[00:02] jfroy: pack doesn't alter what data is visible or how its indexed [00:03] jfroy: all it does is shuffle it around on disk so that fewer I/O operations are needed to get at related data [00:06] well it definitely fixed my problem [00:06] I was able to pull the remote svn branch after packing my local repository [00:06] whereas before the pack it would die on a missing revision exception. [00:11] how confident are you that no other operations were done in between the failure and the pack [00:12] include updates to svn/bzr-svn/bzr [00:17] if pack does fixit there is a critical bug in bzr's repository code [00:22] lifeless: and not bzr-svn's ? [00:22] SamB: bzr-svn implementation a repository proxy around libsvn [00:23] SamB: it doesn't, as far as I know, rummage around through bzr's repository implementation internals [00:23] true [00:23] it doesn't mess with access to a native repository ... [00:24] so if packaging a repository you're pulling into stops a bzr-svn bug happening, packing has changed the behaviour of [some] public interface || bzr-svn is poking under the hood badly [00:25] I think the latter is unlikely and that leaves the former, assuming that jfroy has gathered the evidence accurately [00:25] jfroy: next time this happens, can you take a backup of your bzr repo and your ~/.bazaar/bzr-svn cache die [00:25] *dir* [00:25] jfroy: then, try the pull again, if it fails again, pack and pull [00:26] if that works, backup your backups and then put them back in place [00:26] jfroy: and try the pull again, if it fails once more, we'll have a reproducable dataset to work with [00:26] if the repo is private we'll write code for you to test [00:27] if its not private you can schlep it to us [00:27] is schlepnet fast? [00:28] it can be [00:29] lifeless: will do [00:32] thanks === dereine is now known as dereine[OFF] [01:41] Is there an equivalent to "bzr revno" that shows the revid? [01:41] jfroy: hi [01:41] Well, bzr lookup-revision $(bzr revno) [01:43] yeah, bzr only uses public interfaces [01:47] jfroy: which bug was it exactly that seemed to be fixed by a 'bzr pack' ? [01:54] jelmer_: hey [01:54] Sorry just checked IRC [01:54] jelmer_: https://bugs.launchpad.net/bzr-svn/+bug/364416 [01:54] Ubuntu bug 364416 in bzr-svn "bzr branch fails with "ERROR: The branch has no revision. None" with bzr foreign + bzr-svn 0.6d" [Low,Triaged] [01:58] Peng_: bzr revision-info [01:59] lifeless: Oh, nice. [02:01] jfroy: any chance you can comment on what seems to fix it in the bug report? [02:03] sure [02:03] but I don't know much [02:03] I simply packed to "try see if it does anything", and it unexpectedly did. [02:04] I didn't make a backup of the repository before packing, unfortunately. [02:04] although... [02:04] we may be in luck, actually. [02:04] I backup my documents automatically [02:05] I most likely can revert the repository to its state prior to packing [02:05] and try the sequence all over again [02:05] jfroy: which repository format are you using? [02:06] 1.14 [02:06] (-rich-root of course) [02:12] jfroy: really? bzr info and paste the format details please [02:12] lifeless: 1.14-rich-root or 1.9-rich-root [02:12] ah actually nvm [02:12] its 1.9 repo [02:12] anyways, restoring from backup and trying again [02:12] a new tree format is in 1.14 is all, it supports filters [yay] [02:12] see if I can reproduce [02:14] ahhh, backups :) [02:16] Is there any flags I can use to spit out more info in the bzr.log? [02:16] the repo is confidential, I won't be able to share it [02:16] jfroy: for this operation, nothing offhand [02:16] Although I can certainly paste logs and what not [02:16] we may write some custom patches to debug [02:17] jelmer_: I'll be using your bzr foreign branch and bzr-svn for this [02:17] it's what I usually use [02:22] eh this time it's still dying after the pack [02:22] this is very odd [02:23] mmmm [02:23] interesting [02:23] so after the pack, I still got the missing revision exception [02:24] I then pulled the trunk branch (which was already in the repository) [02:24] repacked [02:24] jfroy: at this point my bzr-foreign branch is basically bzr.dev with bzr+http removed [02:24] and the the branch command worked [02:24] so it may be the trunk pull that fixed it, not the pack [02:25] let me try this again starting at the beginning... [02:25] k [02:26] jelmer_: I think it does a few more things, no? [02:26] jfroy: no, the other things have been merged now [02:26] push working to create new branches and printing the svn revno when pulling [02:26] I see [02:26] gotcha [02:26] good to know [02:26] jfroy: it used to be more [02:26] yeah [02:26] was the fallback credential store stuff merged in too? [02:26] I adopted