[04:22] jelmer: that import is stuck at about average 10 rev per minute at [04:22] jelmer: fetching revisions 149790/264178 [04:22] Should I be trying the "pull 1000 revisions at a time in a loop" trick, maybe? [04:26] (the bzr process is at over 3 GB VSZ, so there's now a lot of swapping going on. It's pretty ugly) [06:47] well that was dissapointing [06:47] bzr ran out of memory =( [06:56] after 15 hours of fast-import xD [06:56] AuroraBorealis: mine hasn't run out of memory, but at 25 hours elapsed it is essentially stuck, constantly swapping [06:56] are you still branching? [06:57] [bad Python VM doing reference count based GC, bad] [06:57] AuroraBorealis: yep [06:57] what are you branching, a git repo? [06:57] AuroraBorealis: the kernel, yes [06:57] as we were discussing yesterday [06:57] yeah thats why its taking so long =P [06:57] i thought you were branching a bazaar repo [06:57] if it's making progress, I don't care. But it's just spinning it's wheels. [06:58] | fetching revisions 151486/264178 [06:58] It was at 140000 this morning [06:58] mine got to 16,000 [06:58] then ran out of memory as it was repacking the repo :> [06:58] 3 GB virtual memory, 1.5 phys... swap swap swap [06:59] I'm going to try the pull a 1000 revisions at a time trick. [06:59] That worked on GTK a couple years ago when pulling from Subversion [06:59] → pub [06:59] heh [07:00] but yes, this could defenitly be faster. and not run out of memory xD [09:41] I'm following the Emacs Bzr guide. It says to do "bzr branch nosmart+bzr+ssh://iaindalton@bzr.savannah.gnu.org/emacs/trunk trunk", but I get Permission denied (publickey). bzr: ERROR: Connection closed: Unexpected end of message. (etc.) Why is it doing this and how can I fix it? [09:50] your ssh key is not present on the server. I don't know why its not, nor the process for getting it on that server - I would hope the guide covers both points. [09:53] It doesn't AFAICT [12:18] Is there any way to know the repository size before downloading it completely? [12:20] rly: "bzr info -v" should tell you the number of revisions [12:21] jelmer: also, how does the whole brz foo:bar thing work? [12:21] rly: do you mean lp: ? [12:21] jelmer: yes [12:22] rly: lp: is a shorthand for addressing branches on Launchpad, provided by the "launchpad" plugin that's bundled with bzr [12:22] hey guys, can someone please help me with an bzr - buildbot issue? It might not be related to buildbot at all, that's why i am asking here [12:22] jelmer: bzr branch lp:calibre is already at 600MB (!). [12:22] jelmer: I want to know how long it will continue. [12:23] I think the Linux kernel is even smaller. [12:24] andy: please ask your question, if anybody here can help we'll reply [12:26] alright, thanks [12:26] i got a buildbot master running. a bzr repo is hooked to it. so the bzr_buildbot.py is my plugins directory. i also adjusted the locations.conf file. but everytime i am trying to check something in, ill get an error: [12:27] $:~/tmp/repo$ bzr ci [12:27] Committing to: /home/.../tmp/repo/ [12:27] modified setup.py [12:27] bzr: ERROR: exceptions.AttributeError: 'Revision' object has no attribute 'get_apparent_author' [12:27] the script i used can be found here: https://raw.github.com/buildbot/buildbot/master/master/contrib/bzr_buildbot.py [12:27] andy: ...What version of bzr? [12:28] andy@europa:~/tmp/buildbot/basedir$ bzr --version [12:28] Bazaar (bzr) 2.3.4 [12:28] Python interpreter: /usr/bin/python 2.7.1 [12:28] Python standard library: /usr/lib/python2.7 [12:28] Platform: Linux-2.6.38-11-generic-x86_64-with-Ubuntu-11.04-natty [12:28] bzrlib: /usr/lib/python2.7/dist-packages/bzrlib [12:28] Bazaar configuration: /home/.../.bazaar [12:28] yeah ok ok ok! [12:28] Bazaar log file: /home/.../.bzr.log [12:28] sry :-) [12:31] buildbot doesn't even have to be running if you want to get that error [12:32] the problem might be here aswell: [12:32] $ cat locations.conf [12:32] [~/tmp/repo] [12:32] buildbot_on = change [12:32] buildbot_server = localhost [12:36] rly: it finished pretty quickly here [12:36] rly: what version of bzr are you running? [12:37] jelmer: are you on a 100Mb network? [12:37] rly: 336.910 