Apr 25 22:45:10 * Now talking on #git Apr 25 22:45:10 * Topic for #git is: 1.7.1 | Homepage: http://git-scm.com | Everyone asleep or clueless? Try git@vger.kernel.org | Git User's Survey 2009! http://git.or.cz/gitwiki/GitSurvey2009 | Channel log http://tinyurl.com/gitlog | Mailing list archives: http://tinyurl.com/gitml | Gits on git: http://tinyurl.com/gittalks | Pastebin: http://gist.github.com/ Apr 25 22:45:10 * Topic for #git set by wereHamster!~tomc@caurea.org at Sat Apr 24 10:10:07 2010 Apr 25 22:45:25 ah, i know where the changes came from... Apr 25 22:45:39 Hello. Could someone tell me if there's a way to tweak parameters to passed upload-pack in gitosis.conf? Apr 25 22:47:11 what kind of tweaking do you want? Apr 25 22:48:13 * urkud (~urkud@212.192.251.49) has joined #git Apr 25 22:48:43 * tzafrir_laptop (~tzafrir@bzq-218-155-145.cablep.bezeqint.net) has joined #git Apr 25 22:48:54 sitaram, want to add a parameter Apr 25 22:49:07 * beni- has quit (Quit: beni-) Apr 25 22:49:49 yeah but what? what would it do? I'm not saying I have an answer to your question but I'm curious... Apr 25 22:49:51 sitaram, http://lists.zerezo.com/git/msg711059.html Apr 25 22:49:57 looking Apr 25 22:50:10 Idea is to be able to pass a pre-/post- hook to upload-pack over the command line Apr 25 22:51:51 even if you found a way to pass it, how do you know the other side will honour it? Apr 25 22:52:07 sitaram: The other side will need to force it (if needed). Apr 25 22:52:09 sitaram, the other side? The hooks will be on the server. Apr 25 22:52:47 * walf443 (~walf443@KD125054058069.ppp-bb.dion.ne.jp) has joined #git Apr 25 22:53:31 Ford_Prefect: I meant the "git-upload-pack" program... I assume you've patched that to execute this new hook, as that thread seems to be saying Apr 25 22:53:45 sitaram, yep. Apr 25 22:54:04 Ford_Prefect: still, from an end user point, what is the purpose? logging? (/me reads slow... still following that thread up the chain) Apr 25 22:54:25 sitaram: Rejecting clones but allowing fetching. Apr 25 22:54:26 * disturb (~user@220.208.94-79.rev.gaoland.net) has joined #git Apr 25 22:54:40 * kkkonrad (~kkkonrad@91-115-69-15.adsl.highway.telekom.at) has joined #git Apr 25 22:54:42 git branch testbranch; git checkout testbranch; touch blabla; git commit -am 'sdfsd'; git push - why does this say: 5d098e7..2281d77 master -> master does it mean that my local branch changes were commited to master? Apr 25 22:54:49 * SlickT10 has quit (Ping timeout: 248 seconds) Apr 25 22:54:53 sitaram: Cloning would be handled by bundles (much lighter on server). Apr 25 22:54:53 ianmcorvidae icwiener idoru iff iizukaway ijuma ijuma_ Ilari ilogger2 imarcusthis impulse9 Ingmar IRSeekBot isf_ ivan Apr 25 22:55:04 metalfan_ - that's what a push does Apr 25 22:55:11 sid3k Sigma simNIX siprbaum sitaram sixteneighty Apr 25 22:55:30 oh - a push pushes all branches Apr 25 22:55:38 perhaps you also had changes in master Apr 25 22:55:47 RandalSchwartz, but i havent merge my testbranch with my master branch on the client side? Apr 25 22:56:00 Ilari: saw that gentoo thing where they dont want initial clone... ok, I get it Apr 25 22:56:01 sure - but you must've had some changes in master Apr 25 22:56:03 metalfan__: You had something on local master that wasn't on remote master. Apr 25 22:56:13 git log 5d09..2281 Apr 25 22:56:23 * sagsousuke (~user@187.57.52.152) has joined #git Apr 25 22:56:34 how to display contents of the git index file in human readable format (something like git cat-file ... tree sha etc)? Apr 25 22:56:47 Ilari, RandalSchwartz thx Apr 25 22:57:01 so the branches never reach the server if i dont merge? Apr 25 22:57:04 * tro has quit (Read error: Connection timed out) Apr 25 22:57:07 I've never thought of the index as a "file" Apr 25 22:57:18 the branches reach the server only when you push them Apr 25 22:57:20 * AkumaStreak has quit (Quit: AkumaStreak) Apr 25 22:57:26 it has nothing to do with merging Apr 25 22:57:33 sitaram: The basic issue is that rejecting clones but allowing fetching would need pre-upload hook, but upload hooks are too dangerous for general use. Thus there are special options to enable them. Apr 25 22:57:54 sitaram: And if you want to restrict fetching, of course you want to force pre-upload hook. Apr 25 22:58:43 * Octalot has quit (Remote host closed the connection) Apr 25 22:58:58 RandalSchwartz, do i need a entry in .git/config like this one: [branch "master"] remote = origin merge = refs/heads/master to push my branch? Apr 25 22:59:02 Ilari: does this have an impact on what we talked about many months ago? restricting read at the branch level. I assume no change, because that's really at a different level (symbolic ref versus real) Apr 25 22:59:41 * adamholt is now known as adamholt_away Apr 25 22:59:46 * adamholt_away is now known as adamholt Apr 25 22:59:52 sitaram: I don't think it would allow restricting that (but that was just a guess). At least it would allow rejecting fetch request. Apr 25 23:00:13 Ford_Prefect: Does the hook allow restricting what is advertised or is it just allow/deny? Apr 25 23:00:15 RandalSchwartz: was this reply intended for me?. if yes, pls feel free to simply ignore the word "file" from my question :-) since some git commands allow to specify an alternate index "file" i think you should not be too strict with me :) Apr 25 23:00:33 Ilari, just allow/deny Apr 25 23:00:45 * tro (trotek@gateway/shell/xzibition.com/x-ljvoxchlmopfmmce) has joined #git Apr 25 23:00:51 Ford_Prefect: Ah, good. There are subtle issues with restricting access. Apr 25 23:00:54 kkkonrad - yes Apr 25 23:00:58 ianmcorvidae icwiener idoru iff iizukaway ijuma ijuma_ Ilari ilogger2 imarcusthis impulse9 Ingmar IRSeekBot isf_ ivan Apr 25 23:01:02 where have you seen index "file"? Apr 25 23:01:13 I hope not in the official docs Apr 25 23:01:19 * engla (~ulrik@90-229-236-112-no153.tbcn.telia.com) has joined #git Apr 25 23:01:31 * engla has quit (Changing host) Apr 25 23:01:32 * engla (~ulrik@wikipedia/Sverdrup) has joined #git Apr 25 23:02:05 kkkonrad: Anyway, $GIT_INDEX_FILE specifies the file to read the index from. And ls-files is the command to dump what index contains (use show or cat-file for contents of files staged). Apr 25 23:02:48 kkkonrad: The default place to read the index file is file 'index' in GIT_DIR Apr 25 23:02:55 RandalSchwartz: let mee poke around in the docs.....ok Ilari seems to have answered my eistopn already. Apr 25 23:02:57 * munichpython (~munichlin@117.195.134.229) has joined #git Apr 25 23:03:09 Ilari: thanks. this anwered my question Apr 25 23:03:10 * j416 (~j416@unaffiliated/j416) has joined #git Apr 25 23:03:10 * pll` (~user@c-24-91-252-122.hsd1.ma.comcast.net) has joined #git Apr 25 23:03:26 s/eistopn/question Apr 25 23:03:35 * pll has quit (Read error: Connection reset by peer) Apr 25 23:04:40 * project2501a (~project25@86.51.114.108) has joined #git Apr 25 23:05:37 * afrasi has quit (Remote host closed the connection) Apr 25 23:05:51 * tro has quit (Read error: Operation timed out) Apr 25 23:06:22 Ilari, sitaram: looks like both gitosis and gitolite will require some patching to modify the upload-pack command line (a config argument, is what I'm thinking) Apr 25 23:06:32 Or am I missing something? Apr 25 23:06:52 * tro (trotek@gateway/shell/xzibition.com/x-ogubacqsvqjjjnoq) has joined #git Apr 25 23:07:16 Ford_Prefect: AFAIK, you