/srv/irclogs.ubuntu.com/2011/01/11/#bzr.txt

mkanatlifeless: Merging that sourcedeps.conf update that I just proposed should fix the KeyError.00:20
lifelessthank you00:22
maxbHrm... help?  I'm hacking on bzr-svn, and I'm getting exceptions a lot.... and every time, bzr fires up apport needlessly00:25
maxbHow do I stop it?!00:25
pooliehi maxb00:25
poolieah00:26
poolieturn on the no_apport debug flag00:26
mkanatlifeless: I'll fix that revision number one, too.00:26
poolieif you check the "don't show this again" that ought to work00:26
pooliefor the particular bzr script in question00:26
maxbpoolie: seems to work on the command line, but not in bazaar.conf00:26
maxband I don't have any 'don't show this again'00:26
pooliereally not in bazaar.conf?00:27
pooliedebug_flags = no_apport00:27
pooliei think i've used that myself so i'm surprised if it's failing00:27
maxbhmm. I think I may have assumed debug_flags was a space-separated list erroneously...00:28
maxb:-)00:28
pooliecommas00:28
poolieor ', '00:28
maxbStill, it does seem a bit quirky that bzr does not honour the systemwide apport enabled status00:28
pooliemm00:28
pooliemaybe00:29
pooliethat one flips as ubuntu goes into release mode, meaning "ubuntu no longer wants so many crash reports"00:29
pooliei see that as a bit orthogonal to whether we want them in bzr00:29
maxbI suppose.00:31
maxbugh, this is mad. bzr-svn is calling svn's get_dir on various things, some of which are files, and only some of the files trigger a SubversionException: ("Can't get entries of non-directory", 160016)00:32
jmlpoolie: you still in #516?00:33
poolieyes00:33
* jml goes to rock a small part of the world.00:34
mkanatlifeless, poolie: Okay, proposed a merge for the "no such revision" oops.00:38
poolienice00:38
mkanatIt actually was good to fix, because it fixes it for every page, since the code is centralized for reading the revno from the URL.00:39
maxbhuh, novel. bzr-svn just managed to eat 4GB of RAM and drive my machine so much into swap death that I had to kill it from a text console :-/01:16
jelmermaxb: what were you trying to do with it?03:29
=== jfroy_ is now known as jfroy
AfCHi05:34
=== luks_ is now known as luks
gthorslundmorning #drizzle! happy bugs week!06:40
gthorslundhehe... morning #bzr06:41
* gthorslund little buggy today :D06:41
maxbjelmer: Run a svn-import with a custom layout class which involved importing a repository with unusually many branches08:54
neaj'lo all -- I have a bunch of config files in .../configs , now i want to 'bzr merge configs/a configs/b'. bzr tells me "Nothing to do", but that's not true. What am I missing?10:30
sobersabrehi.10:31
sobersabreis it possible to see when the last operation of format conversion of a repository has happened ? (local bzr upgrade call)10:32
neajalso, i want to 'bzr get configs/b', but bzr tells me: "bzr: ERROR: Not a branch:"10:32
neaj'bzr get configs' works, but i don't want all the configs here.10:32
kgoetzsobersabre: ~/.bzr.log might carry it10:33
bob2neaj: bzr doesn't currently support checking out part of branches, afaik10:33
bob2I don't think any modern vcs does10:33
neaji believe with svn you can checkout any dir in the repo ..10:34
bob2right10:34
bob2'modern' :-)10:34
neajheh OK10:34
bob2bzr split is a thing, but I don't know how well it works nowadaus10:35
neajOK so how to merge configs/a configs/b ?10:35
neaji can do diff and patch, but that's hardly modern ;-)10:35
neajbzr has no support for merging within a branch?10:39
neajmaking config/a a branch of config/b just to enable merging wouldn't make sense. they're just configs that happen to be mostly similar.10:51
=== Ursinha-afk is now known as Ursinha
catphish_is is possible to specify the local path to a branch in bzr commands?11:54
catphish_ie a path to cwd to before executing the command11:55
bob2pretty sure you can just use a normal path and it'll find the bzr repo above that11:56
catphish_how do you specify that path?11:56
bob2bzr add /who/cares/long/path11:57
catphish_that's useful for add11:57
catphish_but for example 'cat'11:57
bob2it's also useful for cat11:57
catphish_i'll try it11:57
bob2less useful for ci11:57
catphish_so there's no global cwd option?11:58
bob2dunno11:58
catphish_but you can specify it on individual commands as part of the filename11:58
catphish_ok11:58
catphish_it works for cat :)11:59
