[00:00] poolie: more detailed sorting than just importance [00:00] ah no. [00:00] Query on tags? [00:01] Hmmph, already there [00:04] poolie: msged you that part of the discussion [00:09] mars: http://en.wikipedia.org/wiki/Pareto_analysis has some interesting notes :) [00:11] lifeless, and jml posted that nice table of bug data to the mailing list a little while ago... [00:12] :> [00:12] mars, I have a branch on lp:~jml/launchpad/bug-stats that kind of crappily adds such charts to each project on Launchpad [00:13] mars, I'd be very grateful if you made it less crappy and landed it. [00:13] we could use tags to describe caues [00:13] like missing-workflow [00:14] jml, early next year, when I switch to lean tooling as my main priority for a quarter. [00:14] mars: lean tooling? [00:14] leans his tools against the nearest wall [00:14] * jml drags his mind out of the gutter [00:15] mars, I might have made it less crappy myself by then :) [00:15] jml: hi-5 the gutter team ;-) [00:15] spm, :) [00:15] jml, LP Foundations: permanently snowed under [00:16] mars: on a larger scale, foundations might get more help from e.g. code if folk saw that daily ;) [00:16] lifeless, saw what daily? [00:16] * lifeless would actually like to disolve all the subteams. [00:16] mars: bugs in foundations that are critical/high [00:16] lifeless, others have thought the same [00:16] ah [00:17] http://en.wikipedia.org/wiki/Conway%27s_Law [00:17] good night all. [00:17] good night jml [00:18] ... "A real life example: NASA's Mars Climate Orbiter crashed because one team used United States customary units (e.g., inches, feet and pounds) while the other used metric units for a key spacecraft operation." :) [00:18] mars: s/LP Foundations/working at Canonical/ [00:19] lol [00:19] see why it is so difficult to get that critical 20% slack? :) [00:19] I'm serious :) [00:23] anyway [00:23] same [00:25] lifeless, so my wishlist: zero held-over bugs, all problems like OOPSes, testmode, and PQM queue visible on the same screen, 80% utilization for the project. [00:27] mars: I would express it something like 'zero queued defects, all team tasks on one screen, 50% load from stakeholder requests' [00:27] mars: (I think we still estimate too low, and need a buffer while we rearrange things) [00:27] yes, that sounds reasonable [00:28] mars: I use the word defect to separate out from wishlist bugs, which I think many things should just be punted to. [00:28] right [00:29] I think it scares folk though - folk /want/ a cozy little area-of-concern. [00:29] which is the big thing to break though I think. I'll need to be really clear about this in replying to the experiment thread. [00:33] thumper: fwiw https://code.edge.launchpad.net/~vila/bzr/releasing-clarified/+merge/10854 is oopsing consistently (oops-1427ec27) [00:33] ??!? [00:34] * thumper wonders why [00:34] poolie: does it oops on prod? [00:34] yes [00:35] i was playing with vote requests to understand them [00:35] and i think i have the db in a state where it can't render that page [00:35] Module lp.code.model.branchmergeproposal, line 600, in getUsersVoteReference [00:35] query).one() Module storm.store, line 1117, in one [00:35] raise NotOneError("one() used with more than one result available") NotOneError: one() used with more than one result available
[00:38] \o/ [00:46] heh [00:47] poolie: were you reassigning a pending review? [00:47] is that even possible? [00:49] yes [00:50] poolie: so what were you doing? [00:50] i was playing with bug 456643 and bug 487327 [00:50] Bug #456643: Merge proposal claims a review even though I specified a different type [00:50] Bug #487327: Floating "claim review" button [00:50] so requesting reviews of various types from people and teams [00:51] i meant to do this on staging but i think url completion trapped me [01:07] poolie: you found a bug [01:08] poolie: surprise [01:08] poolie: please find bugs by breaking staging next time [01:08] thumper: it may not be possible on staging [01:08] yeah, I know [01:08] but in this case it is [01:09] staging breaks itself