> From: William Hubbs <williamh@gentoo.org> > To: gentoo-portage-dev@lists.gentoo.org > Cc: qa@gentoo.org, William Hubbs <williamh@gentoo.org> > Subject: [gentoo-portage-dev] [PATCH] install-qa-check.d: remove check that bans libtool files and static libs from / > > Most upstreams and build systems do not make this distinction, so this > causes unnecessary hacks in ebuilds. > > Signed-off-by: William Hubbs <williamh@gentoo.org> > --- > bin/install-qa-check.d/80libraries | 10 ---------- > 1 file changed, 10 deletions(-) > > diff --git a/bin/install-qa-check.d/80libraries b/bin/install-qa-check.d/80libraries > index d1d2c4fdd..e59369bf6 100644 > --- a/bin/install-qa-check.d/80libraries > +++ b/bin/install-qa-check.d/80libraries > @@ -152,16 +152,6 @@ lib_check() { > done > [[ ${abort} == "yes" ]] && die "add those ldscripts" > > - # Make sure people don't store libtool files or static libs in /lib > - f=$(ls "${ED%/}"/lib*/*.{a,la} 2>/dev/null) > - if [[ -n ${f} ]] ; then > - __vecho -ne '\n' > - eqawarn "QA Notice: Excessive files found in the / partition" > - eqawarn "${f}" > - __vecho -ne '\n' > - die "static archives (*.a) and libtool library files (*.la) belong in /usr/lib*, not /lib*" > - fi > - > # Verify that the libtool files don't contain bogus $D entries. > local abort=no gentoo_bug=no always_overflow=no > for a in "${ED%/}"/usr/lib*/*.la ; do
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/proj/portage.git/commit/?id=498900e5e51460502d8271f409a4c614a021613b commit 498900e5e51460502d8271f409a4c614a021613b Author: William Hubbs <williamh@gentoo.org> AuthorDate: 2019-10-27 17:40:07 +0000 Commit: Zac Medico <zmedico@gentoo.org> CommitDate: 2019-11-03 19:44:44 +0000 install-qa-check.d: remove check that bans libtool files and static libs from / Most upstreams and build systems do not make this distinction, so this causes unnecessary hacks in ebuilds. Bug: https://bugs.gentoo.org/699252 Signed-off-by: William Hubbs <williamh@gentoo.org> Signed-off-by: Zac Medico <zmedico@gentoo.org> bin/install-qa-check.d/80libraries | 10 ---------- 1 file changed, 10 deletions(-)
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/proj/portage.git/commit/?id=a2506e4ebd47781eb81b7d13b0a1a376ed028c77 commit a2506e4ebd47781eb81b7d13b0a1a376ed028c77 Author: Zac Medico <zmedico@gentoo.org> AuthorDate: 2019-11-03 19:59:06 +0000 Commit: Zac Medico <zmedico@gentoo.org> CommitDate: 2019-11-03 20:00:59 +0000 install-qa-check.d: handle libtool files and static libs from / Fixes: 498900e5e514 ("install-qa-check.d: remove check that bans libtool files and static libs from /") Bug: https://bugs.gentoo.org/699252 Signed-off-by: Zac Medico <zmedico@gentoo.org> bin/install-qa-check.d/80libraries | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/proj/portage.git/commit/?id=ae2a76261fd7e519bfbb3b8c4b41aa511d039245 commit ae2a76261fd7e519bfbb3b8c4b41aa511d039245 Author: Zac Medico <zmedico@gentoo.org> AuthorDate: 2019-11-03 20:12:02 +0000 Commit: Zac Medico <zmedico@gentoo.org> CommitDate: 2019-11-03 20:14:42 +0000 Revert "install-qa-check.d: remove check that bans libtool files and static libs from /" This reverts commit 498900e5e51460502d8271f409a4c614a021613b. This change is still being discussed. See: https://archives.gentoo.org/gentoo-portage-dev/message/6e4cfbb0ef9c36dc6511d4f2003cc458 Bug: https://bugs.gentoo.org/699252 Signed-off-by: Zac Medico <zmedico@gentoo.org> bin/install-qa-check.d/80libraries | 10 ++++++++++ 1 file changed, 10 insertions(+)
So it seems the ban on static libs and libtool files in / is not going to be lifted in general, and FWIW I understand and agree, but... Could some at least user-level override be made available? Maybe QA_STRICT_STATICLIBS=ignore, to sort of match the other QA_STRICT_* vars listed in the make.conf manpage but reverse form from =set, since those force stricter than default while this would be loosening the default? And/or simply make it subject to FEATURES=stricter or some such, which I believe the other QA_STRICT variables are. (I have FEATURES=strict but not stricter set here, and get non-fatal warnings for most of those.) The scenario is bug #701082, filed a couple weeks after this bug in Nov 2019. It seems upstream qt introduced a build-system feature where they detect the existing install location (presumably for qtcore) and install additional modules to the same location. That detects my reverse-usrmerge /usr -> . symlink and installs directly to / in the fake-install image, thus triggering this die where most ebuilds don't, as they're installed to /usr in the fake-install image and only merged directly to / in qmerge, after this QA_ test. But since reverse-usrmerge apparently isn't fully supported and no one else seems to have run into the bug, the bug has pretty much just sat there. FWIW equery f '*' | grep '^/lib64/.*\.a' say I have only four *.a files installed directly to /lib64. All four of those are dev-qt/* three from qtdeclarative, one from designer. There are of course other *.a files in /lib64, but they're actually installed to /usr/lib64 which just happens to be /lib64 on my box, which is fine as they don't trigger the die. Since I troubleshot that bug I've been carrying a local portage patch that s/die/eqawarn/ on the appropriate line in 80libraries in ordered to continue running current qt, required for the live-git kde-* I run from the gentoo/kde overlay, and having some way to user-toggle that check to nonfatal in portage config would allow me to drop that local patch. =:^)
This check is pretty much irrelevant for usr-merged systems, no matter the direction, since /usr/lib and /lib are the same directory.
