IRCloggy #git 2008-11-12

Provider of IRC logs since 2005.

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

	2008-11-12
	← softdrink left	00:01
	← cannonball left	00:01
	← whee left	00:02
	← coderdad left	00:11
	→ EmilMedve joined	00:11
	← loiseau left	00:12
	← towski_ left	00:13
	← gebi left	00:13
	← rtomayko left	00:14
	→ brettgoulder joined	00:16
	→ ebynum joined	00:16
	← codeshot left	00:17
	← EmilMedve left	00:19
	→ therrg joined	00:22
	→ kennethlove_ joined	00:22
	← olly_ left	00:22
	→ gebi joined	00:25
	← Zimm left	00:25
	→ EmilMedve joined	00:28
	← p4tux left	00:29
eventualbuddha	gmcinnes: permissions issue?	00:29
gmcinnes	I dunno. I gave up. Found a nicer way through git+ssh	00:29
eventualbuddha	gmcinnes: dunno what you're doing, but if it's custom git hosting I've found gitosis to be quite nice	00:30
	← gebi left	00:30
gmcinnes	eventualbuddha: nah. just looking to set it up at reamhost	00:31
eventualbuddha	gmcinnes: oh. well, that's what I mean, I think	00:31
	manages repos for you etc	00:32
gmcinnes	dunno why I torture myself :)	00:32
eventualbuddha	it's not exactly _hard_ to do it yourself	00:32
	but it's nice to have a tool that already did it	00:32
gmcinnes	yeah. life is short. if someone else can do it, go for it.	00:33
	which kinda tells me I'm an idiot for trying to set this up :)	00:33
	→ gebi joined	00:34
	→ willb joined	00:34
eventualbuddha	gmcinnes: heh. don't worry. I already did all the work to set it up and use it without for like a month before I actually took the time to learn wtf gitosis was	00:35
	← dancor left	00:35
eventualbuddha	just trying to pass on the lesson ;)	00:35
	← SjB left	00:35
	← supercali left	00:37
	→ supercali joined	00:38
	← kennethlove left	00:38
	→ jackdempsey joined	00:39
	→ bobmcw joined	00:41
	← CodeOfficer left	00:43
	← eventualbuddha left	00:46
	→ towski joined	00:46
	→ spearce joined	00:47
	→ loincloth joined	00:51
	← SjB_ left	00:51
	→ threeve joined	00:59
	← jm left	01:00
	← loincloth left	01:00
	→ loincloth joined	01:01
	← parasti left	01:01
	← patmaddox left	01:04
	→ aspotashev joined	01:05
	← pcapriotti left	01:06
	← ebynum left	01:06
	→ kanru joined	01:07
	← hachi left	01:07
	→ CodeOfficer joined	01:09
	← x_or_ left	01:09
	← withanx left	01:10
	← CodeOfficer left	01:10
	→ coderdad joined	01:10
	→ madewokherd joined	01:11
	→ bartman`` joined	01:11
	← kennethlove_ left	01:13
	← bartman` left	01:17
	→ Rivelli joined	01:21
	← david_koontz left	01:21
	← chris2 left	01:24
	→ CodeOfficer joined	01:24
	← coderdad left	01:24
	← WOP left	01:25
	→ gretch_ joined	01:27
	gretch_ → gretch	01:27
	← bgerber left	01:27
	← unfo left	01:32
	→ Godsize joined	01:33
	→ coderdad joined	01:35
	therrg → therrg\|brb	01:35
	← warthog9 left	01:36
	→ bgerber joined	01:37
	← coderdad left	01:38
Slimer	Where might I file a bug on git?	01:45
vmiklos	send a mail to the list	01:51
	→ p4tux joined	01:52
	→ coderdad joined	01:53
	← Slimer left	01:58
	→ bbs joined	02:01
bbs	can someone really help me with git format patch	02:01
bremner	bbs: what is the difficulty?	02:02
	← coderdad left	02:02
bbs	bremner: i have a file fix which is a patch	02:02
	git diff 2008-11-10-14-54 origin/fixes > fix	02:03
	i did this to get it	02:03
	how could i use that to make it a git format patch	02:03
	← lirit left	02:03
	→ SjB_ joined	02:03
	← litage left	02:03
bremner	bbs: the easiest way is to checkout one branch and git format-patch otherbranch	02:04
	← digitalhobbit left	02:04
bremner	bbs: is 2008-11-10-14-54 a branch or a tag?	02:04
bbs	bremner: tag	02:04
	mmotm	02:04
	→ coderdad joined	02:05
	← markelikalderon left	02:05
	→ popol joined	02:05
bremner	bbs: so you want to end up at origin/fixes?	02:06
	(after applying the patches)	02:06
bbs	yea	02:06
	i want all patches in there	02:07
	they fix reiser4	02:07
bremner	git checkout -b local origin/fixes && git format-patch 2008-11-10-14-54	02:07
bbs	bremner: ok i'll try	02:08
	thx	02:08
	← CodeOfficer left	02:09
	mw → mw\|out	02:11
	→ warthog9 joined	02:11
	→ bogus joined	02:13
	bogus → Guest78129	02:14
	Guest78129 → ridiculousRay	02:14
bbs	bremner: perfect	02:15
	thx	02:15
	← bbs left	02:15
	therrg\|brb → therrg	02:15
	← coderdad left	02:16
	← piclez_ left	02:19
	← supercali left	02:21
	→ coderdad joined	02:23
Godsize	why are grafts entirely local notion?	02:23
	why not a reserve refs/replace/namespace record a new commit object to replace one that already that whose object name is @sha1 as refs/replace/$sha1	02:24
	→ ArdRigh joined	02:25
Godsize	make all the command xcept mosly anything that involves rev-lists objects	02:25
	honor this placement so that anytime you ask for commit $sh1, the object kayer gives you the replacement commit object back	02:25
	so you can cloe or fetch from this repository along with refs anf fsck/prune wont lose its priginal parents because there arent any replacements	02:26
	...	02:26
	?	02:26
	just planting seeds	02:26
	maybe someday they'll take on root	02:26
	you never know	02:26
	← priidu left	02:27
	← Ademan left	02:29
	→ haris joined	02:31
up_the_irons	7pong	02:32
	oops	02:32
	← kanru left	02:35
	← hobodave left	02:38
	← coderdad left	02:39
	→ coderdad joined	02:40
	← joshbuddy left	02:42
	→ moccuo joined	02:43
	← bobmcw left	02:44
	← notbenh left	02:46
moccuo	im new to version control, and im trying to distinguish the benefits of using distributed over central repository. with distributed, lets say we have 10 people working on the same project. where do they all pull from? do they need to find someone with a certain branch and pull from them? or would there be at least some sort of "main" repo everyone goes to?	02:47
	and sorry for being such a noob	02:47
Godsize	you got couple of people they pull the current head, which is hopefully good and tested and they start working on it and they start commitint on it	02:49
ivazquez	Most DVCSes, git included, allow a partially-centralized model. People go off and do their changes locally, then push to the main repo.	02:49
Godsize	and you dont need to dont need to wait for a week till your commits are stable, cause your commits are always local	02:50
	to do*	02:50
	yeah	02:50
	thats what distributed is..theres no central location, everybody is the same and you can merge between yourselves	02:50
	you can commit every line, you can then communite by pulling and merging each others work and one person finds a bug and commits it and tell the other 9 people	02:51
	→ kanru joined	02:52
Godsize	then when they're all done they can tell the main group to pull and they its already tested and works...and you can actually time it before you have to ask anybody else to look at it	02:52
	anyways thats hugely better model for doing development	02:53
	turns out people usually dont need that much power, so people dont pull within a single group, even with huge projects	02:53
	← coderdad left	02:57
	← spearce left	02:57
	→ coderdad joined	02:57
	→ hobodave_ joined	02:58
	→ SoreGums joined	03:00
	← SoreGums left	03:00
Godsize	has anyone here looked at the whole cpu alloc stage 2 patchset?	03:00
	← grahal left	03:03
	→ grahal joined	03:04
	← coderdad left	03:05
	← whee_ left	03:05
	← Godsize left	03:05
	→ imyousuf joined	03:08
	→ hobodav__ joined	03:09
	← mw\|out left	03:09
	← jerbear left	03:09
	← Cacheaway left	03:10
	→ Cacheaway joined	03:10
	sdboyer\|class → sdboyer	03:10
	→ stuffcorpse joined	03:10
	→ GNUix joined	03:11
	← redinger left	03:11
	← flaguy left	03:13
	→ wallerdev joined	03:18
wallerdev	hey, how would i reset changes to HEAD for one file?	03:19
	← p4tux left	03:22
wallerdev	hm interesting	03:23
	git checkout HEAD file	03:24
	aanyway thanks	03:24
	← wallerdev left	03:24
	← GNUix left	03:24
	← brettgoulder left	03:24
	← hobodave_ left	03:25
	→ GNUix joined	03:25
	← ArdRigh left	03:27
	→ qrush joined	03:28
	← foca left	03:31
	← dreiss left	03:32
	→ hobodave joined	03:33
gmcinnes	hmm. gitweb shows none of my projects. any thoughts?	03:34
	→ foca joined	03:36
	→ coderdad joined	03:37
werdna	gmcinnes: ls -la /var/cache/git	03:40
gmcinnes	nothing there.	03:40
	well, just . and ..	03:41
	← stephenos left	03:43
werdna	there's your problem	03:44
	ln -s /path/to/repository /var/cache/git	03:44
	→ ArdRigh joined	03:45
	→ tenpaiyomi joined	03:46
tenpaiyomi	Quick question; What's the command to make a remote repo reflect your local repo exactly?	03:46
	→ abbe joined	03:47
ToxicFrog	rsync -avvPhz --delete local remote?	03:48
	→ ebynum joined	03:48
ToxicFrog	The real answer, of course, is "it depends on what you mean by exactly"	03:48
	← ebynum left	03:49
ToxicFrog	If you want them to match in every respect, including config contents, bareness or lack thereof, etc, that rsync will in fact do it.	03:49
	→ loinclot_ joined	03:50
	← hobodav__ left	03:50
tenpaiyomi	hmmm, well, I guess right now I need to know how to undo a git reset, to bring back some changes, and from there I will want to make a remote branch reflect my local branch	03:52
	→ WOP joined	03:54
ToxicFrog	A hard reset?	03:54
tenpaiyomi	Yea	03:54
	I was able to pull from the reflog	03:54
	So, I got that now	03:55
ToxicFrog	That I really don't know, apart from the fact that it can be done and the reflog is where you start	03:55
	As for syncing local and remote branches - do you have the other repo added as a remote?	03:55
	If so, just: git push <remote> <branch>, eg "git push origin master"	03:55
tenpaiyomi	I finally got it; needed to issue a git push origin master -f, however managed to forget to backup the changes I had made before I did a git reset --hard HEAD^, so had to pull them back with the reflog, then recommit	03:57
	Wanting to keep the commits down to a minimum, and I managed to make a stupid mistake of, of all things, overwriting a readme file in the incorrect place	03:58
	← damog left	03:58
	← Rivelli left	03:58
	← gmcinnes left	03:58
	→ foca_ joined	04:01
	→ dreiss joined	04:01
	→ elmex_ joined	04:02
	← LiamH left	04:04
	← loincloth left	04:05
	← ashleyw left	04:08
dermoth	tenpaiyomi, git push --mirror	04:08
	tenpaiyomi, it will push everything with the sole exception of your stashed changes	04:08
	← moccuo left	04:09
dermoth	tenpaiyomi, It looks like there is also a config option, but I think it's only on very recent versions of Git	04:09
	→ Ademan joined	04:09
	→ csris joined	04:10
	← csris left	04:10
	→ litage joined	04:14
	← elmex left	04:16
	elmex_ → elmex	04:16
	← tenpaiyomi left	04:16
	← kukks left	04:16
ArdRigh	Does anyone here use git on Windows?	04:18
	← Tv left	04:19
ToxicFrog	<--	04:19
	via cygwin, though, not msysgit	04:19
	→ CodeOfficer joined	04:21
ezyang	ArdRigh: I do	04:22
	both ways, but I find cygwin to be more reliable	04:22
ArdRigh	ezyang, how do you generally make your git repository available to other people from your Windows machine?	04:22
	← foca left	04:22
ArdRigh	do they do a pull from you, or do you push	04:22
ezyang	ArdRigh: through a public repos hosting site like repo.or.cz	04:22
ArdRigh	ahh ok	04:22
ezyang	if I need to send one commit	04:23
	format-patch	04:23
ArdRigh	Is it possible to have a git repo on a Windows box that people can pull from ?	04:23
ezyang	I've never needed to give someone ssh access/get ssh access to do a push or pull directly to the other person	04:23
	yep	04:23
ArdRigh	run sshd on windows would be the easiest method?	04:23
ToxicFrog	It should be; if nothing else, cygwin comes with a working ssh server, and gitd would probably build under it too	04:24
ezyang	gitd would be the trick	04:24
	in a worst case scenario, you could even just http publish the repos	04:24
	(if you don't care about performance or pushes)	04:24
	→ perezd88 joined	04:24
	← rmh3093 left	04:24
	← careo left	04:24
	→ rmh3093 joined	04:25
	← perezd88 left	04:25
ArdRigh	gitd ... you mean git-daemon ?	04:28
	git --daemon	04:28
	?	04:28
mugwump	ArdRigh: see man git-daemon	04:29
ArdRigh	yeah I tested that on a linux server	04:30
	thanks for the help :)	04:30
	← ArdRigh left	04:30
	← drewr left	04:35
	→ careo joined	04:38
	→ muthu_ joined	04:49
	← madewokherd left	04:55
	← ahupp\|fb left	04:56
	→ Tv joined	04:57
	→ mczepiel joined	05:00
	→ ahupp\|fb joined	05:09
	← bieneff left	05:14
	→ NemesisD joined	05:24
	← tjafk1 left	05:27
	→ tjafk1 joined	05:27
	← k776 left	05:28
	← pjz_ left	05:32
	→ pjz joined	05:32
	← threeve left	05:34
	← jackdempsey left	05:36
	xaiki → xaik1	05:41
	→ EiNZTEiN joined	05:47
	← niki left	05:47
	← coderdad left	05:49
	← EiNZTEiN left	05:53
	→ cedricv joined	05:59
	← litage left	06:05
	loinclot_ → loincloth	06:09
	← haris left	06:12
	← qrush left	06:13
	→ reithi joined	06:17
	← aspotashev left	06:18
	← loincloth left	06:19
	→ iulian joined	06:21
	← ahupp\|fb left	06:24
	→ ahupp\|fb joined	06:25
	← mosno_ left	06:25
	→ digitalhobbit joined	06:28
	→ gambler joined	06:31
	→ agenteo joined	06:31
agenteo	can I undo a git branch -D ?	06:31
madduck	agenteo: yes, in most cases.	06:32
	you need to find the ID of the commit the branch was at	06:32
	git reflog might help	06:32
	then let's say you know the commit was 83df921, then you just do: git branch branchname Trusted relays that accept mail directly from dial-up connections should not be listed in "internal_networks". List them only in "trusted_networks".	06:33
	oops	06:33
	then let's say you know the commit was 83df921, then you just do: git branch branchname 83df921	06:33
agenteo	preferences is the branch that I've deleted, 7814884... HEAD@{13}: checkout: moving from master to preferences	06:35
madduck	git show 7814884	06:35
agenteo	does that text represent the branch creation?	06:35
madduck	maybe	06:35
	anyway, you don't want branch creation	06:35
	you want branch deletion	06:35
	find the last point where you moved away from preferences	06:35
	or maybe the deletion has its own entry?	06:35
	if all that fails, you can use git fsck to recover the dangling commits and then check them one by one	06:36
agenteo	doesn't look like there is an entry for the deletion	06:36
	last move out is 7814884... HEAD@{12}: checkout: moving from preferences to variants	06:36
madduck	does 7814884 represent the last commit on preferences?	06:37
