/srv/irclogs.ubuntu.com/2011/12/02/#bzr.txt

crtxchi.  Internet where I am is very very unreliable and slow.  Is there a way to resume a branch that has been cut off?  I am trying to branch emacs and it keeps stopping, then I have to start the whole thing over.  I can't find anything via search.08:33
vilacrtxc: you probably want to pull by batches, so:08:48
vilabzr init-repo emacs ; cd emacs ; bzr init trunk ; cd trunk ; bzr pull <trunk url> -r10008:49
vilathen 'bzr pull -r200' until you get to the tip08:49
vilause any increment that your connection can support 1.000 10.000 100.00008:50
vilaonce you get the full history, things should be more reliable as you'll need to download only the new ones08:50
viladiscussions about allowing more robust branching by automating the above occur from time to time but I don't remember anybody going ahead and implement it08:52
mgzmorning all08:58
crtxcvila, oh, ok.  I never came across -r200, that sounds much better for lower the probability of line failure.  Ty very much.09:01
vilamgz: _o/09:03
crtxcvila, is -r revision ARG?  I am trying to find it in the man.09:07
vilayup09:08
crtxcvila, hats off to ya, that is a very clever solution. :)09:09
crtxcty09:09
vilayeah, I learned a few tricks by lurking in this channel so these kudos are really for more than me (/me blushing)09:11
crtxcif you have a web site or a blog you might want to make a small post and get some traffic with that.  I see a few people had trouble with that.  :)  Then kudos to all who have good brains. :)09:12
crtxc:)09:13
* fullermd is willing to accept any and all credit.09:14
crtxc:)09:14
vilaha, right, yeah, that's all fullermd fault, now that you mention it :)09:15
vila's ?09:15
crtxcfullermd has been so kind for hidding the i an n between the m and d.  If fullermd weren't so shy it would be fullermind for all to see.  :)09:17
fullermdHere now.  *I* handle the bad puns around here.  You're on my turf, sparky!09:19
crtxc:)09:19
jelmerhah, finally figured out what the trouble was with those multiple upstream tarballs11:13
vilajelmer: \o/ ?11:14
vilaplanets should be well aligned, my wifi problem just... resolved itself (using canal 6 instead of 13 *cannot* be related right ? Not when the only laptop that was hanging trying to connect is mine, right ?)11:15
mgzwifi goes by water in france :)11:18
quicksilverantenna in your laptop is damaged and finds some channels easier to tune than others?11:23
quicksilverthe breed of chicken you sacrificed was not holy enough to use channel 13?11:24
vilagoats, I always use goats... But now that you mentioned it I may have to audit my provider...11:25
fullermdI guaranteed the goats for SCSI.  I didn't say nuffin' about 802.11.11:38
sorenvila: Not all wifi adapters can go up to channel 13.11:55
sorenvila: I've had a few that only went up to 11.11:55
mgzI thought fullermd said he was in charge of the bad jokes...11:55
vilasoren: good point, but the router upgrade triggered the issue while the previous was already using channel 1311:55
fullermdWhy don't you just raise the frequency of 10 a little, and have 10 as the top channel?11:56
sorenvila: Oh.11:56
sorenfullermd: lol11:56
vilasoren: still,  thanks for pointing that11:56
vilaSCSI ? Come on ! I still have cables and even scsi drives... in my attic ! But I was clear enough when we engage on this goat deal that they won't be used outside of sabbat where virgin (and not goats) are used anyway11:57
vila(the scsi stuff I mean, not the goats)11:58
vilajelmer: what's your feeling about your reconcile proposal and bug #897487 ?11:58
ubot5Launchpad bug 897487 in Bazaar "progress indication for Repository packing and reconciliation over HPSS" [Medium,Confirmed] https://launchpad.net/bugs/89748711:58
vilajelmer: should the bug be a pre-requisite ? Do you think you can fix it soon ? Should we just land and see ?11:59
vila4) should we raise the importance to high ?11:59
vilawriting help thinking, so: I bumped to high and approved ;)12:01
jelmervila: I wonder what the best way is to provide progress reports12:09
jelmervila: if it's sufficient to just stream some data, then we can easily do that, even if it's not very useful data12:09
vilajelmer: I don't know the code well enough to say if multiplexing data and progress is easy or not, so if there is data you can stream, good enough12:10
jelmervila: btw, on another note12:11
jelmervila: if we're looking at fixing the import code in bzr-builddeb, I wonder if we could stream directly from the tarball into commit builder12:11
jelmervila: I've been wondering about that previously as well, because currently we have to check out the base parent on disk, etc12:12
jelmerso it would improve performance, but it would also help my current work12:12
vilamy understanding is that import is not simply a commit, there could be renames and kind changes12:12
vilaand deletions, and so on12:12
vilawhat I don't understand though, is why we can reuse more of our existing stuff. Are renames and the like really the issue again here ?12:13
