=== cinerama_ is now known as cinerama
wgrantjcsackett: Around?00:27
wgrantlifeless: Hi.00:28
wgrantlifeless: Given the utterly disastrous state of QA over the past few days, I am tempted to revert the bad picker fixes.00:29
lifelessdid jons branch fail ?00:31
lifelessI agree we despearately need to get stable00:31
StevenKJon's branch hasn't even been reviewed00:31
wgrantJon's branch is not reviewed, not ec2'd, and not necessarily going to work, and we've already got three rollbacks in the queue.00:32
StevenKAnd 3 more revs to QA00:32
wgrantA fourth to get us back to a reasonable state cannot hurt.00:32
* StevenK finds his rope.00:32
wgrantAnd it's not even easy QA.00:33
lifelessI like rollbacks :)00:37
wgrantOoh, it still reverts cleanly.00:40
* wgrant reverts cleanly.00:40
StevenKwgrant: Are you going to self-review it?00:41
wgrantIt is done.00:45
lifelesswgrant: hi; rabbit00:51
lifelesswgrant: did you have some success ?00:51
wgrantNot really.00:51
lifelesswith it in vms specifically00:51
wgrantI suspect I just need to poke /etc/hosts harder.00:52
lifelessI'm really disappointed with rabbit ease-of-use in this area00:53
wgrantIndeed. It's entirely unnecessary, too.00:54
wgrantYay, create-lxc-aufs still works.00:54
wgrantOK, got it working.00:57
wgrant127.0.0.1 localhost lucid-lp-temp-aufs-7YHq00:57
wgrantDoesn't work00:57
wgrant127.0.0.1 localhost00:57
wgrant127.0.1.1 lucid-lp-temp-aufs-7YHq00:57
wgrantBecause it is hideous and likes to resolve its name both ways :)00:58
timrcwgrant, hey, cody-somerville said you rang00:58
wgrantIt's like kerberos, except worse.00:58
lifelesswgrant: what do you mean?00:58
wgrantHmm, I wonder if it's our fixture, actually.00:59
lifelesswell, u1 had the same ugly01:00
lifelessand we got our fixture's core from u101:00
lifelessbut I'd like to understand what you mean01:00
lifelessdoes it manually parse /etc/hosts ?01:00
lifelessor is it binding to for some bizarre reason?01:00
wgrantNeed to experiment.01:00
lifelessmy lxc doesn't have in 'ip addr'01:01
wgrantIt's not going to.01:01
lifeless    inet scope host lo01:01
wgrantWell, that.01:01
lifelessthe /8 only affects routing01:01
wgrantAnd it will resolve your hostname.01:01
lifeless127.0.1.1 isn't usable unless the address is added somewhere01:02
wgrantYes it is.01:02
lifelessor something daft is going on01:02
wgrant1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN01:03
wgrant    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:0001:03
wgrant    inet scope host lo01:03
wgrant    inet6 ::1/128 scope host01:03
wgrant       valid_lft forever preferred_lft forever01:03
wgrant127.0.0.1/8 is on the interface.01:03
wgrantlo takes the whole /801:04
wgrantAlways has.01:04
wgrantAnd rabbit or the fixture somehow need to be able to resolve it both ways.01:04
lifelesssearching for lo address gets a cee lo news clip ;P01:08
lifelessI'm not sure lo always bound to all the the 127 addresses, but meh01:09
lifelesswgrant: whats odd/confusing is why it *wants*
wgrant127.0.1.1 has been in the default installation forever.01:10
lifelesswgrant: e.g. why its not happy with
lifelesswgrant: no, it hasn't01:10
wgrantlifeless: reverse resolves to localhost first.01:10
lifelesswgrant: feisty/etch brought it in.01:10
lifelessah, so its whinging because it gets a different name back ?01:11
wgrantI presume so.01:12
wgrantIt's possibly the fixture being crap.01:12
StevenK*However*, Feisty only added on new installs01:12
wgrantNot even 12000 OOPSes from yesterday's amusement.01:16
wgrantHow boring.01:16
LPCIBotProject devel build #865: STILL FAILING in 5 hr 50 min: https://lpci.wedontsleep.org/job/devel/865/01:17
mwhudsontotally irrelevant, but apparently the thing is different on os x01:24
=== poolie_ is now known as poolie
wgrantlifeless: Hi.02:12
wgrantlifeless: The problem goes away if we pass fq_nodename instead of just nodename in as RABBITMQ_NODENAME.02:14
wgrantlifeless: rabbitmq will use nodename@hostname if it's specified, otherwise it will do a reverse lookup to guess hostname.02:15
StevenKSo it's a very small patch?02:15
wgrantThat's how the system one works: /usr/lib/rabbitmq/bin/rabbitmq-env constructs rabbit@hostname02:15
wgrantThree characters.02:15
wgrantI think rabbitfixture also leaks tmpdirs.02:19
wgrantNeed to check that out.02:19
wgrantLike, my laptop had several thousand by the end of last week.02:20
pooliehi StevenK, wgrant, lifeless02:21
StevenKO hai!02:21
wgrantHi poolie.02:21
wgrantMm, doesn't seem to leak normally.02:29
wgrantI guess I just killed it too aggressively.02:29
StevenKA thousand times or more02:29
lifelesspoolie: hi02:39
lifelesswgrant: cool02:39
StevenKOrsum. bzrlib.lockable_files._LockWarner is now affecting ~1,500 tests on Jenkins03:08
spivUpgrade your bzrlib :P03:12
StevenKspiv: That's with bzr 2.3.3-0~bazaar1~lucid203:14
spivStevenK: The relevant change landed on trunk about a week ago03:15
lifelessI think I'm blind03:15
lifelessspiv: we run releases in lp though03:16
StevenKI don't think I want to run LP tests against bzr trunk :-)03:16
lifelessspiv: so I think 'release 2.4 please'03:16
lifelessspiv: is a prerequisite03:16
wgrantWe often run betas.03:16
lifelessthe more bzrlib changes things the less I think that that is a good idea03:16
lifelesswhat I mean is that the old one-line-of-releases, each month is as stable as possible, deprecations are always graceful : well that was more amenable to using recent developments.03:17
lifelessits now harder to upgrade bzrlibs03:17
lifelessand we can't be as confident about e.g. wire level support issues till the final release in the series03:18
lifelessI'm not saying this is a problem per se03:19
lifelessbut its a consequence of the changed policies03:19
lifelesszope.testing tests make my eyes bleed03:19
wgrantWhat are you doing to it?03:20
lifelesssee my last comment in https://code.launchpad.net/~lifeless/launchpad/storylayers/+merge/6673803:20
