/srv/irclogs.ubuntu.com/2007/11/04/#bzr.txt

somerville32Hey. I've made Bzr quotes! :]00:01
=== BasicMac is now known as BasicOSX
lifelessnight all03:21
PengGood night.03:22
beunolifeless, you wouldn't happen to be around, would you?   I'm working with Verterok on some changes to bzrlib we would like included before 1.0, and I can't commit to the bzr's branches anymore. Is there a chance I can get activated again?04:25
jelmerbeuno: which bzr branch would you like to commit to ?04:39
jelmerbzr.dev isn't accessible directly - please send merge requests to the list04:40
beunojelmer, no, we want to uplaod a branch to the group so both of us can commit04:40
beunoand it really doesn't make sense to create a new group just for this purpose04:41
beunoso we want to upload it to the bzr group, and work on it there04:41
jelmerah, ok04:41
beuno(seemed to me like the logical workflow)04:42
Kamping_Kaiserwould i be able to install bzr on a sparc running debian Etch?07:04
PengIf Python runs on it, bzr should have no problem.07:07
Kamping_Kaiserthanks07:08
PengBazaar has some C components for performance, but even if they didn't compile for some reason, it doesn't need them (it'll just be slower).07:08
Kamping_Kaiserbzr intalled from backports fine. thanks.07:11
PeakerIf I decide to reject/destroy a branch in a shared repo, I'll get ghost revisions left, right? Is there a way to clean up such ghosts?11:27
LarstiQPeaker: ghost revisions are the opposite, they are referred to by a branch but don't actually exist in the repository.11:29
PeakerLarstiQ: oh, how can that happen?11:29
LarstiQPeaker: after removing a branch you have revisions in your repo that are not referenced instead, and yes you can remove those (although 'bzr gc' still isn't written)11:29
gsuvegre11:29
LarstiQPeaker: one very good source for that is branches converted from arch, where it was almost trivial to get to that state.11:30
gsuvegbzr-gtk is broken?11:30
LarstiQgsuveg: nafaik11:30
gsuveggutsy and from bzr site11:30
PeakerLarstiQ: How can I remove them?11:30
LarstiQPeaker: the remove-revisions plugin11:30
PeakerLarstiQ: ah, thanks11:31
LarstiQPeaker: but be careful, that can also create ghosts11:31
PeakerLarstiQ: if it can't see all the branches?11:31
PeakerLarstiQ: or, how would it cause that?11:31
LarstiQPeaker: if you tell it to remove revisions that are still used11:31
PeakerLarstiQ: Oh, I was hoping its more like "gc" :)11:31
LarstiQPeaker: the step of figuring out which ones to use is not automated afaik11:31
gsuvegbzr-gtk: Depends: bzr (< 0.91~) but 0.92~rc1-1bazaar1 is to be installed11:31
LarstiQPeaker: right :)11:32
gsuvegLarstiQ, sry. me sound like its a bit broken11:32
gsuvegapt sources deb http://bazaar-vcs.org/releases/debs/gutsy ./11:32
LarstiQgsuveg: you're talking about specific situations wrt a distribution, your original statement was way broader11:33
LarstiQgsuveg: so that probably means newer bzr-gtk debs have to be uploaded11:33
LarstiQbut bzr-gtk itself works just fine11:33
datogsuveg: we're waiting that bzr-gtk 0.92 gets releseased, otherwise it can't be packged :)11:33
LarstiQjelmer: ^^11:33
LarstiQdato: oh well, that might also block a bit ;)11:33
gsuvegdato, ah.11:33
gsuvegbut .91 would work with .92 not?11:36
Peakerwhy is ubuntu bleeding-edge not up-to-date with newest released bzr's?11:39
gsuvegdato, and on repo bzr-gtk is only the 0.9012:02
gsuvegyou can edit the dependecy not ?12:02
datogsuveg: well, it's better not to12:06
jelmerLarstiQ, Szilveszter does the releases these days12:16
gsuvegi've back to 0.90 what in gutsy is12:16
LarstiQjelmer: k12:18
lifelessmoin13:57
ryanakcaSince codebrowse appears to be down, how could I get a file from revision 4 if that file is no longer versioned? I've tried bzr revert -r4 file ... but it complains about not being versioned...13:58
datoryanakca: try bzr cat -r 4 path/to/file >file.v414:00
ryanakcadato: thanks :)14:02
lifelessryanakca: bzr revert -r4 file should indeed have worked;14:12
lifelessdesperately seeking reviewers15:08
vilalifeless: hi :) You have two patches pending revisions, I already expressed my concerns about transport.list_dir, I didn't dig enough reconcile nor packs so far to comment on the second (stage 1 of pack reconcile)15:27
