/srv/irclogs.ubuntu.com/2010/12/17/#launchpad-dev.txt

wallyworldcan anyone tell me, given a BranchRevision object, the correct way to figure out if the revision was from merging in another branch and what that other branch is? Do you look for some attribute of any revision parents?03:40
StevenKwallyworld: wgrant says "lol"03:42
wallyworldwhy lol?03:42
wallyworlddid i say something funny?03:42
wgrantYou may be able to use look at the revision's parents' getBranch method.03:43
wgrants/look at //03:43
wallyworldthat's what i'm doing now - i get the first parent and use getBranch() on that03:43
wallyworldi wasn't sure if that was the correct thing to do or not03:43
wallyworldi don't have any good data to check03:44
wallyworldthat my theory is correct that it is the right thing to do03:44
lifelessI suggest discussing this with tim / a bazaar dev03:44
lifelessI would, but its too much like work03:44
wallyworldlifeless: and you are on HOLIDAYS03:45
lifelessexactly03:45
wgrant /abr lifeless03:45
lifelessabr03:45
lifeless?03:45
wallyworldlifeless: so shut down your irc client :-)03:45
StevenKOUT, DAMN SPOT03:45
wgrantlifeless: auto op, ban, remove, de-op.03:46
lifelesswgrant: thats an odd thing to suggest03:47
lifelesswallyworld: On leave != uninterested :)03:48
lifelesswallyworld: just relaxed :>03:48
wallyworldlifeless: well i hope you have a beer in hand or something :-)03:49
lifelessI'm playing wow :)03:49
lifelesswallyworld: while you are here03:50
lifelessbug 69134303:50
_mup_Bug #691343: OOPS page should reflect new bug filing location <Launchpad itself:New> < https://launchpad.net/bugs/691343 >03:50
lifelesswould be easy and extremely useful to fix :)03:50
wallyworldlifeless: ok. i'll take a look03:50
* wallyworld assigns bug ^^^^ to himself03:52
=== almaisan-away is now known as al-maisan
mrevellHello09:08
daniloswgrant, hi, are you perhaps still around?09:20
wgrantdanilos: Hi.10:05
wgrantdanilos: You're wondering about TTBJ.build?10:05
* StevenK looks for a reviewer in here as well ...10:06
StevenKwgrant: Doesn't that just link to the Build table?10:06
wgrantdanilos: I think it should link to the TTB.10:07
wgrantdanilos: But that whole part of the model is being ripped out when I get bored.10:07
wgrantWhich may end up being next week.10:07
wgrantStevenK: Down to 9.1EB yet?10:09
StevenK9223372036695 MB10:09
wgrant:(10:10
StevenKwgrant: You have 23 hours 55 minutes to get here and download the entire Internet10:10
lifelesswtf10:11
StevenKlifeless: So, internet here said I had 500MB download limit. I logged in again after it disconnected me for hitting the limit, and it seems my limit is now 9.2EB10:12
wgrantThe limit also decreased as we used it.10:12
wgrantIt seems to be counting properly.10:13
lifelesswheeee10:13
wgrantDespite being 9.2EB of 500MB.10:13
jpdsHow odd.10:13
StevenKI suspect an overflow bug of some kind10:13
lifeless=== Top 10 Time Out Counts by Page ID ===10:13
lifeless    Hard / Soft  Page ID10:13
lifeless      59 /  242  BugTask:+index10:13
lifeless      31 /  267  ProjectGroupSet:CollectionResource:#project_groups10:13
lifeless      25 / 3459  Archive:+index10:13
lifeless      25 /  289  Distribution:+bugs10:13
lifeless      18 /  237  POFile:+translate10:13
wgrantjpds: Yet remarkably convenient.10:13
daniloswgrant, that's right (sorry, phone and emails :)10:13
lifeless      12 /    1  Bug:EntryResource10:13
lifeless       8 /  120  Question:+index10:13
wgrantlifeless: Nooooooo.10:13
lifeless       6 /   18  DistroSeriesLanguage:+index10:13
lifeless       6 /    0  Person:+leave10:13
lifeless       5 /    2  Person:+bugs10:13
lifelessStevenK: overflow would count down :P10:13
StevenKHaha, Person:+leave10:14
StevenKlifeless: It does count down as it use it ...10:14
lifelesssorry, I meant up10:14
lifelessoverflows would be in 2's complement10:14
StevenKYes. It's very strange10:14
daniloswgrant, also, do you think we could just switch getCurrentBuildFarmJob to get the new-style jobs directly thus not needing to add it for the TranslationTemplateBuildJob?10:14
lifelessI suspect an unsigned where it should be signed ;)10:14
StevenKAnd I don't want to tell them, since then they might fix it10:15
lifelessmeep10:16
wgrantdanilos: If you can find a way to do that, sure.10:16
lifelessfilebug is borked in prod10:16
lifelesson chromium10:16
wgrantdanilos: I'm not sure it's possible yet.10:16
