IRCloggy #git 2010-07-20

Provider of IRC logs since 2005.

WARNING: As Freenode became unjoinable and lost all warnings in topics, we cannot log channels on Freenode anymore.

	2010-07-20
	← steveire left	00:00
	→ steveire joined	00:01
	← steveire left	00:01
	→ steveire joined	00:01
	← hobodave left	00:02
	→ hobodave joined	00:02
	→ Joeconyers joined	00:03
	← scarabx left	00:04
	→ masonkramer joined	00:04
	← priidu left	00:05
	→ ianmcorvidae\|alt joined	00:05
	← ianmcorvidae left	00:06
	← ianmcorvidae\|alt left	00:07
	→ ianmcorvidae joined	00:07
	← ianmcorvidae left	00:07
	→ ianmcorvidae joined	00:07
	→ hearit joined	00:08
	← hobodave left	00:08
	→ hipe joined	00:08
	← unreal left	00:09
	→ unreal joined	00:10
	← amerine left	00:10
	→ frogonwheels joined	00:10
	→ sshc_ joined	00:12
	← godsflaw left	00:12
	cpg → cpg\|away	00:13
	← felipec left	00:14
	→ tyru joined	00:16
	← frakturfreak left	00:24
	→ frakturfreak joined	00:25
	← tstclair left	00:25
	← nevyn left	00:26
	→ nevyn joined	00:28
	→ hasrb joined	00:29
	← jaql left	00:29
	← warthog9 left	00:31
	← eletuchy left	00:31
	→ eletuchy joined	00:31
	← nevyn left	00:32
	← steveire left	00:33
	→ steveire joined	00:34
	← steveire left	00:34
	→ steveire joined	00:34
	→ nevyn joined	00:39
	→ Raydiation joined	00:42
	← alezandro left	00:42
Raydiation	if i push data on the server and i want to update the working tree immediately (i run a webserver) i have to use a post-receive hook right?	00:42
Ilari	Raydiation: Or post-update.	00:43
Raydiation	Ilari: whats the difference?	00:43
Ilari	Raydiation: post-receive gets full update commands (stdin), post-update only branches updated (via arguments).	00:45
	Raydiation: *refs updated	00:45
	← carutsu left	00:46
Raydiation	Ilari: whats better for deploying the files on a webserver?	00:46
	i suppose post-receive	00:46
Ilari	Raydiation: Well, both are about as good.	00:47
Raydiation	ok thx	00:47
batok	I've installed gitolite, in my case I cant that one update hook on the remote ( gitolite controlled ) repo , deploy some files to web server , what would be the best way to do this?	00:49
	s/cant/want	00:49
	→ jmcantrell joined	00:50
	← Tuller left	00:50
Ilari	batok: IIRC, you can just edit hooks that are not reserved by gitolite (reserved hooks include update in any repo and post-update in gitolite-admin. Since deploying files is done from post-receive/post-update, those can be used.	00:50
	And for gitolite-admin, there's second chainload hook for running actions after gitolite post-update has run.	00:51
	Using that secondary hook is pretty rare (but I do use it to reload SRP user databases on config change).	00:52
batok	I ve used update.secondary ( that's how gitolite let do additional things ) but my question is more git related. How in a hook should I get those repo files ?	00:52
Ilari	batok: Search git website HOWTO?	00:52
batok	do I need to make a git archive ?	00:53
Ilari	batok: http://toroid.org/ams/git-website-howto	00:54
	← Yuffster left	00:56
	← eighty4 left	00:56
batok	thank you very much llari	00:58
	← Mage left	00:59
	→ Mage joined	01:00
	→ Camel___ joined	01:00
	← Camel_ left	01:00
batok	so the key is using git checkout -f	01:02
	← hearit left	01:03
	← ExtraSpice left	01:03
	→ alezandro joined	01:03
	← beatak_tksh left	01:06
	← Yuuhi` left	01:08
	← jrmuizel left	01:11
sitaram	possibly with the addition of using GIT_WORK_TREE, because usually the repo you're working off of is a bare repo and should not have a work tree	01:13
	(git checkout should refuse to do that unless GIT_WORK_DIR is set, IIRC)	01:13
	→ hobodave joined	01:14
	→ edt joined	01:14
edt	what does "fatal: This operation must be run in a work tree" mean? And how do I fix it?	01:14
	It does not mater if I trash the tree for subsequent pull - its easy to reclone it.	01:15
	→ Vampire0 joined	01:15
edt	I need to apply a test patch...	01:16
mar77i	edt: well, see if you're in the right directory. either the one where .git is or one where git is in a parent directory.	01:16
edt	git lot works does that not imply I am ok?	01:16
batok	sitaram , have you ever documented anything like that ?	01:17
mar77i	edt: I'm not understanding you. what are you talking about again?	01:17
batok	I mean the "git website HOWTO" adapted to gitolite?	01:17
	← quentusrex_ left	01:18
Ilari	edt: What operation are you trying to run? From hook? From command line?	01:18
edt	gentoo -9999 packages use git.	01:18
	← Vampire0_ left	01:18
edt	I need to apply a patch to one of the git trees	01:18
	I can do git log	01:18
	but trying to do almost any other operation gives the message above	01:19
Ilari	edt: What does 'pwd' give?	01:19
edt	I want to do 'cat ../patch_file \| patch -p1; git commit -a'	01:20
	pwd will give: /usr/src/ati which is a symlink to /usr/portage/distfiles/git-src/xf86-video-ati	01:20
	→ adamm joined	01:20
Ilari	edt: Try changing directory to /usr/portage/distfiles/git-src/xf86-video-ati first.	01:22
edt	going directly to /usr/port... does not help	01:22
	same message	01:22
	← enherit left	01:22
mar77i	I'm off	01:22
Ilari	edt: 'git rev-parse --show-cdup'?	01:22
edt	no output	01:23
Ilari	edt: And also what does 'git rev-parse --git-dir' say?	01:23
	→ tr3buchet joined	01:23
edt	.	01:23
Ilari	edt: That's a bare repository then.	01:24
	→ chrislerum joined	01:24
Ilari	edt: It does not have a working copy and applying patches needs one.	01:24
	← frogonwheels left	01:24
edt	how do I add a working copy?	01:25
Ilari	edt: There is directory called '/usr/portage/distfiles/git-src/xf86-video-ati/objects', right?	01:25
edt	yes	01:25
Ilari	edt: Clone that repository ('git clone file:///usr/portage/distfiles/git-src/xf86-video-ati' from suitable place).	01:26
	← mar77i left	01:27
	← hipe left	01:28
	← Bombe left	01:28
edt	that works. How do I get the fix back into the original repo once I commit the change on the new repo	01:28
	→ Bombe joined	01:28
	→ DrNick joined	01:30
	→ Guest76213 joined	01:30
Guest76213	Hey all	01:30
	Wondering if you can help me with a git checkout on a multicore system, I want git to use all available CPU cores but its only using one	01:31
	← royalty left	01:31
Guest76213	Running on Ubuntu 10.04 LTS Server 64-bit using git-core from repos	01:31
edt	Ilari thanks. Git push worked to add the fix	01:31
	→ friskd joined	01:32
Guest76213	And it's running on EC2	01:32
Ilari	Guest76213: Only some operations like delta compression use multiple cores.	01:33
	← tildeequals left	01:33
Ilari	Guest76213: And usually operations are not CPU bound but either I/O bound or network bound.	01:34
Guest76213	Yeah, looking in top shows high CPU usage	01:34
	I'm checking out a 30GB repository which is why it's taking so long	01:34
	But it's a local repo	01:34
Ilari	Guest76213: Wow... And large binary files do really bad things to git perfromance.	01:34
Guest76213	Yeah I know	01:34
	It's an emergency server move and used git to keep track of local modifications	01:35
	It's just taking so long to checkout...	01:35
	Got it running on a 8 core EC2 instance but as it's single threaded that isn't helping in the slightest	01:35
sitaram	batok: I don't understand... the git website howto applies to any central server, gitolite is just another one, why mix the two. That way I'd be documenting anything you can do with a central git server, no?	01:35
Guest76213	Ilari, any tips to speed this up?	01:35
Ilari	Guest76213: What operation is slow? clone? Checkout?	01:35
Guest76213	checkout	01:36
	It's a 6.5G repository with ~30GB worth of checked out data	01:36
	→ rchavik joined	01:36
Ilari	Checkout should write index followed by working copy files and that should be quite much I/O bound... Hmm...	01:36
	← sshc_ left	01:37
Guest76213	Running the checkout I don't even get the object count	01:37
	It's maxing out CPU and display nothing in terminal	01:37
	→ sshc joined	01:37
Guest76213	On a similar but smaller (2G) repo it works reasonably fast	01:37
AAA_awright	As fast as Git is it's optimized for maximum compression of file size on highly compressible plain text files	01:37
Ilari	Guest76213: How many files there are in the repo?	01:37
Guest76213	Hundreds of thousands	01:38
	Mostly small files <1MB in size	01:38
	Text majority - CSS, PHP etc	01:38
Ilari	Guest76213: That's enough to do bad things to performance also...	01:38
AAA_awright	That doesn't stop me from backing up my ~ directory with it, including GB-size photos	01:38
	Guest76213: How big are your largest directories?	01:39
Guest76213	22G	01:39
Ilari	Guest76213: Strace shows what kind of activity (you can attach it to existing process)?	01:39
AAA_awright	I mean, number of files	01:39
	I have had speed problems with more than thousands of files inside a directory, as if trees aren't as efficient as blobs or something	01:40
Guest76213	Yeah	01:40
	What strace args do you want	01:40
Ilari	Guest76213: 'strace -p <pid>' (where <pid> is some process using lots of CPU).	01:41
	← tedoc2000 left	01:41
Ilari	Guest76213: Collect some syscall messages and then hit CTRL+C to detach (save log to file).	01:42
	→ jrmuizel joined	01:44
	← jksz left	01:44
Guest76213	lstat("path/to/file", 0x7fffdf1a6630) = -1 ENOENT (No such file or directory)	01:45
	Getting a lot of calls like that	01:45
	→ Camel_ joined	01:46
	→ jksz joined	01:46
	← Camel___ left	01:49
	← rayners_afk left	01:50
	→ frogonwheels joined	01:51
	← Guest76213 left	01:56
	← jtdowney left	01:56
	→ Azrathud joined	01:56
	← Azrathud left	01:56
	← edt left	01:59
	← jksz left	02:01
	→ jksz joined	02:02
	→ ignatp joined	02:02
	← parasti left	02:04
	→ treak007 joined	02:05
	→ thwart joined	02:08
	← sh1mmer left	02:08
	→ robbe- joined	02:09
	← flaguy48 left	02:10
robbe-	Hi, I'm looking for a way to pull from gitweb url's. Is that possible? Is it (not) recommended?	02:11
	← saiyr left	02:11
	→ outoftime joined	02:12
outoftime	any ideas on why I might not get color even when color.ui is set to 'auto' ?	02:12
	(but I do when it's set to 'always')	02:12
	I'm running bash with screen on top of it (but I see the same behavior in vanilla bash)	02:12
Ilari	robbe-: It is not possible. You need something like git daemon or sshd for serving.	02:14
	← chrislerum left	02:15
	→ mithro joined	02:15
	→ tildeequals joined	02:16
	→ ajpiano joined	02:17
	← Mage left	02:19
	← Bass10 left	02:19
	→ Mage joined	02:20
	→ cytrinox_ joined	02:20
	← Raydiation left	02:21
	→ jtbandes joined	02:21
robbe-	Thanks Ilari. Any other options worth looking into? I've been using ssh access to clone my own repositories, but now I'd like to enable anonymous cloning/pulling. I'll look into the git daemon for starters.	02:21
	→ sh1mmer joined	02:22
	← ajpiano left	02:22
	→ ajpiano joined	02:23
	← cytrinox left	02:23
	cytrinox_ → cytrinox	02:23
Ilari	robbe-: Yeah, git daemon is for anonymous fetch/clone/pull.	02:26
	← batok left	02:28
robbe-	Thanks a bunch, reading the docs already. :-)	02:28
	→ Bass10 joined	02:28
Ilari	robbe-: And as hint: Whitelist entries better be in sync with --base-path or daemon won't serve anything, and whitelist entries do not set base path.	02:31
	→ batok joined	02:32
robbe-	I don't know yet what you're talking about, but I reckon I will soon enough. I'll definately keep it in mind. :-)	02:32
	→ Camel__ joined	02:33
	← Camel_ left	02:33
Ilari	robbe-: Usually its easiest to use --base-path to restrict what can be served and not use those <directory> arguments at all.	02:33
	← cehteh left	02:34
robbe-	Okay.	02:35
	← masonkramer left	02:36
	← konr left	02:37
	→ konr` joined	02:37
	← Elfe left	02:37
	→ Elfe_ joined	02:37
	← Belna_ left	02:38
Ilari	robbe-: Also useful for testing is running it from command line with --verbose but without --detach. That causes messages to be printed to stdout.	02:38
	→ cehteh joined	02:40
	← hasrb left	02:42
	← khaije\|amalt left	02:43
	→ scarabx joined	02:43
	→ alexmchale joined	02:48
	← kenyon left	02:50
	→ timj_ joined	02:52
	← dget left	02:53
robbe-	I'm modifying my gitweb configuration currently, discovered it can do a bit more than I knew. :-)	02:53
	← timj__ left	02:56
	→ kenyon joined	02:56
	← sh1mmer left	02:56
	← spearce left	02:59
	→ spearce joined	03:00
	← Blackguard left	03:00
	← kukks left	03:01
	← alexmchale left	03:03
	→ lanthan joined	03:11
	→ hasrb joined	03:13
	← fishey left	03:15
	← justin-george left	03:16
	← Mage left	03:17
	→ Mage joined	03:18
	← hyperair left	03:18
	← thiago_home left	03:19
	→ Camel_ joined	03:23
	← Camel__ left	03:23
	← ramsey left	03:23
	→ alexmchale joined	03:23
	← anjen left	03:23
	← AkumaStreak left	03:24
	← alexmchale left	03:27
	← bradallen137_ left	03:28
	← ComaWhite left	03:28
	← unreal left	03:29
	→ unreal joined	03:29
	→ comawhite joined	03:31
	→ adamv joined	03:31
	← spearce left	03:31
	→ bradallen137_ joined	03:33
	→ carutsu joined	03:33
	← tchan left	03:35
	← Joeconyers left	03:36
	→ Joeconyers joined	03:39
	→ hyperair joined	03:41
	← konr` left	03:41
	cpg\|away → cpg	03:41
	← dr_win left	03:43
	jjcm → jjcm_kingOfWit	03:48
	jjcm_kingOfWit → jjcm	03:49
	→ hohoho joined	03:53
	→ welly joined	03:54
	← welly left	03:54
	→ welly joined	03:54
	→ warthog9 joined	03:55
	← jmcantrell left	03:57
	→ montylounge joined	03:59
	→ bandu joined	03:59
	→ guilhermeblanco joined	04:04
	← montylounge left	04:06
	← jrmuizel left	04:07
	→ alexmchale joined	04:08
	← alexmchale left	04:08
	← scarabx left	04:09
	→ johnw joined	04:12
	→ jrmuizel joined	04:13
	bandu → coyo	04:13
	← guilhermeblanco left	04:14
	← aresnick left	04:14
	← jrmuizel left	04:15
	← Joeconyers left	04:15
	→ Camel__ joined	04:21
	← adamm left	04:23
	← zenazn left	04:24
	← Camel_ left	04:25
	→ dr_win joined	04:27
	cpg → cpg\|brb	04:29
	← jfkw left	04:31
	← steveire left	04:33
	→ steveire joined	04:34
	← steveire left	04:34
	→ steveire joined	04:34
	→ pgib joined	04:35
	← BlindOracle left	04:36
	→ sh1mmer joined	04:37
	→ xxthink joined	04:38
	→ BlindOracle joined	04:38
	← xxthink left	04:38
	→ esr joined	04:40
esr	After a git cinversion, I moved a bunch of tags files from refs/remote/tags to refs/tags. But they're not visible in gitk. Any suggestions?	04:44
	→ apsysoev joined	04:45
blueshift	esr: can you see any tags in gitk? istr there may be an option you have to wiggle	04:45
esr	blueshift: You may be right. They show up in git tag --list	04:45
	→ babusri joined	04:45
esr	However, man gitk contains no obvious option	04:47
Gitbot	However: the gitk manpage can be found at http://git.or.cz/man/gitk	04:47
	→ radioman-lt joined	04:47
	← apsysoev left	04:47
blueshift	played around with the gui buttons?	04:47
Ilari	esr: Do they show up in 'gitk --tags'?	04:48
esr	Ilari: Checking...	04:48
	Ilari: Yes. Why that's not on the gitk manpage is now the mystery :-)	04:49
blueshift	because maintaining documentation isn't fun or cool	04:50
Ilari	esr: gitk accepts rev-list revision options. And if tag doesn't point to any commmit that would be otherwise be shown, it isn't shown.	04:51
robbe-	Ilari: I enabled pulling from http using apache configs; basically when you request git.server.org, it gives you the gitweb, git.server.org/repo.git gives the gitweb for 'dumb' clients (using rewrite rules), but it actually serves the repo (well a bare clone of it), so when you request specificfiles which are not index, it works. :-)	04:51
esr	Ilari: OK, this was me not grokking gitk well enough, then. Thanks.	04:51
Ilari	robbe-: Dumb HTTP serving is _deprecated_.	04:52
	robbe-: (at least for servers that can run CGI scripts).	04:52
robbe-	Ilari: Oh, but it was in the gitweb documentation. :(	04:52
	← brosner left	04:53
Ilari	robbe-: And http-backend has optional compatiblity mode for dumb clients anyway.	04:53
	→ aziz joined	04:53
Ilari	robbe-: Performance of dumb HTTP is _bad_.	04:53
robbe-	Darn, well .. at least I learned something about configuring apache. :)	04:53
	→ brosner joined	04:53
	→ zoyd joined	04:54
robbe-	Ilari: Ah, I see, are there more reasons why it is deprecated, except it's slow? (It is only a very small repo I tested it with so I don't noticed it.)	04:54
	didn't*	04:54
zoyd	hi, is "git checkout" the equivalent of a "svn export"?	04:54
blueshift	no, it's the equiv of svn update	04:55
Ilari	zoyd: More like equiv of svn switch	04:55
	zoyd: And svn revert in some other modes.	04:56
cmyers	robbe-: you will notice it more with certain specific useacases. The "smart" http has a script on the server side which can create custom packfiles with exactly what you need. with dumb http it often has to download extra stuff That is why it is slower. YOu might not notice on the initial clone, but for future "fetches" where you only need to pull a few recent changes, there is a big difference.	04:56
zoyd	no way to 'export' a repository then?	04:56
Ilari	robbe-: Slow, bad push support...	04:56
	→ maleknet joined	04:56
Ilari	zoyd: 'git archive'?	04:56
	zoyd: (it even has remote snapshot mode).	04:57
cmyers	zoyd: you want a "checkout" without the history/repo part?	04:57
pgib	zoyd, not so much a problem anyways since git doesn't throw .svn directories all over the place	04:57
zoyd	oh, 'git help' doesn't return an 'archive' .. lemme se	04:57
robbe-	Oh, but I don't want push support :P - of course if it's deprecated I'll switch to the git daemon; Just interested as to why it's deprecated. :-)	04:57
Ilari	zoyd: 'git help archive'.	04:57
	robbe-: Anyway, smart HTTP is superrior to dumb HTTP. And HTTP is bit second-class anyway.	04:58
	robbe-: Might be good idea to offer as secondary to git:// since there are nasty firewalls out there.	04:58
robbe-	Ilari: Yes, I see that completely, thanks to the info you and cmyers are giving me, which is of course very much appreciated.	04:59
	← unreal left	05:00
cmyers	ugh, I am still in 'git svn' hell =( Does anyone know how to make git-svn ignore something it thinks is a branch - that creates those branch@rev remote branches?	05:01
Ilari	Also, AFAIK, currently smart HTTP can't do remote snapshot. Only file://, git://, ssh:// and co (everything supporting CONNECT) can do remote snapshot.	05:01
	→ unreal joined	05:01
zoyd	Ilari: i would think the remote repository needs to allow for 'git archive' to work. because it appears that it's rejecting my request.	05:01
Ilari	Yeah, not all remote repositories allow remote snapshotting.	05:01
	→ freimatz joined	05:02
	← jonmasters left	05:02
	→ pastie joined	05:02
Arrowmaster	i didnt think it was enabled by default	05:03
	zoyd can't afford cloning a 200+megs repo :/	05:03
	← Camel__ left	05:03
Arrowmaster	but you can do a shallow clone then run archive locally	05:03
	→ Camel_ joined	05:03
zoyd	shallow clone?	05:03
	→ doener joined	05:04
robbe-	Okay, so there's good reasons to provide primarily git:// and smart http:// as second option, I'll see to it. Thanks!	05:04
