mwhudson | vorlon: could you look at the r-cran-vctrs autopkgtest failures on s390x? i think that's all that's holding up a bunch of r stuff | 02:56 |
---|---|---|
* mwhudson gives it a retry to make sure it's still broken | 02:56 | |
vorlon | mwhudson: hmm, won't be tonight | 02:57 |
mwhudson | vorlon: sure :) | 02:57 |
mwhudson | although https://github.com/r-lib/vctrs/issues/1353#issuecomment-814097160 | 02:58 |
ubottu | Issue 1353 in r-lib/vctrs "test-partial-factor fails on big-endian machines" [Open] | 02:58 |
apw | are we aware of sound issues (no sources listed in gnome-settings) with jammy today? | 10:53 |
seb128 | apw, no, could you share your recent apt log to see what changed? | 11:29 |
apw | seb128, seems that restarting the various daemons made the sources and sinks come back; so i assume this is an upgrade order issue | 11:30 |
apw | likely we should make the transition request a reboot for sanity. | 11:31 |
seb128 | apw, not really, pipewire isn't supposed to take over sound. Likely you got pipewire-pulse installed by some recent upgrade? | 11:31 |
seb128 | which would be a bug | 11:31 |
apw | i cirtainly have pipewire-pulse installed; it does at least work :) | 11:32 |
seb128 | apw, could you check the apt log to see if it was just pulled in for you? | 11:32 |
seb128 | or is that something you opted in to install before? | 11:32 |
apw | seb128, it seems it was requested by the ubuntu-desktop^ task. | 11:34 |
apw | i routinly install that after a large dist-upgrade to avoid losing thing necessary for reboot | 11:35 |
apw | https://paste.ubuntu.com/p/xbrv4knXhk/ | 11:35 |
apw | seb128, ^ | 11:35 |
seb128 | apw, thanks, it's basically http://bugs.debian.org/999363 | 11:37 |
ubottu | Debian bug 999363 in wireplumber "Shouldn't replace pulseaudio by pipewire by default" [Normal, Open] | 11:37 |
seb128 | apw, I though we wouldn't be impacted yet since wireplumber is still in universe but that's enabled for most users upgrading | 11:37 |
apw | ahh so you don't see it in install testing, but only upgraders. /me enjoys some dogfood. | 11:38 |
seb128 | indeed | 11:38 |
seb128 | also new install would probably work from an user perspective because you don't have the issue of having the sound server changing under you from pulseaudio to pipewire in a session | 11:39 |
apw | well i guess i get to test pipewire-pulse as a thing for a bit; other than the upgrade being "poor", the experience audio wise is ok | 11:39 |
seb128 | apw, anyway, the issue is tracked as part of bug #1949776 for us | 11:39 |
ubottu | Bug 1949776 in wireplumber (Ubuntu) "[MIR] wireplumber" [Undecided, Incomplete] https://launchpad.net/bugs/1949776 | 11:39 |
apw | ack thank. | 11:39 |
seb128 | np, thank you for mentioning it, it's always useful to have issues reported | 11:40 |
ijohnson[m] | slyon: hey when you have a chance could you take a look at this netplan + ubuntu core related problem on the forum: https://forum.snapcraft.io/t/ubuntu-core-20-boot-delayed-by-missing-wi-fi-usb-dongle-despite-optional-true-netplan-config/27403/1 | 13:39 |
slyon | ijohnson[m]: Yes, I will have a look later! | 13:50 |
ijohnson[m] | thank you | 13:54 |
seb128 | bdmurray, hey, iso tracker question, how does one delete a testcase from a testsuite? example, http://iso.qa.ubuntu.com/admin/config/services/qatracker/testsuites/410/edit , I want to remove the live session test from that set | 15:11 |
bdmurray | seb128: Edit then change Status to Disabled should do the trick | 15:17 |
seb128 | bdmurray, ah, thanks, a bit confusing :) | 15:18 |
schopin | Is there an easy way to know if a package is involved in a transition? I'm looking into merging libxmlb, which bumped soname, meaning a small transition (3 rdeps), but I'd like to make sure that this wouldn't conflict with other stuff. | 16:09 |
vorlon | schopin: 'apt rdepends $soname' | 16:16 |
vorlon | I also have a 'misc-transition-script' here that I should clean up and publish, its --dry-run mode would be useful for this | 16:17 |
schopin | That much I have, yes. I was more thinking of something akin to the mention on tracker.debian.org which displays if a package is part of a planned transition. | 16:18 |
vorlon | schopin: ah - the approximation I use is checking whether it's currently in -proposed | 16:18 |
vorlon | (the transition tracker is also written in ocaml, I avoid it whenever I can ;) | 16:19 |
