[00:00] luke-jr: Maybe scmproj might be a solution. http://launchpad.net/bzr-scmproj Though It's not mature yet either. [00:00] You can choose between the lesser of 2 evils.... [00:01] scmproj *is* being actively developed. [00:05] garyvdm: interesting, I hadn't seen that one yet. There's also https://launchpad.net/config-manager. [00:41] is it possible to create a local change that cannot be pushed? === davidstrauss_ is now known as davidstrauss === Mario__ is now known as pygi [05:14] Does anybody with experience in bzr-svn happen to be around? I'm getting "Unable to open an ra_local session" errors when I specify a single file for "bzr diff", but not when I do a directory. (bzr 1.10, bzr-svn 0.5rc1) Same thing for commit, as well. [05:15] headlessagnew, where does the error appear? [05:15] headlessagnew, what platform? [05:17] jelmer: I'm on OSX, running SVN 1.5. The error shows up inside bzrlib/plugins/svn/subvertpy/ra.py at line 43, in RemoteAccess . [05:18] headlessagnew, how did you install bzr-svn? [05:18] running from ~/.bazaar/plugins ? [05:18] jelmer: Systemwide, I'm afraid, but by hand and not with easy_install [05:18] headlessagnew, I suspect you ended up with a broken install then [05:19] there were some fixes wrt the installer in rc1 [05:19] s/fixes/issues/ [05:19] they'll be fixed in the next rc [05:19] jelmer: Oh, I see. Should I pull a new version from launchpad and try that? [05:20] headlessagnew, you can try using the 0.5 branch of bzr-svn [05:20] are you sure you want 0.5 though, not 0.4 ? [05:21] jelmer: I'll confess I don't have any stake in 0.5 over 0.4; I'll drop back to 0.4.16. [05:22] jelmer: much obliged, thanks! [05:22] headlessagnew, np [08:09] hi, trying to install bzr from https://launchpad.net/~bzr/+archive on hardy i get [08:09] bzr: Depends: python-central (>= 0.6.7) but 0.6.5ubuntu1 is to be installed [08:37] Hi - I'm trying to extend some tests in bzr-upload. [08:38] It's a TestCaseWithTransport. [08:39] I'm trying to push to the transport. When I do this, it just freezes. [08:40] This is my code: http://bazaar.launchpad.net/~garyvdm/bzr-upload/nowt/revision/55 [08:42] I stepped through the code to see where it is freezing. It freezes on: [08:42] dir_to = bzrdir.BzrDir.open_from_transport(to_transport) [08:42] line 47 bzrlib/push.py === snaggen1 is now known as Snaggen [10:15] I found my problem with the tests. It's prompting for a password. You cant see the prompt, but it waits for input. [10:15] Does any one know how I should fix this? [10:19] I'm investigating the "bzr commit --fixes= " feature to see it I can get that to change the status of a ticket in trac. Is there any server hook that can do this or any other solution? I think I saw something about this a long time ago, but I cant find anything now... [10:50] moin [10:50] jelmer: ping? === asac_ is now known as asac [12:11] Any way that I can pick which changes I want to go into a send patch? === asac_ is now known as asac [13:49] If I have a bzr repo (repo 'a') which is branched from another repo (repo 'b'), and I commit a changeset to repo 'a', and thereafter changes are made to repo 'b', how do I pull those changesets from 'b' into 'a'? I tried 'bzr merge ', but I think I'm not understanding something... [13:51] first you should probably change the terminology :) [13:51] (to match bzr) [13:51] after doing that, if I do 'bzr st', I get "modified: .... [list of various files]", and "pending merges: ...". [13:52] you need to run bzr commit after bzr merge [13:52] luks: and then all the 'revisions' from repo 'b' will appear in 'a'? [13:53] yes [13:53] well, maybe not in the way you expect them [13:53] they will not be linear [13:54] ok - and how would I get them linear? [13:54] the simple answer is that you can't [13:54] why would you want that? [13:55] And..., if I make another commit to 'a', and want to 'push' both of the changes from a back into b, how does that work? [13:56] after push from 'a' to 'b' (after bzr merge & bzr commit), 'b' will be identical to 'a' [13:56] if you make another commit to a, and it won't be merged to b, bzr will refuse to push [13:57] it will fail with an error about diverged branches [13:58] hmm. it would be cool if you could do something like git or hg rebase - is there something like that for bzr? [13:58] yes, there is bzr-rebase [13:58] but that doesn't "pull" the revisions [13:58] it creates new revisions