IRCloggy #git 2009-04-23

Provider of IRC logs since 2005.

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

	2009-04-23
gitte	doener: GIGO	00:00
	← bremner_ left	00:00
doener	:-)	00:00
	→ bremner_ joined	00:00
gitte	doener: well, thinking about it, Gigo is an acronym I understand. But Gitbot? What does _that_ mean?	00:01
	← HG` left	00:01
	← naeu left	00:02
	← gwoo left	00:02
	→ carldani joined	00:02
carldani	hi	00:02
doener	gitte: hm, in that context, maybe it's just wordssmashedtogether, "git" having its British meaning ;-)	00:02
carldani	is there any way to get the git log for a given repository on disk without having to enter the directory of that repository?	00:04
	→ renan_saddam joined	00:04
doener	carldani: git --git-dir=/path/to/.git log	00:04
jast	gitte: "Garbage In, Typical Bot Output: Trash"	00:04
renan_saddam	how do I merge a branch to another?	00:04
	git merge	00:04
	?	00:04
	→ jerbear joined	00:04
gitte	doener, jast: well, I came up with "Generaly idiotic test, by outputting trash"	00:04
Dashkal	be in the branch you want to merge _into_ then git merge <branch to merge from>	00:05
poseidon	I installed git in Ubuntu, however I can't find git in /usr/bin	00:05
mugwump	you installed the git-core package?	00:05
carldani	doener: thanks!	00:05
renan_saddam	Dashkal: git checkout pt_br; git merge master	00:06
poseidon	mugwump, nope just git. Installing get-core now	00:06
renan_saddam	strange it returns "Already up-to-date"	00:06
	← FreakGuard left	00:06
doener	poseidon: the "git" package used to be "gnu interactive tools". It's called "gnuit" now.	00:06
jast	I'll just go get sleep now to stop myself from coming up with even more ridiculous backronym expansions.	00:06
poseidon	lol, thanks	00:06
doener	poseidon: git, the dvcs, is in git-core, as mugwump said	00:06
Dashkal	renan_saddam: are you sure there's something to merge? The only time I get that message is when the two branches are pointing at the same commit	00:07
renan_saddam	Dashkal: I was on the wrong branch	00:08
Dashkal	that would do it	00:08
	sdiz-afk → sdiz	00:08
renan_saddam	Dashkal: I do a git branch -a and returns "origin/pt_br". I should do a git checkout pt_br, right?	00:09
Dashkal	Are you trying to merge the changes from pt_br into master or the changes from master into bt_br?	00:09
renan_saddam	Dashkal: master to pt_br	00:09
Dashkal	make sure your pt_br is up to date with origin	00:10
	← kalle_ left	00:10
	→ qrush_ joined	00:11
renan_saddam	Dashkal: that's what I'm trying to do... get the last changes from master and merge to pt_br	00:11
	→ [1]intripoon joined	00:11
	← jrmuizel left	00:11
Dashkal	"Already up to date" means the branch you're merging from points to the same commit as the branch your merging into. Nothing to do. Given that that's unexpected, I suspect that either your master or pt_br branch is out of date with origin	00:12
doener	Dashkal: not necessarily the same commit	00:12
Dashkal	doener: My mistake, yes, already merged would do it too	00:12
poseidon	When trying to connect to the sourceforge git repos I get the following "error: src refspec master does not match any."	00:13
doener	poseidon: did you push anything to that repo yet?	00:14
poseidon	no it's empty	00:14
	I'm using the guide at http://apps.sourceforge.net/trac/sourceforge/wiki/Git	00:14
	That was the first time I tried to push on the repo.	00:15
	→ jrmuizel joined	00:15
	← carldani left	00:15
doener	poseidon: ah, sorry, the source refspec was "bad"	00:15
	poseidon: probably means that your local repo is also empty (or has no branch head called "master")	00:15
	→ kalle_ joined	00:16
	← Bass10 left	00:16
	→ mithro joined	00:16
poseidon	O, my my repo is empty. I guess I should fill it with a couple files	00:16
	Whats the difference between add and commit?	00:16
	← bobesponja left	00:17
	→ bobesponja joined	00:17
poseidon	nm	00:17
zpoley	i just did a git svn clone and i have master pointing at one of my subversion branches - is there a way to point master back at trunk ?	00:17
	→ bambule joined	00:17
doener	zpoley: did you do anything else yet? Or just cloned?	00:17
zpoley	just cloned	00:18
	← Student left	00:18
	← geir_ left	00:18
doener	zpoley: then (assuming that you didn't use the --prefix=whatever option) "git reset --hard trunk" should do	00:18
zpoley	doener: nice thank you	00:19
	← ia left	00:19
	← marvil07 left	00:19
	→ wshimmy1 joined	00:20
columbo	?	00:20
	→ bdimcheff joined	00:20
	← qrush left	00:22
	→ marvil07 joined	00:23
	→ gwoo joined	00:24
	→ Leemp joined	00:24
	← gwoo left	00:24
	← bambule4711 left	00:25
	→ javatexan joined	00:28
	← intripoon left	00:29
	[1]intripoon → intripoon	00:29
	→ nis joined	00:32
nis	gitte: hi! What's up?	00:32
	← Modius left	00:33
gitte	nis: time ;-)	00:33
nis	gitte: still stressed?	00:33
gitte	Yep.	00:33
	Still have something to do, but I am dead tired.	00:33
nis	Oh, see you later then.	00:34
gitte	CU!	00:34
	← gitte left	00:34
nis	"CU"?	00:34
	← netoman left	00:34
	← wshimmy left	00:35
	← qrush_ left	00:35
	← travisjeffery left	00:37
nis	could anyone tell me what CU means?	00:37
	← the10nitro left	00:37
	← tatsuyao_ left	00:37
jjuran	"See you."	00:37
	→ Voight-Kampf joined	00:37
DrNick	nis: in this context, "see you" as an abbreviation for "see you later"	00:38
	← orbii left	00:38
nis	Ah, thanks. (IRC and me...) :-)	00:38
