=== cjwatson_ is now known as cjwatson [02:28] with bzr-svn, can i give a specific revision #? i want to clone a repo with thousands of revisions and want to pull 1k revisions per night. [02:28] meteoroid, yes, see -r [02:29] jelmer - with svn-import ? [02:30] or is it easier if i branch each of the root-level 'project' areas with branch, into a shared repo? [02:30] ahh, there's no option for svn-import for that [02:30] yeah see [02:30] i need to svn-import something with over 50k revisions [02:30] and i'd rather not crash their server ;) [02:30] as svn is known to leak and take apache town altogether [02:31] meteoroid: doing `bzr pull -r $x` where $x is incrementing by a few hundred (or whatever) each loop will probably do fine. [02:33] AfC: but, how should i start it all off? i'd be best off, as i understand, with an svn-import, and i'm concerned if the branching will be properly recognized otherwise, which may increaase the amount needing to be transferred, and the storage used, exponentially.. [02:33] i mean this is a huge repo [02:43] meteoroid: just start it off with a branch at -r 10 or something. [02:43] hm. [02:43] maybe just -r 1 [02:43] meteoroid: (actually a `bzr init --rich-root-packs` or whatever == empty branch works too) [02:43] hm [02:43] and then a URL on the first pull [02:43] explain more about the last statement.. [02:43] i'd love to write a general purpose utility for branching large svn repos as a shell script or something based on your advise [02:43] advice [02:44] meteoroid: For what it's worth, I wrote about in making an initial Bazaar conversion of a monster branch (GTK, as it happens) which you might find interesting. http://research.operationaldynamics.com/blogs/andrew/#bzr-branch-of-gtk [02:44] thanks! [02:44] GTK is quite possibly a good bit larger than mine [02:44] how many revs when you started? [02:44] hm 15k revisions [02:45] except for memory limitations on my 1GB Ubuntu slice, I was able to branch over 5k revisions of a 10,5 revision repo [02:45] but i figure if i come up with a gentle strategy for the 52k+ rev repo, it will work for others [02:45] 15k revisions i could probably branch easily on a bigger box in one overnight run [02:46] meteoroid: the key point is that once *someone* has done the conversion once, no one else has to. They can all just `bzr branch` from you, and then later update their branch via `bzr pull svn+ssh://...` [02:46] no i totoally understand [02:47] i'm just concerned that i can't be busying up the server for more than a handful of hours [02:47] 5k revisions took long enough [02:47] i want to spend a month or so pulling around 100k revisions across a few repo [02:59] meteoroid: do you have access to this server? The other way to do this is to get an svn dump, transfer that in toto, and then operate on that elsewhere [03:00] meteoroid: that's how Jc2k built the bzr-mirror.gnome.org set [03:39] AfC : the server admins may be hostile, but i can work on it.. [03:40] i know that'd be the best way. an svndump i could load and then pull the latest from [03:40] i know once i have a copy that updates will be very light [03:40] and that's a big reason i've chosen this approach [03:40] to provide greater branching freedom in our community [03:40] but i'm not sure all people will be welcoming at first [03:40] until we can contribute back.. [03:41] so it's a chicken egg problem and i'd like to just spend a few weeks pulling a couple k rev at a time to not overload upstream server. [03:41] but if we need to surf the social issues, we will === mark1 is now known as markh [04:50] lifeless: are you around? [04:51] Quick question. In the color discussion Jonathan Lange suggested taking the terminal coloring code from twisted.trial: it supports handling things like Win32 already [04:51] What's the bzr policy for taking code from other projects like that? [05:00] twisted license is MIT, ya? [05:08] pickscrape: depends on how it will be taken; if its being copyied it needs to be a) licence compatible and b) not large enough to trigger the need for assignment [05:11] Basically it is two classes from here: http://twistedmatrix.com/trac/browser/tags/releases/twisted-8.1.0/twisted/trial/reporter.py [05:16] If it's not feasible I can work on my own implementation instead. I wasn't going to use them verbatim anyway. [05:24] * meteoroid has had some convo recently with RMS regarding GNU licenses and python "dynamic linking" or loading of .py files, which is fuzzy license territory apparently, despite my previous beliefs.. [05:29] I think I'll just write my own. Less hassle. :) [07:32] lifeless: ? [07:54] I am having problems in building bzr plugins on windows using MinGW (dir_walk plugin), is it a known issue [07:55] Since I have been benchmarking it for emacs repository, any performance gain would help [07:55] I believe that dir_walk on windows does speed up things, am I right in my assumption? [08:02] anyone online