[00:05] * jelmer will have a look at the merge conflict [00:06] Hah. [00:06] I was just going to ask if you were still around. [00:06] Thanks. [00:53] hi wgrant [00:59] Morning poolie. [02:10] wgrant: with the webservice api if something is exported for version 'beta', can i change it? [02:11] Edit the decorator in the interface [02:12] StevenK: you are meant to be away? anyway, you mean edit the decorator to change the export_for_version() ? [02:12] i didn't think that was allowed? [02:13] wallyworld: (Where are your multiple underscores?) Just because I'm holidays and not working doesn't mean I can't answer questions. [02:14] wallyworld: It depends. What are you changing? [02:14] i need to change the semantics of the private attribute on IBranch [02:14] (note that beta is officially EOLed, but 1.0 is still supported, but we will change things still if required) [02:14] Changes to API functions need to be considered very carefully. [02:14] it is read/write (via setPrivate) but i need to to be readonly [02:15] and there's a new explicitly_private attribute instead [02:16] i could keep it read/write and add a new mutator method for explicitly_private [02:16] I would just break compatibility. [02:16] That [02:16] 's the sort of thing that nobody is likely to care about, and those that do can fix their code. [02:16] 1.0 exists mostly to not break code in Ubuntu releases. [02:16] that's what i was hoping for :-) [02:16] None of which uses Branch.private. [02:17] cool. it will still appear the same for r/o access [02:17] but now you write using 'explicitly_private' instead of 'private' [02:17] Is there a git-supporting branch yet? [02:18] wallyworld: Right. [02:18] luke-jr: No. [05:04] Morning! [05:04] wgrant: I'm going to take a screenshot of you telling lifeless to go away :P [05:04] Heh. [05:06] I'm trying to do a html = create_initialized_view(person, '+index')() [05:06] the html is turning out empty. [05:06] (this is in registry) [05:06] what am I doing wrong? [05:21] Is person actually a Person? [05:23] well, yes. I used factory [05:23] This is right, isn't it? person = self.factory.makePerson(time_zone='Asia/Kolkata') [05:24] That is indeed correct. [05:24] Hmm. [05:24] What is returned from create_initialized_view(person, '+index')? [05:25] empty html as far as I can tell from the test failure [05:26] err, empty string [05:29] The user story from the guy earlier is depressing. [05:30] Also, slightly troll-ish. [05:39] Mmm, it's pretty reasonable. [05:39] LP hasn't changed significantly since 2005. [05:39] UI-concept-wise, that is. [05:40] Sure, but LP has a slightly different concept for branches [05:40] Its tied to a project and not a person. [05:40] Although all those fields he talks about are hidden when unset, except when you're looking at your own page... [05:40] Yeah. [05:40] Right, but registering a project and stuff is difficult and awkward. [05:41] Though, I'd love to see UI improvements. [05:41] Does making private branches need an admin? I've never created one or used one. [05:41] let's not go there. [05:42] nigelb: not only is that sequence of lines - wgrant telling lifeless to go away - worthy of being screenshotted. I'm tempted to also print out said shot and place prominently on the wall. for whenever I need giggles. [05:42] spm: That's the point of the screenshotting! Lets frame it ;) [05:42] nigelb: My theory is that Launchpad's privacy features were designed by GitHub. [05:42] framing might be considered excessive. but not by me. good idea™ [05:42] wgrant: Oh. That bad? [05:43] GitHub's team infiltrated Launchpad before GitHub was created. [05:43] And designed Launchpad's privacy features. [05:43] heh [05:43] To make GitHub win,. [05:43] It is probably the most hopelessly incoherent and useless aspect of Launchpad. [05:43] Are UI improvements on the way? Because even bugzilla is getting a facelift soonish. [05:44] huwshimi has plans. [05:44] But we're without him for a while. [05:44] But I don't really think he can do it on his own. [05:44] and UI designers never stay with us for long. [05:44] So the outlook is not stunning. [05:45] Oh. [05:45] Project db-devel build #827: STILL FAILING in 4 hr 41 min: https://lpci.wedontsleep.org/job/db-devel/827/ [05:46] Most of the points in that blog entry are reasonable, apart from eg. condemning LP for not offering free private branches. [05:46] I wish launchpad were more social. [05:47] Ahh, and then he goes into the usual "why can't I check out the branch I created thing". [05:47] GitHub and BitBucket have poisoned everyone's minds :( [05:47] Wait, you can't checkout a branch you just created on github. [05:47] Can't you? [05:47] The difference with GitHub is that you have to do a server-side fork. [05:48] Rather than just pushing up a new branch. [05:48] So people expect that from LP. [05:48] Nobody had a problem with just pushing before GitHub came along. [05:48] Heh. [05:48] The problem is, well, they're good. With good UI. [05:48] Launchpad's UI has never been its strenghts. [05:48] Code review, however, is. [05:49] Eh, not much had good UI in 2005. [05:49] It's just every other website in the world has moved on. [05:49] Heh, that too! [05:49] Anyone tried mergebox yet? [05:50] wgrant: Also, any clues with the test I can't get to work? [05:51] nigelb: Ah, sorry, got distracted. I was interested in the output of create_initialized_view(person, '+index'), *without* the trailing (). [05:51] It should give you a view object. [05:51] The trailing () executes the view to turn it into HTML, which I want to avoid. [05:51] So we can see what the view is. [05:52] Okay, what's the easiest way to output a view object? [05:52] Just get the repr for now. [05:53] "I want to make a branch and then be able to push it back to the original, through the web interface. This should and has to be easy to do and not require juggling between the LP server and my local PC in order to either branch or merge." [05:53] I'm confused. [05:54] Doesn't one normally develop on one's local PC, necessitating that juggling? [05:55] hm, does github/whatever support merging from the web UI? [05:55] I thought not yet. [05:56] I'm not sure. [05:56] But his thing didn't seem to mention actually merging. [05:56] BUt that's what it is right? Create a branch, push back into original. [05:57] but why would you create a branch and push to LP if you don't want an MP. [05:59] Hm. [05:59] AssertionError: 'Asia/Calcutta' not in '' [05:59] Hmm. So that should render to more than ''... [06:00] Yeah, I did try to figure out and then give up. [06:01] I seem to be having a complicated relationship with tests :D [06:02] Launchpad has that effect on people :) [06:02] Heh. [06:03] I thought I'd get this done without asking too much help, etc. Bam! Tests! [06:07] wgrant: What should be the next move? [06:11] I wonder if I can skip tests for this :D [06:21] No, this definitely needs tests :) [06:21] Let's see... [06:22] Almost got my new lucid-lxc-on-oneiric development environment set up adequately. [06:22] I still don't see the charm of developing on bleeding edge. [06:22] Huh. [06:23] It's '' here too. [06:23] \o/ [06:23] I'm not going mad! [06:23] Using oneiric is nice. [06:24] Get to iron out bugs, whine to thumper about the new unity being crap, fglrx 11.8 is slightly less awful than 11.5... [06:24] thumper works on Unity? [06:24] He defected to the Unity team earlier this year. [06:24] * nigelb had a vague feeling he used to be LP or bzr. [06:24] He was lp-bzr. [06:24] Or lp-code, as it was known in its later years. [06:24] Ah [06:25] Right now, there isn't a subteam working on this component is there? All squads working on bits of everything. [06:25] hi nigelb [06:25] Right. [06:25] Hey poolie, Morning! [06:25] (well, its probably evening for you by now :P) [06:26] afternoon [06:26] nigelb: Is your test in the same class as other view tests? [06:27] Yeah, let me grab that class name [06:27] Its inside TestPersonIndexView [06:35] I'm not quite sure what's going on here. I guess it's not using the root template that includes the slots. [06:37] nigelb: Where are you adding this? [06:37] The "Time zone" section under "User information"? [06:39] nigelb: If so, try asking for +portlet-contact-details instead of +index. [06:39] More specific, and it will probably even work. [06:40] wgrant: doing that now [06:41] Although nothing seems to render that directly either. [06:41] You may just have to give up and test an attribute on the view and not the actual display. [06:41] Or resort to a browser test. [06:41] browser test? [06:42] grep for user_browser [06:42] \o/ [06:43] zope.testbrowser is a minimal web browser that exercises most of the Zope stack. [06:43] +portlet-contact-details worked [06:43] So you'll be able to see the output of the view that way. [06:43] Great. [06:44] Aha. [06:46] Gah, so I get to use doctests again. Win. [06:53] ok, looks like everything