[00:00] * mwhudson attempts to figure out how releases on launchpad work these days [00:01] Tak: it's sometimes mentioned in the NEWS file [00:01] Tak: I'm not seeing anything on Lucid [00:03] I thought maybe there was a change in the embedding api, or that I've been Doing It Wrong in a way that just didn't break before (md-bzr), but then a google search is gives hits for similar issues with a fair number of python apps [00:03] (not that I couldn't still be doing it wrong) [00:04] cool, so the api change was 2.1.0b1 - thanks [00:04] hi everyone. I am new to bzr and I have a question... Basically I have this external "branch" which hooks into a larger repository externally -- i would like to do development on my branch locally and then "commit" only to my branch on the remote shell [00:04] how does one go about "committing" to a branch [00:04] if you have any suggestions for doing local development and then getting those changes under version control please let me know [00:10] defn: I'm not sure I understand what you're trying to do [00:11] jelmer: me either! :) [00:11] jelmer: btw, I'm seeing the filehandle leak on both sid and osx 10.6 [00:11] defn: you'd like to do a partial checkout of a remote repository basically? [00:11] Tak: I'm not claiming it doesn't exist on lucid, just saying I haven't seen it yet :-) [00:11] s/seen/noticed/ [00:11] 'morning Ian! [00:11] hi jelmer [00:13] * Tak nod [00:23] jelmer: hmm, im just sort of confused on how my repo was setup initally -- let me do some digging for how it was created and then let me rephrase :) [01:17] hi [01:17] moin poolie [01:18] hi there [01:27] hi poolie [01:28] hi [01:33] poolie: Windows installers built and uploaded for 2.0.5 and 2.1.1 over the weekend [01:33] yay [01:34] how hard/easy would it be to do a 64-bit windows build? [01:34] people have asked a few times [01:35] jelmer, igc_: i think my mail about plugin compatibility was too braindumpy; nobody replied yet [01:36] poolie: I haven't seen it [01:36] but what would you two think about 1- having a blacklist in bzrlib to make it more symmetric [01:36] but it sounds like something I'd have an opinion on [01:36] heh, it does, doesn't it :) [01:36] 2- making the failure less noisy [01:37] I fear that sort of blacklist. It's got a neon sign above it saying "Ask Me About Being Out Of Date" [01:38] fullermd: depends how its phrased and populated [01:38] fullermd: if it was 'plugin X < Ver-Y is bust' then its not stale daya [01:38] < very-Y will stay bust [01:38] s/day/data [01:38] right [01:39] it may be that we add the faciliity and don't make use of it [01:39] * fullermd confiscates lifeless's 'y' key for rampant misuse. [01:39] poolie: I don't know how quiet you want to go. I think a broken plugin should be listed on every invocation of bzr. [01:39] we could put it only in 'bzr plugins' [01:39] even though its noisy, its an aberrant situation. [01:39] and crucially its not easily discoverable [01:39] we may just substitute wonders about why something broke though [01:39] right [01:40] so i think the main thing is to only be giving warnings if the plugin actually would not work [01:40] hmm [01:40] we could [01:40] ie eliminate both type I and II errors [01:40] if exiting with non-zero (e.g. NotBranchError, NoSuchCommand etc) [01:40] also show broken plugins [01:40] That could get irritating in the case where the user can't do much about it though. [01:41] mm [01:41] 'svn:///host/path is not a branch\nWarning: plugin bzr-svn failed to load. See ~/bzr.log for details.' [01:41] They can wind up with nothing to do but set whatever that env var is to disable a plugin, which then becomes an eternally-encysted part of the env. [01:41] fullermd: so, if they are running a non-installed bzr, they can set BZR_PLUGINS_PATH to control the plugins, without disabling just-one [01:42] I was thinking more the case of running an _installed_ bzr, with installed b0rked plugins. [01:42] fullermd: so, a sysadmin fail ? [01:43] In a hard or soft sense, yes. [01:43] I tried fixing such things, but I ran out of bullets :) [01:43] poolie: anyhow, I think both your 1 and 2 could be good [01:43] though I haven't [yet] read the mail to know the context. [01:44] fullermd's concern about staleness of data would be something to cater for in the blacklist design. [01:56] poolie: I'm yet to read it sorry [01:57] poolie: I scanned it and it had good ideas IIRC [02:15] poolie: no opinion on 64-bits builds currently [02:28] how can I merge just a couple of directories into a branch? [02:29] with difficulty [02:30] (assuming the changes are in changesets that altered other files) [02:30] you can merge + revert the bits you don't want (but then the other bits will never be merged) [02:31] huh [02:31] easy [02:31] bzr emrge branch/dir1 [02:31] bzr merge --force branch/dir2 [02:31] bzr commit [02:31] note though, that like *all* partial merges, it won't be recorded as a merge. [02:34] what is the proper way to rename/move A.c to B.c and B.c to A.c ? [02:35] "bzr rename"? [02:35] Use an intermediate name and 3-step it. [02:35] Or use the `bzr xor` trick... ;> [02:35] Oh. You can't do it in one step? [02:36] How would you? [02:36] Err, sorry. [02:36] I was thinking, like, commit wise. Some VCSes don't let you do something like that in one revision. [02:37] that's what i m wondering, though my example was a bit retarded in fact i have to version of the same html page: index.html and index2.html (index.html being the default) - now i want to swap them so index2.html is going to be the default index.html [02:37] bzr mv index.html tmp ; bzr mv index2.html index.html ; bzr mv tmp index2.html [02:38] Peng: one commit is fine. [02:38] You can do it all in one rev, sure. I've... well, I've never done that, but I've moved a file and then created a new file with the old name a lot. [02:38] fullermd: thanks [02:39] Hiya an a Sunday nite! If I 'bzr branch aRepo', then 'bzr mv aRepo/someDir aRepo/newDir', does 'bzr merge aRepo' automagically put changes under version control in the original source's aRepo/someDir in the *renamed* aRepo/newDir? [02:40] Guest87486: I don't know quite what you're asking. [02:40] Guest87486: but I want to know that you can't branch a repo. Youc an only branch a branch. [02:43] lifeless: I'm just trying to understand if bzr is smart enuf to put changes originally meant for "aRepo/someDir" into the directory I renamed it as "aRepo/newDir". [02:44] Guest87486: yes [02:46] * igc_ out for a few hours [02:47] Guest87486: (yes, bzr is smart enough) [02:48] lifeless: I just started with version control in general, and bzr in particular. pretrry amazing once you start to get the hang of it! Thanks. [02:58] hi, what is the submit branch shown in branch.conf? [02:59] It's the default location used for certain operations (send and merge at least, I think) [03:00] any way to move it back to trunk? (i messed up my repository when trying out my first branching operation) [03:01] Using something like `merge --remember` would reset it. You could manually edit it in the branch.conf. [03:03] branch.conf also has some of my commit messages in it [03:03] what does merge --remember do? i don't want to mess things up more [03:03] It shouldn't.... oh, maybe qbzr's uncommit hook stashes them there. [03:04] `merge --remember xyz` does pretty much the same thing as `merge xyz`, except that it stores xyz as the remembered location. [03:04] Without --remember, it will only save it if there isn't already a saved one. [03:10] the submit branch is set to a branch I manually deleted, which is causing me some errors. I have nothing to merge [03:10] What errors is it causing you? [03:11] The stored location is just used as a default location for certain commands when you don't specify one explicitly. Aside from that, it's pretty much cosmetic; it doesn't imply any lower-level link between the branches. [03:11] when i open my trunk it says in bzr explorer "Not a branch: ~/project/trunk/branch1 [03:12] Mmm. Explorer is off my turf. That's igc_'s baliwick, but he's away right now. [03:12] You could just manually delete the line wholesale from the branch.conf. [03:13] right [03:13] also please file a bug on bzr-explorer at bugs.launchpad.net/bzr-explorer [03:15] deleting the submit branch=... line from branch.conf seemed to work [03:15] should i also delete the line below it that just says [commit data]? [03:16] i have a question on launchpad bzr-explorer already [03:16] about this? [03:16] ya [03:16] Probably not... I assume those are saved uncommit messages or something. Might as well leave 'em alone without a good reason to fiddle. [03:32] ok, thx, i