	→ coney joined	00:39
	← Voight-Kampf left	00:39
	amahi → cpg	00:39
coney	so what do i do when i screw up and want to go clean slate	00:39
	← Predominant left	00:40
	← loincloth left	00:40
nis	coney: it depends if you already commited.	00:42
coney	no i never comit :D	00:43
	← dodo_the_last left	00:43
coney	i'm a loser	00:43
nis	coney: seriously?	00:43
coney	XD but i'm tryin to change	00:43
	i'm way off the branch	00:44
	apple on the ground	00:44
nis	coney: what 'git status' says?	00:45
coney	nobranch	00:45
	but its got changes it wants me to save but i dont wanna	00:45
	XD	00:46
	← uberpinguin left	00:46
nis	coney: first, you should commit all your work now	00:46
coney	doesnt that give it upstream?	00:47
nis	coney: no!	00:47
	→ davidad joined	00:47
	→ ngng joined	00:47
	→ scarabx joined	00:47
nis	coney: git commit is a _local_ operation.	00:47
	coney: you should commit often on a dedicated branch	00:49
coney	ok	00:49
nis	coney: done?	00:49
coney	whats that do	00:49
nis	coney: it saves your work in the repo.	00:49
	coney: if you screw up more you'll be able to come back at this state.	00:50
	coney: what 'git branch' says?	00:51
coney	still no branch	00:53
	checkout -b?	00:53
nis	try that	00:53
	coney: the main problem is to find your last "clean slate"	00:54
	coney: do you know why you were not on a branch?	00:55
coney	git reset --hard HEAD looks promising	00:55
	→ egoist joined	00:56
coney	if there was a way to screw up git bet your !#$ i will	00:56
	nothing to commit (working directory clean)	00:56
nis	coney: ok, what is the branch you were working on (i mean before you left it)	00:57
	s/is/was/	00:58
	← foutrelis left	00:58
coney	its in git branch -a	00:58
	→ MarcWeber joined	00:58
coney	git branch -f origin/branch?	00:59
nis	is it in git branch -r too?	00:59
	← bdimcheff left	01:00
coney	of course	01:00
nis	but not in 'git branch'	01:00
	← Dashkal left	01:00
	→ hgimenez joined	01:01
	← solydzajs left	01:01
coney	er wait checkout	01:01
	the docs keep flipin cmds on me	01:01
nis	coney: branches listed in 'git branch -r' are remote branches. you have to create new branches from remote branch to work on it.	01:02
	← Tv left	01:02
	← qhoxie left	01:03
	← marvil07 left	01:05
	→ deskin joined	01:05
nis	coney: e.g. 'git checkout -b mywork origin/interesting'; <work, work, work>; git commit; <work, work, work>; git commit; etc	01:05
	→ Voight-Kampf joined	01:06
	← bremner_ left	01:07
	→ travisjeffery joined	01:09
coney	whoah thats a mouthfull	01:10
	← abra left	01:10
coney	thats it?	01:10
	← kalle_ left	01:10
coney	how do i update	01:10
	from remote commits	01:11
nis	remote commits are stored in the remote branches.	01:11
	← porcnick left	01:11
nis	(well, we can see it like that)	01:12
	→ loincloth joined	01:13
coney	merge?	01:13
nis	?	01:13
	→ tag joined	01:14
	→ sitaram joined	01:14
tag	is there a way I can export just my changesets from the base I cloned?	01:14
	→ arohner joined	01:15
nis	tag: man git-format-patch ?	01:15
Gitbot	tag: the git-format-patch manpage can be found at http://git.or.cz/man/git-format-patch	01:15
	← loincloth left	01:15
	→ kalle_ joined	01:16
nis	tag: something like: 'git format-patch initial_branch..mine'	01:16
	→ solj joined	01:16
	← egoist left	01:16
	→ bremner_ joined	01:16
	→ bremner__ joined	01:17
	→ ia joined	01:18
	← arohner left	01:19
	tedoc2000 → tedo_afk	01:20
	tedo_afk → tedoc2000_afk	01:20
tag	then how do I import those patches? (or do I just apply, commit, apply commit...)	01:21
nis	tag: git am	01:21
	man git-am	01:22
Gitbot	nis: the git-am manpage can be found at http://git.or.cz/man/git-am	01:22
nis	tag: you may also know man git-apply	01:23
Gitbot	tag: the git-apply manpage can be found at http://git.or.cz/man/git-apply	01:23
	→ camwest joined	01:23
	→ bremner___ joined	01:25
	← botanicus left	01:26
solj	hi, i'm getting a segfault when running git svn	01:27
	http://gist.github.com/100214 <== backtrace	01:27
	← javatexan left	01:27
	→ bremner____ joined	01:28
	→ porcnick joined	01:28
	← sgrimm left	01:29
nis	solj: is the repo public?	01:29
solj	nis: yes	01:30
nis	solj: could you give the address and the command you type?	01:30
solj	git svn fetch	01:31
deskin	solj: you'll get farther with a perl debugger; also, the crash looks like it's in the svn bindings (apache portable runtime which they use)	01:31
nis	ah, you already have a clone.	01:31
solj	reposRoot = https://svn.mcs.anl.gov/repos/bcfg	01:31
nis	solj: what your git --version?	01:32
solj	deskin: yeah, i'm pretty sure it's somewhere in the subversion code, but i'm not familiar with debugging perl	01:32
	this is 1.6.0.6 in gentoo	01:32
	→ kaichen joined	01:33
coney	w00t i'm ahead by one commit	01:33
	this is kinda fun	01:33
nis	coney: good. :-)	01:33
deskin	I'm not very either :) what I did was hack the source to dump perl backtraces when crashing, which worked ok; don't have access to the specific method right now	01:33
nis	solj: does gentoo provide a newer version?	01:34
deskin	...actually I think I can find it, if you're interested	01:34
solj	nis: that's the latest unmasked version	01:34
	i can try unmasking a more recent version	01:34
	← WALoeIII left	01:34
nis	solj: or compile from an official repo.	01:34
deskin	I'd try upgrading the svn bindings first, if you can	01:34
	→ Tv joined	01:35
solj	alright, i'll try that	01:35
