=== Meths_ is now known as Meths [02:33] Let's say I make commits (from Base) C1, C2, and C3 and submit them each to an upstream project as a patch. I realize there's an error in C1 and make a fourth commit, C4, to fix it. How can I generate a new patch that includes the changes of C1 and C4 without C2 and C3? [02:33] Do I just have to back everything else out and rebase? [02:35] commit it in a branch just for c1+c4; merge that to trunk, send [c1,c4] as the patch [02:36] Ugh. Making a non-stacked branch takes forever. [02:36] Thanks. [05:01] so I'd like to checkout a copy of an svn repo, work on it a bunch pushing changes to a lp bzr repo, and when I'm done in a few weeks merge back into svn [05:02] http://doc.bazaar.canonical.com/latest/en/user-guide/svn_plugin.html seems close but bzr push sends to svn when I would want it to my lp repo for now [05:03] maybe I should just do an svn checkout then bzr init over the top of the same directories [05:03] using bzr to manage my branch work and when done use svn to merge it back into the svn repo? [05:03] guess that would lose the commit history [05:05] and make it hard to merge [05:05] wildintellect: change your push target while you work on it [05:05] I think I wasn't reading that page close enough [05:06] can I have different push locations for each bzr branch I make? [05:06] wildintellect: check the docs for push_location and submit_branch. (parent_location should be the svn repo) [07:55] hi [07:55] Is there a way to track file permissions with bzr ? [07:55] (I know it's "metadata", but still) [07:58] sobersabre: not built-in. There are tools like etckeeper though. [08:38] spiv: hmmm... nice tool. does it source current permissions into some kind of text file and then processes it to ensure the perms are intact ? [08:44] sobersabre: yep, basically [09:09] jelmer ok. [10:26] is bzr able to perform cross project merges? [10:51] ok, it can [12:56] ed [13:05] jml: old school! [13:08] heh [13:08] I have no idea why I typed that in this channel. [16:22] hi. where I can read about how to do something like bzr status from within python script. [16:22] I want to detect changes/additions/removals. [16:23] ther must be some kind of api aournd. right ? [16:24] sobersabre, yeah [16:24] let me dig that up for you [16:26] sobersabre, http://wiki.bazaar.canonical.com/Integrating_with_Bazaar [16:26] beuno: I am on windows. I have bzr installed, I tried import bzrlib, and it didn't manage. [16:26] has nice examples [16:26] and this is the api documentation: http://people.canonical.com/~mwh/bzrlibapi/ [16:27] I don't know anything about python in windows, so can't help you there [16:30] sobersabre: you want to install the bazaar for your python version, not the all-in-one bundle [16:30] that will put it in your site-packages dir so you can import it. [16:32] ie 'bzr-2.2.3.win32-py2.6.exe' not 'bzr-2.2.3-setup.exe' [16:34] mgz: thanks, man. I did install the right version for "my" python (2.5) [16:34] for some reason bzrlib is not loadable. [16:35] does C:/Python25/Lib/site-packages/bzrlib exist? (adapt for your pythong location) [16:35] ...I type 'pythong' way too much. [16:37] I wear pythongs once in a while. [16:37] anyway, I have no bzrlib in the site-packages dir. [16:38] even though I did install bzr-2.2.1.win32-py2.5.exe [16:39] uninstall and try again, it's possible something got screwed up. [16:39] sobersabre, Eclipse problems? [16:39] awilkins: what do u mean? [16:39] Ah, wrong distro, sorry [16:39] it is an evening, the sun is down... [16:39] moon is visible. [16:40] The exe bundle has some issues with the bzr-xmloutput plugin and hence the bzr-eclipse plugin [16:40] awilkins: hm... distro ? [16:40] pay attention in the installer to where it's putting the files. [16:40] oh! [16:40] hm... [16:40] I do have that plugin (but not on this computer). [16:40] is it possible I will have eclipse plugin punishment, [16:40] but not on the computer it is installed on ? [16:40] you might have a zombie python install somewhere, or a screwed up registry setting. [16:41] :-/ [16:41] mgz any ideas what setting ? [16:41] path ? [16:41] I have at least one bug for the Eclipse plugin, but it only occurs when your project folders are nested more than 1 deeper than your workspace [16:41] python* ? [16:41] well, just try uninstall/reinstall to start with [16:41] ok... it's scary. [16:41] that does a "find where python is" step, if it gets it wrong, then we can go digging [16:48] mgz: I have just rebooted. what step were you talking about? [16:49] off a webpage or something ? [16:49] uninstall your current bzr. then reinstall. may as well use 2.2.3 while you're at it. [16:50] beuno: about windows, as you can see "Me and you are of the same blood" [16:50] and look at what paths it's using in the process. [16:50] mgz: I didn't