IRCloggy #git 2009-05-26

Provider of IRC logs since 2005.

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

	2009-05-26
	→ bremner_ joined	00:00
	→ reprore__ joined	00:02
	← mclem left	00:03
	→ drmclem joined	00:03
	← reprore__ left	00:03
	← bremner_ left	00:04
	→ reprore__ joined	00:04
	→ bremner_ joined	00:04
	← reprore left	00:05
	← drmclem left	00:05
	→ drmclem joined	00:06
	→ scarabx joined	00:06
	→ cads joined	00:06
	← bremner_ left	00:06
cads	how do I create a new git in a directory, where I can add files and stage and commit?	00:07
	I know how to clone from git-hub	00:07
RandalSchwartz	go to the directory, type "git init"	00:07
	then "git add ." and "git commit"	00:07
	← Nazcafan left	00:07
RandalSchwartz	that'll create an initial commit of the current file config	00:07
	RandalSchwartz waits for cads to catch up	00:08
cads	thanks	00:08
RandalSchwartz	if you had gone through teh basic tutorial, you would have seen that.	00:08
	I suggest you do the basic tutorial	00:09
	← djgera left	00:09
RandalSchwartz	definitely not a waste of time	00:09
cads	yeah, I went through some github tutorial	00:09
RandalSchwartz	no - the one that comes with git	00:09
	highly recommended	00:09
cads	where's the basic tutorial	00:09
	→ memiux_ joined	00:10
andres	cads: "git help tutorial"	00:10
	← memiux left	00:10
andres	cads: or man gittutorial	00:10
Gitbot	cads: the gittutorial manpage can be found at http://git.or.cz/man/gittutorial	00:10
RandalSchwartz	but don't look on the website	00:10
	it's on your disk already	00:10
	git help tutorial	00:10
cads	wow, nice	00:11
	→ tvakah joined	00:14
	← dodo_the_last left	00:15
	← tvakah left	00:16
	→ tvakah joined	00:17
	→ rettub_ joined	00:18
Chani	and why does git push not have --quiet either?	00:18
	← mitkok left	00:19
	← drmclem left	00:22
	→ hobodave joined	00:24
	← Mkop left	00:31
	← Zonker left	00:32
	← tntcoda left	00:34
	almostautomated → almostautomated\|	00:37
	→ petercoulton joined	00:41
	→ nighthwk1 joined	00:42
	← WinterWeaver left	00:43
	← memiux_ left	00:47
	→ memiux joined	00:47
	→ mithro joined	00:48
	← memiux left	00:51
	→ memiux_ joined	00:51
	→ _ joined	00:53
	_ → Guest6411	00:53
	→ talexb joined	00:54
	← sfwc left	00:56
Pieter	man git-update-ref	00:56
Gitbot	Pieter: the git-update-ref manpage can be found at http://git.or.cz/man/git-update-ref	00:56
	← Guest6411 left	00:58
	→ Guest6411 joined	00:59
	← cads left	00:59
	→ montylounge joined	01:00
	→ ia joined	01:04
	→ memiux joined	01:05
	← memiux_ left	01:05
	→ bwwww joined	01:07
	← ben_h left	01:08
	← scarabx left	01:11
	← Guest6411 left	01:11
	→ Guest6411 joined	01:11
	← Radar left	01:14
	→ Radar joined	01:15
	← solydzajs left	01:18
	← SRabbelier left	01:18
	← memiux left	01:23
	← markelikalderon left	01:23
	→ reprore_ joined	01:24
	→ memiux joined	01:27
	← jrmuizel left	01:28
	joshthecoder → joshthecoder_afk	01:29
	→ schacon joined	01:30
	← mithro left	01:38
	→ mithro joined	01:38
	← squentin_ left	01:41
	→ markelikalderon joined	01:41
	← montylounge left	01:45
	→ jrmuizel joined	01:46
	← reprore__ left	01:47
	← jrmuizel left	01:47
	← Guest6411 left	01:49
	→ jrmuizel joined	01:49
	→ Guest6411 joined	01:49
	→ reprore__ joined	01:52
	→ WALoeIII joined	01:55
	→ repro____ joined	01:55
	← ajonat left	01:56
	→ cedric_ joined	01:56
	← reprore_ left	01:57
	← codebrulee left	01:57
	→ xorg62 joined	01:57
	← bwwww left	01:59
	→ cemerick joined	02:00
	→ limx__ joined	02:00
	← reprore__ left	02:04
	← Guest6411 left	02:05
	← cemerick left	02:05
	→ jksz joined	02:06
	→ jchris joined	02:07
	→ travisjeffery joined	02:08
	← limx__ left	02:13
	→ Guest6411 joined	02:13
	→ Minimiscience joined	02:14
	→ reprore_ joined	02:14
	→ wmoxam joined	02:16
	→ parth-vader joined	02:19
	← parth-vader left	02:19
	← blaenk left	02:20
	→ cytrinox_ joined	02:21
	→ lmarinho joined	02:27
	← memiux left	02:27
	→ blaenk joined	02:29
	→ repror___ joined	02:31
	← repro____ left	02:31
	→ bgerber joined	02:37
	→ JasonWoof joined	02:38
	→ RurouniJones joined	02:40
	← Guest6411 left	02:42
	→ TekLok joined	02:42
	→ summerbug joined	02:47
	→ Mkop joined	02:48
	← reprore_ left	02:48
	→ memiux joined	02:50
	→ tjafk2 joined	02:52
	→ Guest7710 joined	02:54
Guest7710	hello.	02:55
	Guest7710 → UltraBob	02:55
UltraBob	have a (probably stupid) question	02:55
	→ montylounge joined	02:56
UltraBob	I am in a situation where I am in a subnet and have a local ip address, and it is not feasible to accept ssh connections. I am connecting to a server that does have a public static ip, so basically my connection is one way	02:56
	→ ericindc_ joined	02:56
	→ x_or joined	02:56
UltraBob	I can connect from my local machine to the server, but not the other way around	02:56
	my thought is to have a master repository locally where I work on files and then push them to the server	02:57
	my question is how do I clone my local repository onto the server, and then how do I manage pushing to it?	02:58
cehteh	isnt that the common case?	02:59
	i mean i dont use my server to call my laptop	02:59
	← montylounge left	02:59
cehteh	what do you mean by 'master' repository .. git makes no such distinctions	03:00
UltraBob	ok, that was just a distinction in my mind. like I said probably a stupid question, but I'm not getting my head around it todday	03:00
cehteh	usually you have the one (or more) where you work on and publish your work on some server	03:00
UltraBob	so in this common case how do you do it?	03:00
cehteh	the repository on the server should be bare	03:01
	you work and commit locally and push that up to the server	03:01
	if you server is shared between developers, other may push there too .. in their own repository, or in their own branches, or maybe even in some central merge branch (master?)	03:02
UltraBob	ok, I didn't explain in enough detail	03:02
	I want the rep on the server to be usable as a directory in the website	03:02
	→ ekristen joined	03:03
UltraBob	I am editing plugin files locally and then when I push them, I want them to be in the same form on the server	03:03
ekristen	how do I checkout a repo up to a specific commit id?	03:03
	→ incognit_O_ joined	03:03
cehteh	UltraBob: thats little more complicated because it cant be bare then, actually its not recommended	03:03
	← hobodave left	03:04
incognit_O_	maybe... :-))	03:04
	ati gasit si cripta si mausoleul ?!!?	03:04
cehteh	but well we do that with our website too :) .. usually you reset the working copy in the post-recieve hook , there are examples laying around	03:04
incognit_O_	Maus-o-leul ??	03:04
UltraBob	cehteh I suppose I could have a bare repository on the server and then run a script to log into the server and pull to a working rep on the same server	03:05
	← Minimiscience left	03:05
	joshthecoder_afk → joshthecoder	03:05
	incognit_O_ → umbritool	03:05
UltraBob	cehteh with that second rep being the one in the web directory	03:05
	→ dreiss joined	03:05
cehteh	UltraBob: yes that would work too	03:05
umbritool	La Festa si La Fildes sunt nebunii !!!	03:05
	← umbritool left	03:05
cehteh	http://lumiera.org/.git/hooks/post-receive << our hack, not the best/recommended way but works	03:06
	ekristen: prolly you want to create a branch for that? .. git checkout -b temporary-work $sha1	03:07
ekristen	thanks	03:07
UltraBob	cehteh ok, so if you can help me develop this a little further. assuming I go twith the two reps on the server route. first I create a bare rep on the server, then I need to 1) clone my local rep to it. 2) clone that rep to my working copy 3) figure out how to push to the bare rep. Could you help me with how to do 1 and 3?	03:08
	→ sfwc joined	03:08
cehteh	UltraBob: just similar in a post-recieve hook	03:08
	first you set it up git clone /path/to/bare public_checkout	03:09
	→ summerbug_ joined	03:09
	← blaenk left	03:09
	← sfwc left	03:09
	← jrmuizel left	03:09
almostautomated\|	UltraBob: not sure of your traffic; but you might also want to think of using 3 repos. 1 bare, then two that you'd switch the load to via apache ( or whatever you use ) It'd suck to cause errors for currently connected sessions	03:10
	← chocolate_jesus left	03:10
	→ MachX joined	03:10
cehteh	then you just add cd /path/to/public_checkout; git pull; to the bare post-recieve hook	03:10
UltraBob	cehteh can we use some pseudo connect info so I'm sure I understand? say ssh://my.server/path/to/rep	03:10
cehteh	maybe you have to unset some GIT_* env vars not to confuse git	03:10
	this was server local paths	03:11
	could be almost verbatim used, just with the correct paths	03:11
	← ekristen left	03:12
UltraBob	see the place where I'm confused is with how to get my local data up to the bare rep on the server	03:12
cehteh	and the env unsetting (i never tried it this way, improvizing :))	03:12
	git push	03:12
	.. in steps:	03:12
	on your local	03:12
	git remote add myserver [email@hidden.address]	03:13
	then you can use git push myserver ...	03:13
UltraBob	aha!	03:13
cehteh	with some configs you can configure what to push by default or just give branches explicitly	03:13
UltraBob	that was the part I couldn't get to come to the surface from when I was first experimenting with git a few months ago	03:14
cehteh	you could use that w/o registering the remote too	03:14
	git push [email@hidden.address] ...	03:14
UltraBob	so you can push to a bare rep with nothing in it?	03:14
cehteh	or use ssh:// urls .. whatever you prefer	03:14
UltraBob	no need to establish any connection there first	03:14
cehteh	yes, actually thats the way i prefer to initialize repos	03:15
	mkdir project; cd project; git init --bare	03:15
	← WALoeIII left	03:15
UltraBob	cehteh ok just for further personal fortification. Say that I then tried to run a push from a completely seperate repository to that same bare rep, what would happen?	03:15
cehteh	there are some other spells to do that, rsyncing it up or cloning will work too	03:16
	← lmarinho left	03:16
cehteh	depends on what branches you push and what state they are	03:16
	← jchris left	03:16
cehteh	usually git refuses to do destructive things but allows fast forwards	03:16
UltraBob	say that I have rep a which contains one project and rep b which contains a completely seperate project	03:17
cehteh	you can even configure the server side to refuse to override it	03:17
UltraBob	I set up a bare rep and push to it from a and then push to it from b	03:17
	it would refuse that by default?	03:17
cehteh	in theory you can push completely unrelated prokects into one repo	03:17
	git only knows about heads (branches and tags etc) it doesnt know about projects	03:18
	of course if both projects have a 'master' and you try to push these as master on the remote git will notice that these have nothing in common and refuse to do so	03:18
UltraBob	cehteh, I need to go read the documentation again, but this has been invaluable (I think, actually trying it will be the test) for getting me up and going quickly. Thanks!	03:18
cehteh	well keeping more than one project in a single repo is very uncommon	03:19
	(except kde .. but :P)	03:19
	→ blaenk joined	03:19
	← RurouniJones left	03:25
	→ RurouniJones joined	03:25
	→ bousquet joined	03:26
	→ sakyamuni joined	03:27
bousquet	Why does git think my css file is a single line? Multiple changes to it get recorded as 1 deletion, 1 addition (the whole file)	03:27
offby1	interesting!	03:27
	I dunno but will be curious to hear the solution	03:28
	← kwirky0 left	03:28
	← summerbug left	03:29
	→ sfwc joined	03:30
	→ orafu joined	03:33
	← cedric_ left	03:36
	→ brizly joined	03:36
	→ jony joined	03:37
	← Radar left	03:37
	→ Radar joined	03:37
	→ solydzajs joined	03:38
	← Radar left	03:38
	→ Radar joined	03:39
	← sfwc left	03:40
	← MachX left	03:44
bousquet	Anyone have any ideas on this? - Why does git think my css file is a single line? Multiple changes to it get recorded as 1 deletion, 1 addition (the whole file)	03:44
almostautomated\|	well, that was definitely an interesting read: http://userweb.kernel.org/~junio/200810-Chron.pdf	03:44
	→ reprore joined	03:45
	→ kmap joined	03:46
DrNick	bousquet: are you on a Mac, by chance?	03:46
bousquet	yeah	03:48
	is there a binary flag or something that's getting set accidenatally?	03:48
	→ repr_____ joined	03:50
	← repror___ left	03:50
	→ sfwc joined	03:56
	← memiux left	03:58
	→ cedricv joined	04:00
	→ srid joined	04:02
offby1	perhaps each change is modifying the line separators?	04:03
	← jony left	04:05
	← reprore left	04:09
bousquet	not sure, i haven't been working across win/osx	04:09
	← summerbug_ left	04:12
	→ k0001 joined	04:13
	← Mkop left	04:16
bousquet	DrNick: any ideas?	04:19
	← solydzajs left	04:21
	almostautomated\| → almostautomated	04:21
	← srid left	04:23
	joshthecoder → joshthecoder_afk	04:23
	→ scarabx joined	04:30
	→ bousquet_ joined	04:31
	→ jchris joined	04:32
	← bousquet left	04:32
	← travisjeffery left	04:32
	← schacon left	04:33
	→ schacon joined	04:35
	← bousquet_ left	04:35
	→ bousquet joined	04:35
	← leachim6 left	04:36
	← sfwc left	04:39
	→ sfwc joined	04:40
	→ jrmuizel joined	04:41
	→ kate21de joined	04:42
	← jrmuizel left	04:42
	← schacon left	04:43
cxreg	bousquet: sounds like some autocrlf issue	04:45
	→ schacon joined	04:46
	→ Flathead joined	04:46
	← jchris left	04:47
	→ jrmuizel joined	04:47
	→ priidu joined	04:48
bousquet	so how do I force it to be treated as a text file?	04:51
	← brizly left	04:53
	→ brizly joined	04:54
	← schacon left	04:55
	→ schacon joined	04:55
	→ solydzajs joined	04:57
cxreg	bousquet: is there a .gitattributes file anywhere?	04:57
bousquet	cxreg: not yet	04:57
cxreg	Mac OS used to use only \r as line separators, I wonder if that's what DrNick meant	04:58
	← ericindc_ left	04:58
cxreg	lacking \n, git would think it's just one line	04:58
	← schacon left	04:59
bousquet	ah, that could be it, i don't think this has been touched by windows. which a lot of the newline results mentioned	04:59
	→ schacon joined	05:00
	almostautomated is pretty sick of newlines and windows.	05:00
cxreg	not being a Mac user, I'm not sure how to tell you what's going on there..	05:01
	try something like this?	05:01
	perl -ne 'print if /\n/' my-style.css \| wc -l	05:01
	→ thiago_home joined	05:01
bousquet	0	05:02
DrNick	yeah, git assumes that line endings are LF and has code to convert DOS/Windows CRLF line endings, but no support for old-school Mac CR line endings	05:03
	← solydzajs left	05:05
	← scarabx left	05:05
	→ curvature joined	05:06
bousquet	interesting.	05:06
cxreg	if you dont mind fixing your newlines (and you should), try perl -pi -e 's/\r/\n/g' my-style.css	05:06
	→ ankit9 joined	05:07
	→ bdowning_ joined	05:08
	→ jony joined	05:08
bousquet	cxreg: ok, dope. that looks like it did the trick. git is showing each line now when i diff in the working dir	05:08
	→ davidfetter joined	05:09
bousquet	cxreg: do you guys usually run with config.global autocrlf = true?	05:10
	bdowning_ → bdowning	05:10
	→ smiskis joined	05:10
	→ ajonat joined	05:10
	→ solydzajs joined	05:11
cxreg	I've never used it	05:11
bousquet	i don't have any windows users on the team. not even sure why this newline issue came up	05:12
cxreg	likely someone either with an old mac, or an old editor, or something?	05:12
offby1	bousquet: I have that set when I use git on Windows, which isn't often	05:12
bousquet	yeah, maybe an old editor. this css was given to me from another dev, they probably use dreamweaver or something dumb	05:13
	→ roop joined	05:13
	→ agile joined	05:14
	→ hyperair joined	05:17
	→ jony1 joined	05:18
	← schacon left	05:20
	→ psankar joined	05:20
	→ MrAwesome joined	05:21
	→ jceb joined	05:21
	→ schacon joined	05:21
	→ sea-gull joined	05:21
	→ schacon_ joined	05:22
	→ schacon__ joined	05:23
	← schacon left	05:23
	← schacon_ left	05:23
	← jony left	05:27
	→ girishr joined	05:32
	→ seangrove joined	05:33
	→ robin joined	05:36
	← petercoulton left	05:37
	→ tchan joined	05:39
	← jrmuizel left	05:40
	→ NoOffenc joined	05:40
	← thiago_home left	05:42
	← repr_____ left	05:43
	← bousquet left	05:46
	← solydzajs left	05:47
	← seangrove left	05:47
	almostautomated wonders if there is an automatic way of dealing with line endings in an _organized_ mixed line ending project?	05:53
cehteh	with gitattributes and a lot of magic	05:54
	eh even easier .. rtfm :)	05:56
	→ sunoano joined	05:56
	← brizly left	05:56
almostautomated	hrrmmm auto crlf = input? I'm thinking more of a catch the dev error type thing	05:57
	→ seangrove joined	05:57
cehteh	how do you define a 'error'?	05:58
	← robin left	05:59
almostautomated	The dev clones the repo with cauto crlf = on then tries pushing it; causes lots of yucky modified msgs	06:00
	We have two main dirs	06:00
	← nighthwk1 left	06:00
	← NoOffenc left	06:00
almostautomated	One is from Scintilla and the other is PowerEditor	06:00
	→ nighthwk1 joined	06:00
almostautomated	PowerEditor is crlf and Scintilla isn't	06:01
	So, there isn't an issue if git is setup right; but if it isn't I'm just wondering if there'd be an 'automatic' way to catch it	06:02
cehteh	i am not very used to this crlf think .. but there is either some way to preserve it or you can use a pre-commit hook to check for all kinds of policies you set up	06:03
tansell	what do people use for code review with git?	06:04
almostautomated	tansell: not sure I was surfing google earlier to find something but got side-tracked	06:05
	← nighthwk1 left	06:05
	← bla left	06:06
almostautomated	cehteh: that's what I'm thinking... pre commit hook checking the target dir and verifying the endings	06:06
cehteh	look at gitattributes too ..'filter'	06:06
almostautomated	thanks, will do	06:06
	→ aposada joined	06:09
	← aposada left	06:10
	→ ohadlevy joined	06:10
	→ fridim_ joined	06:10
ohadlevy	hello, anyone willing to help with a strange git svn rebase issue?	06:11
almostautomated	go for it; not many will answer without a specific question	06:12
ohadlevy	I have a git-svn repo, I created a branch, worked on it, checkout master, merged the new branch, adn did dcomit, but now svn rebase fails	06:12
