[04:07] * mwhudson off home [08:32] good morning [08:46] moin adeuring [08:46] hi al-maisan! [09:01] Morning [09:01] Good morning mrevell [09:03] Dobro jutro al-maisan [09:03] oh wow! [09:03] al-maisan, Heh, "Good morning in Bosnian" on Google. [09:04] * al-maisan did not know mrevell as a polyglot :) [09:05] hi all [09:05] I have this failing import: http://launchpadlibrarian.net/38896393/prosody-im-trunk-log.txt [09:06] Hm, that's pretty awesome. [09:09] asabil: jelmer should be able to take a look when he's around, but either way, please file a bug about it so it gets tracked :) [09:11] ok thanks [09:13] bigjools: lamont asked me this morning about publishing copy archives. I said it wasn't currently possible, and redirected him to you. Apparently it's going to be needed soon. [09:14] they're going to be out of luck then :/ [09:15] It's not that difficult... [09:15] gmb: Hi, when you are around, could you initiate the landing of https://code.edge.launchpad.net/~maxb/launchpad/stop-using-deprecated-sets/+merge/18896 (approved; has commit message)? Thanks. [09:15] difficulty is not the issue [09:25] maxb: Will do. [09:26] thanks [10:13] hi, how do I raise an error such that this error is exposed as a 400 HTTP error in the webservice api? === matsubara-afk is now known as matsubara [10:15] thekorn: grep around for 'webservice_error' [10:16] wgrant, ok thanks [10:19] hmm, somehow still now working, it is always a 500 [10:20] ah, nm, I was missing: """This is only effective when the exception is raised from within a published method[...]""" [10:21] thekorn: What were you raising it in? [10:21] A property? [10:22] wgrant, no, in a seperate method which compiles some BugTaskSearchParams for Person.searchTasks() === jamesh_ is now known as jamesh [11:38] The "Request a review" interface is... confusing. [11:46] yes [11:46] the type shouldn't be on the bottom [11:46] blame rockstar and lazr-js [11:47] It would be entirely unobvious what to do, even if the type was at the top. [11:47] why is that? [11:47] Forms normally have buttons. [11:47] right [11:47] blame rockstar and lazr-js :) [11:48] I complained! but there where some limitations, etc [11:48] :( [11:48] wgrant, I'm sure there's a bug for that, and I'm sure we can push it up the priority list [11:49] beuno: It took me a while to work it out, and I've used it before, so it's probably reasonably awkward for newcomers. [11:50] I'm sure it is [12:04] jml: I'm getting hangs in tests and I think it's related to Twisted. Specifically, when TestTwistedJobRunner and doc/externalbugtracker.txt are run, one of them hangs (depending on if I change the layer to TwistedLaunchpadZopelessLayer or not). They both start and stop the reactor. Do you have time to discuss this? If not, I'll talk to abentley later about it. [12:07] allenap, sure. gimme a sec first. [12:11] allenap, what's the branch? [12:12] allenap, and will "bin/test -t doc/externalbugtracker.txt" reproduce the behaviour? [12:12] jml: lp:~allenap/launchpad/twisted-threading-bug-491870 with an additional patch http://paste.ubuntu.com/373163/ [12:12] jml: bin/test -vvct 'externalbugtracker.txt|test_runner' [12:12] allenap, ok. looking into it. [12:13] jml: It'll hang with or without the patch, just in different ways. [12:13] allenap, while I'm fetching branches, I need to get something out of my system that's not particularly helpful but tangentially related to this discussion [12:13] "don't use threads" [12:13] jml: I know :) [12:14] jml: This is Step 1 on the Road To Enlightenment: Getting Twisted's Foot In The Door. [12:14] jml: I want to be able to convert each remote bug tracker one at a time to run async. [12:21] test_timeout hangs? [12:22] jml: Yes. If you revert the patch, externalbugtracker.txt hangs. === mrevell is now known as mrevell-lunch [12:23] allenap, I haven't applied the patch yet [12:23] jml: Ah :) Maybe it's the other way round... [12:24] allenap, looks like the "hanging" is the reactor running with no events to listen to. [12:24] jml: Ah, that's interesting. [12:25] allenap, I found that out using strace and gdb, fwiw. [12:26] jml: I saw that it was hung in a select(...) with strace, but didn't really know what that meant. It feels a little bit obvious now! I don't know how to use gdb :-/ [12:27] allenap, https://dev.launchpad.net/Debugging/GDB [12:27] jml: Thanks. [12:28] allenap, do you ever run the reactor in process manually? [12:29] jml: Do you mean at the REPL? Yes, sometimes. I haven't been doing so for