[00:03] Project devel build #1,014: STILL FAILING in 3 hr 58 min: https://lpci.wedontsleep.org/job/devel/1014/ [01:35] jelmer: maybe we don't need new code import mails now? [01:37] Heh. [01:39] not that i really mind, it's not like they take long to dispatch [02:26] mwhudson, they're probably going to give me Carpal Tunnel Syndrome though from all the clicking needed to delete them :P [02:36] Good morning! [02:36] Evening nigelb. [02:36] cody-somerville: heh, automate with selenium! [02:36] wgrant: I'm taking a shot at TDD today. Writing tests first. I hope that doesn't mean more headdesking :) [02:37] nigelb: It gets better :) [02:37] nigelb: Particularly when they aren't doctests. [02:38] Yeah, I'm wondering why StevenK's deletion spree doesn't delete doctests yet :D [02:39] He spent a week deleting lots of doctests a while ago. [02:40] So there is hope. [02:40] btw, lifeless is proud daddy yet? :) [02:42] How was about last night, so I suspect not, but have not heard either way. [02:43] Ah, ok. === jtv changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: jtv | Critical bugs: 244 - 0:[#######=]:256 [02:53] And I've picked up bug 684548 [02:53] <_mup_> Bug #684548: Display Edit Patch Flag More Prominently

< https://launchpad.net/bugs/684548 > [02:53] Morning jtv! I shall be troubling you in a bit. [02:53] * jtv rushes to change the topic :) [02:54] heh === pjdc_ is now known as pjdc [02:58] Should. Stop. Looking. At. Ivanka's. Pictures. [04:01] Project devel build #1,015: STILL FAILING in 3 hr 58 min: https://lpci.wedontsleep.org/job/devel/1015/ [04:18] Text conflict in test_notification between stable and db-devel. Orsum. [04:40] good morning [04:44] wgrant: could you help me with qa for r13826? It's too early in the morning for me to even understand the related bug... problem is: I am a bit desperate to deploy a revision >= 13822 [04:57] adeuring: It's been broken for nearly 18 months; is it really urgent? [04:57] wgrant: yes, unfortunately [04:57] Intriguing. [04:57] Looking. [04:58] Ah, need to resolve the conflict first, since DF runs db-devel... [05:07] jtv: Are you still using dogfood? [05:08] wgrant: no, I use a piece of cloth now. It doesn't absorb as well but it muffles more effectively. [05:09] Hah. [05:10] So no. [05:10] Meanwhile, I see we have a wonderful merge conflict. [05:10] I'm pushing a branch for that. [05:11] Cool [05:11] But db-devel hasn't been merged in a while, so my 256kbps upstream is being painful... [05:19] adeuring: Hmm, this revision may be bad. We'll see shortly. [05:19] ok... [05:22] Well, maybe not so shortly, since mawson is mawson. [05:24] jtv: http://librarian.dogfood.launchpad.net/71107310/bRhzSbgpU9MP0WU1BbgvqBNgitQ.txt is worrying. [05:24] jtv: The DSDJ runner can't check packageset membership... [05:25] I wouldn't think that would be a new thing. [05:25] That sounds familiar somehow. [05:26] I would suspect that a cowboyed privilege dropped out of the db without the permanent privilege making it in. [05:26] Either because security.py wasn't run, or it failed, or it did not contain that required permission. [05:27] There is no permanent privilege. [05:27] I just reran security.py and checked security.conf :( [05:27] granted select on flatpackagesetinclusion and packagesetsources, and it works. [05:29] And packagesetgroup, it seems. [05:41] Does anybody know how to use Thunderbird's account creation wizard? [05:41] They've removed manual setup, and I can't work out how to use an account on a provider that they don't know about. [05:45] Spend 5 minutes why launchpad.dev wasn't working. Sigh. [05:45] *debugging [05:46] Was it running? [05:46] Yeah, https [05:47] I can do a create_initilized_view for all the portlets too? [05:47] In general that should work. [05:47] tdd is weird, at last at first. [05:47] For some it won't. [05:47] *least [05:53] adeuring: I think that rev is good. [05:53] Which means we can deploy. [05:53] wgrant: great, thanks