IRCloggy #git 2009-01-24

Provider of IRC logs since 2005.

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

	2009-01-24
	← stunnaboi left	00:00
	→ rubbish_ joined	00:03
	→ charlenopires joined	00:04
	→ cytrinox_ joined	00:04
	← charlenopires left	00:04
	Beeet → Beket	00:05
	← svend left	00:05
	← jstad left	00:06
	← johnw left	00:06
	← __doc__ left	00:07
	← bradly left	00:07
	→ rubbish__ joined	00:07
	← ankit9 left	00:07
	← ahupp\|fb left	00:10
	← patrick_ left	00:10
	← WALoeIII left	00:10
	← sdboyer-laptop left	00:11
	← softdrink left	00:12
	← name left	00:13
	→ whee joined	00:18
	← cytrinox left	00:18
	← nothinghappens left	00:18
	← mikearr left	00:18
	→ mikearr joined	00:19
	← exg left	00:19
	← bantic left	00:19
	← Channel6 left	00:19
	← rubbish left	00:20
	→ ahupp\|fb joined	00:20
	← jmatthews-out left	00:20
	→ jmatthews-out joined	00:22
	← mikearr left	00:22
	→ mikearr joined	00:22
gambler	I have a git directory on a mounted cifs filesystem - is that ok?	00:23
	When I try too git pull from it, it is not updating my local copy	00:23
	← korpios left	00:23
gambler	[orion@zeus workspace]$ git diff --cached	00:24
	[orion@zeus workspace]$ pwd	00:24
	/mnt/project/workspace	00:24
	[orion@zeus workspace]$ cd ~/build/workspace/	00:24
	[orion@zeus workspace]$ git diff --cached	00:24
	[orion@zeus workspace]$ git pull /mnt/project/workspace/	00:24
	From /mnt/project/workspace	00:24
	* branch HEAD -> FETCH_HEAD	00:24
	Already up-to-date.	00:24
	[orion@zeus workspace]$ git diff -r --brief . /mnt/project/workspace/	00:24
	← rubbish_ left	00:25
	→ WALoeIII joined	00:25
	→ exg joined	00:26
doener	a) pastebin b) "git diff" is missing (just "git status" would be easier anyway), c) the final "git diff" had no output?	00:26
gambler	the output of the final git diff is a big list of changes	00:28
	← alkemann left	00:28
gambler	doener, (I will pastebin next time)	00:28
	← rubbish__ left	00:29
doener	at least three possible reasons for that: 1) "git diff" isn't empty (--cached doesn't tell anything about the working tree)	00:29
	2) The .git contents differs (not quite unexpected)	00:29
	3) Untracked files	00:29
	← nud left	00:30
doener	oh, and 4) you just have commits in the repo into which you pulled that don't exist in the one from which you pull	00:30
gambler	1 exits silently in both places	00:30
	→ girishr joined	00:31
	← codebrulee_ left	00:32
gambler	how can i resolve #4?	00:32
doener	git fetch /man/project/workspace HEAD; git reset --hard FETCH_HEAD	00:33
	← pjhyett left	00:33
doener	resets your current branch to the same commit as HEAD in the other repo	00:33
	→ pjhyett joined	00:33
doener	and the --hard causes the index and working tree to be updated as well	00:33
gambler	doener, I did that and files are still out of sync	00:34
	← tango_ left	00:35
	← pjhyett left	00:35
	→ pjhyett joined	00:35
gambler	weird. I had to re-add files I already added and commited	00:37
	i wonder if I had a data corruption issue	00:37
	doener: ty	00:37
	→ lorandi joined	00:38
	← WALoeIII left	00:38
	← alperkanat left	00:39
	← perezd88 left	00:44
	← girishr left	00:50
	← Ariens left	00:51
	← krawek left	00:51
	→ nick_h[litage] joined	00:57
	← krh left	00:57
	→ WALoeIII joined	00:59
	→ solydzajs_ joined	00:59
	← solydzajs_ left	00:59
	→ travlr joined	01:03
	← ph^ left	01:04
	← foutrelis left	01:06
	→ mithro joined	01:07
TheGeralt	Hi, I have a strange error " $ git stash	01:07
	/usr/libexec/git-core/git-stash: line 117: /tmp/sample/.git/logs/refs/stash: No such file or directory", I didn't do anything fancy, just running git init, creating a file commiting it, creating another and running git stash. Is this a bug?	01:07
vmiklos	looks like. what version?	01:11
TheGeralt	1.6.0.6	01:11
	"mkdir .git/log/refs" solves it	01:11
vmiklos	that's not even ancient	01:11
	it would be nice if you could try with 1.6.1, though	01:12
	← mithro left	01:12
	→ mithro_ joined	01:12
	← lois[pac] left	01:12
TheGeralt	yes, I'll do that, but tomorrow. time to go to bed, bye	01:13
	← SRabbelier left	01:13
vmiklos	bye	01:13
	← TheGeralt left	01:13
	← travlr left	01:14
	← mbroeker left	01:15
	→ snitko joined	01:15
	← snitko left	01:15
	← solydzajs left	01:17
	→ tango_ joined	01:17
kansan	my first git commit	01:17
	my that was easy	01:17
	→ travlr joined	01:18
	→ bryanl joined	01:23
	← ludde left	01:26
	← markelikalderon left	01:27
	← ahupp\|fb left	01:28
	← tvw left	01:30
	← Tv left	01:31
	→ markelikalderon joined	01:35
	← markelikalderon left	01:35
	→ tvoet_ joined	01:38
	← jpeeler left	01:39
	← kek left	01:39
	← jerome187 left	01:40
tvoet_	can anyone explain why all of a sudden i would get this message upon trying to do a git pull "you asked me to pull without telling me what branch you wanted to merge with	01:41
	this doesn't happen very often, but it just did	01:41
	← pjhyett left	01:42
	statim_ → statim	01:42
	← _slacker_ left	01:45
parasti	tvoet_: "git pull" doesn't work from "where-ever". you need to be on a branch for which tracking config exists	01:45
	→ karant joined	01:45
parasti	tvoet_: i.e., branch.$branch.remote and branch.$branch.merge	01:45
	→ lednerk joined	01:45
tvoet_	parasti: i am in my repo, and on the master branch.	01:46
vmiklos	typical settings are: git config branch.master.remote origin; git config branch.master.merge refs/heads/master	01:46
tvoet_	sorry let me re-phrase, i am in my local clone .	01:46
	with only one branch ( master )	01:47
vmiklos	what is your .git/config look like?	01:47
tvoet_	vmiklos: now it looks like this http://gist.github.com/51300	01:48
	← karant left	01:49
tvoet_	vmiklos: your syntax fixed my problem.	01:49
	thanks	01:49
vmiklos	:)	01:49
	→ markelikalderon joined	01:49
vmiklos	np	01:50
	← markelikalderon left	01:50
tvoet_	vmiklos: i am curious, as to how come those values weren't defaulted to the server i cloned from.	01:50
	vmiklos: any ideas?	01:50
	→ lresende joined	01:51
vmiklos	not really. is this some old version?	01:51
	also you chould check if it's reproducible	01:51
	← ice799 left	01:52
	← wwalker left	01:54
	→ madewokherd joined	01:55
	→ markelikalderon joined	01:57
	← markelikalderon left	01:58
tvoet_	vmiklos: it was a clone of maybe a week or two old ( without work being done on it )	02:00
	→ markelikalderon joined	02:01
tvoet_	vmiklos: as for reproducible, i can try, but don't know how it happened	02:01
	→ jim_c_ joined	02:02
vmiklos	if it isn't reproducible, then there isn't much we can help	02:03
tvoet_	vmiklos: i completely understand. but like i said, i haven't used this repo in close to 2 weeks, and it worked fine last time.	02:05
	← Beket left	02:05
	→ Beeet joined	02:05
vmiklos	then really no idea, sorry:)	02:06
	→ jchris joined	02:06
tvoet_	vmiklos: if it does happen again, i will be sure to post back here. thanks for the fix, it was much appreciated	02:06
	→ Leefmc joined	02:10
vmiklos	actually posting it back to the mailing list is a better idea:)	02:11
tvoet_	vmiklos: which mailing list	02:19
vmiklos	topic :)	02:19
	you don't have to subscribe	02:20
	scientes → foobar	02:21
	foobar → scientes	02:21
	← WALoeIII left	02:25
	→ Tv joined	02:27
	← markelikalderon left	02:28
	→ softdrink joined	02:33
	← inetic left	02:35
neunon	OK. Back. doener: yes, that's what the config says... what should I edit it to be?	02:35
doener	neunon: heh, that's what I call timing. I just returned from the couch :-)	02:36
	neunon: fetch = trunk:refs/remotes/git-svn	02:37
neunon	doener: heh, and my series of phone calls just ended (criminy, where'd 3 hours go?!)	02:37
doener	neunon: then "git filter-branch --subdirectory-filter trunk -- --all"	02:37
neunon	okay.	02:37
doener	and then "rm .git/svn/git-svn/.revmap_*"	02:38
neunon	I've used filter-branch before, but what's this particular filter-branch doing?	02:38
doener	basically it makes the "trunk" directory the new repo root	02:39
neunon	Ah, great. :)	02:39
	Oh joy, 16000 revisions left	02:39
doener	shouldn't take too long. It just needs to write new commit objects, as the tree objects are of course already there	02:39
	← chris2 left	02:40
doener	hm, ok, it's not _that_ optimized. It still creates the index for each revision, as other filters might affect the tree	02:43
	← gwoo left	02:45
	→ gwoo_ joined	02:46
neunon	Almost at 700, heh.	02:49
	At least this operation is significantly less expensive than recloning the whole thing (which was an overnight operation)	02:50
	← gholias left	02:50
	gwoo_ → gwoo	02:51
	→ gholias joined	02:51
neunon	And good gravy, using git cvsimport takes forever. I've been cloning the Bochs CVS repo, and it's been going for -nearly- 24 hours now...	02:51
	→ SudoKing joined	02:54
	← SudoKing left	02:54
	→ Yamakiri joined	02:54
	→ SudoKing joined	02:55
	← SudoKing left	02:55
	← lresende left	02:56
	← orlandov_ left	02:59
	→ reprore joined	03:00
	← parasti left	03:01
	→ markelikalderon joined	03:01
	→ orlandov joined	03:02
doener	hm, reading the index is pretty cheap though... all the objects exist	03:06
	→ reprore__ joined	03:06
	← reprore left	03:07
doener	a quick hack to avoid the index update saves just 4 seconds out of 1m26s	03:07
cpg	hi, all! is it possible to clone only ONE directory of a repo?	03:09
	→ akitada joined	03:09
cpg	if so how?	03:09
	(git noob here)	03:10
doener	no, at least not yet	03:10
	there was some talk about sparse/narrow/name-of-the-day clones, but I have no idea about its status	03:10
cpg	doener: talking to me?	03:11
	the thing is that our repo is somewhat big (50MB) and people are interested in an area of it only	03:11
cehteh	i recently tried such manually with some git wizardy which worked, but certainly not for noobs	03:11
	split it into submodules	03:11
	man git-submodule	03:11
Gitbot	cehteh: the git-submodule manpage can be found at http://git.or.cz/man/git-submodule	03:11
	← aziz left	03:12
	← Yamakiri left	03:12
doener	cpg: 50MB packed?	03:12
cpg	du of the repo is 90MB	03:13
	← markelikalderon left	03:13
cpg	no idea if packed or not (git noob)	03:13
cehteh	du -sh .git/objects	03:13
doener	for the bare repo? (or the .git dir in a non-bare one)	03:13
cehteh	git gc	03:13
	du -sh .git/objects again	03:13
doener	how did the repo come into existance? Import from another scm?	03:14
	← towski left	03:14
doener	s/existance/existence/	03:14
	doener will learn that some day...	03:14
cpg	the objects dir of the bare repo is 89MB	03:15
cehteh	what project is it? sounds relative big	03:15
cpg	it came from svn	03:15
	there is an area of the repo that has a small linux distro in it	03:15
cehteh	ok then	03:15
cpg	however, not all users will need that	03:15
doener	hm, so lots of binary stuff	03:15
cpg	in svn i can check out a directory	03:16
	i figured that would be possible in git	03:16
doener	that distro seems like a good candidate for a submodule	03:16
cehteh	well git submodule is prolly the way to go then	03:16
cpg	well, there are some binary things there	03:16
	but yeah	03:16
	← mvanholstyn left	03:16
	cpg learning about submodules	03:16
doener	cpg: if you have "a moment" of spare time, try "git repack -a -d -f --window=50"	03:16
	→ nothingHappens joined	03:17
doener	might take a while, but if you're lucky, it was just the auto gc during the git-svn import that created a crappy pack	03:17
	← Sigma left	03:17
cpg	the content is 50MB, so i would not expect it to shrink it that much	03:17
	← harinath left	03:18
kevlarman	btw how do people deal with binary files related to a project?	03:19
cehteh	first of all i try to avoid them	03:20
	if there are icons, then version the original svg for example, .fig or whatever	03:20
	→ cytrinox joined	03:20
kevlarman	doesn't work so well for textures	03:21
cehteh	surprisingly if you search for such solutions rarely binary data will be left and then its often something which changes infrequently	03:21
	yes	03:21
kevlarman	it does change very infrequently	03:21
cehteh	so then that should be no problem	03:21
	maybe submodule too or store them somewhere else and only version symlinks	03:22
kevlarman	git can version symlinks?	03:22
cpg	ok, so i cannot as some of these devs to just get a directory out of the repo?	03:22
cehteh	yes	03:22
kevlarman	how does that work with windows?	03:22
	← hobodave left	03:23
cehteh	i dont care :)	03:23
cpg	*ask	03:23
doener	cpg: not as of right now	03:23
kevlarman	cehteh: unfortunately i have to care	03:24
cpg	:( - ok, thanks	03:24
cehteh	kevlarman: well when they change infrequently then it should be not a problem or?	03:24
doener	cpg: btw, for linux-2.6.git my working tree is 370MB, but the .git dir (with almost 4 years of history!) is just 270MB	03:24
kevlarman	cehteh: yeah, i'm leaning towards leaving it as is	03:24
	i was just wondering how people dealt with it	03:25
cpg	thanks doener	03:25
	→ EmilMedve joined	03:25
doener	cpg: so I'd try the repack anyway, there's nothing to lose..	03:25
	← mithro_ left	03:26
	not-xjjk → xjjk	03:26
	→ kansan__ joined	03:27
	← whee left	03:28
cpg	ok	03:30
	let me ask this: can i permanently prune the large directly from the repo, including it's history?	03:31
	*its	03:31
kevlarman	cpg: yup	03:31
doener	yep, using filter-branch	03:31
kevlarman	when it comes to rewriting history, the question is generally not if you can, but how much effort it would take (the answer is generally not a lot)	03:32
