KGB-0 | gnome-user-docs signed tags 41201aa Gunnar Hjalmarsson ubuntu/45.0-1ubuntu2 * gnome-user-docs Debian release 45.0-1ubuntu2 * https://deb.li/4JM | 03:14 |
---|---|---|
=== guiverc2 is now known as guiverc | ||
seb128 | goood morning desktopers! | 07:44 |
lissyx | seb128, hello | 07:59 |
seb128 | lissyx, salut :) | 07:59 |
lissyx | do you think you can help me for https://bugzilla.mozilla.org/show_bug.cgi?id=1855135#c7 and the related PRs ? | 07:59 |
-ubottu:#ubuntu-desktop- Mozilla bug 1855135 in Toolkit "Fetching crash annotations from non-content child processes doesn't work on Ubuntu Snap builds" [--, New] | 07:59 | |
lissyx | https://github.com/canonical/firefox-snap/pull/31 and https://github.com/canonical/firefox-snap/pull/32 | 07:59 |
-ubottu:#ubuntu-desktop- Pull 31 in canonical/firefox-snap "Bug 1855135 - Remove lto=cross (as a temp aworkaround?)" [Open] | 07:59 | |
-ubottu:#ubuntu-desktop- Pull 32 in canonical/firefox-snap "Bug 1855135 - Remove lto=cross (as a temp aworkaround?)" [Open] | 07:59 | |
lissyx | we are lacking crash reports from beta and stable, this is not really good | 07:59 |
seb128 | lissyx, is there a bug about the fact that lto=cross is buggy? | 08:01 |
seb128 | I don't like to apply workarounds if they are used as a pretext to not fix the real issue, that comment on the upstream bug 'Let's keep this bug open until his PRs have landed.' suggest Gabriele plan to close the bug and consider the problem fixed once the workaround is in the snap build? | 08:02 |
lissyx | yes well | 08:04 |
lissyx | enable-lto=cross is not used elsewhere as much as I can find | 08:04 |
lissyx | we can spend time debugging why, but I'm not sure it's the best use of our time | 08:04 |
lissyx | and in the mean time we have users experiencing crashes and we cannot get data on that | 08:06 |
seb128 | I wish I would have the context of why that option was initial added to the snap build before deciding it's wrong :/ | 08:10 |
seb128 | but yes, you are right | 08:11 |
KGB-0 | gnome-control-center ubuntu/latest 9e95dec Sebastien Bacher debian/ changelog patches/series patches/gitlab-fix-vpn.patch * https://deb.li/hBNx | 10:03 |
KGB-0 | * debian/patches/gitlab-fix-vpn.patch: | 10:03 |
KGB-0 | - cherry pick an upstream fix for a recent regression in the VPN | 10:03 |
KGB-0 | configuration (lp: #2037011) | 10:03 |
-ubottu:#ubuntu-desktop- Launchpad bug 2037011 in network-manager-openvpn (Fedora) "Can't create ovpn" [Undecided, New] https://launchpad.net/bugs/2037011 | 10:03 | |
KGB-0 | gnome-control-center ubuntu/latest 49bd756 Sebastien Bacher debian/changelog * releasing package gnome-control-center version 1:45.0-1ubuntu2 * https://deb.li/3S2EE | 10:03 |
lissyx | seb128, looks like there's a regression on mesa 23.0.4.0: https://crash-stats.mozilla.org/signature/?signature=__driDriverGetExtensions_d3d12#aggregations | 10:20 |
lissyx | aggregating on "adapter driver version" reveals only 23.0.4.0 | 10:21 |
seb128 | lissyx, is that ubuntu/snap only? | 10:23 |
lissyx | seb128, I dont think so | 11:03 |
lissyx | seb128, ok it's upstream | 13:34 |
lissyx | seb128, https://bugzilla.mozilla.org/show_bug.cgi?id=1850271#c6 | 13:35 |
-ubottu:#ubuntu-desktop- Mozilla bug 1850271 in Core "Crash in [@ st_ReadPixels]" [S3, New] | 13:35 | |
