/srv/irclogs.ubuntu.com/2009/12/16/#bzr.txt

pooliejames_w: guess you're not around?00:03
pooliespiv, hi?00:03
dOxxxhey poolie00:03
spivpoolie: hey00:05
spivpoolie: I was thinking about the latest ObjectNotLocked bug (this time in cmd_annotate)00:06
pooliewe should have a chat - now if you like, but i might get some coffee first00:06
spivpoolie: now or post-coffee are both fine00:08
spivI already have one ;)00:09
fullermdI find the phrase "post-coffee" so misleading.  It implies that there's such a thing as "pre-coffee".00:09
fullermdThat's like "what happened before time began?"00:09
spmfullermd: +1111100:15
dOxxxWhat version of pycrypto should we be using for bzr 2.0.3 and 2.1.0b4?00:20
dOxxxOr rather, what should I be packaging in the Mac installers?00:20
pooliek00:20
james_whi poolie00:21
james_wI am now00:21
pooliehi00:21
pooliedo you want to talk now, or some other time?00:21
james_wlet's do it now00:23
james_wsave us missing each other again00:23
pooliek00:27
jelmerpoolie, hi00:32
jelmerpoolie: you just changed the product of bug 496827 back to bzr, was that intentional?00:33
ubottuLaunchpad bug 496827 in bzr-svn ""bzr hooks" throws AttributeError exception on Windows" [Undecided,New] https://launchpad.net/bugs/49682700:33
poolieit was accidental00:33
poolienow it won't let me change it back00:34
jelmerah, probably because you're not in the bug maintainers group for bzr-svn?00:34
jelmerI've changed it back00:34
pooliejames_w: http://bazaar-vcs.org/Roadmap00:57
Jordan_Ubzr branch with bzr+ssh is currently using 270m of ram on the server side, is this normal?01:23
spivFor large enough branches, yes, unfortunately.01:24
spiv2.1.0b4 should be a little bit better than 2.0.x on memory consumption, but we still use more than we'd like.01:24
spivOh, and doing cross-format fetches tends to consume more memory too (as well as being slower), so upgrade everything to 2a if you haven't already ;)01:25
Jordan_UThat is unfortunate, it's already triggered the OOM killer twice and is making my computer ( "server" and desktop are the same machine ) unusable01:25
Jordan_Uirssi is about the only usable application right now :)01:26
spivYou can do "bzr branch -r 1000 bzr+ssh://.... foo; cd foo; bzr pull -r 2000; bzr pull -r 3000; ..." etc if necessary.01:26
spivOr maybe even smaller increments.01:26
Jordan_UI assume that also means that subsiquent pulls should reasonable wrt memory usage?01:27
Jordan_U*should be01:28
spivRight, the memory usage is roughly proportional to how much (revisions, file deltas, etc) is being transferred.01:29
Jordan_Uspiv: Thanks01:29
spivSo incremental pulls of the last few revisions new this week will need much less memory than branching the whole thing.01:29
pooliespiv: finished on phone, writing up a mail01:33
spivpoolie: I'm going to grab lunch now; how about I ping you after?02:32
maxbIs it possible to invoke a command from bzrlib.builtins directly from a python script?02:59
maxb..and still get a progress bar?03:05
pooliemaxb: yes03:12
pooliemaxb: to get a progress bar you just need to set up a uifactory03:12
pooliethere is a function to help you do it03:12
maxbmake_ui_for_terminal ?03:14
poolieright03:15
maxbthanks :-)03:15
pooliespiv, sorry, back now03:15
pooliebzr still seems to use pycurl for me03:32
pooliei thought we were going to change to urllib2?03:32
jmlis there a doc / guide on how to maintain a friendly fork with Bazaar03:40
gutworthis there demand for that?03:42
spivpoolie: ok03:47
spivpoolie: shall I call?03:48
pooliespiv, yes, thanks03:51
mwhudsonjml: gtg again?04:07
jmlmwhudson, maaybe.04:07
jmlgutworth, well, I guess I count as demand.04:07
jmlI'm thinking that looms would be a good fit, actually.04:07
dOxxx1uploading mac installer for bzr 2.1.0b4 :)04:14
* fullermd is so happy to have DWIM revspecs...04:20
jfroy|workjelmer: just hit an assertion in bzr-svn's mergeinfo parser, reported in https://bugs.launchpad.net/bzr-svn/+bug/497280; hope it's not going to be too severe to fix :/04:23
ubottuUbuntu bug 497280 in bzr-svn "Assertion hit when pushing merge commit to remote svn repository in parse_mergeinfo_property" [Undecided,New]04:23
poolieomg04:25
poolieis random sniping really _that_ much more fun than writing code?04:26
poolie...04:26
jfroy|workpoolie: was that directed at me?04:26
poolienot at all04:26
pooliesomeone else (guess)04:27
poolienot in this channel04:27
* poolie muffles himself04:27
jfroy|workI'm confused04:27
jfroy|work:/04:27
lifelesspoolie: I think you're turning japanese?04:27
poolieseriously, no offense intended04:27
pooliejust offtopic venting04:27
jfroy|worknone taken, I would write a patch if I had time!04:27
jfroy|workI se04:28
fullermdIf it's not in this channel, it doesn't really exist, does it?   :p04:28
jfroy|work*see04:28
pooliehello lifeless04:28
jfroy|workfullermd: bzrgotism?04:28
mwhudsonpoolie: for some people, it certainly seems that it is04:29
cellofellowis there an easy way to access a SourceForge CVS project with bzr?04:29
dOxxx1lifeless/poolie: I'm running bzr selftest --subunit from lifeless' subunit branch on Vista at the moment.04:29
=== dOxxx1 is now known as dOxxx
lifelesscellofellow: not really04:30
fullermdjfroy|work: If you can get it declared a religion, it's tax-deductible.04:30
lifelessdOxxx: thank you for that04:30
cellofellowbzr-cvsps-import or bzr-fastimport seem close. I don't have commit access to the repo and it isn't very active.04:31
dOxxxlifeless: you're welcome. seems to be working okay so far. I find subunit output rather hard to read though. I must see if I can get the filters working.04:31
lifelessdOxxx: the patch doesn't require subunit output :) - just run as normal.04:31
mwhudsoncellofellow: if you can download the cvs repo, you can use fastimport04:31
dOxxxlifeless: so I didn't need to run with --subunit?04:31
lifelessdOxxx: if you put the filters on your path (set PATH=%PATH%c:\src\subunit\filters) then you should be able to use them.04:32
mwhudsoncellofellow: otherwise you can request an import on launchpad04:32
lifelessdOxxx: indeed :)04:32
lifelessdOxxx: for most people subunit streams will be an implementation detail they rarely need to look at.04:32
dOxxxlifeless: Okay. So I guess I should try --parallel next.04:32
cellofellowok04:33
lifelessdOxxx: you need --parallel=subprocess on win3204:33
dOxxxyeah04:33
lifelesshi poolie04:36
dOxxxlifeless: with --parallel=subprocess, it looks like I'm getting mixed CRLF and LF in the stdout04:36
lifelessdOxxx: what command line did you use?04:36
dOxxxlifeless: bzr selftest --parallel=subprocess > tests.out 2> tests.err04:37
lifelessI wouldn't redirect it04:37
lifelessas it will be doing the fancy bzr progress bar04:37
dOxxxlifeless: ah yes, you're right04:37
dOxxxlifeless: actually, no... I'm getting a lot of spewage04:38
dOxxxlifeless: looks like multiple threads writing to stdout concurrently04:38
dOxxxlifeless: no progress bar that I can see04:39
dOxxxlifeless: in fact I see output very similar to --subunit04:39
dOxxxlifeless: albeit garbled04:39
lifelessok, thats not the goal :)04:39
lifelesssomeone recently had a patch to make --parallel=subprocess work at all on win32; but it should be merged.04:39
dOxxxlifeless: should I try merging trunk into this branch?04:41
spivpoolie: btw, the 405 is apparently from .bzr/branch-format, not .bzr/smart, even so treating it more like a 404 does seem more useful in practice.04:42
pooliei know04:43
lifelessdOxxx: I've merged trunk already04:43
pooliehello lifeless!04:43
pooliespiv, i was going to try TransportNotPossible04:43
lifelesshello poolie!04:43
poolieit's just possible it'll come up in other places04:43
dOxxxin other news, I really need to improve the mac build process so that it checks for missing tarballs *first*04:44
dOxxxwaiting 20 minutes while it builds only to discover that I typo'd a tarball name in the config really sucks04:44
dOxxxdoing this three times in a row..... priceless.04:45
lifelessdOxxx: I'm running --parallel-fork, I am getting the normal progress bar:04:46