	git show 7814884	06:37
agenteo	madduck: I suppose, it's following the text that I've posted before	06:39
madduck	I don't understand.	06:39
	← therrg left	06:39
agenteo	http://pastie.org/312858	06:39
	this is the git reflog	06:39
madduck	you've never made a commit to the preferences branch	06:40
	thus, git branch preferences 7814884	06:41
	is what you want.	06:41
agenteo	that's the weird bit... :) I'm pretty sure I did, but I might be wrong it was some test code nothing too important...	06:41
madduck	reflog dnies it.	06:42
agenteo	yeah... it's prob right	06:42
	madduck: r'u Myles?	06:42
	← meyering left	06:43
madduck	who's Myles	06:43
	→ meyering joined	06:43
agenteo	nevermind, he's somebody with a nickname similar to yours	06:43
madduck	there are many copies but only one original. I am the latter. :)	06:44
agenteo	eheh	06:44
madduck	had this nick since 1992	06:44
	→ Coontact joined	06:44
madduck	i don't know of anyone who had it longer.	06:44
reithi	moin	06:44
Coontact	According to the best selling book on race and intelligence, TheBell Curve, as a group, African Americans have a average mean IQ ofonly 85 points as compared to Europeans who have an average mean IQ of 100-103points.	06:44
	However, in some African countries, the Average Mean IQ Level is even lower--infact, lower than 60 points on average. This occurs in many out of the wayvillages along the Zaire River Basin. Fifty percent of these Africans areeither in the retarded or moron group.	06:44
agenteo	thanks for your help madduck	06:44
Coontact	Back in America, only 16% of African Americans have IQ's above 100 points,which is approximately what is required to achieve a mere D-/F+ in college.Graduation rates of a mere 32% at all-black Howard University bear thisout, and Howard University is definitely not a difficult school to attend.	06:44
madduck	Coontact: that's mighty off-topic here, isn't it?	06:45
Coontact	I am looking to upgrade my african american employees	06:45
	how do I do this?	06:45
madduck	Coontact: please take that elsewhere.	06:45
reithi	how do i push all off my local branches to the remote repo ?	06:47
madduck	read the manpage and discover 'git push --all' ? :)	06:47
Coontact	Hate niggers? Join us at http://www.niggermania.com and http://www.chimpout.org This is an alliance of Whites, Hispanics, and Asians against niggers. Paid for by Chimpout.org.	06:48
reithi	madduck: ups that was to easy thx	06:48
madduck	do not feed the troll guys	06:48
	← agenteo left	06:52
	→ ErrantEgo joined	06:52
	← SjB_ left	06:53
	← Pupeno-M_ left	06:54
	← GNUix left	06:55
Coontact	According to the best selling book on race and intelligence, TheBell Curve, as a group, African Americans have a average mean IQ ofonly 85 points as compared to Europeans who have an average mean IQ of 100-103points.	06:55
	However, in some African countries, the Average Mean IQ Level is even lower--infact, lower than 60 points on average. This occurs in many out of the wayvillages along the Zaire River Basin. Fifty percent of these Africans areeither in the retarded or moron group.	06:55
	Back in America, only 16% of African Americans have IQ's above 100 points,which is approximately what is required to achieve a mere D-/F+ in college.Graduation rates of a mere 32% at all-black Howard University bear thisout, and Howard University is definitely not a difficult school to attend.	06:55
listdata	it's obvious that this is a bot...	06:55
madduck	just ignore, i've alerted freenode stuff and they are on it.	06:56
Coontact	Only 2.3% of all African Americans have an IQ above 115 points, about D+/C-in college. Only 667,000 out of 29,000,000 African Americans measure above115 points as can be seen in the above illustration. This is hardly enoughIQ level in the distribution pattern to justify Affirmative Action Programs.There just aren't enough African Americans to fill the jobs available. Thatis why white women were elevated to "minority status" by affirmativea	06:56
	ction executives, even though white women are far more plentiful than whitemen who are themselves a minority by comparison.*	06:56
	If these facts are true, there is little hope that it can be helped by anymeans known to society.	06:56
	← warthog9 left	06:58
	→ moccuo joined	06:59
	→ daaku joined	07:00
	→ gilimanjaro joined	07:01
Coontact	Lick A Cow's Menstrual Waste to Grow Bigger Balls	07:02
	You've got to hand to those innovative niggers in Africa, huh? Where would the rest of the world be without their mind-boggling advances and discoveries? Who would have thought that licking and eating a cow's menstrual waste could grow bigger he-man sized balls?	07:02
	Now we have it! Video of nigglets in Africa sucking a cow's pussy and showering in cow pee ! Yes folks, this is how niggers live when left on their own away from the 'racism' of whitey.	07:03
	timulate Cows with Oral Sex to Get Them Ready for Mating and Have Sex with Donkeys to Grow Bigger Penises	07:04
	If this nigger innovation ever got out, sales of penis pumps would plummet and cattle herds would double..	07:04
	In a few societies, copulation with animals is reported for African boys, among the Tswana, Riffian, and Masai (Ford and Beach, 1951:p147, 148). Particularly herding boys may be more intimate with cattle than with the opposite sex. Nomad and Nuer herdboys are seen to drink milk straight from camel's udders (e.g., Pavitt, 1997:p147, ill.; Akalu, 1985:p46, ill.)[159]. Young boys are seen performing cunnilingus on cattle to stimulate the moti	07:04
	vation to mate (e.g., Nomachi, 1989 [1990:p45, ill.])[160]. "The shepherd-boys of the Tswana frequently have intercourse with their flocks, but are punished if caught in the act" (De Rachewiltz (1963 [1964:p283]). "Riffian youths who have not yet attained the age of puberty have intercourse with she-asses in order to get sexual capacity and to make the penis grow (Ford, 1945 [1964:p20]).	07:04
	http://www.niggermania.com/tom/lickingcows/niggerslicking.htm	07:05
moccuo	Godsize: hmm, so you only really get a benefit of distributed version control if you have a lot of developers?	07:05
ezyang	moccuo: Nope!	07:06
	DVCS is still helpful in a one developer team	07:06
	← carllerche left	07:06
	← Coontact left	07:06
ezyang	it means that you can go into laptop mode and still commit	07:06
	whereas if you are running SVN off a central server somewhere else, you're outta luck	07:06
	s/laptop/airplane/	07:06
moccuo	aah, gotcha	07:07
ezyang	in fact, that was one of the things that singularly bothered me before I moved to git :-)	07:07
	← ari-_-e left	07:08
moccuo	having a central location where everyone commits to and can access every revision seems like a good way to me though..	07:08
	Ratler_ → Ratler	07:08
ezyang	and generally speaking that's what happens	07:09
	repo.or.cz is a case-in-point	07:09
moccuo	aah, kk. so like github	07:10
	← ErrantEgo left	07:10
	→ Weaselweb joined	07:10
ezyang	yep	07:11
moccuo	so pretty much, if i want to start a working copy of a git repo. i would pull it from there (branch?). and then make commits as i go through (all locally) and then i merge when it's stable?	07:11
	im new to version control in general, and dont have any experience with cvs, svn, or git..	07:11
	so excuse my misunderstanding of things :)	07:12
ezyang	if you're talking about starting things off	07:13
	you'd just find the directory of stuff you want to version	07:13
	and git init	07:13
	that's it	07:13
moccuo	hmm	07:13
ezyang	that doesn't set up the remote stuff, of course	07:13
	but you should read some docs first before asking more questions :-)	07:14
moccuo	yea. but for a completely local repo (which is what im going to be doing now). that's pretty much it?	07:14
ezyang	yep	07:14
	→ EiNZTEiN joined	07:15
moccuo	cool, i watched the talk with torvalds. he just compares to the way cvs and svn did it, which is apparently so awful that everyone who uses it is "stupid" and "ugly"...	07:15
ezyang	:-)	07:16
	That's linus for you	07:16
moccuo	lol	07:17
shd	that talk is very funny and enjoyable	07:20
	→ Ingvae joined	07:32
	← bjeanes left	07:33
	→ ceej joined	07:36
	← bobesponja left	07:39
	→ warthog9 joined	07:43
	→ bentob0x joined	07:45
	→ jackdempsey joined	07:48
	→ idletask joined	07:49
idletask	Hello	07:50
	← ahupp\|fb left	07:50
listdata	i just built git 1.6.0.4 from source and installed it with make install, but git-gui is missing... where do i get the source for git gui?	07:50
	← NemesisD left	07:50
idletask	It is in	07:50
	Did you try "git gui"?	07:50
ezyang	it should be in your git checkout	07:51
listdata	idletask: haha, thx, that works fine	07:51
	i guess hyphenated commands are deprecated now?	07:51
idletask	listdata: the git-* series is deprecated	07:51
listdata	indeed	07:51
ezyang	ya	07:51
	(although you still have to man git-cmd)	07:51
listdata	i see, thx	07:52
idletask	ezyang: yep... It'd be nice if you had, say, gitcommit, gitadd and al	07:52
	Would allow for easy completion :p	07:52
thiago_home	completion is easy	07:53
ezyang	idletask: well, I think someone was working on a completion ext for git	07:53
	I don't know if it's landed in next yet	07:53
thiago_home	completion works just fine on zsh	07:53
idletask	Well, yes, but then why require bash_completion to have it?	07:53
	Or zsh completion system for that matter	07:53
	If you can have it with plain bash...	07:53
thiago_home	git[space]co[tab] <-- shows several possibilities	07:54
drizzd_	idletask: because plain bash is stupid	07:54
thiago_home	including "co -- alias for checkout"	07:54
drizzd_	git expansion also does argument and option expansion	07:54
ezyang	ya; works for me	07:54
idletask	drizzd_: maybe, but it's in use by 99+% of Linux users	07:54
thiago_home	that's because it's the default	07:55
idletask	And nowadays, by more than 90% of Unix-like users in general, I'd say	07:55
thiago_home	most people I know who learnt about zsh and used it, like it and don't look back	07:55
idletask	thiago_home: I'm not one of these "most", then :p	07:55
thiago_home	but let's not discuss it... I guess it's a flame war subject	07:55
	→ bjeanes joined	07:57
	← digitalhobbit left	08:01
idletask	I was just saying that it would be easy enough to make it easier for the common case... And as of yet, the common case is plain bash	08:01
	"Stupid" as though it may be	08:01
	→ charon joined	08:02
thiago_home	it's not difficult to write bash completion code	08:03
	← jackdempsey left	08:03
	→ markelikalderon joined	08:04
madduck	fwiw, I like this: http://xana.scru.org/xana2/quanks/topgitincompletion/	08:07
	it's how I will work with zsh people to implement topgit completion, and I hope that bash will also use the same route, if it can	08:07
	and yes, zsh is better. feed the troll! :)	08:07
	madduck <-- bad	08:08
ezyang	it seems to me	08:09
	that autocompletion info should be data and not code	08:09
	so zsh approach in this case seems to be the correct one	08:09
	← Ingvae left	08:09
idletask	ezyang: what do you mean? Back when I used zsh, I remembered that ./configure --<TAB> invoked ./configure --help, is that code or data for you?	08:10
	← jas4711 left	08:10
	→ jas4711 joined	08:11
ezyang	so, bash_complete looks like a bash script that sets up appropriate functions for autocompletion to work	08:12
	← listdata left	08:12
ezyang	whereas zsh looks at some machine-parseable output of the command to figure out what's going on	08:13
	I probably am misunderstanding?	08:13
idletask	Can't tell, I haven't used zsh for a long time and don't use bash_completion (yet?)	08:16
madduck	idletask: how can you possibly use linux without completion?	08:18
idletask	madduck: I use "classical" bash completion	08:18
madduck	i mean, i am jealous. you are going to experience a revelation. an enigma. it'll be awesome.	08:18
	ah. :)	08:19
	still...	08:19
idletask	I mean, most people don't even know about C-x $, C-x ! and C-x ~	08:19
ezyang	eh, I thought TAB was completion?	08:19
madduck	are those !$, !! and !~?	08:19
idletask	madduck: no	08:19
	Most people also don't know about Esc, C-e as well	08:20
	← pygi left	08:20
madduck	oh, i see. type-specific completion.	08:20
	← ezyang left	08:20
madduck	Esc, C-e does nothing here	08:21
idletask	madduck: it expands subcommands, env vars	08:22
	With zsh, that's only a tab away, sure	08:22
	← normalperson left	08:23
idletask	Anyway	08:23
madduck	ic	08:24
	← ciaran left	08:26
	→ stephenos joined	08:27
	← Tv left	08:28
	← tango_ left	08:29
	← towski left	08:30
	→ Tv joined	08:36
	→ girishr joined	08:36
	← careo left	08:39
	→ aFlag joined	08:40
aFlag	hi. I started changing a few things on my working tree, but then I regreted. I want to save my changes to a new branch so I can get back to it later on, but now I want to do something else. How should I proceed?	08:41
	→ girishr_ joined	08:47
	→ odi joined	08:48
	← abbe left	08:49
	→ cilly joined	08:51
aFlag	if I could somehow commit to a different branch	08:54
	← girishr left	08:54
	→ tvw joined	08:54
jayne	git checkout <good sha1>	08:54
	that will put you on a detached head, from which point you can create a new branch	08:55
	with git branch	08:55
	← CodeOfficer left	08:55
jayne	or git checkout -b	08:55
	merge the changes from master or wherever you didn't want them into this branch	08:55
	then checkout master or wherever you didn't want the changes and git reset --hard	08:56
	→ girishr joined	08:56
aFlag	hm, I just found out about git stash, I think it will work well here :-)	08:56
jayne	oh, you haven't committed?	08:56
	← girishr_ left	08:57
aFlag	no	08:57
jayne	then you can just branch from where you are and commit... but stash works too	08:57
thiago	aFlag: just checkout another branch, commit there, then checkout back	08:57
	→ oops6_4 joined	09:02
	→ boud joined	09:02
oops6_4	Hi I am using git from a log time for my projects but this time I face a problem. when I commit to git it takes long time then the normal is there any way to make it proper	09:04
	→ girishr_ joined	09:05
	← boud left	09:05
	→ lucsky joined	09:06
	→ towski joined	09:06
lucsky	'morning	09:06
	→ botanicus joined	09:08