now? [08:09] xshelf: if you can get bzr 1.6 building it should already have optimized dir walking for win32. [08:10] i am using bzr from the bzr.dev [08:10] right - and can't get the extension building? [08:10] i must say it is much better than I (dhruvakm) did a earlier benchmark for emacs [08:10] yes, I get an error in building the extension [08:11] python setup.py build -c mingw32 [08:11] something to do with Py_ssize type [08:12] hrm - python 2.4? [08:12] it used to build just fine with vc 2003 and my installed python was built using the same compiler [08:12] python 2.5.2 [08:13] strange - it has Py_ssize_t - what is the error? [08:13] i had to update to vc 8, python does not have vc8 distro [08:13] I am therefore forced to use mingw [08:13] error: bzr.dev\bzrlib/_walkdirs_win32.pyx:65:15: Syntax error in C variable declaration [08:16] I'm afraid I really can't help I'm afraid :( I think the author of that code does use mingw, so its possible opening a bug report will get results. [08:16] you could always build your own python with vc 8 ;) [08:17] it needs a whole lot of stuff like sql-lite, bdb... [08:17] yeah - but so would building from vc7 - but I agree its a PITA :( [08:17] i tried hoping it would be quick, just gave up after sometime [08:18] i really like the PERL bundling [08:18] oh - right - if you could use vc7 you wouldn't have to build python itself! [08:18] I have built PERL using cygwin, mingw and msvc with no sweat [08:18] true [08:19] and you cannot build python with mingw [08:19] i wonder if we have some python devs here listening, get python to build with mingw. I build emacs regularly with mingw and it works fine [08:20] some people have managed and contributed makefiles, but they are rarely kept up to date. It all relies on volunteers, and on Windows at least, builds using a non-standard compiler are less interesting as you can't get any pre-built binaries for it. [08:20] you have CMake [08:20] at work, we just moved to cmake [08:20] it generates platform/compiler specific build files from a common set of cmake files [08:21] I am really impressed by cmake. BOOST is moving towards cmake (giving up their home grown bjam tool) [08:22] i never realized you are the same Mark of Python win32 ext [08:22] heh - yeah [08:23] great, you really have brought Python to the windows world, thank you for that [08:23] working on bzr 1.6 binaries as we speak - but msvc7 built ones ;) [08:23] my pleasure :) [08:23] thanks! [08:23] welcome [08:23] yes, building with msvc 7 was easy [08:24] i made a blunder of uninstalling it when i had to install vc8 [08:24] i thought i would save some disk space :-( [08:24] can't you reinstall it? [08:24] i have to run to IT, they will ask questions and get license approvals and all the corporate noise... [08:24] heh - bugger [08:25] i will try that anyway on monday [08:25] so - bzr 1.6 should be much faster than 1.5 even without that extension [08:25] it sure is [08:25] i was part of the emacs mail chain opposing adoption of bzr [08:26] although that extension makes a big different for big trees. Quoting John, its author: " With this patch, doing 'bzr status' in a mysql tree on Win32 changes from: 4.2s => 0.64s, or about 6.5x faster. It is quite noticeable when your command prompt hangs for 4s versus returning in < 1s." [08:26] the turning point was when RMS posted that we have to adopt another GNU project and help make it better [08:26] Yes, I keep running such commands too often in my development [08:27] i thought of writting something like using inotify for windows [08:27] so that, my changes are tracked without walking directories [08:27] The binaries I've got a looking pretty good. If you ping me on Monday I can make sure one is up for you to play with if you like [08:27] i need time to do that, something that sysinternals filemon (procmon) does [08:28] i will do that, thanks. [08:28] I will try getting vc 7 so that I can build and play myself [08:28] yeah, tortoise will grow a file-watcher [08:29] but bzr status is *very* fast - we are already much much faster than tsvn on most trees I've tried [08:29] another question: can bzr seamlessly with perforce? [08:29] I am looking for something like git-p4 [08:29] not afaik, but its an opportunity waiting to be taken, along the lines of bzr-svn ;) [08:30] i am forced to use p4 at work, would love to use just bzr for emacs and work [08:30] but google may well prove me wrong - I'm not that much in the loop that I'd know about it [08:31] i have just started looking at vcp from perforce. The svk folks had used vcp to support seamless operability with p4 but dropped it in their new releases [08:31] let me do some ground work and keep this channel posted [08:32] well, my family is back, will catch you on monday [08:33] have a nice weekend. I have to tear myself away from this computer... [08:33] cheers! [08:45] morning [08:46] i just built bzr from the repo and it bears the