/srv/irclogs.ubuntu.com/2011/12/06/#bzr.txt

Noldorinjelmer, ah ok...00:09
Noldorinso absolute path00:09
Noldorinfair enough00:09
Pegasus_RPGHello. Quick question: if I bzr cp a file to another name, will future changes to the original be propagated to the copy?05:26
vilahi all !07:32
jelmerhi vila07:50
mgzmorning!08:01
jelmermgz!08:01
vilajelmer: regarding bzr-upload, as mentioned in the review, I want to use dedicated series, not maintain compatibility at all costs on trunk08:01
vilahey mgz08:01
vilastandup ?08:01
jelmervila: the plugin's metadata suggests it still supports 2.408:01
vilabecause we didn't bump the internal API for 2.5...08:02
vilaso apples and oranges08:02
vilaand as mentioned in NEWS, the code is still compatible with 2.4 so far, only the tests requires 2.508:02
vilabut one needs to start somewhere :)08:03
vilathe intent is to have a 1.1.0 when bzr-2.5.0 is released08:03
* jelmer looks for his headset..08:05
jelmergot it08:09
jelmervila, mgz: mumble?08:09
vilajelmer: there we are ;)08:09
vilajelmer, mgz : can you have a look at https://code.launchpad.net/~bzr/bzr/doc/+merge/84069 ? Given my implication I can't review it anymore ;)10:54
alansaulHey guys, how can I get previous versions of a file that I accidently deleted?10:54
alansaulI dont want to take everything back, just them, can i just do a revert x.rb10:55
jelmervila: sure10:55
mgz...it didn't seem from the comments I saw go past there were any issues with that mp vila10:55
mgzalansaul: yup.10:55
vilamgz: yup, no issues AFAICS, but voting on it myself seems a bit... bizarre10:55
vilaoh crap a > 2.000 lines proposal :-/ Hmm, at least the submitter signed with his own blood that it's mechanical ;)11:00
mgzwe can throw things at him if it breaks stuff :)11:00
vilaYeah, the most irritating part is to think that annotations will be broken one way or the other :--/11:02
jelmervila: +1ed11:02
vilajelmer: thanks11:02
mgzvila: on the annotations front, just think of it as jelmer accepting all blame from now on11:02
vilahehe11:03
mgzokay, made a teeny bit more work for the pilot, time for lunch13:04
jelmervila: when you have a chance, a review of https://code.launchpad.net/~jelmer/bzr/hpss-call-counts-3/+merge/84494 would be great - it fixes the hpss call count for lightweight checkouts which is causing spurious failures13:30
jam1jelmer, mgz: I'm getting a weird bug with latest bzr.dev14:00
jam1It is trying to take a lock: "~jameinel/u1db/index-transformations/.bzr/branchlock"14:01
jam1shouldn't that be: ~jameinel/u1db/index-transformations/.bzr/branch/lock14:01
jam1?14:01
jam1It looks like a bad trailing-slash normalization bu14:01
jam1bu14:01
jam1bug14:01
jelmerhey jam114:01
* jam1 goes to a speech therapist for stuttering :)14:01
jelmerjam1: looks like you're also hitting a bad trailing-'g' normalization bug :-P14:01
=== jam1 is now known as jam
jelmerit does seem like that could be related to the recent URL changes14:02
mgzwhat fun.14:02
jelmerjam: is this on Windows?14:02
jamcurrently, yes14:02
mgzjam: file a bug with the command to run, we can probably repo and fix.14:03
jamjelmer: http://paste.ubuntu.com/761635/14:03
jammgz: I'm just running 'bzr push' to launchpad :)14:03
mgzo_O14:04
jamk, weird, I also get it with bzr.dev -r-20014:04
mgzr6079 is one known breaker14:04
mgzif r6078 works and that one doesn't, means there's an extra regression the test suite didn't catch14:05
jamwell -200 is 614514:05
jammgz: well, I generally keep up to date with bzr.dev, so 6078 seems a bit to old, but it does indeed work and 6345 does not14:06
jamtoo14:06
jamnow I'm at "nothing to push", so I can't continue to validate it14:06
mgzquick, write more code! :)14:07
wgzokay, I don't get it just pushing to +junk so there must be something specific going on with the branch layout14:12
jammgz: https://bugs.launchpad.net/bzr/+bug/90076414:14
ubot5Ubuntu bug 900764 in Bazaar "failure to push to Launchpad 'branchlock'" [Critical,Confirmed]14:14
wgz...or with u1db14:14
jamLooks like I was using rev 6322 this morning, and I'm using 6345 when I noticed things were broken.14:14
jamwgz: you can't hide from me :)14:15
jameven weirder, I think I had successfully pushed a couple of times. Maybe it is a remote autopack issue?14:15
jambut that shouldn't trigger a bug locking the branch14:15
wgzoo... ooooo14:15
