[09:05] <mgz> morning!
[09:05] <mira|AO> moien
[09:48] <lolek> hello all
[09:48] <mgrandi> hello
[09:49] <lolek> i'm looking for some documentation about, how to supply username aand login for bzr-svn ?
[09:49] <mgrandi> to access a svn server?
[09:49] <lolek> yes
[09:50] <mgrandi> im not an expert on bzr-svn, im not sure if you can do that
[09:50] <mgrandi> but i may be wrong
[09:50] <lolek> well, the problem is that from the error stack trace it seems that bzr is using my system login, and system password
[09:51] <lolek> so there must be some way to use another one
[09:51] <mgrandi> hmm.
[09:52] <mgrandi> http://stackoverflow.com/questions/3166189/how-to-save-subversion-password-with-bzr-svn
[09:53] <mgrandi> maybe authentication.conf?
[09:53] <lolek> ok
[09:53] <lolek> next question, where authentication.conf should be placed cause i don't have that file in: ~/.bazaar
[09:54] <mgrandi> pretty sure in ~/.bazaar
[09:55] <mgrandi> http://doc.bazaar.canonical.com/developers/authentication-ring.html#file-format
[09:58] <lolek> ok and how can i force bzr to use that file, cause it's ... not using it :/
[09:59] <mgrandi> well it should, are you sure the host is the same in the conf and the host you are connecting to?
[09:59] <lolek> yes
[10:00] <mgrandi> are yo using the latest version of bzr-svn?
[10:01] <lolek> Version: 1.2.1-1
[10:02] <lolek> ok, that's weird
[10:02] <lolek> with qbzr it's asking me for a password, but still using the wrong login :/
[10:02] <mgrandi> hmm, i feel like its something silly but im not the expert
[10:02] <mgrandi> jelmer, are you here :o
[10:04] <mira|AO> it's easier to just use ssh keys
[10:04] <lolek> yes i agree, but i don't have that option
[10:05] <lolek> i can only use svn providing login and password
[10:05] <lolek> nothing more
[10:05] <mira|AO> the username's normally in the URL
[10:05] <mira|AO> for svn
[10:05] <mgrandi> bzr branch <scheme>://<user>:<password>@host:port/path
[10:06] <lolek> ok, that's working
[10:06] <lolek> but i'd like to stre it's in authentication.conf
[10:06] <lolek> it should be working
[10:06] <mgrandi> yeah, not sure why its not working with that file
[11:38] <LarstiQ> lolek: if you login with plain 'svn', svn should cache that and bzr-svn will pick up on it, iirc
[11:39] <lolek> LarstiQ: hmm ok, i've read about that, but the problem is that i don't use plain svn login, i do login only from inside eclipse with svn plugin
[11:43] <LarstiQ> lolek: you could try logging in once, as a workaround?
[11:43] <LarstiQ> using something like `svn info` to not do actual work
[11:43] <LarstiQ> or much of it anyway
[11:43] <lolek> hmm
[11:48] <lolek> LarstiQ: ok, i've tried
[11:49] <lolek> and it seems that it gets login from cache
[11:49] <lolek> but it have some problem with password
[11:51] <lolek> LarstiQ: look here: http://pastebin.com/e2pJb3qR
[11:51] <lolek> password contains only ascii characters
[11:54] <LarstiQ> lolek: heuh
[11:54] <lolek> :D
[11:54] <lolek> i knew you'd love this :D
[12:00] <LarstiQ> lolek: next thing I'd do is maybe check bzr/bzr-svn versions and/or see if there any similar bugs on launchpad
[12:00]  * LarstiQ prepares for his Galois Theory course
[12:03] <mgrandi> its not necessarily because you have invalid characters, its just a python 8 bit string vs a unicode string
[12:05] <mira|AO> oh yes, python, the only modern language with a default 7-bit safe string type
[12:05] <mira|AO> and py3k doesn't improve on it
[12:05] <mgrandi> python3 strings are unicode
[12:05] <mgrandi> by default
[12:05] <mira|AO> that doesn't mean you can print them :D
[12:05] <mgrandi> thats not python's fault
[12:06] <mira|AO> that's a matter of opinion
[12:06] <mgrandi> i cant print weird strings on windows cause the console is CP1257 or whatever
[12:06] <mgrandi> and you can't convert it
[12:06] <mgrandi> how would you print the character if the character encoding its outputting doens't have said character?