	thanks	01:35
	← kukks left	01:35
	→ bremner_____ joined	01:36
	← renan_saddam left	01:36
	→ joevandyk joined	01:37
joevandyk	git svn dcommit hangs on me. what should i do??	01:37
	says "commiting to <url> ..."	01:37
	← bremner_ left	01:37
joevandyk	and then nothing	01:37
	← btanaka left	01:38
nis	joevandyk: is the remote repo up?	01:38
joevandyk	yeah	01:38
	i can commit via regular svn	01:38
nis	joevandyk: what's you git --version?	01:38
joevandyk	nis: git version 1.6.2.rc1.13.g25487f	01:38
	← bremner__ left	01:39
nis	joevandyk: did you tried on a newer version?	01:39
joevandyk	first time i've used git-svn and first time i'm trying to dcommit	01:39
	→ dgs__ joined	01:39
joevandyk	nis: thought this one was pretty new	01:39
	but i i will try	01:40
nis	joevandyk: yeah but it's a "rc" release candidate.	01:40
	← dgs_ left	01:40
joevandyk	nis: compiling now	01:40
nis	joevandyk: feel free to feedback. :-)	01:41
	→ javatexan joined	01:41
joevandyk	go go compile	01:41
nis	i'm going nowhere. ,-p	01:42
	← markelikalderon left	01:42
joevandyk	ok, trying to dcommit again	01:42
	same behavior so far. would it normally take a long time the first time to dcommit?	01:43
	→ km2 joined	01:43
solj	nis: fwiw, upgrading to 1.6.2.3 in portage and remerging subeversion with USE=-dso (to disable the runtime module search) fixed it	01:44
	thanks for the help	01:44
nis	solj: np	01:44
joevandyk	yeah, git svn dcommit still hangs	01:44
	there a way to see debugging output? -v does nothing	01:44
solj	joevandyk: strace?	01:45
joevandyk	aha, lots of 4504s	01:45
	404s	01:45
	← bremner____ left	01:46
	→ eno__ joined	01:46
joevandyk	why would i get a 404?	01:46
	→ celling joined	01:47
joevandyk	strace: http://gist.github.com/raw/100227/082932a35d8d3e4ee3ab9bbc162591c2035c3ab6/gistfile1.txt	01:47
nis	joevandyk: are you sure your URL is good?	01:47
joevandyk	nis: i cloned from it	01:47
	same repository url as i'm using for regular svn	01:48
nis	weird	01:48
	→ Zathrus joined	01:48
deskin	joevandyk: you could try looking at the network traffic with wireshark, it dissects webdav pretty well	01:48
joevandyk	there any reason that it would take a long time?	01:49
	← bremner___ left	01:49
joevandyk	this is a fairly large commit	01:49
	← Voight-Kampf left	01:50
	← jpld left	01:50
	← damainman856 left	01:50
	→ Predominant joined	01:50
	→ Student joined	01:51
	← solj left	01:51
nis	joevandyk: not much more than with svn, i guess	01:51
	joevandyk: what's time consuming should mainly be the network traffic	01:52
Zathrus	from work I have a branch called "sita". I did a "git push home". It claims everything up to date. From home, however, git show home/sita doesn't show the latest commit that it shows me at work.	01:52
sitaram	sita? my name's diminutive? wow...!	01:52
	(people call me sita)	01:52
nis	sitaram: :-)	01:53
Zathrus	sitaram: heh. www.sita.aero	01:53
joevandyk	could authentication be a problem?	01:53
nis	Zathrus: is the remote repo a bare repo?	01:53
Zathrus	nis: no, it was what I branched sita off of.	01:53
sitaram	Zathrus: if the branch sita did not exist at home, the push would not have sent it	01:54
Zathrus	ah.	01:54
sitaram	without an explicit refspec, push only sends matching branches	01:54
nis	joevandyk: possibly, don't know. isn't authentication information given in the URL?	01:54
joevandyk	hm, it's not	01:54
	this is a http:// repo	01:54
sitaram	so you need (first time) "git push home sita:sita" or perhaps just "git push home sita"	01:54
	→ bremner_ joined	01:54
	← kaichen left	01:56
	← eno left	01:57
nis	joevandyk: did you ever succes with dcommit?	01:57
Zathrus	sitaram: yeah... I managed to break it. Sigh. I'd recently done a "git push home sita", thought that was wrong, and nuked the branch at home.	01:58
joevandyk	nis no	01:58
sitaram	Zathrus: break?	01:58
joevandyk	tried to update the svn repo url to http://user:pass@url, no luck	01:58
deskin	you don't want user and pass in the url	01:58
	← camwest left	01:58
sitaram	you can push again... unless you mean "yeah but that can only be tomorrow when I get to work", in which case I sympathise!	01:58
nis	joevandyk: does the strace still report 404 this way?	01:58
	← schacon left	01:59
joevandyk	nis: yes	01:59
deskin	git svn will use cached credentials from ~/.subversion/some-file-named-auth-or-something if you have them	01:59
	or I think you can specify --user and get a password prompt; but I'd hope it doesn't hang if you don't	01:59
Zathrus	sitaram: yeah, I already fixed it. I'm VPN'd in (and then ssh'd back, ironically)	02:00
deskin	joevandyk: I definitely recommend using a network sniffer, you'll at least know if it's hanging or still working, even if you don't dig in to the packets	02:01
	→ faithfulgeek joined	02:01
Zathrus	sitaram: ok... so now from work git show home/sita shows the latest checkin. From home git show home/sita shows a stale one.	02:01
	← bremner_____ left	02:01
Zathrus	hrm... nvm	02:01
	nope, thought I fixed it, but it's still showing stale info	02:02
nis	Zathrus: did you checkout?	02:03
	→ xenoterracide joined	02:03
	→ schacon joined	02:04
nis	(at home)	02:04
	← bremner_ left	02:04
Zathrus	nis: I have the master tree at home, as well as a sita tree. I did a git pull home from the sita branch; it updated. git show home/sita from the master branch shows the stale data.	02:05
