[00:06] <BjornT_> mwhudson, al-maisan: sourcecode/lazr-js has been removed from devel. rocketfuel-get in devel will remove the lazr-js from sourcecode/, so if you use the to link the db-devel lazr-js branch to the devel one, it won't work.
[00:07] <thumper> BjornT_: shouldn't you be sleeping?
[00:09] <BjornT_> thumper: probably. but doing sports late in the evening usually causes me to fall asleep late.
[00:09] <thumper> BjornT_: what sport?
[00:09] <wgrant> BjornT_: You have an exaggerated definition of 'late'.
[00:10] <BjornT_> thumper: floorball
[00:10] <thumper> BjornT_: what is floorball?
[00:10] <thumper> BjornT_: indoor football?
[00:10] <BjornT_> thumper: more like indor http://en.wikipedia.org/wiki/Floorball
[00:11] <BjornT_> more like indoor hockey, but without ice and skates
[00:11] <thumper> BjornT_: so roller hockey for those that can't skate?
[00:12] <BjornT_> thumper: yeah, something like that :)
[00:15] <thumper> spm: ping
[00:16] <al-maisan> BjornT_: thanks for the explanation re. sourcecode/lazr-js removal.
[04:07] <thumper> mwhudson: do you know where the fix needs to go for https://bugs.edge.launchpad.net/launchpad-code/+bug/372506  ?
[04:07] <mup> Bug #372506: Don't log OOPSes when people try to delete a directory <codehosting-ssh> <oops> <Launchpad Bazaar Integration:Triaged> <https://launchpad.net/bugs/372506>
[04:08] <mwhudson> thumper: lp.codehosting.sftp probably
[04:08] <thumper> mwhudson: want to take this one?
[04:08] <mwhudson> catch the error in remove or unlink or whatever the method is and error nicely
[04:08] <mwhudson> thumper: yeah, sure
[04:08] <thumper> mwhudson: ta
[04:20]  * mwhudson_ has probably stopped rebooting his modem for now
[04:34]  * thumper whacks up a branch for review
[04:34] <thumper> mwhudson: I don't expect you to do it at 5:30pm
[04:34] <thumper> mwhudson: even though it is smallish
[04:36] <mwhudson_> thumper: i don't see it
[04:36] <thumper> mwhudson: I'm still typing the cover letter :)
[04:37] <mwhudson> ah ok
[04:38] <thumper> email sent, should flow through the system in 2 minutes
[04:41] <thumper> mwhudson: it is there now
[04:41]  * thumper EODs
[04:55] <mwhudson> thumper: eurgh
[05:41] <stub> has anyone elses gpg-agent crapped out today or yesterday?
[05:41] <lifeless> nope
[05:44]  * stub wonders how gpg-agent got uninstalled
[07:07] <henninge> Hi everybody!
[07:08] <henninge> Is there a way to see a log of sql queries issued by a request on launchpad.dev?
[07:09] <henninge> Like the statement log I see in an OOPS report?
[07:13] <henninge> stub: ^ would you know the answer?
[07:13] <henninge> wgrant: ^ ?
[07:14] <henninge> BjornT_: ^ ?
[07:18] <spm_> henninge: there's a switch in postgres somewhere....
[07:18] <henninge> Hi spm_, I was about to ping you next ... ;)
[07:18] <lifeless> henninge: easiest way is to turn on verbose logging
[07:18] <lifeless> in pg
[07:18] <henninge> lifeless: how?
[07:18] <spm_> henninge: http://archives.free.net.ph/message/20030820.033440.c8363e0d.en.html ?? maybe.
[07:18] <wgrant> henninge: It's probably turned on for you.
[07:19] <wgrant> henninge: It was only turned off in lp-database-setup last week.
[07:19] <wgrant> henninge: It's not in the postgres log?
[07:20] <wgrant> (log_statement='all' is the relevant postgres switch)
[07:21] <lifeless> mm
[07:21] <lifeless> https://code.edge.launchpad.net/~lifeless/drizzle/build
[07:21] <lifeless> was pushed to 5 weeks back
[07:21] <lifeless> mtaylor pulled from it
[07:21] <lifeless> where did my content go??!
[07:22] <wgrant> He pulled it alllll awaaaay.
[07:22] <BjornT_> henninge: see lib/canonical/launchpad/pagetests/basics/user-requested-oops.txt
[07:24] <henninge> BjornT_, lifeless, spm_: Thanks. Yes, it is logging for me. I just didn't think about the fact that postgres is not in the LP tree ...
[07:24] <henninge> wgrant: ^
[07:24] <lifeless> spm_: mwhudson: any ideas?
[07:24] <spm_> henninge: LIES!!! EVERYTHING!!! is in the LP tree.
[07:25] <lifeless> mtaylor: I see a repo there...
[07:25] <wgrant> Win. First 3.0 source package in LP.
[07:25] <spm_> lifeless: "This branch has not been pushed to yet." ?
[07:25] <lifeless> spm_: mtaylor branched from it 5 weeks back
[07:25] <lifeless> spm_: and had a copy on his disk, so it most definitely *was* pushed to - and mirrored
[07:26] <spm_> impressive. that sounds like something's zotted it.
[07:26] <lifeless> thats why I'm concerned
[07:26]  * spm_ has a look at crowberry's disk.
[07:26] <lifeless> mtaylor: I'm going to push up to build2, so I don't trash the evidence
[07:26] <mtaylor> lifeless: k. works for me
[07:29] <lifeless> mtaylor: a thought, perhaps I sent you a bundle instead?
[07:29] <lifeless> drizzle is pushing up to build2 now, but will be a while - it can't stack.