Transferred: 390815kB (1160.6kB/s r:390814kB w:1kB) [12:37] jelmer: I have a 6 as a first number? [12:37] rly: what version of bzr are you running? [12:38] jelmer: 2.1.2 [12:38] there have been some major performance fixes recently, perhaps that's relevant here [12:39] rly: I suspect 2.4 will be a lot faster for you, and use less bandwidth [12:39] jelmer: what kind of connection do you have, btw> [12:40] rly: it took 5 minutes at 1.1Mb/s [13:13] do you have any idea if gary poster (the maintainer of the bzr_bulidbot.py) is here sometimes? [13:17] andy: he's often in #launchpad-dev [13:18] andy: though usually during the week rather than in the weekend [13:28] andy: still there? [13:30] andy: the call to get_apparent_author() should be changed to get_apparent_authors()[0] [13:34] Ah, I *knew* there was an API compatibility issue around get_apparent_author(), but I was thinking that it was too new, not that it was too old. It was killed around 2 years ago, wasn't it? [13:35] Peng: yep, and then deprecated for a long time. I didn't realize it had been deleted [15:39] Peng: sorry, was afk. I found the same solution. should i file a bug for this? or are you able to fix it? [15:40] Peng: in the buildbot IRC they told me to file a bug with the fix.... but if you can fix it right away this might be better. [15:51] Sorry, not me. Filing a bug's probably the best bet. [15:51] merge proposal [15:52] Bah. / merge proposal. [15:52] (* Actually, I might have commit rights to it, but I've never touched it.) [15:54] Peng: alright. please give it try, otherwise i am gonna file it [15:59] I don't wanna give it a try. >.> I feel weird jumping in. [16:00] Wait, it's part of buildbot, not part of bzr? Then I certainly don't have commit rights. And I've only used Github once in my life. :P [16:00] ah, sorry, i missunderstood it [16:09] hi [16:14] vila: I'm looking at the UI thing, and I'd like to be sure about you use case [16:14] *your [16:15] on my side, I don't call bazaar from VIM, but rather use VIM python bidings to directly use bzrlib [16:16] I provide my own UIFactory implementation, but since shelf_ui does not use get_boolean, it does not work [16:17] calling bzr from VIM (e.g. :!bzr shelve), does work [16:18] using "yes no | bzr shelve" in a terminal however, does not [16:21] so my question is: looking at your change, it seems that the isatty check is not enough, why? [16:33] bpierre: almost not there, I'll try to be quick [16:33] "yes no | bzr shelve" is seen as 'y' 'e' 's' ' ' 'n' 'o' [16:34] not on the version without your patch [16:34] bpierre: with patch I landed and INSIDE_EMACS=1 it will see 'yes no' [16:34] s//the/ [16:34] sorry, the example was using the unix yes command [16:35] bah, sorry [16:35] :) [16:35] you still need INSIDE_EMACS :) [16:35] so can't have a controling terminal when using shell command in emacs? [16:35] switch to VIM! (kidding) ;) [16:35] hehe [16:36] so now that it's in UI, better to change the environement variable name [16:36] no? [16:36] the trick is that the isatty() is not appropriate under emacs, there is no way to automatically detect the right kind (char or line) [16:36] I know there is one for progress [16:36] ok [16:36] the plan *is* to change that, yes [16:36] gtg, bbl [16:36] ok [16:37] i.e. prpose your branch, we'll sort out the details :) [16:37] bpierre: I'm patch pilot next week anyway, so will have more time [16:37] ok [16:44] bpierre: back, urgent, but quick ;) [16:44] Peng: i made a pull request for the fix. Thanks for your help [16:45] andy: It was mostly jelmer's help :P [16:45] bpierre: start by proposing your branch and if you still want to tweak before it's reviewed, you can. Complete your initial cover letter in the commit messages for your tweaks ans just 'bzr push' [16:45] s/ans/and/ [16:46] ok [16:46] grrr, does someone knows an IRC client with a spell checker ? [16:46] can you just tell me the command you use to test in emacs? [16:46] s/knows/know/ crap [16:46] hehe [16:46] bpierre: it won't work for you :) [16:46] weechat has an aspell plugin [16:47] I tried with M-!, but even with line based input, it does not work [16:47] I don't use emacs at all, have not