otherwise. This makes the code cleaner, and makes sure the check happens all the time.
2001-11-12 09:14:46 -05:00
echo 1>&2 "*** Problem checking deps in \"$$dir\"."; \
Misc. small tweaks. - let USE_MOTIF be sensible: any, openmotif, lesstif. any triggers a lesstif flavor. - show=name as a shortcut. - package-dir-depends to check that packages are okay. - error message for non-existent dirs.
2001-04-18 10:43:55 -04:00
exit 1; \
fi; \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
fi
.endfor
Misc. small tweaks. - let USE_MOTIF be sensible: any, openmotif, lesstif. any triggers a lesstif flavor. - show=name as a shortcut. - package-dir-depends to check that packages are okay. - error message for non-existent dirs.
2001-04-18 10:43:55 -04:00
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
# first stage does not need _RUN_DEP
_build-dir-depends:
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
.for _dir in ${_BUILD_DEP}
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
@echo "$$self ${_dir}"; \
tag cache entries with b(uild), r(un), a(ll). Fixes a rare logic error, where recursive build depends would not trigger all depends because an existing build depends was taken instead.
2004-07-24 09:53:12 -04:00
if ! fgrep -q -e "b|${_dir}|" -e "a|${_dir}|" $${_DEPENDS_FILE}; then \
echo "b|${_dir}|" >> $${_DEPENDS_FILE}; \
streamline flavor_fragment after trying to explain it to sturm@. it makes more sense to have a subdir variable as input, split it into toset and dir as output, and to unsetenv FLAVOR SUBPACKAGE along the lines. End result should be equivalent, but slightly more readable. no objection from my neighbors either...
2005-11-02 15:30:12 -05:00
subdir=${_dir}; ${_flavor_fragment}; \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
toset="$$toset self=\"${_dir}\""; \
if ! eval $$toset ${MAKE} _recurse-all-dir-depends; then \
Move the `check and change directory' part to flavor_fragment, as it's invariably duplicated right after otherwise. This makes the code cleaner, and makes sure the check happens all the time.
2001-11-12 09:14:46 -05:00
echo 1>&2 "*** Problem checking deps in \"$$dir\"."; \
Misc. small tweaks. - let USE_MOTIF be sensible: any, openmotif, lesstif. any triggers a lesstif flavor. - show=name as a shortcut. - package-dir-depends to check that packages are okay. - error message for non-existent dirs.
2001-04-18 10:43:55 -04:00
exit 1; \
fi; \
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
fi
.endfor
build-dir-depends:
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
.if !empty(_BUILD_DEP)
Use trap to remove _DEPENDS_FILE, so that it does not stay around. Since all the code is now in one chunk, put it into _depfile_fragment.
2003-08-04 10:00:46 -04:00
@${_depfile_fragment}; \
tag cache entries with b(uild), r(un), a(ll). Fixes a rare logic error, where recursive build depends would not trigger all depends because an existing build depends was taken instead.
2004-07-24 09:53:12 -04:00
if ! fgrep -q -e "b|${FULLPKGPATH}|" -e "a|${_dir}|" $${_DEPENDS_FILE}; then \
echo "b|${FULLPKGPATH}|" >>$${_DEPENDS_FILE}; \
move _depfile_fragment to pkgpath.mk so that it can be reused more widely. Fix the logic so that the file is only removed where it is set. Add some checks to *-dir-depends for more `global situations'.
2003-08-04 10:37:10 -04:00
self=${FULLPKGPATH} ${MAKE} _build-dir-depends; \
fi
Misc. small tweaks. - let USE_MOTIF be sensible: any, openmotif, lesstif. any triggers a lesstif flavor. - show=name as a shortcut. - package-dir-depends to check that packages are okay. - error message for non-existent dirs.
2001-04-18 10:43:55 -04:00
.else
@echo "${FULLPKGPATH} ${FULLPKGPATH}"
.endif
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
all-dir-depends:
simplify xxx-dir-depends: only rely on _BUILD_DEP/_RUN_DEP. Fold _ALWAYS_DEP.
2006-11-20 04:59:11 -05:00
.if !empty(_BUILD_DEP) || !empty(_RUN_DEP)
Use trap to remove _DEPENDS_FILE, so that it does not stay around. Since all the code is now in one chunk, put it into _depfile_fragment.
