[00:07] I was just talking to a git user who indicated he was turned off by bzr's branches-as-directories model. I'm curious -- has an alternate approach been introduced since I was last current, or is that still the only branching model from a UI perspective? [00:07] i think its the same === thumper_laptop is now known as thumper === rockstar` is now known as rockstar === rockstar` is now known as rockstar [07:03] nDuff: you can work differently, if you prefer [07:03] you can have a shared repo which has all the revisions for all of your branches [07:04] and "treeless" branches inside that. those are subdirectories, but just include the branch info, so they are basically invisible [07:04] then, you can have a single checkout of one of the branches somewhere, which is where you work [07:04] and you can "switch" that checkout from branch to branch [07:04] when you switch, the working directory gets changed to reflect the tip of the branch you just switched to [07:05] and then when you commit, the new revision goes into the shared repository, and metadata for it is added to the branch you are working on [07:05] does that make sense? shout if i have not explained it well [07:06] but this is one of the beauties of bzr - the pieces are orthogonal, so you can have a workflow that suits you [07:06] much more easily than any of the other dvcs' [07:06] Umm, sabdfl, nDuff posted that abut eight hours ago. It's a pretty good explanation though, and I found it quite useful! [07:07] cammoblammo: thank you! he's still logged in as nDuff so perhaps he'll spot the response later [07:07] ;-) [08:31] hello world [08:32] can someone help me get lost in bzr documentation? [08:34] alternatively, can i edit the bzr wiki w/o some magical bits? [08:37] timely: you can easily do both, i think [08:37] are you wanting to contribute to the docs or the wiki? [08:39] i don't really want to do either. but there are grammatical errors on http://bazaar-vcs.org/Integrating_with_Bazaar so i figure if it doesn't require lots of effort, i'm willing to make a wiki account to make some changes [08:39] i'm trying to write some perl based integration to 'dead bzr trees' [08:39] i.e. a file system copy of a bzr repository (where there may or may not be a valid or useful instance of bzr installed) [08:40] mostly, i need to be able to figure out how to generate links to what i guess is bzr web [08:40] although i have no idea what its formal name is [08:41] probably loggerhead actually, i guess [08:43] nice (503) http://www.lag.net/branches/loggerhead/loggerhead_dev/ [08:43] (empty) http://www.lag.net/robey/code/loggerhead/ [08:43] * timely sighs [08:43] timely: might want to look at the branches on launchpad [08:45] i [08:45] 'm there [08:45] i'm quite lost [08:45] all i want to know is [08:46] given (.bzr/*, path/filename) how to calculate a url for a "log view" pinned to path/filename for the version controlled by .bzr/* and an annotation view for the same [08:46] i do not intend to read 1000 branches of a project just to figure out there's no useful documentation :) [08:49] https://bugs.edge.launchpad.net/launchpad-bazaar/+bug/98826 seems to be what i want [08:49] although i can't figure out if the loggerhead i have available supports it [08:51] certainly the one i'm talking to is generating the other url format... which is amazingly unhelpful :( [08:52] anyway... do i need to do more than simply create a wiki account to edit that first url? [08:53] if not, i'll create an account and change it. otherwise, i'll stop considering [08:53] https://code.launchpad.net/loggerhead it says You can _browse the source code _for ... [08:54] is that part of loggerhead's content or is that launchpad itself being clever? [08:56] timely: the link from code.launchpad.net is generated by Launchpad. [08:56] timely: but maybe I'm misunderstanding the question? [08:57] the space after 'code' is underlined as part of the link [08:57] one doesn't typically do that. it's an error in the html [08:57] timely: right. that's a bug. [08:58] either because someone left a space in raw html, an input field, or in the generator [08:58] but i need to know which :) [08:58] timely: I'm not sure why you need to know. [08:58] i only want to send one report to the right team :), but i will report it :) [08:59] timely: oh I can file the bug. [08:59] ok :) [08:59] timely: I'm on the relevant team :) [09:00] timely: https://bugs.launchpad.net/launchpad-bazaar/+bug/236452 [09:01] :) [09:02] ok, so http://bazaar-vcs.org/Integrating_with_Bazaar