maxbThere is no global option, which is quite annoying sometimes. Many commands support -d / --directory, but not all12:16
catphish_ok12:18
catphish_it's useful that you can specify it as part of a filename12:19
catphish_but i might just cd to the path before each command12:19
quicksilveranyone understand bzr merge internals well enough to tell me what it does in this case here? : http://web.archive.org/web/20070603113858/zooko.com/badmerge/simple.html13:01
quicksilverbzr also gets it "wrong".13:25
quicksilverFWIW.13:25
Odd_Blokequicksilver: bzr has several merge algorithms, tried them all?13:53
quicksilverOdd_Bloke: I tried default and weave13:53
=== Ursinha is now known as Ursinha-lunch
jelmermaxb: find_branches() in svn-import has known memory usage issues15:12
maxbI think those are what I saw15:12
maxbjelmer: oh, there was something else I was struggling with, that confused me intensely. The repository I was working with had some doc files at the hierarchy level where branches were.15:15
maxbbzr-svn seemed to be attempting to interpret the paths as branches, and was calling conn.get_dir(...) on them15:15
maxbMuch to my complete confusion, for *some of them only*, this would cause libsvn to error "Can't get directory entries for a non-directory"15:15
maxbDoes anything like that sound at all familiar to you?15:16
jelmervaguely, yeah. IIRC this was a svn server bug15:16
maxbDoes bzr-svn mostly rely on people not having files at places in the hierarchy where it's looking for branches, or did I miss some detail when I went to implement the custom layout class?15:18
maxbThe layout call that seemed to be involved doesn't get passed a revno, so checking in the repo is not feasible15:19
maxbOh, actually, it looked more like finding *tags* was what killed it for me15:25
catphish_what would be the most lightweight way to detect if a repo is empty, and how many commits exist in it?15:26
catphish_*branch15:26
maxbcat .bzr/branch/last-revision     ? :-)15:27
catphish_does last-revision always == number of revisions?15:33
catphish_i assume it does15:33
maxbthe number in there is the number of *mainline* revisions15:33
catphish_mainline? (excuse my ignorance)15:33
maxbi.e. bzr merge bighugebranch; bzr commit will cause it to go up by 115:33
catphish_that's merging from a totally separate branch right?15:34
maxbdepends what you mean by "totally separate", but probably yes15:35
catphish_i'm easily confused because i come from a number of SCMs where one repository has multiple branches15:36
catphish_and while that is true in bzr, the branches are almost entirely separated15:36
=== Ursinha-lunch is now known as Ursinha
catphish_i'm trying to git bzr into an environment where one repository has multiple branches15:37
catphish_which is true of bzr :)15:37
catphish_but in a way i'm not used to15:37
catphish_for example to switch branches, one has to switch to a different location on the filesystem15:38
jmldo you guys know about this warning in natty?15:38
jmlUsing saved parent location: bzr+ssh://bazaar.launchpad.net/~launchpad-pqm/launchpad/devel/15:38
jml/usr/lib/python2.6/dist-packages/Crypto/Util/randpool.py:40: RandomPool_DeprecationWarning: This application uses RandomPool, which is BROKEN in older releases.  See http://www.pycrypto.org/randpool-broken15:38
jml  RandomPool_DeprecationWarning)15:38
maxbjml: It's been around for aaaages, but you don't see it normally because warnings are suppressed in bzr final releases. But natty currently has a beta.15:43
jmlmaxb: ahh, I see.15:43
maxbIIRC it's waiting for the paramiko team to merge the fix15:43
jmlmaxb: Isn't the paramiko team just Robey?15:44
maxbcould be, I don't know15:44
jml:)15:44
jmlmaxb: I guess it's safe for me to ignore the warning for now.15:44
=== Meths_ is now known as Meths
spivjml: yeah, it's safe16:23
pooliemaxb, around?16:31
maxbhi16:31
=== beuno is now known as beuno-lunch
pooliemaxb i think our 2.2.2 SRU should now be unblocked16:32
poolieper https://lists.ubuntu.com/archives/technical-board/2010-December/000632.html16:33
pooliewhat do you think?16:34
maxbI think *technically* we were also waiting for 2.3b4/5 in natty16:34
maxb(Because fixes are supposed to enter natty first)16:34
pooliehm, really?16:35
poolieeven with a MRE?16:35
poolieif we were backporting particular patches that would make sense,16:35
pooliebut if we've generally said "2.2.x is acceptable"16:35