lifelessI am teaching it that if it gets a test with no layer, it should not do layer per-test teardown and setup, regardless of whatever layers are setup.03:20
lifelessits that03:21
lifelessor I can insert an adapting TestResult03:21
lifelessbetween the DocFileTests in a StoryPageTestCase and the zope.testing runner03:21
lifelessbut the adapter would need to merge all the failures in all the elements of the story03:22
lifelessI think this is probably a lesser evil03:22
lifelessanyhow, we have 6 failures in our fork already03:22
lifelessand no idea if they are bitrot, changes, or what have you03:22
lifelessso I'm going to close my eyes and JFDI03:22
benonsoftwareQuestion: What exactly is the Launchpad Community dev team? https://edge.launchpad.net/~launchpad-dev03:31
lifelessits a mailing list03:31
benonsoftwareJust a list?03:31
lifelessyou are welcome to join if you want to discuss the development of launchpad03:31
benonsoftwareI'm alreaady a member on it.03:32
poolielifeless, it would be nice if there was some clearer "what is this team for" concept03:33
pooliesome of them you're welcome to join if you're just interested; some are only for trusted peolpe03:34
poolieperhaps they shouldn't all be presented as teams03:34
=== almaisan-away is now known as al-maisan
benonsoftwareif it just a list they should just have a list section03:34
pooliethere is a section about it being a list03:35
benonsoftwareWhat does biab mean?03:37
lifeless'Back In A Bit'03:38
StevenKspiv: Out of interest, can you link me the fix for that?03:51
spivStevenK: http://bazaar.launchpad.net/~bzr-pqm/bzr/bzr.dev/revision/600203:52
StevenKspiv: That sounds like a good plan.03:53
wgrantProbably makes the GC a bit happier.03:54
spivHmm, probably not so much.  I doubt our lock objects were participating in reference cycles.03:55
lifelessI suspect we have bugs causing us to see the warnings03:57
lifelessok, so this route isn't going to work03:59
lifelessI'll do a big-arse adapter instead.03:59
spivTo adapt it to a smaller arse?04:00
lifelessif only04:01
lifelessthis is z.testing04:01
* StevenK attempts to QA stuff04:23
StevenKwgrant: Can haz blessing to update DF?04:27
wgrantStevenK: doit04:28
StevenKwgrant: I think bug 740208 shouldn't be to bad to QA04:29
StevenKIt's not very clear from the code if it's an e-mail that's in a title or something04:29
wgrantI checked it for descriptions earlier.04:30
wgrantIt seems to be OK.04:30
wgrantBut eeeh.04:30
StevenKwgrant: So it looks fine, you're just nervous about it?04:54
wgrantStevenK: Yes.04:56
wgrantAlso considering tracking down whoever decided that erlang extensions should automatically download their dependencies from git/hg.04:56
wgrantMy eyes04:57
StevenKwgrant: Do you think it's fine to deploy if we keep on an eye on it, or not?04:58
wgrantStevenK: I think it's OK.04:58
wgrantgit clone http://github.com/mochi/mochiweb.git04:58
wgrantCheck out the 'rebar' file.04:58
wgrantrebar seems to be some kind of extension build system.04:58
wgrantBut look at what the file is.04:58
StevenKI'd rather not touch a git repo :-P05:00
StevenKwgrant: And we can probably deploy jtv's change, since it's FF'd05:00
StevenKrvba's change is running glaciall^Won mawson05:00
mwhudsonwgrant: it looks lovely in github's browser, that file05:01
StevenKmwhudson: URL?05:01
LPCIBotProject db-devel build #700: STILL FAILING in 5 hr 40 min: https://lpci.wedontsleep.org/job/db-devel/700/05:01
mwhudsonStevenK: https://github.com/mochi/mochiweb/blob/master/rebar05:01
mwhudson'blob' has rarely been more appropriate05:01
StevenKRight. Now to review my lunch.05:02
ajmitchthat blob looks like a .zip file05:02
wgrantIt is.05:02
mwhudsonit's something erlang related isn't it?05:02
wgrantWhat else...05:02
StevenKwgrant: So, the IDSJ runner just finished on mawson. It looks like it got past the problematic bit, so I'll qa-ok it, but there's another bug there.05:04
* StevenK files it05:04
wgrantStevenK: k05:06
StevenKFirefox, how you annoy me05:10
StevenKWho broke click handling between Firefox 4 and 5, seriously?05:10
StevenKWe are probably deployable in 2 hours05:11
wgrantAlmost there.05:14
=== al-maisan is now known as almaisan-away
wgrantBah, bzr-hg fail.05:32
StevenKWhen r13376 hits qas, we can deploy 42 revisions05:33
wgrantbzr-hg fails to work with rabbitmq-mochiweb.05:35
wgrantIt works with the rest :(05:35
pooliefile a bug, please?05:35
poolieor, find one?05:35
pooliewhich branch?05:36
wgrantCan't branch it directly.05:37
wgrantBut if I try to branch a clone, it does a bit and then:05:37
wgrantbzr: ERROR: An inconsistent delta was supplied involving '<unknown>', 'hg:test'05:37
wgrantreason: Parent is not present in resulting inventory.05:37
StevenKRight, buildbot done, waiting for qas05:43
StevenKwgrant: I have prepared a bug list, too05:44
StevenK42 revisions, 28 bugs05:44
wgrantLess excellent.05:46
pooliewgrant, this rings a bell05:49
pooliewgrant, https://bugs.launchpad.net/bzr-hg/+bug/51336805:50
_mup_Bug #513368: inconsistent delta during fetch <code-import> <Bazaar Hg Plugin:Triaged> <Launchpad itself:Triaged> < https://launchpad.net/bugs/513368 >05:50
wgrantpoolie: Thanks.05:50
* StevenK whimpers at qastaging05:57
StevenKHow did you miss 13376?! HOW?05:58
wgrantbuildbot-poller being slow, maybe?06:01
=== lifeless changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: - | Critical bugs: 204 - 0:[######=_]:256
StevenKwgrant: But surely it pulls from stable?06:03
StevenKBut, perhaps06:04
lifelessit does06:04
lifelessbut perhaps you've forgotten the 30 minute build of the tree on carob06:05
StevenKWhy does stable need to build on carob?06:07
StevenKSince the first thing qas is run buildouyt06:07
wgrantStevenK: buildbot-poller pushes stable.06:09
wgrantStevenK: So buildbot completing is not enough to update stable.06:09
wgrantYou have to hope that buildbot-poller sees it finish in a timely manner.06:10
