/srv/irclogs.ubuntu.com/2008/10/19/#bzr.txt

zacheraok00:00
zacheraimported00:00
zacherawhats next00:00
zachera:D00:00
zacheracreate /aristaeus in /var/subs/trac ?00:00
LarstiQzachera: point trac at the svn repo you just created, yes00:01
zacherasweet00:02
zacherahang on00:02
LarstiQzachera: either use the trac-admin tool, or set 'repository_dir' under [trac] to the correct location00:02
zacherait worked00:02
LarstiQgood00:03
LarstiQzachera: so, solved now?00:03
zacheranot quite00:03
zacherathe website style isnt showing up00:03
zacheralol00:03
LarstiQzachera: may I ask btw why you came here for help? Since this doesn't involve bzr at all.00:03
zacherabecause i installed bazaar in the first place00:03
LarstiQhow did you get the idea to do that? Assembla is svn only afaik?00:04
zacheraIt's simple.00:07
zacheraI saw the words "Bazaar" and "Version Control" and installed it. :P00:07
LarstiQAha :P00:07
zacheraIT WORKEDDDDDDD00:08
zacherahahaha00:08
zachera:D00:08
LarstiQgood00:09
fynnWhat's the canonic (pun not intended) way to perform a message-less commit?01:33
=== Verterok|out is now known as Verterok
=== Spaz is now known as spaz
=== spaz is now known as Spaz
* fullermd watches launchpad ladle out a tarball at <3k/s...06:37
=== fta_ is now known as fta
Odd_Blokejelmer: Are there any instructions on using the Debian bzr packaging team's branches available?15:07
Odd_Blokejelmer: I just hit http://paste.pocoo.org/show/88477/.  Any thoughts?15:30
=== jfroy is now known as jfroy|notworking
bazaarhi, is it possible to find out the revision a specific file was added?18:21
Jc2kbazaar: you should be able to do bzr log filename to see all the revisions  that touched a given file, and hence you should be able to see the one that added it...18:23
Jc2k(i think)18:23
bazaari'll try that.18:23
bazaarworked. thanks.18:23
LarstiQyes, and in most cases it will be the first one, so --limit 1 will work too18:23
Jc2koh, i didnt realise it was oldest first or i would have suggested that ;)18:24
LarstiQJc2k: switchable with --forward18:35
fynnHey. What's the canonical way to perform a message-less commit?18:41
fynn(A commit with no commit message.)18:41
LarstiQfynn: that's normally guarded against18:44
LarstiQI don't think you can get around that without resorting to using bzrlib18:44
fynnLarstiQ: yeah, I'm asking because some developers asked for this feature.18:45
* LarstiQ isn't in the position to check the code right now18:45
fynnthat's OK, I might do that myself. it's Python after all :)18:45
fynnit's just that some people asked for it, and I didn't find any way to do it, so I asked to make sure I didn't miss anything.18:45
LarstiQfynn: hmkay, it seems like a very bad idea to me :) But yes, a custom plugin shouldn't be too hard.18:45
* LarstiQ nods18:46
LarstiQfynn: be aware that the commit message is a property of a revision, you can not change it afterwards.18:46
LarstiQfynn: so if they want to punt till later, they can't.18:46
fynnit's just that some developers work on semi-private branches, and they make a lot of personal commits18:46
* LarstiQ just wildly guesses as to why they'd want that.18:46
LarstiQfynn: what happens with those personal commits later on?18:47
fynnwell, they get merged.18:47
fynnbut I'm thinking about that article someone gave here yesterday18:47
LarstiQok, so they just want to describe the entire package of work during merge time?18:47
fynnabout Loom18:47
fynnyeah, exactly.18:47
LarstiQright, I can see that.18:48
fynnshould I take a look at Loom?18:48
fynnor is there a builtin way of doing that?18:48
LarstiQfynn: do what exactly? Merging is a core bzr capability.18:49
fynnLarstiQ: hm, OK, I'll take a better looks at what _can't_ be done via vanilla merging, before I assume I need Loom :)18:51
fynnalthough Loom does look very interesting: http://bazaar.launchpad.net/~bzr-loom-devs/bzr-loom/trunk/annotate/head:/HOWTO18:51
LarstiQyes, that it is :)18:53
LarstiQbut I haven't gotten further than "cool!", haven't actually used it myself yet18:53
fynnLarstiQ: is there a way to create an alias that would run a shell command?19:07
fynnlike the ! aliases in git?19:07
LarstiQfynn: bzr aliases are restricted to bzr afaik. Do you have an example of such a git alias?19:24
fynnLarstiQ: sure:19:27
fynn git config --global alias.ci '!git commit -m "`date`"'19:27
fynnthis causes `git ci` to commit the current changes with a message equal to the output of DATE(1)19:29
LarstiQfynn: right, I see.19:29
fynnas far as I can see, the easiest way of doing that on bzr would be to write a plugin that would provide that command.19:48
arjhi, I tried upgrading remove sftp tree but it failed and now I have a backup.bzr dir19:58
