[01:14] <db-keen> I never noticed this before: I just branched using bzr-svn and it reported branched through revision 26 of a repository at revision 25. Is it always +1 ?
[01:34] <Odd_Bloke> db-keen: Are you sure that there aren't 26 revisions somewhere in the repository?
[01:34] <Odd_Bloke> I'm just heading to bed, but that'd explain it.
[01:34] <Odd_Bloke> Anyhoo, jelmer is the man to ask.
[01:34] <Odd_Bloke> Night!
[01:35] <Jc2k> db-keen: are there 26 in svn. and 25 in bzr?
[01:35] <Jc2k> (sorry, its later i can't tell if i'm ready it right :P)
[01:35] <Jc2k> gah
[01:35] <Jc2k> reading it right, evn
[01:36] <db-keen> Odd_Bloke, Jc2k: yeah, I've checked and double checked, 26 in bzr, 25 in svn
[01:37] <Jc2k> i think if there is a branching operaion (e.g. svn cp from to) then the revision numbers will be different
[01:37] <lifeless> svn revnos are repository wide
[01:37] <lifeless> bzr revnos are  branch wide
[01:37] <lifeless> you could for instance have 2000 commits in an svn revno, but one branch with only 10, if you bzr branch that bzr will show 10
[01:38] <db-keen> lifeless: but I have an extra revision in bzr, not the other way around
[01:38] <lifeless> oh
[01:38] <lifeless> offhand, no idea.perhaps svn is 0-based?
[01:40] <slangasek> not in my experience, it isn't
[01:41] <Toksyuryel> pretty sure bzr adds the +1 whenever you branch
[01:41] <db-keen> perhaps it does, I just never noticed
[01:42] <Toksyuryel> I think that is what I read in the docs, not completely sure I remember it correctly though
[01:42] <lifeless> 'branch' isn't an action for bzr - two branches one made from another with no commits, have the same revnos
[01:43] <lifeless> branch is an action for svn, but it also adds a revno
[01:44] <lifeless> perhaps its an svn copy to make a branch?
[01:46] <slangasek> that's correct, but how does that fit?
[01:49] <jelmer> 'evening
[01:50] <jelmer> db-keen: if you branched the root of the svn repository, that's correct
[01:51] <db-keen> jelmer: ah, perhaps that's why I never noticed, it's much more common to just branch trunk
[01:52] <Toksyuryel> oh wait I am thinking of merge
[01:52] <Toksyuryel> when you merge the revno is bumped
[01:57] <radix> Toksyuryel: technically, it's when you commit
[02:00] <Toksyuryel> radix: when you commit obviously, but I was thinking of other times the revno is bumped; a merge is one of those times
[02:00] <radix> Toksyuryel: no, it's only when you commit.
[02:00] <radix> Toksyuryel: Just running "bzr merge" doesn't actually modify the branch - it doesn't add a revision
[02:00] <radix> You have to "bzr commit" the merge for it to actually be added to the branch, in the form of a new revision
[02:35] <Keybuk> radix: though it'd be nice if merge did commit ;)
[02:35] <radix> weirdo :P
[02:35] <lifeless> you kike unreviewed code that much?
[02:35] <lifeless> s/k/l
[02:36] <radix> lifeless: I recommend dvorak
[03:22] <chandlerc> anyone know exactly how bzr-svn branches are "supposed" to work?
[03:22] <jelmer> chandlerc, how do you mean?
[03:22] <chandlerc> ie, the whole subversion branching scheme
[03:22] <jelmer> chandlerc: See "bzr help svn-branching-schemes"
[03:22] <chandlerc> i've gotten a weird error with my bzr-svn repository, and i'm wondering if i just set it up incorrectly... (sorry for the bounce.. dunno what my internet is doing)
[03:22] <chandlerc> i just read that
[03:22] <chandlerc> ;]
[03:23] <chandlerc> but how should it look from the bzr side?
[03:23] <jelmer> if you'd just like to check out one branch in svn, branching schemes shouldn't be erelevant at all
[03:23] <chandlerc> currently, all development is done on mainline, but I'm sure that'll change down the road
[03:24] <chandlerc> do you just arrange the paths locally of your bzr branches the same way as the subversion branches are?
[03:24] <chandlerc> or does it only look at the url you push to?
[03:24] <jelmer> there's no requirements as to how you arrange it locally
[03:25] <chandlerc> jelmer: here is the error i'm getting: http://pastebin.org/42285
[03:26] <jelmer> chandlerc: That's unrelated to branching schemes
[03:26] <jelmer> You seem to be trying to create a new branch in svn
[03:26] <chandlerc> ok... i'd still like to understand them, but not right now
[03:26] <chandlerc> and i'm not, i'm running "bzr push"
[03:27] <jelmer> but you're pushing into a svn repository?
[03:27] <chandlerc> yes, but its one i've been pushing too
[03:27] <chandlerc> to too. ;]
[03:27] <tethridge> anybody heard of a clear case plugin for bzr?
[03:28] <jelmer> chandlerc: but the location you're pushing to doesn't contain a branch yet apparently
[03:28] <jelmer> chandlerc, You may have to use "bzr svn-push" rather than "bzr push"
[03:28] <jelmer> since the branch doesn't exist yet
[03:28] <chandlerc> i'm very certain it does
[03:28] <jelmer> bzr is trying to run mkdir since it can't find that directory
[03:28] <chandlerc> Related branches:
[03:28] <chandlerc>     push branch: svn+https://inc.googlecode.com/svn/parser/trunk
[03:28] <chandlerc> ...
[03:28] <chandlerc> # bzr push
[03:28] <tethridge> what is the technical difference between pull and merge?
[03:29] <chandlerc> its the same URL i've been using that command with for 2 months now
[03:29] <lifeless> pull follows one path in the dag
[03:29] <lifeless> merge combines two branches that are separate in the dag but started from some common point
[03:30] <tethridge> does pull autocommit?
[03:30] <lifeless> nothing autocommits
[03:30] <tethridge> to the local branch
[03:30] <tethridge> ok
[03:30] <lifeless> pull replaces the current branch tip
[03:30] <lifeless> but no new commit object is created
[03:30] <chandlerc> jelmer: its one of the more bizarre errors because it started happening without anything changing (that I know of...)
[03:31] <jelmer> chandlerc: What does "bzr branching-scheme svn+https://inc.googlecode.com/svn/parser" return?
[03:31] <tethridge> I find it strange that when I "update" my local mirror of a project that I commit the change
[03:31] <tethridge> I've just been entering a "updated from trunk" message.  Is that what you guys do?
[03:31] <chandlerc> unknown command
[03:32] <chandlerc> jelmer: btw, i'm following the 0.4 release branch, not using fixed releases
[03:32] <jelmer> chandlerc: 'bzr svn-branching-scheme..."
[03:32] <chandlerc> jelmer: Not a branch
[03:33] <chandlerc> for both parser and parser/trunk
[03:33] <lifeless> tethridge: if you have a local mirror, why are you using update & not pull ?
[03:34] <tethridge> well, to be honest, I was confused by whether I should do a merge or a pull.
[03:34] <tethridge> I guess I should be doing a pull
[03:34] <jelmer> chandlerc: hmm, google branch appears to be disallowing operations on the repository root
[03:34] <tethridge> I'm not sure why I need a local mirror
[03:34] <tethridge> why not just have a task branch for each task?
[03:34] <lifeless> tethridge: sure, you can do that
[03:34] <tethridge> is it just a speed thing?
[03:35] <chandlerc> jelmer: would this have been a change on their side to break this, or in how bzr-svn is doing things?
