=== guiverc2 is now known as guiverc [12:15] Hi all [16:02] hi everyone [16:03] RikMills: thanks for the workaround/fix for kio, and another thing: I have detected yesterday a bunch of FTBFSes in our packages [16:03] I think latest dpkg is what triggered them [16:04] in the few build logs I inspected it fails at the autotests part [16:04] santa_: FORTIFY_SOURCE=3 ? [16:05] maybe that's the culprit but I didn't have time to investigate properly yet [16:05] let me give you some links... [16:06] https://area51.tritemio.org/build-status/buildstatus_ubuntu-exp2/ubuntu-exp2_status_frameworks.html [16:06] ↑ this is a test rebuild done a little bit before the last dpkg upload in proposed [16:06] now, after the new dpkg: [16:07] https://watertown.tritemio.org/build-status/buildstatus_ubuntu-exp2/ubuntu-exp2_status_frameworks.html [16:07] also they are plasma and apps packages affected: [16:07] https://watertown.tritemio.org/build-status/buildstatus_ubuntu-exp2/ubuntu-exp2_status_plasma.html [16:07] https://watertown.tritemio.org/build-status/buildstatus_ubuntu-exp2/ubuntu-exp2_status_applications.html [16:09] santa_: there was also new GCC uploaded on the same day as dpkg [16:09] https://launchpad.net/ubuntu/+source/gcc-13/13.2.0-9ubuntu1 [16:10] * For Ubuntu 24.04 LTS and later, switch _FORTIFY_SOURCE to 3 by default [16:10] instead of 2 when optimization is enabled. LP: #2012440. [16:10] -ubottu:#kubuntu-devel- Launchpad bug 2012440 in gcc-13 (Ubuntu) "Please add -D_FORTIFY_SOURCE=3 to default build flags" [High, Fix Released] https://launchpad.net/bugs/2012440 [16:11] many KDE apps are also now segfaulting on startup [16:12] particularly those that use qtwebengine [16:12] https://launchpad.net/ubuntu/+source/dpkg [16:12] s/KDE/KDE and Qt/ [16:12] https://launchpad.net/ubuntu/+source/gcc-13 [16:13] after the dpkg upload (20/12/2023) things started to FTBFS [16:14] the gcc was uploaded a bit earlier (16/12/2023) but that didn't trigger the FTBFSes [16:14] but not before? [16:14] about runtime problems: I don't know [16:14] exactly [16:14] sounds like we have multiple issue then [16:15] and most canicial people are now off until after new year [16:15] the first link that I gave you above is a test rebuild of frameworks done on dec 19 [16:15] canonical [16:15] canicial :) [16:16] :D [16:17] RikMills: any example of apps that segfault? [16:18] santa_: We have a list, let me grab it for you real quick... [16:19] falkon, ghostwriter, digikam (once you have done the startup wizard) and more [16:19] santa_: This is a *draft* list, but it's all the ones we've confirmed so far: https://paste.ubuntu.com/p/vSMfQTG8zp/ [16:19] It's on the Lubuntu-internal notes thing (read: internal because of spammers) so I just copied it over :) [16:20] So, when you add the Plasma widget for displaying a webpage? Crashes Plasma entirely. :P [16:20] it seems to be an apparmor change [16:20] (That's why plasmashell is there and bolded.) [16:20] santa_: https://discourse.ubuntu.com/t/spec-unprivileged-user-namespace-restrictions-via-apparmor-in-ubuntu-23-10/37626 https://launchpad.net/ubuntu/+source/apparmor/4.0.0~alpha2-0ubuntu7 [16:21] https://bugs.launchpad.net/ubuntu/+source/apparmor/+bug/2046844 [16:21] -ubottu:#kubuntu-devel- Launchpad bug 2046844 in apparmor (Ubuntu) "AppArmor user namespace creation restrictions cause many applications to crash with SIGTRAP" [Critical, Confirmed] [16:23] all in all, December 2023 has been bug central for Qt/KDE things :/ [16:23] When it rains, it pours. [16:24] yes, things happened [16:24] speaking of which, we have some KA problem which are now solved in the master branch [16:24] <3 [16:24] * RikMills looks [16:25] I mean that problem with the build status pages and such [16:25] ahh. already get that [16:25] but many thanks again, all the same [16:26] santa_: Not that you particularly care about this side of things, but we've also started on a packagename scheme for KF6 with the Debian folks. The