[00:07] lunchtime [00:25] spm: so is it all back now? [00:25] poolie: yes [00:26] could you please tell identica that? [00:27] sure. give me a few to work my way thru the howto of that... ;-) [00:28] go to identi.ca [00:28] click login [00:30] you're enjoying this.... ;-) [00:30] and you're ok from there? [00:31] login as launchpadstatus [00:31] then type in the little bittum [00:32] right. I'm in. now for some pithy statement [00:35] gosh. that's almost easy. [00:56] hmm. thumper is no longer present. [00:57] what's the ETA for PQM opening? [01:14] * thumper is present [01:19] thumper, oh. I was just thinking about how, with your recent changes to the comment form on the review page, it might be nice to also have an MP status field there [01:19] jml: nah [01:20] jml: not until I split out the code review status from the merge status [01:20] which I still want to do [01:21] I'm going to close irc now to focus on pdr [01:21] thumper, oh right. [01:21] thumper, ok :) [01:21] if anyone needs me because everything has blown up, call or text [01:22] thumper, kerblammo [01:22] thumper, I mean, good bye :) [01:44] beuno: you around? [04:22] * thumper signs off for the year [04:24] thumper, ciao [04:24] thumper, merry christmas & a happy new year. [05:25] * mwhudson EOWs [05:25] mwhudson: enjoy your we === abentley1 is now known as abentley [07:31] spm, still here? spam - https://edge.launchpad.net/buyviagra [07:32] oookay. that's weird. [07:32] gone. the weirdness was the user account was already suspended; had to reset some of the "user" fields for the projet to registry [07:35] thanks [07:37] poolie: fwiw. I do 10am - 7pm these days. Probably till we're out of daylight savings. [07:37] oh ok [07:37] covers the gap 'tween michael and tom, as it were. [07:37] jelmer, hi === spm changed the topic of #launchpad-dev to: Launchpad Development Channel | Week 4 of 3.1.12 | PQM is open; RM: danilos | https://dev.launchpad.net/ | Get the code: https://dev.launchpad.net/Getting | On-call review in #launchpad-reviews | Use http://paste.ubuntu.com/ for pastes [07:48] spm, wuuu [07:49] spm, in danilos still RM? [07:49] jml: sure, it's still week 4, isn't it. [07:49] ? [07:49] henninge, then why is PQM open? [07:49] jml: I figure it's more appropriate for him to demote himself out of the RM role ( danilo_ see topic) :-) [07:50] jml: good point [07:50] jml, well, I believe the process was to stay the RM until the next release; i.e. there might be CPs and such [07:50] jml: you are saying: No more mangement needed once pqm is open? [07:50] danilo_, ahh, ok. [07:51] jml, I am not sure though, it's not that clear, so let's clarify that for the next one [07:51] henninge, no more release management is needed until the next release, which is next year. === danilo_ is now known as danilos [07:51] henninge, which is aaaaaages away. [07:51] jml, years even :) [07:51] yeah [07:52] although not as bad as a few years back where it might have been in the next millenium! [07:52] jml, btw, we are opening PQM because there's no need for a re-roll, at least there was none so far [07:52] henninge, but people still dispute when was that exactly ;) [07:52] danilos: talking about days or years? [07:53] I know, officially it was 2000/2001 [07:53] danilos, even more controversially, when was the first Easter of the new millenium? :P [07:54] jml: .... don't .... open ..... box ..... it's Pandorra's ..... [07:54] jml, heh :) [07:55] * spm opts for the simple answer - when he had the easter long weekend; which year is irrelevant. ;-) [07:56] spm, bah, they didn't even give us an extra day off for the millenium [07:56] OR the olympics. [07:56] shocking really [07:56] jtv, et al: one minor issue on that approve-imports - "-q" <== what's with the warning's??? ;-) [07:56] jml: truly [07:57] spm: it's an operational problem situation that happens once every few months that requires special intervention from a human who understands what's going on with these packages. [07:58] spm: in principle we could turn these into email notifications, but it's just one more thing that could break without us ever noticing [07:59] jtv: :-) I'm teasing. this is endemic across all of LP cronscripts. -q (ie Be QUIET!) doesn't stop warning's - which it should; we should only see ERRORs or higher. [08:00] spm: we just shouldn't issue warnings for things that aren't worth one [08:00] on the more ..... painful scripts that refuse to be quiet we throw to logs and let scriptactivity catch 'em [08:00] right [08:03] * jml submits a bug fix [08:06] jtv: where this particular is even worse - normally all warnings should be STDOUT; ERRORS usually to STDERR; CRITICALS certainly to STDERR. in LP's case; all scripts dump to stderr; which means if we throw to log - we will NOT know if a