IRCloggy #git 2009-05-25

Provider of IRC logs since 2005.

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

	2009-05-25
	→ stuffcorpse joined	00:00
	← bobmcw left	00:00
	→ bobmcw joined	00:00
	→ kukks joined	00:01
	← Eulex_ left	00:03
	→ sfwc joined	00:06
	← stuffcorpse left	00:07
	→ WinterWeaver joined	00:09
	→ stuffcorpse joined	00:13
	elpargo\|lunch → elpargo	00:16
	→ gasull joined	00:17
	→ bremner_ joined	00:17
	→ dodo_the_last joined	00:18
	← stuffcorpse left	00:19
	→ stuffcorpse joined	00:19
	→ rettub_ joined	00:23
	→ scarabx joined	00:25
	→ Murr joined	00:25
	← neunon left	00:27
gasull	http://dpaste.com/47375/ . It seems to be a misconfiguration of the git server. How can I fix it? Thank you in advance.	00:27
	← bremner_ left	00:28
gasull	I meant: Hi, I get this error after git clone, modifying a file, git add ., git commit, git push.	00:28
cehteh	gasull: likely a permission problem	00:28
	did you push as user but once worked in that repository as root?	00:29
mugwump	yes, run 'git config core.sharedRepository group' on the server, plus: chgrp -R group DIR; find DIR -type d \| xargs chmod ug+rwX,g+s	00:29
	as a user in group 'group'	00:29
	you need to run 'git init --shared'	00:30
cehteh	(sometimes git could be more helpful with error messages)	00:30
mugwump	yes, probably	00:30
cehteh	--bare --shared likely	00:30
	← Radar left	00:30
	→ Radar joined	00:31
gasull	cehteh: probably	00:31
	← WinterWeaver left	00:31
cehteh	just try what mugwump saied	00:31
mugwump	saied .. is that lolcat?	00:32
gasull	cehteh, mugwump: OK	00:32
	It's a remote repo in EC2	00:32
cehteh	mugwump: eh what?	00:32
	← kate21de1 left	00:33
mugwump	i sai	00:33
cehteh	sayed?	00:33
	← stuffcorpse left	00:33
mugwump	I wonder when lolcat will be an acceptable language for written examinations in class	00:34
cehteh	my english is far from complete .. but maybe better than your german	00:34
mugwump	txt-speak is already in NZ for non-English papers	00:34
gasull	Is "git config core.sharedRepository group" literal? (I'm still a git noob)	00:34
cehteh	nope its not	00:34
mugwump	gasull: 'group' in that instance is literal - see man git-config	00:35
Gitbot	gasull: the git-config manpage can be found at http://git.or.cz/man/git-config	00:35
cehteh	ah yes :)	00:35
	well i dont do it that way, i like to make such things explicit	00:35
	→ bremner_ joined	00:37
	← sako left	00:38
	→ qrush joined	00:41
	→ stuffcorpse joined	00:42
	→ agoode joined	00:42
SRabbelier	mugwump: nowai	00:44
	mugwump: txt-speak?	00:44
gasull	The group in my gitosis-conf is called mypractice. I tried	00:44
	git config mypractice group	00:44
	error: key does not contain a section: mypractice	00:45
	← dodo_the_last left	00:45
	→ dodo_the_last joined	00:45
	← dodo_the_last left	00:46
mugwump	SRabbelier: http://en.wikinews.org/wiki/New_Zealand_students_able_to_use_txt_language_in_exams	00:47
	→ dodo_the_last joined	00:47
SRabbelier	mugwump: yea rite, you just wrote that yourself :P	00:47
mugwump	yeah, well I'd go and dig out the printed version of the paper from the national archives but you'd say I 'shopped it	00:51
	← jrmuizel left	00:53
	← bremner_ left	00:55
	← ereslibre left	00:55
	← scarabx left	00:57
	→ ekristen joined	01:00
ekristen	good evening, quick question for someone, I cloned a repo and did some work now I want to duplicate all the work and move it to a branch, then re clone the origin as the master branch	01:01
	→ ben_h joined	01:02
	→ ia joined	01:03
	→ loincloth joined	01:04
ekristen	anyone?	01:04
	← nh2 left	01:05
	← loincloth left	01:10
	→ dnyy joined	01:12
	abstrusething\|aw → abstrusething	01:12
dnyy	I'm pretty new to git and am using the jewler gem (generates the file structure for a rubygem, and creates a repo on github). Included in it is a .document file, and I'm not too sure what that's used for. I know about .gitignore, but not .document, and google isn't much help. Any idea what it's used for?	01:13
	← parasti left	01:15
	→ agenteo joined	01:21
	← eletuchy left	01:22
	→ memiux joined	01:23
	← ericindc_ left	01:23
	← k0001 left	01:24
	→ cemerick joined	01:26
	← gasull left	01:26
memiux	Hi, I wondering if it's possible to comparing two repositories or forks, for example: I want to know the differences between http://github.com/adamwiggins/rest-client AND http://github.com/technoweenie/rest-client	01:27
	→ eletuchy joined	01:27
jjuran	How can I omit origin/man and origin/html branches from gitk's display? I deleted the branches, but they come back after git fetch.	01:33
	abstrusething → abstrusething\|aw	01:33
cehteh	jjuran: delete them on the remote too	01:34
	git push remote :branchname deletes a remote branch	01:35
	← stuffcorpse left	01:36
tmz	memiux: if you have each repo as a remote in your local git clone, you can easily diff them. (github may have some way to do that via the web, but I don't use github.)	01:36
	→ stuffcorpse joined	01:36
memiux	thanks tmz	01:37
jjuran	cehteh: I don't think other git devs would like that :-)	01:37
cehteh	jjuran: huh?	01:37
	well then keep it	01:38
jjuran	cehteh: I'm viewing git.git in gitk. I'm porting it to a Unix-like system.	01:38
cehteh	these are then remote branches	01:38
	where you dont have control about	01:38
jjuran	I just don't want to see a bunch of "Autogenerated manpages" commits in gitk.	01:39
pasky	gitk --all ^origin/html ^origin/man perhaps?	01:39
jjuran	pasky: That works, thanks	01:40
	→ k0001 joined	01:40
	→ mithro joined	01:46
	→ scientes_ joined	01:46
	→ wsc joined	01:46
	→ qhoxie joined	01:50
	→ stragerLN joined	01:52
stragerLN	How do I reset a single file so it's the same as the one in the last commit, as if I hadn't edited it?	01:52
	← cemerick left	01:54
	→ cao joined	01:54
	→ kn1ttl joined	01:55
jjuran	stragerLN: git checkout HEAD^ path/to/file	01:55
	→ jhelwig joined	01:56
jjuran	stragerLN: If you've already committed, that is.	01:56
stragerLN	What I mean is	01:56
	I check out. I edit a file.	01:56
	Now I want to undo the edit and go back to the commit, without affecting other files.	01:56
	→ section31 joined	01:57
jjuran	stragerLN: git status tells you.	01:57
section31	How can i force clone a repo to an existing directory	01:57
stragerLN	# Changed but not updated:	01:57
	# (use "git add <file>..." to update what will be committed)	01:57
jjuran	stragerLN: git checkout HEAD path/to/file	01:57
stragerLN	K that worked	01:58
	Thanks. =]	01:58
	→ eletuchy_ joined	01:58
	→ mike_m joined	01:59
Arrowmaster	section31: git clone url dir	02:02
	→ cao1 joined	02:10
	← hacim left	02:10
	← knitt1 left	02:10
	→ jksz joined	02:11
	← eletuchy left	02:13
	← dnyy left	02:14
	→ kuadrosx joined	02:17
	← stragerLN left	02:17
	← Radar left	02:17
	→ perezd88 joined	02:18
	← perezd88 left	02:18
	→ cytrinox joined	02:25
	← kukks left	02:25
	→ girishr joined	02:26
	← cao left	02:27
	← jksz left	02:28
section31	I just installed git on a centos server and when i try to clone i get ndex-pack' is not a git-command, i googled and can't find much. I went down the list of dependencies and it appears i installed them all	02:33
	does anyone know which dependecy might be causing this?	02:33
	→ davidad joined	02:35
	← Murr left	02:43
	→ krawek joined	02:47
	← markelikalderon left	02:48
	→ Jinzhu joined	02:48
	← dodo_the_last left	02:51
cehteh	section31: index-pack or ndex-pack?	02:51
section31	index-pack	02:51
	→ tjafk1 joined	02:51
cehteh	that should be there	02:52
	what git version?	02:53
	→ sitaram joined	02:54
	→ dodo_the_last joined	02:55
	abstrusething\|aw → abstrusething	03:00
	← solydzajs left	03:01
	← SRabbelier left	03:01
	← abstrusething left	03:05
	→ lymeca joined	03:07
	← qhoxie left	03:08
	← zeno^way left	03:08
	← drewr left	03:08
	→ nanothief joined	03:12
	← LiamH left	03:20
	→ Minimiscience joined	03:20
	→ krawek_ joined	03:24
	← krawek left	03:24
	← hummingkid left	03:26
impulze	hm	03:26
	→ hummingkid joined	03:26
impulze	from the manual i cannot quite figure out how to keep git from adding the sender to the cc field in git-send-email	03:27
	is that possible? if so, how?	03:27
mugwump	impulze: seard for sendemail on man git-config	03:28
Gitbot	impulze: the git-config manpage can be found at http://git.or.cz/man/git-config	03:28
mugwump	search, even	03:28
	it might be an envelope destination only	03:28
	ah, there it is: sendemail.suppressfrom	03:28
impulze	yeah i saw that	03:28
sea-gull	hi all!	03:29
impulze	mugwump: oh shush! i was looking at suppress-cc	03:29
	oh perfect	03:30
	thank you mugwump	03:30
mugwump	no worries	03:31
	← k0001 left	03:33
	← dodo_the_last left	03:36
	→ Arafangion joined	03:38
	← Minimiscience left	03:39
Arafangion	Hey, how do I push to an empty, bare repo, using git 1.6?	03:40
	Doh! (Followed the damn instructions that were printed to stdout)	03:40
cehteh	hehe	03:40
	← davidad left	03:40
cehteh	in a few times, git gives useful diagnostics	03:41
	→ davidad joined	03:41
Arafangion	cehteh: Cut me some slack, I use Vista at work... Software on that system _never_ give _useful_ diagnostics, and the help's usually worse, to boot.	03:41
	Force of habit, if you may. :(	03:41
cehteh	Arafangion: you didnt noticed, some time before you joined i told the opposite that git diagnostics need to me improved in many cases	03:42
Arafangion	cehteh: Whatever the error is, it's better than "Error 61232"'	03:43
section31	cehteh: git version 1.6.2.2	03:43
Arafangion	cehteh: (ie, Microsoft Visual Studio)	03:43
section31	sorry just got your message	03:43
	cehteh goes to sleep	03:44
	→ qhoxie joined	03:44
	→ Swimming_Bird joined	03:44
section31	damn	03:44
cehteh	section31: no idea, but should be there, broken package, wrong installation .. whatever	03:44
	→ dodo_the_last joined	03:44
section31	k will try to install again	03:44
cehteh	maybe deinstall and build from source, not terribly complicated with git	03:45
Arafangion	Most complicated part of git's actually getting the help built and installed, but even that's easy.	03:45
cehteh	debians asciidoc package sux ..	03:46
	→ akitada joined	03:46
	← akitada left	03:46
cehteh	first the was no FOP in debian, then they dont package dblatex	03:47
	or at least the asciidoc style was missing	03:47
	n8	03:47
	← dodo_the_last left	03:49
section31	noooooooooooooooooooo git: 'index-stat' is not a git-command. See 'git --help'.	03:52
	after a reinstall from source	03:52
	what the hell is going on	03:52
	← qrush left	03:55
	→ qrush joined	03:56
section31	git: 'index-pack' is not a git-command. See 'git --help'.	03:57
	← qrush left	03:59
	← krawek_ left	04:03
section31	wow the latest git snapshot is fucked...used a diff one and that worked	04:03
almostautomated	section31: by 'diff one' do you mean stable?	04:04
	→ krawek joined	04:04
jjuran	almostautomated: Probably "different"	04:05
section31	false alarm	04:06
	→ Tuomas joined	04:08
	← Jinzhu left	04:12
	← krawek left	04:13
	← summerbug left	04:14
	→ savonarola joined	04:14
	→ krawek joined	04:15
	← savonarola left	04:18
section31	i can only get index-pack to run on root	04:21
	→ Jinzhu joined	04:23
	→ Paf joined	04:25
	→ Tuomas_ joined	04:32
	← hummingkid left	04:34
	cpg → cpg\|biab	04:36
	→ jceb joined	04:38
	← Tuomas left	04:39
	← rlb3 left	04:46
	→ Tuomas joined	04:49
	← resmo left	04:57
	← Tuomas_ left	04:58
	→ psankar joined	05:00
	← cloudhead left	05:02
	← sitaram left	05:05
	→ curvature joined	05:05
	→ Weaselweb joined	05:07
	→ dodo_the_last joined	05:15
	→ srid joined	05:17
	← dodo_the_last left	05:26
	→ bebarino joined	05:27
	→ Radar joined	05:30
	→ robinr joined	05:31
	→ hyperair joined	05:34
	← nanothief left	05:39
	← srid left	05:39
	→ jony joined	05:40
	← altrux left	05:45
	← jceb left	05:47
	← RayMorgan left	05:52
	→ Arafangi1n joined	05:53
	← Arafangion left	05:54
	← robinr left	05:54
	→ Martineo joined	05:55
	→ MrAwesome joined	05:55
	→ dodo_the_last joined	05:55
	→ ph^ joined	05:56
Martineo	i did some git reset head^^ and now when i want to pull it's out of date, what can i do?	05:57
	Arafangi1n → Arafangion	05:58
mugwump	git reset head@{1}	05:58
	will undo that	05:58
	Martineo: see man git-reflog	05:58
Gitbot	Martineo: the git-reflog manpage can be found at http://git.or.cz/man/git-reflog	05:58
mugwump	then use git-gui to stage the changes you made	05:59
	and commit etc	05:59
	← keithhub left	05:59
	→ keithhub joined	05:59
	← dodo_the_last left	06:00
Arafangion	mugwump: I've never been able to get git-gui to work. :(	06:00
	Not for large binary files.	06:00
mugwump	have you written to the list about it?	06:01
Arafangion	mugwump: No, as a matter of fact.	06:01
Martineo	git reset -- head@{1} doesn't work (with git-pull after that)	06:01
mugwump	why did you add the -- ?	06:02
	→ dodo_the_last joined	06:02
	← dodo_the_last left	06:02
Martineo	otherwise it doesn't work at all :)	06:03
	→ dodo_the_last joined	06:03
mugwump	ok, so 'git reset' changed the branch pointer and the index but not the working copy.	06:03
	ah, I meant HEAD@{1}	06:04
Martineo	i'm sorry	06:04
	but still doesn't work :(	06:04
	→ Tuomas_ joined	06:04
mugwump	Look buddy, doesn't work is a strong statement. Does it sit on the couch all day? Is it making faces at you? Does it want more money? Is it sleeping with your girlfriend? Please be specific!	06:08
offby1	mugwump: I resent your tone. I'm not the only one who wants more money	06:09
mugwump	Martineo: why don't you paste some output to gist.github.com so we can see what is going on	06:10
	offby1: taken from a purl factoid :)	06:10
	(purl = irc.perl.org infobot, perhaps one of the oldest in the world)	06:11
offby1	And Martineo's girlfriend is irresistible!	06:11
	→ summerbug joined	06:12
Martineo	:D	06:12
	← Tuomas left	06:13
Martineo	ah, screw it, i will clone it	06:14
mugwump	Martineo: your best bet would be to put some time aside to reading the git user manual or a nice online tutorial like http://www.eecs.harvard.edu/~cduan/technical/git/	06:15
Martineo	yeah, but until then i'll be dead…	06:16
	*until i solved my problem	06:16
mugwump	I think that idiom was lost in translation	06:16
Arafangion	Indeed, and my first language is English!	06:17
Martineo	whatever	06:17
	Tuomas_ → Tuomas	06:17
Ilari	Arafangion: It having problems with "large" binary files isn't very surprising. Git isn't designed to handle those very well.	06:18
	← jony left	06:19
Arafangion	Ilari: By "large", I mean an odd MB or so.	06:19
	Possibly tens of MB.	06:19
	But definetly no more.	06:19
	→ thiago_home joined	06:21
Ilari	Arafangion: What error message?	06:21
Arafangion	Ilari: Just refuses to stage the chunk, I can't stage anything, really.	06:22
	Ilari: Everything works, except that.	06:22
	→ roop joined	06:24
Ilari	Arafangion: Huh, how can one file that it can't stage prevent it from staging other files?	06:24
	← keithhub left	06:24
	→ keithhub joined	06:25
Arafangion	Ilari: I never said that.	06:25
	Ilari: Just the 'large' binary files that I care about.	06:25
	(Which are MS Word files... Not my choice, but bleh)	06:27
Ilari	Arafangion: For me, staging 10MB binary file using git-gui works.	06:27
	→ Eulex_ joined	06:27
Arafangion	Ilari: Must be a Vista thing.	06:27
Ilari	Arafangion: What error message you get?	06:28
Arafangion	Ilari: Nothing.	06:28
	Ilari: It's as if it's ignoring my clicks for such files.	06:28
	→ drizzd joined	06:28
Ilari	Arafangion: For me, staging 10MB file takes few seconds while it calculates the SHA and writes the object.	06:29
Arafangion	I doubt it's even doing that.	06:29
	Anyway, it's a low priority... I'm a command line user anyway, and I'll soon be going back to text files.	06:29
	← piyush left	06:29
	→ angerman joined	06:31
	→ sunoano joined	06:33
	← drizzd left	06:38
	→ giallu joined	06:38
	← memiux left	06:41
	→ godog joined	06:46
	← stuffcorpse left	06:48
	→ stuffcorpse joined	06:48
	← dodo_the_last left	06:48
	← summerbug left	06:49
	→ Martineo_ joined	06:51
	→ lohrun joined	06:53
	← johan-s left	06:53
	← davidad left	06:53
	→ priidu joined	06:54
	→ robinr joined	06:54
	→ davidad joined	06:57
	→ galderz joined	06:57
	← almostautomated left	06:58
	→ meyering joined	06:58
	← robinr left	07:03
	→ summerbug joined	07:05
	← Martineo left	07:07
	→ seangrove_ joined	07:07
	→ doener joined	07:10
	→ Stravy joined	07:12
	← angerman left	07:14
	← Eulex_ left	07:15
	→ mpoullet\|work joined	07:17
	← thiago_home left	07:17
	→ piyush joined	07:19
	→ lucsky joined	07:19
mpoullet\|work	good morning	07:20
	← seangrove left	07:21
mpoullet\|work	how to publish a git tag with git svn, e.g. the same as git push origin tag <tag_name>?	07:21
	→ johan-s joined	07:22
	Martineo_ → Martineo	07:23
	→ ciskje joined	07:23
	→ xrfang joined	07:24
	→ Tuomas_ joined	07:24
xrfang	hello, is there any tool that I can visualize a git tree, and find out the "position" of the working copy on this tree?	07:24
	→ hoohah joined	07:24
hoohah	hello. are any of you familiar with how github ssh works? I'm trying to define the github host in .ssh/config, but am unsure about the exact settings (eg hostname,user etc.)	07:25
	← summerbug left	07:25
hoohah	basically i want to allow compression, and to have it use a different encryption algorithm	07:26
	→ v3s4 joined	07:27
roop	xrfang: did you try gitk or qgit?	07:28
	→ khmarbaise joined	07:29