Committing Changes [09:02] To commit only certain files, we need to provide a list of filenames which we want committing, eg: [09:03] should be "we want committed" or "we want to commit" [09:04] fwiw, "e.g." should be spelled w/ s (always) [09:04] timely: you should just be able to edit the wiki page. [09:04] timely: and yes I agree with both of your points. [09:05] * timely tries to create an account [09:05] gah [09:07] oh brother [09:07] if i error in changing the wiki page, it clears all my changes and says "please specify a password" [09:07] how helpful... [09:12] * timely tries to figure out what to make of "spread over itself" [09:12] it sounds like a harlot [09:12] distributed between itself (?) [09:15] Removing Files [09:16] You can remove multiple files at once. [09:16] what does "at once" mean? [09:16] does it mean "together" or "immediately"? [09:16] together. [09:16] originally i assumed the latter, but.. thanks [09:16] "at the same time" [09:16] should i take a lock on this url so that i don't collide with anyone else? :) [09:16] keep hitting preview. [09:17] i'm using TextEdit.app because my browser sucks :) [09:17] "This gives us a WorkingTree object, which has various methods spread over itself, and its parent classes MutableTree and Tree" could be rephrased to: [09:17] note that there's an instance of "its" that should be "it's" somewhere nearby (already fixed locally) [09:18] (this page clearly wasn't written by someone who cares about English :) [09:18] "This gives us a WorkingTree object, which has methods provided by itself and by its parent classes: MutableTree and Tree." [09:18] sold [09:19] You can rename one file to a different name [09:20] ... can i rename one file to the same name ? :~( [09:20] "You can change the name of a single file" [09:20] thanks [09:20] sorry, it's a lot easier for me to spot bugs than come up w/ proper fixes :( [09:21] timely: btw, looking at the changelog for the page, lots of changes are made by people who don't speak English as their first language. [09:21] fwiw, the document isn't sure if it's speaking in second or first person plural [09:21] (you can, we) [09:21] if you have a preference ... [09:22] timely: I'd default to "you", but just pick whatever you think is best. [09:22] you sounds fine, i'll do that later [09:22] active voice i presume? [09:22] actually, [09:23] * timely goes back and start attacking we's [09:23] timely: standardizing on voice doesn't bother me so much. who cares as long as it sounds good. [09:23] but sure, active works. [09:24] gah.. note to self, watch out for misuse of a(n) [09:33] will preview tell me if i hit a conflict? :) [09:33] IMPORTANT: This should be avoided wherever possible, as it scales with the [09:33] length of history:: [09:34] "scales" ... alternative suggestion? typically people think of "scales" as a positive thing [09:38] http://bazaar-vcs.org/Integrating_with_Bazaar?action=diff&rev2=34&rev1=33 [09:38] i hope that's not too bad [09:38] sadly it doesn't really help me w/ my original problem... but :) [09:40] eww, loggerhead actually requires its own webserver? [09:43] timely: re 'scales': there's not enough info to suggest a replacement. [09:44] timely: "the time/memory/bandwidth this takes increases proportionally to the length of history." [09:44] timely: I don't know which one. [09:44] timely: re loggerhead: it runs it's own webserver but you don't have to present it to the wold [09:44] world, rather. [09:44] you can use Apache's ProxyPass features. [09:45] context was the performance behavior of "branch.revision_history()" [09:46] can it actually be bandwidth? i.e. could i be talking to a remote bzr? [09:46] increases proportionally is indeed better [09:47] timely: you definitely could be talking to a remote repository. [09:47] timely: branch = Branch.open('http://example.com/some/branch') works just fine. [09:49] so i guess i should copy your entire string... ok, when i'm done figuring out which bzrs i actually have locally, i'll worry about that [09:49] timely: oh that reminds me. [09:50] timely: using Python to interact with Bazaar will be faster, easier and more flexible than using Perl. [09:50] that's nice [09:50] sadly, this is just support for an existing tool [09:50] timely: I just thought I'd mention it :) [09:50] and i'm not going to teach myself python just to integrate with one vcs [09:50] which i will use almost never [09:51] * timely nods [09:51] timely: fwiw, you can use Bzr with most VCS's, if that makes it worth learning more