[00:00] ah right, glad we found out how it was happening [00:01] Let me toss that ec2 and then forage for some breakfast. [01:04] Hey guys. [01:05] I've been in the process of registering a new account... [01:06] But I am not allowed to.,. [01:06] Wait, actually, nevermind. [01:07] How do I merge accounts? [01:07] RyuGuns, https://help.launchpad.net/YourAccount/Merging [01:07] :) [01:08] :) [01:24] Funk. [01:24] How do I change my ID? [01:24] Nevermind. [01:54] RyuGuns: bobweaver: you might prefer #launchpad for customer support [01:56] https://launchpad.net/ubuntu+mobile+phone [01:56] Anyone interested? [05:16] wgrant: Free to review, or still distracted by *redacted*? [05:25] And no wallyworld. [05:25] Maybe bigjools' tech has broken TPG. [05:25] StevenK: 'sup? [05:25] wgrant: https://code.launchpad.net/~stevenk/launchpad/hack-itemwidget/+merge/104202 [05:26] oh god [05:26] StevenK: Why are you setting term.name now? [05:27] wgrant: Because the test in test_itemswidgets requires it [05:27] And I'd rather just use self.assertRenderItem [05:30] StevenK: r=me before lp melts again [05:30] Haha [06:59] wgrant: StevenK: have you seen bug emails with the same info twice? eg "** Visibility changed to: Private" is listed twice in the one email [07:00] i seem to recall i may have seen this but am not sure [07:00] I can't recall that one, personally. [07:01] I a note to track down why information type: Public -> Private turns up twice when you change the information type via JS. [07:02] StevenK: that's what i'm seeing also but i'm fairly sure it happened before the info type field was introduced [07:09] I don't recall seeing that before. [07:09] I would suspect a regression. [07:29] wgrant: i seem to recall it may have happened for eg "** Visibility changed to: Private" which was before disclosure [07:29] but i can't be 100% sure [07:30] That stuff was dangerously refactored a few months ago [07:30] It may have been a regression then. === wgrant changed the topic of #launchpad-dev to: http://dev.launchpad.net/ | On call reviewer: - | Firefighting: - | Critical bugs: 3.47*10^2 [08:14] aloha [08:23] czajkowski: hi [08:24] morning [08:40] ok. [08:40] so I'm going to do some Launchpad hacking. [08:40] stand back. [08:40] Uhoh [08:40] does it work in precise yet? [08:40] We're doomed [08:40] Um [08:40] Well [08:40] Maybe. [08:41] You'll need python2.6 [08:41] I have Python 2.6! [08:41] It might Just Work™, then. [08:41] I fixed the other issue yesterday. [08:41] But haven't retried since. [08:43] wgrant: was sure that was one of the resonas why gmb was setting up an Ec2 for uds lp clinic was it not working on precise just yet [08:43] KeyError: 'STORM_CEXTENSIONS' [08:44] Doing what? [08:44] How old is the tree? [08:44] I just pulled [08:44] but I've forgotten everything I need to do to update [08:44] make clean build, to be safe [08:44] ah, ok. [08:45] * wgrant tries. [08:51] Error: pg_config executable not found. [08:51] I get that in the middle of a 'make build' [08:53] Do you have launchpad-developer-dependencies installed? [08:54] I shall return after dinner. Hopefully germanium will give me everything I need while I eat. [08:56] wgrant: heh [08:56] wgrant: ta [08:56] wgrant: no, I don't. It seems to be broken. [09:03] ah, you need python2.6-dev [09:04] I've just got python2.6 from https://launchpad.net/~fkrull/+archive/deadsnakes [09:10] jml: Yeah, I'll probably replace it with python-all-dev and co. once I check it functions without 2.6 [09:10] We mostly just use the default version, but some things still point at 2.6, I believe. [09:12] Everything of note has been updated to just use the default, so it will hopefully work. [09:13] coole. [09:13] I'll keep hacking in my lxc for the moment. [09:13] Probably better for your sanity anyway. [09:21] yeah. [09:30] Interesting [09:30] pg_createcluster now fails if it's run in a nonexistent locale. [09:30] dpkg doesn't notice. [09:31] lifeless: Any objections to dropping !precise host support from /Running/LXC? [09:31] The instructions are like half the size. [09:32] +1 [09:32] * wgrant mauls. [09:54] jml: Just ran various tests successfully without 2.6, so I'll drop it from the deps. [10:01] wgrant: yay [10:28] wgrant: do you know why TestArchivePrivacySwitching is in LaunchpadZopelessLayer? [10:28] jml: Probably because whoever wrote it was wrong. [10:29] wgrant: heh. [10:29] jml: It should be ZopelessDatabaseLayer or DatabaseFunctionalLayer, since it probably