lifelessapparently if you click 'just so', worked when  retried.10:16
wgrantI forget if BuildFarmJob.builder is set during the build.10:17
wgrantAnd at 9:20pm after just getting home from a sprint, I'm not entirely sure that I want to find out.10:17
StevenKLike the sprint was that hard, you slacker10:18
lifelesswgrant: 900ms queries10:19
wgrantdanilos: It looks like BuildFarmJob.builder is not set during the build, so you probably can't do it.10:19
lifelesswgrant: *, well, 1510:19
wgrantdanilos: You have to go through the BuildFarmJobOld10:19
wgrantlifeless: I fail at interpreting OOPS reports. How do I find an OOPS for that pageid?10:20
lifelesswgrant: you start on the oops summary, sent to canonical-launchpad10:20
=== stub1 is now known as stub
lifelesswgrant: then, you look for bugs tagged timeout, and you find (just filed) bug 69147810:20
_mup_Bug #691478: Archive:+index timeouts <timeout> <Launchpad itself:Triaged> < https://launchpad.net/bugs/691478 >10:20
lifelesswgrant: and in there, click on the link :)10:20
daniloswgrant, tbh, I don't really understand most of this, so I am just seeing if there's anything that we really need to do to make it possible for you guys to finish the migration10:20
wgrantdanilos: I don't know exactly what is necessary, sorry. I need to swap that all back in.10:21
wgrantIt's been a few months.10:21
lifelesswgrant: oh, to find the entry in the report, search for the pageid in the summary report.10:21
wgrantlifeless: Hm, OK. Thanks.10:21
daniloswgrant, right, so perhaps just adding the build property to the TTBJ should solve what we can do on our side, and I'll let you worry about the rest10:21
wgrantlifeless: Ow.10:22
daniloswgrant, though, I am not sure how do we switch buildmaster to use new TTBs instead of TTBJs for translation jobs, but I guess that's easy enough10:22
daniloswgrant, also, what happens with all the BFJOld info like score and such, I don't see that in BFJs anymore10:23
wgrantdanilos: The new classes (TTB, BPB, SPRB, etc.) are mostly just used for storing data for histories.10:23
wgrantMost control is still performed by the BFJOld.10:23
wgrantThe score is on BuildQueue.10:24
wgrantI want to merge BuildQueue into BuildFarmJob, but some others do not.10:24
wgrantScore will remain there for now.10:24
wgrantAnd BuildQueue will link to BuildFarmJob.10:24
daniloswgrant, ah, ok, so we are still pretty far off from being able to completely switch to the new model if I got that correctly10:24
wgrantdanilos: Not *really*.10:24
wgrantJust needs some code moved around, basically.10:25
daniloswgrant, heh, ok, then I didn't get that correctly :)10:25
wgrantBut I need to work out what.10:25
wgrantThe job-specific schema changes are done.10:25
wgrantApart from deleting the old tables.10:25
daniloswgrant, right, so do you think it makes the most sense for you to deal with that? would it help if TTBJ.build was provided and pointed at TTB?10:25
wgrantdanilos: Do that for now, since it makes the buildd-manager log sad.10:26
daniloswgrant, ok, I hope I'll be able to come up with a nice simple query to get that (i.e. I am hoping a TTBJ has enough information to do it)10:26
daniloswgrant, thanks for the input10:27
wgrantdanilos: Sorry I can't be more help.10:27
wgrantBut uni pushed all that stuff way down.10:27
daniloswgrant, yeah, don't worry, it's just that _that_ bigjools guys is pushing us to implement stuff that I believe we mostly implemented :)10:28
wgrantdanilos: It's all a bit of a mess, yes.10:28
wgrantWe are really close now, though.10:28
daniloswgrant, and I don't know enough to tell him off, so I am trying to learn all about this in a few hours which turns out is not the simplest thing around ;)10:29
daniloswgrant, not to mention that TTBJ is at the same time a BFJOld and a BranchJob, just to make it more interesting :)10:29
wgrantdanilos: Yeah, I like to blissfully forget that detail.10:29
daniloshey bigjools, I was just gossiping about you, but if I tell you what it was, it won't be proper talk-behind-the-back stuff10:30
lifelesshmm, we should probably log returned rows in the oops query history10:30
wgrantlifeless: The number of returned rows?10:30
lifelessyeah10:30
wgrantIndeed.10:30
lifelessso we can see slow but fast per row10:30
lifelessvs fast but few rows10:31
lifelessvs slow and few rows10:31
daniloswgrant, yeah, I hate that, and I am hoping we can soon just make TTBJ a BranchJob and let any buildd stuff live in TTB10:31
