[00:33] lifeless: Successful builds don't send email. [00:33] Except for recipes. [00:33] But that's about to be fixed. [00:34] maxb: I see the bug on +commentedbugs if I change the filter to not exclude closed bugs. [00:37] Whoops! I didn't even notice that there was a filter [00:39] It's the same default one everywhere. [00:40] There's probably a four-digit bug about that. [00:43] wgrant: is https://bugs.launchpad.net/launchpad/+bug/570846 still in progress? [00:43] <_mup_> Bug #570846: Archive freeze should not affect Partner

< https://launchpad.net/bugs/570846 > [01:06] lifeless: No. [01:10] lifeless: You still can't repro the rabbit failure? [01:10] It fails on both trunks here, but not lucid_lp, so it's something local. [01:12] (I also had problems running this in U1, which is possibly relevant) [01:25] it seems to start OK. [01:25] erm [01:28] wgrant: I can't repro [01:29] It was /etc/hosts. [01:29] 127.0.0.1 has to resolve to the hostname, not localhost. [01:29] Which is not the default config. [01:30] It starts fine on localhost, but then something tries to connect to the node @realhostname instead of @localhost, which fails. [01:32] s/socket.gethostbyname()/localhost/ in fq_nodename works fine. [01:32] But hmmm. [01:33] We should talk to U1, I suppose. [01:34] So, to unblock things let's get pilinut fixed? [02:02] wgrant: I wonder if 127.0.1.1 works too [02:04] No. [02:04] That's there by default. [02:06] Network-manager seems to add to the top of /etc/hosts [02:08] It does some evil stuff, but not that (at least for my wired desktop) [02:11] wgrant: you know a bit about package build stuff, right? [02:11] wallyworld_: Indeed. [02:11] What's up? [02:12] wgrant: bug 732343 [02:12] <_mup_> Bug #732343: Recipe source builds failing because of "Could not build because of missing dependencies" do not send notification email < https://launchpad.net/bugs/732343 > [02:12] aaaaaa [02:12] wgrant: there's a unit test for notifications for dependency issues and i can't see why it would be failing [02:12] by it, i mean the prod code [02:12] i'm sure the test would pass [02:13] wallyworld_: You're looking at a recipe build, not binary builds? [02:14] wgrant: ah that may be it. i think i've got myself confused [02:14] So, situation normal, then? [02:14] This stuff has that effect on people. [02:14] :-P [02:15] so i need to find a sourcepackagebuild class analogous to the binarypackagebuild class? [02:15] sourcepackagerecipebuild. [02:16] right. i'm sad that extends a storm class and has all this business logic in it :-( [02:17] That's how LP works :) [02:17] wgrant: u1 or landscape [02:18] wgrant: i think you meant to type :-( [02:19] No, no, a forced smile of the "I cannot let myself realise how terrible this is or I will surely die" variety. [02:20] zomg, armel is about to catch up. [02:20] 272 builds left. [02:20] wgrant: sort of like when your old aunty gives you a cardigan for xmas and you say "oh how lovely" through a forced smile [02:21] I never had that problem -- it was usually my mother giving me clothes [02:21] wallyworld_: Right. [02:27] maxb: bug 5977 [02:27] <_mup_> Bug #5977: Person's Bugs pages do not show closed bug reports < https://launchpad.net/bugs/5977 > [02:29] That's the one. [02:29] lifeless: So, lucid_db_lp needs sorting out. Can you look at that? [02:31] wgrant: rabbitmq likes neither 172.0.1.1 or the builder's IP in /etc/hosts [02:31] Er, 127.0.1.1 [02:32] StevenK: I had 127.0.0.1 localhost . 127.0.0.1 localhost unbreaks it. [02:35] Hmm, that doesn't help. [02:36] What's the error? [02:37] FAILED - check /var/log/rabbitmq/startup_log, _err [02:37] Need more than that. [02:37] Erm, is this the system rabbit? [02:37] Not the fixture? [02:37] that looks like the system rabbit [02:37] Tis [02:37] which is going to be running shite wrapper scripts [02:38] Yes. [02:38] The fixture rabbit does not require the system rabbit to be running? [02:38] No. [02:38] It starts its own. [02:38] Ah [02:39] makes a homedir for it, unique nodename (by allocating a tmpdir in /tmp) etc [02:39] * StevenK branches devel on this builder so he can test [03:20] the dx channel is much quieter during my work day :-| [03:21] thumper: That's what you get! [03:21] :) [03:25] * thumper moves away from coffee shop [03:25] thumper: what are you hacking on atm? [03:34] Is 7 merge proposals for my very first day of looking at the launchpad code too much? ;-) [03:37] Yes. [03:37] lol [03:40] it shows great enthusiasm [03:40] I hope that any stuff the reviews may need will be shallow [03:40] about half were pet peeves [03:40] and i learned at uds how to get into the code.. so that was a start [03:48] can has review https://code.launchpad.net/~lifeless/launchpad/rabbit/+merge/61057 [03:48] lunch,then some analysis [04:25] lifeless: Analysis of why the build is broken, I hope. [04:27] regressions [04:27] whats up with the build? [04:28] lucid_db_lp is permanently broken by rabbit. [04:28] Please fix it. [04:30] hmm, otp not coming up [05:05] wallyworld_: Any luck with the recipe notification thing? [05:05] wgrant: https://code.launchpad.net/~wallyworld/launchpad/recipe-build-missing-dep-email/+merge/61058 [05:06] would windmill have caught bug 781460? [05:06] <_mup_> Bug #781460: bugttask_index.js hide_assignee_team_selection handling broken