script is barfing (stack trace etc) unless we go looking for it. This is a Bad Thing(tm). [08:07] spm: indeed... we've got the weirdest, most mundane stuff going out to stderr for some reason [08:19] danilos: do you have time for a review regarding the distroseries access when translations are hidden? I went further than fixing the bug and trying to refactor some code... again [08:31] adiroiban, heh, you are unstoppable :) [08:31] adiroiban, sure [08:32] lp:~adiroiban/launchpad/bug-497438 [08:32] * jtv cheers for adiroiban [08:32] well, holydays are comming [08:32] adiroiban, do you have an MP as well? (that's much easier because it keeps the patch up to date) [08:33] danilos: nope. [08:34] i am not sure I did the right thing [08:34] and this is why did not pushed it as a MP [08:34] it's about checkTranslationsViewable [08:34] as it is implemented in 2 classes [08:34] and each class is calling the other class method [08:35] i tried to break that chain and define all the logic required by checkTranslationsViewable in a single place [08:35] not it is split between registry and translations [08:37] s/not/now/ [08:39] adiroiban: not sure what you mean by each class calling the method on the other... all browser code that needs to make this check delegates to the distroseries model class, no? [08:41] checkTranslationsViewable is defined in registry.model.distroseries and also in translations.browser.distroseries [08:42] why not have it defined only in registry.model.distroseries ? [08:43] as checkTranslationsViewable from registry.model.distroseries is calling checkTranslationsViewable from registry.model.distroseries [08:51] adiroiban, first of all, when you inherit a permission, you probably want to use it via super() (i.e. in AdminDistroSeriesLanguage) [08:51] adiroiban, however, since the two permissions are on different objects, you should base AdminDistroSeriesLanguage on AuthorizationBase and keep the code as it is [08:52] uhm, that was for AdminDistroSeriesTranslations [08:53] adiroiban, for AdminDSL, I think it makes more sense to use AdminDistroSeriesTranslations there, even though they are identical right now [08:53] do we have any admin'ing on dsls? [08:54] adiroiban: ok. I will use AuthorizationBase and base DSL to DS [08:55] jtv, well, base both off AuthorizationBase, but call DS one in DSL implementation, if you understand what I mean :) [08:56] danilos: ah yes that's better, thanks :) [08:56] jtv, perhaps we don't yet, but it doesn't hurt to improve it... we also allow admins to take a look at translations for non-open distroseries [08:56] danilos: should we also change this code then? [08:56] class AdminDistributionTranslations(OnlyRosettaExpertsAndAdmins, EditDistributionByDistroOwnersOrAdmins) [08:57] adiroiban, no [08:57] adiroiban, that's exactly the point I am trying to make [08:57] adiroiban, i.e. that one inherits OnlyRosettaExpertsAndAdmins, which doesn't depend on self.obj at all, only on the passed in user [08:58] adiroiban, and on a permission which has a compatible object (self.obj is IDistribution in both) [08:59] danilos: ok :) [08:59] adiroiban, so, that one is fine... but eg. we can't add AdminDistroSeriesTranslations to that list though, because DistroSeries and Distribution are different objects :) [09:00] that's true [09:02] my idea was to make thinks simple, by always inheriting AuthorizationBase, when we are going to overwrite checkAuthenticated [09:04] things simpler [09:04] jml: hi [09:07] Morning! [09:08] bigjools: It works! Thanks. [09:09] * noodles775 thinks bigjools is probably stuffing as much wood into his heater as he can right now... [09:11] * wgrant would prefer that to the weather we had here on Wednesday. [09:12] how hot did it get? [09:12] Only 39°C down here. [09:12] ! [09:12] Hopefully we'll have no 45°C days this summer... [09:13] noodles775: is quite right [09:13] wgrant: you got a 3.0 package in? [09:14] bigjools: I tried one in my PPA, and one was synced a couple of hours ago. [09:14] woohoo [09:14] But gina doesn't seem happy. [09:14] Is iron not done? [09:14] hmmm I asked for all 4 [09:17] The diff worked, though. [09:17] Not sure where that is run. [09:17] wgrant: what makes you think gina is not happy? [09:18] bigjools: It's not imported anything v3, yet you said 11 hours ago it was upgraded everywhere. [09:19] And I thought it was running four times a day now, although even if it's two it's still close. [09:19] wgrant: switch wasn't thrown until 0350 UTC [09:19] yrsh it's 4 [09:19] bigjools: gina is disobedient. [09:19] And does not obey such switches. [09:19] oh of course [09:20] I'll get the log file checked [09:20] Thanks. [10:34] danilos: those changes are not sound, so I will push a new revision. But I am not happy with the current way of