/srv/irclogs.ubuntu.com/2008/03/07/#bzr.txt

Odd_BlokeLeoNerd: I'm going to bed now, if you want to discuss this in more detail, drop me an email at D.M.Watkins@warwick.ac.uk (or email the list). :)00:00
Odd_BlokeNight all!00:00
mwhudson_damn people in london and their sleeping habits00:05
=== mwhudson_ is now known as mwhudson
dashHi. someone remind me how to roll a working copy back to a previous revision? I seem to have svn brain today, I expected 'bzr up -r ...' to do it :)00:25
LeoNerdbzr revert -r ...00:25
dashhah, ok00:25
LeoNerdYou're thinking in CVS-mode :)00:26
=== mw is now known as mw|out
alwynHi everyone06:53
alwynAnyone have success using bzr-svn against a svn repository with authentication?  The passwords are cached by my subversion client, but no joy on both gentoo and mac os x...06:54
* mwhudson_ has been amusing himself this evening: http://people.ubuntu.com/~mwh/really_hacked_up_changelog_view.png 07:44
=== mwhudson_ is now known as mwhudson
=== abentle1 is now known as abentley
asabilanyone knows if there is a way to have bzr behave in a similar way to git ? concerning branches and repositories ?08:37
asabilI mean 1 folder contains many branches08:37
PengNot yet.08:37
PengMaybe in the future.08:37
asabilans is there a way to rewrite the history ?08:47
luksrewrite in what way?08:52
igcmorning08:53
asabilluks: change the commit log08:53
asabilor change the committed files08:54
luksuncommit + commit?08:54
asabilluks: something like : http://www.kernel.org/pub/software/scm/git/docs/git-filter-branch.html08:57
asabilluks: maybe you would want to take a look at this: http://blogs.gnome.org/newren/2008/03/01/happenings-in-the-vcs-world/08:58
asabilit is quite biased toward git, but still contains interesting ideas and complains about bzr08:58
luksumm, I think I really don't want to :)08:58
luksno, bzr doesn't have anything like git-filter-branch, but I think that's good, not bad08:59
asabilluks: that's exactly his complain: the bzr devs don't want to listen:p08:59
luksI'm not a bzr dev, btw08:59
luksbut I wouldn't listen anyway :)09:00
asabilluks: and honestly git-filter-branch can be really useful09:00
lukshow so?09:00
luksisn't VCS about keeping your history?09:00
luksI can understand it's use for migrating or restructuring branches09:00
luksbut not to change commit message or things like you mentioned09:01
asabilyes, but let me give you some use cases09:01
asabillet's say I work for a company A, that uses bzr09:01
fullermdThere are many very good reasons to want to wipe out history   :)09:01
asabilI worked for a long time on a proprietay project for them , and some day they decided it should go open source09:02
asabilhowever, there are some files early in the history of the bzr branch, that are confidential09:02
asabilI still want to pusblish the whole branch with its history09:02
asabilhow do I do ?09:02
luksyes, that's one valid use case09:02
luksbut the first two you mentioned were not09:03
asabilwhich ones ?09:03
luks<asabil> luks: change the commit log09:03
luks<asabil> or change the committed files09:03
asabilluks: sometimes there are typos in the commit log09:03
lukstypo in the commit message is IMO part of the history and should stay there09:04
asabiland sometimes there is a morron in your project who would put crappy commit logs09:04
luksotherwise you break the branch for everybody else09:04
asabiland you want to fix that09:04
luksit's part of the history, don't change the history :)09:04
fullermdDown, boy!  Keep your dogma on its leash!09:05
asabil:)09:05
luksI guess I just don't like git too much, so I'm not the right person do discuss this :)09:06
asabilluks: I don't like git at all, and for good reasons09:06
fullermdIn a lot of cases, breaking the branch is totally acceptable.  In a lot of cases, I know exactly where every copy of the branch is, so remaking them all is a bloody inconvenience, but entirely possible.09:06
asabilbut many people like git, and if you really like bzr, you should listen to their critics and improve09:06
luksI'm not saying filtering shouldn't be possible09:07
luksbut in git it' a common operation09:07
luks*it's09:07
luksbtw, I don't like bzr either :)09:08
asabil...09:08
luksbut it's the only acceptable one09:08
asabilI think there are many things to improve in bzr, maybe also change its name09:08
fullermd"All mail clients suck.  This one just sucks less."09:09
asabilmany people still confuse bzr with bazaar09:09
Odd_Blokeasabil: That's going to be fixed soon.09:10
asabiloh really ?09:11
Odd_BlokeThe Debian 'bazaar' package will install both bzr and baz.09:11
Odd_BlokeAnd eventually just bzr.09:11
asabilthat would be better09:11
PengTypos are one thing, but sometimes I really do want to change the commit message when it's wrong.09:11
johnnyversion the changelogs lol09:12
johnnyerr commit messages09:12
PengThat could work.09:13
Odd_Blokejohnny: The commit messages _are_ versioned, with the content of the trees.09:13
PengIt would, of course, be too complicated, but would be useful.09:13
johnnyi mean seperately Odd_Bloke09:14
johnnybut i personally don't need such a thing..09:14
johnnyyou could find a way to replay the history tho09:14
luksPeng: then uncommit it and commit again09:14
luksif you notice it later and the branch is already published, rewriting is IMO the wrong thing to do09:15
Verterokmorning09:16
asabilI still think that rewriting the history and filtering a branch can be a really neat fature for bzr09:18
asabilalso bzr could improve its handling of Limbo files09:18
asabilmaybe by adopting an interactive ui like darcs09:18