< https://launchpad.net/bugs/781460 > [05:06] lifeless: No. [05:06] There was no test. [05:06] But there is now. [05:09] wallyworld_: So, that's not ideal. [05:09] wallyworld_: Since it affects binary package builds too. [05:09] wallyworld_: And they are automatically retried. [05:09] wgrant: ok. i thought we wanted to send the email both times [05:09] but i guess if they retry [05:10] then no need for binary builds [05:10] Right. [05:10] ok. will fix. thanks [05:10] Some have argued (eg. bug #218261) that depwait should notify after a time. [05:10] <_mup_> Bug #218261: uploaders should get emails/notifications for packages in depwait mode

< https://launchpad.net/bugs/218261 > [05:10] But doing it every time would be bad. [05:11] wow, that's an old bug [05:11] Although we don't have common depwait loops any more, so it might be workable soon. [05:11] wgrant: would that have ever worked ? [05:11] wgrant: #781460: I mean [05:11] <_mup_> Bug #781460: bugttask_index.js hide_assignee_team_selection handling broken

< https://launchpad.net/bugs/781460 > [05:11] lifeless: Yes, but it broke during a YUI upgrade. [05:14] lifeless: Bug #728059 [05:14] <_mup_> Bug #728059: Manually editing the URL for a private branch allows user to access +new-recipe form < https://launchpad.net/bugs/728059 > [05:14] You say it's critical because it exposes the existence of private branches. [05:14] But we don't try to hide private branches. [05:15] we do [05:15] why do you say we don't? [05:16] Because +index gives a 403, as most private objects do. [05:16] In fact, that page 403s for normal users. [05:16] Not critical. [05:16] If you want private branches to 404, that's another bug. [05:16] hmm [05:17] I'll talk to flacoste I think === almaisan-away is now known as al-maisan [05:43] wgrant: why was 714527 a regresion ? [05:44] Bug #714527 [05:44] <_mup_> Bug #714527: owned widget structured strings render as bits of quoted html < https://launchpad.net/bugs/714527 > [05:44] lifeless: sinzui fixed some bug that I don't recall, which ended up breaking that. [05:44] So, it was a regression. [05:46] I think it was when he fixed the XSS in the item widgets that I reported in December. [05:49] spiv: are you still working on 717094 ? [05:49] Bug #717094 [05:49] <_mup_> Bug #717094: InvalidURL OOPS in translatePath because of URL containing non-ascii chars, again < https://launchpad.net/bugs/717094 > [05:55] thumper: what made 721134 a regression ? [05:59] * wallyworld_ does the school pickup run [06:27] wgrant: bug 732442 looks like fallout not a regression [06:27] <_mup_> Bug #732442: disable_existing_builds compares series name to display name < https://launchpad.net/bugs/732442 > [06:27] wgrant: I mean - was something actually broken, or was it just never correct? [06:27] lifeless: It was actually broken. [06:28] Note that a test failed. [06:28] and windmill was disabled at the time ? [06:29] yes. [06:52] lifeless: Bug #728789 was a regression. [06:52] <_mup_> Bug #728789: "Request build" dialog indicates builds are pending when they are not and doesn't allow new builds < https://launchpad.net/bugs/728789 > [06:52] People were unable to request builds that they could previously request. [06:52] Because the new check was buggy. [06:54] was there no previous check ? [06:54] Correct. [06:56] mmm [06:56] so during dev of a beta feature [06:56] I'm torn [06:56] It was a Google beta. [06:57] kindof [06:57] lp-beta-users only [06:57] It was in beta for 12 months. [06:57] It was not a major user-affecting regression. [06:57] But does that mean it wasn't a regression? [06:59] thats not what I meant [06:59] there was an arc of development [06:59] during that arc things were tightened up to enforce new policies [06:59] and some of those changes had bugs [06:59] Yes. [06:59] And they broke some functionality. [06:59] => regression [06:59] they broke two things [07:00] one that unlimited builds in a day could be done manually [07:00] and second that the logic for detecting that was flawed [07:01] which is why I'm saying that I'm torn [07:02] Erm, what's this about unlimited builds in a day? [07:02] This is unrelated to the quota. [07:03] maybe I'm misunderstanding the exact circumstances [07:04] The check was to prevent people from requesting builds that were already queued. [07:05] Unrelated to the silly 5-a-day quota. [07:05] the request builds popup dialog was new and when deployed, there was a bug found, so not a regression [07:05] it was a bug in a new feature [07:05] wallyworld_: Hm? Users were unable to request builds manually where they could before. [07:05] The AJAX form replaced the non-AJAX one, which worked fine. [07:06] the request build popup was totally new - before that it was html without ajax [07:06] Sure. [07:06] It was a new UI,. [07:06] Replacing an existing UI. [07:06] And breaking part of that functionality. [07:07] If I replace apt-ftparchive with NMAF, we lose a lot of functionaliaty. It's totally new, but still a regression. [07:07] the old one was still available. but my view is that a bug in a new feature is not a regression [07:07] There was an unobvious workaround, sure. [07:07] (one which is itself a bug) [07:07] i guess it depends on the definition of regression. imo, regression means a break in existing code [07:08] Users don't care about code. [07:08] They care about functionality. [07:08] not all workj arounds are obvious, but htere was one [07:08] Existing *functionality& [07:08] s/&/*/ [07:08] Before the change: users could request a manual build by clicking the "Request builds" link. [07:09] After the change: users were erroneously prevented from requesting a manual build by clicking the "Request builds" link. [07:09] agreed. but it was a bug in totally new code. so i can see both sides :-) [07:10] Users do not care about code! [07:10] If I delete all of our code, it is not a bug in the code! [07:10] But it is still a regression.' [07:10] replace "new code" with "new functionality" [07:10] it was new functionality [07:10] in a sense [07:10] It was a replacement interface for existing functionaliaty. [07:11] for a feature that was still under development in beta [07:11] Very late beta. [07:11] but technically not officially released [07:11] we should be down the pub having this discussion over a beer :-) [07:12] i'm just stirring the pot a bit :-) [07:12] Heh. [07:12] wallyworld_: Just because the feature isn't released doesn't mean it isn't a regression. [07:12] It is less important. [07:12] But not not a regression. [07:12] i'll go away now. in all honestly, i don't really care one way or the other. just going a bit stir crazy [07:12] * StevenK hits wgrant with the double-negative stick === al-maisan is now known as almaisan-away [07:17] wgrant: why was 766874 a regression ? [07:18] Bug #766874 [07:18] <_mup_> Bug #766874: Inline bug assignment for non-contributors broken in Firefox 3

