[00:05] abentley: Thanks for the review. :) I'll resubmit tomorrow. [00:05] Odd_Bloke: np [00:07] Odd_Bloke: in fact, KeyError is at least a plausible exception for name_hook to throw. === jamesh_ is now known as jamesh [02:16] does anyone have a clue how to do log-type operations without a whole heap of whole-history operations yet? [02:31] New bug: #200412 in bzr "Undescriptive error message when user isn't part of LP team." [Medium,Confirmed] https://launchpad.net/bugs/200412 [02:35] hi, i ran the command " bzr init-repo sftp://username@silenceisdefeat.org/~/repo" and bzr was hung up without any output. I use bzr 1.2 standalone win32 version under cygwin bash shell. How can I debug my problem? [02:36] New bug: #200413 in bzr-webserve "Rss feeds should show revisions" [Undecided,New] https://launchpad.net/bugs/200413 [02:39] have a look in ~/.bzr.log [02:39] thx === bpeterson changed the topic of #bzr to: "http://bazaar-vcs.org/ | Bazaar 1.2 is out! | https://launchpad.net/bzr/1.2/1.2" [03:02] This is the content of .bzr.log: [03:02] bzr arguments: [u'init-repo', u'sftp://username@silenceisdefeat.org/~/repo'] [03:02] looking for plugins in C:/Documents and Settings/jerryc/Application Data/bazaar/2.0/plugins [03:02] looking for plugins in C:/Program Files/Bazaar/plugins [03:02] looking for plugins in c:\Program Files\Bazaar\lib\library.zip\bzrlib\plugins [03:02] Looking for plugins in 'c:\\Program Files\\Bazaar\\lib\\library.zip\\bzrlib\\plugins' [03:02] Names in archive: ['__init__.pyc', 'launchpad/__init__.pyc', 'launchpad/account.pyc', 'launchpad/lp_indirect.pyc', 'launchpa [03:05] I can not figure out from this content. should i install python and bzr source to debug? [03:06] red: Certain -D options might help. [03:06] I dunno. [03:07] ok, I will try _D [03:08] you need a -Dxxx option, but I can't seem to find the list of them anymore [03:08] bzr help global-options [03:08] I dunno which one would help though. [03:08] thx [03:09] None of them seem relevant to me. [03:09] red: Do other things over sftp work? [03:10] red: I'd guess it has to do with prompting you for your password. [03:10] But I dunno. [03:11] is this command correct " bzr -Derror init-repo --no-tree sftp://chenbin@silenceisdefeat.org/~/repo"? I tried "sftp://username:passwd@url [03:11] there is no prompt for password if i run "sftp://username@url" [03:12] Yeah, but -Derror isn't helpful here.. [03:12] Does ~ work? [03:12] I've never used it. [03:12] On sftp it does; bzr+ssh doesn't. [03:12] (but don't worry, that'll be fixed a release or two after the smart server is added) [03:13] I just tried it on both sftp and bzr+ssh and it seemed to screw up badly. [03:13] real command is "bzr init-repo sftp://username@silenceisdefeat.org/~/repo" [03:14] Never mind, I'm an idiot. [03:14] I created repo at first on remote server [03:15] I cannot install software on silenceisdefeat.org [03:15] (I was using "sftp//" instead of "sftp://".) [03:15] does 'ssh username@silenceisdefeat.org ls ~/repo' work? [03:18] sh username@silenceisdefeat.org "ls ~/repo" does work. [03:18] ssh username@silenceisdefeat.org "ls ~/repo" does work. [03:20] I've got it [03:21] this seems a bug of bzr [03:22] I don't know why, but on the dos command shell, I succeeded and got the password prompt. in the cygwin, there is no password prompt and the whole program is hung up. [03:23] As I mentioned af first, I install bazaar 1.2 standalone version. But I run the bzr command in cygwin bash shell. [03:33] I reported the bug at https://bugs.launchpad.net/bzr/+bug/200436. Thank you all. [03:33] Launchpad bug 200436 in bzr "bazaar 1.2 (standalone) failed in cygwin shell" [Undecided,New] [04:14] New bug: #200436 in bzr "bazaar 1.2 (standalone) failed in cygwin shell" [Undecided,New] https://launchpad.net/bugs/200436 === doko_ is now known as doko [08:59] hi guys , where I can set what revision that I want to branch? [09:13] Hopefully he figured it out... [09:15] hi... I'm just learning bzr. how would I find out when (which revnos) was a particular file changed? [09:15] muszek: bzr log file? bzr annotate file? [09:15] Peng: thanks [11:21] dato: ping [11:30] jelmer: pong [11:32] dato: I was wondering if you could sponsor a bzr-dbus upload? [11:40] jelmer: yes, I noted it in my TODO when you sent mail to -maint [11:41] dato: Cool, thanks. [11:41] jelmer: I just need to find a bit of time for it :) [11:41] :-) [11:41] We missed at the sprint last week! [11:42] * dato wonders if "you" or something else is missing above [11:42] uhm, yes :-) [11:43] *g* === mrevell is now known as mrevell-lunch === mvo_ is now known as mvo === mrevell-lunch is now known as mrevell [13:46] any idea when bzr-svn for bzr 1.2 will be available? [13:49] uniscript: Hopefully somewhere at the end of this week [13:49] OK great ta === FreeNode is now known as herb [14:01] jelmer: I'm up to date with http://people.samba.org/bzr/jelmer/bzr-svn/stable/, and I'm getting "using experimental bzr-svn mappings; output may change between revisions" . . . there's nothing to worry about as far as normal usage (esp. with respect to data-loss), is there? :-) [14:03] nevans: you should not be using it for production type data but a release instead [14:04] New bug: #200569 in bzr "[PATCH] v1.2 setup.py fails under python 2.3.5" [Medium,Confirmed] https://launchpad.net/bugs/200569 [14:07] jelmer: thanks. I'll switch to one of the release branches. [14:12] jelmer: should I delete my svn-cache or my bzr repo? could running off of the newer code have "corrupted" either of them so that the older release branch of bzr-svn won't know how to cope? [14:16] New bug: #200575 in bzr ""Address already in use " crashes smart server" [Undecided,New] https://launchpad.net/bugs/200575 [14:17] nevans: In theory, yes. In practice however, there haven't been any regressions recently afaik [14:21] jelmer: s/adaption/adoption/ [14:22] LarstiQ: ? [14:23] jelmer: the title of your blog post :) [14:23] doh! [14:39] jelmer: I forgot to ask you last week. What's the bzr-svn release that corresponds to bzr 1.2? [14:39] AfC: There is none yet [14:40] I had meant to release it last week but was distracted by other things [14:40] I'm sitting here at the GTK hackfest listening to these guys next to me absolutely slaughter themselves fighting with git usage. I'd like to just go "and here it is in bzr" but I kinda need to have my shit together first [14:40] jelmer: certainly [14:40] jelmer: no worries [14:41] * AfC