[00:00] any reason we can't run it right after ? [00:01] It'd run into the 0902 run, so that wouldn't happen. [00:03] wgrant: can has review? [00:03] lifeless: Sorry, missed that. Will look in a sec. [00:03] Just fixing test failures from yesterday's largish branch series. [00:23] lifeless: Could you do the sort of query that jtv asked for https://code.launchpad.net/~jtv/launchpad/bug-613823/+merge/75773? [00:23] lifeless: The approver is cronned on staging. [00:23] wgrant: hmm ? [00:24] lifeless: There are QA instructions in that MP. [00:24] I don't have staging DB access. [00:24] sorry to be a nuisance, but can you translate to sql ? [00:25] SELECT * FROM translationimportqueueentry WHERE error_output IS NOT NULL AND status = 5 ORDER BY date_status_changed DESC LIMIT 10; [00:25] 0 [00:26] That's upsetting. I guess I will check it with hloeung later. [00:29] lifeless: Does oops-timeline really only lose us two lines of code? [00:30] wgrant: yes, but it saves duplication for other folk glueing them together [00:30] Two lines of duplication? [00:31] about 10 [00:31] stsly? [00:31] duplication has never been about size of code though [00:31] srsly? [00:35] lifeless: Do our consumers cope with the s/db_statements/timeline/? [00:36] https://code.launchpad.net/~mwhudson/launchpad/feature-flag-xmlrpc/+merge/75673 if anyone is bored... [01:09] wgrant: oops-tools is a version locked buildout project. [01:10] wgrant: it uses datedir-repo which this change doesn't affect, because its not a labelled key in the rfc822 format [01:11] wgrant: that said, I have a branch for it which updates-and-corrects [01:12] wgrant: that also needs a review :) - code.launchpad.net/~lifeless/oops-tools/timeline/+merge/75941 [01:15] mwhudson: ping [01:16] lifeless: hello [01:17] want to talk (you know, by airwaves) about branch access? [01:18] ah sure [01:18] lifeless: mumble/skype/pots [01:18] ? [01:18] the second door [01:18] mwhudson, just a random idea [01:19] is it cleaner to pass a dict that can contain a username and possibly other stuff? [01:19] i don't know if that will be easier to evolve or not [01:19] poolie: what would the values in the dictionary be? [01:19] objects! [01:19] username='mbp' client_ip='1.2.3.4' [01:19] etc [01:19] ah [01:19] bzr_version='2.4.4' [01:19] lifeless: haha, this is xmlrpc [01:23] #quotes [01:27] lifeless: Ahh, so it doesn't actually make a difference in the serialised format? [01:34] right [01:59] wgrant: were you going to review the oops-tools one too ? [02:01] lifeless: Done. [02:05] Oh look, pocketlint deals with JS. [02:05] If only I knew that on Friday. [02:05] heh [02:06] What happened on Friday? [02:06] Also, I thought pocketlint dealing with JS was one of its biggest advantages. [02:07] I was dealing with the same JS branch, but I had broken the JS badly so Firefox didn't love me and refused to render the page [02:07] Ouch. [02:07] Didn't firebug help? [02:07] No, it just told me I had a syntax error on line 68,000 something [02:08] Which wasn't very handy [02:08] Yeah. [02:08] I like how firebug throws a weird bug when you make a mistake in JSON. [02:08] I spent about 20 minutes for that, until I discovered jsonlint helps. [02:09] StevenK: But if you click on the error it should take you to that line, letting you debug it. [02:09] Not like that's easy either. [02:09] Anyway, pocketlint has solved my issue. [02:10] :) [02:10] I hope noone stabs me for the number of revisions I've made undeployable. [02:11] You've only made 13 undeployable, which is probably the least I've seen from non-APAC breakage. [02:11] I get stabby when it gets above 50. [02:11] non-APAC? [02:12] APAC is Asia-Pacific [02:12] US/EU breakage often gets nasty. [02:12] "technically" I live in APAC [02:12] Well, yeah. [02:12] Just! [02:12] easily [02:12] asia goes -way- west [02:12] Shh. [02:13] Middle east is west of me. [02:13] So, i'm not "just" Asia :P [02:14] isn't it to the bosphorus? [02:14] Yeah [02:14] beautiful river [02:15] lifeless: At some point I need to pick your brain further re:js-oopsd [02:15] But