cmyers	do a full clone to local, then do a clone with '-s' and it will use hardlinks to clone, and barely take up any extra space	05:04
	zoyd reads http://rds.yahoo.com/_ylt=A0oGdE9bLkVMfTkBVrxXNyoA;_ylu=X3oDMTEyaTdsbHRqBHNlYwNzcgRwb3MDMgRjb2xvA3NrMQR2dGlkA1lZWTFfNzg-/SIG=12pufq609/EXP=1279688667/**http%3a//www.kernel.org/pub/software/scm/git/docs/technical/shallow.txt	05:04
cmyers	then you can remove the .git dir and zip it up	05:04
	zoyd reads http://kernel.org/pub/software/scm/git/docs/technical/shallow.txt	05:05
zoyd	cmyers: problem is b/w, storage isn't a problem	05:05
	← meuh left	05:06
cmyers	ah, that is a problem then =/ Is there a remote machine you could use to do the pull?	05:06
	→ meuh joined	05:06
cmyers	if this is a one-time thing, there is that. if this is a regular thing, you only have to clone it once, so maybe paying the b/w once isn't so bad.	05:07
doener	cmyers: -s makes it use alternates, not hardlinks	05:07
cmyers	doener: alternates?	05:07
zoyd	nevermind. i'll ask a friend to git clone for me	05:07
	thanks cmyers, Ilari, pgib, blueshift :)	05:08
doener	cmyers: basically a file that tells git to also look into another repo's object database to find objects it cannot find locally	05:08
cmyers	oh, so like git-specific symlinks	05:08
	→ slonopotamus joined	05:08
cmyers	if that is how you are gonna do it, why not just use actual symlinks?	05:08
doener	cmyers: because you could only have one of those. And you wouldn't be able to have objects local to this repo at all	05:09
zoyd	i suppose repository servers can't handle the archive/compression load either	05:09
doener	cmyers: alternates work in addition to the local object database not instead of	05:10
cmyers	I mean, why not have a symlink for each object in the DB that points to the original object in the source DB. I guess that is a lot of symlinks that way though.	05:10
	← brosner left	05:11
doener	cmyers: any repacking/gc'ing in the other repo would break all the symlinks...	05:11
	← bdiego left	05:13
cmyers	doener: ...oh. that makes sense =)	05:14
	→ jony joined	05:14
doener	cmyers: that's also what alternates offer that hardlinks don't: You don't lose the space savings on a repack/gc run	05:17
	cmyers: but you're dependent on the other repo not suddenly dropping objects it no longer needs, but this repo does (usually that means history rewriting or branch deletion)	05:17
	→ jonmasters joined	05:18
	← zoyd left	05:19
cmyers	well my understanding of that is vastly improved now	05:20
	now if only I understood git-svn well enough to un-fuck it and get my history imported =(	05:20
	→ ProperN[out] joined	05:23
	← ProperN[out] left	05:23
	→ ProperN[out] joined	05:23
	← hyperair left	05:23
	→ tvw joined	05:25
	← ProperNoun left	05:25
	ProperN[out] → ProperNoun	05:25
	← sivy left	05:26
	← batok left	05:26
	→ hyperair joined	05:29
AAA_awright	Hm whhy haven't I read that before... so should it be possible to remove earlier commits from your local repository, by setting a more recent shallow commit, and repacking?	05:30
	→ vu3rdd joined	05:33
	→ tchan joined	05:33
	← urkud left	05:34
coppro	.join #pv6	05:35
	← bradallen137_ left	05:35
	→ LongBeach joined	05:35
	cpg\|brb → cpg	05:36
	→ urkud joined	05:36
	← babusri left	05:36
	→ girishr joined	05:37
	→ poseid joined	05:37
	← coppro left	05:40
doener	cmyers: I had to look up your problem in the logs, and only found a rather vague description... But maybe --ignore-paths?	05:40
	← unreal left	05:40
	→ unreal joined	05:40
	→ coppro joined	05:40
	← hasrb left	05:43
	← hagabaka left	05:44
	← molokai left	05:44
	← adamv left	05:44
	← BlindOracle left	05:44
	← agile left	05:44
	← urkud left	05:45
	→ BlindOracle joined	05:45
	← SlickT10 left	05:45
	→ agile joined	05:45
	← tildeequals left	05:46
	→ roop joined	05:47
	← Camel_ left	05:47
	→ Camel_ joined	05:48
	← Bass10 left	05:48
	← friskd left	05:49
	→ SlickT10 joined	05:50
	→ friskd joined	05:50
	→ Schmallon joined	05:51
	← friskd left	05:55
	→ amerine joined	05:57
	→ nevyn^_ joined	05:58
	← nevyn^_ left	05:59
	→ urkud joined	05:59
	← adamholt left	05:59
	← agile left	06:00
	→ kadoban joined	06:00
	← tehbaut left	06:00
	→ agile joined	06:00
	→ molokai joined	06:01
	→ cilly joined	06:01
	→ adamholt joined	06:01
	→ xrogaan joined	06:02
	← xrogaan left	06:02
	→ xrogaan joined	06:02
	→ tehbaut joined	06:02
	← ignatp left	06:04
	→ Guest49745 joined	06:05
	→ Titosemi joined	06:05
	← Guest49745 left	06:06
	→ cbreak_work joined	06:07
	← urkud left	06:09
	→ urkud joined	06:10
	→ psankar joined	06:14
	→ apsysoev joined	06:14
	→ galderz joined	06:15
	→ [Po]lentino joined	06:16
	→ opalka joined	06:17
	← slonopotamus left	06:18
	→ tildeequals joined	06:20
	← urkud left	06:23
	→ skoop joined	06:23
	→ urkud joined	06:23
	← sako left	06:26
	← aziz left	06:28
	← steveire left	06:28
	← lat left	06:29
	→ steveire joined	06:32
	← unreal left	06:35
	← brntbeer left	06:35
	→ unreal joined	06:35
	→ priidu joined	06:37
	→ brntbeer joined	06:38
	→ Camel__ joined	06:38
	→ rox` joined	06:38
	← urkud left	06:39
	← Camel_ left	06:41
	← unreal left	06:42
	→ unreal joined	06:42
	← bauruine left	06:42
	→ steffkes joined	06:44
	→ mtkd joined	06:45
	← jtbandes left	06:46
	← programble left	06:47
	→ nevyn^_ joined	06:49
	→ cirwin joined	06:49
cirwin	is there an easy way I can see a tree of my branches, without the noise of intervening commits?	06:50
	← thwart left	06:50
johnw	you mean, confine to merge commits?	06:52
cirwin	yes	06:52
johnw	good questio	06:52
	git log --merges --graph?	06:53
cirwin	hrm	06:53
	I think i need to include the heads as well	06:53
johnw	git log --all --merges --graph?	06:53
	→ frakturfreak_ joined	06:54
cirwin	it just shows merges	06:54
	which is half way there :)	06:54
johnw	yeah, you just asked for that	06:54
cirwin	i know	06:54
	but that's not what I meant	06:54
	I'd like to see the tree of branches - which is horrendously complicated - but in order to make sense of that, i'd intuitively show the merge commits	06:54
	→ thwart joined	06:55
	← frakturfreak left	06:55
cirwin	github's format is pretty close to being usable, but I'd like to compress all the commits where there are no merges so that I can see more branches - I don't know whether that makes sense	06:56
johnw	i don't know the answer	06:58
	← johnw left	06:58
	→ johnw joined	06:59
	→ eighty4 joined	06:59
	← thwart left	07:00
	→ giallu joined	07:01
	← project2501b left	07:01
	← Belna left	07:04
	→ thwart joined	07:04
	→ timyhac joined	07:05
	← tvw left	07:06
	→ dpnux joined	07:07
	→ valeo joined	07:07
	← warthog9 left	07:10
	→ plediii joined	07:11
	← __ed left	07:12
sgronblo	Is there a way to show some sort of svn blame output which includes commit messages?	07:13
	→ __ed joined	07:13
	← brntbeer left	07:14
	← thwart left	07:15
	→ aziz joined	07:15
	→ leo2007 joined	07:20
	→ warthog9 joined	07:22
	→ FauxFaux joined	07:22
	→ Camel_ joined	07:23
	← Camel__ left	07:23
	→ Wraithan joined	07:23
Wraithan	Is there a way to do a git log for just one committer?	07:23
	nvm	07:23
rane_	git log --author=Wraithan	07:23
	Wraithan is blind	07:23
	→ ExtraSpice joined	07:25
	← steveire left	07:28
	← Schmallon left	07:28
	→ steveire joined	07:28
	← steveire left	07:29
	→ steveire joined	07:29
	← rchavik left	07:32
	← n2g left	07:35
	← heliostech left	07:36
	← mithro left	07:37
	← Camel_ left	07:37
	→ nevyn^__ joined	07:38
	← carutsu left	07:39
	← ExtraSpice left	07:39
	→ project2501b joined	07:40
	→ ExtraSpice joined	07:40
	← nevyn^_ left	07:40
	→ workmad3 joined	07:40
	← nevyn^__ left	07:41
	← tr3buchet left	07:41
	→ nevyn^_ joined	07:42
	← project2501b left	07:45
	→ project2501b joined	07:45
	→ barismetin joined	07:46
	← nevyn^_ left	07:46
	→ drizzd joined	07:48
	→ shruggar joined	07:49
	← SlickT10 left	07:51
	→ mithro joined	07:53
	→ tahseen joined	07:58
tahseen	wonders if there's a good solution if my index file got zeroed	07:58
cbreak_work	why?	08:00
	← priidu left	08:01
	→ Belna joined	08:03
	← unreal left	08:04
	→ unreal joined	08:05
	← valeo left	08:07
	← steveire left	08:10
	← poseid left	08:11
	→ steveire joined	08:11
	← steveire left	08:11
	→ steveire joined	08:11
	← unreal left	08:11
	→ unreal joined	08:12
	→ trivol joined	08:13
	← johnw left	08:13
	→ urkud joined	08:14
	← steveire left	08:15
	→ steveire joined	08:16
	→ bauruine joined	08:17
	← mattwynne left	08:17
	→ poseid joined	08:18
	→ nicoulaj joined	08:20
	← nicoulaj left	08:21
	→ priidu joined	08:23
	→ PointMan joined	08:24
PointMan	I removed a directory in a branch, then added them as submodules (git repos), then merge the branch into master. Now the directories are shown as modified, I cant rm or git submodule init to get rid of them..ideas?	08:25
	→ masterkorp joined	08:26
	→ ProperN[out] joined	08:27
	← ProperN[out] left	08:27
	→ ProperN[out] joined	08:27
	→ Schmallon joined	08:28
PointMan	ok Ive deleted the directories and committed the changes. However when I do git submodule init and git submodule update, nothing happens	08:29
	ideas please?	08:29
	→ froschi1 joined	08:29
	← ProperNoun left	08:31
	ProperN[out] → ProperNoun	08:31
	→ jony1 joined	08:33
	← jony left	08:33
cbreak_work	no idea about submodules, sorry	08:35
	← _Vi left	08:36
PointMan	I figured it out, I ended up removing the dirs manually and then readded them using git submodule add to get the repos files back	08:36
	← jony1 left	08:37
	← PointMan left	08:37
	← urkud left	08:38
tahseen	cbreak, my colleague's system rebooted and I'm finding the HEAD and index file zeroed out	08:39
	I copied the index from my working dir and it shows up now as a working git repo with old branches .. just not sure if its anything safe that I did :(	08:40
Ilari	tahseen: ext4 or xfs?	08:40
	← poseid left	08:41
	← dpnux left	08:41
tahseen	xfs it seems	08:42
Ilari	tahseen: 'git fsck --full' might be good idea to discover possible more corruption.	08:42
tahseen	it gives the same error :( ... not a git repository	08:43
	thats what I tried first	08:43
Ilari	tahseen: Restore .git/HEAD too ('ref: refs/heads/master' or something like that).	08:45
tahseen	yes did that and then copied my index	08:45
	it seems to be working showing the branch and commits but I'm not sure if this is the best thing to do	08:45
	or its actually gonna screw thigns up more	08:45
	→ mbroeker joined	08:46
	→ tatsuya joined	08:46
Ilari	tahseen: The repository detection is: 1) .git/HEAD exists and is valid. 2) .git/objects exists and is directory, 3) .git/refs exists and is directory.	08:47
	tahseen: You saying 'git log' works but 'git fsck --full' from same directory doesn't?	08:48
tahseen	sorry I missed one thing	08:48
	initially no git command was working	08:48
	← cirwin left	08:48
tahseen	it seeems after restoring .git/HEAD .. git fsck --full is working	08:48
	says notice: HEAD points to an unborn branch (master)	08:49
	fatal: object 03fb0d8fcb23c052836209869c1d7fe686c9e26d is corrupted	08:49
	I think I'm getting somewhere now	08:49
	not sure exactly where	08:49
Ilari	tahseen: '.git/objects/03/fb0d8fcb23c052836209869c1d7fe686c9e26d' is zero-length, right?	08:50
	→ rlotun joined	08:50
tahseen	yup it is	08:50
	tahseen wonders how he guessed that	08:50
	← TDJACR left	08:51
	← unreal left	08:52
Ilari	tahseen: You might generate list of all empty regular files (in .git and subdirs) at once (and pastebin that).	08:53
	→ unreal joined	08:53
	→ curtana joined	08:54
Ilari	tahseen: Considering how HEAD and index file are updated, HEAD or index file being invalid (neither old or new version) points to _serious_ bugs in the filesystem drivers.	08:56
	tahseen: invalid after sudden reboot, that is.	08:56
tahseen	yuck ..	08:57
	and it wasn't even being updated	08:57
	→ petercoulton joined	08:57
tahseen	she just did a fetch-merge and was building it got reboot	08:57
	so you want the zero length files in .git right?	08:58
	and its subdir .. oh got it	08:58
Ilari	I saw XFS do _ABSOLUTELY_ _INSANE_ stuff on sudden system reboot years ago. Any filesystem doing stuff like that is completely unacceptable.	08:59
tahseen	http://pastebin.com/tQhLf1qM	08:59
	← curtana left	09:00
FauxFaux	Nice objects.	09:00
Ilari	tahseen: You have '.git/logs/refs/heads/tms-redesign'?	09:01
tahseen	I remember hearing xfs zeroing unsynched blocks or something on reboot?	09:01
	← barismetin left	09:01
Ilari	tahseen: What I saw was _WAY_ worse than that.	09:01
frogonwheels	tahseen: hey - that's not delphi TMS is it?	09:02
tahseen	I mean does that issue cause this whole zeroing of files? or its even worse then that?	09:02
	→ offby1` joined	09:02
tahseen	nope	09:02
	its a module in our project	09:02
	the file is there just zero size	09:02
	→ bluenovember joined	09:02
tahseen	sorry missed the path	09:02
frogonwheels	:) worth asking.	09:02
tahseen	yup '.git/logs/refs/heads/tms-redesign'? is there	09:03
	so many objects zeroed out	09:03
Ilari	tahseen: Take seocnd hash in last line there and write it to 'refs/heads/tms-redesign'	09:03
	← offby1 left	09:04
Ilari	tahseen: (one of the zero length files). Follow that by LF. The result should be 41 byte file.	09:04
tahseen	./objects/f5/9cd21aaea2de2f1924031d92026e5cc9752ad4 .. you mean the part after f5/ ?	09:04
Ilari	tahseen: No.	09:05
	tahseen: What's the last non-blank line in '.git/logs/refs/heads/tms-redesign'?	09:05
	→ apsysoev1 joined	09:06
tahseen	oh	09:06
	← jtauber left	09:06
tahseen	there's just one long line	09:06
	no blank one	09:06
Ilari	tahseen: That line has two 40 char hash values, timestamp and some message, right?	09:07
	← Mage left	09:07
tahseen	yup commiter and some message	09:07
	so put the second hash in refs/heads/tims-redesign?	09:07
Ilari	tahseen: Yes.	09:08
	tahseen: That should result refs/heads/tims-redesign being 41 bytes.	09:08
tahseen	one thing I don't really need the tms-redesign branch	09:08
workmad3	tahseen: so prune it	09:08
tahseen	the actual development is available in anther repo	09:08
	cool	09:08
Ilari	Bad idea with corrupt repo!	09:09
workmad3	hmm, good point	09:09
	don't prune anything until you've uncorrupted your repo :)	09:09
tahseen	just letting you know if it helps	09:09
	got it :D	09:09
Ilari	tahseen: All those blank files in objects/ look like they should be removed (but don't delete any other files there!)	09:10
	← hobodave left	09:10
divVerent	I have yet another crlf problem	09:10
	→ hobodave joined	09:11
divVerent	is there NO way to get git to handle crlf properly in a repo without havint to manually set config settings?	09:11
cbreak_work	sure. just turn off autocrlf	09:11
divVerent	*.map -crlf	09:11
	*.mapinfo crlf=input	09:11
	I _need_ autocrlf	09:11
cbreak_work	why?	09:11
divVerent	or all the windows users commit CRLFs	09:11
cbreak_work	git can deal with that I think	09:11
divVerent	not really	09:11
cbreak_work	it'll just be ugly	09:11
frogonwheels	yeah you can't have CRLFs committed to a git repo if you want anything to work.	09:11
divVerent	it shows as lots of changes in blame then	09:11
frogonwheels	cbreak_work: nah it can't.	09:12
divVerent	I want the files to stay with LFs in the repo	09:12
frogonwheels	cbreak_work: I mean core git can't, but diff/blame doesn't	09:12
cbreak_work	it will only show up as change if they change lines	09:12
divVerent	some users even commit mixed CRLF-LF files with some IDEs	09:12
tahseen	ok 41 byte it is	09:12
	→ elmob joined	09:12
divVerent	cbreak_work: haha, IU got the opposite here	09:12
frogonwheels	divVerent: yes, you definitely want o have LF only in the repo :)	09:12
cbreak_work	if they only commit theirr own changed lines or new files with CRLF it will work no problem	09:12
tahseen	so now I should all zero length file in ./git/objects Ilari ?	09:12
divVerent	so I am trying to use gitattributes to force the text files to LF	09:12
Ilari	tahseen: Yeah, remove those files.	09:12
tahseen	ok	09:12
divVerent	however, the option crlf=input seems to only work if core.autocrlf is set	09:13
frogonwheels	divVerent: depends on the version.. there's now a git.lf or soemthing	09:13
divVerent	and that defeats the point of setting that in gitattributes	09:13
	does the eol=lf setting work even if core.autocrrlf is off?	09:13
	in that case, I'd at least be fine in git.git :P	09:13
frogonwheels	divVerent: it apparently works ONLY when it's not set	09:13
divVerent	very odd	09:14
frogonwheels	divVerent: but isn't that about local files..not repo files?	09:14
divVerent	crlf=input in the attributes is, undocumentedlys, ignored if core.autocrlf is not set	09:14
	so people STILL commit CRs all the time	09:14
frogonwheels	divVerent: I was under the impression that with eol=* it will always store lf only in the repo.	09:14
divVerent	so was I also under the impression about crlf=input when reading the manpage	09:14
	so I simply want to know - is eol= reliable?	09:14
Ilari	tahseen: 'grep tms-redesign .git/packed-refs'. Any hits (missing file counts as no hits)?	09:15
frogonwheels	divVerent: I have no idea, sorry.	09:15
divVerent	basically, the problem is that the gitattributes file seems to work for some people but not for others	09:15
	← hobodave left	09:15
divVerent	so some commit CR stuff and some not	09:15
	and yet others get merge conflicts between	09:15
	real hell broke out from this	09:15
FauxFaux	Social problem; hurt the Windows users.	09:15
frogonwheels	divVerent: yuck. that sounds like you're gonna have to filter-branch all the crs away :(	09:15
	→ curtana joined	09:16
divVerent	frogonwheels: I'd like to avoid that	09:16
	I am fine if all NEW files are sane	09:16
	I just want to someone enforce no FUTURE CR commits come	09:16
tahseen	there's no .git/packed-refs	09:16
Ilari	divVerent: Hooks on serveside.	09:16
divVerent	just saying: svn can handle this well	09:16
	why not git?	09:16
	← Will\| left	09:17
divVerent	according to the DOCUMENTATION, git c an	09:17
	→ gebi joined	09:17
frogonwheels	divVerent: I think the problem is that there's been a bit of a transition about how this should be handled.	09:17
divVerent	in svn I simply set svn:eol-style for a file, and then it is fine and stays fine	09:17
Ilari	tahseen: Actually, if you don't need tms-redesign, 'mv .git/refs/heads/tms-redesign .git/tms-redesign.head.bak' and 'mv .git/logs/refs/heads/tms-redesign .git/tms-redesign.log.bak'.	09:17
divVerent	and works for about any existinmg svn version in the wild	09:18
	← kenyon left	09:18
cbreak_work	just blame the windows users.	09:18
	let them shoot their feet off	09:18
divVerent	frogonwheels: transition?	09:18
	so you say, this is a bug part of the crlf=input to eol=lf transition?	09:18
cbreak_work	if they want to turn on auto-crlf, they can for their repositories	09:18