[07:30] <mtaylor> lifeless: hrm. perhaps that's what you did
[07:30] <spm_> spm@crowberry:~$ bzr log --short -r -10..-1 /srv/bazaar.launchpad.net/push-branches/00/03/be/17
[07:30] <spm_> bzr: ERROR: Not a branch: "/home/robertc/source/drizzle/autorundeps/".
[07:30] <spm_> lifeless: ^^
[07:31] <lifeless> spm_: ><
[07:31] <lifeless> spm_: bzr info on that?
[07:31] <mtaylor> lifeless: http://paste.ubuntu.com/279365/plain/
[07:31] <lifeless> actually
[07:31] <mtaylor> lifeless: that's what you gave me
[07:31] <lifeless> ls .bzr
[07:31] <lifeless> mtaylor: yeah I gave you a merge directive.
[07:31] <mtaylor> yup
[07:31] <lifeless> so false alarm methinks on the lp losing data
[07:32] <mtaylor> k. well, I've got it now then
[07:32] <spm_> lifeless: http://pastebin.ubuntu.com/303414/
[07:32]  * mtaylor wags finger at lifeless for pastebinning a merge directive instead of pushing a branch ... but gets over it just as quickly
[07:32] <lifeless> mtaylor: hey, drizzle isn't tiny ya know
[07:32] <lifeless> spm_: cat .bzr/branch/format
[07:33] <mtaylor> lifeless: wouldn't have been a problem if we had gotten around to --2a yet.
[07:33] <mtaylor> sigh
[07:33] <lifeless> mtaylor: hint. hint.
[07:33] <lifeless> mtaylor: J F D I
[07:33] <spm_> lifeless: Bazaar-NG Branch Reference Format 1
[07:33] <mtaylor> lifeless: I sent the initial "o hai all ya'll upgrade" email
[07:33] <lifeless> spm_: I pushed a checkout. \o/
[07:33] <lifeless> mtaylor: you're mising the word 'now' ;)
[07:34] <spm_> lifeless: I'm gathering that's officially a "Bad Thing(™)"
[07:34] <lifeless> spm_: its a bug that I keep meaning to track down
[07:34] <lifeless> its a bad thing, yes.
[07:34] <lifeless> its like a symlik
[07:34] <spm_> heh. sounds like you have extra incentive now then huh? :-P
[07:35] <lifeless> spm_: nope, not really - 5 weeks ago occurnce != fresh
[07:35]  * spm_ waves away unimportant minor facts
[08:23] <adeuring> good morning
[08:52] <stub> henninge: There is an environment variable you can set I think
[08:52] <henninge> stub: thanks, I found the log.
[08:53] <stub> henninge: LP_DEBUG_SQL and LP_DEBUG_SQL_EXTRA
[08:53] <henninge> stub: that outputs to stdout?
[08:53] <stub> henninge: Or stderr - its in lib/canonical/launchpad/webapp/adapter.py
[08:54] <henninge> stderr is fine
[09:12] <mrevell> Guten morgen!
[09:19] <henninge> Guten Morgen mrevell!
[09:20] <mrevell> hallo henninge :)
[09:22] <jml> mrevell, henninge: Good morning!
[09:22] <mrevell> yo jml
[09:23] <henninge> Hallo jml!
[09:23] <henninge> ;)
[09:23] <maxb> I was trying to run tests on the python-migration branch last night, and they seemed broken :-/
[09:23] <henninge> and a smile for Mr Re-VELL :)
[09:23] <maxb> Does anyone know about the huge amounts of "warning: Not importing ... missing __init__" ?
[09:23] <jml> maxb, I'm ensaddened.
[09:24] <jml> maxb, yes, I do a little.
[09:24] <jml> maxb, Python generates this error internally. It's a false warning.
[09:24] <jml> maxb, it's caused by lazr.foo being a bit weird.
[09:25] <maxb> Hmm. I thought I saw it trigger at least one "unclean stderr" -> fail
[09:25] <jml> maxb, yes, that's a real problem.
[09:27] <maxb> Life would be so much simpler if zope/lazr hadn't gone for namespace packages, when Python doesn't really support them :-/
[09:34] <henninge> jtv!
[09:34] <jtv> hi henninge!
[09:34] <henninge> Hallo jtv! ;)
[09:35] <jtv> henninge: working on message-sharing memory footprint... forgot I'm also supposed to OCR.
[09:36] <henninge> jtv: working on the +templates page time out - almost done.
[09:36] <jtv> henninge: cool... what are you doing to fix it?
[09:37] <henninge> jtv: getting SourcePacakgeName in the same query as POTemplate for POTemplateSubSet
[09:37] <jml> maxb, yes :)
[09:37] <jtv> henninge: so a prejoin?
[09:37] <henninge> jtv: actually, I have always meant to ask what exactly that is.
[09:37] <henninge> a prejoin
[09:38] <BjornT_> mthaddon: could you please disable edge updates? it seems like current trunk is a bit broken, so i want to make sure edge doesn't get updated until it's fixed.
[09:38] <henninge> jtv: It's a join in "using", is that it?
[09:38] <mthaddon> BjornT_: already disabled
[09:38] <jtv> henninge: you add a join to your query, even though you're not actually using it right there.  But you're pre-seeding the cache with those objects so you don't end up fetching them one by one later.
[09:38] <henninge> jtv: exactly
[09:39] <jtv> henninge: in sqlobject we had a parameter with that exact name, though in Storm what you usually do is fetch tuples of objects.
[09:39] <henninge> jtv: what made is difficult was that the POTemplateSubSet code did not distinguish bewtween the productseries and the distroseries case.
[09:40] <maxb> I'm not in a position to make a branch and MP for it now, but it appears that the build is currently broken (for me, anyway), because lazr-js hasn't been removed from sourcecode/Makefile
[09:40] <henninge> and this only applies to the latter, of course.
[09:40] <jtv> henninge: that's no problem for a left join though!
[09:40] <henninge> jtv: seriously?
