/srv/irclogs.ubuntu.com/2010/02/28/#bzr.txt

Demostheneswow, thats *REALLY* odd. i killed off 4L-gui and a few defunct 4L-cli processes, and bzr works now.00:08
DemosthenesWTF00:08
Demosthenesi don't see any differences in /usr, /lib, or /bin that would indicate i was hacked. the only diffs are .gz files that are package documentation, and they pass gzip -t00:12
Demostheneshas a later bzr been backported to lenny?00:12
ed-209how come whenever I anonymously access launchpad, bzr works fine, but when I try to set up an account with SSH keys and all, it jus fails out?00:13
Demosthenesed-209: ssh -vvv and watch00:13
spivlifeless: I'd welcome patches for news_merge (of course)01:38
spivlifeless: it's pretty hackish in various ways01:38
spivlifeless: Partly because of the stupidly simple parser01:39
spivlifeless: partly because it uses Merge3.merge_regions, but it expects lines and I want to feed it other kinds of blocks of text01:40
spivlifeless: or rather, it doesn't really want to feed Merge3 lines at all, but ideally two-tuples, e.g. [('section', 'Bug Fixes'), ('bullet', '* Fix crash in frobber.'), ...]01:42
lifelessspiv: Activation energy is still too low; I may patch it in future05:24
lifelessspiv: but if you can patch it in work time, we both win :)05:24
lifeless\o/ test suite loop closed.07:05
thumperlifeless: you around?08:52
lifelessthumper: hi, yeah, just making testr rock08:52
thumperlifeless: I'm starting a new python project08:52
thumperlifeless: and I want some test infrastructure08:52
thumperlifeless: what do you recommend?08:53
thumperlifeless: I have nothing right now08:53
lifelesstestrepository08:53
lifelessand testtools08:53
thumperlifeless: automatic discovery if possible08:53
thumperis it packaged?08:53
lifelessuhm, if you want that you can use the discover module08:53
lifeless[I don't really like automatic discovery, but thats me :P]08:53
thumperI can register I suppose08:53
thumperI want a simple way to just run the tests...08:54
lifelesscan I show you something08:54
thumpersure08:54
lifelessbranch lp:testrepository08:54
lifelessinstall python-testresources and python-testscenarios and python-testtools08:54
thumperpython-testscenarios not found08:56
lifelessyou are running lucid, yea?08:56
thumperno08:56
lifelessah08:56
lifelessadd the subunit PPA08:56
thumperwhich is?08:56
lifelesssudo add-apt-repository ppa:subunit/ppa08:57
thumperlifeless: got muffins in the oven08:59
thumperlifeless: got that installed08:59
thumperbut the oven is calling08:59
lifelessok08:59
lifelessI'll be here08:59
lifelessthumper: oh, also you might like lp:commitfromnews; its verra useful :)09:00
thumperlifeless: not a branch09:05
lifelesssorry09:05
lifelessbzr-commitfromnews09:06
thumperwhich does?09:06
lifelessdiffs NEWS to suggest commit messages09:06
lifelessoh, you'll also need python-subunit installed09:07
lifelesssorry, you're getting a full stack here ;)09:07
thumperit better be worth it09:07
lifelessI think you'll like it09:08
lifelessI've just pushed rev 88 of testrepository, probably worth having09:08
lifelessanyhow, there are two main things here09:09
lifelessI want to show you the layout I used in testrepository, as I'm pretty happy with that09:09
lifelessand I want to show you testrepository itself09:09
lifelessas I think its an excellent workflow09:10
lifelessbrb in a couple of minutes, organising dinnery things.09:10
lifelesshave a poke aroun testrepository's code base; try 'make check' etc09:10
lifelessthumper: ok, back09:17
phixhey09:22
phixwhat's new?09:22
lifelesshi09:22
lifelessstuff :P09:22
phixnice09:22
phixummm, I am suprised this channel is still here09:23
lifelesswhy?09:23
phixWell, git has been pretty stable for the last few years now09:23
lifelessso? Thats git, this is bzr.09:23
lifelessWe're gaining users faster than ever09:24
fullermdRCS has been rock-stable for decades   :p09:24
phixyeah but git kind of makes bzr look like cvs09:24
lifelessphix: I think that comparison is false09:24
thumperphix: no one likes trolls09:24
phixok ok, I will give you some credit, more like svn :P09:24
phixhehe09:24
phixthumper: aawwww09:24
phixI m not trolling though09:25
lifelessphix: Is that a considered opinion, or a joke?09:25
phixno no no, bzr is the joke actually :)09:25
