IRCloggy #git 2009-04-25

Provider of IRC logs since 2005.

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

	2009-04-25
	→ gberdyshev joined	00:00
	← juanjoc left	00:01
	← sgrimm left	00:02
	← wshimmy left	00:03
	→ sgrimm joined	00:03
	← jceb left	00:05
	← ajonat left	00:07
	← flujan left	00:10
	→ wshimmy joined	00:12
	← kalle_ left	00:12
	→ kalle_ joined	00:13
	← soycamo left	00:14
	← JohnnyL left	00:15
	→ Fusco joined	00:17
Fusco	hi all	00:17
	i have a little problem and i don't found an good issue on the web	00:17
	i use gitosis and i would like use git daemon	00:18
	when i run a command as	00:19
	sudo -u gitosis git daemon --base-path=/home/gitosis/repositories/ --export-all	00:19
	i had this error	00:19
	fatal: unable to allocate any listen sockets on host (null) port 9418	00:19
	have you an idea ?	00:19
	← Leemp left	00:20
	← Ademan left	00:20
	→ wsc joined	00:21
thiago_home	check if it's not already running	00:23
hachi	doener: solved it.... need to remove the 'fetch' line from the 'bare' remote that I add	00:24
	← bobesponja left	00:24
hachi	otherwise I create a loop where it keeps creating remotes/bare/bare/bare/bare/bare...	00:25
	→ bobesponja joined	00:25
Fusco	no it's not always running :(	00:25
	the 9418 port is open on the firewall	00:26
	ans netstat don't check git but it's in /etc/services	00:26
quentusrex	If I keep doing this command: git pull origin master:refs/heads/master	00:26
	to pull from a remote server, and push back to it easily	00:26
	how do I set it up to only need to type: git pull	00:27
	I know I need to set something up in the config file,	00:28
	→ arohner joined	00:28
quentusrex	but I'm not sure which would allow the simple: git pull	00:28
	and have the default...	00:28
	→ faithfulgeek joined	00:28
	← jfkw left	00:29
	← cpg left	00:30
	→ Leemp joined	00:30
	← KwikkSilva left	00:33
	← MisterN left	00:33
thiago_home	quentusrex: git branch newmaster origin/master	00:34
	git checkout newmaster	00:34
	git branch -d master; git branch -m newmaster master	00:34
	OR: git config branch.master.remote origin; git config branch.master.merge refs/heads/master	00:35
quentusrex	I need to run all of those?	00:35
	ok	00:35
	that last section was awesome...	00:35
	hmm	00:35
	→ bremner_ joined	00:35
	→ Landon joined	00:37
quentusrex	and how do I add an post-update hook that will do a git pull on a directory?	00:37
	I have this:	00:37
	← catalis left	00:37
	← Landon left	00:37
quentusrex	unset GIT_DIR	00:38
	cd /etc/cgit/cgit-admin/	00:38
	git pull --quiet > /dev/null	00:38
	← Casan left	00:39
quentusrex	nvm, fixed it	00:39
	thansk.	00:39
	← stuffcorpse left	00:39
	→ bremner__ joined	00:39
	→ stuffcorpse joined	00:39
	← bremner__ left	00:39
	→ bremner__ joined	00:41
	← stephenjudkins left	00:43
	→ wsc_ joined	00:47
	← wsc left	00:47
	← e_6 left	00:47
	← boto_ left	00:48
	← andrewkasper left	00:49
	← porcnick left	00:50
hachi	how do you ask git if a branch exists?	00:54
	→ nis joined	00:55
	← bremner_ left	00:55
Fusco	git branch [-a\|-r]	00:56
hachi	git rev-parse --verify -q master	00:56
	I guess	00:56
	not git branch	00:56
	that lists you the branches	00:56
	→ bremner_ joined	00:57
	→ gitte joined	00:57
	← faithfulgeek left	00:58
	→ thiago_h1me joined	00:58
	← lois[pac] left	00:58
	← botanicus left	00:58
	thiago_h1me → thiago	00:59
	← thiago_home left	00:59
	→ Landon joined	00:59
Landon	is this the right place to ask about gitosis problems?	00:59
nis	yes	01:00
	← rettub_ left	01:00
	→ rettub_ joined	01:00
Landon	well, I'm having some problems with it on my ubuntu server, it's not updating the conf files	01:00
	← brink left	01:01
nis	what do you do?	01:01
Landon	I've taken a peek at the post-update hook and see that it runs git-update-server-info or something similar, which is a command I can't find at all	01:01
	http://scie.nti.st/2007/11/14/hosting-git-repositories-the-easy-and-secure-way I've been using these instructions step by step	01:01
	← qhoxie left	01:02
nis	from where it doesn't work?	01:02
	→ Evanlec joined	01:02
Evanlec	so my question today is, are there _any_ good Windows clients for Git ?	01:02
Landon	nis: after I push gitosis-admin	01:03
Evanlec	ive been looking at hg and bzr because of this, but i'd really prefer to stick with git	01:03
Landon	I check the .gitosis.conf in the /home/git directory, but it's still how it started	01:03
	nothing's changed	01:03
	← bremner_ left	01:03
	← wsc_ left	01:04
	→ netoman joined	01:04
nis	Landon: what error do you get?	01:04
Landon	no errors	01:05
	it looks like it pushes fine and all	01:05
	it just doesn't update the actual config	01:05
nis	are you sure of your config file?	01:06
Landon	what do you mean?	01:06
	I check /home/git/.gitosis.conf after I push	01:06
nis	perhaps the config file has errors in it.	01:06
	← k0001 left	01:06
Landon	would that matter? it should still push right	01:06
	also, I'm assuming it would push keys from the keydir onto .ssh/authorized_keys, but the only one in there is the one I initialized gitosis with	01:07
	→ solrize joined	01:07
Landon	it's a pretty simple config, I don't see any errors in it	01:07
nis	ok,	01:07
Landon	I'm not sure if that missing command is the problem, but it looks like it could be _a_ problem :\	01:08
nis	please, check if git log in the server is up-to-date.	01:08
Landon	ok, where can I find that?	01:08
	ok, latest commit was "Automatic creation of gitosis repository."	01:09
nis	/var/spool/gitosis/repositories or somthing like that	01:09
	(race condition) :-)	01:10
Landon	I'm looking in /home/git/repositories	01:10
	nothing for gitosis in /var/spool	01:10
nis	ah.	01:10
	did you commit your work?	01:10
	← aziz left	01:10
Landon	yes , I'm pretty sure of that :P	01:11
	→ gberdyshev_ joined	01:11
Landon	let me try again	01:11
nis	git status	01:11
	Landon: is your gitosis server in your local system?	01:12
	← kalle_ left	01:12
	→ kalle_ joined	01:12
	← solrize left	01:12
Landon	oh, I'm getting an error now	01:12
	yeah	01:12
	well, I'm running oall of these commands on my server	01:13
	for now	01:13
nis	ok	01:13
	pastebin the error, plz	01:13
	← Dashkal left	01:13
	→ aziz joined	01:14
Landon	http://landon.pastebin.com/d6ddb27ce	01:14
	← mw left	01:15
nis	did you check the permissions?	01:15
	→ mw joined	01:15
	→ sunblush joined	01:15
Landon	they're at least read for everyone	01:16
	I don't see a directory named gitosis-export though	01:17
	oh , its in the repositories	01:17
	same permissions, -rw-r--r--	01:17
	← gberdyshev_ left	01:17
	→ gberdyshev_ joined	01:18
nis	who owns eht directory?	01:18
	the*	01:18
	and the file?	01:19
Landon	root	01:19
	← bremner__ left	01:20
Landon	not sure how that happened	01:20
	→ wsc joined	01:20
nis	do you have a "git" unix user?	01:20
Landon	yeah	01:20
	→ jerbear joined	01:21
nis	this should be owned by git, i guess.	01:22
	→ dennis__ joined	01:22
dennis__	hi	01:22
Landon	yeah, I just chowned it over	01:22
	tried to push but it told me it was uptodate	01:22
dennis__	i need help.... please	01:22
nis	dennis__: just ask :-)	01:23
dennis__	i have a changes inside a git on the server.... how apply these changes?	01:23
	i type git status and say my changes	01:23
nis	Landon: try 'git push origin master:master'	01:23
dennis__	i type git show and i can see my modification	01:23
	but how now apply this on the server?	01:24
Landon	again, everything up-to-date	01:24
nis	dennis__: man git-push	01:24
Gitbot	dennis__: the git-push manpage can be found at http://git.or.cz/man/git-push	01:24
nis	Landon: is it up-to-date?	01:24
dennis__	ok i push my git and it is now on the repository	01:24
	← romao left	01:25
Landon	nope, gitosis.conf is still old	01:25
dennis__	but no apply changes on the modified files	01:25
	← gberdyshev left	01:25
nis	dennis__: is the remore a bare repo?	01:25
	→ bremner_ joined	01:25
	← dodo_the_last1 left	01:25
dennis__	no i dont share it..	01:25
nis	remote*	01:25
hachi	GUH, crap	01:26
	← jzellman left	01:26
dennis__	yes on remote server by ssh	01:26
hachi	there's a trashed subversion revision	01:26
nis	dennis__: you should have a bare repo	01:26
dennis__	but git push is ok	01:26
nis	dennis__: faq non-bare	01:26
Gitbot	dennis__: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare	01:26
hachi	I know exactly what's wrong with it... is there any way I can force git-svn to skip it or anything?	01:26
	→ dodo_the_last joined	01:27
nis	Landon: did you clone the repo?	01:27
Landon	aha	01:28
	I just cloned it and it's fine	01:28
	the one I pushed	01:28
nis	(so, it's up-to-date)	01:28
Landon	yeah	01:28
	→ jzellman joined	01:28
	← Fusco left	01:29
	← flaguy48 left	01:29
nis	Landon: i guess you messed up your initial clone	01:29
	(it's not an offense)	01:29
Landon	not following you	01:30
	← gitte left	01:30
Landon	I mean, shouldn't the stuff where gitosis lives be updating along with the gitosis-admin repository?	01:30
	→ gitte joined	01:30
nis	not following you	01:30
	;-p	01:30
	basically,	01:30
	← ceej left	01:31
nis	you do the changes in your clone and push to update the gitosis server	01:31
Landon	right	01:31
	← jzellman left	01:31
nis	i'd try a new update from the last clone and push	01:32
	← mw left	01:32
nis	if it works, just remove the initial clone	01:33
	→ altrux joined	01:33
Landon	well	01:33
nis	and mv the good last repo in place of the initial	01:33
Landon	wasn't the initial clone updating correctly?	01:33
	if I cloned it again and it was correct	01:33
	← gaveen left	01:34
nis	maybe. what i've understood from you before is that you initial repo was somewhat broken.	01:34
	← Alinermedia left	01:35
nis	but i may be wrong	01:35
	→ gaveen joined	01:35
Landon	well I think the repo is correct	01:35
	but the post-update hook isn't working	01:35
	and the actual gitosis configuration isn't updating	01:35
	this is the third time I've wiped gitosis and reinstalled, but the same thing keeps happening	01:36
nis	what thing?	01:36
	i mean,	01:36
	if the new clone is up-to-date, it should be ok.	01:37
Landon	well, I make a repo on my laptop	01:38
	which I'vve added to gitosis	01:38
	← dennis__ left	01:38
Landon	but when I try to push the initial commit I get asked for the password for [email@hidden.address]	01:38
	which, it has none, gitosis should have added the key I'm usingfor my laptop	01:38
	← sitaram left	01:39
nis	could you pastebin your config file?	01:39
	← arohner left	01:40
Landon	nis: http://landon.pastebin.com/d4578709b	01:40
	in keydir I have a file [email@hidden.address]	01:40
	→ Alan joined	01:41
nis	you didn't give a name to repo in the "landon" group	01:42
	→ mndoci joined	01:42
Alan	Ooops, i just ended up doing 'git push --mirror', and now i have 'origin/master' in my remote repository...	01:42
	how do i get rid of it?	01:42
nis	something like "writable = landon"	01:42
	← WALoeIII left	01:42
	→ catalis joined	01:43
Landon	nis: I read that it could be used as a meta-user	01:43
	and I could list it in other writables	01:43
	er	01:43
nis	Alan: git remote rm ?	01:43
Landon	in other groups*	01:43
	← bremner_ left	01:43
Landon	so I didn't have to keep listing landon@c29150 and landon@delirium	01:43
Alan	oooh, i think i worked it out	01:43
nis	Landon: http://landon.pastebin.com/m59966057	01:44
Alan	nis: no, from the remote end :\|	01:44
nis	see line 13	01:44
Alan	it's ok now though	01:44
Landon	aha, let me try that	01:44
Alan	git push origin +:origin/master	01:44
Landon	I should do that for the gitosis-admin group too?	01:44
Alan	deleted origin/master on the remote end	01:44
nis	Landon: yes	01:45
Landon	getting somewhere now, I can push from my laptop	01:46
	but I get errors there too	01:46
	offby1` → offby1	01:46
nis	Alan: what do you want exactly?	01:46
Landon	http://landon.pastebin.com/d271c1323	01:47
Alan	nis: basically what happened is this...	01:47
	I have a remote repository, A	01:47
	cloned to a local repository, B	01:47
	in B, i did git push --mirror	01:47
	which copied B's reference to A's master to origin/master on A	01:48
	← aziz left	01:48
	→ sitaram joined	01:48
Alan	I wanted to get rid of that reference	01:48
	And by pushing "no branch" to it, i seem to have managed that	01:48
nis	Landon: git branch	01:48
Landon	what should that be doing?	01:49
nis	list branches	01:49
Landon	nothing	01:49
nis	ok,	01:49
	before push, you must do a commit first.	01:50
	then, you souhld be on branch master	01:50
	→ kodama joined	01:50
nis	then you can push.	01:50
	I usually start with	01:50
	Mr_Grieves\| → Mr_Grieves\|zzz	01:50
nis	git commit --allow-empty -m 'Init'	01:51
Landon	do I need to do the whole push origin stuff?	01:51
kodama	hello, I'm having trouble getting some changes into my git repo. specifically, migrating some perforce code into git.	01:51
	I'm syncing p4 code to /usr/local/src	01:51
	cloning my git repo to /tmp/mypath	01:51
	then running 'cp -rf /usr/local/src /tmp/mypath'	01:51
Landon	like, in the future would I be able to just use "git push" instead of "git push origin master:stuff/morestuff/etc"	01:51
kodama	but git thinks no changes were added?	01:51
nis	Alan: yes, git push origin :branch-to-delete	01:51
kodama	I do a 'git add .' followed by commit	01:52
	but	01:52
	→ antoranz joined	01:52
kodama	no dice	01:52
nis	Landon: yes	01:52
Landon	nis: great, I think I've got everything solved	01:52
	you're the man	01:52
kodama	can anyone suggest what I might be missing?	01:52
nis	Landon: np!	01:52
	← Landon left	01:52
Alan	nis: thanks for verifying that i did it the right way :)	01:52
	I'm pretty happy that i've got to grips with the concepts of Git when I can work those things out	01:53
sitaram	kodama: did the git commit throw up an editor for you to type in a message or error out?	01:54
kodama	sitaram: it just blows up saying "nothing to commit"	01:54
sitaram	and the git add .?	01:54
kodama	sitaram: the git add . runs without saying anything	01:55
sitaram	assuming you're in the right directory, I suggest touching a few files that you are sure have changed and try git add just-those-files first	01:55
	touch file_I_am_sure_has_changed	01:56
	git add file_I_am_sure_has_changed	01:56
	and see if it adds	01:56
kodama	sitaram: good call, let me try...	01:56
nis	sitaram: the touch command?	01:57
kodama	sitaram: touched, edited, added, committed. all worked fine.	01:57
	sitaram: that's quite strange.	01:57
	← gaveen left	01:57
sitaram	kodama: I suspect that the cp is keeping the timestamps intact, and git is optimising by not checking the content if the time stamp is the same	01:58
	maybe the size is also considered, I dont know	01:58
	kodama: try 'find . -type f -print0 \| xargs -0 touch; git add .'	01:58
	don't worry it wont add files that are really unchanged -- but it may be a bit slow as it SHAs each file	01:59
kodama	sitaram: where should I be doing the find from?	01:59
sitaram	/tmp/whatever -- the repo root	02:00
kodama	sitaram: ah	02:00
	→ fujin joined	02:00
kodama	let me try	02:00
sitaram	/tmp/mypath	02:00
	→ gaveen joined	02:00
	→ mw joined	02:00
	→ merlin83 joined	02:00
kodama	'nothing to commit'	02:01
sitaram	nis: what I find strange is that cp is not supposed to preserve timestamps unless you use "-a" or "-p", which he has not used... weird	02:01
kodama	yeah it's like the copy never ran	02:01
nis	sitaram: agreed	02:01
sitaram	wait... but if you do this to a few individual files it worked?	02:01
	← sekimura_ left	02:01
kodama	if I cd to the repo root and touch a file it works just fine	02:01
	I didn't try going to where I had synced new code and add a new file there	02:02
	let me give it a whirl	02:02
sitaram	"where I had synced new code" == ?? (confused)	02:02
kodama	sitaram: oh I see -- maybe I didn't make that clear	02:02
	sorry	02:02
	so	02:02
	what I'm doing is	02:03
	sync P4 to some directory on the box	02:03
nis	kodama: sitaram: don't you have /tmp mounted as shm type with some extra options?	02:03
kodama	git clone my github repo to some dir under /tmp	02:03
	nis: I have no idea	02:03
sitaram	oh you mean /usr/local/src in your case?	02:03
kodama	sitaram: exactly	02:03
	then cp -rf /usr/local/src /tmp/whereiclonedto	02:04
sitaram	nis: you may have hit on it	02:04
	kodama: are you sure the copy completed without errors?	02:04
	← gitte left	02:04
kodama	sitaram: when I run the command nothing explodes	02:04
	sitaram: I'm assuming it'd tell me if it had a problem?	02:04
sitaram	yes	02:05
	it would	02:05
	so let's try and break this down	02:05
nis	kodama: could you try under /home?	02:05
kodama	sure thing	02:05
sitaram	after the cp from usr_local_src to tmp_mypath	02:06
kodama	nis: clone repo to there?	02:06
nis	yep	02:06
Alan	Also, there is something I really love about Git...	02:06
sitaram	do a 'diff -qr usr_local_src tmp_mypath' and see if it reports anything except ".git" and any similar directories for p4	02:06
kodama	nis: let me try	02:06
	← markelikalderon left	02:06
kodama	sitaram: will try that too	02:06
	:P	02:06
	one sec you guys	02:06
sitaram	kodama: do what nis says first	02:06
Alan	Which is that you don't have the "one precious repo we must tiptoe around"	02:06
	I mean, if you really break something in SVN, you break it for everybody	02:07
sitaram	kodama: and if no success then you can do this diff business	02:07
Alan	I mean the repository itself, not just code in it	02:07
kodama	sitaram: okay, one sec then	02:07
	type	02:07
Alan	Even to the point of "oops, accidentally deleted the remote branch"	02:07
	Just get someone else to push it back in	02:08
	or something like that	02:08
	← mw left	02:08
sitaram	Alan: and I always thought that is true of all distributed version control systems, but it appears that the others don't actually clone every branch	02:08
	→ btanaka joined	02:08
nis	Alan: yep, assuming you have people who pulled. :-)	02:08
	(or fetch to be fair)	02:09
	→ mw joined	02:09
Alan	Also, kudos to GNOME for using Git... but i can't find the damn project I'm after in their massive project list!	02:10
	http://git.gnome.org/cgit/	02:10
kodama	nis: sadly...cloning to my home directory did not make a difference. :(	02:10
	-bash-3.1$ git status	02:11
	# On branch master	02:11
	nothing to commit (working directory clean)	02:11
	let me try the diff now	02:11
	← mndoci left	02:12
	← kalle_ left	02:12
	→ kalle_ joined	02:13
kodama	waitwaitwait	02:13
	omg....	02:13
	I think I'm about to be really embarassed.	02:13
	← km2 left	02:13
nis	kodama: we all do crap sometimes. :-)	02:13
kodama	I just noticed a directory where in my git cloned directory that looks suspiciously like it might be named "src".	02:14
	could I just be missing a '.'?	02:14
sitaram	you'd have seen that directory as "new" in that case	02:15
	(in git status I mean)	02:15
kodama	sitaram: oh, oka	02:15
	s/oka/okay/	02:15
sitaram	so try the diff	02:15
kodama	sitaram: wait, though -- that directory should not be present	02:16
sitaram	unless you did that earlier and committed it :)	02:16
kodama	after I do the copy, there shouldn't be a "src" directory in either place	02:16
	sitaram: you know	02:16
	I think I did.	02:16