qrush	i'm curious about this too	03:32
	is it possible to totally remove a file/directory from history?	03:33
cpg	i rather not reimport the whole (minus the large chunk) into the git repo	03:33
kevlarman	qrush: yes	03:34
	git filter-branch	03:34
	Beeet → Beket	03:34
qrush	i'll have to look into it, i've gotten a request for a tip about it	03:35
kevlarman	in fact the man page has an example of how to do just that	03:35
dsal	qrush: You could do a lot of tips about filter-branch.	03:35
	I'm quite happy with my memcached archaeology project. :)	03:35
cpg	doener: repack did not take that long - took it down to 73mb	03:35
	← cytrinox_ left	03:35
qrush	dsal: any you'd like to contribute would be appreciated. :) I need to google around and read up on it	03:37
dsal	It was way ad-hoc. It'd make an ugly post.	03:37
	(or series)	03:37
doener	cpg: ok, not that impressive.	03:38
dsal	qrush: You can see the effects here: http://groups.google.com/group/memcached/browse_thread/thread/a7b153bc087244b1	03:38
qrush	heh. yeah, perhaps if it's deep enough i could do a week's worth of stuff on it...i already want to do a week on hooks	03:38
	dsal: oh, it's you. lol. your github blog is pretty kickass.	03:39
	← kansan left	03:39
dsal	Well, that one was a few different things: Global author and committer fixups, individual author updates, and grafting an older, newly imported history below the sort of "snapshotty" thing.	03:39
	heh, thanks. :)	03:40
	← softdrink left	03:40
	← tvoet_ left	03:40
	→ eno__ joined	03:42
	← ceej left	03:43
	← eno left	03:43
	→ vbabiy joined	03:44
kansan__	why does it seem like i have to : git add filename	03:47
	when i change it in order to commit it	03:47
	when its already existing in repo?	03:47
Arrowmaster	because you do	03:47
kansan__	but that wouldnt seem to make sense	03:47
Arrowmaster	but it does	03:47
doener	kansan__: http://tomayko.com/writings/the-thing-about-git	03:48
context	kansan__: if you knew more about how git works	03:48
kansan__	can you help me to understand	03:48
qrush	kansan__: http://gitready.com/2009/01/18/the-staging-area.html	03:48
dsal	kansan__: add == "add the contents of this file" not "add the filename"	03:49
Arrowmaster	if you dont like the staging area you can use git commit -a	03:50
	← sgrimm left	03:50
	← gholias left	03:50
qrush	heh. i can't imagine someone using git productively and not using the staging area	03:51
	→ gholias joined	03:51
qrush	i do like how they've changed it to `git stage`	03:52
dsal	qrush: ``I don't like running git add'' is one of the surest signs of someone who hasn't used git for more than an hour or two.	03:52
qrush	or rather, aliased it	03:52
	dsal: haha, truth. also why i have a beginner section on gitready :)	03:52
dsal	magit's staging ui is pretty awesome.	03:53
Arrowmaster	dsal: or hasnt explored git add -i or git gui enough	03:53
dsal	Arrowmaster: I can't deal with add -i -- add -p rules. :)	03:53
	← emss left	03:54
kansan__	why is this treu: And because your commits are no longer bound directly to what’s in your working copy, you’re free to stage individual pieces on a file-by-file, hunk-by-hunk basis.	03:56
	why isnt the commit bound to wc?	03:56
dsal	kansan__: Because the staging area is not the working copy.	03:56
	work -> stage -> commit	03:56
kansan__	its the diffs between working copy	03:57
	and what you have?	03:57
	changed?	03:57
dsal	No, it's whatever you've put in there.	03:57
	You can, for example, do some work, get something going, stage it, and then do further work.	03:57
	Later, you can decide you want to go back to what you staged, or commit what you staged and the further work in separate commits.	03:57
kansan__	but i thought the stage is auto generated	03:58
	by what you do?	03:58
doener	no	03:58
	→ jrmuizel joined	03:58
dsal	The easiest way to understand it is to open a file and make two changes (not next to each other), run add -p, say yes to one andno to the other, and then see what happens.	03:58
kansan__	or is that the explicit add command that i'm doing	03:58
doener	that's exactly the point	03:58
kansan__	to add it to staging	03:58
doener	you can control what gets staged	03:58
kansan__	ok	03:58
doener	echo 123 >> foo; git add foo; echo 456 >> foo; git commit	03:58
	that would only commit the addition of the 123 line, not the 456 line	03:59
	→ f4hy joined	03:59
	dsal tends to do a commit -v, which would help illustrate the point	03:59
dsal	``git diff'' (no args) is the difference between the staging area and the working copy. ``git diff --staged'' is the difference between the staging area and HEAD	03:59
kansan__	doener, wow intersting	03:59
	i see	04:00
doener	--cached	04:00
	or was --staged added as an alias there as well?	04:00
	← dreiss left	04:00
Arrowmaster	and git diff HEAD would be the difference between the working copy and HEAD	04:00
doener	ah, yeah, has been added	04:00
Arrowmaster	the 3 different usages of git diff can be confusing	04:01
dsal	Heh, there's more than three. :)	04:01
Arrowmaster	3 main ones	04:01
dsal	I think the different meanings of checkout are even more confusing.	04:01
kansan__	ok i forked from github made some changes and commits	04:02
	and now i want to push changes to github	04:03
	i tried git push	04:03
	with no luck	04:03
doener	define "no luck"	04:03
	← bryanl left	04:03
Arrowmaster	did you use githubs fork feature?	04:03
	→ theColonel joined	04:04
kansan__	fatal: protocol error: expected sha/ref, got '*********'; You can't push to git://github.com/user/repo.git	04:04
	Use [email@hidden.address]	04:04
	Arrowmaster, yes	04:04
	i forked before i checked it out	04:04
doener	well, the error message told you what's up	04:04
	you can't push via git:// only through ssh	04:04
kansan__	i'm a little unsure of how to make that work via ssh	04:05
doener	just use the url in the shown format	04:05
	should be fine to just change the url in your .git/config	04:05
dsal	kansan__: git config remote.origin.url [email@hidden.address]	04:07
	dsal messed that one up...	04:07
doener	s!/!:!	04:07
dsal	kansan__: git config remote.origin.url [email@hidden.address]	04:07
kansan__	yes	04:07
	ok that worked	04:07
	had to add my ssh key	04:08
	but done	04:08
	thanks!	04:08
	← cedricv left	04:09
	→ cedricv joined	04:09
kansan__	what does git diff --cached do that git diff doesnt	04:09
	← greyface left	04:09
	→ mithro joined	04:10
dsal	kansan__: --cached == --staged. It shows you what you've got ready to commit.	04:10
	The staging area has had many names in the past.	04:10
	dsal can sympathize with giving things confusing names	04:10
kansan__	ah ha	04:10
	i am liking this staging thing	04:10
Arrowmaster	stage and index seem to be the common ones	04:10
kansan__	is thatthe current name	04:10
	the idea of cherry picking particular chunks	04:11
dsal	kansan__: That's the name that is being pushed because it describes what it does.	04:11
kansan__	hunks is cool	04:11
	← mithro left	04:11
dsal	kansan__: That was one of my favorite features from darcs. :)	04:11
	→ mithro_ joined	04:11
kansan__	darcs?	04:11
	another vcs?	04:11
Arrowmaster	yes	04:12
RandalSchwartz	"join the darcs side"	04:13
	RandalSchwartz grins	04:13
RandalSchwartz	"git with the program!"	04:13
tango_	I'm going to develop a new vcs and I'll cal it sucs	04:13
	← schacon left	04:13
tango_	Super dUper Control System	04:14
RandalSchwartz	super universal control system	04:14
tango_	sucs : unmatched control system	04:14
dsal	simple, unobtrusive code storage	04:14
	reinvent 9fs	04:14
tango_	9fs the filesystem from plan9?	04:14
dsal	tango_: yeah, append-only	04:15
tango_	what	04:15
	what happens when you run out of disk space?	04:15
Tv	tango_: buy a new storage system, copy things over	04:16
dsal	tango_: They would add more WORM drives.	04:17
	→ amystrat joined	04:17
	→ kek joined	04:18
amystrat	how can I tell if I need to pull before I push? ie, has remote changed?	04:18
tango_	UTF-8 was invented by Ken Thompson to be used as the native encoding in Plan 9 and the whole system was converted to general use in 1992.[6] Note that Plan 9 only supports the codes defined in the Basic Multilingual Plane of Unicode.	04:18
amystrat	trying to make a bash prompt that indicates this	04:18
tango_	wpah	04:18
qrush	amystrat: git log origin..HEAD	04:18
	tango_: plan 9 ftw :P	04:18
tango_	qrush, I'm quite amazed	04:18
dsal	amystrat: push will fail.	04:18
tango_	/proc originated there too ...	04:19
qrush	tango_: my roommate hacks on the kernel, it's hardcore	04:19
dsal	tango_: read the man page for bind. I really miss that from...everything else.	04:19
	← Zimm left	04:19
amystrat	dsal: I know, but I am trying to change my bash prompt to indicate that	04:19
Arrowmaster	but you wouldnt know you needed to pull until youve either tried to push or already pulled	04:19
dsal	amystrat: You'd have to be fetching constantly.	04:20
qrush	amystrat: I just did a post on some bash prompt stuff with git. perhaps you could modify the existing ones? http://gitready.com/2009/01/23/status-in-your-prompt.html	04:20
amystrat	qrush, I just read that! :)	04:20
qrush	:D thanks	04:20
amystrat	it inspired my to try to change my bash prompt	04:20
dsal	I should try out bash someday.	04:21
qrush	I actually use fish normally, but all of this prompt stuff might get me to switch back	04:21
dsal	I used vi/vim for something like 15 years and just recently switched to emacs.	04:21
	← f4hy left	04:21
tango_	I just come from touching a shell that doesn't have 'for'	04:21
	it hurt	04:21
	dsal, YOU SWITCHED TO EMACS?	04:21
	I don't want to touch plan9 anymore	04:22
	;-)	04:22
	← spearce left	04:22
dsal	tango_: Yeah. Friend showed me magit and I figured I'd give it a go. magit is awesome.	04:22
	tango_ has no idea what magit is	04:22
dsal	I'm actually in irc inside of emacs right now -- better client than my previous one.	04:22
tango_	it's funny	04:23
dsal	emacs has vc-mode which is a common interface across all version control systems. You get a consistent view of the world, but can't do too much. magit is more of a git-centric way of doing things.	04:23
Tv	and git's contrib includes git-mode	04:24
tango_	in the auld days in which I used windows I started looking for an open source, portable editor (since I was starting to consider switching out of windows AND i needed solid tools)	04:24
	I tried emacs and vim alternating between the two for quite a while, and I must say I hated vim for being so very unfriendly with its hjkl stuff	04:24
	otoh emacs wasn't really much better keystroke-wise, nor with learning curves, and in the end the vim keystrokes turned out to be way more resting for my wrists, so I switched to vim	04:25
	I do miss some of the emacs goodies though, but I'm used to GUI stuff for IRC and mail, and I do all my vcs from the command line anyway	04:25
	which considering that a friend of mine doesn't want to use git because it has no netbeans interfaces, is a VERY good thing	04:26
	← reprore__ left	04:26
	→ pjhyett joined	04:30
	→ reprore joined	04:31
	→ sgrimm joined	04:32
	← mithro_ left	04:32
dsal	tango_: I'm running some kind of OS X emacs. It's GUI enough for me.	04:32
	← amystrat left	04:33
dsal	The eclipse git stuff seems OK, but I still do most of it from the CLI.	04:33
tango_	this colleague of mine uses netbeans, sadly	04:33
	so we're actually forced to use mercurial	04:33
	or we would be if actually cared about revision control	04:34
	→ DigitalMan joined	04:34
	DigitalMan → DM22	04:34
dsal	I could never understand netbeans.	04:35
DM22	anyone around	04:35
tango_	I have great hopes for the new 'foreign vcs' stuff in git	04:35
dsal	I used it for my sunspots because it's required (annoyed by vendors requiring IDEs).	04:35
DM22	Can someone help me with this:	04:35
	fatal: cannot handle shell internally	04:35
dsal	DM22: context/	04:35
DM22	I get it when I try to push	04:35
	→ archis joined	04:36
DM22	I'm a bit of a n00b at this too but it seems like every time I want to work on my code or push/pull something always goes wrong.	04:36
tango_	working on your code should not be a problem	04:36
	push, depends on your setup	04:36
DM22	well	04:36
	one minute it works fine	04:36
	and then I'll make a change,	04:36
	← nothingHappens left	04:36
DM22	git add file.c, git commit -m 'msg here', git push	04:37
	and I always get some type of new error	04:37
	it's always something different and it's really starting to piss me off :\	04:37
	I've been working on trying to fix git alone for 2 days now :/	04:37
tango_	what is your remote anyway?	04:37
	← reprore left	04:38
dsal	You wouldn't by any chance have been pushing to github, eh?	04:38
DM22	like my clone address? </newb>	04:38
	well heres the thing, as far as I know everything is setup right because it works sometimes and other times doesn't	04:38
	and I don't change any git settings just source code	04:39
dsal	DM22: Pushing where?	04:39
DM22	to git?	04:39
tango_	to what git	04:39
DM22	[email@hidden.address] ?	04:39
	→ Owner_ joined	04:40
tango_	and the error you get is about a missing shell?	04:40
	odd	04:40
dsal	DM22: Yeah, github had a minor outage that lasted about 10 minutes.	04:40
DM22	it is odd	04:40
dsal	affected push	04:40
