IRCloggy #git 2010-12-28

Provider of IRC logs since 2005.

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

	2010-12-28
d0k	you expect something that's advertised as "the information manager from hell" to have any of those properties? ;)	00:01
	← NewtonianB left	00:02
	← chrischambers left	00:03
^Mike	Yes! :D	00:03
charon	d0k: Hell has had a stable API for about 2000 years and reached feature completeness in 1321. not sure where you could find any fault with that	00:04
d0k	:D	00:05
	→ amerine joined	00:05
	← hyperair left	00:05
	→ jarrod_ joined	00:06
	← gilaniali left	00:06
	→ chrischambers joined	00:09
	← chessguy left	00:14
	→ codykrieger joined	00:15
	← tiagobutzke left	00:20
	← icwiener left	00:21
	→ tiagobutzke joined	00:22
kokx	^Mike: well, in the windows sense of library, it should never be a library in my eyes, separate processes for everything are cool :D	00:28
	← petercoulton left	00:28
cbreak	but slow	00:30
	and error prone	00:31
	(because you have to handle all the marshaling)	00:31
	← JEEB left	00:32
	→ sitaram joined	00:32
	→ JEEB joined	00:32
	← sitaram left	00:32
	→ xiangfu joined	00:33
	→ Slix`` joined	00:33
	→ sitaram joined	00:33
	← d0k left	00:34
	← sitaram left	00:35
	← Slix` left	00:36
	→ sitaram joined	00:37
	← xiangfu left	00:40
	→ Christoph^ joined	00:41
	→ sivy joined	00:42
	← homy left	00:43
	← sou_ left	00:44
	→ Textmode joined	00:50
	← flaguy48 left	00:53
	← zxiest left	00:58
	← sivy left	00:59
	→ QinGW joined	00:59
	← Casmoo left	00:59
	→ hyperair joined	01:14
	← tiagobutzke left	01:14
	→ tg joined	01:15
	← ExtraSpice left	01:15
	← codykrieger left	01:16
	← masonkramer left	01:16
	→ masonkramer joined	01:16
	← amerine left	01:16
	→ tiagobutzke joined	01:18
	→ sou_ joined	01:19
	← bcardarella left	01:23
	→ gilaniali joined	01:24
	← gilaniali left	01:25
	→ RobertLaptop joined	01:26
	→ xiangfu joined	01:27
	← Christoph^ left	01:29
	← justin-george left	01:31
	← xiangfu left	01:31
	→ tgunr joined	01:34
	→ tiagobutzke_ joined	01:34
	← tiagobutzke left	01:35
	tiagobutzke_ → tiagobutzke	01:35
	→ whitedragon joined	01:36
	→ Bombe joined	01:39
	→ robotarmy joined	01:39
	→ Mage_Dude joined	01:48
	→ kanru joined	01:49
	← tgunr left	01:49
	→ halstead joined	01:54
	→ Angel-SL joined	01:55
	← charnel left	01:56
	→ wliao joined	02:01
	→ sivy joined	02:02
	← frakturfreak left	02:04
	→ _sh3 joined	02:04
	← zombor left	02:05
	← hal left	02:14
	→ rocketeerbkw joined	02:15
	→ frakturfreak joined	02:15
	→ blaenk joined	02:16
	← vitor-br left	02:20
	← kanru left	02:25
	← kpreid left	02:25
	→ Vampire0_ joined	02:27
	→ zombor joined	02:28
	← Vampire0 left	02:28
	→ jdav_gone joined	02:30
	→ waterbourne joined	02:31
waterbourne	is there a way to merge a submodule into a parent project such that the commit history of the submodule would be visible intelligibly afterwards?	02:33
	← jdav_gone left	02:35
	→ treak007 joined	02:36
	→ hobodave joined	02:38
	← ereslibre_laptop left	02:41
	← sivy left	02:49
	← chrischambers left	02:51
	→ zxiest joined	02:54
	→ cubuanic joined	02:54
	→ kpreid joined	02:55
rocketeerbkw	is there a way to clone a subdir of a repository? I'm on a slow connection and I just want /root/code instead of cloning all /root	02:56
	← tewecske left	02:58
	→ sonnym joined	03:01
	→ MetaCosm joined	03:03
	→ chessguy joined	03:04
	→ stringoO joined	03:07
	← tiagobutzke left	03:07
	→ pyther joined	03:08
pyther	Hello maybe someone can help me	03:08
	I created a new branch, checked it out and have committed my changes to that branch	03:08
	however when I do a git diff master syslinux it doesn't show the changes with my latest commit	03:08
	it still shows old commit code	03:08
	→ sivy joined	03:08
	← Bombe left	03:09
	→ Bombe joined	03:09
Bombe	pyther, have you tried “git diff master..syslinux”?	03:09
pyther	Bombe: no, but I just tried and it is showing the same old code (from the syslinux branch)	03:10
Bombe	pyther, what branch did you commit to?	03:11
	← brosner left	03:11
pyther	I believe the syslinux branch	03:12
	← Amorphous left	03:12
pyther	Bombe: is there a command that'll show me the last commit, so I can double check and make sure?	03:12
Bombe	pyther, does “git checkout” show syslinux as the current branch?	03:12
	uhm, wait.	03:12
	“git branch” of course.	03:12
pyther	Bombe: yes, it has an astrik next to it	03:13
Bombe	Okay, “git show -p” shows the current commit.	03:13
	That should be one of your commits.	03:13
blaenk	hey guys, something that has always confused me: I clone a repository but I wan't to 'run' a branch other than master, how do I do this? say I want to have branch 'apple'	03:13
	something about remote branches or something, I forget	03:13
	→ simplechat joined	03:14
blaenk	assuming branch apple was pushed to teh repository I cloned	03:14
pyther	Bombe: git show, shows some of the new stuff I've added	03:14
Bombe	pyther, okay, then “git diff master” should show the changes you made after you branched from master.	03:15
pyther	errr... I lied git show, shows only my first commit I made today	03:15
blaenk	anyone? :)	03:16
Bombe	pyther, yes, git show only shows the last commit.	03:16
	pyther, try “git show -p HEAD~1” for the one before that.	03:16
blaenk	git branch -la shows: remotes/origin/apple	03:16
Bombe	blaenk, “git checkout -b apple origin/apple”	03:16
blaenk	thanks Bombe	03:16
	then if I make a change on my dev machine to that branch and push it, how would I update? just git pull while being on that branch?	03:17
pyther	Bombe: I guess I somehow never committed my latest change o_o	03:17
	I could swear I did	03:17
	pyther shrugs	03:17
Bombe	pyther, try gitk. Maybe you committed them to some other branch by accident?	03:17
	→ rctay joined	03:23
pyther	Bombe: nope, must just be really "out there" too night	03:23
	← _Vi left	03:23
	→ NewtonianB joined	03:26
	→ accel joined	03:26
	→ Amorphous joined	03:26
accel	without doing a git -rm, how do I tell git to stop tracking a file?	03:26
NewtonianB	if i have C:/dir/gitdir and gitdir contains my .git folder and all my source, can i cut everything and paste it in C:/dir without screwing up everything?	03:26
	← saysjonathan left	03:26
	→ ki___ joined	03:27
accel	is there a way to say "git status" relative to the root, rather than "git status" relative to current directory?	03:27
ki___	Hello. I have two work stations I use. One is mobile. What is the appropriate method to synchronize local branches I may be working on?	03:27
	Do I need the git server portion installed on my work stations or can i synchronize my work some other way?	03:28
	→ genjix joined	03:28
ki___	I have tried copying the files including the .git folder but I had issues getting the files moved set up on the web server ... it was missing some things	03:28
	→ jdav_gone joined	03:30
	← sivy left	03:33
	← Elfe_ left	03:33
	→ ka1ser joined	03:34
	← accel left	03:38
	← stringoO left	03:40
	→ ketas-av joined	03:41
	← treak007 left	03:41
	→ Elfe joined	03:47
rlb	"git checkout FOO -- bar" returns with no message and no error if bar doesn't exist in FOO. Is that intentional?	03:58
SethRobertson	ki___: The easiest way is to set up a remote from each side and "pull" from the new remote when trying to incorporate the changes.	04:02
ki___	SethRobertson, on the remote git server ...	04:03
	← SkydiveMike left	04:06
SethRobertson	While you can use a remote git server for the communication, if you use the public server then everyone else can see the work in progress. Only you can say whether or not this is a good idea.	04:09
	ki___ what I was suggesting was a way for the workstation and the mobile to directly communicate to avoid the public third party	04:10
	→ schambers joined	04:11
ki___	SethRobertson, I am new to git. Should I search for "setup remote git branches"?	04:11
	SethRobertson, or could u kindly refer me to a specific doc if you know what I can read to accomplish this	04:12
SethRobertson	ki___ man git-remote. Specifically the "git remote add" subcommand	04:12
jast	the 'git-remote' manpage can be found at http://git.or.cz/man/git-remote [automatic message]	04:12
rocketeerbkw	is there a way to clone a subdir of a repository? I'm on a slow connection and I just want /root/code instead of cloning all /root	04:15
SethRobertson	Then once you have the remote set up, when you are done one a particular system you commit your work. When you start up on the other system, you `git pull laptop` or whatever to get the latest changes you committed on the other system	04:15
rlb	I guess I would have expected the git checkout to return with something like "bar does not exist in FOO"...	04:16
SethRobertson	rocketeerbkw: You can do shallow clones and sparse checkouts, but you always have the entire contents of the latest revision in the repo at a minimum	04:16
rocketeerbkw	SethRobertson: thx. doing just a fetch instead of fetch+merge won't save me any bandwidth right?	04:20
jast	rocketeerbkw: no.	04:21
SethRobertson	rocketeerbkw: fetch uses the network, merge does not. pull is fetch+merge (or fetch+rebase)	04:22
jast	rlb: I do get an error message	04:22
	error: pathspec 'foobar.h' did not match any file(s) known to git.	04:22
	→ zorzar_ joined	04:23
	← zorzar left	04:26
rlb	jast: what git version?	04:33
jast	rlb: 1.7.3.2	04:34
	← NewtonianB left	04:34
	→ rchavik joined	04:35
	← rsewill left	04:37
	← genjix left	04:39
Fissure	1.7.0.4 has it too... you must be on something pretty old	04:41
jast	rlb: fwiw, that feature was added in git 1.4.4.1 in november 2006 :)	04:43
Fissure	O_o	04:44
	→ entrosca joined	04:44
Fissure	rlb: how's your 2.6.0 kernel holding up? :P	04:45
jast	(I'd like to note that it's pretty amazing that git makes it easy to find out stuff like that)	04:45
Fissure	jast: what did you use to search for it?	04:45
	→ Figaroo joined	04:46
Figaroo	I need help getting started with git. Anyone have any good beginning tutorials?	04:46
jast	Fissure: I went through several steps: git log -S"known to git"; git log -S"match any" -M -- builtin/ls-files.c; git log -S"error-unmatch" -- git-checkout.sh	04:47
	admittedly, I used some prior knowledge, too	04:47
Fissure	Figaroo: there are many; git comes with man gittutorial	04:48
jast	Figaroo: the 'gittutorial' manpage can be found at http://git.or.cz/man/gittutorial [automatic message]	04:48
Fissure	there's also pro git and the git community book if you're looking for something more hefty	04:49
jast	so here's a complete book for you: http://progit.org/book/	04:49
Fissure	and "git for computer scientists"/"git from the bottom up" for a more theoretical introduction	04:49
	though those two are probably a bit dated at this point (but it's moot because the core data structures have stayed the same except for submodules almost since the beginning)	04:50
	the problem may be that git has /too much/ documentation :)	04:51
Figaroo	I'm just not exactly sure on what it IS	04:51
	→ justin-george joined	04:51
	← orafu left	04:52
Fissure	Figaroo: it's magic	04:52
	http://en.wikipedia.org/wiki/Revision_control	04:52
Figaroo	oh god	04:52
jast	more triggers: !cs, !bottomup, !book	04:52
Fissure	"oh god"?	04:53
	Figaroo: what are you trying to do/how did you get here?	04:53
	← pyther left	04:53
	→ orafu joined	04:54
Figaroo	haha, I want to host my code for free and I heard github was better than google code	04:54
jast	so you know about subversion?	04:56
Fissure	i'd say that is quite true	04:56
Figaroo	hardly	04:57
jast	uh, what were you doing at google code if you don't know much about subversion in the first place? ;)	04:58
Fissure	Figaroo: good; you don't have to forget about all the bad things it teaches you	04:58
	→ aah joined	04:59
Figaroo	heh, is git a filesystem, or is that too far of a definition for it?	05:03
Fissure	git is a version control system built on top of a versioned, content-addressable filesystem	05:04
Figaroo	is git a program?	05:04
sitaram	charon: ping...	05:04
Fissure	yes	05:04
jast	I think we can safely say that it is :)	05:04
Figaroo	like it's a program that downloads the files through the git protocol	05:05
jast	but perhaps it has become conscious while we weren't looking	05:05
Figaroo	jast, lol	05:05
Fissure	jast: naw, we gotta wait until aug 29 before that can happen	05:05
jast	oh?	05:05
sitaram	oh? from me too	05:05
Figaroo	so where does it download the files to on my computer?	05:06
	Fissure is proud to say he shares a birthday with michael jackson and skynet	05:06
sitaram	what a combo!	05:06
Fissure	Figaroo: whereever you tell it	05:06
jast	Figaroo: up to you. I suggest you just play with it a little. perhaps look at the book I mentioned.	05:06
	if you are actually interested in doing version control, that is	05:07
	otherwise I don't quite know what exactly you are here for in the first place. :)	05:07
Figaroo	I want to know how it works so I can use it on github	05:07
jast	for what?	05:07
Figaroo	my javascript libraries and stuff	05:08
	← halstead left	05:08
jast	so, you are interested in version control? or just in dumping your files somewhere?	05:08
Fissure	imo, if the thing you're coding is anything other than a shell/perl script that's < 100 lines, you want to be using version control	05:09
sitaram	< 10	05:10
	if it's perl, < 5	05:10
Fissure	i was being generous :)	05:11
jast	depends on your dialect of perl	05:11
	← t0rc left	05:12
jast	I prefer the non-write-only dialect	05:12
Figaroo	I want to dump my files somewhere and would like to update newer versions of my libraries, maybe take feedback somehow from users	05:12
jast	ah well	05:13
	← zombor left	05:13
	→ napster_123 joined	05:13
jast	start with the tutorial, I guess (http://git.or.cz/man/gittutorial)	05:14
Fissure	sitaram: i'd say the first 5 lines of any perl script should consist of shebang, comments, and use {strict,warnings}	05:14
sitaram	jast: I was half joking -- perl and shell are the only things I use anyway :-) I prefer to write readable perl also, but for some people "readable" means "dont use perl idioms at all" :-)	05:15
jast	github has a bunch of guides, too	05:15
sitaram	Fissure: ^	05:15
jast	sitaram: ah well, in that case they might as well use BASIC	05:15
sitaram	COBOL	05:15
	<evil grin>	05:15
jast	Fissure: use common::sense;	05:15
sitaram	jast: I've had people complain that I'm using nested hashes in gitolite, saying that makes it unmaintainable. I basically ignored them!	05:16