StevenKRight, so moving to Jenkins might actually solve this -- since my plan is that when devel builds successfully it will push to stable/db-devel as the last step06:10
wgrantJenkins shouldn't have permission to do that, I don't think.06:11
wgrantBut maybe.06:11
StevenKWell, Jenkins tell tarmac to do it06:11
StevenKUtterly kill buildbot-poller06:12
wgrant3/4 deps packaged...06:16
wgrantAll of which originally grabbed $VCS checkouts of all of their dependencies06:17
LPCIBotProject devel build #866: STILL FAILING in 5 hr 52 min: https://lpci.wedontsleep.org/job/devel/866/07:09
wgrantPackaged rabbitmq-management stack working.07:13
pooliegood night all07:25
wgrantMorning bigjools.07:40
bigjoolswgrant: did you see the QueueInconsistentStateError oops?07:44
wgrantbigjools: Yes, I identified the bug and Steve got the PU rejected.07:45
bigjoolswgrant: good.  is it a new bug?07:45
bigjoolslooked like it was PPA though07:46
wgrantbigjools: Yes. OEM copied a package into a new PPA before it was private (the setup script hadn't called lp_save()), which triggered a delayed copy. It was then rerun with lp_save() added, so it was made private and recopied directly.07:46
wgrantThe delayed copy processed a couple of minutes later, but the direct copy had already been done.07:47
bigjoolsawesome :/07:47
bigjoolsthe story is ....07:47
bigjoolsdelayed copies need to die07:47
wgrantIn other news, I've packaged rabbitmq-management. It's a plugin that provides an HTTP API which will allow the test suite to quickly reset rabbit.07:49
wgrantSo we can have test isolation.07:49
wgrantWhich will be nice.07:49
allenapMorning all.08:04
rvbalifeless: Hi, I modified the code you rejected earlier today to throw a new exception with an additional message instead of using "print" (https://code.launchpad.net/~rvb/launchpad/rabbit-error-plugins/+merge/67000) ... if you think it's still clunky, I'll just forget about it but allenap and I though it would be nice to get a warning about potentially conflicting rabbit plugins.08:09
adeuringgood morning08:10
lifelessrvba: I think there is a better way to tackle this; I will describe it in the proposal08:10
wgrantrvba: We're going to depend on rabbitmq-management soon, so I'll be working out how to run multiple instances tomorrow, hopefully.08:10
wgrantBut there may be others.08:10
rvbalifeless: ok thanks.08:11
StevenKrvba: O hai. I QA'd your change, but hit https://bugs.launchpad.net/launchpad/+bug/80630708:11
_mup_Bug #806307: IDS has no access for DistributionSourcePackage <derivation> <Launchpad itself:Triaged> < https://launchpad.net/bugs/806307 >08:11
rvbawgrant: I suppose the 'conflict' is caused by multiple rabbit instances trying to use the very same port.08:11
wgrantrvba: Right, I need to work out how to configure that.08:12
wgrantNot sure if we can do it through an envvar like with the AMQP listener.08:12
wgrantrvba: Are you running natty?08:12
rvbawgrant: yes.08:13
wgrantNot sure if I want to backport natty's rabbitmq 2.3.1 to everywhere, or take 2.5.0 from oneiric.08:13
wgrant2.5.0 has some major plugin build system changes.08:13
wgrant2.3.1 has everything we need.08:13
lifelessrvba: done08:14
lifelesswgrant: think CATS08:14
wgrantI forget how to check what's there.08:15
wgrantIs there a madison around somewhere?08:15
lifelesspresumably apt-cache search on carob08:15
wgrantOr must I FTP internally?08:15
wgrantcarob only sees stock lucid rabbitmq-server.08:15
rvbalifeless: thanks ... I have used getDetails too ;) ... I guess I'll wait to see what wgrant does about this problem, if he manages to fix this in a generic fashion, the additional warning will be useless.08:15
wgrantPerhaps it's in some other pocket.08:15
* StevenK kills parallel-test and disables the job again08:16
StevenKRunning for 26 hours makes me SAD08:16
lifelessStevenK: sob, thanks.08:16
LPCIBotProject parallel-test build #85: ABORTED in 1 day 2 hr: https://lpci.wedontsleep.org/job/parallel-test/85/08:16
lifelessStevenK: any evidence of how it failed/08:16
rvbaStevenK: it's a good sign that the initialisation started. Thanks for QAing my change.08:16
LPCIBotProject parallel-test build #86: ABORTED in 29 sec: https://lpci.wedontsleep.org/job/parallel-test/86/08:17
LPCIBot* Launchpad Patch Queue Manager: [r=wgrant][rollback=13334] Revert r13334 due to bug #806179.08:17
LPCIBot* Launchpad Patch Queue Manager: [r=gmb][no-qa] New lp.app.longpoll package.08:17
LPCIBot* Launchpad Patch Queue Manager: [r=gmb][bug=805547] Fix packageset's copy in initialise_distroseries.08:17
_mup_Bug #806179: PersonPickers broken from bad passed in config <Launchpad itself:In Progress by jcsackett> < https://launchpad.net/bugs/806179 >08:17
LPCIBot* Launchpad Patch Queue Manager: [r=deryck][bug=740208] obfuscated email addresses for anonymous08:17
LPCIBotwebservice requests08:17
LPCIBot* Launchpad Patch Queue Manager: [r=gmb][bug=802840] Ignore None component/section on sync upload08:17
LPCIBotsource override.08:17
LPCIBot* Launchpad Patch Queue Manager: [r=stevenk][rollback=13356] Rollback revision 13356.08:17
StevenKBah, sorry.08:17
* StevenK smacks LPCIBot.08:17
wgrantlifeless: Possibly interesting is that one of my parallel LXC workers hangs most times.08:17
wgrantlifeless: Only one.08:18
lifelesswgrant: fun08:18
wgrantlifeless: And the whole suite passes if not run with --parallel08:18
lifelesswgrant: you're doing aufs isolation right ?08:18
wgrantlifeless: This is with aufs, yes.08:18
lifelesswgrant: that jenkins job isn't, so its much more fragile08:18
wgrantrvba: You installed rabbitmq-management by grabbing the .ez files from the website and installing them manually?08:19
rvbawgrant: Yes.08:20
rvbaInstalling=moving the files to /usr/lib/rabbitmq/lib/rabbitmq_server-version/plugins08:21
=== almaisan-away is now known as al-maisan
wgrantrvba: How is your JS branch going?08:35
rvbawgrant: the generic part is up for review. https://code.launchpad.net/~rvb/launchpad/lp-app-longpoll-js/+merge/6693608:36
wgrantAh, excellent.08:36