[12:06] <mira|AO> by defaulting to a reasonable encoding
[12:07] <mira|AO> (even on windows, since 1991 the console can do Unicode)
[12:07] <mgrandi> windows doesn't have a 'reasonable' encoding haha
[12:07] <mgrandi> can it? ive had this problem on other languages
[12:07] <mgrandi> what encoding type is that
[12:07] <mira|AO> yeah, just use the API with a W on the end
[12:07] <mira|AO> UTF-2LE
[12:07] <mira|AO> erm no
[12:07] <mira|AO> UCS-2LE
[12:07] <mgrandi> i mean the windows..encoding
[12:07] <mira|AO> so, just the BMP, normally
[12:08] <mgrandi> or what codepage is it
[12:08] <mira|AO> you've got two APIS with, in the case of the console, three encodings
[12:08] <mgrandi> thats what the console uses
[12:08] <mira|AO> one is the API with the A at the end, which uses the "ANSI" encoding (often cp1252)
[12:08] <mira|AO> then there's the OEM encoding (often cp437 or cp850)
[12:08] <mira|AO> then there's the API with the W at the end, which uses UCS-2LE
[12:08] <mira|AO> codepages map between A and W
[12:09] <lolek> ok guys guys guys
[12:09] <lolek> there are no character outside the standard ascii
[12:09] <lolek> in my password
[12:09] <lolek> so this is weird why it's complaining about.. unicode
[12:10] <mgrandi> because the object itself is not a unicode object
[12:10] <mgrandi> aka a bug or something else is going wrong
[12:10] <mira|AO>    I set up the cmd window to use Lucida Console font (and you should do it too, otherwise any attempt
[12:10] <mira|AO>    to see Unicode characters in it is bound to fail!). I realized that it is possible to print wide
[12:10] <mira|AO>    strings directly to the console using functions from conio.h (_cputts, _tcprintf, etc.). Very nice!
[12:10] <mira|AO> like that, it can work
[12:11] <mgrandi> ive tried that but it still doesn't work
[12:11] <mgrandi> and python should see the default code page and adjust to it
[12:11] <mgrandi> what code page do you change it to?
[12:12] <mgrandi> never mind, apparently it doesn't matter? (windows encoding sounds like a cluster****)
[12:13] <mira|AO> WriteConsoleW also would work
[12:15] <mgrandi> that does seem to work
[12:15] <mgrandi> but i dont think that its still python's problem, since you fix it when you change fonts on cmd.exe, nothing on python changes
[12:16] <mira|AO> AUF DROGEN ODER WAS?
[12:16] <mira|AO> oops
[12:17] <mira|AO> sorry, just found some… disgusting code
[12:17] <mira|AO> and forgot I had /query on
[12:17] <mgrandi> like, in "terminal" font, the same python session, gives me
[12:17] <mgrandi> 'garbled characters
[12:22] <w7z> lolek: the get_svn_simple callsite in bzr-svn is misusing the bzr ui, by passing a str not a unicode type
[12:23] <w7z> but the point is you don't want it prompting for a password right? I suspect you have your creds wrong so it's falling back to asking you again.
[12:27] <w7z> lolek: `bzr version` will tell you where to put authentication.conf and check .bzr.log after connecting to see if it works
[12:27] <w7z> I'd avoid putting the username and password in the url
[12:32] <lolek> hmm
[12:37] <lolek> w7z: well i see in the log that it's using the proper login... but i don't see anywhere in bzr.log that it's reading authentication.conf
[12:37] <w7z> ...does it say it connects and does stuff? if so, problem solved, right?
[12:38] <lolek> it's : Obtaining username and password for SVN connection (see only login) and then the error: http://pastebin.com/e2pJb3qR
[12:38] <w7z> okay, so it's not getting the values from the conf file then.
[12:39] <lolek> uhm
[12:39] <lolek> but the question is... why ?
[12:39] <lolek> of course if i put login/password into url it's working ok
[12:40] <w7z> paste your conf with the password xxxed out?