jast	nested hashes? as in $foo->{abc}{def}?	05:16
sitaram	yup	05:16
jast	uh, so what?	05:16
	it's not rocket science	05:16
sitaram	jast: I don't know; I never bothered to respond and find out what their beef was... I think they tried to make some change, failed to grok the data structure I had built, and were venting	05:16
Fissure	i can't see myself making a multi-file perl program without moose	05:17
jast	I can. especially if it uses a framework that doesn't like moose.	05:18
	such as, perhaps, POE (I haven't actually tested whether they are incompatible)	05:19
sitaram	or if it has to run on any perl anywhere, like my aim for gitolite is/was/will be	05:19
	(with no root access or any additional requirements than what git itself needs)	05:19
jast	anyway, I'm gonna hope for perl6 coming out soon so that I don't have to start seriously using moose at all	05:19
	(yeah, right)	05:19
	and then there's my secret project for getting rid of everything that annoys me about programming	05:20
	(also, not very active project)	05:21
Fissure	jast: remind me to ask the bookie what the current odds are on perl 6 vs hurd vs duke nukem forever coming out first	05:22
jast	duke nukem has a release date	05:22
	← kukks left	05:22
	← aah left	05:23
	→ aah joined	05:23
sitaram	jast: that thing we were discussing yesterday, about GIT_PS1_SHOWDIRTYSTATE=ondemand? Won't work -- because it gets called via $(...), it can't maintain state. (I call it via PROMPT_COMMAND)	05:27
	charon: ^ (if you see this)	05:27
jast	I'm not really familiar with the bash completion script	05:29
	not least because I actually use zsh	05:29
	← waterbourne left	05:31
	→ SpookyET joined	05:32
SpookyET	Hello	05:32
sitaram	jast: aah I see; nvm then	05:32
SpookyET	Is it possible to use .gitignore not to ignore files, but to ignore certain lines of a file, for example, authentication in dot files?	05:32
Fissure	SpookyET: no	05:33
	usually you split things out into a template file without credentials which is kept in the repo, and people copy that and fill it in	05:34
SpookyET	Fissure, you mean create a censored branch?	05:35
Fissure	no, you leave the version with the private stuff untracked	05:35
	just keep the file small and updating it when the template changes isn't an issue	05:36
	→ fivetwentysix joined	05:40
	← alezandro left	05:41
	→ j4m3s joined	05:48
rlb	Fissure, jast: thanks; git's 1.7.2.3, kernel's 2.6.32 -- I imagine I was just doing something wrong, but the tree's gone. I'll check more carefully if I notice it again.	05:48
j4m3s	i'm trying to do a git clone from a git repo and i'm unable to do it over git:// (http:// works)	05:49
	i've opened tcp 9418 on my cisco access list	05:51
jast	j4m3s: can you paste the error message to a pastebin?	05:52
	← jfkw left	05:52
j4m3s	jast: yes thx http://pastebin.ca/2031160	05:52
	→ slonopotamus joined	05:55
	← Buglouse left	05:56
j4m3s	jast: and it works if i specify http:// albeit slowly	05:56
	→ zombor joined	05:57
jast	well, either something on your network is still filtering the connection, or routing is broken	05:57
	← justin-george left	05:58
	→ justin-george joined	05:58
jast	if you can't figure out the problem, note that starting from git 1.6.6, HTTP is almost as fast as git:// on servers that support it (that server does)	06:00
	→ pheaver joined	06:00
	← ki___ left	06:00
j4m3s	jast: when i use http it creeps along	06:00
	← _sh3 left	06:02
	→ junyongsuh joined	06:03
	← justin-george left	06:03
	← pheaver left	06:04
	← mmc left	06:05
jast	j4m3s: it's normal for it to take long on forks of linux, no matter the protocol	06:06
junyongsuh	i have a problem while connecting to git server	06:06
	it says,	06:06
	Permission denied (publickey).	06:06
	fatal: The remote end hung up unexpectedly	06:06
	it seems like i have to add some public keys or something to connect to git server from my desktop	06:06
j4m3s	jast: thx	06:07
jast	junyongsuh: what kind of server? one you manage yourself, or a git hosting provider (if so, which one)?	06:07
junyongsuh	i'm using git on my local desktop (mac pro) and, i'm going to GitHub for provider	06:08
	actually, i'm following steps from the book Ruby on Rails Tutorial	06:08
jast	junyongsuh: http://help.github.com/linux-key-setup/	06:09
	← cubuanic left	06:09
jast	should be very similar on osx	06:09
	be careful about deleting existing keys, of course	06:10
junyongsuh	i've read this page,	06:10
	but in ~/.ssh,	06:10
	i only hav know_hosts	06:10
	then should I make other files by meself?	06:10
jast	so, look at the section "Generating a key" :)	06:10
junyongsuh	okay, then it means I don't have any keys in advance so I have to generate one, right?	06:11
jast	yes	06:11
junyongsuh	thanks, i'll give it a shot!	06:11
	→ veroveroyoon joined	06:12
	← j4m3s left	06:15
	← chessguy left	06:17
	← drl left	06:18
	← zombor left	06:19
	→ pheaver joined	06:19
	→ albel727 joined	06:21
	← blaenk left	06:22
	← MikhX left	06:24
	← ceej left	06:28
	→ drl joined	06:30
	→ ki__ joined	06:34
Figaroo	I'm reading progit.org's book. My question at this point is how does each version of a file get saved?	06:34
	if I save a file 5 times, does that create 5 different versions of a file?	06:34
ki__	SethRobertson, would I need gitosis to host the "remote repository"?	06:34
Arrowmaster	ki__: gitosis is deprecated	06:34
Fissure	Figaroo: what do you mean by "save"?	06:34
ki__	Arrowmaster, I want to share a LOCAL branch on my desktop with my laptop, then back to my desktop	06:35
Fissure	writing your editor's internal state to disk has nothing to do with git	06:35
	← robotarmy left	06:35
ki__	SethRobertson, suggested adding a remote	06:35
Figaroo	Fissure, when I modify a file on my machine's HD.	06:35
ki__	I am reading the remote-add man page ... it needs a url for the hosted git remote repo	06:35
SethRobertson	ki___: Several hours ago. I wasn't suggesting adding a remote server, though you could. gitosis is overkill for one person	06:35
	Just add remote laptop on your workstation and workstation on your laptop, then you can pull from the other checked out remote repository	06:36
ki__	SethRobertson, I'm not sure how to get the remote added. It says: fatal: Not a git repository(or any of the parent directories): .git	06:36
Figaroo	If I checkout a project on github, then modify one of the files within that project 10 times, does that mean that there are 10 versions of that file on my computer somewhere?	06:36
Fissure	Figaroo: like i said, doing a save in your editor is tangential to any use of git	06:37
	→ EvilSushi joined	06:37
SethRobertson	ki___: Step one: figure out how to ssh to laptop from workstation and from workstation to laptop. If this cannot be done, then setting up a server repo might be the best approach	06:37
Fissure	you need to invoke git explicitly for a new version to be recorded	06:38
	→ mmc joined	06:39
Figaroo	well if I save a modify a file, shouldn't git keep a copy of it to revert back to?	06:39
	isn't that the whole point of git?	06:39
	← napster_123 left	06:39
Fissure	it keeps the last copy you told it to	06:39
	you could conceivably add a hook in your editor to make a commit whenever you save, but that's not a very good idea	06:40
	← sraue left	06:40
dominikh	the idea is to be able to revert to defined states	06:41
Arrowmaster	Fissure: read this http://tom.preston-werner.com/2009/05/19/the-git-parable.html	06:41
Figaroo	So I have to commit changes of a file to the repositoy?	06:41
Fissure	Arrowmaster: tab completion fail?	06:41
Arrowmaster	sorry, Figaroo	06:41
	yes	06:41
Fissure	Figaroo: yes	06:42
Figaroo	And the repository is both on my machine and on a centralized server somewhere?	06:42
Fissure	there are two repositories, and you may send changes between them	06:42
Figaroo	the two repositories are on my computer and somewhere else?	06:43
Arrowmaster	Figaroo: read this for a bit of the how and why git works http://tom.preston-werner.com/2009/05/19/the-git-parable.html	06:43
ki__	SethRobertson, ssh is set up	06:43
	→ blaenk joined	06:43
SethRobertson	ki____: So on laptop run `ssh workstation ls -d /path/to/repo/.git`	06:44
	← blaenk left	06:44
	→ napster_123 joined	06:44
SethRobertson	ki_: Once that works, the remote URL would be: ssh://workstation/path/to/repo (without the .git)	06:45
Fissure	Figaroo: yes, one on your computer, one on github's	06:45
SethRobertson	ki_____: You should be able to add that remote and the fetch it.	06:45
Fissure	potentially many more elsewhere, if other people have cloned from github	06:45
Figaroo	So, will people be able to commit to my github repo?	06:46
Arrowmaster	no, not unless you let them	06:46
	which i dont think you can do with a free account	06:46
	← slonopotamus left	06:47
Fissure	i think you could several years ago	06:47
	we did it somehow	06:47
	→ MikhX joined	06:47
	← EvilSushi left	06:47
Fissure	though that may have just been "add everyone's ssh key to the same account"	06:47
	i don't think so, though	06:47
	← ki__ left	06:49
	→ _sh3 joined	06:49
	→ rsewill joined	06:50
	← rsewill left	06:51
	→ rsewill joined	06:52
	→ sraue joined	06:54
	→ Adaptee joined	06:55
	→ ki__ joined	07:00
ki__	SethRobertson, git remote add test ssh://HOSTNAME/var/www/working-copy	07:02
	fatal: Not a git repository(or any of the parent directories): .git	07:02
	ssh HOSTNAME; ## Logs in via ssh	07:02
SethRobertson	ki___: you are on your eg. laptop in your git repo on the laptop?	07:03
ki__	ssh workstation ls -d /path/to/repo/.git returns the path to the .git folder	07:03
	SethRobertson, well, I haven't created the repo yet ...	07:03
	normally i would clone from the server	07:04
	do I need to do that?	07:04
SethRobertson	Yes, step zero: Create the repo. Go ahead and clone.	07:04
	from the server	07:04
ki__	can i just make a generic git repo?	07:04
	to test this ..	07:05
Fissure	yes; clone is basically just init + set some config variables + fetch + checkout	07:06
SethRobertson	ki____: You can mkdir test-dir; cd test-dir; git init; git add remote ....	07:06
	ki___: but if you think that is going to be less expensive than the normal clone, you are wrong	07:07
	→ RoboTeddy joined	07:07
RoboTeddy	I've never done it, but it seems easy to overwrite working copy changes with 'git checkout <file>' -- if it were to happen, would there be a way to restore what you had?	07:08
	(it feels like it should be necessary to apply a force flag or something)	07:08
	← drl left	07:10
ki__	SethRobertson, it is cloning now	07:11
	← eletuchy left	07:11
	→ monokrome joined	07:11
monokrome	Hi. I have a file called bin/fix-permissions - but git wont let me add it using this .gitignore: http://paste.pocoo.org/show/310701/	07:11
	I think that my !bin/fix-permissions isn't working. I've also tried using "bin/" instead of "bin/*", but no success.	07:12
ki__	SethRobertson, okay the repository is cloned. I have CD'd into it	07:13
Fissure	if the number of things you want to keep tracked in bin is low, just add them with -f and ignore the whole directory	07:13
SethRobertson	ki_: Add the remote and fetch it	07:14
ki__	git add remote REMOTE.NAME HOST:/path/to/dir	07:14
	fatal: pathspec 'remote' did not match any files	07:14
	crap git remote add	07:14
	git remote add worked	07:15
	at least I think it did	07:15
	→ Maahes joined	07:16
	← SpookyET left	07:16
Maahes	this the first time this is ever happened for me, but git is refusing to connect / timing out on an address I can ping just fine. How can I go about determining what the issue might be?	07:18
	← rchavik left	07:20
ki__	SethRobertson, I got it to switch to the remote branch!	07:20
SethRobertson	Glad I could help	07:20
ki__	SethRobertson, Now I am updating workstation, ie: committing, and i will use git pull to get the updates?	07:20
SethRobertson	Yes. Without a tracking relationship, you will need to name who you want to pull from. `git pull workstation master` or perhaps just `git pull workstation`	07:22
	→ drl joined	07:23
	← Figaroo left	07:25
	→ dc5ala joined	07:25
Maahes	prefix the repo with git:// and it times out, prefix it with http:// and it just complains about missing info/refs	07:26
	nevermind, found a working repository	07:27
	← Maahes left	07:35
	← drl left	07:38
	→ _Vi joined	07:46
	→ guardian joined	07:50
	→ drl joined	07:51
	→ khmarbaise joined	08:01
	← drl left	08:01
napster_123	is there any way to see only the newly added files in previous 10 commits ?	08:02
	i did git diff HEAD~10 HEAD --name-only but it shows all affected files .... not just "newly added files"	08:02
jast	napster_123: --diff-filter=A	08:08
napster_123	jast: ok trying that	08:08
jast	do or do not, there is no try... especially since it works ;P	08:08
napster_123	trying means experiencing it myself.	08:09
	thanks for the switch	08:09
jast	you're welcome, and I'm not really seriously into word acrobatics anyway :)	08:10
	→ psankar joined	08:17
	← MikhX left	08:21
	→ xtuser joined	08:23
	→ MikhX joined	08:25
	→ eletuchy joined	08:36
	→ blaenk joined	08:36
	→ amerine joined	08:37
	→ psoo joined	08:38
	← guardian left	08:40
	← blaenk left	08:50
	← amerine left	08:51
	→ ExtraSpice joined	09:02
	← Slix`` left	09:02
	← junyongsuh left	09:02
	→ Neo\|Desktop joined	09:05
	→ dr_win joined	09:07
xtuser	Hi	09:09
	← ExtraSpice left	09:09
xtuser	I'm trying to find a graphical git client that supports http, but it's not easy	09:09
	This is for windows workstations	09:10
Arrowmaster	git-gui?	09:10
xtuser	Does anybody know of any?	09:10
	Arrowmaster, I've just taken a look, and for a massive deployment, it looks a big deal	09:11
	Isn't there a standalone like tortoise (tortoisegit doesn't seem to support http)?	09:12
	Is it so rare to support https?	09:12
	→ ExtraSpice joined	09:13
xtuser	I support the idea of security through ssh, but not for windozers ;)	09:13
	← hyperair left	09:13
Arrowmaster	have you tried http in tortoisegit?	09:13
	→ hyperair joined	09:14
xtuser	Arrowmaster, yes, but I've been unable to get it work	09:14
	Take that I'm starting at this, and I've just created an empty git repo, but I could have failed	09:14
	I've just done "git init", and the tree has been created	09:15
	I haven't seen anything else to be done	09:15
	Am I missing anything?	09:15
	Of course, I've put that directory under apache and dav according to the info I've found	09:15
