=== beuno_ is now known as beuno === 13WAAIDLL is now known as wallyworld === wallyworld is now known as Guest76072 [04:10] hi! [04:11] I just ran into https://bugs.launchpad.net/bzr/+bug/375013 with tarmac. [04:11] Ubuntu bug 375013 in Launchpad itself "Cannot commit directly to a stacked branch" [High,Fix released] [04:11] the solution is to unstack the branch I presume? [04:34] nigelb: newer bzr should work [04:36] lifeless: hrm, its a lucid server. let me hunt for a ppa. thanks. [09:01] morning all [10:09] hi all ! [10:12] * mgz pats vila on the back [10:24] vila: hello! [10:24] so, another thing about colo branches [10:24] is that I would like to have a better default push location [10:25] when I was using directory-per-branch, then I could easily have an appendpath policy for ~/src such that ~/src/some-project/fix-a-thing would have a push location of lp:~jml/some-project/fix-a-thing [10:25] jml: in locations.conf you mean ? [10:26] what I'd like is for my colo branch of file:///home/jml/src/some-project,branch=fix-a-thing to have a default push location of lp:~jml/some-project/fix-a-thing [10:26] vila: yes [10:26] if you'll excuse a multi-line paste... [10:26] [/home/jml/repos] [10:26] public_branch = lp:~jml [10:26] public_branch:policy = appendpath [10:26] push_location = lp:~jml [10:26] push_location:policy = appendpath [10:26] jml: did you look at {dirname} in recent bzr versions ? Hmm, not it will work for colos though [10:26] that's what I have in locations.conf and it works great for directory-per-branch [10:27] err, basename not dirname [10:27] vila: is that a substitution variable thing? [10:27] yes [10:27] I use it this way: [10:27] [/home/vila/src/udd] [10:27] mypush = lp:~vila/udd/{basename} [10:27] vila: perhaps all that's needed then for what I want is another substitution variable for branch name / nick? [10:28] and then use: bzr push `bzr config mypush` so that I can *still* set push_location in branch.conf [10:29] yeah, nick again... [10:30] brb [10:32] in theory (hand wawe), 'nickname' should be a branch option defaulting to branch.nick or something (evil is in the details) [10:38] I guess I don't really understand the theory [10:39] partly because I don't really understand what 'bzr branches' does. [10:39] jml: as of today, 'nickname' exists in branch.conf only when you use 'bzr nick' or bzr-loom [10:39] so it's not a registered config option and cannot be tweaked like 'basename' is [10:40] hmm, 'basename' is not a registered option either, but that irrelevant (bad faith hand waving) [10:40] vila: should I care about nickname being configurable? [10:40] I don't right now. [10:42] jml: the actual implementation *relies* on nickname being a config option [10:42] vila: oh. implementation. pshaw. :) [10:42] jml: i.e. if 'nickname' exists in branch.conf, that's what it's used ;) [10:42] s/it's/is/ [10:43] urgh, looking at branch._get_nick, it seems to be a bit more complex, but anyhow, the config option can be involved ;) [10:45] roughly (IIRC) the current implementation will query the master branch to get its nick, we've discussed using the config option as authoritative and update it when needed instead [10:45] jml: getting there will mean you'll be able to rely on it always being right and as such use it as {nickname} [10:46] vila: ok. I don't think I've ever noticed it being wrong though :) [10:46] jml: it depends on how you looked at it :) [10:46] bzr nick or bzr config [12:52] vila: I never use bzr config. [12:53] jml: give it a try ;) [12:53] is there a consensus on what's best practice: to merge, resolve conflicts, commit, fix tests, then commit; or to merge, resolve conflicts, fix tests and then commit only once for the whole thing [12:54] jml: I think it's still an open debate and a personal taste, I tend to do the later even if I would prefer the former [12:55] ideally bzr should provide a way to look at how what was done during the 'fix conflicts, tests and code' phase as sometimes there can be non-trivial changes there [12:55] vila: yeah. I just had quite a fiddly merge to do, and I sort of wished I had a diff of the things I was doing to fix the tests, so I could verify their correctness [12:56] vila: right. [12:56] jml: would something like git's staging are be useful for this? [12:56] s/are/area/ [12:57] mgedmin: I don't know enough about git to answer that [13:03] mgedmin: I don't see how the staging area would help in that regard [13:04] hm, 500 again from wiki.bazaar.canonical.com page [13:05] I guess I should bug #is or something? seems random rather than completely reproducable [13:05] well, the staging area separates the changes you've done (merging conflicts) from the changes you're still working on (fixing tests) [13:05] you can see either diff separately [13:06] s/see/review/ [13:06] mgedmin: at that point, but not once the commit has happened [13:06] right [13:06] hmm, or was that what jml was saying? [13:06] "I sort of wished I had a diff of the things I was doing to fix the tests, so I could verify their correctness" [13:07] yeah, that would be possible before commit, but not after [13:07] mgedmin: right, but while you're doiung those things or if you're reviewing somebody else's merge? [13:07] I guess the way I think of it is like a little nested line of revisions [13:07] which I guess is what a branch is [13:07] I guess something like the staging area helps for the first but not the latter [13:45] jml: I think the following should be quite close: merge, fix conflicts, commit, fix tests, uncommit, commit [13:45] the uncommit is optional ;) === yofel_ is now known as yofel [13:47] vila: yeah, that would be quite close. [13:47] * mgedmin was wondering what happens if you uncommit when you have changes in your working tree [13:47] mgedmin: you get more or less changes depending on your changes ;) [13:48] but generally you get more changes and your branch tip is updated to the previous version [13:49] is it possible to undo an accidental uncommit? [13:51] mgedmin: yes, "bzr uncommit" will tell you how to do so [13:53] cool [13:54] huh [13:54] the name of the current branch in a colo is not 'bzr nick' [13:54] is there any command line for getting the name of the current branch? [13:55] jml: 'bzr branches' [13:55] jelmer: just the name of the current branch [13:55] no, there isn't anything that can tell you that at the moment [13:55] jelmer: ta [13:58] jml: in 2.5, the name of the current branch in a colo is the default for 'bzr nick'. [13:59] abentley: I'm using 2.6.0dev1, apparently [13:59] jml: I landed it last week, so it should hit dev the next time someone merges. [14:07] abentley: cool. [14:12] jelmer: I think I may have been unclear in bug #933178. Does it make sense now? [14:12] Launchpad bug 933178 in Bazaar "switch doesn't find sibling branches when colocated" [Undecided,New] https://launchpad.net/bugs/933178 [14:13] abentley: ah, I think I understand what you mean now [14:13] abentley: I think we agree on what the right behaviour should be [14:13] jelmer: great. [14:50] in a colo, how would I 'revert' a file to match the file in another branch? [14:51] (in dir-per-branch, I'd just copy the file) [15:06] jml: `bzr cat` it from the other branch? [15:06] LarstiQ: that's what I ended up doing. not ideal. [15:06] can you use -r branch: to refer to a colo branch? === Quintasan_ is now known as Quintasan === jordan__ is now known as jordan [18:01] is there a way to add custom prefixes to bzr? I'd like to be able to branch from lp like the following bzr branch mandel:ubuntu-sso-client/branch [18:02] mainly because I'd like to have a prefix per member of my team since some of them have very long nicknames.. [18:06] mandel, http://doc.bazaar.canonical.com/plugins/en/bookmarks-plugin.html might help you [18:14] mgedmin, thx! [21:04] hi barry? === lifeless_ is now known as lifeless