/srv/irclogs.ubuntu.com/2008/07/06/#bzr.txt

leo2007does the windoes installer require python to work properly?00:07
Verterokleo2007: there is a standalone bzr.exe and a installer that needs a python installation00:11
Verterokleo2007: bzr-setup-1.5.exe is the standalone installer, and bzr-1.5.win32-[py2.5/py2.4].exe needs a python install00:13
leo2007Verterok: trying to push a local branch to a remote machine through sftp, it didn't succeed00:15
Verterokleo2007: any error msg?00:16
leo2007Verterok: it looks like it succeeded but no files were actually uploaded.00:19
Verterokleo2007: a push don't create/update a remote workingtree00:20
Verterokleo2007: in the remote location you should have a .bzr/ dire00:21
Verteroks/dire/dir/00:21
leo2007Verterok: yes. but my local repo has versioned files00:22
leo2007they weren't uploaded00:22
Pengleo2007: Push doesn't create a working tree. All of the history is in .bzr.00:22
Pengleo2007: If you just want to push and pull from that location, you don't need a working tree.00:23
Pengleo2007: If you really do need one, check out the bzr-upload plugin.00:23
Verterokleo2007: http://doc.bazaar-vcs.org/bzr.dev/en/user-reference/bzr_man.html#push00:25
Verterokleo2007: check the "Description" section, or bzr help push ;)00:26
leo2007Peng: Thank you. That resolved my confusion. I thought the files were being left out00:26
leo2007what tool do you use to manage bzr repo in windows? I have been using the dos window and it is quite inefficient.00:30
Verterokleo2007: I don't use windows, but take a look to qbzr, bzr-gtk and tbzr (tortoise-like)00:34
Verterokleo2007: http://bazaar-vcs.org/QBzr , http://bazaar-vcs.org/bzr-gtk00:36
Verterokleo2007: and http://bazaar-vcs.org/TortoiseBzr00:37
leo2007Verterok: thank you very much. I will take some time to read the doc properly.00:40
Verterokleo2007: y'r welcome :)00:40
leo2007Verterok: good night;)00:42
Verterokleo2007: seeya00:43
rockstarjames_w, are you around?01:42
alencoolhello07:28
alencoolis there anyone in here that could help me?07:29
skavezis there an easy way to install bzr-svn on os x? or is recompiling svn still the only way?08:31
RAOFNewer bzr-svn has it's own subversion bindings; you'd need to build bzr-svn, but you shouldn't need to rebuild subversion.  I think.08:32
lifelessola!08:41
lifelessso08:41
lifelesswho wants to hear some good ne08:41
lifelessws08:41
RAOFlifeless: Oooh!  Oooh!  Me!08:44
lifelessRAOF: on the plane I wrote a new compressor that is (unoptimised) 10% slower than knits, but generates data 10% of the size: smaller than git-repack --window 200 --depth 20008:44
RAOFMy my.08:45
RAOFThat _is_ impressive.08:45
lifelessam now ingensting caffeine and grease08:45
lifelessand will try to make it end user accessible/fix some glaring inefficiencies08:46
RAOFI hope you're rounding that meal out with the other 2 food groups (starch and burnt crispy bits).08:46
lifelessgit  200, 200        638227   0.6MB08:48
lifeless patience all:08:49
lifeless16665951 25.320 197 164 0-mbp%40sourcefrog.net-20050323055643-668814a4d647823508:49
lifelessafter zlib 59903608:49
lifelessand - thats getting some stuff whackily wrong ecause sequencematcher is not designed for relocations08:49
lifelessso am rewriting the match logic entirely08:49
lifelessRAOF: oh, I don't do any of that whacky try-N diffs and pick one stuff either.08:51
RAOFYou're not using the rsync compression algorithm, either :)08:52
lifelessno08:52
lifelessits broadly speaking just sliding window with references to the output08:52
RAOFIs this possibly going to improve initial-branch-from LP speed?08:53
lifelessbut line orientated and entropy coded post processor08:53
lifelesswell, downling 10% of te total data -> should be faster08:53
lifelessbut there is a ways to go from here to production08:53
RAOFRight.  Initial-branch is basically the only bzr operation that takes an apreciable and annoying time for me, so that's the filter I view bzr improvements through.08:55
RAOFI should really learn a little more compression theory.  I'm only familiar with huffman encoding.08:55
lifelessso huffman is probabilistic encoding:08:55
lifelessbuild a tree weighted by the probability of a symbol, and to emit something you name a path in the tree; the deeper the tree the more things you can output with a short path08:56