asabilor having an ibzr command that is fully interactive09:19
Pengluks: The only time I can remember it bothering me was with a personal, non-code branch that isn't published. There had also been commits since it.09:34
LarstiQasabil: handling of Limbo files?09:44
asabilLarstiQ: unknown files09:46
asabilLarstiQ: you know the "Oh sh*t i forgot to add those files before committing"09:47
LarstiQasabil: ah. You confused me since limbo has a specific meaning in bzr, and I have never heard of this use before.\09:49
LarstiQjam: are you running bzr serve or somesuch?09:49
datoLarstiQ: http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/, presumably09:50
Odd_Blokeasabil: There's 'bzr shell', in bzrtools I think.09:51
jamLarstiQ: I am now :)09:51
asabilOdd_Bloke: and ?09:51
asabilOdd_Bloke: ever used darcs ?09:52
LarstiQasabil: maybe you are looking for commit --strict?09:52
asabilLarstiQ: rather something like darcs record09:53
Odd_Blokeasabil: 'bzr shell' is similar to an interactive bzr command, but I should have read more context.09:53
Odd_Blokeasabil: Didn't you write the 'record' plugin?09:54
asabilOdd_Bloke: yes, the record plugin should be renamed btw into bzr make-patch09:54
asabilor something like that09:54
Odd_Blokeasabil: Ah, yes, I recall now.09:54
asabilI wanted it to become like darcs record, but never had time to finish it09:54
* LarstiQ is a bit low on cycles09:55
Odd_BlokeI'd have thought it would be fairly trivial to switch from creating a patch to creating a commit.09:55
LarstiQasabil: I don't know darcs well, what about record is it that you are looking for?09:55
* Odd_Bloke doesn't have any motivation to do it himself, because he wouldn't ever use it.09:55
Odd_BlokeI'd be happy to help someone who does have some motivation though.09:55
asabilLarstiQ: basically darcs is fully interactive, so when you issue a darcs record (the commit equivalent), it will prompt you about the hunks to record09:58
asabil(in a similar manner to bzr shelve), then it will ask you for a record name (commit message), before finishing the record09:59
asabiliirc it will also warn you about unknown files09:59
LarstiQasabil: right, which --strict will also do09:59
asabilLarstiQ: --strict is interactive ?10:00
LarstiQasabil: no10:00
asabilOdd_Bloke: maybe can we cook up something together ?10:00
asabilOdd_Bloke: I want to make the record plugin add a --interactive command to commit10:00
* LarstiQ doesn't think interactivity is desirable in the general case, so if there are useful features of record that can also work non-interactive, that would be nice to factor out10:01
asabils/command/option/10:01
LarstiQor rephrased, not only the interactive users should benefit10:01
asabilmake the interactive users benefit until you find a way to make the non interactive users benefit as well10:02
LarstiQfair enough :)10:02
jelmerdudes!10:02
LarstiQjelmer!10:02
jelmerhow is the sprint today?10:02
LarstiQjelmer: they just did a firealarm test10:03
LarstiQbut announced it so we could just sit and not do the entire evac thing10:03
Odd_Blokeasabil: There was someone in here last night who was also interested in that exact thing.10:04
Odd_BlokeIt was, in fact, LeoNerd (implicit ping). :)10:05
jelmerLarstiQ: ah, that's nice for a change10:14
vilayeah, there were nice enough when I explained why *I* prefer it to be a lighter test...10:15
LarstiQjelmer: do you have anything to say on file copies/path tokens?10:17
jelmerLarstiQ: I'm not sure I entirely remember how they were supposed to work10:18
LarstiQah :)10:18
Odd_Blokejelmer: abentley and lifeless are probably about to have a discussion regarding file copies, hence the question. :)10:19
spivOdd_Bloke: Any chance of you adding a "mirror-update" command to that plugin today? :)10:22
jelmerOdd_Bloke/LarstIq: The wiki page on path tokens doesn't appear to have any actual design details, just use cases :-(10:23
Odd_Blokespiv: That's my next step. :D10:23
LarstiQjelmer: rob's comment was that pathtokens may be a implementation strategy for filecopies, but they don't need to be the way to do it.10:23
spivOdd_Bloke: I have a use-case of a sort10:24
jelmerLarstiQ: makes sense10:24
Odd_Blokespiv: Cool.10:25
spivOdd_Bloke: I have just made a "bzr svn-import" of Twisted's SVN on my laptop, and when I update it (by running bzr svn-import again in the same dir), I'd like a way to update the copy of the import I've put at http://people.ubuntu.com/~andrew/twisted/10:25
spivOdd_Bloke: rsync works atm, but isn't exactly ideal...10:25
asabilOdd_Bloke: let's say I have a patch, how do I commit it (working on --interactive)10:50
jelmertoo bad there's no video feeds10:50
jelmerbut at least there is the BrainStorms page10:50
Odd_Blokeasabil: Well, apply it to the tree and then commit as normal, I guess.10:51
LarstiQjelmer: we could do a conference call I suppose10:51
asabiloki10:52
Odd_Blokespiv: I've just pushed an initial implementation of mirror-update to mirror.dev.10:52
spivOdd_Bloke: thanks!10:53
jelmerLarstiQ: it's being discussed right now?10:58
Odd_Blokejelmer: Nope, plugins ATM.10:59
Odd_Blokespiv: So I'm finding some slightly odd behaviour in that the mirror-update command is pulling the revisions and the history into the branches appropriately but 'bzr log' complains about logging the null revision.11:00
Odd_Bloke'bzr log -r1' shows the right thing, however.11:00
spivOdd_Bloke: hmm, mirroring a rich-root-pack repo doesn't work11:01
spivIt creates pack-0.92 at the destination then barfs.11:01
jelmerhmm, is bundle-buggy down or just being slow11:02
james_wOdd_Bloke: are you using tree.pull() or branch.pull()11:03