label:bzr1.7dev. when is 1.6 going to be released? [08:47] gour: the 13th is the plan (see the release announcement for 1.6rc1) [08:55] spiv: ta. i'm looking at release notes, but there is nothing [08:56] anyway, will bzr come back to regular (aka monthly) releases? [09:45] Feature request proposal: I'd like to have diffs with diff's "-c" option. It should be possible to set this in the conf file, so that not only "bzr diff" but also "bzr commit --show-diff" benefit from it. [10:00] hello -- I'v got a question: what is the best way to incorporate the contents of repository B into A? At the moment I'm only aware of doing a merge... Is it the right way, or is there another approach? [10:00] sorry, I'm confusing repository with branch [10:01] but on the other side doing it with shared repositories would also be interesting [10:02] btw, with "contents" I mean the complete history, too [10:05] jonnydee, I think you could also append the changes of one branch to another branch with re-basing. Haven't done this yet, though. [10:05] rebasing...ok. It sounds like this might be a solution... [10:05] Are the contents of both branches disjunct? [10:06] Yes, they generally are. The reason why I'm asking is: I would like to write a plugin that is able to [10:07] convert a branch/repository in such a way that the root of the repository/branch is moved one level up in the directory tree [10:07] (while preserving the directory tree) [10:09] for example, if I have a branch in: /tmp/repo/jonnydee/ [10:09] then I would like to be able to move /tmp/repo/jonnydee/.bzr to /tmp/repo [10:10] which means the directory jonnydee is now versioned, too [10:10] But repo is not a branch, but a ... well ... repo. [10:11] I think that this would create some administrative collisions, although I must admit that I don't knot the Bazaar interna. [10:11] yes, I know. But I need to handle any combinations.... [10:11] What's the eventual purpose you're thinking of? [10:12] Well the idea came from a discussion here [10:13] someone had exactly this problem: he versione controlled a specific subdirectory and decided to extend version control over to the parent directory [10:13] in his case the solution was simple: [10:15] I can imagine his solution already. You may proceed with explaining a case that is not so simple. ;-) [10:16] well, sorry, I'm a bit confused at the moment -- I think it's too early in the morning. [10:16] But you got the idea [10:16] Not really, because your example can be easily solved without new features. [10:17] the more complicated cases are where one extends version control to a parent directory that has subdirectories containing other branches/repositories.... [10:17] ...and now ;) [10:17] ??? [10:22] with other words, the general case is: you strat from an arbitrarily deep subdirectory in the file system, which contains the repository/branch you want to extend one level upwards. when you go one level upwards you might encounter a shared repository, or nit [10:22] not [10:22] Even then it's simple: Go to the root of your branch (not repo!) and type "bzr mkdir branchname". Move all other top-level files/directories into that dir with "bzr mv". Then move the desired files/dirs from .. into the branch root dir, and add them with "bzr add". Commit everything, voilPONG :niven.freenode.net [10:24] but I do not only want the contents but also their history (if any) [10:25] so lets assume my parent directory is a shared repository. now this means all the branches contained in the shared repository need to be combined into a single branch [10:26] if it is not, then your solution comes into place -- with one exception [10:27] if some subdirectory of the parent directory contains a branch itself (which may be located in an arbitrarily deep subdirectory) then there is no simple way like adding the files [10:27] because we also want the history [10:29] well, while I'm explaining this I am starting to doubt whether such a plugin is woth the work.... [10:29] Do the branches have a common ancestor? [10:30] maybe they have, maaybe not [10:30] In a general case, you might have or are likely to have common ancestors [10:32] such a plugin would be a nice-to-have for me, but the question is: how often is it needed to extend the root to one (ore more) levels up.... [10:33] anyway, I think you see it is not that easy when considering the general case [10:33] ok bye Bronger - cu ;) [10:35] In any case, you have to move all dirs to their final positions in their respective branch, so that there are no collisions anymore when you unite them. If they have a common ancestor, I'd do a merge, if they have not, you have to do a rebase I think. You can rebase in both cases though. Given that this is a rare use case (in case of no common ancestry even *very* rare), and given that it is feasible to do it manually, I don't think that it's worth a plugin [10:38] Bronger, I thin you are right.... You know, I'm trying to get into Bazaar development and I thought writing a "simple" plugin would be a good starting point....but this turns out to not be the simplest plugin.... [10:38] especially in comparison with the benefit one would gain.... :( [10:39] ok, so thank you very much for your help and feedback!!! :) [10:40] and sorry that I've been wasting your time... [10:42] jonnydee: bzr merge -r 0..