[00:32] * poolie thinks, again, there ought to be an incipient mp for every branch [00:32] whether it's been actually proposed or not [00:33] spiv, the thing is writing the extension is easy, deploying it is the hard part [00:34] poolie: but how would you know what the destination branch is? [00:34] maxb: all of them :P [00:34] poolie: why is deploying the extension hard? [00:34] wallyworld: not really [00:34] wallyworld: file one [00:36] i'm just once-bitten by attempting to change the oauth deployment [00:36] how does one document a flag? [00:36] thumper, read dev.launchpad.net/FeatureFlags and if that's not enough, ask me [00:36] ok [00:36] if it's not enough i'd really like to make it better [00:38] poolie: https://launchpad.net/+feature-info lists undocumented features [00:39] poolie: how do they become "documented"? The docs don't cover that [00:40] hm [00:40] ah, see "adding a new flag" [00:40] i'll expand that a bit [00:41] or are you just documenting an existing one? [00:41] I'm wanting to add a new flag, but I should also document the missing ones [00:41] The missing ones are documented in a branch of mine. [00:42] poolie: yeah, that section is very easy to miss [00:42] wgrant: thanks [00:42] poolie: who is able to edit the feature flags on prod? [00:42] losas [00:42] only [00:42] well, ~admin [00:43] i think that's correct [00:43] is that better now? [00:46] thumper, please try to match the naming guide in that page too [00:46] or object to it and we can change it [00:47] But if you invent another non-standard scheme I will cry :) [00:47] exactly, so please don't [00:47] the naked juggling is distracting enough without people criying too [00:52] StevenK: when do you want to chat? [00:52] thumper, all good? [00:53] poolie: the only thing making me cry right now is how we are currently hiding recipes [00:57] thumper: can you remember where that code is where you check for None in the formatter? [00:57] wallyworld: yes [00:58] thumper: and please tell me where it is mr smarty [00:58] wallyworld: lib/lp/code/browser/sourcepackagerecipe.py [00:58] thanks [00:59] archive_picker [00:59] * wallyworld sad - there's actually no existing tests for NoneFormatter [01:26] StevenK: btw [01:26] StevenK: there is a jenkins openid plugin now [02:01] whoa [02:01] where did the loggerhead exceptions go [02:03] StevenK: how's the recipe popup build request review going? [02:04] wallyworld: You mean, the one I wasn't actually doing? :-) [02:04] StevenK: you asked me questions about it so i assumed you had claimed it :-) [02:05] wallyworld: Nope :-) [02:06] StevenK: np. i'll find another sucker^H^H^H^H^H esteemed developer to do it [02:06] wallyworld: I'm a little confused [02:06] wallyworld: I thought you were working on manually requested builds [02:06] lifeless: i always confused [02:07] wallyworld: but I saw you file a but about manually requested daily builds that aren't stale [02:07] lifeless: yes, but the request build popup work was done before that [02:07] which is a /totally/ different use case [02:07] so i have 2 recipe related branches up for review [02:07] the 2nd depends on the first [02:07] due to shared javascript [02:07] lifeless: make sense now? [02:08] wgrant: want a small review? https://code.launchpad.net/~wallyworld/launchpad/add-link-to-noneformatter/+merge/50088 [02:09] * wallyworld has 3 active reviews to try and get off his list [02:20] Yippie, build fixed! [02:20] Project devel build (448): FIXED in 5 hr 59 min: https://hudson.wedontsleep.org/job/devel/448/ [02:20] * Launchpad Patch Queue Manager: [r=leonardr][bug=649252] add an unsubscribe link to non-verbose bug [02:20] notifications [02:20] * Launchpad Patch Queue Manager: [r=leonardr][no-qa] isolate bin/lint.sh's use of grep from the [02:20] environment [02:20] * Launchpad Patch Queue Manager: [r=danilo][ui=none][no-qa] Fix rST and Sphinx errors in our top-level [02:20] docs, and make the main page more interesting. [02:22] whats the glue to say 'does this class implement ISomething or other' - when querying, not declaring [02:36] hi huwshimi [02:36] sinzui: Hey therwe [02:36] *there [02:37] huwshimi: I recall writing about the loss of colour from the involvement portlet. but I cannot find it in my email :( [02:38] huwshimi: So pretending I have dot written about this. do you want to mumble to discuss the issues that users have raised? [02:38] sinzui: Sure, that'd