[03:35] <brandon_rhodes> Is there an easy way to have bzr give me a "bzr log" output with diffs included?
[03:35] <tethridge> so if I'm not connected I could create task branches
[03:35] <chandlerc> jelmer: or in how i'm doing things w/ bzr-svn
[03:35] <brandon_rhodes> Or do I have to run "bzr diff" over and over and over?
[03:35] <tethridge> so nobody knows of a clear case plugin?
[03:36] <lifeless> tethridge: well, I don't know why you are doing $whatever you are doing :P
[03:36] <tethridge> I think I have it figured out now
[03:36] <tethridge> it takes a little bit to switch modes from using something like svn
[03:36] <lifeless> brandon_rhodes: I don't think there is at the moment
[03:36] <lifeless> tethridge: so I think a more useful pull/merge/update explanation for you is:
[03:37] <jelmer> chandlerc: I'm not sure
[03:37] <lifeless> 'pull' is used to maintain an exact copy of another branch.
[03:37] <lifeless> 'merge' is used to combine someone elses work into your branch
[03:37] <jelmer> chandlerc, you may want to run "bzr selftest --starting-with=bzrlib.plugins.svn" to make sure your checkout of bzr-svn works ok
[03:37] <brandon_rhodes> lifeless: drat.  Thanks for the answer, though. :-)
[03:38] <lifeless> 'update' is used when two people are sharing a branch (and thus works like update in cvs/svn)
[03:38] <chandlerc> jelmer: no such option: --starting-with
[03:38] <jelmer> chandlerc: s/--starting-with=bzrlib.plugins.svn/^bzrlib.plugins.svn/
[03:38] <tethridge> lifeless, I don't really see a difference between pull and update
[03:39] <tethridge> I see what you mean between pull and merge
[03:39] <Keybuk> tethridge: pull is different if you branch, instead of checkout
[03:39] <Keybuk> it pulls changes from one branch into another
[03:39] <lifeless> tethridge: say there is a branch at URL
[03:40] <lifeless> tethridge: no working tree, just a branch.
[03:40] <tethridge> ok
[03:40] <lifeless> tethridge: to work on this you can either make a new branch, hack, and then integrate your work into the branch (via push, or by doing a checkout+merge+commit)
[03:41] <lifeless> tethridge: or you can work directly on the branch, by doing a checkout
[03:42] <lifeless> in the latter case you get a local working tree, and unless you specified --lightweight, a mirror of the branch, but bzr keeps it synchronised with URL, any change made locally is made to URL too
[03:42] <lifeless> if two people both do 'bzr checkout URL' and then one of them commits
[03:42] <lifeless> the other one can't commit - because their local working tree does not contain the changes introduced by the others commit
[03:42] <lifeless> to get those changes they do 'bzr update'
[03:43] <tethridge> I see
[03:43] <lifeless> which synchronises their local working tree version with the current branch tip of URL
[03:43] <tethridge> which workflow do most projects use?
[03:43] <lifeless> (and if there is a local mirror branch object, it is synced too). Finally, if they had done an offline commit in the checkout, it gets turned into a pending merge.
[03:44] <tethridge> that makes sense
[03:44] <tethridge> I'm almost finished reading the tutorial, but that helps
[03:44] <tethridge> not tutorial, user guide
[03:44] <lifeless> generally the gatekeeper for a project - the person running trunk - does all the commits to the trunk
[03:45] <lifeless> if there are multiple gatekeepers, then they will have a shared branch -- s ingle branch that the gatekeepers all have checkouts of
[03:45] <chandlerc> jelmer: i'm really hoping to get a good setup going w/ googlecode btw, hoping to get a not insignificant blog post written up about it
[03:45] <lifeless> features are normally done on different branches (made by doing 'bzr branch')
[03:45] <chandlerc> jelmer: to that end, if there is something that needs to be tweaked on their side, let me know! =]
[03:46] <tethridge> I remember reading recently on planet ubuntu about launchpad having something like bugbuddy.  For some reason I thought it was something different though.  You guys know what I'm speaking of?
[03:46] <jelmer> chandlerc: no, I think there's nothing that has to be changed on that side
[03:46] <tethridge> I can't find the post now
[03:46] <lifeless> chandlerc: I saw a blog from a googler about git-with-gcode
[03:46] <jelmer> chandlerc: To be sure you may want to check with a released version of bzr-svn
[03:46] <lifeless> chandlerc: I found that amusing knowing what bzr-svn can do :P
[03:46] <chandlerc> ;]
[03:46] <tethridge> the blog post was mentioning how the conversation about whether or not to approve the bug was in launchpad, which sounds like what bugbuddy does
[03:47] <lifeless> bpeterson: hi
[03:47] <chandlerc> lifeless: i'm going to try and do bzr-svn justice, although my usage is light so far, i'm hoping it'll get increasingly heavy
[03:47] <lifeless> bpeterson: I'm in sydney, ETIMEZONE last night
[03:47] <jelmer> chandlerc: Cool
[03:47] <bpeterson> lifeless: regarding the TooManyConcurrentRequests on ssh error: #125784
[03:47] <lifeless> yes
[03:47] <jelmer> chandlerc: I'm happy to help trying to get it to work for you
[03:47] <chandlerc> =D
[03:48] <chandlerc> it's been working great until the other day when this error started showing up
[03:48] <chandlerc> quite odd
[03:49] <bpeterson> lifeless: I'm in the US
[03:51] <lifeless> anyhow
[03:51] <lifeless> I don't have new info for you
[03:51] <lifeless> have you tried andrews patch?
[03:53] <bpeterson> no
[03:53] <bpeterson> I hope I can merge it cleanly!
[03:53] <chandlerc> jelmer: http://pastebin.org/42288
[03:54] <lifeless> jelmer: what does openchange use for text indices
[03:54] <chandlerc> i don't think any of those are my problem...
[03:54] <jelmer> lifeless, what would we need text indices for ? We don't have a server yet :-)
[03:55] <jelmer> chandlerc, that looks ok
[03:55] <lifeless> jelmer: for answering client folder searches
[03:55] <chandlerc> jelmer: btw, bzr pull fails with the Not a branch error, but no traceback, its just the push that crashes
[03:56] <chandlerc> jelmer: but i'm really mystified as it clearly is a branch... the parent and push branch...
[03:56] <jelmer> chandlerc, please try running with -Dtransport
[03:56] <chandlerc> k
[03:57] <chandlerc> which will provide the best information? the push or the pull? or the svn-branching-scheme?
[03:57] <jelmer> lifeless: I don't think we do search yet
[03:58] <lifeless> k
[03:58] <chandlerc> jelmer: err... nm, none of those commands gave any additional output wth -Dtransport
[03:58] <jelmer> chandlerc, it's writting to ~/.bzr.log
[03:59] <chandlerc> doh, my bad
[03:59] <chandlerc> i remember than now of course... ok, which one would you like, or all of them?
[04:00] <chandlerc> http://pastebin.org/42290
[04:00] <chandlerc> that's push, and pull...
[04:03] <jelmer> chandlerc, can you try "bzr pull svn+http://inc.googlecode.com/svn/parser/trunk" ?
[04:05] <chandlerc> http://pastebin.org/42292
[04:09] <jelmer> without the s
[04:09] <chandlerc> doh
[04:09] <chandlerc> sorry
[04:09] <chandlerc> missed that
[04:09] <chandlerc> yup, that worked
[04:09] <chandlerc> is it my subversion?
[04:11] <jelmer> I suspect a SSL certificate expired or something
[04:11] <chandlerc> mm, lemme try doing an https something via svn
[04:11] <jelmer> python-subversion doesn't pass that information on yet but just returns an error
[04:12] <chandlerc> bingo
[04:12] <chandlerc> wow, so simple...
