IRCloggy #git 2010-03-30

Provider of IRC logs since 2005.

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

	2010-03-30
Hillshum	opcode: I do too	00:00
opcode	yeah, it's a big departure from svn, and I find that I'm constantly getting into situations where I have no clue what's going on and get worried that I'm losing work	00:01
	← jelmer_ left	00:01
	→ qrc1 joined	00:02
cbreak	git does exactly what it says.	00:03
	it's a very simple structure	00:03
	with lots of tools to make that structure comfortable to use	00:03
	→ aryounce joined	00:03
	→ jelmer_ joined	00:03
	← tedoc2000 left	00:04
rudi_s	opcode: And unless you run git reset --hard or git checkout * or similar, Git won't/shouldn't destroy anything.	00:04
mugwump	heh, with svn you didn't need to be worried - you knew that your work would be lost :-)	00:05
	→ scarabx joined	00:05
mugwump	got changes to commit? 'svn update' and pray	00:05
rudi_s	;-)	00:05
opcode	is there any way to see what's gonna happen before I do a git svn dcommit?	00:05
Hillshum	cbreak: Once you understand it. The tools do all sorts of weird things until then	00:05
mugwump	oh... that update just ruined your work. bad luck.	00:05
rudi_s	opcode: git svn dcommit -n	00:05
	→ LoRe joined	00:06
rudi_s	Or maybe the -n before the dcommit, I'm not exactly sure and the man page isn't 100% clear.	00:06
opcode	dcommit -n gives me "diff-tree f21bfc0dbdf0b480be98fd6bae8b19e37fa05142~1 f21bfc0dbdf0b480be98fd6bae8b19e37fa05142" which isn't very descriptive	00:06
	→ jrmuizel joined	00:06
	← jelmer_ left	00:07
	→ gsanderson joined	00:07
rudi_s	opcode: It tells you it will push this commit.	00:07
mugwump	opcode: that's a command, run it	00:07
	prefix with 'git '	00:07
gsanderson	hi, is there any way to make a branch become the master ?	00:07
rudi_s	mugwump: Ah, I didn't even think of that, nice. Thank you.	00:07
	→ jelmer_ joined	00:07
mugwump	gsanderson: sure, git branch -M master; git push -f origin +master	00:08
	very destructive command, but what you asked for	00:08
rudi_s	Just curious, would git reset --hard branchname while on master also work?	00:08
mugwump	rudi_s: yes	00:08
	← andreaa left	00:08
rudi_s	Nice, thanks.	00:08
opcode	that gave me : ":040000 040000 8cb3c037029d7afb13e358f82ace7dd182e8e88e d58d400c77180cff2151f85f6d599e5e590efcb9 M app". How do I see the actual changes to the files?	00:09
	ah, git-diff will show me that	00:09
	← jelmer_ left	00:09
	→ agile joined	00:10
gsanderson	mugwump : thanks !	00:10
	→ Arrgh` joined	00:12
Arrgh`	Hey, I'm a git n00b and I have a hopefully simple question... I cloned a repos from github and now I want to switch my local mirror to one of the branches on github, not create my own new branch	00:13
	In theory 'git checkout <remote-branch-name>' should work but no dice	00:13
mugwump	you need a recent git for that	00:14
Arrgh`	OK, well I have 1.6.3.3 (came with Ubuntu 9.10), is that sufficiently recent? And if not, how do I do the equivalent, e.g. just clone the branch and ignore the rest?	00:14
mugwump	also you need to make sure that you are tracking the person with that remote	00:15
	see eg http://github.com/guides/pull-requests	00:15
	you can use git checkout -b newbranch -t remotename/newbranch	00:15
Arrgh`	Complains that the branch already exists	00:16
mugwump	well, it probably does then! :)	00:16
	inspect the situation with gitk --all	00:16
Arrgh`	I don't want to create a local branch, I just want to do a build from the branch that's on github	00:16
mugwump	just git checkout remotename/branch then	00:17
Arrgh`	I just recloned and did the previous command (checkout -b .. -t) and it worked	00:17
	So thanks :)	00:17
	→ mithro joined	00:18
gsanderson	ok, now I have another problem. I renamed a branch with 'git branch -M master' (and I really did want that branch to become master), but when I that repo its stuck on an old commit	00:19
	*when i clone that repo	00:19
mugwump	you need to push it	00:19
gsanderson	it isn't a local repo, its my main repo	00:20
	→ hipe joined	00:20
	← mjf left	00:20
neuro_damage	what's the appropriate hook file to use when wanting ot send emails of code pushes to master?	00:20
	or any branch?	00:21
rudi_s	neuro_damage: Hook file? Why not git send-email.	00:21
opcode	I keep getting these messages about whitespace: "warning: squelched 65 whitespace errors" and "/home/opcode/blah/.git/rebase-apply/patch:639: trailing whitespace."	00:21
	and when I svn rebase, my editor says the file has been changed, even though no changes have occurred. I think the whitespace is getting changed. how do I prevent this from happening?	00:22
gsanderson	neuro_damage : you can also use post-commit	00:23
	← jjuran left	00:23
	→ jjuran joined	00:23
	m-takagi_ → m-takagi	00:24
	→ enherit joined	00:26
	← justin-george left	00:26
	← gsanderson left	00:27
	→ pheaver joined	00:27
	← auslander left	00:28
	→ gzou2000_ joined	00:29
	gzou2000_ → gsanderson	00:29
	m-takagi → m-takagi_	00:30
gsanderson	sorry, internet bug. So, anyone know what my problem was ?	00:30
mugwump	you can use git update-ref refs/heads/master COMMITID	00:32
	and git symbolic-ref HEAD refs/heads/master to force master to be the default branch	00:32
	← patmaddox left	00:32
	← darwin_ left	00:33
gsanderson	still doesn't work	00:35
	← malumalu left	00:36
gsanderson	although when i cat .git/refs/heads/master its got the right commitid	00:37
	cloning is still stuck on an old commit	00:37
mugwump	try inspecting with git ls-remote URL	00:38
	how are you cloning anyway? http?	00:38
gsanderson	ssh	00:39
	woh, ls-remote shows the refs of the original project i branched off	00:40
	m-takagi_ → m-takagi	00:40
	← hyperair left	00:40
	← Swimming_Bird left	00:41
	m-takagi → m-takagi_	00:42
gsanderson	and it doesn't any of the branches that i created	00:42
	← Helios left	00:42
	← brosner left	00:42
	m-takagi_ → m-takagi	00:43
	→ plovs joined	00:44
	→ selam joined	00:45
selam	hi all	00:45
	← anjen left	00:46
	m-takagi → m-takagi_	00:46
	→ anjen joined	00:46
	→ Helios joined	00:46
	→ IceD^ joined	00:47
	← hobodave left	00:48
	→ frakturfreak_ joined	00:50
selam	i have a question abount branches, i have two branch one off them "master" one of them "cassandra" so cassandra was created two day ago its very diffrents from master so i remove some files from cassandra and pushed to remote server i working on two day and all works commited and pushed to remote branch, than after i change branch to master, i work and commit, now here is my problem: when i changed to cassandra branch from master i see some files owned by master	00:50
	but i dont see this files becouse i confused, so how can i solve this problem? if i clone cassandra branch to another directory ( under /tmp/" ) i dont see these files	00:50
	Arrgh` → Arr`Gone	00:51
gsanderson	what do you mean by 'owned by master' ?	00:53
selam	its must be in only master branch when i working on cassandra branch,	00:53
	← frakturfreak left	00:53
	→ Kelly_zhou joined	00:54
gsanderson	do you mean that there are files that exist in master but not in cassandra ?	00:54
selam	gsanderson: yes	00:54
gsanderson	when you commit, git does not automatically add new files, you have to add them with 'git add /path/to/file'	00:55
	changes made in master do not go into cassandra	00:56
davr	Yeah, it sounds like you may have forgotten to add the files before committing, so they are not on any branch, they are just untracked files hanging around in your working dir	00:57
selam	davr no all files tracking	00:57
	gsanderson: i know, i just dont see it when working :) model/users.py model.user.py one off them use cassandra one off them use sql and i confuse which one i use now :)	00:58
	frakturfreak_ → frakturfreak	00:58
selam	s#model.user.py#model/user.py#	00:58
gsanderson	lol	00:59
	→ brosner joined	00:59
Rawxor	haha	00:59
	nice	00:59
gsanderson	mv user.py users.py; git add users	00:59
	git add users.py	00:59
	or vice-versa	01:00
	← enherit left	01:01
	← smuf left	01:02
gsanderson	you can also update a file with the changes from the other branch with 'git checkout cassandra; git checkout master /path/to/file'	01:02
	← CaptTofu left	01:02
	← Sonderblade left	01:02
	→ Tama02 joined	01:02
gsanderson	so, anyone have any idea on how to fix my ref problem ?	01:03
	← Patrulka left	01:04
selam	i guess i can't tell my issue very well, i play for figure out this and i found	01:04
mugwump	it just sounds like you're getting confused about where the command is taking effect gsanderson	01:05
	← Tama01 left	01:05
mugwump	you need to ssh to the place you want it to work and run it there	01:05
	or, change it locally and push it	01:05
	either that or for some reason the clone url isn't where you're changing	01:05
	← anjen left	01:05
selam	these files not removed from cassandra branch this is reason why i see these files :)	01:06
	→ Patrulka joined	01:06
	← Hillshum left	01:08
	← magn3ts left	01:10
gsanderson	LOL, I just figured it out	01:11
	← Yuuhi left	01:11
gsanderson	I had a symlink pointing to the wrong directory	01:11
	and the directory had an old .git	01:11
	thanks anyway for your help mugwump ;)	01:12
	← flaguy48 left	01:13
mugwump	np gsanderson, thanks for answering questions too	01:13
	→ Swimming_Bird joined	01:13
	→ dreiss joined	01:13
	→ lresende joined	01:14
gsanderson	selam : you can remove files with 'git rm /path/to/file'	01:14
	→ hyperair joined	01:16
	← qrc1 left	01:17
selam	yes i do now	01:17
	ls	01:17
	yeah i gues is done :)	01:24
	thanks a lot	01:24
	→ hobodave joined	01:24
	← yev\|away left	01:24
	← spearce left	01:25
	← kpreid_ left	01:27
	← kukks left	01:28
	← gsanderson left	01:29
	→ _VVSiz_ joined	01:29
	→ jinzhu joined	01:31
	← VVSiz_ left	01:32
	← frakturfreak left	01:32
	← stepnem left	01:34
	← jrmuizel left	01:36
	→ kpreid_ joined	01:37
	→ stepnem joined	01:37
	← selam left	01:37
	← o_a left	01:39
	→ jrmuizel joined	01:42
	← Tommy[D] left	01:44
	← jrmuizel left	01:45
	→ o_a joined	01:45
	→ Tommy[D] joined	01:46
	→ shennyg joined	01:47
	→ jrmuizel joined	01:48
	→ jtaby_ joined	01:52
	← jtaby left	01:52
	← gaveen left	01:52
	← YuriQ left	01:53
	→ jtaby joined	01:54
	← jtaby_ left	01:54
	← novas0x2a left	01:55
	← jrmuizel left	01:56
	→ frakturfreak joined	01:56
	→ rudi_s_ joined	01:57
	← aresnick left	01:58
	→ magn3ts joined	01:58
	← joeconyers left	01:59
	→ joeconyers joined	02:00
	→ jrmuizel joined	02:01
	← rudi_s left	02:01
	→ mikelifeguard joined	02:03
mikelifeguard	Is there a way to create & send patches by email all in one go?	02:03
	← johnw left	02:04
bremner	man git-send-email	02:04
Gitbot	bremner: the git-send-email manpage can be found at http://git.or.cz/man/git-send-email	02:04
mikelifeguard	I did read that	02:05
bremner	ah, so what do you need?	02:06
	mikelifeguard: do you know how to use git-format-patch? git-send-email can take the same parameters.	02:06
	→ aresnick joined	02:07
mikelifeguard	Turns out "HEAD" isn't allowed (O.o) I guess	02:07
Morasique	well, there are no changes between HEAD and HEAD to create patches for :)	02:07
bremner	did you want to make a patch from uncommitted changes?	02:08
	← sh1mmer left	02:08
	← plovs left	02:09
mikelifeguard	no, I wanted the diff between HEAD^ and HEAD :)	02:10
	← hobodave left	02:10
wereHamster	mikelifeguard: HEAD is allowed, but just doesn't make sense (git format-patch HEAD means make a patch for each commit in the range HEAD..HEAD, which is an empty range)	02:10
	← stubbornmule left	02:10
	← tildeequals left	02:10
bremner	mikelifeguard: so git format-patch HEAD^ doesn't do what you want?	02:11
mikelifeguard	it does	02:11
	→ stubbornmule_ joined	02:11
	← parasti left	02:11
	stubbornmule_ → seanc	02:12
	seanc → seancarmody	02:12
	← Morasique left	02:12
	← seancarmody left	02:13
	→ stubbornmule joined	02:14
	→ Morasique joined	02:15
	← meatmanek_ left	02:18
	→ sh1mmer joined	02:18
	→ sh1m joined	02:19
	→ cytrinox_ joined	02:20
	→ kazoo joined	02:21
	→ hobodave joined	02:21
	→ zehrique joined	02:22
	→ magn3ts_ joined	02:22
	← sh1mmer left	02:23
	sh1m → sh1mmer	02:23
	← cytrinox left	02:24
	cytrinox_ → cytrinox	02:24
	→ a2ron_ joined	02:24
	← magn3ts_ left	02:24
	→ magn3ts_ joined	02:25
	← magn3ts left	02:25
	← a2ron left	02:26
	← hobodave left	02:26
	→ meatmanek joined	02:27
	→ novas0x2a joined	02:29
	← Patrulka left	02:29
	← PerlJam left	02:29
	← magn3ts_ left	02:30
	→ megazig joined	02:31
	→ Patrulka joined	02:31
megazig	when trying to setup gitosis with public access to a repo, I try "sudo -u git git-daemon --base-path=/media/old_home/git/repositories/ --export-all" and get "sudo: git-daemon: command not found"	02:32
	now I am under the assumption that git-daemon comes with git by default from the standard Ubuntu apt-get repos	02:32
wereHamster	git daemon	02:32
	→ altogether joined	02:33
mmattice	how do you get git svn to push branches up to a gitosis copy?	02:34
	← adamm left	02:35
	← mikelifeguard left	02:36
offby1	you don't --- instead, you use regular old git	02:37
	"git push name-of-the-gitosis-remote your-branch"	02:37
	note that it's not a great idea though	02:37
mmattice	why's that?	02:37
offby1	you'll be constantly rebasing that svn branch	02:37
	with "git svn rebase"	02:37
	RobertLaptop_ → RobertLaptop	02:37
mmattice	probably not.	02:38
	← Patrulka left	02:38
megazig	after changing git-daemon to git daemon and forwarding the proper port I now get "fatal: unable to allocate any listen sockets on host (null) port 9418"	02:38
	→ jklo joined	02:38
megazig	9418 is set to forward to this machine though	02:38
	→ fool__ joined	02:38
	← jklo left	02:38
	← fool__ left	02:38
	→ fool__ joined	02:38
offby1	mmattice: huh. _I_ always do ...	02:39
	→ jklo joined	02:39
	→ Patrulka joined	02:39
wereHamster	megazig: is there already a daemon listening?	02:40
mmattice	there's no development on that branch. it's a clone of a public project so we can hack stuff on top of it.	02:40
megazig	wereHamster: it would appear so...	02:41
	root 5673 0.0 0.0 108 4 ? Ss 21:14 0:00 runsv git-daemon	02:41
wereHamster	well, there you go	02:41
	← fool_ left	02:42
megazig	nto the same port though. couldn't I have 2 daemons running on separate listening ports?	02:42
wereHamster	sure you can, but why would you do that?	02:43
	← optim0 left	02:44
	→ drunknbass_work joined	02:44
	← fcoury_ left	02:44
megazig	wereHamster: because I'm a confused newb with several thousand lines of code that I'd like to share and am confused about getting public access setup. I have private access going well	02:45
	also, because I can't find the daemon to stop in /etc/init.d/	02:46
	→ patmaddox joined	02:49
	← a2ron_ left	02:49
	← aryounce left	02:49
	← eletuchy left	02:50
	→ anzenketh joined	02:50
	→ marj joined	02:57
mmattice	how do you specify the rev you want in a submodule?	02:57
	← madewokherd left	02:57
frogonwheels	mmattice: you go to the submodule directory, checkout the rev you want ..	02:58
	mmattice: then you go back to the main project and git add submoduledir (without a trailing /)	02:58
	mmattice: and commit it.	02:58
mmattice	submodule status in the / doesn't list that rev	02:58
	← drunknbass_work left	03:00
	← kazoo left	03:01
mmattice	submodule summary does though apparently	03:01
	← pheaver left	03:03
	← gigo left	03:05
	← sh1mmer left	03:07
scavalcante	Hello.. I'm trying to use git to deploy a website using post-receive I'm running git checkout -f	03:09
	I config the worktree, but I'm receiving /usr/bin/git-checkout cannot be used without a working tree.	03:09
wereHamster	did you follow a guide?	03:09
scavalcante	yes	03:10
	http://toroid.org/ams/git-website-howto	03:10
Arrowmaster	most guides on using git to update a website are horrible	03:10
	← carrus85 left	03:11
scavalcante	yes.. I know!	03:11
	← agile left	03:11
Arrowmaster	are you getting that error from the hook?	03:12
	← hyperair left	03:13
scavalcante	from the terminal git checkout	03:14
Arrowmaster	yeah thats why its erroring	03:14
scavalcante	how can I debug the hooks?	03:14
Arrowmaster	read where it talks about GIT_DIR in that guide	03:14
	← felipe left	03:16
	50UAAG9U7 → ilteris__	03:16
	← dsop left	03:16
	← khelll left	03:16
	→ dsop joined	03:16
	← jmcantrell left	03:18