	bc when I look in my repo on github	02:16
	← fujin left	02:17
kodama	I do see that dir	02:17
sitaram	yeehaw! someone just like me -- been there done that got the t-shirt didn't like the color	02:17
kodama	haha!	02:17
	so	02:17
	how do I back this out?	02:17
	and do it right in the future? ;;)	02:17
sitaram	...as long as you did not push or someone did not pull from you	02:17
	we can	02:17
	if someone == a very few people you can personally contact, that also may be ok	02:18
kodama	nobody's pulled from my repo, it's private	02:18
	it looks like I cloned, copied into the wrong place	02:18
sitaram	so, step 1 is to get your local repo back in shape, step 2 is to 'git push -f' that to github	02:18
kodama	committed	02:18
	and then pushed	02:18
	okay	02:18
	step 1 is easy	02:18
sitaram	step 2 is easy; step 1 is not. How thoroughly do you want to clean up? Erase all traces of the extra dir from past commits or just fix the top commit?	02:19
	("not" == relatively)	02:19
kodama	I just want to get rid of the extra dir	02:19
sitaram	I assume you mean "just the top commit being correct is good enough"	02:19
kodama	that's correct	02:20
sitaram	well that's easy. Even otherwise it only means a git filter-branch, but now you just have to git rm -r the-extra-dir	02:20
kodama	the github repo is really just a staging area, so as long as the head is right, things are a-ok	02:20
sitaram	then git status	02:20
kodama	fatal: pathspec 'src/' did not match any files	02:21
	huh?	02:21
	→ cytrinox_ joined	02:21
nis	ls -ld src ?	02:21
	← cytrinox left	02:21
sitaram	kodama: basic sanity check: pwd and make sure you're in the right place	02:21
kodama	wow...my head's about to explode. looks like the dir is bare empty?	02:22
	→ mndoci joined	02:22
kodama	yep, pwd confirms i'm definitely in the root of my app/repo	02:22
sitaram	well that would explain git not knowing, since empty directories are ignored	02:22
	so back to the diff then...	02:22
kodama	wait but whaaaa	02:22
	I'm so confused.	02:22
nis	sitaram: kodama: no, check 'ls -l'	02:22
	the -d option may confuse you there	02:23
kodama	'ls -al' returns just '.' and '..'	02:23
nis	Ah, ok, good to know. :-)	02:23
	rmdir src, then	02:23
sitaram	I assume you did that inside this ghost src directory?	02:23
kodama	sitaram: correct	02:24
	nis: okay, done	02:24
sitaram	now diff -qr root_of_GIT_repo /usr/local/src_where-I_copied_from_p4	02:24
	I can't think of anything else	02:24
nis	what p4 is? (yes i know, it's been a long time we are talking about that)	02:25
sitaram	and if that doesn't work, I'm going to suggest (1) fresh clone (2) diff (3) copy (4) diff again and if step 2 and step 4 give different results my head will explode	02:25
	perforce	02:25
nis	Ah, thans	02:25
sitaram	a proprietary VCS baed on CVS loosely	02:25
nis	thank you :-)	02:26
kodama	nis: sitaram's got it	02:26
sitaram	perl was the most high profile that used it for many years (they made a license exception for perl or something)	02:26
	→ markelikalderon joined	02:26
kodama	sitaram: yeah I think they'll give out licenses for open source	02:26
	but it's def. not a distributed system	02:26
	anywho	02:26
	sorry, I think I've got myself confused on the box here.	02:27
sitaram	none of the older crop are distributed	02:27
kodama	let me back track and get to a clean place.	02:27
sitaram	kodama: yeah often helps to step back and go very slowly	02:27
kodama	I'm re-cloning my repo from scratch and going in to do the rm	02:27
	→ vbabiy joined	02:28
kodama	okay, I did 'rmdir' on the empty directory	02:28
	I'd like to first just get this bad directory removed and pushed to github	02:29
	→ Landon joined	02:29
sitaram	wait wait	02:29
	you got an empty directory on a clone? not possible!	02:29
kodama	sitaram: ?	02:29
Landon	ok, I'm back, I'm having trouble with git-daemon n ow	02:29
	in its logs I get '2009-04-25_02:21:55.53530 [9636] '/home/git/repositories/pysg.git': unable to chdir or not a git archive	02:29
kodama	sitaram: well....ls doesn't lie, right?	02:29
nis	kodama: directories are not tracked	02:29
Landon	and I've looked at the /home/git/repositories/pysg.git/ and it looks like the .git directory after I check out a repo	02:30
kodama	sorry guys, I'm just telling you what 'ls' is telling me	02:30
Landon	is that right?	02:30
	← octagon left	02:30
kodama	let me reclone just to prove it	02:30
	so	02:30
	→ k0001 joined	02:30
nis	Landon: yes	02:30
kodama	i'm in my tmp directory	02:30
	cloning from github now	02:31
Landon	ok, so it	02:31
nis	Landon: that's a bare repository	02:31
Landon	s a valid git archive	02:31
	then I guess the problem is unable to chdir, which I haveno idea about	02:31
kodama	-bash-3.1$ ls -alF src/	02:31
	total 16	02:31
	drwxr-xr-x 2 cp cp 4096 Apr 24 19:30 ./	02:31
	drwxr-xr-x 17 cp cp 4096 Apr 24 19:30 ../	02:31
nis	Landon: check permissions	02:31
Landon	oh, I think I see something, git-daemon is running as "116"	02:31
	going to poke around and see if I can change that	02:32
sitaram	kodama: pwd?	02:32
Landon	should probably run as the git user, right?	02:32
nis	Landon: wait, you're running gitosis, right?	02:32
kodama	sitaram: pwd is the root of where I cloned to -- i.e. I did "git clone git:remote/path myclonedir; cd myclonedir; ls -alF src'	02:33
Landon	nis: right	02:33
	nis: I changed the user of git-daemon andn ow I get a seperate error :D	02:33
	oh, I think I have it now	02:34
nis	Landon: so, what do you do with git-daemon?	02:34
Landon	nis: I would like anon access as well	02:34
	the guide I've been using suggests using git-daemon for that	02:34
nis	Landon: for push or pull?	02:34
Landon	pull	02:34
nis	Ah, ok.	02:35
sitaram	kodama: I am at a loss. Is this a private repo or you just meant no one has cloned it yet? Meaning can I clone it?	02:35
kodama	sitaram: sorry, it's a private repo	02:35
nis	isn't export into gitosis for that purpose?	02:35
sitaram	kodama: np	02:35
Landon	no, that's for user control	02:35
	at least from what I've read	02:36
kodama	sitaram: and I can't share the code (we're hosting proprietary code in there)	02:36
	sitaram: I mean....yeah....this is totally weird.	02:36
sitaram	no no worries -- we do proprietary code too :) that's ok	02:36
kodama	sitaram: :)	02:36
	hmmm	02:36
	so	02:36
	here's one thing I'm thinking -- as a potential	02:36
	→ eno__ joined	02:37
kodama	"this is really friggin' weird but might affect why"	02:37
sitaram	kodama: check all ".gitignore" files, all configs (local and global) for core.excludesfile, and .git/info/exclude for anything suspicious	02:37
nis	Landon: i should say: isn't the "export" support into gitosis for that purpose?	02:37
	or s/support/option/	02:38
	→ fujin joined	02:38
kodama	sitaram: how do I check the configs? I"m pretty new to git	02:38
	← eno left	02:39
nis	Landon: AFAIR, you just have to add a .export-something in your repo.	02:39
kodama	.git/info/exclude looks like it's all just commented out stuff	02:39
	eno__ → eno	02:39
Landon	nis: I'm googling, but not seeing anything for gitosis	02:39
sitaram	kodama: ~/.gitconfig, reporoot/.git/config are the two places	02:39
	.gitignore can be in any directory in reporoot	02:39
	← gberdyshev_ left	02:40
	← catalis left	02:40
	→ cemerick joined	02:40
Landon	nis: the gentoo wiki uses git-daemon	02:41
	on the gitosis page	02:41
kodama	sitaram: the only .gitignores I see are in directories far below the app root -- I'm 99% sure they wouldn't affect anything	02:41
	http://www.pastie.org/457748	02:41
Gitbot	[git 457748]: http://tinyurl.com/cjr2e9 [tree]	02:41
	nis checking	02:41
kodama	is my .git/config	02:41
	no ~/.gitconfig either	02:41
sitaram	kodama: then that's a dead end; all looks fine	02:42
kodama	sitaram: let me paste in what happens in git status after I do the 'rmdir src'	02:42
sitaram	kodama: fine, we'll shelve the "why did clone bring back an empty directory" part for now, and try that diff in the following sequence	02:42
	(1) fresh clone (2) diff (3) copy (4) diff again and if step 2 and step 4 give different results my head will explode	02:43
kodama	sitara: http://www.pastie.org/457751	02:43
	haha okay	02:43
	let me try, one sec.	02:43
	ah	02:43
nis	Landon: did you do http://gist.github.com/101467 ?	02:43
sitaram	looks like src was a symlink then...	02:44
kodama	sitaram: but the thing is -- that empty directory does need to get out of there	02:44
sitaram	kodama: please try 'cd reporoot; ls -al \| grep src'	02:44
kodama	sitaram: after I 'rmdir' or before? I have a fresh clone ready	02:45
sitaram	before	02:45
kodama	I'll send you pastie with the output	02:45
sitaram	ok	02:45
kodama	http://www.pastie.org/457753	02:46
Landon	nis: looking at it, but it looks like just automatic management for git-daemon, I already have the export file in the directory	02:46
	but let me try this , see how it works out	02:46
kodama	sitaram: looks like you expect?	02:47
sitaram	kodama: this is really really weird. How about git ls-tree -l -r -t \| grep src (edit the output to remove or elide confidential names but keep the overall number and structure)	02:47
nis	kodama: no, for me.	02:47
kodama	git didn't like that command -- I'm on 1.6.2.4 on this box	02:48
nis	sitaram: i wonder if there was a src file before, then was rm then a directory with that name creanted.	02:48
sitaram	nis: he got it on a fresh clone; regardless of past history, that's weird!	02:49
nis	ah, beautiful!	02:49
kodama	hopefully this is helping to teach you guys something new and exciting about git. ;)	02:49
sitaram	kodama: what does git say to the ls-tree?	02:50
	oh yes it is!	02:50
kodama	it comes back with a "usage: git ls-tree" message	02:50
nis	seems like (1) the clone process did a mess or (2) the repo cloning from is crappy	02:50
kodama	the thing that's weird is that	02:50
	well let me back up	02:50
	← agile left	02:50
sitaram	kodama: sorry; append HEAD or master or whatever to the ls-tree options -- my mistake	02:50
kodama	I've concluded at least two things so far	02:51
sitaram	s/whatever/ whatever your current branch	02:51
	→ tjafk1 joined	02:51
kodama	so just 'git ls-tree -l -r -t master \| grep src' ?	02:51
	→ Flathead joined	02:51
sitaram	yes, if master is the current branch. Or just use HEAD if in doubt	02:52
	nis: I was going ask for a pastie of the "git clone" output...	02:52
kodama	fact #1) I somehow copied the 'src' directory into a freshly cloned repo, I ran add, git took it and let me commit it and push it	02:52
	I can send you the clone as well	02:52
	→ jboyens joined	02:52
kodama	160000 commit 7d5116aeee97265984e4ea87e661d0a41c673b1a - src	02:52
	(hopefully I didn't just tell you all my secrets there)	02:52
	→ agile joined	02:53
kodama	really, that one fact is all I've found out. : )	02:54
	nis: sitaram: let me do a fresh clone and send you the ouput	02:54
sitaram	kodama: that 160000 commit stuff? is that from ls-tree?	02:54
	you have a submodule!!!	02:54
kodama	sitaram: that's from ls-truee	02:54
	er	02:54
	s/truee/tree/	02:54
	← markelikalderon left	02:54
nis	sitaram: good job :-)	02:54
sitaram	nis: that explains it; I've never seen a "commit" come up in "ls-tree" -- and I vaguely remember that mode 160000 is for submodules	02:55
kodama	bwwaaaaaaah -- I swear! I've done no submoduling!	02:55
sitaram	kodama: cat .gitmodules or .gitsubmodules or some such please	02:55
	← wsc left	02:56
kodama	sitaram: I don't have any files like that.....	02:56
	find . -name "git"	02:57
	→ sekimura joined	02:57
kodama	returns nothing that even looks close to having the word "module" in it	02:57
	all I see are those .gitignore files I mentioned before	02:57
	oh	02:58
	wait...	02:58
	okay...	02:58
	sorry, thinking....	02:58
	hmmm	02:58
	sunblush → sunblush\|away	02:58
kodama	wait, no, nevermind. sorry.	02:59
	→ aspotashev\|eeepc joined	02:59
	nis g2g, sorry	02:59
kodama	nis: np -- thanks for looking	02:59
	← yinkei left	02:59
	← nis left	02:59
kodama	sitaram: nothing else springs to mind?	03:00
	sitaram: I agree this is a pretty weird issue	03:00
sitaram	kodama: 160000 is definitely a submodule thing, as is the finding of a "commit" object in an "ls-tree" output (the second one I am guessing)	03:01
	← Landon left	03:01
sitaram	I have not played with submodules at all, but I would suggest a "git submodule init; git submodule update" on a fresh clone and paste it. I'll be back in about 10 min :(	03:01
kodama	let me try that	03:01
	i'll be back in a sec too	03:02
	thanks for all the help so far	03:02
	sitaram: your head'll explode when you see this -- http://www.pastie.org/457765	03:03
	← sekimura left	03:03
	→ raydeo joined	03:05
madewokherd	looks pre	03:05
	← peff left	03:05
madewokherd	looks pretty safe to heads to me	03:05
	→ Tricon joined	03:06
	← jboyens left	03:06
	→ lazyshrk joined	03:06
kodama	madewokherd: haha -- but we're hot on this submodule track, is why	03:07
	← lazyshrk left	03:07
	← tjafk2 left	03:07
madewokherd	the submodule just isn't configured properly	03:07
	which makes sense :p	03:07
	→ lazyshrk joined	03:07
	→ merlin83_ joined	03:08
	→ dennis_ joined	03:08
dennis_	hi everybody	03:09
kodama	madewokherd: do you have the full context of the conv. though?	03:09
	→ davidad joined	03:09
dennis_	i'm an expert on ROR .. who need help?	03:09
madewokherd	no, just the last 15 minutes or so	03:09
hobodave	ugh	03:09
	I did something stupid and need help before i lose my changes	03:09
kodama	madewokherd: that directory shouldn't even be there -- I got the empty directory on a fresh clone from github	03:09
dennis_	what is your problem?	03:09
sitaram	madewokherd: last 15 min is fine, and if you know anything about submodules please take over :) I'm at the end of my expertise!	03:10
kodama	sitaram: wait wait	03:10
	← Leemp left	03:10
dennis_	but you have call me God always	03:10
kodama	sitaram: don't go just yet! :)	03:10
hobodave	http://pastie.org/private/ksfj9m1fqb9bmbggqjoa	03:10
	can someone help me recover my changes	03:10
	I screwed something up	03:10
sitaram	I'm here not going anywhere, but I am out of ideas :(	03:10
dennis_	what is your problem?	03:11
kodama	sitaram: the empty dir still hasn't been removed and pushed. I'm sure you can at least help me with that part, right?	03:11
	sitaram: or is that actually like completely impossible?	03:11
	sunblush\|away → sunblush	03:11
kodama	(which I assume is "no")	03:11
dennis_	ATTENTION: the best way is formatting your pc and install all again!!!!!!!!!	03:11
sitaram	kodama: that is not just an empty dir; it's a stub for a submodule. "git ls-tree" shows it as a "commit" object	03:11
	well just	03:11
	well just 'git rm src' and see what happens	03:11
	I suspect it'll barf	03:12
kodama	yeah, it says "pathspec didn't match any files"	03:12
	but	03:12
	remember	03:12
	well, actually, no....that wouldn't make any sense.	03:13
hobodave	guys, I committed a revision, then wanted to change my commit message, so I did a 'git reset --soft HEAD^'. I then tried to edit the message by typing 'git commit --amend' but this brought up the previous commit, after quitting out of that, I seem to have lost my most recenet commit	03:13
	please help	03:13
	http://pastie.org/private/ksfj9m1fqb9bmbggqjoa	03:13
kodama	sitaram: oh -- here's one thing	03:13
	sitaram: maybe if I can figure out when the change was introduced that'd be helpful?	03:13
madewokherd	what if you rm --cached?	03:13
	→ schacon joined	03:13
kodama	madewokherd: 'git rm --cached src'?	03:14
madewokherd	yeah	03:14
kodama	nope, still the pathspec error	03:14
	← kukks left	03:14
kodama	git doesn't have a "just nuke it" option? :P	03:14
madewokherd	for rm there's -r, but I don't see why that would help..	03:15
	hobodave: run git reflog	03:15
kodama	madewokherd: I can delete the dir just fine	03:15
madewokherd	hobodave: your commit and its id (which you can checkout) should be near the top	03:15
hobodave	madewokherd: http://pastie.org/private/t8ogeetuzdsq68ehjavjng	03:16
kodama	madewokherd: http://www.pastie.org/457773	03:16
hobodave	c3334fb is the commit I wanted to undo and edit the message of	03:16
	b2ffce3 seems to be me resetting it	03:16
	yet 1f5be9e was a mistake	03:17
sitaram	kodama: I managed to reproduce your situation	03:17
kodama	sitaram: oh? do tell!	03:17
hobodave	and git st doesn't show my local files as having changed	03:17
	do I need to do another git reset?	03:17
sitaram	make a submodule, commit, then delete ".gitmodules", commit, and clone that	03:17
	← scarabx left	03:18
madewokherd	I would reset to c3334fb and then do a commit --amend	03:18
kodama	sitaram: how would I do that?	03:18
madewokherd	but git status already says it's deleted	03:18
	surely all you have to do is commit	03:18
hobodave	madewokherd: git reset --soft c3334fb ?	03:18
kodama	sitaram: I am totally buying you a cake if this works	03:18
hobodave	or hard	03:18
madewokherd	shouldn't matter if all you're doing is a commit --amend, but I'd go with hard	03:19
sitaram	if you have never played with submodules you have been hacked :) Is the only explanation I have now...	03:19
kodama	sitaram: it could very well be that something odd happened....our codebase has a tendency to produce strange effects.	03:19
	← qrush left	03:19
sitaram	if you grep for 160000 in the git sources, it's all about submodules. And I have never seen a commit object show up in an ls-tree, (since I have never played with submodules either)	03:19
hobodave	perfect, t hanks madewokherd	03:19
madewokherd	ah, changed but not updated, you should be able to add that change	03:19
kodama	sitaram: just by peripheral association	03:19
	madewokherd: tharin lies the rub, indeed ;)	03:20
madewokherd	(or git commit -a, as the message says)	03:20
	← jerbear left	03:21
madewokherd	I think commit -a would fix it	03:21
kodama	whoa.	03:21
dennis_	hi madewokherd what is your problem?	03:21
kodama	wild and crazy giggling	03:21
madewokherd	dennis_: I am fine what is your problem sir?	03:22
sitaram	dennis_: it's a submodule that sneaked in somehow and then the .gitmodules file got deleted; can you help?	03:22
kodama	madewokherd: sitaram: git commit -a brought up the commit dialog and let me commit	03:22
	I pushed and now I don't see it anymore.	03:22
	recloning to double-check	03:23
	nope....it's totally gone.	03:23
	head explodes	03:23
sitaram	kodama: one last thing for me:	03:24
kodama	I want to scream, cry, laugh and dance all at the same time.	03:24
madewokherd	submodules are really not all that complicated	03:24
sitaram	kodama: "git log --stat --pretty=oneline -- .gitmodules	03:24
madewokherd	except for the part where they break all sorts of rules	03:24
sitaram	paste please...	03:24
kodama	sitaram: sure thing, one sec.	03:24
sitaram	kodama: just to confirm my sanity :)	03:24
	← dennis_ left	03:24
kodama	sitaram: I'd paste the output, but there is none.	03:25
sitaram	ok my head exploded then... never mind	03:25
kodama	sitaram: well	03:25
offby1	eww	03:25
	offby1 reaches for a damp rag	03:25
kodama	sitaram: if it makes a difference to you, git did put a lot of newlines in	03:25