	complaining that some files do not exists in the index	06:13
	when doing a manual merge, its really strange, as those files were not modified at all	06:13
	in different directories etc	06:13
	← sakyamuni left	06:14
almostautomated	ohadlevy: I haven't used git svn to go back to dcommit yet...	06:16
ohadlevy	almostautomated: well.. dcommit works, just rebase	06:16
	→ ph^ joined	06:16
ohadlevy	another thing to mention, I linked a file to a sub dir of the repo, and worked on it in some part of the time, could that mess up things a bit?	06:17
almostautomated	hopefully someone else will know... just hang out someone will likely answer.	06:18
	← jceb left	06:20
	almostautomated → almostautomated\|	06:20
	→ godog joined	06:20
	← blaenk left	06:22
	→ bschindler joined	06:23
	→ lohrun joined	06:23
	→ sakyamuni joined	06:26
	→ johan-s joined	06:26
	→ stuffcorpse joined	06:27
	→ mbroeker joined	06:28
	→ savonarola joined	06:35
	→ dterei joined	06:35
	← savonarola left	06:35
	→ solar_ant joined	06:36
solar_ant	hi all	06:36
	is it possible that if I have a repository X	06:36
	that has two folder X/a and X/b	06:36
	that i can somehow clone only X/a	06:36
	← kate21de left	06:38
	← schacon__ left	06:40
	← ajonat left	06:44
	→ bla joined	06:47
sdiz	solar_ant: no	06:49
	→ altern joined	06:51
	→ dreiss_ joined	06:52
	← dreiss_ left	06:52
	← altern left	06:52
	← roop left	06:52
	→ galderz joined	06:54
	→ Freenode joined	06:55
	← UltraBob left	06:56
	Freenode → Guest65312	06:56
	→ Freenode_ joined	06:56
	← mbroeker left	06:59
	← hyperair left	07:00
	→ gilimanjaro joined	07:02
	→ j416 joined	07:02
	→ ciskje joined	07:03
	→ hyperair joined	07:05
	→ ben_h joined	07:05
	→ mbroeker joined	07:06
	→ Tuomas joined	07:07
	← ben_h left	07:09
	→ giallu joined	07:09
	→ Alinermedia joined	07:09
	← dterei left	07:09
	→ bmalee joined	07:12
	← bmalee left	07:12
	← gilimanjaro left	07:12
	→ bmalee joined	07:12
cxreg	http://lwn.net/Articles/131657/ is an amusing read	07:12
	← Guest65312 left	07:14
	→ khmarbaise joined	07:14
	→ Stravy joined	07:15
	← tbf left	07:16
	→ tbf joined	07:16
solar_ant	sdiz: I have seen such kind of clone on compiz	07:17
	wonder how they made it possible	07:17
wereHamster	why they?	07:19
	you cloned a repository that was already split up?	07:20
cxreg	git-subtree maybe?	07:21
	→ roop joined	07:22
	→ qhoxie joined	07:22
	→ jmesnil joined	07:23
solar_ant	wereHamster: yes	07:24
	I just google and found out about git submodule add :)	07:25
	ok now I have another challenge	07:26
	I am able to add *~ to .gitignore	07:26
	but i also want to ignore config.php in the current directory	07:27
	and not in other directories	07:27
	i tried ./config.php	07:27
wereHamster	config.php	07:27
solar_ant	wereHamster: wont that ignore like config.php in all the directories?	07:27
	→ madduck joined	07:28
wereHamster	no	07:28
	err, /config.php	07:28
	man gitignore, as always ;)	07:29
Gitbot	wereHamster: the gitignore manpage can be found at http://git.or.cz/man/gitignore	07:29
solar_ant	:)	07:29
	sorry	07:29
	thanks	07:29
	← JasonWoof left	07:36
	→ lucsky joined	07:38
eMBee	good afternoon	07:39
	can i tell git-status to list files that are ignored through gitignore?	07:39
	eMBee can't find such an option in the manpage	07:39
	dwmw2_gone → dwmw2	07:40
charon	eMBee: i think git ls-files can do that, through some combination of -i and possibly --exclude-standard	07:40
	→ marc_kdab joined	07:40
charon	eMBee: looks like git ls-files -io --exclude-standard does it	07:41
	→ jaeckel joined	07:41
eMBee	charon: which version of git do you have? that command gives a usage error in 1.6.2.4	07:42
	← jaeckel left	07:42
	→ jaeckel joined	07:43
	← agenteo left	07:43
eMBee	-i -o works though	07:45
charon	hmm, "git version 1.6.3.1.345.g53491"	07:46
	→ {SD} joined	07:46
	← Alinermedia left	07:46
eMBee	ah, seems the handling of single letter options has been improved :-)	07:47
	→ dvorak1 joined	07:47
	← madduck left	07:47
	→ madduck joined	07:47
	→ Sho_ joined	07:49
	→ psoo joined	07:51
	→ ceyusa joined	07:51
	→ the_zeno joined	07:51
	→ gilimanjaro joined	07:51
	← psoo left	07:52
	→ wshimmy joined	07:52
charon	ce8e88040 is the "culprit"	07:52
Gitbot	[git ce8e88040]: http://tinyurl.com/qfbrh4 -- parse-opt: migrate builtin-ls-files.	07:52
eMBee	:-)	07:52
	← Sho_ left	07:55
	→ cao joined	07:55
eMBee	is there a convenient way to import a whole snapshot of a source tree (like an exported tree). currently i checkout some commit (where i want to apply the changes from the snapshot) then remove everything and untar the snapshot in its place. then commit as is...	07:55
	→ psoo joined	07:55
Arrowmaster	i thought there was something you could just feed archives to	07:56
	→ Sho_ joined	07:56
eMBee	yeah, that would be nice...	07:56
charon	eMBee, Arrowmaster: there are some scripts in <HEAD:contrib/fast-import/>	07:57
Gitbot	charon: I'm sorry, there's no such object: HEAD:contrib/fast-import/.	07:57
charon	meh.	07:57
	<HEAD:contrib/fast-import/import-tars.perl>	07:57
Gitbot	[git import-tars.perl]: http://tinyurl.com/ohuov6 [blob]	07:57
	eMBee looks, thanks	08:00
	jony1 → jony	08:04
	→ rshann_ joined	08:04
	→ GodEater joined	08:05
	← tbf left	08:05
	← wshimmy left	08:06
eMBee	ah, it imports to a seperate branch, good	08:06
	→ shruggar joined	08:07
	→ tbf joined	08:07
	← roop left	08:09
	→ lbt joined	08:10
	→ Ademan joined	08:10
klmann	hi guys. is there a script/command that creates the pull request mails?	08:14
eMBee	hmmm, it creates a seperate branch with no ancestor, not good. unless i change my workflow	08:14
klmann	or is it done manually every time? does not look like that i.e. on the lkml	08:15
charon	eMBee: you can always graft it ;)	08:15
	klmann: man git-request-pull	08:15
Gitbot	klmann: the git-request-pull manpage can be found at http://git.or.cz/man/git-request-pull	08:15
klmann	thanks	08:16
sea-gull	can anybody explain why after checkout 'w' flag in the file mode diseappear?	08:17
	→ gaveen joined	08:18
	→ dwave joined	08:18
eMBee	charon: can you explain how to graft?	08:20
charon	eMBee: suppose $root is from your fast-import experiment, and $last is the last commit along the same ancestry in your "target" history	08:21
eMBee	ok	08:21
Arrowmaster	http://git-scm.org/gitwiki/GraftPoint	08:21
charon	eMBee: then you can do 'echo $root $last >> .git/info/grafts'	08:21
	which effectively tells git to pretend the parents of $root were $last	08:21
eMBee	to pretend, ah, ok	08:21
charon	then verify that the result is correct with gitk or similar, and then finally make them "real" by applying filter-branch	08:22
eMBee	ohhh	08:22
	→ dwave_ joined	08:23
	→ ceyusa_ joined	08:24
	eMBee was going to apply what he learned here: http://colabti.org/irclogger/irclogger_log/git?date=2008-12-19#l459 (esp: http://colabti.org/irclogger/irclogger_log/git?date=2008-12-19#l497 ) to set the commits. but graft and filter-branch seems much easier	08:26
	→ gitte joined	08:27
	← ceyusa left	08:29
	→ tono joined	08:31
	→ targion joined	08:33
	← rshann_ left	08:36
	← Sho_ left	08:37
	→ up_the_irons2 joined	08:38
	← dwave left	08:38
	← psoo left	08:38
	→ Sho_ joined	08:38
	→ doener joined	08:38
	← Sho_ left	08:39
	→ rshann_ joined	08:39
	→ Sho_ joined	08:39
	← Radar left	08:39
	→ Lgb joined	08:39
	→ Radar joined	08:42
eMBee	i em getting: error: bad graft data: ...	08:44
	i just put two commitids on a line	08:44
	← Radar left	08:44
hyperair	where did you see that error exactly?	08:44
	→ psoo joined	08:44
Ilari	eMBee: Do those two commit IDs exist? Are they complete (40 hex chars, with lowercase a-f)?	08:45
	→ naeu joined	08:45
eMBee	yes, i picked them with git show ...	08:46
	→ roop joined	08:47
eMBee	oh, hmm, maybe i picked the wrong thing	08:47
	hmm, no	08:47
	git show only lists the commit id	08:47
	→ tvw joined	08:48
	dwave_ → asksol	08:50
	→ niki joined	08:51
	← disjunctor left	08:51
	→ bx2 joined	08:58
	→ wshimmy joined	08:58
	→ hutingung joined	09:00
hutingung	cool mibbit	09:00
	any1 can tel me which i should use GIT or SVN?	09:01
	reason?	09:01
	任何人？	09:02
Ilari	eMBee: Also, another reason for that might be that you created cyclic reference?	09:02
hutingung	任何人在这里？	09:02
Ilari	EPARSE	09:02
hutingung	测试工具的transalation	09:02
	意思吗？	09:02
	← eletuchy left	09:02
hutingung	无聊频道	09:03
	我退出	09:03
	← hutingung left	09:03
	→ eletuchy joined	09:03
	← Flathead left	09:03
	→ un\|ogic joined	09:04
eMBee	Ilari: hmm, i have A--B--C, and then ran import-tars which created a commit X, and then set X C into grafts. X has no children, so nothing could cause a cycle here	09:05
	X has no other parents either	09:06
smiskis	What is GIT's bigest advantages except being able to have local repositories?	09:07
	← rshann_ left	09:07
	← markelikalderon left	09:07
	→ disjunctor joined	09:07
	→ intripoon joined	09:08
wereHamster	smiskis: being able to have local repositories sums pretty much all the other advantages up	09:08
drizzd	wereHamster: except better merge support	09:09
wereHamster	local repos allow you to edit/rearrange commits before you push those.	09:09
drizzd	a centralized system could do that as well	09:09
wereHamster	ather advantages are: stash, bisect	09:09
eMBee	speed, ability to fix history (git rebase -i is really great), efficiently manage many branches, powerful tools to handle complex commits, bisect is very nice for debugging too	09:10
smiskis	drizzd: Why is merging better.. than lets say SVN?	09:10
drizzd	smiskis: I don't know. I've never used SVN. Everybody just tells me so.	09:11
eMBee	up to 1.5 svn did not track merges at all	09:11
	which means you had to manyally take care of what you wanted to merge	09:11
smiskis	i must say git suffers from unintutive CLI.. just like older SCMs	09:12
drizzd	smiskis: I suppose a major reason is speed itself. Due to the slowness of centralized repositories, many merge strategies are infeasible.	09:12
	← ohadlevy left	09:12
smiskis	okey.. thanx guys	09:12
drizzd	smiskis: improvements are always welcome	09:12
	→ mstrl joined	09:13
eMBee	git's commandline interface it not less intuitive than that of svn, really. it may just be a bit overwhelming if you look at all the commands available. but is easy if you follow a guide which tells you the most common commands and ignore all others (which you'll rarely, if ever, need)	09:14
	→ Octalot joined	09:14
	→ emdete joined	09:14
smiskis	eMBee.. i meant its as overwhelming as older ones.. like SVN	09:15
emdete	when i use git-apply, arent those changes committed? how can i achieve that	09:15
eMBee	smiskis: well, you were comparing to svn at first, so...	09:15
drizzd	emdete: git am	09:15
smiskis	eMBee: could you advice me on such make-it-feel-easy guide?	09:15
	← ceyusa_ left	09:16
emdete	drizzd: thnks...	09:16
drizzd	emdete: git apply doesn't expect commit info	09:16
eMBee	anyone else have a good link for smiskis off hand?	09:16
	→ kraymer joined	09:16
	eMBee would need to look	09:16
	→ KingBug_ joined	09:16
	→ ferdy joined	09:17
drizzd	smiskis: I think the manual (not the manpages) is quite easy to read	09:17
Ademan	if I want to discard the changes I've made on a given file rather than resolving a merge, how would I do that?\	09:19
eMBee	generally, i think it is not relevant to compare the git commandline to gui tools. and as far as the commandline goes git fares well even compared to other modern scms	09:19
	and there are git gui tools (most notably gitk) which is really helpful in understanding the branches and history in your repository.	09:21
drizzd	Ademan: git checkout :2:file	09:21
eMBee	of all scm gui tools i have seen, gitk looks the best (comparing screenhots here)	09:21
drizzd	or is it :3:file?	09:21
	Ademan: it's in the 'how to merge' section of the manual	09:22
Ademan	alright thanks drizzd	09:22
	→ aspotashev joined	09:22
drizzd	ok, 1 is base, 2 is HEAD, and 3 is MERGE_HEAD	09:23
	drizzd has to remember that	09:23
	← {SD} left	09:24
Arrowmaster	isnt there a shortcut for that stuff now?	09:24
	→ d0k joined	09:24
Arrowmaster	like --ours/--theirs	09:25
doener	git checkout --theirs/--ours <file>	09:25
	Ademan: did that "git checkout :3:file" work at all? AFAIK, it should not	09:26
	→ ceyusa_ joined	09:26
doener	and it does not :-)	09:26
	← plasticine left	09:31
Ademan	doener: we tried the "nice" way with mergetool	09:31
	no luck at present though	09:31
	even after resolving the merge he cant commit, it says he still has a pending merge	09:32
eMBee	could grafts error be because there is no common ancestor between the two commits?	09:33
hyperair	no, that shouldn't be an issue	09:33
	you're changing the ancestor by grafting the tree	09:33
jaeckel	smiskis: try to have a look at the 2 git cheat sheets	09:33
	→ legnaleurc joined	09:34
eMBee	right	09:34
jaeckel	there are all often used commands listed and also the possible flows	09:34
	eMBee wonders what's wrong then	09:34
	← dreiss left	09:34
jaeckel	http://git.or.cz/gitwiki/GitCheatSheet http://jan-krueger.net/development/git-cheat-sheet-extended-edition	09:36
	← Stravy left	09:36
	← sakyamuni left	09:36
	→ Stravy joined	09:37
	← kraymer left	09:40
	→ plasticine joined	09:40
	→ struberg joined	09:42
Ilari	eMBee: Some nasty "invisible gremlins" on that line? Try to dump it using 'od -t x1'?	09:43
	eMBee: For instance, its surprisingly easy to type U+00A0 instead of U+0020.	09:43
	← jksz left	09:45
	→ sadleder joined	09:48
	← cedricv left	09:48
	← mithro left	09:48
eMBee	no, possibly the fact that there was a tag and a branch witrh the same name pointing to different commits, causing confusion	09:48
	← bschindler left	09:49
	← ceyusa_ left	09:49
eMBee	now i only get an error if the commits are in the wrong order (even though that should work too	09:49
	but i still do not see a 'new' history	09:50
	git log of the branch with the grafted commit does not show any new parent, nor does gitk	09:50
	→ mithro joined	09:55
mithro	can a branch have a comment?	09:56
wereHamster	mithro: no	09:56
	a branch is just a pointer to a commit	09:57
	→ knitt1 joined	09:58
	→ ekristen joined	10:02
	← ekristen left	10:02
	→ aqmaiu joined	10:02
	→ HG` joined	10:04
	← fridim_ left	10:05
	← targion left	10:14
	← tvw left	10:14
	← dvorak1 left	10:16
	← bx2 left	10:17
	→ cousine_ joined	10:19
eMBee	is there any documentation for the format in the grafts file? i don't get it	10:19
	the parent comes first, and then the child?	10:19
	cousine_ → cousine	10:19
doener	eMBee: <commit> <parent1> <parent2>...	10:22
eMBee	ok	10:22
doener	i.e. the same format you get from --parents (rev-list, log, show, ...)	10:23
	← emdete left	10:25
eMBee	hmm	10:25
	eMBee has no luck making it work	10:25
eMBee	ohh, restarting gitk help	10:26
	with grafts do i replace parents or add the?	10:27
	them	10:27
	→ aziz joined	10:29
	→ HG`` joined	10:32
Grum	eMBee: replace	10:33
	so for a merge you need at least 2 parents	10:33
	and for a root commit 0	10:34
	→ petermarks joined	10:34
eMBee	ok, thanks. seems gitk is a little sensitive when reloading, restarting makes errors go away	10:37
	it becomes much clearer when you can see the effects of the graft actually working	10:37
petermarks	Newbie question: I want to keep track of which version of our code is currently in staging in the remote repository my team shares. I was thinking about using a branch called "staging" and pushing to it just after we deploy. Is this a sensible approach or is there a better way? Also, If we rollback the version in staging, how do I reset the staging branch to an earlier version?	10:39
	→ bschindler joined	10:40
	← HG` left	10:41
	→ Hendu joined	10:42
	→ DavidKlein joined	10:42
drizzd	petermarks: you can always overwrite a branch using git push origin +staging	10:43
petermarks	drizzd: cool, so the + syntax completely overwrites a remote branch?	10:45
drizzd	yes, also push -f, but be careful with that. You don't want to rewrite a branch others are working on	10:47
petermarks	what about a slightly different approach: if I have a tag in my remote repository, can I set a branch in the remote repository to the tagged commit?	10:48
	→ codebrulee joined	10:50
	← sfwc left	10:52
	→ sfwc joined	10:52
	→ parasti joined	10:55
	→ cxvii joined	10:55
drizzd	petermarks: You seem to think that remote repositories are anything special	10:56
	they are not	10:56
	every remote is a git repository just as much as your local repo	10:56
	the concept of remotes is simply there to ease synchonization between repositories	10:57
	anything you can do to your local repo you can do to your remote by pushing (synchronizing)	10:57
	← eletuchy left	11:00
	→ ntoll joined	11:00
	→ mitkok joined	11:01
	← up_the_irons2 left	11:02
	→ up_the_irons2 joined	11:06
petermarks	drizzd: what seems different about a remote repository to a local one is that I can't use most of the git commands on it directly (please correct me if I'm wrong). You seem to have to do things locally first then push them to a remote. However, branches seem to be a repository specific thing so doing something to a remote branch is a confusing concept to me.	11:09
	If I understand correctly, to do what I suggested I have to create a tag locally, push it to the remote. then push my local tag to the remote branch I want to set. Is that correct?	11:10
	← codebrulee left	11:11
	→ adamtanner joined	11:14
	→ adamtanner_ joined	11:16
	← adamtanner_ left	11:17
	→ runasand joined	11:17
	← sgronblo left	11:17
	runasand → ninjarobot	11:18
	← mitkok left	11:19
	← jony left	11:19
	→ markelikalderon joined	11:21
	← ankit9 left	11:28
	KingBug_ → KingBug	11:29
	← cxvii left	11:33
	← adamtanner left	11:33
	→ Moult_ joined	11:34
	← markelikalderon left	11:34
	← giallu left	11:35
	→ giallu joined	11:36
	→ Modius_ joined	11:40
	← smiskis left	11:42
	→ tcurdt joined	11:43
	← ciskje left	11:43
	← psankar left	11:44
tcurdt	hey folks, git-svn question: right now I am tracking only trunk ... how do I make git-svn aware of the fact that a branch was added in svn? so I can also commit to that branch via git-svn	11:44
