[00:00] heh, k [00:09] fullermd: Just a question to check that I've understood correctly how bazaar works: [00:10] In svn, when you've to implement a new complex feature you usually create a branch, commit partial changes to it (updating the branch periodically)... And then merge it into the trunk again. [00:11] In bazaar a working copy is in fact a branch... So you just create your workingcopy ("local" branch), make your changes and then commit them into the trunk with the full history [00:11] Right? I think that this is the key of the two systems... Am I right' [00:11] ? [00:12] you can do that. you could also create a branch next to the public "trunk" and work in that, so people can see your changes as they happen (or as you push). [00:14] you can also use looms instead of branches, afaik [00:15] I still haven't played with that [00:15] the key difference is that branches can be anywhere, rather than just all together in some central repository [00:16] although they can share a repository for efficiency sake ;) [00:18] Well, looms are (IMO) more aimed at the case where you have a larger number of layers. [00:19] looms... googling... I never saw this word in the svn world :) [00:19] svn 1.5 has (AIUI) much better merge tracking than previous versions, so using branches like that probably works better there now. But the branches all still have to be in one single repository. [00:20] Well, that's becaues it doesn't exist in the svn world. Or anywhere else, for that matter :p [00:21] It's a method for automating some of the housekeeping involved when you have a stack of branches, for implementing layers of features. [00:21] ah, ok... Because I've not found any usefull information [00:21] But... [00:21] e.g., you have an "upstream" branch, based on which you have a "feature1" branch to add feature1, based on which you have a "feature2" branch that takes everything from feature1 and adds feature2 on top, based on which you have a 'feature3' branch that needs feature2 and adds feature3 on top, based on which... [00:22] As opposed to doing all the above in a "allmyfeatures" branch. It lets you keep them separate, even though each builds on the other. [00:22] You could do the same thing with just having N branches, but loom automates a bit of the bookwork in keeping them stacked and synced. [00:22] ... in bazaar... When you worked on you working copy (that's a branch) with your commits (aka push?)... And than you commit your final work to the main repository... Do all the changeset get transferred to the main repository too? [00:23] yes [00:23] Good, I thought so... But I needed a confirm ;) [00:24] In bzr, you'd do that by merging [the changes in] your local branch into the trunk. [00:24] So the answer is "yes, but your phrasing doesn't really fit what you do with bzr" :) [00:24] morning [00:25] * fullermd waves at igc. [00:25] fullermd: Yeah, I'm not comfortable with bazaar words yet... :D I'm steel a svn supporter for now ;) [00:26] hi fullermd [00:26] epper: Oh, don't worry about THAT; we've got re-education camps to solve that problem... [00:26] lool [00:27] The first step, really, is to work at thinking in terms of branches, rather than repositories. In SVN, the repo is a sharp boundary for a lot of things; in bzr, it doesn't affect much of anything conceptually. [00:28] Conversely, in bzr, a branch if a boundary from inside, and a unit from outside, whereas in svn, it's a weird construct of a conventional subpart of a repository. Wacky. [00:31] Yes, but I think that this way of handling branches would be usefull adding new complex features and removing all the not-so-stable commits that I could find very often in SVN repositories every day. [00:31] Of course, that also means that in svn you can branch a file within a branch, then branch the branch with a branched file in the branch, and then branch the two branches, one a branch of the other, both with a branch of the file in the branch... [00:31] Lost in branches [00:33] But probably staying here only analyzing bazaar without trying it will not reveal other intresting uses of its featuers... sure of that [00:36] Thanks again fullermd, now I've a better view of bazaar... Maybe better than subvesion one day ;) [00:36] It's time to go