[12:03] <kiko> so tomorrow morning
[12:03] <carlos> I will wake up early tomorrow
[12:03] <carlos> so that should work too
[12:04] <carlos> kiko, thank you very much!
[12:04] <carlos> good night!
[12:21] <Seveas> does malone offer other output formats beside html?
[12:21] <Seveas> or is such a feature planned?
[12:22] <kiko> Seveas, mail?
[12:23] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: Fix up duration of build job in progress, correct trivial bug in builder-index page template. r=salgado (patch-2499: daniel.silverstone@canonical.com)
[12:23] <Seveas> no, xml :)
[12:23] <Seveas> something like bugzillas xml would help
[12:23] <Seveas> I'm porting the supybot bug reporting plugin to malone, but currently the plan is to have a massive regex to filter out the info
[12:24] <Seveas> is such a thing available/planned/not considered yet/may be considered?
[12:25] <bradb> somewhere between planned and not considered, i think :)
[12:25] <Seveas> ok :)
[12:26] <bradb> we'll do XML and RSS feeds like everybody else, i'm sure. not a priority atm though.
[12:26] <Seveas> then i'll hope the output format of malone does not change too heavily too often 
[12:26] <Seveas> thanks for the info bradb 
[12:26] <ajmitch> bradb: xml-rpc interface will be for all functionality of malone, correct?
[12:26] <bradb> Seveas: the emails might still change pretty significantly. well, so might the web UI.
[12:27] <Seveas> k, then I/#ubuntu-desktop will have to live with the fact that the plugin will break once in a while :)
[12:27] <bradb> ajmitch: too early to say, but my guess would be probably, more or less
[12:29] <bradb> Seveas: yeah, sorry, it's possible that it could hurt. after ubuntu main has been on it for 6 months or a year, things will become more predictable. right now, there's just a whole lot of guessing about how people might use the UI, so it's hard to say what will be adjusted still.
[12:32] <Seveas> bradb, that's ok, it was no complaint. It's great to see launchpad getting shape, now if only the hackergotchis would work :)
[12:32] <bradb> Seveas: perms problem, you mean?
[12:33] <Seveas> yeah
[12:33] <Seveas> I can edit it for a team I created, but not my own
[12:35] <Seveas> There are some other issues I collected, I'll file them in malone (if not there yet) tomoorow-ish
[12:37] <bradb> Seveas: if you're interested in the hackergotchi problem, you might want to subscribe to bug #2501
[12:56] <kiko> bradb, what are null bugtasks, again?
[12:57] <kiko> bradb, matsubara has a fix, the test has just been a bit involved to run
[12:57] <bradb> kiko: a bugtask for which no actual bugtask exists (so that you can, e.g., view a bug in a context in which it's not yet been reported.)
[12:58] <kiko> bradb, a bug without a bugtask in the context being currently viewed. ah, ok.
[12:58] <bradb> i.e. it's a null object implementation of IBugTask
[12:58] <bradb> right
[01:10] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: [r=salgado]  make upstream tasks be wiki-mode (patch-2500: brad.bollenbach@canonical.com)
[01:11] <kiko> salgado, https://launchpad.net/errors/showEntry.html?id=1127776181.690.431611962761
[01:12] <kiko> this was a person who followed a link from email
[01:12] <salgado> this was a person who found the +login page on shipit
[01:13] <kiko> no
[01:13] <kiko> look at the referrer
[01:13] <salgado> kiko, he found the +login page and then tried to register on there
[01:15] <kiko> how did he get there from the referer?
[01:15] <bradb> later dudes
[01:17] <salgado> kiko, if you reach the +login page and try to register there, the email sent to you will have a link to https://shipit.ubuntu.com/token/<whatever>, because the application_url on that page is https://shipit.ubuntu.com
[01:17] <salgado> (I mean, if you reach the +login page on https://shipit.ubuntu.com/)
[01:18] <salgado> if you follow the link to launchpad.net/+login and register there, everything will be fine
[01:19] <kiko> salgado, how did the guy find the +login link? it should be pretty easy, we're getting dozens of these in production.
[01:20] <salgado> I have no idea, but this is why we overwrote all that error pages for shipit
[01:20] <salgado> to make sure people won't get a link to the +login page if something goes wrong or they don't have permission
[01:21] <kiko> there must be a straightforward way..
[01:22] <kiko> salgado, just fyi -- even googlebot is finding it.
[01:22] <salgado> dammit
[01:22] <kiko> actually, false.
[01:22] <salgado> we need to talk to SteveA and find a way to disable it in shipit
[01:22] <kiko> <canonical.launchpad.systemhomes.ShipItApplication object at 0x2a9f142650>, name: u','
[01:22] <kiko> NotFound: Object: <canonical.launchpad.systemhomes.ShipItApplication object at 0x2a9f142650>, name: u'robots.txt'
[01:22] <kiko> interesting, eh? a single comma.
[01:24] <salgado> heh
[01:25] <kiko> salgado, here's an interesting one.
[01:25] <kiko> salgado, if you are logged in but don't log out, logging in again doesn't fail.
[01:25] <kiko> (even if you enter an invalid name and pass)
[01:27] <salgado> kiko, how do I get the login page if I'm already logged in?
[01:28] <kiko> salgado, use the back button?
[01:32] <salgado> kiko, well, I guess that's because the form-processing code identifies you're already logged in and do not process the form
[01:32] <kiko> ah, right, perhaps.
[01:33] <kiko> does anyone know when staging last refreshed?
[01:34] <salgado> no idea
[01:35] <kiko> a staging that is a week old is pretty useless
[01:35] <kiko> dammit
[01:37] <kiko> jamesh, ping?
[01:41] <lifeless> jbailey: ping
[01:43] <jbailey> lifeless: pong
[01:44] <lifeless> BAH
[01:44] <lifeless> jamesh: ping
[01:44] <lifeless> jbailey: unping
[01:44] <lifeless> :/
[01:44] <jbailey> lifeless: unpong
[01:44] <lifeless> lol
[01:44] <ajmitch> jbailey: you need to change your name, I think
[01:45] <jbailey> ajmitch: True.  I hear "Andrew" is a good one.
[01:45] <jbailey> Wouldn't that just make #chug confusing?
[01:45] <ajmitch> not at all
[01:46] <jbailey> Oh good.
[01:46] <jbailey> I'm glad you've settled that. =)
[02:22] <jamesh> kiko/lifeless: yeah?
[03:08] <kiko> jamesh, could you commit my patch attached to bug https://launchpad.net/malone/bugs/2540 -- ?
[03:19] <jamesh> kiko: sure.
[03:19] <kiko> thanks!
[03:19] <jamesh> kiko: once we can dump hoary, we can use the system gettext
[03:20] <jamesh> which should make these sort of problems a non-issue
[03:26] <kiko> hmmm
[03:26] <Kinnison> night all
[03:33] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: Make sure slave passes /dev/null as stdin to the subprocess. Rev launchpad-buildd to v7. r=spiv (patch-2501: daniel.silverstone@canonical.com)
[04:06] <dilys> Merge to rocketfuel@canonical.com/launchpad--production--1.34: Cherry pick patch-2493 into production 1.34 (patch-7: rocketfuel@canonical.com, christian.reis@canonical.com)
[04:54] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: [trivial]  Improve script exception output and fix noisy log test (patch-2502: stuart.bishop@canonical.com)
[05:35] <dilys> Merge to rocketfuel@canonical.com/launchpad--production--1.34: Cherry pick patch-2502 into production 1.34 (patch-8: stuart.bishop@canonical.com, rocketfuel@canonical.com)
[05:36] <jdub> jamesh: ping
[06:26] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: [trivial]  Add indexes to the ShipIt tables (patch-2503: stuart.bishop@canonical.com)
[06:56] <dilys> Merge to rocketfuel@canonical.com/pygettextpo--devel--0: [trivial]  disable csharp bindings in included gettext-tools package.  Fixes bug 2540 (patch-7: james.henstridge@canonical.com)
[07:26] <dilys> Merge to rocketfuel@canonical.com/launchpad--production--1.34: Cherry pick patch-2503 into production 1.34 (patch-9: stuart.bishop@canonical.com, rocketfuel@canonical.com)
[07:45] <sabdfl> drumroll...
[08:03] <stub> spiv: Can you poke the librarian to see if it survived the database outage?
[08:05] <spiv> stub: Ok.
[08:07] <spiv> stub: Seems ok.
[08:07] <stub> Cool ;)
[08:08] <stub> sabdfl: Have you seen the new bounty? https://launchpad.net/bounties/chquite
[08:12] <ajmitch> getting a system error now trying to view assigned bugs
[08:12] <ajmitch> url is https://launchpad.net/people/motu/+assignedbugs
[08:14] <stub> ajmitch: ta. I can fix that
[08:15] <ajmitch> thanks
[08:17] <stub> ajmitch: fixed
[08:18] <ajmitch> quick work :)
[08:25] <Seveas> Hey, the 'tabs are a few pixels off' bug got fixed last night :)
[08:25] <Seveas> nice
[08:28] <ajmitch> other ui oddities seem to have been introduced :)
[08:37] <ajmitch> malone uses a different from address for bug emails now? (submitters address)
[08:38] <moyogo> hey where do i go to translate launchpad itself?
[08:39] <sivang> moyogo: I think this is not yet available, but discussions are going to take place on that - stay tuned, this should come up sometime close. (anybody correct me if I'm wrong)
[08:53] <moyogo> sivang: there's version 7 in multiverse but it's broken
[08:53] <moyogo> sivang: i guess i'll have to use evince...
[08:53] <sivang> moyogo: huh? version of launchpad ?
[08:59] <jamesh> I don't think we've looked at setting up the i18n for Launchpad yet
[08:59] <jamesh> some parts of the code have strings marked, and other parts don't
[09:04] <sivang> jamesh: ah, so there's i18n data in multiverse for launchapd ?
[09:07] <jamesh> sivang: not to my knowledge.
[09:09] <sivang> jamesh: k, that's what I thought
[09:16] <Treenaks> I'm getting mail ("Your mail to Launchpad-bugs awaits moderation") when filing bugs in malone
[09:18] <ajmitch> Treenaks: yes, seems the from address for malone bugs has changed :)
[09:19] <bob2> hah
[09:19] <Treenaks> ajmitch: it also breaks people with SPF on their From: domains
[09:19] <Treenaks> +for
[09:19] <bob2> Treenaks: SPF doesn't have anything to do with From:
[09:19] <Treenaks> bob2: MAIL FROM: then
[09:19] <bob2> presumably LP uses it's own envelop-sender
[09:20] <Treenaks> bob2: it changed, yes
[09:20] <Treenaks> bob2: new behaviour breaks RFC 2821 section 3.10
[09:22] <bob2> that is messed up
[09:22] <Treenaks> bob2: already filed a bug
[09:49] <stub> BjornT: ping
[09:56] <BjornT> hi stub 
[09:56] <stub> You broke the bugs mailing list ;)
[09:57] <BjornT> nope, bradb did :)
[09:57] <stub> Or should I say, the changes to bugmail (set reply-to: rather than munging from:) breaks mailman - I've had to open the list.
[09:58] <stub> But I think this is mailman's problem rather than Malones
[09:58] <stub> (because I now have no way of saying 'let through emails from malone but hold everything else)
[09:59] <stub> I might send a use case to Barry
[09:59] <carlos> morning
[09:59] <BjornT> stub: can't mailman look at reply-to instead? or some other header?
[10:00] <stub> BjornT: Nope ;) It would be a simple change through. But currently, it can't.