[12:40] <lolek> k
[12:41] <lolek> http://pastebin.com/MxwVkTDi
[12:45] <w7z> looks okay, maybe bzr-svn just doesn't check
[12:46] <w7z> file a bug against bzr-svn with your traceback and I'll put up a branch that fixes it
[12:46]  * mira|AO eyes “branch that fixes”
[12:47] <w7z> jelmer or vila might know more about svn/auth specifics
[12:48] <lolek> ok, where do you want me to fill a bug ?
[12:49] <w7z> bugs.launchpad.net/bzr-svn
[12:51] <lolek> w7z: https://bugs.launchpad.net/bzr-svn/+bug/452121
[12:51] <lolek> hmm
[12:52] <lolek> https://bugs.launchpad.net/ubuntu/+source/bzr-svn/+bug/532292 <- similar
[12:52] <lolek> lol
[12:52] <lolek> maybe i should wait for new bzr-svn version ?
[12:52] <w7z> no, different bug
[12:52] <w7z> but, you should be able to put the details in the config file and have it work
[12:53] <w7z> the prompt should not happen at all if it can get the values from your config
[12:56] <w7z> do you only get prompted for the password, or did you type something else in first?
[13:05] <vila> w7z: sorry, my memory is fuzzy on that topic. The most common answer I remember is: connect with svn first as it will cache the credentials and bzr-svn can then rely on that
[13:05] <vila> i.e. auth.conf is not used
[13:07] <lolek> w7z: erm... i don't have any prompt
[13:07] <lolek> w7z: i just do: bzr update and i get this error
[13:07] <lolek> no prompt for password
[13:10]  * jelmer waves
[13:10] <vila> I'm not sure 'svn' is a valid scheme though (jelmer ?)
[13:10] <vila> (for auth.conf that is)
[13:11] <jelmer> I don't think I've ever seen (or tested) authentication over svn://
[13:12] <jelmer> I don't see why it shouldn't work, but I don't think I've ever tested it.
[13:12] <jelmer> lolek: ^
[13:13] <lolek> lol :)
[13:13] <lolek> that would explain the problem :)
[13:14] <jelmer> lolek: It shouldn't really be different from http/https though, and those *do* work
[13:14] <jelmer> I aguely recall there were some issues with the authentication interface being inconsistent about whether it expected unicode or byte strings, perhaps that's related.
[13:14] <jelmer> *vaguely
[13:14] <jelmer> what versions of bzr/bzr-svn are these?
[13:20] <lolek> bzr: Version: 2.5.0-2ubuntu2
[13:20] <lolek> bzr-svn: Version: 1.2.1-1
[13:54] <w7z> lolek: okay, can you try a few debug things for me?
[13:54] <lolek> w7z: i'll do my best
[13:55] <w7z> #1: with your authentication.conf in place, run `bzr -Dauth branch svn://YOURDETAILS` and look in .bzr.log for a "Using authentication section" line
[13:58] <lolek> ok,
[13:59] <lolek> the is not such: Using authentication section in the bzr.log
[14:00] <w7z> okay, so that's straight up not being matched
[14:01] <w7z> so, either there's a bug in how bzr-svn uses AuthenticationConfig, or you have the conf file in the wrong place, or the details don't match up
[14:02] <w7z> try adding a section that's just [something] with a username and password set underneath, then branch any (non svn) url with that -Dauth flag and see if that message then appears
[14:03] <w7z> without scheme/host/port/path it should match anything
[14:03] <lolek> ok, give me a second
[14:04] <w7z> also, you still need to file the bzr-svn bug with your traceback? I don't see it.
[14:04] <lolek> ok, that's working
[14:05] <lolek> when i placed: [something] with only user/pass it was working
[14:05] <lolek> ok, i'll fill it up :)
[14:06] <w7z> and you got the line in .bzr.log?
[14:06] <w7z> did you try branching your svn thing, or another random branch?
[14:06] <lolek> yes there is line: using authentication section :  'something'
[14:06] <w7z> I wonder if just the scheme matching is broken or something
[14:07] <lolek> i tried to branch svn thing
[14:07] <w7z> okay, so, try adding back in the host, then the port, then finally the scheme
[14:07] <lolek> ok
[14:07] <w7z> see which one makes it break again
[14:07] <lolek> give me a minute
[14:09] <lolek> it's ... port directive
[14:09] <w7z> fun...
