2012/03/02/#bzr.txt

[00:09] <Omni|Work> I want to create a 'full clone' of an upstream repository that has a number of branches.
[00:09] <Omni|Work> I want copies of all of those branches on my local system.
[00:10] <Omni|Work> Is this possible?
[00:10] <spiv> Currently, not in a single command.
[00:11] <spiv> You can of course make a local repository and then branch each upstream branch into it (and there might be plugins that can help you automate that?)
[00:12] <Omni|Work> How do I do that and ensure that the branches are 'branched into the local repository' rather than each turning into their own standalone repositories?
[00:13] <wilx> Omni|Work: You branch them into a shared repo.
[00:15] <Omni|Work> wilx: bzr init-repo something ; cd something ; bzr branch bzr://repo/a-branch
[00:15] <spiv> Omni|Work: right
[00:18] <Omni|Work> Interesting.
[03:25] <mgrandi> what does this error mean? bzr: ERROR: _ssl.c:319: No root certificates specified for verification of other-side certificates.
[03:29] <spiv> You're trying to connect to an HTTPS server?
[03:29] <mgrandi> i think, tried to just put a github url as the thing to merge with
[03:30] <spiv> It means that for some reason the pycurl library (which bzr uses for https) hasn't got any root certificates configured, thus it can't actually verify the authenticity of any server certificates.
[03:30] <mgrandi> well i'm on a mac, where are these certificates ?
[03:31] <spiv> I don't know, sorry :(
[03:33] <goddard> whats a good bzr tutorial for some one with a short attention span?
[03:33] <mgrandi> the bzr docs? oo
[03:34] <mgrandi> http://mercurial.selenic.com/wiki/CACertificates#Mac_OS_X_10.6_and_higher
[03:34] <mgrandi> needs the keychain, hmm >.>
[03:45] <goddard> mgrandi: thats funny
[03:46] <mgrandi> i mean, if you don't want to read some documentation wehn its pretty simple as that, i dunno what to tell you...
[03:48] <goddard> mgrandi: i didn't say it was me.
[03:49] <goddard> :D
[04:08] <poolie> mgrandi, there's an 'in ten minutes' doc on doc.bazaar.canonical.com
[04:08] <mgrandi> ah, tell that to goddard =)
[04:08] <poolie> i mean, goddard
[04:08] <poolie> :)
[04:08] <poolie> mgrandi, for you, there's an open bug that bzr doesn't know how to talk to the os x keychain, i think
[04:09] <poolie> there's a way to disable it
[04:09] <poolie> which bzr version are you using?
[04:09] <mgrandi> 2.5b6
[04:09] <mgrandi> it says to use -Oca.certs=none or something
[04:09] <mgrandi> which worked
[06:05] <poolie> mgrandi, ok good
[06:05] <poolie> i think that's fixed to be off by default in 2.5.0
[06:05] <mgrandi> using ssl certs?
[06:06] <poolie> yes on mac os
[06:06] <poolie> until it's hooked up
[06:06] <mgrandi> is there an api for the keychain?
[06:07] <mgrandi> how does any other program look up ssl certs then...
[07:14] <goddard> does bzr upload let you have a dev and a live upload?
[07:14] <goddard> or how do you do that?
[07:14] <poolie> goddard, i think you can just give it different urls to upload to
[07:15] <poolie> that's probably what i'd do
[07:15] <poolie> maybe two branches, depending on how much you expect them to vary
[07:15] <poolie> i would aim for not having too much variation
[07:17] <goddard> what is the point of branches?
[07:18] <goddard> i know it is the line of development for a project
[07:18] <goddard> but what should i use it for?
[09:09] <mgz> morning all!
[09:19] <poolie> hi all, and goodbye
[09:30] <wgz> hm, the bzr beta ppa is dead? maxb was maintaining and hasn't been updated since November
[09:31] <wgz> what should be suggested in bug 944557 as an alternative?