[10:00] <stub> Or at least I can't find it
[10:02] <carlos> SteveA, I need a fast review to cherrypick into production before Stub leaves today. It's related to language packs
[10:02] <zyga> morning
[10:02] <zyga> hello carlos :)
[10:02] <carlos> zyga, morning
[10:03] <jamesh> stub: in the mailman admin interface, go to "privacy options -> spam filters"
[10:04] <stub> jamesh: I can't see how to use that though
[10:04] <jamesh> stub: you should be able to configure mailman to accept messages that match an arbitrary header using header_filter_rules
[10:04] <stub> jamesh: I need to say 'let through anything with this', but the spam filters say 'block anything with this'
[10:04] <SteveA> morning
[10:04] <SteveA> carlos: go for it
[10:04] <stub> jamesh: oh! I see
[10:05] <jamesh> stub: really? on the gnome installation, it has a box where you can enter a regexp, and a set of radio buttons (defer, hold, reject, discard, accept)
[10:05] <carlos> SteveA, carlos.perello@canonical.com--2004/launchpad--devel--0
[10:05] <stub> Yes - I didn't notice the radio buttons
[10:06] <stub> jamesh: They are regexps?
[10:06] <jamesh> stub: the text beside the textarea says "Spam Filter Regexp:"
[10:08] <SteveA> carlos: was that the bug about missing translations?  that sometimes the date isn't updated?
[10:09] <carlos> SteveA, right, translators update pofile outside Rosetta
[10:09] <BjornT> stub: how do i create a new dbuser for process-mail.py to use?
[10:09] <carlos> and the revision date is not updated
[10:09] <carlos> so Rosetta ignores that 
[10:09] <stub> BjornT: Add the relevant section to security.cfg, then somewhere in schema.xml to make it configurable
[10:09] <SteveA> so rosetta ignores the new translations because rosetta depends (currently) on the revision date in the po file
[10:10] <carlos> SteveA, right
[10:10] <carlos> SteveA, I just relaxed it a bit and if the revision date is the same we have in our database we accept it 
[10:11] <SteveA> i see
[10:11] <SteveA> what happens if the revision date is a lot in the future?
[10:11] <SteveA> does that stop new translations from occuring?
[10:11] <carlos> we changed our db schema a couple of months ago so reimporting a pofile with the same data will not produce that we lose data 
[10:11] <carlos> SteveA, yes
[10:12] <BjornT> stub: ok, thanks
[10:13] <SteveA> carlos: is that fixed in this patch i'm reviewing?
[10:14] <carlos> SteveA, yeah, that's the point behind that patch
[10:14] <SteveA> ok
[10:15] <SteveA> baz is still working on it
[10:15] <carlos> SteveA, it has also the karma change you cherrypicked last week
[10:16] <SteveA> ok, great
[10:16] <carlos> but it's missing the update you did
[10:16] <carlos> I'm getting the production branch to apply that change too
[10:16] <SteveA> i sure hope bzr comes along soon... baz is getting unbearably slow
[10:19] <stub> BjornT: I'll turn off that deprecation warning on production
[10:20] <BjornT> stub: cool. i'll fix it anyway since i already started
[10:31] <jamesh> SteveA: I think I've got the test suite warning issues sorted -- they weren't due to out-of-process tests
[10:34] <SteveA> jamesh: really?  what was it?
[10:34] <SteveA> is there a tree i can look at?
[10:37] <jamesh> SteveA: haven't mirrored the changes yet
[10:37] <SteveA> ok
[10:37] <jamesh> SteveA: the logging module contains internal state (the configuration of the loggers), and some of the tests were reconfiguring the logger
[10:38] <SteveA> ah
[10:38] <jamesh> so running individual tests wouldn't cause the bug, but running them in sequence would
[10:38] <jamesh> s/bug/output/
[10:39] <SteveA> we could do with a way of registering subsystems to have their state checked and/or cleared between tests.
[10:39] <SteveA> there is something like this for zope3 tests, but i think it is rather messy
[10:39] <ddaa> Grrrr!
[10:39] <SteveA> this could be done with a root-level event subscriber, very simply
[10:39] <ddaa> lifeless: svn import is crap
[10:40] <ddaa> samba (on hoover) and videolan-vlc (on roomba) repeatedly fail with out of memory :(
[10:41] <ddaa> lifeless: need your help about how to diagnose fix the leak
[10:42] <ddaa> Short of  a better idea, I'll try implementing some form of batched fetching (based on svn co) in the hope that it would reduce the problem, but without any proof beforehand that it would.
[10:43] <jamesh> SteveA: for doctests, I wonder if it would be better to configure the logging subsystem to send messages to stdout
[10:43] <jamesh> so that people would need to account for the messages in their tests
[10:43] <SteveA> jamesh: yes, i think that would be good.
[10:47] <stub> jamesh: I fixed some test noise from the librarianformatter tests - log messages were bubbling up to the root logger.
[10:49] <jamesh> stub: I found the doctests in scripts/logger.py were one of the culprits -- they left the root logger configured with the LibrarianFormatter handler attached
[10:50] <stub> jamesh: yes. I've landed a fix that just set 'propagate=False' on the custom logger
[10:50] <stub> so that might conflict with your work
[10:51] <jamesh> stub: nope.  I didn't find any problems with those tests (I did check if they were one of the culprits
[10:56] <stub> jamesh: erm... yer. the doctests in scripts/logger.py are the ones I mean. The propagate=False fix means you don't have to mess with the root logger.
[10:57] <stub> (at least the noise I fixed didn't)
[10:58] <jamesh> stub: the tests in the docstrings for logger_options() and logger() were the ones that were messing with the root logger.
[10:59] <jamesh> stub: my other fix was to configure the root logger again in FunctionalTestSetup.setUp()
[10:59] <stub> ok
[10:59] <stub> You configure it to raise an exception if used ;) ?
[11:00] <stub> Probably too much code to fix if you did that...
[11:00] <jamesh> no
[11:00] <jamesh> it was already being configured to put all the log messages in a StringIO instance
[11:00] <stub> oh... never knew that ;)
[11:04] <jamesh> hmm
[11:06] <jamesh> if you don't call close() on a logging handler, it never gets freed until you call logging.shutdown()
[11:08] <SteveA> carlos: you have new mail (in fact, two new mails, seeing as i forgot to attach the review the first time)
[11:27] <sivang> *launchpad
[11:27] <sabdfl> carlos: so, do we not bring Universe packages into Rosetta?
[11:28] <SteveA> sivang: the spec tracker is a bit like that.
[11:35] <carlos> SteveA, ok, thanks
[11:35] <SteveA> carlos: what's the answer to what mark asked above?
[11:36] <ddaa> SteveA: maybe you can help, do you have any general strategy to help tracking memory leaks in python, that are likely to originate from an extension module?
[11:36] <SteveA> ddaa: like, C code?
[11:36] <ddaa> C++ actually, but yeah
[11:36] <SteveA> here's what i do
[11:37] <SteveA> 1. get a debug build of python
[11:37] <SteveA> that's more than just symbols, it has extra code in to track memory usage / garbage / etc.
[11:37] <SteveA> 2. build your extension against that python
[11:38] <SteveA> 3. run the test suite using the options to run it in a loop, and output object counts after each cycle
[11:38] <SteveA> 4. narrow down the set of tests you need to run in a cycle while still seeing certain objects increase in number
[11:38] <SteveA> 5. write your own test to make it even more minimal
[11:39] <SteveA> 6. scrutinize this minimal code path to check for problems in the extension code along that path
[11:39] <ddaa> that sounds very cumbersome
[11:39] <SteveA> 7. give up and ask jamesh to look at, because he's proved himself very good and finding such problems   ;-)
[11:40] <SteveA> ddaa: it's fiddly, but it uses divide-and-conquor to get the job done
[11:40] <Kinnison> hihi
[11:40] <BjornT> is there a way to connect with a specific dbuser in a doctest?
[11:40] <ddaa> so, I guess the first step is running the test suite in a loop and checking that memory usage grows, right?
[11:42] <SteveA> ddaa: yes, or a subset of the test suite
[11:43] <ddaa> yeah, of course
[11:43] <Kinnison> spiv: Twisted does longest-prefix-first matching when resolving a request for a server which has an object attached to /foo/ and one attached to /, yes?
[11:43] <SteveA> there are specific options to help with this in the zope test runner
[11:43] <SteveA> look at launchpad's test.py perhaps
[11:43] <ddaa> SteveA: well, cscvs is not zope stuff :)
[11:43] <ddaa> but I think I get the idea
[11:45] <spiv> Kinnison: you mean twisted.web?  No.
[11:45] <spiv> It does it pretty much the same way as Zope: path segment by path segment.
[11:45] <ddaa> SteveA: is there a python functionality to monitor memory usage process-wise (i.e. the leak might be in C++ structures that the gc is not aware of)?
[11:45] <spiv> At least, with the default Resources it does.  You could implement all sorts of crazy stuff if you really wanted ;)
[11:45] <Kinnison> spiv: so it does shortest-first?
[11:46] <SteveA> ddaa: i don't know.  ask jamesh.
[11:46] <spiv> Kinnison: The question doesn't really make sense.
[11:46] <Kinnison> okay, allow me to give you context
[11:46] <stub> BjornT: setup = LaunchpadTestSetup(); setup.dbuser = 'foo'; setup.setUp(); .......; setup.tearDown()
[11:46] <ddaa> jamesh: is there a python functionality to monitor memory usage process-wise?
[11:46] <Kinnison> the buildd slave has an XMLRPC interface on /
[11:46] <SteveA> Kinnison: it treats it as a hierarchy, not as an opaque string
[11:46] <Kinnison> I want to add a simple HTTP interface to the filecache on /filecache/
[11:46] <spiv> Kinnison: For a path /a/b/c, it'll do root.getChild('a'), then do .getChild('b') on the result of that, and so on.
[11:47] <stub> BjornT: That should also work with other subclasses, like LaunchpadZopelessTestSetup. LaunchpadFunctionalTestSetup will be problematic though.
[11:47] <spiv> And whereever that stops, it calls render.
[11:48] <stub> BjornT: Although if this is for process-email.py, you might want to use the subprocess module to invoke the actual script so that is tested too (see the test_update_stats.py for example)
[11:48] <spiv> Kinnison: The short answer is: root.putChild('filecache', static.File('blah/blah/filecache'))
[11:48] <Kinnison> root?
[11:49] <spiv> Kinnison: Whatever resource you're passing to server.Site
[11:50] <BjornT> stub: setup = LaunchpadTestSetup(); setup.dbuser = 'foo'; still makes the test pass
[11:51] <spiv> Kinnison: Hmm, XML-RPC is a bit funny, because some clients will hit "http://foo/RPC2" when you tell them the server is at "http://foo/RPC2".
[11:51] <spiv> Kinnison: Let me do a quick patch for you.
[11:51] <stub> BjornT: Does your test do initZopeless itself? If so, you need to change that to initZopeless(dbuser=foo)
[11:52] <BjornT> stub: using subprocess isn't an option here, since I need to look at the email notifications (i guess i could use subprocess for some tests, and call the function directly for some)
[11:52] <BjornT> stub: it did use the normal setup. i change it to do what you suggested, but it seems like the dbuser is ignored
[11:57] <stub> BjornT: How does your test get a connection? Or is it using LaunchpadZopelessTestSetup?
[11:57] <Kinnison> spiv: what will your patch do?
[11:58] <spiv> Kinnison: https://chinstrap.ubuntu.com/~dsilvers/paste/fileLVyrWj.html
[11:59] <spiv> Er, I should have said, 'hit "http://foo/RPC2" when you tell them the server is at "http://foo/"'
[11:59] <Kinnison> right
[11:59] <Kinnison> static.File()
[11:59] <Kinnison> that's a static file *root* yes?
