/srv/irclogs.ubuntu.com/2009/05/18/#bzr.txt

igcmorning00:04
=== abentley1 is now known as abentley
=== davidstrauss_ is now known as davidstrauss
=== abentley1 is now known as abentley
=== abentley1 is now known as abentley
=== abentley1 is now known as abentley
krisfremenis there a way to get only one file from a revision?01:34
Peng_krisfremen: "bzr cat -r 123 foo.py"01:35
krisfremenawesome thanks01:36
AfCkrisfremen: which is also really handy when you have a forest of Branches that do not have Working Trees present with them01:44
Peng_Indeed.01:44
AfCkrisfremen: and you need one or so files out of it01:44
krisfremenyes, that's what i need it for01:45
jelmerigc: hi02:04
jelmerigc: I've just added http://bazaar-vcs.org/ConvertCommand02:04
jelmerigc: it touches on upgrade a bit as well, since they might share common code02:04
igchi jelmer02:04
igcjelmer: thanks for the rio stuff02:05
igcjelmer: one thing that would help though ...02:06
igcis to move the escape logic down somehow ...02:06
igcso that knowing whether to xml_escape or not doesn't need to be handled at higher layers multiple times02:06
jelmerigc: the problem is that we don't know if we're re-serializing an existing revision02:07
igcI had to tweak fast-import to disable it and I imagine ...02:07
igcthat you'll need to do the same in svn-import, etc.02:07
jelmerigc: we could always escape in the XML serializer02:07
igcjelmer: or even the repo layer02:08
igcI just want to get it out of commit.py, fast-import, etc.02:08
jelmerigc: yeah, that would be really useful02:08
igcjelmer: if nothing else, we might need a flag on the format saying whether to escape messages or not02:09
igcjelmer: btw, I fast-import OOo over the weekend02:10
igc36 hours altogether, though only 1 hour for the first 50K so ...02:10
igcthere's room in there to improve things a lot I think02:10
jelmerigc: into dev6 or into 1.9-rr ?02:10
igcdev6-rr02:11
jelmerigc: about the xml char escaping, do you think it would be acceptable to just push it into the serializer and not have commit print warnings about squashed characters?02:13
jelmeralternatively, I guess we could propagate that number somehow02:14
igcjelmer: I think the serialiser is the logical place for it but ...02:14
igcI worry about backwards compatibility02:14
igcthat's why I suggested the repo layer02:14
jelmerigc: backwards compatibility in what sense?02:15
jelmerigc: we don't really escape special characters at the moment, we squash them (we don't escape "\" itself)02:15
igcjelmer: if plugins assume the serialiser does or does not do it02:15
jelmerigc: it's not a problem if plugins escape too02:15
igcjelmer: I guess not02:16
igcjelmer: hopefully our tests we tell us :-)02:17
h6wGday.  I have an existing svn+ssh repos, and I've installed bzr-svn.  What would the url be for the bzr exposure of the svn repos?02:36
jelmerh6w: same URL you would use with svn02:36
h6wbzr+ssh://.....?02:36
jelmerh6w: no, svn+ssh://..02:36
h6wOh really?02:36
h6wSounds a bit confusing, but, oh well.02:37
jelmerh6w: well, it's the svn+ssh:// protocol that you're using with bzr02:37
jelmerh6w: bzr+ssh:// would be the bzr protocol02:37
h6woic.  I installed the bzr-svn plugin on the server, expecting it to do the translation server-side.02:38
h6wSo I'd just use bzr on my client.02:38
jelmerh6w: that might work too (with bzr+ssh:// then, but I've never tried it)02:38
h6wok, will try that.02:39
jelmerh6w: looks like that doesn't work for some reason02:43
jelmerspiv: ping02:43
jelmerh6w: so you'll need bzr-svn on the client instead and use svn+ssh://....02:44
h6wok.  Or else just shift the entire repos.02:47
h6wThe whole point was so that we could do a smooth migration from svn to bzr.02:48
jelmerh6w: you can do such a smooth migration that way, it's just the URL that's different02:49