without much help [01:09] seriously, i did mean to be screwing around on staging [01:09] poolie: can you file a bug with what you most remember doing? [01:09] poolie: tag it with oops and mark it high [01:09] poolie: I'll try to get this fixed early next week for a release-critical [01:09] sure [01:13] how do you reassign a pending review? [01:14] bug 489019 [01:14] Bug #489019: requesting code reviews puts the mp into a state where it can't be displayed [01:14] oh i ese [01:14] see [01:14] that yellow button [01:15] so [01:15] no, i wasn't doing that [01:18] poolie: you used claim review on the team? [01:18] i don't recall [01:18] damn [01:18] i could well have done [01:18] i didn't reassign the reviews because until now i didn't know it was possible [01:18] there's no log? [01:20] no [01:20] well, there may be an apache log [01:20] there is an outstanding bug about history/log [01:20] which is kinda in progress, but not [01:27] thumper: ok i have it [01:27] poolie: yes... [01:29] https://bugs.launchpad.net/launchpad-code/+bug/489019/comments/2 [01:29] Bug #489019: requesting code reviews puts the mp into a state where it can't be displayed [01:33] basically, claim two reviews simultaneously [01:56] beuno: if you get back, and want to talk pics, email me [01:56] s/email/poke/ [02:34] * mwhudson :(s at the failure on buildbot [02:35] * mwhudson also :(s at the way the code import worker tests are parameterized [02:40] i really think we should switch to a single url in the code import code [02:41] not have this svn_branch_url/git_repo_url/... distinction [03:18] mwhudson: that has been requested by stub for the bzr-hg addition [03:21] thumper: yeah i know [03:21] oh, ok [03:25] thumper: when you say in bug 382825 there's no more 'other' section [03:25] Bug #382825: code review confusing 'other' section [03:25] is that because of a new change you've just done? because there is one at the moment [03:25] poolie: yeah, it has another title now :) [03:25] which is? [03:25] poolie: and there is a different bug for that one [03:25] oh ok [03:30] ... [03:31] how'd we end up with so many new bugs... [03:31] "New" not new [03:31] just in case there was confusion ;-) [03:32] I'm sure there was something else I was going to try to fix today [03:32] thumper: which do you mean ? :) [03:32] status: new [03:32] * lifeless waits for apt [03:33] i decided to file bugs rather than grumbling or wondering :) [03:33] but it's ok, you had your revenge by filling my mailbox while i had lunch :) [03:33] poolie: it wasn't just yours I was referring to [03:33] thumper: you can try sticking boto in download-cache as an egg, might fix buildbot [03:34] mwhudson: is that what is missing? [03:34] * lifeless hates on boto [03:34] thumper: yes, the tests now try to import it [03:35] mwhudson: which tests? [03:35] does this have something to do with twisted 9 ? [03:37] thumper: no, this is the fix for the ec2 test --headless problem jtv landed [03:38] mwhudson: didn't you have a fix too? [03:38] thumper: mine didn't have a test [03:38] heh [03:39] * thumper suggests jtv fix it :) [03:39] i guess he'll be up fairly soon [03:44] * thumper is about to hurl in the towel [03:45] for today, or because of something specific? [03:49] for today [03:49] feels like it has been a draining afternoon [03:50] fair enough [03:50] have a good weekend! [03:52] I'll pop back later to try to land those branches [05:16] cheerio thumper [05:16] mwhudson: i can't remember if you actually answered before: [05:17] is me getting rid of python logging going to actually break anything for you? [05:17] re https://code.edge.launchpad.net/~mbp/bzr/remove-logging/+merge/15306 [05:17] poolie: i'm not sure you actually asked, i did wonder why you requested a review from me [05:18] * mwhudson looks [05:23] poolie: i can't see anything obvious [05:24] k [05:24] more just thought i should let you know [05:24] in case you were hooking into bzrlib this way [05:25] we were sort of planning too [05:25] but luckily we never got around to it :-) [06:09] [...] would necessitate a short