maxbyeah, but we're being quite odd by shipping a beta in natty currently16:38
maxbmost packages wouldn't do that16:38
pooliewhy?16:38
pooliei mean why odd?16:38
pooliebecause they're kinda counting on it becoming stable before release?16:38
maxbWell, it's relatively rare for packages to have beta versions in the archive, that's all16:39
poolieright, though not unprecedented16:45
poolieis this a concern to asking for an sru?16:45
pooliei'd just like to get them actually out, now we've done the releases16:45
=== bigjools-afk is now known as bigjools
maxbpoolie: Understood, I just don't much feel like requesting a SRU whilst we're violating point 1 of https://wiki.ubuntu.com/StableReleaseUpdates#Procedure17:01
pooliei see what you mean17:01
pooliei think that procedure doesn't really allow for the existence of MREs though17:01
pooliein fact it pretty much contradicts it17:01
poolieso, i'd like to persuade you to request one and have the conversation with them17:02
quicksilveris there a way to "resurrect" a deleted file in a way that bzr will actually see as an undeleted?17:28
quicksilverof course I can just extract it from an old revision and re-add it, but I'm curious17:28
fullermdrevert it to a rev it was in.17:28
quicksilverah17:28
quicksilverI always forget you can revert single files17:28
quicksilverI always think of doing bzr operations on whole trees17:28
quicksilverthanks17:28
pooliequicksilver, i think there's actually a plugin that adds a 'resurrect' command17:30
quicksilverbzr can't track the case where a file 'becomes two files', can it?17:30
maxbno17:33
maxbHence wailing and gnashing of teeth in the bugtracker concerning 'bzr cp' :-)17:33
poolieso, maxb, about the sru?17:36
pooliedo you have reservations yourself about proceeding with it, or do you just think it will be rejected, or ..?17:37
maxbpoolie: two things: I'm not happy to push this ahead myself without an explicit waiver on point 1 from ubuntu-sru (because my enthusiasm for this SRU is low and I don't see worth in pushing the borders of policy), and additionally, it's not ubuntu-sru's job to upload packages, so you need to speak to ubuntu-sponsors first, I think17:38
poolieok17:39
poolieyour enthusiasm for this SRU is low because...17:39
pooliethe bug fixes in it aren't exciting enough?17:39
maxbI feel it's taken so long, that most people who care are probably on ~bzr/ppa already17:40
pooliehm17:43
pooliewell, in general i do want to get updates out17:43
pooliei want to fix up the sru process so that it there aren't such long delays in future17:44
maxbhmm, actually I'd forgotten the dput hang is in there17:44
pooliei wonder if everybody affected by these things has really upgraded to the ppa17:44
pooliei know, fixed by you even17:44
poolieit's actually the lead bug for the request17:44
maxbSee, my mind is on 2.3 these days, 2.2 seems so old :-)17:44
poolie:)17:44
pooliewe were just saying this morning that if we're not going to push them all the way through the process, we almost might as well not do bugfix releases17:45
pooliei can ask -sponsors to upload it to -proposed17:46
pooliethat might be good practice for me17:46
pooliei just wanted to understand how you thought about it17:46
maxbAlright, so I take back my concerns about it not being SRU-worthy, but I'm still reluctant to press the SRU without conforming to the process w.r.t. fixing in natty first17:46
=== beuno-lunch is now known as beuno
lifelessjames_w: so _export_iter_entries does not permit including . bzrignore17:58
lifelessjames_w: one way to fix things would be to provide a flag to control that17:58
nekohayohey there, I did      echo "the revision ID I got from bzr viz" > .bzr-upload.revid        on my remote host because the files are already there and I don't want to reupload them again18:02
nekohayobut when I try to bzr upload, it says that my local repository has no such revision18:02
beunonekohayo, and you're sure you didn't put in the revno, but the actual revid?18:03
nekohayoyeah, it's foo@bar.com-20110111174334-cg5oj39oq3piua8c18:03
pooliei wonder if it's unhappy about whitespace or something?18:04
nekohayohm, what whitespace?18:04
nekohayooh18:05
nekohayomaybe there's a \n at the end of the file and it doesn't like that18:05
nekohayopoolie, do you have a trick to remove the trailing \n  ?18:05
poolieyes18:05
poolieecho -n FOO > file18:06
nekohayothat worked18:06
