/srv/irclogs.ubuntu.com/2009/08/11/#bzr.txt

spivSamB: well, even if the release is 2.0, I'm happy for the verb to be 1.19.00:03
spivI don't want the name of the verb to actually mislead, though :)00:04
SamBspiv: that's what I'm saying. there's no sense in bumping it from a nonexistant release ;-)00:04
SamBindeed00:04
spiv(And I'm happy to pay the fairly minimal cost to bump it when it misses the intended release.)00:04
SamByou don't want it named for a release before/after the one it debuts in ;-)00:05
* SamB finds 2a pretty slow for just about anything involving conversion at the moment, actually ...00:06
SamB... though it's extremely pathetic for it to be using more downstream than upstream during a push, which I assume your branch eliminates00:07
SamBhard for me to test this, though :-(00:07
lifelessSamB: over the network conversions?00:26
SamBlifeless: any!00:37
SamBover network, local ...00:37
SamBif the 2a repository isn't local, that makes it worse, certainly00:37
lifelesslocal should be ok00:45
pooliehullo again lifeless00:46
pooliedid i see you were working on bug 375013?00:46
ubottuLaunchpad bug 375013 in bzr "Commit doesn't honor stacking invariants" [Critical,Triaged] https://launchpad.net/bugs/37501300:46
igcmorning00:47
AfCClassic. Write a long reply to a bzr mailing list thread, only to realize in the end that if I'm going to pontificate like this, I'd better do it in a blog post instead. Bother.00:49
AfC"Save Draft" think about again later00:49
pooliewhich thread, just for curiousity?00:53
AfCpoolie: GPL licencing00:54
AfCpoolie: nothing to do with bzr00:54
AfCI just liked John's "moral" comment and wanted to add something00:55
AfC[battery, back in 60]00:55
pooliejam, i think explaining the gpl is good but it's not appropriate to be so flippant in the context of a business inquiry01:12
lifelesspoolie: yes01:14
poolieyes you agree?01:14
lifelesspoolie: I'm thinking of just untagging it from 2.001:14
poolieoh ok01:14
lifelesspoolie: [bug 375013?]01:14
ubottuLaunchpad bug 375013 in bzr "Commit doesn't honor stacking invariants" [Critical,Triaged] https://launchpad.net/bugs/37501301:14
poolieright01:14
lifelessI heavyweight checkouts will work as they push via streams01:15
lifelesslightweight checkouts over the network are a bad idea anyway01:15
poolieat least as they currently exist01:16
lifelesslightweight checkouts of stacked branches would seem to be a rare case at the best of times01:16
lifelesswell, indeed.01:16
pooliei think we should make them work in future01:16
poolieit's a reasonable thing to want01:16
lifelessfixing commit to use streams would make them suck less hard01:16
lifelesssure01:16
poolieit may require tweaking the definitions a bit01:16
lifelessat the moment, working content local, all history remote, performs badly01:16
lifelessso even with the bug fixed, its not a recommended way to deploy/use bzr today.01:17
lifelesswhich is why I'm thinking of downgrading it to high, or even leaving it critical but untargeting from 2.0.01:17
lifelesswhat do you think?01:17
pooliei think it might be right01:18
lifeless[and it's not 2a specific, it fails with 1.9 if you have content in the tree]01:18
pooliewould it be possible to detect the problem cases and just fail cleanly?01:25
poolieeg to ban committing in lightweight checkouts of a remote repository?01:25
poolieor remote stacked repository01:25
lifelessremote isn't part of the equation01:26
lifelesswe could put a check in commit.py to say if self.branch.repository._fallback_repositories: bail()01:26
pooliein general i'm more comfortable downgrading a bug if it causes a traceback rather than a problem in what's recorded01:27
lifelessyah01:27
pooliewant to add that check, including a mention of the bug number near it, and then downgrade/untarget?01:28
poolieemmajane: hi, still around?01:28
emmajaneping01:28
emmajanejust eating a cow. or myabe it's a steer. definitely it's beef though.01:29
lifelesspoolie: yes01:31
pooliejam, are you back at all?01:59
jampoolie: I'm mostly just heading upstairs, I may be back in another hour or so02:03
poolienp02:03
pooliei was just thinking about your mail about 2.002:03
jampoolie: about the release schedule, etc?02:03
jamI'd be happy to chat with you about that later02:03
poolieyes, and whether this should be 2.0b102:04
poolieok02:04
pooliejust ping me02:04
spivpoolie: ping, mind if I call now?02:04
pooliesure02:08
emmajaneigc, ping. I'll be up for a little bit and thought I'd check to see if you were around?02:11
igchi emmajane02:17
emmajanehey :)02:17
emmajaneHave you had a chance to play with Sphinx any more?02:19
igcemmajane: a little02:19
* emmajane nods02:20
igcemmajane: you have pretty fine control over the title page for example02:21
emmajaneYou can apply style sheets to sphinx pages though.02:22
igcyes02:22
emmajaneI can't remember if you've got the source for what you sent me as a branch?02:22
emmajaneI htink I just remember seeing output?02:23
igcemmajane: if it helps, have a look at https://launchpad.net/bzr-alldocs02:23
emmajanethanks :)02:23
igcemmajane: grab the trunk and see how it hangs together02:23
igcthe interesting files are en/_templates/index.html and en/_templates/layout.html02:24
emmajaneexcellent.02:24
igcemmajane: I'm just building the 1.18rc1 docs as we speak and tweaking that site to reference them02:24
emmajaneawesome :)02:24
emmajaneI love automatic builds of documentation.02:25
emmajaneI recognize that I need better hobbies, but it's right up there with magic tricks and ponies for me.02:25
igcemmajane: well the alldocs stuff isn't automated *enough* yet but that can come post 2.002:25
* emmajane nods02:26
igcemmajane: e.g. still one site per langu rather than something smarter02:26
* emmajane nods02:26
AfCemmajane: we do automatic (ie, live) _screenshots_ for our documentation. http://java-gnome.sourceforge.net/4.0/doc/api/org/gnome/gtk/TreeView.html :)02:26
emmajaneigc, do you have people working on translations, or is it mostly beuno jsut working on one language?02:26
emmajane\o/ Sovereigns of Canada. :)02:27
igcemmajane: no, several people working on different languages02:27
emmajaneigc, cool02:27
igcemmajane: the alldocs website has been translated to French, Japanese and Russian02:27
emmajanevery nice!02:28
igcemmajane: and it's messy right now - if it were done via LP translations, I'm sure we'd get many more02:28
* emmajane grumbles. the ubuntu package of bzr is old.02:28
emmajaneigc, tough call. Docs are hard to translate. LP doesn't necessarily make it easier (we deal with this for the ubuntu docs team as well)02:29
emmajaneLP is optimized for string translation.02:29
emmajanethe User Guide != strings. :)02:29
igcpoolie: fwiw, the NEWS in bzr.dev has 2 'In development' headings and no 1.18 heading :-(02:29
AfClifeless: c.f. our conversation at dinner a few weeks ago, see emmajane's comment above.02:30
igcemmajane: right, but the wrapper site is really just a few strings. see http://doc.bazaar-vcs.org/en/02:30
AfClifeless: and yes, I realize we're working on the problem.02:30
emmajaneigc, you're right. that's not bad.02:31
igcemmajane: but wrapper site aside, I know at least one person is actively working on translating the bzr docs to Japanese02:31
* emmajane taps her keyboard and tries to decide the most elegant way of upgrading and ignoring ubuntu.02:32
AfCthat's easy. Run Gentoo.02:32
* AfC ducks02:32
emmajaneHAHAHAHA02:32
emmajaneyou're HilARIOUS. :)02:32
emmajaneThe gentoo cow is the best part about gentoo. :)02:33
igcemmajane: the cool thing is, I still haven't announced the alldocs project yet so having those translations being contributed already is awesome02:33
emmajaneigc, that is cool!!02:33
lifelessAfC: which specific comment?02:33
AfCemmajane: [but that also explains why I still don't run Ubuntu, and is the gist of my quick aside to Robert a moment ago. As I said, the packaging issue is being addressed. I hope it works out, because then maybe I could run a Debian distro again]02:33
AfClifeless: "grumbles. the ubuntu package of bzr is old."02:34
emmajaneAfC, APT is made of win.02:34
emmajaneGentoo is made of pain.02:34
AfCemmajane: [but this is otherwise off topic for #bzr, and I'm not a distro zealot]02:34
emmajanelifeless, bzr 1.13.1 on python 2.6.2 (linux2) makes me sad.02:35
lifelessemmajane: what release?02:35
emmajanelifeless, jaunty. up to date.02:35
AfCwow02:35
lifelesssounds about right02:35
lifelessit was current when released02:36
lifelessAfC: old releases of distros stagnating isnt really being worked on; rather we're aiming to have point releases for them, for critical fixes etc02:37
lifelessso in this case, emmajane might be seeing 1.13.2 at this point.02:37
lifelessOTOH we already do point releases when we think its important enough, so I'm not all that sure that that will change at all.02:38
emmajane2.0 will get packaged though, right?02:38
lifelessyes02:38
lifelessbut getting it into jaunty is about as hard as getting 1.17 into jaunty02:38