seb128 | lissyx, ack | 13:36 |
lissyx | seb128, can we also merge https://github.com/canonical/firefox-snap/pull/33 ? | 13:40 |
-ubottu:#ubuntu-desktop- Pull 33 in canonical/firefox-snap "Bug 1855135 - Link with LLD and LTO=cross" [Open] | 13:40 | |
seb128 | lissyx, I'm confused now, didn't we say earlier that we wanted to stop using lto=cross? | 13:44 |
lissyx | seb128, lto=cross with gold | 13:45 |
lissyx | with lld it looks like we are fine | 13:45 |
lissyx | but I can remove it as well ... | 13:45 |
seb128 | well, I'm just not understanding enough those options to have an opinion, but I would expect we want similar build options on the different tracks | 13:45 |
seb128 | no? | 13:45 |
lissyx | seb128, if you prefer, i removed lto=cross | 13:47 |
seb128 | lissyx, what is lto=cross doing and what's the impact on performances or other factors? | 13:47 |
seb128 | also what are the pro and con of building with gold vs lld? | 13:48 |
lissyx | without lld, build is broken | 13:49 |
lissyx | and all linker usage are moving to lld | 13:49 |
lissyx | lto=cross does lto cross language | 13:49 |
lissyx | I dont know the imapct in term of perfs but I can't find us using it in our mozconfig files? | 13:49 |
lissyx | e.g., https://searchfox.org/mozilla-central/search?q=--enable-linker%3D&path=&case=false®exp=false --enable-linker usages | 13:50 |
lissyx | https://searchfox.org/mozilla-central/search?q=--enable-lto%3D&path=&case=false®exp=false | 13:50 |
lissyx | so if it's not part of our own mozconfig ... | 13:50 |
seb128 | ok | 13:51 |
lissyx | lld vs gold because elfhack: https://searchfox.org/mozilla-central/rev/37d9d6f0a77147292a87ab2d7f5906a62644f455/build/unix/mozconfig.linux#18-19 | 13:51 |
seb128 | I don't feel like I understand enough the difference of lld vs gold to be able to give a +1 or -1 :/ | 13:51 |
seb128 | but at the same time I doubt we will have anyone else who will do that review or have an opinion | 13:51 |
lissyx | from https://treeherder.mozilla.org/jobs?repo=try&revision=b6e0072f50325c4ca58ed47ff7864b6d6aab2aed to https://treeherder.mozilla.org/jobs?repo=try&revision=2f0ae43a83621a64488248087ef987e3689097bc | 13:52 |
seb128 | I don't like the idea of having nightly on a different compiler, one of the thing we expect nightly to provide us is to learn about issues earlier | 13:53 |
seb128 | but if we don't use the same toolchain then we will not get notified on issues impacting gold | 13:53 |
lissyx | yet this is also what we do: https://searchfox.org/mozilla-central/source/build/unix/mozconfig.linux#18-19 | 13:56 |
lissyx | anyway the goal is to move to lld everywhere ? | 13:57 |
lissyx | from https://bugzilla.mozilla.org/show_bug.cgi?id=1854047 and https://bugzilla.mozilla.org/show_bug.cgi?id=1839739 | 13:58 |
-ubottu:#ubuntu-desktop- Mozilla bug 1854047 in Firefox Build System "Enable relrhack in local builds when possible" [--, Reopened] | 13:58 | |
-ubottu:#ubuntu-desktop- Mozilla bug 1839739 in Firefox Build System "Switch remaining Linux and Android builds to lld" [S3, Reopened] | 13:58 | |
seb128 | bandali, ^ do you have an opinion on? | 14:03 |
lissyx | seb128, ftr the removal of gold linker was hinted by glandium | 14:13 |
seb128 | lissyx, I would feel better if we were using lld across the channels but that comment state that it's not compatible with the optimisation done in release builds | 14:18 |