are correct. But I say that with Gitosis pretty much unmaintained... Apr 25 23:07:42 Yeah, we're probably moving to gitolite when we get enough time, in Gentoo Apr 25 23:08:18 I'll need to hack something together for either anyway so that I can at least test my changes. Apr 25 23:08:23 Ford_Prefect: possible, if it is a command line argument on the "git-(upload|receive)-pack" command rather than an in-protocol thing. But it's likely to be an easy change... Apr 25 23:08:45 Ilari: git ls-files -s seems to be my friend so far Apr 25 23:08:48 Ford_Prefect: if you ignore gitosis/gitolite, how do you make normal git send this extra parameter?? Apr 25 23:09:50 sitaram, haven't really looked at that option. Only way I can think of is to have some sort of global, root-defined config to allow insecure hooks, and then doing it the way it's done in the patches I submitted. Apr 25 23:09:52 * mefesto (~mefesto@145.160.121.70.cfl.res.rr.com) has joined #git Apr 25 23:10:01 That's not too pretty either, tbh Apr 25 23:10:19 * tvw (~tv@e182073215.adsl.alicedsl.de) has joined #git Apr 25 23:10:31 Both system config and user config can be trusted. Repo config can't. Apr 25 23:11:00 Ah, right Apr 25 23:11:13 * Mage_Dude (~chatzilla@c-24-21-220-30.hsd1.or.comcast.net) has joined #git Apr 25 23:11:29 Can we distinguish between those in the code? Didn't see a way too when I saw the code, but I might have missed it. Apr 25 23:11:39 Ford_Prefect: I may not have the smarts to grok patches to git.git code, but when you get a chance let me know how a normal git client (patched if needed of course) deals with sending this to the server... Apr 25 23:11:52 The problem is, exsting config parse code parses all three... Apr 25 23:12:13 * EricInBNE has quit (Remote host closed the connection) Apr 25 23:12:15 * onigiri_ has quit (Quit: onigiri_) Apr 25 23:13:36 Worse, that option would have to be ssh-specific. git:// can't relay it (and neither can gits:// by virtue of being derived from git://). But then, if you use those, you are going to get it forced anyway if there are such hooks... Apr 25 23:13:38 * sixteneighty has quit (Remote host closed the connection) Apr 25 23:14:33 git:// doesn't look at system config? Apr 25 23:14:51 I think he meant does not pass any extra options? Apr 25 23:14:58 Ford_Prefect: git:// is quite restricted in what it can send. Apr 25 23:15:09 * coppro (~coppro@unaffiliated/coppro) has joined #git Apr 25 23:15:13 * sitaram is still confused who is passing what to whom and controlling what :) but he'll get there by and byb Apr 25 23:15:16 by* Apr 25 23:15:45 Ideally, the client has no control over it, and only the system admin controls it on the server Apr 25 23:15:49 * ajpiano has quit (Remote host closed the connection) Apr 25 23:16:06 Ford_Prefect: in that case why would gitolite/gitosis change at all? Apr 25 23:16:08 Or the hosting admin... Apr 25 23:16:18 sitaram, indeed Apr 25 23:16:33 laziness wins again \o/ Apr 25 23:16:43 * dmg (~dmgr_@77-253-29-61.adsl.inetia.pl) has joined #git Apr 25 23:16:47 sitaram, the ideal case requires the config parser to distinguish system config from the rest of the config sources Apr 25 23:17:05 Which is a rather larger than I'm keen on Apr 25 23:17:22 Which is why I was hoping even a gito{lite,sys}-only option would be okay Apr 25 23:17:52 git-daemon conf is taken from system config only? Apr 25 23:17:56 This is bit crazy, but: What about forcing it in git-shell? AFAIK, gitolite dispatches through that. Apr 25 23:18:13 Ford_Prefect: That reads repo config too (IIRC)... Apr 25 23:18:20 Drat Apr 25 23:19:02 Ford_Prefect: just to be clear what would be in repo config that is dangerous? Apr 25 23:19:02 