=== Ursinha-afk is now known as Ursinha [00:55] wgrant: will look for you in a minute [00:56] mars: oh, you're gone. [01:19] wgrant: cp has been done [01:19] "Julians: Fixes critical bug where arch-independent files are [01:19] getting erroneously superseded across series boundaries, resulting in [01:19] loss of data for PPA users and incorrect package status in Ubuntu [01:19] SteveKs: Don't copy disabled DASes to the child distroseries during [01:19] InitialiseDistroSeries." [01:19] lifeless: They're both on cocoplum and germanium? [01:21] dunno [04:12] Project devel build (95): STILL FAILING in 4 hr 1 min: https://hudson.wedontsleep.org/job/devel/95/ [04:15] !! [04:15] \o/ [04:16] me likey the shouting [04:17] It can also do e-mail, a'la buildbot, I've just not configured it [04:17] .oO a hudson [04:17] StevenK: I like twitter myself [04:17] * StevenK is ignoring twitter in the hopes it all just goes away [04:17] StevenK: we're organising openid [04:18] lifeless: For 1.378 or 'soon'? [04:18] soon [04:18] OpenID for what? [04:18] hudson [04:19] StevenK: please pick a different irc nick for our hudson [04:19] It doesn't support it? That's a bit odd. [04:19] lifeless: Suggestions? [04:19] lpci ? [04:19] 42? [04:19] 101010? [04:19] back soon === Ursinha is now known as Ursinha-afk [04:26] Project db-devel build (53): STILL FAILING in 15 min: https://hudson.wedontsleep.org/job/db-devel/53/ [04:27] Oh, damn, I know why [04:29] Project db-devel build (54): STILL FAILING in 32 sec: https://hudson.wedontsleep.org/job/db-devel/54/ [04:40] StevenK: you are going to install the Chuck Norris plugin for the Hudson instance aren't you? :-) [04:41] http://wiki.hudson-ci.org/display/HUDSON/ChuckNorris+Plugin [04:43] I wasn't planning on :-) [04:47] but it's soooo cool [04:49] that default butler guy is too boring. chuck makes it a lot more interesting [05:22] brucie! [05:24] * StevenK blinks at subunit [05:31] StevenK: the bruce schnier plugin ftw [05:32] Both of which are completly pointless? :-) === almaisan-away is now known as al-maisan [06:20] wallyworld: you have mail [06:24] lifeless: thanks [06:26] lifeless: thanks for your post about a flags remote api; but let's put it on that bug? [06:26] that mp is big enough already [06:28] poolie: sorry :) [06:28] np :) [06:28] it's an interesting point [06:28] i just don't want it lost [06:28] poolie: perhaps you could copy it across? [06:28] sure :) [06:28] thank you [06:33] https://bugs.edge.launchpad.net/launchpad-foundations/+bug/656031 in case you think of anything else or want to prioritize it [06:33] <_mup_> Bug #656031: want non-storm way to get feature rules [06:39] lifeless: Are you up for a quick testfix review for db-devel? [06:41] StevenK: You wouldn't happen to know the publisher code, would you? [06:42] wgrant: What about it? [06:43] StevenK: I think the way that it determines which pockets and archs to generate a Release file for is far more complicated than it needs to be. [06:43] And I'm hoping that someone who still exists has dealt with it before. [06:43] But I doubt it :( [06:43] I suspect Julian, maybe al-maisan [06:44] lifeless: well, 5th time lucky on that sql logging branch. it's in the hands of the ec2 gods again. i've made a sacrifice on an old windows pc to appease them [06:45] Speaking of landing... I have a branch that ec2 has thrown a fit at three times. Could someone try it again? [06:45] https://code.edge.launchpad.net/~wgrant/launchpad/bug-629921-packages-empty-filter/+merge/37339 [06:47] wgrant: Doing so [06:47] StevenK: Thanks. [06:56] StevenK: url [06:57] lifeless: https://code.edge.launchpad.net/~stevenk/launchpad/db-fix-disabled-dases-test/+merge/37819 [06:57] doesn't edwin have the same thing up for review? [06:58] His doesn't do the clean-up that mine does [06:58] (And I'm all for making tests shorter) [07:01] lifeless: And I already spoke to Edwin about it in -reviews [07:04] so InitialiseDistroSeries was the helper? [07:05] lifeless: The helper removed was _create_distroseries() [07:05] StevenK: does it work? [07:06] lifeless: Yes, I verified the test failed before fixing it, and that it passed after fixing it. [07:38] Project devel