Ilari	megazig: One useful thing would be full git-daemon command line. I don't think that process is correct (its just supervisory process to it...)	03:19
scavalcante	Arrowmaster: I'm using http to connect, should I try ssh?	03:23
	← alester left	03:28
	← patmaddox left	03:35
Ilari	megazig: There are for instance some parameter combinations that allow git daemon to start but it will not work.	03:37
	→ hyperair joined	03:38
	→ WALoeIII joined	03:42
	← Theravadan left	03:42
	→ pheaver joined	03:42
mmattice	if I commit, then rebase -i to move that commit, should it move off the HEAD?	03:43
	← opcode left	03:44
	← hipe left	03:46
	offby1 stares blankly	03:46
offby1	what does "move" mean in this context?	03:46
	← stepnem left	03:46
	→ alester joined	03:52
	← shennyg left	03:53
davr	you're ALWAYS on the HEAD. HEAD is just a pointer to your currently checked out commit.	03:54
	← meatmanek left	03:55
	← tetha left	03:57
	→ tetha joined	03:59
	← Twisted_Mentat left	03:59
	← anzenketh left	04:01
	← aresnick left	04:02
	→ jony joined	04:04
traviscline	does anyone know what the right steps toward preventing pushes that would create new commits with zero parent commits?	04:07
bob2	doesn't that fail without -f anyway?	04:08
traviscline	yes	04:08
wereHamster	pre-receive hook, check if <old> is 0000..	04:08
traviscline	wereHamster: cool, that will be true if it's a multi-commit push?	04:09
wereHamster	hm, that would detect new branches, not commits with no parents.	04:09
traviscline	hrm	04:10
	← harinath left	04:11
	→ meatmanek joined	04:12
	→ aryounce joined	04:13
	← scavalcante left	04:16
	← meatmanek left	04:18
	← altogether left	04:18
	← pheaver left	04:21
	→ fool_ joined	04:21
	← fool_ left	04:21
	→ fool_ joined	04:21
	← hyperair left	04:22
	→ pheaver joined	04:22
	→ altogether joined	04:24
	← fool__ left	04:24
	← zehrique left	04:25
	→ harinath joined	04:26
	← pheaver left	04:26
	→ pheaver joined	04:27
	→ shennyg joined	04:30
	← harinath left	04:30
	← joeconyers left	04:30
	← AAA_awright left	04:33
	→ edwefwf joined	04:36
	→ AAA_awright joined	04:37
	→ hyperair joined	04:37
	← scarabx left	04:37
	← alezandro left	04:39
	→ drunknbass_work joined	04:40
	→ hipe joined	04:41
	← edwefwf left	04:42
	← jtaby left	04:42
sitaram	bob2: pushing a new root commit does not need -f	04:44
	→ harinath joined	04:45
	→ meatmanek joined	04:45
	← meatmanek left	04:45
wereHamster	morning sitaram :)	04:46
sitaram	morning wereHamster!	04:47
	→ munichlinux joined	04:47
	→ zjason joined	04:47
traviscline	sitaram: thanks for gitolite, moved to it recently, hot.	04:47
	→ MetaCosm joined	04:47
sitaram	traviscline: you're welcome! and thanks for the compliment	04:48
traviscline	sort of wish I could separate rewinding from deletion	04:48
	but no biggie	04:48
sitaram	traviscline: you know, I think there's no actual way to prevent what you asked. Even with gitolite :)	04:48
	oh wait rewind from deletion? hmmm	04:48
bob2	sitaram: ah, didn't know that, thanks	04:48
sitaram	I thought you were asking about "prevent new root commit"	04:48
	← Cap_J_L_Picard left	04:48
traviscline	sitaram: I think with a custom post-receive you could check that the new commits are ancestors of previous ones	04:48
	sitaram: i was earlier, just a new comment	04:49
	dealing with a large team with client-facing refs -- not everyone is a git pro	04:49
sitaram	traviscline: ok, two separate topics then	04:49
megazig	wereHamster: thanks for the help earlier. I didn't get it working, but I appreciate the knowledge	04:49
traviscline	but we're just going to have the more experienced devs deal with branch deletions	04:49
	yes	04:49
	→ psankar joined	04:49
traviscline	sitaram: i've got to run but if you have thoughts on either i'm all ears, thanks again. back later.	04:50
sitaram	no problem; is your session on and can you check backlog later or should we talk only afte ryou come back?	04:50
	after you*	04:50
	→ Cap_J_L_Picard joined	04:50
	← RobertMeta left	04:51
sitaram	traviscline: prevent new root commits: what you get is an oldsha and a new sha. If the old is all 0-s it may be a new root commit, which can only be known, really, by doing a merge-base with all currently existing heads. Doable, I guess, so I was wrong to say it can't be done	04:52
Mage	does gitosis do private repos yet	04:52
	or am i retarded	04:52
sitaram	.oO(distinguishing deletes from rewinds... hmm... interesting. Can be done. Question is, how to do it with minimum new code and no disruption to existing syntax/semantics!)	04:53
Mage	ok i am thinking about something else	04:53
sitaram	Mage: you're thinking about github perhaps. Gitosis is just software you install yourself, not a service	04:53
Mage	sitaram: not github. there's another one that's /like/ github. i thought it was called gitosis for a second there :p	04:54
	trying to get a good deal for a private repo ;p	04:54
sitaram	Mage: gitorius, indefero, girocco, are what come to mind.	04:56
	not sure about private repos on any of them	04:56
megazig	sitaram: any preference among them ?	04:56
	→ roop joined	04:56
sitaram	github is the only one I have used. I have an indefero account I use as a backup but don't really use as in use their website, login, etc -- just the ssh push	04:57
	← hyperair left	04:57
Mage	sitaram: danke	04:57
sitaram	github does have private repos, for a fee. That is their business model. And I know people sometimes rant about it but I like github a lot. I only wish their issues system would simply tie into email like say bugs.debian.org does, so you don't have to login every time someone posts an issue on your project :)	04:57
traviscline	sitaram: Backlog. And mobile client :) I thought the base git config let you separate deletions from nonff pushes	04:57
	→ jony1 joined	04:58
sitaram	traviscline: man git-config and look for receive.denyDeletes	04:58
Gitbot	traviscline: the git-config manpage can be found at http://git.or.cz/man/git-config	04:58
traviscline	Some dude said earlier that the old 0s may just be a new ref, not an unconnected history line.	04:58
	→ hyperair joined	04:59
	← jinzhu left	04:59
	→ jinzhu joined	04:59
sitaram	traviscline: if you wanted to prevent all branch deletes but allow rewinds, you could give RW+ perms and then set that config on that repo	04:59
traviscline	sitaram: I know, talking about a gitolite context though	04:59
	I want the reverse	04:59
	Disallow force pushes but allow deletes	04:59
sitaram	traviscline: yes; wereHamster did. And he was right. The old/new only refer to "old or new value of this symbolic ref" so 0x40 need not refer to a root commit	04:59
	aah you want the reverse	05:00
	← jony left	05:00
traviscline	( I realize a delete and a repush is eqiuv. )	05:00
	← AAA_awright left	05:00
sitaram	well yes that is the reason I clubbed the two as conceptually the same level of danger from innocent newbies :)	05:00
	→ AAA_awright joined	05:01
sitaram	let me think about this over today... I dont jump that quickly these days :) Gitolite is already "gitoheavy" from all the features I added in the last few months :(	05:01
	← jonmasters left	05:01
sitaram	I may end up saying no first, but some folks here will say that don't mean nothing ;-) and I'll come up with code after a while!	05:02
	→ sde_antrix1 joined	05:02
traviscline	I can alway patch it myself (though im a python dude)	05:02
	Anyhow. Thanks for being responsive and thanks again for gitoheavy ;)	05:03
	→ alezandro joined	05:03
	→ yevlempy joined	05:03
	→ jtaby joined	05:05
	yevlempy → yev\|away	05:06
sitaram	traviscline: would you mind telling me why you want to allow deleting a branch but not rewind? I'm trying to understand the use case better, that's all...	05:06
	→ fool__ joined	05:07
	← fool__ left	05:07
	→ fool__ joined	05:07
	← TheMoonMaster left	05:08
sitaram	the only branches that get deleted often enough that you want to allow any developer to delete, are "personal" branches, for which case even rewind should not really matter... (just thinking aloud)	05:08
better_call_saul	i converted an old cvs repo to git. i cloned it, and when i do a git branch -a i see lots of branches remotes/origin/foo , how doi get a local version of foo?	05:09
	← fool_ left	05:10
	← yev\|away left	05:12
	← o_a left	05:13
	→ pano_ joined	05:13
	→ tasslehoff joined	05:13
	offby1 idly wonders if better_call_saul enjoys a TV series known as "Breaking Bad"	05:15
offby1	git checkout --track remotes/origin/foo, I think	05:15
	→ jonmasters joined	05:15
mugwump	offby1: in newer gits you can use 'git checkout foo'	05:16
offby1	mugwump: I knew there was some simpler way	05:16
	← Swimming_Bird left	05:17
	← pano_ left	05:18
	→ brettgoulder joined	05:18
	→ azol joined	05:19
brettgoulder	I have a weird error when I try to submodule init and update http://gist.github.com/348785	05:19
	it crashes with "fatal: reference is not a tree" then a sha hash	05:20
	what should i do	05:20
offby1	wow	05:20
	never seen that :-\|	05:20
frogonwheels	brettgoulder: I've seen that.	05:20
	← shennyg left	05:21
frogonwheels	brettgoulder: make sure the submodule has been pushed to with the correct version	05:21
brettgoulder	it's weird	05:21
frogonwheels	brettgoulder: ie. whoever updated the submodule reference in the main project needs to make sure they pushed the corresponding changes to the (submodule) repository	05:21
brettgoulder	frogonwheels: what do you mean? like a correct version of git?	05:22
frogonwheels	brettgoulder: .. the main projects refereces a SHA1 in the submodule corresponding to a commit, yeah?	05:22
brettgoulder	ya	05:22
	← Mage left	05:22
frogonwheels	brettgoulder: well that SHA1 hasn't been pushed to the submodules repository.. (hopefully it hasn't been deleted)	05:23
brettgoulder	and if it has been deleted?	05:23
	frogonwheels shrugs.	05:24
	← jrmuizel left	05:24
brettgoulder	can i remove a submodule and readd it?	05:24
frogonwheels	brettgoulder: you can check-out an exisiting reference in the submodule directory..	05:24
	brettgoulder: then git add submodule and git commit ....	05:25
	(being careful not to do git add submodule/ - which is bad)	05:25
	brettgoulder: more details on what happened would aid in diagnosis..	05:26
	← jtaby left	05:26
	→ PointMan joined	05:26
frogonwheels	brettgoulder: who has control of the submodule repository? who committed that submodule reference?	05:26
	← altogether left	05:26
	→ rseifert joined	05:27
frogonwheels	brettgoulder: could the person who committed the submodule reference to the main repository have forgotten to push the changes in the submodule ?	05:27
better_call_saul	offby1: yes, my nick is in Mr. Goodman's honor	05:28
	i thought i tried both those optoins, i'll try again	05:28
frogonwheels	I'm not sure I'm getting the message across.. shrug	05:29
	← meanburrito920 left	05:29
	← Bass10 left	05:29
	→ altogether joined	05:30
better_call_saul	ok it worked with the --track option. it's a pretty new install, what version does just git checkout foo work in?	05:30
	→ yn joined	05:31
	← alezandro left	05:32
	← crankharder left	05:33
	← rseifert left	05:33
	← Heimidal left	05:34
	→ Heimidal joined	05:34
	← r11t_ left	05:35
	← steveire left	05:36
	→ steveire joined	05:37
	← steveire left	05:37
	→ steveire joined	05:37
	← WALoeIII left	05:38
	→ stepnem joined	05:38
	→ ankit9 joined	05:42
	→ khmarbaise joined	05:44
	→ epic_ joined	05:46
	← khmarbaise left	05:47
	← aryounce left	05:47
	← iaj left	05:48
	← epic_ left	05:49
	→ alezandro joined	05:51
	← novas0x2a left	05:52
	← rburton- left	05:54
	→ nataraj joined	05:54
	→ novas0x2a joined	05:54
	← phishy left	05:59
	→ rudi_s joined	05:59
	← rudi_s_ left	05:59
	← pheaver left	06:02
	← novas0x2a left	06:02
	→ maxthoursie joined	06:03
	→ pheaver joined	06:04
	→ sjain joined	06:04
	← rocketeerbkw left	06:04
	→ yevlempy joined	06:04
	→ phishy joined	06:05
	→ drizzd joined	06:07
	← azol left	06:09
	← jklo left	06:09
	→ Weiss joined	06:09
	→ _iron joined	06:09
	→ rocketeerbkw joined	06:09
	← drunknbass_work left	06:15
	→ fool_ joined	06:16
	← fool_ left	06:16
	→ fool_ joined	06:16
	← fool__ left	06:17
brettgoulder	frogonwheels: i think the person who committed did something screwy by deleting the submodule, then cloned the repo in the directory where we keep our submodules	06:17
	frogonwheels: but i managed to fix it up	06:17
	← brettgoulder left	06:18
	← jony1 left	06:20
	→ xiong joined	06:22
	→ skoop joined	06:24
	← Pupeno left	06:25
	← drizzd left	06:26
	→ drizzd joined	06:27
	← yn left	06:27
	← alester left	06:28
	← deanc left	06:28
	→ eletuchy joined	06:29
	← munichlinux left	06:32
	→ tvw joined	06:32
	→ thiago_home joined	06:32
	→ ph^ joined	06:33
	→ epic_ joined	06:33
	→ Paraselene_ joined	06:34
	← nataraj left	06:34
	→ jony joined	06:34
	← Scotto_ left	06:35
	← hyperair left	06:35
	← jony left	06:36
	→ jony joined	06:36
	→ sh1mmer joined	06:36
	→ tbf joined	06:37
	→ steffkes joined	06:37
	→ iaj joined	06:39
	→ mtkd joined	06:39
	← marc_kdab left	06:39
	→ bentob0x joined	06:42
	← ciaranm left	06:45
	→ ciaranm joined	06:45
	← Ingmar left	06:45
	→ Ingmar joined	06:45
	← AkumaStreak left	06:49
	→ WALoeIII joined	06:49
	→ dj_tjerk joined	06:51
	→ tomekw joined	06:53
	→ Twisted_Mentat joined	06:54
	← WALoeIII left	06:54
	→ [Po]lentino joined	06:57
	→ hedgehog83 joined	06:58
	→ d0ugal joined	06:58
	← webchick left	06:59
	→ poseid joined	07:01
	→ Pupeno joined	07:02
	← sde_antrix1 left	07:03
	→ brettgoulder joined	07:03
	→ novas0x2a joined	07:03
	← engla left	07:04
	← xiong left	07:04
	← Tommy[D] left	07:05
	→ priidu_ joined	07:05
	→ kar8nga joined	07:05
	→ svbg joined	07:07
	→ Tommy[D] joined	07:07
	← alezandro left	07:07
	← poseid left	07:07
	← nevyn left	07:08
	→ alezandro joined	07:08
	→ petrux joined	07:08
	→ nevyn joined	07:09
	← svbg left	07:09
	← thiago_home left	07:10
	← merlin83 left	07:13
	← hedgehog83 left	07:16
	→ felipe joined	07:17
	→ patmaddox joined	07:18
	← altogether left	07:18
	→ chittoor joined	07:18
	→ eighty4 joined	07:19
	→ gebi joined	07:19
	← bentob0x left	07:21
eighty4	I've been googling for this but not really sure what to search for. Is there any nice solution that would let me auto clone all existing git repos on a specific server and then keep them up to date, on a second computer. I.e. keeping a backup of all git repos.	07:21
	→ anjen joined	07:22
sitaram	rsync may be all you need then...	07:22
	← dacuca left	07:23
	→ giallu joined	07:23
	→ bentob0x joined	07:24
	→ Theravadan joined	07:25
	→ munichlinux joined	07:25
	→ cedric_ joined	07:28
eighty4	sitaram: yeah, I thought about that but not really what I want	07:29
	preferably something git based, like having one git repo that keeps track of what repos exist and then runs git clone/git pull	07:29
	→ marc_kdab joined	07:31
	← cedricv left	07:31
frogonwheels	eighty4: you could have a git repo with a list of repos to be copied...	07:32
	eighty4: and a cron job that pulls that repo, and then clones/fetches from that list.	07:33
eighty4	frogonwheels: yeah, and then just use a cronjob that... exactly	07:33
	frogonwheels: but how would the list be keept updated on the "server" side?	07:33
frogonwheels	eighty4: you just push changes up to it.	07:34
eighty4	but then I'd have to keep that in mind everytime I create a new repo :)	07:34
	← roop left	07:34
	→ roop joined	07:34
eighty4	seemed like seomthing someone might have done already, thats why I'm asking. (to save me from writing it all) :P	07:35
frogonwheels	eighty4: See if you can add hooks to your server-side stuff (what are you using on the server?)	07:35
	← doener left	07:36
frogonwheels	a custom list could be cool 'cause you could be pulling from different servers.	07:36
	← dreiss left	07:37
gebi	eighty4: if you want to mirror gitweb there is '?a=project_index'	07:37
	gives you a list of all repos in the gitweb site	07:37
	← cedric_ left	07:37
	→ ludde joined	07:38
eighty4	frogonwheels: I can definitely add hooks, the "server" is just a central point where me and some other guys will push/pull to to keep it all simple. It's just an Ubuntu install	07:38
	→ doener joined	07:38
ludde	can i somehow run git filter-branch and keep all branch heads in sync (i.e. update the branch heads to the proper value)?	07:38
	→ andoma joined	07:38
