[00:05] <mwhudson> i don't think loggerhead strictly depends on setuptools but it does use pkg_resources
[00:06] <mwhudson> (which most people have because setuptools itself depends on pkg_resources aiui)
[00:06] <Noldorin> mwhudson: yeah, i'm sure you're right. i included the whole thing just to be safe thuogh
[00:06] <Noldorin> right, now a non dependency-related issue
[00:07] <Noldorin> The specified CGI application misbehaved by not returning a complete set of HTTP headers. The headers it did return are "".
[00:07] <mwhudson> Noldorin: it's not a cgi app
[00:07] <Peng> Yeah, Loggerhead doesn't need setuptools. To run, anyway. I dunno about installing it.
[00:07] <Noldorin> mwhudson: hmm. yeah, it's wsgi. and the server apparently supports that via isapi-wsgi
[00:08] <Noldorin> but i'm not sure how to tell it to use wsgi
[00:08] <Peng> The usual practice is to use HTTP and have the server proxy to it.
[00:10] <mwhudson> Peng: btw, if you want to steal stuff from http://bazaar.launchpad.net/~launchpad-pqm/launchpad-loggerhead/devel/annotate/head%3A/start-loggerhead.py and incorporate it into loggerhead proper that would be great
[00:11] <lifeless> Peng: windows VPS server, nothing usual about it :)
[00:11] <Noldorin> Peng: yeah, so i've been told. but i'm on a shared server
[00:11] <mwhudson> (i'm thinking of the improved logging and the timeout stuff)
[00:11] <Noldorin> lifeless: exactly :S
[00:13] <Peng> lifeless: Ah.
[00:13] <Peng> mwhudson: Hm.
[00:14] <Peng> ...Great, now I'm speaking in two-letter grunts.
[00:14] <mwhudson> Peng: well i guess what i really mean is "have you seen this stuff" ?
[00:14] <Peng> Perhaps it would be easier to get FastCGI working?
[00:14] <Peng> mwhudson: I knew of launchpad-loggerhead, but hadn't looked at it much.
[00:15] <Noldorin> Peng: i'd need a fastcgi -> wsgi interface though
[00:16] <Peng> Noldorin: What? You mean like flup?
[00:16] <lifeless> isapi-wsgi may be that interface already
[00:16] <lifeless> inside the server
[00:16] <Noldorin> Peng: not sure, honestly
[00:16] <Peng> Eh. I'm not going to pretend to know anything about Windows servers.
[00:16] <Noldorin> this is all fairly new to me
[00:16] <Peng> Or that I know what this conversation is about. :P
[00:16] <Noldorin> lifeless: yeah quite possibly
[00:17] <Noldorin> i'm just not sure how to use it :S
[00:17] <Peng> mwhudson: lp-lh is under a different license...
[00:17] <Peng> I think.
[00:17] <Peng> Yeah, Loggerhead is
[00:17] <mwhudson> Peng: oh god it probably is agplv3 isn't it
[00:17] <Peng> GPLv2.
[00:17] <Peng> mwhudson: :D
[00:17] <mwhudson> and loggerhead is gpl
[00:18] <mwhudson> yay
[00:18] <mwhudson> we can probably get this fixes
[00:18] <mwhudson> *fixed
[00:18] <dOxxx> I'm running selftest on windows and I'm getting a lot of these: http://pastebin.com/m3a6c05f0
[00:20] <dOxxx> It's making it pretty hard to see the "real" errors in tests :P
[00:21] <Peng> (Thank goodness for licenses. Otherwise I would've had to come up with an excuse not to work on it. ;-)
[00:23] <Noldorin> lifeless, mwhudson: are you stumped then?
[00:23] <Noldorin> i certainly am now...
[00:23] <Peng> I'm more or less happy with my logging setup, so working on that makes me want to bang my head on the wall without any benefit. :P
[00:23] <Peng> I haven't run into the other things it works around, but they look interesting. 'specially the threadpool_options.
[00:23] <lifeless> Noldorin: I have no idea, haven't had at any point :)
[00:24] <lifeless> Noldorin: I'd guess that you don't have a proper wsgi app setup, *or* that isapi-wsgi doesn't support what we do, or something else.
[00:24] <Noldorin> heh. it's really going to be much less pain using webbr i think
[00:24] <Noldorin> yeah, it's just too tricky on a windows servers, let alone a shared windows service i think
[00:25] <lifeless> I'm sure its doable, but other than saying 'good luck' and 'ask mwhudson if you need wsgi pointers'... I can't do much for you.
[00:25] <mwhudson> loggerhead is set up bit oddly compared to many wsgi things
[00:26] <mwhudson> it's more like a "http server that uses wsgi" rather than "a wsgi app" i think
[00:26] <mwhudson> (this would probably be good and not that hard to fix if you knew how)
[00:26] <Noldorin> lifeless: if i had the time and effort to spare, perhaps
[00:26] <Noldorin> but alas i do not
[00:35] <igc> morning
[00:40] <Noldorin> alright
[00:40] <Noldorin> too late to be hacking now, methinks
[00:40] <Noldorin> thanks for your help, all
[00:40] <Noldorin> 'night
[00:41] <maxb> Does any documentation for bzr-git exist?
[00:42] <lamalex> Hi, i keep getting this bzr: ERROR: [Errno 13] Permission denied when i do bzr update
[00:52] <lifeless> lamalex: check your .bzr.log
[00:54] <jelmer> maxb: the normal bzr documentation should apply
[00:55] <maxb> jelmer: I'm starting to think that despite "bzr git-import" existing, there's no way to pull further revisions from a non-HEAD git branch?
[00:56] <jelmer> maxb: git-import can be run repeatedly
[00:56] <jelmer> but there's no other way to pull from a non-HEAD git branch
[00:56] <jelmer> (since bzr doesn't have a way of addressing colocated branches)
[00:57] <maxb> It would be nice to invent such a thing, even if it only was relevant to bzr-git and bzr-hg
[00:59] <lifeless> it was agreed on thelist ~ a year back
[00:59] <lifeless> I don't recall what the agreed way /way/
[00:59] <lifeless> s/way/was/
[01:00] <jelmer> lifeless: I don't think there was an agreed way
[01:00] <maxb> Was the idea of doing something bzr-git specific shot down already?
[01:00] <jelmer> lifeless: at least I don't recall seeing anything
[01:01] <jelmer> maxb: Yes, that's a bad idea - it means we'll have migration issues later and break existing URLs
[01:05] <dOxxx> I'm getting a lot of TypeErrors about Tuple when running selftest on Windows, any idea what's going? http://pastebin.com/m3a6c05f0
[01:06] <lamalex> lifeless: i dont see on..
[01:06] <lamalex> s/on/one
[01:07] <jelmer> lamalex: should be in your home dir
[01:09] <maxb> jelmer: Well.... it's sad that bzr-git remains crippled for the want of a consensus
[01:10] <maxb>  remote.py |    7 ++++++-
[01:10] <maxb>  1 file changed, 6 insertions(+), 1 deletion(-)
[01:10] <maxb> ^ and I've made it work more or less with merely that
[01:12] <spiv> lamalex: 'bzr --version' should tell you where it is
[01:12] <lamalex> jelmer, lifeless: yah, found it. http://paste2.org/p/541334
[01:12] <spiv> dOxxx: interesting, is this with current lp:bzr ?
[01:12] <dOxxx> spiv: yes
[01:13] <dOxxx> I was running --parallel=subprocess on win32
[01:13] <spiv> dOxxx: are all the C modules up to date?
[01:13] <dOxxx> spiv: this was from source without C modules compiled
[01:13] <spiv> (i.e. have you run whatever the windows equivalent of "make build"?)
[01:13] <spiv> Ah, so no C modules?  Hmm.
[01:13] <dOxxx> spiv: I'm not even sure how to do that on Windows :)
[01:14] <spiv> dOxxx: python setup.py build_ext -i, IIRC, but there may be tricky dependencies or something more to it...
[01:14] <poolie> hello spiv, doxxx, jelmer
[01:15] <dOxxx> spiv: hmmm
[01:15] <dOxxx> hey poolie
[01:15] <spiv> dOxxx: I can reproduce it here
[01:15] <dOxxx> spiv: that's good, I guess :)
[01:15] <spiv> dOxxx: probably a bug in the pure python fallback code for John's StaticTuple work
[01:16] <jelmer> hi poolie
[01:16] <dOxxx> spiv: should I file a bug?