jelmerh6w: if you later convert the repo on the server from svn to bzr using "bzr svn-import" you just adjust the URL and things will still work02:50
h6wI think that's going to have to be the way to go.02:50
h6wI was just hoping to have commits and updates work through both protocols while we tested it.02:50
h6wWe have one client (Launchpad) that doesn't do bzr-svn, only bzr.02:51
jelmerh6w: it's probably an easy fix to support bzr+ssh:// for svn repositories, it's just that nobody has asked for it before02:51
jelmerlifeless,spiv: ^02:52
lifelessjelmer: huh?02:52
lifelessjelmer: do you mean, bzr-svn running on the server?02:52
jelmerlifeless: yeah02:52
h6wYes.02:52
jelmerlifeless:  the smart server raises NotBranchError for some reason, any idea what could be causing that?02:52
lifelessdoesn't really make, sense that would be a site-to-site thing like 3rd-party ftp02:52
lifelessor perhaps I don't understnad what you mean02:53
h6wServing bzr requests via bzr url and piping them to an existing svn repos on the same server.02:53
jelmerlifeless: Using bzr+ssh:// to access a Subversion repo on the server02:53
lifelessjelmer: bzr-svn would need to find an svn repository given a file url02:53
SamBjelmer: why should that be needed?02:53
jelmerlifeless: it can do that02:53
lifelessjelmer: into the repo, not into a tree02:53
SamBwhy not just use svn+ssh ?02:54
jelmerSamB: see above02:54
jelmerlifeless: yes02:54
lifelessjelmer: then it should, in principle, just work02:54
jelmerlifeless: What's the easiest way to debug if it doesn't?02:54
SamBhmm, how do you break an ssh-started program into a debugger ...02:55
lifelessjelmer: use raw bzr://02:55
lifelessjelmer: start a server via bzr serve02:55
lifelessand then use pdb and or ctrl-\ juidiciously02:56
SamBoh, I guess that would work ...02:56
SamBjelmer: why don't you just check ?02:56
lifelessits exceedingly unlikely that ssh is part of the problem02:56
SamBoh, you did02:56
* SamB wonders why he said "if"02:56
lifelessdid you?03:03
jelmerlifeless: it loads the plugin but never even calls the probe function03:03
lifelessjelmer: the smart server uses the standard interface03:04
lifelessjelmer: break in bzrlib.smart.bzrdir .. OpenBranch03:04
SamBlifeless: standard interface ... ?03:04
lifelessSamB: as opposed to a nonstandard one03:05
SamBlifeless: do you have a link to the documentation of this interface?03:07
=== Kissaki is now known as Kissaki^0ff
jelmerlifeless: found the problem03:09
jelmerlifeless: it only tries to do an open with BzrDirMeta103:09
jelmerlifeless: bzrlib/smart/bzrdir.py:5203:09
lifelessoh03:10
lifelessfugly03:10
lifelessiz bug03:10
jelmerfiled03:13
jelmerlifeless: oh, and ChrootTransport doesn't provide local access?03:25
lifelessjelmer: no03:33
lifelessit provides whatever its back on, no more03:34
* SamB wonders what ChrootTransport is ...03:49
AfCWhat's the bzr-svn URL now? I have http://people.samba.org/bzr/jelmer/bzr-svn/trunk/ but it's not redirecting anywhere.03:56
SamBAfC: symlinks don't usually cause HTTP redirection ...03:56
AfCSamB: apache configurations do, though03:57
SamBAfC: well, I believe that's always going to be the trunk of bzr-svn, and I don't think it's ever redirected for me unless I left off the /03:58
AfCSamB: oh; so you believe that URL to be correct?04:00
SamBAfC: assuming you're not getting an error04:00
SamBoh ... actually, I'm using http://people.samba.org/bzr/jelmer/bzr-svn/bzr.dev/04:01
AfCSamB: ok, thanks04:02
* SamB was just confused by AfC's expecting it a redirect04:02
* igc lunch04:20
lifeless-> spain05:12
poolielifeless: ping? still in sydney?06:34
bob214:12:17 < lifeless> -> spain06:34
bob2poolie: ^06:35
poolieah i thought so06:35
lifelesspoolie1: at the airport06:58
lifelesspoolie1: can I do something for you?06:58