[09:41] <henninge> jtv: I could just left join SourcePackageName all the time and it doesn't cost me anything?
[09:41] <jtv> henninge: seriously... if you add an outer join on POTemplate.sourcepackagename, then you'll just get a bunch of nulls instead of SourcePackageName rows in the case where it's null.
[09:41] <henninge> jtv: no performance penalty?
[09:42] <henninge> after all, I am trying to solve a performance problem
[09:42] <jtv> henninge: I can't guarantee that it'll never cost you anything, but I'd say that it's _probably_ not an issue.
[09:42] <BjornT_> mthaddon: could you add a comment saying that it shouldn't be enabled before bug 462502 is fixed?
[09:42] <mup> Bug #462502: Javascript broken when devmode is turned off <Launchpad Foundations:In Progress by bjornt> <https://launchpad.net/bugs/462502>
[09:42] <mthaddon> sure
[09:42] <henninge> jtv: hm, that could make the code a lot simpler, of course.
[09:42] <jtv> henninge: 'xacly
[09:43] <jtv> henninge: I'd suggest that you ask a losa to run some explains to check this.
[09:44] <henninge> jtv: woa, I remember stuart showing that at the epic. But that was about a year ago ...
[09:44] <jtv> henninge: get a sample query that times out; do 2 "explain analyze" runs on it and record the second time; then compare to another "explain analyze" run that has the left join added.
[09:44] <henninge> quite exactly, actually.
[09:44] <henninge> jtv: ah, that is not the problem case
[09:45] <jtv> what isn't?
[09:45] <henninge> jtv: I'd be concerned about productseries +template pages
[09:45] <henninge> but they are not timing out
[09:45] <henninge> concerned, because they don't need the sourcepackagename
[09:46] <henninge> jtv: but maybe that's the answer
[09:46] <jtv> the sourcepackagename lookup is going to be quite fast in cases where the fk is always null, I suspect :)
[09:46] <henninge> jtv: since product series pages don't get that long, there is no performance issue anyway.
[09:53] <jtv> henninge: stub just pointed out the "precache" wrapper for result sets.  Lets you do something close to the classic prejoin, but without having to retrieve tuples that you only need one object out of.
[09:54] <henninge> jtv: sounds interesting, I have some code going into exactly that.
[09:54] <henninge> jtv: do you have an example?
[09:54] <jtv> henninge: looking it up now...
[09:55] <jtv> henninge: lib/lp/services/database/precache.py
[09:55] <jtv> example is in the docstring.
[09:56] <henninge> jtv: cheers
[09:56] <jtv> np
[09:59] <thumper> mwhudson: night all
[10:02] <deryck> Morning, all.
[10:02] <jtv> hi deryck
[10:22] <henninge> stub, jtv: Are you sure that precache works as described in the doc string???
[10:22] <jtv> henninge: see the XXX at the top of the file :)
[10:43] <henninge> jtv: yup, timeout gone. https://translations.staging.launchpad.net/ubuntu/karmic/+templates
[10:43] <jtv> henninge: \o/
[10:43] <jtv> did you get the precache working for this?
[10:44] <henninge> jtv: but that is not using precache yet. Actually, it is failing.
[10:44] <jtv> :-(
[10:44] <henninge> failing tests
[10:44] <henninge> jtv: I got it to work for this page but I am currently running lp.translations tests on it and I already saw a failure.
[10:45] <jtv> :/
[10:45] <henninge> jtv: btw, ordering of the +templates page is next because when it gets that long, it is really useless the way it is now.
[10:45] <henninge> jtv: POTemplateSubSet had an ordering (id,path) ...
[10:46] <jtv> oh, how useful (!)
[10:46] <henninge> jtv: I already got rid of the path part, os now it's only id.
[10:47] <henninge> jtv: I think sourcepackagename.name, potemplate.name sounds sensible
[10:47] <jtv> henninge: quite
[10:47] <jtv> hmm... still seeing timeout on that staging page. :(
[10:49] <henninge> jtv: I don't
[11:04] <jtv> worked for me this time too
[11:22] <jml> allenap, hi
[11:23] <allenap> jml: Hi.
[11:24] <jml> allenap, how's it going with the build eng stuff? love, joy and peace or fire, plague and the sword?
[11:24] <allenap> jml: It's okay, but it's mostly task switching and wtfing :)
[11:25] <jml> allenap, task switching :\
[11:27] <henninge> jtv: I fixed precache, btw.
[11:27] <jml> allenap, nothing would thrill me more than to have a chance to shoot the breeze about this stuff, so please feel free to call me if you want a teddy bear / rubber duck / cardboard cutout...
[11:27] <henninge> jtv: it's in this patch: http://paste.ubuntu.com/303512/
[11:29] <allenap> jml: That would be cool. I'm a bit afk and back today because t'other half is ill and there's no one else to look after the kids, but I should have time and my thoughts in order this afternoon.
[11:30] <jtv> henninge: stub just asked if you were going to add a test too, and you did.  Good show!  :)
[11:30] <jml> allenap, cool. that'd be great.
[11:30] <henninge> jtv: no, the test was there already. I just extended it for the case that came up in our test suite.
[11:32] <jtv> henninge: yes, just heard the same here.  :-)
[11:32] <henninge> stub, jtv: I think I can remove that part of the XXX, then.
[11:32] <henninge> jtv: btw, are you with stub?
[11:32] <stub> +1
[11:33] <jtv> henninge: no, we're just friends
[11:34] <jtv> henninge: yes, we're working on the message-sharing migration footprint issue.
[11:34] <jtv> stub: how's that doing btw?  :-)
[11:36] <henninge> stub: Is the renaming of precache still an issue? That is the only thing left from the XXX (and the bug) that would need doing. If we won't rename it, I'll remove the XXX completely.