[01:16] <jelmer> maxb: agreed
[01:17] <lamalex> anyone know about that error? i dont understand the log output
[01:17] <spiv> dOxxx: yeah
[01:22] <mkanat> Wow, if you run "nick" on a loomified branch, things explode.
[01:23] <fullermd> If you nick us, do we not bleed?
[01:23] <mkanat> lol
[01:24]  * mkanat just realized his bzr is somewhat old, though.
[01:24] <mkanat> 1.16.
[01:24] <dOxxx> spiv: bug 490600
[01:24] <mkanat> Well, 1.16.1.
[01:26] <spiv> mkanat: yeah, I think that may be fixed with current bzr + current bzr-loom
[01:26]  * mkanat nods.
[01:26] <mkanat> Okay.
[01:27] <spiv> dOxxx_: thanks!
[01:40] <shakaran1> Hi, I can have severals parents locations for a branch? I explain that. I have a ppa with bazaar but I need migrate all code to a SVN. If I do a pusth with the url of a SVN, the bazaar branch complains that already exists a parent location and the show a diverged branch
[01:44] <jelmer> shakaran1, hi
[01:46] <jelmer> shakaran1: Does the location you're trying to push to already exist perhaps?
[01:46] <shakaran1> yep, the bazar location exits, and the svn too
[01:47] <shakaran1> I do: bzr push svn+https://forja.rediris.es/svn/cusl4-tivion
[01:47] <shakaran1> And I get: bzr: ERROR: These branches have diverged.  See "bzr help diverged-branches" for more information.
[01:48] <shakaran1> the bzr missing show me: Using saved parent location: bzr+ssh://bazaar.launchpad.net/~shakaran/tivion/tivion/
[01:48] <jelmer> that's a different location
[01:48] <jelmer> than what you're pushing to
[01:49] <jelmer> shakaran1, ^
[01:51] <shakaran1> yeap, Can I do?
[01:51] <shakaran1> I dont know how to do it
[01:51] <shakaran1> I want push all my repo on the SVN
[01:53] <shakaran1> how to do it then?
[01:54] <jelmer> shakaran1: did you read the information in "bzr help diverged-branches" ?
[01:56] <shakaran1>  bzr merge
[01:56] <shakaran1> Merging from remembered submit location bzr+ssh://bazaar.launchpad.net/~shakaran/tivion/tivion/
[01:56] <shakaran1> Nothing to do.
[01:56] <jelmer> shakaran1: What does "bzr missing" on the URL you're trying to push to say?
[01:57] <shakaran1> In this url, I have a SVN repo,  https://forja.rediris.es/svn/cusl4-tivion how to do bzr missing?
[02:01] <shakaran1> I have to do something with this
[02:03] <jelmer> "bzr missing https://forja.rediris.es/svn/cusl4-tivion"
[02:04] <shakaran1> You have 55 extra revision(s):
[02:04] <shakaran1> I need to do a commit?
[02:05] <jelmer> shakaran1: You probably want to push to https://forja.rediris.es/svn/cusl4-tivion/trunk
[02:06] <shakaran1> jelmer: oh perfect! jelmer you save my day!
[02:11]  * igc lunch + est
[02:11] <igc> rest
[02:20] <poolie> spiv, hi - tell me things? :)
[02:21] <spiv_> Well, I just discovered duelling pppd processes, which probably explains why I dropped off adsl a moment ago :)
[02:22] <spiv_> I think I found the cause of https://bugs.edge.launchpad.net/bzr/+bug/490600, and just pushed a simple branch for it.
[02:24] <spiv_> I'm a bit concerned about https://bugs.edge.launchpad.net/bzr/+bug/489211, but it's going to be hard to track down without a way to reproduce it locally.  Coincidentally it's possibly the same area of code as 490600, but that's just coincidence :)
[02:25] <spiv_> I've spent some time staring at the relevant code hoping to spot the bug by inspection, but I think I'll have to make a patch to add more logging for the reporter to try.
[02:35] <poolie> mm i thought that was a dupe
[02:35] <poolie> but apparently not?
[02:35] <poolie> well, they're seeing it in a release that's supposed to have the other bug fixed
[02:35] <poolie> as you say
[02:40] <spiv> poolie: in a release that fixed the other, with a format (2a) that didn't experience the other.
[02:40] <poolie> spiv, i was wondering a bit this morning if there were UDD things more important than these bugs
[02:40] <poolie> but i think these bugs are genuinely high
[02:40] <poolie> so
[02:40] <poolie> it's not a bad use of time to work on them
[02:42] <LaserJock> I'm getting bzr: ERROR: KnitPackRepository is not compatible with RemoteRepository different rich-root support when trying to pull from a branch on Launchpad, any thoughts?
[02:44] <lifeless> the source has upgraded, probably to 2a.
[02:45] <spiv> LaserJock: you have mismatched formats, upgrade one (or both) sides to 2a.
[02:45] <LaserJock> spiv: I think I tried that
[02:45] <spiv> Your local side isn't 2a, judging from that error.
[02:46] <LaserJock> bzr info gives an "unnamed" branch format on both sides
[02:46] <spiv> Try info -v
[02:46] <LaserJock> so yeah, that doesn't sound right
[02:46] <spiv> What's the URL of the remote branch?
[02:46] <LaserJock> format 7 locally
[02:47] <LaserJock> bzr+ssh://bazaar.launchpad.net/~cjwatson/ubuntu-cdimage/mainline/
[02:48] <spiv> LaserJock: "format 7" is the format of the branch, for this it's the repository format that matters
[02:48] <spiv> Right, that remote branch is 2a
[02:50] <spiv> Whereas your local one is probably something like "Packs 6 ..." rather than "Repository format 2a ..." (to use the descriptions info -v will print)
[02:52] <LaserJock> spiv: "Packs containing knits without subtree support"
[02:54] <LaserJock> so do I blow away my repo and start over (not a big deal)?
[02:54] <lifeless> LaserJock: you run 'bzr upgrade'
[02:54] <LaserJock> I did
[02:54] <LaserJock> oh, wait
[02:54] <LaserJock> I ran bzr upgrade on the branch
[02:55] <LaserJock> gah, how confusing
[02:55] <spiv> bzr upgrade, if you have bzr 2.0.0 or newer, will upgrade you to 2a.
[02:55] <LaserJock> ok, running bzr upgrade on the repo dir
[02:55] <spiv> It should upgrade both the branch and the repo, IIRC, that's a bit odd.
[02:56] <lifeless> spiv: it upgrades a single bzrdir last I heard
[02:59] <spiv> Ah.
[02:59] <spiv> That's a pity.
[03:27] <LaserJock> is there any cool ways to make a bzr pull from a repo do pulls from all the branches in the repo?
[03:27] <LaserJock> I'm searching for better alternatives to bzr multi-pull
[03:30] <bob2> repo-push ;p
[03:33] <Peng> spiv: It's unintentional that lp:~spiv/bzr/static-tuple-pure-python-bug-490600 is linked to bug #408193, right?
[03:34] <spiv> Peng: d'oh, yes
[03:34] <spiv> Peng: it's because I copy & pasted the wrong bug number.
[03:35]  * spiv fixes
[03:35] <Peng> :)
[03:55] <zombor> hello, im trying to run svn2bzr on os x, and i get a "ImportError: No module named bzrlib" error, any idea how to fix it? my bzr install works perfectly fine
[04:01] <Peng> zombor: svn2bzr? Isn't that...really old? Why not bzr-svn?
[04:01] <zombor> does that export svn repos into a bzr repo?
[04:02] <Peng> zombor: Yes.
[04:02] <zombor> i thought that was for commiting to svn with bzr
[04:02] <Peng> zombor: It's for bidirectional transfers between bzr and svn.
[04:02] <zombor> yeah, i don't want to do that
[04:02] <Peng> zombor: You don't have to push back to svn if you don't want to.
[04:02] <zombor> i want to take my svn repo, and convert it into a bzr repo
[04:02] <Peng> zombor: So use bzr-svn.
[04:03] <Peng> Did I link to http://bazaar-vcs.org/BzrForeignBranches/Subversion? I meant to link to it.
[04:03] <zombor> im looking at that right now
[04:06] <zombor> hrm, so how do i use this thing...
