IRCloggy #git 2008-09-19

Provider of IRC logs since 2005.

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

	2008-09-19
	← ciaran left	00:00
pcc1	so once I have the output of git show, how can I reimport it?	00:00
	→ christophsturm joined	00:01
mugwump	git-am or git-apply	00:01
	another way to do that is to point and click with gitk	00:02
pcc1	mugwump: thanks	00:02
mugwump	use: gitk rev1..rev2 (ie, your rev-list options from before), then right click the commits you want and hit 'cherry pick this commit'	00:02
	or, git-cherry-pick commitid is a command-line way	00:03
	or git-rebase -i if you want to do it en masse	00:03
	← mithro left	00:03
	← nbrosnahan left	00:07
	→ mithro joined	00:08
	→ zachinglis joined	00:09
	loinclot_ → loincloth	00:09
grumpy	mug: thx for the previous answer, do you know about a git ammend on the remote repo?	00:10
	answers	00:10
	← bdiego left	00:11
	← jasam left	00:14
mugwump	grumpy: you need to use git push -f	00:15
	ie, amend it locally then push	00:15
	if that doesn't work, you can push a delete and then re-push the branch	00:15
	:->	00:15
	→ krawek joined	00:16
	→ tzafrir_laptop joined	00:17
	← Beket left	00:18
tzafrir_laptop	is it possible to convert a repo.or.cz repo from "mirrored" to "push"?	00:18
	I recall something about it in the setup page, but can't find any further reference	00:19
	← cilly left	00:19
	← tongueroo left	00:20
pcc1	mugwump: does git pack-objects and unpack-objects also do what I want?	00:21
	the files it generated were rather small	00:22
mugwump	well, I guess that depends what you want! :)	00:22
	← krawek left	00:22
pcc1	I need to back up only commits that exist within a branch but not in the mainline, so I can restore to the exact same state	00:23
	← timcharper left	00:23
pcc1	(assuming access to mainline)	00:24
mutex	Is there a way to only cherry pick commits to a particular file between branches ?	00:27
	lets say commits to file A were included in commit ID's with changes to several other files	00:27
	but I only want the changes to file A to cherry-pick	00:27
parasti	tzafrir_laptop: it basically says there's current no way to do that, and you have to recreate the repo	00:33
	s/current/currently/	00:33
tzafrir_laptop	parasti, ok	00:34
	thanks	00:34
grumpy	mug: thanks, I'll try it :)	00:35
	→ Flathead joined	00:36
	← elmex left	00:36
	→ BarryCarlyon joined	00:37
	← parasti left	00:48
	k-s → k-s[AWAY]	00:51
	← drizzd left	00:54
	→ capisce_ joined	00:54
	← zachinglis left	00:59
	← capisce left	01:05
	→ vintrepid joined	01:05
	→ VVSiz_ joined	01:05
	← vintrepid left	01:07
	→ capisce joined	01:07
	→ vintrepid joined	01:07
	→ yofel_ joined	01:08
	← whee left	01:11
	← perezd88_ left	01:11
	→ CpuWhiz joined	01:15
grumpy	I'm using git as a frontend to perforce; but due to the sync with p4, I have the library of the different project in the git repository.	01:18
	→ kanru joined	01:19
grumpy	every time I compile the library are shown as modified...	01:19
	← WALoeIII left	01:19
grumpy	which is quite annoying... Is there a way to tell git to ignore those file even if they are in the database?	01:19
Eridius	grumpy: you can tell it to assume a given file is unchanged and never actually check	01:20
	→ tongueroo joined	01:20
Eridius	but the real question is, why are you versioning build results?	01:20
	incidentally, the way you tell it to never actually check if a file is modified is with git update-index --assume-unchanged	01:21
	check the manpage	01:21
mugwump	grumpy: are you looking for .git/info/exclude perhaps?	01:22
Eridius	mugwump: I think he was saying the file is versioned	01:22
	otherwise it wouldn't be "modified"	01:22
	it would be "there"	01:22
	← tongueroo left	01:22
grumpy	eridius: it's a company choice :) if we have the library in p4	01:22
	← capisce_ left	01:23
	← VVSiz left	01:23
mugwump	there might be a post-checkout hook, check Documentation/hooks.txt	01:23
grumpy	that is correct, with the p4 import, the file is in the repository... modifying .gitignore for example doesn't affect it	01:23
Eridius	mugwump: and what would that be fore?	01:23
mugwump	ie, blow the library away in the hook	01:23
Eridius	grumpy: just use git update-index --assume-unchanged like I said	01:24
	mugwump: that's not what he wants to do	01:24
mugwump	ah ok	01:24
	→ jesselucas joined	01:24
Eridius	grumpy: if you mark it as assume unchanged, then git will never tell you it's modified. You can still `git add` it if you want to commit a new version	01:24
	but `git status` will never complain about it	01:24
	← jesselucas left	01:24
	← kukks left	01:24
	→ xinming joined	01:25
	← yofel left	01:25
grumpy	what will happened when I will sync with p4 and will it import a new version of these file? if it does, do I have to rerun the --assume-unchanged commands?	01:25
	← christophsturm left	01:26
Eridius	I actually don't know. My guess is no, but you're just going to have to find out yourself	01:26
	the assume unchanged just marks it as such in the index. It's not a versioned property	01:26
	→ _VVSiz_ joined	01:26
grumpy	in case this create issue, what do I need to do to revert the process?	01:27
mugwump	git-add	01:27
Eridius	mugwump: err no	01:29
	grumpy: git update-index --no-assume-unchanged	01:29
mugwump	thanks	01:29
	mugwump assumed	01:29
neurodamage	how do I setup git email commits, is there any good documentation on that?	01:30
	→ jesselucas joined	01:30
grumpy	thanks a lot eridius, mugwump	01:31
mugwump	neurodamage: see contrib/hooks/post-receive-email	01:32
jesselucas	is there any way to have submodule retain a directory structure? Say I want Super Project > Client Name > Project 1.git . Then the Super Project would keep the folder structure when I cloned it.	01:32
	← irotas left	01:32
Eridius	jesselucas: err... what?	01:32
	← kanru left	01:33
	← joydivider left	01:34
jesselucas	Eridius: I'm confused too :) . Basically I have a folder called Projects, with multiple folders for each client, and each client has many projects. I'd like to make a each project it's own repo, but retain the client folder.	01:35
neurodamage	mugwump: where do I "see" that? I looked on google it's not got solid documentation on there, I don't think.	01:35
grumpy	oh yes, I have a final question, let's say you have a project with files in the repository A and B, and another project with file in repository C and D. A, B, C, and D are at the same level. Is it possible to have 2 git repository, 1 for each project?	01:35
Eridius	jesselucas: ok, sounds fine	01:35
	what's the problem?	01:35
	neurodamage: he gave you a path in the git.git repo	01:35
	grumpy: you could try using submodules to include A and B into a project repo	01:36
jesselucas	Eridius: I'm not setting up submodules correctly. When I git submodule add it creates the project folder in the Super Project directory.	01:36
Eridius	jesselucas: then you're not running the command properly	01:36
	git submodule add repo path	01:36
	the path arg is wrong	01:37
grumpy	mmm ok i ll look into that	01:37
	← xinming_ left	01:37
jesselucas	grumpy: here is a tutorial I've been using: http://git.or.cz/gitwiki/GitSubmoduleTutorial	01:37
Eridius	jesselucas: try reading the manpage for git-submodule	01:38
grumpy	does the git repository name have to be .git? (I was thinking about having a emacs hook which will commit every saved files into a special git repository has backup)	01:38
Eridius	grumpy: err, what?	01:38
	← mw left	01:38
Eridius	bare repos are named foo.git by convention	01:38
	k-s[AWAY] → k-s	01:38
Eridius	non-bare repos use the dir .git to hold the actual repo	01:39
	technically, you can override this by setting the GIT_DIR env var	01:39
	→ ben_h joined	01:39
grumpy	it does make sens it would have been nice to be able to have 2 git repository under the same folder	01:39
	oh? that could be useful then	01:39
Eridius	err, no it doesn't	01:39
	you can't make 2 git repos that share the same work tree	01:39
	← ben_h left	01:39
mugwump	sure you can ;)	01:40
grumpy	eridius: it should be possible, a bit awkward, but possible	01:40
Eridius	mugwump: hush	01:40
mugwump	just be prepared to learn a lot about how git works ;)	01:40
Eridius	grumpy: it's technically possible, but it's ridiculous, and git'll constantly think you have a bunch of untracked files	01:40
grumpy	eridius: i agree every time u change a branch in one repo, it will mess up the other one ;-)	01:41
	alright, I have to go :-) Thanks you very much for all the help. It was very useful. (--assume-unchanged worked great so far, I ll have to check the result after another p4 import)	01:42
jesselucas	Eridius: Can the git repos I want to be submodules be direct children of the superproject?	01:42
Eridius	jesselucas: you can put the submodules anywhere you want in the repo	01:42
grumpy	Have a good evening. Cya	01:42
	← grumpy left	01:43
	← VVSiz_ left	01:43
	→ gretch_ joined	01:43
	gretch_ → gretchen	01:43
	← rmh3093 left	01:43
	→ xinming_ joined	01:47
	← Jacolyte left	01:49
	← vintrepid left	01:55
neurodamage	Eridius: are there scripts already written for this, that would be nice to have instead of having to write all of this, I'd like to have emails with diffs sent to a maillist?	01:55
	→ jnareb joined	01:55
Eridius	neurodamage: did you look at what mugwump suggested?	01:55
	jnareb is ShadeHawk is jnareb is JakubNarebski is Jakub Narębski	01:56
jnareb	finally I have took Git User's Survey 2008 (http://tinyurl.com/GitSurvey2008)	01:56
	quite a bit of time since it was opened...	01:56
neurodamage	Eridius: I missed it and still don't see it?	01:56
	← sgrimm left	01:56
Eridius	neurodamage: did you look where I told you? in the git.git repo?	01:57
neurodamage	Eridius: oh wait, yeah I saw that...but that means I have to write a script for the email process, I'm on archlinux and the .sample I have is empty with one line to uncomment that's a link to a location that doesn't exist :)	01:58
	Eridius: yeah in .git/hooks/post-recieve.sample	01:58
	is where I found it	01:58
Eridius	neurodamage: did you look at the script mugwump told you?	01:58
	that's not what mugwump said	01:58
neurodamage	Eridius: yes	01:58
Eridius	try paying attention	01:58
	no, you didn't	01:58
	he gave you a path which you ignored	01:58
jnareb	Hmmmm... it looks like git doesn't install any contrib...	01:58
Eridius	no, it doesn't install it	01:58
	it's in the git.git repo	01:58
neurodamage	Eridius: right...	01:59
	Eridius: I know that, here i'll paste what I mean ...	01:59
Eridius	neurodamage: the dir contrib/ exists in the git.git repo	01:59
	← xinming left	01:59
neurodamage	http://pastebin.com/d2ad1f666, Eridius	02:00
Eridius	you're not paying any attention to what I'm saying	02:00
neurodamage	that's the post-receive in my .git/hooks, there's not .git/contrib directory in my repo	02:00
Eridius	neurodamage: you're completely ignoring what I'm saying	02:00
	so shut up and listen to me	02:00
neurodamage	ok...	02:00
Eridius	in the git.git repo. This DOES NOT MEAN THE .GIT DIR	02:00
	this means the working tree	02:00
	← jnareb left	02:00
Eridius	there is a dir called contrib/	02:01
	inside there is a dir called hooks	02:01
	inside there is the post-receive-email script	02:01
	ok?	02:01
neurodamage	Eridius: I'm new to git as of today, so I'm not sure I understand what git.git repo is?	02:01
Eridius	foo.git is a convention that means the git repo by the name of "foo"	02:02
	it's called this because the actual URL to the repo ends in foo.git	02:02
	in this case, git.git means the git repo for git itself	02:02
	alternatively, the source tarball for git	02:02
	so you can either clone git://git.kernel.org/pub/scm/git/git.git or just look at the source tarball	02:02
neurodamage	oh ok, well now I understand, ok	02:03
jesselucas	Eridius: The path argument worked great. I'm only having trouble now running the update submodule command from the cloned superproject. Here is a clip of the terminal http://www.pastie.org/275484	02:03
Eridius	jesselucas: did you read the documentation?	02:04
jesselucas	Eridius: yes	02:04
Eridius	you told it that the URL for the repo is "Projects/Barefoot/PointRoll"	02:04
jesselucas	yep	02:04
Eridius	what do you think is wrong with that?	02:05
neurodamage	Eridius: I'm downloading now to try and understand by looking at kernel repo, thanks for some direction	02:05
jesselucas	gotcha. it needs to be absolute. Thanks man!	02:05
	→ ben_h joined	02:05
	→ melvinram joined	02:07
melvinram	hey folks, I've got a git repo on my local machine n github	02:07
	→ rmh3093 joined	02:07
melvinram	the branch i'm on locally doesn't exist on github ... how do I push that branch?	02:08
	when i do "git push origin enrollment" it gives me an error	02:08
	! [rejected] enrollment -> enrollment (non-fast forward) error: failed to push some refs to 'git@github.com:melvinram/brainbank.git'	02:09
	thats the error i get	02:09
	_nn → nn	02:09
melvinram	any ideas?	02:09
ToxicFrog	That seems to imply that it is on github	02:10
jesselucas	melvinram: I just starting using git today and still learning, but I ran across this screencast today http://www.gitcasts.com/posts/empty-branches . Might be able to help with branches	02:10
ToxicFrog	But that there are commits in enrollment there that are not in your local branch	02:10
	Eg, you clone it	02:11
	You make a bunch of commits to enrollment	02:11
	Meanwhile, someone else does the same and pushes	02:11
	Is this a possible scenario in your situation?	02:11
melvinram	Not really	02:11
	Toxic, i'm the only one developing on this entire repo	02:12
	both locally and on github	02:12
ToxicFrog	Oh.	02:12
	And just "git push" skips enrollment entirely?	02:12
melvinram	well i didn't want to take the chance that it would merge it into master	02:13
	so i didn't do it	02:13
	→ vintrepid joined	02:13
ToxicFrog	Push will never cause merges, AFAIK	02:13
	Anyways, the command you're using is correct - "git push <remote> <new-branch>	02:14
	You're sure enrollment can't possibly already exist in the origin?	02:15
	← melvinram left	02:17
	← Man-erg left	02:19
	→ melvinram joined	02:20
	→ bryanl joined	02:20
melvinram	sorry about that... i had a file i was trying to delete and had to restart to figure which program was still holding on to it	02:20
	← jatt left	02:21
	→ imyousuf joined	02:22
melvinram	is there a way to do a pull and have a look at the changes you're pulling before it befores everything?	02:22
neurodamage	Eridius: so I'm going to have to link to git.git repo and keep that source somewhere?	02:23
Eridius	neurodamage: not a bad idea	02:23
ToxicFrog	"before it befores everything"?	02:25
	Before it merges, you mean?	02:25
	git fetch	02:25
neurodamage	Eridius: excellent, cause I was reading it wondering what all the scripts it was relying on were :)	02:25
ToxicFrog	pull is basically git fetch && git merge	02:25
	fetch updates your remote tracking branches, merge merges them into your local branches	02:26
	So you can, say:	02:26
	$ git fetch # update remote tracking branches	02:26
	$ git checkout origin/master # examine remote changes	02:26
	$ git checkout master	02:26
melvinram	how do I delete a branch from github?	02:26
ToxicFrog	$ git merge origin/master # they look good, merge them into line of development	02:26
	→ threeve joined	02:27
neurodamage	Eridius: I see a Config section, it's mentioning hooks.mailinglist etc...are those variables I need to set somewhere?	02:27
ToxicFrog	git push repo :branch	02:27
	melvinram: the man page has the details, but the basic structure is git push remote from:to	02:27
Eridius	neurodamage: yeah, `git config hooks.mailinglist foo`	02:27
	→ xinming joined	02:27
ToxicFrog	to defaults to from, so git push origin master -> git push origin master:master	02:27
	And if you omit the from and just have :to, it deletes it	02:27
	← fade-in left	02:29
CpuWhiz	question: if i add some line to a branch commit, and remove the line in a later commit, then git merge --squash --no-commit master...would the line still be there somewhere and pushed to origin for all to see?	02:29
	and would you have to filter it out somehow or would that solve the issue	02:30
neurodamage	Eridius: how would I set the envelopesender under the hooks in the .git/config to use whatever user is performing the commit?	02:30
Eridius	CpuWhiz: if you squash it, it's not doing a real merge	02:30
	→ capisce_ joined	02:30
melvinram	ToxicFrog: That is a normal push... i need something to delete a branch at an remote server	02:31
Eridius	neurodamage: edit hte script	02:31
	line 628	02:31
CpuWhiz	so if i was to delete the other branch after the merge there would be no trace of the line?	02:31
neurodamage	Eridius: ok cheers, that's all I needed to know :), that's usually the right answer	02:31
Eridius	CpuWhiz: right	02:31
melvinram	I want to delete branch "course" from origin	02:31
Eridius	git push origin :course	02:32
melvinram	how would i do that?	02:32
Eridius	think of it like you're pushing "no reference" to overwrite the remote "course"	02:32
CpuWhiz	so i don't have to worry about my stupid mistake ^^ thanks Eridius	02:32
melvinram	hmm	02:32
neurodamage	Eridius: what takes precedence ~/.gitconfig or repo's .git/config?	02:33
melvinram	hey that worked!	02:33
	← jerbear left	02:33
neurodamage	Eridius: or are they seperate?	02:33
ToxicFrog	melvinram: like I said	02:33
melvinram	Eridius: thanks!	02:33
Eridius	neurodamage: what does the git-config manpage say?	02:33
	← yofel_ left	02:34
neurodamage	Eridius: indeed you have me at RTFM, :), thanks again	02:34
	→ yofel_ joined	02:34
melvinram	ToxicFrog: Yea but i wasn't able to figure it out... i'm a git noob :D	02:35
neurodamage	Eridius: you have seriously sped along my learning, seriously though, thanks	02:35
melvinram	ToxicFrog: Thanks for your help as well	02:35
Eridius	I'm glad you'r elearning	02:35
neurodamage	Eridius: me too, I've been wanting to switch from CVS a long time ago	02:35
Eridius	CVS? shudder	02:36
neurodamage	oh yeah...I've had issues with it, but I also have written elaborate scripts, guess I'll just have to "port" them over	02:36
melvinram	Okay if i'm the master branch, and i pull from a remote server and it fetches the additional branches, how do I create a new branch locally with the remote branch data?	02:39
	is it "git checkout -b newbranch origin/newbranch"?	02:39