at a later point when I'm not sleep deprived! [02:18] reprocess-hwdb-submissions is doing well for itself. [02:18] 90272 None: None [02:19] nigelb: ok [02:22] lifeless: can you tell me the code that gets invoked when serving loggerhead pages like to http://bazaar.launchpad.net/~johndoe/project/files [02:22] wallyworld___: You're trying to add the privacy overlay thing? [02:22] yes [02:22] Run away. [02:23] We have no loggerhead customisation story these days :/ [02:23] * wallyworld___ starts running === wallyworld___ is now known as wallyworld [02:23] Which means we have to copy the LP-specific template and JS into the loggerhead tree. And have it there polluting everyone's installations. [02:23] Even though it's only relevant to us. [02:23] And our theme. [02:23] yuck [02:23] Can't fork it? === wallyworld is now known as Guest73189 [02:24] Forking it is not a good option. [02:24] Like have a branch with launchpad changes, and keep merging trunk into it. [02:24] We should have an alternate theme. [02:24] Well. [02:24] Really we should use LH as a service. [02:24] erm [02:24] And import it into the LP UI. [02:24] right [02:24] if request is a ILaunchpadRequest, what is request.principal? [02:24] But that may happen in three decades. [02:24] the ideal thing is to make LH a background service. [02:24] Until then we need something better. [02:24] mwhudson: *probably* an IAccount. [02:24] in the mean time, we a) maintain loggerhead, b) deploy trunk. [02:24] mwhudson: Or it may have an adapter to IAccount. [02:25] AH [02:25] Guest73189: your nick is bust :) [02:25] yeah :-( [02:25] keeps happening [02:25] So logggerhead won't actually show files form a private branch? [02:25] nigelb: It does. [02:25] nigelb: It just doesn't say that it's private. [02:25] AHH. [02:26] wgrant: ah, it's a ILaunchpadPrincipal [02:26] Stupid question - Isn't loggerhead already a service? [02:26] (well, if it's not unauthenticated) [02:26] mwhudson: Does that handle stuff like IWeaklyAuthenticatedPrincipal? [02:27] (which should really be INotActuallyAuthenticatedAtAllPrincipalButSomePeopleApparentlyCannotHandleEmailSigningBecauseTheyLikeToUseGmail, but that's a bit long) [02:27] heh [02:28] wgrant: basically [02:28] i have a request [02:28] i want a person, or None [02:28] I remember doing this elsewhere. [02:28] nigelb: it is but it displays a web UI of its own [02:29] nigelb: using different templates, middleware, and web stack. [02:29] nigelb: its not a *backend* service that the LP web app can consume [02:29] AHH. [02:29] nigelb: its just a sibling website that happens to be themed the same way [02:29] That means loggerhead should first expose such a service in trunk and then we need to grab that and use that? [02:29] it has a bit of such a service [02:29] it needs more [02:30] and it needs to be -much- faster [02:30] Agree on that one. [02:30] its fast-path is tolerable but its slow path needs to be two orders of magnitude faster [02:34] interesting https://www.facebook.com/notes/facebook-engineering/making-facebook-self-healing/10150275248698920 [02:34] (fwiw, i think IPerson(request.principal, None) is what i want) [02:38] wallyworld__: Can I force an overlay to the top? [02:38] StevenK: you mean so that it always stays on top? [02:39] wallyworld__: Give me a tick, preparing screenshots [02:39] ok [02:41] wallyworld__: http://people.canonical.com/~stevenk/Subscribe-popup.png [02:41] * wallyworld__ looks [02:42] wallyworld__: That's the "before" shot -- it's a private bug and I've clicked the "subscribed to all notifications" link for the popup [02:42] and the confirm overlay appears beneath? [02:42] wallyworld__: http://people.canonical.com/~stevenk/Subscribe-popunder.png [02:42] Needs z-index magic :-) [02:42] StevenK: overlays have a zIndex propety [02:42] * wgrant invokes mpt. [02:42] StevenK: default is 0 [02:43] Oh no, I've angered wgrant. [02:43] Nested overlays... this doesn't seem right. [02:43] StevenK: http://yuilibrary.com/yui/docs/overlay/ [02:43] wallyworld__: Higher is on top, or lower is? [02:44] StevenK: ummmm. higher i *think* [02:45] Higher. [02:45] * nigelb helpfully links to http://www.w3schools.com/cssref/playit.asp?filename=playcss_z-index&preval=2 [02:45] StevenK: sort of agree with wgrant though. i'd like to see the click in the "remove" link result in the confirmation exapnding out below the link via an animation [02:45] Right. [02:45] sort of like for the stuff i did for confirming bug assignees [02:45] The dialog mutate by expanding, or using a second step. [02:45] s/mutate/should mutate/ [02:46] You two suck, or something. [02:46] Dialogs are bad the best of times. [02:46] you've been peeking in my window again haven't you [02:46] Actually, I'll agree with them as well. [02:46] * wallyworld__ needs to close the curtains [02:46] s/two/three/ [02:46] So, StevenK wasted 2 days of JS hacking only to be mpt'd :P [02:47] Not really [02:47] Since there are two callsites [02:47] I always expect a few days of hacking on a new technology to have to be thrown out at least a couple of times. [02:48] And while feature squads seem to like introducing tech debt and terrible UI, I think we should try to do this at least vaguely properly. [02:48] what's tech debt? I see that used in tags. [02:49] Launchpad. [02:49] Heh. [02:49] Launchpad is mostly tech debt. [02:49] http://c2.com/cgi/wiki?TechnicalDebt [02:50] wgrant: I've seen worse. [02:50] lifeless: Huh? [02:50] Where? [02:50] COBOL doesn't count. [02:50] Ah. [02:50] wgrant: minestar [02:50] wgrant: Micropay [02:50] OK, software that isn't meant to be enterprisey? [02:50] wgrant: .... unlike LP ? [02:51] wgrant: win32 api :-) [02:51] Isn't LP the enterprisey codehosting website? ;-) [02:51] LP isn't meant to be enterprisey, I don't think. It is, but I don't think it was meant to be. [02:51] wgrant: and yeah, a few things do come to mind that rival LP for 'wants refactoring soooo bad' [02:51] It may not be meant to be enterprisey, but some of the best features are enterprisey. [02:51] (lp) [02:51] wgrant: see rule #1: all software sucks. [02:52] lifeless: We just have no way to make it not suck. [02:52] Because nobody wants to reduce tech debt. [02:52] Delete more code. That's code that's guaranteed to not suck :-) [02:52] Maintenance squads can't fix it, and feature squads won't. [02:52] => we are fucked [02:52] :) [02:52] Everything will be better when we have no criticals. [02:52] OH WAIT [02:52] Hahahahahahahhaa [02:53] Bwahaha [02:53] I could not help myself ... [02:53] Speaking of which,. [02:53] 390 bugs with tech-debt :/ [02:53] nigelb: And that's mostly just the really bad stuff. === wgrant changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: -| Critical bugs: 260 - 0:[########*** stack smashing detected ***: ./lp terminated [02:54] Bleh, still 260. [02:54] We are going *backward* [02:54] wgrant: Some seem to be not very had. [02:54] *hard [02:54] nigelb: Some aren't, no. [02:54] nigelb: And even the big stuff is not hard. [02:54] I spent a few hours yesterday and very nearly disposed of Zopeless. [02:54] wgrant: How did that turn out? [02:55] StevenK: Some test failures in the last branch in the series. [02:55] Which removes ZTM's commit/abort/begin methods, and initZopeless' return value. [02:55] Leaving ZTM to be moved into DBConfig in the next branch. [02:55] Still have to strip out Zopeless mail somehow. [02:56] But it's all pretty well detangled now. [02:56] great [02:56] initZopeless' callsites are limited to core base classes and buildd-manager. [02:56] Which means we can finally untangle the damned dbuser handling. [02:56] Which is roughly quadruplicated at the moment. [02:57] poolie: is there a reason that there are scopes team:foo but server.edge? [02:57] And it also means the Zopeless layers can die, which simplifies things a bit. [02:57] i.e. ':' vs '.' [03:01] mwhudson, i don't think there's any really good reason [03:01] you could look at it as 'server_edge=True' vs 'server=edge [03:03] o/ 90K hwdb exceptions [03:04] Yup, the reprocessing seems to be going well... [03:04] 181 AttributeError: 'LaunchpadTimeoutError' object