[04:09] <spiv> zombor: bzr-svn?  If you just want to convert one branch "bzr branch svn://.../the_branch the_converted_branch"
[04:09] <zombor> ah ok, it just uses normal svn syntax
[04:10] <spiv> Right.  http: and svn+ssh etc should all Just Work.
[04:10] <zombor> hrm, but i just upgraded to bzr 2.0.2 and it broke my bzr command =(
[04:10] <spiv> Oh?
[04:11] <zombor> ah, nevermind, i needed a shell refresh -_-
[04:11] <zombor> got the svn branching, thanks a lot guys
[04:51] <poolie> getting back into pilotage
[04:54] <poolie> review queue jumpers welcome
[05:03] <_Andrew> Hi guys
[05:03] <_Andrew> Is there a way to export any modified or added files between two different revisions ?
[05:03] <poolie> you want a whole copy of those files?
[05:04] <poolie> or just the diff?
[05:04] <_Andrew> whole copy
[05:04] <poolie> hm
[05:04] <poolie> not built in
[05:04] <_Andrew> I know you can bzr export the whole project
[05:04] <poolie> you could probably script it by reading the output of 'log -v' or 'status -r'
[05:07] <spiv> or maybe do a "bzr diff -rX..Y | lsdiff" to get the list of changed/added files?
[05:08] <spiv> Although I suppose status -r does that too.
[05:09] <spiv> The bzr-upload plugin must calculate something like this too.
[05:09] <JonCruz> so... any recommended way to find out what has changed on the server before syncing it on down?
[05:11] <lifeless> there was a bug report about export I think
[05:11] <lifeless> do do this
[05:18] <spiv> JonCruz: perhaps rsync or the bzr-upload plugin would work for you?
[05:20] <JonCruz> with subversion, "svn status -u" would tell me. I wounder if "-v" on "missing" might help
[05:23] <_Andrew> spiv, thanks i'll check it out
[05:24] <_Andrew> I'm looking at the manual online but I can't find the command just to get the latest revision with bzr log?
[05:25] <spiv> -r -1
[05:25] <_Andrew> ah!
[05:25] <_Andrew> thaks
[05:37] <dOxxx> or -l1
[07:06] <vila>  hi all
[07:06] <lifeless> hiya
[07:09] <vila> revno 4842 broke no-extensions-pure-python support, babune is so confused about that it turned all red
[07:12] <lifeless> vila: https://code.launchpad.net/~spiv/bzr/static-tuple-pure-python-bug-490600/+merge/15470 ?
[07:12] <vila> wow
[07:13] <_Andrew> After I start using a plugin and want to install another do I still need the __init__.py file?
[07:14] <_Andrew> or do I need to merge the two files ?
[07:14] <vila> lifeless: sounds very likely
[07:23] <poolie> hi there vila
[07:23] <poolie> _Andrew: ?
[07:23] <poolie> you should have ~/.bazaar/plugins/one/__init__.py
[07:24] <poolie> and then .../two/...
[07:24] <poolie> etc
[07:25] <_Andrew> oh really
[07:26] <_Andrew> woops
[07:26] <_Andrew> So actually I should put the plugin in it's own folder
[07:38] <Peng> Or call them ~/.bazaar/plugins/one.py and two.py.
[07:42] <_Andrew> but what if two plugins both have __init__.py ?
[07:42] <_Andrew> Surely you can't just rename them?
[07:43] <Peng> _Andrew: So, what poolie said.
[07:43] <_Andrew> Thanks
[07:43] <poolie> _Andrew: they should have folders named eg 'gtk', 'pqm', etc
[07:43] <_Andrew> It's working now. :)
[07:43] <vila> _Andrew: then you'll have one/__init__.py instead of one.py and two/__init__.py instead of two.py
[07:43] <poolie> containing all their files
[07:43] <poolie> yay
[08:05] <bialix> hi
[08:05] <_Andrew> Branch format 6 is old right?
[08:06] <bialix> _Andrew: not really
[08:06] <bialix> it's pack-0.92
[08:07] <_Andrew> Would it be worth upgrading because many people in our office keep complaining it's slow.
[08:07] <Peng> _Andrew: The only thing branch format 7 gets you is stacking.
[08:07] <bialix> Peng is right
[08:07] <_Andrew> Ah, I read that. I don't need that.
[08:07] <Peng> _Andrew: This is about performance? Then the important part is the _repository_ format. Well, unless you want to use stacking too.
[08:12] <_Andrew> I think that if it's not going to make any difference upgrading then I'm not going to touch it
[08:13] <_Andrew> thanks for the help guys
[08:13] <Peng> _Andrew: Upgrading your repository to 2a would probably make a big difference.
[08:13] <_Andrew> oh really?
[08:13] <Peng> _Andrew: Yes. They were able to squeeze much more awesome into it than the previous formats.
[08:14] <Peng> _Andrew: What's your current repository format?
[08:14] <_Andrew> format 7 ?
[08:15] <_Andrew> er
[08:15] <_Andrew> format 6
[08:16] <Peng> Eh.
[08:17] <Peng> _Andrew: What's the first line of "bzr info -v", and what's the "repository:" line?
[08:17] <_Andrew>  pack-0.92
[08:17] <_Andrew> knits without subtree support
[08:18] <_Andrew> We're using bzr for web development
[08:18] <Peng> _Andrew: Okay. So you should see a big performance (and disk space) improvement if you upgrade to 2a. It requires a very recent version of bzr, and upgrading to a rich-root format like 2a is one-way.
[08:19] <Peng> I should rephrase that. You'd have to make the rich-root transition, which is one-way.
[08:20] <_Andrew> We have bzr 2.0.2 installed
[08:20] <Peng> Perfect.
[08:20] <Peng> All of you? There's not that one weirdo who's still using 1.5?
[08:20] <_Andrew> Right now we're all working via ssh off the server
[08:21] <Peng> Does the server have 2.0.2?
[08:21] <_Andrew> yup
[08:21] <_Andrew> How we're setup is that 5 people have all checked out a copy of trunk which they commit and update to
[08:22] <_Andrew> and we have a branch named "live" for stuff that is ready to be uploaded
[08:22] <_Andrew> So what I'm wondering is how would I upgrade this so that it doesn't screw everyone up
[08:23] <Peng> Is bzr really that slow? Is the repo that large?
[08:23] <_Andrew> We have about 600 meg of content I think
[08:23] <_Andrew> Most of the time it's ok but some people you can see that it take a good 5+ seconds
[08:24] <_Andrew> plus my boss keeps using this as a reason to switch to hg
[08:24] <Peng> What takes 5 seconds? Bazaar isn't always a speed demon; 2a might not help all that much.
[08:24] <Peng> _Andrew: Anyway, http://doc.bazaar-vcs.org/latest/en/upgrade-guide/index.html should help.
[08:24] <_Andrew> I can't remember
[08:25] <spiv> _Andrew: definitely upgrade to 2a, it may help a little, or it might help a lot.
[08:25] <bialix> run bzr with cold cache and you get more than 5 seconds only to start
[08:25] <spiv> bialix: yeah
[08:26] <bialix> Peng: "werdo" :-P
[08:26] <spiv> bialix: run *python* with cold cache and you can get that... :/
[08:26] <bialix> spiv: right now I see PyQt starting is very slow
[08:26]  * spiv -> food
[08:26] <_Andrew> So if I upgrade trunk to 2a what happens when others with checked out copies of pack-0.92 try to commit ?
[08:27] <bialix> _Andrew: this require local upgrade
[08:27] <_Andrew> or if I pull changes into my live branch
[08:27] <bialix> or they will get error
[08:27] <_Andrew> So I need to upgrade trunk then make everyone checkout a new copy
[08:27] <bialix> that helps
[08:28] <Peng> _Andrew: The link explains at least some of it.
[08:28] <_Andrew> oh sorry I didn't see that link
[08:28] <bialix> _Andrew: are you using light checkouts or heavy one
[08:28] <_Andrew> looking at it now
[08:28] <bialix> ?
[08:28] <_Andrew> Not sure of the difference
[08:29] <bialix> bzr info tells
[08:29] <bialix> you need heavy co to get the power of local repo to speedup local operations
[08:31] <_Andrew> http://pastebin.com/m3a0e491a
[08:31] <_Andrew> That's the output of bzr info -v
[08:31] <_Andrew> but I don't see anything about heavy or light
[08:31] <bialix> just checkout is heavy
[08:32] <bialix> _Andrew: does you have many binary files?