emmajaneI only care because igc's docs are in 2a for which I need 1.16+02:38
lifelesswhich is to say, very.02:38
lifelessemmajane: use the PPA :)02:39
emmajanelifeless, don't you know people who can do that? ;)02:39
lifelessemmajane: yes. The problem is that the delta is huge - 10's of thousands of lines, and SRU policy is to essentially audit that delta.02:39
lifeless<pain>02:39
emmajanecry02:39
lifelessbackports have no such policy02:39
lifelessand neither do PPA's02:39
* emmajane nods02:39
lifelessthus the answer to 'I /need/ a newer bzr' is - use the PPA.02:40
lifelessthe major thing that is going to help here, isn't the new release schedule02:40
AfClifeless: without any emotional charge whatsoever, and I do understand the engineering decisions that led the situation being the way it is (after all, inherited from Debian),02:40
lifelessits doing less formats.02:40
AfClifeless: I really suggest there is something wrong when Canonical Ltd can't tell it's Linux distro project to backport and include another Canonical project's current version. But you and poolie have heard that enough from me, so I really need to stop bothering you about it.02:40
emmajaneYes. Because distros and packages have taught us that progress and change are both bad. :)02:41
lifelessso, poolie and I sat down with james_w one time, he was offering to walk bzr through a SRU02:41
AfCemmajane: That is also a Debian inheritance.02:41
lifelesshe spoke to the RM, who vets SRU's02:41
lifelessAfC: I think emmajane was being sarcastic :)02:41
emmajaneSRU is a TLA for "pain"?02:42
lifelessI have to say though that I think the SRU (stable release policy) is good, and something many customers depend on.02:42
emmajaneah, right.02:42
lifelessin our case, we *have* regressions vs 1.13.1 at the moment02:42
AfClifeless: yeah, well, it's the reason (and pretty much the only one) we migrated away from Debian in 2002, and so I'm very alert for it changing, because I'd love to go back [or, forward, if you will, to Ubuntu]02:42
igcemmajane: once you get 1.16 or later, grab https://edge.launchpad.net/~ian-clatworthy/+junk/bzr-explorer-docs as well02:42
emmajaneigc, :)02:42
lifelessthere are several bugs in the bug database where users are being hurt hugely by changes we've made02:43
lifelesscorner cases, to be sure02:43
lifelessbut thats what the SRU aims to prevent.02:43
lifelessI think it does a decent job.02:43
AfClifeless: frankly, "we [know we] *have* regressions vs" is actually what's important. That's very responsible, and very cool of you. Most projects aren't that mature.02:43
lifelessthanks :)02:43
AfCHm. I need to stop writing about the weather and write about this topic instead.02:44
AfClifeless: (the seed being this, then going to "... but the problem is what happens when customers running a stable release don't upgrade to the next stable release". But again, off topic)02:45
lifelessfinish your small business accounting package. Please.02:45
poolieigc, the 1.18 branch isn't merged back to bzr.dev yet, when we do that it will fix the headings02:45
AfClifeless: got to finish my word processor first, but I'm actually starting to get excited about working on the accounts package again. I think I've figured out a way to make progress.02:45
emmajaneAfC, you're writing a word processor?!02:45
lifelesspoolie: is there anything delaying that merge-back? I'm about to be blocked on  it..02:45
AfC[and yes, this should be followed by a "what the fuck are you doing writing a word processor"?]02:45
emmajaneLOL02:46
emmajaneclearly you were thinking about writing a book but decided that docbook and OOo sucked so much that you'd just write your own word processor first?02:46
emmajaneNot that I'd know anything about that :)02:46
lifelessemmajane: yak shaving FTW!02:46
AfCemmajane: s/docbook/LaTeX/, and yes02:46
emmajaneLOL02:47
AfCyummy yummy yak02:47
poolieafc, so "why can't Canonical tell X to do Y" has many interesting threads02:47
emmajanethere's nothing like an author trying to avoid writing a chapter. :)02:47
poolieone being that people may object, rightly or wrongly, to canonical making Ubuntu do what Canonical wants02:47
AfCpoolie: I know. I'm approaching this as a management consultant, not a techie.02:48
pooliethe short answer is that i think we should meet in the middle02:48
emmajane(afc: http://www.amazon.com/Front-End-Drupal-Designing-Developers/dp/0137136692 is me)02:48
AfCpoolie: (and I don't pretend to know all the factors, so this *is* interesting to me, even if we're all tired of talking about it)02:48
poolieif we make bugfix-only updates they should go through SRU without needing a big hammer02:48
poolieand if a rubber mallet is needed, canoncial will apply that with discretion02:49
AfCemmajane: very nice02:49
lifelesspoolie: I suspect we'll need further process changes roughly orthogonal to the release schedule changes to help cases like emmajane's.02:49
AfCpoolie: the fundamental problem is that this isn't Cisco IOS. It's open source, and for essentially every open source project with the possible exceptions of (eg) Linux 2.4 vs 2.6 and (once upon a time) Apache 1.3 vs 2.002:50
SamBdocbook is a word processor?02:50
AfCSamB: no, XML documentation fomat02:50
AfCSamB, emmajane: wait one02:50
* emmajane finally gets the PPA installed.02:50
SamBAfC: you forgot!02:50
SamBit's also an SGML documentation format02:50
emmajaneSamB, that's old skool.02:51
AfCpoolie: ... and for open source (and commercial software too, but no one admits it), the bug fixes are in the next version, locked up in new features. There's no difference; the new features are the bug fixes; the bug fixes are new features.02:51
AfCpoolie: (this is one of my keynote themes)02:51
SamBthough at this point I guess you should probably use XML even if you do want to use it with DSSSL sheets02:51
igchi SamB02:51
poolielifeless: if you mean "but i don't want just bugfixes i want the new black" then yes, that needs - either good ppas or backports or whatever02:51
emmajanesometimes I wish error messages had LESS information.02:51
poolieafc, i think in some ways open source is just recognizing reality there02:52
emmajaneTracebacks scare me.02:52
poolieemmajane: yeah that's an interesting question too02:52
emmajane+unexpected tracebacks02:52
igcSamB: I've put some work into fast-import in recent days to make it more reliable coming up to the 2.0 release02:52
pooliepossibly we should hide them unless the user actually wants to report a bug02:52
poolieafc, at the end of the road is something like tom lord's position that everyone should just maintain their own branch and cherrypick the features they want or need02:52
SamBpoolie: I loves reporting bugs if I have a clue what bugs I'm reporting02:52
AfCpoolie: so (and now that it's come up, this is essentially be feedback for the 6-month release thread), I am very wary of such a cycle; for all it's benefits I would suggest that the bzr hacker community will have a hard time mustering much enthusiasm to do anything on 2.stable.x instead of bzr.dev02:52
poolieyou may be right02:53
emmajaneFrom a user perspective it looked like this: bzr: ERROR: exceptions.KeyError: 'Bazaar repository format 2a (needs bzr 1.16 or later)\n' (the slash n is awesome) and the lkjadjkldjlksfadjlksdfljk;asadfjl;sadfjl;sadfljk; for a long time and the error again in English.02:53
pooliei think it's worth a try though02:53
poolie!!02:53
igcSamB: after all, 2.0 will undoubtedly mean more people trying to import into bzr02:53
pooliethat's a bit poor02:53
emmajaneThat's what tracebacks look like to me. :)02:53
SamBigc: mmm hmm02:53
lifelesspoolie: yes, thats what I mean (and my hunch is thats what most folk are asking for)02:54
poolieemmajane: if that's what i think it is should be saying 'unrecognized format' and no traceback02:54
igcSamB: so my priorities for fast-import are:02:54
AfCpoolie: "recognizing reality" yes, exactly. But Debian-esque release freezing and upstream evolving reality just don't seem to jive. I think that's the fundamental problem in the open source sphere currently.02:54
pooliei agree they're gibberish02:54
SamBigc: you want me to try getting that tarball importer working?02:54
igc1. make importing reliable and bug free02:54
igc2. make exporting from other tools really easy02:54
lifelessSamB: are you on Mac OSX ?02:54
AfCpoolie and lifeless: anyway, sorry to get into that.02:54
emmajanepoolie, It's plausible that someone tried to help me at some point and turned on tracebacks by default.02:54
SamBlifeless: no02:54
SamBwhy?02:54
AfCpoolie and lifeless: but always a pleasure to talk with (and learn from) you gents.02:55
igc3. fast-export bugs02:55
lifelessthere is a bug in the python tarball module in some versions of the python Apple ship02:55
poolieafc, i think users will never agree 100% on what's worthwhile fixes and what's unwanted churn02:55
lifelessthey shipped a beta, if I remember the details correctly.02:55
poolieanyhow...02:55