[11:36] <jtv> I vote keep the name-changing part.  :-)
[11:36] <stub> I expect we need a better name since I couldn't remember what it was called earlier. I can't think of a better name though.
[11:36] <stub> Feel free to rename it if you want - there are only two or three callsites at the moment.
[11:36] <henninge> so "prejoin" is not good?
[11:37] <stub> I suspect that was what I intended to call it, but then a brain fade made me type 'precache'
[11:39] <henninge> stub: I hope you are feeling better now ... ;)
[11:43] <stub> $ make harness
[11:43] <stub> >>> import gc
[11:43] <stub> >>> len(gc.get_objects())
[11:43] <stub> 461210
[11:43] <stub> And a memory footprint of 480MB
[11:44] <stub> Just for loading all our libraries and processing that zcml. Sheesh.
[11:49] <jml> :(
[11:50] <jml> stub, to quote _Anchorman_, that's not a small number.
[12:03] <jml> oh wow
[12:03] <jml> python 2.5 has datetime.strptime
[12:04] <jml> O Happy Day!
[12:04] <jml> \
[12:15] <jtv> henninge: that's for cases such as KDE where translations get uploaded as part of one package, but need to be imported in another.
[12:16] <jtv> I always forget the details of how it works.
[12:46] <maxb> barry: Do you know what the last known state of the python-migration branch is? Because it doesn't work for me. (1) apport_python_hook fails the clean_modules test, (2) warning, not importing lazr, missing __init__.py breaks things looking for clean stderr
[13:34] <sinzui> EdwinGrubbs: are you ready for the stand-up call?
[13:34] <sinzui> maxb: barry is not online today
[13:34] <EdwinGrubbs> sinzui: yes
[13:34] <maxb> :-/
[13:34] <maxb> ok
[13:34] <maxb> gary_poster:  Do you know what the last known state of the python-migration branch is? Because it doesn't work for me. (1) apport_python_hook fails the clean_modules test, (2) warning, not importing lazr, missing __init__.py breaks things looking for clean stderr
[13:52] <BjornT_> rockstar, abentley: can you confirm bug 462628 and bug 462632? if they are valid, they need to be dealt with now.
[13:52] <mup> Bug #462628: Commenting on merge proposals is broken <Launchpad Bazaar Integration:New> <https://launchpad.net/bugs/462628>
[13:52] <mup> Bug #462632: Edit commit message on a merge proposal is broken <Launchpad Bazaar Integration:New> <https://launchpad.net/bugs/462632>
[13:54] <abentley> BjornT_: chasing
[13:58] <abentley> BjornT_: I can confirm that inline commenting on merge proposals is broken.
[13:58] <sinzui> reviewers meeting is 2 minutes in #launchpad-meeting
[14:00] <deryck> abentley, BjornT_ -- that commenting is likely broken by my recent branch landing.  I need to get a lazr-js branch landed (if I can now) and both MP comments and commits need updating so the xhtml representation is utf-8 encoded.
[14:00] <BjornT_> abentley: ok. i'd suggest that you talk to the losas and make sure edge doesn't get updated until this is resolved (and the fix is in stable)
[14:01] <deryck> I commented on the encoding bit on thumper's MP for commit message branch.
[14:01] <abentley> deryck: That looks like the cause of the commit message issue, but not the comment issue.
[14:02]  * deryck looks closer at bug
[14:03] <deryck> abentley, yeah, I can't tell much from the Windmill traceback, sorry.
[14:03] <BjornT_> deryck: you can land your lazr-js branch. no you can even land the lazr-js branch first (well, in the future), so that lp won't break.
[14:03] <abentley> losa: can we please disable edge updates-- we have a significant regression we don't want to show users.
[14:03] <mthaddon> abentley: it's already been disabled
[14:03] <abentley> mthaddon: Cool.
[14:05] <deryck> BjornT_, right, I should have landed the lazr-js first, but thought with your work and edge updates being disabled I could beat the rush and get it landed.
[14:05] <deryck> BjornT_, didn't think about having to read up on the new way of doing lazr-js branches ;)
[14:41] <gary_poster> maxb: did you see https://dev.launchpad.net/PythonMigrationStatus and Barry's email?  per those documents, https://code.launchpad.net/~launchpad-dev/launchpad/ztk-2.5 is the primary one we care about now.  Moreover, there will be bit rot, because we can't land it until after the upcoming LP release (again, described in Barry's email).  Dealing with the bit rot and landing that will be my job then (i.e., in about 1.5 weeks)
[14:42] <maxb> right, so I should essentially consider python-migration to be obsolete and "spliced out" of the merge flow?
[14:43] <maxb> Unless there's non-zope stuff that's already landed into ztk-2.5 instead of python-migration already, my problems will still apply there
[14:44] <maxb> Oh, so there are.
[14:44] <maxb> Right
[14:44]  * maxb marks python-migration as Abandoned, and updates wiki and retargets MP
[14:45] <gary_poster> maxb: thank you
[14:49] <maxb> Where has the release schedule gone these days?
[14:50] <maxb> Releases/2009Calendar
[15:07] <maxb> abentley: asdf to you too :-p
[15:07] <abentley> Sorry about that.
[15:07] <maxb> :-)
[15:29] <maxb> grr. The Launchpad Moinmoin CSS has broken strikethrough support.
[16:06] <maxb> gary_poster: Do you suppose you could pre-emptively add the ztk-2.5 new things to lp-source-dependencies at some point?
[16:09] <gary_poster> maxb: my default inclination is to keep them out until I prepare the merge.  I'm probably pretty easy to convince otherwise though.  Why do you want to have them added early?  (One example of a winning argument that need no further explanation is "because I'm going to be keeping the bit-rot away and this would be convenient.")