	sitaram runs off to pick up the pieces and put them back together	03:25
	← vbabiy left	03:25
	← hobodave left	03:25
sitaram	newlines where?	03:25
	← antoranz left	03:25
kodama	i'll paste	03:25
	sitaram: http://www.pastie.org/457778	03:26
sitaram	kodama: take out the --pretty=oneline and try	03:26
	this is getting weirder!	03:26
	git log -- .gitmodules	03:27
	from the repo root of course, kodama	03:27
kodama	sitaram: sure thing, one sec -- I still have to figure out why the heck my copy didn't put things in the right place the first time, but hopefully that's just me borking the syntax of "cp"	03:27
sitaram	(the --stat doesn't matter, in or out)	03:27
kodama	fatal: Not a git repository (or any of the parent directories): .git	03:28
	er	03:28
	wait	03:28
sitaram	huh?	03:28
kodama	tee hee	03:28
sitaram	pwd?	03:28
	→ vbabiy joined	03:28
kodama	forgot to cd into the right directory	03:28
	same thing as before	03:28
jjuran	I have a question about resolving rebase conflicts. I'm familiar with the conflict markers from CVS, where I need to modify the file to look the way I want it to now. But what is git asking for during a rebase conflict?	03:28
kodama	just a lot of blank lines	03:28
	sitaram: it's the exact same output as before	03:28
sitaram	a plain "git log -- .gitmodules"? wow... ok, time for a stiff drink	03:29
kodama	sitaram: hehe	03:29
sitaram	kodama: try the same thing with some nonsense name instead of .gitmodules, something like skjfjksddhfsf	03:30
	jjuran: a rebase is also a merge in a way, so it's the same thing, you look for the same stuff	03:30
kodama	just a sec...I'm recloning so I can give you the output before I commit new code	03:31
	sitaram: I need to push some new code up for deployment :P	03:31
sitaram	a rebase does it commit by commit so it's often easier on conflict resolution (or so I believe)	03:31
	kodama: np	03:31
	kodama: work comes first don't worry	03:31
	← altrux left	03:31
kodama	sitaram: well....this is interesting.	03:32
	sitaram: http://www.pastie.org/457782	03:32
	clearly git is reacting to the .gitmodules	03:32
sitaram	absolutely empty output for a nonsense file? normal	03:32
jjuran	sitaram: I haven't yet wrapped my head around resolving something that I've since modified.	03:32
kodama	sitaram: what I mean is	03:33
sitaram	kodama: clearly it is...	03:33
kodama	sitaram: seems odd to me that it'd act differently	03:33
sitaram	kodama: can you do any git logs at all. Like "git log --stat"	03:34
	and get the normal output?	03:34
	← Tricon left	03:34
kodama	sitaram: totally normally output	03:35
sitaram	you get filenames with ++---- type stuff?	03:35
	if so, grep that for "gitmodules": git log --stat \| grep gitmodules	03:35
	→ faithfulgeek joined	03:36
kodama	sitaram: returns nothing....	03:36
sitaram	sigh. ok, at this point if you fixed the problem, it's probably time to move on; just keep an eye on things	03:37
	I am not sure how git log would return a bunch of blank lines anyway	03:37
kodama	yeah it's pretty weird	03:37
sitaram	unless you;re using some tool to convert from p4 to git and it did something weird	03:37
kodama	no, it's pretty basic and brute force	03:37
sitaram	would explain the submodules too; maybe some aspect of p4 translates to submodules	03:38
	ok	03:38
	nm then	03:38
kodama	sitaram: thanks for sticking with me through it all ;)	03:38
sitaram	kodama: you're welcome -- it's always educational :)	03:38
kodama	as a complete aside, I like your handle -- I like the Ramayana	03:39
sitaram	that's my first name -- in India it is common to name people after mythological characters :)	03:39
	I'll ask you later how you heard about/read the Ramayana :)	03:40
	sitaram has to go drop his son off somewhere; be back in about an hour or so	03:40
	← vmiklos left	03:40
	→ qrush joined	03:42
	→ scarabx joined	03:45
	→ vmiklos joined	03:46
	← sgrimm left	03:47
	→ hax0r1337 joined	03:48
kodama	sitaram: take care!	03:49
	→ catalis joined	03:49
	← kodama left	03:49
	← catalis left	03:49
	→ m104 joined	03:51
	→ sgrimm joined	03:51
	← wshimmy left	03:52
	→ Paf joined	03:59
	→ hobodave joined	04:01
	→ saurabh1403 joined	04:01
	→ wshimmy joined	04:01
EvanCarroll	how do i view the file as it was in HEAD^^	04:01
	→ Zenopus_ joined	04:02
EvanCarroll	git show kind of works	04:03
	ideall i wouldn't see the prepending +	04:04
jjuran	sitaram: I just used git cherry-pick until I passed the conflict and rebased the rest.	04:05
	← Zenopus left	04:06
	Zenopus_ → Zenopus	04:06
	→ gerberb joined	04:07
	← vmiklos left	04:09
	← justatheory left	04:11
k0001	EvanCarroll: what about git ls-tree HEAD	04:12
	→ vmiklos joined	04:13
k0001	EvanCarroll: what about git ls-tree HEAD^^ , and then git show SHA1_OF_THE_BLOB_YOU_WHANT	04:13
	→ jerbear joined	04:13
k0001	WANT* , jeez, is hard to write all in caps	04:13
sitaram	jjuran: rebase is basically a "sequence of cherry picks"	04:13
	→ justatheory joined	04:14
	→ Cynshard joined	04:18
	→ catalis joined	04:20
jjuran	sitaram: Right, the issue is that git stops to ask me for input and I don't understand the question. I understand that I have to revise a patch to resolve conflicts, but I lack the context to know what I'm doing.	04:22
	← bgerber left	04:23
jjuran	I am only an egg, apparently. :-)	04:23
	→ BarryCarlyon joined	04:24
sitaram	jjuran: the times that I have had a rebase stop with conflicts for me, it's been "git status", look for unmerged files, fix them in plain vim or meld if you want to get fancy (meld would be "git mergetool -t meld filename"), git add, and finally git rebase --continue	04:24
jjuran	sitaram: I'll try git mergetool next time.	04:27
	← saurabh1403 left	04:28
	← catalis left	04:28
	← Bass10 left	04:30
sitaram	jjuran: don't try git mergetool -t gvimdiff unless you're a vim jock, though any of the other tools in the list should be fine. I prefer meld when I have to move hunks wholesale to resolve, and vim to clean up any leftovers with a finer "brush" so to speak. And when I say vim, I mean just vim, not "git mergetool -t gvimdiff" :)	04:31
	→ saurabh1403 joined	04:32
	→ catalis joined	04:33
	← saurabh1403 left	04:34
	→ HiBuddy joined	04:35
	← HiBuddy left	04:35
	← girishr left	04:36
	← btanaka left	04:36
	→ qhoxie joined	04:37
	← poseidon left	04:41
	← ashleyw left	04:44
	← scarabx left	04:45
	→ scientes joined	04:46
	← bobmcw left	04:51
	← jerbear left	04:54
	← hyperair left	04:54
	← netoman left	04:55
	bleything → bleything\|away	04:55
	→ astrocub joined	04:57
	→ hyperair joined	04:57
hachi	anyone know what possibly can be done when git-svn hits a bad revision	05:03
	like... I want to just skip it	05:03
	because it's a branch that should have never happened... and was deleted in the next revision anyways	05:04
	→ arohner joined	05:04
	← curvature left	05:06
	→ curvature joined	05:06
	→ lightcap joined	05:06
	← gaveen left	05:07
	→ rylar joined	05:09
	bleything\|away → bleything	05:10
	← kalle_ left	05:12
sitaram	wait for Ilari to show up maybe? I seem to recall he knows a fair bit about git-svn (I myself have no experience of it, and the channel is probably half-asleep or doing its weekend chores!)	05:13
	→ kalle_ joined	05:13
	→ dbb joined	05:14
	← eletuchy left	05:14
	← ia left	05:14
	→ ocornu joined	05:14
	← iamwilhelm left	05:17
	← greyface left	05:22
	→ j416 joined	05:28
	← rylar left	05:29
	→ anjalishenoy joined	05:35
anjalishenoy	Hello	05:36
	I just signed up for a git plan and want to install git on my OSX (TIger) - does git have a binary for this?	05:36
	→ priidu joined	05:37
zpoley	anjalishenoy: i used port for this on one machine, and this: http://code.google.com/p/git-osx-installer/	05:38
	ajalishenoy: on another	05:38
	anjalishenoy: oh and also just by downloading and building from source	05:39
	← lightcap left	05:39
	← qrush left	05:39
zpoley	anjalisehnoy: i'm on leaopard though but i would think they should all work on tiger	05:39
anjalishenoy	zpoley: unfortunately the link you sent me features all installs and downloads for Leopard and not Tiger. I need installs for OSX Tiger	05:40
zpoley	oh sorry	05:40
anjalishenoy	zpoley: guess I'll just have to build from source then?	05:40
zpoley	anjalishenoy: yeah it's just standard configure make install process	05:41
	→ lightcap joined	05:41
anjalishenoy	zpoley: actually, there's dependency issues for Tiger	05:41
zpoley	anjalishenoy: oh really	05:41
	← raydeo left	05:41
anjalishenoy	zpoley: ok - is there a specific directory in which I should install git?	05:41
zpoley	anjalishenoy: i'm just curious, why haven't you upgraded to leopard >	05:41
	anjalishenoy: from source i would do /usr/local/	05:42
	anjalishenoy: that's the standard	05:42
	→ cardioid joined	05:42
anjalishenoy	zpoley: long story short- i have a few macs in my house and want to do a 'family upgrade' and am waiting for snow leopard	05:42
zpoley	anjalishenoy: ah ok	05:42
anjalishenoy	zpoley: as far as tiger dependencies go: check out: http://www.bergek.com/2008/09/22/install-git-on-mac-os-x-104/	05:42
	zpoley: looks like those have to be installed first, THEN git - hence I was asking if you could point me to one complete package	05:43
zpoley	anjalishenoy: ok gotcha, sorry about that	05:43
	anjalishenoy: i've been using os x for about two years and just always keep it up to date, and haven't had such a problem with it	05:44
anjalishenoy	zpoley: which version are you running?	05:44
zpoley	anjalishenoy: 10.5.6	05:45
	← hax0r1337 left	05:45
zpoley	anjalishenoy: my main problem with installing git always had to do with perl	05:45
anjalishenoy	zpoley: that's Leopard (although I don't think that's been out for 2 years)	05:45
	zpoley: what about perl ?	05:46
sitaram	I think you need expat and asciidoc only for documentation -- if you can get by with HTML on the web you can build just the binaries without those two	05:46
	so you won't have local man pages or HTML docs that's all	05:46
zpoley	anjalishenoy: there been issues with perl different perl modules compatibility	05:46
anjalishenoy	zpoley: is this a known issue for tiger or osx in general?	05:47
	← niki left	05:47
zpoley	anjalishenoy: that's just my experience installing it on a couple machines	05:47
anjalishenoy	zpoley: ok. thank you for your help	05:47
zpoley	anjalishenoy: i also had port perl installed though too	05:47
	anjalishenoy: sure hope it helps	05:48
anjalishenoy	zpoley: I can't install it via macports can I?	05:48
zpoley	anjalishenoy: you can try i suppose but i don't know about tiger i guess	05:48
anjalishenoy	zpoley: ok. thanks again	05:48
zpoley	anjalishenoy: sudo port install git-core or sudo port install git-core +svn if you need git-svn	05:49
	anjalishenoy: sure, good luck	05:49
	anjalishenoy: it's awesome when you get it working	05:49
anjalishenoy	zpoley: that's it git-core is all I need?	05:49
	← lightcap left	05:49
	← cemerick left	05:49
zpoley	anjalishenoy: it should be	05:50
anjalishenoy	zpoley: kk cool, wil try it out	05:50
zpoley	anjalishenoy: have fun	05:50
hachi	sitaram: thank you	05:54
	Ilari: if you happen to come around and have a moment, please :)	05:55
Ilari	...	05:56
	→ eletuchy joined	05:57
hachi	oh my	05:57
	I'm looking for a way to bypass a bad pair of revisions with git-svn	05:57
	← aspotashev\|eeepc left	05:57
	→ eno__ joined	05:57
hachi	someone made a bad svn cp, and then deleted it	05:57
	→ aspotashev_ joined	05:58
Ilari	No, I don't know much specifically about git svn. Especally not about internal errors such as bad revisions.	05:58
hachi	oh, there's no error in git-svn itself	05:58
Ilari	hachi: How you even make bad svn cp?	05:58
hachi	the repo is enormous	05:58
Ilari	You saying server goes crazy when it hits that version then?	05:58
hachi	and roughly speaking, someone copied a large portion of it into a smaller part	05:59
	→ kate21de joined	05:59
	← eno left	05:59
	eno__ → eno	05:59
hachi	I can make an example if you'd like... perhaps I should even if you don't	05:59
Ilari	hachi: You mean copied foo to foo/bar or something like that?	05:59
hachi	they copied /a/trunk into /b/tags/something	06:00
	rather than /b/trunk	06:00
	so then it brought in the entire history of /a/	06:00
Ilari	hachi: Can one specify multiple revision ranges (-r)?	06:01
	→ TokyoDan joined	06:01
hachi	I don't believe I can make it skip part	06:02
	it just will wait till the next time I call it and it will try again	06:02
	I might mirror the svn repo and somehow surgically erase the revisions	06:03
	but I'm hoping for a better way	06:03
	→ hobodave_ joined	06:03
hachi	hmm, docs seem to contradict my experience, I will try again	06:04
	→ khmarbaise joined	06:05
	← simosx left	06:08
	← schacon left	06:08
	→ hulo joined	06:10
	← kalle_ left	06:12
	→ kalle_ joined	06:13
	← arohner left	06:14
	← khmarbaise left	06:15
	→ hax0r1337 joined	06:18
	← hobodave left	06:20
hachi	Ilari: you were correct in any case	06:21
	→ aspotashev\|eeepc joined	06:25
	← petercoulton left	06:25
	← aspotashev_ left	06:26
	← hax0r1337 left	06:42
	← stuffcorpse left	06:45
hulo	hello. how do i do a git clone, with creation of the repo in . ie say i don't want creation of a subdir reflecting repo name, but for all dirs and files in git repo to go .	06:45
	→ stuffcorpse joined	06:45
hulo	do ijust specify target directory at the end of the command? ie git clone git@repo . ?	06:46
	← vmiklos left	06:52
	← aspotashev\|eeepc left	06:54
	→ vmiklos joined	06:55
	→ naeu joined	06:56
	← Paf left	06:56
	← anjalishenoy left	06:58
	→ naeu_ joined	07:04
	→ khmarbaise joined	07:05
	→ aspotashev\|eeepc joined	07:05
	→ tvakah joined	07:08
	← hobodave_ left	07:08
	← darrob left	07:09
	→ orbii joined	07:12
	← kalle_ left	07:12
	← naeu left	07:13
	→ darrob joined	07:13
	→ kalle_ joined	07:13
	← naeu_ left	07:15
	→ naeu joined	07:15
	→ Enselic joined	07:17
	← dodo_the_last left	07:20
	→ kraymer joined	07:20
	→ dodo_the_last joined	07:20
Arrowmaster	hulo: yes	07:21
	← naeu left	07:22
	→ Weaselweb joined	07:22
	→ wmoxam joined	07:23
	← neoeinstein left	07:24
	→ neoeinstein joined	07:25
	→ drizzd joined	07:30
	← Enselic left	07:31
hachi	ugh, gitweb README says there are a ton of Environment variables I can set to configure things	07:31
	reading the code... they're all gone	07:31
	→ DavidKlein joined	07:33
	← aspotashev\|eeepc left	07:34
	→ aspotashev\|eeepc joined	07:36
	→ dwave joined	07:37
	→ JensB joined	07:38
JensB	hi everybody	07:38
hachi	hi doctor nick!	07:38
JensB	I accidentally --amended to a commit I had already 'git svn dcommit'ed. It's just the deletion of a single file. but now (of course...) the dcommit fails. how do I undo this?	07:38
mugwump	using the reflog	07:39
	git reset HEAD@{1}	07:39
	← khmarbaise left	07:39
mugwump	(use git reflog to see exactly what value to use	07:39
	)	07:39
	sunblush → sunblush\|away	07:40
JensB	the value should be the one showing "commit (amend): ...", right? this would be HEAD@{2}	07:40
	before that I have HEAD@{1}: checkout: moving from svon to b13dbfc2...	07:41
	and HEAD@{0}: rebase: updating HEAD (this is how I found out it didn't work)	07:41
	mugwump, "git reset HEAD@{2}" did not have any effect (no ouptut, no change in 'git status'), is that correct?	07:42
	→ Davedan joined	07:43
	→ woei joined	07:44
Davedan	I have two separate projects each with its own .git folde. If I have identical file in both projects, does git aware of that and saves only one blob?	07:45
killerchicken	Re-asking a question from earlier: How can I update a bare repository from a remote repository, so that gitweb will show the updates?	07:46
mugwump	JensB: what it should do is make it so that 'git diff HEAD' now shows changes	07:46
	← drizzd_ left	07:46
JensB	mugwump, thanks	07:46
mugwump	then you can 'git commit' and 'git svn dcommit'	07:46
JensB	I'll try, thank for pointing me to git reflog	07:46
	mugwump sees lots of questions, runs	07:47
hachi	mugwump: why am I here?	07:47
	what does this all mean?	07:47
	← Arrowmaster left	07:51
	← wmoxam left	07:51
	→ Arrowmaster joined	07:53
	→ eletuchy_ joined	07:54
	sdiz → sdiz-afk	07:56
	→ mbuf joined	07:56
mbuf	'git reset HEAD file' and 'git rm file --cached' are the same?	07:57
quentusrex	Help. I'm in the middle of a conflicted merge, but I didn't change anything. I'm trying to pull the latest HEAD from a remote repo...	07:58
	Any ideas?	07:58
	Automatic merge failed; fix conflicts and then commit the result.	07:58
	that's the error.	07:58
	→ rylar joined	08:05
	← jamescarr_ left	08:08
	← eletuchy left	08:10
	← dedmajor left	08:11
	← mikl left	08:11
	← qhoxie left	08:11
	→ jamescarr_ joined	08:11
	→ mikl joined	08:12
	← kalle_ left	08:12
	← sunblush\|away left	08:12
	← allagator left	08:13
	← rylar left	08:14
	← eletuchy_ left	08:14
	→ kalle_ joined	08:18
	→ Weiss joined	08:27
	→ solar_ant joined	08:27
	→ transplant joined	08:27
	← mikl left	08:28
	→ jfkw joined	08:29
transplant	google switched to mercurial?	08:31
	→ khmarbaise joined	08:32
	→ RodP2 joined	08:34
	→ trmanco joined	08:35
	← kuadrosx left	08:36
hachi	doener: your example from much earlier today has a flaw in it... I'd love to discuss if you have time sometime	08:38
	← TokyoDan left	08:40
	→ gilimanjaro joined	08:42
doener	hachi: got a "short version"?	08:42
hachi	when you do the clone of the 'bare' repo in your examples	08:43
	the branches all show up as refs/remotes/origin/$SVN_BRANCHNAME	08:43
	→ tlrobinson joined	08:43
doener	yeah, expected	08:43
hachi	but git svn is expecting them by default to be in refs/remotes/$SVN_BRANCHNAME	08:43
doener	hachi: that's why the "git svn init" call at the end gets --prefix=origin/	08:44
hachi	unless this behavior has changed since I last built my git	08:44
	ahh, I missed that detail	08:45
	→ cilly joined	08:46
	← transplant left	08:46
tlrobinson	i seem to be having trouble configuring which remote branch to use for push	08:47
	i have "push = refs/head/experimental" under my "exp" branch	08:48
	→ aspotashev_ joined	08:48
doener	branch? push settings are setup for remotes, not branches	08:48
tlrobinson	but when i do "git push" while that branch is checked out it tries to push master to master	08:48
	oh	08:49
	← AkumaStreak left	08:49
tlrobinson	that would explain it. there's no way to set it on a per-branch basis?	08:49
	← krawek left	08:49
doener	if you just want "push the current branch" behaviour, you can set remote.<name>.push to HEAD	08:49
tlrobinson	ah	08:50
	what if i want "push branch foo to remote branch bar, branch x to remote branch y, etc"	08:52
	← Davedan left	08:52
tlrobinson	(where git push acts on the current branch)	08:52
	← orbii left	08:55
	→ solydzajs_ joined	08:55
	→ SRabbelier joined	08:56
	→ qhoxie joined	08:57
	→ solydzajs__ joined	08:57
	← solydzajs_ left	08:58
	← solydzajs__ left	08:58
	← solydzajs left	08:58
	← solar_ant left	09:00
	← aspotashev\|eeepc left	09:02
	→ solar_ant joined	09:03
	← khmarbaise left	09:08
	← JensB left	09:11
	← kalle_ left	09:12
	← ned left	09:13
	→ jnareb joined	09:14
	← jnareb left	09:16
	→ kalle_ joined	09:18
	← bobesponja left	09:20
	← Octalot left	09:23
