[00:35] igc: ping [00:36] hi spiv [00:36] igc: I've just made a tarball for 0.92rc1 [00:36] hooray [00:37] igc: I don't have write access to escudero though... do you? I know you've been RM in the past. [00:37] igc: I've put it at http://people.ubuntu.com/~andrew/bzr-0.92rc1.tar.gz for now [00:37] yes [00:37] (and there's a .sig as well) [00:37] I'll move it across [00:37] igc: thanks! [00:37] np [00:38] few minutes ... [01:06] spiv: I have the files on esceduro now but not in the right directory :-( Permissions issue ... [01:06] Let me ping a few people [01:07] igc: Ok. [01:09] igc: if necessary I can just have the annoucement and download page point at people.ubuntu.com/~andrew/. That's not great, but at least the tarball is GPG-signed so people will know that I'm definitely the guy to blame ;) === spiv changed the topic of #bzr to: The Bazaar Version Control System | http://bazaar-vcs.org/ | Bazaar 0.92rc1 is out - http://bazaar-vcs.org/Download | Please complete the Bazaar User Survey - http://www.surveymonkey.com/s.aspx?sm=L94RvLswhKdktrxiHWiX3g_3d_3d [02:52] * spiv -> lunch [03:22] hello [03:22] igc, spiv, thanks for making the release [03:22] have a good trip when you leave [03:22] hey poolie [03:22] things are going spiffingly here [03:22] how was your trip? [03:22] excelent [03:24] pretty good [03:24] i'm about to fall asleey [03:24] asleep [03:26] sleep well [03:26] thanks; and say thanks/well done to spiv too [03:53] * igc food === chx_sleeping is now known as chx [05:19] I know I should just RTF mailing list, but what was the most recent pack format change? Just the name? [05:42] The packages for bazaar are listed in ubuntu's repositories with the following warning: [05:43] Unless you have a pressing reason to use bazaar you should use some other revision control system as upstream development has ceased. [05:43] ..this is incorrect, right? [05:43] frogduster: the "bazaar" packages are for the old TLA-derived revision control system. The current tool called "Bazaar" is packaged as "bzr". [05:45] It sounds like the package description for the "bazaar" package should probably say something about the "bzr" package, which is what people are probably looking for when they find the "bazaar" package. [05:46] Ah. Thank you. :-) If the old 'bazaar' is rarely used, it might be wise to put a disclaimer in the description for the package "bazaar" that states that it is not to be confused with the currently-developed package manager packaged as "bzr". [05:46] *nod* [05:46] Hehe. [05:46] Thanks for the info. [05:47] Should I file a bug report / wishlist item on launchpad? [05:48] frogduster: I think so. [05:48] frogduster: assuming it hasn't already been filed :) [05:48] *nod* will do. [05:48] :-) natch. [05:48] frogduster: https://bugs.edge.launchpad.net/ubuntu/+source/bazaar/+bug/143998 looks to be related [05:48] Launchpad bug 143998 in bazaar "rename 'bazaar' package to 'baz'" [Low,New] [05:49] ah. [05:49] You beat me to the search.. :-) But, that's not too bad, as my system is bogged down right now, and firefox is a bit much on top of what it's already doing.. [06:01] Nudged the existing bugreport. :-) [06:01] Thanks, spiv. :-) [07:11] let it be known that I just tested the new pack support of 0.92rc1 and at least for me it didn't work as well as the old one in my working environment. I hit one "lock" error during a bzr+ssh push (which can be part of multiple simultaneous requests) [07:11] anyway, I'e decided for now to continue to use the old format as it seems to be a little more fool proof with regards to locking in my environment. [07:18] dewd__: if you've got time, you should send that as an email to the bazaar mailing list [07:21] it would be more helpful i know. but I have just reverted it back because i was testing it in "production" with a bunch of repos, not one or two. [07:21] so i can't even test/debug it [07:43] Sure [07:59] night all [09:56] hiya [09:56] why does bzr.dev identify as 0.90.0? [09:57] it breaks my newest and greatest bzr-svn [09:57] ooh [09:57] bzrlib: /usr/lib/python2.5/site-packages/bzrlib [09:57] I guess that's why [09:58] how do I make it not use the systemwide bzrlib without uninstalling bzr? [09:58] (the packaged one I mean) [10:01] futzing with python's module path is about all I can suggest [10:02] I've never had that confusion happen, and I use bzr.dev all the time on my workstation... [10:02] I tweaked PYTHONPATH and it appears to work [10:05] hmm, now it doesn't [10:30] bzr bzr-pokerolymp2:2101/> missing --other [10:30] Using last location: svn+https://beta.aimido.de/svn/src2/trunk [10:30] Unhandled error: [10:30] character mapping must return integer, None or unicode [10:30] humm [10:31] I hacked PYTHONPATH into ~/bin/bzr.dev/:$DEFAULTPATH [10:32] hmm, it seems it's bzr-svn that's dying [10:32] jelmer: poke? [10:34] or anyone else, any ideas what might be breaking it? [10:46] argh, bummer [10:46] I need bzr-svn working === bac_afk is now known as bac [11:45] mathrick: pong [11:45] mathrick: can you set BZR_PDB=1 and try again? [11:45] yes [11:46] I just need to switch to 0.92 again [11:49] jelmer: hmm, I merged that branch with upstream using r716 in the meantime, and it doesn't seem to die anymore [11:51] but I've got other fun [11:53] jelmer: http://pastebin.org/6321 [11:55] can you print the contents of elements and existing_elements ? [11:57] (Pdb) p elements [11:57] ['mathrick', 'pokersource'] [11:57] (Pdb) p existing_elements [11:57] [] [11:57] (Pdb) [11:57] jelmer: there [11:58] hmm, but that path (/mathrick/pokersouirce) exists in the repository ? [11:58] jelmer: no, mathrick/ does [11:58] I believe [11:58] uh-huh [11:58] nope [11:58] I wonder why, I created it before [11:59] at least /mathrick will have to exist [11:59] the error should probably be a bit clearer... [11:59] indeed [12:01] jelmer: btw, with 716 it was uploading a lot, and then at the end died with "at least one property change failed, repository unchanged" from subversion [12:01] r716 I mean [12:01] now it did with r761 as well [12:02] jelmer: http://pastebin.org/6322 [12:03] mathrick: can you run with -Dcommit and try again? [12:03] that should print any revision properties set to ~/.bzr.log [12:03] jelmer: that goes as a bzr argument? [12:03] yes [12:08] jelmer: http://pastebin.org/6323 [12:09] ahh [12:11] in ~/.bazaar/subversion.conf, please look for list-QlpoOTFBWSZTWZkHsqUAAAhRgAAQABC6yR4AIAAxTAAAlRkDRtJDWcJGukaEIjw7FScqiDvxdyRThQkJkHsqUA== and change the two '=' at the end into '-' [12:12] oookay [12:13] if that doesn't help, can you please mail me the bit you just pastebinned ? pastebin seems to have a X-character limit [12:14] jelmer: TypeError: Incorrect padding [12:14] list-QlpoOTFBWSZTWZkHsqUAAAhRgAAQABC6yR4AIAAxTAAAlRkDRtJDWcJGukaEIjw7FScqiDvxdyRThQkJkHsqUA-- [12:14] that's what it should look like? [12:14] whoops, that should be two dots at the end [12:14] sorry [12:15] jelmer: still the same with .. [12:16] can you set BZR_PDB=1 again and see where it's breaking? [12:17] **** entering debugger [12:17] > /home/mathrick/Dev/rails/bzr-pokerolymp2/base64.py(76)b64decode() [12:18] and "bt"? [12:19] http://pastebin.org/6326 [12:20] are you sure you're running r761? [12:20] the source for scheme.py:133 is different here [12:22] lemme confirm [12:23] jelmer: it seems that bzr revert on a lightweight checkout will confuse bzr [12:24] gotta report that [12:24] now I've got that mapping thing again [12:24] mathrick: but you were actually running a different version? [12:24] mathrick: which mapping thing? [12:24] jelmer: http://pastebin.org/6329 [12:25] jelmer: yeah, it *said* 761, but was running 716 still I guess [12:26] see /query [12:33] moin [12:34] hey lifeless [12:34] how's the UDS? [12:41] New bug: #158306 in bzr "Reverting a lightweight checkout confuses bzr about revision number" [Undecided,New] https://launchpad.net/bugs/158306 [12:41] good [12:41] jerry is playing with packs [12:45] ^-- lightweight checkouts have very unclear idea of their own status [12:46] well, revert does not change revno [12:47] but you can't change revno on a lightweight checkout otherwise [12:47] since it doesn't have its own identity [12:47] Well, that's what update -r is for. Or would be, if it existed. [12:48] mathrick: well, it does have it's own identity [12:49] lifeless: but tries very hard to hide it :) [12:49] revert does not change it, and is not intended to change it; as fullermd says update is the command that would be suitable for changing it [12:49] mathrick: uhm, not really ;) [12:49] lifeless: Nice :-) [12:50] I wish heavyweight checkouts tried harder to hide it... [12:50] lifeless: try rebinding a lightweight checkout for which the parent branch is inaccessible [12:50] it's not possible [12:50] it's not possible at all by default [12:50] but the plugin to do that fails with inaccessible parents [12:50] mathrick: 'bind' only applies to heavyweight checkouts; I think 'switch' is the right ui for it, but its not done yet [12:51] mathrick: well you have found a bug then :) [12:51] yeah, switch was what I was thinking about [12:51] I know :) [12:51] I thought it's been reported already [12:51] *think [12:51] jelmer: anyway, ping me when there's hot new svn love waiting for me to update :) [12:52] mathrick: The unicode bug should be fixed already [12:52] now looking into the missing prefix error [12:56] jelmer: ah, fine, lemme try that [12:58] jelmer: where do I change the branching scheme again? [12:58] (it doesn't think that mathrick/pokersource/ is a valid branch path) [13:00] mathrick: just remove ~/.bazaar/subversion.conf [13:00] jelmer: oh, it doesn't need the branch specs anymore? [13:01] crap [13:01] it does, but will create it again [13:01] I deleted bazaar.conf instead [13:02] Just goes to show you why you should keep ~/.bazaar in bzr ;) [13:02] mathrick@hatsumi:~/Dev/rails/bzr-pokerolymp2$ BZR_PDB=1 bzr -Dcommit svn-push svn+https://beta.aimido.de/svn/src2/mathrick/pokersource/ [13:02] bzr: ERROR: These branches have diverged. Use the merge command to reconcile them. [13:02] gah [13:03] jelmer: is that related to me having just deleted bazaar.conf? [13:03] mathrick, no, it's shouldn't be [13:03] mathrick, You created just mathrick? [13:03] yeah [13:03] hmm [13:04] but somehow pokersource/ popped up there [13:12] jelmer: see priv [13:23] mathrick: can you still reproduce bug 145171 or is that what we're looking at right now? [13:23] Launchpad bug 145171 in bzr-svn "Invariant break on bzr missing in bzr-svn" [Medium,Incomplete] https://launchpad.net/bugs/145171 [13:24] jelmer: unsure, I looked at it earlier today, but I decided I can't determine if it still happens [13:24] it doesn't look the same [13:25] but you were trying to do the same thing there? [13:27] jelmer: no, I was trying to push to a new branch [13:27] jelmer: oh, you might want to know that I have a merged-into subtree in that repo [13:29] in the bzr repo? [13:29] that you're trying to push? [13:34] jelmer: yes [13:35] mathrick: also in the particular branch you're pushing? [13:36] jelmer: umm, unsure what you mean now [13:37] the pokersource branch you're trying to push, does that include a revision with subtrees? [13:37] the repo is plain SVN, so it doesn't have merged subbranches of course. My branch is derived from that SVN repo trunk, AND merges-into a subbranch, now I want to push the whole thing under branches/mathrick/pokersource/ [13:38] jelmer: what will it take to get rid of bzr svn-push ? [13:40] lifeless: bug 121875 [13:40] Launchpad bug 121875 in bzr "cmd_push() should abstract away transport.mkdir()" [Wishlist,New] https://launchpad.net/bugs/121875 [13:43] lifeless: svn-push is basically a wrapper around the BzrDir.import_branch(path, source_branch) function [13:54] jelmer: so send a patch :) [14:00] lifeless: lots of other things I have to do first, I'm afraid :-/ === cprov is now known as cprov-lunch [14:06] jelmer: it does seem very likely to trip people up; and I don't understand why you don't just decoate the existing push command in bzr-svn as a firststep [14:09] lifeless: I'm focussing on fixing some of the functionality bugs in push that have come up for now [14:11] Hi - does anyone have a moment to answer a question about "bzr status"? [14:12] jelmer: fair enough [14:12] Just ask your question; don't ask to ask [14:12] mrevell: don't ask to ask, just ask. [14:13] * lifeless high fives LeoNerd [14:13] fair enough :) [14:13] Asking to ask asks for axeing. [14:13] I want to know what the Subversion-style status flags are but can't find a definitive list for Bazaar [14:14] can anyone point me in the direction of definitions? [14:14] bzr help st [14:14] ... [14:14] See also: diff, revert, status-flags [14:14] lifeless: Thanks, managed to miss that. [14:19] lifeless: so send a patch :-P [14:19] mrevell: suggestions on making it more clear solicited [14:20] jelmer: fixing performance for e.g. samba first :) [14:20] lifeless: Working on it right now. [14:20] lifeless: ok, I can live with that :-) [14:21] lifeless: Did you discuss the difference in workflow compared to git with Jerry? [14:21] yes somewhat [14:21] (one working tree per repository compared to one per branch) [14:21] big thing appeasrs to be our lack of 'collectionof branches' facilities [14:21] yeah, I blogged about that a while back [14:21] Coming from CVS, I'd say that's true === bac_afk is now known as bac [14:26] If only somebody could rant via email at length on that topic... [14:27] jelmer: I don't see blogs that often; discussion about features is really best on the list IMNSHO :) [14:28] Well, the problem I guess is that it's not a concrete feature request [14:28] more like: these things bother me about bzr at the moment and they all seem to be related to management of branches [14:28] *related branches [14:28] I can forward the blog post to the list, would htat be useful? [14:29] here's a link: http://jelmer.vernstok.nl/blog/archives/192-Bazaar-Need-for-a-Product-object.html [14:30] It was ref'd by igc in the recent thread on the topic. [14:31] jelmer: do you know about path_policy=append ? [14:31] fullermd: :) [14:31] things [14:35] lifeless: no, and cna't find any reference ot it in the bzr source [14:35] with apologies for my bad spelling today [14:35] e.g. [14:35] public_branch:policy = appendpath [14:36] for instances... I have: [14:36] [sftp://bazaar.launchpad.net/%7Ebzr/] [14:36] public_branch = http://bazaar.launchpad.net/~bzr/ [14:36] public_branch:policy = appendpath [14:36] create_signatures = always [14:36] post_commit_to = bazaar-commits@lists.ubuntu.com [14:36] Hmm.. What's that do? [14:37] lifeless: ah, that would at least help with the configuration bit [14:37] but still not allow pushing multiple related branches at once, for example [14:38] LeoNerd: it makes all branches unter that path be given a public location of the public_branch component plus the relative path under the sftp root === bac_afk is now known as bac [14:38] lifeless: also, is this settable in the repository config? [14:39] i don't use locations.conf because it means every time I rename a branch or one of it's parent directories I lose configuration [14:39] I can checkout but not commit to an SVN repository with a self-signed certificate. The latter gives a pycurl error [14:39] gotgenes: what version of bzr-svn are you running? [14:39] I don't think we _have_ a repository config... [14:39] 0.91 with v 0.34 bzr-svn [14:40] gotgenes: not 0.4.3 ? [14:40] I doubt bzr-svn 0.3.4 works with bzr 0.91 [14:41] gotgenes: this is bug 150699, which has been fixed in bzr-svn's 0.4 branch [14:41] Launchpad bug 150699 in bzr-svn "Cannot commit on svn+https checkout" [Low,Fix committed] https://launchpad.net/bugs/150699 [14:41] jelmer: meant v 0.4.3 [14:41] dyslexia [14:41] gotgenes: ok, in that case updating to the 0.4 branch (or 0.4.4, when it's released) should fix that issue [14:42] jelmer: ah, okay [14:42] 0.4.4 will be released later this week [14:42] jelmer: should I just bzr co your branch? [14:43] gotgenes: Yep, just checking it out and linking ~/.bazaar/plugins/svn to it should be sufficient [14:44] jelmer: great! and then it's bzr pull to update from your changes every now and then? [14:44] gotgenes: if you use "bzr