[11:59] <spiv> Kinnison: Right.
[11:59] <Kinnison> coolio
[11:59] <Kinnison> thanks dude
[12:00] <spiv> (Or a single file.  It's happy either way)
[12:01] <BjornT> stub: it's currently using LaunchpadTestSetup, since it needs to look at the email notifications. is there any way of using LaunchpadZopelessTestSetup and have the sent email end up in test_emails?
[12:02] <Kinnison> spiv: was it you who told me to use SafeConfigParser instead of ConfigParser?
[12:02] <BjornT> stub: (the test also needs to use the same security policy as the web app)
[12:02] <spiv> Kinnison: Possibly.  I told someone to use it :)
[12:02] <Kinnison> probably me :-)
[12:03] <stub> BjornT: Hmm... that is working by accident I think. If you need to use the Z3 email stuff, that would require using LaunchpadFunctionalTestSetup to load the component architecture. And there is no way to change the user connected as in that case.
[12:04] <jamesh> ddaa: nothing really accurate (there was a google summer of code project to product a memory profiler, but I don't know the status)
[12:04] <stub> (working by accident because the CA is never actually torn down, which causes problems :-( )
[12:05] <spiv> jamesh, ddaa: You can apparently make valgrind work with Python.
[12:05] <ddaa> jamesh: I just mean something to answer simple question like "How much VM does this python eat?"
[12:05] <jamesh> ddaa: one trick you can use is the fact that instances of heap allocated classes (i.e. ones not defined in C extension modules) hold a reference to their class
[12:05] <jamesh> ddaa: poke around in /proc/$pid/status?
[12:06] <BjornT> stub: don't all system documentation tests use LaunchpadTestSetup by default?
[12:06] <stub> BjornT: I don't think we have any suitable fixtures for what you need at the moment. 
[12:06] <spiv> ddaa: In theory the resource module can do stuff like htat.
[12:06] <ddaa> jamesh: I think that would do the trick. I'm interesting in monitoring all process memory usage including what the GC does not know about
[12:06] <spiv> ddaa: I haven't usedit, though.
[12:06] <stub> BjornT: They use LaunchpadTestSetup and FunctionalDocFileSuite combined, which pretty much duplicates LaunchpadFunctionalTestSetup.
[12:07] <jamesh> ddaa: it'll be quite inaccurate though -- address space isn't returned to the system except for large allocations, so you'll be looking at the high water mark
[12:08] <ddaa> jamesh: I just want to confirm that "running that code causes a memory leak, somehow, somewhere"
[12:08] <jamesh> ddaa: okay.  it might be enough for that kind of thing.
[12:08] <ddaa> THEN I can start talking gc and figure out what kind of leak that is
[12:09] <spiv> ddaa: If you do want to interrogate the gc, http://twistedmatrix.com/users/spiv/countrefs.py may help.
[12:09] <ddaa> spiv: thanks
[12:09] <Kinnison> spiv: does static.File() provide content-length headers properly?
[12:09] <spiv> Kinnison: Yes.
[12:09] <ddaa> mh, no twisted crack in that, good
[12:10] <Kinnison> spiv: cool, ta
[12:10] <spiv> ddaa: Hah :)
[12:10] <BjornT> stub: hmm, i'll look into if i change change my test, otherwise i'll wait with changing the dbuser of process-mail until you implement such a text fixture ;)
[12:16] <carlos> SteveA, the answer to mark's question is: Rosetta is importing universe translations but we don't have language packs for universe so the translations will not reach the archive unless the developer in charge of that package adds it by hand
[12:16] <SteveA> are the MOTU aware of this?
[12:18] <ajmitch> not that I'm aware of
[12:18] <ajmitch> at least we didn't expect to have langpacks for universe
[12:20] <Kinnison> what module has the routine for joining bits of urls together?
[12:22] <SteveA> i mean, are the MOTU aware of how they can get translations out of rosetta for use in the packages?
[12:22] <ajmitch> SteveA: unless it's written up somewhere, I doubt that we are
[12:23] <carlos> SteveA, I think so, but not 100% sure
[12:23] <carlos> we had some mails changed with pitti and sabdfl some weeks ago about this
[12:23] <carlos> but nothing more
[12:24] <Kinnison> aha, urlparse.urljoin, nm
[12:25] <Kinnison> spiv: Given I have two file handles (well, one file, one urllib) what is the accepted way of pumping the contents of slave_file into out_file ?
[12:26] <SteveA> ajmitch, carlos, jordi: i think you guys need to talk about using rosetta for Universe sometime
[12:27] <carlos> ajmitch, We cannot use current language packs or they will grow a lot
[12:27] <ajmitch> something to talk about at UBZ
[12:27] <carlos> ajmitch, jordi I'm a bit busy atm....
[12:27] <Kinnison> stevea: Given I have two file handles (well, one file, one urllib) what is the accepted way of pumping the contents of slave_file into out_file ?
[12:27] <carlos> ajmitch, if it can wait, then yes, better at UBZ
[12:27] <ajmitch> carlos: of course, the universe translations need to be separate
[12:27] <ajmitch> depends if we want to try & get them into breezy
[12:28] <carlos> ajmitch, in the mean time, the only solution I see is to get all translations from Rosetta and add them as a patch to the package
[12:28] <ajmitch> which will be a lot of work if there's many translations, I think
[12:29] <ajmitch> we can try for some key universe apps
[12:29] <ajmitch> is there a quick guide for getting the translations out for a package?
[12:30] <ajmitch> s/out/out of rosetta/
[12:30] <carlos> ajmitch, every sourcepackage has an export/download link that let's you get a tarball with all translations
[12:31] <ajmitch> sounds easy enough for us then
[12:31] <ajmitch> if the source packages are imported for universe & translated
[12:32] <carlos> right
[12:32] <Kinnison> spiv, SteveA: never mind, I've stolen the loop from the librarian uploader
[12:34] <ajmitch> ddaa: how often are source imports processed for a product?
[12:35] <ddaa> ajmitch: syncs are done every day, initial imports are done whenever I come around to it.
[12:35] <ddaa> if a syncing branch appears out of date, I'm the right person to complain to. There are many failed syncs but I just do not have the resources to fix them.
[12:36] <ddaa> If you want an import done in reliable way, ask me. I'll tell you if I hit a roadblock.
[12:36] <ddaa> (or when it's done)
[12:36] <ajmitch> ddaa: right, gnuenterprise is sitting on testing - initial import failed a couple of weeks ago & svn url has changed since then
[12:36] <ddaa> I'll have a look right now.
[12:37] <ajmitch> hm, svn might still be a little broken upstream
[12:37] <ddaa> pleonasm
[12:37] <ddaa> what do you mean, actually?
[12:38] <ajmitch> gnu enterprise server shifted recently, I can't recall if anon svn is working correctly yet
[12:38] <ajmitch> I'll ask you about it when I know it is :)
[12:38] <ddaa> ajmitch: product description: I'm a bit puzzled
[12:38] <ddaa> you say that it's a meta-project, but you say there is a RCS to import
[12:38] <ddaa> that's contradictory
[12:39] <ddaa> GNOME, Mozilla are meta-projects, they do not have  a single RCS to import, they have several products instead
[12:39] <ajmitch> ok, the project has tools & a number of subprojects in the same svn
[12:39] <ddaa> such meta-projects are "projects" in launchpad parlance, specific projects that actually release something are products
[12:40] <Kinnison> elmo, lifeless: Any chance one of you can tickle pqm ?
[12:40] <ddaa> gnue/trunk ?
[12:41] <ddaa> ajmitch: you mean that for release management, security fixes etc, gnue is a single product?
[12:41] <ajmitch> there are a number of packages that get released with different versions
[12:41] <ajmitch> so I'd say not then
[12:41] <ddaa> so, how does the gnue/trunk relate to those packages?
[12:42] <ajmitch> each of them are in a directory under there
[12:42] <ddaa> ajmitch: are you a bit familiar with arch/bazaar?
[12:42] <ajmitch> a bit
[12:42] <ajmitch> I've mainly been using bzr
[12:43] <ddaa> same difference
[12:43] <ddaa> ajmitch: I hope I'm not annoying you by appearing nitpicky
[12:44] <ajmitch> not at all
[12:44] <ddaa> bah, their SVN instruction page is broken...
[12:44] <ajmitch> everything was recently shifted in the last week
[12:45] <ddaa> You can browse the subversion tree online.  -> 404 with a python traceback
[12:45] <ddaa> ajmitch: I have the impression that either:
[12:46] <ajmitch> this was the first new product I did, to learn how it goes
[12:46] <ddaa> 1. the actually trees to import are gnue/trunk/<something>
[12:46] <ddaa> or 2. the gnue release process is a bit broken
[12:46] <ajmitch> importing separate trees & having separate products for each tool is better
[12:47] <ddaa> I mean, either the svn repo hosts several essentially distinct projects, that are packaged independently, and should be imported separatedly, or it does not.
[12:48] <ajmitch> they are separately released & packaged
[12:48] <ddaa> a problem could be for example if the build system relies on stuff at the root, or whatever
[12:48] <ddaa> or if commits are frequently made that spans several projects
[12:48] <ddaa> I'm just trying to grasp what is the model there, to find the best representation for it in launchpad.
[12:50] <ddaa> okay... so let's make them different products
[12:50] <ajmitch> gnue-common does have a script to setup wrapper scripts for all the tools, for running from an svn checkout
[12:50] <ajmitch> that would be the only small complication
[12:50] <ddaa> what are you using to build deb packages?
[12:50] <ddaa> the full checkout or partial checkouts?
[12:51] <ajmitch> using released tarballs at the moment
[12:52] <ajmitch> they use distutils, so releases can be made separately
[12:52] <ddaa> are the release tarballs are made essentially of partial checkouts?
[12:53] <ddaa> The issue here is figuring out the right granularity for malone, rosetta, hct, etc. which all talk in term of source packages.
[12:53] <ddaa> ajmitch: so, one project and multiple products?
[12:53] <ajmitch> by using just the directory for the tool
[12:53] <ajmitch> yes, I'd think so
[12:54] <ddaa> ajmitch: BTW where is the SVN now, it looks like the web page is out date, so I cannot find it.
[12:54] <ajmitch> http://svn.gnuenterprise.org/gnue/trunk is current
[12:55] <ddaa> yeah, totally partial checkouts :)
[12:55] <ajmitch> svn allows for plenty of creativity
[12:58] <ddaa> ajmitch: I'll make you a gnuenterprise project, from the current product details, and assign the existing product to this project, then we can tweak the product to be a real product. You'll need my help to change its name to whatever you want it to be (e.g. gnue-appserver)
[12:58] <ajmitch> alright, thanks
[01:00] <eruin> how do I report this? https://launchpad.net/distros/ubuntu/breezy/+sources/gdm/+pots/gdm/nb/+translate
[01:00] <eruin> "this page and locale in particular constantly spits out system errors" ? :)
[01:01] <SteveA> jamesh: ping?
[01:03] <ddaa> ajmitch: here it is, https://launchpad.net/projects/gnuenterprise/
[01:04] <ddaa> ajmitch: start replacing the details for https://launchpad.net/products/gnuenterprise and I'll take on after that
[01:10] <segfault> is there any problem with LP?
[01:12] <SteveA> working for me
[01:12] <carlos> stub, hi, I have a patchset to cherrypick into production
[01:12] <SteveA> are you seeing a problem?