	→ tzafrir_laptop joined	09:23
tzafrir_laptop	I put a remote copy of a local repo of mine over http . When trying to clone that remote copy I get:	09:24
	warning: remote HEAD refers to nonexistent ref, unable to checkout.	09:24
	I have no problem cloning the same directory when working localy on the server	09:25
	the URL: http://git.tzafrir.org.il/git/asterisk-tools.git	09:25
	← gilimanjaro left	09:26
drizzd	You have to run git update-server-info in the http-hosted repo. Also enable the post-update hook if you don't push to it over http execlusively.	09:27
	jayallen → jayallen-afk	09:27
	→ chalky joined	09:27
	← cilly left	09:28
tzafrir_laptop	drizzd, run where?	09:28
	(and no: this is a read-only repo)	09:28
	← fynn left	09:28
tzafrir_laptop	I push over ssh	09:28
	→ vuf joined	09:28
	→ ned joined	09:29
drizzd	You can push over ssh to a repo which is also accessible via http.	09:29
tzafrir_laptop	ah, ok	09:29
drizzd	In that case you need to run git update-server-info every time you push over ssh, because	09:29
	the ssh protocol does not need it	09:30
	So you let the post-update hook do it for you	09:30
	in any case, it doesn't hurt to enable to post-update hook	09:30
	← ocornu left	09:31
drizzd	I wonder why it doesn't complain about missing info/refs. I'm sure it used to...	09:34
tzafrir_laptop	I ran 'git update-server-info' on the repo on the server, but still get the same error	09:34
drizzd	Here it does, actually. Do you have a very old git version?	09:35
tzafrir_laptop	The one from Debian Stable . 1.5.6	09:35
	1.5.6.5	09:35
drizzd	well, info/refs is still not there. It should have been created by git update-server-info.	09:35
	That's not too old.	09:35
tzafrir_laptop	what does this message mean technically? Which specific ref?	09:35
	err... which HEAD	09:36
	?	09:36
drizzd	it probably refers to the master branch, which it can't find without info/refs.	09:37
tzafrir_laptop	1c19aa9ba240b7a84b431ee1aef57bcb3b256eef refs/heads/master	09:39
	.git/objects/1c/19aa9ba240b7a84b431ee1aef57bcb3b256eef exists as well	09:39
	→ MisterN joined	09:40
drizzd	tzafrir_laptop: I know. You need info/refs for git's http protocol to find it. That's why I told you to run git-update-server-info	09:40
tzafrir_laptop	it's just the all I have is a technical error message	09:41
	the standard google procedure did not help. Thus I try to actually understand it	09:41
drizzd	what's there not to understand?	09:42
tzafrir_laptop	what does 'remote HEAD' in the message refer to?	09:43
drizzd	tzafrir_laptop: it refers to the file HEAD in your remote repo	09:43
	HEAD denotes the currently checked out branch	09:43
	The error message has been fixed since your version of git. It's now "fatal: http://git.tzafrir.org.il/git/asterisk-tools.git/HEAD/info/refs not found: did you run git update-server-info on the server?"	09:44
	← solar_ant left	09:45
	→ solar_ant joined	09:45
drizzd	oops, just http://.../asterisk-tools.git/info/refs	09:45
	→ vienova joined	09:47
	→ yann joined	09:47
tzafrir_laptop	Hmm... seems I had to use: git update-server-info --force	09:47
	→ felipec joined	09:49
drizzd	Interesting. It says "When you remove an existing ref, the command fails to update info/refs file unless --force flag is given." I wasn't aware of that. But I don't see how it can be relevant for your case, since you didn't have an info/refs file in the first place.	09:49
	→ cydork_ joined	09:50
mbuf	if i ammend a commit (by mistake), how can I undo the same?	09:51
	undo, git commit --amend	09:52
drizzd	mbuf: each time you commit the old commit is stored in the reflog	09:52
mbuf	drizzd: i see	09:52
drizzd	you can refer to the previous commit using HEAD@{1}, HEAD@{2} and so on	09:52
	→ eletuchy joined	09:52
drizzd	so you can undo commit --amend by resetting to HEAD@{1}	09:53
mbuf	drizzd: cool	09:53
drizzd	(be careful this removes local changes) git reset --hard HEAD@{1}	09:53
mbuf	drizzd: of course	09:54
tzafrir_laptop	hmm.. no. My mistake. Working in the wrong remote repo.	09:54
MisterN	drizzd: @{n} is reflog, right?	09:55
tzafrir_laptop	and the relevant hook seems to be post-update	09:55
drizzd	MisterN: yes	09:55
	→ aziz joined	09:56
drizzd	tzafrir_laptop: indeed. You still need to run git-update-server-info yourself for the first time.	09:56
	MisterN: see also "git reflog show <ref>"	09:58
	→ aspotashev\|eeepc joined	09:59
sitaram	tzafrir_laptop, drizzd: the -f to update-server-info only affects info/packs, not info/refs	10:00
MisterN	drizzd: possible quite useful. fortunately haven't needed it yet.	10:00
	← aspotashev_ left	10:00
	→ chris2_ joined	10:00
drizzd	sitaram: The man page says something else	10:01
	→ Ademan joined	10:02
sitaram	drizzd: the man page does not even mention what -f does, other than the bug	10:02
	that is, it does not say when, apart from for deleted stuff, does -f do anything	10:02
drizzd	Yes, but either what you're saying is wrong or what the bug description says is wrong.	10:02
	← cydork left	10:03
drizzd	So we should either fix you or the manpage :-)	10:04
sitaram	or the code, since I looked at the code just now :)	10:04
	(before I said that...)	10:04
	→ Iota joined	10:06
drizzd	sitaram: looks like you're right	10:06
Iota	Hi, I've got my git repository working and stuff, but I want it to login for another user I made just for the git repo.	10:07
	→ LotR joined	10:07
	← tzafrir_laptop left	10:07
Iota	When I try to git clone from another computer, it asks me for the password for my personal account on the machine.	10:07
	What can I do to change the user? I've already set the user for the repository as the "git" user.	10:08
drizzd	Iota: man git-fetch, search for user	10:09
Gitbot	Iota: the git-fetch manpage can be found at http://git.or.cz/man/git-fetch	10:09
Iota	Ah, thanks.	10:09
	← trmanco left	10:10
sitaram	drizzd: I prefer to say "looks like the code is right" :) What one could add to the manpage is a description that when packfiles get stale, you want "-f" to remove the corresponding info/packs and start from scratch	10:11
	or some such thing	10:11
	← kalle_ left	10:12
	→ kalle_ joined	10:13
drizzd	sitaram: the bug the manpage is talking about has been fixed in 2005...	10:14
Iota	Sweet, so I can just tell people to git clone git@mymachine:myrepo.git.	10:14
	→ d0k joined	10:14
sitaram	oh wow. Although I don't get it really because there's a function called parse_pack_def that is supposed to check for stale pack names and pretend like "force" was applied if such is found	10:15
	drizzd: ^^	10:15
drizzd	I don't understand it enough to update the message either.	10:17
sitaram	drizzd: so there must be some combination of events that causes stale pack names in info/packs without being detected as stale? maybe... My programming-fu isn't that great these days so I'll just remember that if in doubt, use force :-)	10:18
	make like luke and use (the) force LOL	10:18
	→ Arafangion joined	10:20
mbuf	other than a pattern, do we need to specify anything else to git-grep ? git-grep commit (for example), doesn't return anything in the output	10:20
thiago	git grep <pattern> searches the current HEAD for that pattern	10:21
	additional options are: git grep <pattern> <commit> -- <paths>	10:21
mbuf	thiago: and that includes only the file sources?	10:22
thiago	what is there in addition to file sources?	10:22
mbuf	thiago: .git/ ?	10:23
sitaram	ignored files?	10:24
	→ girishr joined	10:24
	→ catalis_ joined	10:24
	→ trmanco joined	10:24
	← girishr left	10:24
thiago	mbuf: right	10:25
	mbuf: it searches the commit	10:25
	.git isn't part of the commit	10:25
	nor are ignored or generated files	10:25
mbuf	thiago: ok, thanks	10:25
	→ johan-s joined	10:29
	→ Grahack joined	10:34
	← solar_ant left	10:35
	← catalis left	10:36
	← hyperair left	10:39
	→ Frontier joined	10:43
	← Frontier left	10:44
	← davidad left	10:45
	→ solar_ant joined	10:46
jettero	how do you show what a tag points to?	10:47
	git show tagname prolly...	10:47
	→ saurabh1403 joined	10:49
	← johan-s left	10:53
	→ techleopard joined	10:55
	← catalis_ left	10:58
	→ codestr0m joined	10:59
codestr0m	I see pull as a command from git --help, but get the following error on opensolaris # git pull . branch ... git: 'pull' is not a git-command. See 'git --help'.	10:59
	← Grahack left	11:01
Iota	This might be a silly question, but have you tried it with the hyphen?	11:04
	git-pull	11:05
	← woei left	11:05
drizzd	codestr0m: git commands are stored in $prefix/libexec/git-core. Do you have those?	11:05
codestr0m	drizzd: /usr/libexec/git-core/git-pull	11:07
	← sitaram left	11:07
codestr0m	that's not in my path.. (should it be?)	11:07
drizzd	codestr0m: no	11:07
codestr0m	something is broken I think I found it	11:08
	← dodo_the_last left	11:08
	→ rudi_s joined	11:08
codestr0m	thanks	11:08
	← lazyshrk left	11:08
	→ gilimanjaro joined	11:09
	← vuf left	11:09
	← codestr0m left	11:11
	← kalle_ left	11:13
	← mnick left	11:14
	→ froomis joined	11:15
	→ simosx joined	11:15
	→ kalle_ joined	11:18
	← simosx left	11:21
	→ Tuomas joined	11:25
	← aziz left	11:28
	← gilimanjaro left	11:30
	→ gilimanjaro_ joined	11:30
	← chalky left	11:33
	← neoeinstein left	11:35
	→ aziz joined	11:37
	← aziz left	11:46
	→ solar_ant_ joined	11:47
	→ aziz joined	11:49
	← seangrove left	11:53
	→ jceb joined	11:55
	→ fridim_ joined	11:55
	← solar_ant left	11:57
	→ tatsuyao joined	11:57
	→ parasti joined	11:58
	froomis → froom	12:03
	→ corni joined	12:04
	sdiz-afk → sdiz	12:09
	← hummingkid left	12:10
	← froom left	12:10
	sdiz → sdiz-afk	12:11
	← vbabiy left	12:11
	← kalle_ left	12:12
	→ kalle_ joined	12:13
	← j416 left	12:13
	← Arafangion left	12:18
	→ dreamcat4 joined	12:22
dreamcat4	hi when i run "git stash save" it doesnt stash everything, only pre-existing modified	12:24
	not newly created files	12:24
	how best to "stash" those also ?	12:24
	→ chalky joined	12:25
	← jceb left	12:26
	→ antoranz joined	12:27
killerchicken	dreamcat4: looks like that isn't possible atm. See http://kerneltrap.org/mailarchive/git/2007/9/29/325480	12:31
	dreamcat4: The workaround is to git add the files before stashing.	12:31
dreamcat4	thanks @killerchicken	12:32
	so i can if i put them in the index (still uncommitted)	12:32
	well that's good enough	12:32
	← tatsuyao left	12:32
dreamcat4	i can unstage them after reapplying the stash	12:32
	← dwave left	12:32
killerchicken	if necessary, yes.	12:33
dreamcat4	thanks again, gotta go	12:33
	bye	12:33
	← dreamcat4 left	12:33
	→ dwave joined	12:33
	→ voker57_ joined	12:36
	→ fmoreau joined	12:36
	← fmoreau left	12:37
	→ fmoreau joined	12:37
	← fmoreau left	12:38
	→ fmoreau joined	12:38
	→ bryanl joined	12:40
	sdiz-afk → sdiz	12:40
	→ gabeodess_ joined	12:40
gabeodess_	hey, how do I create a branch of someone elses plugin?	12:40
Ilari	gabeodess_: Is it submodule or fully independent repository?	12:41
gabeodess_	Ilari: umm... I'm not sure	12:41
	what's the difference	12:41
	or is there some good documentation you could point me to...	12:41
Ilari	gabeodess_: Do 'git submodule update' (and that kind of commands) sound familiar?	12:42
gabeodess_	no	12:43
	← gilimanjaro_ left	12:43
gabeodess_	basically, there is a plugin I have been using (created by one of my associates), and I want to edit it, but not work on his master copy	12:43
	with the possibility of merging our copies at some point.	12:43
	← dwave left	12:43
Ilari	gabeodess_: If you want just to work on it, just clone it and work on it. Maybe even create second repo somewhere publicly accessable and push it there.	12:44
	← aziz left	12:44
gabeodess_	my git commands have been mostly limited to add, commit, push, pull..	12:44
	← saurabh1403 left	12:44
	→ jceb joined	12:45
gabeodess_	I created a fork... which I think is what you are saying... but he seemed pretty adament that I should do a branch instead, but I'm not sure what the difference is.	12:46
	→ sitaram joined	12:47
	→ khmarbaise joined	12:47
Ilari	gabeodess_: Fork presumably refers to another publically accessable repo. Repository contains one or more branches (and branches are always local in Git).	12:47
gabeodess_	Ilari: okay, so how would I create a branch then?	12:48
	I guess a branch would be a submodule?	12:48
Ilari	gabeodess_: Just clone it. Its another repo, so its also another branch.	12:48
	→ aziz joined	12:48
	← antoranz left	12:48
Ilari	gabeodess_: Or if you have existing clone, use 'git checkout -b newbranchname <startingpoint>'.	12:48
gabeodess_	Ilari: so then when I push, it wont push to the master?	12:49
Ilari	gabeodess_: Yes it would, but you don't have to push.	12:49
	→ Bass10 joined	12:49
gabeodess_	Ilari: but I would want my version under source control	12:49
Ilari	gabeodess_: Also, if you start with multi-branch repos, install git completion package and configure shell propt to display what branch you are on. Less confusing that way.	12:50
gabeodess_	ok	12:50
	→ jceb81 joined	12:50
gabeodess_	Ilari: so you are suggesting that I do not make a branch? just clone and work on it locally?	12:51
	← voker57 left	12:51
Ilari	gabeodess_: You can do either way. Clone to create new repo or create a new branch in existing repo. Either way, it will be a new branch.	12:52
gabeodess_	okay, so how do I create a new branch in the existing repo?	12:52
Ilari	gabeodess_: 'git checkout -b <newbranchname> <startingpoint>'.	12:52
	gabeodess_: You can omit <startingpoint>. Then it starts from what you have currently checked out.	12:53
	→ j416 joined	12:53
gabeodess_	cool, thanks, Ilari	12:53
Ilari	gabeodess_: But as said, multi-branch repos can be bit confusing. Displaying current branch name in shell prompt makes it less so.	12:54
	← jceb81 left	12:54
gabeodess_	ok, I'll look into that	12:55
	thanks	12:55
Ilari	gabeodess_: Many have had puzzling problems, and it eventually turned out that they where running commands on the wrong branch.	12:55
	← aziz left	12:55
	← gabeodess_ left	12:55
	→ saurabh1403 joined	12:56
	→ ben_h joined	12:56
hulo	hello. i need a bit of help setting up a branch. i did the following 1. git init 2 git add test 3 git commit -m 'test commit' 4 git remote add origin [email@hidden.address] 5 git push origin master 6. git checkout -b scratch 7. (modify files) 8. git commit -m 'commit from scratch' 9. git push	12:57
	and that fails to recognise the scratch branch	12:57
	→ altrux joined	12:57
hulo	at step 6, i tried 6. git push origin origin:refs/heads/scratch	12:57
	← solar_ant_ left	12:57
	→ aziz joined	12:58
hulo	error: src refspec origin does not match any.	12:58
Ilari	hulo: It doesn't try to push it because remote end doesn't have scratch branch and you didn't explicitly say you want to push it.	12:58
	hulo: 'git push origin master' means push 'master' to remote repo 'origin'.	12:59
hulo	Ilari: ah, i have to explicitly say it	12:59
	git push origin scratch at 9.	12:59
	?	12:59
Ilari	hulo: Pretty much.	12:59
	hulo: Its quite handy to explicitly have to say, since you might have branches you don't want to push.	12:59
	← sergiodj left	13:00
	→ Nemcue joined	13:00
Ilari	hulo: Better to make software DTRT 99% of time and offer the manual thing for the 1% than do the manual thing 100% of time.	13:01
	→ Hypnos joined	13:01
Ilari	hulo: Well, actually this isn't even the "fully manual" thing.	13:01
	→ solar_ant joined	13:02
Nemcue	Hey, I'm trying to reach the git-repos from a rails script/plugin install, but it doesn't work for some odd reason	13:02
	I can install plugins from other repositorys, but not from git..	13:02
Ilari	hulo: The fully manual thing would let to specify commit by SHA1 expression and specify name of ref to save it as.	13:02
	Nemcue: Define "Doesn't work".	13:03
Nemcue	Oh, sorry	13:03
	well.. it just doesn't do anything	13:03
	Says "plugin: version unknown"	13:03
	and downloads nothing.	13:04
hulo	Ilari: i created a script that sort of automates branch creation (based on a website's guide) http://pastebin.com/f4f47db3	13:04
Ilari	Nemcue: Maybe you need to grab the submodules as well (git submodule update --init)?	13:04
hulo	Ilari: can't that be run at step 6. ?	13:04
	→ leanucci joined	13:04
hulo	ie do i create the remote branch first, or do i create the local one, and then push it?	13:04
	← ben_h left	13:05
Ilari	hulo: No need to fetch. Push creates/updates remote branch mirrors for branches it touches.	13:05
	← dbb left	13:05
	chris2_ → chris2	13:05
	→ jceb81 joined	13:05
Ilari	hulo: And hardcoding local side starting point to 'origin' is bit questionable.	13:05
hulo	Ilari: is any part of that script obviously broke?	13:06
	← ikanobori left	13:06
hulo	Ilari: this is a very basic git repo (for learning purposes) and it'll always be origin	13:06
	← jceb left	13:06
Ilari	hulo: And hopefully that 'origin' means 'refs/remotes/origin/HEAD' and not 'refs/heads/origin'...	13:06
	→ scarabx joined	13:06
	→ solydzajs joined	13:06
hulo	huh?	13:07
	you mean origin is ambiguous?	13:07
	→ mikearr joined	13:07
Ilari	hulo: As ref, 'origin' can expand to 'refs/origin', 'refs/heads/origin' (local branch 'origin'), 'refs/tags/origin' (tag 'origin'), 'refs/remotes/origin' (remote tracking branch 'origin') or 'refs/remotes/origin/HEAD' (remote tracking branch 'origin/HEAD').	13:08
Nemcue	oh. err. wrong git.	13:09
	← Nemcue left	13:09
Ilari	hulo: And clone does set up remote tracking branch 'origin/HEAD' -> can be shortened 'origin'.	13:10
hulo	Ilari: wow	13:10
Ilari	hulo: Origin can also be nickname of repository. In that role, it designates another repository.	13:10
hulo	Ilari: hm, so in the contex tof the script, which seems to create a branch at the remote site (github), which should i be using?	13:10
	i udnerstand now that it shoudl be remote	13:10
	→ orbii joined	13:11
Ilari	hulo: Well, this is about what is the starting point of the new branch.	13:11
	hulo: Note that repositories can have multiple branches. So its not the branch from repository origin.	13:12
	← kalle_ left	13:12
hulo	Ilari: basically, i wanted to create the branch locally, make changes, the push to github	13:13
	*branches	13:13
	→ Kbyte joined	13:13
hulo	i understand now that avoidign the script would be my best bet	13:13
	→ kalle_ joined	13:13
	→ sikanda joined	13:13
	← leanucci left	13:13
	→ montylounge joined	13:13
hulo	as a: git checkout <newbranch_name>; changes; git commit ..; and git push origin <newbranch_name> would do	13:14
	→ dwave joined	13:14
Ilari	hulo: The way its done in 'git checkout'/'git branch' is that if unspecified, the starting point defaults to HEAD.	13:14
	← jceb81 left	13:15
Ilari	hulo: So maybe you want to edit the script so it takes optional second argument that defaults to HEAD, and change that push refspec to be '${second_argument}:refs/heads/${new_branch}' or so.	13:15
	→ jceb joined	13:17
	← corni left	13:17
hulo	Ilari: so, as an example second_argument could be refs/remotes/origin/HEAD ?	13:17
	→ dbb_ joined	13:17
Ilari	hulo: Nope, as an example 'HEAD'.	13:17
	→ bremner_ joined	13:17
Ilari	hulo: Which is its default value.	13:17