	just adding a svn-remote branch in the config?	11:45
wereHamster	tcurdt: add a new fetch refspec	11:46
	→ Modius__ joined	11:47
	→ bx2 joined	11:51
	→ zomg joined	11:52
	→ Nazcafan joined	11:52
Nazcafan	hello	11:52
	How can I track the commits that impacted a particular file ? (knowing that in my case the file is a driver)	11:53
ph^	I'm trying to make a git hook on our shared repository to close redmine tickets, so I figured I should put a hook on the server, but I don't know which hook to edit?	11:54
	post-update? post-receive? I need one with revisions	11:54
	→ ericindc joined	11:54
gitte	Nazcafan: "git log -- <file>"	11:55
Nazcafan	ah thanks	11:55
	← ericindc left	11:55
wereHamster	koenb: ah, that makes sense	11:55
drizzd	petermarks: that's one possibility. or you simply have a local branch pointing to the commit of your choice and push that	11:55
gitte	ph^: the documentation suggests post-receive. Did you read Documentation/githooks.txt?	11:57
zomg	Hi all.. I did a stupid mistake and I need to revert some lines of files back to the way they were before, preferably without killing the new stuff I added	11:57
	I'm quite new to git and not sure how I should approach this	11:58
gitte	zomg: Do you have any uncommitted changes?	11:58
zomg	Nothing that I don't mind losing	11:58
	=)	11:58
gitte	zomg: either stash it or commit it.	11:58
	zomg: that is a good practice in any case, if you want to do something slightly adventurous, start with a clean working directory.	11:59
zomg	Oh and there are few commits of these broken files	11:59
gitte	zomg: then you could use a combination of git checkout and git gui to revert the changes.	11:59
	zomg: in detail: "git checkout <known-good-commit> -- <file>" to get the old version back.	12:00
	zomg: then use "git gui" to unstage all, and then go through the modified files hunk by hunk (or even line by line), and only stage those reverts you actually want.	12:00
	zomg: then commit and stash.	12:01
	zomg: to clean up your branch, you could use rebase -i (but only if you haven't pushed in the meantime).	12:01
zomg	Thanks, that should help me get started =)	12:01
ph^	gitte: thanks, I've tried to read, but I get confused with some of the terminology	12:04
	← Modius_ left	12:04
	← priidu left	12:05
	→ MisterN joined	12:08
	→ codebrulee joined	12:10
	→ Modius joined	12:11
petermarks	drizzd: thanks for your help	12:12
	→ hoohah joined	12:13
hoohah	hello. I'm unsure about a certain aspect of branch merging. Once I perform the merge, will all commit history be retained in addendum to the master, or it be considered as one (massive) commit?	12:14
gitte	hoohah: both.	12:15
	hoohah: it will add a merge commit which has two parents.	12:15
hoohah	gitte: no way to specify one or the other?	12:15
	i'm after the latter effect only	12:15
gitte	hoohah: there is no merge without a merge commit, except when there is nothing to merge.	12:15
	hoohah: probably you are looking for "git merge --squash"	12:15
	→ mitkok joined	12:17
	→ uau joined	12:21
	← Modius__ left	12:21
	→ mobhawk joined	12:23
	→ tfoo joined	12:24
	→ memiux joined	12:24
	← gaveen left	12:25
	← tcurdt left	12:25
	→ gaveen joined	12:25
hoohah	gitte: ah thanks. So --squash basically creates one commit for the whole merge of the branch.. and I guess once that branch is removed, no history of commits can ever be identified with the affected files from branch	12:26
gitte	hoohah: correct.	12:26
	→ name joined	12:27
hoohah	do you have time for me to bother you with one last q?	12:28
	regarding git implementation in my $HOME to manage dotfiles, say I worked with multiple distros sharing the same home path during the day. Would it be advisable to implement git in a way such that master branch in git repo in ~/ contains all similar options, with branches created for the individual distros?	12:29
	and perhaps, somewhere in the distros rc.local, have git checkout respective branch?	12:29
	→ cmarcelo joined	12:33
	← cousine left	12:33
gitte	hoohah: dunno. I'd try to have the distro-specific stuff in the files themselves, but that's just a personal matter. I do not track my $HOME with Git.	12:34
hoohah	gitte: yeah, for a while I was implementing cases left and right, but found that I needed logic beyond shell scripts (e.g. for apps like rtorrent whose version differs across distros)	12:36
	thereby requiring different configuration options	12:37
	i guess i'll just stick with what I'm doing currently	12:37
gitte	hoohah: so you answered your question yourself ;-)	12:37
hoohah	gitte: yeah hehe it seems i always end up doing that	12:37
	gitte: at least with an attentive ear, i can avoid talking to myself :)	12:38
	← girishr left	12:39
	→ angerman joined	12:39
	← memiux left	12:41
	→ LiamH joined	12:41
	← bobmcw left	12:45
	← jmesnil left	12:46
	→ jmesnil joined	12:46
	→ lorandi joined	12:47
	→ tcurdt joined	12:47
	← tcurdt left	12:48
	→ boto joined	12:50
	→ leanucci joined	12:51
	← solar_ant left	12:51
eMBee	yes, lack of versioned build dependencies is really an issue	12:52
	ooops	12:52
	→ hpoydar joined	12:53
	→ ShadeHawk joined	12:54
ShadeHawk	According to Google Summer of Code timeline the "community bonding" period ended, and students should have started working on their projects.	12:55
	→ ben_h joined	12:55
	→ scarabx joined	12:56
ShadeHawk	I wonder what is happening with our GSoC students (they didn't announce and introduce themselves on git mailing list), and with "foreign" GSoC students that work on projects related to git: hg-git from Mercurial, git# from Mono, Build Service from OpenSUSE, GitFS from Plan9, PyVCAL from Python	12:56
	→ gh34 joined	12:56
	wshimmy is quite busily hacking away ;)	12:57
	→ rshann_ joined	12:57
ninjarobot	ShadeHawk: you should poke the students	12:57
	ninjarobot → runasand	12:57
runasand	wonder why I had that nick :]	12:57
	← tfoo left	12:58
frsk	You probably felt like a ninja when choosing it	12:58
runasand	anyways, some students might be busy with exams these days and a friendly poke wouldn't hurt	12:58
	frsk: I always feel like a ninja ;)	12:58
frsk	ShadeHawk: You were asking around for the students about a week ago too, right? (Or was that someone else?)	12:59
	→ qrush joined	12:59
gitte	wshimmy: what are you working on?	12:59
frsk	I recall gitte talking about the progress a few nights ago	13:00
runasand	hm.. there's only two GSoC students	13:00
wshimmy	gitte: at the moment, whipping some diff engine code into shape	13:00
	← struberg left	13:01
gitte	runasand: there are more. With 150 orgs I'd estimate many more than two.	13:01
	wshimmy: oh?	13:01
wshimmy	exactly 1000, IIRC	13:02
runasand	gitte: only two working for git, I mean.	13:02
gitte	wshimmy: can I see some?	13:02
	wshimmy: I am working on a diff engine for JGit as much as my time allows (which is not very much).	13:02
wshimmy	gitte: sure. atm im just trying to adapt/extend some of the code thats used in monodevelop for generating hunks	13:03
ShadeHawk	wshimmy: what GSoC project (and what GSoC org) are you working on?	13:03
wshimmy	gitte: but its a bit old, and since git gives diffing much more exercise than the occasional svn addin, it needs a few more hooks and things	13:03
	ShadeHawk: im that evil mono student	13:03
	→ _rane joined	13:04
	→ BooK joined	13:04
	→ webcoder joined	13:04
_rane	hey, I have 'spec/spec.opts' in .git/info/exclude, but git status keeps saying modified: spec/spec.opts -- any ideas?	13:04
gitte	wshimmy: you are actually the first non-evil one for Mono. You talk to us, at least.	13:04
	wshimmy: that's more than can be said of most of Mono's mentors, too.	13:05
ShadeHawk	_rane: ignored is about files which are not tracked; if it is tracked (known to git) then ignore rules do not matter	13:05
	→ freelook1 joined	13:05
wshimmy	gitte: thanks - i like to think of myself as non-evil the majority of the time :)	13:06
_rane	so how would I get that file ignored locally?	13:06
ShadeHawk	Nice... from foreign GSoC students we have touched base (that I know of) with Build Service from OpenSUSE (on its IRC channel), GitFS from Plan (here); ........... PyVCAL somehow did vanish from list of Python projects for GSoC2009	13:07
	→ ProperNoun joined	13:07
ShadeHawk	I don't know about hg-git project (Git backend to Mercurial frontend), ... and of course our student ...	13:08
gitte	wshimmy: nobody thinks of themselves as evil. Not even certain warlords putting part of the planet into misery while profiting from it.	13:08
Pieter	like Nero	13:08
ShadeHawk	well, I know that with graphical UI there is problem with bikeshedding on git mailing list...	13:08
gitte	ShadeHawk: search the channel log for "sea-gull"	13:09
	→ dodo_the_last joined	13:09
gitte	ShadeHawk: exactly. The UI guy also said beforehand that he'd be busy for one week these days.	13:09
_rane	ShadeHawk: .gitignore works as expected, but because it's tracked in the repo it doesn't produce the wanted results	13:10
	it doesn't seem to make any difference whatever I have in .git/info/exclude	13:10
	← wmoxam left	13:11
zomg	gitte: after a ton of confusion and failure, I managed to fix the problem, create a patch and apply it. Thanks for pointing me to the right direction =)	13:11
ShadeHawk	Thanks for the info gitte	13:11
gitte	zomg: good that you have a solution!	13:12
	ShadeHawk: yw	13:12
zomg	with svn I would've been done an hour ago tho	13:13
	;>	13:13
	but I guess I'll just have to learn	13:13
	hehe	13:13
ShadeHawk	gitte: but sea-gull of GitFS for Plan9 I have met on #git	13:13
	→ memiux joined	13:14
gitte	ShadeHawk: oh, right.	13:14
	ShadeHawk: As for Hg-git, no, we haven't met (yet). According to SRabbelier, the mentor is intimate with the Hg-git project, so I guess we will be able to get into touch.	13:14
	→ krh joined	13:14
andres	How would you estimate the chances that a patch to add builtin support for context diffs is accepted (git and libxdiff)?	13:15
	→ boto_ joined	13:15
ShadeHawk	_rane: hmmm... .gitignore and .git/info/exclude work the same way, so unless you have some conflict (one file excludes, one file unignores) it should work or not work, and not that one works while other doesn't	13:18
	_rane: ignore rules are about unknown (untracked) files _only_	13:18
davidfetter	hello	13:18
	→ kukks joined	13:19
	→ cemerick joined	13:19
davidfetter	i've got a git-cvsimported repo, and would like to get to a particular CVS tag. how to do this?	13:19
	→ jrmuizel joined	13:20
andres	davidfetter: git checkout tags/tagname like normally?	13:21
ShadeHawk	davidfetter: check out "git tag -l" and/or "git branch -a" output	13:21
	davidfetter: where it is	13:21
davidfetter	andres, ShadeHawk, thanks :)	13:21
	→ kwatz joined	13:21
davidfetter	andres, this is in aid of testing the git-to-cvs migration :)	13:21
andres	davidfetter: I doubt that will happen that fast ;-)	13:22
ShadeHawk	davidfetter: and then as andres said: "git checkout <tag>" (it will detach HEAD, c.f. situation with sticky tags in CVS), or "git checkout -b <newbranch> <tag>"	13:22
	→ loincloth joined	13:22
ShadeHawk	davidfetter: git to CVS? not CVS to git?	13:22
davidfetter	er, cvs-to-git	13:23
	davidfetter has not yet had any coffee	13:23
davidfetter	and it's 0623 here in oakland :P	13:23
	→ bdiego joined	13:23
andres	davidfetter: now you just have to convice Davide or somebody else to add context diff support to libxdiff ;-)	13:24
davidfetter	do tell	13:24
	who's Davide and how do i reach him?	13:24
andres	The author of libxdiff	13:24
davidfetter	and is that what comes with git?	13:24
andres	yes.	13:26
ShadeHawk	more or less (git uses modified, enhanced and slimed down version of LibXDiff in xdiff/ directory)	13:26
	→ shana joined	13:27
	← naeu left	13:28
	→ peteinlux joined	13:29
	← scarabx left	13:29
	→ Modius_ joined	13:29
	← khmarbaise left	13:30
	→ pete__ joined	13:30
	→ solar_ant joined	13:30
	← boto left	13:30
	→ ryanew joined	13:31
	← ryanew left	13:32
	→ ryanew joined	13:32
	← peteinlux left	13:32
	→ wmoxam joined	13:32
ShadeHawk	wshimmy: By the way, do you know if there was any usable code produced by previous git# (managed git) projects at previous Google Summers of Code?	13:33
wshimmy	heh	13:33
	→ Arafangion joined	13:33
Arafangion	Hey, want's the easiest way to email a git repo?	13:33
	compress a bare repo and email?	13:33
	← ryanew left	13:33
	hyperair has a horrified look	13:33
hyperair	why would you do that	13:33
	github!	13:33
	or something	13:34
qrush	Arafangion: yep	13:34
wshimmy	ShadeHawk: as far as im aware, thats a negative	13:34
ph^	what is body as opposed to subject in a commit message?	13:34
Ilari	Arafangion: 'git bundle create foo.bundle --all'?	13:34
Arafangion	Ilari: Thanks, that's the example I was hoping for!	13:34
	Ilari: What's the inverse?	13:34
wshimmy	ShadeHawk: there was a student in 2007 that disappeared before the start of coding (got a job or something i think), so produced nothing; and last year 1 of 2 awol'd and the other one wrote garbage	13:35
Ilari	Arafangion: I think bundles support being target of fetch-type operations.	13:35
Arafangion	Ilari: Nice.	13:35
MisterN	wshimmy: "awol"? what's that?	13:35
	← dodo_the_last left	13:36
Arafangion	MisterN: "Went away, without contact".	13:36
wshimmy	absent without leave, sorry	13:36
Arafangion	Usually without notice.	13:36
MisterN	oh	13:36
	← Modius_ left	13:36
ShadeHawk	Arafangion: man git-bundle... or yes, compress and send	13:37
Gitbot	Arafangion: the git-bundle manpage can be found at http://git.or.cz/man/git-bundle	13:37
MisterN	GSOC doesn't seem to be very productive	13:37
wshimmy	MisterN: in general i think the majority of projects are fairly productive	13:37
	→ khmarbaise joined	13:38
wshimmy	it just depends	13:38
Arafangion	ShadeHawk: I think I might go the bundle approach.	13:38
ShadeHawk	MisterN: depending on project...	13:38
MisterN	wshimmy: dunno i've seen some projects have more hassle than benefit	13:38
	and git# adds to the list	13:38
	← aziz left	13:38
	→ Modius_ joined	13:39
ShadeHawk	MisterN: from 6 projects Git had last year, 3 were complete success (builtin merge, fetch in Eclipse, git statistics), 2 produced code but are not included in Git (gitweb caching which is used by git.kernel.org, and unfinished but worked on git-sequencer), with only one fail (GitTorrent Protocol)	13:40
gitte	MisterN: it would be better if you encouraged the student, rather than discouraged him.	13:40
	MisterN: thank you for your cooperation.	13:40
	← pete__ left	13:41
MisterN	gitte: huh? there's a SOC student present? and i think being sceptical is still allowed :P	13:41
wshimmy	skeptical is fine. but IMO its no hassle, and only potential benefit, so it works for me ;)	13:42
gitte	MisterN: you can be sceptical all you want, but if you discourage some GSoC student openly, you will get reprimanded by me.	13:42
	→ dodo_the_last joined	13:42
gitte	MisterN: so, stop it.	13:42
MisterN	gitte: i didn't want to discourage anybody	13:43
	→ Modius__ joined	13:43
	← Modius_ left	13:43
	← shana left	13:45
	→ shana joined	13:47
	→ rudi_s joined	13:47
ShadeHawk	By the way, that reminds me that SoC*Projects git wiki pages are not updated...	13:47
	← gaveen left	13:48
	→ gaveen joined	13:49
	→ gberdyshev_ joined	13:49
	← jrmuizel left	13:52
	← Modius left	13:52
gitte	ShadeHawk: my Git time budget is way negative right now.	13:53
	→ nighthwk1 joined	13:54
	→ komplex joined	13:55
	← rudi_s left	13:55
Arafangion	gitte: How is that possible?	13:56
gitte	Arafangion: very easy. You work too much on Git during the time you had other work to do.	13:57
Arafangion	gitte: Doesn't that affect the /other/ projects?	13:58
	→ priidu joined	13:59
	→ dvorak1 joined	14:00
	→ jony joined	14:01
	jony → jony\|away	14:01
	← hoohah left	14:02
	← gberdyshev left	14:03
	→ squentin_ joined	14:09
	→ tango_ joined	14:11
	→ suman_g joined	14:11
	→ drizzd_ joined	14:12
	→ Modius_ joined	14:12
	→ aziz joined	14:12
	→ alezandro joined	14:13
	← ben_h left	14:13
	← Modius__ left	14:14
	→ girishr joined	14:16
	← hyperair left	14:20
	→ hyperair joined	14:20
	← bx2 left	14:21
	→ jchris joined	14:21
	→ mtvee joined	14:22
	← priidu left	14:22
	← plasticine left	14:23
	← rshann_ left	14:23
	almostautomated\| → almostautomated	14:24
	→ priidu joined	14:24
	← mstrl left	14:24
	← Nazcafan left	14:26
	← Lgb left	14:26
	→ flazz joined	14:27
	→ Lgb joined	14:27
	← solar_ant left	14:27
	→ jrmuizel joined	14:30
	→ Nazcafan joined	14:32
	→ killerchicken_ joined	14:33
	← parasti left	14:34
killerchicken_	hey, I have a commit that fixes stuff from two different earlier commits. Is there a shorthand for doing a split of that commit, and then squashing the two parts into the appropriate commits?	14:34
	like "squash this commit, but only for those lines that affect the earlier commit"?	14:35
	I guess there isn't, but maybe I'm surprised :-)	14:35
andres	killerchicken_: you could split the current commit with add --interactive and then do a rebase --interactive and squash the commits belonging to each other	14:36
killerchicken_	andres: well, the "current" commit is not the topmost commit here. I'd just do an interactive rebase to split the commit	14:37
andres	Well, thats nearly the same ;-)	14:37
killerchicken_	that wouldn't be a problem at all, I was more wondering if there was something more specialized for the situation of fixing part of an earlier commit	14:38
	would make this one faster	14:38
	→ piyo joined	14:38
charon	killerchicken_: isn't the "only those lines that affect the earlier commit" actually AI complete?	14:39
killerchicken_	hm?	14:39
charon	i mean, of course sometimes the overlaps will be textual, but sometimes they're just semantic fixes in other areas of the code, in which case a computer cannot decide which part affects what	14:39
	← DavidKlein left	14:40
killerchicken_	right. For me, they're all overlaps in line numbers	14:40
	→ summerbug joined	14:41
Arafangion	Hey, I've asked someone to compress and email me their bare repo, and I've got it now but it has no HEAD, though a quick glance tells me it obviously has the commits.	14:42