that look like the original ones [13:58] (it is the same in git and hg) [13:59] then, I guess it would be possible to just keep 'a's changes on top of the revisions pulled from 'b'. [13:59] (and keep them linear) [13:59] yes, you can keep the changes, you can't keep the revisions [14:00] you mean revision numbers, or the commit message, etc? [14:00] mainly the revision IDs, which is something you don't normally see, but it tells bzr whether two revisions are identical [14:01] the patch, commit message, author, etc. are all kept when using bzr rebase [14:01] ok, that's good ;) [14:05] What happens when if you have a repo on launchpad, and you rebase your local repository, and try to push the changes back to launchpad -- I presume this won't work, right? [14:05] right, you can push --overwrite though [14:05] and what does that do? [14:05] running rebase on a public branch doesn't make much sense, though [14:05] well, it will overwrite the branch :) [14:06] meaning that it will remove the old revisions and add the new ones from the local branch [14:06] I see -- so everything gets overwritten, or only the 'non-common' parts? (just wondering about how much time it would take, since bzr has been pretty slow for me in the past -- pushing taking like 20 minutes or so). [14:07] (an initial push, that is) [14:07] it will push only the new revisions [14:07] the revisions that are identical in both branches are keps as they are [14:11] luks: ok, good. Thanks for your help. [14:14] one other thing -- how do I chop off (delete) revisions from a repo (starting at a given revno)? [14:15] bzr uncommit, if I understand the question correctly [14:16] that is, if you want to remove revisions from X to the head of the branch [14:16] if you want to remove revisions from 0 to X, there is no easy way [14:18] yes, uncommit should work then -- but I don't see it with the version of bzr I have installed. [14:18] (1.6.1) [14:19] bzr help uncommit should work [14:19] bzr help commands to see all available commands [14:19] never mind. I think I just typed something wrong. [14:29] luks: hmm - rebase seems not to work so well -- I tried 'bzr rebase ", and get a traceback: AttributeError: 'KnitPackRepository' object has no attribute 'revision_parents'. Maybe I'll have to wait on using rebase... [14:31] hm, maybe you are not using compatible versions of bzr and bzr-rebase [14:32] (both installed from ubuntu repositories...) oh well. [14:36] I don't know then, sorry [14:45] lifeless: fwiw, is bzr-search's new index format supposed to be 15% larger? [15:55] hey james_w, I had a thought about the way the config system is represented internally. Might it be better to make it like a dictionary in which the config object is loaded into memory at runtime and options are queried from the control logic, rather than as properties of the config file? [15:57] ronny, pong [15:57] rexbron: yeah, I'm not keen on the config object, I wrote it ages ago, but haven't had the will to change it [15:59] james_w: that probably would address bug 309335 [15:59] Launchpad bug 309335 in bzr-builddeb "Should either accept options with no section in config files, or warn about them" [Medium,Triaged] https://launchpad.net/bugs/309335 [16:02] rexbron: yeah [16:06] james_w: do you mean the bzr-bd config system or the one in bzrlib? [16:08] rexbron: the bzr-bd Config object is the one I don't like [16:09] rexbron: the bzrlib config system, built on configobj, is the one that we will need to investigate to see if we can query for section headers case-insensitively [16:10] james_w: looking quickly, there is bzrlib.utils.configobject that impliments a section based dictionary object [16:11] bzrlib.util.configobj isn't something unique to bzr; it's a packaged copy of ConfigObj. [16:26] jelmer: ping [16:36] ronny, pong [16:38] jelmer: did you write bindings for the svn workdir stuff, too? [16:38] ronny, yep [16:38] jelmer: i didnt yet take a look, but i want to get rid of subprocess.call for anyvc's svn stuff [16:38] ronny, this is all part of subvertpy (not tied to bzr-svn) [16:39] great [16:39] ronny, http://launchpad.net/subvertpy [16:39] the branch is in bzr at the moment, but I'm happy to migrate it to svn if that makes people more comfortable [16:39] oh dammit [16:39] gpl3 is an issue for me [16:40] i need gpl2 support [16:40] Ok [16:40] I think that can be fixed :-) [16:40] Any particular reason you