has no attribute '__traceback__' looks suspect [03:04] lifeless: I pointed that out last week, yeah. The attempted reraising is bad. [03:04] __traceback__ was added in 2.7, wasn't it? [03:04] wgrant: did you file a bug ? [03:04] Maybe even 3.x. [03:04] No. [03:04] wgrant: or reopen it ? [03:04] I had three production breakages. [03:04] I'm not criticising [03:05] I'm asking so I don't duplicate work. [03:05] You should have been :P [03:05] I haven't, no. [03:05] Well, really I wanted to see if anyone else would pick it up. [03:05] IIRC benji worked on it ? [03:05] I think that's right. [03:05] lifeless: So, for science, I'd like to leave it a couple of days and see if a maintenance squad notices. [03:06] well [03:06] I've mailed benji [03:07] :( [03:07] SUCCESS! [03:07] My confirmationoverlay works for +subscriptions [03:08] wallyworld__: *prod* [03:08] StevenK: ouch!~ [03:08] StevenK: excellent [03:09] StevenK: you are now a yui expert :-) [03:09] Trapped! :P [03:09] and i'll assign all the js bugs to you! [03:09] wallyworld__: Suggestions about how to break up the form_content in http://pastebin.ubuntu.com/692723/ ? [03:09] * wallyworld__ looks [03:09] wgrant: Deploying today? [03:10] StevenK: you can use a join [03:10] give me a sec and i'll find an example [03:10] nigelb: Funny you should ask. [03:10] 13:09:58 < wgrant> hloeung: We need to do a nodowntime rollout today so we can turn gina on. [03:10] nigelb: So, you weren't quite perfectly timed, but 'tis acceptable. [03:10] wgrant: Almost immediately? :-) [03:11] what channel was that? Canonical IRC? [03:11] Maybe. Some preparation to do, and we are down to one LOSA this week. [03:11] nigelb: Yes, a private one. [03:11] Yeah, that was #launchpad-ops on the Canonical network. [03:11] mwhudson: don't you hate all the boilerplate glue to say 'this code at that endpoint' ? [03:11] lifeless: YES [03:11] StevenK: ah, I have an example of join somewhere I think. [03:11] i used a bit less boilerplate than some though [03:11] StevenK: search for join('') or join("") in js [03:11] there's quite a few places [03:12] some endpoints have the xml-rpc methods in a view on the application endpoint object [03:12] StevenK: you can use string concat if it's only 2-3 lines. [03:12] i just put the methods on the application endpoint object [03:12] StevenK: ['foo', 'bar'].join('') [03:12] but any longer, a join is recommended [03:12] where foo and bar are 50 to 70 chars each. [03:13] wgrant: I'm technically a jQuery person more, but YUI is quite neat :-) [03:13] err [03:13] wallyworld__: ^ [03:14] nigelb: ? [03:14] wallyworld__: I played around a lot with it for that bug title fix. Its different, but now I think in a better way. [03:16] nigelb: oh, you mean you like yui. it's not too bad actually. i think it's 2nd most popular framework behind jQuery [03:16] wallyworld__, nigelb: Fixed, thanks. [03:16] StevenK: \m/ [03:17] I think I need to revert bug_subscription_portlet changes while I work out how to expand it [03:17] Perhaps I'll do the changes in two branches [03:18] mwhudson: I've done a review [03:19] mwhudson: its probably a bit opaque. Two key things - please don't duplicate code (team: magic string handling), and a bit fuzzier one that you may want to ping me on [03:19] lifeless: you have? i don't see a review [03:19] mwhudson: process-incoming.py [03:19] ah [03:19] lifeless: did you see my response to poolie's comments? [03:20] wallyworld__: I'll grab you after lunch on how to test this thing? [03:20] StevenK: ok [03:20] YUI tests! Fun! :-) [03:20] wallyworld__: No promises on where I'll grab you [03:20] yes, I think dict for scopes was confusing (because scopes are containers), but coming from the same place I am [03:20] :-P [03:20] oh, one can only hope [03:21] mwhudson: docs are in docstrings. [03:22] lifeless: were you under the impression that i know what you're talking about? [03:23] mwhudson: yes! [03:23] mwhudson: poolie pointed you at docs, you found an empty file. [03:23] well, i still don't know what needs to be _updated_ [03:23] mwhudson: I believe they are in module docstrings instead [03:23] all the code i added has docstrings [03:23] yes, but nothing points at your code; thats the point I think. [03:24] ah oik [03:24] __init__.py [03:27] lifeless: were you reviewing the originally proposed version, or r13974 ? [03:27] mwhudson: the mailed out version [03:27] lifeless: ok, look again please [03:29] much better [03:45] mwhudson: back to you :) [03:46] lifeless: lol, guess what my working copy looks like [03:47] default_scopes = (DefaultScope(),) [04:00] mwhudson: :) [04:02] lifeless: back to you (hopefully not much more of this) [04:04] mwhudson: your example is missing a : [04:04] 'codehosting.use_forking_server', ['user:' + user_name]) [04:04] -> [04:04] 'codehosting.use_forking_server', ['user:' + user_name]): [04:05] ah [04:05] ok [04:07] lifeless: thanks [04:08] i wonder if i can still land code ... [04:08] should be able to [04:08] have I put you in the emeritus team yet ? [04:08] don't think so [04:09] want me to? [04:09] lifeless: are you aware of a bug report about this? [04:09] about what? [04:09] oh [04:09] uhm [04:09] there may be one [04:09] lifeless: what impact does being the emeritus team have? [04:09] ok, i'll search [04:09] you'll get review mail which we don't [yet] have a good offswitch for [04:10] you get bugcontrol privs, and push privs to ~canonical-launchpad-branches [04:10] mwhudson: ah you are in it [04:10] oh ok [04:10] https://launchpad.net/~canonical-launchpad-emeritus [04:10] i don't get review mail... [04:10] No PQM access? [04:11] (which i am happy with, before you do anything about that :p) [04:11] mwhudson: its not fully setup up :) [04:11] mwhudson: the intent is to get you review privs, which currently implies mail. [04:11] mwhudson: [sorry] [04:11] lifeless: i appear to be able to approve merge proposals [04:12] i am still in ~launchpad mind [04:12] mwhudson: ah, well then :P [04:12] mwhudson: its moot till you leave that [04:12] ok [04:14] StevenK: yes, if pqm is integrated with LP (which it was till the code decided their old automation design was wrong) [04:15] lifeless: Right, so the intent is that canonical-launchpad-emeritus members continue to have PQM? [04:16] StevenK: the intent is that emeritus devs lose nothing if they are still @ canonical, and lose write-to-deployed-branches only, if they are not. [04:16] e.g. be as close to an open source project as we can with our current constraints [04:19] wallyworld__: *grab* [04:19] wallyworld__: Let me just get some tea [04:19] StevenK: oooh. a little hiher [04:19] * StevenK goes to scrub his hands with solvol and steel wool [04:28] StevenK: so you got your tea? [04:29] wallyworld__: Indeed. Ready when you are. [04:30] StevenK: ok, so you want to write a yui test for your confirmation overlay functionality? [04:30] wallyworld__: Yup. I'm on mumble if that makes it simpler. [04:30] StevenK: ok. i'll just grab my headset. one of the kids has nicked it [04:31] Hah [04:42] Ah, finally. [04:42] ~jans has added their OpenPGP key. [04:43] They've uploaded 18 packages this morning. [04:43] Over several hours. [04:43] Causes lots of process-upload cronspam :( [04:43] Hah [04:43] But maybe it will stop now. [04:50] wgrant: Don't you have filters for that sort of thing? [04:52] nigelb: That's one of the few user errors that still generates emails. [04:53] nigelb: I tried to fix it one day, but then found terrible security holes and never visited that code again. [04:54] Ah, I see that sinzui had a productive anti-tech-debt Sunday as well. [04:58] :) [04:58] wgrant: Oh? [04:58] StevenK: You might recall I was talking about pagetitles last week on the call. [04:58] StevenK: sinzui deleted much of it. [04:59] Neat [05:05] wgrant: can you talk to sinzui about blueprints next? [05:05] Heh [05:05] Haha [05:05] Just delete the folder? [05:05] I can do that. :P [05:05] mwhudson: You want blueprints deleted? [05:05] Damn, ZTM will probably fall tomorrow. Then I'll need to find another target for my wrath :( [05:06] wgrant: lib/canonical [05:06]