/srv/irclogs.ubuntu.com/2008/05/14/#bzr.txt

beunohm00:00
beunoI probably meant 1k00:00
jambeuno: internally, bzr log will be traversing the whole list, as will 'bzr revision-info' if it gets a non-mainline commit00:00
jambeuno: so calling it 1 time with a lot of revisions00:00
jamshould be a lot faster than calling it lots of times.00:00
beunojam, I shoulds absolutely go down that path then00:01
beunoI wonder how hard it would be to change the current code...00:02
nDuffin what contexts is bzr-git intended to be presently usable?00:07
jamnDuff: proof of concept, afaik00:16
jambzr fast-import is the recommended way to convert from git00:16
jambzr-git probably has some bitrot, etc.00:16
bob2I think it could run 'bzr log' on a git repository00:17
jambob2: I believe that was working as well00:19
jambecause it could run "bzr viz" at the time00:19
lifelessyes, my first fraft at EP2006 did that :P00:24
=== mw is now known as mw|out
rockstarabentley, ping02:06
abentleyrockstar: pong02:06
rockstarYou seem to be close to my timezone, and I'm rather interested in getting into Bazaar hacking, probably outside of Canonical time.  I was wondering if we could arrange a mentoring situation.02:07
rockstarI promise to stay out the way as much as possible.02:08
abentleyrockstar: Sure, I'd be happy to.02:08
abentleyDid you have a task in mind?02:08
rockstarNot really.  I looked at the bug list, and I'm not really sure I know enough to say "That's a good starter bug"02:09
abentleyThere are actually a few that are tagged that way.02:09
* rockstar looks again02:10
abentleyLook for the "easy" tag.02:10
rockstarAh, 23 of them.02:11
abentleyAlso "trivial".02:11
abentleyI can't guarantee that the assessment is right-- some are probably only "easy" until you actually try to fix them... :-)02:12
rockstarabentley, yea, that's what I gathered out of some of the trivial bugs.  Didn't see the easy ones02:13
=== kiko is now known as kiko-zzz
rockstarAlright, I'll just take one of these, and start exploring.  Be prepared to be harassed.02:14
abentleyIf you let me know which one you're doing, I can give you some starting points.02:16
rockstarI was thinking this one:  https://bugs.edge.launchpad.net/bzr/+bug/20057502:17
ubottuLaunchpad bug 200575 in bzr ""Address already in use " crashes smart server" [Medium,Triaged]02:17
abentleyrockstar: It seems there's already a patch attached to that bug.02:18
rockstarAh, bugger02:19
rockstarhttps://bugs.edge.launchpad.net/bzr/+bug/8957002:19
ubottuLaunchpad bug 89570 in bzr "unknown format exception could be clearer" [Low,Confirmed]02:19
abentleyrockstar: I'm not sure what more needs to be done there.02:22
abentleyNewer formats do indicate the minimum Bazaar version in their format strings.02:23
abentleyBut people still get thrown from time to time.02:23
abentleyOf course, it's impossible for Bazaar to guess what version is required for an unknown format.02:24
rockstarAlright.  How 'bout I take an hour or so and find a few of them, and them come back to you.  :)02:24
abentleySure.02:25
thumperpoolie: ping02:47
igcthumper: poolie on an leave today02:48
igcs/an/on/02:48
thumperigc: thanks02:48
thumperigc: are you up for a call?02:48
igcI guess so02:49
thumperyou sound so enthusiastic ;-)02:49
igcwell - I'd need to know what's its about02:49
* igc lunch03:45
abentleyBB is temporarily down while I update the machine.04:26
igcout to a medical appointment - bbl05:02
=== mwhudson_ is now known as mwhudson
abentleyBB back.  Failed to install Hardy.06:44
bob2eep06:44
Pengabentley: Upgrading from what to Hardy?06:45
abentleyPeng: Feisty06:45
PengOh.06:46
abentleyNow, it's a Xen box, so not your standard install target.  But still ...06:46
PengGutsy to Hardy was trivial for me, but then, that server (VPS) didn't have much of anything installed at the time.06:47
abentleyPeng: The stuff failing was pretty basic stuff, like initramfs-tools06:48
RAOFYou did go through gutsy, right?06:50
abentleyRAOF: Yep.  Gutsy was where the problems started.06:50
RAOFEh.06:51
RAOFI'm not sure if I've never had problems updating or whether I just don't remember them ;)06:51
=== doko_ is now known as doko
gouri see that opensolaris uses bzr-0.7 when evaluating and it looks speed was their concern  for 'cons'07:10
goursame with mozilla...07:11
bob2'cons'?07:13
gouropposite of 'pro'07:13
bob2ah07:13
Peng0.7?07:13
bob2thought it might be run by lisp nerds07:13
