[00:19] It would appear that Google Maps is rejecting Launchpad's requests [00:19] e.g. https://launchpad.net/people/+me [00:23] Bug 624981 [00:23] <_mup_> Bug #624981: The Google Maps API server rejected your request [00:23] spm: when you are around, I want a profile please [00:23] lifeless: your name is Rob, Your Y yrs old; live in Z ?? [00:31] spm: funny man :P [00:31] spm: on staging [00:31] as soon as it finishes updating [00:31] I want to hit up https://api.staging.launchpad.net/1.0/bugs/414746/attachments [00:31] <_mup_> Bug #414746: speakers cannot be muted when using headphones regression (karmic)

[00:33] also [00:33] https://api.launchpad.net/1.0/1.0/bugs/414746/attachments/+login har! [00:33] <_mup_> Bug #414746: speakers cannot be muted when using headphones regression (karmic)

[00:33] when an API times out [00:34] it gives the usual OOPS [00:34] with a link to login [00:34] which then blows up [00:39] * spm is running out of tabs that aren't doing 'useful stuff' I may have to open a 2nd konsole of 16+ terminal tabs.... [00:41] did you say konsole? [00:42] aye. best terminal around, I've found. [00:43] tabs at the *bottom* <== biggest positive feature over any other I've found [00:43] <- xterm [00:43] * wgrant quickly emails Ng. [00:43] well, technically, uxterm. [00:44] heh [00:44] wgrant: i don't think he cares too much. and well, if it makes him cry? I see that as a plus. [00:45] lifeless: so the restore is gtetting there; just doing the importds atm. maybe 5-15 mins eta? [00:47] spm: cool cool [01:06] The linter doesn't run henninge's thing, does it? [01:06] Would be handy, to stop imports from going bad. [01:07] lifeless: revised eta, another 5-15 mins. sigh. still going. looks to be doing stuff on the staging buildmaster... [01:22] spm: hows it looking ? [01:22] [10:22:28] staging codebrowse just went down, so must be getting closer... [01:23] make build LPCONFIG=staging <== on multiple builds/machines. painfully slow. we should probably look into parallelising that. but /me worries about KISS and debugging obscure staging update woes. [01:29] oh yay, db-devel merge failures [01:29] and so it begins === Ursinha is now known as Ursinha-afk [02:01] lifeless: want to catch up? [02:17] spm: we really really shouldn't be running make build from scratch on every machine [02:17] that's just bong [02:18] and wrong [02:18] thumper: sure [02:19] spm: I can haz profile ? [02:19] one sec, working iwth u1 atm [02:20] lifeless: asuka is doing the nightly.sh run, so be ware it will be slower than usual; if you consider that actually possible [02:21] * spm is still trying to unstack/pop about 4 levels of yesterdays interrupts... sigh/woe is me/ etc etc etc [02:22] spm: yes I still want a profile [02:25] LP needs a GitHub-like commercial offering. [02:27] define that please [02:31] At the moment, you need to get a private project set up manually and poke sysadmins and blah blah. [02:31] And it's far more expensive than GitHub's cheaper offerings. [02:33] lp would also need to start supporting commercial customers better & it's not clear that this is a useful use of resources [02:35] Well, there are no other bzr hosting solutions around. [02:35] And GitHub is a really compelling reason to use Git. [02:36] So the social benefit of causing more engineers to use bzr for their closed-source efforts is expected to outweigh the social benefit of making it easier to host open-source stuff? [02:38] Well, the former would probably cause more wide-spread use of bzr. [02:38] At the moment, if your company wants to use a DVCS, they are probably going to choose Git. Because GitHub is there. That's what I've heard so many people say. [02:39] Personally, I think bzr needs self-hostable hosting solution to properly enable corporate use of bzr [02:39] For many cases, yes. [02:42] win 41 [02:42] Gah. [02:52] so, self hosted via self deployed LP is possible, but traumatic [02:52] it would be nice if that was easier [02:52] wgrant: that story does need to be easier; for sure. [02:52] I guess we need Vostok, but ripping out everything but Codehosting instead. [02:53] there was