[01:37] <mpt__> bluefoxicy, "came up with a way to keep the massive workload of sifting through reports in check"?
[01:37] <mpt__> Tell me about that :-)
[01:38] <bradb> HamAssassin?
[01:39] <bluefoxicy> mpt__: pitti is doing some sort of automated problem reporting in Ubuntu, where it'll catch crashes and get a bunch of debugging data, then turn that into backtraces and crash conditions
[01:39] <bluefoxicy> mpt__:  problem:  Ubuntu puts out updated Firefox.  Update is fucked.  3.2 million reports (same bug) come in in the next hour because firefox crashes.
[01:40] <bluefoxicy> mpt__:  I figure you can't easily automate the process of finding duplicates, because more or less different types of bugs reveal themselves under different conditions and also you can only get "mostly" correct results under specific sets of assumptions
[01:41] <kiko> bluefoxicy, it's interesting that we have /some/ ways of detecting duplicates
[01:41] <bluefoxicy> mpt__:  so what I came up with is, let people SEARCH the reports based on all the characteristics you can come up with on an automated report; and let them chose their heuristic parameters.  Quick and dirty
[01:41] <kiko> bluefoxicy, for instance, if we have crash logs, we can try fingerprinting them.
[01:41] <kiko> bluefoxicy, if we have bugwatches, we can detect if two bugs point to the same bug.
[01:41] <bluefoxicy> mpt__:  at the point that you have a crash reporter, you tell people to stop submitting crashes through the bug reporter and just send them through the automated report system. :)
[01:41] <kiko> agreed there.
[01:42] <mpt__> bluefoxicy, right, I assumed that we don't want to report a bug for every crash
[01:42] <kiko> and indeed part of the strategy for avoiding dupes is to push people to the support tracker
[01:42] <kiko> or some other, separate-from-the-bugtracker, triaging mechanism
[01:42] <mpt__> Nor do we want to report a support tracker request for every crash
[01:42] <mpt__> Launchpad is not the solution to this problem :-)
[01:42] <bluefoxicy> kiko:  nods.  But you can't take every manually-reported "firefox crashed" bug and find that it's a SIGSEGV when the instruction pointer leaves some_function() and attempts to execute {unmapped,non-executable} memory
[01:42] <kiko> crashes are special because we actually know they are crashes
[01:43] <kiko> most bugs are unfortunately more subtle and thus require manual triage
[01:43] <mpt__> bluefoxicy, are you familiar with the Oops system?
[01:43] <kiko> but it's nice if the triage can happen outside of the bugtracker
[01:43] <bluefoxicy> mpt__:  program says "oops I died"?  :)
[01:43] <kiko> bluefoxicy, sure, you can't, but the trick is to get users /not/ to file bugs in that situation.
[01:43] <mpt__> bluefoxicy, when Launchpad crashes it generates an "oops report"
[01:44] <bluefoxicy> kiko:  nods.
[01:44] <mpt__> and matsubara aggregates these each week and reports bugs for each *kind* of crash
[01:44] <bluefoxicy> mpt__:  also "Also I took interest in pitti's automated crash reporter and came up with a way to keep the massive workload of sifting through reports in check" was the full context :)
[01:45] <mpt__> bluefoxicy, I've been designing the GUI for that crash reporter, so how the back end works is highly relevant :-)
[01:45] <bluefoxicy> mpt__:  I was talking about stuff in ubuntu at the time, not launchpad itself, sorry  :)  little off topic.
[01:45] <bluefoxicy> mpt__:  ah.    http://bluefox.kicks-ass.org/stuff/bluefox/slides/pseudo_heuristic_searching/img0.html then :)
[01:45] <mpt__> bluefoxicy, https://wiki.ubuntu.com/CrashReporting
[01:45] <bluefoxicy> and yes slide 5 is lacking data; it's also not very much important
[01:47] <bluefoxicy> mpt__:  https://wiki.ubuntu.com/AutomatedProblemReports  -> https://wiki.ubuntu.com/AutomatedProblemReportsTagging
[01:48] <bluefoxicy> mpt__:  also it would be nice if people brought to my attention that such things are around, re   https://wiki.ubuntu.com/CrashReporting <-> https://wiki.ubuntu.com/AutomatedProblemReportsNotification
[01:49] <mpt__> I could say exactly the same thing
[01:49] <mpt__> oh, I did find it
[01:49] <mpt__> Well, I found the former but not the latter
[01:50] <bluefoxicy> mmm.