vilaAIUI you plan to rework the first to avoid list_dir, if that's the case, I'd be happy to give you a +1 if that makes it happen quicker :)15:28
vilaghaa, s/revisions/reviews/15:29
lifelessvila: well, you know I disagree  on the list_dir usage15:52
lifelessthere is no technical reason I know of yet to avoid list_dir for writable transports15:52
vilalifeless: ha, thanks for the precision, I was wrong then :)15:53
lifelessI plan in a future packs format to see if we can remove list_dir15:54
lifelessbut for the current format there is no alternative15:54
lifelessthat isn't worse.15:54
vilahmmm, so putting the webdav plugin on the back burner still depends on that then15:55
vilaI'll to monitor that point more closely15:55
lifelessstill, I'm seeking a reviewer for the stage1 reconcile packs patch16:44
lifelessjelmer: ping16:55
jelmerlifeless, pong17:02
lifelessjelmer: I've looked at the setup.py patch; and it seems less clean to me than th autotools it's replacing17:19
lifelessjelmer: I wanted to chat interactivvely about this, because I may be missing something17:19
jelmerlifeless: at the moment, the python package doesn't get installed at all17:31
jelmerso you're forced to run out of the source tarball17:32
lifelessright, so I want to fix that17:32
lifelessand I ocmcmented in the bug that I had no objection per se t o moving everything to setup.py if it was a net win; but thats clearly more than just fixing the bug17:32
jelmerAlso, the dependency on automake just to install a few files is a bit heavy17:33
jelmermake check doesn't work currently (for me at least)17:33
lifelesswhat goes wrong ?17:34
jelmerlet me check17:35
jelmerlifeless: never mind, looks like config-manager got deinstalled for some reason17:40
* jelmer wonders if it'd be a better idea to start from scratch17:41
lifelessjelmer: with pqm ?17:42
jelmernot with pqm specifically, but the sort of thing I'm looking for myself17:43
lifelessa little more detail might help me here :)17:45
jelmerI don't need support for baz, config-manager or creating repositories and things like the current testsuite in shell, automake make it hard to add the features I am looking for (bundle support, python gpgv, loading bzr plugins, better status page)17:45
ubotuNew bug: #160012 in bzr "http urllib hides programming errors related exceptions" [Low,In progress] https://launchpad.net/bugs/16001217:45
jelmerperhaps git support17:46
lifelessjelmer: so, pqm is getting cleaned up slowly;17:46
lifelesspqm does load bzr plugins - the plan is to drop all other VCS upport; support git etc via bzrlib plugins17:47
=== phanatic_ is now known as phanatic
jelmerpqm doesn't load bzr plugins at the right time yet, that's what my other patch fixed17:48
lifelessbug #? I'll commit it now17:48
jelmerI don't think there's a bug #.. was a merge request I sent to you on may 2317:52
jelmershould I resend it?17:54
lifelessI've found it I think. one sec, playing with it.17:54
jelmerlifeless: We could still ditch autoconf and automake, use a simple Makefile for building the docbook stuff and setup.py for the python-specific things?17:56
lifelessPEP8 issues but otherwise fine I think18:06
lifeless(VWS after class is needed)18:06
lifelessalso ou create a heavweight checkout not a lightweight one18:07
jelmerlifeless, what about using Makefile for docbook for bug 117197 ?18:10
ubotuLaunchpad bug 117197 in pqm "Doesn't install required Python package" [High,Fix committed] https://launchpad.net/bugs/11719718:10
jelmerlifeless: still alive?18:17
mc_how can i tell bzr which editor to use?18:19
datomc_: there are several ways, all explained in the tutorial. basically some environment variables (EDITOR, BZR_EDITOR)18:20
datomc_: and the "editor" option in ~/.bazaar/bazaar.conf18:21
mc_ty18:23
=== mc_ is now known as mc__
=== BasicMac is now known as BasicOSX
schierbeckjelmer: pingz0r!19:06
jelmerschierbeck, pong!19:11
schierbeckwhat do you think about using commit messages to identify revisions to the user?19:15
schierbeckwhen they have to choose from a revision's parent19:15
lifelessthat might be nice19:17
schierbecki've sent a branch url to the ml19:17
schierbeckhave a look and tell me what you think :)19:18
