/srv/irclogs.ubuntu.com/2009/10/28/#bzr.txt

pooliedone!00:01
poolielow latency fwt00:02
poolieftw*00:02
poolieigc: i commented - what do you think?00:11
igcpoolie: can you explain more?00:16
poolieit seems unfortunate to me that the tests would need to mutate and then restore a library-wide variable00:17
poolieas opposed to just changing their instance of the tree00:17
pooliemaybe i'm massively confused?00:17
poolieor apparently you're relying on python looking up the name both instance-wide and class-wide?00:18
pooliethat works, but we really discourage it00:19
poolieit can be confusing or fragile00:19
igcpoolie: none of that is new. content-filters are registered library-wide and that was the easiest way to test them00:19
pooliei don't see anything else accessing WorkingTree._content_filter_stack00:19
poolieonly on particular instances00:19
igcpoolie: in testing, I need fine control over when filtering is on vs off00:20
igcpoolie: from memory I need to patch in at the class level ...00:21
igcbecause the tree is implicitly created by branching00:21
lifelessigc: filters are registered, but what about being enabled?00:22
lifelessigc: I mean, say I have two WorkingTree instances00:22
lifelesssurely they don't have the same filter stack00:22
lifelessif one tree has eol conversion enabled and the other doesn't00:22
igclifeless: we don't have per tree filtering00:23
lifelessI thought that that was all we had00:23
poolieigc, so it's a bit weird and likely to break eventually00:23
pooliei think at least you should add a comment there00:23
igclifeless: we only have global rules (much to my frustration)00:24
pooliebut it's not new code, so you don't need to clean it up to land this00:24
poolietherefore +100:24
pooliehth00:24
igcpoolie: it does. I'll submit the next patch after this one lands00:24
igcpoolie: can you update the review? I'll add the comment to the tests as requested00:26
pooliek00:31
lifelessjames_w: around?00:52
lifelessigc: I must say I'm confused as to why we have global filters only; I was sure we had agreed on an initial locations.conf/bazaar.conf based setup - which doesn't imply global library state at all00:54
igclifeless: it never got off the ground. I tried a patch along those lines and the design didn't hold up. I can't remember all the details now00:56
=== KhaZ_ is now known as KhaZ
lifelessok00:58
jelmerjohnf: hi02:01
johnfjelmer: howdy. Did you see my bzr check question?02:01
jelmerjohnf: no02:02
johnfone sec02:02
johnfjohnf: anyone know if there is a solution to this problem when running bzr check "reason: Parent not in inventory." It is from a branch that was originally branched from SVN02:02
johnfjelmer: ^02:04
jelmerjohnf: sorry, haven't seen that before02:04
jelmerjohnf: can you reproduce this somhow with a fresh import?02:05
johnfjelmer: problem is we moved on from using SVN as a backend. I suppose I could import and then apply all patches since02:06
jelmerjohnf: Does reconcile fix it perhaps?02:06
johnflet me try that02:07
tedgigc: So I can't get the svn-import to work.  How did you run the svn-import?  (what parameters)02:08
tedgigc: I've been figuring it was me, but I'm getting ready to blame bzr-svn :)02:09
johnfjelmer: nope http://pastie.org/67276202:09
johnftedg: Initially I didn't do an import since we where backing onto svn. I branched it02:10
igctedg: I can't remember. Let me try again and see if that jolts my memory ...02:10
tedgjohnf: ?  I'm not sure what you're saying.  I'm trying to do an import.02:11
arjenAUahoy oh gurus of bzr land, I come to you in a quest for enlightenment02:11
johnftedg: sorry thought you where referring to my problem. Just ignoe me02:11
arjenAUoh hi igc02:11
tedgjohnf: Ah, okay.  No problem :)02:12
igchi arjenAU02:12
arjenAUso, I have a branch with 2 extra local revisions (committed) and 1 remote revision committed. so pull says branches have diverged. fine. so I merge from the remote branch and it says nothing to do. guh?02:12
igctedg: I think I used bzr svn-import --all source dest02:12
jelmerjohnf: You might want to do a clean import using a new version pof bzr-svn and then branch your new work into that02:13
johnfjelmer: sounds like a good idea02:13
tedgigc: Okay, let me try that.  I wasn't using --all, I thought you were grabbing one branch.02:13
igctedg: which version of bzr abd bzr-svn do you have?02:13
igcand02:13
igcbzr plugins will tell you02:14
igcthe bzr-svn version02:14
johnfarjenAU: what does bzr missing --theirs-only say?02:14
tedgigc: 1.0.0dev02:14
jelmerjohnf: you'll want to use a new shared repository02:14
arjenAUjohnf: that i'm missnig one rev.02:15
johnfarjenAU: does bzr status say you have merges pending?02:16
arjenAUnops02:16
tedgigc: That gets me to an exception: AssertionError: Unable to find direct lhs parent for <CachingRevisionMetadata for revision 7680, path tags/start/experimental ...02:16
arjenAUjohnf: nop it seems cleanly committed on both ends, just diverged. so a merge should work I'd say.02:16
johnfarjenAU: ok if "bzr check" says everything is OK then I'm out of ideas and will hand over to a higher power :)02:17
arjenAUjohnf: yep ok02:17