xrfang	roop: I used gitk and giggle, but I don't see how I can visualize a "tree", i.e. a whole repo with branches etc.	07:29
	as I am doing some testing (bug reporting) for wine, take that as an example, I don't know what revision I am on while doing a bisect	07:30
	→ thehcdreamer joined	07:30
xrfang	i.e. gitk reports the actual revision I am working on, while giggle reports the HEAD of master	07:30
	I get lost (I just started using git a few days).	07:30
	→ merlin83 joined	07:31
	← girishr left	07:31
	← Tuomas left	07:32
thehcdreamer	Hello, I pulled from the wrong branch on my repo, and now I would like to rollback to the previous commit. How can I do that given the commit's id? I guess I should use git-reset	07:32
	← hyperair left	07:32
	→ eevar2 joined	07:33
roop	xrfang: so you want to checkout a commit, and want to see which point in the branch you have checked out. while gitk cuts off the branch at the checkout point. is that right?	07:33
	→ ceyusa joined	07:34
roop	thehcdreamer: did the pull do a merge (it does by default, unless you'd asked it to rebase)	07:34
xrfang	roop: exactly, I don't know what is left on the road from my working copy to the head, also I don't know if it can show branches like a tree, and which branch I am on (because wine only have master as I can see)	07:34
thehcdreamer	roop: yes but it failed multiple times	07:34
	roop: I mean with conflicts	07:35
eevar2	possible to move parts of a commit out into a new branch?	07:35
mpoullet\|work	my git-svn setup doesn't seem to provide the git svn tag/branch commands, any idea why: http://pastebin.com/d7e06bd8a ?	07:36
roop	thehcdreamer: if it did do a merge (conflicts resolved manually, i presume), you can git checkout <yourbranch> (it should be a the merged commit now) and then a git reset --hard HEAD^ (meaning: put the repo one commit earlier than present HEAD)	07:36
	← Martineo left	07:36
thehcdreamer	roop: ok I'm trying	07:37
doener	mpoullet\|work: git --version?	07:37
mpoullet\|work	doener: hi, git version 1.6.0.4	07:38
doener	mpoullet\|work: too old ;-) the svn branch/tag command got added in 1.6.2	07:38
thehcdreamer	roop: it worked thanks	07:38
mpoullet\|work	doener: ok, thanks, I'm using ubuntu 9.04 and it seems that's the default version, it sucks...	07:40
	→ rshann_ joined	07:42
	→ cao joined	07:43
	← cao1 left	07:43
v3s4	quick Q: is it possible to clone two repos in the same working dir? I.e. a repo containing C utility macros .h, another repo containing c++ utility macros .hpp files and my own working files from yet another repo	07:43
	three repos actually	07:43
	ok, so it may be possible but is it sound/recommended?	07:43
Grum	why not just symlink? or use submodules?	07:44
	← cao left	07:44
	→ cedricv joined	07:44
	← tango_ left	07:45
	← khmarbaise left	07:45
v3s4	Grum, (on symlinks) because when building a release tarball I must follow symlinks	07:45
	→ khmarbaise joined	07:45
v3s4	Grum, but I'm starting reading on submodules right away	07:45
roop	xrfang: you can see all branches by saying --all	07:46
Grum	you could always make your 'release tarball creation mechanism' smart enough to do the right thing :)	07:46
	← sdboyer left	07:46
	← Tuomas_ left	07:46
xrfang	roop: this is a gitk option?	07:46
roop	xrfang: like in gitk, go to view->new view->Commits_to_include: --all	07:46
xrfang	I will try... thanks	07:47
roop	xrfang: maybe you can create one view that will show with --all, and another that will show with HEAD, so you can toggle between the two views easily.	07:47
	← agenteo left	07:48
	→ cao joined	07:48
v3s4	Grum, I agree but I also must make my git push mechanism smart enough so it pushes to the repo local modifications to the files pointed by the symlinks... and for that I must travel to another cloned repo which is not elegant IMH(and n00b)O	07:48
	→ __iron joined	07:48
xrfang	roop: I am very new to git, need some time to study :P thanks a lot... however, could you please tell me how to do these 2 things by git command line:	07:48
	1) know which branch I am on	07:49
	2) know relative position in the current branch	07:49
Grum	git branch lists a '*' for the current branch	07:49
xrfang	i.e. In a project I may forget that I have rewind to some previous edition	07:49
Grum	relative to what?	07:49
xrfang	relative to HEAD	07:49
Grum	git log HEAD..	07:49
xrfang	if I reset or revert to a previous version, then continue work	07:49
	what will happen ? will a branch be created by default?	07:50
Grum	euh? :D	07:50
xrfang	also I am confused about the differences between reset and revert, and also checkout :(	07:50
Grum	checkout puts the files from a certain version in your working dir, reset moves 'HEAD', revert creates a patch -r changeset for the commit you point at	07:50
	reset has 3 modes btw, it can also muck with your files in the workdir	07:51
	(the difference between --sort, --mixed and --hard)	07:51
	← khmarbaise left	07:51
	→ khmarbaise joined	07:51
xrfang	Grum, I think I need more practice to fully understand the diff... is this a place for asking absolutely newbie questions?:)	07:52
	← cao left	07:52
Grum	you can ask and people might answer yeah :)	07:53
	you might get told to just read the docs however ;)	07:53
xrfang	the doc is a bit scaring I just read git magic, and now plan to read git community book will be back :)	07:54
roop	xrfang: you use checkout to set the files in your dir to some prev rev. if you want to branch off from there, you can use a -b branch with git checkout.	07:55
	→ cao joined	07:55
Grum	its wise to know a bit how git works on the internals, its so simple but it will make lots of other things easier to understand :)	07:55
	← thehcdreamer left	07:55
roop	xrfang: reset is to change the current branch itself, like to cut off some commits	07:55
	xrfang: and dont worry about revert for now :)	07:56
xrfang	roop: ok, I had some vague idea of reset... for now I would like to try branching which is said to be the "core" of git.	07:56
	e.g. I have a tree, with rev 1-2-3-4-5	07:56
	now I checkout rev3	07:56
Grum	roop: revert is actually by FAR the easiest to explain :p	07:56
xrfang	then I work on it then commit	07:56
	what will happen?	07:57
roop	Grum: agree, but i dont think he needs it now, that's why :)	07:57
Grum	if you checked out a specific commit you will have a branch without a name :)	07:57
eb_	hm Grum... 'it's so simple'? not that I know anything about git's internals, but I am looking at the git-internal-docu right now, and it says: If you.re tired of terse man pages or academic white papers, you.ll enjoy more than four dozen colorful diagrams that clearly explain the complicated inner workings of Git.	07:59
Grum	eb_: yes it is that simple	07:59
roop	xrfang: in git, a branch just points to a leaf node in the tree - you trace the branch by tracing it's parents. so in your case, when you created a 3.1, it would've committed fine, but there's no way to reach it from a branch.	07:59
xrfang	actually I don't need most stuff, I just need a smooth transition from svn, :)	07:59
	roop: this means, that will create 3.1, but not 6?	08:00
Grum	a 'branch' is nothing more than a commit with a ref pointing at it with a name like: refs/heads/*	08:00
eb_	xrfang: did you see the svn-git crashcourse?	08:00
Grum	xrfang: you cant smoothly transfer from svn	08:00
	you need to reboot you brain for that :)	08:00
xrfang	eb_, not yet.	08:00
Grum	+r	08:00
eb_	xrfang: there it is: http://git.or.cz/course/svn.html	08:00
roop	xrfang: anyway, so git checkout and change it what you dont do. what you do is, git checkout someplace, and if you want to branch from there, do a git branch, checkout the branch and proceed	08:01
xrfang	Grum: yes, git is too powerful, there are something I didn't imagine from svn, but I doubt I need it right away? e.g. with git it is very easy to erase history:)	08:01
Grum	that is actually stuff you do want	08:01
	that should even be the whole reason to start using git, having local control	08:01
	→ Tuomas joined	08:01
	← giallu left	08:02
	→ Modius joined	08:02
Grum	you dont need to know all of it, you can learn on the way but that is where the major advantage is over for example svn	08:02
roop	xrfang: if your repo is svn, try git svn. you can't meddle with the svn repo however you try, so you're safe.	08:02
Grum	roop: you can meddle with the svn repo?	08:02
	if you dcommit you push stuff over	08:02
doener	xrfang: you have two kinds of history with git. Your local history, which is just _yours_ and you're free to modify it, and the part that is public, and that is generally off limits for modifications	08:02
	→ thebohemian joined	08:03
	← cao left	08:03
Grum	modifications that are non 'fast forward' extensions	08:03
xrfang	roop: ok, what I would like to immediately try with git is to see if it can keep track of a debug version of my application, so that I can use it whenever I need	08:03
roop	Grum: yup, but you can't delete history	08:03
Grum	roop: and THAT is the problem	08:03
	if you make a change you dont want you are rather fucked with svn	08:03
xrfang	doener: that is another issue I am interested and need to figure out : how to setup a git server (preferrably without using something like apache)	08:03
roop	Grum: but you can delete local history because the local repo is git	08:03
Grum	xrfang: google for 'gitosis'	08:04
doener	xrfang: just ssh+git-daemon, or gitosis	08:04
xrfang	just installed that from apt-get :) thanks	08:04
	→ robinr joined	08:04
Grum	oh good luck getting it to work then hehe	08:04
	last time i checked debian kinda mangled that package about :p	08:04
roop	xrfang: didnt get it. why need git to get an application whenever you need?	08:05
thebohemian	hi, while cloning a repository I get the following error: "warning: remote HEAD refers to nonexistent ref, unable to checkout."	08:05
	what should the repo maintainer do to fix this?	08:05
xrfang	roop: that's a bit hard to explain. I mean with svn, I do not keep track of the debug version. whenever I need to debug I add statements into the latest version of my app and send it to customer	08:05
Grum	i'm tempted to link the non-bare faq :)	08:05
	← krawek left	08:06
xrfang	with git, I hope to keep a debug branch, and always be able to merge latest development to the debug branch from master	08:06
	or, do you call this a "rebase"?	08:06
Grum	merge	08:06
	← hoohah left	08:06
Grum	rebasing moves commits	08:06
roop	xrfang: right. rebase is what you need here.	08:06
Grum	or actually sorry, it moves the head of the chain you rebase	08:06
	roop: its not ;)	08:06
	cherrypick or merge	08:06
xrfang	:S confused...	08:07
roop	xrfang: is your central repo on svn?	08:07
xrfang	currently yes, but I hope to get rid of svn while I am comfortable with git	08:07
roop	Grum: obviously, we prefer different ways of working with git :)	08:07
	xrfang: these debug changes, how many are they? 1,2 or 20 and stuff?	08:07
Grum	roop: i guess it matters which side you are going to rebase on	08:07
	→ cao joined	08:08
roop	xrfang: and these debug changes will just add printfs to the code, huh?	08:09
Grum	xrfang: there are usually multiple ways to do the same thing	08:09
xrfang	roop: yes	08:09
	Grum, I hope I don't get confused... :)	08:09
Grum	roop: in that case if those commits are in a single branch somewhere you could indeed rebase that branch to the latest commit	08:10
xrfang	also, in my entire history of using svn, I feel the idea of merge virtually useless,	08:10
	maybe because the project is so tiny comparing to Linux or wine etc.	08:10
Grum	xrfang: how many people are working on your project?	08:10
doener	maybe because merging in svn sucks ;-)	08:10
Grum	that is a given doener ;)	08:10
xrfang	grum, currently only me, but I expect 2.	08:11
	used to be 3, :)	08:11
mpoullet\|work	doener: I've installed the latest packages according to http://blog.avirtualhome.com/2009/05/07/git-163-packages-for-ubuntu-jaunty/ and now everything works fine!	08:11
doener	mpoullet\|work: good :-)	08:11
	→ Sho_ joined	08:11
	→ sitaram joined	08:12
Grum	xrfang: that is something you will unlearn rather fast :)	08:12
	(not using branches that is)	08:12
	(and therefor merging)	08:12
	→ psoo joined	08:12
xrfang	Grum, I love to use that, but don't have this habbit in svn. Another huge issue for me is that with svn there are a lot of garbage (wip, broken versions) in my svn repo	08:12
roop	xrfang: this is my recommendation: use git svn (read up on how to set it up). then, create a debug branch and add your changes there.	08:13
xrfang	with git all that can be solved, but I feel extremely horrified of garble the history :)	08:13
Grum	garbage? you mean you made a change and later another change that 'undid' the change?	08:13
roop	xrfang: whenever you want to sync with the server, do 'git svn rebase' and then 'git rebase master debug'. tada.	08:13
xrfang	Grum, no not undid, but work-in-progress... we used to do a lot of commits during development of a feature, and there are inevitably garbages.	08:14
	← scientes_ left	08:14
Grum	what do you consider garbage?	08:14
xrfang	roop: I will try that	08:14
doener	xrfang: by default, "git push" will refuse to accept non-fast-forward updates (i.e. such that modify the public history)	08:14
Grum	it appears as you say history is garbage	08:14
xrfang	I may also need to ask questions from reading gitmagic... :)	08:14
	Grum: no not at all	08:14
doener	xrfang: so it's unlikely that you really kill any history that's important	08:14
	Grum: he means that with svn, you can't do WIP commits that you later squash to get a single good commit, for example	08:15
xrfang	I feel that is a garbage because there are so many interim versions, and with git, a feature can be a rev (use some tricks like reset, revert etc to manipulate the history). that is what I just learned	08:15
doener	Grum: instead you just have a bunch of garbage revisions	08:15
	→ struberg joined	08:16
xrfang	doener: right that's what I mean, after that feature is done, I hope to merge these WIP commits to one big changeset which dedicates to this feature	08:16
doener	xrfang: well, one commit per feature usually gets "too big to be useful". It's best to trim your commit to be "one logical change that works"	08:17
	→ hyperair joined	08:17
doener	xrfang: besides speed, one of git's biggest advantages compared to svn is that you can actually make use of the history you produced	08:18
xrfang	doener, I agree..	08:18
	I don't understand "make use of the history"	08:18
doener	xrfang: with svn, you basically have a code dump. Even getting a log is slow as hell, and more advanced stuff is basically impossible.	08:18
	xrfang: for example you can't (AFAIK) even get something like "git log -p -- <somefile>" (log with patches) from svn	08:19
	→ tango_ joined	08:20
xrfang	patches, is what I rarely use, although I am familiar with it because of using Linux... When we use svn, we don't produce something called a changeset :p	08:20
	most of the use of svn is to manage windows based projects, but the svn server runs on linux	08:20
doener	xrfang: and git has a lot more to offer. "git blame" works across merges (svn's blame breaks). "git log -S<string> -- <file>" let's you find the commit that added/removed <string> to/from <file>	08:20
	xrfang: so maybe you see a weird "if (foo <= bar)" or so, and have no clue what that means	08:20
xrfang	doener: I think you are getting too advanced here :)	08:21
doener	xrfang: with svn, there's not much you can do. With git, you might do "git log -S'(foo <= bar)'" to find the commit that added that statement	08:21
	xrfang: and the commit message should then tell what that commit was good for, and you should get enough information out of that to understand that piece of code	08:22
	xrfang: IOW you can use the history to find information about the code	08:22
xrfang	ok, I got your idea,	08:22
	← stuffcorpse left	08:23
	→ the_zeno joined	08:23
	→ stuffcorpse joined	08:23
hyperair	doener: more like git blame/praise	08:23
xrfang	there are many many places that I dont' understand until I actually have these requirements.... for example	08:24
doener	hyperair: yeah, blame could also work, but for such cases, I prefer log -S	08:24
xrfang	in the git-magic book	08:24
	it says to "git stash"	08:24
	then work on some bug fix	08:25
	then git stash apply	08:25
	this works on the "master", without an additional branch, right?	08:25
doener	hyperair: being more specific, that function might be recursive now, and maybe had a "while (foo <= bar)" earlier. As the -S only searched for the condition, that would find the right commit. Blame would show the commit that changed it from recursive to iterative ;-)	08:25
	→ aspotashev\|eeepc joined	08:26
hyperair	hmm right. =)	08:26
	← kuadrosx left	08:26
	← hyperair left	08:29
	→ hyperair joined	08:32
	← piyush left	08:37
	→ piyush joined	08:37
	← korren left	08:37
	→ bx2 joined	08:38
	→ k0001 joined	08:39
	→ abstrusething joined	08:41
	← sdboyer-laptop left	08:44
	→ Modius_ joined	08:45
xrfang	ok, I try to continue with the bisect test on wine. I have wine-git on my harddisk, how do I check the revision the working copy?	08:45
