[00:41] poolie: pulls of lp down to 49s [00:41] from? [00:41] poolie: sneaking up on it, 8 seconds wasted on _ensure_real bs [00:41] johnny: approx 2-3 minutes [00:41] yes.. wwe sure are ensuring real bs :) [00:41] johnny: it's not a very scientific test. [00:41] hehe jk [00:42] spiv: you have the same disk format right - knits at both ends ? [00:42] johnny: because I'm not pulling the same revisions each time, I'm just doing updates occasionally. [00:42] lifeless: I think so, I'll double-check [00:42] spiv, reviewed [00:42] if you had a quote bot in here.. that'd totally be in there :) [00:42] lifeless, what bs is that? [00:42] not that i think you should have one.. [00:42] but still [00:42] poolie: thanks! [00:43] so, do you guys participate in that vcs mailing list? [00:43] poolie: its being tracked down still [00:43] revctrl ? [00:43] poolie: I'm guessing tags or something like that [00:43] lifeless: yes, both ends knits [00:43] crazy people ... [00:43] too smart for their own good :) [00:43] johnny: yes, I do [00:43] i know the monotone folks are involved [00:44] that's what i use with other projects [00:44] johnny: theres a strange mix of noise and signal [00:44] monotone is great... but it's written in C++, and has not so good tools :( [00:45] and git is insane... fast, but insane [00:45] now i'm looking into hg as well, which i had never even previously considered [00:46] lots of amazing projects are using all of these systems [00:46] makes it hard to choose.. [00:48] poolie: I'd like a quick call if thats ok [00:48] 1m [00:48] I think you forgot a ^V[[; there [00:49] meh, ^[[; [00:49] heh [00:49] are you controllable by ansi sequences? [00:49] not anymore, I got upgraded [00:50] css? :) [00:54] so - quick call or not ? [00:54] go on then [01:03] thank god for noise cancelling headphones [01:03] my house phone needs a mini-rca jack [01:05] Hmm, revctrl list? I'm in the channel, but it's almost dead. [01:05] What type of noise cancelling headphones? [01:06] Also, crapcrapcrap. I have a Python installation with no bzip2 support, and now bzr requires it. [01:09] Peng, on what kind of machine? [01:09] hm i wonder if we really need it or not? [01:09] Spads, so is that review ok? [01:09] i mean spiv [01:10] poolie: It's a Debian shared web host. They suck at Python (2.3.5 and 2.4.1 installed), so I've compiled 2.5.1 in my homedir, but it winds up without bzip2 (or readline) support. [01:10] Humm. [01:11] Peng: 2.4.1 should work fine for bzr [01:11] poolie: yes, we do - see the thread where I asked first :) [01:11] Peng: actually bzr has required it for some time [01:12] Peng: i guess the bzip2 headers aren't on the host [01:12] Peng: just now needs it in another area [01:12] poolie: yep [01:12] Peng: Sennheiser 350's [01:12] Peng: they make the construction going on upstairs nearly entire go away [01:13] lifeless: ooh, nice [01:13] mwhudson: Yeah, that's probably the case. [01:14] lifeless: Yeah, but I want 2.5. And even if bzr already used bzip2, I'd so far avoided that part of the code. [01:15] lifeless: That sounds good. I have a pair of Sonys that mostly make my loud computer fan tolerable . . [01:19] Peng: what do you want 2.5 for ? [01:23] hi all. umm, if i allready have set up a svn repository - should i use svn as repository? ... i just started reading the user guide ... well, if i dont use svn ... is there any advantage or disadvantage? [01:23] the best way to use bzr is bzr-all-the-way [01:24] lifeless: Just for stuff I screw around with, not bzr. Hmm, I could probably switch bzr to the system 2.4. [01:24] the stuff about svn is because many users we get have been using svn and have a number of systems to migrate [01:24] so its easiest for them to migrate incrementally [01:24] and some users have to work with other folk that use svn, and who won't change, but our users still want to use bzr because its nice :) [01:26] ok so, is it hard to migrate svn repos to bzr ones? [01:27] not at all [01:27] install bzr, bzr-svn and do a bzr svn-import [01:28] do