	hulo: One bad side about that change is that it requires relatively recent Git.	13:18
	hulo: So for exampple 'mkbranch foo bar' invokes 'git push origin bar:refs/heads/foo', and 'mkbranch baz' invokes 'git push origin HEAD:refs/heads/baz'.	13:20
hulo	Ilari: is 1.6.2.4 sufficient?	13:20
Ilari	hulo: According to /topic, its the newest release. It is new enough.	13:20
hulo	git push origin HEAD:refs/heads/${branch_name} still gives me the error	13:22
Ilari	hulo: What error?	13:22
	hulo: Hopefully the final command invoked doesn't have {} with it. Even through '${branch_name}' would be valid name for a branch.	13:23
hulo	fatal: remote part of refspec is not a valid name in HEAD:refs/heads/	13:23
Ilari	hulo: ${branch_name} seems to expand to empty string.	13:24
hulo	ah	13:24
	Ilari: i set branch_name in the script at the pastebin	13:24
	http://pastebin.com/m578e9f42	13:25
	← Hypnos left	13:25
Ilari	hulo: Its HEAD, not HEADS	13:25
	hulo: You are passing a parameter to that script.	13:25
hulo	oh yeah, it's HEAD. i just modified on-site to show that i updated as you advised. small typo. local script has HEAD	13:26
	→ okidogi joined	13:26
okidogi	How can I merge two commits to one, I just forgot the command.	13:27
Ilari	hulo: Also, that pull is bit questionable. It will only get anything if somebody pushes after step 3.	13:27
	→ romao joined	13:27
drizzd	okidogi: git merge	13:27
Ilari	okidogi: Or 'git rebase -i'?	13:27
hulo	Ilari: the website just listed that action as a way to ensure everything was up to date	13:28
	i wish i could find the url again	13:28
	seems the author would have problems with it too	13:28
	→ nothingHappens_ joined	13:28
okidogi	Ilari: I've no idea. I just commit twice but I want merge them together.	13:28
Ilari	okidogi: If thay are two last commits and neither is merge, 'git rebase -i HEAD~2' and mark the latter commit as 'squash'.	13:29
	→ lazyshrk joined	13:29
Ilari	hulo: There are lots of scripts that are hacked together by people who don't really understand what they are doing.	13:30
	→ benn_d joined	13:30
okidogi	Ilari: What do you mean by squash?	13:31
	← mlins left	13:31
Ilari	okidogi: As default, the action for both is 'pick', changing that to 'squash' combines commit with the previous one.	13:31
okidogi	Ilari: Ah, I see, thanks ;-)	13:32
Ilari	okidogi: Rebase interactive mode is swiss army knife of revision history manipulation.	13:33
	→ bremner__ joined	13:33
okidogi	Ilari: This feature seems really convenient, good point.	13:34
	← j416 left	13:34
	→ gabeodess joined	13:35
Ilari	okidogi: The main limitation is that it can't cope properly with merges.	13:35
	← solydzajs left	13:35
okidogi	Ilari: when?	13:35
	← bremner_ left	13:36
Ilari	okidogi: There's some preserve merges option, but it doesn't work in some older versions. Also, nonlinear rebasing would need extensions to script syntax.	13:36
	→ sukanto joined	13:36
okidogi	Ilari: It seems good enough for me now. I will prefer simply skip this. :)	13:37
sukanto	can anyone help me with a git clone issue ?	13:38
Ilari	sukanto: What issue?	13:38
sukanto	tmp $ git clone git://172.16.241.132:9418/linux-2.6.27.4.git Initialized empty Git repository in /tmp/linux-2.6.27.4/.git/ 172.16.241.132[0: 172.16.241.132]: errno=No route to host fatal: unable to connect a socket (No route to host) fetch-pack from 'git://172.16.241.132:9418/linux-2.6.27.4.git' failed.	13:38
	oops	13:39
	tmp $ git clone git : //172.16.241.132:9418/linux-2.6.27.4.git	13:39
Ilari	sukanto: Network-level error.	13:39
sukanto	but i can ping to that host	13:39
	← Bass10 left	13:39
Ilari	sukanto: Something's seriously wrong then. No route to host is layer 3 error. Ping works at layer 4, so if layer 3 doesn't work, ping can't work. Unless that error is spoofed.	13:41
	sukanto: Is it transistent or persistent error?	13:41
drizzd	does the git protocol support the url:port syntax?	13:41
	← aspotashev\|eeepc left	13:41
sukanto	event without port it is not working	13:41
	*even	13:41
	→ sikanda_ joined	13:41
gabeodess	terminal question: how do I remove a directory? I tried rmdir /my_directory and it says it's not empty	13:41
	→ bentob0x joined	13:42
	→ qrush joined	13:42
Ilari	sukanto: For me, that command seems to "hang" with connection in state SYN_SENT.	13:42
lazyshrk	gabeodess; rm -rf /yourdir	13:42
	gabeodess; but be careful with -rf	13:43
gabeodess	thanks, what's -rf?	13:43
	remove files?	13:43
lazyshrk	-r is recurse	13:43
	-f is force	13:43
gabeodess	ah	13:43
	thanks lazyshrk	13:43
Ilari	sukanto: Firewall problems?	13:43
drizzd	rm is remove	13:43
Ilari	sukanto: The usual way to test is via telnet: 'telnet 172.16.241.132 9418'. And err... Is that IP even globally routable?	13:45
	← hulo left	13:45
sukanto	let me check	13:45
lazyshrk	Ilari; it isn't... is it?	13:45
Ilari	sukanto: At least it doesn't have RDNS.	13:46
	→ gitte joined	13:46
	→ plaes joined	13:46
sukanto	even telnet doesn't works	13:46
drizzd	IIRC 172.16-31.x.x is for local networks just like 10.x.x.x and 192.168.x.x	13:46
sukanto	ya it is on a local network	13:47
	host-only network of vmware	13:47
plaes	hey.. any recommendations on how to use git with bash scripts...	13:47
gabeodess	I have a rails app that uses a 'schedulable' plugin. I'm trying to create a branch for the schedulable plugin, so I deleted the plugin, then cloned it, then from within vendor/plugin/schedulable I ran git checkout -b gabeodess. then I tried pushing, and I got "you can't push to git://github.com/user/repo.git Use [email@hidden.address] not really sure what to do with that, help?	13:47
plaes	currently git does not sensible return values	13:47
	→ Leemp joined	13:47
	Mr_Grieves\|zzz → Mr_Grieves\|	13:47
	← agile left	13:48
plaes	any hints how to detect whether the repo has changes?	13:48
lazyshrk	plaes: git status	13:48
	→ Sho_ joined	13:49
drizzd	test -n `git diff-index --name-only HEAD`	13:49
plaes	lazyshrk: did you read stuff I wrote before that?	13:49
MarcWeber	plaes: If you really want to dive in have a look at the git repo itself. It contains many examples	13:49
plaes	MarcWeber: thanks :)	13:49
	Ilari blackholes those networks.	13:50
plaes	well, now that gnome switched from svn to git, I have to rewrite my translation committing scripts :(	13:50
	→ mikl joined	13:51
Ilari	plaes: When using git from bash, avoid high-level operations and use the lower-level ones...	13:52
	plaes: What do those scripts even do?	13:52
	→ jrmuizel joined	13:53
plaes	it pushes translations (after review) from our local svn to gnome one...	13:54
Ilari	plaes: Well, if it just transporting commits raw, its just git push.	13:54
	← Ademan left	13:54
lazyshrk	anyone know why my newly cloned repo (msysgit), has a changed file (that I haven't touched) where it has removed every single line, and added them again, so it's the same file. i've tried to checkout the file, checkout -f, reset --hard, core.autocrlf = input, neither of which does anything. if I reset to the previous commit in gitk, the changes disappear for a second, then just pop back up.. any tips?	13:55
plaes	Ilari: yeah, but I need to do some checking first.. figure out the branch, check for local changes and finally push	13:55
offby1	lazyshrk: I generally set core.autocrlf to false when I use msysgit; give that a whirl	13:56
lazyshrk	offby1; i'll try that.. thanks	13:56
Ilari	plaes: Yeah, checking for local changes is valuable in order to give clearer error messages.	13:57
lazyshrk	offby1; hallelujah.. it worked! thanx	13:57
	← sukanto left	13:58
Ilari	plaes: For true low-level push operation, there is git send-pack. But that can only handle file:///, git:// (very rarely used with it) and ssh://.	13:58
	→ kukks joined	13:59
	← solar_ant left	13:59
	→ markelikalderon joined	14:00
	← Flathead left	14:00
Ilari	plaes: And to really control what goes on, you would need to reimplement pack sending. :-> There's tools to make packs to send (which is the most difficult step).	14:00
plaes	Ilari: I don't really need this stuff atm.. I think git add file; git commit -m file "updated file" ; git push is enough...	14:01
	← mbuf left	14:02
Ilari	plaes: That's bit more than just transporting commits. That's also making them.	14:02
plaes	yeah	14:03
	→ HG` joined	14:03
	→ j416 joined	14:04
	← sikanda left	14:04
	→ bremner_ joined	14:04
	→ solar_ant joined	14:05
	← bremner_ left	14:05
	→ Grahack joined	14:07
	→ iamwilhelm joined	14:09
Ilari	plaes: And "this isn't SVN". Git can do many kinds of multi-repo workflows that are simply not possible with SVN.	14:10
	plaes: And that kind of workflows tend to be needed in order to scale to highly decentralized projects.	14:10
	→ btanaka joined	14:11
plaes	Ilari: I'm just trying to keep things simple..	14:11
	← jrmuizel left	14:12
	← qrush left	14:13
Ilari	plaes: Ah yes, it's Gnome after all. :-)	14:13
plaes	well, to tell you the thruth, each repository conversion has made things a bit more harder..	14:13
	→ Casan joined	14:14
	→ jrmuizel joined	14:15
plaes	with cvs it was possible to download/update only the required file, with svn you could only download whole directory.. and for git you have to download whole source	14:15
	→ qrush joined	14:15
	← bremner__ left	14:15
Ilari	plaes: The smallest transferable unit with git is single branch.	14:15
plaes	yeah.. but see how things have gone actually worse :P	14:16
	translators mostly need to commit one file every once in a while	14:16
Ilari	plaes: At least it merges well.	14:17
rudi_s	plaes: If you just want the latest few commits you can use git clone --depth, so not the whow repository is necessary.	14:17
	s/whow/whole/	14:17
	→ Siavash joined	14:18
Siavash	hi all	14:18
Ilari	...	14:18
plaes	rudi_s: yeah, I know that.. but in order to update a single file I still either need to download whole source or cache it on my hdd :(	14:18
Siavash	is it possible to get the diff of a file on 2 different branches	14:18
	?	14:18
Ilari	Siavash: 'git diff branch1 branch2 -- file'?	14:19
plaes	git diff branch1..branch2 ?	14:19
gabeodess	can anyone help me with this?	14:21
context	if you read the man page	14:21
	it has examples	14:21
	that explain exactly how to compare branches	14:21
gabeodess	I have the gabeodess branch locally...	14:21
	→ jackdempsey joined	14:21
gabeodess	sorry... with this http://pastie.org/458009 ...	14:23
Siavash	thanks	14:23
	← dwave left	14:24
	→ aspotashev\|eeepc joined	14:24
Ilari	gabeodess: You can't push over git:// (at least to github).	14:25
	gabeodess: And addtionally the URL syntax is wrong.	14:25
gabeodess	Ilari: oh, so how do I push my branch?	14:25
Ilari	gabeodess: The Web page for that repository should give the push URL.	14:25
gabeodess	oh...	14:26
	Ilari: I see a 'clone url' where would I find the 'push url'?	14:26
Ilari	gabeodess: Maybe 'git@github.com:keithmgould/schedulable.git' or something.	14:26
gabeodess	Ilari: okay, that gave me a permission denied message.	14:28
Ilari	gabeodess: Except that that command pushes the ref as 'gabeodes'. The default branch is 'master', and there might not be way to change that (with github) and clone doesn't take missing default branch too kindly...	14:28
gabeodess	but the repo is public...	14:28
Ilari	gabeodess: You have account there?	14:29
gabeodess	I have my own account	14:29
Ilari	gabeodess: Presumably not 'keithmgould'?	14:29
gabeodess	but shouldn't I not need one if it is a public repo?	14:29
	→ cilly joined	14:29
gabeodess	Ilari: right, not keithmgould	14:29
Ilari	gabeodess: You need one for pushing.	14:29
	← yann left	14:29
gabeodess	so I can't push to keithmgould?	14:30
Ilari	gabeodess: Its keithmgould/schedulable.git. Wheither you can push to it depends on wheither you have been authorized to do so by admins of that repo.	14:30
gabeodess	I see	14:31
	← Siavash left	14:32
	← j416 left	14:32
	← jamescarr_ left	14:34
	← jceb left	14:35
	→ kuadrosx joined	14:35
gabeodess	Working now. Thanks Ilari.	14:35
	← aziz left	14:35
	← airtonix left	14:37
	← altrux left	14:39
	← eletuchy left	14:42
	← dbb_ left	14:42
	→ Enselic joined	14:43
	→ hacim joined	14:44
hacim	so I converted a svn repository to git and the tags ended up as directories... is there a way I can turn a set of directories into tags?	14:45
Ilari	hacim: You didn't pass -s or the correct layout to init/clone?	14:45
hacim	Ilari: yeah, I think probably not	14:46
Ilari	hacim: That could explain it.	14:46
gabeodess	Ilari: how can I configure to git to push to my branch when I send 'git push' ?	14:46
hacim	Ilari: i dont have the original svn repo available anymore, so I'm not able to re-do it	14:46
	← Enselic left	14:46
	→ schacon joined	14:47
Ilari	gabeodess: You need to configure remote for that repository (man git-remote) and then give it tracking data (branch.foo.remote and branch.foo.merge)	14:48
Gitbot	gabeodess: the git-remote manpage can be found at http://git.or.cz/man/git-remote	14:48
	← btanaka left	14:48
gabeodess	thanks guys	14:48
Ilari	hacim: Then probaly those tags aren't the only problem...	14:48
hacim	Ilari: actually... it looks like I do have it	14:48
	Ilari: now, I just have to do it the right way this time	14:49
	→ IListenU joined	14:49
	← jrmuizel left	14:49
	← aspotashev\|eeepc left	14:50
IListenU	screen: http://parduoduv.puslapiai.lt where's problem? please help.	14:50
Ilari	hacim: And turning set of directories into tags wouldn't have cut it. You would have needed the SVN directory copy data in order to reconstruct the history, as SVN has only single linear history at versioning level.	14:50
	← IListenU left	14:50
hacim	Ilari: ok, i can re-do it, but I want to make sure I do it right this time	14:51
sdiz	what does "git show --raw" supposed to do? it is not documented, and it is different from --pretty=raw	14:51
hacim	Ilari: is there a good guide somewhere that I should use?	14:52
	→ jamescarr_ joined	14:52
hacim	i used this before I think: http://www.simplisticcomplexity.com/2008/03/05/cleanly-migrate-your-subversion-repository-to-a-git-repository/	14:52
Ilari	hacim: Doing it right is more than getting the layout right. Its also about marking merges, getting authors right (--authors-file), stripping metadata, stripping any leading directories, etc...	14:52
	hacim: Its fairly easy to do when you are migrating, but royal pain to do afterwards.	14:53
hacim	Ilari: ok, I know how to do the authors-file bit, but the stipping and merging I'm not sure because that particular blog didn't even mention it	14:55
	it looks like I can use --stdlayout	14:56
	→ dermoth joined	14:56
Ilari	hacim: There's two ways to strip metadata. Either don't generate it in first place or strip it afterwards with git-filter-branch --msg-filter...	14:56
	← okidogi left	14:56
hacim	Ilari: yeah, looks like I used --no-metadata before	14:56
Ilari	hacim: But OTOH, metadata can make fixing the merges easier.	14:56
	hacim: Is that repo even multi-branch? If not, there's no merges to fix.	14:57
	→ jrmuizel joined	14:58
	← jackdempsey left	14:58
	→ bobmcw joined	14:58
	sdiz → sdiz-afk	14:59
	→ bobmcw_ joined	15:00
	sdiz-afk → sdiz	15:01
	sdiz → sdiz-afk	15:01
	→ aziz joined	15:01
	← cilly left	15:02
	→ jackdempsey joined	15:02
	← montylounge left	15:02
	thiago thinks the metadata is very useful	15:04
	← aziz left	15:04
thiago	many commits refer to revision numbers	15:04
	→ brtzsnr joined	15:04
Ilari	If I would retain that data, I would format it in nicer way.	15:05
	← plaes left	15:05
Ilari	Or I would fix the references to other commits.	15:06
	thiago formatted it in a nicer way in svn-all-fast-export	15:06
Ilari	thiago: Example?	15:06
thiago	and, most importantly, I removed the URL to the server	15:06
	that's useless information	15:06
	← dermoth left	15:06
	→ aziz joined	15:06
	→ dermoth joined	15:06
	← saurabh1403 left	15:07
hacim	Ilari: no, there is one branch. however its worse, I only have a checkout of the repository, not the original	15:07
	err, rather there is one additional branch	15:07
thiago	svn path=trunk/somewhere; revision=NNNN	15:07
Ilari	thiago: Is it smart enough to coalesce tags to their parent commits if possible?	15:09
	hacim: Has there been merges between those two branches?	15:09
hacim	Ilari: I dont think so	15:09
Ilari	hacim: Maybe grepping version history with keyword "merge" could be useful.	15:10
thiago	Ilari: SVN doesn't have tags	15:10
	Ilari: and I'm treating that as such	15:10
	SVN only has branches	15:10
	← jackdempsey left	15:10
	← fmoreau left	15:11
	← solar_ant left	15:12
	← kalle_ left	15:13
	→ kalle_ joined	15:13
	← Iota left	15:13
Ilari	In my view, SVN doesn't even have branches. It doesn't even have multiple 1 parent branches with implicit root revision. From that, remove requirements for implicit roots and allow derivation from multiple states and you get git branch model...	15:14
	← aziz left	15:14
thiago	yep	15:15
	I had to write some heuristic code to determine what a copy actually meant	15:15
hacim	hrm, it seems I dont have a complete copy of the original svn repo any longer	15:15
thiago	I think there are 4 types of copy-with-history in SVN (IIRC):	15:15
	- actual branching: copy of a module root to a path where it is known to be a branch	15:16
	- copying inside one repository: source and target paths are inside the same module and branch	15:16
	← bobmcw left	15:16
	← dermoth left	15:16
thiago	- copying of a subtree to a branch: i.e., creating a branch but only of the subtree	15:17
	→ dermoth joined	15:17
Ilari	hacim: Ouch. You are going to need to infer the missing data and fix the git repo accordingly. :-/	15:17
thiago	- supermodule copying: i.e., copying of a directory that contains modules	15:17
	bobmcw_ → bobmcw	15:18
hacim	Ilari: yeah, i've got all the tags imported into a tags dir in the git repo, and I dont care about the history... i'd just like to turn those into proper git tags at this point	15:18
Ilari	hacim: Presumably you also have /trunk and /branches/foo?	15:18
	→ ocornu joined	15:19
	→ dwave joined	15:19
hacim	Ilari: yeah, and actually the only branch in /branches is a junk one, so that can be ignored	15:19
	→ jackdempsey joined	15:20
thiago	hacim: once the import process is done, you can post-process and find proper tags	15:20
	if your "tag" branch contains exactly one commit and it's not making any changes, then it's a Git tag	15:20
Ilari	hacim: So its OK to lose it in import? And do those tags correspond to some versions on trunk?	15:20
hacim	thiago: i've already imported. I've got a tags/ and trunk/ directory in the root of the repo	15:20
Ilari	hacim: Ok, first make a backup of that repo.	15:21
hacim	Ilari: yes, its ok to loose that branch. the tags do correspond to some versions on trunk, yeah	15:21
	Ilari: already done :)	15:21
thiago	hacim: you don't see tags/ and trunk/ in Git	15:21
	they are branches. They'd be in .git/refs/tags and .git/refs/heads	15:21
	← dwave left	15:21
hacim	thiago: no, they are actual directories	15:21
Ilari	hacim: First use filter-branch with --subtree-filter to extract the /trunk.	15:21
thiago	hacim: no... they're not :-)	15:21
