Commit Graph

1184 Commits

Author SHA1 Message Date
espie
083fc635e8 fix subpackage bugs.
- correct syntax for variable (Vadim Zhukov)
- both _DO_LOCK and _cache_fragment want to use traps.
Since that's the only place where the problem occurs, simply put the second
trap in a subshell...
2012-08-17 22:32:29 +00:00
espie
3831fd5d30 - make SUBST_CMD be subpackage dependent, fixes substitution problems in
README-sub (as noticed by aja@)
- stronger checks for X correctly installed: don't ignore ports if X11
is not there, error out right away. Make sure /usr/local/lib/X11/app-defaults
is a link, and that whatis.db is there (as should be fixed  by release in
xenocara)
2012-08-04 14:23:30 +00:00
naddy
0d7d591395 Remove message digest algorithms other than SHA-256 for checksumming
distfiles.  MD5 is known to be insecure and RIPEMD-160 and SHA-1
are considered inferior to SHA-256.

Also, the concatenation of different hashes is not more secure than
its strongest component; see Antoine Joux, "Multicollisions in
iterated hash functions. Application to cascased constructions"
http://www.iacr.org/cryptodb/archive/2004/CRYPTO/1472/1472.pdf

Discussed with many, ok sthen@
2012-07-10 12:04:25 +00:00
espie
a352e9cafe prepare for new pkg_add switch 2012-07-06 12:00:52 +00:00
espie
134ac2cd91 allow users to say MESSAGE-main=
to have empty messages, useful in a multi-package context
(as noticed by aja@)
2012-06-20 13:26:17 +00:00
espie
7b5d36d556 switch to libtool from src, since it will now be in the snaps. 2012-06-19 16:43:47 +00:00
espie
5a916fc963 handle the very special case of ports-readmes, after discussion with
landry, sthen
2012-06-18 12:15:52 +00:00
espie
6ffbfdc3a9 fix sig for LIB_DEPENDS 2012-06-15 10:31:52 +00:00
espie
f217f9b069 match src/ change.
*keep ports and src in synch* or out-of-date won't work at all.
2012-06-14 13:34:24 +00:00
espie
d808116161 Use :Q, it will quote a lot more in a correct way. 2012-06-09 21:14:43 +00:00
sthen
8f529f75b3 quote the permit_cdrom and permit_ftp variables used in the new checksum
code, as they can contain spaces.  ok espie@
2012-06-08 23:51:21 +00:00
espie
98a72af47d typo 2012-06-08 15:17:02 +00:00
espie
3294a50c3d add support for sha256 packages as they're built.
after discussion with aja@
2012-06-08 14:46:54 +00:00
espie
482f234b05 this is tricky enough to warrant an actual comment 2012-05-28 09:54:18 +00:00
espie
b6c9f9baf3 zap old readmes infrastructure, replaced by new age databases/ports-readmes
which is ways more accurate anyways.
2012-05-28 09:38:04 +00:00
espie
d19515c7c3 make it possible for print-plist-with-depends to control lib-depends entirely
(useful for future update-plist)
2012-05-26 12:04:35 +00:00
jasper
fbebf9fd0d don't append ccache to BUILD_DEPENDS if NO_BUILD is set to Yes
ok sthen@
2012-05-16 06:36:56 +00:00
jasper
ec251be3fa "fix" a line which emacs loudly complains about...
ok espie@
2012-05-08 17:38:21 +00:00
halex
1fbdcad4e3 call exit 1 after cleanup in signal traps, or the shell keeps running after
SIGINT and friends

ok espie@
2012-05-07 21:11:43 +00:00
ajacoutot
9eeb9a1c1b Set localstatedir (i.e. BASELOCALSTATEDIR) to /var by default if
CONFIGURE_STYLE is gnu.
Works the same way as SYSCONFDIR, one can append a subdirectory to
change the default localstatedir by using the following construct in the
port Makefile:
LOCALSTATEDIR=	${BASELOCALSTATEDIR}/foobar

Note that this variable in not substituted in PLIST.