Penggour: Bzr has gotten a lot faster, of course, (and better in every other way) but it still has problems with huge projects.07:14
AfCgour: that is true, but then when they were evaluating bzr 0.7 was all that was out. Personally I think they were short sighted in their decision, but it is difficult to fault the "set criterion and then pick the one that scores best" approach. It's called "engineering" :)07:14
gouri do not plan to use bzr for such big project(s), but i'm more interested to hear about other kinds of problem i had with darcs-1 like  doppleganger bug and exponential time07:15
rockstar0.7?  That's older than what's in Gutsy.07:15
james_whi AfC07:15
AfCIt's unfortunate, because the Bazaar project lost out on a significant adoption opportunity there. While no one really cared about OpenSolaris, as such, it was obvious that whatever they picked would be adopted by the Java Engineering team at Sun, which in turn is propagating out via Open Java rapidly.07:16
gourrockstar: see http://www.opensolaris.org/os/community/tools/scm/bzr-eval/07:16
AfCjames_w: hey James. Been a  while. I hope you're well.07:17
james_wI'm good thanks, how are you?07:17
gourAfC: with mozilla it was similar situation?07:18
AfCjames_w: Just about to go for a run, actually. Thought I would chime in before I set off.07:18
james_wso you're ready to make a quick exit07:18
gourso, bzr probably does not have 'doppelganger' problem by design?07:22
james_wgour: not that we know of07:23
gourthat's nice...i was bitten by it in darcs several times...07:24
AfCgour: if you're talking about http://research.operationaldynamics.com/blogs/andrew/software/version-control/red-giant-bugs.html then, no, Bazaar doesn't have that problem07:24
AfCgour: let me put it this way. If you are doing work with Bazaar and you have a project that is so large or a use pattern that is so unusual so as to cause Bazaar problems, I can only suggest that you will likely find interest here and likely will find that if you can offer your repository to the Bazaar hackers to work with that they will be able to use it in their tests and optimizations.07:25
gourhere it is in more detail - http://wiki.darcs.net/DarcsWiki/ConflictsFAQ07:26
gourAfC: thanks. i had similar experience with darcs (droundy was inspecting my repos), but, unfortunately not easy to fix those design-issues07:27
AfCgour: otherwise, asking "is it fast enough" is all a bit silly, really. If it wasn't, there wouldn't be so many people here using it quite happily. Could it be faster? Probably, but that's the wrong question. Does it work reliably? Does it prevent you from shooting yourself in the foot and thereby loosing data? Does it have a clean and consistent user interface? are far more relevant.07:27
gouranother nice thing i see in bzr is bzr-svn (i might soon switch to NixOS and they stil use svn)07:27
AfCgour: as I noted, we used Darcs for years, loved it, and miss it still.07:27
Penggour: Yeah, bzr-svn rocks.07:27
gourAfC: right. git is like automatic gun to shoot your foot all the time :-D07:28
bob2and fast like a greased fox07:28
gourAfC: my 'speed' argument was just in the light of bzr vs. hg for solaris and mozilla07:29
gourfor me it's not so important - finally, that's why we use dvcs ;)07:29
AfCgour: but Bazaar is an improvement in many other areas, so we swallow our impatience with the less-then-Darcs user interface and abysmal lack of cherry picking support in favour of having a robust tool with a vibrant and intelligent group of hackers working on it.07:29
gourAfC: so far (reading user manual etc.) i like bzr, today i'll try with some tests...interface looks nice (i won't even compare it to git's surprises)07:31
* gour --> breakfast. bbs07:31
asabilAfC: maybe I can convince you to hack on bzr-interactive ? :p07:31
fullermdPshaw.  git's interface makes perfect sense if you follow a few simple steps.07:31
fullermdFirst, you become Linus...07:32
AfCfullermd: no, you've got to learn Finnish first.07:32
fullermdAh, no wonder it doesn't work right.  I keep mixing up those Scandinavian countries.07:35
fullermd% git de bork bork bork07:35
fullermdgit: 'de' is not a git-command. See 'git --help'.07:35
* AfC goes out07:37
=== weigon__ is now known as weigon
AfCWhat is the version of bzr-svn you have to have to work with bzr 1.4?08:51
luksAfC: there isn't one that works with 1.4, as far as I know09:03
bob20.4.10 is compatible with 1.4 and 1.509:10
* igc dinner09:10