be great [02:40] lifeless: what do you think about converting feature flag values of "off" and "false" to False ? [02:47] thumper: i'm pretty unkeen [02:47] thumper: what issue are you trying to solve? [02:47] lifeless: I'm adding a feature flag "code.recipe.beta" to control the message and the beta icon [02:48] lifeless: so we'll add a rule that says code.recipe.beta -> "on" [02:48] lifeless: but if we set a rule to say code.recipe.beta -> 'off' [02:48] thumper: theres a rule for this already isn't there ? [02:48] lifeless: I want it to work [02:48] no [02:48] we have code.recipe_enabled [02:48] which is different [02:48] I'd like to rename that to "code.recipe.enabled" [02:49] enabled controls whether we barf on recipe actions [02:49] thumper: so this is to let you go to non-beta without tying it to a deploy ? [02:49] beta is just to show the messages [02:49] yes [02:49] exactly [02:49] and then you'll delete both rules entirely. [02:49] aye [02:49] thumper: can you tell me where our existing web service api tests are? i've had a boy look but can't see them [02:49] I would just do the current thing - 'if : ... ' [02:50] yeah, I know I _can) [02:50] thumper: the default is None/'', and you can add rules of '' [02:50] I just think it would make it more clear [02:50] I have think it makes it more awkward [02:51] getFeatureFlag('some.feature') should be False if some.feature == "off" or "false" [02:51] Not at that layer [02:51] and I really don't see the argument for a higher layer at the moment [02:51] it seems like cruft [02:51] find [02:51] fine [02:51] we can agree to disagree and I'll not change it [02:51] I agree that there is some confusion [02:52] lifeless: what do you have against implementing expected semantics for 'off'/'false' meaning Bool(False) ? [02:52] one of the reasons I have the opinion I have is that 'no rule implies no value' is kind of predictable. 'no rule implies a specific value' needs some sort of schema [02:53] wallyworld: the current /layer/ is simple strings. Its schemaless and typeless. [02:53] wallyworld: this makes it extremely robust in some ways, and frustrating in others. [02:53] ok. [02:54] wallyworld: given that we version the interpreter for this data (the appserver processes) multiple times a day, a failure in this area would be extremely messy. [02:54] yep :-) [02:54] I've no strong objection to a higher layer being built : but I see no benefit to building it either for simple booleans because the current implementation meets our functional needs [02:55] but that's what tests are for :-) [02:55] wallyworld: tests cannot catch this [02:55] we had one failure from features already, took the site down [02:56] -the whole site- [02:56] i'll take your word for it. not familiar enough with it to have an opinion either way. but on the surfaces, everything should be testable in an ideal world [02:56] wallyworld: the rules are in the production DB [02:56] why not also in stable or staging? [02:57] stable is a branch, not a DB, and the rules can have commercial-in-confidence values in them. [02:57] as for staging, it /may/ have the same rules, or it may have different ones [02:57] lifeless: sorry ETERMINOLOGY. i just meant the test databases [02:58] I'd rather see small helpers when folk solve tricky problems here built up than the core of the layer become typed. [02:58] i'd say there's an argument that our test environment should replicate the prod one, in this case having the same core database values. but i take the point about the privacy issue [02:59] and booleans have a comprehensive solution now: there's nothing you can add to make our handling of booleans more pithy, correct or powerful [03:00] wallyworld: folk need to be able to change rules on [qa]staging to be able to check their patches work etc [03:00] wallyworld: it would be disruptive to forcibly sync those on a high frequency basis [03:01] agreed. but people also should do some testing with data reflective of what's in production if possible (akin to progression vs regression testing in a way) [03:02] lifeless: so since i have your attention, do you know where out web service api tests are? [03:04] what sort of such tests [03:04] like [03:04] tests for a given api [03:04] or tests that the api is wired up [03:04] or ... ? [03:05] i want to use a web api client to invoke something annotated with export_write_operation() for example to see that it works [03:05] so i guess tests that the api is wired up [03:05] given that the underlying functionality is unit tested separately [03:06] thats generally done by a smoke test of that api [03:06] AIUI [03:06] but testing that what comes over the wire is as expected [03:06] test.