talk at one point of developing a self-hosted solution for bzr completely separate to LP [02:54] spm: ping me when you're ready [02:55] lifeless: there was but it was shot down with a BIG gun [02:56] I didn't agree with the argument [02:56] I think there should be a simple deployable codehosting solution for corporates [02:56] based on bazaar [02:57] spm: ping me when you're ready [02:57] lifeless: timing is everything.was just loggng into asuka to set that up :-) [02:59] thumper: I think there should be too. I would love it if it had common code with codehosting [02:59] * thumper nods [03:03] lifeless: just about to restart app server [03:04] duh da [03:04] duh da [03:04] duh da duh da duh da [03:04] .... waiting..... [03:04] yay. down. [03:05] starting.... [03:05] I'm glad it starts up quickly [03:06] it lives! [03:06] ok, you can turn that off, thanks. [03:06] anyone fixing the merge conflict? [03:06] anyone? [03:06] lifeless: you may wish to try again; it's only just started working, ish. [03:06] if you can expedite the rsyncing of the trace and oops (OOPS-1700S32) that would be great. [03:06] beuler? [03:06] thumper: its his day off [03:06] haha [03:07] no, I'm not currently fixing it [03:07] lifeless: all done? I'll restart [03:07] spm: yes [03:07] oki [03:07] spm: though I won't know the quality till I get the files ;) [03:07] nmp [03:08] sorry. forgot the 'I wear evil horns' smily there... oh well, it was implied. [03:08] wallyworld: feel like trying out fixing the merge conflicts? [03:08] wallyworld: you can say no if otherwise engaged [03:08] lifeless: they should be synced [03:09] lifeless: is "from datetime import datetime, timedelta" ok with our new import style? [03:09] lifeless: or do I have to put them on multiple lines? [03:09] its fine [03:09] you mean some issues caused by my branch? [03:10] wallyworld: no [03:10] wallyworld: I mean the current merge failure between stable and db-devel [03:10] wallyworld: are you getting those failure emails? [03:11] spm: hmm, i'm not seeing my oops :( [03:11] lifeless: ahh. pebkac. one sec. [03:11] thumper: i've subscribed to canonical-launchpad digest but hadn't been looking in too much detail [03:13] wallyworld: ok, I'll fix it [03:14] lifeless: I was under the impression that the exception was only in place for single-symbol imports. [03:14] wgrant: < 78 chars or whatever [03:14] thumper: i don't mind doing it but if it's time critical then you will be faster than me. but i can look at it np [03:14] lifeless: hrm. not pebkac; looks like a script buglet somewhere. but there now: /srv/launchpad.net-logs/staging/asuka/profiling/ [03:15] wallyworld: I'm sure we'll get some practise next week [03:15] wgrant: frankly I think any time spent talking about it is too much :) [03:15] I'll take this one [03:15] lifeless: https://dev.launchpad.net/PythonStyleGuide?action=diff [03:15] wgrant: sorted one per line when there are many gets better merge conflicts; when there aren't many, anyway its done is ok [03:15] meh [03:16] ok-by-me then [03:16] spm: its thinking rather more now ;) [03:17] spm: its almost like it hasn't been rsyncing since the 3rd [03:17] lifeless: mmmm. possibly. [03:17] -rw-r--r-- 1 robertc robertc 239303 2010-08-27 14:11 2010-08-03_02:04:49.560-MailingListApplication:MailingListAPIView-OOPS-1676S234-Dummy-3.prof [03:17] I dod suspect a script buglet - we have a new shiny ponies script; so... [03:17] was the newest file I had [03:17] note - 27th was when I got it, 08-03 when it was made [03:18] nod [03:22] spm: I has it, thanks [03:40] anyone familiar with bugs code around ? [03:40] I know parts of it. [03:40] What's the issue? [03:40] https://bugs.edge.launchpad.net/malone/+bug/618849 [03:40] <_mup_> Bug #618849: Timeout accessing bugs' attachments using the API [03:41] have a look at that to get the context where I'm looking [03:41] Wow those are fat comments. [03:41] mine? [03:42] All comment fonts are huge. The CSS must be broken. [03:42] you have the beta installed don't you ? [03:42] Yeah. [03:44] Is anybody working on the broken profile page thing? [03:44] ? [03:47] lifeless: All profile