	hacim: if you see them, you did the import wrong	15:22
Ilari	thiago: That repo is imported wrong.	15:22
	thiago: And original SVN repo has been apparently lost.	15:22
thiago	oh	15:22
	that's a bummer	15:22
	directory-filter will help you make sense of the history of each branch	15:22
	the problem will be stitching the history together afterwards	15:23
	← gwoo left	15:23
Ilari	thiago: There's only apparently one. Fortunately. Unfortunately, there's tags.	15:23
thiago	ah	15:23
	if tags were done right, they have exactly one commit	15:24
hacim	Ilari: i think you mean --subdirectory-filter ?	15:24
thiago	and the contents of the tag dir is exactly the contents of trunk at the time the tag was made	15:24
hacim	thiago: right, thats what is in tags/	15:24
Ilari	hacim: Yeah, that.	15:24
hacim	this git repo is published, if you all want to have a look	15:24
	← faithfulgeek left	15:25
Ilari	hacim: The fixed-up thing will be incompatible with it.	15:25
	→ neoeinstein joined	15:25
hacim	Ilari: thats fine, i was just thinking that you might have a clearer view of what I've got if you had it in front of you	15:25
	its git://labs.riseup.net/crabgrass.git if you want to look	15:25
Ilari	hacim: Also, you probably should save output of 'git ls-tree HEAD:tags'. That contains information about where tags point.	15:26
	← dermoth left	15:27
Ilari	hacim: That should dump lots of lines of form "040000 tree 2b615ce7d59f5bc5a4a5ea829c396a2016f63285 tagname".	15:27
hacim	yeah, it does	15:27
	→ dermoth joined	15:27
	→ bcardarella joined	15:27
Ilari	hacim: With that information, its possible to match revisions after rewrite to tags (with some searching).	15:28
thiago	it's easier to match by history	15:28
Ilari	hacim: So save that info somewhere.	15:28
thiago	do a git log -n2 tags/tagname	15:28
	you'll see the commit that created the tag and one commit before	15:28
	the SHA-1 that ls-tree outputs for the tag name should match exactly the ls-tree output of trunk in that commit	15:29
	if that happens, you have a tag	15:29
hacim	Ilari: thiago git log -n2 tags/tagname only shows me one commit (the one before)	15:29
	but wait, we are getting ahead of things here	15:29
	I still haven't figured out how to properly do the git-filter-branch --subdirectory-filter trunk part yet	15:30
	→ ProperNoun joined	15:30
Ilari	hacim: Basically, matching can be done by trying 'git rev-list HEAD~n:' for various values of n (starting from 0 upwards). If you get match to saved ID entry pointed to, then 'git tag tagname <ID>'.	15:31
thiago	hacim: oh, right, of course	15:31
	but do a git log -n2 on that commit now (without the path)	15:31
	← jrmuizel left	15:31
Ilari	hacim: 'git filter-branch --subdirectory-filter trunk master'?	15:31
hacim	Ilari: ok, i've got the ls-tree info saved	15:31
	ok Ref 'refs/heads/master' was rewritten	15:32
	← wshimmy left	15:33
Ilari	hacim: Now it should contain only the trunk branch contents.	15:33
	→ wshimmy joined	15:33
hacim	Ilari: right, it seems to	15:33
	→ djgera joined	15:34
	→ jceb joined	15:35
	→ jrmuizel joined	15:35
Ilari	hacim: For restoring the tags... Take the first tag. Extract ID given by ls-tree and name. Then find n such that 'git rev-parse HEAD~n:' gives such ID.	15:35
	→ aisbaa joined	15:35
Ilari	hacim: Then 'git tag tagname HEAD~n'.	15:35
	→ solar_ant joined	15:35
hacim	where n = the sha1?	15:35
	← jackdempsey left	15:35
Ilari	hacim: n is small integer. Start from 0 and search upwards.	15:36
	← dermoth left	15:36
	→ j416 joined	15:36
Ilari	hacim: I'll clone it and take a look. The IDs resulting from filter-branch should be stable.	15:37
hacim	Ilari: so, first I do git ls-tree HEAD:tags/<tagname> to get the ID?	15:38
	→ dermoth joined	15:38
Ilari	hacim: Use the saved output.	15:38
hacim	Ilari: I have not pushed the results of the filter-branch	15:38
Ilari	hacim: I should be able to reproduce it.	15:38
hacim	Ilari: yes, I'm looking at the saved output	15:38
	if I have: 040000 tree d82a4b0e2da4832719b6800180c1ff6d196dc55ebackupninja-0.3.4	15:38
	err, the tab didn't represent	15:39
Ilari	hacim: It has ID of d82a4b0e2da4832719b6800180c1ff6d196dc55e and name backupninja-0.3.4	15:39
	← dermoth left	15:39
hacim	ok	15:39
Ilari	hacim: That's tree ID. Find the commit that has that as tree.	15:39
hacim	if I do: git rev-parse HEAD~d82a4b0e2da4832719b6800180c1ff6d196dc55e it says unknown revision	15:40
Ilari	hacim: For finding tree id commit has as tree, 'git rev-parse foo:', where foo is any valid commit.	15:40
hacim	ah	15:40
Ilari	hacim: Ah, I figured out simpler way.	15:40
hacim	oh great	15:40
	← wshimmy left	15:41
	→ nis joined	15:41
Ilari	hacim: 'git log --pretty=format:%H\ %T'. That prints tree IDs in right column and corresponding commits in left column.	15:41
	← HG` left	15:42
Ilari	hacim: Grep that for that tree ID.	15:42
hacim	ok, got that	15:42
Ilari	hacim: Then tag the commit given by corresponding left column value.	15:42
	hacim: Err... That repo appears to be post-filter state...	15:43
	hacim: And it doesn't seem to contain any state corresponding what was in that tag...	15:44
hacim	Ilari: errr, i gave you the wrong repo :PP	15:45
Ilari	hacim: Root dir contains: app AUTHORS Capfile CHANGES config db doc lang lib LICENSE log mods public Rakefile README script test tmp tools vendor	15:45
hacim	Ilari: git://labs.riseup.net/backupninja.git <--- thats the right one	15:45
	sorry, bad paste	15:45
	ok, so to tag the commit given by the corresponding left column value, I would do: git tag tagname HEAD~n where n=lefcolumnvalue	15:46
Ilari	hacim: Nope. The left column value is the commit ID.	15:47
	← djgera left	15:47
Ilari	hacim: And you can directly tag that.	15:47
hacim	ok, so just git tag tagname <leftcolumnvalue>	15:47
Ilari	hacim: That first tag gives 9f6af22a4c83e1126cd8b2eea50663e891e8439c	15:48
	→ dermoth joined	15:48
	← sikanda_ left	15:48
	← jrmuizel left	15:48
Ilari	hacim: So 'git tag backupninja-0.3.4 9f6af22a4c83e1126cd8b2eea50663e891e8439c'	15:48
	→ faithfulgeek joined	15:48
hacim	Ilari: got it	15:48
	→ jerbear joined	15:48
thiago	or git tag -a	15:49
	→ jrmuizel joined	15:50
	← dermoth left	15:50
	→ arohner joined	15:51
hacim	Ilari: interestingly, there are a couple in there that have no resulting tree id commit	15:51
Ilari	hacim: Ah, authors are already proper.	15:51
	→ sikanda joined	15:51
hacim	Ilari: the first 8 have them, but then the tag identified as 040000 tree 8d9f06695f0b9aaa5da6842ac7b4d9302fd1ee84 backupninja-0.9.1 doesn't output anything when I do git log --pretty=format:%H\ %T \|grep 8d9f06695f0b9aaa5da6842ac7b4d9302fd1ee84	15:52
Ilari	hacim: They have been modified in tagging then... You need to do commits for those...	15:52
	← fujin left	15:52
	← justatheory left	15:52
	← bcardarella left	15:52
Ilari	hacim: 0.9.3 doesn't have either. Well, maybe do 0.9.2 and rest that have them first.	15:53
hacim	Ilari: yeah 0.9.1 through 0.9.6 don't have them	15:53
	Ilari: i just did the ones that have them	15:53
Ilari	hacim: 0.9.2 does.	15:54
	→ LiamH joined	15:54
hacim	Ilari: yeah, i got that one. i misspoke when I said 0.9.1 through 0.9.6	15:54
Ilari	hacim: 299a43e39e2a2e5bc05b1eca24212d7a39aefde9	15:54
hacim	got it	15:55
	0.9 has one too, but the remaining do not (0.9.1, 0.9.3, 0.9.4, 0.9.5, 0.9.6)	15:55
Ilari	hacim: Going to need that backup of the repo. To find the versions previous to those tagged versions.	15:56
	← vienova left	15:56
	← jrmuizel left	15:56
	→ Tricon joined	15:56
Ilari	hacim: Looking at its history, no wonder it doesn't appear...	15:57
	→ qrush_ joined	15:58
	→ fujin joined	15:58
hacim	Ilari: in the svn repo I have tags for 0.9.5 and 0.9.6	15:58
	→ dermoth joined	15:58
Ilari	hacim: Apparently all those where tagged incorrectly and manually fixed later.	15:59
hacim	the joys of learning svn	15:59
	→ hummingkid joined	16:00
Ilari	hacim: Ok, found base commit for 0.9.1.	16:01
	→ Octalot joined	16:01
	→ jzellman joined	16:02
Ilari	hacim: Maybe do something like 'echo Actual 0.9.1 tag \| git commit-tree 8d9f06695f0b9aaa5da6842ac7b4d9302fd1ee84 -p 0e1bd269884c234dd76b00fa30bae674e82892bf' to make a commit for 0.9.1	16:02
	← mw left	16:04
Ilari	hacim: I found that 0e1bd... thingy by asking history for tags/backupninja-0.9.1 in pre-rewrite repo and picking first entry that has subtree tags/backupninja-0.9.1 that's also found from trunk and then do tree ID to commit ID lookup for that.	16:04
	hacim: That commit-tree outputs commit ID you can tag.	16:04
	→ solydzajs joined	16:05
	→ agile joined	16:07
	hacim was just taking notes on this	16:07
	→ HG` joined	16:08
	← dermoth left	16:08
	→ bcardarella joined	16:08
hacim	Ilari: ok, tagged 0.9.1 with that	16:09
	→ dermoth joined	16:09
	→ bcardarella_ joined	16:09
Ilari	hacim: Well, maybe commit message like "Backupninja 0.9.1" could have been better...	16:09
	← faithfulgeek left	16:09
	→ stmg joined	16:09
Ilari	hacim: At least better than 'Actual 0.9.1 tag'.	16:10
	neoeinstein is listening to an interesting talk at BarCamp Boston (#barcampboston) on Git feature branching.	16:10
	→ aspotashev\|eeepc joined	16:10
hacim	Ilari: hehe, well no matter	16:11
	← kate21de left	16:11
hacim	Ilari: ok, help me step through how you got that 0e1bd so I can do it for the other missing ones	16:11
	← kalle_ left	16:13
	→ kalle_ joined	16:13
Ilari	hacim: The process for 0.9.3 is: First search history of tags/backupninja-0.9.3. The first commit doesn't have match, so try second one (95b5f3b632fbf492b41e8c353553a17582ccc281). Looking up 95b5f3b632fbf492b41e8c353553a17582ccc281:tags/backupninja-0.9.3 yields b5b5b296c3e266570a842d732ae7ace3f1c50029. Now doing tree-ID to commit-ID on that yields cf0e2ac5d5efba9bbc2efc79901467fd94cf3f15	16:13
	hacim: (that's for the 0.9.3 tag).	16:14
	← stmg left	16:14
hacim	Ilari: searching the history of tags/backupninja-0.9.3 is doing: git log -- tags/backupninja-0.9.3 ?	16:14
Ilari	hacim: Yes. I noticed that second one yields no matches for 0.9.5, you need third one.	16:15
	hacim: 0.9.6 is the fifth(!) one.	16:16
	hacim: Yes, it was fixed 4 times.	16:16
hacim	Ilari: when you said abotu 0.9.3: 'the first commit doesn't have match' you are taking the commit ids in that history and then doing a git log \| grep <commit id> in trunk to see if you find a match?	16:17
	← gitte left	16:17
	→ abbe joined	16:18
	→ neoeinstein_ joined	16:18
	→ niki joined	16:18
	← dermoth left	16:18
Ilari	hacim: The way I'm looking it up is: "git log --pretty=format:%H\ %T \| grep `git rev-parse 95b5f3b632fbf492b41e8c353553a17582:tags/backupninja-0.9.3`". I got that 95b5f3b632fbf492b41e8c353553a17582 from git log.	16:19
	→ dermoth joined	16:19
Ilari	hacim: That yields: cf0e2ac5d5efba9bbc2efc79901467fd94cf3f15 b5b5b296c3e266570a842d732ae7ace3f1c50029	16:19
	← ProperNoun left	16:19
	→ jnareb joined	16:20
	← jnareb left	16:21
	← solar_ant left	16:21
	→ stmg joined	16:21
	→ greyface joined	16:21
	→ mw joined	16:22
	← qrush left	16:22
Ilari	hacim: Then its something like 'git tag backupninja-0.9.3 `echo Backupninja 0.9.3 \| git commit-tree 58fe867c71ac295ed1b15f8e306099bfa0e37d2c -p cf0e2ac5d5efba9bbc2efc79901467fd94cf3f15`'.	16:22
hacim	Ilari: hmm, 'git rev-parse 95b5f3b632fbf492b41e8c353553a17582:tags/backupninja-0.9.3' yields b5b5b296c3e266570a842d732ae7ace3f1c50029 here and then git log --pretty=format:%H\ %T \| grep b5b5b296c3e266570a842d732ae7ace3f1c50029 doesn't output anything	16:22
	→ gyukaku joined	16:22
Ilari	hacim: What repo are you running that log command?	16:22
hacim	Ilari: in my un-filtered backup	16:23
	→ Zenopus_ joined	16:23
Ilari	hacim: Run it in filtered repo.	16:23
hacim	ah	16:23
	that works	16:23
Ilari	hacim: Now make the commit and tag it.	16:24
	← dermoth left	16:24
	→ dermoth joined	16:25
gyukaku	Can I embed revision id like what cvs does with $Id: $ ?	16:25
	← jzellman left	16:25
	← bcardarella left	16:25
rudi_s	gyukaku: You can but you shouldn't.	16:25
Ilari	gyukaku: And it isn't commit ID. Its blob ID.	16:26
gyukaku	uh	16:26
Ilari	gyukaku: That is, it identifies file contents, not the commit it is from.	16:26
gyukaku	hmm	16:26
hacim	Ilari: i'm catching up... doing this on 0.9.4 now	16:26
	→ feanil joined	16:27
	→ eilartt joined	16:27
rudi_s	gyukaku: http://git.or.cz/gitwiki/GitFaq#head-4a0afe71a2bb7734777a8b0b345e0308aefdbd40	16:27
Ilari	hacim: I get 787e4...	16:27
hacim	Ilari: got that too	16:27
	→ jzellman joined	16:28
	→ woei joined	16:28
Ilari	gyukaku: The right way to do it is rely on VCS services while its in VCS, and add the revision info to files on export. git archive supports various keywords (like for revision ID).	16:29
	← rudi_s left	16:29
hacim	getting a little confused here	16:29
Ilari	hacim: With 0.9.5?	16:29
hacim	what is the first id that should be passed to git commit-tree?	16:29
	Ilari: no, still trying to get the tag right for 0.9.4, i have the 787e4	16:30
Ilari	hacim: The one from saved tags file.	16:30
hacim	ah, right!	16:30
Ilari	hacim: 71d02...	16:30
	hacim: With 0.9.5, note it was fixed _twice_	16:30
hacim	so: git tag backupninja-0.9.4 `echo Backupninja 0.9.4 \| git commit-tree 71d0292668701652ab7ea489721d320b583eb4b5 -p 787e4aadd2aa9786c95d897d781ebd67d7dd5875`	16:31
	→ fujin_ joined	16:31
	← neoeinstein_ left	16:31
gyukaku	Ilari: thanks. I will read about git archive	16:31
Ilari	hacim: Based on quick look, seems sane. And executes without errors.	16:31
	gyukaku: Also see man gitattributes (export-subst) and man git-log	16:32
Gitbot	gyukaku: the gitattributes manpage can be found at http://git.or.cz/man/gitattributes	16:32
hacim	ok, so because 0.9.5 was done twice, what does that mean I should do with that? it looks like I need the third commit	16:32
	← Zenopus left	16:32
	← benn_d left	16:32
	Zenopus_ → Zenopus	16:32
Ilari	hacim: Yes.	16:32
	→ tvakah_ joined	16:33
Ilari	hacim: (and 0.9.6 is even worse. It was fixed _four_ _times_)	16:33
	← neoeinstein left	16:34
	← tvakah left	16:34
	← orbii left	16:34
Ilari	gyukaku: man git-log because that format string uses the same syntax as --pretty=format: of git log	16:34
	→ solar_ant joined	16:34
Gitbot	gyukaku: the git-log manpage can be found at http://git.or.cz/man/git-log	16:34
hacim	Ilari: git tag backupninja-0.9.5 `echo Backupninja 0.9.5 \| git commit-tree d3897513b979dd90dd84f81d15f48a6ee392f658 -p cb2de66fc83c10fce3b8d6f1d5e439c9d9890f4c`	16:35
	Ilari: can you check my "math" on that? :)	16:35
	→ poseidon joined	16:36
Ilari	hacim: The IDs seem correct.	16:36
	← eilartt left	16:38
hacim	hm, i'm not finding it for 0.9.6	16:38
Ilari	hacim: I get ddb9dff1fbb36e658e80b7d065058bb6dcb3424c -> a325d02a4efadd60215a7be57fdfc65dc6a093aa -> d404d8b754758dd3ac27c0be280f3ef9187ffce4.	16:39
hacim	Ilari: ah, i didn't do ddb9	16:40
Ilari	hacim: And the proper tree ID is ebad12f2c58a4f221509f111d190d905946e62e3	16:40
	← arohner left	16:40
	→ hax0r1337 joined	16:40
hacim	git tag backupninja-0.9.6 `echo Backupninja 0.9.6 \| git commit-tree ebad12f2c58a4f221509f111d190d905946e62e3 -p d404d8b754758dd3ac27c0be280f3ef9187ffce4	16:40
	phew	16:41
Ilari	hacim: Seems reasonable.	16:41
	→ ashleyw joined	16:42
	← sikanda left	16:42
	→ neoeinstein joined	16:43
	← bcardarella_ left	16:43
	→ botanicus joined	16:43
	← ashleyw left	16:44
	← bobmcw left	16:44
	→ benn_d joined	16:44
	← Casan left	16:44
	→ eletuchy joined	16:44
hacim	Ilari: ok, now that I have this, i guess i'll have to do a git push --force to get it published	16:45
Ilari	hacim: Also explicitly list the tags in order to get those pushed.	16:46
	← fujin left	16:46
Ilari	hacim: So something like 'git push origin +master refs/tags/:refs/tags/'.	16:47
	→ bcardarella joined	16:47
hacim	Ilari: ah I just did 'git push --force --tags'	16:47
Ilari	hacim: There is also debian branch there.	16:47
hacim	Ilari: looks like I need to do the same thing for that branch	16:48
	as there are tags and a trunk/	16:48
	← neoeinstein left	16:48
	→ raydeo joined	16:49
Ilari	hacim: Yeah. Apparently. Hopefully there aren't lots of fixups there.	16:49
	→ saurabh1403 joined	16:49
	← peeja left	16:50
hacim	Ilari: I tried to re-write the trunk to debian: git filter-branch --subdirectory-filter trunk debian and it says, "Namespace refs/original/ not empty"	16:50
	→ killerchicken_ joined	16:51
Ilari	hacim: Delete that backup ref.	16:51
hacim	ok, i can just rm it	16:51
Ilari	hacim: Not proper way but should work.	16:51
	← fujin_ left	16:52
	→ neoeinstein joined	16:53
Ilari	hacim: Or this version I have can be forced with -f.	16:53
hacim	Ilari: i just tried the first tag, and there was only one commit, but it didn't have a matching tree	16:54
Ilari	hacim: Only 0.9.6-4 of those has it.	16:54
hacim	Ilari: ok, i tagged that one... what do you think I should do about the others? just loose them?	16:57
Ilari	hacim: Maybe use backupninja-0.9.5 as base for 0.9.5-1... Something like "git tag backupninja-0.9.5-1 `echo Backupninja 0.9.5-1 \| git commit-tree 813e9933ef7698dc5ac489dc26a8e3b6388d1563 -p backupninja-0.9.5`"	16:58
	← neoeinstein left	16:58
Ilari	hacim: And then 0.9.5-1 as base of 0.9.5-2, etc...	16:59
hacim	yeah, but that would miss the bits that were unique in the debian branch	16:59
	← jzellman left	16:59
	→ solar_ant_ joined	16:59
	← killerchicken left	17:00
	← solar_ant left	17:00
	← bentob0x left	17:00
	→ ashleyw joined	17:00
hacim	Ilari: i have the svn repo with all those tags actually	17:01
	could I use that to construct it?	17:02
	← bcardarella left	17:02
Ilari	hacim: master and debian diverged only in 2009(!) accoding to gitk.	17:03
hacim	Ilari: probably when I did the git-svn	17:05
	→ d0k_ joined	17:05
Ilari	hacim: Those tags all seem to appear in single revision.	17:05
hacim	Ilari: yeah, i might have done some rewriting or something to create that branch :P	17:06
Ilari	hacim: First it nukes all tags and then adds those tags.	17:06
	← sitaram left	17:07
hacim	Ilari: ok, well i have the svn pieces for the debian/tags directory, maybe that can be used in an interesting way?	17:08
Ilari	hacim: You would need to have the history for that to get anything interesting out of it.	17:08
hacim	ah and svn history would be contained on the remote	17:09
Ilari	hacim: What I cloned doesn't appear to contain any interesting history relating to those tags.	17:11
hacim	yeah	17:12
	← solar_ant_ left	17:12
	→ solar_ant joined	17:12
	← d0k left	17:12
hacim	Ilari: could I do something like git filter-branch --subdirectory-filter tags/0.9.5-1 debian ; git tag debian/0.9.5 and then re-do that for each tag from a fresh unfiltered repo?	17:13
	a bit of a pain, but it would at least get each tag's contents to be associated with the files	17:14
	→ petercoulton joined	17:14
Ilari	hacim: That would be even worse than chaining those tags.	17:15
	hacim: Since with that filter-branch, it would be just a single commit.	17:15
gyukaku	Ilari: it worked. thanks. I set export-sust in the gitattributes, then git-archive with commit id	17:17
	export-subst	17:17
hacim	Ilari: hrm. i could also just tag the entire repository setup as it is now, with a trunk/ and a tags/ dir, so that those older tags could be available if needed, and then move on from there	17:17