to sleep here, thanks... and good bye ;) === mw is now known as mw|out === kiko is now known as kiko-zzz === kiko is now known as kiko-zzz [01:40] poolie: did you know that the nightly ppa isn't updating? [01:43] thumper: i did not [01:43] statik runs it [01:44] yeah, I have pinged statik but no answer yet [02:36] thumper: i lost my hard drive in the computer that was running the nightly bzr sourcepackage builds, i've moved things to a slicehost machine and just haven't finished. i'll finish now [02:36] statik: thanks [02:36] statik: I'd also like to talk to you about LP reviews if you have a few minutes [02:36] thumper: np. funny, this is the first time i've signed onto freenode in a long time [02:37] statik: you have been missed :) [02:37] thanks man, i will try to connect to freenode more [02:37] thumper: give me 10 minutes to finish this, then i we can talk code reviews [02:38] statik: ack [03:15] hopefully easy question: is there a way to show all logs relating to any file/directory in a tree of directories? [03:18] poolie, thumper: nightly bzr package build seems to be safely back in cron, and packages should hit the PPA in 15 minutes or so [03:19] w00t [03:20] thanks [03:27] 'bzr diff directory' does show the differences in all files in directory 'bzr log directory' only shows logs from when the directory itself was altered. Can I make bzr log behave like bzr diff? [03:31] Ah I guess I found the answer: https://lists.ubuntu.com/archives/bazaar/2008q3/046146.html [03:35] :( [03:53] :( indeed [03:53] hm, are the milestones not being updated on https://launchpad.net/bzr ? [03:55] rocky, they look like the are [03:56] looks pretty bare https://launchpad.net/bzr/+milestone/1.9.1 [03:56] rocky, 1.9.1 hasn't been released yet [03:56] oh i thought it had [03:56] bah [03:56] sorry [03:57] which branch is active dev happening on now? [03:57] just lp:bzr? when will that get branched as 1.10 ? [03:58] rocky: normally when 1.10 goes into release candidate [03:58] gotcha [04:01] i merged changes from a svn working copy to my bzr branch with the the svn plugin - they are showing up as pending merges - what does that mean? [04:01] dstanek: you still need to do a bzr commit to your local branch [04:02] ah i c [04:02] rocky: yep thanks,that was it [04:05] thumper: does everything that goes into bzr go through a blueprint first? [04:06] no [04:06] rocky: most go through bugs [04:06] rocky: but certainly not all [04:07] blueprints are mostly about more sweeping changes? [04:07] rocky: I'm not sure how much the blueprint feature is used [04:07] for bzr [04:07] * rocky is trying to understand bzr and it's use in and for launchapd [04:07] *launchpad [04:07] thumper: this looks like it's being neglected (which would support your comment) ... https://blueprints.launchpad.net/bzr/+spec/0.10-release-process [04:07] rocky: if you want to talk about launchpad, we could go to #launchpad [04:07] well, this is about how bzr uses launchpad ... so i dunno where i should talk ;) [04:08] :) [04:08] I think bzr uses bugs more than blueprints [04:09] is there a non-automated roadmap someplace i could look at? [04:12] i just realized i was mistaking 0.10 for 1.10 [04:15] rocky: what kind of question specifically? [04:17] i just want to see what's planned for the next major release of bzr [04:22] the best place to look is here: http://doc.bazaar-vcs.org/bzr.dev/en/release-notes/NEWS.html [04:56] beuno: Sorry, but I have no idea how Google found the broken URL. [06:15] spiv, still around, want to talk before I finish up? [07:01] Good morning all [07:14] oh hai [07:14] why haz bzr log not to be having useful (i.e. portable) revision identifiers even with -v? [07:18] elmo: the argument --show-ids is what you want. [07:18] (assuming "revid"s are what you are looking for, but you _rarely_ need them) [07:32] AfC: it's what I'm looking for - I was put off by the "internal use" in the --help output [07:33] AfC: how else should I specify the revision I'm talking about with someone who has the same history, but not the same tree, so revnos don't match? [07:33] if you have a copy of the branch the revnos will be the same up to the point of divergence. [07:34] I don't quite now what I have vs what barry has. it's launchpad, so