DM22	because I made a successful commit a couple minutes ago	04:40
	← drewr left	04:40
DM22	oh your right	04:41
	it was an outage	04:41
	just went through :D	04:41
tango_	something happened on gitorious instead	04:41
dsal		04:41
	<qrush> dsal: oh, it's you. lol. your github blog is pretty kickass. [19:39]	04:41
	*** kansan (n=kansan__@97-113-97-64.tukw.qwest.net) has quit: Read error: 113	04:41
	(No route to host)	04:41
	<dsal> Well, that one was a few different things: Global author and committer	04:41
	fixups, individual author updates, and grafting an older, newly	04:42
	imported history below the sort of "snapshotty" thing.	04:42
	<dsal> heh, thanks. :) [19:40]	04:42
	*** softdrink (n=softdrin@c-67-186-205-161.hsd1.ut.comcast.net) has quit:	04:42
	*** tvoet_ (n=tvoet@modemcable085.233-37-24.mc.videotron.ca) has quit:	04:42
	"Ex-Chat"	04:42
	*** eno__ (n=eno@adsl-70-137-177-148.dsl.snfc21.sbcglobal.net) has joined	04:42
	channel #git [19:42]	04:42
	*** ceej (n=ceej@cpe-76-94-228-101.socal.res.rr.com) has quit: [19:43]	04:42
	*** eno (n=eno@nslu2-linux/eno) has quit: Read error: 104 (Connection reset by	04:42
	peer)	04:42
	*** vbabiy (n=vbabiy@pool-71-244-118-231.albyny.fios.verizon.net) has joined	04:42
	channel #git [19:44]	04:42
	<kansan__> why does it seem like i have to : git add filename [19:47]	04:42
	<kansan__> when i change it in order to commit it	04:42
	<kansan__> when its already existing in repo?	04:42
tango_	ehrm	04:42
dsal	<Arrowmaster> because you do	04:42
	<kansan__> but that wouldnt seem to make sense	04:42
tango_	dsal, that was absolutely uncalled for	04:42
dsal	<Arrowmaster> but it does	04:42
	← jrmuizel left	04:42
dsal	<doener> kansan__: http://tomayko.com/writings/the-thing-about-git [19:48]	04:42
	<context> kansan__: if you knew more about how git works	04:43
tango_	use pastebin for more than 3 lines of paste	04:43
dsal	<kansan__> can you help me to understand	04:43
	<qrush> kansan__: http://gitready.com/2009/01/18/the-staging-area.html	04:43
tango_	somebody please stop him	04:43
DM22	.	04:43
dsal	<dsal> kansan__: add == "add the contents of this file" not "add the filename"	04:43
	[19:49]	04:43
	<Arrowmaster> if you dont like the staging area you can use git commit -a	04:43
	[19:50]	04:43
	*** sgrimm (n=koreth@outbound285a.pasd.tfbnw.net) has quit: Read error: 110	04:43
DM22	/ignore ftw	04:43
dsal	(Connection timed out)	04:43
DM22	lol	04:43
	← kukks left	04:43
dsal	*** gholias (n=gholias@cpe-069-134-008-128.nc.res.rr.com) has quit: Read	04:43
	error: 104 (Connection reset by peer)	04:43
	<qrush> heh. i can't imagine someone using git productively and not using the	04:43
	staging area [19:51]	04:43
	*** gholias (n=gholias@cpe-069-134-008-128.nc.res.rr.com) has joined channel	04:43
	#git	04:43
DM22	lawd.	04:43
dsal	<qrush> i do like how they've changed it to `git stage` [19:52]	04:43
	<dsal> qrush: ``I don't like running git add'' is one of the surest signs of	04:43
	someone who hasn't used git for more than an hour or two.	04:43
qrush	dsal: oh hai	04:43
dsal	<qrush> or rather, aliased it	04:43
	<qrush> dsal: haha, truth. also why i have a beginner section on gitready :)	04:43
	<dsal> magit's staging ui is pretty awesome.	04:44
	<Arrowmaster> dsal: or hasnt explored git add -i or git gui enough [19:53]	04:44
	<dsal> Arrowmaster: I can't deal with add -i -- add -p rules. :)	04:44
	ERC> /ping dsal [19:54]	04:44
	*** Ping time to dsal is 00:00.01	04:44
	*** emss (i=emss@d118-75-37-96.nap.wideopenwest.com) has quit: Connection	04:44
DM22	dude	04:44
	srsly	04:44
	stop	04:44
dsal	timed out	04:44
	ERC> /whoami [19:55]	04:44
qrush	this is why all irc clients need paste protection	04:44
dsal	*** dsal is unknown (n=user@adsl-69-230-8-158.dsl.pltn13.pacbell.net)	04:44
	*** dsal is on channel(s): #buildbot #memcached #github #git	04:44
tango_	DM22, I doubt he can stop, he just copypasted	04:44
dsal	*** dsal is/was on server kubrick.freenode.net (Los Angeles, CA, US)	04:44
	*** dsal is an identified user	04:44
tango_	well, he's an emacs user, what can you guys expect	04:44
dsal	*** dsal is an identified user	04:44
	*** dsal has been idle for 00:00.55, on since 13:47:41 2009/01/23	04:44
DM22	I know how he can stop	04:44
dsal	ERC> /help [19:56]	04:44
	<kansan__> why is this treu: And because your commits are no longer bound	04:44
DM22	/quit <-	04:44
	:P	04:44
dsal	directly to what’s in your working copy, you’re free to stage	04:44
	individual pieces on a file-by-file, hunk-by-hunk basis.	04:44
	<kansan__> why isnt the commit bound to wc?	04:44
	<dsal> kansan__: Because the staging area is not the working copy.	04:44
tango_	maybe he'll actually manage to paste his password even and we can ghost him	04:44
dsal	<dsal> work -> stage -> commit	04:45
	<kansan__> its the diffs between working copy [19:57]	04:45
	<kansan__> and what you have?	04:45
	<kansan__> changed?	04:45
	<dsal> No, it's whatever you've put in there.	04:45
	<dsal> You can, for example, do some work, get something going, stage it, and	04:45
ia	hello. if i want to change working dir state at some previous commit, then i should use "git revert <commit id>", right?	04:45
nick125	DM22: Some clients are retarded and won't send the QUIT command until the buffer is cleared out	04:45
dsal	then do further work.	04:45
	<dsal> Later, you can decide you want to go back to what you staged, or commit	04:45
	what you staged and the further work in separate commits.	04:45
	<kansan__> but i thought the stage is auto generated [19:58]	04:45
	<kansan__> by what you do?	04:45
	<doener> no	04:45
	*** jrmuizel	04:45
tango_	ia, no, that actually creates a commit that nullifies the commit you're reverting	04:45
dsal	(n=jrmuizel@CPE001f5be79d0f-CM0017ee62f8b0.cpe.net.cable.rogers.com) has	04:45
RandalSchwartz	eww. what is all this paste?	04:45
DM22	23:45:20 up 82 days, 12:39, 2 users, load average: 0.00, 0.00, 0.00 <- Win. :)	04:45
dsal	joined channel #git	04:45
	<dsal> The easiest way to understand it is to open a file and make two changes	04:45
tango_	ia, git checkout is your friend	04:45
dsal	(not next to each other), run add -p, say yes to one andno to the	04:45
	other, and then see what happens.	04:45
	<kansan__> or is that the explicit add command that i'm doing	04:45
	<doener> that's exactly the point	04:45
	<kansan__> to add it to staging	04:46
tango_	we need an op in this channel	04:46
dsal	<doener> you can control what gets staged	04:46
ia	tango_: oh, right.	04:46
dsal	<kansan__> ok	04:46
	<doener> echo 123 >> foo; git add foo; echo 456 >> foo; git commit	04:46
ia	tango_: thanks	04:46
	RandalSchwartz tries to wipe the paste from his screen	04:46
DM22	even better,	04:46
dsal	<doener> that would only commit the addition of the 123 line, not the 456 line	04:46
	[19:59]	04:46
tango_	ia, np	04:46
DM22	an oper	04:46
dsal	* dsal tends to do a commit -v, which would help illustrate the point	04:46
	*** f4hy (n=f4hy@ip68-111-237-5.sd.sd.cox.net) has joined channel #git	04:46
	<dsal> ``git diff'' (no args) is the difference between the staging area and	04:46
	the working copy. ``git diff --staged'' is the difference between the	04:46
tango_	let's see if there's freenode stuff around	04:46
dsal	staging area and HEAD	04:46
RandalSchwartz	dsal - wake up	04:46
	leave now	04:46
dsal	<kansan__> doener, wow intersting	04:46
	<kansan__> i see [20:00]	04:46
RandalSchwartz	before the big guns come out	04:46
dsal	<doener> --cached	04:46
	<doener> or was --staged added as an alias there as well?	04:46
	*** dreiss (n=dreiss@64.201.253.161) has quit: Read error: 113 (No route to	04:46
	host)	04:46
	<Arrowmaster> and git diff HEAD would be the difference between the working	04:46
	← dsal left	04:46
DM22	THANK GOD	04:47
	:P	04:47
RandalSchwartz	i imagine it was a bad x11 paste	04:47
	when i used x11, i was always doing that	04:47
	it's a lot harder in osx	04:47
	→ dsal joined	04:47
RandalSchwartz	dsal - bad x11 paste?	04:47
DM22	>:O	04:47
tango_	dsal, you just flooeded the channel with hundred of lines of paste	04:47
dsal	RandalSchwartz: Ugh, something horrible happened in erc. Sorry. :(	04:47
	← archis left	04:47
tango_	dsal, so tell me again about the wonders of using emacs as an irc client	04:47
	RandalSchwartz wipes the paste from his creen	04:48
RandalSchwartz	tango_ - I don't seem to have problems	04:48
	but i'm not using erc	04:48
	I'm using irc.el	04:48
	i still can't use erc. it's too different	04:48
tango_	so there's actually two irc clients for emacs?	04:48
	ehe	04:48
RandalSchwartz	there's about five	04:48
	tango_ refrains from the typical joke	04:48
dsal	My problem is that I still try to hit command-v to paste sometimes. And when I do, it kind of freaks out, and apparently copies stuff randomly.	04:48
DM22	thanks for your help guys, cya	04:48
	dsal was trying to paste a URL to DM22	04:49
	← DM22 left	04:49
nick125	Wow, emacs has five IRC clients..too bad it just doesn't have a decent editor yet ;-)	04:49
tango_	nick125, that's the joke I didn't want to make 8-D	04:49
nick125	I figured someone had to :D	04:49
tango_	I'll be happy if they ever implement vim in emacs though	04:49
	actually the only thing I want is vim in bash	04:50
	as in being able to navigate the shell with vim keys	04:50
	the vi mode doesn't work very well	04:50
	← simplechat left	04:50
RandalSchwartz	go native	04:50
	emacs is best used when you don't try to work around it :)	04:50
tango_	too much strainn on my wirst	04:51
	nick125 doesn't really like vim or emacs	04:51
tango_	nick125, you use ed?	04:51
nick125	nano :D	04:51
	→ careym joined	04:52
nick125	I've used vim and emacs, but they're just overcomplicated in my opinion.	04:52
	eno__ → eno	04:53
careym	hi all, I have two seperate local repos and want to merge them into a single history any pointers to instructions? the ones at https://kerneltrap.org/mailarchive/git/2007/12/9/484613/thread in Alex Risens email fail when I try to fetch from the remote project	04:54
	→ yacc_ joined	04:54
careym	wait maybee git-stitch-repo is what I am after .. reading	04:56
dsal	careym: You just need to add a graft and do a filter-branch	04:56
careym	dsal: thanks any pointers on where I can find out about graft and filter-branch	04:57
dsal	careym: git help filter-branch should tell you all you need to know.	04:58
	careym: That's one of the easiest things to do with it.	04:58
	Also note that you can just log after doing the graft and see if it's what you want.	04:58
	→ jstad joined	04:59
	→ rufnut joined	05:00
	← Leefmc left	05:03
	→ khaije1 joined	05:07
	← dsal left	05:07
	→ bieneff joined	05:07
	→ bdimcheff joined	05:10
	← yacc__ left	05:12
	← lorandi left	05:23
	← batok left	05:24
	← bieneff left	05:25
	→ tjafk2 joined	05:27
	← tjafk1 left	05:27
	← cthompson left	05:29
	← Owner_ left	05:30
	→ dreiss joined	05:31
	← gholias left	05:32
	→ zapada joined	05:33
	← theColonel left	05:35
	→ gholias joined	05:37
	← pjhyett left	05:41
	← scientes left	05:47
	→ joshua__ joined	05:48
	← jm left	05:48
scrooloose	im in a directory and i wanna know who made the last change to every file in that dir, someone point me in the right direction?	05:48
joshua__	Hi folks; I have a file that is akin to a binary that changes with some frequency (it's a hex file that's generated, and needs to be in version control for people who don't have a toolchain to build the hex file). Is there a way to force git to treat it like a binary?	05:49
careym	dsal: thanks ended up doing a pair of `git-fast-export --all --date-order` importing the first file, adding a branch repo1, checking out branch repo1, deleting master, importing the second file, merging the two branches	05:49
	and the creating a new master	05:49
joshua__	Ah, .gitattributes did what I wanted, if anyone else is listening to that has this problem	05:53
	← joshua__ left	05:53
	← khaije1 left	05:55
qrush	scrooloose: just need to bend git log to your will	05:56
	→ synsol joined	05:58
qrush	scrooloose: if you're just looking for one author, you could use --author	05:58
	so	05:58
	git log --author=scroolose	05:58
	← curvature left	06:00
	→ curvature joined	06:00
scrooloose	qrush: yes, i can use that to get the commits by me, but its not quite what im after	06:00
	← rufnut left	06:02
qrush	could use bash or pipes to go over each file then use git log to pull the committer/email out	06:02
	i'm not a sed/awk guru, sorry :/	06:03
	→ thumper joined	06:03
	→ chuck joined	06:04
chuck	what are some good code review systems that work with git?	06:04
	(preferrably something written with an easy to install language like PHP)	06:05
tango_	I use gitweb + irc	06:06
chuck	well i'd rather have something where you can actually mark statuses on certain commits and post comments, etc	06:07
	→ GeertB joined	06:07
	→ markelikalderon joined	06:07
cehteh	chuck: there are some .. forgot the names but they are listed in the tools section on the git website	06:07
	→ joshdavey joined	06:08
chuck	cehteh: which site? git.or.cz or git-scm.com	06:08
	oh wait they're the same now	06:08
	nevermind	06:08
	chuck thought they were different .-.	06:08
chuck	cehteh: what section would they be under?	06:10
	→ cantropy joined	06:10
chuck	oh sweet, there's a version of rietveld for git	06:11