this. [12:29] allenap, hmm. looks like it. afaict, updateBugTrackers calls reactor.run() [12:30] allenap, I mean in the tests. [12:30] allenap, there's this shitty thing about Twisted where you can't restart the reactor reliably [12:31] allenap, Trial works around this mostly by out-evilling everything around it. [12:31] jml: Heh :) Trial FTW. I wondered if there was an issue there. [12:32] jml: Is there an easy way to run these tests in a subprocess? ISTR that the Zope test runner does something along those lines when it can't tear down a layer? [12:32] hmm. [12:32] allenap, that, I don't know. [12:32] you could jigger something up with subunit, I guess. [12:33] jml: Yes, that would work. [12:33] subunit.IsolatedTestCase() around the doctest. [12:34] jml: Do you know how it works in TestTwistedJobRunner? Is it simply that that's the first kid on the block to start and stop the reactor so it's been working fine so far? [12:34] jml: Ooh, that's cool :) [12:34] hmm. [12:35] allenap, I think so. [12:35] IMHO, TestTwistedJobRunner should be tweaked to not call reactor.run either. [12:36] jml: Use a reactor test double instead? [12:36] allenap, maaybe. [12:36] allenap, I'm not sure. [12:36] allenap, I don't really have the headspace atm to figure out The Right Way of testing This Sort of Thing [12:37] jml: Okay, it's good to know that I shouldn't pursue the path of trying to get the reactor to restart. I'll look into IsolatedTestCase I think. [12:38] jml: Thanks. [12:40] allenap, np. [13:07] allenap, hey, I think a addressed all your comments on https://code.edge.launchpad.net/~thekorn/launchpad/make_iperson_ihasbugs/+merge/18541 however there is one thing missing, I'm still relying on sample data - I forgot to mention this in my comment to the merge request [13:08] but I can try to find out how to produce some data for testing purposes later === mrevell-lunch is now known as mrevell === jamalta-afk is now known as jamalta [14:07] Nothing landed today? is PQM busy doing private things? [14:09] maxb, PQM is pretty empty [14:09] hmm. Maybe it'll do my branch soon then :-) [14:10] maxb, it's not queued up [14:11] who sent it to PQM? [14:11] ah [14:11] it is queued up [14:11] 12 minutes ago [14:11] ah, great [14:11] it'll take a little while [14:11] it's the only branch on the queue [14:25] should I worry about seeing this on pqm.l.n ? http://paste.ubuntu.com/373241/ [14:29] jml: should I worry about seeing this on pqm.l.n ? http://paste.ubuntu.com/373241/ [14:30] kfogel: it means that the patch before yours failed because it had db changes [14:30] BjornT: Oh! Thank you. I would never have guessed that from the output (it looks like both the failure and my change are in the "Now Playing" category, and thus I thought the failure was due to my change.) [14:30] s/.)/)./ [14:31] thekorn: Awesome. I'll look at your branch this afternoon. [14:31] allenap, super thanks [14:32] kfogel: well, they are. the problem is that it's a tail from the pqm log, which doesn't clearly indicate where a new merge request starts [14:32] BjornT: ah, so the "cleaning working directory" part (say) is about my change, but stuff farther above is just from earlier in the log? [14:34] kfogel: i think so, but i'm not sure === matsubara is now known as matsubara-lunch [15:44] wgrant, beuno, I have a fix for the "Request a Review" picker, but it's ugly, and I'm tired of writing ugly javascript. I just need to get some time to clean it up. [15:57] rockstar, cool [15:57] while you're here [15:57] could you or abentley take a look at: https://code.edge.launchpad.net/~facundo/ubuntuone-client/lr-clean-trash [15:58] beuno, crapola. :/ [15:58] rockstar, any idea how that got in that state? [15:58] Oh wait, this isn't an artifact from the other day. [15:59] beuno, no, no idea how it got like that, but I'd be curious to find out. [15:59] rockstar, AFAIK, just pushing to it === matsubara-lunch is now known as matsubara [16:00] beuno, can you break the lock? I'm under the impression that the mirrored copy is the locked copy. [16:01] gary, you know how the version marker interfaces subclass each other for convenience? [16:01] it turns out there's a place where that's very _in_convenient [16:02] heh, what is it? [16:02] is there a method like providedBy which only returns true if the object provides the interface itself as opposed to providing a subclass? [16:03] i thought there was something like directlyProvidedBy but apparently not [16:03] Yeah I don't think so [16:06] gary: the problem is in