	← albel727 left	09:16
Arrowmaster	might need to make a commit first	09:16
	← psankar left	09:18
	→ guardian joined	09:19
	← monokrome left	09:19
	← napster_123 left	09:23
	→ nevyn^_ joined	09:28
	← RoboTeddy left	09:33
	→ mmw joined	09:37
	← dr_win left	09:38
	← _Vi left	09:39
f00li5h	hrm	09:41
	how can I commit and then push and be told that everything is up to date?	09:41
	→ Zephyrus joined	09:42
doener	f00li5h: the commit didn't affect a branch head that is to be pushed	09:43
f00li5h	how can that be?	09:49
ukleinek	f00li5h: how did you push?	09:50
f00li5h	by typing git push	09:50
jast	"git push" updates all branches that exist on both sides	09:51
	→ Seb joined	09:51
jast	i.e. if you are on a new branch, it does nothing	09:51
f00li5h	there are no branches ...	09:52
	I mgiht be pushing to somewhere silly	09:52
Seb	is there really no way to perform git operations outside of the top-level directory of my project ? For isntance, if ~/foo is that directory (result of a git clone), can't I "git add foo/bar" from ~, insead of having to cd into foo first ?	09:53
_ikke_	f00li5h: You always have one brnach	09:53
	Seb: There are ways	09:53
Seb	_ikke_: dirty ones ? :)	09:53
_ikke_	Seb: Nope	09:53
Seb	_ikke_: k, I'm listening heh	09:54
_ikke_	git --git-dir=~/foo ...	09:54
	r setting the git_dir environment variable	09:54
Seb	I see	09:54
	→ _Vi joined	09:54
Seb	not really convenient, but it does answer my question !	09:55
jast	f00li5h: to push a new branch, git push someremote branchname	09:55
Seb	i was thinking maybe some config option would have git somehow try a couple of things automatically	09:55
	but I can settle for that :)	09:56
jast	nah, git doesn't ever go looking for a repository in subdirectories	09:56
Seb	all right	09:57
	jast, _ikke_: thank you guys	09:57
	→ tewecske joined	10:00
	← Seb left	10:03
	→ tizbac joined	10:04
	→ RoboTeddy joined	10:04
	← mmw left	10:06
	← hyperair left	10:06
	→ psankar joined	10:06
	→ mmw joined	10:07
	← pheaver left	10:07
	→ pheaver joined	10:16
	← aah left	10:17
	← psankar left	10:20
	← _Vi left	10:22
	← mmw left	10:28
	→ mmw joined	10:28
	← fivetwentysix left	10:29
	← QinGW left	10:29
	← mmw left	10:33
	→ mmw joined	10:34
	← veroveroyoon left	10:34
	← pheaver left	10:34
	← tewecske left	10:36
	→ petercoulton joined	10:37
	→ tewecske joined	10:38
	→ refund joined	10:43
refund	how do i get the uri for this working tree/repository?	10:43
	so that i can do a clone somewhere else	10:43
	→ [Po]lentino joined	10:44
codejunky	git remote -v	10:44
refund	thanks	10:46
	→ Figaroo joined	10:49
Figaroo	so why would I get an error saying that I could not lock .git/config: No such file or directory when I try git config user.name "Figaroo"?	10:51
cirwin	you didn't do it from inside a git repository	10:53
	you probably wanted git config --globla ?	10:54
	*global	10:54
Figaroo	cirwin, yes --global works. So repositories are contained in a folder right?	10:55
cirwin	yes	10:55
Figaroo	and .git is the database?	10:55
cirwin	yup	10:56
Figaroo	so all the old file versions are stored in .git?	10:56
cirwin	yes	10:56
Figaroo	everything out side of .git is the latest version of files?	10:57
cirwin	indeed — though there are some "special" files, like .gitignore that are used for git configuration. though it's only the latest version of that file too :)	10:57
Figaroo	where's git installed?	10:58
cirwin	depends on your computer — wherever applications are normally installed	10:58
Figaroo	I'm using cygwin	10:59
	so it's "like" unix	10:59
	or linux	10:59
cirwin	it's probably spread around /usr/bin and /usr/lib, not sure	10:59
	why do you ned to know?	11:00
Figaroo	linux's FSH is very confusing	11:00
cirwin	I agree	11:00
Figaroo	I think macs might have the best FS	11:02
cirwin	why?	11:02
	→ napster_123 joined	11:11
Figaroo	because they're not affraid to use long names; they don't abbreviate everything	11:13
	anyway	11:15
	g2g to bed	11:15
	night	11:15
	← Figaroo left	11:15
	→ ch077179 joined	11:16
	← krosenvold_ left	11:16
	← napster_123 left	11:21
	→ Casmoo joined	11:22
	→ napster_123 joined	11:24
	← mmw left	11:25
	← wliao left	11:30
	← doener left	11:30
	→ doener joined	11:30
	→ seutje joined	11:30
xtuser	How can I clone a https repository, bypassing the certificate check?	11:33
	← nevyn^_ left	11:34
seutje	I'm going crazy, I added "foo" and even "pee/poo/foo" to my .git/info/exclude, but it's still showing pee/poo/foo in git status, not using a .gitignore file	11:36
Bombe	seutje, is it modified?	11:37
seutje	well yea, but I don't want to commit that change	11:38
Bombe	seutje, then don’t commit it but you can not ignore already tracked files.	11:38
	Either you want to track them or don’t.	11:38
seutje	ah boo, so I can't just edit it locally and ignore that change and only commit the rest without explicitly unstaging it every time I "add ." ?	11:39
cirwin	seutje: unfortunately not	11:40
cbreak	gitignore does not ignore changes	11:40
	it ignores files	11:40
Bombe	seutje, IMO it’s bad to use “git add .”. Only add the files you want to commit.	11:40
cbreak	and obviously, you can not track and ignore a file at the same time	11:40
cirwin	that's a bit inconsistent with the rest of git ;)	11:40
Bombe	Too many stuff might creep in using “add .”.	11:40
seutje	Bombe: but then I'd have to type crazy long paths all the time	11:40
Bombe	That’s a subversionism.	11:40
seutje	I always check the diff before commiting	11:40
Bombe	seutje, you might use a GUI. Or copy & paste from git status.	11:40
cbreak	just use git add -p	11:41
Bombe	Or add -p, right.	11:41
cbreak	I don't think I ever used git add . after the initial commit	11:41
Bombe	seutje, or: Create a branch for that change alone, do your development on master, and merge master into your one-change-branch every now and then, compiling/using stuff from that branch to test.	11:41
rudi_s	Is git's completion able to complete modified files? (I know zsh's completion can do it, but I don't know bashs)	11:42
Bombe	cbreak, me neither. :)	11:42
seutje	thanks, now u gave me a headache :P	11:42
Bombe	I’m always glad to help. :)	11:42
	I have such a setup for local configuration stuff that is not going in any other repositories.	11:42
cbreak	anyway, just git rm --cached the file	11:42
seutje	yea, I'll just alias that	11:43
cbreak	it won't be tracked anymore, and it'll be deleted from the repository	11:43
	and it won't show up for you either	11:43
seutje	I don't wanna delete it from the repo	11:43
	I'll get lynched by my coworkers	11:43
cbreak	then don't modify it :)	11:43
Bombe	:)	11:44
	← Adaptee left	11:44
seutje	I have to, it asplodes my php with some murky segmentation fault I don't wanna debug	11:44
cbreak	I bet they are similarly unhappy about a modified version	11:44
seutje	I don't want to modify it in the repo, I just want to use a modified local version, leaving the one in the repo untouched	11:45
cbreak	segmentation faults are problematic.	11:45
seutje	tbh, php is problematic at its core ;)	11:45
Bombe	Oh yaah.	11:45
cbreak	when the stuff I develop crashes, I stop all other development until it's fixed	11:45
seutje	it's just an ini_set, which I've already set to that value in my global php.ini, so I don't even need that it	11:46
	→ kwijibo joined	11:46
seutje	which is why I wanted to remove it and ignore the change, but guess I'll just make an alias to unstage it every time	11:46
cbreak	don't stage it in the first place	11:47
seutje	seems easier to stage all and unstage 1, than to stage all one by one	11:48
	anyway, senks guys/girls/whateveryouliketobeaddressedas <3	11:51
	← Elfe left	11:53
kwijibo	hi does git-svn have to be used with a whole repository ? what I want is to just have a directory in my git repo synched with trunk of an svn repo	11:53
	→ Elfe_ joined	11:53
	→ sveinse joined	11:53
sveinse	Hi. I've run "git clone --mirror" from an external git repo. Can I somhow update it incrementally again with git pull or similar?	11:54
	(In that respect, what's the difference between --bare and --mirror?)	11:55
	← hobodave left	11:59
bremner	kwijibo: should work fine	12:09
	kwijibo: just give the right svn url to git svn init	12:09
	← dc5ala left	12:10
kwijibo	hmm	12:11
	i get error: Can't use --stdin-paths with --no-filters	12:11
	then a usage message	12:11
	then	12:11
	hash-object -w --stdin-paths --no-filters: command returned error: 129	12:11
	and	12:11
	error closing pipe: Bad file descriptor at /usr/local/git/libexec/git-core/git-svn line 0	12:11
	→ goshawk joined	12:11
kwijibo	is my installation b0rked ?	12:11
bremner	what did you type?	12:12
	→ chrischambers joined	12:15
	← goshawk left	12:17
	→ flo` joined	12:17
flo`	hi	12:17
	i'm new to git, and i'd like to put a project under the git system...	12:18
	→ MatheusNg joined	12:18
flo`	can i tell git somehow not to monitor binaries like foo.o or myprog (the ELF-file)?	12:18
	or must i do a make clean manually each time?	12:18
bremner	flo`: man gitignore	12:18
jast	flo`: the 'gitignore' manpage can be found at http://git.or.cz/man/gitignore [automatic message]	12:18
	← MatheusNg left	12:18
	→ albel727 joined	12:19
flo`	thanks	12:20
	← mmc left	12:29
	→ goshawk joined	12:30
	→ rseifert joined	12:32
	→ d0k joined	12:36
	→ SkydiveMike joined	12:37
	← goshawk left	12:37
	→ nevyn^_ joined	12:38
	→ TheFox21 joined	12:40
	→ hyperair joined	12:40
	← entrosca left	12:41
	→ yorickpeterse joined	12:42
	← yorickpeterse left	12:42
	← Mage left	12:43
	← napster_123 left	12:54
	→ napster_123 joined	12:56
	→ mmc joined	12:56
	→ _Vi joined	12:57
	→ entrosca joined	12:59
	← entrosca left	13:01
	→ entrosca joined	13:01
	← entrosca left	13:02
kwijibo	bremen: I typed git-svn clone -s http://moriarty.googlecode.com/svn/trunk moriarty	13:03
	→ entrosca joined	13:03
	→ gebi joined	13:03
	→ TSK joined	13:04
charon	kwijibo: look at this thread: http://thread.gmane.org/gmane.comp.version-control.git/141468/focus=141469	13:06
	kwijibo: it sounds like your git-svn already has patch 2/3, but the git installation used to run git-hash-objects does not have patch 1/3	13:07
	hence this can't work	13:07
kwijibo	ah thanks charon	13:07
bremner	Is there a git way of extracting the changes from a given commit (which won't cherry-pick) to a given file? Or am I stuck editing diffs?	13:12
	I suppose I can checkout that commit, and reset.	13:14
	→ tiagobutzke joined	13:14
	← tiagobutzke left	13:14
	→ rchavik joined	13:16
	← masonkramer left	13:16
charon	bremner: you mean do a "limited cherry-pick"? try 'git show --format=email -- files \| git am'	13:16
	→ masonkramer joined	13:16
bremner	ah, cool	13:16
charon	er, with a commit argument for show, sorry	13:17
bremner	understood :)	13:17
	→ segher joined	13:18
	← napster_123 left	13:20
bremner	ah well, no common history, so I ended up needing to use git diff \| git apply	13:24
	← TSK left	13:27
	← kpreid left	13:29
	← ben_alman left	13:32
	→ NewtonianB joined	13:32
	← _Vi left	13:38
charon	bremner: how does (lack of) common history affect this?	13:39
bremner	well to be precise, the patches don't apply with git-am	13:40
	→ rns joined	13:41
	→ tomaw joined	13:42
bremner	oh, it looks like one of the patches is creating the file. I guess this is an artifact of cloning an svn branch with git-svn; history seems to have been squashed.	13:42
	← kwijibo left	13:45
NewtonianB	hey, i have a git bare repository on my server that i push to, then i run a git archive master \| tar -x C /mywww/folder/ the problem is it just overwrites or creates new things but if i delete files they dont get deleted	13:46
	what can I do without having to delete the entire directory everytime because it contains things i need not deleted	13:47
	→ mstsirkin joined	13:50
wereHamster	NewtonianB: rm -rf /mywww/folder; git clone /path/to/the/bare/repo /mywww/folder; and then to update: cd /mywww/folder && git pull	13:52
	or, you could use a proper deployment tool which does that for you	13:52
	→ joel_falcou joined	13:53
NewtonianB	wereHamster: thanks, whats a proper deployment tool?	13:53
joel_falcou	hey, how can i retrieve the latest commit SHA-1 to include into a .h ?	13:53
wereHamster	capistrano, fabric etc	13:53
	joel_falcou: git describe --always	13:54
	joel_falcou: and use tags, to make the output form git descrbe meaningful	13:54
joel_falcou	hmmm ok	13:54
	git tag i assume	13:56
	thx wereHamster	13:58
	← NewtonianB left	14:03
	→ sivy joined	14:03
	← sivy left	14:03
	← mstsirkin left	14:03
	→ selckin joined	14:08
	← tg left	14:09
	→ gigamonkey joined	14:16
	← normanrichards left	14:22
	→ napster_123 joined	14:23
	← napster_123 left	14:23
	→ rolfb joined	14:24
	← rolfb left	14:24
	→ rolfb joined	14:24
	← rolfb left	14:24
	→ rolfb joined	14:24
	← rsewill left	14:26
	← tizbac left	14:30
	← gebi left	14:32
	← rchavik left	14:34
	→ zombor joined	14:39
	← sraue left	14:41
	→ aah joined	14:43
	← nevyn^_ left	14:50
	→ nevyn^_ joined	14:51
	→ normanrichards joined	14:51
	→ runvnc joined	14:52
	→ t0rc joined	14:52
	→ sraue joined	14:53
	← rolfb left	14:53
runvnc	having trouble transferring a subdirectory. i used git add subdir, git commit, went in and added all the files, the pushed. it keeps saying everything is up to date, but the directory on the other repo is empty after I pull	14:54
	→ _Vi joined	14:54
	→ goshawk joined	14:56
selckin	needs more details	14:56
wereHamster	runvnc: is that subdir a git repo on its own?	14:57
runvnc	werehamster I hope not.. all I did that I remember is say git add mydir	14:57
	and then commit and push	14:57
	maybe it didnt really add it	14:57
wereHamster	ls subdir/.git; does it list any files?	14:58
runvnc	although seems like it	14:58
	one second	14:58
	crap	14:58
	there is a .git in the subdir is that bad	14:58
wereHamster	yes	14:59
	git addid it as a submodule	14:59
runvnc	how did I do that	14:59
wereHamster	git did it	14:59
runvnc	or nevermind how did I do that.. main thing is I need to transfer the files by pulling	14:59
	or I could just zip them	15:00
	how hard is it to fix so it is part of the other repo	15:00
	→ bmomjian joined	15:00
	→ icwiener joined	15:00
wereHamster	not at all	15:00
runvnc	git mergebackintomainmodule mydir	15:01
bmomjian	Is there a way to use an external diff tool for 'git show' like is possible with 'git diff'? Right now, I can only get git show to show unified diffs	15:01
	→ rsewill joined	15:02
