IRCloggy #git 2010-03-07

Provider of IRC logs since 2005.

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

	2010-03-07
	→ PHLAK joined	00:00
	← hyperair left	00:01
przemoc	frogonwheels: cat file \| tr -d \\r \| git am	00:01
	← PHLAK left	00:01
	→ PHLAK joined	00:01
	← rettub left	00:01
przemoc	frogonwheels: sorry, you want stored with crlf	00:02
	so my ex is the opposite	00:02
	← airfoil left	00:02
	→ airfoil joined	00:02
	→ hakunin joined	00:03
przemoc	frogonwheels: your repo has CRLF, but your patch has CRLF or LF?	00:03
	← Paraselene_ left	00:04
	→ rettub joined	00:04
frogonwheels	when the patch was generated, it was lf. but I used vim to crlf it	00:04
	er them.	00:05
	przemoc: so CRLF	00:05
	← parasti left	00:06
	→ Mike_lifeguard joined	00:06
przemoc	frogonwheels: so you solved your problem? BTW you can use sed to do it. sed -i 's/$/\r/' files	00:06
	(I mean instead of vim)	00:06
frogonwheels	przemoc: no. when it patched, it patched with LF only	00:06
	.. so the patched bits have LF only	00:06
	(there were whitespace warnings)	00:07
	przemoc: in vim: :argdo set ff=unix ;)	00:07
	→ seathief joined	00:08
frogonwheels	.even though the patch-files themselves had been converted to CRLF	00:08
	(They came from an autocrlf=yes repository).	00:09
	← git__ left	00:09
	→ kelvie joined	00:09
	→ Paraselene_ joined	00:09
	← jasonn1234 left	00:10
przemoc	frogonwheels: it's by design	00:10
	← infogulch1 left	00:10
	← adac left	00:11
frogonwheels	przemoc: Because they came from autocrlf=yes repo? or because I have autocrlf=no ?	00:11
	is there a way around it?	00:12
przemoc	frogonwheels: http://kerneltrap.org/mailarchive/git/2006/5/26/205677/thread	00:14
	← tectonic0 left	00:15
przemoc	frogonwheels: change autocrlf just for this git am	00:17
mr_danie1	what is the current working directory of the post-update hook, which lies in GIT_DIR/hooks/post-update ?	00:18
	← apol left	00:18
frogonwheels	przemoc: trying.	00:18
	przemoc: hmm.. but won't that mean any files will be committed with only lfs?	00:18
	I don't see how it helps?	00:18
	→ hgimenez joined	00:19
mr_danie1	and why can I find to many example scripts which begin with a unset GIT_DIR ?	00:20
dsop	whats the best way to reset a working dir to the current head if there are modifications?	00:20
wereHamster	mr_danie1: my guess that the working directory is either the root of the working tree or the .git directory	00:20
	dsop: reset --hard	00:21
	→ pantsman joined	00:21
dsop	hm why did I think that i need to specify somethig for git reset	00:21
mr_danie1	does 'git checkout -f' has the same effect as 'reset --hard'?	00:22
wereHamster	yes	00:22
	→ co_sndr1r1an joined	00:22
wereHamster	well, not quite. reset --hard has an implicit HEAD, checkout -f uses the index	00:22
przemoc	frogonwheels: you're right, I'm bumbling	00:23
frogonwheels	przemoc: I had found that particular email before, btw,	00:23
mr_danie1	understoos, thanks wereHamster	00:23
frogonwheels	przemoc: I'd oped that a few years might have seen a solution. I'd settle for a command-line switch :\|	00:23
przemoc	but CRLF is evil and should be killed off	00:24
frogonwheels	przemoc: unfortunately CRLF being the wrong thing 100% of the time in unix land, it's the similar in DOS land. LF only is the wrong thing.. well 99% of the time.	00:24
	przemoc: sure, it's redundent and painful. but unlikely to be killed off any time soon :\|	00:25
przemoc	frogonwheels: if you have right tools, they can correctly work with LF alone.	00:25
	← co_sndr1r1an left	00:25
mr_danie1	przemoc: 100% agree; 'killing it' would make life much easier	00:25
	← GarethAdams\|Home left	00:26
frogonwheels	przemoc: Well, I'm programming in Delphi, and my team would be rather annoyed if I decided to strip CRs.	00:26
przemoc	obviously m$ notepad doesn't support lack of CR...	00:26
	→ git__ joined	00:26
przemoc	frogonwheels: I understand your PoV	00:27
frogonwheels	przemoc: it's tough enough convinving them that git is the right choice when it soo obviously is for many, many other easons	00:27
	→ aef joined	00:27
	→ SteveDekorte2 joined	00:27
SteveDekorte2	Is there any way to just test if you have push permissions on a git remote branch?	00:27
aef	hi, why is there no way to sign each commit via gpg, why only tags? (or did i miss this?)	00:27
wereHamster	because it does not make sense to sign commits	00:28
przemoc	frogonwheels: so maybe (I'll be banished for sure) mercurial? though I don't know is it any better w.r.t. CRLF...	00:28
	→ patthoyts joined	00:28
	← patthoyts left	00:28
	→ patthoyts joined	00:28
przemoc	frogonwheels: have you tried raw patch command?	00:28
aef	so when someone forks my rep, and then modifies my commits, someone else might think i did the changes and has no way to be sure	00:29
wereHamster	aef: that's why you tag your releases and sign the tags with gpg	00:29
frogonwheels	przemoc: looks like I've got 54 patches to go through and manually patch. woopee	00:29
	← froschi left	00:29
aef	wereHamster: but i'm not talking about releases	00:29
	→ froschi joined	00:29
aef	i'm talking about development, like it is in github	00:29
	→ GarethAdams\|Home joined	00:29
przemoc	frogonwheels: it can be easily scripted, just test whether patch does the right thing	00:30
aef	people fork each other's reps all the time and the evil guys could fuck all that up	00:30
wereHamster	aef: true, but the evil guy would also have to convince everybody to pull from their repo	00:31
frogonwheels	przemoc: It probably depends on which 'patch' you use - I'm sure I can find one which does...	00:31
przemoc	I mean just 'patch'	00:31
aef	wereHamster: yes, easy, just be the one with the latest commit on some well known github rep and people start pulling from you	00:31
	they even write you many pull requests	00:32
	← anxiolytic left	00:32
frogonwheels	przemoc: using a 'unxutils	00:32
	' patch works fine.	00:32
	← bambule left	00:32
	← dgl left	00:33
wereHamster	aef: and how would gpg signed commits prevent that? You would have to give every user your gpg fingerprint, otherwise they wouldn't have a way to verify that the commits are yours.	00:34
	← JonathanT left	00:34
wereHamster	aef: and I don't see any difference between giving the users your fingerprint and giving them the url to your own repo and tell them to not pull from anyone else	00:34
unregistered	how do I get my local repo to forget about a remote branch?	00:34
aef	yes, as it is with tag signing and downloading things that are signed from the internet	00:34
wereHamster	aef: after all, users can just as easily ignore both and get the 'bad' commits anyway	00:35
aef	the idiots will always be fucked, but with signed commits you can offer people a difference	00:35
wereHamster	.. as can you by giving them the url to your very own repo	00:35
aef	apart of this, in github, i don't tell anyone to get the files explicitly, they just pull	00:35
	← DaveIngram left	00:37
aef	cryptography is the only way i know, which is able to solve the problem, not that easy, but at least possible	00:37
	→ simplechat joined	00:39
	← kelvie left	00:39
	→ kelvie joined	00:40
	← pantsman left	00:40
offby1	unregistered: you can remove the entire remote, but I don't think you can remove a specific remote branch.	00:42
	unregistered: but I doubt it's hurting you any	00:42
	← froschi left	00:44
	→ reprore_ joined	00:45
git__	i'm trying to create a central system configuration using git	00:46
	can one checkout HEAD from a bare repo without its commit history?	00:47
	← erikvold left	00:47
	← altogether left	00:48
offby1	sort of	00:49
	you can do "git archive" which will create a tar or zipfile.	00:49
	I wouldn't call that "checking out", but perhaps I'm just being pedantic	00:49
	← mebus left	00:50
	→ archis joined	00:51
	← wwwald left	00:51
	← Blackshark left	00:52
	← archis left	00:52
	← jakob left	00:52
	← brizly left	00:52
	→ alexmchale joined	00:52
wereHamster	git__: man git-archive, pipe that directly to tar -xf - -C ...	00:53
Gitbot	git__: the git-archive manpage can be found at http://git.or.cz/man/git-archive	00:53
	← a3li left	00:53
	→ a3li joined	00:53
	← freedrull left	00:54
	← kelvie left	00:55
	← mr_danie1 left	00:55
	→ aresnick joined	00:55
	→ kelvie joined	00:56
unregistered	offby1: the thing is, i'm seeing branches that were deleted on the remote	00:56
	→ eMBee joined	00:56
wereHamster	unregistered: git remote prune (deletes all remote branches which have been deleted in the remote)	00:56
unregistered	ahah. thanks	00:56
	← mbroeker left	00:59
	← enigmus left	01:00
przemoc	good night	01:01
	przemoc is afk	01:02
	→ mr_daniel joined	01:02
	→ offby1` joined	01:03
	← imajes left	01:04
	→ imajes joined	01:05
	← offby1 left	01:06
	← Theravadan left	01:07
	← pingveno left	01:09
	← PHLAK left	01:09
	← rettub left	01:10
	→ fool_ joined	01:11
	← fool_ left	01:11
	→ fool_ joined	01:11
	← seathief left	01:12
	→ jakob joined	01:13
	→ rettub joined	01:13
	← richcollins left	01:13
	← fool__ left	01:14
	→ bburhans joined	01:15
	← corni left	01:15
	→ bond` joined	01:16
	← mr_daniel left	01:17
	← patthoyts left	01:17
	→ mebus joined	01:17
	→ anjen joined	01:18
	→ mr_daniel joined	01:18
	→ infogulch joined	01:19
	→ Tureba joined	01:22
	← jelmer left	01:22
	→ phaedra joined	01:22
	→ jelmer joined	01:25
	Mike_lifeguard → Mike\|\|gone	01:27
	← airfoil left	01:29
	→ ijuma__ joined	01:31
	← kelvie left	01:32
	← ijuma left	01:32
	→ aleray joined	01:33
	← dcampano left	01:35
	→ dcampano joined	01:36
	→ anxiolytic joined	01:36
	→ kelvie joined	01:37
	Mike\|\|gone → Mike_lifeguard	01:37
	← alexmchale left	01:38
frogonwheels	ok. I just found a post from linux about autocrlf=false .. Many people in many places say autocrlf=false should be best for windoze-only projects ..	01:43
	← douilspee left	01:43
frogonwheels	... and Linus disagrees - stating that so much of git support tools &c rely on LF as the line ending - and that the only reaso CRLF works a lot of the time is cause it includes LF.	01:44
	→ bronson joined	01:44
	→ parasti joined	01:44
frogonwheels	.. so I'd better change my options on my repo and re-export my repository. It only takes 12hours or so.	01:44
	→ douilspee joined	01:47
	← SteveDekorte2 left	01:49
	← Bass10 left	01:50
	← Paraselene_ left	01:51
cbreak	or...	01:51
	you just use LF in your windows tools too	01:52
	→ dreiss joined	01:52
	→ DaveIngram joined	01:53
	← DaveIngram left	01:54
	← aleray left	01:55
	→ toxx_ joined	01:57
	← kpreid_ left	02:01
	→ alexmchale joined	02:02
	← toxx_ left	02:05
	→ kpreid_ joined	02:07
frogonwheels	cbreak: not an option. delphi. other programmers. fat chance	02:08
	cbreak: also I believe .dproj files are effectively DotNet make files, and I believe they have to stick to crlf (also with a utf8 bom at the beginning)	02:09
	← lightcap left	02:10
frogonwheels	.. really there has to be another name for a utf8 bom. seing as it's not really a byte-order-marker.. but that's another story.	02:10
cbreak	yay for XCode.	02:11
	seems it's project files are UTF-8 encoded with LF line endings	02:11
	it is a byte order mark	02:11
	not sure if it's actually used as such	02:12
	→ _iron joined	02:12
frogonwheels	.. utf-8 wouldn't have different byte-orders would it?	02:12
cbreak	probably only in UTF-16	02:12
frogonwheels	yep. exactly.	02:12
	← alexmchale left	02:15
	→ amerine joined	02:15
	← amerine left	02:17
	→ VVSiz_ joined	02:24
	hipe_ → hipe	02:25
	→ Textmode joined	02:26
	← anxiolytic left	02:26
Textmode	morning all	02:27
	← VVSiz left	02:27
	→ anxiolytic joined	02:27
	→ airfoil joined	02:29
	← anxiolytic left	02:31
	→ crankharder joined	02:36
	→ anxiolytic joined	02:36
	← kpreid_ left	02:39
	→ yacoob joined	02:42
yacoob	Hmm. I used to say 'git checkout foo' and that was automatically creating local foo, tracking origin/foo	02:43
	can anyone tell me which version of git introduced that?	02:43
bob2	look for 'auto tracking' or so in the changelog	02:44
	← Unirgy left	02:44
bob2	actually, nevermind :)	02:45
yacoob	1.6.0.4 doesn't seem to have it, just been wondering whether this is expected, or is it just broken :D	02:45
	→ mr_danie1 joined	02:49
	← mr_daniel left	02:51
	← Thell left	02:52
sergio_	Are there any "must read" stuff on git or should I just read whatever google returns ?	02:56
	→ owen1_ joined	02:56
rudi_s	sergio_: progit is IMHO a very good guide for git.	02:56
	offby1` → offby1	02:56
	← offby1 left	02:57
	→ offby1 joined	02:57
owen1_	i checkout older sha1. how to turn it into master instead of the original master?	02:57
sergio_	rudi_s, thanks	02:57
offby1	owen1_: "git checkout master; git reset --hard OLDSHA1"	02:57
rudi_s	np	02:58
owen1_	offby1: thanks!	02:58
offby1	sergio_: 1) Read "Git For Computer Scientists" (http://eagain.net/articles/git-for-computer-scientists/); 2) run "gitk --all"	02:58
	← drizzd left	02:58
yacoob	sergio_, progit is good, git internals is also interesting	02:59
	← phaedra left	03:00
sergio_	yacoob, I was starting to read the "Community Book"	03:00
	but i'll bookmark those two	03:01
yacoob	community book is neat, but quite chaotic	03:01
spearce	yea, progit was a commercial effort so scott and i put a bit more effort into making it easy to follow	03:02
	← mcdebugger left	03:03
	← owen1_ left	03:03
yacoob	it's a bit sketchy in few places, but I liked it :)	03:03
offby1	spearce: I seem to recall that you're working on a Java reimplementation of git. If I'm right, does this bode well for Windows -- will your version perhaps run faster than (say) msysgit?	03:07
spearce	eclipse.org/jgit. but its also been ported to C#/.Net as GitSharp.	03:08
	← parasti left	03:08
	← reprore_ left	03:09
	→ alexmchale joined	03:11
	→ mklappstuhl joined	03:14
	→ jargan joined	03:15
	→ AAA_awright_ joined	03:16
offby1	oh, it's runnable now? That actually didn't occur to me :)	03:17
	← rettub left	03:17
offby1	ah, but it's just a library; not a command-line suite.	03:17
	Bummer	03:17
	→ badboy__ joined	03:18
spearce	jgit does have a command line tool. less friendly on window though because the tool is actually a bourne shell script with a standalone JAR concated on the end.	03:19
	not sure how to do something nifty like that with ms-dos...	03:19
offby1	shame	03:19
	← AAA_awright left	03:19
	← lemonchicken left	03:19
	→ rettub joined	03:19
	← mklappstuhl left	03:20
	→ cytrinox_ joined	03:20
	← jast left	03:21
	→ lemonchicken joined	03:23
	← cytrinox left	03:23
	cytrinox_ → cytrinox	03:23
	← AAA_awright_ left	03:26
	→ AAA_awright joined	03:28
	→ dys` joined	03:30
	→ kpreid_ joined	03:31
	← dys left	03:33
	Mike_lifeguard → Mike\|\|gone	03:40
	→ amerine joined	03:43
	← kukks left	03:43
	← git__ left	03:45
rtyler	spearce: figure i'll ask this here instead of PM; do you know offhand if the git makefiles contain packaging commands for platforms?	03:45
	or is that something that's all handled downstream?	03:45
	← alexmchale left	03:46
spearce	its all handled downstream	03:47
	← anxiolytic left	03:48
unregistered	przemoc: i got sick of fiddling with svn and tried redmine per your advice. the source browser isn't as nice and is slower than I am used to :(. but at least it works.	03:51
	← spearce left	03:53
rtyler	drats, there's a hardcoded /usr/bin/python in these makefiles t hat breaks on FreeBSD	03:55
	(it's not shipped with a Python distribution, one is from ports in /usr/local/bin/python)	03:55
	← airfoil left	03:55
	→ juan_arandaalvar joined	03:56
	← timja left	03:56
	← sjuxax left	03:59
	→ sjuxax joined	03:59
	← sjuxax left	04:00
bcoca	.	04:02
	→ andres_ joined	04:03
	→ erikvold joined	04:03
	← eno left	04:04
	→ scarabx joined	04:05
	→ eno joined	04:05
andres_	is there a way to undo a $git reset --hard ?	04:06
mspang	had you run git add on the file you want back	04:07
andres_	mspang: yeah they should all be in the repo	04:08
	mspang: I thought reset --hard moves you back to the previous revision?	04:08
mspang	it does both	04:08
	type git reflog	04:09
	→ timja joined	04:09
mspang	you can do git reset --hard HEAD@{something} to get back to a commit you had before	04:09
offby1	andres_: yeah, what mspang said	04:09
	→ rsewill joined	04:11
andres_	so what exactly does a simple git reset --hard do?	04:12
	i.e. with no arguments?	04:12
	← rsewill left	04:12
andres_	(btw, that worked, thx)	04:12
mspang	blows away uncomitted changes	04:13
andres_	ok	04:13
	no trouble then	04:13
	any experience with gitolite? I got it working yesterday, but today after someone else cloned one of the repos I'm hosting, I can't do anything with it	04:13
sitaram	I have some...	04:14
	:)	04:14
andres_	haha	04:14
	you wrote it?	04:14
	or at least you have a very similar name :)	04:14
	unable to chdir or not a git archive	04:14
	keep on getting this... is there a log I can look at to know what's going on?	04:14