cedricv	oops6_4: you need a "git gc" maybe	09:08
	→ ciaran joined	09:11
	← daaku left	09:11
	← aggieben left	09:11
	← girishr_ left	09:13
	← girishr left	09:14
	→ girishr_ joined	09:15
	→ cradek_ joined	09:16
	← cradek left	09:16
	← ceej left	09:17
	→ parasti joined	09:18
	→ ceej joined	09:18
	→ naeu joined	09:21
	← mczepiel left	09:21
	→ bschindler joined	09:21
	← dreiss left	09:21
bschindler	hi - in subversion there are those $ Revision ... $ tags, is there something similar in git?	09:21
	← bjeanes left	09:23
	← ridiculousRay left	09:24
idletask	No	09:26
	← Lyfe left	09:26
idletask	Git doesn't track individual files	09:26
	It tracks the repository as a whole	09:26
	→ bjeanes joined	09:27
bschindler	I know git tracks content. That's why I asked here	09:28
	but thanks	09:28
	← Tv left	09:30
	← cilly left	09:30
idletask	There is an ongoing discussion on the git ml about these	09:30
charon	bschindler: you can use export-subst during git-archive, see man gitattributes	09:31
	http://www.kernel.org/pub/software/scm/git/docs/gitattributes.html (where'd the bot go?)	09:31
	← girishr_ left	09:32
	← quoin left	09:32
	→ pcapriotti joined	09:35
	← ceej left	09:36
	→ Tv joined	09:36
	→ mosno joined	09:38
	→ pavelgolubev joined	09:40
	← stephenos left	09:42
	← cradek_ left	09:43
	← pavelgolubev left	09:48
	→ priidu joined	09:49
	→ naeu_ joined	09:50
	← naeu_ left	09:50
	← bartman`` left	09:56
	← naeu left	09:57
	→ shodges joined	09:57
	← shodges left	09:58
	→ koke joined	10:03
	dwmw2_gone → dwmw2	10:07
	→ pygi joined	10:11
	← cykl left	10:17
	→ cilly joined	10:21
	← doener left	10:22
	→ gour joined	10:27
gour	vmiklos: hi, i'm testing darcs-fast-export with bzr, but errors are all i get	10:28
	vmiklos: see e.g. http://rafb.net/p/x8IJQX57.html	10:28
kupsch	charon: the gitattribute 'ident' is really nice. isn't there something for the HEAD hash as well?	10:30
charon	only at git-archive time, where TFM says it's $Format:%H$	10:31
	(i don't use any of those features)	10:31
	← ciaran left	10:32
kupsch	cheers	10:32
gour	vmiklos: similar with git - http://rafb.net/p/G8X3L141.html	10:33
cehteh	morning gour	10:33
gour	cehteh: morning	10:33
	→ ciaran joined	10:36
	→ nud joined	10:37
	← hobodave left	10:39
	← odi left	10:39
	→ hobodave joined	10:39
	→ Numbers joined	10:40
	← iulian left	10:40
	→ hobodave_ joined	10:41
Numbers	has 'Already up-to-date. Yeeah!' always been there, or did 'Yeeah' just magically appear? :/	10:41
kupsch	you could try to bisect the git sources and find out yourself	10:43
Numbers	heh, I could.. but I probably wouldnt understand the code and b) I don't have time at the minute :P	10:44
Arjen	It was introduced in 1c7b76be7d620bbaf2e6b8417f04012326bbb9df	10:44
Numbers	lol	10:45
parasti	no, it was in the shell script before that	10:45
	→ metric joined	10:45
	← mosno left	10:46
	← cilly left	10:47
	← imyousuf left	10:47
charon	oldest i can find is 839a7a06f3 from april 2005... i guess it's always been there	10:47
Arjen	Yup	10:48
Numbers	Must be my bad eyesight then	10:48
	→ namenlos joined	10:48
Numbers	it miraculously sparked an hour long debate this morning in the office :D heh	10:49
	← namenlos left	10:49
	← pygi left	10:49
kupsch	your bad eyesight or the 'Yeaah!'?	10:49
	→ odi joined	10:49
Numbers	the 'Yeaah!'	10:49
	seems nobody saw it	10:50
charon	not all code paths have that apparently, several just say 'Already up-to-date.'	10:50
	→ girishr joined	10:51
	← bschindler left	10:52
	← girishr left	10:54
	→ girishr joined	10:56
	← hobodave left	10:58
charon	Numbers: TFS says that it uses the Yeeah! only if you requested an octopus merge	10:59
	cute :)	10:59
	→ FunkeeMonk joined	11:04
vmiklos	gour: what does darcs check say?	11:05
	IOError: CRC check failed 0x327ba8f != 0xd257b5eL	11:05
	← stuffcorpse left	11:05
	→ stuffcorpse joined	11:06
vmiklos	sounds like a corrupted patch	11:06
	← Garoth left	11:06
dmiles_afk	my remote ssh repo is Writing objects: 3% (2524/76037), 14.30 MiB \| 37 KiB/s .. rsync gets 168k.. should i rsync it?	11:06
cehteh	dmiles_afk: git gc	11:07
Numbers	charon, cheers	11:07
dmiles_afk	cehteh, oh right, i think that will shave off 200mb	11:07
cehteh	:)	11:08
	→ girishr_ joined	11:08
	→ ph^ joined	11:08
dmiles_afk	ok i see it getting 157-173kbs now	11:08
	→ not-xjjk joined	11:09
cehteh	what filesystem btw?	11:09
dmiles_afk	ext2	11:09
	nomrally i use ext3 but i dont trust this hd	11:09
	→ Kbyte joined	11:09
	→ scientes joined	11:10
dmiles_afk	funny, i find more often i lose ext3 journals then have unrecoverable ext2s	11:10
	→ quoin joined	11:11
	← girishr left	11:11
dmiles_afk	1.7G 2008-11-12 02:59 gitzip.zip	11:13
	hrrm right now i am git push --all dmiles@denton_tx:/usr/shared/repos/javaCycGit .. i wonder which would be quicker	11:14
	also was it right to make it like: git init --bare --shared ?	11:16
	on denton_tx	11:17
	→ scientes_ joined	11:17
dmiles_afk	also i am going to 'git gc' in a fs copy of it just to see really how much it shrinks	11:18
	← scientes left	11:18
gour	vmiklos: all repos are consistent...	11:20
	→ mosno joined	11:22
	→ cilly joined	11:22
gour	vmiklos: with one very small repo it works, but i've problem with --export-marks=file option, anything wrong in invocation ' darcs-fast-export.py --export-marks=./marks ~/repos/darcs/sweh \|(cd sweh.bzr; bzr fast-import -)' ?	11:22
	← xjjk left	11:23
	← girishr_ left	11:24
	← moccuo left	11:24
dmiles_afk	are packs temporary or are they the actual content?	11:26
	→ awarde joined	11:27
	→ lorandi joined	11:27
	← cilly left	11:28
	→ moccuo joined	11:28
	← muthu_ left	11:28
dmiles_afk	for updating or remote syncing it makes .pack to upload.. then it removes them?	11:28
	← xenoterracide left	11:29
	→ xenoterracide joined	11:30
	→ aspotashev joined	11:32
	← ciaran left	11:33
	→ Sho_ joined	11:34
	← towski left	11:36
	← xenoterracide left	11:39
parasti	dmiles_afk: there's some info about packs in man git-pack-objects	11:40
	→ xenoterracide joined	11:40
parasti	dmiles_afk: and maybe even in the manual, but I haven't looked	11:40
	← zewerson left	11:40
	← popol left	11:41
	→ zewerson joined	11:42
dmiles_afk	parasti, thanks, i decided to instread of just pushing to a remote system takes a few hours.. i'd create a LAN site to push to .. i am doing this from a differnt .git dirrecory as to not confuse the process...	11:42
	← selckin1 left	11:42
dmiles_afk	fatal: pack has bad object at offset 1596795887: inflate returned -3 \n error: pack-objects died with strange error \n error: failed to push some refs to 'root@localhost:/javaCycGit/'	11:43
	now i am trying to find reading material to learn how to repair this	11:44
	→ gitte joined	11:46
	→ cilly joined	11:46
dmiles_afk	i think in the past someone was pushing stuff to me with a network error, or this was from a bad fsck	11:46
	so maybe a git checkout each version and commit each version to a new .git would eventaully fix this?	11:48
corecode	does somebody have a pointer on how I could split one root/branch into two, using filter-branch, i guess	11:48
gitte	corecode: depending on how specific you want to be, it might be a better idea to use fast-export && edit && fast-import	11:49
corecode	mhm	11:49
	that means i should hack it into the repo conversion instead of post-processing it	11:49
	i can do that as well	11:50
gitte	No, if you want automated, it is better to use filter-branch.	11:50
corecode	it is a one time thing	11:51
	how would I do that with filter-branch?	11:51
gitte	You'd need some sort of "parent filter" and a function mapping original parents to desired parents.	11:51
corecode	that isn't clear to me	11:51
	okay	11:51
gitte	parent filter should be pretty intuitive, no?	11:52
	Just look for --parent-filter in the man page for filter-branch; that explains it better than I ever could.	11:52
corecode	yah, i just got confused by the description that the children would then get the parents assigned automatically	11:52
	how is the run time of filter-branch vs fast-import?	11:52
gitte	I do not see the word children there. But maybe I am too dumb.	11:53
corecode	same? faster/slower?	11:53
gitte	Depends. On a lot of factors.	11:53
	Use of RAM disk.	11:53
	Which filters you use.	11:53
	← cilly left	11:53
corecode	yah, ram disk...	11:53
gitte	Whether fast-export is appropriate at all (if you have to fear that your automated edits bork the syntax, it is not for you!)	11:53
	Then it comes down to the filters you use.	11:54
corecode	nono, i would hack it into fromcvs/togit	11:54
gitte	tree-filter being the most expensive one, of course.	11:54
	corecode: I kinda assumed that.	11:54
	→ johnflux joined	11:54
johnflux	Hey all	11:54
	I have a file, and I want to know who added/changed a particular line	11:55
	→ boto joined	11:55
johnflux	how can I do this please?	11:55
corecode	blame	11:55
	← xenoterracide left	11:55
johnflux	neat - thanks	11:55
	← Kbyte left	11:55
	→ Kbyte joined	11:56
	→ pygi joined	11:57
	→ selckin1 joined	11:58
	→ ciaran joined	11:59
	→ xenoterracide joined	12:03
	→ fisons joined	12:04
gitte	corecode: did you manage to write that parent-filter?	12:04
corecode	still waking up	12:05
	mainly idling	12:05
	so might take a bit :)	12:05
	now thinking of it it shouldn't be too hard	12:05
	i just need to find a way to find out how to split the branch automatically	12:05
	i.e. look at the changed filenames or so	12:05
	or do that before	12:06
	← aspotashev left	12:07
wanders	Is there some command to get git repo root?	12:07
	← gilimanjaro left	12:08
	→ Ryback_ joined	12:08
moccuo	if all commits in git are local then how are we able to access them in github for example?	12:09
	"commit history"	12:10
	are there different kind of commits that go to a central server?	12:10
	← bjeanes left	12:10
madduck	no, github is also a local repository	12:11
	(albeit a bare one)	12:11
	github then simply has a web interface allowing you to view this local repository	12:11
	so you just push your commits to the other repository	12:11
	from your perspective, it's remote	12:11
	but if you took the .git directories on both sides, they would have the same contents	12:12
	(almost)	12:12
	→ hrajchert joined	12:12
	→ kermit_ joined	12:13
	← Flathead left	12:15
	← LuboC left	12:15
wanders	To answer my own question. "git rev-parse --git-dir"	12:17
	→ komplex joined	12:17
	← koke left	12:18
corecode	gitte: i have the feeling that filter-branch is definitely slower	12:18
	i get 2 commits/sec	12:18
	← xenoterracide left	12:18
hrajchert	Hi, i have a doubt about normal use of git. I have a project that has 2 branches, master (or stable) and myTestingBranch (that rather than new shiny code has things to help portability), i found myself with a bug (havent quite found the solution) in both branches but its easier for me to work from myTestingBranch... continues...	12:19
corecode	hrajchert: best is to fix the bug where it appeared and merge that into both branch heads	12:20
	→ xenoterracide joined	12:20
hrajchert	Because i dont want to mess up the myTestingBranch i would like to create a branch from there, try all this wacky things to find and solve the bug, then clean up that branch and merge it with myTestingBranch... but the bug fix is sure to be put in master, how can i do that? i dont want to merge myTestingBranch and master, just the bug fix part	12:21
	corecode: the bug is in both branches	12:21
corecode	yah	12:21
	but did it appear separately	12:21
	or did it appear before testing branched from master	12:22
johnflux	hrajchert: you can merge just a single commit	12:22
hrajchert	it appear when in the myTestingBranch after ive updated some external dependencies...	12:23
kupsch	hrajchert: I would do a git rebase --interactive and only pick the commits that actually fix the bug	12:23
hrajchert	johnflux you are saying the same as kupsch right? ill look how to do that	12:24
johnflux	hrajchert: yes. you don't need to look how to do that, he just told you how to do that ..	12:25
corecode	gitte: i'm thinking of doing a fast-import of only the commits, instead of running a filter-branch, and use the old pack for tree/blobs	12:25
hrajchert	johnflux: i mean that im going to test that out on a sandbox project :P	12:26
johnflux	oh right :)	12:27
	not-xjjk → xjjk	12:27
	→ bjeanes joined	12:28
	→ gerontosepp joined	12:28
	cbrake_away → cbrake	12:29
	← mosno left	12:30
	← gerontosepp left	12:31
	← EmilMedve left	12:32
jnl_	helo, im a bit confused about some output from git ls-tree	12:34
	mkdir -p x/d x/e; echo foo > x/f1 ; echo bar > x/d/f2 echo abr > x/e/f3	12:34
	git init -q ; git add . ; git ci -q -m . ; git ls-tree -d HEAD x x/d	12:34
	the result im getting is that i lists x/d and x/e	12:34
	i was sort of expecting it to list x and x/d, and to nt list x/e... or have i misunderstood the -d option?	12:34
	→ LuboC joined	12:34
jnl_	s/nt/not	12:35
	→ juanjoc joined	12:36
	← harinath left	12:41
	← selckin1 left	12:43
	← tvw left	12:44
	← grahal left	12:44
	→ Pupeno-M joined	12:44
	→ grahal joined	12:44
	← gretch left	12:45
	→ tvw joined	12:46
	→ namenlos joined	12:46
	→ Pupeno-M_ joined	12:48
	→ selckin joined	12:49
	→ Arafangion joined	12:50
Arafangion	Hey, what's wrong with my commit input to git-fast-import?	12:50
	http://rafb.net/p/YrsY5U57.html	12:50
madduck	jnl_: very weird.	12:51
	sorry, no idea	12:51
idletask	gitte?	12:51
	← tvw left	12:51
	→ tvw joined	12:52
Arafangion	Oh stupid.	12:53
	<censored> python inserting newlines when I didn't want it.	12:53
jnl_	madduck: it happens when one of the direcotries are a subdir of the other. if i do them one by one, or for example x/d and x/e, it works a i expected..	12:53
	← oops6_4 left	12:53
moccuo	can you merge a branch if it has a file that has also since been modified on the master branch?	12:55
	or is that why im getting a merge conflict? :)	12:55
Arjen	Fix the conflict and commit	12:57
Arafangion	What's wrong with this input to fast-import? http://rafb.net/p/GpK4H155.html	12:57
	I am currently stumped.: (	12:58
moccuo	Arjen: yes i know. but does it always conflict?	12:58
madduck	Arafangion: patience. if noone answers, noone knows.	12:58