[08:32] <_Andrew> i'm not sure of the ratio but probably. Things like videos, images, etc
[08:33] <bialix> hg is not much faster with binary files as I know
[08:35] <_Andrew> Yeah, I'm not too sure either but i'd prefer to keep using bzr
[08:37] <bialix> you can upgrade to 1.9 format, it has non-rich-root variant, but 2a is recommended default today, so...
[08:38] <_Andrew> I think it won't be too much trouble
[08:38] <bialix> you have pack-0.92 which is old, 1.9 is better
[08:38] <_Andrew> Just make sure everyone commits beforehand
[08:39] <bialix> will be nice to know how much 2a will be faster for you
[08:41] <Peng> 1.9 isn't all *that* much better, though.
[08:41] <Peng> Well, it mostly is. Just not compared to 2a.
[08:41] <_Andrew> Well, I asked the developer who was having the problem and he said every command was slow. I'll let you know when we do the update
[08:44] <_Andrew> I guess he means bzr up, st, ci etc
[08:49] <Peng> _Andrew: If he's using a lightweight checkout, using a heavyweight checkout or regular branch would probably be faster, since bzr wouldn't have to contact the network so much.
[08:50] <lifeless> vila: I think you should land spivs fix, its nearly 8 here
[08:51] <vila> lifeless: yup, I was just checking pqm
[08:56] <bialix> Peng: I disagree with you, but have no time to argue
[09:15] <sjamaan> Hi
[09:15] <sjamaan> I'm trying to push to an FTP server using "bzr branch . ftp://some-location"
[09:16] <sjamaan> It seems to hang on "Get stream" for a long time
[09:16] <sjamaan> Now that disappeared and doesn't seem to do anything
[09:16] <lifeless> sjamaan: yes, ftp uses append, there is a bug open about switching to PUT, for most of teh data copying.
[09:16] <lifeless> just be patient.
[09:17] <lifeless> also, you would normally use 'bzr push ftp://some-location', not 'bzr branch . URL' to do a push.
[09:17] <sjamaan> lifeless: I'm making an initial branch
[09:17] <sjamaan> Not pushing to an existing branch
[09:17] <Peng> sjamaan: You'd still use push for that.
[09:17] <sjamaan> ic
[09:17] <sjamaan> Why is that?
[09:17] <Peng> Why not?
[09:17] <sjamaan> Because it's a new branch
[09:18] <sjamaan> Push is for updating
[09:18] <lifeless> sjamaan: push will make a branch if one doesn't exist.
[09:18] <Peng> ISTM the bzr way is easier.
[09:19] <sjamaan> hm, it finished but didn't make a working copy
[09:19] <Peng> sjamaan: That's correct.
[09:19] <sjamaan> I only see an option for --no-tree, but not the inverse
[09:20] <Peng> sjamaan: Why do you want one?
[09:20] <lifeless> sjamaan: we dont make working copies over non-local-filesystems
[09:20] <sjamaan> Peng: I want to make a deployment of a website on a shared hosting provider
[09:21] <sjamaan> I'd like to be able to push updates easily and thought bzr would be ideal for that
[09:21] <sjamaan> I only have ftp access, no ssh
[09:21] <Peng> sjamaan: You should SSH to the server and run "bzr checkout" and maybe install the bzr-push-and-update plugin -- ah.
[09:21] <Peng> sjamaan: Use the bzr-upload plugin.
[09:21] <sjamaan> heh :)
[09:21] <sjamaan> thanks for the pointer
[09:22] <Peng> sjamaan: It doesn't create working trees on remote FSes because it would be tricky to do things like resolve conflicts.
[09:22] <sjamaan> Ah, of course
[09:22] <sjamaan> In my case that's not relevant, but I understand the general case is hard to get right
[09:26] <Rune> which format does bzr upgrade to when I do not specify a format? bzr 2.0.0
[09:27] <vila> Rune: 2a starting with bzr-2.0
[09:29] <vila> lifeless: thanks for adding me to c-a-c-o team, but I just wonder if I indeed apply them as I wanted to or if you did that on your own ?
[09:29] <vila> s/them/there/
[09:30] <lifeless> vila: see my mail to the list (not sent yet :)) - and the internal wiki page on contributor agreements
[09:30] <vila> lifeless: come to think of it, I think I didn't find a way to apply, so certainly you did that on your own :)
[09:30] <jml> how do I get the bound location of a branch?
[09:30] <lifeless> branch.get_bound_branch(), I think.
[09:30] <jml> thanks
[09:30] <vila> lifeless: lol synchronicity/telepathy on so on ;)
[09:32] <sjamaan> Peng: The upload plugin works fine. Thanks again for the tip!
[09:35] <Peng> sjamaan: Great. :)
[11:40] <vila> spiv: I've landed lp:~spiv/bzr/static-tuple-pure-python-bug-490600, it helps reduce the ~4000 errors, but still 2000 are left :-/
[13:19] <Morbus> g'day. i'm a newb. how does one "set a configuration option"? http://doc.bazaar-vcs.org/plugins/en/email-plugin.html
[13:19] <beuno>  Morbus it's usually bu adding it to ~/.bazaar/bazaar.conf
[13:20] <beuno> or if it's for a specific branch, in ~/.bazaar/locations.conf
[13:20] <beuno> under the branch's entry
[13:20] <Morbus> beuno: doesn't that make it compulsory on the user though? i want anyone who uses the master to have the commits mailed out, regardless.
[13:20] <Morbus> in a svn-based environment, this would be editing the post-commit on the master repo itself.
[13:21] <beuno> Morbus, it does. I know there's a server-side alternative, but I don't know about it
[13:21] <Morbus> hrm. alright.
[13:21] <Morbus> i'll poke around some more.
[13:21] <RenatoSilva> Morbus: .bzr/branch/branch.conf. Using bzr-email?
[13:22] <Morbus> RenatoSilva: yeah.
[13:22] <Morbus> so stick the post_commit_to config option in there?
[13:23] <RenatoSilva> Morbus: I wrote a patch to allow customization of subject and body: https://code.launchpad.net/~renatosilva/bzr-email/mail-customization
[13:23] <Morbus> i just want to get it to work first, then i'll worry about beautifying it ;)
[13:23] <RenatoSilva> Morbus: yes, set the options inside the branch, and anyone using it will be affected
[13:24] <Morbus> so, where does the plugin get installed?
[13:24] <Morbus> .bzr/branch/plugins? ;)
[13:24] <Morbus> or do i just install via the setup.py on the master box?
[13:25] <Morbus> looks like it.
[13:25] <RenatoSilva> Can we install plugins inside branches? O.o I know of ~/.bazaar/plugins, and <bzr_home>/plugins
[13:26] <Morbus> yeah, i just checked out, and python setup.py install.
[13:26] <Morbus> stuck it in /usr/lib/python2.4/site-packages/bzrlib/plugins/email/tests/ for this box.
[13:26] <Morbus> so, now just to add the config and hopefully i'm dandy.
[13:29] <RenatoSilva> email/tests? it should not be in a subdir, it should be in email
[13:29] <Morbus> yeah, it is. i just randomly pasted one of the installation lines.
[13:30] <Morbus> is there anyway to debug the sending? i just ptu the config in, made a commit from an external checkout, and no mail. whee.
[13:31] <Morbus> sending mail manually using /usr/bin/mail does work...
[13:36] <RenatoSilva> how long you've been waiting?
[13:36] <Morbus> hrm. nope. not working.
[13:36] <Morbus> i've been checking the mail log after each send.
[13:36] <Morbus> i've tried /usr/bin/mail, /bin/mail, and smtplib.
[13:36] <RenatoSilva> smtp server log? see bzr.log
[13:36] <Morbus> for the value of post_commit_mailer
[13:37] <Morbus> what should i be looking for in there?
[13:38] <RenatoSilva> clean bzr.log and do a commit, then check what's outputted there
[13:38] <Morbus> there are no errors jumping out at me.
[13:38] <Morbus> nor are there any indications that email was attempted.
[13:38] <Morbus> (though a bzr help email does display the help text, suggesting that the plugin was installed)
[13:38] <Morbus> i'm the only one using the repo at the moment, so there's no confusion about what messages relate to what (re, clean bzr log)
[13:39] <Morbus> should i be seeing activation of the email plugin in there?
[13:39] <RenatoSilva> Morbus: $bzr plugins, email should be listed
[13:39] <Morbus> it is.