need GPLv2 ? [16:41] hg is gpl2only [16:45] jelmer: well, since bzr got better im more comfortable with bzr than with svn [16:46] ronny: Yeah, looks like relicensing should be possible since I'm the only copyright holder. [16:46] jelmer: "gpl2 or later" would be fine [16:47] Yeah, that's what I would go with [16:47] its unlikely that any of the hg/bzr libs will ever be lgpl, so there aint need to be lax [16:49] jelmer: btw, im wondering, would it be possible to mimic a svn server using a bzr repository + the common conventions for branches + painfull magic [16:49] ronny, yes, we've done some work in that direction [16:49] sounds like much pain tho [16:50] bzr-svn includes a "bzr svn-serve" command that provides a subversion server that accesses a bzr branch [16:50] it only does "/trunk" at the moment though, not any of the merged revisions yet [16:50] hmm [16:51] im kinda on war with bzr-svn, for some reason every time i installed it somethign in bzr broke [16:51] and it's quality doesn't reach beyond proof-of-concept atm [16:52] most likely due to version missmatches === fawek_ is now known as fawek [16:53] ronny, yeah, version mismatches are soon a problem because bzr-svn hooks into really a lot of the internal bzr APIs [17:02] jelmer: building subvertpy complains about a missing apr-config, what package ships that? [17:03] ronny: Install libsvn-dev or whatever. It should pull in everything else. [17:03] ronny, libapr1-dev IIRC [17:05] libsvn-dev pulls in all dependencies afaik [17:05] works fine now [17:05] thanks [17:06] hmm [17:06] jelmer: btw, could the tests be moved out of the main package? [17:07] and how fast could one get subvertpy packages into the distros [17:10] I'm planning on uploading to debian/ubuntu before the next version of bzr-svn [17:10] since it needs it [17:10] Any reason why you would want to see the tests moved out of the main package ? [17:11] just seems more clean to have them out [17:11] also makes comparing stuff more simple [17:12] hmmok [17:13] I don't have any objections to doing that per se, but need to make sure that certain utility classes/functions in tests/ are still installed, since they are used by bzr-svn for its tests [17:13] oh, if its a dep, then you might either leave them just in, or make a util module === fawek_ is now known as fawek [17:16] jelmer: is there any doc for subvertpy [17:17] ronny, not much at the moment. Some of the basie guys mentioned they were working on docs but didn't hear much from them recently. [17:17] ronny, the tests should help to some degree (not sure how much tests there are for the working copy stuff though) [17:19] there is one for add [17:19] not much else [17:20] I'll see if I can add some more + fix the licensing after dinner [17:20] back later [17:20] k, thanks === asac_ is now known as asac [19:43] anyone can tell me where that proposal for after-commit metadata is? === _thumper_ is now known as thumper [20:37] Hello everybody. [20:37] I've a quick question :) [20:37] I created an account+ project on launchpad === Mario__ is now known as pygi [20:38] My source code is also hosted on launchpad and I want to learn how to delete a branch through my command line ? [20:39] I tried: bzr remove-tree lp:myuser/myProject/branch_name [20:39] but it doesn't work [20:40] you can [20:40] er [20:41] you can't delete a branch using bzr [20:41] ah [20:41] luks so I always should do it through the lp website ? [20:41] yes [20:41] ok no problem. Thank you luks ;) [20:42] moin [20:42] Peng_: its not surprising that it is [20:43] Btw, bzr rox ... it's so f***** cool ! [20:43] I just left googlecode ... no longer wanna use svn :/ [20:43] * Goundy is going back to work [20:43] see you [20:53] lifeless: OK. [20:58] hi everyone! [21:00] Does someone know a way to export a changeset (every modified file of a changeset) into a file tree? [21:00] for example, if in the last commit I changed file "A/b/c" and added file "d" [21:01] I would do something like "bzr export -r 23..24 lastCommit/" [21:02] and the file "lastCommit/A/b/c" and "lastCommit/d" are created [21:08] so, maybe the "-r X..Y" and "-c X" options should be feature requests for the "bzr export" function :) [21:09] Mmm. Only if export did something totally different from what it does... [21:09] Enisseo: what would the content of the files be? [21:09] So, export complete copies of only the files that changed, or create