Arafangion	Apparently I haven't provided a 'committer', but it's right there on line two!	12:58
Arjen	2committer ?	12:58
Arafangion	madduck: Alternatively, they didn't respond because I didn't provide enough info.	12:58
	Arjen: '2' is my mark.	12:58
Arjen	What's the error message?	12:59
Arafangion	fatal: Expected committer but didn't get one	12:59
	Ok, seems that marks are described below, it's not just a number, but has it's own format, thanks Arjen.	13:00
Arjen	Yeah, mark:$num	13:00
	moccuo: Has the file been changed on both branches?	13:00
moccuo	Arjen: yes	13:01
Arjen	moccuo: If the changes are in the same area on both branches, then it will conflict (obviously :-)	13:01
	What do you mean by 'always'?	13:01
moccuo	Arjen: i thought it would only conflict if it was the same line or something.. lets say for example in one brance i changed something in the <head> of an html document. and in the other branch i changed something in the middle somewhere.. it will still conflict?	13:02
jnl_	i think i have "fix" for it, just not sure if the original code is doing what was intended to or not	13:03
Arjen	moccuo: Probably not	13:03
	← Pupeno-M left	13:03
moccuo	Arjen: heh, well that's what i did.	13:03
	i'll try it again. might have done something wrong. after all its my first time playing with a VCS	13:04
Arjen	Heheh :-)	13:04
	← Modius left	13:05
gitte	idletask: yes?	13:07
	→ mw\|out joined	13:07
idletask	gitte: saw my last mail? The question that buggers me for now is "why twice the number of lines in the source file" (for the hashmap size). I can't get my head around this one... I mean, this has to have some sort of mathematical background or something, and I don't know enough about mathematics to understand your argument	13:09
	Second: what do you think about my "building a (dest ordered) list of lines using a list_head" stuff?	13:10
Arafangion	What's the difference between .git/branches and .git/refs?	13:14
	What /is/ .git/branches, rather?	13:15
	→ girishr joined	13:15
	mw\|out → mw	13:17
	→ icwiener joined	13:18
Arafangion	I'm trying to work out how I should store a "remote" branch.	13:19
	→ aspotashev joined	13:19
charon	6687f8fea22	13:19
gitte	idletask: a hashmap is typically twice the size of expected entries so that the runtime has a chance to get optimal.	13:19
	→ whee joined	13:19
gitte	idletask: remember, expected runtime for insertion and lookup is _constant_.	13:20
	→ EmilMedve joined	13:20
gitte	idletask: you can't have that in a next to full hashmap, though, for obvious reasons.	13:20
charon	Arafangion: meh, the bot is offline. but the message of 6687f8fea22 explains a bit of that	13:20
gitte	idletask: as to the second point: wasn't there some explanation as to how you could get at the longest common sequence on Alfredo's blog?	13:21
Arafangion	charon: You just memorized the hash?	13:21
charon	Arafangion: no, i used 'git log --grep=".git/branches"'	13:21
Arafangion	charon: Thank goodness. :)	13:21
idletask	gitte: not really... But I understand now about the first point	13:22
	→ vasandgvd joined	13:23
	← Lemurnomicon left	13:24
vasandgvd	how can i see the diff of a file after a commit?	13:25
	→ Lemurnomicon joined	13:25
Arjen	git show $commit	13:25
idletask	gitte: there is some talk about this, but it is blurry at best	13:25
	← pygi left	13:25
Arjen	vasandgvd: git log -p	13:25
	← icwiener left	13:26
gitte	idletask: lemmesee	13:26
	← schlort left	13:27
	→ SjB_ joined	13:27
vasandgvd	idletask: thank you!	13:27
	→ girishr_ joined	13:27
gitte	vasandgvd: you meant Arjen.	13:27
idletask	What I was about to say	13:28
vasandgvd	yeap... sorry! thank you Arjen!	13:29
Arjen	vasandgvd: :-) np	13:29
	→ dlowe joined	13:29
idletask	gitte: it gives me enough information to add comments :p This is my pet peeve... I want to know why it was designed/coded this way, and this applies to what I code as well	13:30
	Even if it's thrown away at the end - at least I know I have understood what I coded	13:31
Arjen	You really appreciate the code archeology capabilities of git when you have to use other systems and search for things in the history	13:31
	→ Zimm joined	13:32
gitte	idletask: sure, but for example this comment: "@newrec: a pointer to a xrecord_t" does not add any information to what is present in the function signature anyway.	13:32
	idletask: that's what I meant with "adding too much documentation". If it only clutters otherwise readable code, it should not be there.	13:32
idletask	gitte: bah, you can't get it perfect the first time, let me some time :p	13:33
	The day when I fully master what I code, the comment will be different, promise	13:34
gitte	idletask: course I do! Why do you think I try to answer your questions? :-)	13:34
	← girishr left	13:34
gitte	idletask: http://alfedenzia.com/images/patience.gif	13:34
	idletask: "The algorithm builds a set of piles, with each card being placed on the left-most pile whose top card is greater than it. When a card is placed on a pile, a backreference is made to the top-most card on the previous pile, if any. At the end of the algorithm, we work backwards from the top-most card on the right-most stack, thus building a longest increasing subsequence."	13:34
	gitte is sorry for the long lines	13:34
idletask	gitte: I'm not even there, I'm at the diagrams right above where it says that a preamble input is the sequence of lines ordered by what is seen in the destination	13:35
	Once I get that right, I'll see about the piles	13:35
moccuo	Arjen: hmm, it worked this time. strange.	13:35
idletask	(and probably change my inner loops once more)	13:35
	And, BTW, I'm still interested by a testing environment	13:36
	← SjB_ left	13:36
	← binjured left	13:36
gitte	idletask: where do you read the word "preamble" on that page?	13:36
idletask	I haven't even had the time to build one :/	13:36
	gitte: I understood "preamble" as "necessary step before going further"	13:36
gitte	idletask: ah, that's what you mean.	13:37
	→ girishr__ joined	13:37
	→ deskin joined	13:37
idletask	Yes, "sorry"	13:37
gitte	idletask: np	13:37
	idletask: but the sorting is pretty easy: you just keep a "next" pointer in the hashmap entries.	13:37
idletask	Which means, I want to get up to point n before going n+1, even if going n+1 means what I understood as "going to n" is wrong in the first place :p	13:37
gitte	idletask: you'll need a prev pointer, too.	13:38
idletask	gitte: and that's where I say that using list.h is WAY easier	13:38
gitte	idletask: but it is not necessary, and like you said, it only works on Linux.	13:38
idletask	For the reasons I outlined already	13:38
	No, it only works on _gcc_	13:38
gitte	idletask: besides, double-linked lists are sooooo easy.	13:38
	→ koke joined	13:39
gitte	Does not matter. gcc, Linux, does not matter.	13:39
	→ binjured joined	13:39
idletask	gitte: and what I say is, turning from gcc only to "not only gcc" is doable	13:39
	I've done it in the past	13:39
	The only thing that needs changing is container_og	13:39
	container_of	13:39
gitte	static void remove_entry(struct entry *entry) { entry->prev->next = entry->next; entry->next->prev = entry->prev; }	13:39
	← Zimm left	13:40
idletask	gitte: and you lose the generality	13:40
	And you don't know what the head is	13:40
gitte	idletask: why do I need any generality? I don't!	13:40
idletask	list.h doesn't have this problem	13:40
	gitte: today, probably no	13:40
	Tomorrow?	13:40
gitte	idletask: sure, go ahead, use list.h. I do not care, I'll just hack it to my likings on this end.	13:40
idletask	I mean, this is a state of the art linked list implementation, which only requires a little hacking to make it work on non gcc based compiled environments	13:41
	Why not use it?	13:41
Arafangion	My goodness, it's insanely easy to write a fast-import importer.	13:42
	→ Zimm joined	13:42
	← girishr_ left	13:43
idletask	gitte: see, you have to define a function to remove an entry from a list - with list.h, _you don't even have to_	13:45
	It's builtin	13:45
gitte	Arafangion: isn't it?	13:45
	idletask: besides, I have no idea why we discuss technicalities instead of the algorithm.	13:46
Arafangion	gitte: It is, still figuring things out, but I'm impressed.	13:46
	→ girishr_ joined	13:46
	→ coderdad joined	13:46
idletask	gitte: because this might help in the long run, that's why - and also, the fact that list.h has a head allows me to build a dest-ordered list from which to extract subsequences	13:47
gitte	Arafangion: say thanks to spearce, he invented the format.	13:47
	→ ankit9 joined	13:47
idletask	Because I only have to define the head _once_	13:47
	← ankit9 left	13:48
	→ coderdad_ joined	13:48
	← coderdad left	13:48
	→ LiamH joined	13:48
	→ ankit9 joined	13:48
idletask	gitte: see my mail again about building an ordered list	13:48
	From this list, building the rest is easier	13:48
	But I understand now the argument for twice the number of source lines for the hasmpa - key distribution	13:49
	The thing is, at the end, since you have all relevant entries linked as a list, AND the list is ready-ordered for the rest, the next steps will be even easier	13:50
	← gour left	13:53
gitte	idletask: and here I disagree. The next steps will be easier once we have an understanding how to get the longest common sequence.	13:53
	idletask: and then we can ask ourselves what data structure to use.	13:53
	idletask: and whether to have next/prev pointers that we can override in meaning at will.	13:53
Arafangion	Hmm, I'm getting _many_ git commits that don't have associated files. (Each file is empty)	13:54
	← girishr__ left	13:54
Arafangion	That is, I'm committing empty dummy files, but not all commits have such files, when they should.	13:54
gitte	Arafangion: guess: they are identical, both in content and name?	13:54
Arafangion	gitte: Yikes, git's that smart?	13:55
	→ roolio joined	13:55
Arafangion	gitte: They are identical, yes, both in content and name, however the first commit should have two files (which are empty and subsequently rewritten in a later checkin).	13:56
	But the first commit doesn't.	13:56
	In essence I have: http://rafb.net/p/Em9guJ96.html	13:57
idletask	gitte: OK, I think I need to read a little more about that LCS problem... But can you do me a favour and provide the way to do a test case? I want at least to see whether I can obtain the output from step 2 (dest ordered output)	13:57
Arafangion	(That's a portion, showing the first and part of the second commit)	13:57
idletask	gitte: once I'm there, I may have a better understanding of the rest, and why it's needed	13:58
	→ Pupeno-M__ joined	13:59
gitte	Arafangion: heh, I bet you verified with "git log --stat"	13:59
Arafangion	gitte: gitk, actually.	13:59
gitte	Arafangion: try "git log -p"	13:59
	→ jm joined	14:00
idletask	And, well, I just think I have figured out how to give up finding an LCS :p	14:00
Arafangion	That's more like it. Why don't they show up in gitk, though?	14:00
	→ nice_burger joined	14:01
	→ stampp joined	14:01
	← quoin left	14:02
stampp	Hey guys -- how do I unstage a 'git rm' ?	14:02
	→ loincloth joined	14:02
	← girishr_ left	14:02
	← johnflux left	14:03
idletask	gitte: guess what? I have the solution for the longest subsequence problem, I think.. And it involves list.h	14:04
	← loincloth left	14:04
	idletask thinks he'll have a go as the test case by himself :/	14:05
gitte	idletask: I am writing the test case for you right now!	14:07
	← bjeanes left	14:07
Arafangion	In gitk, why doesn't it show the files that were modified by the first commit, if one is looking at the first commit?	14:08
gitte	idletask: just have a look at http://repo.or.cz/w/git/dscho.git?a=shortlog;h=refs/heads/patience	14:08
	idletask: if you want to work on it, I have enough other stuff to do.	14:09
Arjen	stampp: git add ?	14:10
	← jas4711 left	14:10
gitte	idletask: okay, until you answer the question whether you want me to continue to work on it, I will work on something else.	14:10
	← stampp left	14:10
	→ harinath joined	14:10
	→ jas4711 joined	14:11
	→ bjeanes joined	14:11
	← aspotashev left	14:11
	← harinath left	14:11
	→ farrukh_najmi joined	14:12
	→ bschindler joined	14:12
	→ harinath joined	14:12
	→ thebolt joined	14:12
	→ schlort joined	14:14
	← Pupeno-M_ left	14:15
idletask	gitte: as I've said earlier, I have yet to understand the full algorithm, and I've only understood 1..n and not n..n+1, so maybe I'm completely misguided, maybe not	14:15
	← scientes_ left	14:15
idletask	But dismissing list.h is... Well... Shortsighted imho	14:15
gitte	idletask: I do not care about list.h at the moment. For me, this is not an issue right now. The issue right now is to get a hashmap going. Then to verify that we find all unique lines in text1. After that, we can go further.	14:16
	→ bartman`` joined	14:16
gitte	idletask: so do you want to make the test case yourself? I'm fine with that.	14:16
idletask	gitte: no	14:17
	You understand libxdiff much better than I do, and in particular its interface	14:17
	I don't know right now how to "transform" two files into two xdfile_t and then an xdenv_t :(	14:18
	← cliffstah left	14:18
idletask	If you can spare the time, or have done this already, this is what I'd like :(	14:18
gitte	idletask: have you looked at the link I posted 10 minutes ago?	14:18
idletask	Not yet, not the time yet :(	14:18
	gitte: if it's there already, tell me - I won't bother you anymore and will have a look	14:19
gitte	idletask: http://repo.or.cz/w/git/dscho.git?a=shortlog;h=refs/heads/patience	14:20
	idletask: just have a look at the newest commitdiff.	14:20
madduck	yay, patiencediff	14:22
	go gitte go!	14:22
	→ quickdev joined	14:22
quickdev	Hey	14:22
	← kermit_ left	14:23
	Arafangion has figured out his issues.	14:23
gitte	madduck: hey, haven't seen you in a long, long time!	14:23
	→ softdrink joined	14:23
gitte	Arafangion: what were they?	14:23
	← jm left	14:24
Arafangion	gitte: For now, using an incrementing integer for file data, so that it's easy to see how git stores the file, etc - real data will be used later - as for gitk, it seems that the 'Tree' lists the files in question - but not patch. Presumeably, if there is no previous commit, it won't bother generating a patch. (Which, imho, is inconsistent with other tools, but anyway)	14:25