	→ nevyn^ joined	07:39
eighty4	gebi: I not really following you here. Could you explain?	07:39
	← tvw left	07:39
eighty4	gebi: where would I run that?	07:39
	→ mastro joined	07:40
	→ voxpelli joined	07:41
frogonwheels	eighty4: yeah- but presumably you'll use gitosis / gitolite or something else to provide that part of it.	07:41
gebi	eighty4: http://git.kernel.org/?a=project_index	07:41
eighty4	gebi: yeah, but wouldn't that require me installing gitweb as well?	07:42
gebi	sure	07:42
	eighty4 is very much a n00b when it comes to git	07:42
gebi	you really want some webthing to look at the repos too	07:42
	though problematic with private repos	07:43
frogonwheels	eighty4: git by itself doesn't behave like a 'server' and do security &c. you need something else to do that.	07:43
	← eletuchy left	07:44
	← voxpelli left	07:45
	→ eletuchy joined	07:45
eighty4	frogonwheels: yeah, that part I know. But not sure I need gitosis/gitolite at the moment. Just a couple of users that all have acccess to git using ssh	07:45
	→ steffkes- joined	07:45
	→ whitman joined	07:47
	→ nick3 joined	07:47
	← nick3 left	07:48
	← gebi left	07:48
	→ hyperair joined	07:48
	← steffkes left	07:48
eighty4	but i'd like it if the users didn't need to clone and remember to pull	07:50
	← steffkes- left	07:50
frogonwheels	eighty4: remember that pull is a fetch+merge.	07:51
	eighty4: the merge will be the gotcha - don't want that to be done auto-magically	07:51
	← yevlempy left	07:54
	← dj_tjerk left	07:55
	→ brrt joined	07:56
	flity_ → flity	07:57
brrt	ping mugwump	07:57
	→ nicoulaj joined	08:01
	← mtkd left	08:05
	→ shruggar joined	08:06
	→ sde_antrix1 joined	08:09
	→ ixti joined	08:09
	→ mtkd joined	08:10
	→ Textmode joined	08:11
maxthoursie	when I change the url in .gitmodules, and run git submodule sync, .git/config isn't updated	08:12
	am I missing something?	08:12
	← brrt left	08:12
Textmode	morning all	08:13
	→ lucsky joined	08:14
	→ ArdRigh joined	08:17
eighty4	frogonwheels: thanks for the input. I'll play with it a bit and see what I come up with	08:19
frogonwheels	eighty4: np. gl.	08:19
	→ poseid joined	08:20
	← hyperair left	08:20
	→ cedricv joined	08:20
	→ steffkes joined	08:26
	→ byang joined	08:27
byang	hi	08:27
Ilari	hi	08:28
	→ everzet joined	08:29
	← Patrulka left	08:29
	→ hedgehog83 joined	08:29
	→ hyperair joined	08:30
Ilari	maxthoursie: There's URL in .git/config as well.	08:30
	→ cilly joined	08:31
	← Twisted_Mentat left	08:31
	→ Patrulka joined	08:32
Ilari	maxthoursie: And IIRC, that URL is the one that's actually used.	08:32
maxthoursie	Ilari: yes, it is, but I thought git submodule sync was suppoesed to update it	08:32
Ilari	maxthoursie: AFAIK, it isn't.	08:34
maxthoursie	sync	08:34
	Synchronizes submodules' remote URL configuration setting to the value specified in .gitmodules. This is useful when submodule URLs change upstream and you need to update your local repositories	08:34
	accordingly.	08:34
	Ilari: Am I making the wrong interpretation of the man page, or does it not work as intended?	08:35
Ilari	maxthoursie: remove "URL configuration setting"... The way I read it, it is different setting...	08:36
	← ankit9 left	08:36
	← novas0x2a left	08:36
	ambuli → jelari	08:36
	jelari → ambuli	08:36
	→ hedgehog831 joined	08:36
	→ novas0x2a joined	08:37
maxthoursie	Ilari: hm, perhaps, dunno which setting that should refer to though	08:38
Ilari	maxthoursie: There's remote URL in submodule itself and there is submodule URL on parent project config.	08:38
	maxthoursie: Plus URL in parent .gitmodules	08:38
	← hedgehog83 left	08:39
maxthoursie	Ilari: ah, sorry, now I see	08:39
	Ilari: the url in submodule/.git/config was indeed updated	08:39
	but whats the url in <parent>/.git/config used for then?	08:40
	← sjain left	08:40
Ilari	maxthoursie: At least for cloning the submodule...	08:40
maxthoursie	Ilari: ofcourse, but once cloned	08:41
	→ voxpelli joined	08:41
	→ project2501a joined	08:43
doener	maxthoursie: "update" seems to lookup submodule.<name>.url to check whether the submodule has been initialized	08:44
	← voxpelli left	08:44
maxthoursie	doener: so it needs to be updated to then?	08:45
	so when I change the url of a submodule I have to update .gitmodules, then I have to tell my coworkers to remove the submodule section in their config, rerun git submodule init and then run git submodule sync.	08:46
	Or is there a way to simplify that?	08:46
doener	maxthoursie: the value isn't actually used there. It's checked checked to be non-empty	08:47
	← matthieu left	08:48
	← whitman left	08:48
	← byang left	08:49
	← ArdRigh left	08:49
maxthoursie	doener: ah, ok. So if my coworkers already have the submodule with the old url, a git submodule sync should suffice, and they can ignore the old value in theire .git/config ?	08:49
doener	I guess so	08:50
	doener only skimmed the git-submodule source	08:50
	→ byang joined	08:50
byang	Generally, blame -M is actullay detect same code but not code movement.	08:51
	→ froschi joined	08:51
maxthoursie	doener, Ilari: Thanks for your help	08:52
	← infogulch left	08:53
	→ simey00irc joined	08:54
	← simey00irc left	08:55
	→ simey00irc joined	08:56
	← simey00irc left	08:57
bentob0x	what's the best way to keep tracks of notes and things to watch depending on the various patches and progressions of a project?	08:58
shruggar	a bug tracker?	08:59
bentob0x	something easy to use along Git	08:59
	I mean that doesn't require me to learn another software to start working on describing the things to do	08:59
Rawxor	a text editor	09:00
shruggar	there is no agreed-on standard or even set of conventions. Bug tracking is an unsolved problem, not just in git, but in general. There are several projects which are designed with git in mind, but last I looked they were not anywhere near usable	09:00
Rawxor	Bugzilla is easy enough to get going	09:00
bentob0x	unless there is an easy way to sync up a bug tracker and git	09:00
bremner	I don't know it well, but there is ticgit	09:01
bentob0x	can you sync up bugzilla and git easily?	09:01
shruggar	what do you mean by "sync up"?	09:01
	what interface are you hoping for?	09:01
bentob0x	every time a new patch is entered in git, bugzilla would 'read' it and know there is a new patch and then updates itself	09:02
	interface is irrelevant, console will do just fine	09:02
shruggar	I think there's some post-commit hooks in contrib that will do that. At least, I assume there are, since they're so easy to write. Alternatively, you could write one	09:03
bentob0x	just a link between patch #12345678ABCDEF and a note like: "make sure all CSS declaration #divTest is removed everywhere => run a grep"	09:03
	for instance	09:03
bremner	err, use decent commit messages?	09:03
charon	bentob0x: maybe look at git-notes	09:03
	but yeah, that should go into the commit message	09:03
bentob0x	git-notes it is, thx charon :D	09:04
	← byang left	09:04
	← munichlinux left	09:04
bentob0x	yes the commit messages are already good to undersand what is in the patch	09:04
	→ ankit9 joined	09:04
	→ byang joined	09:04
	→ bluenovember joined	09:04
bentob0x	but sometimes, you want to add notes for the few 'tricky' patches	09:04
shruggar	add notes like what?	09:05
	I just don't understand why they would need to be outside of the commit message	09:05
bentob0x	like: "when merging this patch, make sure all client-specific templates are all set with the new div naming convention" for instance	09:06
shruggar	and, as was mentioned, if you really need them to be, git-notes does that	09:06
bentob0x	yes, that's what I understood, going to read and most likely use it :)	09:06
	→ doobry joined	09:07
bentob0x	thx shruggar, bremner, charon	09:07
	← pheaver left	09:08
	← Heimidal left	09:10
	→ level09 joined	09:11
level09	any 1 got a mac here :D	09:12
	→ GarethAdams\|Work joined	09:12
sitaram	eighty4: you still around?	09:12
	→ pheaver joined	09:14
	→ curtana joined	09:15
	→ wolog joined	09:15
	→ munichlinux joined	09:15
	→ Azeroth-IsWorkin joined	09:15
	→ dytrivedi joined	09:16
	← _VVSiz_ left	09:18
	← maattd left	09:19
	← Azeroth-Working left	09:19
pom	I merge as branch into master by mistake. Can I remove theat merge-commit?	09:19
	← Patrulka left	09:19
	← marj left	09:20
avtobiff	pom, just git checkout HEAD^ or git reset --hard HEAD^	09:20
	just rollback as any other commit	09:20
pom	Oh, I have only used revoke for that before. Nice new trick. Thanks.	09:21
	← novas0x2a left	09:22
	← IceD^ left	09:22
	→ Patrulka joined	09:22
	→ jas4711 joined	09:23
	→ novas0x2a joined	09:23
	→ dmgr joined	09:24
	→ zjason` joined	09:27
	← wolog left	09:28
	← zjason left	09:31
	→ wolog joined	09:31
	→ psoo joined	09:31
	→ trivol joined	09:32
	→ darwin_ joined	09:32
	← kar8nga left	09:34
	← unreal left	09:35
	← frogonwheels left	09:38
	← psoo left	09:38
	→ unreal joined	09:38
	→ struberg joined	09:38
	← hyperair left	09:39
	← level09 left	09:39
	→ marj joined	09:39
	→ anno^da joined	09:40
	→ groove joined	09:40
	← anno^da left	09:40
	→ Adaptee joined	09:40
	→ anno^da joined	09:41
groove	hi, i'm trying to run a git->svn conversion (don't ask) and during the "git rebase" step i get an error "fatal: unable to write file foo mode 120000"	09:42
charon	avtobiff, pom: careful there, 'git checkout HEAD^' does not remove the merge from the branch, it merely moves you to a detached HEAD (look it up in the FAQ if you don't know the term!) that doesn't have the merge yet.	09:42
	→ galderz joined	09:42
groove	i've googled it but what i've read doesn't make sense to me	09:42
	← dmgr left	09:42
	→ frogonwheels joined	09:42
eighty4	sitaram: yeah	09:44
sitaram	eighty4: seems like all you need is (1) a script to trawl a directory tree on the server looking for bare repos (2) and for each of them run a "git clone --mirror" the first time and a "git fetch" (or perhaps a "git fetch --prune") on an ongoing basis	09:45
	eighty4: anyway, just wanted to say the git specific part of what you want is likely to be those 2 commands; the rest has to be done outside git I'm afraid	09:46
	→ shame joined	09:46
	→ dmgr joined	09:48
	→ amanda9 joined	09:48
amanda9	http://www.mdhjakten.se/dela/?id=dti2d6s	09:48
	hard stuff	09:48
	← trivol left	09:50
	→ d0k joined	09:54
	→ hyperair joined	09:56
	→ psoo joined	09:58
	→ trivol joined	09:58
	← Adaptee left	09:59
	← amanda9 left	09:59
	← Patrulka left	09:59
	→ fool__ joined	09:59
	← fool__ left	09:59
	→ fool__ joined	09:59
	→ Patrulka joined	10:01
	← brettgoulder left	10:02
	→ zjason`` joined	10:02
	← fool_ left	10:02
	→ aziz joined	10:02
	← hyperair left	10:05
	← sven^ left	10:05
	← zjason` left	10:05
	[Po]lentino → [Po]lentino\|away	10:07
avtobiff	charon, thanks	10:09
	← patmaddox left	10:12
	→ cuco joined	10:13
cuco	hi all, i have this git project i started on a windows machine	10:13
	← hedgehog831 left	10:13
cuco	how can i clone it to linux? i can't use git clost ssh/http(s)... none of them are available on the windows machine	10:14
	→ Azeroth-Working joined	10:14
curtana	cuco, you could share the directory containing the repoistory via windows file sharing, then mount it on your linux system, and clone from the filesystem	10:16
cuco	curtana: how lame is it to just copy the directory to a usb-disk and then... ?	10:17
curtana	that would also be fine	10:17
cuco	really? no harm will be done?	10:17
curtana	probably not. worth a go anyway ;)	10:17
	if you intend to do much development on both systems then you are best off creating a shared, bare repository and cloning from it on both machines	10:17
	← ludde left	10:17
	← Azeroth-IsWorkin left	10:18
bremner	cuco: I seem to recall some problems on non-case sensitive file systems. But Windows users could tell you more about that.	10:19
	← Kelly_zhou left	10:19
cuco	bremner: when copying from linux to windows, i am doing the other way. I am comint into the light, not running away from it.	10:19
	← poseid left	10:20
curtana	cuco, if you are really concerned, make a .tar of the directory and copy _that_ do the usb disk	10:20
	or format a partition on the disk as ext2 and install ext2fsd so you can write to it from windows	10:20
	or if there's an ssh server on the linux system you can add a remote repository and push to it	10:20
	← darwin_ left	10:21
	→ tatsuya joined	10:22
	→ mithro_ joined	10:22
cuco	so basically, pull and push do the same but in reverse?	10:22
curtana	hm	10:23
	not quite. but almost.	10:23
	pull = fetch followed by merge. push is like the opposite of fetch... maybe.	10:23
	that's probably a bad explanation	10:23
	← mithro left	10:25
	→ parasti joined	10:25
	→ ntoll joined	10:26
eighty4	sitaram: thanks for the input :)	10:26
sitaram	you're welcome!	10:27
bremner	cuco: fetch and push are closer to opposite	10:30
	→ neen joined	10:31
eighty4	one more question do you actually have to stand in the git repo to be able to do a git fetch? I cant find anything about specifying a path to the git repo you want to work on	10:31
	← project2501a left	10:32
neen	hi guys, i know git doesn't support keyword expansion. i'm looking for some kind of 3rd party (preferrably commandline) script that allows me to do it	10:33
	i dont need a pre-commit hook or anything like that	10:33
	← ntoll left	10:34
	← munichlinux left	10:34
	← struberg left	10:34
	→ mattions joined	10:34
	→ Yuuhi joined	10:35
	← mtkd left	10:35
	← neen left	10:35
	→ neen joined	10:35
	→ tobiassjosten joined	10:36
mattions	anyone using the python binding for git?	10:36
bremner	which one?	10:37
	→ mithro__ joined	10:39
	→ khelll joined	10:39
	→ xiong joined	10:41
	← mithro_ left	10:42
	→ naeu joined	10:43
	→ mtkd joined	10:43
	← mtkd left	10:43
	→ o_a joined	10:44
sitaram	eighty4: your PWD should be the destination git repo of course; the source one can be specified by URL (any of the URLs syntaxes you see in man git-clone)	10:44
Gitbot	eighty4: the git-clone manpage can be found at http://git.or.cz/man/git-clone	10:44
	doener mumbles something about --git-dir and GIT_DIR	10:45
	← xiong left	10:45
	→ xiong joined	10:45
eighty4	sitaram: I.e. no git fetch --repo /path/to/repo ? Or am I really stupid here	10:48
	mithro__ → mithro	10:48
	← ericlee left	10:48
wereHamster	eighty4: what doener said...	10:48
	← cuco left	10:48
	→ ericlee joined	10:48
doener	see git(1)... but in practice, I'd most likely go with "cd /path/to/repo; git fetch" instead...	10:49
mattions	bremner, http://gitorious.org/git-python	10:50
	there are other around?	10:50
eighty4	doener: right.	10:50
	→ R_Rios joined	10:50
	← o_a left	10:50
bremner	mattions: there is also dulwich, which I guess is more a reimplimentation than bindings. I have to confess I don't know much more than that they exist.	10:51
	← mithro left	10:52
sitaram	eighty4: git fetch URL_TO_REPO master:master # but this doesn't take "--all" argument	10:52
mattions	bremner, any hint how to choose between the two?	10:53
	← jinzhu left	10:53
	← mastro left	10:54
bremner	mattions: well, I guess if the bindings do what you want, they are less likely cause corruption since the underlying code is widely used. OTOH, some google project is using dulwich, so it could be worth at least checking out.	10:55
	← Theravadan left	10:56
	→ ntoll joined	10:56
	→ bourbon joined	10:58
	→ malumalu joined	10:59
mattions	bremner, thanks	11:00
	← neen left	11:01
	← disappearedng__ left	11:03
	← jds left	11:03
	← disappearedng left	11:03
	← parasti left	11:04
	→ ehabkost joined	11:04
	→ parasti joined	11:05
	→ bx2 joined	11:05
	← chittoor left	11:08
	← parasti left	11:08
	→ hedgehog83 joined	11:08
	→ parasti joined	11:08
	← JEEB left	11:09
	→ disappearedng joined	11:09
	← jas4711 left	11:10
	← psoo left	11:10
	→ mastro joined	11:11
	→ jas4711 joined	11:13
	← xiong left	11:13
	← pheaver left	11:13
	→ bdiego joined	11:13
	← teukka left	11:13
	→ teukka joined	11:13
	→ JEEB joined	11:13
	→ xiong joined	11:13
	← dytrivedi left	11:14
	→ dajero joined	11:14
	← JEEB left	11:16
	← teukka left	11:17
xiong	test?	11:17
	← xiong left	11:17
	→ teukka joined	11:17
	→ disappearedng__ joined	11:17
	→ mithro joined	11:17
eighty4	sitaram: you suggested I should use "git clone --mirror" if a repo doesn't exist. That does a "--bare" and that would result in people not being able to use the checked out files, right?	11:18