	← schambers left	15:04
jaeckel	runvnc: http://stackoverflow.com/questions/1759587/un-submodule-a-git-submodule	15:04
	→ schambers joined	15:04
runvnc	thanks jaeckel i was just reading that	15:04
	I dont even know how I made it a submodule in the first place though	15:04
wereHamster	bmomjian: does git_external_diff not work?	15:05
bmomjian	wereHamster: no, it works for git diff, but not git show, which seems odd	15:05
wereHamster	bmomjian: indeed. Send an email to the mailing list	15:06
bmomjian	ok, which email list?	15:06
wereHamster	the git mailing list	15:06
runvnc	except I have no .gitmodules so maybe its a main module	15:07
	except its not in /home/git/repositories so its not	15:07
	when i go into the subdir it says On branch master. Your branch is ahead of 'origin/master' by 2 commits	15:09
	→ MattDiPasquale joined	15:10
bmomjian	wereHamster: email sent, thanks	15:10
runvnc	oh. lol. its slickgrid	15:10
	i forgot i cloned it from github	15:11
	then i edited the files	15:11
	← entrosca left	15:12
	← goshawk left	15:13
runvnc	I'm sorry, how can I tell git that this directory is not related to the github slickgrid and is just full of normal files, so that I can then add it normally to my own repo?	15:14
charon	bmomjian: note that git-show is essentially git log -p --cc --no-walk, so any proposed solution would have to make sense in the presence of multiple arguments (showing several commits) and options like --walk (which again makes it behave more log-like in that it walks history)	15:15
	er, --do-walk (why on earth is it not named --walk)	15:15
	i suspect your average use of git-difftool is not amenable to walking history	15:16
runvnc	maybe if i just delete the .git	15:16
	← jas4711 left	15:16
	← multigrain left	15:17
	→ vezult joined	15:17
	→ multigrain joined	15:17
	→ kylehayes joined	15:18
vezult	how do I teach git that it is incorrectly detecting a rename?	15:18
charon	vezult: you don't, it's all detected after the fact. what context? if it e.g. shows a rename happened in the status report after 'git commit', don't worry, it does not affect the outcome of git-commit in any way	15:19
uau	vezult: git does not support renames, only some commands have heuristics to guess after the fact	15:19
runvnc	ok so what I did was I cloned slickgrid, now I need that grid which is showing my client's specific database, to work on the production machine. how can I take the modified slickgrid files from my subdirectory and remove them from the association with the github slickgrid so that I can add them into the repo in the parent folder and then deploy them	15:20
	I appreciate your help guys	15:21
	→ FYLGood joined	15:21
	→ int3__ joined	15:25
	← multigrain left	15:25
bmomjian	charon: understand	15:26
	Can I use git diff to just show the lines from a single commit? I couldn't figure out the syntax	15:26
runvnc	nm i am an idiot it is in fact a submodule	15:26
	→ ciupicri joined	15:27
runvnc	even though it was cloned and there is no .gitmodules	15:27
charon	bmomjian: git diff C^ C	15:27
	or various other forms, including the sometimes-accidentally-working C^!	15:27
	← FYLGood left	15:28
bmomjian	charon: yeah, I guess that is the way; kind of odd I have to specify the tag twice, but it works	15:28
	→ ceej joined	15:28
ciupicri	I have branch A and branch B derived from branch A; I've rebased A (squashed some commits) and now I can't rebase B from A	15:28
bmomjian	charon: I just wrote a script for it	15:29
ciupicri	I have a conflict that I've solved with git mergetool -t kdiff3 and when I run git rebase --continue it tells me "No changes - did you forget to use 'git add'?"	15:29
charon	bmomjian: why is that odd? there are two sides to a diff ;)	15:29
bmomjian	odd I am retyping the same long tag twice for such a common operation	15:30
vezult	charon, uau: excellent, thanks!	15:30
	→ rolfb joined	15:31
	← rolfb left	15:31
	← sonnym left	15:32
	→ bcardarella joined	15:32
	← int3__ left	15:32
jason237	bmomjian: I use git show to look at a single commit; git log -p -1 would also work	15:35
	ciupicri: yes, after you resolve the conflict you need to git add the affected files before rebase --continue. or you can just git add .	15:36
ciupicri	jason237, I've added the file, but it hasn't changed	15:36
	→ goshawk joined	15:37
jason237	you mean you did git add after merging?	15:38
ciupicri	jason237, yes	15:38
	git mergetool -t kdiff3 -y ; git add etc/cobbler/settings	15:38
jason237	ah, if there are no changes try rebase --skip	15:39
ciupicri	jason237, I think it worked; thanks!	15:40
	← segher left	15:41
	→ ben_alman joined	15:44
	← ben_alman left	15:44
	→ ben_alman joined	15:47
	seutje → frigginSmoke	15:47
	→ Buglouse joined	15:48
	← Buglouse left	15:50
	← ch077179 left	15:50
runvnc	thanks again	15:52
	← runvnc left	15:52
	→ Buglouse joined	15:52
bmomjian	jason237: git log -p does not use the external diff program, so that does not help me	15:53
ciupicri	how can I find out what (remote) branch a local branch is tracking?	15:53
jason237	bmomjian: it should if you use --ext-diff	15:54
bmomjian	oh, wow it did, thanks	15:55
	ah, this works too: git show --ext-diff	15:56
	nice. I suppose there is no way to default --ext-diff for those commands like there is for git diff	15:57
	→ fowlduck joined	15:58
	→ mstsirkin joined	15:59
	→ primalmotion joined	16:02
fr0sty	bmomjian: you can create an alias...	16:03
	frigginSmoke → seutje	16:03
fr0sty	[alias] \n eshow = show --ext-diff	16:04
	← flo` left	16:04
fr0sty	or something like that...	16:04
bmomjian	interesting; I was not aware of that ability	16:04
	← jaeckel left	16:04
	→ jaeckel joined	16:04
fr0sty	man gitconfig and look for alias.*	16:04
	← sou_ left	16:05
	→ sonnym joined	16:06
	→ ziro` joined	16:06
ziro`	is it possible to work on a submodule and commit to it?	16:06
fr0sty	ziro`: yes, but.	16:06
	man git-submodule and look and understand what you are doing first.	16:07
jast	the 'git-submodule' manpage can be found at http://git.or.cz/man/git-submodule [automatic message]	16:07
fr0sty	there are a few gotchas:	16:08
	the submodule is usually on a detached head	16:08
	committing in the submodule does not automatically update the superproject	16:08
	committing the superproject doesn't update the subproject	16:08
	pushing the superporject without first pushing the subproject will break other people's repos	16:09
	might be some more, too...	16:09
	fr0sty cues SethRoberson for his pitch re: gitslave	16:09
fr0sty	ahem, SethRobertson, that is...	16:09
SethRobertson	wha?	16:09
	Oh, yes, submodules suck. gitslave great.	16:10
	gitslave (http://gitslave.sf.net) is useful when you control and develop on the subprojects at more of less the same time as the superproject, and furthermore when you typically want to tag, branch, push, pull, etc all repositories at the same time.	16:10
ziro`	ok, so how does one commit to a submodule?	16:10
SethRobertson	git-submodule is better when you do not control the subprojects or wish to fix the subproject at a specific revision even as the subproject changes	16:10
ziro`	right	16:11
	ok	16:11
	fine	16:11
fr0sty	but, yes you can, and people are doing it every day without issue.	16:11
SethRobertson	To commit to a submodule just check out the branch, commit, and then update the superproject to the new point	16:11
ziro`	right, so i can't actually commit to a submodule, i must clone the repo seperately	16:12
SethRobertson	You already have cloned the repo. that is what the submodule is.;	16:12
fr0sty	eh? you can commit in the submodule. it works.	16:12
ziro`	well git status in the submodule is empty	16:12
fr0sty	this is why you should read the fine manual	16:12
SethRobertson	Or use gitslave	16:13
fr0sty	man git submodule init might be helpful...	16:13
jast	the 'git' manpage can be found at http://git.or.cz/man/git [automatic message]	16:13
ziro`	git submodule status /path/to/submodule is also empty	16:13
fr0sty	man git-submodule	16:13
jast	the 'git-submodule' manpage can be found at http://git.or.cz/man/git-submodule [automatic message]	16:13
ziro`	git-submodule - Initialize, update or inspect submodules	16:13
fr0sty	ack! update, not init...	16:14
	update Update the registered submodules, i.e. clone missing submodules and checkout the commit specified in the index of the containing repository.	16:14
	→ WonTu joined	16:15
	← WonTu left	16:15
ziro`	i'm still not seeing anything that actually refers to COMMITing TO the submodule	16:16
fr0sty	a submodule is just another repository.	16:16
SethRobertson	Read my message above or one of the fine submodule tutorials.	16:17
	→ Thell joined	16:17
	→ crazed joined	16:17
fr0sty	ziro`: look here: http://book.git-scm.com/5_submodules.html	16:18
ziro`	sigh git status is empty, although I know for a fact a file differs	16:21
	thanks for your help guys	16:21
	← Textmode left	16:22
crazed	what are some good options for public git repositories with multiple people committing and a code browser?	16:22
	← mstsirkin left	16:23
	→ alezandro joined	16:24
ciupicri	crazed, github is nice	16:24
	crazed, FOSS projects are hosted for free, others cost money	16:24
crazed	i understand github is nice, and the code should be foss, but still looking for a host it myself type solution	16:25
	what if github is down?	16:25
ciupicri	crazed, so you want to run your own git hosting service?	16:25
crazed	no	16:26
ciupicri	crazed, or you want just a backup solution?	16:26
crazed	more of a solution that i can use at work	16:26
ciupicri	crazed, git is a distributed VCS, so each checkout (clone) is a backup	16:26
	← refund left	16:26
crazed	i'm not really worried about backups so much as just hosting our own git repo	16:27
ciupicri	crazed, if github fails you use git over ssh to fetch or push the code from your colleagues	16:27
crazed	basically i've setup git-daemon and set guid on the repo to enable commits from anyone in the proper group	16:28
ciupicri	crazed, there was also another hosting service which happened to use an open source solution, but I can't remember it right now	16:28
crazed	but i'm looking for the code browsing	16:28
	→ Schmallon joined	16:28
ciupicri	crazed, use Trac or Gitweb	16:29
	→ bentob0x joined	16:29
ciupicri	crazed, http://progit.org/book/ch4-6.html	16:29
crazed	oh wow	16:31
	what the hell doesn't git come with	16:31
	does it send email yet?	16:31
ciupicri	crazed, yes, I think it does	16:32
Bombe	crazed, man git-send-email	16:32
jast	crazed: the 'git-send-email' manpage can be found at http://git.or.cz/man/git-send-email [automatic message]	16:32
crazed	lol oh god	16:33
	that was a joke, but awesome	16:33
	→ corni joined	16:36
ciupicri	crazed, be careful what you wish, it might become true :-)	16:37
crazed	now to get gitweb working with nginx	16:37
	probably have to reverse proxy unfortunately	16:37
	← ciupicri left	16:38
ziro`	how do i make a newline in a git commit messagE?	16:38
	google fails	16:38
Bombe	git commit -m "First line	16:38
	second line."	16:38
ziro`	how do i do that in the cli?	16:39
Bombe	Or omit the message altogether and type it in the editor.	16:39
ziro`	vim is an abortion	16:39
Bombe	ziro`, exactly that way. Type that line, press return, type the second line.	16:39
tango_	WHAT	16:39
crazed	WHAT	16:39
bremner	ziro`: so use some other editor	16:39
Bombe	bash (and every other shell) will handle that correctly.	16:39
	ziro`, set your EDITOR to something else, then.	16:39
tango_	ziro`: you can configure your preferred editor anyway	16:39
Bombe	ziro`, you might also use git-gui.	16:39
bremner	ziro`: anyway, see the -F option in man git-commit	16:39
jast	ziro`: the 'git-commit' manpage can be found at http://git.or.cz/man/git-commit [automatic message]	16:39
tango_	I hear notepad is very strong among programmers	16:40
crazed	export EDITOR=nano	16:40
bremner	"git commit -F -" should work even in braindead shells.	16:40
ziro`	nano is the only editor which actually works as i expect (that i've tried)	16:40
	seutje → frigginCommute	16:40
crazed	:/	16:40
	← corni left	16:41
Bombe	Yes, vi(m) has a learning curve.	16:41
	It’s worth it, though (IMO).	16:41
ziro`	omfg	16:41
crazed	100% worth it	16:41
	← psoo left	16:47
fr0sty	crazed: at least that...	16:48
	← aah left	16:50
	← bmomjian left	16:51
ziro`	sorted it out in the end guys! thanks!	16:54
	someone needs to make a javascript based vi trainer	16:55
	← rseifert left	16:56
Bombe	It’s good to have a colleague who knows vi inside-out. :)	16:57
	I learned most of what I know about vim today from him.	16:57
	→ justin-george joined	16:59
	→ hobodave joined	17:02
	← Zephyrus left	17:03
	← fowlduck left	17:06
	→ pgib joined	17:09
	→ Zephyrus joined	17:10
	→ NewtonianB joined	17:13
	→ segher joined	17:14
	→ amerine joined	17:16
consumerism	ziro`: vimtutor	17:17
ziro`	nice! thanks!	17:17
	→ tiagobutzke joined	17:18
	→ xchg joined	17:18
xchg	Hi. How to push to server if I did "commit --amend" ? It's rejecting me	17:18
	← nevyn^_ left	17:19
Bombe	xchg, you probably shouldn’t.	17:20
	→ ki_ joined	17:20
	← Buglouse left	17:21
Bombe	xchg, please see the discussion in man git-rebase.	17:21
jast	xchg: the 'git-rebase' manpage can be found at http://git.or.cz/man/git-rebase [automatic message]	17:21
xchg	well, I pulled, fixed conflicts, pushed and it works :)	17:21
Bombe	Great. :)	17:22
	→ robotarmy joined	17:26
	→ \|RicharD\| joined	17:29
\|RicharD\|	hi i have a problem	17:29
	when i do git commit -m "message" it say:	17:29
	nothing to commit (working directory clean)	17:29
	why ? and it not add files :(	17:30
uau	xchg: that probably wasn't a good idea	17:30
	if you pull after amend you'll get history with both the amended AND non-amended versions of the commit	17:30
	→ Buglouse joined	17:31
uau	xchg: if you're willing to change history on the server you're probably better off resetting to the first amended version	17:32
	and then pushing that to the server too	17:32
xchg	I just started my project so I can just create new repo	17:32
uau	well doing "git reset --hard <commit>" to reset local branch state to the first amend commit	17:33
	and then "git push <remote> +<branchname>" should be enough to fix things	17:33
	→ AlexP_ joined	17:34
uau	the '+' there means to overwrite existing history	17:34
Bombe	\|RicharD\|, you need to add files to commit first.	17:34
\|RicharD\|	i have do it	17:34