[09:31] <ubot5> Launchpad bug 812749 in bzr-builddeb "duplicate for #944557 Misuses bzr 2.4's new set_commit_message hook to disable editor prompting for a commit message entirely" [High,Fix released] https://launchpad.net/bugs/812749
[09:32] <wgz> ...you're being too clever ubot5, I wanted a link to the dupe
[10:25] <soren> I'm having trouble with bzr behind a proxy.
[10:26] <soren> bzr branch lp:reincarnate
[10:26] <soren> bzr: ERROR: Certificate error: hostname 'proxy-sjc-1.cisco.com' doesn't match either of '*.launchpad.net', 'launchpad.net'
[10:26] <Daviey> soren: Do they have ssl bump proxy?
[10:26] <soren> Both $http_proxy and $https_proxy read "http://proxy-sjc-1.cisc.com/"
[10:26] <soren> Daviey: It works well enough to talk to launchpad in w3m.
[10:27] <soren> Sorry, that was a lie. Both $http_proxy and $https_proxy read "http://proxy-sjc-1.cisc.com:80/"
[10:27] <Daviey> soren: right, but are they MITM'ling you?
[10:27] <soren> Daviey: Not enough to make w3m upset at all.
[10:28] <soren> wget is also perfectly content fetching things over https.
[10:28] <soren> So no. They're not.
[10:28] <Daviey> soren: right, but if *.cisc.com is a valid cert, it won't complain, right?
[10:28] <soren> Daviey: Then why would bzr?
[10:28] <Daviey> It seems to me that bzr has stricter checking.
[10:29] <soren> It really looks like the checknig is backwards.
[10:30] <soren> It gets bzr branch lp:reincarnate
[10:30] <soren> bzr: ERROR: Certificate error: hostname 'proxy-sjc-1.cisco.com' doesn't match either of '*.launchpad.net', 'launchpad.net'
[10:30] <soren> Gah, whoops. Sorry.
[10:30] <soren> It gets the '*.launchpad.net', 'launchpad.net' from the cert, but compares it to the hostname of the proxy.
[10:30] <soren> Which seems wrong.
[10:31] <wgz> soren: file a bug, the urllib2 checking is new and python doesn't help out much, so there are probably issues.
[10:31] <soren> *sob*
[10:31] <soren> I guess maybe corkscrew will work for now.
[10:31] <wgz> soren: it would probably be useful to try with pycurl as well and confirm that that works
[10:31] <Daviey> soren: try, -Ossl.cert_reqs=none
[10:32] <soren> Daviey: But I *do* want cert checking :)
[10:32] <soren> Daviey: But it works.
[10:32] <Daviey> soren: and you DO want to be MITM'd? :)
[10:34] <soren> Er.. no.
[10:34] <soren> I thought we agreed I wasn't being MITM'd?
[10:34] <Daviey> soren: but yes, it would seem it doesn't take proxy into account.
[10:35] <Daviey> soren: i didn't agree that..
[10:36] <Daviey> soren: it's likely you are correct, but they could be doing. http://wiki.squid-cache.org/Features/SslBump
[10:38] <wgz> soren: try this as a cute workaround:
[10:38] <wgz> BZR_LP_XMLRPC_URL=http://xmlrpc.launchpad.net/bazaar/
[10:39] <wgz> nope, wrong value...
[10:40] <wgz> annoying that there's no way to override the connection method of that
[10:41] <soren> uite.
[10:41] <soren> Quite, even.
[10:43] <wgz> soren: I take it `bzr info https://bazaar.launchpad.net/` also complains about the cert?
[10:43] <wgz> (just to confirm this isn't xmlrpc specific)
[10:46] <soren> wgz: Hm.  No.
[10:46] <soren> $ bzr info https://bazaar.launchpad.net/
[10:46] <soren> bzr: ERROR: Not a branch: "https://bazaar.launchpad.net/".
[10:46] <wgz> okay, then it's launchpad plugin specific
[10:47] <wgz> certainly a bug in our code or a dependancy then
[10:47] <wgz> please file :)
[10:48] <soren> wgz: Already did.