=== mario_ is now known as pygi
gouri see that bzr help formats shows plethora of different formats...will this trend to continue or is it reasonable to expect bzr to settle on some proven one?09:50
Penggour: Well, they probably thought they were going to settle on each one as they created it.10:09
Penggour: But yes, it should go down.10:09
gourPeng: the 0.92 format is that git-like 'pack' format?10:10
Penggour: I'm sure the implementation is different, but it's the same idea, yeah.10:11
Penggour: BTW, there are 5 or 6 more hidden formats! ;)10:11
gouronly?10:12
gour:-D10:12
Penggour: Whenever they've changed the branch, repository or working directory format, they've added a new format to the formats list, and pretty much all of them are still supported.10:13
Penggour: E.g., I think dirstate is just knit with a different working tree format. dirstate-tags is just dirstate, only the branch supports tags. pack-0.92 uses the same branch and working tree formats as dirstate-tags, but an entirely new repo format.10:14
gourin any case, 0.92 is default now and recommended one?10:16
Penggour: Also, there are now multiple variations of the same base format: There's pack-0.92; rich-root-pack, which is the same but supports slightly more meta data; and pack-0.92-subtree, which has experimental support for stuff similar to svn:externals.10:16
Penggour: pack-0.92 is.10:16
Penggour: They want to switch the default to a rich-root format soon, which will simplify things somewhat. (Rich roots are perfectly stable, but you can't downgrade from a rich-root format to a non-rich-root format.)10:17
gourPeng: in 1.6 ?10:30
Penggour: Maybe.10:31
gourit would be nice10:31
elmobzr: ERROR: Transport error: Server refuses to fullfil the request10:44
elmo'sup with that?10:44
Pengelmo: Are you the guy from the mailing list?10:47
elmoPeng: err, i don't know? if you mean, have I posted about this problem on the mailing list, then no10:47
elmoI only ran into it 10 seconds ago10:47
Pengelmo: Then, someone else has the same issue on the mailing list. :)10:47
elmook10:47
elmoapparently it's a 1.4 regression too10:47
elmosomeone else with 1.3 claims it WF him10:48
elmoI'll try downgrading10:48
Pengelmo: Run bzr with -Dhttp -Dtransport and pastebin the relevant section of your .bzr.log (warning, really long).10:48
Pengelmo: Are you experiencing this with the same branch?10:48
elmoPeng: same branch as what, sorry?10:48
elmowow, it really does work with 1.310:51
james_welmo: I presume this isn't python-apt--debian-sid from mvo's people.debian.org you are trying to grab?10:52
elmoit's http://bzr.debian.org/da-tools/da-tools/userdir-ldap-common10:53
elmo1.4 seems to a POST?10:53
elmowhich gets 403ed10:53
james_wdid 1.3 -> 1.4 have the addition of probing for the smart server?10:53
james_wah, yes, it seems so.10:53
james_welmo: I think nosmart+http:// is the transport that forces that off, which should allow you to work around this.10:54
elmojames_w: 'apt-get install bzr/hardy' is good too :-) thanks10:54
elmofiled as #23022311:01
james_wthanks11:03
=== mwhudson__ is now known as mwhudson
awilkinsjam: Ping?11:19
KemuriiHi, I'm having a rather small Mercurial repo, and need to move it to Bazaar.. So, I'm using fast-import:   bzr init-repo new.bzr ; hg-fast-export.py --repo=../old.hg | bzr fast-import -  .. I copied first all files from old.hg..  but, it says "bzr: ERROR: No WorkingTree exists for" ... I'm missing a step somewhere?11:48
mwhudsonKemurii: init-repo doesn't make a branch or a tree11:49
mwhudsontry bzr init-repo new.bzr; bzr init new.bzr/import-branch; cd new.bzr/import-branch11:49
mwhudsonthen the same commands again11:49
Kemuriiah, mm11:49
PengKemurii: Just out of curiosity, why move from hg to bzr?11:51
goursame question here as well11:51
KemuriiPeng: company policy, I'm sticking with Mercurial for personal stuff11:51
PengKemurii: Oh. That sucks.11:52
PengAt least the company policy isn't VSS or something.11:52
Kemuriiwell, 'company policy', nothing says I have to do it in bzr, but well.. :)11:52
Kemuriimwhudson: hehe, that somehow worked :)  thanks!11:53
mwhudsonKemurii: no worries11:53
PengKemurii: Oh, good.11:55
Kemuriimtaylor: here?11:55
Kemuriiwoohoo, pushed to parent, nice nice11:59
Kemuriinow I have to learn typing 'bzr' instead of hg or bk12:01
* awilkins proposes the evil solution of aliasing hg and bk to something that mocks and ridicules you12:04
awilkinsrandom-insult.sh should do12:04