build (96): STILL FAILING in 3 hr 25 min: https://hudson.wedontsleep.org/job/devel/96/ [07:38] * Launchpad Patch Queue Manager: [r=EdwinGrubbs][ui=none][bug 645702] Forward small messages to the [07:38] moderation queue. [07:38] <_mup_> Bug #645702: oops in holdMessage storing large message [07:38] * Launchpad Patch Queue Manager: [r=thumper][ui=none][bug=652626] Upgrade windmill from r1440 to r1544 [07:38] to fix an issue preventing some new launchpad tests from running [07:38] Botwar! [07:38] Haha [07:38] don't encourage them! [07:39] What's with those utf8 failures? [07:39] Mmmm. The SCM messages are not so handy [07:42] do i need to mark a bug fixreleased after qa, or just qa-ok? [07:42] does a bot do the former? [07:42] poolie: Just qa-ok [07:43] great, bug [07:43] Fix Released should only happen once it's on prod. [07:43] bug 615740 is done then [07:43] <_mup_> Bug #615740: test_on_merge.py doesn't handle eintr [07:44] Parts of NMAF look far too much like are an initial hacked-together implementation that was meant to be immediately followed by a proper one. [07:44] Hacks hacks hacks. [07:44] EVERYWHERE. [08:24] Project db-devel build (55): STILL FAILING in 3 hr 53 min: https://hudson.wedontsleep.org/job/db-devel/55/ [08:53] good morning [09:01] Hallo [09:01] Hullo. [09:01] lp is readonly? no warning? ffs. [09:02] It was on identi.ca. [09:02] And the blog. [09:02] But no in-app warning that I saw. [09:02] tweeting 17h in advance is kind of missing the point of the medium [09:02] Heh. [09:03] i'll do it now [09:16] bigjools: Sorry I missed your ping last night; I'd left about three minutes earlier. [09:18] nae bother [09:19] I believe the CPs are done? [09:19] yes [09:19] Phew. [09:19] although my fixit query didn't work, I need to look at it again [09:19] What was the query? [09:20] http://pastebin.ubuntu.com/507347/ [09:20] we cancelled it after a couple of hours :/ [09:20] Ah, that kind of not working. [09:21] What if you just do the usual 2.5-minute SELECT into a temp table, then UPDATE from that? [09:22] Also, dateesuperseded. [09:22] You're not unsetting that. [09:22] ah true [09:22] yeah I'll do the extra temp table [09:23] You might also want to check datepublished IS NOT NULL [09:23] (since it's possible that it superseded a Pending pub) [09:24] Setting that to Published would be confusing, since it would leave a file missing from disk, but it wouldn't really break anything. [09:24] hmmm yes [09:24] Staging was timing out on pillar pages yesterday, same today. Is this a known problem, or is there some nastiness in db-stable? [09:24] I hate our model [09:24] I mean, it's unlikely, but possible. [09:25] bigjools: Sometimes I wonder why status is explicit. [09:25] wgrant: to help query performance I would guess [09:25] I'd rather have more states [09:25] More!? [09:25] yes [09:26] it's a farce at the moment - superseded, but then you also have to look at dateremoved to see if it's really gone, for example [09:26] Was ArchiveRemovalRedesign before your time? [09:27] I think it might have been. [09:27] yes [09:27] Back in the good old days we had Pending, Published, PendingRemoval, Removed. [09:27] Then ArchiveRemovalRedesign introduced the three removal candidate states, and removed the explicit on-disk state. [09:28] So we are now stuck half-way between having an explicit status field, and implying status from dates. [09:28] I'm not sure if that was always the plan. [09:28] Or if it only got taken half-way. [09:28] exactly [09:29] I guess removing Pending would make it pretty much sensible. [09:31] Having Pending but not Removed is stupid. [09:31] But having Published/Superseded/Deleted/Obsolete makes sense. [10:03] hello... === maxb_ is now known as maxb [10:06] wgrant: ok I've fixed staging, let's check it out [10:07] The publication which alerted me to everything is fixed. [10:07] That is encouraging. [10:08] hurray [10:08] That leaves no borked pubs in the primary archive? [10:08] I hope so [10:11] we're down to 20k rows that match the first query [10:12] ie. the first temporary table containing all supersed arch-indep pubs since 2010-08-09? [10:13] yes [10:14] shockingly the 2nd query now