[01:50] <mpt__> I had no idea that you were interested in the subject, let alone what your nickname was.
[01:51] <bluefoxicy> ah, yeah.  Actually my nickname is registered on launchpad
[01:51] <bluefoxicy> amusingly I'll talk to people about specs and they'll be like "yeah that guy is a nut" not realizing ... :>
[01:52] <mpt__> So for Edgy we can't do anything fancy, just make it easy for people to report a bug if they want to
[01:52] <bluefoxicy> http://bluefox.kicks-ass.org/stuff/bluefox/slides/pseudo_heuristic_searching/img12.html  <-- the notes are fugly but this is one interesting case, sometimes a SIGSEGV and a SIGILL are the same bug.
[01:53] <mpt__> and hopefully I'll get to discuss this with sfllaw and pitti (and maybe matsubara?) at the next UDS.
[01:53] <mpt__> or jamesh
[01:53] <bluefoxicy> yeah, edgy is way too close to do any fancy stuff
[01:53] <bluefoxicy> I was thinking about trying to sneak into the next developer's summit, except I have no way to get there
[01:57] <bluefoxicy> mpt__: I'm also looking at documenting crashes that indicate possible security flaws
[01:58] <bluefoxicy> any crash that indicates either the instruction pointer moved in ways it shouldn't or that someone was writing somewhere they shouldn't be (i.e. unmapped memory) shows a way to mess with program flow, possibly hijack the program
[01:58] <bluefoxicy> which makes things very fuzzy
[01:59] <bluefoxicy> since that's just about every crash (SIGILL and SIGSEGV)
[02:00] <bluefoxicy> but double-free() and stack smashes are explicitly detected; again, though, this is fuzzy since now we KNOW there's a hole but we also know that we stop it if someone attacks it so we don't care quite so much as we might.
[02:00] <bluefoxicy> But that's my end goal :)
[02:47] <Sp4rKy> hi
[02:48] <Sp4rKy> i'm ubuntu member since a few weeks but i haven't yet my @ubuntu.com email adress
[02:48] <Sp4rKy> is it normal ?
[02:53] <kiko> Sp4rKy, nope, it's a bug. you can file a bug on elmo though
[02:54] <Sp4rKy> ok
[02:55] <Sp4rKy> just send it an email
[02:56] <Sp4rKy> should i manually add my ?
[02:57] <Sp4rKy> oups
[02:57] <kiko> well
[02:58] <kiko> you can mail new@bugs.launchpad.net yes
[02:58] <Sp4rKy> k
[02:58] <Sp4rKy> i've wrote a query to elmo 
[02:59] <kiko> right
[02:59] <kiko> well it's kinda late around there so you could try tomorrow morning
[02:59] <kiko> or monday
[03:00] <Sp4rKy> yes
[03:00] <Sp4rKy> but as he's only away , i 've wrote to him
[03:00] <Sp4rKy> and i just let my computer started
[03:00] <Sp4rKy> thx
[03:01] <kiko-zzz> sure thing
[11:53] <sabdfl> help
[11:54] <SteveA> hi mark
[11:54] <SteveA> what's up?
[11:55] <SteveA> hi sabdfl 
[11:55] <sabdfl> what's a zope Interaction?
[11:55] <SteveA> are you seeing a test failure in the karmacategory.txt doctest?
[11:55] <sabdfl> nup
[11:55] <SteveA> I've had that fail in PQM a few times.  Looks like an ordering issue.
[11:56] <sabdfl> long pause, then a crash in webapp/launchbag.py
[11:56] <sabdfl> far, far, far from anything i'm touching
[11:56] <sabdfl> line 67
[11:56] <SteveA> you need to update your zope checkout
[11:56] <SteveA> stub mailed the list -- needed to pull a more recent upstream to get some sessions fixes
[11:57] <SteveA>   bzr pull sftp://devpad.canonical.com/home/warthogs/archives/rocketfuel-built/launchpad/sourcecode/zope
[11:57] <SteveA> or just bzr pull from your zope directory
[11:57] <SteveA> should do it
[11:58] <sabdfl> ok, thanks
[11:58] <sabdfl> i just recently rsync'd down from the rocketfuel-built dir, though
[11:58] <SteveA> seeing as you asked, the interaction is a thread-local that keeps track of the request in progress on the current thread and the principals associated with it
[11:58] <sabdfl> ah
[11:58] <SteveA> did you rsync it from chinstrap?