patches or something? [21:09] diffs, or the files at the last revision? [21:09] the files [21:09] not diff [21:09] you could write a plugin to do that without too much trouble [21:09] the idea is to zip and distribute the zip file with the files [21:10] You could also make the plugin zip 'em. [21:10] for an user to unzip it in the right directory and overwrite existing files [21:10] How would that deal with moved or deleted files? [21:10] Peng_: yes, indeed, as the export function does [21:11] fullermd: moved or deleted files would not be managed [21:12] james_w: yes, I will if no one else does know an existing plugin [21:13] You could hack up a prototype with just a little sh scripting. Grab the output of stat between those revs, then cat the files into a tree. [21:14] Lot faster to write it directly on bzrlib of course, but... [21:14] fullermd: I use Windows ;) [21:14] makes sense to be part of export to me [21:17] To me too, since I tried "bzr export -r X..Y" directly without reading the doc [21:17] Where is iter_tree documented? [21:17] erm, iter_changes, rather [21:19] * nDuff finds the relevant docs in InterTree.iter_changes [21:29] Anyone mind lookin at a little mistake command i did, and tell me where bzr pushed it to? I accidentally pushed a branch to "ssh".. ? [21:29] http://dpaste.com/105372/ [21:30] well, you gave the url to push to as 'ssh' [21:30] so it pushed to 'ssh [21:31] well yea.. but where the hell is ssh? See i've had scenarios like this where i typo the push location, but it pushes anyway.. i wonder if its going somewhere weird on my comp, or my target comp, etc.. ? [21:31] leefmc: ls -l ~/projects/uioli/ssh [21:31] leefmc: its ./ssh [21:31] ah hah, ty [21:32] Now, while we are on the subject of pushing, is there any shorthand for sshing to home? I've tried lee@addy/~/repo but that didnt seem to work, i had to push to ssh lee@addy/home/lee/repo, thoughts? [21:35] mwhudson, hi [21:35] leefmc: homedir isn't support on bzr+ssh yet, I believe spiv is working on that atm [21:35] how's it going? [21:36] lifeless: K ty, so the way i was forced to do is the best/only way then? [21:36] at the moment [21:36] hey beuno [21:36] hey hey lifeless [21:36] ty [21:36] lifeless, how was your break? [21:37] excellent [21:37] cool, happy to see everyone back :) [21:38] mwhudson, I want to have a skype chat with you today about LH, if you're up to it [21:49] beuno: hello [21:57] Question: I'm having trouble adding one repo into another, anyone know why this is? Eg, i have a projects folder which is a repo. This repo holds branches of my projects. One of these projects has many different aspects (design/programming/etc), so the programming side of it has a bzr repo and branches of its own. Its this programming repo i cannot seem to add to the project branch [21:57] http://dpaste.com/105382/ [21:57] Thoughts? I've done this before with no trouble.. but im not sure what was different [21:58] if it's versioned separately you can't add it to the enclosing repository [21:58] unless [21:58] you're using the experimental subtrees feature [21:59] or you might be better using the scmproj plugin [21:59] well i've got almost this identical setup in another project branch.. i don't get why its working there but not here [22:00] hell, i was told this was a fine solution a while back in this channel heh. [22:00] This allows me to have entire projects be tracked/shared between all my computers, and have specific repos just focused on programming, allowing me to push to real repos and whatnot [22:01] I think we'd need more info on your working environment, I strongly suspect cross-wires of some sort [22:01] entirely possible [22:01] Im switching computers though, so i'll brb [22:02] Though.. i cant actually push my project like i would normally heh, so i gatta pindrive it :o [22:03] brb [22:09] lifeless: Ok, so you mentioned cross wired.. Howso? Perhaps i did something wrong on this latest repo [22:09] lifeless: Because i do have a different project branch, that has a repo inside of it which is use to track the actual programming. It works fine, no problems sharing, and no problems pushing/pulling [22:09] well bzr allows you to nest repostories, but it doesn't allow you to version a branch A as a element of branch B at the moment [22:11] so you *can* have A - a branch and A/B another branch, but to push them around you need to seperately push A and B [22:11] lifeless: So you mean nesting branches, like repo/branchA/branchB vs nesting repoA/branchA/repoB/branchB ? [22:11] or use multipush or something like that [22:11] because atm, i have repoA/branchA/repoB/branchB working great, just not on my latest project.. and im confused as to why [22:12] Leefmc: by working great, what do you mean ? :> [22:14] lifeless: In repoB/branchB i am able to make modifications to the code/programming, commit them, just like i want. Then, if i need to share the entire project which has my files/etc, i can add/commit changes to branchA and commit them, then push them to my local server to be pulled by a different machine of mine [22:14] Leefmc: can you run 'bzr info -v' in branchA ? [22:14] pastebin the result [22:16] An example of what i would need this for is i use an ide, and the ide has a project file for the ide. Now this file has my personal settings and junk in it, stuff i wouldn't want to submit to the public if i pushed the code to launchpad. So i actually track that aswell, in what i call a "project branch". This way i can commit&push the ide files and everything else that are just "mine" to my local server, for sharing with all my computers, a [22:16] nd i still can have a repo inside the project for actual programming. Ie, stuff i would push to launchpad [22:16] k one sec [22:17] lifeless: I forgot, it didnt push when i switched computers (hence the problem), though on my other project im not seeing the project. Perhaps it ghosted me.. and i thought it was working? [22:18] lifeless: So your saying this workflow is not possible in bazaar? [22:18] Leefmc: bzr won't manage the link between the branches for you today; it is planned but not done yet [22:18] I've actually grown quite accustom to not having to thumbdrive my working files around computers.. [22:19] gotcha [22:19] Leefmc: you can certainly have the two branches laid out on disk in that manner [22:19] you just have to run bzr push in each branch yourself [22:20] lifeless: Not sure if its worth it, nesting branches and pushing to multiple places just to change computers seems odd [22:21] Leefmc, I'm just curious as opposed to likely to be helpful -- but what's the purpose of having repoB, as opposed to putting branchB in repoA? [22:21] Ie, i'd be pushing incomplete code to launchpad if i wanted to change comps, or i'd have to manage multiple places to push to (incompelete/complete/etc) [22:22] nDuff: I think its so that personal stuff doesn't pollute the main tree [22:23] nDuff: e.g. in the example above IDE settings in a containing branch, code in a subordinate one [22:23] nDuff: Well for one, im a bit of a nut with clean directory structures. In this case, my project repo is strictly projects, which houses all my personal settings/notes/files for an actual project. Anything from reference images to example code to whatever. Inside that project is where i actually do work, so that may have a repo of its own where the programming is actually done. Its a matter of having strict places where things belong [22:24] I've always had this project repository idea, but i've never revisioned it. Revisioning it offered insane ease of use when sharing between multiple computers, not to mention nifty bonuses (.. reivisioning) for my project itself. [22:25] I probably "could" have a master repository, but then i'd be maintaining multiple sets of commits/pushes for the sake of sharing between computers and whatnot. [22:26] Ie, if i have some code that is not commit-able, but i want to switch to a different computer, i'd rather not commit&push it to launchpad. So then i start managing two places to push this single branch, along with having poor commit reasons, etc. [22:27] Its not impossible/implausible by any means, i just prefer a certain workflow and project organization [22:36] the bad thing is now i've become addicted to the ease of pushing/pulling when i need to share my projects between my computers.. oh how will i get out of this now :o [22:37] Im trying to figure out of git can handle this now [22:46] Leefmc: it has submodules, but AIUI you have to push each seperately too [23:04] lifeless: I think i am just going to go with something a bit more advanced like some type of syncing software. [23:06] Leefmc: Unison is pretty nice [23:06] Howdy, Does anybody know if there is a way to install Bazaar on Windows without admin rights? [23:09] the_Lorax, Bazaar