returns zero :) [10:14] I hope that's the one with dateremoved IS NULL [10:15] yes [10:24] anyone fixing the test failure in devel? [10:25] or the one in db-devel, for that matter. [10:25] StevenK had a branch for the IDS db-devel failure. [10:25] Not sure about devel. [10:25] wgrant: thanks. [10:26] I've forced a rebuild of both db-lp buildslaves [10:26] I guess I'll fix devel [10:27] jml: I have a branch for devel, I was waiting for codehosting [10:27] db-devel is in buildbot now [10:27] StevenK: oh cool. the test_remote failures? [10:27] jml: Yes [10:27] StevenK: yeah, but buildbot died :\ [10:27] StevenK: sweet. may I see the merge? [10:27] jml: Pushing it now [10:28] StevenK: thanks. [10:28] Oh, so Hudson wasn't on crack? Those were real failures? [10:28] that reminds me. [10:28] time to run the full test suite against stable & db-stable to see if there's anything else to be done [10:31] jml: https://code.edge.launchpad.net/~stevenk/launchpad/fix-ec2test-utf8/+merge/37831 [10:32] looking [10:32] jml: But that looks a lot bigger than what I had locally .. [10:32] yeah. [10:32] StevenK: it also looks suspiciously like mars's branch. [10:32] jml: Yes ... [10:33] StevenK: why is that? I thought that mars's fix had already landed... [10:33] Wrong target... [10:33] So did I [10:33] wgrant: ta [10:34] StevenK: ok. get the right diff & we'll figure out what's making this more interesting than it needs to be. [10:40] Hm. [10:40] We can now kill 2.5 off, can't we? [10:40] Since wildcherry no longer fails. [10:44] !@!!@! [10:44] I just had a branch fail EC2 for the fifth time. [10:44] Each time with a different error. [10:44] Grrr. [10:45] wgrant: I'm sorry. [10:46] wgrant: lifeless is going to make it better. [10:46] Heh. [10:46] wgrant: until then, can I retry your branch for you. [10:46] jml: devel's still broken, isn't it? [10:46] (also, what's the deal with this paramiko error) [10:46] wgrant: I'll run it against stable [10:46] of course [10:47] stable might be broken [10:47] Heh. [10:47] Yes... [10:47] https://code.edge.launchpad.net/~wgrant/launchpad/bug-629921-packages-empty-filter/+merge/37339 [10:48] jml: I'm not sure why that diff is so wrong [10:48] StevenK: Targetted at lp:launchpad. [10:48] And db-devel is missing three devel revs. [10:48] StevenK: it's targeted to db-devel rather than devel. [10:48] Oh, ... [10:50] jml: https://code.edge.launchpad.net/~stevenk/launchpad/fix-ec2test-utf8/+merge/37832 [10:50] (That's what I get for filing MPs while on the phone to bigjools) [10:50] StevenK: uhh ok. [10:51] StevenK: do those tests pass locally for you? [10:51] jml: Yup [10:51] hmm. [10:51] I suspect platform variation. mars changed the 'utf-8' to 'utf8' to fix a test failure [10:51] * jml tries [10:52] anyone getting paramiko errors? http://paste.ubuntu.com/507903/ [10:52] Oh, are you kidding me? Now it wants utf8, rather than utf-8 [10:53] yeah. let me look at the tests. [10:53] wgrant: I'm going to run that fixit query in an hour or so, I think it's ok unless you have any further misgivings? [10:53] I reckon we should relax the constraint somehow. [10:53] bigjools: pastebin pls. [10:53] Just to be safe. [10:54] StevenK: ok. here's what we should do. [10:54] "charset=utf8" is wrong, though. [10:54] Anything that wants that is broken. [10:55] StevenK: split body['Content-Type'] by ';', assert that the first part is 'text/plain' [10:55] StevenK: don't care about the second part. [10:56] wgrant: http://pastebin.ubuntu.com/507877/ [10:56] * wgrant WTFs at the differing theme between paste.ubuntu.com and pastebin.ubuntu.com [10:56] wgrant: for this, we just don't care. if people start having encoding issues with their ec2 mail then we'll worry then. [10:58] bigjools: +1 doit. [11:00] jml: MP updated [11:00] StevenK: thanks. [11:00] StevenK: have you run the tests locally? [11:00] Yes [11:01] bin/test -vv -m devscripts -t test_remote is happy [11:02] StevenK: humour an old man, run bin/test -vv devscripts [11:02] ... Aren't you younger than me? [11:02] jml: Ran 146 tests with 0 failures and 0 errors in 15.604 seconds. [11:03] Almost 10 tests a second? What sorcery is this? [11:03] ... A fast machine? [11:03] You