/srv/irclogs.ubuntu.com/2010/11/01/#launchpad-dev.txt

wgrantWhat's happening with vostok these days?01:03
LPCIBotProject parallel-test build (12): STILL FAILING in 2 hr 11 min: https://hudson.wedontsleep.org/job/parallel-test/12/01:08
lifelessmwhudson: hey01:26
lifelessmwhudson: so in the person merge code01:26
lifelessthere is an XXX from you01:26
lifelessabout recipes and a function to write.01:26
lifelessbut no bug #01:26
mwhudsonah oops01:26
lifelessdid you forget to put the bug # in, or is the bug mythical? It seems like an important before-live-todo, to me01:26
mwhudsoni don't remember filing a bug :/01:27
lifelessperhaps you could ?01:27
mwhudsonthis is about person merge not blowing up if you merge two accounts that have a recipe of the same name01:27
lifelessI would assume so01:27
lifelessalso, I have sent a needs-to-be-digested mail to the list about code structure01:28
lifelessI'm totally sure its impenetrable due to jetlag :)01:28
mwhudsonwell, i'm pretty sure it would be to me :-)01:28
lifeless    Hard / Soft  Page ID02:02
lifeless     266 /    0  Archive:EntryResource:getBuildSummariesForSourceIds02:02
lifeless     162 /  260  BugTask:+index02:02
lifeless     133 /  161  CodeImportSchedulerApplication:CodeImportSchedulerAPI02:02
lifeless     121 /   43  Person:+commentedbugs02:02
lifeless     107 /  282  Distribution:+bugs02:02
lifeless     100 /    0  MailingListApplication:MailingListAPIView02:02
lifeless      29 /   53  Archive:+packages02:02
lifeless      26 / 1956  Archive:+index02:02
lifeless      21 /   25  Milestone:+index02:02
lifeless      15 /   85  POFile:+translate02:02
thumperlifeless: did arsenic get new code?02:15
thumperlifeless: MailingListApplication:MailingListAPIView lives there02:16
thumperlifeless: and while I did only one method, it should have helped02:16
thumperlifeless: see my note about getting the method used into the pageid for xmlrpc calls02:16
lifelessno02:17
lifelessits a SPOF at the moment02:17
lifelessthere is an RT to fix that02:18
lifelesswhen thats done, it will be in the nodowntime set02:18
lifelessthumper: I replied about the xmlrpc pageid02:18
lifelessthumper: 'yes we can change it, there is a bug already I think02:18
lifelessparaphrasing02:18
thumperlifeless: did you? must of missed it02:19
lifeless11:58 < lifeless> whats arsenic02:19
lifeless11:59 < lifeless> yes, there is a bug already02:19
thumperah02:19
thumperyes, see it now02:20
* thumper goes to make a coffee02:20
lifelesswgrant: when are your exams done with ?02:29
wgrantlifeless: 23rd.02:29
lifelessof nov?02:30
wgrantJa.02:31
lifelesskk02:31
lifelesslots of prep time then ;)02:31
lifelessgl02:31
wgrantThanks.02:31
thumperwgrant: how many exams?02:32
wgrantthumper: Just the three.02:32
thumpernot too bad then :)02:32
wgrantSince the fourth subject was the project, which lacks an exam.02:32
wgrantIndeed.02:33
wgrant10, 11, 2302:33
thumperI remember my first year exams, Saturday am, Tuesday am, Tuesday pm, Thursday am, Thursday pm, Friday pm, then two weeks break before the last two02:43
wgrantWeekend exams? Cruel.02:44
thumperyeah02:45
spmtry having your 21st birthday in the middle of exam week. :-(02:51
thumperspm: I got married the day before I turned 21 :)02:53
spmthumper: so what I'm hearing, unhappy for pretty much the same reason??? :-P02:53
thumperno comment02:53
spmheh02:53
thumperthere is something wrong if the builds are taking 7 hours :(02:57
EdwinGrubbslifeless: sinzui mentioned that you did some research into the whole distro milestone +index timeout when there are lots of bugs. It seems that it is just spending a ton of time creating all the storm objects. Do you have any other insight?03:46
* thumper marvels at the amount of yak hair in blueprints03:51
* thumper has the clippers out03:51
lifelessEdwinGrubbs: I setup eager loading03:54
lifelessEdwinGrubbs: for both bugs and milestones03:54
lifelessEdwinGrubbs: but I have only turned it on in one code path (to ensure I could actually deliver)03:55
lifelessEdwinGrubbs: which pageid / bug are you looking at.03:55
EdwinGrubbslifeless: https://lp-oops.canonical.com/oops.py/?oopsid=OOPS-1765ED125803:55
EdwinGrubbsbug 63892403:55
_mup_Bug #638924: Milestone:+index timeouts with many bugs <pg83> <timeout> <Launchpad Registry:In Progress by edwin-grubbs> <https://launchpad.net/bugs/638924>03:55
