| enr0n | @pilot out | 00:47 |
|---|---|---|
| === ChanServ changed the topic of #ubuntu-devel to: Archive: Open | Devel of Ubuntu (not support) | Build failures: http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of Focal-Oracular | Patch Pilots: schopin | ||
| === tsimonq2_ is now known as tsimonq2 | ||
| schopin | @pilot out | 08:36 |
| === ChanServ changed the topic of #ubuntu-devel to: Archive: Open | Devel of Ubuntu (not support) | Build failures: http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of Focal-Oracular | Patch Pilots: N/A | ||
| adrien | I'm looking at migration of coq packages: it seems they have a random suffix which I guess is determined at build time and stored as a Provides: and I guess that when a dependant package is built, it depends not on the Package: but on the Provides: | 09:10 |
| adrien | but Build-Depends is on the Package: | 09:10 |
| adrien | and that would mean that the result depends on the build order | 09:11 |
| adrien | can someone confirm or infirm that? | 09:11 |
| adrien | and in that case, what should be done? no-change rebuilds I guess, but is there a rule for which packages to rebuild and in which order then? | 09:12 |
| adrien | hmm, I see something different: looks like you need to craft triggers but then you need to find the right ones due to the dynamic Provides! | 09:38 |
| paride | adrien, autopkgtest does retry to install the test deps with "all proposed", and that normally should workaround that kind of issue | 10:22 |
| paride | adrien, however it does not seems to be sufficient in this specific case | 10:22 |
| paride | adrien, I need to verify a bit more in detail what is happening | 10:23 |
| adrien | as I mentioned elsewhere, coq-hott/8.20-1build1 coq/8.20.0+dfsg-1 2024-11-21 09:15:36 UTC 0h 02m 57s - pass 56c4953a-afc4-4a00-b291-927dcf236133 log artifacts | 10:23 |
| adrien | erf | 10:23 |
| adrien | https://autopkgtest.ubuntu.com/packages/c/coq-equations/plucky/arm64 shows that coq-hott was added as a trigger and that's a good move but now I'm wondering who did it and how | 10:24 |
| adrien | I think the easiest/clearest/simplest rule is that if coq triggers a coq-equations test, you look at coq-equations and see it depends on coq-hott, so you trigger with the added dependency | 10:25 |
| adrien | (I need to afk for a bit) | 10:26 |
| paride | adrien, that was done by britney | 10:26 |
| paride | adrien, but you are asking, "why not in the first try?". maybe there were missing binaries? | 10:26 |
| adrien | oh? automatic retries by britney then? | 10:26 |
| adrien | well, first try would be good but second or third is already pretty good! | 10:27 |
| adrien | (I really need to afk but I'll be back in an hour or so) | 10:27 |
| paride | not "true" automatic retries of failures, but looks like something changed (a new binary got published?) and that made britney re-queue a test | 10:28 |
| paride | (true auto-retries of failures are currently being worked on) | 10:28 |
| adrien | I've seen coq-equations/coq-hott tests succeed | 10:28 |
| adrien | I had no idea that could happen but that can make sense | 10:29 |
| bdrung | @pilot in | 12:17 |
| === ChanServ changed the topic of #ubuntu-devel to: Archive: Open | Devel of Ubuntu (not support) | Build failures: http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of Focal-Oracular | Patch Pilots: bdrung | ||
| brlin | bdrung: Hello, I have submitted a debdiff to Bug #2076869 and would like to ask if there's anything I can do at my end to accelerate the process to have it merged? | 12:49 |
| -ubottu:#ubuntu-devel- Bug 2076869 in mutter (Ubuntu Noble) "When changing file or folder names using the Nautilus GUI, Chinese/Japanese input using the IBus input method isn't possible." [Medium, Triaged] https://launchpad.net/bugs/2076869 | 12:49 | |
| jbicha | adrien: for building coq packages for a transition, you just need to follow the order at https://ubuntu-archive-team.ubuntu.com/transitions/html/coq.html | 12:58 |
| jbicha | if you hover over coq-deriving, it says it needs ssreflect to be built (and published) first | 12:58 |
| jbicha | the coq packages also show on the ocaml transition page but the ocaml transition page doesn't have the correct build order | 12:59 |
| bdrung | brlin, pinging the patch pilot is a way to speed up the process. i'll have a look. | 13:01 |
| brlin | bdrung: Thanks for the help, appreciated! | 13:03 |
| adrien | jbicha: interesting, thanks; does that mean that coq packages sometimes need to be rebuild or is it somehow always fine (maybe because the uploads in debian are sorted too and delayed until their dependencies have built)? | 13:06 |
| ginggs | adrien: launchpad autosyncs from debian every six hours in batches, so there's no guarantee things will be built and published in the correct order here | 13:16 |