rvbaThe remaining parts (that I shall address after the DD work is finished) are:08:36
rvba- fix up a service for the async frontend08:37
rvba- use the generic stuff in the MP page (this will be just cleaning up the prototype that we've done in Dublin)08:37
wgrantlifeless: Only 2.2.0 is in lucid-cat-landscape. That's just new enough that I can probably get this plugin stack backported to it. But natty's 2.3.1 backports to lucid fine without even a rebuild, and I'm not sure if 2.2.0 will work for us.08:44
wgrantI guess I should test.08:44
jtvAnyone mind if I kick dogfood?  bigjools, StevenK, wgrant?08:51
wgrantNot I08:51
bigjoolsjtv: are you updating it?08:51
bigjoolscool, was gonna do that anyway08:52
StevenKI updated it a few hours ago08:52
StevenKSo you may not need to08:52
bigjoolsjtv: remember which user you're supposed to use ;)08:52
jtvI made doing it right easier than doing it wrong, remember?08:52
lifelessjtv: hi08:52
jtvhi lifeless — saw your note, thanks for that.08:52
lifelessjtv: I'd love to know what I missed08:52
lifelessjtv: (I'm assuming you tried that and it didn't work...)08:53
jtvOh, no.  My query was just a quick scribble on a late-night train.  :)08:53
lifelessI have an observation about postgresql - if there is an ORDER BY matching an index, pg will almost exclusively use it08:54
jtvWhich makes a lot of sense.08:54
lifelessso if you order by id with a date constraint, it will use the primary key index08:54
jtvYes, I was simply too focused on getting to the id side of things when I wrote that version.08:55
lifelessjtv: I think the heuristic is too strong, but I haven't had the time to dig into what/why is going on08:55
jtvWell... it can save you some very painful plan nodes such as hash joins.08:55
lifelessjtv: anyhow, it means that I now have a spinal reflex: to force an index, order by it08:55
=== jtv is now known as jtv-eat
LPCIBotProject db-devel build #701: STILL FAILING in 5 hr 41 min: https://lpci.wedontsleep.org/job/db-devel/701/10:43
=== jtv-eat is now known as jtv
=== al-maisan is now known as almaisan-away
jtvlifeless: I think I just remembered why I wrote that query the way I did… it performs reasonably for reasonable inputs even without the new index.  The simpler version is abysmal without the index no matter what the input is.  I think I'll land it my way in devel and then do it your way in db-devel after adding the index.11:23
jtvNo, that wasn't it.  :(11:25
jtvIt's been said so many times: if only the statistics would recognize monotonic columns.  :(11:27
bigjoolsjtv: fancy doing a review?11:28
jtvbigjools: if it's not too big11:29
bigjools~280 lines11:29
bigjoolsbut a lot of refactoring11:29
* jtv has at it11:30
bigjoolsjtv: the previous_series stuff that I mentioned earlier11:30
jtvah yes11:30
lifelessjtv: we can add the index live11:30
jtvoh that'd be great11:30
jtvdo we have a Procedure for that yet?11:31
lifelessjtv: land it on devel with a non -0 patch number11:31
bigjoolslifeless: slony copes with index changes now?11:31
lifelessnormal review (stub + me for db, launchpad-reviewers for the code changes), and stub or a losa can create the index live at any point11:31
jtvSo the -0 patches go out as db-devel rollouts?11:31
spivdanilos: I see https://code.launchpad.net/~danilo/launchpad/expander-anim isn't proposed for merging yet... what's the delay?  https://code.launchpad.net/~spiv/launchpad/bmp-inline-diffs/+merge/66634 depends on it.11:32
lifelessso the plan is to make -0 verboten11:32
spivdanilos: I'd like to get my shiny diffs landed :)11:32
lifelessjtv: https://dev.launchpad.net/Database/LivePatching11:32
danilosspiv, I'll be getting it landed today, some tests missing still11:32
jtvlifeless: ta11:32
lifelessbigjools: slony doesn't see the index11:32
spivdanilos: woo!11:32
lifelessbigjools: what we do is CREATE INDEX CONCURRENTLY on all the replicates11:32
spivCool, I'll nag people to review mine tomorrow then.11:33
bigjoolsah ok11:33
lifelessmaster first if its an index that can fail (e.g. UNIQUE), then the readonly replicas11:33
stubbigjools: Slony doesn't care about indexes. it does care about constraints, which are sometimes tied up with the indexes.11:33
* lifeless hands over to stub11:33
lifelessI get up in 6 hours :(11:33
bigjoolsfurry muff11:34
lifelessjtv: anyhow, moral of the story is the index can be made live right now if needed.11:34
jtvgreat news11:34
bigjoolslifeless: night, see you at 6am :)11:34
bigjoolsunless you meant move the meeting TODAY?11:35
lifelessbigjools: that would be wonderful but I left the suggestion too late I think :)11:35
* lifeless puts hand to forehead, palm out11:35
danilosspiv, also, fwiw, you can land your branch without depending on mine, and mine will seamlessly provide nice animations for you when it lands :)11:44
spivdanilos: well, mine merged from yours, so I can't easily land it without landing yours too :)11:47
=== almaisan-away is now known as al-maisan
=== matsubara-afk is now known as matsubara
LPCIBotProject devel build #867: STILL FAILING in 5 hr 47 min: https://lpci.wedontsleep.org/job/devel/867/12:57
deryckMorning, all.13:06
abentleyderyck: Morning.13:07
bigjools+N  dist/rabbitfixture-0.2.1.tar.gz13:07
bigjools+N  dist/rabbitfixture-0.3.tar.gz13:07
jcsackettmorning, all.13:12
jcsackettwgrant: you still up, by any chance?13:12
deryckMorning, jcsackett13:12
jtvbigjools: some problems with your branch I'm afraid.13:16
jtvSee MP.13:16
bigjoolsjtv: music to my shell-like13:17
jtvNot tufted?13:17
bigjoolsjtv: I don't understand your comment13:18
jtvArcher ref.  Think ocelot.13:18
wgrantjcsackett: Hi13:18
bigjoolsjtv: not that one, the one in the MP13:19
wgrantallenap: Do you have any more changes to rabbitfixture planned?13:19
jtvbigjools: which comment?  I've been a little distracted by something or other wrong with my innards, so I may be talking nonsense.13:19
jcsackettwgrant: hi. i see you pinged me last night--was it just about the rollbacks, or something else?13:19