2003-08-04 10:00:46 -04:00
@${_depfile_fragment}; \
move _depfile_fragment to pkgpath.mk so that it can be reused more widely. Fix the logic so that the file is only removed where it is set. Add some checks to *-dir-depends for more `global situations'.
2003-08-04 10:37:10 -04:00
if ! fgrep -q "|${FULLPKGPATH}|" $${_DEPENDS_FILE}; then \
echo "|${FULLPKGPATH}|" >>$${_DEPENDS_FILE}; \
self=${FULLPKGPATH} ${MAKE} _recurse-all-dir-depends; \
fi
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
.else
@echo "${FULLPKGPATH} ${FULLPKGPATH}"
Basic new lib dependency code for package. LIB_DEPENDS are no longer part of ALWAYS_DEPEND for the new-depends code, but use their own code. The code does scan the list of libraries for dependent packages, and insert corresponding @libdepend lines into the resulting package. There are a few important consequences: - no libdepend lines are inserted if only a static library is found. So RUN_DEPENDS are now needed to supplement LIB_DEPENDS if a package requires another package for something that isn't a library. - dependency checking for installed stuff can go one step further, since we have the major/minor number of the libraries used for the build. At the moment, pkg resolve dependencies does nothing smart with the inserted libdepends, but it will (soon). In fact, for most libdepends, if the major/minor scheme are correct, no check on the installed pkgspec ought to be necessary... which is why the `default pkgspec to check' is pkg-* for libraries. - resolving recursive run dependencies need to go one step further: a RUN_DEPENDS pkg may have some LIB_DEPENDS, and the RUN_DEPENDS pkg will be needed to explicitly write the required information into the resulting package... Hence the necessity of being able to tune the list of libraries to ask the uninstalled package (and to make sure the uninstalled package is built). This is only preliminary work. Currently, this doesn't yield any real benefits to the old approach, as a few semantic details need to be sharpened out. Also, there are now quite a few old targets that are a complete mess and will need to be cleaned up/removed entirely. Thanks to naddy@ for testing various preliminary versions of this patch and helping me iron bugs out.
2001-11-12 09:32:52 -05:00
.endif
link-categories/unlink-categories target: populate ports tree with symlinks.
2000-12-16 10:49:12 -05:00
link-categories:
.for _CAT in ${CATEGORIES}
@linkname=${PORTSDIR}/${_CAT}/`basename ${.CURDIR}`; \
if [ ! -e $$linkname ]; then \
echo "$$linkname -> ${.CURDIR}"; \
mkdir -p ${PORTSDIR}/${_CAT}; \
ln -s ${.CURDIR} $$linkname; \
fi
.endfor
unlink-categories:
.for _CAT in ${CATEGORIES}
@linkname=${PORTSDIR}/${_CAT}/`basename ${.CURDIR}`; \
if [ -L $$linkname ]; then \
echo "rm $$linkname"; \
rm $$linkname; \
if rmdir ${PORTSDIR}/${_CAT} 2>/dev/null; then \
echo "rmdir ${PORTSDIR}/${_CAT}"; \
fi; \
fi
.endfor
switch to emacs Makefile mode, at the request of Han Boetes. Kill extra ending spaces.
2003-07-18 15:02:13 -04:00
Add rules to create a list of homepage links for all ports.
2002-03-13 08:51:59 -05:00
homepage-links:
.if defined(HOMEPAGE)
@echo '<li><A HREF="${HOMEPAGE}">${PKGNAME}</A>'
.else
@echo '<li>${PKGNAME}'
.endif
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
#####################################################
# convenience targets, not really needed
#####################################################
checkpatch:
@cd ${.CURDIR} && exec ${MAKE} PATCH_CHECK_ONLY=Yes patch
clean-depends:
pull a few more targets out of user-visible space. Sprinkle a few more cd ${.CURDIR} && exec move distclean into convenience targets area.