Odd_BlokeI get "bzr: ERROR: Repository KnitPackRepository('file:///var/tmp/foo/.bzr/repository/') is not compatible with repository KnitPackRepository('file:///var/tmp/blah/.bzr/repository/')" when doing that.11:03
Odd_Blokespiv: ^11:03
Odd_Blokejames_w: branch.pull()11:03
spivOdd_Bloke: right.11:03
james_wOdd_Bloke: you want tree.pull() anyway.11:03
Odd_Blokejelmer: abentley was using it earlier, but I don't know about ATM.11:03
Odd_Blokejames_w: How's that going to cope with branches without trees, which is probably what we want from this (as the eventual use-case is switching into the mirrored branches)?11:04
james_wOdd_Bloke: well, use open_tree_or_branch() and then you know whether there is a tree there already, and then switch between tree.pull() and branch.pull()11:05
james_wOdd_Bloke: and I guess you have to solve that branch.pull() problem.11:06
james_wI can show you some code that does the former if you like.11:06
Odd_Blokejames_w: Sure, please. :)11:06
Odd_Blokejelmer: BundleBuggy WFM ATM TLA.11:06
jelmerOdd_Bloke: hmm, looks like it's just being very slow to me11:07
Odd_Blokespiv: Yeah, it'll be fun. :p11:08
jelmerPeng: Do you have pqm access or should I submit the branch for "http://bundlebuggy.aaronbentley.com/request/%3C47C5D9C0.5080702@mattnordhoff.com%3E" ?11:11
jameshjelmer: the debian/watch files for bzr-email and bzr-pqm in Debian seem to be pointing at the wrong LP pages11:13
jameshaccording to http://qa.debian.org/developer.php?login=pkg-bazaar-maint@lists.alioth.debian.org11:13
jelmerjamesh: as far as I know there is no watch file for bzr-email and bzr-pqm11:14
jameshjelmer: maybe that site creates automatic watches then.  My mistake :)11:15
jelmerjamesh: I wonder how though11:16
jelmersince it seems to be using launchpad urls11:17
datoigc: ayt?11:17
LarstiQheya dato11:17
datohey LarstiQ11:17
jameshjelmer: the home page from the package metadata11:18
ubotuNew bug: #199440 in bzr "Traceback if authentication.conf contains section without password " [Undecided,New] https://launchpad.net/bugs/19944011:20
igchi dato11:22
jelmerjamesh: and it then makes assumptions about the releases being on launchpad as well?11:22
igcdato: spoke to abentley re the best way to do fast-export ...11:22
datoaha11:22
igctwo revision-trees and changes_from is fine11:22
igcabentley also mentioned iter_changes11:23
igcand had some idea down the track11:23
datoI saw iter_changes, but only as a _underscore_method, so I went with public stuff11:24
abentleychanges_from is a bit ugly and lossy.11:25
igcs/idea/ideas/11:25
igcre more efficient ways11:25
igcI imagine that two rev-trees + changes_from is fast enough for any projects though ...11:25
igcso we ought to stick with that initially I feel11:25
igcs/any/many/11:25
datoabentley: lossy?11:25
LarstiQOdd_Bloke: one of the log messages of bisect we looked at mentioned working with merged revisions, what was that about then?11:25
ubotuNew bug: #199442 in bzr "every command twice encode sys.stdout?" [Undecided,New] https://launchpad.net/bugs/19944211:26
Odd_BlokeLarstiQ: I dunno, I thought it coped with them too.11:26
Odd_BlokeI may have screwed the test up. ^_^11:26
datoigc: now, I have bzr-crude-export.py in a state that does everything I'd need, so I'm not sure how much more time I'd like to put into it. otoh, I'd like to release it. so, how about I drop a mail about it in bazaar@ and git@, and mention that it'll probably need a new home in the future, possibly bzr-fastimport?11:26
abentleydato: There are some combinations with kind changes that aren't recorded properly by changes_from.11:27
datoabentley: ok. my script doesn't look at kind_changes atm, so...11:27
igcdato: I definitely want it in fastimport and I'm happy to own it11:28
datoabentley: btw kind_changed seems missing in TreeDelta's __doc__11:29
Odd_Blokespiv: I'm not going to work that mirror bug out now, as I think the git-style branches discussion will have some bearing on it.11:30
datoigc: ok. shall I keep calling it -crude, or would be -fast ok with you? I'll mention in __doc__ that later on it'll live in launchpad.net/bzr-fastimport11:30
igcdato: fast is ok. If you send me the code or link, I'll include it real soon now11:32
igcbbiab11:32
asabilOdd_Bloke: ok, done I will publish my interactive commit crack11:33
Odd_Blokeasabil: Cool. :)11:34
asabilOdd_Bloke: any name to suggest ?11:35
asabilbzr_interactive ?11:35
Odd_Blokeasabil: interactive-commit, or somesuch.11:36
asabilOdd_Bloke: it also contains a record-patch command11:36
Odd_Blokeasabil: I dunno then.11:36
Odd_BlokeIt can always be changed later. :p11:36
asabilabentley: what do you think about adding the record-patch command to bzrtools ?11:44
beunoLarstiQ, *ahem*debconf*ahem*11:45
abentleyI'm not keen on making it easy for people to do partial commits.11:47
LeoNerdIt's already easy... bzr ci some files here11:53
LeoNerdThe hunk selection only extends a bit further, to do changes -within- files.. It's really no worse in principle11:53
LeoNerdIt just depends on the (fairly-arbitrary) way that the user has split their content across files11:54
abentleyLeoNerd: Different-in-degree is good enough for me.11:57
asabilabentley: oO record-patch is about creating patches manageable with quilt11:59
LarstiQbeuno: yes, thank you :)12:02
abentleyasabil: You should really look into looms, then.12:12
abentleyThat should be a much more reliable way of generating quilt-compatible patches.12:13
datoigc: ok, I CCed you. I'm sorry is a git repo, but feel free to use git-fast-export! :-P12:14
asabiloh, oki didn't know aboutit12:14
asabilthanks12:14
spivOdd_Bloke: makes sense12:20
