=== thumper is now known as thumper-afk === thumper-afk is now known as thumper === hloeung_ is now known as hloeung [06:16] hi. [06:17] I am on linux, using bzr 2.3->2.6 and have "beyond compare" diff tool. [06:17] when I'm using bcompare for bzr diff of a single file, everything is OK. [06:18] but, when I'm doing this for multitude of files, bzr is spawning each FILE in its own window. [06:18] This is even more annoying, b/c it does so sequentially: first file, then I need to close the next one, etc. [06:19] What I'd want to have is something I have with mercurial: diffing creates 2 tmp dirs, and spawns beyondcompare SMARTLY to compare folders. [06:19] This means I get to see a bigger picture before diving into the diffs of individual files. [06:20] Is there a hack to make bzr do this ? [06:22] (I'm reading ext-merge page to see whether it can do this now... seems it's MERGE oriented) [06:22] Could maybe script something up around export. [06:23] It is what I'm hoping to FIND, not script :) [06:24] I'm not aware of anything to be found like that (though I've never looked for any such thing, so that's not exactly definitive) [06:24] I am ... a bit surprised nobody is bothered by this. [06:26] Well, I'm pretty sure I could count on one hand the number of times I've seen beyondcompare mentioned ;) [06:26] And probably have enough left over to play a mean game of Hungry Hungry Hippos. [06:56] fullermd: not aware of hungry hungry hippos, but I'm still googling... [06:58] It's not particularly relevant to anything version-controlish :p [08:02] Having two branches with a lot of diffs, I would like to generate a patch with only some of them. Like running a qdiff, and being and being able to select changes I want and click "generate patch" :P [08:06] How can I do this ? I was thinking about some tricks like merging my second branch in the first on, and then revert all the things I don't want. But this sounds a little tricky.. [08:10] maybe you want filterdiff [08:14] bob2: How does that work ? [08:24] vila: An idea about how I can do this with Bazaar N [08:26] Merwin: well, depending on your needs, I can think of 2 ways: a manual one editing the patch file, an more automated one relying on setting up your tree so that it contains only the parts you want in the patch [08:26] Merwin: or may be just use 'bzr shelve' [08:27] Changes are already commited so shelves can't work [08:27] Merwin: i.e. shelve the changes you don't want in the diff [08:27] (Unless I missed something ?) [08:27] Merwin: uncommit in a temp branch ? [08:27] There's a lot of commits to uncommit :p [08:27] That would hardly work for scattered bits through the history... [08:27] Merwin: but I'm surprised by your use case, if it was committed why isn't it worth sharing ? [08:27] I'd just make a temp branch, merge, then start merging out the bits you don't want. [08:28] In fatch I have to brand : the main one and one for feature A. The problem is that a small feature B has been made in branch of feature A. Now, I ant to extract this features from this branch to put it in the main branch... [08:28] In fatc I have two branches* [08:29] Merwin: If I had to do it myself, I'm sure I won't be able to get it right manually at the first try so I'd use a dedicated tree for that to make sure I can refine the patch [08:29] So basically, I made a diff between main branch and branch A (containing feature A & B), but I just want to keep feature B, and generate a patch for it :p [08:30] Merwin: right, so *I* would make sure that feature B can me merged cleanly to trunk by.... first building theat featureB branch without relying on feature A and I'll get the diff from there [08:31] Yes, seems a good idea, but... how to build branch B... I can't cherrypick commits from branch A because it's badly commited (mixed features in commits...) [08:31] Hum, maybe create a branch B from main branch, these merge branch A in it, and before commiting merge, shelving changes I don't want ? [08:31] Merwin: so, like fullermd said, start a new branch with feature A & B and remove the unwanted bits [08:32] Merwin: yeah, something like that [08:32] okay thank you vila and fullermd [08:32] Merwin: and to avoid issues in the future, you probably want to not record that first merge with 'bzr revert --forget-merges' so you end up with a "stand-alone" patch [08:33] This won't revert changes, just remove the "pending merge" status ? [08:34] Merwin: failing to do so, once feature B is part of trunk, merging feature A will raise conflicts: "You did merge feature A in the past and removed bits, now you're merging feature A again. Do you want to add the removed bits again or leave them out ?" [08:34] Merwin: yes, it would forget which revisioins were merged, keeping only the changes as if you applied them manually [08:34] hum ok, interesting I didn't know it [08:34] Thank you [08:37] Merwin: this is better understood by toying around with the various workflows in temp branches and looking at the result with 'bzr qlog' === wedgwood_away is now known as wedgwood [15:41] hi all. when you do "bzr qlog ... " you see a diagram representing the relations btween all mentioned branches (the DAG actually). I always loved that feature. Is it documented somwhere in the website? I want to show it to a friends but I am too lazy to make the minimal example, take screenshots and all that. Is there any pretty pic on the web ? [15:42] * ggherdov also: why http://bazaar.canonical.com/ shows just file listing? wasn't it a nice homepage back in the days ? [15:42] * ggherdov looks like the redirection to http://bazaar.canonical.com/en/ is broken ... [15:43] it does indeed look like that's