wgrantdanilos: I will hopefully be deleting TTBJ.10:31
wgrantAlthough it may end up staying, but not as part of the build farm model.10:32
daniloswgrant, well, we can't delete the branchjob stuff, that's what schedules the build10:32
wgrantRight.10:32
daniloswgrant, right, we should just rename it10:32
wgrantTTB will probably know about it, but nothing else.10:32
lifelesswgrant: there wqere some dups10:32
lifelesswgrant: but only 2.5 seconds worth10:32
wgrantlifeless: yeah, max of 3 reps.10:33
wgrantNot 100.10:33
lifelesswgrant: max of 26 perhaps10:33
lifelessrow 4 of duplicates10:33
daniloswgrant, right, though perhaps not even that (i.e. we use it to add a TTB to the queue, but that should be it)10:33
lifelessah nope, wrong column10:33
lifelesswgrant: so I think we've still got a per-row-in-ui just-in-time query being triggered10:34
* bigjools is not here10:34
wgrantlifeless: Possibly. Can I see recent OOPSes for Archive:+packages somehow?10:34
wgrantI want to compare them to Archive:+index.10:35
lifelesswgrant: log into devpad, grep.10:35
wgrantYay.10:35
wgrantI will just trigger new ones instead.10:35
wgrantOn DF.10:35
lifelesswgrant: or10:35
lifelessOOPS-1811H189610:35
lifeless25 seconds!10:35
lifeless(top 10 durations in the lpnet summary)10:35
wgrantMinimal reps.10:36
wgrantJust... very slow.10:36
wgrantHm.10:36
wgrantThat OOPS must be wrong.10:36
* wgrant blames the GIL.10:36
wgrant5s gap after a 200ms query.10:36
lifelessthere's a bug for getting per thread times10:37
lifelessjust needs tuits10:37
wgrantAnd a couple of other similar ones.10:37
lifelesswgrant: large row counts coming back through storm can legitimately show up as that, fwiw10:37
wgrantOh, that +index OOPS is the same.10:37
wgrantMulti-second gaps after ~200ms queries that wouldn't return that many rows.10:37
lifelessheh10:38
wgrantLike <100.10:38
lifelesswell, someone should merge my single threaded appserver branch :)10:38
=== al-maisan is now known as almaisan-away
=== matsubara-afk is now known as matsubara
henningedanilos: do you know about LaunchpadWriteTarFile?11:38
daniloshenninge, now I do :)11:38
henningedanilos: it's in lp.translations.utilites.translation_export11:39
henningebut I think it belongs under lp.testing.11:39
daniloshenninge, right, and it might, though testing does sound weird, let me look11:39
henningedanilos: oh, right. I thought it was only used in tests11:40
daniloshenninge, it might make sense to move it to lp.services though11:40
henningebut the language pack exporter uses it, too.11:40
henningedanilos: yes, I agree.11:40
henningedanilos: I'll put it in lp.services.tarfile11:41
daniloshenninge, sure, though you might want to avoid using the same name as the global tarfile module imho :)11:42
henningeok11:42
henningelptarfile11:42
henninge?11:42
daniloshenninge, i.e. you can't do "from lp.services import tarfile" and "import tarfile" then11:42
daniloshenninge, I don't know, it's also only for writing11:43
henningeI would not recommend taht anyway11:43
henningeimporting the module, I mean.11:43
henningetarfile_helpers11:43
daniloshenninge, heh, I am not saying I'd recommend it, it's just that it'd be unobvious to a reader of the code which tarfile module is being used in that case11:43
henningetrue11:44
henningedanilos: how is tarfile_helpers?11:44
daniloshenninge, probably good enough, go with that and see if your reviewer thinks otherwise11:44
henningecool, thanks11:44
deryckMorning, all.11:59
maxb*blink*   Is PPA +copy-packages supposed to offer o-series and p-series as destinations?12:31
jelmermaxb, there's a bug open about that12:33
jelmermaxb, we previously didn't really support distro series that were present but not initialized12:34
=== almaisan-away is now known as al-maisan
gmbgary_poster: Are you the right person to talk to about making lxml a dependency for Launchpad?14:12
=== salgado is now known as salgado-brb
gary_postergmb, I guess14:12
gmbgary_poster: I ask because I've got a branch in review that depends on it; bac has pointed out that we don't necessarily have it on our appservers.14:13
gary_postergmb, lxml is notoriously annoying to build as a Python egg.  Is the Deb-packaged version acceptable?  Also...14:14
gmbgary_poster: There's now some discussion in launchpad-reviews about this...14:14
gary_posterISTR that we have had an lxml dependency in the past14:14
gmbORLY.14:14