divVerent	BTW, does it make sense to specify BOTH attributes in the .gitattributes file?	09:18
cbreak_work	it has to be done at commit time.	09:18
Ilari	tahseen: And also similarly for .git/refs/remotes/yan/tms-redesign and .git/logs/refs/remotes/yan/tms-redesign	09:18
	→ HG` joined	09:19
divVerent	cbreak_work: I want it enforced by the repository	09:19
Ilari	divVerent: update hook.	09:19
divVerent	not have to tell them to set an obscure option in git	09:19
	isn't that exactly what .gitattributes is for?	09:19
FauxFaux	What editor are they using that doesn't support global configuration against crlf?	09:19
	← unreal left	09:19
	→ unreal joined	09:20
Ilari	tahseen: After that, you should only have empty FETCH_HEAD and empty index file.	09:20
	tahseen: (oh and one of those moveaside files can be also empty)	09:20
	→ Paraselene_ joined	09:21
tahseen	So I now have FETCH_HEAD and index as zero length	09:22
	→ didlix joined	09:22
tahseen	do I run fsck now ?	09:22
Ilari	tahseen: 'git fsck --full'.	09:23
	→ GarethAdams\|Work joined	09:24
	→ docelic joined	09:25
docelic	Hey folks, how do I see the diff between the branch I am currently on, and another branch of choice?	09:26
tahseen	notice: HEAD points to an unborn branch (master)	09:26
	fatal: index file smaller than expected	09:26
divVerent	no idea which editor	09:26
	but anyway	09:26
Ilari	tahseen: Remove .git/index	09:26
	tahseen: That should allow it to proceed and not hit that second error.	09:27
	→ malditogeek joined	09:27
tahseen	index is just for staging right, its not part of the committted data is it?	09:27
	→ kenyon joined	09:28
shruggar	correct	09:28
	← unreal left	09:28
tahseen	there's just lots of dangling blob and the error about HEAD	09:29
	→ unreal joined	09:29
shruggar	index is a "faster to use, but less-sensible-looking" format, which also contains some extra data (like mtime) which commits don't care about	09:29
	← jksz left	09:29
Ilari	tahseen: 'git fetch yan'.	09:29
tahseen	so when I do a git log does it look at index then rather than looking at the commits?	09:29
	← hyperair left	09:29
shruggar	tahnseen: the index isn't involved in "git log" in any way, and by default the commit-based equivalent (trees) are not involved either	09:30
malditogeek	Hi guys, weird problem here, I accidentally created a branch called "-t" but now I can't switch, rename or delete that branch! the cli interprets the name "-t" as an option and does nothing. Also tried escaping the dash but then it can't find a branch with that name… any clue?	09:30
tahseen	where does it get the info then I wonder	09:30
divVerent	worst part about the crlf breakage is that even git reset --hard does not clean up the work tree	09:30
	thwere seems to be no way to recover	09:30
	→ barismetin joined	09:31
tahseen	ok Ilari .. now git branch and git commands on other branch except master seem to work	09:31
	why 'git fetch yan'? to fix master?	09:31
shruggar	tahseen, what info?	09:31
tahseen	'git log'	09:31
shruggar	that's not info, that's a command :P	09:31
Ilari	tahseen: Try restoring the backups of refs/heads/tms-redesign and logs/refs/heads/tms-redesign (you moved those aside, right?)	09:32
	← vu3rdd left	09:32
tahseen	did 'git fetch yan' .. still same issue with master	09:32
shruggar	tahseen, what info used by "git log" are you talking about?	09:32
tahseen	yup backed them up	09:32
	I meant where do 'git log' gets the data to generate the log :)	09:32
	sorry if I'm not making any sense	09:32
Ilari	tahseen: Then try 'git fsck --full' and see if it throws more complaints than about HEAD and about dangling stuff.	09:32
	→ powr-toc joined	09:33
	→ jksz joined	09:34
	← Zenopus left	09:34
shruggar	tahseen, by default "git log" only shows authors, dates, and messages. Those are all part of the commit object. The tree objects are where file names and hierarchies are stored. Information derived from them can be viewed by "git log", but by default they are not used.	09:34
	← bauruine left	09:34
	→ bauruine joined	09:35
tahseen	so when is index used apart from when I add staff and look at status or diff --cached?	09:35
shruggar	tahseen, when you commit stuff, it is used to generate the tree which gets referenced by the commit object which is created	09:35
tahseen	yup thats sort of what I understand	09:36
	Ilari, still the same errors on fsck	09:36
Ilari	tahseen: Change HEAD to read something like 'ref: refs/heads/tms-redesign'. That should take care of HEAD error.	09:37
shruggar	tahseen, I think merges also take place directly in the index	09:37
tahseen	luckily it seems she didn't any uncommited works on master	09:37
	the other branches seem to be accessible, so those should work fine right?	09:37
	→ d0k joined	09:37
	→ Zenopus joined	09:37
	→ nevyn^_ joined	09:38
	← jksz left	09:38
	← roger_padactor left	09:39
	→ jksz joined	09:40
	→ kazoo joined	09:42
	← slakmagik left	09:43
Ilari	tahseen: If you only get dangling errors after that, do 'git reset' (no other parameters or arguments!!!)	09:45
	→ poseid joined	09:45
Ilari	tahseen: (that should rewrite the index file)	09:45
	→ slakmagik joined	09:46
	→ MattDiPasquale joined	09:48
	← steveire left	09:48
	→ steveire joined	09:49
	← steveire left	09:49
	→ steveire joined	09:49
tahseen	well the dangling blob's are still there on git fsck but thigns seem be working just fine with them	09:49
Ilari	tahseen: Dangling blobs are normal and are generated a lot.	09:50
	tahseen: Then look at 'git status' and see what local modifications it prints.	09:51
	← unreal left	09:51
	→ unreal joined	09:51
	← MattDiPasquale left	09:52
robbe-	Ahh, some rest can do wonders, I think I got it all working now. Trying to configure apache and not knowing a lot about it can be a pain, apparently. :D	09:52
	Is there a way I can be /sure/ that I'm serving smart http btw?	09:52
tahseen	its just showing the regular files that got zeroed	09:53
	thanks a lot Ilari :) for taking such patience to walk me through with it	09:53
	seems I better work hard in getting rid of these xfs partitions at the office	09:53
	whats the reputation of ext4 on such issues? do you find that safe enough or recommend ext3 ?	09:54
	→ Yuuhi joined	09:55
Ilari	tahseen: On latest kernel versions, ext4 is sane, but on earlier versions it was bad.	09:55
	tahseen: Of course, ext3 is known to not have insane stuff in it...	09:56
tahseen	2.6.30 something should be ok right?	09:56
	→ courtain joined	09:56
tahseen	gotta run, seems I need to shutdown the machine	09:57
Ilari	tahseen: I think that's new enough...	09:57
tahseen	once again thanks a lot, though I have to wonder about what we did :)	09:58
	but talk on that later	09:58
	← tahseen left	09:58
	← curtana left	09:58
	← courtain left	10:01
	← mithro left	10:01
	→ mattwynne joined	10:02
	← steveire left	10:02
	← brizly left	10:02
	→ brizly joined	10:03
	→ steveire joined	10:03
	← steveire left	10:03
	→ steveire joined	10:03
	→ jtdowney joined	10:03
	← unreal left	10:04
	→ unreal joined	10:04
	→ [1]intripoon joined	10:07
	← petercoulton left	10:07
	← poseid left	10:09
	→ tcurdt joined	10:09
	← kazoo left	10:09
	→ urkud joined	10:09
	← intripoon left	10:10
	[1]intripoon → intripoon	10:10
	→ mountains joined	10:10
mountains	Hi - if I want a git repo to use a different branch than "master" by default, is this possible?	10:10
shruggar	mountains, as in, when you clone?	10:11
mountains	Yes.	10:11
shruggar	mountains, edit the file "HEAD" in the .git directory. The format should be obvious	10:11
cbreak_work	you can rename branches	10:11
shruggar	mountains, (this assumes a bare repository)	10:12
	→ poseid joined	10:12
cbreak_work	I think if you rename them with git branch, git will adapt the settings for it, and other things	10:12
	maybe also the bare HEAD	10:12
	→ courtain joined	10:12
cbreak_work	there's git remote set-head. Probably not neded if you do what shruggar said	10:13
shruggar	cbreak, afaik it does not. iirc, I have pushed into a bare repository which never had a "master" branch, and it still defaulted to master	10:13
cbreak_work	:/	10:13
	→ yaotti_aw joined	10:14
	yaotti_aw → yaotti	10:14
shruggar	(as in, it gave an error when trying to check out master after clone)	10:14
	set-head seems like a more sensible way of doing things, though	10:14
	← poseid left	10:16
	→ adrien joined	10:16
	→ tvw joined	10:18
Ilari	AFAIK, if remote HEAD points to existing branch, it doesn't just default to some nonexistent branch and complain on clone.	10:19
	→ Dave^\| joined	10:19
	→ justin-george joined	10:21
	← eighty4 left	10:26
shruggar	to reproduce what I was talking about: http://gist.github.com/482787	10:28
	→ eighty4 joined	10:31
	→ marcob joined	10:31
	yaotti → yaotti_aw	10:31
	← docelic left	10:32
	← mbroeker left	10:32
	→ murr4y joined	10:34
	→ malumalu joined	10:34
	→ alexmchale joined	10:35
Chaser_	Hi. I have a bare repository of an upstream one. To update the bare repository's master branch to upstream's master I need to do git fetch origin master - correct ?	10:37
	← bauruine left	10:39
	← steveire left	10:40
	→ steveire joined	10:41
	→ poseid joined	10:42
	→ gemmes joined	10:43
	→ Joeconyers joined	10:44
gemmes	Hi all, I just installed git and seems to have went well. Should I now remove the git folder and files I used for the install that is in my root folder?	10:45
	← adrien left	10:46
Pepsi1x1	If I want to make a .patch file to send to the owner of a repo, it the correct command to issue: git diff -p > patch.diff	10:51
	say I have local changes that I want to send to them so they can apply it	10:51
	because apparantly im being told that a corrupt patch has been produced	10:53
shruggar	Pepsi1x1, usually one uses: git format-patch <upstream>	10:54
Pepsi1x1	ok ill give that w whirtl then, thanks shruggar	10:55
shruggar	gemmes, how did you install git? what root folder?	10:55
Pepsi1x1	a whirl*	10:55
gemmes	default folder when i login	10:55
shruggar	gemmes, your home directory?	10:56
gemmes	yes sorry	10:56
robbe-	Hmm, is there a way one can pull in changes in a bare repo, or is the only option to push them?	10:56
	← malumalu left	10:56
shruggar	gemmes, how did you install git? What steps were involved? were you following a guide?	10:56
	→ malumalu joined	10:56
	→ ereslibre joined	10:56
	← ereslibre left	10:56
	→ ereslibre joined	10:56
	← cworth left	10:56
	→ treak007_ joined	10:57
gemmes	yes and git is working	10:57
	← gebi left	10:57
	← steveire left	10:58
shruggar	gemmes, I didn't ask any yes-or-no questions	10:58
	→ cworth joined	10:58
robbe-	Well, actually, the last one is a yes-or-no question.	10:58
	robbe- ducks	10:58
gemmes	long story, i installed x2, one way very long using curl second way shorter.	10:59
shruggar	gemmes, I assume the answer is "yes, you can remove all the files you used to install git or subdirectories of your home dir", but I have no idea what steps you took to install git, what guide you may have been following, etc, so I have no idea.	10:59
	← mattwynne left	10:59
gemmes	first way : http://www.angry-fly.com/post.cfm/installing-git-on-centos-5	11:00
shruggar	robbe-, okay, you're right :)	11:00
	→ steveire joined	11:00
	← treak007 left	11:00
	robbe- comes out of hiding	11:00
robbe-	:-)	11:00
	→ felipec joined	11:00
	← eighty4 left	11:01
	← GarethAdams\|Work left	11:01
	← offby1` left	11:01
	← DrNick left	11:01
	← tyru left	11:01
	← JEEB left	11:01
	← binjured left	11:01
	← leavittx left	11:01
	← nsussman left	11:01
	← tango_ left	11:01
gemmes	I cant find page i used for 2nd install but it was a lot shorter. yum install blah then make alias, etc	11:02
	cpg → cpg\|away	11:03
	→ eighty4 joined	11:03
	→ GarethAdams\|Work joined	11:03
	→ offby1` joined	11:03
	→ DrNick joined	11:03
	→ tyru joined	11:03
	→ JEEB joined	11:03
	→ binjured joined	11:03
	→ leavittx joined	11:03
	→ nsussman joined	11:03
	→ tango_ joined	11:03
shruggar	gemmes, well if you just did "yum install git", than there's no $HOME-related files involved. so yeah, you can delete those :)	11:04
gemmes	thanks all, fingers crossed	11:05
	shruggar wants the whole world to install things like GoboLinux does	11:05
	yaotti_aw → yaotti	11:06
Pepsi1x1	ok so to clarify my earlier question	11:06
	the method with format patch is	11:06
	commit	11:06
	git format-patch master	11:06
shruggar	Pepsi1x1, please type a complete thought, then hit enter	11:07
	→ JEEBcz joined	11:07
	→ herdingcat joined	11:07
Pepsi1x1	hehe ok	11:07
shruggar	'cause I don't know if you're done asking your question yet :)	11:07
Pepsi1x1	To clarify my earlier question, to produce a patch file, I would first commit my intended changes, then I should issue the command git format-patch master. And this will produce a patch file for each commit	11:08
shruggar	usually the workflow is: git checkout -b sometopic origin/master; (do some work); git commit; (do some more work); git commit; git format-patch origin/master	11:09
	and that will produce a patch ready for sending via git send-email, one file per commit	11:09
	← JEEB left	11:10
Pepsi1x1	fantastic, i feel confident in that now! Thank you for your help once again shruggar.	11:10
shruggar	Pepsi1x1, you may want to use git format-patch --signoff, depending on what project you're submitting to	11:11
gemmes	removed files, git still working :) thanks again.	11:12
	→ ralf_e joined	11:12
	← [Po]lentino left	11:13
	→ Paraselene__ joined	11:14
	→ thiago_home joined	11:15
	← Paraselene_ left	11:15
	Paraselene__ → Paraselene_	11:15
	→ treak007 joined	11:17
	← treak007_ left	11:20
	frakturfreak_ → frakturfreak	11:21
	→ slonopotamus joined	11:24
	→ telemachus joined	11:26
	→ sylr joined	11:32
	← alexmchale left	11:33
	← urkud left	11:33
	→ _Vi joined	11:34
	→ urkud joined	11:34
	← priidu left	11:34
	→ Raydiation joined	11:35
	← psankar left	11:35
	→ roger_padactor joined	11:35
	← opalka left	11:36
	→ hyperair joined	11:37
Raydiation	if i do git --bare init in a full directory will it delete all files?	11:38
	← telemachus left	11:38
Raydiation	and whats the difference to normal git init?	11:38
	→ opalka joined	11:39
shruggar	Raydiation, 1) no 2) it makes a repository with no work-tree (ie, just the ".git" directory's files, no .git folder, no checkout)	11:39
	Raydiation, so it won't delete all files, but it won't actually be useful	11:40
Raydiation	shruggar: because i set up a repository with git init, added files, commited and then cloned it to my pc. after pushing some changes it said it will fk up the index if git reset --hard isnt called	11:41
	← slonopotamus left	11:41
	← Determinist left	11:41
shruggar	ah, that's nice of it. I didn't know it had a sane error message.	11:42
	Raydiation, faq non-bare	11:42
Gitbot	Raydiation: Pushing to non-bare repositories is discouraged. See https://git.wiki.kernel.org/index.php/GitFaq#non-bare	11:42
Raydiation	the error msg was really verbose but it didnt explain good enough why it happened, more what to do next	11:43
shruggar	Raydiation, in short: "push" doesn't have anything at all to do with work-trees. It only pushes refs. So if you push into a ref which is checked-out, the work-tree doesn't actually get updated.	11:43
	Raydiation, the intended workflow is to push to something or somewhere that doesn't have a work-tree, then merge that from wherever you want it to be. The simplest (in terms of explanations) way to do this is to create an intermediate "bare" repository	11:44
	← jtdowney left	11:45
	→ raichoo joined	11:48
	← __ed left	11:48
Raydiation	shruggar: ah k, i suppose i can use the post-update hook linked in that side, i have a django programm on that server and i want to update it via git	11:49
	yaotti → yaotti_aw	11:49
shruggar	Raydiation, it's a common need. Keywords for googling include: post-update hook, post-receive hook, and deployment	11:49
Raydiation	shruggar: yes i used a very easy hook but i still got the msg, thats why i was buffled	11:50
shruggar	Raydiation, please note that many people recommend using "git pull" in deployment scripts, which utterly stupid and will break your server	11:50
Raydiation	cd .. && env -i git reset --hard	11:50
	→ telemachus joined	11:50
shruggar	looks right to me	11:50
Raydiation	why will it break it?	11:51
	for instance i want my friend to get a backup too with git	11:52
	and he should be able to commit changes too	11:52
shruggar	two reasons I can think of off-hand: "pull" is actually a merge, which can have conflicts. And non-fast-forward pushes will not be properly handled.	11:52
Raydiation	that sounds like its nearly impossible to do that :/	11:53
	← Aikawa left	11:53
shruggar	What's impossible?	11:54
Raydiation	shruggar: work together with my friend updating the scripts via git	11:54
	→ slonopotamus joined	11:54
	← HG` left	11:54
shruggar	what about it do you think would be non-straightforward?	11:55
	← raichoo left	11:55
	→ mru joined	11:56
	← henkboom left	11:57
Raydiation	shruggar: ?	11:57
shruggar	Raydiation, I haven't got the slightest clue why you think it would be impossible, or difficult, or anything less than easy	11:58
	← gemmes left	11:59
Raydiation	shruggar: ah ok, i missunderstood the stuff with the deployment scripts	11:59
	→ thinker_by joined	11:59
	→ henkboom joined	12:00
	→ priidu joined	12:00
	→ Dawgmatix joined	12:00
	→ antgel joined	12:00
	← courtain left	12:01
	→ courtain joined	12:01
	→ newbie\|2 joined	12:01
	courtain → cortana	12:02
	→ alexmchale joined	12:02
	← sylr left	12:03
deavid	about gitolite: it says it requires git 1.6.2, but i feel it's because there are problems cloning empty repos in prior versions... if it's the only reason, I could use older git versions in some situations (by the way, i don't know if there's a better channel for gitolite specifics)	12:03
Raydiation	shruggar: ah ok, heres the problem again http://pastebin.com/T8ddDNAw	12:03
	i used the post update script	12:04
shruggar	deavid, this is the right channel to ask about gitolite. And how old a git version are we talking about?	12:04
Raydiation	do i have to explicetly silence it because it shows this msg every time?	12:04
	← mru left	12:04
sitaram	deavid: I got tired of the support load, and since 1.6.2 is old enough, I just went with that restriction. You can try and make it work by removing that line if you like, but if it's older 1.6 then you can expect other issues	12:04
deavid	shruggar: 1.5.x i think. Debian lenny's version	12:04
	← telemachus left	12:04
sitaram	deavid: sorry... upgrade!	12:04
	(or at least I can't support it... no time!)	12:05
	← thinker_by left	12:05
shruggar	Raydiation, that error message tells you exactly what to do if you want to ignore the message. But instead: DON'T IGNORE THE MESSAGE :P	12:05
	→ hipe joined	12:05