sitaram	eighty4: all server repos should be bare so what you're mirroring from should also be bare... repos with working directories are always considered personal	11:19
	← priidu_ left	11:19
sitaram	unless I misunderstood your original request	11:19
	← hedgehog83 left	11:19
sitaram	eighty4: "clone all existing git repos on a specific server" <-- to me that means bare repos already :)	11:20
eighty4	sitaram: yeah, the servers repos are bare. But I'd like those I clone to be accessible for the users	11:20
	← PointMan left	11:20
sitaram	eighty4: accessible via clone or fetch, sure...	11:20
eighty4	i.e. the user getting the "git clone" repos should be able to browse the files.	11:20
	But from what I understand doing git clone --mirror will create a bare checkout?	11:21
	→ dytrivedi joined	11:21
	→ apa_ joined	11:21
	→ mtkd joined	11:22
	← mugwump left	11:23
apa_	hi, if i have repo A, and B, and from A, i push changes to B, does it cause different SHA1 in B, compared i pull changes in B from A ?	11:23
	→ mugwump joined	11:23
	→ nelmwyn joined	11:24
	→ flaguy48 joined	11:25
	→ munichlinux joined	11:25
	→ chittoor joined	11:25
	→ PureVirtual joined	11:27
	← kpreid_ left	11:27
	← ph^ left	11:27
shruggar	apa_, when you push, no	11:28
	apa_, when you pull, technically no	11:28
	→ kar8nga joined	11:28
	→ smuf joined	11:29
doener	the commits that got fetched by the fetch part of pull will be the same as in the remote repo	11:29
apa_	shruggar: i see.. now what i see is different SHA's in different repositories, and still same comments and codes. :( :( :(	11:29
	← Patrulka left	11:29
doener	if the pull merges, you might get an extra merge commit.	11:29
	→ hyperair joined	11:29
shruggar	apa_, when you pull, you may wind up with HEAD on a different commit because a merge may have happened, what he said.	11:29
doener	if the pull rebases, your local history gets recreated, so you get new commits for that part	11:29
	sounds like one repo saw a bad rebase	11:30
apa_	i write what has happened:	11:30
	→ pano_ joined	11:31
	→ Patrulka joined	11:31
	→ level09 joined	11:31
	→ viniciusluiz joined	11:31
	← riz2 left	11:32
apa_	I took checkout from svn with git svn from machine A with machine B, then i clone this checkout to memory stick C, clone this stick content to to machine D, where i did changes, push/pull (cant remember) back to stick, where it changes has push/pull to machine B, where i committed those to SVN to machine A.	11:32
doener	the dcommit creates new commits	11:32
apa_	When the stuff has moved from stick C to machine B, the SHA has changes	11:32
	→ ph^ joined	11:32
doener	and those new commits will of course have different sha1s	11:33
apa_	the sha1 change was happened before dcommit	11:34
	→ sjain joined	11:36
	→ rlotun joined	11:40
	← malumalu left	11:40
	← MetaCosm left	11:41
	← kar8nga left	11:41
	← jas4711 left	11:42
	← psankar left	11:43
	→ JEEB joined	11:45
	← epic_ left	11:45
	→ TadasEta joined	11:48
TadasEta	Hi	11:48
	How do I checkout previous revision of my project (example)?	11:49
apa_	TadasEta: maby you mean git reset --hard SHA1	11:50
	← dmgr left	11:50
	→ MetaCosm joined	11:51
segher	that _throws away_ everything newer	11:51
TadasEta	apa_: I have GUI program which can CHECKOUT to previous revision. Know I am searching how to do that in CLI	11:52
segher	you want git checkout xxxxx	11:52
	git checkout HEAD^	11:52
	caret means "first parent of"	11:52
TadasEta	segher: What if I want to checkout some specific revision (not previous). How do I identify it?	11:52
segher	git checkout sha-of-that-thing	11:53
	there are lots of ways to name commits	11:53
	← steveire left	11:54
apa_	hmm.. what i have used git reset --hard, it doesnt throw away nothing.. i can always go back "newer" with reset..	11:54
segher	http://www.kernel.org/pub/software/scm/git/docs/git-rev-parse.html has a big list of ways to name commits	11:56
	apa_: if it hasn't been gc'ed yet, sure	11:56
	→ steveire joined	11:56
	← steveire left	11:56
	→ steveire joined	11:56
apa_	TadasEta: You can find the commits by "git log" or "git reflog" etc.	11:56
TadasEta	Thanks	11:56
segher	but it changes the head of your current branch, which is not a good idea to do in general	11:56
TadasEta	One more question. What is quick way to discard any local changes made?	11:57
	← [J]stringo0 left	11:57
	← stringo0 left	11:57
	← stubbornmule left	11:57
apa_	reset	11:57
	it depend	11:58
	← pano_ left	11:58
apa_	if you like to keep the changes what you have made, then checkout, but if you absolute sure you like to discard them, reset.	11:58
segher	git checkout some-file will do it, if you haven't staged it yet	11:58
	"git status" will tell you what to do, always	11:59
	← Patrulka left	11:59
apa_	the git-rebase manual uses word <upstream>, what does it means?	12:01
	→ Patrulka joined	12:01
mapreduce	I find that git status shows a file is modified, but there's no output from git diff, git checkout thatfile doesn't help, and git config core.autocrlf true, false or input, don't help.	12:03
	Any suggestions?	12:03
	jony → jony\|away	12:05
	→ dianat joined	12:05
	→ cemerick joined	12:06
segher	git diff --cached ?	12:06
mapreduce	I fixed it by setting core.autocrlf to false and then doing a git checkout .	12:07
	and a git reset --head thatfile	12:07
	s/--head/HEAD/	12:07
cemerick	I'd like to apply a set of patches to multiple refs -- essentially, 'rebase other-ref', without moving the ref I'm on currently. Hints?	12:07
	→ PointMan joined	12:07
	→ mosno joined	12:09
	→ engla joined	12:10
TadasEta	apa_: I want to discard local changes	12:10
	← engla left	12:10
	→ engla joined	12:10
TadasEta	git reset seems to leave them...	12:10
	→ dmgr joined	12:10
apa_	TadasEta: there is also git-clean	12:11
segher	bad plan	12:12
	read what "git status" tells you. do that.	12:12
charon	apa_, TadasEta: to throw away all uncommitted state since the last commit, 'git reset --hard'. i hate to say it but all other advice about checkout, reset etc is wrong. (it _is_ possible with checkout, but you need 'git checkout HEAD -- .')	12:13
	apa_: <upstream> is notionally the branch that your current branch was based on when you created it. ("notionally" because that doesn't have to be the case, depending on the effect you want to achieve)	12:14
apa_	segher: Lets think that you have just tried something fancy with your project, without any ideas to save you changes, how git status will clean your code back to last clean state??? What i can see, the only way to clean is git-clean or git soft reset	12:14
charon	apa_: read the section headers that 'git status' shows.	12:14
	← viniciusluiz left	12:15
segher	git status does not clean your code. it tells you the status of your files. it tells you how to undo stuff as well	12:15
	→ viniciusluiz joined	12:15
charon	apa_: and please, do not tell others dangerous advice as long as you are not absolutely sure about it. it's very easy to lose data/work with git-reset and git-clean.	12:15
segher	git-clean should only work if you pass the --yes-i-really-mean-it option three times	12:16
	← hyperair left	12:16
segher	way too dangerous	12:16
charon	well, it already requires it once	12:16
segher	still :-)	12:17
apa_	What is the part in git-status header which tell me how to discard changes?	12:17
	→ oddmunds joined	12:17
segher	# Changed but not updated:	12:18
	# (use "git add <file>..." to update what will be committed)	12:18
	# (use "git checkout -- <file>..." to discard changes in working directory)	12:18
apa_	i see	12:18
TadasEta	segher: How to define that "<file>" to include all project root?	12:18
segher	# Changes to be committed:	12:18
	# (use "git reset HEAD <file>..." to unstage)	12:18
apa_	Then maby i have little old version from git becuse it doesnt show that..	12:18
	sorry	12:19
TadasEta	seems like git checkout -- . is what I needed	12:19
	x)	12:19
	← BP{k} left	12:19
	← flaguy48 left	12:21
	← maiOo left	12:21
	← Pev_away_ left	12:22
	→ maiOo joined	12:22
	→ Pev_away joined	12:22
lack	apa_: What version do you have? 'git --version'	12:23
apa_	git version 1.6.0.4	12:24
	→ BP{k} joined	12:25
	→ pheaver joined	12:25
charon	how unlucky, 1.6.1 has the improved status... consider upgrading to 1.7.x while you're at it	12:26
	← cemerick left	12:26
	← dianat left	12:26
	← mosno left	12:26
apa_	Well.. personally i have get used to work with git reset and clean, so there is no big need for better status..	12:26
	← megazig left	12:27
apa_	But yes, i should need newer git, newer linux version, and newer laptop. =)	12:27
	→ flaguy48 joined	12:27
	→ crankharder joined	12:28
	→ malumalu joined	12:31
	→ dforsten joined	12:34
	← roop left	12:35
	→ joeytwiddle joined	12:35
	← viniciusluiz left	12:36
	→ priidu_ joined	12:36
	→ viniciusluiz joined	12:36
	→ fcoury joined	12:43
	← fcoury left	12:43
	→ fcoury joined	12:43
	← fcoury left	12:44
	→ fcoury joined	12:44
	← fcoury left	12:44
	→ fcoury joined	12:44
	← fcoury left	12:45
	← apa_ left	12:45
	→ fcoury joined	12:45
	← fcoury left	12:45
	→ Bass10 joined	12:45
	→ fcoury joined	12:46
	← fcoury left	12:46
	→ fcoury joined	12:46
	← level09 left	12:47
	→ kukks joined	12:47
	→ auslander joined	12:48
	← auslander left	12:48
	→ auslander joined	12:48
	→ hyperair joined	12:48
	← sjain left	12:49
	→ merlin83 joined	12:52
	← merlin83 left	12:52
	→ merlin83 joined	12:52
	→ adamholt joined	12:52
	← tasslehoff left	12:52
	→ sjain joined	12:54
	← R_Rios left	12:54
	→ camonz joined	12:55
	→ gebi joined	12:56
	← munichlinux left	12:57
	→ o_a joined	12:57
	fcoury → fcoury\|away	12:58
	← nelmwyn left	12:58
	← byang left	13:01
	→ altogether joined	13:02
	← gebi left	13:02
	→ gebi joined	13:03
	→ YingFan joined	13:05
YingFan	How can I be ahead of a branch I never work on, and only do an occasional pullupdate on? Does that mean they did some kind of rollback or rebase?	13:06
Aides	YingFan: pastebin the output of git status	13:07
	← wolog left	13:07
Aides	you can see the commits that you have and the remote doesn't with something like: git log origin/master..	13:07
	(assuming origin/master is the remote branch you need)	13:08
	→ HG` joined	13:09
	← lucid_ left	13:09
	→ kpreid_ joined	13:10
	→ mm_work joined	13:12
	→ byang joined	13:12
YingFan	Aides just says on branch master..., Your branch is ahead of 'origin/master' by 4 commits. , nothing to commit..clean	13:12
	← mm_work left	13:12
YingFan	checking the log you suggested...	13:12
	→ hugoxrosa joined	13:13
	← gebi left	13:13
	→ Axius joined	13:13
	→ wolog joined	13:14
YingFan	hmm, how to sort that log on date?	13:14
	date-order..., reading the manual, how terrible ;-)	13:15
	→ infogulch joined	13:15
	→ sagsousuke joined	13:16
	→ meanburrito920 joined	13:16
	← meanburrito920 left	13:16
	→ meanburrito920 joined	13:16
YingFan	git diff HEAD..origin/master	13:17
	→ Mezner joined	13:17
	← froschi left	13:17
	← groove left	13:18
	→ gebi joined	13:18
	← alezandro left	13:19
	← ph^ left	13:20
byang	charon: I have updated my application according your feedback and I will public my work in this summer in github,	13:21
	← dytrivedi left	13:21
	→ ph^ joined	13:21
byang	charon: If there is no other issues, I will submit my application. :)	13:21
	← YingFan left	13:21
	← sjain left	13:21
	→ YingFan joined	13:21
charon	byang: traditionally people have used repo.or.cz for git "forks"/topics, but it's not like this really matters	13:22
	→ PerlJam joined	13:22
charon	other than that, go for it :-)	13:22
byang	charon: I have make a mirror from github to make a repo.or.cz , too. :)	13:22
	I found most of git hackers like repo.or.cz :-)	13:23
	← gebi left	13:24
	← imajes left	13:24
	← iaj left	13:24
	→ UncleCJ joined	13:24
byang	Michael has also talked with me about that. :)	13:24
UncleCJ	Hi all. Probably super-simple thing, but git log -1 always shows the most recent commit, even if I use --reverse, wtf?	13:25
charon	UncleCJ: that's just how -<n> and --reverse are defined to interact. it first limits the number of commits, then reverses the list	13:26
	→ megazig joined	13:26
UncleCJ	charon: See, I said it was silly of me. Of course it works like that. So how do I reverse and then limit?	13:26
	→ imajes joined	13:27
charon	UncleCJ: no really good way, unless you restrict yourself to one-line output, as in the case of 'git rev-list HEAD \| tail -1'	13:27
doener	git rev-list HEAD \| tail -n 10 \| git log --no-walk --stdin	13:28
UncleCJ	doener: Wow	13:28
	chaoflow_ → chaoflow	13:28
doener	IIRC the --stdin for log needs a somewhat recent git	13:28
	charon bows to doener's --stdin-fu	13:28
UncleCJ	doener: Yeah	13:29
doener	OTOH, th	13:29
	← Patrulka left	13:29
UncleCJ	charon: Yeah, I guess I can do oneline instead. Just run it twice instead	13:29
doener	is should also do: git log --no-walk $(git rev-list ... \| tail ...)	13:29
	doener is fully used to his new keyboard yet...	13:29
	→ gebi joined	13:30
doener	s/is/isn't/	13:30
	see? ;-)	13:30
UncleCJ	doener: Stop it, you're making me hot :-)	13:30
byang	charon: any interest to have a look at http://article.gmane.org/gmane.comp.version-control.git/143179 , I really think the -M option does more than it should...	13:31
UncleCJ	charon: I'm essentially doing something to create changelogs and it would be nice to work with --since, --before and whatever to just give one resulting commit in all cases...	13:31
Tass`	how to get changed loc for a commit?	13:31
	→ Patrulka joined	13:31
_rane	git show	13:32
charon	byang: i gave it a brief look but haven't replied so far because i couldn't explain it any further than what you already write in the email	13:32
shruggar	Tass`, please rephrase	13:32
byang	charon: Ah, I am wondering if some patch to make -M do what exactly it should do make sense ?	13:33
charon	byang: don't think so, unless you can significantly cut down the slowdown incurred by -M	13:33
	byang: the options aren't so much about progressively shifting blame "elsewhere", than about spending progressively more time trying to find a "better explanation"	13:34
	← PerlJam left	13:34
charon	i think of it as -M enabling an O(file size) algorithm, -C an O(#files), -CC an O(#files^2) and -CCC an O(#files^2 * #commits), even though i haven't actually checked that these are the correct bounds	13:35
	→ Swimming_Bird joined	13:36
mmattice	offby1: still around?	13:36
	← jony\|away left	13:36
byang	I think -C O($modified files) :-)	13:36
	→ sjain joined	13:36
	← Axius left	13:36
charon	er, yeah, that sounds more sensible.	13:36
	← gebi left	13:36
Tass`	shruggar, need the number of changed loc of a commit	13:36
charon	byang: if you can at least find a good estimate for the running time, actually documenting those might be a good idea though	13:36
shruggar	Tass`, perhaps "git log -1 --stat thecommit" ?	13:37
byang	charon: yeah, so, documenting these means to write them to user manual ?	13:37
charon	byang: yes	13:37
shruggar	Tass`, I am assuming when you say "loc" you mean "lines of code"	13:38
Tass`	shruggar, yes, I do.	13:38
byang	charon: And maybe update the explanation of '-M' is also necessary. :)	13:38
charon	byang: if it benefits the user :)	13:39
	though i can actually see it become simpler	13:39
byang	ok, thanks charon	13:39
	← YingFan left	13:40
	← dmgr left	13:40
	→ gebi joined	13:42
	→ spearce joined	13:43
	→ roop joined	13:45
	← codeswing left	13:46
	← mtkd left	13:46
	← gebi left	13:47
	→ gebi joined	13:47
	→ dmgr joined	13:47
	→ Sonderblade joined	13:49
	← mw left	13:49
	← dmgr left	13:53
	→ perlpilot joined	13:53
	← cedricv left	13:54
	← Patrulka left	13:54
	→ Patrulka joined	13:55
	→ CyuLat joined	13:55
	→ alexmchale joined	13:56
	← CyuLat left	13:57
	← PureVirtual left	13:57
	→ nicklodd joined	13:57
	← sylr left	13:57
	→ mtkd joined	13:58
	→ codeswing joined	13:59
charon	byang: btw, finding evil merges is easy, running 'git log --grep=evil' immediately turns up e2d2e38	13:59
Gitbot	[git e2d2e38]: http://tinyurl.com/ylah8nn -- Merge branch 'jc/maint-1.6.4-grep-lookahead' into jc/maint-grep-lookahead	13:59
	← tomekw left	14:01
	← dajero left	14:01
_rane	e	14:01
	← PointMan left	14:02
	← bx2 left	14:03
	← Caius left	14:03
	← Belna left	14:04
	→ jtaby joined	14:04
