/srv/irclogs.ubuntu.com/2012/07/20/#bzr.txt

Noldorinno one :(00:05
lifelessNoldorin: its 1am or something for Jelmer00:31
=== jamesh__ is now known as jamesh
lifelessNoldorin: and you were addressing your discussion to him, so its not surprising :)00:31
Noldorinlifeless: 1am is the middle of the day for me :)00:31
fullermdBest time to lay out and work on your startan.00:34
Noldorinstartan?00:34
fullermdWell, with my skin, it's more of a starburn, but...00:35
Noldorinoh00:35
Noldorinheh00:35
Noldorinfullermd: what's the 'index' in bzr btw?01:11
=== zyga-afk is now known as zyga
mgzmorning08:16
=== frankoid_ is now known as frankoid
abentleymgz: jelmer wondered if maybe you'd take a look at this MP: https://code.launchpad.net/~abentley/bzr/branch-store/+merge/11561814:41
mgzwill have a look.14:46
mgz(relook)14:46
abentleymgz: ty.14:51
=== AfC1 is now known as AfC
pfsmorigohi, I want to understand the "branches" command. I just created a branch using "bzr branch repo1 repo2". If I execute "bzr branches", on both, will show the following message "* (default)". What this means?16:34
jelmerpfsmorigo: each of those directories has only one branch in it, the default branch16:34
pfsmorigojelmer: so, how do I create another branchs in it?16:35
jelmerpfsmorigo: if you have a shared repository with multiple branches in it, they should show up in 'bzr branches'16:36
jelmeror you can create new colocated branches (in newer versions of bzr), by running "bzr branch . co:newbranch" or "bzr switch -b newbranch" (which will switch to the new branch too)16:37
pfsmorigojelmer: the "co:" command doesnt worked by the second worked (at least the branch now have a new name)16:44
pfsmorigojelmer: so, I have two options. create a directory copy using "bzr branch oldcopy newcopy" or this colocated option. I'm right?16:45
pfsmorigojelmer: two directories for each branch or one for both, right?16:47
jelmerpfsmorigo: yes, and the one-directory-per-branch approach is used most commonly16:50
pfsmorigojelmer: I was reading bzr use guide and there is just the "two directories" approach... http://doc.bazaar.canonical.com/latest/en/user-guide/organizing_branches.html16:52
jelmerpfsmorigo: colocated branches (multiple branches in a single directory) is still experimental, and therefore not yet documented or recommended16:52
pfsmorigojelmer: hmm... now the parent branch is "."16:52
jelmerpfsmorigo: if you use the shared repository approach (which should be documented), "bzr branches" will report sibling branches16:53
pfsmorigojelmer: how can I keep the default branch with the original parent branch?17:25
jelmerpfsmorigo: how do you mean?17:33
jelmerpfsmorigo: the parent branch is the branch that a particular branch was based off17:33
pfsmorigojelmer: https://gist.github.com/753752177417d6c5b03c17:34
pfsmorigojelmer: there isn't a default branch anymore...17:34
jelmerpfsmorigo: the defualt branch is like HEAD in git17:44
jelmerpfsmorigo: it's only shown if it isn't referring to one of the named colocated branches17:44
pfsmorigojelmer: ok. I will read a bit more the help and test some commands. tks for the support!17:50
leo2007hello19:09
mgrandihello19:10
leo2007what is the best way to move a commit (someone else's) from one branch to another?19:11
leo2007I need to backport a commit from emacs-trunk to emacs-24.19:11
mgrandimerging?19:11
mgrandimake another branch of the branch from emacs-trunk at the commit you want to move, then merge it into emacs24?19:12
leo2007sounds a bit too heavy for just one commit.19:24
jelmerleo2007: it sounds like you want to cherrypick a single commit ?19:26
leo2007jelmer: yes19:26
leo2007how to do that?19:29
jelmerleo2007: bzr merge -c REV-TO-CHERRYPICK BRANCH-TO-CHERRYPICK-FROM19:30
jelmerleo2007: bzr commit -m "Cherrypick REV-TO-CHERRYPICK"19:30
mgrandiso that just takes the diff/changes of that revision and merges that, rather then trying to merge the entire tree at that revision?19:32
jelmermgrandi: yes19:33
leo2007jelmer: that seems to work but it didn't keep the authorship19:35
jelmerleo2007: right, it doesn't - you can use the --author argument to respecify the author19:36
leo2007a lot of hassle.19:37