< https://launchpad.net/bugs/766874 > [07:19] lifeless: wallyworld had fixed another regression (I forget the number... the one about inline assignment to a non-contributor not giving a warning), and it was borked in Firefox 3. [07:19] Windmill caught it. [07:19] But was disabled at the time. [07:20] what about bug 768336 [07:20] <_mup_> Bug #768336: Subscribing to a bug doesn't show display name on FF4

< https://launchpad.net/bugs/768336 > [07:20] I wasn't involved in that one. I don't know if it was related to the new subscription work. [07:20] I really wish I could click on links after viewing the merge diff [07:21] Hmmm, I saw that once locally. [07:21] Last week. [07:21] Thought it was a browser glitch. [07:21] But I guess not. [07:28] wgrant: what makes bug 780429 a regression (other than being a daft change) [07:28] <_mup_> Bug #780429: CopyChecker.checkCopy uses check_permission badly < https://launchpad.net/bugs/780429 > [07:28] wgrant: was it functionally wrong, or just high risk / would break if we changed other assumptions? [07:29] lifeless: The fact that it called check_permission was high risk. [07:29] The fact that it always used the primary archive was a security vulnerability. [07:29] ok [07:34] https://dev.launchpad.net/Regressions2011#preview [07:35] You win lots of points for doing that, but lose several for unreadability. [07:35] That page is barely readable [07:36] yes [07:36] its a draft [07:36] And 780429 should be badtest [07:36] notest sideffect 777766 the list of subscribers for gtk bugs is empty [07:36] badtest, surely. [07:37] StevenK: there was a test for permissions on different archives? [07:37] There certainly was a few testcases for rvb's work, I can't recall what they did [07:37] A relevant test was added at the time of the change. [07:38] But it had the same misunderstanding. [07:39] so [07:39] uhhm [07:39] if you know better, please fix [07:39] I intend badtest to be when the test intent would cover the thing that was faulty [07:40] Ah. [07:40] so a test whose intent was to check something funky [07:40] and it successfully checked that [07:40] is not a badtest [07:40] the real problem had no test aiming to cover it [07:40] and the tests have to exist before the change [07:41] otherwise its notest [07:41] I'll add some prose to that effect after dinner [07:41] notest sideeffect 754089 launchpad only remembers alternate landing targets in merge proposal creation page for one day [07:41] [r=jcsackett, [07:41] sinzui][ui=none][bug=541713] Skip targets from old proposals. [07:41] Looks more deliberate than a sideeffect. [07:42] it was meant to clean up after a while [07:42] the knob is wrong [07:42] Hm. [07:42] That rev set it to 90 days. [07:43] rofl [07:43] could be badtest [07:43] No, it *says* it set it to 90 days. [07:43] But it set it to 1. [07:44] The relevant line (with days=1) was last changed in r12165.1.2, with a commit message of "The widget for selecting merge targets ignores proposals older than 90 days" [07:44] So, not a side-effect, nor really a badtest, because the relevant test was only added then. [07:44] Perhaps reviewfail. === nigel-cloud is now known as nigelb [07:50] well [07:50] I think I'd argue badtest [07:50] the test that the thing is preserved doesn't really keep the user story intact [07:50] anyhow, please fix it up [07:50] I put it on the wiki deliberately :) [07:51] Yeah. [07:51] Well. [07:51] I'm fixing the bug instead :) [07:52] as for the sideeffect [07:52] there was a bug that they accumulated forever [07:52] with no way to edit/fix [07:52] this regression was added while fixing that bug [07:53] That is cutting it very fine. [07:53] But OK. [07:59] lifeless: Bug #778847 broke notifications for everybody for any bug that had an affected subscriber. [07:59] <_mup_> Bug #778847: Muting a bug subscription for a team with a contact address crashes getRecipientFilterData

