[02:05] how to change the parent branch of bzr? [02:10] xiaohui: bzr pull --remember [02:12] jelmer, thank you [02:15] abentley, around? [02:15] jelmer, do you know much about bzr hooks? [02:16] rockstar: hi [02:16] rockstar: a bit [02:17] jelmer, so, I've got a class that inherits from bzrlib.hooks.Hook, and it registers two hook points. How do I fire those two hook points? [02:18] rockstar: the hook points are iterables over the registered hooks [02:18] so you can just do something like: [02:19] for hook in myhooks_instance['hook_name']: [02:19] hook(42) [02:19] bzrlib.hooks.HookPoint has a hook method, but not an fire method. [02:19] jelmer, you would think that the HookPoint would know how to fire itself. :/ [02:20] rockstar: no, you have to call the individually registered entries yourself [02:20] although I guess it would indeed make sense for HookPoint to have such a method [02:22] jelmer, yea. Something like HookPoint.fire(*args) [02:25] jelmer, I'll implement create a subclass of HookPoint in Tarmac, try it out, and if it works, I'll create a patch for Bazaar. === Kissaki is now known as Kissaki^0ff [07:39] Is there a simple way to add revisions to a branch created with BzrDir.create_branch_convenience ? [07:44] Attempting 1.15rc1 release and I keep getting "bzr: ERROR: socket.error: (104, 'Connection reset by peer') " several network, all peered differently, even 1 network from a large hosting provider in the midwest (US). Something wrong with LP tonight? [07:45] BasicOSX, on push? [07:46] or branch or pull [07:46] BasicOSX, hm, I haven't had any problems, and I've been using it steadily tonight. [07:51] rockstar: what do you mean by add revisions [07:51] rockstar: is this test code or real code [07:51] lifeless, test code. [07:51] BasicOSX: what aare you getting that from [07:51] bzr branch http://code.launchpad.net/~bzr/bzr/bzr.1.15 [07:51] gives that to me [07:51] rockstar: and what do you mean by add revisions [07:52] lifeless, basically, I want to test that two branches are merged. I was planning on going to find the merge tests in bzrlib and see how they work. [07:52] lifeless, if I create two branches with BzrDir.create_branch_convenience and try to merge them, it throws NoCommits (which makes sense) [07:52] A and B are merged if len(heads(a.tip, b.tip)) == 1 [07:52] lifeless: attempting ~/projects/bzr/bzr.1.14/bzr branch http://code.launchpad.net/~bzr/bzr/bzr.1.15 to see if it's a problem(?) in bzr.dev (revision: 4369) [07:53] rockstar: tree1 = self.make_branch_and_tree('1'); tree1.commit(''); tree2 = tree1.branch.push('2') [07:53] tree1.commit('next') [07:53] lifeless, yes, so I want to add some revisions b so that I can actually make a merge (but not necessarily commit) [07:53] tree2.commit('next in 2') [07:53] tree2.merge_from_branch(tree1) [07:53] lifeless, I'm not using the bzr test suite. [07:54] rockstar: oh; any reason preventing you? [07:54] bzr: ERROR: Connection error: Couldn't resolve host 'bazaar.launchpad.net' (-2, 'Name or service not known') things keep getting stranger [07:54] lifeless, um, mostly because I'm too lazy to go try it out, and I probably don't need most of it. [07:54] I did just recently upgrade to 2.6.24-24-server on my hardy system, but that's the only change [07:54] lifeless, but mostly lazy. :) [07:55] rockstar: from bzrlib.tests import TestCaseWithTransport [07:55] thats all thats needed :) [07:55] lifeless, I am stealing the Hook/HookPoint stuff though. [07:55] Well, not stealing, but using. [07:58] rockstar: to add revisions you need to do commits or pull/push from another branch [07:58] rockstar: by far the easiest way is using the bzr test harness [07:58] lifeless, okay, that's simple enough. [08:02] BasicOSX: anything I can do, let me know [08:03] BasicOSX: but I think you have some local trouble ;P [08:03] I'd like to think so, but 4 different computers on 4 different network on 4 different ISPs? [08:04] Comcast, UUnet, Qwest, and Sprint all having problems? could be, heck google went down the other day :-) [08:15] lifeless: 373455 when from Fix Released to Fix Committed, that normal? [08:15] bug 373455 that is [08:15] Launchpad bug 373455 in bzr "brisbane-core doesn't support stacking but should" [High,Fix committed] https://launchpad.net/bugs/373455 === sabdfl1 is now known as sabdfl [08:30] BasicOSX: no [08:31] BasicOSX: fix