[13:40] <Morbus> post_commit_to = devteam@example.com
[13:40] <Morbus> post_commit_sender = noreply@intra.example.com
[13:40] <Morbus> post_commit_mailer = smtplib
[13:40] <Morbus> is in my /path/to/master/branch/.bzr/branch/branch.conf
[13:41] <Morbus> i've also tried /bin/mail for the mailer, which works when I run "/bin/mail -s whee morbus@disobey.com" etc.
[13:44] <RenatoSilva> .bzr/branch/branch.conf? where's smtp_server?
[13:44] <Morbus> just added that to localhost, did a commit, checked mail and bzr.log and no change.
[13:46] <RenatoSilva> try localhost:25
[13:47] <Morbus> reading through https://bugs.launchpad.net/bzr-email/+bug/307988 at the moment
[13:48] <RenatoSilva> $bzr version?
[13:48] <Morbus> 2.0.2
[13:48] <RenatoSilva> os?
[13:48] <Morbus> centos, also mentioned in the above bug
[13:49] <Morbus> i am checking out from an external client - a different machine entirely. one of the reports here suggests that it'll work from a checkout on the server itself - i'll test that out shortly.
[13:52] <RenatoSilva> are you using bzr commit to test it? push/pull notification is disabled by default
[13:52] <Morbus> yes, using bzr commit, and just added push/pull too.
[13:52] <Morbus> still nothing. but i haven't tested a server checkout yet.
[13:52] <RenatoSilva> tried localhost:25?
[13:52] <Morbus> yes.
[13:52] <Morbus> and i'm going over bzr+ssh
[13:53] <RenatoSilva> man, it's like email plugin is not loaded
[13:53] <RenatoSilva> crazzy
[13:53] <Morbus> should i be seeing it loaded in the bzr log?
[13:53] <Morbus> i see:
[13:53] <Morbus> 0.023  looking for plugins in /home/bzr/.bazaar/plugins
[13:53] <Morbus> 0.023  looking for plugins in /usr/lib64/python2.4/site-packages/bzrlib/plugins
[13:53] <Morbus> 0.084  looking for plugins in /usr/lib/python2.4/site-packages/bzrlib/plugins
[13:53] <Morbus> and email is in one of those dirs.
[13:53] <Morbus> but it never says "found email!" ;)
[13:54] <RenatoSilva> Morbus: $bzr plugins, email should be listed, by that I mean run the command 'bzr plugins'
[13:54] <Morbus> yes, and i did, and it was.
[13:54] <Morbus> ran that command on the branch server.
[13:55] <RenatoSilva> try creating a simple branch, for test
[13:56] <Morbus> and, what, put in the branch.conf again?
[13:56] <Morbus> RenatoSilva: here's a question.
[13:57] <Morbus> RenatoSilva: the checkout on machine B was checkedout BEFORE I made the changes to branch.conf on master branch machine A.
[13:57] <Morbus> does that matter?
[13:59] <RenatoSilva> I don't know, but I don't think so. Try not using checkout, try using the branch directly
[14:00] <Morbus> gasp, something worked.
[14:01] <Morbus> determining what it was.
[14:02] <RenatoSilva> if doesn't work, try a fresh branch: bzr init test; copy your lines to ./bzr/branch/branch.conf; create a file, bzr add; bzr commit
[14:03] <bialix> hello again
[14:04] <Morbus> RenatoSilva: hrm. while i figure out what made it work, is there anywy to get the diff inline instead of as an attached file?
[14:04] <bialix> I've just had a situation when after merge new code from source branch was not merged to destination branch
[14:04] <Morbus> RenatoSilva: bah, looks like that's a symtop of using smtplib.
[14:04] <Morbus> *symptom
[14:05] <bialix> any core devs around?
[14:05]  * Morbus goes back to debugging.
[14:05] <RenatoSilva> Morbus: I don't think so, unless you patch it as I did
[14:05]  * Morbus nods.
[14:07] <bialix> oh, nm, PEBKAC
[14:07]  * bialix bangs head by wall
[14:08] <RenatoSilva> Morbus: https://bugs.launchpad.net/bzr-email/+bug/399398
[14:09] <Morbus> annoying enough, i just made a third ocmmit (the last two got emailed), and so far, no email. heh.
[14:09] <Morbus> still waiting.
[14:09] <RenatoSilva> what made it work?
[14:09] <Morbus> no idea.
[14:10] <Morbus> the last two changes i added was commit on push/pull and localhost:25. i wanted to comment one out and see if i could figure out the exact culprit.
[14:10] <Morbus> but, before i did so, i wanted to do one more commit test to ensure it still worked.
[14:10] <Morbus> and i'm still waiitng for that email to come in ;)
[14:13] <Morbus> ok. there it is.
[14:13] <Morbus> took about 8 minutes to get here.
[14:14] <Morbus> so, with push/pull true and localhost:25, it worked.
[14:14] <Morbus> i'm now gonna comment out push/pull and try it again ;)
[14:15] <vila> jam: ping
[14:15] <Morbus> note to self: 11628 (both enabled), 11629 (no push/pull)
[14:16] <RenatoSilva> that bug you linked, the suer still had the problem in last comment, so I don't know if it's invalid
[14:17] <RenatoSilva> s/suer/user
[14:17] <Morbus> yeah, but he was always using mail.
[14:17] <Morbus> i've been using smtplib
[14:17] <Morbus> that's the diff between the two there.
[14:17] <Morbus> i need the utf8 capability of smtplib.
[14:18] <RenatoSilva> if you can understand his problem clearly (did he explain it well? I'm not use, I don't use "centralized" model), and if you can *reproduce* the bug, you can change the status from invalid to confirmed
[14:18] <jam> morning vila
[14:19] <vila> morning jam !
[14:19] <RenatoSilva> s/not use/not sure
[14:20] <Morbus> RenatoSilva: well, i read his use of centralized as the same thing i came in with: i want emailed commits to be in the master branch, not optionable on the client side.
[14:20] <vila> jam: did you see bug #490600
[14:20] <Morbus> RenatoSilva: ok, so confirmed.
[14:20] <Morbus> post_commit_push_pull = True
[14:20] <Morbus> it was that.
[14:20] <vila> jam: spiv proposed a partial fix, but there are still ~2000 errors (from the ~4000 initial ones :)
[14:20] <Morbus> 11629, which didn't have that, didn't come through. 11630, with that, DID come through.
[14:21] <jam> I did, I'm also thinking that spiv's proposal was probably at the wrong layer
[14:21] <Morbus> now i'll test with and without port :25 ;)
[14:21] <vila> jam: babune agrees with you, it's still all red :)
[14:21] <Morbus> 11631 (without 25)
[14:21] <RenatoSilva> Morbus: comment there too, saying you also have the same problem and mainly why the invalid status is invalid
[14:21] <Morbus> 11632 (without smtp_server)
[14:22] <Morbus> RenatoSilva: yep.
[14:22] <Morbus> i will.
[14:22] <jam> vila: all red? Or just red on the slave that doesn't compile the extensions?
[14:22] <siks> what's the proper way to do a init-repo with a unc path? \\server\path and //server/path don't seem to work
[14:22] <vila> Morbus: are you using host:25 notation, I know some parts of the code can't grok the embedded port and requires specific variables for the port
[14:22] <Morbus> RenatoSilva: well, at this point, post_comit_push_pull seems required, even though the docs suggest that it isn't. so i'm not sure it IS invalid.
[14:22] <Morbus> vila: i'm testing with and without at the moment.
[14:22] <vila> jam: all slaves run without extensions, so they are all red :)
[14:23] <jam> vila: really? that seems foolish
[14:23] <jam> given that our preferred way of running is with extensions...
[14:23] <RenatoSilva> Morbus: try answer the first comment that was the one arguing it was invalid
[14:23] <vila> jam: which one should you  chose to run without extensions ? What guarantee will that give you ?
[14:23] <Morbus> RenatoSilva: yeah, without an explanation of what he consideres "appropriate", it's a bit hard ;)
[14:24] <RenatoSilva> Morbus: I'm not sure it is invalid because the user is not satisfied in the last comment, so something's wrong
[14:24] <Morbus> ok, the smtp_server value had nothing to do with it.
[14:24] <vila> jam: anyway running lots of combinations *is* foolish, that's why we use a bot :D
[14:24] <Morbus> i tested with localhost:25, localhost, and no smtp_server at all, and all messages came through.
