[01:33] so, for some reason the 'git' git repo won't fast-import into bzr, but other smaller ones will [01:36] mgrandi, there have been bugs in the past where git wouldn't import because it has strange historical artifacts [01:36] due to buggy versions of git being used on it [01:36] bzr has similar things [01:36] i thought it was fixed [01:36] so you could file a bug [01:36] yeah [01:36] if there's not already one [01:36] i'll look, its pretty vague [01:36] just a 'keyerror' [01:46] gonna see if i can trace it out~ === bradm_ is now known as bradm [03:30] so it seems that the reason the fastimport is failing is because the commit a certain tag is referencing is not there... [03:38] so, not sure what bzr is supposed to do in that circumstance [07:56] hi all [07:58] hi there vila [07:59] poolie: hey ! [07:59] hello [08:00] morning all! [08:01] hi there [08:01] let's hang out [08:17] :-( === jelmer_ is now known as jelmer === jml` is now known as jml [11:00] * mgrandi is awake [11:06] fastimport is hard. [11:06] let's go shopping? [11:07] heh [11:08] mgrandi: what about it, specifically? [11:08] one sec , let me play the 'which tab is playing music' game [11:10] well, i tried importing the 'git' git repo again, and it failed at the end because of a key error [11:10] http://bpaste.net/show/25513/ [11:10] i made it drop into pdb, it seems that there is no 'mark' for '2', so im guessing that means a tag is referring to a commit that no longer exists? [11:11] ...I was just going to suggest BZR_PDB [11:12] well it happened at a specific spot so i just made it drop into pdb when it caught a keyerror exception [11:12] and you already did it. you're almost and expert now :) [11:12] the "command" its on, is this: http://bpaste.net/show/25520/ [11:13] yay! [11:13] okay, running my cache summary script on your dump from last night [11:14] ...MemoryError [11:14] wat [11:14] you ran into a memory error running the script? [11:14] I probably just ran out of memory :) [11:15] you should post that script, it sounds useful :o [11:16] it's a bit of a hack from before I had a neater way of doing it, but will put up for you to have a go [11:18] http://float.endofinternet.org/temp/cache_summary.py [11:19] will try it on this box too... after building meliae [11:21] ...keyerror [11:22] fun never ends! [11:22] although key error was what i was getting hwen i let it finish, maybe related? [11:23] also do you have a link to my dump? i cant login to my ftp server for some reason [11:23] no, just an error in how the script pokes around with the objects [11:23] I had a better way of doing some of this... [11:24] mgrandi: http://float.endofinternet.org/temp/dump.json.gz [11:36] okay, three bugs in short succession, should just fix meliae first [11:37] 3 bugs you just found in meliae? [11:37] well, I probably already filed them [11:42] hmm [11:43] bug 871450 is particularly annoying but probably easy to work around, every attribute has '"' stuck on the end [11:43] Launchpad bug 871450 in Meliae "Test failures from incorrect json string escaping" [Undecided,New] https://launchpad.net/bugs/871450 [11:48] hmm [11:49] any easy fix we can do right now? [11:50] yeah, fiddle with the json import which there's also a bug for [11:53] so it just not doing it right because simplejson is not there and therefore its not using anything to encode json? [11:57] right. [11:57] okay, let's try this update... [12:02] doesn't python have a json api? [12:02] built in? [12:03] indeed it does these days, and based on the module that meliae is using [12:03] so a little import dance is enough [12:03] bug 871451 [12:03] Launchpad bug 871451 in Meliae "Use json module from stdlib as simplejson fallback" [Undecided,New] https://launchpad.net/bugs/871451 [12:04] they both use the same api, so i can just say simplejson=json ? [12:04] import json as simplejson? [12:11] well, the reverse makes more sense, but that's what I did, yeah [12:12] now just working out some other changes [12:12] ok [12:12] IndexError: _LRUNode(4423523216 80B 5refs 2par) has only 5 (not 5) references [12:12] funny. [12:12] o.o [12:14] it's just the error forgetting to compensate for 0-based indexing :) [12:16] hehe [12:18] how are you supposed to run the unit tests? i can never remember the one liner to do that [12:18] for meliae (and most other python projects) [12:19] okay, this looks like it mostly works now, just the id type issue left... [12:20] mgrandi: usually something like 'make check' or 'trial meliae' [12:20] mgrandi: there's a script called run_tests.py in the root dir :) [12:21] yeah, but those don't get copied to the site-packages dir =P [12:21] or don't get built, so i was wondering if i just had to copy them and then run them or something easier [12:30] wgz: you're still having weekend ? :-P [12:30] vila, wgz: I wonder if I could tempt one of you into doing some reviews some time this week [12:31] I was about to say I would code for reviews, but that seems kindof pointless... [12:31] jelmer: :) [12:31] jelmer: will do my best [12:35] mgrandi: http://paste.ubuntu.com/892105/ [12:35] and the script in my temp folder is updated to produce that [12:36] so, there's 100mb just in decompressed string caches [12:36] did you fix up meliae in the process? [12:36] some of it, I worked around a few things too [12:36] and i remember python using around 500 mb total when i stopped it [12:36] so 1/5th isn't good [12:41] and what you told me about the meliae test suite, this is what i was looking for =P Oct 02 04:57:14 can do that with... `cd meliae` `Python27\python setup.py build_ext -i` `Python27\python -m unittest meliae.tests.test_suite` [12:43] but now i sleep for another hour === mgrandi is now known as mgrandi|away === Pikkachu is now known as PikkachuAway === PikkachuAway is now known as Pikkachu [15:12] vila vila [15:12] something is burning ? [15:12] when you froze at the end of the hang out this morning [15:13] I was about to ask if you would help me write some test case things [15:13] oh, hopefully ;) [15:13] if I commit what I have, can you look over it and make suggestions on how to solve my problems? [15:13] sure [15:13] okay, will split this file up into several bits [15:36] hey, I have a few patches for pidgin I need to maintain, and I wonder what's a good way to manage them [15:37] people from #ubuntu-motu said it's hard [15:37] I'm slightly thinking of something with Bazaar branches though, anyone has some experience? [15:47] Pikkachu: the best way is to submit them upstream ;) [15:47] Failing that, you should probably do it the debian/ubuntu way and even better the udd way [15:48] googling for udd should give you the pointers to the existing docs [16:01] thanks for trying [17:30] hello [17:33] wello [17:34] so [17:35] there are like 5 bugs related to fast import that are preventing us from actually solving said memory problem [17:35] well, it doesn't help, but it's not a complete stop [17:35] have you filed all of 'em yet? [17:35] well [17:35] one was the json thing which i think you said was filed [17:35] and might of fixed when i was in class/sleeping haha [17:36] the timezone one is filed, the memory one is filed, and im not 100% sure what is wrong with the one i ran into yesterday [17:37] from what i can tell from the git fast-import man page and the fast-import code, this bug: http://bpaste.net/show/25508/ is caused by a tag referring to a "mark", ( a commit i think) not existing anymore [17:38] i want to make sure thats right before i go reporting it [17:39] seems possible [17:39] ideally you'd just make a minimal repo that had the problem [17:39] but what you've got is enough for a reasonable [17:39] ...bug report [17:39] right, just going into the garden briefly [17:40] thing is i dunno how to create a repo with that problem [17:40] or rather, know enough about git [17:41] that was easier than it could have been [17:41] had not buried himself anywhere obscure [17:41] like, how do you just delete a commit? ovo [17:42] without deleting the tag that references it [17:44] or, this could like the case with the invalid timezone, that this is a git bug and fastimport just needs to be tolerant of it === r0bby_ is now known as robbyoconnor [18:00] so i have no idea if you know much about the fast import format, i think jelmer does though [18:01] I'm reasonably familiar with it [18:01] it's all relative [18:04] what is? [18:04] how much somebody knows about fastimport [18:04] ah [18:04] well i just meant [18:05] i dunno if you saw the stack trace i pasted above, but basically its just saying that a mark (2) is not found in the dict of marks [18:05] marks -> bzr revision ids [18:05] im assuming that means that bzr did not find a revision with mark 2, and thats why its not there -> why it throws a key error [18:06] I haven't looked at the traceback, but that sounds plausible.. [18:06] http://bpaste.net/show/25508/ if you didnt see it [18:08] my question is before i look silly on the bug report, is can you do this in git? just remove commits? or is this just another possible bug with git that fastimport needs to be tolerant of === yofel_ is now known as yofel [22:29] hi all [22:29] 'mÃ¸rning poolie [22:35] UnicodeDecodeError [22:37] wgz.decode('wgz') [22:52] hi jelmer, wgz [22:52] :) [22:56] poolie, am I understanding google plus correctly in that it's actually impossible to do any fixed-width/code formatting? [22:57] so, it's not really possible to do a post with snippets or diffs or console output such like? [22:57] i think that's correct