wonders if his distro has backported the subversion changes that bzr-svn relies on. [14:41] AfC: what distro are you running? [14:41] I hope so. I had a very poor time trying to get subversion working built from source. [14:41] jelmer: Gentoo [14:41] AfC: There is an overlay (not sure what that means exactly) that has them [14:41] https://edge.launchpad.net/bzr-gentoo-overlay [14:42] jelmer: overlay is the equivalent of "well, I posted a patch". Until it's into Gentoo's official tree it's meaningless. [14:42] ah, ok [14:42] It's like expecting people to depend on a .deb that's not official. It may exist, but it's not usable without a fight [14:43] Don't worry, it'll all be fixed when svn 1.5 is released... I'm sure it'll happen in a week or two, no problem. [14:43] fullermd: oh really? That's encouraging [14:44] [Part of this is that someone did an import of GTK using launchpad, but that's really not any use so long as the upstream project is using Subversion] [14:44] Well, it's been a couple weeks out for, like, 18 months now... [14:44] I suppose I could risk trying that overlay. I don't know who wrote it, what's in it, or whether it is trustworthy. [14:44] fullermd: that too [14:45] Now if launchpad used bzr-svn, THAT would be useful. [14:46] Using other overlays in Gentoo is pretty easy, IIRC. [14:46] More like adding something to your /etc/apt/sources.list. [14:46] I didn't say it wasn't easy. I said it wasn't sound [14:46] Ok. [14:47] [and even more pragmatically, in my experience when someone does a private overlay the problem goes away for them so they no longer are incented to push their work to upstream (in this case, the distro proper) - which means the problem remains unsolved for everyone else] [14:48] [so, for me to go and use that overlay would mean I would no longer be agitating for the problem to be fixed, and the quality level for everyone else remains low. Far better for bzr if the necessary fixes get into Portage officially] === bigdo2 is now known as bigdog [15:10] AfC: LP does SVN imports... [15:11] Odd_Bloke: but you don't mean "using bzr-svn to publish a branch which is actually backed by an external project that is in Subversion" do you? [15:11] AfC: Probably not, but I'm not sure I understand exactly what you mean. :) [15:12] import (as I understand it) is a one way transform for a project that is switching VCS. bzr-svn allows you to use Bazaar against an upstream project that uses (and will continue to use) Subversion [15:12] what I was just suggesting would be interesting & cool would be [15:12] AfC: LP does tracking imports, so that any new upstream revisions are mirrored. [15:12] if Launchpad hosted the bzr-svn infrastructure and sync'd against upstream [15:12] Hm. [15:13] AfC: it has been discussed in the past [15:13] jelmer: I didn't really think it was likely to be an original idea [15:13] I'm probably only sensitive to this because it's taken a while for me to get bzr-svn up and running [15:13] AfC: However, the restrictions that bzr-svn puts on the way conversions are done make it impossible to be used by launchpad [15:14] Well, I'm just as happy to leave them out of it. [15:14] AfC: bzr-svn uses deterministic revision ids so can't change the way mappings are done without changing the revision ids [15:14] jelmer: oh? What restrictions are these? [15:14] jelmer: interesting [15:15] abentley: I meant the referential integrity has to be maintained [15:15] Why is that a problem? [15:17] it makes it impossible to change the mappings halfway through [15:17] without changing revids [15:17] if I add support for svn:keywords in bzr-svn now, I need to change the revision id prefix it's using [15:18] wheras launchpad could just add support for it and add it in the new revisions it imports without users getting funny "Branches diverged" errors [15:20] Otoh, deterministic conversions are useful, and Launchpad might be willing to put up with that upgrade pain for such a gain. [15:21] abentley: True, but afaik that was the reason launchpad decided against bzr-svn a while back [15:22] abentley: btw, did path tokens get discussed later in the week during the sprint ? [15:22] No, they were never discussed. [15:22] Could someone vote 'Resubmit' on http://bundlebuggy.aaronbentley.com/request/%3C20080310005229.11704edf@lapbert.oxbridgetech%3E for me? [15:23] Odd_Bloke: Just submit the fixed version. [15:23] abentley: :-( [15:24] abentley: Sure. :) [15:24] hmm, I think this is the first year I haven't brought up revision id aliases [15:24] :-P === weigon__ is now known as weigon [16:01] Earlier I started with putting an existing project folder under version control (init, add, commit), now I'd like to have that folder as a repository and move the existing files to a subfolder ('main') to allow me to create branches in separate folders next to the 'main' subfolder. How should I proceed? Is it possible to do this afterwards and keeping the current revisions? [16:04] Prodoc: Create a separate shared repo (using 'bzr init-repo repo') and 'bzr branch repo/main'. [16:06] Odd_Bloke: performing 'bzr branch repo/main' before or after I copied the existing project (including the .bzr folder I assume), to the 'main' folder? [16:07] Prodoc: The 'bzr branch' will do the copying, as well as using the shared repository. [16:09] so I just have to perform the 'bzr init-repo repo' on the current version controlled folder? [16:10] Prodoc: It's best to do it outside of the version-controlled folder, to avoid confusion. [16:10] You can move the repository (and it's branches) where you want later. [16:12] ah yes, now I see what you mean, cheers [16:14] ow, and the existing revision history will be retained? [16:14] yes [16:14] hi poolie [16:18] How is bazzar diff from git ? [16:18] How is an orange different to an apple? [16:19] ok I will get to the point. [16:20] git has a lot more citric acid? :p [16:20] I found these distributed scms suffer compatibility between windows and linuxes [16:20] not really distribted. [16:20] Also bear in mind that while we're bzr users in here, many of us won't know git that well [16:20] Or at all [16:21] basically how easy to get my revisions between windows and linux machines. [16:21] so I don't have to worry about setting up ssh or starting a deamon to pull/push to and from. [16:21] afaik, bzr still has better windows support than git [16:21] Bazaar has pretty good Windows compatibility, but not many of the developers