wgzrepacking would be extra fun.14:15
jamwell the failure is early on in bzrlib.push._show_push_branch14:16
jamduring the first remote write lock call on the Branch.14:16
jamwell, something is working now...14:19
wgzhow do we get logs from the launchpad side?14:19
jamI wonder if it was a display bug, coupled with a temporary Launchpad failure.14:19
wgzhm.14:20
wgzthat's an interesting thought, and testable.14:20
jammgz/wgz: Interestingly, I have 4 emails in my inbox from Launchpad talking about failures to generate merge proposals.14:22
jamlooks like something with the librarian there14:23
jelmeryeah, the librarian was down for a while becuase of hardware problems14:28
wgzI got a 503 from the librarian just before you posted.14:28
jelmerit's not involved in code hosting though, as far as I know14:29
wgzcoming back and looking at the bug report again, I don't think it can be a bzr.dev regression14:42
wgzthe LockDir url is constructed on the launchpad side and just being reported to the client14:43
vilajelmer: done15:00
jelmervila: thanks15:03
mgzvila: for some reason, I'm not getting your mail sent to the bzr list15:06
mgzcan read it in the html list archive, but neither of your last two "week in" emails got to my inbox15:07
vilao_O15:07
vilamail log confirms the one for this morning was sent15:08
vilanot received but I'm pretty sure that's how I configured my subscription15:09
vila(i.e. don't send me my own mails, thanks)15:09
mgzit got as far as the list, but didn't then get back out to me15:13
mgzand your last week's one had the same issue, I only saw it existed because I got the reply-to-the-reply15:14
vilabah, vila, learn to read, he said the mail went to the list, not a problem on your side15:14
vilamgz: that leaves: 1) an issue at your provider (aggressive spam filtering), 2) a wrong filtering on your side ?15:15
vilamgz: they don't like French maybe ? ;-p15:15
mgzit's canonical then google, and didn't get spamified15:16
vilaoh, on google ?15:16
mgzI was wondering if there was anything else odd about either message that might cause it to get dropped15:16
vilasee ? They know far too much already :)15:16
vilahmm, I can't imagine what can be different, let me check again15:16
vilamgz: here is my local copy: http://paste.ubuntu.com/761701/15:18
vilamgz: the only lines that you probably won't find in your copy are the 'X-Draft-From' and 'Xref' ones which are local15:19
vilayou should have more Received: lines though15:19
vilain other news, consistent oops on https://code.launchpad.net/~gz/bzr/mv_removed_non_ascii_898541/+merge/8460915:20
vilasome one is trying to get between us, mgz, what a nefarious (but hopeless) conspiracy !15:21
vilathere is a suspicious "Warning: <type 'exceptions.KeyError'>: 'branch'" in the oops page...15:23
vilajust after a: "Warning: Macro expansion failed"15:23
vilaany recent bzr deployment on lp ?15:23
mgzdon't think so, but given jam's earlier experience something is up with the launchpad codehosting15:25
* vila nods15:25
vilathe librarian failure seem to have triggered a bunch of import failures.... and the importer caught them as lp down times and happily retried them, ending up with 0 additional failures \o/15:28
vilait did flip-flop a lot (back/down/back/down) but still managed to retry all the right ones15:29
mgzwhat's one of your oops # vila?15:30
vila(Error ID: OOPS-cbc7444249203b34468eab8a253cf7f1)15:30
mgz...actually it's so consistent I can also get my own on that mp15:31
vilathere are several urls with %7Egz (I thought we settled on ~ being legal and as such never quoted anymore, may be not landed on lp ??)15:32
mgz...the oops page is really not very browsable15:35
vilayup :)15:36
jelmervila: yes, that's not on LP yet15:36
vilawith 'that' being 'stop quoting ~' ?15:37
mgzalright, is librarian related, forgot they put the diffs there15:37
vilayeah, librarian related but librarian is back. Did you propose while it was down ?15:38
mgz(what jam was saying earlier about getting diff generation failure email should have twigged earlier)15:38
vilas//submit/15:38
mgzit lost some files apparently, they need reuploading15:38
mgzanyway, they're on it, so won't bug 'em further15:39
vilajelmer: wow, that was confusing (the test rewrites), I got it now15:52
mgzvila: oh conflicts expert15:52
mgzhow in the hell you you get to the state:15:53
