Compiling kde-plasma/kwin-6.0.90.1 in the llvm profile giving the follow error bellow: ``` FAILED: src/CMakeFiles/kwin_wayland.dir/watchdog.cpp.o (truncated) error: use of undeclared identifier 'geteuid' 47 | qCInfo(KWIN_WATCHDOG) << "Watchdog: enabled. Interval:" << watchdogIntervalInUs << t->interval() << qgetenv("NOTIFY_SOCKET") << geteuid(); | ^ 1 error generated. ``` Kwin was compiled with the follow useflags: caps handbook lock screencast shortcuts systemd I gonna compile with gcc to check if changes anything.
Created attachment 894607 [details] kwin-6.9.90.1-build.log.gz
Created attachment 894608 [details] emerge_info.txt
Compiling with gcc it did worse (I was expecting that). I notice that systemd flag was added, gonna try compile kwin without to see if the problem is there. I think is better to add this bug as block to BUG 926828
Ok, disabling systemd useflag on kwin results in a successful compiled, but very strange the build switch is failing.
Can confirm. Also ended up figuring the systemd USE flag was turning on the new watchdog option, so I turned it off and kwin built fine.
(In reply to unhappy-ending from comment #5) > Can confirm. Also ended up figuring the systemd USE flag was turning on the > new watchdog option, so I turned it off and kwin built fine. I didn’t analyze the issue further because plasma-workspace is broke after the update and now I rebuild all plasma-meta dependencies, maybe some missing dependencie.
Is everyone currently CCed using systemwide-libcxx?
(In reply to Andreas Sturmlechner from comment #7) > Is everyone currently CCed using systemwide-libcxx? At least me and unhappy-ending are
Created attachment 894761 [details, diff] includes.patch https://invent.kde.org/plasma/kwin/-/commit/05a3e2bad9a3baf27293f0b5ed4bef4f38952804 This adds use of geteuid without the needed headers. Please try the attached untested patch.
Patch is all good!
*** Bug 933131 has been marked as a duplicate of this bug. ***
Can confirm this patch work and solves BUG 933131 See also: https://bugs.gentoo.org/933131#c7
The libcxx type thing is separate and I think upstream KDE have done all they can there really. We just need to figure out something on the Gentoo side which we're already discussing in another bug, right? Anyway, thanks for testing, I'll submit it.
(C++ doesn't have implicit function declarations so I would be very surprised if includes for C functions like this could make any runtime difference.)
(In reply to Sam James from comment #14) > (C++ doesn't have implicit function declarations so I would be very > surprised if includes for C functions like this could make any runtime > difference.) Sam this will be a bit off topic but what are the problems of using `MYCMAKEARGS="-DLIBCXX_TYPEINFO_COMPARISON_IMPLEMENTATION=2"`? For now to solve the right click BUG 923292 on plasma cause by qtwayland, I recompile both with the `MYCMAKEARGS` and then recompile libcxx without the `MYCMAKEARGS` to not interfere with other packages.
(In reply to Gonçalo Negrier Duarte from comment #15) > (In reply to Sam James from comment #14) > > (C++ doesn't have implicit function declarations so I would be very > > surprised if includes for C functions like this could make any runtime > > difference.) > > Sam this will be a bit off topic but what are the problems of using > `MYCMAKEARGS="-DLIBCXX_TYPEINFO_COMPARISON_IMPLEMENTATION=2"`? > > For now to solve the right click BUG 923292 on plasma cause by qtwayland, I > recompile both with the `MYCMAKEARGS` and then recompile libcxx without the > `MYCMAKEARGS` to not interfere with other packages. I don't know, I'm afraid. I have no idea if it breaks ABI or what the consequences of setting it are. I (or someone) needs to ask some C++ gurus.
(In reply to Sam James from comment #16) > (In reply to Gonçalo Negrier Duarte from comment #15) > > (In reply to Sam James from comment #14) > > > (C++ doesn't have implicit function declarations so I would be very > > > surprised if includes for C functions like this could make any runtime > > > difference.) > > > > Sam this will be a bit off topic but what are the problems of using > > `MYCMAKEARGS="-DLIBCXX_TYPEINFO_COMPARISON_IMPLEMENTATION=2"`? > > > > For now to solve the right click BUG 923292 on plasma cause by qtwayland, I > > recompile both with the `MYCMAKEARGS` and then recompile libcxx without the > > `MYCMAKEARGS` to not interfere with other packages. > > I don't know, I'm afraid. I have no idea if it breaks ABI or what the > consequences of setting it are. I (or someone) needs to ask some C++ gurus. Ok thanks anyway, until futher notice I will use my method at least qtwayland compiles fast and I don't risk breaking my system.
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=abbe800563fc9c87b9072e14387cab15feb9ab8a commit abbe800563fc9c87b9072e14387cab15feb9ab8a Author: Sam James <sam@gentoo.org> AuthorDate: 2024-06-13 00:45:36 +0000 Commit: Sam James <sam@gentoo.org> CommitDate: 2024-06-13 00:46:15 +0000 kde-plasma/kwin: fix USE=systemd w/ libcxx Already merged upstream in https://invent.kde.org/plasma/kwin/-/merge_requests/5811 but filed a backport MR which is pending at https://invent.kde.org/plasma/kwin/-/merge_requests/5892. Closes: https://bugs.gentoo.org/933120 Signed-off-by: Sam James <sam@gentoo.org> kde-plasma/kwin/files/kwin-6.0.90.1-libcxx.patch | 27 ++++++++++++++++++++++++ kde-plasma/kwin/kwin-6.0.90.1-r3.ebuild | 4 ++++ 2 files changed, 31 insertions(+)