concentrate on Windows, and it has case-insensitive issues. [16:21] The repo format is portable, the program is portable... to my knowledge though, bzr won't itself handle linefeed issues [16:21] Peng: indeed [16:22] LeoNerd: we want to do that though [16:23] I don't mind even if would handle ascii files as binary objs. [16:23] The current workaround is using a text editor that doesn't suck. :P [16:23] Oh, that's not necessarily a requirement. People could even use emacs... [16:24] or vim. [16:24] tsk :P [16:24] (g)vim can easily cope with either line format anywhere [16:24] But.. I imagine you're going to be storing source code, etc..? [16:24] the problem is other tools, like, say, patch [16:24] Make sure the compiler/interpreter on each platform can cope [16:24] LarstiQ: patch will happily convert line endings [16:25] luks_: does it nowadays? It sure used to convert everything to native on windows. [16:25] at least the one on windows does [16:26] anyway, for interested people, http://bazaar-vcs.org/SprintLondonMarch08/Brainstorms and http://bazaar-vcs.org/LineEndings would be starting points for implementing it [16:26] I just need a simple tool to efficently transfer versioned binary diffs between windows and linux and not worry about setting up a whole lot of servers doing push and pull it should simply ask during syncing chose the remote and local when conflicts arrive. [16:26] Pengwn: that's as trivial as running "bzr send" [16:27] LarstiQ: the problem with implementing in is that bzrlib reads from the working tree all over the source base :( [16:27] there is no central place for such operations [16:27] when I do bzr send from a window machine how would the other window machine recieve it? [16:28] luks_: reads or writes? [16:28] both, and md5sums [16:28] luks_: the idea is to have a transform layer via tree operations [16:28] I've tried it a few months ago, but I gave up as it would mean refactoring quite a lot of code [16:29] Pengwn: "bzr send" works over email [16:29] luks_: did you document your experiences somewhere? [16:29] so yuou get the patch by mail, and then "bzr pull" it [16:29] LarstiQ: nope [16:29] ok [16:29] just in my private and probably already deleted branch :) [16:29] doh :P [16:30] to put it a little simpler I would like to have versioned files with unison syncing and samba transparency. [16:30] any tool that would remove the trouble of setting up samba would be awesome. [16:30] not sure if that's simpler [16:30] * luks_ doesn't know that means :) [16:31] Pengwn: bzr is a version control tool, not a backup solution and not directory sync tool [16:32] if your only intention is to move files between machines, bzr or any other vcs is probably not going to make you happy [16:32] yep that's what I am looking for then a versioned diff backup solution able to sync from any linux window system. [16:35] unison ? [16:35] I've only used it between Linux boxen, but I know it's designed for Lin/Win or Win/Win too [16:37] problem with unison it goes thru the whole directory structure and takes to much band width this way git handles better with sha1 comparison but pulling and pushing to linux is ok but to windows is a pain in the *** [16:38] Umm... [16:38] unison runs on both machines [16:38] It does local timestamp and checksumming, nowhere near the network [16:39] You have to run it _on_ both machines, though [16:39] Running it on one side to do a local-to-SMB isn't going to gain you anythingh [16:42] how hard is it to setup bazar to do bzr send and get ? [16:43] Pengwn: pretty easy, assuming there is a reasonable way to send mail from the source [16:43] like either sendmail or an smtp server [16:44] can I unison to sync directories then use bzr on top of it ? [16:44] I mean bzr locally on windows and linux boxes? [16:45] sync directories or repos i.e. [16:45] another option would be "bzr serve" if you can temporarily open a port on one of the machines, and they are accessible from each other [16:45] bzr serve is a python server? [16:45] or c server? [16:46] Python [16:46] almost everything in bzr is python [16:46] is bzr server a push pull thing or I can commit and checkout from it like cvs? [16:47] both [16:47] cool. [16:47] but it doesn't handle any authentification [16:47] fine with me. [16:47] atleast plain text passwords? [16:47] No [16:47] nope, nothing [16:48] Use bzr+ssh if you want security [16:48] ssh is not for windows :) [16:48] There are SSH daemons for Win32 AFAIK [16:48] another big problem otherwise would have stuck to git. [16:48] You can certainly use windows as a client to bzr+ssh [16:49] https://code.launchpad.net/~parker-friikz/bzr-auth/trunk [16:49] * awilkins has done it [16:49] can I limit the clients to bzr serve? [16:49] client connection i.e.? [16:49] say to one. [16:49] heh [16:50] maybe I should continue with that :P [16:50] Pengwn: are you planning to use it on a central server? [16:50] Parker-: you shuold! :) [16:50] is it usable yet? [16:50] mostly yes [16:50] :D [16:51] but.. it works [16:51] Pengwn: if it's a central server and it runs any unix based system, you can just use bzr+ssh [16:51] which works fine on windows as a client [16:52] need to leave train.. laters.. [16:53] Ok this is the scenario. I work on windows , linux boxes all over the place I want to commit/checkout on any box and before leaving in the evening to save my code/commits on any spare box available at the client side. I work mostly in a vpn so I can connect to any box but don't want to go thru headace setting up ssh/samba what ever crap. [16:56] Pengwn: if ssh is installed on your target psuh box, your headache is limited to generating keypair for yourself and adding it to the authorized_keys list in your /home/.ssh folder on the target [16:56] And then running pageant / putty on the Win32 bopx [16:56] or just to use password :) [16:56] luks_: Ah, well, I learned to do it before bzr had auth callbacks :-) [16:57] It finally pushed me over to using PK auth for everything, none of my boxen support password anymore. [17:00] Anyway, yes, Pengwn, you just need plink (from putty suite) to use win32 as an ssh client. Small, no-install executable. [17:00] actually, you don't [17:00] Other client in mind? [17:00] the bzr installer comes with paramiko installed [17:00] which provides ssh support without any external dependencies [17:01] * awilkins ... is out of date because of the auth-callbacks thing [17:01] I needed something with an ssh-agent [17:01] paramiko _can_ use putty's ssh agent :) [17:02] awilkins: the headace there is I am working on a windows box all day after ftping my repo now I come to my hotel room connect to slow vpn and save changes to a windows laptop. ftp is again damn slow and remote windows box doesn't have ssh setup. [17:03] Have to go catch train back later [17:04] Pengwn: bzr will download only as much as it needs, which usually isn't that much after one working day [17:04] you mean with bzr serve? [17:04] or bzr-ssh? [17:04] any transport [17:04] you would simple replace the 'ftping my repo' step with 'bzr push' [17:04] and then 'bzr pull' on the laptop [17:05] cool. [17:05] I will try that out then. [17:05] you can use sftp://, ftp://, bzr+ssh://, bzr:// [17:05] thanks luks. [17:06] so bzr is well supported on windows and linux very well? [17:06] I mean my bzr serve should hang I am screwed again. [17:07] bzr works equally on windows and linux [17:07] the only things that are different are caused by missing features of either OS [17:07] the best windows scm server is p4s.exe. perforce's. Never hanged on me maybe once. [17:08] basically I have to say how well is python on windows : ) [17:08] but probably the best you can do is to always push the changes to one server and use that server from each machine you are working on [17:09] that's the problem. servers sometimes in different location country and bandwidth sucks. [17:10] Pengwn: it's not clear to me exactly what problem you are trying to solve. [17:11] admin problems automation problems etc. etc. huge repo like 2GB and need to transfer around on different client places. [17:13] yeah maybe I should stip down the take away the exe/binaries but damn lazy to do that so code and installables all in one repo.:) [17:33] New bug: #200599 in bzr "exception error "Connection reset" caused by push" [Undecided,New] https://launchpad.net/bugs/200599 === FreeNode is now known as herb [18:00] what the difference between init and init-repo? [18:00] Bloguero__Connor, init-repo is a shared repository [18:01] so, init is only for private? [18:01] Bloguero__Connor, I wouldn't call it private, it's jsut for one repository [18:01] the otbher one shares the revisiones [18:02] so it saves space if you're branching the same repo [18:02] Er. init-repo creates a shared repo; init creates a branch. [18:02] wich one saves space? [18:03] and init will create a non-shared repo if you are not in a shared repo when you call it. [18:03] Bloguero__Connor, it depents on what you're going to do ;) [18:04] "bzr help repositories" [18:08] OK. If someone creates a sharedrepo at http://someplace.com/repoX, and I want to colaborate, should I use "bzr branch http://..."? [18:08] Bloguero__Connor, share repos are only relevant locally [18:08] why? [18:08] Bloguero__Connor: A 'shared' repo is only called 'shared' because multiple branches use it to store revisions. [18:09] Those branches share the repository. [18:14] If I have sftp access to the shared repo, what diff make if it is local or in another machine? [18:15] Bloguero__Connor, none [18:16] beuno, you said before that share repos are relevant only locally [18:16] Bloguero__Connor, right, because they save space when creating new branches in the shared repo [18:16] they re-use the same common revisions [18:16] so you don't duplicate them [18:16] but it makes no difference when branching out externally [18:18] Bloguero__Connor: I think you may be being confused by the use of the word 'repository'. In bzr, a repository is _only_ a place to store revisions. Branches are what make sense of these revisions (and are what you can share in the way you seem to mean). [18:18] so share repo is called shared because it shares files between branches, not because it shares branches between users. [18:19] Bloguero__Connor: Yes. :) [18:20] guys back. [18:21] I did bzr serve --directory=c:\myhome [18:21] how do i access or sync with this from another windows machine? [18:23] Pengwn, bzr://host/branch [18:23] I am getting errors saying bzr: ERROR: not a branch "C:/myhome/" [18:24] Pengwn: Is C:/myhome/ a branch? [18:24] using bzr pull bzr://141.144.64.18:4155/myhome [18:24] nope directory. [18:25] Pengwn: That'll be your problem then. You can only pull from branches. [18:26] is branch nick and branch one and the same? [18:27] Pengwn: no [18:27] ok how to create a branch and serve it? [18:27] there's no default branch? [18:28] Pengwn: "cd branch; bzr serve" [18:28] or "bzr serve --directory=c:\location\of\branch" [18:29] Pengwn: bzr init & bzr add (or be more specific here) & bzr commit [18:29] cd c:\myhome; bzr init ; bzr add; bzr commit ; bzr serve ?? [18:30] then on my other machine bzr pull bzr://141.144.64.18:4155/myhome ?? [18:30] Pengwn, that's right, if you just want to have one branch, yes [18:30] you probably don't want to do add your whole home directory, though [18:31] luks_, it's a windows home, it's probably empty :p [18:32] james_w, does bazaar still use absolute paths with the smart server, or deso it do relative now? [18:32] beuno: in what sense? [18:32] beuno: I don't know I'm afraid. [18:32] if I do bzr serve --directory ~/src/bzr ; then bzr branch bzr://larstiq/bzr.dev works [18:32] LarstiQ, that's it [18:33] thanks :D [18:33] beuno: how is Prague? [18:33] james_w, absolutely great. Having a chance to get back some brain-functions working :p [18:33] :-) [18:34] james_w, tomorrow is your last day at previous work? [18:34] james_w: hey, meant to ask the other day: are you coming to debconf? [18:34] Does anyone know who admins planet.bazaar-vcs.org? [18:34] LarstiQ: same for you [18:34] beuno: yes it is \o/ [18:34] dato: I wasn't planning on it, but it sounds like there may be a chance that I am now. [18:35] james_w: ok :) [18:35] james_w, so from when on can I start complaining about Ubuntu doing things wrong to you? [18:35] beuno: um, you'd be better of complaining to someone else? :) [18:36] oh damn do i have to bzr init on both machines? [18:36] Pengwn, no, just do branch instead of pull [18:36] (once, to create the initial branch) [18:36] dato: I think it would be great if at some point you could send a message to the list outlining the features that made git so compelling for you. [18:37] dato: I realise you have mentioned a couple already, which is better than most people who just say that there are things without really pointing them out, so thanks for that. [18:38] ok, I could think about