phixwell no, the fact that ppl still use it is I suppose09:25
phixany way09:25
phixI am off :)09:25
phixsee you guys in #git soon for sure09:25
lifelessphix: thats either hostile or a troll, and neither is particularly friendly or encouraging.09:26
phixoh09:26
phixrevert phix09:26
fullermdOh, sweet.  Now I have the perfect answer to people who tell me I'm not funny!09:26
phix:P09:26
lifelessfullermd: I think you mean reset --hard -q --patch ./09:27
fullermdI...  do?09:28
lifelessfullermd: I picked a classic command new git users get confused by09:28
fullermdWell, it apparently works, 'cuz I'm confused why I'd mean it   :p09:32
lifelessfullermd: its 'revert' spelt git style09:34
fullermdOh.09:36
fullermdI didn't say 'revert'.09:36
lifelessfullermd: what was your perfect answer then ?09:53
fullermdWell, that I use bzr, so obviously I'm a joke.  Presumably a good one, naturally.09:55
lifelessah09:55
lifelessunless its dark humour09:55
lifelessuhm09:57
lifelessthis is a wtf moment:09:57
lifelessmake docs09:57
lifelesspython -c "import shutil; shutil.copyfile('NEWS', 'doc/en/release-notes/NEWS.txt')"09:57
lifelessI think if we have Make, we can surely assume 'cp' ?09:57
fullermdThat should be implemented with inline ASM, to be more optimized.10:03
lifelessyou're here all week right? So you might improve.. :)10:06
fullermdDon't forget to tip your waitress!10:06
lifeless<3 commitfromnews10:08
bialixhi igc10:17
Kamping_Kaiseris there a bzr plugin to mark bugs in debbugs pending on commit to a bzr repo? (I can't imagine i'm the only person to wonder this, but i've not seen it asked before)10:17
lifelessbzr builddeb would be a good place to put that10:17
lifelessits really 'bzr-debian' :P10:18
Kamping_Kaiserok , I'll check to see what it does :)10:18
Kamping_Kaiserhehe10:18
igchi bialix10:25
bialixMarco Polo is good codename, the first great explorer, right?10:25
* bialix uploading installer now10:25
bialixigc: installer done10:27
igcbialix: certainly one of the better known early explorers10:28
igcbialix: thanks for the installer too10:29
bialix:-)10:29
igcbialix: http://en.wikipedia.org/wiki/Exploration10:29
bialixigc: I have an theory that every "project" in human life takes about 9 months to get somewhere. your bzr-explorer is another confirmation ;-)10:30
lifelessheh10:30
lifelessI'm only just getting to where I want testing to be, after 14 years10:30
igcbialix: I just hope it's stable enough to warrant 1.010:31
bialixlifeless: I don't say about maturity10:31
lifeless:P10:31
fullermdOgod, I hope this project will be done in less time than that   :(10:31
bialixthe question is what this for you10:32
* bialix waves hi to all10:32
lifelessigc: grats on 1.0.011:20
dsuchHi, I'm using bzrlib in a piece of software of mine and I'm wondering if it's safe to keep a reference to a WorkingTree instance in multiple processes?12:23
dsuchAm I not going to get hit by some weird race condition sooner or later? (Just wondering, not that I have seen anything like that so far)12:23
dsuchI'm currently using 2.1.012:24
dsuchOr, alternatively, can those subprocesses have their own separate WorkingTrees pointing to the same repository location? I guess that should be safe?12:28
lifelessdsuch: do you mean 'are WorkingTree methods correctly locked' ?12:38
lifelessdsuch: and the answer is 'we think so' ;)12:38
dsuchyea, guess that was the question :)12:39
dsuchthanks12:39
dsuchlifeless: and if I don't share those objects, there sure will be no problems at all, right?12:41
lifelesswhat do you mean by share12:41
dsuchshare as in multiple threads have the same WorkingTree instance assigned to them12:42
dsuchbut it's not that I really want it12:43
dsuchin fact, I could as well have each thread/subprocess use its own WorkingTree object so there would be no sharing of a WorkingTree instance's state at all12:43
lifelessso12:44
dsuchand that, if I understand it correctly, should be perfectly safe?12:44
lifelesson unix, os locks are not exclusive across threads12:44
lifelessyou can stomp on yourself trivially.12:44
lifeless*processes* are safe.12:44
lifeless*threads* are not.12:45
dsuchmhm12:45
lifelesseven if you have separate WorkingTree objects, in different threads, you could zerg your dirstate file12:46
dsuchgot it12:46