sitaram	you're using the wrong ssh key; see doc/6 //gld6s	04:15
	you're using the wrong ssh key; see doc/6 http://github.com/sitaramc/gitolite/blob/pu/doc/6-ssh-troubleshooting.mkd	04:15
	→ t0rc joined	04:16
	→ rsewill joined	04:17
	← Tureba left	04:18
	← rsewill left	04:19
sitaram	and by the way, someone else cloning cannot cause this; something else happened	04:19
	← Pupeno left	04:19
	← kelvie left	04:20
	→ cloudhea joined	04:23
	→ apipkin joined	04:23
	← imajes left	04:24
	→ Wolfman2000 joined	04:24
	→ imajes joined	04:26
	→ richcollins joined	04:26
	→ kelvie joined	04:27
	← cloudhead left	04:27
andres_	sitaram: pass all the check except for the ssh gitolite info	04:29
apipkin	anyone ever have an error using github?	04:29
andres_	sitaram: which outputs: bash: info: command not found	04:29
apipkin	i'm brand new to git, and i did a git push origin master, and got "fatal: The remote end hung up unexpectedly"	04:29
	→ Plasmastar joined	04:31
	→ alezandro joined	04:33
	→ rsewill joined	04:34
fryguy	apipkin: do you have ssh keys setup correctly and such?	04:37
apipkin	i do not know	04:37
	i'm not sure what that means	04:37
	i'm following the next steps from github	04:38
	→ fool__ joined	04:39
	← fool__ left	04:39
	→ fool__ joined	04:39
	← fool_ left	04:39
andres_	all the priv keys are listed in ssh-add -l	04:40
	← t0rc left	04:40
andres_	and can ssh gitolite no problem	04:40
	→ airfoil joined	04:40
apipkin	ssh-add -l	04:41
	i get "The agent has no identities"	04:41
	→ codeswing joined	04:41
	← juan_arandaalvar left	04:42
	→ segher joined	04:42
	→ project2501a joined	04:42
	← rsewill left	04:42
	→ rsewill joined	04:43
	→ Hard joined	04:43
andres_	apipkin: that wasn't actually for you, but it's actually your problem, you need to add your keys. You probably already created a private/pub keypair doing something like ssh-keygen -t rsa	04:44
artagnon	thiago_home: I found http://repo.or.cz/w/svn-all-fast-export.git -- could you kindly clarify the license?	04:44
sitaram	andres_: try the sshkeys-lint program	04:44
andres_	and copy/pasted the pub in github	04:45
sitaram	andres_: run it without arguments to see how to use it; then use it	04:45
	← Hard left	04:45
andres_	now you have to ssh-add ~/.ssh/id_rsa or whatever you named it	04:45
sitaram	andres_: after that, (or before), do "ssh -v gitolite" and see which actual pubkey offered is being accepted by the server	04:46
	→ johnw joined	04:49
	← Sigma left	04:52
	← mr_danie1 left	04:53
	← rsewill left	04:53
	→ mr_daniel joined	04:55
	← project2501a left	04:55
	← sagsousuke left	04:56
artagnon	thiago_home: Never mind, I found it in the header. GPL v2 or later. Thanks :)	05:02
Wolfman2000	evening/morning. Is it ever common to want to clone your own repository instead of just making a branch?	05:02
artagnon	Wolfman2000: Depends. What's the situation?	05:03
	Wolfman2000: It's quite queer though.	05:03
	← mr_daniel left	05:03
Wolfman2000	I have put the vast majority of the source code to a website of mine on a git repository. I have been given enough tools to make a second website off of the existing one, with minor changes.	05:04
	The way I normally program with git, I try to eventually have everything lead back to one branch.	05:04
	→ codeshepherd joined	05:04
artagnon	Wolfman2000: Maintain two separate branches then.	05:04
	One for the old website and another for the new	05:04
andres_	sitaram: I take it it's not available in mac os x	05:04
artagnon	git.git has 4 (5?) permaent branches, including master, next, todo, pu. Why shouldn't you?	05:05
Wolfman2000	artagnon: The "old" website is still going to be in use for the record.	05:05
	→ mr_daniel joined	05:05
artagnon	Wolfman2000: If you want to use both simultaneously, I don't see how you can avoid cloning.	05:05
Wolfman2000	As far as why shouldn't I...one of my common patterns is to want to merge my branches when I'm done with my work. I'd have to remember to not do that.	05:05
	artagnon: I am trying to avoid branching, not cloning.	05:06
andres_	sitaram: as for the pub key that it's using is the id_rsa, not andres	05:06
Wolfman2000	I always thought branches were for implementing new things in the primary program	05:06
andres_	sitaram: looks like that's the problem?	05:06
Wolfman2000	and then when you get done, remove the branch	05:06
artagnon	Wolfman2000: Not necessarily.	05:06
	That's a common usecase, but it's not always the case	05:06
	Like I said, see git.git for an example of an alternative usecase.	05:06
	→ mcdebugger joined	05:07
Wolfman2000	I can't view www.git.git	05:07
bob2	what is www.git.git?	05:08
Wolfman2000	bob2: a non existant website	05:08
	← nevyn left	05:08
Mage	uh	05:08
	i don't imagine www.git.git would work	05:08
Wolfman2000	I found the repo anyway. http://repo.or.cz/w/git.git	05:08
andres_	sitaram:	05:09
Mage	yeah...	05:09
	most websites have directories and such (:	05:09
Wolfman2000	...wooooooooooboy. The git-browser lags my browser for git's heads	05:10
andres_	sitaram: the authorized_keys had a key above gitolite begin which was the original id_rsa key I had used to ssh into the server, I removed it and it seems to be working fine	05:10
	→ Theravadan joined	05:11
andres_	sitaram: is that good/bad/uninportant and it really seems like nothing had changed since it was working, pretty strange	05:11
	← disappearedng left	05:12
andres_	I'm moving a repository into gitolite, but I had a bunch of branches in this repo, how can I make sure the structure is preserved when moved over?	05:13
	anyone?	05:13
	apipkin: did that work?	05:13
	→ Twisted_Mentat joined	05:14
	→ novas0x2a1 joined	05:14
andres_	another general question: what's the best way to have other repositories i.e. libraries added to a git repo (these are just checked out from git or hg)	05:15
apipkin	andres_: no still having the same issue :-/	05:15
andres_	apipkin: did you add them to ssh-add?	05:16
	also	05:16
	← richcollins left	05:16
andres_	try doing eval `ssh-agent` (those are backquotes)	05:16
apipkin	i dont understand what i'm suppose to be adding	05:16
andres_	what does	05:16
	ssh-add -l show	05:17
	← novas0x2a left	05:17
apipkin	2048 cf:e1:5b:c6:89:af:6d:1b:79:17:c2:7c:ad:d2:66:89 apipkin@apipkin-laptop (RSA)	05:17
	← ilteris_ left	05:17
	→ ilteris_ joined	05:17
	→ disappearedng joined	05:19
andres_	ok	05:19
	did you paste the contents of apipkin-lapto.pub onto github?	05:20
	also	05:20
	did you create a new repository in github?	05:20
	← spacebat left	05:21
apipkin	andres_: this is what i'm following... http://gist.github.com/324171	05:21
	ah... didn't do line 3 or 4	05:21
	← ilteris_ left	05:22
	→ spacebat joined	05:22
apipkin	do you know what line 5 means? or is that what you are asking me to do	05:23
	cloudhea → cloudhead	05:23
andres_	yes	05:24
	line 5 is what I'm asking you to do	05:24
	so you if you log into github	05:24
	account settings	05:24
	> account overview > ssh public keys > add another public key and paste in there the contents of	05:25
	cat apipking-laptop.pub	05:25
	be careful not to add extra lines, etc	05:25
apipkin	cat apipkin-laptop.pub says "no such file or directory"	05:26
	do i need to change my path?	05:26
	← hakunin left	05:26
apipkin	andres_: is this it?	05:27
	http://help.github.com/linux-key-setup/	05:27
andres_	yup	05:27
	→ hakunin joined	05:28
andres_	although you don't have to delete the old keys	05:28
	you can reuse one if you want	05:28
apipkin	just the generating a key part	05:28
andres_	oh yes	05:28
	if you don't have any keys in .ssh	05:28
	then you need to generate one	05:28
	→ Sigma joined	05:29
andres_	you can do it with ssh-keygen -t rsa	05:29
	anyone know a good way to include and hg repo inside a git repo?	05:29
apipkin	in ~/.ssh i have id_rsa, id_rsa_pub, and known_hosts	05:29
johnw	andres: well, you could import it using git-hg, and then regard it as a submodule with subsequent updates	05:30
	you want to use hg-fast-export.sh	05:30
sitaram	andres_: mac os? the sshkeys-lint is part of gitolite	05:31
	andres_: when you move a repo, (at the OS copy level) everything is preserved as is, but you do need to make sure the update hook exists and is the right link	05:32
	→ mr_danie1 joined	05:32
	← orafu left	05:32
	→ patbam joined	05:33
patbam	is there a way to commit all the files listed one-per-line in a file?	05:33
	→ orafu joined	05:33
ToxicFrog	cat file \| xargs git add; git commit	05:33
	← amerine left	05:33
andres_	sitaram: I didn't move it at the os level, I just changed the origin master in the .git/config and pushed it to the gitolite repo, creating a new one, should I have copied it over to gitolite on the server?	05:33
	← mr_daniel left	05:34
unregistered	is git clone a lot slower on windows?	05:34
patbam	ToxicFrog: will that submit the lot of them in a single commit?	05:34
ToxicFrog	patbam: yes? git add stages files for commit, git commit commits them	05:35
	If you're worried, do the cat \| xargs git add first, then git status to make sure it looks like what you want	05:35
unregistered	i'm getting 1 tenth the transfer rate on windows than I have been otherwise	05:35
patbam	ToxicFrog: ok, thanks	05:36
	→ Qvintvs joined	05:38
	← SpookyET left	05:39
sitaram	andres_: in that case "push --all" will do	05:39
	andres_: and perhaps "push --tags"	05:39
Qvintvs	how would I write .gitignore rules to ignore all files in a log/ folder except one? would it be log/ <newline> !log/exception.file ?	05:39
unregistered	heh. I did a clone. didn't touch anything. but git thinks 4 files have been modified	05:41
	i think somehow the endlines got replaced on those files..	05:42
	← hakunin left	05:42
unregistered	core.autocrlf was set to input. was that a bad idea?	05:44
	← patbam left	05:44
andres_	sitaram: after removing the first line in the authorized_keys (above #gitolite start), i.e. the rsa key I had used to set up access for git, it worked, I could interact with the repo, but after running sshkeys-lint it said I could only use the andres pub key to interact with gitolite. I readded the rsa key to authorized keys, and it stopped working again, which one is the correct setup?	05:46
	→ skrit joined	05:46
	← codeshepherd left	05:47
	→ codeshepherd joined	05:47
andres_	sitaram: here's the output of sshkeys-lint http://dpaste.com/168912/	05:47
	← aresnick left	05:48
sitaram	andres_: that looks fine; can you paste ~/.ssh/config also?	05:49
	andres_: it should have a "host gitolite" para, so "ssh gitolite info" will hit gitolite, and "ssh git@server info" will hit bash	05:50
andres_	johnw: can you elaborate a bit more on that? I would export it to a git repo, but then I can't do subsequent git pulls directly? I would have to first update the hg repo and then reexport to git?	05:50
sitaram	andres_: in turn, that means you use "git clone gitolite:reponame" etc, not "git clone git@server:reponame"	05:50
johnw	you'd have to use fast-export to constantly refresh your git mirror of the hg repo	05:51
	your use of git would be read-only in this case	05:51
andres_	johnw: also, how is having it as a submodule better than just copying it in?	05:52
johnw	if you copy it in, you can't update it again later using fast-export	05:52
cxreg	sitaram: supposedly there's a difference between fetch + rebase, and pull --rebase. do you know what that is?	05:52
sitaram	cxreg: the latter helps to overcome a rebase on the server side; it finds a merge-base byb looking at the reflog of the tracking branch I think	05:53
	cxreg: s/server side/upstream	05:53
	s/overcome/compensate	05:54
cxreg	hm	05:54
	sitaram types faster than he can think :)	05:54
andres_	sitaram: I think it looks fine http://dpaste.com/168913/	05:54
sitaram	cxreg: it's hard for me to explain; wish doe_ner or Il_ari were here (not highlighting them to avoid disturbing them)	05:54
cxreg	i think i get it	05:54
sitaram	andres_: everything looks quite normal for an "admin workstation" (aka "gitolite easy install was run from this user/workstation")	05:55
andres_	sitaram: that's correct	05:56
sitaram	andres_: so... is there a problem still? something not working?	05:56
	← tzafrir_laptop left	05:57
	← Twisted_Mentat left	05:58
andres_	sitaram: well, if I leave the initial rsa key I used when setting up gitolite, it gives me the not a git archive error	05:59
apipkin	andres_: thanks for the help! i got it uploading :)	05:59
andres_	sitaram: so not sure if I should remove or keep the public key in the server side?	05:59
	apipkin: welcome, glad I was of help	06:00
sitaram	andres_: what command are you using when you get the error?	06:00
andres_	sitaram: git pull	06:01
	← skrit left	06:01
	← apipkin left	06:01
sitaram	andres_: what does "leave the initial rsa key..." mean? is that "id_rsa.pub"?	06:01
andres_	sitaram: yes, the one I initially cat'ed into the authorized_keys on the server, to get ssh access without a passwd	06:02
	sitaram: for the git account	06:02
sitaram	andres_: what does "git remote -v" say?	06:03
	← LiamH left	06:03
andres_	origin gitolite:gitolite-admin (fetch)	06:03
	origin gitolite:gitolite-admin (push)	06:03
	→ Twisted_Mentat joined	06:03
sitaram	andres_: and "ssh gitolite info"?	06:04
andres_	sitaram: could the version of git in the server be a problem? I don't think 1.7.0 made it to the stable ubuntu repos	06:04
	sitaram: bash: info: command not found	06:05
	it's basically giving me a shell	06:05
	if I do ssh gitolite	06:05
sitaram	andres_: and "ssh-add -l"?	06:05
andres_	but if I remove the other id_rsa.pub, it give me repo info	06:06
	ssh-add -l shows me id_dsa, id_rsa and andres	06:06
	sitaram: the three keys	06:06
sitaram	that is weird... can you paste "ssh -v gitolite pwd"	06:07
	→ project2501a joined	06:07
	→ skrits joined	06:08
andres_	sitaram: leaving the id_rsa.pub in or removing it?	06:08
sitaram	leaving it in	06:08
	← codeswing left	06:08
andres_	sitaram: http://dpaste.com/168915/	06:09
	oh	06:09
	sitaram: yeah, that one is correct	06:10
sitaram	notice line 33? and then 35/36? for some reason sshd is not accepting the andres key	06:10
	can you paste the same command after taking out the id_rsa?	06:11
andres_	sitaram: http://dpaste.com/168916/	06:14
	it won't let me ssh in and give me a shell	06:14
	I agree it's rejecting my andres key	06:15
	but don't know why	06:15
	→ webchick joined	06:16
	→ pingveno joined	06:17
	← Qvintvs left	06:18
sitaram	andres_: can you paste youor entire ~/.ssh/config?	06:18
	(or at least temp'ly remove everything except the "host gitolite" para and try all this again?)	06:19
	→ Tekk_ joined	06:19
andres_	sitaram: http://dpaste.com/168919/	06:20
	← krobertson left	06:20
	← Tekk_ left	06:20
	← project2501a left	06:20
sitaram	andres_: I have no idea what's going on; are you sure you're removing the right key from the authkeys line on the server side? I also have go offline for an hour or so (although the session will be running)	06:22
andres_	yeah, I believe I'm deleting the right key	06:23
sitaram	andres_: after deleting it, copy the authkeys file back and run sshkeys-lint again; see what it says?	06:23
	← ttvd left	06:23
andres_	there is only my key and the one I added for my coworker	06:23
	→ ttvd joined	06:24
	← codeshepherd left	06:24
	→ project2501a joined	06:24
	→ amerine joined	06:25
andres_	sitaram: http://dpaste.com/168920/ here you go, ok, I'll keep on poking around. I'm on a laptop and should also move, but will be back online in a little bit	06:26
	sitaram: I might just reinstall gitolite, I'd be done by this point :(	06:26
sitaram	andres_: do you have a passphrase on these keypairs?	06:27
andres_	I think so	06:27
	sitaram: could that be a problem?	06:28
sitaram	andres_: if not, can you "ssh-add -D" and then try? I dont understand why "andres" is being offered twice in both those pastes, and only the second offer succeeds	06:28
chuck	What does this error mean exactly? http://gist.github.com/324213	06:28
sitaram	andres_: not a problem per se, but getting ssh-agent out of the way might give us some info	06:28
	chuck: faq non-ff	06:29
	er	06:29
	faq ff	06:29
	hmm	06:29
	anyone know the keyword for the non-fastforward faq	06:29
chuck	nevermind, I just reset the commit that was causing that	06:30
andres_	sitaram: ok, deleted them, http://dpaste.com/168922/	06:31
	sitaram: ok, now it seems to be working, I can git pull and still ssh andres and git	06:32
	sitaram: seems like git-add was messing things up?	06:32
	→ anxiolytic joined	06:34
	← andres_ left	06:35
	← scarabx left	06:36
sitaram	andres: did you by any chance move the actual files around (rename/swap) after doing the "ssh-add"?	06:37
	← mcdebugger left	06:42
	← yacoob left	06:44
	← pingveno left	06:46
	→ fool_ joined	06:46
	← fool_ left	06:46
	→ fool_ joined	06:46
	← fool__ left	06:47
	← dreiss left	06:51
	→ Adman65 joined	06:52