the protocol in my keychain store and it's working great [02:27] jfroy: the trunk pull is likely it [02:27] It makes a lot more sense, yes [02:27] yes, yes it does :) [02:28] pack should not fundamentally change the content of the history [02:28] right [02:28] it just makes it tidier [02:29] confirmed [02:30] the trunk pull fixes the branch command [02:31] I also confirmed I cannot pull the branch in a new repository without having trunk in it [02:31] if trunk is not there, it always dies with a no such revision exception [02:31] so it's unlikely that the svn cache is involved here either [02:34] jfroy: are you pulling from svn [02:34] or from bzr [02:34] there is a bugfix in bzr (server side) for 'NoSuchRevision' pulling from a stacked branch [02:35] pulling from svn [02:35] in all cases [02:35] jelmer_: confirmed, I just made a brand new repository [02:36] tried to pull the svn branch in it, died with the usual error [02:36] I then pulled the svn trunk in that repository and then pulled the svn branch again [02:36] and that worked [02:36] so it seems bzr-svn is failing to pull certain revisions it needs [02:36] I've updated the bug with that finding. [02:39] thanks [02:40] if you need revprop / file props to debug this when you get around to it, let me know [02:56] will do, thanks [03:55] lifeless: btw, did I mention dpush to remote git branches works now? [03:56] the only thing remaining now is 'proper' push [03:58] ye [03:59] it would be nice if somebody could fix up hg support at some point [04:01] I think the API changed or something but it can't do much else than view revision history atm === dereine[OFF] is now known as dereine [08:58] hola, I have a stupid question that I couldn't figure out from "Bazaar in 5 minutes".. If I want to force checkout the latest revision of file abc.py in current folder, what is the command? [08:59] if you want the one corresponding to the currently checkout'd revision, bzr revert abc.py [09:00] (ie it will undo all uncommited local changes) [09:00] pzico: Bazaar considers the entire tree 1 object. If you are using a checkout `bzr update` will make it up to date with the master location. If you're using a branch, `bzr pull` will pull in all the new revisions from your parent location. [09:00] pzico: or possibly what bob2 said, from your wording it isn't clear to me what effect you're looking for [09:03] Well, I haven't yet learned to think "bazaar" way. So I just made simple project and used bzr init, add and commits.. now I just want to revert back to previous committed version.. [09:03] and only for one of the files I have edited [09:04] pzico: ah ok, if the most recent commit is the one you want to go back to `bzr revert abc.py` [09:04] It seems that the revert command did the job. Thanks. [09:04] pzico: or if you would like to bring it to an arbitrary revision, `bzr revert -r 123 abc.py` [09:04] cool [09:54] Does anybody know of an Xcode plugin for Bazaar? === dereine is now known as dereine[OFF] === dereine[OFF] is now known as dereine [13:51] hmm [13:52] anyone aware of an api that kind of combiness Workingtrees iter_changes and list_files ? [13:53] i need renames and unknown files at once if possible [14:25] ronny: TreeDelta? [14:25] of working tree against the basis tree [14:25] currently im using wt.iter_changes(wt.basis_tree()) [14:26] ronny: can't you ask for unknown in iter_changes? [14:27] want_unversioned=True [14:27] ah, [14:28] hmm [14:28] i have that, yet its still somehow missing something [14:29] * ronny rechecks his test helpers [14:30] hmm, i have include_unchanged and want_unversioned [14:33] do i have to invalidate a workingtree after a commit? [14:35] ronny: how do you mean [14:36] hmm, i think im confusing something about the api here [14:39] ok, i think i found my error [14:39] lifeless: i didnt notice i had a broken test [14:40] hi guys - a question: is there much difference between push/pull and commit/checkout when working on a remote repo with others? [14:44] oh darn [15:43] vxnick: Not really.. they're basically the same... A "bound" branch is just one that pushes after it commits. etc... [15:43] LeoNerd: thanks, I thought it might've been something like that [17:31] guys - I think I'm going crazy here. I've created a shared repo and a trunk on a remote server, but can't checkout locally - I get the "bzr error not a branch" error [17:35] vxnick: are you checking out the branch location? [17:35] I mean, the location of trunk [17:36] jelmer: I am indeed [17:36] jelmer: I 'bzr init-repo project_name' and then 'bzr init trunk" within that dir [17:36] vxnick: so the url ends in /trunk? [17:36] jelmer: that's right [17:37] does bzr info work on the URL? [17:37] nope [17:37] not a branch [17:38] oh dear [17:38] the path was wrong [17:38] *embarrassed* [17:38] thanks for your help [17:38] np [18:53] bzr-upload stores the remote location by default, but is there a way that I can switch between locations, perhaps by using some form of alias? I think I might have read about this somewhere but can't find anyhting [18:53] Sometimes I wonder if 'bzr update' should automatically fall back to 'bzr pull' when the branch isn't bound. === zirpu2 is now known as zirpu [18:57] hi, I'm new to bzr and distributed vcs all together, I've created my first bzr repo with a friend an we are using the decentralized workflow, we wanted to use PQM as our gate keeper but I'm having trouble even getting started, is there a good how to available, I can't find one [18:59] vxnick: have a look at bzr-bookmarks, it allows you to refer to different locations by short names [19:00] james_w: thanks [19:00] cdecarlo: I don't know of one I'm afraid [19:19] is --fixes lp:xxx just supposed to link the branch? [19:20] that's all it does currently [19:20] there are plans to be a bit more useful I believe [19:20] ok, I thought it would set fix committed too [19:20] * Laney does so [19:58] jelmer: ping? === vxnick is now known as vxnick_away [20:34] is it possible to manage permissions/access rights with a bazaar smart server? [20:35] dirkD: yes [20:35] LarstiQ: :O [20:35] how? [20:35] just with unix permissions? [20:35] (i can't find it in the user guide) [20:37] LarstiQ: I'd say "yes" is a kind of cheating, until the smart server still has VFS calls [20:40] luks: i'm using bzr 1.14 [20:40] so... RPC calls? [20:41] dirkD: you can ignore me, that was purely technical :) [20:41] what kind of access rights do you want to setup? [20:42] i want to give some users access to a branch, and some users not [20:42] or groups... [20:43] the easiest way is probably to use contrib/bzr_access [20:43] assuming you want to use bzr+ssh [20:43] indeed [20:44] is there any documentation on bzr_access? [20:44] (it's a plugin i assume? or a patch?) [20:44] there are some examples in the file itself [20:45] you basically give users ssh access and restrict the commands they can execute [20:45] ah [20:45] it's in the source code tarball [20:49] luks: looks fine :) [20:49] but they're talking about .ssh/authorized_key [20:49] is that about ~/.ssh [20:50] + ? [20:50] so i do have to edit authorized_keys for every user? [20:50] (or create a symlink to a system-wide one of course [20:50] + ) [20:52] sorry, I don't know more about that [20:52] maybe there is a global way, I don't know [20:52] ok, thanks for your help [20:52] hello, how can i create plugin on the bzr server? [20:53] i've got bzr+ssh server - on the client side pre_commit works as expected, but i cannot make it run on server on (bzr co, bzr up, bzr ci -m "" flow) === vxnick_away is now known as vxnick [20:58] is this possible? [20:59] bluszcz: Plugins don't do that. Consider the SFTP transport -- it *couldn't*. [21:01] I think there is a post push hook [21:02] hm, no, that's not the one [21:02] On the client, maybe. [21:02] luks: I see what you mean, and what I was thinking of is also not directly helpful (Launchpad/ClueBzr style smartserver) [21:03] hmmmm... [21:03] luks: post_branch_tip_change or something like that [21:03] yeah, that one [21:03] CardinalFang: plugin's don't do what? sorry, but i didn't understand [21:03] bluszcz: `bzr help hooks` should mention which ones get run on the server [21:03] ok [21:04] strange, I can see post_tip_change mentioned in tests, but not the code [21:04] i was reading BranchHook.__init__ ;) [21:05] let me fish up how I run bzr-email on the server [21:07] ok, i read help hooks but still have no idea how to install server side hook [21:07] bluszcz: in all ~user/.bazaar/locations.conf I have: [chroot-*:///srv/bzr/]\npost_commit_push_pull = True\npost_commit_url = bzr+ssh://server/srv/bzr\npost_commit_url:policy = appendpath\npost_commit_to = commits@ [21:07] etc [21:08] bluszcz: the plugin you execute must be loaded by the bzr your users execute [21:08] (on the server) [21:08] bluszcz: other than that, that should be it [21:08] reading [21:09] hm, bit complicated [21:09] true. [21:09] some of that is due to how bzr-email works [21:10] bluszcz: if you write your own plugin, the key thing is to make sure your plugin code gets loaded, and that you configure for the right branch [21:10] if you do it for all branches, that last bit doesn't matter. [21:15] hm [21:15] my configuration is: [21:15] i've got ssh acount on my server [21:15] my user [21:15] ;) [21:16] there are branches [21:16] which are checkout locally and using it as i write above (locally make commit for