	← vbabiy left	06:12
cehteh	http://git.or.cz/gitwiki/InterfacesFrontendsAndTools#head-43a31dc491d3e4c38b9ff48bedf36c1a25c0495e	06:12
	ok found your way	06:12
	← cantropy left	06:17
	← jstad left	06:18
	→ muthu_ joined	06:25
	→ drobbins_ joined	06:26
	← bdimcheff left	06:26
	drobbins_ → drobbins	06:27
	← plasticine left	06:30
	← Circuitsoft left	06:31
	→ Circuitsoft_ joined	06:31
	Circuitsoft_ → Circuitsoft	06:31
	→ ekontsev1y joined	06:31
	← ekontsevoy left	06:33
	→ theColonel joined	06:34
chuck	if I were to write a code review web application for git, what would be the best way of being able to use it with a remote repository?	06:37
	chuck isn't quite sure how to phrase that	06:37
RandalSchwartz	remote repos are generally for fetching or cloning	06:38
	you can't really do anything else	06:38
	so concentrate on local repo	06:38
chuck	with SVN, you can interact with a remote repository without checking it out, but i'm not sure how I could do something like that in git .-.	06:38
	RandalSchwartz: what method do you think I should use, then?	06:39
	→ davidfetter joined	06:39
RandalSchwartz	method of what?	06:40
	you need the objects locally	06:40
chuck	yeah, that's what I was thinking, but how... would I seriously have to make the site run git pull origin master periodically?	06:40
offby1	chuck: with svn, you can do 'svn ls' on a remote repository; with git, you could in theory do "ssh server git ls-tree"	06:40
	probably not worthwhile though	06:41
RandalSchwartz	git fetch is a fast op	06:41
	especially if it's all up to date	06:41
	so just fetch from time to time	06:41
	before you would be running a report	06:41
	i don't see how that would be hard	06:41
chuck	RandalSchwartz: how is git fetch different from git pull?	06:42
RandalSchwartz	and remember you don't want pull	06:42
	you want fetch	06:42
	pull = fetch + merge	06:42
	you don't want to be merging	06:42
chuck	so how would fetch work? git fetch origin master?	06:42
RandalSchwartz	just 'git fetch'	06:42
	defaults to origin	06:42
	and then defaults that to all public branches at remote	06:43
	exactly what you want	06:43
	at least, by default	06:43
	you can tweak nearly anything :)	06:43
chuck	is git fetch supposed to not say anything	06:44
RandalSchwartz	so essentially, "git fetch" makes local repo have all objects that remote "origin" repo has	06:44
	it's quiet if you're up to date	06:44
	chuck wonders if he has any repos checked out that aren't up to date	06:44
chuck	wow, is it really that hard to find a repo that has had updates :P	06:45
RandalSchwartz	git checkout origin/master; git reset --hard HEAD^3 :)	06:46
offby1	if you, like me, forget where you put your repository:	06:46
	command locate --basename '\.git' \| sort \| uniq	06:46
	offby1 nods sagely	06:46
chuck	okay i think it worked perfectly	06:46
	now to see if ruby git supports it	06:47
	yep, it does	06:47
	→ priidu joined	06:48
chuck	RandalSchwartz: thanks for the tip, i think it'll work perfectly	06:48
	RandalSchwartz bows	06:48
	RandalSchwartz hits head on desk	06:48
chuck	git fetch runs really quickly when there's nothing to update, so i think that'll work	06:48
RandalSchwartz	ouch!	06:48
chuck	lol	06:48
	offby1: mac os x's locate does not have --basename :<	06:50
	would piping it through basename work?	06:50
RandalSchwartz	because osx is sane	06:50
	and not using gnu-hacked tools	06:50
offby1	chuck: doubt it	06:50
RandalSchwartz	as in, "cat" doesn't have 18 options :)	06:50
offby1	my cat has 18 toes	06:50
chuck	lol	06:50
	hmm	06:51
	now, the big decision	06:51
	← markelikalderon left	06:51
chuck	do i make this code review system as a mediawiki extension (integrated with our authentication, yet a really annoying bitch to program), or do I write it using ruby and sinatra? (good code, yet not integrated with authentication)	06:52
	hmm...	06:52
RandalSchwartz	or write it in smalltalk, instead of ruby, since ruby is just a bastardized slow smalltalk. :)	06:52
	all the syntax of perl with all the speed of lisp! :)	06:53
	it's not surprising that the fastest ruby today is translating ruby to smalltalk bytecodes, then running it on the GemStone/S smalltalk vm	06:54
	aka "maglev"	06:54
	chuck doesn't like smalltalk already	06:55
chuck	their site is ugly, and I can't find any examples of the code easily	06:55
RandalSchwartz	"their site"?	06:55
	which site are you on?	06:55
chuck	smalltalk.org	06:56
RandalSchwartz	wikipedia "smalltalk" would be a good start	06:56
	oh. smalltalk.org is owned by some crazy smalltalk consultant	06:56
	he's not very cooperative	06:56
	doesn't even come to the conferences any more	06:56
	wants a huge pile of money to sell it though	06:57
offby1	squeak makes me wince	06:57
	chuck does not like smalltalk's syntax xD	06:57
RandalSchwartz	the point of smalltalk's syntax is that it's very simple	06:57
	all the power is in how it is used	06:57
	i can teach you entire smalltalk syntax in 20 minutes	06:57
	takes a week or two with perl	06:58
	or ruby	06:58
chuck	really?	06:58
RandalSchwartz	yes	06:58
	chuck thinks ruby is easier than that to learn :P	06:58
chuck	though, there are a lot of weirdnesses that it has	06:58
nick125	A week with Ruby? does that include all of the stupid quirks and nasty shortcuts?	06:58
RandalSchwartz	oh. thought you were challenging smalltalk in 20 minutesw. :)	06:59
	"this is a comment"	06:59
	anObject unaryMessage	07:00
	anObject + anotherObject "binary message"	07:00
	anObject with: another and: aThird "keyword message with:and:"	07:00
	→ ekontsevoy joined	07:00
offby1	and he's off!	07:00
RandalSchwartz	foo := anObject "assignment"	07:00
chuck	19:30 left	07:00
RandalSchwartz	foo := [anObject aMessage] "block aka lambda"	07:01
	result := foo value "invoke block"	07:01
	← yann left	07:01
RandalSchwartz	that's 70% of it therre	07:01
	just a few more subtleties about "." between statemtns, and how to declare locals and args	07:02
	foo := [\| local1 local2 \| anObject thisMessage. anObject thatMessage]	07:02
	foo := [:bar \| bar thisMessage. bar thatMessage] "argument is bar"	07:03
	→ drobbins_ joined	07:03
RandalSchwartz	result := foo value: anObject "pass anObject into bar"	07:03
neunon	doener: Heh, remember that filter-tree you had me do? It's still going. _	07:03
Ilari	neunon: If it is tree filter and the tree is large, its going to take a while...	07:04
neunon	I blame the PowerPC G4 ;)	07:04
Ilari	neunon: Is it even CPU bound or disk I/O bound?	07:05
	← ekontsevoy left	07:05
neunon	I'm pretty sure the CPU is the limiting factor here. CPU usage is staying at 100%.	07:05
	← libc left	07:05
Ilari	neunon: What kind of filter it is running?	07:06
neunon	"git filter-branch --subdirectory-filter trunk -- --all"	07:06
	It's on commit 13K of 16K or so	07:06
	I should have run it on my Core 2 machine instead, heh	07:07
	theColonel → thechuck	07:07
	thechuck → canuck	07:07
	canuck → colonel	07:08
Ilari	neunon: Ah. Subdirectory filter... That should be (relatively) fairly fast (when compared with really slow things like tree filter)...	07:08
	colonel → Guest36735	07:08
	← neoeinstein left	07:09
	← drobbins left	07:09
neunon	Yeah, the G4 really takes its time on things. Sad thing is, it's my git server. It's particularly bad when it's compressing objects for a clone and whatnot	07:09
	→ NSaibot joined	07:11
chuck	neunon: I didn't know people still used ppc G4 :P	07:11
	← Guest36735 left	07:11
	chuck thought he was the only one that was still on powerpc ;-)	07:11
	← ekontsev1y left	07:11
neunon	I got it just recently so I could have a PowerPC mac to play with	07:11
	particularly for cross-development	07:12
chuck	oh :P	07:12
	← davidfetter left	07:12
neunon	turns out that I could have just built apps for PowerPC and tested them under rosetta...	07:12
	shrug	07:12
	→ libc joined	07:14
	← madewokherd left	07:15
	→ drobbins joined	07:16
	→ davidfetter joined	07:20
	← NSaibot left	07:21
	→ NSaibot joined	07:21
	← bobesponja left	07:22
	→ eno__ joined	07:22
	← orlandov left	07:24
	→ ahupp\|fb joined	07:28
	← Niamor left	07:29
	→ Niamor joined	07:29
	← jim_c_ left	07:32
	← Whtiger left	07:33
	← drobbins_ left	07:34
	← nick_h[litage] left	07:34
	← eno left	07:34
	→ joseferr joined	07:38
	← ahupp\|fb left	07:39
	→ gaveen joined	07:42
	→ Whtiger joined	07:44
	← cygnus__ left	07:45
	→ chalky joined	07:51
	→ orlandov joined	08:00
	← Flathead left	08:02
	← pigeon left	08:02
	→ heren joined	08:14
	→ benjo2 joined	08:27
	benjo2 → benmcw	08:28
	benmcw → benw	08:28
	→ NSaibot_ joined	08:28
	benw → bennyw	08:28
	→ plasticine joined	08:29
	→ drobbins_ joined	08:31
	← drobbins_ left	08:32
	← drobbins left	08:32
	→ orlandov_ joined	08:36
	← orlandov left	08:37
	→ ciskje joined	08:39
	→ peritus_ joined	08:41
	← NSaibot left	08:41
	← peritus_ left	08:41
	← zpoley left	08:56
	→ alley_cat joined	08:57
	→ zpoley joined	08:58
	→ peritus_ joined	08:59
	← ashleyw left	09:01
	→ tvw joined	09:03
	→ hobodave joined	09:06
careym	#join nouveau	09:11
	→ ashleyw joined	09:11
careym	bye	09:11
	← careym left	09:12
	← k0001 left	09:12
	curvature → curvature_	09:19
	curvature_ → curvature	09:20
	← stephenlb left	09:20
	→ Fullmoon joined	09:22
	→ solydzajs joined	09:23
	← Beket left	09:25
	→ SRabbelier joined	09:26
	← joshdavey left	09:27
	← dreiss left	09:28
	← orlandov_ left	09:32
	→ ludde joined	09:36
	→ j416 joined	09:39
	→ rvhi joined	09:41
rvhi	hi, i accidentally modify a file before commit, how do I rollback?	09:42
j416	rvhi: reset --soft HEAD^	09:42
	will reset to the state before commit, but not modify your index or working dir	09:42
Arjen	Correct the file, git add $file, and git commit --amend	09:42
	→ deavid joined	09:43
j416	that's another way	09:43
Arjen	Saves typing in the commit message again	09:43
	→ pigeon joined	09:44
rvhi	my question was not clean	09:44
	→ ph^ joined	09:44
rvhi	i didn't do any commit yet	09:45
	just modify the file	09:45
	want to get the original one back	09:45
j416	rvhi: you want it back from your repo?	09:46
	or from the index?	09:46
	git checkout yourfile	09:46
	will restore yourfile from the head	09:46
rvhi	checkout did it :)	09:46
j416	great	09:47
rvhi	is it repo or index? Not sure the difference...	09:47
j416	git checkout yourfile gets the file as it looks in the last commit	09:48
rvhi	one more question, how do i let git ignore files .o and .so?	09:49
j416	add it to .gitignore	09:50
	→ codestr0m joined	09:50
j416	there are two other ways, you can have a global gitignore, and you can have a gitignore that isn't included when cloning the repo	09:51
	read all about it in man gitignore	09:51
Gitbot	j416: the gitignore manpage can be found at http://git.or.cz/man/gitignore	09:51
j416	read all about it in man gitignore > rvhi	09:51
Gitbot	j416: the gitignore manpage can be found at http://git.or.cz/man/gitignore	09:51
j416	hm, what's the syntax for Gitbot o_O	09:51
rvhi	.gitignore works perfect!	09:54
	.gitignore won't ignore modified files, is that normal?	09:58
j416	it will not affect already tracked files.	09:59
rvhi	is there a way?	10:00
neunon	any idea why git svn rebase would simply hang?	10:01
j416	rvhi: remove the files from the repo? :)	10:01
	← heren left	10:04
	← pigeon left	10:06
neunon	hrm. why the hell is git svn rebase hanging like this? D:	10:06
	→ pigeon joined	10:08
	→ kumbayo joined	10:10
codestr0m	anyone have any idea why git is growing in popularity?	10:10
	I mean. what makes it so successful?	10:10
	← codestr0m left	10:10
	← jepler left	10:12
	→ gouki_ joined	10:13
	→ masak joined	10:15
Pieter	it has a critical mass now	10:16
	→ ulf__ joined	10:17
	→ d0k joined	10:17
thiago_home	I was going to say that it's because it's good, fast and works, but yeah, that too :-)	10:17
vuf	unfortunatley, both "git" and "mercurial" are bad Google Trends words ...	10:26
	→ muthu joined	10:26
vuf	it seems that git is super popular in Turkey	10:27
	← solydzajs left	10:28
	← gouki left	10:28
	→ solydzajs joined	10:28
Pieter	yeah, if yu do a google blog search you get all kinds of turkish crap	10:29
masak	vuf: there seems to be a music album with that name in Turkey. http://tr.wikipedia.org/wiki/Git_(alb%C3%BCm)	10:29
	→ Voker57 joined	10:29
	vuf wonders what it means	10:31
	← jmspeex left	10:32
Aides	vuf: http://www.langtolang.com/?selectMenuLang=1&txtLang=git&selectFrom=Turkish&selectTo=English&submit=Search&selectMenuLang=1	10:34
vuf	thanks	10:36
	← j416 left	10:37
	← muthu_ left	10:37
	← ashleyw left	10:39
	→ Aikawa joined	10:40
	← joseferr left	10:41
	← rvhi left	10:46
	← alley_cat left	10:46
	→ alley_cat joined	10:49
	→ hingwah joined	10:50
	← gaveen left	10:51
	← glommer left	10:51
	→ gaveen joined	10:56
neunon	Right. Seems git svn rebase has gotten into an infinite loop. I've let it go for an hour and it's done nothing productive. D:	10:57
	→ glommer joined	10:57
	← hingwah left	10:58
	← gebi left	10:58
	→ mithro_ joined	11:00
drizzd	neunon: can you run git-rebase directly instead?	11:00
	→ fruit joined	11:00
neunon	I'm going to just rm -rf the repo and re-git svn clone	11:00
drizzd	neunon: git-svn rebase seems to do both fetch and rebase, so it could be a network problem.	11:01
neunon	well, it was using 100% CPU for an hour straight and my network monitor showed no activity. And I checked via browser, the SVN repo seems fine	11:02