nekohayowondering if I should file a bug though18:06
james_wlifeless, that would work18:07
poolienekohayo, hm, generally speaking hacking internal files isn't documented18:07
poolies//isn't a bug18:07
pooliebut you could put up a merge proposal that adds it to the documentatino18:08
nekohayobut for some use cases like this, I presume it can't hurt to strip out the trailing \n ?18:08
nekohayooh well18:11
=== Ursinha is now known as Ursinha-afk
nmendeshi, can anyone help me? I'm having this problem:21:07
nmendesbzr branch "svn://localhost/[delphi] dws gestor de licenças/trunk" C:\[dataworks]\Repositórios\test21:07
nmendesERROR: Unsupported protocol for url "svn://localhost/[delphi] dws gestor de licenças/trunk"21:07
maxbvila: Really, Won't Fix, on bug 701581? I'd say Importance Wishlist, give us a branch if you care21:07
ubot5Launchpad bug 701581 in bzr Upload plugin "Escape \n line returns in .bzr-upload.revid" [Undecided,Won't fix] https://launchpad.net/bugs/70158121:07
maxbnmendes: Do you have bzr-svn installed? Run 'bzr plugins' and see21:08
nmendesk, gonna try, tx21:08
vilamaxb: well.... what will be the point of making the plugin *less* robust ?21:08
maxbvila: Lots of bzr internal files are line based. I wouldn't begrudge a .strip() call in there21:09
vilamaxb: but this can only happen if the file has been... tampered ?21:10
maxbtrue, but people are encouraged to tamper with ~/.bazaar/*, .bzr/branch/branch.conf, so it's not that great a leap for people to try this21:11
vilamaxb: bzr-upload runs on a rather thin ice relying on this file to make sure it was the one doing the upload, editing this file will never be recommended so I don't want to encourage it either21:11
maxbBasically I agree it's not worthy of yours or my time to make bzr-upload more tolerant here, but I was surprised at the implication we'd refuse a branch even if offered21:12
vilamaxb: well, I'm not ready to accept such a branch so instead of saying no if one is proposed, I thought it was more honest to say upfront: "no, this is not a desired feature"21:13
maxbfair enough. I'd have suggested it was an acceptable change, but I doubt I'll ever use the plugin, so I don't have strong feelings on it21:14
vilamaxb: and it's probably the first time I used WontFix21:14
vilamaxb: not mentioning that rsync will probably does a better job reagrding bandwith than the plugin, so...21:15
maxbok :-)21:15
nmendesI have the bzr-svn plug in installed, the problem is the special char "ç" on the svn path, looks like it is tanslated  to "ç" and so the command fails21:16
nmendessame command with svn paths without those king of chars work as expeted21:17
maxbThat is odd. I would not have expected " Unsupported protocol for url " even if there was a charset bug of that sort21:18
nmendesjust did it:21:20
nmendesC:\Users\Nuno Mendes>bzr branch "svn://localhost/[delphi] dws meteo/trunk" "C:\U sers\Nuno Mendes\Desktop\[delphi] dws meteo" Branched 13 revision(s).21:20
nmendesC:\Users\Nuno Mendes>bzr branch "svn://localhost/[delphi] dws gestor de licenças /trunk" "C:\Users\Nuno Mendes\Desktop\[delphi] dws gestor de licenças" bzr: ERROR: Unsupported protocol for url "svn://localhost/[delphi] dws gestor de  licen├ºas/trunk"21:21
nmendeslooks like a bug, right?21:23
maxbIndeed, that looks like a bug21:23
maxbCould you try creating a bzr branch with an accented character in its name, to discover whether it is a bzr or bzr-svn bug?21:23
nmendesk21:24
jelmernmendes: what version of bzr-svn?21:24
nmendesC:\Users\Nuno Mendes>bzr init-repo --format=default "C:\Users\Nuno Mendes\Deskto p\a name with a ç just to test" Shared repository with trees (format: 2a) Location:   shared repository: Desktop/a name with a ç just to test21:26
nmendesinit -repo works ok21:27
nmendesTortoise Bazaar 0.5.821:28
nmendesbzr-svn 1.0.421:28
nmendesbzr branch "svn://localhost/[delphi] dws meteo/trunk" "C:\Users\Nuno Mendes\Desktop\brance_with_a_ç_to_test" Branched 13 revision(s).21:32
nmendeslooks like it's on bzr-svn21:33
pooliehi vila, all21:35
pooliei'm still upstairs21:35
jampoolie: be that way, then21:40
jam:)21:40
lifelessits all me :P21:42
lifelessjelmer: ncommander in the arm room would benefit from some 1:1 on bzr-builddeb - how to get it going on existing source+packaging branches that don't have import revisions etc21:42
jelmerlifeless: hey21:42
