| seb128 | bdmurray, hey, I'm unsure how you keep up with SRU bugs comment, so please have another look to bug #1832457 (also would be nice to not block SRUs in such cases when it's clear that the new serie isn't going to miss the update/fix) | 08:33 |
|---|---|---|
| ubottu | bug 1832457 in glib2.0 (Ubuntu Disco) "[SRU] 2.60.4" [Undecided,In progress] https://launchpad.net/bugs/1832457 | 08:33 |
| juliank | This KDE phabricator thing is crazy | 10:19 |
| juliank | When submitting it rewrites your last commit | 10:31 |
| juliank | It's like it's written for a single commit | 10:31 |
| === cpaelzer__ is now known as cpaelzer | ||
| === weedloser is now known as epicman | ||
| === epicman is now known as weedloser | ||
| === weedloser is now known as epicman | ||
| === epicman is now known as eqeqe | ||
| ahasenack | hi, is there precedent in transforming a native ubuntu package into a non-native in ubuntu? Both in a devel release, and in srus? | 13:00 |
| ahasenack | is that frowned upon? | 13:00 |
| ahasenack | by sru, I mean a package that is native currently in a stable release, becoming non-native via an sru (there are other changes in the sru though, not just this) | 13:01 |
| === ricab is now known as ricab|lunch | ||
| ximion | hey Laney :) | 14:08 |
| ximion | I was just made aware that the LMDB dependency of AppStream is actually an issue for Ubuntu, as LMDB isn't in main (yet?) | 14:09 |
| Laney | correct | 14:10 |
| * Laney has tried to shift the paperwork onto acheronuk | 14:12 | |
| ximion | Laney: of course in my opinion LMDB should be in main (you could enable it for bind9, postfix etc. as well then), but if there's anything I can help with, let me know | 14:17 |
| ximion | (in the very worst case, AS could also run without cache, but that would be significant effort to make it work for terrible performance and memory usage) | 14:17 |
| ximion | fortunately, LMDB is tiny, depending only on libc | 14:18 |
| Laney | usually the main problem is if the MIR team decides to ask for a security review | 14:18 |
| Laney | then it can take a while | 14:18 |
| Laney | ximion: it's fine, we have this semi frequently, but it may delay any updates to appstream for a long while | 14:23 |
| Laney | so in general (if you want to care about this situation), making new runtime deps conditional is nice | 14:24 |
| ximion | Laney: I don't forsee any additional dependencies for AppStream - the library should really be quite small, all the annoying stuff (like font rendering and image scaling) should end up in asgen anyway | 14:26 |
| ximion | a security review is sensible - I was just looking on whether RHEL has it yet, but it's only in EPEL there, at least for the current version | 14:26 |
| Laney | I think suse has a similar procedure? | 14:27 |
| ximion | jup - SUSE already has LMDB in the trusted zone (at least for openSUSE) though due to KDE depending on it | 14:29 |
| ximion | (also RPM has a LMDB backend) | 14:29 |
| ximion | I couldn't find any info on what SLE does though, which is more relevant | 14:29 |
| sil2100 | Laney: hmmm, it looks like britney dies for xenial since over a week: https://people.canonical.com/~ubuntu-archive/proposed-migration/log/xenial/2019-06-19/13:46:56.log | 14:36 |
| sil2100 | Laney: do you, by any chance, know this part of britney? | 14:37 |
| Laney | sil2100: not off the top of my head, but I think we have a FauxPackage for snapd if the message just before the traceback is relevant | 14:38 |
| Laney | those are in b1 | 14:38 |
| sil2100 | Laney: ah, indeed! You think removing the faux package might help? I guess I can try that, since snapd now builds on powerpc | 14:43 |
| sil2100 | (at least it provides 'some' powerpc binaries) | 14:43 |
| Laney | is that what changed around the time it started breaking? | 14:43 |
| sil2100 | Laney: yeah, that would fit, since the new snapd landed on the 11th, which was when the last time britney ran without crashing on xenial | 14:44 |
| * sil2100 reverts and hopes for the best | 14:44 | |
| sil2100 | Laney: do you know if such britney crashes are sent somewhere when they happen? Like some notification | 14:46 |
| Laney | sil2100: probably the mailbox on snakefruit that nobody reads | 14:47 |
| Laney | We just find out when someone notices :( | 14:47 |
| === ricab|lunch is now known as ricab | ||
| sil2100 | vorlon: hey! Do you know what could be the reason for this? https://people.canonical.com/~ubuntu-archive/proposed-migration/log/xenial/2019-06-19/16:09:59.log | 16:16 |
| sil2100 | I removed snapd from the FauxPackages because britney on xenial was crashing | 16:16 |
| sil2100 | And now it's just crashing totally | 16:17 |
| vorlon | whee | 16:17 |
| vorlon | checking | 16:17 |
| sil2100 | I feel strange because I thought I just reverted the snapd addition from some time ago ;p | 16:18 |