wgrantthumper: Was Blueprint touched significantly after its initial drop?03:59
thumperwgrant: not much no03:59
thumperwgrant: I now have a pipeline: ~/src/launchpad/blueprint-yak-shaving04:00
thumperif anything was a test for not needing a review, this is04:00
wgrantHeh.04:05
StevenKthumper: So, I've done something surprising.04:06
StevenKthumper: Robert Ancell was talking to me about specs not being exported over the API, so I made them be, in anger, on two plane flights.04:06
thumperStevenK: you've moved to NZ?04:06
lifelessthumper: its certainly a consideration04:07
lifelessStevenK: have you moved the security checks?04:07
StevenKlifeless: There's security checks?04:07
lifelessStevenK: james_w has a branch that exports specs over the API; its blocked on that change.04:07
StevenKRargh, didn't know that04:07
thumperStevenK: yeah, too many checks done in the browser code04:07
StevenKSpeaking of, I should charge my laptop04:07
thumperI've started shaving blueprints...04:07
thumperbut it will take some time to see the shape of things04:08
thumperunder all that hair :)04:08
thumperStevenK: but just think of the fun you had :)04:08
StevenKHah04:08
=== EdwinGrubbs is now known as Edwin-zz
=== Edwin-zz is now known as Edwin-zzz
StevenKIt wasn't that bad, I'm just curious why I could only get close, but not all the way04:08
lifelessStevenK: old LP code did security checks at the browser layer04:09
lifelessStevenK: blueprints is old LP code04:09
StevenK(I can grab a specifications object from lplib, and iterate over it, but it returns nothing)04:09
thumperheh04:10
thumperStevenK: it has an entry, but nothing exported04:10
thumperStevenK: mildly shit04:10
StevenKthumper, lifeless: To be perfectly honest, I did it for shits and giggles, while I was bored, so I'm happy enough to toss the work away.04:11
lifelessStevenK: no worries04:11
lifelessStevenK: I would love to see it happen04:11
StevenKthumper: Oh, I fixed that. I can post a diff if you have tissues to stem the bleeding04:11
thumperStevenK: I've more a problem with who sets what04:12
thumper:)04:12
StevenKI was also curious why I couldn't export Choice() elements where the vocab is in quotes, but can if it's a real object.04:12
StevenKRight, now I get what james_w is blocked on.04:13
lifelessEdwin-zzz: ah, yes04:13
lifelessEdwin-zzz: so thats high python time04:13
thumperStevenK: it'll have to do with the method trying to work out what is valid04:13
StevenKlifeless: When did you get home, by the way?04:13
lifelessEdwin-zzz: I'd get a profile trace of that on [qa]staging. Will mail you.04:13
StevenKthumper: Hm, I hadn't considered that, but it would make sense04:15
StevenKStarting to seriously consider a nanna-nap, and it's only 3:16pm04:16
lifelessStevenK: 10:20am arrival04:25
StevenKlifeless: At home, or the airport?04:25
lifelessjtv: so, we're experimenting in testtools and a few other projects with04:25
lifelessStevenK: airport, home @1104:25
StevenKlifeless: Landed at 0825, home at 114004:26
wgrantStevenK: The empty collection is probably because of the lack of a launchpad.View adapter for ISpecification.04:26
lifelessjtv: wishlist = fallback, medium = 'something we thought worked doesn't', critical = 'blocks project doing a release'04:26
jtvlifeless: the awkward marriage between the "problem reporting" and "task tracking" faces of bug trackers has bothered me for a long time, so I'm all on board… if you've been thinking about this, I'd be happy to introduce you to kirit for constructive discussion.04:26
lifelessjtv: I had a great discussion with persia - theres a commons problem too.04:26
lifelessbugs are global scope; work prioritisation is local (FSVO local) scope04:27
StevenKwgrant: Ahh. However, I'm now not going to talk about it, since james_w has already done work on it.04:27
stubspm: Do you know much about Puppet, and is it something we should be looking at for Launchpad configuration, scheduling or anything?04:27
lifelessstub: I know a bit :) - no, no and no.04:27
lifelessstub: the sysadmins will be using it to manage a growing set of services; for what we're doing/developing we should effectively ignore it, for now.04:28
stubwondering if it was useful for readonly switch, or LPCONFIG type stuff.04:28
lifelessit may be able to replace the deployment scripts eventually.04:29
