[01:12] bzr: ERROR: No module named configobj [01:12] is there some way to make that more verbose? [01:13] ah, .bzr.log has a traceback [01:13] -Derror [01:13] hi, bzr started crashing on me today when i tried to commit [01:14] I guess bzr-fastimport is out of date wrt bzrlib, or something [01:14] any ideas? trackback is here: http://pastebin.com/aiUXNkWm [01:14] idnar: I think we stopped bundling configobj [01:15] idnar: fastimport should be importing the system configobj, if it wants to play with configobj directly. [01:15] lifeless: yeah, this code tries to do "import bzrlib.util.configobj.configobj as configobj" [01:15] yeah, naughty. [01:15] tct_: most folk are on leave with easter [01:15] tct_: you might file a bug [01:18] thanks, will do [01:19] what... the heck is fast-import doing [01:19] I suppose I should just create the destination myself === ctrlsoft_ is now known as ctrlsoft === ctrlsoft is now known as Guest44441 === Guest44441 is now known as ctrlsoft [02:04] poolie: I <3 google - lmirrors pypi page is the first hit for 'lmirror' :> [02:05] that's pretty quick === BasicPRO is now known as iMacBasic [07:21] hi all ! [09:16] Hi all [09:25] hello guys [09:30] Hello reyman [09:35] i have question about bzr, and i see anything about that in help ... [09:35] i'm working with launchpad, and i want to work in a decentralized workflow [09:36] hi all [09:37] i checkout a trunk branch, but when i try to duplicate in a local branch with branch command, bzr says me " trunk is not a branch" :/ [09:37] hi bialix [09:40] reyman: I don't understand your problem [09:40] maybe it helps if you pastebin shell session [09:41] or example of commands [09:42] i make first a "bzr checkout /myproject/trunk trunk" , a mirror if i understand correctly the help. [09:43] then i want to work in decentralized workflow, users commit on local branch before mergin with human gate keeper on launchpad [09:43] checkout is the checkout or bound branch. It's not exactly mirror [09:43] hi bialix [09:44] you can convert your checkout to the simple branch with command `bzr unbind` [09:44] hi GaryvdM ! [09:44] ok [09:45] about mirror branch the help says : "To create a mirror branch, set-up a shared repository (if you haven’t already) and then use the branch (or checkout) command to create the mirror" [09:46] but why create a checkout branch, if i need to unbind after to duplicate the checkout with branch command [09:47] it's impossible to make a "bzr branch trunk newTrunk" when trunk is a "checkout mirror" [09:47] reyman: why it's impossible? it's not true AFAIK [09:47] No it's not... [09:48] hi fullermd ! [09:53] ello [09:53] olle [09:53] lole? [09:54] leol? [09:54] I hate to admit, I've been against Bazaar for a while, but now thinking of setting up a new revision control system.. Been comparing bzr and git.. Bazaar looks nice, but I'm just wondering if there are any updated benchmarks? [09:54] * AndrewLuecke has a 1-2GB repo [09:55] * bialix does not think bzr can beat git on such repo [09:55] Everyone seems to be saying Git is much faster, (and they both seem to have good features, with bzr more polished).. But the other benefits I see for bazaar seem to put it in the running [09:56] Yeah.. but I don't need it to beat all cases, but just not be much slower than SVN.. [09:56] I doubt it's arguable that git will be faster. But will git be pleasant, and bzr unpleasant? [09:56] Doubt you could say without trying. [09:57] bialix: so ok it's possible, but i try and i don't understand why i fail with this command :s [09:57] well, been going through the features, addons, etc. [09:57] reyman: pastebin please [09:57] oki thx bialix [09:57] reyman: also I'm not native english speaker it's hard to understand you what's wrong [09:58] bla - I don have my gpg key here. going to pop in to work to get it. See you all later. [09:58] i'm not native english too :] with bad bad bad english [09:59] from your opinions though, does Bazaar feel like it gets really slow? [09:59] !pastebin [09:59] For posting multi-line texts into the channel, please use http://paste.ubuntu.com | To post !screenshots use http://tinyurl.com/imagebin | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic. [10:00] reyman: ^ [10:00] Well, I've never used it with a 2 gig repo. There are a lot of variables other than size that have huge effects on speed too. It depends on shape of history, size and shape of tree, yada yada. [10:01] AndrewLuecke: if you want to know what is slow, you can try to run benchmarks on Windows [10:02] yeah.. only prob is, I'm in Australia, so quota'ed internet.. Must save downloads.. Not sure I could run a good comparison (maybe a local one I guess, but not sure how useful that would be) [10:02] but anyway, if it was woefully slow, I guess there wouldn't be as many people here, as there are now.. [10:03] * AndrewLuecke put up with SVN for this long anyway [10:04] one of my past coworkers said one day: the result of meterings is highly dependent on the methods of measuring. [10:04] so mostly all benchmarks lies [10:04] only your PC and the operations you're interested is make sense [10:04] yeah.. Normally I would stick the tree into both.. But hard to say [10:05] fact is, we don't know yet fully what is needed, because we are forking an open source project [10:07] What's the bzr equivalent of git reset --hard ? [10:07] meh.. fedge it.. maybe I will give them a proper test (technically, the code can