Adman65	Hey guys, I've made some changes. There are a few files I want to commit, and alot of changes I just want to undo. If i add the files I want saved to a commit and commit, then reset --hard HEAD will that blow away all the changes I don't want ?	06:53
	→ priidu joined	06:54
	← cloudhead left	06:56
	→ WebDragon joined	06:56
	→ Enno_ joined	06:58
	→ lll joined	07:04
	→ SteveDekorte2 joined	07:05
amerine	Adman65: That should work.	07:05
SteveDekorte2	Is there any way to just test if you have push permissions on a git remote branch?	07:05
Adman65	amerine: and it did perfectly, except I had to deleted the untracked files myself. Is there a way to blow those away?	07:06
amerine	Adman65: It wont delete any files you haven't added to the repo though, so if you were trying out a plugin in that you didn't add you'll have to manually delete it	07:06
	Adman65: Yeah, git clean	07:06
Adman65	cool	07:06
	ty	07:07
	← alezandro left	07:08
	← a2ron left	07:08
	→ priidu_ joined	07:09
	→ foo joined	07:11
foo	I'm looking at installing a VCS. I've read a few documents. The use is for myself, possibly a few other people in the future. Any reason I shouldn't be using git?	07:12
	← amerine left	07:14
	→ amerine joined	07:14
	← airfoil left	07:15
Mage	foo: if you want to enforce everyone to commit to the same centralized location, svn might be a better pick	07:15
	or another non-distributed vcs	07:15
foo	Mage: I see. Why would I want to do that? (I'm very open-minded, still trying to figure this out)	07:15
Textmode	yes, git can do centralized, but its not what its built for.	07:16
foo	I see	07:16
Mage	I don't know. some people like having some sort of control in that aspect	07:16
Textmode	(I mean, it is a distributed version control system...)	07:16
foo	Textmode: :D	07:17
	Mage: thank you, interesting. /me thinks	07:17
	← godsflaw left	07:17
Textmode	other than that, the only think I can think of is the the initial clone can be a bit slow, since its basically the entire history...	07:17
Mage	indeed	07:17
amerine	foo: I would argue that wether you choose a centralized model or not, there are definite technical reasons to use GIT over SVN, the biggest in my book being able to access the entire history offline.	07:17
Textmode	which might be an issue if you have a lot of short-term contributers.	07:17
amerine	s/wether/whether	07:18
	← Twisted_Mentat left	07:18
foo	Textmode: aha, I see	07:19
	← eno left	07:19
foo	Mage: let's say I'd like to have the control of knowing it's on a centralized system. even if it is, won't someone still be able to download everything to their machine? or no?	07:19
Enno_	so today msysgit did this thing where it left me with a directory on my disk that had no owners, that i could not delete, change the ACL on or do anything. is this a common thing?	07:19
	→ codeshepherd joined	07:19
Textmode	Enno_: :/	07:20
amerine	foo: If they have access to the repository location, yes.	07:20
	foo: But that same things exists for every SCM	07:20
	← Plasmastar left	07:21
	→ Plasmastar joined	07:21
foo	amerine: so, there is a way to use svn, let someone write code/fix code, and not give them the repository location?	07:21
	(thanks for being open to me talking about svn guys)	07:21
amerine	foo: And thats a liberal usage of everything, I should say common opensource SCM's	07:21
	→ eFfeM1 joined	07:21
	→ eno joined	07:21
Enno_	guess i'm alone on that. well, i hope it doesn't happen again, it was a pain in the neck :-)	07:22
eFfeM1	hi, i want to import/checkout a subdir that exists in the main branch but not in my private branch, but can't really find the command to do so (except for checking out the whole main branch,stash away the files I need and checking out my branch again)	07:23
amerine	foo: If you expose yrou SVN repo through something like mod_svn, It's my understanding there is some sort of ACL you can use. Beyond that your question is something I've never had to deal with	07:23
eFfeM1	anyone an idea what command to issue ? guess there should be an easier way	07:23
foo	amerine: alright, thank you, I appreciate it	07:23
jhelwig	eFfeM1: You can use checkout to checkout a subdir and everything under it from a specific revision. git checkout <revision> -- path/to/files/	07:24
amerine	eFfeM1: something like git checkout master -- /path/to/dir	07:24
Textmode	foo: generally if you want to do that sort of thing, you give them some workable copy of the source (such as a bog-standard source tarball), have them write their patch to that, and send you (or another maintainer) a diff, which is vetted and applied by them.	07:24
	(er..."them" being the maintainer)	07:25
Enno_	that does happen, but it's a pita when the patch is huge	07:25
eFfeM1	jhelwig: amerine thanks willl try	07:25
foo	Textmode: I see, thanks. That makes sense	07:25
Textmode	true.	07:25
Enno_	for example, we have a contractor that does porting of our application to another OS	07:25
Textmode	you generally only do that with "outside" contributers, though.	07:26
Enno_	he is not on site,	07:26
eFfeM1	amerine: jhelwig that did it, thanks alot!	07:26
Enno_	but he still gets enormous value out of the history for example.	07:26
	so for him to use git makes perfect sense.	07:26
	and svn just isn't workable in that scenario.	07:27
	← codeshepherd left	07:27
	→ pingveno joined	07:27
Enno_	and neither are patches, because that guy touches _everything_	07:27
foo	Alright, thinking more about my use... let's say I'm using Drupal, and I'd like 2 people to work on different parts (and not be able to see each others work because there isn't much of a need) - would svn or git be better for the job? I'm thinking svn, but I don't have enough intimate understanding of things, I think	07:29
bob2	don't bother being silly about it	07:29
foo	bob2: was that for me?	07:30
bob2	yes	07:30
	but git or svn would work	07:30
foo	ok	07:30
bob2	if they're non-techncial, svn might be simpler	07:30
foo	being silly = making it so they can't see each others stuff?	07:30
bob2	(since it is probably simpler to use in the common case, and has better windows/ os x integration)	07:30
	yeah :)	07:31
	← greg8si left	07:31
Enno_	if they don't need to see each other's work much, and they are far apart, frequent svn updates of giant image libraries will drive them nuts. but that is probably equally or more annoying with git.	07:31
foo	gotcha :)	07:31
	→ alezandro joined	07:32
	← fool_ left	07:34
foo	it's mostly drupal stuff, so it's not too large (I don't think)	07:34
	foo does some more googling and reading	07:35
foo	thanks guys, I really appreciate it	07:35
	→ fool_ joined	07:36
	← fool_ left	07:36
	→ fool_ joined	07:36
	← Adman65 left	07:40
	→ vivien` joined	07:42
	→ yevlempy joined	07:42
	→ dytrivedi joined	07:44
	← vivien` left	07:46
foo	ok, I'm using git. I think it'll be better in the long run. I'm not too familiar with work flow with a CVS. So, I'm creating a repository right now. Is there any problem with making it /var/www/html/drupal-website ? eg. making it the actual apache root dir? or is that ok?	07:46
	→ brizly joined	07:46
	→ vivien` joined	07:46
foo	hm, actually, I probably need to learn the difference between a repository and a branch. I think branches are inside of repositories.	07:47
	foo googles	07:47
	→ hyperair joined	07:47
bob2	don't do that	07:48
	they are	07:48
	you're not meant to push to a non-bare repository	07:48
	non-bare means 'repository that has all your files sitting in it'	07:48
foo	bob2: ok, thank you. why don't I want to do that? I have a dev environment that I was going to do with it.	07:48
	aha	07:48
	← vivien` left	07:48
	→ vivien` joined	07:49
foo	I see	07:49
bob2	I think the go is to have a bare repository that you push to	07:49
	and have a hook that updates some other non-bare repository	07:49
foo	bob2: gotcha, makes sense. I can do that with some bash magic, I think	07:50
bob2	yes	07:50
	iirc there's an example in the faq	07:50
foo	perfect	07:50
	bob2: so, let's say I have a dev environment + live environment (drupal). I should create a separate directory where I work on drupal modules, then I commit as I work, then when ready I push to dev module branch, and then my hook script copies over the data I push to my actual dev environment, then I can test in the browser?	07:52
	foo learning about workflow with a cvs	07:52
bob2	yes	07:53
	the hook script probably only does 'cd /www/live ; git pull'	07:53
	← VVSiz_ left	07:53
	→ VVSiz joined	07:53
foo	bob2: ok, and when I work on my own code, I'm working blindly, I shouldn't have another drupal instance in that to test, or do some people do that?	07:53
artagnon	foo: Keep one dev branch, one production branch.	07:54
foo	artagnon: ok, that's what I was thinking. thinks he's getting on the same page	07:54
bob2	ah, sorry, I assumed you'd have a live and dev branch	07:54
foo	bob2: ah, I do.	07:55
bob2	push to dev: hook script updates dev	07:55
	push to live: hook script udpates live	07:55
foo	mhmm, ok	07:55
artagnon	foo: the production instance of drupal can track the production branch, and the development instance can track the development branch.	07:55
	→ abadr joined	07:55
artagnon	either use hooks or git pull manually. I prefer the latter personally.	07:55
foo	artagnon: I see, thank you	07:56
	← abadr left	07:56
artagnon	np	07:56
	→ abadr joined	07:57
	← abadr left	07:58
MetaCosm	what is the right way to commit as someone else (on their behalf)	07:59
	nm	07:59
jhelwig	MetaCosm: man git-commit You can set the author/committer info using some env vars.	07:59
Gitbot	MetaCosm: the git-commit manpage can be found at http://git.or.cz/man/git-commit	07:59
foo	So, to make sure I got it, I have 2 drupal instances, 2 branches (dev/live). The repository (to start) should be an empty directory somewhere on my system. When I work on my own branch (doing all the work on the same system), I edit code, then commit it to track changes, (I do this all blindly without testing in dev instance), then I push to dev, then a hook script pulls into my actual dev instance, then I can test in drupal dev instance. sound good?	07:59
bob2	yes	08:00
artagnon	MetaCosm: Ask them to git-format-patch and git-apply the patch	08:01
bob2	ideally you're running all your unit tests on your machine, so things Just Work on dev ;p	08:01
artagnon	foo: Yes. You could improvise by running gitosis though :)	08:01
foo	bob2: aha, that's what I'm wondering. so on the code that I'm working on, I should have another drupal instance to test that (that way I can test + commit and then push when ready) - right?	08:02
bob2	well, you can if you like	08:02
	drupal/php seem to make that a bit annoying	08:02
foo	ok, I think this fits into my current workflow well. Right now, I have 3 drupal instances: playground (I do a ton of testing/break things/implement/etc... all things are born here), then I have a "staging" platform where I have a "healthy" instance, this is where I migrate stuff from live and try it with my new addons, then I have live. Now, with git, my two branches should be staging and live, and then playground will be where I actually do all my work + ...	08:04
	... commits, then when I push it goes to staging, I test, then when that's good I roll over to live. am I thinking about this properly?	08:04
bob2	sounds like a good start	08:04
	→ alley_cat joined	08:05
	→ ab3 joined	08:06
foo	perfect, thank you bob2. Now that I understand the conceptual aspect, I think it'll be easier to set up	08:06
	→ rado1 joined	08:07
	← yevlempy left	08:08
	→ kumbayo joined	08:08
foo	I'm following this - http://blogs.koolwal.net/2009/08/07/learn-git-series-part-1-installing-git-on-debian/ - how does this typically work, does a repository need a branch? or are branches optional?	08:08
	foo doesn't see option to create a branch here	08:09
bob2	mostly yes, repositories have branches	08:09
	→ yevlempy joined	08:09
bob2	'git init' creates a branch called 'master', by default	08:09
	← daaku left	08:10
foo	bob2: Aha! I see. I want to leave master branch blank, and create a staging and live branch	08:10
	I believe	08:10
	← _Vi left	08:10
	← gbacon left	08:14
foo	hm, git checkout -b staging - does this sound right? /me scratches head	08:15
bob2	yeah	08:15
	→ godsflaw joined	08:16
foo	bob2: do I need to make a staging dir, or does git do that?	08:16
bob2	where?	08:17
	oh right	08:17
	you're used to svn	08:17
foo	I've only done very little with svn, just wondering if I need to make a "staging" directory in my repostory directory, or something	08:17
bob2	so, branches are not directories, they're config info in .git	08:17
	no, you don't	08:17
	ypu put your code directly in the repository root	08:17
foo	aha!	08:17
	I see	08:17
bob2	'git branch' shows the branches in the current repository	08:18
	← webchick left	08:19
foo	bob2: I made my repository in /home/www/repository which is empty. now, my playground environment is in /home/www/playground.com/sites/ - if I want to work on my code here, do I need to create the branch there?	08:19
bob2	simply, you need a repository there	08:20
	← bweikulrich left	08:20
bob2	cd /home/www ; git clone repository playground.com	08:21
	ie clone 'repository' to 'playground.com'	08:21
foo	bob2: ah, I see. If that clones my repository to be playground.com, then what do I do with my live.com folder? do I make that another repository?	08:22
	(so staging + live would be 2 repositories as opposed to 2 branches)	08:22
	→ bambule joined	08:23
bob2	meh	08:24
	you need one repository per php thing you want to run	08:25
	maybe that's clearer	08:25
foo	bob2: ah. how would you define a branch?	08:25
	maybe that's where I'm getting confused	08:25
bob2	I don't know what you mean	08:25
foo	in the context of git, what is a branch?	08:25
bob2	a pointer to a revision	08:25
	anyway	08:26
	think of it this way	08:26
	each repository can only have one branch checked out at a time	08:26
	→ wwwald joined	08:26
bob2	ergo you need a separate repository for each version of your code that you would like apache to serve	08:26
foo	aha, I think that makes better sense. staging and live each have their own vhost in apache, which alone would probably mean I need one repository for each, I think	08:27
bob2	yes	08:28
	xactly	08:28
foo	ok, great, ok, thanks for rexplaining bob2	08:28
	← amerine left	08:28
foo	now, I did already create /home/www/repository; then I did git init, and I noticed it created a .git directory and all that good stuff. if I need to clone playground.com in a repository, should I just delete the repository directory, create it again, and then run the git clone stuff (like you said above) ?	08:29
rado1	foo, a branch would be used when say you want to work on some new site feature that will take a week to accomplish, but in the meantime you may have a few things you want to change on production, so you create your new feature branch, work a little, checkout the master, add your production small change, push that live, checkout feature branch, work more, etc.	08:30
foo	rado1: aha! thank you, that makes sense. /me will come back to that once he's got this basic install / repository set up	08:30
	→ ixti joined	08:31
	← killerchicken_ left	08:32
foo	in other words, the .git directory created within a repository directory is the only thing that tracks that a repository exists, right? eg. if I delete the .git directory, then I delete all of git's data for that repository, right?	08:34
rado1	the .git data is where all the history of everthing is kept, if you delete that then you deleted that repositories history	08:36
	→ nanoo joined	08:36
rado1	yes	08:36
foo	rado1: aha, I see. so git stores data for the repository in another place? or is the history it?	08:36
	← dytrivedi left	08:36
rado1	git store everything in the .git directory	08:37
foo	gotcha	08:37
rado1	well	08:37
	unless you clone a repository using hardlinks	08:37
foo	gotcha	08:37
	← skrits left	08:38
rado1	I would create the live repo first	08:38
foo	before the dev? ok	08:38
rado1	ya	08:38
	foo deletes /home/www/repository	08:38
	→ GodEater joined	08:38
	foo makes /home/www/live-repo/	08:38
rado1	do you just want live and playground environments, where you work in playground then push that live?	08:39
	or live, qa/staging, and playground?	08:40
foo	rado1: thanks for helping, I'm trying to wrap my mind around this... making some progress thanks to you guys + this guide on google. I'm thinking I can have staging + live, and I'll actually work on the code + commit in playground. so, I code in playground, commit, when it works, I push into staging, when it's good in staging, I push to live	08:41
	→ aziz joined	08:41
rado1	ok, for that, you first create the live-repo(which you did already), clone that to get your staging repo, then clone that to get your playground repo	08:42