jelmervila: deletions and so on shouldn't be a problem,12:13
vilaas in: some operations cannot be automatically deduced12:13
jelmervila: ? I don't see how that matters, the only difference is that we're reading the contents for the commit from a tarball rather than off disk.12:14
vilaha, let's sync12:15
vilaI was referring to the actual implementation that seems to use a transform tree to better handle renames and file-id reuses12:15
vilayou're talking about using a different source for commit, which indeed sounds like a good plan (gmta ;), but what about file-id reuse and renames ?12:16
vilaare the existing tests good enough to cover that ? (obviously nothing covers bug 800270)12:18
ubot5Launchpad bug 800270 in bzr-builddeb "import-tar incorrectly handles symlinks turned into directories" [High,Triaged] https://launchpad.net/bugs/80027012:18
vilajelmer: still there ?12:32
mgzI think being able to do certain transforms without having a tree on disk is something we need.12:32
jelmervila: sorry, back now12:34
jelmervila: I'm pretty sure there are some tests for preserving file ids12:35
vilaand renames ?12:36
vilamgz: wow, you're raising the bar now ;)12:37
vilabah, pqm not using news_merge is now really going on my nerves12:38
vilahmm, could it be that pqm is using a too old bzr version...12:39
vila... and we don't want to upgrade until that subunit/testools compat bug is fixed, I've lost track :-/12:39
jelmervila: I think the plan also was to look at tarmac12:40
vilahehe, reading mgz proposals on the active review page is... funny: fix-bzr-subprocess followed by don-use-bzr-subprocess-after-all :)12:41
vilajelmer: yeah, better12:41
vilawwooooow, lunch time12:42
mgzhey, I said it was a bug in two halves, I just fixed both of them :)12:46
crtxchi.  I cloned emacs.  I am not a developer.  I only want to make a copy, build emacs, then use the original to update any changes on the parent repo.  Did I do this right?  http://codepad.org/et7ETDzS12:48
mgzcrtxc: you don't really need any of that except `bzr branch` for your use12:51
mgzall I'd suggest is using the launchpad mirror rather than savannah may still be faster12:51
crtxcoh.  So I just do bzr branch everytime I update?12:52
mgzno, you branch once, then you just `bzr pull` to update12:52
crtxcoh.  I overcomplicated it.12:52
mgzright, but the extra bits you did aren't harmful12:53
crtxcok :)12:53
mgzthey're just for emacs developers' funny work flows.12:53
crtxcok.  I never did this before I only use bzr to track my text files.12:55
crtxcoops.  I just tried to bzr and I got an error.  http://codepad.org/0qXcOk7112:57
crtxctried to bzr pull^12:57
fullermdunbind it12:57
crtxcok.  ty.  unbind.12:57
crtxcwell.  Ty very, very much for your insights all.13:01
crtxcok.  That is it.  I should do some reading on shell scripting and c++ but my eyes are burning so ty very much13:14
fullermdA little C++ will fix that right up.13:14
fullermdThere's no burning left after you gouge them out.13:14
crtxc:)13:14
crtxcI am trying to learn shell scripting, python, and c++.13:15
=== gthorslund_ is now known as gthorslund
mgzlunchtime, prize if anyone can guess where I'm going with these mps13:45
vilalol, I was reading it and was about to ask :)13:46
vilaha, I read only one13:46
jmlwhat's the udd mailing list?13:47
jmlfor lp:udd in particular, I mean.13:47
vilaubuntu-distributed-devel@lists.ubuntu.com13:48
vilamgz: getting encoding right in more contexts ?13:49
fullermdmgz: To lunch.  What do I win?13:49
vilamgz: rats, lunch of course13:50
chromaticwtwhy learn c++ before lisp?14:38
=== beuno is now known as beuno-lunch
=== beuno-lunch is now known as beuno
vilamgz: I'm still holding my breath16:43
wgzit will be very exciting... maybe16:43
vilahehe16:43
fullermdI dunno, I think holding your breath that long is probably pretty boring.16:44
fullermdMaybe not for your heirs, I guess.16:44
vilashe looks worried indeed16:44
lamalexjml, is there a convenient way to call testtools.run from python on all of my test_ modules?16:45
jmllamalex: 'testtools.run discover [package]' iirc16:45
jmllamalex: http://testtools.readthedocs.org/en/latest/for-test-authors.html#running-your-tests will get it righter than me :)16:46
lamalexjml, thats from a shell though16:46
jmllamalex: what are you trying to do?16:47
jmllamalex: or, is testtools.run.main() not convenient enough for you?16:48
lamalexjml, nevermind that discover thing was exactly what i needed16:49
=== supton_ is now known as supton
lamalexjml, setUp runs for every test_ method, is there a TestCase wide set up?17:08
jmllamalex: no.17:08
jmllamalex: having such a thing is a bad idea17:08
lamalexwhy's that17:10
jmllamalex: because it makes test isolation very hard17:11
jmllamalex: which in turn makes it very easy to write tests where a failure or mistake in one corrupts the results of others17:11
lamalexjml, yah but we're not writing traditional unit tests17:12