[10:48] <wgz> soren: thanks!
[10:48] <soren> https://bugs.launchpad.net/bzr/+bug/944696
[10:48] <ubot5> Ubuntu bug 944696 in Bazaar "When using a proxy to talk https, cert checking validates against proxy host name" [Undecided,New]
[10:49] <wgz> confirming that your browsers are happy talking to https://xmlrpc.launchpad.net/bazaar/ is probably also a good idea
[10:50] <soren> wgz: It just says "Your request didn't fit the protocol expected by this server."
[10:51] <soren> wgz: Which is also does from somewhere else, so that's fine.
[10:51] <wgz> yup, that's expected.
[10:54] <wgz> soren: the easy work around for now, provided you can ssh out, is to set your launchpad login so the lookup over http step isn't required
[10:55] <wgz> otherwise you can use the http url that lp:reincarnate resovles to to avoid the lookup for that particular branch
[10:58] <soren> wgz: I just added some more info to the bug (with a better stack trace)
[10:58] <soren> wgz: I can't ssh out, unfortunately.
[10:59] <wgz> well, `bzr branch http://bazaar.launchpad.net/~soren/reincarnate/trunk/` should work at least
[11:00] <soren> wgz: Indeed it does.
[11:00] <wgz> loses much of the convience of lp: urls but generally they get remembered after the first time
[11:46] <Pikkachu> I want to fork a branch but it's overkill, I just want one file...what should I do?
[11:46] <Pikkachu> is it possible to create a separate branch which has nothing to do with it for then merging on a per file basis?
[11:53] <jelmer> Pikkachu: branching is cheap, even for a small change it shouldn't have much overhead
[11:53] <jelmer> Pikkachu: can you perhaps try to explain what you're trying to do exactly?
[11:56] <Pikkachu> jelmer: patching purple plugin pack, it' a whole pack but I just need one single file from it
[12:00] <jelmer> Pikkachu: what is the problem with simply merging that file?
[12:03] <Pikkachu> jelmer: the problem is a common ancestral to begin with
[12:04] <Pikkachu> jelmer: I wanted to move irchelper/irchleper.c to the top but it seems to confuse bzr completely, to the point of saying non-sense. I'm gonna try keeping file structure...
[12:05] <jelmer> Pikkachu: you can force common ancestry by using "bzr merge -r0..-1"
[12:24] <Pikkachu> (bzr 2.3) bzr: ERROR: No final name for trans_id 'new-4', file-id: None, root trans-id: 'new-0'
[12:27] <jelmer> Pikkachu: what are you trying to merge into what exactly? I can give it a try with 2.5.
[12:27] <Pikkachu> the problem is how to being
[12:27] <Pikkachu> begin
[12:27] <Pikkachu> I have a branch with lots of stuff
[12:27] <Pikkachu> I jus want a new branch with only of those files, what to do?
[12:28] <Pikkachu> I want to merge updates to that secific file later
[12:34] <jelmer> Pikkachu: I would suggest just writing a script that pulls in the files you need
[12:36] <nisu> how is nested tree support comming for bzr? It's hard to find comprehensive up-to-date imformation about that
[12:39] <jelmer> nisu: there are small incremental steps towards nested tree support
=== Pikkachu is now known as Pikkachu[Away]
[12:39] <jelmer> with a bit of luck, 2.6 will have nested tree support
[12:39] <jelmer> I don't think we've published much about it
[12:40] <jelmer> I should finish off the update to the developer docs document about nested trees...
[12:41] <nisu> jelmer: sounds good
[12:41] <nisu> does that include tracking remote repositories of lets say subversion code as nested trees?
=== Pikkachu[Away] is now known as Pikkachu
[12:54] <Pikkachu> hm ok jelmer thanks anyway
[12:56] <fayaz> hi, is there any way i can convert a shared repository into one with no-trees?