wereHamster	xrfang: do you want to display the version at which your working copy is?	08:48
	← Modius left	08:49
xrfang	wereHamster: exactly	08:49
	also, I would like to be able to checkout any version by its tag, e.g. wine-1.1.21 etc.	08:50
	→ gilimanjaro joined	08:51
wereHamster	xrfang: git describe	08:53
xrfang	ok, thanks	08:53
wereHamster	xrfang: git checkout -b your-local-branch-name-of-1.1.21 wine-1.1.21	08:54
	→ prateeksaxena joined	08:58
	prateeksaxena → prtksxna	08:58
xrfang	-b means create a branch?	08:59
doener	yup	08:59
	← k0001 left	09:00
doener	"git checkout -b <name> <start>" is a shortcut for "git branch <name> <start> && git checkout <name>"	09:00
	→ markelikalderon joined	09:00
	→ zwu joined	09:01
	← markelikalderon left	09:01
	→ seangrove joined	09:02
	→ droidcore joined	09:02
	→ gitte joined	09:03
	→ balint joined	09:04
	→ giallu joined	09:06
	← bobesponja left	09:06
	→ ankit9 joined	09:12
	← stuffcorpse left	09:13
	→ stuffcorpse joined	09:14
	← ankit9 left	09:14
	→ ankit9 joined	09:15
	→ marc_kdab joined	09:15
	← seangrove_ left	09:16
	→ aziz joined	09:16
	→ s0ber joined	09:17
	→ Eulex_ joined	09:18
xrfang	doener: what is <start> it is just the word "start"?	09:19
	but for the purpose of wine dbg, I can also do without branching, right?	09:20
	← sunoano left	09:20
	← rshann_ left	09:28
sitaram	<start> was intended as the notional name for some branch or commit or maybe tag (a "ref" in git terms)	09:30
	← eevar2 left	09:31
	← struberg left	09:36
	→ ankit9_ joined	09:39
	→ intripoon joined	09:41
	← Modius_ left	09:42
	→ Modius joined	09:42
	→ sadleder joined	09:42
wereHamster	xrfang: it can even be the sha of a commit	09:43
	← Paf left	09:44
	→ nh2 joined	09:48
nh2	I committed to a local repository with the wrong email address, so I want to rewrite history like in here: http://coffee.geek.nz/how-change-author-git.html	09:49
	but there is that umlaut (ü) in my name and git complains it was not UTF-8 (but afaik it is?!)	09:49
roop	xrfang: amazingly, gitk --all works too :)	09:50
nh2	any idea how to tell git it might please also properly consider ä,ö,ü,ß and so on?	09:51
	← ankit9 left	09:52
	→ rudi_s joined	09:53
	→ naeu joined	09:54
	→ dodo_the_last joined	09:54
	→ tbf joined	09:59
	← sadleder left	10:01
	← Radar left	10:01
xrfang	roop: I don't know what you mean here...	10:02
	→ j416 joined	10:03
	← cao left	10:04
	→ cao joined	10:07
roop	xrfang: you asked me, right if you can pass --all to gitk directly, like "gitk --all"? apparently, you can.	10:11
	→ WinterWeaver joined	10:11
xrfang	roop: ok, thanks.	10:11
	actually I have already tried that.	10:12
	;)	10:12
	→ alanhaggai joined	10:18
alanhaggai	Hi. Is there any way that I can rewrite a password which is in a file back in history?	10:19
	→ hatseflats joined	10:22
hatseflats	morning' everyone	10:22
	this may be a very stupid 'man git' question, but I have a local change which I want to revert, but when I 'git revert', it fatals out with a file-not-found error	10:23
	← balint left	10:23
rudi_s	alanhaggai: If you haven't pushed since the change have a look at man git-rebase (have a look at -i).	10:25
Gitbot	alanhaggai: the git-rebase manpage can be found at http://git.or.cz/man/git-rebase	10:25
	abstrusething → abstrusething\|aw	10:25
doener	hatseflats: "git revert" is to revert commits	10:25
hatseflats	I figured I was getting terminology mixed up with subversion	10:25
alanhaggai	rudi_s: Thanks. I will check it out.	10:25
hatseflats	still transitioning	10:26
doener	hatseflats: if you want to get a "pristine" copy of a file from a commit or the index, use "git checkout"	10:26
rudi_s	alanhaggai: np	10:26
doener	hatseflats: e.g. "git checkout -- file" to get file from the index	10:26
	hatseflats: or "git checkout HEAD -- file" to get file from HEAD (into the index and working tree)	10:26
hatseflats	that did the trick, thanks doener.	10:26
	abstrusething\|aw → abstrusething	10:26
doener	hatseflats: IIRC "git revert $commit" is like "svn merge -c -<rev>"	10:27
hatseflats	oh jeez, I suppose my tree is messed up then	10:27
	ah well, that's what we've got revisions for	10:27
	hatseflats checks	10:28
	← bebarino left	10:31
	← Jinzhu left	10:31
	→ azumanga joined	10:32
	→ parasti joined	10:33
azumanga	I'm looking for a good guide for a svn-like process (central repository all developers clone + push and pull from). This is just about working, but we find some weird things happening, and I can't find a good guide that specifically deals with this setup.	10:35
	→ ntoll joined	10:36
sitaram	what sort of weird stuff? I use it here for many projects...	10:36
azumanga	Commits that look like:	10:36
	commit 08534716d937a11a395b749c0ec27ccba51666a2	10:37
	Merge: 51c6003 8a1c3de	10:37
	Author: Christopher [email@hidden.address]	10:37
	Date: Fri May 22 18:11:36 2009 +0100	10:37
	Merge branch 'master' of https://<servername>/repositories/minion	10:37
doener	well, Christopher pulled	10:37
azumanga	Appearing in everyone's log, which I wouldn't expect	10:37
doener	and his branch and the "central" branch diverged, so a merge commit was created	10:38
	after all, pull is fetch + merge	10:38
ricky	Try git pull --rebase, perhaps?	10:38
azumanga	doener: But, it's confusing (to me) having such things in the log.. they don' tell you what actually happened or anything	10:39
Weiss	yes they do	10:39
doener	azumanga: they tell you that a merge happened	10:39
Weiss	they record the development history	10:39
sitaram	azumanga: when in doubt, 'gitk --all' :-)	10:40
azumanga	OK, so is it expected that in a large project, the main development branch will have many commits that look like that, from many different developers	10:40
	→ icwiener joined	10:40
Weiss	in a large project, it's often expected that a contributor will "clean up" their branch before committing, by rebasing their work against the latest "tip" of the central branch	10:41
azumanga	Ah, so rebasing come up again!	10:41
	So we should be doing git pull --rebase, instead of just git pull?	10:41
	(Current using git pull / git push)	10:41
doener	azumanga: usually around one merge commit per feature. You create a feature branch, work on it, merge it, done.	10:41
Weiss	that gains a bit of clarity, but loses a bit of history. often, it's worth recording the fact that the contributor's workflow went "write something, test it, merge the latest central branch in, fix things, re-test, do some more work, etc etc etc.. commit"	10:41
	abstrusething → abstrusething\|aw	10:42
doener	azumanga: sometimes you get a fast-forward (no merge commit), so on average you should get less than one merge per feature	10:42
	azumanga: doing "git pull" all the time like you did "svn up" is harmful, as it just clutters the history with sh!tload of perfectly useless merge commits	10:43
azumanga	doener: So you would suggest git pull --rebase instead?	10:43
	Or, just not pulling at all in short-lived branches?	10:43
	I have to pull before I push (either pull or pull --rebase)?	10:43
doener	azumanga: I use "git rebase -i <opts>" to keep my history clean, and _sometimes_ I use rebase to "forward-port" my stuff (primarily when working as a by-email contributor or with git-svn)	10:45
	azumanga: the important point is just "don't merge when you don't have to".	10:45
	→ fujin joined	10:45
doener	azumanga: about 99% of the time, it's makes no difference at all whether you merge the stuff the others did, because they simply touched a part of the code that has no meaning for the stuff you're working on	10:46
azumanga	I would have thought frequent merging to find conflicts would be a good idea	10:46
	abstrusething\|aw → abstrusething	10:46
sitaram	a rebase finds the same conflicts	10:47
azumanga	Anyway thanks, at least I have a better idea what's going on, and will certainly look into rebasing	10:47
j416	looking at the git project repository, there is a certain workflow used (branches like maint, next, todo etc.), where can I learn how this workflow works?	10:47
doener	you can frequently do integration merges on a temporary testing branch, just don't keep all that crap in your actual branch	10:47
	sitaram: those and more! ;-)	10:47
	→ lfoppiano joined	10:47
sitaram	j416: man git-workflow (or some such)	10:47
j416	oh	10:47
	thank you!	10:47
doener	man gitworkflows	10:47
Gitbot	doener: the gitworkflows manpage can be found at http://git.or.cz/man/gitworkflows	10:47
lfoppiano	Hi all. here's my problem...I have no idea about what solution ... I have two branches.	10:47
	master and branch1	10:47
	in branch1 there are several files I modified starting from master version	10:48
j416	great	10:48
	thank you doener	10:48
lfoppiano	and I want to merge only 1 file (a subset of branch1 branch) into merge...	10:48
	how I can do?	10:48
	← eletuchy_ left	10:48
lfoppiano	I have no idea...I saw only how to merge between branches...	10:48
	but not on single files	10:48
	→ Fullmoon joined	10:50
	← __iron left	10:51
	← prtksxna left	10:54
	→ Moult joined	10:56
azumanga	doener: Could you suggest your short workflow for "branch from origin/master, make some changes, put back onto origin/master" ?	10:56
Moult	i have an account for a shared hosting server. is it possible to put my git repository on that site? what are the requirements?	10:57
alanhaggai	lfoppiano: I think you want to cherry pick the commits. (If I am right.)	10:57
	→ francois__ joined	10:57
	← khmarbaise left	10:58
	→ mtvee joined	10:58
doener	azumanga: for a really short-lived branch: git checkout -b new_feature origin/master; work; git rebase -i <whatever> (to clean things up); git pull; git push origin HEAD:master	10:58
	→ khmarbaise joined	10:59
lfoppiano	thanks alanhaggai I'll search it ;-)	10:59
	← hyperair left	10:59
azumanga	doener: What's the diff between git push origin HEAD:master and git push origin master (which seems to work)	10:59
	→ hyperair joined	10:59
doener	azumanga: "git push origin master" pushes master	10:59
	azumanga: but in the above example, all the new stuff is in "new_feature" so I push the checked out branch (identified by HEAD, and is new_feature in this case) to the remote's "master"	11:00
	← khmarbaise left	11:00
doener	azumanga: the refspec for push is "source:destination", "master" is a shortcut. If a branch head refs/heads/master exists locally, then "master" as a push refspec is the same as "refs/heads/master:refs/heads/master"	11:01
azumanga	Ah, I see! that shortcut has been confusing me I think, I'll go and have another go.	11:02
	One thing is for sure, this is much more complex than svn :)	11:02
	← xrfang left	11:02
	→ kn1tt1 joined	11:02
doener	azumanga: well, there's more that you can do. But overall, most stuff is pretty logical. Sadly, a lot of stuff on the net doesn't explain anything but just has "recipes"	11:03
	← rudi_s left	11:03
doener	azumanga: e.g. "git pull" is a convenience command, and one should know what the commands that it wraps (fetch + merge/rebase) do first, and then learn about the convenience stuff	11:04
azumanga	I think that's my problem, I'm sure how branches interact is actually quite simpler, and easier tha learning some simplified fragile patterns	11:04
	Yes, I think telling people to fetch + rebase is generally what they want	11:05
	and avoids the possibility they forget to --rebase their pull	11:05
	→ reel_ joined	11:05
doener	not really, it depends on where in the developer hierarchy you are	11:05
	(and how you want your history to look)	11:05
azumanga	We are a fairly small bunch of developers, who all commit frequently with no control / central developer to a project. Fairly close to standard svn behaviour	11:06
	→ rshann_ joined	11:07
azumanga	But we are now wanting to move into medium-length branches, which svn is really, really terrible at	11:07
	← Arafangion left	11:08
alanhaggai	lfoppiano: You are welcome. (Sorry for the late reply. I was away. :-)	11:10
lfoppiano	;-)	11:10
	→ mbroeker joined	11:10
	→ cemerick joined	11:11
	← kn1ttl left	11:12
	→ MisterN joined	11:12
	→ kraymer joined	11:12
sitaram	doener: you said (back up ^^ a bit) that a rebase catches more conflicts than a merge. Is this because a rebase goes step by step, and if there are 2 compensating conflicts (canceling each other out), it'd needlessly do both?	11:15
	→ kraymer_ joined	11:15
sitaram	I always thought rebase will show less conflicts because it is doing it one commit at a time	11:15
	← thebohemian left	11:16
doener	sitaram: yep, let's say upstream changes 123 to 456	11:17
	→ bschindler joined	11:17
doener	sitaram: your commits might first change 123 to 666 and in a later commit you change 666 back to 123 (for whatever reason)	11:18
	sitaram: a merge would see "no change" on your branch, and just accept the change from upstream	11:19
	sitaram: a rebase would conflict, maybe even twice	11:19
	→ markelikalderon joined	11:20
sitaram	ok so I guessed sort of right :)	11:20
	→ jaeckel joined	11:21
sitaram	but not counting cases like this, I thought resolving conflicts one commit at a time is easier than resolving all of them at once... Is that not true?	11:21
	(like 10 commits, one each, sum of effort/complexity, is less than effort/complexity of one monolithic merge)	11:22
doener	sitaram: it really depends on what you have changed.	11:22
	← kraymer left	11:23
doener	sitaram: for example, about a month ago, I basically rewrote some largish function (~100 lines of code), got splitted into 3 functions, and the code was quite different afterwards	11:23
	← francois__ left	11:23
doener	sitaram: rebasing caused a conflict for basically every commit	11:24
Pieter	also, with rebases, after you have fixed a conflict, the next patch might not apply anymore, so you'll have to fix conflicts there that you'd never encounter if you'd just merged	11:24
j416	how can i show commits on a branch that are local to that branch? For example, if I have a branch with a lot of merges, how can I have it to just show the merge commits, together with any branch-local commits?	11:24
sitaram	Pieter: that makes sense too...	11:24
doener	sitaram: merging was trivial, because I could look at the upstream commit that changed some piece of logic from A to B, and I could just replicate that semantic change in the new code	11:24
wereHamster	~.	11:24
sitaram	doener: but was that manual ("look at...")?	11:25
	wereHamster: didn't work; hit enter first then ~.	11:25
	:)	11:25
doener	sitaram: hm? That was "gitk --merge <file>", see what got changed, fix it	11:25
sitaram	oh ok	11:25
	→ markelik_ joined	11:25
wereHamster	sitaram: yep :)	11:26
	→ marvil07 joined	11:26
doener	sitaram: with the rebase is was more manual, because --merge won't work	11:26
sitaram	(brb)	11:27
doener	sitaram: and of course I had to fix basically the same problem over and over again (as Pieter said, the patches simply didn't apply anymore, and as I moved the code around, the post-image didn't have the fix anymore about 25% of the time)	11:28
	← hyperair left	11:28
	→ boto joined	11:28
	→ hyperair joined	11:28
	→ Jinzhu joined	11:31
	→ madewokherd joined	11:32
	→ gejr joined	11:32
j416	did it	11:32
	git log --pretty=oneline --abbrev-commit --graph -50 \| grep ^\* \| tr "\|*" " " \| tr -s " "	11:32
	ugly but it works hehe	11:32
	ah, \ also needed in first tr	11:33
	well well	11:33
Pieter	hu	11:34
	what are you trying to do?	11:34
	← jeekl left	11:35
	← markelik_ left	11:35
j416	Pieter: 13:24 j416: how can i show commits on a branch that are local to that branch?	11:36
sitaram	doener: (sorry, back now) wouldn't rerere help somewhat here?	11:36
	← nh2 left	11:37
doener	sitaram: no, it wasn't the same textual conflict, just the same logical conflict	11:37
	← markelikalderon left	11:38
doener	sitaram: rerere only helps if you see the same textual thing more than once, e.g. when doing test integration merges over and over again	11:38
sitaram	oh... but with a merge you'd just use gitk --merge as you said? (I'm not very clear on --merge, in practice. In theory I know what it is... heh!)	11:39
charon	j416: for a cleaner approach try: git log HEAD --not $(git for-each-ref --format="%(refname)" \| grep -v "^$(git symbolic-ref HEAD)$")	11:40
	j416: that pretty much spells out "commits on HEAD but not on any other ref"	11:40
j416	oh ok	11:40
charon	you'd have to modify it slightly for the detached case	11:40
j416	thank you!	11:40
	looked more complicated though hehe	11:40
	but cleaner indeed	11:40
doener	sitaram: --merge just limits the output to interesting commits, so you can more easily identify which commit is responsible for the conflict and what might be the correct resolution	11:41
sitaram	but the application of that resolution is still vi?	11:41
doener	yeah, you still get to fix the stuff yourself, gitk --merge just helps you to find the right solution	11:42
j416	looking at the maint branch of the git project, i see it contains severl "Merge branch 'maint'", how can maint be merged into itself? o_O 8146f19 is an example.	11:42
doener	and rerere saves you from having to do that over and over again, because it records the resolution you have chosen on the first try	11:42
sitaram	hmm; I guess I still have to do some really big stuff with this. Darn it my teams are too small :-)	11:42