	← madaemo left	08:43
rado1	then you work in playground, commit it to the playground repo, git push to the staging repo, then when it's ready you git push to the live repo	08:44
foo	rado1: aha, thank you! I see. so, let's say I am working in /home/www, and I have dirs playground/ staging/ live/ . I'm going to do something like: mkdir live-repo; git clone live-repo live; cd ..; mkdir staging-repo; git clone staging-repo staging; cd .. ... and then do that for playground. sound right?	08:45
rado1	well mkdri live-repo; git init; git clone live-repo	08:46
	you can't clone just a directory, it has to be a git repo	08:47
	→ mcdebugger joined	08:47
foo	rado1: aha, right, forgot the init	08:50
rado1	cd /home/www; cd live; git init; cd ..; git clone live staging; git clone staging playground; cd playground, start working	08:50
	← yevlempy left	08:51
foo	rado1: so, essentially, I'll be having 3 repos. now, as of right now, I'm working on the server this is on with ssh. let's say someone else wants to download the repo locally. how would that work? they download the playground repo, work on it in their environment, then they push to the playground repo on the server, I test it, I push it into staging ... sound right?	08:51
	→ voxpelli joined	08:52
rado1	they would do git clone ssh://yourserver.com:/home/www/staging/.git	08:53
	if you want them to push directly to staging	08:53
	or git clone ssh://yourserver.com:/home/www/playground/.git if you want them to push to playground	08:54
	← jmcantrell left	08:54
rado1	that's assuming they have ssh access to the box	08:54
foo	aha, I see. if they were to commit, that would be local, if they push, it gets pushed to me, right?	08:54
	← wwwald left	08:54
rado1	right	08:54
foo	aha, I see	08:55
	ok, cool, /me does this now	08:55
	rado1: will you be around for the next 10-20 min or so?	08:55
rado1	probably not	08:56
foo	in that case, let me ask one more question on the next step after I get this	08:56
	(if you still have time)	08:56
rado1	ok	08:56
foo	so, after I do the clone command for playground, playground-repo will contain all the code that playground/ has, I'll edit the code in playground-repo, I'll make changes, commit, when I think it's good (without testing) I'll push to playground instance, and then the file will get put in playground/ in the respective folder - sound right?	08:58
rado1	so then you have 4 repos?	08:59
	→ bentob0x joined	08:59
rado1	live, staging, playground, playground-repo?	08:59
foo	sorry, no, I thought I only had 3	08:59
	live, staging, and playground	08:59
	→ drizzd joined	09:00
rado1	I'm assuming playground and playground-repo are the same thing	09:01
foo	yeah, playground-repo is a repo for playground (generated via git clone playground-repo playground	09:01
	sorry	09:01
	← ab3 left	09:02
rado1	so then you do have 4 repos, right	09:02
	?	09:02
	→ mbroeker joined	09:02
rado1	well if you clone playground then the command would be git clone playground playground-repo	09:03
foo	ahh! I got it backwards	09:03
	hmm, darn, maybe I'm getting confused. rado1, I respect your time constraint. Maybe I'll catch you another time (unless you're ok to keep chatting)	09:03
rado1	thanks, I'm going to head to bead	09:03
	bed	09:04
foo	good night :)	09:04
	thanks again rado1	09:04
	← rado1 left	09:04
	→ fool__ joined	09:05
	← fool__ left	09:05
	→ fool__ joined	09:05
	→ tvw joined	09:06
	← fool_ left	09:07
	→ v0n joined	09:10
	→ apol joined	09:15
	← erikvold left	09:15
	→ andres_ joined	09:15
	← alezandro left	09:17
	← hyperair left	09:18
	→ Yuuhi joined	09:20
	→ Twisted_Mentat joined	09:20
	→ khmarbaise joined	09:20
	→ hyperair joined	09:21
	→ fragje joined	09:22
	← tvw left	09:22
foo	unregistered: haha, so I thought :) Ended up going with git, thanks	09:23
	← bambule left	09:25
	→ bambule joined	09:25
	→ froschi joined	09:26
	→ d0k joined	09:27
	← hgimenez left	09:27
	→ hgimenez joined	09:27
	→ airfoil joined	09:28
	→ Mooseous joined	09:28
	← Mooseous left	09:30
	← mr_danie1 left	09:31
	→ aspotashev\|eeepc joined	09:33
chuck	Do you guys know why "git push origin master" would be saying "Already up to date" even though there are commits to be pushed?	09:35
	→ mr_daniel joined	09:35
	← aziz left	09:36
	→ ehabkost joined	09:37
	← pingveno left	09:37
	→ mithro joined	09:38
	→ christophsturm joined	09:38
	→ fool_ joined	09:38
	← fool_ left	09:39
	→ fool_ joined	09:39
andres_	what's the easiest way to update git to 1.7.0 on ubuntu?	09:39
	chuck: are you sure you're on the right branch?	09:39
	→ aziz joined	09:39
andres_	chuck: and did you actually commit the changes?	09:39
	← fragje left	09:41
	← fool__ left	09:42
thiago_home	andres_: get the sources and compile	09:43
andres_	thiago_home:	09:43
	thiago_home: gotcha....	09:43
	← d0k left	09:43
	→ d0k joined	09:43
andres_	sitaram: question about adding a submodule from gitolite, do I want to refer to it as git submodule add gitolite:submodule.git submodule or do I instead want to qualify it with the full url, since other people who might check out the repo will likely not have the gitolite alias in the ssh config?	09:45
	← alus left	09:45
	maattd\|away → maattd	09:49
	→ pantsman joined	09:49
	→ bburhans_ joined	09:50
	→ Leonidas joined	09:51
	ijuma__ → ijuma	09:51
	→ stuffcorpse joined	09:52
	← bburhans left	09:52
	→ Mooseous joined	09:53
	a3li → a4li	09:53
	← Twisted_Mentat left	09:54
	a4li → a3li	09:54
	← Mooseous left	09:54
przemoc	andres_: https://launchpad.net/~git-core/+archive/ppa	09:59
andres_	przemoc: many thanks :)	10:00
	przemoc: is it advisable to upgrade to 1.7 or is 1.6.2 enough?	10:00
	→ fragje joined	10:01
przemoc	andres_: since 9.10 you can just do: add-apt-repository ppa:git-core/ppa and then apt-get update && apt-get upgrade	10:01
	← Enno_ left	10:02
przemoc	andres_: 1.7.0.1 is currently upstream maintained version, but of course 1.6.x is slightly more tested. so do as you like, but I prefer newer version	10:03
	→ malumalu joined	10:04
andres_	add-apt-repository not found... i'm on 9.04..	10:04
	przemoc: thx for the info	10:04
przemoc	Personally I have 1.6.5 on my debian lenny from backports, because lenny's repo has quite old 1.5.6.5, 1.7.0.1 on ubuntu karmic and own built 1.7.0 on slackware 12.1	10:06
	→ giallu joined	10:07
	→ Zephyrus joined	10:07
	→ gbacon joined	10:09
	← airfoil left	10:14
	→ airfoil joined	10:17
	→ Paraselene_ joined	10:18
	← airfoil left	10:18
	→ fmardini joined	10:21
	← SteveDekorte2 left	10:22
	→ palisv joined	10:26
	← orafu left	10:27
	→ orafu joined	10:27
	← bigpresh left	10:27
	→ bigpresh joined	10:30
	→ migge_ joined	10:30
	← migge left	10:31
	→ nadim joined	10:31
mr_daniel	Usually the work tree and the repo are in the same place. For example the work tree is ~/git/myproject/ and the repo is in ~/git/myproject/.git	10:32
	is there a way to 'deattach' them, so that ~/git/myproject/.git can be in another folder?	10:33
przemoc	mr_daniel: just move .git folder and set GIT_DIR environment to it	10:33
mr_daniel	also whats very important in that case: what is the role of GIT_DIR in that case? to which folder GIT_DIR points? and how sets GIT_DIR, and how uses it?	10:33
	I find settings environment variables is not very handy, is it instead possible to write the ne GIT_DIR in a git config file?	10:35
	← palisv left	10:35
	→ palisv joined	10:35
przemoc	mr_daniel: but git config file is in GIT_DIR :)	10:35
mr_daniel	ok :)	10:35
	just to checking if I understood: every work tree has a GIT_DIR, and GIT_DIR points to the repo of the work tree, right?	10:36
	← froschi left	10:37
przemoc	GIT_DIR is just an environment variable and if it is not set, then git goes up until it finds .git dir (or not)	10:37
	→ vals_ joined	10:37
przemoc	and possibly internally sets it then	10:37
	← apol left	10:38
	← tango_ left	10:38
przemoc	of course it can recognize if you're in a bare repo too, dunno what files/dirs it's looking for to be sure	10:39
mr_daniel	till now I have read a lot of docu, but was no able to find a single page explaining the purpose of GIT_DIR	10:40
	do you have a good page to read about it przemoc ?	10:41
	→ Mooseous joined	10:42
wereHamster	mr_daniel: why don't you try google? about the fourth hit is 'Introducing GIT_DIR environment variable.'	10:43
mr_daniel	right wereHamster , just found it myself ...	10:44
	→ futoase joined	10:44
	← futoase left	10:45
	→ e6La3BaNo joined	10:47
	→ kumbayo_ joined	10:47
	← kumbayo left	10:47
	← johnw left	10:47
	← hyperair left	10:48
	jargan → jast	10:49
	→ EricInBNE joined	10:49
	← Mooseous left	10:50
	→ Octalot joined	10:50
	→ yann joined	10:51
przemoc	mr_daniel: you can even detach .git/config via GIT_CONFIG and you don't have to be in the work tree, because you can set core.worktree or GIT_WORK_TREE	10:51
	→ dgl joined	10:51
przemoc	but AFAICS you cannot set repository location in git config (assuming GIT_CONFIG != $GIT_DIR/config), so using GIT_CONFIG alone won't work here	10:54
	→ kipras joined	10:55
	→ futoase joined	10:56
	← futoase left	10:56
	migge_ → migge	10:57
	→ radarek joined	10:59
	→ aresnick joined	10:59
	→ ph^_ joined	11:03
navap	Is there a better way to change the author of the two latest commits then this? http://coffee.geek.nz/how-change-author-git.html	11:04
	→ xrogaan joined	11:04
	→ Hackykid joined	11:06
wereHamster	navap: you couuld also use interactive rebase and then commit --amedn --author=...	11:07
	← e6La3BaNo left	11:08
radarek	navap: I would do this: create new branch on top HEAD^2 (we want to ommit 2 last commits), next: git cherry-pick each of those missing commit and afer each git commit --author ...	11:08
	rebasing seems to be simpler	11:08
	*after each git commit --ammend --author=...	11:08
navap	Would either of those keep the commits I created intact? I edited the same files in the commits and I want to keep the changes separate.	11:09
	← Mage left	11:10
	← dsop left	11:10
	← Cesario_ left	11:10
	← tg left	11:13
przemoc	navap: yes	11:13
	→ tg joined	11:13
	← bambule left	11:14
	← fragje left	11:15
	→ bambule joined	11:17
	← Yuuhi left	11:19
	→ voker57 joined	11:21
voker57	how to reject all local changes to branch and revert it to origin's version?	11:21
	like git -D master && git co -b master origin/master	11:21
Aides	git reset --hard ?	11:22
	git reset --hard origin/master to be specific	11:22
voker57	right, thanks	11:22
	→ yevlempy joined	11:24
	→ PointMan joined	11:24
	→ mydog2 joined	11:24
mydog2	morning..	11:24
	→ dsop joined	11:34
	← guardian left	11:36
	→ hypera1r joined	11:36
	→ Pupeno joined	11:36
	maattd → maattd\|away	11:37
	← Boriskr left	11:38
	→ Boriskr joined	11:39
	→ gebi joined	11:41
sitaram	andres_: I have never really used submodules, but I always thought the submodule mechanism had ways to allow different people to refer to the submodule using different URLs	11:45
	← hypera1r left	11:47
	→ corni joined	11:47
	→ parasti joined	11:48
	→ hypera1r joined	11:51
	← magn3ts left	11:53
	→ yhager joined	11:53
PointMan	how do I fetch a file from a diff branch that im currently on? For example if someone has merged a file update into master that I want into my branch (which will later be merged into master)	11:54
yhager	I installed gitolite on a server with a few existing repositories. How do I add them to gitolite without creating from scratch?	11:54
	← DrNick left	11:54
thiago_home	PointMan: git checkout otherbranch -- filename	11:55
PointMan	cool thanks	11:55
	→ nizuri joined	11:55
	← kipras left	11:55
	→ zoniq joined	11:55
	← hypera1r left	11:57
	← anjen left	11:59
	maattd\|away → maattd	12:01
	→ fragje joined	12:01
	← badboy__ left	12:04
	→ badboy_ joined	12:04
	← fragje left	12:05
	→ hypera1r joined	12:06
	→ fragje joined	12:06
	→ mklappstuhl joined	12:07
sitaram	yhager: (1) copy them at the OS copy level, into $REPO_BASE (default ~/repositories), (2) add them to conf/gitolite.conf, commit, push, (3) run easy install once again, exactly as done in the first place	12:08
	yhager: make sure they are bare repos whose directory names end in .git	12:09
yhager	(1) was already done, since I set up REPO_BASE to point them (/var/git)	12:09
	i just did (2) - what do I need (3) for then?	12:10
sitaram	yhager: set up the update hook	12:10
	yhager: without which none of the branch level controls will work	12:10
yhager	btw, kudos for the easy install script! very well made	12:11
sitaram	:)	12:11
	thanks	12:11
yhager	so I have set it all up - now where is the daemon that listens to incoming git:// connections?	12:12
	→ thrashold joined	12:13
thrashold	I have cloned some git repository some time ago. How do I get which revision it is?	12:13
jast	git log	12:13
yhager	sitaram: gitolite works with git daemon, or just ssh?	12:14
thrashold	jast: Thank you very much	12:14
sitaram	yhager: ssh only	12:15
	← Theravadan left	12:15
yhager	sitaram: ok, this is actually good. I never came about to install and run the git daemon. not sure what I need it for	12:16
sitaram	yhager: git daemon is read-only; for every repo it's either served, or not served, that's all. Most people in "internal" (corporate) settings dont need it. It's basically for public-read repos	12:16
yhager	what happens if I commit to gitolite managed repos using a different user (not the gitolite user)/	12:18
	← fragje left	12:18
yhager	ah, I guess I can't if those repos are 700 permissioned	12:19
	← parasti left	12:19
	← dsop left	12:20
	← palisv left	12:20
	→ dsop joined	12:20
PointMan	thiago: error: pathspec 'vendor/plugins/searchlogic' did not match any file(s) known to git. I get this error. Its a new directory I merged into master	12:21
	new file	12:21
sitaram	yhager: yeah that depends on the OS perms at that point	12:22
	can your userA routinely touch files in ~userB? :)	12:22
	→ Amitaibu joined	12:23
Amitaibu	yhager, ping	12:23
yhager	sitaram: I set up gitolite, and it works for my user, but another user gets "$REPOBASE/$repo not found in compiled config"	12:24
	← nizuri left	12:24
	→ engla joined	12:25
sitaram	yhager: from your previous question, I have to first ask what you mean by "another user" :)	12:26
	→ brizly1 joined	12:27
	← aspotashev\|eeepc left	12:27
yhager	sitaram: he ran 'git clone ssh://git@server:port/var/git/repo.git', and has his ssh key in keydir, and his name in gitolite.conf	12:28
	another user = Amitaibu	12:28
Amitaibu	hola! :)	12:28
	← brizly left	12:28
	→ fragje joined	12:28
sitaram	yhager: why does his clone URL appear to be including the $REPO_BASE value? it should not...	12:28
	← project2501a left	12:29
yhager	sitaram: because I need the port	12:29
	maybe we should try it with an alias in .ssh/config	12:30
	→ psoo joined	12:30
sitaram	yhager: what has the port got to do with it? please try ssh://git@server:port/repo.git	12:31
yhager	sitaram: I already tried, it assumes repo.git is at the root of the server.	12:32
sitaram	yhager: then he is bypassing gitolite; what does "ssh git@server info" show from his machine?	12:32
yhager	Amitaibu: ^^	12:33
	sitaram: adding the port info to .ssh/config worked. could it be some problem with the url parsing?	12:34
sitaram	yhager: as I said, what does "ssh -p port git@server info" show from his account?	12:35
	← mklappstuhl left	12:35
	→ ab3 joined	12:36
Amitaibu	hello amitai, the gitolite version here is v1.1-1-g42cc720	12:37
	you have the following permissions:	12:37
	R W REPO	12:37
	@ @ testing	12:37
	← priidu_ left	12:37
	→ t0rc joined	12:37
	← priidu left	12:37
	→ priidu_ joined	12:37
sitaram	Amitaibu: and yet "git clone ssh://git@server:reponame" does not work? can you paste the exact output?	12:38
yhager	oh, I think it was my mistake - I asked him to use the /var/git/repo.git syntax	12:39
	→ cuco joined	12:39
yhager	the reason I did that was that from my account it didn't work without it, but that is because I indeed bypassed gitolite with my key.	12:39
cuco	yhager: yo bitch	12:40
sitaram	Amitaibu: can you try ssh://git@server:port/repo.git	12:40
	(git clone I mean)	12:40
	(and the .git at the end is optional actually)	12:40
yhager	how can I possibly bypass gitolite?	12:41
	→ doobry joined	12:41
Amitaibu	sitaram, thanks for your help! seems to be working now	12:41
sitaram	yhager: because you installed gitolite, you have two keys there, the first one (probably id_rsa) for shell access	12:42
	→ _Vi joined	12:42
	← fmardini left	12:42
yhager	sitaram: oh, that explains it. Thanks!	12:43
	→ fmardini joined	12:44
	← TDJACR left	12:44
bremner	sitaram: I was thinking it might eliminate some key management confusion to have people create a second ssh/config stanza for gitolite-install, which the appropriate identityfile	12:46
	← t0rc left	12:47
	→ TDJACR joined	12:47
	→ Yuuhi joined	12:47
	→ matthieud joined	12:47
	← matthieud left	12:49