[01:12] <segfault> can you try to access the main url? https://launchpad.net
[01:12] <carlos> stub, which branch is being used?
[01:12] <segfault> System error here.
[01:12] <eruin> I get system error on main
[01:13] <segfault> not only me, then
[01:13] <stub> carlos: launchpad--production--1.34. Just land it in launchpad--devel--0 and I'll pick it from there
[01:14] <segfault> even this url doesn't work https://launchpad.net/distros/ubuntu/breezy
[01:14] <carlos> stub, ok
[01:15] <zyga_> why did karma got re-set recently?
[01:15] <zyga_> carlos: anyone?
[01:15] <carlos> zyga_, because a bug I introduced, we added too much karma in some situations
[01:16] <carlos> and we were not able to remove only the bad one
[01:16] <carlos> so we choose to reset it now that it's fixed
[01:16] <zyga_> carlos: awww
[01:16] <zyga_> carlos: I lost all my translation karma and I'm a karma whore ;)
[01:16] <carlos> zyga_, you know how to fix that.... do more translations!
[01:16] <carlos> :-P
[01:17] <carlos> zyga_, anyway, all people had that "problem"
[01:17] <carlos> so you didn't lose any place in the karma ranking ;-)
[01:17] <ajmitch> carlos: they did? I don't think I lost karma :)
[01:17] <zyga_> carlos: yeah, I'll prepend space to everything, upload, remove the space, upload,..., *PROFIT*
[01:17] <carlos> ajmitch, only Rosetta's karma was removed
[01:18] <carlos> zyga_, you are sooo evil sometimes....
[01:18] <carlos> :-)
[01:18] <ajmitch> carlos: oh that's alright then :)
[01:18] <zyga_> carlos: I should be a manager
[01:20] <carlos> zyga_, :-P
[01:20] <carlos> stub, merge request sent
[01:20] <carlos> stub, it includes to the patch that you cherry picked on Friday for the karma fix, so perhaps you will get a conflict
[01:21] <stub> carlos: it will conflict then. 
[01:21] <carlos> stub, after that code is on production, we need to execute an UPDATE on production DB, will send you by email with the final patchset
[01:21] <carlos> stub, it's exactly the same patch so the conflict should be easy to solve
[01:21] <stub> ok. 
[01:32] <segfault> can anyone please restart LP?
[01:32] <segfault> :P
[01:34] <SteveA> segfault: why?
[01:34] <segfault> https://launchpad.net/distros/ubuntu/breezy
[01:34] <segfault> gives a system error.
[01:35] <SteveA> carlos: https://launchpad.net/errors/showEntry.html?id=1127820881.540.0383009149773
[01:35] <SteveA> segfault: that is strange.  it doesn't do so for me.
[01:35] <SteveA> can you try again please?
[01:36] <carlos> SteveA, ?
[01:36] <segfault> i tried, same error
[01:36] <SteveA> carlos: /distros/ubuntu/breezy/+lang/fr
[01:37] <spiv> SteveA: Hmm, I'm getting a system error from the front page!
[01:37] <SteveA> is giving SQLObjectMoreThanOneResultError: 2 rows retrieved by selectOneBy  errors
[01:37] <SteveA> spiv: hmm, same error
[01:37] <spiv> From Expression: <PathExpr standard:'view/ubuntu_languages'> in root-portlet-rosetta.pt
[01:37] <spiv> , due to the same error...
[01:37] <segfault> shouldn't it be nice to see the error dump?
[01:38] <carlos> SteveA, yeah, but I don't understand why...
[01:38] <spiv> SteveA: Really quick hack would be to relax selectOneBy, I guess :/
[01:38] <SteveA> spiv: having selectOneBy give the SQL query in the error would help
[01:38] <spiv> SteveA: Hmm.
[01:39] <SteveA>  Module canonical.launchpad.database.distrorelease, line 230, in getDistroReleaseLanguage
[01:39] <SteveA> languageID=language.id)
[01:39] <SteveA> anyway
[01:39] <SteveA> that's the thing
[01:39] <SteveA> i guess we have duplicate language ids
[01:39] <SteveA> or that query is wrong
[01:39] <carlos> ok
[01:39] <carlos> we are missing a constraint there
[01:40] <SteveA> so, i expect people who see errors are those who have certain language settings in their browsers
[01:40] <SteveA> that's why i can't see the error
[01:40] <SteveA> segfault: thanks for reporting the problem.
[01:41] <SteveA> spiv: see the same thing on staging?
[01:41] <segfault> no problem
[01:42] <spiv> SteveA: Hmm, no, I don't.
[01:42] <spiv> Very odd.
[01:42] <ddaa> ajmitch: I'd like to make your imports, really... but I need your product description stuff before :)
[01:42] <SteveA> spiv: do you have read access to the production DB?
[01:42] <lifeless> he does
[01:42] <lifeless> via auth server
[01:42] <spiv> At least some bits of it.
[01:42] <ajmitch> ddaa: sorry, been talking with gnue upstream :)
[01:42] <spiv> And via the librarian.
[01:42] <carlos> stub, do you have a minute and we solve that missing constraint?
[01:42] <lifeless> more can be enabled if needed
[01:42] <spiv> I don't know if that gives me read access to all tables?
[01:43] <lifeless> check security.conf
[01:43] <ajmitch> ddaa: discussing whether they could use malone as their bug tracker
[01:43] <ddaa> we'd certainly love that
[01:43] <stub> carlos: sure
[01:44] <ddaa> ajmitch: but your description suggest they're hardcore GNU, so the fact that Malone code is not released might be a problem.
[01:44] <carlos> ok
[01:44] <ajmitch> ddaa: gnu project, malone not free software.. so obstacles there
[01:44] <ddaa> ajmitch: what I said :)
[01:44] <ajmitch> you just typed faster :)
[01:46] <ddaa> Much as I respect him, we surely do not want RMS to be out to get us because some GNU maintainers are depending on Launchpad.
[01:46] <lifeless> ddaa: ?
[01:46] <ajmitch> lifeless: if a gnu project decided to use malone for its bug tracker
[01:46] <ddaa> lifeless: svn import code is leaky, preventing import of samba on hoover. Out of memory. Same condition for vlc on roomba.
[01:46] <ddaa> lifeless: any clue?
[01:48] <lifeless> ddaa: is it an error in our code, or are we dosing the svn server ?
[01:48] <ddaa> local OOM
[01:48] <lifeless> ouch
[01:48] <lifeless> what rev # ?
[01:48] <ddaa> variable
[01:48] <ddaa> at least on hoover
[01:48] <lifeless> well, pick a lower number, and use spivs object refcounting magic to print the refcounts
[01:49] <lifeless> iz possibly apache portable runtime leakage too, might want to valgrind a very small import
[01:49] <ddaa> on samba/hoover it's around 8-10k revisions, signal 6 or OSError
[01:50] <ddaa> (in os.fork)
[01:50] <spiv> ddaa: SIGABRT?  That's exciting :)
[01:51] <ddaa> videolan fails with OSError in os.fork at about 2-3k revisions
[01:51] <ddaa> spiv: I suspect it's OOM killer kiking in when other syncs run and want some memory
[01:52] <ddaa> lifeless: what does the apache portable runtime have to do with that?
[01:53] <ddaa> lifeless: I've been trying to loop the svn_oo test suite to see if I can reproduce the leak there, but on the second run I get plenty of failures with
[01:53] <ddaa> ClientError: Unable to open an ra_local session to URL
[01:53] <ddaa> Unable to open repository 'file:///home/david/home/devel/canonical/cscvs/,,onefilereporo/modulefile1
[01:54] <ddaa> I suspect your cached repo magic is brittle...
[01:54] <carlos> segfault, could you try it now?
[01:57] <segfault> carlos: niiice. works. thanks!
[01:59] <carlos> segfault, stub fixed it, so say thanks to him
[01:59] <carlos> stub, thanks
[02:04] <spiv> SteveA: want to review andrew.bennetts@canonical.com/sqlobject--selectOne-better-exception-message--0?  Try it with e.g. PYTHONPATH=lib python -c "from canonical.database.harness import *; EmailAddress.selectOneBy(personID=22)"
[02:04] <SteveA> spiv: about to go for lunch.  can you mail me a request to review, and i'll do it when i get back?
[02:04] <spiv> SteveA: Sure.
[02:08] <carlos> pqm is stalled
[02:08] <carlos> lifeless, ^^^
[02:17] <dilys> Merge to rocketfuel@canonical.com/launchpad--production--1.34: Turn off initZopeless warnings (patch-10: stuart.bishop@canonical.com)
[02:26] <salgado> stub, around?
[02:26] <stub> yes
[02:26] <ddaa> OMG, svn_oo is such a fat pig! VmSize after svn_oo test suite iterations: 0: 130808 kB 1:  225244 kB 2: 319668 kB
[02:27] <salgado> stub, can we make everything under /shipit to not raise RequestExpired()?
[02:27] <ddaa> it's not a leak, it's a torrent!
[02:28] <SteveA> salgado: not really.  talk to me about it when i come back from lunch please.
[02:28] <salgado> SteveA, sure
[02:28] <stub> salgado: What isn't working?
[02:28] <SteveA> stub: go to sleep ;-p
[02:29] <stub> I can login to shipit and view my order
[02:29] <salgado> stub, some (all) queries to display pending/approved orders
[02:29] <salgado> yes, for the users themselves it's not a problem. the problem is for Marilize
[02:30] <stub> salgado: Can you confirm it is happening in the last few hours? I added a load of indexes to the shipit tables as part of the rollout
[02:31] <salgado> well, I just tried it
[02:31] <stub> salgado: Otherwise we need to fix the screens so I know what the queries it is trying to do are
[02:31] <stub> erm... I mean I will need urls and such so I can determine what the queries are.
[02:32] <stub> Or just some tracebacks if they are included
[02:32] <salgado> stub, ShippingRequestSet.searchCustomRequests() and ShippingRequestSet.searchStandardRequests()
[02:33] <stub> I'm more interested in the sql
[02:33] <salgado> stub, there you'll see the sql
[02:34] <salgado> stub, the traceback I got in /errors doesn't have the SQL that caused the problem
[02:35] <stub> The query in ShippingRequestSet.searchCustomRequests needs to be rewritten as a group by 
[02:35] <stub> as does the searchStandardRequests
[02:36] <salgado> hmmm. I have the feeling I don't know how to do that
[02:40] <stub> I'm trying to workout what ShippingRequestSet.sarchStandardRequests is actually trying to do? 
[02:42] <salgado> stub, we have some standard options of orders for people to choose. that method will query all orders whose number of CDs match the number of CDs of the standard options
[02:44] <Kinnison> If anyone here fancies an easy review (and with tests this time -- wahey) take on my buildd-slave patch I just added to the general queue on PendingReviews
[02:44] <stub> That doesn't explain the options (status, omit_cancelled, standard_type)
[02:45] <stub> ok
[02:45] <salgado> you can filter the list using these options
[02:47] <stub> And why do you want to return all custom requests, or all standard requests? Those lists are going to be huuuuge
[02:49] <stub> I guess you need a list of all pending (or at least the first few)
[02:50] <salgado> that's what she does --she queries all pending requests, most of the time
[02:50] <salgado> sometimes she queries the approved ones, to make changes
[02:50] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: [trivial]  Even spiv can get it wrong. rev launchpad-buildd to v8, ensure we pass .fileno() not the file object. (patch-2504: daniel.silverstone@canonical.com)
[02:50] <salgado> anyway, today I was going to add more filter options so she can get useful results
[02:51] <stub> ok. I think she will need them ;)
[02:51] <salgado> yes, she will. there's a bug open on that
[02:54] <stub> salgado: Are you batching the results?