[04:13] <chandlerc> ahhh, now i get a much more interesting failure
[04:13] <chandlerc> so how would i set two different branching schemes for two different "modules" in my subversion repo?
[04:14] <chandlerc> i have svn/parser/{trunk,branches,tags}, but also svn/wiki which doesn't have the {trunk,branches,tags} scheme
[04:14] <chandlerc> when i 'bzr push' to svn/parser/trunk, it inspects the svn/wiki area, and errors because branches is missing...
[04:15] <jelmer> you can't use a different branching schemes within the same repository
[04:15] <chandlerc> that's... a problem for googlecode
[04:16] <jelmer> chandlerc: but every project has its own repository
[04:16] <chandlerc> yes, but every repository has a "wiki" directory
[04:16] <chandlerc> with no trunk, branches, or tags
[04:16] <bpeterson> lifeless: Andrew's patch works for me! Can we get it applied?
[04:16] <chandlerc> so if the project wants branches...
[04:16] <jelmer> The fact that "branches" is required is a bug in bzr-svn
[04:17] <chandlerc> http://pastebin.org/42294 is the traceback
[04:17] <chandlerc> but i'd really like to be able to bzr branch 'svn+https://inc.googlecode.com/svn/wiki' ... will that not work?
[04:18] <jelmer> chandlerc, bug 235301
[04:19] <jelmer> chandlerc, "bzr branch svn+http://inc.googlecode.com/svn/wiki" works fine here
[04:19] <chandlerc> mk... so i just can't "branch" the wiki root
[04:19] <chandlerc> which is fine honestly
[04:19] <jelmer> wiki root?
[04:20] <chandlerc> sorry, so i can't bzr branch from that URL, and then svn-push into a different path to create a subversion "branch"
[04:20] <chandlerc> because it doesn't follow the branching scheme
[04:20] <jelmer> you _can_ branch from that URL
[04:20] <chandlerc> yea, i misspoke, its pushing a new subversion branch that i was anticipating breakage
[04:21] <jelmer> no, that shouldn't be a problem either I think
[04:21] <chandlerc> ? then what does the differing branching schemes do?
[04:21] <jelmer> it's relevant for merge tracking
[04:22] <chandlerc> hmmm, ok
[04:22] <chandlerc> i think i follow that
[04:22] <jelmer> branching schemes should go away in the future
[04:22] <chandlerc> cools, simpler is better. =D
[04:23] <jelmer> I'm considering just hiding them for 0.4.11 since they appear to cause more confusion than they're worth
[04:23] <chandlerc> hehehe
[04:24] <chandlerc> any ETA on the bug? its a blocker for me, i'd just like to know.
[04:24] <chandlerc> additionally, if i can help work up a patch for it, i'd be happy to
[04:27] <jelmer> not sure, haven't had time to look into it yet
[04:27] <jelmer> I'll see if I can fix it for 0.4.11
[04:28] <chandlerc> =D
[04:28] <chandlerc> it'll sadly break 90% of googlecode projects.
[04:29] <chandlerc> i'm gonna poke at it some, but i may get lost in the python (not my strongest language)
[04:31] <jelmer> If there's anything I can help with, let me know
[04:31] <lifeless> bpeterson: could you mention that in the bug
[04:31] <jelmer> The sun appears to be coming up here in .ie though, time for some sleep
[04:31] <jelmer> g'night *
[04:37] <lifeless> beuno: just remembered, you did see my plugin-info plugin announcement right ?
[04:37] <lifeless> night jelmer
[04:46] <chandlerc> sweet, fixed it
[04:54] <lifeless> grats
[04:54] <lifeless> open sores wins again
[05:04] <Keybuk> I dislike this time of year
[05:04] <Keybuk> the Sun stays up and fools you into thinking it's mid evening when in fact it's closer to midnight
[05:04] <Keybuk> then, after you get a bit of darkness to get a few hours decent work done
[05:04] <Keybuk> it comes up again
[05:04] <LaserJock> heh
[05:11] <lifeless> back on SST huh ?
[05:12] <Keybuk> yeah
[05:12] <Keybuk> clearly my personal timezone has already made the shift from management to development
[05:12] <Keybuk> even if my employment status hasn't yet caught up
[05:15] <lifeless> :P
[05:17] <Keybuk> which reminds me
[05:17] <Keybuk> really must write my job description
[06:49] <toyto1> hello guys, any help? I tried branching Tortoise for Windows XP by 'bzr branch https://code.launchpad.net/~amduser29/tortoisebzr/trunk'
[06:50] <toyto1> after it i issue command 'python TortoiseBZR.py' but got an error saying 'Failed to find bzrlib module! Include the path to bzrlib in PYTHONPATH environment...'
[06:51] <toyto1> I set the PYTHONPATH to 'set PYTHONPATH=C:\Program Files\Bazaar\lib' but still it shows that error
[06:51] <toyto1> any idea?
[06:52] <lifeless> what is in C:\Program Files\Bazaar\lib
[06:52] <lifeless> specifically is there a C:\Program Files\Bazaar\lib\bzrlib ?
[06:52] <toyto1> first, thanks for reply lifeless. oh no, actually I didn't find bzrlib under ...\lib directory
[06:53] <toyto1> I found this library.zip, does it contains the bzrlib? let me check
[06:53] <lifeless> if the library has bzrlib in it
[06:53] <lifeless> then put the library on the path - C:\Program Files\Bazaar\lib\library.zip
[06:53] <toyto1> oh yeah, I found a directory named bzrlib inside that library.zip, so I need to extract it in that directory under ..\lib?
[06:54] <toyto1> ah i see
[06:54] <toyto1> okay trying...
[06:54] <lifeless> I don't use windows
[06:54] <lifeless> so can't offer more than generic help for you though, sorry
[06:54] <lifeless> markh is doing a lot on this stuff I believe
[06:56] <markh> it will need to be in your "global" environment, so processes like explorer.exe see the env var
[06:56] <markh> just setting it at a command-prompt will not work properly
[06:56] <toyto1> indeed me too :( my friends wants me to try it in windows.
[06:56] <toyto1> markh: what do you mean setting in the command prompt?
[06:56] <markh> how are you setting the PYTHONPATH variable?
[06:56] <toyto1> so something like going to My Computer by right clicking it... blah2... way.
[06:57] <markh> ok - sorry - I just re-read the above.  It sounds like you are having trouble with registration anyway.
[06:57] <toyto1> yeah, I set the PYTHONPATH thru command prompt, I set actually under the My computer... Advance... Environment variables but the path is C:\
[06:57] <markh> so, once you set the variable, you try and re-execute the 'python' command?
[06:57] <markh> in a command-prompt?
[06:58] <lifeless> speak their name and they appear; the wonders of the net.
[06:59] <markh> :)
[06:59] <toyto1> markh: actually I did just under command prompt, it executse exlcuding that error, but when I hit the carriage return there's a Message box so I just click 'OK' but it does execute
[06:59] <toyto1> ah markh it says The procedure entry point ?PyWinObject_AsHANDLE@@YAHPAU_OBJECT.... something like that
[07:00] <toyto1> but after clicking ok the TurtoiseBZR.py executes its scripts
[07:00] <markh> bugger - it sounds a little like the pywin32 install is screwed up :(
[07:01] <markh> oh - right-  you are trying to use tbzr with a binary install of bzr itself
[07:01] <markh> that might be a struggle :(
[07:01] <toyto1> markh: oh, that's a bug? ah what do you mean by that?
