/srv/irclogs.ubuntu.com/2007/11/24/#bzr.txt

elijahHow would I determine how much space the svn-bridge information takes under .bzr?00:12
elijah(For a repository imported via bzr-svn.)00:13
lifelesswhat do you mean 'svn-bridge' ?00:20
elijahIf I import with bzr-svn, I may want to later do an incremental update.00:21
elijahDoes bzr-svn store any information to make later updates easier?00:22
elijahIf so, how much space does such information take?00:22
elijah(i.e. it's not fair to compare svn storage requirements to bzr storage requirements using bzr-svn on an existing svn repository if bzr-svn is storing extra information about bridging to an alien repository)00:23
elijahFrom poking around under .bzr, it appears the answer is not much, if anything...but I just want to be sure.00:24
lifelesselijah: the mapping data you are talking about goes in ~/.bazaar/00:24
lifelesselijah: and is regenerated on demand00:24
elijahah, so if I have a svn checkout of foo in foo-svn and a bzr-svn checkout of foo in foo-bzr, then it is fair to compare 'du -hs foo-svn' and 'du -hs bzr-svn', yes?00:25
elijah(Well, keeping in mind the fact that bzr stores all information on the branch instead of just a copy of the latest version)00:25
james_welijah: yes.00:27
elijahcool, thanks.  :)00:27
james_welijah: however it may not be representative of what you would have with a pure bzr branch of the same revisions, as bzr-svn may have some overhead, but it is all mixed in with the .bzr metadata00:28
elijahdoh00:28
elijahwell, I probably can't measure that easily, so I guess I'll have to do without.00:28
lifelessalso00:29
lifelesswe have annotations cached in .bzr00:30
lifelesssome formats make them optional, if storage is a premium00:30
lifelessand in about 2 months we're expecting a 50% size reduction in .bzr00:30
elijahwith the knit-pack format?00:30
beunolifeless, would you have a few minutes to talk about the XML/Output Abstraction patch?   We're still unsure on to procede...  :(00:33
lifelessbeuno: yes, but not right now sorry00:42
lifelesselijah: with a successor based on it. new delta logic and ordering00:42
lifelesselijah: its a minor tweak now we have the infrastructure in place.00:42
lifelesslater all00:42
beunolifeless, np, tnx anyway00:44
elijahlifeless: cool00:44
elijahthanks for the help, lifeless and james_w00:44
AfCDoes[n't] Launchpad have a bzr:// server running?04:07
jameshAfC: no04:13
jameshjust http: and bzr+ssh: at the moment04:14
jameshand sftp:04:14
jameshyou can access all branches via http and bzr+ssh04:14
AfCjamesh: yeah, but http:// is deadly slow04:14
jameshtry bzr+ssh then04:15
AfCI'm surprised at them for not running a Bazaar server. Hello, dogfood.04:15
jameshwe are running a Bazaar server04:15
AfCThere are some pretty evident differences between invoking a single bzr serve for a short duration over an SSH tunnel, and having a bzr serve daemon running.04:16
jameshAfC: we did have bzr+http set up a while back, but it seems to have bit-rotten04:16
AfCjamesh: (yeah, I gave up trying to make that work)04:16
AfC(too)04:16
jameshthere do seem to be some bugs in Bazaar related to closing connections04:17
AfCjamesh: Incidentally, in reply to your response of "try bzr+ssh://" to "my http:// is deadly slow", my comment was really more directed along the lines of "everyone is publicly being told (by Launchpad itself, and everything else)04:17
jameshAfC: well, we're looking at changing those instructions to "bzr branch lp:whatever"04:18
AfCjamesh: to do bzr clone http://whatever ... and thus the very first impression they get of Bazaar as a result is pathetic performance. We observed an improvement of the order of 22 minutes to ~105 seconds by switching to a public bzr:// instance04:19
AfCjamesh: yeah, I filed one myself about CLOSE_WAITs.04:20
jameshAfC: sure.  http or sftp are not really comparable to bzr or bzr+ssh04:20
AfC[which is usually a pretty clear flag or such]04:20
AfCs/or/of04:20
AfCThe Bazaar hacker crew it talking about 1.0 any day now; it's things like what we've been speaking about which I would encourage be cleaned up lest the marketing opportunity they are hankering after be wasted.04:21
AfCs/it/is04:21
AfCApparently I haven't had enough coffee yet today04:21
jameshI wonder if we have a bug for adding plain bzr:// protocol support04:21
AfCI'll file one if you want, but that's really your internal business.04:23
jameshplease do so04:23
jameshAfC: how many squares were on the senate ballot paper for you?04:26
AfCjamesh: Zero. I'm not an Australian citizen.04:27
jameshah.04:27
jameshyou should fix that :)04:27