be broken into 3 trees anyway, so each would only be 600MB anyway I guess. Most of its mozillas code I guess [10:08] artagnon: I believe that would be bzr pull --overwrite -r [10:09] pff i delete all my branch, checkout and restart cmd, and it's ok ... duplicate checkout in another local branch is ok.. sorry for inconvenience. [10:09] most likely bzr pull . --overwrite -r [10:10] Well, yeah. My brain typed the period; not its fault if my fingers didn't comply :p [10:10] reyman: there were gremlins [10:11] fullermd && bialix: Got it, thanks :) [10:11] artagnon: always happy to help [10:11] Huh? I just got a huge bunch of conflicts! [10:13] Did you have local changes? [10:13] Yeah :| [10:13] Mmm. pull may try to merge them. I can never remember if --overwrite affects that or not. [10:13] Well, git reset --hard blows them away too AFAIK, so you can just revert to get back to a pristine rev WT. [10:14] * artagnon nods [10:14] so bzr pull . --overwrite -r isn't the equivalent [10:14] Not quite. pull ; revert (or revert ; pull) would be I guess. [10:15] git reset blurs the lines. In bzr, pull affects the branch, and revert the WT. [10:16] ok, got it [10:17] what do I do now though? I just want to hard reset to HEAD and try all this [10:18] revert with no args will blat a pristine state of the current rev onto your working tree. After the pull, that 'current rev' will be what you set for there, so just 'revert' should do it. [10:20] Ok, nice. Now bzr status indicates a lot of unknown files though- how do I get rid of them? [10:21] Mmm. Probably a construction of bzr ls --someargs | xargs rm [10:21] artagnon: I haven't tried it before, but 'bzr clean-tree' is probably what you're looking for. [10:22] Probably --unknown -R ? [10:22] rm `bzr unknowns` [10:22] Oh look, another ls bogusness I find in checking that... how surprising. [10:22] `bzr clean-tree` FTW [10:23] in help there are a chapter "advanced layout", and the speak about the launchpad model, but when you use launchpad to host the code, you doesn't have choice with your layout for repository, no ? [10:25] Is it sound to pass a URL to bzrlib.config.LocationConfig()? [10:26] jszakmeister: works perfectly, Thanks :) [10:26] artagnon: you're welcome! [11:17] Hi all -- I have a branch, of a branch, of a "master" repository -- This branch is on a server. How can I create a branch of my remote branch so I can work locally, and then push my commits into my the remote branch [11:18] so right now I have branch-foo -> branch-bar -> master, and i want to have branch-local --|--> branch-foo -> branch-bar -> master [11:18] does that make sense? === Kamping_Kaiser is now known as Bambi_BOFH [11:20] defn bzr branch server/master [11:20] defn: you will then have a master branch on your computer [11:21] defn: I think that it is easier than you are expecting it to be :-) [11:52] GaryvdM: :) thanks [11:57] hi people, I'm having problems commiting to a launchpad branch, is launchpad down or something? [11:58] ah, yes, it seems so [12:18] GaryvdM: so would it look like: bzr branch user@remotehost:dir/of/my/branch [12:19] defn: maybe bzr branch sftp://user@remotehost/dir/of/my/branch [12:20] defn: how is the branch shared? [12:20] GaryvdM: im not sure im new to bzr [12:20] i have a branch in another user's branch, which is pulled into the master [12:22] defn: The branch can be shared over ftp, ssh(sftp), http, or anything that you can mount. [12:22] i have a dir with a .bzr/ dir in it, and then i created a branch named v1.0 in the same directory which has its own .bzr directory inside [12:23] that branch was created from the other user's branch [12:23] does any of what im saying make sense? [12:23] :) [12:24] defn: no, rather don't move the .bzr dirs arround yourself. [12:24] i havent touched the .bzr dirs [12:24] im getting "Not a branch" [12:25] defn: What are we trying to do again. I thought you were trying to get a branch from a server [12:25] bzr branch bzr+ssh://user@server/dev/repo/v1.0 [12:25] GaryvdM: im trying to do local development of that remote branch, like so i can commit locally when im not on the LAN [12:25] and then when i get to work i can commit those changes [12:26] And what happens when you do bzr branch bzr+ssh://user@server/dev/repo/v1.0 ? [12:26] server does not understand bazaar network protocol 3, ... error: not a branch [12:27] Who created the branch? maybe they did not? [12:27] is bzr installed on the server? [12:27] lifeless: yes but it's an older version i think [12:27] 1.3.1 [12:28] * GaryvdM -> lunch [12:31] okay so here is the layout... [12:32] another user on the same remote server created a branch for me in his home directory. on that server i created a branch in my home directory by doing: bzr branch /home/otheruser/dev/repo/branch branch-name === reyman is now known as reyman_aw [12:35] okay i think i got it [12:36] i guess my next question is, now that i've got a local copy of my branch, how do i commit locally only, so i dont need to have access to the remote server at all times? [12:37] if you downloaded the brach using 'bzr branch', just use 'bzr commit' [12:38] and then how, when im on the lan, do i push those commits to the remote branch? [12:38] push? [12:39] bzr push [12:39] yes [12:39] okay cool [12:39] thanks everyone === Bambi_BOFH is now known as Kamping_Kaiser [13:20] i don't understand why i have "sproject.dev/trunk" and not "/trunk" when i make bzr checkout "bzr checkout