deavid	sitaram: i don't want so much support for it, I only wanted to know if there's a big reason for that	12:06
	→ bentob0x joined	12:06
shruggar	Raydiation, just push somewhere other than the checked-out ref. Or check out another ref. reset --hard some/other/ref works fine and doesn't change what ref you're on	12:06
deavid	i guess i'll end updating git packages from backports in these workstations	12:06
shruggar	deavid, if you're using git 1.5.x it will be hard to get support in here for anything at all, really. :) Everyone will eventually say "wait, what version are you using?" followed by "upgrade!"	12:07
	← cortana left	12:07
deavid	shruggar: ok, so i guess it will give some strange errors	12:07
	→ cortana joined	12:07
deavid	i wanted to use a newer version for server, but i know some clients are old	12:07
	i can give support to my users if are common things on git	12:08
FauxFaux	Speaking of releases, how does the rc process work? I notice 1.7.2 is up to rc3.	12:08
	Not that it has any cool features or fixes in. :)	12:09
antgel	hi people. please see my paste at http://paste.scsys.co.uk/46910. this pull command used to work, i'm sure	12:09
	← cortana left	12:09
Pieter	antgel: you probably checked out a different branch	12:09
FauxFaux	antgel: git remote -v probably lists some other names for that url.	12:09
	→ sylr joined	12:09
antgel	where is my "default configured remote" specified? it might have changed when knittl helped me fix the repo recently	12:09
Raydiation	shruggar: what do you mean with push somewhere other? do i have to initialize a a branch on the server which then merges with the master branch after post-update?	12:10
antgel	Pieter: i only have one branch (master)	12:10
	→ Ryback_ joined	12:10
	→ curtana joined	12:10
	→ skratool joined	12:10
	← sylr left	12:10
	← unreal left	12:10
skratool	how can I checkout one file from other local branch, to local working tree?	12:11
knittl	antgel: shouldn't have changed	12:11
	→ mcella joined	12:11
antgel	FauxFaux: that command lists that url twice, as fetch and push	12:11
knittl	but it's in .git/config	12:11
	→ unreal joined	12:11
slonopotamus	skratool: git checkout <branchname> -- <path>	12:11
knittl	antgel: this is a different repo now, isn't it?	12:11
	→ gebi joined	12:12
	→ sylr joined	12:12
antgel	knittl: no, same one. but in the public post, i substituted $something with "munge" :)	12:12
knittl	antgel: what branch are you on? (git branch)	12:13
antgel	knittl: master	12:13
	git remote returns "origin"	12:13
knittl	proto.munge.com is not your default origin	12:14
	→ Bass10 joined	12:14
skratool	slonopotamus: thanks!	12:15
knittl	antgel: you'd have to do git pull <repo> master	12:15
	→ beni- joined	12:15
	→ telemachus joined	12:15
antgel	knittl: http://paste.scsys.co.uk/46911	12:15
	ah	12:15
	looking good. my crib sheet must have been incorrect	12:16
knittl	but it is your default remote. strange it wouldn't pull from master	12:16
	antgel: make sure to not pull outdated/removed history ;)	12:17
antgel	knittl: don't, just don't :)	12:17
knittl	hm?	12:17
	← telemachus left	12:17
	→ parasti joined	12:17
antgel	as in, don't talk to me about fubaring the repo	12:18
knittl	i'm simply telling you to check history after pulling	12:18
	← sylr left	12:18
antgel	knittl: british humour	12:18
	← henkboom left	12:18
knittl	after all you don't want to go through the same procedure of rebasing again ^^	12:18
antgel	looks good	12:18
knittl	ok, the brits	12:18
	→ sylr joined	12:19
	← alexmchale left	12:20
	→ henkboom joined	12:21
	yaotti_aw → yaotti	12:22
	← poseid left	12:23
	→ mar77i joined	12:25
	← sylr left	12:25
	→ TheNewAndy joined	12:26
	← skratool left	12:26
	→ sylr joined	12:26
	← urkud left	12:27
	→ JohnTheSavage joined	12:29
	→ josh0x0 joined	12:29
	← hipe left	12:31
	→ raichoo joined	12:31
	→ sylr_ joined	12:32
	← sylr_ left	12:33
	→ Adaptee joined	12:33
	← pgib left	12:34
	← steveire left	12:34
	→ steveire joined	12:36
	← steveire left	12:36
	→ steveire joined	12:36
	← girishr left	12:36
	← felipec left	12:40
	→ t0rc joined	12:40
	→ Bass2 joined	12:40
	→ tchan1 joined	12:40
	← Bass2 left	12:41
	← andrewl left	12:42
	→ Bass2 joined	12:42
	← travlr left	12:42
	← tchan left	12:42
	→ travlr joined	12:42
	← Bass10 left	12:43
	→ bauruine joined	12:43
	→ Tricia joined	12:44
	→ IslandUsurper joined	12:46
Rhonda	I have a question about this blog article: http://feeding.cloud.geek.nz/2010/07/cherry-picking-range-of-git-commits.html	12:46
	The commit IDs are given in short form - how to I get that? The mentioned "tree" alias doesn't give it for me.	12:46
frsk	Does it work if you do "git log --pretty=oneline --decorate --graph"?	12:47
mar77i	that's the alias :)	12:48
	→ LiamH joined	12:48
mar77i	it's funny, because I get all sha1 sums, too	12:48
Rhonda	frsk: That gives me the complete commit hashes, not the short ones that are displayed in the blog entry. :/	12:48
	← steveire left	12:48
frsk	aha	12:48
	So it's not an issue with aliases, but git log	12:48
	Rhonda: git version?	12:49
Rhonda	1.7.1	12:49
	frsk: Does it give you short commit IDs?	12:49
frsk	Rhonda: No, but if I do --oneline instead of --pretty=oneline, it does :-)	12:50
	→ dropdrive joined	12:50
frsk	Consistent behaviour..	12:50
Rhonda	Oh, right!	12:50
	--online works for me too.	12:50
frsk	great =)	12:50
	→ steveire joined	12:50
Rhonda	Now could this be considered a bug when --pretty=online behaves different to --online?	12:50
dropdrive	Is there a way to extract (from a huge repo) a new repo that contains only the history for one file? filter-branch seems to work on subdirectories only.	12:51
Rhonda	HAH!	12:51
	frsk: --abbrev-commit	12:51
	man git-log said --online is shorthand for --pretty=online --abbrev-commit	12:51
Gitbot	Rhonda: the git-log manpage can be found at http://git.or.cz/man/git-log	12:51
mountains	dropdrive: Start it from the root of the repo?	12:51
mar77i	cool	12:51
	--oneline	12:51
mountains	dropdrive: filter-branch is the right tool for this.	12:51
	Rhonda thanks frsk for pointing right direction. :)	12:51
frsk	Rhonda: And thanks for explaning, I got a bit confused myself :)	12:52
mar77i	not --online xD	12:52
dropdrive	mountains: Okay, thanks, I'll give it a shot.	12:52
	→ jtauber joined	12:53
Rhonda	mar77i: feel free to keep my typos ;)	12:53
	← aziz left	12:54
	→ aziz_ joined	12:54
	→ tstclair joined	12:55
mar77i	chuck norris doesn't approve typos :)	12:55
	→ nataraj joined	12:58
Rhonda	mar77i: It seems git does neither. …	12:59
	fatal: unrecognized argument: --online	12:59
	← bauruine left	13:00
	→ bauruine joined	13:00
	→ simmel joined	13:04
	→ TeckniX joined	13:04
	← slonopotamus left	13:05
	← justin-george left	13:05
nataraj	How do i 'push' to origin branch on remote host?	13:06
workmad3	nataraj: depends on if your local branch is a tracking branch or not	13:06
simmel	I tried to search in the man-page, but what takes precedence of the git-configs? .git/config -> ~/.gitconfig → /etc/gitconfig?	13:06
	← marc_kdab_ left	13:06
	→ hipe joined	13:08
elb0w	Does git have file size limits?	13:08
	→ flo` joined	13:08
flo`	hi	13:08
	→ marc_kdab_ joined	13:09
	→ jrmuizel joined	13:09
flo`	may i make within a git tree, and do i have to make clean before using git pull or git bisect?	13:09
simmel	Seems like it was how I thought it was.	13:09
	← hipe left	13:09
nataraj	workmad3, local master has to be tracking branch by default?	13:10
workmad3	your local master will track origin/master by default, yes	13:11
	← simmel left	13:11
workmad3	unless you created the repo locally and are pushing for the first time	13:11
nataraj	if i merge to local master, i can then merge to the remote/origin?	13:11
workmad3	nataraj: sort of... what you'd do is merge your branch into master and then push your master to origin/master	13:12
	→ hipe joined	13:12
	← marc_kdab_ left	13:12
	← steveire left	13:12
	→ rseifert joined	13:12
cbreak_work	you can not merge to remotes. Merging is local only. What you can do is fast-forward, which is what push does	13:12
workmad3	^^	13:13
	← hipe left	13:13
	→ steveire joined	13:13
	← steveire left	13:14
	→ steveire joined	13:14
workmad3	nataraj: or you could just push your local branch to the remote with a 'git push origin local_branch_name:remote_branch_name'	13:14
	and I think that will then make that local branch a tracking branch of the new remote branch	13:15
nataraj	ok	13:17
	← squentin left	13:17
	→ marc_kdab_ joined	13:17
	← jrmuizel left	13:17
	← marc_kdab_ left	13:18
	← nataraj left	13:18
	→ marc_kdab_ joined	13:18
	→ sivy joined	13:18
	← roop left	13:19
	← marc_kdab_ left	13:19
	→ pgib joined	13:19
	→ marc_kdab_ joined	13:20
	← powr-toc left	13:20
	← marc_kdab_ left	13:23
	→ brntbeer joined	13:23
	→ tizbac joined	13:23
	→ marc_kdab_ joined	13:24
	← marc_kdab_ left	13:25
	→ ChUrP joined	13:25
	→ HG` joined	13:26
	← pgib left	13:26
	→ marc_kdab_ joined	13:26
	← marc_kdab_ left	13:28
	→ marc_kdab_ joined	13:29
	← maleknet left	13:29
	← marc_kdab_ left	13:31
	→ marc_kdab_ joined	13:32
	→ spearce joined	13:32
	← spearce left	13:32
	→ dids joined	13:32
	→ spearce joined	13:33
	← raichoo left	13:33
	← ChUrP left	13:33
	← marc_kdab_ left	13:33
	→ sammcd joined	13:33
	→ martengo joined	13:34
	→ mFacenet joined	13:34
	→ ChUrP joined	13:34
	→ marc_kdab_ joined	13:34
	← didlix left	13:35
	dids → didlix	13:35
	→ squentin joined	13:35
	→ khelll joined	13:37
	→ MattDiPasquale joined	13:37
	→ __ed joined	13:38
	← mFacenet left	13:38
	→ ChUrP1 joined	13:39
	← unreal left	13:39
	→ unreal joined	13:40
	← ChUrP left	13:40
	← sylr left	13:41
	yaotti → yaotti_aw	13:41
	→ sylr joined	13:43
	← aziz_ left	13:45
	→ aziz_ joined	13:45
	← tizbac left	13:45
	→ scarabx joined	13:46
	→ level09 joined	13:46
	← brntbeer left	13:47
	→ thomas_b joined	13:47
	→ girishr joined	13:50
	→ mFacenet joined	13:50
	→ kukks joined	13:52
	→ hugoxrosa joined	13:53
	← level09 left	13:53
	→ steffkes- joined	13:54
	→ ChUrP joined	13:54
	← scarabx left	13:55
	← unreal left	13:55
	→ rayners joined	13:55
	→ unreal joined	13:56
	← steffkes left	13:56
	← ChUrP1 left	13:56
	← bauruine left	13:57
	→ masonkramer joined	13:57
	← flazz left	13:57
	→ durarara joined	13:58
	← TheNewAndy left	13:58
	← masonkramer left	13:58
	→ flazz joined	13:59
	→ beatak joined	13:59
	→ ehabkost joined	13:59
	→ xuxc000 joined	13:59
	→ level09 joined	14:00
	→ bdiego joined	14:00
	→ loincloth joined	14:01
	← freimatz left	14:01
	→ bauruine joined	14:01
	← kukks left	14:03
	← _Vi left	14:03
	← ianmcorvidae left	14:03
	→ ianmcorvidae joined	14:03
	→ _Vi joined	14:03
charon	elb0w: not really but most operations require at least twice the size of the largest blob processed in terms of RAM	14:04
	← merlin83 left	14:04
	← _Vi left	14:04
charon	elb0w: so working with large (hundreds of MB or more) files isn't so practical unless you have a ton of memory	14:04
	→ _Vi joined	14:04
	← mFacenet left	14:05
	← EricInBNE left	14:06
	→ smuf joined	14:06
	→ Remixman joined	14:06
	← HG` left	14:07
	← unreal left	14:08
	→ unreal joined	14:08
	→ EricInBNE joined	14:09
	→ mFacenet joined	14:09
	→ gbacon joined	14:10
	→ merlin83 joined	14:10
	← merlin83 left	14:10
	→ merlin83 joined	14:10
	→ smple joined	14:10
	→ altrux joined	14:10
	→ ChUrP1 joined	14:11
	← ChUrP left	14:14
	← hugoxrosa left	14:15
	→ roop joined	14:15
smple	Hey, I'm trying to set up something simple and i'm falling on my face. I'm wondering if you guys could point me in the right direction of my screwup	14:17
	← project2501b left	14:17
	→ kpreid joined	14:17
	← Ryback_ left	14:17
workmad3	smple: possibly... describe your problem? :)	14:18
smple	pretty much I'm trying to set up a remote repository with two branches, stable and testing (i'm using gitosis), which i can then pull out as two separate repos on my end	14:18
	← hohoho left	14:18
smple	however i can't seem to pull in a branch on the servers side, nor can i seem to create one on my end	14:18
	→ Ryback_ joined	14:18
cbreak_work	just push branches to the other repo	14:18
smple	creating one on my end (via git branch -a testing) gives me a "fatal: Not a valid object name: 'master'." error	14:18
workmad3	smple: you don't get separate repos from branches on the server	14:18
cbreak_work	and creating on your side is with git checkout -b or git branch	14:18
	you can make separate repos if you want	14:19
smple	workmad3, i understand that	14:19
	so git checkout -b testing, then git push?	14:19
workmad3	smple: good :) and then as cbreak_work said... 'git checkout -b testing; git push origin testing:testing'	14:19
cbreak_work	you'd have to git push destrepo branchname	14:19
workmad3	ah yeah, you can shorten it to git push origin testing	14:20
smple	does "error: src refspec testing does not match any.: mean anything to you guys?	14:20
	→ adrien joined	14:20
	→ ChUrP joined	14:20
cbreak_work	how did you create your repo?	14:20
smple	via gitosis	14:20
cbreak_work	does git log show anything?	14:20
	without commits, you can't make branches.	14:21
smple	it shows my initial commit on testing	14:21
cbreak_work	ok	14:21
smple	waaaait	14:21
	what	14:21
	one second	14:21
	why is it that i can git checkout testing and still be on branch master?	14:21
	→ jrmuizel joined	14:21
	← ChUrP1 left	14:22
cbreak_work	getting an error?	14:22
smple	no	14:22
	git checkout testing	14:22
	→ batok joined	14:23
smple	(no error) git status #On branch master	14:23
	← unreal left	14:23
	→ unreal joined	14:23
charon	is 'testing' a file but not a branch?	14:23
smple	hmmmm	14:24
	checkout -b didn't put me onto another branch?	14:24
adrien	_Vi: external storage for git is pretty nice but there a few problems: mostly, I'm trying to get and older commit and git status keeps on complaining, disabling the clean filter for a minute solves it but there are definitely some things to iron out ;-)	14:24
smple	git branch -a still fails	14:24
	and git branch doesn't show me as being on testing	14:24
workmad3	smple: can you create a gist with your sequence of commands so far, please?	14:25
_Vi	adrien, You can use some stronger command (like "git stash" or "git update-index").	14:25
smple	git init	14:25
	→ brntbeer joined	14:25
smple	git branch -a testing	14:25
	[failure]	14:25
workmad3	smple: right, so you haven't got any commits...	14:25
	← Adriaan left	14:26
smple	i can pull from the remote repo	14:26
workmad3	smple: so what exactly do you expect git to branch?	14:26
smple	I want a branch	14:26
	called test	14:26
	*testing	14:26
_Vi	adrien, Is it still using SSH or you have switched it to some other protocol (rsync?)	14:26
smple	that i can push onto the other side	14:26
workmad3	smple: yes, but a branch needs a commit to point to...	14:26
smple	or well a branch called anything that i can push it to the remote side as testing	14:26
adrien	_Vi: git stash and git reset --hard didn't work, but I'm not using the "anti-clean" filter (I've forgotten the name)	14:26
smple	workmad3, can I have differently named branch names?	14:26
_Vi	adrien, "smudge".	14:26
workmad3	smple: a branch is, after all, a branch-point in your commit history	14:26
smple	yeah	14:27
workmad3	smple: but you have no commits, so there is nothing to branch	14:27
smple	i'm a little unsure as to the syntax of doing that	14:27
adrien	_Vi: right	14:27
	_Vi: it's local so using cp :P	14:27
	← kpreid left	14:27
workmad3	smple: or, in other words... add something to your repository before trying to branch it!	14:27
smple	ok	14:27
	i've added something locally	14:27
workmad3	and committed it?	14:27
smple	yep	14:27
	added the remote origin	14:27
	← bauruine left	14:28
workmad3	ok, so now do 'git checkout -b testing'	14:28
smple	tried to "git push remote origin testing"	14:28
	and it failed	14:28
_Vi	adrien, Why not using "smudge"? What you do after checking out old commit?	14:28
smple	hmmmm	14:28
	→ tedoc2000 joined	14:28
smple	workmad3, won't this break the other branches though?	14:28
workmad3	smple: no	14:28
	smple: why would it affect any other branches?	14:29
	← unreal left	14:29
workmad3	you're creating a new branch	14:29
smple	on the remote side	14:29
	it wouldn't be a fast forward merge	14:29
	on master	14:29
	→ unreal joined	14:29
workmad3	smple: does the repo exist on the server?	14:29
	smple: and if so, why on earth are you doing a git init locally?	14:29
	← flo` left	14:29
charon	[16:25] <smple> git branch -a testing	14:30
	you know, you could read the message that gives:	14:30
	"fatal: -a and -r options to 'git branch' do not make sense with a branch name"	14:30
workmad3	smple: if you have a remote repo, you do 'git clone <url>' to get your repository, not a new 'git init'	14:30
adrien	_Vi: because only having external storage is more practical for me: I have several projects with different needs wrt what to download and when, so managing it by myself and not in git works better for me	14:31
	my problems might well be the absence of the smudge filter however	14:31
smple	hmmmmm	14:31
	ah	14:31
	workmad3, that makes more sense now	14:31
	← tlir left	14:31
_Vi	adrien, "clean" is to turn the file into it's hash. "smudge" is to turn the hash into a file back. You will get a oneliner with digits instead of a file unless you use "smudge" filter.	14:32
	→ bradallen137_ joined	14:32
workmad3	smple: yeah... if you create a new repository, you automatically aren't in a fast-forward situation ;)	14:33
adrien	_Vi: yes, of course: what I meant is that it was better for me to do something like: if [ ! -e $file$sha1 ]; then wget --quiet -O $file$sha1; fi	14:33
smple	hmmm	14:34
	→ fleet1ng joined	14:35
	← ShooterMG left	14:35
	← fleet1ng left	14:36
_Vi	adrien, Would it be much better if Git supported such tricks [semi-]officially?	14:36
	→ ChUrP1 joined	14:37
	← herdingcat left	14:37
	→ thwart joined	14:38
	← flazz left	14:39
	← tcurdt left	14:39
	← galderz left	14:39
	→ flo` joined	14:39
flo`	hi	14:40
	← ChUrP left	14:40
	→ t0nic[w] joined	14:40
flo`	i'm git-bisecting atm. how can i find out at which commit i am atm?	14:40
	→ flazz joined	14:40
	→ galderz joined	14:40
flo`	i.e., how can i find out some string that identifies the program version i just compiled?	14:40
	← unreal left	14:40
	→ bauruine joined	14:40
	→ unreal joined	14:41
shruggar	flo`, git rev-parse HEAD	14:41
flo`	shruggar, so, i can later tell git bisect that the string git rev-parse told me is good or bad, right?	14:42
adrien	_Vi: yeah, most probably: it works pretty well but doing it in git directly, without relying on the filter functionnality would probably be better, and it would probably make it possible to decide when to get the actual file: always when cloning, with a command ('git extstorage get path/to/file' for instance), never...	14:42