jelmerleo2007: there is the 'bzr replay' command that's part of bzr-rewrite too, but it doesn't deal with conflict19:38
luke-jrPossible to get bzr to use diff3 for merge conflicts?19:59
maxbjelmer: Do you have an unpushed revision for lp:~jelmer/ubuntu/precise/bzr/2.5.1-sru for the -0ubuntu2 upload?19:59
jelmerluke-jr: bzr merge --diff3 ?19:59
mgrandi    --diff3             Merge using external diff3.19:59
luke-jrjelmer: I tried that, it doesn't seem to work19:59
luke-jrI still get non-diff3 conflicts20:00
mgrandido you have diff3 installed?20:00
luke-jryes20:00
mgrandislash can it find it? (look in .bzr.log to see if it reported anything)20:00
jelmerluke-jr: ah, you mean diff3 as a command line utility?>20:00
luke-jrthe log doesn't suggest it tried20:00
jelmerluke-jr: bzr merge --using=diff3 perhaps?20:00
luke-jrjelmer: I mean showing me TREE, common ancestor, and merging branch20:01
jelmerluke-jr: ah, --show-base will do that20:01
luke-jraha20:01
jelmermaxb: lp:~jelmer/ubuntu/precise/bzr/2.5.1-sru was just for the -0ubuntu120:02
jelmerI don't think I used UDD for the -0ubuntu2 upload20:02
maxbfair enough20:08
maxbjust assessing how best to represent things in the ppa-oneiric branch20:08
=== yofel_ is now known as yofel
Noldorinooh, busy tonight :)20:54
jelmerhey Noldorin20:56
Noldorinheya20:57
Noldorinjelmer: was takign a look at some bzr-git stuff yesterday after you were gone20:57
Noldorinjelmer: i'm tempted to say the way forward is using bzr-fastimport actually20:57
Noldorinin combination with git import20:57
Noldorinand whatnot20:57
Noldorinit's quite versatile in fact20:57
Noldorincan be extended to hg and other VCSs too in fact20:57
Noldorinthere are some derivative projects that do cool things actually20:58
jelmerNoldorin: bzr-fastimport is not bzr-git :)20:58
Noldorinjelmer: no but they can accomplish many of the same things ;)20:58
Noldorinthis is my point20:58
Noldorinand bzr-fastimport has versatility on my side20:59
Noldorinjust found out about it20:59
Noldorinbut it seems like an interesting way to do various vcs-interop tasks.20:59
jelmerNoldorin: bzr-fastimport is mostly interesting if you're using git and want to interoperate with bzr20:59
Noldorinjelmer: it can be used to mirror branches both ways21:00
Noldorinit's bidirectional ability is very good actually21:00
Noldorinjust saying :)21:00
jelmerNoldorin: git->bzr only works if you're importing into the same bzr branch you originally exported from21:02
jelmerNoldorin: so it's mostly useful if you're the only one using bzr, or if you're the only one merging from git21:11
Noldorinjelmer: it's good enough in the vast majority of cases21:11
jelmerNoldorin: depends on your use case21:12
Noldorinonly for intricate cross bzr-git workflows do you need more21:12
Noldorinjelmer: well i've not seen one person lamenting the lack of features so far :)21:12
Noldorinmore than enough for me21:12
Noldorinin fact21:12
Noldorinit's really nice21:12
Noldorinheh21:12
jelmerNoldorin: for e.g. contributing a simple fix for an existing git project, bzr-fastimport is suboptimal21:18
Noldorinyou just use git for that :P21:19
Noldorinha21:19
Noldorinno point using anything else21:19
jelmeryou first have to clone the git repository, then git fast-export it, then bzr-fastimport it, then make your change, then bzr-fast-export your change and git-fast-import it21:19
jelmerNoldorin: That's the use case dpush was designed for21:19
Noldorinyeah but dpush is horribly broken as we know :)21:19
Noldorinhttps://github.com/alexreg/git-bzr-ng21:19
Noldorinreally nice ^21:19
Noldorinsimplifies import/export tasks21:20
jelmerNoldorin: dpush doesn't handle a few corner cases well21:20
jelmerNoldorin: but the same is true for bzr-fast-import21:20
Noldorinmate, stop calling them corner cases21:20
Noldorinw21:20
Noldorine21:20
Noldorinwe know that's not true21:20
Noldorinbzr-git fails on the bzr branch itself, and two of my branches21:20
Noldoringit-bzr-ng and fastimport works on all of them :)21:20
Noldorinit has to be said21:20
jelmerNoldorin: if you're just pushing one or two changes (and don't care about renames, since you're contributing to a git repo anyway) that's unlikely to matter21:21