pqm managed if that makes a difference. all I know is that he has the same history, so I could specify the revid and it worked for him, but his revnos were completely divergent [07:35] sorry, same history probably isn't accurate in VCS speak [07:35] I mean, he has at least all the changes I have, I guess === jszakmeister is now known as jszakmeister|awa [07:42] elmo: if you have a shared mainline, anything on the mainline can be considered stable [07:42] elmo: revid are globally unique identifiers, so you can use them across different branches, if "he" doesn't have "your" revid, "he" misses part of "your" history === jamesh_ is now known as jamesh [07:43] elmo: also, after you've cloned his branch, doing `bzr missing --line ../path/to/his/branch` can be very useful. [07:46] right, sorry for wasting your time [07:46] apparently barry did have the revno after all, and I'm just a muppet (too) and didn't notice him saying so (later) [07:49] Hooray for the Muppets! [07:50] http://www.evula.org/solarsystem/Resources/muppet.jpg [10:04] poolie: ping [13:40] I seem to be having a problem with authentication.conf ; it's not being read at all as far as I can mak eout. [13:40] Hold on, test a little more I wil [13:43] Ok, I'm trying a pull from a smart http server, it's throwing back a 401, and bazaar is just stopping there. [13:44] It's not even trying to access authentication.conf (confirmed by using a filesystem debugger) === mm-mysql__ is now known as mm-mysql === mw|out is now known as mw [14:04] awilkins: At one point, http wouldn't even try auth unless the username were given in the URL. Maybe that's still the case? [14:04] fullermd: The documentation says otherwise, but I'll try it [14:05] Yes, it reads the file and uses the password [14:05] Bah [14:07] Well, that's just NAUGHTY, not doing what the docs say. [14:12] Looks like _SharedConnection() needs to try matching authentication.conf in transport/_init_ === thekorn_ is now known as thekorn === kiko-zzz is now known as kiko === kiko is now known as kiko-fud === evarlast_ is now known as evarlast [15:50] Hi, is trigger plug-in included into bzr? I saw https://lists.ubuntu.com/archives/bazaar/2006q1/007285.html, but it seem to be a little bit obsolete. [15:58] alum: I haven't heard anything about trigger since that 2006 thread [16:08] jam: I'll appreciate it in main bzr. Is there any plans? [16:09] alum: as I haven't heard of it for 2 (almost 3) years, nobody seems to be pushing to get it included. [16:09] If you are interested, you could put together a patch for it [16:09] and we would review it like everything else [16:09] personally, I find the functionality fits very well as a plugin [16:10] I'm willing to be persuaded as to why it should be a core feature, though. [16:15] I need to create a copy of a file, but there's no bzr copy command. What should I use instead? [16:16] cp file new-file [16:16] bzr add new-file [16:16] but will the revision history follow the copied file? [16:16] No [16:16] That's unfortunate [16:16] gotgenes, bzr manages renames, but not copies [16:16] But there is no feature that will make it, alas. [16:17] Why do you need to copy the file? [16:17] It's my CV [16:17] (just to get a better idea of what alternatives may be pertinent) [16:17] I need to pare it down to a resume [16:17] Would a "resume" branch of the main "CV" branch be more useful? [16:18] Mmm, let me think. [16:18] Then changes to CV could be merged to "resume" (but you'd probably get a reasonably ;arge conflict set) [16:18] (CV?) [16:18] probably would get a large conflict set [16:18] Curriculum Vitae [16:19] Commonly used UK (Europe) term for what 'mercans call "Resume" [16:19] awilkins: I suppose I'm probably not going to merge the two ever, being practical and forward thinking. [16:19] I'm presuming it's even in a format that merges well :-) [16:19] Huh, yeah, I've never heard it called that here. [16:19] awilkins: LaTeX [16:20] That would count :-) [16:20] You in the molecular biology field then? [16:20] awilkins: Yeah [16:20] well, bioinformatics [16:20] jam: From my point of view it could be very interested, because I wit it I can automatically re-create source package after commit. This is something I need. Or it can send me notification