sitaram	bremner: "people" meaning the other users?	12:49
	→ Tureba joined	12:50
bremner	sitaram: people = "gitolite admin"	12:50
	to install/upgrade "./src/gl-install git gitolite-install admin-name"	12:50
yhager	is it gl-auth-command which gives me the shell access?	12:52
sitaram	bremner: I dont understand...	12:53
	yhager: not unless your userid is listed in $SHELL_USERS in the rc file (~/.gitolite.rc on the server)	12:53
selckin	B3	12:54
bremner	sitaram: my .ssh/config has "host gitolite-install\n identifyfile gitolite_setup" where gitolite_setup is the key that gives me shell access	12:55
sitaram	bremner: when would you set this up? think like a new user who's just downloaded (cloned) the gitolite source to his workstation...	13:00
	← kumbayo_ left	13:00
bremner	right, exactly then, during the initial install, before running easy-install for the first time. It only makes sense if you re-install/upgrade several times.	13:01
	→ pulse00 joined	13:01
	→ roop joined	13:02
	← roop left	13:02
sitaram	bremner: I'm afraid I don't agree. Let me explain...	13:02
pulse00	hi all. i have branch A which contains the folder "SomeFolder". everything in the folder is under version control. in branch B, the folder has been deleted, and replaced by a submodule. how can i merge B into A, so that the folder remains a submodule?	13:03
sitaram	bremner: easy install assumes normal shell access already exists, and uses it. Setting that up separately from gitolite makes it a pure ssh problem and people are able to deal with it independently. I don't want to make that step any more complicated than the bare minimum	13:04
	→ alexmchale joined	13:04
yhager	sitaram: I don't have my user as $SHELL_USERS in ~/.gitolite.rc on the server, but still I can ssh git@server	13:04
sitaram	yhager: doc/6	13:05
	→ mklappstuhl joined	13:06
bremner	sitaram: ok, well I don't feel strongly about it, it has just helped me a bit. Otherwise every time I seem to end up fussing with ssh agent more than necessary.	13:07
	→ patgodo joined	13:09
	→ abstortedminds joined	13:09
	→ aspotashev\|eeepc joined	13:09
yhager	sitaram: great documentation!	13:10
	← abstortedminds left	13:10
	← fragje left	13:13
bremner	sitaram: ok, now a serious question :). Does it make sense to you that new repos would only be created in the first of two urls pushed two? (this is part of my gitolite mirror experiments)	13:15
	→ project2501a joined	13:15
	← yevlempy left	13:15
	→ Tureba_ joined	13:15
bremner	pushing here is gitolite-admin repo.	13:15
	← Tureba left	13:17
	← cuco left	13:20
sitaram	yhager: I spent more time on docs than code, I am pretty sure :)	13:20
	bremner: what does that mean? I suspect you're confusing client side and server side stuff	13:21
	← pulse00 left	13:22
	Mike\|\|gone → Mike_lifeguard	13:23
bremner	sitaram: In order to have new repos created on the mirror, I added a second pushurl to the gitolite-admin repo. this seems to update both copies of the gitolite-admin repo ok, but the post-update hook is run only in the first copy.	13:25
	← frogonwheels left	13:25
sitaram	bremner: are both push urls pointing to the same user@server?	13:26
	→ kipras joined	13:26
bremner	sitaram: no, different ones (well, same user, same ssh_key, different servers)	13:27
sitaram	bremner: rather, how did you create the "mirror", is it on some other machine?	13:27
	ok	13:27
	← [reaVer] left	13:27
sitaram	bremner: rather, how did you create the "mirror", using easy install as before?	13:27
bremner	I did a second install with gl-easy-install, yes	13:27
	bremner pastebins his notes	13:27
yhager	sitaram: this is great thinking. docs are very important, a lot of people tend to forget that (me too, when I am on the writing side)	13:27
bremner	sitaram: notes in progress http://paste.debian.net/62919	13:28
sitaram	bremner: if the second url was also a gitolite instance created using easy install, that should also run its own post-update hook. In fact, the two servers dont even know about each other, so "first" and "second" doesnt make sense	13:29
	→ frogonwheels joined	13:30
bremner	sitaram: that was my understanding too. But I wonder if the local ref gets updated after the first push (whichever happens first) so the post-update hook is not triggered on the second.	13:30
	I re-installed the one that was not running the hook, just to be sure	13:30
	so maybe it is not a gitolite specific question, but rather about post-update hooks	13:32
sitaram	bremner: re last 4 lines of your notes paste, I hope you did not mess with the post-update hook of the gitolite-admin repo itself...	13:34
bremner	fwiw, here is what a push looks like http://paste.debian.net/62920	13:34
	← patgodo left	13:34
bremner	sitaram: no, just in one regular repo so far.	13:34
	I verified that both copies of gitolite-admin are the same using git ls-remote	13:35
sitaram	bremner: that was a push of the admin repo, not a regular repo, right?	13:37
bremner	correct	13:37
sitaram	bremner: and you're seeing it pushing to only one of your push urls...?	13:38
bremner	well, it is a puzzle, because like I say, both get the same SHA1 of HEAD	13:38
sitaram	bremner: I think this is neither a gitolite question nor a post-update hook question but a pushurl question	13:38
bremner	ack	13:38
	→ bluenovember joined	13:39
bremner	sitaram: ok, you're off the hook then :)	13:39
sitaram	bremner: nothing in the docs indicates that you can have multiple pushurls	13:40
bremner	I'll try to get more output from push. But first I need to commit something sensible.	13:40
sitaram	(I often "git commit --allow-empty -m empty" just to change the SHA)	13:40
bremner	sitaram: hmm. It is a bit ambigous, but git remote set-url has a --add option, which makes more sense for pushurl than fetch	13:41
	(git 1.7.x)	13:41
	→ sagsousuke joined	13:41
sitaram	bremner: ack, but you may have to wait for the real experts to pop in here or ask on the ML about multiple pushurls	13:42
	→ hipe_ joined	13:42
	← kipras left	13:43
	← Yuuhi left	13:44
	← NetHawk left	13:44
	→ incluye joined	13:45
	← hipe left	13:45
	← psoo left	13:45
Rolenun	hehe, you mean you've come across a question you cant answer sitaram :D	13:45
bremner	with --verbose, we can see both pushes are happening http://paste.debian.net/62922	13:45
	→ kipras joined	13:46
sitaram	Rolenun: why thank you! But it's not modesty that compels me to say I'm not an expert	13:46
	← alexmchale left	13:46
sitaram	it's honesty ;)	13:46
Rolenun	hehe	13:46
bremner	sitaram: ok, thanks for your help so far. I know a few people here at least know about multiple pushurls, so I'll see if anyone chimes in about hooks	13:46
sitaram	bremner: check your post-update hook on the admin repo on the slave; I'll bet it's your one-line code there, not gitolite's long perl code	13:47
Rolenun	to be honest, I'm not sure anyone is a real expert. Each learns the area they wish to. Trying to memorize 160+ commands with multiple arguments is time consuming and quite difficult :D	13:47
	← incluye left	13:47
sitaram	Rolenun++	13:47
	→ incluye joined	13:48
bremner	sitaram: basically 5 lines, ending with gl-compile-conf	13:49
sitaram	bremner: then I dont see how you're getting line 14 of http://paste.debian.net/62922	13:49
	bremner: somewhere in your master, slave, mirror, etc., you've got something pointing to the wrong place	13:50
bremner	you mean a bad hook?	13:51
Rolenun	which questions are still outstanding? You guys held a conversation above :D	13:51
sitaram	Rolenun: still holding it ;)	13:51
bremner	Rolenun: I have a repo with multiple pushurls, but it seems to only run post-update hook in one	13:51
sitaram	bremner: no idea; but something is wonky somewhere; check gitolite log files on both systems...	13:51
	→ karpar joined	13:52
Rolenun	aha. I wonder if it has to do with thread priority.	13:53
bremner	sitaram: interesting. No record of that push in .gitolite/logs. Maybe a key issue	13:53
	→ Groenleer joined	13:54
	← khmarbaise left	13:54
Ilari	IIRC, up to date messages are generated by client which will then break the connection before even sending update orders (so hooks do not run).	13:55
	← malumalu left	13:55
	→ alexmchale joined	13:55
bremner	to me, naively it looks like push 1 updates the local ref, which confuses push 2	13:55
	→ malumalu joined	13:56
	→ mklappst1hl joined	13:56
bremner	hmm. hang on, something changed when I deleted a key from authorized_keys	13:56
Ilari	I regularly do pushes to remotes with 2 urls. Don't have noticed any problems besides those remote helper reset issues (as first of those URLs usually has remote helper).	13:58
	→ Wizzup_ joined	13:59
	← mklappstuhl left	13:59
bremner	ah, no same problem. So Ilari, in particular post-update hooks run in both urls?	14:00
	← Wizzup left	14:00
Rolenun	he probably doesnt use hooks :D	14:01
	Wizzup_ → Wizzup	14:01
	← project2501a left	14:02
	→ NetHawk joined	14:03
	→ mahound joined	14:03
	← mklappst1hl left	14:04
	→ mklappstuhl joined	14:06
	→ mutex_ joined	14:06
Rolenun	bremner: if I can ask, what are you doing in the post-receive hooks? is it the same operation, etc	14:06
bremner	Rolenun: both are gitolite instances	14:07
sitaram	Ilari: "up to date messages are generated by client..." <-- can you elaborate (I didn't understand)	14:07
Rolenun	you're creating the instances, pushing to gitolite, etc?	14:08
	I am wondering if the issue may not be with git, but with the received result from the hooks	14:08
	→ matthieu_ joined	14:09
bremner	so if I reorder the pushurls in .git/config, then the other url has it's hook run	14:09
	Rolenun: the post-update hook is supposed to create new repos when added to the gitolite config	14:09
Rolenun	for instance, two simultaneous pushes to a server restricting the max number of connections causes one to go thump :D	14:10
bremner	Rolenun: I don't think its server specific, because I can choose which url runs its hook by putting it first	14:11
	and they are in different servers in different countries	14:11
Rolenun	ok, that's what I was wondering. different servers will help to narrow it down heh	14:11
	← maattd left	14:12
sitaram	19:39:22 < bremner> so if I reorder the pushurls in .git/config, then the other url has it's hook run	14:14
	← mutex_ left	14:14
sitaram	bremner: this says to me there's a possible bug somewhere in git... (or in my understanding)	14:14
	→ mutex_ joined	14:15
Rolenun	sitaram: yeah, it's sounding more and more like a thread priority issue	14:16
	something in the thread causes it to go thump heh	14:16
	← dsop left	14:17
Rolenun	could be what Ilari referred to earlier	14:17
	→ dsop joined	14:18
Ilari	What OS on client?	14:18
	→ tvw joined	14:19
	→ reprore joined	14:20
	→ gko joined	14:20
	← dsop left	14:23
	← sagsousuke left	14:24
	→ dsop joined	14:24
	← Hackykid left	14:25
	→ Hackykid joined	14:26
	→ frogonwheels_ joined	14:27
	← frogonwheels left	14:28
	→ reaVer joined	14:28
	← reaVer left	14:28
	→ reaVer joined	14:28
bremner	debian	14:31
	testing, if it matters	14:31
	← vivien` left	14:31
Ilari	So Linux (or close enough)...	14:32
bremner	yes	14:32
	git 1.7.0	14:33
	→ SpookyET joined	14:34
	→ kar8nga joined	14:34
	→ khmarbaise joined	14:38
	← mklappstuhl left	14:40
	← Amitaibu left	14:42
	← simplechat left	14:44
	← imajes left	14:46
	← alexmchale left	14:46
	→ alexmchale joined	14:47
	→ imajes joined	14:48
	→ jfkw joined	14:48
	← Hackykid left	14:48
	→ Hackykid joined	14:49
	→ mklappstuhl joined	14:51
	← mklappstuhl left	14:56
	← Hackykid left	14:57
	← tvw left	14:58
	← kpreid_ left	14:59
	← aresnick left	15:00
Ilari	bremner: What happens if you point second URL at some local filesystem path (bumping the second URL third)?	15:01
	← darwin_ left	15:03
	← mithro left	15:05
bremner	trying...	15:07
	← seanius left	15:07
	→ mklappstuhl joined	15:08
	← bigpresh left	15:08
bremner	Ilari: hook is still run only on first.	15:09
	← reaVer left	15:10
	← zenwryly left	15:10
	→ jasonn1234 joined	15:12
	→ kukks joined	15:12
	→ bigpresh joined	15:12
	← jasonn1234 left	15:13
	→ darwin joined	15:15
	hypera1r → hyperair	15:17
	→ kpreid_ joined	15:20
	→ mithro joined	15:23
	← hipe_ left	15:28
	← Tureba_ left	15:30
	→ Bass10 joined	15:32
	← karpar left	15:32
	← alexmchale left	15:34
	← kpreid_ left	15:34
	→ clintongormley joined	15:36
clintongormley	hi all - needing some basic git help	15:36
	i'm trying to write a script to update the current checked out branch from a remote repo	15:36
	so, for instance, on my demo website, i've created a 'demo' branch that tracks remotes/origin/demo	15:37
	so on my laptop, i make the changes i want to demo, push them to remotes/origin/demo, then on the demo webserver, i want to run a script which shows a log of the changes about to be put live, then updates the checkout	15:38
	conflicts should be ignored. it should just be updated to the latest version	15:38
	so i'm doing: git fetch; git log --online HEAD..origin; git reset --hard # (to remove temporary local changes); git pull	15:39
	but that complains about conflicts	15:39
	→ metalfan_ joined	15:40
metalfan_	hi	15:40
clintongormley	so i tried: git reset --hard origin , but that seems to reset to origin's master	15:40
	→ robins joined	15:40
cbreak	you can git reset --hard origin/demo	15:40
metalfan_	ive did a "git pull" but my drive was full. another git pull fails with: Merge with strategy recursive failed. what can i do?	15:40
cbreak	that would reset to origin's demo	15:40
	also, it should be --oneline, not --online	15:41
clintongormley	cbreak: ok - is there a way to do that for $current_remote_branch, without specifying it explicitly?	15:41
	(sorry yes, typo)	15:41
cbreak	hmm.	15:41
clintongormley	otherwise i need to do something like: git branch -vv \| grep '*' to get the remote tracking branch, no?	15:41
	← _Vi left	15:41
cbreak	you could use git config to find out what branch you are tracking	15:41
clintongormley	git config? git config -l doesn't seem to contain the relevant info	15:42
	← GarethAdams\|Home left	15:44
	→ kumbayo joined	15:44
cbreak	there should be something like branch.master.remote=origin that says that the master branch tracks origin	15:44
	→ kpreid_ joined	15:45
cbreak	and branch.master.merge=refs/heads/master, that says some more	15:45
	I've not worked with that directly though, so I am not sure if you can use it	15:45
clintongormley	yeah - i think the git branch -vv may be simplest	15:46
	thanks for the help	15:46
	← nanoo left	15:47
Ilari	bremner: The two others get up to date message? Even if the second repo is empty?	15:47
	← robinsmidsrod left	15:47
	robins → robinsmidsrod	15:48
	← khmarbaise left	15:49
Ilari	bremner: And with -v, does it show the branches as '= [up to date]' on other repos?	15:49
bremner	ah, no. I get [new branch] on the empty repo	15:49
Ilari	bremner: If you do modification, push to first repo by URL (not by remote) and then try to push by remote, you get that both are up to date or that only first is up to date?	15:51
	← metalfan_ left	15:51
bremner	Ilari: just the last one http://paste.debian.net/62948	15:51
	→ metalfan_ joined	15:53
metalfan_	sry, computer crashed	15:53
Ilari	bremner: 'git push -v gitolite:gitolite-admin master' followed by 'git push -v origin master' (with some changes) gives what?	15:53
bremner	http://paste.debian.net/62949	15:54
metalfan_	ive did a "git pull" but my drive was full. another git pull fails with: Merge with strategy recursive failed. what can i do?	15:54
jast	read the rest of the message, perhaps ;)	15:55
	→ ilteris_ joined	15:57
kristian42_	How do I reset remotes/origin/master to point to a different commit ?	15:58
	(Without pushing)	15:58
wereHamster	kristian42_: why would you want to do taht?	15:59
Ilari	kristian42_: 'git fetch origin'?	15:59
kristian42_	I'm trying to simulate a remote commit for a tutorial	15:59
Ilari	kristian42_: Just create the remote repo on local filesystem?	15:59
wereHamster	kristian42_: if you really insist: git update-ref refs/remotes/origin/master <the new commit>	15:59
kristian42_	Thanks, it's all about cheating ;)	16:00
Ilari	bremner: If you ls-remote gitolite:gitolite-admin and gitolite-slave:gitolite-admin, they show different values for refs/heads/master even after "successful" push.	16:01
	→ drry joined	16:01
SpookyET	man git log is 18 freaking printed pages	16:02
bremner	Ilari: all 3 show the same SHA1 with git ls-remote	16:03
selckin	SpookyET: agreed, how dare they document things	16:03
SpookyET	selckin: Too much functionality. Way too much crap to print some text.	16:04
	→ coolcat joined	16:04
	← ixti left	16:05
engla	man git-log	16:05
Gitbot	engla: the git-log manpage can be found at http://git.or.cz/man/git-log	16:05
Ilari	bremner: Try making a change, pushing it to origin (no URL push this time). Does ls-remote still claim the same values?	16:05