	How do I fix this? (HEAD points to refs: refs/heads/master	14:42
	→ RedShift joined	14:42
RedShift	hi	14:42
Arafangion	But it doesn't exist) - what is a liekly mistake? They used 7zip to compress it.	14:42
	← Stravy left	14:43
	→ rudi_s joined	14:43
ShadeHawk	Arafangion: you can use git-symbolic-ref to change where HEAD points to	14:43
wshimmy	Arafangion: if its a bare repo you don't have a working directory	14:43
RedShift	I've got a question, so git is built for branching and merging, but what if I want to 'tag' a certain version of all files as a release?	14:43
	→ altrux joined	14:43
RedShift	like in CVS you have 'tags'	14:43
Arafangion	wshimmy: But surely it has a refs/master	14:43
PerlJam	RedShift: git has tags too	14:43
Arafangion	(Which it doesn't)	14:43
	← d0k left	14:44
RedShift	I'm reading http://www-cs-students.stanford.edu/~blynn/gitmagic/ but it doesn't mention tagging as far as I've read	14:44
ShadeHawk	Arafangion: if it is bare repository, then "current branch" doesn't make sense (no branch is checked out), so HEAD might not exist	14:44
ToxicFrog	RedShift: man git-tag	14:44
charon	wshimmy, Arafangion: even in bare repos, HEAD is useful as it indicates (to some extent, the protocol isn't good enough to handle it all) the default branch that new clones should check out	14:44
Gitbot	RedShift: the git-tag manpage can be found at http://git.or.cz/man/git-tag	14:44
Arafangion	wshimmy: I tried to clone it, but it fails.	14:44
RedShift	thanks	14:44
ShadeHawk	Arafangion: refs/master instead of refs/heads/master is likely a mistake (e.g. in push refspec)	14:45
wshimmy	it should definitely have the refs	14:45
Arafangion	charon, wshimmy: There are /no/ branches, though it obviously has the contents.	14:45
ShadeHawk	Arafangion: dod you check out output of "git branch -a" and/or refs/packed-refs file?	14:45
ToxicFrog	RedShift: you can create light named tags (which are basically just alternate names for given commits), as well as heavier, cryptographically signed tags with message content.	14:45
Arafangion	It was a git --bare init; then from the original repo, it was a git push /path/to/bare/repo master	14:46
ToxicFrog	The man page has all the gory details.	14:46
Arafangion	ShadeHawk: The refs directory is entirely empty.	14:46
	← ph^ left	14:47
	← khmarbaise left	14:47
ShadeHawk	Arafangion: push, not fetch? Because fetch without refspec saves fetched head only in FETCH_HEAD	14:47
	→ khmarbaise joined	14:47
Arafangion	Actually, the utterly freaky thing is... All the directories are empty! (Wierd, I'm blaming 7zip!)	14:47
	ShadeHawk: Push, to populate the bare repo.	14:48
charon	Arafangion: umm, how do you know it has any contents when it's empty? :)	14:48
Arafangion	charon: I did an ls, and saw all the sha1's. :)	14:48
	charon: And missed that I shouldn't have seen them... As they should've been in objects.	14:48
charon	maybe you unpacked it "flat" to a single directory? the objects should all be in objects/xx/sha1	14:48
Arafangion	charon: Looks that way, in which case 7zip is utterly idiotic.	14:49
charon	and you most likely have a ./master that should have ended up in refs/heads/ or so	14:49
Arafangion	charon: I expect so.	14:49
RedShift	ToxicFrog: nice, ty	14:50
	← Lgb left	14:50
SamB	Arafangion: you did it that way with the GUI, or the CLI?	14:50
	→ d0k joined	14:51
Arafangion	SamB: The problem is 7zr, not git.	14:51
SamB	Arafangion: okay, the CLI then	14:51
	I was talking about 7-zip's GUI and CLI	14:51
ShadeHawk	RedShift: See also my answer about difference between Git and CVS here: http://stackoverflow.com/questions/802573/difference-between-git-and-cvs/824241#824241	14:51
	← memiux left	14:52
Arafangion	SamB: The originator probably used the GUI, I'm using the CLI.	14:52
	→ willb joined	14:52
	← Modius_ left	14:53
Arafangion	SamB: Do you happen to know how to tell 7zr to behave?	14:53
SamB	Arafangion: yeah, you often have to be careful when using archivers that originated on the PC ...	14:53
	→ Lgb joined	14:53
Arafangion	SamB: I can understand that, but this is frankly idiotic!	14:53
	→ Modius_ joined	14:54
SamB	Arafangion: oh, you use the "x" command	14:54
	not the "e" command	14:54
	→ cannonball joined	14:54
Arafangion	SamB: Wow. Thanks for your help!	14:54
SamB	according to 7zr --help	14:54
	← asksol left	14:54
	← lohrun left	14:54
Arafangion	SamB: I saw 'e Extract'	14:55
SamB	unrar uses the same commands for those activities ...	14:56
	Arafangion wishes everybody used .tar.gz, as God Intended. (The GNU version)	14:57
	← johan-s left	14:58
Arafangion	Thanks SamB :)	14:59
	← summerbug left	14:59
Slashbunny	I have a bunch of local branches associated with remote tracking branches- Is there an easy way to fetch/merge them all in one shot? It looks like the way to do it would be to check each one out and "pull" individually- is that how I should do it?	15:00
	→ Modius__ joined	15:02
	→ feanil joined	15:02
	← roop left	15:03
	→ UncleCJ joined	15:03
offby1	I think there's a "git remote update"	15:04
UncleCJ	Hi. Is there a good way to address the "branching point" commit between the current and some other branch?	15:04
	← Modius_ left	15:04
offby1	UncleCJ: git merge-base	15:04
UncleCJ	Hmm...	15:04
charon	Slashbunny: if you want to update the local (tracking) branches safely, yes ... you can optimize away the checkout by manually verifying that it'll be a fast-forward, but that's more or less it	15:05
	you could just drop the local branches and only use the remote// family though, if you only fast-forward them anyway	15:06
	→ memiux joined	15:06
ShadeHawk	UncleCJ: git-merge-base (or a..b, usually), or branch.<name>.merge config variable	15:06
UncleCJ	ShadeHawk: What does branch.<name>.merge have to do with it, I thought that was for tracking branches?	15:07
	→ leo2007 joined	15:07
	← leo2007 left	15:07
UncleCJ	ShadeHawk: What does branch.<name>.merge have to do with it, I thought that was for tracking branches?	15:08
	offby1: Indeed, merge-base does the trick!	15:08
	← girishr left	15:08
	→ leo2007 joined	15:09
	← gaveen left	15:09
sadleder	Slashbunny: the branches must be tracking	15:10
ShadeHawk	UncleCJ: to know the branch for git-merge-base	15:10
Slashbunny	charon: thanks.. They will usually be fast-forward only, but sometimes not... I guess under normal circumstances I wouldn't even need the local branch until I have to do something.. then checkout -b would work	15:10
sadleder	Slashbunny: see "git config -e", there should be definitions like [branch "master"] remote = origin merge = refs/heads/master	15:11
	Slashbunny: for all branches you want to track	15:11
	← petermarks left	15:11
	← Arafangion left	15:11
	→ xrfang joined	15:12
xrfang	hello what is the difference between HEAD and ORIG_HEAD?	15:12
UncleCJ	ShadeHawk: Ah... wonder if that is relevant for me, I don't rally let tracking branches deviate from each other and in this case wanted to find the merge-base to do a cleanup interactive rebase against it	15:12
	xrfang: Just the question I was abotu to ask!	15:12
xrfang	:)	15:13
charon	Slashbunny: note that non-ff merges/pulls always risk a conflict and may be rather hard to automate. though in your case, you can probably merge them all in turn and stop if a conflict happens; then simply fix the conflict, commit the merge and re-run the script, as it'll just do nothing on the already-merged branches	15:13
	→ priidu_ joined	15:13
Slashbunny	sadleder: yeah.. let's say I have master, dev, stuff, remotes/origin/master, remotes/origin/dev, remotes/origin/stuff. And I have master checked out. If I do "git pull", it will do a fetch from origin and update all the remotes/... branches, and then merge master with origin/master. But it won't touch local branches "dev" and "stuff", unless I check them out and do git pull or git merge origin/x	15:13
xrfang	I am confused while doing some wine test. I did git checkout wine-1.1.1, after I am done with it, I do: git checkout HEAD, it still stay at 1.1.1, and then I did git checkout ORIG_HEAD, it goes to 1.1.17, but the head should be 1.1.22!	15:14
	← jony\|away left	15:15
Slashbunny	charon: Yeah, that's what I was planning on doing, I was wondering if there was a magical way- It makes sense though, you don't want to "pull all" the way "push all" works (push.default matching) because it's going to create a bunch of merge commits which may or may not blow up in the process	15:15
	→ balint joined	15:15
charon	xrfang: you're mostly confused about HEAD. HEAD is always the commit (actually it knows the branch) currently checked out, so checking out HEAD is a no-op. ORIG_HEAD is mostly a relic that is superseded by the reflog	15:16
SamB	charon: dunno about that "checking out HEAD is always a no-op"	15:16
charon	Slashbunny: pull is simply not the opposite of push, so the opposite of "push all" is actually "fetch"	15:16
xrfang	charon: what is the reflog and relic etc? what I wanted to do is to checkout the latest revision in the current branch, that's all. what should I do?	15:17
	→ jony joined	15:17
charon	SamB: a plain 'git checkout HEAD' should be	15:17
	→ section31work joined	15:17
charon	of course it's slightly different if you give it path arguments	15:17
SamB	I keep finding myself with a repository who's HEAD is rather out of sync with the index	15:17
section31work	How can i remove a file from the entire repository	15:17
xrfang	charon: so HEAD means "current working copy", NOT the last commit in the current branch??	15:17
charon	xrfang: by saying 'git checkout wine-1.1.1' (i.e., a tag), you switched to _no_ branch (a detached HEAD). you no longer have a current branch. tell git to go back to some with, say, 'git checkout master'	15:18
SamB	xrfang: it means whatever revision you're supposed to have checked out	15:18
charon	SamB: it will just compare HEAD and index, figure they're different, and re-merge the changes between (the previous) HEAD and index onto (the new, but same) HEAD, thereby doing nothing but a status display.	15:18
xrfang	charon: I have heard about this, but I don't really understand... they say you should use git checkout -b mybranch wine-1.1.1 etc.	15:19
	to create a "local branch"	15:19
SamB	where by "supposed to have", I mean you can easily confuse git by using git update-ref on the branch you've got checked out ...	15:19
xrfang	but, what if I want to remain in master, just get back in time?	15:19
	→ bcardarella joined	15:19
charon	(please don't tell newbies about update-ref....)	15:19
SamB	charon: sorry!	15:19
charon	:)	15:19
Slashbunny	charon: good point. Thanks for the help.	15:20
SamB	I actually was hoping you'd tell me about a safer way to do that, that wouldn't confuse git so much	15:20
xrfang	I really hope there are some good visualization tool (gitk is not enough)... I don't care about how git does its work under the hood, but I would like a big picture of "branches" etc.	15:20
	→ Modius joined	15:21
charon	SamB: check out said ref and use merge or reset (--hard usually, but that's also dangerous), depending on what you want to do	15:21
	xrfang: read man gittutorial and friends, and/or google for "git for computer scientists"	15:21
xrfang	ok, gurus... how about my questions? 1) check out a history version, and remain in master; 2) go back to the LATEST commit of the current branch...	15:21
Gitbot	xrfang: the gittutorial manpage can be found at http://git.or.cz/man/gittutorial	15:21
ShadeHawk	UncleCJ: but git-rebase finds merge-base by itself...	15:21
	→ eno__ joined	15:21
xrfang	ok, thanks	15:22
SamB	xrfang: what, you want to change master to point at the ancient revision ?	15:22
charon	xrfang: your point (1) is already a contradiction. please read the tutorial, it should become clear	15:22
xrfang	SamB: my scenario is, I want to do a bisect but I forgot which version the bug appears	15:22
SamB	xrfang: you don't need to stay in "master" to do that	15:23
xrfang	say, wine is at 1.1.22 now, I want to rewind to 1.1.1 to test whether the bug appears or not, then start a bisect	15:23
	← jony left	15:23
xrfang	so I need to checkout that version	15:23
SamB	sure!	15:23
	there's no harm in having a non-branch checked out, really	15:23
xrfang	SamB: I don't need to stay in master, yes, but I don't see a need to create a branch for this! do I have to ?	15:23
SamB	it's just that certain things are irrelevant and therefore don't happen	15:23
	xrfang: not really	15:24
	not if you're just going to test the workingness of the commit pointed at by the tag	15:24
ShadeHawk	xrfang: why not give 1.1.1 as appropriate bad or good revision, or starting point in bisect instead? man git-bisect	15:24
Gitbot	xrfang: the git-bisect manpage can be found at http://git.or.cz/man/git-bisect	15:24
	→ jony joined	15:24
SamB	ShadeHawk: he wants to make sure it actually is good, first	15:24
xrfang	ShadeHawk: I must find out a 100% good revision to start bisect	15:24
	← priidu left	15:24
xrfang	yes	15:25
SamB	anyway, there's no need to have a branch if all you're going to do is look	15:25
	a branch is only needed if you're going to start making commits	15:25
xrfang	ok. this is good. so now, someone said above that after a "branch-less" checkout, I need to type "git checkout master" to return to the latest commit of master, right?	15:26
SamB	(and even then, it's not absolutely needed unless you want to switch to some other commit ...)	15:26
	xrfang: yeah	15:26
	and get that to be the branch that moves when you commit, and so forth	15:27
xrfang	ok, so HEAD is the pointer of current working copy, it really means nothing, (i.e. it is variable, not constant) right?	15:27
SamB	(if you have checked out from a tag or SHA1, it won't move any branch, of course)	15:27
	→ sakyamuni joined	15:27
	→ ciskje joined	15:28
SamB	xrfang: well, it means something	15:28
xrfang	I am confused about HEAD, ORIG_HEAD and origin/master etc.	15:28
SamB	it can point at either a commit or another ref	15:28
xrfang	maybe they are not related at all, but...	15:28
SamB	if it points at a ref, that ref gets updated when you commit	15:28
	← hpoydar left	15:28
xrfang	SamB: ok, I will ignore that for not, "ref" confused me even more ;)	15:29
	last question for now:	15:29
SamB	xrfang: sure!	15:29
	→ ruz_mac joined	15:29
xrfang	I did git branch to see the status	15:29
SamB	you don't have to understand it all right now	15:29
xrfang	the * is on (no branch)	15:29
	what on earth is no branch?	15:29
	why ?	15:29
SamB	it just means you aren't on a branch at the moment	15:29
	→ hpoydar joined	15:29
SamB	if you look in .git/HEAD, it presumably contains an SHA1	15:30
xrfang	so why? I checked out a history version on master, then I lost my track?	15:30
ShadeHawk	(no branch) means that HEAD points directly to commit (so called 'detached HEAD'), and not to branch which in turn point to commit	15:30
	← RedShift left	15:30
ShadeHawk	if you do "git checkout <tag>" or "git checkout <remote tracking branch>" you would land on detached HEAD	15:30
SamB	xrfang: the past commits along a branch aren't really associated with the branch per-se anymore in git's way of thinking	15:30
ShadeHawk	git-rebase and I think also git-bisect make use of detached HEAD (git-bisect used to use 'bisect' branch)	15:31
	→ dwave joined	15:31
SamB	they're just parents of the branch tip	15:31
ruz_mac	hi, is there way to get next commit after X so I can walk towards to HEAD one commit at a time?	15:31
xrfang	SamB: you mean, if I visualize the git tree, there is actually no concept of a tree	15:31
	e.g. in wine. 1.1.22 is master, but 1.2.1 or 1.1.1 is no longer "master"?	15:31
SamB	xrfang: I mean, that it doesn't remember what branch something was first on, except in the reflog	15:31
	which is a local thing	15:32
	I doubt 1.1.22 is master	15:32
	well, could be	15:32
	→ mytruehero joined	15:32
SamB	I guess if it's like Linus' kernel tree ...	15:32
	anyway, yeah, historical commits don't remember what branch they were made on	15:33
	← Modius__ left	15:33
xrfang	ok, consider a chain.... can't I say that a revision is like a "node" on that chain? there is no idea of "parent" as in other VCS?	15:33
SamB	sure there is an idea of parent	15:33
xrfang	but you said "anyway, yeah, historical commits don't remember what branch they were made on"	15:33
	charon points at "git for computer scientists" again	15:33
	→ fridim_ joined	15:33
	→ secleinteer joined	15:33
	← gitte left	15:33
xrfang	ok, charon, I think I'd better do more reading ;)	15:33
mytruehero	I have a commit that I merged into my master branch and pushed. I decided that I don't want to deploy that commit yet, however, so I cherry-picked it into a separate branch. How can I remove that commit from master? Or do a commit that essentially un-does the commit I want to remove, while allowing me to merge it back in later from the branch it now resides in?	15:34
xrfang	thanks all	15:34
	bye	15:34
SamB	but the branch name just refers to one node at a time -- hopefully, public branches only move from one commit to (a child of )*that commit	15:34
	(that parenthesized bit with the asterisk is in regex notation)	15:34
wereHamster	xrfang: git revert $commitsha	15:34
xrfang	?	15:35
	→ Dvyjones joined	15:35
secleinteer	hi, i used git-svn to download an svn repository. now i'd like to commit my changes to a _different_ svn repository. i used --stdlayout when i initially downloaded the repository. when i committed to the new repository, it didn't translate back to stdlayout (i.e., "branch", "trunk", "tags" directories in svn). does anybody know why this is?	15:35
Dvyjones	How do I pull changes from another repo?	15:35
hyperair	Dvyjones: git pull git://otherrepo branch	15:35
wereHamster	xrfang: git-revert creates a new commit that reverts the changes of another commit	15:35
Dvyjones	hyperair: branch is the branch I want to pull it into?	15:36
hyperair	Dvyjones: no, the branch you wanna pull from.	15:36
Dvyjones	hyperair: Oh, ok.	15:36
hyperair	Dvyjones: you're expected to be on the branch you want to pull to	15:37