Ilari, hehe, that would work. Does it look at any config? Apr 25 23:19:34 sitaram, repo admin can add arbitrary scripts that would be executed as the user who is doing the upload Apr 25 23:19:37 Ford_Prefect: Well, one wouldn't have to... Apr 25 23:20:01 Always pass the argument if the file exists? Apr 25 23:20:23 Ford_Prefect: in the config? how? Apr 25 23:20:29 * pantsman (~pantsman@pdpc/supporter/active/pantsman) has joined #git Apr 25 23:20:34 Ford_Prefect: And git-daemon2 can chain to arbitrary executable (that knows about SSH_ORIGINAL_COMMAND and how git uses it). Apr 25 23:21:19 sitaram, in the current patch, if the hook is defined, it is executed. And the repo admin can change that. Apr 25 23:21:29 * bx2 (~bx2@host-79.173.26.10.tesatnet.pl) has joined #git Apr 25 23:22:00 Ilari, git-daemon2? Apr 25 23:22:23 Ford_Prefect: gits:// server (can also handle git:// because gits:// is superset of git://) Apr 25 23:22:26 sitaram, makes sense? I'm afraid I'm probably not explaining the case very well. Apr 25 23:22:45 Ford_Prefect: I checked. Gitolite does chain into git-shell. Apr 25 23:22:59 Ford_Prefect: and "defined" is in the repo config (among other places)? Or "defined" == a file with a specific name exists and "repo admin can change that" == "something in config" Apr 25 23:23:19 * tro has quit (Read error: Connection timed out) Apr 25 23:23:34 * domcat (~domcat@62.197.198.36) has joined #git Apr 25 23:23:36 Ford_Prefect: And gitosis does too. Apr 25 23:23:54 sitaram, the first Apr 25 23:24:27 Ford_Prefect: ok Apr 25 23:24:32 sitaram, so sorry about the confusion, but the second. :) Apr 25 23:24:50 Ford_Prefect: the confusion was probably my convoluted wording anyway :) Apr 25 23:25:06 sitaram, nah, I'd forgotten my patches. Getting time to look at this after looong. Apr 25 23:25:10 Ilari, sweet. So that would actually be the path of least change. Apr 25 23:25:34 * coolcat_ (~zimbres@189-55-163-18-nd.cpe.vivax.com.br) has joined #git Apr 25 23:25:43 ianmcorvidae icwiener idoru iff iizukaway ijuma ijuma_ Ilari ilogger2 imarcusthis impulse9 Ingmar IRSeekBot isf_ ivan Apr 25 23:25:57 Ilari, but this means git-shell would have to look at a non-repo path for the hook? Apr 25 23:25:58 * fcoury (~fcoury@189.61.192.220) has joined #git Apr 25 23:26:00 * coolcat_ is now known as Guest11304 Apr 25 23:26:09 * saccade_ (~saccade@dhcp-18-111-66-219.dyn.mit.edu) has joined #git Apr 25 23:26:10 Ilari: is there a command or option which goes even closer to the bare metal on git ls-files? Apr 25 23:26:12 Ilari, as in a path that the repo admin can't play with Apr 25 23:26:24 * tro (~tro@gateway/shell/xzibition.com/x-rkncpladyjumnpxr) has joined #git Apr 25 23:26:25 * sh1mmer (~croucher@adsl-71-134-247-112.dsl.pltn13.pacbell.net) has joined #git Apr 25 23:26:30 Ilari: e.g show stored modification times and such? Apr 25 23:27:04 kkkonrad: There are no modification times. AFAIK, the only thing it doesn't show are internal cache data and extension data. Apr 25 23:27:31 * aheinecke (~quassel@f048148219.adsl.alicedsl.de) has joined #git Apr 25 23:27:40 * segher (~segher@84-105-60-153.cable.quicknet.nl) has joined #git Apr 25 23:27:42 * aheinecke has quit (Read error: Connection reset by peer) Apr 25 23:27:50 Ilari: i thought it considered mod times when looking which files changed in the work dir Apr 25 23:27:58 Ilari (at least on linux) Apr 25 23:28:10 kkkonrad, content only. Try touch'ing a file. Apr 25 23:28:25 kkkonrad, nm. Apr 25 23:28:45 kkkonrad, realised I don't know what I'm talking about. :/ Apr 25 23:28:56 * Mage_Dude has quit (Read error: Connection timed out) Apr 25 23:29:00 Ford_Prefect: dont worry Apr 25 23:29:04 * cpg|away is now known