[15:07] hi, any bzr gurus out there [15:08] hugh: it's better to just ask your question, and wait until somebody who knows is available to answer it [15:11] ok; I temporarily renamed a directory using mv (i.e. not using bzr), and then later I did a commit; bzr decided I intended to remove the dir [15:11] directory [15:12] so, I've tried to use revert to bring the directory back [15:12] but, I get an error message [15:12] hello [15:12] bzr: ERROR: Path(s) are not versioned: py/coe/signalprocessing [15:12] I've recently been talking to people interested in using bazaar. One question they asked was whether it had a command to totally rebuild the working tree. [15:13] something like bzr revert, except that it removes all files not in version control, even ignored ones. [15:13] so, how do I bring back a directory that bzr decide to remove [15:13] hugh: bzr revert perhaps? [15:13] two of asking about revert at the same time! [15:14] no, revert didn't work, said: bzr: ERROR: Path(s) are not versioned: py/coe/signalprocessing [15:14] hugh: oh, I've just read your full explanation. [15:15] hugh: one thing you could try is 'bzr uncommit' (if moving the directory was something you did in the most recent commit) [15:15] hugh: do this [15:15] hugh: let's say the commit where the directory was accidentally removed was 123 [15:15] bzr merge -r 123..122 . [15:16] jml: clean-tree ? [15:16] that can remove ignored files [15:16] ok data, thank, I thought that might be the recommended route [15:16] I'll give it a try [15:16] jelmer: awesome. [15:16] hugh: better mv out the directory first [15:16] hugh: bzr will restore it [15:17] oh, I have the moved dir, I just want to restore it with its history rather than re-adding it; I'll be careful [15:19] jelmer: I guess the exact thing they would want is 'bzr revert && bzr clean-tree --ignored --unknown --detritus === bac_afk is now known as bac === bac_afk is now known as bac [15:23] hugh: or, just branch locally [15:23] hugh: bzr branch foo foo-clean [15:24] hugh: foo-clean will have a rebuilt working tree, with no ignored files etc [15:24] I think you mean jml [15:26] quicksilver: oh yeah, good point. [15:26] jml: you want "bzr revert; bzr clean-tree" [15:26] (clean-tree is provided by bzrtools) [15:26] hugh: oh, you could run bzr uncommit indeed, if it was the last revision. [15:26] (and it wasn't pushed anywhere) [15:26] dato: clean-tree doesn't delete ignored files by default. [15:26] hugh: sorry. [15:27] jml: I did indeed mean jml :) [15:27] jml: well, delete-tree --ignored --unknown ? [15:28] dato: :) [15:30] hm, bzr branch would be the right answer if tree building was faster [15:44] Hi again, I just fixed a small bug in the bzr-email plugin when using smtp.gmail.com. I published my branch on launchpad. Is that enough to get my changes reviewed and hopefully going into trunk if there are no issues ? === cprov-lunch is now known as cprov [15:53] kaaloo: probably best to mail the bazaar list i'd think [15:54] ok thanks, I just sent out an email to Robert, but I'll do so next time [16:13] jelmer: ping [16:14] schierbeck, pong [16:15] jelmer: are you acquainted with the ancestry tree drawing code of the viz? [16:16] schierbeck: no, not at all [16:16] darn [16:17] i'm investigating whether it's feasible to refresh the graph when new revisions are available, or if i have to redraw the entire widget... [16:17] garyvdm is the person you'd want to talk to [16:17] yeah, but i haven't seen him on the channel recently [16:18] refreshing the viz is crucial if we want to pull and push from within the viz [16:19] hell, the current compact-view-toggle is just a hack that forces destroys the treeview, then creates a new one [16:19] *-forces [16:20] I think that's acceptable for now [16:20] even when we start doing push/pull from viz [16:55] jml: clean-tree can do it, combine the two is straight forward. Or you could write a plugin trivially. [16:57] lifeless: thanks. [17:08] well back to bed for a bit [17:27] meh [17:27] I'm awake [17:29] abentley: ah, the summary is a link, and thats why I didn't notice it as something to copy n paste from [17:33] jam-laptop: how goes dirstate-merge? [17:33] ok, i was benchmarking your patch after a merge of dev showed me just how slow things had gotten [17:34] :) [17:34] the feedback that its radically faster for you is interesting [17:34] how many packs do you have? [17:35] 11 [17:35] cool [17:35] sorry, 13 after testing [18:21] New bug: #162702 in bzr-email "Does not use bzrlib.smtp_connection" [Undecided,New] https://launchpad.net/bugs/162702 === mw is now known as mw|food [18:35] New bug: #162707 in bzr "bzrsvn: permission denied on win32" [Undecided,New] https://launchpad.net/bugs/162707 [18:37] jam-laptop: ping; can we have a short skype chat ? [18:37] lifeless: how about after 30 minutes? [18:37] I'm spinning on reconcile