	→ dmgr joined	14:05
	→ Caius joined	14:05
	← sde_antrix1 left	14:06
	→ alezandro joined	14:06
	→ jschoolcraft joined	14:06
jschoolcraft	Is there a decent story for git on windows beyond: it sucks? Have a client that needs to update some HTML on a project I have in git. They're on Windows and of the MS VS persuasion. So was hoping something like tortoisegit existed and worked alright.	14:07
mapreduce	Git Bash seems great.	14:08
shruggar	tortoisegit exists and works alright	14:08
	doesn't it?	14:08
jschoolcraft	shruggar: it exists, not sure about the works alright part	14:08
mapreduce	Oh, you mean people who like mice.. dunno.	14:08
jschoolcraft	rather like to just give them one, simple option and be done.	14:09
mapreduce	It's not particular to HTML, but IntelliJ IDEA's git support seems pretty good.	14:09
	I've prepared installation and git clone instructions before for a less-techie person and they didn't struggle.	14:09
steffkes	jschoolcraft, tortoiseGit is fine for them	14:10
jschoolcraft	mapreduce: not sure I can get them to install IntelliJ in addition to all their Visual Studio stuff.	14:10
	steffkes: thanks, will start there.	14:10
shruggar	tortoisegit seems to work well enough for basic use. I've been trying to work out how to tell people to use it in a actually useful manner and have yet to	14:11
	→ andrewthorp joined	14:11
	→ sven^ joined	14:11
andrewthorp	Hey guys	14:11
	Is anyone using git with ruby on rails?	14:12
thiago	gitorious is	14:12
jschoolcraft	andrewthorp: loads of people	14:12
andrewthorp	I guess a better way to ask my question is this:	14:12
	I backup some of my database in fixtures, and one of them in particular became a yaml file that is about 10.5 MB -- I know this is way too large	14:13
	So in my branch I have this file committed, and I need to rebase it on master	14:13
	git rebase master from this branch takes FOREVER though (for obvious reasons)	14:13
	how can I speed it up?	14:13
wereHamster	git --version ?	14:14
andrewthorp	1.6.6	14:14
wereHamster	I seem to remember a bug/issue with rebase which caused it to be very slow, but I don't remember if/when it was fixed	14:15
shruggar	rebase used to not fast-forward when it could have, that's been fixed within the past couple of months	14:15
	← jschoolcraft left	14:16
engla	that's probably fixed already in 1.6.6	14:16
	← pheaver left	14:16
	← murr4y left	14:17
byang	charon: I think they are mostly evil merge, and I admit they should also be considered by line level browser	14:17
	→ yevlempy joined	14:17
	← shame left	14:17
	← codeswing left	14:17
andrewthorp	so	14:22
	any ideas	14:22
	how I can speed that up	14:22
	git will sit there and process rebase for 30 minutes if i let it	14:22
	but its only slow when I have these 10MB files	14:22
	that basically every line is changed in	14:22
	← ph^ left	14:22
	→ ph^ joined	14:23
charon	andrewthorp: haven't really thought about it but try 'git rebase -m ...'	14:24
shruggar	perhaps there is a way of telling git that they are binary files and shouldn't be merged?	14:24
charon	shruggar, andrewthorp: there's already been a thread on the list that basically concluded libxdiff lacks the "fast mode" of gnu diff, which makes diff performance abysmal on such files. so the culprit is format-patch, which is used in ordinary rebases	14:25
	i'm not sure if merge has the same issues, but it's worth a try	14:25
	← skoop left	14:26
	→ rduplain joined	14:27
	→ joeconyers joined	14:29
	← joeconyers left	14:29
	← ph^ left	14:29
	→ t0rc joined	14:29
	→ aryounce joined	14:30
	→ Daenyth joined	14:32
	→ nanoo joined	14:32
andrewthorp	gr	14:32
	git rebase -m is taking forever too	14:33
Daenyth	Hi, I'm trying to set up a pre-commit hook and I'm wondering what commands I can use to do it. Basically I/we use git to manage our package repository build scripts, and I'd like to make the pre-commit script check to make sure all required source files are checked in, or will be	14:33
andrewthorp	/frustrating	14:33
Daenyth	How can I look at the names of files in the index, and in the current tree (as git sees it)	14:33
	→ endpoint_david joined	14:33
	→ simplechat joined	14:33
	→ anno^da_ joined	14:33
	← simplechat left	14:33
	→ simplechat joined	14:33
	← sagsousuke left	14:34
Daenyth	I'd like to check the commit for any/all files named PKGBUILD, then extract the sources from there, then check that all sources are either in the index to be committed or are already existant in the repo	14:34
endpoint_david	is it possible to do a `git rebase` purely in the index, i.e. leave the workdir alone?	14:34
	→ codeswing joined	14:34
Daenyth	Aha, does git ls-files --stage show me the list of files in the index to be committed?	14:36
	← wolog left	14:36
Daenyth	There we go...	14:36
endpoint_david	(trying to push some changes, but non-ff upstream, and don't want to merge, but it's a live site, so don't want changes to occur purely from the local commit shuffling)	14:36
	← anno^da left	14:37
charon	andrewthorp: any change if you set proper .gitattributes that make the files binary, as shruggar suggested? you may have to set them in .git/info/attributes to avoid the change being reversed when jumping around in history	14:37
	← ilteris left	14:37
	ilteris__ → ilteris	14:37
charon	(of course that won't merge any more)	14:37
	→ murr4y joined	14:37
	→ loincloth joined	14:37
andrewthorp	hm	14:37
	im not following	14:38
	would the file essentially be ignored?	14:38
	that is - any changes to them from here on	14:38
charon	it would generate binary diffs	14:38
	← ankit9 left	14:38
charon	which are (IIRC) "remove these contents, add these other contents"	14:38
andrewthorp	oh	14:38
	ok hm sec	14:38
charon	so you lose the ability to merge over changes, but diff generation should be essentially instant	14:38
	that's my second educated guess though, if that fails too i'll have to start testing ;)	14:39
	← yevlempy left	14:39
	→ ilteris___ joined	14:39
	→ disturb joined	14:39
andrewthorp	ok so just so i follow	14:39
	I should be editing .git/info/attributes in my feature branch	14:39
charon	that's an untracked file, so it doesn't live in any branch/commit	14:40
andrewthorp	Ah	14:40
charon	does the same thing that .gitattributes does, but is not versioned	14:40
andrewthorp	hm	14:40
	so that would be in my root directory	14:41
	not project	14:41
	right?	14:41
charon	huh?	14:41
mmattice	why does git rebase have trouble applying empty commits?	14:41
charon	well, it's at the root of your repository, where the .git lives	14:41
andrewthorp	that file doesn't exist in my repo	14:41
charon	that's ok, but .git/info should already be there	14:41
andrewthorp	ya it is	14:41
	so vim attributes	14:41
	?	14:41
\yrlnry	Suppose I created a git repo but forgot to make it bare. What do I need to do to it to make it a bare repo? Is it enough to mv repo-dir JUNK; mv JUNK/.git repo-dir; rm -rf JUNK ?	14:41
charon	andrewthorp: yes	14:41
	← marj left	14:41
andrewthorp	and then how do i tell it .yaml is binary	14:42
	?	14:42
	*.yaml	14:42
wereHamster	\yrlnry: no, you also have to set core.bare to true	14:42
andrewthorp	*.yml	14:42
	lol	14:42
	tmz_ → tmz	14:42
charon	andrewthorp: *.yml binary (cf man gitattributes)	14:42
Gitbot	andrewthorp: the gitattributes manpage can be found at http://git.or.cz/man/gitattributes	14:42
wereHamster	\yrlnry: and disable reflog (bare repos have reflog disabled)	14:42
\yrlnry	wereHamster: Awesome, thanks. Where was that documented?	14:42
andrewthorp	binary	14:42
	wohops	14:42
wereHamster	\yrlnry: in my brain	14:43
\yrlnry	Anywhere else?	14:43
	← t0rc left	14:43
wereHamster	maybe in the faq	14:43
\yrlnry	Did it arise spontaneously and originally in your brain, or did it perhaps migrate there from elsewhere?	14:43
wereHamster	from experience	14:43
	← endpoint_david left	14:44
andrewthorp	ok, I added attributes in info and it only contains 1 line	14:44
	*.yml binary	14:44
	do I need to tell git to use this file, or am i good to go?	14:44
\yrlnry	wereHamster: thanks again	14:44
	← mtkd left	14:44
	→ wolog joined	14:44
mmattice	what'll happen if I rebase --continue over a commit (that's empty anyway)?	14:45
charon	andrewthorp: git will use that automatically	14:46
andrewthorp	k	14:46
\yrlnry	wereHamster: It is indeed in the FAQ. https://git.wiki.kernel.org/index.php/GitFaq#How_do_I_make_existing_non-bare_repository_bare.3F	14:46
	→ agile joined	14:46
	→ groove joined	14:46
	→ Ryback_ joined	14:47
groove	can anybody point me in the direction of solving the error "fatal: unable to write file foo mode 120000" from a git rebase	14:47
	noodl_ → noodl	14:47
tango_	disk full or wrong permissions?	14:47
andrewthorp	still very slow	14:47
	hm	14:47
	wow	14:48
	git rebase -m	14:48
	was very fast	14:48
groove	tango_: i'm fairly certain the disk is not filling, the repo is only about 1GB and i have 20GB free	14:48
	tango_: what kind of permissions problems would cause this?	14:49
	→ hebz0rl joined	14:49
	← eighty4 left	14:49
tango_	no idea	14:49
	→ ToreadorVampire joined	14:50
groove	should the file rather be 120777 ?	14:51
	→ irc2samus joined	14:51
mmattice	why does a rebase fail on empty commits if a --continue is all it takes to keep going?	14:51
	→ bdimcheff joined	14:51
ToreadorVampire	Hey all - does anyone know of/can anyone recommend a nice tool for building changelogs?	14:51
	→ DavidKlein joined	14:51
wereHamster	ToreadorVampire: git log	14:51
andrewthorp	charon: thanks for all of your help	14:52
	→ patmaddox joined	14:52
andrewthorp	charon: you've been great	14:52
	← jtaby left	14:52
mmattice	should that be considered a bug?	14:52
	→ project2501a joined	14:53
byang	Ah, the gsoc application editor is really not user friendly...	14:53
ToreadorVampire	wereHamster: Oh, yes, I see what you mean, but that's not quite what I meant, I probably confused the issue by misusing the word "changelog"	14:53
charon	andrewthorp: glad to hear it worked. you can put the attributes in a .gitattributes in the relevant directory (or just at the repo root) to make them tracked. (i wanted it to be untracked so as to avoid any surprise when rebase checks out older revisions)	14:53
andrewthorp	ya	14:53
ToreadorVampire	I mean - a tool that would help me write good commit messages?	14:53
	What has happened is that over the past day I have done a stack of work (90% of which has been focussed on a single feature, which is quite big and required a couple of db refactorings and other things) ... this means I have changed rather a lot of my project :s	14:54
	← flaguy48 left	14:55
	← merlin83 left	14:55
	→ zjason``` joined	14:55
	← kpreid_ left	14:55
ToreadorVampire	It trivially easy to bang in a two-line commit message along the lines of "Massive refactoring and feature addition, see #374 on the tracker, addition of 'get my own status' method"	14:55
	But I really want to be able to put more meaningful comments against the files that I changed (and ended up refactoring along the way) - I guess what I want is a file-by-file display, with ability to see a diff (reminding me what i changed) and then a slot to type in a change/commit entry for that file	14:57
	← zjason`` left	14:57
	← jsquared left	14:57
wereHamster	commit messages are per-commit, not per-file. But you are free to make more granular commits.	14:57
	← byang left	14:58
	← alezandro left	14:58
mmattice	awww, hell, is doing a rebase --continue over empty commits dropping them?	14:58
ToreadorVampire	wereHamster: Yeah, I suppose what I'm saying is "I left it too long between commits" and my commit message (summary of what I changed in this commit) is pretty massive, and I'd like something to remind me of everything I did over the past 24h ;)	14:59
	→ jsquared joined	15:00
shruggar	mmattice, unless you tell it to keep them, yes	15:00
	→ ankit9 joined	15:00
wereHamster	ToreadorVampire: you can split the commit up into smaller pieces	15:00
	→ lanthan__ joined	15:01
	→ byang joined	15:01
	← _iron left	15:02
	→ richcollins joined	15:02
	→ _iron joined	15:02
mmattice	shruggar: how would one go about that?	15:03
	← everzet left	15:03
shruggar	mmattice, I may have been mistaken, I really thought there were such an option, but skimming through the manpage I did not see such a thing	15:03
	← byang left	15:03
	→ dajero joined	15:03
	← nicklodd left	15:04
	← UncleCJ left	15:04
	→ Danielpk joined	15:04
Danielpk	how i remove a branch?	15:04
mmattice	scaring the crap out of me :)	15:04
wereHamster	Danielpk: man git-branch	15:04
Gitbot	Danielpk: the git-branch manpage can be found at http://git.or.cz/man/git-branch	15:04
	← lanthan_ left	15:04
	→ webchick joined	15:04
	→ mtkd joined	15:04
	← kkoehne left	15:05
Danielpk	wereHamster, thx =)	15:06
	← giallu left	15:07
	← gebi left	15:07
	← andrewthorp left	15:07
	← codeswing left	15:08
	→ adamm joined	15:08
	→ gaveen joined	15:09
	← dmgr left	15:09
	→ kpreid_ joined	15:10
	← meanburrito920 left	15:10
	← richcollins left	15:10
	→ drunknbass_work joined	15:11
	→ cedricv joined	15:12
	→ gebi joined	15:13
	→ alezandro joined	15:14
	→ munichlinux joined	15:15
	← ankit9 left	15:15
	← Pupeno left	15:16
	→ Pupeno joined	15:16
	→ dytrivedi joined	15:17
	→ icefox joined	15:17
	→ simNIX joined	15:19
	→ florianb joined	15:19
florianb	hi there, i have a question. i pushed a wrong commit to a remote master instead to a topic.	15:20
	→ nakkor joined	15:20
icefox	florianb: you can force push the correct branch to the remote master	15:21
	→ TheMoonMaster joined	15:21
florianb	i reverted the commit local with git-reset --hard HEAD^^ (while the 1st ^ was a merge-commit)	15:21
	→ codeswing joined	15:21
	→ voodootikigod_ joined	15:22
florianb	when i'm trying to push the corrected master branch again, i get the xy-commits behind message. if i would try to megre now, i would get a new (unwanted) merge-commit don't i?	15:23
	← mahound` left	15:23
	← robinbowes left	15:23
	← fcoury\|away left	15:24
florianb	the question i have is, how do i revert these two commits in the remote with (as best) no further commits?	15:24
	→ fcoury joined	15:25
florianb	icefox: you mean i should just "git-push -force origin master"?	15:25
voodootikigod_	is there a good way to reduce the size of .git/objects/pack ?	15:25
\yrlnry	florianb: You can push -f; that will reset the remote branch to point to the same commit as the local branch.	15:25
	→ kipras joined	15:26
icefox	florianb: yah, if you want to overwrite what is on the remote side you use -f	15:26
	← jspiros left	15:26
	→ jspiros joined	15:26
shruggar	voodootikigod_, the "good" ways, git does automatically, afaik	15:26
voodootikigod_	mine is up to 199mb	15:27
	seems a bit big	15:27
florianb	okay - and what is about the 'Your branch is behind 'origin/master' by 154 commits'-message. Wouldn't i kill 154 commits from other devs?	15:27
icefox	yup	15:28
	← anno^da_ left	15:28
florianb	so couldn't i just fetch first, doing no merge and then push?	15:29
icefox	a better alternative would be to revert the commit and push that	15:29
shruggar	voodootikigod_, matter of opinion I guess. That doesn't sound big to me at all	15:31
voodootikigod_	cool	15:31
	thanks	15:31
florianb	can i undo the reset?	15:32
	← sh1mmer left	15:33
shruggar	voodootikigod_, git wastes a lot of space with history as part of its fundamental design. It's just a "historical accuracy is more important than the penny it costs to get another gig of space" issue. I'd prefer it weren't there, but it is	15:33
	→ dmgr joined	15:33
	← groove left	15:33
	→ tedoc2000 joined	15:34
engla	git doesn't waste space. it's amazing how much it can fit in that little space(!)	15:34
karihre_	florianb: git pull origin master ?	15:34
	→ yn joined	15:35
florianb	thanks..	15:35
icefox	Anyone familiar with git-am that could take a look at a patch for some feedback on it before I submit it to the ml? http://git.meyerhome.net/icefox/git/commit/fe4a20e38f39f7020c9b1a599824bdbb3097a264	15:37
	→ robinbowes joined	15:38
	→ krawek joined	15:38
	← mtkd left	15:39
	→ ankit9 joined	15:40
	← drunknbass_work left	15:40
	→ psankar joined	15:40
	← orospakr left	15:40