AfC[This is almost, but not quite a bitch for me. Same problem in New York. I lived there years. I PAY TAXES. And yet I am unable to exercise representation. Taxation without representation has been the singular most dominating issue driving western politics in the last 300 years. Even religious conflict is usually cover for this issue]04:28
jameshI remember one of my former colleagues who moved states in Canada04:29
jameshthe two states had different enrollment rules, so he got taken off the roll in the old state, but had to wait a while before they'd put him on the roll in the new state04:30
jameshso he missed the election04:30
AfCuh huh04:30
jamesh(can't remember which type of election it was though)04:31
AfCI'm pretty much a permanent expatriate now. It's ok. I kinda like it that way.04:31
AfCI love how when I try to report a bug in launchpad it crashes my web browser.04:32
AfCjamesh: https://bugs.launchpad.net/launchpad/+bug/164790 filed04:34
ubotuLaunchpad bug 164790 in launchpad "Launchpad should run a Bazaar server and advertise bzr:// URLs." [Undecided,New]04:34
jameshwith the changes I did to the launchpad plugin for Bazaar, it'd be trivial to get people using bzr:// instead of http:// for lp: URLs purely through changes on the server04:36
AfCjamesh: that makes sense04:38
AfCjamesh: I'm mostly an outsider on all this, but we made an early commitment to Bazaar for what we believe are the right reasons. My colleagues and I just want to see Bazaar put forward as successfully as possible as a result.04:39
AfCie removing cruft, etc04:39
jameshAfC: yep.04:40
jameshhopefully you'll find some of the new Launchpad features in the coming months useful04:40
AfC[IMO (nothing humble about it) merely calling it "1.0" has little to do with achieving that. I know that'll bring more users, but on the other hand04:41
AfCthere are so many warts (and the two competitors are improving in usability rapidly enough) that it's really quite underwhelming as a result. Little things matter.04:41
jamesh1.0 is not the end by any means though04:42
AfCBut the nature of leadership making is that people make input and then a decision is taken. Martin has made his call, fine]04:42
AfCjamesh: I know. But all you're setting yourself up for is to get FUCKING PANNED in the trade press. Why would you invite this?04:42
jameshyou'd have to ask one of the Bazaar guys about release planning -- I'm primarily a Launchpad developer04:43
AfCSure04:44
AfCs/you/Canonical/04:44
* AfC is used to getting panned and mocked when he makes releases. Alas.04:44
AfCThat's Open Source for ya.04:45
fullermdAfC: That's why I take great care to get my code ALMOST ready for release, and then never touch it again.  It's not laziness, it's careful PR!  ;)07:43
AfCfullermd: ha07:43
fullermdHaving spent the last couple days working on code in mtn, I'm reminded of how much I prefer bzr.07:44
fullermdBut I do agree that a current 1.0 isn't just tempting fate, it's slapping a 'Kick Me' sign on yourself.07:46
Ownerhello, I've got a question about using bazaar with asp.net+IIS projects, anyone around?07:56
=== Owner is now known as bp
fullermdI dunno what the question is, but I'll bet the answer involves holy water...07:57
bp:) it's not as bad as you think, infact I kind of actually have a solution to the problem, but I wanted to know if bazaar supports the fix "officially"07:59
bpThe problem is that IIS seems to reject applications that contain directories with a "." (period) in their names. I.E. the .bzr folder08:01
bpthere seems to have been a bug report filed back in 2006: http://bazaar-vcs.org/UnderscoreControlDir   but I can't find any recent mention of it.08:02
fullermdWhat, it ignores the whole directory because there's one file like that in it??08:02
bpnot ignores, per say, just acts wierd in certain situations08:03
bprenaming .bzr to _bzr (or anything else for that matter) fixes the problem, but of course this breaks the repo.08:03
fullermdThat's fwacky.08:03
fullermdI do remember the _bzr discussion, but it's certainly not currently there.  I don't remember if it just petered out, or got voted down for some reason, though.08:04
bpyup, so basically I'm asking if bazaar supports renaming the repo folder.08:04
fullermdAFAIK, no.08:05
fullermdYou might be able to put a wrapper around all bzr commands that mv it from _bzr to .bzr, do the bzr stuff, then mv it back.08:05
fullermdOr just keep the VCS separate from the deployment.08:06
beunounder apache, he means  :p08:06
fullermdThe former has its problems, but may work.  The latter's how I do stuff, but it does mean a little more infrastructure work.08:06
fullermdWell, obviously, apache is the Right Answer  ;)08:06
