| Summary: | sys-apps/systemd-215 error generating polkit policy files due to intltool bug | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Stefan G. Weichinger <lists> |
| Component: | [OLD] Core system | Assignee: | Freedesktop bugs <freedesktop-bugs> |
| Status: | RESOLVED UPSTREAM | ||
| Severity: | normal | CC: | alexander, genzilla, gottlieb, marduk, mike, ssuominen, systemd |
| Priority: | Normal | Keywords: | NeedPatch, UPSTREAM |
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| URL: | https://bugs.launchpad.net/intltool/+bug/605826 | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: | build.log for systemd-215 | ||
|
Description
Stefan G. Weichinger
2014-07-04 09:39:50 UTC
Ah, I see, I checked the wrong directory. It fails in /var/tmp/portage/sys-apps/systemd-215/work/systemd-215-abi_x86_64.amd64 and the missing file is: /var/tmp/portage/sys-apps/systemd-215/work/systemd-215-abi_x86_64.amd64/src/hostname/org.freedesktop.hostname1.policy the whole subdir "hostname" is missing in src It seems as if it should chose /var/tmp/portage/sys-apps/systemd-215/work/systemd-215 instead of systemd-215-abi_x86_64.amd64, right? I was able to compile it after doing this: cd /var/tmp/portage/sys-apps/systemd-215/work/systemd-215-abi_x86_64.amd64 mkdir src/timedate mkdir src/login mkdir src/locale chown -R portage:portage src Don't know if this is correct. Thanks, Stefan I have this issue as well. It built on one system, but failed with the error on another. The only difference I saw was that the working systemd had USE=-filecaps where the breaking system had the flag on, but when I tried USE=filecaps on the breaking system it still fails with the same error. Finally I saw that the two systems had different python's eselected but changing that didn't make any difference either. Seems like a dupe of #493542 although the work-around still didn't work for me. (In reply to Albert W. Hopkins from comment #4) > Seems like a dupe of #493542 although the work-around still didn't work for > me. In my initial report I already tried MAKEOPTS="-j1" without success. (In reply to Stefan G. Weichinger from comment #0) > On one of my ~amd64 systems it builds fine, on another it does not. > > > Reproducible: Always > > Steps to Reproduce: > 1. # MAKEOPTS="-j1" emerge -1 systemd > 2. fails with build.log posted below > > Actual Results: > # last lines of build.log Post complete build.log, please, it's always useful when looking at build errors, it will directly show on top of it what USE flags were used as well as the ./configure output This is likely related to the new --enable/--disable-nls flag upstream introduced that is directly related to the --enable/--disable-polkit Created attachment 380176 [details]
build.log for systemd-215
(In reply to Stefan G. Weichinger from comment #7) > Created attachment 380176 [details] > build.log for systemd-215 on my 2 thinkpads it built fine, just tested on the L520. build.log for one of those as well? (In reply to Stefan G. Weichinger from comment #1) > the whole subdir "hostname" is missing in src There is probably some missing make dependency that allows intltool-merge to be called before mkdir -p src/hostname. It is quite strange that you get this with MAKEOPTS=-j1. Err.. actually that directory should get created by configure. So this quite a mystery indeed. Hmm... configure only creates it if dependency tracking is enabled; econf disables this by default. Try this as a kludge/workaround: EXTRA_ECONF="--enable-dependency-tracking" emerge -1 systemd (In reply to Mike Gilbert from comment #12) > Try this as a kludge/workaround: > > EXTRA_ECONF="--enable-dependency-tracking" emerge -1 systemd The above workaround works for me, although I'm still confused as to why it's not needed on some systems but is for others. (In reply to Albert W. Hopkins from comment #13) It just causes configure to create the directory instead of deferring it to the (broken) Makefile. (In reply to Mike Gilbert from comment #12) > Try this as a kludge/workaround: > > EXTRA_ECONF="--enable-dependency-tracking" emerge -1 systemd works here as well. If I run "touch /usr/bin/intltool-merge" so that it has a newer mtime than the .policy files in the systemd tarball, I can consistently reproduce this failure. I have added --enable-dependency-tracking to the ebuild for now. Another viable workaround would be to do an in-source build. The real solution would be to fix the rule in m4/intltool.m4 to create the parent directory, but I'm not sure how to do so. *** Bug 493542 has been marked as a duplicate of this bug. *** Bug for the upstream intltool project: https://bugs.launchpad.net/intltool/+bug/605826 suggesting to close this one as OBSOLETE? (outdated version of package ... not in portage anymore) |