Noldorinyeah but git-bzr-ng is just as easy21:21
Noldorinto be fair21:21
jelmerNoldorin: isn't git-bzr-ng a git plugin?21:26
Noldorinyep21:27
Noldorinso?21:27
jelmerNoldorin: if you're happy using the git executable and are storing your main code in git, what are you using bzr for?21:28
Noldorinhaha21:28
Noldorinjelmer: no-one said anything about storing main code in git21:28
jelmerNoldorin: that doesn't make much sense to me, but okay21:32
Noldorinexplain then21:32
Noldorinif you're confused21:32
jelmerNoldorin: that's okay, that bit doesn't really interest me; it's more the different approaches for exporting to git21:33
Noldorinjelmer: this *is* an approach for exporting to git :P21:33
jelmerNoldorin: it's the different approaches for exporting to git I'm interested in, not really on why your main branch is in the 2a format rather than in the git format21:38
Noldorinthis is a different approach21:38
Noldorinso i don't get you21:39
jelmerNoldorin: anyway, I seem to be the de-facto maintainer for both bzr-fastimport and bzr-git; I just think bzr-git is the cleaner approach21:39
Noldorinyeah i saw that ;)21:39
Noldorinso it's not like i'm slagging off one over the other21:39
Noldorinfastimport just seems rather more reliable at htem oment21:39
Noldorinat the moment*21:39
Noldorinand indeed versatile21:39
maxbThere are some rather perplexing test failures for bzr-builddeb in oneiric and natty :-/21:41
jelmermaxb: can you give an example?21:41
maxbIOError: [Errno 21] Is a directory: u'/tmp/testbzr-LexZzZ.tmp/bzrlib.plugins.builddeb.tests.test_util.TreeContainsUpstreamSourceTests.test_empty/work/.bzr/checkout/dirstate'21:41
maxb!21:41
maxband some sort of treetransform fallout in natty:21:42
maxbOSError: [Errno 20] Not a directory: '/tmp/testbzr-kPYNe4.tmp/bzrlib.plugins.builddeb.tests.test_revspec.TestRevisionSpec_package.test_from_string_package/work/tree2/.bzr/checkout/limbo/new-1'21:42
jelmerthat second one looks vaguely familiar21:43
jelmermaxb: I mean, the "Not a directory" bit looks vaguely familiar21:43
jelmerI'm afraid I don't remember specifics though :-/.21:44
* maxb reviews some previous versions21:44
maxbThe same test failed in a different way in the previous bzr-builddeb release on natty21:46
maxband before that, a string of failures for other reasons21:46
jelmerI think it had something to do with tarballs or zip files21:48
maxbFrom the oneiric failure:21:48
maxb101.551  trying to create missing lock '/tmp/testbzr-LexZzZ.tmp/bzrlib.plugins.builddeb.tests.test_util.TreeContainsUpstreamSourceTests.test_empty/work/.bzr/checkout/dirstate'21:48
maxberr... what? Something's amusingly confused there !21:48
jelmerurgh, yep :)21:57
maxbHm, looks like the dulwich testsuite has hung on maverick & lucid too21:59
maxbThere was I thinking this series of PPA updates was going OK21:59
jelmermaxb: that's probably one of the compat tests22:03
jelmermaxb: you might need a minimal version of git-core in order to be able to run it22:03
Noldorinjelmer: hmm is colocated stuff still in a separate plugin?22:32
Noldorinon the latest version i have on mac, i don't see it there22:32
maxbThe bzr-builddeb failures are feeling like tests that don't properly support concurrency22:41
maxbAlso these builds would be a lot faster if they didn't run the tests 4 times22:45
Noldorinlifeless: hye22:48
Noldorinlifeless: just saw http://doc.bazaar.canonical.com/developers/colocated-branches.html…22:49
Noldorinlifeless: you may want to update that re: the branch separator… even though i prefer the simple ',' ;)22:49
jelmerNoldorin: it's in core, but only 2.623:12
Noldorinah right23:12
Noldorinjelmer: was just curious why it wasn't included as a default plugin in 2.5...23:13
jelmerNoldorin: bzr-colo is very different from the colocated branch support in the core23:13
Noldorinoh i see23:13
Noldorinhow so, briefly?23:13
jelmerNoldorin: bzr-colo was designed to not depend on any changes to the core23:13
jelmerbut it's a bit of a hack, it adds an entire new set of commands to deal with colocated branches and has more overhead than the native colocated branches23:14
Noldorinright23:15