awilkinsKemurii: Your company policy is highly enlightened. I know shops that are still using VSS and moving to TFS12:05
awilkins(blech)12:05
KinnisonKemurii: because my prompt knows what vcs I'm using, I am tempted to start using 'vcs' as the command and let my shell sort it all out for me :-)12:06
Kemuriiawilkins: not sure it's policy yet, just 1st option for VCS now12:07
KemuriiKinnison: now that's an idea :)12:07
awilkinsAt least in SVN shops you can still use bzr.12:09
awilkinsI'm the svn server admin here, and I use bzr these days.12:09
Kinnisonvcs () { if [ "x$_VCS" = "x" ] then bzr "$@"; else $_VCS "$@" ; fi }12:10
Kinnisonthat should do the trick for me12:10
Kemuriihehe12:10
PengTFS?12:12
Kemuriianyway, thanks again guys, back to work, I mean lunch, cheers12:14
awilkinsTeam Foundation Server (with all new VSS! Blech!)12:20
lambydato: I'm still failing at accessing that repo over http. I'm using your exact command too.13:35
pickscrapeIs there some way to suppress the deprecation notice output when running bzr?13:47
kaalooHi, I was wondering if there was an issue with bzr-svn using authentication.conf ?  I'm using 1.3.1-113:50
datolamby: odd13:51
datolamby: tried from another host?13:51
lambydato: Ah, that seems to have helped. I'll poke a little more.13:57
awilkinskaaloo: I'm not sure it even uses authentication.conf14:03
pickscrapeIs there way to define location 'bookmarks', in the same way that the launchpad plugin lets you reference lp: ?14:04
pickscrapee.g. so I can define the 'mine' refers to bzr+ssh://example.com/bzr, and subsequently use bzr branch mine:trunk14:06
Pengpickscrape: THere's a bzr-bookmark or bzr-bookmarks plugin. Not sure how exactly it works.14:07
pickscrapeOh, cool14:08
Odd_Blokelamby: What bzr version are you using?14:53
lamby1.4.14:53
lambyNow this is interesting, can reproduce on 1.5~rc1 from a different host.14:55
lambyWould you like a login? It's a testing machine anyway.14:56
james_wlamby: hi14:56
lambyGood afternoon.14:56
james_wsorry, I missed the start of this14:56
james_wis this a problem with http?14:56
lambyThis started on pkg-bazaar-maint.. So the testcase is "bzr get http://bzr.debian.org/pkg-bazaar/trac-bzr/unstable/ trac-bzr".14:57
james_wah yeah, I saw that14:58
james_wis there a difference in whether pycurl is installed on these machines?14:58
lambyHeh, psychic debugging.. Installing pycurl made it work.14:59
Odd_BlokeI was going to suggest that yesterday, but thought it might be _too_ psychic to be right. :p15:00
james_wbug 230223 may be relevant then15:00
ubottuLaunchpad bug 230223 in bzr "smart server probing in 1.4 breaks check outs of short bus http repositories [regression]" [High,Confirmed] https://launchpad.net/bugs/23022315:00
lambyAh yes, seems to be the same bug. tcpdump is confirming that only the POST is attempted, which 403s.15:01
james_whopefulls spiv or vila will be able to find a fix soon.15:03
lambySubtle ping.15:04
lambyjelmer: bzr-gtk just uploaded with DM-Upload-Allowed: yes.  I'm still seeing the breakage when visualising via `olive-gtk` (when `bzr viz` works fine). Did you try a fix? Shall I file a bug?15:04
jamjames_w: well, there is already the workaround of nosmart+http://15:05
james_wjam: true15:06
vilajames_w: you got the targets right :-) My plate is already pretty full so I'm not sure I can fix it quickly. But I think the diagnosis points to hpss probing, also bug #229076 is pretty similar15:06
ubottuLaunchpad bug 229076 in bzr "'Connection reset by peer' error when branching repository" [Undecided,New] https://launchpad.net/bugs/22907615:06
vilahehe, nice point jam, indeed nosmart for the win :)15:06
vilaphoen15:06
* dato marks "upload bzr-gtk" off his TODO :)15:07
james_wvila: yeah, I think that's what it is. It also appears to be a difference in the handling of 403 response between urllib and pycurl.15:07
* abentley really hopes we can get rid of pycurl soon.15:08
datosigh, I forgot to push to the bzrtools packaging branch again15:28
orospakrHey, is there a bazaar equivalent to git-archive?  I'd like to spit out a tarball of all the files that are in version control.15:33
james_wexport15:34
james_w"bzr export <location> [<branch>]"15:34
james_wlocation can end with .tar.gz to make a gzipped tarball15:34
james_wnote that it exports the last revision, not the working tree, so uncommitted changes won't be included.15:34
orospakrthat's exactly what I want. Thanks! :D15:37
