[00:18] Hmm. bzr branch -v lp:gcc-linaro/4.5 as part of creating a mirror branch is sitting there at 1kB/s after about 300 kB. I see the same on a machine with a good network connection like chinstrap [00:19] Ah! It hit ~380 kB, finished the 'fetching revisions' stage, and is now at ~1MB/s === oubiwann is now known as oubiwann-away [03:09] Hi there. I could use some help with performance. [03:09] I'm working on the gcc-linaro branch, which is huge, and pull and/or checkout operations take a long time [03:10] A pull on one machine has been in the 0/3 merge phase for the last 15 minutes [03:10] 2.2b4 is much better, but still too slow with how I'm using it [03:48] spiv, can you help michael? [04:05] ref http://www.mail-archive.com/bazaar-commits@lists.canonical.com/msg07202.html <== I'm sure there's a page on the bzr wiki/docco somewhere that describes this - using bzr/tc for slow networks etc, but am having finders failure; anyone have a link handy? [04:11] hi spm [04:11] heya poolie [04:11] well the specific thing that patch is updating is http://doc.bazaar.canonical.com/bzr.dev/developers/testing.html#simulating-slow-networks [04:12] there may be other pages that mention it [04:12] ahh perfect. that looks like the one I was looking for. ta! [04:12] i'd like to believe there's a good reason the syntax is so obtuse :) [04:13] "because" [06:53] This will sound really stupid, but I want to get the grub 2 graphical manager from a bazaar branch, and I can't figure out how [06:59] maybe "bzr branch lp:grub" [07:00] or maybe the "graphical manager" is something else [07:01] Oh, btw, someone at Canonical should really redraw the icon of the "VCS Imports" user, it does not make sense anymore since we switched away from GNU Arch. === beaumonta is now known as abeaumont_ [09:33] Hmm, I have some productively failing tests. I guess that's a good time to clock off for the day. === oubiwann-away is now known as oubiwann [14:47] Having trouble using mv --after here... can anyone help? http://pastebin.com/wvjdKR2R [14:48] VSpike: That file isn't versioned, you can't use 'bzr mv' to move it [14:50] jelmer: if you look at the "bzr st" output, it lists deployment/connections.geonixlocate.config as a removed file [14:51] VSpike: if it's removed, you have to revert it again first before moving it I think [14:52] Hm I'm sure I've never had this problem before. In fact, I succesfully did it in this branch with other files that I'd moved and then wanted to tell bzr about [14:55] I'm not sure in that case. [15:27] jelmer: this doesn't look good, does it? http://pastebin.com/GS4ztGni [15:27] I moved the files back to where they were before and that is the result [15:29] VSpike: You need to re-add ("bzr add") them [15:29] VSpike, Right now some of those files are present but not versioned, hence them showing up in "removed" and "unknown" [15:30] If I re-add them, then the show in removed and added [15:31] But surely I will still lose the history of the file that way? [15:32] VSpike: "bzr revert" should be able to help [15:32] Normally if you remove a file and bzr st shows it removed, if you put it back it just vanishes from the bzr st output [15:34] Yeah, I was thinking that. Because they were moved and changed, I'll have to rename the new versions, revert the removed ones, copy the changes in, then bzr mv them [15:34] Well, worth a try anyway [15:42] seems to have worked :) === mnepton is now known as mneptok [16:18] hi jelmer [16:18] (4am is usually a bit early for me to respond :) === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha [16:57] jam: 'morning! [16:59] Hi everyone [16:59] hi sinasalek [16:59] I have a question about branches located inside repositories [16:59] When i push a branch into a repository it act like a working copy [17:00] Is it possible not to have files checked out on branches inside repository. [17:00] ? [17:01] Is my question clear?! [17:01] If you e.g. ssh on to the repository server and "bzr checkout" in that branch directory, a checkout will be created yes [17:01] But... "bzr push" from elsewhere (or the implied push that happens during a bound commit) will not keep that checkout updated [17:02] So it will become stale, unless you find some other mechanism to "bzr up" it each time === oubiwann is now known as oubiwann-away [17:05] Tx but that