[00:40] bbiab [00:45] hi === mw is now known as mw|out [02:14] Is there any mechanism for handling hook script in bzr? For example: I want bzr send me a email when somebody commits his change. ( same function as CVSROOT/loginfo or commitinfo) [02:15] mrbig, there is a plugin that does that in: http://bazaar-vcs.org/BzrPlugins [02:16] beuno: thx, I check it now [02:16] mrbig, :D [02:16] ;) [02:59] * igc lunch & some xmas shopping - bbl [03:19] mrbig: hi nhan! [04:01] opps [04:01] Hi bac [04:01] :) [04:01] mrbig: i just came here to find an answer for you but saw you had already done so. i hope one of those plugins works for you. [04:02] bac: I'm playing with this [04:02] bac: bzr is really cool [04:03] mrbig: great! i hope it works out for you! [04:05] bac: yeah, I've told my team try to use it already. We'll have a small internal presentation about bzr tomorror === bigdog is now known as bigdog_ [04:06] mrbig: i look forward to hearing how that goes. will you be at the LUG dinner tomorrow night? [04:08] bac: I don't know about it, it's a weekly cafe I guess [04:11] bac: ah just got email, sure I'll come [04:16] New bug: #177592 in bzr "Commit fails after reconfiguring from light- to heavy-checkout" [Undecided,New] https://launchpad.net/bugs/177592 [05:12] http://sourcefrog.net/tmp/bzrsite-draft/ [05:12] hi bac [05:12] retro r us [05:13] quick, publish it! [06:01] New bug: #177605 in bzr-cvsps-import "module help too bare bones" [Undecided,New] https://launchpad.net/bugs/177605 [06:01] New bug: #177606 in bzr-cvsps-import "no install guide" [Undecided,New] https://launchpad.net/bugs/177606 [06:06] New bug: #177607 in bzr-cvsps-import "no setup.py" [Undecided,New] https://launchpad.net/bugs/177607 [06:45] can someone tell me a more friendly name for branch format: Bazaar Knit Repository Format 4? [06:45] poolie: hi [06:47] thumper: It's the repository format used in "rich-root". [06:48] abentley: ah, thanks, is this provided through a plug-in? [06:48] No, it's in Bazaar 1.0 [06:48] thumper: no, it's part of 1.0 [06:49] thanks [06:49] * thumper expects LP errors to go away then [06:49] abentley: isn't it almost 2am for you? [06:50] Yep. Just off to bed. [06:50] :) [06:50] night [07:15] New bug: #177613 in bzr "Crash when merging" [Undecided,New] https://launchpad.net/bugs/177613 [07:19] ok, i'm going away for a while [07:19] see you later [07:20] poolie: hi... [07:20] poolie: give me a shout when you get back. i see you pinged when i was out for lunch [07:21] mm [07:21] not that i remember [07:21] i'm leaving for the holidays tomorrow [07:21] have a good holiday [07:21] poolie: 00:12:39 hi bac [07:21] i was just saying hello [07:22] ah, ok. hello and good bye. :) [09:02] poolie: if you're still online, have a good holiday, xmas and New Year. [09:02] * igc dinner [09:14] * quicksilver thinks everyone should have a good holiday xmas and new year, including but not restricted to poolie [09:15] Sorry, there's only one good holiday to go around this year, and it's already been assigned to poolie... [09:17] ah well, I'm off to poolie house to bask in reflected happiness, then [09:39] do tracebacks like [09:39] File "", line 4, in set_last_revision_info_write_locked [09:39] File "/srv/sm-ng/production/launchpad/sourcecode/bzr/bzrlib/branch.py", line 1412, in set_last_revision_info [09:39] assert len(history) == revno, '%d != %d' % (len(history), revno) [09:39] AssertionError: 1368 != 1370 [09:39] look familiar to anyone here? [09:40] I've seen it before, yes [09:40] not sure what causes it though [09:42] it is currently screwed up mirrored branches on launchpad [10:59] jelmer: I can reproduce it with: http://pastebin.com/m189c0ec9 [11:00] that branch may have been modified by .dev versions of bzr [11:02] Reconciling the local copy and reconciling the source to pull --overwrite from doesn't fix it. [11:05] Upgrading