for a long time [16:48] emacs sets INSIDE_EMACS=1, so *I* don't have to type but you do. After that you just use ./bzr selftest -s bt.test_shelf_ui [16:48] to test your changes [16:49] or if you want to observe the behavior of 'bzr shelve': 'INSIDE_EMACS=1 ./bzr shelve' [16:49] by how do you control the shelf ui from emacs? [16:49] oh ! Sorry [16:49] with VIM, :!bzr uncommit [16:49] I run a shell under emacs [16:49] works ok [16:49] so how do you start the shell, that's all I'm asking! [16:49] a regular bash shell *except* the input is line based [16:50] M-x shell [16:50] M for meta, 'ESC x' works too [16:50] ok! and not M-! command [16:50] got it [16:53] bpierre: just curious, how well do you know emacs ? [16:53] hmm, you have it installed obviously [16:54] no well at all! just installed it, for that and testing notmuch, started with it, before converting to vim :) [16:54] 10 years ago [16:59] bpierre: OMG, you went to the Dark Side ! vi, vi, vi VIVIVI is the editor of The Beast ! [17:15] bpierre: I've seen rms in flesh as St IGNUcis for the first time recently, the guy has a strong sense of auto-derision ;) http://stallman.org/saint.html [17:18] yeah [17:37] bpierre: A *good* way to understand what my patch fix: with an older bzr version: add a line in a file, try to shelve it. [17:38] bpierre: and watch your added line still present 8-) [17:38] should i report a bug if bzr ran out of memory doing something? [17:39] bpierre: you just type 'y\n' and shelve sees 'y' '\n' (default n) '\n' (default n) and think you don't want to shelve it finally :) [17:41] AuroraBorealis: es, that would be helpful and awesome! Especially if you mentioned your hardware config ( mem size, proc, etc) and the software one (bzr version, plugin versions, etc) and of course the data (urls, formats, revision ids, etc) so we can reproduce and compare [17:42] AuroraBorealis: we may need time to fix it but we'll know how close we get along the way ;) [17:42] k [17:42] i was trying to...convert the git repo for the linux kernel [17:42] and first i ran out of memory, then i ran into this bug =P https://bugs.launchpad.net/bzr/+bug/541626 [17:42] Ubuntu bug 541626 in Bazaar "'BTreeBuilder' object has no attribute '_find_ancestors'" [High,Triaged] [17:46] AuroraBorealis: 2.1.0 ?? [17:47] do you mean thats the bzr version i'm using? [17:47] i'm using 2.4.0 [17:47] AuroraBorealis: You'd really get better performance and memory consumption with 2.4 which is stable or 2.5b1 which is beta [17:47] oh, known bug ! Sorry, let me finish reading [17:47] xD [17:49] i also have 8 gb of memory, but i guess the version of python bzr uses is 32 bit [17:49] AuroraBorealis: sry, gtg, metoo the bug and subscribe [17:50] kk! [17:50] AuroraBorealis: and the informations mentioned above (and check the comments ?) [21:06] so I have a recipe set up for my project, and the source is being imported from a git repository by launchpad [21:06] I need to add a tag "upstream-0.2.4t" [21:06] I don't know anything about bzr, but I assume that's to tag a particular revision as 0.2.4? [21:07] since the debian packaging info says it's building 0.2.4 [21:09] I'm not sure how I'd do this since the branch is coming from git [21:16] udp: hi [21:16] udp: still there? [21:16] Hi jelmer, yeah I am [21:17] udp: git imports should import tags as well, so if you create one in Git it should create a matching tag in bzr [21:17] udp: that said, why do you need that particular tag? It shouldn't be necessary for the recipe. [21:17] ah, I didn't realise it did [21:17] thanks - well, I'm getting "bzr: ERROR: No such tag: upstream-0.2.4t hooks - Stage 5/5" [21:18] udp: try building with --allow-fallback-to-native [21:18] I guess it'd be building HEAD if I did that? [21:20] udp: well, this is about how it determines the upstream source [21:21] udp: if you specify --allow-fallback-to-native then it will simply create a "native" debian package if it can't find the upstream source [21:23] OK, I don't get the error if I use --allow-fallback-to-native [21:24] I'm fixing the error message to be a bit clearer. [21:27] can I add --allow-fallback-to-native to the recipe itself? [21:27] udp: no, but