lifelesspoolie1: you have a 10 minute window before I'm forcible un-wifi'd07:00
lifelessok, boarding call ;)07:07
lifelessciao07:07
Peng_Bye. :)07:10
lifelessactually, misheard :)07:12
lifeless16:30 I have to board07:13
Peng_14 minutes from now?07:16
Peng_Well, enjoy your last 13.5 minutes of wifi. :)07:16
lifelessyes07:16
lifelessthough pinging my home client could be wrong :)07:16
abeaumontgood luck with jet lag... since it's 8:17am here07:17
Peng_So, um, why are you guys always flying all over the world?07:18
lifelesswe get together for sprints07:19
lifelessdifferent areas of the company do this at different times07:19
lifelessthis particular one is everyone in the company, which is way cool07:19
Peng_Oh, okay.07:19
lifelessand happens ~ once a year07:19
lifelesswow, DCO's look evil07:20
poolie1hello lifeless07:20
poolie1no, nothing important07:20
poolie1DCOs?07:20
lifelessdevice configuration overlay07:20
lifelessfor vendors that want to sell '60 GB drives' but their supplier gives them 65GB drives.07:21
lifelessapply a DCO and most everything sees it as 6007:21
lifelessjust reading up on HPA's due to finding the BIOS HPA support on my desktop flaky, I decided I should know more about it07:21
lifelessand the best doc I've found so far is from the journal of investigative forensics :)07:23
* lifeless waves07:32
poolie1fly well07:39
Peng_Fly well? The airline industry is doing so badly that passengers have to flap their arms now?07:40
awilcoxHi, is there a facility on Bazaar similar to subversion's tags?08:08
awilcoxI would ask my bzr repo's maintainer but he has been offline for quite a while so I cannot08:08
Peng_awilcox: Bazaar supports tagging, but branches are more similar to svn tags.08:10
awilcoxAh.08:10
fullermdPeng_: Well, if you believe the version control/airlines thing...08:10
awilcoxMy project is closing in on a release and I'd like to "branch" the tree I guess, so that this version can be out of trunk and be worked on separately08:11
Peng_awilcox: Then use a branch.08:11
awilcoxThat way, last-minute bug fixes can be done by the QA devs on the branch, and me and the rest of the team can be adding shiny new features to 2.008:11
awilcoxOkay.08:12
fullermdHave a branch for the release engineering, and tag the release on it, is SOP.08:12
awilcoxfullermd: is there a tutorial for doing this?  Because I looked on the wiki and it does not say how to make a branch08:12
awilcoxI am an utter n00b when it comes to bazaar.  I mean, my repo maintainer taught me how to use checkout and resolved and said "anything else read the manpage"08:13
awilcoxAnd the manpage doesn't seem to cover tags and just mentions what branches are, not how to to create one =/08:13
fullermd`bzr branch foo bzr`08:13
fullermdAargh.  Dangit.  bar.  bAr.  With an A.08:14
Peng_:D08:14
awilcoxHah.08:14
awilcoxSo it would be like bzr branch trunk release ?08:14
awilcoxWhere do I run that from?  I have ssh to the server...08:14
fullermdWell, one general way would be to put it right next to trunk.08:15
awilcoxRight.08:15
awilcoxIt would be like foo.com/project/trunk  or foo.com/project/release08:15
Peng_Unless you only want to make one release, it might be better to name the branch "1.2" or whatever.08:16
fullermdOr better, since you're going on with 2.0, name the branch RELENG_1 or something.  Then you can add branches off that for 1.0.x or 1.1.x releases or whatever's needful later.08:17
awilcoxHe said make a branch for releases and tag the release on it08:17
Peng_Well, it's your choice. Some projects just have one "stable" branch for the current release.08:17
awilcoxOkay so we'll have a RELENG_108:17
awilcoxOkay I am at /home/awos-bzr/repos/awos08:18
awilcoxI see: branches and trunk08:18
Peng_This isn't CVS, you don't need to keep caps lock on. :P08:18
awilcoxDo I just cp trunk branches/1.0 ?08:18