think its fixed [03:33] does anyone know what the colored circles indicate in bzr explorer when viewing your log history? One of my branches is colored pink and one is colored sky blue. [03:36] I think it uses some heuristic to guess which branch different revs came from, and colors them based on that. I'm not sure it has any deeper meaning. === timchen1` is now known as nasloc__ [05:01] can bzr apply patches? [05:02] GungaDin: yes, bzr help patch i think [05:02] hm [05:02] thx [05:02] If you just have a plain diff, you can use the usual 'patch' command... [05:02] maybe that's from some plugin actually... [07:27] I am trying to run 'bzr log' with bzr from trunk. but thats failing with api not compatible error. version and st work and pull says there is nothing to pull http://pastebin.com/Hbgt3KRM [07:28] am i missing something? [07:28] hi all ! (new DST here for those who care :) [07:28] parthm: upgrade or disable your plugins until they update their checks against bzrlib API [07:30] vila: ah ok. so a plugin probably updated the log command. it works fine with --no-plugins. thanks. [07:31] parthm: just loading the plugins (even without modifying log) can trigger that [07:31] vila: i should have thought of that earlier :) [07:33] How can I change the path to Python that Bazaar uses? [07:33] davidstrauss: an installed bzr ? [07:33] davidstrauss: or running from source? [07:34] lifeless: Either, preferably installed [07:34] lifeless: I need to run Bazaar under Python 2.6 for bzr-svn support for one use. [07:34] user* [07:34] well [07:35] you'll need to installed it under python 2.6 [07:35] unless you're using packages, which should have instaleld it under all versions [07:35] the #! line at the top of /usr/bin/bzr controls the interpreter used to run bzr [07:35] but you can also do 'python2.6 /usr/bin/bzr ....' [07:38] lifeless: Running the "python2.6 /usr/bin/bzr" complains about not having bzrlib in the PYTHONPATH [07:39] davidstrauss: then its not installed for python2.6 [07:40] davidstrauss: is it ubuntu packages that you've installed it with ? [07:40] or something else? [07:40] lifeless: This is with the Four Kitchens Yum repo packages for RHEL/CentOS [07:41] * Peng declares dealing with plugin API compatibility the most annoying thing ever. [07:44] * fullermd waves at vila. [07:45] lifeless: I think I figured it out. [07:45] * vila waves back ! [07:47] davidstrauss: I don't know anything about CentOS python package sorry [08:06] lifeless: teddybear about apport [08:06] with my current fix for bug 528114 we can get ubuntu-bug sending the crash files to lp again [08:06] Launchpad bug 528114 in bzr "apport complains "This problem report applies to a program which is not installed any more."" [High,In progress] https://launchpad.net/bugs/528114 [08:06] but there are two drawbacks [08:07] firstly they are filed against the ubuntu package when they're probably upstream bugs [08:07] secondly they are private by default, pending retracing [08:07] and there seems to be a long delay there [08:08] back [08:08] with a running system this time :-) [08:14] poolie: hi [08:14] poolie: uhm, some apport is better than no apport [08:14] yeah that was basically the question [08:14] poolie: what caused the issue? Have you corresponded with pitti ? [08:14] previously; not today [08:14] i think i just need to fix a larger bug [08:15] well yes, I know previously :) [08:15] pitti is up , I saw him in another channel; might make sense to have a hallway chat with him about this now. [09:56] * igc1 dinner [10:27] Hi everyone [10:27] Is it okay to use bzr on a low-bandwidth connection ? [10:28] * writer is on a 512 Kbps connection, and getting 'bzr: ERROR: Connection error: while sending GET /ikarus.dev/.bzr/repository/indices/a2ff5c6db2537db1740e0e9ba5216e52.six: [Errno 4] Interrupted system call' with bzr 2.1.0-2 on Arch GNU/Linux [10:29] thats not going to be your connection per se [10:29] perhaps there is an intercepting proxy at your ISP or something [10:30] bzr is usable with lower speed connections than yours [10:30] lifeless: there is not any, AFAIK, and I'm getting download speeeds like 2KB/s, 1KB/s and then also get spikes like 67KB/s, 22KB/s [10:40] Wait a minute. "Interrupted system call"? As in EINTR? [12:01] Hey, I accidentally forgot to specify files