	← xinming_ left	02:39
ToxicFrog	melvinram: close; add a --track there so that pull will handle it properly	02:41
	You can also do it in two steps:	02:41
	git branch --track foo origin/foo	02:41
	git checkout foo	02:41
	← bryanl left	02:41
	← neurodamage left	02:41
ToxicFrog	See also: man git-branch, man git-checkout	02:41
	← capisce left	02:41
Gitbot	ToxicFrog: the git-branch manpage can be found at http://git.or.cz/man/git-branch	02:41
ToxicFrog	Thank you, gitbot.	02:42
	→ neurodamage joined	02:42
melvinram	ToxicFrog: I think it automatically assumed the --trace	02:42
	ToxicFrog: because i did it without it and it worked	02:42
ToxicFrog	track, not trace, and pull worked too?	02:43
	--track isn't necessary to actually create or check out the branch; what it does is update the configuration so that "git pull" will update the new branch along with the rest.	02:44
melvinram	oops, yea thats waht i meant	02:44
	it gave me this output	02:44
	Branch EnrollMe set up to track remote branch refs/remotes/origin/EnrollMe.	02:44
	Switched to a new branch "EnrollMe"	02:44
tpope	--track is the default now	02:44
	→ capisce joined	02:45
ToxicFrog	tpope: aah? Cool.	02:46
	→ xinming_ joined	02:49
	→ ttuttle joined	02:50
ttuttle	Hey.	02:50
	I installed git from the 1.6.0 tarball onto a server, and set up a git repos. When I try to clone it from another box, I get "git-upload-pack: Command not found." No surprise, git didn't install that command. How do I fix this?	02:51
	← tzafrir_laptop left	02:51
mugwump	that's an ssh oddity, you need to make an /etc/ssh/environment file	02:51
ttuttle	I can't control /etc/ssh/environment. Where else can I put it?	02:51
	(I've just got an account. git is installed in my homedir.)	02:52
mugwump	~/.ssh/environment	02:52
	and put your PATH in there	02:52
ttuttle	mugwump: What is the actual problem? Oh.	02:52
mugwump	or maybe symlink ~/.ssh/rc to your bashrc	02:52
ttuttle	Ah.	02:52
	← capisce_ left	02:52
mugwump	oh, actually scratch that last idea	02:53
	but the environment one should work	02:53
ttuttle	Oh :(	02:53
	So can I do something like PATH=$PATH:blah?	02:53
mugwump	I'd guess not	02:53
	the default is really dumb, like '/usr/bin:/bin'	02:53
	hmm, yet, my own .bashrc gets sourced when I: ssh somehost 'echo $PATH'	02:54
	→ sgrimm joined	02:54
	mugwump is confused now	02:55
ttuttle	Yeah, same here.	02:55
	Hold on.	02:55
	There actually isn't a git-upload-pack command.	02:55
	There's git-upload-archive only.	02:56
mugwump	I can only speculate as to why	02:56
	maybe it didn't 'make install' fully ?	02:56
	I really doubt that command has gone away	02:57
ttuttle	Nah, the C file is there.	02:57
	→ xinming__ joined	02:58
ttuttle	Okay, I just compiled and installed git-1.6.0.2 using the instructions in INSTALL, and there's no git-upload-pack. WTF?	03:00
	Why is this not working?	03:01
	→ offby1` joined	03:02
	← xinming left	03:02
	→ synsol joined	03:03
	← melvinram left	03:04
mugwump	ttuttle: does the file exist in your build directory?	03:04
ttuttle	Yes.	03:05
	It just doesn't get installed.	03:05
	Also, WTF is $X in the Makefile?	03:05
	← kef left	03:08
mugwump	that will be .exe on windows	03:12
	cygwin	03:12
	→ zumbrunn_ joined	03:12
mugwump	and mingw	03:12
	can you nopaste your 'make install' output?	03:12
	ttuttle is not liking git.	03:13
	← rdavila left	03:13
mugwump	yeah, nobody likes a learning curvbe	03:13
ttuttle	mugwump: Eh, what can I say.	03:13
mugwump	but don't worry, there's a good view from the top	03:13
ttuttle	mugwump: I'm sure git is great. It's a shame the interface is inscrutiable to newbies.	03:14
	mugwump: So none of the programs in PROGRAMS seem to be installed.	03:14
	← xinming_ left	03:14
	→ bogolisk joined	03:15
mugwump	ah. perhaps it's in gitexecdir	03:16
ttuttle	?	03:16
bogolisk	ne1 ever try git reflog with an annotated tag? I don't know how to read the output!	03:16
ttuttle	mugwump: you mean $PREFIX/libexec/	03:16
	?	03:16
mugwump	right	03:16
ttuttle	Nope.	03:17
	← offby1 left	03:17
ttuttle	That's where all the other commands went.	03:17
mugwump	that's really strange, first time I've heard of that happening	03:18
	ttuttle curses the jackasses who run my school's server for not providing git.	03:19
	→ timcharper joined	03:19
	← eternaleye_ left	03:20
	→ eternaleye_ joined	03:20
	← imyousuf left	03:20
	← zumbrunn left	03:20
	← bogolisk left	03:21
MaddHatter	ttuttle: what school (just curious)?	03:21
ttuttle	Ah.	03:21
	I found the problem.	03:21
	install is broken.	03:22
	← madewokherd left	03:22
ttuttle	It only installs the first argument.	03:22
	→ FunkeeMonk joined	03:22
ttuttle	Yay, it's working.	03:24
	Expect me to be back with more questions over time ;-)	03:24
MaddHatter	Is tehre a way to get 'git show' to show renames/copies/moves as such instead of as two diffs against /dev/null?	03:26
	← cbreak left	03:28
	← CpuWhiz left	03:29
	k-s → k-s[AWAY]	03:29
MaddHatter	ah, it does take -M/-C	03:30
	I need to start submitting patches for all these man pages	03:31
	→ yofel__ joined	03:34
	← threeve left	03:34
robbyonrails	hey there	03:41
	how do i go about removing tags on a remote?	03:41
ToxicFrog	Same way as removing a branch, IIRC	03:42
	git push remote :tagname	03:42
	← drewr left	03:43
robbyonrails	we accidently added a remote branch and tag with the same name	03:43
	and it caused some confusion/issues	03:43
	← jaredj left	03:44
	← kyofel left	03:47
robbyonrails	okay, so that deleted the tag	03:47
	→ mediogre joined	03:47
robbyonrails	but when i try to push the branch back it says: error: src refspec staging matches more than one	03:47
	→ foca joined	03:47
	→ kyofel joined	03:49
mugwump	robbyonrails: delete it locally too	03:49
	git tag -d	03:49
robbyonrails	did that	03:49
mugwump	use git show-ref \| grep tagname	03:50
robbyonrails	you're right, ti was still showing... must have came back after i did a fetch again	03:50
	durr	03:50
	→ perezd88 joined	03:52
robbyonrails	thanks!	03:53
	eternaleye_ → eternaleye	03:53
mugwump	no worries	03:53
justacrab	hi mugwump.	03:57
	→ bobmcw joined	03:58
mugwump	hi crab	03:59
	← yofel_ left	03:59
	→ schacon joined	04:04
	→ tjafk1 joined	04:10
neurodamage	Eridius: so you don't know of any scripts though that email the commits? or does that script do it and I just need to config the heck out of it?	04:12
Eridius	I think that script does it	04:12
	but I've never actually used it	04:12
neurodamage	Eridius: huh didn't on the email I received, showed the log, but no diff...i'll double check, thanks again	04:14
	← schacon left	04:15
neurodamage	Eridius: when I make changes to the repo .git/config file, they don't take?	04:15
	is that because I HAVE to use git-config?	04:15
	→ reithi joined	04:16
Eridius	neurodamage: git-config just writes to .git/config	04:16
	however, if you edit .git/config yourself, you might screw up the syntax	04:16
	← LostInTheWorld left	04:16
neurodamage	Eridius: ok, huh, so that's interesting, when I check out a local copy of the gem.git repo	04:16
	I find that the .git/hooks, isn't the same as the server?	04:17
Eridius	nope	04:17
	clone doesn't copy .git/hooks, .git/config, or .git/info	04:17
neurodamage	oh ok, then no worries, if that's how it's supposed to be	04:17
	ahhhh...excellent	04:17
Eridius	or, well, it actually doesn't copy anything in .git. It only fetches the objects	04:17
neurodamage	right...I've noticed that	04:17
	does git-fetch do that?	04:17
Eridius	yeah, git-fetch copies the objects	04:17
	git clone inits the repo and then calls fetch	04:18
	→ Bass10 joined	04:19
neurodamage	Eridius: is it ok to put strings around variables? with git config --global hooks.emailprefix="[GEM] "	04:20
Eridius	neurodamage: that will work just fine	04:20
neurodamage	Eridius: so when I run git config, it updates the repo right away right?	04:20
Eridius	actualy it won't work	04:20
	replace the = with a space	04:20
neurodamage	oh ok, excellent	04:21
Eridius	and why are you using --global?	04:21
	that writes ~/.gitconfig	04:21
neurodamage	it wrote it like so emailprefix="[GEM] "	04:21
	under the [hooks]	04:21
	awesome.	04:22
	Eridius: I'm the only developer :) for my company, it's kind of insane that way :)	04:22
	→ marj joined	04:22
neurodamage	Eridius: well how do I make sure the repo is updated with the configuration changes?	04:22
Eridius	neurodamage: what do you think happens when you run git-config?	04:22
neurodamage	Eridius: I'm setting git config options	04:23
	for git settings.	04:23
Eridius	right. and it writes out the config file immediately	04:23
neurodamage	default is current repository that's checked out	04:23
Eridius	now, how on earth could that not "update the repo" immediately?	04:24
neurodamage	Eridius: right, to the local copy, but how do I push that to the repo?	04:24
	oh ok	04:24
Eridius	push? what push?	04:24
	git's config isn't shared	04:24
	it's local to your repo	04:24
neurodamage	Eridius: well I'm checking out the repo and then running the git-config	04:24
Eridius	you don't push config	04:24
neurodamage	ok	04:24
Eridius	if you want to config your remote repo, you have to ssh to that repo and config it there	04:24
	.git/config is never pushed	04:24
neurodamage	oh ok, excellent, now I understand how it works, gotchya	04:25
	:)	04:25
	so basically everytime I check out a fresh copy the ~/.gitconfig is where the settings will be for everything right?	04:25
	→ jeukku joined	04:25
neurodamage	because when I checkout a fresh copy the .git/config will be different	04:25
Eridius	you do know what ~/.gitconfig is, right?	04:26
	it's your per-user global config	04:26
	it's not per-repo at all	04:26
neurodamage	Eridius: right.	04:27
Eridius	so why are you bringing it up?	04:27
	you only want to put stuff there that should always be set for any repo	04:27
	like, say, color.diff	04:27
	← tjafk2 left	04:27
Eridius	or your author name	04:27
neurodamage	oh ok, my bad	04:27
	Eridius: ok all done, thanks again. just have to read that script now to figure out how to insert diffs in there :)	04:29
	← Oompa left	04:31
	→ xinming joined	04:32
neurodamage	Eridius: where would I find a list of git-config variables?	04:32
Eridius	the core ones are documented in git-config. Generally each command documents the ones it uses, though	04:32
	← xinming__ left	04:34
	← niki left	04:34
neurodamage	Eridius: can't find project name?	04:34
Eridius	what?	04:34
neurodamage	setting the project's name	04:34
	the emails I receive are saying UNAMED PROJECT	04:34
Eridius	so read the script	04:34
neurodamage	k	04:34
	so I guess that's just a file in .git/description?	04:35
	→ CodeOfficer joined	04:36
	→ imyousuf joined	04:38
pjz	if I've done a clone of one remote repo, how do I pull in a branch from a different remote repo (so I can eventaully make a 3rd branch that's a merge of the two)	04:39
	← ben_h left	04:40
	→ ben_h joined	04:41
	← mithraic left	04:41
	→ ph^ joined	04:42
	← capisce left	04:43
	→ capisce joined	04:44
	→ xinming_ joined	04:46
	→ kanru joined	04:49
	← ben_h left	04:50
	→ ben_h joined	04:50
	→ harinath joined	04:51
	→ SRabbelier joined	04:55
neurodamage	Eridius: how do I get git to ignore files with status, etc...	04:56
	for example .o and .pyc	04:56
Eridius	man gitignore	04:56
Gitbot	Eridius: the gitignore manpage can be found at http://git.or.cz/man/gitignore	04:56
neurodamage	I already have a .gitignore	04:56
	oh ok	04:56
	← loincloth left	04:56
neurodamage	cheers again Eridius, I owe all my git learning to you and docs right now :P	04:57
	→ t3mp3st joined	04:59
t3mp3st	is it appropriate to ask git questions in this channel?	04:59
neurodamage	Eridius: if I make a change to the remote git repo how do I get an update of those changes?	04:59
	Eridius: in my local check out?	05:00
Eridius	neurodamage: you really should start your search in the documentation and in tutorials	05:00
t3mp3st	Eridius: I've looked through the docs, but not found an answer	05:00
Eridius	t3mp3st: did you seriously ask if it's appropriate to ask git questions in a channel named #git	05:01
	?	05:01
t3mp3st	haha, I've gotten yelled at before :P	05:01
	so here's the deal. Created an experimental branch, left at HEAD. Next, switched back to mater and did a hard reset to remove the last commit (so that the branch solely contains that history). Now I'd like to make the remote repository exactly mirror my local one. But I'm unable to get that to happen	05:03
	← xinming left	05:03
t3mp3st	by "the branch", I meant, "the experimental branch"	05:03
	→ mbuf joined	05:03
t3mp3st	another detail: the remote repository mirrored the state of the local repository before the creation of the branch and the hard reset on master.	05:04
	→ archis joined	05:05
	→ k776 joined	05:06
	← softdrink left	05:09
	← CodeOfficer left	05:10
	→ BierSchaumSchlue joined	05:10
	← qrush left	05:10
BierSchaumSchlue	http://www.pennergame.de/change_please/2199735/ ?!?	05:10
	← t3mp3st left	05:11
	← eikonos left	05:11
	→ capisce_ joined	05:12
neurodamage	Eridius: that one directory contrib contains all the nicties provided by others 3rd party right? I'm not missing the script that does the cvs diff in the git-push, I only ask so that I can save myself sometime on coding the scripts	05:12
Eridius	cvs diff? what?	05:12
neurodamage	Eridius: I want to also email the diffs of the files being commit, ie. the changes made in the code.	05:13
Eridius	so ask it for git diff	05:13
	→ xinming joined	05:14
	offby1` → offby1	05:14
	→ GNUix joined	05:14
	← BierSchaumSchlue left	05:14
	→ Ademan joined	05:15
neurodamage	Eridius: ok, once I'm done where/whom would I submit patch to ? andy parkins email is not listed :), of course	05:17
Eridius	<Git:Documentation/SubmittingPatches>	05:18
Gitbot	Eridius: I'm sorry, there's no such object: Git:Documentation/SubmittingPatches.	05:18
Eridius	hrm, did I do that wrong?	05:18
	<Git:HEAD:Documentation/SubmittingPatches>	05:18
	hmm	05:18
	anyway, Documentation/SubmittingPatches	05:18
	← capisce left	05:19
	← selckin left	05:20
Eridius	<Git HEAD:Documentation/SubmittingPatches>	05:20
	<Git master:Documentation/SubmittingPatches	05:20
	<Git master:Documentation/SubmittingPatches>	05:20
	Eridius pokes Gitbot	05:20
Mikachu	tr G g	05:20
Eridius	it responded to my first attempt without a branch name	05:20
	oh wait, that's the squelching code in there. damnit!	05:21
neurodamage	heh	05:21
	well played sir, seems I'm not the only who may be tired :)	05:22
	← vintrepid left	05:22
	→ t3mp3st joined	05:24
t3mp3st	can you cause a remote repository to do a hard reset?	05:24
	→ vintrepid joined	05:24
t3mp3st	I'm one of two developers, and we're still setting things up-- and I'd like the remote repository to look just like my local one	05:24
Mikachu	t3mp3st: faq non-bare	05:26
Gitbot	Mikachu: You shouldn't push to a non-bare repository. See http://git.or.cz/gitwiki/GitFaq#non-bare	05:26
	← xinming_ left	05:26
neurodamage	Eridius: would I have to pass the diffs from the post-commit to post-receive? or is there a way to generate a diff with any of these variables (oldrev, newrev, refname) with git	05:27
Eridius	neurodamage: if you think really hard, you might be able to come up with some possible way to ask git to give you a diff between two revisions	05:28
neurodamage	well that question is to anyone actually I need to stop directing ALL my questions to you	05:28
	Eridius: awwww...ok	05:28
t3mp3st	Mikachu: hm-- but my remote repository is bare at the moment	05:28
	am I misunderstanding this	05:28
Mikachu	then what do you want to reset?	05:28
	→ mountie joined	05:28
t3mp3st	well, by bare, I mean that the remote repository was created using --bare	05:29
	I'd just like to "undo" the latest commit in the remote repository	05:29
	← Zarin left	05:30
Mikachu	push -f theremote a_sha1:thebranch	05:31
	a_sha1 can for example be thebranch^ for the parent of the current commit	05:31
	you might want to talk to the other guy first to make sure he doesn't commit just before, then you would delete his stuff	05:32
	→ mczepiel joined	05:32
	→ xinming_ joined	05:32
t3mp3st	Mikachu: thanks so much-- no worries with respect to the other guy (we're just getting started)	05:32
	← werdnativ left	05:33
	← xinming left	05:34
	← spearce left	05:35
	← capisce_ left	05:35
	→ peritus joined	05:36
	→ capisce joined	05:37
	→ xinming joined	05:39
	← Bass10 left	05:39
t3mp3st	Mikachu: doesn't appear to be working (newly cloned repos contain the "removed" commit). I'm also getting the following message: "error: denying non-fast forward; you should pull first"	05:40
Mikachu	t3mp3st: then you didn't use the command i gave	05:41
	→ Zarin joined	05:41
Mikachu	t3mp3st: i think	05:41
t3mp3st	git push -f origin master\^:master is what I used	05:41
Mikachu	okay, then your server isn't allowing you to delete commits	05:41
t3mp3st	any override aside from rebuilding the server's repo?	05:41
	← perezd88 left	05:42
Mikachu	it's a flag in the config file in the server's repo	05:42
	something obvious like denyNonFastForwards = true probably	05:42
t3mp3st	lol, okay-- I'll go monkey around	05:42
	thanks for the help :)	05:42
	→ aiur joined	05:43
	← reithi left	05:46
	← timcharper left	05:46
	← aiur left	05:51
	← xinming_ left	05:52
	→ xinming_ joined	05:53
	→ gittish joined	05:58
	← Zarin left	05:59
	→ gbucher joined	05:59
	← xinming left	05:59
	→ namenlos joined	06:03
	→ BugeyeD_ joined	06:04
	← samurai left	06:06
	← BugeyeD left	06:06
	→ deavid1 joined	06:13
gbucher	Help ! I did a "git reset" on the wrong branch...	06:15
DrNick	which is why reflogs exist	06:18
gbucher	What are reflogs ?	06:19
DrNick	a sequential log of values for each reference	06:19
Eridius	man git-reflog	06:19
Gitbot	Eridius: the git-reflog manpage can be found at http://git.or.cz/man/git-reflog	06:20
johnw	usually git reset --hard HEAD@{1} will undo what you just did	06:20
DrNick	git log -g	06:20
johnw	ohh, awesome, I never knew about -g!	06:20
	← foca left	06:20
johnw	but in his case, what if his reset orphaned the previous HEAD?	06:20
	will log -g still show it?	06:20
DrNick	yes	06:20
johnw	sweet	06:21
gbucher	I don't know if that's the best way to do, but I did "git stash"	06:21
	git pull	06:21
	git stash apply	06:21
	→ Zarin joined	06:21
gbucher	and I am rewriting the last commit	06:21