Gitbot	[git 8146f19]: http://tinyurl.com/ogxsjd -- Merge branch 'maint'	11:42
charon	j416: probably that happened on master and was later merged to maint	11:42
	→ DavidKlein joined	11:42
	→ agenteo joined	11:42
doener	j416: maint got merged to master, and later maint got fast-forwarded to master	11:43
	← Stravy left	11:43
j416	oh..	11:43
charon	actually, that _certainly_ happened on master, otherwise the subject would be "Merge branch 'maint' into 'foo'"	11:43
agenteo	Hi, I accidentally did a "git revert HEAD" is there anyway I can undo it?	11:44
j416	charon: does git automatically append "into 'foo'" or is this just a convention used here?	11:44
charon	j416: happens automatically unless it's into master	11:44
j416	agenteo: git revert just records a commit, do git reset --hard HEAD^	11:45
	agenteo: this also resets your workdir, sorry: git reset HEAD^	11:45
charon	well, actually it doesn't put quotes around the foo. shrug	11:45
j416	charon: cool, thanks :) so 'master' is a little magic afterall then	11:45
agenteo	j416: yeah a commit that deleted what I committed in the last revision...	11:47
	doing git reset HEAD^, presents me the list of those deleted files	11:47
j416	agenteo: take a look at your log, you will see that the last commit is you revert operation	11:48
agenteo	yep exactly	11:48
j416	undoing that last commit will undo that revert	11:48
	so, if you reset it to revertcommit^ it will be fine	11:49
charon	actually, if you want to discard the revert you'll have to reset --hard, otherwise your working directory still has the changes	11:50
	← abstrusething left	11:50
charon	(danger, loaded gun)	11:50
j416	agenteo: or, you can revert your revert. :D	11:50
	→ orafu joined	11:51
agenteo	that sounds a bit confusing j416 :)	11:51
	I did reset HEAD^	11:51
j416	agenteo: yes, and probably not what you want	11:51
	charon: i figured he might have done other changed, didn't want to be responsible for lost work hehe	11:51
	s/changed/changes	11:51
agenteo	and the commit from revert HEAD (the one that removed my change) is gone	11:51
j416	agenteo: that's right	11:52
agenteo	and reset --hard restored the state b4 I fucked it up	11:52
j416	if you haven't done any changed since that last revert, you can do "git reset --hard" to get rid of those files	11:52
agenteo	:)	11:52
charon	j416: well, true, but on the other hand, with a plain 'reset' he'll have the reverting changes lingering with the others, so it's kind of a lose-lose and probably best to stash any intervening changes and reset --hard anyway	11:52
agenteo	basically I had some untracked files, and I wanted to get rid of them	11:53
j416	charon: yep, I was kind of trying to (failingly) imply stash hehe	11:53
	agenteo: get rid of untracked files with git-clean	11:53
agenteo	it partially worked, it deleted the files in the project root but I still have some dirs	11:55
	that do contain files	11:55
	→ khmarbaise joined	11:55
	→ disjunctor joined	11:56
j416	agenteo: man git-clean	11:57
Gitbot	agenteo: the git-clean manpage can be found at http://git.or.cz/man/git-clean	11:57
agenteo	cheers j416 :)	11:58
j416	;)	11:58
	→ bobmcw_ joined	12:00
	← bobmcw left	12:00
	→ akitada joined	12:00
	→ rudi_s joined	12:02
	← Eulex_ left	12:04
akitada	git stash pop gets me back to the state I did git stash but it seems clearn staging/unstaging status of files	12:11
	Can I completely get back to the state where I run git stash?	12:12
	re-staging modified files are cumbersome.	12:12
	→ ricky_ joined	12:14
	← disjunctor left	12:14
	ricky_ → ricky	12:15
wereHamster	akitada: If the --index option is used, then tries to reinstate not only the working tree's changes, but also the index's ones. However, this can fail, when you have conflicts (which are stored in the index, where you therefore can no longer apply the changes as they were originally).	12:16
sitaram	git stash apply --index?	12:16
	← rshann_ left	12:16
sitaram	(not sure if it works with "pop")	12:16
	→ daveok joined	12:16
wereHamster	you can always use apply --index; drop	12:16
sitaram	true	12:18
j416	git stash pop --index works great	12:18
kusma	I'm having some troubles with git-svn here. I'm doing "git svn rebase", and I get an index and checksum mismatch. Something like this: http://gist.github.com/117520	12:18
	Anyone with a clue on what might be going on?	12:18
	← piyush left	12:18
sitaram	j416: good to know; was going to try :)	12:18
j416	;)	12:18
	→ disjunctor joined	12:19
	→ bentob0x joined	12:19
akitada	I thought i cannot use --index with pop, seems it just works. thanks guys!	12:19
	← marvil07 left	12:19
j416	since pop is just short for apply; drop, there would really be no reason for not allowing --index :)	12:19
	← priidu left	12:20
akitada	j416: didn't know that. how did you know that?	12:21
j416	speaking of which, is there a way to get a dropped stash back?	12:21
sitaram	sure; grub about in the reflog :)	12:21
	→ marvil07 joined	12:21
doener	sitaram: hm? Stashed stuff does only go into the stash reflog, and from there it is dropped when you use "git stash drop/pop"	12:22
j416	akitada: if you try git stash apply; git stash drop, you will get the same results as if you do git stash pop :)	12:22
sitaram	j416: http://sitaramc.github.com/concepts-and-tips/recover-dropped-stash.html :-)	12:22
j416	ooh	12:22
sitaram	doener: you're right	12:22
	sorry...	12:22
	but take a look anyway :)	12:23
j416	this should be added to the git stash command list	12:23
	git stash revive	12:23
	lol	12:23
akitada	j416: no doc on that?	12:23
j416	akitada: man git-stash	12:23
Gitbot	akitada: the git-stash manpage can be found at http://git.or.cz/man/git-stash	12:23
	← keithhub left	12:23
sitaram	j416: there's a git stash revive?	12:23
	→ memiux joined	12:24
j416	sitaram: no but it would be handy	12:24
	← memiux left	12:24
sitaram	j416: oh, I'm slow today :)	12:24
akitada	j416: it just says " Remove a single stashed state from the stash list and apply on top of the current working tree state. When no <stash> is given, stash@{0} is assumed. See also apply."	12:24
j416	i run in to this from time to time, when I apply (pop) stashed changes that do not merge cleanly and I want to just undo it all and have my stash back the way it was	12:24
	akitada: yes, and git apply does "apply on top of the current working tree state" and drop does "remove a single stashed state from the stash list"	12:25
doener	sitaram: git fsck \| grep commit \| cut -d' ' -f3 \| while read hash; do git rev-parse --verify --quiet $hash^2 && echo $hash; done \| xargs git show	12:25
j416	logical :)	12:25
doener	sitaram: that's what I suggested back in March. Just exploits the fact that stash entries are mege commits, so you can limit the output	12:25
sitaram	doener: not smart enough; you'll have to explain the rev-parse bit	12:25
	s/not/me not	12:26
doener	sitaram: checks if the given commit has a second parent, i.e. if it is a merge commit	12:26
akitada	j416: I see, but git stash pop is short for git stash apply; drop would be nicer explanation.	12:26
	← the_zeno left	12:27
sitaram	doener: but they needn't all be stashs, right -- could be plain merges that got discarded somewhere	12:27
j416	akitada: i think git pop also checks if the stash applies cleanly	12:27
doener	sitaram: sure. The thing is, yours generates even more false positives ;-)	12:28
j416	(which indeed solves my problem.. lol, didn't think of that)	12:28
	→ bremner_ joined	12:29
sitaram	doener: yes, I don't check for merges, but boy it looks very cool ;-) And I limit it with --since	12:29
	→ the_zeno joined	12:30
doener	sitaram: heh, maybe go and create a best of both worlds ;-)	12:30
	sitaram likes the GUIs. Maybe too much :-) Sort of a late bloomer in the "likes GUI" category of people?	12:30
sitaram	doener: yes I've already added the rev-parse; have to test and push...	12:30
	← bremner_ left	12:31
akitada	anyone succeeded to get your repository at work switched to git from svn or cvs? I'm trying but failing.	12:31
sitaram	politically or technically?	12:32
	:)	12:32
akitada	seems most people don't care much about vcs	12:32
	← freelook left	12:32
akitada	sitaram: i think it's due to both	12:32
MisterN	akitada: you can "stealth convert" :)	12:32
	akitada: i.e. use git svn to use the svn repo	12:32
sitaram	if they don't care then they're not using it. I've actually told one project that is basically using it as a "backup system" they don't need anything more than WinZip or something :)	12:33
	got them really mad at me for implied insult -- dont do this!	12:33
	← cemerick left	12:33
akitada	MisterN: that's exactly I'm doing now. but git-svn is a hack at most. I should be missing something using it and feeling I'm living hard way	12:34
	sitaram: 100% agree... I recently found a guy commit his changes with a message "backup" :p	12:35
sitaram	ooh that's one better than my example!	12:35
broonie	A former coworker used to always use the log message "New version".	12:35
sitaram	my guys used to say "end of day <date>", "going for lunch", "weekend", etc!	12:36
akitada	sitaram: that's funny	12:36
jjuran	I remember trying to introduce CVS at work, as it was an improvement.	12:36
sitaram	glad to see I'm not the only one seeing these things :)	12:36
MisterN	jjuran: over what? PVCS or so? :P	12:36
sitaram	MisterN: I think jjuran meant 'over having none' :)	12:36
	we've got projects like that too...	12:37
akitada	So, I'd really appreciate if anyone could give me any tips to convert them to git	12:37
jjuran	MisterN: In one case, Visual SourceSafe, and in the other, folder copies.	12:37
sitaram	akitada: if you clone my github repo, the latest commit on "odp" branch has an ODP you may want to use	12:37
	(you dont have to clone, the web interface will work)	12:37
MisterN	jjuran: folder copies! probably still the most popular system :)	12:38
	→ bremner_ joined	12:38
MisterN	jjuran: and to be honest, with a good diff/merge tool, folder copies aren't _so_ bad :D	12:38
	at least compared to svn :D	12:39
jjuran	MisterN: And some of the backup copies found their way into the folder that was getting backed up...	12:39
	→ d0k joined	12:39
akitada	one thing I hate about svn is that its tags/branches. that's too unusable	12:40
sitaram	akitada: http://github.com/sitaramc/git-notes/raw/bbd3f2fe747117de85a9e869798e4fc730cf6dfd/git-versus-svn.odp -- did it yesterday for a similar situation; haven't presented it yet though (i.e., untested)	12:40
	← kn1tt1 left	12:40
sitaram	akitada: that's because SVN mixes up directories with branches, I think... (ooh nice point, missed it in my ODP!)	12:41
jjuran	sitaram: Perforce makes the same error.	12:41
	→ b_jonas joined	12:42
	→ cemerick joined	12:42
sitaram	hmm	12:42
	bobmcw_ → bobmcw	12:42
sitaram	I can understand some of the tech, but this always seemed a little... weird	12:42
	s/tech/tech decisions	12:42
jjuran	sitaram: Your Web server isn't supplying a Content-type header for that URL.	12:44
	sitaram: Never mind: I can't read.	12:44
sitaram	jjuran: you're right; just "wget" it then...	12:44
	→ Arafangion joined	12:44
jjuran	sitaram: Content-Type: text/plain; charset=utf-8	12:44
	→ drewr joined	12:45
sitaram	I think github's serving of RAW data doesn't interpret what the data is, because in a git repo mostly it will be text anyway	12:45
daveok	hi all - I hope someone can help me with a git permissions issue we have on our shared repo. We changed our repo so it uses core.sharedRepository, but we still can't push to another users branch. The error we get now is: Couldn't set refs/heads/BRANCH_NAME. Anyone got any ideas?	12:46
	sitaram is going home	12:47
daveok	we USED to get a permission denied error on refs/heads/BRANCH_NAME, but we sorted the perms out there. Then we got an error in the logs/refs/heads directories, and we fixed THOSE permissions. And now that we get this 'could not set' error, we're stumped :(	12:47
	→ mfilipe joined	12:48
jjuran	Filename extension mapping would go a long way.	12:48
	→ francois__ joined	12:50
	← cemerick left	12:50
doener	daveok: did you also fix the permissions of the directories? Or just the files? The files (except for the logs) are usually read-only, but the directories need write permissions, as the files in there are just replaced	12:51
	→ FreakGuard joined	12:51
	→ hoohah joined	12:53
hoohah	hello. git does not consider file permissions when push/pulling, right?	12:53
MisterN	hoohah: it does, but when committing.	12:53
	hoohah: push/fetch does not consider files at all	12:54
	pull does because it merges	12:54
daveok	doener - we only changed the directory perms, we're looking at changing the files too	12:55
doener	daveok: hm, AFAICT that error message means that the creation of the lock file failed	12:55
hoohah	MisterN: what is the purpose of doing that during commit?	12:56
daveok	ah, cool	12:56
doener	daveok: so file permissions shouldn't matter at all	12:56
MisterN	hoohah: well, at least if you have specified an option like -a	12:56
doener	daveok: oh, actually, it's not the creation, but the renaming	12:56
	← bremner_ left	12:56
	→ MrN joined	12:57
hoohah	MisterN: would there be a problem if my files were only read/write/executable by owner?	12:57
	← tjafk1 left	12:57
MisterN	hoohah: i don't think so.	12:57
	← mithro left	12:58
daveok	doener - afaik, the permissions on files and directories are correct.	12:58
	→ tjafk1 joined	12:58
	→ kn1tt1 joined	12:59
alanhaggai	Hi. Is there any way to change the author/commmitted by information recorded in a past commit?	13:00
	← tjafk1 left	13:00
	← Fullmoon left	13:00
	→ mfilipe_ joined	13:03
	→ priidu joined	13:05
	→ tjafk1 joined	13:06
	← mtvee left	13:06
Rhonda	alanhaggai: rebase --interactive, but that will only help for not yet pushed/cloned stuff.	13:06
	→ Fullmoon joined	13:08
alanhaggai	Rhonda: What I want to do is to not change the commit messages, instead, change the author/committer information of an earlier commit.	13:09
Rhonda	Yes, I understood that.	13:10
	Again, interactive rebasing is what you want. :)	13:10
hoohah	for what appears to be timeouts during git push, I'd C-c to get out of it. Will the push still be reported as complete even if only a partial push was done? (if this is even possible)	13:11
alanhaggai	Rhonda: Oh I see. How should I use it?	13:11
hoohah	is there anyway i can crosscheck hash of files of local repo against hash of remote files?	13:11
rudi_s	alanhaggai: Have a look at man git-commit-tree, you need to use enviroment variables when using git commit --amend (after you used git-rebase -i).	13:12
Gitbot	alanhaggai: the git-commit-tree manpage can be found at http://git.or.cz/man/git-commit-tree	13:12
Rhonda	alanhaggai: You pick the commits in rebase -i for which you want to change it, and then commit --author="" them.	13:12
	erm, --amend involved of course.	13:13
	→ kukks joined	13:13
alanhaggai	Rhonda: Thank you. :-)	13:13
	→ reprore joined	13:14
	← MisterN left	13:15
Rhonda	alanhaggai: The cool thing with git is that you can easily just create a new branch and fiddle around in there without any worries to break anything. ;)	13:16
	← mfilipe left	13:18
alanhaggai	Rhonda: Yes. Git is excellent. :-)	13:20
	← hyperair left	13:21
alanhaggai	rudi_s: Thanks.	13:23
rudi_s	alanhaggai: np	13:23
	← priidu left	13:28
	→ qrush joined	13:29
	← Sho_ left	13:30
	→ Sho_ joined	13:30
	→ the_zeno_ joined	13:31
	← Jinzhu left	13:31
	→ mithro joined	13:31
	← Fullmoon left	13:32
	→ Fullmoon joined	13:32
	← daveok left	13:32
	→ cmarcelo joined	13:32
	← hoohah left	13:33
	→ Fullmoon_ joined	13:33
	→ priidu joined	13:33
	→ mtvee joined	13:36
	→ hyperair joined	13:37
	← Fullmoon_ left	13:42
pasky	repo 20152 58.3 73.4 4476636 2985756 pts/3 D+ 15:30 4:42 git pack-objects	13:43
	grumble	13:43
	damn thing	13:44
	how to make it use less memory? :)	13:44
	← the_zeno left	13:44
Pieter	use less objects	13:44
	;)	13:44
	or smaller window/depth	13:45
pasky	how do I pass to git gc?	13:45
	*that	13:45
	the docs don't explain how to specify custom window/depth	13:45
Pieter	I don't know	13:46
	I guess you can put defaults in a configfile	13:46
	→ repror___ joined	13:47
	← naeu left	13:47
	← reprore left	13:47
	→ scarabx joined	13:48
pasky	2.6G /srv/git/gcc.git	13:48
	i just need to clean this up...	13:48
	→ hugoxrosa joined	13:49
	← reel_ left	13:49
	the_zeno_ → the_zeno	13:50
	← v3s4 left	13:50
	← Fullmoon left	13:50
	→ cemerick joined	13:51
	← psankar left	13:52
	→ mitkok joined	13:52
pasky	repo@rover:/srv/git/gcc.git$ ~pasky/src/git-fresh-inst/bin/git repack -l -A -d --window-memory=2g	13:52
	let's see	13:52
	→ abstrusething joined	13:54
	→ aspotashev joined	13:54
	← akitada left	13:54
	→ plasticine_ joined	13:56
	← aspotashev\|eeepc left	13:56
	← kn1tt1 left	13:57
	← agenteo left	13:58
	→ altrux joined	13:59
	← roop left	14:02
	→ soul9 joined	14:03
	→ memiux joined	14:04
	→ plasticine__ joined	14:06
	→ sdboyer-laptop joined	14:06
	← plasticine left	14:08
	→ kn1tt1 joined	14:10
	← mitkok left	14:10
	← icwiener left	14:13
	→ reprore joined	14:13
	← plasticine_ left	14:13
	→ repr_____ joined	14:14
	← repror___ left	14:14
	← ankit9_ left	14:17
	→ bdiego joined	14:20
	← ph^ left	14:20
	← priidu left	14:21
	← cemerick left	14:22