Python [09:51] no [09:51] :) [09:51] for my day job, i use c++/js [09:52] and which vcs there? [09:52] in my free time, i work on a fairly old and relatively stable tool (mxr.mozilla.org) [09:52] well, at times it's using hg, or git, or cvs, or svn, or bzr, or non vcs'd stuff [09:52] i think i have trees from p4 [09:53] and typically the integration hooks aren't things that are really known or usefully knowable from most tools [09:53] since it's "where's your web ui" "how do i ask for blame" "how do i ask for log" [09:53] those are rarely usefully standardized (if ever) [09:53] true [09:53] heck, in a lot of cases there's no link between the vcs and the web ui at all :0 [09:54] as for pull commands, my standard cases don't use simple pulls [09:54] they use complicated (or very complicated) checkout scripts [09:54] sorry... i'm the edge case... i know most people do benefit from such hints... [09:55] ok, quick question folks [09:55] bzr: ERROR: Cannot lock LockDir(http://bazaar.launchpad.net/%7Elibburnia-team/libisofs/mainline/.bzr/repository/lock): Transport operation not possible: http does not support mkdir() [09:55] when I try to push to lp:libisofs [09:55] I also tried lp:///libisofs [09:56] pygi, you need to run: bzr launchpad-login username [09:56] it's defaulting to http as you haven't set up a username :) [09:56] * timely presumes it needs https [09:56] it will switch to bzr+ssh when you have logged in, assuming you've setup an ssh key [09:58] AFAIK, bzr 1.5 does a better job at explaining this. Are you using 1.3, pygi? [09:58] beuno, on this machine, yes :) [09:58] * sabdfl thinks we should push bzr 2.0 out to stable distros when it's baked [09:58] sabdfl, beuno : thanks ;) [09:59] sabdfl, at this rate, it sould be able to get into Lenny, and, well, I presume you'll be able to get it into Hardy as well :) [09:59] pygi: opet se srećemo ;) [09:59] gour, hehe :) [10:00] can i ask questions about loggerhead here? (again, i don't speak python) [10:00] timely, you can try :) [10:00] timely: yeah, this is the best channel for loggerhead questions. [10:00] i read the README and made a loggerhead.conf [10:00] i did not run make because the README made no mention of it [10:00] pygi: kupih samsunga 204...prži sve u 16. stara pržilica mi izderala par cd/dvd-a [10:00] it did say to run start-loggerhead [10:01] File "./start-loggerhead.py", line 3, in ? ... ImportError: No module named pkg_resources [10:01] sudo apt-get install python-pkg-resources ? :) [10:01] i don't own this computer (i never own computers), i did not configure or install bzr (although some random unknown version is installed), there are many random (unspecified/unknown) versions of python here [10:01] no :) [10:02] timely, loggerhead has a set of dependencies which aren't very clearly layed out. You need turbogears and sqlite to start with [10:02] let me check what else [10:02] keep in mind what i said. i don't own this box. that means i'm not and can not be root :) [10:02] timely, ah, that may be a problem then :) [10:02] timely: then ask your admin to pop in here? [10:02] i'm not specifying the linux distribution (it is linux) [10:03] in general when things are installed by the admin, they make things worse, not better :) [10:03] so, if there are instructions for local install, i'll follow them :) [10:03] it's 2am, i'd hope he's sleeping :) [10:04] timely, well, loggerhead is complicated enough to get working with root access, I suspect that with just local install, it may prove to be a challenge [10:05] ... [10:05] we'de really like to trim down dependencies for it, but it currently has a set of performance issues which are more pressing [10:05] note: if i actually could get the admin here, i could just ask him to upgrade his loggerhead install :) [10:06] in which case... why would i bother setting it up in the first place? :) [10:06] also... something that needs root access doesn't exactly inspire trust :) [10:06] beuno, he could install py and all the deps in his home dir [10:06] [timeless@landfill mxr-test]$ which hg [10:06] timely, there is another web interface for bzr: http://bazaar-vcs.org/WebInterface [10:07] /home/timeless/bin/hg [10:07] http://goffredo-baroncelli.homelinux.net/bazaar [10:07] beuno: ... the point of this is to integrate w/ existing things... [10:07] i suppose i could setup that one just to integrate w/ it [10:07] it's basically a plugin for bzr, so it should work pretty much out of the box [10:08] but my target is integrating http://mxr-test.landfill.bugzilla.org/bugzilla-bzr/source/ with http://bzr.everythingsolved.com/bugzilla/trunk/changes [10:09] interesting [10:09] otoh, there's also bugzillamozilla-bzr which doesn't seem to have any web ui [10:09] so i suppose i can simply run WebInterface for that and integrate w/ it first [10:10] timely, there is some work going on in Loggerhead (I've been doing lots of it these past weeks), so it should be looking better very soon [10:10] maybe i should leave an email address and you can ping me when i should look again.. [10:10] i was getting annoyed w/ hg and decided i'd look to see how hard integrating bzr would be [10:11] it doesn't seem promising :) [10:11] timely, well, bzr does have an xml-output plugin, which is commonly used to integrate into different external tools [10:11] * jml is going offline. [10:11] http://mxr-test.landfill.bugzilla.org/mozilla-central/source/xpcom/Makefile.in [10:11] maybe that is worth looking at [10:12] timely: good luck with the integration. [10:12] the integration for hg is basically the "HG Log" and "HG Blame" links at the top [10:12] jml: thanks [10:12] beuno: how many output flavors does the plugin generate? [10:13] so far i've seen 2 different .bzr formats [10:13] and i have no faith in xml flavors [10:13] (i've also seen and support 2 .svn formats) [10:13] timely, it doesn't relate to bzr's format, it works on a higher level [10:13] yes, but have you *changed* the xml it generates? [10:14] and will you change it *again*? [10:14] the wonderful thing about xml is that no one seems to care about not breaking compatibility [10:14] just like internal repo formats [10:14] timely, it hasn't in at least 8 months, and, it's commonly used, so we're *very* careful not to [10:14] 8months? hah [10:14] the projects i work on are 10+years old :) [10:15] things that only change every 8months.... [10:15] that'd set me back years :) [10:15] I've been using it for over a year in integrating into a PHP aplication, and the Eclipse integration uses it too [10:15] but yeah, i understand... and if every month you say it hasn't changed in 8+n months, then eventually that's a good thing :) [10:15] timely, well, the plugin is barely over a year old, so... [10:15] sorry, i'm jaded... [10:15] heh [10:15] you can always *not* upgrade and keep using the same format [10:16] nope [10:16] i don't control what other people use [10:16] i'm a middle person [10:16] if someone tries to use my tool, they'll use it w/ whichever version of other tools their distro provides [10:16] which on average will be *different* from the one i picked [10:16] and no. i don't want someone forking mxr [10:16] there are already way too many forks of lxr, thank you very much :) [10:17] alright, well, what can we do to help you? [10:17] well... do most of these web uis accept the "global repo version" for any file path? [10:18] (minus the ones that seem to require some private id instead of a file path) [10:18] if so, how do i get that? [10:18] i really suspect all i need is that one identifier [10:18] (and a non sucky version of loggerhead, which seems to be something i'll have to sleep on) [10:18] well, it uses file_ids, but it wouldn't be too hard for it to accept paths as well [10:19] which gets fun with history viewing, the path in _this_ version, or in _that_ version? [10:19] https://bugs.edge.launchpad.net/launchpad-bazaar/+bug/98826 [10:19] LarstiQ: why is it fun? [10:19] i only ever have a single version of the repo in front of me [10:20] if you'd like, you can drop me an email (or, to the public bzr list), with what you need, and I can try and fit it in [10:20] LarstiQ, I think he just wants HEAD [10:20] a user says "" " for it" [10:20] i'm not sure it's necessarily head [10:20] although i don't want to look up your definition of HEAD [10:20] but i can assure you that the repo will be in a state so whichever it is, the repo will know [10:20] timely: I don't know how mxr works, does it not do any navigating in history at all? [10:20] correct [10:21] it merely links to someone else who provides that [10:21] all it works on is a flat checkout of a [10:21] how convoluted foopy may be depends on the checkout scripts [10:22] there might be a mix of .hg,.svn,.git,.bzr,CVS,.p4,nothing w/ random inconsistent versions for CVS and SVN [10:22] ugh [10:22] well, not your problem then. [10:22] right... it's not a problem [10:22] as long as i can get the identifier that defines the .bzr state for the directory containing the file [10:22] i'm perfectly happy [10:23] and i believe *that* shouldn't