igcarjenAU: sounds weird ...02:18
arjenAUigc: hehe yea. running 2.0 on jaunty from ppa.02:18
igcarjenAU: here's what I'd do ...02:19
arjenAUigc: i possibly forgot something, just can't think of what. it look sfine just doesn't fly02:19
arjenAUok listening02:19
igcgrab a copy of the remote branch locally02:19
igctry merging from that copy into your branch02:19
arjenAUahye and pull merge. ok02:19
igcarjenAU: I recall some weirdness ages ago where the branch had a change but the working tree wasn't refreshed02:22
igcso merge got confused and said nothing to do02:22
arjenAUnothing to do still...02:22
arjenAUshould I try merging the other way?02:23
igcarjenAU: one way to ensure the tree is in sync with the branch is to 'bzr revert', assuming you have no local changes02:23
arjenAUwell I do have local changes is the ponit02:24
arjenAUbut not uncommitted02:24
igcrevert is safe if your local changes are committed02:24
lifelessarjenAU: run 'bzr revision-info' in your local branch02:24
lifelessand with the url of youre remote branch02:24
lifelessalso pastebin 'bzr info' for both branches02:25
arjenAUurl of remote branch where?02:25
arjenAUrevision-info doesn't accept the bzr+ssh syntax02:25
tedgigc: Okay, it seems that appending part of the svn path helps.  Then it's not looking for the odd repository layout from the CVS import.  It's coping revisions now.  Which means my lap is getting hot from the laptop :)02:26
igctedg: :-)02:27
tedgOh my, bzr is using 1.5GB of RAM...02:27
igctedg: trunk has lots of memory improvements thanks to jam02:28
igctedg: if required, you can convert using that (safely)02:28
tedgI might have to... we'll see.  I only have 2GB physical.  If it hits swap, it'll never finish :)02:29
tedgNope.  Error'd out :(02:30
tedgHmm, why don't errors generate crash files?02:31
igctedg: if you throw the error in a bzr-svn bug report, jelmer is pretty good at solving them02:31
tedgI can't imagine this error is very useful: ERROR: Must end write group before releasing write lock on CHKInventoryRepository02:32
tedgWithout even a backtrace.02:32
tedgThat's all I got.02:33
igctedg: if you run with -Derror, you should see a backtrace02:33
igctedg: though it's surprising you didn't get one anyway02:33
arjenAUlifeless: revision-info doesn't accept the bzr+ssh syntax02:34
fullermdarjenAU: You'd have to use the '-d' arg.02:35
arjenAUlifeless: http://pastebin.org/4886702:43
fullermdarjenAU: What branch are you merging?02:44
arjenAUfullermd: remote into local02:45
fullermdYeah, but what's 'remote'?02:45
arjenAUthe one with the url02:45
fullermdThere are two URLs, and they're different.02:45
arjenAUoh both02:45
arjenAUbother sorry02:45
arjenAUhang on02:45
fullermdAre you sure you're missing'ing the same branch you're merge'ing?02:46
arjenAUindeed apparently not.02:46
tedgigc: Not enough memory to get the backtrace.  Tried to upgrade to nightly, but now bzr-svn won't work because of API version issues.  :(02:46
tedgWhy can't this information be encoded in to the packaging so that it's known at install?02:47
tedgIt's really annoying that bazaar breaks even when apt is happy.02:47
arjenAUfullermd: the plot thicken02:47
arjenAU+s02:47
SamB_XP_tedg: I'm going to blame jelmer ;-P02:48
lifelesstedg: what ppa are you using?02:48
SamB_XP_if only for not setting the PPA managers straight02:48
igctedg: ok, I have plenty of memory here. I'll try doing some of the conversion now02:48
tedglifeless:  http://ppa.launchpad.net/bzr-nightly-ppa/ubuntu02:48
lifelessso, nightly friction02:49
tedgigc: This is what works for me: $ bzr svn-import --all -Derror inkscape.svn/inkscape inkscape.bzr02:49
=== sidnei_ is now known as sidnei
tedgigc: Well, where "works" is generates an out of memory error :)02:49
arjenAUok fullermd lifeless igc the prob came from the fact that we have push, parent and submit branch urls... so it gets confusing.02:49
tedglifeless: Sure, but apt should stop bzr from installing then.  Or atleast warn me that the others need to be removed.02:50
lifelesstedg: nightlies dont have the right metadata02:51
lifelesstedg: they are pulling trunk, and there isn't a connection between trunk commits changing deps and the apt packaging02:52
lifelesstedg: same issue that bobthebuilder recipes need updating for02:52
tedglifeless: yes, but it seems that the API version is related to the bzr version.  So it would seem that the bzr-svn package in Karmic should have a depends on bzr (= 2.0.0) if it's going to only request and deal with a specific version.02:54
lifelesstedg: no, the API version is like a soname, it changes when we break things, not when a release is done per se02:54
lifelesstedg: what does dpkg -l bzr02:55
lifelessshow?02:55
tedgbzr            2.0.0+4766+12902:55
lifelessso02:55
lifelessdpkg thinks that you still have bzr 2.0.0 on your system02:55
lifelessas I said, metadata mismatch02:55
SamB_XP_lifeless: perhaps this metadata should be derived from something within the bzr tree somehow ?02:57
lifelessSamB_XP_: perhaps02:57
SamB_XP_like, maybe the API version ;-P02:57
lifelesstedg: you shouldn't use nightlies unless you're willing to deal with such mismatches, which will occur02:57