| vorlon | sil2100: did you make your change locally or in the bzr branch? | 16:19 |
| vorlon | because there's a conflict in the checkout | 16:19 |
| sil2100 | vorlon: ugh, I did a fresh bzr branch of lp:~ubuntu-release/britney/britney1-ubuntu, modified and bzr pushed | 16:20 |
| sil2100 | https://code.launchpad.net/~ubuntu-release/britney/britney1-ubuntu | 16:21 |
| vorlon | rather strange, the branch history looks like all the changes should've been commits, not cowboys | 16:21 |
| vorlon | anyway, I've resolved it now | 16:21 |
| Laney | bet someone cowboyed and then committed the same thing | 16:22 |
| sil2100 | I only pushed this to the bzr branch | 16:22 |
| sil2100 | Can we kick britney for xenial somehow manually? Or should I find a xenial kernel to publish to -proposed ;p? | 16:22 |
| sil2100 | Since I wanted to see if this change makes xenial not-crash | 16:22 |
| sil2100 | vorlon: I already poked La_ney about that, but I guess it would be nice for us to get some kind of a notification whenever britney crashes, or maybe even something more sublime like 'no successful britney run after x days' or something | 16:27 |
| sil2100 | vorlon: I guess I could write something up quickly for our stable series | 16:27 |
| sil2100 | Since for devel it's easy to notice, but for stable we rarely look directly at excuses | 16:28 |
| sil2100 | Right now for instance, because of this we could have published some xenial updates that didn't have their autopkgtests ran and might have regressed | 16:28 |
| sil2100 | And it's the second time I see such a situation | 16:28 |
| sil2100 | Anyway, let me card it | 16:28 |
| sil2100 | vorlon: hmmm, so britney now is crashing for gnome-shell apparently - I wonder if maybe the gnome-shell fauxpkg entry should also be checked? | 16:54 |
| sil2100 | vorlon: I wonder though why it is failing now suddenly, but maybe there were some local changes cowboyed that now got removed? | 16:55 |
| sil2100 | vorlon: latest log: https://people.canonical.com/~ubuntu-archive/proposed-migration/log/xenial/2019-06-19/16:45:31.log | 16:55 |
| sil2100 | vorlon: what were the changes that were conflicting in the production branch? Was that something that could explain the issue here? | 16:56 |
| sil2100 | Laney: ^ you're probably EOD as well, right? | 17:01 |
| sil2100 | Do we have fauxpackages per-series? | 17:02 |
| vorlon | sil2100: the conflicts were between snapd being shown as locally added and gnome-shell being added to the merge source, which didn't make any sense in the end | 17:12 |
| vorlon | sil2100: is it possible it's crashing because gnome-shell/s390x exists as a real package in xenial? | 17:13 |
| vorlon | sil2100: otoh gnome-shell has been in fauxpackages since april, without problems | 17:15 |
| vorlon | sil2100: this is what ends up in xenial-proposed/Packages_s390x; I don't understand why the architecture string is what it is, I don't recall this detail of fauxpackages | 17:34 |
| vorlon | Section: faux | 17:34 |
| vorlon | Version: 3.18.5-0ubuntu0.3 | 17:34 |
| vorlon | Architecture: all | 17:34 |
| vorlon | Package: gnome-shell | 17:34 |
| * Faux twitches again. | 17:35 | |
| xnox | Faux: sorry for all the highlights! | 21:59 |
| infinity | vorlon: Oh, I just noticed the backscroll WRT fauxpackages. I did indeed cowboy removal of the gnome-shell one when eoan opened and we realised it was breaking xenial. I should have committed that, but I was hoping to dig deeper and figure out why. | 22:00 |
| infinity | I think. | 22:01 |
| infinity | Memory is fuzzy. | 22:01 |
| vorlon | ah | 22:01 |
| vorlon | Laney: ^^ | 22:01 |
| Laney | makes sense | 22:01 |
| Laney | so let's commit removing that | 22:01 |
| infinity | I think I may have cowboyed instead of comitted because I feared I'd need to flip it back in and out if it was needed for a larger migration. | 22:01 |
| infinity | (but also, Debian seemed on the cusp of fixing gnome-shell on s390x at that point) | 22:02 |
| infinity | Anyhow, removing it should just render a couple of things uninstallable on s390x-only, which is fine. If it causes migration issues for newer versions of those same things, I'll reevaluate. | 22:04 |
| infinity | The real problem here comes from our clever use of one britney base and multiple config files to handle N releases. Maybe we could extend FauxPackages to have a series filter, if we cared, but we use it so infrequently... | 22:06 |
| infinity | Laney: Did you want me to commit the revert, or was "let's" you saying you were doing so? | 22:07 |
| Laney | infinity: Go for it | 22:07 |
| infinity | Laney: Done. | 22:08 |
| Laney | Ta | 22:08 |
| GunnarHj | bdmurray: Still there? | 23:43 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!