Odd_Blokespiv: That's what I thought.  However, it seems like the discussion may not happen after all...12:21
spivOdd_Bloke: we'll see...12:23
Odd_Bloke*ominous music*12:23
jmlhow do I list all commands provided by a plug in?12:43
spivjml: "bzr help commands | grep PLUGIN-NAME12:47
beunojml, or use lifeless' plugin-info plugin: https://launchpad.net/bzr-plugin-info12:47
jmlspiv: doesn't work12:48
jml<-- still not stupid12:48
Odd_Blokeabentley: http://pastebin.slackadelic.com/426 is a patch that removes a couple of Python-2.5isms from your LP directory service patch.12:48
jmlspiv: because help wraps12:48
abentleyOdd_Bloke: tx12:52
Odd_Blokeabentley: NP.12:52
spivjml: "grep -1 PLUGIN_NAME" ;)12:56
=== mrevell is now known as mrevell-lunch
jmlspiv: wrath12:57
jmlspiv: and violence.12:59
LarstiQjml: :)13:08
jmlspiv: (it doesn't always wrap)13:12
=== mw|out is now known as mw
=== FreeNode is now known as herb
=== mrevell-lunch is now known as mrevell
VSpikei'm using bzr to control a web application.  My hosting co only gives me access to the webspace by ftp.  Can I use bzr to push changes up to the hosting space?13:56
james_wVSpike: yes. You can push over ftp.13:57
VSpikei tried "bzr init ftp://user:pass@hosting.com" but I get an error "File exists: '/.bzr': 550 /. bzr: access is denied"13:57
james_wVSpike: that doesn't look good.13:57
VSpikeI'm using bzr 1.2.0 on Windows13:57
james_wVSpike: also, bzr will not create a working tree on the ftp server, so I don't think it will do what you want.13:58
beunoVSpike, and also, it will have the bzr repository online, making it easy to download the source13:59
beunoVSpike, we are working an a plugin to tackle the website-uploading workflow13:59
beunoso keep an eye out for it  ;)13:59
VSpikebueno the whole source is always in the webspace anyway, and I can specifically block access to .bzr14:00
VSpikeHmm I don't think even rsync can do ftp14:02
VSpikerats14:02
beunoVSpike, right, well, you still have the working tree issue, which is the actualy files14:03
beunobut, again, a solution to that is very close by14:03
beunojust not today14:03
VSpikeI'm not sure i understand the working tree issue?14:04
beunoVSpike, the working tree is the actual files14:04
hmelandThe "bzr push" will only push the repository part, i.e. stuff under ".bzr".14:04
beunoso when you push, you just send the repository files, which are the ones under .bzr14:04
VSpikeah right14:05
beunoVSpike, so you would need ssh access on the other end to do this now14:05
beuno(if you're willing to block the .bzr directory via htaccess or something)14:05
VSpikeUnderstood14:06
VSpikeTimestamps never seem to match either so probably the easiest/safest thing to do it use a sledgehammer and just write a script to ftp everything up14:06
beunoVSpike, right, at the moment14:07
VSpikeokay.  thanks!14:07
hmelandAny ftp mirroring tool should be able to do the job.14:07
beuno(I'm personally doing php magic to do it on top of bzr, but that's not available yet either :p)14:07
awilkinsIs this stuff convered in the FAQ? Because it's sure as heck a FAQ in here.14:07
awilkinsOh, and while we're talking about timestamps, was it a conscious design decision for timestamps in working trees not to match the timestamp of the files last revision?14:08
=== thumper_laptop is now known as thumper
beunoawilkins, we're aiming at solving the actual problem, so I suppose a FAQ wouldn't make any sense at the moment14:09
hmelandI seem to recall there being a bug on the file-timestamp issue; don't know the status, though.14:09
LarstiQawilkins: What are they instead and what case are you thinking of?14:10
awilkinsI find the timestamp thing annoying because my tree compare tool is time sensitive14:10
awilkinsLarstiQ: The timestamps are "time of writing to disk" from the build phase14:10
LarstiQawilkins: ah right, so clean checkout you mean.14:10
awilkinsLarstiQ: Yup14:11
LarstiQawilkins: Doesn't sound like a desing decision, but rather a bug I think.14:11
awilkinsIt feels like a bug to me :-)14:11
awilkinsALthough AFAIR it's optional behaviour in TSVN... I'll look at the settings dialogue14:11
hmelandOf course, in a distributed system, using the "commit time" for file timestamps could lead to clock skew related problems.14:11
awilkinsYech, there is that14:12
awilkinsOk, it's optional default to "off" in TSVN14:12
awilkinsBut I find myself comparing local file trees a lot more with bzr because branching is much more frequent14:13
hmelandOut of interest, which timestamp-sensitive tree compare tool are you using?14:14
awilkinsBeyond Compare 214:14
LarstiQawilkins: any chance that could use bzrlib?14:15
awilkinsIt has a rules-based compare as well, which gives more accurate results, but it still marks the "newer" file if they are different14:15
lifelessawilkins: timestamps are not set to time of last revision deliberately14:15
lifelessawilkins: /away14:15
awilkinsWould you be against an option to do so?14:16
lifelessawilkins: without a good reason, yes. :). We do have an open bug that the times should all be identical14:16
awilkinsIs this because it's expensive to determine the last revision of individual files?14:18
awilkinsOh, anyone who does BzrGtk here?14:19
hmelandI've never used "Beyond Compare", but according to http://diablopup.blogspot.com/2007/07/product-recommendation-beyond-compare.html it would seem that it has an "ignore timestamp differences" option?14:19
beunoawilkins, jelmer and phanatic14:19
phanaticwhat did i wrong this time? :)14:20
beunophanatic, bzr-gtk, apparently  :p14:20