pages are broken due to the Google Maps thing. [03:47] They pop up an alert(). [03:47] so, its a maps problem [03:47] for now we're waiting to see if they resolve it [03:47] Is it? The Google thing that Curtis mentioned looks irrelevant to me. [03:47] AIUI [03:48] why? didn't fit the symptoms ? [03:49] Well, it seems specific to another Google service. But perhaps it is more general. [03:49] omg [03:49] omg omg omg omg omg [03:49] lib/lp/bugs/browser/bug.py line 522 [03:50] *this is not safe* [03:50] ..... [03:50] Haha. [03:50] lets go and pull 140+ objects from the database [03:50] the font size issue is known [03:50] lifeless: oof [03:50] and being worked on I believe [03:51] lifeless: makes me think of some of this spec code [03:51] "find all specs named $x then filter out the ones that aren't targeted at $foo" [03:51] (fortunately given how specs are used this probably isn't so bad, but...) [03:51] mwhudson: In Python, not SQL? [03:52] wgrant: yes [03:53] Yay. [03:55] mwhudson, specs get used (or not) the way they do because of the limitations of the impementation, rather than the intent of the users. [03:55] persia: well, maybe, but also it's actually unlikely that there are many specs in the system with the same name [03:58] There are social conventions in place to avoid that, leading to exceedingly frustrating names for Ubuntu specs (${release-target}-${responsible-team}-${relevant-area}-${real-spec-name}) [03:58] persia: well yeah, but even if you just used ${real-spec-name} i conjecture that there would be <10 duplicates for any given name [03:59] it's not like lifeless's code that could be getting 100s of objects for no good reason [04:00] Potentially. I was involved in the nomenclature discussion, but most example cases were 2-3 with the same name, rather than >10. I agree it's good optimisation, I just think it's dangerous to base "how to design blueprints" based on current blueprints usage (as opposed to theory, which is safe) [04:01] persia: you're not suggestion that stakeholders are consulted are you? that's: 1. crazy talk and 2. heresy! [04:01] persia: sure [04:02] * mwhudson has one of these "how does this code work at all" moments [04:03] spm, No. I'm just asking for blueprints design based on blue-sky theory, rather than partial analysis of current usage in an attempt to divine the potential desires of conceptual stakeholders. [04:03] awesome! [04:03] it doesn't [04:04] mwhudson: What's broken? [04:04] Apart from the entirety of Blueprint? [04:05] wgrant: go to add a dependency to a blueprint, click choose, type some text and search [04:05] though it's timing out for me actually, not exploding quite like i expected [04:05] That's blueprints. Expected behaviour [04:06] ok, it's not completely broken [04:06] just stupid [04:10] mwhudson: can I encourage you to do one thing. [04:10] lifeless: sure [04:10] mwhudson: add query-capped tests to the views you touch. [04:10] lifeless: um [04:10] mwhudson: doesn't matter what the count is, just put a ratchet there ;) [04:10] heh [04:10] actually probably not touching any views really [04:10] but ok [04:19] mwhudson: we currently have no insurance for query blowouts [04:20] mwhudson: its all 'someone analysed this once and made it good', then something changes, and because our code has the property that what looks like good python performs terribly, *boom* [04:21] if I have a method that yields and is used as a generator [04:21] and I have an edge case where I don't want to yield anything [04:21] what do you do? [04:21] raise StopIteration? [04:21] return [04:21] thumper: 'return' [04:21] it will be a generator because of the yield statements [04:21] ok [04:22] mwhudson: so anyhow, I think its sensible to put *an* insurance policy in place. [04:23] lifeless: ok [05:09] what is salgado used for in tests ? [05:11] too much [05:11] the default webservice caller uses him i think [05:11] which is crazy, because he's an admin in sample data :( [05:38] how do we ask that we are running in a test environment? [05:38] I have a view that uses the slave store [05:38] but for my test I need to use the master [05:38] to