bigjoolsjtv: "you seem to test for the presence of a release date as an implicit side effect of SQL comparison semantics for nulls"13:20
allenapwgrant: No, none yet. What have you got planned?13:20
bigjoolsjtv: also you didn't read my blurb properly13:20
jtvbigjools: oh, that one.  Well, you seem to rely on "DistroSeries.date_released < foo" to filter out DistroSeries whose date_released is null.13:20
wgrantjcsackett: Just the rollback.13:20
bigjoolselse you'd not have made the second comment :)13:20
jcsackettwgrant: dig.13:20
wgrantallenap: A trivial change to make it more robust against suboptimal /etc/hosts configs, and a more involved and as-yet unwritten one to use rabbitmq-management for cheap reset.13:21
bigjoolsjtv: self.context.series[0] is correct, see the "addendum"13:21
allenapwgrant: Tip top.13:21
bigjoolsjtv: ok so I need an extra "Not(DistroSeries.datereleased is None)" clause?13:23
wgrant!= None!13:23
wgrantStorm can't override is.13:23
wgrantSo that will, er, not do what you want.13:23
bigjoolsI hate Storm13:24
jtvIn Storm you say != None.13:24
bigjoolsI am going to start writing store.execute everywhere13:24
al-maisanbigjools: :)13:26
bigjoolsal-maisan: yes, something you can appreciate :)13:26
al-maisanwell, I had moments when I thought the same :)13:27
bigjoolshehe, I don't hate it really, but there are some weird bits13:28
al-maisanlike with every tool in existence13:28
deryckabentley, ping for new standup time.13:31
bigjoolsjtv: http://pastebin.ubuntu.com/638897/13:34
matsubaraanyone that could help me out with a test failure I got (https://pastebin.canonical.com/49400/) for this branch: https://code.launchpad.net/~matsubara/launchpad/39605-bugtask-tooltip/+merge/66928?13:34
matsubaraI'm not sure how to keep the query count under the limit13:34
=== sidnei-away is now known as sidnei
stubmatsubara: I'm seeing a number of Person rows being retrieved one at a time. Pulling them in a batch will shave off some queries.13:42
stubmatsubara: Similarly BugActivity, Milestone and BugWatch13:43
matsubarastub, hmm this is probably the for loop iterating on activities.order_by(Desc('datechanged')), right?13:44
stubMaybe some code traversing attributes causing late loading of database objects? Pull them into the cache earlier.13:44
stubmatsubara: I don't know. I just skimmed the SQL log looking for repeated stuff.13:44
matsubarastub, how can I pull everything I need on a single query?13:45
jtvstub o/13:48
stubperson_ids = [activity.personID for activity in activities if activity.target == targetname]13:48
stubstore.find(Person, Person.id.is_in(person_ids)13:48
stubmatsubara: That might do it13:48
stubjtv: o/13:48
jtvstub: just wrote a "lightweight patch" for the schema, but find that "make schema" breaks because of a CREATE INDEX CONCURRENTLY.  Should I do it conventionally (non-concurrently) in the patch?13:49
jtvAh, I think I see that in the documentation now.13:49
matsubarathanks stub, I'll give it a try13:49
stubjtv: Don't put the concurrently in there. I add it in when applying stuff manually13:49
jtvIt's a bit difficult to make out what's relevant for development and what for deployment.13:49
wgrantstub, matsubara: There's a helper for that: load_related.13:49
stubjtv: Just a normal patch ending with something other than -0. I do the rest.13:50
matsubarawgrant, cool. I'll look for it13:50
wgrantmatsubara: However, the important thing is not to do this in a single query, but in a constant number of queries.13:50
wgrantmatsubara: Which means that you're going to need to calculate it outside BugTask.13:50
wgrantSo the page doesn't scale by number of tasks.13:50
jtvstub: I don't see any other traces of lightweight patches in devel, of self-assigned revision numbers.  Am I doing it wrong or am I getting a "first post" on both?13:51
matsubarawgrant, are there some examples of how that's done somewhere else?13:51
wgrantmatsubara: There should be a fair bit of it in BugTaskVie13:52
stubor any other of the non -0 patches in the tree...13:53
jtvstub: I don't see any in devel though — I thought lifeless said it could land in devel.13:54
stubI don't think they can land in devel yet13:54
wgrantstub: Didn't you fix the hook?13:54
jtvMaybe we should just try.13:54
stubDid I?13:54
wgrantYou certainly asked a LOSA to do it a few weeks back...13:54
stubI must be running low on drugs13:55
wgrantOr maybe you only got hold of the config file.13:55
stubIt must have been terribly exciting. I vaguely recall getting pqm tweaked a while a go but don't recall what the change wass13:56
wgrantStatus: resolved13:56
stubSo yeah, what my frontal lobe extension said.14:00
jtvstub: can I have db review?  https://code.launchpad.net/~jtv/launchpad/index-798297/+merge/6704614:02
stubjtv: message__datecreated__idx is preferred naming14:03
jtvstub: thanks & sorry14:03
jtvstub: done14:04
stubjtv: So the basic reason we need this is that ORDER BY id DESC sucks when filtering the date range?14:05
jtvstub: not quite — but it is one of those correlated-columns problems.14:06
jtvI'm trying to find the oldest Message before a particular timestamp.14:06
stubOldest message in what context?14:07
jtvSorry, newest.14:07
jtvI use the id of the newest older-than-X Message to filter the message FKs on DSDComment.14:07
jtvbigjools: no test for what happens when distroseries.datereleased is null?14:08
bigjoolsjtv: look again14:08
bigjoolsI inserted an extra series, it is not returned14:09
stubjtv: The id of the newest older-than-X message will either be max(id) or NULL14:09
jtvstub: yes, the code is prepared for that14:09
jtvbigjools: argh, that's hard to see!  Implicit is better than explicit etc.14:10
bigjoolsjtv: how explicit do you want?!  It tests that only stuff with datereleased is returned14:10
stubjtv: I'm just wondering then why you can't do max(id) and throw away the result if it is younger than the cuttoff.14:11