[16:09] <gary_poster> needs
[16:10] <maxb> I figured I'd try my hand at resolving some of the "Bustimications" - that _might_ evolve into keeping pace with devel :-)
[16:12] <gary_poster> ok fair enough.
[16:12] <gary_poster> I'll do that today.
[16:39] <al-maisan> Mould somebody please help me with create_initialized_view() usage? I am adding unit tests (http://pastebin.ubuntu.com/303650/, line 62-63) and the setUp() method fails with: http://pastebin.ubuntu.com/303651/
[16:39] <al-maisan> *Could
[16:39] <al-maisan> *sigh*
[16:41] <gary_poster> maxb: rev 87 of lp-source-dependencies has new ztk deps
[17:22] <allenap> jml: Are you around for a bit? If so, do you fancy having a chat?
[17:23] <jml> allenap, I am. I'm on a call now though. I'll ping you when I'm off.
[17:23] <allenap> jml: Cool.
[17:26] <jml> allenap, ping!
[17:27] <allenap> jml: pong
[17:29] <allenap> jml: Or a mobile, if someone has sold you one yet. I have near infinite free credit for mobile-to-mobile calls.
[18:12] <mrevell> I'm gonna head off before Xchat craps out again. Night all
[18:24] <thumper> deryck: can I have a very quick call with you after the TL call?
[18:25] <deryck> thumper, I have a quick chat with BjornT_ and then we can.
[18:25] <thumper> deryck: or I could talk to you in 1.5h
[18:26] <thumper> as it is breakfast time here :)
[18:27] <deryck> thumper, I have to be EOD in 1.5 hour with family responsibilities today. anytime before 1.5 hour from now.
[18:27] <thumper> :(
[18:27] <thumper> deryck: how long is the BjornT_ likely to take?
[18:28] <deryck> thumper, 5 mins, tops.  He has to go, too.
[18:28] <thumper> ok
[18:28] <thumper> I'll wait
[18:39] <mwhudson> good morning
[18:43] <BjornT_> deryck: so, what's up?
[18:43] <thumper> mwhudson: early start?
[18:43] <deryck> BjornT_, so I submitted a branch to lazr-js trunk via pqm.  Saw it playing there.  Then lunched, then it's finished in pqm....
[18:44] <deryck> BjornT_, but no sign of it in lazr-js trunk, no email saying success or failure.
[18:44] <deryck> BjornT_, where can I look to work out what happened?
[18:44] <mwhudson> thumper: yeah, emma is off to a meeting today, left at 6:30
[18:44] <thumper> :)
[18:45] <thumper> deryck: I'd say chase a losa for the pqm logs
[18:46] <BjornT_> deryck: yeah, what thumper said. you might be able to look at the logs yourself. they might be on devpad
[18:46] <thumper> deryck: wild stab in the dark, but there was speed and connectivity issues at some stage yesterday
[18:46] <thumper> deryck: perhaps the push failed?
[18:46] <BjornT_> yeah, we have seen buildbot fail pulling a few times lately
[18:47] <deryck> thumper, BjornT_, ok, I can try the submit again, and then chase a lose if no luck.  Or look for logs on devpad.  Thanks, guys.
[18:47] <thumper> deryck: I'd look at the logs first
[18:48] <deryck> thumper, ok
[18:48] <deryck> thumper, shall we do a call now?
[18:48] <thumper> deryck: it may be that your branch has landed on the pqm copy already
[18:48] <thumper> deryck: yep
[18:48] <deryck> BjornT_, thanks for the chat time!
[18:49] <deryck> thumper, on skype?
[18:49] <thumper> deryck: I'm trying to get skype to give me sound
[18:49] <BjornT_> deryck: "different rich-root support"
[18:49] <BjornT_> deryck: in patch.1256740771.log
[18:50] <deryck> ah, suck.
[18:50] <deryck> I know what I did.
[18:50] <deryck> got a repo doesn't support 2a format, and couldn't tell, so I went one format back in the help text and forgot to ask :)
[18:50] <thumper> we should upgrade lazr-js to 2a
[18:50] <deryck> indeed we should
[19:38] <mwhudson> i love the way all the launchpad developers answer "What’s more important? Principle or pragmatism?" by bashing the question
[19:40] <intellectronica> yeah, i'm still waiting for someone to just make a simple choice, a one word answer
[19:40] <intellectronica> of course, by not answering the questionaire i gave up the opportunity to do that myself :-/
[19:40] <rockstar> mwhudson, :)
[19:48] <flacoste> intellectronica: yeah, i should have answered 'Five pounds of flax'
[19:49] <intellectronica> :)
[19:50] <flacoste> kfogel: do you have a LWN subscribtion? did you look at http://lwn.net/Articles/359013?
[19:51] <kfogel> flacoste: I don't have a subscription.  let me see
[19:51] <kfogel> flacoste: oh, it's paywalled
[19:51] <mwhudson> flacoste: i pasted a free link into my warthogs mail
[19:52] <mwhudson> kfogel: ^^
[19:52] <kfogel> mwhudson: thx
[19:53] <jml> ISTR I answered "principle"
[19:54] <jml> and then talked and talked and talked
[19:54] <intellectronica> jml: that's encouraging
[19:55] <jml> intellectronica, "Five pounds of flax" would have been a better answer.
[19:55] <intellectronica> anyway, is it not five tons?
[19:56] <mwhudson> jml: hello
[19:56] <mwhudson> jml: i wanted to talk to you about something, i wonder what it was
[19:56] <intellectronica> or are you just being pragmatic?
[19:56] <mwhudson> jml: oh yes, maybe it was https://bugs.edge.launchpad.net/launchpad-code/+bug/450998
[19:56] <mup> Bug #450998: branch puller can try to pull the same branch twice at once <Launchpad Bazaar Integration:Triaged> <https://launchpad.net/bugs/450998>
[19:58] <jml> mwhudson, I'd love to. I'm going to talk to lifeless first though.