	gitte: I'm also tagging each and every commit for good measure, so that it's easy to determine which commits correspond to which p4 commits.	14:25
	→ scientes_ joined	14:25
	← willb left	14:26
	← EiNZTEiN left	14:26
gitte	Arafangion: I think that might be an issue in the --stat code that does not detect +-0 patches (i.e. it ignores the creation of an empty file)	14:26
madduck	gitte: phd.martin-krafft.net keeping me busy.	14:26
	gitte: you ought to enjoy the time while I am not here and charge up for my return! :)	14:26
	madduck is the general troublemaker, HARHARHAR	14:26
gitte	madduck: where's the git repository with the .tex files?	14:28
madduck	gitte: ssh://git.madduck.net/srv/git.madduck.net/phd/thesis.git	14:28
	→ gasbag joined	14:28
Arafangion	gitte: I note that git-fast-import is using alot of memory, though, should i be concerned about that?	14:28
madduck	sorry, gitte: ssh://git.madduck.net/srv/git/git.madduck.net/phd/thesis.git	14:29
	without the *'s	14:29
Arafangion	16881 arafangi 23 0 457m 427m 18m R 89.9 67.7 7:47.61 git-fast-import	14:29
	That's the line from top for the memory usage.	14:29
gitte	madduck: bah, no git:// URL? ;-)	14:29
Arafangion	git-fast-import is definetly the bottleneck. :(	14:29
madduck	gitte: if you really wanted it, i could consider it. but then I would expect feedback! :)	14:30
gitte	Arafangion: AFAIR you can limit the memory, but the more you give it, the faster it runs.	14:30
	← ankit9 left	14:30
gitte	madduck: ouch.	14:30
	→ ankit9 joined	14:30
madduck	gitte: or just wait, it won't be long until it's done and then it'll be all public and Free, including the data.	14:30
	there's a lot of /git/i matches in the data. :)	14:30
	→ EmilMedve1 joined	14:31
	← rmh3093 left	14:31
gitte	madduck: well, I'd be most interested in a short version, so I'll probably have to wait for the papers...	14:31
	← EmilMedve left	14:32
madduck	first round data:	14:32
	lapse:~\|master\|% n=0; grep -hic git ~/mail/phd/cur/:2,F \| while read i; do n=$(($n + $i)); done; echo $n	14:32
	93	14:32
	→ SjB joined	14:32
madduck	gitte: the thesis will be first. if there are going to be papers, then later. i doubt i will write real papers though, i hate academia too much.	14:33
	but the website will have the most important information in a concise format	14:33
gitte	The common sequence: thesis > papers. However, there are niceties in academia.	14:33
madduck	really? :)	14:33
gitte	For example, you do not get to bore yourself to death, as in private industry.	14:33
idletask	gitte: I'll have a look tonight, am @work right now - thanks for the link?	14:34
	s,\?$,!,	14:34
gitte	For example, if you feel like it, you typically can come in for lunch instead of just after breakfast.	14:34
madduck	true; anyway, my only condition to my supervisor when I started (yeah, i know, it's usually the other way around) is that it'd be a monograph and I would not have to write a single paper.	14:34
gitte	madduck: as long as you get your stuff done... which is quite challenging if you're involved in Git.	14:34
madduck	gitte: i work in private industry too	14:34
	... from home. :)	14:34
idletask	(and sorry for sounding like an ass)	14:34
gitte	idletask: an ass goes eee-haaah.	14:35
Arafangion	gitte: Seems that checkpointing each commit is very expensive.	14:35
	← gasbag left	14:35
Arafangion	gitte: :)	14:35
gitte	Arafangion: yep, its only use is for debugging.	14:35
Arafangion	gitte: Only? Don't even need to do it at the end?	14:36
gitte	Not that I know of. Should be mentioned somewhere in the manpage IIRC.	14:36
	→ jm joined	14:37
	→ Yuuhi joined	14:37
Arafangion	I should re-read it when I'm clearer headed, should get some sleep.	14:37
	Thanks heaps gitte :)	14:37
gitte	Arafangion: what's your tz?	14:38
	← deskin left	14:39
Arafangion	gitte: GMT +10	14:39
	gitte: I think it's daylight savings atm, though.	14:39
gitte	Oh, that explains it.	14:39
	Ah, I remember seeing your name with an .au email	14:39
Arafangion	:)	14:39
	→ name joined	14:40
Arafangion	Those were my first patches to an OSS project, btw. :)	14:40
	→ Garoth joined	14:41
	→ simmerz joined	14:43
madduck	congrats!	14:43
	and not just any, but the best of them! :)	14:43
	→ scientes__ joined	14:43
simmerz	is it possible to remove a commit that has been made previously?	14:43
	← thebolt left	14:44
Arafangion	madduck: Thanks. :) Is it odd that I'm committing patches while I'm doing a rewrite?	14:44
madduck	simmerz: did you push it anywhere?	14:44
	Arafangion: nothing is odd in OSS	14:44
simmerz	madduck: yes. so I think what I want to do is merge back to before that change was done	14:45
	I'm used to svn, so learning git	14:45
madduck	simmerz: no, then just use git-revert	14:45
moccuo	is it possible to revert back to an older revision, but only on certain file(s)?	14:45
madduck	moccuo: not really	14:45
	git checkout rev -- file	14:45
	but then you have to commit the changes	14:45
simmerz	madduck: a bit confused how i use git-revert	14:46
	what is a <commit-ish> ?	14:46
Arafangion	simmerz: A mark, tag, branch, sha1...	14:46
madduck	a commit id or branch name	14:47
simmerz	the one I want to revert to?	14:47
madduck	not revert to	14:47
	simmerz: okay, let's take a second here...	14:47
	assume you have commits A,B,C,D,E in that order, and you wanted to revert C	14:47
	do you want A,B as result, or A,B,D,E?	14:48
	← scientes_ left	14:48
simmerz	I would "like" ABDE	14:48
madduck	then you do	14:48
	git-revert C	14:48
	git revert C even	14:48
	this will create ABCDEF	14:48
	where F is !C	14:48
	F reverts C	14:48
	it's the proper way of doing it	14:48
simmerz	is a merge but no -m option was given, apparently	14:48
madduck	aha.	14:49
	well, read about the -m option and then figure out what it wants. :)	14:49
	probably -m0 but you need to see	14:49
	← Arafangion left	14:49
madduck	anyway, if you screw up, you can always rescue it	14:49
	do	14:49
	git tag safety-net	14:49
	now	14:49
	so that you can always restore to the current state later	14:49
	(git checkout master; git reset --hard safety-net)	14:50
	do not use git reset unless you know what it's doing otherwise	14:50
simmerz	ok	14:50
madduck	especially not with --hard	14:50
	and make that s/;/\&\&/ in the above :)	14:50
	← Sho_ left	14:51
madduck	-m1, not -m0	14:52
	sorry	14:52
	who designed this? :)	14:52
simmerz	ok. I've now got a conflict in a swf (binary) file. what to do? it suggests git add <path> and git rm <path> but there is only one file there	14:53
	← ankit9 left	14:54
	← majoh left	14:55
	→ majoh joined	14:55
	→ Daenyth joined	14:59
simmerz	next question. how do i take a patch that I've created and merge it in? I got giggle to create the patch for me	14:59
Daenyth	I accidently uploaded a branch with the wrong name to my github repo, how can I remove it?	14:59
simmerz	ah. git-apply	15:00
Daenyth	is there a simple way to do that?	15:00
madduck	Daenyth: git push origin :branchname	15:01
Daenyth	simple :)	15:02
	thanks man	15:02
	→ EiNZTEiN joined	15:02
	← Daenyth left	15:02
	→ EiNZTEiN_ joined	15:02
	→ cilly joined	15:02
madduck	i'm not a man	15:04
	→ kmap joined	15:04
	← bjeanes left	15:04
	foca_ → foca	15:04
imarcusthis	a duck?	15:05
	→ loincloth joined	15:05
madduck	very good!	15:05
	← hobodave_ left	15:05
imarcusthis	your ircname says martin. strange name for a duck.	15:06
madduck	prejudice. why can't ducks have nice names too?	15:06
imarcusthis	I mostly know them as "dinner"	15:06
	madduck gasps	15:06
	→ hobodave_ joined	15:07
	→ aspotashev joined	15:07
	→ d0k joined	15:07
Arjen	Heheh	15:08
	→ rmh3093 joined	15:10
simmerz	final question I can't work out: how do push a branch to remote? I've done a git commit -a and a push then whilst in master, but it hasn't pushed the branch I created	15:11
broonie	simmerz: git push <remote> <branch>	15:12
	→ willb joined	15:13
broonie	By default it will only push branches that already exist, you need to explicitly tell it to push a new one.	15:13
	→ krh joined	15:13
simmerz	ok	15:15
	→ withanx_ joined	15:16
	← Niamor left	15:17
	→ NiaGhost joined	15:17
	NiaGhost → Niamor	15:17
samjam	What does DAG mean?	15:17
Arjen	Directed Acyclic Graph	15:18
	→ clairvy joined	15:19
	→ jackdempsey joined	15:19
	→ spearce joined	15:19
samjam	thanks.	15:19
	Anyone know the spell to make stgit just shut up and start again?	15:20
	I daren't do "stg commit" in case it does something	15:20
	← kanru left	15:20
samjam	for some strange reason; "git fetch" did a sort of merge and broke stgit	15:20
	← moccuo left	15:23
	→ ivaxer joined	15:23
	→ grithe joined	15:24
	← vasandgvd left	15:24
	→ kanru joined	15:25
	← spearce left	15:25
	→ careo joined	15:26
	← hobodave_ left	15:26
madduck	topgit. :)	15:27
	→ harinath1 joined	15:29
samjam	maybe I should	15:29
withanx_	you have to do adds at the end in git? If you do a git add file, then modify the file, it won't commit the modifications until you re-add it, right?	15:31
	← lorandi left	15:31
madduck	right	15:31
	→ lorandi joined	15:31
madduck	one of the best features of git	15:31
	alongside git add -p :)	15:31
	← harinath left	15:32
withanx_	git add -p does what?	15:32
	→ koke_ joined	15:32
	→ akahn joined	15:32
madduck	check it out	15:33
	make some changes all over and then call it.	15:33
	← FunkeeMonk left	15:33
akahn	How do I merge when what I'm pulling in will overwrite a binary file I have? I'm okay with that, so I want the merge to go ahead and do it.	15:33
madduck	use a merge driver like 'theirs'?	15:34
akahn	what is a merge driver?	15:34
madduck	man gitattributes and git-merge	15:35
	the one you want is probably 'binary'	15:35
akahn	ok. i read git help merge and it didn't seem to mention a way to use 'theirs'	15:35
madduck	theirs doesn't exist	15:35
	sorry	15:35
	'binary'	15:35
	→ iulian joined	15:36
madduck	strategy btw	15:36
	git merge -s binary ...	15:36
akahn	madduck: i don't see that strategy in git help merge	15:36
madduck	man gitattributes	15:36
	wait	15:37
	i think i am confusing things	15:37
	← cilly left	15:37
	← koke left	15:37
	→ SgtPepperKSU joined	15:37
madduck	what is the relation between merge drivers and strategies?	15:37
simmerz	ok. confused now again. sorry! If I do git-branch -r -d origin/somebranch, why does it not disappear off the github branch list?	15:39
	EmilMedve1 → EmilMedve	15:39
madduck	because it's a separate repo	15:40
akahn	so i see where the man page is talking about the binary driver. how do I set git to use that driver for certain files/directories?	15:40
madduck	man gitattributes	15:40
	i need to stop this now	15:40
	i never get any work done otherwise	15:40
	what an addictive thing, this Git.	15:41
	gitte: you should have warned me! :)	15:41
	← EmilMedve left	15:41
	→ Modius joined	15:41
	→ kukks joined	15:43
	← joshdavey left	15:45
	← jpieper left	15:46
	→ jpieper joined	15:46
	→ cannonball joined	15:49
	← softdrink left	15:50
	→ pygi joined	15:52
gitte	madduck: why should you have any advantage over me? I had to unsubscribe from the mailing list, back then.	15:52
madduck	gitte: you are older than me? it is your duty to protect me from the evil!	15:52
	i get to learn from your mistakes.	15:53
	→ softdrink joined	15:53
madduck	or rather: you get to make them and i get to avoid making them. :)	15:53
	← sergiodj left	15:55
	→ EmilMedve joined	15:55
	→ foca_ joined	16:00
	← dotsintacks left	16:01
	→ threeve joined	16:01
	← parasti left	16:02
	→ bobesponja joined	16:04
	← LuboC left	16:05
	→ evansparkman joined	16:05
evansparkman	hey guys	16:05
	could someone help me on this error? http://pastie.org/313083	16:05
	→ bobmcw joined	16:05
madduck	ask in #ruby?	16:06
evansparkman	no	16:06
	this happens when i do a cap deploy	16:06
madduck	i know git, but not what a cap deploy is.	16:07
	→ LuboC joined	16:07
broonie	Automated deployment of an app.	16:07
madduck	in ruby?	16:08
broonie	It's a ruby thing, yes.	16:09
	well, rails anyway.	16:09
madduck	see above. :)	16:09
	← foca left	16:09
	foca_ → foca	16:09
	← rmh3093 left	16:10
	→ stephenos joined	16:11
	→ CodeOfficer joined	16:12
	← ia left	16:13
	→ ia joined	16:14
	→ aroben joined	16:17
bremner	is there a #capistrano or #rails?	16:18
danderson	try it?	16:18
	← jackdempsey left	16:18
	← pygi left	16:19
	bremner is not personally interested, just as a suggestion for places to ask capistrano and rails questions	16:19
	← simmerz left	16:19
	→ racerx joined	16:19
bremner	#rubyonrails ftw	16:20
	→ muthu_ joined	16:20
	→ Voker57 joined	16:20
racerx	can someone help me with y a command to set up a repository is not working?	16:21
	→ blithe joined	16:21
racerx	here is the command: sudo -u git git-daemon --base-path=/home/git/repositories/ --export-all	16:22
	i'm running it on ubuntu distro	16:22
	→ sergiodj joined	16:22
	→ hobodave joined	16:22
blithe	Can anyone point me to a method to pull a directory (and it's history) out of a current repository and add it into a new seperate repository?	16:23
	← evansparkman left	16:23
blithe	I started a project in another repository that now needs it's own repo.	16:23
madduck	git-filter-branch maybe?	16:24
tokkee	blithe: git filter-branch --subdirectory-filter	16:24
	← roolio left	16:25
madduck	blithe: but be careful of the consequences	16:25
blithe	madduck: Oh?	16:25
madduck	this rewrites history and requires you to make sure that everyone who is following the repo reclones basically	16:25
blithe	I'm the only one working the repo at the moment.	16:25
Mikachu	you want to clone it first, and then filter-branch	16:25
madduck	otherwise you will diverge with other team members.	16:25
Mikachu	don't filter in the original repo	16:26
madduck	Mikachu: not even if you tag the current commit before?	16:26
racerx	can someone show or point me to a tutorial on how to set up a git protocol on my repository?	16:26
madduck	racerx: http://git.or.cz	16:26
Mikachu	madduck: well, maybe, but it's still a pretty good idea to make a new repo	16:26
madduck	you just don't want bug reports from angry users. :)	16:27