	→ LiamH joined	14:23
	← j416 left	14:23
	→ priidu joined	14:23
	→ cemerick joined	14:24
	← rudi_s left	14:25
	← sitaram left	14:26
	← ekristen left	14:27
	→ ericindc joined	14:27
	← scarabx left	14:27
ericindc	Is there a way to move committed changes from one branch (master) to another?	14:29
	← alanhaggai left	14:29
b_jonas	yes	14:30
	merge or rebase them, depending on whether you want to change the first branch too	14:30
	← jaeckel left	14:30
ericindc	b_jonas: I simply want to take changes I committed and pushed to the master branch and add add them to a different branch.	14:31
	← bx2 left	14:31
ericindc	I'll read up on rebase	14:31
	→ tchan1 joined	14:31
ericindc	It also seems like cherry-pick would work since I have the commit ID	14:32
	← tchan1 left	14:33
	→ tchan1 joined	14:34
	tchan1 → tchan	14:36
	→ Yuuhi joined	14:37
	→ markelikalderon joined	14:38
	→ davidfetter joined	14:41
	← reprore left	14:46
	→ balint joined	14:47
	→ d0k_ joined	14:49
	← repr_____ left	14:49
	→ reprore_ joined	14:49
	→ eno__ joined	14:50
	← mbroeker left	14:52
	→ tvw joined	14:52
	→ loincloth joined	14:52
	← priidu left	14:55
	← d0k left	14:56
	→ mbroeker joined	14:56
	→ jrmuizel joined	14:56
	← reprore_ left	14:56
	→ memiux_ joined	14:58
	→ priidu joined	14:58
	← memiux left	14:58
	→ Korren joined	14:59
	← johan-s left	14:59
	← hugoxrosa left	15:00
	← azumanga left	15:01
Gitzilla	pasky: take a look at git-pack-objects.c, starting at line 1814.	15:02
	→ knittl joined	15:03
	→ Bass10 joined	15:06
	← abstrusething left	15:09
	← bschindler left	15:09
davidfetter	hello	15:10
	short of filterdiff, is there any way to get context-style diffs from git?	15:10
	→ k0001 joined	15:11
	eno__ → eno	15:14
	→ jcervidae_ joined	15:14
jcervidae_	does anyone know if google code have any plans to support git?	15:14
	I don't really see how converting to and from subversion into their repositories is a good thing :(	15:14
	and I know mercurial but prefer git	15:15
b_jonas	did you try writing to them? they usually have a feedback form hidden somewhere. (I don't know if they read it or not though.)	15:18
	→ prateeksaxena joined	15:18
	→ wmoxam joined	15:18
	← plasticine__ left	15:19
	← prateeksaxena left	15:19
jcervidae_	b_jonas: no, but I will. I much prefer google code to sourceforge but I can't use it until they support git	15:20
hyperair	github!	15:20
	i prefer sourceforge to google code actually	15:20
	no, actually i prefer launchpad above everything, but launchpad doesn't have git =(	15:20
	← mbroeker left	15:21
	← kn1tt1 left	15:21
	→ limx joined	15:21
limx	gi	15:21
	hi	15:22
wereHamster	gi -> git	15:22
limx	wanted to write hi :)	15:22
wereHamster	hi git :)	15:22
limx	hehe	15:22
	i got problems pushing a git repo over ssh	15:22
	it says	15:22
	Agent admitted failure to sign using the key.	15:23
	Korren → korren	15:23
	→ icwiener joined	15:23
limx	at my own repo as well at the github	15:23
	← mfilipe_ left	15:23
limx	keys are the same	15:24
	mtvee → mtvee\|away	15:24
	→ abstrusething joined	15:24
limx	any idea? :-/	15:24
wereHamster	limx: did it work at a point in the past? Or never?	15:24
	← memiux_ left	15:24
limx	wereHamster: it worked months ago.. but i reinstalled everything today	15:25
	do i have to copy the whole id_rsa.pub from my local machine to the server?	15:25
	or just parts of it	15:25
	→ mbroeker joined	15:25
wereHamster	you need your private key (~/.ssh/id_rsa)	15:26
	← priidu left	15:28
	→ sea-gull joined	15:28
	→ lll joined	15:28
	← abstrusething left	15:29
limx	wereHamster: not the public key?	15:30
frsk	You have to send the public key to the server in order to use it	15:30
	The private key is private	15:30
wereHamster	to connect to github, you need your private key	15:30
frsk	..and github needs your public key	15:31
limx	what now? :)	15:31
	← s0ber left	15:31
limx	public or private?	15:31
frsk	limx: You have ~/.ssh/id_rsa, right?	15:31
limx	right	15:31
frsk	limx: Then github needs to know what ~/.ssh/id_rsa.pub looks like	15:32
limx	github has the public key allready	15:32
frsk	But it isn't working?	15:32
limx	nope	15:32
	ive got the same problem	15:32
	with an other server	15:32
wereHamster	maybe the private and public key don't match	15:32
limx	i allready tried creating completly new keys	15:33
	didnt work	15:33
	← ben_h left	15:33
frsk	limx: Try ssh [email@hidden.address]	15:33
	→ solydzajs joined	15:33
	← giallu left	15:33
frsk	It lets you know whether or not ssh is able to authenticate	15:33
wereHamster	throw in a -v to get verbose output	15:34
limx	one moment please	15:34
	i paste the output	15:34
	← mithro left	15:34
	→ KingBug_ joined	15:35
	→ cloudhead joined	15:35
limx	wereHamster: http://paste.pocoo.org/show/hNw1lADvreSYub2PTH39/	15:37
frsk	key mismatch	15:37
limx	but why? the keys in id_rsa.pub and on the server are the same	15:38
	are there other locations with keys maybe?	15:38
frsk	But does id_rsa.pub actually originate from your id_rsa?	15:38
limx	hm i'll try making new keys	15:39
	frsk: "ssh-keygen -t rsa" asks me for a password	15:40
	→ aspotashev_ joined	15:40
limx	should i leave it blank?	15:40
frsk	limx: it's up to you	15:41
	← tvw left	15:41
limx	ok but i dont have to, right?	15:41
	→ galtgendo joined	15:41
frsk	No	15:41
	→ MisterN joined	15:41
frsk	Just make sure it isn't available to anyone else	15:41
limx	okay	15:41
	frsk: next step.. simply cat id_rsa.pub and put output to github.com?	15:42
frsk	limx: yeah	15:42
	← WinterWeaver left	15:43
limx	frsk: the same problem again..	15:43
	Agent admitted failure to sign using the key.	15:43
frsk	limx: hm, at github.com/account, you can see your RSA fingerprint, right?	15:43
limx	frsk: yep	15:43
	← tango_ left	15:44
	→ tango_ joined	15:44
limx	maybe the ssh program on my local machine is broken?	15:44
	→ s0ber joined	15:44
limx	i am using ubuntu jaunty	15:44
frsk	oops, I was wrong, you can't see the fingerprint..	15:45
	It's their fingerprint	15:45
	→ naeu joined	15:45
limx	frsk: i see the their fingerprint and my pub key is in the list	15:45
frsk	limx: does ssh-keygen -l -f ~/.ssh/id_rsa and ssh -l -f ~/.ssh/id_rsa.pub give the same output?	15:45
	limx: uhm, hold on.. do you have id_rsa and id_dsa laying around?	15:46
	Your paste says id_dsa	15:46
	← mpoullet\|work left	15:46
limx	id_dsa?	15:46
galtgendo	can anyone tell whether gtksourceview module optional dep of gitview is one from pygtksourceview or from gnome-python-desktop ?	15:46
limx	not in .ssh	15:46
frsk	limx: hmkey. what about those ssh-keygen-commands?	15:47
	← orafu left	15:47
limx	frsk: the second throws an error	15:47
galtgendo	if so, can somebody add a note to the docs about it ?	15:47
limx	frsk: ssh: Could not resolve hostname /home/chans/.ssh/id_rsa.pub: Name or service not known	15:47
frsk	oh? what?	15:47
	→ orafu joined	15:47
frsk	limx: ..but I thought you had your id_rsa.pub which you cat'ed and gave to github?	15:48
limx	right	15:48
frsk	It's not in ~/.ssh?	15:48
limx	it's in .ssh	15:48
frsk	hm, then how come ssh-keygen -l -f ~/.ssh/id_rsa.pub won't work..	15:48
limx	i put ~/.ssh/id_rsa.pub on github	15:49
	the file exists	15:49
	oh	15:49
	now it works	15:49
	strange	15:49
frsk	Well, as long as it works, right? :)	15:49
limx	frsk: no, i mean ssh-keygen -l -f ~/.ssh/id_rsa.pub throws no error :D	15:50
	pushing repos still doesnt work	15:50
frsk	ah, heh :)	15:50
limx	frsk: output of both commands is equal	15:50
	← s0ber left	15:50
frsk	limx: Do you have a ~/.ssh/authorized_keys2?	15:50
	Try cat ~/.ssh/id_rsa.pub >> ~/.ssh/authorized_keys2	15:51
limx	frsk: nope	15:51
frsk	And then ssh localhost	15:51
	It's a bit useless, but will show if the key is actually working or not	15:51
	→ elbryan joined	15:52
	← psoo left	15:52
limx	frsk: i dont have ssd running	15:52
	should i install it?	15:52
	*sshd	15:52
elbryan	hi everybody.. I want to switch account that I use to commit changes on a svn repo ..	15:52
	KingBug_ → KingBug	15:52
elbryan	how to switch username? ^^	15:53
thiago	with git-svn?	15:53
elbryan	(username plus password, of course)	15:53
	thiago: yes	15:53
frsk	limx: hm, you could try it and then remove it if you'd like?	15:53
thiago	elbryan: man git-svn. Search for rewriteRoot	15:53
Gitbot	elbryan: the git-svn manpage can be found at http://git.or.cz/man/git-svn	15:53
	→ abstrusething joined	15:53
	← aspotashev left	15:54
limx	frsk: ok i installed sshd	15:55
	frsk: same error	15:55
	frsk: Agent admitted failure to sign using the key.	15:55
	frsk: but im able to login	15:55
	although the error apears	15:55
frsk	limx: using your password?	15:55
	Or without it?	15:55
limx	the ssh key has no password	15:56
	my local account has one	15:56
elbryan	thiago: have I to reinitialize it?	15:56
limx	of course	15:56
thiago	elbryan: no, just set the configuration	15:56
frsk	limx: Yeah, so you did have to enter a password?	15:56
limx	frsk: when i connect to my root server the error doesnt appear	15:56
	frsk: yep	15:57
frsk	limx: In which case, you authenticated with your password, not your key	15:57
	hm, strange	15:57
	→ bx2 joined	15:57
limx	frsk: should i try reinstalling the ssh package of my distro?	15:58
frsk	limx: nah, not worth it	15:58
limx	hm	15:58
	every time "Agent admitted failure to sign using the key." comes up	15:58
elbryan	thiago: sorry but I really don't get it ..	15:59
frsk	Ah, great, I should've googled a little bit earlier :)	15:59
	→ almostautomated joined	15:59
elbryan	isn't it a parameter of clone command?	15:59
thiago	elbryan: did you search for rewriteRoot (not rewrite-root)?	15:59
frsk	limx: https://bugs.launchpad.net/ubuntu/+source/openssh/+bug/201786	15:59
limx	https://bugs.launchpad.net/ubuntu/+source/openssh/+bug/328445	15:59
	lol	15:59
frsk	limx: You're might being hit by a bug	15:59
limx	found the same	15:59
frsk	:-P	15:59
limx	hehe	15:59
thiago	elbryan: it's a config option for git-config	15:59
limx	next time i use google first xD	16:00
	→ reprore_ joined	16:00
limx	yeah	16:00
	that might be	16:00
	i am using jaunty 64 bit	16:00
	← marc_kdab left	16:01
elbryan	thiago: This allows users to create repositories from alternate URLs.	16:01
	← reprore_ left	16:01
thiago	elbryan: exactly	16:02
	elbryan: you want to change the URL (the username)	16:02
elbryan	how this could help me switching username?	16:02
	→ s0ber joined	16:02
elbryan	the point is.. actually it is automatic.. don't know where git stores these informations	16:02
thiago	by changing the username in the URL	16:03
	use git-config to change the config	16:03
	→ hpoydar joined	16:03
frsk	limx: Heh, some of the comments suggests restarting X.. -- Good luck making it work :-)	16:03
	frsk jumps back to reading for exams	16:03
elbryan	thiago: yea.. but I think git stores my informations somewhere	16:03
limx	frsk: thx for your help	16:03
frsk	limx: np	16:03
limx	i'll try it	16:03
thiago	elbryan: in .git/config	16:03
	← limx left	16:03
frsk	sorry it didn't work out, though :p	16:03
	→ scientes joined	16:03
elbryan	thiago: there isn't my username and password there.. but commits are automatic	16:04
thiago	elbryan: so change the username in the URL	16:04
	← MrN left	16:04
	→ limx joined	16:04
limx	re	16:04
elbryan	thiago: can I simply delete all informations about my old account?	16:05
limx	frsk: ha, working now!	16:05
	:D	16:05
frsk	limx: :-D	16:05
limx	some things can be so simple	16:06
	restart x	16:06
	→ stuffcor1se joined	16:06
limx	strange bug anyway	16:06
frsk	Sounds like a messed up bug	16:06
limx	yeah	16:07
	← stuffcorpse left	16:07
	← gitte left	16:08
	← bx2 left	16:11
	← khmarbaise left	16:11
	← robinr left	16:13
	← b_jonas left	16:14
	← Arafangion left	16:17
	← lfoppiano left	16:18
	← tango_ left	16:18
elbryan	thiago: rm -rf ~/.subversion/auth is the way	16:19
	→ wagle joined	16:22
limx	finally.. :-) http://github.com/limx/buteo/	16:22
	→ krawek joined	16:29
	← abstrusething left	16:32
	SamB wishes git update-ref would do something sensible when you modify the ref you've got checked out ...	16:37
SamB	(like refuse to go on without some flag, maybe ...)	16:38
	← balint left	16:38
Pieter	wow.. I'd never heard of git check-ref-format before	16:38
	→ sunoano joined	16:39
	← sunoano left	16:40
	← elbryan left	16:40
	→ abstrusething joined	16:41
	→ Alinermedia joined	16:41
	← cemerick left	16:47
davidfetter	hello	16:48
	how would i set my .git/config file to make context-style diffs instead of unified?	16:49
SamB	davidfetter: why would you want to set it that way for every repository you use ?	16:50
	oh, wait.	16:50
	SamB silly	16:50
davidfetter	SamB, ?	16:50
SamB	er, anyway ...	16:50
	→ thiago_home joined	16:51
davidfetter	let's say there are people on a project that rhymes with PostgreSQL who are used to seeing context-style diffs and will be much happier to switch to git if they can be provided by default	16:51
SamB	I got confused, sorry, yeah, that's sensible	16:51
	← ciskje left	16:52
	← mbroeker left	16:52
SamB	should be in either git-diff(1) or git-config(1), if not both ...	16:52
	→ memiux joined	16:53
	← solydzajs left	16:53
	→ radarek joined	16:53
SamB	huh ... is there no such option ?	16:54
	← galderz left	16:54
	→ mbroeker joined	16:54
broonie	It's the first time I've ever heard of anyone wanting anything except unified diff, really.	16:54
davidfetter	i don't know. i'd like to see to it that such diffs appear by default. it will really help drive git adoption on that project	16:55
broonie	davidfetter: wv :)	16:55
davidfetter	broonie, ?	16:55
broonie	well volunteered	16:55
davidfetter	iirc context-style diffs used to be an option	16:56
	dunno where they went	16:56
SamB	could be that they're still in there somewhere ...	16:56
	→ voker57_ joined	16:57
SamB	davidfetter: well, you could have it use an external diff tool if nothing else ...	16:57
	← memiux left	16:58
	→ rshann_ joined	16:58
	← mbroeker left	16:58
	→ brizly joined	16:59
davidfetter	how would that work?	16:59
SamB	dunno!	16:59
	but I saw something in git-config	16:59
	← qhoxie left	16:59
SamB	that is, in it's manpage	16:59
davidfetter	heh	16:59
	davidfetter Rs TFM	16:59
	← aspotashev_ left	16:59
	→ aspotashev joined	17:00
	← loincloth left	17:00
	← gilimanjaro left	17:02
	→ elpargo_ joined	17:04
	→ binjured joined	17:07
	→ mitkok joined	17:08
	← LiamH left	17:09
	→ LiamH joined	17:11
	→ r2p2 joined	17:11
	← cmarcelo left	17:11
	← voker57 left	17:11
r2p2	hello, i was wondering if anyone could help me with installing git....i am using os x 10.5 and when i run the installer dmg it says everything installed but it doesnt seem like its working	17:13
	→ Dashkal joined	17:13
wereHamster	r2p2: is there no readme in the dmg?	17:13
r2p2	wereHamster: there is a readme and i followed everything in it there is only 3 steps	17:14
	→ ph^ joined	17:16
almostautomated	Hey all; as ome of you may know the Notepad++ team is thinking/working on switching to git; and have worked lately ( with major help from doener, Ilari, and nis and the rest of the gang in here ) on getting all setup on github. Well now I get the privilege ( :P ) of getting the project founder on 'the right track'...	17:17
	He is 'interested' but not 'sold' yet. I've forwarded the links for the Linus Talk and the Google Git Talk to him; and was wondering if there are some other good 'passive' resources to pass his way?	17:18
	cpg\|biab → cpg	17:18
	→ wu-lee joined	17:19
almostautomated	I guess I should have said... both the Linus and Randal Tech Talks	17:19
	← elpargo left	17:19
	→ __iron joined	17:20
	→ fridim_ joined	17:20
r2p2	i found git in my /usr/local/git folder but when i try and use git in the command line it says "command not found"	17:25
	→ ToxicFrog joined	17:25
r2p2	any ideas on whats happening?	17:25
	→ memiux joined	17:25
	→ sergiodj joined	17:25
thiago_home	r2p2: where is the "git" executable?	17:25
r2p2	i beilive it is /usr/local/git/bin	17:26
	← fujin left	17:26
	→ khmarbaise joined	17:27
offby1	type "type -ap git" to see which version your shell chooses	17:27
	← sergiodj left	17:27
	→ sergiodj joined	17:27
r2p2	offby1: nothing showed up when i did that	17:28
	→ mbroeker joined	17:28
wereHamster	r2p2: is /usr/local/git/bin in your PATH?	17:28
	→ kuadrosx joined	17:28
r2p2	wereHamster: i didnt put it there so i assume not how would i go about adding it?	17:29
	wereHamster: or could i copy it to somewhere else?	17:29
wereHamster	r2p2: have you restarted your terminal after installation?	17:29
	→ cemerick joined	17:29
r2p2	wereHamster: yes i have	17:30
	→ agile joined	17:30
r2p2	wereHamster: i even restarted the computer	17:30
	→ Modius_ joined	17:30
wereHamster	r2p2: ah, for the terminal, you need to edit your PATH in ~/.bashrc, for other applications read the readme	17:31
	← Modius_ left	17:32
	→ Modius_ joined	17:32
r2p2	wereHamster: ill see what i can do thanks for your help	17:32
	← ceyusa left	17:33
	← sergiodj left	17:33
	→ Paf joined	17:33
	→ sergiodj joined	17:33
	→ Modius__ joined	17:38
r2p2	wereHamster: what if i dont have a ~/.bashrc? haha i have a .bash_login	17:38
	→ is_null joined	17:38
	→ tango_ joined	17:38
thiago_home	type "man bash" and read its manual, especially the part where it lists which files bash reads	17:39
is_null	hi, i can't find the url to the official git-etc website with google ... any tip please?	17:39