RAOFRight.08:57
lifelessI'm reasonably sure I have http://www.amazon.com/Data-Compression-Book-Mark-Nelson/dp/1558514341 first edition around at home somewhere08:58
lifelessor a t least, I have *a* textbook on dc08:58
lifelessanyhow dictionary based compression you have a dictionary, and the compressor outputs keys in the dict08:59
lifelessmost compressors are dynamic these days - turning the state with the symbols rom the input stream, and the decompressor mirrors that08:59
RAOFOk.09:00
lifelessanyhow, what I do is:09:01
lifelessoutput the first text09:01
lifelessdiff the next against that, output the diff09:01
lifelessdif the next against the total output so far09:01
lifelessand so on09:01
lifelessso the output is a combination of refernces to the output stream so far, and new symbols09:02
lifelesswhich is basically an infiinte window  dictionary cmopressor09:02
lifelessthe only reason to do this in preference to e.g. lzma is that I know I don' need the intermediate texts09:03
lifelessso decompression is basically one read, split the bytes into 'text before the delta we want, and the delta w want'09:04
lifelesspasrese the delta09:04
lifelessapply09:04
lifelessalso, typign from London; excuse the spellink09:04
RAOF:)09:06
RAOFRight.  So, nice and quick unpacking.  Cool.09:07
lifelessyah, typical dictionary decompressiors need to care about the entire input stream09:09
lifelesswhereas my references are not to an internal symbol able, but to the actual output09:10
lifelessless efficient,. but for some work sets I'm seeing 200:1 compression *before* zlib is applied09:10
RAOFI suppose revision data is likely to be an ideal candidate for compression.09:11
lifelessone insight the git devs had that is very useful is that decreasing size makes most patches a series of deletes09:11
lifeless(or more deletes than adds)09:12
lifelessI *think* I can make that largely irrelevant, but until I finish replacing the 'does this work' sketch with something that handles moves it will bite09:12
skavezargh! does "bzr: ERROR: exceptions.ImportError: cannot import name make_file_factory"  sound like something i can fix? i'm getting it while trying to branch an svn repo with bzr-svn 0.4.11 and bzr 1.6b209:15
lifelessskavez: you need bzr.dev for that version of bzr-svn I think09:23
BjornTdo i need any special configuration for using 'bzr send'? when i try 'bzr send --mail-to=email@example.com', all that happens is that a terminal with mutt is opened (listing my inbox), and no mail is sent09:25
skavezlifeless: ah ok. i'll give that a shot09:26
lifelessBjornT: the hlep should describe it BjornT ; its trying to use mutt as your mailer; perhaps you have misconfigured xdg-utils or something?09:27
BjornTlifeless: well, i'd love to use mutt :), as long as it would allow me to send the e-mail.09:28
lifelessBjornT: poke at the help, and the mnual, failing that file a bug09:29
BjornTlifeless: at least it works if i specify mail_client=mutt09:31
BjornTxdg-email email@example.com doesn't work either, so i guess that's where the bug is09:33
lifelessBjornT: I would guess so09:40
lifelessRAOF: lp:~lifeless/+junk/bzr-groupcompress09:40
lifeless-<> gate A18 now :)09:40
=== mwhudson_ is now known as mwhudson
=== Tank|Awa1 is now known as Linnk
j^is it a common problem that the trac plugin gets really slow for the /timeline view?13:55
j^when i started a new repos it was ok, but it seams it gets slower with each commit i make13:56
jettohello15:15
jettoI come from cvs and Clearcase(base not ucm), and I'm looking to use a more advenced scm tool.15:16
jettoI'm very confused with bzr concept I want to do some test one a real work15:18
jettoI try to hack a game : rrgbis. I got sources from SF and do some fix. then upstream release a new version than include some of my fix. I would like merge then continue to hack.15:21
jettoHow can I do this on bzr ? I mean having something like a vendor branch on CVS15:22
Odd_Blokejetto: What SCM tool are they using on SF?15:34
uwscvs :)15:45
Odd_BlokeHmm, I'm not sure of the best way to deal with CVS is.15:46
bigdoggood morning15:49
bigdogis this the place to ask about a problem I am having pushing a branch up to launchpad?15:49
bigdogI am using windows15:50