tedglifeless: Believe me, I don't want to use nightlies :)02:58
lifelessso why are you?02:58
tedglifeless: Because apparently they have memory fixes in them, that might bring down the memory of my import.02:58
tedgAs apparently I don't have enough physical memory.02:58
SamB_XP_tedg: why are you using physical memory anyway ?02:59
lifelessthere are significant memory improvements in trunk, thats true02:59
lifelesshowever igc is doing it now, right?:)02:59
tedglifeless: And I'm uninstalling the nightly as we speak :)02:59
igclifeless, tedg: rsynch'ing the inkscape svn repo as we speak02:59
tedgigc: If you could casually watch the memory usage as you're running, that'd be helpful.  If I'm close, I'd like to get this working as we're going to have to update with the final release hopefully in a couple days as well as pulling other sub-repos out of the SVN tree.03:03
tedg(website, docs, etc.)03:03
igctedg: sure03:03
igcbbiab03:06
* igc lunch03:38
=== mwhudson_ is now known as mwhudson
MTecknologySo.. loggerhead04:01
MTecknologybetter installed from bzr instead of apt?04:01
spivMTecknology: hmm, I've never tried installing it :)04:05
MTecknologyspiv: I'm trying to remember this and from what I recall from a year ago you run it and you don't really have control over what it listens to; I want it to run throguh apache..04:07
spivI'm pretty sure you can control that.04:08
MTecknologyok, cool04:08
spivSee the loggerhead.conf.example04:08
spiv(http://bazaar.launchpad.net/~loggerhead-team/loggerhead/trunk-rich/annotate/head%3A/loggerhead.conf.example)04:09
MTecknologythanks :)04:09
spivThe README has a section on serving it from behind apache, too.04:09
MTecknologyOH!04:11
MTecknologyI remember this now... apache proxy04:11
MTecknologyspiv: what's with the .1 files in there?04:21
MTecknologylike serve-branches.104:22
spivNo idea.04:22
mwhudson_man pages!04:22
spivOh, man pages.04:22
MTecknologymwhudson_: the readme file?04:23
mwhudson_MTecknology: no04:25
MTecknologymwhudson_: install from repos and use the man page then..?04:25
mwhudson_MTecknology: i may be missing bits of this conversation, i was updating the firmware on my router04:26
MTecknologymwhudson_: I'm using the head branch of loggerhead and I'm going to have it proxy throuch apache and hopefully make apache use system accounts to log into that (force ssl of course)04:28
MTecknologymwhudson_: I was just wondering whta th .1 files were for in the branch04:28
mwhudson_MTecknology: the .1 files are source for the man pages (as in the unix comand 'man')04:28
MTecknologyoh04:28
MTecknologymwhudson_: OH!04:28
MTecknologyLOL04:28
MTecknologythat's just funny now - I thought you were telling me to look at the man pages04:29
mwhudson_aaa04:29
mwhudson_sorry :)04:29
MTecknology:P04:29
MTecknologynow make ssh work for this other guy I'm trying to help. I'm looking on the server and the permissions are perfect - just not logging in using the key04:30
=== mwhudson_ is now known as mwhudson
pooliehi spiv?05:21
MTecknologyHow can I remove loggerhead? I ran the setup install but I'd rather run it right from teh directory..06:19
MTecknologyloggerhead is acting like it's running correctly, but it seems to be lying06:25
MTecknologyI'm getting this error now... I'm getting close :) Unsupported configuration: PasteDeploy not available, but loggerhead appears to be behind a proxy.06:43
MTecknologyYES!06:45
MTecknologyNow.... Is there any way you guys can have loggerhead show commit messages the way launchpad does on the project page?06:47
MTecknologyI know you can view a revision; it just won't let you see all the commit messages in one nice pretty list06:48
MTecknologyhurray....07:10
MTecknologypermission errors07:11
MTecknologyI really with there was an easier way to manage permissions - like the ability to manage permissions over http the way svn lets you...07:13
MTecknologyI'm going to need to look at how launchpad does it because this isn't working....07:13
fullermdPermissions...   pah.  Yet another pointless attempt to solve a social problem via technical means...07:14
fullermdYou can totally skip worrying about that sort of thing if you just kill off everybody who isn't authorized to access $WHATEVER.07:15
MTecknologyor can i... use apache permissions to manage directory access (which is already there) and run bzr over http07:15
MTecknologyfullermd: hm?07:15
fullermdIt's uncanny how many problems can be solved by the judicious application of mass murder.07:16
MTecknologyfullermd: The group is staying the same because I set the 'sticky bit' on the bzr directory. But that doesn't work because the sticky bit doesn't retain owner07:16
AfCfullermd: there's good money in that line of work, death-as-enforcement-of-filesystem-permisions07:16
MTecknologyI like the idea of being able to trust all developers.....07:17
fullermdchmod u+leadpipe colonelmustard.07:17
fullermdI suspect you're setting the setgid bit, not the sticky bit...07:18
MTecknologyI'd like to manage permissions the same way launchpad does but.... that would be hard07:18
