[00:00] awilkins: any chance you can file a bug with that patch attached before you go to sleep? [00:04] jelmer: Hmm, this is odd ; bzr update says tree is up to date, but there's a huge diff [00:04] smore like it - diff -r 877 [00:04] awilkins: what do you expect "bzr update" to do? [00:05] it only makes sense if you have a bound branch [00:05] I was expecting it to update my working tree to the latest revision in the repository [00:05] it will only do that if you used "bzr co" [00:05] if you used "bzr branch" to create the branch you need to use "bzr pull" [00:06] No revisions to pull , my tree is at 877 but my repos is at 891 ; hence a huge diff unless I specify 877 [00:07] awilkins: A repository doesn't have a tip, not sure I'm following... [00:08] The source in my working tree (plugin folder) is at 877 (release 0.4.7) [00:08] But the repo has revisions up to 891 (HEAD) [00:08] If I diff the tree, it includes reverse diffs all the way back to 877 [00:09] Launchpad is fubar [00:09] Where do you diff against? [00:10] Also, what do you mean by repo? [00:10] The revisions in the .bzr folder (it's a standalone branch) [00:11] bzr log -r -1.. shows revision 891 [00:11] But the source is at 877 [00:11] "bzr revno" gives 891 ? [00:11] ItYes [00:12] The source being at 877 was intentional, I wanted the release tag [00:12] Ah, you ran "bzr update -r877" or something? [00:12] or perhaps "bzr revert -r877" ? [00:12] bzr pull -r 877 I think [00:12] Maybe revert [00:12] revert would do that [00:13] if you can send in a diff against 877 that would help too [00:19] jelmer: sorry, downtime, ignored my battery warning for too long. [00:19] Back up again now [00:19] welcome back :-) [00:28] #188233 [00:29] thanks [00:29] * jelmer waits for ubotu to announce [00:31] Grr, I've reverted up to 891 and now gnu patch won't apply [00:35] New bug: #188233 in bzr-svn "PATCH : fix some win32 testing issues" [Undecided,New] https://launchpad.net/bugs/188233 [00:38] Hooray for ~1~ files [00:47] Hm, if I make several revisions to a branch, how can I merge those changes to a parent branch without losing the original commit messages? [00:47] SVK has "--verbatim --incremental" [00:48] rolly: Just use "bzr merge" [00:48] bzr log will then show the merged revisions indented [00:48] I can't, because there is a revision that must not be propagated [00:48] i have to specify a range [00:48] rolly: that would be a cherrypick [00:49] rolly: tracking cherrypicks is not supported yet [00:49] OK thanks [00:49] This puts a damper on my plans for world domination [00:49] :p [00:49] Do the merge and reverse merge the "bad" revision? === mw is now known as mw|out [00:50] jelmer: Do you prefer bzr send output or bzr diff? [00:50] That would be only slightly better [00:51] awilkins: bzr send please [00:51] I'm just going to have to refactor this code so there aren't any "special" files that can't be propagated [00:51] * awilkins attaches [00:52] Ah, the old "special" files with the CEO's password in. [00:52] yep yep [00:52] nuclear launch codes [00:58] awilkins: For "bzr send", you have to commit ifrst [00:58] you now attached an empty bundle :-) [01:00] I wondered why it was so short, I was really impresed :-) [01:05] * awilkins attaches a patch with stuff in it [01:09] thansk, merged [01:15] is there a roadmap/timeframe for 1.2? [01:25] https://launchpad.net/bzr/1.2/ [01:35] awilkins: thanks [01:45] Say, what are the *.~1~ files? No mention of them in the reference guide [02:20] anybody know what this means: "not updating child fraction" [03:15] mtaylor: A progress bar issue [03:36] rolly: "bzr revert". [03:42] Peng: thanks [03:42] (handy feature) [05:12] floam likes to touch himself at night [05:12] don't listen to him [05:12] can bazaar automatically throw the current revision into files at some spot I wish somehow? [05:12] god damn it [05:25] automatically? there are hooks you can use [05:29] floam: There's no built-in support for keywords like $Id$ though. [05:29] rolly: I just meant, magically have a number pop up in some file under version control without having to call bzr revno or anything [05:29] Peng: ok, thanks [05:30] Oh I misread you. Something like CVS then eh [05:30] floam: Yeah, currently you have to do something like execute "bzr revno" in your Makefile. [05:32] yeah, I was doing that. It just means I need to think a bit harder, because I was hoping the number would be there when my users run a setup.py. I'll just need to keep a static file in the distribution with the number in it [06:06] floam, things