bphmm, yeah those are valid solutions, but I think I might just attempt to apply the patch in the aforementioned link. Thanks for your help.08:07
fullermdWell, the actual changes are probably a lot more involved than that suggestion there.  I'm pretty sure that only gets one place of the number that look for .bzr.08:07
bpoh, well. I guessed as much, especially since that was back in 2006. Meh, no big deal, just a minor annoyance.08:09
beunobp, you could serve it through ftp08:12
beunoanonymous ftp if you need public access08:12
bpoh, no I'm not trying to make the repo public. Just trying to run the project (asp.net) in my dev environment.08:13
beunobp, well, you can serve bzr through ftp then08:14
beunoyou can push/pull through ftp08:14
* fullermd blinks.08:14
fullermdI don't think bzr access is in any way his problem...08:14
beunooh, I might of understood it wrong then08:14
fullermdIt's that for whatever stupid reason, IIS goes nutty simply because of the presence of a '.bzr' directory, even if it's never accessed.08:14
beunoaaah, sounds like fun08:15
fullermdWhich sounds insane to me.  Even for IIS.08:15
beunoI was under the impression he wanted to branch from http08:15
* beuno reads it through again and realizes it was pretty clear08:18
beunono more IRC for me after 4am08:18
bp:) thank's for trying beuno. hmm, just noticed Martin Pool (poolie) and Robert C (lifeless?) are on the channel, I think I'll just lurk for a few hours and hope they see the convo08:18
fullermdPshaw.  That's the best time   ;)08:19
fullermdWell, it's Saturday night for them.  They may check in, but...08:19
lifelessbp: I've seen that you had one, but not the content, and I'm on the way out to dinner08:19
lifelessbp: mailing the ist would be a good thing to do08:19
fullermdlifeless: asking after _bzr support08:20
fullermdDid that discussion just peter out, or did we decide against it?08:20
lifelessI think its got enough UI problems we are more against than for08:21
lifelessbut its not an impossibility; whats the reason - broken webserve or 'dont want an empty dir' ?08:21
lifelessactually, I have to go; sorry08:21
* fullermd nods.08:22
bpnp, thanks lifeless, I think I'll put this issue on the lists08:22
fullermdbp: Yeah, a summary to the list is probably the best thing.08:22
bpyup08:22
ubotuNew bug: #164810 in bzr "knit chunk should contains inside all meta information that index has" [Undecided,New] https://launchpad.net/bugs/16481008:55
ubotuNew bug: #164814 in bzr "osutils.splitpath lose info about root directory" [Undecided,New] https://launchpad.net/bugs/16481409:00
=== cfbolz_ is now known as cfbolz
ubotuNew bug: #164825 in bzr-push-and-update ""pushed 0 revisions" incorrect" [Undecided,New] https://launchpad.net/bugs/16482510:16
=== weigon__ is now known as weigon
denndaHi there. I have a problem that isn't bzr-related at all, but maybe you can help me. I heard bzr uses ElementTree a lot. Is that correct?11:05
denndaI parsed a svg-file (which is xml) with ElementTree and without altering it saved it back to the disk. the result is: http://pastebin.ca/795717 <- I want to know how to get rid of ElementTrees namespace-behaviour. It adds them automatically, which is not what I want. (It breaks the svg.) - If anyone has an Idea I would be glad if you shared it (maybe in query. Don't want to spam your channel)11:10
fullermdWell, I'm no developer or expert.  But I was under the impression that bzr didn't use ElementTree to write the XML; just to parse it for reading.  I think the writing is done manually.11:14
=== kiko is now known as kiko-phone
=== kiko-phone is now known as kiko
=== kiko is now known as kiko-afk
=== cprov-away is now known as cprov-lunch
datooh, no jam.14:08
datoah, weekend. right. :)14:08
rtoAre some of you well versed in the bzrlib?14:09
rtoOk, i'll just try asking a question.14:11
rtoThe thing is, i would like to find out which revision are "local" to a branch.14:11
rtoThat is, when a branch is branched from another branch, i would like a way to find the revision commit'ed on the new branch - rather than all the branches from both the new branch and the branch i branched from.14:12
rtoBasically i want a bzr log --local command or something similar14:13
james_wrto: bzr missing --mine-only otherbranch?14:32
rtojames_w: That seems to be somewhere along the lines of what i want, but it only lists revisions that differ between the two branches.14:35
james_wso you want all revisions that were committed to the local branch and not another branch.14:36
rtojames_w: Yep, that is basically it. Something like bzr log --local14:36
