linker now uses exec-only by default - amd64, aarch64, riscv64,
powerpc64 - so that when packages are next rebuilt, pkg_add -u will
replace all packages with the new ones.
(additionally, the generated Makefiles to create debug packages changed
from binutils strip+objcopy to the LLVM versions to avoid damage done to
GNU_RELRO, so all packages which have debug-package support in their
build need to be replaced for that reason also; only aarch64 and amd64
have debug-packages so they're covered by the above bump too).
Make use of new features in xz 5.4.0:
* Use xz, which now has upstream support for pledge(), instead of xzdec.
* Enable threaded decompression for ports that use parallel make.
* Use xz instead of lunzip to decompress .lz files.
ok kn@ sthen@
script infrastructure previously used for USE_WXNEEDED to allow adding
--no-execute-only by setting USE_NOEXECONLY=Yes in the port.
This isn't ideal for compilers (in cases where those can't be made to work
with exec-only they need patching to add to the internally constructed linker
command line, though it does at least help with initial testing there),
and is useful for some other ports.
ok tb@
Historically, bsd.lib.mk used to build lib*_pic.a libraries, except
on mips64. Ten years ago, bsd.lib.mk stopped creating lib*_pic.a
files and the corresponding PFRAG.no_mips64 bits were removed from
the ports tree. Remove the _PKG_ARGS handling that was forgotten
at the time.
From Brad, ok sthen@
separate commands for files, files with .part suffix, and directories.
fixes "make clean=dist" in sysutils/telegraf and probably some other
ports using go.port.mk. (one big 622K command split into 228+246+147K).
use ${MAKESUMFILES:=.part} instead of regex for files with .part suffix,
it is reliably faste^H^H^H^H^Hnot quite as slow
discussed with landry, phessler
as sdk@ noticed, there are a lot of subdirectories in gorust land.
removing every subdirectory with rmdir -p is easy to do (thx semarie@)
and since we start in a subdirectory, rmdir -p cannot recurse up to /,
so sthen@ gave his approval.
uses a different name or a subdirectory for the file.
Adjust some ports to use it: p5-DBI (different name), p5-Template (xs
subdirectory). There are others that could use this.
OK bluhm@
from base Perl. The newer version of the header fixes compiler warnings
(-Wcompound-token-split-by-macro) which are noisy in build logs.
Tweaked from diff from espie@ via bluhm@, earlier version ok afresh1@
format to ${PKGDIR}/(class).login, this also allows getting rid of the
LOGIN_CONF variable so that no Makefile changes are needed, only adding
the @sample.
the symbol changes in libcompiler-rt.a for emutls are causing problems
and at least 250 packages maybe affected so this is the most sensible
way (we were overdue a bump to this anyway after other toolchain
changes, it's better if people are running code built with current
toolchain rather than a mixture).
discussed with robert@ tb@ espie@ naddy@
Export some ExtUtils::MakeMaker and Module::Build recognized environment
variables so they are picked up even if we don't run the configure
script directly.
This is useful for things that are not perl modules themselves, but include
perl modules, some of which seem to configure them at build time instead
of configure.
This is the first step to moving perl to have a vendor lib that
ports install into that is separate from where CPAN will install things.
Thanks to kmos@ for a bulk build that didn't turn up anything untoward
as discussed with sthen@)
also: kill some special useless casing... the regexp for adding
REVISION/EPOCH at the right location works with and without flavors
(no difference shown on make show=PKGNAMES through a full tree)
comes from REVISION and EPOCH properly. The most crucial check is probably
checking the end "flavor_ext" indeed comes from flavors (with the notable
exception of MULTI_PACKAGES' default FULLPKGNAME, as per ispell)
from gettext, to add a few more gettext scripts that we see sometimes, and
to add asciidoc. (has been in i386 bulks for months and forgotten about!)
ok espie, "I'm fine with the gettext-tools part and have no opinion on asciidoc" naddy@
Add powerpc to LLD_ARCH, and bump _SYSTEM_VERSION-powerpc so people
will upgrade to lld-linked packages when they exist. Remove
-Wl,--relax from several ports; ld.bfd needed this flag, but ld.lld
can't take it.
lang/gcc/8 needs USE_LLD = No, because LLD can't link C++ code from
gcc (error: relocation refers to a discarded section: ...), and also,
the adastrap is too old.
To avoid linker errors on macppc, you need a snapshot where
/usr/bin/ld is LLD.
ok sthen@