toWADL(). i go through the registered adapters looking for entry adapters. [16:06] IBug_betaAdapter, IBug_10Adapter, and so on [16:06] rockstar, I'll get him to try [16:07] it's odd that locks would get mirrored though [16:07] the problem is that if you ask for WADL for 'beta', it matches both IBug_betaAdapter and IBug_10Adapter, because IBug_betaAdapter is registered under the 'beta' marker interface, and IBug_10Adapter is registered under a subclass of that interface [16:07] so you get an error saying that there are two classes called 'bug' [16:07] beuno, no, I'm assuming it got locked so that the mirrorer could mirror the branch. [16:09] aha [16:09] rockstar, I'll ge thim to try that and let you know [16:14] leonardr: thinking. do you already have a work-around in mind? [16:14] gary: no, i'm stumped. going through zope.component looking for a function that might help [16:14] ok [16:16] leonardr: Can you give me a code snippet to look at of the pertinent lookup? [16:16] gary, sure [16:19] gary: http://pastebin.ubuntu.com/373322/ [16:19] looks [16:22] so leonardr you need ``if not version.providedBy(self.request):`` to be more restrictive, right? [16:22] gary: right [16:29] gary: zope.interface implements providedBy by getting some kind of specification object and returning "self in spec._implied" [16:29] if i could get that spec there might be a spec._directly [16:31] leonardr: you can call zope.interface.providedBy(obj) which will return a spec, which has an __iro__ that would have the right ordering. Pretty roundabout but maybe that's all we can do. [16:31] i'll try it [16:40] gary: changing to "if not version in providedBy(self.request):" worked [16:41] leonardr: great! === jamalta_ is now known as jamalta [18:03] Whereabouts in the PQM queue is lp:~maxb/launchpad/stop-using-deprecated-sets ? [18:04] maxb, PQM is empty [18:05] argh [18:05] gmb landed my branch on db-devel :-( === deryck is now known as deryck[lunch] [18:06] maxb, I could try and land it against devel, but it will take me a while [18:06] maybe someone else here? [18:06] * beuno looks around [18:11] EdwinGrubbs: With your OCR hat on, could you do a PQM submit for me? I have a branch which has already been through ec2test successfully (against devel), but somehow then went on to land on db-devel. I'd like it to re-land on devel, if that's ok. lp:~maxb/launchpad/stop-using-deprecated-sets is the branch. I can forward you the test results email if you'd like some corroboration concerning skipping ec2test [18:11] g'night all [18:15] maxb, if nobody volunteers in the next 20 minutes, ping me and I'll do it [18:16] It is not that urgent. I'd like it done if someone's free, else I'll hassle gmb again tomorrow morning [18:16] maxb, I'll do it now [18:17] I don't want you to feel slowed down, we appreciate all your work a lot :) [18:17] I'm in the London office anyway, so it should be speedier [18:18] maxb: I can do it. [18:19] * beuno control+c's and moves on [18:21] maxb: out of curiosity, what is the problem with just waiting for the rollout to get it into devel, since your branch should already be on staging for testing? [18:23] I have another branch that conflicts with it that I also want to land, and following on from that my aim is to start trying to run the testsuite under Python 2.6. I guess I could just move all my efforts to being based on db-devel for this cycle, but that seems a bit awkward [18:23] Especially with it being week 1 [18:29] maxb: makes sense, I'll land it now. [18:29] thansk [18:36] thumper, ping [18:36] rockstar, ping [18:36] https://code.edge.launchpad.net/~facundo/ubuntuone-client/lr-clean-trash [18:36] is still trashed [18:57] maxb: your branch can't be landed on devel because changes files in the database/schema directory. [19:00] oh [19:01] ugh [19:05] EdwinGrubbs: Thanks for pointing this out, I failed to consider that, since I was merely tidying scripts, not actually thinking about db schema changes === deryck[lunch] is now known as deryck [19:08] good morning [19:13] mwhudson, hey hey [19:13] whenever you have enough coffee in you [19:13] could you take a peak at: https://code.edge.launchpad.net/~facundo/ubuntuone-client/lr-clean-trash [19:14] beuno: hmm [19:19] beuno: i got the puller to try again and it seems happier now [19:19] mwhudson, that's very nice of you, thank you [19:20] any idea how it got itself into that situation? [19:36] +filebug is timing out on malone a lot today. I wonder if we broke something. === matsubara is now