uwsbigdog: yes, or #launchpad15:50
bigdogI am using bzr 1.515:50
bigdoguws: I will start here, If I describe my problem, and I should go to #launchpad, let me know15:51
bigdogWhen I was using bzr 1.2, on windows, with ssh key,  I was able to push branches with no problem (a couple of months ago)15:51
uwsbigdog: i'm not the one who can help you I'm afraid15:51
bigdoguws: ok15:52
uwsbut usually the people in this channel are quite responsive and helpful15:52
uwsso do proceed15:52
bigdogthanks15:52
bigdogSo everything worked a couple of months ago.  windows + pagaeant  .15:53
bigdogyesterday I had some time to push a new branch.  I noticed there was a new version of bzr, so I grabbed and ran the installer15:53
bigdogH:\launchpad\txcomputegrid>bzr launchpad-login -v  bigdog15:54
bigdogbzr: ERROR: The user bigdog has not registered any SSH keys with Launchpad.15:54
bigdogI have the same ssh key, that is registered with paegent15:54
bigdogsame response for set BZR_SSH=plink15:55
bigdogand set BZR_SSH=paramiko15:55
bigdogany insight?15:55
bigdoguws: Sunday morning is busy day for people,  I will try #launchpad16:01
Odd_Blokebigdog: That'll probably be a result of the Debian SSL issues.16:02
Odd_Blokebigdog: Check that you actually _do_ have a LP key registered.16:02
bigdogOdd_Bloke: thanks16:08
Tim3393Hello,16:19
Tim3393I have a question regarding the way bazaar does handle duplicate files when storing them in a repository.16:19
Tim3393In my project there are often a number of identical files in different locations of the working tree.16:20
Tim3393As I do not want the repository to be unnecessarily blown up I would like to know if bazaar is so smart that it will detect the identical copies and actually only store their content once?16:20
clementeBazaar stops when doing some operations (ex: branch) and starts running again after I press ENTER four times...17:35
clementeIn fact it's doing read(0,...) according to strace17:36
clementeDo you also get this behaviour? I can always reproduce it17:40
clementeBoth with 1.5 and with latest version from today17:40
clementeIt is probably related to dbus...17:50
antoranzhi guys! Do you know when 1.6 is commign out?18:14
leo2007how to set up emacs to use bzr in windows?18:28
clemente(I submitted bug 246052 for that problem of halting on STDIN)18:34
ubottuLaunchpad bug 246052 in bzr "Bazaar halts and silently expects input from STDIN" [Undecided,New] https://launchpad.net/bugs/24605218:34
* Foskasse ▄█▀ █▬█ █ ▀█▀19:07
KinnisonHow delightful19:09
* Foskasse █▬█ █19:26
KinnisonMy god it's expensive to convert from weave to btree-plain19:28
kgoetzhi hackers! i seem to have exploded bzr :( http://paste.ubuntu.com/25500/ i installed python 2.5 recently. could this be the cause?20:19
gourKinnison: when is btree supposed to go into trunk?20:23
Odd_Blokekgoetz: That error looks familiar, but I can't remember what causes it. :(20:28
kgoetzOdd_Bloke: incase it helps: i merged from the repos default branch, then realised i had no local changes, then pulled, and you see the result there.20:29
Kinnisongour: no idea20:34
Odd_Blokekgoetz: I'd recommend looking through LP for bugs and looking at the ML.20:35
kgoetzOdd_Bloke: i'll try and give LP a trawl.20:36
kgoetzthis bug looks similar, but seems totally unrelated. https://bugs.launchpad.net/bzr-gtk/+bug/23720520:45
ubottuLaunchpad bug 237205 in bzr-gtk "error starting webbrowser.GenericBrowser in python2.4" [Medium,Fix committed]20:45
mwhudsonany devs awake?21:44
=== sdboyer is now known as sdboyer-laptop
=== sdboyer-laptop is now known as sdboyer
=== sdboyer is now known as sdboyer_
=== sdboyer_ is now known as sdboyer
abentleymwhudson: I'm not *really* awake...21:50
mwhudsonabentley: so bzr get nosmart+<http url that redirects> bombs out pretty messily21:51
mwhudsonabentley: any obvious ideas for places to look?21:51
mwhudsonoh21:52
* mwhudson finds a fixme21:52
mwhudson            # FIXME: If 'transport' has a qualifier, this should21:52
mwhudson            # be applied again to the new transport *iff* the21:52
mwhudson            # schemes used are the same. Uncomment this code21:52
mwhudson            # once the function (and tests) exist.21:52
mwhudson            # -- vila2007021221:52