lifeless[407 in 1m18s] per_repository.test_repository.TestRepository.test_item_keys_introduced_by(Reposit04:46
lifelesstrying subprocess now04:46
lifelessdOxxx: subprocess working for me04:48
lifelessdOxxx: I'll speculate that stdout in the child processes isn't wired up correctly.04:48
lifelesspoolie: which PPA, if any, would you like me to upload python-testtools into, and which versions of ubuntu for?04:51
pooliegood idea doxxx04:51
pooliei'd like it to be possible to run selftest on the packaged bzrs04:52
poolietherefore, for it to be in at least ~bzr for ubuntu versions that don't ship it04:52
pooliewell, i guess ~bzr-beta-ppa04:52
pooliethough the betas may not be in there yet04:52
pooliethere are also ppa dependencies, but i think that's only for building?04:52
lifelessso, hardy, jaunty, karmic ?04:52
poolieppas have been sufficiently successful to create new problems :/04:53
poolieyes04:53
lifelesslucid will have the right versions once various archive syncs catch up04:53
dOxxxlifeless: silly question, but are you actually running on a multi-CPU machine?04:54
dOxxxlifeless: now I'm not getting the problem...04:55
lifelessdOxxx: yes, 2 cpu04:56
dOxxxlifeless: I mean the garbled output problem. it seems coherent now, but still subunit-like spam.04:56
dOxxxand no progress bar04:56
dOxxxlifeless: I'm not running it through any subunit filter. Should I be?04:57
lifelessno04:57
lifelessit wires one up internally04:57
lifelessthere are two possibilities04:57
lifelesseither stdout on the child is attached to the terminal, not a pipe to the parent04:58
lifelessor something is introducing noise into the pipe and confusing the parser04:58
dOxxxcould it be my method for installing subunit? I just copied the python modules into my site-packages dir since I couldn't use the makefile04:58
lifelessdOxxx: that should have worked fine04:59
lifelessat the moment there isn't any C code needed04:59
dOxxxit looks like I only get the garbled output when I press Ctrl-C to stop05:00
lifelessah05:00
dOxxxI suspect that stderr and stdout are mixing05:00
lifelessperhaps the chilren are not being killed appropriately or something05:00
dOxxxif I recall correctly, ctrl-c is sent to all subprocess associated with the terminal, in win3205:02
* lifeless shrus05:02
lifelessif, when you don't hit ctrl-C,05:02
lifelessyou get a process count and test name showing05:02
lifelessthen robert is happy05:02
dOxxxthis is what I'm getting, and lots of it: http://pastebin.com/d7623943b05:03
pooliei was hoping someone would say "how about you help?" - go nmb05:05
jmlwhat does "IllegalUseOfScopeReplacer" mean? http://paste.ubuntu.com/342421/05:06
jmlI'm trying to follow a review suggestion to use lazy import05:06
dOxxxpoolie: I'm still trying to refine the process of building mac packages. :P05:07
jmlffs05:07
pooliejml it means05:09
poolieoh, blah05:10
poolieit means fiddle with the way you're using it05:10
pooliemake it a global05:10
pooliei've read the review but i don't recall the details05:10
poolieessentially i think you should have your __init__ only lazy_register_command()s05:10
poolieand then the command module can load its heavy dependencies05:11
dOxxxuploading bzr 2.0.3 package for mac :)05:12
jmlpoolie, lazy_register_command might be the ticket, thanks.05:12
lifelessjml: it means that something outside the module accessed the lazy import05:12
lifelessdOxxx: that pastebin is raw subunit05:14
jmllifeless, so what should I do about that?05:14
lifelessjml: don't do that :). its likely one of your other modules. They should import (lazily if possible) it themselves.05:14
jmlpoolie, lazy_register_command doesn't seem to be a thing05:14
lifelessrather than doing foo.bar.baz05:14
jmllifeless, hmm. why isn't this a problem if a plugin fails to lazily import trace, say05:15
lifelessjml: I'm leaving the house, you can ring me and I will help you happily, if you give me 3 minutes before ringing.05:16
jmllifeless, ok, thanks.05:17
dOxxxlifeless: so I guess that means that the filter is failing05:17
lifelesspoolie: on the subunit branch for bzr, I hope I'm not coming across as pushy, I'm just trying to execute on somethings you've been saying for a bit.05:18
poolieyeah05:18
pooliei think you're being appropriately pushy05:18
pooliei just wanted to make my comment more clear05:19
pooliei'd rather merge it now than after the last beta05:19
pooliejml if you push or pastebin or gobby your code i will look05:19
jmlpoolie, thanks.05:19
jmlpoolie, this diff: http://paste.ubuntu.com/342430/ from https://code.edge.launchpad.net/~jml/bzr/lp-login-oauth-2/+merge/15897 -- branch lp:~jml/bzr/lp-login-oauth-2 is up-to-date05:22
poolieoh you're trying to lazy import a submodule of the current module?05:25
jmlpoolie, I guess I am.05:25
poolieor whatever the right word is05:25
pooliethis is from05:25
poolie<jam> Also, it seems like it might be nicer to lazily import lp_registration.05:25
poolie?05:25
poolies/from/motivated by05:26
jmlpoolie, that's right.05:26
pooliequick call?05:26
jmlpoolie, sure. +6136264317005:26
dOxxxmac installers are build and uploaded so I'm going to bed. good night!05:41
pooliejml http://en.wikipedia.org/wiki/List_of_Ubuntu_releases#Release_history05:58
poolieso quite a few are still supported: dapper (server only), hardy, intrepid, jaunty, karmic05:59
poolieand in a sense lucid05:59
jmlpoolie, citation needed!05:59
poolie:)05:59
poolie"moving violation"05:59
jmlpoolie, where are they getting our support policy from?06:00
spivjml: if you do "from from bzrlib.plugins.launchpad import lp_registration as _lazy_lp_registration" in the __init__.py, does that fix it?06:01
jmlspiv, maybe.06:01
pooliei don't know06:01
poolieit's consistent with my memory of it06:01
pooliespiv i think we solved it06:01
jmlspiv, I've abandoned the lazy registration angle, I'm afraid.06:01
spiv(and s/lp_registration/_lazy_lp_registration/, obviously)06:02
spivjml: :)06:02
jmlhi06:37
jmlhttp://paste.ubuntu.com/342486/ -- from win32utils06:37
jmlis the second import as unnecessary as it looks?06:37
jmlalso, is "WindowsError" a builtin on win32?06:38
* spm must fight temptation to suggest it's a ground state of being....06:48
spivjml: file:///usr/share/doc/python2.6/html/library/exceptions.html#exceptions.WindowsError implies yes06:51
jmlspiv, thanks.06:51
jmlspiv, that's probably a bug in pyflakes then06:51
spivjml: yes, that second import looks redundant (and weird)06:52
jmlspiv, I actually deleted the first one06:52
spivWell, arguably it's the first import -- right.06:52
jmlhmm.06:57
jmlactually, it's probably not a bug.06:57
jmlit's just annoying.06:57
pooliehm07:02
pooliegoogle code is either flaky or intermittently banning me07:02
poolieor both07:02
pooliecould someone test https://spyderlib.googlecode.com/.bzr/branch-format for me?07:23
poolieactually nm07:23
pooliereally just down07:24
vilahi all07:27
jmlvila, hello07:42
vilahi jml ! You're up early !07:42
vila:D07:42
jmlvila, well, the sun is shining07:43
vilajml: Lucky you07:43
jmland it's well over 20 degrees celsius07:43
jmlI thought07:43
jmlwhy not be awake07:43
jmlhttps://code.edge.launchpad.net/~jml/bzr/lp-login-oauth-2/+merge/15897 -- I've replied to John's review07:43
jmlit's ready for another round.07:43
vilaOh yeah, 3 minutes ago :)07:44
jmlvila, why wait even that long!07:45
pooliehello vila07:45
pooliejml, i'm pretty much ok with it but07:45
pooliei greedily want to finish my own patch first07:45
jmlvila, it's very common for Launchpad branches to be landed within a day after submission for review.07:45
poolieand to talk to vila!07:45
* jml is off to eat food.07:46
vilaIt looks like you don't need piloting, you're doing well and already have one core reviewer :) jam sounded ok so I don't see why he would not approve.07:46
vilaYour branch is on my radar anyway, I've just not come to it yet07:46
vilajml: Do you need two votes here ? You have pqm access right ?07:47
jmlvila, I have PQM access.07:47
jmlvila, I only need one vote07:47