	with git add .	17:34
	and git add *	17:35
uau	\|RicharD\|: well that message says you haven't added any modifications	17:35
	and that the working directory is unmodified too, so there are no changes you _could_ add	17:36
\|RicharD\|	very strange	17:36
	i put manualy this files into directory	17:37
	where i had cloned a git reposity	17:37
	so now i want add this in my reposity	17:37
	i had done git add .	17:37
	and then commit and then push	17:37
	but not add nothing	17:37
	→ steph021_mac joined	17:37
uau	well i can see 3 possible alternatives:	17:37
	→ eno joined	17:38
uau	1) there are actually no such files in the directory, 2) they've already been added to the HEAD commit, or 3) the files match ignore rules	17:38
	← jdav_gone left	17:38
\|RicharD\|	1)no	17:39
	2)no	17:39
	3) how i can see it ?	17:39
uau	ignore rules can be specified in a couple of locations	17:39
	if you just cloned the repo and haven't added anything yourself then they should be in .gitignore only	17:40
\|RicharD\|	how i can see	17:40
	← ben_alman left	17:40
\|RicharD\|	if there are already in HEAD COMMIT ?	17:40
uau	'git log -- <filename>' for example	17:41
\|RicharD\|	i see	17:41
	2 commit	17:41
	so why it not do push in server ?	17:41
jast	did 'git commit' give you an error message? if not, the commit was created successfully	17:41
	if it did, looking at the message will help	17:42
\|RicharD\|	git commit is created well	17:42
	# On branch master nothing to commit (working directory clean)	17:42
	→ ben_alman joined	17:42
uau	how did you try to push the changes to the server?	17:43
	and what check did you do that made you think they had not been successfully pushed?	17:43
	→ jdav_gone joined	17:44
\|RicharD\|	why if i cloned	17:44
	in a different position	17:44
	is it empty	17:44
uau	ok, how did you try to push the changes?	17:46
\|RicharD\|	git push	17:46
uau	you should avoid using 'git push' without arguments	17:47
	try 'git push origin master'	17:47
	does that produce any output?	17:47
	(in that command 'origin' identifies the remote to push to, and 'master' is short for 'master:master', meaning to push the local branch 'master' to remote branch 'master')	17:48
	→ pheaver joined	17:48
\|RicharD\|	Total 925 (delta 327), reused 0 (delta 0)	17:49
	fatal: The remote end hung up unexpectedly	17:49
	fatal: expected ok/error, helper said '2004�oRI=�ٻ4�'�%��M��M>�:N��,�$�' �ѯ	17:49
	�ߴi��~� '	17:49
	fatal: write error: Broken pipe	17:49
uau	well something's broken...	17:50
	← alezandro left	17:50
\|RicharD\|	lol i see :P	17:51
	→ vitor-br joined	17:51
	→ aah joined	17:52
	→ alezandro joined	17:53
\|RicharD\|	any idea ?	17:53
	→ sapristisaucisse joined	17:54
	→ Plasmastar joined	17:59
	← Plasmastar left	18:00
	→ dr_win joined	18:00
fr0sty	paste the output of 'git remote show origin' to gist.github.com	18:00
\|RicharD\|	oky	18:02
fr0sty	and tell us what the link is.	18:02
	← pheaver left	18:03
\|RicharD\|	https://gist.github.com/757490	18:03
hobodave	hey guys, I don't know what is going on here. I'm trying to cherry-pick a specific revision out of another branch, and it is silently failing	18:05
	→ Chadams joined	18:05
hobodave	I've tried "git cherry-pick 4c5bf18b90625be48c0f7ee407c3a1b542be140f" and "git cherry-pick qa~2"	18:06
	all I get in output is the equivalent as if I typed "git status"	18:06
	and nothing is changed, as confirmed by the log	18:06
	→ rolfb joined	18:06
	← rolfb left	18:07
	→ mmw joined	18:07
	→ rolfb joined	18:07
	← rolfb left	18:07
hobodave	I don't understand what I'm missing, I'm looking at the man page	18:07
	→ rolfb joined	18:07
	← rolfb left	18:07
	→ rolfb joined	18:08
	← rolfb left	18:08
	← Chadams left	18:08
SethRobertson	Could that commit already be contained by your branch?	18:08
hobodave	SethRobertson: definitely not	18:08
	→ rolfb joined	18:09
hobodave	SethRobertson: git log --pretty=oneline \| grep 4c5bf18b90625be48c0f7ee407c3a1b542be140f (no output)	18:09
SethRobertson	Try --full-history after git log, just to be sure	18:10
hobodave	are you just guessing?	18:11
	it doesn't have any results either	18:11
	→ juangiordana joined	18:11
SethRobertson	Yes, but failure to use that flag will cause reverted commits to disappear, which can cause cherry-pick to appear to fail	18:11
	→ fowlduck joined	18:11
SethRobertson	Under some circumstances, I mean	18:12
	← SkydiveMike left	18:12
fr0sty	\|RicharD\|: I would be very suprised if you could connect to http://infinity@*/airrus.git	18:12
SethRobertson	hobodave: What version of git are you using?	18:12
fr0sty	fix that url and things should go more smoothly...	18:12
\|RicharD\|	...	18:12
	→ ProLoser\|Work joined	18:12
	← Buglouse left	18:12
\|RicharD\|	i had edit it :P	18:12
hobodave	SethRobertson: 1.7.3.2	18:12
ProLoser\|Work	i tried removing a submodule but now i get this: No submodule mapping found in .gitmodules for path 'app/plugins/filter'	18:12
	← simplechat left	18:12
\|RicharD\|	the connection is oky	18:12
	→ Buglouse joined	18:12
fr0sty	\|RicharD\|: what happens when you try to push now?	18:13
\|RicharD\|	fatal: expected ok/error, helper said '2004��>i?�#��O��*þ�X��0�\�H�M�/�+�E�`	18:13
	fatal: write error: Broken pipe	18:14
fr0sty	what did you change the url to? what does 'git fetch origin' do?	18:14
hobodave	SethRobertson: https://gist.github.com/757504 my active branch is e59cc89 I'm trying to cherry-pick 4c5bf18 into it	18:14
	does it matter that 4c5bf18 is a submodule update commit?	18:15
	← sapristisaucisse left	18:15
	→ sapristisaucisse joined	18:16
	← t0rc left	18:16
\|RicharD\|	where i do	18:16
	git fetch origin ?local or remote ?	18:16
hobodave	sigh, so frustrating	18:17
SethRobertson	hobodave, perhaps. I have never tried the combination. You could try picking some non-submodule commit to see if that would work	18:17
hobodave	it does wrk	18:18
	so, wtf	18:18
	I'm confused about what's happening here	18:18
	oh this makes my head hurt	18:19
	← sapristisaucisse left	18:19
SethRobertson	Perhaps gitslave would be less confusing than submodules?	18:20
hobodave	e59cc89 was also a submodule update commit, but it commits a revision that was a child of the 4c5bf18 commit I was trying to cherry-pick	18:20
	so the changes by 4c5bf18 are already in here, just invisible unless I look at the submodule commit tree	18:20
	← MikhX left	18:21
hobodave	thanks for the suggestion SethRobertson	18:21
	→ rseifert joined	18:22
	← rseifert left	18:23
	→ sapristisaucisse joined	18:23
	← guardian left	18:25
	→ deek0146 joined	18:26
deek0146	Hey	18:27
	I know that git diff --cached is the difference between the working tree and HEAD	18:27
	→ levifig joined	18:27
deek0146	What does git diff display?	18:27
fr0sty	\|RicharD\|: local	18:27
ProLoser\|Work	i have been failing to remove git submodules using --cached, is there a way i can peruse the -cached to see the other ones i may have left behind?	18:27
\|RicharD\|	it not say nothing	18:27
bremner	deek0146: do you know what the index is?	18:27
deek0146	Emm	18:27
	when i said working treei meant index	18:27
bremner	ok ;)	18:28
deek0146	Index is stuff i've told git about but not commited	18:28
fr0sty	\|RicharD\|: can you update the gist with the new 'git remote show origin' information?	18:28
	'git remote show' would also be useful...	18:28
deek0146	Is git diff the difference between the contents of the actual directory and the index?	18:29
	(with no parameters)	18:29
	← NewtonianB left	18:29
	→ t0rc joined	18:29
albel727	deek0146: yep. exactly.	18:29
	→ NewtonianB joined	18:30
	→ disappearedng_ joined	18:30
bremner	deek0146: see the first case in man git diff	18:30
jast	deek0146: the 'git' manpage can be found at http://git.or.cz/man/git [automatic message]	18:30
	→ foo joined	18:30
	→ AMDmi3 joined	18:30
albel727	man git-diff	18:31
jast	the 'git-diff' manpage can be found at http://git.or.cz/man/git-diff [automatic message]	18:31
deek0146	It has a list of things git diff shows	18:31
	← t0rc left	18:31
foo	I did git reset files/here ... I do git commit -a , and then when I do git status I see all the files in files/here/SDKFJSKDJF as untracked... I don't want them tracked... am I doing something wrong?	18:31
bremner	yeah, and read "This form is to view..."	18:31
deek0146	a tree and the working tree, a tree and the index file, or the index file and the working tree	18:31
AMDmi3	hi; I'm truing to clone git repo via ssh, however I get "git: 'index-pack' is not a git-command" error; my idea is that that's because client is 1.6.5.7 and server is 1.7.3.3	18:31
bremner	keep reading	18:32
deek0146	Theres no blank form :(	18:32
AMDmi3	is there a way to work with repo witout updating git?	18:32
	← sapristisaucisse left	18:32
bremner	deek0146: you know what the [] mean, right?	18:33
deek0146	Optional	18:33
	So its the first one?	18:33
bremner	yes	18:33
deek0146	Hokay	18:33
	← Thell left	18:34
deek0146	And 'working tree' is the files in their current state in the directory?	18:34
	→ jrmuizel joined	18:34
bremner	yes	18:34
deek0146	Cool	18:34
	Thanks :)	18:34
	← NewtonianB left	18:34
bremner	deek0146: see also man gitglossary	18:34
jast	deek0146: the 'gitglossary' manpage can be found at http://git.or.cz/man/gitglossary [automatic message]	18:34
deek0146	Huh cool	18:35
	← ProLoser\|Work left	18:35
	→ darrint joined	18:35
darrint	If I have a commit id, how can I tell which tags are it's neighbors in history?	18:35
jast	"neighbors"?	18:36
	→ tty1 joined	18:36
darrint	jast: Maybe it's clearer if I ask what software release has this commit in it? Which tags have this commit in it's history?	18:37
tty1	what is the standard approach in git for handling a repository with large binary files (As part of the source code) that changes frequently and needs to be version tracked (like bitmaps in a video game or something)?	18:37
fr0sty	tty1: man git tag	18:37
jast	tty1: the 'git' manpage can be found at http://git.or.cz/man/git [automatic message]	18:37
fr0sty	look at --contains.	18:37
bremner	other than checking every file, is there a way to get a list of all of the files having "git svn propget" containing a given property?	18:38
darrint	ttr1: I think that was meant for me.	18:38
	tty1: even.	18:38
tty1	fr0sty: was that the right man page the bot showed?	18:38
fr0sty	yeah, darrint, that was for you...	18:38
jast	tty1: if they're not extremely large, just treat them like any other file. a couple of megabytes should be fine in any case.	18:38
tty1	oh	18:38
	lol	18:39
jast	(per file per version)	18:39
tty1	jast, the resulting repo is approaching gigs already	18:39
darrint	fr0sty: Is that a new feature? I don't see it. I have 1.5.5.6.	18:39
bremner	fr0sty: I'm pretty sure you mean to say man git-tag	18:40
jast	fr0sty: the 'git-tag' manpage can be found at http://git.or.cz/man/git-tag [automatic message]	18:40
	← Angel-SL left	18:40
fr0sty	darrint: that version is older than dirt.	18:40
	bremner: you are correct	18:40
	jast: thanks, again.	18:40
	← tiagobutzke left	18:40
fr0sty	tty1: if you are trying to store gigabytes of image files you are going to end up with a huge repository.	18:41
	→ tiagobutzke joined	18:41
tty1	fr0sty exactly and thats a huge problem, wasnt a problem on svn.. maybe ill need to migrate back, if i even can, ive been on git for a while now	18:41
fr0sty	why it is a huge problem?	18:42
	← deek0146 left	18:42
tty1	fr0sty: not all of our devs have the luxury of cloning from a huge repo, they have limited bandwidth	18:42
	← AMDmi3 left	18:42
fr0sty	you can mail them a DVD	18:42
	← dr_win left	18:42
darrint	thank you rhel.	18:42
fr0sty	can't beat sneaker-net for bandwidth...	18:42
	← tiagobutzke left	18:42
	→ MikhX joined	18:43
tty1	fr0sty: this is open source, we have to consider the everyday user too who might want the repo, etc/	18:43
fr0sty	except with station-wagon-net ;-)	18:43
tty1	lol	18:43
fr0sty	tty1 linke?	18:43
	also, how big is the working copy?	18:43
	→ NewtonianB joined	18:44
tty1	fr0sty: this one is only about 1/4 of a gig so far, but its one example: http://wiki.syncleus.com/index.php/dANN ive been working on splitting it into submodules to make it more reasonable or even splitting out some of the old history.. but thats more of a hack than a real solution	18:44
darrint	fr0sty: git-tag --contains answered my question. Thanks!	18:46
fr0sty	tty1: there is always a shallow clone...	18:46
tty1	fr0sty: yup i considered that, but they arent fully functional and fairly useless for devs	18:47
	← NewtonianB left	18:47
	→ saysjonathan joined	18:48
fr0sty	tty1 is there any value to the old binary assets?	18:48
darrint	tty1: Long long ago I solved this problem for version control of linux distros. We had a "repository" for big binary files keyed on sha1sum. We pulled files from there on demand.	18:48
tty1	fr0sty the primary value is that it allows you to check out old versions and still have them work (kinda important if someone wants version 1.1 for some legacy app or something)	18:48
fr0sty	tty1: how would you expect that to work if a dev did not have the full history?	18:49
darrint	tty1: We used git for the "source" of the distro (which packages you wanted and build scripts) and pulled all the binaries on demand.	18:50
fr0sty	I like darrint's approach.	18:50
tty1	fr0sty: well on svn its easy, the dev doesnt need to pull the entire repo to work with the repo.. they can review full history without pulling down any actual files and then when they want a particular version they pull only that version and nothing more. another words with SVN they download the parts of the repo they need on demand, but with git you pull the whole thing at once	18:50
	fr0sty the problem with darrint approach is it just skirts the issue. in the case of an application with many bitmaps and other binary files that are source (they need to be tracked and get modified as often as the source) then pulling int he binaries on demand just skirts the issue and you basically need a secondary non-git repo to address the shortcommings of git	18:52
	seems kind of pointless to have to keep half your files in svn and the other half in git just because git doesnt do something in an elegant way that svn can	18:53
	besides, i hate svn and would rather not have to resort to it anyway	18:53
fr0sty	based on the stories I have seen from svn refugees, 'svn' and 'elegant' don't belong in the same sentance ';-)	18:53