should try it [11:03] Ah, and no DB too, I guess. [11:04] yeah. they are tests for the devscripts stuff [11:04] wgrant: Only unittest layer, so quick [11:04] actually the test_remote tests are quite slow because they make real branches [11:08] jml: Should I land that as a testfix? [11:08] StevenK: yes please [11:08] StevenK: sorry for the delay [11:08] jml: Do you want to stamp it? [11:08] StevenK: sure. [11:09] StevenK: done [11:09] sorry about the delay. multitasking way too much. [11:11] jml: Tis cool. I've tossed it at PQM [11:27] no one knows anything about that paramiko failure [11:28] Hm, that's an odd one. [11:29] yeah. I haven't sat down to serious debugging yet. Want to flush out my mail queue first. [11:54] wgrant: production fixed [11:55] bigjools: Looks good, thanks. [11:55] can you grab counts of remaining broken publications? [11:56] yeah already done a while ago [11:59] Morning, all. [12:06] Project db-devel build (56): STILL FAILING in 3 hr 42 min: https://hudson.wedontsleep.org/job/db-devel/56/ [12:06] Launchpad Patch Queue Manager: [testfix][r=lifeless][ui=none][no-qa] Fix the InitialiseDistroSeries [12:06] test for disabled DASes. [12:08] funk6 [12:08] err funky, even [12:12] jml: Did you end up sending my branch off? [12:12] It may have some hope.. [12:13] wgrant: can we talk about your bug-655614-disabled-arch-indices branch [12:13] bigjools: I opened that 30 seconds ago and was about to ask you the same thing. [12:14] :) [12:14] It's not so much a branch as a set of a few patches that we might want bits of. [12:14] r9787 is confusing me [12:14] It's probably too revolting to land (but no worse than parts of the existing code :/) [12:15] Hm? [12:15] henninge, so, in what sense does +translations on a distroseries fail? [12:16] bigjools: I removed a duplicate loop and added an arch.enabled guard. Previously it was looping across the archs once to publish indices, and again to request Release files. [12:16] wgrant: why the changes other than the .enabled check? [12:16] ah [12:16] It was that or add the check twice. [12:17] got it [12:17] is it all tested? [12:18] 9788 has some quick partial tests, but they're not thorough or pretty. [12:18] not series.getDistroArchSeries(architecture[7:]).enabled): [12:18] eeeurrrrghh [12:18] bigjools: I stole the architecture[7:] from existing code! [12:18] I know :( [12:18] I originally did .split('-')[0], but decided to go with the... preferred method. [12:19] danilos: on [12:19] * bigjools puts fingers in ears and goes blahblahblah [12:19] danilos: 'templates' does not exist [12:20] bigjools: I have formulated a plan to remove all that stupidity, but it will require a bit of a-f disentanglement. [12:20] So it can't be done quickly. [12:20] wgrant: yes [12:20] jml: Great, thanks. [12:20] danilos: that's when rendering DistroSeriesView. [12:21] StevenK: build failure just sent to the list [12:21] wgrant: if you get this branch finished I will run it on dogfood while you get it reviewed [12:21] danilos: that's the test that's failing, actually the last "test_" method im the file. [12:21] danilos: http://pastebin.ubuntu.com/507871/ [12:21] bigjools: I'll write the tests properly now it's no longer 3am. [12:22] jml: That's lp, not lucid_lp [12:22] But I think it's good, so please try on DF, yes. [12:22] StevenK: I'm a little bit behind ... does lp build slave still matter? [12:23] wildcherry was the end of 2.5, I believe. [12:23] So we may be safe. [12:23] danilos: read the backscroll on #lp-reviews for more detail ;-) [12:23] We're all on lucid now, I think [12:24] wgrant: :) [12:24] danilos: bac has ended his day in defeat, so it is not super-urgent atm. [12:25] henninge, fwiw, I can simply create a distribution and distroseries using factory methods and browse to +translations on it [12:25] danilos: that's odd. How do you do that exactly? [12:26] (I know how to use the factory methods) [12:26] bigjools: I'd test PPA and primary archives both with and without active publications in the disabled DASes. Check that the DAS' directories aren't recreated, and confirm that the series' Release doesn't include those arches. [12:26] I will return in 