	This seemed the safest way to be sure my "master" is in sync with the remote.	06:22
	← k776 left	06:22
johnw	gbucher: i wouldn't really say safer, just avoiding Git's own facilities for solving the problem	06:23
	→ foca joined	06:23
gbucher	or using commands I really understand...	06:23
johnw	it's very worth it to master the reflog	06:23
	you can undo _almost_ anything	06:24
DrNick	except branch deletions	06:24
	there probably should be a branch deletion log	06:24
	← GarethAdams left	06:24
	← foca left	06:24
	← harinath left	06:25
gbucher	I will dive into the reflog to understand it (and also the HEAD@{1} ref which is cryptic to me at the moment)	06:25
Eridius	HEAD@{1} means the first entry in HEAD's reflog	06:26
	i.e. the previous value of HEAD	06:26
	← bobesponja left	06:26
	→ harinath joined	06:26
gbucher	Eridius: nice.	06:26
	← jesselucas left	06:28
	→ priidu joined	06:28
	← johan-s left	06:32
	→ reel_ joined	06:33
	→ reithi joined	06:34
	← Ademan left	06:36
	→ bentob0x joined	06:39
	→ jesselucas joined	06:42
	← gittish left	06:44
	← jesselucas left	06:46
peritus	is there any stgit tutorial around, someone can recommend ? I simply don't get which problems stgit solves and why I need it (which I think).	06:46
johnw	i doubt you need it	06:46
	branches + rebase effectively replaces every need for it I've ever found	06:47
peritus	hm. ok then. thx.	06:48
	→ Epichero joined	06:51
	← deavid left	06:51
	→ deavid joined	06:51
	→ nud_ joined	06:52
	→ Tux32 joined	06:56
namenlos	johnw: me, too, was looking at stgit and couldn't find a reason for it.	06:56
johnw	there's one use case it makes easier	06:58
	if you have a stack of 9 patches and suddenly want to pull a middle patch temporarily	06:58
	but you can also just git revert that commit and then reset HEAD^	06:58
	→ Tux32- joined	06:59
	← Epichero left	06:59
	→ Weaselweb joined	07:00
thiago_home	you can create a new branch off HEAD, then rebase -i and remove that patch	07:00
	← Tux32- left	07:01
	→ Tux32- joined	07:01
	← Tux32- left	07:04
	→ Tux32__ joined	07:04
	← peritus left	07:05
	→ Tux32- joined	07:07
	← Tux32__ left	07:07
	→ LostInTheWorld joined	07:07
	← Tux32- left	07:09
	→ Tux32- joined	07:09
	→ drizzd joined	07:12
	→ Tux32__ joined	07:12
	← Tux32__ left	07:15
	→ Tux32__ joined	07:16
	← Tux32 left	07:16
	← Tux32__ left	07:17
	→ Tux32__ joined	07:18
	← Tux32__ left	07:20
	→ Tux32__ joined	07:20
	← litage left	07:21
	← Tux32__ left	07:21
	← gbucher left	07:24
	← t3mp3st left	07:28
	← Tux32- left	07:29
	→ gittish joined	07:35
	→ koke joined	07:36
	→ shodges joined	07:37
	← gittish left	07:39
	→ robinr joined	07:39
	← Fullmoon left	07:39
	← robinr left	07:40
	← vintrepid left	07:41
	→ Fullmoon joined	07:42
	→ robinr joined	07:43
	← ben_h left	07:44
	→ ben_h joined	07:47
	→ gbucher joined	07:51
	← mczepiel left	07:51
	→ BabelO joined	07:53
	← johnw left	07:53
bentob0x	github down?	07:54
	→ reel__ joined	07:54
doener_	worksforme	07:54
bentob0x	hmm	07:54
doener_	at least the start page	07:54
bentob0x	:)	07:54
	doesn't seem to work for me	07:55
	can't pull + no website	07:55
	hmm	07:55
doener_	cloning a random project also works here	07:55
bentob0x	might be a temp network problem	07:56
	→ loiseau joined	07:56
	→ kef joined	07:57
	→ charon joined	07:57
	→ Pupeno joined	07:58
Pupeno	Hello.	07:58
jaalto	How do people deal with massive rebases? I mean, if you have 20-30 branches to upstream code and then new version comes.	08:00
gebi	jaalto: git-wtf for the summary	08:03
	→ Hongbo joined	08:03
neurodamage	Eridius: so I have it posting the diffs in the email now, I'm wondering though, the output for --color are those terminal colors?	08:03
Eridius	what?	08:03
Pupeno	When I do a git svn fetch, the SVN branches like trunk or others get update but master remains untouched, right?	08:03
Rhonda	Pupeno: you need to do a git svn rebase afterwards.	08:04
neurodamage	Eridius: I modified the script but when it sends emails I want the stuff colored niceley :), so I was wondering if you knew about a git tool that did that? html colored rendering so I don't have to code all that in there :)	08:04
Eridius	I don't know	08:04
	normal coloration is for terminals	08:04
	Rhonda: afterwards? git svn rebase is the usual command. if you've done git svn fetch, you can do a simple rebase	08:04
Pupeno	Rhonda: I did a git merge trunk on master, would that be the same?	08:04
	→ trochala joined	08:04
charon	neurodamage: http://en.wikipedia.org/wiki/ANSI_escape_sequence	08:05
	iirc it uses those	08:05
Rhonda	neurodamage: git diff \| vim -	08:05
	neurodamage: :runtime! syntax/2html.vim	08:05
	← Hongbo left	08:06
Rhonda	Pupeno: Propably, but rebase is encouraged, especially if you have local commits not dcommit'ed yet.	08:06
neurodamage	Rhonda: woah...that's crazy vim stuff :), thing is I need it to automate that process for a git commit emailer :)	08:06
	Rhonda: but that's a start	08:06
Rhonda	Eridius: afterwards like in git svn fetch and ... well, then. :)	08:06
neurodamage	Rhonda: heh, I knew about ANSI Color escape sequences etc...but I didn't know that vim was so cool like that :P	08:07
Pupeno	Rhonda: ok, thanks.	08:07
	I don't plan to dcommit ever (I don't have write access to svn), would it be problematic to work on master?	08:07
Rhonda	Don't think so, but rather use rebase than merge, just in case.	08:08
	svn rebase that is.	08:08
neurodamage	Rhonda: so how would I automagic that whole process on the command line in one statement ?	08:08
Pupeno	ok, thanks.	08:08
	→ GarethAdams joined	08:09
	→ capisce_ joined	08:09
	← Fullmoon left	08:09
	← reel_ left	08:10
Rhonda	neurodamage: git diff \| vim -c 'runtime !syntax/2html.vim' -c 'w diff.thml' -c 'q' - # and then use diff.html	08:10
	→ markkald_ joined	08:10
Rhonda	hmm, wait. that didn't color it, what the ...	08:10
neurodamage	git diff --color	08:11
Rhonda	ah	08:11
	← markkalderon left	08:11
	← mithro left	08:12
	← capisce left	08:12
	→ johnw joined	08:12
Rhonda	neurodamage: git diff \| vim -c 'TOhtml' -c 'w! diff.thml' -c 'qa!' - # might overwrite previous existing diff.html files	08:13
	→ elmex joined	08:13
neurodamage	Rhonda: git diff should only pipe one or all the files diffs right?	08:13
	Rhonda: not keep overwriting them?	08:13
Rhonda	Everything you want to have htmlized, so tweak the git diff part to what you aim for.	08:14
neurodamage	Rhonda: ok awesome, thanks, that saved me some time :)	08:14
	→ Fullmoon joined	08:15
	← Fullmoon left	08:15
	→ snitko joined	08:18
robinr	omNatten	08:18
neurodamage	Rhonda: I'm getting errors about output not being to a terminal for vim?	08:21
Rhonda	Then you are missing the -	08:22
robinr	bah	08:24
	← ben_h left	08:24
	→ ben_h joined	08:25
	← ben_h left	08:25
	← codeshot left	08:25
	→ drizzd_ joined	08:27
	← priidu left	08:27
neurodamage	Rhonda: I was testing your script you gave me it doesn't take in the content from the git diff --color files \| vim	08:28
	Rhonda: oh ok	08:28
	← Pupeno left	08:28
neurodamage	Rhonda: is that runtime standard with vim or do I have to download it from somewhere BECAUSE it's not converting the content into HTML?	08:29
	git diff --color choices.py \| vim - -c 'runtime !syntax/2html.vim' -c 'w! diff.html' -c 'q'	08:29
Rhonda	That's part of the vim core packages.	08:30
	Do my later part.	08:30
	→ tzafrir_laptop joined	08:30
Rhonda	neurodamage: \| vim -c 'TOhtml' -c 'w! diff.thml' -c 'qa!' -	08:30
neurodamage	Rhonda: that worked except it didn't remove the ANSI escape sequences lol	08:31
	→ Man-erg joined	08:32
Rhonda	neurodamage: You used git --color switch in there?	08:32
neurodamage	Rhonda: yeah ...	08:33
Rhonda	Go figure ;)	08:33
neurodamage	yeah...but I mean I want it to be colorized :)	08:33
	Rhonda: seems no simple script is going to do it, I'm going to try this :), http://www.linuxjournal.com/content/convert-diff-output-colorized-html	08:33
Rhonda	Yes, but not ansi coloured but vim syntax coloured.	08:33
	← BarryCarlyon left	08:34
Rhonda	Just drop the --color switch from git diff. :)	08:34
	→ Ademan joined	08:35
neurodamage	Rhonda: I had that working before :)	08:35
	← pygi left	08:35
neurodamage	Rhonda: if i place a script in the hooks folder I can call it and it won't interfere with anything right?	08:36
	← snitko left	08:38
jogla	Can i revert the last commit and have the changes back in the index?	08:38
charon	reset --soft HEAD^	08:38
	(note that "revert" means something else in git)	08:39
jogla	Thank you	08:39
charon	also note that there's commit --amend which might be easier in this case	08:39
Rhonda	neurodamage: Depends. I think some of the hook scripts depend on return codes so make sure there won't be anything going wrong.	08:40
	→ priidu joined	08:40
	← loiseau left	08:41
neurodamage	Rhonda: ok where's a good place to do things like that?	08:41
	put scripts/custom ones?	08:41
	i'll just put it in /usr/bin	08:41
	← reel__ left	08:42
	→ reel_ joined	08:42
	← drizzd left	08:42
Rhonda	neurodamage: I'm not that confident with git itself that than I feel confident in vim usage, so you might address that question to the general public than to me specific. :)	08:43
	← ajonat left	08:44
	→ sson joined	08:45
	→ loiseau joined	08:46
sson	How do you set up so that a clone of a git-svn repo can be used to commit to svn directly as well?	08:46
	i suspect i need to add something to .git/config?	08:46
Mikachu	run git svn init on the clone with the same original arguments as the git svn clone	08:46
	then you might need to fiddle a bit with remote tracking branches too	08:47
	it's easy if you only track one branch, just copy master to remotes/git-svn	08:47
	it's easier if you use fetch refs/:refs/ when cloning the first repo maybe	08:47
Rhonda	sson: You just have to have to use the appropriate url set through which you can commit, git svn dcommit is what you are looking for then.	08:47
Mikachu	note that you can't do that with clone, so you'd have to use init + remote add + fetch	08:47
	and that i haven't tried it :)	08:48
neurodamage	heh Rhonda looks like I have the cool diff.html file I'm looking for, BUT it's not sending the email mimetype as html content lol	08:48
	→ capisce joined	08:48
sson	hmm. okay. this will be tricky and not that good for teasing coworkers into start using git.	08:49
Rhonda	sson: It's really	08:49
	→ botanicus joined	08:50
Rhonda	sson: It's really just that, git svn clone, git commit, git dcommit	08:50
	Not really sure what Mikachu makes it sound more complicated for, did work for me just like that.	08:50
	erm, git svn dcommit of course.	08:51
Mikachu	Rhonda: we want to clone the git svn clone with git clone, and then dcommit from there	08:51
Rhonda	oh, ah.	08:52
	← statim left	08:52
	→ statim joined	08:52
	→ BarryCarlyon joined	08:52
	← capisce_ left	08:53
	→ capisce_ joined	08:53
sson	hey! git svn fetch just worked from the git-clone after some git svn init and copying .git/svn/ stuff from the master git svn clone.	08:58
	← johnw left	08:58
	← capisce left	08:58
	→ johnw joined	08:59
	← priidu left	09:00
neurodamage	Rhonda: success! :)	09:02
	← koke left	09:03
	→ joelr1 joined	09:04
	← Araneidae left	09:05
joelr1	good morning! how do i clone a remote branch (on my disk, actually) of a repo? if i just clone i get the currently active branch and see the "remote" branches with git branch -r	09:05
	← joelr1 left	09:07
	← GarethAdams left	09:08
	→ Araneidae joined	09:09
	→ pengo joined	09:09
	← BarryCarlyon left	09:11
	→ WildPikachu joined	09:11
WildPikachu	can one ignore the trailing space warning? (busy checking in config files)	09:12
doener_	--whitespace=nowarn or --whitespace=fix	09:12
	(assuming that you're talking about git apply)	09:12
WildPikachu	fanks man	09:13
jogla	Is there a git equivalent for svn export?	09:14
doener_	there are also some config options related to various whitespace warnings/errors, see git-config(1), I don't remember them off-hand	09:14
	jogla: git archive	09:14
jogla	doener_: Thank you	09:14
WildPikachu	now ... to ignore/strip file exec permissions ... hrmmm	09:14
	→ capisce joined	09:14
WildPikachu	gotcha	09:15
jogla	doener_: but that does also create an archive	09:16
Mikachu	jogla: git archive $args \| tar -C /targetdir x	09:17
jogla	Heh, ok, thank you :)	09:18
doener_	the plumbing way is something like: GIT_INDEX_FILE=/tmp/foo git --work-tree=/path/to/target --git-dir=/path/to/.git read-tree -u --reset <tree-ish>; rm /tmp/foo	09:22
	← tzafrir_laptop left	09:22
doener_	if you really badly want to avoid the tar call ;-)	09:22
	→ niki joined	09:24
johnw	that didn't work for me doener	09:25
	GIT_INDEX_FILE=/tmp/index.$$; git --work-tree="$2" read-tree -u --reset $1	09:25
	→ Sho_ joined	09:25
johnw	called: git export HEAD /tmp/foo	09:25
	→ BarryCarlyon joined	09:26
Mikachu	did you try the direct call?	09:26
johnw	direct call?	09:26
Mikachu	not a wrapper script	09:26
johnw	yeah, it doesn't do anything	09:27
	GIT_INDEX_FILE=/tmp/index; git --work-tree="/tmp/foo" read-tree -u --reset HEAD	09:27
doener_	johnw: you need the --git-dir part (otherwise it looks for .git in the work tree)	09:27
johnw	do I have to have --git-dir?	09:27
	there is a .git in my work tree	09:27
Mikachu	i don't think doener_ wrote it for fun	09:27
	you can't just remove args and say it doesn't work :)	09:27
doener_	johnw: and if you put the ";" after the env var setting, you need to export it	09:27
johnw	ohhh	09:28
	gotcha	09:28
	tripped up by sh 101	09:28
	that was it	09:28
doener_	johnw: you have a .git in /tmp/foo?	09:28
johnw	oh, I see what you mean	09:28
	do I need to mkdir /tmp/foo?	09:29
doener_	yep	09:29
johnw	hmmm.. i still can't get this to work	09:29
bentob0x	github still down?	09:29
johnw	http://gist.github.com/11576	09:29
	← capisce_ left	09:30
johnw	doener_: does that look right to you?	09:30
	→ Signum joined	09:30
doener_	johnw: try "$(git rev-parse --show-cdup)/.git" instead of $PWD	09:31
johnw	fatal: Not a git repository: '/.git'	09:31
	i'm inside a working tree as I run this	09:32
Mikachu	bentob0x: #github?	09:32
johnw	ah, I see what it does	09:33
	but still no workie	09:33
doener_	johnw: sorry, meant "$PWD$(git ...)/.git"	09:33
	but yeah, that somehow still fails	09:34
	ah, no, that was my dumbness, I had an extra ) in there	09:34
johnw	i updated teh gist, but that command still does nothing	09:35
	← imyousuf left	09:36
doener_	http://git.pastebin.com/m7aa817ce	09:37
johnw	a more glorious version of the same, I'm afraid	09:38
	still no exported files	09:38
	→ bcarlyon\|laptop joined	09:38
doener_	worksforme	09:38
johnw	hmm	09:38
doener_	how do you call it?	09:38
johnw	i see what's happening here	09:39
	it's resetting my working tree	09:39
	which, thanks to my being uncautious, just wiped out about an hour of coding	09:39
doener_	ouch, sorry about that :-(	09:40
johnw	n/p, I deserved it	09:40
	must always git-stash before playing with git scripts	09:40
	doener_ thankfully did a "cd ~/src/git" first, nothing to lose there	09:40
johnw	argh, recoding is one of the least pleasant things to do	09:40
doener_	an intermediate version here had "WORK_TREE" instead of "GIT_WORK_TREE" and also messed up things here	09:40
johnw	maybe Time Machine can save me a few minutes, brb	09:41
	→ ciaran joined	09:41
	→ ponch84 joined	09:42
doener_	johnw: wait a second	09:42
johnw	yeah?	09:42
	← BarryCarlyon left	09:43
	bcarlyon\|laptop → BarryCarlyon	09:43
doener_	johnw: hm, nevermind, I guess... I thought your "GIT_INDEX_FILE=...;" might have caused trouble	09:44
	johnw: but even unsetting that here still gives me a broken index... hu?	09:44
	ah, lol I managed to export into "xdiff" at some point, that's why I saw the untracked files...	09:45
	didn't pay attention that the xdiff prefix was on all files...	09:46
johnw	ah, good old Time Machine	09:46
	last backup was 2.5 hours ago, but even still this work started about 6 hours ago	09:46
doener_	basically, your script is just missing a / from mine (after $PWD, only required when you're not at the top-level of the repo)	09:47
	dunno why that fails for you :-(	09:48
johnw	well, I guess the glamour of discovery fails me at the moment, but it's early	09:48
	← mediogre left	09:54
	← reithi left	09:57
	→ koke joined	09:58
	→ ereslibre joined	10:00
	→ cilly joined	10:01
	→ Zdra joined	10:06
	← Laibsch left	10:06
Zdra	Hi, I created a git repository using "git-svn clone <url>" a long time ago. But now I realise that I don't get SVN branches into my git repository	10:08
	I see that the doc tells to use "-t tags -b branches -T trunk" options to create the repository and if I reclone the repository with those options I get branches and tags	10:09
xinming_	I've imported an repository, What's the best way to checkout git version one by one? and see what is changed? for now, what I do is, 'git log' and get the $first_sha1, then git checkout, after finished, the 'git checkout master' and do git log again.	10:09
	Is there a better way?	10:09
	← trochala left	10:09
Zdra	However I have lots of contributors having branches based on my git repository. Is there a way to add SVN branch to my existing git repo?	10:09
	→ DrFrasierCrane joined	10:10
	← soul9 left	10:12
	← gbucher left	10:13
	← johnw left	10:16
	→ jmd joined	10:20
	→ vintrepid joined	10:20
doener_	Zdra: yes, check the .git/config in both versions, the "full" one should differ in three things: fetch says "trunk" instead of "git-svn" and there are branches and tags lines in the svn-remote section	10:20
	adjust the "partial" clone to have those branches and tags lines and then "git svn fetch" (maybe try on a copy of the repo first)	10:21
	Zdra: if it differs in the "url" line as well, you need to adjust a bit more, but the idea stays the same	10:22
	xinming_: "git log -p", maybe? I don't really understand what you want to do	10:23
	→ johnw joined	10:25
	← jonkristian left	10:25
	→ Lash\| joined	10:27
jmd	How can I undo a push?	10:28
Rhonda	You can't, from remote.	10:28
Pieter	sure you can	10:29
Rhonda	Oh?	10:29
Pieter	git push -f origin old-remote-sha:master	10:29
	will push an older sha to the branch 'master' on remote	10:29
Rhonda	But that will not delete what got pushed, it will just change the ref, not?	10:30
Pieter	yes	10:30
Rhonda	Then it's no undo, IMHO. :)	10:30
Pieter	it's probably good enough	10:31
	→ jammyd_home joined	10:31