SamBlifeless: hmm ... iirc, the script in question is written in perl02:55
igcSamB: any help you can offer on making import highly reliable would be great02:55
AfCpoolie: there's also the "corporate sysadmin" vs "developer" vs "end user". All three have radically different needs.02:55
fullermdSamB: Pfft.  SGML roolz   :p02:55
AfCanyhow02:55
igcSamB: that 'tag from ref' bug or whatever it was sounds important if git-fast-export is now producing that more reguarly02:56
SamBfullermd: well, I don't know if they've been updating the SGML DTDs and those are kind of important if you're going to use abbreviated forms ...02:56
igcSamB: but in general, any help on bug fixing or adding tests is ultra-welcome02:56
AfCSamB, emmajane: I'm happy to talk to you about what I'm working on, I _have_ done up a very preliminary web page, but it's not at "I can download it and play with it myself stage" yet so I'm reluctant to talk it up too much.02:57
fullermdSamB: Well, I don't either; I don't use Docbook.  I was just making a general trol^Wcomment  ;p02:57
SamBigc: actually, that was generated by a slightly-altered form of the tarball import script from git's contrib tree02:57
igcSamB: I'll like to release bzr-fastimport 0.9 next Monday say ...02:57
AfCSamB, emmajane: but this isn't the place to talk about it further, so after a couple URLs if you want to chat with me about it (please do!) -> #java-gnome on irc.gimp.net02:57
lifelessSamB: you know that bzr can import from tarballs itself?02:57
igcSamB: to give some time to get it packaged into karmic, etc.02:57
SamBlifeless: it's nice to just set it loose on a bunch of 'em, though ...02:57
AfCSamB, emmajane: general motivations http://blogs.operationaldynamics.com/andrew/#one-year-and-more and what isn't linked to from there because I'm not promoting it anytime soon is http://research.operationaldynamics.com/projects/quill/02:58
poolieafc, that's nice02:59
pooliei've always thought the thing about 'a man's reach exceed his grasp' is especially relevant to software or similar creations03:00
SamBlifeless: oh, also this fast-import thing gets the datestamps right on the commits, I think ...03:01
SamBthat is, it takes them from the tarballs03:02
lifelessSamB: thats a fairly esoteric definition of 'right' ;)03:02
SamByeah ;-P03:02
AfCpoolie: it's been a fascinating couple of weeks, because I met a poet who understands (well, almost) what open source is. Really great discussions.03:03
SamBigc: so ... if I were going to send you a modified version of a git contrib/ script to include, where should I put it?03:03
SamBoh, exporters03:03
SamB:-)03:03
emmajanepoolie, where would the traceback stuff be turned off/on? I don't see anything in bazaar.conf03:04
SamBhmm ... is there some easy way to import a single file's history from another branch?03:04
SamBa completely unrelated branch?03:04
poolieemmajane: ah i might have been talking about what should happen rather than what does happen03:04
poolieclassic major Australian isp service announcement a while ago "no email should have been lost during the upgrade"03:05
poolieyeah :)03:05
emmajanepoolie, ahhhh03:05
poolieemmajane: we could change trace.py to not show them and only log them03:05
emmajanepoolie, that'd be lovely.03:05
pooliethere's a flag that does the opposite, -Derror, always shows them03:05
pooliewant to send an rfc mail?03:05
emmajanesure.03:05
SamBpoolie: it should be easy to configure it to behave like now, I think03:06
SamBI prefer not to have to dig too deep for my tracebacks03:06
poolieit'd be a small change to trace.py03:06
SamBat least, when it's a bug-type traceback03:06
poolieoh i see03:06
emmajane"the version you want is newer than the version you're using" isn't a bug IMO.03:06
pooliewe could add another flag or somethnig then you could set in bazaar.conf03:06
* emmajane nods. Flags work.03:07
lifelesswe should make unknown formats not traceback, if they are.03:07
SamByeah, that's not a bug per-se03:07
lifelesswe can render it reasonably well03:07
SamBjust an absent feature03:07
emmajaneI'll paste the full message with the RFC.03:07
emmajanethere are kwargs involved.03:08
SamBso ... how would I import a file *with* it's entire history from a branch ... will that thing in that mail I saw recently do it?03:09
fullermdFiles don't have history; history has files.03:11
SamBfullermd: maybe that's true in the soviet union ;-P03:11
fullermdIt's also true in bzr   :p03:12
fullermdAnd mergitatone, for that matter...03:12
SamBwell, *I* was speaking english03:20
SamBnevermind that the file in question is actually under git ;-P03:21
emmajanepoolie, RFC sent. I'm not sure that it quite captures things correctly, so feel free to tell me what my opinion is supposed to be if I got it wrong. :)03:26
SamBlol03:26
* emmajane scrolls up and notes the references to oppressive regimes and figures it all fits quite nicely together. :)03:27
lifelesspoolie: is there a list of failing testids with your plugin?03:55
poolienup, not up to date03:55
pooliei can attach it to that bug03:55
lifelessplease03:55
pooliemany are now fixed so i'll re-run it03:55
lifelesseven a larger list would be helpful - massively reduces the window to look at03:55
pooliei'm tempted to do subunit-sort first, it woul rock for that kind of thing03:55
pooliebut i won't03:55
lifeless:)03:55
pooliemerge back to trunk sent04:07
igcpoolie: new docs built and uploaded - email sent to list04:09
* igc lunch04:09
poolieok04:09
poolieigc my merge will incidentally fix news04:09
* emmajane shuts down for the night.04:17
jamhey poolie, still working?04:23
jamnight emmajane04:23
pooliei am04:23
pooliestruggling to keep awake at 1:23pm :)04:23
pooliemore seriously i would like some lunch soon....04:24
pooliei was going to work this afternoon but away from email and irc04:24
AdysI thought bzr pack was supposed to compress/garbage-collect a repository? After running it on an old one, repo size jumps from 41mb to 54mb04:25
lifelessAdys: it consolidates the storage to reduce roundtrips when accessing it over the VFS04:25
poolieAdys: there are probably some pack files in obsolete_packs that are retained in case someone's still using them04:25
pooliebut that will be removed soon04:25
lifelessAdys: basically you never need to run it04:25
poolieand what he said04:25
pooliejam, did you want to talk?04:25
Adysum, does that mean I should revert back to the old one or it doesnt matter?04:26
poolieit doesn't matter04:27
lifelessAdys: its not harmful04:27
Adysfair enough04:27
lifelessAdys: but its not usually helpful either - bzr will pack behind the scenes when it needs to.04:27
pooliesee bug 304320 and bug 37353904:27
ubottuLaunchpad bug 304320 in bzr "bzr pack should (optionally?) delete obsolete packs" [High,Confirmed] https://launchpad.net/bugs/30432004:27
ubottuLaunchpad bug 373539 in bzr "Add a `clean` command that deletes obsolete packs and uploads" [Medium,Confirmed] https://launchpad.net/bugs/37353904:27
pooliei agree it's surprising/confusing04:27
jampoolie: I'm happy to chat a bit if you want04:29
pooliei was going to ask you about the 1.18rc1 thing04:29
jambut if you prefer to head off, that's fine04:29
pooliei replied04:29
poolieat any rate i don't plan to do anything more on it before i fly so it's not urgent04:29
poolieunless you really want to talk04:30
jampoolie: no, I just read your reply, and it fits pretty well my understanding04:31
jamas far as it goes, I don't really care if we just do a 1.18 and make the big change at 2.004:31
lifelessI'd prefer to make the change at 2.004:31
jamit feels a bit more straightforward, to be honest04:31
poolieok04:31
lifelessavoids problems with debian version numbers around the version number04:32
lifelessnot to mention giving packagers warning04:32
poolietherefore we'll do a 1.18 in a week, and then a 2.0beta1 following that?04:32
poolieafter the default changes04:32
jampoolie: sounds ~ right, though a bit of a stepped up schedule when the next release would hit04:33
jamare you thinking to get a beta out before 2.0-final?04:33
poolieit might be good04:33
pooliei think a bit of extra exposure after we change the format could be good04:33
poolieeven if we don't think every change is finished04:33
jamI'll think about it04:34
jamI was hoping to see 2.0 be the next release04:34
poolieok04:34
poolieme too04:34
pooliei wouldn't have a one-month beta04:34
jambut we'll have to see how things are shaping up04:34
pooliemaybe we can just do an rc, or a couple of rcs, and then see04:34
poolielet's see when things land04:34
pooliebiab04:34
SamBpoolie: 2.0 had better have spiv's branch landed, that's all I can say!04:38
jamSamB: stop working cross format, and it won't matter04:39
jamif you want 2a, use it04:39
jam:)04:39
SamBjam: so ... it doesn't use VFS calls over the hpss if you're using just 2a?04:39
jamSamB: yep04:40
jamwell, no more than the current one does, which is only edge cases IIRC04:40
SamBbut it seems kind of dumb to have it using 95% VFS calls over the hpss like it is ...04:40