arjhow do I remove that and try again?19:58
Flare183When I try to push one of my infobot projects I get this error: bzr: ERROR: Invalid http response for http://bazaar.launchpad.net/%7Erichardson183/ubuntu-bots/flarebot/.bzr/branch-format: Unable to handle http code 502: Bad Gateway19:59
Flare183Why do I get this error?19:59
Flare183dres: Can you help me?20:01
LarstiQfynn: yes, though extending current alias support with ! shouldn't be too hard20:03
fynnLarstiQ: yeah, probably :)20:09
muntyan_hi guys. is it easy to have multiple branches inside one repository, and have those branches synchronized between multiple repositories?21:40
muntyan_to achieve something like svn but without a real central server21:40
lifelessmuntyan_: yes21:51
muntyan_lifeless: any docs about it? i was googling around, and just found that what i want is actually impossible21:52
muntyan_some stuff about shared repositories and whatnot21:52
LeoNerdHi guys.. I have a slightly odd question. I'm looking at wrapping bzrlib in perl for some perl code.. I've found Inline::Python which basically seems to Just Work. Just with one problem.22:31
LeoNerdGiven an object reference to a python object, I can call methods on it, but not access bare attributes. Some of the bzrlib stuff needs these, in places. Anyone any ideas around it?22:32
lifelessLeoNerd: teach Inline::Python about __getattr__22:47
LeoNerdOh.. is that a method on python objects anyway though?22:47
LeoNerdCould I  $object->__getattr__("attrname")  ?22:47
lifelessLeoNerd: (there is no such thing as bare attribute access - foo.bar is getattr(foo, "bar") under the hood22:47
lifelessand getattr calls into foo.__getattr__("bar") - using getattr(foo, "bar") is the standard though22:48
LeoNerdYes.. but that form I can do for free22:48
LeoNerdWrapping getattr() as a plain perl function is More Work22:49
lifelesswell, up to you22:49
lifelessprobably just doing foo.__getattr__("bar") will work22:49
spivActually, getattr calls __getattribute__.22:50
lifelessah22:50
lifelessthanks :)22:50
LeoNerdOK. But that's just a plain object method on the object, right?22:50
spivA pure-python translation of getattr (and type.__getattribute__) is surprisingly messy! :)22:50
LeoNerdso e.g. at the perl level, I could   my $repository = $revision->__getattribute__("repository");  and it would work?22:51
lifelessLeoNerd: its on the class I think22:51
spivLeoNerd: If you can make Inline::Python use the getattr builtin function, that would be best I think.22:51
LeoNerdspiv: That requires Much Voodoo22:51
LeoNerdAnyone who works on Inline::Python has to know a lot about both perl and python's C-level representations22:51
LeoNerdThere's not many who do22:51
spivLeoNerd: well, I mean, use its API to invoke getattr, rather than the (presumably hard) task of fixing Inline::Python.22:52
lifelessLeoNerd: another thing you could do is write a small python helper module22:52
LeoNerdlifeless: I've pondered that.22:52
LeoNerdI also pondered just adding a 'get_repository' method to Revision22:52
LeoNerdSo I can call it :)22:52
LeoNerdspiv: Yes, but that means I have to export getattr. It seems Inline::Python doesn't let you just plain export that function; I'd have to write my own which calls it, that it then exports for me.22:53
LeoNerdSeems messy that way22:53
spivLeoNerd: that seems like a large gap in Inline::Python!22:57
LeoNerdYes. A horrendous one22:58
LeoNerdbut as I said, very narrow field of people who could fix it22:58
spivLeoNerd: A quick glance at the docs I found with google suggest you could do it with py_call_function22:58
LeoNerdIt requires someone who is an expert in both perl and python; who knows a lot about the sorts of things programs in each language would do, _and_ who knows about the C-level representations internally.22:58
LeoNerdGiven enough hackery, I could justabout make such objects respond to hash fetch representations... i.e.  $obj->{key}22:59
LeoNerdBut knowing what to call into python to do that; no idea22:59
spivLeoNerd: the "package" arg would need to be '__builtins__'23:00
LeoNerdspiv: Ohsure.. there's a number of hackish ways to do it.. I've got a list of four or so candidates.23:00
LeoNerdI'd like to know the best of these, or how to do a Real Solution. but as I said, a real one probably involves C-level hackery23:00
spivLeoNerd: it doesn't seem so hard to make a perl helper called py_getattr(obj, attr_name) than invokes getattr(obj, attr_name) then?23:01
LeoNerdNot at all..23:02
LeoNerd$object->__getattribute__("foo")   seems already to do it though.23:02
LeoNerdsub getattr { some perl level hackery };   then  getattr($object, "foo")  being another23:02
=== TheMuso_ is now known as TheMuso