	→ ke_ joined	02:05
	eno__ → eno	02:06
	← poseidon left	02:06
sitaram	Zathrus: at home, there should not be a branch called home/sita	02:06
	Zathrus: should be just sita	02:07
Zathrus	there's both	02:07
sitaram	because you pushed it	02:07
Zathrus	let me pastebin...	02:07
sitaram	Zathrus: also git branch -a -v if you don't mind	02:07
Zathrus	sure	02:07
	← spearce left	02:09
sitaram	Zathrus: also, I have to leave and go to work; will stay logged in and reconnect from there but that will be about 30-35 min later :(	02:09
	30-45*	02:09
	→ loincloth joined	02:09
	→ hax0r1337 joined	02:10
Zathrus	http://pastebin.centos.org/25826	02:11
	← kalle_ left	02:11
joevandyk	hm, so i'm using wireshark and capturing traffic, but don't quite know what to do about what i'm seeing	02:11
Zathrus	sitaram: np	02:11
	→ qrush joined	02:11
	← loincloth left	02:11
	→ kalle_ joined	02:11
deskin	joevandyk: it depends how much webdav you know or want to learn. But the most important thing is that packets are continuing to be sent; if that's the case, git svn isn't hanging; infinite-looping, maybe, but not hanging	02:12
joevandyk	it's trying to find files on the remote server that don't exist yet (because I'm adding those files)	02:12
	← schacon left	02:12
nis	joevandyk: is one repo inside the other one?	02:13
joevandyk	what?	02:13
nis	Oops, nvr	02:13
deskin	then wait patiently for it	02:13
	← hax0r1337 left	02:13
Zathrus	nis: if that was to me -- no.	02:14
nis	Zathrus: yep, ok.	02:14
	→ javatexan1 joined	02:14
nis	Zathrus: which one is at home?	02:14
	← ke_ left	02:15
	← javatexan left	02:15
Zathrus	nis: those are actually both at home. But the "gittmp" directory is not part of the repo.	02:16
joevandyk	so it's usual behavior for it to check for the existance of every new file, one by one?	02:16
	← octagon left	02:16
	→ orbii joined	02:17
joevandyk	(there's a couple thousand new files)	02:17
nis	Zathrus: you're pushing from gittmp?	02:17
deskin	ask #svn, I dunno	02:17
	but until it's git-svn hanging, we can't help you much	02:18
	← orbii left	02:18
deskin	remember, someone else could've ninja-updated all those files between your last commit from svn and your dcommit	02:18
	→ orbii joined	02:18
Zathrus	nis: no, it was pushed from work. Yes, I know, I'm doing wonky things.	02:19
nis	Zathrus: so, gittmp is up-to-date?	02:20
joevandyk	woohoo,it's creating new stuff now!	02:20
	jesus	02:20
deskin	yay!	02:20
joevandyk	was not expecting it to take that long	02:20
Zathrus	nis: yes; I did a git pull home thinking that it would update the home/sita tracking branch	02:20
	nis: and I needed it updated anyway	02:21
nis	Zathrus: i'm lost with your repo	02:21
	Zathrus: why did you do a gittmp?	02:21
	→ [1]intripoon joined	02:21
Zathrus	heh	02:21
	→ harv_ joined	02:23
nis	Zathrus: is 0754b58 the good last wanted commit?	02:23
	Zathrus: or 8912209?	02:23
Zathrus	0754b58	02:24
nis	hm, it's in both repos.	02:24
Zathrus	I'm just going to nuke the gittmp dir and re-pull entirely. It was initially created as a testing ground anyway, and I don't actually need it.	02:24
	→ cytrinox joined	02:25
nis	Zathrus: at least, it will clear my mind! :-)	02:25
Zathrus	the important ones (master, sita, origin/*) are sane. Speaking of which, how can I rename origin, since I named it that stupidly while being overly pedantic following the github guides	02:26
joevandyk	man, git is so much faster than csv. i'm still trying to dcommit my first patch	02:26
Zathrus	you mean cvs?	02:26
nis	Zathrus: in the .gitconfig file	02:26
Zathrus	nis: that's easy	02:27
joevandyk	oops, subversion, not csv or cvs	02:27
Zathrus	that's one heck of a typo :)	02:27
joevandyk	Zathrus: long day	02:27
	but i'm glad i'm finally commiting my git work into the svn repo	02:27
	so thanks all for your help	02:27
Zathrus	it's not in the .gitconfig file	02:28
nis	Oops, typo: .git/config	02:28
	→ Voight-Kampf joined	02:29
Zathrus	just the [remotes "origin"] line or the fetch = line as well?	02:29
nis	Zathrus: wait,	02:29
	→ Ramonski_ joined	02:29
	← Ramonski_ left	02:30
nis	Zathrus: just do git remote rename old new	02:30
	← parasti left	02:30
Zathrus	yeah, not in my version... must've been added after 1.5.5	02:31
nis	Zathrus: ah.	02:31
	→ schacon joined	02:31
nis	Zathrus: can't you upgrade?	02:33
Zathrus	nis: not easily.	02:33
nis	well, save your repo and try that:	02:34
	Zathrus: rename origin refs in the .git/config file (both refs)	02:35
deskin	joevandyk: git-svn is a lifesaver. Glad it's working for you now	02:36
	← schacon left	02:36
nis	Zathrus: and cd .git/refs/remotes ; mkdir origin newname	02:36
	Zathrus: sorry	02:37
Zathrus	mv origin newname	02:37
	I figured I'd need to do something like that	02:37
nis	Zathrus: yeah, I would do cp -a insted	02:37
	instead*	02:37
	← cytrinox_ left	02:37
	← intripoon left	02:39
	[1]intripoon → intripoon	02:39
	→ damainman856 joined	02:39
	← dreiss left	02:40
	→ neunon joined	02:41
nis	Zathrus: then you should do 'git remote show' ; 'git remote show newname'	02:41
	→ wsc joined	02:42
	→ reprore joined	02:42
nis	Zathrus: i'm still on your pastebin, it seems that the push just worked	02:42
	Zathrus: the branch sita is up-to-date, right?	02:43
Zathrus	yes	02:43
	← yec left	02:43
nis	Zathrus: the push problem is resolved?	02:44
	→ yec joined	02:44