as cpg Apr 25 23:29:15 kkkonrad: if you want a "raw" tree object, 'git cat-file tree HEAD | xxd' (xxd to view it as hex dump) Apr 25 23:29:43 they're compressed, I think Apr 25 23:29:47 at least, they're all binary Apr 25 23:29:51 engla: yes but i would like it "not so raw" (read: human understandable) Apr 25 23:29:54 * kof (~kof@37.201.broadband13.iol.cz) has joined #git Apr 25 23:30:05 * kof (~kof@37.201.broadband13.iol.cz) has left #git ("Leaving") Apr 25 23:30:09 git ls-tree might be what you wnat Apr 25 23:30:15 * sh1mmer has quit (Client Quit) Apr 25 23:30:16 kkkonrad: What data are you looking to dump? The working copy cache data from index (used for change detection)? Apr 25 23:30:24 * Mage_Dude (~chatzilla@c-24-21-220-30.hsd1.or.comcast.net) has joined #git Apr 25 23:31:18 ls-tree has exactly the same information I think. Apr 25 23:31:30 Ilari: i am just poking around and find out why some operations are somewhat slower on msysgit vs. linux Apr 25 23:31:50 kkkonrad: The biggest reason: Windows VFS sucks and Linux VFS rocks. Apr 25 23:32:03 kkkonrad: in my case, I think it's because Windows is horribly slow :-| Apr 25 23:32:07 * devians[w] has quit (Ping timeout: 264 seconds) Apr 25 23:32:11 offby1: cat-file doesn't output compressed data, commits, blobs and trees are not compressed in their real form Apr 25 23:32:22 Ilari: that might be true but our users are on windows..... Apr 25 23:32:28 * devians[w] (~devians@58.96.60.131) has joined #git Apr 25 23:32:34 * _rane has quit (Quit: leaving) Apr 25 23:32:45 engla: must have been binary, which I assumed was compressed. Apr 25 23:33:02 * rane_ (~rane@cs69040.pp.htv.fi) has joined #git Apr 25 23:33:06 engla: and there is always the source code of course Apr 25 23:33:17 Ilari: might it not also be because git is carefully optimized for Linux, but (presumably) not optimized for Windows? Apr 25 23:33:22 but before trying that i thought i would better ask here :) Apr 25 23:33:25 * rane_ is now known as _rane Apr 25 23:33:35 * _rane is now known as rane_ Apr 25 23:33:37 * robin_ (~robin@148.122.35.132) has joined #git Apr 25 23:33:42 git assumes the that operating system has good caching. if it doesn't, everything is slower. Apr 25 23:34:00 I know y'all gonna laugh, but I'm surprised that windows is so slow Apr 25 23:34:02 and I think some of the OS features git uses just don't exist on windows, but I'm not that much of an expert on windows internals Apr 25 23:34:53 jast: indeed. it is not soooooo much, the most prominent speed difference shows up for me on git clone where it creates hard links on linux ... Apr 25 23:35:09 * shennyg (~shennyg@adsl-71-130-122-57.dsl.irvnca.pacbell.net) has joined #git Apr 25 23:35:34 * Mage_Dude has quit (Ping timeout: 276 seconds) Apr 25 23:36:15 * StaticShock (~anton@ip72-192-14-24.ri.ri.cox.net) has joined #git Apr 25 23:36:23 * robin_ has quit (Remote host closed the connection) Apr 25 23:36:24 * Mage_Dude (~chatzilla@c-24-21-220-30.hsd1.or.comcast.net) has joined #git Apr 25 23:37:27 so, svn has clean, linear checkin numbers that start at 1. does git have any tool to enumerate the check-ins on some branch in a similar way? Apr 25 23:37:47 StaticShock: not that I know of Apr 25 23:37:52 ok Apr 25 23:37:59 those clean, linear checkin numbers are an evil lie, in my humble opinion Apr 25 23:38:20 how so Apr 25 23:38:21 Ilari, so if I understand the final resolution, we make git-shell look for the existence of the hook and then pass it on the command line to upload-pack? Apr 25 23:38:40 * |kukks| (~Guenter@p57A08C55.dip0.t-ipconnect.de) has joined #git Apr 25 23:38:52 * keystr0k has quit (Quit: keystr0k) Apr 25 23:40:04 * robinr (~robin@148.122.35.132) has