i need to install bzr on the server? [01:28] it depends how you configure it - and that depends on what you want. Bzr can run over sftp which requires no special server side support, or over ssh which does want a copy of bzr on the server [01:30] can bzr have its own authentication mechanism like with svn and http? [01:31] bzr uses your authentication; you can have it be authenticated by apache, - if what you mean is 'does bzr have to use my passwd database' the answer is no. [01:31] in fact someone has recently done a ssh server for bzr with its own auth, but I've not tested that myself. [01:31] i dont want to give anyone shell access - sftp also uses ssh authentication [01:32] PAM [01:32] or whatever [01:32] i need an own authentication mechanism for bzr repos [01:32] sure, that ssh server I mention will do that [01:32] maybe i should read the admin guide if there is any for bzr [01:32] it doesn't use pam [01:33] and you can do restricted ssh keys by the way - that prevents anyone having shell access (you can have /bin/false as the shell too for extra security) [01:33] docs.bazaar-vcs.org IIRC [01:35] ok thank you === Verterok is now known as Verterok_ [02:10] umm, does the trac-bazaar plugin work and is it as good as the subversion one? [02:13] abentley: shaved about 100ms off of 5.5 seconds user time for indexes so far - using checkout ---lightweight of a treeless branch as my test [02:14] abentley: still mainly refactoring to make it amenable to overhauling [02:14] c1|freaky: sorry, I don't use trac, can't really comment [02:14] ok bo [02:14] ^^ [02:14] np [02:15] it seems bzr.dev push hasn't received the same love as bzr.dev pull yet? [02:16] * mwhudson watches revisions.kndx trickle upstream [02:19] mwhudson: not yet [02:19] mwhudson: although packs make it better [02:20] i'm sure [02:20] i guess i should convert my devpad repo over to packs [02:21] lifeless: Wow, I can't get such steady timings on that workload to see 100 ms go away. [02:22] Still, it sounds like movement, so that can't be bad. [02:24] Ha! I got it to work with Python 2.4 and virtual-python. [02:28] abentley: the userspace timing is pretty consistent for me [02:28] abentley: the wall clock is pretty jumpy [02:28] Ah. [02:28] I found while optimising commit to just ignore wall clock [02:28] and run it several times [02:29] that or close everything down [03:12] i just did a "bzr add foo" and "bzr add bar", but changed my mind about adding these file before doing a commit [03:12] how can i tell bzr not to version these files after all ? [03:12] bzr remove --keep [03:13] thank you [03:13] what about "bzr remove --new" ? [03:14] how is that different from --keep ? [03:16] remove --new automatically selects the files just added [03:16] so i think you want --keep --new [03:17] i'm going to look into the pqm failure i got the other day, merging some already-approved branches [03:17] --new is not needed [03:17] if he names the files it's not needed [03:17] unless you want bzr to scan [03:18] spiv, ping me when you've merged your thing, please [03:18] poolie: ok [03:18] lifeless, random comment [03:19] i find pqm is disruptive to my concentration because it makes me open my mail client, which is the mind-killer [03:19] i should probably procmail them into a special box and look at only that [03:19] poolie: fear! [03:19] poolie: but thats a good comment [03:20] poolie: pqm should blog or something [03:20] yeah [03:20] theres an open bug for showing recent results in the status page [03:20] * fullermd doesn't wanna let pqm pass thru him. [03:20] or even dump them in an indexed web directory [03:20] i realized this is more than half of my unhappiness about its asynchronicity [03:20] thanks for your help, guys [03:21] after a few hours of reading i still dont know if i should switch to bzr from svn or not ... what do you say? [03:21] switch [03:22] you'll love it [03:25] just try it out, it's very easy to make the transition [03:25] my only problem, at the moment is [03:25] c1|freaky: And you can