lifelessdunno about the readonly angle; offhand anyway.04:29
lifelessperhaps setting a feature flag would be best for readonly.04:29
lifelessit would propogate to the slave, be found there and be locked on until someone pokes more directly. [/wildarse thought from jetlagged person]04:30
stub'cept feature flags are in the database, and changing to read-only mode switches to a database that was broken out of replication before the switch was set ;)04:30
lifelessstub: is that ordering changable ?04:30
stubNo - breaking the slave out of replication involves removing the replication triggers on every table and subsequent locking.04:31
jtvAnd I guess it complicates the case where you go to read-only because you lose the master.04:31
stubAnyway - we can't really ask the database if we are in read-only mode, because we need to know if we are in read-only mode to decide what database to talk to.04:32
lifelessgotcha04:34
lifelessthe ini file you added is probably simplest for now04:34
StevenKwgrant: And ohloh is still counting lines of source code, after 1 restart and, what, 3 weeks?04:34
spmstub: not vast amounts no; but will be in Syd next week doing a course on same.04:35
wgrantStevenK: At least it's on step 3 now!04:35
StevenKBut it has been for a week04:35
stubI think we should stop using Zope's email delivery, instead storing emails in the database and having a cronscript or daemon that does the delivery. Pretty trivial, easier to improve later with things like delivery failure, and sidesteps the SMTP ZCML gumph.04:35
wgrantStevenK: I think it's even slower than Launchpad git imports used to be.04:35
lifelessstub: I'd like to know whats up before we reimplement exim/postfix - if you see what I mean.04:36
StevenKstub: If you suggest writing EmailJob, I Will Find You04:36
stubNothing so complex :)04:37
stublifeless: At the moment appservers spool emails to the SMTP server during request. If we fix queued delivery, the appservers spool emails to the SMTP server in a separate thread. But why have the appservers do it at all?04:38
lifelessstub: the expansion from intent->mails I'm delighted to move out of webapp request handling04:39
lifelessstub: I don't really care if thats a Job or a thread.04:39
jtvspm: just realized the humorous chance I missed Sydney…  to go around telling everyone how great their city and country are and, if this is just a local capital, how incredible Canberra must be then.04:40
lifelessstub: mail sending is sometimes massively slower than expected, and we should fix that.04:40
lifelessstub: at the moment the appservers spool *after* the commit *before the request ends*04:40
lifelesswhich is a little odd :)04:40
stublifeless: Yes. In request delivery is slow, queued delivery is broken. We can fix queued delivery upstream, or move to a different model which I think is better in the long run.04:40
stubI'm suggesting still allowing the appservers to assemble the actual emails, but stuff them into the database rather than do the delivery themselves.04:41
stubProbably the same amount of work as fixing queued delivery.04:42
lifelessstub: perf wise I'd expect it to be slower04:45
lifelessstub: local mail spools /should/ be millisecond fast to accept things04:45
lifelessanyhow04:46
lifelesshalt()04:46
wgrantAnything sending more than a couple of emails from a request is probably a bug.04:46
stubpickle vs. store.execute()04:46
StevenKwgrant: Such as, say, updates to bugs?04:47
wgrantStevenK: They don't.04:47
wgrantStevenK: They use an external process.04:47
wgrantStevenK: But still manage to be mortifyingly slow.04:47
StevenKPity, I was hoping to be ironic04:47
stubwgrant: Agreed. But we can't control how long the SMTP server we are talking to will take before the email is handed off, unless we get a dedicated one we can turn off stuff like DNS checks etc.04:48
wgrantstub: True.04:48
lifelessstub: we have a dedicated SMTP server04:49
lifelessstub: *per appserver*04:49
stubWe talking to localhost now? Ok.04:49
lifelessyeah04:59
stubthumper: Just reading meeting notes. Thanks for reducing the mailman query counts! Its a big burden on the db.05:09
spmjtv: heh05:11
spmjtv: it's .... different. big country town with the amenities (to some degree) of a city.05:12
jtvYou're saying it's a bureaucratic amusement park at the end of a fat fibre-optic cable?05:12
spmclose05:13