	← sonnym left	18:54
tty1	fr0sty, no svn isnt elegant at all, ia gree, thats why i moved to git, but in this single, very important area, svn works, git doesnt.	18:54
fr0sty	darrint's approach doesn't involve svn	18:54
darrint	tty1: We were willing to do the work to keep big binaries out of our source repositories. We were innoculated really early with the doctrine of no bins in source.	18:54
tty1	and id rather find a git solution.. if there is one.. but most of what i find are, at best, third-party additions and not standard git	18:54
jast	I could imagine a way to make it happen, using git notes	18:55
tty1	fr0sty darrint approach essentially involves "some mysterious repo that holds the binaries seperatly".. which if that isnt svn it is some SCM (unless you dont have multiple versions of the binaries, which of course we do, and most projects would)	18:55
jast	it's clunky, though	18:55
bremner	git-annex is directed at this.	18:55
	← sveinse left	18:56
jast	annex seems to require a separate transport mechanism for the large files	18:56
darrint	tty1: You need some unambiguous way to identify exactly which bins you want and that is in your source. So you do have history of your bins.	18:56
	← rctay left	18:56
tty1	darrint, but if you keep binaries out of git they still need to go somewhere, and if the binaries are versioned and change along with each commit in git and they need to be synced up (as most binaries do such as library dependencies or images compiled into your program) then the binaries still need to go in some repo somewhere	18:57
	darrint, right and at that point it basically comes down to making your own SCM of some sort just to handle larger files	18:57
jast	actually, upload-archive might help with making my idea less clunky	18:58
	← rolfb left	18:58
jast	that way the transport part could still be done by git, you just need a wrapper that takes care of associating the large files with git commits	18:58
	→ rolfb joined	18:58
	→ hacim joined	18:59
darrint	tty1: That git annex thing looks nice. http://git-annex.branchable.com/	18:59
tty1	even if your keeping the large files on an FTP, and pulling it in by script and you organize your binaries on the ftp into folders for each "Version" of the binary that corelates toa vommit or seriesof commits in your software.. at that point your basically doing version control by hand so its still a SCM, just one you maintain manually using a directory structure (sounds hackish to me still and sorta against the point of having a repo int he firs	18:59
	t place)	18:59
hacim	why does 'git log -- files/certs/roots/CA.pem' not show any log entries when there was a rename operation that I can find in the git log?	19:00
fr0sty	hacim: add --follow	19:00
hacim	fr0sty: aha! thanks!	19:00
tty1	darrint, yea ive been looking over git-annex as a possiblity for a few days now.. im nto fully familiar with it yet but by the looks of it it seems to be 1) third party (ie if anyone wants to use the repo they need a special privilaged version of git) 2) requires the server to be configured to have the client as a remote and vice versa 3) doesnt support bare repositories	19:01
	→ tiagobutzke joined	19:01
jast	there is no non-third-party solution, period	19:01
	← goshawk left	19:01
tty1	jast, i think that should be a good indication that git needs to be fixed then to provide a non-third-party solution .. but that aside, i dont see a viable third-party solution either (as you can see there were other reasons to reject the annex option besides it just being third party)	19:02
fr0sty	jast: there are second-party solutions. but they are not conceptually pure	19:02
	hence unacceptable, it seems...	19:02
	→ tizbac joined	19:03
fr0sty	tty1: how can git be fixed?	19:03
jast	"fixed" is a very subjective thing here	19:03
fr0sty	commit and checkout are local-only operations	19:03
	← tiagobutzke left	19:03
tty1	fr0sty huh? it has nothing to do with conceptually pure, the fact is, this is a design flaw in git that needs fixing by the sound of it. rather than skirting the issue perhaps its best to figure out how git can be fixed in the future so as to respolve this issue natively	19:04
	← juangiordana left	19:04
fr0sty	tty1: it is not a design flaw. It was a conscious design decision and a very publicized limitation of git.	19:04
jast	that is a huge undertaking, it would essentially mean rewriting large parts of git	19:04
	← tizbac left	19:05
tty1	fr0sty well my solution would be to make it so git wouldnt need to pull the whole repo in one shot when you git clone (or simple fix --depth so as not to criple the repo).. instead the objects directory shouldnt be populated at all IMO and instead should get populated on demand	19:05
	→ tizbac joined	19:05
bremner	tty1: do you know how to do these things, or is this just a list of wishes?	19:05
tty1	i cant speak for the internals of git well enough to know how much of an undertaking it would be.. but seems like something that is major enough, and common enough to be worth the effort even if it is a large effort	19:05
fr0sty	tty1: that does not work with git's data model.	19:06
	← ExtraSpice left	19:06
	← tizbac left	19:06
tty1	bremner if everyone knew every detail as to how to do these things it would be done already. thats why there is a development process for any feature	19:06
	← jrmuizel left	19:07
	→ tizbac joined	19:07
jast	the fetch/push protocol assumes that you have the entire history of a given branch present locally, in order to be able to negotiate a list of objects to be transferred	19:07
tty1	fr0sty it seems to me it could be done almost transparently to git with a few tweaks.. froma filesystem perspective if the objects directory files werent populated with data until the object itself is open ont eh filesystem for reading (at which point the appropriate file was fetched when opened) it would work just fine without needed to do much of anything to git internals.. int heory	19:08
darrint	tty1: I think it's a big deal to get done and these guys didn't win: http://caca.zoy.org/wiki/git-bigfiles	19:09
tty1	jast, all it really needs to know is the existing objects and their sha though, the physical contents are only needed for the changed files	19:09
jast	tty1: and in practice that kind of "lazy fetching" would mean that previously fast operations would now be dog slow even on repositories with much smaller files	19:09
	← tizbac left	19:09
jast	tty1: the index itself is pretty big already	19:09
foo	I did git reset files/here ... I do git commit -a , and then when I do git status I see all the files in files/here/SDKFJSKDJF as untracked... I don't want them tracked... am I doing something wrong?	19:10
tty1	jast, well im not suggesting lazy fetching be the only option or replace the current approach.. the way i see it the currernt approach of cloning would be fine, but perhaps hdd a git clone --lazy tag that tells it to do it lazy, and while this may slow down some operations (only at first since once the file is fetched it would remain and wouldnt need to be refetched) it would also significantly reduce bandwidth as many of the old files in history a	19:10
	re never even touched (Atleast their content isnt in most cases just the sha needs to be known)	19:11
	← alezandro left	19:11
	→ alezandro joined	19:11
jast	tty1: at any rate, lazy fetching would mean that you might have to download several megabytes of data for an innocent command like git whatchanged	19:12
tty1	long story short, git shouldnt need the full content of files (objects) for things that arent likely to be checked out or modified.. the only thing git really needs is the sha of such files for many operations.	19:12
	jast, for any operation that turely needs the content of all or most of the objects in the history then git wouldnt need to download anything more than what a full clone would need anyway, so that seems like a moot point to me	19:13
	← rolfb left	19:13
jast	and then a 'git diff' or 'git merge' has to fetch the files	19:13
	→ rolfb joined	19:13
	← rolfb left	19:13
jast	tty1: the point is that it gets hard to anticipate which commands will result in huge downloads and which won't	19:13
	→ rolfb joined	19:13
	← rolfb left	19:13
tty1	jast, right, git diff would need to fetch the files, but how often do you need to diff against a file int he repos history that was checked in 10 years ago? probably not much, if ever, yet git clone downloads it everytime anyway	19:14
jast	right now, virtually all commands work locally and are thus very fast	19:14
	→ rolfb joined	19:14
jast	well, even a simple git diff HEAD~10 might net you a download of 50 megabytes	19:14
	and that's an extremely generous estimate	19:14
tty1	jast, whatever the download might be, it cant possibly be more than the download of a git clone operation anyway, so still seems like a moot point to me	19:15
jast	you are missing my point	19:15
	lazy fetching makes it unpredictable when the transfer is going to happen	19:15
	→ coletg joined	19:15
fr0sty	tty1: if the history is unimportant just truncate it.	19:15
	→ sonnym joined	19:15
	→ sveinse joined	19:15
jast	I think it would be much more worthwhile to try and add limited support for pushing from repositories that used --depth	19:16
tty1	fr0sty it isnt unimportant though, it is just underused (99% of people wont need history beyond about the past 10 commits or so, but there is still that 1% that do)	19:16
fr0sty	tty1 how many of those 99% are incapable of submitting patches?	19:16
tty1	fr0sty what does that have to do with it?	19:16
	→ necromancer joined	19:17
fr0sty	you can commit and work in a shallow clone just fine	19:17
	your limitation is in pushing back from it.	19:17
tty1	jast, if --depth was able to push and otherwise not be crippled in any major way id be satisffied, doesnt seem like an ideal solution but it would be good enough for me TBH	19:17
jast	at any rate, lazy fetching would once again be vastly inferior to subversion in a repository with lots of large files	19:17
fr0sty	you have to sed patches	19:17
	s/sed/send	19:17
	if people are capable of doing that shallow clones are a suitable solution to your 'problem'	19:17
jast	because you'd probably end up accidentally downloading tons of data anyway	19:17
	which will not happen in svn	19:18
tty1	fr0sty oh right, still seems like an example of how git is broken in this regard though.. because your right, they could jsut submit patches, but why should they have to?	19:18
	← rolfb left	19:18
	→ rolfb joined	19:19
	← rolfb left	19:19
	→ khmarbaise_ joined	19:19
	← MattDiPasquale left	19:19
	→ rolfb joined	19:19
	← rolfb left	19:19
tty1	i only remark ont his because i love git, it is a great deal better than svn and cvs is almost every way except for this one glaring problem (Well that and the windows EOL bug no one seems to want to fix).. if git worked in this regard too id be as happy as can be.. but as it stands right now its a huge thorn in my side	19:19
	→ rolfb joined	19:19
	← zxiest left	19:20
	→ drizzd_ joined	19:20
	← khmarbaise left	19:21
	khmarbaise_ → khmarbaise	19:21
	→ guardian joined	19:21
	→ juangiordana joined	19:22
tty1	i have one app, like i said, ic ant even move over to git because of this. it consists of lots of bitmaps and wireframes and similar graphic design elements that sit along side the source code. they need to be versionc ontrolled with the source code as they change as frequencly as the source and the two are closely coupled. moving to git would make the thing be hundreds of gigs in size and since they change so frequencly moving the binaries to a no	19:23
	n source controlled location isnt viable	19:23
	so in short with that particular project git simple can not work, which i think says something about the neccesty of finding a solution to this shortcomming	19:23
jast	well, at any rate, it's not something the usual active developers of git are particularly interested in. if you came up with a sound proposal and contributed patches, I'm sure we'd get it included anyway.	19:24
	← drizzd left	19:24
cbreak	there are hacks around it	19:24
	external storage of big files	19:25
fr0sty	cbreak: save your breath. If it ain't built-in it ain't acceptable...	19:26
cbreak	why not?	19:26
tty1	jast, im not sure why an active developer wouldnt be more concerned about a design flaw in git that essentially prevents it from being able to be even remotely functional for certain types of projects and requires hacks for the vast majority of other projects. it would seem to me to be something that should be a top priority for a fully functional SCM. but i guess that answers my question. as much as id hate to i guess i need to consider moving bac	19:26
	k to svn then	19:26
	fr0sty i said nothing of the sort, dont put words in my mouth	19:26
fr0sty	tty1: then why the hell won't you just use one and quit whining.	19:27
tty1	being built0-in is a big plus yes.. but as i said all the third-party solutions ive seen so far have their own problems	19:27
	→ MattDiPasquale joined	19:27
necromancer	is this the right place to ask questions about gitweb?	19:27
fr0sty	?	19:27
tty1	fr0sty ive went over several of the third party solutions that ive explored and explained why they dont seem able to work for this partiular problem.. not sure why your getting all defensive rather than being constructive	19:27
fr0sty	I use [ant\|make\|gnome\|firefox\|etc], it sucks, but I get on with my life...	19:28
jast	tty1: well, git contributors are mostly motivated by their own problems, not other people's problems. it's as simple as that. :) you become a contributor by solving a problem that you consider worth solving. that doesn't obligate other contributors to have the same priorities.	19:28
necromancer	i don't think git requires hacks for the "vast majority" of other projects	19:28
	because i've been using git for a few years now and still haven't used a hack	19:28
tty1	jast, yea thats fair, and im certainly not suggesting the contributors have an obligation to solve or do anything. it just seems to me they might want to address certain known problems is all.	19:29
jast	I am somewhat interested in it just for the heck of it, but I have many, many other things to do, and I'd have to spent several weeks of time on something like that; time that I just don't have	19:29