jamSamB: so I optimize the cross-format conversion to be fast when you have 2 local branches or "bzr upgrade"04:41
jambecause cross-format over long distances didn't seem that important04:41
SamBjam: that's not very fast either!04:41
jamif you want to upgrade, then you upgrade04:41
jamSamB: it could be a *lot* slower04:41
SamBjam: so?04:41
jamthe primary issue is getting the data out of the old format04:41
SamBI should steal some of your RAM04:41
jamabout 60+% of the time04:41
SamBjam: that's not saying much04:41
SamBfor me, it's at roughly ∞ anyway ...04:42
SamBfor bzr.dev-sized repos04:42
SamBjam: also, you guys haven't upgraded lp:bzr yet04:44
igcpoolie: I think the bzr-windows team has got off to a good start - we have 22 members now05:17
igcpoolie: I'd like to do the same for os x05:17
igcpoolie: do you agree? if so, bzr-mac or bzr-os-x as the name?05:18
igcpoolie: once again, topic #1 will be packaging/installers, following a similar RFI/call-for-volunteers process as the Windows team05:19
bialixigc: I don't quite understand your mail about windows installer changes05:28
igcbialix: which bit?05:28
bialixmaybe I need some coffee first05:29
bialixbtw, Inno can't build msi, but it definitely build one exe05:29
bialixitem #2 is unclear for me, WDYM?05:29
bialixand you talk about consensus. Does there already consensus reached?05:30
igcbialix: Well, if the consensus is that msi is what we want, then we need to look at how we build that05:30
igcbialix: there was a fair amount I think, e.g. ...05:31
bialixmsi is long standing wish with lowest priority as I can say05:31
igcone installer is promoted; easy-install for the simple command-line case05:32
bialixigc: btw qbzr 0.13 is out, it just lacks announce mail05:32
igcbialix: also Explorer & tbzr both in with the latter off by default05:32
igcbialix: oh, ok - I'll roll out a new explorer release tonight05:32
bialixigc: it seems sidnei want to throw away bzr.exe, but awilkins want it stay.What is consensus here?05:33
igcbialix: we keep it I think05:33
bialixigc: TBZR today is barely usable buggy alpha software05:33
bialixit's shame that nobody works on it, but Mark has put very high bar it seems05:34
bialixso I think wil be better to remove it from installer entirely for some time05:34
bialixis it too extreme?05:34
igcbialix: maybe not. post to bzr-window asking if anyone prefers it to explorer? If no-one says yes, then perhaps we could leave it out altogether05:36
bialixwell Naoki seems to trying maintain it05:36
lifelessit was asked for /a lot/ when it didn't exist05:36
lifelessand folk seem to be using it05:36
igcbialix: otoh, maybe plenty of silent users are relying on it being there05:36
igcbialix: I'd be hestitant to remove it given it's part of our current default install05:37
bialixlifeless: in the past there was no alternative. today we have bzr-explorer05:37
bialixigc: perhaps I;ve lost the track of user needs, I'm happy with my custom installer, so I think from my egoistical POV05:38
pooliebialix: i think at least making it optional would be a good step05:39
poolieas i wrote on the list05:39
poolieigc, +1 to a mac os list05:39
bialixigc: if I'l have some free time at lunch time I'll try to consolidate the feedback05:39
pooliethe metric i'm really watching is whether more people post05:39
pooliebut we'll see05:39
bialixpoolie: it's already optional05:40
pooliei think being able to build an installer using only the free compiler would be good05:40
poolieoh one more random thought here - i wonder if we can run win32 python under wine well enough to test?05:40
bialixI have no experience with COM programming, so I have no idea is mingw could compile this stuff05:41
bialixhg guys seems to run windows tests on wine with good success05:41
bialixrun CLI bzr should be no problem for wine05:42
bialixbut somebody should just try05:42
igcpoolie: so preferred list name? bzr-mac or bzr-os-x or bzr-mac-os-x06:07
pooliethe first is shorter :)06:08
igcpoolie: I probably prefer it06:08
=== cprov is now known as cprov-zzz
lifelesspoolie: bug 375013 fix up for review.06:47
ubottuLaunchpad bug 375013 in bzr "lightweight checkout commit to a stacked branch does not work" [Critical,Triaged] https://launchpad.net/bugs/37501306:47
vilahi all07:11
igchi vila07:16
=== loxs_wrk is now known as loxs
lifelessvila: I wonder if I could get a review from you on my merge for bug-37501307:42
vilalifeless: hmm, given the subject, I'm afraid I can't do that dave, but I'lll have a look07:44
vilalifeless: merge request URL ?07:44
lifelessc.l.n/~lifeless/bzr/bug-37501307:45
lifelessis the branch07:45
lifelessclickyclicky from there07:45
vilagot it07:45
vilalifeless: err is the 2.0.0 stuff from your or another mad case ?07:47
lifeless2.0.0 ?07:47
vilahttps://code.edge.launchpad.net/~lifeless/bzr/bug-375013/+merge/996707:47
vilabzr and bzrlib/__init__.py07:47
lifelessignore those07:47
lifelessits trunk being old07:47
vilalifeless: so, if I read that right, this is a one-line fix (in get_commit_builder) and then fixing failing tests ?07:58
lifelessyes07:59
lifelessand a NEWS entry about it.07:59
vilalifeless: but it seems you change some tests to use a checkout instead of a branch ? (I may need to read more deeply, just a quick confirmation)07:59
lifelessI got better at the fix later on07:59
vilalifeless: right the NEWS entry is important07:59
lifelessbut they are == in terms of speed so didn't bother to make them all the same08:00
vilathe feeling I have is that you have hijacked some tests instead of making new ones, am I wrong ?08:02
=== loxs_wrk is now known as loxs
lifelessI think you're wrong08:02
lifelessI don't think I changed the intent of any test.08:02
lifelesssome tests were doing things that they didn't claim to test, so I pruned them back to what they claimed.08:02
vilaha, good, not obvious at first read, hence the question08:03
lifelessI'm off for a while. ciao.08:04
vilalifeless: meh, don't you want an approve ?08:04
LarstiQspm: how did the pqm upgrade go?08:43
bialixigc: bzr-explorer 0.6 small issue with bzr.exe09:03
igcbialix: namely?09:03
bialixAbout dialog fails with import error: no module named platform09:04
bialixbzr.exe lacks std platform module09:05
igcbialix: damn I hate how bzr.exe doesn't have all the python libs :-(09:05
bialixreally? hate?09:06
bialixI can package it into installer for bzr-explorer if you wish09:06
igcbialix: ok, I guess we can catch the import error in explorer and not report the python version?09:06
bialixor into qbzr installer09:06
igcinto the bzr-explorer installer sounds good to me09:07
bialixwhy not report?09:07
bialixsys.version is not good for you?09:07
igcbialix: it will do. I was copying some code from a book sample and I just did as it did09:07
bialixI can look at the code and tweak it (not yet)09:08
igcbialix: please09:08
bialixshould I push tag then or you prefer 0.6.109:08
igcbialix: 0.6.109:08
bialixok, so I'll fix the issue and make new release with installer for 0.6.1 only09:09
bialixand you can announce it tomorrow09:09
bialixok?09:09
igcbialix: yes09:09
bialixgaryvdm about to announce qbzr 0.13 right now09:10
bialixigc: actually I've thought you want to report OS version, like "Windows XP SP3"09:11
bialixplatform module can provide this sort of info09:11
bialixpython version available via sys09:11
igcbialix: I think that would be good but not for 0.609:11
bialixI'm just trying to understand your intent re platform module09:12
bialixI have no chance to look at code yet09:12
* bialix pulls and looks09:12
bialixigc: post 0.6.x I'd like to cleanup tests09:14
bialixthere is ./tests and ./lib/tests09:14
bialixwe have to merge them into one09:14
igcbialix: go for it. Just ./tests is the best direction I think09:15
bialixwhy not ./lib/tests ?09:15
bialixwhat's the reason to keep in the root of the tree?09:15
igcbialix: it just feels more symmetric to me: data, doc, lib, tests, etc,09:16
igcbialix: "lib" is effectively "src"09:16
bialixok09:16
bialixqbzr question: one man building qbzr .deb for Debian Lenny and Etch09:17
bialixwhere he can publish them?09:17
bialixLP PPA can be used?09:18
* igc dinnner09:34
bialixigc: fix for about dialog has pushed09:59
=== cprov-zzz is now known as cprov
bialixLarstiQ, jelmer: hi11:04
LarstiQhey bialix11:11
bialixLarstiQ: are you debian dev/maintainer?11:11
garyvdmbialix: Just updated http://groups.google.com/group/qbzr/web/screenshots11:27
bialixvery nice, thanks11:27
bialixAfC has mentioned they have auto screenshot updater for the docs in java-gnome. I'm not sure how it could be done, but definitely could11:28
bialixespecially for new site with user docs11:29
garyvdmbialix: It was difficult to choose what to include.  I don't have any thing for qcat.11:29
LarstiQbialix: no, I don't have upload rights.11:29
LarstiQbialix: but I do have some involvement11:29
bialixLarstiQ: what is usual practice?11:29
LarstiQbialix: I think I'm missing some context, what do you want to do?11:30
garyvdmbialix: is this to get qbzr into debian?11:30