for your checks! [05:53] Sorry it took so long. High startup time and lots of checks to do. [06:06] so, I create a bug. [06:06] I add an attachment [06:06] now, how do I get the comment of that attachment? [06:08] (I feel more lost writing tests first) [06:13] nigelb: BugAttachment.message [06:13] ah [06:13] need to look at models deeper. [06:14] nigelb: I often find it's clearer to check the DB structure. [06:14] \d bugattachment [06:15] I'm not well versed with postgres enough.' [06:15] yet. [06:15] psql launchpad_dev [06:15] \d bugattachment [06:15] Done :) [06:15] wow [06:16] I did not know about psql. [06:16] This, is neat. [06:28] nigelb: I told you that exact same trick! === jtv is now known as jtv-eat [06:31] jtv-eat: well, I didn't know how to use it correctly :) [06:35] psql is love [06:37] hehe, so with tests first, I can head desk first [06:37] and then when I fix it, there's more joy [06:37] Gah, I can't do this apparently. [06:38] html = create_initialized_view( self.bugattachment.message, name='+index', )() [06:40] nigelb: Why not? [06:40] nigelb: It's odd syntax, but it should work. [06:40] Oh, hmm. [06:40] There's no +index for Message itself. [06:41] ah. [06:41] You'll need to get the BugMessage. [06:41] (Message just represents an email in LP. May be on a bug, merge proposal, answer, mailing list...) [06:42] oh. [06:46] ForbiddenAttribute: ('BugMessage', ) [06:48] getUtility(IBugMessageSet).getByBugAndMessage(some_bug, some_message) [07:09] hehe [07:09] awesome traceback [07:09] Oh? [07:09] http://dpaste.com/605694/ [07:17] wgrant: Thoughts? :) [07:17] nigelb: Hm, that's pretty nice. [07:18] Haha [07:18] nigelb: It looks like you're calling it with a bugattachment, not bugattachment.message. [07:18] aaah. [07:22] Sigh. [07:22] http://dpaste.com/605703/ [07:24] Why do I end up wasting more time than finding a fix in tests :| [07:25] Ah, it looks like you actually want BugComment:+index, not BugMessage:+index. And BugComment creation is not for the faint of heart :/ [07:25] You're possibly better just rendering BugTask:+index. [07:26] Well, the problem is the edit shows up in a portlet as wwell. [07:26] I don't watch to get that. [07:27] OR, I'll have to count the number of appearances. [07:27] Or add a class to the find and use soupmacheers [07:27] Or you can use BeautifulSoup/soupmatcher to look only in the comment list. [07:27] There should be an existing class/ID you can use. [07:27] Ah. [07:29] Ah, we deployed. [07:29] * nigelb checks his fix. [07:29] We did. [07:29] It looks good. [07:29] \o/ [07:57] Project devel build #1,016: STILL FAILING in 3 hr 55 min: https://lpci.wedontsleep.org/job/devel/1016/ [08:06] I have to agree with bigjools, soupmatches is quite rad. [08:11] Hello [08:13] Hello mrevell! Good morning :) [08:14] Hey nigelb! [08:15] nigelb: I'm sure you will come to love soupmatchers ;) [08:15] rvba: Heh, already loving it! [08:15] nigelb: ;) === almaisan-away is now known as al-maisan [08:16] heh, I summoned bigjools? :P [08:16] I was rubbed [08:17] xkcd ruined that reference for me. [08:19] Anyway, I was mentioning how I like soupmatchers :) [09:28] jtv: Could you please have a look at this (tiny) MP when you get a chance: https://code.launchpad.net/~rvb/launchpad/masscreate-bug-835040/+merge/73495 ? [09:54] wgrant: GAH. [09:54] I'm doing something wrong, I'm getting hideous tracebacks. [09:55] wgrant: while you're being distracted, can I interest you in reviewing my DB privilege fix? https://code.launchpad.net/~jtv/launchpad/bug-837893/+merge/73502 === jtv changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: - | Critical bugs: 244 - 0:[#######=]:256 [09:55] My nightmares with tests continue. Sigh. [09:56] Or I could ask danilos, who is OCR'ing today. [09:56] jtv, oh, he is? :) [09:56] Apparently. [09:56] At least on Gregorian Wednesday. === danilos changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: danilos | Critical bugs: 244 - 