up" if you've "bzr co" earlier [14:44] "bzr pull" if you used "bzr branch" earlier [14:44] jelmer: ah, okay [14:45] jelmer: no, its locations.conf - not coupled to repo's [14:45] most confusing thing in the bzr UI at the moment, imho [14:47] anyone know a workaround for bug 107593? [14:47] Launchpad bug 107593 in bzr "bzr unable to ask password for access over bzr+ssh:// or sftp:// when plink.exe used as SSH client" [Low,Confirmed] https://launchpad.net/bugs/107593 [14:48] I literally can't use bzr on win32 to push branches [14:48] I'd consider it a showstopper for win32, but then again, I guess much less people use plink than I thought [14:51] jroes: any reason you can't use paramiko? === mw|out is now known as mw [15:01] well, tbh, I simply downloaded a binary dist of bzr, and I'm guessing paramiko is just bundled in there somewhere [15:01] I don't honestly know what paramiko is besides "ssh2 for python" :) [15:01] jroes: SSH and SFTP for Python. [15:06] jroes: it is a bit of a showstopper for plink users; I guess we don't have all that many, or they are using the putty agent [15:07] plink is the putty agent? [15:07] oh, pageant is [15:07] pageant is the putty agent [15:07] I don't know the binary names, I just click the pretty icons :) [15:08] I think the mistake is that we use plink if we can find it, even though paramiko works better :) [15:08] so, I guess I'm confused, I just use pageant, something else makes the decision as to whether or not to use paramiko or plink [15:08] but I'm not the one making that decision, and I don't know how to choose :) [15:08] we see if we can find plink in the path [15:08] ah [15:09] I don't know that we have a way to disable it .... [15:09] I guess the assumption is that if it is available you have it configured [15:09] ah ha [15:09] certainly both paramiko and plink will talk to pageant [15:09] ok, removing plink from path :) [15:10] that should do it :) [15:10] yes, it sure did, awesome :) [15:11] thank you very very much :) [15:11] You can override the choice of SSH bzr uses with an environment variable IIRC. [15:12] set BZR_SSH=paramiko [15:12] I think that would do it. [15:13] excellent. I'd probably use that if I needed plink :) [15:13] (It'll be fractionally faster too, because it won't need to autodetect which SSH to use) [15:14] How much does paramiko hurt bzr's SSH performance? [15:16] Peng: probably only slightly; I believe most of the heavy lifting is done in the PyCrypto module, which is written in C. [15:16] Hmm, pycrypto is a good point. [15:18] is there any way to have bzr use xxdiff to compare changes instead of the regular diff program? [15:20] pattern: the difftools plugin IIRC [15:21] ok, cool [15:21] thanks [15:58] New bug: #158350 in bzr-eclipse "import wizard does not display" [Undecided,New] https://launchpad.net/bugs/158350 [16:17] New bug: #158333 in bzr "bzr-rebase-0.2 crashes with bzr-0.92rc1" [Undecided,Confirmed] https://launchpad.net/bugs/158333 [16:20] Can anyone tell me where bzr help draws the "usage" line from? [16:21] mrevell: it creates them by looking at attributes of the command. [16:22] thanks james_w. [16:26] mrevell: it generally looks at "takes_args" and "takes_options" [16:26] though takes_args is the one to generate Usage [16:27] mrevell: bzrlib.commands.Command._usage() [16:27] jam-laptop: Thanks. [16:27] Command.get_help_text() is what you would change [16:28] if you wanted to move the SeeAlso stuff [16:28] or bzrlib.help_topics.help_as_plain_text() [16:28] no, it is the get_help_text() [16:28] I was just confused by the if blocks [16:29] jam-laptop: hi. lets talk pack2 [16:30] lifeless: sure [16:30] It is always fun to see you on at the same time as I am [16:31] :) [16:31] so, zlib, I think thats kinda interesting to do (but the current packs format is frozeN) [16:32] I'd really like to see xdelta/mutiparentdiffs/arbitrary knit diffs though [16:33] mrevell: I'll reply to you with a simple patch [16:33] lifeless: I'm curious if the 10% zlib difference would scale for first commit [16:33] mrevell: Thanks. I may need