* poolie will look, it's paged in07:47
vilapoolie: ok07:47
jmlI'd really love to rewrite the Launchpad plugin from scratch.07:48
jmlexcept that's always a terrible idea07:48
vilajml: pilot work done :-) I will ensure you enjoy your flight anyway :)07:48
* jml off to eat food, for real07:48
jmlvila, :D07:48
pooliejml did you ever work out how to test lplib?07:49
vilawhat ? no tests from jml ? Surely he forgot to push the right thread ? :D07:50
vilajml: before you raise some ill-founded objections, I'll be ok with a soft dependency on a locally installed launchpad :D07:51
PengWoah, an lp-mirror command. Awesome!07:54
poolievila: there was a thread a while ago about 'how on earth do you test lplib?'08:00
vilahmm, before the Mooloolaba sprint ?08:00
vilare-phrasing: before jml tought me how to installl lauchpad ?08:01
pooliei should go soon...08:01
poolieprobably yes08:01
jmlvila, the way to test it is to use a mocked launchpadlib, basically08:02
jmlvila, launchpadlib has support for this, but this support is not available in karmic.08:02
jmlvila, nor in any released version of launchpadlib.08:02
vilajml: that's one way, the one I was hinting above is more like the local_test_server plugin approach were you require a true server locally running and inject it in the bzr test suite08:03
vilajml: whatever works !08:03
vilaminimizing entropy, etc08:04
jmlpoolie, I've just pushed up a change to the NEWS file.08:09
pooliek08:11
pooliejml, this is the point where we poke the 'diffs get out of date' bug?08:12
poolieactually are you really sure you pushed?08:12
poolieit doesn't even show the revisions08:12
jmlpoolie, ahh.08:14
jmlpoolie, bzr: ERROR: These branches have diverged.  See "bzr help diverged-branches" for more information.08:14
jmlpoolie, injudicious use of uncommit, plus errors all looking the same in text08:14
pooliemm08:14
jmlpoolie, I just successfully pushed.08:15
pooliegood for you :)08:24
poolieme too08:24
pooliehttps://code.edge.launchpad.net/~mbp/bzr/497274-http-405/+merge/16229 may unblock bzr-git imports from google code if we're lucky08:24
MvGGood morning, or whatever time of day might be appropriate to you!08:25
MvGWould I be correct to assume that methods decorated with needs_read_lock will ALWAYS unlock their object once they are done?08:26
pooliethey will decrement its lock counter08:26
poolieif it goes to zero they will unlock it08:26
MvGpoolie: Ah. Didn't see any lock counter in the Branch implementation...08:27
MvGBut otoh, branch is mostly forwarding to its control files, it seems.08:27
MvGOK, so it might be I forgot to lock things...08:28
ZapeliusI have a driver-foo.c -file, now there is a modified new hw revision and I would need a slightly modified driver for that. how do I copy that driver-foo.c file with history to driver-foo-a.c ?08:50
spivZapelius: bzr doesn't track file copies, unfortunately08:51
spivhttps://bugs.edge.launchpad.net/bzr/+bug/26909508:52
ubottuUbuntu bug 269095 in bzr "bzr missing "cp" command for forking files /w history" [Medium,Confirmed]08:52
spiv(currently the top bug on https://bugs.edge.launchpad.net/bzr/+bugs?orderby=-users_affected_count !)08:52
Zapeliusyep, found that too. there's no work-around-hack?08:53
spivNot that I can think of, but what behaviour are you after?08:54
spivJust that 'bzr log FILE' and 'bzr annotate FILE' would reflect the history from before the copy?08:54
spivOr do you expect something special to happen on 'bzr merge' when the file is changed?08:55
ZapeliusI have a branch that I'm working on, and I just wanted to have the entire history for the new file too, as it's up to a few lines identical to the old one09:03
* spiv nods09:04
spivIt's a reasonable enough request, just we haven't implemented it yet.09:04
Zapeliusk09:05
ZapeliusI have an option to make the driver itself dynamic, selecting the hw rev with an option09:06
CaMasonHi guys. I'm at r30, and I've got a changeset at r24 that I want to 'revert'. Would revert be the correct command?09:18
spivCaMason: probably you want "bzr merge -r 24..2309:24
spivCaMason: probably you want "bzr merge -r 24..23 ."09:24
CaMasonspiv, ah, thanks :)09:24
spiv(I think "bzr help revert" mentions this case :)09:25
CaMasonhm not working as intended :/10:01
CaMasonif I use `bzr diff -r 26..27` I see the changes I made. Those are the changes I want to get back. However, `bzr marge 26..27 .` I don' get them10:02
CaMasonbzr merge* even10:02
bob2-r10:02
CaMasonyes, sorry was using that10:03
CaMasonaha.. gotta use the revisions the other way around10:03
CaMasonworked perfectly :D10:04
spivCaMason: great :)10:09
CaMasonthanks10:09
=== Adys_ is now known as Adys
=== idnar_ is now known as idnar
bialixhi all10:32
johnflifeless, jelmer: Thoughts on http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=55846010:47
ubottuDebian bug 558460 in bzr "bzr: Needs python-pycurl to verify SSL certificates, but only Suggests it" [Important,Open]10:47
A4Techhi all!11:02
A4TechHow can I synchronize a local code so that updated my partner on launchpad?11:03
lifelessjohnf: according to policy, suggests is right11:04
lifelessits optional, so its not requires11:04
lifelessand in fact we find urllib to be more reliable11:04
bialixA4Tech: use pull and push commands11:04
A4Techlifeless: you can help me? :)11:04
A4Techusing pull  i'm get error11:04
bialixpastebin11:05
A4Techok11:05
A4Techbialix: wait  a second:)11:06
* bialix twiddles the thumbs11:07
A4Techbialix: in general, I get an error saying that I have not selected a working branch.11:08
bialixbzr pull URL11:08
bialixensure you're invoking this command from your local branch11:08
A4TechIt was after this command and received this error. Now make a new commit, I will show an error, but then I got involved in the latest version of code11:09
bialixpastebin?11:13
johnflifeless: we have Recommends for ca-certificates though11:54
johnfso sureley recomends for pycurl makes sense. Other wise the certs are fairly useless right?11:54
lifelessask vila11:55
A4TechWhat should I do if I wrote "branches have diverged" as bzr merge says "Nothing to do"12:01
beunoA4Tech, when do you get the diverged message?12:01
beunoit may be trying to merge from a different location?12:02
A4TechJust now I made the changes and then commit, and my friend also made a change, and now he wrote it.12:03
beunoso he is getting the message?12:05
A4Techno my fiend12:06
beunook, so he has to merge from you12:07
vilajonhf: Hmm, it's a bit messy...12:09
johnfvila: how so?12:10
* beuno waves at vila and lifeless 12:10
vilaToday, only pycurl do certificate validation, urllib don't12:10
vila\o12:10
vilaToday, only pycurl does certificate validation, urllib doesn't12:10
A4Techbeuno: and how?12:10
vilaIf you don't have ca-certificates, I'm pretty sure pycurl will whine loudly12:10
beunoA4Tech, bzr merge URL12:11
beunoA4Tech, are you using launchpad?12:11
A4Techyes12:11
A4Techbut12:11
vilaNow, if you don't have pycurl, then all is "well":  certificates are just ignored12:11
A4Techmy friend use ArchLinux :)12:11
beunoA4Tech, so "bzr merge URL_ON_LAUNCHPAD12:12
johnfvila: Right. So currently ca-certificates is a Recomends, while pycurl is a Suggests. So either we should make ca-certificates a Suggests or pycurl a recommends. I'm leaning towards the latter12:12
A4Techbeuno: start from 122 line http://pastebin.com/d155c387312:13
vilajohnf: Well, we plan de deprecate pycurl once we support certificate validation in urllib, so the current rules are not that far away from what we want...12:14
vilas/de/to/12:14
johnfvila:  ok. Well I've uploaded 2.0.3 anyway so it can wait till next time :)12:14
vilajohnf: all of that being said, yes, both should be in the same group until then12:14
vilawhether that group is Recommends or Suggests is a bit more subjective :-) If you care about certificates or not I think :)12:15
vilajohnf: While I have the dependencies ouder the eyes... celementree still needed for python-2.5 ? Really ?12:16
vilas/ouder/under/12:16
* vila shouts at his new keyboard12:16