them [18:38] * dato takes a note [18:38] dato: thanks. [18:42] dato: I had been filling in fields in penta slowly, I need to complete that. [18:46] Wow bzr is cool on windows. [18:46] hope it's equivalently cool on linux :) [18:47] by next best tool :) [18:47] Pengwn, like everything else, even cooler :p [18:48] beuno: now now :P [18:48] but seems I have to go the physical c:\myhome directory and do bzr pull bzr://hostname:4155 [18:48] * beuno goes back to his corner and sits back down again [18:49] the bzr://hostname:4155/myhome doesn't work. [18:49] says c/myhome/ is not a branch. === ja1 is now known as jam [18:50] jam: o/ [18:50] Pengwn: what does `bzr info` say when in c:\\myhome ? [18:51] Odd_Bloke: hi [18:51] jam: Thanks for the reviews earlier. :) [18:51] np, trying to get the merge queue down to something reasonable [18:52] Odd_Bloke: did you have a good train journey back home? [18:52] jam: Yeah, managed to get a seat and everything. :p How was your flight? [18:52] branch root: . [18:52] Related branches: [18:52] parent branch: bzr://141.144.65.18:4155/ [18:53] hi jam. [18:53] The flight was rather uneventful, they managed to arrive 30-min early, but then take >1hr to get my luggage unloaded [18:53] hi james_w [18:53] Perhaps the luggage arrived on time. :p [18:54] Pengwn: and you did "cd c:; bzr serve" first? [18:54] (if you did cd C:\myhome, then the appropriate branch is "bzr branch bzr://hostname/" [18:54] no cd c:\myhome; bzr serve [18:54] Since paths are relative to the cwd [18:55] otherwise it would be branching "c:\myhome\myhome" which obviously doesn't exist [18:57] hey thanks [18:57] bzr://hostname:4155/myhome works [18:57] LarstiQ: ok :) [18:57] after cd c:\ ; bzr serve [18:58] how the hell it bzr know c:\myhome is the bzr root without being in it? [18:58] Pengwn: Because that's where you ran 'bzr serve' from? [18:58] Peng: it maps URLs to file system paths [18:58] erg. pengwn [18:59] ahh got now cd c:\myhome; bzr://hostnaem:4155/ fails :) [19:02] The logic is then If bzr serve find the BZR ROOT client cannot use the path info. If bzr serve doesn't find the BZR ROOT the client has to tell it where to find it. :) [19:04] Pengwn: bzr serve doesn't care where it is run from, it will take an incoming request and then look in $serveroot + $relpath [19:04] Pengwn: so yes [19:05] hello jam [19:05] Pengwn: you could also (on a unix system) do `bzr serve --directory=/` and that way expose your entire filesystem, but you probably don't want that [19:05] hi poolie [19:05] How was your travel back? [19:06] LarstiQ: what is the $serverroot on windows it just says . [19:06] not too bad [19:06] you? [19:06] Pengwn, wherever you specified bzr serve to run from [19:06] Pengwn: it is either the directory you run bzr serve from, or what you specified with --directory [19:07] I gues it cannot transalte c: and d: etc then ? [19:07] Pengwn: so 'cd c:/myhome/; bzr serve', $serverroot would be 'c:/myhome'. [19:07] Pengwn: what do you mean? [19:08] Pengwn, you might be able to use shortcuts to link to d:, but windows is a bit uncertain for me [19:08] LarstiQ, I think he means server contents from c:/foo and d:/bar [19:08] beuno: hmm [19:08] you know, there is no / in win [19:10] beuno: there is if your machine meets the Windows Vista Ultimate POSIX Professional Edition minimum hardware specs. [19:10] mneptok, you mean kubuntu with the vista skin? :p [19:11] welcome to emacs [19:11] once the bzr serve finds the BZR root it always give $server + $relpath ans not a branch. [19:12] it recongnizer "c:/myhome" and not "c:\myhome" [19:13] so once it finds the root and i say bzr pull bzr://hostname:4155/proj1 it doesn't find it. [19:13] i have to cd c:\myhome\proj1 and then do bzr pull bzr://hostname:4155 [19:13] Pengwn: this sounds wrong [19:14] Pengwn: you have a bzr server running from c:/myhome, right? [19:14] windows is always wrong for me :) [19:14] yep. [19:14] Pengwn: could you then try mkdir c:\temp; cd c:\temp; bzr branch bzr://hostname/proj1 [19:14] bzr serve --directory="c:/myhome" [19:15] I am in C: [19:15] Pengwn: could you try my way first? I'd like to rule out funky window pathnames first [19:16] C:\myhome>bzr pull bzr://141.144.65.18:4155/proj1/ [19:16] bzr: ERROR: Not a branch: "bzr://141.144.65.18:4155/proj1/". [19:17] statik: ping [19:17] Pengwn: that's not really what I asked you :P [19:17] * LarstiQ is tempted to boot into windows and check [19:20] I just have to be out of the location where there is no .bzr directory and then the bzr://hostname:4155/path works. [19:25] some how my PWD is overiding the branch infomartion to the bzr serve. [19:26] ok got to go now. [19:28] You shouldn't have to specify the port number [19:29] poolie: overall uneventful, which is how I prefer travelling :) [19:30] hello afc [19:30] Good evening Martin [19:31] hey poolie [19:47] poolie: hi. How are you? [19:48] poolie: who is it that admins the bzr planet? [19:48] good, thankyou [19:48] though in an unusual personal timezone [19:48] canonical IS [19:48] I can pass on requests or comments [19:48] poolie: I thought it was a little early for you. [19:49] I would like to be added if you approve, does that need to go through them? [19:51] that's probably easiest if you don't find them here [19:51] lamont (in us/mountain time) can do it for example [19:52] can you pm me the urls [19:54] or mail [19:55] abentley: ping [19:55] * lamont notes that #canonical-sysadmins is where one can generally expect to find zero or more awake sysadmins [19:55] (this server) [19:55] jelmer: pong [19:55] abentley: Do you know how to get kid to not insert a HTML header above pages? [19:56] * jelmer has a bundlebuggy patch which adds rss feeds and this is the last bit he can't get to work [19:56] lamont, james_w: sent [19:56] The header will be coming from your master template. [19:56] poolie: thanks. [19:56] abentley: I mean the line on top, which is inserted even while it is not there in my kid template [19:57] jelmer: one sec [19:57] did i show this screenie off in here yet? http://people.ubuntu.com/~mwh/hacked_up_changelog_view_3.png [19:57] mwhudson: released yet? :) [19:58] poolie: well, i released 1.2 and 1.2.1 yes [19:58] poolie: that changelog view is not even committed to a branch though i think, it's a total hack [19:58] oh, interesting [19:58] jelmer: expose it like this: "@expose(template='livelistings.templates.atom_events', format='xml', content_type='application/xml')" [19:59] mwhudson: fair enough :) [19:59] abentley: Thanks! [19:59] mwhudson: i don't think you need different font sizes btw [19:59] The content_type would probably be different. [19:59] they're already distinguished by placement and weight [20:03] abentley: yep, application/rss+xml probably [20:03] it works \o/ [20:06] grr === mwhudson_ is now known as mwhudson [20:07] did i miss anything? [20:08] Thunder. Lightning. Dancing girls. 