*webservice, spread throughout the tree [03:07] oh. thanks, i'll take a look. i just wanted a starting point to current practices before diving down the wrong rabbit hole :-) [03:08] lifeless: ah i see my problem now. there's web services already on the domain model i'm working with but no tests written, so i couldn't find any examples to start from [03:09] for the new web services i'm adding [03:09] Argh. Where is the representation of an archive as a link hiding? [03:09] StevenK: PPAFormatterAPI ? [03:09] are [03:09] canonical.librarian.ftests.test_web.LibrarianWebTestCase.test_404 [03:09] canonical.librarian.ftests.test_web.LibrarianWebTestCase.test_oldurl [03:09] still known as intermittent failures? [03:09] lifeless: Yes, they failed yesterday [03:09] StevenK: or something like that? [03:14] huwshimi: http://pxtoem.com/ [03:19] wallyworld: Thanks, but that isn't in the trace back [03:19] StevenK: pastebin? [03:19] AH! [03:20] The information is spat out before the traceback [03:21] wallyworld: You could give me a clue tracing back from the page template to the code, though. [03:22] StevenK: ok. wanna throw the tb in pb? [03:22] wallyworld: [03:22] wallyworld: I think that's the line. If not, it's sourcepackagerecipe-index.pt, unchanged from devel [03:23] StevenK: should we mumble? [03:23] wallyworld: Sure [03:29] simple review lifeless? https://code.launchpad.net/~thumper/launchpad/recipe-feature-flag/+merge/50092 [03:30] just pushing some lint cleanup [03:31] thumper: get in line :-P [03:31] take a number [03:31] wallyworld: ok, you do it :) [03:31] thumper: i'll do yours if you do mine :-) [03:31] ok [03:31] which one? [03:32] thumper: Check your indenting line 43 of the diff [03:32] thanks. i've got 3 and our wip limit is exceeded on the board [03:32] StevenK: that was in the lint cleanup [03:32] StevenK: the diff is updating as we speak [03:34] thumper: which one? there's the simple link formatter which could be done first [03:34] wallyworld: hit me [03:34] thumper: https://code.launchpad.net/~wallyworld/launchpad/add-link-to-noneformatter/+merge/50088 [03:34] the recipe ones will take a lot more effort [03:34] but still need to be done [03:37] wallyworld: TestXHTMLRepresentations? [03:37] wallyworld: shouldn't that be TestNoneFormatter or something? [03:37] ? [03:37] * wallyworld looks [03:37] wallyworld: have you looked at test_tales? [03:40] wallyworld: 'cause I don't understand you test_shorten_traversal [03:40] thumper: didn't see test_tales. since tales.py was in lib/app/browser i looked for existing tests in lib/app/browser/tests [03:40] * thumper nods [03:40] from lp.testing import test_tales [03:40] or more accurately [03:41] from lazr.restful.testing.tales import test_tales [03:41] thumper: the test_shorten_traversal was my attempt to unit test that method based on my reading of it's intended behaviour [03:41] wallyworld: since there is no docstring, I can't tell what it should do [03:41] :) [03:42] thumper: yeah, but there's no docstring on the traverse() method that is being tested either, so it's catch 22 [03:43] thumper: i'll look at the existing test_tales stuff and fix the tests accordingly [03:45] \o/ tests pass [03:51] wallyworld: ok [03:52] thumper: the TestXHTMLRepresentations was a cut'n'paste gone wrong :-( [03:52] wallyworld: good :) [03:52] wallyworld: I was wondering if I was understanding it :) [04:02] thumper: mp diff updating as i type this [04:06] * wallyworld dislikes how long it takes for the diff to be updated [04:07] wallyworld: Patches welcome? [04:07] StevenK: isn't it controlled by a cron script? [04:07] Two, in fact [04:08] hey, remember when Launchpad devs didn't use merge proposals at all? those where the days [04:08] hi beuno [04:08] ah, when i was a wee lad, i had to walk to school through 10 feet of snow with no shoes [04:08] beuno: notice we just went over the 50k mark? [04:08] and speaking of merge proposals! [04:08] thumper, I did, and I \o/ed [04:09] :) [04:09] it kinda blew my mind, I didn't realize the numbers where growing so quick;y [04:10] beuno: still mostly canonical people using them [04:10] I