[02:54] <salgado> stub, not yet
[02:55] <stub> salgado: I think that is the problem then. The queries arn't actually that slow (I need to do some more complete tests though). But rendering a list of 2000 shipments in tal will take ages.
[02:56] <salgado> for sure that's part of the problem, but I don't think that's the main cause
[02:56] <salgado> stub, can you query all orders with only 5 x86 CDs on production?
[02:57] <stub> Can you give me the SQL?
[02:57] <segfault> when will the blue and purple strings be merged to the packages?
[02:58] <salgado> stub, nevermind. there must be thousands of orders with only 5 x86 CDs
[02:59] <salgado> stub, I tried searching for another standard option and got 670 results. works just fine
[03:00] <stub> 4166 if my query is correct
[03:00] <salgado> it probably is
[03:00] <salgado> that was a bad example I choose
[03:01] <Kinnison> *g*
[03:05] <stub> salgado: ok. I now think the searchStandardRequests SQL is fine. I can add some indexes to help.
[03:07] <salgado> stub, and the searchCustomRequests is fine too? 
[03:07] <salgado> I'm adding the batching already
[03:08] <stub> one at a time ;)
[03:10] <mpt> stub: How do I convert the current.sql password field from a sampledata account into an actual password? Alternatively, is there a list somewhere of sampledata accounts and passwords? (the only ones I know are foo.bar and carlos)
[03:11] <stub> mpt: You can't reverse them. Ask here and maybe someone remembers.
[03:11] <mpt> heh
[03:11] <stub> mpt: If there is a page test that uses them, the Authorization: line can be reversed
[03:12] <mpt> ok, so how do I do that?
[03:12] <carlos> mpt, I think all accounts have the password 'test'
[03:12] <mpt> Authorization: Basic amVmZi53YXVnaEB1YnVudHVsaW51eC5jb206dGVzdA==
[03:13] <carlos> at least daf, carlos, foo.bar and non-priv are using that one
[03:13] <stub> salgado: if shipit starts supporting more distroreleases or flavours, we will need to redo these indexes
[03:14] <stub> mpt: 'amVmZi53YXVnaEB1YnVudHVsaW51eC5jb206dGVzdA=='.decode('base64')
[03:14] <Kamion> I'd like to report some Rosetta bugs, but https://launchpad.net/products/rosetta gives a system error
[03:14] <salgado> stub, no problem, I guess. we'll need to redo a lot of things to support that
[03:14] <Kamion> known?
[03:14] <mpt> carlos: jdub doesn't
[03:15] <carlos> mpt, just change it to test and fix any test that fails or look at those test to know the right password....
[03:16] <SteveA> there's a list of accounts and passwords in the sampledata
[03:16] <mpt> carlos: Yes, I already asked how to get the password from the Authorization field
[03:16] <carlos> Kamion, it works here...
[03:16] <mpt> jdub's causing 4 test failures and I don't know why
[03:16] <SteveA> i think (although i haven't checked) that it is possible to use plaintext names and passwords
[03:16] <SteveA> and if not, i have a branch waiting to be merged for that
[03:16] <Kamion> oh, works fine now
[03:16] <Kamion> how strange, it definitely failed earlier - oh well :)
[03:17] <carlos> Kamion, from what I see seems like you got a timeout
[03:17] <carlos>  RequestExpired: The current request has expired  
[03:18] <Kamion> I'm using an HTTP proxy over a VPN, I guess it could be misbehaviour there
[03:18] <mpt> stub: thankyou
[03:19] <mpt> huh, that does return "test", but trying that login doesn't work on localhost
[03:24] <stub> salgado: Even without the indexes, the queries are running fast. I can speed them up, but I don't think they are a significant part of the rendering time.
[03:24] <BjornT> SteveA: in our version of zope you can't use plaintext names and passwords, but you can in 3.1
[03:25] <SteveA> BjornT: thanks
[03:25] <salgado> stub, if the queries are running fast, then let's see if the batching solves the problem
[03:26] <SteveA> BjornT: http://localhost:8086/distros/ubuntu/hoary/+sources/evolution/+bugs
[03:26] <kiko-zzz> salgado, we need batching there -- definitely
[03:26] <kiko-zzz> and searching
[03:26] <SteveA> that page looks very empty
[03:26] <kiko> hey ho
[03:27] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: [trivial]  Fix kiko's easy-to-miss misspelling (patch-2505: daniel.silverstone@canonical.com)
[03:28] <stub> salgado: So... should a shipment request change from being a custom request to a standard request if someone goes and changes the standard request definitions?
[03:28] <kiko> Kinnison, what did I do now? :)
[03:29] <stub> salgado: If not, we should be storing a 'standard' boolean in shippingrequest table, that would also make these queries trivial
 salgado: So... should a shipment request change from being a custom request to a standard request if someone goes and changes the standard request definitions?
[03:29] <kiko> stub, the answer is probably yes
[03:29] <BjornT> SteveA: yeah, thanks. i'll fix it
[03:29] <kiko> silbs and marilize are looking at the orders to decide which options should be standard
[03:30] <carlos> kiko, hi, SteveA did the review this morning, so don't worry about that
[03:31] <SteveA> cprov: how's that sample data patch going?
[03:31] <kiko> carlos, rock
[03:31] <kiko> thanks SteveA you saved my life
[03:31] <cprov> SteveA: testing locally, 10 mi
[03:32] <SteveA> thanks cprov 
[03:33] <cprov> SteveA: the addition of a sampledata key brakes some tests, repairing them
[03:33] <SteveA> ok
[03:33] <SteveA> probably a good thing to add the sample data
[03:33] <SteveA> as it means more realistic code-paths are being tried
[03:35] <stub> salgado: Anyway - I've sped up those queries by 33%. We can do more, but I wouldn't worry for now (test queries are running in around 0.25 seconds)
[03:36] <salgado> stub, that's great. I think we won't have more problems as soon as we have batching
[03:37] <stub> salgado: ohh.... you shouldn't select a list of ids, and then do ShippingRequest.select('id in (%s)' % ids)
[03:37] <stub> . That is unnecessary and could give you problems.
[03:38] <salgado> stub, how can I do that, then?
[03:38] <salgado> using a temporary table?
[03:38] <stub> salgado: You should be able to do it using select and clauseTables
[03:39] <SteveA> cprov: sourcepackagerelease-index.pt.  what url can i go to in order to see it?
[03:40] <stub> salgado: Although you might need to resort to using a view, as SQLObject will give you trouble otherwise
[03:40] <cprov> SteveA: distribution/distrorelease/+sources/sourcename/releaseversion .. I think
[03:40] <cprov> SteveA: it must be published in that distrorelease
[03:41] <SteveA> a specific URL in the sample data
[03:41] <cprov> SteveA: are you working in canonical_url for srcrelease ?
[03:41] <SteveA> no, although if there isn't on, i'll need to make one
[03:41] <SteveA> i want to see that page template rendered
[03:41] <SteveA> so i can adjust its menus
[03:41] <salgado> stub, I guess the view is the best thing to do. I'll probably do this today, as part of the work I'll do to improve the searches
[03:41] <SteveA> adn see the results
[03:42] <stub> salgado: https://chinstrap.ubuntu.com/~dsilvers/paste/file2hWTGJ.html
[03:43] <SteveA> Kinnison: how big is vegastrike-data ?
[03:43] <stub> vegastrike still fit on a CD last time I saw it...
[03:46] <cprov> SteveA: difficult to do, many moving parts involved, I wonder if we can do it properly in the sampledata and not mess up with publising history, but I think we have some 'wrong' stuff for hoary (mozilla or pmount) in the sampledata 
[03:47] <SteveA> cprov: we should be able to see all the page templates at URLs on localhost, based on the default contents of the sample data.  otherwise, it makes any problems really hard to debug, and it means we probably don't have page tests for these things.
[03:48] <SteveA> cprov: can you let me know when you have the signed coc sampledata mirrored, and i'll merge directly from your branch.  also, send it to pqm.  i just want it earlier.
[03:49] <cprov> SteveA: yes, I know 
[03:49] <cprov> SteveA: sure, test still wrong, more some minutes
[03:49] <SteveA> thanks
[03:58] <Kinnison> SteveA: between 150 and 250 megs, I forget exactly
[03:58] <SteveA> i'm suprised that didn't easily fit across xmlrpc
[03:58] <SteveA> it's not ideal, but should't have caused a barf
[03:59] <Kinnison> it comes across, gets decoded, gets stringIOed, OOMs
[03:59] <Kinnison> kiko: When you "cleaned up" you changed a 'binarypackagerelease' to a 'binarypackage' by mistake
[04:00] <Kinnison> kiko: It's the kind of mistake I make too, it was just a debugging statement but I fixed it
[04:00] <cprov> SteveA: celso.providelo@canonical.com/launchpad--soyuyUI--0--patch-15
[04:01] <cprov> SteveA: should not product any conflicts, but I'm merging from RF now
[04:02] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: Reject onlly the pofiles that have dates older than last import and accept the ones that are equal or newer. Also, we give karma only to translations done inside Rosetta. r=SteveA (patch-2506: carlos.perello@canonical.com)
[04:02] <SteveA> thank you cprov 
[04:02] <cprov> SteveA: nop
[04:03] <SteveA> cprov: will you look into that published release thing too?
[04:03] <kiko> Kinnison, sorry -- and being a debug statement, tough to test, too :-(
[04:03] <Kinnison> kiko: aye, s'okay, I spotted it while doing a publishing run and fix0red it
[04:03] <kiko> Kinnison, perhaps using smaller functions the undefined name would come up in pyflakes
[04:03] <cprov> SteveA: you should be able to see sourcepackagerelease-index in sampledata, I think
[04:04] <Kinnison> kiko: perhaps.
[04:04] <stub> carlos: That the patch that needs cherry picking?
[04:04] <SteveA> cprov: what URL ?
[04:06] <cprov> SteveA: http://localhost:8086/distros/ubuntu/hoary/+sources/pmount for sourcepackage only, you are right, we can't see sourcepackagerelease yet
[04:07] <SteveA> okay.  i want to be able to get a URL where i can see that template, based on the sample data
[04:07] <carlos> stub, yeah, I just sent you an email
[04:08] <carlos> stub, if you could apply that before leaving...
[04:10] <SteveA> carlos: https://launchpad.net/errors/showEntry.html?id=1127830161.350.075378138752
[04:10] <SteveA> jamesh: 
[04:10] <SteveA> https://launchpad.net/errors/showEntry.html?id=1127830025.820.952379835119
[04:13] <carlos> SteveA, it expired already
[04:15] <SteveA> carlos: works for me
[04:16] <carlos> Exception traceback
[04:16] <carlos> The specified log entry was not found. It may have expired.
[04:16] <SteveA> https://chinstrap.ubuntu.com/~dsilvers/paste/fileXnH5ac.html
[04:16] <SteveA> look there carlos 
[04:17] <cprov> SteveA: today, I won't have time to do it, but woul be nice it you can submit a bug for it (it will be postponed probably until "basic test coverage for soyuz" task)
[04:17] <kiko> SteveA, carlos: I believe that bug is reported
[04:17] <kiko> search for rosetta bugs containing newlines or unix
[04:18] <SteveA> kiko: any idea about this one? https://chinstrap.ubuntu.com/~dsilvers/paste/fileXsuZBb.html
[04:19] <carlos> kiko, yeah, you already reported it
[04:19] <kiko> SteveA, looking
[04:21] <ddaa> gah
[04:21] <ddaa> apparently, the svn_oo memory leak is scared away by valgrind...