jamvila, james_w: there also seems to be a problem where you get the 403 a long time after you started downloading15:38
jamAt least, that is what Ben Finney's logs seem to indicate15:39
jamI'm guessing a permissions issue, but I'm not positive15:39
=== kiko-zzz is now known as kiko
james_wjam: yeah, I saw that just now, I was going to look at the permissions on the directory that you pointed to, but I'm currently locked out of that machine.15:39
awilkinsjam: I've been having some success with your service plugin as a backend for bzr-eclipse15:40
ricardokirknerhi, what is the best way to configure bzr over http with authentication? fastcgi or mod_python? or something else?15:41
awilkinsjam: I had to change it to threads rather than forks because the library implements ForkTCPServer with fork() (duH). No forks on Windows.15:42
jamawilkins: I will warn you that bzrlib isn't particularly thread safe15:42
awilkinsjam: It should all be single-threaded access.15:42
jamand you can fork under cygwin, so it *is* possible, but AIUI they had to jump through a lot of hoops to get it to work15:42
awilkinsjam: Yeah, I mean "the os module in CPython doesn't do fork()"15:43
awilkinsFor the purpose I'm using it for, thread isn't so bad ; it gives charge of the process to the Eclipse instance with no extra steps involved.15:44
awilkinsDammit, IO blocking again15:45
jamawilkins: I would just mention that you probably want a plain TCPServer not a Forking* or a Threading*15:46
jamI believe if you use the base class, you will get a single-threaded server15:46
jamwhich is what you need anyway.15:46
awilkinsPerhaps I'll change it down one then...15:47
awilkinsIt so annoying ; it works lovely when you run the tests one-class-at-a-time... then it blocks on IO when you run them as a block15:49
awilkinsIs JUnit multithreaded?15:50
jamawilkins: I don't know about Junit, but enforcing multithreaded test running without requesting it seems like bad mojo15:51
awilkinsOr bad "juju" for JUnit :-)15:52
kaalooawilknis: thanks, I'll check the code, maybe there is a fix15:53
ricardokirknerI am trying to set up a bzr repo for my project, which I migrated from svn. the bzr stuff is working ok, but I need to be able to checkout/checkin through http so as to mimick the previous setup. checkout through http is ok, but in order to commit, do I need webdav or anything like that? (I dont want to give the apache user full write permission to my repository)15:53
ricardokirknerok, I managed to checkout my branch using bzr+http and mod_python... now a strange thing happened... if I try to check out a normal branch, it complains about KnitpackRepository is not compatible with RemoteRepository, but if I checkout a lightweight branch, then it works...16:02
ricardokirkneris this a feature or a bug?16:02
james_w"checkout a lightweight branch"? I'm not sure what you mean, "checkout --lightweight"?16:02
ricardokirkneryes16:03
james_wand yes, there are two options, bzr+http with write access, or webdav. I've not used either, so I don't feel confident to explain the pros and cons.16:03
jamricardokirkner: you converted from svn16:03
jamwhich means you need to use 'bzr init-repo --rich-root-pack' locally16:03
ricardokirkneryes :$16:03
jamthe problem is that the RemoteRepository is hiding the "real" versions on each side to give you a clearer error16:04
jamyou can also do "bzr upgrade --rich-root-pack" in your local repository16:04
jamand things should work16:04
jamricardokirkner: I would probably recommend using bzr+http with write access, but I would *probably* do it as bzr+https: and make sure you have proper authentication at the http layer16:04
ricardokirknerjam, but then every developer will have to know that the project originally started in svn, and prepare a local repository by hand? isn't that a bit of counterproductive?16:04
jamricardokirkner: bzr-svn "jumped the gun" and started using some development formats16:05
jamsupported, but not default16:05
jamwe are working on making it the default format for 1.616:05
ricardokirknerjam, ok I think I understand. the correct behaviour should be that the bzr server tells the client which format that branch is, so that the client can create the repository in the correct format instead of assuming the default format. am I right?16:06
jamricardokirkner: you did 'bzr init-repo' on the local side first16:07
jamif you did not do that16:07
jamthen doing "bzr branch" would preserve the source format16:07