implement the security check for distroseries translations [10:52] adiroiban, well, there are many things I am not happy with in the code, but what works today is always much better than what might be there in the future :) [10:54] I am not sure if moving the security checks in the model is the right thing to solve the problem [10:54] adiroiban, btw, sorry, I had a call so couldn't continue looking at it [10:54] danilos: np [10:56] adiroiban, so, in general, I am not sure why do you need permissions on the model here? [10:56] adiroiban, what we should have is launchpad.View permission always allow those with launchpad.TranslationsAdmin permission [10:56] adiroiban, at least for all the translations objects [11:00] in the model for distroseries we have checkTranslationsViewable, which raised an exception if those translations should not be displayed to the user [11:00] and showing or hiding the translations depends on the user [11:03] adiroiban, so, that approach is ugly :) [11:03] and right now, before calling that fuction, the view checks if the users have the rights to see the translations and if does, that method will not be called [11:04] and instead of checking for permission in a single place (like security proxy) [11:04] we check in the view and the model (beside the usual check from the security proxy) [11:06] adiroiban, well, it doesn't sound as something too hard to fix, except that we'd have to do it for all the objects we are rendering in a single distroseries (i.e. pofiles, potemplates, DSLs, templatesets, etc.) [11:07] yes, and there are also some test that will fail [11:07] i will try to fix it, but in another branch [11:07] adiroiban, basically, we'd need a generic permission for all of these to allow those with check_permission(TranslationsAdmin), but for objects which can be somewhere else (i.e. pofiles and potemplates can be in a product), we'd have to have custom permissions [11:07] adiroiban, yeah, please keep the fixes separate from refactoring, that makes it much simpler [11:11] danilos: another note. I think we should stop using admin_browser and switch to an translationsadmin_brower, or better to rosettaexpertes_browser and distrotranslationsadmin_browser [11:12] maybe we can find shorter names [11:12] adiroiban, we should, true [11:12] adiroiban, many tests might need rewriting though [11:13] i think that for most we can just replace admin_browser with newadmin_browser [11:14] newadmin_browser = rosettaexperts_browser [11:14] and when we are dealing with ubuntu or a distribution, use utc_browser [11:14] or dtc_browser (distributiontranslationscoordinators_browser) === jelmer_ is now known as jelmer [11:33] adiroiban, it's not that simple, I am afraid... some tests might be using admin_browser to change stuff that rosetta_browser can't (i.e. setting up a bzr branch for a productseries); I am just guessing, I don't know how many cases like these we have [11:34] danilos: I agree [11:34] adiroiban, anyway, stories are not meant to be comprehensive [11:34] then where should we do those tests? [11:35] adiroiban, well, we should do basic tests for major user stories like you suggest [11:35] adiroiban, we should not comprehensively test all bits and pieces... i.e. if we test that UTCs have privileges, we should probably not test that rosetta experts have, and so forth... i.e. what you said yourself :) [11:37] adiroiban: ok. but can we have rosetta_browser and utc_browser created by default in pagetest, so that we don't need to create them in each test? [11:39] I can just start migrate the current test as I come across them [11:39] with the latest change in permission system, most of them are no longer valid [11:44] adiroiban, yeah, we can [11:45] adiroiban, but, since those are based on sampledata, I'd rather just have a utility function which constructs them [11:45] adiroiban, i.e. sampledata that doesn't exist === mrevell is now known as mrevell-lunch === vila is now known as vila-lunch [12:26] danilos: True. But then we need to have something like this at the beginning of each test for distributions http://paste.ubuntu.com/343746/ [12:27] should we but it in factory.py ? [12:29] adiroiban, oh, I said a utility method so you should replace that code with utc_browser = setup_distribution_translation_admin_browser() [12:29] adiroiban, this should be a pagetest helper, not part of factory [12:30] yep. [12:31] adiroiban, it should be added to lib/canonical/launchpad/testing/pages.py and to the setUpGlobs so it's accessible widely [12:31] thanks! [12:31] i was just searching for that :) [12:32] adiroiban, I just don't want utc_browser to be set-up for every pagetest, because it's complex to generate (and would slow everything down) [12:32] adiroiban, so, don't set it up like the other