http://bazaar.launchpad.net/~reyman64/sproject/sproject.dev" [13:22] actually you should have only sproject.dev [13:22] the last part of URL used by default as the directory name [13:23] yes i have only a branch sproject.dev in serie trunk [13:24] hi [13:24] i want to make a human gate keeper flow, with a trunk only available for merge by me [13:24] i have problems with qlog because of that error: https://bugs.launchpad.net/ubuntu/lucid/+source/qbzr/+bug/544928 [13:24] Ubuntu bug 544928 in qbzr "qlog fails with a special combination of PyQt4 and Qt" [Critical,Confirmed] [13:25] ubottu: exactly that problem, any whay to solve that? [13:25] Error: I am only a bot, please don't think I'm intelligent :) [13:25] ^^ [13:25] lol [13:25] reyman_aw: you may want to push your branch as lp:~reyman64/sproject/trunk [13:25] is there any way known to solve that problem [13:25] nomatter001: downgrade the PyQt [13:25] bialix: no other way? [13:26] manual patch in pyqt/qbzr? [13:26] GaryvdM working on this, he's away for lunch now [13:26] bialix: yes, with no confirmation for me, and restriction review for other push'er [13:27] reyman_aw: lp:~reyman64/xxx is only writable by you [13:27] reyman_aw: check this page and read https://code.launchpad.net/~reyman64/sproject/sproject.dev [13:27] ok thx bialix [13:27] bialix: does it work with pyqt-4.7.0-2? [13:27] reyman_aw: when I'm looking at your branch I see the following message: "You cannot upload to this branch. Only sebastien.rey can upload to this branch. " [13:28] GaryvdM: are you here? [13:28] nomatter001: Yes [13:28] Gary the Wizard came to the room [13:28] GaryvdM: ok thx [13:29] and, so i have sprojet.dev in my repository after checkout, but why there are trunk folder into, don't understand ..if i want to create a fix , i need to branch the sprojet.dev/trunk or sprojet.dev only ? [13:29] reyman_aw: just advice: use english for the project description [13:29] or at least english + french [13:29] ok bialix thx for advice, i translate later ;) [13:30] reyman_aw: are you using Bazaar Explorer? [13:31] no [13:31] command line for understand :) [13:32] (try to understand ...) [13:32] wait a sec, I'll try to reproduce [13:33] bzr 2.1? [13:40] yes bialix [13:40] latest bzr on windows [13:46] okay [13:46] the same here [13:46] reyman_aw: you have trunk directory in your branch [13:47] hum :/ [13:47] the sproject.dev is the checkout root [13:47] do following: [13:47] cd trunk; bzr mv * ../ [13:47] cd .. [13:47] bzr rm trunk [13:47] bzr commit [13:48] ok so i make noob mistake in my manipulation when i create a trunk folder , arg [13:48] thx a lot bialix, it's clear now :] [13:48] sorry for this noob inconvenience [13:49] no problem [13:49] reyman_aw: remember: bzr info is your friend when you lost in the bzr forest [13:49] ok thx for advice :) [13:49] `bzr info` tells you what bzr object you have and where its root [13:50] i see yes [13:51] reyman_aw: you can try Bazaar Explorer [13:51] it provides some hints [13:52] ok [13:52] back to your question about workflow [13:52] you'd better create shared repo for local work [13:53] bzr init-repo sproject [13:53] cd sproject [13:53] i create a shared repo yes [13:53] bzr checkout lp:sproject trunk [13:53] the last command will create local checkout of your lp branch [13:54] then if you need to create new feature branch you do: bzr branch trunk xxx [13:54] ok [13:55] to merge your branch back to trunk do: cd trunk; bzr merge ../xxx [13:55] ok [13:55] after merge you need to check the result, resolve conflict if needed, and commit [13:55] and if i want to propose merge on initial checkout trunk [13:56] i make this with launchpad interface? [13:56] what it means? [13:56] if you want to create merge proposal? [13:56] or somebody else? [13:57] only ~reyman64 have the access of initial trunk, but if other want to create feature and want to merge, [13:57] yes bialix exactly [13:57] usually to create merge proposal one have to push his branch to lp [13:57] then do: bzr lp-open [13:58] the latter command will open the branch page, like https://code.launchpad.net/~reyman64/sproject/sproject.dev [13:58] there will be a button to create merge proposal [13:59] and ~reyman64 will receive the mail about merge proposal [13:59] ok :] [13:59] perfect ! [13:59] then to merge the branch from lp you need to cd to shared repo sproject [13:59] then: bzr branch lp:other-branch [14:00] cd trunk [14:00] bzr merge ../other-branch [14:00] and commit as usual [14:00] bialix: thx a lot for this tutorial, there are no real tutorial for decentralized workflow :/ [14:00] on launchpad [14:00] reaaly? [14:00] there is tutorials on Bazaar site [14:01] reyman_aw: http://doc.bazaar.canonical.com/bzr.2.1/en/tutorials/index.html [14:01] yes but, not really for launchpad, so i don't understand the layout branch for launchpad [14:01] http://doc.bazaar.canonical.com/bzr.2.1/en/tutorials/using_bazaar_with_launchpad.html [14:01] ask your questions, give us feedback [14:02] send proposals to ML how the docs should be improved [14:02] there was nice tutorial on arstechnica [14:03] anybody has the link to that article? [14:03] ok ! i search on arstechnica [14:03] i have this review http://arstechnica.com/business/news/2010/02/an-introduction-to-collaborative-development-with-launchpad.ars [14:04] yes, that's it [14:08] reyman_aw: what do you mean by "the layout branch for launchpad"? [14:14] bialix: in help there are a chapter " advanced layout" [14:14] link please [14:15] ah the end : [14:15] http://doc.bazaar.canonical.com/latest/en/user-guide/shared_repository_layouts.html [14:15] -h+t [14:15] layout " Simple developer naming (project/joe/foo, project/barry/bar) " [14:15] this is not very clear for newbies [14:15] like me :) [14:16] Btw, has anyone tested both HTTPS and ssh, which was faster? [14:16] Actually, nevermind, I'll just use SSH [14:16] bzr+ssh [14:17] reyman_aw: yep, there is the picture of tree of folders [14:17] reyman_aw: I think you don't need advanced things while you're newbie [14:18] reyman_aw: in short about lp: it uses the simple naming scheme: lp:~USER/PROJECT/BRANCH [14:18] now i understand that the folder tree is a snapshoot of the launchpad ,no ? [14:18] I'd say that doc should be improved [14:19] * vila is stunned by bialix explanations, who needs doc after that ? [14:20] bazaar have a great doc, and bialix help a lot ! Perhaps a tutorial on "how participate" on real launchpad project for dub :p [14:20] make things more clear [14:21] reyman_aw: patches welcome ! :) [14:21] eheh [14:21] I'm correct in believing bzr_access only allows r/rw/nothing right? Not anything special thats per directory (unless I create many users).. That is correct right? [14:22] reyman_aw: I'm only half-joking here, at this instant, *you* have the clearer idea about what is missing... [14:23] nevermind.. doesn't matter anyway :P [14:23] AndrewLuecke: roughly yes, there was some discussion about going further but I can't remember it if was 1) with bzr-access, 2) with another script, 3) only discussed [14:23] yes, i'm pionner in my entreprise, so i probably make a tutorial for the other later :] [14:23] reyman_aw: keep notes ! [14:23] and why not for communities [14:23] I think I heard somewhere vila, there is was ACL's, is that what you are refering to? [14:24] AndrewLuecke: not ACL as the OS ones, but yes, something about allowing more fine-grained control at branch level [14:24] ok [14:25] AndrewLuecke: but nothing *below* the branch level in case that's what you're after [14:25] vila: the bug for 1st april: https://bugs.launchpad.net/bzr/+bug/553955 (unfortunately today is 2nd) [14:25] Ubuntu bug 553955 in bzr "bzr 2.1: commit incorrectly reports missing files as modified" [Undecided,New] [14:26] nah.. It was more of a theoretical.. But I guess it doesn't matter, as I'll only probably have 1 primary repository anyway I guess [14:26] i have a problem bialix, i delete the false trunk in my sproject.dev, make commit, but i have BZR ERROR: cannot lock dir : transport operation no possible : http does not support mkdir() ( ?? ) [14:27] AndrewLuecke: I'm 80% sure someone has a working solution, I just can't remember the details (Alzheimer, etc) [14:27] reyman_aw: because you did checkout from http branch, which is read-only [14:27] hmm [14:27] ok vila.. It will probably pop up anyway [14:28] reyman_aw: do the following: bzr switch lp:xxx (where xxx is your lp branch) [14:28] ok [14:28] * AndrewLuecke is still wrestling with debian anyway [14:28] bialix: naah, you can't delete foo.txt, that's not allowed [14:29] AndrewLuecke, vila: there is ClueBzrServer project which I'm still planning to try out [14:29] thx bialix it's ok :) [14:29] vila: seriously? %-) [14:29] cool [14:30] bialix: no :) 'bzr remove foo.txt' is so engrained in mind that I cannot encounter such a bug, checking [14:30] phew [14:30] * bialix used Windows Explorer, but don't say anybody [14:33] * AndrewLuecke debates switching his server to ubuntu again [14:33] bialix: feel free to mark such bugs as confirmed and set an importance (I just did) [14:34] * vila needs to reboot bbiab [14:48] vila, come back [14:50] Bug 544928 Fixed. [14:50] Launchpad bug 544928 in qbzr "qlog fails with a special combination of PyQt4 and Qt" [Critical,Confirmed] https://launchpad.net/bugs/544928 [14:51] bialix: Because it affects so many people, I'm going to release 0.18.5 now. That ok? [14:52] GaryvdM: sure, please also put in the release my fix for backslashes [14:52] Yes [14:52] GaryvdM: congratulations on the bug fix! [14:53] I've planned to do 0.18.5 anyway tomorrow [14:53] GaryvdM: you can leave it for me [14:53] * GaryvdM grumbles about qt.... [14:53] * bialix grumbles about all new versions of all software === Kamping_Kaiser is now known as Bambi_BOFH [14:54] I'm developing a love hate relationship with qt. [14:54] (excluding qbzr of course) [14:54] or PyQT? === Bambi_BOFH is now known as Kamping_Kaiser [14:59] jam ? [15:26] vila: I'm loving BZR_PLUGINS_AT [15:27] GaryvdM: I'm glad you like it, but I wonder how you avoided bug #552922 though.... [15:27] Launchpad bug 552922 in bzr "BZR_PLUGINS_AT can be fooled by os.listdir order" [Critical,In progress] https://launchpad.net/bugs/552922 [15:28] GaryvdM: fix available at https://code.edge.launchpad.net/~vila/bzr/552922-plugins-at/+merge/22697 [15:28] wow, nice feature [15:28] vila: I don't know [15:44] woot.. I am such a fool.. Nobody cares, but its hard to explain the satisfaction you get when ssh actually authenticates properly [15:44] AndrewLuecke: You're not alone :-) [15:45] AndrewLuecke: about the satisfaction that is... [15:45] AndrewLuecke: nah kidding, I'm so a fool too :-P [15:45] serves me right for using 2 different texts to set up ssh (one told me that my home dir should be my repo, and I forgot I did that so clearly, authenticated keys wouldn't work) [15:46] ooh.. Import