remote branches) [21:17] generally at this time, i am not interested in any wide system/multi user solution, only "wide" in my needs is that i want to run some plugins on the server during commiting to that branches, all branches. [21:19] (reading http://bazaar-vcs.org/ConfiguringBz) [21:24] i found also this: [21:24] https://lists.canonical.com/archives/bazaar/2008q3/044390.html [21:27] pre_change_branch_tip (Branch) [21:29] bluszcz: does that get run on the smartserver? [21:31] jelmer: so what's new and sexy about bzr-git 0.3.0 ? [21:33] yeah [21:34] moin [21:35] bzr: ERROR: Received bad protocol version marker: 'sage 3 (bzr 1.6)\n\x00\x00\x00\x1dd16:Sof' [21:35] lol [21:35] server: Bazaar (bzr) 1.14.1 [21:35] local: $ bzr version [21:35] Bazaar (bzr) 1.14.1 [21:35] so, wtf? [21:35] bluszcz: its been mangled [21:36] bluszcz: you have something outputting to stdout while you login to your machine, if you're using bzr+ssh [21:36] that will also break sftp [21:38] stdout from my local? [21:38] the server [21:39] what protocol are you using? [21:39] yes, pluging is talking to me [21:39] bzr+ssh, ssh2 [21:39] ok, if you do "ssh host: echo" [21:39] do you get anything shown ? [21:39] sorry, get rid of the : there [21:39] bluszcz@atomic [(nie maj 10) 22:39]:~/projekty/blogi [21:39] $ ssh bzr-bluszcz echo [21:39] bluszcz@atomic [(nie maj 10) 22:39]:~/projekty/blogi [21:39] $ [21:40] only \n [21:40] ok that looks fnie [21:40] *fine* [21:40] must be something else [21:40] yes, i now [21:40] how can i pass -vvv for ssh in bzr+shs? [21:40] ssh [21:40] if you have a plugin, which execs something and that writes to stdout, it can also interrupt a bzr session [21:41] because the bzr protocol is writing to stdout [21:41] hm [21:41] so, i should use stderr? [21:41] oh yes [21:41] to write information about "your commit message is empty?" [21:41] in a plugin on the server? [21:41] btw, lifeless - nice nick ;) [21:42] lifeless: yes, on the server [21:44] one sec I'm looking up a bug for you [21:44] ok [21:44] it looks like working now [21:45] :) [21:45] https://bugs.edge.launchpad.net/bzr/+bug/364908 [21:45] Ubuntu bug 364908 in bzr "server side 'progress'" [High,New] [21:45] thats a closely related issue [21:46] however, if your plugin is actually stopping the commit/push, you should probably just raise an exception - except that won't get formatted well. hm time for another bug [21:46] yea [21:47] raise "You should check first you *py files, or die" [21:49] if this is a pre-tip-change hook [21:49] there is a specific exception for that you can raise [21:49] which I think gets formatted nicely === beuno_ is now known as beuno [21:49] errors.TipChangeRejected [21:51] https://bugs.edge.launchpad.net/bzr/+bug/374605 [21:51] just filed that for you [21:51] Ubuntu bug 374605 in bzr "way for plugins to present non InternalError from hpss" [High,Confirmed] [21:53] :) [21:54] Robert Collins 2 minutes ago Changed in bzr: [21:54] importance: Undecided → High [21:54] status: New → Confirmed [21:54] things really goes fast [22:03] ok, now i have to read how to read file liste [22:46] ech, it has only: # (params) where params is a ChangeBranchTipParams with the members # (branch, old_revno, new_revno, old_revid, new_revid) [22:46] no way i suppose to check the commited files [22:47] is it possible that getting a dev6 branch over http would use very large amounts of RAM? [22:48] dev6? [22:48] user ram on which element? client or server? [22:49] bluszcz: the development6 format [22:49] "uses" [22:49] mwhudson: if it is file a bug [22:49] mwhudson: it might if transcoding [22:49] lifeless: ok, finding out now [22:49] lifeless: it really really shouldn't be doing that [22:49] but i guess i can find out [22:51] <[tpd]> Hey folks, does anyone here use bazaar for website deployment? [22:51] [tpd]: some people do, and they find the bzr-upload plugin useful [22:52] <[tpd]> Ah, is that a standard way of doing it? I'm new to vcs. [22:55] [tpd]: it's what the plugin is for [22:56] <[tpd]> I understand that, but is this how it "should be done"? [22:56] i don't know of any better way [23:31] how is the best way to gest list of modified files using bzrlib.? [23:31] in a commit? [23:31] si [23:31] bzr st -c rev [23:31] oh, bzrlib [23:31] si [23:32] tree1, tree2 = bzrlib.repository.revision_trees([rev1, rev2]) [23:32] changes = tree2.iter_changes(tree1) [23:32] checking, thanx [23:33] no such revision_trees hm [23:34] sorry [23:34] I meant 'repository.revision_trees' [23:34] where repository is a bzrlib.repository.Repository object [23:34] if you have a branch its branch.repository [23:34] if y ou have a working tree its wt.branch.repository [23:34] working tree is easiest i think