browsers are set up [12:32] agreed [12:36] something like this: test.globs['setup_utc_browser'] = setup_utc_browser [12:36] just a reference to the function [12:36] ? === mrevell-lunch is now known as mrevell === vila-lunch is now known as vila [13:51] danilos: regarding bug 359180, it's ok to go with yui event-key and have keybindings similar to poedit? [13:51] Bug #359180: Missing keyboard shortcut for navigation [14:13] andrea-bs: so you've already got the ui branch? [14:14] intellectronica, I've just pushed it, but actually I haven't written code yet [14:15] ah ok :) [14:15] just howl if you need any help [14:15] oh and pqm is open. i'll send your model branch for test and landing [14:48] adiroiban, depends, do they conflict with any regular bindings? [14:48] adiroiban, i.e. what is Ctrl+C for? [14:48] :) [14:50] my fault. it should have been Alt+C === EdwinGrubbs_ is now known as EdwinGrubbs [14:54] adiroiban, I've added a comment to the bug, fwiw :) [14:54] adiroiban, Alt + C might also interfere with what people might be doing in their browsers (depending on the translation of their menus) [14:56] but those shortcuts will only be available on +translate page [14:57] if we are going to make some complicate keybindings just to avoind any browser in any language, they will no longer be shortcuts [14:57] don't know what to say === salgado is now known as salgado-lunch [15:33] does 'ec2 land' work for anyone? i get this errror when trying to use it: http://paste.ubuntu.com/343765/ === EdwinGrubbs is now known as Edwin-afk === salgado-lunch is now known as salgado === Edwin-afk is now known as EdwinGrubbs [17:43] beuno: you around? [18:20] Pilky, hey [18:31] beuno: hey, just wanted to discuss a crazy idea I had with you [18:34] Pilky, I'm eating lunch, but, shoot [18:34] I'll read on and off [18:36] ok, well I was thinking about launchpad UI again last night and the thought occurred to me that trying to use regular web design for it is doing it a great injustice, as it isn't a web site, but a web app on the more complex end of the scale [18:37] so I had the crazy idea of attempting to make a cappuccino UI for launchpad, using the API [18:37] and designing it as though it is an actual application rather than a powerful website [18:37] Pilky, interesting [18:37] the benefit is, it would be able to be a completely separate thing, by going through the API the UI would be able to be a separate project [18:38] would that mean making all pages ajaxy? [18:38] meaning a smallish team could work on it and provide consistency [18:38] well not as such, have you heard about cappuccino? [18:38] I haven;t [18:38] I expect it's a framework? [18:38] well, basically it is the closets to a desktop calibre API for building high end web apps [18:38] yeah [18:39] it basically has two components: cappuccino the framework and Objective-J the language [18:39] Pilky, it's an interesting idea, I'd pitch it to the -dev mailing list [18:39] Cappucino is to Cocoa as Objective-J is to Objective-C [18:41] and it allows for applications along these lines: http://280slides.com/ [18:41] Pilky, does it scale well? [18:41] into a gazillion requests per minute, etc? [18:41] well I'm not sure, I mean the application, once loaded, runs entirely in the browser as it is basically pure javascript, such things would have to be tested [18:41] splitting the UI that much is certainly an interesting idea to explore [18:41] but the benefit is, rather than replacing the current LP UI it can be an alternative [18:41] but it would make it feel much more like an application, which is what it really is [18:43] the benefit of cappuccino of course is that it is LGPL so stuff can be added if needed [18:43] yeah, would allow more experimentation [18:44] yeah, I've been looking for a project to allow me to really test out cappuccino and something like this would be ideal [18:44] I'll post something to the -dev list and see if anyone else would be interested in helping out one something like this in the new year [18:46] sure, I'd have to look into objecive-j as well :) [18:46] heh, well the website for it is: http://cappuccino.org/ [18:47] and the IDE they're making that is currently in beta is at: http://280atlas.com [18:47] the benefit to me is that Objective-J/Cappuccino basically are Obj-C/Cocoa for the web so my Mac/iPhone development knowledge is quite easily transferable [19:03] Pilky, I will poke around, i always like learning new things [19:04] cool [19:29] beuno: ok I've posted something to the list. I'm really piling up a large amount of open source projects on my plate at the moment :P [19:30] this will be the 4th one I'm working on, not including all the various bits of code I release === salgado is now known as salgado-afk === mup_ is now known as mup