Peng_awilcox: "bzr branch".08:19
fullermdWell, you maybe _could_, but using 'branch' would be better.08:19
awilcoxOkay08:19
awilcoxbzr branch trunk branches/1.0 ?08:19
Peng_IMO it's silly to have a "branches" directory, but yes.08:21
fullermdUsers get offended when it's called "oldcrap" instead...08:21
awilcoxBranched 670 revision(s).08:21
awilcoxSo now if I commit to the branch, it will have revision 671, and that will be separate from trunk revision 671, yes?08:24
nok4hi, i don't want bzr to track the file mode/ownership changes, how to do that?08:25
Peng_nok4: All it tracks is the exec bit. If you want to turn that off, um, I don't know how.08:28
Peng_awilcox: Yes.08:28
awilcoxPeng_: Awesome08:28
Peng_awilcox: Y'know, you can merge the release branch back into the trunk when you make changes.08:28
awilcoxThank you Peng_ and fullermd for your help, it is much appreciated :D08:28
nok4Peng_: ok, i'll try chmod the x bit to work around it.08:30
Peng_Why does the index have to be probed to figure out that there's nothing to pull?08:46
thekornhi, I've got a question about using hooks: is it possible to ship a hook within a branch?08:54
thekornas far as I understand the docs a hook has to be in ~/.bazaar/plugins08:54
thekornbut I'm looking for hooks on a per branch basis, so something like mybranch/.bzr/hooks08:55
awilcoxPeng_: How do I merge?09:04
Peng_awilcox: "bzr merge". :D09:04
awilcoxahhhhhh09:04
Peng_awilcox: Read the tutorial and/or help.09:04
awilcoxThat should have been obvious.09:04
awilcox:P09:04
* igc dinner09:18
awilkinsthekorn: No, you can't ship hooks in a branch unless you do it and get the users to run a script that installs them09:52
awilkinsthekorn: Bazaar will by design not execute stuff in a branch automatically09:53
thekornawilkins, thanks, so it won't make sense to file a wishlist bug, as this is by design, correct?09:54
awilkinsthekorn: I'd agree with that09:54
fullermdWell, it's probably something we'd like more mechanism for.  So it's wishlist in a sense.10:00
thekornit would be nice to have especially for hooks which don't do any harm, like commit_message_template10:02
thekornbut I can live without it, just curious10:03
pygithekorn: plugin plugin plugin10:15
pygiyou can do almost anything in a plugin10:15
thekornpygi, I'm sure you can, so write me a plugin which allows me a define a commit_message_template hook within a branch ;)10:16
pygithekorn: w00t xD10:17
pygicatch me during UDS, we might talk if its really something you need10:17
pygi:p10:19
* pygi hides10:19
thekornpygi, ok, ok, I'm also not sure if I really need it10:20
pygiehhe10:22
* pygi shoots thekorn 10:22
fullermdWell, it's kinda a bootstrap problem   :p10:22
* thekorn switches over to git10:22
fullermdMaking a plugin that forces somebody not having the problem to get the plugin has minor temporal difficulties.10:23
fullermds/problem/plugin/10:23
thekornright, I think I will just trust the contributors to the project to always use the template for commit messages we agreed on10:25
fullermdYah.  Social problems are best handled with a cattle prod   8-}10:26
KinnisonForcing a given template for commit messages seems a bit draconian. Is this a commercial project?10:26
KinnisonBut as fullermd says, cattle prods are useful.10:27
thekornno, I just want them to mention all files they changed10:27
KinnisonIn what sense?10:28
Kinnisonthe change surely does that for you.10:28
pygithekorn: hm, whats the point if bzr can tell you that for every revision? o.O10:28
=== Kissaki^0ff is now known as Kissaki
thekornpygi, I'm using bzr in a computer sience class, and I want the student to actually describe their changes in detail10:35
pygithekorn: ah, ok!10:35
thekornso just for teaching purposes10:35
Peng_bzr-gtk has a nice commit UI, doesn't it?10:37