Noldorinjelmer: not to worry. i presume colo branches will be adequately documented when 2.6 comes out :)23:15
jelmerNoldorin: note that the ocument you just linked to is the spec for colocated branches, not the user documentation23:16
Noldorinindeed23:16
Noldorini knew that was just a whitepaper heh23:17
Noldorinjelmer: only thing i'm curious about… did you decide on the "," seperator (as in the spec) or ",branch=" as before?23:17
jelmerNoldorin: at the moment it's ",branch=foo"23:18
Noldorinoh23:18
Noldorinokay23:18
jelmerNoldorin: we decided to use that as it's the safer option23:18
Noldorinthat's unfortunate… i'm sure you have a reason though23:18
jelmerit's always possible to add support for ",foo" in addition to ",branch=foo" later23:18
Noldorinindeed23:18
Noldorinthat's true23:18
jelmerNoldorin: there are other situations in which you might want to use ,23:18
jelmerNoldorin: e.g. "file:///tmp/bzr,revision=4323:19
jelmerNoldorin: e.g. "file:///tmp/bzr,revision=43"23:19
Noldorinhmm, yeah23:19
Noldorindoes that already work?23:19
Noldorini would suggest file:///temp/bzr?r=43 is better23:20
Noldorinbut fair enough23:20
jelmerNoldorin: what happens when things are nested?23:20
Noldorinlike?23:21
jelmerNoldorin: if you hasve a bzr tree inside another bzr tree23:21
Noldorinha23:21
jelmeryou might want to specify the revision of the inner tree23:21
Noldorinthen you've fucked up? ;)23:21
jelmerthis is the reason we picked , - that's what's specified by std66 for per-path-element parameters23:21
jelmerNoldorin: also, ? is interpreted by shells and often can't be used without quoting i23:21
Noldorinfair enough23:22
Noldorinjelmer: so maybe simply ,b= and ,r= would be better23:22
Noldorinat least23:22
Noldorina23:22
Noldorins23:22
Noldorin as aliases*23:22
jelmerNoldorin: those aliases can always be added later23:22
Noldorinyes that's what i just said23:22
jelmerNoldorin: also, that's a fair bit less readable23:22
Noldorin:P23:22
Noldorinjelmer: ah, but is readability the key here? to me, definitely not23:23
Noldorinsuccinctness is preferable23:23
jelmerNoldorin: to occasional bzr users coming across these for the first time23:23
jelmerNoldorin: another issue is whether "r" is going to be short for revno or for revid23:23
Noldorinyes but weigh it up against the verboity23:23
Noldorinverbosity23:23
Noldorindefinitely worth considering23:23
Noldorinit shouldn't be dismissed quickly!23:24
jelmerNoldorin: let's worry about that when it actually becomes an issue - colocated branch support has to be finished first23:24
Noldorini'm not worried23:24
Noldorini'm just throwing ideas at you!23:24
Noldorinthought you might be grateful23:24
Noldorin*shrug*23:24
jelmerI'm not working on this stuff :)23:24
Noldorinjelmer: you replied to me :P23:24
Noldorinwas initially at lifeless23:25
Noldorinbut yeah23:25
jelmerNoldorin: you asked about my design considerations for colocated branches23:25
Noldorineither way is cool23:25
jelmeranyway, time for some sleep!23:25
jelmerg'night23:25
Noldorinyes, after lifeless was away23:25
Noldorinjelmer: you reply, you involve yourself! ;)23:25
Noldorinjust try and be more receptive23:26
Noldorinbureaucracy is the downfall of all entreprises23:26
Noldorinafter all23:26
jelmerNoldorin: what's bureucratic here?23:26
Noldorini'll let you figure that out for yourself mate23:26
* jelmer sighs23:27
Noldorinhah, go sleep on it or something23:27
Noldorindon't sigh23:27
Noldorin*think*23:27
Noldorinit's good.23:27
jelmerI don't care :)23:27
jelmer/me sleeps23:27
Noldoringood23:27
Noldorinthen don't ask23:27
Noldorinelse you'll give people conflicting views23:27
Noldorin*laughs*23:27
* maxb looks at distro-info for the first time23:30
* maxb boggles that anyone ever thought writing it in Haskell was a good idea23:31
Noldorinmaxb: writing anything practical in haskell is probably a bad idea23:45
stumblesHi, I'd like to copy a configuration file from a template whenever I make a new branch of a repository. Ideally the behaviour would be embedded in the repository itself, rather than my user configuration. Is this something I can do with a plugin?23:56

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