At kde overlay 20c63cd8e (emails despammed, metadata update after): * commit d1f87928d |\ Merge: 967b4cbb9 9f1b1d7a4 | | Author: Repository mirror & CI <repomirrorci@> | | AuthorDate: Thu Apr 11 20:48:23 2024 +0000 | | Commit: Repository mirror & CI <repomirrorci@> | | CommitDate: Thu Apr 11 20:48:23 2024 +0000 | | | | Merge updates from master from upstream plasma-desktop (emails despammed): commit b4fe3e911 Author: Jakob Petsovits <jpetso@> AuthorDate: Thu Apr 4 19:03:49 2024 -0400 Commit: Jakob Petsovits <jpetso@> CommitDate: Tue Apr 9 21:59:55 2024 -0400 kcms/mouse: Remove unused xorg-evdev dependency It's not necessary to access it directly anymore, now that we only expose settings for libinput. kcms/mouse/CMakeLists.txt | 3 --- kcms/mouse/backends/CMakeLists.txt | 1 - kcms/mouse/inputbackend.h | 1 - 3 files changed, 5 deletions(-) plasma-desktop-9999 is failing rebuild. At that commit the patch failure is with kcms/mouse/CMakeLists.txt which that commit changes. However, HEAD has an additional failure at the root CMakeLists.txt , resulting in: Applying plasma-desktop-5.90.0-override-include-dirs.patch ... patching file CMakeLists.txt Hunk #1 FAILED at 224. 1 out of 1 hunk FAILED -- saving rejects to file CMakeLists.txt.rej patching file config-gentoo.cmake patching file kcms/mouse/CMakeLists.txt Hunk #1 FAILED at 32. 1 out of 1 hunk FAILED -- saving rejects to file kcms/mouse/CMakeLists.txt.rej patching file kcms/touchpad/CMakeLists.txt patching file kcms/touchpad/backends/x11.cmake [ !! ]
Oh the joys/challenges/follies of running a live-git plasma desktop via the gentoo/kde overlay ebuilds (and patches)... Yes, I /did/ know what I was signing up for! =:^\ The trouble is that while the complexity of the override-include-dirs approach (not just a patch, a tarball with the substitute headers...) does make a longer than usual delay here entirely understandable, I'm still sitting here watching plasma-desktop get further and further behind upstream, worrying that they are going to do one of their inter-package component-moves or some other major multi-package-synced-change and I'll end up with a broken desktop due to plasma-desktop being behind and out-of-sync. So better to be proactive and at least try to come up with a hack to get a synced plasma-desktop merging again for me and avoid that potential problem, even if such a hack isn't gentoo/kde commit-worthy... but of course the additional complexity of this particular case hits me too! Anyway... The above upstream commit deleted the three evdev lines in kcms/mouse/CMakeLists.txt thus triggering that hunk failure. Applying a patch-to-gentoos-patch removing those lines... -- if (NOT EVDEV_FOUND) -- list(APPEND MISSING_DEPS "xorg-evdev") -- endif() ... allows the build to continue without further issue at that commit and that's what I just installed now. (I just left the existing tarball including the evdev header, etc, in place.) But as mentioned there's also the main CMakeLists.txt hunk failure from a later commit, which I've yet to investigate... I'm passed upstream b4fe3e911 now tho, progress to report.
Upstream f5f803832 (kcms/keyboard but touches the main CMakeLists.txt) is the second culprit, for the main CMakeLists.txt chunk patch failure. Note that the immediately previous 8608d7034 patches (with the evdev fix as in comment #1) but will fail to build (at least with USE=-X) due to an upstream bug fixed in (later) 27eec5ff1. So I applied 27ee as a patch and now have 8608 installed (with the fix in comment #1), but haven't yet looked at second-patch-failure-culprit f5f8 to see what we need to do with it. But it's midnight and I'm supposed to be up in ~4 hours to get ready for work, so further examination of f5f8 will have to wait...
Created attachment 892128 [details, diff] kcms/mouse/CMakeLists.txt no evdev patch Here's the first override-include-dirs patch patch, removing evdev from the mouse kcm cmakelists.txt
Created attachment 892131 [details, diff] main cmakelists.txt xkbregistry patch Here's the second override-include-dirs patch patch, adding xkbregistry to the main cmakelists.txt. This (with the first patch, and the later upstream 72ee commit still applied as a patch) allows upstream f5f803832 to patch, build and merge successfully. And after remembering to disable that 72ee patch now that I can get past where it's applied upstream, HEAD ( 9680bc8fe ) now builds and merges for me. =:^)
Very good, I came to the same conclusion. One more thing to cleanup upstream though before respinning the patch.
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/proj/kde.git/commit/?id=336670543dcd9cbcacaff655a482320a88232853 commit 336670543dcd9cbcacaff655a482320a88232853 Author: Andreas Sturmlechner <asturm@gentoo.org> AuthorDate: 2024-05-20 18:44:09 +0000 Commit: Andreas Sturmlechner <asturm@gentoo.org> CommitDate: 2024-05-20 19:18:32 +0000 kde-plasma/plasma-desktop: Drop IUSE X, bump override-include-dirs to v4 IUSE X was very weak in the first place since still depending on libX11, and not even correct in some of the thought-to-be optional deps. Drop EVDEV, it is gone, backport a couple commits to 6.0 branch for dual use of the respun tarball. Closes: https://bugs.gentoo.org/929969 Signed-off-by: Andreas Sturmlechner <asturm@gentoo.org> kde-plasma/plasma-desktop/Manifest | 2 +- ...asma-desktop-5.90.0-override-include-dirs.patch | 117 --------------------- .../files/plasma-desktop-6.0.5-unused-dep.patch | 85 +++++++++++++++ .../plasma-desktop-6.0.49.9999.ebuild | 31 +++--- .../plasma-desktop/plasma-desktop-9999.ebuild | 30 ++---- 5 files changed, 109 insertions(+), 156 deletions(-)