almostautomated	In the 'contributing with git' talk @ about 20 min in; there is reference made to the 'invisible' part of a cimmit message? what is that?	17:40
offby1	huh	17:40
r2p2	i think i got it working!! i just added /usr/local/git/bin to the PATH string in my ~/.bash_login file and now i can use git anywhere	17:41
offby1	good question indeed	17:41
	r2p2: sounds right	17:41
r2p2	thanks for the help guys!	17:41
	← ph^ left	17:41
	← Modius_ left	17:43
	← MisterN left	17:43
	← rshann_ left	17:43
	→ MisterN joined	17:44
	← r2p2 left	17:44
	→ memiux_ joined	17:45
	← memiux left	17:45
	→ ekristen joined	17:45
ekristen	how do you create a branch on a remote repository?	17:46
is_null	ekristen, do you mean "how to push a branch"?	17:46
offby1	create it locally, then push it	17:46
	is_null: jinx	17:46
	→ tobiassjosten joined	17:46
tobiassjosten	Does git rebase change the commit timestamps?	17:47
ekristen	is_null: sure, basically I have master on my local and push it to origin	17:47
thiago_home	tobiassjosten: the author timestamp is kept	17:47
	tobiassjosten: the committer timestamp is changed	17:47
	← Modius left	17:47
almostautomated	offby1: if you get the chance to watch the vieo the reference is made @ +20:09	17:48
ekristen	I want to create another branch locally and push it up to the remote, but not merge it to master on the remote, I want to keep it separate	17:48
tobiassjosten	thiago_home: Care to explain the difference? I didn't know there was more than one.	17:48
ekristen	is_null: ?	17:48
	→ priidu joined	17:49
ekristen	offby1: how do you push a local branch to a remote and have it stay separate from master on the remote?	17:50
offby1	git push name-of-the-remote refs/heads/whatever:refs/heads/whatever	17:52
ekristen	so for example git push origin refs/heads/testing:refs/heads/testing ?	17:53
offby1	suer	17:53
	sure	17:53
	Try It And See™.	17:53
ekristen	why not just git push origin testing? and why refs/heads ( I mean I am definitely not an expert :) )	17:53
	offby1: I really don't want to screw up my remote	17:54
thiago_home	tobiassjosten: the one is the date in which the contents were first created	17:54
offby1	"git push origin testing" might well work	17:54
straszheim	<- is getting pushback because git "isn't good enough on windows"	17:54
thiago_home	tobiassjosten: the other is the actual commit	17:54
offby1	ekristen: make a toy repository and try out your push to _it_ first.	17:54
	straszheim: well, it's true.	17:54
thiago_home	tobiassjosten: it's changed when you rebase, cherry-pick or use "am" to apply patches	17:54
straszheim	offby1: just doesn't really work?	17:54
offby1	straszheim: it does, but it's just a little trickier	17:55
straszheim	i see this msysgit thing	17:55
offby1	configuring ssh, e.g., can be baffling (granted that's not git's fault)	17:55
	yeah, msysgit is fine	17:55
	it's _definitely_ worth trying.	17:55
	straszheim isn't a windows guy but is having to deal with a lot of ignorant mfckers	17:55
offby1	straszheim: trouble is, if you want a GUI interface -- and many people do -- your choices are git-gui, which works fine but is _very_ different from typical Windows GUIs; or TortoiseGit, which is kind of unpolished at the moment	17:56
	now now, they're not mf**ckers just because they're not Linux whizzes	17:56
	→ tvakah joined	17:56
	← LiamH left	17:56
straszheim	nah, they're proactively ignorant about version control, release process, you name it	17:56
	okay. well thanks for the background about the windows stuff.	17:57
offby1	I think you should try both msysgit and tortoisegit	17:57
	you might find they're perfectly fine	17:58
	→ LiamH joined	17:58
	→ drizzd joined	18:00
	→ robinr joined	18:02
	→ rshann_ joined	18:03
	→ hobodave joined	18:04
	← memiux_ left	18:05
	→ nadim joined	18:08
	← lohrun left	18:08
	→ jksz joined	18:10
	→ memiux joined	18:13
	← brizly left	18:14
	→ brizly joined	18:15
	→ sdboyer joined	18:17
	→ ph^ joined	18:18
	← brizly left	18:19
	→ mclem joined	18:19
	← mbroeker left	18:21
	→ bobesponja joined	18:23
	← memiux left	18:25
	elpargo_ → elpargo	18:25
	→ solydzajs joined	18:26
	← mclem left	18:26
	→ brizly joined	18:27
	→ insulina joined	18:28
insulina	hello, i have a git repository in sourceforge, and now i want to put exactly what i have in the local dir to sf	18:30
	how can i do that ?	18:30
Ilari	insulina: Add whatever content you want there, commit and push?	18:32
	→ SRabbelier joined	18:32
Ilari	insulina: Hint: doing git add on directory is recursive add of every non-ignored file.	18:32
	← mitkok left	18:34
	→ mbroeker joined	18:36
	← sfwc left	18:37
	→ schacon joined	18:37
insulina	when i do git push ssh://... , it tells me that everything is update	18:39
	but it isnt	18:39
Ilari	insulina: You have new commits to push?	18:40
	→ memiux joined	18:40
Ilari	insulina: Push transfers whole commits.	18:40
	→ toxx_ joined	18:40
	← cemerick left	18:41
	→ r0bby_ joined	18:43
	← r0bby_ left	18:44
	→ r0bby joined	18:44
	→ sfwc joined	18:47
v0lksman	how do you remove a file from a repo without removing it from working dir?	18:47
Ilari	v0lksman: 'git rm --cached p/f'.	18:48
offby1	.oO("p/f"?)	18:48
v0lksman	Ilari: thx...	18:49
thiago_home	offby1: path/to/file	18:49
	→ elpargo_ joined	18:50
	→ khmarbaise_ joined	18:51
	← robinr left	18:53
	← khmarbaise left	18:55
	← memiux left	18:56
insulina	how do i clone my local repo to sf repo ?	18:56
thiago_home	sf as in sourceforge.net?	18:57
insulina	yes	18:57
	← tobiassjosten left	18:57
thiago_home	didn't know they had Git	18:57
	has the repository been created on their website?	18:57
insulina	they have now , and a lot of others stuff	18:58
	→ sakyamuni joined	18:58
	→ memiux joined	18:58
	→ balint joined	18:58
insulina	i have a git repository there , and i put there git stuff , but now my local git and sf git is diferente , and i dont know how to put it equal to my local repo.	18:58
thiago_home	push from your local repo there	18:59
insulina	i alredy did the push , but deletes dont get any efect	18:59
thiago_home	what deletes?	18:59
	→ kahmalo joined	18:59
	→ ashH joined	19:00
insulina	file and folders deletes	19:00
	← SRabbelier left	19:00
	→ robinr joined	19:00
thiago_home	you push commits, not files	19:00
	← memiux left	19:00
thiago_home	can you see the branches on their website updated to what you wanted to push?	19:00
	→ memiux joined	19:00
almostautomated	Just finished the tech talk: 'Contributing with Git' ... nice.	19:02
	→ ronny joined	19:03
	→ Chani joined	19:03
ronny	hi	19:03
ft	google tech talk? is it going to be publically available?	19:04
ronny	is there any way to get the full status of the workdir in less than 3 subprocess calls?	19:04
	← abstrusething left	19:04
thiago_home	git-status is one subprocess	19:04
ronny	thiago_home: not exaclty helpfull for processing the data	19:05
thiago_home	git ls-files then	19:05
	→ cxvii joined	19:05
ronny	currently i do ls-tree -r HEAD, ls-files -c -<some more> and ls-files -<some others> (without -c) to get all data i need	19:05
	← elpargo left	19:06
	← rshann_ left	19:07
	← memiux left	19:08
ronny	thiago_home: i havent found any way to go with less than 3 calls, and im not yet sure i really get all data i need	19:08
	→ memiux joined	19:08
Chani	what's the most efficient way to tell git "remove all local changes, committed or otherwise, and make this look like a fresh checkout of origin"? I imagine it would involve git clean and checkout or reset... probably checkout...	19:09
offby1	"git reset --hard origin ; git clean -dxf" prolly	19:09
Chani	ah. thanks	19:10
thiago_home	the other way around	19:11
	git clean -xdf ; git reset --hard origin	19:11
	← memiux left	19:12
Chani	thiago_home: why?	19:12
	→ nh2 joined	19:13
	← marvil07 left	19:13
thiago_home	because of overwriting untracked files	19:14
	git reset refuses to overwrite untracked files	19:14
offby1	ah	19:14
	→ wshimmy joined	19:14
thiago_home	so you must get rid of all of them before resetting	19:14
Chani	ahhh	19:14
offby1	also for some reason that I cannot articulate, I wonder if "git checkout" would be less harsh than "git reset"	19:14
	git reset abandons the current commit, you see	19:14
	it could leave that commit unreferenced.	19:15
	which is a minor pain	19:15
Chani	I'm not 100% certain that committed-but-not-push changes should be reset by this portion of scripty, as svn has no such concept... but I figure it's the safer route. nuke anything unexpected, start clean.	19:15
nh2	Wine wiki says: "If you've found the commit which is causing your problem, revert it with git show b821e839bb33ac8f56939cc582010ecf4d9c25d4 \| patch -p1 -R". Can anyone explain how this is different of git revert? Is it just that it does not record a commit?	19:15
	← drewr left	19:15
offby1	apparently.	19:15
	I suspect "git revert -n" would be exactly equivalent	19:15
	← naeu left	19:15
	→ drewr joined	19:15
nh2	I'm not quite sure, has git revert been there right on from the beginning? maybe the wiki is just not up to date	19:16
	→ memiux joined	19:16
thiago_home	Chani: just make scripty never leave unpushed commits	19:16
offby1	nh2: I was wondering exactly that	19:16
thiago_home	Chani: in case it does, the next run will just discad them and do the work again	19:16
impulze	hm	19:17
	→ icwiener_ joined	19:17
nh2	oh no, I see it has been git revert -n before, but for some reason they changed it	19:17
Chani	thiago_home: there seems to be some redundancy in script already with unnecessary repeat cleanup	19:17
impulze	is git send-email --compose supposed to overwrite the summary in the message?	19:17
almostautomated	ft; it already is... you tube	19:17
Chani	hmmm.	19:18
ft	almostautomated: oh, okay then. thanks. :)	19:18
Chani	so, branches.	19:18
	← icwiener left	19:18
nh2	the comment of the wiki edit was: "Don't use 'git revert' - it creates a new commit" - but this is not true if you use git revert -n right (which apparantly was in there before)	19:19
	← markelikalderon left	19:20
	→ wferi joined	19:20
	→ abstrusething joined	19:20
nh2	maybe it does affect something if you change something else?	19:21
Chani	scripty is only interested in a couple of the remote branches. and update_xml itself is only interested in one particular branch. does git have a way to only download the branch I'm interested in or should I clone the whole repo and checkout the branch to be used..? my knowledge of remote branches is fuzzy	19:21
thiago_home	Chani: you can download as many branches as you want, including just one	19:21
Chani	does clone get all of them, or..?	19:22
	← almostautomated left	19:22
	← wferi left	19:22
	← abstrusething left	19:23
	← agile left	19:24
	Chani does some tests	19:24
	→ dyjytyl joined	19:25
dyjytyl	im getting this error when i try to git push: ! [rejected] master -> master (non-fast forward) ... error: failed to push some refs to	19:25
ToxicFrog	You pulled master, made some commits, and are now trying to push	19:26
	However, in the meantime, someone else did push	19:26
	You need to re-pull to merge your changes with theirs	19:27
insulina	thiago_home: when i browse git in sf , it seems good, but when i try to push it or clone the repo, it doesnt give me what i see in the browser	19:27
thiago_home	insulina: do you mean pull?	19:30
	→ ciskje joined	19:30
	← brizly left	19:31
insulina	thiago_home: i got it now ... made a new clone repo, git is confusing ....	19:32
thiago_home	insulina: well, I still haven't understood you	19:32
insulina	i am trying to put my local changes in my laptop to sf , and get the repo to my desktop from sf	19:33
	→ RayMorgan joined	19:33
	← schacon left	19:34
Chani	well, this is odd. I checked out some of my code from gitorious, created a new branch, tried to push it, and got "fatal: The remote end hung up unexpectedly	19:34
galtgendo	can anyone tell whether gtksourceview module optional dep of gitview is one from pygtksourceview or from gnome-python-desktop ?	19:35
	→ bx2 joined	19:35
galtgendo	if so, can somebody add a note to the docs about it ?	19:35
	← hobodave left	19:36
	→ henux joined	19:36
	→ mozinator joined	19:36
	→ talexb joined	19:36
henux	will git handle binary files (png images) fine automagically just when i `git add` and commit them or should i do something extra?	19:37
thiago_home	Chani: pushing via [email@hidden.address] ?	19:37
	→ schacon joined	19:37
Chani	hrrrm. committing to master works from my other checkout of the same repo	19:38
pasky	sorry, i think i just broke bunch of git/mingw forks on repo.or.c	19:38
	working on fixing it up	19:39
cYmen	Imagining I had a really huge history in a repository. Is there any way to get to the current version except cloning the entire repository?	19:39
	→ avenger_ joined	19:39
thiago_home	Chani: check that you're pushing via SSH, not git://	19:39
Chani	oh	19:40
	how do I get git to tell me what url my remote uses?	19:40
	← dyjytyl left	19:40
thiago_home	git remote show origin	19:40
Chani	thanks	19:41
henux	does anyone know?	19:41
	→ galderz joined	19:41
pasky	should be fixed again	19:41
	→ limx_ joined	19:41
thiago_home	henux: it will handle them	19:41
Chani	yeah, Iused git:// on hte broken one	19:41
thiago_home	henux: there's no secret	19:41
	Chani: yep	19:41
	Chani: you can't push via git://. Gitorious doesn't accept it.	19:41
henux	thanks	19:42
	just wanted to make sure	19:42
	→ davi joined	19:42
	← DavidKlein left	19:42
	← limx left	19:42
thiago_home	Chani: to change: git config remote.origin.url new-url-here	19:42
avenger_	I put up a gitosis server, it works well, now I want to use eclipse IDE with Jgit, but it doesn't work right	19:42
Chani	oh cool	19:43
	→ almostautomated joined	19:43
davi	How to git-pull a new branch which has been created in the remote repository?	19:43
thiago_home	davi: usually, git fetch remotename will be enough	19:43
	davi: that will create the local copy of the remote branch	19:43
davi	Let me try	19:43
	thiago_home, # git fetch stable-001	19:45
	fatal: 'stable-001': unable to chdir or not a git archive	19:45
	fatal: The remote end hung up unexpectedly	19:45
	# git fetch refs/heads/stable-001	19:46
	fatal: 'refs/heads/stable-001': unable to chdir or not a git archive	19:46
	fatal: The remote end hung up unexpectedly	19:46
Grum	what is the name of the remote?	19:46
avenger_	does anyone use eclipse IDE with Jgit?	19:46
	→ snitko joined	19:47
Chani	nooooo, I'm out of time :(	19:47
	Chani curses reality	19:47
davi	http://git.savannah.gnu.org/gitweb/?p=gnuherds-app.git	19:47
	→ rudi_s joined	19:47
	← rudi_s left	19:48
	← Sho_ left	19:48
	← cxvii left	19:48
	→ name joined	19:48
thiago_home	davi: remotename is the name of the remote	19:48
	davi: what is your remote called?	19:48
	davi: probably "origin"	19:48
	← Weaselweb left	19:48
davi	thiago, The remove branch I want to fetch is 'stable-001'	19:49
thiago_home	davi: not remote branch name	19:49
	Chani wonders whether it'd be faster for scripty to have a separate clone for each branch, or one clone and checkout each branch in turn to work on them	19:49
thiago_home	davi: I meant the name of the remote repository	19:49
Chani	probably the latter	19:49
thiago_home	davi: it's "origin" for the repository you cloned.	19:50
davi	ssh://git.sv.gnu.org/srv/git/gnuherds-app.git	19:50
thiago_home	davi: did you clone that?	19:50
	Chani: no separate clones	19:50
	Chani: that's a waste of disk space	19:50
davi	thiago, I cloned git://git.sv.gnu.org/gnuherds-app.git	19:51
SamB	Chani: it partly depends on how many files are different between branches	19:51
thiago_home	Chani: either switch branches when necessary, or use separate workdirs	19:51
	davi: then it's "origin"	19:51
Chani	"separate workdirs"?	19:51
davi	# git fetch git://git.sv.gnu.org/gnuherds-app.git	19:51
	From git://git.sv.gnu.org/gnuherds-app	19:51
	* branch HEAD -> FETCH_HEAD	19:51
thiago_home	see the git-new-workdir script	19:51
	davi: "origin"	19:51
Chani	but yeah, saving disk space is good. I'm just concerned about how slow scripty is	19:51
SamB	thiago_home: you can do that ?	19:51
	what happens to HEAD when you do that ?	19:51
Grum	if you have a clone just do git pull or fetch and you should get all the new stuff	19:52
thiago_home	SamB: one HEAD per workdir	19:52
SamB	how's it work ?	19:52
Grum	read the source	19:52
thiago_home	SamB: it creates a .git subdir that consists of mostly symlinks	19:52
	SamB: but not everything.	19:52
SamB	sounds disgusting!	19:52
thiago_home	SamB: .git/HEAD is not a symlink	19:52
Chani	oh. neat.	19:53
thiago_home	Chani: unless you're trying to parallelise scripty	19:53
	→ knweiss joined	19:53
Chani	thiago_home: no, I'm not	19:53
thiago_home	Chani: then one checkout should be fine	19:53
	Chani: you can always do rm -rf *	19:54
	Chani: another possibility is to create a checkout in /tmp, work on it, then delete	19:54
wereHamster	what about git-new-workdir?	19:54
jcervidae_	Can you guys recommend the best place to host my opensource projects that are managed with git? Sourceforge is OK but alternatives to choose from would be cool	19:55
thiago_home	Chani: you can even clone your clone into /tmp, do your work, push to the server, then rm -rf the temporary	19:55