2003-07-30 15:59:48 -04:00
@cd ${.CURDIR} && exec ${MAKE} clean=depends
distclean:
@cd ${.CURDIR} && exec ${MAKE} clean=dist
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
delete-package:
@cd ${.CURDIR} && exec ${MAKE} clean=package
reinstall:
* force uninstall in reinstall * fix typo in clean=install handling discussed with sturm@
2003-08-28 19:42:44 -04:00
@cd ${.CURDIR} && exec ${MAKE} clean='install force'
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
@cd ${.CURDIR} && DEPENDS_TARGET=${DEPENDS_TARGET} exec ${MAKE} install
repackage:
@cd ${.CURDIR} && exec ${MAKE} clean=packages
@cd ${.CURDIR} && exec ${MAKE} package
rebuild:
@rm -f ${_BUILD_COOKIE}
@cd ${.CURDIR} && exec ${MAKE} build
uninstall deinstall:
@${ECHO_MSG} "===> Deinstalling for ${FULLPKGNAME${SUBPACKAGE}}"
remove "-f" from calls to pkg_delete, fixes "make uninstall" ok espie@
2004-01-28 15:16:14 -05:00
@${SUDO} ${PKG_DELETE} ${FULLPKGNAME${SUBPACKAGE}}
clean-up and speed-up dependency generation. thanks to nikolay, naddy, and others for comments.
2003-07-30 15:31:31 -04:00
move error handling later for greater flexibility. Move default NO_* after including local Makefiles.
2001-10-03 04:41:16 -04:00
.if defined(ERRORS)
.BEGIN:
. for _m in ${ERRORS}
tag error messages with PKGPATH, as they get hard to decipher when walking the tree.
2004-08-03 17:18:24 -04:00
@echo 1>&2 ${_m} "(in ${PKGPATH})"
move error handling later for greater flexibility. Move default NO_* after including local Makefiles.
2001-10-03 04:41:16 -04:00
. endfor
Handle ERRORS+='Fatal:...' as well
2002-03-17 20:52:46 -05:00
. if !empty(ERRORS:M"Fatal\:*") || !empty(ERRORS:M'Fatal\:*')
move error handling later for greater flexibility. Move default NO_* after including local Makefiles.
2001-10-03 04:41:16 -04:00
@exit 1
. endif
.endif
convenience target for porters: peek-ftp connects to the first MASTER_SITES and leaves you at the ftp prompt in the directory where stuff is fetched from. (I find myself using this quite often to figure out which version of stuff exists, especially for CPAN).
2006-07-01 07:08:05 -04:00
peek-ftp:
let peek-ftp cd ${FULLDISTDIR} first. Add a message to that effect so that users don't get confused. (specifically, make itself does not change dirs, so if you do ^Z, you're not where you think you are...)
2006-10-17 18:32:45 -04:00
@mkdir -p ${FULLDISTDIR}; cd ${FULLDISTDIR}; echo "cd ${FULLDISTDIR}"; \
for i in ${MASTER_SITES:Mftp*}; do echo "Connecting to $$i"; ${FETCH_CMD} $$i ; break; done
convenience target for porters: peek-ftp connects to the first MASTER_SITES and leaves you at the ftp prompt in the directory where stuff is fetched from. (I find myself using this quite often to figure out which version of stuff exists, especially for CPAN).
2006-07-01 07:08:05 -04:00
new target, show-required-by, that can give you a full list of everything a port will affect. name by mbalmer@
2005-11-01 09:04:21 -05:00
show-required-by:
@cd ${PORTSDIR} && make all-dir-depends | perl ${PORTSDIR}/infrastructure/build/extract-dependencies -r ${FULLPKGPATH}
introspection mechanism: dump-vars
2006-07-09 07:10:16 -04:00
show:
.for _s in ${show}
@echo ${${_s}:Q}
.endfor
verbose-show:
.for _s in ${verbose-show}
. if defined(${_s})
@echo ${_s}=${${_s}:Q}
. endif
.endfor
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
subdump-vars:
zap a few non relevant lines, make _ALL_VARIABLES* internal. Use TMPDIR in pkgpath.mk
2006-11-01 07:41:34 -05:00
.for _s in ${_ALL_VARIABLES}
introspection mechanism: dump-vars
2006-07-09 07:10:16 -04:00
. if defined(${_s})
@echo ${FULLPKGPATH}.${_s}=${${_s}:Q}
. endif
.endfor
add an extra level of indirection for dependency lists: _xxx_DEPLIST. allows us to get rid of some NO_DEPENDS tests later, also allows for RUN_DEPENDS to become indexed on SUBPACKAGE. LIB_DEPENDS and WANTLIB are going to need more surgery, since they need to be checked during build and packaging...