lissyx | seb128, the expert regarding linker and elfhack is glandium | 14:18 |
lissyx | when I have a linkage issue and he suggests to remove gold, I listen :p | 14:19 |
seb128 | :) | 14:19 |
lissyx | usage of lld will flow to beta and stable after, I guess | 14:19 |
lissyx | as soon as nightly train advances | 14:19 |
lissyx | and time to leave | 14:20 |
seb128 | talk to you later! | 14:20 |
lissyx | seb128, back from school | 14:39 |
bandali | hi, i don't have the full context sorry, will catch up in a bit. but in general i'm no compile/linking expert, and i'd be very much inclined to listen to glandium's suggestions, they've helped me/us in the past already :-) | 15:21 |
seb128 | bandali, well, I merged the changes from lissyx now | 15:26 |
bandali | seb128, ha okay great, thanks :-) i'm catching up with the context to generally be in the know about it | 15:29 |
KGB-0 | gnome-shell ubuntu/latest 46ceed3 Jeremy Bícha debian/patches/ series layout-Destroy-panel-barrier-on-shutdown.patch overview-Handle-unredirection-in-OverviewShown-state-mach.patch * Cherry-pick 2 commits from future 45.1: * https://deb.li/OCWi | 18:21 |
KGB-0 | gnome-shell ubuntu/latest 07fdf07 Jeremy Bícha debian/gbp.conf * debian/gbp.conf: Branch for mantic * https://deb.li/3Ky8h | 18:21 |
KGB-0 | gnome-shell ubuntu/latest 91c46c0 Jeremy Bícha debian/changelog * releasing package gnome-shell version 45.0-1ubuntu2 * https://deb.li/v1L2 | 18:21 |
KGB-2 | gnome-shell pipeline Jeremy Bicha 586259 * running (extract-source: running; build: created; build i386: created; build source: created; test-build-any: created; test-build-all: created; test-crossbuild-arm64: created; reprotest: created; lintian: created; autopkgtest: created; blhc: created; piuparts: created) | 18:22 |
KGB-2 | gnome-shell pipeline Jeremy Bicha 586257 * [13 minutes and 8 seconds] failed (extract-source: success; build: failed; build i386: failed; build source: success; test-build-any: skipped; test-build-all: skipped; test-crossbuild-arm64: skipped; reprotest: skipped; lintian: skipped; autopkgtest: skipped; blhc: skipped; piuparts: skipped) | 18:34 |
KGB-2 | mutter signed tags 95e8501 Jeremy Bicha ubuntu/45.0-3ubuntu1 * mutter Debian release 45.0-3ubuntu1 * https://deb.li/3Iwln | 22:44 |
KGB-0 | mutter ubuntu/latest 5c073a6 Jeremy Bicha * pushed 38 commits (first 5 follow) * https://deb.li/Mugs | 22:45 |
KGB-0 | mutter ubuntu/latest 83829c3 Simon McVittie debian/changelog * New upstream release * https://deb.li/ikHJU | 22:45 |
KGB-0 | mutter ubuntu/latest 4529b8c Simon McVittie debian/patches/ (6 files in 2 dirs) * d/patches: Drop patches that were applied upstream * https://deb.li/7uQG | 22:45 |
KGB-0 | mutter ubuntu/latest d21b14e Simon McVittie debian/changelog * Update changelog * https://deb.li/3rH9M | 22:45 |
KGB-0 | mutter ubuntu/latest d56a98d Simon McVittie debian/patches/ series x11-events-NULL-Check-input-event-when-getting-name.patch * Don't crash when profiling with sysprof 45~rc (mutter#3030) * https://deb.li/ywSy | 22:45 |
KGB-0 | mutter ubuntu/latest 9ebdd2f Simon McVittie debian/changelog * Update changelog * https://deb.li/3Mfn4 | 22:45 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!