bialixLarstiQ: one man from ru_bzr want to build deb package for qbzr for Debian Lenny and Etch, he asking me about advices how to publish his work11:31
bialixgaryvdm: yep11:31
bialixI'll start procedure of buying domain11:32
garyvdmbialix: Well I can't see a reason that he could not publish it to the ppa.11:32
bialixlp guys said there is no PPA for Debian11:33
garyvdmoh11:33
bialixyou can look at irclogs of #launchpad11:33
garyvdmbbl11:33
LarstiQbialix: I'm willing to review the packaging work11:33
bialix[11:27]<wgrant>bialix: Launchpad can't currently build PPA packages for Debian. However, if qbzr is Python as I assume, Ubuntu packages should work on Debian.11:33
LarstiQbialix: finding someone to sponsor the upload to backports shouldn't be too hard either11:34
wgrantRight, no Debian PPA support at this point.11:34
bialixLarstiQ: can I give your contact to that man?11:34
LarstiQbialix: sure11:35
LarstiQbialix: hmm11:35
bialixso you can provide some hints11:35
LarstiQbialix: qbzr doesn't seem to be in Debian currently?11:35
bialixare you asking me about this?11:35
bialixI don;t know even where to look to check?11:35
LarstiQbialix: direct him to me, we'll figure it out then :)11:35
bialixokay11:36
* LarstiQ is best contactable on irc btw11:36
* bialix nods11:38
bialixLarstiQ: you talk Serbian?11:38
LarstiQbialix: no, I guess I should remove that, I added that for testing bits of Rosetta11:39
* LarstiQ does so now11:39
bialixnp11:39
jelmer:-)11:39
jelmerbialix: qbzr is indeed not in Debian11:40
LarstiQbialix: I do since speak a little bit of Finnish though ;)11:40
bialix:-)11:40
LarstiQupdated11:41
* jelmer speaks 4 words of russian, not sure how to write them though :-)11:48
bialixjelmer: there is usual practice on writing them using latin alphabet11:55
* LarstiQ heads to lunch12:00
=== bac` is now known as bac
bialixgaryvdm: so, what's our plan for today?13:10
=== deryck_ is now known as deryck
garyvdmbialix: http://launchpad.net/qbzr/trunk/0.13.1/+download/qbzr-0.13.1.tar.gz13:25
bialixok13:25
bialixdo you have a dedicated branch for this?13:26
bialixfound https://code.launchpad.net/~qbzr-dev/qbzr/0.1313:26
bialixgaryvdm: does QtTest is used in qbzr now?13:30
bialixgaryvdm: windows installer has been uploaded13:36
=== mrevell is now known as mrevell-lunch
garyvdmbialix: I forgot to include in NEWS something that say we now require Qt 4.4. I'll include it in the announcement mail.14:13
=== nevans1 is now known as nevans
bialixgaryvdm: cool graph on ohloh, indeed!14:37
=== mrevell-lunch is now known as mrevell
jammorning vila15:11
jamfjalvingh: are you around?15:11
vilamorning jam15:11
fjalvinghjames_w: Hello, yes; im around15:16
fjalvinghjam, I mean 8-(15:16
=== EdwinGrubbs is now known as Edwin-afk
jamgaryvdm, bialix: I just say 0.13.1 announced, is that different from the 0.13 you told me about last night?15:33
bialixнуы15:33
bialixyes15:33
bialixgaryvdm has fixed regression introduced before 0.13 release15:34
igcnight all15:34
garyvdmjam: tisk tisk tisk lots of code just before a release...15:35
jam:)15:39
jamnight igc15:39
fullermdThat's what releases are for, aren't they?  To get people to put in new code?15:39
bialixjam: it seems like your comment about new features before release... it was evil eye15:39
jambialix: it was experience15:39
cody-somervilleI'm having much difficult reverting a single revision15:40
jamcody-somerville: Assuming you mean "I'm having trouble removing the changes introduced in a single revision"15:42
jamthen you probably want15:42
jam"bzr merge -r REV..REV-1"15:42
cody-somervilleBut I have no idea what REV-1 is for 625.1.115:42
jamcody-somerville: before:625.1.1 works fine15:42
jamso "bzr merge -r 625.1.1..before:625.1.1"15:43
cody-somervillethanks15:43
cody-somervillethat worked15:43
cody-somervillebut was way too difficult15:43
=== deryck is now known as deryck[lunch]
garyvdmbailix: There are lots of bug that I want to change to "wish list", so that I can see the real bug at the top of the page.16:23
garyvdme.g. bug 30668816:23
ubottuLaunchpad bug 306688 in qbzr "Allow diff style selection to be remembered or configured" [Medium,Confirmed] https://launchpad.net/bugs/30668816:23
garyvdmIs that ok>16:24
LarstiQgaryvdm: are you using the same bug workflow as bzr?16:39
garyvdmLarsitQ: workflow?16:39
garyvdmLarstiQ: I was looking for what to work on next. I would like to fix bugs first, before working on new features.16:40
garyvdmIt would be nice to be able to prioritize Wishlist bugs...16:41
LarstiQgaryvdm: doc/developers/bug-handling.txt16:52
cr3darn you guys! now I can't type "bar" anymore, it always comes out as "bzr"!16:54
* awilkins has also noticed that bzr has become a short reflex loop16:54
Takfzo, bzr, bzz16:54
garyvdmLarstiQ: Hmm - we are using wishlist for features, bzr doc describes them as very "wishfull thinking buddy...."16:57
garyvdmaka very low Importance.16:58
LarstiQgaryvdm: right16:59
=== deryck[lunch] is now known as deryck
garyvdmHi bialix17:42
bialixevening garyvdm17:42
* bialix reviewing qexport17:43
bialixtoday I'm thinking about automating screenshot work17:44
bialixwe can use qbzr trunk to illustrate some interesting dialogs17:45
bialixand some dialogs could be created programatically17:45
bialixthen only printscreen capture need to be automated17:46
=== kiko is now known as kiko-fud
bialixI hope there is some programmable tools to do this work17:46
bialixwill be nice to get this functionality directly from qt though17:47
garyvdmbialix: That would be cool!17:47
bialixgaryvdm: do you aware about creating screenshots of qt windows?17:47
garyvdmSorry no17:47
bialixI think such feature exists in Tkinter, but I don't remember for sure17:47
luksthere is something in QDesktopWidget17:58
lukshm, maybe not there17:58
luksbut it's somewhere17:58
luksQPixmap::grabWindow17:59
bialixluks, garyvdm: http://doc.trolltech.com/4.5/desktop-screenshot.html18:00
bialixoriginalPixmap = QPixmap::grabWindow(QApplication::desktop()->winId());18:00
bialixluks: yeah18:00
=== beuno is now known as beuno-lunch
luksTestCaseWithTransport automatically creates a standalone branch in the root of the test directory, is there a clean way to disable that?18:27
luksI need to make a failing blackbox test for code that uses WorkingTree.open_containing and with that branch there it will always find something18:28
LarstiQdoes it really?18:29
james_wyeah18:29
luksyes18:29
james_wit's there for safety18:29
james_wand you would be writing a test that would fail on some systems18:29
bialixluks: unfortunately you can't18:29
LarstiQjames_w: ah, so it doesn't recurse upwards infinitely?18:29
james_wperhaps creating a type of bzrdir that says "stop looking here please" for use in tests would help18:30
james_wLarstiQ: exactly18:30
luksI'm not sure I follow18:30
Takwait, what system allows infinite upward recursion?18:30
luksit's hard to believe that no blackbox test in bzr ever needed that18:31
luksbut I really can't find anything18:31
luksI wonder what would happen if I remove the root .bzr directory18:32
statikhi bzr hackers! on one of my machines, i'm trying to use bzr and bzr-gtk, both installed from https://edge.launchpad.net/~bzr/+archive/ppa. I get this failure: Unable to load plugin 'gtk'. It requested API version (1, 13, 0) of module <module 'bzrlib' from '/usr/lib/python2.6/dist-packages/bzrlib/__init__.pyc'> but the minimum exported version is (1, 17, 0), and the maximum is (1, 17, 0)18:32
luksor perhaps just move it and then move back18:32
statikI'm trying to do this on jaunty (9.04), and having some trouble figuring out how to get a usable version of bzr-gtk18:33
luksbah, so this means I can't write the test18:34
bialixluks: I suppose it's for safety, so your test don't open the branch outside you testing sandbox18:35
luksbialix: yeah, I just realized that18:35
bialixI've stuck with the same probkem recently18:35
bialixproblem18:35
luksI'll just have to ignore that case18:35
jamluks: why don't you just create another dir that is a repository18:36
jamwithout a working tree18:36
jamself.make_repository('.')18:36
jamthat should trap WT.open_containing() so it doesn't find the one in the test root18:36
luksoh, that will work?18:36
jamluks: try it, but I'm 90% sure it will work18:36
lukscool, thanks18:36
luksworks great18:38
garyvdmstatik: I'll probable end up trying to convince you to try qbzr, but let me try help with bzr-gtk first. What version of bzr-gtk do you have installed? This can be seen from bzr plugins -v18:38
jamvila: are you still around? (not that you should be)18:38
jamstatik: you need to get the latest bzr-gtk from its ppa, IIRC18:39
jamknown issue that bzr-gtk had not actually released code for a long time18:39
statikgaryvdm: thanks for the help. jam: ah, I didn't realize bzr-gtk had it's own ppa, I was confused because I saw bzr-gtk packages in the main bzr ppa18:39
jamstatik: well, I'm not positive about that18:39
jamI'm looking18:39