designs [18:37] k [18:37] so essentially blocked until I validate an approach [18:38] 30 minutes is fine; earlier is better. (But I do mean short) [18:38] I'll ping you as soon as I'm ready [18:39] thanks! [18:41] by the way, the discard_merge_parents is causing the 'b' directory to disappear [18:41] I'm not sure why yet [18:41] it is claimed to be present in this [18:41] ohkies [18:41] the b array, or the entry for b in the root array ? [18:41] the entry for b in the root array [18:41] blech [18:41] before: p [[(z[1][0][0],z[0][:2]) for z in a[1]] for a in self._dirblocks] [18:41] [[('d', ('', ''))], [('a', ('', 'a')), ('d', ('', 'b')), ('a', ('', 'rootfile'))], [('r', ('a', 'b'))], [('f', ('b', 'b'))]] [18:42] after: p [[(z[1][0][0],z[0][:2]) for z in a[1]] for a in self._dirblocks] [18:42] [[('d', ('', ''))], [], [], [('f', ('b', 'b'))]] [18:42] 'I shall not write untested helper functions' * 500 [18:42] It is a bit hard to parse [18:42] but that is the status in this tree [18:42] followed by the path [18:42] for each directory block [18:43] and you can see that the root dirblock still exists, as does the 'a' and 'b' dirblocks [18:43] but the root dirblock is now empty [19:04] lifeless: I'm logging into skype now === mw|food is now known as mw [20:03] jam-laptop: thanks! [20:03] now we just need to figure out why you woke up at 3am :) [20:04] if it was an issue of you already being awake, I think most of us would understand, but just waking up....? [20:05] jam-laptop: 4am today :) [20:05] jetlag [20:05] veryt tired at 9pm [20:05] so went to sleep [20:05] 7 hours later [20:05] tada I'm awake :) [20:06] I forgot about the daylight savings time switchover [20:08] lifeless: found it... man is it weird [20:09] you have a call for [20:09] for pos, entry in enumerate(block[1]): [20:09] ... [20:09] if len(deleted_positions) == len(block): [20:09] del block[1][:] [20:09] do you see the bug? [20:09] (len(deleted_positions) == len(block[1])) [20:12] fark [20:12] great catch [20:12] so, what is a good way to test that [20:12] dirstate specific test I think [20:12] setup a failure and call discard directly [20:12] yeah, that is what I was thinking [20:13] maybe have a few permutations of tests for _discard_merge_parents [20:13] 05:42 < lifeless> 'I shall not write untested helper functions' * 500 [20:13] lifeless: speaking of which ['I shall not write untested helper functions'] * 500 is probably better [20:13] or at least put a '\n' at the end [20:13] rofl [20:28] hello everyone [20:29] I was wondering if there is anyone here who can help me with something [20:30] try asking the question :) [20:31] I need version control for a project of mine, I'm working together with another person. We were searching for possible solutions and I found the "partner" workflow. [20:31] I just can't figure out how one can practicaly make this work :/ [20:32] well, what have you tried so far? [20:32] well, I haven't tried anything specific, I'm a little lost [20:32] sorry if this sounds too general [20:32] (and if my english isn't totaly correct) [20:32] got a url to whats confusing you? [20:33] your english is better than my french [20:33] I'm pretty sure he is talking about http://bazaar-vcs.org/Workflows [20:34] Probably explicitly: http://bazaar-vcs.org/Workflows#head-55f80fdd1cfe8503703241ae491839c943327000 [20:34] hm, yes, that's where I found it [20:34] wow thats a confusing document [20:35] I mean, I know this stuff but the pictures are hard to interpret [20:35] my problem is that I don't understand the way you can synchronize [20:36] we should like to the tutorials, or have more detail there. [20:36] time to reboot... :( bbiab [20:36] geekfish: bzr push to publish your work somewhere your other person can access, and they use bzr merge URL + bzr commit, to sync [20:37] time for some food, I'll be back in about 15; try reading the tutorials they will give you details on using bzr [20:37] the workflow is about organisation not specific commands [20:37] hm, yes, I understand [20:38] I suppose I'll have to search most of it by myself [20:38] thanks :) [20:42] morning all [20:43] hi [21:10] igc: so I'm going to nag you about doing some code too [21:11] lifeless: I'm working on bug 6700 right now [21:11] Launchpad bug 6700 in bzr "bzr diff -r 10..20 http://foo" unsupported" [High,Confirmed] https://launchpad.net/bugs/6700 [21:11] hope to have a patch out today [21:11] igc: woot! [21:11] if you want to chat about it, I'd be delighted to [21:12] thanks - I'm right so far. Made lots of progress yesterday === cprov is now known as cprov-away [21:33] Hi, I couldn't find the cvs-like workflow explained in commands. Is bzr commit; bzr push - and on the other end bzr pull the equivalent to cvs commit, cvs update? [21:34] wam: if you want cvs workflow, just use the commands cvs has - [21:34] checkout; commit [21:34] lifeless: ah - the commands are on the arrows ;) Ok - understood now. Thanks ;) [21:37] I'm using it for a year now and meanwhile it's the only vcs I'm using. I even could encourage other peope to move to bzr. Right now I'm having to use other workflows due to customer's wishes. Thanks for that great software ;) [21:43] With which software did you create the graphics? Which iconsets are these? http://bazaar-vcs.org/Workflows [21:45] probably inkscape and tango [21:53] jam-laptop: I think with a little horizontal extraction we could have just one lru class and two helpers [21:53] with less duplication [21:53] probably [21:53] jam-laptop: anyhow, I've put it up, same fileids. [21:53] it looked good enough to merge to me [21:53] I was trying to make the check fairly fast [21:53] so if you want to +1 I'll push it in. [21:56] hello, anyone can tell me how to resurrect a directory? [21:56] several revisions ago, I incorrectly removed a directory while doing a merge [21:57] so uncommit is impractical here [21:57] i want to say "please add back that directory as if I had never touched it" [21:57] flacoste: bzr revert [21:57] jam-laptop: i thought that revert only removed changed in the working dir [21:58] flacoste: 'bzr revert -r -10 directory' [21:58] hmm, the help says Revert files to a previous revision. [21:58] should do what you want [21:58] right, thanks a lot! [21:58] for whatever -r before you removed it [21:58] any quick thought on how I could find that out? [21:59] * igc food [22:00] bzr log tells me that the Path does not have any revision history === phanatic_ is now known as phanatic [22:03] hgrep would help :) [22:03] flacoste: bzr log -v | less [22:03] flacoste: look for the path you want [22:03] flacoste: or use bzr ls -r path and do bisection [22:06] lifeless: thanks bzr log -v worked [22:41] quicksilver: Cart actually uses loggerhead as its source viewing library. It's coming along quite nicely, I think. Once I'm done with schema changes, I plan to release it. [22:48] abentley: it's still taking most of your hacking time ? [23:00] lifeless: No, sadly. I don't seem to get as much hacking time as I used to. [23:09] What's the current schedule for the next version? I'm trying to decide whether it's worth trying to implement rich-root, no-subtree formats. [23:10] poolie knows more than I [23:11] I think though that as its 1.0 its better not to rush through to such a thing; 1.0 should be a matter of polish not of changing default formats [23:12] If I did it, it would be non-default, but at least the knit version would not be flagged experimental. [23:12] I'd really like to get the patch from LarstiQ [23:13] LarstiQ: btw, where is your branch; I don't care if its not finished at the moment noone else can collaborate on it [23:13] AIUI subtrees are usable with that patch. [23:13] lifeless: Well, the last time I put it up for review, no one did. [23:13] And it's quite a lot more involved than introducing a new format. [23:13] abentley: yah, it was a bad time [23:13] at least for me I was under a tad of pressure [23:14] The point being that, although it would be awesome to get it merged, it's a lot more work to get ready and to get merged. [23:15] And pragmatically, we can't afford to keep making svn converts use an experimental format. [23:19] I worry that we will have holes in our rich root no-subtrees stuff, because we only really test all or nothings [23:24] Well, that's reasonable. Obviously, I'll add it to the repository_implementation and interrepository_implementation tests. Any suggestions what else I should do? [23:26] I thought there was some discussion a while back about making subtree capability a branch flag or something, rather than a repo format. [23:29] fullermd: Yeah, and the outcome was that we should make it be determined by the working tree. I said I'd do that. I didn't. I'd rather mess with repo formats than dirstate. [23:30] Sounds like a good choice to me. Dirstate likes jumping out at me and going "boogieboogieboogie", and I'm not even poking at its innards with a knife. [23:40] abentley: I don't see that you'd have to touch dirstate [23:40] Does bazaar support any type of virtual accounts for authenticating users that are doing commits or must they be system accounts? (I'm a subversion user, but the bazaar project looks really interesting to me so I'm doing some research to see what I would do to convert my svn repos) [23:46] tanderson: Bazaar supports any authentication system that you can set up with ssh, ftp or (with a plugin) WebDAV. [23:48] Certainly there are ftp and WebDAV servers that allow you to set up authentication that's not based on system accounts, and I believe there are SSH/SFTP servers that do the same. [23:50] abentley: That is great. I know that I can do virtual accounts with my FTP server but I've never looked into doing this with ssh/sftp