spm"You're saying it's a bureaucratic amusement park at the end"05:13
jtvIt just became a whole lot less impressive.05:21
wgrantIt's Canberra. It is far less impressive than even your least impressive impression.05:22
spmsome of the more ... rigid bureaucratic side has changed, but it's still very much a town that revolves around federal govt. and that has implications right thru the entire place. unf, usually in painful ways.05:27
spmflip side. I'm a 20min drive from the city centre; yet from our back deck can see paddocks with sheep/horses across the valley. We get Kangaroos < 50metres from our front door (I have pics!). That's pretty special.05:28
mwhudsonhaven't we fixed this: http://launchpadlibrarian.net/58406579/vcs-imports-wordpress-trunk.log at least twice already?05:37
mwhudsonspm: ^^05:38
spmugh05:38
spmactually - I wonder - I came across this doing some simple bzr ops last week - don't recall which server offhand.05:38
spmmwhudson: which importd what that?05:39
spmwas* that05:39
mwhudsonspm: i'm a twenty minute _walk_ from wellington cbd and not much further from rural bliss than you :-p05:39
spm:-)05:39
mwhudsonspm: well, https://code.launchpad.net/~vcs-imports/wordpress/trunk05:39
mwhudsonspm: ahh, seems there's more than one kind of failure :(05:40
mwhudsonspm: seems it's galapagos that's doing the nowhoami thing05:40
spmta05:41
lifelessspm: I'm 20m from another entire city ;)05:48
thumperstub: np, I might tackle the methods one by one, but I want the xmlrpc server less loaded06:31
mwhudsonhttps://launchpad.net/~mwhudson/+karma -> "Michael Hudson-Doyle's karma has expired. "Total karma: 20838"07:43
mwhudsonum?07:43
henningeDoes anybody have an idea how we can get out of testfix again?07:43
henningemwhudson: could that mean "Some of you karma has expired?"07:44
mwhudsonmayhap07:44
wgrantIt can also mean that the karma cache is being updated.07:45
wgrantSometimes a couple of the karma totals will vanish for a few minutes while that happens.07:45
wgrantOr, in this case, all of them.07:45
mwhudsoncould be07:46
wgrantEveryone is similarly afflicted.07:46
wgrantWhich is rather suspicious indeed.07:46
jelmerIt looks like more windmill test failures.07:55
* jelmer waves to mwhudson, wgrant07:56
wgrantMorning jelmer.07:56
mrevellHello09:12
henningeHi mrevell!09:37
mrevellhello henninge09:38
henningemrevell: I just thought that you might not yet be aware of the fact that we moved the 10.11 release back by a week.09:38
wgrantThe guy in #launchpad looks like he could do with some commercial subscription help.09:39
henningemrevell: sorry for having kept you out of the loop here.09:39
mrevellhenninge, Thanks for letting me know now. So we're going for the 17th?09:40
henningewgrant: thanks, will talk to him.09:40
mrevellwgrant, thanks09:40
mrevellheh09:40
henningemrevell: yes, I just emailed the list.09:40
mrevellAh, I see your mail henninge09:40
henningemrevell: you go and talk ;)09:40
wgrantWhat's this December "Bug Jam" thing? Does the following cycle not start until January?09:41
mrevellhenninge, I saw a discussion where I think the release was going to 00.00 UTC. Is that still the case?09:42
henningemrevell: no, I am proposing 10:00 UTC09:43
mrevellwgrant, The idea was that rather than having a mini-cycle in December, we'd spend a couple of weeks doing nothing other than close bugs (not necessarily fix...)09:43
wgrantHeh.09:43
wgrantOK.09:43
henningemrevell: I had proposed 22 UTC for the 10th because I already knew that Tom would not be around.09:44
mrevellwgrant, So, we could just have one week of bug jam and a full-sized December cycle09:44
mrevellhenninge, Ah, I see.09:44
henningemrevell: but with more LOSAs not there, we scratched the 10th completely and since Tom is there on the 17th, we can do it at 10:0009:44
mrevellhenninge, Are we pretty confident of the date and time? I mean, can I announce it?09:44
henningemrevell: it's not finalized yet, no09:44
henningeoop, public channel ...09:44
henninge;-)09:45
henningebut I don't expect any more problems.09:45
mrevellheh, that's not a problem :)09:45
=== almaisan-away is now known as al-maisan
wgrantjelmer: Hi.10:51
mrevellHey, I'm running make on a new branch on a clean Maverick install ... I'm getting a permission error when I run make (clean, schema and run so far)10:51
mrevellAny ideas?10:51
jelmerwgrant: Hi10:51