[19:58] <mwhudson> jml: ok
[19:59] <mwhudson> it's not really week 2 still is it?
[20:00] <flacoste> intellectronica: : it is 5 tons!
[20:02] <intellectronica> actually, according to google it's 3 pounds in the original zen story, and 5 tons in the principa discordia, so i guess that's a hybrid answer
[20:02] <flacoste> right, i mixed it up :-)
[20:03] <kfogel> mwhudson, flacoste: responded
[20:03] <kfogel> mwhudson, flacoste: I mean "responded on warthogs", not publicly (and see warthogs response for why)
[20:05] <flacoste> kfogel: note that this will not paywalled in one week
[20:05] <kfogel> flacoste: yes, but by then it'll be old news :-)
[20:05] <kfogel> flacoste: not completely joking -- one effect of paywalls is to fracture the audience and reduce feedback effects.
[20:06] <flacoste> i never thought of it that way, but that's an interesting point
[20:07] <intellectronica> are we in testfix mode?
[20:08] <intellectronica> oh yes we are
[20:08] <abentley> intellectronica: Just hit out of space on the builder.
[20:08] <intellectronica> bööö
[20:57] <thumper> rockstar: don't see you on line
[20:57] <thumper> rockstar: on skype that is
[20:58] <rockstar> thumper, I didn't leave it up.  One sec.
[21:00] <rockstar> thumper, you aren't picking up.
[21:08] <rockstar> barry, are we having a meeting this week.
[21:12] <abentley> rockstar: sinzui ran this morning's meeting because barry is unavailable.
[21:12] <sinzui> oh
[21:12] <sinzui> thanks for the reminder
[21:13] <rockstar> sinzui, no problem.
[21:30] <Ursinha> hi gary_poster :) do you have a moment to take a look in an oops, please?
[21:30] <gary_poster> sure, Ursinha
[21:30] <Ursinha> gary_poster, I have an oops, https://lp-oops.canonical.com/oops.py/?oopsid=OOPS-1389H3247
[21:31] <Ursinha> this one happened yesterday
[21:31] <Ursinha> about 40 occurrences
[21:31] <gary_poster> Ursinha: and is new?
[21:32] <gary_poster> (there's some recent work on that page, possibly CP)
[21:33] <Ursinha> looking for a bug, I found another bug that references https://lp-oops.canonical.com/oops.py/?oopsid=OOPS-1396L2872
[21:35] <Ursinha> gary_poster, I just can't find the bug # again :/ still looking
[21:35] <sinzui> Working in blueprints is a deep well of sadness.
[21:35] <Ursinha> gary_poster, well, my question is if these two are related
[21:35] <sinzui> I am writing so many basic tests for existing features so that I can 1 attribute
[21:37] <gary_poster> Ursinha, I believe they are unrelated.  Let me look at one more thing.
[21:37] <Ursinha> sure gary_poster
[21:37] <Ursinha> thanks
[21:42] <gary_poster> Ursinha: no they are unrelated except in the basic mechanism of the problem.  For the ResetPasswordView, next_url is an @property that does not have a setter defined, and yet _cancel in the superclass BaseTokenView tries to set next_url.  Boom.  That's a Foundations bug.  If you could create the bug, I'll assign it to myself and fix it tomorrow.  Now I'm looking at the other one.
[21:43] <gary_poster> Ursinha: eh, without looking at the code, that's a bugs issue
[21:43] <gary_poster> They are trying to do the same thing, but the only similarity is the kind of error in our code.
[21:43] <Ursinha> gary_poster, I see.
[21:44] <Ursinha> gary_poster, I found bug 54387, is it related to the resetpassword one?
[21:44] <mup> Bug #54387: Race condition in reset password workflow leads to an oops. <oops> <Launchpad Foundations:Triaged> <https://launchpad.net/bugs/54387>
[21:45] <gary_poster> Ursinha, no, the new OOPS is simpler and unrelated other than being in the same vague area of code
[21:46] <Ursinha> gary_poster, all right, filing a bug..
[21:46] <gary_poster> thank you
[21:48] <jml> mwhudson, hi
[21:53] <mwhudson> jml: hi
[21:53] <jml> mwhudson, you wanted to talk about a bug?
[21:53] <mwhudson> jml: only if it's not too late for you
[21:54] <mwhudson> jml: it's the "pull a branch twice at once" bug
[21:54] <jml> mwhudson, sure.
[21:56] <Ursinha> gary_poster, bug 462923
[21:56] <mup> Bug #462923: AttributeError in +resetpassword <oops> <Launchpad Foundations:New> <https://launchpad.net/bugs/462923>
[21:57] <gary_poster> Ursinha: thank you!  triaged and assigned.
[21:57] <Ursinha> gary_poster, thank you very very much
[21:57] <gary_poster> :-)
[22:08] <Ursinha> gary_poster, considering you'll fix soon, can I target that bug to the 3.1.10 milestone?
[22:08] <gary_poster> :-) sure
[22:08] <Ursinha> gary_poster, thank you :)
[22:09] <gary_poster> Ursinha: I can only improve a few aspects of my managing habits at a time--milestones will have to come in a later revision of my software. ;-)
[22:09] <Ursinha> gary_poster, hahahaha :)
[22:16] <gary_poster> night all
[22:21] <thumper> mwhudson: about that vocabulary branch...
[22:22] <mwhudson> thumper: ah yes
[22:25] <mwhudson> thumper: i didn't get a mail about your comment :/
[22:26] <thumper> mwhudson: I bet it got filtered
[22:27] <thumper> abentley: bug 462628, is this the one you have just fixed?