schierbeckcurrently only the toolbar buttons use it19:18
jelmerlifeless: see my comments from about an hour ago19:18
schierbeck?19:20
schierbecki only just sent it in19:20
schierbeckdamn you're fast!19:20
datoheh19:20
jelmerschierbeck, that was about some pqm stuff19:20
schierbeckoh19:20
schierbeck:)19:20
lifelessjelmer: Sorry, eating etc19:21
lifelessI've merged your patch, it's off to people now19:21
=== me_too is now known as turbo0O
=== turbo0O is now known as me_too
jelmerlifeless: any thoughts wrt using make+setup.py rather than automake?19:21
jelmerlifeless, thanks for merging it19:22
lifelessdone, revno 17219:22
lifelessnetwork here sucks worse than well most things19:24
lifelessI don't entirely get the objection to automake; automake isnot huge, it's well understood19:28
jelmerIt's not really aimed at installing python packages19:32
jelmeryou'd have to add some magic to find the right directory to install it in, the right python executable to use, etc19:33
jelmeralso, automake sucks in general imo19:36
schierbeckjelmer: don't you think we call the navigation buttons in the viz something other than "back" and "forward"19:46
schierbeckit doesn't fit with the visual representation of the history19:47
schierbecki'd rather have "up" and "down"19:47
jelmeryeah, up and down sounds fine to me19:47
datobut history is *time* independently of how you repreesnt it19:47
jelmerI'm not so sure about using commit message to identify revisions19:47
datoand time goes forwards and backwards, not up and down19:47
datoIMHO :)19:47
jelmerthere are a lot of people that don't use clear enough commit messages19:48
schierbeckjelmer: i still think it's MUCH better than, say, revision id19:50
schierbeckand i think it's good enough for now19:50
schierbeckuntil we find something better19:50
schierbeckdato: but normal people don't think that way19:51
schierbeckthey say "i want to go down on this list"19:51
datowell. you won't convince me but, alas, I'm not the bzr-gtk maintainer. :)19:51
schierbeckuser interfaces aren't about what's right, they're about what's right for the users19:51
schierbeck:)19:52
jelmerschierbeck: I'd rather not experiment like this in the branch that's used for releases19:53
phanaticschierbeck: i'm about to release 0.92 and jelmer and i were talking about broken lines (ui option to disable it would be nice to have)19:54
jelmerschierbeck: I think revno's would be a better temporary solution19:55
schierbeckjelmer: compromise -- revno's *and* messages19:57
schierbeckphanatic: i'm working on options ui right now19:57
schierbeckjelmer: you have to agree, revids are completely useless for this purpose19:58
jelmeryeah, revids aren't useful for this sort of stuff, indeed19:59
jelmerschierbeck, I think a combination is a good compromise19:59
schierbeckokay20:02
schierbecki'm currently looking at including the branch nick as well20:02
schierbeckit works pretty good20:02
Peakeris there a way to retroactively delete large files to save space in the repo?20:15
jelmeryou can remove revisions from history using the remove-revisions plugin20:16
jelmerbut it's very very slow (rewrites the whole repository)20:16
Peakerso if I remove revisions that added large files, the continuations of those will be rebuilt without those large files?20:18
Peakeror will the continuations/children just be broken?20:18
jelmerIn theory, yes20:20
jelmerHaven't ever done that yet20:20
Peakeryes broken or yes deleted? :)20:21
jelmerthe continuations will be ok20:22
Peakerah cool, I hope he didnt add more stuff with the big files20:24
schierbeckphanatic and jelmer: how should i turn off brokenlines?20:27
schierbeckif you could add a gproperty to viz.treeview.TreeView it would be pretty easy20:28
phanaticschierbeck: we were pondering about a button on the toolbar which would toggle broken lines (and also remember its state)20:29
schierbeckphanatic: i'd rather have it in the menu bar20:30
schierbeckpeople probably won't toggle it every few minutes20:30
jelmerschierbeck: that means the menu bar would have to be merged before 0.9220:31
Peakerdo you think it would be a good idea to create a plugin that removes a file retroactively by recreating the entire history in the repo excluding the addition and any changes to that file?20:31