3 pirouetting elephants. [20:08] cool [20:09] i guess i should set up a proxy one of these days [20:19] anyone available to make a small change to the bazaar user-guide so I don't have to get the source first and create a patch and stuff? [20:20] Prodoc: sure [20:21] with sufficient detailed suggestions I can actually do that :) [20:22] LarstiQ: the following sentence in http://doc.bazaar-vcs.org/bzr.dev/en/user-guide/index.html#id41 doesn't really flow: 'On the other hand, Mary might want to merge into her branch the work Bill has done in his.'. The following makes more sense imho: 'On the other hand, Mary might want to merge the work Bill has done in his branch into her branch.' [20:23] hmm [20:23] nitpicky I know ;-) [20:23] I had to read it over 3x [20:24] I actually prefer the first sentence. [20:24] but I'm not a native speaker, and I know my mind sometimes works rather strangely [20:24] or leave out 'in his branch' in the suggested sentence [20:24] http://bundlebuggy.vernstok.nl/bzr-gtk/rss?selected=pending&unreviewed=n [20:26] that would make: 'On the other hand, Mary might want to merge the work Bill has done into her branch.', a lot clearer imo [20:29] jelmer: Cool. [20:30] jelmer: But I won't be able to put it on the main BB instance, because that query is really hard on the DB. [20:30] abentley: it should be the same query as for the index page [20:31] jelmer: Yes. The index page is really hard on the DB. [20:31] ouch :-( [20:31] Yeah. [20:31] Mainly because that box is underpowered, I think. [20:32] It runs at a decent speed on my 3-year-old laptop. [20:32] The first entry produces a parse error for me: XML Parsing Error: not well-formed [20:32] Location: file:/// [20:32] Line Number 60, Column 7: [20:32] what box is it running on? [20:33] It runs on aaronbentley.com, which is a virtual machine hosted by unixshell [21:00] abentley: I believe poolie discussed getting it hosted by Canonical servers, has anything come of that? [21:01] I know one concern is just getting enough admin resources for a Canonical machine [21:01] They like to lock the machines down pretty tight [21:01] jam: more recently, we've been talking about using a different host. [21:01] poolie was just talking to me about it. [21:14] hey guys [21:14] in a post-push hook [21:14] is there any way to get ahold of the revnos that the revisions will be in the remote tree? [21:15] jam, i think it's just you and me to talk today, as spiv is in canada and igc is on leave [21:16] is Rob doing ok? [21:16] in what way? [21:16] he seemed ok yesterday [21:16] poolie: he wasn't too well the last days I saw him [21:17] oh right [21:17] i think he was mostly over it by the time we left [21:17] though i am feeling a bit sick myself today [21:18] poolie: ai, beterschap [21:36] moin [21:36] hi lifeless. [21:36] are you feeling better now? [21:36] comes and goes a bit [21:37] hello [21:37] may head up to doctor later today and get a considered opinion [21:37] hi Stavros [21:37] i would like to create a setup where the live server automatically updates its working tree with any revision tagged "release", is this possible? [21:38] yes; it will need a cronscript at this point, or a client side hook [21:39] aha, well, since it has to pull from another repo, i'm thinking cronscript [21:39] but how can i read the tag? [21:41] Stavros: does "bzr tags" help? [21:42] not really, i need to do it programmatically [21:42] i'd probably need a cronjob to pull at regular intervals and a hook to update the tree on release/restart apache? [21:42] Stavros: are you doing it in shell, or in python? [21:43] Stavros: because you could just do "bzr pull -r tag:release" from shell [21:43] From python it would be more: [21:43] jam: probably in python, because i need tags preserved (i'm not entirely sure how tags work), so the tags would be something like "Release 1.2.2" [21:44] tag_revision = branch.tags.lookup_tag('release') [21:44] Ah, you want Release XXX not just "Release" [21:44] yes, something like that [21:44] it's the same thing in python, that's why i didn't mention it [21:44] So in that case, you could do: [21:44] b = bzrlib.branch.Branch.open('remote_location') [21:45] actually i should read up on bzr scripting because i have no idea how to do this [21:45] tags = b.tags.get_tag_dict() [21:45] for name in tags: [21:45] if name.lower().startswith('release'): [21:45] .. [21:45] it sort of depends how you want to decide to use 1.2.2 over 1.2.1 [21:45] etc [21:46] i shouldn't have any problems with the actual script, i just want to know where to put it/how you go about writing hooks/etc [21:46] is there a webpage that explains that stuff? [21:46] is it this? http://bazaar-vcs.org/WritingPlugins [21:47] Stavros: that has some of it. I don't know if it will have all that you need though. Feel free to ask questions in here if not. [21:48] will do, thanks [21:49] 7w 21 [21:49] Gah. [21:59] wow, you can version your bzr plugins with bzr [21:59] i think my head just exploded [22:00] are bzr tags just text that is associated with a release number? [22:02] Stavros: tags are just a name associated with a revision_id [22:02] err yes, why do i keep calling revisions releases, i'll never know [22:02] that is quite elegant [22:02] And yes, all of my tags are bzr branches in ~/.bazaar/plugins [22:02] and tags are unique? [22:02] jam: what was that? [22:02] tags are not universally unique [22:02] hmm [22:02] sorry, all of my plugins are branches .... [22:02] not tags :) [22:02] ah [22:03] tags aren't unique? how does that work then? [22:03] they are unique to a branch, and a merge may cause a tag conflict that you have to resolve. [22:04] oh [22:04] well, a branch is as universal as it gets, isn't it [22:04] so if I tag one revision as "release 1.1" and you tag a different one as "release 1.1" then there needs to be intervention when we merge to say who was right. [22:04] true [22:24] can someone kindly tell me what's the difference between --no-trees and the default in init-repo command? won't they