jmllamalex: are you not interested in the results of these tests?17:12
vilawgz: encoding guessing... to cope with people using wrong fs encodings or not setting any ?17:36
vila(I had to breath .... ;)17:36
wgznot quite yet, but that one is on the list :)17:38
vilaoooh, at least properly reporting the full string on unicode exceptions ?17:39
wgzyup, and full unicode output for all errors.17:40
* vila whistles17:41
* vila hugs mgz too17:42
fullermdPah.  7 bits should be enough for anyone.17:43
vilanow, given that nexuiz-data needs more than 24h and that we kill imports that exceed their 24h quota... we need a finer-grained configuration ;)17:43
vilajelmer: pretty cute failure: http://babune.ladeuil.net:24842/job/selftest-chroot-precise/lastFailedBuild/testReport/junit/bzrlib.tests.blackbox.test_checkout/TestSmartServerCheckout/test_lightweight_checkout/17:50
vilajelmer: looks like you did even better than you thought ;-D17:50
jelmervila: argh17:51
jelmervila: we'll have to change the lower limit to 34 I guess17:52
vilajelmer: yup, not a problem17:52
jelmervila: there's a lot of variation in a single "bzr co --lightweight" command17:52
vilawhich is not ideal for a test but meh17:52
vilathe point is to not go up17:52
vilado you really need the 9 commits ?17:53
jelmervila: this is before I worked my magic btw ;-) the hpss-get-inventories branch brings it down to 1517:53
vilawoohoo17:54
vilajelmer, mgz : would you mind clearing the active reviews page guys, you have 7 approved reviews ready to land there :)17:55
wgzI'm trying to get everything ready all at once :D17:55
vilahehe, beware of conflicts when landing... the more you land before hand the less conflicts you bump into ;)17:56
vilaon top of that you have one for 2.4 that will need to be merged up ;-p17:57
=== Quintasan_ is now known as Quintasan
jelmerhmm19:07
jelmervila: I wonder if that RevisionNotPresent issue we've been hitting with the importer is related to the repository not clearing it's caching of negative revids in the parent map cache19:08
jelmervila: it looks like the "PoMerger created" message made it into bzr.dev, btw19:33
kirklandokay, a couple of bzr fast-import questions, from git20:30
kirklandI've done the fast-export from git, and then fast-import into bzr20:30
kirklandI now have a directory of a bunch of branches/tags, I think20:30
kirklandI guess I'm trying to figure out what/how to do with those20:33
jelmerhi Dustin20:37
jelmerkirkland: You should be able to just use those branches like you normally would20:38
jelmerkirkland: if they don't have a working tree yet, you can use "bzr co" to create one20:38
kirklandjelmer: oh, hmm, okay20:38
kirklandjelmer: let me try those20:38
kirklandjelmer: all of them looked "empty" except for .bzr20:38
kirklandjelmer: i guess i just needed to co them20:38
jelmerkirkland: we should have bzr-fastimport hint that the branches created are tree-less, just like most other bzr import tools do20:42
jelmerkirkland: this is done for space reasons, btw - you don't want to end up with 20 copies of the kernel if you're fastimporting a kernel repo with 20 branches20:43
vilajelmer: I left it there on purpose, I still have doubts about whether it's created only once and it's the cheapest way to gather data. Once I get this feedback, I'll address the issue one way or the other (remove the message or fix the duplicated creation)20:56
vilajelmer: hence my reply to your review: "It's not useless noise" ;)20:56
jelmervila: it should at least be hidden behind a debug flag in that case I think20:58
jelmervila: it's showing up for every revision processed by the udd importer20:58
vilait's one line per merge and will be useless if hidden behind a debug flag...20:58
Noldorinwhat's the easiest way to run Bzr server as a Windows Service?20:58
jelmervila: that's the case for lots of things20:59
jelmervila: doesn't mean we have them enabled by default :_)20:59
vilathen consider that I put a debug flag active by default and that this will be revisited21:00
jelmervila: sure, I'll file a bug21:00
vilaif you're really searching to reduce the size of the import logs, there are far bigger targets though21:01
jelmervila: sure, I'm just saying it's noise unless you're actually trying to debug something related to pomerge21:04
vilaI *am* :)21:04
vilaI mentioned that I've seen at least one case where the message where there twice for the same merge21:05
vilabut to avoid chasing ghosts for too long I submitted without digging further (nothing in my proposal should trigger such a behavior)21:06
vilaand there is one spurious merge in this sentence21:06
vilaghaa, where not merge21:06
vilaway past EOD ;)21:07
jelmervila: I still don't think that's a good reason, by that reasoning we might as well enable a whole bunch of things because they will help with debugging - like -Dhpss and -Devil21:08
jelmervila: if there was a -Dpo_merge flag, *you* could enable it21:08
jelmeranyway, it is indeed well past EOD :-)21:08

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