outage. I'd estimate ~ 10-20 seconds. [06:09] hope springs eternal ;) [06:09] :-) [06:10] it should be just shutdown apache; start it up with shiny new configs and things just work. [06:11] But it's a major - potential - disruption to a critical service, so not one I'm willing to JFDI [06:12] agree [06:13] ideally - we'd schedule it for when I'm in a car and driving up to sydney for the xmas party and thus unable to assist in any failures. but that'd be lazy/optimisim at work ;-) [08:30] goodmonring === jamesh_ is now known as jamesh [08:53] golly... you guys should come up with some way to help users see if there's an existing bug that matches the subject of the one they want to report to help eliminate duplicates... [08:53] something like.. idk.. you type the short description; click a button before continuing that gives you a list of possible dupes.. [08:57] -_- | sorry, just getting irritated when there's 32 dupes of a single bug that very accurately describes the issue. Even if they don't see that one, they should at least see the 32 others [10:16] flacoste, does having testtools on PyPI make it easier to use as a Launchpad dependency? [10:37] jml: it shouldn't matter much where it's located. the most important thing is that it has a proper setup.py, so that eggs can be generated. having it on PyPI only makes it slightly easier to download the sdist. [10:46] BjornT_, thanks. [10:53] BjornT_, you might be interested in the assertThat stuff that lifeless recently added to testtools. [11:00] jml: what does it do? or where can i easily see it? [11:01] BjornT_, http://bazaar.launchpad.net/~testtools-dev/testtools/trunk/annotate/head:/MANUAL is probably the best starting point [11:02] BjornT_, it's a way of avoiding adding lots of domain-specific assertFoo methods to a base class by moving the assertion logic out to separate classes that implement a well-defined interface === matsubara-afk is now known as matsubara === danilo_ is now known as danilos [12:23] what needs to be done to close off bug 139855? [12:23] Bug #139855: Display stats about PPA usage [12:24] I think it just needs scheduling - afaik, wgrant has done a lot of the required backend work for it. [12:25] noodles775, that's good. :) [12:25] Well, when I say 'just', I'm sure it depends on exactly *what* stats are required etc. :). [12:25] noodles775, but I actually meant, what if someone off the street wanted to fix it. [12:25] noodles775, jml: The log parser has been refactored, so the backend work is just a couple of dozen lines of code. But then there's UI. [12:26] The UI shouldn't be complicated - we've left space in the portlet on the PPA index page for it. [12:26] wgrant, what is it that makes the UI more complicated than just a number? [12:26] jml: A number where? [12:26] Per binary? [12:26] Per binary name? [12:26] Per source? [12:26] Per release? [12:26] Per version? [12:26] Argh. [12:26] lol [12:26] I see. [12:27] Yay Soyuz/ [12:27] what do you think people care about? [12:27] So it's aggregating the appropriate stat that's the issue? [12:27] noodles775: Right. [12:27] jml: I do not know. [12:29] wgrant, were it me, I think I'd like to have the raw data available for all of those, and I'd probably like some kind of aggregated statistic. [12:30] jml: Right. [12:30] so how do we go about picking one? [12:30] The stats will more than likely be keyed on (day, country, archive, binarypackagerelease), and from there you can do a lot. [12:32] is this the kind of change that would show up on edge before a rollout? [12:32] Not unless the backend makes it onto production first. [12:32] and that's work that's not yet done, right? [12:32] Correct. [12:33] Most of it is done (the refactoring was the big bit), but not the actual feature addition. [12:33] ok. I was just wondering, maybe we could land an obviously wrong UI shortly after the release, and then learn from our experiences using it on edge. [12:34] by "obviously wrong" I mean "our best guess made with little deliberation", which is roughly the same thing. [12:34] Right. [12:37] we could