vila2.4.2 is still in -proposed ?15:54
mgzadded: config.py / conflicts: Conflict adding file config.py. Moved existing file to config.py.moved.15:54
vilamgz: different file-ids for the same file15:54
mgzI can only get to a added: removed: conflicts: state.15:54
mgznot a modified one.15:54
mgzwithout the removed.15:55
vilawhat branches do you encounter that in ?15:55
mgzthe repo of a bug in qbzr.15:55
mgzI have the guy's branch state, but not clear instructions on how he got there, so I'm not sure how to write the test.15:56
vilabzrlib.tests.test_conflicts15:56
vilaha, crap, you don't like the way these tests are written15:56
vilaI had an idea about how to rewrite them but I've yet to do it ;)15:57
vilaat least you'll get the bits you need15:57
vilaalso, you can have a look at po_merger tests which are script-based and easier to grasp,15:57
vilabasically you create a branch with all the stuff you need and then branch from the "right" revisions15:58
mgzanyway, just creating two branches with the same file and merging doesn't do it.15:58
mgzas in, diverging from the same root15:58
vilamwhahaha, I want some of either 1) what you're smoking, 2) the tweaked bzr addressing the parallel import issues you're using ;)15:59
vilaif merge doesn't conflict merging the same path with two different file-ids it's probably because both paths use the same file-id :)15:59
mgzit conflicts alright, but not the right way16:00
vilatext conflict ?16:00
mgzI get added/removed, the example branch just has modified16:00
mgzbut with the conflict saying .moved has been created16:00
mgzthe diff is just a line in the file16:00
mgzmy diff is two files, one added, one removed16:01
vilameh, I'm lost, can you paste your test ?16:01
mgz...I should look at the history in qbzr16:01
mgzo_O there is only one revision16:02
wgzvila: zip attachment in bug 81582216:03
ubot5Launchpad bug 815822 in QBzr "AttributeError: 'NoneType' object has no attribute 'is_ignored'" [Undecided,Incomplete] https://launchpad.net/bugs/81582216:03
vilajelmer: looking at https://code.launchpad.net/~bzr-core/ubuntu/oneiric/bzr/sru-2.4.2/+merge/81039 , shouldn't some sru team be subscribed there ?16:03
wgzedit testrepo-fb-checkout/.bzr/branch/location to point to the right abspath of where you unpack it, then inspect testrepo-fb-checkout16:04
wgzokay, I think I have an idea16:06
vilagrrr, of course I read that after doing it :)16:06
kirklandjelmer: hey there!  back to my line of bzr/git import questions16:06
vilawgrant: obviously the .moved file is not there anymore16:07
kirklandjelmer: so i did my 'bzr up' in each of a dozen or so of the dirs that I imported from git into bzr16:07
vilawgrant: sry, bad completion16:07
vilawgz: obviously the .moved file is not there anymore16:07
kirklandjelmer: now, it is customary to push each of those to its own branch in LP?16:07
kirklandjelmer: for history/tracking purposes?16:07
mgzvila: it's just an idea, I still want the expert opinion :)16:07
kirklandjelmer: or how do others handle that?16:07
mgzI think the way I was doing it was wrong, I have two child branches both create a file then merge one to the other16:08
jelmerhi kirkland16:08
kirklandjelmer: howdy :-)16:08
mgzperhaps what's needed is a child branch to create a new file, move it over the file in the parent branch, then merge child to parent16:08
kirklandjelmer: so there are a handful of tags (v1.5, v1.4.3, v1.7, etc)16:08
jelmerkirkland: tags will be stored in the branch as well ("bzr tags")16:09
vilamgz: or create a new file with the same path in branch A and B, then merge ../B16:09
jelmerkirkland: but branches from git are usually pushed as separate branches onto Launchpad16:09
kirklandjelmer: as I described?16:09
mgzvila: that gives me the conflict I had, which is added f, removed f16:09
vilamgz: but from a quick look, the stuff in error.zip is incomplete16:10
mgzrather than modified f16:10
jelmervila: see poolie's comment on that MP16:10
jelmerkirkland: it depends on your situation a bit - most people register launchpad imports to track branches in git16:10
jelmerkirkland: what's the situation in which you're using bzr-git?16:11
kirklandjelmer: I'm helping my new company migrate from git to bzr/Launchpad16:11
vilajelmer: I followed the comment to bug #848064 and there he says (and I pretty much agree) that this is not specific to bzr (or at least we don't have evidence so far)16:11