committed means 'it has landed in bzr.dev' [08:31] But bzr uses "Fix Released" for that. [08:32] oh yes, I'm explaining badly [08:32] what I mean is, if its not ix released, its not in trunk [08:33] I don't think we've finished fixing stacking [08:33] I could be wrong === BasicOSX changed the topic of #bzr to: Bazaar version control system | 1.15rc1 released 15 May, 2009 | http://bazaar-vcs.org | http://irclogs.ubuntu.com/ | http://planet.bazaar-vcs.org/ [08:46] Congrats. :) [08:47] Now the scary part.. merge back into trunk :-P === Kissaki^0ff is now known as Kissaki === Spaz is now known as Fapjack [10:42] is there a way to tell bzr to ignore changes in permissions? I just copied everything to a fat32 usb stick, and it's telling me that the entire repo has been changed [10:43] alsuren: the only permission bzr tracks is the executable bit [10:44] alsuren: are you using bzr under windows? [10:46] no, I'm on ubuntu. Problem is that I made *some* changes [10:50] oh, I just read the bit on status-flags. I reckon I could hack something together with grep to work out which files I *actually* modified [10:58] yeah, so parsing the output of status -S is actually really easy. Why isn't this the default? [10:58] actually, I'm quite glad that it's not [10:58] ignore me [11:09] :D [11:09] okay, so actually I still need help [11:09] $ bzr revert [11:09] bzr: ERROR: [Errno 1] Operation not permitted [11:10] there is a bug in launchpad about it [11:10] https://bugs.launchpad.net/bzr/+bug/190725 [11:10] Ubuntu bug 190725 in bzr "Bzr can't init branch on ntfs-3g filesystem" [Medium,Triaged] [11:12] (for reference, I used cp -R proj/.bzr /media/disk to do the copy) [11:14] does anyone know a way around this, or am I just screwed? [11:25] alsuren: well the bzr metadata should be just fine [11:25] alsuren: you could try branching fresh [11:27] *bets he'll run out of space if he does that, but tries it anyway* [11:27] no. still operation not permitted [11:31] oh, ololol: it's symlink that's throwing === Fapjack is now known as Spaz [11:36] okay, so removing the symlinks from the repo makes revert work, but I still need some way to stop bzr from trying to track the execute bit [11:38] Hmm, content filters? :D [11:40] Peng_: https://bugs.launchpad.net/bzr/+bug/190725/comments/8 suggests a plugin. how do I need to change it? [11:40] Ubuntu bug 190725 in bzr "Bzr can't init branch on ntfs-3g filesystem" [Medium,Triaged] [11:42] (I'm thinking that the real solution would probably be retro-fitting permissions to FAT at the operating system layer, but we all know that's not going to happen [11:46] hrrrm... I guess I could monkey-patch os.stat to clear the execute bit, but that's *really* horrible [11:47] o [11:47] jelmer: I just hit http://dpaste.com/44691/ when trying to push to Debian SVN. [11:48] so the basic problem is, fat is returning garbage data, and we assume when the platform is unix that all fs's track X reasonably well [11:48] we have code to work around no X tracking but only activate it on windows [11:48] file a bug please [11:48] I suggest we add a a config setting/environment variable to force that workaround code on in this sort of situation [11:49] What about explicitly checking if the fs supports it? [11:49] Same way you do the CIFS checks? [11:49] Peng_: what CIFS checks [11:50] Peng_: do you mean the case insensitive ones? [11:50] Yeah. [11:50] we need either reference data or to write [11:50] its possible in principle [11:50] lifeless: should I paste this conversation to the bottom of my bug report? [11:54] lifeless: where is this no-X-tracking magic, and can I enable it with a plugin in the short-term? [11:54] Offtopic : Is there an ability in the SFTP protocol to just get a hash or crc of a file, not the file? [11:55] Thinking about it, is tracking x-bit a security risk, and should it be done at all? [11:58] awilkins: where did you get that idea from? [11:59] bzr is used for pushing source code around, which is compiled and executed on the user's system. There are far more dangerous things that you can do than twiddling an X bit [11:59] alsuren: Yes, put that way it does seem less significant [12:05] alsuren: its mainly in workingtree*.py [12:05] awilkins: don't think so [12:05] [to both SFTP and x-bit security] [12:05] bzr doesn't run the users code [12:05] its why we don't have in-tree plugins, for