jamit may just be that nobody packaged the latest release18:40
statikgaryvdm: i had the version from the ~bzr ppa, but i think i might try switching to qbzr or bzr-gtk trunk18:40
statikgaryvdm: if i'm running bzr 1.17 is it better to use qbzr from trunk or from a ppa somewhere?18:40
garyvdmjam: If so, someone should either copy the new ppa to the bzr ppa, or delete the old ones.18:40
jamgaryvdm: I don't think there is a bzr-gtk ppa after all18:41
jamvila was kind enough to force an actual release of bzr-gtk code18:41
jambut I don't see any .debs around18:41
garyvdmstatik: the one in the bzr ppa is 0.12, which is fine for 0.17. The latest release is 0.13.1, which is only in the qbzr ppa18:41
jamgaryvdm: other than the fact that I just hit "copy" for 0.13.1 => bzr's ppa18:42
statikgaryvdm: ~qbzr-dev is the qbzr ppa you are referring to?18:42
jamI guess Is hould have only copied it to the beta18:42
garyvdmjam: Oh - cool18:42
jamsince 1.18rc1 isn't in the bzr ppa...18:42
jamdoes 0.13.1 work with 1.17?18:42
jamstatik: I do see: http://bzr.debian.org/pkg-bazaar/bzr-gtk/ubuntu/18:43
garyvdmjam: 0.13.1 should be fine with 0.16, and 0.1718:43
jamI don't really know what's up with that18:43
jamgaryvdm: well, hopefully 1.16 and not 0.1618:43
jam:)18:43
jamthough if it still works with 0.16, great18:43
garyvdmjam: err- yes - I get so confused.18:43
jamgaryvdm: yeah, the numbers are "close" but still not in sync18:43
jamuh-oh18:44
jamit looks like *I* was the last person to upload bzr-gtk18:44
jamthough that was back in 2008-08-2618:44
statikhaha, with qbzr 0.12-1 'bzr help qbzr' says "Not finished -- DON'T USE" in the description18:44
jamstatik: interesting. certainly should be updated. It is the default package on windows for a while now, and is in rather good shape18:45
garyvdmstatik: bzr qbzr is our main window, which is not finished. All the other command are very useable18:45
statikbut, yay, qannotate works nicely and I can get back to spelunking through my code reviews :) thanks for the help!18:45
statikgaryvdm: ah, that makes more sense18:45
garyvdmstatik: e.g. qlog, qcommit, etc.18:45
jamgaryvdm: you might want to reconsider the name of your main window, so that you get help on the plugin rather than the command18:46
garyvdmjam: yes, I plan to rename it to qmain.18:46
ryanakcaIs there an equivalent to git add -p, that's to say, only commit parts of the changes in a file?18:46
LarstiQstatik: contrast `bzr help plugins/qbzr`18:46
garyvdmbialix: ^^^ any objection to doing that now...18:47
jamryanakca: not directly, but you can use "bzr shelve" to back out parts of the change temporarily and then "bzr unshelve" it later18:47
bialixone sec18:47
LarstiQryanakca: on _add_?18:47
jamLarstiQ: 'git add' stages things to actually be committed18:47
jam(as in 'git add' stores stuff in the repository)18:48
statikLarstiQ: ah, thats what i was looking for. thanks!18:48
jambut doesn't reference it until 'git commit'18:48
ryanakcajam: but to shelve just say, lines 8-12 , but not lines 25-75 ?18:48
bialixgaryvdm: objection for what?18:48
=== beuno-lunch is now known as beuno
jamryanakca: bzr shelve is interactive18:48
garyvdmbialix rename qbzr to qmain18:49
ryanakcajam: Awesome, thanks18:49
jamwith the main advantage that you can then *test* what you are committing18:49
bialixgaryvdm: ah, no, of course, it's hidden anyway. But please write NEWS, because some people seems to try using it18:49
LarstiQryanakca: although it doesn't support hunk splitting atm18:49
garyvdmbialix: lots of new users type bzr help qbzr, and get a not so nice message about  not been complete.18:50
ryanakcaLarstiQ: OK18:50
bialixyes, let's do it18:50
SamBjam: it would be nice if bzr revert was interactive too ...18:52
SamBor could be18:52
mzzcouldn't you shelve the changes you want to keep, revert everything else, then unshelve?18:55
mzzI think I've actually done that18:55
=== kiko-fud is now known as kiko
* SamB wants "bzr vizmissing" 18:55
LarstiQmzz: yes18:55
jamSamB: bzr qlog branch1 branch218:56
jama bit of extra info18:56
jambut would show you what is on one branch and not the other18:56
SamBjam: I don't think I have enough free disk space to install qbzr18:56
jamSamB:  I thought viz may have also been updated to handle multiple branches18:57
SamBoh, true18:57
SamBtoo bad I can't do branch1...branch2, though :-(18:57
jamSamB: bzr log -r -1:../branch1..-1../branch2 ?18:57
jamnot sure how that works with viz18:57
garyvdmSamB: bzr qlog . branch218:58
SamB"bzr viz ../trunk .&" does something ;-)18:59
meoblast001hi, i have a question18:59
garyvdmSamB; you can do that with vis if the missing revisions from branch2 are available in branch1's repo18:59
meoblast001suppose i have some binary files in my program18:59
jammeoblast001: I have a dream :)18:59
meoblast001if i modify and push those binary files regularly, does that increase the size of the branch rapidly?19:00
SamBgaryvdm: they are in the same repo, actually -- are you saying that viz only looks in one repo?19:00
meoblast001and make people have to download more19:00
garyvdmSamB: I the revisions are not there, it will fail silently :-(.19:00
SamBmeoblast001: it depends on how much serialization changes when you change the file19:00
SamBmeoblast001: what sort of binary files?19:00
meoblast001.blend files19:01
meoblast001it's a game19:01
garyvdmSamB: yes, vis only looks at 1 repo, qlog is able to look at multiple.19:01
SamBhow does .blend work?19:01
mzzmeoblast001: I'd expect it to depend heavily on the file format (if the file's compressed it'll suck, for example)19:01
LarstiQ.blend is not compressed19:01
LarstiQand it's chunk based19:02
meoblast001is that good?19:02
SamBprobably!19:02
LarstiQmeoblast001: I think it should work reasonably well, but I haven't tried it in practice yet.19:02
jammeoblast001: I would say... It depends19:02
jamthe latest format (--2a) should do better at producing binary deltas than previous formats which did line-deltas19:02
meoblast001is all the data for revisions held in .bzr, i could check19:02
lukshm, what do I have to do to resubmit a merge request on LP?19:02
jamluks: mark it "resubmit"19:03
jamversus Approve/etc19:03
LarstiQmeoblast001: in the .bzr of the repository yes (ie, the .bzr with .bzr/repository/)19:03
meoblast001yuck19:03
meoblast00127 megabytes19:03
SamBmeoblast001: does that include obsolete_packs ?19:03
luksjam: thanks19:03
LarstiQmeoblast001: for what treesize?19:04
meoblast001and the entire directory for my game is 2819:04
meoblast001actually, that doesn't seem right19:04
LarstiQmeoblast001: at how many revisions?19:04
garyvdmSamB: I would like to get some feedback, how come you are using bzr-gtk, as appose to qbzr?19:04
meoblast001about 5019:04
LarstiQmeoblast001: ok19:04
SamBgaryvdm: qt is too big19:04
meoblast001so is that bad?19:04
jamgaryvdm: hysterical raisins, I imagine19:04
garyvdmSamB: ok19:04
meoblast001it seems pretty bad19:04
LarstiQmeoblast001: a bit hard to tell yet19:04
SamBand my /usr is pretty full19:04
meoblast001at revision 1000 we'll have several gigs of stuff19:05
LarstiQmeoblast001: well, how quick has it been growing?19:05
meoblast001slowly19:05
=== Edwin-afk is now known as EdwinGrubbs
meoblast00150 revisions over the course of a half year19:05
LarstiQmeoblast001: no I mean, has it been 27meg from rev1 and stayed the same?19:05
meoblast001would you recommend removing all the art and putting that in a seperate location?19:05
mzzmeoblast001: I don't follow. Does 28 for the "entire directory" include those 27m for .bzr?19:05
meoblast001mzz: yes19:05
mzzok, that's odd19:05
LarstiQmeoblast001: ah, I thought without .bzr was 2819:06
meoblast001wait a second19:06
meoblast001this is weird19:06
jammeoblast001: just paste the results of "du -ksh .; du -ksh .bzr"19:06
meoblast001hm, my tree is 27 MB19:06
mzzmeoblast001: are you sure the directory wasn't much larger in the past? 27m of .bzr for 50 revisions of a tree that's on average 1m seems a bit larger than usual.19:06
meoblast001so let me check .bzr19:06
mzzor that.19:06
meoblast001it is also 2719:06
meoblast001that's weird19:07
mzzyeah, that sounds more reasonable.19:07
garyvdmSamB: I have spoken to many people that have that reason. So many I often consider creating a gtk port of qbzr. (I think that would be eaiser than porting qbzr's feature to bzr-gtk.)19:07
mzzmeoblast001: data inside .bzr *is* compressed, so it's possible for it to be a bit smaller than your source tree is, if you don't have many revisions and they compress well.19:07
LarstiQgaryvdm: eh, that sounds like a lot of work19:07
lukscome on, QtGui+QtCore is like 10M or so19:08
SamBgaryvdm: well, another approach might be to get the python qt bindings split into smaller pieces19:08
meoblast001well, 50 revisions, .bzr is about the same size as the rest of the source tree19:08
luksI doubt that's the real reason19:08
meoblast001does that sound like it's good?19:08
mzzmeoblast001: that sounds about right to me, yes. But I'm no expert :)19:08