StevenKmrevell: Can you pastebin the trace?10:52
wgrantjelmer: I've addressed your comment on https://code.edge.launchpad.net/~wgrant/launchpad/destroy-distroseries-lucilleconfig/+merge/38648. Anything else?10:52
jelmerwgrant: Yes, I need to land your branch. :-) I'll send it off to ec2 today.10:53
mrevellStevenK, Thanks, Here it is: http://pastebin.ubuntu.com/523776/ .... that's just the last few lines ... do you need the full thing?10:54
wgrantjelmer: Thanks!10:54
wgrantmrevell: I normally fix that with "sudo rm -r /var/tmp/bazaar.launchpad.dev".10:55
wgrantI think Apache creates it or something.10:55
mrevellOh, simple as that eh? Thanks wgrant, I'll give that a go.10:56
mrevellwhoop, that worked for make clean ... thanks wgrant and StevenK10:57
deryckMorning, all.11:15
=== al-maisan is now known as almaisan-away
=== mrevell is now known as mrevell-lunch
=== deryck_ is now known as deryck
=== mrevell-lunch is now known as mrevell
* lifeless yawns14:57
=== stub1 is now known as stub
=== almaisan-away is now known as al-maisan
lifelesshi jml15:25
jmllifeless: hi15:25
lifelessflacoste: jml: since we're all awake, perhaps we should continue our long running voice discussion15:27
jmllifeless: on another call right now.15:31
lifelesskk15:37
deryckgmb or allenap, could one of you spare time for an easy review?15:56
allenapderyck: Sure.15:56
deryckallenap, thanks!  https://code.edge.launchpad.net/~deryck/launchpad/allow-reporter-fixed-released-unsetting-664096/+merge/3975415:56
jmllifeless: maybe we could have a call later today16:03
lifelesssure16:03
jmlsinzui: hello16:06
sinzuihi jml16:07
jmlsinzui: when do you think we're supposed to have our call16:08
sinzuinext week -1 hour?16:08
jmlsinzui: I'm happy with that. Britain changed timezones during my flight home today16:09
=== benji is now known as benji-lunch
sinzuiokay16:09
jmlsinzui: so I'm not sure when anything is16:09
sinzuijml: you are my only meeting on Mondays. You can fit me in at your convenience16:10
jmlsinzui: cool.16:10
jmlsinzui: did you want to chat today?16:10
sinzuisure16:10
sinzuinow?16:10
jmlyeah.16:10
=== deryck is now known as deryck[lunch]
lifelessUrsula: hi16:22
lifelesshttps://devpad.canonical.com/~lpqateam/qa_reports/deployment-stable.html is empty?16:22
lifelessUrsula: bah, browser fail. ignore that question.16:22
lifelessUrsula: however16:24
lifelessUrsula: please do look - Revision 11801 is bad: possible blocker. is reported 4 or 5 times :)16:24
jcsackettgary_poster: hey, an LP user has asked me for any info on this issue in #launchpad https://answers.launchpad.net/moin-openid/+question/110123; you were looking at it at one point, any notions about it or who i might point the user to for more info?16:26
lifeless Bug 664060 needs QA16:29
_mup_Bug #664060: bug supervisor should be able to configure bugtracker for projects <qa-needstesting> <Launchpad Bugs:Fix Committed by brian-murray> <https://launchpad.net/bugs/664060>16:29
jmllifeless: so, now that edge is almost gone, are you going to tackle the other subdomains next?16:42
lifelessjml: thats a more complex question, and would benefit from specific discussion about it.16:43
lifelessjml: the complexities for it are:16:44
lifeless - are they part of the UI/ease of use story?16:44
lifeless - impact on performance (more domains == more SSL handshakes, but also potentially more parallelism [though we don't do this :P so its a bit of a no-brainer today]16:44
lifeless - impact on vostok16:45
lifelessjml: I have no immediate plans to do anything to/for/or about the bugs etc domains16:45
lifelessjml: but perhaps you meant some other set of subdomains16:45
lifelesshttps://bugs.launchpad.net/software-center-agent/+bug/627608 needs qa16:56
_mup_Bug #627608: Got a 401 on a fresh purchase <qa-needstesting> <Software Center Agent:Fix Released> <Soyuz:Fix Committed by michael.nelson> <software-center (Ubuntu):Fix Released> <https://launchpad.net/bugs/627608>16:56
lifelessjml: We did ~ 2200 commits to trunk in the last year.16:58
lifelessjml: our velocity in terms of commits-per-month for the last two months is about the same as for the last 12.16:59
lifelessjml: 20 hours per commit16:59
lifelessper dev16:59
lifeless(given a nominal 2000 hours/year work period)17:01
jmllifeless: interesting17:16
jmllifeless: because anecdotally I'd say committing to trunk is getting harder.17:16