0:[#######=]:256 [09:56] :-) [09:57] jtv: You're marked under Europe? HA [09:58] I'm sort of inbetween. [09:58] jtv, r=me [09:58] thanks danilos! [09:58] jtv, which one of us? [09:58] wgrant: it's done—you go help nigel or something. :) [09:58] for danilo in danilos: danilo.thank() [09:58] heh [09:59] if I create a bug with factory, I can call its +index right? [09:59] * nigelb checks zcml [10:00] oh, I need to call the +index of the bugtask? [10:00] nigelb: Bug has no +index, right. [10:00] It redirects to default_bugtask's +index. [10:01] so, create_initialized_view(self.bug.default_bugtask, name='+index', )() [10:01] I got a scary traceback for that one. [10:02] http://dpaste.com/605821/ [10:02] I wonder if this has something do with layer = LaunchpadFunctionalLayer [10:03] I think I've mostly used DatabaseFunctionalLayer before. [10:04] danilos: you'll notice I left one review on the queue. It looks like nobody's worked on it but it's got other reviewers' names on it. [10:04] jtv: thanks for the review mon cher. [10:04] rvba: ton /quoi/? [10:04] Très cher jtv ;) [10:04] le horreur [10:04] I thought I was doing this work at quite a reasonable rate. [10:05] jtv is actually cheap [10:05] Oh now that's rich. [10:05] And is it le horreur or l'horreur? [10:05] hehe [10:05] "l'horreur" indeed. [10:05] FTR this "cher" was "dear". [10:06] Funnily enough, "dear" can have both meanings in English. [10:06] I was enunciating all the syllables [10:06] (although "expensive" is somewhat archaic) [10:06] bigjools: I see. Maybe you should enunciate them twice in succession: le horreur, le horreur [10:07] Wasn't until that bit in the book that I realized that the film was based on it. [10:09] “le horreur, sans blague”? [10:10] nigelb: I /think/ the problem you have is due to the fact that you don't pass any 'principal' to create_initialized_view. [10:11] rvba: principal being a person? [10:11] nigelb: yes. [10:11] rvba, nigelb: It defaults to anonymous. [10:11] This is possibly another view that doesn't like being instantiated directly :/ [10:11] It's a fairly horrific view. [10:12] I tend to pick the fun ones, yeah. [10:12] I'm pretty sure I had this exact same problem. [10:12] Someday I'll learn how to do this right :| [10:12] Ha, solution? [10:13] nigelb: please try principal=factory.makePerson() [10:15] mmm [10:15] NoCanonicalUrl: No url for None because None broke the chain. [10:19] nigelb: Looks like something is trying to redirect to None. [10:20] rvba: Oh. [10:20] That's wweird :| [10:21] I should have started with fixing the bug isntead of tests [10:21] jtv, on line 84 of the diff on https://code.launchpad.net/~jtv/launchpad/bug-834388/+merge/73483, "removables" is not a ResultSet, right? [10:21] I've spend 8 hours and I've nothing to show for it, except a broken test :) [10:22] danilos: no, it's a set that's first painstakingly constructed by hand. :/ [10:22] Sorry, list! [10:22] Not set. [10:23] jtv, ok, fair enough (btw, I am surprised you didn't know about resultset.set() :P) [10:23] Well, now I do! [10:23] nigelb: It's a good habit to start with the tests. And its absolutely normal to have trouble to write tests at first. [10:23] I have to leave very very soon btw [10:24] jtv, sure, I am almost done with this second review [10:24] Great, thanks [10:24] jtv, just one more non-critical question, why the changes to logger statements? [10:24] They annoyed me. [10:24] Why have special provision for debug [10:24] but none for the other log levels? [10:25] Someone's going to do a logger.info or a self.warn and find it doesn't exist. [10:25] Or do you mean the string interpolation? [10:25] Apparently the "proper" way to do it is logger.info("format %s string", variable) [10:25] Not logger.info("format %s string" % variable) [10:26] Hrm, I wonder what I do next. [10:26] I can't get the view to get a test. My test is supposed to just check the existence of a URL. [10:26] btw guys, what is the best