* vila looks defiantly at beuno waiting for comments12:17
* beuno giggles12:17
johnfvila: funny you should ask12:17
johnfhere are two more  bugs12:17
johnfhttp://bugs.debian.org/cgi-bin/bugreport.cgi?bug=55533612:17
ubottuDebian bug 555336 in bzr "bzr: Embedded code copy of python-configobj" [Normal,Open]12:17
johnfhttp://bugs.debian.org/cgi-bin/bugreport.cgi?bug=55534312:17
=== jelmer_ is now known as jelmer
ubottuDebian bug 555343 in bzr "bzr: Embedded code copy of python-elementtree" [Normal,Open]12:18
vilaI'm pretty sure we have some minor changes from configobj but I don't remember which12:18
vilafor the later... what ? We have that ? Where ?12:19
vilajohnf: right, so for the later, the copy is there for python-2.4 support, so youcould mark that bug as invalid or whatever the idiom is there, I don't think debian policy requires you to do the kind of brain surgery required to get rid of it12:25
johnfvila: ok thanks12:27
lifelessvila: actually, debian is pretty aggressive about duplicated libraries12:28
vilalifeless: I can understand that and why, but here... 2.5 being supported and upstream supporting another version, how do they address that kind of problem in other contexts ?12:29
lifelessvila: file a bug, talk, talk some more. And if you're lucky, talk some more.12:31
vilalol, you mean: until upstream stop trying to the packaging itself ? :D12:32
lifelessdeath by discussion12:33
lifelessseriously though, they would patch it out in debian12:33
lifelessand file a bug upstream12:33
vilaAFAICS, in that particular case, we may get rid of the python library in util/. The only one that will need it is RedHat were there should be a way to add celementtree in the packaging.12:33
lifelessvila: yeah. still, E-On-Leave :P12:34
vilalifeless: thought so :)12:35
vilalifeless: enjoy !12:35
lifelessvila: you might like my subunit branch of bzr :)12:35
lifelessvila: if you could give it, with subunit 0.0.3 & testtools 0.9.2 a spin on babune that might be good to catch anything up with it.12:35
vilalifeless: I have my eye on it since the begining but gave priority to non-core while pp'ing and then.... time flew :-/12:36
vilaBut yes, that was my plan12:36
vilaand sorry for delaying that kind of check so long :-/12:36
lifelesshey nothing to apologise for12:37
lifelesstesttools was only released this morning12:37
vilawell, I was waiting for some release of test ... exactly, so the delay is ... today :)12:37
vilalifeless: so that includes you protocol branch right ?12:37
vilas/you/your/12:38
lifeless0.0.3 does, yes12:38
lifelessif you want to see how it looks, break a test, run selftest --parallel, and look - you should get the log passed across nicely.12:38
vilaI think my big question about your bzr branch was how it fall backs if neither subunit nor testtools is available and how it behaves when wrong versions are available12:39
lifelessso, it doesn't alter the subunit aspect at all12:40
vilaother than that, if you get rid of the horrors of get_log in bzr, I'd love it :-D12:40
lifelessolder subunit will degrade cleanly - you get the log, but as a 'mega backtrace'12:40
lifelessolder testtools, and bzr selftest will die horribly while loading the tests.12:40
* vila reading lifeless/bzr/subunit12:47
vilaBZRTransformingResult death ? \o/12:47
vilalifeless: I'm pretty sure you didn't fix the XXX in _get_log(), if further content has been added returning the cached content is still too aggressive12:54
vilalifeless: E_OUT_OF_SCOPE accepted as long as you restore the comment :)12:55
lifelessvila: the xxx was wrong12:58
vilayou mean self._log_contents can contain the content flushed ? How ?12:59
lifelessit gets cached when the log is deleted12:59
vilayeah, and what happens to writes that occur *after* that ?12:59
lifelessmutter will fail ?13:00
vilaI'm pretty sure it didn't when I added that XXX13:00
lifelessthe statement was that unflushed stuff wouldn't be seen13:00
lifelessonce the log is closed, there cannot be unflushed stuf13:01
vilathe method is called _get_log() not close_log()13:01
lifelessif you mean 'mutter can be called after the log is closed and stuff goes nowhere', thats fine - but it belongs totally in a different place.13:01
lifelessvila: it is, but it serves multiple masters.13:01
vilaincluding tests that just want to peek at intermediate point13:02
lifelessvila: not anymore13:02
lifelessvila: and if they do, it *does not cache the result*13:02
vilanor provide a way to reset the cache :-/13:02
lifelessvila: no point once the log is gone.13:03
lifelessvila: the file is deleted from disk and all13:03
lifelessvila: I am asserting that the XXX did not describe a bug in this function, when I saw the function - in trunk.13:03
vilaI know, that's the problem ! While I don't like tests that look inside the log, I had to write one yesterday13:03
lifelessvila: my replacement API doesn't have this issue13:04
lifelessvila: but even in trunk, the XXX was complaining about a non issue.13:05
lifelessif you don't believe me, audit the code.13:05
lifelessthe cache was only set when the log was deleted13:06
vila!13:08
lifelessI believe you saw an issue, but I think you misdiagnosed13:09
lifelessand an XXX that is wrong is not helpful to fixing issues.13:09
rubbsgood morning13:11
vilalifeless: ok, get it, and addDetail use keep_log_file=True. I may have misunderstood the keep_log_file semantic at the time and never re-read the code carefully enough as I missed the cache created only if file deleted part13:14
vilaon the other hand, I think the relationship between bzrlib.trace._trace_file and self._log_file_name wasn't and is still not that obvious :-/13:15
lifelessvila: yes, that would fit my observation ... that and having to mention it 3 times here n now :P13:15
lifelessvila: I have 0 objections to you cleaning it up :P13:15
vilahehe13:15
lifelessjust let me land my patch first so we're not treading on each other.13:15
vilaHAHA13:15
vilaapart from the get_log needed to avoid duplication as other said, the last bit is that deleted addCLeanup13:17
lifelesshmm ?13:17
vilain test_trace_nesting, I'm pretty there was a file leak there13:17
lifelesshow so?13:17
vilabecause I found a file with the related content13:18
vilalifeless: hmm, may be you fixed the problem, I should have asked first: why did you delete that addCleanup ?13:20
lifelessbecause it was wrong :)13:20
vilabefore or after ?13:21
vilaor both ?13:21
lifelessit was deleting a log file from a test that had been finalised13:21
lifelessso I moved it into the teset itself13:22
lifelesssee the hunk above where I add it.13:22
lifelessso I haven't deleted it so much as moved it to be closer to the thing it deletes.13:22
vilayeah, I just realized that13:23
lifelesshave you had your coffee?13:23
vilaI'm going from the diff to the code, that move wasn't obvious :)13:24
lifelessI'm -> sleep. Let me distract you : http://build.robertcollins.net - see e.g. http://build.robertcollins.net/job/testtools-2.4/13:24
vilaOf course now that I re-read the diff, it *is* obvious ;)13:24
* vila cries13:25
lifelessvila: ?13:26
vilalifeless: by the way, do you version any of that stuff (i.e. have you sorted out which files need to be versioned and which are not interesting)13:26
vilalifeless: I have hudson very on my plate except I can't find the time to get back to *that* plate :-D13:26
lifelessvila: which stuff?13:27
vilausing hudson13:27
vilaerr13:27
lifelessno, I don't version the config13:27
vilawhich files contains the job description or any part you modify in your hudsonsetup13:27
vilaok13:27
lifelessI keep the config lean13:27
lifelessand if its complex make a shell/python/whateve file to source - and that I version13:28
vilaFamous last words :-D13:28
lifelessis babune ubuntu ?13:28
vilayup13:28
vilawell, not only by definition, depends on what you call babune, but the host supporting most of the slaves is karmic13:28
lifelessvila: on the master, deb http://hudson-ci.org/debian binary/13:29
vilaalready installed13:29