2006-11-19 07:32:53 -05:00
.for _s in ${_ALL_VARIABLES_INDEXED}
cheat a bit: make sure the right variables are used for MULTI_PACKAGES.
2006-10-09 14:11:38 -04:00
. if defined(${_s}${SUBPACKAGE})
@echo ${FULLPKGPATH}.${_s}=${${_s}${SUBPACKAGE}:Q}
. endif
.endfor
introspection mechanism: dump-vars
2006-07-09 07:10:16 -04:00
Declare phony target as phony targets. This may avert a few subtle bugs, and it documents behavior as well.
1999-02-24 07:34:46 -05:00
.PHONY: \
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
_build-dir-depends _fetch-makefile _fetch-onefile \
_print-packagename \
remove source-based lib-depends-check. We always check the package now.
2005-01-31 04:58:22 -05:00
_recurse-all-dir-depends \
kill code that is no longer used.
2004-11-27 09:07:19 -05:00
_recurse-run-dir-depends _refetch \
new support routine _print-package-args, that prints new-style package dependencies. _syslib_resolve_fragment that can resolve system libs.
2004-11-21 05:43:48 -05:00
addsum _print-package-args \
redo the list of phony targets, in alphabetical order
2003-07-30 15:51:11 -04:00
all all-dir-depends build \
build-depends build-depends-list build-dir-depends \
checkpatch checksum clean \
clean-depends configure deinstall \
pull a few more targets out of user-visible space. Sprinkle a few more cd ${.CURDIR} && exec move distclean into convenience targets area.
2003-07-30 15:59:48 -04:00
delete-package \
depends \
redo the list of phony targets, in alphabetical order
2003-07-30 15:51:11 -04:00
describe distclean distpatch \
do-build do-configure do-distpatch \
do-extract do-fetch do-install \
do-package do-regress extract \
fake fetch fetch-all \
fetch-makefile full-all-depends full-build-depends \
full-run-depends homepage-links install \
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
lib-depends lib-depends-check lib-depends-list \
redo the list of phony targets, in alphabetical order
2003-07-30 15:51:11 -04:00
link-categories makesum manpages-check \
pull a few more targets out of user-visible space. Sprinkle a few more cd ${.CURDIR} && exec move distclean into convenience targets area.
2003-07-30 15:59:48 -04:00
package patch \
redo the list of phony targets, in alphabetical order
2003-07-30 15:51:11 -04:00
plist post-build post-configure \
post-distpatch post-extract post-fetch \
post-install post-package post-patch \
post-regress pre-build pre-configure \
pre-extract pre-fake pre-fetch \
pre-install pre-package pre-patch \
- new read-only variable, PKGNAMES. - plist repository under PLIST_DB (optional). - print-package-signatures shows what's used for signatures. - make fetch shows full url you can copy/paste, simpler to look for typos that way.
2005-09-04 18:32:37 -04:00
pre-regress print-build-depends print-package-signature \
print-run-depends \
readmes is yet another target that should recurse.
2006-09-19 16:10:40 -04:00
readmes _readme rebuild \
pull a few more targets out of user-visible space. Sprinkle a few more cd ${.CURDIR} && exec move distclean into convenience targets area.
2003-07-30 15:59:48 -04:00
regress regress-depends \
redo the list of phony targets, in alphabetical order
2003-07-30 15:51:11 -04:00
reinstall repackage run-depends \
introspection mechanism: dump-vars
2006-07-09 07:10:16 -04:00
run-depends-list run-dir-depends show verbose-show dump-vars \
redo the list of phony targets, in alphabetical order
2003-07-30 15:51:11 -04:00
uninstall unlink-categories update-patches \
advisory license-check: warns if a package is apparently built from bad sources. Will returns some false positives.
2003-08-11 16:10:41 -04:00
update-plist \
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
license-check _license-check \
_internal-extract _internal-distpatch _internal-configure \
_internal-build _internal-all _internal_install _internal-fake \
fix a bunch of issues: - make sure locks happen when they should, including for lib-depends-check. - make sure various targets set PACKAGING as they should. - let BULK=Yes be invoked only from make package. Fix for the BULK_COOKIE and UPDATE_COOKIE issue. They can no longer depend on PACKAGE_COOKIES, since this needs to get built with PACKAGING set. So instead, simply remove the cookies when we create a new package succesfully. By any kind of reasoning, both bulk and update are `reset' when a new package appears, so this seems to be the right semantics in most cases. Thanks for wilfried and nikolay for error spotting. Hoping this will be what's needed, finally.