[07:02] <markh> iiuc, tbzr basically expects to be running from a source distro of bzr
[07:02] <lifeless> markh: completely separate to the current discussion you might like to send commits for the stuff you are doing to the bazaar-commits list; we encourage contributors to do this as a way of getting visibility
[07:03] <lifeless> re: binary install - the 'binary' of bzr has only a small subset of the regular python libraries
[07:03] <lifeless> it should be possible to do an add-on installer to add what tbzr needs as well as tbzr itself etc etc, but I don't believe anyone has done that so far
[07:03] <markh> lifeless: yes, I do need to do that - I'll be making some announcement this week, then next week I go to sydney to work with a few of the guys
[07:04] <lifeless> markh: including me ;P. Though the vf merge and stacking will be consuming me
[07:04] <markh> lifeless: oh cool :)  I'm not sure who you even are ;)
[07:04] <lifeless>  /who lifeless
[07:04] <markh> non-obvious nicks ;)
[07:04] <markh> pidgin doesn't know what /who means :)
[07:05] <lifeless> oh, right click and info on my nick?
[07:05] <lifeless> (gui clients, feh) :)
[07:05] <markh> :)  Ahh - cool - will be good to meet!
[07:06] <markh> so yeah, we do need better binary "packaging" story on Windows IMO, especially keeping plugins in mind
[07:06] <lifeless> yup
[07:06] <lifeless> alexander basically designed it all
[07:06] <lifeless> and AFAICT his design is sane, but noone stepped up to do actual packages
[07:08] <markh> I'm wondering if py2exe is a good idea if we really want plugins to be seamless.  An alternative is to install a (basically) full stand-alone Python tree, running from source code, with a few front-end .exe files that are actually the entry-points.  Kinda 1/2 way between py2exe and a source distro.  Possibility is things like eggs etc would work seamlessly
[07:08] <markh> but I've no firm ideas or opinions
[07:08] <markh> IIUC, setuptools already provides a tool to generate such executables
[07:09] <lifeless> I have a bad opinion of eggs; but perhaps they are the best solution for windows.
[07:09] <lifeless> broadly though many plugins have dependencies that are a superset of pywin32+bzr
[07:09] <lifeless> e.g. they want 'patch' or 'pysvn' or '...'
[07:09] <markh> eggs or not, the idea is that it looks alot more like a "normal" python distro, so any way of getting packages on is likely to work, or nearly work :)
[07:10] <lifeless> so however we approach it we need the act of 'install plugin X' to also 'install missing deps for X'
[07:10] <Keybuk> what's cheese?
[07:10] <markh> ideally yeah, but I fear with py2exe, some plugins would fail to work even if you manually attempted to install all plugins
[07:10] <markh> sorry
[07:11] <markh> install all deps
[07:11] <lifeless> to my mind, the py2exe thing isn't really a problem as a result, because we can pickup missing standard library modules as part of their dependencies
[07:11] <lifeless> markh: are there other causes for failure that I'm not aware of then ?
[07:12] <markh> what happens when a plugin needs an obscure python lib module we didn't include in the .zip file?  Or we just ensure the entire lib is there?
[07:12] <lifeless> (I guess I should be more explicit - if we did what py2exe/freeze do on a per-plugin basis, then subtract the set we know ship in the bzr py2exe .zip, we get the extras)
[07:12] <toyto1> markh: I uninstalled bzr, the package with Windows_Installer_bzr.exe, then install the bzr-1.5.win32-py2.5.exe
[07:13] <toyto1> would that be fine? I try going to command prompt and issuing 'bzr' but it is now not recognize, how to make it recognize or use bzr in that way?
[07:13] <lifeless> markh: right, I'm saying that the .zip for a packaged plugin potentially includes std library modules too.
[07:13] <markh> toyto1: so you need to say "python c:\path\to\bar"
[07:14] <markh> lifeless: right - if it is OK that only plugins which have had the special "packaging for windows" magic done before they can be used, I'd expect that would work fine.
[07:14] <toyto1> markh: what path, only for python? I already set the path PATH=C:\Python25;%PATH%
[07:14] <markh> windows doesn't look at the shebang line
[07:15] <markh> sorry - /path/to/bzr
[07:15] <lifeless> markh: so I think that installing plugins by 'bzr branch' can only be expected to work for source installs of bzr and folk willing to manually install deps. binary installs (both unix and windows) should work with the recommended $target binary package of bzr
[07:15] <markh> so "c:\src\bzr\bzr", assuming that file exists
[07:15] <markh> (the source distro of bzr has the entry point without a .py extension)
[07:16] <markh> lifeless: right - I guess I need to understand the scope of plugins better
[07:17] <lifeless> markh: Its pretty broad.
[07:17] <toyto1> markh: Sorry I didn't notice it but found the 'Start -> Program Files -> Bazaar -> Start bzr' then it shows me the command prompt at directory C:\Python25\Scripts
[07:17] <lifeless> markh: bzr-svn is one of the most cool plugins I know of.
[07:17] <lifeless> (in terms of technology, and in terms of what it allows users to do)
[07:18] <toyto1> so setting or adding like PATH=C:\Python25\Scripts;%PATH% would recognize 'bzr' in command prompt, isn't it?
[07:18] <markh> toyto1: no - the cmd prompt doesn't consider a file named 'bzr' could be executable
[07:18] <lifeless> no, windows needs a .bat or .com or .exe file to do that
[07:18] <lifeless> setup.py can create a bzr.bat
[07:18] <lifeless> uhm
[07:18] <lifeless> python setup.py build --in-place
[07:19] <lifeless> not sure if that will work, but it is worth a shot
[07:19] <markh> I don't recall doing any build at all :)
[07:19] <lifeless> markh: we don't require one.
[07:19] <lifeless> markh: but you can if you wish, to build the C accelerator modules.
[07:20] <toyto1> markh: ohmm I have that notion because when I try installing the bzr as package w/o python, then trying at command prompt issuing bzr it works actually but I unisntalled it. let me try
[07:20] <lifeless> one advantage of binary installs is that the C modules are prebuilt
[07:20] <markh> toyto1: yes, the binary install installs a bzr.exe
[07:20] <markh> I expect that windows users will very rarely run from source unless they are interested in hacking on bzr itself
[07:21] <lifeless> yup, me too. But we should keep it easy to do, to make casual contribution straight forward.
[07:22] <markh> It is far far easier to make the decision to run from source on a linux box.  IMO we want to avoid saying "to do X, you must be running from source" on Windows if we can (which isn't to say anyone is planning that or anything :)
[07:23] <toyto1> markh: ah its now recognize at command prompt after I append C:\Python25\Scripts at Environment Variables
[07:23] <markh> right - I guess its finding the bzr.bat
[07:23] <lifeless> markh: oh, completely agree, for unix too. source installs should only be needed for development-of-bzr.
[07:24] <lifeless> markh: I think every bzr dev will agree :)
[07:24] <kumi_> Well, I run source on windows because the official release is broken :)
[07:24] <toyto1> markh: and now when I try to python TortoiseBZR.py the error that I told you that shows with 'OK' button is now gone. Thanks guys :)
[07:24] <markh> cool :)  The next thing will be to check it actually works with explorer :)
[07:25] <lifeless> kumi_: you have filed bugs?
[07:26] <kumi_> Nope, it was that merge bug you called my attention to
[07:26] <kumi_> You stated it was already fixed in HEAD so I didn't bother
[07:27] <toyto1> markh: ohmm I tried editing a file, then right clicking it then I clicked BZR Diff it says 'BZR Error ... Unknown Diff Command'
[07:27] <toyto1> any idea?