can be run directly out of a source directory if you're desperate, though you need to have a Python interpreter handy. [23:09] yeah, don't have python installed at work either :( [23:10] Managed to get an older version of bzr installed into a personal directory & working. [23:10] Wonder if I can just install it on a different PC and then copy over the program directory? [23:16] morning [23:16] the_Lorax: I'm not sure, but if there isn't an option please file a bug askinf for it :) [23:17] the_Lorax: because I can see that being generally useful [23:17] hi igc [23:17] igc: morning. [23:18] hi lifeless, spiv! [23:21] Lifeless: I guess I'll try and if I can't get it working I'll file a bug and see if anything happens. Thanks [23:22] the_Lorax: the shell extensions may be tricky [23:22] lifeless: don't particularly care about the shell extensions - command line is just fine for me. [23:30] lifeless: You know that projects repository i had earlier? deleting the .bzr in that will not harm any sub-repositories/files correct? [23:30] Also, unison seems perfect, ty [23:30] Leefmc: if there are branches uses the repository you will break them [23:30] you can check by running 'bzr info' in the sub things and see what repository they report using [23:31] but by "break them" you mean the bzr commands will not work correct? No harm to the actual files [23:31] I mean break [23:31] If they contained things have their own repository they will not be affected at all [23:32] if they are using the repository you want to delete, clearly they will stop working when you remove the repository [23:32] yea [23:32] I don't know what files in specific you mean [23:32] if you mean 'are my checked out files on disk going to go away when I remove .bzr' -> no [23:32] my bad :) [23:32] yea, more like that [23:33] if you mean 'are the files that bzr uses independent of a containing repository' -> depends on if they are using it [23:33] if i see files, deleting .bzr wont touch it. :) [23:33] right, bzr isn't hooked into file system at that level ;P [23:33] haha [23:33] good thing i got backups [23:34] I forgot i don't have any working trees on my server, where i just deleted the .bzr .. :D [23:34] :> [23:35] This is nice though, assuming unison can work the way i want it (pretty sure) i can continue to use bzr. [23:35] Which is good, i've come to like it [23:35] Though eventually i'll be moving back to git im sure. Once google adopts it [23:35] they're working on supporting a dvcs now, and we all assume its git [23:37] what about google code do you like most ? [23:41] Well for one, launchpad has annoyed me with its speed. It has been horrendously slow so often that i wanted to peel my eyes out heh. Thats mostly it for me. A guy who leads another project i work on is very adamant about how launchpad is confusing to clients though.. i don't personally see it that much, but google code is much cleaner/simpler when thinking from the client side [23:42] Personally, i found launchpads branch system a bit confusing, and never delved that deeply into it, but from what i gather of it i actually like it. (Despite what i lack in understanding) [23:43] Thats something i dont think google code even offers, where as many sites are. Forking in github, etc [23:43] plus Google Code's web based code viewer is much cleaner/friendlier than i found logger to be. I found logger to be a bit ugly, too spaced out, etc. [23:43] Not that easy to navigate around a code base [23:44] But really, my main issue is the speed heh [23:44] speed being, the time it takes for a page to load. [23:45] Leefmc: what in particular did you find confusing about the branch system? [23:47] jml: That was just very early on, but i dont recall the specifics. It just took some getting used to, how a project is actually owned by a user, and not owned by the project, etc. [23:48] Leefmc: yeah, I got tripped up by that early on. [23:48] And if were on the topic of making constructive criticisms here, does launchpad offer a wiki? IIRC, it does not, and for all it does offer, i was honestly very surprised it doesn't. It should, in my opinion [23:48] Leefmc: it doesn't, and pretty much everyone agrees it should. :) [23:49] It seems to integrate so much, but a very common feature, a wiki, is not included [23:49] jml: I actually found it simply odd that it didnt. I saw so much, so i kept assuming i was missing it :o