	→ ashleyw_ joined	17:18
Ilari	hacim: What I would do is to build 0.9.5-1 on 0.9.5, then 0.9.5-2 on 0.9.5-1, etc...	17:18
	← gyukaku left	17:18
hacim	Ilari: but doesn't that loose the changes that would have been different in 0.9.5-1 from 0.9.5?	17:19
	← bryanl left	17:19
hacim	(the one good thing is that the differences are always contained in the subdirectory called debian in their respective tag directories)	17:19
	→ masak joined	17:20
Ilari	hacim: Ah, there's debian branch imported as well...	17:21
	→ sitaram joined	17:21
Ilari	hacim: And it has commits like "[svn-buildpackage] Tagging backupninja (0.9.6-4)"	17:22
	→ sikanda joined	17:23
Ilari	hacim: So the repo indeed _has_ interesting history for those tags...	17:23
hacim	Ilari: right, hm should I just take those commit ids and try to get their tree ids?	17:23
	→ mnick joined	17:24
Ilari	hacim: But those don't have commit IDs unless you extract those branches as well...	17:25
	← nis left	17:25
hacim	Ilari: extract those branches... i'm not sure what branches you are referring to here	17:27
	← stmg left	17:28
	→ stmg joined	17:29
	→ AkumaStreak joined	17:30
	← ashleyw left	17:31
	→ Chaser_ joined	17:33
Ilari	hacim: branches/debian	17:34
	← scientes left	17:35
	→ sikanda_ joined	17:35
	→ jackdempsey joined	17:38
	← sitaram left	17:39
	← jackdempsey left	17:39
	→ lightcap joined	17:40
	← lightcap left	17:42
	→ gaveen joined	17:44
	→ jzellman joined	17:46
	→ radarek joined	17:47
	→ bobesponja joined	17:47
	→ lightcap joined	17:49
	← ashleyw_ left	17:52
	← doener left	17:52
	→ doener joined	17:52
	→ ashleyw joined	17:53
	← Kbyte left	17:55
	→ neoeinstein joined	17:56
	→ thumper joined	17:56
straszheim	% git submodule update --init	17:57
	Submodule 'spec' (git://sodium.resophonic.com/spec) registered for path 'spec'	17:57
	shift: 634: can't shift that many	17:57
	update --init looks broke	17:57
	→ jceb81 joined	17:57
straszheim	submodule init ; submodule update <- works	17:57
	← jceb left	18:00
	← sikanda left	18:01
	→ Blah12309 joined	18:01
Blah12309	×amsg×	18:01
	→ dwave joined	18:02
Blah12309	×amsg×	18:03
ehird	what	18:03
	← solar_ant left	18:03
Blah12309	×amsg×	18:03
	← dwave left	18:03
Blah12309	×amsg×	18:03
	×amsg×	18:03
	×amsg×	18:04
ehird	Blah12309: is this one of those crash bug things?	18:04
	fairly obnoxious.	18:04
	→ bcardarella joined	18:04
Blah12309	lol	18:04
	P	18:04
	O.o	18:04
Ilari	ehird: IIRC, those crash bugs where CTCPs, not regular messages to channle.	18:04
	*channel	18:04
ehird	Ilari: there may be some other kind, I suppose.	18:04
Blah12309	×amsg×	18:06
	→ Louie` joined	18:06
ehird	Blah12309: go away	18:07
	→ davidad joined	18:07
	← jzellman left	18:10
	← Chaser_ left	18:11
	→ nosatalian joined	18:14
	→ tlyng joined	18:14
Mr_Grieves\|	Quick question -- I want to pull an old version of a repo on git hub, but I can't seem to figure it out :-/ it should be fairly simple, right?	18:14
	← feanil left	18:15
Blah12309	×amsg×	18:15
Mr_Grieves\|	I've tried checking out an older branch (git checkout -b <hash>), but stuff newer than <hash> was in my copy afterwards?	18:16
Ilari	Mr_Grieves\|: That's syntax to create branch named <hash> from current version.	18:16
	Mr_Grieves\|: You probably want 'git checkout -b branchname <hash>'?	18:17
	→ sfwc joined	18:17
Mr_Grieves\|	Aha :) Thanks	18:17
Blah12309	v	18:19
	→ krawek joined	18:20
Mr_Grieves\|	Awww, Blah12309 thinks he's clever...how cute.	18:20
Ilari	Mr_Grieves\|: BTW, <hash> is IIRC valid name for a branch. :-)	18:20
Mr_Grieves\|	Ilari -- It is :) The second syntax worked perfectly.	18:20
	I was close, at least ;-)	18:20
	→ gaveen_ joined	18:21
Ilari	Mr_Grieves\|: Yup, verified with git-check-ref-format. There are total of 88 ASCII characters that may appear in git branch names. Plus loads more from Unicode...	18:22
Blah12309	everyone!	18:22
	← Blah12309 left	18:23
Grum	another fool who found out /amsg :(	18:24
	← khmarbaise left	18:24
Ilari	What's /amsg? How it is different from /msg?	18:24
	→ feanil joined	18:26
	← neoeinstein left	18:28
Ilari	Ah, apparently send to all channels you are on command...	18:29
voker57_	sends message to all channels	18:29
	→ gilimanjaro joined	18:30
Mr_Grieves\|	No ops here to kick/ban?	18:31
Ilari	Mr_Grieves\|: They got disconnected already.	18:31
	→ bcardarella_ joined	18:32
	Mr_Grieves\| is a little slow... :P	18:32
	→ bobmcw joined	18:32
	→ fynn joined	18:34
	→ eletuchy_ joined	18:35
	← eletuchy left	18:35
	← gaveen left	18:36
argonel	i just did a git reset <commit>, which was clearly the wrong thing to do.. how can i undo this?	18:38
	← bcardarella left	18:39
	← sikanda_ left	18:40
	→ neoeinstein joined	18:40
	→ remitaylor joined	18:41
	← gilimanjaro left	18:41
	← thumper left	18:42
remitaylor	does anyone know if there are hooks that get called even when everything is up-to-date? ie. can i run a script when a pull/push is done, even if there's nothing to pull/push because everything is up-to-date?	18:42
	→ Modius_ joined	18:42
drizzd	pre-update or something like that? not sure	18:43
	→ jackdempsey joined	18:43
wereHamster	argonel: git-reset just unstages files from the index. Simply git-add the files	18:43
drizzd	no, that doesn't exist	18:43
	how about pre-receive?	18:43
wereHamster	hm, git-reset HEAD@{1} would probably solve as well	18:44
remitaylor	drizzd: trying those	18:44
	← jamescarr_ left	18:45
	→ sikanda joined	18:46
remitaylor	nope. might not exist. kindof a bummer ... you'd think there'd be a hook that catches pretty much all connection attempts.	18:46
argonel	wereHamster: it undid my most recent commit	18:47
remitaylor	drizzd: i'll make sure i'm implementing all of the available hooks ... will post back if one works	18:47
wereHamster	argonel: did you use --hard?	18:47
argonel	wereHamster: nope	18:48
	→ Modius__ joined	18:48
wereHamster	argonel: look at the reflog (git log -g) and reset back to the commit you were before	18:48
argonel	wereHamster: thanks, head@{1} and that commit were the same :)	18:50
	→ bcardarella joined	18:53
	← saurabh1403 left	18:53
	← Modius_ left	18:54
	→ Paf joined	18:55
	→ ia joined	18:56
	→ jamescarr_ joined	18:59
	← Modius left	18:59
	← ashleyw left	19:02
	← j416 left	19:02
Ilari	remitaylor: Are there even any hooks that get run on fetch-type op?	19:02
Modius__	Git as I am using it seems to diff by comparing lines - is there a way to change that (specifically, to consider indentation insignificant if a large expression is indented) or is that kind of structurally carved in stone?	19:02
	← bcardarella_ left	19:03
	→ wsc joined	19:05
drizzd	Modius__: git diff -w	19:05
	← bcardarella left	19:05
remitaylor	Ilari: not that i can find, nope	19:05
	← mw left	19:06
	Modius__ → Modius	19:06
Modius	drizzd: Thanks - is there a way to get Git GUI to use that for its diffing?	19:06
	→ bcardarella joined	19:07
	gaveen_ → gaveen	19:07
drizzd	Modius: don't think so	19:08
	← poseidon left	19:09
drizzd	would be pretty counter-intuitive to hide those changes, since the main purpose for git gui is adding/removing changes.	19:09
Modius	drizzd: Guess I'm fundamentally wondering if Git, internally, works in terms of changes to lines, i.e. with lines, and if that is itself adjustable.	19:10
	← nosatalian left	19:10
Ilari	Modius: It doesn't. It works with whole-tree snapshots.	19:11
Modius	Ilari: If I have a file, with a line in it, and I modify that line's indentation, will it store a new copy of the full line?	19:11
thiago	basically, each file you add and commit is kept in entirety	19:11
	yes	19:11
	but that's before compression	19:12
	← Grahack left	19:12
Ilari	Modius: It essentially needs to store new version of the entiere file. It might be presented in terms of the other revision, but its still its own object.	19:12
Modius	Ilari: Understood.	19:12
Ilari	Modius: And when working on storage, all files are binary.	19:13
Modius	In my normal workflow, something that changes the indent of a whole function (common in lisp or certain functionalish code, more than in C) is highlighted as the whole function guts being removed and readded as something different.	19:13
	← kalle_ left	19:13
	← woei left	19:13
Ilari	Modius: You could even write your own diff program and have git call that.	19:14
	→ kalle_ joined	19:14
Modius	Ilari: At the repo level or just wrt my local workings?	19:14
Ilari	Modius: At object storage level. Even tag messages are 8-bit clean.	19:15
	← schacon left	19:15
Modius	Ilari: Any keywords I can search on to see examples of anyone doing that?	19:15
	← remitaylor left	19:15
	← sfwc left	19:16
Ilari	Modius: Probably no. Really using the fact that tag messages are 8-bit clean is quite insane.	19:16
	Modius: For commit messages, it wants some character set, but if character set with single-byte characters and all characters valid exists...	19:17
	→ ceej joined	19:22
jettero	I'm looking for a way to list all files from all revisions...	19:25
wereHamster	jettero: git-rev-list + git-ls-files	19:26
jettero	ls-files can list files from other revisions?	19:27
Ilari	jettero: '(for i in `git rev-list --all` ; do git ls-files -r --full-tree ; done) \| sort \| uniq'? Untested.	19:27
jettero	Ilari: something is missing from that lsfiles .... $i before the ;?	19:28
Ilari	jettero: (for i in `git rev-list --all` ; do git ls-tree -r --full-tree $i ; done) \| sort \| uniq	19:28
	← Leemp left	19:29
	→ saurabh1403 joined	19:29
	← sikanda left	19:29
Ilari	jettero: Doesn't work very well...	19:29
	jettero: Add 'cut -f 2' to pipeline before the uniq.	19:30
	← davidfetter left	19:30
Ilari	jettero: Actually, before sort.	19:30
jettero	?	19:31
	← bcardarella left	19:31
	→ dwave joined	19:31
jettero	k, first of all it looks like it would just list the files for the current HEAD, but for each revision....	19:31
	→ bla joined	19:31
jettero	oh, I see, nm	19:31
	git ls-tree --name-only seems to negate the need for a cut	19:32
	→ qrush joined	19:35
	→ robinr joined	19:35
	→ neoeinstein_ joined	19:35
	→ bryanl joined	19:36
	→ qrush__ joined	19:37
	← neoeinstein_ left	19:38
	→ neoeinstein_ joined	19:39
	← neoeinstein left	19:40
	neoeinstein_ → neoeinstein	19:40
	tedoc2000 → tedoc2000_afk	19:40
	→ Leemp joined	19:44
	→ mxey joined	19:45
	← qrush_ left	19:47
	← saurabh1403 left	19:47
	→ boto joined	19:49
	← robinr left	19:50
	← gabeodess left	19:51
	← raydeo left	19:53
	→ precision joined	19:54
	← boto left	19:54
	← qrush left	19:56
	→ gauge joined	19:57
	→ sikanda_ joined	19:57
gauge	killer	19:57
	→ hulo joined	19:58
hulo	hello. i understand the concept of .gitigore, but i was wondering if there was a reverse, like gitallow?	19:59
	if there isn't, is it possible to get alisting of untracked files in a repo, and dump that to .gitignore?	19:59
Ilari	hulo: 'git ls-files --untracked' or something like that?	20:00
	hulo: And .gitignore supports negative rules (prefix with '!'). Plus there's git add -f.	20:00
neoeinstein	`git clean -dn` would give you a listing of all untracked files not excluded by a .gitignore rule	20:04
	I think you can actually leave off the 'd' too.	20:04
	→ faithfulgeek joined	20:04
	← qrush__ left	20:06
	← ThaDon left	20:06
	→ pencilo joined	20:10
	← Paf left	20:11
	→ Grahack joined	20:11
	← faithfulgeek left	20:13
	← kalle_ left	20:13
	→ bcardarella joined	20:14
	← kraymer left	20:15
	→ kalle_ joined	20:18
	← bcardarella left	20:18
straszheim	ls	20:20
	→ mfilipe joined	20:23
	→ bcardarella joined	20:23
	→ kraymer joined	20:24
	→ scroogey joined	20:25
	← ceej left	20:25
	→ Dashkal joined	20:26
	← Weaselweb left	20:27
	→ kate21de joined	20:28
	→ ceej joined	20:28
	← gaveen left	20:29
	→ yann joined	20:29
	→ gilimanjaro joined	20:29
	← sikanda_ left	20:30
	→ coldwaterlover joined	20:30
scroogey	hello	20:30
	i had a question about git diff	20:30
	or more gitx or gitk	20:30
	→ sobersabre joined	20:30
sobersabre	hi. I'm on debian, and I've read the git-svn man page. it referred to something named "git-svnimport"	20:31
Ilari	scroogey: What question?	20:31
scroogey	you can use git svn clone to import svn repos	20:31
	Ilari: well, i have made changes	20:31
Ilari	sobersabre: Its deprecated.	20:31
scroogey	but it always says the whole changed	20:31
	instead of the lines	20:31
sobersabre	Ilari: ok.	20:31
Ilari	sobersabre: Sounds like CRLF problems or reidentation.	20:32
sobersabre	so IIUC: git svn clone URL would get all the history of an svn repo into a given git repo.	20:32
	Ilari: I think you're mistyping initial nicks all the time...	20:32
Ilari	sobersabre: Note that you need to specify the layout (there's dedicated switch for default layout).	20:32
scroogey	yeah like git svn clone -s http://alivepdf.googlecode.com/svn/trunk/ alivepdf-read-only	20:33
	well with out the trunk then	20:33
	← purplepenguins left	20:33
Ilari	scroogey: Sounds like CRLF probs or reindentation.	20:33
scroogey	Ilari: aah. bummer.	20:33
sobersabre	scroogey: how are the svn's usernames mapped into authors/committers ?	20:33
scroogey	sobersabre: no idea :>	20:33
sobersabre	is there a way to prepare some kind of map ?	20:33
	→ arohner joined	20:33
Ilari	sobersabre: Well, two nicks active both beginning with Sierra.	20:33
	sobersabre: There's authors file.	20:34
sobersabre	Ilari: where ?	20:34
Ilari	sobersabre: At least git-svn has option to take one in.	20:34
sobersabre	Sierra Oscar vs. Sierra Charlie	20:34
	Ilari: thanks, I'll RTFM the git-svn man to the end then.	20:34
	thanks all	20:35
	Out.	20:35
	;-]	20:35
	← sobersabre left	20:35
Ilari	:-)	20:35
scroogey	time to find out how to make an archive of HEAD~1	20:36
	bbl	20:36
	← abbe left	20:36
	→ Ian00 joined	20:36
Ilari	scroogey: 'git archive HEAD~1'?	20:37
scroogey	of course!	20:38
	too dumb :+	20:38
	→ sikanda_ joined	20:38
	← dwave left	20:39
scroogey	thanks Ilari	20:39
	bloody apple notebook	20:39
	Ilari: doy ou know a command to cleanup identation and CRLF	20:41
	when commiting?	20:41
Ilari	scroogey: There is CRLF cleanup mode that only cleans them on add. But git doesn't want to readd files that don't have modifications.	20:42
	scroogey: Few days ago I had problems with having some files that had CRLFs. Especially as Kate liked to strip them away.	20:42
scroogey	Maybe I am using the Mac and the other developer Windows? I think Linux/Mac is not using CRLF but only CR?	20:43
	because	20:43
	← aisbaa left	20:43
Ilari	scroogey: On Windows, the default is to enable CRLF translation both ways...	20:44
scroogey	aah.	20:44
	oh well.	20:44
	i will see if i am having the issue when my own projects	20:44
	let me check in GitX for my rails project	20:44
	← ned left	20:45
	← AkumaStreak left	20:45
	→ AkumaStreak joined	20:46
scroogey	there it works :)	20:46
	good	20:46
	→ dwave joined	20:48
	→ firewing1 joined	20:49
firewing1	I hope this is the right place to ask this... I'm wondering what your opinion/experience is with Github	20:50
	(I'm thinking of moving my local git repository for an open-source project there)	20:50
	→ seangrove joined	20:50
scroogey	works fine here	20:50
	← coldwaterlover left	20:53
neoeinstein	personally, I like github and find it simple and user-friendly to setup and work with.	20:54
firewing1	Alright, thanks for the info :)	20:54
	firewing1 hits "Sign up"	20:55
scroogey	:)	20:55
	yes, it's nice	20:55
	only govt doesn't allow me to use it :(	20:55
Ilari	scroogey: Some port blocking or policies?	20:56
	→ gitte joined	20:56
scroogey	security issues yes	20:56
	not allowed to store IP stuff on non-govt systems	20:57
	especially not outside the country	20:57
	→ ashleyw joined	20:57
scroogey	probably something to do that the govt got hacked from middle east :+	20:57
	by people	20:58
Ilari	scroogey: And probably no way to get internal git server? :-/	20:58
scroogey	we got internal one :)	20:58
	→ dodo_the_last joined	20:58
	← krawek left	20:58
	→ Enselic joined	20:59
scroogey	it was opensource and free	20:59
	that liked that	20:59
	they	20:59
	yet to find a way to export to perforce	20:59
	← dwave left	20:59
Enselic	Sometimes you only want to change the commit message of a commit a few commits back, is there a nice workflow for this other than git-rebase -i ?	20:59
Ilari	Enselic: Rebase -i is the easiest way probably.	21:00
wereHamster	filter-branch is probably overkill	21:00
	→ ned joined	21:01
	← bcardarella left	21:01
Enselic	I'd just like to double-click on the message area in gitk and start to edit, and when I'm done git does the rebase silently for me	21:02
wereHamster	Enselic: that's what one of the git gsoc projects is about :)	21:02
Enselic	wereHamster: yeah I saw that one, I have high hopes for it	21:02
	→ bryanl_ joined	21:03
	← yann left	21:06
scroogey	i like the git monster on the website :>	21:06
	← fynn left	21:08
	→ fynn joined	21:08
	→ robinr joined	21:09
	← robinr left	21:09
	→ robinr joined	21:10
	← bryanl left	21:10
	← radarek left	21:10
Ilari	Yeah, quite obiviously suffers from indeterminate growth from eating trees like candy. :->	21:12
	← kalle_ left	21:13
scroogey	yep :>	21:15
	← Grahack left	21:15
	sikanda_ → brianjlandau	21:17
	→ octagon joined	21:18
	→ kalle_ joined	21:19
	→ schaefer joined	21:19
offby1	I have never understood that monster.	21:20
	"eats trees and makes logs" ?	21:21
Fissure	it's also a "blob"	21:21
	→ sobersabre joined	21:22
sobersabre	hi.	21:22
schaefer	hello... i'm very interested in the status of the git eclipse plugin. does anyone know what the plans are to make the latest code available from an update site?	21:22
sobersabre	I ran git svn clone <repourl> -Aauthors	21:22
Ilari	sobersabre: -s missing?	21:23
sobersabre	I expected to run upto HEAD revision, but after r1000 it stopped, and started "auto packing your repository"	21:23
Ilari	sobersabre: Yeah, it does that and then resumes.	21:23
sobersabre	what do you mean "resumes" ?	21:24
	It exitted.	21:24
	does it revive on itself ?!	21:24
	← replaca left	21:24
sobersabre	and what is "-s" ?	21:24
Ilari	sobersabre: Use standard layout.	21:24
	→ vbabiy joined	21:25
firewing1	Is it possible to selectively push only certain commits to a remote origin?	21:25
	→ schacon joined	21:25
Ilari	sobersabre: Standard SVN layout.	21:25
firewing1	For example, I'd like to push only the last 5 commits to github	21:25
Ilari	firewing1: You need to rebase those commits to be first unpushed commits.	21:25
wereHamster	firewing1: certain branches, yes	21:25
sobersabre	I thought if my repo is based on trnk,branches,tags ... git would be smart enough to pickup that part...	21:25
	→ yann joined	21:25
Ilari	sobersabre: There's switch for that layout as its most common.	21:25
firewing1	Ilari: Will I have to redo the 'git add' steps after I rebase?	21:26
sobersabre	ok...so, what do I do to continue the svn clone from 1000 upto the HEAD of svn repository ?	21:26
	run it again ?	21:26
	← schaefer left	21:26
Ilari	sobersabre: If you did it without specifying layout, do it again, what it did is quite insane.	21:27