[22:27] <mup> Bug #462628: Commenting on merge proposals is broken <Launchpad Bazaar Integration:New> <https://launchpad.net/bugs/462628>
[22:29] <mwhudson> thumper: i guess i'll just settle for making you put in a new bug report and an XXX
[22:30] <thumper> mwhudson: what for?
[22:30] <thumper> mwhudson: why is it an issue there
[22:30] <thumper> ?
[22:30] <mwhudson> thumper: the problem is your fix is both too narrow and too broad
[22:30] <thumper> mwhudson: so just right then
[22:30] <mwhudson> too narrow: it won't help other callers of getByUrl, for example the merge proposal handling code
[22:30] <mwhudson> too broad: it will mask bugs in getByUrl
[22:31] <mwhudson> thumper: sadly, more than one axis
[22:31] <thumper> mwhudson: I don't get the to narrow bit
[22:31] <thumper> s/to/too
[22:31] <mwhudson> thumper: it would be far better if getByUrl raised a predictable set of exceptions
[22:31] <thumper> mwhudson: it is a predictable set, just about 10 different ones
[22:31] <thumper> mwhudson: I could name them all
[22:32] <thumper> mwhudson: I just didn't think it was worth it
[22:32] <mwhudson> thumper: if you send in a merge proposal somehow that says "target_branch = lp:", it looks very much like the mail handler will oops
[22:32] <thumper> mwhudson: and they don't have a common base class
[22:32] <thumper> mwhudson: most likely
[22:33] <mwhudson> thumper: maybe getByUrl itself should catch them and re raise?  there are basically three answers: 1) a branch 2) not found 3) wtf are you on about
[22:33] <mwhudson> 200 OK, 404 Not Found, 400 Bad Request in http terms
[22:34] <thumper> mwhudson: you are suggesting having getByUrl only raise one type of Bad Request error?
[22:34] <mwhudson> thumper: yeah, i think so
[22:34] <mwhudson> (it returns None rather than raising NotFoundError, I take it?)
[22:34] <thumper> hmm...
[22:35] <thumper> yeah
[22:35] <thumper> so
[22:35] <thumper> Branch if found, none if valid name but not found, and WTF if bad input
[22:35] <mwhudson> i see it returns None when passed an invalid URI :/
[22:36] <thumper> mwhudson: we can change that
[22:36] <mwhudson> thumper: so perhaps to be consistent, getByLPPath errors should be masked too
[22:37] <thumper> yeah...
[22:37] <thumper> so much for a simple branch
[22:37] <mwhudson> sorry :/
[22:37] <mwhudson> thumper: i'd be happier about just sticking the try:except: around getByLPPath
[22:37] <mwhudson> in getByUrl
[22:41] <mwhudson> thumper: replied on the mp
[22:41] <thumper> mwhudson: ok ta
[22:42] <Ursinha> hey sinzui, do you know what's https://lp-oops.canonical.com/oops.py/?oopsid=OOPS-1396EA105 about?
[22:42] <mwhudson> thumper: maybe i didn't get the email before because i hadn't claimed the review
[22:42] <mwhudson> (i have now)
[22:42] <thumper> mwhudson: yes, but the email would have been sent to the list
[22:42] <mwhudson> thumper: i'm not subscribed to the list any more
[22:42] <sinzui> Ursinha: intellectronica has a fix for that already approved to land
[22:42] <thumper> mwhudson: ah ha, that'll be why
[22:43] <Ursinha> sinzui, great, is there a bug for that?
[22:43] <sinzui> Ursinha: the subscribe/unsubscribe link is fails for anonymous users
[22:44] <sinzui> Ursinha:  bug #462742
[22:44] <mup> Bug #462742: OOPS getting a project index page as an anonymous user <oops> <Launchpad Registry:In Progress by intellectronica> <https://launchpad.net/bugs/462742>
[22:44]  * Ursinha stabs launchpad search
[22:45] <thumper> mwhudson: so are you suggesting changing the except clause in branchlookup.py:231 to be Exception? m
[22:45] <thumper> mwhudson: or a list of exceptions raisable ?
[22:45] <mwhudson> thumper: list of exceptions raisable
[22:45] <mwhudson> thumper: consider that part optional though
[22:45] <thumper> I think a list is fine
[22:45] <mwhudson> cool
[22:47] <thumper> mwhudson: it is except (A, B, C): isn't it?
[22:48] <mwhudson> thumper: yes
[22:51] <Ursinha> thanks sinzui :)
[23:18]  * thumper afk for lunch and class
[23:29] <intellectronica> how come we're still in testfix mode? waterfall looks green
[23:29] <elmo> wgrant: around?
[23:30] <wgrant> elmo: I am.
[23:30] <elmo> wgrant: how familiar are you with buildd-manager?
[23:30] <wgrant> elmo: Far too...
[23:30] <elmo> awesome
[23:30] <elmo> it's SNAFU on production
[23:30] <elmo> and I'm epically failing to debug it
[23:30] <wgrant> Ew. What's it saying?
[23:31] <wgrant> It can be really really really obscure when it fails.
[23:31] <elmo> wgrant: it's not saying anything
[23:31] <elmo> it's only schedulding builds to gold
[23:31] <elmo> it's ignoring every other buildd
[23:31] <wgrant> Huh.
[23:31] <elmo> even though they're marked as active and ok
[23:31] <wgrant> A month ago it was scheduling builds to everywhere *except* gold.
[23:32] <wgrant> Let's see...
[23:32] <elmo> http://paste.ubuntu.com/303891/ <-- log is just that
[23:32] <wgrant> elmo: A restart doesn't help, I suppose?
[23:32] <elmo> nope
[23:32] <wgrant> Yup, I love that verbose logging.
[23:32] <Ursinha> kfogel, hi :)
[23:32] <wgrant> OK.
[23:33] <wgrant> Might be best to flip the logging up higher.
[23:33]  * wgrant tries to remember how.