drizzd	neunon: on which platform is this?	11:03
neunon	Mac OS X	11:03
drizzd	hmm, no idea then. But running fetch and rebase manually will narrow it down	11:04
fruit	eclipse git plugin users: does context menu commands work for you (commit, update, etc.)?	11:04
drizzd	fruit: yes	11:04
neunon	I suspect that the filter-branch I did earlier mucked something up, but shrug	11:04
	← shad left	11:04
fruit	drizzd, aha - tnx	11:05
drizzd	fruit: having trouble?	11:05
	neunon: that's likely, since you can't change svn history	11:06
	but still it shouldn't hang	11:06
	→ shad joined	11:06
fruit	drizzd, history and refresh work, but if i press something else, IDE freeze ("kill" helps ;>)	11:07
drizzd	hmm, I used eclipse JEE version 3.4.1 and jgit v0.4.0	11:07
neunon	drizzd: I agree, it shouldn't. The problem was that I cloned an SVN repo which didn't have the 'branches' or 'tags' folders, so it thought 'trunk' was simply a folder rather than a branch. had to use a filter-branch to move it. It was annoying.	11:07
drizzd	jgit/egit v0.4.0 rather	11:08
	neunon: weird, I don't know anything about that	11:08
neunon	Also, is it just me or is the PowerPC G4 -ridiculously- inefficient?	11:09
drizzd	fruit: oh and apparently you should install JDK 1.6	11:09
fruit	drizzd, ok... thanks	11:10
	→ hobodave_ joined	11:11
	→ not-xjjk joined	11:12
	→ charon joined	11:13
	mithro_ → mithro	11:14
	→ heren joined	11:16
	← priidu left	11:21
	← xjjk left	11:24
	→ yann joined	11:25
	→ CardinalNumber joined	11:25
	→ Sigma joined	11:27
	→ priidu joined	11:28
	→ protorom joined	11:30
	← yacc_ left	11:31
	← hobodave left	11:31
	← protorom left	11:33
	→ protorom joined	11:33
	→ hobodave joined	11:33
	→ mbroeker joined	11:33
	← Niamor left	11:34
	→ Niamor joined	11:34
	← protorom left	11:35
	→ protorom joined	11:35
	← protorom left	11:37
	→ Acry joined	11:37
	← Acry left	11:38
	→ protorom joined	11:38
	→ Ariens joined	11:38
	→ greyface joined	11:40
	→ name joined	11:41
	← ProperNoun left	11:42
	→ TheGeralt joined	11:42
	→ \|Voker57\| joined	11:43
	→ bentob0x joined	11:43
TheGeralt	Hi, is it possible that you can't use "git stash" in 1.6.0.6 to save your work in progress? I have to use "git stash save" and the manual http://www.kernel.org/pub/software/scm/git/docs/user-manual.html uses just "git stash"	11:43
\|Voker57\|	i use git stash with 1.6.0.6 ok	11:45
TheGeralt	\|Voker57\|: does it also work if your index is empty, i.e. no changes are added to it?	11:45
\|Voker57\|	% git stash	11:47
	No local changes to save	11:47
	→ Yuuhi joined	11:47
	← priidu left	11:48
TheGeralt	That's the same message I got, but according to the documentation it saves all your changes if you type "git stash" and if you type "git stash save" it works even if there are no changes added to the index	11:48
	← kanru left	11:48
\|Voker57\|	what would it save if no changes were added?	11:48
TheGeralt	Oh yeah it only stashes changes in the index, not the ones in the working tree.	11:49
	\|Voker57\|: the status of your working tree, but it seems to do that you have to type in "git stash save"	11:49
	← hobodave_ left	11:49
	gouki_ → gouki	11:50
charon	TheGeralt, \|Voker57\|: 'git stash' is definitely about both working tree _and_ index changes	11:50
TheGeralt	charon: ok, but why does the following only stash the 'bar' file and not the 'test' file? "cd /tmp; mkdir sample; cd sample; git init; echo 'hello' > greeting; git add greeting; git commit -m 'Initial'; echo 'foobar' > bar; git add bar; echo 'another one' > test; git stash"	11:52
	← protorom left	11:53
	← muthu left	11:54
	→ Ratler_ joined	11:55
	← greyface left	11:57
	← Ratler left	11:57
	← fruit left	11:57
	← Voker57 left	11:58
	← Eridius left	11:59
	→ theCarpenter joined	12:03
theCarpenter	are cogito and stacked git still alive?	12:03
broonie	cogito no, stgit yes	12:03
	→ ProperNoun joined	12:04
	→ Eridius joined	12:05
theCarpenter	it looks like the first hit on google for stacked git is dead	12:06
	http://www.google.com/search?q=stacked+git&ie=utf-8&oe=utf-8&aq=t&rls=com.ubuntu:en-US:unofficial&client=firefox-a	12:06
	→ aziz joined	12:07
drizzd	git-scm.com has the official link	12:08
	apparently the site is down...	12:08
	← mikl left	12:09
	→ hobodave_ joined	12:09
	→ Weaselweb joined	12:09
	← theCarpenter left	12:13
	→ warlock_mza joined	12:15
	← hobodave left	12:16
	→ greyface joined	12:16
	→ hobodave joined	12:16
	→ cilly joined	12:17
	→ theCarpenter joined	12:17
	← theCarpenter left	12:17
	→ Leefmc joined	12:17
	← CardinalNumber left	12:17
	← mithro left	12:19
	→ mithro joined	12:20
	→ j416 joined	12:23
charon	TheGeralt: because 'test' is untracked and in git's eyes doesn't have anything to do with the repo	12:26
TheGeralt	charon: so if I would change the greeting file 'git stash' would store its changes even if I don't add it to the index? Well, I guess that makes sense, because untracked (i.e. new) files don't affect older revisions of my project and thus it's safe to let them there	12:28
charon	TheGeralt: indeed, because 'greeting' is tracked. 'git stash' more or less tucks away all changes shown by 'git status'	12:29
	... except untracked files.	12:29
	(almost forgot that status.showUntrackedFiles is on by default...)	12:29
TheGeralt	charon: ok, then git stash's behaviour makes sense, thank you :)	12:30
	← heren left	12:35
	→ foutrelis joined	12:36
	← hobodave_ left	12:36
	→ NSaibot joined	12:37
	→ cbreak\|evil joined	12:38
	← peritus_ left	12:39
	→ bryanl joined	12:46
	← NSaibot_ left	12:48
	→ jonalv joined	12:49
	← \|Voker57\| left	12:49
	← mxey left	12:52
	← cbreak left	12:54
	← cilly left	13:02
	← kumbayo left	13:04
	→ jackdempsey joined	13:05
	→ chris2 joined	13:10
	→ CardinalNumber joined	13:12
	← hobodave left	13:13
	→ icwiener joined	13:14
	← Ratler_ left	13:16
	← Ariens left	13:19
	→ Ratler joined	13:19
	← foutrelis left	13:24
Leefmc	Question: How does git handle large (200G) amounts of binary data? For whatever reason, im a bit curious about protecting the integrity of my music library by using Git to revision it. Any comments on using git for this odd application?	13:26
bremner	Leefmc: not so great is the concensus	13:29
	→ snitko joined	13:29
Leefmc	haha, im not surprised, but why exactly? Performance wise, i wasn't sure, but i did assume git would store large amounts of data for itself	13:29
bremner	Leefmc: mp3's don't diff very well as I understand it.	13:30
Leefmc	I started thinking, and aside from some drastic times where i need to edit large amounts of meta-data on the files (Say a few albums artist name, or something), but in general the edits are few and far between, so i assumed git's size wouldn't expand exponentially.	13:31
drizzd	compressed things usually don't diff very well	13:31
bremner	Leefmc: just try with one or two files, update metadata and see how much data is schlepped around	13:31
Leefmc	bremner: Well actually i wasn't going to diff at all	13:31
bremner	Leefmc: internally git does, that's the point	13:31
Leefmc	bremner: But good call. Aside from storage size, do you think git could handle it?	13:31
	→ Zimm joined	13:31
Griffon26	Leefmc: you'd know when it was tampered with, but there would be no error correction. Aren't you better off with some kind of error correction code?	13:31
bremner	Leefmc: I'm just not sure it's a win if you are storing the the full files for each version anyway, some other tool might be better	13:32
	→ drewr joined	13:32
drizzd	Griffon26: that would increase it's size even more though	13:33
Leefmc	Ah, when i heard diff i was thinking more along the lines of "compare "A" commit, to "B" commit.", but by diff you are simply referring to git's revisioning. Correct? Eg, its ability to move back and forth between two file states	13:33
	← ProperNoun left	13:33
Leefmc	I actually thought git excelled at that, simply because i often heard bits about git/linux/binary	13:34
	→ foutrelis joined	13:34
Griffon26	drizzd: sure. But I'm wondering what kind of functionality he is looking for that he thought he'd find in git	13:34
	→ cbreak joined	13:34
	← neunon left	13:35
Leefmc	Griffon26: Functionality wise, basically just a saftey net, at minimal size. If a modification is made, 6 months down the road i could have git revert back to the good state. Or if a player corrupts a file, i can revert it. Or unintended deletions, etc.	13:35
drizzd	Leefmc: if you're just concerned about corruption it would be more efficient simply to compute and store hashes for each file. You can do that using git hash-object	13:35
bremner	Leefmc: I'm talking about the internal representation of a git repo http://eagain.net/articles/git-for-computer-scientists/	13:36
Leefmc	But really, this is just something i thought of _now_. I've never used git for raw binary files (eg, none text data) so i had no idea if it was possible, how possible it was, etc. Which is why i am here :)	13:36
bremner	Leefmc: sure it is possible. Git just isn't as fantastic with binary data as we have grown used to with text	13:36
	(as far as performance_	13:37
	)	13:37
	→ Owner joined	13:37
	← Ratler left	13:37
	→ Ariens joined	13:37
Leefmc	drizzd: Pardon my ignorance, but You can't actually do anything with hashes right? I mean.. sure i can know if the file changed, but thats about it.. no? (Ie, i can't repair the file, simply by knowing the hash)	13:37
	→ naeu joined	13:38
drizzd	Leefmc: that's correct	13:38
	but if the repository is corrupted, you cannot restore it either	13:38
	(unless you have a backup)	13:38
	→ cbreak\|evil_ joined	13:38
Pieter	if it's a single bit swap you might be able to brute force it	13:39
	→ Voker57 joined	13:39
drizzd	but I think you're not talking about corruption in the common sense. You just mean (technically) intended modifications.	13:39
Leefmc	bremner: Any idea if git scales well? Eg, if i could try it on say 1 album, do a number of tests to see gits performance, storage size, revision tests, etc, and then scale the math up to the size of my full library. Think that would be a decent scale test?	13:39
	drizzd: Well yea, theres any number of problems heh. Anything from a program messing with a file, to it being deleted by me, etc. The real problem is that these things can happen in a large library, and you may never know. I keep backups, but each time i update my backup hardrives, i don't actually listen to each song. And there in lies the real problem, i can be backuping up something broken. Revisioning would be awesome for this	13:40
	→ jmspeex joined	13:41
Leefmc	if it worked, ofcourse :o	13:41
bremner	Leefmc: It's only a guess, but I suppose it should be now worse than linear	13:41
	s/now/no/	13:41
heipei	Leefmc: keeping mp3s with git is one of the most stupid ideas ive heard, to be honest	13:41
	← Ademan left	13:41
bremner	heipei: hey, now, you need to get out more :-)	13:41
heipei	Leefmc: what you could do is keep a list of the files with git, generated via crontab	13:41
Leefmc	heipei: Lol, i was just curious man, god forbid i try out a goofy idea.	13:41
heipei	Leefmc: no, im just saying, i had that idea too once	13:42
drizzd	heipei: do you have a better alternative?	13:42
Leefmc	(Note that this was not a "LOI" but a True "LOL" case, so that was infact a laugh out loud)	13:42
heipei	Leefmc: but think about, git does nothing that applies to the scenario of binary files which _dont_ change in their content, aren't diffable or compressable	13:42
	drizzd: yeah, just back-up the mp3s, and if you want to track the dir content use a list of the files which you can then track with git	13:43
drizzd	heipei: but they do change in his case	13:43
bremner	maybe some version file system is usable? http://en.wikipedia.org/wiki/Versioning_file_system#Linux	13:43
heipei	drizzd: oh, then i didnt read that	13:43
Leefmc	heipei: Man your right, if only i would have thought this out with other smart individuals who know git better then i. If only if only i knew of a place where i could do this at. Perhaps the people at #git have an idea.. people other than you, ofcourse.	13:43
	← cbreak\|evil left	13:44
heipei	Leefmc: ok, let me ask you this: why would you want to _version_ your mp3s? (not backup, the versioning part)	13:44
Leefmc	bremner: Good idea	13:44
	← jonalv left	13:45
Leefmc	heipei: To, as efficiently as possible, backup all versions of a file, as to have a history of any problems/modifications that may have taken place.	13:45
drizzd	the problem with binary files in general is that you cannot easily see _what changed_	13:45
Leefmc	drizzd: Oh that i am aware	13:46
	drizzd: But if there was a history, and you knew something was up with the file you didnt like, you can just trace back time and try each version until you found the correct one.	13:46
Ilari	When it comes to MP3's I don't think editing metadata would cause large internal changes...	13:47
Leefmc	Ilari: They don't, but none the less	13:47
	← foutrelis left	13:47
heipei	Leefmc: but for an mp3?	13:47
Leefmc	Ilari: Im a nut like that ;), i'd like to have the power to revert things if needed	13:47
bremner	Leefmc: Actually, if you work a bit, you could have a seperate tree of metadata that would version nicely	13:48