[07:27] <lifeless> kumi_: oh right
[07:27] <lifeless> kumi_: I thought you means some specific to windows errata
[07:30] <kumi_> well there is a slight win32 bug I noticed. The included start_bzr.bat in bzr.dev calls bzr.exe, which doesn't exist
[07:31] <lifeless> ah interesting
[07:31] <lifeless> if there isn't a bug, having one would be good
[07:32] <kumi_> where do I file it?
[07:32] <lifeless> launchpad.net/bzr
[07:34] <markh> hrm
[07:36] <markh> toyto1: I'm not sure where that message would come from - its in a message box?
[07:37] <kumi_> https://bugs.launchpad.net/bzr/+bug/238277
[07:37] <fdv> Hi. I'm still trying to convert an svn repo to something else, bzr being the first choice. I have run into a problem, though, that several revisions are broken in the svn repo, and generate errors when one tries to check them out. Does anybody know whether it's possible to make bzr-svn ignore some revisions and how that would work?
[07:38] <fdv> (that is, running bzr branch)
[07:39] <lifeless> fdv: oh interesting
[07:39] <lifeless> fdv: actual svn corruption?
[07:39] <fdv> lifeless: yes, it is :p
[07:40] <lifeless> fdv: I think the usual answer would be to do a svndump, svnfilter them out, then restore the dump
[07:40] <lifeless> fdv: if you can't do that, i think you'd need to raise a question on questions.launchpad.net/bzr-svn
[07:40] <lifeless> I am sure noone has needed this facility to date; and it has implications with the revision mapping logic
[07:40] <toyto1> markh: yeah, python error related. I think it shows that error because the bzr was not recognize under python scripts, what do you think? because after I reinstalled that bazaar package installer w/no python then install the bzr-...pythonwin32.exe it's okay now
[07:41] <toyto1> I mean okay that it doesn't show that message prompt anymore.
[07:41] <fdv> uh
[07:41] <fdv> pain :p
[07:41] <fdv> lifeless: ok, thanks, then I might have a solution, at least :)
[07:41] <markh> toyto1: does it work though?
[07:41] <toyto1> but my problem now is using that tortoise with Bzr Commit and Bzr Commit says Uknown command commit/diff
[07:42] <toyto1> nah :( even I have already installed the paramiko and pycrypto libraries
[07:42] <fdv> lifeless: yes, I'd think it raises some issues
[07:42] <lifeless> my spam filters die on weekends
[07:42] <lifeless> I feel a strong urge to violence to spammers
[07:42] <toyto1> I tried under command prompt and it's okay actually. I did commit and diff or issuing bzr command and successfully interacted to the server w/ repositories
[07:45] <markh> toyto1: that error message could be due to any error, and tbzr sadly doesn't indicate any details about the error :(
[07:46] <markh> (I found where it comes from, and it isn't pretty :)
[07:47] <kumi_> maybe your .bzr.log has a clue as to why
[07:47] <toyto1> markh: what do you think where it came from? I have no idea actually, since I assume it would work since I assigned globally to the environment variable %PATH% the C:\Python25\Scripts where bzr resides
[07:48] <toyto1> knowing the bzr would be recognize as a command
[07:48] <toyto1> kumi_: may I know where that .bzr.log resides? I use the Windows explorer oh is it in the .bzr directory?
[07:48] <markh> toyto1: its an "internal error" in tbzr itself, not related to bzr itself.  Its not getting as far as executing any bzr commands.
[07:49] <markh> IIRC, the log file will be in your "My Documents" folder
[07:49] <markh> I doubt it will have anything useful, but its worth a look :)
[07:51] <toyto1> markh: let me check it. :)
[07:51] <lifeless> 'bzr --version' will list the log file location
[07:56] <kumi_> yep, look for "Bazaar log file" in the output :)
[07:58] <kumi_> BTW, the tortoisebzr README says "To see debugging output run python tortoise-bzr.py --debugging-output"
[07:58] <markh> sadly that error message comes from a bare 'except' clause that doesn't log or display or do *anything* with details about the error :(
[07:59] <toyto1> guys kindly take a look at this page -> http://pastebin.org/42313
[07:59] <toyto1> that's the log file I found under my documents folder
[08:00] <kumi_> maybe that's the wrong location, try bzr --version as was suggested
[08:01] <kumi_> mine is not in My Documents, but in C:\Documents and Settings\user\bazaar\2.0
[08:04] <toyto1> kumi_: ah yeah kumi_ that's the location actually the bzr --version showed to me.
[08:06] <markh> so, I'm guessing another dependency is missing :(
[08:07] <markh> try executing: "python.exe -c "import commands.CommitCommand"
[08:07] <kumi_> I get a traceback when I run the TortoiseBZR Diff command from the commandline
[08:07] <markh> I see:
[08:07] <kumi_> "C:\bin\Python25\python.exe" "C:\bin\bzr.dev\bzrlib\plugins\tortoisebzr\TBZRCommand.py" /command:TBZRDiff /filepath:"C:\Temp\BZR57A3.tmp"
[08:07] <kumi_> that results in:
[08:07] <markh> Traceback (most recent call last):
[08:07] <markh>   File "<string>", line 1, in <module>
[08:07] <markh>   File "commands\CommitCommand.py", line 48, in <module>
[08:07] <markh>     import gtk
[08:07] <markh> ImportError: No module named gtk
[08:08] <kumi_> gah n/m
[08:08] <toyto1> markh: ah i see, what gtk is that do you think? let me check it.
[08:08]  * kumi_ disappears into slumberland
[08:09] <clemente> Hi, I'm seeing „AttributeError: 'GitRepository' object has no attribute '_format'“. See http://pastebin.org/42314
[08:09] <clemente> jelmer: I think it may be after your change in revision 16 to the plugin „git“
[08:21] <toyto1> now clicking the BZR Diff works, but BZR Commit still an unknown commit command
[08:28] <markh> it kinda sucks, but the only thing I can suggest it to get a command similar to 'python.exe" "C:\...\TBZRCommand.py" /command:Commit /filepath:"commit.tmp"' working, so you get the same error (commit.tmp must exist with the full path of each file you want to commit per line).  Once you see the same error, we can then modify a .py file so it actually prints the error itself
[08:29] <AfC> clemente: the bzr-git capability is still very young...
[08:29] <markh> I know it shouldn't be this hard :(  We are working on fixing it, but I'd expect it to be at least a few weeks before I've something to show for myself.
[08:35] <clemente> AfC: however it's the best I found for moving from git to bzr
[08:36] <clemente> Although the other methods also didn't work
[08:36] <AfC> That's a surprising endorsement to hear.
[08:37] <AfC> clemente: (yeah, I ran into the same problem. I still have one project whose history is abandoned. Someday I hope to convert it and merge/graft/rebase/pray-to-the-gods/sacrifice-a-small-goat/whatever together to the just add it all at once I was forced to do instead)
[08:39] <clemente> Can't you import the history with git-fast-export + bzr fast-import ?
[08:39] <clemente> I run into encoding problems due to invalid UTF-8
[08:40] <toyto1> markh: is it /command:Commit or /command:TBZRCommit ?
[08:41] <markh> it looks like TBZRCommit, yeah
[08:41] <markh> indeed, it looks like XYZYCommit would also work (ie, first 4 chars can be anything :)
[08:42] <toyto1> markh: thanks, i'll try
[08:44] <toyto1> markh: it says 'python.exe Unable to Locate Component .... This application has failed to start because libglib-2.0-0.dll was not found. Re-installing the application may fix this problem'
[08:45] <toyto1> where I can get that dll by the way? any idea?