	← t0nic[w] left	14:42
	→ earthome joined	14:42
shruggar	flo`, yes. You can also just say "git bisect good" or "git bisect bad", and it defaults to the one you are currently on	14:43
adrien	btw, it got a 4.2GB repo down to a 45MB one, which compresses down to 3MB for transfer, so it's definitely worth it in my case	14:43
	:-)	14:43
flo`	shruggar, good. the thing is, i actually am tracking three errors.	14:43
	and as long as they are all good, it's okay. but as soon as they diverge, i need to remember the current point and track them seperately	14:44
shruggar	flo`, well track one at a time, or "git bisect" won't work ;) you can use "git bisect skip" to mark the commit you're currently at as "I can't tell from this one"	14:44
flo`	thank you ;)	14:44
	→ hipe joined	14:44
	← galderz left	14:44
flo`	i'm off again. thx	14:45
	← flo` left	14:45
_Vi	adrien, May be the most simple way would be marking of blobs as unpackable, e.g. they will always be separate files without delta packing attempts.	14:45
shruggar	_Vi, "unpackable" would mean "unfetchable", would it not?	14:45
	← eighty4 left	14:45
_Vi	E.g. usual source code is usual blobs, "media" blobs are always objects.	14:45
	← aziz_ left	14:46
_Vi	shruggar, Or pack each of them to the separate trivial pack (without any compression algorithms).	14:46
	← ustunozgur left	14:46
_Vi	(Or with some specialized compression algorithm that works with big aligned blocks)	14:47
	→ ustunozgur joined	14:47
adrien	for _my_ case, making them unfetchable and actually copying them to another place (http, ftp, rsync, nfs...) would be good, I don't know if it'd still give a kind of pointer/reference to the original object (the sha1 for instance)	14:48
	← xuxc000 left	14:48
	← opalka left	14:48
_Vi	shruggar, "Unfetchabe" is current idea with filters. Git only stores hash of a file, external tools manage how to get them (and where to put).	14:48
	← beni- left	14:48
	← ustunozgur left	14:48
	← Remixman left	14:49
_Vi	E.g. to add special new object type "pseudo-blob" that will tell git that this blob is not in object database and should be fetched from external source.	14:50
	← unreal left	14:50
	→ unreal joined	14:50
	← ChUrP1 left	14:51
	← JohnTheSavage left	14:52
	offby1` → offby1	14:52
	← offby1 left	14:52
	→ offby1 joined	14:52
	→ MetaCosm joined	14:53
shruggar	_Vi, probably makes more sense to add a filter to contrib	14:53
	← unreal left	14:54
	← tstclair left	14:54
	→ unreal joined	14:54
	← malumalu left	14:56
	← mFacenet left	14:57
	→ kpreid joined	14:57
	→ warlock_mza joined	14:57
	← outoftime left	14:57
	→ outoftime joined	14:58
	→ JohnTheSavage joined	14:58
	← outoftime left	14:58
	← JohnTheSavage left	14:58
	→ ChUrP joined	14:59
	← sylr left	14:59
	→ outoftime joined	14:59
smple	Hey, if i have a remote server that has multiple branches	14:59
	← outoftime left	14:59
	→ project2501b joined	14:59
smple	is there any way to pull a specific branch down?	14:59
_Vi	shruggar, OK. Should I send a patch that adds it to "/contrib"?	14:59
smple	git clone hasn't pulled it	14:59
Pieter	smple: git pull origin branchname	14:59
	→ marcopolococo joined	14:59
	← mountains left	15:00
	→ outoftime_ joined	15:00
	→ hagabaka joined	15:00
	← hagabaka left	15:00
	→ hagabaka joined	15:00
shruggar	_Vi: sure. If nothing else, it will generate discussion, which is good.	15:00
smple	Pieter, ok, i'm still on branch master on my repo	15:00
	if i git pull will it work nicely?	15:00
	→ SlickT10 joined	15:00
	← thwart left	15:01
Pieter	smple: sure, it'll merge the changes of that branch into your master branch	15:01
FauxFaux	Heh, I just tried the gsoc svn-dump-fast-export. It's anihilatingly fast. Now to just go add support for generating branches so I can actually check-out my repo.	15:02
	→ galderz joined	15:02
	← galderz left	15:02
	→ galderz joined	15:02
	← apsysoev1 left	15:02
	→ Fangorn_ joined	15:02
smple	sweet :d	15:03
	→ thm joined	15:03
adrien	something I would have liked was to clone a repo which has blobs in it and get blobs stored outside of the clone, for that, filters did not work (not triggered)	15:03
thm	how can I reject (via a hook, maybe) a push that is a descendant of some well-known revision (hash)?	15:04
Ilari	smple: Does 'git branch -r' list it?	15:04
	→ tizbac joined	15:04
	→ Imran-UK joined	15:04
Ilari	thm: Does the repo contain that commit already (reachable from any of its refs)?	15:04
	thm: Ah, aren't there ready-made "blacklist" hooks?	15:05
thm	REPO1 contains it, REPO2 should not, and devs are pusing from REPO1 to REPO2	15:05
	+h	15:05
	→ alsuren_ joined	15:06
	← galderz left	15:06
	→ galderz_ joined	15:06
smple	yep	15:07
_Vi	shruggar, Where should I send it?	15:07
	← khelll left	15:07
FauxFaux	Also, interestingly, the repack -a -d'd git repo is bigger than the source svn one.	15:07
alsuren_	I want to have an alias grep = grep -C2 -n but the closest I can get is gr = grep -C2 -n	15:08
	← bradallen137_ left	15:09
shruggar	_Vi, see Documentation/SubmittingPatches in git.git ( if my gitbot-fu is up to it: b7c600f132949b7f995ac2643ab9728272d3aaab:Documentation/SubmittingPatches )	15:09
Gitbot	[git b7c600f13]: http://tinyurl.com/27ddta8 -- Merge branch 'master' into next	15:09
_Vi	shruggar, OK, sending to general list.	15:09
shruggar	hm apparently not: http://repo.or.cz/w/git.git/blob/b7c600f132949b7f995ac2643ab9728272d3aaab:/Documentation/SubmittingPatches	15:09
_Vi	(when it be ready)	15:09
FauxFaux	GIRL	15:09
	(Wrong window)	15:10
alsuren_	is there any way to change the default options for git grep, or do I just have to train myself to use an alias?	15:10
charon	shruggar: you can say <HEAD:Documentation/SubmittingPatches>	15:10
Gitbot	[git SubmittingPatches]: http://tinyurl.com/ybt3xrd [blob]	15:10
	→ adamv joined	15:10
Ilari	thm: Anyway, there are "blacklist" hooks that can do it.	15:10
shruggar	charon, ah, <> is the key, then. I coulda sworn I had tripped it by accident without those, before. Maybe I was giving a tree's sha, though.	15:11
	→ h32Lg joined	15:11
	← apsysoev left	15:11
	→ jfkw joined	15:11
	← ChUrP left	15:12
charon	alsuren_: no. aliases cannot override builtins because that would make git very hard to script.	15:12
	shruggar: i think the <> are required unless it's an sha1 prefix of an object... those are parsed anyways (which is what happened for you)	15:13
shruggar	testing: b7c600f132949b7f995ac2643ab9728272d3aaab:/Documentation/SubmittingPatches	15:14
Gitbot	[git b7c600f13]: http://tinyurl.com/27ddta8 -- Merge branch 'master' into next	15:14
	→ raichoo joined	15:14
shruggar	just wondered if missing the / had anything to do with it	15:14
	← level09 left	15:14
	← cbreak_work left	15:14
	→ andrewl joined	15:15
	→ ChUrP joined	15:19
	← kpreid left	15:19
	← radioman-lt left	15:19
	→ radioman-lt joined	15:20
	→ kipras joined	15:20
	→ poseid joined	15:21
	→ Nemurenai joined	15:22
	→ zhou__ joined	15:22
	→ Yuffster joined	15:24
	← zhou__ left	15:24
	→ Xiaobo joined	15:24
	← jspiros left	15:24
	→ alex__c2022 joined	15:25
	→ jspiros joined	15:25
	← poseid left	15:25
	→ ChUrP1 joined	15:25
	← tahorg left	15:25
	← ChUrP1 left	15:26
thm	and, slightly OT, does someone know whether git hooks on sourceforge can use perl?	15:28
	→ hugoxrosa joined	15:28
	← ChUrP left	15:28
	← skoop left	15:31
	→ cirwin joined	15:32
	→ aspotashev joined	15:32
	→ aziz_ joined	15:34
	← aspotashev left	15:34
	→ kenichi joined	15:35
	→ aspotashev joined	15:35
deavid	sitaram: it is ok to create repositories in subdirs? i already have them in dirs and i wish to preserve them	15:35
	(in gitolite)	15:35
	← aspotashev left	15:35
	← amerine left	15:35
	→ hohoho joined	15:35
	← warlock_mza left	15:35
	→ loinclot_ joined	15:35
sitaram	deavid: take a wild guess...	15:36
	→ flo` joined	15:37
flo`	hi	15:37
deavid	i could try	15:37
flo`	can git bisect be restricted to certain files?	15:37
	← TeckniX left	15:37
	IslandUsurper → IslandUsurperAFK	15:37
flo`	if i know that there's a bug in the wlan driver, i assume i have to do less bisections as if i didn't know, right?	15:37
deavid	sorry for asking but i dind't find a short version of the docs with the specs... i'm reading the entire documentation and that's very slow	15:37
charon	flo`: hint: the git-bisect manpage synopsis says: git bisect start [<bad> [<good>...]] [--] [<paths>...]	15:38
sitaram	deavid: precisely... so much doc ==> so many features, right? And you think I'd miss something so fundamental? ;-)	15:38
flo`	oh... oops ;) thank you charon ;)	15:38
	→ Determinist joined	15:38
	→ aspotashev joined	15:38
	← smuf left	15:39
	← loincloth left	15:39
deavid	sitaram: i don't know. a short document with specs, limitations, and complex examples would be faster for me to get an idea of what gitolite can do. But i guess i should keep reading the docs	15:40
flo`	charon, when specifying a path, will then only that path be updated, or the whole tree?	15:40
deavid	anyway, big documentation is far better than only short documentation or no docs at all :-)	15:41
	→ tahorg joined	15:41
sitaram	deavid: specs, limitations, complex examples? and short? you don't want much do you?	15:41
charon	flo`: updated during checkout? the whole tree of course, git needs to check out the full snapshot of that commit	15:41
	→ joelmccracken joined	15:41
flo`	okay. thanks ;)	15:41
	→ rajeshsr_ joined	15:41
	← joelmccracken left	15:42
charon	flo`: it's merely a filter that says: commits that do not affect <paths> cannot have introduced the bug	15:42
sitaram	deavid: the "specs" are complex enough that most of them I could only put into the example config file -- rather like squid's config file, it's commented code that explains what it does as it goes.	15:42
flo`	okay ;)	15:42
sitaram	deavid: so technically that's not even in the docs, so you could just read example/* and be done	15:42
deavid	sitaram: so that's the problem, i didn't found that example earlier ... hmmm	15:42
adrien	sitaram: btw, gitolite will fail if you have a ~/.ssh/authorized_keys2 file and not a ~/.ssh/authorized_keys (no '2')	15:42
deavid	i guess is somewhere in the debian package	15:43
	← steffkes- left	15:43
deavid	sitaram: i found them :-) they were in /etc/gitolite and, yes, that was what i wanted to read :-)	15:44
	→ opalka joined	15:45
	→ epicmath joined	15:45
	→ cytrinox_ joined	15:45
	→ xiong joined	15:45
	← loinclot_ left	15:45
epicmath	How can I make git-instaweb recognize httpd as apache2? My distro includes httpd (apache2) but it isn't recognized.	15:45
	← brntbeer left	15:45
	→ loincloth joined	15:46
	→ kpreid joined	15:46
sitaram	adrien: isn't that obsolete now?	15:46
	→ tstclair joined	15:47
	← Joeconyers left	15:47
adrien	sitaram: it is but some might still have that file, so it's compat	15:47
	← Wraithan left	15:47
adrien	basically: authorized_keys doesn't exist, try authorized_keys2	15:48
	← cytrinox left	15:48
	cytrinox_ → cytrinox	15:48
sitaram	adrien: man ssh, sshd, ssh_config, sshd_config, none of them mention authorized_keys2	15:48
	→ tjholowaychuk joined	15:48
adrien	http://sial.org/howto/openssh/#s6 -> "The authorized_keys2 file has been deprecated since the OpenSSH 3.0 release (2001). " but I had it on one computer	15:49
	← flo` left	15:49
adrien	I was only suggesting that if it took one or two lines of code, as far as I'm concerned, I wouldn't make big changes to support that	15:50
sitaram	adrien: I don't even support git older than 1.6.2, and thats a darn sight newer than 2001 ;-) sorry...!	15:51
	adrien: it's not the # lines of code. It's the testing and double checking and all that, and understanding the precedence, and so on...	15:52
	→ bradallen137_ joined	15:55
	→ carutsu joined	15:56
adrien	sitaram: hahaha, absolutely no problem ;-)	15:56
	← epicmath left	15:56
sitaram	:)	15:56
	→ amerine joined	15:57
	→ GDorn joined	15:59
	→ babusri joined	15:59
	← roop left	16:00
	← elmob left	16:00
	← rox` left	16:00
tjholowaychuk	is it possible to have a submodule for only a portion of the tree? for example if you have ./lib/mylibrary, can I have a submodule that is only "mylibrary" and does not contain all the cruft above that	16:01
	← MattDiPasquale left	16:01
tjholowaychuk	hard to explain	16:01
	→ MattDiPasquale joined	16:02
charon	tjholowaychuk: not as such, though you can do a "reverse subtree merge" style dance to make a modified copy of the sub-repo which only "sees" 'mylibrary'	16:02
	→ poseid joined	16:03
	→ brntbeer joined	16:03
charon	or maybe i'm misunderstanding. is lib/mylibrary an existing submodule content, or the path in the super-repo where you want your submodule to reside?	16:03
	→ TeckniX joined	16:04
	← steveire left	16:04
	→ g0bl1n joined	16:05
	← Xiaobo left	16:05
	→ rox` joined	16:06
	→ justin-george joined	16:08
	→ hobodave joined	16:10
	← sunn left	16:11
	→ Modius joined	16:11
	← martengo left	16:11
	← Modius left	16:12
	→ sunn joined	16:12
	← marcopolococo left	16:13
	→ Modius joined	16:13
	→ simplechat joined	16:14
simplechat	Hey guys	16:14
	← girishr left	16:14
shruggar	.	16:14
offby1	!	16:14
	← smple left	16:14
shruggar	?!	16:14
	→ MattDiPa_ joined	16:15
simplechat	My git repository is now completely and utterly broken :(	16:15
shruggar	simplechat, nuh-uh!	16:15
	← Yuffster left	16:15
shruggar	simplechat, be more descriptive, we'll prove you wrong ;D	16:15
simplechat	seriously	16:15
	i push on one branch	16:16
	i can pull it on two and both work	16:16
	brb	16:16
	→ brosner joined	16:16
charon	simplechat: pushing "on" a branch already indicates you misunderstand push.default in man git-config	16:16
Gitbot	simplechat: the git-config manpage can be found at http://git.or.cz/man/git-config	16:16
	← priidu left	16:16
	← MattDiPasquale left	16:17
	← thiago_home left	16:17
	→ Yuffster joined	16:18
	← EricInBNE left	16:19
jason237	shruggar: I completely broke a git repository once by setting it to be its own alternate and running git gc	16:19
	oops	16:20
	← raichoo left	16:20
shruggar	jason237, sounds like a bug to me	16:21
	→ flo` joined	16:21
flo`	hi	16:21
	→ steveire_ joined	16:21
	← steveire_ left	16:21
	→ steveire_ joined	16:21
charon	shruggar: not really, when setting alternates you really have to know the implications	16:23
	→ resmo joined	16:23
charon	see e.g. the huge warning against clone --shared in the manpage	16:23
flo`	how can i display all commits with their short description (like "added foo") that affect a certain file	16:23
	or a certain dir?	16:23
	← hohoho left	16:23
	← plediii left	16:23
shruggar	charon, sure, you need to know the implications, but "I don't need my own objects, I can use myself for reference!" sounds detectably stupid	16:23
	→ sukima joined	16:24
	← Dawgmatix left	16:24
	→ hohoho joined	16:24
jason237	flo`: git log FILE?	16:24
shruggar	flo`, git log --oneline -- <file>	16:25
flo`	oh, thanks	16:25
sukima	Odd question but can you move a commit history from one branch to another? Say I have a bunch of commits which I want to keep but realized they are on the wrong `branch (ie master vs develop) can I move them?	16:25
flo`	i couldn't find a FILE option in the manpage :(	16:25
jason237	sukima: sounds like a job for rebase	16:26
	→ ProperN[out] joined	16:27
	← ProperN[out] left	16:27
	→ ProperN[out] joined	16:27
sukima	jason237 perhaps my knowledge of rebase is somewhat limited as I don't see how it can do a "move"	16:27
shruggar	sukima, you want git rebase --onto	16:27
	sukima furiously starts reading the man page again. Thanks!	16:28
	→ krawek joined	16:30
	← ProperNoun left	16:30
	ProperN[out] → ProperNoun	16:30
	→ corni joined	16:30
jason237	question: is there any way to query what remote branch my current branch merges with? that is, the ref in my repository, rather than branch.foo.merge which is the ref on the remote	16:30
	← hipe left	16:30
shruggar	jason237, I thought it was "git branch -v", but that doesn't appear to be showing the ref, only the upstream commit	16:33
	jason237, git config branch.<branchname>.merge will show it, but sounds a bit low-level for my taste	16:33
	→ nevyn^__ joined	16:34
jason237	shruggar: and that's the ref on the remote, which isn't the same	16:34
	← nevyn^__ left	16:34
	← rajeshsr_ left	16:34
jason237	i.e. it's refs/heads/foo rather than refs/remotes/origin/foo	16:34
	← bball__ left	16:34
shruggar	ah, of course this will work: git rev-parse --symbolic-full-name @{u}	16:35
	→ bball__ joined	16:35
flo`	may i use git-revert when i only did the following: git-bisect, create a .config, make	16:36
	or do i have to clean first?	16:36
	← alester left	16:36
	← barismetin left	16:37
shruggar	a bit brute-forcily, though :)	16:37
	flo`, what?	16:37
	→ tmz joined	16:37
jason237	shruggar: ah, I didn't know about @{u}, thanks!	16:37
flo`	well, i simply clean up before :D	16:38
	← nevyn^_ left	16:38
	→ asarih joined	16:38
	← trivol left	16:38
asarih	hi. how do you delete a remote branch on the remote repository? (not the reference to a remote branch on the local repo.)	16:39
	→ ShooterMG joined	16:40
	← whaley left	16:41
	← kpreid left	16:41
curtana	asarih, git push --delete remote branchname	16:43
asarih	curtana: ah. thanks!	16:43
curtana	if you have a slightly old version of git then you'll have to do: git push -f remote :branchname	16:43
asarih	let me try the 'new' way.	16:43
curtana	if you have a slightly older version again: git push remote +:branchname	16:43
	they all do the same thing, but you can see the evolution of the syntax of the command :)	16:43
adrien	is there a way to make it impossible to write/commit on a branch? I'd like to be able to get a remote but be forbidden to push to it (don't have gitolite where I'd push)	16:44
	← shruggar left	16:44
	→ sako joined	16:45
	→ tomasm- joined	16:45
Ilari	adrien: You want to set so you can't accidentially push to some remote?	16:45
	adrien: Or just for some branches?	16:46
tomasm-	is there a git browser module for cakephp (or php in general)?	16:46
	→ ProperN[out] joined	16:46
	→ rajeshsr_ joined	16:46
	← ProperN[out] left	16:46
	→ ProperN[out] joined	16:46
adrien	Ilari: to some branch, but to some remote would work too (either would work for me)	16:46
	← cirwin left	16:47
	← thoraxe left	16:47
	← warthog9 left	16:47
	→ thoraxe joined	16:48
	← ereslibre left	16:48
Ilari	adrien: Well, if you have shell access to remote side, you could use update hook on remote side. For preventing push to local side, set the pushurl for it to something invalid.	16:48
adrien	right, good idea, thanks :-)	16:48
asarih	curtana: great. It worked. thanks again.	16:49
	← adamv left	16:49
	← ProperNoun left	16:49
	ProperN[out] → ProperNoun	16:49
	→ adamv joined	16:49
Ilari	adrien: 'invalid:invalid' should be one URL usable for invalid URL.	16:49