ubot5Launchpad bug 848064 in Ubuntu Distributed Development "Revision not present branching from udd-imported branches on lp" [Critical,Confirmed] https://launchpad.net/bugs/84806416:11
kirklandjelmer: they have one project (private/proprietary in Launchpad)16:12
jelmerkirkland: ah, I see - in that case what you described indeed makes the most sense - push each separate branch import from git as a separate branch on Launchpad16:12
kirklandjelmer: great16:12
jelmerkirkland: the tags should come along16:12
kirklandjelmer: okay, I'll have a look16:12
vilamgz: ctrl-alt-del, looks like I misunderstood16:12
jelmerkirkland: (and new versions of bzr-git will in fact no longer create separate directories for tags)16:12
kirklandjelmer: oh, interesting16:13
kirklandjelmer: okay, thanks for your help, cheers!16:14
vilamgz: .moved is the THIS file (confusing)16:14
vilamgz: therefore, the added one should be the OTHER file16:15
vilamgz: bbiab16:16
jelmervila: argh, Option didn't have a help argument in 2.4 :-(16:38
=== deryck is now known as deryck[lunch]
mgzha! got the repo16:57
mgz...that was the most convoluted thing ever16:57
vilajelmer: and you encounter that in what context ?16:59
vilamgz: so, fresh from the bug, he did delete .moved file and AFAICS we don't have the branch he was merging from17:02
vilamgz: additionally, his dirstate file seem to miss the revid for the merge too :-/17:02
vilamgz: on the other hand, he provided the .zip to reproduce a bug which is not related to conflicts.17:03
vilamgz: in a nutshell, the branch he provided may not be one you can re-create17:04
mgzvila: I don't have the exact dirstate he got, the 'modifed' thing is mystifying as the bug needs no file-id to repo17:07
mgzbut I have a series of steps that gets that problem17:07
vilamgz: with bzr-2.3.4 ?17:08
mgzwith current bzr, the logic error in qbzr is reasonably clear17:08
mgzit's just less obvious how to actually trigger it17:08
mgzI'll wave the test case in front of you17:09
mgz(shortly)17:09
vilak17:09
jelmervila: current bzr-svn trunk is broken with bzr 2.417:13
vilaha, because of the new config support ?17:15
vilajelmer: I thought I mention the compatibility part not being addressed when I first submitted but then thought you had different series targeted at 2.4 and 2.5 and finally, I forgot about it :-/17:15
jelmervila: np, it's not too hard to work around17:16
vilajelmer: so, more than help should cause problems no ?17:17
jelmervila: some other minor regressions, fixing them now17:20
vilajelmer: what kind ?17:20
jelmervila: not related to config, but apart from that bzr-svn trunk is 2.4-compatible17:21
vilabut not 2.3 ? Where do you draw the line ?17:21
vilajelmer: stable + trunk ?17:22
jelmervila: 2.4 and bzr.dev at the moment17:22
vilak17:22
jelmervila: it varies, I dropped 2.3 support a month or so ago17:22
jelmervila: I usually aim for the latest stable series and bzr.dev, but it depends on how hard it is to stay compatible17:23
vilanow that you mention it, you probably want to add a .id attribute in your store17:24
jelmervila: anyway, I just noticed it as I was trying to actually switch over some code to use the new options stuff17:24
vilaright, it's opt-in at the option level17:25
vilabut even the new options can still be read/modified with the old implementation17:25
jelmervila: right, but how well will 2.4 work?17:27
vilabzr-2.4 use the new config implementation for only one (or a few) options17:28
vilaso it should work at it works before :)17:29
vilaworked17:29
vilaas far as bzr-svn is concerned, get_config_stack will not be called and since no branch option use get_config_stack in bzr-2.4 nothing has changed17:29
vilajelmer: unless you encounter a scenario I can't imagine ?17:30
jelmervila: I'm wondering whether I should start using get_config_stack in bzr-svn, or if that will affect compatibility with bzr 2.417:38
vilathat should even break with bzr-2.4, there is no get_config_stack there17:39
jelmervila: there is on SvnBranch :-)17:39
vilaooooh, that17:40
vilahmm17:40
vilathat's a tricky question :)17:40
* vila looks at bzr diff -rbzr-2.4.2.. bzrlib/config.py17:41
vilajelmer: registering options looks suspicious17:43
vilano cmdline overrides17:43