to commit. Can I un-commit? [12:01] oh [12:01] uncommit heh [12:02] I didn't expect that to be there :o === radoe_ is now known as radoe === tumbleweed_ is now known as tumbleweed [14:46] night all === igc1 is now known as igc === salgado is now known as salgado-lunch [16:40] Hello. It's impossible for me to to even a simple bzr branch, I always get http://pastebin.com/mrm62Bd1 . However, 1. my public key is in ~/.ssh/launchpad_id_rsa.pub, 2. It matches what LP outputs when I click on "SSH keys: ronj@launchpad.blob", and 3. I tried removing it from LP and reimporting it. Help! [16:41] Using Ubuntu 9.10 x86 with bzr 2.0.2 [16:42] And I forgot to mention: it used to work. I don't know what caused it no now fail === deryck is now known as deryck[lunch] [16:59] hi guys, i currently have a bzr branch that's pack-0.92 that i want to convert to rich-root-pack so i can pull; how can I do this? [17:00] davertron: bzr upgrade --rich-root-pack [17:00] davertron: or, alternatively -- 'bzr upgrade --2a' if you're on bzr >= 2.0 [17:00] jelmer: is that a pretty safe avenue? [17:00] jelmer: should i worry about the possibility of a corrupted branch or anything like that? [17:00] davertron: yeah, though I'd really recommend using 2a if you can - it's much faster === IslandUsurper is now known as IslandUsurperAFK [17:01] jelmer: looks like i'm using 1.13.1 === IslandUsurperAFK is now known as IslandUsurper === salgado-lunch is now known as salgado === beuno is now known as beuno-lunch === deryck[lunch] is now known as deryck [18:19] For remote access, I know I can use bzr+ssh or sftp. What's the _difference_ between the two? When should I use which? Does it matter? [18:20] sftp uses a dumb filesystem on the other end; namely the SFTP server [18:20] bzr+ssh shells in to the remote server, runs bzr, and uses it as a smart agent [18:20] LeoNerd: so bzr+ssh is a better opt, it seems [18:20] bzr+ssh costs more in terms of resources on the server, but the smart agent can help reduce network usage or roundtrips [18:21] Can be. Depends on your setup [18:21] between my desktop and my server I use sftp, because my server is a celeron 333 with 784MiB of RAM accessible over a 100Mbit LAN with <1msec of latency. [18:21] sftp is finished well before that celery has even fork()+exec()'ed a python process. [18:24] LeoNerd: Thanks for the explain! [18:34] does anybody know if horizontal scrolling with the mouse wheel doesn't work in qbzr because of qbzr or Qt itself? [18:37] I'm using a touchpad, but I know that's not the problem because Chrome and Firefox scroll horizontally when I use it [18:43] IslandUsurper: where in qbzr? [18:43] it works for me in qdiff [18:43] (using a touchpad) [18:44] also works in qlog [18:49] luks, well, I was looking at qdiff [18:49] However, I'm using Windows [18:55] ah, can't test on windows right now [18:56] when I get home, I guess I can test it on Linux === beuno-lunch is now known as beuno === radoe_ is now known as radoe [19:09] Hi, I've just found a crash in qbzr owing to lazy-importing bzrlib.revision.NULL_REVISION. Is lazy-importing a constant ever the right thing to do? [19:19] vila, around? can you have a peek at https://answers.edge.launchpad.net/launchpad/+question/105604? [19:19] (I just ran 'bzr get lp:vm' and it worked just fine) [20:28] jam: does that loggerhead merge proposal have any particular bzr version requirements? [20:34] we are looking for a version control system for our developers to install on their machines. Some dev work is done directly on a web server though. can Bazaar integrate directly to keep a backup of files on a remote web server? [20:45] GaLiLe0, you have a plugin called "bzr-upload" which will just upload the working tree [20:47] Hi all. [20:47] hey Gary [20:47] * GaryvdM came to find bailix [20:47] Hi Jelmer [20:48] \o/ Built qbzr inno installer under wine :-) [20:49] Bye - rebooting to test in windows.... [21:04] jam, ping === Toksyury1l is now known as Toksyuryel [21:08] jelmer: we have some fun problems with bzr-git [21:08] jelmer: in particular the size of the cache for the incremental kernel import [21:13] hi beuno [21:13] mwhudson: I'm sure it does, I don't know it offhand, probably bzr 2.1? [21:13] maxb: it would be reasonable to lazy-import the module [21:14] jam, 1) You rock! Thank you for the loggerhead branch. 