	→ apsysoev joined	16:49
adrien	thanks, I'll do that in a few minutes	16:49
	adrien is moving repos and branches around	16:49
	← grumpytoad left	16:50
	IslandUsurperAFK → IslandUsurper	16:50
	→ m104 joined	16:50
	← steveire_ left	16:51
	← malditogeek left	16:52
	→ rpg joined	16:53
	→ kpreid joined	16:53
mar77i	w00t	16:53
	mar77i seinen matetee schlürft	16:53
	← gebi left	16:53
	← brntbeer left	16:54
	→ Eaven joined	16:54
	→ grumpytoad joined	16:54
	→ brntbeer joined	16:55
	← brntbeer left	16:56
	← tomasm- left	16:57
	← rlotun left	16:57
	← henkboom left	16:58
	→ henkboom joined	17:00
	→ aresnick joined	17:00
	← Schmallon left	17:01
	← Imran-UK left	17:01
	→ p3rg2z joined	17:02
	→ nevyn^_ joined	17:02
rajeshsr_	hi all	17:02
	is there any lib to write git hooks?	17:03
	would prefer in python or perl..	17:03
	← mcella left	17:03
simplechat	Is there any way to force a reset?	17:06
	error: Entry 'stable' not uptodate. Cannot merge.	17:06
	when i'm trying to pull on stable	17:06
	how am i supposed to update something if i can't update it?	17:06
	→ ntoll joined	17:06
rajeshsr_	simplechat: git reset --hard	17:07
	?	17:07
	→ ProperN[out] joined	17:08
	← ProperN[out] left	17:08
	→ ProperN[out] joined	17:08
	← ProperNoun left	17:11
	ProperN[out] → ProperNoun	17:11
simplechat	rajeshsr, thanks :D	17:12
	→ rubydiam_ joined	17:14
	→ Dawgmatix joined	17:16
Ilari	rajeshsr_: You can write git hooks in any language that can be compiled to binary or supports shellbang execution.	17:16
	→ xperience joined	17:16
	← nevyn^_ left	17:16
	← asarih left	17:17
Ilari	rajeshsr_: The git just checks if hook is executable and if it is, it uses execve(2) to run it. Whatever kernel can run can be used as hook.	17:17
rajeshsr_	Ilari: yes, i understand that. But it can get easier to write good hooks with an lib, like vim-python..	17:17
	← cbreak left	17:18
	→ jtbandes joined	17:18
	← Ryback_ left	17:19
	← adamv left	17:20
	← m104 left	17:20
	→ cbreak joined	17:21
	adamholt → adamholt_away	17:21
	adamholt_away → adamholt	17:21
	← brosner left	17:22
	→ Ryback_ joined	17:23
	← ntoll left	17:23
	→ mrball joined	17:23
	→ brosner joined	17:24
	→ coachz joined	17:24
coachz	you guys have any software for tracking an overview of all the functions a site has and to track tasks to be completed/priorities etc that integrates with git ?	17:24
	← jspiros left	17:25
	→ jspiros joined	17:25
	→ hipe joined	17:25
	← xperience left	17:28
jason237	rajeshsr: there is a git perl module included with git	17:28
	→ hagebake joined	17:31
	← hagebake left	17:31
	→ hagebake joined	17:31
	→ zenazn joined	17:31
	→ m104 joined	17:32
	← tatsuya left	17:32
	← hagabaka left	17:33
	→ nick_ joined	17:33
	nick_ → Guest14320	17:34
	← coachz left	17:34
	← SlickT10 left	17:34
rajeshsr_	jason237: cool! any doc on how to use it?	17:34
	→ Dunemaster joined	17:35
jason237	rajeshsr: man Git	17:35
	(with the capital G)	17:35
rajeshsr_	jason237: tats too interesting! thanks	17:37
	← mquin left	17:38
	→ dreiss joined	17:38
flo`	how can i set the state to a certain commit?	17:39
rajeshsr_	git revert ?	17:39
	or even checkout	17:39
	if you dpn't want to get things affected..	17:39
flo`	rajeshsr, i also thought of that, but the manpage cunfuses me.	17:39
rajeshsr_	*don't	17:39
	→ Andy_ODT joined	17:39
flo`	"and record a new commit that records it"	17:40
Dunemaster	Hi, I am having this problem with msysgit. I have installed it and everytime I run Git from command line I keep getting message that chcp is not a name for command and the same mesage for reg. I was trying to google this problem, but it seems that google does not know anything about chcp and keeps giving me results for dhcp.	17:40
flo`	i don't want to record a new commit!	17:40
Andy_ODT	how do I revert to a revision by date like you can do in svn with svn update -r#### ?	17:40
rajeshsr_	flo`: checkout won't do anything.	17:40
	← Nemurenai left	17:40
rajeshsr_	git checkout -b newBranch hashCommit	17:41
	→ justicefries joined	17:41
justicefries	so if I have some unversioned files in a branch...how do I..version them?	17:41
charon	Andy_ODT: that's a rather ill-defined request in nonlinear history. the best you can do is find one commit made after <date> by using 'git log -1 --after "<date>" --all' (or similar) and then 'git checkout <sha1>'. note that the latter will detach your HEAD, make sure you check out a branch again before doing any work	17:41
rajeshsr_	Andy_ODT: I think you can search the commits by date using log	17:42
	← GDorn left	17:42
flo`	ah, thanks ;)	17:42
Andy_ODT	thank you	17:42
sukima	justicefries: git add <file>	17:42
justicefries	thanks.	17:42
	→ Evanlec joined	17:42
Evanlec	hey guys, how can I see the entire history of a particular file? (i.e. show me all commits +diffs that affect this file, since day 1)	17:43
rajeshsr_	git log -- file	17:44
Evanlec	oh, thats it	17:44
charon	... and if you believe there are commits missing from this, look up --full-history	17:44
Evanlec	well that wasnt too hard ;p	17:44
	→ nevyn^_ joined	17:44
	→ rorist joined	17:44
	← marc_kdab_ left	17:45
Evanlec	thanks ;)	17:45
	→ eighty4 joined	17:45
	← Raydiation left	17:46
	← rayners left	17:46
Dunemaster	/join github	17:46
rajeshsr_	Welcome!	17:46
Dunemaster	Hi, I am having this problem with msysgit. I have installed it and everytime I run Git from command line I keep getting message that chcp is not a name for command and the same mesage for reg. I was trying to google this problem, but it seems that google does not know anything about chcp and keeps giving me results for dhcp.	17:47
	→ rayners joined	17:48
Dunemaster	if you know the answer, pls email me at honza.bok(at)gmail.com	17:48
	← Dunemaster left	17:49
	← rpg left	17:50
	← antgel left	17:51
	→ cirwin joined	17:51
	→ thiago_home joined	17:51
	→ marc_kdab_ joined	17:52
	→ richardbronosky joined	17:53
	← binjured left	17:53
	← marc_kdab_ left	17:53
	thiago_home → thiago_cgh	17:54
	← __ed left	17:54
	adamholt → adamholt_away	17:54
	adamholt_away → adamholt	17:54
	→ marc_kdab_ joined	17:55
	→ sylr joined	17:55
	← marc_kdab_ left	17:56
	← mastro left	17:56
	→ rseifert_ joined	17:56
	→ marc_kdab_ joined	17:57
	← marc_kdab_ left	17:58
	← unreal left	17:58
	← Goneri left	17:58
	→ marc_kdab_ joined	17:58
	→ unreal joined	17:58
	→ binjured joined	17:59
	← marc_kdab_ left	18:00
	← froschi1 left	18:00
	← curtana left	18:01
	→ marc_kdab_ joined	18:01
	← marc_kdab_ left	18:02
richardbronosky	I'm seeing examples of how to inspect the commit message with the commit-msg hook, but I can't figure out how to alter the commit message.	18:02
	→ marc_kdab_ joined	18:02
thiago_cgh	richardbronosky: answer: you simply don't	18:02
	← marc_kdab_ left	18:03
	→ Goneri joined	18:04
	← rubydiam_ left	18:04
	→ rubydiam_ joined	18:04
	rubydiam_ → rubydiamond	18:04
	→ marc_kdab_ joined	18:04
richardbronosky	"you simply don't" was the answer to many needs before git was created. That is why git was created.	18:05
	← Guest14320 left	18:05
	← marc_kdab_ left	18:05
	→ SlickT10 joined	18:05
	→ Sou\|cutter joined	18:05
doener	richardbronosky: well, the manpage tells you to simply edit the file... does that fail?	18:06
Sou\|cutter	Is there a way to remove the last 4 commits in a branch from the index completely?	18:06
richardbronosky	for some reason I did not have .git/hooks/prepare-commit-msg.sample	18:06
	→ marc_kdab_ joined	18:06
Sou\|cutter	I don't want that history in there at all, so a new commit reverting the changes won't work for me	18:06
patrikf	Sou\|cutter: if you haven't pushed them yet, git reset HEAD^4 (maybe even --hard)	18:07
doener	Sou\|cutter: did you already push those commits? If not, just reset your branch head back	18:07
	patrikf: HEAD~4, not ^4	18:07
patrikf	yes, doener is right	18:07
	← marc_kdab_ left	18:07
Sou\|cutter	nah, I haven't pushed yet	18:07
patrikf	(or HEAD^^^^)	18:07
Sou\|cutter	cool, I thought there would be a way, I just couldn't figure it out meself... trying now	18:07
	→ marc_kdab_ joined	18:08
richardbronosky	I just found it. looks like that hook receives the temp files as argv[1]. thiago_cgh I hope you get a chance to use that info someday.	18:08
	→ Nemurenai joined	18:08
Sou\|cutter	whew, you guys are lifesavers :) thanks!	18:09
	← marc_kdab_ left	18:09
Sou\|cutter	I almost blew my cover grin	18:09
	← thiago_cgh left	18:09
	← unreal left	18:10
	→ unreal joined	18:10
	→ Belna_ joined	18:10
patrikf	Sou\|cutter: actually, you should have used git push [email@hidden.address] --all	18:12
	→ marc_kdab_ joined	18:13
Sou\|cutter	patrikf: I'll do that now!	18:13
	← Belna left	18:14
	← marc_kdab_ left	18:14
	← cirwin left	18:14
	← sako left	18:14
Sou\|cutter	wait a second....	18:14
	Sou\|cutter grins	18:14
	→ marc_kdab_ joined	18:14
	← mrball left	18:14
	→ webus joined	18:15
	→ LeMoNiCe joined	18:15
	← LeMoNiCe left	18:15
	← Ryback_ left	18:15
	→ LeMoNiCe joined	18:15
	← LeMoNiCe left	18:15
	→ LeMoNiCe joined	18:15
	← marc_kdab_ left	18:15
	← TBH left	18:16
	→ marc_kdab_ joined	18:16
	← h32Lg left	18:16
	→ TBH joined	18:17
	← alsuren_ left	18:17
	← marc_kdab_ left	18:17
	← loincloth left	18:17
	→ marc_kdab_ joined	18:18
	← marc_kdab_ left	18:20
	→ marc_kdab_ joined	18:21
	→ Liquid-Silence joined	18:21
Liquid-Silence	hi guys	18:21
	I call git add .	18:21
	and then git commit -a -m "test"	18:21
	but my github account show no changes	18:21
	wtf	18:21
	← flo` left	18:22
Liquid-Silence	anything I am missing?	18:22
	or does it just take a while to update?	18:23
redfox	did you pushed your repo?	18:24
	→ chrislerum joined	18:25
Liquid-Silence	ah so its git commit -m "message" && git push?	18:25
	nevyn^_ → nevyn^	18:25
redfox	yep	18:25
	→ _2of1_ joined	18:25
_2of1_	hi	18:25
	← didlix left	18:26
_2of1_	quick question:	18:26
	how do I apply an old git patch<?	18:26
	meaning i have a patch from quite a few commits/reveisions ago	18:26
	is there any way i can find out what revision the patch applies to, checkout that revision, apply the patch and then do a git up or something?	18:26
	→ becomingGuru joined	18:27
	← rajeshsr_ left	18:28
	← durarara left	18:28
	← rayners left	18:30
_2of1_	?	18:30
	→ linux454 joined	18:30
	← simplechat left	18:30
	→ cirwin joined	18:31
	→ rolfb joined	18:32
	← bluenovember left	18:32
	← gbacon left	18:33
_2of1_	anyone?	18:34
linux454	Greetings all. I'm writing a release script to handle creating releases of my projects. I'd like to store some project specific metadata (project name, archive base name, etc) in the .git/config file and have those properties propagate to my public repository when I do a push. Is there any way to get git to push these parameters to another repository when it is cloned/pushed/pulled?	18:34
	→ Ryback_ joined	18:34
	→ brntbeer joined	18:35
	→ Titosemi_ joined	18:35
patrikf	linux454: nah... maybe store them in a dotfile?	18:36
	← Titosemi left	18:36
	Titosemi_ → Titosemi	18:36
	← tvw left	18:38
linux454	Yeah, I didn't think about using the --file option to git config, that should do the trick and buys me the ability to still use git config to manipulate the values.	18:38
	→ khmarbaise joined	18:41
	→ khmarbaise_ joined	18:42
	cpg\|away → cpg	18:42
	→ priidu joined	18:43
	→ durarara joined	18:44
	← khmarbaise left	18:46
	khmarbaise_ → khmarbaise	18:46
Sou\|cutter	_2of1_: just apply it with --3way and take care of any conflicts	18:46
	← cxreg left	18:46
Sou\|cutter	or am it maybe, you get the idea	18:47
	→ roxoft joined	18:47
Sou\|cutter	_2of1_: I did this like 20 minutes ago, easy	18:47
	→ mastro joined	18:47
	← linux454 left	18:48
	← cirwin left	18:48
	← LeMoNiCe left	18:49
	← unreal left	18:49
	→ unreal joined	18:49
	← hebz0rl left	18:50
	← rox` left	18:50
	← Liquid-Silence left	18:51
	→ didlix joined	18:51
	← unreal left	18:51
	→ unreal joined	18:51
	← _Vi left	18:51
	→ marienz joined	18:52
	→ MrPunkin joined	18:53
MrPunkin	Is there a way to see a log of when a file was last updated locally by a git pull (so not the git commit log)	18:54
	or a list of those dates, I guess the modified date would show the latest pull	18:54
	← kelvie left	18:55
	→ JohnTheSavage joined	18:55
	← durarara left	18:56
	→ rayners joined	18:57
	← JohnTheSavage left	18:57
_2of1_	Sou: will try that now.. - thanks! :)	18:58
	→ JohnTheSavage joined	18:58
	← JohnTheSavage left	18:59
	→ ive joined	18:59
	→ JohnTheSavage joined	18:59
	← JohnTheSavage left	18:59
	← mtkd left	19:00
	→ aspotashev\|eeepc joined	19:00
	→ JohnTheSavage joined	19:00
	→ montylounge joined	19:00
	→ loincloth joined	19:01
	← JohnTheSavage left	19:01
	→ JohnTheSavage joined	19:01
	→ treak007_ joined	19:02
	→ ajsie joined	19:02
	← Sou\|cutter left	19:02
ajsie	what does this mean: git diff shows the diff between HEAD and the current project state	19:02
	→ ProLoser joined	19:02
ProLoser	hallo	19:02
	i'm having trouble setting up winmerge as my diff tool on msysgit	19:03
ajsie	the HEAD is the current active commit object? what does "current project state" means?	19:03
ProLoser	i opened up the settings on git gui and dropped the path to the app in the box, but it says the tool is unsupported	19:03
	what is a supported diff tool i can use, i don't care aout complexity or features too much, i just want to continue working with little to no configuration	19:03
	← treak007 left	19:04
	← aspotashev\|eeepc left	19:04
	← aspotashev left	19:04
	→ aspotashev joined	19:04
	← leo2007 left	19:05
	← marienz left	19:05
	← JohnTheSavage left	19:06
	→ khelll joined	19:07
	→ marienz joined	19:08
	→ cirwin joined	19:09
	← khelll left	19:10
drizzd	ProLoser: do you mean Merge Tool?	19:11
ProLoser	yes sorry	19:11
	→ voxpelli joined	19:11
ProLoser	mergetool	19:11
drizzd	vimdiff	19:11
	← Yuuhi left	19:11
ProLoser	i need a graphical tool	19:11
drizzd	ProLoser: see the paragraph on merge.tool in man git-config	19:12
Gitbot	ProLoser: the git-config manpage can be found at http://git.or.cz/man/git-config	19:12
	→ cxreg joined	19:12
ProLoser	so when it says valid built in it means it will know the path and the arguments to send to the tool?	19:13
drizzd	ProLoser: yes	19:13
ProLoser	hmm... even the path? like if i'm on a 64 bit os	19:13
	→ dhaza joined	19:13
drizzd	well, it should. if it doesn't you can still configure a custom mergetool	19:14
dhaza	is it possible to have two files in git mirror each other?	19:14
	like if there is a foo/a/config.ini, and a foo/b/config.ini that i want to be the same	19:14
	something like a symlink in git?	19:15
drizzd	dhaza: you can have symlinks	19:15
	but not hardlinks	19:15
ProLoser	error: More than one value for the key merge.tool: =	19:15
	how can i fix this?	19:15
	i thought i was properly setting the merge.tool config	19:15
drizzd	ProLoser: looks like you added it twice. How did you do that?	19:16
ProLoser	no idea	19:17
drizzd	no, I'm asking which command/gui did you use?	19:17
ProLoser	well i was doing git config merge.tool = something	19:17
	i went into the config file and deleted the two lines, it's fixed now	19:17
drizzd	ProLoser: you should not use an '=' sign here	19:18
ProLoser	i realized that	19:18
	→ malditogeek joined	19:18
ProLoser	i'm gonna install diffuse and use that	19:19
	i had winmerge installed	19:19
	← grumpytoad left	19:19
	→ jaql joined	19:20
	← henkboom left	19:20
	← malditogeek left	19:21
dhaza	ahh	19:21
ProLoser	anyone here use diffuse?	19:21
	→ tvw joined	19:21
drizzd	I've never used a merge tool, but if winmerge is a common one, we could add it to the builtins	19:21
ProLoser	winmerge is very popular among windows	19:21
	← mastro left	19:22
ProLoser	second to turtoisemerge	19:22
	i was just lucky in that diffuse has a win32 variant	19:22
	→ raichoo joined	19:22
ProLoser	but yes, winmerge support would be much appreciated :)	19:22
	← aspotashev left	19:22
ProLoser	now i just need to figure out why diffuse has 4 panes :/	19:23
	→ henkboom joined	19:23
	→ simosx joined	19:23
	← simosx left	19:24
	→ simosx joined	19:24
	→ dropandhop joined	19:24
drizzd	ProLoser: an extra one for the common ancestor maybe?	19:24
ProLoser	hmm	19:25
	this thing isn't that easy to use	19:25
	maybe i should just get turtoise, i had very little trouble using it in the past :/	19:25
dropandhop	hey all...I'd appreciate some help setting up http-backend with a bare repo that should be read/write via http and ssh	19:26
drizzd	ProLoser: defining a custom mergetool is not hard though	19:26
	← dhaza left	19:26
ProLoser	i understand that	19:26
dropandhop	i thought i had it setup correctly, but I'm getting the following error just when i try to clone via http: warning: remote HEAD refers to nonexistent ref, unable to checkout.	19:26
ProLoser	but the discussions online aout configuring winmerge are actually quite varied	19:27
dropandhop	I would really appreciate any help you can offer	19:27
	→ erythros joined	19:27
	→ mastro joined	19:27
drizzd	dropandhop: cd to the served repo and do "git rev-parse HEAD"	19:27
ProLoser	drizzd, if i have turtoise merge, how can i tell what path the git tool will look for it at? there's no associated msi installer	19:28
erythros	i hope i'm not out of line but does anyone know how to have git covert time to single local time?	19:28
ProLoser	out of line?	19:28
drizzd	dropandhop: doesn't sound like a problem specific to http btw.	19:29
dropandhop	drizzd- thanks, i'll do that right now...i just read the doc but don't fully understand what it does...could you possibly explain it to me?	19:29
cmyers	erythros: time is part of the commit metadata which is hashed - are you talking about when you type "git log" you want to see times in local time?	19:29
dropandhop	drizzd- i agree, i thought it was a git repo problem...but i can clone via ssh	19:29
erythros	a few of my users are in different time zones	19:29
drizzd	dropandhop: ok, but the clone via http doesn't abort either, does it?	19:30
erythros	so when i look at the logs some commits show up as being several hours old	19:30
	→ robertkeizer joined	19:30
erythros	even though they were just pushed up	19:30
	← rubydiamond left	19:30
drizzd	ProLoser: I don't know	19:30
	→ sagsousuke joined	19:30
cmyers	erythros: the times have the local timezone in them	19:30
	like: 'Fri Nov 13 18:05:23 2009 -0800'	19:31
	→ Heimidal_ joined	19:31
drizzd	dropandhop: maybe you have permission problems? is the head readable for the webserver?	19:31
cmyers	git is designed to not favor any one timezone over any other since it is a distributed VCS.	19:31
drizzd	*the HEAD file	19:31
dropandhop	drizzd- when i clone via http it creates the directory...but nothing in it...one sec on the permissions	19:32
	→ tongueroo joined	19:32
drizzd	dropandhop: except for the entire repo	19:32