lifelessjml: there sure is a lot of friction we need to fix.17:16
jmllifeless: fwiw, I meant the bugs etc subdomains.17:18
jmllifeless: but I do realize it's a big question.17:18
lifeless11824-7230 = 4594 over the last two years17:18
lifelessjml: I think the team size has grown a bit over two years?17:19
=== benji-lunch is now known as benji
=== deryck[lunch] is now known as deryck
=== lifeless changed the topic of #launchpad-dev to: Performance Tuesday | Launchpad Development Channel | Week 1 of 10.11 | PQM is open | firefighting: - | https:/​/​dev.launchpad.net/​ | Get the code: https:/​/​dev.launchpad.net/​Getting
jmllifeless: I'm not convinced it has. Anyway, it's easy enough to see how the number of distinct committers has changed17:25
lifelessjml: indeed, I'm doing very crude stats just now - log | less :)17:25
jmllifeless: well, easy-ish. bzr has a way to go before querying logs is as easy as SQL17:25
=== al-maisan is now known as almaisan-away
jmllifeless: reckon it's worthwhile to file bugs on the remaining parallel testing work?17:41
* jml gone, back later17:52
lifelessjml: not really; there is a parallel test job; looking at that will show whats broke18:03
lifelessstatik: shall we skip our call today? we caught up late last week ..18:15
statiklifeless: that sounds good to me18:16
SpamapSare there known bugs in reading archives of private mailing lists that a lp user is subscribed to?18:27
SpamapSI'm unable to pass the openid "Continue" button.18:27
lifelessugh18:29
lifelesshttp://python.mirocommunity.org/video/1186/djangocon-2009-django-is-obsol is quite interesting18:30
SpamapSIs this a follow up to iamcal's "Why I Hate Django" ? ;)18:31
lifelessno18:31
lifelessits about evolution18:31
SpamapSin a way, so was cal's talk18:32
lifelesswell, it might be then :P18:32
SpamapSbut, I'll refrain from commenting further until I've watched the presentation18:32
marslifeless, would be nice if there was a transcript there - it would be nice to just skim the facts of the argument18:37
SpamapShah.. onions spoil all kinds of apps18:38
SpamapSSome might say "minimize window" is an onion.18:38
lifelessSpamapS: you'll like the 'templates will finally die' line18:45
SpamapSI stopped at about 15 minutes.. I get his point.. rethink.. multi-threaded.. etc. ;)18:47
lifelessSpamapS: he switches themes18:47
SpamapSlifeless: even livejournal figured out that you can do the various bits of the template asynchronously server-side using gearman.18:47
SpamapSlifeless: oh very well.. starting back up.. ;)18:48
lifelesslol18:48
lifelessSpamapS: only if you're interested18:48
SpamapSI'm very interested18:48
SpamapSOne thing I am starting to think about.. is why hadoop is the only one anybody mentions18:48
SpamapSmap/reduce isn't that hard!18:49
lifelesshe mentioned cassandra dinomite etc18:49
SpamapSYeah I mean the map/reduce frameworks.18:49
SpamapSlifeless: so the last time I heard this argument, it was for xslt18:55
SpamapSCSS is obviously a much simpler way to do presentation than xslt..18:56
lifelessyeah18:56
lifelessflacoste: ping19:13
lifelessawol for a bit19:41
cr3leonardr: hi there, is there a way to define a method in lazr that returns void? I have a write_operation method which seems to cause a subsequent GET on the server because launchpadlib seems to do self.resource.lp_refresh() if the return code is 200 and the method is post20:05
cr3leonardr: I tried to redefine the method in question as a factory operation but then it expects the method to return the object created rather than void, which results in a ComponentLookupError for None20:06
leonardrcr3: https://bugs.launchpad.net/launchpadlib/+bug/24995020:06
_mup_Bug #249950: Stop refreshing after every POST named operation <launchpadlib :Triaged> <https://launchpad.net/bugs/249950>20:06
leonardrno :|20:07
cr3leonardr: thanks for the reference, good to know20:07
leonardrcr3, hopefully it's just an annoyance?20:08
cr3leonardr: a performance annoyance where each call results in two calls20:09
cr3leonardr: a crappy workaround might be to convert the method from a write_operation to a read_operation, but I'm not sure how much data I could encode in the url20:10
leonardrcr3: if it helps, i'm about to start on a big web service project and i could fix this problem as a side effect20:11