seb128 | I was going to say, check if it's proposed and what is listing it on update_excuses | 16:19 |
vorlon | schopin: here's my terrible script, if it would be useful to you: https://paste.ubuntu.com/p/vr5p2R62Xw/ | 16:28 |
schopin | thanks! | 16:30 |
vorlon | (needs improvement: auto-detect the devel release; take a list of package exclusions as commandline arguments; take a changelog override as commandline argument; then I would push it somewhere public and maintained) | 16:32 |
teward | ddstreet: fyi i'm still wanting to backport vmfs6-tools because its needed in 18.04. but i'm maintianer for that in Debian too so | 16:45 |
vorlon | schopin: now with those improvements: https://paste.ubuntu.com/p/WzRSFrpSCs/ | 17:01 |
=== genii-core is now known as genii | ||
vorlon | has anyone written an apt-bisect? Would be useful... | 20:29 |
bdmurray | I'm not certian I know exactly what you might but bryyce might have done something like what I think you mean | 20:33 |
bdmurray | s/might/mean/ | 20:33 |
sarnold | you could probably get there with an aptly mirror or snapshots.debian.org without too much trouble. given how quickly ubuntu reaps replaced packages off the mirrors I think it'd be pretty hard to do over here | 20:33 |
vorlon | bdmurray: libpillowfight/amd64 has some kind of misbuild; the last good binary was built in the middle of the hirsute release; building it on released hirsute fails, building it on released groovy succeeds - bisect the apt dist-upgrade | 20:34 |
vorlon | bryyce: ? | 20:34 |
vorlon | sarnold: in this case I would be happy just bisecting the set of packages selected for upgrade between the releases | 20:35 |
vorlon | ... beacuse it doesn't make sense why *any* of the packages would have broken the build | 20:36 |
sarnold | vorlon: oh! that's a different angle than I expected :) yeah that'd be useful. | 20:36 |
vorlon | ok just confirmed it wasn't linux-libc-dev that broke it, phew | 20:37 |
vorlon | at this point I'm assuming what's going to happen is I'm going to upgrade ALL of the packages on top of groovy and it's still going to rebuild successfully | 20:38 |
bryyce | vorlon, I do have a menagerie of scripts for the ppa transition but alas no equivalent of an apt-bisect which does sound handy | 20:48 |
vorlon | good news, I reproduced the failure! bad news, the set of packages I upgraded that triggered it are crazypants | 20:49 |
bryyce | I'd also love a cli to generate a listing of packages involved in a transition ala the transition tracker, although I find the tracker's list is more of a starting point; there's more packages involved than that | 20:49 |
vorlon | and the winner is.... <drumroll> | 20:55 |
vorlon | dpkg-dev | 20:55 |
vorlon | a dpkg-dev upgrade results in a misbuild of python3-pypillowfight | 20:56 |
vorlon | ... because LTO \o/ | 21:37 |
sarnold | oh boy oh boy :) | 21:38 |
vorlon | doko: so libpillowfight regresses on amd64 due to LTO, only detected at runtime by the autopkgtests. Aside from turning off LTO in the package build, what do you want done with this? | 21:49 |
lucas | vorlon: bisecting by timestamps using snapshot.debian.org works reasonably well (for Debian) | 21:50 |
* vorlon nods | 21:50 | |
sarnold | there's a lto-disabled-list package that may need some updates: https://wiki.ubuntu.com/ToolChain/LTO | 21:54 |
vorlon | sarnold: my understanding from doko's email to the list is this was a temporary measure to be superseded by per-package overrides anyway | 21:56 |
sarnold | vorlon: aha | 21:57 |
Eickmeyer[m] | Indeed. I have a couple of packages that require overrides on LTO (the upstream hates the very idea of LTO). | 22:01 |
bryyce | dovecot has optimize=-lto, but the new version from debian fails its tests due to lto issues, even with that. | 23:00 |
bryyce | in this case, looks like the dovecot maintainers enabled it (adding -flto), as it worked with gcc 10. Doesn't work with gcc 11 though. | 23:03 |
vorlon | mwhudson: https://launchpad.net/ubuntu/+source/r-cran-vctrs/0.3.8-1ubuntu1 you made me write in R >_< | 23:31 |
sarnold | that's the most R i've ever seen in one place before | 23:33 |
vorlon | a worse language to google about than go | 23:35 |
sarnold | oh my | 23:36 |
vorlon | hmph, just found a bug in the patch, ohwell | 23:36 |
vorlon | (the bug does not stop the tests from passing) | 23:36 |
=== genii is now known as genii-core |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!