see the new data [05:39] stub: any magic I can use? [05:39] In Soyuz we just commit. It's ugly, but I think it's better than a special case that might break. [05:40] thumper: Sounds like you need a database policy that returns the master store even when the slave is requested. I think there is one in dbpolicy.py already. [05:41] stub: where is that? [05:41] thumper: nah.... need to add MasterOnlyDatabasePolicy - SlaveOnlyDatabasePolicy can be cargo culted for that. [05:41] thumper: lib/canonical/launchpad/webapp/dbpolicy.py [05:42] with MasterOnlyDatabasePolicy(): [...] [05:42] stub: this is for a pagetest, is your solution still sane? [05:42] I don't know [05:42] I think the answer is no [05:42] can I ask the config which environment we are in? [05:42] is that sane? [05:42] or just insane? [05:43] I've done it before, but then the code your testing isn't the code that will run on production [05:43] How come you can't just commit the changes you made to the master so they are available to the slave? [05:52] I'll just commit [06:23] hmm, exported_as doesn't work if there is an attribute with the same name that isn't exported. [06:23] what project should I put the bug on? lazr.restful ? [06:23] yes [06:30] https://bugs.edge.launchpad.net/lazr.restful/+bug/625102 \o/ [06:30] <_mup_> Bug #625102: exported_as does not handle overriding an unexported attribute [07:02] Morning [07:33] thumper: When I want to split a pipe into two, i add a new pipe before the last one, then how did you interactively include only certain changes? It was something similar to shelve? [07:34] thumper: sorry, just realised you've probably EOD. nm, enjoy your evening! [07:35] Ah, merge -i... wonderful. [07:35] Yep, merge -i is pretty awesome. [08:14] wow [08:14] adding one attachment adds 10 queries >< [08:37] good morning === almaisan-away is now known as al-maisan [11:37] stub: do you know if it's possible to use the result from store.execute() like a ResultSet? [11:38] The interface is a little different [11:38] I need it to work in the batch navigator [11:38] so count and slicing is all that's needed I think [11:38] It might work [11:39] I shall give it a go then :) [11:39] Otherwise convert it to a store.find (store.find(Foo, SQL("hairy where clause"))) [11:39] (I'm trying to put the results of findBuildCandidate into a page) [11:40] I doubt that would work, take a look at findBuildCandidate... :/ [11:48] FYI, am getting a Javascript alert complaining about an invalid Google Maps key when navigating to https://edge.launchpad.net/~nick-moffitt [11:54] Are we using Vouchers or is that dead code? [11:58] They're still used for commercial subscriptions, AFAIK. [11:58] lifeless: any reason why installFixture should not return its argument? === al-maisan is now known as almaisan-away [12:24] mrevell: Any comments on the help bubble so far? I put the branch up for review, since it's a substantial code improvement in any case. If there's anything you _hate_ about it I can still opt not to land, or if there's something that's not quite the way you want it then we can do that as a separate bug. [12:36] What's wrong with EC2? I'm getting "remote host identification has changed" errors all the time, meaning I have to delete ~/.ec2/known_hosts again. Also, startup takes ages. [12:40] Also, does anyone know how I can choose a different EC2 site? I can think of one that must be at least 4 megameters nearer than the one I'm getting now. [12:46] jtv: IIRC you can change EC2 sites in the console. IIRC there are only sites in east/west coast of the US and Europe though [12:46] jelmer: and singapore! [12:46] thanks [12:47] Singapore is quite, quite nearby in terms of internet infrastructure… Some ISPs here will probably take traffic for a US EC2 _through_ Singapore. [12:48] (And then it gets slow due to poor bandwidth allocation, so in principle I could speed things up by running a proxy in the Singapore EC!) === matsubara-afk is now known as matsubara === Ursinha-afk is now known as Ursinha [13:52] Where can I find the Bazaar SSH smart server integration into Twisted Conch? [13:54] Is it the Poppy