MTecknologydrwsrws--t 5 shane kalliki-suite 4096 2009-10-27 23:56 .bzr07:19
fullermdThe sticky bit is unlikely to do anything useful for you there (and the setuid probably does nothing at all)07:19
awilkinsbzr-migration-docs : not building properly for me ... it leaves the body of the text out of the html page ; what's wrong? Wrong version in my toolchain?07:20
fullermdActually, ISTM I've heard of some combinations of VFS patches and config files on some OSen that cause the FS to force user ownership based on setuid dirs...07:21
fullermdBut that's pretty niche stuff.07:21
MTecknologyI was going for simple07:22
MTecknologyI was hoping I set permissions once and it just works07:22
AfCI'm not sure why sticky group isn't working for you, though07:22
MTecknologybut since you're right that setuid isn't doing jack.....07:22
AfCthat's what we use07:23
MTecknologythe gid is being retained07:23
MTecknologyjust not the uid07:23
fullermdAfC: On a current project, I've actually roped my neighbor into doing some work; by a weird coincidence, she happens to have some experience we can use on this project.07:23
fullermdIn giving her access to some files last week, she was all asking me what sort of legal blahblah I wanted her to sign about not touching unrelated stuff etc.07:24
AfCshared repository is mode 02775 group "bzr" and we make sure users' umask is 0022 and that they're all in group "bzr". Works like a charm07:24
fullermdI just gave her this quizzical look, and said "I know where you live."07:24
AfCheh07:24
AfCfullermd: yes, exactly. You don't need elaborate legal structures when the practical consequences are self-evident.07:25
MTecknologyAfC: can anyone at all touch any repo?07:25
fullermdDon't need to worry about umask [for commits in an existing repo], since bzr will handle retaining those perms itself based on...  uh...   not entirely clear.  Either .bzr or .bzr/repository or .bzr/repository/packs...07:25
AfCMTecknology: you mean any Branch in that Repository? Yes. But it is of course plain courtesy not to mess with other peoples' branches07:26
AfCMTecknology: without explicit permission07:26
AfCMTecknology: otherwise, fullermd's "I know where you live" kicks in quite admirably07:26
AfCMTecknology: not to mention `userdel offender`07:26
vilahi all07:26
MTecknologyAfC: I was referring to the repository, like /bzr/repo1 /bzr/repo207:27
AfCfullermd: not sure I'd trust that, even if I believed it, but good to know. Thanks.07:27
fullermdThere's no problem so big it can't be solved by killing the user off, removing their account, and reporting their REAL income to Internal Revenue.07:27
* fullermd waves at vila.07:27
MTecknologyAfC: Each repo will have a different group, but I want the uid to always be the same07:28
AfCMTecknology: in this case the repository in question is /bzr/java-gnome/hackers/...07:28
MTecknologyAfC: once uid is retained when a new file is created, I'll be happy and set07:28
fullermdMTecknology: You're going to have an...  'interesting' time sweet-talking a *nix system into letting you create a file as a user other than yourself...07:28
AfCie /bzr/java-gnome/hackers/andrew/enchant-updates and /bzr/java-gnome/hackers/guillaume/bug-455532-fixes07:29
MTecknologyfullermd: this kinda sucks - I need to keep this fine grained because I'll have to deal with people handling touchy info that I won't be able to fight back if they screw around07:30
* fullermd . o O ( cp /bin/sh /tmp/sh ; chmod +s /tmp/sh ; chown root /tmp/sh ; <fun> )07:30
lifelessMTecknology: bzr doesn't edit files, why do you need the uid kept?07:30
MTecknologyfullermd: My only other idea is to run it over apache07:30
MTecknologylifeless: when a user adds data, the next user can't pull it07:30
lifelessMTecknology: I think what you what is umask, not super special uid magic07:30
fullermdThe only way to get the files owned by a single uid is to only have them accessed by a [bzr] process running as that uid.07:30
lifelessMTecknology: uhm, details please07:30
MTecknologylifeless: h on07:31
MTecknologyhttp://paste.ubuntu.com/303413/07:31
MTecknologyactually... I'm not sure if that is a permission issue anymore07:31
MTecknologyumask helped a lot07:33
MTecknologylifeless: for something like this, do you think running the permissions over http might be better?07:35
MTecknologyor not really07:35
lifelessMTecknology: well, over http all your users would be the same user07:35
lifelessso have identical access and so forth07:36
* igc dinner07:36
lifelessMTecknology: that error is on push, in insert_stream07:36
lifelesscheck that the upload directory (.bzr/repository/upload) has group +w07:36
MTecknologylifeless: except that I could make each user authenticate as themselves07:36
lifelessyou'll want chmod -R g+w .bzr07:36
MTecknologynothing in there07:36
lifelessMTecknology: then it would be the same as ssh, wouldn't it?07:36
MTecknologyexcept same user/group on the system07:37
lifelessMTecknology: permissions, not content ;)07:37
MTecknologydrwsrws--t 2 root kalliki-suite 4096 2009-10-28 02:33 upload07:37
lifelessMTecknology: thats what I said though, 'all your users would be the same user'07:37