phanaticagreed. but merging the menubar branch just before release wouldn't be a good qa prectice :)20:31
phanaticpractice20:31
jelmerschierbeck: which should really be released asap before bazaar 0.92 is released20:31
schierbeckjelmer: okay, we can move it to the menubar after the release20:32
schierbeckwhat persistence backend should we use?20:33
schierbeckthe conf files?20:33
jelmereither the conf files or gconf, not sure what would make the most sense20:33
jelmerusing the conf files is consistent with bazaar, using gconf is consistent with GNOME20:34
phanaticjelmer: gconf is not available on win3220:34
phanaticafaik20:34
jelmerok, so we would at least have to support the conf files20:34
jelmerguess we can always add gconf support later, if there's a good reason for it20:34
schierbeckdo we have to reload the treeview when toggling?20:35
schierbeckor can it redraw dynamically20:36
jelmerreloading would be acceptable for now I think20:36
schierbeckokay20:37
schierbeckby the way, what do you think about the menubar approach? is it something i should pursue?20:38
phanaticschierbeck: from the olive point of view, a menubar on a child window looks quite interesting :)20:39
schierbeckphanatic: i know, but i hate olive20:39
schierbecki never use it :)20:39
phanaticmany people do, love it or hate it :P20:39
schierbecki think it's important that we can set options directly from viz20:40
schierbecksuch as show/hide columns20:40
schierbeckand a go menu really makes sense20:40
schierbecki'm only asking because i'm making some other major improvements on that branch20:41
jelmeryeah, we have to keep olive in mind20:41
phanaticthat sounds reasonable. i don't have any other complaints about the menubar on viz20:41
jelmeranother option would be to add a LogWindow that is specific to olive and is somewhat simpler20:42
phanaticbut maybe you could allow it to be disabled20:42
jelmernow that we have all the different components available as separate widgets, that should be quite easy20:42
schierbeckyup, that sounds good20:44
schierbecki'll continue with my efforts then20:44
schierbecki'd like to switch to using signals and properties only20:45
schierbeckthe revision-/logview especially is bad20:45
schierbeckwith a custom callback mechanism20:45
schierbeckwhen's the next release due?20:46
jelmerall synchronised with bazaars' releases20:46
jelmers/bazaars'/bazaars main/20:46
Peakerit could be nice to pop up kompare as the differ in the "bzr vis" dialog20:47
phanaticwe agreed to do synchronised releases till bzr 1.020:48
datoand then?20:49
phanaticbzrlib should be api stable by then, so we can walk on our own path20:50
phanaticjelmer, schierbeck: i'll do a release tomorrow, if you don't mind...20:53
schierbeckit's fine20:53
jelmersounds good to me, provided that the broken lines feature can be turned off somehow20:53
schierbeckis there a setting one can give to the cell renderer that disables the newish stuff?20:54
schierbecki'm not too comfortable with the cairo stuff20:54
jelmerschierbeck: you can specify the behaviour of the broken lines code using a parameter to TreeView's constructor20:55
schierbeckokay20:55
schierbeckshould it be None when turned off?20:57
jelmeryes20:57
schierbeckokay20:57
schierbeckand just hardcoded to 32 otherwise?20:57
schierbeck(short-term)20:57
jelmerI'm not sure what the best value for it would be, it would depend on the branch I think21:06
jelmerI just kept 32 as default since that's what gary used21:06
schierbeckokay21:10
schierbeckjelmer: what should the toggle button be labeled?21:10
jelmer"Compact View" or something?21:12
schierbeckokay21:15
schierbeckjelmer, phanatic: pushing to lp right now21:22
schierbecki still need to implement persistence21:22
phanaticschierbeck: cool21:22
schierbeckphanatic: i'd like to have a meeting some time in the not-too-distant future21:24
schierbeckabout coding guidelines21:24
phanatic:)21:24
schierbeckthings like the naming of callbacks21:24
schierbeckthere are like 3 different ways being used right now :)21:24
phanaticyeah, it would be useful21:25
schierbecki'll send a message to the ml tomorrow21:25
phanaticright21:25
schierbeckhttp://bazaar.launchpad.net/~dasch/bzr-gtk/brokenlines21:26