	i say this as with my mdadm maintainer hat on. i would love if people did their mdadm commands on clones instead of live systems. :)	16:27
racerx	I have a repo setup. I need to make it so that you say git clone git://... the emphasis being on git protocol	16:27
blithe	Thanks guys, it worked great. Thanks Mikachu for the advice on cloning first.	16:27
madduck	racerx: i am sure the wiki or homepage explain this.	16:27
	man git-daemon	16:28
racerx	I tried to follow a tutorial using this commanline on ubuntu and it did not work. sudo -u git git-daemon --base-path=/home/git/repositories/ --export-all	16:28
madduck	on ubuntu, just installing and configuring in /etc should work	16:29
	← koke_ left	16:30
madduck	this is how to use it with xinetd:	16:30
	http://1t2.us/152904	16:30
	in /etc/xinetd.d/local-git-daemon	16:30
	arguably, that should be a different user though	16:31
	→ fynn joined	16:32
fynn	yay git update	16:32
	so git would be even bettah	16:32
	→ rmh3093 joined	16:32
	← ivaxer left	16:33
	→ koke joined	16:35
	← ciaran left	16:36
	→ ciaran joined	16:36
	← koke left	16:37
	→ tango_ joined	16:37
	→ koke joined	16:38
	← reithi left	16:39
	→ bieneff joined	16:42
	→ qrush joined	16:42
	→ spearce joined	16:42
	sdboyer → sdboyer\|soup	16:42
	→ digitalhobbit joined	16:44
	← namenlos left	16:44
	→ j_engelh joined	16:46
	→ pygi joined	16:49
	→ rubydiamond joined	16:49
	→ jmatthews_ joined	16:52
	← idletask left	16:53
	→ ahupp\|fb joined	16:53
	→ radarek joined	16:54
	→ jackdempsey joined	16:55
	→ DrFrasierCrane joined	16:55
	→ kergoth joined	16:56
kergoth	if i have a list of random commits, taken from the history of a single ref, is there an easy way to order them by accessibility? the order they occurred in the history?	16:57
	← muthu_ left	16:57
j_engelh	--date-order	16:57
	or...	16:58
	you could do this:	16:58
vmiklos	--date-order is the default iirc	16:58
j_engelh	no, topo-order is default	16:58
	git rev-list TheSingleRef \| grep -Ff textfile-of-your-random-commits	16:59
kergoth	that only works with log or rev-list, no? i have a number of commits here, they didn't occur one after another, or in this order, and i don't want to see any commits other than these.	16:59
vmiklos	ah, yes	16:59
kergoth	ah, thats a good idea	16:59
	→ koke_ joined	17:00
	→ paltman joined	17:03
	→ theefer joined	17:03
paltman	what's the difference between HEAD and master?	17:03
	→ mediogre joined	17:04
kergoth	paltman: HEAD is what you currently have checked out. that isnt necessarily master	17:04
	← rmh3093 left	17:06
	→ cilly joined	17:06
	j_engelh → Hatler	17:07
	Hatler → j_engelh	17:07
	→ rmh3093 joined	17:07
	← koke left	17:07
	← DrNick left	17:08
	→ DrNick joined	17:08
	← rmh3093 left	17:09
	← nud left	17:09
	→ rmh3093 joined	17:09
	→ tewk_ joined	17:11
	← SgtPepperKSU left	17:14
	→ ceej joined	17:15
	← odi left	17:16
	→ patmaddox joined	17:17
	← jackdempsey left	17:17
	← akahn left	17:17
	← bschindler left	17:19
	→ ceej_ joined	17:20
	← tewk left	17:23
	← gitte left	17:23
racerx	I'm getting an error when trying to clone a git repository. the error is "errno=No route to host". Is it something on my end or the server because I can do git clone username@//ip_adress:rep. But no git clone git://ip../repo.	17:24
	→ idletask joined	17:24
	harinath1 → harinath	17:25
	→ Modius_ joined	17:26
	← ceej left	17:27
	→ joshbuddy joined	17:29
	→ Pistos joined	17:30
Pistos	The docs are not really clear: in git config files, is a tab character required, or can spaces be used to indent?	17:30
	← binjured left	17:31
	→ netoman joined	17:32
	→ binjured joined	17:32
	← Modius left	17:32
	→ emss joined	17:38
	→ Modius joined	17:40
	→ p4tux joined	17:42
	← hrajchert left	17:42
	→ deskin joined	17:43
	← racerx left	17:45
madduck	space is fine	17:45
	← aFlag left	17:46
	→ ceej joined	17:46
madduck	i don't think you even need any space at the start of the line	17:46
	← PerlJam left	17:47
	→ PerlJam joined	17:47
	← cilly left	17:48
	→ digitalhobbit_ joined	17:49
	→ Modius__ joined	17:49
	→ cilly joined	17:50
	← Numbers left	17:51
	← ceej_ left	17:52
	← Modius_ left	17:52
	→ Modius_ joined	17:53
	← Kbyte left	17:54
Pistos	madduck: Nice. Thanks.	17:54
	→ niki joined	17:55
	→ kraymer joined	17:55
	← ahupp\|fb left	17:56
	→ adamb90 joined	17:56
	← Modius_ left	17:57
	→ Modius_ joined	17:58
	→ redinger joined	18:00
	→ nud joined	18:00
	← zewerson left	18:00
	← lorandi left	18:01
	← ged left	18:01
	→ lorandi joined	18:01
	→ Nies joined	18:02
	← digitalhobbit left	18:02
	→ zewerson joined	18:02
	← coderdad_ left	18:02
	→ jackdempsey joined	18:02
	← markelikalderon left	18:03
	← Pistos left	18:03
	← harinath left	18:03
	← Modius_ left	18:05
	← clairvy left	18:06
	← Modius left	18:06
	→ gilimanjaro joined	18:06
	← Modius__ left	18:10
	← quickdev left	18:13
	→ ashleyw joined	18:14
	→ iiijjjiii joined	18:16
	← ben_h left	18:17
	← cilly left	18:18
	→ litage joined	18:19
	← DrFrasierCrane left	18:21
	← jm left	18:22
	→ drizzd joined	18:23
	← spearce left	18:25
	→ farrukh_najm1 joined	18:26
	← farrukh_najmi left	18:27
	← deskin left	18:28
	← Voker57 left	18:29
	← mediogre left	18:30
	→ Voker57 joined	18:31
	→ spearce joined	18:34
	← charon left	18:39
pygi	hi hi spearce :)	18:40
	← qrush left	18:40
	← ashleyw left	18:42
	← drizzd_ left	18:44
	aroben → aroben\|meeting	18:44
	← idletask left	18:46
	← iiijjjiii left	18:48
	→ nixbox joined	18:50
nixbox	hi	18:50
	→ dreiss joined	18:50
	← Tv left	18:50
nixbox	i am doing a git pull , it does something and then terminates saying that a file xyz.c needs update	18:50
	→ \|\|seb\|\| joined	18:50
nixbox	what do i need to do to fix this error?	18:50
\|\|seb\|\|	Is there any reason git is better for kernel development while mercurial is better for userland apps? (I'd prever to use git for everything)	18:51
	prefer*	18:51
marcan	I use git for everything...	18:51
	even slideshows :P	18:51
bmalee	Well, git is better for the Linux kernel 'cause it's what Linux uses...	18:52
\|\|seb\|\|	marcan: hmm...then why is mercurial being used by a lot of projects...i wonder if the perception is that "git is for kernel dev" in open sourceworld	18:52
PerlJam	\|\|seb\|\|: Not around here.	18:52
bmalee	Beyond that, I don't know any reason why one would be better than others in some situations but not others.	18:52
PerlJam	\|\|seb\|\|: mercurial is used because people like it.	18:52
	→ qrush joined	18:53
\|\|seb\|\|	PerlJam: i'll likely have to learn both to cover all my bases	18:53
	← koke_ left	18:53
marcan	I haven't dealt with hg, but I'm sure we'll all agree that git beats the pants off of cvs and svn. Too bad sourceforge doesn't offer git hosting though; some cling to that fact to avoid being assimilated by git	18:54
	→ jm joined	18:54
j_engelh	sf will eventually get there, just scream louder	18:54
\|\|seb\|\|	marcan: i'm amazed how fast linus blew the pants off BitKeeper and moved on	18:54
j_engelh	also, I am kinda thankful I don't host git at sf... sf would probably prohibit push -f	18:54
marcan	yeah	18:54
	→ schacon joined	18:55
PerlJam	marcan: why would sf offer git hosting? github already exists :-)	18:55
	→ david_koontz joined	18:55
	← spearce left	18:55
marcan	PerlJam: legacy users afraid of change and stuck with SF :-)	18:56
j_engelh	you'd be surprised...	18:56
	like one or two years ago, I ran a sample on SF... and it turned out 95% of projects ran on CVS rather than svn...	18:56
nice_burger	i want to find a function in a file. the function was deleted several commits ago - how do i find out which commit still has this function? some sort of grep file_name from some git output, but what exactly?	18:56
\|\|seb\|\|	how soon before we see hghub? :)	18:57
schacon	is there an easy way I'm not thinking of where you can list all the commits over several branches only if they have a certain sha in their history?	18:57
j_engelh	\|\|seb\|\|: probably never.	18:57
PerlJam	nice_burger: git help grep	18:57
\|\|seb\|\|	j_engelh: all it takes a one PC with a new connection...doesn't seem that hard	18:57
marcan	the guys I get to deal with every now and then still use SF CVS	18:57
\|\|seb\|\|	s/new/net	18:57
marcan	they don't like me because I improve their stuff and throw it on a public git repo	18:57
\|\|seb\|\|	marcan: bah..CVS	18:57
schacon	seb: bitbucket is basically github for hg	18:57
marcan	of course the git repo kind of supersedes their CVS repo	18:58
	you can just checkout origin and get their CVS master, updated hourly	18:58
	→ spearce joined	18:58
PerlJam	I was at a meeting last year where the presenter was telling everyone about this process he'd built up around CVS. About mid-way through the presentation, he finally asked how many people used CVS. Not a single person raised their hand.	18:58
marcan	hahahah	18:58
nice_burger	PerlJam: thanks, that's a bingo. i see the line i'm interested in, but how do i know what commit that is?	18:59
PerlJam	nice_burger: um ... good question.	19:00
	nice_burger: maybe use "git bisect" in conjunction with "git grep" /	19:01
	?	19:01
	→ amystrat joined	19:01
nice_burger	PerlJam: how bout this as a solution- you know a way to tell git grep to give me N lines and N lines after the matching line? then i could just see the function and copy it, and nevermind what commit it was	19:03
amystrat	Having strange problem with a few users, git pull once..... have conflict in random file no merge.... do git pull again, it merges fine	19:03
	→ ahupp\|fb joined	19:03
amystrat	when the users had no local changes	19:03
PerlJam	nice_burger: sure, you use the same options you would for unix grep -A, -B, or -C (lines after, lines before, or both)	19:04
nice_burger	thanks, just found that, was gonna tell you	19:05
	(i'm new to grep in ANY context hehe)	19:05
	→ ashleyw joined	19:13
	fynn is adapting to Git by finding cute nicknames for huge SHA hashes	19:17
fynn	for example, f909356 is like "f, gogo, and the number of days in the year"!	19:17
	but gog instead of gogo :\|	19:18
	but now you guys all know what I'm committing :/	19:18
bremner	fynn: tags are nicknames for hashes, the cute part is up to you :-)	19:18
fynn	bremner: yeah, but it's not like it's a good idea to create a huge bunch of them...?	19:19
bremner	fynn: well, you can delete them	19:19
fynn	hm, probably a good idea	19:20
	I'm still going to refer to that one as fgog. he's sort of a personal friend.	19:20
	though admittedly of a somewhat imaginary nature.	19:21
	← \|\|seb\|\| left	19:21
	sdboyer\|soup → sdboyer	19:21
bremner	on IRC, no one knows your a tag	19:22
	→ ciaran_ joined	19:22
	→ sdboyer-laptop joined	19:23
	mw → mw\|food	19:23
	← rubydiamond left	19:23
	→ coderdad joined	19:24
vuf_	fynn: does your leap year have 356 days?	19:25
j_engelh	No.	19:25
	even my normal years don't have 356 ;)	19:25
fynn	vuf_: I always confuse them, so it wfm	19:26
	→ scrooloose_ joined	19:26
gilimanjaro	all my years have 356	19:27
	most have more even	19:27
bremner	gilimanjaro: so far anyway	19:27
gilimanjaro	true	19:28
j_engelh	like 9-10 extra days	19:28
gilimanjaro	there aprox 97% probability that I will experience one year of less then 356 days	19:28
j_engelh	there is a 0% chance you'll ever get a year with less than 364.	19:29
	← Ademan left	19:29
gilimanjaro	nope	19:29
	in fact	19:29
bremner	j_engelh: whoosh	19:29
gilimanjaro	i've already experienced one year with much less	19:29
vuf_	and if you go to Venus, you will get many more	19:29
j_engelh	well yeah if you travel all day you can have a year with a little more than 130 days	19:29
gilimanjaro	aprox 160 days that year	19:29
	no	19:30
	my first year had less	19:30
	and my last will probably have less also	19:30
j_engelh	-.-	19:30
bremner	j_engelh: is that the displeased pug dog smiley?	19:30
	→ Cacher joined	19:31
	stick → stick[T15]	19:31
	stick[T15] → stick	19:32
	→ cilly joined	19:32
	← coderdad left	19:32
	← ciaran left	19:33
	ciaran_ → ciaran	19:33
	→ odi joined	19:36
	→ jesselucas joined	19:36
	→ kumbayo joined	19:36
	→ ivazquez1 joined	19:38
	→ coderdad joined	19:38
	→ Daenyth joined	19:38
Daenyth	I'm trying to use git diff to see all changes done to a portion of my tree since a specified date, but I can't figure out the syntax	19:39
	how would I do that	19:39
	something like git diff --from="Oct 10"	19:39
	for example	19:39
gilimanjaro	git diff {2008-10-10}..HEAD	19:40
	might work	19:40
	man git-rev-parse	19:40
	dwmw2 → dwmw2_gone	19:40
Daenyth	thanks	19:41
	← sgrimm left	19:41
Daenyth	hrm	19:42
gilimanjaro	isn't working eh	19:42
	i'm trying what will work	19:42
Daenyth	says bad revision	19:42
	← ivazquez left	19:42
gilimanjaro	ah	19:42
	--since	19:42
Daenyth	ahh	19:42
gilimanjaro	erh	19:43
	jese	19:43
	no sorry	19:43
Daenyth	it's not in the manpage 0.o	19:43
	ah	19:43
gilimanjaro	that's an argument to git-rev-parse	19:43
Daenyth	is that for git log?	19:43
	ivazquez1 → ivazquez	19:44
vuf_	just find a commit from 10-10 and diff to that	19:44
gilimanjaro	ah	19:45
	I almost had it	19:45
	git diff @{2008-10-10}	19:45
Daenyth	ahh	19:45
gilimanjaro	i forgot the @	19:45
	shoudl work with git log too	19:45
Daenyth	win	19:45
	many thanks, will save me a lot of work	19:46
gilimanjaro	no	19:46
	problem	19:46
	:)	19:46
	→ daniel-w_ joined	19:46
Daenyth	:D	19:46
	← Daenyth left	19:46
	→ charon joined	19:47
	← ceej left	19:47
	→ thedoctor joined	19:48
	← scrooloose left	19:48
thedoctor	Hi, can anyone explain git's mmaping behavior? I can't find anything on google.	19:48
vuf_	what do you mean by behaviour?	19:49
thedoctor	what does git mmap in particular and what operations is it particularly trying to speed up	19:49
	things relative to the index?	19:49
vuf_	I don't know	19:50
	← amystrat left	19:51
thedoctor	hrm... well, what got me started on this was the initial email thread when Linus was first starting git	19:51
	and he mentioned something about keeping his whole kernel tree in memory and git leveraged that, so I got looking.	19:51
	the particular email is here: http://lkml.org/lkml/2005/4/8/205	19:53
	and that got me interested both in what exactly Linus is doing and what git does	19:54