known as matsubara-afk [20:11] gmb, ping [20:11] deryck: Hi [20:11] * gmb sees previous message [20:11] gmb, so +filebug seems to be eternally timing out today... [20:11] Oy [20:11] deryck: OOPS me. [20:12] gmb, I can only get the simplest of dupe searches to run. [20:12] gmb, that was my question... how do I get an OOPS number now? :-) [20:12] Hah. Good question. [20:12] There is a way, I think, hang on... [20:15] deryck, gmb: either of you know wiki syntax for an xml charref? e.g., Łukasz Czyżykowski :-) [20:15] Any way to convert that to something that will display right? [20:16] kfogel, you know, I don't know. I proudly don't speak moin fluently. [20:16] kfogel, I assume the syntax guide was no help? [20:17] deryck: not so far; googling [20:17] deryck: Ah, rats. No there isn't, because it's all compressed code so it's impossible to put a breakpoint in. I think it's a job for hte OOPS report. [20:18] gmb, and we don't have a way to get a list of OOPS other than the OOPS report? [20:18] deryck: Well, rooting around on devpad I found OOPS-1502EB617 for a start. [20:19] deryck: But no. And that sucks. [20:19] gmb, at least one OOPS is good though, thanks! Some place to start debugging. [20:20] deryck: Ouch. 17000ms on the query to find possible dupes. [20:20] deryck: It's an FTI search though, so it's always a bit sluggish. [20:21] gmb, yeah. But something has to be wrong'er than normal. :-) [20:21] deryck: Possibly. [20:22] deryck: if you grep -r for '+filebug-show-similar' in devpad:/srv/launchpad.net-logs/edge you'll find the OOPSes. you'll need to discard the launchpad access log matches. [20:22] Obviously you can also limit by date; that helps :) [20:22] gmb, cool, thanks. [20:22] gmb, also, I'm trying queries on staging DB for a point of comparison. [20:23] deryck: Good idea. [20:24] * gmb switches machines; brb [20:25] gmb, the 17000 ms query was 11064 on staging, which smells wrong. queries always take longer on staging, no? [20:26] and repeats are wayyyyyyy faster. [20:28] * gmb returns [20:28] deryck: That could be a caching / buffering / something-else-about-postgres-I-don't-understand thing, though. [20:29] Maybe its a db load thing...? [20:29] * gmb clutches at straws [20:29] gmb, yeah, I'm sure it is. But why not see the same on re-checks on lpnet? and yeah, load is what I'm wondering too. [20:29] Hmm [20:30] deryck: Well, the request should go to a slave, right, because it's a SELECT. So maybe the re-checks go to a different slave. [20:30] But this is all guesswork. [20:30] staging is usually much less loaded than prod... [20:30] Right [20:31] right, but +filebug for me on staging never works, but does in prod... so if staging queries run in half the time as lpnet today... anyway, this is just guess work at this point. [20:33] ah ok, that does sound a bit odd [20:33] and hi mwhudson, btw :-) === salgado is now known as salgado-afk [21:35] jelmer: are you going to land a change to use lp:bzr-git tip for lp or should i? [21:36] mwhudson: is there sense in landing something now, or could we wait until closer to the rollout? [21:37] jelmer: testing on staging, maybe? [21:37] not sure that's needed here [21:37] "being sure we won't forget" is the main one i guess [22:48] jml: I'm sure you're not there, but if you are: https://code.edge.launchpad.net/~kfogel/launchpad/cc-script-new-world/+merge/19060 [22:49] That contains, among other things, the community-contributions.py improvement we were discussing yesterday. [22:52] kfogel, cool. I'll do the code review tomorrow. Hip-deep in my own thing right now. [22:52] (we gonna learn you some Python idioms!) [22:52] jml: I see you must have looked at it already :-). [22:52] jml: looking forward to the review, thanks. [23:00] kfogel: Is that really the latest code? It calls people like Daniel Silverstone non-LP devs (which is wrong), but /Contributions/Draft correctly omits their commits from the list. [23:02] wgrant: don't know how Daniel Silverstone ended up on the list. I'll remove him. It is the latest code, and so I wonder why Silverstone is omitted when he should be (wrongly) included right now. [23:03] wgrant: did he formerly work for canonical on a non-lp team? [23:03] wgrant: also noticing david allouche, same situation [23:04] kfogel: Daniel Silverstone did used to do LP work [23:04] kfogel: He was pretty much the primary Soyuz developer at the start, so he was very much LP. [23:04] ddaa was Code. [23:04] thumper: *nod*, fixing, thanks [23:04] wgrant: ddaa is david