doesn't need librarian etc. [10:29] wgrant: yeah, trying DFL now. [10:29] The difference between those two is now pretty much just the permission model. [10:29] Zopeless is still PermissiveSecurityPolicy [10:30] ah cool. [10:31] wgrant: how much work does "pretty much" hide? [10:31] jml: Working out what FunctionalTestSetup etc. does that is different. [10:32] I merged most of it in a large branch series in September, but am yet to obtain the courage to complete the merge. [10:32] Becauze Zope testing setup stuff is somewhat intimidating. [10:32] It's also not clear quite how it's best to do it. [10:33] yeah. [10:33] But I suspect an attribute on the test class to specify the security policy might work. [10:33] s/work/not be as terrible/ [10:33] that's what I was thinking [10:33] and then one by one change the test classes to do something better. [10:34] I wonder if there's a better mechanism than attribute (subclassing maybe?) [10:35] StupidLegacyTestCase? :) [10:35] well, we need to separate concerns [10:35] StupidLegacyPermissiveTestCase [10:35] Heh [10:36] actually, you could probably do something with testtools's `run_tests_with` thing. [10:36] publisher.prepareBreezyAutotest() [10:36] That's actually not a terrible idea. [10:36] Run away [10:36] oh yeah, that needs the librarian [10:36] It does, yeah. [10:37] Because inserting fake LFAs is hard, I guess. [10:37] It tries to upload chroots. [10:39] It could be taught not to participate in such stupidity, though. [10:40] I'm not sure if our test suite's antics are back to being hilarious rather than depressing yet. [10:45] I'm guessing getPubSource has side-effects that belie its name [10:45] Oh yes [10:46] makePubSourcePlusSomeOtherBitsActually [10:46] You can usually get away with the factory's makeSourcePackagePublishingHistory these days [10:46] SoyuzTestPublisher is deprecated and new usage of it incurs wrath. [10:49] oh [10:49] that's nice [10:49] ISTR the last time I tried this sort of thing the accepted wisdom was "use SoyuzTestPublisher because you'll never be able to get it right any other way" [10:52] is there a safe way to use the Librarian in tests without using a layer that provides it? [10:59] no [10:59] the layer is a thin shim on the fixture now [10:59] jml: The appropriateness of STP depends on the test. [10:59] but you still need all the bits joined together [10:59] You can sometimes use the FakeLibrarian. [11:00] Translations uses it for something in like one test. [11:02] If there were a safe way, then lots of test runs could be made a bit faster easily. I could, for example, switch this test to ZopelessDatabaseLayer and then just use that safe way to run this test with the librarian. [11:02] zopeless is no cheaper than zopeful [11:03] lifeless: I wasn't contending that point. [11:03] ZopelessDatabaseLayer is heaps cheaper than LaunchpadZopelessLayer though. [11:04] librarian, memcache and rabbit all take their toll. [11:09] I don't seem to have much luck with FakeLibrarian, usually. Maybe I'm not smart enough. [11:09] I tend to get incomprehensible storm tracebacks. [11:25] battery threat! === rick_h_ changed the topic of #launchpad-dev to: http://dev.launchpad.net/ | On call reviewer: rick_h | Firefighting: - | Critical bugs: 3.47*10^2 [11:30] https://code.launchpad.net/~jml/launchpad/narrow-commercial-celebrity/+merge/104236 up for review. [11:30] rick_h_: poor chap, you always seem to end up with my sad offerings. [11:30] jml: hah, all good [11:30] I have to go find electricity. Will be back sooner than you'd like. [11:30] and yet he comes back for more [11:30] cant be that bad [11:31] jml: ok, going to grab some grub, so will take me longer than you'd like :P [12:00] ok. I'm back. [12:02] rick_h_: are you far into reviewing that branch, because I thought of some cleanups that I can tack on to it. [12:02] jml: we have date/times now for the LP clinic at uds https://wiki.ubuntu.com/UDS-Q/LaunchpadClinic [12:03] jml: I'm peeking at it, but please go ahead and update and I'll get back to it in 10 then [12:28] rick_h_: ok, finally done [12:28] jml, ok, working on another review atm, but will get to it shortly [12:28] rick_h_: 30s per test run has a surprisingly strong negative effect on speed. [12:29] jml: tell me about it :/ [12:30] jml: I'm pretty sure it hasn't become worse since you defected. I guess you've just repressed the tramautic memories :( [12:31] wgrant: oh, I'm sure that's the case. [12:31] The brain does repress bad