SpookyET	engla: ?	16:05
	→ ixti joined	16:05
engla	sorry I was lazy and just wanted the URL to the html docs	16:06
	→ reaVer joined	16:06
	← reaVer left	16:06
	→ reaVer joined	16:06
bremner	Ilari: yes, still the same	16:08
Ilari	bremner: Are you sure you don't have some hooks serverside that try to cross-mirror those gitolite-admin repos or something?	16:09
	bremner: Or that those SSH hostaliases don't point to same host?	16:10
bremner	facepalm	16:10
Ilari	bremner: They point to the same host?	16:11
bremner	yeah. /me is blushing	16:11
	seems to work fine. Double-checking with a meaningful commit.	16:13
	yup. it works fine. sorry to have wasted your time with that.	16:15
MetaCosm	bremner, we have all made those mistakes.	16:19
	Some of my favorites are when you are trying to explain to a dense coworker on the whiteboard something about the codebase, and midway through you go.. "ohh, ugh, nevermind"	16:19
	→ Paul___ joined	16:20
	← Paul___ left	16:20
bremner	Luckily I didn't get as far as thinking anyone was dense today :)	16:21
	← Rolenun left	16:24
	→ tedoc2000 joined	16:25
	← kpreid_ left	16:25
	← hgimenez left	16:26
	→ Rolenun joined	16:26
	→ Tureba joined	16:27
	→ hgimenez joined	16:27
	→ Erica1981new joined	16:29
Erica1981new	is here a some normal people??? 8)	16:30
	→ _Vi joined	16:30
locklace	no	16:30
	→ mr_danie1 joined	16:31
	← Groenleer left	16:31
	→ carolyn joined	16:31
	← mr_daniel left	16:32
Erica1981new	fuck	16:32
	were is a normal people?	16:32
locklace	#ubuntu	16:32
	← VVSiz left	16:33
Erica1981new	no, they banned me esterdaay	16:33
	→ koltroll joined	16:34
Erica1981new	koltroll hi	16:35
	→ yevlempy joined	16:36
koltroll	Erica1981new, hi (?) :)	16:36
	← tedoc2000 left	16:37
	→ nanoo joined	16:38
Erica1981new	koltroll lets kill all human???	16:40
	→ koliz joined	16:40
	← koliz left	16:40
koltroll	Erica1981new, that sounds like an awful idea!	16:41
	also I'm most offended by you even asking me that type of question. assuming I would be some sort of - god forbid - terrorist.	16:42
Erica1981new	koltroll it was just test. so youre nit a real troll. ahahahahaha	16:43
koltroll	Well. Fine.	16:43
	← Mike_lifeguard left	16:44
Erica1981new	What mean "kol"?	16:44
	MetaCosm sees if he can set someone on fire with is mind...	16:44
MetaCosm	is it getting warm Erica1981new?	16:45
	bremner recommends /ignore for hours of peace and quiet	16:45
	→ Mike_lifeguard joined	16:45
koltroll	heh	16:45
Erica1981new	MetaCosm no)) i am not warm, i am just cool	16:45
	MetaCosm failed again.	16:46
koltroll	bremner, I'd say that's to chicken out, to go the /ignore way.	16:46
	however, you are right.	16:46
MetaCosm	attempt #817, marked as a failure, but I am making a note here, "huge success", I will do what I must, because I can!	16:46
Erica1981new	KTO TUT GOVORIT PARUSKI??????? CHISTA INTERESNA)))) KTO TUT GOVORIT PARUSKI??????? CHISTA INTERESNA)))) KTO TUT GOVORIT PARUSKI??????? CHISTA INTERESNA)))) KTO TUT GOVORIT PARUSKI??????? CHISTA INTERESNA))))	16:47
cbreak	I know why they banned you, you idiot	16:47
MetaCosm	laugh.	16:47
	← WebDragon left	16:47
Erica1981new	cbreak what did you say?	16:48
	→ hipe joined	16:48
	← fr left	16:49
	→ alezandro joined	16:49
	→ fr joined	16:50
	→ petercoulton joined	16:50
	← Erica1981new left	16:50
	→ froschi joined	16:54
	← sunaku left	16:56
metalfan_	jast, the message tells me that flexyz would be overwritten	16:57
	but since i know that everything is fine with overwriting, how do i force it?	16:58
	yevlempy asks if anyone can give him a link regarding making a patch and submitting that to his project	16:58
	yevlempy is new to git	16:59
	← thiago_home left	16:59
metalfan_	ive did a "git pull" but my drive was full. another git pull fails with: fatal: Untracked working tree file xyz...(i can remote this file, then the name changes) Merge with strategy recursive failed. how can i get rid of the last changes from pull?	17:01
bremner	yevlempy: git format-patch and git send-email are two useful commands.	17:01
yevlempy	bremner, ok	17:01
jast	metalfan_: if you're sure that all your work is safely committed, just delete the file and try again	17:02
	→ scarabx joined	17:03
	→ robinr joined	17:03
	← migge left	17:04
	→ stringo0 joined	17:05
	→ jasonn1234 joined	17:05
	→ thiago_home joined	17:06
	→ kpreid_ joined	17:08
	← ttvd left	17:08
metalfan_	the last commit was from another repo, so yes. its all fine	17:09
	jast, there are quite some files in the last pull that could not be downloaded because of not spac	17:09
	e	17:09
	← fmardini left	17:09
	→ VVSiz joined	17:10
	← EricInBNE left	17:12
jast	git should just fetch them the second time	17:13
	I gotta go though	17:13
metalfan_	thjx	17:13
	thx	17:13
	← Adaptee left	17:14
	← petercoulton left	17:15
	→ a2ron joined	17:15
	→ cloudhead joined	17:17
	→ webchick joined	17:18
	← jast left	17:18
	← disappearedng left	17:18
	→ mr_daniel joined	17:19
	← mr_danie1 left	17:20
	→ Adaptee joined	17:21
	← hipe left	17:21
	→ hipe joined	17:22
	→ randy7 joined	17:22
	kmap_ → kmap	17:22
	→ Yuuhi joined	17:23
	→ fcuk112_ joined	17:24
	← fcuk112_ left	17:24
	→ fcuk112_ joined	17:25
	→ dreiss joined	17:25
	← fcuk112_ left	17:25
	← ddp left	17:27
	← kpreid_ left	17:27
	→ fcuk112_ joined	17:28
	→ poseid joined	17:28
	→ t0rc joined	17:29
	← fcuk112_ left	17:30
	→ disappearedng joined	17:33
	← tetha left	17:33
	→ tetha joined	17:34
	← t0rc left	17:35
	← incluye left	17:36
	→ kpreid_ joined	17:36
	→ incluye joined	17:36
	← incluye left	17:37
	→ incluye joined	17:37
	← Fullmoon left	17:39
	→ Cesario_ joined	17:39
	← jasonn1234 left	17:40
randy7	hi guys, can I ask a question?	17:41
	← incluye left	17:42
	← twmi left	17:42
randy7	(newbie) I have a repository on disk, and would like to upload it to github. what do I need to do to connect the two?	17:42
	I also generated a secure key, but don't know how to use it yet for transfer	17:43
MetaCosm	randy7, github has really good docs on all the steps	17:43
	→ aresnick joined	17:43
	→ agile joined	17:44
randy7	even if I created it before I had the github account?	17:44
	ok, i'll take another look.	17:44
	Thanks	17:44
	← zoniq left	17:44
	← anxiolytic left	17:45
MetaCosm	randy7, there are lots of great docs, learn.github.com	17:47
	← a2ron left	17:49
	← hgimenez left	17:49
	Mike_lifeguard → Mike\|\|gone	17:50
	→ hgimenez joined	17:50
	← Vortex35 left	17:51
	→ jMyles joined	17:51
	→ anxiolytic joined	17:51
	→ LeMoNiCe joined	17:52
	→ jmcantrell joined	17:52
LeMoNiCe	hi	17:52
	is there a way to use versioning with my .PHP files? like i write: $Id$ and git replaces it with a version number	17:53
offby1	yes, but it's widely considered a bad idea	17:53
	→ psoo joined	17:53
offby1	although to be honest I can never remember why.	17:53
	Linus has a whole rant about why it's bad	17:54
	← mklappstuhl left	17:54
	→ ruslang joined	17:54
Mage_Dude	Maybe because the file 'version' is already tracked by a SHA1 value?	17:54
	→ nanooo joined	17:56
offby1	maybe; honestly I can't remember the gist.	17:56
	← unijunction left	17:56
offby1	Certainly LeMoNiCe (and I) have a real problem, for which versions-embedded-in-the-files is an appealing solution: you want to know which version of a file you're looking at!	17:57
	→ twmi joined	17:57
offby1	I usually wind up having some step in my build process that sticks the commit ID someplace convenient.	17:57
	Like I have a web app which has a "version" command: when you type "version" at it, it echoes back the SHA1 of the commit.	17:57
	that seems to be enough for me	17:57
LeMoNiCe	mh	17:57
	ok	17:57
offby1	of course that's useless if you deploy a version that never existed in the repo: i.e., if you've edited some of the files "in production"	17:58
	→ codeswing joined	17:58
offby1	but with some care you can avoid that, I suppose	17:58
	← bambule left	17:58
	← nanoo left	18:00
	→ __iron joined	18:00
	→ nanoo joined	18:00
	← nanooo left	18:01
	← randy7 left	18:01
	← thiago_home left	18:02
	← _iron left	18:03
	← disappearedng left	18:04
	→ Vortex35 joined	18:04
	→ tvw joined	18:08
	→ smuf joined	18:08
	← kar8nga left	18:09
	→ khmarbaise joined	18:10
	→ BeerSerc joined	18:10
	→ DrNick joined	18:10
	→ mklappstuhl joined	18:10
	→ unijunction joined	18:11
	→ alexmchale joined	18:11
	← sdboyer left	18:12
	← ab3 left	18:12
	→ sdboyer joined	18:13
	← sdboyer left	18:13
	→ sdboyer joined	18:13
	← alexmchale left	18:14
	← gko left	18:16
	→ jast joined	18:17
	← SpookyET left	18:17
	Mike\|\|gone → Mike_lifeguard	18:17
	→ jasonn1234 joined	18:18
	← voxpelli left	18:19
	→ mrbrdo joined	18:19
mrbrdo	how can i "untrack" a file but without marking it as deleted (leave it in repo), just not update it with my changes	18:19
	i put the file in .gitignore but it's already being tracked.. if i do git rm --cached it shows at deleted (i don't want to delete it in repo)	18:20
	→ Pewpewarrows joined	18:20
	→ disappearedng joined	18:21
	→ palisv joined	18:25
	→ thiago_home joined	18:25
	← aresnick left	18:26
offby1	I don't think that's possible.	18:29
MetaCosm	mrbrdo, are you trying to stash it?	18:29
	to go do something else?	18:29
offby1	If git is tracking the file, and you've made changes to that file ... then git's gonna wanna tell you all about those changes	18:29
	→ seanius joined	18:29
offby1	git will assume that you eventually will want to commit those changes, or throw them away	18:29
	mrbrdo: what kind of file is this? Is it, by any chance, a configuration file for a web app?	18:30
mrbrdo	yes	18:30
offby1	score!	18:30
	← Sonderblade left	18:31
offby1	I can't think of an easy way; we have the same situation at work, and every so often people accidentally check in their hacked-up meant-only-for-their-private-use versions of those files :-\|	18:31
	ideally, we'd change our project so that we only checked in _templates_ of those files, and had a sort of "build step" that filled those templates in with sensible values	18:31
	that would solve the problem, but it'd be a fair amount of work	18:31
mrbrdo	well	18:32
	→ amerine joined	18:32
offby1	what I do is: I keep my hacked-up private versions in a stash, and regularly apply that stash	18:32
	then I have to try hard not to commit those changes	18:32
mrbrdo	i used git update-index --assume-unchanged file..	18:32
	it seemed to work	18:32
offby1	huh	18:33
	interesting	18:34
jast	--assume-unchanged is not completely straightforward... be sure to read the manpage	18:35
	I think that git reset for example kills the flag	18:36
	→ alexmchale joined	18:37
	← unijunction left	18:39
	→ ntoll joined	18:39
	← hipe left	18:39
	→ kar8nga joined	18:39
	← jelmer left	18:41
	← mklappstuhl left	18:41
	← matthieu_ left	18:41
	→ jelmer joined	18:42
	→ patcito joined	18:43
	← v0n left	18:43
	← jasonn1234 left	18:44
	→ nadavoid joined	18:44
	→ bamccaig joined	18:46
bamccaig	I just committed to what I thought was branch 'string', then pushed string to origin. Except the change I committed didn't go.	18:47
	← mebus left	18:47
wereHamster	'git branch' to see on which branch you are	18:47
bamccaig	Right. git status said I was on no branch...	18:48
wereHamster	bamccaig: faq detached	18:48
cbreak	git status is most likely right	18:48
	did you check out a tag for instance? or a hash directly?	18:48
	or are you on a submodule?	18:48
	→ flaguy48 joined	18:48
cbreak	if you are not on a branch, everything you commit vanishes	18:48
	(write down the hash if you want to salvage it)	18:49
	→ maattd joined	18:49
	← bernie left	18:49
jast	bamccaig: http://sitaramc.github.com/concepts/detached-head.html	18:50
bamccaig	wereHamster: Thanks.	18:50
	← alexmchale left	18:50
	← poseid left	18:50
bamccaig	cbreak: I should still have it in my terminal. >_<	18:50
	jast: Already found it, thanks. ^_^	18:50
jast	right :)	18:50
wereHamster	there's always the reflog...	18:51
cbreak	first thing: make a tag out of your detached head so it won't be GCed by accident	18:51
	right, true	18:51
offby1	well, "vanishes" is putting it somewhat strongly	18:51
	← carolyn left	18:52
	→ meanburrito920_ joined	18:53
	← reprore left	18:53
	→ Alagar joined	18:54
	← gunmuc left	18:54
	→ unijunction joined	18:54
	← Yuuhi left	18:55
	→ cedricv joined	18:57
	← patcito left	18:57
	→ aresnick joined	18:58
	← mcdebugger left	18:59
	→ Yuuhi joined	18:59
	→ incluye joined	19:00
	→ mcdebugger joined	19:00
cxreg	sometimes there's been talk about supporting "reverse" rebase, or cherry-pick on a range	19:00
	anyone know if either of those ever went anywhere?	19:00
	→ tedoc2000 joined	19:01
	→ gunmuc joined	19:01
	offby1 stares blankly	19:02
wereHamster	'reverse' rebase?	19:02
	← aresnick left	19:02
offby1	I seem to recall I saw them getting on a train bound for Baton Rouge	19:02
	→ jceb joined	19:03
MetaCosm	wereHamster, he want to rollback his changes... in reverse, against a branch the changes never happened on, in order to make it look, like the other branch, before the change	19:04
	obvious really.	19:04
	← thiago_home left	19:06
wereHamster	git merge -s ours .. ?	19:08
	← darwin left	19:09
	← cedricv left	19:09
	← psoo left	19:11
	→ cedricv joined	19:12
cxreg	wereHamster: instead of moving the head of the branch being rebased, move the head of the destination	19:12
	→ darwin joined	19:13
cxreg	it'd be about the same as saying "git cherry-pick foo..bar", if that was possible	19:13
	→ bcardarella joined	19:14
	→ psoo joined	19:14
MetaCosm	wereHamster, I was kidding :) I have no idea what he wants :)	19:15
	cxreg, lets just use the term "replay" -- you want to replay the changes from <what branch> onto <what other branch>?	19:15
cxreg	here's a previous thread talking about it	19:16
	http://lists.zerezo.com/git/msg627749.html	19:16
	i did not invent the terminology	19:16
MetaCosm	I know you didn't, it was just for my gratification	19:16
	→ rado1 joined	19:17
offby1	sounds like a kind of rebasing	19:17
metalfan_	ive did a "git pull" but my drive was full. another git pull fails with: fatal: Untracked working tree file xyz...(i can remote this file, then the name changes) Merge with strategy recursive failed. how can i get rid of the last changes from pull?	19:17
cxreg	normally "git rebase --onto foo bar baz" moves the head "baz" to the end result	19:17
	the "reverse" version would simply move "bar" instead	19:17
	as long as that's a local branch	19:17
	→ thiago_home joined	19:17
	vals_ → tango_	19:17
	← tedoc2000 left	19:18
cxreg	metalfan_: did the pull succeed?	19:18
metalfan_	not, drive was full	19:18
	no	19:19
cxreg	did you have any local uncommitted changes?	19:19
metalfan_	no	19:19
cxreg	try "git reset --hard"	19:19
MetaCosm	cxreg, would baz be a branch of foo, or a branch of bar?	19:19
metalfan_	the last commit to the "main" repo was from another repository	19:19
cxreg	MetaCosm: it doesn't matter	19:19
	but most likely of bar	19:20
metalfan_	cxreg, still the same error	19:20
	→ alexmchale joined	19:20
jhelwig	metalfan_: If "git show ORIG_HEAD" looks like where you were before the pull, you can do "git reset --hard ORIG_HEAD", then do the pull over again.	19:20
	← bamccaig left	19:21
cxreg	i wonder if the pull partially succeeded and left a file that would have appeared in the new commit in his working tree, and now that file conflicts	19:22
metalfan_	maybe i should post more infos: three pcs, from one pc i did a commit to the "main" repo. from pc2 i did a pull which failed because of insufficient disk space. now i need to get the changes that this pull did out of the system somehow	19:23
jhelwig	cxreg: So the new file got created, before the index failed to be updated? Sounds very possible.	19:24
metalfan_	cxreg, thats my idea too. because git asks me that some files would be overwritten	19:24
cxreg	metalfan_: if you really dont have any changes at all you want to keep, do "git reset --hard; git clean -xdf"	19:24
metalfan_	to what does git reset reset to?	19:24
cxreg	HEAD	19:24
metalfan_	from the local repo?	19:24
cxreg	er, yes. whatever commit is currently checked out. you can check this with "git show HEAD"	19:25
metalfan_	seems to work	19:26
	have to read another howto about git, still dont really understand what i just did	19:26
	thx	19:26
cxreg	np	19:26
badboy_	http://progit.org/book/	19:26
cxreg	mainly, the reset changed tracked files in the index and work tree	19:26
	the clean removed untracked files	19:26
metalfan_	thats new, will take a look	19:26
cxreg	it sounds like the pull introduced a new file, which is "untracked" in the previous commit	19:27
	← AkumaStreak left	19:27
	← rado1 left	19:28
metalfan_	the commit from that local repo?	19:30
cxreg	whichever	19:30
	← mydog2 left	19:30
	→ nikki93 joined	19:31
nikki93	Why does git appreciate complexity over intuitiveness?	19:31