allouche? [23:04] kfogel: yes [23:04] And carlos Translations. [23:04] perello? [23:04] kfogel: it is wgrant being a young person and abbreviating everything he can :) [23:05] Malcolm Cleaton was Soyuz. [23:05] thumper: Pfft. [23:05] thumper: stp mkng fn of ppl [23:05] wgrant: thank you! I'm fixing these (moving them to the known_canonical_lp_devs list) as you say them [23:05] any others? [23:05] kfogel: you have a surperflous o in there [23:06] thumper: it's not superfluououous [23:06] I wish I got red underlines under incorrectly spelt words [23:06] Some of the others were probably on LP ages ago, but I couldn't be sure. [23:06] wgrant: I'm sure I'll hear about them eventually. [23:07] They seem to be magically not showing up anyway. [23:07] mwhudson: did you want to have a call this afternoon about the build from recipe stuff? [23:08] wgrant: that worries me; I wonder why not [23:08] wgrant: maybe because their bzr identities were set up in some other way? [23:08] mwhudson: I was going to take the car down to get looked at, but we could do after [23:08] In fact, despite the list being quite wrong, the produced page contains not one person who even might have been on the LP team. [23:08] thumper: yeah, sounds sane [23:09] wgrant: that makes me worry if it's leaving out people who should be there [23:09] thumper: i'm not going anywhere, when is good for you? [23:09] mwhudson: later :) [23:09] wgrant: feel free to do a code review on https://code.edge.launchpad.net/~kfogel/launchpad/cc-script-new-world/+merge/19060 -- I could well have mussed something up. [23:09] thumper: :-) [23:10] wgrant: in fact, if you want, I'll put in a review request [23:10] (so it's tracked) [23:10] wgrant: (draft wiki page regenerated, btw) [23:11] Oh, also, Code Somerville is in both lists (the first one twice), when he should just be in the non-LP one. [23:12] I didn't think John Lenton was ever LP, and Michael Casadevall never was. And neither was Sidnei, AFAIK. === jamalta is now known as jamalta-afk [23:15] wgrant: fixed (about to commit and push up) [23:15] wgrant: any others you spot? [23:16] wgrant: oh that's right, sidnei works elsewhere -- I just talked to him a bunch before, so thought we were on the same team. I guess I'm pretty informal about these boundaries :-). [23:18] Elliot Murphy is in both. I'm not sure about him. David Murphy I'm not sure about. Jonathan Knowles was an LP dev. [23:19] Oh, and Michael Vogt should be added to the non-LP list. [23:19] wgrant: thanks. I guess this is proof that we really should be using Launchpad to do these identifications! [23:19] David is non-LP [23:19] david murphy was an lp dev for a while [23:20] * james_w backpedals [23:20] james_w, mwhudson: see, yeah, that's the thing -- the script has no way of dealing with people who changed roles. [23:20] * maxb wonders what the cc script does re db-devel landings [23:20] maxb: punts [23:20] maxb: see note at top of wiki page [23:20] maxb: when they make it to devel, then they show up [23:20] kfogel: it's pretty hard to figure that out i can see [23:20] ah right, I'll get a bonus point in three weeks then :-) [23:21] It should actually be reasonably easy to make it consider both, I think. [23:21] Maybe I'll have a look at that later. [23:21] wgrant: you'd be my hero [23:21] wgrant: its current state is a classic example of tech debt by design, IMHO [23:22] maxb: right :-) [23:24] * maxb crosses fingers and commences a test run under python 2.6 [23:25] maxb: Is all your 2.6 stuff now merged? [23:25] (apart from the actual version switch, of course) [23:36] wgrant: no, I have two branches totally pending plus 1 landed on db-devel not devel [23:38] One of which can't land because of bugs in the tests of initialize-from-parent.py [23:40] maxb: What's the bug? [23:41] It intends to check that the parent has no pending builds. That check was broken. I fixed it. However, the use of it in the testsuite actually does invoke it on a parent with pending builds. [23:41] Ew. [23:44] * maxb wonders why ubuntu/hoary in the sampledata has pending builds [23:45] The Soyuz sample data is awfully broken. [23:45] It's odd that a test uses ubuntu rather than ubuntutest, though. [23:51] oh no, it uses both :-) [23:51] it creates a new series in ubuntutest, parented on one in ubuntu [23:54] man, soyuz is so hobbled [23:58] I believe ubuntutest exists because ubuntu got too broken in the sample data. [23:58] but too many tests depend on it [23:58] araragh [23:58] * mwhudson lunches [23:59] Right.