Zathrus	ok, I have no idea how I managed to do THIS. from /home/tom: git remote show home -- URL: /home/tom	02:46
	→ Tricon joined	02:47
nis	git remote show?	02:47
	→ hax0r1337 joined	02:47
	← ceej left	02:47
nis	is "home" in the list?	02:48
	(i guess)	02:48
	← airtonix left	02:48
nis	Zathrus: could you check your .git/config?	02:49
	→ baetis-fly joined	02:49
Zathrus	http://pastebin.centos.org/25829	02:50
baetis-fly	Question: I have a fork of a project on github. I branched, implemented a feature. In the mean time, my master got out of sync. Best way to get everything in sync? git checkout master - git pull upstream master - git checkout topic - git pull master?	02:50
nis	Zathrus: could you pastebin your .git/config?	02:50
	baetis-fly: not exactly.	02:51
	baetis-fly: what you probably want is	02:51
	→ tjafk1 joined	02:51
nis	baetis-fly: git chcekout master ; git pull upstream master ; git checkout topic ; git rebase master	02:52
Zathrus	nis: http://pastebin.centos.org/25832 -- fixing the origin idiocy right now	02:52
baetis-fly	nis: was thinking that, but I expect conflicts, still ok to rebase?	02:52
nis	baetis-fly: yes. you'll have to resolve the confilcts, btw.	02:53
baetis-fly	ok, but the rebase is still ok. got it.	02:53
	many thanks.	02:53
nis	baetis-fly: if you just want to know how far you diverged from upstream,	02:53
	baetis-fly: you could also do:	02:54
sitaram	Zathrus: I just finished looking at the backlog and I must say I'm a little confused too :) Are you putting your entire $HOME under git control by any chance?	02:54
nis	baetis-fly: git checkout topic ; git merge master ; (see if conflicts are hard) ; git reset --hard HEAD^	02:55
baetis-fly	nis: thanks. going to backup and try the rebase.	02:56
nis	baetis-fly: np; have fun	02:56
	→ ben_h joined	02:56
Zathrus	sitaram: I'm putting my dotfiles under git control	02:57
	e.g. -- .bashrc, .vimrc, etc	02:57
nis	Zathrus: are you using linux?	02:57
Zathrus	but .gitignore has /*	02:57
	nis: yes	02:57
sitaram	I quickly found that got too confusing and irritating; it seems to be better to make a proper repo and symlink everything that needs to be revision control to $HOME	02:57
	and make that symlink also automatic with a couple of lines in .bashrc	02:58
nis	Zathrus: the way i do that is by mount --bind into a dedicated repo.	02:58
	← Leemp left	02:58
Zathrus	nis: yeah, that's not an option. My work home dir is a NIS (NFS) and SMB mount	02:58
	I can go down the symlink path if needed, but since one of my target platforms _is_ Windows (for vim) I'd rather not since NTFS symlinks are so hosed	02:59
nis	Zathrus: i don't think it's a problem for bind, really.	02:59
	← niki left	03:00
Zathrus	nis: btw, renaming origin worked fine	03:00
nis	Zathrus: Ah, windows may be a problem. :-)	03:00
Zathrus	isn't it always? :)	03:00
nis	Zathrus: don't you use cygwin?	03:01
	→ niki joined	03:01
Zathrus	yes, but I use Win32 native vim/gvim. msysgit is synching the .vimrc/.vim dir just fine	03:01
sitaram	Zathrus: yeah my method is not so good with windows; I agree	03:01
Zathrus	the gittmp dirs were created initially to get down the process for pulling a git repo into the home dir -- git init; git add remote foo; git fetch foo; git checkout -b bar foo/bar	03:03
	in any case, it looks like the home/* branches are bogus and could be removed	03:04
	← KwikkSilva left	03:04
Zathrus	(from the master)	03:04
	← antoranz left	03:04
nis	can't say. :-/	03:05
	← javatexan1 left	03:06
	← damainman856 left	03:07
	← orbii left	03:07
Zathrus	well, I had the backup already, so I did a git remote rm home and it seems ok	03:07
	← tjafk2 left	03:07
Zathrus	nis, sitaram: thanks for the help	03:07
	← mithro left	03:07
nis	Zathrus: np; good luck	03:08
sitaram	good luck!	03:08
	→ girishr joined	03:09
	→ sgrimm joined	03:11
	← Student left	03:14
	→ kaichen joined	03:14
	→ amahi joined	03:20
baetis-fly	nis: that worked like a charm. is it safe to do a commit --amend at this point if I haven't pushed to public yet?	03:21
	← qrush left	03:21
nis	baetis-fly: yes	03:21
baetis-fly	nis: what a strange time for whitespace warnings...	03:21
	← kaichen left	03:21
	→ dreiss joined	03:22
	→ ceej joined	03:24
	→ schacon joined	03:24
	← tag left	03:25
	→ tag_ joined	03:25
nis	baetis-fly: you can configure whitespace issues	03:25
	baetis-fly: see man git-rebase ; man git-apply and man git-config	03:25
Gitbot	baetis-fly: the git-rebase manpage can be found at http://git.or.cz/man/git-rebase	03:25
	← schacon left	03:26
	→ spearce joined	03:26
	← hax0r1337 left	03:26
	← cpg left	03:26
baetis-fly	nis: you've been very helpful. many thanks.	03:27
	amahi → cpg	03:27
nis	baetis-fly: Oh, np!	03:27
	← ceej left	03:27
	→ Alien_Freak joined	03:30
	← kuadrosx left	03:30
Alien_Freak	can a git repo contain a .git directory... will it cause confusion?	03:30
max_ep	Alien: You want a submodule	03:31
Alien_Freak	here's my setup right now.....	03:31
	← jerbear left	03:31
Alien_Freak	I have a ~/workspace and different directories under each one... each .git was made using a git svn clone of a particular directory.... now rather then keeping in sync with workspace/data workspace/foobar etc.. I'd like to have a .git in ~/workspace then I can just clone from a remote box	03:32
	is submodule what I want? ...	03:32
	→ kuadrosx joined	03:33