be so hard [10:23] right [10:24] * timely presumes that bzr pull for an unmodified .bzr is functionally equivalent to cvs update for an unmodified CVS [10:25] timely: ignoring weird cvs update corner cases, yeah [10:25] then why didn't it work? ;-) [10:25] "just give me the same state as they have over there, kxthbye" [10:26] [timeless@landfill bugzilla]$ [ -d .bzr ] && bzr pull [10:26] bzr: ERROR: No pull location known or specified. [10:26] timely: I don't know, what is it not doing you expected it to do? [10:26] timely: right, as the error said [10:26] my cwd is the parent of http://mxr-test.landfill.bugzilla.org/bugzilla-bzr/source/.bzr/ [10:26] timely: it doesn't know where to pull from [10:27] man says: [10:27] bzr pull [LOCATION] [10:27] Turn this branch into a mirror of another branch. [10:27] traditionally [] means optional [10:27] am i misreading man? [10:27] timely: it is optional if the pull location is remembered [10:27] http://mxr-test.landfill.bugzilla.org/bugzilla-bzr/source/.bzr/branch/bound [10:27] exists [10:27] and points to what i think is the right answer [10:27] timely: you're using a checkout? [10:27] timely: try 'bzr update' instead then [10:28] ok, that worked [10:28] if you're always using checkouts, stick with update [10:29] timely: just to have mentioned it, although you don't need it, if you do 'bzr pull ' once, it will remember that as the pull location. [10:30] timely: If you want to then once pull from somewhere else, 'bzr pull ' will get data from location2 but still remember location1 as where it should normally pull from [10:30] timely: bzr pull --remember to override that [10:31] so what did i do wrong the first time such that it didn't remember? [10:31] also, does bzr update understand if it's talking to a tty? [10:32] specifically, i don't mind the output at the end where it says what merged, but i don't want the progress indicator since this is being redirected to a file [10:33] timely, adding --quite should do the trick [10:34] --quiet / -q, i found that [10:34] but i kinda expect that to silence everything [10:34] oops [10:34] * timely kicks self [10:34] * timely has a slightly too greedy regexp [10:34] * timely tries to figure out how to handle it [10:35] or you can use BZR_PROGRESS_BAR=tty,dummy,none,dots [10:35] well, I'm off to bed again [10:36] beuno: sleep well [10:36] timely, if you have any remaining questions/requests on loggerhead, don't hesitate to mail the list, or myself (argentina@gmail.com) [10:36] timely: why pull didn't have a pull location at first, you used a checkout instead of a standalone branch, which is a different model [10:36] LarstiQ, thanks, and good to see you around here :) [10:36] timely: if you were to 'bzr branch ' instead of 'bzr checkout ', it would have it's default pull location set [10:37] timely: if you try do pull in a checkout, it will instead run pull on the branch it is bound to [10:37] * timely boggles [10:38] distributed rcs is too complicated for my tiny little mind [10:38] beuno: thanks, [10:39] timely: well, you're mixing centralized and decentralized workflows :) [10:40] well... i'm a readonly consumer [10:40] i should only ever have one upstream, and i just want the latest and greatest [10:40] oh, and i'm a robot, did i mention that? [10:40] thinking is really way too complicated for me :) [10:41] timely: right, so 'bzr checkout ' once, 'bzr update' always after that, should cover your use case perfectly. [10:41] right, which is probably why i used bzr checkout in the first place [10:41] stupid question. would it hurt bzr to suggest "did you mean bzr update" when i do bzr pull? [10:41] if this tool were an opensolaris tool, i suspect it'd have said that [10:42] note: that's a non interactive error, the user should be forced to type the correct thing at a new prompt [10:43] timely: maybe [10:43] in unrelated bits, can i demote a pull to a checkout? [10:44] since i think the second bzr repo i made was a branch instead of a checkout [10:44] timely: detecting the case where you're in a checkout and have no pull location would be doable. But what if the pull location _is_ set? [10:44] timely: you can convert a standalone branch into a checkout and vice versa with bind/unbind [10:44] i'm not sure what that means :) [10:44] (to the former) [10:45] (which doesn't mean i can't parse your English, just that i don't speak bzr) [10:45] timely: that means I'm not sure if suggesting to use update instead of pull is a good thing to try to do. [10:45] no, that part i understood [10:46] what i don't understand is what it means conceptually to have both a checkout and a pull source [10:46] timely: Ah. [10:47] timely: Are you sure you want to know this? It gets rather decentralized :) [10:47] i'll live, if it means i get a chance to convince you to improve the error reporting :) [10:47] all of the work i'm doing these days is decentralized [10:48] ok [10:48] but... i don't drink, so stop if it sounds like you'll force me to drink :) [10:48] timely: First, remembering a pull location or the bound branch is just storing data, that in itself doesn't change the behaviour. [10:49] "which behavior" :) [10:49] timely: right. The behaviour of it beign a standalone branch or a checkout. [10:50] timely: In a standalone branch, your workflow would be: branch once, pull/push after that, respectively to the remembered pull and push locations. [10:51] timely: in a checkout, your workflow would usually be to checkout once, or otherwise bind a standalone branch, and then use update/commit for sending over revisions. [10:51] timely: (obviously, in a standalone branch you also need to commit, but the act of sending it across the wire is decoupled into pull, whereas with a checkout a commit makes a new revision _and_ pushes it out to the master) [10:52] timely: Now, using pull in a checkout is also possible, but it doesn't do the same thing as in a standalone branch. [10:53] timely: pull in a checkout will update the branch it is bound to, as well as the local version === pbor|out is now known as pbor [10:53] unless I've been gone too long and I'm severely mistaken [10:54] timely: as beuno said, I haven't been around in a while [10:54] * timely missed that last bit by beuno [10:54] but ok :) [10:54] anyway [10:56] timely: I'm going to start my day in a couple of minutes, do you have enough information for now? [10:57] so pull in a checkout is kinda like changing branches (but potentially or likely changing repos which isn't something that i care about) [10:58] i think it's a non issue [10:58] anyone doing pull in a plain checkout will get an error the first time [10:58] with a suggestion to do update [10:58] anyone who actively uses pull w/ a url "knows what they're doing" [10:58] and the next time they do a pull, they'll get what they want [10:59] however, anyone doing pull in a checkout w/o a url doesn't know what they're doing [10:59] I'm not so sure. [10:59] (i.e. me) and just wants an error w/ a suggestion [10:59] timely: it's perfectly legitimate to run pull without any arguments in a checkout. [10:59] but it errors! [11:00] that's what started this conversation :) [11:00] timely: only in this very specific case [11:01] timely: And as I said, that can be detected and warned about. But I fear there are more cases where someone doesn't know what they're doing and we can not detect it. [11:01] personally, my general view is that you should add warnings/errors where possible [11:01] timely: but yeah, I don't know. [11:02] instead of throwing up your hands and saying "we can't be perfect" [11:03] * LarstiQ can't think this through right now. [11:03] bzr pull [LOCATION] [11:03] --remember Remember the specified location as a [11:03] default. [11:03] timely: but feel free to suggest it to someone who has actually been working on bzr lately :P [11:03] English nit: "how many defaults can bzr have for a single repo"? [11:03] "a default" ^ [11:04] timely: unlimited [11:04] but I see what you're getting at. [11:04] good. because i really don't want to hear the explanation for your answer :) [11:05] there is no limited to the amount of branches in a repository ;P [11:05] I do think you meant to say branch, and in that case, it's not unlimited. [11:05] vcs lingo yay [11:05] i didn't say branch [11:06] right [11:06] oh [11:06] timely: and the default is on a per branch basis. [11:06] bah. i was trying to be so careful [11:06] i should have said .bzr/ [11:06] would that have been any more or less ambiguous? [11:07] well, there is {.bzr/repository,foo/.bzr/branch} etc [11:07] timely: more ambiguous [11:07] gah. [11:07] timely: You find a .bzr in 3 locations: repositories, branches, and working trees. [11:07] i'll settle for you understanding what i meant and praying you'll find someone who can fix the docs to be less strange :) [11:08] you find 'a default' strange? [11:08] yes [11:08] I'm not native enough to do so. [11:08] normally you'd write "the default" [11:09] if you don't mean "the default" then you should write "the default for " [11:09] right [11:09]