MetaCosm	cxreg, I am trying to grok you intentions (mainly so I can learn a little more about rebase)... the two ways I saw in the docs where -> http://gist.github.com/324585 ... you want a third way, but I don't exactly understand what you want the result to look like.	19:31
	← froschi left	19:31
MetaCosm	nikki93, so you don't lack the feature you absolutely need.	19:31
cxreg	nikki93: nice troll	19:31
nikki93	Nah, I just want to see what to tell my friend who told me this. :)	19:31
	← scarabx left	19:32
MetaCosm	nikki93, tell you friend RCS rocks, and (s)he should use it, everything past straight RCS has been overcomplicated nonsense.	19:32
	→ mklappstuhl joined	19:32
cxreg	MetaCosm: take the first graph, and say you dont want to move topic, but you want to end up with master pointing to where topic was moved to	19:33
nikki93	Is git really 'more complex' and 'less user-friendly' than Hg?	19:34
cxreg	those 2 "versions" really aren't any different	19:34
	nikki93: no	19:34
	→ Hackykid joined	19:34
MetaCosm	nikki93, not too much worse, and Hg is retarded when it comes to branching.	19:34
cxreg	nikki93: that was plausibly true 2 years ago	19:34
MetaCosm	ie: clone over branch adds a lot of complex for my dev (lots of new directories)	19:35
nikki93	Check out the last post on here: http://www.ogre3d.org/forums/viewtopic.php?f=4&t=53129&start=275 what do you think?	19:35
cxreg	git has made very significant usability improvements since 1.5.0	19:35
	we're on 1.7.0 now	19:35
	← koltroll left	19:35
MetaCosm	I taught someone the basics in about an hour	19:36
	he was using it (branching, merging, sharing) in an hour	19:36
cxreg	nikki93: it smells of bias, and not facts	19:36
	its hard to argue with someone like that	19:36
	it's true that Hg is a little more polished on Windows	19:36
	but msysgit is pretty good	19:37
MetaCosm	cxreg, msysgit is "daily usable" for sure.	19:37
	not as fast ... but depends what you compare it to	19:37
	→ abadr joined	19:37
MetaCosm	greased lightning next to svn...	19:38
	← abadr left	19:38
MetaCosm	but, slower than native git on linux	19:38
nikki93	Also - if git is 'different' in some way - is it just because it 'wants to be different' or it's just 'it's philosophy'	19:38
	← alexmchale left	19:38
nikki93	cxreg: Yeah I use msysgit too..	19:38
MetaCosm	... neither.	19:38
	→ GarethAdams\|Home joined	19:38
cxreg	nikki93: they're more similar than different, really	19:38
nikki93	I haven't had problems for it.	19:38
cxreg	the biggest advantage git has over Hg is the index	19:38
	and then branching	19:38
MetaCosm	git didn't try to clone anything, it was built to meet the needs of a group of people, everything it does is based on that.	19:39
	the only pattern I have heard mentioned consistantly is "what wouldn't CVS/SVN do"	19:39
cxreg	heh	19:39
MetaCosm	cxreg, back to your example, since I am now full on playing with rebase (I generally just merge unless working with git-svn)	19:40
offby1	nikki93: git's UI and documentation are certainly both complex. But its internal model is quite simple; I've never really understood hg's internal model (although to be fair I haven't spent that much effort trying)	19:40
Mike_lifeguard	that it is written in python is actually pretty nice	19:40
offby1	nikki93: and I've found that, now that I understand git's innards, I don't really care that the UI and docs are complex; I can figure out what I need	19:40
Mike_lifeguard	makes it very easy to convince people it doesn't suck on Windows	19:40
tango_	honestly, though, I never found git more complex than mercurial	19:41
cxreg	a year ago or so, there was a "10 things that are better in Hg than in git" blog post	19:41
MetaCosm	cxreg, you want to leave topic right where it is, a branch of next... but you want to somehow route master to it?	19:41
offby1	Mike_lifeguard: is hg reasonably fast on Windows? Git and Subversion both (in my experience anyway) are painfully slow	19:41
tango_	(and I was actually an hg fan a long time ago)	19:41
MetaCosm	offby1, you still find msysgit painfully slow (or cygwin git?)	19:41
cxreg	at the time, some of them were simply not true	19:41
offby1	MetaCosm: yep	19:41
cxreg	within several months, none of them were true	19:41
offby1	MetaCosm: about as slow as subversion.	19:41
Mike_lifeguard	offby1: I don't know, I don't use Windows.	19:42
MetaCosm	offby1, I haven't found that to be the case (git-svn, so a fairly clear mapping) with about 25k files.	19:42
	offby1, some operation that would be a good example case?	19:42
offby1	MetaCosm: I wonder if you're using a very different version than I am. I am unable to upgrade, unfortunately, due to a bug in git-svn that prevents it from properly reading our svn repo :-\|	19:42
	→ ttvd joined	19:42
cxreg	MetaCosm: in that gist you linked, look at the "after" in use-case 1. the plae where it says "topic", it would say "master"	19:43
	← ttvd left	19:43
offby1	MetaCosm: any command at all. For example, "git status" takes ... I dunno, ten or twenty seconds.	19:43
	← clintongormley left	19:43
nikki93	The problem they have with git and Linux it seems is 'evangelists'.	19:43
	I don't want them to blame git and Linux for that. :(	19:43
offby1	On essentially the same repo, on *nix (on a fast box, admittedly; but not THAT much faster) it's way under one second	19:43
	→ ttvd joined	19:43
MetaCosm	offby1, 25k files, about 1.8 seconds on my windows 7 box.	19:43
bremner	nikki93: I don't give a damn what software anyone uses. Does that help? :)	19:44
offby1	MetaCosm: huh. I wish I know how you were doing that	19:44
MetaCosm	git version 1.6.5.1.msysgit.1	19:44
offby1	I don't have my Windows version handy, but I suspect my version isn't much older than that.	19:44
	In fact it might be exactly the same.	19:44
nikki93	bremner: Unfortunately I have an emotional attachement to these people and Ogre because I've learnt computer graphics/math/3d stuff and programming with it. ^_^	19:44
	bremner: Started making games when I was 12 around 5 years ago. Great fun. At the cost of spamming I'm gonna link to my site here: http://nikki93.github.com/	19:45
MetaCosm	offby1, maybe I got some caching already wired up and it is slow on a cold read and I just didn't notice, let me find a project I haven't touched in a long time	19:45
cxreg	nikki93: python evangelists are way worse than git ones	19:45
nikki93	bremner: Check out 'GraLL 2', my latest project.	19:45
	offby1 points at nikki93 and intones: Spammer!	19:45
nikki93	offby1: ^_^	19:45
	offby1 hurls a heavy dusty rulebook	19:45
	nikki93 ducks	19:45
	→ owen1 joined	19:45
offby1	no, no: nikki93 _evades_ :)	19:46
	nikki93 points at offby1 and laughs	19:46
owen1	after modifying .gitignore, do i need to commit it?	19:46
	offby1 can't. They appear to be cursed.	19:46
offby1	owen1: yes	19:46
	nikki93 throws a smoke bomb and escapes	19:46
owen1	ok, thanks	19:46
offby1	if you want other people to benefit from it.	19:46
	← bcardarella left	19:46
owen1	offby1: but if i don't?	19:46
nikki93	owen1: It'll be... 'local'.	19:46
owen1	nikki93: but it shows as modified.	19:47
	(on git status)	19:47
nikki93	owen1: So you haven't commited the changes yet. It's just modified.	19:47
owen1	i mean, the .gitignore itself	19:47
offby1	owen1: you can put similar information into .git/info/exclude	19:47
nikki93	owen1: Other people that pull won't have the .gitignore then, and if they push, boom.	19:47
offby1	then git won't keep nagging you about uncommitted changes	19:47
owen1	nikki93: so commiting my gitignore will not effect others?	19:48
	each one has it's own gitignore?	19:48
	← kumbayo left	19:48
nikki93	.gitignore is a 'normal file'.	19:49
	→ mydog2 joined	19:49
offby1	owen1: yes. If you commit .gitignore, everyone will see it.	19:49
	Just like any other file in your work tree	19:49
owen1	offby1: ok	19:49
offby1	the only thing that's special about .gitignore is that git consults it when considering which files to ignore.	19:49
	← bentob0x left	19:50
MetaCosm	cxreg (thanks for bearing with me), so, in the after, now topic is called master (has become master), so what becomes of the formerly named master branch (top row in ascii art)?	19:50
owen1	got it, thanks	19:50
	→ froschi joined	19:50
cxreg	MetaCosm: it remains as ancestor commits of the new master.	19:50
MetaCosm	so it becomes one unbroken row, correct?	19:51
cxreg	pretty much yeah	19:51
	→ alexmchale joined	19:51
cxreg	and topic is left where it was. not moved, not deleted.	19:51
MetaCosm	just... copied	19:51
	← crankharder left	19:52
cxreg	well sure. but that's what rebase and cherry-pick do.	19:52
	→ kumbayo joined	19:52
MetaCosm	that was my understand, which is why I am trying to grok the "reverse" part	19:53
	← unijunction left	19:53
	← nikki93 left	19:54
	→ AkumaStreak joined	19:54
cxreg	MetaCosm: it only refers to which head is moved, not what work is actually done	19:54
MetaCosm	standard rebase, or what you want?	19:55
	→ jdt joined	19:55
cxreg	huh?	19:56
	it could be spelled --into, as in "git rebase --into foo bar baz".	19:57
	offby1 has never understood --onto	19:57
offby1	I don't even understand the problem it solves	19:58
	← incluye left	19:58
locklace	i think describing as the same as the notional "cherry-pick a..b" makes the intended meaning obvious	19:58
	← voker57 left	19:58
cxreg	locklace: yes. it would be fabulous if cherry-pick actually did that.	19:58
	jdt → incluye	19:58
locklace	cxreg: yeah, cherry-pick should support ranges and lists	19:58
cxreg	offby1: --onto specifies where to start applying the commits. it defaults to <upstream>	19:58
	offby1: mainly, <upstream> and <branch> are used to identify the set of commits to rebase, and --onto says where to deliver them	19:59
	offby1: git pull --rebase uses that form in a particularly clever fashion	20:00
	← incluye left	20:00
MetaCosm	cxreg, would an example use-case be pulling a fix out of a branch and applying it to master (maybe that fix was over a dozen or more commits?)	20:02
	→ Hackykid_ joined	20:02
	← Hackykid left	20:04
cxreg	MetaCosm: something like that, yeah	20:04
MetaCosm	cxreg, can't use use format-patch and am to do the exact thing you want?	20:04
locklace	having to do that is a hack	20:05
	→ Twisted_Mentat joined	20:05
	→ mebus joined	20:05
cxreg	there's at least 4 ways to do it	20:06
	having a one step process would be great	20:06
MetaCosm	I agree, cause just talking about has me all confuzzled.	20:06
	you could cherry-pick one by one (scripted probably)	20:07
	you can format-patch, then am it...	20:07
	→ kingfishr joined	20:07
MetaCosm	what are other two ways?	20:07
cxreg	you could rebase the target branch on top of master, check out master, and merge the previous rebase	20:07
	and i forget what i was thinking of as the fourth one :)	20:08
offby1	cxreg: well, that was a clear explanation! Thanks	20:08
	← thiago_home left	20:08
cxreg	offby1: hope so ;)	20:08
MetaCosm	ugh, can't decide if #2 or #3 is more hacky	20:08
	→ unijunction joined	20:09
	nadavoid → nadavoid_afk	20:09
	→ Hackykid joined	20:10
	← Hackykid_ left	20:10
	→ greg8si joined	20:10
	→ fullerton joined	20:11
	→ aryounce joined	20:11
cxreg	MetaCosm: if i was doing #3, i'd probably /actually/ do "git rebase --onto destination upstream some_branch^0; git checkout destination; git merge HEAD@{1}"	20:12
	heh	20:13
	→ rudi_s_ joined	20:13
cxreg	the ^0 is key, because it detaches HEAD to avoid moving some_branch during the rebase	20:13
	cxreg wonders if "^0^" is valid...	20:14
cxreg	haha, it is. \m.	20:14
	\m/ even	20:14
	offby1 wonders if o/ is valid...	20:15
	← Hackykid left	20:15
cxreg	git reset --hard \o/	20:15
	← nadavoid_afk left	20:15
	← ruslang left	20:15
	← rudi_s left	20:16
	→ Hackykid joined	20:16
	← psoo left	20:17
	→ dj_tjerk joined	20:17
	→ Ilari joined	20:17
	→ thiago_home joined	20:18
	← thiago_home left	20:18
	→ thiago_home joined	20:18
	← yann left	20:20
	→ bambule joined	20:20
	→ Hackykid_ joined	20:22
	← Hackykid left	20:23
	← fullerton left	20:24
	→ intripoon joined	20:24
	→ Theravadan joined	20:25
	← Hackykid_ left	20:25
	← mrbrdo left	20:26
	→ fullerton joined	20:26
	← amerine left	20:27
	← ajpiano left	20:30
	← radarek left	20:31
	→ ajpiano joined	20:33
	→ napster_123 joined	20:34
napster_123	hi, is there any server that is basic and simple to host a local git project ?	20:35
	only local users could pull and push from ?	20:35
	← unijunction left	20:35
	→ WaltzingAlong joined	20:36
unregistered	napster_123: on a unix. you typically make a bunch of users with write access to a folder and place a repo in the folder	20:37
	← kelvie left	20:37
napster_123	unregistered, is there any tool on windows ?	20:37
unregistered	git comes with "git serve"	20:38
	→ JonathanT joined	20:38
	→ lightcap_ joined	20:38
napster_123	unregistered, ok let me try it	20:38
	← lightcap_ left	20:38
napster_123	unregistered, it says... serve is not a git command	20:39
unregistered	yeah sorry i was thinking of hg	20:40
	→ Hackykid joined	20:40
	→ kelvie joined	20:40
napster_123	unregistered, that's exactly what i was looking for	20:41
	Mike_lifeguard → Mike\|\|gone	20:41
	← thiago_home left	20:42
	→ thiago_home joined	20:42
unregistered	i'm not sure how git:// protocol servers are done. i'm surprised no one has chimed in yet.	20:43
	← Hackykid left	20:43
	→ Hackykid_ joined	20:43
locklace	napster_123: man git-daemon	20:43
Gitbot	napster_123: the git-daemon manpage can be found at http://git.or.cz/man/git-daemon	20:43
	← kelvie left	20:43
napster_123	locklace, does it work on windows ?	20:43
	→ aresnick joined	20:44
	← d0k left	20:44
	← mutex_ left	20:44
	← locklace left	20:44
	→ mutex_ joined	20:45
	← alexmchale left	20:45
	→ richcollins joined	20:45
	→ kelvie joined	20:47
	→ locklace joined	20:47
	← Mage_Dude left	20:48
	← nanoo left	20:48
	← codeswing left	20:49
	→ zenwryly joined	20:50
	→ unijunction joined	20:51
	← tildeequals left	20:52
	← khmarbaise left	20:53
	→ alexmchale joined	20:54
	→ mutex__ joined	20:54
	← mutex_ left	20:54
	→ Mage_Dude joined	20:54
	← jceb left	20:55
	← dreiss left	20:55
	← malumalu left	20:56
	Mike\|\|gone → Mike_lifeguard	20:56
	→ scarabx joined	20:56
RobertMeta	napster_123, you can just use a samba (smb) share	20:57
	napster_123, git uses minimal file locking	20:57
napster_123	RobertMeta, any url on it's setup etc?	20:58
unregistered	napster_123: just place a repo in a windows share he means	20:58
RobertMeta	napster_123, umm... you have a directory, you share it... got create the repo there	20:58
	done.	20:58
	give everyone who needs write access to it, write access to the directory	20:58
napster_123	RobertMeta, ok got your point	20:58
	← yevlempy left	20:59
	← zenwryly left	20:59
	← flaguy48 left	20:59
RobertMeta	make that a bare repo (just for sharing) and you can have seperate local repos too	20:59
	that don't require network access	20:59
napster_123	RobertMeta, but i'll have to install samba client on every other pc in my network right?	21:00
RobertMeta	... smb is windows native sharing	21:00
	nothing to install.	21:00
	→ Mage joined	21:00
RobertMeta	napster_123, there are 1000 ways to do it	21:01
	← unijunction left	21:01
RobertMeta	create read only access on both sides, and you guys can pull	21:01
napster_123	RobertMeta, ok so simple windows sharing wizard must do it as well right?	21:01
	→ quag joined	21:01
RobertMeta	but not have to worry about pushes	21:01
	which might be easier	21:01
	you just pull off other person when you want thier changes	21:01
	→ dreiss joined	21:02
napster_123	RobertMeta, ok ... will try this	21:02
	← mutex__ left	21:02
	→ mutex__ joined	21:04
	← aresnick left	21:04
	→ fool__ joined	21:04
	← fool__ left	21:04
	→ fool__ joined	21:04
	← fool_ left	21:04
	→ zenwryly joined	21:05
quag	Hi git wizards, something that has escaped me, is how to switch branches without modifying the working tree. Something like a git checkout, but it doesn't swap all the files over. Ideas?	21:06
	Alternatively, a way to do a commit and manually tell it the parent would also work.	21:07
Pieter	git reset --soft?	21:07
quag	oh.... I see. git reset is the tool for moving the HEAD around?	21:08
Pieter	no, git reset does a lot of different things	21:08
	it depends on the invocation on what it will touch	21:09
quag	heh	21:09
	← smuf left	21:09
	→ magn3ts joined	21:10
wereHamster	quag: if you really feel up to it: man git-commit-tree	21:11
Gitbot	quag: the git-commit-tree manpage can be found at http://git.or.cz/man/git-commit-tree	21:11
	→ jceb joined	21:11
quag	yeah, might be time to learn commit-tree	21:11
	:)	21:11
	→ smuf joined	21:11
	→ rindolf joined	21:11
	→ johnw joined	21:12
	→ nabeel_ joined	21:13
	→ parasti joined	21:13
rindolf	Hi all. How do I ignore files that end with ~ in .git? I tried setting core.excludesfiles in .gitconfig to a location that contains ~, but it still doesn't ignore a ~ file.	21:13