MTecknologywhy's that?07:37
lifelessMTecknology: so, is the user getting the error a member of kalliki-suite07:37
MTecknologyyup07:37
lifelessok07:38
lifelesscheck their ~/.bzr.log07:38
lifelessmay have more detail07:38
MTecknologyhttp://paste.ubuntu.com/303417/07:39
MTecknologyoh... it was svn that I remember being able to handle granular permissions07:40
pooliespiv: nice mail re imports, thanks!07:40
lifelessMTecknology: on the server, the .bzr.log07:40
MTecknologyhttp://paste.ubuntu.com/303419/07:41
MTecknologyalmost 3am again07:42
MTecknologyI really never can just sit down and hammer something out quick07:42
MTecknologyIsn't there any documentation to handling groups that access bzr repos?07:42
lifelessMTecknology: its permissions07:42
MTecknologyI'd have no issues starting fresh07:42
lifelesschmod group write access within the repo07:42
MTecknologyg+w?07:43
lifelessyes07:43
MTecknologysame thing07:43
MTecknologyit was already that way :(07:44
lifelessls -lR .bzr/repository07:44
lifelesspastebin that07:44
MTecknologyhttp://paste.ubuntu.com/303421/07:45
AfCroot?07:46
MTecknologyoh...07:46
lifelessMTecknology: and the directory itself07:47
lifelessMTecknology: ls -ld .bzr/repository07:47
MTecknologydrwsrws--t 7 root kalliki-suite 4096 2009-10-28 02:06 bzr/suite-dev/.bzr/repository/07:48
lifelessok07:50
lifelessso07:50
lifeless-rw-rwx--- 1 michael kalliki-suite  181 2009-10-28 02:31 3c29bfe02aba60925bb1e0d04349b994.rix.1864.carpo.jr7tsonnb7.tmp07:50
lifelessis noise from one of the errors07:50
lifelessyou can see that its an index that your root user has already written07:51
MTecknologylifeless: can I just delete it?07:51
lifelessMTecknology: hell no07:52
MTecknologyok07:52
lifelessso, you definitely don't want the s bit set07:52
lifelessthats setuid07:52
MTecknologyu-s07:53
MTecknologyor ug-s ?07:53
lifelesson files07:53
lifelesson dirs +s is ok07:53
lifelessthe t bit is what is making you get errors07:54
MTecknologyoh07:54
lifeless"For directories, it prevents unprivi‚Äź07:54
lifeless       leged users from removing or renaming a file in the directory unless they own the file or the directory; this is called the restricted deletion flag07:54
lifeless       for  the  directory"07:54
lifeless(see man chmod)07:54
MTecknologythat was the problem.........07:54
lifelessyes07:55
MTecknologyso, when a different user tries to commit, will there be an issue?07:55
MTecknologyif one user commits a brand spankin' new repo and another user tries to make a commit over it07:55
MTecknologyI suppose as long as the group is retained07:57
MTecknologylifeless: thanks - I love you07:57
poolielifeless: good night08:02
lifelessniht poolie08:03
bialixhi all11:01
=== obstriege is now known as obst
jmlNumber of times I have run pqm-submit, got an "public branch out of date" error and _not_ wanted to immediately push to the public branch:11:40
jmlZero.11:40
spivjml: shell scripting to the rescue! ;)11:42
spivjml: btw,11:43
SamB_XPspiv: how is shell scripting going to help? you can't integrate that into bzr very easily ...11:43
spivjml: https://code.edge.launchpad.net/~spiv/bzr-pqm/non-local-submission/+merge/1333311:43
spivSamB_XP: it has while loops, bzr has exit codes, ta da!11:43
spivSamB_XP: (also, note the wink...)11:44
SamB_XPbut it's not an 0.6 wink ...11:45
spivIndeed.  Perhaps I should segment my eyelids...11:46
spivUntil then, I think I'll stick with whole winks.11:46
jmlspiv, I notice that that patch doesn't change any tests11:48
spivjml: true, and I didn't even try running them.  I just bashed it into shape enough to solve my immediate problem.11:49
jmlspiv, heh :)11:50
spiv(Which was there were a handful of things in bzr's review queue ready to land that I wanted to send to PQM without mucking about... I guess hacking on bzr-pqm means I failed ;)11:50
spivI expect the next person (possibly me) that needs this will bash on it again, hopefully touching tests :)11:51
jmlspiv, I've reviewed your patch, fwiw.11:53
spivjml: heh, ta11:54
spivWow, I'd already forgotten most of this patch :)11:55
spivI really should extract the config improvements, they'd be handy for writing some kinds of hooks I think.11:56
spivAnyway,11:57
spivg'night11:57
=== lamont` is now known as lamont2
=== lamont2 is now known as lamont
Bear10When does bazaar lock files?12:14
=== mrevell is now known as mrevell-lunch
=== mrevell-lunch is now known as mrevell
=== vds1 is now known as vds
vilajam: ping14:14
jammorning vila14:15
vilamornign jam !14:15
awilkinsMmmm. Morning jam. On toast.14:21
matkorwhere is doc explaining what is needed when migrating from 1.x to 2.x ? And what are gains (except being in supported version), and potential pitfalls ? TIA15:22
Peng_matkor: http://doc.bazaar-vcs.org/latest/en/upgrade-guide/index.html15:29
Peng_matkor: You mean upgrading your bzr install or your files?15:30
matkorPeng_: Yes, thank you15:32
=== deryck is now known as deryck[lunch]
=== beuno is now known as beuno-lunch
=== cjwatson_ is now known as cjwatson
=== james_w` is now known as james_w
faheemhi. was running bzr upgrade on the mysql repos from lp. been running overnight and is consuming 1.6G memory. is this normal?16:52