both start out with emptry directory? [22:25] huslu: they will. The difference is what happens when you bzr init/branch branches into it [22:27] LarstiQ: thanks but what is the difference when you branch (or push?) into it (with when init-repo was started with --no-trees) [22:28] huslu: with --no-trees you'll just get a branch but no working tree. Ie, branchname/.bzr/ but no other files under branchname/ [22:28] huslu: which goes nicely with having a lightweight checkout of that branch somewhere else on your filesystem where you do actual development [22:32] LarstiQ: k, so if i understand correctly, all the --no-trees does is, keeps the revision history in a different place? [22:32] lifeless: can i call? [22:33] huslu: no, the revision history is kept in the same place. The only difference is to wether create a working tree at the same place as the branch or not. [22:34] huslu: but you can see for yourself, bzr init-repo repo; bzr branch lp:bzr-gtk repo/gtk; bzr init-repo --no-trees repo-no-trees; bzr branch repo/gtk repo-no-trees/gtk [22:34] huslu: and then ls repo/gtk repo-no-trees/gtk [22:34] wait, to create a hook all i need to do is write a function and register it? [22:35] LarstiQ: right, so the difference is that the files are not in the repo-no-trees directory? [22:37] huslu: that sentence is a bit amibguous. All the files needed to reconstruct a working tree are in both repositories [22:37] huslu: you can get the same effect on repo-no-trees by doing: cd repo-no-trees/gtk; bzr checkout . [22:37] huslu: but the default will still be to create tree-less branches then [22:38] huslu: the other way around, cd repo/gtk; bzr remove-tree [22:38] does a "pull" update the working tree if it had no local changes? [22:38] Stavros: yes, also if it did. [22:38] Stavros: that is, non-committed local changes [22:38] LarstiQ: it just overwrites the changes? [22:38] that can't be right [22:39] Stavros: if you have diverging commits, it will tell you it has diverged [22:39] ah [22:39] Stavros: no, it merges the content [22:39] LarstiQ: ok, that helps, now, can you help me to see the light when or why is it beneficial to not to store the working tree there? [22:39] how do i make it so it doesn't update the wc? [22:39] huslu: each working tree consumes diskspace, you don't always need them [22:40] Stavros: I'd have to check to be sure, but I'd think invoking the method on Branch instead of WorkingTree [22:41] huslu: or for example if you use a workflow where you `bzr switch` between branches but have only 1 working tree [22:41] hmm [22:42] LarstiQ: ok, i haven't gotten around to that part yet. but your explanation made things more clear, i'll keep learning. [22:42] hmm, well, this doesn't help my plan for the self-updating servers... [22:42] Stavros: alternatively, you could decouple the branch and the checkout, update the branch, and then update the working tree to the latest release tag [22:43] huslu: np [22:43] LarstiQ: you mean push to the branch> [22:43] ? [22:43] Stavros: no, pull [22:44] how would i do it so the wc wasn't updated, though? [22:44] Stavros: I'd go with a branch without wc, and the wc in a different location [22:44] ah [22:44] not very elegant, though [22:45] or more elegant, depending on how you look at it :) [22:45] well, the more elegant solution would be one that didn't require me to keep the repo and the files in separate locations and copy between them :p [22:45] Stavros: or if you could go without local changes, pull, and then revert to the latest release tag [22:45] Stavros: copy? Eh, no [22:46] Stavros: branch in one location, lightweight checkout in another [22:46] oh [22:46] does anyone know if python-glade2 2.8.6 will be enough for bzr-gtk [22:46] hmm, what was that other solution? pull and revert? [22:46] re bug 200366 [22:46] actually that should work [22:46] Urg. Not very well. [22:47] Stavros: the drawback of that is that is going to do more writes to disk than needed [22:47] fullermd: to poolie's question or Stavros comment? [22:47] The latter. [22:47] LarstiQ: it doesn't matter much, it's not that many files [22:47] fullermd: what would you recommend? [22:47] hmm, any way to use bzr (for commits, reverts, etc) while using 'bzr gdiff'? [22:48] mamato: does it lock? [22:48] I haven't been much paying attention. But you'll work yourself into an unpleasant corner using rever tlike that. [22:48] yep! not sure why... [22:48] fullermd: the server doesn't have any local changes, and this is the best solution so far... [22:48] the absolute best would be if there was a "bzr pull --no-update" option [22:48] so it didn't update the working copy [22:48] fullermd: the use case is only updating the working tree to revisions tagged with a release [22:49] and then i manually reverted to the latest revision whenever i wanted [22:49] How do you define a 'release'? [22:49] Stavros: it's not like gdiff allows you to do much :S [22:49] Stavros: well, in my mind that is exactly served by having branch and tree not co-located [22:49] LarstiQ: hmm, i should read up on a lightweight checkout [22:49] fullermd: a release is the well-tested code [22:50] that should run on production [22:50] all other commits should run on dev [22:50] I know, but what in bzr terms makes the release? A sliding tag? Somebody updating a revno in a control file? [22:50] fullermd: oh, a tag [22:50] not sliding, just tag "release x.x.x" [22:50] or even sliding, same thing [22:50] The problem is that `revert -r` is not `update -r`; you'll wind up in Conflict City every time you pull. [22:50] oh [22:50] that won't do, then [22:51] I'd just make a sliding LATEST_RELEASE tag, and pull -rtag:LATEST_RELEASE. [22:51] why would it conflict, though? there are no local changes [22:51] There are, though. revert makes local changes. [22:51] ah [22:51] (you may need pull --overwrite because of the rough corners of tags) [22:51] fullermd: something I wanted to ask early, does it do the lookup of tag:LATEST_RELEASE in the parent branch? [22:51] i would like to be able to track releases, so i'm not very partial to the sliding tag idea [22:51] LarstiQ: Yah. pull -rwhatever refers to revs in the branch pulled from. [22:52] Well, the idea is that you update the sliding tag to point at a release when you make it and are sure it's ready; that way it all Just Happens. [22:52] fullermd: oh, doh [22:52] If you don't slide a tag, you have to go update the tag in the pull'ing script every release, since