the local branch to rich-root format fixes the problem. === `6og is now known as Kamping_Kaiser [11:10] spiv: upgrading to dirstate-tags is enough [11:10] i really think it must be revision-history vs last-revision [11:11] Right [11:11] and so it's probably not as bad as i thought: it's only breaking mirroring of all dirstate branches on lp [11:14] * spiv wonders why _lefthand_history is disagreeing with revision_history [11:14] oh, it's not even that bad [11:14] jelmer@samba.org-20051106183643-4864eee4ce83e874 is the revision it's omitting. [11:14] although there are 2278 failures today [11:15] it is actually only 13 branches that are failing [11:16] for some reason they are not being marked as failing, and are being attempted every time the puller runs [11:21] Hmm!: [11:21] $ bzr revno [11:21] 37 [11:21] $ bzr revision-history | wc -l [11:21] 36 [11:21] well, that looks bad [11:21] Indeed. [11:22] That's after upgrading to dirstate-tags. [11:22] i suspect that we're hitting an old, maybe unreleased, bzr bug that's lead to bogus data [11:22] If I leave it in the original format, they're both 37. [11:22] Yeah. [11:22] does 'bzr check' pass in both branches? [11:23] The disagreement is basically that jelmer@samba.org-20051106183643-4864eee4ce83e874 is sometimes in the revision-history, and it shouldn't be, because it's not a mainline revision I think. [11:24] mwhudson: it does in the original [11:24] It also passes in the dirstate-tags branch [11:24] feh [11:24] (despite "45 inconsistent parents", which reconcile can fix) [11:32] jelmer: deleting jelmer@samba.org-20051106183643-4864eee4ce83e874 from your .bzr/branch/revision-history corrects the problem [11:32] jelmer: I think... [11:32] jelmer: I'm fairly sure that's correct. [11:33] Current bzr probably ought to give a better error than an assert for this case. [11:33] ok [11:34] mwhudson: I need to stop for the night, can you file a bug and make sure jam knows about it [11:34] mwhudson: maybe even mail the list. [11:34] jelmer: Please keep a copy of the branch as is, though :) [11:34] ok, I'll just leave ti for now [11:34] jelmer: thanks! [11:35] mwhudson: bzr pull --overwrite . should fix it [11:35] What is happening is that history was not normalized on one side [11:35] We should make "bzr check" check that .bzr/branch/{revision-history,last-revision} is correct [11:35] jam: great timing [11:36] spiv: you pinged my name :) [11:36] hello jam! [11:36] jam: it's amazing what technology can do :) [11:36] It used to happen in bzr pre 0.8 or so [11:36] and it just sort of stayed there for a long time [11:36] 'bzr.dev pull --overwrite .' crashes too [11:36] (And we should make "bzr reconcile" fix it) [11:37] well, internally it is [11:37] b = bzrlib.branch.Branch.open('.') [11:37] Anyway, that's it for me for the night. [11:37] b.generate_revision_history(b.last_revision()) [11:38] spiv: good night [11:38] mwhudson: it happened in a recent patch to just call "set_last_revision" [11:38] jam: sounds like an easy thing to teach bzr check/reconcile... ;) [11:39] spiv: should be fairly trivial [11:39] * spiv -> zzz [11:39] part of the problem is that we are trusting the source branch to have the correct revno and last revision [11:40] But there seem to be a lot of Branch5 out there with incorrect revision-history [11:40] anyway, the problem is nowhere near as bad as i feared [11:40] yeah, we just used to allow you to pick any path through history you wanted [11:40] but we moved away from that [11:41] mwhudson: are you willing to put together a patch for check & reconcile? [11:41] jam: not right now [11:41] Speaking of reconcile, there's something there that bugs me... [11:41] It seems like 'reconcile' always takes longer than 'check', even when there's nothing to do. [11:41] jam: i want to fix the two problems that made this so scary for me [11:45] fullermd: really? in what repository format? [11:45] I've always found "check" to take longer [11:45] because it checks all texts match their sha1sum [11:46] All knit, probably. [11:46] reconcile doesn't, because at the moment, there isn't really anything you can do if there is a mismatch [11:46] Don't remember runs back in weave days. [11:49] (And I've not had the time/adventuresomeness to try pack) [11:49] knit reconcile << pack reconcile. pack reconcile is very fast, according to the intel users. [11:49] night all [11:49] * fullermd gets whiplash watching the drive-by. [11:50] Knit and pack reconcile both do the same thing, right? [11:51] The end result is the same, presumably, but the mechanics are probably a lot different. [11:55] Peng: they work very differently [11:55] IIRC knit reconcile is a bit more brute force [11:56] Is that good or bad? [11:57] Peng: well it makes knits slower [11:57] knits extract all inventories and work out what files have changed, etc. [11:57] packs do a similar action, but without having to extract all fulltexts [12:00] If I'm converting from knits to packs, would it be a good idea to reconcile both, just to be careful? [12:11] New bug: #177643 in bzr "http auth pycurl is broken" [Undecided,New] https://launchpad.net/bugs/177643 [12:30] Peng: you could, though I don't think the final effect is any different [12:34] is there a way to purge a version of a file from a repository? I hold some sample media files in bzr and would like to keep the version overhead small [12:35] I'm aware that this goal conflicts with having a file under version control at all [12:36] use case here would be 'I'd like to have the latest 3 versions of this audio files available, but not all of them' [12:39] TeTeT: generally, no [12:39] for referential integrity we don't yet allow removing nodes [12:39] jam: ok [12:40] TeTeT: you might be able to put something together with 'rebase' but it would get pretty clumsy [12:40] You can also look forward to [12:40] http://bazaar-vcs.org/HistoryHorizons [12:40] ok [12:40] which should allow people to do partial downloads [12:40] sorry the correct link is: http://bazaar-vcs.org/HistoryHorizon [12:40] I expect them to be finish in the first quarter next year [12:41] for projects like the desktop training where graphics go through 2-4 versions each easily I guess it's best to split the media files from the text files === mw|out is now known as mw === mvo_ is now known as mvo [15:45] Can someone give me some pointers. I'm getting an error when I do bzr launchpad-login username [15:46] specifically bzr: ERROR: Connection error: curl connection error (problem with the SSL CA cert (path? access rights?)) [15:46] wharp: are you a launchpad beta tester? [15:46] (very random guess) [15:46] don't think so [15:46] is it possibly a problem with my key? [15:47] No, it's pycurl not being able to verify the chain to LP's cert. [15:47] no, it sounds like pycurl being a pain [15:47] well from what i can tell I don't have pycurl [15:48] or is it not a package? [15:48] It's just hiding 'cuz it knows you're pissed. [15:48] python -c 'import pycurl' [15:48] Where's my trout. [15:49] tried that, still nothing [15:49] do I possibly need to remove curl? I did install that [15:50] if that didn't fail, then you have pycurl installed [15:50] it didn't fail [15:51] at least, it didn't return anything [15:51] is there a way to test? [15:51] do you have the 'ca-certificates' package installed? [15:51] i do now [15:51] does it help? [15:52] yep, sure does [15:52] thanks, nothing I'd read mentioned that [15:52] just another piece of random knowledge floating around [15:52] just out of curiosity, should I have known I needed that? [15:52] wharp: probably not [15:52] lol [15:53] well, put another way: yes, but it's our fault not yours that you didn't :) [15:53] I'll blame it on Mr. Shuttleworth [15:54] If