cjwatsonhow do I find out what version of bzr is running on a remote smart server?17:01
cjwatsonI don't have shell access, but do have bzr+ssh access17:01
cjwatsonnever mind, hacked bzrlib to tell me :-)17:05
beuno-lunchfaheem, it is17:07
beuno-lunchif it fails, maybe we can help you and upgrade it in the datacenter17:07
=== beuno-lunch is now known as beuno
=== rockstar is now known as rockstar-afk
=== khmarbaise__ is now known as khmarbaise
=== deryck[lunch] is now known as deryck
=== abentley is now known as abentley-lunch
=== phinze_ is now known as phinze
=== rockstar-afk is now known as rockstar
=== abentley-lunch is now known as abentley
faheembeuno: ok18:35
faheemthanks18:36
faheemi'm not clear what this bzr upgrade is for, though. can someone clarify?18:38
faheemi'm using bzr 2.0.1. just bzr branched mysql repos from lp18:38
faheembzr: ERROR: Views are not supported by <WorkingTree4 of /tmp/mysql-server>; use 'bzr upgrade' to change your tree to a later format.18:39
faheemthat's the error message i get18:40
faheemi think i was trying to do bzr view or some such18:40
* awilkins has never used `bzr view` and only just learned of it's existence18:47
awilkinsLooks like it's the bzr answer to the "staging area" in git18:48
awilkinsOr something like it18:48
awilkinsfaheem: It may be that `view` requires an experimental working tree format18:49
awilkinsBut it seems to work on Working Tree Format 618:52
=== asac_ is now known as asac
igcmorning19:04
igcawilkins: views will work in 2a19:05
awilkinsMorning Ian19:05
igcawilkins: it's a bit rough around the edges though - some edge cases don't work frm memory19:05
igce.g. shelve?19:06
faheemyes, but i'm not sure from what to what the upgrade is. it's not very explicit19:06
faheemi assume there are repos formats?19:06
igcawilkins: for simple stuff, it should be fine19:06
awilkinsfaheem: Ah, that's an upgrade of repository and working tree to the newer format. For MySQL it could take a while....19:06
igcfaheem: MySQL are upgrading soon I believe19:07
igcfaheem: if you branch in bzr, it keep the source format19:07
igckeeps19:07
faheemawilkins: i think it has been running overnight.19:07
faheemlooks relatively close to completion19:07
faheemhow many different active repos formats does bzr have?19:08
awilkinsigc: Looking at the VSS converters I think vss2svn is clearly the best of the bunch... all the ones that depend on SS.exe are hobbled by the privacy of much of the code in the stack19:08
faheemis bzr view a builtin?19:08
igcfaheem: there's only 1 default at a time19:08
awilkinsigc: It's come on a lot since I last used it - full rewrite :-)19:08
faheemi recompiled on debian lenny with unstable sources19:08
igcfaheem: for bzr 1.x, it was "pack-0.92"; for bzr 2.x, it's 2a19:09
faheemdidn't ask me for anything for viewing stuff19:09
igcfaheem: bzr view is builtin, but only supported on 2a formats19:09
faheemigc: how many of these are supporting wrt to conversion?19:09
awilkinsfaheem: view is native ; but it's new to me, I've never really had a need for it19:09
igcfaheem: you can forward convert from pack-0.92 to 2a but not go back the other way19:09
faheemigc, awilkins: i see19:10
faheemso two formats?19:10
igcfaheem: there were additional formats released over the 1x timeframe but they never became the default19:10
faheemigc: what happens if a repos is using them?19:11
faheemi mean wrt conversion?19:11
awilkinsfaheem: If your local repository is a different format, it converts revisions on the fly19:11
igcfaheem: and it was messy - some had "rich roots" as needed by bzr-svn while others didin't19:11
awilkinsfaheem: This won't work backwards from 2a to 0.92 though19:11
igcfaheem: 2a fixes that mess - everything is "rich root" from now on19:12
awilkins2a is a rich-root format, 0.92 is not19:12
faheem awilkins,  igc : i see. thanks.19:12
faheemso you are stablizing on the 2a format?19:12
faheemany further repos format changes anticipated?19:12
awilkinsOn past history, you can expect more optional formats that add performance or features19:13
faheemis there a history of these changes on the net? just curious.19:13
awilkinsIn newer versions of Bazaar... I don't think they'll be backported to 2.0.x series releases19:13
awilkinsfaheem: `bzr help formats`19:14
faheem awilkins : great! thanks.19:14
awilkinsLooks like that help doc needs a bit of tickling, it's out of date WRT to 2.0.x19:15
igcfaheem: we may introduce another format in the future but that won't happen for at least 12 months19:18
awilkinswhat, no 2.1-super-fast-omg-ponies  ??  :P19:19
igcfaheem: if we do, it will be to add new features (and there will be some performance/storage benefits most likely as well)19:19
igcawilkins: no, not new format in 2.119:19
awilkinsigc: I think the only thing I really want to see improve right now is nested tree support and packing times19:20
igcawilkins: nothing public at least - there may be some evolution of the "development" one19:20
awilkinsbzr pack is significantly slower on 2a than the equivalent 1.14-rich-root for a fairly minimal space gain in most of my own cases.19:21