[14:10] <lolek> yeah
[14:10] <w7z> so, with the scheme and the host it works, but plus port then doesn't?
[14:10] <lolek> when i added port=... it goes with the error  (pass is not unicode..)
[14:10] <lolek> correct
[14:10] <w7z> great, I can imagine the bug now
[14:11] <w7z> "3690" != 3690 I bet
[14:12] <lolek> hmm
[14:12] <lolek> ok, give me second
[14:12] <w7z> so, for you right now, just omitting the port is fine
[14:12] <w7z> but can I request two bug reports from you
[14:12] <w7z> one for the password traceback against bzr-svn
[14:13] <w7z> and one against bzr for the autentication.conf section not matching when you supply a port
[14:13] <lolek> but the password traceback is only visible when the port directive is given
[14:13] <lolek> there is no problem if there is no port directive :)
[14:13] <lolek> aaa ok
[14:13] <lolek> i understand, ok
[14:15] <lolek> https://bugs.launchpad.net/bzr-svn/+bug/1076386
[14:16] <lolek> here
[14:21] <lolek> https://bugs.launchpad.net/bzr-svn/+bug/1076388
[14:31] <w7z> thanks lolek!
[14:31] <lolek> you welcome :)
[14:33] <lolek> ok, i'm going... bye :)
[14:33] <lolek> thx for helping me out :)
[15:40] <jetsaredim> is it possible to create a branch based on the tree of a certain release of a given package?
[15:41] <jetsaredim> like - i want to try to fix something in a quantal package but there is already a raring version of the same
[15:41] <jetsaredim> so i'm assuming if i just branch the project it will give me the raring version
[15:48] <mgz> you can branch lp:ubuntu/SERIES/PROJECT instead of lp:ubuntu/PROJECT
[15:49] <jetsaredim> i seem to be getting an error with bzr
[15:50] <jetsaredim> http://paste.ubuntu.com/1342817/
[15:52] <mgz> jetsaredim: you don't have ssh configured correctly or haven't given launchpad the right ssh key
[15:52] <mgz> use `ssh -vv YOURLPUSERNAME@bazaar.launchpad.net` to debug, it's not a bzr related problem
[15:53] <jetsaredim> mgz: this would be the key I uploaded to launchpad vs my gpg key locally?
[15:55] <mgz> jetsaredim: this is your ssh key, from ~/.ssh, which is a different thing from your gpg key. see https://help.launchpad.net pages for walkthrough
[15:57] <jetsaredim> ahh - much better
[15:59] <jetsaredim> so - once i download the release-based tree for a given package i can then branch it locally for my own changes?
[16:00] <w7z> jetsaredim: see <http://developer.ubuntu.com/packaging/html/patches-to-packages.html> for details
[16:01] <w7z> be a little careful with terms 'tree' and 'branch', they mean specific things for bzr, <http://doc.bazaar.canonical.com/bzr.dev/en/user-guide/core_concepts.html>
[16:04] <w7z> also you need to go through the SRU process for quantal, so might want to check in #ubuntu whether the change you're planning is suitable for that, see <http://developer.ubuntu.com/packaging/html/security-and-stable-release-updates.html>
[16:04] <jetsaredim> w7z: its a pretty simple change to the build config
[16:05] <jetsaredim> and its already in raring
[16:05] <w7z> that doesn't sound security or serious data loss related
[16:06] <jetsaredim> the package is broken for certain functionality due to a config option not being set
[16:08] <w7z> jetsaredim: try asking in #ubuntu-devel
[16:22] <delinquentme> is there some reason that I need to pass a --no-strict .. to push to a parent when I've got locally uncommitted changes?
[21:15] <mathrick> hmm, I have a colocated project created on windows, which I just zipped and unpacked on linux
[21:16] <mathrick> and I can't get it to work, since it can't find the repo/branches with differing absolute paths
[21:17] <mathrick> http://pastebin.com/JwMDwvpK
[21:17] <mathrick> how do I fix it?