cr3leonardr: in the meanwhile, how acceptable would it be for me to define my own decorator called export_void_operation which would derive export_write_operation and setStatus(201) in the overridden call method?20:13
leonardrcr3: that's not good. you don't want to send 201 at all--you'd be lying20:14
leonardrhow bad a performance problem is this?20:14
leonardrif it's absolutely awful i can spend some time pretty soon figuring out the best way to fix it20:14
* mars \o/ devel is Green once again!20:15
cr3leonardr: it's not that awful, it's just that I know myself well enough that I'll probably forget to fix this later. I can take a note and wait, no worries20:16
leonardrcr3: what we want is something like export_void_operation + sending a Cache-Control header or something that says 'you just made the object dirty, you need to refresh'20:17
cr3leonardr: if I want to return an object which contains a reference to another object in the same call, can that be done?20:31
leonardrcr3: you mean send both objects at one?20:31
leonardrthat's not possible20:31
leonardrunless you can return a list of objects--that might work, but i  doubt it20:32
cr3leonardr: yeah, both at once to spare the subsequent get on the referenced object20:32
leonardrcr3: there's a general feature planned that will let a client tell the server it wants certain objects expanded, but it's not planned for a long time--i need to rationalize the web service design first20:33
cr3leonardr: maybe it's a lesson for me to create even coarser grained objects20:34
leonardrcr3: could be, if this is a really bad problem. would it help if this data was cached on the client side, or is even getting it once a big performance problem?20:35
lifelessleonardr: whats the big web service project?20:36
leonardrlifeless: refactoring the named operations into a few 'families' (search/filter operations and so on) so that each resource type in the web service doesn't have 5-20 miscellaneous and unstandardized operations dangling from it20:39
leonardrmake it easier to learn and use20:39
leonardri'm working with jml on it, but the desktop thing has intruded for several months20:39
jcsackettlifeless: per the db-patch rules, i believe i need a review from you on this: https://code.edge.launchpad.net/~jcsackett/launchpad/migrate-official-bool-data-627631/+merge/3958320:43
lifelessjcsackett: you need one from stub to assign the patch id, to land it20:44
jcsackettlifeless: i know, but the docs say i need one from the architect as well. that would be you. :-)20:44
lifelessyes20:44
lifelessbut not to land20:44
lifelessthe docs also say that :)20:45
jcsackettdig. i was just pointing out the mp to you. my understanding is that stub is out, so i'll have to chase him down later.20:45
jcsackettthere's already a request for him on the MP as well.20:45
lifelessthere is a 'request a review' buttton on the mp - my preference is for you to use that for me too :). Thanks for letting me know.20:46
jcsackettlifeless: i used that for you too, but it was while you were at uds, so this was sort of pushing it back into awareness. :-)20:46
wallyworld_rockstar: abentley: thumper: standup?21:06
EdwinGrubbsmatsubara-afk, Ursula: ping21:16
wgrantjelmer: Did you get around to EC2ing that branch?21:53
lifelesssinzui: hey22:11
thumperlifeless: so what is the process then for landing something without a review?22:31
thumperlifeless: I have my very boring enum change for blueprints22:32
lifelesshttps://dev.launchpad.net/PolicyAndProcess/OptionalReviews22:33
lifelessActivities22:33
lifelessSubmit the branch to create an MP (our toolchains can look at this and it provides a location for a post landing review if the branch has that done to it). Self review with review type 'unreviewed'. Land via the normal landing process.22:33
lifelesswe may find this needs tuning22:34
lifelesse.g. review type none/code22:34
lifelessjml: have you crashed?22:36
wallyworld_thumper: one of my branches got merged into db-devel rather than devel. i need to do stuff that depends on it. do i have any option other than to work on a db-devel branch?22:41
lifelesswallyworld_: land it again.22:41
lifelesswallyworld_: to the right place.22:41
thumperwallyworld_: did you branch from db-devel?22:41
wallyworld_ok22:41
lifelesswallyworld_: just be sure that you don't suck all of db-devel in.22:41