code? [13:58] davidstrauss: Hi [13:58] jelmer, hi! [13:59] davidstrauss: No, poppy is the server code that's used for package uploads. [13:59] jelmer, ah [13:59] jelmer, what is the twisted daemon for branches? [13:59] davidstrauss: My guess is lp.codehosting.sshserver [14:02] jelmer, thanks === almaisan-away is now known as al-maisan [15:44] I received an error when trying to land a cherry pick on production devel that I could use some help sorting out [15:45] All lines of log output:["PQM Cannot merge between different VCSsystems. [15:45] 'bzr+ssh://bazaar.launchpad.net/~brian-murray/launchpad/cherry-pick-bug-modifier'(pqm.Baz1_1Handler) and [15:45] '/home/pqm/archives/rocketfuel/launchpad/production-devel'(pqm.Bazaar2Handler) are different [15:45] bdmurray, are you sure 'bzr+ssh://bazaar.launchpad.net/~brian-murray/launchpad/cherry-pick-bug-modifier' exists? [15:46] bdmurray, pqm seems to think there is a bazaar 1 ("baz") repository at that location [15:48] jelmer: yes, I'm sure - would the fact that it is a private branch affect it? [15:48] bdmurray: probably - is the launchpad pqm able to access that branch? [15:48] bdmurray: I think it has to be subscribed. [15:48] jelmer: no, I'll give that a shot then [15:49] that's a rather unhelpful error message [15:51] bdmurray: It probably defaults to thinking there is a "baz" repository there if it can't find anything else. [15:58] jelmer: any idea how I could try resubmitting the branch? bzr lp-land is failing since it is a private branch [15:59] bdmurray: You can use "bzr pqm-submit", which is part of the bzr-pqm plugin. [15:59] Alternatively, you should be able to construct a PQM email manually. === salgado is now known as salgado-lunch === salgado-lunch is now known as salgado [16:38] bdmurray: you can subscribe launchpad-pqm to the branch and it should then work. [16:39] * noodles775 realises that's been tried. === Ursinha is now known as Ursinha-lunch [16:46] * rockstar physically relocates === salgado is now known as salgado-lunch === beuno is now known as beuno-lunch === benji is now known as benji-lunch === matsubara is now known as matsubara-lunch === Ursinha-lunch is now known as Ursinha === salgado-lunch is now known as salgado === al-maisan is now known as almaisan-away === beuno-lunch is now known as beuno === benji-lunch is now known as benji === matsubara-lunch is now known as matsubara [18:54] morning [19:02] hello [19:03] can anyone explain to me what is this http://is.gd/eH471 [19:03] ? [19:13] yes, we believe its a known outage at Google [19:14] https://bugs.edge.launchpad.net/launchpad-registry/+bug/624981 [19:14] <_mup_> Bug #624981: The Google Maps API server rejected your request [19:14] * sinzui is experimenting with a brute force use of featureflags to control gmp2 [19:15] I was thinking along the same lines. Seems like a perfect use for them. [19:16] lifeless, thanks [19:19] benji, the controller objects are not all in production. I am writing my code to what I see in a production test instead of what is in our development tree [19:19] mmm [19:20] sinzui: \o/ [19:20] I would suppose your result will be trivially translatable to the real thing once it's deployed. [19:20] yes [19:20] this is not the first time google has failed us [19:21] we talked about a way to toggle maps on/off last year === Ursinha-afk is now known as Ursinha [19:27] sinzui: a flag ? [19:28] ah [19:28] thats what gmp2 is [19:28] ? [19:29] lifeless, there are many versions of gmap. we use a specific version. We need to track their version. Having flags for the version we want is ideal [19:29] cool [19:29] I think feature flags can return a value [19:29] if they can't, its a small tweak to make them do so [19:29] or you can use different flags, one for each version [19:30] correct. We had some adventures shortly after we added maps and the versions changed. [20:31] if I run testr -t my_application, it seems that 62 tests match, so is there a way to specifically run the tests for my application? [20:32] testr -- -p packagename [20:33] it depends on your app structure really [20:33] lifeless: I'll give that a try [20:34] its