way to find out what handles/generates a particular page? [10:27] configure.zcml [10:27] it will be inside browser folder [10:29] jtv, I meant both, the fact that logger.info considers the first string a format string doesn't make it proper in my book, but I don't care either way :) [10:29] nigelb: ah ha, thanks! :) [10:29] G: :) [10:29] jtv, it'd make sense if it was something like structured() or sqlvalues() where stuff needs to be interpolated some how [10:29] s/ // [10:29] danilos: I never bothered to ask really, it just allows for slightly better formatting. :) [10:30] jtv, heh, sure, formatting is a good argument [10:30] jtv, anyway, r=me already, but I assume you know what will this do for performance :) [10:30] thanks! [10:30] It'll improve the normal case, make the weird case worse. [10:31] And it shows future engineers how to fix the weird case. === jtv is now known as jtv-afk [10:34] wgrant / rvba: Anything more I could try? [10:34] nigelb: please paste your diff. [10:36] http://dpaste.com/605832/ [10:39] hmmm took a look at bug 61428 and I think I've identified a 3 line fix [10:39] <_mup_> Bug #61428: Want a "subscribed to teams" portlet...

< https://launchpad.net/bugs/61428 > [10:40] The 'proper' way of interpolation with the logger API saves interpolation time, but has a habit of hiding interpolation bugs as we end up with untested code paths. [10:42] ahhh actually, not that simple [10:49] danilos: if you've got a moment, question about the above mentioned bug, the portlet while nearly identical for a subscribed bugs page, is nearly the same as the current assigned bugs (person-portlet-team-assignedbugs.pt) is the attitude to create a seperate portlet, or try and introduce logic to the current portlet [10:50] gmb: Bug comment lazy loading - what will search engines see if we turned this on for everybody? [10:51] stub: Uhm... Well, at the moment it's just JS on top of what's currently there. But fair point; if we did it automatically then it would be not good for our Googlejuice. [10:51] G: I am looking at it [10:52] gmb: Or using text-only browsers :) [10:52] nigelb: also true. [10:52] gmb: Yer. We might need to be more intelligent. Default batched comments with a sane batchsize, and js magic to improve on that if you have js. [10:52] Thankfully, we're talking about something I haven't done yet :). [10:52] heh [10:52] stub: Agreed. It wouldn't be too hard to do that, actually. [10:53] danilos: a second portlet, would basically be :%s/assigned/subscribed/g w/ a little bit of extra CSS for the new div class id's [10:53] For insane people who want to scroll through the entirety of Bug #1 [10:53] <_mup_> Bug #1: Microsoft has a majority market share (at least my intpretation) [10:53] argh! [10:54] But there are probably Ubuntu bugs worse than that. [10:54] rvba: Were able to spare a minute to look into that diff? [10:54] G: right, I think you are better off reusing it, and just introducing a few variables like CSS class and target view to decide what to do in the template [10:54] G: though, that entire portlet looks like a very bad UI, but if you find it useful, it's better than no UI :) [10:55] nigelb: Yes, nothing seems bad with your diff but this "NoCanonicalUrl: No url for None because None broke the chain." is a tricky error. [10:55] nigelb: hang on. [10:55] okay, so I'm not going completely crazy :) [10:56] danilos: so just cp it, and change where it's needed [10:57] G: nope, with "reusing" I meant use the same template (perhaps rename it), but modify it so you can pass in the bits like the div ID and "+assignedbugs" part of URL from the view [10:57] ahh yep, I get you [10:57] so basically the second option (introduce logic/pass variables) [10:58] G: btw, would it not be better if there was a page listing all the actual bugs any of your teams is assigned to/subscribed to? [10:58] G: yes [10:59] danilos: you mean something like +subscribedbugs as is, and then something