lifelessshove that in sources.list or a .list file in sources.list.d13:29
lifelesswell then, you're all done :P13:29
vilahehe, no, I want to run a specific instance in a dedicated user and version the relevant bits :)13:29
lifelessthe deb creates a dedicated user13:29
lifeless/var/lib/hudson if you want to version stuff, but honestly, why?13:30
vilaENOTNOW !!!13:30
lifelessvila: 5 minutes, I'll bootstrap you.13:30
vilaWhy ??? Oh my, if only to rollback when I break stuff !13:30
lifelessexclude jobs13:31
lifelessactually13:32
lifelessinclude jobs13:32
vilalifeless: I have an instance running on locahost:8080 (11 days ago it says even :)13:32
lifelessexclude 'workspace' inside each job13:32
lifelesshave you setup security?13:32
vilaApart from the port not being visible on internet, no13:32
lifelessgo to /configure13:33
lifelessfor security realm, select 'hudsons own user db', and 'allow people to sign up'13:33
lifelessselect 'logged in users can do anything'13:33
lifelesssave the config (button at the bottom)13:33
lifelessclick on sign up in the top right13:34
lifelessgive yourself a username(e.g. vila) and passsword etc.13:34
lifelessnow go back to /configure13:34
lifelessand click on 'matrix based security'13:34
lifelessit will probably have just anonymous there13:35
lifelessif so, give anonymous both read permissions (second and middlish positions)13:35
lifelessadd your username13:35
lifelessand click every tick box in that row13:35
lifelessadd 'authenticated' as a user, and give it both read permissions13:35
lifelessclick save13:35
lifelessthat should take you back to the main screen13:36
lifelessand still give you a 'manage' link13:36
lifelessyou can add more admins such as john or me via that form.13:36
viladone13:37
vilaerr, except the last :)13:37
lifelessinstall the bazaar plugin from http://build.robertcollins.net/pluginManager/13:37
lifelesswe'll need to add our own accounts first, no rush,13:37
lifelesssorry, /pluginManager13:37
lifelessonce its installed, click the restart button, or use invoke-rc.d hudson restart13:38
lifelessthen visit /view/All/newJob13:38
lifelessjob name 'bzr.dev', type 'build a free software project', OK13:39
lifelesssource code management -> bazaar, repository url 'lp:bzr'13:39
lifeless'poll scm', @hourly13:40
lifelessadd build step 'execute shell', put in13:40
vilawilll trigger only if changes appear there right >13:40
lifeless./bzr selftest --parallel=fork --subunit | subunit2junitxml -o tests.xml13:43
lifelessclick 'publish junit test result report'13:43
lifelessput in *.xml as the field13:43
lifelessclick on save13:43
viladone13:44
vilaWell, I added a 'make' build step first13:44
lifelessok13:44
lifelessso, assuming you have subunit and python-junitxml installed, this will just work.13:44
vilathey are installed only for my user :)13:44
lifelessok, so ideally you'd install them13:45
vilaso I guess it will just breaks ?13:45
lifelesshowever13:45
lifelessanyhow - make the build shell be what I pasted above13:52
lifelessand it will grab our deps from lp13:52
lifelesssave the config13:52
lifelessclick on build now13:52
vilait complains about -f13:52
vila./bzr selftest --parallel=fork --subunit | ./subunit/filters/subunit2junitxml -o tests.xml -f | ./subunit/filters/subunit2pyunit13:52
lifelessreally ?13:53
lifelessoh bugger13:53
lifelessoptparse faaail13:53
lifelesssubunit 0.0.4 will be coming to a doorway near you tomorrow13:54
lifelessadd 113:54
lifeless-f 113:54
jelmerhah, boolean option?13:55
lifelessyeah13:55
lifelesswell, meant to be :P13:55
vilarunning13:56
vilainstant feedback via console output, good13:57
vilaprogress bar as barbershop, less good13:57
vila:D13:57
lifelessvila: should be nearly done ;)13:59
lifelesson your beefy machine13:59
* vila tracking the python processed by use hudson14:00
vilaurgh, zombies ? 3 of them, bad14:01
vilasome other processes are still running though, so let see14:01
vilano more feedback in the uhdson console :-/14:01
lifelessdoes hudson think its still running?14:01
vilastill the spinning wheel but no more *test* names14:02
lifelesspossibly the setup test14:02
lifeless:P14:02
* vila kills the build14:03
vilano more zombies, good14:03
lifelessthats a little harsh :P14:03
lifelessok, put test_plugin or something on the selftest line14:03
vilafar too long14:03
lifelessso I can go to bed ;)14:03
lifelesssee, the US is getting up. Now I know its late.14:04
vilahehe14:04
vilaRan 0 tests in 0.172s14:04
vilaOK14:04
rubbslifeless: haha14:05
lifelessvila: thats perhaps a little too fine a test selection14:05
viladichotomy :)14:05
vilaRan 208 tests in 1.496s14:05
vilaOK14:05
vilaRecording test results14:05
vilaFinished: SUCCESS14:05
vilalifeless: sleep well :)14:06
lifelessvila: right, now on the project page14:06
lifelessthere should be a test graph14:06
lifelessand a 'last test result'14:06
vilagraph, yes, latest, yes14:06
lifelessclick, click enjoy.14:06
vilarhaaaaaa lovely :)14:06
lifelesswhat else do you need to know?14:07
vilaactivating remote slaves, I read somewhere you should do that early14:07
lifelessok14:07
lifelessare you using vmware? kvm? ec2?14:08
vilavirtualbox, real boxes so far, ec2 very soon for desolation14:08
lifeless#hudson btw, good folk there if you need a hand.14:08
vilagood14:08
lifelessso, virtualbox oesn't have a plugin yet14:08
lifelessdo you start the slaves and leave them running?14:08
vilafor vbox, yes,14:09
lifelessok14:09
lifelessgo to /computer/new14:09
vilaexcept lately were I need to reboot them far too often, but that's most probably a vbox bug14:09
lifelessselect 'dumb' slave14:09
lifeless(vs automatic provision like ec2 does)14:09
lifelessexecutors is concurrency on the slave14:10
lifelessfs root - e.g /home/babune/hudson14:10
lifelesschange launch method to ssh14:11
lifelessoh14:11
lifelessyou probably want the platform labeller plugin14:11
lifelessfor tying jobs to os's14:11
lifelessbut it won't really matter at this point14:12
vilaoh I would certainly like that14:12
lifelessyou can tie by the node name too14:12
lifelessso if your node is called 'freebsd' you'll probably be ok for now :)14:12
lifelessbut install platform labeller, its mine ;)14:12
vilaok, noted14:13
vilaprivate key path, damn, needs to run as babune again :)14:14
lifelessln :P14:16
lifelessor cp probably better14:16
lifelessK, crashing.14:17
lifelessall you need now is java installed on the slave14:17
vilasure, thanks a lot, that helped !14:17
lifelessand you can either add a new copy, as a copy of this one14:17
lifelessand 'tie to a node' in both jobs14:17
lifeless*or814:17
lifelessyou can play with matrix builds14:17
lifelessI suggest the former, for now. Its easier to grok.14:17
lifelessif a tad more manual.14:17
vilayeah a bit too much clicks for me :-) Another reason to version the stuff, hopefully the xml bearable...14:18
vilais14:19
lifelessthis process - /usr/bin/python /var/lib/hudson/jobs/bzr.dev/workspace/bzr --no-plugins serve --port localhost:0 - is stuck for me14:19
vilaoh my, it's installing the jdk :-)14:19
lifelessI suspect its what you had too ?14:19
vilayup,14:19
vilaI didn't mention it earlier but it appeared early in the game14:20
lifelesstest_breakin14:21
lifelesstest_breakin_harder14:21
lifelessthey were the problem14:21
vilahehe, back they are :0)14:21
vilaGood,looks like it's mainly config.xml fileS under /var/lib/hudson for the main, the jobs and the hosts14:22
lifelessthis is what test failures look like: http://build.robertcollins.net/job/bzr.dev/lastCompletedBuild/testReport/14:22
lifeless(with my patch they will have the log)14:23
vilagee, /systeminfo, refreshing transparency....14:27
vilaDisk space is too low. Only 1.012GB left. Come on hudson....14:29
lifelessvila: /computer/configure14:36
lifelessturn off things to your hearts content14:36
lifelessand thats my final note; consider yourself unblocked:)14:37
lifelessinfuture, remember I'm happy to share knowledge!14:37
lifelessciao14:37
vilaI'm there, I'd like to tune instead, but thanks a ton for all the hints, highly appreciated !14:39