xrfang	wereHamster: yes I got some idea, but I am a bit confused about the status of working copy and the repo etc. quite hard to explain now... I think I need just to read the manual... anyway, this seems is not my current question? ；）	15:37
Dvyjones	hyperair: Doesn't seem to work...	15:37
	hyperair: I want to pull this: http://github.com/danlucraft/redcar into this: http://github.com/dvyjones/redcar	15:37
wereHamster	xrfang: git-scm.com <- there's plenty of documentation	15:38
	← balint left	15:38
	→ ph^ joined	15:38
	→ netoman joined	15:38
ShadeHawk	Dvyjones: First, use git URL, and not GitHub URL, as an argument to git-pull / git-fetch / git remote add	15:38
hyperair	Dvyjones: git clone git://blabla/dvyjones/blabla.git; git remote add danlucraft git://blabla/danlucraft/blabla.git; git pull danlucraft	15:38
	er i forgot to cd blabla	15:39
ShadeHawk	dvorak1: git pull git://github.com/danlucraft/redcar.git	15:39
	Dvyjones: git pull git://github.com/danlucraft/redcar.git	15:39
Dvyjones	ShadeHawk: I am :P	15:39
	It says "already up-to-date"	15:39
	*Everything	15:39
xrfang	wereHamster: thanks	15:39
ShadeHawk	so that means that there aren't any new commits to be fetched... nothing to be done	15:39
	→ aspotashev_ joined	15:40
Dvyjones	Well, there's a lot of commits missing...	15:40
	Wait, nevermind...	15:40
	D'oh	15:40
	→ jacobat joined	15:40
	← aspotashev_ left	15:41
	← mitkok left	15:41
	→ aspotashev_ joined	15:41
	→ com4_ joined	15:42
	← jony left	15:42
	← aspotashev_ left	15:42
	→ jony joined	15:44
	← ciskje left	15:44
	→ aspotashev_ joined	15:44
	← xrfang left	15:44
	← mytruehero left	15:45
	→ miriamg joined	15:45
	← aspotashev_ left	15:45
	→ aspotashev_ joined	15:45
	← jony left	15:46
	→ ciskje joined	15:46
ShadeHawk	Some of git web interfaces (GitHub, Gitorious) present a kind of "tree blame" in a 'tree' view (probably following what ViewVC and other web interfaces for per-file SCMs), with "date the file changed" beside each file (and sometimes "commit that changed the file")	15:47
	Is such information useful? Would it be worth adding it (as an option) to gitweb, for example?	15:47
	← sfwc left	15:49
	← giallu left	15:50
	→ sfwc joined	15:50
	← boto_ left	15:50
	→ hobodave joined	15:51
	← aspotashev left	15:54
	→ parasti joined	15:54
	→ RayMorgan joined	15:55
	← gilimanjaro left	15:56
	→ juanjoc joined	15:56
	→ Yuuhi joined	15:56
	← Nazcafan left	15:57
	→ schacon joined	15:58
	← sfwc left	15:59
	→ abstrusething joined	15:59
	→ akitada joined	15:59
	→ sfwc joined	15:59
freelook1	I'm trying to track a remote git repository but make my own custom changes to it. The problem is that I have a desktop and a laptop, and so I'd like to keep changes synced in between them. Any ideas on a good way to do this? I've got a local server as well. Perhaps having a repository on the server is a good way of doing it?	15:59
	joshthecoder_afk → joshthecoder	16:00
	→ carrus85 joined	16:01
	← abstrusething left	16:01
Aides	just push and pull your changes from/to desktop to/from laptop	16:02
	→ Bass10 joined	16:03
	← priidu_ left	16:03
Aides	best way of course would be to setup remotes on both sides and pull	16:03
	because pushing to non-bare repository could be not nice	16:04
	← gberdyshev_ left	16:04
	→ georgyberdyshev joined	16:04
	thiago also recommends pulling only	16:04
freelook1	That could work except my desktop is not always on, and my laptop is often away. It would be nice to have a central point for myself to pus from/pull to	16:04
	push*	16:04
thiago	then do it	16:05
SamB	freelook1: you could set up something on github or gitorious or repo.or.cz (not sure repo.or.cz can do that today, though -- it was partly-down for maintenance yesterday)	16:05
Aides	then yes, local server is nice	16:05
	just make it bare, so you can push safely	16:05
	→ WinterWeaver joined	16:05
freelook1	ok, I'll try that. Thanks a lot	16:05
	→ priidu_ joined	16:06
	← agile left	16:06
freelook1	Aides: in the case of a local server then, is gitosis the recommended way?	16:06
	← psoo left	16:07
	← WinterWeaver left	16:07
Aides	freelook1, if you'll be using it yourself, just use plain ssh	16:07
	drop it somewhere, say /home/user/repo	16:07
	and then access as ssh://server/home/user/repo	16:08
freelook1	great. thanks Aides	16:08
Aides	gitosis is nice for several users, because git loves to screw file permissions	16:08
freelook1	ahh	16:08
	← bschindler left	16:10
	→ georgyberdyshev_ joined	16:10
	→ JasonWoof joined	16:11
	→ DavidKlein joined	16:12
	→ webcoder_ joined	16:13
	← qhoxie left	16:13
akitada	When I work with code using git, I always try to make a good commit, which is a fix or chagne of some specific problem/enhancement. Git is great for this. I usually create a branch for each fix and try to make up a good commit. but there's one problem. How I can work with multiple branches.	16:13
	As I said, each branch is for a specific problem. Sometimes a fixing a bug takes a lots of time and during working on that, sometimes I have to work with another problem.	16:15
wshimmy	akitada: that's the whole point of working with branches in git ;)	16:16
	akitada: man git branch and man git checkout should help you	16:17
akitada	In that case, I would: git stash; git checkout -b another_bug; hack hack; git add; git commit; git checkout -b branch-i-came-from; git pop --index; hack again...	16:17
	→ WinterWeaver joined	16:17
wshimmy	akitada: sounds about right, except you don't want the -b in checking out branch-i-came-from	16:18
Slashbunny	I noticed git can create physical paths for branches, which is pretty cool- Like git branch majorchange/subchange1, git branch majorchange/subchange2 . Let's say for example one of those changes is going to require more work, or more experimental tests done on it- Following this heirarchy, I would create another sub folder... Now I understand why git branch majorchange/subchange1/subsubchangeA doesn't	16:19
	work on a technical level (it's trying to create a directory where a file is), but how do you think I should handle this? make a subchange1/master branch and a subchange1/subsubchangeA branch? Has there ever been talk of having branches represented as directories to begin with and having a directory/.hash file... I assume this would be more clunky and slower... just curious	16:19
ShadeHawk	freelook1: Take a look at GitTips page on git wiki: http://git.or.cz/gitwiki/	16:19
akitada	and here's my question. Am I using git right? I think I could do it better and like to have some advice from git experts.	16:19
	wshimmy: oops, that's typo...	16:19
wshimmy	akitada: and you can use git checkout -b another_bug master, for example, if you don't want the new branch to have the commits on your current branch, which is probably the case	16:20
	← altrux left	16:20
	← webcoder left	16:20
	→ eletuchy joined	16:21
akitada	wshimmy: oh, that's nice. is that cause the branch is based on master? cool.	16:21
	→ johan-s joined	16:21
wshimmy	akitada: git help branch	16:22
	← WinterWeaver left	16:22
akitada	wshimmy: ah, <start-point>. cool...	16:22
	→ mgerwitz joined	16:24
	← lorandi left	16:24
wshimmy	Slashbunny: do you really need that much hierarchy?	16:25
akitada	git checkout -b yet-another; hack hack; git add; git commit; git checkout master; git merge yet-another;	16:25
	→ lorandi joined	16:25
akitada	this is my workflow.	16:25
wshimmy	Slashbunny: if so, I would probably just put a master under my branch dirs, just from a taxonomic point of view	16:27
SamB	I wish git stash could handle in-progress merges :-(	16:28
	← aspotashev_ left	16:30
Slashbunny	wshimmy: yeah, that seems reasonable	16:30
	→ aspotashev joined	16:31
	← georgyberdyshev left	16:31
	→ schacon_ joined	16:31
miriamg	anyone know how to specify a specific revision to pull a submodule at? i see "git submodule add -b <branch>", but nothing to specify a revision...	16:32
	← angerman left	16:32
ShadeHawk	SamB: yes, unfortunately git-stash cannot handle in-progress merges... but that was brought into attention; it is not technically impossible (each index stage could be separate parent, for example)	16:32
	← georgyberdyshev_ left	16:32
	→ georgyberdyshev_ joined	16:32
	→ Alinermedia joined	16:33
	← khmarbaise left	16:33
SamB	ShadeHawk: I know that!	16:33
	it's certainly something it could do	16:33
	it'd just need more work	16:33
wshimmy	rumour has it they accept patches ;)	16:33
SamB	I dunno if I can do the work or not, though	16:34
	← priidu_ left	16:34
akitada	learning git is fun. cool features are still there even in "git-add". learning something new everyday :)	16:34
	← leanucci left	16:34
	SamB sometimes wishes git didn't have so many "features"	16:35
	→ leanucci joined	16:35
ShadeHawk	SamB: as in... if you wait, perhaps it would be soon in git (I mean stashing away merges...)	16:35
SamB	(where feature means something like "projecting corners upon which to injure oneself" ...)	16:35
	← schacon left	16:37
	→ jayallen joined	16:38
	← mgerwitz left	16:38
	→ qhoxie joined	16:38
	← davidfetter left	16:38
	→ davidfetter joined	16:39
	→ azol joined	16:40
	← dwmw2 left	16:40
	← georgyberdyshev_ left	16:41
	← marc_kdab left	16:41
	→ Galaxor joined	16:42
Galaxor	Hey. I've got a git repo that I can't push to anymore. It says: error: unable to create temporary sha1 filename ./objects/19: File exists	16:42
	What can I do about that?	16:42
	→ WinterWeaver joined	16:43
	← ciskje left	16:43
	← shruggar left	16:43
Galaxor	the repo I'm trying to push from is on a machine with git 1.6.0.1, while the intended recipient is running 1.6.0.	16:43
	→ kraymer joined	16:45
	→ scientes joined	16:46
	← ntoll left	16:47
	← akitada left	16:48
ShadeHawk	Galaxor: it looks like something f**ked up on remote side. Is objects/19 a file, or a directory?	16:49
	→ javatexan joined	16:49
	→ FauxFaux joined	16:49
	→ random8r joined	16:49
	→ lois[pac] joined	16:49
Galaxor	A directory.	16:49
	It's got one file in it.	16:50
ShadeHawk	hmmm...	16:50
doener	hm, looks like the usual "Permission denied" being half-fixed	16:50
	→ abstrusething joined	16:50
	→ Modius_ joined	16:50
	← fujin left	16:50
ShadeHawk	well, strange error message it is...	16:50
	→ naeu joined	16:51
doener	the "Permission denied" message was also wrong. The fallback code eats the original error message	16:51
	now if I could recall where that code is...	16:51
	← hpoydar left	16:51
Galaxor	Aha. Permission denied indeed. I found that some of the objects directories were owned by "root:root" instead of "git:git".	16:51
	I changed it and it now accepts pushes.	16:52
	Permission denied would've been a much more helpful error.	16:52
	→ hpoydar joined	16:52
	→ drmclem joined	16:52
doener	Galaxor: yeah, your version probably still has the bug that made it always try to create the directory when creating the file fails	16:53
	Galaxor: and that kills the original error message	16:53
	→ dreiss joined	16:54
Galaxor	Aha. Cool, I'll try to upgrade when I've got some time. Thanks.	16:54
doener	Galaxor: got fixed in 1.6.0.5	16:55
	Galaxor: (server side is what matters here)	16:55
	→ solydzajs joined	16:55
	→ thiago_home joined	16:55
Galaxor	doener: Okay, thanks.	16:56
	→ Modius__ joined	16:56
	→ voker57_ joined	16:57
	← tono left	16:57
	→ rshann_ joined	16:58
	→ solar_ant joined	17:00
	→ Weaselweb joined	17:02
	← random8r left	17:05
	← schacon_ left	17:05
	→ schacon joined	17:06
	→ factotvm joined	17:07
	← Modius left	17:07
	← Yuuhi left	17:08
	→ dwmw2 joined	17:09
	→ cloudhead joined	17:09
miriamg	ah, I found the answer here (if anyone else was wondering): http://woss.name/2008/04/09/using-git-submodules-to-track-vendorrails/. clearly I still don't really understand submodules	17:10
	→ __iron joined	17:10
	← voker57 left	17:10
	→ tvw joined	17:11
	← drmclem left	17:11
ShadeHawk	Does anybody know why Bazaar-NG wasn't considered as viable choice for Linux kernel at the time git was being created?	17:11
	(July 2005)	17:11
	← j416 left	17:11
FauxFaux	SLOW	17:12
	→ Dashkal joined	17:12
	← Modius_ left	17:13
	← Modius__ left	17:14
	→ krawek joined	17:15
secleinteer	hi, i used git-svn to download an svn repository. now i'd like to commit my changes to a _different_ svn repository. i used --stdlayout when i initially downloaded the repository. when i committed to the new repository, it didn't translate back to stdlayout (i.e., "branch", "trunk", "tags" directories in svn). does anybody know why this is?	17:15
	← memiux left	17:16
	mtvee → mtvee\|away	17:16
	→ ceez joined	17:17
	→ cxvii joined	17:17
	← Galaxor left	17:17
	→ Fullmoon joined	17:19
	← jmesnil left	17:20
	→ laj joined	17:21
	→ agile joined	17:23
	→ knweiss joined	17:24
	→ sgrimm joined	17:25
	→ moonflux joined	17:26
wereHamster	secleinteer: git-svn doesn't automatically commit all branches to the svn repo, only the currently active one	17:26
	→ memiux joined	17:28
	← jrmuizel left	17:33
	→ towski joined	17:33
	← Hendu left	17:34
	→ jrmuizel joined	17:34
ShadeHawk	FauxFaux: You mean that Bazaar-NG was too slow for Linux kernel?	17:35
	Errr... Bazaar: Initial release 14 December 2007	17:35
	didn't exist then	17:35
	(and Arch has too complicated UI)	17:35
doener	ShadeHawk: that was 1.0	17:36
	ShadeHawk: http://doc.bazaar-vcs.org/bzr.1.15/en/release-notes/NEWS.html#bzr-0-0-0-69-2005-03-22 -- first public release 2005-03-22	17:36
	← memiux left	17:37
	→ memiux joined	17:37
ShadeHawk	ah... so it did exist then...	17:38
	→ eletuchy_ joined	17:38
	→ adamtanner joined	17:39
	→ kuadrosx joined	17:39
	← legnaleurc left	17:40
	← dreiss left	17:40
	→ JEEBcz joined	17:41
	← juanjoc left	17:42
	← sadleder left	17:42
	→ juanjoc joined	17:42
	→ salty-horse joined	17:44
	← sunoano left	17:44
	→ djgera joined	17:45
secleinteer	wereHamster: is there a way to change that behavior?	17:45
	← adamtanner left	17:46
	← sakyamuni left	17:46
	→ sakyamuni joined	17:46
	bleything\|away → bleything	17:48
	→ sunoano joined	17:49
almostautomated	Hiya all; could someone configrm that when viewing diff out that the highlighted '^M' I'm seeing is a marker designating whitespace at EOL ?	17:51
	→ FylGood joined	17:51
doener	almostautomated: it's a CR	17:51
wereHamster	secleinteer: I don't think so. Do you want to mirror the svn repo through git to another svn repo?	17:52
	← rshann_ left	17:52
almostautomated	doener: Thanks; I knew I had mixed line endings ( expected ) yet wasn't expecting that. Thank you.	17:52
	→ robin joined	17:53
	→ jwheare joined	17:54
	← eletuchy left	17:55
	← bcardarella left	17:55
jwheare	i just made 5 commits that i want to completely remove, not just git revert	17:55
Aides	have you pushed them?	17:55
jwheare	no	17:55
	← towski left	17:55
	← memiux left	17:55
Aides	then git reset --hard HEAD^^^^^	17:55
	← cxvii left	17:56
jwheare	i guess i can replace HEAD^^^^^ with the sha that i want to reset to	17:56
	cheers	17:57
	→ psoo joined	17:58
	← psoo left	17:58
	← FylGood left	17:58
	→ jspiros joined	17:58
	bleything → bleything\|away	17:59
	bleything\|away → bleything	18:00
knweiss	Or shorter: git reset --hard HEAD~5	18:00
	← ruz_mac left	18:02
	→ lanwin joined	18:02
	← agile left	18:02
sakyamuni	I want to undo a checkin. I haven't done a push. I did git reset HEAD^1.. what's the next command that I should do?	18:02
	→ psoo joined	18:03
lanwin	hi a question. Is there an command to add all deleted files to the next commit without calling git rm for each separatly?	18:04
frsk	sakyamuni: well, git reset head^ removes the commit.. what more do you want to do? =)	18:04
	→ kobani joined	18:04
jaeckel	sakyamuni: why not only amending?	18:04
Aides	lanwin, git add -u	18:04
sakyamuni	jaeckel: never done that, how is that accomplished?	18:04
	→ khmarbaise joined	18:04
sakyamuni	frsk: i thought --hard removes it whilst the way I did it, soft, it's still "sorta" still there.	18:05
frsk	sakyamuni: --hard reverts changes in the file as well	18:05
jaeckel	in git gui changing radio button from 'new commit' to 'amend last commit'	18:05
sakyamuni	I dod a git diff and I see the changes, should I just do git reset --hard now?	18:05
jaeckel	:)	18:05
	→ bradly joined	18:05
jaeckel	don't use commandline for committing	18:05
	← sfwc left	18:05
frsk	sakyamuni: If you'd want to throw the changes away, yes	18:06
	→ SRabbelier joined	18:06
sakyamuni	frsk: i want to throw 'em away so git reset --hard worked	18:06
frsk	sakyamuni: If you're not 100% sure, you could always make a backup of the files your first time, though =)	18:06
	okey, great =)	18:06
jaeckel	but my git cheat sheet says commit -a --amend	18:06
Aides	cheater!	18:07
jaeckel	=D	18:07
	← jwheare left	18:07
jaeckel	well	18:07
	finishing time	18:07
	→ lymeca joined	18:07
jaeckel	bye	18:07
	← jaeckel left	18:08
	← naeu left	18:08
	→ sfwc joined	18:10
	← factotvm left	18:13
	→ agile joined	18:13
lanwin	ah thanks Aides	18:14
	→ mozinator joined	18:16
	← abstrusething left	18:16
	→ abstrusething joined	18:20
	→ boto joined	18:20
	→ altrux joined	18:22
	→ bschindler\|lapto joined	18:25
	← solar_ant left	18:26
	← boto left	18:27
	→ boto joined	18:27
	→ solar_ant joined	18:28
	← Dashkal left	18:29
	→ Dashkal joined	18:29
	← bschindler\|lapto left	18:33
	→ bobesponja joined	18:35
	← robin left	18:36
	→ dreiss joined	18:37
	← jayallen left	18:38
secleinteer	wereHamster: yes, that's right	18:39
	i use git locally to work on a public project	18:40
	we are temporarily managing a branch locally due to some licensing issues, using svn as well	18:40
	no one else will use git :(	18:40
	← aqmaiu left	18:44
	← HG`` left	18:48
	→ icwiener joined	18:48
jbwiv	is it possible to edit timestamps in a git repository? Say I wanted to rewrite history and have every commit occur between 4 pm and 6 pm? Strange question, I know	18:51
Chani	hmm. there's no git cp? is this because git will magically know I've copied?	18:51
Grum	Chani: yes	18:52
Chani	cool.	18:52
	← wshimmy left	18:52
SamB	jbwiv: it's possible, but you're going to have to come up with a better use-case if you want me to say how...	18:52