joined #git Apr 25 23:40:50 Ford_Prefect: Maybe. But idea was forcing it on git-shell (unconditionally or on some condition). Apr 25 23:40:52 * hyperair has quit (Ping timeout: 264 seconds) Apr 25 23:41:20 Ilari, but upload-pack doesn't know it's being called from git-shell, right? Apr 25 23:41:40 * fisch_ (~Thomas@ppp-93-104-141-70.dynamic.mnet-online.de) has joined #git Apr 25 23:41:45 * kukks has quit (Ping timeout: 276 seconds) Apr 25 23:41:46 ianmcorvidae icwiener idoru iff iizukaway ijuma ijuma_ Ilari ilogger2 imarcusthis impulse9 Ingmar IRSeekBot isf_ ivan Apr 25 23:41:48 Ilari, So we _could_ always pass .git/hooks/pre-upload on the command line. And check for existence in upload-pack Apr 25 23:41:54 Ford_Prefect: Its (to enable extra hooks) command line option to upload-pack? Apr 25 23:42:13 * ttvd (~ttvd@CPE0023691ddfd1-CM001404948d9e.cpe.net.cable.rogers.com) has joined #git Apr 25 23:42:20 Having a tree with some random tags on different branches, how can I get the diff between the source tree at tag1 and the source tree at tag2? Apr 25 23:42:29 Ilari, y'know that's so much smarter and simpler than what I was thinking. :) Apr 25 23:42:38 Aleric: git diff Apr 25 23:42:51 heh, I knew it had to be simple... Apr 25 23:43:14 * tro has quit (Read error: Connection timed out) Apr 25 23:43:27 ianmcorvidae icwiener idoru iff iizukaway ijuma ijuma_ Ilari ilogger2 imarcusthis impulse9 Ingmar IRSeekBot isf_ ivan Apr 25 23:43:58 Ilari, so upload-pack gets and --enable-hooks command line argument. Is it possible for anything other than git-shell to force this? Apr 25 23:44:15 (looking for possible security problems with the approach) Apr 25 23:44:28 Ford_Prefect: Client on ssh:// could set it (but would require client mods). Apr 25 23:44:49 * tro (trotek@gateway/shell/xzibition.com/x-opwklugggcrdhvdq) has joined #git Apr 25 23:45:08 * Belna_ has quit (Ping timeout: 258 seconds) Apr 25 23:45:25 ianmcorvidae icwiener idoru iff iizukaway ijuma ijuma_ Ilari ilogger2 imarcusthis impulse9 Ingmar IRSeekBot isf_ ivan Apr 25 23:45:33 * walf443 has quit (Ping timeout: 260 seconds) Apr 25 23:46:11 Ilari, right. Can make it so if the client didn't send the command-line option and the pre- hook exists, it errors out Apr 25 23:46:13 ianmcorvidae icwiener idoru iff iizukaway ijuma ijuma_ Ilari ilogger2 imarcusthis impulse9 Ingmar IRSeekBot isf_ ivan Apr 25 23:46:20 Ilari, warning in the same situation for the post- hook Apr 25 23:46:43 * fcoury is now known as fcoury|away Apr 25 23:49:17 Ford_Prefect: wouldn't that require everyone to use a patched client? Apr 25 23:49:21 StaticShock: in subversions, checkin numbers are not actually per branch, see... apart from that, git stores history differently, namely as a graph of commits for each branch. that is, a commit refers to the commit it is based on, and merge commits refer to several commits. there is no unique way of numbering those commits. Apr 25 23:49:40 sitaram, I'll include a patch. Apr 25 23:49:53 * tzafrir_laptop has quit (Ping timeout: 268 seconds) Apr 25 23:50:15 * owen1 (~oren@cpe-72-129-82-140.socal.res.rr.com) has left #git Apr 25 23:50:28 sitaram, so yes, the admin is basically saying "you need at least git 1.7.