guess the biggest consumer of MPs is LP? [04:11] I bet U1 can give LP a run for its money [04:11] thumper: just to double check - you don't know of any existing web service tests for recipes that i can't find? [04:11] LP only has like 7000 branches. [04:11] beuno: I wonder how to test that theory [04:11] So it can't be many of the MPs. [04:12] https://lpstats.canonical.com/graphs/BranchMergeProposalsNewPerDay/20100218/20110218/ [04:12] I'm pondering crafting some SQL to get the answer [04:12] NotFound: Object: , name: u'lp.js' [04:12] ? [04:12] * beuno tries to manually add up all the branches for u1 [04:12] thumper: That's a really encouraging graph. [04:13] https://lpstats.canonical.com/graphs/BMPNewRegistrants/20100218/20110218/ [04:13] wgrant, how do I find out how many branches a project has? [04:13] hi all [04:14] it only seems to tell me about active branches [04:14] https://lpstats.canonical.com/graphs/BranchMergeProposalsProjectsNewMergeProposals/20100218/20110218/ [04:14] hi beuno [04:14] heya poolie! [04:14] beuno: select show all branches [04:15] I did, "Any status" [04:15] Ubuntu One Servers has 345 active branches owned by 31 people and 3 teams. There were 716 commits by 19 people in the last month. [04:15] but that's not useful [04:15] flacoste: hhhaa - XXX flacoste 2008/04/24 This should be moved to a [04:15] flacoste: I've half-implemented said setTarget [04:15] In SQL: "PackageBuild.archive = Archive.id AND False" ... D'OH [04:15] 10294 | launchpad | f | 5103 [04:15] 9514 | ubuntuone-servers | f | 3462 [04:15] 1186 | bzr | f | 1473 [04:15] 8920 | drizzle | f | 1279 [04:15] 23444 | examiner | f | 1109 [04:15] 13681 | openlp | f | 939 [04:16] 12784 | ubuntuone-client | f | 754 [04:16] From October. [04:16] Right column is the count [04:16] So U1 probably wins by now. [04:16] :( [04:16] wallyworld: U1 client already does? [04:16] BAH [04:16] wgrant: ^ [04:16] right, we have desktopcouch, bindwood, android clients, iphone [04:17] beuno: Are they in a project group? [04:17] wgrant, yes, all under ubuuntuone I think [04:19] name | count [04:19] ------------------------------+------- [04:19] launchpad-project | 5385 [04:19] ubuntuone | 5042 [04:19] bazaar | 2339 [04:19] But that is outdated :( [04:19] well, still surprising to see more activity in LP [04:19] October is not that far away [04:20] wgrant: Tempted to get that run on prod :-) [04:20] wgrant: With "LIMIT 3" or so [04:20] wgrant: go on, ask spm [04:21] thumper: staging's only a week out of date; you can do it :P [04:21] wgrant: ok [04:21] From the start of 2010: [04:21] launchpad-project | 2375 [04:21] ubuntuone | 2271 [04:21] bazaar | 1079 [04:21] ayatana | 959 [04:21] So we still win!? [04:22] thumper: SELECT project.name, COUNT(*) FROM branchmergeproposal JOIN branch ON branch.id=branchmergeproposal.target_branch JOIN product ON product.id = branch.product JOIN project ON project.id = product.project GROUP BY project.name ORDER BY COUNT(*) DESC; [04:22] thumper: watch out with your recipe FF-only branch. [04:22] thumper: request-daily-builds uses the config option. [04:23] And we can't make an FF apply only to a script, AFAIK. [04:23] launchpad-project | 8912 [04:23] ubuntuone | 6695 [04:23] bazaar | 4294 [04:23] I wrote it from scratch [04:23] [15:20:50] wgrant: go on, ask spm <== have I finally got you lot scared of me. *FINALLY*!?!? [04:23] spm: Ask wgrant? [04:23] * StevenK cackles [04:23] heh [04:24] * StevenK notes that Storm doesn't like 'is' [04:24] But it returns a BOOLEAN! [04:26] its interesting - bazaar has 1/5th the devs, and 1/2 the mps [04:26] lifeless: But Bazaar also doesn't suck. [04:28] I guess bazaar probably has the most community as well [04:29] True. [04:29] spm: r12395 to nodowntime, pls. [04:30] and, launchpad was the first of all those to start using mps, if we look at totals [04:30] bazaar slowly moves away from bundlebuggy after a while [04:30] True. [04:30] That's why I took the numbers from 2010 onwards. [04:30] When everyone should have been using MPs. [04:31] * beuno nods [04:31] \o/ [04:32] one query for the milestone targeted bugs portlet. [04:33] lifeless: But how slow? [04:34] wgrant: https://bugs.launchpad.net/launchpad/+bug/717394/comments/2 [04:34] wgrant: I expect to be close to that, am assessing the generated sql now [04:35] argh, search builder fail [04:35] constraining on distro series using any -> noop, no