Rhonda	It goes the security-by-obscurity path. If the intention of the question was to not have something published that shouldn't get out, you can't do that remotely.	10:31
jammyd_home	git-branch doesn't seem to like GIT_DIR being set. No matter what I set it to, it complains that the directory isnt a git repo	10:31
	git-branch ; GIT_DIR=$(pwd) git-branch	10:32
	can someone try that to verify it for me?	10:32
Pieter	Rhonda: if you change the ref, others can't access it anymore	10:32
xinming_	doener_: No, I want to check the local copy, Eg, to see what the directory structure like	10:32
Rhonda	jammyd_home: Shouldn't that be $(pwd)/.git?	10:32
Pieter	Rhonda: so with plain git:// access it'll still be safe	10:32
jammyd_home	Rhonda: oooh :P	10:32
	indeed it should, my bad :)	10:32
Rhonda	Pieter: Not? Can't they explicily ask for that sha1sum, if they know about it?	10:32
jammyd_home	thanks	10:32
Pieter	Rhonda: no	10:33
xinming_	doener_: Since for my old subversion repo, I've changed layout many times, So, I want to see the change one by one for local copy.	10:33
Pieter	Rhonda: the sha's they ask for have to be reachable from refs	10:33
	Rhonda: otherwise the fetch is cancelled	10:33
Rhonda	hmm	10:33
	→ parasti joined	10:36
	← cilly left	10:39
	→ tvw joined	10:40
	→ cilly joined	10:40
	→ jezdez_ joined	10:43
	→ trochala joined	10:43
	← jezdez_ left	10:45
	← BarryCarlyon left	10:49
	→ cbreak joined	10:53
	← namenlos left	10:56
	→ ciaran_ joined	10:59
	← ciaran_ left	11:00
	→ ciaran_ joined	11:00
	→ ciaran__ joined	11:02
	← ciaran_ left	11:03
	→ warlock_mza joined	11:03
	← markkald_ left	11:06
	← botanicus left	11:08
	→ bcarlyon\|laptop joined	11:14
	← ciaran left	11:16
	→ christophsturm joined	11:16
	→ EmilMedve joined	11:16
	bcarlyon\|laptop → BarryCarlyon	11:17
	← gretchen left	11:19
	→ pehlert joined	11:20
	← koke left	11:23
	→ schierbeck joined	11:25
	← vintrepid left	11:27
	→ hurikhan\|Work joined	11:28
jmspeex	I just received a patch that doesn't apply. Any way of finding on what commit it's based?	11:30
doener_	jmspeex: does it have an "index: " line?	11:30
jmspeex	It has one index line for each file	11:31
	doener_: i.e. it was generated with git format-patch	11:32
doener_	then try "git am -3"	11:32
	that causes git to try to build a fake ancestor and do a 3-way merge	11:32
Mikachu	doesn't git format-patch put the actual commit in the header too?	11:32
	From ce425c61a0ba6c6daa44bf51a98cc50812cfe247 Mon Sep 17 00:00:00 2001	11:33
jmspeex	error: libcelt/cwrs.c: does not match index	11:33
	and so on	11:33
Mikachu	that's a weird date...	11:33
	anything i produce with format-patch gets that date, anyone else get that?	11:34
doener_	Mikachu: yep, same date, different commit	11:34
Mikachu	i wish i would stop finding bugs everywhere	11:35
doener_	just close your eyes ;-)	11:35
Mikachu	die("bogos committer info %s\n", committer);	11:36
	see, just trying to find this bug i found another bug	11:36
charon	sounds like a use-case for automated bisection? ;)	11:36
doener_	Mikachu: should be "bog os", means Windows	11:37
Mikachu	it will probably show something like "builtin git-format-patch rewrite in C" :)	11:37
doener_	Mikachu: log-tree.c: printf("From %s Mon Sep 17 00:00:00 2001\n", name);	11:38
	→ stouset joined	11:39
Mikachu	- printf("From %s Thu Apr 7 15:13:13 2005\n",	11:40
	+ printf("From %s Mon Sep 17 00:00:00 2001\n", sha1);	11:40
	but no explanation	11:40
	Mikachu continues blameing	11:40
	← ph^ left	11:40
Mikachu	3eefc189 puts the 2005 there also without explanation	11:41
Gitbot	[git 3eefc189]: http://tinyurl.com/4mn78h -- Tentative built-in format-patch.	11:41
Mikachu	see, i was right :)	11:41
doener_	5c2c972f474f22894f5cbcc1234b5dc022de5a33	11:44
Gitbot	[git 5c2c972f4]: http://tinyurl.com/4f59mq -- [PATCH] format-patch: --mbox and --check.	11:44
doener_	that seems to be the first version that adds that "From " line	11:45
	uses the 2001 date	11:45
	and no, no explanation either :-)	11:45
Mikachu	hm, i guess that is some internal mbox format then, and it was changed to put the sha1 in there just for fun later?	11:46
	← Flathead left	11:46
	← EmilMedve left	11:47
	→ koke joined	11:47
charon	the From line is what separates mails in mbox... not sure why it puts in a bogus date though	11:47
	→ sednivo joined	11:48
robinr	presumably because a timestmap is expected	11:49
	→ ben_h joined	11:49
	← johnw left	11:49
charon	yes, but why not use the _real_ date?	11:49
	→ glommer joined	11:51
	→ spook joined	11:52
spook	it might just be my tiredness, but i cant find any documentation on how to setup a git repo	11:52
	← jeukku left	11:52
xl0	spook: git-init-db?	11:53
doener_	git init	11:53
	git init-db is oooold	11:53
parasti	init-db? woah	11:53
	→ Yuuhi joined	11:53
xl0	Ugh, that's how I was always doing it..	11:54
	What's the difference?	11:54
Mikachu	three letters	11:54
spook	anything else i need to do? perhaps a service?	11:55
Pieter	yugh, letters	11:55
Mikachu	- is now a letter	11:55
	nud_ still uses git-init-db	11:56
nud_	hold habits last...	11:56
	xl0: none	11:56
	apart from the fact that init-db is "obsoleted" by init	11:56
xl0	Hey, it's just a hard link!	11:56
	→ kukks joined	11:58
	→ cannonball joined	11:58
	→ boto joined	11:58
	← spook left	11:59
Zdra	doener_: Hm, it does not fetch branches	11:59
	now my .git/config looks like that:	11:59
	http://pastebin.be/13850	11:59
	← deavid1 left	12:00
xl0	spook: If you want to serve the repo to others, you might consider running git-daemon.	12:00
Zdra	doener_: I simply added the branches and tags lines	12:00
doener_	Zdra: you checked with "git branch -r", right?	12:00
	→ elliottcable joined	12:01
Zdra	doener_: yes	12:01
	git svn fetch did nothing	12:01
	← huperniketes left	12:01
Zdra	doener_: in fact the URL should be changed since I still have /trunk	12:01
doener_	Zdra: yeah, if the url line differs, then you need more adjustments	12:02
Zdra	doener_: what do I need?	12:02
	← elliottcable left	12:02
doener_	Zdra: the fetch line also needs to be changed then to accomodate the changed url line	12:02
Zdra	if I just change the url git svn fetch still does nothing	12:02
doener_	Zdra: basically, using the config from the "full" clone should do, except that the right hand side of the fetch line should say "git-svn" instead of "trunk"	12:03
	(you can also rename your git-svn remote tracking branch and do some other stuff, but that is annoying, so if you can live without that...)	12:03
	→ johan-s joined	12:03
Zdra	doener_: So now I have that: http://pastebin.be/13851	12:06
	neurodamage → neurodamage_away	12:06
Zdra	but git-svn fetch still does nothing and branches does not appear	12:06
	→ ebel joined	12:07
doener_	Zdra: I'll clone here	12:08
	→ lorandi joined	12:08
Mikachu	i think you need to create the tracking branches manually first	12:11
	← BarryCarlyon left	12:11
xl0	pasky: http://git.slind.org/git <- Is this "category" feature available in the mainline gitweb?	12:11
Mikachu	xl0: almost all git binaries are hardlinks	12:11
xl0	Ugh, right. ;))	12:12
Mikachu	not sure if there's a syntax in bash, but in zsh for example, ARGV0=git-init git-add works fine too :)	12:12
	→ foca joined	12:13
Zdra	doener_: I tried several things, but each time git-svn re-downloaded the whole svn history instead of basing the branches on the commits I already have in my repository	12:14
	k-s[AWAY] → k-s	12:15
Mikachu	Zdra: try git fetch origin 'refs/remotes/:refs/remotes/' or so (maybe)	12:16
doener_	Mikachu: we're talking about extending a git-svn repo so that it fetches branches and tags, besides the trunk which it already has	12:17
Mikachu	oh okay, i thought this was the clone of git-svn repo person :)	12:18
	→ powr-toc joined	12:18
powr-toc	Is there anyway to merge a branch (with no shared ancestry) into a subdirectory within another branch?	12:19
doener_	Zdra: worksforme	12:19
	oh wait, it starts doing something weird now	12:20
powr-toc	Kinda like a filter-branch, but I don't want to construct a new repo, and I don't want to merge into the root directory of the other branch.	12:20
pasky	xl0: no	12:21
	there are some patches floating around	12:21
doener_	Zdra: ah, no, that was just the svn tag being broken	12:21
powr-toc	Do I want a subtree merge strategy?	12:21
Mikachu	powr-toc: yes, <HEAD:Documentation/howto/using-merge-subtree.txt>	12:21
Gitbot	[git using-merge-subtree.txt]: http://tinyurl.com/6o88by [blob]	12:21
powr-toc	Mikachu: cheers :-)	12:22
xl0	pasky: Is there a good reason they didn't find their way into the mainline?	12:22
doener_	Zdra: yup, worked out just fine, AFAICT from "gitk --all"	12:22
	→ ruphy joined	12:22
doener_	Zdra: same .git/config	12:22
	well, except for the url, where I used anonymous svn:// access	12:22
Zdra	doener_: Oh, I remember that I hacked a bit my repository when I created it, by adding that: http://pastebin.be/13852. Now that I removed that your solution seems to work	12:23
	← stouset left	12:23
doener_	hm, interesting	12:23
	I don't see how that would conflict, but anyway, glad it works now	12:23
	→ iulian joined	12:26
	← cilly left	12:27
pasky	xl0: no	12:27
	i think at least	12:27
	noone got around to review and push them through	12:28
justacrab	hi pasky.	12:28
pasky	hi	12:28
	→ stouset joined	12:29
	→ cilly joined	12:30
	← jmd left	12:31
	→ drewr joined	12:32
	→ BarryCarlyon joined	12:33
	→ gbucher joined	12:35
powr-toc	Mikachu: I'm following the subtree merge example, almost to the letter and stage<2> fails with a "Merge strategy ours failed." error. Any ideas?	12:35
	← deavid left	12:35
Mikachu	did it print anythign else?	12:36
xl0	pasky: If I find the author and ask him to resend the patch, would you merge it?	12:36
powr-toc	Mikachu: no, nothing	12:36
	← sednivo left	12:38
powr-toc	Mikachu: presumably I need to commit my changes before doing the merge?	12:38
Mikachu	what changes?	12:39
powr-toc	changes in the branch I want to subtree merge into	12:39
	← jas4711 left	12:40
	← trochala left	12:40
	→ EmilMedve joined	12:40
powr-toc	Mikachu: yeah, that seems to have been the problem...	12:41
	→ Bass10 joined	12:41
	→ mcella joined	12:44
powr-toc	Mikachu: Thanks for the help! It worked a treat! :-)	12:46
Mikachu	np :)	12:46
powr-toc	git, yet again proves that it rocks!	12:46
	← flaguy left	12:46
	→ deavid joined	12:47
	k-s → k-s[AWAY]	12:52
	→ capisce_ joined	12:52
	← deavid left	12:53
	→ deavid joined	12:53
Zdra	doener_: anyway, it works fine! thanks you very much!	12:55
	doener_: just a last question: If I want to push a branch in svn how can I do that?	12:55
	← sanity_ left	12:55
	→ irotas joined	12:55
Zdra	I do "git checkout -b my-branch git-svn" to and then I make some commits, and finally I would like to push it on my svn repository as a new branch, not in trunk	12:56
	→ whee joined	12:57
	← whee left	12:57
	→ madewokherd joined	12:57
	sson → snos	12:59
	← deavid left	12:59
	→ gittish joined	12:59
	← robinr left	12:59
	→ deavid joined	12:59
doener_	Zdra: there were patches for that on the mailing list	13:00
	Zdra: for now, I'm using this script http://git.pastebin.com/m48d204e4	13:00
	→ cypromis joined	13:01
cypromis	I have a question regarding git-svn	13:01
	→ LiamH joined	13:01
	→ whee joined	13:02
doener_	Zdra: and you either need to create and fetch the branch first, so your work is already based on it, or you'll have to rebase it onto the new branch before dcommitting	13:02
cypromis	I cloned a svn repo	13:02
	and now want to redistribute it with git-daemon	13:02
	so copied the .git dir into project.git	13:02
	and the redistribution is fine	13:02
	but how do I update the project.git dir with the latest svn repo changes ?	13:02
Zdra	doener_: in fact I just want to branch for GNOME 2.24 release, so I can even manually branch on svn, then "git svn fetch" will import it in git	13:03
	doener_: like that I guess I can do "git checkout -b my-stable-branch-fix gnome-2-24", make some commits, and git-svn dcommit will push commits to the "gnome-2-24" svn branch, right?	13:04
Pieter	http://pastie.org/275731	13:05
	heh	13:05
	→ etgryphon joined	13:06
	← capisce left	13:06
Mikachu	Pieter: if you write git aaaaaaaaaaaa, does it suggest format-patch for you?	13:06
Pieter	Mikachu: no, it suggests nothing	13:06
Mikachu	hm, i must have busticated something then	13:06
Zdra	doener_: also: Let's say someone commited in trunk a patch that is useful for the stable "gnome-2-24" branch. "git-svn rebase" will import that patch into my "git-svn" git branch. Can I cherry-pick the commit into my "gnome2-24" git branch and then "git-svn dcommit" will push it to the "gnome-2-24" svn branch?	13:07
	← gbucher left	13:07
Mutru	Umh, I'm pretty new to git. I wanted to cherry-pick some patches to an existing release branch, so I did 'git co -b rel10 origin/REL10', then 'git cherry-pick -x <several commits>' and then 'git push'. It claims that "Everything up-to-date". Any ideas what could be causing this?	13:07
	→ warthog19 joined	13:07
Mutru	gitk shows that the origin/REL10 branch is NOT up-to-date with rel10.	13:07
fujin	rel10 != REL10	13:08
	git push origin REL10	13:08
	← warthog9 left	13:08
fujin	rel10 => REL10	13:08
Mikachu	Pieter: are you on next or master?	13:08
Pieter	Mikachu: next	13:08
ciaran__	Zdra: you should make a new branch based on the branch in svn, and then rebase your changes onto that	13:08
cypromis	hmmm	13:08
Mikachu	hm, extra weird then	13:08
doener_	Zdra: a) yes, b) yes, but maybe use -e with cherry-pick and remove the git-svn-id line, just to make sure	13:08
Mutru	fujin: rel10 is my local branch, REL10 is remote. That doesn't work either: there's no local branch named REL10.	13:08
doener_	Zdra: git-svn should recognize that the git-svn-id line is wrong, but better safe than sorry	13:08
Pieter	Mikachu: maybe my aliases patch messed something up? not sure if that's in next	13:09
doener_	Zdra: you can also use "git svn dcommit -n" to just see what it will do	13:09
fujin	Mutru: uh? I thought you're trying to push to the remote REL10 branch - what are you tyring to push to if not that?	13:09
Mikachu	Pieter: it was the same before that too	13:09
Pieter	hm	13:09
Mutru	fujin: Pushing my local rel10 to remote origin/REL10, yes. :) 'git push origin REL10' gives me "error: src refspec REL10 does not match any".	13:09
Mikachu	Pieter: it's probably my fault somehow	13:10
Pieter	Mikachu: :D	13:10
	Mikachu: I lag a little on next	13:10
fujin	Mutru: ehm; try git push origin rel10:REL10	13:10
	← stouset left	13:11
Mutru	fujin: Ok, thanks, that was it.	13:12
	→ feugan3333 joined	13:12
fujin	Mutru: sorry yeah	13:12
	src refspec gave it away - check git-push manpage for more details	13:13
	→ ciaran joined	13:13
	→ snitko joined	13:14
	→ mw joined	13:15
meyering	hi git crowd! is there a better way to test for is_commit_with_2_or_more_parents in Bourne shell code than sha1=HEAD; p=$(git log -1 --pretty=format:%p $sha1); case $p in ' ') echo yes;; *) echo no;; esac -- FYI, this is for use in an update hook to prevent pushing accidental merge commits	13:15
Zdra	doener_: many thanks!! I owe you a beer	13:16
	doener_: ah, the last question: If I remove a branch from SVN, will it be removed from git too? How can I tell git-svn to purge removed branch?	13:18
doener_	no	13:18
	← ciaran__ left	13:18
	→ vbabiy joined	13:18
	→ p4tux joined	13:20
doener_	meyering: git rev-parse --verify --quiet $sha^2 && echo 'Is merge' \|\| echo 'Is not a merge'	13:20
	→ M_J_G joined	13:21
doener_	meyering: but note that that will say "Is not a merge" for invalid $sha values as well	13:21
	→ GarethAdams joined	13:22
meyering	doener_: perfect! thank you.	13:22
feugan3333	Hi all. I'm trying to create a repository where user push to. I'm will use the repository with git-svn so I don't want users merging branches. I set the receive.denyNonFastForward setting to True. But it does not work as expected. It still allows users to push merges, but when I do the pull from that repo into my git-svn rep then I get the error: ! [rejected] master -> master (non fast forward). Can anyone explain this?	13:22
Zdra	doener_: so there is no way to remove branches tracking a deleted svn branch?	13:23
doener_	feugan3333: "git svn rebase" and "git svn dcommit" both rewrite history	13:23
	Zdra: you can manually do that. "git branch -r -d remotes/foo"	13:23
Zdra	doener_: ok	13:23
feugan3333	doener_: So what should I do if I want to prevent users from using branches?	13:24
doener_	feugan3333: setup a hook that denies pushing merge commits. You cannot generally prevent them from using branches (and using branches alone does not mean non-linear history)	13:25
	feugan3333: maybe talk to meyering, he seems to be doing something similar	13:26
feugan3333	doener_, Ok thanks. I thought receive.denyNonFastForward was supposed to do exactly that.	13:27
doener_	no, that prevents you from, well, non fast-forward pushes	13:27
	faq non-ff	13:27
Gitbot	doener_: Your push would lose changes on the remote. See http://git.or.cz/gitwiki/GitFaq#non-ff	13:27
	← marj left	13:28
	← ben_h left	13:28
	→ jburd2 joined	13:28
	→ ben_h joined	13:29
	← cilly left	13:29
doener_	feugan3333: rule of thumg: Don't push anything that was not dcommitted, or be prepared to deal with non-ff stuff and know what it means and how to deal with it	13:29
	→ ben_h_ joined	13:29
doener_	s/thumg/thumb/	13:29
jburd2	I'm trying to add a submodule with git version 1.6.0.1 using 'git submodule add ~/subtut/public/a' where ~/subtut/public/a is a git repository that exists, but I get an error instead: http://rafb.net/p/c5LaHU15.html	13:30
	ttuttle -> campus.	13:30
ttuttle	er	13:30
	← kef left	13:30
ttuttle	wrong window	13:30
Pieter	jburd2: you have the syntax wrong.. see man git-submodule	13:30
Gitbot	Pieter: the git-submodule manpage can be found at http://git.or.cz/man/git-submodule	13:31
doener_	jburd2: it's [add <repo> <path>], you only supplied <repo>	13:31
jburd2	It's exactly this line from the gitsubmodule tutorial: $ for mod in a b c d; do git submodule add ~/subtut/public/$mod; done	13:32
	Ok, I'll add the repo.	13:32
doener_	justacrab: says "git submodule add ~/git/$i $i" here: http://www.kernel.org/pub/software/scm/git/docs/user-manual.html#submodules	13:33
	← madewokherd left	13:33
jburd2	a crab?	13:33
	err, ok. heh. I get it.	13:34