gary_posteryeah, I saw, gmb, I'm not entirely sure where to write :-)14:14
gmbOver there. There are more people paying attention :)14:14
=== flacoste changed the topic of #launchpad-dev to: Launchpad Development Channel | BUG JAM! http://mumak.net/lp-bugjam-2010/ 73 so far | New starter this week: wgrant | PQM open for business | firefighting: - | Get the code: https:/​/​dev.launchpad.net/​Getting
maxbgmb: Remember to tag the launchpad-dependencies branch :-)14:59
gmbmaxb: Done, thanks for the reminder. There are two sets of instructions for maintaining that .deb and on of htem is wrong.15:03
gmbI'll update it now...15:03
maxbgmb: where is the second?15:03
gmbmaxb: Well, the first is here: https://dev.launchpad.net/Hacking#Changing%20the%20launchpad%20dependency%20debs (and was wrong). The second (correct; or at least more up-to-date) version is here: https://dev.launchpad.net/LaunchpadPpa#launchpad-dependencies15:05
gmbI'd got all the way through the process before finding hte second.15:05
gmbI've updated /Hacking to point at the LaunchpadPPA page.15:06
maxbhah, nice. I never knew of the first, when I wrote the second15:07
maxbNeither did anyone who was asking me if there were any instructions for doing it :-)15:07
gmbHeh.15:08
gmbWell, they were similar enough that I didn't screw things up. The first just used the old, non-bzr-builddeb way.15:08
=== salgado-brb is now known as salgado
henningedanilos: fxing bug 61912116:08
_mup_Bug #619121: Inactive template in rosetta: Confusing wording <bugjam2010> <lp-translations> <trivial> <ui> <Launchpad itself:In Progress by henninge> < https://launchpad.net/bugs/619121 >16:08
henningedanilos: do you agree with "Template active?" instead of "Accept translations?"16:08
daniloshenninge, yes16:18
daniloshenninge, or "Template is active"16:18
henningewithout the '?'16:19
henningeI think I like that even better16:19
daniloshenninge, I don't, I think it should just be "???"16:19
henninge;-)16:20
henningedanilos: bug 67082116:20
_mup_Bug #670821: translator comment formating lost <lp-translations> <trivial> <ui> <Launchpad itself:Triaged> < https://launchpad.net/bugs/670821 >16:20
henningedo you know of any reason not to use a <pre> here?16:20
daniloshenninge, well, one: comments in PO files commonly have some whitespace padding as in:16:21
danilos#.   This is all so cool16:21
danilos#.     that it doesn't make sense at all16:21
danilos(imagine those two were aligned: basically, we'd need equivalents for "rectangle whitespace" handling)16:22
daniloshenninge, but, it's probably an improvement to just use <pre>, which is why it's 'trivial' :)16:22
henningedanilos: but doesn't <pre> give us that?16:22
daniloshenninge, well, my point was that it'd be nice to strip initial spaces (same amount from every line) and then put it into <pre>16:23
daniloshenninge, i.e. think of a case like this as well:16:24
danilos#.    numbered items:16:24
danilos#.      1. one16:24
danilos#.      2. two16:24
daniloshenninge, it'd be nice if we stripped those 4 leading spaces from each line, but not a big deal imo16:24
henningeno, not really. ;)16:25
=== matsubara is now known as matsubara-afk
daniloshenninge, yeah, so go for the trivial :)16:25
henningewe can do that when we redesign the +translate page ... ;)16:25
daniloshenninge, heh, right, "when" :)16:25
henningemeaning 'if'?16:25
daniloshenninge, well, meaning "we" :)16:26
danilosor rather, "we?" :)16:26
henningethey!16:31
danilosMutatedCelebrityError :(17:27
=== deryck is now known as deryck[lunch]
maxbhah17:35
maxbthat is an AMAZING classname :-)17:35
danilos:)17:41
danilosI'd probably enjoy it more if we didn't get ~800 OOPSes with that :)17:41
benjiI'm reminded of the TemporalParadox raised when you tried to edit a past version of an object in the Zope 2 history tab.17:45
=== benji is now known as benji-lunch
=== benji-lunch is now known as benji
LPCIBotProject db-devel build (231): FAILURE in 3 hr 38 min: https://hudson.wedontsleep.org/job/db-devel/231/20:19
=== Ursinha is now known as Ursinha-afk
LPCIBotProject devel build (313): FAILURE in 3 hr 13 min: https://hudson.wedontsleep.org/job/devel/313/21:59
LPCIBotLaunchpad Patch Queue Manager: [r=bac][ui=none][bug=685624] Provide21:59
LPCIBotTranslationTemplatesBuildJob.build property linking to21:59
LPCIBotTranslationTemplatesBuild objects to stop polluting21:59
LPCIBotbuild-manager logs.21:59

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