it needs to go somewhere different. [22:52] yes, but i've had times when i needed to revert to the previous release [22:53] fullermd: i was thinking of a plugin that parsed it and selected the latest revision tagged with "release" [22:53] fullermd: `bzr tags --sort=time | tail -1 ` ;) [22:53] Launchpad bug 200366 in bzr "Ubuntu package bzr-gtk for Dapper depends on package not available for Dapper." [Medium,Confirmed] https://launchpad.net/bugs/200366 [22:53] something like dato's [22:53] Then you'd just slide the tag back, and the pull'ing copy will follow next time it runs (with --overwrite in that case, definitely) [22:53] fullermd: yes, but the problem is remembering where the tag was [22:54] Who needs to remember? It was the same as rel_3_4, then rel_3_5, then rel_3_6, then OOPS 3.6 was horked up, move it back to rel_3_5 [22:54] fullermd: that's not sliding, though [22:54] or can you have multiple tags per release? [22:54] Sure it is. You have defined tags for each release, *AND* a DEPLY_THIS_CODE tag pointing at whatever you want the server running at the moment. [22:54] err, revision [22:54] ah, THAT i like [22:54] Sure. You could have a thousand tags pointing at one rev. [22:54] great gret [22:55] And you should, actually. I don't think anybody's done it, so we need a guinea pig ;> [22:55] so i tag a rev in my local repo and push it to the main one [22:55] haha, damn you [22:55] then i do bzr pull -rtag:PRODUCTION ? [22:55] * fullermd nods. [22:55] aha, let me try that [22:55] Probably need pull --overwrite; otherwise it won't step backward if you have to move the tag back. [22:55] aha, okay [22:56] (and it may not pull the tag quite right either, due to above) [22:56] does "bzr tag tag1 tag2" tag with two tags, or tag with one tag with a space in it? [22:57] I think it may give an error... [22:57] aha [22:57] also, are tags fully reversible? i can delete them without any trace, no? [22:57] I'd do something like `bzr tag -r-1 RELEASE_3_5 ; bzr tag -rtag:RELEASE_3_5 PRODUCTION` [22:58] why -r-1? [22:58] oh [22:58] i see, yes [22:58] Well, or -rwhatever; find the rev to tag the release, then use the release tag to set the PRODUCTION tag. [22:58] yes [22:58] I _think_ pull --overwrite doesn't delete tags gone upstream. It should move moved ones, though. [22:59] is it possible to have bzr commit display the diffs in the bottom part of the file it has you edit for comment, where the list of files is? [22:59] Tags got to that awkward state where they're plenty good enough for 90% of uses, and the other 10% isn't near the top of anybody's priority list :| [22:59] mamato: bzr commit --show-diff [22:59] hehe, nice! [23:00] offtopic, does anyone use fish? [23:01] james_w argh... not available... i guess my bzr is too old :( [23:01] Mmm... NEWS says it was in 0.91. That's a fair couple steps back now... [23:02] i use what ubuntu gave me (and it's up to date)... it's 0.90.0 :( [23:02] it seems to be working [23:02] mamato: yeah, ubuntu is a bit bacjk [23:03] use the repository on the site [23:03] fullermd: is there any way i could push to that server and have it detect if the PRODUCTION tag changed revisions and restart servers/etc? [23:03] i mean, do all that on push instead of pull [23:04] With push? I dunno... I s'pose you could parse the output at least, looking for an 'update' (or whatever it says) keyword. Not sure if the exit code says anything. [23:04] Stavros: the normal way to do this would be using branches I think. However that may not help you. [23:05] hmm [23:05] so the easiest way would be to write a crontab to pull every x minutes? [23:05] Yeah, that's how I'd do it. [23:05] (well, it's not of course, but it's how I'd do it given you general setup ;) [23:05] how would you do it? [23:05] with another setup? [23:06] Well, I do all my deployals via install(1) :p [23:07] pff :p [23:07] rrrrrhhh... newer bzr requires newer libc and python-central... [23:09] oops... nevermind [23:18] any way to have --show-diff be the default? [23:19] aliases won't even work transparently since option has to come after command :( [23:20] defaults? [23:21] Err, I dunno. That's an hg thing. :P [23:22] heathen! [23:22] hmm... should have looked more carefully at hg before choosing bzr i guess ;) [23:22] sacrilege [23:23] actually i think they're more or less the same [23:23] bzr has better merging [23:23] Peng: what is a hg thing? [23:24] Huh? Why don't aliases work? [23:29] well you cant alias it as 'bzr commit' [23:30] I'm pretty sure you can. [23:30] Sure you can. [23:30] Wait. bzr aliases, not shell aliases. [23:30] mamato: are you referring to shell aliases? did you realise that bzr itself supports aliases? [23:31] was talking about shell aliases [23:32] mamato: http://bazaar-vcs.org/CommandAliases [23:32] Oh. [23:32] Bazaar aliases can do this fine. I just tested it. [23:32] yeah, I have that alias set up. [23:33] nice! so bzr does allow default options! [23:34] And it automatically supports inverse options, so that you don't have to worry about doing unalias to get around any issues. [23:35] for instance if you alias "commit = commit --strict" this won't let you commit with unknown files in your working tree. However if you want to do it once for some reason you can run "bzr commit --no-strict" which expands to "bzr commit --strict --no-strict" and the latter takes precedence. [23:39] while i'm at it, is it possible to add colors or bolds to diff? (w/o launching gtk) [23:43] mamato: The 'cdiff' command in bzrtools adds colour. [23:45] thx! [23:49] just in case, you got a trick to colorize the --show-diff of commit in vim too?! [23:52] mamato: a vim plugin to do that may be quite straightforward, but I haven't heard of one yet. [23:53] hehe oki [23:53] I think I saw one being posted on Google code [23:53] a generic one for vcs'es [23:54] one last thing then... a replacement for less which would work with colorized output? [23:54] jelmer: vcscommand? [23:54] james_w: yeah, probably [23:55] somebody posted a patch that added bzr support for it [23:55] jelmer: yeah, I think that provides :vcs commit or similar, but it may have what mamato wants. [23:56] yeah, it allows you to run vimdiff between the base of the working tree and the working tree [23:56] is there a switch to commit the wc just like it is, even if it's a few revisions back or whatever? [23:56] mamato: bzr cdiff | less -R do what you want? [23:56] héhé, found my 'less -R' for now [23:56] yep, thx!