go back if you want to. [03:25] how can i have people to easily authenticate for submit [03:25] I hit that snag too. I use bzr+ssh:// [03:26] in a way without PAM authentication [03:26] i dont want to give everyone shell access who should be able to access a repository [03:26] c1|freaky: there was a post on the list a few days ago about a plugin that implements its own ssh server [03:26] c1|freaky: *also* you can "Use PAM but not give everyone shell access" [03:27] there are myrisad ways to restrict ssh [03:27] they'd still have ftp access and a home directory [03:27] that sucks [03:27] *myriad [03:27] c1|freaky: Your system requires all users to have FTP access? [03:27] no [03:28] http://tadek.pietraszek.org/blog/2006/10/18/restricted-shell-account-ssh-and-subversion/ [03:28] adapt to bzr. [03:28] thx [03:29] im still not sure [03:29] ill have a rest now [03:29] and continue later [03:30] 20 minutes or so [03:38] oh i see, having times in the trace log makes it a bit harder to test, foo [03:41] Anyone know anything about Bug #189478? [03:41] Launchpad bug 189478 in bzr "man page has errors when typing "man bzr"" [Undecided,New] https://launchpad.net/bugs/189478 [03:42] c1|freaky, the issue is never between bzr and subversion.. bzr always beats subversion :) [03:42] i installed bzr manually because the ubuntu version still is 1.0.0 [03:42] its' between bzr,mercurial,git,and monotone [03:42] with subversion i can do http authentication [03:43] i want protected repositories where the user don't have ssh shells [03:43] an own authentication system [03:43] the rest of svn won't make that worth it [03:43] monotone has it's own [03:43] based on public keys [03:43] git can use http iirc [03:44] on the downside, the openssh people are almost certainly better at writing authentication systems than anyone else [03:45] the new monotone uses ssh agent for local keys which is nice [03:45] * johnny doesn't like having system accounts either [03:46] how do i merge a patch into parent branch? [03:46] im using trac and i just want to provide some friends with repositories, and with the best possibilities [03:47] trac will always work best with subversion, cuz they designed it badly [03:47] it wasn't abstracted enough [03:47] so.. for 0.12 it should work nicer with distributed vcs [03:47] there is a bzr plugin [03:47] as well as git and mtn ones [03:47] Aloha: merge parent into your branch, then push to the parent [03:47] Aloha: or checkout parent, merge your branch into it, then push [03:48] bob2, the patch is from someone else. how do i merge their patch? [03:48] Aloha: is it really a patch, or is it a bzr branch or bundle? [03:49] bob2, its a bzr -o send patch [03:50] Aloha: "bzr merge patch", where patch is the file generated by "bzr send -o" [03:50] Aloha: "bzr merge patch_file" [03:50] spiv, bob2, thnx [03:50] so merge that into my branch and then push, yeah? [03:51] yup (don't forget to commit after merge) [03:51] Aloha: a merge is like any other change to your working tree, you need to commit it. [03:51] Aloha: but otherwise yes. [03:52] bob2, doesn't bzr -o send create a patch based off of parent? [03:52] so does that mean my local branch has to be synced with parent? [03:53] it has to be at or ahead of it, but you'd need that before you could push to it anyway [03:53] Aloha: no it doesn't have to be synced [03:53] Aloha: it uses the revision graph to generate a good diff [03:54] lifeless, thats freaking cool [03:54] * Aloha loves bzr [03:54] lifeless: it'll automaticaly fetch needed revisions? [03:55] bob2: if the other branch is ahead, there is nothing to fetch; if the other branch is behind the new revisions are in the local branch [03:57] hey guys, ill migrate from svn to bzr but i need some help. i have a 3 svn repositories on my server. i now want to "convert" them to bzr. how to do that? (i have one of them checked out) [03:58] (i just reinstalled kubuntu on my laptop that's why i dont have them all yet [03:58] read the docs [03:58] there's a migration section [03:58] in which documentation? [03:59] onthe main site [03:59] for bazaar [03:59] user guide? [04:00] no [04:00] it's right on the site [04:00] http://bazaar-vcs.org/BzrMigration [04:00] does the pre-commit hook allow for modification of a versioned file, or addition of an unversioned file? [04:00] oh ok found it [04:00] :d [04:00] thank you [04:01] rolly: no, I don't think so [04:01] modifying versioned files === bBAD [04:01] BAD!!!!!! [04:01] haha [04:01] this isn't CVS [04:01] rolly: we're looking at a start-commit hook for things like that [04:01] it's not ALWAYS bad [04:01] keywords will never be expanded [04:01] igc: that would be so awesome [04:01] it would make my dreams of versioning my database come true [04:02] jelmer recently raised a bug on lp for that [04:02] * igc looks [04:03] rolly: not currently; we will add one soon [04:03] * johnny tries to think of a way you could modify a versioned file that was already committed [04:03] lifeless: cool! Yet one more step ahead of SVN [04:03] as that sounds like it would invalidate it [04:04] bug 186422 [04:04] Launchpad bug 186422 in bzr "Ability to modify the tree from a pre-commit hook" [Wishlist,Triaged] https://launchpad.net/bugs/186422 [04:04] oh.. pre commit would be fine, but you'd have to be super careful :) [04:05] i idled in #monotone in otfc for awhile, those guys were always talking versioning theory [04:05] pretty intense stuff [04:05] wait, so I _can_ modify with pre-commit? [04:05] * rolly reads [04:05] i never rode the svn train [04:05] i knew it was a lame idea [04:06] johnny: you can't alter history [04:06] johnny: but versioned file refers a file that is managed by bzr in general [04:06] johnny: so altering a versioned file in a working tree is what you do with emacs/vim etc all day [04:07] Maybe I can get by with a little hack to bzr.bat [04:09] about 150ms down stably now [04:20] when i checked out the svn repositories to a local place ... now i'd have to create the bzr repos on the server, right? but before that, i need an authentication mechanism and i still dont know which i could use [04:20] :(( [04:20] i dont want ssh or any other PAM authentication [04:22] It sounds like an excellent solution to me. I mean, that's the whole point of PAM... [04:23] i dont want to give everyone ssh access [04:23] :(( [04:23] You don't have to. You can just give them bzr+ssh access. [04:25] how to do that? [04:25] and where should i create the repos? should i create a user called "bazaar" [04:25] and in that directory create the repository [04:25] for the different projects? [04:25] 200ms now [04:25] c1|freaky: I suggest you read the users guide and play with bzr *before* migrationg [04:26] c1|freaky: start with it on your workstation and get a feel for how it works. This will help you a lot when you start thinking about group setups [04:26] :( [04:26] i have no project im starting right now i just want to migrate [04:26] hmm [04:26] You can use PAM to authenticate against a database or another file or something other than the system password file. You can set allowable commands for the users via ssh. They don't have to have home directories or real shells. [04:27] c1|freaky: then just do 'bzr branch svn://urlhere local-disk-path' [04:27] c1|freaky: and you'll have your existing projects to play with [04:27] ok cool [04:27] :D [04:27] ill try that, thank you [04:27] c1|freaky: I suggest that you learn the basic tool first because otherwise you are going to be learning new concepts in several directions all at once. [04:28] i've allready read the whole user guide [04:28] c1|freaky: and thats tough; we have much more facility than svn, but we *different* - you don't /need/ a "repository" for instance. (You can have one, but its not needed) [04:28] ok ill remove my svn checkouts [04:28] and use the bzr branch dirs then [04:28] instead [04:28] and later when i created the repos on the server [04:29] ill use the bzr repos and commit them [04:29] yes [04:30] abentley: still around? [04:30] lifeless: Yup. [04:30] Bazaar has encountered an internal error :\ [04:30] ubotu: paste [04:30] pastebin is a service to post large texts so you don't flood the channel. The Ubuntu pastebin is at http://paste.ubuntu-nl.org (make sure you give us the URL for your paste - see also the #ubuntu channel topic) [04:30] PROPFIND request failed [04:31] abentley: total time for bzr.dev's lightweight checkouts - of that what fraction is index from memory? [04:31] abentley: (I'm not using your branch yet) [04:31] I've shaved 5% off basically at this point [04:31] http://paste.ubuntu-nl.org/55055/ [04:31] On my branch, it's 50%. [04:31] c1|freaky: what url did you us ? [04:31] Total index. [04:31] abentley: right, I'm asking bzr.dev :) [04:31] abentley: IDK is fine [04:32] i used: http://code-1.de/svn/freaky-tcl/ [04:32] c1|freaky: try svn+http:// [04:32] k [04:32] c1|freaky: I hear that that may work better [04:32] failing that, jelmer ^^ [04:32] lifeless: I use dev as my sample data :-) [04:33] abentley: yes, but bzr.dev's _code_ [04:33] lifeless: still the same [04:33] c1|freaky: do you have write access to the branch? [04:33] c1|freaky: if so what url do you use ? [04:33] i use http://code-1.de/svn/freaky-tcl/ i _have_ write access but it doesnt ask for login information? oO [04:33] still anyone can read [04:33] oh interesting [04:34] I understood, was explaining my dumbness. [04:34] so i dont really need login information [04:34] abentley: ok, I don't understand and am confused :) [04:34] c1|freaky: its not that, I was hoping for a different url. We'll need some help from the bzr-svn author at this point I think [04:34] c1|freaky: that SVN repo has many projects in it. [04:34] dev spends 53.43% in bisect_multi_bytes. [04:34] abentley: thanks [04:35] spiv: i just put them all together as one projects - my tcl scripts [04:35] -s [04:35] c1|freaky: https://answers.edge.launchpad.net/bzr-svn [04:35] I suggest asking a question there, and looking in the bug tracker too [04:35] (The percentages are similar for dev and my code because I also eliminated some lookups) [04:35] At least, I assumed so. [04:36] If it's cached, maybe there's another reason. [04:36] c1|freaky: hmm, so there's no trunk/ or branches/ or tags/? Just one branch at the root of the repo? [04:36] spiv: yea [04:36] abentley: if bzr.dev spends agreater fraction in the index layer, its more efficient at the other things [04:36] it's just TCL [04:36] abentley: unless the extra index use is disproportionately slow [04:36] abentley: anyhow, my problem for now, and I guess 10% improvement in that region [04:37] if I assume 2.5 seconds in index, I'm down to 2.3 [04:40] The percentage bisect_multi_bytes for my branch is 50.76, I think. Getting some contradictory output. [04:41] Iter_entries time for dev is 57%, for mine, 51%. [04:42] But recall that I'm batching my index queries. [04:42] So that may help on the initial bisect. [04:46] batched queries are a huge win [04:48] I want! http://awaregeek.com/funny-stuff/usb-panic-button/ === asac_ is now known as asac [05:20] lifeless: Would there be a performance win if indexes calculated build-dependencies all in one go? [05:21] abentley: possibly; its been my intent to put some tasteful graph query support into the index layer [05:22] jml, my car is going to be ready in a bit, so i'll walk up and get it [05:22] i got another question: say 2 people want to code ona project together, and they want to use a server, do you have to create a user account for every project your users make together, in addition to their normal user accounts just for the projects - because they both need access to the repository which can only be on a account on the server to which they both have access [05:22] the only possible win though is slightly fewer round trips in extreme cases [05:22] you're welcome to stay here [05:22] or can u do that using groups? [05:23] abentley: we could in fact choose to not resolve references to merges but only to the actual trees. [05:23] needs this refactoring finished