[23:33] <wgrant> 'cause all the buildds look fine...
[23:33] <mwhudson> probably pass -v when you run it
[23:33] <mwhudson> hm, maybe not
[23:33] <wgrant> No.
[23:33] <wgrant> It's hardcoded.
[23:34] <wgrant> elmo: You'll have to hack lib/lp/builddmaster/manager.py, in _setupLogger. s/INFO/DEBUG/
[23:34] <elmo> \o/
[23:34] <wgrant> Something like that.
[23:37] <wgrant> So, that should convince it to actually tell you what it's doing each cycle.
[23:39] <wgrant> Hm.
[23:40] <wgrant> elmo: Giving you anything useful?
[23:40] <elmo> sorry, got distracted, trying now
[23:41] <elmo> also, p.s. die twisted logging scheme, DIE DIE DIE DIE
[23:41] <wgrant> Heh.
[23:41] <mwhudson> yes
[23:41] <mwhudson> all our (code's) twisted stuff logs via stdlib logging
[23:41] <wgrant> So does buildd-manager.
[23:42] <elmo> 2009-10-28 23:42:31+0000 [-] Considering rothera
[23:42] <elmo> 2009-10-28 23:42:31+0000 [-] Builder is not available, ignored.
[23:42] <elmo> OH REALLY
[23:42] <elmo> how about, not
[23:42] <wgrant> rothera unavailable. That's nice.
[23:43]  * wgrant checks the code.
[23:44] <wgrant> Ah.
[23:44] <wgrant> That's fine.
[23:44] <wgrant> A builder must be IDLE to be available.
[23:45] <elmo> oh, bad example
[23:45] <wgrant> Rather.
[23:45] <elmo> it says it about all of them
[23:45] <elmo> e.g.
[23:45] <wgrant> How about you pastebin an entire cycle, if there're no private builds around.
[23:45] <elmo> unfortunately there are
[23:46] <elmo> 2009-10-28 23:45:07+0000 [-] Considering mercury
[23:46] <elmo> 2009-10-28 23:45:07+0000 [-] Builder is not available, ignored.
[23:46] <elmo> better example
[23:46] <wgrant> Let's see...
[23:46] <elmo> 2009-10-28 23:46:07+0000 [-] Checking zirconium
[23:46] <elmo> 2009-10-28 23:46:07+0000 [-] Checking mercury
[23:46] <elmo> 2009-10-28 23:46:07+0000 [-] Checking rhenium
[23:46] <elmo> sorry, I should pastebin more
[23:46] <elmo> anyway, that last part is interesting
[23:46] <elmo> because it doesn't do:
[23:46] <elmo> 2009-10-28 23:46:07+0000 [-] Adding DistroArchSeries ubuntu/hardy/amd64
[23:46] <elmo> that
[23:46] <elmo> which it does fgor gold
[23:47] <wgrant> Argh, buildergroup. die die die
[23:47] <wgrant> Those three 'Checking ' lines are contiguous?
[23:48] <elmo> yeah
[23:48] <elmo> AHA
[23:48] <elmo> I bet the freaking slave died
[23:48] <elmo> hmm
[23:48] <wgrant> On all of them?
[23:48] <wgrant> That seems unlikely.
[23:49] <elmo> maybe not
[23:49] <elmo> wgrant: oh, sorry, context
[23:49] <elmo> wgrant: is that we fucked up the networking earlier and dropped the majority of the buildds off net for ~5 minutes
[23:49] <elmo> during which time cesium wouldn't have been able to contact them
[23:49] <wgrant> Oh.
[23:49] <wgrant> But that should have just marked them un-OK.
[23:49] <elmo> right, which we undid
[23:49] <elmo> but it's certainly related
[23:49] <wgrant> Yeah.
[23:50] <elmo> as f.e. gold is in the DC that wasn't dropped off net
[23:50] <wgrant> What if you bounce the lp-buildd on one of them?
[23:50] <elmo> sure, trying
[23:50] <wgrant> There's not too much state on the master side.
[23:51] <elmo> ok, that didn't help
[23:51] <wgrant> I can't think why slaves would have died from that, though...
[23:51] <elmo> I bounced palmer, no luck
[23:51] <wgrant> I was hoping only the virt builders would be dead...
[23:53] <wgrant> Hmm hmm hmmmmmmm.
[23:53] <wgrant> It looks like it is still actually scanning properly?
[23:53] <wgrant> eg. rothera's build log is updating a bit.
[23:54] <wgrant> Yeah, so it's just the dispatching...
[23:54] <elmo> right
[23:54] <elmo> the buildds that didn't drop off net are fine and building and uploading
[23:55] <elmo> god I really badly needs a TAGS-ified LP tree
[23:55] <wgrant> Ah, didn't know the uploading bit worked as well. So it's not really really sick.
[23:55] <elmo> grep is sooooooooooooooo slow
[23:55] <wgrant> elmo: This is why you need lots of RAM.
[23:55] <mwhudson> elmo: make TAGS works
[23:55] <mwhudson> i think
[23:55] <wgrant> And to grep through just lib/lp/buildmaster and lib/lp/soyuz.
[23:55] <elmo> mwhudson: yeah, but I'm looking at code on production ;-)
[23:55] <mwhudson> elmo: i see :)
[23:56] <elmo> and besides my team give me grief whenever I install a real editor on the servers
[23:56] <wgrant> elmo: All the lines you pasted here so far look normal.
[23:56] <elmo> wgrant: yeah
[23:56] <wgrant> And the buildds are certainly OK in the DB.
[23:58] <wgrant> elmo: There are no attempts to dispatch?
[23:58] <elmo> wgrant: to the working buildds? yes
[23:58] <wgrant> It will fail really strangely if, for example, the builders can't talk to librarian or ppa.launchpad.net
[23:58] <elmo> see e.g. gold's history