rtoAnd preferably something where you don't have to give it another branch (or more) to compare with.14:37
rtoUp til now i have been able to use branch-nick to something along the lines, so local revisions are the ones where the branch nick on a revision is the same as the branch's nick.14:38
rtoBut nick's can change, and i think there should be some way to tell revisions committed on the branch from revisions gotten from another branch.14:39
james_wrto: well what you are asking for is not that easy to define. All the information required to do this is certainly not stored by bzr.14:40
james_wthe nick is the closest you can probably get, but as you say it's not perfect.14:40
james_walso bzr log --short is kind of like that too, but again it depends on the situation.14:41
rtoHmm, it just seems like something you would want to know, for instance bzrweb could show more relevant information by only showing the revisions on the particular branch.14:44
james_wrto: well you normally define that in relation to another branch, as missing does.14:47
rtojames_w: But won't you end up with having to list all the branches that you have ever merged from in the past to show just "your changes"?14:48
abentleyrto: No, merges don't show up in the lefthand ancestry.14:49
abentleyThat is, the revisions you merge don't show up there.14:49
james_wrto: yes, it doesn't scale to many branches.14:50
abentleySo all you need to know is when your branch started.  Everything on the lefthand ancestry since your branch started is stuff you committed.14:50
rtoabentley: Ok, so i only need to list the branch i branched from14:51
abentleyYes.  Or know the revision you branched at somehow.14:51
rtoabentley: Now _that_ sounds like something bzr should be able to tell me, the point at which i branched, or whether this branch is a branch of something else in the first place.14:53
abentleyIt would make sense to store that, I agree.14:53
abentleySome of our early decisions were based on the assumption that we wanted "cp -a" to be the standard way of branching.14:54
abentleyThat's why we don't have a branch ID or a branch-point marker.  Those things are useless if copied verbatim.14:55
abentleySince cp -a is not commonly used for branching, we could revisit that.14:56
rtoabentley: Ahh, that makes sense.14:57
pygihello all16:57
pygiwould anyone be interested in writing a bzr adapter for rvcs? estimate is that it's around 500 lines of code =)17:01
james_wpygi: the Road Vehicle Certification System?17:06
pygijames_w: actually, it's not :)17:07
pygiRuby Version Control System Abstraction Layer17:07
james_wpygi: what does it take?17:08
pygijames_w: what do you mean?17:08
james_wyou need a ruby mapping to all concepts? Or just a way of getting the latest version of the code for build scripts or what?17:09
james_wand where's the code?17:09
pygijames_w: gimme a second :)17:09
pygisvn checkout http://rvcs.rubyforge.org/svn/trunk17:10
james_wah, ok. Doesn't look too hard, but I don't know ruby, sorry.17:12
pygijames_w: it's not too hard ... I'd implement it myself, but currently playing with the abstraction code itself17:12
pygiI think that's a good way to bring bzr to ruby community =)17:13
loswillioshej17:14
loswilliosI'm having some problems with bzr, seems that something isn't in its correct path: http://pastebin.com/m4eb18ec317:19
loswillioscan somebody help me with that?17:19
james_wloswillios: how did you install bzr?17:20
loswilliosaptitude install bzr17:20
loswilliosI have bzrlib in /usr/lib/python2.4/site-packages/ and /usr/share/pycentral/bzr/site-packages/, I guess I need to put that in some $PATH17:21
james_wloswillios: python --version?17:22
loswilliospython -V: Python 2.4.417:23
loswilliosurgs17:25
loswilliosVersion: 0.11-1.117:25
loswilliosI guess that's way too old right?17:25
pygiis that bzr version loswillios? :P17:25
pygiyes, that's way too old :D17:25
loswilliosdebian...17:26
loswilliosthanks james_w17:26
loswillioshmmz17:31
liwI'm doing a "bzr add" in a source tree that happens to contain sub-directories which contain .svn dirs, and bzr-svn (I assume) wants to connect to the svn server -- is this expected behavior?17:31
loswillioseven with 0.91-2~bpo40+1 from backports.org I have that problem17:31
loswilliosweird stuff17:32
loswilliosanyone another idea on that?17:40
pygiso anyone knows a bit of ruby and willing to write a bzr module? ^_^17:42
abentleyloswillios: It sounds like bzr thinks you're trying to add files to the svn checkout.17:43
abentleysorry, I meant liw17:45
james_wloswillios: can you run python and then run import sys; print sys.path17:45
liwabentley, that perhaps be a reasonable assumption, I guess17:47