awilkinsphanatic: Nothing, I just have a suggestion to make tool shelling a bit more flexible ; I'm currently operating with a very crude custom hack so I can use my 3-way merger of choice14:21
awilkinsUsing subprocess with a list of args isn't compatible because it uses an arg format that isn't MFC flavoured14:22
awilkinsI've just hacked in a custom line for the tool I'm using using a % format string, but that obviously isn't portable to anyone else14:22
awilkinshmeland: Thanks for that, it's nice to learn new things about your favourite tools (even thought they are blindingly obvious)14:24
hmelandawilkins: No problem, glad to help :-)14:24
awilkinsI just wish it did 3-way merge (have to wait for v3 to do that....)14:26
lifelessawilkins: several reasons.14:41
lifelessawilkins: one is that different machines have datestamp skew14:42
lifelessawilkins: another is that datestamp of commit is not sufficient to ensure correctness for build systems it in fact makes things worse than what we do today)14:42
awilkinsYes, I can see that too ; different platforms with different timezone behaviours too.14:42
lifelessawilkins: that too14:43
* awilkins curses Win32 for having stupid TZ behaviour and has for many years14:43
awilkinsDo the dates in bzr try to be UTC internally, btw?14:44
luksawilkins: they are in local time + timezone14:46
lifelessNg: want to stab loggerhead btw?14:46
Nglifeless: sure14:46
lifelessawilkins: anyhow, what would settingthe datestamp for you buy you?14:47
Nglifeless: done14:47
awilkinslifeless: the red "newer" highlights in my compare tool would be on the right side14:47
lifelessawilkins: ah; so the tool is broken ? :>14:59
LarstiQwell, if it's just a general tree compare tool, where else would it get the information from?14:59
lifelessLarstiQ: newer from datestamps is inherently meangingless in the presence of diff & patch14:59
lifelessawilkins: a plugin to 'reset' a tree timestamp state should be easy enough to write15:00
awilkinslifeless: I thought the same myself15:00
awilkinslifeless: "newer" has value if you are (e.g.) uploading content to an FTP with limited bandwidth and youre target FTP server does not support CRC15:01
ubotuNew bug: #103199 in bzr-gtk "diff window should not block typing in gcommit" [Medium,Fix released] https://launchpad.net/bugs/10319915:06
Odd_Blokeigc: I'm pretty happy with the 'mirror' command available at https://code.edge.launchpad.net/~daniel-thewatkins/bzr-mirror/mirror.dev .  If you want to have a play around with it and give me some feedback, that'd be good. :)15:18
ubotuNew bug: #118461 in bzr-gtk "symlinks not handled correctly" [Low,Confirmed] https://launchpad.net/bugs/11846115:36
ubotuNew bug: #120524 in bzr-gtk "Fold arrow in commit dialog serves no purpose" [Low,Fix released] https://launchpad.net/bugs/12052415:36
ubotuNew bug: #120526 in bzr-gtk "Not possible to specify an external diff viewer to use" [Low,Confirmed] https://launchpad.net/bugs/12052615:42
beunoVerterok, https://bugs.launchpad.net/bzr/+bug/8665216:28
ubotuLaunchpad bug 86652 in bzr "commit option to set revision properties" [Undecided,New]16:28
beunowould your patch fix that?16:29
jelmerPeng: ping16:36
jelmerbeuno: No, that only supports registering custom revision property displayers if we're talking about the same patch16:38
Verterokbeuno: nope, I'm working on showing the properties16:38
beunojelmer, right, but that would be a start, wouldn't it?16:38
Verterokjelmer: thanks ;)16:38
* beuno pokes ponders asigning the bug to Verterok16:39
* Verterok hides, and run away :)16:40
* beuno stops bug triaging and goes back to coding16:40
Verterokbeuno: but I can put some time in adding the option, but I need someone to review my code16:41
ubotuNew bug: #131471 in bzr-gtk "bzr gconflicts outputs error messages when pushing green button and no files are in conflict" [Low,Fix committed] https://launchpad.net/bugs/13147116:41
ubotuNew bug: #199513 in olive "gcommit crashes: ERROR: dbus.exceptions.DBusException" [Undecided,New] https://launchpad.net/bugs/19951316:41
beunoVerterok, right, just kidding (not really)16:41
jelmerbeuno: I actually think #86652 would be a bad idea16:42
jelmerbeuno: I've just commented16:42
Verterokok16:42
ubotuNew bug: #147011 in bzr-gtk "[viz] date format is too verbose" [Wishlist,Fix released] https://launchpad.net/bugs/14701116:46
fredreichbierhello16:53
phanaticsquashing bugs is fun16:54
fredreichbieri have a problem: i and another developer use launchpad+bazaar for the code. now he did some changes and pushed them as revision 11. i did some changes, too, and so i committed by own changes in my own branch and then merged his and my changes together. but after pushing it to launchpad, his revision disappears and is replaced by my revision. what's wrong? :)16:55
=== mrevell is now known as mrevell-post-box
LarstiQfredreichbier: It didn't disappear but is merged in, have a lookg at bzr log, it should show his commit indented16:56
fredreichbierright, but it disappeared in launchpad16:57
beunofredreichbier, well, launchpad doesn't show sub-revisions16:58
beunoyou will probably see it with "code browse"16:58
beunojml, ^   :p16:58
fredreichbierah16:58
fredreichbierah yes in 'browse code' it's visible. thank you :)16:59
beunofredreichbier, welcome16:59
beunojam, https://code.launchpad.net/landscape  == no code. I'm I looking in the wrong place?17:07
=== mrevell-post-box is now known as mrevell
ubotuNew bug: #199527 in bzr-gtk "viz tracebacks when refreshing on an uncommitted and recommitted branch" [Undecided,New] https://launchpad.net/bugs/19952717:21
trohow would i uninstall bzr 1.2 if i installed it from source (ie. python setup.py install) into the default directory?17:23