igcawilkins: can you report a bug with the figures re pack?19:21
igcawilkins: jam or lifeless will take a look most likely19:22
jamawilkins: 'bzr pack' on 1.14 rich-root probably had 0 space gains19:24
jam'bzr pack' in 2a has huge wins, except we recompress on the fly, and recompress after a conversion19:24
jamso all the gains have already been 'one'19:24
jam'won'19:24
jamif you give it a couple months, you could see more significant changes19:25
jamthough we certainly could have a "bzr pack --lite" which uses our current 'recompress-on-the-fly' code19:25
awilkinsjam: I think I saw about 100MB -ve delta out of 800M repo size which is definitely smaller, but the packs take a lot longer19:26
jam(recompress only groups that don't look compressed)19:26
jamawilkins: well as we are actually doing real work in 2a, I expect the time to go up :)19:26
jam'bzr pack' is ~ git repack -a -f19:26
awilkinsjam: It was enough to provoke me to split the repo into mutliple repos (one for each tree) instead of hosting all the trees in a single repository19:27
jamawilkins: why the need to force repack?19:27
awilkinsjam: Not just forced repacks, autopacks too19:27
jam(as in, it shouldn't be something that you really need to do)19:27
awilkinsjam: They're noticeably a lot longer19:27
jamautopacks would be slower than before, but not something I would expect to be noticeable except on major boundaries19:27
jamthe 1 time I've seen it is when I rolled over 1k revisions to repack19:28
faheemwho is the current bzr lead developer? is martin pool still working on it?19:28
awilkinsjam: I might be an opposite corner case ; my trees are not that deep, but they are quite wide in many cases19:28
jamfaheem: poolie is our leader, yes, though he doesn't do quite as much coding as he used to19:28
awilkinsSome of those trees have 1.9GB of text files in them :-)19:29
jamawilkins: sure, but how much 'churn' do you have in that 2GB?19:30
awilkinsjam: Oh, not a lot, except when a new folder with 150MB of text gets added... but just it's presence in the same repo as other trees seemed to slow down autopacks19:30
awilkinsI know this is all wishywashy with no metrics, but it's an assertion based on 1.14-rich-root autopack times never bothering me enough to go "grr"... and 2a does on occasion.19:31
gioelewill 2.1 be the next stable version or are you adopting a odd-dev/even-stable numbering scheme?19:32
awilkinsRight, time to eat some pizza.... sorry to duck out, back later19:32
jamgioele: 2.1 will be next stable, 2.1.0b1 is the 'unstable' new release19:32
jamfollowed by b2, b3? rc1 then 2.1.0 final19:32
gioeletime based betas?19:33
jamgioele: yeah, monthly releases19:33
gioeleI see19:33
jamsame as before, we are just labeling them differently now that we hit 2.019:33
jamwe'll also be doing ~ monthly 2.0.x releases19:33
jamwith only bugfixes19:33
gioeleThat seems a very good thing to do. Suggests more stability than the 1.x cycle19:34
gioeleanother question: what is the correct way to generate the apidocs? pydoctor refuses to work and "make api-docs" seems no longer supported19:35
jamgioele: use epydoc < 3.0?19:36
jamISTR you filing a bug report about epydoc 3.0 not working19:36
jam:)19:36
jamI would guess 'make api-docs' is supposed to be the way to do it, and we just need a patch to ensure compatibility with newer epydoc versions19:37
gioelejam: that is available only on dapper19:37
mwhudsongioele: in what way is pydoctor barfing?19:37
jamISTR we had to monkeypatch some doc generation tools because they didn't handle some of our formatting19:38
jamwe had a 'foo-bar' in one of our keys that wasn't liked19:38
gioelemwhudson: I created this .cfg following other bzr plugins: http://pastebin.com/d3c084a7919:41
gioelethen pydoctor quits with Exception: you must pass a package directory to preprocessDirectory19:41
mwhudsongioele: you don't want the "prependedpackage" bit for bzrlib itself19:42
gioelemwhudson: even without it complains in the same way19:42
mwhudsongioele: what commandline are you using?19:43
jammwhudson: would it be "packages: bzrlib" and no prependpackage?19:43
gioelemwhudson: pydoctor -c bzr.cfg --make-html19:43
mwhudsonjam: ah, yes, probably19:44
gioelejam: yes, that seems to work19:44
mwhudsonah man, i need to find some time to give pydoctor some love :/19:44
jammwhudson: you have 10 minutes starting....19:45
jamnow19:45
jam:)19:45
gioelere19:48
mwhudsondear me, i'd forgotten how slow the rest parser is19:51
mwhudson"1320/3756 pages written"19:51
gioelemay I suggest this tiny patch to Makefile: http://pastebin.com/d37a5e57a ?19:54
=== khmarbaise_ is now known as khmarbaise
meoblast001wtf does this mean "This transport does not update the working tree of: bzr+ssh://meoblast@bazaar.launchpad.net/~mysticgalaxies/amethyst-mm/libaaf/. See 'bzr help working-trees' for more information."20:32
luksmeoblast001: that the remote working tree on the server will be out-of-date20:39
luksbut in the case of launchpad, I don't know :)20:39
meoblast001luks: i just downgraded the new tree to 1.920:39
nyuwhile importing svn into a pristine 1.14-rich-root using bzr 2.0.1: http://pastebin.com/m2f1454f321:25