launchpad has it enabled already [21:30] ah ok - cheers [21:41] odd, there's a *_source.changes file being generated in the working directory when I run dailydeb - it includes an incorrect Changed-By email address which is causing lintian to fail (it uses my local hostname) [21:42] it's not the address from `bzr whoami` [21:44] udp: I'm not entirely sure what the logic behind that is [21:57] jelmer i posted my bzr version to that MemoryError bug [21:59] never mind, starting a new terminal solved it - I tried to build on launchpad again, but it failed with "bzr: ERROR: Failed to apply quilt patches" [22:00] there aren't any patches [22:00] udp: is there a debian/patches/series file? [22:00] yes, it's blank [22:00] udp: launchpad is still running a fairly old bzr-builder which has problems with that [22:02] udp: if you remove debian/patches it should work [22:11] AuroraBorealis: ha, I should have refreshed the bug page [22:11] AuroraBorealis: thanks [22:11] i have 8 gb of memory, so i would think that it shouldn't..run out o.o [22:12] AuroraBorealis: I'm not sure about bzr-fastimport, but I remember importing linux with bzr-git at some point [22:12] you need to install it youself from parts if you want that to be true. [22:12] auroraborealis: I'm also pretty sure the Launchpad vcs-imports don't have that much memory [22:12] yeah i would think not [22:12] because bzr only provides 32-bit windows installers [22:13] it seems that the importing was fine [22:13] but repacking it caused it to run out of memory [22:13] jelmer: launchpad doesn't run on widnows. [22:13] ...or windows [22:13] true [22:13] maybe John will have something useful to say about this on Monday [22:13] I posted some observations to the bug. [22:14] there are like, three things we could fix that I can see, but I don't know which are important without actual memory measurements [22:15] is there a way to do memory measurements? [22:15] is that what lp:meliae is? [22:15] jelmer: It built successfully on launchpad - thanks for your help :-) [22:15] AuroraBorealis: yep, meliae can indeed help with memory usage analysis [22:15] yup, if you have that when you OOM it does a dump. [22:16] udp: cool, glad I could help [22:16] ok i shall try this out [22:16] i might run out of disk space though haha [22:16] the fast export file is 20 gigs >.> [22:16] AuroraBorealis: I'd really like if you tried a 64-bit bzr if you're okay with building python things yourself [22:16] yeah i can try that [22:16] not sure how i do that on windows [22:17] you need a 64 bit python and the compiler it built with for starts... [22:17] wgz: ah, you're Martin! [22:18] which compiler is that? [22:18] wgz: What made your "m" go upside down? [22:18] weekend! [22:18] ...I didn't quite stick to irc hours discipline [22:18] yeah, it's hard on all of us :) [22:20] AuroraBorealis: let me check, I'm slightly concerned that the old free beer MSVCs were 32 bit only [22:20] i *should* have access to all of the microsoft visual studio stuff, just need to ask my school [22:21] ..I've typed "54 bit" like, three times now [22:22] okay, this is the first thing I've turned up on google that looks even vaguely useful: [22:22] as with all such things, some of what it says is probably out of date [22:23] ive heard that vs2008 is the correct one too [22:23] this no doubt still stands though "Visual C++ express works, as long as it’s 2008." [22:23] so, you should be able to get it to work. [22:24] ok [22:24] so then, i just download bzr from source [22:24] and then what do i compile? [22:25] hm, jelmer, do you know if the C-from-pyrex sources in the tarball can be used straight up? [22:26] wgz: I think so; the ugliness of the generated code is in part due to it trying to support all possible compilers [22:26] if so, you should just then be able to do `...Python2.7/python.exe setup.py` for Bazaar... then you'll need to do a similar thing for the two fastimport packages [22:27] and meliae... anything else? [22:27] *`python setup.py install` [22:28] ask in here if you get stuck at any point or need more help. [22:28] haha [22:28] which fast import packages do i need? [22:28] or are those included in the