broken.. [15:44] yep [15:44] not good for the press ;-) [15:46] ok ok i am doing my own screenshots... :-) [16:03] err... can anybody try to connect to the link http://doc.bazaar-vcs.org/bzr.dev/en/user-guide/index.html#team-collaboration-distributed-style ? it takes 100% of my CPU and 95% of disk I/O... what the heck? I am trying on my ubuntu VM , firefox 20.0 and on my winXP workstation, same version of FFox. Whazat ? [16:04] does it work fine to you? [16:06] update, you dont need the fragment #team-collaboration-distributed-style in the url, it's just http://doc.bazaar-vcs.org/bzr.dev/en/user-guide/index.html that goes crazy. [16:07] redirect loop, really firefox should do something sane with that... [16:07] you want http://doc.bazaar.canonical.com/bzr.dev/en/user-guide/index.html [16:09] mgz: yep redirect loop. All firefox does is freezing my OS... good boy :-) thanks for the direct link [16:35] in the command "bzr qlog ... " , is there any requirement on 's ? like, do they need to share some history ? or can the be like random branches not even branched from the same origin ? like, can I do "bzr qlog lp:mysql-server lp:bzr" i.e. compare totally unrelated things? [16:36] * ggherdov_ is gonna find out very soon [16:37] they can actually be whatever I believe [16:37] it's a qlog only wonderment [16:37] mgz ah nice [18:16] mgz: are you still around? just to say that you were right! you can actually do "bzr qlog lp:bzr lp:mysql-server" look: http://imgur.com/rDyYLIA [18:17] I just put some gimp incantation to make it fit in a viewable size [18:18] * ggherdov gotta go [18:18] neat :) [19:26] I'm trying to setup Jenkins to commit code to branches. I have jenkins setup on a machine. Right now there are two branches. Each branch needs a different 'launchpad user' to do the comitting. I see how to use different email addresses, but I'm not finding any help with how to use two different LP users to commit code. Is this possible? [19:33] the easiest way would be to supply different BZR_HOME values [19:34] then they could have different launchpad_username values set in each ~/.bazaar/bazaar.conf [19:43] Hi all. Git folks have this page http://gitolite.com/1-basic-usage/tias.html that show a super simple way to simulate a real world workflow scenario. You do that and immediately can start playing and grasp what the system does in some situation that you might not understand very well. Is there a similar page for bzr? or, what are the correspondent commands to play the TIAS game in bzr (Try It And See) ? [20:02] bzr init-repo b ; bzr branch b c1 [20:03] why do I have "Not a branch: "b/.bzr/branch/": location is a repository." ? [20:04] ...because you created b as a repo, not a branch [20:05] compare to `bzr init b; bzr branch b c2` [20:05] msg ah. thank you. [20:12] msg by the way, just to be clear about my intent, I'd like to create a ... $thing (repository? branch?) called "b" to act as the "production", or "reference" line of development (like a central server let's say), and then to "fork" it into what I called "c1", which simulate one of the N developers that will contribute to it. The first names that came to my mind are "repository" for "b", and... "branch" for c1, .., c_n. Is my understanding wrong? [20:13] basically c1, ..., c_n use b to communicate btween them, via push/pull [20:14] ggherdov: see [20:14] msg ok [20:15] * ggherdov calling mgz msg bcse of his dyslexia :-) [20:28] mgz: I think I got it. "bzr init-repo" create a repository, that can potentially contain *many* branches. Thus, "bzr branch" on a repo doesn't make any sense (I guess one can specify a branch inside that repo somehow). OTOH, "bzr init" create a single-repo-single-branch, which is a sort of "limited" repo since it's bound to have only one branch inside. Now, "bzr branch" on it makes sense since bazaar understand that you wanna fork the one and only [20:28] branch in that repo. Am I close? [20:29] right. [20:29] thanks [20:33] mgz: is there a way to have each launchpad_username use the correct ssh key? [20:35] if both keys are available, they should just use the one that works, but I imagine you can set up something sane with the ~/.ssh/config [20:40] mgz: nooo [20:41] username exchange happens before key probing [20:41] mgz: (or I may misunderstand what you meant there) [20:41] the username comes from the bazaar config [20:42] we don't expose a way to pick which key to use, as far as I know [20:42] but ssh will try all keys available locally, so if one launchpad account has one key, and the other has a different one, provided both are loaded it should be fine, no? [20:43] I'm getting pubkey errors [20:44] can you do `ssh -vv USER_A@bazaar.canonica [20:44] two keys exist. one key is on one LP account, the other is on another LP account [20:44] *@bazaar.launchpad.net [20:44] and the same for USER_B? [20:45] mgz: yes, listing all the keyrings should work fine [20:45] or you can just use the same key on both LP accounts. [20:45] they bot gave http://paste.ubuntu.com/5699705 [20:46] s/bot/both [20:46] cjohnston: I brainoed, you want the correction, bazaar.launchpad.net [20:48] http://paste.ubuntu.com/5699711/ http://paste.ubuntu.com/5699712/ [20:49] the first works, the second doesn't have the key [20:49] correct [20:49] what's the name of the second private key locally? [20:49] qajenkinsbot [20:50] you just want IdentityFile that then in .ssh/config [20:50] you can have multiple IdentityFile entries per host [20:51] ie: [20:51] Host bazaar.launchpad.net [20:51] IdentityFile ~/.ssh/id_rsa [20:51] IdentityFile ~/.ssh/qajenkinsbot [20:51] or similar [20:52] ok.. ty === wedgwood is now known as wedgwood_away