memories. [12:31] otoh, the nice thing with working on LP is there's so much obvious stuff to do [12:31] it's very hard to get stuck. [12:31] Indeed. [12:31] Indeeeeeeed. [12:31] So much to delete. [12:32] Like Translations, Answers, Blueprints. [12:32] And Soyuz to rip out. [12:32] or update, or fix, or refactor. [12:32] yeah [12:32] ripping out the buildmaster would be fun [12:32] step 1, rationalize the transactions [12:32] It, Soyuz and Translations have no business at all being in LP core. [12:32] The others might have some justification. [12:32] step 2, xmlrpc (or whatever) interface [12:32] step 3, yoink! [12:33] Yep [12:33] It just turns into a process which bridges the LP and buildd XML-RPC interfaces. [12:33] chunks of lp.codehosting could be ripped out pretty easily too. [12:33] I'm likely to rip out lp.services.sshserver into a separate project in the next few months. [12:33] As part of destroying poppy. [12:33] It's mostly self-contained. [12:33] wgrant: oh, I made lp:txsshserver once upon a time with roughly that intent [12:33] wgrant: maybe you should take over that project when you do it? [12:34] Maybe indeed. [12:34] it has no users. [12:34] All the better! [12:34] hmm. maybe I killed it :\ [12:34] One thing I do like about Code(hosting) is that the architecture isn't entirely stupid, unlike say most of the rest of LP. [12:34] apart from branchrevision. [12:35] wgrant: aww shucks. === matsubara-afk is now known as matsubara === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away [13:03] is the a graph of LP LoC over time? [13:04] , but cjwatson had a better one of the delta [13:04] I'm not sure how accurate ohloh's is. [13:05] heh, wasn't ready to seem my name up there in that [13:06] but as a re=me so it's ok :) [13:07] jml: so you're not removing the concept of commercial ppas, but this single getCommercialPPA helper then? [13:07] rick_h_: that's right. [13:08] ok, cool [13:08] ohloh's one tends to be out-of-date and imprecise [13:08] rick_h_: if we remove the commercial concept entirely, well, this is half (hah!) the work [13:08] jml: just making sure I'm following, saw the sea of red but some bits left and wanted to make sure I was 'getting' it right [13:08] :) [13:08] jml: The Soyuz's team blood, sweat and tears (mostly tears) went into making IArchive.commercial! [13:09] StevenK: sunk cost fallacy. [13:09] Woah [13:09] StevenK: Sorry, I meant to say that we really appreciate it and hope some day we can return the favour. [13:09] Never seen someone brave enough to invoke that on LP before :) [13:09] jml: Oh, I know it's a sunk cost, I'm just saying think of our tears. :-) [13:09] tears wash away easily with some soap and water...all good [13:10] wgrant: You mean what I said? [13:10] No, the sunk cost fallacy. [13:10] Ah. [13:10] It could be applied to harmlessly remove 90% of our functionality :) [13:10] how dare you use logic and management terminology in here! :) [13:11] * StevenK removes 90% of the JS, thereby removing the need for convoy or jsbuild. [13:11] * rick_h_ watches the site go boom! [13:11] It already was earlier today. [13:12] phew, glad I missed it then [13:12] Production thrice, qastaging once. [13:12] wgrant: well, there you'd compare cost/benefit of deleting with cost/benefit of user migration [13:12] Fun day. [13:12] ugh, what went kaput? [13:12] oh, also, perplexing thing: https://code.launchpad.net/~jml/txsshserver/trunk (200); https://code.launchpad.net/txsshserver (404) [13:13] jml: Deactivated project, probably. [13:13] ah ok [13:13] "This project is currently inactive" [13:13] rick_h_: anyway, I don't have commit so if you approve of that branch please land it for me. [13:13] jml: ah ok will do [13:13] going to have to footnote you on my ec2 bill :P [13:14] rick_h_: I left two LP instances running accidentally, I think to support some pair programming. I have a $300+ bill from last month. [13:14] ouch! [13:14] wgrant: how come you are so low on ohloh? [13:14] jml: Hm? [13:15] I'm constantly afraid of that. I have two of my own I keep running to have to keep thinking "hmmm, 2 ok, >2 doh!" [13:15] wallyworld, got a bill for $1,500 [13:15] It ran for 45 days [13:15] oooh...45 days?! [13:15] sinzui: for ec2? [13:15] that must have been pre the large instances? [13:15] wgrant: you don't appear on http://www.ohloh.net/p/launchpad/contributors [13:15] yes [13:15] ah, my bill [13:15] I'd have thought 45 days would be higher [13:15] jml: I'm probably on like