-1 path/to/otherbranch [10:42] Not at all. It help me to understand those DVCS more and more. [10:44] thinking positive is always a good idea ;) [10:44] jonnydee: the key thing there is revision 0, being a virtual revision every revision dag inherits from [10:45] jonnydee: so even if the two branches don't have anything in common otherwise, you can force them to merge that way [10:45] LarstiQ: When using this command, do I get the complete history from the other branch, too? [10:45] jonnydee: as came up (I didn't read all backlog), you might want to move directories beforehand to keep things from clashing. [10:45] jonnydee: yes. [10:46] that sounds like magic :) [10:46] well, it's just normal operation really :) [10:46] * LarstiQ runs off for the day [10:46] jonnydee: have fun :) [10:46] thanks larstiq [10:46] :) [10:48] well, I'm going to play a bit. maybe, I'll implement simple cases, just for fun....and if I still have fun, then maybe I consider more complex cases... [10:49] Thanks a lot to Bronger an to LarstiQ :) [10:52] jelmer: I have a couple bzr-svn questions. With that root revision patch, what exactly was broken, and did data get corrupted or anything, or would bzr just traceback? [10:54] jelmer: Also, with changing the file ID map version, what exactly does that mean? [10:54] Peng_: The root revision patch is more just paranoia at this point [10:55] gour: yes, back to monthly releases is the plan [10:58] jelmer: Okay. [11:00] Peng_: the file id map version was changed because older versions of bzr-svn may write invalid data to it in some corner cases [11:01] rare, but still [11:03] jelmer: I liked the suggestion for bumping the version to 1.0, btw. [11:05] I dunno about 1.0, but I would vote for 0.5. [11:11] spiv, I'd rather wait with any sort of major version change until the mappings change again [11:12] jelmer: fair enough. But 1.0 soon is justified I think, it is a pretty mature tool now. [13:11] markh: ? [14:55] jelmer: remember the other day i said i was working on a bzr-svn blog post and would like you to review it before i published it? i have it ready... don't suppose you have an email i could send it to? [14:55] rocky, Yeah, sure - my email is jelmer@samba.org [14:58] sent [15:00] brb [15:10] rocky, just curious - what sort of blog software do you use? [15:10] TextPress [15:10] i'm a python fanatic [15:10] www.serverzen.net [15:11] i just blogged on settnig up TextPress ;) [15:14] rocky, I don't see any errors [15:14] any obvious misuse of bzr? i mean, am i using bzr in an acceptable fashion? :) [15:15] rocky: heh, no that all looks alright [15:15] jelmer: I think Wilmer van der Gaast sits a few places from me at work. ;-) [15:16] dato: Ah, cool [15:16] dato, he's a good friend of mine [15:16] dato, I was over there at Google a couple of weeks ago [15:16] O'RLY? [15:16] jelmer: alrighty then [15:16] we missed the chance to meet, then :( [15:16] now the question is do i wait a bit since i blogged just a few days ago, or post immediately :) [15:16] dato, You're working for google these days? [15:17] jelmer: I'm doing an internship this summer, at least for now. [15:17] dato, ah, nice [15:26] * pickscrape gets bzr diff --color working for the first time :) [15:28] rocky, thanks for the link to textpress, I'll check it out [15:28] dato, ask him about french-irish girls in boots :-) [15:28] jelmer: np, my blog post should get you started fast [15:58] hi -- is there any way I can use 'bzr branch' to read a .bzr repo which is served by a webserver at https://server/site/.bzr -- when this is passwd protected? [15:59] I can access the dir using a browser by entering credentials into the basic-auth dialog [16:00] nexus10, sure, bzr branch https://user@server/site [16:01] it should just ask you for the password [16:01] beuno: :-) [16:01] beuno: now I feel *really* smart... [16:02] nexus10, I went through the same thing before I learned that, so, welcome to the club :) [16:03] :-) [17:22] beuno: that bzr branch https://user@server/site worked splendidly, thanks a lot. [18:41] Hi all. [18:41] How do I use bzr-svn ? [18:46] rindolf: http://www.serverzen.net/2008/08/09/starting-with-bazaar-bzr-svn [18:48] rocky: thanks. [18:53] beuno: Loggerhead question: If a commit message is "foo\n\nbar", the /changes page will show "foo" in the summary. Then if you click the expand button, it changes to "foo bar". Is that intended? [18:54] rocky: python: subversion/libsvn_ra/ra_loader.c:944: svn_ra_get_log2: Assertion `*path != '/'' failed. [18:54] Peng_, it's expected rather then intended [18:55] we strip out all HTML to show in that context [18:55] including \n [18:55] so we can show a plain commit message [18:56] there may be smarter ways of doing that though [18:56] pickscrape, ping [18:56] Ah.. [18:57] Well, I'm gonna go. Have a nice day. :) [18:57] Peng_, you too [19:03] https://bugs.launchpad.net/bzr-svn/+bug/234010 - hmmm... [19:03] Launchpad bug 234010 in bzr-svn "abort: svn_ra_get_log2: Assertion `*path != '/'' failed (dup-of: 229419)" [Undecided,New] [19:03] What should I do? [19:03] Launchpad bug 229419 in bzr-svn "WorkingSubversionBranch.test_create_checkout fails" [High,Fix released] [19:08] Anyone? [19:15] rindolf, jelmer is the right person to talk to, but probably not on a weekend [19:16] he'll see the bug eventually, and help you out [19:16] beuno: thanks. [19:17] :) === _mathrick is now known as mathrick [20:01] is it normal that the bzr PPA for hardy doesn't actually contain bzr ? [20:03] Heh. [20:04] Oh, I'm using the beta PPA. I forgot. [20:04] asabil: Yes, it is. Someone published 1.6b3 for Hardy, and you can't revert it back to 1.5. [20:04] asabil, it will contain only releases from now on [20:04] so beta PPA for betas/rc's, and bzr PPA for releases [20:05] asabil, https://edge.launchpad.net/~bzr-beta-ppa/+archive [20:05] beuno: hmm ? but it doesn't even contain bzr 1.5 [20:05] asabil, right, as Peng_ pointed out, something went wrong, and it had to be removed. 1.6 *will* be on there [20:05] beuno: I don't really want bzr 1.6 [20:05] and everything should be fine from now on [20:05] oh oki [20:22] Hello All [20:42] beuno: pong [20:42] pickscrape, hi :) [20:42] Afternoon :) [20:42] I've been looking at your patch/branch [20:43] I can't figure out *what* it does [20:43] It converts the path string at the top of the file and directory browser pages into breadcrumb links [20:44] So you can click on the parts of the path to browse to that directory [20:44] * beuno tests again [20:46] pickscrape, cool. Can you make the file browsing do that too? [20:47] I thought I'd got them all... Which view are you looking at? [20:47] pickscrape, when you end up in the actual branch [20:48] It should already work [20:48] in /files [20:48] it doesn't for me :/ [20:50] I just browsed into loggethead/loggerhead/static/css locally. At the top I see "viewing /loggerhead/static/css for revision 206" [20:50] loggerhead, static and css are all links for me [20:51] pickscrape, aaaaaaah [20:51] I was expecting the path to continue [20:51] so when I jumped into the file view [20:51] I expected to be able to go back [20:52] Ah, you mean the branch name part to the left of the 'viewing' [20:52] right [20:52] it deserved some thought [20:52] Yes, I didn't want to play with that because my aim here was to add the links without changing the presentation at all [20:53] But I suppose I could add the path to the branch to the left of the branch name. [20:53] The branch name itself should probably be a link too. [20:54] yeah, this is a big improvement [20:55] The problem is once you add paths, you start to expect seeing the branch's directory name, and not nick. [20:56] So it might be better doing it with pure directory names and displaying the nick somewhere else. [20:56] I have to pop out for 5 minutes... brb [20:57] pickscrape, ok, cool. Just one comment on the branch, and it's mergeable. You have no link to the root [21:08] pickscrape, I gotto run, but if you can add a link to the root, that'll be great. If not, I'll merge anyway when I come back, as this is very practical :) [21:19] beuno: see my export patch ? [21:25] beuno: not sure if you mean a link to the branch root or to the root directory that loggerhead is serving. [22:47] hello, wondering if I can nest branches one into another. i.e. in my project folder I could have a subfolder which contains libxxx. I'd like libxxx to be referring to an external branch. is that possible ? === Snaggen_ is now known as Snaggen [23:26] trying again: hello, wondering if I can nest branches one into another. i.e. in my project folder I could have a subfolder which contains libxxx. I'd like libxxx to be referring to an external branch. is that possible ? [23:27] tacone: Well, you can put a branch inside of another one. The outer one will completely ignore the inner one. [23:27] <[cliff]> hi all, I'm trying to push my project onto a new launchpad branch and I'm getting this error: Transport operation not possible: http does not support mkdir(). any thoughts? [23:28] [cliff]: Run "bzr launchpad-login your_username" [23:28] <[cliff]> aha! :) [23:28] hm, why is an svn-import pulling only one portion of a repo? i branched earlier, deleted that branch - maybe the svn cache is fudged? [23:28] jelmer? [23:29] Peng_: is it possibile to get the outer branch to "include" the code of the inner one ? [23:30] <[cliff]> Peng_, good stuff, everything works. thanks! [23:32] only getting 82 rev of a repo with 577 [23:34] even with --all