[14:24] <bialix> siks: is there any error you get?
[14:24] <Morbus> so it's definitely the push/pull config.
[14:25] <Morbus> that was the clincher for em.
[14:25] <RenatoSilva> Morbus: I'm not sure I can get all that info well as I don't use checkouts/centralized model/mail program. I think you can get it better
[14:25] <Morbus> *me.
[14:26] <RenatoSilva> Morbus: you have to convince Robert Collins it it not invalid, just because he's the developer :P
[14:26] <Morbus> aye ;)
[14:26] <Morbus> how would i set it back to status confirmed?
[14:26] <Morbus> i got not widgets.
[14:26] <Morbus> *no widgets.
[14:26] <Morbus> ah, way at the top there.
[14:26] <siks> bialix: [Error 58] The specified server cannot perform the requested operation: u'//127.0.0.1/temp/bzr'
[14:26] <Morbus> anyways.
[14:27] <bialix> siks: do you have permissions rights?
[14:28] <jam> vila: I would run 1 bot per platform with extensions. One bot on one platform w/o extensions. One bot on one platform for 3 major python versions (with extensions). That is N platforms + 1 + 3 bots. Not 3*N or 3*N*2 bots, etc.
[14:28] <siks> bialix: i do
[14:28] <RenatoSilva> Morbus: either way, if the bug is really invalid, it would be nice if that user and you created an answer for people having the same syntoms. That is, provide detailed context of what you were doing, and the resulting problem: no email, no error. Then provide what is the root cause of that. All this of course, if you find that root cause (ps: maybe what's causing your problem is not causing that user's, so there may be need for two q/a, try contacting him
[14:29] <Morbus> he's sub'd to the bug, and so am I, so hopefully he'll see it. i certainly will, if he updates.
[14:29] <bialix> siks: I have slightly different error: bzr: ERROR: No such file: u'//127.0.0.1/temp/bzr/': [Error 53] The network path was not found: u'//127.0.0.1/temp/bzr/'
[14:29] <vila> jam: yeah, I thought so too for some time, and then I started getting weird failures and I came to the conclusion that trying to be smart is not the point. The not point is to avoid regressions, and you never know *where* the regression will come from. That's why we test.
[14:30] <vila> s/not point/point/
[14:30] <bialix> siks: it seems like a bug for me
[14:30] <siks> bialix: i get that with //127.0.0.1/temp/bzr and the error i showed with \\server
[14:30] <bialix> siks: I'm sure it worked in the past when I wrote UNC support
[14:31] <siks> i'm investigating bzr because i really want a new version control system for the company i work for
[14:31] <siks> really tired of freevcs
[14:31]  * bialix said on core devs with disapprobation
[14:31] <vila> jam: there will always be cases like: oh, look, python2.6.2 has a bug but not python-2.6.4 and Fedora will use 2.6.2 as part of its LTS policy, etc
[14:31] <bialix> siks: as workaround you can assign your UNC path to some letter
[14:31] <bialix> and file a bug!
[14:32] <siks> the thing is, we need a central server (which of course is np), and a somewhat central place for the projects
[14:32] <vila> jam: and that's without mentioning pqm not running py2.4 anymore...
[14:32] <siks> freevcs has a hierarchy but i don't see a problem why our core bazaar rep can't be a net dir
[14:33] <bialix> siks: oh, no
[14:33] <bialix> siks: wait
[14:33] <jam> vila: mthaddon just changed pqm over to python2.4 as of.. yesterday I believe
[14:33] <bialix> it works for me with real remote machine
[14:33] <siks> strange
[14:33] <bialix> siks: does not work with 127.0.0.1
[14:33]  * bialix trying with localhost
[14:34] <bialix> ok, I got error if the path not found, but I can't reproduce your case
[14:34] <vila> jam: good to know :) But the point is still valid: whatever assumption *I* will make, I will forget some cases and be doomed when the failure is revealed :) So as long as it's a bot running the tests, the more the merrier !
[14:35] <bialix> siks: with localhost and real existing path I have another error:
[14:35] <bialix> C:\>bzr init-repo \\localhost\share\bzr
[14:35] <bialix> bzr: ERROR: No such file: u'//localhost/share/bzr/': [Error 52] You were not connected because a duplicate name exists on the ne
[14:36] <bialix> twork. Go to System in Control Panel to change the computer name and try again: u'//localhost/share/bzr/'
[14:36] <bialix> siks: works for me with 127.0.0.1 and real shared folder
[14:36] <jam> vila: except you also are a big proponent of test isolation
[14:37] <jam> having 20 bots fail because of one bug
[14:37] <bialix> you may want to look into .bzr.log and see if there something suspicious about error 58
[14:37] <jam> doesn't give us defect localization :)
[14:37] <bialix> siks: do you need password to connect your server?
[14:37] <siks> nope
[14:38] <bialix> maybe if you pastebin the relevant part of .bzr.log I may suggest something
[14:38] <siks> it's not that big a problem anyway, i was just trying out if unc works at all. if we end up using it, it'll be a proper remote host anyway
[14:38] <vila> jam: you got that wrong I'm afraid, defect localization is about having more failures not less, especially with eager tests where the first failure masks the other ones
[14:39] <siks> and i can test it locally without unc
[14:39] <bialix> siks: just a thought: we are trying to make .bzr folder hidden after creation
[14:39] <jam> vila: if the only bot to fail was the one running without extensions, then it is clear that the bug is when running without extensions
[14:39] <vila> it's about making tests more focused with more precise contexts
[14:39] <jam> having 20 bots go red doesn't tell you that
[14:39] <bialix> siks: perhaps your server refuse to change file attributes?
[14:39] <jam> it gives you 20 test files to go sift through to see what is wrong
[14:40] <vila> jam: it's 6 not 20 and it tells me it's it's about running without extensions. *I* read the first report and that's enough to know the cause, I don't need to read the other ones
[14:40] <bialix> siks: so either mkdir on the server does not work, or something other
[14:40] <jam> vila: it is 6 because you don't *have* 20 bots...
[14:40] <bialix> siks: but I see it works for me
[14:40] <jam> I was using it as "BIG_NUM"
[14:41] <vila> But if the failure were only on one slave *then* by virtue of defect localization, we'll know it's related to that slave and not only to running without extensions
[14:41] <Morbus> can i edit a commnt on launchpad? i forgot to sanitze some emails.
[14:42] <bialix> Morbus: uncommit, commit, push --overwrite
[14:42] <Morbus> bialix: no, in a comment on a bug, not in a commit ;)
[14:42] <vila> If we run it on a single slave and it doesn't fail, *we* lose
[14:42] <bialix> Morbus: you can edit first message, but I don't know the way for other comments
[14:42] <siks> bialix: hm i guess i'll investigate a little more tomorrow
[14:42] <siks> running out of works hours soon for today :)
[14:42] <bialix> siks: np
[14:44] <Peng> Even if you edit the first message, anyone can click a link and see the first one.
[14:44] <Peng> Plus a dozen people probably got emailed a copy...
[14:51] <bialix> today is not my day
[15:01] <jam> Morbus: also, note that for users that aren't logged in, all things that look like emails are sanitized
[15:13] <chrisw> hmm, why does Bazaar Explorer on windwos drop a dos box? Why isn't pythonw used instead of python?
[15:16] <jam> chrisw: it requires a bit of change to how the script is built, and just hasn't been done yet
[15:17] <jam> I tried working on it once, and ran into problems with py2exe + pyqt + python2.6
[15:17] <jam> so I couldn't get the build working locally
[15:21] <chrisw> I also tried to click the "config" button and it crashed...
[15:21] <jam> vila: so do you run a second time with extensions?
[15:22] <vila> jam: first with extensions, second with C locale, third without extensions
[15:24] <chx> how can i edit the subject line of the commit email? maybe a plugin? what should i rtfm?
[15:36] <bialix> anybody can suggest a way to compress bzr revision-id to hexascii form, and make sure the string will be smaller at least in half?
[15:36] <bialix> is it possible in theory?
[15:36] <bialix> I'm using revision id in version of the program (bzr version-info output) and it looks long
[15:37] <bialix> so I think if I can compress it and later decomress
[15:38] <tumbleweed> howdy. can anyone give me some good reasons why I should upgrade my projects trunk to 2a instead of 1.6? (1.6 because it makes stacked branches possible)
[15:38] <jam> bialix: zlib.compress(revision_id) ?