Ilari	You could test this by taking some MP3 file, making two copies, editing metadata of one. Then try how well they diff against each other using test-delta.	13:48
Leefmc	heipei: "But for an mp3?"?	13:48
heipei	Leefmc: what "problems" arise when you change stuff an mp3?	13:48
	Ilari: or just have a look with a hexeditor	13:48
	← mithro left	13:48
	← snitko left	13:48
Ilari	Leefmc: One problem that adding MP3's to git repository would run to is that large files are inefficient to diff, thus blowing up repack times.	13:49
	→ ProperNoun joined	13:49
heipei	Leefmc: you could write a tool that (assuming you dont rename/move files often) puts out the metadata for each of your mp3s and then just versions this output	13:49
Leefmc	bremner: Not sure, i havent even considered separate tracking solutions like that because i'd like the flexibility of players, etc. Eg, im assuming with something like that, i would have to make a special plugin for any media player to handle the remote metadata. Not to mention apply it to the mp3s, before putting them onto ipods and such, etc.	13:50
bremner	Leefmc: nah, just have a script that syncs your mp3s with the metadata	13:51
Leefmc	Anyway, this wasn't a serious question, i was mostly just looking to ping the idea off of some folks here :). Thanks to bremner, drizzd, and Ilari :)	13:51
	← cbreak left	13:51
Leefmc	bremner: True	13:51
	If it turned out to be easily possible (eg, git handled binary data superbly) then hey, win win. If not, nothing lost :)	13:52
bremner	anyone remember the name of that project that seemed to be building versioned file system on top of the git object store, something to do with gnome?	13:52
Ilari	Leefmc: Of course, it could be possible to directly write pack representations using custom algorithms... After all, even after metadata editing, MP3's should have very large common chunk...	13:52
Leefmc	Im surprised to see there are no versioning linux file systems	13:53
	→ jm joined	13:54
	→ ankit9 joined	13:54
	→ parasti joined	13:54
	→ resmo joined	13:55
Ilari	Leefmc: But the problem with that is that on each repack, you likely need to write a huge file anew...	13:55
Leefmc	Yea, i was assuming that even if git could handle it, i would have a huge git repo	13:55
Ilari	Leefmc: Most data could be straight-out copied, but it would still be huge.	13:55
Leefmc	Which may be worth it, if it did amazing things	13:56
	→ peritus_ joined	13:57
	→ cantropy joined	13:58
Ilari	Leefmc: OTOH, writing custom content-addressable filesystem would likely not be that hard...	13:59
Leefmc	It'll be nice when we get to the point in time where its standard to have the file system revision files. Imagine having every file, at all times, versioned. Hell, even windows would be safer with this technology heh	13:59
	→ Ratler joined	13:59
	← alley_cat left	14:00
	← jm left	14:01
	← xenoterracide left	14:01
	← CardinalNumber left	14:01
	→ xenoterracide joined	14:01
	→ nud joined	14:01
vomjom	Leefmc, it already exists	14:01
	in some log structured filesystems	14:01
	→ Owner_ joined	14:01
	→ harinath joined	14:02
vomjom	i think nilfs is the more notable one right now	14:02
	← cantropy left	14:02
vomjom	although it's slow in practice	14:02
Leefmc	vomjom: I simply mean when its standard, eg, on all systems. For whatever reason, its not now. I'll assume its a case of technology limitations (too many resources), along with implementations simply not good enough, etc.	14:02
	yea	14:02
	→ alley_cat joined	14:02
	→ jrmuizel joined	14:02
vomjom	Leefmc, it works well on SSDs where random access is cheap	14:03
	but for regular hard drives, you'll get a lot of seeking	14:03
	if you version everything	14:03
Leefmc	vomjom: Interesting	14:03
	vomjom: Then perhaps its coming sooner than i even assumed. Given that SSD is right around the corner	14:03
	← plasticine left	14:03
vomjom	nah	14:03
	→ Owner__ joined	14:03
vomjom	i think snapshotting is more useful to most people	14:04
	and that's coming with btrfs	14:04
	btrfs will be the new standard :P	14:04
	← bentob0x left	14:04
Leefmc	snapshotting the entire hd?	14:04
vomjom	if you want	14:05
	it's a pretty cheap operation	14:05
Leefmc	http://btrfs.wiki.kernel.org/index.php/Main_Page nice	14:05
vomjom	btrfs was merged in the kernel recently	14:05
	as of a few weeks ago	14:05
	it's not stable yet, though :P	14:05
	← warlock_mza left	14:05
Leefmc	yea	14:06
icwiener	Is it possible to disable branch point detection in git-svn? I only want to track trunk of a project and need to keep the archive's size low.	14:06
Leefmc	vomjom: http://btrfs.wiki.kernel.org/index.php/Development_timeline Q408 heh :o	14:06
vomjom	most of that's already done, actually	14:06
	the disk format is stable already	14:07
	they just haven't updated the wiki	14:07
Leefmc	vomjom: Any thoughts on when it'll be fully usable/supported? by 2010 perhaps? or more likely 2012	14:07
vomjom	i think mid-late 2009	14:07
Leefmc	.. that'll only give us a year to use it. Given that the world will end before 2013.. ;D	14:07
	nice	14:08
	→ icwiener_ joined	14:08
vomjom	it's pretty usable right now if you want to risk your data :P	14:08
	grub doesn't support it yet, though	14:08
Leefmc	haha, considering the insane solution to my data protection i was considering, nope. :o	14:08
	Anyway, thanks for the discussion all, back to work for me	14:08
	→ Acry joined	14:11
	← jrmuizel left	14:12
	← Acry left	14:13
	→ LiamH joined	14:16
	← Owner left	14:16
	← gmarik_ left	14:17
	cbreak\|evil_ → cbreak	14:20
	← Owner_ left	14:21
	← nud left	14:23
	→ nud joined	14:24
	← icwiener left	14:24
	→ flaguy48 joined	14:25
	→ gmarik_ joined	14:26
Arrowmaster	so everything git pull does is from git-pull.sh in the source right?	14:30
	→ gitte joined	14:31
	→ jrmuizel joined	14:35
	→ srp joined	14:37
	← gaveen left	14:38
srp	i pushed a bad commit to my remote repo by mistake. git reset --hard HEAD~1 on my local branch reset's my changes locally. How do I reset the remote branch on the repo ?	14:38
Voker57	git push -f maybe	14:39
srp	ok thanks - that worked.	14:40
	← jrmuizel left	14:42
	→ DrFrasierCrane joined	14:47
	← ivaxer_ left	14:52
	← ankit9 left	14:55
	← srp left	14:58
	→ gaveen joined	14:58
	→ srp joined	14:59
	← jefferai left	14:59
	→ jefferai joined	15:00
d-snp	Leefmc: there are plenty snapshotting filesystems out there already	15:03
	you don't have to wait for some beta fs :)	15:04
	← solydzajs left	15:04
Leefmc	d-snp: Snapshotting was not the original conversation, just a bit of side talk	15:04
	d-snp:The conversation was about file system versioning.	15:04
d-snp	right, which you usually do by snapshotting right?	15:05
	→ solydzajs joined	15:06
	← Owner__ left	15:06
Leefmc	Well i had thought snapshotting was more like mirroring, rather than true revisioning	15:07
	→ Owner__ joined	15:07
Leefmc	Eg, the vfs wiki entry says there are no linux versioning file systems currently	15:07
	but honestly im far from knowledgeable on the subject. This was all just conversation, spawned off another topic.	15:08
d-snp	ok :P	15:08
	it was about backing up your mp3's?	15:09
	← Niamor left	15:09
	→ Niamor joined	15:09
d-snp	I have a friend who backs up his large (multiple tb's) anime database with snapshots	15:11
	so he doesn't accidentily delete stuff	15:11
	→ vbabiy joined	15:12
	→ nick_h[litage] joined	15:16
	→ CardinalNumber joined	15:17
	← cbreak left	15:17
	→ rootPosixXombie joined	15:17
rootPosixXombie	I had deleted accidentally a subdirectory of a local copy of the linux kernel git tree	15:18
	what should I do to restore it ?	15:18
Voker57	git checkout <dir>	15:18
	→ Tuttle joined	15:18
Tuttle	How can I quickly know what remote branch is tracked by my local current one?	15:19
rootPosixXombie	Voker57, thnks	15:19
wereHamster	Tuttle: look at .git/config	15:19
gitte	Tuttle: git config branch.<name>.merge	15:19
	Tuttle: where <name> is the name of the current branch	15:20
Tuttle	it's empty	15:20
	therefore the local branch is just local, no remote peer set, right?	15:20
Ilari	Tuttle: Local branches are local branches. They may or may not have default pull source set... But the source can be always specified in pull command...	15:22
	→ Beeet joined	15:23
	← unreal left	15:24
Ilari	Tuttle: And some commands would display the relation of local branch and what it is marked to track.	15:24
	→ abbe joined	15:24
	→ unreal joined	15:25
Tuttle	apparently i created new branch locally with git checkout -b newbr, so now i need to create the branch with the same name in the remote repo and make local branch to track the remote.	15:25
	← NSaibot left	15:27
Ilari	Tuttle: Push that branch to remote and add the two settings ('branch.<name>.remote' and 'branch.<name>.merge').	15:27
	Tuttle: Note: 'branch.<name>.merge' is name of local branch on _REMOTE_ side, not name of local remote tracking branch!	15:27
	Tuttle: So something like 'git config branch.foo.remote origin', 'git config branch.foo.merge refs/heads/foo'.	15:28
jast	hmm, I guess a feature to start tracking a remote branch after the fact would be nice	15:28
Ilari	Also feature to have push optionally add remote tracking data on push.	15:29
Tuttle	jast: yes. this is not an easy operation for us newbies. and starting the local branch with later wish of publication is not so rare workflow i think.	15:29
gitte	jast: IIRC I provided a patch for that ages ago.	15:30
	← lednerk left	15:30
Tuttle	Ilari: the first my command will be : git push origin origin:refs/heads/newbr right?	15:30
Ilari	Tuttle: Nope: 'git push origin newbr'.	15:30
jast	gitte, quite possible, I'm afraid I don't always remember all patches ;)	15:30
	← Owner__ left	15:31
jast	git push --track or something, perhaps	15:32
Tuttle	i have branch.autosetupmerge = true, i expect no need of --track	15:33
	eno__ → eno	15:33
Tuttle	looks like there's is no --track in git push	15:33
jast	yeah, that's what I was suggesting to add	15:33
gitte	jast: you are allowed to not remember: http://article.gmane.org/gmane.comp.version-control.git/13735	15:34
Tuttle	additionally i expected git branch will dump the info about tracking in verbose mode.	15:35
	or verbose verbose mode :)	15:35
jast	oh, back then I wasn't even born yet	15:35
	← ProperNoun left	15:35
	→ cbreak joined	15:37
heipei	Signed-off-by: Johannes "always off by one" Schindelin	15:37
	lol	15:37
gitte	jast: lol!	15:37
Tuttle	:))	15:37
	→ cthompson joined	15:37
	← Fullmoon left	15:38
	→ jrmuizel joined	15:39
	→ pi-meson joined	15:41
pi-meson	I have a branch that I created fro master a few days ago, and since have committed ~10 commits to add a new feature. I would like to merge this branch back into master, but as a single commit. Can git help me here?	15:42
jast	"as a single commit" meaning you want to not have proper merge history? merge --squash	15:42
	→ static^ joined	15:43
pi-meson	jast: the "proper" there suggests that this is a bad thing?	15:44
jast	it depends	15:44
drizzd	pi-meson: do not forget your history so you can learn from it	15:44
Griffon26	pi-meson: or this: http://blog.madism.org/index.php/2007/09/09/138-git-awsome-ness-git-rebase-interactive	15:44
jast	if you want to merge again between the two branches later on, it tends to be a bad thing	15:44
	if the new branch is a throw-away thing, it doesn't really matter	15:44
pi-meson	right, no this is a "throw-away" branch that I was hacking on a brand-new feature	15:44
jast	generally I would advise against huge commits since they make finding bugs later on harder than it needs to be	15:45
drizzd	pi-meson: it's not only that. If you want to find a bug later, it's better to have fine-grained changes	15:45
jast	of course it's up to you to decide what "huge" means	15:45
thiago_home	on the other hand, having too-fine-grained can be too much information	15:45
	you don't want "compile" and "forgot to add a file" commits	15:45
pi-meson	thiago: yea, i tend to have lots of those commits. I'm a bit sloppy with my git use sometimes.	15:46
	→ CardinalFang joined	15:47
CardinalFang	kernel.org git web interface b0rken? http://git.kernel.org/	15:47
	→ jonalv joined	15:48
	← Beeet left	15:49
jast	CardinalFang, works for me	15:49
Ilari	Quick'n'dirty remote tracking branch setting script: http://git.pastebin.com/m623baab3	15:50
jast	looks good	15:51
CardinalFang	jast, Huh. wget, "Connecting to git.kernel.org\|130.239.17.7\|:80... connected.\nHTTP request sent, awaiting response... No data received."	15:51
jast	CardinalFang, git.kernel.org appears to resolve to different addresses depending on your location	15:52
CardinalFang	Mine is git4.	15:52
jast	try git.eu.kernel.org	15:52
	→ warlock_mza joined	15:52
Tuttle	Ilari: it looks like only thing i really needed to do was git push origin newbr i adviced me. no changing of config, still there is no mention of newbr. but everything works: my local branch commits to remote newbr. why should i set the config as you told me then?	15:53
Ilari	Tuttle: Without it, you can't just say 'git pull' (you would have to explicitly specify it as 'git pull origin newbr') to pull branch 'newbr' from remote 'origin'.	15:54
	If you want to make a new script that's subcommand of git, name it 'git-foo', put it into some directory that is in $PATH and make it +x.	15:55