and tweaked further I think. [05:24] abentley: what would be a good result - (key, selected_reference, value) in an iterator? [05:25] and [05:25] how can i fix this: Unable to load 'bzr-gtk-0.93.0' in '/home/uwe/.bazaar/plugins' as a plugin because file path isn't a valid module name; try renaming it to 'bzr_gtk_0_93_0'. [05:25] Wow, PQM is taking an hour and a half to run tests for a single merge. [05:26] Possibly. Is selected_reference for accelerating future queries? [05:26] c1|freaky: do what it says: rename the bzr-gtk-0.93.0 directory in plugins so that it doesn't have dots or hyphens. [05:28] i removed it [05:28] i installed it as root [05:28] did that in the wrong direcotry [05:28] thank you [05:29] and where do i install that PQM thingy? on the server? what does it compile? do i have to create testing-config files i mean, how to compile programs etc. or are there default ones which just work? [05:29] that's all so much work for a private thing [05:29] but i want something neat for coders on my server [05:29] c1|freaky: in fact you should name that directory 'gtk' :) [05:30] c1|freaky: I use group permissions for shared repos on my server [05:30] works great [05:30] mtaylor: and where do you put the directories for the shared repos? [05:30] Blasphemously enough, I've got some in /home/cvs/bzr/[...] [05:31] c1|freaky: I made a /var/lib/bzr [05:31] c1|freaky: then every time I add a new repository, I go there and do chgrp -R src repos_name [05:31] and then chmod -R g+s repos_name [05:31] ok, so i can put several projects on there, one repository (/var/lib/bzr) and in there the different projects on which different users could work on. and for every project i would have to create a group on the system [05:31] and then I never think about it again [05:32] c1|freaky: I make a new repos for each unrelated project myself [05:33] c1|freaky: and if you really wanted to restrict access to project to a different set of people, then yeah, each one would want its own group [05:35] any chance that "InvalidRevisionNumber: Invalid revision number 5946" means something other than what it says? [05:36] ok [05:36] now i need to create a repo on the server [05:37] how do you do that? do you first create the repo on the server itself? [05:37] yes, the same as you would with svnadmin [05:37] c1|freaky: you can do it multiple ways, but I usually just go to /var/lib/bzr and do bzr init-repo --no-trees repo_name [05:38] poolie: at this rate, PQM should have merged my fix at about 6pm. [05:38] what does no-trees do btw, i didnt udnerstand it yet [05:38] spiv, +1 my pqm speedup patch then :) [05:39] c1|freaky: doesn't include a checkout-out copy of the source code in the repository [05:40] poolie: I've got about 10% so far [05:40] mtaylor: and when that repo is created, how to you put content on it? [05:40] poolie: I'm calling it a day at this point [05:40] poolie: I should have the refactoring finished tomorrow [05:41] c1|freaky: then you can just push a branch there from somewhere eles. [05:41] i have a branch on my pc [05:41] abentley: its my index.range_map if you want to give it a whirl [05:41] c1|freaky: so if you do cd /var/lib/bzr ; bzr init-repo --no-trees foo on your server [05:41] Hmm, SyPy tonight. [05:41] but the server cant access my pc [05:41] c1|freaky: doesn't need to [05:41] lifeless: Where? [05:41] c1|freaky: go to your pc, and go into your branch [05:42] c1|freaky: and do bzr push bzr+ssh://yoursever/var/lib/bzr/foo/branch_name [05:42] abentley: https://code.launchpad.net/~lifeless/bzr/index.range_map [05:42] abentley: since I fixed register_branch I register them all :) [05:42] abentley: alternatively, because 6 hour mirroring is bog [05:42] *bong* [05:42] you'll want the external url http://people.ubuntu.com/~robertc/baz2.0/index.range_map [05:42] push is always like svn commit right? [05:42] c1|freaky: no, commit is like commit [05:43] but if im not bound to the repo [05:43] i can do commit and it wont commit to the server? [05:43] c1|freaky: right. [05:43]