vilayou saved me hours14:39
=== vednis is now known as mars
=== weigon_ is now known as weigon
eric_fI'm curious if there's a way I can register nickname for branches more globally? I can use bzr nick trunk, bzr nick eric to allow me to easily switch between my "trunk" and "eric" branches, but when I'm bound to "eric" and want to merge in "trunk" I have to always use bzr+ssh://username@host:port/repos/project/trunk, I just want to simply type in "bzr merge trunk"15:48
eric_fNote that both branches are remote ones on a central server15:49
rubbsI'm not sure if this is what your looking for, but I'm sending it just in case15:55
rubbshttp://doc.bazaar.canonical.com/bzr.2.0/en/user-reference/index.html?highlight=intent15:55
=== AnMaster_ is now known as AnMaster
jamlifeless: we get ^M output on the stream until we run a test that ends up setting sys.stdout to binary16:06
jam\o/ for test suite isolation16:06
jam(That is CRLF until a expect binary output is run, and then we get LF only on the output)16:06
jamI think the test is "test_send" but any time you run_bzr with something that has output_encoding=exact it potentially causes it16:06
jamoh, and you can't run the filters via "C:\Path\to\subunit2pyunit" you have to use "C:\Python26\python C:\Path\to\subunit2pyunit"16:08
jamI alias them locally16:08
ad-530hiho16:09
ad-530need help with bzr builder/dailydeb16:09
intellectronicajames_w: can you maybe help ad-530 with some bzr builder problems he's having?16:09
james_whi ad-53016:12
ad-530hi james16:12
ad-530can i start?16:12
james_wad-530: of course :-)16:13
ad-530ok ^^16:13
ad-530i want to setup daily builds with bzr dailydeb from a sf.net svn repos and publish them in a ppa16:14
ad-530i've made a local branch of the svn repos and keep it up to date with bzr pull16:15
ad-530the svn working copy resides in /home/ad-530/prog/dangerdeep/trunk/16:15
ad-530the local bzr branch in /home/ad-530/prog/dangerdeep/bzr/16:15
ad-530my recipe resides in /home/ad-530/prog/dangerdeep/trunk/dangerdeep/packaging/ubuntu/daily/recipes16:16
ad-530the debian folder for packaging is located here /home/ad-530/prog/dangerdeep/trunk/dangerdeep/packaging/ubuntu/daily/karmic/debian16:17
ad-530the recipe content is:16:17
ad-530# bzr-builder format 0.2 deb-version 0~dangerdeep{revno}+0.0ubuntu~9.1016:17
ad-530/home/ad-530/prog/dangerdeep/bzr/dangerdeep/trunk/dangerdeep16:17
ad-530nest packaging /home/ad-530/prog/dangerdeep/trunk/dangerdeep/packaging/ubuntu/daily/karmic .16:18
ad-530now i've two problems16:18
ad-530fuck16:18
ad-530james_w, what was the last you'd recieved ?16:27
james_w<ad-530> ok ^^16:27
ad-530lol16:27
james_wnot good :-(16:27
james_wcould you repeat what you said?16:28
ad-530are the script kiddies attacking freenode excelusively or just their sponsors and the splits are a side effect?16:28
ad-530sure16:28
ad-530<ad-530> i want to setup daily builds with bzr dailydeb from a sf.net svn repos and publish them in a ppa16:28
ad-530<ad-530> i've made a local branch of the svn repos and keep it up to date with bzr pull16:28
ad-530<ad-530> the svn working copy resides in /home/ad-530/prog/dangerdeep/trunk/16:29
ad-530<ad-530> the local bzr branch in /home/ad-530/prog/dangerdeep/bzr/16:29
ad-530<ad-530> my recipe resides in /home/ad-530/prog/dangerdeep/trunk/dangerdeep/packaging/ubuntu/daily/recipes16:29
ad-530<ad-530> the debian folder for packaging is located here /home/ad-530/prog/dangerdeep/trunk/dangerdeep/packaging/ubuntu/daily/karmic/debian16:29
ad-530<ad-530> the recipe content is:16:29
ad-530<ad-530> # bzr-builder format 0.2 deb-version 0~dangerdeep{revno}+0.0ubuntu~9.1016:29
james_wad-530: I'm not sure, I would suspect freenode itself16:29
ad-530<ad-530> /home/ad-530/prog/dangerdeep/bzr/dangerdeep/trunk/dangerdeep16:29
ad-530<ad-530> nest packaging /home/ad-530/prog/dangerdeep/trunk/dangerdeep/packaging/ubuntu/daily/karmic .16:29
ad-530now i've two problems16:29
ad-530first is the nest line16:29
ad-530is "." right?16:29
ad-530or is the "nest" itself right?16:30
ad-530most howtos are using merge but that gives an error16:30
ad-530not the same ancestor16:30
ad-5302nd problem is when running bzr dailydeb with that recipes it gives the following error msg:16:31
james_wyou can't nest to ".", not16:31
james_werr, "no"16:31
james_wyou can nest to "debian"16:32
james_wso the line would be16:32
james_w nest packaging /home/ad-530/prog/dangerdeep/trunk/dangerdeep/packaging/ubuntu/daily/karmic debian16:32
james_whowever, it depends on what is in that branch as to whether that will do the right thing16:32
ad-530well, in all the howtos the guys are merging a "packaging" branch into16:33
ad-530thought it's a good idea to have several packaging files for several distributions16:34
ad-530but merging gives an error about the two branches don't have the same ancestor16:34
james_wyou can do that16:34
ad-530with the new nest line there's no lock error furthermore16:36
ad-530but that produces /home/ad-530/blubb/karmic-0~dangerdeep1+0.0ubuntu~9.10/debian/debian16:38
ad-530guess i need to change the line to: nest packaging /home/ad-530/prog/dangerdeep/trunk/dangerdeep/packaging/ubuntu/daily/karmic/debian debian16:39
ad-530ah yes16:41
ad-530works16:41
ad-530thank you very much16:41
ad-530what is the manifest file used for?16:45
=== jam1 is now known as jam
=== abentley1 is now known as abentley
nekohayohey folks, what's the status of the bzr nautilus integration?17:16
=== beuno-lunch is now known as beuno
jelmernekohayo, there's something basic, but no developmens in the last little whle17:21
nekohayoand it seems like there's a new kid on the block, bzr explorer?17:22
nekohayois it out to replace olive?17:22
=== elmo__ is now known as elmo
maxbThe bazaar website move has broken the bzr-eclipse releases update site, is it available anywhere else?17:27
jamnekohayo: bzr explorer >> olive (at the moment)17:28
nekohayo>> ?17:28
jammuch much better17:28
nekohayooh17:28
nekohayowhat becomes of olive?17:29
rubbsit's till around, but it's mostly suggested that people switch to explorer.17:30
ad-530is it possible to specify ignores in builder/dailydeb ?17:32
james_wad-530: not currently, what do you want to ignore?17:49
ad-530a folder17:57
ad-530that contains a lot of data files that should better be in an own package17:58
=== mrevell is now known as mrevell-dinner
=== deryck is now known as deryck[lunch]
ad-530the trunk has a data/ and a src/ folder18:00
ad-530and a makefile of course18:01
=== mbarnett` is now known as mbarnett
ad-530best thing would be to have the local repos and inside one branch for the trunk/src/, one for the trunk/data/ and one for the makefile(s) inside trunk/18:06
ad-530that's not possible, right?18:06
james_wad-530: not currently, no18:11
rellisIs there a way to get bzr to convert windows newline character to unix?18:25
rellisWhen any random person commits..18:25
rellisI guess I'm looking for a pre-commit bit to turn CRLF -> LF.18:30
rellisnm found it18:31
=== deryck[lunch] is now known as deryck
lifelessvila: thats the idea; you should learn and tune, but basic setup a quickstart is a good idea20:37
lifelesshi jam, so there is a problem on win32?20:37
jamlifeless: there is a problem on all platforms20:37
jamGCCHKPacker isn't passing the reload func to the vfs20:38
jamso they are aborting20:38
jamraising the wrong exception20:38
jamthe fix is one-line20:38
jamlifeless: I'd love to chat about how to test it20:38
lifelessjam: I mean about subunut :P20:38
jamah, different patch :020:38
lifelessjam: but the autopack one. sure20:38
lifelessuhm20:38
jamhere is a start comment20:38
jamSo, IMO, to have a stable test, you'd ideally like a high-level per-repository test that a concurrent 'Repo.autopack()' doesn't fail with an error. Creating this situation across all repository formats seems difficult to do. You could create threads, but then you introduce timing race conditions. Monkeypatching to inject the race seems brittle against the repository format.20:38
lifelessso20:40