Tuttle	Ilari: ah, tried it. forgot. thanks for your time	15:55
	Ilari: this all can't be done any easier?	16:00
	i guess then that starting new branch as remote is quickier way to define...	16:01
	← CardinalFang left	16:01
	→ mckoan joined	16:02
Ilari	Tuttle: Well, I hacked up script related to that (see the git.pastebin.com link)...	16:02
Tuttle	oh, i missed that it was on me	16:02
	→ sergio joined	16:02
	← vbabiy left	16:04
	→ muthu_ joined	16:06
	← static^ left	16:06
	→ PloenK joined	16:08
akitada	git svn rebase stopped working for me with this error http://gist.github.com/51474	16:11
	Any clues?	16:11
Ilari	akitada: Isn't that one line duplicated?	16:12
Tuttle	it looks like putting working script git-foo to PATH and make it +x is not sufficient for git 1.5.4.3 to consider it as a subcommand. but it does not matter, can be run directly	16:12
Ilari	Tuttle: Works for me (v1.6.1.<something>)...	16:13
	'git-foo' appears as subcommand 'foo'.	16:13
	→ riverrat66 joined	16:13
akitada	Ilari: That's what it is on my disk	16:13
Ilari	akitada: Try deleting one of those duplicates?	16:14
	→ reprore joined	16:15
akitada	Ilari: added http://gist.github.com/51474	16:15
	btw, I'm using 1.6.1	16:16
	← mckoan left	16:20
	→ k0001 joined	16:21
	← awarde left	16:21
	← riverrat66 left	16:21
	← Leefmc left	16:23
	← ciskje left	16:24
	→ NSaibot joined	16:25
akitada	"git svn clone something; Ctrl-C" will reproduce this state.	16:27
	← kek left	16:28
parasti	akitada: sure. no local branch was created and thus nothing is checked out	16:28
	akitada: git svn fetch should continue to function, though	16:29
	← srp left	16:29
akitada	parasti: seems working. thanks.	16:30
	→ inetic joined	16:35
	← bobmcw left	16:36
	← tango_ left	16:39
	← TheGeralt left	16:40
	← pi-meson left	16:40
	→ madewokherd joined	16:40
	← Niamor left	16:41
	← Aikawa left	16:43
	← felipec left	16:43
	→ felipec joined	16:44
	← j416 left	16:44
	→ Owner__ joined	16:44
	→ ProperNoun joined	16:45
	→ Aikawa joined	16:46
	← CardinalNumber left	16:47
	← cedricv left	16:50
	← Tuttle left	16:50
	← punto left	16:51
	→ _mike_m_ joined	16:51
	← rootPosixXombie left	16:53
	← mike_m left	16:53
	→ codebrulee joined	16:55
	← codebrulee left	16:55
	→ quickdev joined	16:55
quickdev	Hey	16:55
	I've got two branches, A and B. A has many differences in subdirectories from B. I'd like to merge a certain commit from A to B. What's the best way to accomplish that?	16:57
vmiklos	quickdev: man git-cherry-pick	16:57
Gitbot	quickdev: the git-cherry-pick manpage can be found at http://git.or.cz/man/git-cherry-pick	16:57
quickdev	vmiklos, thanks, great	16:57
	→ Niamor joined	16:58
	→ gol joined	17:00
	← Yuuhi left	17:00
	→ jonalv` joined	17:03
	→ bobmcw joined	17:04
quickdev	vmiklos, after cherry-picking and running git push, it says "Everything up-to-date", although the commit is not pushed. What did I forget?	17:05
	→ Leefmc joined	17:07
vmiklos	are you on a detached head?	17:07
	← pasky_ left	17:08
	→ scientes joined	17:08
	→ pasky joined	17:08
	← scientes left	17:09
	→ scientes joined	17:10
quickdev	vmiklos, what does that mean?	17:10
	→ kukks joined	17:10
vmiklos	http://www.kernel.org/pub/software/scm/git/docs/gitglossary.html#def_detached_HEAD	17:10
	git branch has "no branch" in its output in that case	17:11
	→ Yuuhi joined	17:11
quickdev	vmiklos, no, git branch says "fso_ms5", which I checkout out with git checkout fso/milestone5 -b fso_ms5	17:11
	so it's not detached	17:11
	→ FreakGuard joined	17:12
vmiklos	ok. is there such a branch on the server already?	17:12
quickdev	vmiklos, yes - I checked it out - so it already has to be there? ;)	17:13
	→ vbabiy joined	17:13
vmiklos	hm, ok.	17:13
	what does git push -v say?	17:14
quickdev	vmiklos, http://rafb.net/p/gIT1GU23.html	17:15
	← pasky left	17:15
	→ pasky joined	17:15
quickdev	vmiklos, this: http://rafb.net/p/OjduiK17.html	17:16
vmiklos	see if the hash of 'git rev-parse fso/milestone5	17:17
	'	17:17
	and 'git ls-remote <url>' matches	17:17
	→ __name__ joined	17:18
	← NSaibot left	17:18
	→ softdrink joined	17:19
	← name left	17:19
	→ schacon joined	17:19
	__name__ → name	17:19
	← jonalv left	17:19
	→ jim_c joined	17:19
quickdev	vmiklos, it's the same, although my commit through cherry-pick is on top when executing git log.	17:19
vmiklos	aah	17:20
	git push origin fso_ms5:fso/milestone5 :)	17:20
	← schacon left	17:21
quickdev	vmiklos, that did it. Thank you very much :)	17:22
	← cbreak left	17:25
	← pasky left	17:25
	← gitte left	17:26
	→ nthalk joined	17:28
nthalk	Anyone know how to put remote repositories within a parent repository?	17:28
	→ ankit9 joined	17:29
	→ schacon joined	17:30
	← schacon left	17:30
nthalk	Hello!	17:31
	→ pasky joined	17:31
nthalk	git pull room	17:31
	← LotR left	17:32
nthalk	Hmmm.	17:32
	← warlock_mza left	17:32
nthalk	rm -rf room; git clone #git room	17:32
	Anyone alive?	17:33
Aides	fl	17:33
	kr	17:33
	→ cbreak joined	17:33
Aides	forgot to escape # ;)	17:33
nthalk	Was that a faint flicker of life I just saw?	17:33
Ilari	nthalk: 'git remote add'?	17:33
nthalk	Ilari: doesn't that just add at root?	17:34
Ilari	nthalk: Or 'git submodule add' for submodules?	17:35
nthalk	Sweet, that's what I am looking for!	17:35
	Thanks!	17:35
	Googling subrepository, subclone, subcheckout yielded nothing.	17:35
	Thanks alot!	17:36
	→ LotR joined	17:36
	← muthu_ left	17:39
	← masak left	17:43
	← jonalv` left	17:44
	→ amahi joined	17:45
	→ jm joined	17:47
	→ foutrelis joined	17:48
	← cpg left	17:48
	→ a-priori joined	17:55
	→ oudeis joined	17:57
	← solydzajs left	17:57
	→ solydzajs joined	17:58
	→ rvhi joined	17:59
	→ johnw joined	18:00
Circuitsoft	Hello	18:01
	← nick_h[litage] left	18:01
Circuitsoft	I think I found a bug in git, although you could (sort of) accuse me of trying to break it.	18:01
	git checkout -b alex\'s_test_branch froze.	18:01
	→ drizzd_ joined	18:02
	← Strogg left	18:02
Ilari	Circuitsoft: What git version? That works for me...	18:07
	← mikearr left	18:07
Circuitsoft	1.6.0.4	18:07
	Oh, Nevermind, I'm dumb.	18:08
	I'm on Cygwin.	18:08
	I forgot that Windows can't handle arbitrary characters in filenames.	18:08
	← solydzajs left	18:08
Circuitsoft	Does that restriction cause a race of some kind?	18:09
	← a-priori left	18:09
	← naeu left	18:10
	→ solydzajs joined	18:10
	amahi → cpg	18:11
	→ a-priori_ joined	18:12
	→ mikearr joined	18:16
	→ warlock_mza joined	18:17
	→ muthu_ joined	18:18
	→ urbanmonk joined	18:19
	← drizzd left	18:20
	→ SjB joined	18:21
	← SjB left	18:21
	→ SjB joined	18:22
	→ trofi joined	18:23
	← rvhi left	18:23
	→ halfdan joined	18:24
	← ankit9 left	18:24
halfdan	ahoi	18:24
Ilari	...	18:24
	→ ankit9 joined	18:24
halfdan	is there a recommended directory-structure for a git repository?	18:24
Ilari	halfdan: Everything at root.	18:25
	halfdan: :-)	18:25
	halfdan: I presume you have used SVN before?	18:25
trofi	hi! i am happy git-1.6.1 user. when i use git-pull - i get master branch fast-forwarded (if untouched), but other tracked branches (created by git checkout -b l_b origin/r_b) do not update automatically. How to beat that?	18:25
halfdan	indeed	18:25
kevlarman	trofi: pull = fetch + merge	18:26
trofi	yep, i understand	18:26
kevlarman	you need a working copy to perform a merge, so only the checked out branch gets merged	18:26
offby1	trofi: I assume you need to do something like "for b in $(git branch -a); do git checkout $b; git pull; done"	18:26
doener	trofi: you mean "git pull" tells you that you didn't tell it what to do?	18:26
Ilari	halfdan: In SVN, branches are top of versioning. Git does not work that way.	18:26
	halfdan: Instead, it supports branches at versioning level.	18:27
kevlarman	halfdan: you're better off forgetting everything you know about svn while you try to learn git (i learned this the hard way)	18:27
trofi	doener: no	18:27
	→ tvoet_ joined	18:28
trofi	offby1: it is understood, but i tought there is yet builtin solution (some magic git fast-forward-all-fast-forwardable-local-branches)	18:28
halfdan	kevlarman: i'm trying. i just watched the google tech talk by linus about git	18:28
doener	halfdan: http://eagain.net/articles/git-for-computer-scientists/	18:29
	halfdan: http://www.newartisans.com/blog/2008/04/git-from-the-bottom-up.html	18:29
Ilari	'git fetch . refs/remotes/origin/:refs/heads/'?	18:29
	Oops, that doesn't do sane things if you have branches without local equivalents...	18:30
halfdan	doener: thanks	18:31
	← pasky left	18:32
	→ pasky joined	18:32
	← inetic left	18:33
offby1	trofi: I don't know that there is one built in, but I bet it wouldn't be hard to kludge one up	18:33
	dunno _how_ ...	18:33
Ilari	Some useful commands for that sort of script: merge-base (for fast-forwardness testing), for-each-ref (grabbing list of local branches) and update-ref (doing the actual update)...	18:34
	→ EmilMedve1 joined	18:39
	← a-priori_ left	18:40
	→ marvil07 joined	18:40
	← EmilMedve left	18:41
	← nadim__ left	18:41
	→ tsaarni joined	18:42
	→ ashleyw joined	18:45
	→ a-priori joined	18:45
	→ simosx joined	18:46
	→ ceej joined	18:46
	← muthu_ left	18:48
	→ ceej_ joined	18:52
	→ plopix joined	18:53
doener	trofi: for something really ugly: http://git.pastebin.com/m61cda5a1	18:58
	→ priidu joined	18:58
trofi	push?	18:59
	← nadim_ left	18:59
	← a-priori left	18:59
	→ pyalot joined	19:00
doener	trofi: yeah, local push... like fetch, that refuses to do non-fast-forwards by default. So that's an extra safety net	19:00
	← bobmcw left	19:00
doener	push just felt more natural to me than fetch	19:00
	→ hobodave joined	19:00
doener	anyway, it's just to get an idea what's needed	19:00
trofi	ah yes, sounds really good	19:01
doener	it still lacks a bit of error checking, and might do funny stuff when you have branch.<name>.remote set to an url instead of an remote	19:01
	← solydzajs left	19:02
	→ solydzajs joined	19:03
trofi	i suspect all this can be done with one push command (with --all option)	19:03
doener	uhm, no?	19:03
	that would just locally push master to master	19:04
	← jrmuizel left	19:04
doener	and so on for the other stuff	19:04
	but you need to push from the remote tracking branch to the local branch	19:04
trofi	yes i understand	19:04
doener	in a test run I had, for example:	19:04
	5dc1308..21c042e origin/pu -> foo	19:04
Gitbot	[git 5dc1308]: http://tinyurl.com/cd6zta -- Merge branch 'js/patience-diff'	19:04
doener	damn bot	19:04
	← pyalot left	19:04
	→ a-priori joined	19:05
Voker57	lol collision	19:05
trofi	what did he show?	19:05
	→ etienne9 joined	19:05
trofi	ah git.git	19:07
	if git push can push all refs, why git pull can't? :]	19:09
	→ SOFY joined	19:09
doener	push can't cause conflicts	19:09
SOFY	HOLA	19:10
	← ceej left	19:11
SOFY	HOLA	19:12
	← SOFY left	19:13
	→ SOFY joined	19:13
	→ drobbins joined	19:13
Voker57	git fetch is opposite of git push	19:13
	← SOFY left	19:13
	← urbanmonk left	19:15
	→ sergio_ joined	19:15
Ilari	If one used fetch for the task instead of push, then one wouldn't have to check for current branch if one doesn't mind about one error message...	19:16
Voker57	well not exact opposite...	19:17
trofi	fetch syncs real remote repo with local remote/ state, no?	19:18
doener	trofi: that's just what the default fetch config for a remote does	19:18
	trofi: if you look at, for example, remote.origin.fetch, you'll probably see "refs/heads/:refs/remotes/origin/"	19:19
RandalSchwartz	test-ctype is untracked after current head fetch	19:19
	should i just remove it?	19:19
doener	trofi: and _that_ is why "git fetch origin" updates the remote tracking branches origin/*	19:19
RandalSchwartz	or is that a bug?	19:19
	→ dreiss joined	19:21
trofi	fetch is perfectly understood. it does not touch local state at all	19:21
	and remote :]	19:21
	(remote host)	19:22
drizzd_	RandalSchwartz: untracked files can 'magically' appear if the ignore rules change	19:22
Ilari	Looks like test-ctime is new file and not in .gitignore...	19:24
doener	trofi: the remote tracking branches are also "local". And when you setup fetch to "refs/heads/:refs/heads/" (eg. a bare mirror repository) then fetch will of course also update the local branch heads	19:24
	← warlock_mza left	19:24
Arjen	Does anyone know what 'xfrm' means in 'xfrm_status' and 'xfrm_msg' in diff.c and diffcore.h ?	19:25
	← sergiodj left	19:25
offby1	presumably it's an abbreviation of "transform"	19:26
trofi	doener: ah, good, what fetch will do if local state isn't fast-forwardable?	19:26
offby1	.oO("Heh. 'yaph.org'")	19:26
Arjen	Hm?	19:26
doener	refuse by default unless you use -f or prefix the refspec with a +	19:26
	offby1 raises an eyebrow at RandalSchwartz	19:26
Arjen	Heheh	19:26
RandalSchwartz	an eyebrow?	19:26
doener	so, for a mirror, it's better to have the + prefix	19:27
trofi	to add new branches	19:27
	← vbabiy left	19:27
doener	adding new branches is not non-fast-forward, so the prefix doesn't matter there	19:27
	that's already handled by the * wildcard	19:27
offby1	ok, ok	19:27
	offby1 raises a stein and waves it at RandalSchwartz	19:28
RandalSchwartz	oh. ok	19:28
	hey	19:28
drizzd_	Ilari: yeah, it's a new build target	19:28
trofi	so, what does + prefix?	19:28
	→ cantropy joined	19:29
	→ jrmuizel joined	19:29
	→ nick_h[litage] joined	19:30