the third page. [13:15] * wgrant hunts. [13:15] wgrant: 2nd [13:16] when I see these start ups with 100s of ec2 servers I cringe at that monthly bill [13:16] Ah, not too bad. [13:16] I am the top committer of the last 12 months, though :) [13:16] I think I'm on like the 6th page. :-( [13:16] We can all run starry-eyed into the arms of canonistack [13:16] Although sinzui is close behind. [13:16] hmm. [13:17] sinzui: yea, are there instructions for that? [13:17] I don't know how to get that report from ohloh [13:17] sinzui: Sounds good to me. Except no one seems to care. [13:17] jml: Code Analysis [13:17] jml: You can click on people in the legend to turn them off. [13:17] (ie. to dispose of PQM) [13:17] I'm not quite sure how it picks who to show. [13:17] Looks vaguely like the top 5 of the last year and then the top 5 [13:18] of all time [13:18] me? I have done bugger all for 12 months. I am atrifying. [13:18] jml: so side curiousity here, is this how people are getting listed in the software center? They have a commercial ppa and that's listed out as apps in software center? [13:19] rick_h_: yeah, that's right. [13:19] rick_h_: but only for commercial (i.e. they cost money) apps [13:19] jml: right, the newish paid apps stuff [13:19] yep [13:20] ok, just curious, wasn't sure how that stuff worked and this review has the brain thinking [13:20] jml: Aren't the proprietary free apps also there? [13:20] eg. Vendetta Online [13:20] Or whatever that free one is [13:20] this work we're doing with LP now is aimed at people being able to upload apps to developer.ubuntu.com and have them packaged and published to a PPA automatically [13:21] wgrant: I don't know about proprietary free. [13:21] wgrant,jml: http://people.canonical.com/~cjwatson/tmp/loc.png is a currently-up-to-date graph of delta per commit. But it's kind of hard to read; it looks mostly nice and negative but the net change over that time period is in fact +2755. [13:21] wgrant: I know that libre+gratis goes to extras (or the archve) [13:21] jml: I like to pretend that extras doesn't exist :) [13:21] :) [13:21] cjwatson: :( [13:22] cjwatson: do you have a graph of the integral of that? [13:25] jml: that's ohloh's graph, isn't it? [13:25] I guess you might want zoomed in and zero-based [13:25] oh yeah. [13:25] I think I'll just hook up bzr & ggplot2 in my own spare time :) [13:25] Some counter on lpqateam.c.c would be nice. [13:25] what I want is a command that tells me my score. [13:26] haven't you heard, you need graphite graphs these days to be cool [13:26] I just want to see if I'm negative. [13:26] with hourly commit points and fancy colored moving charts [13:27] and we all put them on a display on the wall over our desks [13:27] rick_h_: Speak for yourself, hipster. [13:27] lol [13:27] I'm on LP, I think that disqualifies me for any hipster movements [13:27] Haha [13:28] Nah [13:28] GitHub's too mainstream [13:28] when we get some BDD JS going then we'll chat about rushing house hipster [13:28] StevenK: yeah, if I'm negative that's a good thing to know. But I also want to be able to compete with people [13:28] where's jono when you need some new achievement badges [13:29] "negative LoC leader for month of May" [13:29] If we hook the LOC metric into Ubuntu Achievements, someone will get fired. [13:29] lol [13:29] StevenK: how so? [13:29] ... or I'll quit in disgust. [13:30] jml: There's an implied :-P [13:30] jml: but, http://people.canonical.com/~cjwatson/tmp/loc-cum.png [13:30] StevenK: oh, I just didn't (don't yet) get the joke. [13:31] cjwatson: thanks! [13:31] I'm waiting for my UI work to end on prod. [13:31] I wonder how it changes if we filter out database/schema [13:31] Since that's append-only. [13:32] Then I can rip out 2,000 lines of stuff. [13:33] jml: (easy once I remembered 'smooth cumulative'. gnuplot is ugly as sin but quick to hook trivial stuff up in.) [13:33] cjwatson: ah cool. [13:34] bzrlib + matplotlib ftw [13:35] yeah. I've enjoyed the stuff I've done recently with R & ggplot2 [13:35] Enjoying R? I think you're doing it wrong. [13:35] webulating graphs is too hard though [13:35] i.e. taking a one-off graph made and turning into something that can be hit on the web and is always relatively up-to-date [13:37] wgrant: the thing I really like is that it's really easy to use my data analysis again at a later point. Mostly my Python graphing stuff doesn't seem to work out like that. [13:37] Yeah [13:37] Maybe I should try