	→ memiux joined	18:52
	← bgerber left	18:53
	← agile left	18:55
	→ cousine_ joined	18:55
	cousine_ → cousine	18:56
	← hyperair left	18:56
	← psoo left	18:57
	← memiux left	18:58
	→ bgerber joined	18:58
	→ khmarbaise_ joined	18:59
ShadeHawk	Chani: git would turn of copy detection if you enable / request it; git cp is just cp + git add.	18:59
	← cousine left	18:59
	→ cousine joined	19:00
	← schacon left	19:00
ShadeHawk	jbwiv: man git-rewrite-hist + GIT_AUTHOR_DATE and GIT_COMMIT_DATE environmental variables	19:00
Grum	ShadeHawk: rm+add	19:00
ShadeHawk	Grum: 'git cp' if it existed would be equivalent to 'cp' + 'git add'	19:01
Grum	errr yes sorry not mv ><	19:01
ShadeHawk	On the other hand 'git rm' and 'git mv' has built in extra safeties	19:01
	→ psoo joined	19:01
	← flazz left	19:02
	→ flazz joined	19:02
	← abstrusething left	19:03
	→ hyperair joined	19:05
	← solydzajs left	19:06
	→ mge joined	19:08
	← khmarbaise left	19:08
	→ jetru joined	19:09
	→ grahal joined	19:09
mge	so I accidentally did a git checkout github/master and reverted my commited changes I had loally to what was on the github server. Any way to recover mylocally commited changes?	19:09
	→ blaenk joined	19:10
	→ Modius joined	19:11
maxb	git checkout whatever-branch-you-were-on-before ?	19:12
	Unless you were working on a disconnected head?	19:13
FauxFaux	git fsck --unreachable, rite?	19:13
mge	I am a complete newb.. this is my first attempt at merging changes remotely with my local changes	19:13
	git checkout master maybe?	19:13
hiffy	mge: yup	19:14
ShadeHawk	mge: 'git checkout master' (assuming that you were on branch 'master' before 'git checkout github/master')	19:14
hiffy	if it got comitted you should be ok	19:14
ShadeHawk	mge: git would refuse to checkout if it would mean losing your local changes	19:14
hiffy	so in this case this means that github/master is not mge's origin/master?	19:14
	oh right	19:15
	it prolly makes a distinction b/w origin master and the local master	19:15
mge	it looks like git checkout master got me back to where I was	19:15
	→ omakase joined	19:15
	→ solydzajs joined	19:16
	→ wshimmy joined	19:16
	→ icwiener_ joined	19:17
	→ jayallen joined	19:18
	← icwiener left	19:18
	→ drdave joined	19:20
	→ jetienne joined	19:20
jetienne	q. is there a way to measure the size of a change between version ?	19:20
	→ walken joined	19:20
walken	hi.... I've been wondering, what is .git/branches for ? all my branches show up in .git/refs/heads instead.	19:21
	bleything → bleything\|away	19:21
drdave	Hi, I'm trying to write a script that exports a revision of a given file - say the third from the HEAD revision. I'm trying things like:	19:21
thiago_home	jetienne: and how do you measure change?	19:21
	walken: legacy. It's not used anymore.	19:22
drdave	git show Court~3:"opn_Plugins/opn_Text/rev/object/View\|Layouts\|Horizontal\|Three/group_1207"	19:22
	or git show HEAD~3:"opn_Plugins/opn_Text/rev/object/View\|Layouts\|Horizontal\|Three/group_1207"	19:22
thiago_home	drdave: I don't think wildcards work there	19:22
	bleything\|away → bleything	19:22
thiago_home	drdave: just tested: they don't. You should pass the full path.	19:23
drdave	ok - trying...	19:23
	← webcoder_ left	19:23
	bleything → bleything\|away	19:24
	← jetienne left	19:24
Chani	how do I set what files git ignores in ~/.gitconfig? the gitignore man page explained how to do it in .gitignore but that's different...	19:25
	← rudi_s left	19:25
	← lorandi left	19:25
drdave	fatal: ambiguous argument 'Court~3:/[[full_path]]': unknown revision or path not in the working tree.	19:27
	Use '--' to separate paths from revisions	19:27
	The [[full_path]] exists and is inside the git repository	19:27
salty-horse	is there a way to set svn props with git-svn?	19:28
thiago_home	salty-horse: not that I know of	19:28
	drdave: can you do git shouw Court~3 ?	19:28
Chani	.gitconfig could really do with some more documentation...	19:29
thiago_home	Chani: man git-config	19:29
Gitbot	Chani: the git-config manpage can be found at http://git.or.cz/man/git-config	19:29
	bleything\|away → bleything\|away\|a	19:30
Chani	thiago_home: yeah, I looked there too	19:30
drdave	Yes - though as it is not referencing a given file - it does not return the text of the revision, but a load of stuff :)	19:30
	→ dedmajor joined	19:30
Chani	unless gitignore meant that I should set core.excludesfile	19:30
	andthen have another file in hte format of .gitignore	19:30
drdave	PS - I'm on OSX	19:31
	← Alinermedia left	19:32
	→ Alinermedia joined	19:32
thiago_home	Chani: git config core.excludessfile ~/path/to/global-gitignore	19:32
drdave	I get the result i want if i use a form like:	19:32
	git show f1a5079bbbb772641a512692d007e4d64721bb56:"opn_Plugins/opn_Text/rev/object/libOPN_GIT/card_1002"	19:32
Chani	ooh, it worked :)	19:33
thiago_home	Chani: with one s less	19:33
	← khmarbaise_ left	19:33
Chani	no more .svn or .swp files :)	19:33
thiago_home	drdave: ok	19:33
	drdave: but I can vouch that git show REF~3:path/to/file works	19:34
drdave	OK - so you think could be an OSX thing?	19:34
thiago_home	drdave: try placing the entire argument in quotes	19:35
drdave	OK	19:35
thiago_home	or, just test: echo Court~3:opn_Plugins/opn_Text/rev/object/libOPN_GIT/card_1002	19:35
	what do you see?	19:35
	→ giallu joined	19:36
	← cousine left	19:36
	← bleything\|away\|a left	19:37
	→ bleything joined	19:37
	→ rshann_ joined	19:38
	← lanwin left	19:40
drdave	Been testing... will try ur suggeestion in a sec	19:41
	Everything failed - single quotes, double, full and relative paths...	19:41
	this sort of works:	19:42
	← javatexan left	19:42
drdave	git show "HEAD~3:opn_Plugins/opn_Text/rev/object/libOPN_GIT/card_1002"	19:42
	BUT - it returns revision 0 and not 3 back from HEAD	19:42
	← salty-horse left	19:42
drdave	Either single or dounle quotes - (I changed to another file without "\|" in the filename)	19:42
	→ memiux joined	19:42
	→ lorandi joined	19:43
	← lorandi left	19:43
	→ lorandi joined	19:43
	→ khmarbaise joined	19:44
drdave	not sure what you want me to "echo"	19:44
	ls -la opn_Plugins/opn_Text/rev/object/libOPN_GIT/card_1002	19:44
	→ javatexan joined	19:44
drdave	returns as expected - text file is there and exists	19:45
	I'm at the git root	19:45
	repository root	19:45
	← j-b left	19:46
almostautomated	hey guys; quick question to satisfy a curiosity; if someone creates a commit with a bunch of diff stuff in it; and we only want parts of that single commit is it fairly simple or is it easier to have them redo it and re-submit?	19:48
	→ cytrinox joined	19:50
	→ eht joined	19:50
	→ rlorandi joined	19:52
	→ tlrobinson joined	19:52
tlrobinson	why does git sometimes complain that a file isn't uptodate and can't be merged when pulling the first time, but if i immediately pull again it works?	19:53
	(and git status doesn't show the file was modified)	19:53
thiago_home	"not up to date" means you have changes that haven't been committed	19:54
	Git won't touch modified files	19:54
tlrobinson	yeah, but "git status" shows nothing has changed. and if i immediately try pulling again then it works the second time	19:55
	← com4_ left	19:55
	→ priidu joined	19:55
tlrobinson	know what i mean? does that happen to anyone else?	19:55
	→ memiux_ joined	19:55
thiago_home	which version are you using?	19:56
drdave	Ok - getting somewhere - relative files seem to work in uncomplicated cases (without quoting) need to do some more testing	19:56
tlrobinson	1.6	19:56
thiago_home	tlrobinson: no, I haven't seen this problem before	19:56
drdave	git version 1.6.2.4	19:56
	← memiux_ left	19:56
	→ bx2 joined	19:56
	→ memiux_ joined	19:56
tlrobinson	well, its not a big deal, it's just curious	19:59
	→ schacon joined	20:00
	→ plasticine joined	20:00
	← bradly left	20:01
drdave	Ok seems to work with relative paths and second part quoted - rest of it must have been my mistake. Thanks.	20:02
	← lorandi left	20:03
	← rshann_ left	20:04
almostautomated	what type of search terms can lead me towards an answer to:	20:04
	hey guys; quick question to satisfy a curiosity; if someone creates a commit with a bunch of diff stuff in it; and we only want parts of that single commit is it fairly simple or is it easier to have them redo it and re-submit?	20:04
	→ ciskje joined	20:05
	→ HG` joined	20:05
	→ bcardarella joined	20:05
drdave	Anyone recommend a good cross-platform way to view git diff / changes in text files cross-platform? I'm using "FileMerge" on OSX (part of developer tools) - looking to distribute script of Linux and Windows...	20:05
SamB	almostautomated: have him redo it!	20:05
thiago_home	drdave: "less" is very nice to read diffs	20:06
	← cytrinox_ left	20:06
	← memiux left	20:06
drdave	:)	20:06
SamB	yeah, just using PAGER is best	20:07
almostautomated	SamB: and is there a man page I can refer him to on how to seperate out the different parts of the code changed in one given file? ( perhaps git add -i )	20:07
SamB	almostautomated: sometimes I do that with git-gui	20:07
EvanCarroll	I want vimdiff to work with flat calculated diffs.	20:08
	=(	20:08
	→ unreal_ joined	20:08
	→ memiux joined	20:08
almostautomated	SamB: K; I'll see if I can lead him in that direction. Yet as you know; sometimes people say you want it here it is; in that case is there an effective way to parse out _parts_ of the commit?	20:09
	← memiux_ left	20:10
SamB	almostautomated: copy and paste from the diff ?	20:10
	but never encourage contributors to make things difficult for you	20:11
almostautomated	LOL; I was afraid of something like that. Definitely not, I'll absolutely ask that it be redone	20:11
	thanks for the input	20:11
	→ TheUni joined	20:12
TheUni	i'm trying to use git svn to clone a repo that's obviously corrupt somehow. every time it hits a certain file, git barfs. is it possible to ignore a file when cloning?	20:13
	→ cytrinox_ joined	20:13
thiago_home	TheUni: ignore? no	20:13
	but you can start the clone after it	20:14
SamB	TheUni: there's not a chance that it's a bug in git-svn ?	20:14
thiago_home	drop everything before that	20:14
	→ gigi joined	20:14
SamB	I'd post to the git mailing list with instructions to reproduce	20:14
TheUni	thiago_home: don't think so.	20:14
	same file every time	20:15
	and if i skip it, the next time it's referenced it does the same thing	20:15
	← tvw left	20:15
SamB	TheUni: what's the filename ?	20:15
TheUni	mplayer.dll	20:15
SamB	okay ...	20:15
	and I guess it's probably not over 2GiB	20:15
	;-)	20:15
TheUni	heh, no. nothing weird about it	20:16
	and for the revs that fail, i can grab the file from svn just fine	20:16
SamB	so you can check out those entire revs just fine ?	20:16
	→ schacon_ joined	20:16
SamB	sounds like a bug in git-svn to me ...	20:16
	← schacon left	20:16
SamB	so, report it on the list	20:17
	I assume it's a public SVN repo?	20:17
TheUni	yea	20:17
	http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=512710	20:17
	mtvee\|away → mtvee	20:18
	→ clamothe joined	20:19
	← dedmajor left	20:19
	← gigi left	20:20
clamothe	Using git-svn, how should i go about merging two branches? They are remote 'svn' branches, and I have local branches for both of them.	20:20
	just merge my local branches and dcommit, or will that be messy?	20:20
	→ dedmajor joined	20:22
	← memiux left	20:24
	→ cytrinox1 joined	20:24
	→ Toad joined	20:25
Toad	hey all. if I've done a merge using git merge-tool, but then decided I want to change my mind and merge it a different way, is there a way to mark the file as "needing merge"	20:26
	getting "file does not need merge" since merge-tool git-added it when I was done	20:26
	→ radarek joined	20:27
	→ j416 joined	20:27
	← cytrinox left	20:28
	→ Radar joined	20:29
	← SRabbelier left	20:30
	← solydzajs left	20:31
	→ lightcap joined	20:31
	← clamothe left	20:32
	→ gitte joined	20:33
	→ tote joined	20:34
tote	What could be wrong when i get: git: 'http-push' is not a git-command. See 'git --help'.	20:34
ToxicFrog	What command are you issuing?	20:35
tote	git push	20:35
	ive cloned a repo from an http url	20:35
JasonWoof	almostautomated: git add -i is nice for committing some, but not all changes to the working tree. if you reset the commit you'd like part of, you should be able to use git add -i (then press 5 then *) to have it ask you which diff chunks you'd like added to the index	20:36
	tote: you can't put over http	20:36
	s/put/push/	20:36
	http is read-only	20:36
	← lucsky left	20:36
tote	JasonWoof: why is there a git-http-push command then?	20:37
doener	JasonWoof: dav?	20:37
JasonWoof	tote: there isn't. that's why you got a message that it's not a command	20:38
doener	tote: self-built git, or from distro packages?	20:38
JasonWoof	doener: git uses http in a read-only way	20:38
tote	doener: debian package	20:38
JasonWoof	afaik	20:38
doener	JasonWoof: man git-http-push	20:38
Gitbot	JasonWoof: the git-http-push manpage can be found at http://git.or.cz/man/git-http-push	20:38
tote	and dav is installed on the http-server	20:38
almostautomated	JasonWoof: selection by diff hunk - that's on a per file basis or per hunk within the file's diff? Either way; that gets us closer. Thanks.	20:39
JasonWoof	almostautomated: "git add -i" can do per file or per hunk	20:39
	almostautomated: option 5 in the menu is per-hunk, which I like a lot	20:39
	almostautomated: then it asks you which files you'd like to pick and choose hunks from	20:39
doener	tote: lenny?	20:40
	← galderz left	20:40
almostautomated	fantastic... I'll play with that tonight! Thanks again JasonWoof	20:40
tote	doener: squeeze	20:40
	← cytrinox_ left	20:40
	← knweiss left	20:41
wshimmy	"git add -p file" gets you there too, if you have a tracked file modified that you want to add hunks of	20:41
doener	tote: hm, does /usr/lib/git-core/git-http-push exist? Does "dpkg -L git-core" show it?	20:41
JasonWoof	I stand corrected, git can push to DAV. Although the manpage doener sent me to says that it's disabled if your libcurl is older than 7.16	20:42
	← tlrobinson left	20:42
gitte	JasonWoof: how old is 7.16?	20:42
tote	doener: yes its there	20:42
gejr	how can i set the name of the project in gitweb? It's always named ".git"..	20:43
	← DavidKlein left	20:43
	← priidu left	20:43
almostautomated	wshimmy: I had read that somewhere but didn't see it in the add man page...	20:43
	← rlorandi left	20:43
doener	tote: hmm... what does "git --exec-path" say?	20:44
JasonWoof	gitte: no idea	20:44
	← ShadeHawk left	20:44
	→ cousine joined	20:44
	→ lorandi joined	20:44
doener	gitte, JasonWoof: 2006-10-30	20:44
tote	doener: /usr/local/libexec/git-core/	20:45
	← miriamg left	20:45
	← Toad left	20:45
gitte	JasonWoof: pity you did not even research that, otherwise I would have blasted you less for touting a _completely obsolete_ version of cURL.	20:45
doener	tote: so you have another git installed, presumably self-built	20:45
gitte	doener: thanks for doing JasonWoof's job.	20:45
tote	doener: hmm :/ gotta look into it	20:46
doener	tote: and that might be lacking the http-push support. I don't recall if fetch and push over http have different requirements	20:46
	tote: or maybe the version in /usr/local/ got installed after you did the clone	20:47
	unreal_ → unreal	20:47
tote	doener: any tip on have to remove the self-built version?	20:47
	← lorandi left	20:47
	→ lorandi joined	20:47
JasonWoof	I just figured if it was completely irrelevant, it wouldn't be in the manpage.	20:48
	→ rshann_ joined	20:50
doener	tote: http://git.pastebin.com/m340a1a74 -- that and the man pages	20:50
	tote: IIRC that's all that gets installed	20:51
tote	doener: ok, thanks :)	20:51
	← gh34 left	20:52
	→ Zenopus_ joined	20:52
	→ jmatthews joined	20:53
	← HG` left	20:54
	→ azumanga joined	20:59
	← Weaselweb left	20:59
azumanga	In a default git setup, is there anything that stops someone from deleting all the branches from a remote repository they have write access to?	20:59
alip	is it possible to do a rebase after merging from upstream several times?	21:00
DrNick	no	21:00
azumanga	While I'm sure I'm stating the obvious, isn't the fact that branches can be deleted and not recovered a fairly major hole in a version control system?	21:01
	→ keystr0k joined	21:02
azumanga	I'm genuinely interested what the reasoning was behind this, rather than versioning branches the same as files	21:03
keystr0k	I have a bare repository with all of the latest commits. What's the best way to reset my local working tree to a previous commit... and if I like the changes that I end up making, to it, commit them as a new commit into the bare repo?	21:03
gitte	keystr0k: clone the bare repo.	21:04
	← sfwc left	21:04
gitte	keystr0k: make the changes.	21:04
	keystr0k: commit them.	21:04
	keystr0k: push them.	21:05
	→ FreakGuard joined	21:05
keystr0k	gitte: and it won't overwrite any of the other commits ? do I checkout that specific commit number? Currently the bare repo and working tree are exactly the same... I was going to reset to the earlier commit, make my changes, etc... but that doesn't seem right... branches?	21:06
gitte	azumanga: if you like a system disallowing branches, and thereby forcing people to blow away repositories that they resurrect with _just_ the branches they like, just say so.	21:06
doener	azumanga: unless you have something like gitosis, write access may as well mean that the user can do "rm -rf repo"	21:06
	Zenopus_ → Zenopus	21:06
doener	azumanga: and if you don't trust your devs set receive.denyDeletes	21:06
azumanga	doener: Really?? I'd assumed, like svn, they couldn't delete history	21:07
gitte	keystr0k: you can do branches, of course, but if you want to keep the other commits, you will have to merge the branches and then push.	21:07
keystr0k	gitte: mmkay	21:07
azumanga	So, I should add some extra version control for my version control?	21:08
gitte	azumanga: you can use subversion. That would be a solution, no?	21:09
	→ hcl2 joined	21:09
gitte	azumanga: ask on #subversion, I am sure they'll agree.	21:09