works. [06:53] * nigelb forces a failure [06:54] aaargh. lint failure. [06:54] I should be fixing lint failure that I didn't cause, right? [06:56] Ok, fixing this lint failure might lead to breakage. [06:59] wgrant: I get a bunch of "assigned to but never used" lint failures. [06:59] nigelb: In doctests? [07:00] Its unrelatd to code I touched. [07:00] But in the files I touched [07:00] If it's not a doctest, you should just be able to remove the assignment. [07:00] ie, s/a = b/b/ [07:01] A bunch of these http://pastebin.ubuntu.com/677037/ [07:01] Those should all be trivially fixable. [07:02] liene 308 is interesting. [07:02] I think the view is created so that a login token exists. [07:03] Yeah, you probably need to continue creating the view, but you don't have to assign it to anywhere. [07:03] The return value is not important. [07:03] aah [07:08] wgrant: what does the last one in that lint failure mean? [07:09] nigelb: Normally that there's a = or + without a space on both sides. [07:14] wgrant: Could you review https://code.launchpad.net/~nigelbabu/launchpad/188187-time-zone-offset/+merge/73168 [07:14] stub doesn't seem to be here yet. [07:20] Morning all. [07:33] wgrant: not entirely sure that defected is the right word [07:34] thumper: Escaped? [07:34] heh [07:35] thumper: "ran away" [07:35] how about, "needed a change" [07:35] Bah. [07:36] A lot Canonical seem to be Launchpad defectors. :P [07:36] Heh. [07:36] wgrant: You picky picky person :D [07:38] I do that :) [07:38] when I do tal:foo [07:38] foo is a variable? [07:39] It's irrelevant. [07:39] means nothing at all. [07:48] wgrant: Updated! [07:48] poolie: ooh, that's a nice next bug :D [07:48] the team timezones thing? [07:48] yeah [07:49] it has limits [07:49] especially as you know nothing about people's habitual online hours [07:49] True [07:49] I set my timezone on Launchpad as UTC. [07:49] good morning [07:49] i think thisi s one thing people specifically said they missed when we took out the google maps display [07:49] hi abel [07:50] Why did we take out maps? I remember there was some javascript error and then it wasn't around the next day. [07:50] i think part of it was about google maps over https being a for-pay service [07:50] nigelb: Google wanted to charge a heinous sum for HTTPS maps. [07:50] and maybe some licencing problem [07:50] But I believe that is no longer an issue. [07:50] Oh. [07:51] people complained about it being nonfree [07:51] As they have decided that HTTPS should be everywhere now. [07:51] But it has been suggested it should be replaced with OSM, yes. [07:51] I'm always about what works vs non-free :) [07:51] they have decided SPDY should be everywhere [07:51] We had something like that in Loco directory using google maps. [07:52] An OSM developer requested we not use OSM because that's not what its meant for. [07:52] oh? [07:52] it's meant for precise positioning? [07:52] It's interesting that LP has been HTTPS-only forever, and many people used to complain, and now the rest of the world is finally starting to go HTTPS-only. We were years ahead of the curve on one thing, at least. [07:53] yes, which is quite cool [07:53] wgrant: We're ahead only in one curve :) [07:53] of course many of these maps are not all that accurate [07:53] poolie: Because OSM is not ready to get hit with lots of people loading pages with it or something. [07:53] nigelb: Indentation on prejoins has the same indentation issue. [07:54] GRAR [07:54] just showing "... which is on the south-east coast of india" would be enough [07:56] wgrant: okay, fixed. [07:56] poolie: But still, that doesn't help when you want to figure out when is a good time to set a meeting with someone :) [07:56] poolie: If we wanted to be invasive, we probably could work out an awful lot about peoples habitual online hours :-) [07:57] s/invasive/Google/ [07:57] No, please don't. My hours are horrible. [07:57] I slept at 4 am, woke up at 11, and I'm now going to work at 1:30 pm. Fun. [07:58] perhaps it can be a separate service [07:58] oh i guess if we had a timeline view [07:59] Heh. [07:59] timeline of commits? that should be fun. [07:59] should also overlap with marking who's on vacation when. [07:59] Award karma points if you don't obtain any karma between midnight and 8am in your designated timezone. [07:59] badges [08:00] fix one bug every hour for 24h [08:00] heh. [08:00] Double karma between 10 pm and 8 am your timezone. [08:00] I prefer