jam'init-repo' is a good thing, but 'bzr branch' isn't going to upgrade your repo for you16:08
jamas if you are using a specific format, there may be a reason16:08
jam(old clients, etc.)16:08
ricardokirknerno jam, I did not issue init-repo on the local side first16:08
ricardokirknerI have a repo on the server side, but I just did a bzr co16:08
jamricardokirkner: hmm... that seems a bit weird, I'll try to check the code for a sec16:08
kaaloook, update on my bzr-svn problem, of it not using authentication.conf, actually pycurl was not installed, now I'm getting a "certificate verification failed" error from pycurl, which is actually true since I had to deal with that using svn co, but here pycurl simply fails16:09
ricardokirknerjam, if I try a bzr branch, then I get a traceback16:10
jamricardokirkner: ok, it does seem to be a problem with the RemoteRepository not reporting the right source format, I can reproduce it here trivially16:10
jamricardokirkner: interestingly 'bzr branch' works fine16:10
jamricardokirkner: can you paste your traceback16:10
jam!paste16:10
ubottupastebin is a service to post multiple-lined texts so you don't flood the channel. The Ubuntu pastebin is at http://paste.ubuntu.com (make sure you give us the URL for your paste - see also the channel topic)16:10
ricardokirknerjust a sec16:10
ricardokirknerjam, http://paste.ubuntu.com/12071/16:13
gouranyone uses fish shell? i just noticed it does not have bzr-completion :-(16:13
kaaloook this is an old bug that hasn't been fixed in pycurl : https://bugs.launchpad.net/bzr/+bug/8208616:15
ubottuLaunchpad bug 82086 in bzr "pycurl transport causes tracebacks if the server's SSL cert cannot be verified." [Medium,Confirmed]16:15
datojam: your latest import cleanup to stats broke things... http://dpaste.com/49901/16:18
jamdato: update to trunk, you are using the wrong one16:19
jamlp:bzr-stats is the official trunk16:19
jamnot the one on bzr.arbash-meinel.com16:19
jamI'll try to fix my branch16:19
datohm16:20
datook, sorry for the noise.16:20
jamdato: you aren't the first one, I was just being lazy about my branch16:21
jamI'll redirect it to launchpad16:21
datoI supposed update-mirrors now lives in... lp too?16:21
jamdato: actually, I don't know if I ever moved that one, I should16:22
datohaha16:22
jamdato: some old code there :)16:22
jamI believe Jelmer was interested in helping with stats, so I moved it to lp:16:22
jamdato: I haven't touched update-mirrors since 2006-06-1916:22
jamso almost 2yrs ago16:22
jamI'm amazed it still works properly16:23
jamI guess we have at least *some* api stability16:23
dato*g*16:23
datoit certainly does work, against 1.5 even16:23
jamdato: can you try doing an update/pull/branch from my bzr.arbash-meinel.com/plugins/stats ?16:25
jamI believe it should redirect you to lp:///bzr-stats now16:26
datostray \n in location, I'd say16:26
jamdato: try again16:27
datoworks :)16:28
jamdato: does it give you any indication, or just "work" but point at a different location when it is done16:29
jammy understanding is it just works, but if you do "bzr info" you will find out it went somewhere else16:29
datoyes, it changes your url16:30
kaaloohmm, unfortunately the pycurl http transport implementation doesn't seem to use AuthenticationConfig as does the urllib2 implementation :(16:37
denndahttp://paste.pocoo.org/show/50195/ <-- Urg, what's wrong?16:50
dennda(This is arch linux)16:51
denndaThe very same command works well on my ubuntu box16:52
denndabzr 1.316:53
kaaloook, I think I see what's happening, the svn repository is being accessed through WebDav over https, and it requires authentication to checkout a revision.  Running bzr branch https://...  is not getting through to bzr-svn because of the authentication issue, when starts to check the branch format, it coughs on a 401.  On the other hand running bzr branch svn+https://.. tries to go through the svn smart server protocol which of course fails 16:53
james_wdennda: it's something to do with paramiko.17:01
james_wif you search for the error string you should find the bug report.17:02
mxpxpodhow would I undo a merge?17:59
=== kiko is now known as kiko-fud
LeoNerdrevert? uncommit..?18:00
asabil_mxpxpod: bzr revert18:01
LeoNerdmerge backwards?18:01
asabil_if you didn't commit18:01
mxpxpodasabil_: revert still leaves pending merges in the status18:01
mxpxpodah, wait18:01
mxpxpodn/m... I was doing bzr revert *18:01