doener_	oh, oops, tab-completion shows its love for me again	13:34
	← pehlert left	13:35
jburd2	So I guess someone should edit the submodule tutorial docs	13:35
doener_	how about you? :-)	13:35
jburd2	alright. :-)	13:36
	→ softdrink joined	13:36
feugan3333	doener_, I'm rather confused. The command that I'm using is git pull ../blah.git master:master and I'm getting that non fast forward error. The repo that I'm pulling from is not my origin, can I still use that git fetch command to see the differences?	13:40
	→ dramsay joined	13:40
Tv	feugan3333: don't fetch changes directly into local branches	13:40
doener_	feugan3333: sure. either add a remote for "../blah.git", or do "git fetch ../blah.git master", then FETCH_HEAD will refer to what you fetched	13:41
feugan3333	doener_, ok	13:41
doener_	totally OT: Anyone knows a good, small (7-8pt) font for coding? I'm getting kind of annoyed by terminus for some reason	13:42
	→ capisce joined	13:43
	→ stouset joined	13:43
	← GarethAdams left	13:45
feugan3333	Tv, What should I be doing? I'm tying to fetch changes from a group of developers (in a shared git repo) and commit those changes to an upstream svn server (using git-svn).	13:46
jburd2	doener_: ayuthaya, monaco, liberation mono, inconsolata	13:46
feugan3333	doener_, So I've done the fetch and am looking at the git log HEAD..FETCH_HEAD, I see a merge and several commits. Is the problem that it contains a merge?	13:48
doener_	feugan3333: I'd rather try looking at it with "gitk HEAD...FETCH_HEAD"	13:48
	feugan3333: maybe even with --left-right	13:48
	feugan3333: then you can see which commits are on each side and thus which you would lose	13:48
	← BarryCarlyon left	13:49
	→ peritus joined	13:50
	← capisce_ left	13:50
doener_	jburd2: hm, cannot seem to find ayuthaya anywhere. monaco and liberation aren't quite what I want, and while I've seen nice images of inconsolata, it looks like crap here, no matter what I do	13:51
	→ BarryCarlyon joined	13:51
doener_	jburd2: thanks anyway :-)	13:51
	← pengo left	13:52
christophsturm	is there an easy way to put a git repo into one file to mail it to someone?	13:53
	← GNUix left	13:53
Pieter	christophsturm: see git-bundle	13:54
	→ qrush joined	13:55
	← foca left	13:56
christophsturm	thanks	13:57
jburd2	doener_: Do you have font smoothing without hinting turned on for your terminal?	13:57
softdrink	Happy International Talk Like A Pirate Day!	13:58
doener_	jburd2: tried with AA on/off, subpixel stuff on/off and any hinting level that freetype offers	13:58
softdrink	ProFont is quite nice	13:58
	so is ProggyTiny	13:58
jburd2	doener_: You might want to edit /etc/fonts/conf.d/... to disable hinting. Subpixel hinting + no hinting looks pretty cool.	13:58
	Mikachu doesn't use xft fonts in the terminal, too slow	13:58
jburd2	But good for your eyes though. :-)	13:59
gittish	Mikachu types too fast :)	13:59
	→ ez joined	13:59
doener_	jburd2: at around 12pt, it looks bearable, though the subpixel stuff breaks coloring badly. But at 8pt, it's looking like broken pixels	13:59
softdrink	http://www.tobias-jung.de/seekingprofont/	13:59
jburd2	doener_: Can you show us a screenshot	13:59
gittish	doener_: depends on your DPI.	14:00
softdrink	see also: http://www.proggyfonts.com/index.php?menu=download	14:00
doener_	gittish: 94x92	14:00
gittish	see, to low	14:00
	← stouset left	14:00
gittish	with 150 it is still nice	14:01
	→ selckin joined	14:01
gittish	but then the screen is only 15"	14:01
doener_	gittish: buy me a new screen ;-) but anyway, Inconsolata at 8pt is even smaller than eg. Consolas at 8pt, so there's definitely something wrong	14:01
gittish	i assume you mean pt as it is ment, excluding any bugs in font definitions or wherever it is	14:03
softdrink	ProFont in xterm on mac os x: http://jocafa.com/temp/profont.png	14:04
	← mbuf left	14:04
softdrink	it's kindof a pixel font, but it's antialiased in all the right places	14:04
gittish	looks a bit fuzzy to me	14:05
doener_	gittish: I mean that pt that is about 0.139 inch. So I expect each font at 8pt to be about the same size	14:05
	jburd2: http://people.linux-vserver.org/~doener/xterm.png	14:07
gittish	doener_: yes, I understand. I've seen the phenomenon. Cannot comment on those fonts in particular since I don't have them installed.	14:07
	→ spearce joined	14:07
	← niki left	14:10
	→ stouset joined	14:11
bentob0x	would there be any reason why a git diff doesn't show me anything but in the commit there is a diff?	14:14
doener_	you already added the changes to the index	14:15
	try "git diff --cached" for a diff between the index and HEAD	14:15
bentob0x	git diff => nothing then git commit -a -v => diff in log	14:15
	ah	14:15
Pieter	or git diff HEAD	14:15
bentob0x	ok that works but what happened?	14:15
	in the index you mean in what will be comitted?	14:15
doener_	"git diff" => working tree against index, "git diff --cached" => index against HEAD, "git diff HEAD" => working tree against index	14:16
	→ cilly joined	14:16
bentob0x	ah ok	14:16
	just did the test	14:16
	funny I never spotted that	14:16
	← softdrink left	14:17
	→ softdrink joined	14:17
	→ bartocc joined	14:17
bentob0x	thx doener_, Pieter	14:18
jburd2	doener_: http://ifile.it/1qbk8xu/screenshot-weechat_0.2.6.png	14:20
	← rmh3093 left	14:20
jburd2	That's 8pt on my machine.	14:20
	← BugeyeD_ left	14:21
	→ sander__ joined	14:21
	→ BugeyeD joined	14:21
	→ Oompa joined	14:22
	← bartocc left	14:23
	→ priidu joined	14:24
doener_	jburd2: IIRC windows still uses a fixed dpi value to do font size calculations, so the font sizes are not necessarily correct	14:24
	← FunkeeMonk left	14:26
	← ponch84 left	14:28
	→ KiBi joined	14:29
	← cypromis left	14:30
	← softdrink left	14:31
	→ ponch_ joined	14:31
	→ Fullmoon joined	14:32
	→ softdrink joined	14:32
	ponch_ → ponch	14:33
	← BarryCarlyon left	14:34
	→ olauzon joined	14:34
	← cilly left	14:35
	→ elzurk joined	14:35
	→ kmap joined	14:35
elzurk	hello all, is there a cannonical hg -> git converter?	14:35
	→ cilly joined	14:36
Pieter	elzurk: yes, hg-fast-exprot	14:36
	*export	14:36
	→ mithro joined	14:37
	→ icwiener joined	14:37
elzurk	Pieter: thanks	14:37
	→ bcarlyon\|laptop joined	14:38
	← Sho_ left	14:39
	→ aroben joined	14:40
	→ bartocc joined	14:43
bartocc	hi all	14:43
	is there a way to tell git to "git rm" all the files that are marked as deleted, but not yet staged ?	14:44
vmiklos	git add -u?	14:44
doener_	git add -u	14:44
bartocc	great !!	14:45
	thx guys	14:45
	→ advany joined	14:48
	→ ciaran_ joined	14:48
	← johan-s left	14:49
	← Ademan left	14:49
	ebel → lucky5	14:49
	← snitko left	14:54
	← nud_ left	14:54
	→ Ace_NoOne joined	14:54
	→ bobesponja joined	14:54
Ace_NoOne	hi there - we recently changed a commit message in our SVN repo - is there a way to update my local git log without doing an all new cloning?	14:55
Mikachu	you can reset the tracking branch to before that commit and then refetch	14:56
	i think	14:56
Ace_NoOne	trying that	14:56
	← archis left	14:57
	→ nud_ joined	14:57
Ace_NoOne	also, when I just did a git svn fetch, it seemed to be fetching missing older revisions - how come those were missing from the clone in the first place?	14:57
	lucky5 → ebel	14:57
doener_	Ace_NoOne: from other branches?	14:57
	Ace_NoOne: "git svn rebase" is "smart" and only fetches the stuff it needs to work	14:57
	Ace_NoOne: so older revisions from other branches get fetched later	14:57
Ace_NoOne	doener_: oh I see - thanks	14:57
	← cilly left	14:58
Ace_NoOne	Mikachu: git reset <rev>; git svn rebase -- oddly enough, that still has the old commit message	14:58
Mikachu	Ace_NoOne: that's not what i said to do	14:59
	Ace_NoOne checks	14:59
doener_	Ace_NoOne: you need to reset the remote tracking branch, not just your local branch	14:59
Ace_NoOne	how do I do that	14:59
Mikachu	git push -f . <rev>:refs/remotes/foo	14:59
	← snos left	14:59
Ace_NoOne	hmm	15:00
	← ciaran left	15:00
	→ rtomayko_ joined	15:01
Ace_NoOne	Mikachu: git push -f . 111[...]0a2:refs/remote/trunk; git reset 111[...]0a2; git reset --hard; git svn rebase	15:03
	that didn't do it either	15:03
doener_	softdrink: btw, thanks, I'm using proggy tiny sz for now, let's see how that works out	15:03
softdrink	np	15:03
doener_	Ace_NoOne: "111...0a2" is just before the rewritten commit?	15:04
Ace_NoOne	yes	15:04
doener_	Ace_NoOne: try deleting .git/svn/$branchname/.rev_map.* as well	15:05
	→ Ilari joined	15:05
doener_	that's the only other thing that git-svn uses to keep track of svn revisions I know of	15:05
	← resmo_ left	15:06
Mikachu	you should ask if he uses nometadata before suggesting that :)	15:06
Ace_NoOne	that didn't work either - most curious	15:06
	→ cryo_ joined	15:06
doener_	Mikachu: hm, right. I always forget about that, because no-metadata only worked for one-shot imports for me, not with git svn fetch/rebase/dcommit	15:07
	→ CodeOfficer joined	15:07
	← reel_ left	15:07
doener_	Ace_NoOne: so you did the push/reset stuff _and_ the deletion, right?	15:07
Ace_NoOne	yes doener_ - deleted the revmap, then the command chain as described above	15:08
doener_	hm, weirdo	15:08
	ehrm, s/weirdo/weird/	15:08
Ace_NoOne	hehe I figured	15:08
shodges	lol	15:08
	← Oompa left	15:09
	doener_ needs to watch his fingers	15:09
	→ Oompa joined	15:09
Ace_NoOne	so what's this no-metadata thing	15:09
doener_	that stops git-svn from adding the git-svn-id lines	15:09
Ace_NoOne	I have those	15:09
	← koke left	15:10
doener_	Ace_NoOne: hm, "git push $oldrev:refs/remotes/branchname; rm .git/svn/branchname/.rev_map.*; git svn fetch" did it for me	15:10
Ace_NoOne	trying that	15:11
	neurodamage_away → neurodamage	15:11
	→ cilly joined	15:12
	← cilly left	15:12
	→ rdavila joined	15:12
Ace_NoOne	ahh I think I got it	15:12
doener_	ehrm, well, with the push command fixed, so it includes the "-f ." part	15:12
Ace_NoOne	fatal: The remote end hung up unexpectedly	15:12
doener_	yeah, that's the broken push...	15:12
Ace_NoOne	for git push 111[...]0a2:refs/remotes/trunk	15:13
	so that's ok?	15:13
doener_	git push -f . 111....	15:13
Ace_NoOne	hah nope, still not	15:13
doener_	btw, is refs/remotes/trunk actually correct for you?	15:13
Ace_NoOne	I would think so	15:13
	→ jackdempsey joined	15:14
doener_	I always use --prefix=svn/ with git-svn, so I get refs/remotes/svn/trunk	15:14
	← ciaran_ left	15:14
doener_	hm, try "git log remotes/trunk"	15:14
	does that still point to the old commit?	15:14
Ace_NoOne	nope	15:14
	that's the correct commit msg there	15:15
doener_	"correct" == the new one?	15:15
Ace_NoOne	yup	15:15
doener_	ah, ok, so you probably just looked at "git log", after the "git svn fetch", right?	15:15
Ace_NoOne	yes	15:15
doener_	the fetch only updates the remote tracking branch, not your local one	15:16
elzurk	anyone have a good text output for looking at the result of a merge? Kind of like a diff, but with left-parent, right-parent, common-ancestor and resulting file data in it	15:16
	→ krawek joined	15:16
elzurk	code reviewing merges is a PITA	15:16
doener_	try "git svn rebase" or just "git reset --hard remotes/trunk" now on your local branch	15:16
elzurk	at least with the current tools I'm using	15:16
Ace_NoOne	doener_: the reset did it	15:16
	thank you very much (I'm not entirely sure how we got here though - guess I got a lot more to learn about Git... )	15:17
	→ magpieu1 joined	15:18
	← mithro left	15:18
magpieu1	is it possible to reset a remote branch ?	15:18
	→ ciaran joined	15:18
	← ciaran left	15:19
	→ ciaran joined	15:19
doener_	elzurk: try "git diff-tree -p -c $commit" (or -m or --cc instead of -c, for different diff formats)	15:20
	→ mithro joined	15:20
	k-s[AWAY] → k-s	15:20
	← ciaran left	15:20
	→ ciaran joined	15:20
	→ Ryback_ joined	15:20
	→ ciaran_ joined	15:22
	← qrush left	15:22
	← bobmcw left	15:23
feugan3333	doener_, I figured out why my pull was failing, after a previous pull I did a git svn dcommit, and since it does a rebase my commit tree no longer looks like the original. So on the next pull things are mess.	15:24
	→ abjdshfsd joined	15:24
tokkee	How do I suppress messages like "268fb8f..fbe19d2 master -> origin/master" when doing git-pull?	15:25
	--quiet does not help.	15:25
	→ bryanl joined	15:25
	← kanru left	15:26
	→ jesselucas joined	15:26
Ilari	tokkee: If you are trying to script it, there is stuff like fetch-pack...	15:26
	← ez left	15:26
	→ kanru joined	15:27
	← olauzon left	15:28
tokkee	Ilari: Hrm ... so there is no way to shut up git-pull completely?	15:29
Ilari	tokkee: Well, git pull is designed to be used interactively, not via scripts...	15:30
	← nixbox left	15:31
tokkee	Ilari: Okay - thanks.	15:31
	→ bdiego joined	15:36
	→ clairvy_ joined	15:36
	→ gilimanjaro joined	15:36
	← jackdempsey left	15:37
	← bcarlyon\|laptop left	15:37
	← sander__ left	15:37
	← M_J_G left	15:37
	← irotas left	15:37
	← EmilMedve left	15:37
	← magpieu1 left	15:37
	← Lash\| left	15:37
	← hiffy left	15:37
	← krh left	15:37
	← tmm1 left	15:37
	← johannes left	15:37
	← andres left	15:37
	← Tv left	15:37
	← stick left	15:37
	← jettero left	15:37
	← thiago_home left	15:37
	← daniel-w left	15:37
	← Eridius left	15:37
	← mcr left	15:37
	← clairvy left	15:37
gilimanjaro	hello folks; anyone here who can help me get started with egit?	15:37
	→ krh joined	15:37
	← ponch left	15:37
	→ svend joined	15:37
	→ daniel-w joined	15:38
	→ ciaran__ joined	15:38
	→ jettero joined	15:38
	→ thiago_home joined	15:38
	← ciaran left	15:39
	← ciaran__ left	15:39
	→ ciaran joined	15:39
gilimanjaro	so i guess not eh	15:39
	→ jesselucas_ joined	15:40
tokkee	gilimanjaro: 1.) nobody can help you, if you don't ask any questions 2.) you should be just a little bit more patient than waiting for three minutes ...	15:40
	← kanru left	15:40
	→ Eridius joined	15:42
	← jammyd_home left	15:42
	← krh left	15:42
	→ stick joined	15:43
	→ magpieu1 joined	15:43
	→ jackdempsey joined	15:43
	→ bcarlyon\|laptop joined	15:43
	→ sander__ joined	15:43
	→ M_J_G joined	15:43
	→ irotas joined	15:43
	→ Lash\| joined	15:43
	→ mcr joined	15:43
	→ hiffy joined	15:43
	→ krh joined	15:43
	→ tmm1 joined	15:43
	→ johannes joined	15:43
	→ clairvy joined	15:43
	→ Tv joined	15:43
	→ andres joined	15:43
	→ tmm11 joined	15:43
	← ciaran left	15:43
	→ krh_ joined	15:43
	← jackdempsey left	15:43
	← sander__ left	15:43
	← irotas left	15:43
	→ jackdempsey joined	15:43
	→ qrush joined	15:45
gilimanjaro	tokkee: ok; sorry... the problem is a mysterious error when i try to 'Fetch From' on a seemingly correctly shared project. The first window properly lists the configured remote repository. But when I click next, the new window pops up an error: "Cannot get remote repository refs." with a 'String index out of range: -1'. I've a feeling something is not going well with the ssh connection jgit is using. The same connections works fine from a ter	15:45
	← Ace_NoOne left	15:45
spearce	gilimanjaro: can you look in the Error Log view and see if there is a stack backtrace for that exception?	15:46
	→ hiffy_ joined	15:46
	← segher left	15:47
xinming_	I have an git repo, when I use 'git archive' command, Is there a way to also specify it to ignore the .gitignore and gitmodules files?	15:48
	← johannes left	15:48
	← hiffy left	15:48
	hiffy_ → hiffy	15:48
gilimanjaro	spearce: (I'm relatively new to Eclipse & Git, just so you know...) I've opened the Error Log view (under Show View, Other), and managed to get the error again, but nothing shows up in my error log view sadly...	15:49
	← krh left	15:49
	← bcarlyon\|laptop left	15:49
spearce	fark. what is the url you are using for that remote?	15:50
jesselucas_	Anyone familiar with gitosis? I'm trying to figure out a way to give a group write access to all repositories. Basically writable = * . Is that possible with gitosis or do I have to add each new repository each time to the groups writable?	15:50
	→ sander__ joined	15:51
	→ andres_f joined	15:51
	← mcr left	15:52
	← andres left	15:52
	andres_f → andres	15:52
	← clairvy left	15:52
gilimanjaro	spearce: i have the same issue with 2 different ssh repo's. one is on my own server (boba:src/egit, my own clone of your project), and the other is internal to our company (blackhawk:git_tree/main)	15:52
	→ mcr joined	15:52
	← bartocc left	15:53
gilimanjaro	spearce: what they do have in common, is that both have relative paths on the ssh server. i just noticed this. let me try it with an absolute path	15:53
	← magpieu1 left	15:53
spearce	fwiw, relative should work. obviously you have found a real bug.	15:54
Tv	jesselucas_: writable = * would mean people create new repos if they typo the url; that is a non-goal for gitosis	15:54
	→ metze joined	15:54
	→ johannes joined	15:54
	→ uau joined	15:54
gilimanjaro	spearce: yes, absolute also fails	15:54
Tv	jesselucas_: it would also make everyone be able to write to gitosis-admin, which is just not a good idea	15:55
spearce	gilimanjaro: can you edit the egit code and retest with the modified copy?	15:55
gilimanjaro	spearce: any way you would like me to report this properly? any more info i can help you with?	15:55
spearce	we're missing an Activitator.logError("foo", cause) on line 234 of RefSpecPage	15:55
	that should at least give you the stack trace in the error view.	15:55
	← qrush left	15:55
gilimanjaro	spearce: ehrm... i guess. though i've only had experience with plugins through update-site's	15:56
spearce	there's like a thousand lines of code covered by that try/catch block so its hard to narrow down with the trace. :-\|	15:56
	→ niki joined	15:56
jesselucas_	Tv: Ahh, I guess I was just looking for a way to give just give an admin group the ability to write to any repository that was created up there. Then setup other groups with restricted access .	15:56
	← sergiodj left	15:56
	→ sergiodj joined	15:56
Tv	jesselucas_: well just have members include @admin for every group you config	15:57
vmiklos	heh	15:57
gilimanjaro	spearce: gotcha. i'll add that line then, and have a friend explain how to deal with local plugins. good learning experience for me :)	15:57
vmiklos	http://git.pastebin.com/ffe8ea1d	15:57
spearce	gilimanjaro: building the plugins is fairly simple; import the egit directory, make the edit, open the egit-updatesite/site.xml, click build all, then point your eclipse installer at that directory to install the plugins.	15:57
