[01:12] StevenK: https://code.launchpad.net/~wgrant/launchpad/bi-dependencies-1/+merge/188489 and https://code.launchpad.net/~wgrant/launchpad/bi-dependencies/+merge/188493 [01:12] Ursinha, wgrant: What do you think about reverting r16775? [01:13] The fix is probably very simple [01:13] Not sure how Ursinha's going with that, though. [01:13] wgrant: https://code.launchpad.net/~stevenk/launchpad/drop-kick-getBugTasks/+merge/188490 [01:13] wgrant: Oh, is that a minus one, as opposed to dash one? [01:13] StevenK: k [01:13] StevenK: Er, yeah, I missed a - [01:14] wgrant, it depends on how much you guys need to deploy new stuff today [01:14] Was meant to be dash-minus-1 [01:14] Ursinha: It would be nice to deploy, but we can wait a bit [01:15] Ursinha: I have my own QA to finish off, so please don't rush on my account. :-) [01:16] okay [01:21] 471+ self.assertEqual(('snap crackle pop', ''), output) [01:21] wgrant: You must answer for your crimes. [01:22] StevenK: Hey, the test had 'pop' in the final position, and there was a hole in the test coverage for buildd_name, so the test string ended up with three elements, 'pop' in the last. My hands were tied. [01:23] wgrant: Tell it to the judge. [01:24] wgrant: r=me for -1, with one niggle [02:23] wgrant, so, PackageUploads that refer to copies don't have a SourcePackageRelease, and attachTranslationFiles is a spr method [02:23] should I move it somewhere else? or add one for copies? [02:30] Ursinha: It might be worth moving somewhere else, but you will still need to find a SourcePackageRelease. [02:30] I suspect. [02:30] Unless you can just use the source name, let me see. [02:31] Hmm [02:31] Ursinha: I'd move attachTranslationFiles elsewhere, yeah. It doesn't really need a SourcePackageRelease; you have all you need on the PackageCopyJob too. [02:31] I'd just put it in a non-method function somewhere, I think. [02:31] StevenK: Thanks. [02:33] wgrant, what I did so far is to add a component_name and target_distroseries properties in PackageUpload, following the pattern of other properties that vary according to them being copies or not [02:34] Ursinha: PackageUpload.distroseries already exists [02:35] wgrant, I saw that, but it made no sense to me that other parts of the code query sourcepackagerelease to know the target [02:35] But you'll need to look up the component name from the relevant SourcePackagePublishingHistory in the target. I'm not sure if there are existing methods to help with that. [02:35] so I thought this might mean something else [02:35] Ursinha: It used to make sense, sort of. [02:35] haha [02:35] This code was already buggy in some copy circumstances. [02:35] can I safely use PackageUpload.distroseries as target then? [02:35] Yes. [02:36] okay. [02:36] PackageUpload.archive, PackageUpload.distroseries, PackageUpload.pocket are accurate [02:36] But you can't use sourcepackagerelease.component. You'll probably need to find the latest SPPH for the relevant (archive, distroseries, pocket, sourcepackagename), and look up the component from that. [02:37] That component lookup was already buggy for some security updates, but it is a lot more important now. [02:37] wgrant, so it was buggy before? [02:38] Yes, for source copies. [02:38] This is probably actually more of a problem than I thought. [02:38] hehe [02:39] Or maybe not. [02:39] The component check only applies to pre-oneiric, so its brokenness isn't too bad. [02:39] But it's still only a few lines to fix, so we probably should. [02:42] StevenK: Do you have a testfix? [03:06] wgrant, should I use the SPR.publishings or query the archive with getPublishedSources? [03:11] Ursinha: You'll have to use getPublishedSources. [03:22] wgrant: Looking now, I was grabbing lunch. [03:22] Sounds like a few more deletions are needed. [03:34] wgrant: testfix == r16787 [04:16] wgrant: With the death of proc_families, should I QA to packagecloner on DF? [04:16] s/to/the/ [04:17] StevenK: Yeah. Can hopefully do just a small PS. [04:18] IDS{,J} got the same treatment, but I don't really want to do that. [04:26] StevenK: IDSJs are easy and cheap to test [04:27] StevenK: Just create one with a tiny PS [04:27] I normally create a µbuntu containing only bzr packages, IIRC [04:33] * StevenK stabs SSO until it bleeds to death [04:35] Why? [04:36] wgrant: Because of the stale page error if you mis-type or don't hold down the yubikey long enough and it puts the default token in, and because once it logs me in it wants me to update my details and there's no way to say "Go away and auth to the site I was originally at" [04:37] Ah, handy [04:37] Filed a bug? [04:42] Not yet [04:42] wgrant, is that ok if I move the content of attachTranslationFiles to run() in the PackageTranslationsUploadJob? having a packageupload at that point is enough (it seems), so I'd provide the packageupload and the libraryfilealias to the job [04:43] Ursinha: It might be useful to have it in a separate function in the job's module so you can continue to test it directly. But otherwise that sounds fine. [04:43] * StevenK tries to remember how to create an IDSJ [04:44] StevenK: Create a new distro, create a series within it, initialise series. [04:46] Hm [04:46] * StevenK anoints himself [04:49] wgrant, okay, done :) now, the importer is currently sourcepackagerelease.creator, who should it be