lifelesswhats the clearest statement we can make about the race20:40
lifelessheres my stab20:40
lifeless'when one pack completes other pack operations may be reading from a file it tries to obsolete' ?20:41
jamSo we have a test like:     def test_concurrent_pack_triggers_reload(self):20:42
jamBut it passes, because the Repo.texts etc all have reload func set appropriately.20:42
jamIt is only the GCCHKPacker code that suffers from this20:42
jamSo it has to be 2 concurrent pack operations20:42
jamand not just pack + some other op20:42
jammaybe just r1.pack(), r2.pack()?20:43
jamlifeless: I think your summary is reasonable20:43
lifelessdo they have to be partial packs, or will an autopack do ?20:43
lifelessanyway20:43
jamwell, autopack is triggering it today :)20:43
lifelesshere is a way to trigger it20:43
lifelessfirstly we should parameterise to just pack based repos, like there are just chk based tests. test_pack_repo.py already exists and we can use that I guess.20:44
jamyeah20:44
lifelessso, if we use a VFS20:44
jamper_pack_repository now20:44
lifelesswe can make io block, call other operations and whatnot20:45
lifelesshere is a simplish way to make a non-threaded race20:45
lifelessall we need is a way to figure out the name of one of the packs20:45
lifeless write a transport decorator which can trigger a callable on a specified trigger.20:46
lifelesse.g. t = decorate(self.get_transport())20:46
lifelesssetup a repo, two commits and then two repo objects20:46
lifelessthen do20:46
lifelesst.set_trigger('readv', one_pack_name, repo2.pack)20:47
lifelessrepo1.pack()20:47
lifelessthis will cause repo2 to repack under repo1's attempt to pack20:48
lifelessnice and synchronously20:48
jaminteresting point, this code:20:49
jamhttp://paste.ubuntu.com/342979/20:49
jamIn that order, all formats pass20:49
jamreversing those two lines20:49
jamall formats fail20:49
jamwell, 7 failures at least20:49
lifelessthats odd20:51
lifelessthat test is also pretty good, but it assumes that once the pack name index is read it isn't re-read till later20:51
jamyeah20:51
lifelesswhich the callback style trick I suggest above won't assume.20:51
lifelessas AIUI we now know matters20:52
lifelessor we've done something such that the existing test isn't strong enough20:52
jamsure, though I think I've also exposed another bug, since all the formats fail20:52
jamthe existing tests mix packing with record streaming20:52
jamand the bug in the current format is about mixing packing with packing20:52
jamwell, and reading indices20:53
jamit mixes a get_parent_map() call in one test and a get_record_stream().next() in the other20:53
lifelessso20:55
lifelessif you use the trace decorator20:55
lifelessand copy tree20:55
lifelessyou can find all the i/o with some granularity that pack does20:55
lifelessand then do it again N times20:55
lifelesswith each time a different IO hooked20:56
lifelessjam: do you know the name of an xfail test offhand ?20:56
jamI believe there is one in test_merge20:57
jam    def test_modify_conflicts_with_delete(self):20:57
=== ubott2 is now known as ubottu
lifelessthanks20:59
lifelessthanks21:00
jamlifeless: just to clarify, the trigger code would be a new class, right?21:00
lifelessyeah21:00
lifelessa new transport decorator21:00
lifelessthat with no triggers, does nothing unusal21:00
jamthat's what I thought, I just know that 'trace+' already exists21:00
lifelessbut with a trigger, fires a callback21:00
lifelessand (to follow a regular convention) can have the trigger kept if what you return evaluates True21:01
jamlifeless: you don't seem to have a "subunit.__version__" string. Is that intentional?21:04
jam(I often use something like that to figure out versions I have installed.)21:04
lifelessI haven't been doing that, I've started adding them to projects as I notice the lack.21:04
jamconsider this an observation of the lack :)21:05
lifelessjam: heh21:07
justdaveis there any way to make bzr pass specific SSL certificate handling options to libcurl when it communicates with an https repository?21:07
lifelessjustdave: not at the moment, i think its why we want to drop libcurl21:08
lifelessyou can make it not use libcurl21:08
justdavethat's probably work.21:08
justdavegot a repo with a local-CA-signed certificate, and it's been working fine for months...21:08
justdaveupdated the OS on the webserver, and it picked up a new version of libcurl and now I can't bzr up anymore :)21:09
lifelessyou could add that to your certificate store21:09
lifelessthe upgrade probably discarded it21:09
justdaveI figured out how to specify the root ca cert to check it against via an environment variable, which was a good temporary solution21:10
justdavebut it's also got a hostname mismatch actually.21:10
lifelessrotfl21:10
lifelessok21:10
lifelessso, I think its urllib+http://...21:10
lifelessits in bzr help urlspec, I think.21:10
justdavemakes it a little harder problem :)21:10
justdavein a few weeks this problem will go away when the repo moves to a dedicated server21:11
justdavein the meantime I need to convince it to work :)21:11
PengThe recent progress bar changes have lead to some suckiness for me: " M  NEWSKB     0KB/s | Pull phase:Merge phase:Resolution pass 1/10 0/1263sion paths 0/1"21:12
PengAh, only happens when piped to less.21:13
justdavedon't see anything mentioning urllib in bzr help urlspec, and tells me "Unsupported protocol" when I try it anyway21:14
justdaveprobably an old version of bzr or something21:14
lifelessno21:14
jamlifeless: so it turns out the reason why the '.pack() + .pack()' fails, is because .pack *doesn't* handle retry, only autopack does...21:14
justdavelooks like it's 2.0.121:14
lifelessjam: woo :P21:14
lifelessjam: do you remember justdaves thing21:14
jamjustdave: http+urllib:// iirc21:15
jamnot urlib+http21:15
jamyep: register_transport_proto('http+urllib://',21:15
justdavethat works.21:15
justdaveAll changes applied successfully.21:16
justdavewoot21:16
justdavethanks21:16
justdaveit's funny...21:16
justdave[root@mradm02 bugzilla.mozilla.org]# bzr pull --overwrite https+urllib://dm-bugstage01.mozilla.org/bmo/3.4/21:16
justdavehttps://dm-bugstage01.mozilla.org/bmo/3.4 is permanently redirected to https://dm-bugstage01.mozilla.org/bmo/3.4/21:16
justdavedidn't I specify it with a / on the end?21:16
justdave:)21:16
lifelessit gets normalised21:17
DanChow do I check out a copy of the python branch of lp:zim? i.e. https://launchpad.net/zim/pyzim21:18
DanCoh. duh.21:18
justdaveok, so how do I re-parent my local branch to that urlspec?21:19
justdave(so I don't have to specify it every time)21:19
justdavebzr switch says I can only do that to a checkout21:19
DanC$ bzr branch https://launchpad.net/zim/pyzim21:19
bialixpull --remember21:19
DanCthat worked; I'm kinda curious how that works. but I'm happy21:19
justdave--remember does the trick.  thanks!21:19
* bialix mutters: gary gary where are you21:20
jambialix: hiding from you.21:20
jam /wave21:20
jamby the way21:20
bialix?21:20
jambialix: I'm saying hi, and teasing you that gary is hiding from you21:20
bialixah21:21
jamsorry if it didn't translate well21:21
bialixhi jam!21:21
bialixsometimes I think I did something not good21:21
lifelessDanC: launchpad puts a bzr style symlink at series paths21:21
DanC"symlink"?21:21
lifelessyeah, with scare quotes ;)21:21
bialixjam: thanks for new instalers, waiting for announce now :-D21:22
DanCI know how filesystem symlinks work21:22
lifelesstechnically a branch reference object21:22
lifelesshas a .bzr/branch which is a branch reference, same as lightweight checkouts use.21:22
DanCwhat does it look like on the wire?21:22
DanCah.21:22
* bialix sighs: this new colored chatzilla is full of colors, so strange21:22
bialixluks: ping21:29
maxbI've been looking at bzr-svn trying to understand why there are trunk0, trunk1, trunk2, etc. layouts. I can't see any reason why there couldn't just be a single trunk* layout. Am I missing anything?21:36
lifelessusers are insane21:37
maxbusers?21:37