vmiklos	intresting:)	15:57
spearce	gilimanjaro: there is an issue tracker at code.google.com/p/egit if you can paste the stack trace there once you get it	15:57
gilimanjaro	spearce: I'll do that! thank you so far!	15:58
Pieter	vmiklos: nice :) and with more aggressive parameters?	15:58
vmiklos	i can put out the repo then you can play with it if you want ;)	15:59
gilimanjaro	spearce: I was just told it's beer'o'clock, so it will be later tonight (18:00 here)	15:59
Pieter	hmm.. i don' really care that much :)	15:59
vmiklos	i just remember from the mailing list that --aggressive is rarely a good idea. and now it was.	15:59
	it's a git-bzr import;)	15:59
Pieter	ah	15:59
	→ carrus85 joined	16:00
	→ EmilMedve joined	16:00
vmiklos	and given that bzr repo is about 100M i though it's impossible that the git one is 300+ :P	16:00
	← shodges left	16:00
Pieter	yeah.. I guess it might be more compressable	16:00
	git <-> bzr ratio is usually 1:3 or so for source repositories	16:01
	← boto left	16:01
	← rdavila left	16:01
Ilari	vmiklos: Even better (but very CPU intensive) would have been something like 'git repack -a -d -f --window=250 --depth=250'...	16:02
	→ rdavila joined	16:02
jesselucas_	Tv: Gitosis is amazing! I was just being really lazy and wanted to setup the gitosis-admin's with the ability to have writable to all repositories lol. Just so I didn't have to edit the gitosis.conf every time I created a test repo, etc..	16:02
	clairvy_ → clairvy	16:03
vmiklos	Pieter: btw, when estimating the size of the bzr repo, is it fair to just du -sh the .bzr dir as well?	16:04
Pieter	vmiklos: probably.. though you could perhaps check if there are some .bzr/*backup files there	16:05
vmiklos	there are none.	16:06
	← ciaran_ left	16:06
Pieter	what project is it?	16:07
vmiklos	gnash	16:08
Pieter	a	16:08
vmiklos	actually what i care about a lot more is cloning time, and this bzr over http is slow as well ;) (especially since i don't have to use git over http)	16:08
Pieter	do they have a large history?	16:08
	cloning bzr over * is slow	16:08
vmiklos	$ git rev-list HEAD\|wc -l	16:09
	10626	16:09
	but i guess the repo is large (well, not that large) is because of the tests	16:09
	those are binary swf files	16:09
	← tmm1 left	16:10
bobesponja	is there a way to know if a branch is fast-forward without merging it?	16:10
Tv	bobesponja: git rev-list HEAD ^branch is empty	16:10
bobesponja	Tv: thanks	16:11
thiago_home	that means HEAD is a subset of "branch"	16:11
vmiklos	"list commits which are in HEAD, but exclude the ones in branch"	16:12
	erm, you did not ask, you said:)	16:12
tokkee	Hrm ... I'm unsure if I understood fetch-pack correctly: Is it my responsibility to update the refs in refs/remotes/?	16:12
	← advany left	16:13
	← rtomayko_ left	16:13
spearce	tokee: if you are calling fetch-pack directly, yes	16:13
	← com4 left	16:14
bobesponja	thiago_home: so is there a way to do so?	16:14
thiago_home	do what?	16:15
tokkee	spearce: Okay - and I assume I just need to strip off refs/heads/, etc. from the refs returned by fetch-pack?	16:16
thiago_home	bobesponja: aside from what Tv told you?	16:16
Zdra	Hi, is it possible to make git-svn respect the author of the git commit when pushing to SVN?	16:16
thiago_home	Zdra: it's not git-svn's fault	16:16
	Zdra: it's SVN's fault.	16:16
	Zdra: to solve that problem, stop using SVN.	16:16
spearce	tokkee: why strip them? the old fetch process just passed the whole name to git-update-ref	16:16
	→ mokmok joined	16:16
Ilari	tokkee: 'refs/heads/foo' is local branch 'foo'.	16:16
spearce	tokkee: oh, wait i take that back	16:17
mokmok	hi all, anyone know how to pull up to a specific tag?	16:17
bobesponja	thiago_home: yes	16:17
spearce	yes, you need to mangle refs/heads/foo into refs/remotes/origin/foo	16:17
Ilari	tokkee: 'refs/tags/foo' is tag 'foo'.	16:17
Zdra	thiago_home: I cry everyday for GNOME to stop using the worst CVS ever, but nobody wants to move...	16:17
	→ bronson joined	16:17
thiago_home	bobesponja: where did Tv's solution not work?	16:17
Zdra	so there is no hack or magic stuff to do?	16:17
	it's just impossible?	16:17
	→ zpiro joined	16:17
tokkee	spearce, Ilari: Good - so I was assuming right ;-) Thanks.	16:17
thiago_home	Zdra: I have talked to some GNOME devels who were trying to convert to Git.	16:17
	for GNOME it's actually easy because they don't have modules the same way we do in KDE.	16:18
	so you can't have copy-with-history and the layout is sane.	16:18
feugan3333	I'm sure you could write a hack with commit hooks.	16:18
bobesponja	thiago_home: it just sounded like from your answer like it wouldn't always work so I was just wondering, so far it's working here	16:18
thiago_home	bobesponja: no, I was interpreting it	16:18
	← etgryphon left	16:19
bobesponja	ok, thanks	16:19
thiago_home	bobesponja: a fast-forward is only possible when the current head is a strict subset of the new head	16:19
	that's exactly what git rev-list HEAD ^branch does	16:19
	→ sdboyAAARR joined	16:19
bobesponja	k	16:19
thiago_home	you can also do git merge --ff and check if it worked	16:19
	actually, no, that's not what --ff does	16:20
	ZorbaGama → ZorbaTHut	16:20
	← nud_ left	16:21
Ilari	bobesponja: One way to check wheither A is strict subset of B is to check wheither their merge-base (git merge-base) is SHA1 of A.	16:21
	jburd2 → jburd	16:22
jesselucas_	What is the easiest way to change the remote location for a multiple sub modules? Basically my super project uses submodules all from the same server and I'm trying to figure out an easy way to switch all the remote locations.	16:22
	← feugan3333 left	16:24
	→ tomtt joined	16:26
tomtt	I would like to do the following: get a list of all the files in some/subdir and some/other/subdir that were changed in rev aaa until fff. My git ninja skills are not developed enough yet to figure this one out...	16:29
	← mokmok left	16:31
spearce	tomtt: git log --name-status aaa ^fff -- some/subdir some/other/subdir ?	16:31
	or maybe git log --raw -r ... ?	16:32
	→ WALoeIII joined	16:32
Mikachu	tomtt: if you don't care if they changed and then changed back again, git diff --name-only aaa fff	16:32
spearce	true, but then you lose the commits they were changed by.	16:33
	depends entirely on what tomtt needs. :)	16:33
Mikachu	he didn't say he wanted the commits :)	16:33
spearce	i assumed he did for some reason. perhaps more coffee is required.	16:34
tomtt	Mikachu: i think your approach is good enough for me	16:34
vmiklos	Ilari: hm yes, 59MB->48MB, not that much difference, though.	16:35
	(compared to 300+->59 :) )	16:36
tomtt	Mikachu: can i narrow it down to only those commits that I made?	16:36
Mikachu	not with diff	16:36
thiago_home	Ilari: right, but that's a more expensive operation	16:36
	← svend left	16:36
thiago_home	Ilari: git rev-list -1 $newrev..$oldrev	16:37
	that works best	16:37
	← charon left	16:37
tomtt	hmm git-log --name-status is too verbose, diff can't narrow it down to just my commits...	16:38
doener_	specify a revision range, or use --author, depending on what you mean by "just my commits"	16:39
	← lu_zero left	16:39
	warthog19 → warthog9	16:39
Mikachu	tomtt: i think log takes --name-only too, maybe	16:39
	% git log --name-only --pretty=format: HEAD\~50 --author=mikachu\|sort -u	16:40
	that kind of works i guess	16:40
	→ radarek joined	16:40
	← M_J_G left	16:41
	→ bts-_ joined	16:41
	← bts- left	16:41
	→ thannoy joined	16:43
	← Man-erg left	16:47
tomtt	Mikachu: sweet, thx: this suits me just fine: git log --name-only --pretty=format: aaa..HEAD --author=Tom\|sort -u	16:48
	← mw left	16:49
	← ben_h_ left	16:49
	→ ben_h_ joined	16:49
	→ markkalderon joined	16:51
	← rdavila left	16:51
	→ stouset_ joined	16:51
	→ mw joined	16:53
	← stouset left	16:53
	← BabelO left	16:53
	← LostInTheWorld left	16:56
	→ segher joined	16:56
	→ rmh3093 joined	16:57
	← capisce left	16:57
	← tomtt left	16:57
	← peritus left	16:57
	→ capisce joined	16:59
	← warlock_mza left	16:59
	sdboyAAARR → DrunkOnGrog	17:00
	← tvw left	17:00
	→ tvw joined	17:01
	← Zarin left	17:01
	← ruphy left	17:02
	← mcella left	17:02
	→ BabelO joined	17:02
	→ vintrepid joined	17:03
	← powr-toc left	17:04
	→ bobmcw joined	17:04
	→ joydivider joined	17:05
	→ Eludias joined	17:07
	→ etgryphon joined	17:07
	w00t_ → w00t	17:09
	→ huperniketes joined	17:09
	← blowery left	17:09
	blowery_ → blowery	17:09
	← ben_h_ left	17:09
huperniketes	how do I checkout a specific branch when I clone a repo?	17:09
	← ben_h left	17:10
Tv	huperniketes: man git-checkout	17:12
Gitbot	huperniketes: the git-checkout manpage can be found at http://git.or.cz/man/git-checkout	17:12
	→ AI0867 joined	17:14
huperniketes	Tv: so I should clone w/o a checkout (git-clone -n), then do the checkout of the specific branch separately?	17:14
Tv	huperniketes: no	17:14
	→ johnjosephbachir joined	17:15
Tv	huperniketes: what you should do is read the tutorial	17:15
	← johnjosephbachir left	17:16
huperniketes	Tv: I read the tutorial, and I've been able to do checkouts from remote repos, but this situation is different	17:17
Tv	you can't checkout remote repos	17:17
	→ com4 joined	17:17
	← gittish left	17:17
AI0867	git-am is stripping the ')' character from my name	17:17
	author Alexander van Gessel (AI/AI0867 [email@hidden.address] 1221835717 +0200	17:17
	committer Alexander van Gessel (AI/AI0867) [email@hidden.address] 1221843665 +0200	17:17
	→ ciaran joined	17:17
	→ blowery_ joined	17:17
AI0867	is this intentional or am I doing something wrong?	17:18
huperniketes	Tv: In the present case, someone sent me a repo which I put in the dir where I keep my repos from other SCMs	17:18
	← priidu left	17:19
Tv	huperniketes: how does that change anything?	17:19
	← ebel left	17:21
huperniketes	Tv: now I'm trying to do a checkout of a non-master branch, but git-checkout doesn't accept the branchname because the working dir hasn't been set up yet	17:21
Tv	if you got a bare repo from someone, clone it	17:21
ZorbaTHut	okay, I got a weird question. I have a git repo, I did some work on it. the last eight commits or so turn out to be unnecessary, and I want to remove them, without them being in the main history at all. however, I do want to keep that little fork in case I do need it at some point in the future	17:22
	how can I go "back in time" and redefine an old point as the new master, preserving the no-longer-needed commits as a branch?	17:23
Tv	ZorbaTHut: git branch -m master just_in_case && git checkout -b master HEAD~8	17:23
	← KevBurnsJr left	17:24
AI0867	Tv: any input on my issue? should I do something differently or can I fix this somehow?	17:24
Tv	AI0867: i don't use git am	17:24
	AI0867: email the list with copy-paste of a minimal session that reproduces the bug	17:25
	← EmilMedve left	17:25
	→ srid joined	17:25
AI0867	to [email@hidden.address]	17:26
ZorbaTHut	tv, done, but when I push, it gets rejected	17:26
	→ ponch_ joined	17:26
Tv	ZorbaTHut: it seems you already pushed your unwanted changes	17:26
ZorbaTHut	("non-fast forward" seems to be the error)	17:26
Tv	ZorbaTHut: sucks to be you	17:26
AI0867	ZorbaTHut: you already published those 8? because rewriting history now will get you conflicts everywhere	17:26
	→ eikonos joined	17:27
Tv	AI0867: he's not rewriting	17:27
AI0867	resetting	17:27
	almost the same thing	17:27
ZorbaTHut	I'm not rewriting, I want to switch the current "master" to be something else, and take over "master" with a new branch	17:27
Tv	ZorbaTHut: "git push origin +master" will overwrite, but that's dangerous if someone else already fetched the changes	17:27
ZorbaTHut	I know nobody else has because the public access has been broken since before I started this, so at least that's convenient :P still, no clean way to do it?	17:27
Tv	what did i just say?	17:28
capisce	+ forces a push?	17:28
Tv	capisce: man git-push	17:28
ZorbaTHut	well, you said "sucks to be you", which indicates that you're feeling kind of smug about the whole thing. I thought I'd double-check because there might be a "more difficult" way to do it that you're just not telling me.	17:28
Gitbot	capisce: the git-push manpage can be found at http://git.or.cz/man/git-push	17:28
ZorbaTHut	I've seen it happen before, and I've occasionally gotten better answers by doublechecking :P	17:29
com4	can i import my git repository into a new blank svn repository?	17:29
Tv	com4: svn can't represent git history, in general	17:29
com4	so i can only commit changes to svn from git?	17:30
Bryson	git checkout <sha hash for revision number> -- <filename of file to revert>	17:30
	how do i tel that what branch to check that file out to?	17:30
Tv	Bryson: did you read man git-checkout	17:31
Gitbot	Bryson: the git-checkout manpage can be found at http://git.or.cz/man/git-checkout	17:31
	→ juanjoc joined	17:31
	→ svend joined	17:31
Ilari	com4: There's git2svn, don't know how well it does the conversion...	17:33
com4	Ilari: Cool. Thanks	17:34
	DrunkOnGrog → sdboyAAAAAR	17:36
	→ danten joined	17:36
	← srid left	17:40
	← hurikhan\|Work left	17:40
	→ mcella joined	17:41
	→ charon joined	17:43
	← mcella left	17:44
	→ foca joined	17:44
	→ tommorris joined	17:47
	→ ruphy joined	17:49
	jburd → po	17:49
	← po left	17:51
	→ Dvyjones joined	17:51
	← BabelO left	17:51
	← Fullmoon left	17:52
Dvyjones	I just installed git 1.6.0, but I can't find git-config. Is that a seperate program?	17:52
doener_	"git config", no dash	17:52
	Dvyjones: see first paragraph in the 1.6.0 release notes. http://www.kernel.org/pub/software/scm/git/docs/RelNotes-1.6.0.txt	17:52
	→ xjjk joined	17:54
	→ jburd joined	17:55
jburd	Can one get git to automatically git submodule init; git submodule update on clone/pull?	17:55
	→ jburd1 joined	17:57
AI0867	Tv: mail sent to git@vger	17:57
	← tvw left	17:59
	← AI0867 left	18:01
	→ jnareb joined	18:01
	jnareb is ShadeHawk is jnareb is JakubNarebski is Jakub Narębski	18:02
huperniketes	can I get git-clone to copy all the branches from a local repo, or should I just cp it?	18:03
doener_	"git clone" does transfer all local branches, they're stored as remote tracking branches (see "git branch -r")	18:05
	← danten left	18:06
huperniketes	doener_: ah, it's in origin/ tks	18:07
	→ stouset joined	18:07
	→ kixx joined	18:07
jnareb	unless of course you would do a bare, or mirror clone	18:07
	but bare additionally mean _without working area_	18:07
	you use bare repositories as publishing repositories, not repositories you do work in	18:08
	→ danten joined	18:08
huperniketes	jnareb: right. someone transferred maintenance of the repo to me, which I now keep in my repo store	18:09
jnareb	Hmmm... number of respondents per day in Git User's Survey 2008 (http://tinyurl.com/GitSurvey2008) has fallen to around 20 per day and lower... but this is expected	18:09
huperniketes	jnareb: so I'm trying to set up a working dir for a particular, non-HEAD, branch	18:09
jnareb	git checkout <branch>	18:10
	→ mediogre joined	18:10
jnareb	or if you want to create a new branch and start using it, "git checkout -b <newbranch>"	18:10
	If you want to work based on some remote branch, then "git checkout -b <branchname> origin/<branchname>"	18:11
	→ _graham_ joined	18:11
jnareb	you would have local branch <branchname> following remote-tracking branch origin/<branchname> (which follows <branchname> at origin)	18:12
	HTH	18:12
	→ akahn joined	18:12
akahn	What's the command for taking my repo back to a previous revision?	18:12
Pieter	git reset	18:13
doener_	your repo? you can move your branch head, or checkout some older revision	18:13
Pieter	or git checkout	18:13
jnareb	"git reset --hard ORIG_HEAD" <-- user's beware, read man git-reset first	18:13
Gitbot	jnareb: the git-reset manpage can be found at http://git.or.cz/man/git-reset	18:13
jnareb	I think	18:13
sdboyAAAAAR	anybody know how to get git-cvsimport to work directly into a bare repo? thus far i've had to git-cvsimport into a normal repo then push that into a bare, i'd like to eliminate the intermediate step if possible	18:13
jnareb	try other cvs import tools, if possible, perhaps	18:14
	← jburd left	18:14
drizzd_	umh, which branch contains v1.6.0.2?	18:14
sdboyAAAAAR	other tools, as in, cvs2git? that's the only other one i'm aware of	18:14
jnareb	cvs2git (i.e. cvs2svn, git output), parsecvs; see http://git.or.cz/gitwiki/InterfacesFrontendsAndTools	18:16
sdboyAAAAAR	jnareb: great. thanks	18:17
jnareb	drizzd_: I guess 'maint'	18:17
drizzd_	that's what I thought, but it doesn't	18:17
jnareb	sdboyAAAAAR: not sure if it would help; from what I know only git-cvsimport is incremental although worst (from what I heard) in dealing with CVS quirks	18:17
sdboyAAAAAR	jnareb: it does get a little quirky, yeah :/ but i've been able to deal as thus far it's been consistently quirky	18:18
jnareb	git name-rev --refs refs/heads/* v1.6.0.2	18:18
	← niki left	18:18
jnareb	v1.6.0.2 maint~19	18:18
Ilari	drizzd_: v1.6.0.2 ^origin/maint contains no commits...	18:18
drizzd_	it does for me, somehow...	18:19
	→ mattf joined	18:20
akahn	Pieter: how do I specify a particular commit to reset to?	18:22
Pieter	akahn: git reset <commit>	18:22
	← harinath left	18:22
akahn	Pieter: right, what does <commit> actually look like?	18:23
Eridius	huh, interesting. when git-config parses a file, it treats ^M as a space	18:23
	← Eridius left	18:23
	→ Eridius joined	18:24
	→ prof_freud joined	18:24
Eridius	I'm not sure if that last message went through: huh, interesting. when git-config parses a file, it treats ^M as a space	18:24
	→ schacon joined	18:25
Ilari	akahn: It can be full 40-digit SHA1, any unique prefix of it, any ref (including name of branch) and any valid expression can be postfixed by nth parent operator (^n), 1st parent operator (^) or n iterated first parents operator (~n)...	18:26
	akahn: Maybe look at 'git-log' or 'gitk' and copy the commit ID displayed there...	18:27
	→ rutski joined	18:28
	→ thorny_sun joined	18:28
rutski	I'm going to be learning git by developing a project locally (no public repo)	18:28
	← jnareb left	18:28
	← jburd1 left	18:28
rutski	but then in the future I do eventually want to upload the repo to a public server	18:28
	← ciaran left	18:28
	→ daaku joined	18:28
Ilari	rutski: Not a problem...	18:28
rutski	however, my local machine has git-1.6, while the server I eventually want to upload to has 1.4	18:28