schierbeckcheck it out21:26
schierbecki'll go eat something :)21:26
schierbeckjelmer: do you know how to implement the persistence layer?21:55
schierbecki haven't looked at the api21:55
jelmerphanatic was looking at that21:57
schierbeckokay21:58
schierbecki'll just take a look myself, it seems that phanatic has left21:59
schierbeckwell, that was pretty easy22:07
jelmeryou may both be working on the same thing22:07
schierbeckwell, i'm already done :)22:12
schierbecki've pushed it to the branch22:12
schierbeckhttp://bazaar.launchpad.net/~dasch/bzr-gtk/brokenlines22:12
lifelessola22:13
Verteroklifeless, hola22:24
lifelessjamesh: hi, how did pending reviews go on further changes ?22:32
jameshnot sure22:35
lifelessjelmer: we need a better name for olive22:36
lifelessjelmer: it's still showing as 'olive' rather than e.g. 'bzr GUI' or something22:36
jelmerlifeless: please discuss that with Szilveszter, I don't use/develop Olive22:36
lifelessahha, ok22:37
jelmerpersonally, I would rather see the integration in nautilus improved than having a standalone app22:37
jelmerbut I think Olive is nice to have too - it's certainly more mature than nautilus-bzr22:38
schierbeckjelmer: i agree completely22:39
lifelessI agree that iing the nautlius integration is important22:39
lifelesss/ii/fixi/22:39
schierbeckwe need more work on nautilus-bzr22:39
schierbecki should eventually replace olive completely, if possible22:39
schierbeck*it22:39
jelmernot sure about that - olive also works on Windows and XFCE22:40
schierbeckwe need better integration with the Windows file browser, too22:40
jelmerso there is certainly a place for it, until we get thunar and windows shell integration done22:40
lifelesswe have a bzr tortoise for inwdows now22:40
jelmerlifeless: tortoise isn't anywhere near olive wrt functionality though22:40
lifelesstill; native UI++22:41
schierbeck+122:41
* jamesh wonders why all Windows VCS plugins are called tortoise22:41
jelmeryeah, I agree22:41
schierbeckbtw, i got a tango guy working on some icons for us22:41
jelmerjamesh: they're all slow22:41
lifelessouch22:42
jelmerschierbeck: ah, nice :-)22:42
schierbeckthought so :)22:42
fullermdI wish one of them were easier to get going, though.  Every time I look at the installation instructions, it makes ME twitch, and I'm the *nix guy.  The idea of trying to point some of the graphicist Windows people I'd want to use it at that installation isn't confidence-inspiring.22:42
jelmerwe need more people developing on Windows22:42
fullermdOr less people running Windows   ;)22:43
jameshby that I take it you mean other people?22:43
schierbecki'm with fullermd...22:43
schierbeckhehe22:43
fullermdI mean, there can't be more than a billion or so of them, right?  All it takes is a little rounding error to turn 1 into 0...22:43
schierbeckhow are we doing on OS X22:43
jelmerjamesh: Afraid so, yep22:44
schierbeckjelmer: should we review and merge the brokenlines-option branch tonight?22:45
jelmerschierbeck: as long as it happens before 0.92 :-) Not sure what Szilveszters plans wrt the relaese date are22:47
schierbecki think he wants a release tomorrow22:48
lifelesslater all, meeting time22:54
schierbeckbb22:54
schierbeckwell, i'm off as well22:55
schierbecksee y'all!22:55
ubotuNew bug: #159997 in bzr "`bzr push` fails with error 530 (login incorrect)" [Undecided,New] https://launchpad.net/bugs/15999723:00
ubotuNew bug: #160081 in bzr "KnitCorrupt: incorrect number of lines" [Undecided,New] https://launchpad.net/bugs/16008123:00
siretartjelmer: is bzr-svn supposed to support push over svn+ssh:// urls?23:08
jelmersiretart: yes.23:09
siretartjelmer: http://paste.debian.net/4150523:09
siretartnot sure what's going wrong here23:09
siretartsvn ls svn+ssh://svn.debian.org/svn/pkg-wpa/wpasupplicant/trunk works at least23:10
jelmersiretart: please file a bug23:10
siretartok. what information do you need?23:10
siretart(apart from the traceback)23:10
jelmerthe backtrace should be sufficient23:10
siretartok23:10
siretartthanks!23:10
siretartfiled as bug #16008523:13
ubotuLaunchpad bug 160085 in bzr-svn "push over bzr+ssh:// crashes" [Undecided,New] https://launchpad.net/bugs/16008523:13
jelmerthanks23:13

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