lifelessdirstate2, *when* we get to it, will not use OS locks at all, and won't have this risk.12:46
dsuchI fear to ask what a dirstate is :) I take it it's some important internal repo structure?12:47
lifelessyes12:47
dsuchok12:47
lifeless.bzr/checkout/dirstate12:47
lifelessit lists all the files in the tree12:47
lifelessand their inode fingerprint12:48
lifelessmerge status12:48
dsuchokay so it is important :)12:48
lifelesssemantic changes to this file are locked with a dir-lock, which is safe even in threads.12:48
lifelesshowever, readonly locks can do cache-updates to this file12:48
lifelessand they are only protected by an OS lock12:48
lifelesswhich as I mentioned is not actually as safe as one might hope12:49
dsuchI see.12:49
lifelessgnight12:53
lifelessif you have more questions, just ask, I'm sure someone will answer eventually :)12:53
dsuchsure, thanks again lifeless12:53
dsuchhope you get better ;-)12:54
dsuch(sorry, couldn't resist)12:54
felixhummel_hi! how would you put files under version control that belong to one project, but live in different directories scattered across the whole file system?13:30
felixhummel_e.g. all hand-edited configuration files of a ubuntu box13:31
jelmerfelixhummel_: Generally the configuration files of a ubuntu box would be in /etc13:32
jelmerfelixhummel_: other than that, you should be able to use symlinks to a versioned directory13:32
felixhummel_jelmer: so you would do ``cd /etc; bzr init .; bzr add hosts``?13:34
jelmerfelixhummel_: basically, yeah13:34
jelmerfelixhummel_: I'm using the etckeeper app to maintain my /etc13:34
jelmerit wraps around bzr and takes care of e.g. invoking bzr commit when you upgrade or install packages13:35
felixhummel_looks like etckeeper fits my use case perfectly. thanks, jelmer!13:36
jelmerfelixhummel_: you're welcome13:36
jelmermaxb: the bzr-rewrite trunk should be fixed now14:42
ajeanshello15:56
ajeansI have a question about "./bzr selftest", and yahoo couldn"t find the answer for me15:57
ajeans:)15:57
ajeans".bzr selftest" runs all the bzr unit tests (> 25000), what is the command to run one15:58
ajeans(sorry) only one of the unit tests (e.g. tests contained in bzrlib/tests/test_osutils.py) ?15:59
fullermdajeans: Either giving it a substring as an argument, or using -s and a class prefix.16:20
ajeans@fullermd: From the root, I am trying "./bzr selftest -s TestStatus" and also tried "./bzr selftest bzrlib/tests/test_status.py", with no success16:29
ajeans@fullermd: Thanks for the information, you can pass the method name to only run it16:35
ajeans./bzr selftest --help had some information about it.16:35
ajeansI was hoping to simply pass the Class name to execute all the tests within, but I can still list all the methods manually... thanks again16:36
=== khmarbaise_ is now known as khmarbaise
fullermdajeans: You can use -s to pass the class name, it just has to be fully qualified.  e.g., "-s bzrlib.tests.test_foo.TestFoo....."16:52
fullermdOr 'selftest test_foo'.  The former is faster because it doesn't have to build the full list of all the tests then search through it.16:53
mgedminlong-term wishlist: automatic mirroring of my bzr branch16:57
mgedminsort of like bzr bind or manual bzr push after each commit, but without having to wait for the push16:57
mgedminlet the push happen in the background, opportunistically16:58
=== khmarbaise_ is now known as khmarbaise
ubuntujenkinsif I do bzr merge lp:~ubuntu-manual/ubuntu-manual-screenshots/bg -r 0..-1 does that update the status of the branch in lanuchpad?17:14
jelmerubuntujenkins: it might mark a merge proposal as "merged" if there is a merge proposal for that merge17:16
jelmerotherwise it won't affect the status of the branch17:16
ubuntujenkinsweird I the status has changed from devolpment to merged. I don't know why its anoying as I have to change it back on 47 branches now17:17
ubuntujenkinsthanks for your help jelmer17:23
ajeans@fullermd: thanks for the explanation, it works perfectly :)17:49
maxbjelmer: Excellent! But, now there's a 'trunk', a 'rebase-trunk', and a 'trunk-mirrorred' ... it's a bit confusing17:55
jelmermaxb: trunk is the right one17:55
jelmerit's also registered as the development focus17:55
maxbSure.. but why the others?17:56
jelmerhistorical reasons17:57