	I'm sure similar things apply to other contributors	19:29
	tty1 nods	19:29
	← rolfb left	19:29
	→ rolfb joined	19:30
	← rolfb left	19:30
	→ rolfb joined	19:30
	← rolfb left	19:30
tty1	its just frustrating have a SCM that does things so well in almost every regard and to have one major flaw that is stopping me from using it effectivly. Hell id pay a git developer to fix the problem if it would get it into git :) ::stares at jast:: :)	19:31
	→ rolfb joined	19:31
jast	well, I'm pretty expensive :}	19:32
tty1	jast and im pretty rich, so sounds like a good match :)	19:32
	jast, actually my company "donates" regularly to open-source projects in order to get specific features included. we are an open-source oriented company so its a common practice for us	19:33
cbreak	tty1: what hacks?	19:33
	→ jrmuizel joined	19:35
sveinse	Hi. I've run "git clone --mirror" from an external git repo. Can I somhow update it incrementally again with git pull or similar?	19:35
	← petercoulton left	19:35
	→ abetaha joined	19:36
jast	sveinse: git fetch	19:36
cbreak	mirroring isn't usually to get something you fetch from	19:36
tty1	cbreak, essentially the scenario where you have a project that has large binaries that need to be version controlled along side the rest of the source (another words if on many commits the binaries change as oftena s the rest of the source and the version of the binaries and the rest of the source need to stay in sync). A common example is source that includes libraries that get statically compiled in and are upgraded to the latest version every fe	19:36
	w weeks, a less common example but more severe is the case of a 3d game where the wireframes and bitmaps exist along side the source and are updated very often.	19:36
cbreak	it's to create a complete mirror	19:37
	!!!tty1 I don't see that as common problem	19:37
	I never had to do any hack	19:37
sveinse	What is the real difference between --mirror and --bare exactly?	19:37
cbreak	one creates a mirror, the other one does not	19:37
	→ deek0146 joined	19:38
deek0146	The git glossary is amusing	19:38
	← cking_nj left	19:38
tty1	cbreak, seems common to me, ive worked on half a dozen repos with this problem so far, more than those that dont have the problem (almost any java app still using ant for example will have this problem, and virtually all 3d and 2d games would have this problem)	19:38
sveinse	Point is, I do want it to be a mirror. I'd hoped I could update the mirror instead of refetching (or more precisely cloning) it all over again	19:39
	→ brosner joined	19:39
	← brosner left	19:39
	← jrmuizel left	19:39
sveinse	jast: git fetch doesn't work, since its not a working tree	19:39
	← vezult left	19:39
cbreak	tty1: I never had that problem	19:40
	committing binaries seems dumb to me in any case	19:40
	there's no way to use that binary in a situation on a different platform for example	19:40
jast	well, if you have versioned binaries, you have to do something	19:40
necromancer	cbreak: what would you call PSD/EPS/AI files?	19:40
cbreak	media	19:41
	ai is text.	19:41
jast	binary != executable	19:41
necromancer	well, git thinks they're binary files	19:41
deek0146	binary == executable	19:41
necromancer	and so does svn	19:41
	and cvs	19:41
deek0146	just executable also equals other things	19:41
tty1	cbreak, yea but how often have you had to statically compile in other libraries that you keep up to date and modify concurrently? i know nothing about your usage	19:41
necromancer	which is all the SCMs I've worked with ;)	19:41
	→ ExtraSpice joined	19:41
	→ brosner joined	19:41
jast	deek0146: not according to the way 'binary' is used by git as a term	19:41
	← brosner left	19:41
deek0146	Oh right	19:41
cbreak	never	19:41
jast	'binary' just means: non-textual data	19:42
cbreak	libraries get included and as source	19:42
deek0146	Sorry of course, came into a conversation halfway through	19:42
	Yea	19:42
tty1	when i say binary im including images, bitmapes, wireframes, etc.. there are lots of things	19:42
cbreak	and compiled from it	19:42
	→ brosner joined	19:42
deek0146	Just wondering, does git have capabilities to store any image files not as binary?	19:42
	Or any other media	19:42
	Or is it just text	19:42
necromancer	i mean, i store photoshop/illustrator sources in almost all my repos. if i fuck up and save accidentally i can revert back to the original file that i was given	19:42
tty1	deek0146, yea like svg it could handle as text fine	19:42
jast	deek0146: you can use custom filters. most of the time, simply storing them as-is works fine, though.	19:43
deek0146	What about bmps or jpegs	19:43
necromancer	storing bmps & jpgs as text rather than bin?	19:43
deek0146	I know i'm just thinking it'll make the history huge (on disc)	19:43
jast	sveinse: git fetch works in bare repositories	19:43
	git pull doesn't	19:43
sveinse	ah, my bad.	19:44
jast	also, it doesn't refetch; it reuses existing data	19:44
	← rolfb left	19:44
jast	so it's exactly what you want, methinks :)	19:44
tty1	deek0146, yea thats the very problem im facing, and trying to discuss here (Although i think we exhausted the conversation, loong story short, git doesnt workw ell with it, it will make the repo huge, and pretty much make it unmanagable in time)	19:44
sveinse	thanks man!	19:44
jast	the difference between --bare and --mirror is that --mirror includes more refs, e.g. remote branches in refs/remotes in the original repository	19:44
deek0146	Yea. So its just best not to tell git about media resources?	19:44
tty1	deek0146, well seems the best soltuion is not to include large binaries in git of any kind	19:45
jast	depends on the size and volatility of yoru media files	19:45
	gotta step out for a few minutes	19:45
tty1	jast, thanks for all your help	19:45
fr0sty	:jk:q	19:46
	heh, don't mind me...	19:46
deek0146	Kk. Its just I'm working on a game and my artist is a programmer so git won't be tough for him, i thought it would be cool to use git but it occured to me that it could pile up especially if he keeps changing them	19:46
necromancer	captain vim over there	19:46
cbreak	deek0146: git stores everything as binary	19:46
deek0146	Boo.	19:46
	← ExtraSpice left	19:46
tty1	deek0146, thats pretty much the exact same complaint i had.. and there dont seem to be any elegant solutions, those that exist are third party and have their ownshort commings	19:47
deek0146	I was thinking third party	19:47
	Whats wrong with them? :(	19:47
cbreak	git makes some assumptions about the data it stores	19:47
tty1	deek0146, well depends, git annex is one for example, but it seems to require the server to know about the clients (and add them as remotes) and the server cant be a bare repo .. which rules that one out for me	19:47
cbreak	that assumption does hold for normal source code, but it does not hold for computer generated data in general	19:47
deek0146	I suppose it wouldn't work well for compressed image files	19:48
cbreak	such as image files, computer generated text files or similar	19:48
jast	oh well	19:48
deek0146	It would work with bitmaps	19:48
tty1	deek0146, there is also submodules, but that just breaks the problem up into more than one repo, but the overall size isnt reduced if you need it all	19:48
	deek0146, etc	19:48
jast	turns out that I'm still pretty good at slamming my forehead into door frames	19:48
deek0146	The diff log would look like pixel 24,25 changed to colour 30,60,255	19:48
	But with compressed images like png's	19:48
	One change to the image would change the layout of the file hugely	19:48
	← brosner left	19:49
deek0146	Ah well	19:49
	I'll just remove the images from the repository	19:49
	Bye bye, got 20 minutes battery left ><	19:50
	→ ExtraSpice joined	19:50
jast	now I won't get to any shops in time. I guess that means no food today. :/	19:50
	← deek0146 left	19:50
	→ tzaeru joined	19:52
fr0sty	tty1: what is wrong with a solution like darrint's from earlier?	19:52
	→ madewokherd joined	19:53
fr0sty	aside from being hand-rolled, naturally...	19:54
cbreak	I would just store big binary files in an external store	19:54
	← ExtraSpice left	19:57
fr0sty	cbreak: you could do one better and git add a file with filename->sha1 for the current commit.	19:57
cbreak	problem is that git add adds the file into git	19:58
	I'd use git hash-file or what ever	19:58
	← steph021_mac left	19:58
fr0sty	add a pre-commit/post-checkout hook to make sure the files and hashes match up. (store the files in the store by sha1)	19:58
	→ ExtraSpice joined	20:00
	→ dreiss joined	20:00
	→ Mahjongg joined	20:00
ziro`	guys, i've got a workflow conundrum to discuss. I've got two servers, production (master) and staging (staging). When a new feature is developed (feature1), it needs to be merged with staging for sign off, once signed off it needs to be merged with master for roll out on production	20:00
Mahjongg	hello, is it possible to clone from a local git repository	20:01
ziro`	the issue i've got is that some features never make it past staging	20:01
Mahjongg	how?	20:01
ziro`	Mahjongg: clone ../my/other/repo ?	20:01
Mahjongg	cool	20:02
jast	that even reuses files so it uses less space, at least if your filesystem supports hardlinks	20:02
	← eletuchy left	20:02
fr0sty	ziro`: if you don't like the feature just reset to before the merge...	20:04
	or create test-integration branches and release those to the staging server.	20:04
ziro`	the problem with reseting is that one feature may be signed off before another	20:05
fr0sty	ziro`: then merge the branches into production individually as they are signed off	20:06
ziro`	e.g. feature1 > staging ... feature2 > staging > feature2 signoff ... feature1 declined	20:06
fr0sty	ziro`: how are you 'declining' features?	20:06
ziro`	e.g. the client changes their mind	20:06
fr0sty	no, how do you remove them from the staging branch?	20:07
	→ krosenvo- joined	20:07
	← ExtraSpice left	20:07
	→ vvsh joined	20:07
ziro`	that's a good question	20:07
fr0sty	that is where the reset comes in...	20:07
	in my mind, at least...	20:08
	← _Vi left	20:08
	→ _Vi joined	20:08
ziro`	can you reset individual commits?	20:08
	← Mahjongg left	20:09
wereHamster	ziro`: what do you mean by 'reset'	20:09
	?	20:09
	→ ch077179 joined	20:10
ziro`	remove, negate, pretend it never happened	20:10
wereHamster	ziro`: man git-revert	20:10
jast	ziro`: the 'git-revert' manpage can be found at http://git.or.cz/man/git-revert [automatic message]	20:10
ziro`	revert	20:10
	→ ExtraSpice joined	20:10
	← mmc left	20:11
fr0sty	ziro`: that will create a new commit which undoes the previous one.	20:11
ziro`	ok fine	20:11
tty1	fr0sty i dont recall a solution other than saying "put the files in an external locationa nd pull it in via a script" or something to that effect the problem with that is that it works fine for a few binaries.. but if you are updating your binaries every week or so, and the version of the binary needs to be synced with specific commits int he repo (which is common for most cases where this would be needed) then you wind up needing to manually maint	20:11
	ain an external REPO that does version control much like git does except you cant use git since the files are big. so not really a solution	20:11
fr0sty	tty1: look up at what jast, cbreak and I said recently...	20:12
	you would end up with a listing, inside your current repository, detailing what versions of what files were associated with that commit.	20:12
	→ Bisu[Shield] joined	20:12
jast	well, it's not really manual. you can write a program to do it. takes a few hours, tops.	20:13
fr0sty	a script would go to the external pile of binaries, and copy/link the appropriate ones.	20:13
	→ kukks joined	20:13
	← ExtraSpice left	20:13
fr0sty	the pile of binaries would be labelled by their sha so you sould just need to keep them backed up somewhere, not version controlled.	20:13
consumerism	is there a way to make git status always pipe to less if output is bigger than a screenful?	20:13
fr0sty	consumerism: man git-config	20:13
jast	consumerism: the 'git-config' manpage can be found at http://git.or.cz/man/git-config [automatic message]	20:13
	necromancer → necromancer`bzzy	20:14
jast	consumerism: not that I know	20:14
	it's all or nothing	20:14
fr0sty	look at the pager options. pager.<cmd> might work...	20:14
jast	you could write your own pager, though, that wraps your favourite real pager	20:15
	and, in fact, I think some pagers have an option like that	20:15
consumerism	so - how come git log does this automatically?	20:15
jast	git log always paginates for me	20:15
consumerism	jast: me too	20:15
	but not git status	20:15
jast	well, you can turn on paging unconditionally for status, making it work the same way as log	20:16
consumerism	i mean it paginates if the log is bigger than my screen	20:16
	→ ExtraSpice joined	20:16
jast	then that's some pager-specific configuration you've got	20:16
consumerism	hm, i don't think so - piping ls for instance which is less than a page doesn't act the same	20:17
	→ thiago_home joined	20:17
fr0sty	this might be a function of the arguments given to less.	20:17
consumerism	fr0sty: how might i learn more about that	20:18
jast	less has the -F option for this	20:18
	← TheFox21 left	20:18
fr0sty	git sets LESS to FRSX by default.	20:19
	according to the git-config manpage, at least.	20:19
	← ExtraSpice left	20:20
	← pgib left	20:20
	→ Rickardo1 joined	20:21
	thiago_home → thiago	20:22
	→ ExtraSpice joined	20:23
jast	but only if LESSOPTS is not defined, I think. that would explain why it always paginates for me.	20:28
	I do have LESSOPTS set.	20:29
	→ rseifert joined	20:29
	← rseifert left	20:31
ziro`	my problem is at work we're using svn. We have a staging branch and 'feature' / 'developer' branches which are merged with the staging branch as needed. Pushing to live is a case of doing manual diff from staging to production over sftp	20:31
	the deployed version doesn't represent the staging branch very well	20:32
	the deployed / production code isn't actually SCM'd at all	20:32
	→ eletuchy joined	20:33
	← ExtraSpice left	20:35
cbreak	that sounds dumb	20:39
	→ dr_win joined	20:40
	→ Unknown[OFF] joined	20:42
	→ sapristisaucisse joined	20:43
	← eletuchy left	20:45
cbreak	why don't you just scp over the whole staging branch?	20:46