Peng_I mean, for messages. It's the source of all of those really detailed messages in bzr.dev, isn't it?10:37
* Spaz tickles everyone10:43
Spazoops10:43
Spazwrong channel10:43
pygiSpaz: :p10:44
pygilol10:44
Spazyou don't deserve my tickling! >:(10:44
Kamping_Kaiser:(10:45
fullermdI always thought commit messages were meant for long rambling harangues about the stupid things the client made you do...10:47
Odd_BlokeBonus points for using profanity as shorthand.10:52
awilkinsBonus points for writing comments that when they are acronymized are an insult to the irritating client that cause the patch to be necessary10:57
awilkinsthekorn: I remember a plugin that automatically listed all changed files in the comment10:59
awilkinsthekorn: Wouldn't be hard to write though10:59
awilkinsthekorn: I remember another one that parsed the --- below here --- part of the comment that was previously ignored and used it as a commit list, so you just deleted the line for any file you didn't want committed11:00
fullermdI remember _discussion_ about that, but I'm not sure it really exists.11:01
bialixhi, I have a question about English term. What is blueprint? Re: blueprints at launchpad11:09
fullermdbialix: A set of plans for construction, as of a building.11:10
bialixso it's more like "plan"?11:11
bialixor "planned features"?11:11
bialixor closer to "specification"?11:11
fullermdProbably specification.11:11
bialixI need this for Russian translations11:11
fullermd"plan" can get too general; a blueprint is a plan for building a specific thing, whereas "plan" could also mean broader what things you want to build.11:11
fullermdOnly downside to 'specification' is that it could also mean a description of something that already exists, rather than the desired specifications for something to be made.11:13
fullermdBut I don't know whether that ambiguity would translate.11:13
bialixso "blueprints @ launchpad" -- plans?11:13
fullermdYeah, that would probably get the right idea across.11:15
bialixok11:15
bialixfullermd: is it somehow related to "request"?11:18
=== Kissaki is now known as Kissaki^0ff
fullermdbialix: No, it's more like "We've decided to do this (or at least explore it in-depth), here is the behavior the feature should have and something of how we'll implement it at the code-level"11:21
fullermdSo it's more of an "implementation plan", maybe.11:22
bialixlooks good11:24
knielsenjam: I tried to use bzr merge-into to work-around bug 375898 as you suggested. Unfortunately using bzr merge-into gives the exact same problem :-(11:59
ubottuLaunchpad bug 375898 in bzr "bzr merge fails: bzr: ERROR: No final name for trans_id 'new-1'" [Undecided,New] https://launchpad.net/bugs/37589811:59
knielsenjam: I have added to the bug report the exact steps needed to get the error, including the merge-into step and subsequent patches applied.12:00
vilajam: ping13:28
Peng_jelmer: ping13:38
jelmerPeng_: yo13:38
Peng_Hi! :)13:39
Peng_Oops, wait, never mind.13:39
beunovila!13:40
vilabeuno !13:40
vila:)13:40
beunovila, I'm packing to go see you!13:40
Peng_jelmer: Sorry. I wanted to mention a little error in bzr-svn, but you've already changed that code and fixed it. Oops, should've checked before pinging you.13:40
vilabeuno: I'll do the same very soon (still a couple of hours hacking while re-checking my survival kit :)13:41
beunovila, you've got a slightly shorter flight than I do  ;)13:41
vilabeuno: we should sit and talk again about bzr-upload, I'll took notes last time but I need more details to refresh my memory :-/13:42
vilabeuno: sure thing, the flight is shorter, yet I have to leave in the morning anyway because there isn't a lot of flights from here...13:42
beunovila, would be great13:42
beunothere's that bzr sprint going on during UDS13:42
beunoI haven'g signed up because I'm over-booked elsewhere13:43
beunoBUT13:43