Chani	if I delete it, then I have to redownload it all the next day.	19:55
	oh, cloning hte clone...	19:55
	haha	19:55
	→ Zonker joined	19:55
thiago_home	Chani: for that, I'd recommend bare clones for the caching	19:55
	Chani: and don't forget the -s option to clone (to share objects)	19:56
Chani	eh. clean&reset is probably faster	19:56
thiago_home	yeah	19:56
Chani	aanyways, I was leaving	19:56
thiago_home	rm -rf && reset even faster	19:56
	or not	19:56
knweiss	jcervidae_: Did you try githost.com?	19:57
thiago_home	jcervidae_: what do you need from your hosting?	19:57
knweiss	jcervidae_: I meant github.com. Sorry.	19:57
	→ dduncan joined	19:58
	→ ilogger2 joined	19:58
wereHamster	jcervidae_: sourceforge does support git now. Another alternative to github or sourceforge are gitorious or gitosis	20:02
jcervidae_	thiago_home: somewhere to put stuff really. somewhere people will notice it is good	20:03
	→ naeu joined	20:06
pasky	Ok. I have started a fresh full repack job on repo.or.cz, with fork-safe pruning enabled. If anyone noticed a repository gone bad, please tell me. :)	20:08
	→ MisterN joined	20:09
SamB	jcervidae_: along with repo.or.cz, of course ...	20:10
	though you can't make a repo now, I think ;-P	20:10
	judging by what pasky just said	20:11
	pasky: do you think that gdb.git could be moved to gdb/kde.git, and gdb/gnu.git to gdb.git ?	20:12
cYmen	Imagining I had a really huge history in a repository. Is there any way to get to the current version except cloning the entire repository?	20:12
jcervidae_	github is a bit web 2.0 for my eyes	20:13
	yes it looks nice on an e-commerce site i want my project site to be more like google :)	20:13
pasky	yes, repositories management is disabled now, while i'm performing some major optimization tasks and upgrade of the web interface	20:13
cYmen	jcervidae_: Don't be racist just because something is web 2.0 doesn't make it bad.	20:13
pasky	pushing/pulling should work as usual at all times	20:13
SamB	pasky: yeah, I saw on the site	20:13
pasky	SamB: ok, will confirm with owners then will do	20:13
wereHamster	cYmen: you can use --depth to limit the amount of history, but then you can't push back	20:14
mugwump	pasky: my mirror at http://git.utsl.gen.nz/mirror/repo.or.cz/ should be current	20:14
	hmm, actually the last time the job ran it returned some errors	20:15
pasky	mugwump: great! can you disable the mirroring for a bit now? i'm doing some massive repacking, so it's good to have a backup ;)	20:15
mugwump	that old "remote: aborting due to possible repository corruption on the remote side." message	20:16
pasky	hmm	20:16
cYmen	wereHamster: any other way to interact with a repository without downloading it? some branching magic or something?	20:16
mugwump	in, eg, jCrypto.git	20:16
	→ adamtanner_ joined	20:17
mugwump	also qgit4/qgit4-libgit-thin-port.git	20:17
pasky	i know there is one corrupted repository (not sure which it is, though ;)	20:17
SamB	pasky: I don't suppose your repack takes into account the sharing of objects between related repositories ?	20:17
wereHamster	cYmen: no. If the repo has a web interface (gitweb/cgit) you can use that to download individual files	20:17
pasky	SamB: that's the point, that it does :)	20:17
wereHamster	cYmen: oh, there's also git-archive --remote (if the remote supports it)	20:17
pasky	it would be very disastrous on repo.or.cz if it didn't	20:18
mugwump	hmm, those might be empty repos	20:18
cYmen	wereHamster: how about creating a local repository and push merging it to the main?	20:18
	if that makes any sense at all ;)	20:18
SamB	pasky: well, I've noticed that uploading a new repo seemed to involve uploading everything ...	20:18
wereHamster	cYmen: nope, doesn't make sense to me, sorry	20:18
pasky	mugwump: qgit4/qgit4-libgit-thin-port.git fscks ok	20:18
SamB	even if it was forked	20:18
pasky	mugwump: jCrypto.git is corrupted	20:18
mugwump	yeah, I fsck that one ok but no refs	20:19
pasky	SamB: that used to happen long ago, but shouldn't anymore	20:19
	mugwump: i'm stumped about how jCrypto.git could get corrupted	20:19
SamB	but I might have been confused due to gdb.git not actually being the upstream of gdb/SamB.git ...	20:19
	→ Masca joined	20:19
SamB	since gdb.git is that somewhat-stale-looking KDE-related repository	20:20
mugwump	pasky: http://gist.github.com/117723	20:20
pasky	jCrypto2.git has the same master but isn't corrupted	20:20
mugwump	pasky: did you fsck --full ?	20:20
pasky	no i didn't ;)	20:20
	grm	20:20
	a single missing blob	20:21
	i'll write to the owners	20:21
mugwump	ok. is that blob the same as the one on the nopaste?	20:21
Masca	I have an empty 'unmerged' file. I have just done a reset --hard and then merge with upstream and it filed. Do you know how can I fix this?	20:21
mugwump	I've disabled the mirroring job btw	20:21
	(if nothing else, because it repacks etc :))	20:21
cYmen	wereHamster: hm...downloading 300M of git rep for the linux kernel seems pretty huge...	20:22
Masca	err an empty diff from an unmerged file.	20:22
knweiss	Given a dirty working directory: Sometimes 'git-checkout anotherbranch' succeedes (even without -f!) and sometimes I'll get the "You have local changes..." error message and I would like to know what makes the difference? Is it allowed as long as all dirty files are identical on both branches?	20:22
SamB	pasky: and you can't get it from the other repo ?	20:22
	→ altern joined	20:22
offby1	knweiss: I wonder, too. Perhaps it's when checking out would overwrite changed files	20:22
knweiss	From reading the git-checkout man page I would have expected that git-checkout should fail everytime without -f.	20:22
mugwump	pasky: there's a full log of the mirror run at http://git.utsl.gen.nz/mirror/repo.or.cz/mirror.log	20:23
	→ sfwc joined	20:23
pasky	SamB: sorry?	20:23
SamB	pasky: the missing blob	20:23
offby1	where!?	20:24
pasky	SamB: what other repo?	20:24
	offby1 hides under the bed	20:24
pasky	mugwump: did you spot anything else interesting there?	20:24
SamB	pasky: you mentioned jCrypto2 ?	20:25
pasky	SamB: that's the other one	20:25
	SamB: this one was in libqt	20:25
mugwump	pasky: well, search for 'corrupt', there's a handful	20:26
pasky	SamB: and curiously, both jCrypto* repositories ahve the same corruption	20:26
parasti	why do I keep expecting this to not fail: "git checkout -f -b $existingbranch"	20:26
pasky	mugwump: i can see no other ones	20:27
	→ cxvii joined	20:27
	← adamtanner_ left	20:27
	→ voker57 joined	20:27
mugwump	pasky: true. no other issues known, I haven't pored over the log though	20:28
wereHamster	cYmen: either you need the history, in which case 300MB isn't that much, or if you don't need the history then don't use git and download snapshots	20:29
mugwump	gitte wrote that mirroring script	20:29
	wereHamster: or use clone --depth :)	20:29
	→ plasticine joined	20:30
	→ NoOffenc joined	20:30
	→ markelikalderon joined	20:31
knweiss	offby1: I think so. But I would like to be sure because it really took me by surprise when I noticed this today. Regarding the -f option (off by default) the git-checkout man page says "When switching branches, proceed even if the index or the working tree differs from HEAD." I.e. why does the checkout proceed without -f?	20:31
altern	How would I tell git on windows to change working directory to the one from where I call 'git' command?	20:31
offby1	knweiss: good point	20:31
	→ Bombe joined	20:32
offby1	if I were you I'd write a little shell script that creates a trivial repository, with two branches, and does 'git checkout' ... to see if I could pin down the cause of that message	20:32
cYmen	wereHamster: I just don't see why I can't edit code without having the entire history.	20:33
	→ Sonderblade joined	20:33
wereHamster	cYmen: that's the nature of git	20:33
	offby1 nods sagely	20:33
NoOffenc	Hi, I'm thinking of setting up a backup+sync scheme for my home directory. The idea is to have a shared repository on a lightweight usb drive, then carry it back and forth between home and work, update from it at the start of work and commit at the end.	20:34
	However, it's not obvious to me how to proceed to do this with git (I can imagine how to do it with Svn, which I am a longtime user of. Any clues? Should I create a 'bare' repository on the usb drive?	20:34
wereHamster	cYmen: do you want to submit the code upstream or are these just your own personal modifications?	20:34
offby1	NoOffenc: seems reasonable	20:34
soul9	NoOffenc, yeah, though if you have changing binary files your repo wil blow up	20:35
	→ bx2 joined	20:35
NoOffenc	but do I have to have a working copy in my home directory? this would double its size	20:35
soul9	by blow up i mean take vast amounts of space	20:35
Sonderblade	How do you configure git to use colors everywhere? All my git commands are monochorme. :(	20:35
soul9	NoOffenc, no, you can export a working tree from git	20:35
	(see git archive)	20:35
cYmen	wereHamster: well it's actually a purely theoretical question - somebody pointed out that this is possible with svn and not with git and I'm now trying to figure out why because I don't see how it contradicts any of git's underlying principles	20:35
NoOffenc	ok I'll read up on git archive	20:36
wereHamster	cYmen: that somebody is correct.	20:36
soul9	you can edit a git archive export of the working tree and make a patch, but you can't commit on it	20:36
NoOffenc	wrt changing binary files the only ones that could pose problems would be perhaps virtual machines	20:36
	my home dir is about 80 gig + 20 gig VM	20:36
offby1	gaaahhh	20:37
wereHamster	cYmen: git needs to know the whole history in order to create a new commit that can be merged back into the original repository. svn doesn't have such restriction	20:37
	← Masca left	20:37
NoOffenc	and the usb drive is 500 gig, sould be enough for a while	20:37
soul9	heh, well, you don't want to put that in git, the history would blow into 500G pretty fast imo	20:37
offby1	NoOffenc: I really don't think you want to put all that in a single git repo	20:37
	NoOffenc: I thought you were referring to just your dot files.	20:37
	NoOffenc: use rsync or unison or something	20:37
	→ gitte joined	20:37
NoOffenc	but most of my home dir is unchanging	20:37
	← cxvii left	20:38
Grum	i've used git for stuff with a source but no versioning publicly available; just unpack the source; git init git add .; git commit -m"version X"; then git branch and do my local modifications in there, whenever there is a new version of the software, i just checkout master, unpack the new source, git add .; git commit -m"version" and rebase my local changes onto the new HEAD	20:38
cYmen	wereHamster: but why can't a git patch be created without the knowledge of all history	20:38
NoOffenc	most of this 80 GB is MP3s and movies which never change	20:38
soul9	this is the misconception a lot of people have: i say i want to sync my homes, and i'm directed to git or mercurial all the time	20:38
	that's _wrong_	20:38
NoOffenc	I am doing this with rsync right now, actually	20:38
	I am just afraid that one day I'm going to blow up royally	20:38
soul9	git and mercurial are for text files, they work well with those.	20:38
NoOffenc	rsync in the wrong direction or something	20:38
tmz	Sonderblade: set color.ui auto. you can do this for ~/.gitconfig with: git config --global color.ui auto	20:38
offby1	"blow up?"	20:38
soul9	NoOffenc, then set up a nice backup system like venti :)	20:39
NoOffenc	the idea is that git keeps history and keeps me safe from shooting myself in the foot	20:39
offby1	true	20:39
wereHamster	cYmen: of course a single patch can be created (from a git clone with --depth=0 for example). But you can't git push the new commit back into the original repository, you have to create the patch and send that via email to someone with access to the full history	20:39
Grum	or just suck it and download the full history once	20:40
offby1	"Plan 9 from User Space" ... brilliant	20:40
Grum	if 300mb is a problem there are many other problems ;)	20:40
	→ Zonker joined	20:40
soul9	offby1, yeah, http://www.urbandictionary.com/define.php?term=blow+up	20:40
NoOffenc	what's up with the blow up? I'm not a native English speaker :-)	20:41
soul9	offby1, is that sarcastic? :)	20:41
Sonderblade	tmz: it worked thanks.. but uh the colorization isn't as colorful as i'm used to in older git versions	20:41
	gitte is thinking about commit-less tree objects for large-blob tracking.	20:41
soul9	NoOffenc, meaning that space usage will expand really quickly	20:41
offby1	soul9: that link was completely unilluminating.	20:41
soul9	well, yeah, i think i'm not really using any of those meanings actually :)	20:42
	#3 is pretty close though	20:42
	sorry, #2	20:42
NoOffenc	soul9: but even with changing VMs, won't git just store deltas?	20:42
gitte	or #1	20:42
	→ sakyamuni joined	20:42
	gitte could continue with #0, #-1, etc	20:42
gitte	NoOffenc: Git never stores deltas.	20:42
	NoOffenc: they are generated after the fact.	20:42
NoOffenc	gitte: well then stupid me, I was under the impression it did	20:43
robinr	as always it's more complicated	20:43
soul9	that haskall vcs does store stuff in plain diffs though iirc	20:43
	haskell*	20:44
	darcs is it's name.	20:44
robinr	most vcs's do	20:44
soul9	ok, but darcs is a distributed vcs aswell...	20:44
gitte	soul9: it is also slow as a dead snail.	20:44
soul9	:)	20:45
	i said nothing about it being good at all	20:45
robinr	that's orthogonal to distributed/centralized	20:45
soul9	though if someone ever checked out kernel sources using git...well...the first clone takes a while.	20:45
vmiklos	soul9: though it has its own (evil) diff format, it's not a standard diff. ;)	20:45
soul9	heh, it does?	20:45
drizzd	soul9: seriously? I doubt it's much slower than downloading and unpacking a tar	20:46
vmiklos	the bonus is that the format has no specification and continously changes:)	20:46
soul9	drizzd, er, it's waaaaaaaaaaaaay slower. and i have a 100M line.	20:46
	→ almostautomated joined	20:46
	→ rombus joined	20:47
	→ freelook joined	20:47
soul9	it downloads around 500M of history i think	20:47
	maybe more, dunno	20:47
drizzd	Ok, when I cloned it it was much less than that.	20:48
soul9	:) that was probably a while ago?	20:48
	let me see the size...	20:49
	yeah, 725M	20:49
drizzd	2007?	20:49
soul9	with a working tree ofc.	20:49
	:D	20:49
pasky	you don't download the working tree	20:49
Grum	you do in the packfiles :p	20:49
drizzd	.git has 525mb	20:50
pasky	really, that much?	20:50
	rombus → karbassi	20:50
pasky	that's a bit surprising	20:50
	gitte thinks it has tons of packfiles.	20:50
thiago_home	try repacking	20:50
drizzd	yeah, about 10	20:50
pasky	my linux-2.6.git didn't come under the new pruning repack job on r.or.cz so i can't check ;)	20:50
NoOffenc	this gibak stuff seems promising, and I suppose I could exclude the VMs from backup and keep rsyncing them the way I am doing now	20:51
vmiklos	wow, what a rare moment. ;) (/me rarely has time to watch here on #git, and pasky as well ;) )	20:51
	→ j-b joined	20:53
pasky	:P	20:54
	→ squentin_ joined	20:54
pasky	gitte: i used your hardlinking scheme from 2007 after all ;)	20:54
drizzd	ok, a fresh pack out of the oven has 323mb	20:54
gitte	pasky: oh no! ;-)	20:54
pasky	gitte: hardlinking all objects and packs to all forks, then repack + prune	20:54
	i like the idea :)	20:54
gitte	pasky: but why is it now better than in 2007?	20:55
pasky	i don't know why did i reject it then :)	20:55
gitte	pasky: it is ugly as hell.	20:55
pasky	but it is simple and fast	20:55
	→ ingo86 joined	20:55
ingo86	hi all	20:55
	← almostautomated left	20:55
gitte	pasky: it wastes i/o cycles.	20:55
	→ almostautomated joined	20:56
pasky	i'm not really sure about reasonably simple more i/o-efficient options	20:56
gitte	pasky: what was wrong with my --local-and-packed idea?	20:56
pasky	the hardlinking itself is trivial	20:56
ingo86	i'm new to git, i would like to ask if a centralized repository where wvwry developer could push changes is adviced	20:56
	*every	20:56
pasky	gitte: for many administrative reasons, this time around i'm trying hard to avoid having customized git except gitweb	20:57
j-b	how can i rewrite a commit log on a distant repository ?	20:57
thiago_home	ingo86: it works if that's what you want/need	20:57
	j-b: fetch it, rewrite, push -f	20:57
gitte	pasky: this option would benefit others, too.	20:57
thiago_home	j-b: note that anyone who had downloaded this repository will have problems after you push	20:57
j-b	thiago_home: doesn't push -f screw up a lot of stuffs?	20:57
thiago_home	j-b: yes. What I said above.	20:58
soul9	does anyone know why git doesn't work in a fuse-mounted directory?	20:58
	← almostautomated left	20:58
soul9	specificallt an sshfs mounted directory	20:58
	(other than fuse sucks ;)	20:58
pasky	gitte: yes, but unfortunately i have very little time to spare :( i have many tasks in the suse/glibc land nowadays and try to do very different things in my free time as well	20:58
j-b	thiago_home: ok, thanks :D	20:58
gitte	ingo86: it really depends, often it is better to use forks.	20:58
wereHamster	soul9: with some options tweaked it works alright	20:58