wallyworld_no i branched originally from devel22:42
thumperwallyworld_: or more importantly, did you branch or merge from db-devel since the last rollout?22:42
thumperwallyworld_: in which case, land it on devel as well22:42
lifelessthumper: 'release' - not rollout ;)22:42
thumperwallyworld_: bzr's merge detection will handle it22:42
wallyworld_with that branch i have only merged into it from devel22:42
wallyworld_not sure why it landed in db-devel22:43
lifelessthe default is bong.22:43
lifelessmailing22:43
thumperlifeless: the default is due to branch stacking22:43
thumperlifeless: and that is all22:43
wallyworld_thumper: i think we are still in textfix mode so i'll have to wait to land it again22:44
lifelessthumper: its the series, we can change it right ?22:44
thumperlifeless: if we had an easy way to say push and stack on db-devel when needed (that is tested and simple to use) then we could switch relatively easily22:45
lifelessthumper: I realise we need to do a repo fetch22:45
lifelessthumper: we can switch trivially22:45
thumperlifeless: devel isn't stacked22:45
lifelessthumper: right, but branches stacked on db-devel may need db-devel only revisions.22:45
thumperlifeless: right22:45
lifelessso its devel.fetch(db_devel), then change the default series.22:45
lifeless4 lines of python + a losa22:46
thumperwe don't need any code changes22:46
lifelessright22:46
lifelessjust the fetch22:46
thumperno22:49
thumpernot even the fetch22:49
thumperbranches are stacked on aboslute paths22:49
thumpernot shortcuts22:49
thumperit is more the overhead of new db-devel branches22:49
thumperbut if we don't care...22:49
* thumper shrugs22:49
lifelessthumper: devel branches are the common case22:49
lifelessthumper: they pay massively today ;)22:50
lifelessthumper: please do it22:50
lifelessthumper: and let launchpad-dev know22:50
* thumper nods22:50
lifeless(unless you think its wrong to change it)22:50
wgrantNothing depends on the lp:launchpad alias?22:53
lifelessflacoste: ppr needs 'max', I think :)23:11
lifelesslosa ping23:18
spmyo23:18
lifelesshttps://pastebin.canonical.com/39248/23:18
lifelesscan you add that into23:18
lifelesshttps://edge.launchpad.net/+feature-rules23:18
spmsure. what does it all do? timeouts against specific pages?23:20
lifelessyes23:20
lifelessas per https://dev.launchpad.net/LEP/FeatureFlags23:20
spmjust never seen that style before23:20
lifelessif you look at the lpnet timeout report23:20
lifelessthese are the current hard timeouts23:20
wgrantlifeless: Exempting all the main timeouts and slashing the limit again?23:20
lifelessI looked at the PPR to get a feel for how things are going23:21
lifelesswgrant: raising slightly for stuff hurt by 8.423:21
spmhaha23:21
wgrantBah.23:21
spmlifeless: OOPS-1766ED229523:21
lifelesswgrant: will lower the timeout again once we've caught up with that23:21
lifelessspm: rotlf23:21
lifelessspm: that was near-instant, right ?23:22
spmyah, purty much23:22
lifelessspm: what was the traceback ?23:22
spmRetry: duplicate key value violates unique constraint "feature_flag_unique_priority_per_flag"<br /> <br />23:22
lifelessah23:22
lifelessI had 4 duplicated23:22
spmyes. given the timeouts...23:22
lifelesschange the second ' 4 ' to ' 5 ' and so own down the page.23:22
lifelesss/own/on/23:22
spmhrm/ I think that's applied. it doesnt' exactly let you know that it has done anything.23:24
lifelessplease do file bugs on this interface23:24
spmyeah, has.23:24
lifelessits your primary knob for controlling LP in real time.23:24
wgrantHm,.23:24
wgrantDo we really need such a massive freeze now that most stuff should be QA'd immediately?23:24
lifelessits bare bones now, but it doesn't have to stay this way.23:24
wgrantA week to QA db-devel changes seems excessive.23:24
lifelesswgrant: the thing I proposed has a much smaller critical section23:24
lifelesswgrant: we're still a week behind in QA right now.23:25
lifelesswgrant: until we've caught up, we haven't caught up.23:25
wgrantTrue.23:25
lifelessspm: thanks, in 60 minutes we should see a drop to near-zero hard timeouts on lpstats23:26
wgrantlifeless: Don't worry, Soyuz will fix that :)23:27
wgrantsinzui: Why does DistributionMirror live in Registry?23:34
lifelessaren't you meant to be stufying ?23:37
wgrantSilence.23:37

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