vilayes, hopefully we'll find time in the two coming weeks13:43
beunoI will run away occasionaly and hang out13:43
jelmerhmm, bb is down again :-(13:48
=== Kissaki^0ff is now known as Kissaki
=== vila is now known as vila-dentist
=== `6og is now known as Kamping_Kaiser
=== vila-dentist is now known as vila
vilajam: ping15:33
jamhi vila15:34
ph8Hi all, i'm new to bazaar - am actually just trying to checkout a copy of the planet project (planetplanet.org)'s repo - i've used bzr branch http://www.gnome.org/~jdub/bzr/planet/devel/trunk/ ./ - but it seems to be sticking, does it take time to examine files or is it likely my http proxy isn't setup correctly?16:04
=== Spaz is now known as spaz
_amanica_ph8, it can be your proxy settings17:00
KinnisonEww, planetplanet is using an old repo format17:06
Takwhat's the (practical) difference between default-rich-root and rich-root-pack?18:14
jamTak: in the future --default-rich-root may point to a different format18:17
jam--rich-root-pack will always stay the same18:17
jamas of *today* there is no difference18:18
Takhmm - which one would be a better default for a user who's unlikely to know the difference?18:20
jelmerTak: where are you using them for?18:50
TakI'm looking at adding "init" to md-bzr18:51
jelmerTak: --default-rich-root seems like the most sensible choice then18:51
Takok18:51
=== fta_ is now known as fta
las3rHello all.  I'm trying to get the bzr "svn" plugin working, but it tells me "No module named filters".  I'm using bzr-svn-0.5.4 w/ bzr 1.13.1 on Fedora.22:31
las3rAnd indeed, there is no module bzrlib.filters.22:32
jelmerlas3r: for bzr 1.13.1 you need bzr-svn 0.5.3 (see the bzr-svn wiki page)22:33
las3rAh, I see.  I was looking in the "requirements" section, which wasn't so specific.  Now I understand what the release list was telling me.  Thanks!22:33
las3rOkay.  Now I'm trying to import from an https:// subversion repository using a self-signed cert.  Curl is complaining ("Peer certificate cannot be authenticated...").  Is there any easy to way to pass the equivalent of 'curl --insecure' to this process?22:37
jelmerlas3r: try prefixing the URL with svn+22:47
las3rjelmer: Huh, seems to work, but with the deprecated warning.  I had just finished patching the code to use connection.setopt(pycurl.SSL_VERIFYPEER,0) and then ran into authentication issues...22:48
las3rjelmer: Other than using svn+https, is there any way to provide a username/password to svn-import?22:48
las3rWell, I'll try the mailing list, because it's failed on the next step...23:08
ronnyjelmer: are you doing weird stuff to dulwich? every time try i pull the bzr branch, it tells me "branches diverged"23:19
davidstraussI'm getting "bzr: ERROR: Server sent an unexpected error: ('error', "ImportError instance has no attribute 'message'")" when checking out from my server over ssh23:31
Odd_Blokedavidstrauss: Can you try running a similar operation locally on the server?23:34
jelmerronny: Shouldn't have happened recently, I mean, not in the last two weeks23:34
davidstraussOdd_Bloke: Works23:34
davidstrauss(from local access)23:35
Odd_Blokedavidstrauss: Hmm, that's the only debugging idea I had.23:36
Odd_BlokeI know nothing about the smart server. :p23:36
Odd_Blokedavidstrauss: Is this bzr+ssh or sftp?23:37
ronnyjelmer: im not very regular updating, since busy with other stuff23:37
davidstraussOdd_Bloke: bzr+ssh23:37
Odd_Blokedavidstrauss: Try sftp.23:37
davidstraussworks fine over sftp23:37
=== Kissaki is now known as Kissaki^0ff
jelmerronny: So, the rebasing should've stopped by now23:39
jelmerronny: it was mainly caused by the fact that dulwich is maintained in both bzr and git23:39
Odd_Blokedavidstrauss: Right, now I'm out of ideas. :p23:39
jelmerronny: and should stop now23:40

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