instance [12:06] Fairy snuff. [12:06] as long as bzr doesn't run code from a branch, there isn't a risk that is bzr's fault ;) [12:06] now,if bzr tracked owners [12:06] or setuid [12:06] bzr actually can try to run some python code :) [12:07] that would be different [12:07] luks: from the users branch? [12:07] yes [12:07] how?! [12:07] via imports [12:07] thats the system python path [12:07] outside bzr's control; file a bug on your OS/python install :) [12:07] sure, but it can happen :) [12:08] same as bash can run some code from there [12:08] but bash is expected to do so [12:08] no [12:08] having '.' in your path is widely recognised as a security law [12:08] *flaw* [12:08] same goes for . in your python path [12:09] except that's the default setting for python [12:10] luks: well if you fancy porting bzr to python3 and using "from . import whatever" all over the place, then go for it [12:12] I'm not saying anything except that it's possible for bzr to run user's code right now [12:12] I know about it and I can live with it [12:14] that's actually something I'd never thought about before. [12:16] I can't even reproduce it right now, so maybe there was something done to fix it [12:16] I've seen a few people to have that problem on the ML in the past [12:18] luks: reading the tut, that's not how it works [12:18] "Actually, modules are searched in the list of directories given by the variable sys.path which is initialized from the directory containing the input script (or the current directory)" [12:20] so . isn't actually in your path, only $ dirname `which bzr` [12:22] can't really find the posts [12:23] alsuren: that doesn't protect against it :P [12:37] Odd_Bloke: hi [12:37] Odd_Bloke: that's the no 1 bug for bzr-svn, should be fixed before 1.15 [13:08] jelmer: Is there any way I can work around it? [13:12] Odd_Bloke: you can comment out the assertion [13:12] it should generally be harmless [13:20] jelmer: Cool, thanks. [13:29] lifeless: if . isn't in your path, then surely you're pretty safe (sorry, went to lunch) [13:35] https://bugs.launchpad.net/bzr/+bug/377243 [13:35] Ubuntu bug 377243 in bzr "bzr (linux) doesn't handle FAT's limitations (X-bit and symlinks) well. Should behave like on windows." [Undecided,New] [13:41] lifeless: can't find the win32 hacks you speak of [14:40] I have a --no-trees shared repo (bzr init --no-trees). Is there an option to `bzr branch` to override the --no-trees, eg. the opposite of `bzr branch --no-tree` ? `bzr brancj --tree` doesn't complain about the not documented --tree option, but it also doesn't create a tree? [14:41] (I know I can just bzr co afterwards, so not big deal, just wondering if I missed the option) [14:42] knielsen: I don't think there is anything like that atm, but I agree it makes sense [14:42] ok [14:42] Maybe an oversight, since the --tree option is there somehow but seems to do nothing [14:43] might be a good starter project to get into the code ;-) [14:48] knielsen: Yeah, this might be a trivial one to fix === pst is now known as pst555 [15:20] i wonder how i could accomplish something like repo1/lib where lib being a link to repo2 [15:28] is a bzr co lib inside repo1 and then a bzr link-tree the right thing to do? [17:02] jelmer: care to elaborate on "dpush" ? :) [17:03] rocky: what about it? [17:03] just trying to understand it [17:06] rocky: it just pushes into a foreign vcs (such as svn) losslessly (i.e. what can't be represented in that VCS is lost) [17:07] ah metadata can be loss [17:07] hmmmm [17:07] * LarstiQ wonders about a /dev/null foreign vcs [18:33] good evening london [18:35] hi? === Jesse is now known as jesse === jesse is now known as jesse_1 [18:48] hi [18:48] hi [18:49] is it possible to configure bazaar to use a username for ssh to certain hosts? [18:49] i tried authenticantion.conf [18:49] but bzr still uses the $USER value [18:49] I don't know, but on unix with openssh you can use ~/.ssh/config [18:50] aha [18:50] dumb me, i thought everything's configured in bzr [18:50] i am still confused why authentication.conf didn't work [18:51] i am using private key for sftp, but am tired of typing my username in the url [18:51] so i put a section in my auth conf [18:52] with scheme, user & host === krisfree is now known as krisfremen [21:37] Hi everybody... [21:37] I need some help, about Eclipse+Bzr [21:41] ??