[04:22] <ddaa> SteveA: is it acceptable to run production services under valgrind to fix a memory leak?
[04:23] <kiko> matsubara, bug 1419 is a dupe of another already reported rosetta bug
[04:24] <SteveA> if elmo is happy with valgrind installed on a server, it gets the job done, and it doesn't piss on anyone's shoes, sure
[04:24] <kiko> matsubara, https://launchpad.net/malone/bugs/908
[04:24] <cprov> SteveA: thanks for # 2603
[04:24] <SteveA> np
[04:24] <ddaa> right... what did I expect from the guy who fixes bugs by running "python -O" :)
[04:25] <SteveA> i fully expect at least one of the three conditions to be false
[04:25] <ddaa> it would get the job done, given 100x the hardware and a lot of time
[04:25] <SteveA> cprov: which user now has a signed Co
[04:25] <SteveA> cprov: which user now has a signed CoC?
[04:26] <cprov> SteveA: name16
[04:26] <SteveA> ddaa: if you had 100x the hardware, the memory leak wouldn't matter
[04:26] <SteveA> thanks cprov 
[04:26] <carlos> SteveA, stub https://launchpad.net/errors/showEntry.html?id=1127831112.880.702582916245
[04:26] <carlos> we still get timeouts...
[04:26] <cprov> SteveA: is it enough ?
[04:28] <SteveA> cprov: hmm
[04:29] <SteveA> i still can't see a way to get a page that uses signedcodeofconduct-portlet-actions.pt
[04:29] <SteveA> maybe it isn't used
[04:29] <cprov> SteveA: could be, mpt have modified some things after i land it
[04:30] <SteveA> actually,
[04:30] <SteveA> i see a page where name16 has signed a CoC
[04:30] <SteveA> and a checkbox to deactivate it
[04:30] <SteveA> but no page to actually get to the CoC itself
[04:31] <cprov> SteveA: /codeofconduct/console/
[04:32] <SteveA> what goes in the "from:" box?
[04:32] <SteveA> aha -- i have the page i need.  thanks.
[04:32] <SteveA> that "From:" and the search form is confusing
[04:33] <SteveA> maybe get mpt's opinion on it, and file a bug on how to improve it?
[04:35] <cprov> SteveA: again, AFAIK he already spend some time on it, but ok, it still needing his love, bug him, if there is something I must he will let me know  
[04:35] <SteveA> ok
[04:35] <SteveA> the thing is, i found the "From:" form item confusing
[04:35] <SteveA> i tried to enter dates in there
[04:36] <SteveA> then tried to leave it blank, expecting to get a batch of all the results
[04:36] <SteveA> then tried name16
[04:38] <cprov> SteveA: a form help message would solve the issue, I see you point 
[04:39] <mpt> What kind of data is supposed to go in there?
[04:48] <mpt> cprov?
[04:48] <mpt> I can fix it right now, I'm in [trivial]  mode
[04:50] <cprov> mpt: oh, sorry afk ... displayname or name, let me verify 
[04:52] <cprov> mpt: uses person fti
[04:52] <mpt> fti?
[04:54] <cprov> mpt: the postgres word indexing, i.e, name, displayname, etc whatever they configure the indexing engine
[04:55] <mpt> So "Enter all or part of a person's name."
[04:57] <kiko> SteveA, that bug is also reported
[04:57] <SteveA> kiko: and assigned to jamesh?
[04:58] <kiko> hmmm
[04:58] <cprov> mpt: yes, name16, foo, bar, Foo, Bar, etc work
[04:58] <kiko> let me find out
[04:58] <kiko> SteveA, I discovered there's a launchpad-cal product?
[04:59] <cprov> mpt: ahhh, '%' is everything
[04:59] <mpt> cprov: and "" is also everything, I hope
[04:59] <SteveA> "" is nothing :-(
[05:00] <mpt> that's bad
[05:00] <cprov> mpt: no, as SteveA said "" is nothing 
[05:00] <cprov> mpt: do you want me to change it ?
[05:01] <mpt> yes, I think it would be better if it worked like a filter
[05:01] <mpt> so if you don't enter anything, you get everything
[05:02] <cprov> mpt: ok, will do it now as trivial 
[05:03] <mpt> cprov: Does the search include e-mail addresses as well?
[05:04] <cprov> mpt: I'm not sure if the Person.fti includes person email, maybe salgado knows ?
[05:04] <salgado> cprov, no, it doesn't
[05:05] <salgado> cprov, mpt, AFAICT we don't have any cross-table fti
[05:06] <Kinnison> SteveA: When you asked about vegastrike-data earlier, was that because you'd decided to take on my branch for review?
[05:06] <cprov> salgado: yes, exactly ... thank you 
[05:06] <SteveA> Kinnison: no
[05:07] <Kinnison> SteveA: *sob*
[05:08] <Kinnison> SteveA: Celso has a patch which I want in RF. It's a stop-gap measure until the uploader is in place. it's more than a few lines, but it'll be thrown away when the uploader is done. Can he merge it [trivial]  or rs=dsilvers ?
[05:10] <SteveA> Kinnison: what modules does it touch?
[05:10] <Kinnison> just buildd master slave scanner IIRC
[05:12] <kiko> SteveA, jamesh: what do we do about the launchpad-cal product?
[05:12] <kiko> SteveA, https://launchpad.net/products/launchpad/+bug/2166
[05:12] <SteveA> kiko: we should have two products: launchpad and rosetta
[05:12] <SteveA> well, and shipit
[05:13] <kiko> there's many malone bugs
[05:13] <kiko> I think it deserves its own product
[05:13] <SteveA> Kinnison: if the only files it touches are in that subsystem, then that's fine.
[05:13] <Kinnison> SteveA: which, [trivial]  or rs=dsilvers ?
[05:13] <kiko> SteveA, I'm going to assign that bug to neuman, it's an easy fix
[05:13] <SteveA> kiko: do we want end users to distingush between whether they're in malone code or in launchpad code?
[05:13] <SteveA> kiko: make sure it gets a test ;-)
[05:14] <SteveA> Kinnison: rs=stevea
[05:14] <kiko> SteveA, the test is quite trivial -- just add to xx-notfound-traversals
[05:14] <Kinnison> SteveA: thanks
[05:15] <kiko> SteveA, I think bug reporters know if the bug occurred while tracking a bug, doing a translation, or using any other part launchpad
[05:15] <kiko> SteveA, until we get moving around products however it's going to be painful
[05:15] <kiko> gneuman, you have bug 2166 too
[05:19] <kiko> thanks bradb you're a star
[05:21] <cprov> SteveA: could you verify the review you did yesterday for me, I've improved the tests and added some librarian setup code for use when running the test alone
[05:23] <SteveA> anyone available to review some code?
[05:23] <SteveA> cprov: okay.  what do i need to look at?
[05:24] <cprov> SteveA: https://chinstrap.ubuntu.com/~dsilvers/paste/fileOXR1j8.html
[05:27] <SteveA> cprov: they look okay.  i can't really concentrate on them well, as i wrote too much code today already :-/  using 'assert' in doctests seems odd to me.
[05:29] <cprov> SteveA: assert is only used for things I can't be sure, like buildduration, datebuilt (timestamps) and buildlog (librarian alias)
[05:31] <SteveA> i don't know what you mean by "used for things I can't be sure"
[05:32] <cprov> SteveA: buildduration and datebuilt are timestamps, usually "NOW" how to check it ? i could use type, but it still being inside an assert 
[05:34] <cprov> SteveA: is it ok then ?
[05:35] <cprov> salgado: have time for a 50 lines review to unblock mpt ?
[05:35] <SteveA> it looks okay to me, but i didn't look at it very deeply.  i think it is okay to merge, but you might want to ask salgado for a second look at it.
[05:36] <cprov> SteveA: fair enough, let's see if he has time ... thank you
[05:36] <mpt> cprov: I'm not blocked, I'll just merge assuming that you also merge soon :-)
[05:36] <mpt> "Leave blank for a complete list."
[05:38] <cprov> mpt: ok, it was just a excuse to unblock myself ;) 4 trees with pending stuff, brain is screaming ...
[05:39] <cprov> salgado: ping?
[05:40] <salgado> cprov, where's the diff?
[05:40] <cprov> salgado: https://chinstrap.ubuntu.com/~dsilvers/paste/filenv7z3Z.html
[05:44] <kiko> gneuman, when you send me code, please cite the bug number it fixes -- matsubara, thanks for including the bug #
[05:50] <kiko> matsubara, I decided that bug 1419 is not a dupe of 908, but only derived from it
[05:50] <kiko> I'll undupe
[05:53] <kiko> matsubara, when was the last time you merged from rocketfuel?
[05:54] <kiko> I think those links are disabled
[05:54] <matsubara> i merged friday
[05:55] <matsubara> but the links wasn't disabled yet, cause steve was waiting for a revision
[05:55] <ddaa> OMG... python-subversion memory management is... inexistent...
[05:56] <kiko> matsubara, ah
[05:57] <matsubara> anyway, i re-enable it because it's working now and the page is self-explaining (steve was complaining about that) as you can see in the bug comments
[05:57] <kiko> matsubara, I think the latest code in RF has them disabled in browser code.
[05:57] <kiko> bradb, SteveA: I got a bad conflict when star-merging -- can someone send me the current version of browser/bugtask.py?
[05:58] <bradb> kiko: the quickest would probably be to get $rocketfuel, because I'd have to do that too, to ensure I have the most recent
[05:59] <bradb> and gets don't normally take long
[05:59] <SteveA> quicketst -- get on chinstrap
[05:59] <mpt> go go pqm
[05:59] <SteveA> ssh it across
[05:59] <kiko> fucking baz
[06:00] <mpt> bradb: https://launchpad.net/products/malone/+bug/2608
[06:00] <bradb> yeah, looks like somebody broke that
[06:01] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: [r=BjornT]  support tracker cleanup; [trivial]  fix bug 1389 (patch-2507: mpt@canonical.com)
[06:02] <kiko> mpt, I just noticed the same 
[06:05] <mpt> ha
[06:06] <mpt> I was just about to report a bug about a Malone notification having a malformed Reply-To header, then I realized the notification containing the malformed Reply-To header was notifying me that Bjorn had just reported the bug about the malformed Reply-To header
[06:07] <kiko> heh
[06:12] <kiko> cprov, ping?
[06:13] <cprov> kiko: pong
[06:13] <kiko> can I assign matsubara to bug 2609?
[06:13] <SteveA> mpt: you changed lib/canonical/launchpad/templates/ticket-portlet-actions.pt
[06:14] <SteveA> mpt: what did you change?
[06:14] <cprov> kiko: sure
[06:15] <kiko> thanks
[06:15] <mpt> SteveA: from memory, it was just s/ticket/request/g
[06:16] <SteveA> okay, i have a diff of it
[06:16] <SteveA> i have totally converted that into menus
[06:16] <mpt> was I right? :-)
[06:16] <SteveA> you did more than that
[06:16] <SteveA> but i can change the menu
[06:16] <mpt> ok, sorry about that
[06:17] <mpt> that branch was started before the actions portlet nuking
[06:18] <kiko>  EmailAddressAlreadyTaken: The email address azam.r@rediffmail.com is already registered.  
[06:18] <kiko> salgado?
[06:18] <bradb> darn, system error assiging specs to milestones
[06:19] <ddaa> memory managemen by svn: freeing memory? hu? what for?
[06:20] <bradb> How do I access the error logs for production? https://launchpad.net/errors gives me a 403.
[06:22] <mpt> kiko: See, I'm not the only one!!