[08:45] <markh> ack - so I guess that is part of gtk or something :(
[08:45] <lifeless> thats bzr-gtk yes
[08:45] <toyto1> I see. that would be PyGTK ? which interacts to gtk2 library isn't it?
[08:45] <lifeless> installing bzr-gtk and adding it to the path will work
[08:46] <lifeless> yes, gtk and glib are kindof twinned :)
[08:46] <toyto1> lifeless : ah so installing it let me try
[08:52] <toyto1> lifeless: it's still 'Unknown Commit Command' but I found that libglib...dll and it's under Application Data\bazaar\2.0\plugins\gtk\_lib\gtk-2.0\gobject
[08:54] <clemente> I have installed latest bzr in a local directory which now has its ./bin, ./lib, ./man. Now how do I run this instead of the system bzr?
[08:56] <lifeless> clemente: put that bin directory on your path ahead of wherever your system bzr is
[08:56] <toyto1> and issuing the 'python TBZRCommand.py /command:TBZRCommit /filepath:"%USERPROFILE%/Desktop/bzrtemp.tmp"' would still say that the lib is not found
[08:56] <lifeless> toyto1: like I say, I don't use windows, so I'm not in a good position to help :(
[08:56] <toyto1> lifeless: oh sorry :)
[08:57] <markh> toyto1: try adding Application Data\bazaar\2.0\plugins\gtk\_lib\gtk-2.0\gobject to your PATH
[08:58] <markh> just at the cmdprompt is fine: set PATH=%PATH%;Application Data\bazaar\2.0\plugins\gtk\_lib\gtk-2.0\gobject
[08:58] <markh> (obviously the fq path is needed)
[08:58] <clemente> lifeless: Then the old bzrlib from /usr/share/pyshared/bzrlib will be used
[09:09] <lifeless> clemente: have you tried? we have some logic for handling this
[09:10] <lifeless> clemente: alternatively, just put the source directory you downloaded bzr from on your path
[09:10] <lifeless> the bzr there will pickup the bzrlib there automatically
[09:15] <clemente> : /n/bzrbzr ; PATH=/n/bzrbzr PYTHONPATH=/n/bzrbzr bin/bzr
[09:15] <clemente> bzr: WARNING: bzrlib version doesn't match the bzr program.
[09:15] <clemente> .... bzrlib from ['/usr/lib/python2.5/site-packages/bzrlib'] is version (1, 5, 0, 'final', 0)
[09:16] <toyto1> markh: it works fine but when the window showed up, clicking the commit button after I provided a message prompts an Uknown error saying 'NoneType' object has no attribute 'encode'
[09:16] <lifeless> clemente: when I run from source I usually just do 'bzr branch <> local_path'; 'ln -s ~/bin/bzr local_path/bzr' and that works fine
[09:17] <markh> toyto1: is there a traceback?
[09:28] <clemente> lifeless: Well, for me it works if I do this:   PYTHONPATH=/n/bzrbzr/lib/python:/usr/lib/python2.5:/usr/lib/python2.5/lib-dynload /n/bzrbzr/bin/bzr
[09:28] <lifeless> clemente: ouch; well time for a shell alias I guess
[09:29] <clemente> Just executing ~/bin/bzr still uses the system bzrlib
[09:29] <clemente> Or better look for a clean solution...
[09:30] <toyto1> markh: where I can find the traceback?
[09:30] <toyto1> markh: the error won't show if I check the 'commit locally'
[09:30] <lifeless> clemente: the problem is that 'setup.py install' splits the command and the library
[09:30] <lifeless> clemente: which is why I suggested pointing at the source, not the 'installed copy'
[09:31] <toyto1> markh: however if I try to bzr log to the server, there's no commit or changes happens, only at that local directory/file even I tried again right click the directory -> BZR Commit
[09:31] <clemente> lifeless: ok, that's true, by using the source directly you save problems
[09:32] <beuno> lifeless, the plugin-info announcent from a few months back?
[09:32] <lifeless> beuno: yes; it is rather relevant to the plugin stuff ;P
[09:32] <toyto1> markh: when I edit a file, let's say test.txt, then right click that file BZR Commit will show the window, then I click the commit locally (because it will show that error stuff if not)
[09:32] <lifeless> beuno: but I had _no_ feedback, felt a bit like it went into a vacuum
[09:33] <toyto1> markh: after that, I again right click the file -> BZR Commit -> and it works okay but when I 'bzr log sftp://URL' to the server, the file was unchanged :(
[09:34] <beuno> lifeless, ah, it's perfect from what I'm working on. I thought I had answered, but I'll find the thread and reply. It's definetely something that should get implemented. If got merged into the docs though, didn't it?
[09:34] <lifeless> beuno: uhm, docs stuff was different; the plugin has a command to scan plugins
[09:35] <lifeless> :!bzr plugin-info .
[09:35] <lifeless> search(1.6.0dev0) from . supplies commands("index","search"), control formats(), checkout formats(), branch formats(), repository formats().
[09:36] <lifeless> :!bzr plugin-info ../../loom/trunk/
[09:36] <lifeless> loom(1.4.0dev0) from ../../loom/trunk/ supplies commands("combine-thread","create-thread","down-thread","loomify","record","revert-loom","show-loom","up-thread"), control formats(), checkout formats(), branch formats("Loom branch format 6","Loom branch format 1"), repository formats().
[09:36] <beuno> lifeless, ah, I *did* miss the announcement then
[09:37] <markh> toyto1: now tbzr is actually asking bzr to do something, that log might hold some clues
[09:37] <lifeless> Message-Id: <1204311754.28682.80.camel@lifeless-64>
[09:39] <lifeless> beuno: as in 'stuff should get implemented, I did the freaking thing'
[09:41] <beuno> lifeless, heh, yeah. Branched it, I'll reply and try to help you get some movement there. Now, it's almost 6am, so I'm off to be for a while.
[09:42] <beuno> s/be/bed
[09:42] <beuno> night!
[09:44] <lifeless> jelmer: is svn://gwalcmai.vernstok.nl/bzr/trunk you ?
[09:44] <lifeless> beuno: gnight
[09:49] <toyto1> markh: i'll send the log and that problem to tbzr
[10:17] <toyto1> markh: I just notice the tbzr doesn't put any logs or report logs under .bzr.log isn't it? I notice it just now, because everytime I do something with tbzr the log file doesn't change
[10:18] <toyto1> else, if i do some stuffs at command line, then the log file changes
[10:19] <vila> lifeless: ping, a couple of questions regarding webdav/list_dir()/stat()
[10:22] <vila> lifeless: I've now implemented list_dir for webdav, but the test suite now fails because it want stat() also. I'm pretty sure packs doesn't requires stat() (list_dir was required for clearing obsolete_packs.
[10:25] <vila> the current transport API has several flaws here: stat() usage is more or less implied if the transport is writable, it seems to be used to allow S_ISDIR and may be file  size (at least remote seems to only implement file_size and chmod bits only)
[10:26] <vila> thoughts ?
[10:27] <vila> and once again, I wish a plugin could say: this test would fail with this set of parameters
[10:28] <vila> so that I can say: webdav does not implement delete_tree and copy_tree for example
[10:28] <vila> which AFAIK are not used by bzr
[10:29] <lifeless> well
[10:29] <lifeless> delete_tree and copy_tree are implemented on the base class I thought, in terms of other ops
[10:29] <vila> hurray ! You're not asleep yet :)
[10:29] <lifeless> so webdav has no reason to not have them
[10:29] <lifeless> yes stat is implied, what remote does is indeed sufficient
[10:30] <vila> lifeless: yup, but the base implementation needs a way to distinguish between files and directories which stat() provides
[10:32] <vila> I can find a work around to implement only file_size and dir bit in the chmod bits but only for apache2 (AFAICT after tests against apache2 and lighhtp, the later having more bugs that make it even harder to support for the webdav plugin)
[10:32] <vila> lifeless: oh, I misread " is indeed sufficient" as "is indeed *not* sufficient"
[10:33] <vila> So a work-around for apache2 as said above will be ok ?