charon	icefox: i think the patch itself is ok, though you could probably do away with the awk and just use sed to grab the sha. but do we really want to show that sha unconditionally? when applying other people's mails, the sha will be meaningless and in many cases absent	15:41
	→ orospakr joined	15:41
	← Daenyth left	15:42
icefox	charon: I mostly am using it when I rebase, outputing Applying [subject] isn't very helpful when there is a conflict. Being able to see the original sha and look at what it was trying to do is often helpful to me to resolving the issue	15:42
	charon: any way to know if I am in the middle of a rebase?	15:42
charon	icefox: $dotest/rebasing will exist, there's plenty of checks for this already :)	15:43
icefox	cool, i'll change that then, thanks for checking it out	15:43
	← patmaddox left	15:47
	← ToreadorVampire left	15:48
	→ lobolife joined	15:48
	← alezandro left	15:49
lobolife	I'm working off the same git repository with a couple friends. It seems two of us can push and pull fine, but on one computer the recent commit's won't pull. It says up to date even when we know there are changes on the remote. Any ideas of why one wouldn't recognize changes on the remote?	15:50
	← megazig left	15:50
icefox	lobolife:what is the error that is outputed with it 'wont' pull'	15:50
	← nanoo left	15:51
lobolife	no error. That's the really odd thing. It just says "up to date"	15:51
	but it's not	15:51
	← EricInBNE left	15:51
	← Patrulka left	15:52
	→ t0rc joined	15:53
icefox	so in git log there is no merge?	15:53
	← wolog left	15:53
bremner	umm. so what is the current state of ticgit? is it dead? There seems to be a few forks on github. Is there a sucessor project?	15:54
	→ wolog joined	15:55
shruggar	I think ticgit is dead	15:55
	← hebz0rl left	15:55
	→ a2ron joined	15:56
	fcoury → fcoury\|away	15:56
	← robinbowes left	15:57
	← chittoor left	15:58
	← dmgr left	15:58
	→ dmgr joined	15:59
bremner	hmm. ditz might be worth a look.	15:59
	← Pupeno left	15:59
	← khelll left	15:59
	→ sylr joined	15:59
	← fujin left	15:59
	→ jrmuizel joined	16:00
	→ LiamH joined	16:01
	→ robinbowes joined	16:02
	→ hobodave joined	16:03
	← lobolife left	16:04
	← Blaine left	16:04
	→ lhz joined	16:04
	→ pano_ joined	16:05
	→ erikvold joined	16:06
	→ trochala joined	16:06
	← munichlinux left	16:07
	← yatiohi left	16:07
	← wolog left	16:08
	← steffkes left	16:08
	→ Blaine joined	16:08
	← harinath left	16:09
	← nevyn^ left	16:09
	→ dj_tjerk joined	16:09
	→ wolog joined	16:10
	→ Crell joined	16:10
	← bentob0x left	16:11
	← gebi left	16:11
	→ agib joined	16:12
	← hyperair left	16:13
	→ tobier joined	16:13
	← kpreid_ left	16:14
	→ alley_cat joined	16:14
	→ sh1mmer joined	16:15
	← ntoll left	16:15
	← jrmuizel left	16:17
	→ khelll joined	16:17
	← Crell left	16:18
	← wald0 left	16:18
\yrlnry	bremner: I was talking to Chacon about ticgit not long ago. Just a minute while I recover the discussion.	16:19
	→ jklo joined	16:19
bremner	thanks!	16:19
	← nakkor left	16:20
\yrlnry	bremner: http://gist.github.com/349253	16:21
	It's not much, but at least it's more than you had before.	16:22
	→ oal joined	16:22
\yrlnry	(Schacon is the original author of ticgit.)	16:22
oal	When I do "git status" all my files are red, and it says "Changed but not updated" above them. Why do I have to "git add" them every time I commit?	16:23
	→ harinath joined	16:23
bremner	\yrlnry: thanks again.	16:23
\yrlnry	oal: You don't. You can git commit -a and it will automatically commit everything it knows about.	16:24
oal	What about files that are marked with "delete:" under "Changed but not updated" will they be committed?	16:24
\yrlnry	oal: But more generally, having a separate "add" step allows you to commit some, but not all files, or some , but not all changes in a file.	16:24
	← petrux left	16:25
	→ shennyg joined	16:26
	← Yuuhi left	16:26
	→ patmaddox joined	16:26
	← eletuchy left	16:26
\yrlnry	oal: if you git commit -a, the removal of the file will be committed.	16:26
	Or if you "git add" the deleted file, then commit, the removal of the file will be committed.	16:26
	Otherwise, not.	16:26
oal	\yrlnry: Thanks. When I do git push I always get an error that I did not specify refspecs to push or something	16:27
\yrlnry	Show your complete git push command and the complete error message.	16:27
oal	I just to git push	16:27
\yrlnry	You didn't say where to push to.	16:27
	Mayb eyou want "git push origin"?	16:27
oal	\yrlnry: http://dpaste.com/177762/	16:27
	→ enherit joined	16:27
\yrlnry	Maybe you want "git push origin master"?	16:28
oal	I have checkouted this project from my server, and I'm pushing it back to the server	16:28
	← priidu_ left	16:28
oal	cloned*	16:28
perlpilot	oal: you can always follow the instructions given in the warning you know	16:28
oal	I tried to do a git push matching, but It didn't push anything, I think	16:29
perlpilot	oal: git config push.default matching # will get rid of the warning because you will have configured the setting	16:29
oal	Ok, thanks	16:30
	:)	16:30
	→ ph^ joined	16:30
perlpilot	oal: any time you see git warn about some configuration setting, you'll want to use "git config" to rectify things	16:30
	→ nakkor joined	16:31
oal	Thanks	16:31
	← florianb left	16:31
	← doobry left	16:32
\yrlnry	"How can I make git behave more like SCCS?"	16:33
	→ doobry joined	16:33
	← d0ugal left	16:33
	← dytrivedi left	16:34
\yrlnry	Also I was thinking about writing a program that takes a repository and rewrites all the commits so that the merge history is as convoluted and as bizarre as possible, but so that the resulting tree is the same. Does something like that exist already?	16:35
	← naeu left	16:35
	← simplechat left	16:35
	webchick → webchick\|fooood	16:35
	← codeswing left	16:37
	→ merlin83 joined	16:38
	← merlin83 left	16:38
	→ merlin83 joined	16:38
	← shruggar left	16:39
	← tstclair left	16:39
Fissure	is there a way to force git to output text diffs for files it thinks are binary?	16:40
	adamholt → adamholt_away	16:41
Fissure	er, i'm blind	16:42
	nvm	16:42
	must have skipped that option in the manpage	16:42
	→ domcat joined	16:42
	← project2501a left	16:43
	← marc_kdab left	16:44
	→ dreiss joined	16:44
	← sh1mmer left	16:46
	→ akahn joined	16:47
	perlpilot → PerlJam	16:48
	← sdboyer left	16:49
	← trivol left	16:50
akahn	How come I can't change branches if I have local changes? Why can't those changes just stay in that branch?	16:51
ToxicFrog	You can't change branches if you have local changes if those changes would be overwritten by the branch switch.	16:51
	← Mezner left	16:51
ToxicFrog	Specifically, the problem is this:	16:52
doener	a branch head references a commit, so naturally, uncommitted changes can't be part of a branch. They exist solely in you working tree/index	16:52
ToxicFrog	You are on branch A. It has file foo.	16:52
	Branch B has a different version of foo.	16:52
	You modify, but do not commit your changes to, foo, then try to checkout B.	16:52
	→ johnw joined	16:52
	→ bradhaydon joined	16:52
	← galderz left	16:52
ToxicFrog	If you do that, it has to overwrite foo with B's version, destroying your changes - and since you didn't commit them, they can never be recovered.	16:52
	To get around this, either reset those changes (if you don't want them), stash them (if you want to move them to a different branch), or commit them (if you want them on the current branch).	16:53
	→ wolog_ joined	16:53
	← wolog left	16:53
	← lresende left	16:54
	→ froschi joined	16:54
	→ magn3ts joined	16:54
	← HG` left	16:54
	← zenwryly left	16:55
	→ sh1mmer joined	16:56
	→ mindworx joined	16:57
akahn	thanks, that explains it	16:58
	→ eletuchy joined	16:58
akahn	there's no snapshot of those changes unless i stash or commit	16:58
	→ Theravadan joined	16:58
akahn	so they won't be there when i get back	16:58
mindworx	I have a repo on my server that i use to do all work for my clients.. I have a particular client that would like the dev environment set up on their local box.. I pulled from the repo to set up the rails site.. should i create a branch that they can work off of to make changes?	16:58
	so that way they are not pushing to master	16:58
	← roop left	16:59
	→ static^ joined	16:59
	→ richcollins joined	17:00
	← RobertLaptop left	17:01
	← rgr left	17:01
mindworx	?	17:01
	← kipras left	17:02
sitaram	dont let them push to you at all; tell them to allow you to pull from them	17:03
	because if you allow them to push to your server saying "oh by the way please dont push to master", then sure as hell someone will, sooner or later :)	17:04
	→ timj joined	17:04
	→ eletuchy_ joined	17:05
	→ aresnick joined	17:05
	← sjain left	17:07
	← eletuchy left	17:07
	→ sjain joined	17:07
	← aresnick left	17:07
	→ pantsman joined	17:08
	← magn3ts left	17:08
	→ rkenned joined	17:09
	→ meanburrito920 joined	17:09
	← meanburrito920 left	17:09
	→ meanburrito920 joined	17:09
	→ dole joined	17:10
dole	is anyone open to help a newbie in using git on a windows machine?	17:11
	→ jrmuizel joined	17:11
	← nicoulaj left	17:12
	→ priidu_ joined	17:12
	← meanburrito920 left	17:13
	← fivetwentysix left	17:13
	→ RobertLaptop joined	17:14
ToxicFrog	dole: go ahead	17:15
	→ bx2 joined	17:15
ToxicFrog	More generally, you're more likely to get help by asking than by asking whether you can ask.	17:15
	→ maxter joined	17:16
	→ aresnick joined	17:17
	→ hyperair joined	17:17
bremner	\yrlnry: I made a summary of ticgit and similar things at http://www.cs.unb.ca/~bremner/blog/posts/git-issue-trackers/	17:18
	← dionoea left	17:19
	→ dionoea joined	17:19
	→ Pupeno joined	17:21
	→ Chillance joined	17:21
	→ chittoor joined	17:21
	→ rgr joined	17:22
	← wolog_ left	17:23
\yrlnry	bremner: Many thanks!	17:24
	← curtana left	17:25
	→ codeswing joined	17:25
	← codeswing left	17:25
	→ codeswing joined	17:25
	→ AAABeef joined	17:25
	← AAABeef left	17:25
	← bluenovember left	17:26
	← rkenned left	17:27
	→ AAABeef joined	17:28
	← bx2 left	17:29
	← yn left	17:29
	→ kpreid_ joined	17:30
	→ marc_kdab joined	17:31
	← rocketeerbkw left	17:33
	adamholt_away → adamholt	17:33
	→ thiago_home joined	17:34
	← domcat left	17:35
	← doobry left	17:36
	← stamina left	17:37
	← richcollins left	17:39
	← t0rc left	17:41
	← tatsuya left	17:41
	→ mssm joined	17:41
	→ nanoo joined	17:42
	← mindworx left	17:43
	→ mindworx joined	17:43
	→ dgnorton joined	17:43
	← phishy left	17:44
	→ t0rc joined	17:44
dgnorton	is there a way to list the files that have changed from in a range of commits? sort of like ... git log --stat --name-only HEAD~3..HEAD except only list the file name once even if it change in more than one commit?	17:45
wereHamster	dgnorton: git diff --name-only HEAD~3..HEAD	17:45
dgnorton	wereHamster: thanks!	17:46
	← rlotun left	17:46
	← nakkor left	17:47
	→ lresende joined	17:47
	← lresende left	17:48
	→ tstclair joined	17:48
	→ chrisb joined	17:48
	← o_a left	17:50
	← hobodave left	17:50
	→ fool_ joined	17:51
	← fool_ left	17:51
	→ fool_ joined	17:51
	← novas0x2a left	17:51
	← bdiego left	17:52
	← fool__ left	17:55
	← khelll left	17:58
	← ph^ left	17:59
	← engla left	18:00
	webchick\|fooood → webchick	18:01
	← chrisb left	18:03
	→ jklo_ joined	18:03
crankharder	I'm trying to permanently convert a svn repo to git -- git svn clone -s ... works (and git br -av) shows that it correctly pulled in all the branches/tags -- What do I do at this point?	18:04
	→ domcat joined	18:04
crankharder	some places suggest using clone --bare, but that seems to wipe out my branches/tags	18:04
	← hyperair left	18:04
	← jklo left	18:05
	jklo_ → jklo	18:05
	← Swimming_Bird left	18:07
icefox	crankharder: start hacking?	18:07
	→ novas0x2a joined	18:09
jaalto	github gurus: I wan to collaborate with UPSTREAM "A". So I forked the project to "B". I have cloned both A and B to loca disk, and done at local disk "B" repository: "git remote add ../A upstream". What is the best way to stay in touch with the latest upstream chnages?	18:09
	→ HG` joined	18:09
wereHamster	fetch the changes, merge the changes	18:09
jaalto	Can i rebase somehow to the remotes/upstream/master?	18:10
wereHamster	git rebase remotes/upstream/master ?	18:10
jaalto	That branch need not to be "git checkout --track" first?	18:11
wereHamster	no	18:12
jaalto	Ok	18:12
	← gk__ left	18:14
	→ gk joined	18:14
	← gk left	18:16
	← t0rc left	18:20
	→ Axius joined	18:21
	← DrNick left	18:24
	→ brizly joined	18:24
	Arr`Gone → Arrgh`	18:25
	← Arrgh` left	18:26
	→ spawnyd joined	18:29
	← dole left	18:29
	→ webus joined	18:29
	← domcat left	18:31
	→ meanburrito920 joined	18:32
	← meanburrito920 left	18:32
	→ meanburrito920 joined	18:32
	← bourbon left	18:33
sylr	Hi	18:33
	← mcdebugger left	18:34
sylr	is it possible to revert only parts of commits	18:34
	?	18:34
wereHamster	git revert -n; git reset -p; git commit	18:34
	→ mcdebugger joined	18:34
	→ amorphous0815 joined	18:34
	← Axius left	18:35
sylr	what does git reset -p ?	18:35
thiago_home	reset -p? that's new for me	18:35
wereHamster	like add -p	18:35
	→ anzenketh joined	18:35
	→ engla joined	18:36
	← engla left	18:36
	→ engla joined	18:36
	← meanburrito920 left	18:36
	→ Patrulka joined	18:36
	→ DrNick joined	18:38
	→ UncleCJ joined	18:39
	→ eighty4 joined	18:39
	← dajero left	18:41
	→ Zephyrus joined	18:42
	→ novas0x2a1 joined	18:43
	← amorphous0815 left	18:43
	← novas0x2a left	18:44
	→ giallu joined	18:44
sylr	wereHamster, git reset HEAD, then git reset -p gives me No changes.	18:45
	git reset HEAD --no-commit ?	18:45
	← maxthoursie left	18:45
	← sjain left	18:48
	→ ph^ joined	18:50
	→ sixteneighty joined	18:52
	→ fujin joined	18:52
	← cedricv left	18:53
	← weary left	18:54
wereHamster	did I say you should do reset HEAD?	18:54
	← mindworx left	18:57
	→ weary joined	18:59
	→ khmarbaise joined	19:00
	← neuro_damage left	19:01
	→ khelll joined	19:01
	← sixteneighty left	19:02
	← ph^ left	19:02
	→ ph^ joined	19:03
	← ph^ left	19:03
	→ ph^ joined	19:04
	→ joeb joined	19:04
joeb	Is it possible to add a submodule into a project but only a specific directory of that submodule?	19:04
	→ mindworx joined	19:04
	→ sixteneighty joined	19:05
	← runasand left	19:05
	← LotR left	19:05
	← pcc1 left	19:05
drizzd	not without filter-branch	19:05
	← chuck left	19:05
joeb	I've never used filter-branch before, I'm assuming it's name is pretty much self explanatory of it	19:06
	→ mw joined	19:06
	← tychoish left	19:06
	← nothingmuch left	19:06
drizzd	you'd need filter-branch --subdirectory-filter	19:06
	→ iaj joined	19:06
joeb	drizzd: Can I perform it on a submodule just like normal?	19:07
	→ rkenned joined	19:07
	← gkelly left	19:08
	→ magn3ts joined	19:08
	← ScottMac left	19:08
	→ tychoish joined	19:08
drizzd	yes, but that implies that you first added the entire repo as a submodule. in your case I'd create a separate repo for the "submodule", apply filter-branch (you will have to do that again for each update), then include the result as a submodule.	19:08
joeb	Okay. I will try it out	19:09
	← mastro left	19:09
drizzd	actually, it wouldn't work otherwise, since you probably want to remove the repository state prior to filter-branch, making the old submodule state invalid as well.	19:09
	← psankar left	19:10
joeb	Interesting.	19:10
	Yeah, I'm doing filter-branch and it's iterating over every commit it looks like	19:10
	And, there's quite a bit, heh.	19:10
	→ ScottMac joined	19:11
	← engla left	19:11
	← mcdebugger left	19:11
	← tychoish left	19:11
joeb	So I should create a new repo, do all this (submodile, filter-branch) and then use that newly created repo as a submodule for other projects... have I got that right?	19:11
drizzd	yes, it's slow	19:11
sylr	wereHamster, sorry, I meant git revert HEAD	19:11
	→ Aikawa joined	19:11
	→ pcc1 joined	19:11
	→ mcdebugger joined	19:11
	→ chuck joined	19:12
	← jklo left	19:12
	→ runasand joined	19:12
	→ LotR joined	19:12
	→ tychoish joined	19:12
drizzd	joeb: you should have a separate repo, which will become a submodule, but is not a submodule per se. Use filter-branch on that, then add it as a submodule.	19:12
	→ jklo joined	19:12
drizzd	to get updates from the original repo, you will have to do the same again, except that you merge the result	19:13
joeb	basically the repo for a framework I use has things in it that I don't need, just its core libraries. I'd like to be able to submodule those libraries.	19:13