Ilari	Alien_Freak: Every directory with .git is usually either ignored or submodule boundary point.	03:33
max_ep	Ok, a submodule will let you make a 'master' you can clone...	03:34
	→ Alinermedia joined	03:34
max_ep	However, they will still be separate repos	03:34
	the 'master' will only contain a reference to them	03:34
	→ damainman856 joined	03:35
max_ep	The other option is to lose your history and create a new repo	03:36
	← spearce left	03:36
	← faithfulgeek left	03:37
	→ snowman856 joined	03:38
	← orafu left	03:38
	→ orafu joined	03:38
	← lois[pac] left	03:39
Alien_Freak	ooh.. that's I think exactly what I want	03:39
	no.. I don't want to lose the history.. submodule sounds much better	03:39
	→ tag joined	03:42
tag	git svn clone can take forever on a large repository holy crap	03:42
	it'd be sweet if it didn't download everything.	03:42
	→ tstellar joined	03:42
tag	or atleast, if I could ask it to explicitly	03:43
deskin	tag: then it wouldn't be git	03:43
Ilari	tag: Or if actually used pipelining... Pipelining with SVN is just bit too insane...	03:43
deskin	you can use -rm:n where m and n are integers to get a range of revisions, or omit one and the colon for just that revision	03:43
	it's not a bad way to get up and running, and fetch the full history later and format-patch \| am to move your work	03:44
tag	hey that's great	03:46
	→ schacon joined	03:46
tstellar	I have a patch file in my git repo, but when I make changes to it and generate a diff, git apply can't apply the changes, becuase it gets confused by the file's contents(since the file itself is a patch) any suggestions?	03:47
	← scarabx left	03:48
nis	tstellar: are you diff'ing a patch file with a regular file?	03:49
Alien_Freak	I think you can tell it to just pull head	03:49
	← zpoley left	03:52
	→ h0nk1 joined	03:52
tstellar	nis: I am creating a diff of a diff file (if that makes sense).	03:52
	→ amahi joined	03:53
	→ KwikkSilva joined	03:53
h0nk1	hi, is there any porcelain to rebase a tree of branches onto another branch?	03:53
	→ arohner joined	03:53
deskin	tstellar: patches are just files and git should be able to parse diffs of diffs fine; what's git apply complaining about?	03:54
	→ repror___ joined	03:54
	← reprore left	03:54
	→ mediogre joined	03:54
	→ nis_ joined	03:55
	← damainman856 left	03:55
	→ alanhaggai joined	03:57
nis_	(i've been disconnected, sorry)	03:57
	tstellar: no, it doesn't make sense.	03:58
tstellar	Hold on, I think generated the patch backwards.	03:58
	→ ajonat joined	03:58
deskin	that'll do it	03:58
h0nk1	this is basically a stgit functionality, but I tend to do this quite often with interdependent topic branches I still work on	03:58
deskin	h0nk1: never used it myself, but topgit may have what you want	03:59
	this sounds like its use case	03:59
kold	Hi all, I'd like to find the first instance that a line of code in a file first appeared in git....can anyone offer suggestions on how I'd do that other than a 'git log -p \| grep 'code I am seeking'	03:59
	← ned left	03:59
tstellar	Yeah, I generated the patch in reverse, problem solved thanks.	04:00
deskin	kold: git blame	04:00
	er man git-blame	04:00
Gitbot	deskin: the git-blame manpage can be found at http://git.or.cz/man/git-blame	04:00
kold	deskin: thanks	04:00
h0nk1	deskin: looks good, thanks	04:00
kold	wasn't aware that is what blame was for, but it makes sense now	04:01
	→ ashleyw joined	04:02
	← halfmouse left	04:02
	← repror___ left	04:03
	→ Zenopus_ joined	04:03
	← jrmuizel left	04:04
deskin	:) you can also spell 'blame' as 'annotate' but that's just more to type	04:04
noclue	is it possible to tell git to store work files in a different directory than the repo files are stored in?	04:05
nis_	noclue: no. what do you want to do?	04:06
	deskin: how do you tell git-blame to find the first instance?	04:06
noclue	nis_: say i have a repo at /foo/.git and i want the work files stored in the foo directory instead of .git	04:07
	→ solar_ant joined	04:07
	← nis left	04:07
	→ jrmuizel joined	04:07
	← intripoon left	04:07
nis_	noclue: you may use a symlink	04:07
deskin	nis_: I'm not sure what you mean; the commit it shows is by definition the first instance; unless, you have the same 'line' in multiple places, that I'm not sure about	04:07
	noclue: you can use a different working tree, man git IIRC tells you the environment variables to set	04:08
	← cpg left	04:08
	→ alester joined	04:08
nis_	deskin: what i understood from kold is that he wants to find the first instance of a content in the history range.	04:09
noclue	nis_,deskin: thanks.	04:09
nis_	noclue: np!	04:09
	→ highandwild joined	04:09
highandwild	hi all	04:09
	I think I screwed up :)	04:09
nis_	deskin: which is not possible AFAIK.	04:09
highandwild	I have a refactor branch. I did: git-rebase -s ours master thinking ours referred to refactor, but it has refactored it to master !! how do I undo this !	04:10
sitaram	kold: (sorry; just looked at the backlog) git gui blame is even nicer -- reasonably interactive :)	04:10
	coney → allagator	04:10
sitaram	highandwild: did you start from a clean tree?	04:10
kold	ok, so I have blame telling me who authored the current line. the --since flag won't work since it will always show the latest....I need to know when a line first entered the file	04:10
	← kalle_ left	04:10
highandwild	sitaram: yes	04:10