maxboh. stacked branches prevent deletion?17:57
maxbperhaps set them as 'Abandoned' if they are no longer relevant?17:58
jelmermaxb: feel free to do so17:59
maxbjelmer: I'm pretty sure only the branch owner can18:00
jelmermaxb: the branch status isn't really useful at the moment imho, and I have too much branches to set all of their statuses appropriately18:00
maxbI find the branch status quite useful - it would be nice to at least have ones with 'trunk' in their name which are no longer relevant hidden from the project's default branch listing18:02
jelmermaxb: isn't that what the development focus is for though?18:02
lifelessmoin18:04
jelmermaxb: I've changed the status for these two branches to abandoned, although I don't really see the problem tbh18:05
jelmerhey lifeless18:05
jelmeroh, looks like I could remove them after all18:06
jelmerlifeless: if you could give some quick feedback on my named-branches patch, that'd still be much appreciated :-)18:24
lifelessjelmer: ah yes, I knew I forgot something this weekend18:28
lifelessjelmer: shall do18:28
=== radoe_ is now known as radoe
jelmermaxb, if your merge request is still relevant, can you please resubmit it against lp:bzr-rewrite?19:45
mgedmintime bzr rocks on a Nokia N900 --> 1.72 seconds wall-clock time20:50
mgedminseems a bit on the high-side20:50
mgedminit's a 600 MHz ARM CPU20:51
lifelessif you run it again?20:53
mgedminlifeless, varies between 2.32s and 1.38s (I suspect cpufreq)21:19
lifelesswow21:19
mgedminthis is 2.0.2, I should package a newer version...21:19
jonohey all22:24
jonoanyone here used the bzrlib python module?22:25
lifelessyes22:32
lifelessbah22:33
thumperlifeless: hey22:38
lifelesshi22:38
thumperlifeless: when we did stacking we decided that the server should hit to the bzr client to stack22:38
thumperlifeless: do you know if the --no-stacked (or similar) option can or does ignore the suggestion?22:38
lifelessit should, it doesn't22:39
thumpers/hit/hint/22:39
thumperlifeless: is there a bug?22:39
lifelessI'm sure of it22:39
slangasekis anyone here using pristine-tar for importing of upstream tarballs outside of LP?22:55
lifelessaie aie22:56
slangaseklifeless: I was trying to work off http://www.advogato.org/person/robertc/diary/130.html to do bzrification and debianization of a new source package, but import-dsc doesn't seem to have usefully imported the autotoolage22:56
slangasekis there an undocumented step I'm missing?22:57
slangasekrather, it imported it as part of a single commit with the Debian stuff22:57
lifelesslet me just see what post that is22:57
lifelessrbtcollins.wordpress.com has better urls22:57
lifelessah yes22:58
slangasekwell, rbtcollins.wordpress.com is not what you have linked from Planet Debian :)22:58
lifelessmeh :P22:58
lifelessok, so import-dsc does two imports22:58
lifelessa tarball, accessible from 'bzr tags'22:58
lifelessand a debianised22:58
lifelesslook at 'bzr log -n0' or at 'bzr tags'22:59
slangasekneither shows me what I'm looking for22:59
lifelessthe upstream-XYZ tag should be precisely the tarball contents22:59
slangasekdoes it matter that this is 3.0 (quilt) ?22:59
lifelessprobably, AFAICT quilt-3.0 is a <censored>-<censored>22:59
lifelessuhm23:00
slangasekthere's no 3.0 (bzr) format23:00
slangasekand 3.0 (native) is n/a23:00
slangasekand I have an upstream tar.bz223:00
lifelessslangasek: pastebin bzr tags and bzr log -n0 somewhere23:00
slangasekhttp://paste.ubuntu.com/385962/, http://paste.ubuntu.com/385963/23:02
lifelessok, no upstream- tag23:03
slangasekI'll mention that when following the directions as written, import-dsc first gave me an error about 'bzr add'ing debian/changelog first23:03
slangasekand rather than doing this, I moved my debian/ dir aside23:03
slangasekso if I'm meant to actually do the 'bzr add', I can go back and try that23:04
lifelessdo you have the session in history, so I can see it?23:04
slangasekI can uncommit and do it again? :)23:04
lifelessno23:04
lifelessyou can start from a fresh branch though23:05
slangasekok23:05
lifeless*don't* branch from this branch23:05
slangasekyep23:05
slangaseklifeless: http://paste.ubuntu.com/385965/23:06
lifelessslangasek: and bzr+ssh://bzr.debian.org/bzr/pkg-samba/cifs-utils/trunk is an *upstream* branch - no packaging in it at all ?23:07