jtvbigjools: which is the one that isn't returned based on this — ds3 or ds4?14:11
jtvstub: You mean max(id) WHERE datecreated < foo?14:12
bigjoolsjtv: I don't think you understand what is going on here14:12
jtvProbably not.  Sorry.14:12
stubjtv: no, max(id). and if the corresponding datecreated is too young replace it with NULL14:12
bigjoolsjtv: I shall add more comments and then it will become clear, this is my fault14:12
jtvbigjools: no need to go all mea culpa.  :)  But I think it would be better to split these issues up into separate tests.14:14
jtvTranslations experience involves repeated headaches and nightmares caused by single-run, multiple-data tests.14:14
bigjoolsjtv: I don't think that will help, it will be the exact same tests over and over14:14
jtvstub: I'm afraid I don't follow…14:15
bigjoolsin fact they'll be slightly less useful as they won't test real world data14:15
jtvDistroSeries relationships aren't that intricate, are they?  I would think what the factory spits out is good enough to test for this.14:16
bigjoolsit is not14:16
bigjoolsthe factory is naive14:16
bigjoolson top of this, pulseaudio is pissing me about14:16
jtvIt likes that.14:17
bigjoolswhat's the ui thing that configs it?14:17
bigjoolsI am stuck with output going to speakers even when a headset is plugged in14:17
jtvErr… KDE Sound Preferences?14:18
* jtv is blatantly bluffing14:18
stub'the id of the newest older-than-X message' == message = store.find(Message, ...).order_by(Desc(id)).first(); if message.date_created < X: message = None14:19
jtvMe, I just shout a lot and hope some neighbour will call the other and say "please tell your friend to be quiet, he keeps shouting PLEASE RESTART DOGFOOD like some crazy person."14:19
jmlbigjools: it's sound preferences in normal Ubuntu. 'pavucontrol' is the one provided by pulseaudio itself.14:19
stubnow I'm confused.14:20
bigjoolsjml: thanks, I'll try it14:20
jtvstub: I suspect that I wasn't clear enough in explaining.14:20
jcsackettis anyone available to review https://code.launchpad.net/~jcsackett/launchpad/button-configs-break-pickers/+merge/67047?14:20
jcsacketti am happy to review something in exchange. :-)14:21
stubMy antihistamines more likely. Too much grass here.14:21
bigjoolsjml: unfortunately no luck.  There is something else somewhere that tells PA what devices to use IIRC14:22
nigelbbigjools: skype?14:22
bigjoolsnigelb: no, as soon as it detects PA it won't use anything else14:23
bigjoolsand now phonon won't show me anything else14:23
bigjoolsthis is infuriating14:23
nigelbbigjools: see if you can find dtchen in ubuntu-devel, he probably knows more about sound than almost most of us14:23
jtvstub: I think you're querying for the newest Message, or None if that one is too new.  I'm looking for the newest Message that is older than a specific date.14:23
jmlbigjools: pavucontrol normally does that on the Playback & Recording tabs. I find it hard to believe that Kubuntu shipped without an equivalent14:23
bigjoolsnigelb: thanks14:23
stubjtv: Right, so we need the index. yes.14:24
bigjoolsjml: gah, you have to be *in a call" for the option to appear in pavucontrol14:24
jtv(Since DSDComments are a new thing, I was tempted to hard-code a cutoff number as a shortcut :-)14:24
jtvWhat the zark is pavu?14:24
bigjoolskubuntu does it differently14:25
* jtv is amazed14:25
jmlbigjools: that sucks. I reckon it would be worthwhile seeking out whatever the KDE equivalent of Gnome Sound Preferences is14:25
* deryck switches work locations, back soon14:25
jmlbigjools: because Sound Preferences actually works in natty.14:25
bigjoolsjml: it's the config for phonon - it has a preferential list of devices to use in different scenarios (comms, multimedia, games etc) but they all got replaced by just "pulseaudio14:25
bigjoolsso it's a lot better than pavucontrol - when it ****ing works!14:26
jmlbigjools: is phonon the central sound control for kde?14:26
jmlbigjools: ahh.14:26
bigjoolsI need a phonon expert I think14:26
jmlbigjools: see, the gnome sound preferences now (afaict) actually configures pa, and everything uses pa, so it's all win.14:26
jmlbigjools: *nod*14:27
nigelbjml: When do you make the switch out of LP?14:27
bigjoolsthe phonon config is supposed to show all the pulse devices too14:27
jmlnigelb: already done14:27
nigelbjml: Ah! Nice :)14:28
jmlbigjools: so it's buggy?14:28
bigjoolsand of course, the next call goes back to the speakers14:28
bigjoolsfuck sake14:29
nigelbThis sounds like the time my mic took input from my speaker.14:29
bigjoolsjml: well it might be - it was all fine a couple of days ago :/14:29
=== abentley changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: abentley | Critical bugs: 204 - 0:[######=_]:256
benjiI need to prepare.  How about I call you in about 5 minutes?  Skype?14:43
flacostebenji: sure14:43
benjiflacoste: what is your skype user name?14:44
flacostebenji: fjlacoste14:44
=== al-maisan is now known as almaisan-away
danilosspiv, to be able to get my branches landed, I had to remove some bits that are in your branches as well (eg. the branch.revisionexpander.js stuff); expect conflicts soon :)15:00
jcsackettabentley: have room in your queue for https://code.launchpad.net/~jcsackett/launchpad/button-configs-break-pickers/+merge/67047?15:00
abentleyjcsackett: sure.15:00
jtvdanilos: getting more fallout from that Google bug we discussed.  It's like half of India is trying to create mail accounts for me.15:05
jtv(I know, half of India is a lot of people but… ever break a spammer's mailbox and see their ability to send email crumble to dust?)15:06
abentleyderyck: I am looking into https://code.launchpad.net/~matsubara/launchpad/39605-bugtask-tooltip/+merge/66928 and BugActivity scares the hell out of me.15:20
matsubaraabentley,thanks for looking that merge proposal but adeuring already reviewed it15:20
deryckabentley, yeah, it can be a bit hairy.15:21