hcl2	for git merge or pull the --no-commit option doesn't seem to do anything.	21:09
doener	azumanga: If you're scared of your devs you should probably rather hire new ones... But to protect from pure mistakes, what's the problem with receive.deny{Deletes,NonFastForwards}?	21:09
	← cousine left	21:09
azumanga	doener: I'm going to have a look at those	21:09
walken	azumanga: git beginner here. but, I'd assume you can clone the repo so you have a backup.	21:10
azumanga	But seriously, part of the point of version control, is to avoid people's mistakes and allow old history to be dug back up in the case of accidents	21:10
	→ riddochc joined	21:10
	← tbf left	21:10
azumanga	Perfect devs wouldn't need version control, they'd just edit the code live	21:10
gitte	azumanga: BS	21:10
	azumanga: Perfect devs still would need version control	21:11
	azumanga: obviously you haven't thought things through...	21:11
	← qrush left	21:12
	doener .oO( hm, live editing and interpreted code... a whole new dimension for race conditions )	21:12
gitte	azumanga: let me guess: you are a project manager, not a dev.	21:12
	← drdave left	21:12
azumanga	gitte: No	21:12
gitte	azumanga: with an MBA as background. Am I right? ;-)	21:12
riddochc	That assumes that version control is only for the purpose of knowing what's been changed.	21:12
azumanga	I can list some code I've written if you really want me to	21:12
gitte	azumanga: then how can you say that?	21:12
	← loincloth left	21:12
squentin_	azumanga: you can also use git where each dev has its own public repo, and you pull from them	21:13
	→ dmoerner joined	21:13
	← keystr0k left	21:13
azumanga	Well, I think (although I'm looking at how I'm going to work around it) that allowing branches to be deleted, never to be recovered, puts a serious hole in version control	21:14
gitte	squentin_: although to prevent "data loss" azumanga is so concerned about, you'd still have to "git fetch" from those pretty often.	21:14
	azumanga: didn't you say that already?	21:14
	azumanga: but you haven't offered any rationale for that.	21:14
	azumanga: in contrast, I can give you at least one _very_ _good_ reason why you should be able to delete branches.	21:15
	azumanga: trade secrets. You could spill them unintentionally.	21:15
azumanga	Do I need a rationale? My belief is that people make stupid mistakes, and someone might well at some point accidentally delete a branch which actually they, or someone else, still needed	21:15
gitte	azumanga: happened to me, I was told that I could commit the stuff, but then asked to delete it later.	21:16
	azumanga: obviously you do not need a rationale.	21:16
	azumanga: and obviously I wasted my time on you. Have a fun time.	21:16
azumanga	The same reason we back up users's home directories, because one day someone is going to accidentally rm -rf in the wrong directory, and bang, home directory gone.	21:16
	→ gerberb joined	21:16
riddochc	deleting a branch in git should only result in a loss of data if nothing else refers to the same commit, and you've got reflogs off, and... probably a few other things, too.	21:16
doener	azumanga: and you exclude your code repositories from the backup exactly why?	21:17
Ilari	azumanga: If someone else needed it, its likely has remote mirrors (git fetch won't delete those without mirror mode), or better yet, has local working branch based on it.	21:17
	azumanga: And IIRC, mirror mode is pretty much for bare repos only.	21:17
azumanga	We certainly don't exclude them, but (although I haven't tried) splicing a deleted branch from the backup back into the git server might prove fun	21:17
doener	azumanga: put the backed up repo somewhere, push, done	21:18
Ilari	azumanga: Or just take some clone and push its tracking branch.	21:18
	← rshann_ left	21:19
wshimmy	azumanga: for it to really be data loss, you need to delete it from the server. then everybody who has a clone of the repo has to delete the branch. and you have to wait for the reflog to expire or for a gc on the server.	21:19
azumanga	Anyway, it seems the option I would like, which is that branches are not actually deleted, but versioned and recoverable, like svn, doesn't exist in git. I'll investigate the various options you have suggested, in particular recieve.denyDeletes looks very promising	21:19
wshimmy	azumanga: if somebody does a git push origin :master	21:19
	azumanga: i would imagine they'd realize right away	21:19
azumanga	wshimmy: The problem is that maybe you don't realise you wanted that branch for months	21:19
wshimmy	azumanga: and then its just a git push origin master away from being fixed	21:20
	← cemerick left	21:20
wshimmy	azumanga: and there isn't any clone anywhere with those commits?	21:20
	→ sfwc joined	21:21
azumanga	wshimmy: Well, we are working in hypotheticals here.. but I would expect there to be, except of course if there was a backup.	21:21
	This is just a feature of svn which I'm, perhaps unnecessarily, afraid of loosing when moving to git.	21:21
Ilari	azumanga: If you want on-line "second copy" repo, just periodically fetch the other repo?	21:21
wshimmy	azumanga: im just saying, if that's the case, i think your workflow has bigger problems than git allowing you to manage your repo yourself	21:22
azumanga	wshimmy: You really can't imagine someone might do away with a branch, thinking it was uninteresting, then months later wanting to query something about it?	21:22
wshimmy	gitte: sorry i disappeared earlier. you still interested in seeing diff engine code?	21:22
doener	azumanga: why does a branch that nobody wants end up in a "central" repo anyway?	21:22
	← Fullmoon left	21:23
walken	azumanga, I suspect the idea is that devs will create and delete branches, nilly-willy, on their own cloned repos.	21:23
wshimmy	azumanga: i cant imagine someone doing away with a branch on the central server that hasn't been merged or tracked elsewhere	21:23
walken	so you want branch creation to be easy because people will do it a lot - but they shouldn't have to do it much at all on the central repo.	21:23
azumanga	doener: Well, I'm imagining that at one point lots of people were interested in it, perhaps preparing for a new release or feature or some such.	21:23
	← codebrulee left	21:23
	← jrmuizel left	21:24
	→ gberdyshev joined	21:24
gitte	wshimmy: sorry, I am a bit too annoyed with clueless persons, have no mind for the diff engine.	21:24
	→ okram joined	21:24
wshimmy	gitte: heh, no problem	21:25
azumanga	OK, a different concrete comment. Do you think (I may e-mail the git mailing list anyway) an option in git which moved all deleted branches into some kind of "attic", from which they could be recovered for all time, would have any chance of acceptance? Obviously such an option would not be the default.	21:25
	← azol left	21:26
tote	how can i setup gitweb to have a little git link for every project like on http://git.kernel.org/ ?	21:26
Ilari	azumanga: As cronjob command, something like 'git fetch <repo-URL> refs/:refs/remotes/repoX/`date -u +%Y%m%dT%H%M%SZ`/' run from dedicated bare repo?	21:26
riddochc	azumanga: rather unlikely	21:26
wshimmy	azumanga: probably not, for the reason gitte gave earlier about accidentally pushing sensitive information	21:26
doener	azumanga: dunno... deleting branches in a remote repository is probably one of the less used commands	21:26
	s/less/least/ even	21:27
Ilari	azumanga: Can be done using hooks.	21:27
	→ memiux joined	21:27
azumanga	llari: Ah, thanks, that looks like a good idea. I'll give that a try	21:27
gitte	wshimmy: I would have more reasons, too, but I have better things to do than to waste my time, thankyouverymuch.	21:28
	← kwatz left	21:28
Ilari	azumanga: Branch deletes are visible to *-receive hooks as ordinary update instructions, with new sha values being all zeroes (40 of them).	21:28
	gitte whistles "Do I need a rationale?" No, obviously not. D'oh!	21:28
azumanga	gitte: Do you really think I haven't provided a rationale?	21:28
walken	hmmmm. I have a question too. been wondering what .git/branches is for (all my branches seem to end up in .git/refs/heads instead). And in particular, why git-new-workdir does not symlink .git/branches.	21:28
azumanga	Anyway, you aren't being helpful, so it doesn't really matter	21:29
Ilari	walken: Its obsolete.	21:29
	← bgerber left	21:29
walken	Ilari, thanks, that's what I thought but I wasn't sure.	21:29
gitte	azumanga: you do _not_ have any rationale, you have _not_ thought things through, otherwise you would _not_ insist on that _stupid_ insistence that allowing to delete branches would be a hole in a VCS.	21:29
azumanga	Thanks doener, llari, I have some good things to look at now. I will report back if it works smoothly, as other people might want it.	21:30
gitte	azumanga: it is _completely_ moronic to assume so, for the same reason it would be _completely_ stupid to disallow deleting files.	21:30
azumanga	gitte: Deleting files is completely different to deleting branches in git	21:30
gitte	azumanga: and now think things through and don't accuse people who did as being unhelpful.	21:30
Ilari	walken: Hasn't been used since at least 1.5.0.	21:31
gitte	azumanga: you can also go to #subversion, as far as I am concerned, just stop wasting time on #git. You have a brain, use it before asking questions.	21:31
azumanga	gitte: because if I delete a file, I can get it back, for all time. Surely you wouldn't suggest that deleted files can't be recovered, because why would anyone delete a file if they might want it back later?	21:31
	→ Gnutoo joined	21:31
gitte	azumanga: "Deleting files is different from deleting branches".. pfff!	21:31
	azumanga: I say "pfff!"	21:31
	azumanga: OF COURSE you need to be able to delete files _AND LIKEWISE_ branches.	21:32
	azumanga: and you cannot get back a deleted file if you did your job properly.	21:32
azumanga	gitte: Yes, but part of the point of a VCS is that you can rewind time, and recover previously deleted files	21:32
gitte	azumanga: IN CONTRAST you can get back a branch in git.	21:32
Gnutoo	hello, I don't know why but I've a file that I don't want included and that is included(like when you include files with git add)...how do I get rid of it(I've done git commit but I've not pushed yet...)?	21:32
gitte	azumanga: so _STOP WASTING TIME ON THIS CHANNEL_	21:32
	→ zedr0n joined	21:33
bleything	gitte: lay off the caps and underscores. if you feel like your time is being wasted, go do something else.	21:33
gitte	Gnutoo: you cannot, according to this azumanga guy here in this channel who wastes all of our time.	21:33
	→ qrush joined	21:33
	← zedr0n left	21:33
Ilari	Gnutoo: rebase -i ?	21:33
wshimmy	Gnutoo: you can git reset it and then git commit --amend if it's the latest commit	21:33
gitte	bleything: I am here for some other reason than to hear rationale-less ranting.	21:33
Gnutoo	it's the last commit	21:33
	I've to do only one commit	21:34
Ilari	Gnutoo: If there's only one, 'git rm --cached' it and then 'git commit --amend'.	21:34
Gnutoo	thanks a lot	21:34
Ilari	Gnutoo: Also works if its only in latest commit (and latest commit hasn't been pushed out).	21:34
Gnutoo	I didn't push yet...I always review my commits before pushing	21:35
Ilari	Gnutoo: Good idea.	21:35
riddochc	Gnutoo: You might want to look at git gui - makes editing previous commits, and knowing what you're committing, a lot easier.	21:36
gitte	Gnutoo: which commit is it? The latest one?	21:36
Gnutoo	that's what I use	21:36
	yes the lastest one	21:36
	→ bremner_ joined	21:36
gitte	Gnutoo: Ilari's advice was very good: rm --cached the file.	21:36
tote	how can i setup gitweb to have a standard clone url for every project it serves?	21:36
leo2007	do I need to treat binary files such as .png or .pdf files carefully?	21:37
gitte	Gnutoo: then you need to "git commit --amend", which basically redoes the latest commit according to the current staging area.	21:37
	← radarek left	21:37
gitte	Gnutoo: but verify that everything worked as you wanted it to, by "git show HEAD" after the commit.	21:37
	→ killerchicken__ joined	21:37
Ilari	leo2007: If you want, you can mark them as binary. But usually not needed. Especially if on unix.	21:37
Gnutoo	thanks	21:37
riddochc	Gnutoo: in the 'commit' menu of git gui, there's "amend last commit"	21:38
Gnutoo	git rm deleted the file not the change...I'll reset what I've done and do it again	21:38
	hopefully I saved the commit message	21:38
azumanga	doener: Thanks, I've tested receive.denyDeletes on our public repo, and that does exactly what I want.	21:38
Ilari	Gnutoo: Resurrect the file from latest revision?	21:38
Gnutoo	good idea	21:39
Ilari	Gnutoo: And then rm --cached it.	21:39
	→ okram_ joined	21:39
Gnutoo	git checkout -- file if I remember	21:39
	okram_ → mfontani	21:39
Ilari	Gnutoo: Actually, you need 'checkout HEAD -- file' since you want it from latest revision.	21:39
	→ hoohah joined	21:39
Ilari	Gnutoo: 'checkout -- file' grabs it from the index (and git rm likely nuked that).	21:40
	← jayallen left	21:40
Gnutoo	ok thanks	21:40
Ilari	leo2007: Except that one needs to be careful when modifying them, since they can't be merged.	21:40
hoohah	hello. how git handle symlinks in a repo? when i clone it, will the symlink remain intact?	21:40
	wshimmy points to his previous suggestion of git reset :)	21:40
Ilari	hoohah: Symlink targets are treated as opaque blobs.	21:40
leo2007	Ilari: yes, that's my concern them.	21:40
	Ilari: how to modify them	21:40
hoohah	Ilari: what you mean?	21:40
Ilari	hoohah: Symlinks can be added. They are stored as symlinks, but no further interpretation is done.	21:41
	← bremner_ left	21:41
	→ solydzajs joined	21:41
Ilari	leo2007: You get nasty merge if you modify binary file in different ways on two branches and then merge those two branches.	21:42
gitte	Gnutoo: you can always see what the new commit would look like with "git diff HEAD^"	21:42
Gnutoo	ok thanks	21:42
hoohah	Ilari: ah good. that is what i after	21:43
	i'm	21:43
gitte	doener: seems that I had to shout before your answer was finally accepted anyway.	21:43
leo2007	how to better handle them, Ilari	21:43
Ilari	leo2007: PDFs and pngs are essentially unmergeable. There's essentially no way around it. If the PDF is generated from other files, maybe its better not to put the PDF there?	21:44
pasky	why are they unmergeable?	21:45
	just choose one version	21:45
	esp. for png that's usually just fine	21:45
	→ matthewmcculloug joined	21:46
	← sakyamuni left	21:46
	← killerchicken_ left	21:46
leo2007	does that mean if I have many branches I'd be better modify them in just one branch?	21:46
	→ WALoeIII joined	21:48
	← feanil left	21:48
	→ qrush_ joined	21:48
	← leanucci left	21:48
	→ jayallen joined	21:49
Ilari	leo2007: Actually, maybe just propagating changes to all branches as soon as possible is enough. But problem turns bit nastier this is multiple-dev project.	21:50
	← mozinator left	21:51
	← qrush left	21:51
leo2007	Ilari: if I modify them on one branch, would that prevent me from all these problems?	21:52
	← john3909 left	21:52
	→ kallepersson joined	21:52
	→ Minimiscience joined	21:52
leo2007	I'd like to stick to one practice to avoid problems. I'm working on a latex project so I'd need to include some images in .png and .pdf format	21:52
	← solydzajs left	21:52
Aides	convert them to tikz ;)	21:53
	← okram left	21:53
azumanga	Are these .pngs and .pdfs auto generated?	21:53
	I assume the .pdfs are the paper itself, latexed?	21:53
kallepersson	Hi! A very basic question: If I pulled the code 10 minutes ago, my friend comitted a change 5 minutes ago (which means my code doesn't have the changes he made), how would I to do push my code without overwriting his changes?	21:54
azumanga	I've found it best to not store the .pdfs of the paper, assuming it is easy to regenerate them from the late source, to avoid conflicts	21:54
Ilari	kallepersson: pull again?	21:54
	← suman_g left	21:54
kallepersson	Ilari: wouldn't that overwrite my changes?	21:54
Ilari	kallepersson: Commit first.	21:54
leo2007	azumanga: no	21:54
kallepersson	Oh!	21:54
squentin_	leo2007: I think what you should do is just never modify them, just use a new filename if you want to modify one	21:55
leo2007	azumanga: no the .pdf are images exported from other programs	21:55
	← mfontani left	21:55
Ilari	kallepersson: And to push the changes, you would have needed to commit anyway.	21:55
kallepersson	True	21:55
leo2007	squentin_: that's very problematic, that means I also need to modify the .tex source.	21:55
Minimiscience	If a subdirectory of a Git repository evolves into its own project, what would be the best way to give it its own repository with history?	21:55
	Should I just clone the repo and delete everything outside of the directory, optionally rebasing away the history of the other files, or is there a cleaner way to do it?	21:55
azumanga	leo2007: Ah, sorry. in that case, things do get more complicated. Obviously, if you don't put different copies in different branches, everything will just merge fine. It's only if you change the picture in multiple branches, in which case you will need to think about which one you want to merge anyway.	21:55
Ilari	leo2007: What about after doing each modification, merge/rebase all affected branches immediately before further changes? That should ensure no conflicts on those files.	21:56
Aides	Minimiscience, there is git filter-branch example on this	21:56
squentin_	leo2007: keeping track of changes in text file is what git knows how to do	21:56
Aides	Minimiscience, you're looking for --subdirectory-filter option	21:56
gitte	Minimiscience: --subdirectory-filter	21:56
Minimiscience	I did not know that; it looks like it should work.	21:56
	← name left	21:57
	→ clamothe joined	21:57
leo2007	squentin_: I'm not questioning that. Just that this approach is not efficient. it requires unnecessary changes to the source.	21:57
Ilari	leo2007: If all non-mergeable files really have linear history (per-file history) just keep with seperate filenames?	21:58