the first variant to try and stop burnout rather than encourage it :) [08:01] Heh [08:01] I'll then reset the timezone to the timezone my brain lives in. [08:01] "Oh, I'm working in Hawaii time today. Iceland tomorrow!" [08:01] I've filed 50 LP UI bugs in 24 hours, but not fixed 24... [08:02] Yer, need roving to cope with my 25 hour cycle :-/ [08:03] Laters! Off to lunch and then work. [09:20] wgrant: Gah. formatimports. I knew I missed something. === henninge changed the topic of #launchpad-dev to: #launchpad-dev ist: https://dev.launchpad.net/ | On call reviewer: henninge | Critical bugs: 237 - 0:[#######=]:256 [09:31] nigelb: When you fix, I will send it off to ec2. [09:31] Although it won't land until jelmer hopefully sorts out the testfix on db-devel. [09:32] I'll fix when I get home. In about 5 hours. [09:32] nigelb: Thanks! [09:32] One more inch closer to 128. [09:47] wgrant: whoops, I hadn't seen that yet [09:49] jelmer: Ah, wasn't expecting you around yet. [10:41] testfix is on the way [10:45] jelmer: Thanks! === henninge is now known as henninge-lunch [11:04] in other news, bzr code imports seem to work on qastaging :) [11:04] https://code.qastaging.launchpad.net/~jelmer/bzr-svn/revprops [11:20] wgrant: and it looks like staging was updated? === luke-jr_ is now known as luke-jr [11:43] jelmer: Indeed, spm unbroke it this afternoon :) [11:45] jelmer: What UI is there for them? "Import details" doesn't seem to show anything... [11:45] wgrant: The UI bit hasn't landed yet, the only way to create them on qastaging is using the API [11:45] jelmer: Ah, that would do it. [11:45] wgrant: I have another branch up for review that changes the web UI to register "bzr" code imports rather than mirrors [11:46] Great. [11:46] then someday, perhaps, we can get rid of the branch puller [11:47] Well, we'd need to fix the way imports work first. [11:47] They still use the puller :( [11:48] yeah, that bit I'm still pondering [11:48] But that is a significantly easier problem than general branch mirroring. [11:48] a related issue is that it would be really useful to do stacking of code import branches, but the current architecture makes that hard (especially if the development focus is not a code import) [11:48] Indeed. [11:49] I think they should probably just be bzr+ssh clients or so. [11:49] Or maybe use the internal branchfs. [11:49] adeuring, hey, good morning. could you take another look at that oops-tools branch you reviewed on Friday? https://code.launchpad.net/~matsubara/oops-tools/829460-typeerror-qsmmx/+merge/73062 [11:49] matsubara: sure, but I'd like to have lunch first :) [11:49] Assuming I don't accidentally delete all the hosted areas again. [11:50] adeuring, sure. [11:50] wgrant: :) [11:51] jelmer: I guess we probably don't have to consider stacking on private branches. [11:51] Which might make things easier. [11:52] wgrant: What about private users who are importing multiple branches? [11:53] jelmer: We don't support credentials in URLs, do we? [11:53] So private imports are not hugely likely at the moment. [11:53] wgrant: We do, at least for bzr-svn/bzr-git/bzr-hg imports (some people have silly requirements like logging in with guest/guest. Not sure about mirror URLs. [11:53] Bah. [11:53] OK. [11:54] So we just have to grant the service user access to dev focus branches sometimes too, I guess. [11:56] wgrant: hmm, yeah [11:57] wgrant: I guess just supporting stacking on public dev focus branches would be easiest for now, and already be a huge improvement over the current situation. [11:57] It's all going to be special-cased anyway, because branches can be owned by anyone. [11:57] But only owners can write. [11:57] does zope support impersonation ? :) [11:58] It can, but LP doesn't use it. [11:58] eg. SCA is special-cased in security adapters. [11:58] I imagine this would, for now, be something similar. [11:59] ah, interesting [12:00] We just killed bazaar-experts, now you want to resurrect it? [12:01] :) I'm glad that's gone.. [12:01] A celebrity for the importds is probably best for now, I fear. Unless we want to give them access to the branchfs, which sounds bad. [12:02] jelmer: The team still exists, since lifeless is a wimp. But it is no longer a celebrity. [12:02] wgrant: That makes me very sad. === benji changed the topic of #launchpad-dev to: ist: https://dev.launchpad.net/ | On call reviewer: henninge, benji | Critical bugs: 237 - 0:[#######=]:256 [12:02] StevenK: For now. [12:02] StevenK: Until we have a less bad permission system. [12:03] "now" in Launchpad time -- that is, at least a decade. [12:04] wgrant: I agree, about having a celebrity === henninge-lunch is now known as henninge [12:05] wgrant: Is it just my change of perception, or is Launchpad development quicker these days? [12:05] benji: O hai -- did Irene spare you her wrath? [12:05] jelmer: Hmmm. [12:05] jelmer: I think it depends where you're working. [12:05] And how you're working. [12:05] and what you're doing. [12:05] But it may be. [12:06] StevenK: yep, it wasn't too bad here; we just got a good dose of rain [12:06] benji: Ah, rain is easily coped with. [12:06] jelmer: Code's always been quicker than the rest, I think. [12:06] jelmer: Because it's been designed. [12:07] At least a bit. [12:07] At least I've always found it nice. [12:07] henninge: benji Hi guys, could any of you have a look at this (Javascript) MP https://code.launchpad.net/~rvb/launchpad/confirmation-overlay-bug-830982/+merge/73078 ? [12:08] rvba: I'm available. [12:08] wgrant: hmm [12:08] benji: Thanks a lot! [12:08] benji: I'd like to take ;) [12:08] henninge: be my guest [12:08] wgrant: can you do me a favour tomorrow? put the performance tuesday tag on in your morning [12:08] lifeless: Only if you go away, but sure :) [12:08] heh [12:09] moar screenshots. [12:09] wgrant: given I just drove back from the hospital and am about to crash for 6 hours; I can guarantee I am going away. [12:09] :) [12:09] Night! [12:09] night, thanks! === wgrant changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: henninge, benji | Critical bugs: 244 - 0:[#######=]:256 [12:10] danilos: Hi. [12:28] Hmmm. [12:28] This is interesting. [12:28] Trying to upload via FTP to a poppy running in LXC, and I have the hanging-on-the-last-kilobyte bug. [12:29] bigjools: ^^ [12:29] nice [12:29] Reproducible too. [12:29] repeatable for any upload? [12:29] * wgrant debugs. [12:29] Does suggest it's odd networking, though. [12:29] yes [12:29] does the LXC NAT? [12:30] I'm using libvirt's bridge. [12:30] I'll lay real money on it being a nat type issue [12:30] It does NAT, but this shouldn't be using it. [12:30] Because it's host-to-guest. [12:30] Haha. How much real money? [12:30] 2 pence [12:30] So should just go straight over the internal bridge without NAT. [12:30] it'll be doing something internally still [12:30] So given the UK currency that's about 0.5 Australian cents [12:30] Let's hope. [12:31] StevenK: miaow [12:31] Wow. [12:31] Wireshark doesn't hang Unity! [12:32] A complex application that doesn't hang oneiric's unity is a rarity these days. [12:32] well, its moved from simple to complex, that's worth rejoicing. [12:32] Ah, there we go, lagging up a bit now. === jtv is now known as jtv-eat [12:39] Hmm. [12:39] So it does the gpg verification, then doesn't send anything back. [12:40] wgrant: I wonder if an SFTP upload also hangs with LXC involved [12:40] I doubt it. [12:41] It works. [12:41] remember that this happened with the zope server too [12:41] * bigjools -> food [12:42] Yeah, that is pretty odd. [12:53] So, it turns out that in this case it's the GPG verifier crashing. [12:54] Nice! [12:54] Do you have a traceback? [12:54] But I can't see the exception by default... [12:54] Not until I add any extra log entry right before the crash. [12:54] once I do that, I get an error traceback. [12:54] WTF twisted? [12:55] Hmm. [12:55] I wonder if logging doesn't get completely initialised at startup. [12:56] Writing to the Python log somehow finishes setup, allowing the Twisted logger to log. [12:56] Can you paste the traceback? [12:57] The problem with it is clear: the "key not registered" error is unicode, while twisted wants a str. [12:58] But this doesn't explain all the hangs, sadly. [12:59] It may explain why we can't work them out, though. [12:59] I wonder if that can tested sanely [13:02] Morning, all. [13:03] matsubara: r=me [13:03] morning deryck [13:03] thanks adeuring [13:15] mm, LP commits visualized. http://www.youtube.com/watch?v=9Pe70fw83Zs === jtv-eat is now known as jtv [13:23] adeuring, hey, could you join me in mumble for 2 seconds? Just need to test I'm working. [13:31] adeuring, ping for standup. [13:31] deryck: ok [13:33] henninge, benji: can I add a MP to your queue? [13:34] jelmer: sure === henninge changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: benji | Critical bugs: 244 - 0:[#######=]:256 [13:34]