matsubarait wouldn't hurt to have some additional feedback though :-)15:21
deryckabentley, not to mention, it's such a pain to get any useful info from it.15:21
abentleymatsubara: why did adeuring review it?15:22
adeuringabentley: matsubara asked me :)15:23
matsubaraabentley, I asked him on #launchpad. I looked at the topic and saw his name there and thought he was the OCR15:23
abentleymatsubara: ah.  No, I am.15:23
matsubaraby the time I noticed that, adeuring already had done the review15:24
abentleymatsubara: adeuring's review looks similar to what I was in the middle of writing.15:24
matsubaracool. I'm working on the changes15:24
abentleymatsubara: This is what I was going to propose: http://pastebin.ubuntu.com/638927/15:26
abentleymatsubara: Actually, ISTM that you should just grab all bug activities and find the latest for a given whatchanged string.  You can answer all information about what changed then without further queries.15:31
abentleymatsubara: Since I assume you care about all bug tasks on that bug.15:31
danilosjtv, yay google ;)15:34
abentleyjcsackett: Was "hacro" a deliberate combination of "macro" and "hack"?15:37
jcsackettabentley: as i don't remember ever typing that, i would say it's more likely a typo. but if you like it, we can pretend it was intentional. :-P15:38
jcsackettabentley: i think it's probably best that i fix that comment, though. :-)15:38
abentleyjcsackett: also the "a a".15:40
jcsackettabentley: indeed.15:40
abentleyjcsackett: bzr gannotate shows that the unused mailing_list is hysterical raisins.15:45
abentleyjcsackett: I'm not very comfortable with having "true" and "false" as Python member variables.  Seems like that would be hard to use in Python code.15:53
abentleyjcsackett: I would think you could calculate the whole config server-side and then simplejson.dumps it.15:54
jcsackettabentley: i suppose you could--this branch is just a quick fix to get the other personpicker stuff fixed quickly. bug 799847 concerns redoing the form-macros stuff so we just put important things in JSONCache and then don't need to do any view/js passing.15:59
_mup_Bug #799847: form-picker-macros should not do a bunch of manual js assembly <person-picker> <Launchpad itself:Triaged> < https://launchpad.net/bugs/799847 >15:59
jcsackettat which point the simplejson.dump would also be unnecessary.16:00
abentleyjcsackett: I don't really understand how you'd use the jsoncache that way.  Would you have one JSONCACHE entry per picker?16:01
adeuringabentley: your paste is roughly what I had in mind :) But I would remove a given attribute from the set "attributes" once it is found and terminate the loop when the set "attributes" becomes empty16:02
jcsacketthm. crap, that's actually fair point. benji suggested it once upon a time, but we were looking at a case with only one picker on the page.16:03
jcsackettstill, i think cleaner form-macro stuff is more in scope for that bug (which i'll be addressing sometime shortly).16:03
jcsackettin my mind, your suggestion (or other implementations) fall in there.16:03
abentleyjcsackett: Okay, fair enough.16:03
jcsackettabentley: i may well be pinging you to pick your brain for preimp on that bug, if you're game for it.16:04
abentleyjcsackett: sure.16:04
=== beuno is now known as beuno-lunch
abentleyjcsackett: This should have tests, though.  I don't think I can approve it without tests that would fail if the Python True and False were used.16:07
jcsackettabentley: sure, though i'm not sure how to test this without using windmill or something to do app-server + yui.16:08
jcsackettabentley: and to my knowledge, we don't have the new implementation for those tests in place yet, do we?16:08
abentleyjcsackett: You should be able to render it and check for the strings, no?16:09
* jcsackett headdesks.16:09
jcsackettyeah, actually, i can just do a view test.16:09
jcsackettabentley: i will do that. i'm not sure why i was thinking only YUI stuff. :-P16:09
abentleyjcsackett: At 192 of the full diff, it says "instead if" but should say "instead of".16:11
jcsackettabentley: ack.16:11
abentleyjcsackett: Your XXX should also follow https://dev.launchpad.net/PolicyandProcess/XXXPolicy (currently it lacks a bug)16:12
jcsackettabentley: so it does.16:12
jcsacketti'll fix that as well.16:12
jcsackettin the YUI knowledge sharing thing right now, so changes won't be made until after that.16:13
abentleyThanks.  Ping me again when it's updated, and I'm sure we can wrap it quickly.16:13
jcsackettabentley: sounds good, thanks. :-)16:13
LPCIBotProject db-devel build #702: STILL FAILING in 5 hr 43 min: https://lpci.wedontsleep.org/job/db-devel/702/16:26
=== salgado is now known as salgado-doctor
=== beuno-lunch is now known as beuno
cr3hi folks, has anyone tried building launchpad on oneiric? I'm wondering if you also get this error, probably because of spriteutils: IOError: decoder zip not available18:06
=== salgado-doctor is now known as salgado
deryckcr3, I know sinzui runs on oneiric.  I don't know if he had that issues.18:33
deryckcr3, unfortunately, he's not around this week.18:33
jcsackettabentley: i have had time to circle back to this mornings review and fix the branch: https://code.launchpad.net/~jcsackett/launchpad/button-configs-break-pickers/+merge/6704718:42
abentleyjcsackett: ACK.18:44
abentleyjcsackett: r=me18:47
jcsackettabentley: thanks!18:47
jcsackettlifeless: ping.18:48
jcsacketthi, lifeless. i've got a question about how to go about landing my fix branch regarding bug 801388 and bug 806179. yesterday you told me to land with rollback, but wgrant already rolled back that branch.18:50
_mup_Bug #801388: some person pickers show "assign me"/"remove assignee" when that makes no sense <bad-commit-13334> <disclosure> <person-picker> <qa-bad> <ui> <Launchpad itself:Fix Committed by wallyworld> < https://launchpad.net/bugs/801388 >18:50
_mup_Bug #806179: PersonPickers broken from bad passed in config <Launchpad itself:In Progress by jcsackett> < https://launchpad.net/bugs/806179 >18:50