abentleymwhudson: I haven't see the actual traceback...21:54
mwhudsonsorry21:55
mwhudsonabentley: http://pastebin.ubuntu.com/25504/21:55
abentleyYeah, so I think you've already found the relevant bit.21:55
mwhudsonthough the exception is from before the comment21:56
mwhudsonbut yeah, it seems the problem is somewhat known21:56
mwhudsoni guess vila is the man to hassle about this22:00
abentleymwhudson: So what's being done there is a bit bogus.22:00
abentleyBecause there's no reason to assume that it's a relative path.22:00
mwhudsonright22:00
abentleyIt could legitimately be on a different host, for example.22:00
mwhudsonindeed, in http its really really not supposed to be22:01
abentleySo it should probably be using urlutils.relative_url instead.22:01
mwhudsonsomething a little funky is happening, because http redirects do actually work most of the time22:01
abentleyThat operation doesn't error when the urls are unrelated.22:01
abentleyOkay, here's what I think happens.22:03
abentleyrelpath is being used in a check.22:03
abentleyspecifically, if I try to get foo/bar/baz, I should get redirected to */baz22:04
mwhudsonright22:04
abentleyAnd there's a discrepancy between what e thinks the url was and what the transport thinks it was.22:05
mwhudsonhm hm22:10
mwhudsonabentley: i think the problem is more basic:22:14
mwhudsonabentley: http://pastebin.ubuntu.com/25513/22:15
* mwhudson winds up a poking for spiv when he wakes up22:15
abentleymwhudson: How does that contradict me?22:16
mwhudsonabentley: it doesn't22:16
abentleyOh.  Okay.22:17
mwhudsonwell, hm22:17
* mwhudson thinks22:17
abentleyI think that it is correct to not treat http://foo/bar as a child of bzr+http://foo22:17
mwhudsonp = '...'; get_transport(p).relpath(p) shouldn't fail though, should it?22:19
mwhudsonthis is tripping up the puller where we force all http: urls to be nosmart+22:19
mwhudsonmaybe there's a better way to do that22:20
abentleymwhudson: Yes, it's legitimate for that to fail.22:20
abentleyget_transport may canonicalize the url, and it may do a lookup.  e.g. lp:bzr22:20
mwhudsonhm22:21
abentleytransport.base will give the the final value the transport used.22:22
mwhudsonso yeah, this stuff in redirected() does look a bit bogus then22:25
jettohum bzr looks like too hard form me :-(22:26
mwhudsonjetto: ?22:27
jettoI would like to do some thing very simple and it's not clear how to do it :22:28
jettoI get source as a tgz from upstream. I hack a little, then the upstream release a new version. I merge and continue to hack. How to do this ?22:31
mwhudsonjetto: i think the 'bzr import' command can help with this22:32
jettoI mean I would like to follow pure upstream change (maybe in a branch) and do my change in another22:33
mwhudsonjetto: something like bzr init-repo repo; bzr init repo/upstream22:33
mwhudsoncd repo/upstream; bzr import ~/release-0.1.tgz; bzr add; bzr ci -m 'relese 0.1'22:33
mwhudsoncd ..22:33
mwhudsonbzr branch upstream22:33
mwhudsonbzr branch upstream my-branch (sorry)22:34
mwhudsoncd my-branch22:34
mwhudsonhack hack hack22:34
mwhudsoncd ../upstream22:34
mwhudsonbzr import ~/release-0.2.tgz; bzr add; bzr ci22:34
mwhudsoncd ../my-branch22:34
mwhudsonbzr merge ../upstream22:34
mwhudson(disclaimer: i haven't done this)22:34
jettowell I'll test this, thanks.22:35
jettoIt's hard for me to adapt to decentralized SCM22:37
jettoeven choosing one is very hard ;-)22:38
jettodo I need to do a init-repository ?22:39
mwhudsonyou don't need to make a repository no22:40
mwhudsonit's just a way of sharing storage, it takes up less disk space and makes making new branches quicker22:40
mwhudsonbut i guess in your case, history is going to be pretty shallow22:40
* Foskasse 1:20 para fazer anos!!!22:49
jettomwhudson, your pocedure rocks, thanks23:25
mwhudsonjetto: glad i helped23:25
* Foskasse quase quase quase a fazer anos!!! :D :D :D23:28
Pilkyjetto: if it helps at all you can use bzr as a centralised VCS, with all the improved branching/merging benefits23:31
Pilkyin some ways bzr is one of the best centralised VCSs out there ;)23:31
jettoPilky, I think I'll do this but not for home work23:32
luisshola23:53
luisshola23:53
luisshola23:53
pooliehello23:54
Verterokmornin' poolie23:55

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