see 2.2.3 for py 2.5 on win the last time I installed... [16:50] it's new-ish. [16:50] http://launchpad.net/bzr/2.2/2.2.3/+download/bzr-2.2.3.win32-py2.5.exe [16:52] alrightie. [16:52] I used the main canonical page last time... [16:52] so where can I look at to see how to answer your 'find where python is' q ? [16:53] I know where the python is. [16:53] it's in D:\Python25 [16:54] the installer doesn't ask you, it looks for itself. [16:54] hm, now I can practically SEE the files are being copied and compiled to D:\Python25\Lib\site-packages\bzrlib [16:54] feels good. [16:54] let's see it works too. [16:54] make sure it's beliefs and yours coincide [16:55] okay, that sounds good. [16:55] -' dammit. [17:02] sobersabre: import works now? [17:02] mgz: oh, it does! [17:02] I like. [17:02] :) [17:03] I am one happy geek now. [17:03] preparing a deployment fab. [17:03] it's my 2nd month with python at work, and I like it. [17:03] too much. [17:05] so, to get started, you want something like: [17:05] from bzrlib.workingtree import WorkingTree [17:05] tree = WorkingTree.open_containing(directory)[0] [17:05] mgz I don't do anything like that :) [17:05] I prefer to use the long names. [17:06] then you can inspect the working tree, and see bzrlib.status for what the ui does. [17:06] mgz I think it is better to use changes. it gives an iterable thing. [17:07] I will count what I want. if the counter is > 0 it's the time to act. [17:07] Any known issues with bzr doing a checkout over sftp with a password protected rsa key? [17:07] basically there's: changes.has_changed [17:07] :) [17:08] if that's all you need then fine, status does some more work on top of that as well. [17:08] mgz: the problem with status is its result. [17:08] I don't want to print out something. [17:08] I need to do or not do something depending on the changes. [17:09] I was suggesting you read the code, not used it as-is. [17:09] iter_changes may be all you need. [17:10] AdamDV: as in, does it work, or as in, are there known security issues? [17:10] if you just want to know if it works, try it and see. [17:11] Well, right now its looking like it does not. [17:11] Even though I've used *without* a password protected key before. [17:11] Let me pastey the errors. [17:12] AdamDV: I think you should be more asking about paramyko. I remember it did have some troubles with default values long ago. [17:12] paramyko is the module for ssh communication. [17:12] bzr is just using it. [17:12] (unless the same people who develop bzr also work on paramyko) [17:13] AdamDV: did you look at the code of paramyko you're running ? [17:13] http://pastebin.com/pvn4p8X3 [17:13] I did not? [17:13] it works fine connecting to ssh. [17:13] BUt yes, it looks like a paramiko problem. [17:14] And if sftp+password key arent going to work, what other options do I have? Use a password-less key? [17:14] AdamDV: also, what kind of network media are you on with this operation ? [17:15] network media? [17:15] As in how I'm connecting? [17:15] Ethernet, WiFi, analog modem ? [17:15] I'm on WiFi in Toronto, server is on ethernet in Dallas. [17:15] AdamDV: what OS ? [17:16] I'm on Maverick, servers on Lucid [17:16] ok. [17:16] 10.04.1 is the server, to be specific. [17:16] on the client: does ifconfig show you any errors ? [17:16] Nope. [17:16] TX/RX ... [17:16] ? [17:16] sure ? [17:17] I was using this yesterday, without a password on the key. We re-imaged the server and I re-generated my key and added a password, now its not working :) [17:17] bug 647916 is similar, check the last comment there? [17:17] Lemme pastebin ifconfig [17:17] Launchpad bug 647916 in Bazaar "Push command fails with "Garbage packet received" error" [Undecided,Incomplete] https://launchpad.net/bugs/647916 [17:17] http://pastebin.com/2gh1VPsq [17:18] mgz: I looked at it... [17:18] Oh. [17:18] That could be it. [17:18] My shell is not bash, its actually a custom internal shell we use here. [17:19] *changes back to bash* [17:19] sounds likely. [17:19] AdamDV: I would not use anything non-standard to debug such problem. [17:19] and, permission denied is kinda interesting. [17:20] Are you sure the user you're using to run this has access to /home/adam/.ssh/id_rsa ? [17:20] AdamDV: is it possible it is chmodded 600, but belongs to ... a different user ? [17:21] Changing to bash on my server user fixes the problem. [17:21] AdamDV: is it possible your "supershell" uses the same UID you have as user 'adam' ? [17:21] I mean it DOESN'T use it. [17:22] As in server user id is the same as my local user id? [17:22] Changing to bash on my server user fixes the problem. <- good good. [17:22] It sounds to me that AdamDV's custom shell is not supporting the invocation of sftp on the server properly [17:22] Yea. Its a 100 line shell written in PHP for the