jcsackettso now my branch includes the fixes that wallyworld put in--should i just link it to both bugs and land normally?18:51
lifelessjcsackett: right, so you can just land normally once you've bypassed the merge.18:51
jcsackettdo i need to remove the qa-bad tag on 801388, or will qatagger handle all of that later?18:51
lifelesswe've deployed past it, and its ok (so aren't rolling back)18:52
lifelessso you can drop the bad-commit tag if you want18:52
jcsackettok. thanks.18:52
lifelessqatagger will replace the qa-bad tag when it sees the commit18:52
lifelessyou can drop it by hand too if you want, but its not needed18:52
cr3deryck: I finally managed to reproduce on oneiric server but it worked fine on oneiric desktop, strange18:52
lifelesscr3: missing dep in the brand new world of onei?19:20
abentleyjcsackett: I don't understand the purpose of https://code.launchpad.net/~jcsackett/lpreview-body/body-callback-template/+merge/6612519:21
cr3lifeless: seems hairier than that, installing ubuntu-desktop didn't solve the problem but I could've sworn spriteutils worked when I installed a fresh desktop19:21
jcsackettabentley: it may not be a great piece of code--it was my first stab at plugin code. in short, it was to allow a means of specifying a different template for merge proposals than the one that's hardcoded currently.19:22
abentleyjcsackett: Right, but why?19:22
jcsackettfor example, i prefer reST style headers, so i've put together a template that does that.19:22
abentleyjcsackett: We have a standard template that's supposed to be used, and that's what it provides.  If we change to reST headers, then we should update it to use that all the time.19:23
jcsackettabentley: fair point. i didn't realize the current format was mandated, given many people use different ones. (e.g. sinzui uses one that specifies his rules for a completed branch).19:24
abentleyjcsackett: it may be honoured more in breach than observance, but yes, that's the standard.19:24
jcsackettabentley: dig. than obviously the merge proposal should not be accepted. :-)19:25
abentleyjcsackett: thanks.  Rejected.19:28
jcsackettis there a clear demo of using the expander widget in lp.app.javascript anywhere? i can't find anything that uses it...19:53
jcsackettbut i suspect my grep fu is week.19:53
dobeyderyck: boo. my bug is not low. :(20:04
deryckdobey, then go fix it. We're open source. ;)20:05
deryckdobey, Seriously, though, I'm sorry this disappoints you. We have way more bugs then we'll ever fix and until this happens more frequently or causes more problems I doubt it will be more than a low bug.20:06
dobeyderyck: if this was the matrix, and i could inject all the knowledge of the launchpad source directly into my skull, i totally would :)20:06
deryckdobey, fair enough :)20:06
dobeyderyck: well, it actually breaks a feature in tarmac, and therefore blocks branch landing :(20:07
deryckdobey, sounds like a bug in tarmac to me ;)20:07
dobeyhow is launchpad returning bad data, a bug in tarmac?20:08
deryckdobey, that was largely a joke.20:08
deryckdobey, though I'm not clear why a person etag matters for landing branches.20:08
dobeyjokes don't work on irc, unless it's "that's what she said" :)20:09
deryckdobey, why does tarmac care about the etag?  is it changing something about a person and needs to verify before landing a branch?20:09
dobeyderyck: because we check that people have signed the contributor agreement, via team membership, where cla is required. and lazr.restfulclient checks the etag in its Entry.__eq__ override20:10
dobeyderyck: no, we're not changing the person at all. my bot user doesn't have those privs :)20:10
dobeybut the eq compares self_link, http_etag, and an internal dirty dictionary, and etag being different causes it to fail20:11
dobeysince apparently the person != the person20:11
deryckdobey, so tarmac code is not referencing the two versions of people, lazr.restfulclient is checking the team member against the person collection for you?20:12
lifelessdobey: perhaps you could include your sample code?20:12
dobeywell, in tarmac we're getting both objects and just doing ==20:12
_mup_Bug #806163: Different http_etag for same person resource when accessed via people collection or team object <api> <Launchpad itself:Triaged> < https://launchpad.net/bugs/806163 >20:14
dobeythe subteam == person there is failing, when it shouldn't be, because of the etag issue20:15
deryckdobey, so check on ID, not object.  bug in tarmac. ;)20:16
deryckdobey, again a partial joke.  just check on something unique for each user, like self_link or id (if we expose that) rather than doing an object to object comparison.20:21
dobeywell, yes, i could work around it, and have temporarily. but workarounds suck20:22
dobeybut users' accounts being broken is, i think, a more important issue. i'm surprised more people aren't hitting this or similar situations on LP20:24
dobeyalthough, i don't know why exactly we haven't had an issue until now20:24
=== matsubara is now known as matsubara-afk
=== abentley changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: - | Critical bugs: 204 - 0:[######=_]:256
wgrantMorning all.21:08
deryckLater on, everyone.21:09
LPCIBotProject devel build #868: STILL FAILING in 6 hr 12 min: https://lpci.wedontsleep.org/job/devel/868/21:14
=== salgado is now known as salgado-afk
jcsackettlater, all.21:56
nhandlerNot sure how many people are subscribed to debian-derivatives@lists.debian.org, but there was just some discussion about possibly getting lintian to be run against ubuntu packages in Launchpad. I just thought it was worth noting here in case anyone with more experience is interested in getting involved in the discussion (although it will probably get brought to your attention soon enough)22:33
=== poolie_ is now known as poolie

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!