Verteroktro: in what OS?17:24
trolinux17:24
Verteroktro: as root?17:25
troi have bzrlib in /usr/lib/python.2.5/site-packages17:25
troya17:25
Verterokrm bzrlib, and the bzr executable17:25
Verterokthat's all17:25
troo ok. thanks17:25
phanatictro: manually remove the bzrlib directory there + bzr executable + bzr manpage -- that's pretty much all i guess17:25
beunoOdd_Bloke, would you take a look at: https://bugs.edge.launchpad.net/bzr/+bug/19051217:25
Verterokphanatic: thanks17:25
ubotuNew bug: #199539 in bzr "plugins command crashes complaining about "verbose"" [Undecided,New] https://launchpad.net/bugs/19953917:25
ubotuLaunchpad bug 190512 in bzr ""bzr: ERROR: No such file" on bzr push" [Undecided,New]17:25
beunoit might be realted to the bug you just posted17:25
troman, you'd think setup.py would be able to record where it installed the files and have an "uninstall" command or something17:26
Odd_Bloketro: That's what your package manager is for. :)17:29
troOdd_Bloke: yeah, i guess. i wonder if checkinstall can handle setup.py installations17:30
fredreichbiertro: as far i know it can17:31
awilkinsAnyone know if the capability for symlink on Vista (NTFS 6) has shown up in CPython?17:38
Odd_BlokeVerterok: beuno: ISTR https://bugs.edge.launchpad.net/bzr/+bug/199539 being something to do with xmloutput.  Am I just inventing that?17:38
ubotuLaunchpad bug 199539 in bzr "plugins command crashes complaining about "verbose"" [Undecided,New]17:38
beunoOdd_Bloke, I'd like to go for the "inventing" bit, but I can change my mind if you point me to what makes you think that17:40
VerterokOdd_Bloke: thanks, the xmloutput it's fixed, maybe it's a old version?17:41
* awilkins discovers that symlinks for Vista have not shown up yet17:41
batomsi had asked this a while back but is it possible with bazaar to branch from one remote tree to another remote tree without going through the local machine18:12
batomse.g. can i branch from a launchpad tree to another launchpad tree without first downloading the first tree and reuploading it18:12
awilkinsbatoms: You could probably do it if you had ssh access to the target machine, but otherwise, no18:13
awilkinsHow does run the tests from inside a python console?18:15
* awilkins is testing IronPython to see how lacking it is to run BZR18:15
LarstiQawilkins: os.system('./bzr selftest')? ;)18:15
LarstiQawilkins: I think we looked at IronPython yesterday and got a bit scared at the differences.18:16
awilkinsHeh, I don't think the win32 "os" has system18:16
LarstiQawilkins: you should be able to do the same thing as cmd_selftest from bzrlib/builtins to kick off the testsuite.18:16
LarstiQawilkins: I'm rather sure it does.18:16
awilkinsI tried import bzrlib18:17
awilkinsthen bzrlib.test_suite()18:17
awilkinsRan into an error :-18:17
awilkinsAh, I need to chdir to where bzr is18:17
awilkinsLarstiQ: Maybe the IPY implementation of "os" has no .system() call.18:18
LarstiQawilkins: from bzrlib.tests import selftest; selftest()18:18
LarstiQawilkins: hmm, could be18:18
awilkinsNo module named fcntl18:19
LarstiQhttp://www.codeplex.com/IronPython/Wiki/View.aspx?title=Differences&referringTitle=Home iir18:19
LarstiQawilkins: oh bother18:19
LarstiQawilkins: you have some work cut out for you18:19
awilkinsI'm testing with IP 2.08 A which is a bit closer18:19
awilkinsIt's mostly idle interest and the promise of 1.8x performance (which is probably not true beacuse the performance critical stuff should be C)18:20
* LarstiQ wasn't aware there was a 2.x line?18:20
* awilkins didn't bother with the 1.1 line18:20
awilkinsOne problem is that it completly foobars most of your platform capability detection by reporting that sys.platform == 'cli'18:21
awilkinsMost of your code uses osutils which could be patched, but most of the tests directly check sys.platform18:22
awilkinsAnd of course, sys.platform is still 'cli' even when you run IPY on Mono/Linux18:22
* awilkins has a bzr.iron branch with comments on each platform check.18:23
LeoNerdhttp://changelog.complete.org/posts/698-If-Version-Control-Systems-were-Airlines.html   <== Quite favourable to bzr, I think :)18:23
awilkinsHeh, that's the third time I've seen that here today18:23
awilkinsHeh, this first error is a problem with the CPython platform detection ; subprocess thinks it's running on something posix18:26
awilkinsWhere's the msvcrt module? Builtin?18:37
fredreichbiermsvcrt is a builtin module18:37
* awilkins has now discovered this18:38
awilkinsDarn18:38
lifelesstchau, see you on the flipside18:41
batomscould anyone ssh to launchpad for me and branch a tree for me19:02
batomsmy connection is too slow19:02
batomsi've been waiting on a branch for hours19:02
=== statik` is now known as statik
phanaticbeuno: http://www.ubuntu.com/news/landscape19:24
=== BasicOSX is now known as Starving
beunophanatic, yeah, but jam mentioned the client was open source19:45
phanaticbeuno: it seem it's "just free"19:45
phanaticfor my fellow "beer hackers": http://xkcd.com/323/19:59
LarstiQOdd_Bloke: http://thedailywtf.com/Articles/SQL-Sentences.aspx20:01
Odd_BlokeLarstiQ: A true WTF.20:06
Vantagehi all, is bzr cvsps import working with bzr 1.1/1.2?20:12
Vantagei know it worked with 0.18 and broke with 0.96.  I'm just wondering what the current status is20:13
* LarstiQ doesn't know, but it should obviously20:14
Stavros1hello20:17
Stavros1i would like a setup where i review changes made by people before committing to the master repo, how would i do that with bzr?20:18