sobersabre	hmm.. you mean reinit the git repo, etc. ?	21:27
Ilari	sobersabre: 'git svn clone --stdlayout -Aauthors <URL>' or something.	21:28
sobersabre	ok... retrying...	21:28
	Ilari: thanks.	21:28
	← solydzajs left	21:29
	→ schaefer joined	21:29
Ilari	sobersabre: I have done some imports long enough to trigger that repack in the middle. For those imports after repack finished, it continued importing.	21:30
	← Enselic left	21:30
sobersabre	weird...	21:30
	I do as follows:	21:30
	mkdir -p workdir; cd workdir; git init; git svn clone <flags> https://hostname/svn/repos	21:30
	it writes:	21:31
	Initialized empty Git repository in /home/username/workdir/repos/.git/	21:31
Ilari	sobersabre: git svn clone does git init...	21:31
sobersabre	sh1te... ok.	21:31
	5th time!	21:31
Ilari	sobersabre: You can move it afterwards.	21:31
sobersabre	I want all the stuff to be in ~/workdir	21:32
	and I don't want any ~/workdir/repos to be created.	21:32
	shall I do:	21:32
	cd	21:32
	mkdir -p workdir; cd workdir; git init; git svn clone <flags> https://hostname/svn/repos workdir	21:32
Ilari	sobersabre: Well probably easiest to mv it after clone finishes.	21:32
sobersabre	sorry,	21:32
	I didn't need anything except the git svn clone and it's parameters.	21:33
	right ?	21:33
Fissure	right	21:33
sobersabre	ok... seems it does what I wanted.	21:34
	now, again, I don't understand the gc thing.	21:34
Ilari	sobersabre: git svn clone creates directory where it clones stuff. And IIRC it really wants to create it, otherwise it complains.	21:34
sobersabre	ok, now, what do I do if it ends the process after the r1000 ?	21:34
	I want everything upto HEAD, which is about 220 revisions further :)	21:34
Ilari	sobersabre: Pastebin the (end of) the output?	21:34
sobersabre	I will (when it ends). is there any special git-bin ?	21:35
Ilari	sobersabre: You can use any pastebin...	21:35
	sobersabre: /topic has URL of one, but there are others.	21:36
	→ sergiodj joined	21:37
	← ehird left	21:37
	← Tricon left	21:37
	→ ehird joined	21:37
	← feanil left	21:39
	← hax0r1337 left	21:39
	← masak left	21:40
	← ehird left	21:41
sobersabre	I'm watching it run... and I think something weird again happened.	21:41
	I think I saw git run over the same rev. numbers 2 times.	21:41
wereHamster	that's normal	21:42
sobersabre	as if it did some kind of loop.	21:42
	is it possible ?	21:42
	→ didi joined	21:42
sobersabre	I mean, in svn - it's all monotonic increment of revisions.	21:42
Ilari	sobersabre: It loops over various branches.	21:42
sobersabre	oh! now I saw it again.	21:42
	it went back to r400 something third time.	21:43
	very intriguing...	21:43
Ilari	sobersabre: That can happen in multi-branch repos.	21:43
	← yann left	21:43
	← brianjlandau left	21:43
	← arohner left	21:43
	→ yann joined	21:43
sobersabre	it's more intriguing than watching a progress bar progress...	21:44
	:)	21:44
	oh. now it's autopacking on r811....	21:44
didi	Is someone making a git server with gitosis? I am trying and although I am able to clone the gitosis-admin, make changes and commit, I am unable to push new repositories as get this response: ERROR:gitosis.serve.main:Repository read access denied	21:44
sobersabre	it's different than the 1st two clones...	21:44
	how can this be ?	21:44
	← offby1 left	21:45
Ilari	sobersabre: --stdlayout?	21:45
	← stuffcorpse left	21:46
Ilari	didi: Did you authorize yourself to read or write them?	21:46
	didi: And you pushed your changes right?	21:46
didi	Ilari: I followed this tutorial: http://scie.nti.st/2007/11/14/hosting-git-repositories-the-easy-and-secure-way	21:47
sobersabre	Ilari: yes. http://gist.github.com/101766	21:47
didi	Ilari: And added the supposed lines to gitosis.conf	21:48
Ilari	sobersabre: Messages would indicate it is switching branches.	21:48
	didi: Then did add, commit and push?	21:48
didi	Ilari: Yes.	21:48
sobersabre	Ilari: it hasn't completed the cloning process. so shall I do as it advised: run git gc ?	21:49
Ilari	sobersabre: It ran that.	21:49
sobersabre	so... do I get this straight: git svn clone didn't manage to complete.	21:49
	?	21:50
	→ Enselic joined	21:50
didi	Ilari: Is there a known bug or something?	21:50
Ilari	didi: Can you pastebin gitosis.conf and the URL you are trying to clone?	21:50
didi	Ilari: Sure.	21:50
Ilari	sobersabre: It exited immediately after that?	21:50
sobersabre	yes.	21:50
Enselic	Sometimes when I am doing a git-rebase and then git-rebase --continue, I get this error message "fatal: unable to create '.git/index.lock': File exists". It seems to happen randomly. I don' have gitk running. Anyone know why this is?	21:51
Ilari	sobersabre: Maybe try running 'git svn fetch --fetch-all' in that repo and see if it continues?	21:51
Enselic	I don't know what happens if I git-rebase --continue now	21:51
Ilari	Enselic: That would indicate git command having crashed in middle of updating index or some git command updating the index.	21:52
sobersabre	with -A../authors --stdlayout --fetch-all URL ?	21:52
	← ceej left	21:52
Enselic	Ilari: hmm could it perhaps be that I have EDITOR=gedit?	21:52
didi	Here is my gitosis.conf: http://pastebin.com/d5b743a6a	21:52
Ilari	sobersabre: Just 'git svn fetch --fetch-all'.	21:52
wereHamster	sobersabre: cd workdir/; git svn fetch --fetch-all	21:52
Ilari	didi: And what URL you are trying to clone?	21:53
didi	Ilari: Just a sec.	21:53
sobersabre	wereHamster: thanks, it continues from the last rev. it stopped furing git svn clone...	21:53
	hoping for the best now.	21:53
scroogey	:)	21:54
sobersabre	OK! it passed rev 1000 :)	21:54
Ilari	sobersabre: Still quite concerning that it stopped halfway without error messages...	21:54
sobersabre	Ilari: it wrote something weird: Following parent with do_switch	21:55
	whatever that's supposed to mean.	21:55
Ilari	sobersabre: do_switch is some SVN feature to speed up branch switching...	21:55
sobersabre	I do have a comment though	21:56
didi	Ilari: Here are the commands: http://pastebin.com/d4b0e7351	21:56
sobersabre	while running the initial command, I got a prompt, with no comment. it went like	21:56
	Password:	21:57
	I had no idea what password it was supposed to get.	21:57
	eventually it was supposed to be the ssl cert password.	21:57
	It's maybe some perl lib that does svn authentication, I don't know, but this is a very ... incomplete prompt.	21:57
firewing1	Hm, another question - I'm in the process of rewriting one of my programs from Python to C++... Am I better off creating branches/ and trunk/ directories, or using git's branch features?	21:58
	→ offby1 joined	21:58
	← moh left	21:58
sobersabre	firewing1: what a terrible thing to hear.... (I mean from python to C++)	21:58
firewing1	hehe... I need speed though :(	21:58
	← gitte left	21:59
offby1	hope you only wrote a little bit of the tight loop in C++ ... 95% of your program's time is spent in 5% of the code, so the cliché goes	21:59
	→ benn_d_ joined	22:00
Ilari	firewing1: Git's branching. branches/ and trunk/ is SVN braindamage.	22:00
	← benn_d left	22:00
didi	Ilari: Did you see something wrong?	22:00
firewing1	Ilari, OK thanks - I'll lookup a guide on that	22:00
	offby1: It's a backup program though, so just overall Python is unsuitable for what I'd like to do	22:00
	ie, tight SSH and SSL integration, fast file copies (with the rsync delta algorithm), etc	22:01
offby1	I wouldn't expect a backup program to use lots of CPU. Is it spending all its time figuring out which files to copy?	22:01
Ilari	didi: The git account on server should have raw copy of gitosis.conf. See if it matches what it should be (thats what it uses for access control).	22:01
	sobersabre is listening to Shpongle - "Nothing Lasts, but Nothing's Lost"	22:01
sobersabre	while git is looping back to rev. 350...	22:02
	hell.	22:02
firewing1	offby1: I'm not sure, actually, but Python is just terrible at it... for example when I did benchmarks, unpacking a tar archive takes 1m+ in Python but ~20s natively	22:02
Ilari	sobersabre: At least it did one branch...	22:02
sobersabre	Ilari: it did more than this, but I'm sitting here watching it with awe	22:02
	:)	22:02
offby1	firewing1: I can't argue with the numbers :) But I wonder if the problem is due to Python being slow, or if instead it's something else	22:03
ToxicFrog	That seems really weird, since in an untar disk IO should dominate over everything else.	22:03
Ilari	firewing1: Git doesn't have tools to deal with stuff like branches/ and trunk/ (outside git-svn importing), exactly _because_ its braindamaged.	22:04
didi	Ilari: Hum... I see a .gitosis.conf inside /srv/git (which is the user git's home). I did a "sudo cat .gitosis.conf" and the output was the old gitosis.conf, not the one I commited.	22:04
firewing1	I found it a bit odd too... but a lot of time with Python is spend dealing with the files, not copying the files themselves	22:04
sobersabre	Ilari: please correct me: git's concept of "branch" is very different than svn's, so it recreates all the possible histories of a given branch, and this is why it repeats some stuff.. right ?	22:04
firewing1	I wrote a simple file copy in C and it was more or less the same in Python IIRC	22:04
	← scroogey left	22:04
firewing1	but when you deal with 1000 small files, Python is so slow... It's just terrible	22:05
Ilari	didi: Assume git user and 'git show HEAD:gitoisis.conf' in gitosis-admin repo. Does that match what was pushed?	22:05
sobersabre	firewing1: it's very weird that untarring takes so much time in your python script than in C, since... they're supposed to use the same.	22:05
	firewing1: what file access routines are you using ?	22:05
firewing1	sobersabre: Nope, Python has it's own "tarfile" module	22:06
Ilari	sobersabre: More like its faster to follow one branch and then switch to others than grab whole revision at once and then move to next.	22:06
	← jayallen-afk left	22:06
	→ ceej joined	22:06
sobersabre	Ilari: If I understood hash idea - the branches should be linked to the subtrees of given node in the revisions graph.	22:06
ToxicFrog	firewing1: as far as branch management goes, the git tutorial (http://www.kernel.org/pub/software/scm/git/docs/gittutorial.html) has a decent intro, although it does kind of assume no preconceptions about what a branch is (ie, it doesn't explain them, it just shows how you behave and waits for you to infer)	22:06
	← kraymer left	22:07
sobersabre	so, I'm confused: revisions are also unique...	22:07
	oh... I'm not.	22:07
Ilari	sobersabre: Well that "looping-back" behaviour is essentially because its faster to do it that way.	22:08
sobersabre	if I am committing something under git, another son of my dir's parent doesn't "feel" it.	22:08
	right ?	22:08
ToxicFrog	The git user manual (http://www.kernel.org/pub/software/scm/git/docs/user-manual.html) has some more in depth stuff, and Git for Computer Scientists (http://eagain.net/articles/git-for-computer-scientists/) I found very helpful in grokking both how git stores stuff internally, and what git branches are.	22:08
Ilari	sobersabre: Branches are independent. Git doesn't have hierarchial branches.	22:08
firewing1	ToxicFrog: So I assume that in this case, I would branch my currently empty repo, commit the master as the Python version and work on the 'cplusplus' branch independently	22:08
Ilari	Also "Git from the Bottom Up".	22:08
firewing1	When I'm done, I just delete master and make 'cplusplus' master?	22:08
ToxicFrog	firewing1: yeah, or perhaps make the python one "python" and the c++ one 'master'	22:09
didi	Ilari: The "git show" show me the gitosis.conf that I pushed, but the /srv/git/.gitosis.conf has the first version of it. The .gitosis.conf also links to a inner .gitosis.conf (repositories/gitosis-admin.git/gitosis.conf) which also has the older version. I am confused...	22:10
	← offby1 left	22:11
	→ offby1` joined	22:11
Ilari	didi: It uses hook to update the gitosis.conf. Probably that hook didn't fire for some reason. Is that partition mounted 'noexec' (it doesn't work then)?	22:11
	offby1` → offby1	22:11
didi	Ilari: Hum. I don't know. How can I know it? It is on the same partition of my whole system.	22:12
sobersabre	run: mount	22:12
tony_	in a commit, can i commit some files, and reset the others?	22:12
sobersabre	didi: is your all system on 1 partition.. are you sure ?	22:13
wereHamster	tony_: selectively git-add the files you want to commit	22:13
	← bobesponja left	22:13
	← kalle_ left	22:13
Ilari	didi: Maybe assume git user, cd to gitosis-admin and run 'hooks/post-update' (or whatever it used).	22:13
	→ kalle_ joined	22:13
	→ bobesponja joined	22:13
sobersabre	Ilari: where does git's hash come from ?	22:13
didi	sobersabre: I have some partitions, but /srv/git is in the same of, say, /usr, /dev and /tmp	22:14
sobersabre	I mean it's a nice rate of quite a "carefully selected random hashes" rate.	22:14
didi	Ilari: Doing this now.	22:14
Ilari	sobersabre: Hashing object contents together with some headers.	22:14
tony_	wereHamster: thanks	22:14
	→ bcardarella joined	22:14
sobersabre	Ilari: I mean the rate of their generation is quite high.	22:14
	I doubt my machine has an RNG hardware.	22:14
	→ travisjeffery joined	22:15
sobersabre	COMPLETED!	22:15
	:)	22:15
Ilari	sobersabre: They aren't random but fully deterministic.	22:15
	sobersabre: Well, there is timestamps involved...	22:15
	← Enselic left	22:15
sobersabre	Ilari: I think the term "carefully selected randoms" is a good one.	22:15
	(it's not mine, my friend's)	22:16
Ilari	sobersabre: If two hashes are different, whatever they point to differs in some way.	22:16
	→ moh joined	22:17
wereHamster	.. with a probability of [nearly one, but not quite]	22:17
ToxicFrog	didi: make sure that hooks/post-update is marked executable in the first place, too; ISTR hearing that in some cases the installer will not mark it executable as it should.	22:17
didi	Ilari: Ok. It now works.	22:17
	ToxicFrog: Yes, that was it.	22:17
	ToxicFrog: Ilari remembered me of that. A chmod did it.	22:18
	Ilari: Thank you.	22:18
	My own git server. Nice! :-D	22:19
sobersabre	Ilari: it did import everything, but authors is messed up :(	22:20
	I mean, it didn't work.	22:20
	I have an authors file with 3 lines of the format:	22:20
	username = John Doe [email@hidden.address]	22:20
	I added it as -A../authors argument.	22:21
Ian00	how can i grep file, and every version of it in my repo and its history	22:21
Ilari	sobersabre: I don't know very much about git-svn... Sorry. :-/	22:22
	← Sho_ left	22:23
	← jceb81 left	22:24
	← HG` left	22:27
Ian00	how can i git-grep through history?	22:29
	← Leemp left	22:29
wereHamster	Ian00: git-rev-list \| git-grep	22:30
	← hdl left	22:33
Ian00	wereHamster, git-rev-list needs options. Also, it looks like git log -Sstring will find what i need, but i'd like it to search all branches, not just he history from the one i'm on. git-rev-list man page might help, but if you know, please speak up	22:33
	← dodo_the_last left	22:35
wereHamster	Ian00: for rev in `git rev-list --all`; do git grep whatever $rev; done	22:35
	← neoeinstein left	22:35
	→ patrikf joined	22:36
Ian00	wereHamster, nice thank you. looks like you can just pipe it to git grep also.	22:37
	→ Enselic joined	22:39
Enselic	I have made a git-rebase and lost a commit. I have the SHA1 of the commit. git-merge 54A1.. causes massive conflicts. Any trick so I can find the commit that was HEAD before my rebase so that I can redo it?	22:40
wereHamster	Enselic: reflog (git log -g)	22:41
	← vmiklos left	22:42
Enselic	wereHamster: interesting, will look closer at that	22:42
	→ Shadow_Master joined	22:43
	→ peeja joined	22:44
	← gilimanjaro left	22:44
	← priidu left	22:45
	→ nick_e joined	22:45
	→ matthewmcculloug joined	22:45
	← bcardarella left	22:47
	← parasti left	22:47
	← benn_d_ left	22:53
	→ benn_d joined	22:53
sobersabre	thanks Ilari	22:53
	and wereHamster	22:53
	← sobersabre left	22:53
	← DavidKlein left	22:56
	← vbabiy left	22:59
	→ noz joined	22:59
	→ redacted joined	23:00
	← patrikf left	23:00
	← d0k_ left	23:01
	→ vbabiy joined	23:02
noz	Using git svn (dcommit), I got the message "Unable to determine upstream SVN information from xxxxx history." Where should I look for the problem? In .git/svn?	23:03
	The repo was created with a git clone, then git svn init	23:04
Enselic	wereHamster: thanks, that did it	23:05
	← Enselic left	23:05
	OO11 → sejo	23:07
	← robinr left	23:08
	← kate21de left	23:08
	← techleopard left	23:11
	← jackdempsey left	23:12
	← kalle_ left	23:13
	← octagon left	23:17
	→ kalle_ joined	23:19
	← vbabiy left	23:19
drizzd	noz: just guessing here, but how is git supposed to know how your git history relates to the svn history?	23:21
noz	The original git repo was created from the svn repo, and after cloning, I did a 'git svn init' - I thought that was suppoed to do it?	23:22
	→ vmiklos joined	23:23
drizzd	probably not	23:25
	you could have done that with any git repo, totally unrelated to the particular svn repo	23:26
	did you do a git svn fetch?	23:26
noz	Yes - the full sequence was: git clone, git config --add remote.origin.fetch '+refs/remotes/:refs/remotes/', git fetch origin, git svn init ..., git svn rebase	23:28
	→ stuffcorpse joined	23:29
	← AAA_awright left	23:29
	offby1 → offby-laptop	23:29
	→ offby1 joined	23:30
drizzd	where did you get that? I don't think refs/remotes are advertised	23:30
	← offby-laptop left	23:30
	→ lea_w joined	23:30
	← lea_w left	23:30
	→ robinr joined	23:30
drizzd	so the ... is where you called git svn fetch?	23:30
	← iamwilhelm left	23:30
	← Octalot left	23:31
noz	Sorry - no. No git svn fetch was done. A git fetch origin, but not a git svn fetch	23:31
	← firewing1 left	23:31
	← didi left	23:32
	← qhoxie left	23:32
noz	The ... were just options to git svn init	23:33
drizzd	well, like I said. This probably doesn't do what you think it does.	23:33
noz	Thanks for your help.	23:33
drizzd	I'm not familiar with git-svn myself, but I'm sure there are tutorials out there	23:34
noz	That's the trouble trying to use someone else's recipe when you don't fully understand the concepts...	23:34
	→ justatheory joined	23:34
drizzd	I used "git svn clone" once, and it seemd to do the trick.	23:35
	→ sobersabre joined	23:44
	→ AAA_awright joined	23:44
sobersabre	hi. I have run git svn fetch --fetch-all,	23:44
	← redacted left	23:44
sobersabre	and got all the revs of svn repository.	23:44
	my problem is that authors file didn't "catch", i.e. I don't have authors as I stated in the authors file.	23:45
	Is there a way to go through all the revisions and substitude old author data with another ?	23:45
	I don't mind rebuilding the whole history during all this.	23:46
	→ ashleyw_ joined	23:46
	→ bcardarella joined	23:47
	← Modius left	23:49
	→ gautam_c joined	23:49
	← kuadrosx left	23:51
	→ offby1` joined	23:52
	← offby1 left	23:53
	offby1` → offby1	23:53
	← romao left	23:56
	← AAA_awright left	23:58
	→ AAA_awright joined	23:58

Logs Search ←Prev date Next date→

Channels Documentation