2006-10-23 09:42:50 -04:00
_internal-package _internal-package-only \
_internal-fetch _internal-checksum \
start making a big distinction between lib depends evaluated at build time and at `run(packaging)' time.
2006-11-19 11:39:41 -05:00
_internal-depends _internal-lib-depends _internal-buildwantlib-depends \
_internal-runwantlib-depends \
_internal-buildlib-depends _internal_runlib-depends \
better semantics for library checking: just check the pkgspec for recursive building, and then do one big check of all specs in LIB_DEPENDS and wantlib. This should be faster and more accurate.
2005-10-09 09:31:50 -04:00
_internal-build-depends \
Lock infrastructure for the ports tree. If LOCK_CMD, UNLOCK_CMD and LOCK_DIR are defined, those are used to perform `big-lock' style locking on top-level targets, such as extract, patch, build. The internals of the ports tree do not use any finer grained locking. Those top-level targets now redirect to _internal-targets, without any behavioral change. Any dependency computation will recurse to another directory, and invoke a top-level target, thus triggering the locking of the dependency. All locking is done using FULLPKGNAME, except for fetch with uses the DISTFILES names for independent files. If no locking is desired, the top-level targets simply redirect to the _internal-targets. The cost is close to zero: make just needs to handle an extra ~20 phony targets. Much testing and approval by brad@, naddy@, pval@, fries@ and other people. Thanks to niklas@ for some useful discussion.
2004-07-18 18:44:36 -04:00
_internal-run-depends _internal-regress-depends \
_internal-regress _internal-clean _internal-lib-depends-check \
newlib-depends-check that takes the semantics-to-be into account.
2004-11-21 06:26:22 -05:00
_internal-newlib-depends-check \
Add a simple update target, that relies on pkg_info -e pkgpath to know if there is something to update...
2004-11-10 05:16:41 -05:00
_internal-manpages-check _internal-plist _internal-update-plist \
tweak _print-package-args a bit. Add internal NODEPS to turn off computing dependencies. Write code similar to _print-package-args that is strictly ports based, and hence really walks the wantlibs, and needs the pkg_create stuff to get at the real ports contents. Recode print-package-signature to refer to the ports tree contents exclusively (pkg_info -S for the package). Thx bernd@ for trying that out. This should make for a slow, but accurate, out-of-date printer.
2005-09-24 15:46:56 -04:00
_internal-update update print-plist print-plist-contents \
new target, show-required-by, that can give you a full list of everything a port will affect. name by mbalmer@
2005-11-01 09:04:21 -05:00
_list-port-libs _print-package-signature-lib _print-package-signature-run \
somewhat big change: zap the package-link/package-unlink stuff, turn these into real targets. - now make package will create missing ftp/cdrom links when PERMIT says so. - PKG_ARCH=* packages get built into packages/no-arch, and linked from elsewhere. Everything gets based off PACKAGE_REPOSITORY, PKGREPOSITORYBASE and friends get ditched. All `package' targets go through the same routine: iterate through every subpackage with PACKAGING set. This includes describe, lib-depends-check, package, dump-vars, install-all, which will iterate through subdescribe, sublib-depends-check, subpackage, subdump-vars, and install. (names are subject to change). Much simpler logic, plus hey, you get an install-all target ! Might be some minor breakage, most stuff appear to work just fine...
2006-09-17 18:08:19 -04:00
show-required-by peek-ftp _internal-subpackage subdescribe \
sublib-depends-check subdump-vars _internal-install-all install-all \
add regress-dir-depends/full-regress-depends, similar mechanism to run-depends, only the basic test changes.
2006-11-05 10:46:51 -05:00
subpackage _internal-subupdate _internal-update \
regress-dir-depends _recurse-regress-dir-depends \
shorten dependency names, so that regress and run can share. add print-plist-all, for better introspection (to be used by pkg_mklocatedb).
2006-11-11 11:10:36 -05:00
full-regress-depends print-plist-all
Reference in New Issue Copy Permalink