runnin bin/test under the hood [20:40] lifeless: ./bin/test --help seems to indicate that -p is for progress (where testr -- -p package gets translated to xvfb-run ./bin/test --subunit -p package [20:41] cr3: the -- says 'to the right, pass onto the test process [20:41] cr3: I was told -p was package [20:41] perhaps its -m you want [20:41] anyway, all I'm saying is that its a bin/test problem for defining 'my application' [20:42] lifeless: heh, understood :) [20:42] I'm no expert on bin/test :P [20:44] what's the difference between lib/lp//scripts/tests and lib/lp//scripts, where both sometimes contain test_*.py files? [20:45] the first path is apparently recommended for containing tests as detailed here: https://dev.launchpad.net/Testing [20:45] actually, my mistake, scripts doesn't contain test_*.py files (not that I've seen anyways), my mistake. nevermind :) [20:46] :P [20:46] lifeless: I'm obviously writing my first test for launchpad [20:49] lifeless: how can I use testr to show the details about the tests that were run instead of just the number of tests? [20:50] apply one of jmls patches I haven't merged [20:50] or [20:51] subunit-ls < .testrepository/XX [20:51] testr will show failures always [20:52] lifeless: looking good, thanks! [20:58] is there a document on dev.launchpad.net for modifying the database schema? I see lots of files under the database/schema directory and I could tentatively create a patch with a sequentially numbered filename, but I'm not sure that'll make everyone happy [20:59] cr3, https://dev.launchpad.net/PolicyAndProcess/DatabaseSchemaChangesProcess [20:59] nevermind, found the answer in the README file conveniently located under that directory [20:59] (from database/schema/README) [20:59] :) === jcsackett is now known as jcsackett|afk === jcsackett|afk is now known as jcsackett [22:23] I'm running preliminary tests and would like to access the database, so I'm using getUtility(IStoreSelector).get(MAIN_STORE, DEFAULT_FLAVOR). but, when I try to retrieve rows, I get: ProgrammingError: permission denied for relation result [22:23] might there be an obvious reason for this? [22:31] in a test, setting a breakpoint with pdb.set_trace() doesn't seem to show anything when running testr. should this be done otherwise? [22:33] you haven't granted permission in security.py [22:33] do that [22:33] cr3: are you running as a special db user do you know? [22:33] and run make schema again [22:33] aha! nevermind that last question, running ./bin/test directly and not piping to testr load works fine [22:33] cr3: ?! thats new, and I have no idea how that could be happening [22:34] james_w: I looked around for special db privileges and didn't find anything, but it certainly feels that way since I'm getting an error from storm execute [22:35] cr3: take a look at database/schema/security.cfg [22:35] there is where the permissions for various tables are granted to various users [22:35] if you aren't running a script in your test then it is likely that you are just running as the "launchpad" user === matsubara is now known as matsubara-afk [22:36] james_w: crap, I didn't get to step 13 yet in the database policy. my bad, thanks! [22:36] add public.result = SELECT, INSERT to [launchpad_main] and you should be good to go [22:37] james_w: awesome, I was just wondering how to prevent UPDATE and DELETE on the table so that solves another problem :) [22:38] james_w: while I have your attention, is there a special user with autocommit isolation level? [22:38] don't know [22:38] I'll grep around, that should be easy to find if such a user at least exists [22:38] you probably don't want that if you are running the webapp context are you? [22:39] in some cases, it is desirable to run some statements that way [22:39] well, you have transaction.commit() [22:39] you may want to defer some of your processing in to a script outside the webapp context [22:39] * james_w -> dinner [22:40] maybe I'll have a function that checks for the conflict exception, whatever that might be. lots of exploration to do, we'll see [22:44] security.cfg did the trick, thanks folks!