kold	sitaram: thanks, looking at git gui now	04:10
nis_	highandwild: check your git reflog	04:11
sitaram	highandwild: then 'git reflog show HEAD@{now}' will show you the commit you want to reset to	04:11
Fissure	kold: git log -S	04:11
	might be helpful as well	04:11
	→ kalle_ joined	04:11
kold	Fissure: thanks	04:11
highandwild	I see a long git reflog	04:12
	71 lines	04:12
	→ Evanlec joined	04:12
deskin	nis_: you mean, first instance of a line in some unspecified file? That I don't know; the range can be done like git blame from..to file	04:13
Evanlec	How can i merge in 1 file from another branch into my current branch?	04:13
Fissure	Evanlec: not easily; git deals with whole trees	04:13
Evanlec	Fissure: ah	04:13
offby1	Evanlec: just copy the file	04:13
Evanlec	Fissure: alright, #mercurial didnt have an answer for me either, i thot it could be done in git tho	04:14
nis_	deskin: but how do you find the first instance of the line?	04:14
Fissure	you can use git show to get the other version and use rcs merge...	04:14
offby1	betcha bzr won't have an answer either	04:14
highandwild	sitaram: the refactor was ahead of the master, and I think one or two commits are missing from it now	04:14
Evanlec	offby1: makes sense	04:14
sitaram	highandwild: page it, look from the top until you see something useful or pastebin it and I will helpyou interpret it	04:14
Evanlec	Fissure: simple is better ;)	04:14
deskin	nis_: git blame; that's the same question you asked before, what do you mean more specifically?	04:15
highandwild	sitaram: ok, suppose I want to reset to the last line which says: "3 months ago". Will I get my current branch back to the state it was ?	04:15
	→ ned joined	04:15
nis_	deskin: let's say you added a function new() at a point.	04:16
	← Zenopus left	04:16
sitaram	highandwild: if the line above that gives you a time that roughly corresponds to your incorrect rebase, yes	04:16
	Zenopus_ → Zenopus	04:16
nis_	deskin: since you did a lot of commit.	04:16
	deskin: how can git blame help you to find the commit where new() was added?	04:17
highandwild	sitaram: but then, that would be the first line which says "8 minutes ago"	04:17
	← KwikkSilva left	04:17
Fissure	nis_: well, git works on lines and has no knowledge of functions	04:18
sitaram	highandwild: if you have nothing between 8 minutes ago and 3 months ago, and it was 8 minutes ago that you did this, then yes -- you need to reset to that 3 months ago SHA	04:18
Fissure	blame will tell you when the line containing the function header was created, though	04:18
nis_	Fissure: thank you. I'm on an example. :-)	04:18
sitaram	though I assume you should sanity check the idea that nothing happened on this repo for the past 3 months :)	04:18
deskin	nis_: that's not the same as the first instance of a line :/ I'd probably use git log -u \| grep	04:18
highandwild	sitaram: http://pastie.org/private/42velhfxlkk6i7flogvka	04:19
sitaram	looking	04:19
deskin	if new()'s prototype never changed, you can just look at the commit for that line in git blame	04:19
nis_	Fissure: and if you don't have header?	04:19
	Fissure: deskin: you're both assuming it is C content, which may not be the case.	04:20
sitaram	highandwild: last sane point before most recent activity for master seems to be line 4 (SHA 579520) and for refactor, line 3 (a57bdf5)	04:21
nis_	Fissure: deskin: what i understood from kold is that he wants to find when "new()" was added the first time.	04:21
highandwild	sitaram: so I just do a git reset [ref] ?	04:22
	← jchris left	04:22
	← Alinermedia left	04:22
	← Kimmie left	04:22
Fissure	highandwild: you probably want --hard	04:22
deskin	no, I'm not; same applies to java, python, javascript; I guess perl is exempt	04:22
nis_	Fissure: deskin: AFAIK, there is no way to do that easily with git	04:23
sitaram	highandwild: I'd suggest adding '--hard'	04:23
highandwild	success !	04:23
	sitaram, Fissure: thanks	04:23
deskin	you are correct, for some definition of 'easily'	04:23
sitaram	you're welcome ,,,	04:23
	← noclue left	04:23
	→ damainman856 joined	04:23
nis_	deskin: since when python has headers?	04:23
highandwild	a related quesiton. How do I merge a single file from master into refactor ?	04:23
deskin	there is no well-defined way, certainly; you'll have to look at diffs or iterate on git blame somehow	04:23
Fissure	nis_: well, do you expect git to have intimate knowledge of all the intricacies of all programming languages?	04:23
sitaram	highandwild: let's not use the word merge -- that has a specific meaning in any VCS	04:24
deskin	prototype != header	04:24
highandwild	sitaram: ok ... should I just export the diff as a patch and apply it ?	04:24
deskin	when you define the function, you have to tell its arguments	04:24
nis_	Fissure: no, certainly not.	04:24
sitaram	highandwild: what do you actually want to do? take the version from master and put it on refactor?	04:24
highandwild	sitaram: actually, I really do want to rebase refactor upon master, and I want the resulting HEAD to be the HEAD of refactor	04:24
Fissure	nis_: then how could it give you a complete answer on whether a function was added?	04:25
	it only knows about _lines_	04:25
highandwild	sitaram: or even a merge, with the resulting head = refactor's HEAD	04:25
sitaram	highandwild: git checkout refactor ; git rebase master	04:25