*, for example) Apr 25 23:50:29 * treak007 (~jim@h240.208.130.174.dynamic.ip.windstream.net) has joined #git Apr 25 23:50:30 StaticShock: apart from that, different clones of a repository can have commits in different places in a branch's history graph due... that's a specialty in decentralized revision control. Apr 25 23:50:35 * ddunbar has quit (Ping timeout: 245 seconds) Apr 25 23:50:50 StaticShock: so it's entirely possible that your revision number 51 is someone else's revision number 53... Apr 25 23:50:52 StaticShock, bzr numbers branch revisions sequentially as a convenience but this isn't the primary identifier - and can't be in any distributed versioon trontrol system Apr 25 23:51:05 Ford_Prefect: "recent git needed" is not a bad thing. "Patched git needed" I'm not so sure of Apr 25 23:51:12 And git has no such convenience Apr 25 23:51:35 sitaram, indeed. This would require the server-side and client-side patches to be merged. Apr 25 23:51:53 Ilari, do we want to always pass that option from git-shell? What if the system/hosting admin doesn't want to allow these hooks? Apr 25 23:52:01 * ThunderChicken has quit (Remote host closed the connection) Apr 25 23:53:22 * nevyn^_ is now known as nevyn^ Apr 25 23:53:49 * tzafrir_laptop (~tzafrir@bzq-218-155-145.cablep.bezeqint.net) has joined #git Apr 25 23:54:05 Ford_Prefect: If it is system like gitolite, the admin either has the hooks (and wants to force them) or does not have them. Apr 25 23:54:26 Ilari, repo admins can't add the hook themselves? Apr 25 23:54:49 * ddunbar (~daniel@adsl-71-134-245-151.dsl.pltn13.pacbell.net) has joined #git Apr 25 23:54:56 * sagsousuke has quit (Ping timeout: 265 seconds) Apr 25 23:57:13 * dj_tjerk (~pietjebel@5354947C.cable.casema.nl) has joined #git Apr 25 23:57:29 * mw has quit (Quit: mw) Apr 25 23:57:44 Ford_Prefect: One who has shell access to gitolite account has high-level access.. Apr 25 23:58:00 ianmcorvidae icwiener idoru iff iizukaway ijuma ijuma_ Ilari ilogger2 imarcusthis impulse9 Ingmar IRSeekBot isf_ ivan Apr 25 23:58:00 ianmcorvidae icwiener idoru iff iizukaway ijuma ijuma_ Ilari ilogger2 imarcusthis impulse9 Ingmar IRSeekBot isf_ ivan Apr 25 23:59:18 * ice799 (~joe@67-23-10-49.static.slicehost.net) has joined #git Apr 25 23:59:31 * ThunderChicken (~bofh@kernel-panic/member/ThunderChicken) has joined #git Apr 25 23:59:32 * munichpython has quit (Remote host closed the connection) Apr 25 23:59:33 * rgr (~user@85.183.18.158) has joined #git Apr 26 00:00:00 * walf443 (~walf443@KD125054058069.ppp-bb.dion.ne.jp) has joined #git Apr 26 00:00:12 Ilari, looking at Nicolas' mail here - http://lists.zerezo.com/git/msg711058.html - I was under the impression that an in-repo hook is a bad thing Apr 26 00:01:15 * tro has quit (Read error: Connection timed out) Apr 26 00:01:42 * v0n (~vdidel@85-170-186-241.rev.numericable.fr) has joined #git Apr 26 00:01:46 * Rene (~Rene@cs181081047.pp.htv.fi) has joined #git Apr 26 00:01:53 good evening all! Apr 26 00:02:10 * homer5439 has quit (Quit: Page closed) Apr 26 00:02:32 * g3d (~g3d@85-222-76-212.home.aster.pl) has joined #git Apr 26 00:03:08 does anyone know any cure for the msysgit package so that it would handle the windows "home"-directory correctly? Apr 26 00:03:31 * blaines has quit (Ping timeout: 276 seconds) Apr 26 00:03:50 my home-dir contains the -character, and both bash and the windows command-prompt does not recognize this path correctly.. Apr 26 00:04:15 Ilari, okay, I think I see what we're saying here - the system admin can only enable this if (s)he trusts /all/ the repo admins on the box. Users need to trust admins of their repositories. Apr 26 00:04:35 * darwin_ has quit (Quit: Leaving...) Apr 26 00:04:58 * darwin (~darwin@a40-prg1-17-141.static.adsl.vol.cz) has joined #git Apr 26 00:05:20 * tg has quit (Remote host closed the connection) Apr 26 00:05:22 Ilari, that said, having it on-always also means that the sysad can't disable it Apr 26 00:06:04 * tg (irc@x-net.hu) has joined #git Apr 26 00:06:31 * tro (trotek@gateway/shell/xzibition.com/x-ymtmionxauytiwrv) has joined #git Apr 26 00:06:58 * Ford_Prefect has the sinking feeling we're back to square 1