	← cilly left	19:54
	→ rtomayko joined	19:55
	← daniel-w left	19:57
	← komplex left	19:58
	→ k776 joined	19:58
	→ ceej joined	19:59
broonie	h	20:02
	← tvw left	20:03
broonie	thedoctor: git is optimised to work well when the filesystem cache can cache the entire .git	20:03
	thedoctor: so all memor	20:03
	← fynn left	20:03
broonie	thedoctor: so reading data becomes extremely cheap once the cache is warm.	20:03
	→ _graham_ joined	20:03
	→ fynn joined	20:04
fynn	Can I ref (tag) a specific file at a specific revision?	20:04
	i.e. "the file foo.py at its state after changeset 23"	20:04
thedoctor	broonie: ok, so git tries to mmap the entire repo?	20:04
charon	fynn: you can tag the blob of the file _contents_ at that commit. that tag won't "know" the commit it took the file from, however.	20:05
broonie	or just read it, the important thing is that it doesn't have to hit disk	20:05
	← lorandi left	20:06
thedoctor	broonie: what do you mean by read it?	20:06
broonie	using normal I/O functions rather than mmap	20:06
thedoctor	so are you saying git will read the repo and cache information in a separate way in RAM?	20:07
broonie	No.	20:07
thedoctor	I'm sorry I don't see what you're saying...	20:07
broonie	git relies on the filesystem caching that the OS does.	20:07
fynn	charon: so basically, I can get a tag called "foo.py at revision 23" but it won't have any internal reference to revision 23...	20:07
thedoctor	oh, ok	20:07
	I guess I'm understanding Linus's email more now...	20:08
	do you know what he meant by keeping kernel trees in memory?	20:08
charon	fynn: yes. git doesn't version files separately, only trees of files. so you can tag blobs (contents), trees, and commits.	20:08
thedoctor	does he mean tmpfs or disk buffers?	20:08
vuf_	thedoctor: buffers	20:08
charon	er, s/yes/exactly/. i suck at english semantics.	20:08
broonie	buffers.	20:09
thedoctor	so is there some way to tweak the buffering behavior?	20:10
	← jas4711 left	20:10
thedoctor	or is he just assuming that once he accesses the kernel directory the kernel will do the right thing?	20:10
fynn	charon: what does it mean that I can tag a tree?	20:10
vuf_	for Linus, that is not an assumption	20:10
	← ashleyw left	20:11
broonie	thedoctor: Details of the OS disk cache will depend on the OS.	20:11
DrNick	non-Linux operating systems do tend to have terrible caches, though	20:11
thedoctor	I guess what I'm getting at is how exactly can you be sure the OS will cache a given directory?	20:12
	assuming OS==linux	20:12
	← kergoth left	20:12
broonie	You can't; move it to tmpfs or something else memory backed if you're concerned.	20:12
charon	fynn: a tree is git's way of storing the filenames. it's just a list saying "file 'foo.py' had contents $sha1". i haven't seen anyone tag trees in practice though. you should stick to commits :)	20:13
broonie	but not doing that kills performance for lots of apps so there's a very good chance the OS will DTRT as much as it can.	20:13
thedoctor	broonie: yeah, that's what I'm trying to deduce.. does Linus just have a ton of RAM and knows that it will get cached right, or is he using tmpfs	20:13
	because if anything will speed up compilation and ease my disk's duties, I'm interested :P	20:14
broonie	you only need a few hundred megabytes even for the kernel tree.	20:14
thedoctor	broonie: an entire 2.6 git tree is about 1.2G, though	20:14
	is caching smart enough to keep from hammering the disk with linux? I'm assuming it is because I don't hear it grinding when I compile, but I've never gotten real numbers.	20:15
	← pygi left	20:15
	→ Mario_ joined	20:15
fynn	charon: right, because probably the tree is the entire tree.	20:15
broonie	yes	20:15
	← schacon left	20:15
broonie	note that .git plus the actual code you build is the only interesting bit here.	20:15
fynn	charon: so if I have a repo with more than one file, tagging the tree is always going to refer to more than one file.	20:15
thedoctor	true, and what I'm having a hard time understanding is if .git is over a gigabyte, how is caching affected?	20:16
charon	yes. why do you want to tag a (commit,file) pair anyway? doesn't the entire context matter? and if it doesn't, why aren't the contents of the file enough?	20:16
	← kraymer left	20:16
	aroben\|meeting → aroben\|lunch	20:16
thedoctor	assuming we might not have enough space to cache all of .git	20:16
	which, I guess, brings me back to the question of "what does git mmap?"... I saw things saying it make 2 250MB mmap files, what for?	20:17
	scrooloose_ → scrooloose	20:18
	→ ashleyw joined	20:18
charon	it mmaps whenever it can... for example, to compute sha1s of a file, it just mmaps the entire file instead of reading it	20:18
thedoctor	ok.	20:18
	I think I've got the picture now	20:18
	thanks!	20:18
	← Mario_ left	20:19
vuf_	thedoctor: do realize that it is mostly stat information that is cached, git only looks at files if they actually were modified	20:20
thedoctor	vuf_ : yeah, that was something I noticed mentioned in the thread.	20:21
charon	you don't usually need to have all of .git cached anyway, for example, writing a new commit should just need .git/index, the loose objects and the pack indexes (i haven't looked, but it doesn't have any use for the rest of the data)	20:21
	← nice_burger left	20:22
thedoctor	that would make sense as mostly you'll be needing to look at the index relative to what you're doing.	20:22
	→ damog joined	20:22
fynn	charon: isn't tagging the commit focused on the changeset, rather than the resulting state?	20:23
vuf_	fynn: not really, you tag for example v1.0	20:24
charon	fynn: changesets are an illusion given by git-diff (and supported by other tools such as 'am' and 'rebase'). git always tracks the entire contents at that commit	20:24
	mw\|food → mw	20:24
	← thedoctor left	20:25
	→ parasti joined	20:25
	← bentob0x left	20:25
	→ Russel-Athletic joined	20:27
Russel-Athletic	hiho	20:27
	← adamb90 left	20:27
	→ adamb90 joined	20:27
fynn	charon: so it's like the Matrix?!	20:28
Russel-Athletic	perhaps this is a stupid question but i have now git on my laptop and on my pc. both repositories differ now in such a way, that a merge doesn't work. how can i just say something like: forget what i did on my laptop, just use the stuff on my pc	20:28
fynn	charon: j/k, so that's what is meant by the notion that "git tracks snapshots"?	20:28
charon	exactly	20:28
	→ gour joined	20:28
	→ jas4711 joined	20:28
gour	vmiklos: hi, here is the public url of that repo - darcs get http://code.haskell.org/xmobar/	20:29
	← hiffy left	20:29
	← Voker57 left	20:29
gilimanjaro	Russel; you can discard your laptop repo and clone it from your pc	20:30
Russel-Athletic	well that would be a possibility but isn't there a nicer way?	20:30
gilimanjaro	define nicer :)	20:31
jptix	hmm, if want to basically do `git log -p \| grep foo`, but also see what commit the match comes from..any good ways to do that?	20:31
Russel-Athletic	well something without loading the whole repository	20:31
	→ ciskje joined	20:31
vuf_	Russel-Athletic: look up "merge ours"	20:31
jptix	besides `grep "^commit\|foo"	20:31
gilimanjaro	if you know when the two started to diverge, you can first reset your laptop to that commit	20:31
	and then pull in the changes from your pc	20:31
Russel-Athletic	but it would also help if somebody could name me a tool or a way to see the differences in the file better then just the <<<<< and >>>> signs in a file	20:32
jast	jptix, you mean git log --grep?	20:32
gilimanjaro	russel: gitk may be of help?	20:32
vuf_	Russel-Athletic: git mergetool perhaps? (I never learned to like it, though)	20:32
jast	(or -S and the --pickaxe-* options)	20:33
jptix	jast: sounds promising	20:33
Russel-Athletic	thanks vuf_	20:33
gilimanjaro	sometimes rebasing on batch of commits on top op the other clears things up a bit also	20:33
	s/on/one/	20:34
Russel-Athletic	and git merge --s ours just gave me a message: "you are in the middle of a conflict merge"	20:34
	→ sgrimm joined	20:36
gilimanjaro	russel; well, you may want to back out of the merge before deciding on a course of action	20:36
	but git-mergetool was very good advice from vuf i think	20:37
	never used it myself, but it sounds like what you need	20:37
Russel-Athletic	yes it is a start and i am trying to merge it at the moment with this	20:37
	unfortunatly the tool confuses me now	20:38
	but thanks for the help :)	20:38
gilimanjaro	read it's manpage; you can use it with any diff viewer you want	20:38
Russel-Athletic	i know	20:39
	but this is just the time to learn this new cool fancy emacs merge thingy :)	20:39
jptix	jast: thanks, -S is what i'm looking for.	20:40
	sneaky thing doesn't accept a space before the search string…	20:40
	took me a while to see that	20:40
	← lucsky left	20:43
	→ pjhyett joined	20:43
	← adamb90 left	20:44
	← ashleyw left	20:45
	→ lfaraone joined	20:46
lfaraone	Hey, I apologize for my cluelessness, but how do I "git update" from our master tree when I have local changes that havn't been merged?	20:47
	→ kansan joined	20:48
shd	lfaraone: do you mean local changes that have not been committed?	20:48
	→ ashleyw joined	20:48
lfaraone	shd: yeah.	20:49
shd	lfaraone: git stash -> you get a clean working directory. then "git pull" to merge the master.. handle the merge. then "git stash apply"	20:49
	or, maybe this:	20:49
	throwaway branch for temp changes: git checkout -b newtmpbranch ; git commit ; git checkout master ; git pull frommaster ; then get the stuff from newtmpbranch	20:50
	→ Tv joined	20:50
shd	lfaraone: "git stash apply" will restore your dirty files with 3-way patch	20:50
	→ ciskje_ joined	20:50
shd	lfaraone: stash is a kind of temporary working area that can later be "merged" on top of any branch	20:51
PerlJam	I think of a stash as a kind of nameless branch.	20:52
	← ciaran left	20:53
shd	PerlJam: the difference to a branch is, it does not have new commits	20:54
	→ ciaran joined	20:54
shd	i often use stash, but sometimes i use named temporary branches to do the same thing..	20:54
vuf_	I think of stash as confusing	20:55
gilimanjaro	i often use my stash, but get the munchies after	20:55
lfaraone	shd: ah...	20:55
fynn	charon: OK, so a "changeset" is really a snapshot + the diff that created that snapshot...?	20:55
charon	the diff is generated on the fly	20:55
	a commit just knows the snapshot and its parent(s)	20:56
fynn	OK, what are the "parents"?	20:56
gilimanjaro	previous commits	20:56
lfaraone	shd: crap, there's a conflict...	20:56
gilimanjaro	previous snapshops i should say	20:56
lfaraone	shd: (I tried the "stash")	20:57
fynn	gilimanjaro: so basically, the repo has a linked list of all previous snapshots?	20:57
shd	lfaraone: it's not a silver bullet.. you have to handle those things anyway	20:57
	lfaraone: what git stash saves you is a series of cp and mv commands ;)	20:57
gilimanjaro	fynn: sort of, but it's not a proper linked list, as a commit can have multiple parent commits if it's a merge commit	20:57
fynn	gilimanjaro: ah, so it's more like a tree.	20:58
gilimanjaro	fynn: and multiple children if it's being branched from	20:58
fynn	right.	20:58
	CS speaking, it could still be a linked list with several parents :)	20:58
gilimanjaro	fynn: a bit, but also not really, because two branches can merge back together	20:58
fynn	(but not with several children)	20:58
	oh snap	20:58
parasti	everyone calls it a DAG -- directed acyclic graph	20:58
fynn	now I'm going to find another analogy and you'll just say "a bit but also not really because it can shoot unicorns"	20:59
gilimanjaro	fynn: running gitk on a proper project with branches visualizes it best usually	20:59
	fynn: it can, if you want it too	20:59
fynn	parasti, gilimanjaro: thanks, I'll look up those newfangled DAGs	20:59
	in my day, everything could be either a list or a tree and we were pretty happy about it.	21:00
	← netoman left	21:00
gilimanjaro	fynn: the storage system behind git is quite interesting, though hard to wrap your head around if you're used to other systems	21:00
	→ felipec joined	21:00
gilimanjaro	fynn: yeah, i know. took me a whole lot of reading before i got into the spirit of this. it's sorta cool though	21:01
fynn	gilimanjaro: yeah, I was really surprised. git sort of has the reputation of something being quickly hacked together for the primary objective of organizing the kernel source.	21:01
parasti	http://eagain.net/articles/git-for-computer-scientists/ (for whoever hasn't seen this yet)	21:01
gilimanjaro	fynn: well, that's how the kernel itself started right :)	21:01
fynn	yeah, now that I dived into it, it now seems like possibly the best engineered DVCS around.	21:02
vmiklos	gour: ok, i'll try to reproduce :)	21:02
gilimanjaro	fynn: whatever linux hacks together is usually a lot better than what other people spend years on	21:02
	s/linux/linus/	21:02
	vmiklos: i've been trying for years	21:02
fynn	gilimanjaro: was about to say that possibly Linus just has a knack for doing top-notch design work really fast.	21:03
	:)	21:03
vmiklos	gilimanjaro: are you a spambot?	21:03
	← Russel-Athletic left	21:03
gour	vmiklos: thanks	21:04
gilimanjaro	vmiklos: no way. but me and the misses have been trying to reproduce for a while... :)	21:04
	← elmex left	21:05
	← ciskje left	21:05
lfaraone	gilimanjaro: what's the square root of 8?	21:05
	← Ryback_ left	21:06
	→ markelikalderon joined	21:06
gilimanjaro	lfaraone: 2.82842712474619 according to Calculator.app	21:06
	lfaraone: why? :)	21:06
lfaraone	gilimanjaro: just wondering what responce I'd get.	21:07
cardioid	I'm a lot better with the cube root	21:07
lfaraone	cardioid: bah!	21:07
gilimanjaro	lfaraone: ok... well, if you need another square root, just let me know	21:08
	← iulian left	21:08
gilimanjaro	i'm gonna pass an inverse turing test	21:08
	convincing humans i'm a bot	21:08
	← ciaran left	21:09
	→ ciaran joined	21:10
nud	apps whose name end with .app are sort of evil	21:10
gilimanjaro	not if you're on OS/X	21:11
	in which case the rest of the world is evil	21:11
nud	os whose name end with X are sort of evil	21:11
gilimanjaro	hehe	21:11