thorny_sun	i would i go about populating a subdir with its contents as they were in a previous commit?	18:29
Eridius	hrm, turns out it interprets any whitespace char as a single space	18:29
Ilari	rutski: I think that should work...	18:29
rutski	Ilari: is your answer the same even after the my mentioning the version difference?	18:29
	ah, ok; cool	18:29
	← markkalderon left	18:29
	← mithro left	18:30
	→ nud_ joined	18:31
Ilari	rutski: Well, at least as long as smart protocols are used. Using push over HTTP could lead to trouble if client uses stuff that server does not understand...	18:31
	← stouset_ left	18:31
rutski	indeed	18:31
	I don't plan to play with HTTP based serving though	18:31
	also, I can always circumvent the server's package manager and just install 1.6 from source I suppose	18:32
	← ijcd left	18:33
	← LiamH left	18:33
	← kixx left	18:34
	→ botanicus joined	18:35
	→ koke joined	18:36
	→ gbucher joined	18:36
gbucher	Hi ! I want to merge part of a file (a chunk) from one branch to the other. How can I do this ?	18:36
Ilari	rutski: Well, the server could have trouble if the project is extremely large (>4GiB repo)...	18:37
rutski	oh, heh; no way	18:37
	it's not a problem in my case	18:38
	but I wonder if it's an issue with large game projects that include lots of media	18:38
drizzd_	my experiments with git fetch --depth apparently caused my problems above. everything is fine with a new clone now.	18:38
rutski	perhaps they don't bother storing the median in the scm though?	18:38
Ilari	rutski: Git does not like binary files very much. Especally if they are large, compressed or encrypted. Also, merging binary files could be nasty...	18:39
rutski	definitely	18:39
	I suppose keeping the media elsewhere can't hurt	18:39
gbucher	Any clue to merge a chunk from one branch to the other ?	18:40
Ilari	rutski: The delta algorithm can deal with binary files, but only if they are "code-like" in changes (to file itself).	18:40
	→ neoeinstein joined	18:40
	tmm11 → tmm1	18:40
	→ Fullmoon joined	18:41
rutski	which they probably aren't if they're video or image files	18:41
Ilari	rutski: BMP files are pretty much code-like. GIF, PNG and pretty much all videos files are not...	18:41
rutski	ah, good to know	18:42
Ilari	rutski: As said: Compression. In uncompressed and unencrypted stuff, localized changes to what file represents tend to cause localized changes in file itself. If it is compressed, everything tends to change...	18:43
rutski	ah, right; good point	18:43
	→ mithro joined	18:44
	→ polpak joined	18:44
	← floppyears left	18:44
	→ komplex joined	18:44
Ilari	rutski: There are filters support, so one can dissect the file (if one knows the format) and internally uncompress it (I have done this for OOo documents)...	18:44
	← komplex left	18:45
Ilari	rutski: Don't store big files. They really cause pain. Compressed stuff just tends to bloat the repository. Big stuff is much worse...	18:45
	← Fullmoon left	18:46
	← warthog9 left	18:47
	→ andersbr joined	18:47
	← adamm left	18:47
	← jackdempsey left	18:48
akahn	Ilari: thanks	18:49
	← DrFrasierCrane left	18:49
	→ Tammy-- joined	18:51
andersbr	n00b question: I set up a repo on a remote server, cloned to a local dir, made some changes, did a commit, then a push. I expected it to update the remote repo, but it didn't. What was my mistake?	18:51
	It says "everything up to date", which doesn't seem to be true.	18:52
charon	faq non-bare	18:52
Gitbot	charon: You shouldn't push to a non-bare repository. See http://git.or.cz/gitwiki/GitFaq#non-bare	18:52
	→ kixx joined	18:53
andersbr	I believe it is bare.	18:53
	→ BarryCarlyon joined	18:53
Ilari	andersbr: How did you tell it "didn't update"?	18:53
	→ entie joined	18:54
andersbr	Ilari: deploying the app through Capistrano.	18:54
	← icwiener left	18:54
capisce	?	18:54
andersbr	Capistrano, ruby-based tool for deploying web apps. It does a 'git clone -q' from the remote repo to check out the latest version of the code.	18:55
	→ gittish joined	18:56
	← gbucher left	18:56
Ilari	andersbr: What if you run 'git log' on server, does it show the new commits?	18:57
	← ponch_ left	18:58
andersbr	Ilari: Nope. :)	18:58
	← nud_ left	19:00
	→ etgryphon_ joined	19:02
	→ rlorandi joined	19:03
	← lorandi left	19:03
	→ floppyears joined	19:04
	← thorny_sun left	19:04
	← entie left	19:04
	→ ice799 joined	19:05
ice799	Hi - I am a total git newb, but I am trying to setup a really simple backup script that will clone my git repo to a backup cifs store that i have mounted on my box	19:06
	→ Fullmoon joined	19:06
	→ nacho joined	19:06
nacho	Hi	19:06
ice799	now when my script says git clone .... to clone the repo, everything works fine except that the permissions on the files are all changed	19:07
	so all the files show up as a modified and then running git pull afterwards fails. is there some way of telling git pull that i don't care about file permissions ?	19:07
nacho	anybody knows howto make a git-svn dcommit with a different transfer method ? Right now I only can make git-svn rebase with url = http://.... but if I change it to svn+ssh:// it does not work	19:07
MaddHatter	ice799: the only permission it looks at, iirc, is the execute bit	19:08
	andres → andrezZz	19:08
	andrezZz → andres	19:08
MaddHatter	so just don't make stuff executeable that's not (and vice-versa)	19:08
andersbr	I think the problem was, I did 'git commit' without the -a flag. I just figured that was the default behavior.	19:09
ice799	MaddHatter: I think the issue is that cifs shares dont respect unix permissions because they are network storage -- they don't know about the uid/gids on the box	19:09
	→ loincloth joined	19:09
	← akahn left	19:09
ice799	MaddHatter: So I'm not sure how to fix this issue.	19:09
MaddHatter	yeah, but git doesn't care about uids/gids	19:10
	→ foca_ joined	19:10
doener_	ice799: git only cares about +/-x, uid/gid don't matter	19:10
ice799	MaddHatter: but I think you are right about the execute bit, because git diff says: old mode 100644 new mode 100755	19:10
doener_	ice799: try "clone --mirror" and "git fetch" instead of "git clone" and "git pull"	19:10
	← loiseau left	19:11
ice799	doener_, MaddHatter: Yeah I understand what you are saying, but what I'm saying is that I think any files written to the cifs share will automagically be given those permissions	19:11
	simply because the cifs share is dumb.	19:11
MaddHatter	as for a solution, tell cifs to stop making everything executable, or just chmod -R a-x . all the time	19:11
	samba can be told to be less dumb in that regard	19:11
doener_	ice799: that creates a bare repo (no working tree at all), which is just fine for backup purposes	19:11
schacon	is there any way to emulate a subversion interface to a git server? like git-cvsserver, but with svn?	19:11
ice799	MaddHatter, doener_: Thanks I'll play around a bit with cifs to see if I can make it not dumb.	19:12
	← svend left	19:12
	← synsol left	19:14
doener_	ice799: I'd really look into "clone --mirror" instead. That is probably better than a "normal" clone, and has no x-bit problems	19:14
	→ anttih joined	19:19
	← etgryphon left	19:19
anttih	what permissions do I need to set to a bare git repo in order for the group `git` to run read operations on it?	19:20
	g+r?	19:20
	← etgryphon_ left	19:21
anttih	that doesn't seem to be enough	19:21
ice799	doener_: I am checking it out now.	19:21
charon	nacho: when changing the git-svn urls (not just for protocol changes but also moves and such) you need to rewrite the git-svn-id: lines with git-filter-branch --msg-filter, then rm -rf .git/svn/ so it regenerates its metadata cache	19:23
	→ jackdempsey joined	19:23
Ilari	anttih: Are you sure some files don't get wrong permissions?	19:24
charon	nacho: both are quite destructive so please make a backup, but i've done it in the past, it works.	19:24
anttih	Ilari: with g+r?	19:24
	← vbabiy left	19:24
Ilari	anttih: Are all files g+r with group 'git'?	19:24
anttih	Ilari: what do you mean?	19:24
nacho	charon, but if I remove the .git/svn is it going to work ?	19:24
Ilari	anttih: In the repository (.git), that is...	19:25
charon	nacho: well, removing .git/svn makes it look at git-svn-id again	19:25
	which is where gfb comes in	19:25
anttih	Ilari: they are yes. but I get the "not a git repository error"	19:25
charon	be sure to use -- --all so it also rewrites the git-svn remote	19:25
anttih	but with another user the command works	19:25
charon	unfortunately i have to catch a train so i can't talk you through it, but i'm sure someone else can help with gfb usage if you need more assistance	19:26
	good luck :)	19:26
	← charon left	19:26
nacho	charon, thansk	19:26
	← mediogre left	19:27
	← mithro left	19:27
	→ lima joined	19:30
nacho	anybody can help me with what charon said?	19:30
	I removed the .git/svn dir but now I get: Unable to determine upstream SVN information from HEAD history.	19:30
	This is happening me when I change the url from http:// to svn+ssh	19:31
	not sure how to fix it	19:31
	← foca left	19:31
	→ venger joined	19:32
	→ DrFrasierCrane joined	19:33
	→ mithro joined	19:34
	→ qrush joined	19:35
	→ vbabiy joined	19:35
	→ nacho_ joined	19:35
	← jackdempsey left	19:36
	← nacho left	19:36
	← stouset left	19:37
	← prof_freud left	19:38
	→ jackdempsey joined	19:38
	← Lash\| left	19:41
ice799	doener_: why did you suggest that I do a git fetch after going a git clone --mirror? I'm unclear why doing a fetch is worthwhile there... Sorry, I'm a huge git newb.	19:42
	→ EmilMedve joined	19:47
venger	is `git branch -r -D origin/branchname` the proper way to delete a remote branch? git does output "Deleted remote branch origin/branchname" and git branch -r reflects that its missing from the list, however a git fetch brings it right back as it was.	19:47
DrNick	no	19:47
	git push origin :branchname	19:48
venger	with the ':' ?	19:48
DrNick	yes	19:49
	the normal syntax is git push remote local_branch:remote_branch, which sets remote_branch to local_branch	19:49
	but in this case you're setting remote_branch to an empty ref, which the (somewhat logical) way to delete remote_branch	19:49
venger	alright that did the trick. its odd the other command would report successful deletion though and not do something else	19:53
	← BarryCarlyon left	19:53
	← qrush left	19:55
	← ToxicFrog left	19:55
xinming_	After understanding some of git's internal work, I'd say, It's really a good design for implementing vcs such way.	19:58
	→ markkalderon joined	19:58
Ilari	xinming_: AFAIK, Git originally started as 'build-your-SCM'-kit.	19:59
xinming_	BTW, for .git/objects/pack/, How do I combine 2 packs into one?	20:01
	→ LostInTheWorld joined	20:01
xinming_	Oh, -a option to git-repack	20:01
	thanks	20:01
Ilari	xinming_: 'git repack -a -d'?	20:01
	→ advany joined	20:02
	→ jerbear joined	20:03
xinming_	Ilari: Thanks, Only -a not worked. :-)	20:03
Ilari	xinming_: -a without -d is pretty strange combination. Make a new pack, but don't delete anything...	20:04
	xinming_: New pack collecting entiere reachable history...	20:04
Mikachu	Ilari: maybe if you want to be nice to existing http:// clones	20:05
	though that doesn't make sense now that i think	20:05
	→ stouset joined	20:06
	→ nbrunson joined	20:07
	← spearce left	20:08
	→ spearce joined	20:09
	→ ToxicFrog joined	20:12
	← nickh left	20:12
	← Tammy-- left	20:14
	← andreaja left	20:15
	→ bbuss joined	20:16
	→ avu joined	20:17
	← Bryson left	20:17
	← avu left	20:18
	→ BarryCarlyon joined	20:19
	→ boto joined	20:20
	→ koke_ joined	20:21
	→ hurikhan\|Work joined	20:21
	← koke left	20:21
	→ warthog9 joined	20:25
	← markkalderon left	20:26
	← polpak left	20:26
	→ nickh joined	20:27
	← hurikhan\|Work left	20:29
	→ markkalderon joined	20:30
	← gittish left	20:31
	→ bcarlyon\|laptop joined	20:34
	← bentob0x left	20:34
	→ robinr joined	20:34
	← mithro left	20:34
	← bbuss left	20:36
	→ capisce_ joined	20:36
	→ bbuss joined	20:38
	← BarryCarlyon left	20:39
	bcarlyon\|laptop → BarryCarlyon	20:39
	← DrFrasierCrane left	20:39
	← capisce left	20:40
anttih	Ilari: I just can't make the repo available for the group	20:41
	→ Millow joined	20:42
anttih	any ideas?	20:42
	→ ukleinek joined	20:43
Millow	Hello, anyone here?	20:43
	← Fullmoon left	20:43
schacon	Millow: how you doing?	20:43
Millow	! [rejected] master -> master (non-fast forward)	20:43
	error: failed to push some refs to '****:proj.git'	20:43
	→ tiglionabbit joined	20:43
	→ bryanl_ joined	20:43
Millow	I'm getting this error when trying to push to my repository server	20:43
Pieter	Millow: see faq non-ff	20:43
Gitbot	Millow: Your push would lose changes on the remote. See http://git.or.cz/gitwiki/GitFaq#non-ff	20:43
schacon	generally means you have to pull and then push again	20:44
tiglionabbit	I have all these files that are "deleted" under "Changed but not updated". How do I get this information committed?	20:44
	→ mithro joined	20:44
Millow	thx	20:44
Pieter	tiglionabbit: use 'git add -u' to stage them	20:44
	← vintrepid left	20:44
anttih	do you guys know which permissions do I need to set for a bare repo so that the group who ones it can access it? basicly just read stuff.	20:45
ukleinek	I want to test a modification to topgit. Is there an easy way to do so without the need to install it somewhere?	20:45
anttih	*owns it	20:45
	→ tiglionabbit_ joined	20:46
ukleinek	anttih: so the group should be able to commit, or read only?	20:46
anttih	ukleinek: read only	20:47
ukleinek	so you can just do chmod -R g=o-w $repo	20:47
anttih	ukleinek: right now, I get "fatal: Not a git repository: "	20:47
tiglionabbit_	Pieter: that's just giving me a usage response	20:47
	what is it expecting?	20:47
Millow	Pieter: well, that didnt help much :(	20:47
	jackdempsey → jack_afk	20:48
ukleinek	anttih: you have a repo and another group member tries to clone it, right?	20:48
anttih	ukleinek: nope, just basically `log`	20:49
	well, I figured this has something to do with permissions but I'm not sure	20:49
ukleinek	anttih: so the group member does: cd $repo; git log; right?	20:49
anttih	ukleinek: he can't even `cd` :-(	20:50
	→ pjhyett joined	20:50
Millow	Iäm getting the non-ff error all the time, even if I fetch first and get some changes	20:50
ukleinek	anttih: then some parent dir misses executable permissions for the group member	20:51
	← tcoppi left	20:51
anttih	ukleinek: ah, let me check	20:51
	← statim left	20:52
	← bobmcw left	20:52
	→ statim joined	20:52
	← capisce_ left	20:53
anttih	ukleinek: yep, that did it. thanks!	20:54
	→ capisce joined	20:54
	→ tcoppi joined	20:55
	→ agib joined	20:56
	← cannonball left	20:56
	← agib left	20:56
	→ agib joined	20:56
ukleinek	anttih: np	20:57
	← agib left	20:57
	→ agib joined	20:57
	→ vintrepid joined	20:58
	← dramsay left	20:59
tiglionabbit_	okay, sorry, my connection is all flakey	21:00
	→ LiamH joined	21:00
tiglionabbit_	so, my question was, when I say git status I see a bunch of things as deleted under "Changed but not updated". How can I get these deletions committed?	21:00
Millow	If I clone my repo and then try to change something and commit , I get this non-ff error	21:01
	nothing seems to help...	21:01
	hmmz	21:01
	← spearce left	21:01
tiglionabbit_	non-ff?	21:01
	← tiglionabbit left	21:02
Millow	! [rejected] master -> master (non-fast forward)	21:02
	error: failed to push some refs to	21:02
	← rlorandi left	21:02
schacon	Millow: what does 'git branch -a' give you?	21:03
Millow	schacon:	21:03
	* master	21:03
	origin/HEAD	21:03
	origin/master	21:03
tiglionabbit_	what about me?	21:03
doener_	tiglionabbit_: "git add -u" didn't work for you? Which git version is that?	21:03
schacon	tiglionabbit_: 'git add -u'	21:03
tiglionabbit_	1.5.0.5	21:04
	git add -u just gives me a usage exit	21:04
Millow	schacon: any ideas?	21:04
DrNick	tiglionabbit_: git commit -a or git rm --cached [bunch of stuff] && git commit	21:04
schacon	Millow: 'git remote show origin'?	21:04
tiglionabbit_	but I don't want to commit all these untracked files	21:04
	→ Pupeno joined	21:04
doener_	hm, that's at least an ancient version that I have not yet seen being used	21:04
tiglionabbit_	I just want to git rid of the deleted ones, and there's a lot of them	21:04
DrNick	tiglionabbit_: then do the second half of the or	21:05
Millow	URL: git_server:foo.git	21:05
	Remote branch merged with 'git pull' while on branch master	21:05
schacon	tiglionabbit_: commit -a won't add untracked files	21:05
Millow	master	21:05
	Tracked remote branch	21:05
	master	21:05
tiglionabbit_	schacon: yeah it will =[	21:05
	← advany left	21:05
	→ deskin joined	21:05
	← jerbear left	21:05
	→ andreaja joined	21:06
tiglionabbit_	huh, so I have to copy-paste all the deleted filenames into a rm command?	21:06
DrNick	tiglionabbit_: or just git commit -a, now that I've reread the manual	21:06
	k-s → k-s[AWAY]	21:06
schacon	tiglionabbit_: commit -a does not add untracked files	21:06
Millow	somehting is worng, but what... :[	21:07
schacon	Millow: one sec	21:07
Millow	ye	21:07
DrNick	Millow: non-fast forward means that the remote branch has changed and you need to pull it before you can push	21:07
tiglionabbit_	oh good	21:07
Millow	DrNick: yes I know, except that it wont work...	21:07
	→ dkowis joined	21:07
DrNick	Millow: have you pulled?	21:07
Millow	yes	21:07
	already up to date	21:07
dkowis	hey guys, I've got a small issue with git	21:07
	it's making too much noise (namely it's not changing a thing) when passing -q to git pull	21:08
	I have far too noisy cron scripts :(	21:08
	mw → mw\|food	21:08
dkowis	supposedly, according to the manpage, -q is supposed to "pass --quiet to git-fetch-pack and silence any other ineternally used programs"	21:09
	well it doesn't	21:09
	git pull spews forth the same amount of noise as it does without quietness	21:09
	→ david_koontz joined	21:09
	← Dvyjones left	21:10
Millow	Could it be that I moved from github to my own server and some things was left on github	21:10
	← david_koontz left	21:10
	→ david_koontz joined	21:10
schacon	Millow: can you run this 'git rev-list master \| grep `git rev-parse origin/master`'	21:10
	is there any output?	21:10
Millow	yes	21:11
	an id	21:11
	of some sort	21:11
schacon	hm, that is curious	21:11
	← venger left	21:11
	→ fanthomas joined	21:12