2) Is it compatible with previous bzr'z? [21:14] but yeah, the constant is bad [21:14] beuno: prob bzr 2.1 [21:14] ah, mwhudson already asked that :) [21:14] beuno: actually, I see "KnownGraph" in one of my bzr 2.0 branches, let me look closer [21:15] KnownGraph was introduced in bzr 1.18 [21:15] sounds ok to me then [21:15] me too [21:15] ps: does someone want to maintain loggerhead instead of me? [21:15] however, KnownGraph.get_parent_keys is in 2.1, but not 2.0 [21:16] i don't feel like i've been doing a very good job lately [21:16] jam: ah [21:17] I vote for Peng [21:17] let me see if there is a way to work around it in an obvious manner [21:18] mwhudson: both the C and python versions expose self._nodes which is the internal graph [21:18] jam: I've bumped qbzr min required bzrlib version due to that. [21:19] however, the objects in there aren't identical wrt the C vs python versions [21:19] so it isn't great to grab that [21:19] I could certainly write (ugly) code that would work under all versions [21:19] if 2.0 compatibility is important [21:20] (pure python stores a reference-by-key , Pyrex stores a reference-to-the-node) [21:21] jam, I think we can just say that the next release of LH only works with 2.1 [21:21] and it's twice as fast :) [21:22] well, this isn't the history cache loading, just the generation without-cache step [21:22] I'm toying with ideas about how to handle a graph data store that isn't "1 rev-at-time" while also not being "all revs, all the time" [21:22] I plan on writing something up for the list [21:23] I think the first hit is usually what gives people the impression it's slow [21:23] though honestly, the loggerhead code doesn't really need to be calling "self._load_whole_history_data()" all the time [21:23] which could help a lot [21:23] The main page doesn't seem like it uses most of that data [21:23] jam: i think hazmat has been working on that [21:24] mwhudson: assuming that is the same Kapil that was posting to canonical-tech, that is actually what inspired me to look into this code [21:25] jam, mwhudson i just removed it from a views that where obvious, i'm not going to be actively pushing it, my knowledge of bzr internals is pretty rudimentary [21:25] however, his discussion was "use chameleon, disable all the stuff that seems to be O(history)" [21:25] jam, thanks for having a look at that [21:25] vs, how keep the info, without having to load the *entire* history [21:25] i checked out your branch, it does seem to better on the cache computation [21:26] but yeah.. that's an accurate summary :-) [21:26] mwhudson: so, other than pushing data into production, and letting people bang on it, what sort of testing does loggerhead have? [21:26] specifically, if I just started hacking in 'history.py' to make it not load the whole graph [21:26] jam: yeah, that's about it [21:26] how would I find out where the gotchas are [21:26] jam: i don't know [21:27] code-inspection it is, I guess [21:27] the code makes me very sad in general [21:27] btw, there is one other part of my patch I didn't mention, which is the "try: import loggerhead.app.branch" stuff [21:27] there's not actually that much code at least [21:27] for some reason on my machine, "try: import loggerhead" succeeds [21:27] but "try: import loggerhead.apps.branch" fails [21:27] oh yes [21:27] that end of things is all screwed up too :/ [21:28] thumper: Hi [21:28] jelmer: on a call right now, with you shortly [21:29] anyway, with that patch "bzr serve --http" works for me, so I'm happy enough with it :) [21:29] thumper: If it's a stand-up, mwhudson knows about it :-) [21:30] mwhudson: what is "cache_key", it seems to be BranchWSGIApp.friendly_name, but that still doesn't help much [21:30] is it ~ the path on disk? [21:32] and, IIRC, the mapping in the sql data structure is "branch_tip_revision => revision_graph_of_whole_ancestry" right? [21:32] (so the data currently stored cannot be parsed in anything less than all history) [21:34] jam: it's supposed to identify the branch [21:34] mwhudson: in theory this could also be applied to your valid concern about the branch => ancestry table [21:34] since that is O(branches*ancestry) [21:36] jam, hi? [21:36] hi poolie [21:36] shall we talk? [21:36] are you off DST yet? [21:36] not yet, next weekend [21:36] so