	→ cedricv joined	19:13
	→ nothingmuch joined	19:13
joeb	Okay, I think I got it.	19:13
	→ steelwil joined	19:13
	→ Pupeno_ joined	19:13
drizzd	you can even merge changes back using subtree merge later	19:13
steelwil	is this the right place to ask questions?	19:13
	← disturb left	19:13
	→ gkelly joined	19:13
wereHamster	sylr: did you use git revert HEAD or git revert -n HEAD ?	19:14
	→ sgh joined	19:14
joeb	drizzd: Thanks for the help!	19:14
drizzd	yw	19:14
	← joeb left	19:14
sylr	wereHamster, the first :P	19:14
	→ alester joined	19:15
	← Aikawa left	19:15
	→ fahadsadah joined	19:15
sylr	wereHamster, but it's okay, I reset and did it with the -n	19:15
	← Pupeno left	19:15
	Pupeno_ → Pupeno	19:15
	→ Aikawa joined	19:15
sylr	wereHamster, thanks	19:16
steelwil	I have two branches I want to merge just a certain file from the one branch to the other, how do I do that	19:16
wereHamster	you can't	19:16
drizzd	steelwil: that is per definition not a merge	19:16
	you want "git checkout <branch> <file>; git commit"	19:16
	oh	19:17
	well, maybe not exactly	19:17
steelwil	what I did was check out the one branch, copy and paste file to another directory	19:17
	ok drizzd I thought there would be a more elegant way of doing it	19:18
drizzd	but you can use "git merge-file"	19:18
	← HG` left	19:18
	→ beni- joined	19:18
	← spawnyd left	19:19
	→ Heimidal_ joined	19:19
steelwil	I will look up the git merge-file docs, thnks	19:19
	← altogether left	19:21
	→ Alagar joined	19:22
drizzd	steelwil: actually, it would be easier, starting with a clean worktree, to do "git merge <branch>; git reset; git add <file>; git checkout .; git commit"	19:22
	→ engla joined	19:22
	← engla left	19:22
	→ engla joined	19:22
drizzd	oh and you probably want to use git merge --squash instead	19:23
	← eletuchy_ left	19:23
drizzd	jeez, forgot --no-commit too, so it's "git merge --squash --no-commit <branch>; git reset; git add <file>; git checkout .; git commit"	19:23
	← maxter left	19:24
steelwil	ok thanks drizzd	19:25
	← Pupeno left	19:26
	→ Pupeno joined	19:26
	← kpreid_ left	19:27
	→ eletuchy joined	19:28
	← mindworx left	19:28
	→ coyo joined	19:29
	→ eletuchy_ joined	19:32
	← werdan7 left	19:33
steelwil	drizzd wont the "git reset' reset the branch to the last committed state?	19:34
	← nanoo left	19:34
	→ SimManiac joined	19:34
	← eletuchy left	19:34
SimManiac	hi	19:35
drizzd	steelwil: git reset HEAD (the default) never modifies the branch, since HEAD is already the current state	19:35
	agib → agib\|away	19:36
steelwil	ok	19:36
drizzd	the point of git reset there is to undo the changes git merge made to all the files other than <file>.	19:37
	→ kpreid_ joined	19:37
	← kpreid_ left	19:38
steelwil	what is stopping <file> from also being reset?	19:38
drizzd	unfortunately it also removes the three-way merge result in the index, so you can't do a three-way diff after that, but at least you have the merged file in the work tree	19:38
	→ icwiener joined	19:38
drizzd	steelwil: file is also being reset, that's why you re-add it after.	19:38
steelwil	ok now I understand	19:39
	→ Boriskr joined	19:39
drizzd	you could also do "git checkout HEAD -- $(git diff --name-only HEAD\|grep -v '^<file>$')" (not tested), which would preserve the index	19:40
	← webus left	19:41
	→ kpreid_ joined	19:43
charon	drizzd: far easier to GIT_INDEX_FILE=scratch_index git checkout HEAD -- .; rm -f scratch_index	19:44
	(or did i miss some intermediate step?)	19:45
	← Boriskr left	19:47
	→ project2501a joined	19:47
drizzd	charon: we're trying to merge only a single file instead of the entire branch	19:48
	← johnw left	19:48
	← [Po]lentino\|away left	19:48
charon	oh	19:48
	my usual take on that is 'git show --format=email HEAD...branch -- file \| git am'	19:49
	er	19:49
	charon should make up his mind on whether to use show or diff	19:49
	→ johnw joined	19:50
charon	git diff --format=email HEAD...branch -- file \| git apply	19:50
	the log -p \| git am form can be useful to make a "filtered rebase", though	19:50
drizzd	but that would potentially create a lot of out-of-context commits	19:50
	ah, ok	19:50
	but that last one won't do a three-way merge	19:51
charon	true. IIRC you can still feed a diff to am -3 if you fix the commit message afterwards, though	19:52
	← mssm left	19:52
	→ meanburrito920 joined	19:53
	← meanburrito920 left	19:53
	→ meanburrito920 joined	19:53
Ilari	spearce: Was the missing commit problem also fixed? I know missing trees/blobs was...	19:53
	→ jklo_ joined	19:53
drizzd	right, so there are a few ways to do it, but no command. I wonder if something like that would be useful. But then I don't know how I feel about encouraging people to use it, since it is essentially a cvs style merge, without merge history.	19:53
	← AAABeef left	19:53
	→ AAABeef joined	19:54
drizzd	but then so is cherry-pick, which does the same thing, only in the time dimension instead of the tree dimension	19:54
	← jklo left	19:54
	jklo_ → jklo	19:54
steelwil	well it all seems complicated	19:55
wereHamster	yeah, because that's not what git was written for	19:55
drizzd	I don't know about that. We could easily to it, we just haven't implemented it as a single command.	19:56
	← anzenketh left	19:56
	→ Louie` joined	19:57
steelwil	wereHamster explain the correct way, many commits?	19:57
	← alexmchale left	19:58
drizzd	steelwil: The point is that it gets really ugly when you try to track merges on file-level. I'm sure there is a large flaming thread in somehwere in git's mailing list history.	19:58
wereHamster	merging a single file from another branch was not a use case that linus had in mind when he initially wrote git	19:59
	→ Twisted_Mentat joined	19:59
	← jklo left	19:59
	← zjason``` left	19:59
	→ AkumaStreak joined	20:00
	→ phishy joined	20:00
	→ jklo joined	20:01
	← eighty4 left	20:01
steelwil	I would like do do something like git checkout experimental1branch theFile; git add theFile; git commit -m "added theFile into current branch"	20:01
wereHamster	you can do that	20:02
	← project2501a left	20:02
drizzd	but it's not a merge	20:02
	unless the file didn't change in the current branch	20:02
wereHamster	that, however, will not preserve the history (you will later not see that the file was copied from the other branch)	20:02
steelwil	drizzd what is it then?	20:02
wereHamster	that is a plain simple commit	20:03
drizzd	you take the version of the other branch and overwrite the version in the current branch	20:03
	← auslander left	20:03
drizzd	a merge, on the other hand, takes changes from both branches	20:04
steelwil	ok	20:04
	← UncleCJ left	20:05
steelwil	I can see that a merge would be handy	20:05
fr0sty	steelwil: Something like this might work: 'git checkout -b temp_branch <merge_base>; git checkout <other_branch> TheFile; git commit TheFile; git checkout <my_branch>; git merge temp_branch'	20:06
	agib\|away → agib	20:06
fr0sty	but as others have mentioned this gets ugly, messy, and is probably not what you really, truly want.	20:07
	← pkahle_ left	20:07
	← dgnorton left	20:08
	→ maxter joined	20:09
	→ acalbaza_ joined	20:09
acalbaza_	is it possible to svn dcommit a specific hash?	20:09
	or should i just checkout a quick branch, make my change, commit, then revery back to my work?	20:09
steelwil	I would get both versions into kdiff and merge that way	20:10
	→ altogether joined	20:10
	→ Eaven joined	20:11
avar	acalbaza_: When I work on git-svn I do all my work on a branch and cherry-pick stuff there before pushing it	20:14
	→ Scotto_ joined	20:15
	← froschi left	20:16
	→ unhygienix joined	20:18
	← meanburrito920 left	20:18
	→ meanburrito920 joined	20:18
	← meanburrito920 left	20:18
	→ meanburrito920 joined	20:18
Scotto_	whats the best way to get a list of commits and nested under those the list of files comitted?	20:19
	← jklo left	20:19
	→ jklo joined	20:20
	→ jtaby joined	20:20
thiago_home	git whatchanged	20:20
Scotto_	that can be applied to all files?	20:20
	← jklo left	20:20
Scotto_	niice	20:21
jtaby	Hey, I'm having a problem which I can't fix myself: When I try to do a 'git push', it says it failed because it's not fast forward, which implies there's a newer version on the server. When I do a git pull, I get an error telling me that I didn't tell it which branch I want to merge with	20:21
Scotto_	i just wish i could exclude the initial commit	20:21
jtaby	when I do a git fetch, nothing comes down	20:21
	any ideas?	20:21
Scotto_	as its very long	20:21
acalbaza_	avar: true... but it might be safer to stash my stuff then checkout a master branch and recommit.	20:21
Scotto_	thiago_home: how about just the commits, so i can then list all the files under a certain commit	20:22
	← timj left	20:23
	← magn3ts left	20:23
	→ magn3ts joined	20:23
	← rkenned left	20:23
thiago_home	git diff --name-status commit^..commit	20:23
	→ sjain joined	20:23
	→ nanoo joined	20:24
	→ pklingem joined	20:24
Scotto_	my window buffer can only hold so much :(	20:24
	→ jlilly joined	20:24
Scotto_	whatchanged should display in the opposite order, so that the most recent commit ends up displaying last	20:25
	→ atealtha joined	20:25
	← altogether left	20:25
	→ mjf joined	20:26
	← steelwil left	20:26
thiago_home	--reverse	20:26
	→ Swimming_Bird joined	20:26
atealtha	I'm having issues doing 'git clone ssh://git@localhost:1234/blah.git' even though I can clone from a remote host: 'git clone ssh://git@remotehost...' I can ssh inside the box itself too. anyone know what might be the issue here?	20:26
	← weary left	20:27
atealtha	oh well I can't ssh into itself because of gitosis, but it still responds	20:27
	→ weary joined	20:27
	→ jaql joined	20:27
	← dmgr left	20:29
	→ dmgr joined	20:30
	← meanburrito920 left	20:30
	→ hobodave joined	20:31
	← kpreid_ left	20:31
	← AkumaStreak left	20:32
	→ AkumaStreak_ joined	20:32
	← beni- left	20:33
Ilari	atealtha: What does 'ssh -p 1234 git@localhost' output (the error message)?	20:33
	→ meanburrito920 joined	20:33
atealtha	Ilari: http://nopaste.voric.com/paste.php?f=zu3wvj	20:34
	← magn3ts left	20:34
Ilari	atealtha: And what does 'git ls-remote ssh://git@localhost:1234/blah.git' give?	20:35
atealtha	ssh: localhost:15556: Name or service not known	20:35
	→ josh0x0 joined	20:35
atealtha	fatal: The remote end hung up unexpectedly	20:36
Ilari	atealtha: 'echo $GIT_SSH' (it may be blank)?	20:36
atealtha	yes it is blank	20:36
	→ timj joined	20:37
Ilari	atealtha: Looks like git->ssh invocation is done incorrectly...	20:37
atealtha	in the box I tried remotely, $GIT_SSH is also blank	20:38
Ilari	atealtha: Blank means default (ssh).	20:38
atealtha	oh	20:38
	so it's something on git's side	20:39
Ilari	atealtha: You could put something like this to ~/.ssh/config: 'host git-localhost' 'hostname localhost' 'port 15556' 'user git'. Then try 'ssh git-localhost'.	20:40
crankharder	how should I set up permissions on /home/git (and the repos) such that my various users (fred, bob, etc..) can clone w/ git clone [email@hidden.address] ?	20:41
atealtha	Ilari: one moment	20:41
crankharder	add them to the git group?	20:41
Ilari	crankharder: Use gitolite? Or add them to git group (and create repos shared).	20:41
	← dmgr left	20:42
atealtha	Ilari: same output from when I tried to SSH manually	20:42
Ilari	atealtha: Then try 'git ls-remote ssh://git-localhost/blah.git'.	20:42
atealtha	ERROR:gitosis.serve.main:Repository read access denied << interesting	20:43
	→ mindworx joined	20:43
atealtha	I just checked my gitosis.conf	20:43
	→ dmgr joined	20:45
	← drizzd left	20:45
	→ deadroot joined	20:46
atealtha	Ilari: uhh, it actually WAS my gitosis.conf that was bad	20:46
	← cilly left	20:46
atealtha	sorry for all the trouble, thanks for the help	20:46
deadroot	hi all	20:47
	i'm having a problem with git am	20:47
	it's saying that my "patch does not apply"	20:47
Ilari	atealtha: There was also some ssh problem...	20:47
deadroot	what should i do?	20:48
atealtha	Ilari: you're right, ugh	20:49
	ls-remote works now though	20:49
	→ Fullmoon joined	20:50
	← codeswing left	20:50
Ilari	atealtha: If ls-remote works, clone should as well.	20:50
atealtha	if I use git-localhost it works, but not git@localhost:15556	20:50
	possibly an old version issue?	20:50
	← jaql left	20:51
Ilari	atealtha: What git version? What OpenSSH version?	20:51
	← iaj left	20:52
	→ iaj joined	20:53
atealtha	git: 1.5.2.2; OpenSSH: < 5.x. so yeah, old ssh	20:53
Ilari	Also, git is old version.	20:54
	How you manage to have that old version now? Even Debian Stable has newer version...	20:54
	← static^ left	20:54
	← iaj left	20:54
atealtha	coughslackwarecough	20:54
mugwump	git turns 5 soon	20:54
	→ iaj joined	20:54
	→ bentob0x joined	20:55
atealtha	I'm just too lazy to upgrade	20:55
	→ yn joined	20:56
atealtha	Ilari: but thanks for the .ssh/config tip, that really helped	20:56
	← Zephyrus left	20:56
	← Danielpk left	20:56
acalbaza_	avar: ahh... i just looked at cherry picking again... looks like it is really the straightforward way to do what i want	20:58
	← sjain left	21:00
	← lucsky left	21:01
	← AlexanderS left	21:01
	← bburhans left	21:01
	← ruphy left	21:01
	← badboy_ left	21:01
	← jakob left	21:01
	→ dirker joined	21:01
	→ jakob` joined	21:01
	→ bburhans joined	21:01
	→ patrikf joined	21:01
	← iaj left	21:01
	→ badboy_ joined	21:01
	→ wereHamster joined	21:01
	← yn left	21:02
	→ charon joined	21:02
	→ iaj joined	21:02
	→ ruphy joined	21:02
	← mindworx left	21:03
	→ jast joined	21:03
deadroot	could my problem be because of a mix of lf and crlf?	21:03
Morasique	deadroot: it could be anything; for whatever reason, the patch you're trying to apply doesn't fit your current HEAD	21:04
	→ daifuku joined	21:04
	← iaj left	21:04
	← xou left	21:04
	→ xou joined	21:04
	← viniciusluiz left	21:05
avar	acalbaza_: yay	21:05
	→ AlexanderS joined	21:05
	← Fullmoon left	21:05
deadroot	Morasique: how can i find out what is the cause?	21:06
daifuku	Q: There's a folder that stays listed by `git status` in the "untracked files" section... even though i checked in (supposedly) all the files in the folder. Any ideas on what could cause this?	21:06
wereHamster	you did not (check in the files)	21:07
deadroot	git apply works with the same patch	21:08
acalbaza_	avar: however... when i try it, it says automatic cherry-pick failed...	21:08
	but not sure why	21:08
	→ kipras joined	21:08
	→ gebi joined	21:09
daifuku	@wereHamster: I did a `git add <foldername>\*` and `git commit` and the files are listed by `git log -1 --name-only`	21:09
avar	you didn't give it a sha as an argument	21:09
	← voodootikigod_ left	21:09
acalbaza_	i did actually.. Automatic cherry-pick failed. After resolving the conflicts,	21:09
	but the files show up an unmerged and "deleted by us: ..."	21:10
	← lhz left	21:10
daifuku	Is it possible to see why git thinks the folder is "untracked" ... i've tried looking for hidden files, etc and didn't see anything	21:10
wereHamster	daifuku: windows?	21:10
daifuku	@wereHamster: yep, windows vista :\|	21:10
	← rgr left	21:11
	→ shingo joined	21:11
	← dforsten left	21:11
	→ mindworx joined	21:11
	← pklingem left	21:11
	→ rgr joined	21:11
	← brizly left	21:13
acalbaza_	hmmm.. i had a merge conflict in one file, but cherry-pick didnt tell me which one. odd that it shows all files as "deleted by us:" i dont understand what that means.	21:13
mugwump	"us" probably means the branch you are merging into	21:14
	← akahn left	21:14
	← Aikawa left	21:14
	→ kpreid_ joined	21:16
	→ Aikawa joined	21:16
	← Bass10 left	21:16
	← ThiefMaster left	21:17
	← rgr left	21:18
	← Paraselene_ left	21:20
traviscline	is it possible to use mergetool when git am fails?	21:22
	it leaves a clean index and working dir	21:22
daifuku	I've done a `git ls-files` in that folder and compared it with the output of a normal directory listing... no differences	21:22
	→ ThiefMaster joined	21:23
	→ tvw joined	21:23
	→ nanooo joined	21:24
doener	traviscline: only when using -3, which needs a) git generated patches and b) the base objects for the patch to exist in your repo	21:26
traviscline	ok cool	21:26
	← chittoor left	21:26
	← nanoo left	21:27
	→ novas0x2a joined	21:27
	← camonz left	21:28
	← maxter left	21:29
	← irc2samus left	21:29