LarstiQmeoblast001: that sounds belieavable, yes19:08
meoblast001so i should continue the way i'm going?19:08
SamBluks: the python bindings are monolithic, though19:08
mzzmeoblast001: I would, but ymmv19:09
SamBluks: in Debian, at least19:09
meoblast001ymmv?19:09
luksSamB: in debian, they can be built separately19:09
mzzmeoblast001: your mileage may vary19:09
garyvdmluks, SamB: Yes, thats why U loged bug 38140919:09
meoblast001ok19:09
ubottuLaunchpad bug 381409 in python-qt4 "python-qt4 depends on phonon" [Undecided,New] https://launchpad.net/bugs/38140919:09
garyvdm*I19:09
SamBgaryvdm: ah, thanks19:09
SamBI was just about to go see if I had gotten around to it ;-)19:10
garyvdmlol19:10
LarstiQgaryvdm: how does it pull in phonon?19:10
garyvdmLarstiQ: I describe it in the bug19:10
SamBgaryvdm: hmm, is there a debbug for that too?19:10
* LarstiQ has python-qt4 installed on Lenny, no phonon19:10
luksphonon is a part of Qt19:10
luksit was added in 4.5, maybe you have something older?19:10
SamBI think python-qt4 also pulls in the qitchen sink19:11
mzzprobably19:11
LarstiQluks: yup, 4.419:11
mzzafaict it's quite possible to split PyQt4 into smaller binary package just like Qt4 itself is19:11
mzzpackages, even19:11
bialixpython experts help needed: is it valid to write: def func(self, workdir=None, *args): ?19:11
jambialix: I believe it is "def func(self, *args, workdir=None)"19:12
mzzbialix: yes, but it doesn't do what you want.19:12
jamnot sure, though19:12
jamah right19:12
bialixmzz: it's already splitted in Windows package19:12
mzzbialix: what jam said is a SyntaxError. What you said works but if you call it with two arguments "workdir" eats the second one, not "args"19:12
jammzz: right, I forgot19:13
mzzbialix: you want python 3's keyword-only arguments. Without those the best you can do is take **kwargs and parse them manually.19:13
jamwhich is why we went with "def func(self, *args, **kwargs)" and then inspect kwargs in some of our code19:13
mzzexactly19:13
bialixI can relax and don't set workdir as None19:13
mzzbialix: yeah, making workdir non-optional makes it work as expected too19:14
bialixok19:14
bialixthx19:14
garyvdmLarstiQ: How knowledgeable are you with debian packaging? Could you have a go at fixing bug 381409 in Debian?19:15
ubottuLaunchpad bug 381409 in python-qt4 "python-qt4 depends on phonon" [Undecided,New] https://launchpad.net/bugs/38140919:15
vxnickis there any plugin docs other than the two pages on the bzr website? Struggling a bit to make a plugin, despite checking things like builtins.py19:15
SamBmzz: yeah, and it's presumably not that hard to split libc6-dbg just like libc6 is split ...19:15
mzzpython 3 lets you write both what you said and what jam said, and they'll behave differently when they get a single positional arg, but you probably can't use that yet.19:15
mzzand I have no idea if that feature will be backported.19:15
LarstiQgaryvdm: I expect that to be fairly involved19:15
SamBmzz: I don't think it will be19:15
mzzSamB: oh, I guess it'd be backwards incompatible for things like the inspect module, bleh.19:16
SamBmzz: I was just thinking that probably that kind of CODE already means something else in Python200019:16
mzzSamB: no, it's a SyntaxError19:16
SamBoh19:16
garyvdmLarstiQ: Do you think you could push the bug to Debian's bug db, and nag a dd to work on it for me?19:16
SamBin that case I could see it being backported19:16
LarstiQgaryvdm: sure19:16
garyvdmLarstiQ: Thanks19:17
mzzSamB: I'd have to check, but I bet it'd make it impossible for older code to introspect functions making use of the new feature.19:17
SamBmzz: it might cause difficulties, true19:17
SamBbut that's not nearly as bad as breaking everything!19:17
SamB(potentially)19:18
mzzSamB: fwiw: http://paste.pocoo.org/show/133700/ (python 3.1)19:19
SamBI don't suppose anyone here would have any sway on http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=520680 ?19:19
ubottuDebian bug 520680 in libc6-dbg "libc6-dbg: Debug symbols not broken up by runtime package" [Normal,Open]19:19
SamBit's marked wontfix19:19
LarstiQSamB: I'm afraid that is not likely to change19:21
mzzSamB: his arguments aren't that good imho, but I can't help you there.19:21
LarstiQmzz: not good?19:21
SamBhmm, I guess I'll just see if I can get #gdb involved ...19:21
mzzSamB: specifically I'm pretty sure those mozilla-related -dbg packages are huge just because the debug info for those packages is, not because there's one debug package per source package19:22
SamBsome of them work at redhat, which seems to have a nice solution to the "too many -dbg packages" problem ...19:22
mzzSamB: and I don't know enough about debian to understand why "increase the number of packages in the archive" is problematic19:22
SamBmzz: psh, I don't care about the mozilla ones19:22
LarstiQmzz: the point is, _all_ -dbg packages follow one-per-source19:22
mzzSamB: yeah, but that's over half of what he listed in response to your "what other packages are you referring to"19:22
SamBI just want libc6-dbg split by architecture like the runtimes are19:22
mzzLarstiQ: yes! it should be fixed for all of them :)19:22
SamBLarstiQ: surely that can be fixed19:22
LarstiQmzz: eh, no19:23
SamBand he never mentioned *that* as an argument, either19:23
SamBI would have recognized that as a valid, reasonable, concern!19:23
LarstiQeuh, I thought he said that19:23
* LarstiQ reads again19:23
SamBnot that I could see19:23
mzz"debugging packages in Debian are all using the same layout: one per19:23
mzzsource package"19:23
LarstiQSamB: he did: Yes, debugging packages in Debian are all using the same layout: one per19:23
mzzugh, linebreak, sorry.19:23
LarstiQsource package. A lot of them are a lot bigger than libc6-dbg.19:23
SamBoh, he did ?19:24
LarstiQmzz: number of packages in the archive is an ongoing concern.19:24
mzzand yes, many others are bigger, but the ones he listed are bigger just because they're bigger, not because they're from a source package that's split into many binary packages.19:24
SamBoh, but then he went into that fallacy "other packages are doing the same thing, so it's fine if I do!"19:24
mzzSamB: not necessarily. It's entirely possible this is policy, not just convention. I wouldn't know.19:24
SamBif he hadn't gone and pointed the finger at irrelevant packages I would have been more well-disposed towards his argument19:25
bialixgaryvdm: in side-by-side qdiff color lines becomes thickier, it's intended behavior19:25
bialixgaryvdm: ?19:25
mzzSamB: that doesn't matter if it is indeed policy. You'd have to get policy changed.19:25
LarstiQSamB: I think you parsed that wrongly19:25
bialixgaryvdm: I mean green and red lines19:25
SamBLarstiQ: could be!19:25
SamBbut I don't care about mozilla symbols19:26
SamBthe way it dies is that I kill it for it's RAM anyway19:26
garyvdmbialix: It happens on the side where there is nothing.  The change was not intended, but, I was not too worried about it.19:27
garyvdmbialix: Would be easy to fix.19:27
LarstiQSamB: if you can get an ok from a ftpmaster up front about splitting into more packages, you could bring that to the table19:27
SamBLarstiQ: I don't see any reason for that, though19:27
LarstiQSamB: for what?19:27
SamBwhy duplicate the same source tarball three times?19:27
bialixgaryvdm: I don't mind it stay, bt I've noticed because it seems I've used to old behavior19:28
LarstiQno, not source tarball19:28
mzzSamB: again, I don't know about debian, but if the concern is about disk usage, not number of packages, I'd expect this kind of split to not hurt significantly19:28
bialixgaryvdm: it's not big deal19:28
LarstiQSamB: splitting the -dbg package19:28
SamBoh, you meant .debs?19:28
LarstiQmzz: the concern is number of packages19:28
mzzdrat.19:28
SamBsomething about "users"19:28
SamBwas mentioned19:28
LarstiQSamB: you are argueing with the maintainer about a point that I think will be upheld even if you escalate to a wider audience19:29
LarstiQSamB: how small is your disk anyway?19:30
mzzthe size of my /usr/ isn't that much of an issue next to the size of my /home if I don't clean it up regularly19:30
SamBoh, apparantly I reported the python-qt4 thing as #53575919:31
* mzz prods ubottu 19:31
SamBoh, apparantly I reported the python-qt4 thing as debbug #535759, that is19:31
SamBhow does this thing work?19:31
* mzz gives ubottu a friendly pat on the head19:31
SamBmzz: I have /home on a bigger disk19:32
SamBI might possibly be able to grow /home, I'm not sure19:32
SamBer.19:32
SamBgrow /usr19:32
mzzyay lvm19:32
SamBfor instance, if my old /home is right after it, or my ancient HURD partition ...19:33
SamBmzz: hmm, I'm not using that19:33
SamBis it easy to start?19:33
LarstiQSamB: parted?19:33
mzztoo bad! if you were you'd be able to grow your /usr partition onto the other drive.19:34
fullermdlvm?  Is that what people use who can't get ZFS?  ;p19:34
mzzconverting to it is rather a hassle though.19:34
SamBmzz: I think I'd be more likely to grow /home on that one ;-)19:34