even blog about it and post to the mailing list to get feedback from users who aren't intimately involved with Launchpad development. [12:37] I know the Bazaar guys really want this, and I'm confident they'll have informed & interesting opinions :) [12:37] Definitely good ideas. [12:37] I'm glad you think so. :) [12:38] so, about that backend work... [12:39] wgrant, are you planning on doing it before the rollout? [12:39] jml: I'm not. I'm busy with real work and other stuff. [12:40] jml: Basically, it needs PPA versions of cronscripts/parse-librarian-apache-logs.py and lib/canonical/launchpad/scripts/librarian_apache_log_parser.py [12:40] Plus the DB tables. [12:41] * jml looks. [12:42] There's not much remaining in either of them. [12:44] indeed not. [12:47] wgrant, which DB tables, exactly? [12:48] jml: ArchivePackageDownloadCount, or something along those lines. [12:48] LibraryFileDownloadCount is the existing one. [12:51] wgrant, thanks. I've added a comment to the bug summarizing this discussion, in the hope that I or someone else will act on it. [12:52] jml: Looks good. [13:01] * wgrant sleeps. [13:02] wgrant, g'night [14:09] Hi BjornT_ , would you be able to look over the last comment on bug 487009 (and the linked diff) with your SA goggles on? [14:09] Bug #487009: Generalise IBuilder [14:28] jml: testtools in PiPy: it makes thing a little bit easier, in that you can rely on the default buildout behaviour to get updates, but once the release is in the download-cache, it doesn't make a big difference [14:37] noodles775: it seems a bit odd to adapt a job type. adapting an actual job seems more natural [14:37] BjornT_: yes - that would be much better (I'd headed down that path as I initially had a factory there). [14:38] BjornT_: although, that would then require adding a factory that knows about the different types, wouldn't it? [14:38] That was what I was trying to avoid too (having multiple places that need to know about the different types) [14:39] * bigjools-afk wonders if he has some electromagnetic field that causes power cuts where ever he goes [14:39] noodles775: no. you can mark the job as providing the right interface, depending on which job type it is [14:40] BjornT_: ah, great! [14:40] noodles775: you can see an example in BugTask._init() [14:40] * noodles775 looks [14:42] noodles775: same sort of thing happens in Archive.__init__ [14:43] Right. === bigjools-afk is now known as bigjools [15:40] hi guys, im doing the canonical-application-support sec (https://blueprints.edge.launchpad.net/ubuntu/+spec/server-lucid-canonical-application-support) I was wondering why launchpad doesnt use python-subversion rather than python-svn since python-svn is not in main? [15:45] zul, If I had to guess, it's because either launchpad-cscvs or bzr-svn depends on python-svn in particular [15:45] jml: thats reasonable then [16:01] zul, it's probably worth sending an email to mwhudson (and the launchpad-dev list) about this. [16:01] jml: will do [16:01] zul, if the python-svn dep is because of launchpad-cscvs, it's possible we'll phase this out. === beuno is now known as beuno-lunch === beuno-lunch is now known as beuno [18:06] rockstar, how's your UI review crusade going? [18:27] lifeless, I can't remember if it was you or poolie who suggested I open N tabs of bugs in (firefox, chrome) and time how long it took for all the tabs to load [18:27] does anyone have a moment to verify a JavaScript bug on db-devel? I'm getting it consistently on local, but staging is clean. [18:28] lifeless, but for ten tabs, ~24s for firefox, ~13s for chrome. same bugs. [18:28] ah, wait a minute, it might be the devmode switch [18:28] * mars tries it [18:29] jml, is that with a clean firefox profile? [18:29] mars, no, it's with my dirty chrome & firefox profiles. [18:29] ah. Just wondering. Things like Firebug can really mess with the results of such a test. [18:31] mars, kind of. [18:32] mars, depends on what I'm measuring, really. I'm not going to uninstall firebug just because it makes Launchpad faster :) [18:32] :) [18:33] ok, the