popping up and thinking "harder" is what leads yourself to touch yourself at night [06:07] That's a bit of stretch [06:08] ... [06:26] dude. I walked in on the wrong conversation [08:10] rolly: ~1~ files are "revert" leftovers [08:41] jelmer: I have a new trace from bzr seltest svn on win32 if you want it. [08:42] Not that you have a windows box :-) === cfbolz_ is now known as cfbolz [10:18] jelmer: The major error now appears to be that the support for is_executable relies on the inventory, but it's being called in the read_inventory routine (in add_file_to_inv) [11:46] hello [11:46] i am thinking of using bzr for regular backups [11:46] does it only keep the diffs of changed files, or the whole files? [11:47] Stavros: diffs. [11:47] even for binary files? [11:48] Stavros: yes, but the diffs are not very efficient in that case. They can be very suboptimal in fact. [11:48] ah hmm [11:48] and it keeps the history locally as well, so it wouldn't be very efficient for large files, correct? [11:49] Stavros: you can use a local checkout that only stores the history remotely, but requires network access for operations that query history or make changes (e.g. commit) [11:50] james_w: checkins query history for diffs, don't they? [11:50] well, the commit has to put the data on the server, so it has to have the network access. [11:51] before putting the data, doesn't it have to read the old ones to see what's changed? [11:51] yep. [11:52] that's a bit much... but i guess my current rsync setup also does it [11:52] oh hmm no, rsync is per-file [11:55] Stavros: it knows which files have changed, do it only has to find the old versions of those ones. [11:55] hi [11:55] oh aha [11:57] is there a way to create a branch "inline", that is to have multiple branches in the same directory? [12:01] Leonidas: no, that is not currently possible. [12:01] Leonidas: or do you mean use the 'git' workflow of having one tree for multiple branches and just switch the tree. [12:02] james_w: oh, thats a pity. I wanted to do a feature branch, but making it in a new directory would be a lot of work because webserver configurations would need to be changend [12:02] james_w: yaeh, I want to do something like in git - switching the branch with git checkout [12:02] Leonidas: ah, you can do something like that with a bit of work using checkouts and the 'switch' command. [12:03] you basically create two treeless branches elsewhere on the filesystem and then have a checkout in the location you want and use 'switch' to jump between the branches. [12:03] I think there is a 'cbranch' command in bzrtools that makes this a bit easier as well. [12:04] james_w: ah, ok. Does not look too pretty, but this sounds good. [12:04] james_w: I also found http://doc.bazaar-vcs.org/latest/en/user-guide/index.html#reusing-a-checkout [12:04] how would i go about doing local checkouts? [12:04] or is that what "co" or "bind" does [12:04] ? [12:05] Stavros: exactly. You need to use co --lightweight to have the history only stored remotely. [12:05] ah, thanks [12:06] Stavros: you can also see the reconfigure command which might help if you have branches already. [12:06] james_w: as far as I see, cbranch is quite what I was looking for - thanks, I'll give it a try :) [12:07] Hi, I am using the ppa apt repository, but it complains about the crypto-key. Can I add ppa's key to my apt keys? Where can I find it? [12:08] simony: good question. Does it complain on update or install? [12:09] james_w, on install [12:09] simony: and it says it doesn't have the public key, or that they are unsigned? [12:10] it says: Install these packages without verification [y/N]? [12:10] (and also, before it says: WARNING: The following packages cannot be authenticated!) [12:11] simony: I think this is https://bugs.launchpad.net/soyuz/+bug/125103 [12:11] Launchpad bug 125103 in soyuz "ppa archives are not signed" [High,Confirmed] [12:11] simony: so I guess that there is no public key to import to solve the issue. [12:11] can't you ignore crypto warnings on apt-get?? [12:11] minus the second question mark [12:12] (sorry, I haven't used PPAs, so I don't really know what the solution might be). [12:12] simony: #launchpad might be able to give you a better answer. [12:12] I can ignore them, yes. Was hoping not to :-) Thanks [12:13] Stavros: I think there is --allow-unauthenticated, but that is just the same as typing yes, so it doesn't really solve the problem. [12:14] ah [12:31] re [12:42] awilkins: Yeah, I would be interested in seeing that output [12:54] jelmer: hi. when I branch with bzr and svn-psh to a svn repository, it seems to get one spurious revision. see revs 423-424, and 415 in https://forja.rediris.es/svn/csl2-minirok for an example. [12:54] jelmer: I would've expected 424 to be 423 (that is, cp+that change already) [13:12] dato: Hi [13:12] dato: Please file a bug :-) [13:12] This does look like a bug, indeed. [13:47] jelmer: http://filebin.ca/nzoebw/bzr-svn-win32-test.zip [13:47] awilkins: Thanks [13:49] awilkins: The _inventory bit should be fixed in newer revisions [13:49] Was just about to mention it :-) [13:53] awilkins: I've just committed another revision that should help fix the executable bit [13:53] awilkins: Any chance you can run selftest again on a newer revision? [13:54] Sure [13:55] There's a lot of "permission denied" problems cleaning up temp dirs, but not fatal [13:55] And not really a bzr-svn test problem ; it's in the base test class [13:56] Well, the cleanup probably fails because those files are still open by bzr-svn [13:57] I see that a lot on *nix code that gets ported to win32 [13:57] svk has a load of gaffes like that too. [13:57] I guess it's just a lot less graceful about unlinking files [14:03] It's the difference between unlinking and deleting I think [14:03] jelmer: there's another minor thing that I don't know if you've noticed, or consider it "candidate for fixing". [14:03] unlinking (as happens on POSIX) allows you to remove a path while there are still people who have the file that path refers to open [14:04] jelmer: if you give the bzr commit message with the editor instead of -m, in the svn log always appears an empty trailing line in the log, even if there wasn't any in the editor. [14:04] Yeah, windows doesn't let you do that. [14:05] I'm not sure which OS behaviour I like best, to be honest [14:05] dato: I think it's svn that does that [14:05] dato: bzr-svn doesn't add trailing whitespace [14:06] ah, maybe svn unconditionally adds a \n [14:06] awilkins: both have their advantages I guess [14:06] jelmer: maybe you could strip one \n at the end of the message? [14:07] If the \n is added unconditionally, sure. [14:07] please file a bug though :-) [14:08] ok [14:10] New bug: #188353 in bzr-svn "Branching with bzr and svn-push'ing produces a spurious revision" [Undecided,New] https://launchpad.net/bugs/188353 [14:27] jelmer: http://filebin.ca/kqujcc/bzr-svn-win32-test2.zip [14:32] thanks === jelmer_ is now known as jelmer [17:54] Is it possible to delete the parent info from a branch? [17:59] Not sure what you mean. [18:01] I had my project in an SVN repo. With "bzr branch", I created a local branch. WIth push, I created a copy of it on Launchpad. This still has the SVN set as its parent. Now I wonder whether I can detach it from the SVN. [18:03] you can remove it from .bzr/branch/branch.conf, but the parent location doesn't really mean the branch is "attached" to the SVN [18:04] So it doesn't do any harm? [18:04] you mean removing it or keeping it? [18:04] Keeping it. [18:04] not at all [18:04] Okay, thank you. [19:13] hello, I want to publish (push) into a ftp directory, but, my username have a '@' in the username (that is typical for a shared hosting) and bzr don't understand it. [19:14] BlogueroConnor: try push ftp://you%40domain.com@server.com [19:14] %40 is like @? [19:14] encoded? [19:14] will try it [19:15] YES. it worked! [19:16] But I think there is an error in this tutorial: http://doc.bazaar-vcs.org/bzr.dev/en/mini-tutorial/index.html [19:16] because I tried... [19:16] bzr push --create-prefix sftp://your.name@example.com/~/public_html/myproject (with my data of course) [19:17] but I got: [19:17] bzr: ERROR: Target directory XXX already exists, but does not have a valid .bzr directory. Supply --use-existing-dir to push there anyway. [20:13] BlogueroConnor: bzr will only push a branch to a directory that doesn't exist yet or is empty [20:14] BlogueroConnor: the tutorial about that bit looks ok to me [20:30] jelmer, but this directory is empy. (it was, now I used the --use-existing-dir and it worked) [20:31] jerlmer: I can give you an ftp account in my site for you to see. [20:35] Now I understand [20:37] the problem was that I was using mysite.com instead of mysite.com/branchname, because the user I was using, had direct access to mysite.com/branchname when you log in (the base directory of that user is branchname, so I dont have to specify it there). [21:34] I'm a longtime svn user. I used svn2bzr some months back on a repository, and now I need to merge