like +subscribedbugs/+all ? [10:59] G: yeah, something like that [10:59] actually, +indirectsubscribedbugs [11:00] danilos: couldn't that be quite 'expensive' if someone did that against say Mark Shuttleworth who I'm guessing is a member of a lot of teams? [11:01] G: it could be nontrivial, sure, but if that's the question you really want answered with this page, it should be doable, imho [11:01] (expensive as in time to generate & amount of data that could be retrieved from the DB for some people) [11:02] G: of course, I am not suggesting you should block on this, but just wondering if that's what you really want? [11:02] G: also, as far as data being transferred from the DB, we do batch most of the big requests [11:03] nigelb: Why are you rendering a view on bugTask in test_bugattachment_edit_view? [11:03] rvba: I should move it to another file, but I wanted to write it some place where attachment bits existed. [11:03] Is there a better place already existing? [11:04] danilos: actually seems +bugs does that already [11:04] nigelb: I don't know that code (bugs) very well but I /think/ the bug is not properly setup for you to render bugTask:+index [11:04] hmm. [11:05] I should perhaps tell you what I'm testing so you can suggest a better, working alternative. [11:05] nigelb: My advice would be to find in the code a properly setup bug (one that ressembles what you want to test). [11:06] When you attach a patch on LP, it shows up in the comments, but the edit link is a portlet on the left. I'm adding that edit link to the link in the patch as well. [11:06] Ah, perhaps I should try that. [11:06] G: yeah, though I'd never claim to know what exactly "related" means there [11:06] nigelb: I see. [11:06] G: at least not before reading the code carefully :) [11:07] danilos: I think 'related' is something all could easily be lost in translation [11:07] G: I also dislike the concept of me being related to a bug ;) [11:07] G, danilos: "Related" is a combination of all the filters. [11:07] s/combination/union/ [11:07] danilos: but the way I see it is: all bugs (x) has had something to do with [11:07] Stuff you're subscribed to, or assigned to, or reported, or commented on. [11:08] hrm, I could do a browser test I guess. [11:08] wgrant, that simple? no team subscriptions considered? [11:08] danilos: Correct. [11:08] wgrant, right, thanks [11:08] ahhh so it doesn't consider via teams [11:08] nigelb: that's already what you tried to do. [11:08] G: Good to see you tackling more issues. :-) [11:09] StevenK: seems a bit of a waste to sign a document to contribute and stop after one issue :) [11:09] rvba: No, the from lp.testing.BrowserTestCase [11:09] StevenK: Isn't it fun that there are two new people contributing and both of us have the same first name ;) [11:09] G: so, my gut feel is that we should be able to generate a single pages with all related bugs for any of the filters which includes your team subscriptions: basically, the "only" difference would be a join through TeamParticipation table :) [11:10] nigelb: That isn't confusing *at all* [11:10] heh [11:10] G: s/single pages/self-contained pages/ [11:10] StevenK: that and, I really like Launchpad, and the code is much better than another unnamed set of tools are :) [11:11] wgrant: Can I hope for mildly better success with BrowserTestCase? :) === henninge is now known as henninge-lunch [11:12] danilos: actually, what I was wondering is if the best way is to create an AJAXy portlet that uses the team list, to basically: "Also show bugs assigned to: " [11:13] or even a "Showing directly (assigned|subscribed) bugs, [link: include indirectly (assigned|subscribed) bugs] [11:14] G: yeah, though choosing through a bunch of checkboxes is a nightmare imho :) [11:15] danilos: yeah, I agree [11:15] but with an on/off type thing, it'd also be possible to still list the user's teams that implements the existing behaviour as well [11:17] G: yeah, though user's