time.. Time to see how fast I can import from songbird's svn to prgmr [15:46] AndrewLuecke: 'ssh -v' is your friend too [15:47] it was a server side issue.. I thought I did the authorised keys thing wrong.. So dumb as I was.. I figured if I redid the same thing 5 times it would work.. [15:48] AndrewLuecke: good policy, served me right for the last 20 years, I use 2 instead of 5 though :-D [15:48] * vila kills more chicken [15:49] Speaking of chicken.. If I'm not here tomorrow, its because the chicken in the fridge I purchased 2 or 3 days ago wasn't safe anymore [15:50] tsk tsk, always sacrifice *live* chicken so you're sure they are fresh when you eat them... [15:50] * AndrewLuecke needs to find out how big the SVN tree is his importing, he suspects it wont fit on his VPS [15:51] Nah.. no need [15:51] Danger is my middle name... [15:52] btw.. there is a broken link on: http://doc.bazaar.canonical.com/bzr.2.1/en/admin-guide/migration.html#subversion-conversion [15:52] it points to http://doc.bazaar-vcs.org/en/migration/foreign/bzr-on-svn-projects.html [15:52] which isn't found.. [15:52] AndrewLuecke: please file a bug on lp [15:52] I assume we don't have access to edit that [15:53] How do I file a bug? [15:53] * AndrewLuecke is joking... [15:53] AndrewLuecke: the admin-guide is part of the sources, so you can even provide a fix if you know the right url :) [15:54] doc/en/admin-guide/migration.txt [15:54] not sure yet.. Anyway.. priority #1 is to start tearing the SVN to pieces (because that might take a day or so) [15:55] sure, I'm just trying to help you find something to do while waiting for the import :-D [15:55] Haven't started the import yet.. I'm concerned about waking up tomorrow and finding my VPS full.. Trying to find out how big it is first.. [16:06] Hrmm, I remember there being some caveats in the difference between lightweight/heavyweight checkouts that don't seem to be mentioned on http://wiki.bazaar-vcs.org/CheckoutTutorial [16:07] * AndrewLuecke wonders when google wave got proper extensions [16:07] and I vaguely remember things like "rebase" sortof ... break in a heavyweight checkout. [16:07] AndrewLuecke: "proper" extensions? It's had extensions from the start. [16:07] yeah.. but now, it actually has some in the list.. not only 2 or 3 (lots of em) [16:12] vila: can you help? It seems I found rather serious bug in bzrlib [16:12] April 2th, really 2th ! [16:13] bialix: don't ask to ask :-) [16:13] oh, it seems I understand [16:14] bug about relative path to the master of light checkout [16:14] rats [16:16] it kills mew [16:16] me [16:16] I think someone reported a bug about that recently or at least mentioned it on the ML or here [16:18] I can work with light checkout at all if it uses absolute path to another windows drive!!! [16:24] bialix: Have you noticed a difference in qbzr startup time? [16:24] bialix: Have you noticed a difference in qbzr startup time? [16:25] no, I havent [16:25] o I havent [16:25] GaryvdM: it's twice as slow ? [16:25] GaryvdM: it's twice as slow ? [16:25] no, I havent [16:26] :-( Sorry about the dup, I push up+enter, the wrong window had focus :-) [16:26] we're joking [16:26] relax [16:26] :-) [16:26] :-) [16:26] :-) [16:26] no difference though? [16:27] * bialix just hits the wall with relative paths [16:27] GaryvdM: in which branch/revno? [16:27] bialix: I thought poolie may have landed a related patch recently but I can't find it.... [16:28] vila: I'm not sure [16:28] bialix: anyway if things have changed and broke your workflow, raise a bug [16:28] bialix: in trunk since rev 1229 (2010/03/25) [16:28] vila: things are not changed I belivev, I've just found another use case when absolute paths are show stopper for me [16:29] GaryvdM: oh, I'm often working/dogfooding with 0.18 [16:29] bialix: haaa, then I may be I've got everything backwards and the discussion was about using relpaths instead [16:30] GaryvdM: but no, I don't think I've noticed any highly visible improvements [16:30] GaryvdM: do you have benchmarks? [16:30] vila: relative paths will helps [16:32] GaryvdM: when you'll finish with 0.18.5 can you look at my question in qbzr ML about qlog+colo? [16:32] please [16:33] Sure, I have a half finished draft [16:33] Such a change will need quite a bit of plumbing work in qlog... [16:33] bialix: did you see my reply? [16:34] jelmer, is it you? [16:34] ctrlsoft: no, I don't [16:35] reply for what? [16:35] bialix: your email about colo support in qlog [16:35] s/qlog/qbzr [16:36] ctrlsoft: no, is it recent? [16:36] bialix: a couple of days ago I think [16:36] I've sent my question just yesterday [16:37] I don't see any your mails, ctrlsoft [16:37] ctrlsoft: can't see it either. Maybe got filtered by google groups spam filter. [16:37] hmm :-( [16:38] bialix: to summarize it; bzr 2.2 has support for colocated branches in its API [16:39] I have to admit that google groups became worse in last weeks [16:39] bialix: it would be nice if qbzr and bzr-colo supported that, that would also give you support for colocated branches in git repositories [16:39] ctrlsoft: that sounds great [16:40] can you CC your mail to bzr ML, please? [16:41] ctrlsoft: you have joined to qbzr ML with your samba mail [16:41] your mail have to arrive, I believe === salgado is now known as salgado-lunch [16:42] bialix: For inno setup, should I us the unicode, or non unicode version? Does it make a difference to the output? [16:42] *use [16:43] GaryvdM: currently I'm using 5.3.3 which is pre-unicode IIRC [16:43] * bialix is just lazy to upgrade [16:44] in short: I dunno [16:44] OK [16:44] vila: is it new thing that push to another drive letter don't create working tree? [16:45] meh, I don't think so, are you sure you're not pushing into a --no-trees repo ? [16:46] * bialix checks [16:46] vila: info -v says: Create working tree for new branches inside the repository. [16:47] bialix: no idea then, but I don't recall any change in this area either [16:47] okay, time for bug report [16:49] hmmm, can't reproduce [16:50] with simple branch I can't reproduce [16:50] bialix: kill more chicken [16:50] what's that? [16:51] bialix: black magic against the gremlins [16:51] oh, cool [16:51] * bialix notes [16:52] level-up! [16:52] :-) [16:53] OIC, I'm pushing from colo workspace, which uses treeless branches and light co [16:53] * GaryvdM checks if he is in the right channel.... [16:54] bialix: oh - thats how bzr qlog sees the colo branches... [16:55] GaryvdM: bzr qlog colo: ? [16:56] yeah === IslandUsurper is now known as IslandUsurperAFK [16:56] vila: yes, I can reproduce [16:57] bialix: fyi: I get an error with the unicode version of inno. It went away when I installed the non unicode version... [16:58] GaryvdM: ok, thanks for the info [16:58] perhaps wine issue? [16:58] maybe wine related... [16:58] yes [16:58] http://pastebin.org/131345 [17:00] btw - it rocks that I can now build the windows installer on ubuntu. [17:00] strange [17:00] indeed, rocks [17:00] wine is cool [17:01] GaryvdM: you should 1) backup your HD, 2) wipe it clean 3) insall Ubuntu, 4) install a windows VM :) [17:02] bialix: you too :-P [17:02] vila: :-P [17:03] and who will report about windows bugs then? [17:03] bialix: you can still use windows in the VM :-) [17:03] vila: The problem is that I have allsorts of windows apps that I'm not sure If I have instalers for. [17:04] bialix: Some addictions just can't be cured [17:04] * bialix prouds to be addicted [17:04] :) [17:04] vila: It would be cool if I could boot my existing windows installation in a vm... [17:05] GaryvdM: That will be a good occasion to cleanup :) [17:05] I'm using windows also because there is a lot of PC games for my daughter [17:05] (and sometimes for me) [17:05] GaryvdM: I've never done it, but I think it can be done [17:06] vila: I pritty much weened of it now. The last real thing that I need is to be able to run classic asp on linux. [17:06] bye all, and thx for the advice ! [17:06] For maintenance. [17:07] * bialix waves [17:07] exit [17:08] GaryvdM: just checked, VirtualBox can use raw host hard disk from a guest, either an entire physical hard disks or selected partitions [17:09] Oh cool - I'll try that... [17:10] GaryvdM: but if I had to try that, I'd make sure to have backups, just in case something weird happen like windows suddenly thinking that it's running on a new hardware because the video card is not the same anymore or something like that [17:10] GaryvdM: this is from chater 9.5 Advanced storage configuration in the vbox doc by the way [17:16] qbzr 0.18.5 code name Silver terminalia [17:16] nice! [17:26] bialix, GaryvdM : Can you run python -c 'import paramiko; print paramiko.__version__' and tell me which version is shipped with bzr ? === deryck is now known as deryck[lunch] [17:27] vila: no [17:27] hehe [17:27] it's not easy with py2exe [17:27] wait a minute [17:30] vila: it seems 1.7.6 (Fanny) [17:30] bialix: cool, thanks [17:31] * bialix was lazy to install depends plugin === IslandUsurperAFK is now known as IslandUsurper [17:33] vila: 1.7.6 (Fanny) (on lucid) [17:34] * bialix eods [17:34] bye all [17:34] bialix: [17:34] GaryvdM: yeah, I know, babune reports a test failing on lucid and so far I've got 1.7.4 passing on karmic and 1.7.6 failing on lucid, but maybe it's pycrypto instead [17:34] bignose: enjoy your week-end !! [17:34] too late :-/ [17:34] and wrong nick :-( [17:35] sorry bignose that was for bialix [17:36] jam: ping, I seem to remember you mentioned something about paramiko/pycrypto on windows but I can't remember the details [17:37] hi [17:37] jam: something about either patching one package or the other or going with a workaround [17:37] yeah ! jam is here ! Good morning ! [17:37] so paramiko uses a RandomPool from pycrypto IIRC [17:37] and [17:37] yeah, along these lines [17:37] 1) With older pycrypto and MS < Windows [17:37] it loops around time.clock() 100 times, with 15ms granularity (taking 1.5s to start) [17:38] 2) With MS Vista at least (sorry time.time()) granularity goes to 1ms [17:38] so it is only 100ms to start [17:38] 3) We can patch pycrypto to use time.clock() [17:38] 4) Newer pycrypto doesn't have the bug, but has completely deprecated RandomPool [17:38] right, but during the discussion I think it was mentioned another point about some, yeah that [17:38] so raises a warning that paramiko is using it [17:39] 5) So we have to patch paramiko to work around *that* [17:39] So at the moment, we stick with older pycrypto and patch it [17:39] We could upgrade to newer pycrypto and patch paramiko somehow [17:40] right, so across all our platforms, freebsd, gentoo and now lucid are failing because paramiko use RandomPool [17:41] (not 100% sure for lucid, but gentoo and freebsd now says: PID check failed. RNG must be re-initialized after fork(). Hint: Try Random.atfork() [17:42] with a traceback from crypto to paramiko [17:43] jam: ok, so patching paramiko seems to be the only viable alternative there, I'm EODing, but I'll file a bug first thing when I came back [17:44] bye all, have a nice week-end ! [17:46] jam: just one last thing, when/where do we *require* paramiko ? [17:47] we have to have it for sftp support [17:47] night [17:47] and thanks [17:47] we *can* use it as an ssh agent instead of openssh [17:47] so bzr+ssh shouldn't require paramiko if you have openssh [17:48] jam: right, thanks for confirming === salgado-lunch is now known as salgado === deryck[lunch] is now known as deryck === radoe_ is now known as radoe [20:50] hey guys [20:52] is there a way to display just *my* commits [20:52] I'm expected to provide a log of my work each day and it would be ideal if I could run something like bzr log -u email@example.com [20:54] bzr log -m email@example\\.com [20:58] lifeless: doesn't seem to work on this server. is there a special requirement? [20:59] * psynaptic pasted http://pastie.textmate.org/private/gepmld86ygjflga6qro3a [20:59] this is what I tried [21:00] uhm, it might not search author/committer; file a wishlist bug [21:00] ok, thank you [21:21] vila: ping [21:31] lifeless: can you tell me how to get a diff of the changes since a given revision MINUS the changes merged in from trunk? [21:32] uhm [21:32] make a temp branch [21:32] of that rev [21:32] yep [21:32] merge trunk to it [21:32] make a temp branch of tip [21:32] merge trunk to that [21:32] diff --old first-temp --new second-temp [21:33] i'd tried *most* of that. let me try. thanks [21:33] (this feeds into why looms and pipelines have things merged together) [21:34] I'm having problems when pulling from a repo. [21:35] bzr: ERROR: KnitPackRepository('local-repo') is not compatible with CHKInventoryRepository('lp-repo') [21:35] different rich-root support [21:41] magcius: the remote repository has more information than the local one, so the revisions can't be losslessly stored locally [21:41] magcius: you might want to upgrade to a newer repository format [21:41] ("bzr upgrade") [21:41] ctrlsoft: um [21:41] ctrlsoft: you should have a better error message, or do that automatically [21:41] ctrlsoft: because it was an older version before, so when I try to pull it should try to upgrade [21:41] magcius: I agree a better error message would indeed be a good idea, I think there's an open bug about that. [21:42] magcius: we don't want to upgrade automatically because it will make the repository unreadable by older versions of bazaar [21:58] lifeless: pong [21:59] hai [21:59] bzrlib's HTTP transport [21:59] is it threadsafe? [22:00] and if I were to do a 'get' on it, does it buffer the full response still ? [22:00] vila: ^ [22:00] hmm, I can't think of any part that shouldn't. [22:00] vila: persistent connection cache ? [22:00] pycurl buffers, urllib doesn't [22:01] the connection is shared between the transports that use it, there is no global cache [22:01] lifeless: the only weak point is when the server and the client are in the same process but I don't think you care about that [22:02] and even there it's more a python bug than unsafe thread code [22:04] lifeless: when you do a get, the urllib transport try to make all reads goes to the socket without buffering (at least for the data, headers are a different story but again you should not care about that) [22:05] can someone explain this though? "We identify revisions using sequential numbers per branch, not per repository." [22:05] I thought branch was synonymous with repository? [22:06] magcius: bzr help init-repo [22:07] http://wiki.bazaar.canonical.com/MatthewFuller/SpotDocs/PiecesInBrief [22:08] can bzr revision numbers theoretically get to 1.6.12.46.76.1.6.78/123 and so on? [22:08] s:/:.: [22:08] seriously, read http://wiki.bazaar.canonical.com/MatthewFuller/SpotDocs [22:08] short answer, no, they can't [22:09] magcius: no, they are either 123 or 123.456.789 i.e. one or three numbers [22:09] wait, what? [22:09] sorry, I'm a bzr noobie [22:09] magcius: yeah, luks gave you a very good url for that [22:09] vila: right, I don't care about that [22:09] vila: is there control over cache-control yet ? [22:10] got an older revision, since that revision I have removed and created new files. How to restore the older revision? [22:10] lifeless: no, we explicitly disable the caching unconditionaly [22:11] RumblePure: bzr revert -r X path/to/file [22:11] if I understand correctly that you want to undelete a file [22:11] vila: yeah, I thought so [22:12] vila: is it at least in a function I could override ? [22:12] lifeless: see -urlib2_wrappers.AbstractHTTPHandler._default_headers [22:12] How are the pieces stored on the disk? Is it pickle or something? [22:12] which pieces? [22:12] hello all. i was looking at bug #518609 [22:12] Launchpad bug 518609 in bzr "Unicode exception occurs by "version-info"" [Medium,Confirmed] https://launchpad.net/bugs/518609 [22:12] vila: bug 120697 [22:12] luks: the files, the commit metadata [22:13] Launchpad bug 120697 in bzr "want a way to mark transport operations as cacheable" [Medium,Confirmed] https://launchpad.net/bugs/120697 [22:13] lifeless: overriding AbstractHTTPHandler.http_request should be a good starting point [22:13] it seems that the exception occurs due to log.add('message', message) in format_rio.py but message is already unicode [22:13] http://pastebin.com/XP5VfTJt [22:13] hth debbian things its fixed, I dunno [22:13] any ideas what the problem might be? [22:14] magcius: there are various repository and branch formats in bzr [22:14] parthm: it shouldn't be unicode [22:14] luks: simple enough. thx for your help. :) [22:14] lifeless: yes I remember this bug [22:14] luks: is there any documentation on them, if I wanted to, say, implement a bzr web viewer in PHP or Perl? [22:14] magcius: only the source code, I'm afraid [22:14] lifeless: especially comment #9 :) [22:15] but reading them directly is not going to be easy [22:15] lifeless: oh. so does bzr store message as regular 8-bit strings? [22:15] byte strings [22:15] luks: ouch [22:16] magcius: bzr, like all VCS's, is a database. [22:16] magcius: bzr is more a vcs framework than a simple application :) [22:16] magcius: would you write a viewer for Postgresql or MySQL databases in PHP/Perl ? [22:16] lifeless: it's been done [22:16] the interface (both user and API) was always more important than the file formats [22:16] magcius: well, its that order of complexity. [22:17] you need to implement the B+Tree index logic, groupcompress compression engine, pack file format, to have any hope [22:17] and then on top of that you can start implementing user visible things. [22:17] those three things are well documetned, in pydoc. [22:18] and hope bzr will not change it's default format in the next version :) [22:18] lifeless: do you have a library written in something like C that I can get bindings to? [22:18] luks: that doesn't really matter, a given website can stay on any format it wants indefinitely. [22:18] magcius: you can call python from C; so our existing Python code is accessible from C. [22:19] lifeless: um [22:19] and bzr is structured very cleanly as front-end and library. [22:19] lifeless: I assume if somebody invests time in writing a reader for the 2a format, they want to actively use it [22:19] there is also bzr-xmloutput, xml bindings for most commands. [22:19] and bzr doesn't tend to support old formats indefinitely [22:20] magcius: loggerhead is a bzr web viewer written in python https://launchpad.net/loggerhead [22:20] luks: we still support the very earliest alpha formats [22:20] parthm: yes, I've used loggerhead and seen it break thousands of time [22:20] lifeless: usually only for conversion [22:20] luks: its only the absolutely oldest ones (that couldn't even do networking or merge) that are readonly [22:20] you wouldn't want to use them [22:20] luks: !citation time, you're making FUD [22:20] magcius: bug reports and patches welcome :) [22:21] lifeless: :) [22:21] a little too serious? :) [22:21] vila: I made a feature request a while ago: indentation lines or a left margin, it's awful reading Python code in Loggerhead because I can't tell where global scope is [22:22] luks: he's right, the test suite guarantees that and fully pass for every commit on mainline... [22:22] vila: second, fix your URLs, I don't want to see %2F or whatever : is in my URLs. Third, I don't like how you have different URLs for reading files or trees. I should be able to change the URL to delete the filename and make it work. [22:22] vila: yes, I know they are fully working [22:22] vila: but they are inneficient [22:22] magcius: I didn't write loggerhead [22:22] luks: magcius have you filed a bug for that ? [22:23] sorry [22:23] magcius: have you filed a bug for that? [22:23] vila: and they get more inneficient as new formats get developed [22:23] lifeless: I was told that both bugs already existed earlier [22:23] luks: well I surely the newer formats are more efficient [22:23] luks: they shouldn't be more inefficient than their intrinsic capabilities [22:23] lifeless: you are right `log.add('message', message.encode('ascii', 'ignore'))` fixes the issue. but is this the right place to convert the message to ascii? or do i need to find out why the message is becoming unicode in the first place? [22:23] luks: we have removed unsafe optimisations that we used as a stopgap while preparing better formats [22:23] magcius: I don't clearly see how reading the low-level bzr data will help you there ? [22:23] vila: knits were getting alower along with packs development, the original packs were getting slower along with ... [22:24] parthm: I have no idea what you're doing, so I can't really answer that. [22:24] parthm: perhaps mail the list ? [22:24] vila: anyway, I never wanted to argue about this :) [22:24] parthm: unless vila has time to dig into this - I don't right now,s orry. [22:25] I only said that this would be a reason for me to not implement bzr's internal file formats [22:25] luks: I won't deny that but I won't say that the degradation were more important than the optimizations either [22:25] lifeless: i will do that. thanks for the pointer on message being ascii. [22:25] luks: that's for sure :-) [22:25] lifeless: mailing list is probably good :) [22:26] vila: so - http://launchpadlibrarian.net/10490569/120697.patch [22:27] lifeless: yup, still valid (still not usable as is but in essence that's what controls the caching) [22:27] * vila giggles at the # Begin bundle, long time not seen :) [22:28] vila: I've suggested an API [22:29] lifeless: I'll happily review any patch you can come with :-P [22:37] lifeless: but roughly, caching packs/* and indices/* should cover at least 90% of the possible gains no ? [22:53] vila: yes [22:53] I don't plan to do that [22:53] I want caching in another project, dealing with large data sets