	dropandhop: it just does not know which branch to checkout, because there is no HEAD	19:32
cmyers	erythros: looks like you can do `git log --date=local` and it will show them times in local timezone. for more, see `git help log`	19:33
dropandhop	drizzd- the HEAD file is owned by git:git and www-data is part of the git group	19:33
	drizzd- the permissions are: -rw-rw-r--	19:33
drizzd	dropandhop: ok, so that's not it. So I'm guessing git rev-parse HEAD does not output any errors either?	19:34
erythros	heres the out put on a few updates in the order they were posted	19:34
	37 min ago erythros	19:34
	27 hours ago Emrys	19:34
	3 hours ago erythros	19:34
drizzd	dropandhop: how about git show-ref HEAD	19:34
	does it refer to an existing branch?	19:34
dropandhop	if i run git show-ref HEAD on the server i don't get anything back	19:34
drizzd	you should also try "git show-ref origin/HEAD" in the "failed" clone	19:34
dropandhop	ok	19:34
richardbronosky	is there a command that would allow me to do a "rebase" but create a duplicat branch instead of moving it?	19:35
drizzd	dropandhop: show-ref should output something like this 9918285fb10d81af9021dae99c5f4de88ded497c refs/heads/master	19:35
Gitbot	[git 9918285fb]: http://tinyurl.com/33bsveo -- Git 1.7.2-rc2	19:35
charon	richardbronosky: just make a new branch first	19:35
drizzd	dropandhop: if it doesn't output anything, then git rev-parse HEAD should complain too	19:35
charon	richardbronosky: rebasing will only move one of the branches, and the other keeps the old commits	19:35
richardbronosky	in git help rebase, where they introduce branch "next"	19:35
	← robertkeizer left	19:35
	→ kelvie joined	19:36
	← galderz_ left	19:36
richardbronosky	charon: so I could do: git co topic; git co -b topic-merge; git co master; git rebase master topic-merge;	19:37
	→ gbacon joined	19:37
dropandhop	drizzd- i just tried to run show-ref HEAD on that failed http clone and it doesn't give me any ouput	19:37
	→ whaley joined	19:37
drizzd	dropandhop: oops my bad, git show-ref --head HEAD	19:37
dropandhop	ahhh!	19:38
	hehe- thanks	19:38
richardbronosky	and topic would still be on next and topic-merge would be on branch. and the changes introduced to next by topic would be applied to master via topic-merge?	19:38
drizzd	anyways, the HEAD of the failed clone is not interesting	19:38
	the HEAD of the remote is interesting	19:38
	← aziz_ left	19:38
charon	richardbronosky: yes except that this can equivalently be written as: git checkout -b topic-merge topic; git rebase master	19:38
dropandhop	drizzd- gotcha...nothing with that command on the failed clone...on the server i get the following: a5d318461fe051bdde23426de919eebbf207174e HEAD	19:39
charon	richardbronosky: note that a branch is not really "on" any other branch, they may just have the same commits. and also note that master will not magically pick up the commits, you have to (fast-forward) merge topic-merge to master first.	19:39
	→ sako joined	19:39
drizzd	dropandhop: ok, how about "cat HEAD"	19:39
	→ froschi1 joined	19:39
dropandhop	drizzd- thanks for your patience with this...i really appreciate it	19:39
	→ aziz_ joined	19:41
dropandhop	drizzd- the following command doesn't seem to work: git show-ref --cat HEAD	19:41
drizzd	dropandhop: yeah, just "cat HEAD", without git or anything	19:41
dropandhop	k	19:41
	→ tar- joined	19:42
	← erythros left	19:42
dropandhop	derrr!	19:42
	i wasn't paying attention...sorry	19:42
	it's: ref: refs/heads/master	19:42
drizzd	ok, and that branch exists?	19:42
dropandhop	ya, it's the only one we use	19:42
ProLoser	The merge tool tortoisemerge is not available as 'tortoisemerge'	19:43
	any idea why i'm getting this?	19:43
drizzd	what's the git version on the client? the one that's cloning the repo?	19:43
	→ froschi2 joined	19:43
	← sagsousuke left	19:43
	→ sagsousuke joined	19:43
	→ Tureba joined	19:44
	← earthome left	19:44
	← froschi1 left	19:44
	← chrislerum left	19:44
	← cirwin left	19:44
dropandhop	1:1.7.0.4-1 on the server and on the client	19:44
drizzd	ProLoser: git does "type tortoisemerge" to determine if the merge tool is available	19:45
	ProLoser: so you should probably have tortoisemerge in your PATH	19:45
ProLoser	i tried	19:45
	do i put the path to the exe	19:45
	or the folder it contains?	19:45
drizzd	ProLoser: the folder which contains the executable	19:45
dropandhop	drizzd- are these permissions ok on the HEAD file? -rw-rw-r--	19:45
ProLoser	it didn't work :/	19:46
drizzd	dropandhop: yes	19:46
	ProLoser: how about type tortoisemerge.exe	19:46
dropandhop	drizzd- k...how can i confirm that i have a master branch setup on the server?	19:46
ProLoser	hmm	19:46
	it worked now	19:46
drizzd	dropandhop: in the "failed" clone, does git branch -a show origin/master?	19:46
	← Nemurenai left	19:46
ProLoser	i musta used the wrong %path% thingy	19:46
	← MattDiPa_ left	19:47
drizzd	"git branch -a"	19:47
	→ timcharp` joined	19:47
	← timcharp` left	19:47
drizzd	ProLoser: well, if it still doesn't work, you can override mergetool.tortoisemerge.cmd	19:47
	→ MattDiPasquale joined	19:47
	→ timcharp` joined	19:48
dropandhop	drizzd- do i just run the following b/c it doesn't give me anything: git branch -a	19:49
	→ marsilainen joined	19:50
	← rolfb left	19:50
drizzd	dropandhop: ok, strange	19:50
	looks like it doesn't find any branches	19:50
	what happens if you access http://$repourl/info/refs directly?	19:51
	I mean, point your browser to that address	19:51
ProLoser	drizzd, naw got it to work	19:51
	← opalka left	19:51
ProLoser	the path fixed it just had to double check it	19:51
drizzd	dropandhop: I wonder if maybe the http-backend script is not executed, and git just tried to clone using the old protocol	19:52
	ProLoser: ok, great	19:52
	→ _Vi joined	19:52
dropandhop	drizzd- if it's easier i can tell you all the steps i did to set this up, since i just did it	19:52
	← unreal left	19:52
	→ gebi joined	19:52
	→ unreal joined	19:52
	→ earthome joined	19:53
drizzd	dropandhop: I'm not sure it would help. But please check info/refs above, and maybe check your webserver logs to see if http-backend actually runs	19:53
dropandhop	gotcha...let me try to hit the location in my browser...just give me a few	19:54
drizzd	I wonder if we have a --no-dump-http switch somewhere	19:54
	← coyo left	19:55
	→ programble joined	19:56
	← mar77i left	19:57
	← MattDiPasquale left	19:58
	→ MattDiPasquale joined	20:01
	→ urkud joined	20:02
	→ cirwin joined	20:03
dropandhop	drizzd- i can't seem to see anything of interest in the apache log! how can that be	20:04
	← cannonball left	20:04
dropandhop	weird!	20:04
	drizzd- i can give you the big picture of what i did to set this up...if you have the patience	20:04
drizzd	dropandhop: what about info/refs?	20:08
	→ bdrewery joined	20:08
dropandhop	hitting them in my browser?	20:08
	→ mcella joined	20:09
drizzd	you can also check in the repo directly	20:09
bdrewery	How can I automatically check if a merge would present conflicts or not, without changing anything (or rolling back)?	20:09
drizzd	it should _not_ be there	20:09
	← mcella left	20:09
	→ warthog9 joined	20:09
drizzd	(in the repo)	20:09
	bdrewery: git merge --no-commit	20:10
	→ portal joined	20:10
Ilari	Aren't there some debug options in remote-http?	20:10
dropandhop	drizzd- the only file i have on the server in the info dir is exlude....is that what your getting to?	20:10
	← jtbandes left	20:10
dropandhop	Ilari- ahhh that would be really helpful!	20:10
portal	git-diff isn't showing changes which exist in the index, grr...	20:10
	git-status shows an "M" next to "foo.c" for example, and that "M" is shifted left, so I know that "foo.c" change was already added to the index	20:11
drizzd	dropandhop: yes, so git seems to clone something, but it ends up without any refs. I'm puzzled.	20:11
bdrewery	drizzd: what could I run to detect conflict?	20:11
portal	but I want to see which changes exactly are indexed	20:11
	→ steveire_ joined	20:11
	← rseifert_ left	20:11
Fangorn_	git diff --cached	20:11
Ilari	dropandhop: I don't remember exactly how to activate it, but I remember seeing debug traces.	20:11
drizzd	bdrewery: what I said	20:11
portal	ah, sweet; hmm, I wonder how I missed that when reading the mna page	20:12
	*man page	20:12
bdrewery	I'm basically writing a script to git merge only if a conflict is found, but to not if no conflict	20:12
dropandhop	drizzd- i do have some stuff in my refs folder on the server....i have a heads directory with a file called master in it	20:12
	← justicefries left	20:12
Ilari	dropandhop: That's branch master.	20:13
	→ drako joined	20:13
drizzd	dropandhop: maybe simply try git clone --verbose	20:13
dropandhop	drizzd- so it should be git clone http://repourl ?	20:14
	← urkud left	20:14
dropandhop	sorry	20:14
	git clone --verbose http://repourl	20:14
drizzd	yes, just add --verbose	20:14
dropandhop	thanks!	20:14
Fangorn_	portal: you can also used git diff --staged to diff between working and index	20:15
Ilari	dropandhop: The fallback (or no) to dumb http will be visible in requests made.	20:15
	→ jtbandes joined	20:15
dropandhop	I tried the clone in verbose mode, but it didn't provide any more info	20:16
	→ khmarbaise_ joined	20:16
	→ vpovirk joined	20:16
	← khmarbaise left	20:16
	khmarbaise_ → khmarbaise	20:16
vpovirk	I have a list of sha1 commit id's, and I want to sort them chronologically (or topologically, which should be equivalent); is there an easy way to do that?	20:16
dropandhop	i bet you this is something silly i missed...should i lay out how i set this up?	20:17
	→ Schmallon joined	20:17
	→ sixteneighty joined	20:18
	→ khaije\|selen joined	20:19
drizzd	dropandhop: try "git clone -v -v $url"	20:19
	← Belna_ left	20:19
dropandhop	k!	20:20
	← drako left	20:20
	→ drako joined	20:20
dropandhop	didn't seem to give any more info then before	20:21
	← ProLoser left	20:23
	→ unreal_ joined	20:24
	→ Nemurenai joined	20:24
	→ elmob joined	20:25
	← unreal left	20:25
	unreal_ → unreal	20:25
	→ baba_bubba joined	20:26
	→ odie5533 joined	20:26
	→ shruggar joined	20:27
	← marsilainen left	20:28
	← khmarbaise left	20:28
	← ive left	20:28
	→ rpg joined	20:28
dropandhop	drizzd- could it be a file permission thing? i feel like i should have set everything to g+s and i didn't?	20:29
	→ khmarbaise joined	20:30
	← _2of1_ left	20:30
	→ rbronosky joined	20:30
	← montylounge left	20:32
Ilari	dropandhop: Its $GIT_CURL_VERBOSE	20:32
	← richardbronosky left	20:34
	rbronosky → richardbronosky	20:34
	← g0bl1n left	20:35
drizzd	well, if I do git clone -v -v http://github.com/drizzd/diffit.git it at least tells me this	20:36
	Server supports multi_ack_detailed	20:36
	Server supports side-band-64k	20:36
	Server supports ofs-delta	20:36
	want 773ef1ccaa2569b088b6d421b3f4b07dacd32686 (refs/heads/master)	20:36
	done	20:36
	POST git-upload-pack (116 bytes)	20:36
	→ mar77i joined	20:36
dropandhop	i just tried -v -v and got nothing extra back on the clone...just the same error	20:37
	how would I use that $GIT_CURL_VERBOSE ?	20:37
	← sixteneighty left	20:38
Ilari	dropandhop: Its an environemnt variable.	20:38
drizzd	GIT_CURL_VERBOSE=1 git clone $url	20:38
	→ rolfb joined	20:38
	→ treak007 joined	20:38
	← corni left	20:38
portal	Fangorn_: aww, sweet, --stated; /me makes a mental note	20:38
dropandhop	ok- that environment variable got a lot of stuff!	20:39
	← LongBeach left	20:39
Ilari	dropandhop: Pastebin the output?	20:40
dropandhop	comin right up!	20:40
	← treak007_ left	20:41
dropandhop	http://pastebin.com/RvmvGy01	20:42
patrikf	vpovirk: for-each-ref, maybe?	20:42
	yaotti_aw → yaotti	20:42
	← treak007 left	20:42
	→ treak007 joined	20:43
	→ stamina joined	20:43
Ilari	dropandhop: It tries smart http but falls back because MIME type in response is wrong.	20:44
	dropandhop: Is that public repo?	20:45
dropandhop	ahh- so this is an apache issue it sounds like	20:45
	→ crazed joined	20:45
dropandhop	it's on a public server	20:45
crazed	if i've got a bunch of garbage in my git repo, how can i remove it without removing the files from my working dir?	20:45
	→ ericindc joined	20:45
patrikf	crazed: if all the files you want to keep are tracked, git clean might be what you want	20:46
Ilari	dropandhop: Mind if I try looking at what it actually sends for smart transport refs request?	20:46
	← cilly left	20:46
	→ urkud joined	20:46
ericindc	If I have branch X which I created a branch Y from it's HEAD and made a series of commits to Y, how do I place the changes from Y onto the end of X	20:47
patrikf	ericindc: rebase, and/or merge	20:47
dropandhop	Ilari- no, i don't...thanks very much	20:47
Ilari	dropandhop: Also 200 response...	20:47
crazed	patrikf: i have a bunch that i don't want tracked as well, i can match them with a glob, is it easy to just untrack them?	20:48
	← avar left	20:48
ericindc	patrikf: i tried git rebase X with Y checked out, but it says it's up to date	20:48
	so i assume that means i was rebasing in the wrong direction	20:48
	→ felipec joined	20:48
ericindc	rebase Y and I get " Needed a single revision"	20:48
patrikf	ericindc: I think you want to merge Y into X.	20:48
Ilari	dropandhop: The result looks like Field15 home page.	20:48
	← whaley left	20:49
Ilari	dropandhop: So apache is sending totally wrong data as response.	20:49
patrikf	crazed: just "git rm" them and commit...	20:49
	← treak007 left	20:49
ericindc	patrikf: can it not be rebased?	20:49
crazed	that removes them from the working directory though	20:49
	→ avar joined	20:50
	← elmob left	20:50
	← rayners left	20:50
patrikf	ericindc: you misunderstand the idea of rebase, I'm afraid. if Y was rebased on X, then merging Y into X will be a fast-forward merge (i.e. a "silent" merge, a merge without a merge commit)	20:50
	← timcharp` left	20:50
patrikf	crazed: if you just want to untrack them, rm --cached	20:50
	→ scootsm joined	20:51
patrikf	crazed: ah, now I see what you mean. forget about git clean	20:51
	cpg → cpg\|brb	20:51
scootsm	Is there a tool that pretends to be SVN, but is actually Git?	20:51
crazed	ah --cached	20:51
	that sounds useful	20:51
ericindc	patrikf: you're right, i'm misunderstanding something	20:51
	→ thiago_cgh joined	20:51
patrikf	crazed: they'll still be there in your history though. if you want to rewrite history as if they never were there (useful with sensible information, or large amounts of data), filter-branch can do that	20:52
Ilari	scootsm: There were plans for that, but AFAIK no implementation.	20:52
	→ rayners joined	20:53
	→ treak007 joined	20:53
	→ leo2007 joined	20:53
scootsm	Damn it.	20:53
	Hmm, are there any servers that can receive SVN commits, but actually does Git commits--to stick it to the man?	20:54
	I suppose that's why God made Google	20:54
	→ psoo joined	20:55
	→ Math joined	20:55
dropandhop	Ilari- we take any request that doesn't exist and send them back to our homepage...so that makes sense	20:55
	should i just add .git to the known file types?	20:55
	← jaql left	20:55
patrikf	scootsm: is this (company) politics, or why on earth would you want that? just curious.	20:55
	→ bousquet joined	20:56
	← Heimidal_ left	20:56
Math	got some problems, someone got a conflict, didnt resolve it properly and pushed it, but now even reverting his commit doesnt fix the screwups, any hints?	20:56
	guess I could diff it	20:56
Ilari	dropandhop: info/refs in git repo should point to http-backend cgi script.	20:56
	← unreal left	20:57
	→ unreal joined	20:57
	← ericindc left	20:57
	← simosx left	20:57
scootsm	patrikf: Column A, Column B. It's a bit weird. We have an SVN server, internally, and separate (private) Github accounts. I want to push my code to Github. But I also just want to use Git, because Git is Great. While those are all very simple problems to solve, I'm using a text editor called "Coda" which is a really great text editor. But it supports SVN, not Git.	20:57
dropandhop	Ilari- sorry to be slow...what does that mean?! hehe	20:57
	→ ericindc joined	20:58
scootsm	patrikf: Complicating matters further, my project is not meant for our SVN repo. Rather, it's a meant to stay quiet internally and, therefore, should go on my Github.	20:58
	→ treak007_ joined	20:58
scootsm	patrikf: Yet another complication is that my company is quite strange.	20:58
	← treak007 left	20:59
patrikf	scootsm: so its politics, as guessed. no further questions ;-)	20:59
Ilari	dropandhop: It means that smart-http is not set up correctly... I have no clue of how to configure apache, sorry.	20:59
	← froschi2 left	20:59
dropandhop	Ilari- that's ok! thanks for everything so far	20:59
	← krawek left	20:59
	← justin-george left	20:59
scootsm	patrikf: Oh nuh-uh, you dun di'ant just tl;dr me	21:00
	← ralf_e left	21:00
scootsm	I also speak jive	21:00
	← redondos left	21:00
	→ redondos joined	21:01
patrikf	that's too distorted for my bad english. how do you pronounce the semicolon?	21:01
scootsm	It's "Too Long; Didn't Read"	21:02
	→ justin-george joined	21:03
	→ krawek joined	21:03
patrikf	scootsm: oh, jargon fail for me. indeed i read the bulk of your text only after responding. ;-)	21:05
	→ rseifert_ joined	21:05
	← rseifert left	21:05
	→ Heimidal_ joined	21:05
jtbandes	What's the difference between merge and pull?	21:05
	Like, can I pull from local branches?	21:06
patrikf	jtbandes: pull = fetch + merge	21:06
jtbandes	Oh, is that all?	21:06
patrikf	yes, it's that boring	21:06
	← richardbronosky left	21:06
patrikf	saves you some typing, that's it. :)	21:06
jtbandes	Hm, I'm seeing "Simple merge did not work, trying automatic merge." & "Merge made by octopus.", what does that mean?	21:06
	← brntbeer left	21:07
bdrewery	it's trying different merge methods	21:07
patrikf	jtbandes: see man git-merge, "MERGE STRATEGIES"	21:07
Gitbot	jtbandes: the git-merge manpage can be found at http://git.or.cz/man/git-merge	21:07
	← unreal left	21:07
jtbandes	Oh, just because I did more than one branch	21:07
	→ unreal joined	21:08
	→ malumalu joined	21:08
	← unreal left	21:09
	→ toabctl joined	21:09
	→ unreal joined	21:10
	→ richardbronosky joined	21:10
	← khmarbaise left	21:11
	← coppro left	21:11
	→ brendon_work joined	21:12
	← Heimidal_ left	21:13
	← thiago_cgh left	21:15
	← Nemurenai left	21:15
	← leo2007 left	21:15
	→ treak007 joined	21:16
	→ Mage joined	21:17
	← radioman-lt left	21:17
	cpg\|brb → cpg	21:17
	→ leo2007 joined	21:18
	← flazz left	21:18
	→ grumpytoad joined	21:18
	← richardbronosky left	21:18
	→ flazz joined	21:19
	← rolfb left	21:19
	← treak007_ left	21:20
	← earthome left	21:20
	← treak007 left	21:20
	→ rbronosky joined	21:20
	→ jtdowney joined	21:21
	← spearce left	21:21
	← rbronosky left	21:21
ajsie	what is the difference between HEAD and master?	21:21