[11:58] <sabdfl> hmm... maybe
[11:59] <sabdfl> no, sodium
[11:59] <SteveA> use sodium (aliased as devpad.canonical.com, to avoid distruption in future changes of machine)
[11:59] <SteveA> hmm
[11:59] <SteveA> that is strange
[12:00] <SteveA> stick the full traceback on https://sodium.ubuntu.com/~andrew/paste/
[12:00] <SteveA> and I'll see if it makes any sense to me
[12:01] <sabdfl> oh, sec
[12:01] <sabdfl> there's a crash just before that
[12:02] <sabdfl> an IntegrityError tht *is* related to what i'm touching
[12:02] <sabdfl> then it pauses, and crashes on the redirect
[12:02] <sabdfl> interesting
[12:02] <SteveA> hmm.  maybe it hoses the database connection, and then doesn't do an abort properly in the publisher
[12:03] <SteveA> something like that
[12:03] <SteveA>   latest revision: Fri 2006-07-28 08:27:39 +0100
[12:03] <SteveA> that's the latest revision in my zope tree, fwiw
[12:03] <SteveA> although it sounds like you've found the problem
[12:03] <SteveA> I'm still interested to see the TB of the interaction error
[12:04] <sabdfl> i have the same revision
[12:05] <sabdfl> will paste you the TB shortly
[12:06] <sabdfl> http://sodium.ubuntu.com/~andrew/paste/fileWlooqp.html
[12:07] <sabdfl> https://sodium.ubuntu.com/~andrew/paste/fileWIooqp.html
[12:07] <sabdfl> sorry, the latter
[12:12] <SteveA> ah, okay
[12:12] <SteveA> so, database connection was hosed.  we try to render an oops page in the same transaction
[12:12] <SteveA> but the transaction won't accept any more statements/queries
[12:12] <SteveA> we didn't yet load the Person who is logged in
[12:13] <SteveA> so we get an error when trying to display information about the user
[12:13] <sabdfl> interesting
[12:13] <SteveA> what we should do is start a totally fresh transaction for the error handling
[12:13] <SteveA> including clearing caches etc.
[12:13] <sabdfl> it seems that SQLobject happily passed None to a field which had been told should be notNull
[12:13] <SteveA> that way, the error page will always render
[12:13] <SteveA> nice.  a whole bunch of bugs.
[12:14] <SteveA> I'll file a bug about error handling and transactions, anyway.
[12:14] <SteveA> lunching.  back later.
[12:27] <Kamion> Has stub been around at all recently?
[12:31] <Kamion> I need somebody to do the DELETE of a bogus distrorelease that kiko mailed the launchpad list last night about
[12:32] <Kamion> because its presence is crashing the publisher in a way I don't think can easily be worked around
[12:38] <Kamion> (very urgent, unfortunately)
[12:42] <elmo> 'cos it's a weekend and we can't expect our DBAs to be around, I'm going to cowboy this on production - I've tested it in a transaction and 'dapper.0' is referenceless, so it's a single delete which I can undo if it causes problems
[01:25] <sabdfl> yeeeeHAAA!
[01:25] <sabdfl> go elmo
[01:32] <quail> evening all
[02:00] <plhardy> sadbfl: hello i am brand new and wanted to join ubuntu team and i wanted especially you to convince me to do so.
[02:01] <plhardy> too bad he's already away
[02:09] <dsas> plhardy: The best way to "join" the ubuntu team is to just start contributing to things and helping out. See http://www.ubuntu.com/community/participate
[02:09] <dsas> (incidentally, this is probably the wrong channel for discussing such things) 
[02:20] <lifeless> dsas: rather than saying its the wrong channel, you might try saying '#ubuntu-motu is a channel with many people that would love to talk about that' or something
[02:21] <lifeless> dsas: just saying its the wrong channel is not all that helpful
[02:21] <dsas> lifeless: I thought of that, but it depends on how you want to get involved. The page I linked to links to team specific pages which tell you the right IRC channel.