that panda thing GvR posted on G+ about. [13:37] also, R is actually pretty good for interactive mucking around with data [13:38] even if it's rather poor considered as a language. [13:39] distributionmirror_prober: is that a candidate for splitting out? [13:40] Yes. [13:40] It should in all probability be a separate database. [13:40] The interactions with LP internals are minimal. [13:40] But it should probably be deleted and replaced with an existing third-party solution [13:41] ooh [13:42] its tests are an affront. [13:43] I'd like to fix them, but removing them & the code they test entirely would be much better. [13:44] Exactly. [13:45] Lots of Launchpad components like to pretend they're solving a unique problem. [13:46] So stuff like that happens :) [13:46] :) [13:46] james_w: hello [13:46] hi [13:46] I'm wondering whether I should make a celebrity team for which membership means "you can set commercial" [13:47] or whether I should re-purpose the software_center_agent celebrity somehow [13:47] Or add a field to distribution which permits it. [13:47] wgrant: e.g. Distribution.commercial_team [13:48] ? [13:48] nasty_proprietary_software_people, but yeah :) [13:48] hmm. [13:49] wgrant, hey. I saw something in scrollback about lp being down for you guys. are there incident reports for this? or anything I can look at. [13:49] I need a better name than either of them. [13:50] deryck: https://wiki.canonical.com/IncidentReports/2012-04-26-LP-release-performance-issues covers the first two of the day's issues, the other one is in the hands of IS, I believe. [13:50] 'commercial_admin' would be pretty good if it weren't already a heavily-loaded term [13:50] Yeah [13:50] wgrant, thanks! [13:50] deryck: Alternatively we can replace commercial with a shut_up flag that anyone can set, I guess. [13:51] s/anyone/the archive owner/ [13:51] Bah [13:51] jml: ^^ [13:51] heh. [13:51] * deryck votes for shut_up flah, too, FWIW. :) [13:51] flag [13:51] jml: I thought my team name was pretty catchy, though. [13:51] bigjools seemed to dislike the idea [13:52] wgrant: sure, it has zing. It could be easily misunderstood though. [13:52] abentley: ping, can you take a peek at this for me when you get settled? https://code.launchpad.net/~wallyworld/launchpad/revoke-access-delete-subscriptions-job/+merge/104198 [13:53] rick_h_: Yes, I'll have a look at that. [13:53] abentley: want to make sure it's all celery approved, not sure about the any special needs/better ways of doing things for that. [13:53] thanks! [13:53] wgrant: as it would read to me as "all of the propriety uploaders to d.u.c" rather than "the d.u.c bots" [13:53] True, true. [13:53] jml, I'd lean towards re-using the celebrity, but wouldn't be against a field on distribution I don't think [13:54] james_w: one thing about the celeb is that (I think) it's a Person not a Team [13:54] jml: It can be either. [13:54] wgrant: I mean that currently it is a Person. [13:54] People are teams, war is peace, etc. [13:54] Ah [13:54] wgrant: I don't know what the fallout of changing it would be. [13:55] It would break. [13:55] I think. [13:55] well, yeah, that's generally what we mean by "fallout of change" [13:55] Since I believe the XML-RPC API may actually set up an interaction as it. [13:55] Which won't work if it's a team. [13:55] But I may be misremembering, and it just passes it in as arguments to horrid methods, in which case it might work. [13:56] There's one easy way to find out... [13:56] I'm more worried about how we use it on our side [13:56] that's harder to co-ordinate. [13:56] Howso? [13:56] How do you use it on your side? [13:56] I thought you mostly poked the xmlrpc-private API [13:57] I don't know. [13:57] * jml looks [13:58] Ah [13:58] There's a single xmlrpc-private call, getOrCreateSoftwareCenterCustomer [13:58] Rest must be through lazr.restful. [13:58] yeah [13:58] So, you could create a team, add the agent to it, then make that team the celeb. [13:59] jml: have a bug to tie the MP against please? [13:59] well guess there's not a lot of QA for this removed is there :/ [14:01] rick_h_: https://bugs.launchpad.net/launchpad/+bug/992622 [14:01] <_mup_> Bug #992622: getCommercialPPAs is not needed < https://launchpad.net/bugs/992622 > [14:01] rick_h_: as you wish. [14:02] jml: yea, that ec2 land wants things to all be setup right, picky little thing. :) thanks. Landing now [14:02] rick_h_: thanks! :) [14:08] hmm [14:09] I have to write this code somehow. [14:11]