[13:10] <jelmer> nisu: that's a separate topic, although in the case of bzr-svn and bzr-git most of the code to support svn-externals and git submodules is already present (just waiting for the support from bzr core)
[13:13] <nisu> jelmer: that is not exactly what I meant, I am more interested to know if it will be possible to import a plain svn or git branch using bzr-* into a nested tree of a pure bzr project
[13:14] <jelmer> nisu: ah, I see
[13:14] <jelmer> nisu: yes, that would be possible
[13:15] <nisu> jelmer: nice, can't wait to try it out :)
=== yofel_ is now known as yofel
[13:38] <Pikkachu> bzr branch pack fork; mv fork/sub/file .; rm -r fork/*; mkdir fork/sub; cp file fork/sub; cd fork; bzr add; bzr commit -m "Pack purge except for file"; bzr tag PURGE; <patches here>; bzr commit -m "my fork"; bzr merge ../pack/sub/file
[13:38] <Pikkachu> is this workflow ok?
[13:44] <jelmer> Pikkachu: sure
[13:50] <Pikkachu> how about making 'fork' --stacked for having it even more lightweight?
[14:02] <jelmer> Pikkachu: there is a --stacked option to 'bzr branch', if that's what you mean
[14:03] <Pikkachu> that's what I mean, I would like to stick the branch on exactly where I diverge
[14:04] <Pikkachu> so when I branch it from LP it's lightweight
[14:04] <Pikkachu> (I likely won't be any interested in changes prior to the divergence and purge)
[14:05] <Pikkachu> but I'm a bit afraid after reading the docs...:(
[14:12] <jelmer> Pikkachu: how do you mean?
[14:17] <Pikkachu> I'm just afraid of using it, nothing much...
[14:33] <james_w> if running some code using bzr is exiting and just printing "Killed" to the terminal, what would that indicate?
[14:36] <mgz> james_w: SIGTERM I'd guess
[14:37] <james_w> hrm
[14:44] <mgz> james_w: SIGKILL rather
[14:45] <james_w> I wonder if celery is getting involved even though I'm not actually going through it in this case
[14:45] <mgz> I was going to suggest you installed a signal handler, but doesn't help with that case
[14:48] <james_w> it's pretty hard to branch large branches if you can barely get the revision history from the server prior to whatever timeout is kicking in :-)
=== Pikkachu is now known as Pikkachu[Away]
[14:59] <james_w> aha
[14:59] <james_w> oom killer
[15:00] <mgz> ehehe
[15:00] <mgz> an old friend
[15:19] <abentley> jelmer: the daily builds of pqm for oneiric-lucid are failing.  I can't reproduce the errors locally-- even with no email set, I get a default.
[16:07] <jelmer> abentley: I think it might be related to the version of bzr
[16:07] <jelmer> let me see if I can reproduce it here
[16:09] <goddard> bzr just chokced on me over a file name
[16:09] <goddard> InvalidEntryName
[16:10] <jelmer> goddard: what's the file name?
[16:13] <abentley> jelmer: Yeah, I reverted to the version of bzr specified by the bzr package name, but still couldn't get it.
[16:16] <jelmer> abentley: it seems to've built fine here
[16:16] <jelmer> and there is a new build pending
[16:16] <jelmer> http://people.canonical.com/~jelmer/recipe-status/bzr.html
[16:17] <abentley> jelmer: unless you've changed those tests, I'd expect the pending builds (except precise) to fail.
[16:18] <jelmer> abentley: unfortunately I can't see any of the old builds on the daily build page (https://code.launchpad.net/~bzr/+recipe/bzr-pqm-daily)
[16:18] <abentley> jelmer: No, you can get to them via the PPA package listings, though.