	→ offby1` joined	19:34
doener	trofi: it forces non-fast-forwards	19:34
trofi	ah. it is meaningful only for fetch?	19:35
	but push should use -f for the same?	19:35
doener	no, same for push	19:37
	so you can do eg. "git push master bla +foo"	19:37
	that way, only "foo" is forced	19:37
	while -f would be "global"	19:37
	→ bobesponja joined	19:41
	← offby1 left	19:48
	← a-priori left	19:49
	← cantropy left	19:53
	← ankit9 left	19:53
	← quickdev left	19:55
	← Aikawa left	19:57
	→ Aikawa joined	20:00
	→ Pengwn joined	20:01
	offby1` → offby1	20:02
	→ spooneybarger joined	20:06
	← oudeis left	20:07
	→ a-priori joined	20:09
	→ ankit9 joined	20:12
	← abbe left	20:16
	→ CardinalNumber joined	20:17
	← ProperNoun left	20:17
	← Pengwn left	20:17
	← gholias left	20:19
	← ijcd_ left	20:21
Circuitsoft	I just did "git checkout -b my_new_test_branch" and it took about 30 seconds.	20:21
	That's rather unusual, isn't it?	20:21
	← bobesponja left	20:22
	← SjB left	20:23
	← mattikus left	20:26
	→ bar joined	20:27
	→ bobesponja joined	20:27
bar	what is the canonical way to get data from and push back to a cvs repository?	20:28
	→ balint joined	20:28
bar	or to put it another way, i want to create a reference git repo from which i can then pull	20:28
	→ essaii joined	20:29
bar	(or others)	20:29
Arrowmaster	to create a repo to act as a central repo? git --bare init	20:29
bar	Arrowmaster, from a cvs master	20:30
	should i just cvs-import?	20:31
Arrowmaster	oh, no clue i dont use cvs	20:31
	never have, hope i never will	20:31
bar	or is there a better way (unlikely, but i'd rather ask _before_ i commit to a particular course of action)	20:31
	i used to use it	20:31
	but..	20:31
	bar hates	20:31
	→ tango_ joined	20:33
Arrowmaster	umm theres a git cvsexportcommit	20:33
	← essaii left	20:34
	← a-priori left	20:37
	→ towski joined	20:38
	→ a-priori joined	20:38
	→ kevwil joined	20:40
	← softdrink left	20:45
	← kevwil left	20:46
	← dreiss left	20:47
	→ tchan1 joined	20:51
	← balint left	20:52
	→ schacon joined	20:53
halfdan	hrmpf	20:54
	why is TeX stealing me the space after \TeX?	20:54
	got "\TeX in" and it gets converted to TeXin	20:55
bremner	halfdan: because \TeX is a macro	20:55
	halfdan: use \TeX{}	20:55
halfdan	how do i keep the space?	20:55
	ok	20:55
	← tsaarni left	20:55
bremner	halfdan: err, we're a bit offtopic for #git	20:56
	← ceej_ left	20:56
halfdan	err damn	20:56
	sorry	20:56
	← tchan left	20:56
	← tvw left	20:57
	→ kevwil joined	20:58
	→ ceej joined	20:59
	← ashleyw left	20:59
	→ naeu joined	20:59
	← kevwil left	21:00
	← bobesponja left	21:03
johnw	schacon: ping	21:03
schacon	hey	21:03
johnw	hey there!	21:03
	had a github query	21:03
	← gwoo left	21:03
johnw	is there a way in my dashboard to separate out a listing of "my" repos (those I originating or maintain) from those I simply mirror?	21:03
schacon	not right now	21:04
	← zeom left	21:04
schacon	we're working on a new dashboard that should give everyone a lot more options	21:04
	← ceej left	21:04
johnw	ah, thanks	21:04
	thought you would know :)	21:04
	→ gwoo joined	21:05
	→ bobesponja joined	21:06
	← tango_ left	21:07
	→ g0bl1n joined	21:08
	→ vegard joined	21:09
vegard	hi. I made some commits at the head of no branch, and now I've lost them. is there a way to see "loose ends"? I am quite confident that the commit is there somewhere, but I don't have the sha1. help?	21:12
Aides	vegard: git reflog	21:12
context	or git fsck --unreachable	21:13
vegard	thanks! :)	21:13
	← solydzajs left	21:15
	→ solydzajs joined	21:15
	EmilMedve1 → EmilMedve	21:15
	← solydzajs left	21:17
	← marvil07 left	21:21
	← Aikawa left	21:26
	→ bobmcw joined	21:27
	← drewr left	21:28
	← DrFrasierCrane left	21:28
	← ia left	21:33
	← Dreamer3 left	21:33
	→ ia joined	21:34
	→ msaizar joined	21:34
	← g0bl1n left	21:37
	→ dreiss joined	21:38
	→ troynt joined	21:41
troynt	Can anyone recommend a plugin for git for eclipse or aptana ?	21:41
context	eclipse has a plugin in the software add/update listing	21:42
	i cant vouch for it though	21:42
	eclipse is a PoC	21:42
	peice of crap! not proof of concept	21:42
Voker57	lol	21:42
	i second that	21:43
troynt	context: what IDE do you use?	21:43
	← alley_cat left	21:43
	← CardinalNumber left	21:43
doener	vim!	21:43
troynt	doener, how do you do a find and replace on all files in your active project w/ vim?	21:43
	→ Aikawa joined	21:44
troynt	anywho.. back on topic... Git for Eclipse?	21:44
Aides	find and replace on all files? why would anyone ever need that?	21:45
	← naeu left	21:46
troynt	Aides: You've never had to fix someone elses code?	21:46
doener	troynt: git ls-files \| xargs \| sed ;-)	21:46
Aides	by replacing something in all files no	21:47
doener	troynt: but IIRC there's also some script that allows to define a project and do operations on all files	21:47
Voker57	Aides: what if you need to change copyright for example?	21:47
doener	and I don't do java, so I don't refactor my whole project every other minute	21:47
troynt	doener: neither do I	21:48
	← reprore left	21:48
	→ zeom joined	21:48
Aides	Voker57: does eclipse support multiline replaces?	21:48
Voker57	Aides: dunno, i'm talking just about such a need	21:49
	" why would anyone ever need that?"	21:49
troynt	I just think it is silly to diss Eclipse when your using VIM. Don't get me wrong VIM is nice, but when working with a lot of files and directories, it is just faster to click...	21:49
Voker57	Eclipse is slow as hell, that's the problem	21:50
kevlarman	never thought i'd hear that from a channel full of emacs users	21:50
	kevlarman ducks	21:50
troynt	kevlarman: ;)	21:51
Aides	Voker57: i exaggerated a bit, my point was it isn't the thing you need every day	21:51
doener	troynt: not for me. tabcompletion for the filename is a lot faster for me.	21:51
troynt	man a programmers IDE is like his religion .. lol it is very polarizing.	21:52
Aides	clicking is never faster, unless you have a spare hand for the mouse	21:52
kevlarman	that would rock	21:52
doener	Aides: I have a trackpoint, also for my desktop box, but yeah, it's still slower than typing	21:52
	→ softdrink joined	21:53
troynt	trackpoint the joystick on keyboard?	21:53
	mmk yea from google search	21:54
	→ naeu joined	21:54
troynt	doener: what keyboard do you have that gives you one for your desktop?	21:54
	Aides smells a trackpoint vs. touchpad holywar	21:55
doener	troynt: ibm ultranav keyboard	21:56
	troynt: the buttons for the trackpoint seem to suck though. They broke for me after about 6 months	21:56
	← a-priori left	21:56
doener	troynt: so I have to use the trackpoint mouse buttons, which unfortunately means no middle mouse button	21:56
	and the net seems to agree that the button breakage is a common problem	21:57
	tried cleaning the contacts, but that didn't help	21:57
	and it's kinda expensive, about 90 Euros IIRC	21:57
	If I didn't have one, I'd buy it again anytime though	21:57
troynt	looks nice, minus touchpad	21:58
doener	even with the broken "upper" buttons, it's still a lot better than using a mouse	21:58
troynt	I'm a trackpoint fan.. i wish more laptops had them	21:58
doener	yeah, I would love it a lot more if it wouldn't have the touchpad	21:58
	hate that thing	21:58
kevlarman	<insert xkcd link here>	21:58
doener	and I can't even turn the damn thing off, as the buttons stop working then	21:59
	(or I'm just to stupid to get the X.org config right)	21:59
	← naeu left	21:59
	→ naeu joined	22:00
doener	s/to/too/	22:00
	← naeu left	22:01
FreakGuard	doener, "emulate third mouse button"	22:02
	→ a-priori joined	22:03
doener	FreakGuard: yeah, but the trackpoint part has a native middle button	22:03
	FreakGuard: but as those buttons are broken, I need the touchpad buttons (2) with emulated middle mouse button	22:04
FreakGuard	hm	22:04
	← Aikawa left	22:04
doener	FreakGuard: and that way you can't disable the touchpad, which is pretty annoying when your wrist starts to move the mouse pointer	22:04
	← troynt left	22:05
doener	FreakGuard: ie. basically, I'd want the touchpad to be disabled, but still use the buttong	22:06
	s/g$/s/	22:06
	anyway, pretty offtopic by now	22:06
bar	what is the canonical way to get data from and push back to a cvs repository?	22:06
	or to put it another way, i want to create a reference git repo from which i can then pull	22:07
	→ Aikawa joined	22:07
	← a-priori left	22:13
	→ srid joined	22:14
	→ ashleyw joined	22:19
bremner	interesting git low level library in python: http://samba.org/~jelmer/dulwich/	22:19
doener	yeah, gitte added that to the wiki lately	22:19
	https://kerneltrap.org/mailarchive/git/2009/1/16/4750174	22:20
srid	bremner: will it be able to programatically do 1) commits 2) pull ?	22:21
	and 3) add	22:22
	← johnw left	22:24
	→ gebi joined	22:25
bremner	srid: dunno. I am just spamming the channel with random crap I found.	22:25
	→ seangrove joined	22:26
doener	bremner: lol	22:27
	→ bartocc joined	22:27
	drobbins → drobbins\|away	22:28
argnel	is it possible to split already committed changes as if git-add -p had been used?	22:29
doener	you can reset and re-do the commit(s)	22:29
	with --mixed (the default) the chanegs are kept in the working tree	22:29
	so you could do: git reset HEAD^; git add -p ...; git commit; git add -p ...; git commit	22:30
	← travlr left	22:30
	tchan1 → tchan	22:30
doener	and have the commit split into two	22:30
	← idl0r left	22:30
doener	you can reuse the commit message/date/etc. using the -c/-C option to git commti	22:30
	← Owner__ left	22:32
argnel	doener: so much to learn, so little time.. thanks :)	22:34
danderson	I'm assuming there is no way to make certain repo hooks executable on clone?	22:35
	← lordpil left	22:36
	→ lordpil joined	22:37
danderson	that is, I want to enable the pre-commit hook in repositories being cloned from the source. Is the best way to do that just to have a script in the repo and tell people to execute it after cloning?	22:37
doener	danderson: yep	22:37
	danderson: having hooks cloned and enabled by default is bad	22:38
	danderson: post-commit: rm -rf ~	22:38
	oops	22:38
danderson	sure, as with all other scms.	22:38
	Just wondering if there was some preferred way to get this done in git	22:38
doener	danderson: I'd probably have a second history (ie. another root commit) in a special branch that holds the hooks	22:39
	danderson: that way, it doesn't clutter the "real" project, but is still easily accessible	22:39
telmich	doener: maybe it even makes sense to keep them in the real project, if they are needed to work with it	22:42
	→ marvil07 joined	22:43
	→ Strogg joined	22:44
	← marvil07 left	22:48
	→ pyalot joined	22:51
	← bar left	22:54
	→ plasticine joined	22:59
	← ankit9 left	23:00
	← Ariens left	23:00
	→ EvilGuru joined	23:01
	← ashleyw left	23:01
	← foutrelis left	23:03
EvilGuru	I am having trouble with git-daemon and inetd. My line in /etc/intet.conf is: git stream tcp nowait nobody /usr/bin/git git daemon --inetd --verbose --export-all --interpolated-path=/home/freddie/public_html/%D /home/freddie/public_html/ but when I do git clone git://freddie.witherden.org/git/doublependulum.git I get an error in /var/log/daemon being: '/home/freddie/public_html/git//git/doublependulum.git': unable to chdir or not a git archive; w	23:04
	→ Beeet joined	23:06
Circuitsoft	git clone git://freddie.witherden.org/doublependulum.git	23:06
	→ TokyoDan joined	23:06
	← Weaselweb left	23:07
	← srid left	23:07
EvilGuru	Circuitsoft: Works a treat; I know. However, I would like the /git/	23:07
	← simosx left	23:07
EvilGuru	so git://freddie.witherden.org/git/doublependulum.git	23:08
	← Leefmc left	23:08
	→ simosx joined	23:08
Circuitsoft	I'm not actually sure if that's fully supported.	23:08
	I don't think I've ever seen it.	23:08
	If I were you, I'd rebuild git with debugging info, then start git-daemon under gdb or ddd and trace through it.	23:09
	→ ashleyw joined	23:09
	→ foutrelis joined	23:10
Ilari	EvilGuru: Wonder where it gets that '/git/' part from...	23:19
EvilGuru	Indeed	23:19
	← parasti left	23:20
	→ parasti joined	23:20
LiamH	Is there a way to bring back a previous version of a single file without checking out the entire (old) tree?	23:30
Ilari	LiamH: 'git checkout <revision> -- file'?	23:31
	← doener left	23:33
	→ doener joined	23:34
LiamH	Ilari: really? that sounds like what I want, but --file doesn't show up as an option in my git help checkout.	23:34
kevlarman	LiamH: not --file	23:35
danderson	where are hooks invoked from? The root of the repo?	23:35
	I mean, what's the working directory?	23:35
	← bartocc left	23:36
	→ grzywacz joined	23:37
LiamH	kevlarman: oh I see, proportional fonts are not a good idea on IRC clients	23:37
	→ bartocc joined	23:38
kevlarman	s/on irc clients//i :P	23:38
	← charon left	23:41
Circuitsoft	EvilGuru - Try cloning git://freddie.witherden.org/a/path/something.git	23:41
	See if you get /a/path/a/path or /a/a/path/path/ or something else	23:41
EvilGuru	Circuitsoft: Will do	23:42
	→ Yuuhi` joined	23:44
	← pyalot left	23:48
	← nick_h[litage] left	23:51
	→ Flathead joined	23:51
	→ dsal joined	23:52
	← LotR left	23:55
	← mbroeker left	23:55
	→ plediii joined	23:55
	← Yuuhi left	23:59

Logs Search ←Prev date Next date→

Channels Documentation