lifelessslangasek: the debian dir shouldn't be present: the revert step in my instructions is to remove the packaging (because it said debianise as normal :P)23:08
slangasekcorrect23:08
igcmorning23:08
slangasekbut 'bzr revert' doesn't delete new files23:08
lifelessslangasek: heh23:09
lifelessslangasek: anyhow, yes, you are right to not have the debian dir present23:09
slangasekso I inferred, when I got the error, that this is what you wanted, so I moved debian/ away :)23:09
lifelessok23:09
lifelessand what happens then23:09
slangasekhttp://paste.ubuntu.com/385967/23:11
slangaseksingle commit, no upstream tag23:11
lifelesscan you paste the dsc23:11
slangasek(this is with bzr-builddeb 22, btw; Debian unstable seems to be a version behind Ubuntu)23:11
lifelessand check there is only one dsc in the dir above23:12
lifelessslangasek: generally you will want bzr-builddeb trunk :P23:12
slangasekhttp://paste.ubuntu.com/385968/23:12
slangasekno, generally *I* want to be dogfooding what's actually in the distro ;)23:12
slangasekyes, there's only one .dsc23:12
lifelessthis is very strange23:13
lifelessI'd like you to try trunk.23:13
lifelessif it doesn't work, file a bug.23:13
slangasekok - url?23:13
lifelessif it does, file a different bug23:13
lifelessbzr branch lp:bzr-builddeb ~/.bazaar/plugins/builddeb23:14
slangaseklifeless: yep, works with trunk23:19
lifelessslangasek: so; you want a newer package ;)23:19
slangasek(another wrinkle: had an out-of-date version of bzr in my Debian env, updated from bzr 2.0 to 2.1 to get builddeb trunk to work; and I can't seem to use the old builddeb in unstable with new bzr...)23:19
lifelessjml: are you online?23:19
slangaseklooks like bzr-builddeb 2.3 is in experimental, I'll double-check with that also23:20
slangasek2.3 also works correctly23:21
slangasekso someone needs to put 2.3 in unstable23:22
slangasekjelmer: ^^ :)23:22
jono_mwhudson, around?23:26
mwhudsonjono_: just about to have lunch, but can hang around a bit ... how can i help?23:30
jono_mwhudson, I am just starting hacking with bzrlib - is there a short example of checking out a branch?23:30
jono_mwhudson, actually, forget it23:31
jono_I figured it out :)23:31
mwhudsonjono_: heh23:31
jono_thanks!23:31
mwhudsonjono_: generally to find out how command 'foo' works23:31
mwhudsonlook for cmd_foo in bzrlib/builtins.py23:31
* mwhudson afk for a few then23:31
jono_right thanks!23:31
jono_cheers pal23:31
jono_I will add some examples to python-snippets23:31
jono_would be awesome if you could submit some more :)23:32
DavieyThere is no stopping jono_ now!23:32
jono_Daviey, :)23:32
jono_just hacking Acire to check out the code directly from bzr23:32
jono_then no dailly PPA is needed :)23:32
lifelessjono_: hi23:33
jono_hey lifeless23:33
lifelessjono_: your mail server is naffed23:33
pooliehello jono, lifeless23:33
jono_lifeless, yeah, I know23:34
lifelessjono_: I emaile you a week ago with the answer to this :P23:34
jono_1and1 having troubles23:34
jono_lifeless,23:34
jono_ahhh23:34
jono_can you forward it to jono AT ubuntu DOT com?23:34
lifelesswhen you pinged me, then disconnected ;P23:34
jono_hah23:34
lifelessjono_: sure23:34
jono_thanks!23:34
lifelesspoolie: there is a gift for you in the tribunal review queue ;)23:36
poolieheh, thanks23:38
pooliebtw did you see my grief on friday was due to a fsck bug?23:39
lifelesspoolie: I didn't see the resolution of it, no.23:39
lifelessthats good news.23:39
pooliemy machine is still down but at least the problem is known23:41
poolieit's actually fairly obvious where it must be23:41
poolieironically similar to some bugs we had in dirstate23:41
pooliein this case, it incorrectly sorts things that are lexicographically before '.' i think23:42
lifelessheh23:42
lifelessso 'do not fsck and you will be fine'23:42
lifeless ?23:42
pooliedo not use -D23:44
poolie*afaik, there may be other problems23:45
poolieactually since the manpage says it may resort directories sometimes even without -D23:45
pooliei don't think that's enough23:45
poolieand you can't upgrade without fscking23:45
lifelessmwhudson: is the link in http://doc.bazaar.canonical.com/bzr.dev/developers/#developing-using-bzrlib stale? (to the api docs)23:58

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