LarstiQStavros1: you can do that in a multiple of ways. bzr.dev development does that with a combination of pqm and bundlebuggy20:20
Stavros1hmm, i don't know what any of those are20:20
LarstiQ:)20:20
LarstiQStavros1: bzr.dev can only be committed to by pqm, which is a bot that you send a merge requested, it merges your branch when it gets to it in it's queue and  enforces policy (in our case, it runs the testsuite)20:21
LarstiQStavros1: if it succeeds, it commits, otherwise, it rejects the merge20:22
Stavros1oh hey, that's great20:22
Stavros1does it notify you as well?20:22
LarstiQStavros1: bundlebuggy is a tool to help the review process, you send bundles to a mailing list, and it tracks if it's merged or not20:22
LarstiQStavros1: yes20:22
Stavros1awesome20:22
LarstiQStavros1: see http://bundlebuggy.aaronbentley.com/20:22
Stavros1hmm i just looked at that, but i don't know how i can roll stuff back if it doesn't pass the code review20:23
Stavros1(or something equivalent)20:23
LarstiQStavros1: roll stuff back?20:23
Stavros1roll the commit back, if it shouldn't be in the repo20:23
Stavros1i want to sort of check the commits to see if they should be committed to the main repo20:24
Stavros1like pqm, only manual20:24
LarstiQStavros1: that would be a classical gatekeeper workflow methinks20:24
Stavros1ah, i saw some info on that but i couldn't find the page now20:24
Stavros1oh found it, workflows20:25
Stavros1yes, that's it20:25
Stavros1i don't want to set up an entire system for that though, it's going to be just me and someone else at first20:25
Stavros1so what can i do? create a branch and always have them push there?20:26
Stavros1and rebranch if i don't want to include the commit?20:26
LarstiQStavros1: well, they can publish their own branch, and you then merge from it, review, accept/reject by committing or reverting20:27
LarstiQStavros1: or, get the other person to use `bzr send`20:27
Stavros1bzr send sounds nice20:28
Stavros1it sends me mail and then what do i do?20:28
LarstiQStavros1: you can read the diff and see if you want to have it, then bzr merge it20:29
Vantagehow do you guys deal with having large quantities of binary files that are versioned (e.g. images and flash files for a website) along with your code.  These would make branches fairly large.  Is there a way to have make having multiple branches on a machine a little less disk space expensive?20:29
LarstiQStavros1: or merge it and then do some checks, you have options :)20:29
LarstiQVantage: yes20:30
weigoncan you see in a bzr push which revno I got on the remote-tree ?20:30
LarstiQVantage: you can use a shared repository to share revisions between branches20:30
LarstiQVantage: and if it's about working trees, see link-tree from bzrtools20:30
VantageLarstiQ:  and just use bzr switch to switch between them?20:30
LarstiQVantage: that's an option, yes20:31
VantageLarstiQ: that was the one I thought of. Any other options?  I'm googling link-tree right now...20:31
LarstiQVantage: I don't know right now20:32
* LarstiQ gets back to hacking for a while20:33
VantageLarstiQ: ok, thanks.  When you said "an option" I thought you meant you knew of others :)20:33
jelmerOdd_Bloke: What exactly does "owner" on the brainstorm page mean?20:40
Odd_Blokejelmer: Person in charge of making it happen, I guess.20:44
Odd_Blokeigc added it.20:44
jelmerah, ok20:45
LarstiQjelmer: xp style terminology I think20:52
Stavros1i have a server running linux, what are my options for a read-only bzr repo?20:55
Odd_BlokeStavros1: HTTP. :)20:56
Stavros1Odd_Bloke: oh, with apache? does it also support auth?20:56
jelmerOdd_Bloke: and contributors? The people who contributed to the discussion @ the sprint or the people interested in contributing?20:56
Odd_BlokeStavros1: I'm not sure about auth, I'm afraid.20:56
Stavros1does it use apache?20:57
Odd_Blokejelmer: People who are anticipated to help out in the implementation (I guess).20:57
Vantagecouldn't you just set permissions on the tree and use ssh?20:57
Odd_BlokeStavros1: Posting to the ML or asking a question is probably the best thing to do, most of the Bazaar guys are heading to the pub right now (end of sprint :D).20:58
james_wjelmer: I think the "owner" is the person that you should talk to if you are interested in the feature. Contributors are those that are interested in it.20:58
Stavros1Odd_Bloke: ah :p20:58
james_wThough the owner are the ones that are probably going to drive it if it is going to happen.20:58
Stavros1Vantage: i probably could... i don't want to give people access to the machine, though20:58
VantageStavros1: you could put that copy on a separate machine (or virtual machine) and just push to it when you do releases20:59
Stavros1ah, i don't have another machine (even virtual) to spare...21:00
VantageStavros1: well they'll have to access somewhere ;)21:00
Stavros1they can access the server, just not all of it :P21:01
Stavros1i wonder if i can chroot it21:01
VantageStavros1: I remember something called scponly which might work21:08
Stavros1aha, i'll check it out, thanks21:08
igcjelmer: owner means "I'll guide/monitor this"21:08
VantageStavros1: you can also do lighter virtual servers with linux-vserver or openvz I believe.  They don't do hardware vm21:08
jelmerOdd_Bloke, james_w, igc: Thanks!21:09
igcjelmer: Contributor means "I'll  do the work if the owner doesn't" :-)21:09
Stavros1Vantage: well the apache solution sounds nice, do you have any info on that?21:09
beunojelmer, so being the owner is where you want to be21:09
Stavros1or actually i can just use ftp on another branch, i guess21:09
LarstiQbeuno: unless you have zero contributors.21:09
igcs/if/IF/21:10