garyvdmSamB: You have to prefix with bug. Bug 53575919:34
ubottuError: Launchpad bug 535759 could not be found19:34
SamBgaryvdm: doesn't work for debian bugs!19:34
LarstiQfullermd: LVM is much older19:34
SamBdebian bug 53575919:34
ubottuDebian bug 535759 in python-qt4 "python-qt4: package too monolithic -- should be split up to match dependencies" [Normal,Open] http://bugs.debian.org/53575919:34
SamBokay, that works19:34
LarstiQfullermd: even the Linux implementation, but I was thinking HP-UX (or maybe AIX)19:34
fullermdI know.  That's no good reason not to troll a little though   :p19:35
SamBfullermd: I thought it was called "tease"19:35
fullermdSamB: You'd have to be a lot better looking before I upgraded you to 'tease'.  Sorry.19:35
SamBI'm not even the one using LVM19:36
SamBI'm still using IBM/MS partition tables19:36
fullermdThat's why we're all making fun of you, see   8-}19:37
SamByeah, as well you should19:37
SamBthat primary/extended/logical thing is *so* stupid!19:37
fullermdPshaw.  It's brilliant; you can run DOS and CP/M on the same machine!  What more do you want?19:38
SamBhow can you run CP/M on it?19:38
SamBand since when does CP/M support fixed disks ?19:38
fullermdI dunno.  I can't afford enough drugs to want to try.19:38
SamBI thought CP/M was for Z8019:38
fullermdIt probably wouldn't run bzr anyway...19:38
SamByeah, getting it to work nicely on DOS would be hard enough19:39
SamBit's not even worth trying on CP/M19:39
mzzI don't think bzr would like the 8.3 filesystem much, even if you managed to find a python that runs on dos.19:43
garyvdmhttp://mail.python.org/pipermail/python-list/2004-December/295818.html19:46
mzzI didn't say 16 bit dos (you could run it in an extender)19:50
SamBmzz: how about DOS 7, then?19:51
mzzhttp://www.caddit.net/pythond/19:51
lukswow, threading?19:52
LarstiQopengl?19:52
* bialix hopes igc will be glad to see qexport landed; it turned out to be more work than I've expected19:52
SamBhmm, the Debian policy manual doesn't have anything obvious about debug packages ...19:54
bialixgaryvdm: we have 26 public q-commands now, 23 of them are equivalents to bzr CLI19:56
bialixwoot!19:56
garyvdm:-)19:56
lukssomebody should port qbzr to git19:57
bialixgtk lags behind ;-P19:57
bialixluks: there is QGit19:57
bialixbut it's only ~ qlog19:57
luksbialix: I don't care much about using Qt, but having sensible UI19:57
bialixwell, ok19:58
bialixso I've misunderstood you19:58
bialixwhat's the sense to port bzr-specific tool then?19:58
lukswell, I miss the tools I know :)19:58
bialixyou don't use bzr at all?19:59
luksI do19:59
luksbut I had to deal with git too recently19:59
garyvdmluks: use bzr-git20:00
luksthat would be too slow :)20:00
garyvdmbialix: I think one of the things that sets qbzr apart from bzr-gtk this that the ui is very consistent, and consistent with the cli20:04
bialixgaryvdm: ?20:04
bialixoh20:05
bialixI've misread20:05
garyvdmbialix: bzr-gtk's ui is very inconsistent.20:05
bialixI've read bzr-git instead of bzr-gtk20:05
bialixyeah, I've got it20:05
bialixwell, bzr-gtk was collection of different tools at first20:06
garyvdmyes20:06
bialixor maybe still is20:06
SamBbialix: still seems like it20:06
bialixso we have a big bonus in our design20:06
SamBfor instance, why doesn't gmissing look anything like viz?20:07
bialixgaryvdm: we have several blueprints filed, as I've discovered20:07
bialixI'll attach bug reports to them, so we can track it20:07
garyvdmbialix: For tag from qlog, I recommend working from lp:~garyvdm/qbzr/logslots . It should merge into trunk soon. The code is much easier to read.20:15
bialixgaryvdm: I'll wait then20:16
bialixgaryvdm: I'll start working on saving more commit data, including bugs20:16
garyvdmbialix: That will be cool.20:16
bialixI've promised to Craig to implement this20:17
bialixI should do it finally20:17
garyvdmbialix: I did mention this in a mail, but I just want to bump it. It would be cool if the was a api in bzrlib for plugins to provide a default message.20:18
bialixthere is some hook20:19
garyvdmbialix: bzr-builddeb provides a default message (which it pulls from the debian/changelog file.) It would be nice if qcommit could default to that.20:19
* garyvdm looks at the bzr-builddeb code.20:20
bialixI remember there was hook, but documentation does not mention it20:20
bialixjelmer: ping20:20
bialixjames_w, jelmer: someone of you has implemented hook for commit message20:21
bialixI don't remember details20:21
garyvdmbialix: http://bazaar.launchpad.net/~bzr-builddeb-hackers/bzr-builddeb/trunk/annotate/head%3A/__init__.py#L10520:22
bialixMessageEditorHooks, it's about providing template for edit it by user20:23
bialixyes, it's one20:24
* bialix bbl20:25
=== Pilky_ is now known as Pilky
Noldorinhi lifeless20:52
garyvdmNoldorin: CTCP time reply “Wed Aug 12 05:53:21 2009” from lifeless20:53
Noldoringaryvdm: heh, fair enough20:54
Noldorinforgot he's an aussie :)20:54
moldyhi20:57
moldycan i get bzr to show relative pathnames in e.g. "bzr st"?20:57
mneptokNoldorin: uh oh ...21:01
mneptokNoldorin: lifeless is a New Zealander. and quite passionate about that point.21:02
mneptokNoldorin: he's going to go all Morgoth on you now21:02
Noldorinmneptok: woops, lol21:04
Noldorinit's all the antipodes :)21:04
garyvdmmoldy: I don't think so. Why do you want to do that?21:04
moldygaryvdm: because i commonly work from a directory that is not the root of my branch21:05
Noldorinmneptok: tolkien fan too, i must presume?21:05
LarstiQmoldy: not atm, but you can restrict status output to a subdir21:05
moldygaryvdm: having relative pathnames would allow copying and pasting, and possibly make shell completion better21:05
moldymy zsh installation does not do bzr completion well :(21:05
moldyLarstiQ: ok, thanks21:06
LarstiQmoldy: looking up the bug, just a sec21:06
LarstiQmoldy: bug 3015921:07
ubottuLaunchpad bug 30159 in bzr "paths are always from root of branch" [Low,Confirmed] https://launchpad.net/bugs/3015921:07
garyvdmmneptok, Noldorin: I think lifeless currently resides in Sydney21:07
mzzmoldy: yeah, the zsh I have here kinda sucks at it too. I keep typing "bbzr ..." so I get regular filename completion to work around it.21:07
Noldorinmneptok: hah, so i was right! :)21:07
LarstiQmzz: compdef -d bzr too radical?21:08
moldymzz: hehe, ya :)21:08
NoldorinNoldorin: well, at least an aussie in residence21:08
mzzLarstiQ: if that does what it sounds like: perhaps21:08
mzzLarstiQ: some of its completions are kinda ok, although frequently slow, like completing new files for "bzr add"21:09
moldyLarstiQ: glad to hear that this is going to change eventually21:09
LarstiQmoldy: welll21:11
LarstiQmoldy: as you see it is a low priority bug21:11
LarstiQmoldy: it would be _nice_ to fix21:11
LarstiQmoldy: but it's not entirely trivial21:11
fullermdAnd some of us really like root-relative paths  :)21:11
LarstiQmoldy: it lacks someone to dedicate time towards it21:12
moldyLarstiQ: i understand21:13
mneptokgaryvdm: he does reside there, but is a New Zealander. if you'd like to dispute his nationality with him, i'll start making your funeral arrangements. :)21:14
garyvdmlol21:14
moldyi will keep it in the back of my head, maybe i can find the time to come up with a patch21:14
LarstiQmoldy: cool. If you do, feel free to bug me21:15
garyvdmmoldy: you could try the branch linked to that bug.21:15
garyvdmlp:~jdobrien/bzr/bug3015921:15
moldyi will see. right now, i have other stuff to do (the stuff i am actually using bzr for ;), but i will bookmark it and _maybe_ come back to it ;)21:16
LarstiQgaryvdm: I don't think much happened there21:16
LarstiQsome tests21:16
LarstiQmoldy: sure, cool :)21:16
* LarstiQ continues preparing for HAR21:17
fullermdWell, once you have tests, it's easy.  You just bang up a little code randomizer and run it 'till the tests pass.21:17
moldyhehehe21:17
lifelessmneptok: thanks ;)21:20
lifelessNoldorin: hi21:31
lifelessNoldorin: what was the bug number for this problem?21:31
garyvdmHi lifeless :-)21:40
lifelesshi garyvdm21:41
lifelesshows things?21:41
garyvdmGood. You21:41
garyvdm?21:41
lifelesspretty good; the weather can't decide if its appalling or sunny though :(21:42
lifelessjml: hi; enjoying your self/21:46
lifeless?21:46
nekohayohey there, how is the status/health of the bzr git translation thing?22:48
mneptoklifeless: i know the difference between Hobbiton and Alice Springs ;)22:55
lvhhi23:03
nekohayoI tried to branch pitivi's git repository by installing bzr-git and doing a "bzr branch git://git.pitivi.org/git/pitivi.git pitivi-bazaar", but I just get a huge traceback, including NameError: global name 'name' is not defined23:14
thumperjelmer: ping23:23
thumperjelmer: I was going to also ask about bzr-git and git://git.pitivi.org/git/pitivi.git23:24
thumperjelmer: probably some weird edgecase23:24
Noldorinlifeless: hey, you there?23:50
lifelessyes23:51
lifelessdid you see my question?23:51

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