	or beginning with X	21:11
lfaraone	gilimanjaro: binaries should have no extention.	21:11
gilimanjaro	true	21:11
nud	so evil I forgot an 's' in both sentences	21:12
gilimanjaro	but who ever said an application always is a binary?	21:12
lfaraone	gilimanjaro: although I _do_ like OSX's portability...	21:12
	gilimanjaro: binaries == applications for our purposes. :P	21:12
	gilimanjaro: where does a python script that you want to run go? /usr/bin.	21:12
gilimanjaro	In OS/X an application is a 'bundle' (directory actually) which contains binaries along with other resources.	21:12
lfaraone	(if you want the world to be able to run it)	21:12
	← ashleyw left	21:13
gilimanjaro	yes, but those are (mainly) command line binaries	21:13
	that's just not how OS/X graphical applications work	21:13
nud	OS/X sounds so much like OS/2	21:14
gilimanjaro	hmm... the name yes, but the mechanics, not so much	21:15
	i worked on OS/2 a lot when i worked at IBM	21:15
	← lresende left	21:15
gilimanjaro	OS/2 was more windowzy	21:16
	but that may have been LANMAN mostly	21:16
	← name left	21:18
	→ name joined	21:20
	← name left	21:20
	→ name joined	21:21
	gour enjoyed os2 for many years	21:24
	← ciaran left	21:27
	← rtomayko left	21:27
	← jesselucas left	21:27
	← scrooloose left	21:27
	← ahupp\|fb left	21:27
	← p4tux left	21:27
	← patmaddox left	21:27
	← damog left	21:27
	← ia left	21:27
	← LuboC left	21:27
	← foca left	21:27
	← careo left	21:27
	← schlort left	21:27
	← LiamH left	21:27
	← grahal left	21:27
	← priidu left	21:27
	← mika left	21:27
	← Ratler left	21:27
	← morphir left	21:27
	← hurikhan\|Work left	21:27
	← dwave left	21:27
	← Weiss left	21:27
	← tcoppi left	21:27
	← tchan left	21:27
	← Tekerson left	21:27
	← qiyong left	21:27
	← hidek_ left	21:27
	← chrizz left	21:27
	← fynn left	21:27
	← jdavis__ left	21:27
	← mugwump left	21:27
	← blithe left	21:27
	← dirker left	21:27
	← iabervon left	21:27
	← jptix left	21:27
	← mutex left	21:27
	← jones- left	21:27
	← malahal__ left	21:27
	← UukGoblin left	21:27
	← marcan left	21:27
	← dermoth left	21:27
	← jarod left	21:27
	← thommey left	21:27
	← Ned_ left	21:27
	← cehteh left	21:27
	← esden`away left	21:27
	← Whtiger left	21:27
	← IRSeekBot left	21:27
	← jql left	21:27
	← BeerSerc left	21:27
	← halfline left	21:27
	← ddollar left	21:27
	← pasky left	21:27
	← felipec left	21:27
	← k776 left	21:27
	← daniel-w_ left	21:27
	← coderdad left	21:27
	← odi left	21:27
	← david_koontz left	21:27
	← kansan left	21:27
	← digitalhobbit_ left	21:27
	← PerlJam left	21:27
	← bieneff left	21:27
	← juanjoc left	21:27
	← nick_fn left	21:27
	← zumbrunn left	21:27
	← petteri left	21:27
	← wagle left	21:27
	← lll left	21:27
	← IronHeart left	21:27
	← a-atwood left	21:27
	← Leonidas left	21:27
	← da5id left	21:27
	← incidence left	21:27
	← doiuf left	21:27
	← com4 left	21:27
	← worch left	21:27
	← pilpi_ left	21:27
	← elliottcable left	21:27
	← Tv left	21:27
	← cannonball left	21:27
	→ blithe joined	21:28
	→ olly joined	21:28
	← hobodave left	21:29
	→ Tekerson joined	21:29
	← gilimanjaro left	21:29
	→ wagle joined	21:31
	→ ddollar joined	21:31
	→ dermoth joined	21:32
	→ patmaddox joined	21:33
	← SjB left	21:34
	→ jesselucas joined	21:34
	→ adsworth joined	21:34
	← dlowe left	21:36
	aroben\|lunch → aroben	21:36
	→ davidfetter joined	21:38
	← dmlloyd left	21:40
	← Toad left	21:40
	→ Whtiger joined	21:40
	→ thommey joined	21:42
	→ jdavis__ joined	21:43
	→ priidu joined	21:44
	→ schlort joined	21:46
	→ damog joined	21:46
	→ rtomayko joined	21:46
	→ scrooloose joined	21:46
	→ ahupp\|fb joined	21:46
	→ p4tux joined	21:46
	→ ia joined	21:46
	→ LuboC joined	21:46
	→ foca joined	21:46
	→ careo joined	21:46
	→ LiamH joined	21:46
	→ grahal joined	21:46
	→ mika joined	21:46
	→ Ratler joined	21:46
	→ morphir joined	21:46
	→ hurikhan\|Work joined	21:46
	→ dwave joined	21:46
	→ Weiss joined	21:46
	→ tcoppi joined	21:46
	→ qiyong joined	21:46
	→ Ned_ joined	21:46
	→ hidek_ joined	21:46
	→ chrizz joined	21:46
	→ dirker joined	21:46
	→ iabervon joined	21:46
	→ jptix joined	21:46
	→ jones- joined	21:46
	→ malahal__ joined	21:46
	→ jarod joined	21:46
	→ BeerSerc joined	21:46
	→ UukGoblin joined	21:46
	→ esden`away joined	21:46
	→ halfline joined	21:46
	→ cehteh joined	21:46
	→ pasky joined	21:46
	→ Toad joined	21:46
	→ Tv joined	21:46
	→ cannonball joined	21:46
	→ dmlloyd joined	21:46
	→ mutex joined	21:46
	→ ashleyw joined	21:47
	→ mugwump joined	21:47
	← olly left	21:47
	→ marcan joined	21:47
	→ tchan joined	21:47
	→ felipec joined	21:47
	→ kansan joined	21:47
	→ k776 joined	21:47
	→ daniel-w_ joined	21:47
	→ coderdad joined	21:47
	→ odi joined	21:47
	→ david_koontz joined	21:47
	→ digitalhobbit_ joined	21:47
	→ PerlJam joined	21:47
	→ bieneff joined	21:47
	→ juanjoc joined	21:47
	→ nick_fn joined	21:47
	→ zumbrunn joined	21:47
	→ petteri joined	21:47
	→ lll joined	21:47
	→ IronHeart joined	21:47
	→ pilpi_ joined	21:47
	→ a-atwood joined	21:47
	→ Leonidas joined	21:47
	→ incidence joined	21:47
	→ da5id joined	21:47
	→ elliottcable joined	21:47
	→ worch joined	21:47
	→ doiuf joined	21:47
	→ com4 joined	21:47
	→ Weasel[DK] joined	21:48
	← bieneff left	21:48
	← Weaselweb left	21:49
	→ bieneff joined	21:49
	→ adamb90 joined	21:51
	← WOP left	21:52
	← ahupp\|fb left	21:52
	→ olly_ joined	21:53
	← d0k left	21:55
	← Pupeno-M__ left	21:56
	← farrukh_najm1 left	21:56
	→ farrukh_najmi joined	21:57
	→ mosno joined	21:58
	→ camwest joined	21:59
	← joshbuddy left	22:01
	→ joshbuddy joined	22:01
	sdboyer → sdboyer\|class	22:01
	→ clairvy joined	22:01
	← pjhyett left	22:02
	← Cacher left	22:02
	← olly_ left	22:03
	→ pygi joined	22:07
	← LiamH left	22:07
	→ entie joined	22:08
	→ padski joined	22:11
	→ mosno_ joined	22:12
	← gour left	22:12
	→ joshbuddy_ joined	22:13
	← joshbuddy left	22:13
HopsNBarley	i ran git repack on a repo, but it looks like an http fetch of the repo isn't using the pack.	22:15
	ideas?	22:15
	← mosno left	22:15
	mosno_ → mosno	22:15
	→ voodootikigod joined	22:17
	← joshbuddy_ left	22:17
	→ joshbuddy joined	22:18
charon	did you git update-server-info?	22:19
HopsNBarley	yeah.	22:20
shd	HopsNBarley: you may want to setup a hook to automatically git update-server-info on commit/push	22:20
	← voodootikigod left	22:20
	← camwest left	22:20
	→ pjhyett joined	22:20
HopsNBarley	ah well, it works, it's just slow.	22:20
	← sdboyer-laptop left	22:20
	→ flaguy joined	22:21
	← lfaraone left	22:21
	→ chris2_ joined	22:22
	← boto left	22:22
paltman	have a coworker trying to make a fresh clone on his laptop of our private repo, and it getting index-pack fatal errors (http://gist.github.com/24286) any ideas? he is running 1.6.0.4	22:22
	→ joydivider joined	22:22
	← joshbuddy left	22:23
	→ joshbuddy joined	22:23
	← juanjoc left	22:24
Bombe	Hmm, damn. "git filter-branch HEAD" (with a .git/info/grafts file) outputs the parameter list of rev-list, including the error "could not get the commits". What did I do wrong?	22:24
cbreak	do you want to filter the branch from HEAD to HEAD?	22:25
	there are no commits in between	22:25
Bombe	No, I want to filter all commits.	22:26
	chris2_ → chris2	22:26
cbreak	hmm...	22:26
Bombe	"git filter-branch" gives the same result.	22:26
paltman	no one knows anything about this the git bug with fatal index packing?	22:27
	← joshbuddy left	22:28
	→ olly_ joined	22:28
	→ joshbuddy joined	22:29
adsworth	hi I'm just starting with and from I've read it's all about branches, so should one branch even if it's my repository and I'm the only one working on it. or is it ok to work directly on the master branch?	22:30
	starting with git that is	22:31
	← qrush left	22:32
	← bobmcw left	22:32
	→ therrg joined	22:33
	→ eventualbuddha joined	22:34
	← Nies left	22:34
parasti	adsworth: sure you can work directly in master. it's just another branch	22:34
	partha → partha-bbl	22:34
padski	how can I list the commits that touch a given filename	22:34
mugwump	padski: git-whatchanged -- filename	22:35
	or git-log, even	22:35
	throw a --all in there before the -- to see all branches	22:36
padski	mugwump, thanks. I couldn't figure out how to do it with git-log	22:36
adsworth	parasti: thanks	22:36
	→ ijcd joined	22:37
ijcd	Last fetched revision of refs/remotes/20081112-release was r30367, but we are about to fetch: r30333!	22:38
	anyone know how to fix that?	22:38
	using git svn	22:38
	I used "git fetch" on a repo cloned from another that had full svn history in it... seems things have gotten out of whack	22:39
	this is while trying to run "git svn rebase"	22:39
	→ agenteo joined	22:39
	→ sverrej joined	22:41
ijcd	my svn commands are taking a long time right now too... maybe the server is busy	22:42
	22:43:18 up 78 days, 3:53, 21 users, load average: 4.91, 3.25, 2.90	22:43
	yeah	22:43
	partha-bbl → partha	22:43
ijcd	oops	22:43
	← cmarcelo left	22:45
	← charon left	22:48
	← EiNZTEiN_ left	22:49
	← tcoppi left	22:50
	→ tcoppi joined	22:50
	→ EiNZTEiN_ joined	22:50
	cbrake → cbrake_away	22:51
	← aknowles left	22:53
	← odi left	22:53
	← jesselucas left	22:53
	← ph^ left	22:54
vuf_	any experts here? :)	22:55
pygi	just ask	22:55
adsworth	bye	22:56
	← adsworth left	22:56
	→ carllerche joined	22:57
vuf_	that is going to be a waste if there aren't experts around	22:58
	← Yuuhi left	22:58
	← EiNZTEiN left	22:59
	→ entie_ joined	23:00
	← krh left	23:00
	← entie left	23:00
	← joshbuddy left	23:00
Mikachu	vuf_: we prefer if you waste your time rather than our	23:00
vuf_	okay, we have changed an out-of-repo symlink into a directory with the files in-repo, and now a pull on another branch fails with a conflict	23:01
paltman	are there any hooks that get called when a tag is the fetch?	23:02
vuf_	if I make the pull be a fast-forward, there is no problem, but merging finds conflicts in these files that are new on one branch	23:02
	→ joydivider_ joined	23:03
	← Weasel[DK] left	23:03
	→ EiNZTEiN joined	23:05
	← jelmer left	23:07
vuf_	argh, minimal test case did not work	23:07
	→ johnw joined	23:08
	← EiNZTEiN_ left	23:08
	← pygi left	23:09
	← ebneter left	23:09
vuf_	is this supposed to happen during pull? "fatal: git write-tree failed to write a tree"	23:11
werdna	vuf_: I presume not :P	23:11
_graham_	Are the permissions on the .git directory correct?	23:12
	← padski left	23:12
vuf_	yup	23:13
	← name left	23:15
	← paltman left	23:15
kandjar	wow	23:15
	I have git rebase failing	23:15
	its running out of memory....	23:15
	(already did a git gc)	23:15
	← adamb90 left	23:17
	← loincloth left	23:17
	← _graham_ left	23:17
	→ loincloth joined	23:18
	→ markstos joined	23:18
	← eventualbuddha left	23:18
	← david_koontz left	23:19
	← threeve left	23:19
	← joydivider left	23:20
markstos	I've got this workflow where I keep submitting patches to a project, and they always refine them heavily and publish the result. When I pull that, I get these big conflicts to manually resolve. What I would like to do is to throw away the patches I sent at that point and just accept the upstream versions. Is there an easy way to do that?	23:20
	→ pygi joined	23:21
vuf_	maybe revert before you pull?	23:22
jast	in that case i would actually suggest rebasing away the patch	23:22
	or, if it's the newest patch, using reset HEAD^	23:22
	aroben → aroben\|meeting	23:22
parasti	markstos: put your patches into a separate branch, then rebase that against the branch you pull into, and use rebase --skip..?	23:23
	→ jql joined	23:23
markstos	Thanks. It's no wonder I was confused between whether to use revert, rebase and reset. :)	23:23
jast	well, effectively, you've got an old version of the commit	23:23
	and you gave it to them (and probably nobody else), and they committed something different based on it	23:24
	i think in that case it's very reasonable to say that you don't need the old commit anymore, so revert isn't the solution of choice	23:24
markstos	I'm collaborating via github, so even after I get my local tree in order, I have to deal with getting my published github tree in order, which will also have the conflicting patch already pushed there.	23:24
jast	well, in that case i'd suggest either reverting the old commit before merging, or just fixing the conflict	23:25
markstos	Ok. I also got confused looking at the conflict, because in a distributed system, it wasn't clear whether HEAD indicated local or remote, and on the other side of the conflict was the hash string, which I also couldn't readily identify as being local or remote.	23:27
	→ JavaGeek joined	23:27
	← ahhsm left	23:27
markstos	But I can figure that out.	23:27
JavaGeek	greetings	23:27
markstos	Thanks everyone for the help	23:27
JavaGeek	wow, so many people	23:27
jast	one helpful way to look at conflicts is using git diff --ours and --theirs	23:27
	or using git mergetool (configured to use a nice merge tool, e.g. kdiff3)	23:28
	← farrukh_najmi left	23:28
	→ farrukh_najmi joined	23:29
JavaGeek	I'm using git locally and svn remotely (client requirement). If I created a local branch in git, and now the client wants me to create the same branch on svn, is there a command to 'push' the local branch to the remote svn repo?	23:29
	← felipec left	23:30
	← coderdad left	23:30
	← kumbayo left	23:30
	← priidu left	23:31
parasti	JavaGeek: create a branch in svn, fetch it, branch it, then arrange your commits on top of it, and dcommit... or something along those lines	23:31
	→ coderdad joined	23:32
	← loincloth left	23:35
	→ hobodave joined	23:36
	← nud left	23:36
	→ ebneter joined	23:37
	← olly_ left	23:40
	← jackdempsey left	23:40
	→ olly_ joined	23:40
	← entie_ left	23:41
vuf_	okay, I get "Skipped <filename> (merged same as existing)" from merge debug info ... but the file is new in one branch!?	23:42
	← ciskje_ left	23:43
	← willb left	23:44
	→ quoin joined	23:45
	→ doener joined	23:46
	→ haris joined	23:49
	← coderdad left	23:49
	← jm left	23:52
	→ ahhsm joined	23:54
	← olly_ left	23:56
	← therrg left	23:56
	→ therrg joined	23:57
	← therrg left	23:57
	← joydivider_ left	23:58
	→ qrush joined	23:58
	← cannonball left	23:58
	← SRabbelier left	23:59

Logs Search ←Prev date Next date→

Channels Documentation