[15:39] <bialix> jam: if I have 53 chars in orig revid, then zlib makes result 61 chars long
[15:39] <bialix> in hexascii I'll have 122
[15:40] <bialix> jam: maybe your groupcompress?
[15:40] <jam> bialix: there is a fair amount of non-randomness in revision-ids. If you compress them together it will probably shrink, but there is little that could be done for a single string
[15:41] <bialix> tumbleweed: 1.9
[15:41] <jam> tumbleweed: 2a is generally significantly smaller on disk than 1.6. Depends on the project
[15:41] <jam> but bzr went from 200MB => 35MB, etc.
[15:42] <bialix> maybe I need to reformat timestamp
[15:42] <tumbleweed> bialix: there's no FAQ on this, what's the major improvement in 1.9?
[15:42] <jam> tumbleweed: 1.9 introduced btree indices
[15:42] <jam> but I would still recommend 2a first
[15:42] <jam> only  reason not to is if you expect people to be using old bzr clients
[15:42] <tumbleweed> jam: yeah, I'm feeling the pull to 2a, but we don't have a massive code base
[15:42] <tumbleweed> yes, don't want to make life hard for committers
[15:43] <bialix> jam: maybe it's possible to do as plugin
[15:43] <bialix> hmmm
[15:43] <jam> bialix: possible to compress revision ids? I'm saying that the string is not very compressible
[15:46] <bialix> :-/
[15:47] <jam> if you want to include version-info output, the best you could probably do is to map revision-ids to simple integers, and then reference them that way
[15:47] <jam> then at least you don't end up with the same revision-id multiple times
[15:48] <bialix> and how I can found from simple integer the real revision id? by creating the map again? so it's basically as sha(revid) to get the index?
[15:49] <jam> bialix: you store the map in the output
[15:49] <bialix> no, I can't
[15:50] <bialix> I've got this program version from customer
[15:50] <bialix> in this case I need anyway full revid
[15:57] <jam> bialix: you were saying that the revid is "too big" and wanted to compress it. You could compress the whole output of 'bzr version-info' which would give a smaller total content.
[15:57] <jam> There isn't a trivial way to shrink a 53-byte string that has as much entropy that revids have
[15:57] <bialix> ok,  understand
[15:58] <bialix> I may imagine the way to shrink it based on semantic: email-timestamp-random
[15:58] <bialix> but this semantic not always will be true
[16:11] <bialix> it seems when I merge 2 branches which have tag conflicts I don't get any error message. Is it intended?
[16:35] <lugo> hi
[16:38] <lugo> i'm having trouble with https://bugs.launchpad.net/bzr/+bug/156462
[16:38] <Noldorin> hello. what's the best way to interface with bzr if i want to extra information about certain commits/revision/files/etc. ?
[16:38] <NfNitLoop> Noldorin: bzr log -v ?
[16:39] <NfNitLoop> Noldorin: what sort of extra info?
[16:39] <lugo> since latest entries in that bug have been posted some time ago, has anyone workaround for that?
[16:42] <lugo> or informations on how to update using bzr+ssh instead of http? the cmdline output is http://dpaste.org/NwPm/
[16:42] <Noldorin> NfNitLoop: the sort of info loggerhead extracts
[16:42] <lugo> *information ;)
[16:42] <Noldorin> i've been tempted to cobble together something in asp.net that is vaguely equivalent
[16:42] <beuno> Noldorin, using python is probably the best way  ;)
[16:42] <beuno> or
[16:42] <Noldorin> beuno: mm, so i suspected
[16:42] <bialix> lugo: what tells bzr info on your local checkout?
[16:42] <beuno> bzr-xmloutput
[16:43] <Noldorin> does bzr expose some sort of api though?
[16:43] <Noldorin> ah, interesting
[16:43] <beuno> Noldorin, yes, it has a python API
[16:43] <beuno> you can use xmloutput to get XML, but you may need to tweak it to get extra information
[16:43] <Noldorin> bueno: if i had ironpython on the server, then it would be win. alas, i don't
[16:43] <Noldorin> i see
[16:43] <beuno> right
[16:44] <beuno> verterok built xmloutput to interface with java, and I piled up on top of it for PHP  ;)
[16:45] <Noldorin> so in theory, interfacing it with asp.net shouldn't be a problem either
[16:45] <Noldorin> beuno: what did you write in php on top of it?
[16:47] <beuno> Noldorin, I did xml > mysql
[16:47] <beuno> and then used that data in an internal workflow software
[16:47] <beuno> (match commits to projects and people)
[16:48] <beuno> to some extent, that's what Launchpad does, but without the xml bit
[16:50] <Noldorin> ah i see
[16:50] <Noldorin> beuno: well i'm just using this for webbzr accesss, so i don't really forsee a performance issue here
[16:50] <Noldorin> i think i'll give that a go
[16:50] <Noldorin> thanks :)
[16:51] <lugo> bialix, http://dpaste.com/127539/ , checkout of branch: http://bazaar.launchpad.net/%7Eexaile-devel/exaile/exaile-0.3.x/
[16:51] <beuno> Noldorin, good luck!
[16:51] <beuno> (and maybe verterok will some day get xmloutput in the core!)
[16:51] <Noldorin> cheers. i'll report back if i have any success :P
[16:51] <bialix> lugo: based on the comments in that bug you need to "fix" your master branch url.
[16:52] <bialix> lugo: open .bzr/branch/branch.conf and find bound_location entry; change %7E to ~
[16:52] <bialix> save and try again
[16:54] <lugo> bialix, oh i read tha part but i seem to have misunderstood it, it's working now thank you :)
[16:54] <bialix> please add the comment to the bug
[16:54] <bialix> so other people will understand the workaround
[17:00] <lugo> done
[17:01] <bialix> thx
[17:17] <vila> jam: well done, only 22 left \o/
[17:21] <jam> vila: can you point me to the remaining failures?
[17:22] <vila> jam: http://paste.ubuntu.com/332463/
[17:25] <vila> jam: safely ignoring the spurious error: Socket is closed, this heavily points to the smart server serialize/deserializer not supporting StaticTuple ?
[17:36] <jam> vila: right
[17:40]  * vila EODing
[17:40] <vila> g'night all, may the force be with jam ;)
[17:41] <jml> I wrote this script today that almost does something useful: http://paste.ubuntu.com/332474/
[17:41] <jml> vila, g'night.
[17:43] <fullermd> Whew, vila's gone.  Now it's safe to come out.
[19:16] <james_w> anyone know if it's possible to just have loggerhead serve a directory listing?
[19:16] <james_w> It's serving all the branches below a directory, but I just want to allow it to access something that's not a branch as well
[19:19] <beuno> james_w, not currently supported, no
[19:19] <james_w> ok, thanks
[19:19] <beuno> my gut feeling is that it shouldn't be tto hard
[19:19] <beuno> *too
[19:21] <james_w> I'll find another way around it
[19:22] <james_w> I just wanted a quick hack to avoid troubling IS
[19:35] <emmajane> james_w, have you tried just throwing a .htaccess file into the dir?
[19:35] <emmajane> (with an allow directive)
[19:36] <james_w> so, I've placed the directory I want to serve under .bzr
[19:36] <james_w> because apache doesn't then proxy that to loggerhead
[19:36] <james_w> now I've got a permissions issue
[19:36] <emmajane> http://viralpatel.net/blogs/2009/03/htaccess-directory-listing-enable-disable-allow-deny-prevent-htaccess-directory-listing.html <---- try that?
[19:36] <james_w> because it won't follow the symlink out of its tree I think
[19:36] <emmajane> .htaccess containing: IndexOptions +FancyIndexing
[19:37] <emmajane> (they may have an allowoverride set to None, but it's worth a shot if that's all you're trying to do)
[19:37] <james_w> and the main one has                AllowOverride None
[19:37] <emmajane> grr
[19:37]  * emmajane shakes her fist
[19:37] <emmajane> nevermind then. :(
[19:38] <james_w> so I have http://package-import.ubuntu.com/failures/.bzr/
[19:38] <james_w> but you can't see the symlink there
[19:38] <emmajane> which shows no content?
[19:38] <emmajane> yeah
[19:38] <emmajane> what are the permissions on the target of the symlink?
[19:38] <NfNitLoop> and is your web server set to follow symlinks?
[19:39] <emmajane> I had problems with something similar on DreamHost. It wouldn't do Bazaar properly when symlinked outside of the root directory. I couldn't be bothered to figure out if it was Apache or Bazaar though.