< https://launchpad.net/bugs/778847 > [08:00] It was probably the most critical regression we have had in months. [08:00] It was only by very good luck that it didn't became a serious incident. [08:00] If any regression this quarter needs analysis, it is that one. [08:07] Ahh, that MP target test was flawed in a couple of other ways, too. [08:08] killall launchbag [08:33] good morning [08:36] hi adeuring [08:36] hi jtv! [08:43] wgrant: so, it could only affect muted teams, which could only happen to beta users [08:43] lifeless: No. [08:43] wgrant: no? [08:43] lifeless: It affects all notifications to anyone from bugs which might have sent notifications to a muted team. [08:44] The mute breaks recipient calculation for the whole bug. [09:06] Yippie, build fixed! [09:06] Project devel build #719: FIXED in 5 hr 30 min: https://lpci.wedontsleep.org/job/devel/719/ [09:06] !!!!! [09:06] Well done StevenK. [09:07] * StevenK takes a bow. [09:08] StevenK: cool [09:08] StevenK: thanks! [09:09] is bug 740584 a regression as well? [09:09] <_mup_> Bug #740584: Build lacks a corresponding source publication < https://launchpad.net/bugs/740584 > [09:09] ah yes [09:09] win [09:10] Yes :( [09:10] That bug makes me sad [09:10] wgrant: could you add to https://dev.launchpad.net/Regressions2011#preview ? [09:10] The rationale given for the assertion was false. [09:10] Sure. [09:11] Done. [09:12] * wgrant stabs bug heat. [09:12] DIE === almaisan-away is now known as al-maisan [09:17] wgrant: what triggered that outnyurst [09:17] Clearly, bug heat did *something*. [09:18] lifeless: Timing out on DF. [09:18] In particular, selecting the maximum heat for a context. [09:18] But it is horrible on qas as well. [09:18] When updating. [09:18] Hallo [09:28] spiv: are you still working on bug 717094 ? [09:28] <_mup_> Bug #717094: InvalidURL OOPS in translatePath because of URL containing non-ascii chars, again < https://launchpad.net/bugs/717094 > [09:34] lifeless: I don't think I ever started working on that, so no :/ [09:34] spiv: it is assigned to you [09:34] spiv: by poolie [09:34] Yeah, I see the bug history [09:34] :) [09:43] wgrant: do you remember if there are bugs filed about the soyuz uploader oopses that are not oopses? [09:49] hi all === henninge changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: henninge | https://code.launchpad.net/launchpad-project/+activereviews [09:56] lifeless: how much longer are you around? [09:56] I'm not at all, but I can be if needed, if that makes sense. [09:56] not urgent, I just wanted to fill your head with something I learned at UDS [09:57] some nice architecture that the Landscape guys use, which I also want to use [09:57] the notification server? [09:57] yarp [09:57] yeah [09:57] I'm aware of that [09:57] ahy coolio [09:57] we have a few things we'd need heavy lifting on to start using it [09:57] - need to arrange open sourcing of it [09:58] totally [09:58]