	→ spearce` joined	21:21
	← Math left	21:21
	→ treak007 joined	21:21
	← ehabkost left	21:22
	← hohoho left	21:22
tango_	master is the master branch	21:22
IslandUsurper	ajsie, master is a branch, HEAD points to the current branch	21:22
	→ rbronosky joined	21:22
halfline	ajsie: master is a branch, HEAD is just a special name for the last commit on the current branch	21:22
tango_	HEAD is the commit currently checked out	21:23
IslandUsurper	tango_'s definition of HEAD is probably the most accurate	21:23
	← rook2pawn left	21:23
leo2007	HEAD is more like a sym link	21:24
	→ coppro joined	21:25
selckin	implementation detail	21:25
	← gbacon left	21:25
ajsie	okay ... master is a branch...and HEAD is the last commit	21:25
shruggar	HEAD is a pointer to the currently-checked-out ref, or, if no ref is checked out, the currently checked-out commit	21:26
ajsie	so HEAD is always the LAST commit? it cant be the second last?	21:26
	→ thiago_cgh joined	21:26
	← spearce` left	21:26
ajsie	hmm	21:26
FauxFaux	reset --hard will change HEAD.	21:27
davr	no, HEAD is not always the LAST commit. It's always the commit that you currently have checked out	21:27
	git allows this funny thing where you can check out a certain revision but not be on a branch	21:27
	it's called "detached head", there's a FAQ entry that explains in more detail I think	21:27
leo2007	there are quite a few videos that explain this well.	21:27
shruggar	FauxFaux, unless you're on a detached HEAD, reset --hard will NOT change HEAD, though it will change the commit which HEAD eventually points to	21:28
ajsie	okay thanks guys	21:28
	← sylr left	21:28
	← s0ber left	21:29
	→ zenazn_ joined	21:29
	← zenazn left	21:29
	← unreal left	21:29
	→ unreal joined	21:29
	→ sylr joined	21:30
	→ s0ber joined	21:30
	← dropandhop left	21:31
shruggar	FauxFaux, technically you can point HEAD to ref: refs/heads/foo, and point refs/heads/foo to ref: refs/heads/master ;)	21:34
	I don't know what that feature is intended to be used for, though	21:35
	FauxFaux ponders.	21:35
jtbandes	If I merge a topic branch in, does it stay around? or is there a way to merge it in and get rid of it?	21:35
FauxFaux	Oh, I'm being dumb.	21:36
shruggar	jtbandes, merge it, then delete it? the "branch" is just a ref.	21:36
	jtbandes, if you don't want the conceptual merge, ie, the graph, then merge with --squash	21:36
	← aziz_ left	21:37
ajsie	are u all using git commands in the cli?	21:37
	← Schmallon left	21:37
ajsie	isnt it easier with GUI tools	21:37
	like Netbeans Git Plugin	21:37
	← sylr left	21:37
	← sftp left	21:38
	← sammcd left	21:38
mar77i	ajsie: depends on what your definition of ease would be. plugins always prove an additional source of failure and an at least minor loss of flexibility:)	21:38
	← nevyn^ left	21:38
	→ sylr joined	21:39
	← m104 left	21:41
shruggar	ajsie, I'll use a git gui as soon as I see one which makes any sense at all ;)	21:41
	← bgy_ left	21:41
ajsie	:)	21:41
	→ skoop joined	21:42
	← shader left	21:42
mar77i	s/prove/provide/ seems to make more sense	21:42
	← poseid left	21:42
	→ sylvainr joined	21:42
jtbandes	Is that the right workflow? if I have some topic branches, to just checkout master and merge them in?	21:43
	← toabctl left	21:43
shruggar	jtbandes, when theyre ready to be merged into master, yeah. As opposed to what?	21:43
	← tstclair left	21:45
jtbandes	shruggar: rebasing? I dunno	21:46
	← Dave^\| left	21:46
shruggar	jtbandes, well rebasing is just outright the totally wrong thing, if that's what you're asking	21:46
	← tar- left	21:46
jtbandes	I've read things about people using it with/before merges	21:47
shruggar	rebasing can be thought of (in a gross oversimplfication) as the opposite of a merge	21:47
bdrewery	I know what rebasing is, and that's just confusing to try and figure out what you mean by that	21:47
vpovirk	rebasing makes sense if you want to keep a linear history for a project, and you are keeping your topic branches private	21:47
	→ Pewpewarrows joined	21:47
vpovirk	and you have to be sure you rebase the right thing onto the right thing	21:47
	→ ignatp joined	21:48
shruggar	jtbandes, often people will rebase with upstream prior to submitting to upstream, as it ensures that merge-conflicts are taken care of by the maker of the topic branch instead of by the maintainer of the upstream branch. But its not necessary, it's just sometimes considered polite	21:48
	← rbronosky left	21:49
jtbandes	Ah I see	21:49
	→ Yuuhi joined	21:49
	→ nakkor joined	21:49
	→ richardbronosky joined	21:49
jtbandes	But that could be a good idea locally too	21:49
	To rebase an older branch before merging into master	21:49
vpovirk	you pretty much have to if you're submitting changes as a patch series; otherwise, they won't apply cleanly	21:49
shruggar	if you have a long-running topic branch, I think it makes the history look better to occasionally rebase with upstream, rather than merging upstream	21:50
	→ rolfb joined	21:50
jtbandes	shruggar: you mean instead of merging upstream into the topic before working on the topic more?	21:50
	← rolfb left	21:51
shruggar	jtbandes, right. It should still be reintegrated using merge	21:51
jtbandes	At the end.	21:51
shruggar	righ	21:51
	← sylr left	21:51
	← nakkor left	21:51
	→ nakkor joined	21:52
	→ sylr joined	21:52
	→ bgy_ joined	21:52
	← nakkor left	21:52
shruggar	jtbandes, think of it like this: what is going to make the most sense when you're viewing history, or trying to track down a bug?	21:52
	→ EricInBNE joined	21:53
jtbandes	Yeah, you could rebase first to make it more obvious what was changed	21:53
	← rorist left	21:53
vpovirk	afaik your two choices are just "rebase" and "merge", there are slightly different ways to do them that will give you the same result, and I always thought people who merged topic branches were crazy	21:54
	unless it's a long-running and public topic branch	21:54
	→ flaguy48 joined	21:54
	→ poseid joined	21:57
	→ jortel joined	21:57
	← warthog9 left	21:58
	→ xuxc000 joined	21:58
	← urkud left	21:58
	← IslandUsurper left	21:59
	← flazz left	21:59
	← zenazn_ left	22:00
	→ nevyn^_ joined	22:01
	yaotti → yaotti_aw	22:01
	← josh0x0 left	22:02
	← sukima left	22:03
	← batok left	22:03
	← Paraselene_ left	22:03
	← poseid left	22:03
	→ Blackguard joined	22:03
	← cirwin left	22:03
	← sylvainr left	22:04
	nevyn^_ → nevyn^	22:05
	→ alexmchale joined	22:05
	← bousquet left	22:07
	← jortel left	22:08
	→ bousquet joined	22:08
	← alexmchale left	22:08
	← priidu left	22:09
	→ simplechat joined	22:09
	→ Joeconyers joined	22:09
	→ warthog9 joined	22:09
	← warthog9 left	22:09
	← richardbronosky left	22:10
	← altrux left	22:10
	← vpovirk left	22:11
	← newbie\|2 left	22:11
	yaotti_aw → yaotti	22:12
	→ priidu joined	22:14
	← loincloth left	22:14
	← Titosemi left	22:17
	← masterkorp left	22:17
	→ warthog9 joined	22:19
	→ cirwin joined	22:19
	← portal left	22:19
charon	by that logic, most of git.git would be crazy	22:20
	← bauruine left	22:21
	→ Paraselene_ joined	22:21
	→ brntbeer joined	22:22
	→ Heimidal_ joined	22:22
	← sako left	22:24
	← tvw left	22:24
	← Heimidal_ left	22:25
	→ neeraj joined	22:25
	← eighty4 left	22:25
	← drako left	22:25
	← kipras left	22:26
	← jonaskoelker left	22:26
	→ mackstann__ joined	22:26
hachi	why does git-rev-parse --tags print out the usage information for me?	22:27
	rev-list	22:27
	sorry	22:28
	# git rev-list --tags	22:28
	usage: git rev-list [OPTION] <commit-id>... [ -- paths... ]	22:28
mackstann__	say i have revision A that made one desirable change. then, much later, i have a giant revision that changed tons of stuff, including reverting revision A. i now want to essentially re-apply revision A, but the git-merge man page tells me it'll skip it in this scenario. how do i force it to re-apply that old revision?	22:28
hachi	I can understand that I don't have any tags for candidate, but shouldn't that just mean it gives me zero lines of output?	22:28
	← psoo left	22:28
	← sylr left	22:29
	→ bauruine joined	22:29
neeraj	hi I am trying to pull someone else changes/commits from a git repository using git pull	22:29
	git pull git+ssh://neeraj-guest@git.debian.org/git/collab-maint/sugar-record-activity.git	22:29
	You asked to pull from the remote 'git+ssh://neeraj-guest@git.debian.org/git/collab-maint/sugar-record-activity.git', but did not specify	22:29
	a branch. Because this is not the default configured remote	22:29
	for your current branch, you must specify a branch on the command line.	22:29
	← drizzd left	22:29
	adamholt → adamholt_away	22:29
	adamholt_away → adamholt	22:30
davr	mackstann__: have you already pushed the giant revision to a public repo? if not, i'd reccomend rebasing & splitting it apart into multiple commits	22:30
	← baba_bubba left	22:30
shruggar	mackstann__, try cherry-pick?	22:30
mackstann__	davr: yes, the giant commit is pretty much set in stone	22:30
davr	neeraj: you need us to explain the error to you in more detail or something? it seems pretty clear to me	22:30
	← hugoxrosa left	22:30
mackstann__	shruggar: ah, looks like cherry-pick is exactly it	22:31
	→ enherit joined	22:31
	→ masterkorp joined	22:31
davr	i thought cherry-pick skipped things you already had, but maybe I am mistaken. can't hurt to try it	22:31
	← roxoft left	22:32
shruggar	I thought it might detect an already-applied patch (as rebase apparently does), but I couldn't see an option to disable that behaviour, so I'm guessing it doesn't	22:32
	→ gbacon joined	22:32
	← nevyn^ left	22:32
hachi	how does one figure out what branch is being tracked? do I have to use 'git config' to discover this	22:32
	or is there a faster way?	22:32
	programmatically I mean	22:32
neeraj	davr, yes some detailed explanation or relevant link giving details would be much appreciated :)	22:32
hachi	I don't want to parse the output of 'status'	22:32
mackstann__	well it says it applies an existing commit, which seems to imply that it would re-apply it	22:33
shruggar	hachi, git rev-parse --symbolic-full-name @{u}	22:33
	mackstann__, "existing" doesn't imply "existing in the current line of history"	22:34
davr	neeraj: basically you just need to specify which branch to pull from. you just told it to pull from a repo which may have many branches in it, git doesn't know which remote branch to merge into your current one	22:35
	→ nakkor joined	22:35
hachi	shruggar: what's the @{u} thing called? I want to read about it...	22:36
	oh	22:36
	I see it now	22:36
	→ zorzar joined	22:36
	← jfkw left	22:36
	cpg → cpg\|brb	22:39
	← jkp left	22:39
	← zorzar_ left	22:39
	← MrPunkin left	22:40
	→ Oxf13 joined	22:40
mackstann__	shruggar: perhaps not. the wording is ambiguous	22:40
Oxf13	sitaram: ping are you around? I'm about to try out latest HEAD from master...	22:40
	→ sako joined	22:41
	← jrmuizel left	22:42
neeraj	ok,, I have no idea what all branches this git repository has. I had creates this using git push only and I was able to pull changes by other people by using git pull	22:43
	← steveire_ left	22:45
neeraj	s/creates/created	22:45
	← brntbeer left	22:45
mackstann__	hm, indeed, cherry-pick will not re-apply a change that is an ancestor	22:47
	cpg\|brb → cpg	22:49
	← tizbac left	22:49
	→ vinicius joined	22:52
	← LiamH left	22:52
shruggar	mackstann__, git diff <commit>^!\|git apply I suppose is a fall-back	22:52
vinicius	hi... is this list for git support or development?	22:53
shruggar	mostly support	22:53
vinicius	shruggar, nice...	22:53
	im trying to build a little post-update script	22:53
mackstann__	shruggar: yeah but then it's just a 2-way merge which should mean lots of conflicts, right?	22:53
	← xiong left	22:54
vinicius	ive got a web server for production, and i want to have some repos available for an apache folder, so i set something like an "auto-repo" throught "git pull" on post-update hook	22:55
shruggar	mackstann__, as opposed to what? There are three trees involved in the command I gave	22:55
	vinicius, don't use "pull" in non-interactive scripts	22:55
vinicius	shruggar, humn... ok	22:55
	what, then?	22:56
	← masterkorp left	22:56
	← becomingGuru left	22:56
neeraj	Ok, I am using gbp-clone --pristine-tar and will commit my changes	22:56
	← flaguy48 left	22:56
shruggar	vinicius, for deployment, git reset --keep is safer, since it won't ever leave your tree in a conflicted state, and will respect non-fastforward pushes	22:56
vinicius	shruggar, THAT was what im look for!	22:57
	← raichoo left	22:57
vinicius	ill try it	22:57
	=D	22:57
	← bradallen137_ left	22:58
	← giallu left	22:58
	← shruggar left	22:58
	← TeckniX left	22:58
	← rpg left	22:59
	→ sftp joined	23:00
	→ zenazn joined	23:02
	← gebi left	23:03
	← didlix left	23:03
	← d0k left	23:04
hachi	how do you determine if the working tree is dirty as a boolean?	23:04
	also, is there a way to perform the actions of a pull excluding the fetch step?	23:07
	← Ryback_ left	23:07
hachi	I would love to invoke a command and have it do the rebase or merge depending on what is in the configs, not manually choose myself	23:07
	← scootsm left	23:08
	← MattDiPasquale left	23:09
	→ g0bl1n joined	23:09
	← andrewl left	23:10
	→ andrewl joined	23:10
	yaotti → yaotti_aw	23:11
	← skoop left	23:14
	← Eaven left	23:17
vinicius	ok.. my script now copy the git repo to a clone on a web server...but how can i use paths on post-update while im not on a terminal like tortoise git, for example?	23:18
	← nakkor left	23:18
	→ Eaven joined	23:18
	← jtbandes left	23:19
	← workmad3 left	23:19
hachi	is `git merge @{u}' basically what git pull calls after calling git fetch?	23:20
	← priidu left	23:23
	← apsysoev left	23:23
	← voxpelli left	23:23
	← malumalu left	23:25
	← bentob0x left	23:27
	← Paraselene_ left	23:27
	→ arooni joined	23:28
arooni	hi folks; i want to completley reset my local copy to the latest remote copy; how can i do this?	23:28
	← rayners left	23:29
	← marienz left	23:29
cmyers	arooni: use `git log` to figure out the hash you want to set it to, then `git reset --hard <hash>`	23:30
	→ marienz joined	23:30
cmyers	if you haven't committed anything, then HEAD is probably what you want, so you can do `git reset --hard HEAD`	23:30
	→ Tuller joined	23:32
	→ rpg joined	23:33
	→ tty1 joined	23:34
tty1	hey guys im having a weird probem... is there a way to specify ont he git command line what private ssh key to use? i cant seem to clone from this box and it seems to be some sort of permission problem	23:35
	← neeraj left	23:35
	→ ralf_e joined	23:36
	→ mattwynne joined	23:36
	← Haaarg left	23:37
	← mattwynne left	23:37
frsk	You can specify it in .ssh/config instead	23:38
	→ nakkor joined	23:38
odie5533	what personal data does git pull from my computer?	23:39
	does it use my computer name anywhere?	23:39
cmyers	odie5533: I'm not sure that question even makes sense. Git pulls data from other computers. If you `push` your changes, it pushes the data in those changes, is that what you mean?	23:40
	Standard commit metadata contains whatever you set as your "username" and "email", and the date and time.	23:40
odie5533	okay, that was what I wondering yeah.	23:40
cmyers	the default email is "$USER@$HOSTNAME"	23:41
	so if you want someting else, be sure to set it. see `git help config`	23:41
odie5533	ooo. that's not good..	23:41
	→ anathematic joined	23:41
	← felipec left	23:41
	← dreiss left	23:42
odie5533	can I change the username/email of a commit after it's been pushed to a server?	23:42
	and does it use any key to sign commits?	23:44
AAA_awright	You might be able to --force it but that could cause problems for people pulling	23:44
	No	23:44
	odie5533: Re-create the commit with git commit --amend, check it with git log, re-push it with git push --force	23:46
	That's not something that is normally sugguested but it should be possible	23:46
tty1	hmm	23:46
	→ mquin joined	23:46
	← frogonwheels left	23:47
	→ frogonwheels joined	23:47
tty1	i just did a clone on a blank repo (just inited ont he server side as a bare repo).. i added some files and tried to commit then push.. during push i got the error: No refs in common and none specified; doing nothing.	23:47
	Perhaps you should specify a branch such as 'master'.	23:47
Ilari	tty1: Yeah, .ssh/config is the only way excluding remote-ext (which is not in master).	23:47
tty1	how do i do that? just switch branches	23:47
	← thm left	23:47
tty1	or maybe i need to make the master branch a remote or something	23:48
	← warthog9 left	23:48
Ilari	tty1: 'git push <remote> <branch>',	23:48
tty1	i seem to always forget	23:48
	so what would remote and branch be if its the master though?	23:48
	← vinicius left	23:49
	← sivy left	23:49
tty1	there doesnt seem to be a master	23:49
	→ dyjytyl joined	23:50
Ilari	tty1: 'git remote' to see valid remotes, and 'git branch' to list valid branches.	23:51
tty1	hmm	23:52
	there is just origin in the remotes and nothing int he branches	23:52
	hmm	23:53
odie5533	when connecting to github, do I first create the git thing on my own computer with git init, or does it does that for me?	23:53
	← jtdowney left	23:53
Ilari	tty1: You are saying output of 'git branch' is empty?	23:54
	odie5533: You need local repository, either created by init (create new repository) or clone (copy existing repository).	23:54
tty1	Ilari: it actualy seems to ahve a master now (must have been one o the commits i was trying)... so i tried git push origin master.. and it is chugging away but produced the error: error: insufficient permission for adding an object to repository database ./objects	23:55
	so i traded one error for a new one it seems	23:55
odie5533	will git init add all the files in my directory?	23:55
	← dyjytyl left	23:55
Ilari	tty1: Permissions errors on remote side.	23:55
tty1	Ilari: cant imagine why.. let me try this over...	23:56
Ilari	tty1: Are you sure you got the right user account on remote side?	23:56
cmyers	odie5533: all `git init` does is create an empty repo - you then add what you want.	23:56
tty1	Ilari: seems to work fine for all the other repos i have ont he same remote	23:56
cmyers	odie5533: I suggest reading the Git Magic tutorial, it has a really good begining guide: http://www-cs-students.stanford.edu/~blynn/gitmagic/index.html	23:56
odie5533	thanks for the link, reading now	23:57
cmyers	I think specifically chapter 2 has the most important bits, and chapter 3, if you want to jump right to it.	23:57
	but the entire thing is a great read.	23:57
	← xuxc000 left	23:58
Ilari	tty1: That error comes if trying to create files on remote repository gets access denied error from operating system.	23:58
tty1	hmm	23:59
	→ Haaarg joined	23:59

Logs Search ←Prev date Next date→

Channels Documentation