changes from svn into my bzr working set. [21:34] bzr merge doesn't work. [21:34] do I need to bzr branch then merge that new branch into my existing bzr branch? [21:47] does bzr merge -r n..m work? [21:47] bkc: that only works with bzr-svn [21:47] bkc: not with svn2bzr [21:49] yes, I have bzr-svn already installed. I just used svnimport to create the original bzr import [21:50] I guess bzr co svn doesn't support https [21:50] bkc: "bzr svn-import" != svn2bzr [21:50] if you used svn2bzr to create the original import, you can't use bzr-svn now [21:51] bkc: bzr-svn should support https fine [21:51] ok, so I guess I need to diff3 by hand for every file then? [21:51] there have been changes in the bzr repository and the subversion one, I need to get the subversion changes into bzr [21:52] You could try "rebasing" first [21:52] regarding bzr-svn with https, I have an https url to a subversion repository. How do I check it out using bzr? [21:52] I have a svn password on it, but I've already used svn info to cache the password [21:53] bkc: You should be able to just use the https url [21:53] bkc: If that doesn't work, please pastebin the error [21:53] I tried bzr branch https:// ... and I get bzr: ERROR: Invalid http response for https:/ .. tal2xslt/.bzr/branch-format: Unable to handle http code 401: Authorization Required [21:54] bkc: Try prefixing the URL with "svn+" [21:54] svn+ .. I get SubversionException: ("Undefined tunnel scheme 'https'", 125002) [21:54] bkc: Your local svn libraries aren't build with SSL support [21:55] I think [21:55] what version of bzr-svn are you using? [21:55] the subversion client, or the bzr build? [21:55] it could also be caused by using a version of bzr-svn < 0.4.3 [21:56] yeah, it's 0.4.1-1 I'm on feisty using standard packages [21:57] but maybe I could svn co, then use bzr merge from the svn checkout directory? [21:57] bkc: a .svn directory doesn't contain any historical data [21:57] it will still have to go to the server and that will fail [21:58] if it is at all possible, I would recommend upgrading to a newer version of bzr-svn [21:58] I'm worried about breaking bzr-tools, which I need for eclipse [21:59] the bzr-eclipse plugin you mean? [21:59] I thought that only depended on bzr-xmloutput plugin [21:59] yes, apparently the version I have calls out to bzrtools [21:59] maybe my plugin is old.. [21:59] beuno: ping [22:07] hi, what is the best way to covert a cvs repository to bzr [22:08] especially if you have just have checkouts :) [22:08] the cvsps document seems to imply that you need server side access, since it doesn't support :ext: or :pserver: [22:16] where can i find resources on embedded bzr? [22:17] * johnny_ has never even heard of that [22:17] to quote: "If you want intelligent version control embedded into your application or content management system, Bazaar has the architecture you need." [22:18] sounds like you can use it as a lib [22:19] or backend [22:24] You can do "import bzrlib". [22:24] and fiddle with it? [22:25] bzrlib is the entire codebase. [22:25] The "bzr" script is just a little bit of startup code. [22:31] and then go where? [22:31] theres a lot of submodules [22:34] Aviator, gonna have to read the pydocs i assume [22:38] Aviator: http://bazaar-vcs.org/Integrating_with_Bazaar might be a start === bob2_ is now known as bob2 [22:40] call me stupid. I have a checkout of an lp branch I host. I have committed my changes, but how do I upload the changes back to lp? lp's page still doesn't show my revs. [22:41] if it's a checkout, they're uploaded on commit [22:41] oh .. they finally shows up. I guess a commit on a checkout doesn't require a push.. just slow [22:41] right [22:41] thanks, I'm still trying to get the hang of bzr vs svn [22:42] jelmer: thanks for your help too.. was a good time to learn how to use kdiff3 [22:42] "bzr log sftp://bazaar.launchpad.net/foobarbaz" will probably show them instantaneously [22:50] so, is there a way to import a cvs repository without the full cvs root? [22:53] I think https://launchpad.net/bzr-cvsps-import works over the network [22:54] this page says it doesn't support :ext: or :pserver: http://bazaar-vcs.org/CVSPSImport [22:56] hm, right [23:27] jelmer, pong [23:30] johnny_, bzr-eclipse doesn't depend on bzrtools at all [23:31] just bzr-xmloutput [23:33] huh ? bz-eclipse ? [23:33] why did i say anything about bzr-eclipse [23:38] johnny_, ahm, sorry, wrong person [23:38] * beuno should read backlogs more carefully [23:48] beuno: why does bzr-xmloutput depend on bzrtools then?