only we had a FAQ where we could list such things... [15:54] I really didn't see that error mentioned anywhere. [15:54] Is the FAQ a wiki? [15:55] Another question. It's adding bazar.launchpad.net, etc to the host....should it do something after that or will that take a while? [15:56] embarrassingly, given this is my area of the code, i don't really know what it does [15:56] lol [15:57] but basically it just sets a configuration option, it shouldn't take very long [15:58] yeah, palintheus is getting a timeout so I think there might be an issue network wise [15:58] Im geting: ssh: connect to host bazaar.launchpad.net port 53000: Connection timed out [15:59] port 53000 ?? [15:59] thats what Isaid [16:00] * Palintheus double checks /etc/ssh/ssh_config [16:00] well, that makes sense on one level: [16:00] there's nothing listening on bazaar.launchpad.net:53000 [16:03] now when it's pulling something down, is there a progress bar or anything? [16:11] gah! had remnants of an old work-around I needed, now its just sitting there, hope its working [17:39] Is there a way to get `bzr st` relative to your current directory? I.e., not show anything in parent folders, and show all paths relative to the current directory? [17:39] bzr st . ? [17:39] That shows all the changes in the whole tree. [17:39] And all paths are relative to the top most directory. [17:40] er no [17:40] it show only changes below the cwd [17:40] but it does show paths relative to the root [17:40] which is a bug, i guess [17:40] Ah. [17:41] It makes it really tough to look down the st list and type the paths. [17:42] what pycurl version is running on pam ? I have a test involving http proxy failing in a rather strange way [17:42] s/pam/pqm/ ghaa [17:44] vila: whatever's in dapper i'd guess [17:45] wow, how do I find what version of pycurl is in dapper ? ;-) [17:45] packages.ubuntu.com ? [17:46] excellent, thanks [17:46] 7.15.0 [17:51] vila: yes, 7.15.0-1ubuntu1 according to the sysadmins [18:17] will bazaar look into shipping some of the more popular plugins, like git, svn, and bisect along with the standard source distribution? [18:17] because that would be pretty awesome. [18:37] rebase is tooootally confusing. [18:42] bzr: ERROR: dirstate: inconsistent delta, with tree 0. 'fbreader/qvfb/fbreader.desktop' 'fbreader.desktop-20071220183852-09pyx75ogdtn8vpl-6' :( [18:42] * mgedmin has the ability to break software [18:43] what I did: took my old bzr repositories (that could very well have internal inconsistencies in their data structures), ran bzr upgrade, then tried to remove, replace and add some files [18:43] the old repositories were created with an ancient version of bzr that didn't support symlinks, and my source tree had symlinks [18:45] * mgedmin wonders if "101 inconsistent parents" in the output of bzr check is bad [18:46] * mgedmin assumes that BzrCheckError: Internal check failed: Mismatched basis inventory content. is bad [18:47] to fix the dirstate you can try 'bzr branch broken new' [18:47] but I have no idea about the output from bzr check [18:48] * mgedmin should probably just rm the old history [18:51] That check error is probably from the broken dirstate. [18:52] no, that's in a different repo [18:52] er. branc [19:10] mwhudson: thanks for the confirmation (was dining) [19:17] Hi, deepjoy and myself are trying to integrate WebDAV with a bzr centralized server.. Could we get some help? [19:18] brilliantnut, using the webdav plugin? [19:19] yes, or trying to.. :) [19:19] vila is the person you'd want to talk to [19:20] brilliantnut: what bzr version are you using ? [19:20] 1.0 [19:20] :-) [19:21] Are you server's admins ? [19:21] yeah,. [19:21] use bzr+http then ;-) [19:22] We have an ldap server we set up for other authentication [19:22] the plugin doesn't support bzr-1.0 until one bug is fixed and you'll get far better results with