pooliei'm going to see about getting lifeless some help for his stiff neck^Wback21:43
* jelmer goes by the arm room21:43
vilapoolie: don't worry we're waiting for you to open the door21:46
poolie?21:47
poolieseriously? didn't i give you the card?21:47
vila:)21:47
vilasorry21:47
poolie:)21:47
vilaof course you did, we're in, just teasing you for making my laptop ring :)21:47
TheTinyToonHi Everyone - a collegue and me are testing Bazaar in a partner workflow and have a question regarding the steps to get started22:35
TheTinyToonFrom what I get, I first initialize a local repo, make some changes and commits and send a tarball over to my collegue22:36
TheTinyToonhe then branches from my tarball, makes some changes himself while I work on my local branch22:36
TheTinyToonour changes are then send by mail via bzr send and merged in our local branches22:36
TheTinyToonhowever, every time I try to send a mail after making some changes, I always the the warning "Bundling 0 revision(s)"22:37
TheTinyToonAny hints?22:37
lifelessyou probably are not holding a mirror of your partner's branch22:39
lifelessbzr send wants to compare two branches to see what needs to be sent22:39
TheTinyToonlifeless: I tried it locally by branching my repo into another directory, made some changes in the new branch and tried to send those back to myself - same error.22:40
TheTinyToon"bzr status" shows me the correct parent branch path22:40
lifelessyou may need to show a pastebin of your commands or something22:40
TheTinyToonk, one sec22:41
TheTinyToonhttp://pastebin.com/Z4QLz77s22:44
lifelessjam: might be interesting to run http://draketo.de/proj/hg-vs-git-server/test-results.html agains bzr22:45
TheTinyToonwe are currently evaluating Bazaar as a method for our students to get them used to effectively work in teams22:45
maxbTheTinyToon: It's often more convenient to use some shared server than email to pass around changes. Is there a particular reason you are focussing on 'bzr send' ?22:46
TheTinyToonmaxb: no particular reason - we just want to start without a central server22:47
TheTinyToon... or public repos at all22:47
spivjam: http://paste.ubuntu.com/552997/22:49
maxbOK, well the two arguments to bzr send are the submit branch and the public branch22:49
maxbThe submit branch is the branch you are submitting the changes to - or a mirror thereof22:49
TheTinyToonaah, switched the arguments22:50
maxbThe public branch is a public location holding a copy of the branch you are currently in - it's optional unless --no-bundle is in use22:50
maxbThe problem with having no central server is that it is hard for a group to have a good collective idea of what the latest master version is. Use of sent bundles generally works best when there's a clear hierarchy of how changes flow from person to person, which there generally isn't in a student project22:52
maxbUnless the project has deliberately been structured to exemplify this workflow22:52
TheTinyToonmaxb: yes, we are thinking about showing several workflows, one of which should be the "manual one"22:53
maxbNB that even when using 'bzr send' to send changes *up*stream, there's usually a read-only public location from which people can fetch changes downstream22:54
maxbi.e. developer A sends in a merge request to the project leader, developer B later pulls/merges from the project leader's public branch22:55
TheTinyToonyeah, classic human gatekeeper workflow - will definetly be shown.22:56
maxbIn which case it's pretty much "cd mybranch && bzr send ../gatekeeper-mirror"22:57
TheTinyToonjup - thanks for the help. We're testing it now...22:57
vilamaxb: just thought I'd mention your fix for the ssh leak associated with socketpair is about to land and the next submission in the queue should address the last failure with py27 on natty23:00
maxbaha23:01
vila:)23:02
jmlbtw23:32
jmlreally like seeing shelf status from 'bzr st'23:32
jmlthanks.23:32
jelmerjml: Blame Parth :-)23:34
jmljelmer: ok23:34
=== mnepton is now known as mneptok
jmlhmm23:40
jmlhow do I see the diff for a shelved item?23:40
mwhudsonbzr unshelve --preview23:40
spivwot 'e said.23:41
jmlahh yes23:41
jmlI don't know how I missed thta23:41
jmlmwhudson, spiv: thanks.23:41
lifelessI missed it the other day23:41
spivPerhaps because you didn't look at unshelve because you specifically didn't want to unshelve yet :)23:42

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