| ginggs | for added fun, we can get version skew across architectures | 13:17 |
| adrien | Indeed, thanks for the precisions. Are these version skews something that only manifests (and lasts I guess) for coq, or for basically any package? | 13:18 |
| jbicha | I took care of coq level 2 yesterday and dok_o did rebuilds for level 3 today | 13:19 |
| jbicha | this kind of problem is particularly common for the things listed as permanent trackers at https://ubuntu-archive-team.ubuntu.com/transitions/ and fairly rare elsewhere | 13:20 |
| adrien | it makes a lot of sense but I had never read these things and didn't know they had to be handled in practice, thanks for all the explanations :) | 13:21 |
| * adrien will have a lengthy +1 report | 13:21 | |
| bdrung | brlin, https://bugs.launchpad.net/nautilus/+bug/2076869/comments/10 - are you okay with the changes I did to the debdiff? If yes, I'll sponsor it in this version. | 13:37 |
| -ubottu:#ubuntu-devel- Launchpad bug 2076869 in mutter (Ubuntu Noble) "When changing file or folder names using the Nautilus GUI, Chinese/Japanese input using the IBus input method isn't possible." [Medium, Triaged] | 13:37 | |
| bdrung | my local test build was just successful. | 13:37 |
| brlin | bdrung: I'm fine with the change, thanks! | 13:38 |
| bdrung | brlin, and upload. thanks for your contribution | 13:48 |
| brlin | bdrung: Again, thanks for the help! mOwOm | 13:49 |
| brlin | I suppose the next step is to verify the build in the noble-proposed pocket? | 13:51 |
| bdrung | brlin, that is step 2. the next step is that a SRU member accepts the upload. | 13:54 |
| brlin | bdrung: Interesting...thanks for the info. | 13:55 |
| dviererbe | tsimonq2: Hey, I want to fix pinentry, which stuck in -proposed for a month. As you are the last person that touched pinentry, could you review and sponsor my merge proposal? https://code.launchpad.net/~dviererbe/ubuntu/+source/pinentry/+git/pinentry/+merge/476772 | 14:32 |
| adrien | ginggs: in https://bugs.launchpad.net/ubuntu/+source/pmix/+bug/2078275 , I see that you've removed pmix binaries on armhf but I also see that the current pmix upload in proposed attempts to build them. Is the rationale that with no previous build on armhf, this "Dependency wait" will have no impact in practice? | 14:58 |
| -ubottu:#ubuntu-devel- Launchpad bug 2078275 in pmix (Ubuntu) "Please remove pmix on 32-bit architectures" [Undecided, Fix Released] | 14:58 | |
| ginggs | adrien: yes, there will be no impact. 'rmadison libpmix-bin' shows no armhf binaries in oracular and plucky | 15:01 |
| ginggs | the deb-wait is on architecture-is-64-bit, which by definition does not exist on 32-bit architectures | 15:02 |
| ginggs | dep-wait, even | 15:02 |
| adrien | It's somewhat confusing to see LP attempt to build it but I see how it makes sense. I don't have all the reflexes yet. | 15:03 |
| === pushkarnk1 is now known as pushkarnk | ||
| bdrung | @pilot out | 16:33 |
| === ChanServ changed the topic of #ubuntu-devel to: Archive: Open | Devel of Ubuntu (not support) | Build failures: http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of Focal-Oracular | Patch Pilots: N/A | ||
| juliank | LocutusOfBorg: so we just looked at the libgit2 llhttp bug and it seems we have a green light to now do the vendoring of llhttp from the security team in https://bugs.launchpad.net/ubuntu/+source/libgit2/+bug/2080872 | 16:52 |
| -ubottu:#ubuntu-devel- Launchpad bug 2080872 in node-undici (Ubuntu) "libgit2: replace unmaintained http-parser dependency with llhttp" [Undecided, New] | 16:52 | |
| juliank | LocutusOfBorg: Stands to reason maybe we should have some fake Build-Using-like thing to make sure we keep track of it | 16:54 |
| UnivrslSuprBox | juliank: The security team tracks vendored copies of other software, like in Firefox and Thunderbird (before they were snaps). How do they do that? | 16:58 |
| enr0n | ahasenack: so regarding the openssh blocker to noble-updates (bug 2087551), what bar do we need to clear to be able to progress the SRU? As it stands, none of the reporters have been able to give enough information to reproduce, and I have tried many different configurations. | 20:30 |
| -ubottu:#ubuntu-devel- Bug 2087551 in openssh (Ubuntu) "OpenSSH server config broken on unattended update" [Critical, Incomplete] https://launchpad.net/bugs/2087551 | 20:30 | |
| enr0n | I don't want to release something broken of course, but the SRU contains some important fixes, so I don't want to hold it forever either | 20:31 |
| tsimonq2 | dviererbe: Retroactively, LGTM. Thanks for the ping! | 20:49 |
| tsimonq2 | bdrung: Sorry to step on your toes; I uploaded a fix for bug 1596220 last night, plymouth's fix migrated but it looks like you already had a dracut upload ready. | 20:50 |