[10:33] <vila> I'll try to report bugs against lighttpd then.
[10:33] <lifeless> sounds reasonable to me
[10:34] <vila> lifeless: good.
[12:57] <aantn> hello
[12:57] <aantn> how can I apply a patch generated with bzr diff?
[12:58] <aantn> when I use bzr patch I get the following
[12:58] <aantn> http://rafb.net/p/W8hLCl96.html
[13:03] <Odd_Bloke> aantn: Why are you using 'diff' and 'patch' to communicate changes?
[13:03] <aantn> Odd_Bloke: Its from a person who's interested in getting involved and just made their first change to the code
[13:04] <aantn> Odd_Bloke: I figured out how to apply it; I needed to specify the right directory
[13:05] <fredreichbier> hi all ;)
[13:08] <Odd_Bloke> aantn: If both parties are using bzr, there are better ways to communicate changes than diff/patch.
[13:09] <aantn> Odd_Bloke: do you mind explaining?
[13:09] <Odd_Bloke> However, I need to grab a shower before heading off to cook for 30 people, so now is probably not the best time for me to help you out.
[13:09] <aantn> Odd_Bloke: fair enough
[13:09] <Odd_Bloke> aantn: Briefly, you want to use 'bzr merge', because it will retain history (whereas a dumb diff/patch won't).
[13:10] <aantn> Odd_Bloke: I know, but this is for a minor ten line patch
[13:11] <Odd_Bloke> aantn: OK, I'd still tend to use 'merge', if only because it (a) doesn't require remembering how to create and apply patches properly, (b) retains history (however little of it there may be), and (c) will have much better conflict resolution if you do get problems.
[13:11] <Odd_Bloke> Anyhow, now I'm really *GONE*.
[13:11] <aantn> Odd_Bloke: ok
[13:11] <aantn> thanks for the advice
[13:21] <Pieter> so there's no real way to work with patches like git and mercurial allow you to?
[13:23] <Jc2k> Pieter: define "work with patches"?
[13:23] <Jc2k> there is a "bzr patch" command in bzrtools that i've never used before
[13:24] <Jc2k> there are also cherry picking and shelves to help break a patch up into pieces, i guess.
[13:25] <lifeless> Pieter: bundles are patches+rich metadata that transfer all history, merges etc
[13:25] <lifeless> Pieter: they can send a single commit, or an entire branch
[13:29] <Pieter> lifeless: how can they send a single commit? it always tries to send all commits differing from my upstream
[13:29] <lifeless> Pieter: bzr send -r -2..-1
[13:30] <lifeless> that will create a cherrypick, with enough data to ensure it can be applied whether or not they have the basis available
[13:36] <Pieter> lifeless: ok, but if I merge that it doesn't keep my commit message
[13:36] <lifeless> Pieter: yes, that is a current limitation of bzr - what happens is that the cherrypick is not put into the merge graph
[13:37] <lifeless> Pieter: the commit message *is* in the repository of the person that merged it, but all our current revision graph pointers are transitive
[13:37] <lifeless> cherrypicking is on the fairly-high-TODO-list to put a comprehensive solution in place for. Unfortunately the semantics of merge while clear to a human, are not clear at the algorithmic level [in the present of cherrypicking even more so]
[13:38] <lifeless> or to put it another way: when we fix cherrypicks to be better, the bundles that are created today can give better results immediately
[15:37] <lifeless> gnight
[15:37] <lifeless> tomorrow, I shall teach search to query the index.
[15:37] <lifeless> anyone wanting extremely early previews - pull my commit diffs off the commits list :P
[15:38] <lifeless> actually, I'll push it to lp:~lifeless/+junk/bzr-search now
[15:44] <toyto1> lifeless: any idea why I receive bzr: ERROR: Path(s) are not versioned: "testing.txt"
[15:44] <mathrick> lifeless: will it eat my data if I run it on my repo?
[15:46] <mathrick> hmm, it doesn't do anything for me
[15:46] <mathrick> no errors, but no results either
[15:57] <MvG> Someone willing to review http://tinyurl.com/3zbfun related to bug #128496? Or what's the normal workflow here?
[16:15] <vila> MvG: the normal workflow for the bzr-svn plugin is to have jelmer informed, since he participated to the bug discussion, I think he is :) If not having mention his name should be enough to summon him
[16:23] <dato> Pieter: hm
[16:23] <dato> Pieter: I fixed bug 219042 without knowing you had too
[16:25] <dato> Pieter: I did the same as you first, but there're cases when that can go wrong
[16:26] <dato> Pieter: http://dpaste.com/55476/
[16:26] <fredreichbier> what's wrong with this hook? http://paste.pocoo.org/show/65053/
[16:26] <MvG> vila: The diff in the tinyurl above is against bzr.dev, not bzr-svn.
[16:27] <MvG> So it's the bzr workflow I'm interested in.
[16:29] <vila> MvG: ha, sorry, missed that, then maling bazaar@lists.canonical.com is the first step, if you think it's ready to be merged (doest it include tests for example) you can put [MERGE] in the subject, anyway, mailing the list will at least get the discussion started
[16:29] <vila> s/maling/mailing/ s/doest/does/ s/$/typos are accepted, don't worry/
[16:30] <MvG> vila: OK, the I'll do so once the full test suite runs here have completed and I know there were no failures introduced by this patch.
[16:31] <vila> ok, extra bonus points if you had a test failing without your patch and succeeding with it
[16:45] <MvG> vila: Hmmm... tricky, as such tests would require a locale besides the default POSIX locale to be supported and installed. That's not a requirement I'd make in a test.
[16:46] <vila> MvG: Mention that in your mail then
[16:46] <MvG> And writing tests supposed to expectedly fail on most systems except those in a specific locale seems like a bad idea as well.
[16:46] <MvG> OK, will do.
[17:06] <afflux> hi. I have a "trunk" bzr repo and I created a branch for it. Now I did changes to my branch and the trunk was changed too. I merged the trunk stuff to my branch using "bzr merge $trunk". My problem is now the following: my branch was merged into the trunk and some parts of it were changed.
[17:07] <afflux> Finally, my parts were removed again. So now, I want to re-use my old branch. I want to use the current trunk as a new "base" and re-start to diverge from there.
[17:08] <afflux> how do I do that? Doing "bzr pull $trunk" fails with "bzr: ERROR: These branches have diverged. Use the merge command to reconcile them."
[17:13] <MvG> I'm not sure I understood you correctly, but if you want to re-start to diverge from some other branch, it would seem sensible to create a new branch.
[17:14] <MvG> "bzr pull --overwrite" might do the same thing, looking at its documentation.
[17:16] <afflux> MvG: looks good, thanks!
[17:41] <fredreichbier> are there apps like loggerhead for other languages, e.g. php?
[19:45] <fredreichbier> a question to bzrlib: how do i get a complete diff between two revisions?