[06:22] <mpt> bradb: Install the Launchpad client cert. in your browser
[06:22] <kiko> bradb, doh. 
[06:22] <ddaa> maybe if I just fork cscvs for each revision that would make it possible to keep the leak in control...
[06:24] <bradb> Hey, cool, Safari in Tiger has "private browsing". Worry-free pr0n surfing.
[06:25] <SteveA> argh -- seeing &#8217; in page template source makes them much harder to read and understand
[06:26] <SteveA> worse than other things because they interrupt a word
[06:27] <salgado> kiko, ?
[06:27] <kiko> salgado?
[06:28] <kiko> have you seen that error before?
[06:30] <mpt> SteveA: Actually, I can fix that by specifying an encoding in main-template.pt
[06:30] <mpt> since we're serving XHTML-as-HTML
[06:31] <SteveA> does that mean that the files would no longer be ascii
[06:31] <SteveA> ?
[06:31] <salgado> kiko, I don't remember seeing it before
[06:31] <bradb> Can someone privmsg me the client cert password?
[06:31] <mpt> SteveA: correct
[06:31] <kiko> me neither
[06:31] <SteveA> my point is, would it mean that ordinary programmers would not be able to type a consistent "'" ?
[06:31] <SteveA> no, don't do that
[06:32] <SteveA> i think apostrophes should be the "'" character in launchpad
[06:32] <bradb> kiko: Thanks
[06:32] <kiko> SteveA, why not use something like DPoT to convert?
[06:32] <mpt> SteveA: No, it wouldn't mean that
[06:33] <SteveA> kiko: this stuff is coded directly into templates
[06:33] <SteveA> it makes them hard to read
[06:33] <kiko> SteveA, that I don't like very much either
[06:33] <mpt> So does all this <a href="..."> and <code> and <div> goop
[06:33] <kiko> SteveA, btw, is there a cookbook on working around "Constraint not satisfied"?
[06:33] <SteveA> they don't occur in the middle of a word, mpt
[06:34] <SteveA> i don't object to "nice quotes" around a word
[06:34] <SteveA> but, to interrupt a word in the middle is awful
[06:34] <SteveA> kiko: how do you get that?
[06:35] <bradb> Hm, where are we at with fixing the problem with error logs, and getting a random error log when you visit launchpad.net/errors?
[06:35] <dilys> Merge to rocketfuel@canonical.com/launchpad--production--1.34: [trivial]  production cherry picks (patch-11: carlos.perello@canonical.com, stuart.bishop@canonical.com, rocketfuel@canonical.com)
[06:35] <kiko> SteveA, matsubara gets it when uploading a larger hackergotchi than allowed.
[06:36] <SteveA> so, working as it was designed?
[06:36] <kiko> SteveA, yes, apart from the horrible error message
[06:36] <SteveA> aha
[06:36] <kiko> IIRC stub explained how to avoid that
[06:36] <SteveA> i'd have to research it.  i don't have the answer in front of me.  maybe BjornT knows.
[06:37] <kiko> stub knows
[06:37] <BjornT> kiko: you should raise a LaunchpadValidationError in the validator
[06:37] <kiko> aha
[06:38] <kiko> salgado, matsubara: is that easy to do for you?
[06:39] <mpt> SteveA: When did you last land menus work?
[06:40] <SteveA> bjorn is reviewing the next lot now
[06:40] <SteveA> this basically finishes the conversion
[06:41] <mpt> Sorry, I meant, when did you *last* land menus work, as in, did you get my fix for the Rosetta status legend
[06:41] <mpt> someone's just reported a bug on it
[06:42] <SteveA> i have no idea what "Rosetta status legend" means
[06:42] <SteveA> best thing to do is
[06:42] <SteveA> 1. log into chinstrap
[06:42] <SteveA> 2. baz get rocketfuel@canonical.com/launchpad--devel--0
[06:42] <mpt> SteveA: The stuff you merged from me yesterday included a change to rosetta-status-legend.pt
[06:43] <SteveA> 3. look in the resulting tree to see if it is there
[06:43] <salgado> kiko, yes, that is. I'll guide matsubara to fix that
[06:43] <SteveA> i haven't had pqm merge things since yesterday
[06:43] <mpt> ok
[06:43] <kiko> thanks salgado -- you rock as usual
[06:43] <mpt> thanks
[06:43] <SteveA> everything that was on your Menus branch today will be merged later today or early tomorrow
[06:44] <SteveA> mpt: i've just added a custom error page for requests timing out
[06:44] <SteveA> you'll probably want to improve the text on it
[06:44] <mpt> okie dokie
[06:45] <SteveA> but the idea is to let people know that it is a different kind of error, and may well be transient, and also to let the developers know more quickly the nature of the problem.
[06:59] <SteveA> ddaa: ping
[06:59] <ddaa> ding dong
[06:59] <SteveA> apr_initialize() ; stuff ; apr_terminate()
[06:59] <ddaa> makes no difference
[06:59] <SteveA> try that
[06:59] <SteveA> okay
[07:00] <ddaa> that's why I ommited it, it's not part of the minimal snippet :)
[07:09] <SteveA> cprov: the console stuff for signed cocs -- is it accessible by admins only?
[07:09] <cprov> SteveA: yes 
[07:09] <SteveA> thanks
[07:30] <kiko-fud> bradb, please don't assign bugs to mark.
[07:32] <bradb> who should it be assigned to?
[07:33] <kiko> anyone but him
[07:33] <kiko> I'll fix it
[07:33] <kiko> don't worry
[07:33] <kiko> just use that for future reference
[07:34] <bradb> ok
[07:42] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: [trivial]  Unpublishing a missing file should be a noop. (also fix silly typo in debug message in death row processor) (patch-2508: daniel.silverstone@canonical.com)
[07:47] <Kinnison> ciao dudes
[07:47] <mpt> tchau
[07:56] <zyga> hello
[08:02] <zyga> https://launchpad.net/products/launchpad/+bug/2612
[08:03] <zyga> how to add requests for enchancement?
[08:04] <kiko> zyga, use malone, it's practically the same.
[08:05] <zyga> hmm, isn't this the same thing *exactly*?
[08:11] <kiko> that's a good question
[08:12] <kiko> mark has the idea that enhancement/feature requests are different than bugs
[08:12] <kiko> it may be that he's changed his mind now that he has a spec tracker
[08:21] <bradb> aw damn
[08:21] <bradb> monster pqm queue of death, and now i see that i'm going to conflict with BjornT's patch
[08:23] <bradb> so, who wants to bounce pqm? :)
[08:23] <kiko> pas moi
[08:23] <bradb> ciboire
[08:25] <mpt> bradb: Did you realize how many things would have broken if you *hadn't* set up a redirect for /malone/bugs/x? :-)
[08:25] <bradb> there was no question about the redirect
[08:26] <bradb> it would have been just plain rude any other way
[08:26] <mpt> Should the URLs included in e-mail notifications change now?
[08:26] <mpt> (and if so, to what?)
[08:26] <bradb> i'm not so sure
[08:27] <bradb> maybe to just /bugs/42 ;)
[08:28] <mpt> or bugs.launchpad.net/42
[08:28] <bradb> that would be a lot harder
[08:28] <bradb> because it would imply that bugs.launchpad.net works
[08:28] <kiko> right
[08:28] <mpt> true
[08:29] <bradb> unfortunately, the bugmail address already gives that illusion :/
[08:30] <kiko> true
[08:30] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: [trivial]  change a .remove to .pop to fix strange pool bug (patch-2509: daniel.silverstone@canonical.com)
[08:31] <mpt> well, move the "bug" from after the @ to the start of the address, and drop the "s." ;-)
[08:31] <bradb> dilys: are you serious?
[08:31] <kiko> Kinnison, with fixes?
[08:31] <BjornT> bradb: hmm, maybe your patch is better, i simply disabled the huge lists. did you make them batched, though? otherwise the new page probably will take too long to render.
[08:31] <bradb> BjornT: no batching. batching hurts.
[08:31] <bradb> BjornT: i broke it into two reports though
[08:32] <bradb> because a "bugs on my software" thing is important to keep
[08:33] <kiko> BjornT, yeah, bradb's patch is nicer..
[08:35] <bradb> mpt: possibly. or i was thinking bug.42@... might be a slightly cleaner (looking) way to do it.
[08:36] <bradb> of course, *ideally*, i would love it if bugs.launchpad.net actually *did* work :)
[08:36] <BjornT> bradb, kiko: the number of bugs directly assigned to a user is fairly small, though. which means that the new page will cause a system error, just like +assignedbugs did. is that really what we want?
[08:37] <bradb> BjornT: i think so. currently, the "Assigned Bugs" report shows more than just assigned bugs, which seems confusing. we can tune the reports further, if needed.
[08:38] <mpt> hooray for that
[08:38] <kiko> BjornT, I see your point -- he moved everything out of there, is that it?
[08:38] <BjornT> bradb: i agree that +assignedbugs should show only bugs assigned to the user. the question is, do we want to add a page which probably will cause a system error?
[08:39] <bradb> BjornT: no, but we want to have a page to view bugs on software you maintain. if it still causes a system error, we'll optimize it.
[08:39] <BjornT> kiko: yeah, he move the part that took long to render
[08:39] <BjornT> bradb: then i suggest, temporarily disable it, optimize it, enable it. how does that sound?
[08:40] <bradb> BjornT: why not see if it works first? :)
[08:41] <bradb> i suspect one of the nastier culprits on that page was the "bugs with shared interest" bit
[08:41] <bradb> e.g. i can render the +assignedbugs report when i'm not logged in, but when i login, that "bugs with shared interest" query gets run, and whoops, timeout error
[08:43] <zyga> bradb: what is pqm?
[08:43] <BjornT> bradb: from my testing with sample data, viewing /people/bjornt/+assignedbugs shouldn't show that part. maybe i'm wrong, though.
[08:43] <bradb> BjornT: the query gets run if you're logged in
[08:43] <mpt> zyga: Patch Queue Manager
[08:44] <BjornT> bradb: searching for bugs also takes longer while you're logged in, so it could be the query for private bugs that slow things down.
[08:44] <mpt> zyga: It takes all the changes we're making to the Launchpad code, applies them in order, and for each one tells us if the change was applied successfully
[08:44] <BjornT> bradb: ouch, it shouldn't... but did you remove that part, or is that on the new page as well?
[08:45] <bradb> BjornT: it should. that's the way the tal:condition is setup. that bit is completely gone on my branch.
[08:46] <bradb> (the code behind the scenes is still there if we want to display that info somewhere else, but it doesn't seem to make sense on an "assigned bugs" report, nor on a "bugs on software i maintain" report.)
[08:46] <BjornT> bradb: when i said "shouldn't" i didn't mean according to the tal:condition, but according to common sense :)
[08:48] <BjornT> bradb: in fact, it isn't being run, only the view method is called which returns early
[08:51] <bradb> the method always gets called when you're logged in, as best i can tell. i can see though that it short-circuits when current user == the person of the page being rendered
[08:52] <BjornT> bradb: so what do you think. should we go for your patch and see how it goes, or go with mine and fix it properly later?
[08:52] <zyga> mpt: hmm so you send a changeset to this PQM and it applies them in submission orted?
[08:52] <zyga> order
[08:52] <mpt> yes
[08:52] <zyga> seems random :D
[08:53] <mpt> well, usually we don't conflict with each other, so it's not a problem
[08:53] <zyga> how big is the launchpad tree?
[08:53] <mpt> 30,000 lines, iirc
[08:53] <zyga> nice
[08:54] <bradb> BjornT: i'd kind of prefer to leave it in there, because then at least we can test it, e.g. on staging. i can probably fix it proper before the next prod rollout.