error is still present on my local setup [18:34] so, can anyone spare a moment to verify that the bug is present on db-devel when run locally? [18:34] I can have a go [18:34] jml, cool, thanks [18:34] I might give up if it's too hard for a friday evening [18:34] jml, shouldn't be [18:35] ~17s on "safe mode" firefox, no extensions or themes. [18:35] jml, please pull the latest db-devel, edit configs/development/launchpad.conf and shut off devmode, make schema && make run, and visit any bug. [18:36] mars, this will take a while. I haven't run make clean in my db-devel branch since the python switch [18:36] no problem [18:36] thanks for the heads up though. I'm doing the same here. [18:39] and the bug is /still/ there [18:39] mars, is it in db-stable? [18:40] hmm, haven't tried that yet [18:43] jml, watched your "Teach me Packaging" talk. That was interesting, and a nice presentation idea. It worked really well with the skills of the audience. [18:43] mars, thanks [18:44] mars, I stole the presentation idea from Steve Holden, who did a "Teach me Twisted" talk at the preceding PyCon. [18:44] mars, he had two hours, and a bottle of whisky. [18:44] heh [18:49] jml, so could you successfully create your own python packages after that talk? [18:49] I know other people appear to have an easy enough time creating them [18:50] mars, almost. It's not actually a great way for me personally to learn -- I should have been more prompt in writing up the notes and blogging them. [18:51] and the bug is present on db-stable as well [18:53] with devmode on and off [18:54] but not on staging [18:55] thus, it might be caused by something in r8722..8729 [18:55] yes [18:56] when does the next staging update take place? [18:57] that, I don't know. [18:57] something to add to the future developer control-panel [18:58] in the name of making blockers visible [18:58] losas, do you know when the next staging update will be? [18:58] jml, any luck reproducing the error on your local system? [19:01] mars, I'm looking at a bug page. What's the error? [19:01] jml, A JavaScript error, 'display_name' is undefined [19:02] mars, yes, I get that in firefox [19:03] jml, ok, thanks. So it does happen locally. And I don't know why I am not seeing it on staging. [19:03] it is present on r8721 as well, so before the last update. [19:07] mars, http://pastebin.ubuntu.com/329617/ does rather look like the bug subscriber code has changed [19:08] mars, which seems to be where the js error is coming from [19:11] jml, yes, so perhaps the CSS classes are generated differently now [19:13] yep, quite possibly. Line 769 of the diff switches the CSS to use a different execution path [19:13] unlike line 756, which uses the old code-path [19:15] I am still suspicious though, because r8721 locally shows the error. This diff is for r8722.. [19:34] mars, I don't get it in r8721 [19:35] jml, bzr branch -r8721 ? [19:35] I did have that correct? [19:35] mars, 'bzr revert -r8721' in db-stable is what I did [19:35] then Ctrl-Shift-R to be sure I wasn't getting stale js [19:36] yes, I did control-F5 [19:37] oh wait [19:38] I'm seeing it in trunk as well [19:38] so it must not be a YUI3 problem [19:38] no, I do get it in r8721 [19:39] ok [19:39] mars, if you are seeing it in trunk, then I recommend using the stable branch and winding back (starting from 9944, which is 8722 in db-stable) until you don't see the error any more [19:39] mars: staging update is finishing up now, should be done in 2-3 hours or so [19:40] Chex, great, thank you [19:40] jml, ok, bug filed: #489342. Thanks for all the help so far. [19:40] Bug #489342: "'display_name' is undefined" JavaScript error on development [19:40] np [19:41] * jml off downstairs to stay awake for another couple of hours === matsubara is now known as matsubara-afk [21:08] zul: launchpad (through cscvs) depends on *both* python-svn and python-subversion [21:09] zul: if from this you deduce that cscvs isn't the greatest thing ever, well... [21:41] jml: me [21:42] jml: poolie may or may not have suggested it; I khnow I have :)