to ask what seem like simple questions though :) [16:35] Is a branch allowed to have more than one root commit? (i.e. commit with no parents) [16:35] jam-laptop: easy enough to test; once you have it in a different format we can convert moz to it :) [16:36] james_w: yes [16:36] mw: are you really talking to yourself? [16:36] lifeless: my branch has it in the zlib format at the moment [16:36] james_w: You can merge 2 ancestries together [16:37] so you can get two root commits [16:37] james_w: Sure, I do it all the time. [16:38] lifeless: what's that? [16:39] thanks. I guess I set parent_ids to an empty list, and last_revision_info to (0, NULL_REVISION). [16:39] mw: lifeless was actually trying to point out to mrevell that he wrote "mrevell:" rather than probably replying to me [16:39] I'll be right back, we're grabbing foodstuffs [16:39] Lordy [16:40] mw: he meant to address mrevell [16:40] people confuse us all the time [16:40] (that is a lie) [16:41] Oh, you can't trust any of those 'm' people anyway. [16:43] mw: I meant mrevell === cprov-lunch is now known as cprov [17:01] I'm back [17:02] jam-laptop: so, where is your branch? I'll convert lp to your new format [17:03] lifeless: http://bzr.arbash-meinel.com/branches/bzr/0.92-dev/knit_parents [17:03] If you prefer, I could push it to LP or something [17:04] thats fine [17:04] does it affect knitpack-experimental, or is it a new repo format ? [17:22] The help for bzr inventory has a line that reads, "It is also possible to restrict the list of files to a specific set. For example: bzr inventory --show-ids this/file" Can anyone explain "this/file" to me, please? [17:23] mrevell: if it is "this/" then it will only show files/directories underneath this/ [17:23] I presume it means the path of a file/subdir in the branch... [17:23] mrevell: try it :). I think its just a filter on the path. [17:23] if it is "this/file" then it would only show this/file [17:23] (so I'm pretty sure it is 90% useless to give it a filename, but there are edge cases where it could be useful) [17:24] When I tried it I got a "paths are no versioned" message. [17:24] s/no/not [17:25] Works for me with .dev and 0.91. [17:26] fullermd: Me too now, I made a typo. Cheers. === thumper_laptop is now known as thumper [17:31] jam-laptop: so is it a new repo format or does it alter knitpack-experimental [17:31] lifeless: it alters knitpack-experimental, but uses a different format string on disk [17:31] Which means you can't pull from a pack into a pack [17:31] but it was reasonable to experiment with [17:31] mrevell: well was "this/file" versioned? [17:31] ok, so I won't overwrite my devpad test branch :| [17:32] good idea :) [17:32] jam-laptop: In the end, yes, once I understood properly and also corrected a typo. The use of "this" threw me a bit. === thumper_laptop is now known as thumper [17:45] lifeless: the argument for packs is still --knitpack-experimental, is it a good idea to upgrade yet? [17:45] jelmer: knitpack-experimental is frozen now [17:46] jelmer: however it's not suitable for uninformed use because of e.g. obsolete_packs, slow annotate[no cache], etc [17:46] what are the chances of corruption? [17:47] reconcile your knits branch before you convert [17:47] after that it should be fine [17:47] I've been dogfooding for a couple of months, no corrpution [17:47] I'm considering moving my Samba branches to packs, but would rather not lose any data in the process :-) [17:47] here is the acid test: [17:47] ok, I'll give it a go then. most of my code is mirrored in svn anyway [17:47] convert to packs, then push to a new pack branch [17:48] if that works (and it will if you've reconciled first), then no problems [17:48] read the docs :) (doc/developer/knitpack.txt) [17:48] 'k, here's a random thing that's bugged me for a little while... [17:48] Why does 'reconcile' exist? [17:48] Which is to say, "Why isn't there 'check --fix-broken-crap'?" [17:49] 'check' implies readonly in the name. [17:49] in terms of code, we should probably have Reconcile/Reconciler --check-only [17:49]