[08:54] <zyga> the largest project I've worked on has 48K lines
[08:55] <bradb> zyga: did the users like it?
[08:55] <BjornT> bradb: this patch should get cherry picked into production, it's not good to have pages that produce system errors
[08:55] <zyga> bradb: its a very advanced linguistic search engine, all C
[08:55] <zyga> the GUI's are not counted into this
[08:56] <zyga> bradb: the users love it, they don't have a choice ;=)
[08:56] <zyga> but anyway it was really nice, all the similar stuff was either close sourced or really insignificant
[08:56] <zyga> or both
[08:58] <mpt> boom, firefox go bye bye
[08:59] <Hieronymus> The search function in Malone isn't working for me
[09:01] <bradb> BjornT: *shrug*. not sure what to tell you. both of our patches make the same change (it sounds like) to +assignedbugs. neither of our patches will guarantee that the system error goes away. mine keeps a +packagebugs page there too, so that we might at least start to see a pattern emerging of where the problem is. (don't forget that the privacy query gets run on both our +assignedbugs patches anyway.)
[09:02] <mpt> DIE PLONE.CSS DIE
[09:02] <bradb> kill it
[09:03] <bradb> baz rm plone.css; baz commit -s "buhbye"
[09:03] <kiko> bradb, it's your day to do a drive-by review
[09:03] <BjornT> bradb: well +assignedbugs will contain a small number of bugs, so it shouldn't be a problem. how about if i leave my patch in pqm and get it cherry picked into production. you resolve the conflicts and resubmit your patch, which we can test on staging?
[09:03] <kiko> are you ready?
[09:04] <bradb> BjornT: yeah, that seems ok to me
[09:04] <bradb> kiko: sure
[09:05] <BjornT> bradb: cool
[09:05] <kiko> bradb, it's for fixing that bug about the assignee widget javascript -- remind me of the number again?
[09:06] <bradb> kiko: cool. #2107.
[09:08] <BjornT> kiko: how's it going with my reviews?
[09:09] <kiko> BjornT, I'm going to start after fixing this, you should see how much more pull office people have for reviews :-P
[09:13] <bradb> BjornT: thanks for filing the bug reports about the bug reports. btw, the report seems to have some attachment pollution in it. e.g. https://launchpad.net/products/malone/+bug/2614
[09:13] <bradb> maybe that's a known problem
[09:14] <kiko> bradb, sending diff off
[09:15] <bradb> cool
[09:15] <BjornT> bradb: yeah, i should remove that one, it's quite trivial.
[09:18] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: [trivial]  fix temporary buildd result collector to accomplish with publisher, its a worarround before landing uploader properly. (patch-2510: celso.providelo@canonical.com)
[09:24] <kiko> bradb, received it?
[09:25] <bradb> kiko: yeah, just looking now
[09:25] <kiko> thanks
[09:31] <bradb> kiko: can you make it select that specific radio button in onKeyPress radio event? (e.g. the one with id assign_to)
[09:32] <kiko> hmmmm
[09:32] <kiko> using getElementById()?
[09:32] <kiko> perhaps
[09:32] <kiko> bradb, sent another patch to fix a traversal issue I just found
[09:33] <bradb> also, shouldn't it be on the onFocus event? e.g. onFocus="javascript: getElementById('assign_to').checked = true"
[09:35] <kiko> I don't think so
[09:35] <bradb> hm, actually, neither of those events seems to focus the correct radio button if you use "(Choose...)" to pick the assignee
[09:36] <kiko> I think I can work around that
[09:44] <kiko> BjornT, how do I place an explicit semicolon in a tal:attributes string declaration?
[09:46] <BjornT> kiko: a quick google search told me to double the semicolon (;;)
[09:47] <kiko> sorry :-(
[09:53] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: [r=stevea]  Release DEPWAIT and CHROOTWAIT jobs in slavescanner, only block BUILDERFAIL ones (patch-2511: celso.providelo@canonical.com)
[09:53] <bradb> zope 3 is usually not very googlable
[09:54] <kiko> that's why I didn't try
[09:56] <BjornT> true. TAL isn't specific to zope 3, though, so it's easier to find information about that.
[09:56] <bradb> one other thing (sorry, was on phone there for a bit), but it looks like you removed a space above ISinglePopupView that leaves only one space between it and the previous class's method declaration. normally it's 2 blank lines between class definitions
[09:58] <bradb> same with top-level function defs
[09:59] <kiko> perhaps I did
[09:59] <kiko> but that file is all busted anyway
[09:59] <kiko> fixed
[10:00] <salgado> kiko, let's review https://chinstrap.warthogs.hbd.com/~jamesh/pending-reviews/guilherme.salgado@canonical.com/launchpad--shipitng--1/filtered-diff together?
[10:01] <kiko> yes
[10:01] <kiko> let me finish my patch for bradb
[10:05] <bradb> BjornT: hm, is it possible for the "name" passed to a traverser function to ever be None? that doesn't make sense to me.
[10:06] <kiko> bradb, did you see the URL I provided in the summary?
[10:06] <bradb> yes
[10:07] <BjornT> bradb: i don't think so. why do you ask?
[10:07] <kiko> that system error outs
[10:07] <bradb> BjornT: just curious, because kiko was putting checks against None in this traversal patch
[10:08] <kiko> because if you don't, it blows up.
[10:08] <bradb> kiko: name isn't None in that case. name is +bug
[10:08] <BjornT> kiko: which URL? if it blows up, there should be a test causes None to be passed as the name.
[10:09] <bradb> and then we try to consume the next path step inside the traverser to get the right bugtask to show
[10:09] <kiko> bradb, really?
[10:09] <bradb> yeah
[10:09] <kiko> that's not what the traceback tells me.
[10:09] <bradb> that's what _consume_next_step is all about
[10:09] <kiko> I think you're confused.
[10:10] <kiko> bradb, https://launchpad.net/products/launchpad/+bug
[10:10] <kiko>     *  if nextstep.isdigit():
[10:10] <kiko> AttributeError: 'NoneType' object has no attribute 'isdigit'
[10:10] <bradb> right, nextstep, not name :)
[10:10] <kiko> oh, sorry.
[10:10] <kiko> that was just me being extra-cautious :)
[10:13] <bradb> kiko: also, the _traverse_plus_bug function you created has three args, but your callsites are only passing two
[10:13] <bradb> not sure what "self" is all about
[10:13] <kiko> I nuked that, I was on crack
[10:14] <bradb> can you send me a new version of that patch? this code is b0rked dude. :P
[10:14] <kiko> no
[10:14] <kiko> all I did was remove the self method
[10:14] <bradb> then it's still broken
[10:15] <kiko> really?
[10:15] <bradb> it's referring to "request" in that function, when request isn't defined
[10:15] <kiko> yes, bradb, with the obvious fixes
[10:15] <kiko> anything else?
[10:15] <kiko> I feel like invoking the BRUCE on you
[10:15] <bradb> you mean you renamed "nextstep" to "request"?
[10:16] <kiko> yes, bradb, the obvious fixes (the patch hadn't been run)
[10:17] <bradb> kiko: you can remove the "name is not None" bit of that if statement, in case you hadn't already
[10:17] <kiko> sure
[10:17] <kiko> the name will never be none, right?
[10:17] <bradb> according to BjornT and I, it wouldn't make sense. you can blame us later if we turn out to be wrong.
[10:18] <kiko> sure
[10:18] <bradb> other than that, i'd add a check_not_found test for each +bug traversal in xx-notfound-traversals.txt
[10:18] <kiko> will do
[10:18] <kiko> thanks
[10:18] <bradb> np
[10:21] <bradb> p.s. PQM IS WAITING FOR YOU
[10:21] <kiko> bradb, fixed the other issue with the JS patch
[10:22] <bradb> sweet. so, it'll work with Choose...?
[10:22] <kiko> yep.
[10:22] <kiko> bradb, is PQM hung then?
[10:22] <salgado> no, it's not
[10:22] <bradb> kiko: no, i guess not. just taking astonishingly long.
[10:23] <bradb> there are three hour old requests still in the queue
[10:23] <salgado> the queue's been full all day
[10:24] <asmodai> kiko: you happy? :)
[10:24] <kiko> asmodai, you rock!
[10:24] <asmodai> kiko: hahaha
[10:25] <asmodai> kiko: you're welcome, hope it works out to something nice for you guys. ^^
[10:25] <kiko> me too!
[10:25] <asmodai> Michael is a cool guy to deal with, passionate about things, which you might've guessed from his email
[10:25] <asmodai> No idea on the pure business side of things though, he at least seems to be very concrete.
[10:26] <kiko> yeah, seems very sane
[10:26] <kiko> I need to talk to gabriel here to see who we can allocate to this
[10:26] <asmodai> *nods*
[10:27] <asmodai> Done much on BSD yet?
[10:28] <kiko> we have two projects on openbsd
[10:28] <asmodai> Oh nice.
[10:29] <asmodai> If there's any issues I can help out with, just bellow. 
[10:29] <kiko> thanks, you rock
[10:30] <asmodai> de nada
[10:32] <asmodai> kiko: And if stuff goes as it should you will wind up doing Ubuntu related coding too for them I guess ;)
[10:32] <kiko> hopefully!
[10:32] <asmodai> It's in your hands ;)
[10:45] <kiko> bradb, damn, I can't find out if the onkeypress() method is standard or not.
[10:50] <bradb> hm, it's mentioned here: http://www.w3.org/TR/WD-html40-970917/interact/forms.html#edef-INPUT but the wording isn't written in a way that clearly says to me "this is standard"
[10:50] <bradb> seems like a definite maybe
[10:52] <kiko> bradb, AFAICS that refers to the event, which does exist and is standard
[11:07] <sabdfl> mpt: ! you karma monster!
[11:07] <mpt> What have I done?
[11:07] <sabdfl> dude, you're like in the top 5 on LP!
[11:08] <sabdfl> pretty awesome
[11:08] <mpt> oh, right :-)
[11:08] <mpt> from all my bug reporting, I guess
[11:09] <mpt> and the occasional fix doesn't hurt, either
[11:09] <janimo> ping lifeless
[11:09] <kiko> I should get into translating
[11:09] <kiko> it gives you lots of karma
[11:09] <kiko> bug work just doesn't pay off
[11:10] <janimo> ping jblack
[11:10] <mpt> yeah, I have 938 and jordi has, like, 9857
[11:10] <janimo> are any of the bazaar importer people here?
[11:11] <dilys> Merge to rocketfuel@canonical.com/launchpad--devel--0: Add batch navigation to the list of shipit orders. r=kiko (patch-2512: guilherme.salgado@canonical.com)
[11:13] <salgado> hey kiko, what about that code review?
[11:14] <kiko> salgado, almost there
[11:15] <kiko> janimo, hmm ddaa or jblack are the guys
[11:15] <salgado> kiko, are you reviewing it already?
[11:15] <kiko> salgado, no, finishing my patch
[11:16] <janimo> kiko, thanks is there a launchpad m-l?
[11:17] <kiko> janimo, you can file bugs on launchpad, and I can make sure they are fixed
[11:18] <janimo> ok thanks, I was about to ask about importing from external svn to baz but will get jblack tomorrow
[11:19] <kiko> sure thing
[11:45] <jordi> so much jealousy
[11:45] <jordi> good night folks
[11:46] <kiko> gnight
[12:01] <sabdfl> spiv: ping
[12:01] <sabdfl> night kiko
[12:01] <sabdfl> jordi: dude. #1
[12:01] <sabdfl> taste it