wereHamster	echo "*~" > .gitignore	21:13
rindolf	wereHamster: I want it global.	21:15
	wereHamster: for all git repositories present and future.	21:15
	→ nabeel__ joined	21:15
rindolf	Seems like git doesn't like my ~/ path.	21:16
	← thiago_home left	21:16
rindolf	Is there anyway to make it respect a $ENV home variable?	21:16
	← napster_123 left	21:16
	← kipras left	21:16
	→ unijunction joined	21:16
rindolf	I mean a HOME environment variable.	21:16
	foo resumes his git learning	21:16
foo	Ok, let's say I do mkdir site-repo; git clone mysite site-repo - this should clone mysite to my git repository, right?	21:17
	← ph^ left	21:17
	← ph^_ left	21:17
	dutchie → EmperorDutchie	21:17
	→ WAAHS93 joined	21:18
	→ QAAEE4E joined	21:18
	← nabeel_ left	21:18
	← froschi left	21:18
unregistered	why does vim color the second line in my commit messages red?	21:18
mspang	leave the second line blank	21:18
WaltzingAlong	unregistered: git commit messages are generally 1st line-summary, 2nd line-blank, 3rd and so on-detailed descr of what/why	21:19
	EmperorDutchie → dutchie	21:19
	→ thiago_home joined	21:19
unregistered	ah. I'll start following that convention. thanks	21:19
	← nabeel__ left	21:20
unregistered	i guess I would have known that if I read the commit man page	21:21
rudi_s_	unregistered: And if possible first line below 50 characters, the rest below 72 characters per line.	21:21
unregistered	just read that paragraph in the discussio section of man commit :)	21:21
	← jceb left	21:22
rindolf	Seems like this feature was added to a recent version of git. :-S	21:22
	← Zephyrus left	21:22
rindolf	I mean the ~ in core.excludesfile.	21:23
	maattd → maattd\|away	21:23
foo	I tried to clone my live website into a repo. I did: git clone live repositories/live... it returns: fatal: failed to open /home/www/live/objects - any idea why?	21:23
mspang	you can't clone a directory that is not a git repository	21:24
	← brizly1 left	21:24
foo	mspang: hm, so if I'm cloning something, I have to clone a git repository to another git repository, correct?	21:25
mspang	yes	21:25
WaltzingAlong	foo: you git clone an existing git repository (not just a random directory) into a new location (the cloned copy)	21:25
	→ nadavoid joined	21:26
	← locklace left	21:26
foo	Aha, I see. When I work on my code, do I normally work on it in the repository?	21:26
WaltzingAlong	foo: yes	21:27
	foo: have you worked with versioning control systems before?	21:27
foo	aha, ok. I'm still learning git and am trying to understanding conceptually how it works. I think I have an ok grasp on it, just want to confirm a few things. WaltzingAlong, very very basic stuff with svn (checkout, commits, updates, really)	21:27
	→ napster_123 joined	21:27
rudi_s_	foo: I could recommend progit as a good guide for git.	21:28
WaltzingAlong	foo: ok well in the case of svn, you were usually working inside your own 'working copy'. since git contains the entire history, your what-would-be svn working copy in git is a full-blown repository	21:28
foo	rudi_s_: aha, thank you.	21:29
	WaltzingAlong: aha, that makes sense	21:29
rudi_s_	np	21:29
	← unijunction left	21:30
	← Adaptee left	21:30
	← aryounce left	21:32
	→ kipras joined	21:32
foo	Ok, so, I'm going to attempt to explain what I'm looking to do. I have a playground directory (drupal) where I work on the code, a staging area where I migrate in live data/test, and a live site (all drupal). I'd like to be able to make my playground directory a repository (so I can edit the code + test it in the playground drupal instance), then I guess I'd need to make a repository for staging and live - correct?	21:32
bob2	yes	21:33
	← yhager left	21:33
foo	aha! bob2 is around	21:33
	← alexmchale left	21:33
foo	bob2: the clone command, I thought I would use that to clone my playground environment to a new repository, but I guess clone can only clone a repository to another repository. I think I missed something last night	21:33
WaltzingAlong	foo: could also be branches within the same repository. then where you have the different storage locations, different repositories checked out/following the different remote branches	21:34
	foo: right. copying a directory is different than cloning a repository	21:34
	→ froschi joined	21:35
foo	WaltzingAlong: can you have sub-branches?	21:36
	← jas4711 left	21:36
	→ _numbers joined	21:37
unregistered	foo: FYI http://pastebin.com/uSaZDBj0 :P	21:37
	that didn't even catch all of them	21:37
foo	unregistered: haha	21:38
	← priidu_ left	21:38
foo	to my understanding, I would want to create a branch on the fly if I begin working on something, but don't finish it and want to return to it later	21:38
WaltzingAlong	foo: nothing wrong with that. create a branch. work on that branch. want to forget about it? work on a different branch	21:39
	← pantsman left	21:39
foo	WaltzingAlong: aha, I see. Hm, I'm thinking repositories might be better for the sake of organization. that way I won't get confused on which branches in the same repository belong to live/staging/etc	21:42
	WaltzingAlong: thank you. /me making progress	21:42
WaltzingAlong	foo: i was thinking of doing the same - just having the other repositories do nothing other than track the various different branches.	21:43
	but whatever works for you! foo	21:43
foo	thanks WaltzingAlong :)	21:43
	→ roll joined	21:44
foo	WaltzingAlong: ok, and repositories are meant to be empty, right? eg. when I create them.	21:44
	→ unijunction joined	21:45
	→ locklace joined	21:45
WaltzingAlong	foo: mkdir myplayground; cd myplayground; git init; yeah that will still have no files in it but will have a .git directory	21:45
	maattd\|away → maattd	21:46
	← Tureba left	21:46
foo	WaltzingAlong: gotcha. now, I'd like my repository to be my apache vhost so I can edit code + test it in drupal - is that ok?	21:46
roll	hi all, i converted my svn repo to git and want to commit to an svn branch "backtracking" with: $ svn dcommit backtracking. but i get an error (dcommitted on a detached HEAD because you gave a revision argument.). wht would be the correct way to doi it? thanks in advance for your help	21:47
	→ tildeequals joined	21:47
WaltzingAlong	foo: ok with me :D	21:47
foo	WaltzingAlong: hm, in that case, why can't I do git init in my drupal playground folder (vhost) which I already have?	21:48
	← maattd left	21:49
roll	sorry, a typo: the command is: git svn dcommit backtracking. any ideas how to commit to svn branch?	21:49
WaltzingAlong	foo: check out man git init or git init --help	21:50
	→ c_axis joined	21:51
foo	WaltzingAlong: I see " Create an empty git repository" - I guess my question is, if I want my code to be my playground webroot, then how do I do that? maybe, create an empty repo, cp -Rv my code in, then change vhost to match the repo dir?	21:51
	← richcollins left	21:51
	→ pantsman joined	21:51
WaltzingAlong	foo: could do that. create the git repo (new folder/empty), then copy/move your files in there. then you need to tell git to track those files (see git add)	21:52
	→ jas4711 joined	21:52
	← _Vi left	21:52
foo	WaltzingAlong: gotcha, ok. any disadvantage for doing that with all 3 of my directories (playground/staging/live) ?	21:52
	→ alester joined	21:53
	← mcdebugger left	21:53
unregistered	roll: i don't remember how that worked.. something like make a separate branch and use git reset to link that branch to the remote svn branch	21:54
	→ MacGyverNL_ joined	21:55
	→ Matthew joined	21:55
	Matthew → MWB	21:55
unregistered	roll: http://www.robbyonrails.com/articles/2008/04/10/git-svn-is-a-gateway-drug	21:55
roll	thanks unregistered, will give it a try :)	21:55
	→ nadavoid_ joined	21:55
MWB	is there a way I can specify a user with the git commit command? so if a I had multiple authors they could specify their name for the commit?	21:56
	← MacGyverNL left	21:57
	← __iron left	21:58
	→ MacGyverNL joined	21:58
unregistered	MWB: glancing at the man page --author stands out	21:58
	← c_axis left	21:58
	← nadavoid_ left	21:58
	→ c_axis joined	21:58
MWB	I tried that unregistered and I get fatal: malformed --author parameter,	21:59
	← MacGyverNL_ left	21:59
unregistered	MWB: well how did you form it?	21:59
drizzd	MWB: you need "name <mail@address>" format	21:59
unregistered	the quotes are probably important too	22:00
MWB	okay.	22:00
	← c_axis left	22:00
	→ mcdebugger joined	22:00
	← giallu left	22:01
	→ MacGyverNL_ joined	22:02
	← tildeequals left	22:02
	→ flaguy48 joined	22:03
	← unijunction left	22:03
	← eFfeM1 left	22:04
Textmode	night all	22:04
	← Textmode left	22:04
	→ alexmchale joined	22:05
	→ jasonn1234 joined	22:05
	→ Tureba joined	22:05
	← johnw left	22:06
	← MacGyverNL left	22:06
	→ nabeel_ joined	22:07
	→ aryounce joined	22:09
	→ neunon joined	22:09
	← corni left	22:10
	← aziz left	22:10
	← napster_123 left	22:11
	← cedricv left	22:11
	← kumbayo left	22:12
	← rindolf left	22:12
	← nadavoid left	22:13
	← pantsman left	22:14
	← dgl left	22:15
	← alexmchale left	22:18
	← nabeel_ left	22:18
	→ unijunction joined	22:18
	→ aziz joined	22:20
	→ niquil joined	22:20
	→ _Vi joined	22:23
	← LeMoNiCe left	22:24
	← ilteris_ left	22:24
	← kar8nga left	22:25
	← uau left	22:26
	← unijunction left	22:27
	← thrashold left	22:28
	→ bentob0x joined	22:28
	← WaltzingAlong left	22:28
	← JonathanT left	22:29
roll	unregistered: it worked :) thanks again	22:31
	→ uau joined	22:31
	→ JonasF joined	22:31
	→ CO_gairah_tinggi joined	22:31
	← ixti left	22:31
	← dys` left	22:32
	← hgimenez left	22:33
	← mcdebugger left	22:34
	← roll left	22:35
JonasF	Hi, can .gitconfig read environment variables? I couldn't get core.excludesfile to work with the line "excludesfile = ~/.gitignore" or ".. = $HOME/.gitignore", though /home/j/.gitignore works. But I'd like to use the same gitconfig on other hosts too, where $HOME might be different.	22:36
	→ hgimenez joined	22:36
	frogonwheels_ → frogonwheels	22:39
	→ unijunction joined	22:42
	← Hackykid_ left	22:42
	→ johnw joined	22:42
	← webchick left	22:44
	→ mcdebugger joined	22:45
	← hyperair left	22:48
	← doobry left	22:48
	→ Surma joined	22:49
Surma	Hey guys. Is there a quick way to find out, which HEADs point to the same commit? (i.e. which remotes and locals are in sync?)	22:49
	← unijunction left	22:50
	→ mark[oz] joined	22:51
wereHamster	JonasF: have you tried to use a relative path (relative to ~/.gitconfig) ?	22:53
	Surma: google 'git wtf'	22:54
foo	I'm working with Drupal code. Is there any reason not to have my repository directory be a vhost accessible via http? (so I can write code + test it)?	22:55
	← Tureba left	22:55
	← Surma left	22:55
	← froschi left	22:58
JonasF	wereHamster: I just tried, ".gitignore" or "./.gitignore" doesn't work neither	22:58
	← _Vi left	22:59
	→ _Vi joined	22:59
wereHamster	JonasF: which OS?	23:00
JonasF	wereHamster: ubuntu 9.10	23:00
wereHamster	btw, ~/.gitignore should work (the code does expand ~ to $HOME)	23:00
JonasF	okay, strange	23:01
wereHamster	are the files you are trying to ignore already tracked?	23:01
JonasF	no, I have .swp in ~/.gitignore, if I write [core]\n excludefiles = ~/.gitignore in ~/.gitconfig, .swp files are listed as untracked by git status, if I write excludefiles = /home/j/.gitconfig, they are not	23:03
wereHamster	git --version ?	23:04
	← sshc left	23:04
JonasF	1.6.3.3	23:04
	→ sshc joined	23:04
	← mklappstuhl left	23:05
JonasF	could it be of avail that .gitignore and .gitconfig are symlinks?	23:06
	→ unijunction joined	23:06
wereHamster	shouldn't be a problem	23:06
	<395de250>	23:07
Gitbot	[git 395de250]: http://tinyurl.com/yhurhzl -- Expand ~ and ~user in core.excludesfile, commit.template	23:07
	← johnw left	23:07
wereHamster	you have to update git	23:07
	v1.6.5.5 or later	23:07
foo	if I do mkdir repo; cd repo; git init; cd ..; mv repo repo-dir; ... that's fine, right? since the data is in .git in the directory (eg. want to make sure I can change the name without breaking anything)	23:09
JonasF	okay, just realized that it worked on my arch linux box, with git version 1.7.0.1.. thank you!	23:09
	→ Hackykid joined	23:09
wereHamster	foo: sure, no problem there	23:09
foo	wereHamster: thanks :) do you happen to know about my other question? regarding having a repository be a webroot	23:09
	← aziz left	23:09
	← tg left	23:09
	← _numbers left	23:10
	→ tg joined	23:10
wereHamster	foo: if you don't mind potentially exposing the .git directory to users, you can do that	23:10
foo	wereHamster: I basically want to be able to edit my code + see repsonses in drupal. I can't think of another way to do this (using drupal)	23:11
	wereHamster: still learning the ropes with git	23:11
wereHamster	foo: google 'git website howto'	23:11
unregistered	damnit. i wish git extensions didn't prompt you about line endings and just left line endings alone	23:11
	someone just commited a whole bunch of files with new line endings. after doing a pull on windows	23:12
foo	wereHamster: perfect, thank you - that's the kind of direction I was looking for. I see a few good results, reading now	23:12
	→ mr_danie1 joined	23:13
	← mr_daniel left	23:13
	← anxiolytic left	23:15
	← bentob0x left	23:15
	← intripoon left	23:16
	← niki left	23:16
	← flaguy48 left	23:17
	ChanServ set mode: +o	23:17
	wereHamster changed the topic to: 1.7.0.2 \| Homepage: http://git-scm.com \| Everyone asleep or clueless? Try [email@hidden.address] \| Git User's Survey 2009! http://git.or.cz/gitwiki/GitSurvey2009 \| Channel log http://tinyurl.com/gitlog \| Mailing list archives: http://tinyurl.com/gitml \| Gits on git: http://tinyurl.com/gittalks \| Pastebin: http://gist.github.com/	23:17
	→ pantsman joined	23:17
	ChanServ set mode: -o	23:17
	→ pingveno joined	23:19
	← pingveno left	23:19
	← pantsman left	23:20
	← ology left	23:20
	→ simplechat joined	23:23
	← gebi left	23:23
	← MrFixit left	23:24
	← niquil left	23:24
	→ ice799 joined	23:25
	← ice799 left	23:25
	→ Sonderblade joined	23:27
	→ niki joined	23:28
	→ ice799 joined	23:29
	← Twisted_Mentat left	23:33
	→ pantsman joined	23:33
	→ SpookyET joined	23:34
	→ daaku joined	23:34
	→ Twisted_Mentat joined	23:35
	→ Chaser_ joined	23:37
	← hgimenez left	23:39
	→ halfdime joined	23:40
	← MWB left	23:41
foo	wereHamster: so, I read this: http://toroid.org/ams/git-website-howto - the thing is, I don't have remote and local, I'm doing everything locally on the server via ssh... which might be why I'm confusd	23:41
	← tvw left	23:42
halfdime	Quick question [I hope]: if I want to split out part of an existing git repository to a repository of it's own with the history for those files, how would I do that?	23:44
	I have read all about subprojects and so forth but cannot find anywhere this is mentioned.	23:45
	→ claudine joined	23:45
wereHamster	halfdime: are all those files you want to split up in a subdirectory of your existing project?	23:46
halfdime	yeah.	23:46
	→ hgimenez joined	23:46
halfdime	I tried using git archive but no .git directory came with the subdir.	23:46
wereHamster	halfdime: man git-filter-branch and use the subdirectory filter	23:46
Gitbot	halfdime: the git-filter-branch manpage can be found at http://git.or.cz/man/git-filter-branch	23:46
wereHamster	halfdime: git archive is for something completely different	23:46
halfdime	wereHamster: you are the best.	23:46
	nod	23:46
	→ webchick joined	23:47
halfdime	my version of git (1.6.4) doesn't show filter-branch when I do git help. sigh	23:47
	← andres_ left	23:47
	→ andres_ joined	23:48
wereHamster	halfdime: git --help --all	23:48
halfdime	oh, and again. you rock!	23:48
	→ tildeequals joined	23:48
	→ Hackykid_ joined	23:49
	← Hackykid left	23:50
	← jasonn1234 left	23:50
wereHamster	foo: sure you can have the git repository right there	23:50
	→ erikvold joined	23:51
foo	wereHamster: ok, I don't recall anything saying that it's bad to have my repository be my webroot. I assume that's ok?	23:51
	maybe I can hide .git somehow	23:51
wereHamster	well, it's entirely up to you if you want to hide the .git directory..	23:52
	git doesnt' mind if other processes have read access to it	23:52
foo	wereHamster: what's the worst someone can do if they view it via http? or know what to look for in it? history of my changes?	23:52
wereHamster	yes	23:53
	and anything you have stored in the history (including files with passwords which would otherwise be protected by drupal etc)	23:53
foo	wereHamster: I see. yeah, assuming I want to hide that, what would be the right way to do it?	23:54
wereHamster	.htaccess ..?	23:54
	← alester left	23:54
	← palisv left	23:54
foo	wereHamster: duh... didn't think about that. thanks.	23:55
	← darwin left	23:55
	→ EricInBNE joined	23:58
	← Paraselene_ left	23:59

Logs Search ←Prev date Next date→

Channels Documentation