pooliehi jam21:25
nyuand bzr-svn 1.0.021:25
jamhey poolie21:25
thumperpoolie: can I have a call with you soonish?21:31
jamigc: are you around?21:31
pooliethumper, yes, i was just going to suggest that21:34
poolienow works for me21:34
=== jam changed the topic of #bzr to: Bazaar version control system | 2.0.1 and 2.1.0b1 are official! | try https://answers.launchpad.net/bzr for more help | http://bazaar-vcs.org | http://irclogs.ubuntu.com/
Peng_jam: Congrats on the releases. :)21:54
jamyeah, finally21:56
jamonly took me 2 weeks21:56
Peng_jam: The CHK index thing sounds awesome too. Have I mentioned that I love you? :D21:58
jamand now that 2.1.0b2 was supposed to be released yesterday...22:06
Peng_Heh, oops.22:06
Peng_How current is the code in 2.1.0b1? Is it mostly 2 weeks old?22:07
jamPeng_: 2.1.0b1 was 'cut' 2 weeks ago22:09
jamso no changes since then22:09
Peng_Ah.22:09
jamjust had to get the installers built22:09
jamand announcements made22:09
jam2.1.0b2 will include the StaticTuple stuff22:09
Peng_When are you going to do 2.1.0b2, then?22:09
jamI posted to the list22:10
jamWe'll find out22:10
jam2.1.0b1 was already 2-weeks behind when it went gold, so it is officially 1 month late22:10
jamso either we'll release 2.1.0b2 right away to catch up, or wait a month22:10
jamanyway /me gone22:10
Peng_jam: See you later. :)22:10
Stavroshello22:13
Stavroshas anyone managed to push to github with bzr-git?22:13
Stavrosanyone?22:17
Stavrosjelmer?22:20
hnoDownload link for 2.1.0b1 is wrong on the home page.. links to /2.0/ instead of /2.1/..22:22
Stavrosis bzr-git also supposed to be able to commit to git branches?22:26
lifelessjam: do you think chk specific index is a big enough win to be doing another repo format?22:27
lifelessjam: or are you just tracking down all the memory inefficiencies as aggressively as possible?22:27
lifelessStavros: I believe so22:27
Stavroslifeless: turns out i have to do git+ssh, but then it can't find the repo :/22:28
lifelessStavros: the path is from /, not ~22:28
Stavrosoh good, it works22:29
Stavroswhat do you mean?22:29
Stavrosoh right22:29
Stavrosso /Stavros instead of :stavro22:29
Stavrosthat's how it worked, thanks22:29
Peng_hno: Eh. I'll send a merge proposal, unless someone else gets to it first.22:29
Peng_lifeless: ping22:38
lifelessPeng_: ?22:38
Peng_lifeless: I'm being pushy, but if you have rights to the website, could you land a trivial patch that fixes a broken link on the homepage? It's kind of bad. :\ https://code.edge.launchpad.net/~mnordhoff/bzr-website/2.1.0b1-link/+merge/1412122:40
lifelessI'm not entirely sure where the new site is deployed to22:40
lifelesspoolie: ^22:40
hnoheh, not every day I stir up this much over a single digit..22:41
lifelessPeng_: it would be good if you can also submit a merge to update the docs to note that:22:41
lifeless - releasing requires access to deploy a new website22:41
Peng_hno: :D22:41
lifeless - the release manager needs to do this22:41
Peng_Eh.22:41
* lifeless loathes static websites22:41
lifelessPeng_: have I misunderstood whats wrong?22:42
Peng_Yeah, me too. I don't have an anti-static wrist strap.  (Sorry, that was terrible.)22:42
Peng_lifeless: No.22:42
Peng_Well, the website *was* deployed. There was just a typo.22:42
hnohi lifeless btw22:42
lifelessok22:42
lifelesshi hno :)22:42
poolieyou just need to merge it to the branch22:42
poolieto the bzr-website trunk22:43
poolienothing else is required22:43
lifelessoh ok22:43
lifelessthats good22:43
pooliepatches to make it dynamic welcome ;)22:43
Peng_releasing.txt already mentions it:22:43
Peng_#. Announce on the `Bazaar website <http://bazaar-vcs.org/>`_. This page is edited via the lp:bzr-website branch. (Changes pushed to this branch are refreshed by a cron job on escudero.)22:43
lifelessPeng_: are you able to push to the trunk ?22:43
lifeless(and if not, why not :P)22:44
Peng_lifeless: I'm not a member of ~bzr.22:44
lifelesshmm, we should fix that22:44
lifelessnevertheless22:44
Peng_(That was three lines on my clipboard. How did it get magically converted into one nicely-formatted line?)22:45
Peng_lifeless: Well, I wouldn't mind joining. I never thought about it before.22:45
lifelessyour branch is pushed22:48
Peng_lifeless: <322:48
lifelessand you should join, I think.22:48
lifelesspoolie: ^22:48
Peng_I wonder how often the cronjob runs?22:48
poolie+122:52
poolielifeless: you already merged it to trunk?22:52
poolieroughly every hour22:52
pooliethis is a great example of why it should be scripted btw22:52
poolieand why a wiki is no better22:52
lifelesspoolie: yes22:53
lifelessI'm popping out to get my hair cut; mobile is on.22:54
Peng_Ah, the website was rebuilt at XX:50:03, which was immediately after the new revision was pushed.23:16
Peng_So the cronjob runs at least every 10 minutes.23:17
hnobzr+bzrtools 2.1.0b1 now packaged for Fedora-13/Rawhide.. and 2.0.1 pushed for Fedora-12 which means it should make it into the final release (2.0.0 currently there).23:21
Peng_Cool. :)23:24
* igc out for a while23:43

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