asabil_bzr revert --forget-merges18:02
asabil_only reverts the merge status18:02
mxpxpodnice, thanks18:02
abentleymxpxpod: "revert" with no arguments will forget the merge status.  Supplying arguments prevents that.18:12
mxpxpodabentley: thanks :)18:12
abentley(revert with no arguments reverts all files too)18:12
jamricardokirkner: the exception shown here: http://paste.ubuntu.com/12071/ is a bug in 1.2.0 which has been fixed in later releases. You can get 1.5rc1 from http://launchpad.net/~bzr/+archive, or we can try to work out something else.18:27
jamdennda: this is actually an interaction with a newer paramiko18:27
jamif you use paramiko 1.7.2 you don't have that problem18:27
jamparamiko 1.7.3 causes the "no attribute "get_name"" traceback18:28
ricardokirknerjam, thank you for your help. what version do you suggest me to use? I was using 1.4 previously, but downgraded due to some issues with that version too.. what is the most stable version (with a good amount of support for svn) for use in production right now?18:28
=== kiko-fud is now known as kiko
denndajam: uff, I doubt I can install the older version through pacman that easily18:57
denndaIs there some sort of push-history in bzr?19:08
jamricardokirkner: sorry, about the delay, I have a sick child here.19:43
jamanyway, what problem were you having with 1.4?19:43
jamdennda: push-history?19:43
denndajam: bzr stores the last push location19:51
denndaI wonder if it stores the last N locations19:51
denndaaccessible via push N19:51
denndaI will drop the net now, brb19:52
Stavroshello20:20
Stavrosi want to use bzr for backups, to version a folder on a different hard disk, is there a way not to have the entire repository on both disks?20:20
Stavrosi would like to minimize disk space20:21
Stavrosso if i could have JUST the repo on the backup disk and JUST the working copy on my normal disk it would be ideal20:21
Stavrosis that possible?20:21
beunoStavros, maybe something along the lines of a lightweight checkiut20:21
beunobut, if you have large binary files, it might not be a good idea to use bzr20:22
Stavrosbeuno: hmm, yes, how do i do that?20:22
Stavroshmm20:22
Stavrosi have large binary files but they don't change much20:22
Stavrosthey're mostly pictures and such20:22
beunoStavros, bzr uses a lot of ram for binary files20:22
Stavrosi'd use git for backup (is that better for binaries?) but its windows support is lacking20:22
Stavrosoh20:22
Stavroshmm20:22
beunois this going to be one way syincing?20:22
Stavrosyes20:22
Stavrosmostly added files20:23
Stavrosno changed ones20:23
beunoStavros, bzr will use aprox 3x the file size in RAM20:25
demodsounds like a job for rsync or rsnapshot20:25
beunojust so you can calculate what it will take20:25
beunobut yeah, rsynce sounds like a better candidate20:25
Stavrosi do use rsync, i wanted a better solution because rsync takes ages20:25
=== juliank0 is now known as juliank
demodthats odd, rsync should be rather fast since it's only checking file size and modification time (iirc)20:28
Stavroshmm, indeed20:30
Stavrosmaybe it's because around 10 mb20:30
=== fullermd_ is now known as fullermd
brilliantnuthi, I've been around asking this a long time ago, but what is the best way to have a central repository with LDAP based authentication?21:37
brilliantnutbasically, we looked at bzr+http earlier, but there were some open bugs concerning those, so we left it at that... this was around 1.021:42
jambrilliantnut: you could use 'bzr+ssh' and have ssh use pam+ldap for authentication22:37
jamI believe you can even put ssh keys into ldap22:37
pickscrapeAnyone know the bzr-svn syntax for subversion.conf for explicitly specifying the branches in the svn repository?22:38
pickscrapebzr svn-branching-scheme --set is crashing for me22:39
brilliantnutincidentally, this was the exact suggestion we got the last time around, but we came back to bzr+http for some reason... We really need to bzr+http to work...22:42
brilliantnutjam: we have followed instructions on http://doc.bazaar-vcs.org/bzr.dev/en/user-guide/index.html#serving-bazaar-with-fastcgi22:44
=== c1|freaky_ is now known as c1|freaky
jambrilliantnut: I'm not sure what your current state is. You followed the instructions and it didn't work?22:56
brilliantnutjam: yes. We followed the instructions on the link that I just posted.23:02
brilliantnut/home/tech-admin/data/bzr is the repository, and is mapped to http://ourserver/bzr, /home/tech-admin/data/bzr/HEAD is the root branch.23:03