	← dr_win left	20:46
	← sapristisaucisse left	20:47
WildPikachu	in a rebase when one rebases a branch onto another branch (feature onto master, from maint), what would the meaning of the ^ be in this .... git rebase --onto master maint^ feature ?	20:48
cbreak	WildPikachu: ^ means always the same	20:49
	"first parent of"	20:49
	WildPikachu: man git-rev-parse	20:49
jast	WildPikachu: the 'git-rev-parse' manpage can be found at http://git.or.cz/man/git-rev-parse [automatic message]	20:49
	← thiago left	20:51
WildPikachu	hrmmmm , so one could just rebase against maint, and leave out the ^ when transplanting the commits onto the other branch?	20:51
	→ Kobaz joined	20:51
cbreak	yeah	20:52
	try :)	20:52
WildPikachu	awesome :)	20:52
Kobaz	so i just did a stash, and then pull, and stash pop... which had some conflicts, which are now fixed...	20:52
WildPikachu	I'm reading a book :) just making sure I understand this 100%	20:52
	thanks cbreak	20:52
Kobaz	so now i do a git status... and i see three modified files... but i do a git diff... and diff shows no changes	20:52
	→ thiago joined	20:53
fr0sty	'git diff' compares index <-> working copy	20:53
Kobaz	but then i do a git commit -a -v... and i get a diff	20:53
fr0sty	'git diff --cached' compares index <-> HEAD commit	20:53
Kobaz	how do i get the same diff	20:54
	oh okay	20:54
fr0sty	'git diff HEAD' does working-copy <-> HEAD	20:54
Kobaz	so what's the index defined as... when doing a straight git diff	20:55
	→ `andy joined	20:56
	→ tclarke joined	20:57
tclarke	I'm using git 1.7.3 with a git-svn mirror and I'm having some trouble	20:58
	working on a subversion branch (corresponding git branch is active)	20:58
	made some changes to the repo and checked them into the git branch	20:58
	git svn dcommit	20:58
	I get "Committing to https://svn.repo/path ...	20:59
Kobaz	fr0sty: where could i read about the different parts of the commit tree, like index versus cached, vs HEAD	20:59
tclarke	M path/to/a/modified/file	20:59
	Incomplete data: Delta source ended unexpectedly at /usr/libexec/git-core/git-svn line 573	20:59
cbreak	Kobaz: index is staging area. all you have added.	20:59
fr0sty	Kobaz: the Git Book is good	20:59
	Pro-Git as well	20:59
	any book on git will have a decent explanation of that.	21:00
Kobaz	i've read some pieces of pro git, k	21:00
	so regular git diff, by itself, will compare the changes made compared to the base files before those changes?	21:02
cbreak	git diff compares working files with staging area	21:05
fr0sty	Kobaz: git diff compares working copy to index. if you have changes you have not added (with git add) they will show up.	21:05
	once you add them they will not show up in 'git diff'	21:05
Kobaz	oh	21:05
fr0sty	but you can see the changes which are in the index, but not yet committed by using 'git diff --cached'	21:06
Kobaz	ooh okay	21:06
	ziro` → lukebarton	21:06
guardian	i used git-svn to clone some svn repo hosted on code.google.com — looks like it worked but now i would like to update it; when i type "git pull" it says there is no remote repo specified, what should i do please?	21:06
fr0sty	if you want to ignore the index and see the difference between the working copy and the last commit (HEAD): 'git diff HEAD'	21:06
Kobaz	k	21:07
fr0sty	guardian: you probably want git svn [rebase\|update]	21:07
	:q	21:07
	crap!	21:07
guardian	fr0sty: oh i thought i could use it as a normal git repo after git svn clone	21:08
fr0sty	guardian: you can use the repository normally (sort of), but if you want to pull changes from the svn repo you need to use 'git svn' commands.	21:09
	← darrint left	21:10
guardian	ok	21:10
	thx	21:10
	← madewokherd left	21:16
	← [Po]lentino left	21:16
	← abetaha left	21:16
	→ abetaha joined	21:16
	← lukebarton left	21:17
	necromancer`bzzy → necromancer	21:17
	→ jrabbit joined	21:21
	← Zephyrus left	21:24
	← jrabbit left	21:26
	→ acalbaza joined	21:30
acalbaza	im in the middle of resolving conflicts during a rebase, but for some reason, when i run git difftool, i get nothing	21:30
	no diffs open with a difftool	21:31
	they just dump to console	21:31
	any hints?	21:31
	→ FYLGood joined	21:32
acalbaza	nevermind.... mergetool is what i need	21:32
	dug	21:32
	duh	21:32
	← FYLGood left	21:32
	→ datafirm joined	21:32
datafirm	Hi	21:33
	I have some unstaged changes, all of which I'd like to push to a new branch. Is this possible?	21:33
	Ic an create a stash and apply the stash on a new branch.. that should work.	21:34
fr0sty	datafirm: will the new branch be based off of your current commit?	21:39
datafirm	fr0sty: The stash trick worked!	21:39
	← fowlduck left	21:39
	→ eletuchy joined	21:40
fr0sty	if you were starting the new branch from the current commit: 'git checkout -b newbranch' would have worked it.	21:41
datafirm	fr0sty: I had not made a commit.. I'd modified some files and realized I wanted a new branch with those changes based off master.	21:41
fr0sty	were you on master?	21:42
tclarke	getting "Incomplete data: Delta source ended unexpectedly at /usr/libexec/git-core/git-svn line 573" when doing git svn dcommit to a subversion branch	21:42
	any ideas?	21:42
fr0sty	tclarke: don't know about git svn, sorry.	21:44
	← thiago left	21:47
	← tclarke left	21:47
	← coletg left	21:51
	→ lousygarua joined	21:52
lousygarua	is there a way to download only a certain revision from a repository so i won't have to download the whole history of the project?	21:52
	→ flaguy48 joined	21:54
tty1	fr0sty: right thats the solution i will wind up going with. my point is just that it is essentially just a homebrewed SCM your making at that point that is script handled. So while it may be the right solution right now it still feels like a hack to get around a fault in git. so i still feel it should be addressed by git, but thank you for the solution	21:54
fr0sty	tty1: it is a homebrew backing store, really. git is still managing the file versions, you are just moving the contents out of .git/objects/.	21:56
tty1	fr0sty: well i guess thats where annex comes in.. it is intended to do that just in a git native way... which is fine. i guess i just feel annex went about it the wrong way and should be native to git is all.	21:57
	→ hunterloftis joined	21:57
hunterloftis	about 5 commits ago a team member deleted some files he thought we weren't using anymore	21:58
	Is there a simple way to restore them if I know their path & the commit where they did exist?	21:58
	→ ericdc joined	21:58
tty1	hunterloftis, just do a revert most likely	21:58
fr0sty	haven't investigated annex in depth, but it looks like a similar approach.	21:58
hunterloftis	tty1: you can revert just a single file?	21:59
Bombe	hunterloftis, git checkout <commit-id> -- <file1> <file2> ... <fileN>	21:59
	hunterloftis, afterwards git-add & git-commit to your pleasure.	21:59
tty1	hunterloftis, if the commit is just a few files you wanna restore but not all of it you can do a checkout of that commit for just those specific files	21:59
hunterloftis	Bombe: baller, thanks	21:59
tty1	then recommit them	21:59
fr0sty	tty1: home-brew is better in some ways though, because you can distribute it within your repo instead of having to tell people to upgrade their git.	21:59
Bombe	hunterloftis, reverting that commit might be a cleaner solution.	21:59
	Unless the commit does more than only deleting those files.	22:00
	← lousygarua left	22:00
hunterloftis	Bombe: yeah but it was several commits ago	22:00
tty1	fr0sty: yea there is that. of course then if you use a shell script you have platform dependence issues to consider too	22:00
fr0sty	also true...	22:01
	still boils down to the fact that git doesn't play well with big files....	22:01
	← necromancer left	22:01
	← segher left	22:01
Bombe	hunterloftis, doesn’t matter. git-revert reverts that specific commit.	22:02
	→ Tommy[D] joined	22:02
hunterloftis	thanks for your help guys	22:02
	← saysjonathan left	22:03
tty1	fr0sty yea, i think for the past stuff ill just use submodules and for future stuff im considering a script or other homebrew, maybe annex i dunno	22:03
	→ steph021_mac joined	22:04
	← guardian left	22:04
	→ guardian joined	22:05
	← tty1 left	22:05
	→ sterlo joined	22:09
fr0sty	tty1: best of luck to you.	22:10
	← datafirm left	22:10
fr0sty	I gotta jet.	22:11
	← sonnym left	22:12
	← ben_alman left	22:12
kylehayes	I feel like git ought to be on OS X out of the box	22:12
	→ blaenk joined	22:15
	← alezandro left	22:15
	→ rgr joined	22:15
	← RoboTeddy left	22:18
	→ programmerq joined	22:19
	ChanServ set mode: +v	22:20
wereHamster	kylehayes: better to compile yourself, osx tends to be out of date with certain packages	22:21
kylehayes	gotcha	22:22
programmerq	Hello. I have a subversion repository that has a non-standard structure that's just awful. I'm trying to get 'git svn clone' to make me a copy, but I'm not sure how. Here is what the directory structure looks like:	22:22
	trunk branch1 branch2 branches/branch3 tag1	22:22
	"git svn clone -s https://svnserver/path/to/module/" doesn't cut it here. Am I stuck with only partial branch discovery and/or treating a tag like a branch?	22:24
	or is there some option that can allow me to specify the location of a single branch?	22:24
	→ Slix` joined	22:28
wereHamster	programmerq: don't use -s	22:32
	← Tommy[D] left	22:33
	→ realseb joined	22:34
programmerq	wereHamster: right-- my question is what would be correct to use?	22:34
	"git svn clone -T trunk --branches / --tags / <url>" doesn't seem to fit either	22:35
	because at the top level of this module there are branches, tag(s), and trunk all in one directory. Not to mention another directory called branches which isn't a branch, but contains a branch.	22:36
	← sveinse left	22:37
programmerq	I'm stumped because I figured there'd be an option like --branch that lets me specify a specific branch location (and can specify multiple times for multiple branches)	22:37
	← Bisu[Shield] left	22:37
programmerq	yet there doesn't seem to be such an option	22:37
	→ Tommy[D] joined	22:39
programmerq	"git svn clone -T trunk --branch branch1 --branch branch2 --branches branches --tag tag1 --tags tags" is basically what I was hoping for, but the manpage, help text, documentation, etc don't seem to have something like --branch, so I was hoping that there would be an equivalent	22:39
	← sraue left	22:40
	← zombor left	22:42
	← Buglouse left	22:45
	→ Buglouse joined	22:47
	← khmarbaise left	22:50
programmerq	well, asking my question made me think more anout my problem, and I've thought of a different way to get what I need. Thanks for letting me monologue!	22:51
	← programmerq left	22:51
	← doener left	22:53
	→ sraue joined	22:53
	← guardian left	22:54
	→ guardian joined	22:55
cbreak	not using svn would be a good way :/	22:55
	or not attempting to do branches with subversion	22:55
	or at least doing it with some kind of structure :/	22:55
	→ Figaroo joined	22:59
Figaroo	I'm getting an error	22:59
wereHamster	you can keep it	22:59
Figaroo	lol	23:00
	hang on I got to figure out how to copy from my shell	23:00
cbreak	windows?	23:00
	if so, good luck! you'll need it	23:00
	microsoft has it's problem copying copy&paste :/	23:00
Figaroo	cygwin actually	23:01
	→ Mage joined	23:02
	→ kylehayes_ joined	23:05
	← kylehayes left	23:08
	← kylehayes_ left	23:10
	→ tiagobutzke joined	23:10
	← unreal left	23:13
	← dreiss left	23:13
	→ unreal joined	23:14
	→ sonnym joined	23:15
	← juangiordana left	23:23
	→ tizbac joined	23:23
	← joel_falcou left	23:24
	← tizbac left	23:25
	← acalbaza left	23:25
	→ RoboTeddy joined	23:26
Figaroo	git help remote is giving me "bash: get: command not found", is that normal?	23:26
FauxFaux	No. It's probably trying to luanch man or your browser.	23:27
	→ akoskm joined	23:27
Figaroo	browser = terminal?	23:27
	← Nikoli left	23:27
FauxFaux	No, msysgit launches a webbrowser with the html docs.	23:27
	← Tommy[D] left	23:28
Figaroo	so what do I do?	23:28
FauxFaux	Panic.	23:29
	→ Nikoli joined	23:29
Figaroo	I'm one step ahead of you there.	23:30
FauxFaux	Work out what it's trying to do and how it's failing at escaping. Or just run man git-remote if you're on a manny platform.	23:30
jast	the 'git-remote' manpage can be found at http://git.or.cz/man/git-remote [automatic message]	23:30
	→ joel_falcou joined	23:30
	← Buglouse left	23:32
	→ Buglouse joined	23:34
	→ tzafrir_laptop joined	23:39
	← mmw left	23:40
	← MattDiPasquale left	23:41
WildPikachu	warning: remote HEAD refers to nonexistent ref, unable to checkout. <= anyone seen that when cloning using rsync?	23:42
tzafrir_laptop	any RTFM as to converting a SCCS repo to git?	23:45
FauxFaux	1) Hunt down whovere decided to use SCCS. 2) Hurt them. 3) ... 4) Profit?	23:46
tzafrir_laptop	FauxFaux, 3) that guy then gives consulting for maintaining SCCS. 3a) get your share from that	23:47
	← Nikoli left	23:48
	← akoskm left	23:49
	→ Nikoli joined	23:50
	→ dreiss joined	23:54
Figaroo	okay why is this happening "$ git push origin master	23:56
	error: cannot run ssh: No such file or directory	23:56
	fatal: unable to fork	23:56
	"?	23:56
daed	what happens when you type "ssh" in the console	23:56
fr0sty	Figaroo: 'which ssh' ?	23:56
	← Nikoli left	23:56
Figaroo	-bash: ssh: command not found	23:56
FauxFaux	Figaroo: What platform are you on? i.e. which distro?	23:56
Figaroo	is what happens	23:56
	mintty on cygwin on windows	23:57
fr0sty	that might be your problem...	23:57
FauxFaux	Lolz. Yeah. What he said.	23:57
	→ zombor joined	23:58
Figaroo	aw, what can I do	23:58
	besides panic?	23:58
FauxFaux	Uninstall cygwin and use msysgit instead.	23:58
fr0sty	uh, Don't panic ?	23:58
	→ Tommy[D] joined	23:59
Figaroo	Okay takes deep breath	23:59
	← robotarmy left	23:59
fr0sty	Figaroo: what does 'echo $PATH' give you?	23:59
Figaroo	/usr/local/bin:/usr/bin:/bin:/cygdrive/c/Program Files/Common Files/Microsoft Shared/Windows Live:/cygdrive/c/Program Files (x86)/Common Files/Microsoft Shared/Windows Live:/cygdrive/c/Windows/system32:/cygdrive/c/Windows:/cygdrive/c/Windows/System32/Wbem:/cygdrive/c/Program Files (x86)/Common Files/Intuit/QBPOSSDKRuntime:/cygdrive/c/Program Files (x86)/Common Files/Roxio Shared/10.0/DLLShared/:/cygdrive/c/Program Files (x86)/Common Files/Roxio Shar	23:59

Logs Search ←Prev date Next date→

Channels Documentation