[19:45] <beuno> fredreichbier, let me cook that up for you real quick
[19:46] <fredreichbier> :)
[19:49] <beuno> fredreichbier, something like this should work: http://paste.ubuntu.com/18544/
[19:49] <fredreichbier> thank you very much :)
[19:50] <beuno> you're very welcome
[19:50] <beuno> Verterok, mornin'
[19:51] <Verterok> beuno: mornin'
[21:42] <lifeless> mathrick: check in .bzr/bzr-search/indices
[21:47] <mathrick> lifeless: it has things like 0408nfiguration0176
[21:47] <mathrick> that seems a bit garbled
[21:51] <lifeless> it has \x00 delimiters
[21:51] <lifeless> use a binary safe viewer, e.g. vim or some versions of less
[21:52] <lifeless> its a very crude index at the moment, see DESIGN for some notes
[21:52] <lifeless> (but thanks for giving it a spin and trusting it wouldn't eat stuff!)
[21:57] <thumper> morning
[21:58] <mathrick> lifeless: hehe, you're welcome
[21:58] <mathrick> I always appreciate cute tools like that
[21:58] <lifeless> you can also use bzrlib to query it
[21:59] <lifeless> no query language at the moment so its -
[21:59] <lifeless> from bzrlib.plugins.search.index import open_index_url
[21:59] <lifeless> index = open_index_url('.')
[21:59] <lifeless> terms = dict(index.all_terms())
[22:00] <lifeless> mathrick: if I can ask, what branch did you index?
[22:00] <lifeless> bzr itself, a project of yours?
[22:00] <mathrick> my project
[22:00] <mathrick> a very small one, too
[22:00] <lifeless> cool
[22:00] <mathrick> lifeless: ah, I used bzr search
[22:00] <lifeless> it probably returned near-instantly
[22:00] <mathrick> yeah
[22:01] <mathrick> what is the syntax for search?
[22:01] <lifeless> well, I'm writing that bit now (my sign-off comment was that that bit was next) :P
[22:01] <lifeless> I'm going to do the most trivial thing to start with - a list of terms, and just do a set union
[22:04] <lifeless> hi thumper - holiday for me today :P
[22:04] <thumper> lifeless: yeah, I know
[22:04] <thumper> lifeless: doesn't stop you being here though, does it?
[22:05] <lifeless> so I'm going to do what I did yesterday - play WoW, and when on a bird, code a search plugin for bzr
[22:06] <thumper> on a bird?
[22:06] <lifeless> long distance travel
[22:11] <chandlerc> is there a bisecting plugin for bzr?
[22:11] <lifeless> yes, bzr-bisect
[22:11] <chandlerc> sweet... part of bzrtools, or separately available?
[22:11] <lifeless> just separate
[22:12] <lifeless> its on the plugins page, packaged for debian/ubuntu too
[22:12] <chandlerc> k
[22:14] <chandlerc> hmm, any particular reason lots of stuff in the bazaar gentoo overlay doesn't have amd64 keywords?
[22:15] <lifeless> I don't know what that means
[22:16] <chandlerc> in gentoo, packages are marked with keywords to indicate what "platform" they are available for
[22:16] <Jc2k> chandlerc: why would python stuff need to special case for amd64?
[22:16] <chandlerc> Jc2k: it shouldn't which is why i don't know why its *only* exposed for x86
[22:16] <lifeless> chandlerc: most of bazaar is pure python
[22:16] <Jc2k> packaging bug :p
[22:16] <chandlerc> unfortunately, the existing gentoo paradigm requires opting into each and ever platform explicitly
[22:17] <lifeless> chandlerc: sounds like a bug to me
[22:17] <mathrick> chandlerc: that's a good question for #gentoo, asking anyone else is a bad idea
[22:17] <chandlerc> well, the gentoo overlay is managed by bazaar folks, not gentoo folks
[22:17] <mathrick> ah
[22:18] <chandlerc> lifeless: i'll file a bug with that project on launchpad suggesting to always use the same set of platform keywords as python
[22:18] <lifeless> chandlerc: please do
[22:18] <lifeless> chandlerc: its actually gentoo folk that manage it - gentoo folk interested in bazaar AIUI
[22:18] <chandlerc> fair enough, but i suspect there may be more gentoo folks interested in bazaar here than in #gentoo. ;]
[22:18] <lifeless> indeed ;P
[22:19] <chandlerc> potentially because i'm here, and not there... oh the joys of a tiny distribution
[22:26] <beuno> lifeless, Loggerhead would rock with a good search plugin, so you can count on me implementing it as soon as it's usable  (that would eliminate the last need for a sqlite cache, fulltext indexing)
[22:29] <mwhudson> beuno: but there's still the second-to-last (files changed between revisions) to go :)
[22:30] <beuno> mwhudson, absolutely, and good morning
[22:30] <mwhudson> hi
[22:30] <beuno> I was thinking that if the current way of getting diffs works out to be a winner, we can do the same for history navigation
[22:31] <beuno> and get the files changed on a per-revision basis
[22:31] <beuno> and maybe get rid of the cache that way
[22:31] <beuno> until bzr can provide that information fast enough
[22:32] <mwhudson> yes that's definitely something to consider
[22:33] <beuno> mwhudson, I've been cleaning the branch up lately, and, even though I still have to tweak some things, if you could start reviewing it, it would help me quite a lot :)
[22:34] <mwhudson> beuno: ok, i'll have a look in a bit
[22:35] <beuno> mwhudson, cool, thanks. I'll be around the following hours working on this.
[22:40] <lifeless> beuno: well grab the plugin, patches appreciated
[22:42] <beuno> lifeless, it's already out there?
[22:42] <lifeless> a sketch yeah
[22:42] <lifeless> lp:~lifeless/+junk/bzr-search
[22:43] <lifeless> there is plenty of surface area
[22:46]  * beuno branches
[22:46] <beuno> any luck on the python text indexing library?
[22:47] <lifeless> see DESIGN :P
[22:48] <lifeless> when did for x in y: else: get added?
[22:52] <beuno> lifeless, interesting, you're pretty deep into it already
[22:54] <lifeless> beuno: indeed there would not be much point asking for patches until ~ where it is now
[22:55] <beuno> right, I was under the impression it was more of an idea than 1k of code  :)
[22:55] <lifeless> well
[22:56] <beuno> I'll play around with it and hopefully be able to provide some patches to it
[22:56] <lifeless> its the weekend, and I did not get much coding on the last flight :P
[23:14] <Verterok> mornin'
[23:14] <Verterok> lifeless: Hi
[23:14] <lifeless> hi
[23:14] <Verterok> lifeless: about the fulltext indexer, I just found: http://swapoff.org/wiki/pyndexter
[23:18] <lifeless> http://swapoff.org/wiki/pyndexter/docs/pyndexter.indexers.builtin :(
[23:19] <Verterok> http://swapoff.org/browser/pyndexter/trunk/pyndexter/indexers/_builtin.py
[23:21] <lifeless> oh god it uses pickle
[23:25] <Verterok> it's that too bad?
[23:26] <lifeless> its possibly ok
[23:26] <lifeless> we'd have to write our own backend from the looks of it anyhow
[23:27] <lifeless> (or have it be local disk only)
[23:27] <lifeless> and it doesn't have any idea of expoential backoff on storage
[23:36] <vila> woot! webdav plugin passing bzr tests again :)
[23:36] <lifeless> grats
[23:38] <vila> the plugin test server itself need to be fixed though I used a local apache2 test server (with the local_test_server plugin) to validate the new implementation, but that should be easy and can wait tomorrow :)
[23:39] <lifeless> beuno: a good thing to work on would be to extend the term posting lists created such that file texts, file paths etc are getting indexed
[23:39] <lifeless> thats really quite orthogonal to the 'engine' work
[23:41] <beuno> lifeless, sure, I'll take a look at it and see if I can move forward
[23:42]  * beuno adds that to his ToDo
[23:49] <mwhudson> jelmer: hey, do you know if pysvn is prone to giving you unicode strings?