liw(why people put .svn dirs in release tarballs I do not know; I only put CVS ones in by mistake, but some people do it intentionall, go figure)17:47
loswilliosjames_w: ['', '/usr/lib/python24.zip', '/usr/lib/python2.4', '/usr/lib/python2.4/plat-linux2', '/usr/lib/python2.4/lib-tk', '/usr/lib/python2.4/lib-dynload', '/usr/local/lib/python2.4/site-packages', '/usr/lib/python2.4/site-packages', '/var/lib/python-support/python2.4']17:48
abentleyliw: It means that you can immediately start working on the code from the tarball.17:48
abentleyI was actually going to propose that bzr export start doing that.17:48
liwabentley, ugh17:49
abentleyA lightweight checkout doesn't require very much space.17:49
loswilliosjames_w: bzrlib is in /usr/lib/python2.4/site-packages/17:49
abentleyAnd it means you can update to the latest code if you want.17:49
abentleyOr convert it to a branch with "bzr reconfigure --tree"17:50
james_wloswillios: so can you do 'import bzrlib'?17:50
liwabentley, as long as you make it optional, I won't mind17:50
loswilliosjames_w: nope, ImportError: No module named bzrlib17:50
liwif I want a branch, I'll make a branch; I use "bzr export" to make release tarballs, and I don't want any vcs metadata there17:50
james_wloswillios: weird.17:52
=== me_too is now known as too_short
=== too_short is now known as me_too
loswilliosindeed17:53
abentleyliw: You aren't the person who will be using the tarball.18:11
liwabentley, yes I am18:11
abentleyReally?  No one else uses your software?18:12
liwthere are other users (at least sometimes), but that doesn't mean I don't use the tarball myself18:12
abentleyWell, I am thinking of software that has many users who are not the author.18:14
liwabentley, I am not opposed to having an option to include .bzr/ in the export, if the user wants it18:15
liwabentley, but if you don't retain the current behavior, then I think you're making a mistake18:15
abentleyAll I'm talking about is bringing it up for discussion.18:16
abentleyBut I think that it's in-line with open-source ideals to make it easy to start hacking on the software you've just downloaded.18:17
liwand therefore you want to make it difficult to not include the .bzr stuff even for people who don18:18
liw't want it?18:18
abentleyliw: I think that it might be nice to make it a lightweight checkout by default.18:19
abentleyBecause I think that in the common case, the author will not be the tarball user.18:19
liwand I think having any vcs metadata in a release tarball is bad, since it messes up with the downloader's ability to set up their own version control18:21
abentleyCould you expand on that?18:23
liwI've never had any use for any .svn, .bzr, .git or other such metadata included in a release tarball; rather, they interfere with things when I want to import the sources into a new branch I've created myself18:25
liwor prevent me from creating such a branch before I clean up18:25
liwif I want to branch off upstream's branch, I already have tools for that; putting extra stuff in tarballs does not give me new functionality, but does make more work for me18:26
liw(further, not all bzr branches are for software projects, of course; the .bzr stuff is of zero use for, say, when I create a zip of a document project to send to my father)18:29
=== me_too is now known as too_short
=== too_short is now known as me_too
=== me_too is now known as too_short
=== too_short is now known as me_too
=== abadger1999 is now known as abadger_afk
=== cfbolz__ is now known as cfbolz
=== cfbolz_ is now known as cfbolz
=== cfbolz is now known as cfbolz_____
nekohayohello, is there any reason for my 3 bugs with patches (for example, https://bugs.launchpad.net/bzr-gtk/+bug/151824 ) still lying around, in an unconfirmed state, after one month?22:13
ubotuLaunchpad bug 151824 in olive "use single click for bookmarks" [Undecided,New]22:13
nekohayodid I file them on the wrong components?22:13
pygihello22:13
pygianyone with at least little ruby knowledge willing to write bzr adapter for rvcs??22:13
beunonekohayo, let me take a look. It might be due to 1.0 release being so close by22:25
beunoa lot of attention is diverted toward that22:25
beunonekohayo, seems filed correctly, maybe you could email the list with the bug numbers so they can be looked at?22:26
pygianyone? :P23:04
nekohayobeuno: wha, 1.0 already? of bzr or bzr-gtk you mean?23:18
jelmernekohayo: manpower mostly23:19
beunonekohayo, bzr23:19
jelmernekohayo: there's only a couple of us working on bzr-gtk, and we're quite busy with other things as well23:19

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