constraino. [04:35] beuno: ... [04:36] launchpad-project | 3879 [04:36] ubuntuone | 3525 [04:36] bazaar | 1548 [04:36] ( [04:36] SELECT BugTask.distroseries, COUNT(BugTask.bug) FROM BugTask, Bug WHERE Bug.id = BugTask.bug AND ((BugTask.status = 10) OR (BugTask.status = 15) OR (BugTask.status = 20) OR (BugTask.status = 21) OR (BugTask.status = 22) OR (BugTask.status = 25)) AND Bug.duplicateof is NULL GROUP BY BugTask.distroseries [04:36] for branches created after the start of 2010 [04:36] ah, laz U1 devs! [04:36] *lazy [04:37] * thumper is getting the munchies [04:37] still, i need to go; I'll finish it off later [04:38] Mmmmmm, smells like rain here [04:39] wgrant: pushing to lp:~lifeless/launchpad/bug-717394 if you are interested. Its a little ugly, but EMENOCARE [04:40] pushed [04:41] wgrant: I thought you requested a deploy? [04:41] lifeless: It's on LPS [04:42] lifeless: henninge added one to the deployed section of LPS, but didn't poke anyone about it. I extended it to the latest rev and moved it to the right place. [04:42] Just now. [04:42] wgrant: cool; probably want to poke someone [04:42] I did. [04:42] lifeless: He did? [04:43] spm: Are you sufficiently poked? [04:43] Haha [04:44] "Are you not entertained!?" [05:04] thumper, so lifeless's point was to just keep using non-empty strings as boolean true? [05:05] poolie: yes... although I agree with you [05:05] * thumper EODs [05:05] good night [05:05] poolie: we can talk about this more tomorrow if you like [05:07] Can haz review from someone? [05:09] lifeless: You know, I think identi.ca doesn't know your new suburb [05:22] wow [05:23] wgrant, the lag seemed to be suddenly going up to several seconds [05:23] Yeah. [05:23] I guess Mumble isn't so good when you're on the same side of the world. [05:23] adequate consultation is good [05:23] routing via london's not optimal [05:23] Not quite. [05:24] i guess we possibly could have an australasian server [05:24] i wonder how hard it is [05:24] Huh. [05:25] What happened to the loggerhead OOPSes/ [05:25] ? [05:25] There were <1000 yesterday. [05:26] lib/lp/contrib/ [05:26] Oxymoron! [05:26] wgrant: Naming is hard [05:27] StevenK: We already have lib/contrib :( [05:27] Yes, and it needs to die too [05:27] Now I have two directories to destroy :( [05:28] Awesome, lib/contrib/oauth.py has no license [05:29] wgrant: Seems like spm is not sufficently poked. [05:30] Insufficiently present, would be my guess. [05:30] poolie: Is there something wrong with the oauth 1.0.1 egg? [05:30] Or did you just try to use a package because it is less evil? [05:30] wallyworld, hi [05:30] poolie: yellow [05:31] the thing with tim's patch is when he says 'on/off' he doesn't actually mean the strings 'on' or 'off' [05:31] how silly of you :) [05:31] we can either fix it so you can actually use those values [05:31] which is the bug i mentioned [05:31] i thought that's what gerFeatureFlag() returns? [05:31] or we can update his docs [05:31] you're right [05:31] it just returns the string at the moment [05:32] so unless it's handled at a lower level he needs to check == 'on' [05:32] so his condiitonal will fail [05:32] as written [05:32] well, it won't work in the way you'd expect [05:32] how do you say that in TAL? [05:32] i think it's a bit longwinded [05:32] wgrant, i believe the egg still has the bug in it [05:32] ubuntu's package is 1.0+a bit [05:33] poolie: We have the 1.0 egg. [05:33] poolie: so, won't "if not getFeatureFlag(RECIPE_ENABLED_FLAG)" give a crap result if the ff is set to "off" ???? [05:33] There is a 1.0.1 egg aka 1.0a, which is probably what Ubuntu has. [05:34] wgrant: Can you see the point of lib/canonical/lazr/security.py at all? [05:36] StevenK: Probably to support callsites that were added before stuff was moved to lazr.restful. [05:36] I can only see one left, though, in a test. [05:36] Right [05:36] I'm tempted to just fix it [05:36] Delete delete delete. [05:36] You have my full support. [05:36] For once. :-P [05:36] You have my full support to delete anything in lib/canonical, as long as it doesn't break tests :P [05:37] I'll delete lib/canonical/database, then. [05:37] That won't break tests, that will break EVERYTHING. [05:37] Indeed. [05:38] * wallyworld goes to take kid to soccer [05:40] wgrant: So, fix the test to import from lazr.restful or just bin it completly? [05:42] StevenK: LP doesn't use protect_schema directly. [05:42]