lifelessof svn21:37
maxbSorry, I'm not questioning why flexible layout handling needs to exist. I'm questioning why the trunk layout needs parameterization with a fixed integer21:41
maxbEspecially since I have a repository with projects at varying depths.21:41
lifelessI don't recall why sorry21:41
lifelessjelmer: ^21:41
jelmermaxb: hi21:42
lifelessit may be part of the older styles no longer recommended21:42
maxbhello21:42
jelmermaxb: what if you have a file named trunk in a directory named trunk ?21:42
maxbBut no one would be silly enough to do that, right? :-)21:42
lifelessmaxb: see under insane21:43
jelmersimilarly, not all directories in a repository usually follow the trunk, branches, tags scheme21:43
maxbWell, I guess I'll have a go at writing a trunk* layout, since I need it, and it seems mostly plausible21:43
jelmerso some people have a directory "releases" and having a flexible scheme would require us to descend into that directory to see if there happens to be a idirectory there somewhere named "trunk"21:44
maxb*Ah*21:44
maxbThat's the thing I was missing21:44
jelmerso, to summarize: a flexible schema would cause confusing behaviour and potential performance issues.21:45
jelmerargh21:46
jelmermaxb, did you see my last lines?21:47
maxb<jelmer> so, to summarize: a flexible schema would cause confusing behaviour and potential performance issues.21:47
maxbOK, I now understand why a flexi-depth layout isn't a good all-purpose solution21:47
maxbI think I can still make it work for me, however21:48
lifelessmaxb: note that if you do that, only you can use that svn repo with bzr, until you get the patch in other folks plugin21:48
maxbTrue, but it beats having to edit .bazaar/subversion.conf depending on which project I'm currently working on in that repository21:49
jelmerlifeless, that's not true anymore21:49
lifelessjelmer: oh, cool21:50
jelmermaxb: you can set a list of wildcard expressions21:51
maxbI could, but it just seems so messy21:52
maxbAlso, the wildcard layout does not push non-lefthand ancestry21:53
maxbIs there any flag to make bzr log show bugs-fixed info?22:05
bob2don't think so22:07
bob2qlog does iirc22:07
lifelessmaxb: it would be nice to have that22:10
PengI think there's a bug open about it.22:10
bialixmaxb: qlog22:18
maxbbialix: Yup.. except when I'm working over ssh :-)22:19
bialixqlog REMOTE_URL?22:19
maxbhmm... that's... a wonderful idea :-)22:22
bialix:-)22:22
* maxb fights trying to get qbzr-eclipse installed.22:23
maxb(for an unrelated matter)22:23
bialixmaxb: there was some issues with redirection from qbzr-eclipse@b-v.o to real location22:24
bialixmaxb: http://permalink.gmane.org/gmane.comp.version-control.bazaar-ng.general/6478322:25
bialixmaybe this related to your fight?22:25
maxbI think I've hopelessly confused my eclipse by trying to install both qbzr-eclipse and bzr-eclipse22:25
bialixlast time I've tried do this it succeed22:25
bialixthis summer22:26
=== Leonidas is now known as Xenonidas
maxbOh. No, it has worked, but I was confused because qbzr-eclipse doesn't seem to be a proper "Team" plugin22:28
maxbIt puts all its menus and options elsewhere22:28
bialixit puts its menu to main menu line22:29
maxbIndeed. Disconcertingly bizarre to someone used to how standard Eclipse VCS integration plugins22:29
maxbwork22:29
bialixsighs22:30
=== Xenonidas is now known as Leonidas
pooliehello maxb, bialix22:31
maxbhello22:31
poolieis there a problem with the list?22:31
bialixmorning poolie22:31
bialixthe list?22:32
bialixthe?22:32
maxbMailing list? I have an email from half an hour ago22:32
bialixconfirm22:32
bialixfreenode has some problems though22:33
poolieoh i was asking about22:44
poolie> bialix: maxb: there was some issues with redirection from qbzr-eclipse@b-v.o to real location22:44
bialixpoolie: http://permalink.gmane.org/gmane.comp.version-control.bazaar-ng.general/6478322:45
bialixjam sent the answer22:45
phoenixzfullermd: Hi there, dunno if you remember me, I switched from SVK/SVN to BZR a couple of weeks ago..22:46
=== jam changed the topic of #bzr to: Bazaar version control | try https://answers.launchpad.net/bzr for more help | http://bazaar-vcs.org | http://irclogs.ubuntu.com/ | Patch pilot: vila | bzr 2.1.0b4 and 2.0.3 released
phoenixzfullermd: Im having 2 problems with BZR, dunno if you might be able to help out?22:46
bialixphoenixz: just ask, there many people around22:46
phoenixzSure22:48
phoenixzproblem 1) I have my branch, and I have a branch (without WT) on a central server that acts as trunk. Communication is over bzr+ssh:// Now, the second developer joined the group, we try to push and pull but the second person gets access denied all the time, even though I made group permissions of the entire server brach "development" and we are both member of that group..  I chgrp development .bzr -R again, the second person has access, but as soon as I do22:50
phoenixz an update, we have the very same problem again.. How can I fix this problem?22:50
phoenixzDunno if I put that clear...22:50
fullermdUse a group with the two of you.22:51
phoenixzIts a permission problem when 2 persons try to push over bzr+ssh:// to a central repository ..22:51
fullermdIf should preserve the.22:51
fullermds/the/that/.22:51
phoenixzfullermd: eh, sorry, don't understand what you are saying..22:51
fullermdThough you may need g+s on a filesystem with SysV semantics; do you have that?22:51
jamphoenixz: did you setgid ?22:52
jamchmod 2775 -R .bzr22:52
fullermd(also, I presume you made it group writable, as well as chgrp'ing)22:52
phoenixzjam: setgid?22:52
phoenixzfullermd: correct, I added group write permissions22:52
jamphoenixz: If the bit is set, it sets the group for newly created flies22:52
jamto inherit from the directory22:52
fullermdWell, I wouldn't do it that bluntly, just on GP.  A find | xargs of just the dirs rather.22:52
phoenixzjam: ah, that sounds interresting...22:52
jamfind .bzr -type d -print0 | xargs -0 chmod 277522:53
jamfind .bzr -type f -print0 | xargs -0 chmod 66422:53
jamphoenixz: what is prob 2, btw22:53
phoenixzjam: well, Id like all this only to be available for development users.. could I also do 2770 and 660?22:53
phoenixzjam: eh... problem 2... crap.. I forgot, heheheh...22:54
jamphoenixz: sure22:54
phoenixzThere were 2 things, but I just cant get what the 2nd thing was..well, not too important then.. :)22:55
jamyou *might* need to look into setting the umask for the bzr process, but I think the chmod + chgrp should be enough22:55
phoenixzjam: fullermd: Going to try this.. thanks!22:57
phoenixzjam: ah, got the other problem!22:59
phoenixzone time I did bzr tag 0.3.0... then I made modifications and I needed to tag THAT one as 0.3.0, so, according to documentation I did bzr tag -f 0.3.0 to force retagging..23:00
phoenixzever since, I get Conflicting tags:23:00
phoenixz0.3.023:00
phoenixzhow can I fix this?23:01
fullermdYou'll have to move (or eliminate) it on the branch it's set at the old rev on.23:02
bialixdoes it only for me? bzr: ERROR: Unable to connect to SSH host bazaar.launchpad.net; (10061, 'Connection refused')23:02
phoenixzbialix: denyhosts?23:03
spivbialix: it's down for maintenance23:03
bialixthanks spiv23:03
spivbialix: topic in #launchpad says it should be back by 23:59 UTC23:03
bialixhmm, there is nothing on twitter channel23:04
jamspiv: so another... 2 hours or so?23:09
jamanyway, thanks to everyone who made 2.1.0b4 and 2.0.3 happen. I'm out for tonight.23:10
jamI should mention, the turnaround on binaries was pretty darn good this time. Thanks!23:11
sproatyis the web browisng interface part of bzr itself?23:12
spivjam: less than 40 minutes left, 'date -u' will show you the time in UTC23:22
lifelessspiv: assuming estimates are accurate predictors :P23:27
spivlifeless: :P23:27
Takssh: connect to host bazaar.launchpad.net port 22: Connection refused23:55
Takoh, damn - if there hadn't been so many splits, I would've seen that in the scrollback23:56

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