smart server anyway [19:22] so Apache is able to authenticate against it for other use [19:22] we're running a smart server right now [19:22] deepjoy: you can still use that [19:22] performance seems good [19:23] with the webdav plugin ?!?!? [19:23] sry, miss one line [19:23] no, we haven't been able to integrate the webdav plugin yet. [19:23] we've set up the bzr:// smart server, and its working cool for us. [19:24] Never mind webdav, could you point us to some documentation for setting up bzr+http [19:24] using ldap? [19:25] bzr+http with use http authentication, if your Apache is configured to use ldap, there should be nothing to do [19:25] so we put the bzr repository directory to be authenticated against [19:25] I think so [19:26] and the the bzr client should work with the bzr+http:// protocol? [19:26] that's the idea ;) [19:26] cool [19:26] we'll give it a try [19:26] :-) thanks a ton [19:26] thanks vila [19:26] have you read doc/en/user-guide/http_smart_server.txt [19:27] deepjoy, brilliantnut: happy to help (TM)(jam) [19:27] well, modulo spiv's recent patch to make it all work smoothly [19:29] jam: I can't remember if it was server-side client-side or both ? [19:29] I see a change to SmartClient [19:29] and one to SmartWSGIApp [19:29] so it looks like both [19:29] http://bundlebuggy.aaronbentley.com/request/%3C20071214015112.GC14963@steerpike.home.puzzling.org%3E [19:29] and [19:29] http://bundlebuggy.aaronbentley.com/request/%3C20071214011937.GA26039@steerpike.home.puzzling.org%3E [19:30] vila: I'll try to get them reviewed today [19:30] which would mean they get merged for 1.1 [19:30] ok, thks for that [19:32] jelmer: hey, could you have a look at bzr-gtk bundlebuggy? i get a 503 http error... [19:32] phanatic, sure, will do [19:33] abentley: Any specific reason rich-root-pack is still marked as experimental ? [19:33] jelmer: thanks. i'd like to go through the pending stuff... [19:35] phanatic: fixed [19:35] jelmer: cool, thanks :) [19:42] Hi. I'm having trouble writing a plugin hook, or rather, I have a plugin written and I think I found a bug. [19:42] phanatic: can you also merge the two requests you've just approved? [19:43] hi CardinalFang [19:43] """bzrlib.errors.UnknownHook: The Hooks hook 'post_commit' is unknown in this version of bzrlib.""" [19:43] jelmer: sure [19:44] phanatic, I think it would be nice to maintain the same policy as core bzr: if the original author is a team member, they do the merge, if the original author isn't, the second approver does [19:45] CardinalFang: What are you calling to install the hook? [19:45] jelmer: okay :) [19:49] jelmer, hooks = bzrlib.branch.Hooks(); hooks.install_hook("post_commit", foo) [19:50] CardinalFang, You don't have to create an instance of hooks [19:50] Ah. [19:50] just something like this should do: [19:50] Branch.hooks.install_hook('post_commit', branch_commit_hook) [19:51] jelmer, TypeError: unbound method install_hook() must be called with Hooks instance as first argument (got str instance instead) [19:53] CardinalFang: What are you calling exactly now? [19:53] note that hooks is with a lowercase h [19:55] jelmer, Yes. install_hooks() is a method of a Hooks class. Calling it as a class method causes that error. [19:55] (My first way was where I roamed after this error message.) [19:55] CardinalFang: sorry, I should've been a bit clearer [19:56] there's no need to create your own instance of hooks [19:56] rather, use the one in the Branch class [19:56] see my example [19:57] jelmer, I think I know the problem. I was testing my code my running it standalone. $ python .bazaar/plugins/foo.py [19:57] "by running" [19:58] no, standalone should work as well [19:58] jelmer: patches applied... [19:58] yep, does here [19:58] phanatic, Thanks! [20:39] hi vila, this is brilliantnut, and deepjoy, back with some questions. [20:40] brilliantnut: pong [20:40] in bzr-smart.py, what should we set the root and prefix parameters to? [20:41] .. [20:41] smart_server_app = wsgi.make_app( [20:41] root='/srv/example.com/code', [20:41] prefix='/code/', [20:41] .. [20:42] so, our repository is rooted at /home/vcs/bzr/ , and the trunk is at /home/vcs/bzr/branches/HEAD [20:42] so, what do we set root and prefix in this configuration to, and what is the significance.. [20:43] BTW, we have apache 2.2, and the is /bzr [20:43] and we don't have any rewrite rules defined.. is that required? [20:44] /bzr points to /home/vcs/bzr/ ? [20:45] in the file-system or in the apache conf? [20:45] :) Whatever [20:46] I guess I'll try both [20:46] :-) [20:47] I'd say root='/bzr' prefix='/bzr' but you may be hitting bugs recently fixed by spiv, in which case you'll need to update to bzr.dev or wait for bzr 1.1 [20:47] prefix='/bzr/' [20:48] hmm. [20:48] have you read doc/en/user-guide/http_smart_server.txt [20:48] yes, I was quoting that earlier. [20:49] ok, just making sure, in the urls mentioned by jam earlier, spiv said: This, combined with the small change to fix the client-side path calculations [20:49] I've posted separately, makes bzr+http actually work in non-trivial situations. [20:49] [20:50] brilliantnut: I think before Andrew's patches, you have to set "root = /path/to/repo" and that must line up with "http://host/" [20:50] spiv is the definitive reference on these problems anyway [20:51] In other words, you set root to the location that is the root of public space [20:51] spiv == Andrew [20:51] After spiv's patches land (I'm just finishing a review of one) [20:51] Is he here? [20:51] brilliantnut: he is, but is in Australia [20:51] so we won't be online for at least another hour [20:51] anyway, after spiv's patches land, you can do: [20:51] root = /path/to/repo [20:52] prefix = '/repo/' [20:52] And have that exported as [20:52] http://host/repo [20:52] brilliantnut: am I making sense? [20:52] you are, but would this also require some special handling in httpd.conf? [20:53] or a blind redirect for everything in the /bzr context to the bzr-smart.py would do? [20:53] the latter is what we have set up. [20:53] brilliantnut: I *think* a blind redirect works [20:54] I can't manage to get hooks to work. Can someone give me a dummy "pass" example? [20:55] jam: before Andrew's patches, (i.e. on 1.0), if we set root = /path/to/repo, what should we set prefix to? [20:55] It has been a *long* time since I set that up, give me a sec [20:55] CardinalFang: what hook [20:55] (just to give you an example you want) [20:57] FYI we keep getting [20:57] bzr: ERROR: Not a branch: "bzr+http://user%40comain.com@dev.domain.com/bzr/branches/HEAD/" [20:57] when we try a checkout or a branch or a bind to that url [20:58] brilliantnut: so I set it up to be [20:58] if we add or subtract from the URL we get 404's [20:58] root = '/home/jameinel/dev/bzr' [20:58] prefix='/bzr/' [20:58] and in http.conf [20:58] I have [20:58] Alias /bzr /home/jameinel/dev/bzr [20:59] $ cat dummy.py [20:59] def notify_list(local, master, old_revno, old_revid, new_revno, new_revid): pass [20:59] import bzrlib.branch [20:59] bzrlib.branch.hooks.install_hook("post_commit", dummy) [20:59] $ bzr plugins [20:59] Unable to load plugin 'dummy' from '/home/cmiller/.bazaar/plugins' [21:00] brilliantnut: do you have "GET" and "POST" installed [21:01] yes [21:01] CardinalFang: check ~/.bzr.log [21:01] it should explain why it is failing [21:01] but you should do [21:01] bzrlib.branch.Branch.hooks.install_hook... [21:01] bzrlib.branch [21:01] is a module [21:01] bzrlib.branch.Branch [21:01] is the class [21:01] brilliantnut: make sure this works [21:01] echo "hello" | POST http://localhost/path/.bzr/smart [21:01] It should return something like "ok2" [21:02] Technically, it returns [21:02] ok\012\n [21:02] well, ok\0x012\n [21:02]