| -ubottu:#ubuntu-devel- Bug 1596220 in dracut (Ubuntu) "plymouth-set-default-theme is missing - trigger dracut problem" [High, Confirmed] https://launchpad.net/bugs/1596220 | 20:50 | |
| tsimonq2 | The dracut patch fixing it shouldn't be forwarded to Debian anyway. :) | 20:51 |
| ahasenack | enr0n: I understand, but we had multiple people comment on that bug, it's quite scary to have an openssh update leave the system unreachable | 20:51 |
| vpa1977 | @pilot in | 20:51 |
| === ChanServ changed the topic of #ubuntu-devel to: Archive: Open | Devel of Ubuntu (not support) | Build failures: http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of Focal-Oracular | Patch Pilots: vpa1977 | ||
| ahasenack | ahasenack: the only other thing I can think of is that their system was already broken, and that the restart just made it obvious | 20:51 |
| ahasenack | like when you make a config change but don't restart the service. You will only notice the next time it's restarted, which could be days later | 20:52 |
| ahasenack | enr0n: perhaps can you bring up this possibility in the bug, see how they respond? | 20:52 |
| enr0n | ahasenack: sure, that seems worth asking | 20:52 |
| enr0n | ahasenack: but yeah generally I am just trying to understand *what* we need to do to gain confidence in the update. From your perspective, is anything besides "find a reproducer and fix the root cause" acceptable? | 20:55 |
| ahasenack | enr0n: we won't get new reports in that bug likely because the update has been moved back to proposed | 20:56 |
| ahasenack | enr0n: so that data is what we have to work with | 20:56 |
| ahasenack | how hard? I would like to give it a try, but didn't have time this week | 20:56 |
| ahasenack | something else we could do is split the update | 20:56 |
| ahasenack | instead of fixing 4 in one update, go one by one | 20:56 |
| ahasenack | another thing still, is to ask those who were affected to retry with the package that is in proposed right now, see if it triggers the problem again | 20:57 |
| ahasenack | then zero in on that | 20:57 |
| enr0n | doing incremental updates sounds reasonable to me | 20:59 |
| enr0n | I think those affected would still have the -proposed version on their systems, unless they purged it manually, no? | 21:00 |
| ahasenack | maybe, I guess some disabled socket activation | 21:00 |
| ahasenack | we could ask them to revert and retry the proposed one | 21:00 |
| enr0n | I had one of them restore socket activation and it worked from the sounds of it, but I will post that again. Hopefully someone responds | 21:03 |
| enr0n | I guess I can try spinning up a bunch of containers and enable noble-proposed for unattended-upgrades, and see if I hit it. Maybe it's not consistently reproducible | 21:06 |
| ahasenack | fwiw, I updated my systems when it was in proposed, nothing broke, but I also don't have custom port configurations | 21:06 |
| jbicha | tsimonq2: can we revert the indi transition? I think this is blocking Edubuntu daily ISO builds because stellarium doesn't build with the new indi & indi is stuck in proposed anyway | 21:24 |
| tsimonq2 | jbicha: Sure, we can do that | 21:24 |
| bdrung | tsimonq2, sorry for stepping on your toe. I did not check for upload in between when I did my upload. your plymouth fix looks good. | 21:59 |
| bdrung | tsimonq2, my wish for upstream is to replace the plymouth module by an improved variant. see https://github.com/dracut-ng/dracut-ng/pull/442 | 22:06 |
| -ubottu:#ubuntu-devel- Pull 442 in dracut-ng/dracut-ng "feat(plymouth): install plymouth files via dracut-install" [Open] | 22:06 | |
| bdrung | tsimonq2, https://bugs.launchpad.net/ubuntu/+source/plymouth/+bug/1596220/comments/6 - i failed to get the plymouth theme displayed with this plymouth module above ^ | 22:07 |
| -ubottu:#ubuntu-devel- Launchpad bug 1596220 in dracut (Ubuntu) "plymouth-set-default-theme is missing - trigger dracut problem" [High, Confirmed] | 22:07 | |
| bdrung | so i am interested in any analysis results | 22:08 |
| tsimonq2 | bdrung: No worries, I'll continue working at it and let you know what my findings are. | 22:14 |
| tsimonq2 | Thanks for that PR link, it seems to align with what I've been thinking so far. | 22:14 |
| tsimonq2 | bdrung: At the very least, it doesn't *fail* now, so a step forward :) | 22:15 |
| Eickmeyer | jbicha, tsimonq2: That's not the reason. It's the exiv2transition blocking, which has a few moving pieces. vorlon had mentioned a hint, but apparently boost and openmpi need to migrate as well. | 22:28 |
| sudip | I think exiv2 now has two revdeps left, stellarium is blocked by indi and python3-py3exiv2 is by boost. | 22:42 |
| sudip | iiuc, python3-py3exiv2 can be removed from release, it has no reverse dependency. but stellarium is the main blocker | 22:44 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!