[16:19] <abentley> jelmer: https://code.launchpad.net/~bzr/+archive/daily/+packages?batch=75&memo=75&start=75
[16:22] <goddard> this is what it is giving me
[16:22] <goddard> %\^5C5^5C597A95%?min_videos_edit.tpl.php
[16:23] <goddard> looks like a cached file to me
[16:25] <jelmer> goddard: that file seems to contain a backslash, which is a prohibited character
[16:36] <goddard> that is why i got the error
=== deryck is now known as deryck[lunch]
=== deryck[lunch] is now known as deryck
[20:07] <lamalex> Does anyone know when the -c flag was added to bzr merge?
[20:07] <jelmer> lamalex: it's really old - it's been there for as long as I can remember
[20:08] <lamalex> far out
[20:08] <lamalex> i wish i knew this ages ago. i thought you had to do bzr merge -r Y..X barnch
[20:09] <jelmer> there are a few other commands that support -c too
[20:09] <jelmer> 'bzr diff' and 'bzr log' in particular
[21:16] <wilx> Hi.
[21:16] <wilx> Is there any explanation anywhere about the three different merge algorithms?
[21:17] <wilx> Merge3, LCA, Weave?
[21:19] <wilx> Never mind, I have found this: http://doc.bazaar.canonical.com/developers/lca-merge.html
[21:23] <wilx> Hmm, still not enough information about the --weave algorithm.
[21:23] <wilx> Also, what is the --reprocess switch about? The help text is rather uninformative.
[21:24] <fullermd> Just try using it.  Then you'll be doubly uninformed.
[21:24] <wilx> Heh.
[21:24] <wilx> I want to know what to expect.
[21:24] <fullermd> There are some fair explications of weave merges out there.  I think --weave is not actually precisely a weave merge anymore, but it's comparable.
[21:25] <fullermd> If you can dig up some of the talk about "Precise Codeville Merge" from years back, I believe that's essentially a reinvention of weave merges.
[21:31] <wilx> I see.
[21:31] <wilx> Isn't LCA what monotone uses?
[21:33] <fullermd> I think mark-merge is a bit different.  But I was a little hand-wavy on the details when I had it all loaded into my brain, and that was years ago.
[21:36] <wilx> http://revctrl.org/FrontPage
[21:36] <wilx> I hate wiki spammers :(
[21:36] <wilx> They destroy so much useful stuff.
[21:36] <fullermd> Sadly, that's been abandoned to the spammers for many years now   :|
[21:50] <smoser> hey all.
[21:55] <wgz> hey smoser
[21:57] <smoser> trying to get a dailyb build going
[21:57] <smoser> and its not being nice
[21:59] <smoser> http://paste.ubuntu.com/865797/
[21:59] <smoser> any ideas ?
[22:02] <wgz> hm.
[22:03] <wgz> the error message seems pretty clear, but I'm not sure what you want.
[22:04] <wgz> throw a quick mail to the udd list
[22:04] <smoser> why would there be a tag for that thing ?
[22:05] <smoser> that string 'upstream-0.9.4+r4177' is only a creation of this daily build recipe
[22:05] <wgz> I don't know, I haven't done any of the fancy recipe things
[22:13] <wgz> jelmer will be able to point you in the right direction, and I'm sure he'll look at email over the weekend :)
[22:15] <smoser> ewll, i sent email just now, but i think i must have unsubscribed from that list at some point
[22:16] <smoser> and i just tried re-subscribe but still haven't gotten the conf mail
[22:16] <smoser> so dont know if my mail will go through
[22:17] <wgz> it arrived.
[22:17] <smoser> well, thanks for confirming that.
[22:17] <wgz> thanks, that's a clear problem report
[22:18]  * smoser steps away for weekend.
[22:18] <wgz> part of my issue with builder problems is they tend to get asked on irc, quickly answered by jelmer, then my only recall of them is searching through irc logs :)
[22:18] <wgz> have a nice weekend!
[22:38] <Amoz> jelmer, I'm trying to fix the bzr-gtk FTBFS in precise, and xvfb-run -a ./debian/testsuite.sh works well if I run it manually in precise, but fails in pbuilder due to an assertionerror
[22:38] <Amoz> any pointers of what could be happening here?