secretarys to use. [17:22] maxb: I know how this can happen: [17:22] * maxb runs away screaming [17:22] the "supershell" prints out too many lines of output. [17:23] I mean too many chars of output. [17:23] maxb: Yea. [17:23] AdamDV: can you make your logon shell quieter ? [17:23] ? [17:24] I'd rather use BASH anyway :) [17:24] AdamDV: if you can - do, if you cannot - make sure the shell shuts up upon invokation. [17:24] I remember having added lots of stats upon logon (with colors, etc.) this made sftp inoperable. [17:24] I have no control over the development of our shell, unfortunatly. If I did, it wouldn't exist.. [17:25] I'm off, mgz thank you! [17:25] Thank you for your help mgz and sobersabre [17:25] :) [17:26] bye both. [17:29] * AdamDV idles === r0bby is now known as robbyoconnor === tchan1 is now known as tchan [19:56] okay, so I may be in for a heck of a pickle... I have some fairly old Bzr repositories which where converted from TLA using baz-import .. a while ago.. [19:56] was trying to work at upgrading them, and a bzr log in them claims that it can not locate an ARCH-1 revision near the end of the log [19:57] http://dpaste.com/370125/ <- any idea why this happens when I try to pull a branch? [19:57] major: what version of bzr are you using? and what's the exact error message? [19:58] MTecknology: the local branch doesn't support a particular feature that the remote branch uses [19:58] MTecknology: you can upgrade the local branch and try again [19:58] the error message could admittedly use some work.. [19:58] oh.. thanks :) [19:58] works perfect now [19:59] bzr version 2.2.2 [19:59] bzr: ERROR: Revision {Arch-1:email@domain--project%package--1.0^Cversion-0} not present in "KnitRepository('file:///home/user/devel/user.bzr/repository/')". [19:59] erm [19:59] erm [19:59] * major sighs. [19:59] that ^C shouldn't be there [20:00] it is has '--' in that spot [20:00] I tried to hide the email addresses and accidentally mangled the error a touch [20:01] major: was the revision present originally? [20:01] missing revisions ("ghosts") were pretty common in baz [20:01] well, it didn't used to give this error... [20:01] and the original repositories where converted from TLA [20:02] I suspect they where external archives.. [20:02] the original repository (the bzr one converted from baz) doesn't give that error? [20:03] they where all TLA archives, and still are [20:03] I still have them [20:03] they where retained for historical purposes [20:03] baz was only used during baz-import, back when it was still part of the bzrtools [20:04] it'd be interesting to see if that revision was actually present in baz [20:04] or if it wasn't present and older baz/bzr simply didn't mention it [20:04] well, I can definately look at it [20:04] and it is definately there [20:05] it just isn't actually in the bzr repository [20:14] is there any documented way to fix ghosts like this? [20:17] major: baz-import shouldn't have created a ghost if the revision was present in the original baz repository [20:19] what if the revision was in a separate tla archive? [20:19] though its location was registered [20:19] major: no idea (I've never used baz) [20:20] hmm [20:20] it will create ghosts [20:20] it doesn't traverse archives [20:20] you need to import each archive [20:21] if I'm remembering correctly [20:27] hmm.. [20:27] I suppose that feels mostly right.. [20:27] just makes this pretty.. well .. funky [20:28] I don't suppose importing the parent repository ontop of the already converted repository will make it "just work" ? [20:29] major: importing it elsewhere and then running "bzr fetch-ghosts " from the original repo should work [20:29] where new-repo is the bzr repo created from the parent baz repo [20:30] nifty [20:31] emphasis on /should/, I've never tried that before [20:32] major: yes, it should just work. [20:32] emphasis on the should [20:32] heh [20:32] well, I have a number of backups of this stuff already [20:32] we haven't done much tla imports for about 5 years :) [20:32] yah .. [20:33] trying to revive a project that hasn't been touched in almost that long [20:41] cool [20:44] does bazaar have the same issues as mercurial with tagging.. that a tag actually does an invisible commit, and that you can't edit history easily of past revisions ? [20:50] Ramosa: a tag doesn't do an invisible commit, it's just a named pointer to a particular revision [20:51] Ramosa: history can't be edited, but you can rewrite it [20:51] well .. it is chugging away at this, hopefully it all works... [20:52] will likely take all day to do all these repositories though [20:52] but that is better than losing the history [20:54] whoa.. that's really slow :-( [20:54] well, a lot of repositories to fix, with a lot of branching inside of them [20:58] I suspect the SCM history