it's 7:30 here [21:36] poolie: I have nothing to say to you :) [21:36] sure [21:37] poolie: how about skype? [21:37] sure [21:37] vila: thanks for jumping in on that question [21:38] thumper: I'm ill, so don't hope too much either :-/ I just woke up after falling alseep instead of dining (to give you some idea ;-) [21:39] poolie: I logged in, but I don't see you on skype yet [21:39] likewise [21:39] vila: sorry you're feeling crap, get well soon [21:41] thumper: no worries, I'm rarely ill and it rarely last long :) (crossing fingers, killing some chicken, etc) [21:47] I love qbzr. Not only is it an amazing tool, but I'm repeatedly astonished by how fast bugs are turned around. <4 hours filed->fixed in this case :-) [21:52] maxb: I was doing the release, and it was a easy thing to fix. [21:53] maxb: Thanks for the bug report. [21:54] You deserve the praise - this isn't the first time I've had a bug turnaround measured in hours :-) [21:54] maxb: However, we do have our share of old bugs, and open bug seems to grow all the time :-~ [21:55] So do all projects, I fear [21:55] moin [21:56] moin lifeless [22:12] can someone review my apport mp? [22:17] jelmer: If I want to add an entry to debian/changelog, but I am not going to be uploading, should I just exclude the -- Gary... line? [22:24] GaryvdM: Just include it - whoever will actually do the upload will change it to their name [22:25] jelmer: Ok thanks === salgado is now known as salgado-afk [23:07] Hello everyone. I am still a bit new to bzr and trying to get my head wrapped around how I should best use it. In my case, I have a distributed team work on a base project (PHP website). Most of the work goes into the base product. But, from time to time, we need to branch and create specific version for a client. Most of the changes are cosmetic, CSS, images, etc. Some are code changes. Either way, we would like to be able to track the changes in th [23:11] is there an equivalent to bazaar explorer for Gtk+? [23:13] timClicks: olive, which is a part of bzr-gtk [23:13] GaryvdM: ty [23:15] timClicks: You can also configure bzr explorer to use the bzr-gtk dialogs, but it will still require that you have qt installed (which i guess is what you don't want) [23:16] well, I'm fine mixing and matching but used a Windows machine for checking out some code yesterday [23:16] and was really impressed by Bazaar Explorer [23:16] and wanted to have something that easy for my Ubuntu machine [23:17] timClicks: bzr explorer/qbzr will honner your gnome styles, if thats whats important [23:18] timClicks: see http://doc.bazaar.canonical.com/explorer/en/visual-tour-gnome.html [23:24] * timClicks didn't realise that bazaar explorer was available for GNOME [23:30] timClicks: bzr explorer is built with qt. (qt!=kde) [23:30] * timClicks nods [23:30] understood [23:31] I actually thought it was a native MS Windows app [23:31] Oh [23:31] things make much more sense now [23:38] Hi, how does one start bazaar explorer in OS X? I installed the desktop bundle and qt libraries (things like bzr qinit work ok from a terminal) [23:39] jrib: run bzr explorer [23:40] jrib: there is allready a bug for adding a application shortcut [23:41] GaryvdM: hmm, I'm getting "bzr: ERROR: unkown command "explorer"". I installed 2.0.1 + desktop bundle [23:43] jrib: thats odd. Do you see explorer under bzr plugins ? [23:43] GaryvdM: bzr plugins | grep -i explore returns nothing [23:45] GaryvdM: it is possible that I installed bzr some other way and now installed the bundle on top of it (I don't use OS X and am just trying to get some simple instructions down for a colleague). bzr currently points to /usr/local/bin/bzr [23:45] jrib: sorry - I don't own a mac. Not sure [23:46] GaryvdM: ok, thank you. In any case, it's clear now that if I did have explorer installed it should be listed under bzr plugins, right? [23:46] jrib: yes [23:47] jrib: If you have pyqt installed, then you should be able to just put bzr-explorer in the plugins dir [23:47] Not sure where that would be though [23:52] GaryvdM: thanks, copying it to the plugins folder was sufficient. Now I have to redo it without using a terminal to explain it to my friend :) [23:53] jrib: Not sure why + desktop bundle did not work for you. [23:54] I ran it on a family members mac once, an it worked for me