wasn't what i meant. lets ask it in another way. How can clone a branch but only clone its working tree and not the checkout files inside it? === beuno is now known as beuno-lunch [17:10] And another way! [17:10] Reproducing [17:10] 1. Create a shared repository [17:10] 2. Create a trunk branch inside it [17:10] 3. Push a checked out branch else where to trunk [17:10] What i expected? [17:10] I expected push to only copy the working tree of the "checked out branch" and not the checked out files inside it. [17:10] Why do i need it? [17:10] To preserve disk space, checkedout files are not compressed === oubiwann-away is now known as oubiwann [17:12] Any idea? [17:17] In subversion repository is only repository you can't use it directly for development. If you want to you have to checkout a branch on another location. In bazaar however it's different. [17:18] How can i make bazaar repository to act similar to subversion (to only contain working trees and not checked out files!) [17:20] chx: I'm looking for a command or a switch to ignore the checked out files [17:24] anyone up for looking at problem i'm having with bzr merge? explanation at http://dpaste.com/222815/ [17:25] in summary: diff of two files looks fine, but the conflict shown by bzr merge does not (it's missing a line). i guess i must be doing something wrong... but don't know what. [17:29] http://pastebin.com/KFWfDGVW [17:38] Does any one have time to have a look at my repo structure and lighten with his vauable advices? http://pastebin.com/VfK7KCbY [18:11] sinasalek: I'm not sure I understand your question, you are looking to make a repo that is just for servers? one without a working tree? You just want the revision history so people can checkout from it correct? [18:12] sinasalek: if so, than bzr init-repo --no-trees will create a repo with no working files, just the revision history. === beuno-lunch is now known as beuno [18:27] rubbs: yes exactly! thanks [18:28] sinasalek: np [18:29] sinasalek: also, there is no magic repo structure, anytime you branch ( or clone) from the repo, you create your own repo. It's kind of hard to understand at first, but once you understand it, it becomes that much more clear [18:31] rubbs: Yes it seems that i still have problem understanding how bazaar works exactly :) i thought that working tree was the same thing as revision history in bazaar!! [18:31] sinasalek: it's hard, took me a while. Just keep asking questions, someone here will help you out [18:32] basically think of each branch as it's own repo. You get your own personal repo each time you branch [18:32] you actually merge repos when you merge branches... or at least that's how I conceptualize it [18:35] rubbs: i see, with people like you around i'm not worried at all ;) [18:35] That... sounds like it would just be more confusing in the long run... [18:36] fullermd: well, I should clarify, that's how i used to conceptualize it. it's enough to get started at least [18:36] Actually, it makes me crosseyed on the short run too :p [18:37] fair enough. [18:37] DVCS is hard to nail down one good way to conceptualize IMHO [18:37] sinasalek: I'd like to know if http://wiki.bazaar.canonical.com/MatthewFuller/SpotDocs/PiecesInBrief is any help to you. [18:39] rubbs: Aha! it's good! i read dozens of documents but i didn't see this one! [18:39] oh, thank fullermd not me ;) [18:39] :) [18:40] I didn't know about that one either, fullermd. I'll use that for further questions, thanks. [18:40] fullermd: Aha! it's good! i read dozens of documents but i didn't see this one! DVCS is a hole great new world [18:40] fullermd: why don't yuo move it to official doc? it's very helpful for newbies [18:41] The SpotDocs are an ongoing (in the sense of "not done", sadly not in the sense of "I get a lot of time to keep working" :( ) project of mine to try and give complete answers to specific questions. [18:41] The PiecesIn* bits are groundwork below that to nail down vocabulary and base concepts. [18:42] fullermd: This is great work you're doing. Are you on Bazaar doc team? [18:43] Oh, no. I've just used bzr for a while. [18:44] rubbs: i just tried bzr --remove-tree an it has fixed!!! This was one of my few major issue with bazaar. Thanks man you rock [18:44] sinasalek: np. glad to hear you're making progress [18:45] fullermd: You should notify them about your work, they might be able to give you a hand and include at least part of it in official doc [18:46] Oh, it's known. It's jut very much a WIP (and I'm not quite sure it really fits into any good slot in the existing official doc gestalt) [18:54] My article has quite some visits, i included a link to your doc page. least i can do. keep up the good work :) http://sina.salek.ws/node/1314/revisions/1459/view?diff=1 [19:20] Do you consider Bazaar 2.2b4 stable enough to use? [19:21] Well, I track the dev head, so... [19:21] sinasalek: yes [19:21] the beta means that we are allowed to change internal apis [19:21] so plugins may not work [19:21] etc [19:22] it doesn't mean the code is actually unstable [19:22] jam: I tought that was what alpha is for [19:22] fullermd: You mean you're using dev?! [19:22] No, alpha is for sacrificing on an altar to try and get people to buy Itanium ;p [19:22] sinasalek: bzr.dev (the trunk branch) must pass its full test suite for every revision [19:22] most of the developers use it (dogfooding) [19:23] jam: That's great, i'm ok with crashes as long as it does not corrupt my repo [19:23] sinasalek: nothing to worry about for repo corruption [19:24] jam, fullermd: thanks , lets try it then :) === Meths_ is now known as Meths [23:00] bzr bd - I've got a branch which is working fine for both me and several other people, but one guy gets: [23:00] mtaylor: bzr: ERROR: Unable to find the needed upstream tarball: swift_20100728203750.orig.tar.gz. [23:01] it's using pristine-tar/import-dsc and all... he has the same version of bzr-builddeb I have - any ideas on where to look? [23:03] mtaylor: do you already have the .orig.tar.gz laying around on your system? [23:03] If you move it out of the way, and run again, does 'bzr bd' still work ? [23:03] mtaylor: perhaps it's an out-of-date version of builddeb? [23:03] hi poolie [23:03] jam: I tried in a completely fresh dir [23:03] otherwise i'd look at the traceback in that error to see just what it's checking [23:04] mtaylor: does that include not having it in the parent dir? [23:04] IIRC it always looks in ../ for the .orig.tar.gz [23:04] jam: yup [23:04] jam: I went to a completely unrelated tree [23:04] poolie: same version of builddeb - although I will say that what apt says the version is and what bzr says the version is do not match [23:04] mtaylor: and your bd output *does* include a line about 'pristine-tar' ? [23:04] mtaylor: that would indicate a plugin vs a system-wide install [23:04] (probably) [23:05] jam: started back from scratch and repulled the code. A dependency error came up for python-all. Fixed that and everything worked ok [23:05] mtaylor: well then, no problems here :) [23:06] jam: $ bzr plugins | grep builddeb [23:06] builddeb 2.2.0 [23:06] apt-cache policy bzr-builddeb [23:06] Installed: 2.4.2 [23:06] also - I cleaned out my ~/.bazaar/plugins dir [23:06] this isn't bothering me - I just thought I'd let you know about it while we were talking [23:09] mtaylor: bzr plugins -v should also show you what plugin it is loading [23:09] It is also possible that somebody updated one version string without updating the other. [23:10] jam: /usr/lib/python2.6/dist-packages/bzrlib/plugins/builddeb [23:10] I think I saw something recently from statik that hard-coded a version in setup.py and didn't use the one from package/__init__.py, which is certainly the type of thing that can cause skew. [23:10] * mtaylor pokes statik in the eye [23:10] I'm not sure if that is bzr-builddeb or another project right now [23:12] poolie: you seem to be back in your regularly scheduled TZ. How were your flights? [23:24] hi there jam; they were pretty tolerable [23:24] qantas sent me a voucher as an apology for the long delays on the way over, which was nice [23:24] poolie: I'm past due on stopping for the day, but it would be fun to chat when we can get a chnace [23:24] chance [23:25] ok, i'm away tomorrow to visit igc [23:25] i'll try to be online early tuesday and we could talk then [23:25] or you can call me later tonight [23:26] mtaylor: it's possibly an issue with the version of pristine-tar, but I guess you are on the same version? [23:27] james_w: it got sorted... somehow it was a missing python-all [23:27] ah [23:27] odd [23:33] python-all was not installed?