beunoLarstiQ, that just means you have to blog a lot21:10
VantageStavros1: well apache is just putting it up on a publically accessible webserver, not sure about authentication support for bzr with that though, probably...21:10
Stavros1Vantage: oh duh, you're right :/21:10
Stavros1bzr has http access...21:10
VantageStavros1: yup :)21:10
Stavros1so i can just point apache to the repo21:10
Stavros1well that's simple enough21:10
Stavros1actually i have trac with the bzr plugin, does that expose the functionality?21:12
Stavros1i mean the repo21:12
VantageStavros1: they can browse it, but I don't think they can branch from it (but I could be wrong)21:14
Stavros1hmm, i was wondering if it has a url that points to the bare directory21:15
zepardhi21:17
VantageStavros1: it might.  Shouldn't be hard to set one up if it doesn't though21:18
LarstiQjelmer: oi, you added me to Line Endings ;P21:20
jelmerLarstiQ: Yeah, I initially assumed this was about who contributed to the actual discussion21:20
jelmerLarstiQ: You shouldn't be there any more now21:20
Stavros1Vantage: indeed, i'm doing it as we speak21:21
Stavros1it would just be nicer if it had one21:21
Stavros1i'll open a ticket21:21
igcnight all21:22
james_wfooooooood!21:22
beunofood+121:22
LarstiQfair enough21:22
phanaticfood++21:23
Odd_Blokehttp://icanhascheezburger.files.wordpress.com/2008/01/funny-pictures-pandas-eating-noms.jpg21:24
Stavros1it works! thanks for your help21:28
Stavros1auth works too, since it's a normal directory21:28
datooh, no james_w22:19
abentleydato: They were off to the pub last I heard.22:34
datook, thanks22:34
blueyedCan you query the last "push" to a repo (or "commit" for a checkout)? Is this "latest revision" in "bzr info -v"?22:35
bpetersonblueyed: yes22:38
blueyedwell.. unfortunately "bzr info" is probably slower than "bzr commit". is there a shorter path for getting this info?22:45
datoblueyed: sorry, I did not understad your problem very well?22:46
datomaybe you want `bzr log -r -1 | grep timestamp` ?22:47
datogotta go now22:47
blueyeddato: I'm thinking about optimizing the add-5-a-day tool: only commit if the last commit is older than 1h.22:47
blueyedyes.. makes sense and is faster.22:48
bpetersonblueyed: I'd try bzr log22:48
awilkinsblueyed: how about bzr revision-info23:14
awilkinsblueyed: forget it, that doesn't have the TZ offset in the time23:15
awilkinsBut it is UTC though....23:15
blueyedawilkins: it's quite some faster.. There is probably something better even in bzrlib though, which does not require using the email module to parse the output of a subprocess.. ;)23:21
awilkinsEmail module? Just take the split between the first and second "-"23:28
awilkinsOr a regexp of 14 digits beginning with "20"23:29
blueyedawilkins: yes, from rev-info, but not from log.. a subprocess it bad though, isn't it - if I'm already in python.. can't I use bzrlib for this?23:29
awilkinsblueyed: Yes, of course you can23:29
ubotuNew bug: #199654 in baltix "bzr on fat32/ntfs from ubuntu  works only with sudo init/commit" [Undecided,New] https://launchpad.net/bugs/19965423:29
ubotuNew bug: #199655 in bzr "bzr status shows pending merges with incorrect indentation" [Undecided,New] https://launchpad.net/bugs/19965523:29
ubotuNew bug: #199659 in bzr "editor backup file left over after bzr commit" [Undecided,New] https://launchpad.net/bugs/19965923:30
blueyedawilkins: any hints where to look in bzrlib?23:34
awilkinsbranch.py is where I'm looking23:34
awilkinsfrom bzrlib import branch ; b = branch.Branch.open(branchRoot) ; b.last_revision_info() ; # that seems to work23:37
awilkinsDoesn't work for alternative revid formats like bzr-svn though23:41
jelmerThere is no guarantee that standard revids follow that format either23:42
awilkinsHow do you make a python object tell you what members it contains?23:44
awilkinsI can't be arse to pick through all the source looking for what revision_history() emits23:45
Odd_BlokeEvening all.23:46
blueyedawilkins: I use ipython and tab for that..23:46
blueyedipython is really awesome wrt to tab-completion.. in fact it's also a shell replacement I've heard.. ;)23:47
awilkinsbah, theyre just strngs anyway23:48
blueyedawilkins: yes, b.last_revision() seems to be the closest so far..23:48
awilkinsblueyed: But you can't rely on the format and it's just a string23:49
jelmerb = Branch.open("foo")23:51
jelmerb.repository.get_revision(b.last_revision())23:51
Odd_Blokejelmer: o/23:51
jelmerthat last line will return a revision object which contains the commit timestamp among other things23:52
jelmerhey Daniel23:52
* awilkins discovers __dict__23:52
blueyedjelmer: thanks, r.timestamp is it..23:53
Odd_Blokejelmer: LarstiQ, Verterok, beuno, phanatic and I are sitting in the bar at the hotel with our laptops out. \o/23:54
jelmerOdd_Bloke: Which is why you're currently talking with me on IRC ? ;-)23:54
beunoOdd_Bloke, with girls, of course (?)23:54
jelmer*without laptop* >-)23:54
jelmerah23:55
jelmerout as in you have them in front of you23:55
jelmernot as in you have them turned off23:55
jelmerwell, I'm getting some sleep23:56
jelmerslept a total of 8 hours in total in the last two days and not much during the sprint either23:57
jelmerSay hi to the other folks over there for me!23:57
beunojelmer, how was your exam?23:57
jelmerbeuno: It went ok, but not brilliant.23:57
jelmerwith a bit of luck, I'll pass it23:58
beunojelmer, well, bzr-gtk got better, so that might help for your conscious23:58
jelmer:-)23:58
james_whi jelmer. Sorry I missed you at the sprint.23:59
james_wSleep well.23:59

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