brilliantnutbzr branch http://<credentials>@ourserver/bzr works just fine23:04
brilliantnutactually scratch that23:06
brilliantnutbzr branch http://<credentials>@ourserver/bzr/HEAD works just fine23:07
brilliantnutbzr branch bzr+http://<credentials>@ourserver/bzr/HEAD gives an error: bzr: ERROR: No repository present: "bzr+http://<credentials>@ourserver/bzr/HEAD/"23:07
brilliantnutI could pastebin the bzr-smart.py and relevant portions of our httpd if needed...23:10
bob2also logs23:10
brilliantnuthttp://paste.ubuntu.com/12141/23:19
brilliantnutThere are no relevant logs in error.log with that error... access.log shows a 401 followed by several 200s... post which I get the "No repository present" error.23:20
bob2adoes http://blahblah/.bzr/smart 404?23:21
brilliantnutno, no 404.23:21
brilliantnut401 Unauthorized23:21
bob2great23:22
brilliantnutwhich is fine, since it tries to access without auth first, then I provide credentials, and it goes ahead to 200s23:23
brilliantnutall the 200s are http://blahblah/bzr/HEAD/.bzr/smart23:24
bob2perhaps "bzr -Dhpss co http://blah" will be enlkightening23:25
deepjoythis is what we saw as brilliantnut described http://paste.ubuntu.com/12142/23:26
brilliantnutdeepjoy is with me, by the way...23:26
jambrilliantnut, deepjoy: There is a plugin you might try23:30
jam'bzr co lp:bzr-hello ~/.bazaar/plugins/hello23:30
jamit adds a "bzr hello URL" command23:31
jamwhich just tries to send the "hello" request to the server23:31
jamthat can be a "is the smart server running" check23:31
jamIt does look like the POST is getting through23:31
jamYou have a "/bzr/.bzr/repository/" directory?23:32
jam(I'm just checking where the actual repository is at, and that the smart server was told that it has access to it)23:32
jamOtherwise you might have restricted the smart server to only serve things underneath /bzr/HEAD23:32
jamwhich would prevent it from reaching the repo at /bzr/...23:32
deepjoyjam: we can branch using the http protocol23:36
deepjoyjam: it's only when we use bzr+http that we have trouble23:36
deepjoyjam: no the config does not include HEAD anywhere23:37
jamdeepjoy: I understand that. I'm saying that you may have configured the bzr server with the wrong chroot23:37
jamI don't know for user23:37
jamsure23:37
jamyou could try sending a23:38
jambzr hello bzr+http://.../bzr/23:38
jamversus23:38
jambzr hello bzr+http://.../bzr/HEAD23:38
jamif the former fails but the latter works23:38
jamthen you are unable to contact the smart server via bzr/.bzr/smart (as opposed to bzr/HEAD/.bzr/smart)23:38
jamdeepjoy: alternatively, it could be a problem with your Apache config23:39
jamdepending on what rule you used to redirect POST => .bzr/smart to the cgi script23:39
deepjoyI'll post the relevant configs. just gimme a moment23:39
jamI'm not a huge expert on this, spiv knows more23:39
jamI'm just trying to point out some things to try23:39
deepjoyapache conf http://paste.ubuntu.com/12145/23:42
jamdeepjoy: that apache conf *looks* correct to me. Can you try installing the 'bzr-hello' plugin and see if you can talk to "bzr+http://..../bzr" ?23:43
jamdeepjoy: and, of course, make sure you've restarted apache since whenever you last changed the config files23:43
jamApache config always feels like shooting in the dark to me. As a stray '/' can totally break things.23:44
deepjoybzr-smart.py http://paste.ubuntu.com/12146/23:44
deepjoyjam: yes we restarted apache even if the change was only at the py file23:44
deepjoyapache is serving up trac on the same virtualhost also on mod_python23:45
jamdeepjoy: I'm wondering about the line in smart.py:23:45
jamprefix='/bzr/',23:45
jamYou might try23:45
jamprefix='/bzr',23:45
jamI'm not guaranteeing anything here, just something to try23:46
brilliantnutwe've tried both of them :)23:46
jamspiv should be on in another 15min or so, and he's had more experience with this23:46
jambrilliantnut: do you have the "POST" program installed?23:46
brilliantnutyes.23:46
jamyou should be able to do23:46
jamecho "hello" > POST http://.../bzr/.bzr/smart23:46
jamIt should  return "ok\012"23:48
jamsorry \001 or \x01 or ^A depending on how you want to view that character23:48
jambut on your terminal, you should see something "ok 2"23:48
brilliantnutjust a sec, sanitizing the conf again.23:50
jamspiv, igc, poolie: are you guys there?23:58

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