[02:22] <sabdfl> plhardy: welcome aboard :-)
[02:23] <sabdfl> SteveA: is there good documentation for converting tests from the older format to the new one?
[02:38] <kiko-zzz> sabdfl, yeah, there's a wiki doc.
[02:39] <kiko-zzz> lemme find it
[02:40] <kiko-zzz> https://launchpad.canonical.com/PageTests?highlight=%28test%29
[02:53] <sabdfl> thanks kiko
[02:53] <kiko-zzz> enjoy
[02:54] <sabdfl> what's the standard way to test for a page that should require login?
[02:54] <sabdfl> let me know and i'll update the doc
[02:54] <sabdfl>  (different to testing if your login actually has permission to see a particular page)
[02:58] <kiko-zzz> sabdfl, log in using anon_browser and verify the redirect to +login?
[02:59] <sabdfl> i guess...
[02:59] <kiko-zzz> seems to me like the only way to actually test it
[03:03] <sabdfl> damn, this is a very nice system
[03:05] <kiko-zzz> sabdfl, it's super-nice actually
[03:05] <kiko-zzz> okay catch you later time to hit the mountains
[03:46] <sabdfl> anybody know how to get a particular form with the testbrowser, when there is no name or id on the form?
[04:20] <Mez> hey, what prequisites do you need to work on a team branch ?
[04:21] <Mez> and are branch pushes on the supermirror as they used to be?
[04:22] <sabdfl> Mez: no prereqs, and yes I believe everything is as it was
[04:22] <Mez> sabdfl: including the amount of time to wait between pushing and seeing it via HTTP ?
[04:22] <Mez> (and I mean - for a team branch - you have to be in the team to push to it i presume?)
[04:22] <Mez> oh, and hi btwe
[04:24] <LarstiQ> yes, I do believe the time before it is mirrored should decrease
[04:24] <LarstiQ> but it works fine ime
[04:25] <Mez> the old supermirror time to mirror was a joke :P
[05:02] <Mez> anyone able to remove stuff from the bazaar? it seems a push failed and it wont let me push again 
[05:02] <Mez> lots of errors
[05:27] <Mez> who actually runs the supermirror
[05:28] <LarstiQ> Mez: you should be able to use an sftp client to move stuff out of the way
[05:28] <LarstiQ> Mez: if that doesn't work, ask one of the launchpad admins
[05:29] <Mez> what stuff do I need to move out of the way
[05:29] <LarstiQ> just the branch dir should suffice
[05:33] <Mez> how do i get rid of it it's telling me it's failing
[05:35] <LarstiQ> what is the exact command and error you are getting?
[05:36] <Mez> bzr: ERROR: File exists: '/~katapult-dev/katapult/0.3.x-dev': mkdir failed: unable to mkdir
[05:37] <LarstiQ> that should not be a problem if you remove the '0.3.x-dev' dir using a normal sftp client (as bundeled with openssh)
[05:37] <LarstiQ> however, it looks like katapult might have set up some vcs-imports or the like
[05:38] <LarstiQ> https://launchpad.net/people/katapult-dev/+branch/katapult/0.3.x-dev
[05:38] <Mez> hmm
[05:38] <Mez> how do i do an rm -rf on sftp ?
[05:39] <Mez> I dont have access to delete it
[05:40] <LarstiQ> hmm, I can't find that either
[05:40] <Mez> I dont have access to delete it just delete the files underneath it
[05:41] <LarstiQ> Mez: I don't see one of the lp devs that could do this for you around right now
[05:42] <LarstiQ> stub would be the one to ask I think, untill it is possible to do so yourself
[05:43] <LarstiQ> Mez: perhaps a support ticket is the best option for now?
[05:43] <Mez> hmmles
[05:43] <Mez> it's ok I know why
[05:43] <Mez> I just need a client that has an ability to remove lots of things
[05:51] <Mez> bah
[05:52] <Kamion> Mez: one moment
[05:53] <Kamion> Mez: hmm, ok, neither sftp nor psftp (putty-tools) seem to be able to do that
[05:53] <Kamion> Mez: try mounting it using Places -> Connect to Server and using nautilus?