current proposal (from a much newer member) is to go with kf6-$UPSTREAM_NAME, you don't have any huge objections to Kubuntu mirroring that, do you? [16:27] IMO that seems reasonable, and hopefully this helps set some rough expectations on KF6 support in KA :) [16:36] tsimonq2: I do care about these things, because whatever wrong thing they do probably will end up here (not by my hand, but probably by the hand of others) [16:37] tsimonq2: also it's relevant for the KA support like you said [16:39] santa_: Here's the exact proposal that Rik and I generally agree with; what are your thoughts? :) https://paste.ubuntu.com/p/3pzJgzpq2D/ [16:40] The only visible query on that is whether the source package names will exactly match the repo names, I think *that* is undecided as of yet. [16:40] tsimonq2: that seems to me like the right solution (new source packages), at least for frameworks [16:42] they are some things to consider here: [16:42] - one thing is the source package name [16:42] - another thing is the upstream tarball name [16:42] - another thing is the distro(kubuntu,debian) git repo name [16:43] the source package name is not important for KA and automation in general (or it shouldn't be) [16:43] the distro repo name is relevant [16:44] when you do a git-clone-all, KA retrieves the list of upstream tarballs in KDE's FTP [16:44] from that list it clones the distro git repos [16:46] it uses the same name than the tarballs, unless yoy override the thing in ka-metadata [16:46] s/yoy/you( [16:46] s/yoy/you/ [16:46] dammit XD [16:46] let's see if I can find the file in question [16:50] https://git.launchpad.net/~kubuntu-packagers/ka/+git/ka-metadata/tree/name-conversions/kde-tarball-to-distro-git/kubuntu.json [16:50] ↑ here, the exceptions for git-clone-all [16:51] I'll leave a comment there suggesting we keep the repo name the same as the source package name :) [16:52] santa_: I really appreciate the feedback! [16:52] tsimonq2: all of that being said, kf6-$UPSTREAM_NAME is a good proposal for distro git repos because we could make git-clone-all -r frameworks6 automatically preprend '-kf6' [16:53] Right, exactly :) [16:53] * 'kf6-' [16:53] and also it doesn't force us to use a separate name space for the git repos [16:55] the good thing of also using kf6- prefix for source package names is that allows the packages to co-exist with the kf5 counterpart [16:55] (but you don't need the KA developer to tell you that :D) [16:56] I have been reading the things in salsa a couple of days ago byt the way [16:56] s/byt/by/ [17:10] RikMills: I think most of the Qt crash-on-start problems are resolved by building qtbase-opensource-src with _FORTIFY_SOURCE=2. I don't know about the qtwebengine stuff though. [17:14] santa_: "I can't speak for Debian's automation, but for Kubuntu's automation (as currently designed), we have to specifically map packages whose source names do not match their repo names. If it isn't terribly inconvenient, could we just keep the repository names the same as the source package? Thanks for your consideration." sent :) [17:14] santa_: oh you're in the Salsa team? cool :) [17:14] santa_: didn't know if you were - makes both of our lives easier :D [17:14] https://salsa.debian.org/qt-kde-team/kde-6-planning/-/issues/2 [17:14] -ubottu:#kubuntu-devel- Issue 2 in qt-kde-team/kde-6-planning "KF6 source package names" [Opened] [17:15] tsimonq2: I'm not but I can see the repos, the issues... it's puiblicly open [17:15] I remember sending that link to someone last week and the issue must have been not open, or something. [17:15] My bad :) [17:16] tsimonq2: regarding the mismatching repo names, we could make them match, yes, but keep in mind they are renames [17:16] e.g. kgamma5 was recently renamed as kgamma, so the fastest solution was just adding an entry in that file [17:17] but yeah we could clean up that thing in the long term [17:17] Makes sense :) [17:17] * tsimonq2 finds an early lunch...