	← webchick left	21:29
	← novas0x2a1 left	21:30
	→ jackieO joined	21:30
	agib → agib\|away	21:30
	adamholt → adamholt_away	21:30
	agib\|away → agib	21:30
	→ rotty joined	21:37
	agib → agib\|away	21:37
	agib\|away → agib	21:37
	→ rgr joined	21:38
rotty	how does one do away with a merge -- i.e. keep the history as is but for omittting the second parent of the merge commit	21:39
	?	21:39
selckin	rebase?	21:39
wereHamster	rotty: merge -s ours ?	21:40
rotty	selckin: rebase gives me conflicts for some reason -- I'd guess this should be possible without conflicts	21:40
	← johnw left	21:41
	← rduplain left	21:41
rotty	wereHamster: I don't think that will do what I want	21:41
	wereHamster: IIUC, this creates a merge commit, while I want to get rid of one (or rather the "merge aspect" of it)	21:41
wereHamster	well, then rebase is what you want	21:42
	← tstclair left	21:42
	← ph^ left	21:43
	← atealtha left	21:43
	→ git__ joined	21:44
git__	when git does cloning, does it download one file at a time?	21:44
wereHamster	no	21:45
selckin	define file	21:45
git__	based on filesystem	21:45
selckin	define one at a time	21:45
git__	it goes fetch a file, download, save to local, fetch a file, download, save to local	21:46
	like that	21:46
wereHamster	clone downloads a single pack (and puts it into .git/objects/pack/)	21:46
selckin	which is basicly a single file, but probably not in the way you think of it	21:46
jtaby	i can freely move files within a git repo, right?	21:46
	← meanburrito920 left	21:46
wereHamster	yes	21:46
git__	a single pack is like a single tar.gz file?	21:47
wereHamster	yes	21:47
git__	then local would tar zxvf	21:47
	if i have a clone a 2GB repo, it's a go or no go?	21:47
rotty	wereHamster: rebase seems to follow both branches; my history looks like this: http://i.imgur.com/NXe57.png	21:47
git__	if it gets interrupted half way?	21:47
	err "if i clone a 2GB ..."	21:47
	→ werdan7 joined	21:47
wereHamster	clone can't be restarted (yet)	21:47
git__	ah	21:48
selckin	so why can't you just ask if you can resume interrupted clones	21:48
git__	no wonder why it's taking forever when I try to clone	21:48
	selckin, hehe :P	21:48
selckin	stupid people	21:48
rotty	wereHamster: and I want to just get rid of the black arrows to/from the [WIP] commit, keeping the history of the manpage branch otherwise intact	21:48
git__	:)	21:48
wereHamster	rotty: git checkout manpage; git rebase master; ?	21:49
rotty	wereHamster: I get conflicts regarding the [WIP] commit if I try that	21:49
wereHamster	then resolve the conflicts	21:49
rotty	there should be I way without having me to do that :-/	21:50
	← nanooo left	21:50
wereHamster	no, there is no way to guarantee that there won't be conflicts	21:50
rotty	wereHamster: I /did/ fix those now, but couldn't git apply the diff between every commit, considering just the master branch?	21:51
	erm, manpage branch	21:51
wereHamster	which diff should git apply first, the WIP commit or the 'ui cmdline split off...'?	21:53
	← ixti left	21:53
wereHamster	there is no guarantee that when you apply one and then the other that there won't be a conflict	21:53
rotty	the split off, then diff between that and the merge result	21:53
wereHamster	ah, so you want to squash those two commits into one?	21:54
rotty	if you put it it that way, yes	21:54
wereHamster	I guess you could use rebase -i and then read-tree the merge result instead of fixing the conflict	21:55
	← acalbaza_ left	21:55
rotty	I guess I'll rebase -i on the result of what I've now	21:56
	→ majd joined	21:56
	← loincloth left	21:58
	← jtaby left	21:58
	← sixteneighty left	21:59
	← robinbowes left	21:59
	← khmarbaise left	21:59
	← hugoxrosa left	22:00
	→ joeconyers joined	22:01
	← Louie` left	22:01
rotty	wereHamster: thanks for your hints! (problem solved now, next time I'll try rebase -i)	22:03
	→ Bass10 joined	22:03
	← PerlJam left	22:03
	← dj_tjerk left	22:03
deadroot	oh screw the history. recommitting one by one	22:04
	thanks	22:04
	← deadroot left	22:04
	← Bass10 left	22:05
	→ robinbowes joined	22:05
	→ Bass10 joined	22:05
	→ sagsousuke joined	22:06
	→ psoo joined	22:06
	→ EricInBNE joined	22:07
	→ altogether joined	22:08
	← pantsman left	22:08
	← AkumaStreak_ left	22:08
	← pano_ left	22:09
	← psoo left	22:09
	← majd left	22:11
	→ rook2pawn joined	22:12
rook2pawn	iS there a way to turn off compression on certain files when sending up to the repo?	22:12
	→ Yuuhi joined	22:12
	← lucs left	22:13
wereHamster	I think git will always compress objects with zlib, but you can tell git to not try certain things (like diff, delta compression) on files by setting gitattributes	22:14
	← Pupeno left	22:15
	→ meanburrito920 joined	22:18
	← meanburrito920 left	22:18
	→ meanburrito920 joined	22:18
	← dmgr left	22:18
	← sagsousuke left	22:18
Ilari	The delta compression is the one that can really eat CPU time...	22:19
	→ sjain joined	22:19
rook2pawn	How do i set core.compression or pack.window to 0 ? I'd ideally like this just to affect this one giant .rar file we keep	22:20
Ilari	rook2pawn: '*.rar -delta' to .git/info/attributes (or what it was...)?	22:21
rook2pawn	-delta means delta only?	22:22
Ilari	rook2pawn: Disable delta compression on matching files.	22:22
rook2pawn	Oh okay. Is there a way to turn off zlib as well?	22:22
Ilari	rook2pawn: No there isn't.	22:23
	→ jstephenson joined	22:23
	→ flaguy48 joined	22:23
	← ehabkost left	22:24
wereHamster	the real question is, why are you trying to add rar files to a git repo?	22:25
	→ o_a joined	22:26
rook2pawn	we have a collection of .png's but there are tens and thousands of them. sending them to git took so long, the transfers just died somewhere in the 2 -3 hour mark	22:27
	← Theravadan left	22:27
rook2pawn	so i turned off compression on .rar, added the recovery record index, and decided this is better than .tar, sent it up, and it worked, about 15 minutes	22:27
	→ bauruine joined	22:31
	← agile left	22:32
	← _iron left	22:33
	← sjain left	22:33
	← robinbowes left	22:33
	← altogether left	22:35
	→ manyoso joined	22:35
	← thiago_home left	22:35
manyoso	anyone know of a quick git command to tell whether to repo's share any common objects?	22:36
	should be, "to tell whether two repo's..."	22:36
	→ Rick_ joined	22:38
	→ robinbowes joined	22:38
	→ tek0 joined	22:38
wereHamster	fetch one repo into the other, check if any two branches have a common merge-base	22:38
Textmode	night all	22:38
Rick_	I'm new to git and I have a question about a workflow model. What would be the best way to have both a free version and pay for version of an application in one repository?	22:39
manyoso	wereHamster: faster to just iterate over objects dir and compare?	22:39
	← Textmode left	22:39
	→ fool__ joined	22:39
	← fool__ left	22:39
	→ fool__ joined	22:39
wereHamster	manyoso: maybe..	22:39
	Rick_: have two branches, one for the free, one for the pay version	22:40
Rick_	wereHamster: Should I develop in the free branch, and then merge changes to the paid branch for changes that would apply to both?	22:40
	wereHamster: Also, how should I handle tagging version numbers then?	22:40
wereHamster	does the free version contain things that the paid version doesn't?	22:41
	→ sgh_ joined	22:41
	← jrmuizel left	22:41
Rick_	It will contain very small number of things that the paid doesn't like for example, a screen about the paid version	22:41
	← sgh left	22:41
wereHamster	then you need three branches, one common where you develop and merge into the free and paid branches	22:42
	→ SpaT joined	22:42
Rick_	so the main develop branch should be the least common denominator	22:42
	correct?	22:42
wereHamster	yes	22:42
	← fool_ left	22:42
Rick_	For a tagging system should i do something like git tag -a paid-v1.0 -m 'version 1.0 (paid)" ?	22:42
wereHamster	yes, I'd tag the paid and free versions separately	22:43
SpaT	what happens if i have modified files in the workcopy and switch branches?	22:43
Rick_	Would you say there is a naming convention thats generally accepted for that already or not?	22:43
	→ WALoeIII joined	22:43
wereHamster	most projects use rX.Y or vX.Y for tags, but what you proposed looks fine	22:44
Rick_	Thanks wereHamster , I really appreciate it	22:44
frogonwheels	SpaT: if it has to switch to a different revision, it wont let you .you need to stash or commit	22:44
wereHamster	you could even use git tag -a paid/v1.0 and tag free/v1.0 if you want to split the tags into different namespaces, but that's entirely up to you	22:45
Rick_	How do the namespaces apply to git specifically?	22:45
	← tobiassjosten left	22:45
SpaT	frogonwheels: thought it would do something sensible like that thanks	22:46
wereHamster	Rick_: it's like directories	22:46
Rick_	alright I think I will do it that way	22:46
frogonwheels	SpaT: in the case where you've made changes, want to branch before you commit, you can happily create a new branch where you are, check it out and commit..	22:47
wereHamster	Rick_: branches live in refs/heads, tags in refs/tags, remote braches in refs/remotes (run git show-ref in your repo and look at the output)	22:47
Rick_	I see, so then it would split the tags into two seperate subdirectories?	22:47
wereHamster	internally, yes	22:47
	← Aikawa left	22:47
Rick_	Works for me, thanks for the tip	22:48
	→ Pupeno joined	22:48
	→ Aikawa joined	22:48
wereHamster	it's mostly just for convenience (so you could push only the free tags with a single command, git push <...> refs/tags/free/*)	22:48
Rick_	Oh yes, that makes sense now	22:48
	→ pklingem joined	22:49
	← Aikawa left	22:50
	→ ntoll joined	22:51
	← Ryback_ left	22:51
	→ Mage joined	22:51
	→ Ryback_ joined	22:52
	← sgh_ left	22:52
	← simNIX left	22:52
	← Mage left	22:52
	← alley_cat left	22:52
	agib → agib\|away	22:53
	→ Mage joined	22:53
	shingo → aikawa	22:56
	← gebi left	22:56
	← git__ left	22:57
	← Ryback_ left	22:57
	→ mike8901 joined	22:59
	→ mykeus joined	22:59
mike8901	If I checkout an old commit, how can I get git log to show _ALL_ commits, including ones after the current one?	22:59
	← tvw left	22:59
bob2	git log thebranch	23:00
	← unhygienix left	23:02
	← aziz left	23:02
	→ unhygienix joined	23:02
	← d0k left	23:02
SpaT	frogonwheels: that is the most likely case I will use it. Always refactoring and messing up a already working copy	23:03
mugwump	mike8901: see man git-rev-list	23:04
Gitbot	mike8901: the git-rev-list manpage can be found at http://git.or.cz/man/git-rev-list	23:04
mugwump	there is --all if you want all commits on all branches	23:04
	← drd_ left	23:04
	← SpaT left	23:06
	← unhygienix left	23:07
	← patmaddox left	23:07
	← oal left	23:08
	← Yuuhi left	23:10
	→ shingo joined	23:10
	← kipras left	23:11
	← jstephenson left	23:11
	← Chillance left	23:11
	→ agile joined	23:13
	← SimManiac left	23:15
	→ pax- joined	23:17
pax-	so.. I commit'ed and now want's to push, git is saying "already up-to-date". Easyest way to solve?	23:18
	← mattions left	23:18
	← cedricv left	23:19
wereHamster	tell git what to push (git push <remote> <branch>)	23:19
	← icwiener left	23:19
	→ auslander joined	23:19
	← auslander left	23:19
	→ auslander joined	23:19
pax-	"everything up-to-date" when I do git push origin master	23:20
	oh wait	23:20
wereHamster	git doesn't lie to you...	23:20
	pax- facepalms	23:20
pax-	I was working on another branch, thanks anyway	23:20
	→ rburton- joined	23:21
	← ntoll left	23:21
	← khelll left	23:22
	→ ben_h joined	23:23
	← AAABeef left	23:24
	→ loincloth joined	23:25
	← alester left	23:25
	← mithro left	23:25
	→ gasull joined	23:27
	→ d0ugal joined	23:27
	← auslander left	23:28
gasull	Hi all. How can I remove a file with sensitive information from the git repo removing previous versions too? Thank you.	23:28
wereHamster	gasull: man git-filter-branch	23:28
Gitbot	gasull: the git-filter-branch manpage can be found at http://git.or.cz/man/git-filter-branch	23:28
	← Eaven left	23:28
	← frogonwheels left	23:30
	→ madewokherd joined	23:30
	→ frogonwheels joined	23:30
	→ juan_arandaalvar joined	23:32
Rick_	wereHamster: Is there an easy way to rename old annotated tags?	23:32
mugwump	you can use git update-ref refs/tags/newname oldname^{tag}	23:33
Rick_	Is ther ea way to change the message associated with the tag?	23:34
wereHamster	no	23:34
mugwump	http://utsl.gen.nz/git/git-amend will do ti	23:34
	however the tools make this deliberately difficult	23:34
Rick_	Basically I have a bunch of old version numbers but now my client wants a free and a paid app so i want to chnage all the old tags to be paid/oldtaghere	23:35
mugwump	clients for instance will not overwrite tags they already have when pulling	23:35
Rick_	maybe ill just delete all the old tags since they were beta builds anyway	23:35
	← bauruine left	23:35
	→ warlock_mza joined	23:43
	← floppyears left	23:44
	→ pheaver joined	23:45
	→ floppyears joined	23:45
gasull	wereHamster: thanks. Maybe my question should be "how can I remove a file with sensitive information from a remote git repo?"	23:46
wereHamster	gasull: fetch the remote repo, run filter-branch, push the result back	23:46
	← aryounce left	23:47
	← dpino left	23:49
	→ magn3ts joined	23:49
gasull	wereHamster: I get the error "A previous backup already exists in refs/original/	23:50
	Force overwriting the backup with -f". But I don't know where -f should go. I get errors wherever in the command "git filter-branch --tree-filter 'rm fabfile.py' HEAD" I put the -f.	23:50
wereHamster	that's an option to git-filter-branch (so it goes right after the command)	23:51
	← AAA_awright left	23:51
	→ patmaddox joined	23:51
gasull	wereHamster: $ git filter-branch -f --tree-filter 'rm fabfile.py' HEAD	23:52
	Rewrite c64e434ea03602d8371dc20cee77bb135e2ce54f (1/14)rm: cannot remove `fabfile.py': No such file or directory	23:52
	tree filter failed: rm fabfile.py	23:52
wereHamster	also, the man page has an example how to remote a file from the history, and you're not doing it like there described	23:52
	(and that's why it fails, btw)	23:52
gasull	wereHamster: can't find the word "remote" in that link	23:52
wereHamster	remote? Why would you look for that?	23:52
	← aikawa left	23:52
	← shingo left	23:52
gasull	wereHamster: mmm... because it's a remote repo?	23:53
	→ altogether joined	23:53
gasull	maybe my concept of "remote" is wrong in git jargon. But my understanding it's that it is a remote repo since it isn't in my laptop	23:54
wereHamster	is it a bare repo?	23:54
	→ AAA_awright joined	23:54
	→ unhygienix joined	23:55
	→ robinr joined	23:55
	← LiamH left	23:55
gasull	wereHamster: I'm not sure what "bare repo" means. If you mean if it's new, it's almost new. I committed there a few times.	23:55
	← malumalu left	23:55
frogonwheels	gasull: you looking at it over a file share? or didyou 'clone' it from the original	23:56
mugwump	gasull: clean it up in your own repository. push to the remote one. log into the remote one, clear logs, and run 'git repack -a -d' and 'git prune'	23:56
wereHamster	well, I have absolutely no idea which repos you have, in which you work and what you expect.	23:56
	I can, however, tell you how to remove a file from the history (or rather, the git-filter-branch man page will tell you that)	23:56
gasull	frogonwheels: the original repo is the one in my laptop. I didn't clone it from the remote	23:56
	← giallu left	23:57
wereHamster	you go into the repo which has the commits, you run the filter-branch command (like described in the man page), then you remove all refs in the refs/original namespace, done	23:57
	← Swimming_Bird left	23:57
gasull	mugwump, wereHamster: the remote repository is in http://www.projectlocker.com, so I cannot ssh into it. I hoped to be able to fix this running some git command locally.	23:58
	← rook2pawn left	23:58
wereHamster	then clone the repo, run the filter-branch command, and push the result back	23:58
	→ naturalethic joined	23:59
mugwump	well, clean it up, push - the last steps involving logging in are if you want to be really really sure	23:59
	but automatic GC will get rid of them eventually	23:59
	they will be hard to find. depends on how sensitive the information is, really	23:59
	← pax- left	23:59

Logs Search ←Prev date Next date→

Channels Documentation