[05:53] <Kamion> failing that at least sftp and psftp both support globbing in the rm command, even if not recursion
[05:55] <Mez> Kamion - globbing worked... but still wont let me remove the dir :(
[05:55] <Mez> globbing doesnt work for rmdir though
[05:56] <Mez> oh, and Kamion: I'm not a Gnome User + I'm at an internet cafe and they wont let me use linux here
[05:56] <Kamion> psftp does support rmdir globbing
[05:56] <Kamion> at least according to http://the.earth.li/~sgtatham/putty/0.58/htmldoc/Chapter6.html#psftp
[05:56] <Kamion> you should be able to get psftp for Windows
[05:57] <Mez> yeah - using psftp - sftp onlinux doesnt
[05:57] <Mez> (ssh -> sftp)
[05:57] <Kamion> apt-get install putty-tools <- psftp on Linux
[05:57] <Mez> but meh - I still dont have permission
[05:57] <Mez> Kamion: not my server i was connectign to
[05:57] <Kamion> extract the binary from the .deb then :)
[05:58] <Mez> Kamion: I'm lazy - If I wanted to I could login to a pbuildf, install it on that and use it from there :D
[05:58] <Mez> but I'm lazy
[05:58] <Mez> anyways ... I got to go
[06:18] <plhardy> i saw a specification https://launchpad.net/distros/ubuntu/+spec/old-kernels-removal i think it should make clear that no user compiled is ever removed with that method.
[06:18] <plhardy> i don't now how to add this comment on the spec without any rights...
[06:21] <Kamion> (a) there's little point until it's fleshed out with a wiki page; subscribe to it and you'll be told if it gets anywhere beyond a braindump; (b) I think that spec might well be superseded by something like dependency-removal although I'd have to go hunting to be sure; (c) that's an Ubuntu specification, so is not really on-topic for #launchpad - this channel's for the infrastructure
[06:23] <plhardy> you (a) point answerd was then on-topic ? 
[06:24] <Kamion> *shrug*
[06:26] <plhardy> i ever do many mistakes but normaly never twice the same :-)
[06:45] <LarstiQ> https://launchpad.canonical.com/MaloneReleaseManagement does not seem publically viewable :(
[07:18] <sabdfl> LarstiQ: what do you need to know there?
[07:18] <sabdfl> happy to tell you what that's about
[07:20] <sabdfl> it's to make it easier for the distro team to track bugs that are important for them to fix before a release
[07:20] <LarstiQ> sabdfl: currently, our workflow with bzr is to keep track of bugs in the 'current' version, and will close them if fixed there. This doesn't allow for bugs to be kept for different versions.
[07:20] <sabdfl> and the same for upstream
[07:20] <sabdfl> ok, this is what you will want, then
[07:20] <LarstiQ> possibly series and milestones already do this, but I wanted to look at that spec
[07:21] <sabdfl> you create a series, then you can nominate and approve/decline bugs to be fixed in that series
[07:21] <LarstiQ> and if malone didn't have it planned yet, it might be an idea to start one (perhaps based on the debian bts)
[07:21] <sabdfl> start one what?
[07:21] <LarstiQ> spec
[07:21] <sabdfl> i think this will do what you want
[07:21] <sabdfl> it's pretty simple
[07:21] <LarstiQ> sabdfl: so, will that enable bugs to stay open for 0.8, and be closed for 0.9?
[07:22] <sabdfl> yes
[07:22] <LarstiQ> great :)
[07:22] <sabdfl> you can track bugs in 0.8 and 0.9 separately
[07:22] <sabdfl> and trunk, of course
[07:24] <LarstiQ> sabdfl: having that concern adressed, is there any reason to not have all launchpad specs public?
[07:24] <LarstiQ> or well, the wiki content
[07:24] <LarstiQ> perhaps to motivate work on having the wiki be superseded by blueprint itself?
[07:25] <sabdfl> LarstiQ: there is private code in some of the specs, and features that are confidential
[07:25] <sabdfl> how else are we to surprise our users at christmas?
[07:26] <LarstiQ> hehe :)
[07:26] <LarstiQ> fair enough
[10:48] <lucasvo> is it possible to select a i18n-po file for rosetta out of a mirrored bzr branch?