jelmervila: that's actually something that seems to work17:43
vilak17:43
jelmervila: did cmdline overrides work at all in 2.4?17:44
vilait didn't exist17:44
vilajust listing17:44
vilaha, remote (smart) branches broken, no impact or bzr-svn ?17:44
jelmervila: nope - remote branches can't access svn stuff anyway, not even in bzr.dev17:45
vilajelmer: and no expansion either, so overall I saw nothing likely to fail, the basic pieces were in place and if some internals have changed, they should already work in 2.417:46
jelmervila: cool17:46
vilaand given the current read/write scheme using both implementations should be ok17:47
vilaonce this model change (so bzr.dev only) it will become dangerous to access one option with both implementations17:47
vilabut even there, that will require modifying the value, so unlikely to be encountered soon (except for GUIs may be)17:48
=== deryck[lunch] is now known as deryck
mgzokay, have a test18:00
mgzdon't really understand all this monkeying but hey18:00
mgzvila: do you have a sec to look at a diff and tell me ways to make it saner?18:07
vilamgz: shoot18:07
mgzvila: http://pastebin.ubuntu.com/761875/18:10
vilathe 'remove/add/rename_one' dance you mean ?18:12
mgzyeah, and any of the other branch creaty bits.18:13
vilawell, the branch creation bits are not that bad (unless you can create a richer branch in make_branch_and_tree)18:13
mgzI guess remove/os.rename/add would work18:13
vilayup, was about to ask/suggest about using os. everywhere in tree nowhere18:14
vilaas I understand it, error.zip was obtained after shell commands, with no bzr commands involved18:14
vilaso qbzr may be confused because some files referenced in the dirstate are not there anymore and some other have totally different content18:15
mgzyeah, but I can't get into that state, though I can do this similar one.18:15
vilaoh, misread, make_branch_and_tree is the basic bzr one18:17
vilaso po_merger tests use branch_builder and scripts which can be more readable, but your tree setup being 10 lines long... that would be overkill18:18
vilayou can also add assertLength(1, tree.conflicts()) to complement self.assertPathExists("parent/f.moved")18:22
vilaEOD'ing18:23
mgzthanks vila.18:28
vilamgz: rhaa, my comments were confusing. One thing you should think about is that the bug may be caused by some difference between the dirstate content and what qbzr thinks about the wt. As it's written, your test may not reflect that as it uses an in-memory tree which *may* not trigger the same behavior18:31
vilatest scripts running complete commands are less likely to suffer from that which is why I mentioned them as opposed to the bzr-upload ones18:32
vilathere, that's should be clearer ;)18:32
vilahmm, almost, by in-memory, I mean a wt with its internal state as opposed to loaded from the dirstate18:35
lamal666lifeless, is pyjunitxml already integrated in testtools somehow?18:50
=== lamal666 is now known as lamalex
jelmerlamalex: it's a separate package18:55
lamalexi know, but im wondering if there's a simple integration path that i haven't figured out18:56
lifelesslamalex: in what sense integrated?19:02
lifelessif you mean 'report tests using it',most folk I know have been setting that up in their local test glue, or if they have non via subunit19:03
lifelessthere is a merge proposal just waiting copyright signoff to add a junitxml.main as a convenience for folk too19:04
lamalexah, i guess we dont really have any test glue we've just been doing python -m testtools.run discover19:04
lamalexwhich is probably why im having a hard time figuring out how to fit the two together :P19:04
lifelessright now the easiest thing will be19:04
lifelesspython -m subunit.run discover | subunit2junitxml -o tests.xml19:05
lifeless(I think, thats from memory - use -h and season to taste)19:05
lamalexyah19:05
lamalexok, thanks lifeless19:05
lamalexi honestly would have never figured that out19:05
lifelessheh19:07
lamalexlifeless, does testtools output subunit by default/19:32
lifelessno, subunit.run does (which is testtools compatible)19:32
lamalexahh19:35
lamalexwoop dere it is19:36
pooliehi all21:23
wgzhey poolie21:43
pooliehi there wgz, how's it going?21:43
wgzgood, a little bit of an interesting launchpad day. how was your weekend?21:46
pooliereally nice21:55

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