	← icwiener_ left	21:58
	← walken left	21:58
kallepersson	Thanks	21:58
	← kallepersson left	21:58
	← clamothe left	21:58
squentin_	leo2007: well you can modify them, it's just that when you want to merge 2 branches with different version of them, you'll have to choose one	21:58
leo2007	and git knows that are binary	21:59
squentin_	yes	21:59
Ilari	leo2007: And since those files are autogenerated, presumably you could generate merge result somehow?	21:59
leo2007	ok that sounds good	22:00
	thank you all for clearing my concerns	22:01
Ilari	leo2007: Or just take one of the merged versions as the new version.	22:02
	leo2007: (git checkout --ours/--theirs -- file').	22:02
	← lightcap left	22:02
	← krh left	22:03
azumanga	Thanks for comments everyone. Goodnight	22:03
leo2007	Ilari: thanks, I'll keep that in mind. I haven't used the --ours/--theirs options before	22:03
	← azumanga left	22:04
	← Minimiscience left	22:04
Ilari	leo2007: If you have old version, you might need to upgrade git.	22:04
leo2007	Ilari: I'm using 1.6.2.4	22:05
	← Sho_ left	22:05
	→ solydzajs joined	22:07
	→ dmlloyd_ joined	22:08
Ilari	leo2007: Should be new enough for most features. Of course, 1.6.3.1 is out...	22:09
	← d0k left	22:09
leo2007	Ilari: thanks. I'll update it later on.	22:10
	→ SRabbelier joined	22:10
	← Tuomas left	22:10
	dmlloyd_ → dmlloyd	22:12
	← dwmw2 left	22:12
	← riddochc left	22:12
	← bcardarella left	22:13
	← ph^ left	22:13
	→ naeu joined	22:14
	→ ericindc joined	22:15
	← LiamH left	22:15
	← cmarcelo left	22:15
	gitte still tries to think about parent-less commits as ways to track large binary files	22:17
	→ azumanga joined	22:18
	← __iron left	22:19
wshimmy	gitte: where they would just be replaced rather than compared in a merge?	22:22
gitte	wshimmy: yes, the idea is to avoid having the whole complete history of the binary files.	22:23
	wshimmy: there was a talk about tracking huge binary files with Git at the GitTogether 2008.	22:23
	→ lightcap joined	22:24
wshimmy	gitte: whats the downside to the parent-less commits method?	22:24
	gitte: other than being unable to commit a binary and a source file in the same commit id imagine?	22:25
	→ sergiodj joined	22:26
	→ scarabx joined	22:26
azumanga	I have pulled a update to a repository. Someone else as added various tags, which I can see with gitweb, but "git pull" seems to have only pulled about a third of the tags to my computer. Any ideas why, and how I get the rest?	22:26
	← dwave left	22:27
	← bdiego left	22:27
	← lorandi left	22:28
wshimmy	azumanga: git pull --tags	22:28
	azumanga: see git help push, it explains what gets pulled and what not	22:28
	err, obviously git help pull	22:28
gitte	wshimmy: the downside is that merging is not possible. That would need the history.	22:29
azumanga	Ah, thanks, I missed that entirely!	22:29
	Wonder how they ended up historyless	22:29
	← bx2 left	22:29
azumanga	I believe they came from a svn import, so possibly something weird happened there	22:30
wshimmy	gitte: but this is for files that can't be merged anyways?	22:30
azumanga	Yes, they were 'svn tags' that someone commited changes to (as a tag in svn is just a directory you can commit to), which, not unsuprisingly, confused the conversion process (I think)	22:31
	← giallu left	22:33
	← qrush_ left	22:33
	← ciskje left	22:33
	→ qrush joined	22:34
	← WinterWeaver left	22:34
	← psoo left	22:34
	→ mitkok joined	22:34
	→ lourense joined	22:34
	← naeu left	22:36
	← SRabbelier left	22:38
	→ SRabbelier joined	22:38
hoohah	hello. how do i get the basename of the git repo I'm currently in?	22:39
gitte	wshimmy: binary files can be merged if you have an appropriate merge helper.	22:40
hoohah	also, if i'm in separate non-git dir, and I want to perform a git ls-tree on a git repo dir, how do i do so?	22:40
wshimmy	gitte: ah, right	22:40
hoohah	(for purposes of scripting)	22:40
	← kraymer left	22:41
	→ sakyamuni joined	22:41
gitte	wshimmy: we would have a GSoC project to implement domain-specific merge helpers, if the student hadn't been led down the wrong path by someone on the mailing list.	22:42
	← flazz left	22:43
	→ leanucci joined	22:43
SRabbelier	gitte: why so sour?	22:43
wshimmy	gitte: how manual is this merge process with a helper?	22:44
	← Gnutoo left	22:44
gitte	SRabbelier: I could have used those merge helpers myself.	22:44
	wshimmy: depends on the merge helper.	22:44
wshimmy	gitte: i would think that in a lot of cases they would be mainly just 'converting' the binary format into something that could be compared or merged by hand	22:47
	but perhaps im just unimaginative ;)	22:47
	← khmarbaise left	22:47
gitte	wshimmy: well, in the most general case, the merge _must_ be manual.	22:47
	wshimmy: however, very often, a clever program can help to resolve the common conflicts almost automatically.	22:48
wshimmy	gitte: but for a completely manual case, you don't need the history anyways since you're not doing a three-way merge, right?	22:48
	gitte: that seems very impressive. i have not thought about this in the slightest, clearly. :P	22:48
	← wmoxam left	22:49
gitte	wshimmy: a merge is only necessary when there are two different lines of development.	22:49
	→ qrush_ joined	22:50
gitte	wshimmy: so it would very much be a 3way merge.	22:50
	wshimmy: that's what our default merge does with text files.	22:50
	← hpoydar left	22:51
	← javatexan left	22:52
wshimmy	...right, that was dumb	22:52
	← qrush left	22:52
	→ richcollins joined	22:53
	→ jayallen_ joined	22:53
richcollins	How do I handle this? git pull steve master:new-db	22:53
	! [rejected] master -> new-db (non fast forward)	22:53
	← azumanga left	22:54
	← sunoano left	22:54
richcollins	I would expect a merge conflict message if that were a problem	22:54
Aides	git checkout new-db; git merge steve/master	22:54
	not neccessarily	22:55
	but probably	22:55
	it means that a merge is required, it doesn't mean that there will be conflicts	22:55
	→ fooljay_ joined	22:55
Ilari	richcollins: That syntax looks odd.	22:56
richcollins	Aides: That is fine, why doesn't it go ahead with the merge and tell me if there are conflicts	22:56
	Ilari: It might be wrong	22:56
	← cannonball left	22:56
Ilari	richcollins: Merge requires working tree.	22:56
Aides	probably because its not checked out	22:56
richcollins	I am pulling from the remote named steve and attempting to merge the brand master on steve with the branch new-db locally	22:56
	I did check it out	22:56
wshimmy	gitte: but i still don't see why that is merge-prohibitive	22:56
richcollins	* new-db	22:56
wshimmy	gitte: if i understand the 3-way merge git uses correctly (and i hope i do)...take the two differing parents (but none of their histories) and do its merge thing, right?	22:57
richcollins	Ilari: Is that the correct syntax for the operation I described?	22:58
gitte	wshimmy: actually, take the base and both new versions, and present only the differing developments.	22:58
Ilari	wshimmy: Having common ancestor version available lets to resolve many kinds of conflicts.	22:58
Aides	richcollins, you need to run git merge steve/master now	22:59
wshimmy	gitte: so if i want to merge my parent-less binary commit abcde with your parent-less binary commit 12345, it would need only abcde and 12345 to reach the final version	22:59
Ilari	richcollins: Usually pull is invoked with designation of remote repo and branch on remote repo.	22:59
	→ agile joined	22:59
Ilari	richcollins: So usually something like 'git pull foo bar'.	22:59
wshimmy	gitte, Ilari: well, im speaking only in the very manual merge resolution case	22:59
gitte	wshimmy: one thing would be to help merging OpenOffice files by presenting the user with a way to see and resolve the conflicting changes.	22:59
richcollins	Ilari: How do I know which local branch it will be merged with in that case?	22:59
	I added :new-db to explicitly merge with new-db	23:00
gitte	wshimmy: but the parent-less merge (i.e. a 2way merge) would not be able to know which side to take.	23:00
Ilari	richcollins: Current branch.	23:00
	richcollins: And if you want to try to fast-forward some local branch, use fetch, not pull.	23:00
richcollins	Ilari: OK I just wanted to be explicit to avoid merging with the wrong branch	23:00
gitte	wshimmy: in a way, a 2way merge would be possible, by assuming an _empty_ base.	23:00
wshimmy	gitte: but you would be presented all of the conflicting changes by the merge driver and choose it yourself	23:00
richcollins	Ilari: I wanted to fetch and merge	23:01
	it isn't a fast forward	23:01
	almostautomated → almostautomated\|	23:01
gitte	wshimmy: that's also what we do with text files, when merging two branches where the conflicting file did not exist in the merge base.	23:01
	wshimmy: yes, something like that would be needed.	23:01
wshimmy	gitte: seems better than nothing (or are there already binary merge drivers im ignorant of that work just fine atm?)	23:02
gitte	wshimmy: (The same as with lacking file in merge base happens when merging independent branches, i.e. branches without common history)	23:02
	wshimmy: AFAIR there are some very special merge helpers that haven't been published.	23:02
	wshimmy: I started something for .odt files, but it never was stable enough to be used properly.	23:03
wshimmy	gitte: those not withstanding, the current procedure for binaries is just 'replace blob'?	23:03
	← juanjoc left	23:04
gitte	wshimmy: no... the current procedure is: here are the two versions, <file>~HEAD and <file>~deadbeef, choose your poison.	23:04
wshimmy	gitte: er, right	23:05
Ilari	richcollins: You were probably lucky it hit non-ff. Because pull syntax you used _still_ merges with current branch.	23:05
richcollins	I thought the refspec syntax was src:dst	23:06
	which dst is it referring to?	23:06
wshimmy	gitte: so i guess i would just see the merge drivers as "choose your poison...carefully"	23:06
	← static^ left	23:06
Ilari	richcollins: It is. pull passes the refspec down to fetch and then merges what was fetched into current branch.	23:06
gitte	wshimmy: heh, _much_ more carefully ;-)	23:06
	← doener left	23:07
gitte	wshimmy: same as presenting conflict markers in text files is way superior than just giving the user both versions of the file.	23:07
wshimmy	gitte: so it wouldn't allow for the automagic. but i would still think of it as a pretty big improvement	23:07
	→ bcardarella joined	23:07
richcollins	So the refspec refers to the remote but not the local?	23:07
altrux	i need architectural help.	23:07
richcollins	And the : syntax doesn't apply to fetch	23:08
altrux	i have a set of projects, which all share code. and i want to factor a large part of that shared code out	23:08
gitte	wshimmy: oh, but it could allow for automagic. Think .zip files: they can be merged easily if only timestamps differ.	23:08
wshimmy	gitte: but without parents you still don't know which one to take, or?	23:08
altrux	and put it i its own repo. the problem is it isn't in a single directory, its spread, willy, nilly, across a few directories, mixed into the other projects. i can not change this, due to the framework.	23:08
Ilari	richcollins: And push and fetch interpret refspec 'master' quite differently.	23:08
wshimmy	gitte: er, a mutual ancestor	23:08
gitte	wshimmy: I know what you meant.	23:09
richcollins	Ilari: shocking	23:09
	ok I think I've got it ... thanks for your help	23:09
	← jayallen left	23:09
gitte	wshimmy: without merge base, you do have to assume that the common ancestor is empty.	23:09
	← MisterN left	23:09
altrux	is there a way to have submodules, or some such, cover just some files, and not any others, and have the main project ignore those files, without creating some crazy ..gitignore of specific files and such?	23:09
	→ drdave joined	23:09
Ilari	richcollins: While push copies src to dest if there is no RHS, fetch fetches to FETCH_HEAD if there is no RHS.	23:10
gitte	wshimmy: so that the parts that are common between both versions of the file are recognized as identical changes, and therefore unambiguous.	23:10
	wshimmy: only that this might be very difficult with binary files.	23:10
	wshimmy: for example, if you want to merge bitmap images, a naive way would be to merge pixel-wise.	23:10
	→ aspotashev\|eeepc joined	23:11
	← jayallen_ left	23:11
drdave	Hi, I am trying to display the previous revision of a particular file:	23:12
	using the	23:12
	git show HEAD~1:relativePath syntax	23:12
	← bcardarella left	23:13
	→ context joined	23:13
drdave	The problem is if I commit a different file - then HEAD~1 is no longer the previous revision	23:13
Ilari	Simple bitmap merge algorithm: Given bitmaps A and B with ancestor C, compute pixel-channel-wise: X = C + (A - C) + (B - C). :-)	23:13
drdave	I need to use HEAD~2	23:13
Ilari	drdave: Then HEAD~2:path'?	23:14
drdave	how do I get the previous revision of a given file?	23:14
gitte	wshimmy: anyway, time to get some sleep. Thanks for the interesting conversation!	23:14
	← gitte left	23:14
	→ drudge joined	23:14
drdave	The nly way I can think of doing it at the moment is getting the log and moving down a couple of lines and extracting the hash...	23:15
	Must be an easier way?	23:15
	And faster :)	23:15
	→ _swien_ joined	23:15
_swien_	hi	23:16
	← aspotashev left	23:16
Ilari	drdave: The syntax of ':' operator is '<committsh>:<path-to-file>'. HEAD~2 is valid committish assuming 1st-parent of 1st-parent of current commit exists.	23:16
	← SRabbelier left	23:16
drdave	I can't use HEAD~2:path - as all I want is the previous text - and I don;t know how many other files have been changed.	23:17
	← solydzajs left	23:17
Ilari	drdave: '`git rev-list -1 HEAD -- file'^:file' or something like that?	23:18
drdave	AFAIK - the integer in HEAD~2 refers to the was the file was 2 commits ago - but 2 commits for any change / file	23:18
	← j416 left	23:18
drdave	I want 2 commits for that particular file	23:18
	Wow.... that one is new / makes no sense at all to me :)	23:19
	← fooljay_ left	23:19
drdave	I'll take a look at "rev-list"	23:20
	Can you translate?	23:20
	especially the last bit - -- file'^:file'	23:20
Ilari	drdave: `` is shell substitution operator. Shell runs the command inside and pastes its stdout over it and then runs the whole command.	23:20
drdave	k	23:21
Ilari	drdave: And that first one had one ' where there should have been `	23:21
drdave	I'm going to need to pass an actual path somewhere - can you give us an example with a path in it?	23:22
Ilari	drdave: For instance 'echo foo`echo bar`baz' prints 'foobarbaz'.	23:22
	← matthewmcculloug left	23:23
	← agile left	23:24
Ilari	drdave: git show `git rev-list -1 HEAD -- builtin-mv.c`^:builtin-mv.c	23:24
drdave	thanks.... will wrap my head round it :) Not quite got logic of shell substitution operator yet :)	23:25
Ilari	drdave: There's also another substitution operator: '$()'. That one can also be nested (and presumably has some other differences as well).	23:25
drdave	k	23:26
	← wshimmy left	23:26
drdave	How would git show `git rev-list -1 HEAD -- builtin-mv.c`^:builtin-mv.c be written long hand then	23:26
	→ solydzajs joined	23:26
	← dodo_the_last left	23:27
Ilari	Quite useful to have nested substitutions when playing with commit-tree...	23:27
drdave	true	23:27
	→ wshimmy joined	23:27
drdave	ah... think I'm getting it - its a sort of merge	23:27
	← joshthecoder left	23:28
wshimmy	arg	23:28
Ilari	drdave: On my git.git copy (bit out of date), the rev-list resolves to 5aed3c6ab834367292bd7a9b0894177a07afd910, so the shell runs 'git show 5aed3c6ab834367292bd7a9b0894177a07afd910^:builtin-mv.c', which is equivalent to 'git show c8ba6b1b199bf00f227e8fdad4adb5a2147c8160:builtin-mv.c', which resolves to 'git show bce9959293e30925c4b16c40ac33a3f2e0474e30'	23:29
Gitbot	[git 5aed3c6ab]: http://tinyurl.com/q8ac93 -- builtin-mv.c: check for unversionned files before looking at the destination.	23:29
drdave	everything between `xxxxxx` gets run and subsitituted - then the outer shell command is run - something like that ?	23:29
	← fridim_ left	23:29
Ilari	drdave: Yes.	23:29
drdave	Great	23:29
	← moonflux left	23:30
drdave	thanks a bundle	23:30
	→ joshthecoder joined	23:30
	← altrux left	23:32
	← _swien_ left	23:34
	→ keystr0k joined	23:35
keystr0k	Git doesn't track empty directories (or so I've read). Does this mean it won't delete them if I go back to an earlier commit that doesn't have them?	23:36
mugwump	no, it doesn't mean that	23:37
	→ Fullmoon joined	23:38
	← mithro left	23:39
keystr0k	mugwump: darn.	23:39
mugwump	it won't delete them if there is content which is not tracked there though	23:39
	ie, it removes files it put there. if there is nothing left in a directory, the directory is removed	23:40
	→ cpg joined	23:40
	→ shame joined	23:40
	shame → agile	23:40
keystr0k	mugwump: so that kinda stinks... My web app has a bunch of empty folders that might be needed sometime... and I want to preserve them. So I have to add a hidden file to each or something?	23:40
	→ eross joined	23:40
	→ shruggar joined	23:40
mugwump	yes, eg .gitignore	23:41
keystr0k	mugwump: mmkay. thanks	23:41
eross	I'm used to the debian apt-get, what is git?	23:42
	→ SRabbelier joined	23:42
cxreg	"I'm used to food, what is gasoline?"	23:43
mugwump	eross: git-core	23:43
	see also apt-cache rdepends git-core	23:43
eross	git-core is already built-in?	23:43
	← SRabbelier left	23:44
mugwump	it's already packaged for debian, yes	23:44
	if you're on debian stable then try the versions on backports.org	23:44
eross	ok just one thing i read touted in a linux journal about the latest thing	23:44
	→ SRabbelier joined	23:44
	→ Voight-Kampff joined	23:46
	← eross left	23:47
	← hoohah left	23:48
	← Alinermedia left	23:48
	← plasticine left	23:49
	→ cytrinox joined	23:50
	the_zeno → zeno^n8	23:51
	→ plasticine joined	23:51
	→ mike_m joined	23:51
	← mitkok left	23:52
	← richcollins left	23:53
	← Fullmoon left	23:54
	→ cytrinox_ joined	23:56

Logs Search ←Prev date Next date→

Channels Documentation