bulk tested by landry (on a previous diff)
ok jasper@ sthen@
2012-04-28 10:50:35 +00:00
espie
c1c992e6a5 dump-vars should show VMEM_WARNING 2012-04-22 10:39:48 +00:00
espie
390cba80df handle modbuild (perl module) stuff similarly to what gnu does.
there's no actual reason not to do it.
2012-04-17 09:31:47 +00:00
espie
b411a2917d strip PORTSDIR (or equivalent) from DESCR the same way we compute PKGPATH.
avoids random changes of INDEX depending on where the ports tree is installed
and configured.
2012-04-16 09:21:39 +00:00
espie
f456512a27 error out if PORTSDIR is not set to the real directory.
There are at least two or three ports that don't like symlinks. ;/
2012-03-22 14:08:50 +00:00
espie
aa4fe831d9 give FULLPKGPATH in warning message. Redundant, but may be useful for multi. 2012-02-23 08:32:17 +00:00
espie
bf99288338 scrap old fetch targets
add a warning when LIB_DEPENDS doesn't register (for sthen@)
tweak FLAVOR message for consistency
2012-02-17 07:33:04 +00:00
espie
baeec434e0 PKGFILES, why not ? 2012-01-29 11:29:51 +00:00
espie
16fd5981f1 now that update sees all pkgpaths, we can still remove duplicates. 2012-01-28 08:39:40 +00:00
espie
9971f34ba5 prevent some errors I've run into recently
- add ERRORS framework to bsd.port.subdir.mk
- trying to add a FLAVOR or a SUBPACKAGE to an intermediate SUBDIR is an
error (see editors/vim-spell,af recently)
- re-including bsd.port.mk/bsd.port.subdir.mk after either of them is an
error (can happen when one moves stuff to Makefile.inc without really thinking
about it, ends up with PKG_ARGS holding some contents twice and make package
erroring out with duplicate contents)
2012-01-21 14:44:40 +00:00
espie
0739a30939 pass PERMIT_DISTFILES* to dpb fetch 2012-01-14 12:22:07 +00:00
espie
2fe091cc70 recognize pkgpath changes when updating 2012-01-14 12:21:13 +00:00
miod
95cad4b2a7 In MAKE_ENV, quote PICFLAG as it may contain spaces. Other variables might
require similar quoting in the future, but so far this is enough to repair
port building operation on m68k.
ok espie@
2011-12-21 05:16:04 +00:00
espie
0603716dc7 - tweak CONFIG_SITE_LIST to run under infra/db.
- recognize absence of dependencies on gsed, gtar, ggrep,
and prevent more tests from seeing these.
2011-12-10 11:15:16 +00:00
espie
949fe672d3 make it possible to assemble a CONFIG_SITE instead of having one
single file.
approved by aja@, sthen@, jasper@...
2011-12-09 09:42:44 +00:00
espie
ab6b512da4 move the architecture *constants* to a private arch-defines.mk file
That way, bsd.port.arch.mk can still make available in a Makefile, but
we can also include them early in bsd.port.mk, so that they will always
be available from modules and Makefile.inc.
2011-12-08 08:15:36 +00:00
espie
1c7aab78c9 sort -u requires one parameter to not hang on stdin...
allow SUBDIR=archivers/arc make show-prepare-results to work
2011-12-02 15:14:20 +00:00
sthen
e7fe871f2c replace "!defined(MASTER_SITE_OVERRIDE)" check for MASTER_SITES[0..9] with the
new ${MASTER_SITE_OVERRIDE:L} == "no" construct, problem reported by naddy@.
2011-11-27 21:04:34 +00:00
espie
03d1bfec7f generate a tempfile, THEN move to saved_libs if things are okay 2011-11-27 17:21:44 +00:00
espie
51892d0631 explicitly ask for stdin 2011-11-27 16:52:04 +00:00
espie
a3733c98dd for :patch ports, show-prepare-results should also include the list of
installed dependent ports, so we have no choice but ask, which is fairly
easy to do.
2011-11-26 13:52:52 +00:00
espie
03b76c3a13 - tweak "prepare" stage to give more accurate default/dependency result directly
code should still work with older pkg_info, but you need uptodate pkg_info
for best messages, e.g.,:
===> kdelibs-3.5.10p14 depends on: gettext->=0.10.38 -> gettext-0.18.1p0
===> kdelibs-3.5.10p14 depends on: metaauto-* -> metaauto-1.0
===> kdelibs-3.5.10p14 depends on: autoconf-2.61 -> autoconf-2.61p3
===> kdelibs-3.5.10p14 depends on: gmake-* -> gmake-3.82

- adds target: show-prepare-results that yields the list of installed packages
deduced by prepare

- kill undocumented CLEAN_PLIST_OUTPUT
- remove old internals
- ensure IS_INTERACTIVE and REGRESS_IS_INTERACTIVE are always defined
- fix a bug in FETCH_MANUALLY and BATCH interaction
2011-11-25 13:58:13 +00:00
espie
636426623c grmf, managed to reintroduce two PKG_DBLOCK 2011-11-24 19:24:54 +00:00
espie
ff9908d97d bug-fix, behavior for dependency check with "_DEPENDS_TARGET=package" is
disturbingly complicated (as reported by aja@, naddy@, and others. Sorry,
was looking at the wrong code, so I didn't see the bug)
2011-11-24 18:12:28 +00:00
espie
752bf17c1c zap/protect some stuff.
keep FTP_KEEP_ALIVE for now, until we sort it out.
2011-11-24 17:49:58 +00:00
espie
825f591fa4 BASESYSCONFDIR points to /etc (global user settings), and SYSCONFDIR
is derived from it.
old march discussion, prodded by fgs@ about it.
2011-11-21 16:42:52 +00:00
espie
0ed05101b4 nicer less confusing messages 2011-11-21 12:26:05 +00:00
espie
f5458c8a3e better check for PREPARE_CHECK_ONLY: assume we're second_pass already in
the check_install case.
That way, we get the full error report.
2011-11-21 12:20:53 +00:00
espie
eba34186b3 Speed up prepare stage: simplify the old loop, remove old cruft.
In the case where we want an install, run one single pkg_info instead of two.
If this fails during the second pass, we will run it again to figure out
whether the install or the default is wrong, but this is an error condition,
so this need not slow down the general case.

Add PREPARE_CHECK_ONLY knob, to be used by dpb: dpb already installs what
it needs, so for install dependencies to be missing during the "prepare"
stage is an error in dpb's logic...
2011-11-21 12:16:42 +00:00
espie
cb960e35d0 mv _cache_fragment to pkgpath.mk
create a new category of recursive targets that can use this fragment,
and use them, e.g., print-package-signature from top-level should benefit
a lot.
2011-11-19 11:33:39 +00:00
espie
4c9d60034c do the no_mips64 dance here, no need to duplicate it in lots of Makefiles 2011-11-18 11:01:47 +00:00