if it's a copy? [04:50] PCJ.requester [04:50] Or something close [04:51] Probably PU.findPersonToNotify [04:51] Which uses PCJ.requester when appropriate [04:51] I think we already use that for the job creator, don't we? [04:51] lib/lp/soyuz/model/packagecopyjob.py doesn't mention findPersonToNotify [04:52] Not PCJ, PTUJ. [04:52] lib/lp/soyuz/model/packagetranslationsuploadjob.py also doesn't mention findPersonToNotify [04:53] And findPersonToNotify wants an SPR, it doesn't deal with PCJs [04:53] blamee = packageupload.findPersonToNotify() [04:53] getUtility(IPackageTranslationsUploadJobSource).create( [04:53] sourcepackagerelease, libraryfilealias, blamee) [04:53] Ah right [04:53] At this point we have an SPPH, though [04:53] wgrant, attachTranslationsUpload uses the requester, it's another thing [04:54] So it'll use SPPH.creator, which should be the PCJ requester. [04:54] I recall the same confusion during the sprint [04:54] it rendered a testfix that time [04:54] :) [04:55] Grr. [04:55] * StevenK pokes around for pubconf [04:55] Ursinha: Use PTUJ.requester, rather than SPR.creator. [04:55] StevenK: Pubconfs can be set in the UI [04:56] wgrant: But the details don't matter? [04:57] StevenK: Not particularly, unless you want to publish it. [05:00] The initialization of Turgid Turtle has been scheduled and should run shortly. [05:00] I blame infinity [05:00] ... [05:00] µbuntu [05:00] Not microbuntu! [05:01] We aren't Fedora [05:01] Our infrastructure likes Unicode. [05:01] wgrant: It didn't for the name [05:01] Not for the name, no. :( [05:02] wgrant: How about that, then [05:02] Superior. [05:12] wgrant: Hm, where do OOPSes go? :-( [05:13] StevenK: Should be /srv/launchpad.net/dogfood-logs/ [05:13] ProgrammingError: permission denied for relation packageupload [05:13] WAT [05:13] Oh [05:13] I was trying to remember why I had that diff on DF [05:13] That was it [05:14] So it was your fault? :-( [05:14] Not exactly, I just locally patched that permission in last time I was testing this [05:14] Because it had broken due to PCJ changes, IIRC. [05:14] Specifically CUC [05:14] I think [05:15] Do we want to land a security.cfg change? [05:15] Test on DF and land, yeah [05:15] Yup [05:15] Digging out the job so I can mangle it back to Waiting [05:15] I'd just create a new one. [05:17] Hm, IDS has SELECT on packageupload* [05:18] Ah [05:18] It wants INSERT on it and PUC [05:24] OH [05:24] And UPDATE [05:24] Damn searchable_names :-P [05:40] wgrant: https://dogfood.launchpad.net/microbuntu/turgid [05:40] wgrant: That's IDSJ run, and then harness'ing up updatePackageCount() [05:41] eh [05:41] Heh [05:41] So you know how I said our Unicode support was good [05:41] https://dogfood.launchpad.net/microbuntu/turgid/+source/2ping [05:42] HAHA [05:43] Fortunately Python 3 prevents people from writing such retarded code. [05:43] wgrant: How tasty are those words you're now eating? [05:44] Not entirely. [05:46] wgrant: Landing security.cfg change [05:47] turgid eh [05:47] lifeless: It was infinity's suggestion for what T would be [05:48] And it's a crappy test series on DF, so meh? :-) [05:48] I think you've been trolled [05:48] wgrant: Are you bouncing the appserver? [05:48] I am [05:48] It didn't come back up... [05:48] Haha [05:50] Hm, that looks better [05:51] wgrant: Facepalm at the diff [05:52] It won't affect many places [05:52] Because .format is backported from 3, it is slightly more Unicode-pedantic. [05:52] % works fine [05:55] wgrant: So I'm happy enough to qa-ok it, any objections? [05:56] Sure [05:56] As long as you've tested the other changed bits [05:57] wgrant: IArchive API bits, editing an archive and changing restricted processors, creating a distribution and doing the same, IDSJ and package cloner [05:57] Right [05:57] Sounds reasoanble. [05:57] I don't think there is anything else [06:26] wgrant: Is http://lpbuildbot.canonical.com/builders/lucid_lp_lxc/builds/892/steps/shell_9/logs/summary your fault? [06:29] StevenK: I don't believe so. [06:29] I haven't touched BuilderSlave, and that one has been occasionally failing 4eva. [10:53] cjwatson, I'm going to take a nap and should return soon, hopefully the fix for bug 1201485 will be able to land today [10:53] <_mup_> Bug #1201485: Need to import translations for the unity daily builds

[10:54] OK, great [10:54] Happy to help reviewing === matsubara is now known as matsubara-lunch === matsubara-lunch is now known as matsubara === rpadovani is now known as WebbyIT === _mup__ is now known as _mup_ === elmo__ is now known as elmo [22:18] wgrant: https://answers.launchpad.net/launchpad/+question/236697 when you get a chance [22:19] wgrant: some of them I can do and am doing [22:19] wgrant: but I think a little help will be needed === Nigel_ is now known as G [23:41] lifeless: Do you just need the tuskar -> tripleo merge done? [23:42] wgrant: and python-tuskarclient reowned [23:44] lifeless: You can't ask the current owner to do that? [23:44] wgrant: sure, they're at an offsite teambuilding thing today :P [23:44] wgrant: but one thing at a time; the team merge first [23:47] lifeless: tuskar has a mailing list. [23:48] gnar [23:48] ok, I'll just manually add the people I guess [23:50] lifeless: API it? [23:50] lifeless: Note that it still owns a couple of trunks, I think. [23:51] wgrant: meh, mirrors from git