e-mail with changes. [16:20] but I started in molec. bio. [16:20] I got turned down for a bioinformatics research post once :-) [16:21] awilkins: Where? Hopefully not here. [16:21] Manchester Uni [16:21] (VBI) [16:21] alum: I would look closely at the post-branch-tip-changed plugin hook, or bzr-email [16:21] awilkins: Ah, yeah, they have a big outfit there. [16:21] but it depends on what level you want to be implementing your custom work, I guess [16:21] I didn't have enough propellor-headed C for them [16:22] Loong tme ago [16:22] They probably have more money now [16:22] It seems demand doesn't go down. [16:22] Does LaTeX support the means to hide bits of your CV by applying anything like a stylesheet? [16:22] Too much data. [16:22] awilkins: I think it does actually. [16:23] That's how Beamer works at least. [16:23] I don't use mine enough to do this, but I always think about writing it in XML and using XSLT on it [16:23] But it's not something I know how to do at the moment and being pragmatic, I should at least get it done now, then go back and do it "right" when time's on my side. [16:23] jam: Thanks, I'll look at it [16:23] awilkins: well, considering that Tex is Turing complete (IIRC) I'm pretty sure you could do lots of things with it [16:24] Anyway ; anyone want to comment on https://bugs.launchpad.net/bzr/+bug/300347 [16:24] Yeah, I have no idea what the best format to do it in would be. [16:24] Launchpad bug 300347 in bzr "HTTP transport does not use authentication.conf unless you supply a user name." [Undecided,New] [16:24] alum: I also seem to remember a plugin that extended plugin hooks to allow it to run shell scripts [16:24] vila: ^^ bug 300347 seems right up your allye [16:24] Launchpad bug 300347 in bzr "HTTP transport does not use authentication.conf unless you supply a user name." [Undecided,New] https://launchpad.net/bugs/300347 [16:24] Thought vila was the man for this one [16:25] well he did write the authentication.conf support [16:25] He seems to be the go-to guy for anything transport related [16:26] at first read that sounds really strange... do you use the pycurl or the urllib based http client ? [16:26] urllib [16:27] pycurl causes worse problems for me [16:29] (my other problem is proxy settings - neither of them understand PAC scripts and it seems to be impossible to use environment variables to tell them to NOT use a proxy) [16:30] They take the default IE settings and fail. :-( [16:30] is it more worser [16:32] awilkins: isn't there a "none_proxy" or something equivalent [16:32] to explicitly list hosts that shouldn't be proxied [16:33] jam: Hmm, I'll look through the urllib source [16:33] jam: I was looking for values of HTTP_PROXY like "DIRECT" (which is what PAC scripts return) [16:33] (magic values) [16:35] awilkins: PROXY env vars *are* magic (as in never specified and implemented in gazillion different ways by all browers and their sister) [16:35] looking at urllib.py [16:35] I see it has code to read the magic registry entries on windows [16:35] vila: I got that feeling but I'm only looking for magic that works with Bazaar :-) [16:36] which python version are you using ? [16:36] jam: Yes, it does, it does a reasonable job of reading the IE proxy values, but they are a problem [16:36] vila: 2.5 [16:36] though it has both "getproxies_environment()" and "getproxies_registry()" [16:36] from the look of it [16:36] env vars override the registry [16:36] as long as they have something defined [16:36] jam: ENV overrides the resgitry, but I want NO proxy [16:36] set http_proxy='' [16:37] isn't enough to get that? [16:37] I also see a "getproxy_bypass" [16:37] jam: Problem is that IT services force proxy settings on use via group policy - the server is available direct from both sides of the network so no proxy is required [16:37] which looks at the ProxyEnable and ProxyOverride settings in the registry [16:38] And their proxy setting is a PAC script anyway, so Python can't use it [16:38] Hmm, I've been setting the env variable from the shell, maybe I can set an empty value from the GUI [16:38] awilkins: use no_proxy=host:port [16:39] ProxyOverride seems like something worth exploring [16:40] ProxyOverride="host;" seems like it might work [16:42] awilkins: did your patch fixes *your* problem ? Because it looks a lot like code already in _urllib2_wrappers (in get_user_password) [16:43] vila: Yes, it fixes my specific problem ; I did peruse the code in _urllib2_wrappers before I wrote it [16:43] It's all rather complicated, hence not sending a merge to the list. [16:44] * vila scratching head [16:45] * fullermd scratches vila's head too, for luck. [16:45] awilkins: you can use -Dauth to see which sections are accessed in auth.conf [16:45] With and without patch? [16:45] Without patch I'm afraid it never touched auth.conf [16:46] Can you try, without patch, with -Dauth and -Dhttp and send me the .bzr.log (sanitize appropriately depending of your auth scheme, etc) [16:47] then, some thing but *with* your patch [16:47] s/some/same/ [16:48] bug #300055 [16:48] Launchpad bug 300055 in bzr ""bzr log --forward FILE" crashes for revision range if first revision is not mainline" [Medium,Confirmed] https://launchpad.net/bugs/300055 [16:58] vila: I've attached a log to the bug at http://launchpadlibrarian.net/19813229/auth-conf-http-log.zip [16:59] vila: Watching the file with a filesystem monitor, the process does not attempt to access auth.conf at all during the session. [17:07] And some combination of your "no proxy" tips PLUS me realizing that I'm a moron with the wrong IP in his hosts file has fixed my problem. [17:08] (with proxies) [17:08] I shall have to clean it all up and see which tip is most effective for least effort [17:15] anyone familiar with loggerhead? [17:15] File "./start-loggerhead.py", line 3, in ? [17:15] import pkg_resources [17:15] i can't start the server [17:16] do you have the dependencies installed? [17:17] heh [17:17] which dependencies? [17:17] there's none noted in the docs [17:17] awilkins: thanks for the log, I'll look into fixing the bug asap, keep your patch if it suits you for the time being [17:17] beuno: ^^ ? [17:17] I don't remember the exact ones [17:17] but simple-tal comes to mind [17:17] Kobaz: looks like setuptools... [17:17] (aka: easy_install) [17:17] k [17:18] Kobaz, we have noted them on the README [17:18] you need python-paste and python-simpletal [17:18] # grep -i setuptools README.txt [17:18] # [17:18] It's actually the first thing the readme says ;) [17:19] right, READ the README [17:19] 11 GETTING STARTED [17:19] i did, there's nothing in there about deps [17:19] 12 --------------- [17:19] 13 [17:19] 14 Loggerhead depends on. [17:19] 15 1) SimpleTAL for templating. [17:19] 16 on Ubuntu package `sudo apt-get install python-simpletal` [17:19] 17 or download from http://www.owlfish.com/software/simpleTAL/download.html [17:19] 18 2) Paste for the server. (You need version 1.2 or newer of Paste.). [17:19] 19 on Ubuntu package `sudo apt-get install python-paste` [17:19] 20 or use `easy_install Paste` [17:19] which readme are you reading? [17:19] Kobaz, where did you get loggerhead from? [17:19] https://launchpad.net/loggerhead/+download [17:20] version? [17:20] erm [17:20] 1.2.1 [17:20] right [17:20] I'd say, either get 1.6.1 [17:20] or trunk [17:20] yeah, that is trunk [17:20] which is why i downloaded it, thought it was the stabl [17:21] right, it's confusing [17:21] let me delete those... [17:21] there [17:21] now, get 1.6.1 :) [17:21] beuno: LOL [17:21] heh, k [17:21] or branch from trunk, which is pretty close to a release :) [17:21] sorry about that Kobaz [17:21] hehe [17:22] okay [17:22] now i see the dependencies [17:22] :) [17:23] is there a way i can use it to serve all repos automagically [17:23] this reminds me why we changed *everything* [17:23] instead of configuring each thinger seperatly or spawning a new daemon for each one [17:23] Kobaz, sure, just run "serve-branches", and it will server anything in that dir [17:24] oh okay [17:26] so what was the reasons behind the design decision to make it an independent daemon rather than a series of web scripts and let apache handle the rest [17:27] well, we sort of inhertied the project, so a lot of the decisions came pre-made [17:27] but I think we wouldn't want people to need mod_python [17:27] and we really wanted loggerhead to be standalone [17:28] apache is an overkill in many cases [17:28] you can still use apache though [17:28] yeah [17:28] well it's like... reinventing the wheel versus using an existing codebase [17:29] in what way? [17:29] ie: not having to worry about keeping track of client connections, etc [17:29] right, paste does that for us [17:29] oh [17:29] playu [17:29] er [17:29] off the home row [17:29] ah, okay [17:30] mmm [17:30] it's working [17:31] yay [17:33] :) === kiko-fud is now known as kiko [17:34] is there a nice way to have the daemon write a log file [17:35] instead of dumping everything to stdout [17:35] i can do a redirect to a file, but then you can't rotate the log file without restarting the daemon [17:37] it should log by default as well [17:37] and rotate logs [17:38] hmm [17:38] where does it write the logs [17:39] Kobaz, you can also do --log-folder=LOG_FOLDER [17:39] k [17:40] that's not in --help :P [17:40] it is here! [17:40] not 100% sure if maybe that wasn't released in 1.6.1 [17:40] 1.6.1? [17:40] heh [17:40] so, again, I'd recommend branching trunk [17:40] bzr branch lp:loggerhead [17:41] oh, that's what you mean [17:55] okay so.... i have the apache proxy going [17:55] all my links point to 127.0.0.1 [17:58] * Kobaz pokes beuno [17:58] * beuno pokes Kobaz back [17:59] oh, you weren't here when i wrote [17:59] okay so.... i have the apache proxy going [17:59] all my links point to 127.0.0.1 [17:59] did you install python-pastedeploy? [18:00] hnmm [18:00] apparently not, installing it now [18:00] :) [18:01] okay [18:01] the links work now [18:01] kinda [18:01] i have a https url [18:01] and all the links are the right hostname, but http [18:02] ah [18:03] you may have to tinker with loggerhead itself to get it to work with http [18:03] patches welcome :) [18:03] heh [18:03] k [18:03] so next question [18:04] [18:04] ProxyPass http://127.0.0.1:8082/ [18:04] that setup [18:04] doesn't seem to work, and i get this: [18:04] File does not exist: /home/web/bzr.local/static [18:05] in my error logs [18:05] if i do... [18:05] [18:05] then it works out [18:05] oh wait, is static a dir in loggerhead [18:05] i may need to alias that [18:09] okay, aliasing it worked nicely [18:10] beuno: what chunk of code makes the links [18:13] your code isn't really commented, heh [18:14] Kobaz, off the top of my head, apps/config.py or apps/branch.py [18:15] heh, well, again, it's partially because it was inherited [18:15] yeah i'm looking at apps/branch.py [18:15] i found the url function [18:16] Kobaz: newcomers are the best comment writers ! I'm sure beuno will love your patches :) [18:16] yeah but [18:16] i've never written python in my life [18:16] so this may be a little painful [18:17] Kobaz, python is fun! [18:17] Kobaz: for comments, python is quite simple and pretty much like bash: start with a '#' and add poetry :) [18:17] yeah [18:18] i'm a c/c++/perl/php/bash/tcl/asm guy [18:21] Kobaz: I was a perl guy for ~10 years before trying python, the transition is *not* painful, once you get rig of ';' and the sigils, things are even pretty fun [18:22] heh [18:22] You just have to banish the very thought of "variables" out of your head. [18:22] whitespace dependency has always weirded me out [18:22] And lose a whole load of other concepts while you're at it [18:26] Kobaz: whitespace dependency worried me for a couple of days... until it just stopped worrying me leaving me confused about *why* I was worried ;-) [18:27] beuno: can loggerhead and bzr branches live at the same location? (so that I can branch from the url of what I'm looking at in loggerhead) [18:29] LarstiQ, they can, yes. And, there's a bug open for displaying the "public location" on the web UI [18:29] if you're feeling hackery-ish :) [18:29] maybe tomorrow :) [18:29] * LarstiQ is starting to get hungry, and should head home [18:37] bzr push bzr+https://user:pass@bzr.local/bzr/repo/ [18:37] that works fine [18:37] but then i do a "bzr push" [18:37] Using saved push location: push bzr+https://user:pass@bzr.local/bzr/repo/ [18:37] bzr: ERROR: Generic bzr smart protocol error: Invalid http response Unable to handle http code 401: Authorization Required [18:37] it doesn't save the user/pass === bac is now known as bac_lunch [18:49] so hmm, does noone know how to preserve the login to a repo? === mw is now known as mw|food [19:09]