LeoNerdA third would be to override specific cases; so     class bzrlib.Revision: def get_repository(self): return self.repository;23:03
LeoNerdThen I can  $revision->get_repository()23:03
LeoNerdThe first two of these require much annoyance at every single callsite.23:03
LeoNerdThe latter requires adjustment of every one of bzrlib's API points that I want to call.23:03
pooliethis is silly, let's just talk here23:08
spivpoolie: Hi :)23:10
pooliehm it's a bit disturbing that in the month 8.10 is meant to release neither of my machines will boot reliably :/23:10
lifelessso23:10
poolieso lifeless i saw your mail, and i have the numbers23:10
lifelessI finished the conversion by hand and sent stats23:10
lifelessI haven't altered the cron job in any dimension23:10
pooliei was wondering if i should restart the cron job which seems to be blocked atm23:10
lifelessso if it was left enabled it should be continuing23:10
pooliemaybe because the lock file wasn't removed?23:10
lifelesscould be23:10
lifelessthere should be disk space to run23:11
lifelessnote that ~/run can get kinda full easily :>23:11
pooliei'll check on that too23:12
poolieso for myself today i was going to do more packaging of 1.8, and see about automating packaging of bzrtools and maybe other plugins23:12
poolieand work on the review queue23:12
lifelessanyhow, when I get back to this, I think we have solid progress - a significant drop in uncompressed data size23:12
lifelessLeoNerd: the silly poolie refers to is a voip epic failure.23:13
poolieyes23:13
poolieInline::Python may also be silly ;-)23:14
LeoNerdIt's... not bad.23:14
LeoNerdIt's quite cute apart from this annoyance on attrs23:14
poolielifeless: so today was going to be sacrificed to administration demons?23:16
lifelessyes23:17
spivOn Friday I did a couple of reviews, and also some improvements to HPSS client-side error translation (removing duplication and bizarreness in bzrlib/transport/remote.py especially).23:18
pooliecool23:20
poolieit is a bit bizarre still23:20
poolieit would be nice to have less of the return values handled per-rpc23:20
pooliewhat's on the menu today?23:21
spivRight.  That's actually caused at least one bug in bzrlib/remote.py, in RemoteBzrDir.__init__.  So next on my hit list is fixing that bug by refactoring that file so that the "try: self._client.call(...) except ErrorFromSmartServer, err: self._translate_error(err, ...)" blocks are repeated everywhere.23:22
spivEr,23:22
spiv*aren't* :)23:22
spiv(The flaw in RemoteBzrDir.__init__ is that it does a HPSS call without catching ErrorFromSmartServer)23:23
pooliewoo23:23
spivI think I should do more reviews today as well, the queue is pretty large.23:23
poolieit is23:24
pooliei should too23:24
spivMuch of that is Aaron's shelving stuff which I'm looking forward to having merged, but I'm not sure I know an awful lot about the relevant code.23:24
pooliemaybe lifeless can when he's done with paperwork23:24
poolieso i think a review from an interested (relative) layman is better than no review at all23:25
pooliemaybe just asking questions etc23:25
spivOk.23:25
poolieok, we can sit down now23:26
pooliehave a good day :)23:26
spivpoolie: thinking of administrivia23:31
pooliemm?23:31
spivpoolie: the mailing list tagging regexes still need some tweaks23:31
pooliedo you have any expenses etc outstanding yourself?23:31
poolieare you referring to the utf8 thing?23:31
poolieit's probably technically possible to write a regexp for it23:32
spivpoolie: the [Success!] mails (and also my [bzr-usertest:MERGE] mails) got tagged everything else23:32
poolieah, ok23:32
spivThe utf8 thing just plain sucks, and needs mailman support to fix; in theory mailman could decode the unicode I think and run the regex on that :)23:33
pooliemm23:33
pooliethough arguably you might also want to examine the uninterpreted form23:34
spivThe [Success!] one is probably a straightforward tweak to fix.  For things like [bzr-usertest:MEGE] I guess we want to settle on a convention for "please review this code, but don't bother bundlebuggy with it"23:34
pooliespiv, i thought bb left the [merge] behind in the success mails?23:34
pooliei guess not23:34
pooliedo you want to change this?23:34
spivIt does, but somehow they are still mistageed.23:34
spivWell, I got mail from someone saying "I didn't want to receive those two patches"23:35
poolieactually what i meant was, will you fix this or should i?23:35
spivOh, I see.  I don't know that I have the rights to fix it?23:35
spivI didn't realise there was a possibility that I might :)23:36
spivpoolie: ok, I can't think of an improvement to the regexes, I've replied on the list to the complaint I got.23:47

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!