bazaar source [22:29] lp:bzr-fastimport and lp:python-fastimport [22:29] k, and do i need VS2008? [22:30] yup. [22:30] so what exactly do i build with vs2008 [22:30] i know i run setup.py for most everything [22:31] setup.py will call the compiler for you, for the modules that need compiling [22:31] (*should, see that blog post if it doesn't work) [22:32] yay! [22:33] * AuroraBorealis starts the download [22:56] wgz: mgz by another nick? [22:59] lifeless: otherwise I'd forget to logout and be ghosting come Monday all the time [22:59] wgz: welcome to Canonical! [23:00] i can't seem to branch lp:bzr [23:00] it just sits there at "starting" =/ [23:00] AuroraBorealis: I suggest you get the tarball, will save you needing cythong [23:00] AuroraBorealis: are you behind a proxy perhaps ? [23:00] i'm not [23:01] lifeless: and welcome to not being able to stay logged out over the weekend? :) [23:01] wgz: working on something we're passionate about, from home, tends to lead to blurred lines ;) [23:01] [23:01] you have already right? [23:02] yeah [23:02] wgz: for instance, I've just triaged a hundred or so stale LP bugs [23:03] lifeless - he's passionate about stale bugs [23:03] they eat funny things in nz... [23:03] bzr seems to be trying to open the repo with subversion or something [23:03] wgz: :P I'm passionate about LP [23:03] :P [23:04] !pastebin the log quickly AuroraBorealis? [23:04] wgz: I am only a bot, please don't think I'm intelligent :) [23:04] !pastebin [23:04] For posting multi-line texts into the channel, please use http://paste.ubuntu.com | To post !screenshots use http://imagebin.org/?page=add | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic. [23:06] this is unrelated to the memory error thing, its just not branching lp:bzr and creating a repo has weird output o.o http://paste.ubuntu.com/700769/ [23:06] yup. [23:06] ugh, yeah, that's not pretty. [23:07] no idea why its doing it, i'm just creating a new repo >.> [23:08] it also seems to be denying my public key [23:11] back to what i was doing, its saying that no cython or pyrex is installed. do i need those? [23:12] shouldn't say that if you went with the tarball? [23:12] otherwise yeah, you'll need one [23:12] yeah its with the tarball [23:13] just did python27_64 setup.py and its saying "no cython, trying pyrex...pyrex is not available" [23:13] http://www.lfd.uci.edu/~gohlke/pythonlibs/#cython [23:13] did it then fail? [23:13] * wgz checks the setup.py logic [23:14] i was just checking the arguments [23:14] it said that then printed the usage [23:14] ah, it looks like it'll say that [23:14] but then build anyway [23:14] so, don't worry [23:14] now its saying cython is there [23:14] so yay [23:15] booooo [23:15] "valueerror" [23:15] Obviously, it thinks you have questionable values. [23:16] pastebin traceback? [23:16] did you get the windows 7 sdk? [23:16] http://paste.ubuntu.com/700774/ [23:16] no i didn't [23:17] * AuroraBorealis installs that [23:17] is what you want I think [23:17] ^that error might be one of the ones the blog post from earlier had a workaround for [23:18] yeah i'm installing it now [23:18] see the "Trick distutils" section [23:18] which comes after "Install the Windows 7 Platform SDK" [23:19] * wgz hugs fullermd and his underscore [23:19] annnnd installation failed on the sdk [23:20] * AuroraBorealis cries [23:20] * wgz offers emotional support [23:21] the error was "see the config html page for more information" [23:21] thanks microsfot [23:21] soft* [23:25] did it dump a html log anywhere obvious? [23:25] i have the log [23:25] no idea what its actually doing [23:26] http://paste.ubuntu.com/700777/ [23:26] Aiee! There's an underscore on my back?? Get it off! Get it off!! [23:27] * wgz swats at it [23:27] * fullermd_ spins around barking and biting as his tail. === fullermd_ is now known as fullermd [23:27] found something promising on google, one se [23:27] c [23:28] http://blogs.msdn.com/b/windowssdk/archive/2009/09/16/windows-7-sdk-setup-common-installation-issues-and-fixes.aspx [23:28] the list of errors there is amusing [23:28] the fix is diving deeep into the registry [23:34] and now i apparently need to reboot