schacon	can you run 'git fetch origin', then run that command again?	21:12
	← Eludias left	21:12
Millow	same id	21:12
	← fanthomas left	21:12
schacon	that should be a fast-forward then.... weird	21:13
doener_	Millow: did the "fetch" complain about something being rejected?	21:13
Millow	doener_: nopes	21:13
	but I dont know it the master is on the same server or something	21:14
	because I had the project on github.com	21:14
	then I moved it to my own server, if something went wrong there or somehting	21:14
	except that I get the same error for the gitosis-admin project that's only been on my own git server	21:14
	→ joydivider_ joined	21:15
doener_	Millow: "git remote show origin", does the URL there look right?	21:15
schacon	hm, maybe it can't write on the server or something?	21:16
Millow	the url looks right for origin to my server	21:16
doener_	Millow: and you just do "git push" or "git push origin", right?	21:16
Millow	both	21:17
	same error	21:17
	schacon: it has permissions to write to the server	21:17
	it worked first day, but not after that	21:17
doener_	ok, try this please: git fetch origin master; git branch tmp FETCH_HEAD; git push . master:tmp	21:17
Millow	* branch master -> FETCH_HEAD	21:18
	Total 0 (delta 0), reused 0 (delta 0)	21:18
	To .	21:18
	c728069..5bb4ac5 master -> tmp	21:18
drizzd_	git push origin master:refs/heads/master	21:18
	→ capisce_ joined	21:18
	→ niki joined	21:18
Millow	hmm	21:19
	Counting objects: 10, done.	21:19
	Compressing objects: 100% (6/6), done.	21:19
	Writing objects: 100% (6/6), 2.20 KiB, done.	21:19
	Total 6 (delta 4), reused 0 (delta 0)	21:19
	*** Project description file hasn't been set	21:19
	error: hooks/update exited with error code 1	21:19
	error: hook declined to update refs/heads/master	21:19
	To git_server:foo.git	21:19
	! [remote rejected] master -> master (hook declined)	21:19
Tv	well that's pretty obviously the hook	21:19
	← warthog9 left	21:20
Millow	what hook?	21:20
	can I change the hook?	21:20
	← brendon_work left	21:21
Tv	you just enable hooks blindly?	21:21
	oh well	21:21
	chmod a-x it	21:22
schacon	you can unset the execute flag on the [git_dir]/hooks/update file	21:22
	← vbabiy left	21:22
Millow	hmm	21:22
	→ jnareb joined	21:22
	← pjhyett left	21:23
Millow	schacon: update.sample?	21:23
jnareb	I wonder if there would be video and/or slides from GitTogether 2008...	21:23
deskin	jnareb: I hope so; CA is a long way from home for me	21:24
jnareb	deskin: if you are in US, even more longer for me (EU)	21:25
deskin	I am	21:25
	← capisce left	21:25
drizzd_	Millow: if it's only the .sample file, it's disabled	21:26
	Tv bets Millow is looking at his local repo	21:27
Millow	can I start over from my github project or something	21:27
	← glommer left	21:27
schacon	Millow: the [git_dir]/hooks/update file where [git_dir] is the repo on your server, not your clone	21:28
Millow	Maybe I should export my repo form github and start over	21:28
	← CelticSoul left	21:28
schacon	why move from github?	21:29
	→ iabervon joined	21:29
Millow	i'm useing remine	21:29
	redmine	21:29
	and it doesnt support github so I have to use my own server for it	21:30
	← joydivider left	21:33
	stouset → stouset\|away	21:33
schacon	jnareb: deskin: i think they're planning on taping at least some of it	21:34
	→ peritus joined	21:36
jnareb	schacon: that is what I understood too (that some of talks on GitTogether could be recorded as Google Tech Talks)	21:36
dkowis	anyone got any clue as to why git pull isn't obeying it's -q flag?	21:36
	← nacho_ left	21:37
	← bryanl_ left	21:39
	← bryanl left	21:39
	← Weaselweb left	21:39
	← danten left	21:39
	← peritus left	21:40
	← agib left	21:40
	→ johan-s joined	21:41
	← mithro left	21:42
	→ danten joined	21:42
	→ ajonat joined	21:43
	→ mithro joined	21:44
	← LiamH left	21:45
	deskin → deskin_away	21:45
	← ukleinek left	21:45
	← danten left	21:46
Millow	schacon: solved it	21:46
schacon	yay!	21:47
Millow	started over and cloned my project from github	21:47
	removed .git	21:47
	removed the repo from my own server	21:47
	and runned git init on the project and added remote directly to my own server	21:47
	now it seems to work :)	21:47
schacon	yay?	21:48
Millow	WTF!	21:49
	getting the same error again	21:49
	I'm getting this all the time now, even on new repos	21:49
	→ advany joined	21:50
Millow	schacon: any ideas? :)	21:51
schacon	did you fix the 'update' execute thing?	21:53
	← tsb left	21:53
jnareb	update should be executable and without .sample prefix (executable by the other side)	21:53
Millow	how do I check that?	21:54
	→ warthog9 joined	21:54
	← grahal left	21:54
schacon	ssh into your server, cd to the git dir, ls -la hooks/	21:54
	← foca_ left	21:55
	← sgrimm left	21:55
Millow	git dir as the user, repo or the git app?	21:55
ttuttle	Hey.	21:55
	I made a commit, then I realized I forgot a file in it.	21:55
	So I did "git reset", which the man page says will let you redo a commit.	21:56
	Now one of the files I wrote is missing.	21:56
Millow	schacon: I see the update file	21:56
	dkowis → hax_control	21:56
Millow	what should I do with it?	21:56
schacon	chmod a-x update	21:56
ttuttle	In fact, all the files I committed are missing.	21:56
	But the commit is still in "git log". How do I fix this?	21:57
	← krh_ left	21:57
	deskin_away → deskin	21:57
	→ tiglionabbit joined	21:58
deskin	ttuttle: what was the exact command line you used with 'git reset'?	21:58
ttuttle	deskin: "git reset"	21:58
Millow	schacon: That Helped!	21:58
ttuttle	deskin: no arguments	21:59
deskin	what does git-log --stat HEAD show? do you see the files you added (except for the one you forgot)?	21:59
	→ thelaptop joined	21:59
	← thelaptop left	21:59
ttuttle	deskin: They're still there.	21:59
	deskin: I want to get the files back, then remove the commit without touching my working directory.	22:00
	← EmilMedve left	22:00
ttuttle	deskin: So I think they're stored somewhere, just not here.	22:00
	← nutrimatt left	22:01
iabervon	I think "git checkout HEAD -- ." should get them back from the commit, and then you should use "git commit --amend" to correct the commit.	22:01
	jack_afk → jackdempsey	22:01
ttuttle	iabervon: okay. Can you explain what that command (git checkout HEAD -- ." is doing?	22:01
deskin	right; I think git-reset --hard HEAD; git-reset HEAD~1 will do what you want; first, re-checkout the files, then throw away the commit but leave the files in the working tree	22:01
Millow	schacon: thank you very much for the help	22:02
schacon	Millow: no problem	22:02
ttuttle	Okay, git checkout HEAD -- . worked.	22:02
	But why?	22:02
	→ doener joined	22:02
iabervon	ttuttle: "git checkout <path>" is like "cvs checkout"; it checks out the content of those paths from the repository into your working tree.	22:03
ttuttle	Okay.	22:03
	But I thought HEAD was the last commit you made?	22:03
iabervon	Approximately; it's the commit the current branch is at.	22:04
ttuttle	Okay.	22:04
	So how did it get my files back when I had never committed them?	22:04
iabervon	You said they were there in the log output for HEAD, so...	22:04
ttuttle	Ah.	22:04
iabervon	You did commit them, and your reset hadn't really uncommitted them.	22:05
ttuttle	But wouldn't git reset have changed the current branch to the commit before?	22:05
	Oh, the reset just brought the working directory to that state, not the index?	22:05
iabervon	You need to use --hard to modify the branch (which, fortunately, you didn't do).	22:05
ttuttle	Ah.	22:06
iabervon	Or something. I'm not sure what actually happened there.	22:06
ttuttle	Wow, git is confusing.	22:07
iabervon	I don't normally use "reset"	22:07
warthog9	just a word of warning / caution / heads up	22:07
ttuttle	warthog9: Oh?	22:08
iabervon	I think "reset" is confusing, since one command can manipulate three different things depending on an option that doesn't explicitly mention any of them.	22:08
warthog9	git.kernel.org is now properly doing geodns lookups	22:08
ttuttle	Okay, /me is going to do his first "git push".	22:08
warthog9	so in theory you will get a more local machine when you do a pull, or look at gitweb	22:08
ttuttle	Oh, damn, I forgot to set my name/email :(	22:08
	How do I undo this right?	22:08
	I'm back where I started -- I need to undo the last commit, but I haven't screwed it up yet.	22:10
	→ mfwitten joined	22:10
mfwitten	So I've 'cloned' (partially) an svn repo with git-svn and I'm making commits like crazy. I plan on making a bunch of patches to send to the main developer of this project, so that he can apply them to the svn repository. Is there anything special I need to do (I imagine he'll use the unix `patch' program)? What if I want to change the name of a directory? Can the regular unix patch program handle that? Thanks for the help!	22:10
iabervon	The easiest thing is actually that "git commit --amend" will let you fix the last commit without undoing it per se.	22:10
	→ polpak joined	22:11
ttuttle	iabervon: Ah.	22:11
	iabervon: I'd still like to know how to undo though.	22:11
	iabervon: Besides, I need to change my name/email on the commit.	22:12
	← neoeinstein left	22:12
ttuttle	git reset --soft HEAD^1 seems to do it.	22:12
	← tiglionabbit_ left	22:12
iabervon	I think the replacement that --amend generates will have the new committer. Not sure about author.	22:12
	Yeah, that looks right.	22:12
	→ CelticSoul joined	22:13
	→ entie joined	22:13
deskin	you can alter author with 'git-commit --amend --author="A U Thor [email@hidden.address]	22:14
	← doener_ left	22:14
ttuttle	Too late, I reset.	22:14
iabervon	I generally change history by making a new branch and adding the old branch's changes, rather than resetting. resetting only lets you get a single set of to-be-committed changes, which is a bit limiting.	22:15
mfwitten	ttuttle: It's easy to get it back.	22:16
	← ThunderChicken left	22:16
mfwitten	ttuttle: If I recall	22:16
	mw\|food → mw	22:16
ttuttle	mfwitten: It's okay.	22:16
mfwitten	ok	22:16
iabervon	Oh, and if you uncommit with reset, you have to write the commit message again.	22:16
	ttuttle push.	22:17
	← vintrepid left	22:18
	→ cryolab joined	22:19
ttuttle	Okay, one of my collaborators just got a message "error: Entry 'path/to/file' not uptodate. Cannot merge."	22:19
	Can I force it to overwrite the changes?	22:19
	← boto left	22:19
	← Pupeno left	22:19
	← hax_control left	22:20
ttuttle	Anyone?	22:20
iabervon	That probably means your collaborator has uncommitted changes to that file.	22:21
deskin	tuttle: pushes can't change the working directory on the remote; your collaborator has to merge into their own branch (or rebase)	22:21
	← aroben left	22:22
ttuttle	iabervon: I want to ignore them. (It's sort of a temp file.)	22:22
	Hold on, if I add a file to .gitignore, how do I remove it? Just git rm?	22:22
	← mfwitten left	22:24
iabervon	git rm will make git stop tracking it. Adding it to .gitignore will, if it's not currently tracked, skip it as a file that could be worth tracking.	22:24
Millow	how can I setup so I can get my repos trough http://git.domain.com/repo.git?	22:25
	shat do I need?	22:25
	← entie left	22:25
iabervon	Millow: your web server has to serve repo.git at that URL.	22:26
doener	and you need to run "git update-server-info"	22:27
	← cryolab left	22:27
Millow	iabervon: yes, but are there any speciell tools for it?	22:27
doener	once initially, and then after each update of the repo	22:27
	Millow: hm? Special tools? You mean, like apache?	22:28
Millow	nope...	22:28
	I want to use that url to commit and clone and so on	22:28
iabervon	The HTTP method is designed to work without any server-side special tools.	22:28
Tv	Millow: you're in for a world of hurt, and the name of the pain is webdav	22:28
doener	to fetch (which includes cloning), just serve the directory and run "git update-server-info"	22:28
	and you cannot "commit" to a remote, only push	22:29
Millow	yes push	22:29
	clone, fetch and push	22:29
	that's what I need	22:29
doener	well, over http, it's what Tv said	22:29
	→ madewokherd joined	22:29
Millow	:P	22:29
	← jnareb left	22:30
	← ajonat left	22:30
	← advany left	22:32
	→ akahn joined	22:34
anttih	why is `git --git-dir='/path/to/repo/repo.git' rev-parse --git-dir` exiting with an error code for me?	22:35
	127	22:35
	it outputs the git-dir correctly	22:35
akahn	How can I reset to a previous ref in a particular directory, and have the tree update as well? git reset 1stheuth34sth . reports "x file needs update" for each file. how do i update them from the index?	22:37
doener	anttih: exits with 0 for me	22:37
	git version 1.6.0.2.287.g3791f	22:38
	← cryo_ left	22:38
	→ Ademan joined	22:39
	← joydivider_ left	22:39
	→ foca joined	22:40
doener	akahn: git reset --hard	22:40
	→ joydivider joined	22:40
	→ nil joined	22:40
nil	hi!	22:40
	is there a way to change the email used in past commits? the git tree only exists on my own machine, no clone	22:41
Tv	nil: rebase -i	22:41
doener	filter-branch	22:42
	→ lu_zero joined	22:42
nil	hmmm ok I'll search from that	22:42
Tv	yeah depending on whether it's a few commits or all of them	22:42
nil	all of them	22:43
	I did all my commits with a bad .gitconfig	22:43
	whaow, filter-branch looks great!	22:44
	← Millow left	22:45
deskin	nil: not used filter-branch much myself, but something like 'git-filter-branch --commit-filter "GIT_AUTHOR_NAME=foo GIT_AUTHOR_EMAIL=bar git-commit-tree" looks promising	22:46
	maybe even without the environment vars, if you set the config correctly first	22:47
	→ tiglionabbit_ joined	22:47
doener	deskin: no, the variables are required. Otherwise, it would always make all commits look as if they're from you	22:47
deskin	doener: I think that's what he wants though	22:48
	→ ckm joined	22:48
nil	deskin: they are all from me, in fact	22:48
doener	deskin: yeah, I'm just saying that filter-branch cannot just rewrite all commits to the author that is set in the config. In the general case, that's not what you want	22:48
nil	it's a personal development	22:49
doener	and the --env-filter might be better suited	22:49
nil	deskin: why do you add "git-commit-tree"?	22:50
	← mattf left	22:51
deskin	doener: good point about env-filter	22:51
nil	ah, ok, got it	22:53
	deskin → deskin_away	22:53
anttih	dotech: duh, I was running git through `env`	22:54
	oops, doener	22:54
	← kixx left	22:54
anttih	actually that should've worked	22:56
doener	anttih: env git --git-dir=... also returns 0 here	22:56
anttih	doener: yep, I was running it with the web server user and something is wrong there	22:57
	← tiglionabbit left	22:59
	→ qrush joined	23:00
nil	`git-filter-branch --env-filter "GIT_AUTHOR_EMAIL=my_email" branch1 branch2 branch3 && rm -rf .git/refs/original` was perfect. thanks	23:02
	→ spearce joined	23:03
	← not_benh left	23:04
parasti	though usually you never want to run rm -rf .git/refs/original before you've made sure filter-branch didn't mess up	23:04
	← softdrink left	23:09
	→ camwest joined	23:09
	← tiglionabbit_ left	23:10
	← nbrunson left	23:10
	→ priidu joined	23:11
	← morphir left	23:13
	→ eyeris joined	23:15
eyeris	How can I get a copy of a file as it existed after commit X	23:15
	?	23:15
	← camwest left	23:15
	→ tiglionabbit joined	23:16
Eridius	git show X:file	23:16
	or git cat-file blob X:file	23:16
	← drizzd_ left	23:16
eyeris	Okay, thanks. I new there had to be an easier way that feeding git diff into patch, etc :)	23:16
schacon	i asked a bit earlier, but i don't think anyone knew - is there an SVN frontend similar to git-cvsserver around?	23:16
Eridius	not that I know of	23:17
	← carrus85 left	23:17
	→ EmilMedve joined	23:18
	→ camwest joined	23:18
	← thiago_home left	23:18
	← eyeris left	23:20
	← bbuss left	23:22
	← polpak left	23:23
	→ elliottcable joined	23:26
	← mithro left	23:28
	→ nbrosnahan joined	23:32
	← nbrosnahan left	23:32
	← jackdempsey left	23:34
spearce	schacon: there was a gsoc project idea to create a svn frontend, but its harder than it sounds and the project didn't get off the ground	23:34
	→ mithro joined	23:34
spearce	so the answer is no, there isn't such a beast	23:34
	← Ryback_ left	23:35
	← juanjoc left	23:37
	← koke_ left	23:38
	→ flaguy joined	23:38
schacon	hm	23:40
	<-- want	23:40
spearce	why? i don't think you've thought through what you lose by trying to represent git with the limitations of svn	23:42
	← radarek left	23:42
	← joydivider left	23:42
schacon	no, I hate svn, but there are some people in love with tortoiseSVN - it would be nice to let them use that for simple commit/checkout ops	23:42
	designers and such	23:43
	→ softdrink joined	23:43
schacon	writing that layer is preferable to writing (gasp) windows code to emulate tort (to me)	23:44
	← priidu left	23:45
	→ pygi joined	23:46
	← david_koontz left	23:47
spearce	from what i understand svn has two protocols too; the http-web-dav-lets-support-the-kitchen-sink and its own native thing. which makes implementing it more interesting.	23:47
	→ david_koontz joined	23:48
schacon	i'm thinking the dav would be easier	23:48
	if i'm only supporting a few basic things, i was thinking it wouldn't be horrible	23:48
	then, no COM programming! yay!	23:49
albertito	I took a look a while back and the native protocol doesn't seem too complicated. Parsing it is easy (s-exp) and the set of commands to support basic operations is not too big	23:50
doener	schacon: http://utsl.gen.nz/git/git-svnserver.txt -- might be interesting	23:51
schacon	saw that, but no codez :(	23:51
	→ leachim6 joined	23:51
	→ jmd joined	23:51
jmd	Where can I find the format for the .gitrc file ?	23:52
	← leachim6 left	23:52
Eridius	jmd: gitrc?	23:52
spearce	jmd: you mean .gitconfig ?	23:52
Eridius	maybe you're thinking of ~/.gitconfig	23:52
spearce	man git-config or git config --help	23:52
Gitbot	spearce: the git-config manpage can be found at http://git.or.cz/man/git-config	23:52
Eridius	in which case, it's the same as .git/config, but you should use git-config	23:53
jmd	Thanks.	23:53
Eridius	incidentally, earlier today I finished writing a parser in OCaml that handles git config files	23:53
	Eridius believes he got it to match git-config's parsing behavior exactly, but he's not sure yet	23:53
	← thannoy left	23:57
	← Yuuhi left	23:58

Logs Search ←Prev date Next date→

Channels Documentation