alone could be used for an interesting white paper on development work-flow [20:59] hehe [20:59] the original sources started off in RCS, later converted to CVS, then pulled into Arch via cscvs (a tool to compute change-sets within CVS) [21:00] from there a number of developers created branches from the original Arch archives, and then those where converted to bzr via baz-import [21:00] and a few developers developed from multiple locations, which in Arch gives them unique global identifiers [21:01] really not certain if it could get much more chaotic === Ursinha-afk is now known as Ursinha [21:43] hello all [21:44] g'morning poolie [21:44] hi there jelmer [21:51] How can I write a revisionspec that means 'since this branch was forked from its parent'? [21:51] (And I mean without looking that point up beforehand, of course.) [21:52] Ah, ancestor:, maybe. [21:52] quotemstr: yeah, generally ancestor: [21:53] But I still hae to type out the parent branch in that case. Ahwell. [21:54] Hrm. bzr log -r ancestor:./../trunk.. seems to not be pegging the CPU and not making any forward progress [22:04] jelmer: lp:bzr-hg seems to have a pdb.set_trace() in it [22:08] is there a bzr concepts/internals guide? [22:40] mwhudson: oh, whoops [22:40] mwhudson: I'm about to land a branch that fixes 75% of the hg imports [22:41] jelmer: \o/ [22:43] quotemstr: ancestor::parent [22:43] * jelmer waves to spiv [22:44] Good morning :) [22:44] Thanks. [22:44] So ':parent' is magical. [22:48] it's a location alias i think the name is [22:48] there are a few [22:48] :push and :submit are others [22:53] spiv: That doesn't DTRT at all. [22:54] It gives me some random commit from today. [23:00] quotemstr: perhaps :parent isn't what you expect? Check 'bzr info' [23:00] Actually, after bzr log -r ancestor::parent.., I get "bzr: ERROR: Start revision not found in left-hand history of end revision. [23:00] parent is set correctly. === spiv changed the topic of #bzr to: Bazaar version control | try https://answers.launchpad.net/bzr for more help | http://irclogs.ubuntu.com/ | Patch pilot: spiv | 2.2.3 is officially out ! (rm vila) [23:01] Oh, log is fussy like that. abentley added some new revision specs that help with that. [23:02] Wuh? Why *wouldn't* that work? [23:02] And why do you need new syntax to make it work? [23:02] mainline: is the revspec I'm thinking of. [23:03] I agree, I think log should cope, but currently it doesn't. [23:05] For non-trivial history exploration I tend to use 'bzr qlog' (from the qbzr plugin) or 'bzr viz' (similar thing in the bzr-gtk plugin). [23:07] Ah. Anything native for OS X or Win32? [23:07] (Well, I can use X with both. It's just a pain.) [23:07] qbzr looks pretty slick on Win32 at least (it and bzr-explorer and some other plugins are bundled in the bzr installer for that platform) [23:08] IIRC it works ok on OS X too, although there may be some fiddly dependencies that I'm ignorant of. [23:09] Thanks. I'll take a look. [23:09] I'm leery of GUIs generally though. [23:17] use a good solvent [23:29] we does my checkout of lp:loggerhead whinge about not being able to break locks when i run bzr up? [23:29] i suspect the fact that it appears to be making three connections to bazaar.launchpad.net is related [23:30] ugh [23:30] what's so special about it that it needs 3 connections? [23:30] don't know [23:30] the working tree is out of date with the local copy of the branch now [23:31] hm, unbind/bind fixed it [23:31] mwhudson: :( [23:35] mwhudson: note that I'm about to push 1.18 to trunk, or something like that [23:36] how can I "fork" my project at a particular revnum? I tried branch but then why I try to commit changes I get an error "bound branch ... is out of date". Basically what I want is a copy-paste of everything at a chosen revnum, so that it can stay independent. [23:37] monkey_d_luffy: "bzr branch -r REVNO" [23:37] but that didn't work... [23:38] How did it go wrong? [23:38] that's the problem I'm having now [23:39] Your error says "bound branch" though, which isn't something that "bzr branch" (by itself) could give. [23:40] Did you use "bzr checkout" instead, or perhaps "bzr bind" or "bzr reconfigure" after "bzr branch"? [23:40] I did a branch of my project bzr branch -r 60 trunk trunk_new then I renamed both directories. And now when I try to do a commit in my new dir, I get that error [23:41] it could have been a checkout, but I'm pretty sure it was a branch... will try again then [23:41] Hmm. Well, you should be able to "bzr unbind" (or "bzr reconfigure --tree") to change it. [23:41] And "bzr info" can tell you what it is now. [23:42] I did a bzr info now... it was a checkout after all :| [23:46] so a branch makes it completely independent from the parent branch (which may even be deleted) correct? [23:50] Yes. [23:51] ok thansk