teams are listed elsewhere as well (I'd even argue that it's misplaced on +assignedbugs now) [11:18] danilos: well for someone with a lot of team memberships, it allows drilling down to individual teams easily, but I agree to that view as well [11:18] G: true, true [11:18] G: well, I think providing the list won't be bad or wrong, so I suggest you just go for it :) [11:19] (I just realised as well, that it'd likely have to be a non-AJAXy because the searching is also non-AJAXy [11:20] nigelb: I think the only place where this (adding a comment with a attachement) is tested is bugtarget-filebug-views.txt === al-maisan is now known as almaisan-away [11:21] doctest :( [11:21] Indeed. [11:22] Maybe just a portlet that basically has: http://dev.nigelj.com:81/bugportlet.txt [11:23] nigelb: my advice would be to see what your fix does to that test. If you only need to add a couple of lines in there, it's fine. [11:23] or is that getting a little too complex? [11:23] anyone else seeing update-sourcecode breaking when run from rf-get? [11:23] bigjools: the cache file? [11:23] yes [11:23] yeah, I saw that yesterday [11:23] I was told to just use the merged version in devel [11:24] there's a conflict [11:24] Someone probably updated it on an i386 machine. [11:24] So the order is different. [11:24] nigelb: I'm sorry but I'll have to leave you to it now. Maybe danilos can give you a hand if you need more help. [11:24] rvba: Sure, no problem, I'll dig deeper. [11:25] bzr resolve --take-other FTW [11:25] yeah, that's what I did [11:25] *too [11:25] danilos: does that look semi-sane, or a bit too long? [11:25] This may come off as exceedingly stupid. How do I run a doctest? [11:26] G: I think it'd be fine [11:26] nigelb: bin/test -t [11:26] G: and definitely an improvement over what we have now [11:26] bigjools: ah, thanks! [11:27] danilos: and basically the idea is for it be no-AJAX, and just add a includeindirect=1 to the request to +subscribedbugs [11:27] nigelb: remember that doctests are loaded by unit tests [11:27] nigelb: ./bin/test -cvv -t xx-doctest.txt [11:27] (or assignedbugs for that matter) [11:27] G: I'd also do it in two steps (branches) to make it easier: generalize what we have in one branch and make it work for the subscribed bugs case, and then another one to provide the indirect bug searches [11:28] G: yeah, that sounds good, though that may conflict with the search parameters [11:28] danilos: so basically 1 branch to include it for subscribedbugs, and then a second branch to implement show indirect [11:28] hmm, is something up with loggerhead on staging? [11:28] bigjools: oh. that's why everyone tells me doctests having a slightly weird loading. [11:28] G: yep, that also ensures you solve at least one problem [11:28] danilos: ahhh a good point === matsubara-afk is now known as matsubara [11:29] G: (if the "include indirect" turns out to really be too expensive) [11:29] jelmer: Probably. asuka has been in swapdeath twice in the last 24 hours. [11:29] jelmer: Stuff may need restarting. [11:29] jelmer: although codebrowse is tellurium, so I am stupid. [11:29] Gah, I fixed it and I don't see a text breakage. [11:29] *test [11:29] The one time I'm unhappy a test didn't break. [11:29] nigelb: this is an example of one: lib/lp/archivepublisher/tests/test_publisher_documentation.py [11:29] wgrant: ah, ok [11:29] danilos: alright I'll type it up in launchpad [11:29] wgrant: either way, I guess should ask a losa [11:30] wgrant: what was up with asuka? [11:30] G: cool, I'll be switching locations soon, but if you want me to comment on the bug as well, just let me know :) [11:30] jelmer: I'm not sure that was ever determined. [11:30] bigjools: Ooooh. thanks! [11:30] danilos: thanks! [11:32] wgrant: hmm, that makes me feel a bit uneasy about the stuff that is currently on staging.. [11:33] jelmer: You could obtain process listings. [11:33] jelmer: I believe they are taken every minute or 5. [11:33]