[21:27] <saedelaere> hi
[21:28] <saedelaere> i have a question to bzr-cia plugin. somehow when my changelist is long not all changes are posted to my irc channel. anyone seen the same behaviour?
[21:28] <ckontros> Can anyone tell me what I'm missing here? (the bzr ERROR) http://paste.ubuntu.com/332619 (fresh Inkscape compile from BZR) I'm wondering if I stumbled across an issue for them or just user error.
[21:45] <jelmer> saedelaere, bzr-cia only submits on new commits
[21:46] <saedelaere> sure jelmer, i make a commit. this commit has a long changelist. then not all changes are posted to my irc channel. they get cut off at some point.
[21:47] <saedelaere> by the way you are the creator of this plugin right?
[21:47] <saedelaere> nice work!
[21:47] <jelmer> saedelaere, what do you mean with changes ? the commit message?
[21:47] <jelmer> saedelaere, thanks
[21:47] <saedelaere> yes the commit message
[21:47] <jelmer> saedelaere, bzr-cia submits the full commit message
[21:48] <jelmer> the cutting off is done somewhere inside of the cia IRC bot
[21:49] <saedelaere> ah ok interesting. i just looked in my cia.cv account, but couldn't find anything to configure this. perhaps it is even a limitation of my irc channel.
[21:50] <NfNitLoop> IRC does limit the size of a single message.
[21:50] <NfNitLoop> as well as rate-limit messages.
[21:51] <NfNitLoop> so it could be getting chopped down the line.
[21:56] <saedelaere> this must be the source of the problem. anyone an idea how i could circumvent this?
[21:57] <NfNitLoop>  I don't know bzr-cia.  Is there a way to set a max message length and tell it to split long commit messages into multiple IRC messages?
[21:57] <NfNitLoop> That could get annoying though if someone commits something with a huge message.
[21:58] <NfNitLoop> bzr commit -m `cat war_and_peace.txt`
[21:58] <jelmer> NfNitLoop, it doesn't have anything to do with bzr-cia
[21:58] <jelmer> saedelaere: You might want to ask in #cia
[21:59] <saedelaere> jelmer: you are right better to ask there :) thanks for your help guys and have a nice evening all...
[21:59] <NfNitLoop> Oh.  I didn't realize what cia was.  I thought bzr-cia included its own bot.  heh.  Nevermind. :)
[22:21] <poolie> hi all
[22:46] <jam>  hi poolie
[22:51] <dOxxx> heya
[22:52] <dOxxx> is it safe to use os.makedirs to make the .bzrmeta dir?
[22:53] <jam> dOxxx: since it is only a single directory, what is wrong with 'os.mkdir' ?
[22:53] <jam> but since it is in a working tree, yes it is generally safe
[22:54] <jam> though I believe makedirs raises an exception if the dir exists
[22:54] <jam> so you still have to check first
[22:54] <dOxxx> jam: yeah, I was intending to do that
[23:01] <dOxxx> I noticed there isn't any other mention of .bzrmeta in trunk. Am I pioneer? :P
[23:01] <spiv> Good morning.
[23:06] <starenka> hi, i can't get bzr stable docs in chm (wanna read them on android - pdf viewers suck bigtime) - links are dead, directory listing havent revealed anything either, any chance gettingthem?
[23:06] <dOxxx> jam: or should I be using tree.mkdir instead?
[23:07] <jam> starenka: http://doc.bazaar-vcs.org/bzr.2.0.2/downloads/ ?
[23:08] <jam> or http://doc.bazaar-vcs.org/bzr.2.1.0b2/downloads/
[23:08] <starenka> oh, ok http://doc.bazaar-vcs.org/latest/downloads/ linked from website are bad than (at least for chm)
[23:09] <starenka> jam thank you!
[23:19] <poolie> hello jam
[23:25] <poolie> jam/lifeless: what is ben's actual objection to launchpad? that we should create an faq?
[23:25] <poolie> i mean, that i feel like we should create an faq
[23:25] <poolie> i think it's just that remembering another password would be annoying?
[23:25] <jelmer> poolie, yeah, that's it basically - and another account to keep up to date
[23:26] <dOxxx> perhaps he doesn't like giving out personal info?
[23:26] <Peng> CoughOpenIDCough. :P
[23:26] <jelmer> dOxxx, there isn't much personal info you have to give out
[23:27] <dOxxx> just trying to get inside his head :)
[23:28] <lifeless> poolie: what objection ?
[23:28] <lifeless> poolie: oh, he has a principled stand.
[23:28] <poolie> but it's an oddly pragmatic principle to defend
[23:29] <poolie> i'm just trying to work out what the principle is and I think it is "i don't want to create another account"
[23:29] <lifeless> hes expressed it before as 'just emailing should be enough' too
[23:29] <lifeless> he likes debbugs, for instance.
[23:30] <dOxxx> while an email interface is cool, it is somewhat crude.
[23:31] <lifeless> lamson + html email with javascript.
[23:31] <lifeless> ajax-over-email.
[23:31] <lifeless> would be an interesting hack.
[23:31] <dOxxx> heh
[23:31] <fullermd> A matter of perspective...   it's a cheap universal API.  I broke down and cried the day NSI stopped supporting their email interface for registration manglement  :|
[23:31] <jelmer> is there some way to disable the write-traceback-to-file stuff ?
[23:31] <dOxxx> this .bzrmeta/ignore thing is trickier than it looks. a lot of stuff is broken by the extra directory in the path.
[23:32] <dOxxx> makes tests messier
[23:32] <fullermd> I must say, in this day and age I'm a little creeped out by email control interfaces that don't require crypto.
[23:32] <jelmer> debbugs seems to be doing (mostly) ok without crypto
[23:32] <dOxxx> doesn't launchpad though? I thought that's why you could set your PGP key in your launchpad account?
[23:32] <rockstar> fullermd, the Launchpad interface does if you want to change something.
[23:33] <fullermd> Yah, but that requires either luck (e.g., in the form of being 'not an interesting target'), or a lot of monitoring vigilence.
[23:33] <fullermd> To say nothing of the ability to speel...
[23:33] <Peng> My favorites are the ones where you can make changes that actually cost money over email.
[23:34] <fullermd> Of course, that also ties in with my creepy feeling seeing patches to add TLS to bzr://.
[23:34] <Peng> Or those domain registrars/registries where you can change your nameservers.
[23:34] <dOxxx> would it be cheating if, in bt.test_ignores.TestTreeIgnores.test_add_to_existing if I used tree_ignores_add_patterns to create the initial contents of the ignore file before using tree_ignores_add_patterns again to test adding to an existing file?
[23:34] <fullermd> Shouldn't have to be patches.  Any protocol written since 2000 should be encrypted by default from the get-go...
[23:35] <Peng> fullermd: I'm glad the bzr developers don't have to worry about getting encryption right, though.
[23:36] <fullermd> Except DNS, of course, 'cuz if I have to see one more DNSSEC argument, I'm going to have to kill somebody.
[23:36] <fullermd> (which of course means I'll probably be indicted by the end of the week, but still...)
[23:36] <spiv> fullermd: perhaps we just need a smart server mode that verifies gpg signatures on revisions (and checking the gpg keys are from an authorised set) before inserting them ;)
[23:38] <fullermd> Well, you'd have to go the other way too; smart servers are used for pulling as well as pushing.
[23:39] <fullermd> I just figure the question should never be "why crypto"; it should be "why not crpyto", and absent a really compelling reason it should just be the default posture.
[23:39] <Peng> spiv: That feature already exists; it just doesn't work/.
[23:39] <dOxxx> does this make your skin crawl? http://pastebin.com/m240ae4e7
[23:39] <dOxxx> in a test
[23:42] <dOxxx> as opposed to this: http://pastebin.com/d53952ce2
[23:44] <dOxxx> I'm trying to decide if being abstract about the ignore filename/path is worth the extra hassle.
[23:45] <spiv> dOxxx: I prefer the latter in this case
[23:45] <spiv> dOxxx: because the abstraction actually makes it harder to understand the intent
[23:45] <dOxxx> yeah and to be truly abstract about the path to the ignore file, I'd have to handle the case where there are multiple missing parent directories, which is just insane.
[23:46] <dOxxx> time for a global search and replace :P
[23:58] <dOxxx> I need bzr qshelve :)