	→ almostautomated joined	20:58
wereHamster	but I don't remember which ones you need to tweak :(	20:58
soul9	wereHamster, git needs tweak, or mount options?	20:58
wereHamster	soul9: I think git (in the config file)	20:59
gitte	pasky: oh, hopefully you do not have to cope with that German who is so full of himself...	20:59
soul9	wereHamster, ok, i'll look around, thanks	20:59
ingo86	thiago_home, gitte: as i'm new I'm not sure this is what I want. I need to setup a redmine installation integrated with git. I'm coming from subversion so it's really difficult for me to imagine a repo without a central point	21:00
MarcWeber	Do you know a simple fast gui editor for windows which does not fork and can cope with unix text files?	21:00
ingo86	thiago_home, gitte: a central repo could be useful for backup too, do you agree?	21:01
gitte	ingo86: well, if you're unwilling to go with a decentralized setup, that's it, then.	21:01
Tuomas	soul9: I think you need -o workaround rename with sshfs	21:01
pasky	gitte: i do :) but on sunday we actually had a short meaningful discussion on the mailing list! i was quite happy :)	21:01
	it was the first time ever, i think	21:02
ingo86	gitte: do i need gitosis to manage the central repo?	21:02
gitte	pasky: my heartfelt condolences.	21:02
	ingo86: I forgot my CVS past, you might want to ask others ;-)	21:03
thiago_home	ingo86: I don't see how backup and central repository have anything to do with each other	21:03
pasky	ingo86: central repo is fairly common setup	21:04
	ingo86: I'd say if in doubt, go for central repo	21:04
	IMO, at least	21:04
knweiss	ingo86: You can always use a server where everybody who's involved in the project publishes his/her own public repository. Then backup this collection of repositories.	21:04
ingo86	thiago_home: maybe because the central repository with git means that someone has the updated repo on his computer?	21:05
pasky	ingo86: if your developers aren't proficient with distributed version control enough to demand a different setup, i think you will make them happiest by using a single repository	21:05
	→ john3909 joined	21:05
thiago_home	ingo86: I don't see what you mean	21:05
gitte	pasky: but in that case, it might make them even happier to stay with svn... pfff.	21:06
pasky	ingo86: of course we would need to have a more detailed account of your development processes (e.g. review requirements etc.) to give you a competent advice	21:06
	← karbassi left	21:06
pasky	gitte: surely not :)	21:06
	git gives them flexibility to do it differently when they _need_	21:06
gitte	pasky: well, you know as well as I do the people who'd rather not learn new things, even if that means that they are more limited in their freedom.	21:07
	→ hyperair joined	21:08
ingo86	pasky: well, we are a small team, we work producing frequent releases but we would like to associate every new feature/bug closed with a commit	21:08
	pasky: we start from a long list of single atomic features and then we produce quicky a working release, then we proceed adding the rest	21:09
	pasky: i'm new to git, i read only partially the docs but i have seen that git make my life easier, and every developer could work branch by branch, he can create a branch every time he's working on a new release, that's terrific	21:10
mugwump	distributed vcs doesn't preserve incomplete work very well c.f. centralised, I'd call that a feature	21:10
	→ reprore joined	21:11
pasky	ingo86: with central repository, usually anyone can push whatever they implement with no review or review outside of the VCS; the main alternative is to have a single maintainer (let's call him L) who periodically pulls from repos of the developers, reviews the changes for general sanity and then merges them all together	21:11
	hmm, i didn't need to call him L after all	21:11
	:(	21:11
	so anyway	21:12
	repo@rover:/srv/git$ du -sh linux-2.6.git/objects/pack/	21:12
	317M linux-2.6.git/objects/pack/	21:12
mugwump	with few enough eyeballs reviewing code, even the most active projects are buggy	21:12
pasky	finally repacked :)	21:12
gitte	pasky: and now with --aggressive	21:13
	gitte ducks	21:13
mugwump	that's 70mb smaller than my clone	21:13
pasky	gitte: and now the other thousands of projects :P	21:14
ingo86	pasky: as said, we're a small team and we didn't need review or something else right now (maybe in the future), however i though about adopting git because every clone has the full repo, so every developer could work and turn back without accessing the central repository	21:15
	pasky: this is useful enough to make me change scm	21:16
mugwump	heh ... small team so don't need review, that's funny	21:16
	suppose you don't need tests or documentation either, small project?	21:16
	→ cousine joined	21:17
pasky	ingo86: then i'd say go for single central repository	21:17
ingo86	mugwump: 2 people don't need review, and we write tests before	21:18
	pasky: thank you	21:18
mugwump	every time you look at history you're reviewing	21:19
	when you annotate a piece of code to understand why a change is there, that's review	21:19
ingo86	mugwump: we don't use a review mechanism like pasky said, then	21:21
	→ snitko joined	21:21
	→ Ryback_ joined	21:21
	← NoOffenc left	21:22
	→ psoo joined	21:22
	→ \|almostautomated joined	21:24
	← markelikalderon left	21:25
	→ tbf joined	21:29
	← almostautomated left	21:30
	← freelook left	21:30
	← snitko left	21:31
	→ almostautomated joined	21:32
	→ Mkop joined	21:34
Mkop	is anyone here familiar with jgit?	21:34
mugwump	505 people, apparently	21:36
Mkop	mugwump: not necessarily do all those people use jgit in particular	21:37
	in fact, I would venture that only a minority have ever used it	21:37
	does anyone know how to connect to a repo using an ssh key?	21:37
mugwump	ok, misread it as a dumb question	21:38
	← sakyamuni left	21:38
Mkop	ok, no problem	21:38
	→ bobmcw joined	21:38
	→ blaenk joined	21:38
blaenk	hey guys, is git like svn in that I can check out/clone a repository of a folder and use it in production? since subversion files are hidden	21:39
	→ r0bby joined	21:39
drizzd	Mkop: egit, which uses jgit, can do it	21:40
Mkop	drizzd: yeah, I'm using egit. How do I do that?	21:40
mugwump	blaenk: if you do that for a webroot, make sure you disallow access to the .git folder	21:40
blaenk	so it is possible then, thank you mugwump	21:40
drizzd	Mkop: workspace preferences, under General > Network Connections > SSH2	21:41
	at least that's what shawn suggests in http://n2.nabble.com/-egit--how-to-use-a-ssh-private-key-td2966741.html	21:43
	← cousine left	21:44
	← MisterN left	21:44
	→ MisterN joined	21:45
	→ vuf joined	21:47
	← psoo left	21:48
	→ psoo joined	21:48
	← Ryback_ left	21:49
vuf	hmmm, I seem to have lost my work ... can I grep through the reflog for a filename?	21:50
mugwump	vuf: try git whatchanged --all -g	21:50
	→ repror___ joined	21:50
Mkop	they key is listed, but it's prompting for the repo password instead of private key password	21:51
	s/they/the/	21:52
vuf	mugwump: it's not there. I figure that I lost it in a rebase somehow.	21:52
mugwump	mkop, can you enable verbose ssh handshake ?	21:52
Mkop	I have no idea	21:52
	→ WinterWeaver joined	21:52
mugwump	vuf: if you lost it in a rebase it should be in the reflog	21:52
	otherwise, you could try git fsck and check the dangling commits/trees/blobs	21:53
vuf	mugwump: okay, I think that I now located it in the reflog ... my problem was that I did not now which entry it was, so I had to look at them all one by one	21:54
mugwump	yeah	21:55
	← altern left	21:55
mugwump	I'd tend to use gitk and brute force it somehow	21:56
	eg find .git/logs -type f \| xargs awk '{print $1}' \| grep -v '^0*$' \| sort -u \| xargs gitk --all	21:56
	then chuck the filename in the box	21:56
	→ markelikalderon joined	21:56
vuf	mugwump: that's cool, and pretty much what I was looking for ... I will save that command for later. Thanks!	21:57
pasky	550M webkit/qt.git/objects/pack/pack-8f780f11a36bc874dccb9af0d1a3ab6fe431256d.pack	21:58
	← markelikalderon left	21:58
	→ cousine_ joined	21:58
pasky	wth _are_ these guys keeping in their repos?	21:58
mugwump	qt? uh-oh	21:58
	cousine_ → cousine	21:59
	← cousine left	21:59
nh2	I did something in a branch and accidentally git-fetched upstream changes into that branch instead of master. Which is the best way to roll that back?	22:00
vuf	I am looking forward to the trash reflog so that I will never ever be afraid of losing stuff :)	22:00
mugwump	nh2: see man git-reflog	22:01
Gitbot	mugwump: the git-reflog manpage can be found at http://git.or.cz/man/git-reflog	22:01
pasky	-rw-rw-r-- 1 repo repo 830529410 2009-04-26 10:27 mpls-kernel.git/objects/pack/pack-ac6f31bf30189ce053fbb9c603222ca15daeaa0a.pack	22:01
	this thing stinks too	22:01
mugwump	Gitbot: I told nh2, why are you telling me?	22:01
	stupid bot	22:01
nh2	thx ^^	22:02
mugwump	pasky: that's huge! my clone is also big	22:02
	it should be a fork of linux-2.6.git at the least	22:03
	wow ... look at commit fab657fc0ed	22:04
	Looks like that tree has been through perforce and back	22:04
	gitte suspects a nice backup copy of a nice moving picture.	22:04
Mkop	http://imagebin.org/50322 is the error message I'm getting	22:05
Grum	you got the wrong url	22:06
	the gitweb url is not the 'clone url'	22:06
mugwump	yeah that url is all wrong	22:06
Grum	normally people give tyou the proper url to clone from on the webpage	22:07
	← naeu left	22:07
	← reprore left	22:07
	→ __iron joined	22:08
	← ingo86 left	22:14
	→ MrAwesome joined	22:15
	→ solydzajs joined	22:16
Zonker	hi, what is the recommended way to work as a team using git with a huge svn repo? specifically, i've cloned the svn repo using git-svn (~100'000 commits, a couple dozen branches, ~1GB of source data) into a bare git repo on an NFS server. we are planning to use that as a "buffer" and both clone isolated branches we are working on. however, updating the "buffer" repo via git-svn makes trouble.	22:16
	← gitte left	22:16
Zonker	I'm running something like: git --bare svn fetch --all in the directory of the bare repo, but I don't seem to get updates any more if I judge by gi	22:17
	git log	22:17
	Any ideas?	22:17
	→ SRabbelier joined	22:17
	→ mitkok joined	22:17
	→ memiux joined	22:18
	→ giallu joined	22:18
blaenk	hey guys can someone please tell me how to add all of the files in a current directory	22:21
	I just created a git repository in a folder containing my source	22:21
	git init	22:21
	← bx2 left	22:21
blaenk	but I do not know how to mark all files/subfiles as new files, to be added to the repository	22:21
	git add * ?	22:21
	← repror___ left	22:21
numbah	.	22:22
blaenk	git add . , alright thanks	22:22
jcervidae_	ok i take it back, the flashing web 2.0 front page is horrible but once you login the web 2.0 ness is much nicer and it functions very well	22:22
blaenk	worked perfectly thanks numbah	22:22
jcervidae_	github is a great site and i am putting my projects there	22:22
	almostautomated → almostautomated_	22:23
mugwump	Zonker: judging by git log --all, you mean?	22:23
	→ markelikalderon joined	22:24
	\|almostautomated → almostautomated	22:24
	→ djgera joined	22:27
Zonker	mugwump: yeah, but am I doing the right thing? The updates seem stuck at some past revision, and a git branch -r does not show more recently created svn branches which should've been pulled.	22:29
	→ kate21de joined	22:29
	→ offby1` joined	22:30
	→ fujin joined	22:30
	offby1` → offby1	22:31
	→ kwirky0 joined	22:32
	← almostautomated left	22:34
	→ sakyamuni joined	22:34
	almostautomated_ → almostautomated	22:36
Chani	is there an easy way to tell git, when I'm making a local branch from a remote branch, that I want push and pull to use that remote?	22:37
andres	Is it possible to configure diff.external but have "git diff" default to the internal diff much like git log/git format-patch do?	22:38
	→ offby1` joined	22:38
	→ codebrulee joined	22:38
	→ dodo_the_last joined	22:39
Chani	oh, it seems it is automagically set up. iinteresting.	22:40
	← dodo_the_last left	22:41
	→ dodo_the_last joined	22:41
Chani	huh. and now it works even for master.	22:41
	← psoo left	22:42
	Chani is surprised	22:42
Chani	I guess my previous repos suffered from not being done right at the beginning	22:42
	← offby1 left	22:44
	offby1` → offby1	22:44
	→ leachim6 joined	22:44
	→ mjf joined	22:49
	→ fingo_ joined	22:50
	→ Nazcafan joined	22:50
Nazcafan	hello	22:50
	← plasticine left	22:50
fingo_	so I initialized a bare repository, and now I want it to actually contain the working files, is there an easy way to do this?	22:50
Chani	why doesn't git branch accept --quiet?	22:51
	→ plasticine joined	22:51
	← dodo_the_last left	22:52
Nazcafan	I got an error while trying to pull linux: git://gist.github.com/117790.git	22:52
	How may I solve this?	22:52
	← almostautomated left	22:53
	→ robin joined	22:53
drizzd	are you pulling from an official server?	22:54
	→ dodo_the_last joined	22:54
	→ WinterWeaver_ joined	22:55
	← MisterN left	22:55
drizzd	there is a how-to recover from corrupt repos in the git-faq	22:56
	→ chocolate_jesus joined	22:56
Nazcafan	drizzd, I don'e	22:57
	drizzd, I don't remember where I am pulling from	22:57
	:-(	22:57
	→ francois__ joined	22:58
Nazcafan	drizzd, what is the FAQ url?	23:00
	← WinterWeaver_ left	23:00
drizzd	faq fix-broken-repo	23:01
	http://git.or.cz/gitwiki/GitFaq#Howtofixabrokenrepository.3F	23:02
	→ almostautomated joined	23:03
	→ Ryback_ joined	23:03
	← mjf left	23:04
Nazcafan	ah thanks	23:04
	← hyperair left	23:04
	← dodo_the_last left	23:04
	→ dodo_the_last joined	23:07
	← WinterWeaver left	23:07
	→ Radar joined	23:07
	→ WinterWeaver joined	23:08
	← MrAwesome left	23:08
	← markelikalderon left	23:08
	→ reprore joined	23:09
blaenk	hey guys, I'm trying out github. I did: git remote add github [email@hidden.address]	23:11
	I can do git push github master	23:11
offby1	you can?	23:11
blaenk	but I have to do it each time, can't I just do git push ?	23:11
offby1	blaenk: if you edit your .git/config, you can indeed	23:12
blaenk	yeah I have	23:12
	as shown here: http://github.com/guides/setting-up-a-remote-repository-using-github-and-osx	23:12
offby1	[branch "master"] ... remote = gitub	23:12
	something like that	23:12
davi	How to clone the non-master remote branch?	23:12
	→ ben_h joined	23:13
blaenk	ohhh offby1 so it's not how it says it in that url, okay thanks, I'll have to look for a guide on that then thanks	23:13
Pieter	davi: git clones all branches by default	23:13
offby1	davi: you don't clone branches; you clone repositories.	23:13
	blaenk: your 'guide' is: man git-config	23:13
Gitbot	blaenk: the git-config manpage can be found at http://git.or.cz/man/git-config	23:13
offby1	that bot's like reading my mind, maaan	23:14
SamB	offby1: oh, and here I thought you had it do that on purpose	23:14
	that's what I thought the second colon was for ;-)	23:14
	→ jrmuizel joined	23:14
davi	Pieter, offby1, I see the remove branches as origin/name. How I checkout such non master branch in the cloned repository?	23:14
	← giallu left	23:15
Pieter	git checkout --branch name origin/name	23:15
davi	thank Pieter	23:15
Pieter	or git checkout --track origin/name if your git is new enough	23:15
davi	I will try both to practice	23:15
	← sakyamuni left	23:16
davi	$ git checkout --branch origin/stable-001	23:16
	error: unknown option `branch'	23:16
mugwump	you know, there's this helpful way of seeing what the commands really are, it's called 'man git-checkout'	23:17
Gitbot	mugwump: the git-checkout manpage can be found at http://git.or.cz/man/git-checkout	23:17
	← sfwc left	23:18
davi	Pieter, This seems to work, thanks:	23:18
	$ git checkout --track origin/stable-001 -b stable-001	23:18
	→ francois___ joined	23:19
	→ markelikalderon joined	23:19
	← francois___ left	23:19
	→ sfwc joined	23:20
	← markelikalderon left	23:23
	← dodo_the_last left	23:27
Nazcafan	drizzd, I tried to follow the steps in the Faq but now I am in the git fsck --full part and I get very different results from what the faq states:	23:28
	← __iron left	23:29
	→ kate21de1 joined	23:31
	→ markelikalderon joined	23:31
	→ dodo_the_last joined	23:32
	← memiux left	23:34
	→ memiux joined	23:34
	→ disjunctor joined	23:34
	→ cxvii joined	23:35
	→ AAA_awright joined	23:36
	← francois__ left	23:36
	→ ajonat joined	23:37
	← cxvii left	23:37
	← Ryback_ left	23:38
	→ tntcoda joined	23:38
	→ agenteo joined	23:39
tntcoda	Hi guys, very newbie question that im stuck with. I checked out an older version of a file, and i want to revert to the newer version without commiting anything, hows this done? git log doesnt have any record of the newer commit, it only goes upto the current commit	23:40
offby1	try "git log --all"	23:42
	→ eletuchy joined	23:44
Ilari	tntcoda: Also, 'git reflog'.	23:45
tntcoda	brilliant thanks guys /feels silly	23:46
	almostautomated → almostautomated\|	23:46
	almostautomated\| → almostautomated	23:47
	← kate21de left	23:47
	← reprore left	23:49
	→ gberdyshev joined	23:51
	← fingo_ left	23:52
	← robin left	23:56
	→ mclem joined	23:57
	← kate21de1 left	23:58
	→ reprore joined	23:59

Logs Search ←Prev date Next date→

Channels Documentation