20. [12:27] henninge, in 'make harness' using http://paste.ubuntu.com/507948/ [12:27] wgrant: yup [12:27] henninge, oh, if you know how to use factory methods, perhaps you were asking how do I browse to it? :P [12:27] danilos: exactly ;) [12:28] henninge, well, I open a web browser and go and look at it :) [12:28] danilos: I was not sure which changs to the db are presereved after make harness. [12:28] danilos: ok, simply 'make run' [12:28] henninge, all changes that you've committed are preserved [12:28] henninge, right, but I do that even before 'make harness': it's basically like two appservers talking to the same DB [12:29] I see [12:29] henninge, I've already looked at that test for bac, but now that I think about it, it's probably a missing commit somewhere [12:29] or at least a .sync() [12:31] danilos: we were also wondering why DistroSeriesTranslationsMenu.templates is @enabled_with_permission('launchpad.Edit') when still it is shown to everybody. [12:32] henninge, probably because it's not useful to anybody else all that much [12:32] henninge, i.e. perhaps shouldn't be like that [12:32] danilos: no, I am saying that the link (and the page) is shown to unauthorized users, too. [12:33] danilos: *although* it is @enabled_with_permission('launchpad.Edit') ... [12:33] henninge, oh, then the decorator needs to go [12:33] danilos: you mean it has no function any more? [12:33] danilos: btw, a commit makes the test pass. [12:33] ;-) [12:34] danilos: where would the sync() go? [12:34] henninge, where the commit is, on the object that needs syncing (sometimes a sync() is not enough, but you could try self.distroseries.sync()) [12:35] henninge, that also means that all the other tests (for non-LAUNCHPAD usage values) pass by accident :) [12:35] sync() is useless now I thought? [12:36] indeed [12:36] store.flush() FTW [12:36] good to know ;) [12:36] danilos: A commit does *not* help. btw. I had forgotten that I had commented out the failing parts of the template ... [12:36] :( [12:37] bigjools, ah, right, thanks :) [12:37] henninge, sorry to hear that [12:37] henninge, does removing the permissions decorator on the property help? [12:37] nope [12:38] danilos: if I comment out the context/menu:navigation/templates line in the tmplate [12:38] danilos: it fails on context/menu:navigation/langpacks [12:38] ... [12:38] danilos: My suspicion is it has the wrong menu. But why and how do I prove that? [12:39] henninge, right, we should just get rid of the context/menu:navigation use anyway, but... that doesn't help with this branch [12:39] henninge, hum, "wrong" menu is a potential issue if the entire page is included into distribution:+translations, but I am pretty sure it isn't [12:43] henninge, fwiw, I'd suggest trying setting a breakpoint and going through the test with pdb (or trying it manually in 'make harness') [12:45] bigjools: so you've got the buildd-slavescanner work now? [12:45] bigjools: I can remove it from my todo? [12:45] jml: I'll sort it, yes [12:45] bigjools: ta [12:46] jml: btw have you been playing Uncharted2 then? :-) [12:46] * wgrant returns. [12:47] bigjools: not recently. and actually it was mostly watching two of my flatmates play it [12:48] F1 2010 is keeping me up way too late [13:09] bigjools: Hm, the Release update didn't work? [13:09] Or have you just deleted but not republished yet? [13:10] wgrant: still trying to get the f***ing publisher to do something [13:10] Heh. [13:10] ♥ dogfood [13:11] I've got a pending publication but p-d ignores it [13:11] Pending publications, or pending uploads? [13:12] It's easy to forget p-a... [13:12] already ran p-a [13:12] there's a pending source and binary now [13:12] So even phase A doesn't see them? [13:13] nup [13:13] this is w/o your patch too [13:13] Yeah. I didn't touch phase A. [13:13] Come on DF... show me your queue. [13:14] Uh, is DF's appserver hung? [13:14] There is little response. [13:15] +queue? [13:15] Ah, there we go. [13:15] * bigjools restarts DF [13:15] Yeah, that. [13:15] It's alive now. [13:15] +queue is slow slow slow [13:15] It was going for well over a minute. [13:15] Which should have timed out. [13:15] But it's rendered now. [13:16]