nis_	Fissure: that's partially what i would answer to kold. :-)	04:26
	→ gwoo joined	04:26
	→ bentob0x joined	04:26
Fissure	nis_: claiming to be a proxy arguer doesn't excuse you from being wrong	04:27
nis_	Fissure: where?	04:28
	← Tricon left	04:28
highandwild	sitaram: sheesh :)	04:29
	→ moh joined	04:30
	→ intripoon joined	04:30
	→ bob2 joined	04:31
nis_	so, does git blame tell you who/when a line was _created_?	04:31
	bob2 → Guest37195	04:32
deskin	yes; it's the first commit where the line existed	04:32
nis_	deskin: find that commit is the goal.	04:33
	← solar_ant left	04:34
	→ hax0r1337 joined	04:34
sitaram	nis_: the gui version will let you follow the history of that line and surrounding lines	04:35
Arrowmaster	in which gui?	04:35
sitaram	git gui blame	04:35
nis_	sitaram: yeah, that's the best way i know	04:35
	→ aspotashev\|eeepc joined	04:36
	→ reprore joined	04:36
nis_	sitaram: but if you have dozen or hundreds commits to forward, it's not pretty	04:36
sitaram	forward?	04:37
	sitaram didn't understand... :(	04:37
nis_	(to follow)	04:37
	→ faithfulgeek joined	04:38
	offby1 bets that "forward" is a common term with some other system	04:38
offby1	and we git-heads simply don't know what it means	04:38
	probably something like "cherry-pick"	04:38
	← snowman856 left	04:38
nis_	offby1: no, it's my english knowledge	04:39
sitaram	as in "step through" perhaps...?	04:39
nis_	yeah	04:39
	← highandwild left	04:39
sitaram	yeah in that case I agree, but then you need "gitk --all -- filename"	04:40
nis_	it could help.	04:40
	→ solar_ant joined	04:41
nis_	but i'm still not convinced as the file history could go trough hundreds of commits	04:41
	s/trough/through/	04:41
	parsing git log -p could be easier	04:42
	← neunon left	04:42
deskin	nis_: I'm not sure why you need to look through commits given git-blame; doing it iteratively is a fine approach if you want the non-trivial first appearance of a function	04:44
nis_	iteratively with git blame?	04:45
	or git gui?	04:45
sitaram	nis_: if I have 100s of commits I'd use 'git log -S'	04:46
deskin	sure: run git blame. decide that the version of the function isn't what you want; so, pick the commit next to it, and do git blame <commit>~1 ...	04:46
	rinse & repeat	04:46
sitaram	then -- if needed -- zero in on that hash using gitk, or "git gui blame SHA file" to start the gui blame at that hash	04:46
deskin	I don't use git gui so can't help you there	04:46
sitaram	deskin: you don't need to restart -- just click the links in git gui blame	04:47
	deskin: don't == won't or can't ? (no offense; just curious)	04:47
	← Voight-Kampf left	04:47
sitaram	that rinse and repeat is what guis are made for :)	04:48
deskin	haven't found a need, and am comfortable with the command line, so more of the former; though won't is a little too strong	04:48
	true that	04:48
	← arohner left	04:49
sitaram	deskin: try it once. Git's guis converted me from an exclusive command line jock to understanding that a GUI has a place (outside firefox and mplayer and gimp!)	04:49
Fissure	i'm unsure why you're including mplayer in that list...	04:50
deskin	I use gitk a reasonable bit	04:50
	I guess I worry I'll end up needing a gui to do basic stuff, and then being sshed to a machine but unable to get any work done	04:50
	→ mithro joined	04:51
	← Guest37195 left	04:52
sitaram	Fissure: gmplayer?	04:52
	sitaram nods to what deskin said about ssh	04:52
sitaram	Fissure: I see what you mean; I never use the mouse for gmplayer anyway :)	04:53
	→ jpld joined	04:53
	← wsc left	04:56
	→ reprore__ joined	04:56
	← solar_ant left	04:58
	→ xtc joined	04:58
	→ solar_ant joined	04:58
	→ jceb joined	04:59
	→ girishr_ joined	04:59
	← girishr left	05:00
	← reprore left	05:02
Fissure	what's wrong with ssh -X?	05:03
nis_	it's not very fast	05:03
	→ Pewpewarrows joined	05:03
deskin	not always an option; bandwidth, or maybe the remote lacks the gui/ x client code; or maybe I'm using cygwin o_O	05:04
	← petercoulton left	05:04
	→ naeu joined	05:04
deskin	...or my handheld	05:04
Fissure	i opened gitk up over wireless earlier today :)	05:05
	← curvature left	05:05
	← solar_ant left	05:05
	→ curvature joined	05:05
	harv_ → harv	05:05
	← naeu left	05:05
deskin	sure, but I'm not going to do it on my iPAQ; doesn't even have a proper x server	05:05
	← schacon left	05:06
	← nis_ left	05:07
	← jrmuizel left	05:07
	← xtc left	05:09
	→ reithi joined	05:09
	← gberdyshev left	05:09
sitaram	ssh -C -X works sort of ok but it isn't snappy enough -- and that "snappy" makes all the difference to usable GUI :(	05:09
	→ jrmuizel joined	05:10
sitaram	however, my stand is that stuff that is used rarely, and for which I know the CLI eqvt (if any) I don't mind using the GUI	05:10
	→ disappearedng joined	05:10
	← kalle_ left	05:10
	→ kalle_ joined	05:10
disappearedng	Hey how does git clone automatically check out the new patches	05:11
RandalSchwartz	git clone doesn't check anything new out	05:11
	it just clones	05:11
disappearedng	like I have the original code, and someone made a commit, if I do a checkout will the diff be downloaded	05:11
	hm	05:11
RandalSchwartz	you want git fetch once you ahve a repo	05:11
sitaram	you need a pull or a fetch	05:11
RandalSchwartz	don't mention the "p" word to newbies	05:12
disappearedng	I am n00b btw, how do I create a repo?	05:12
RandalSchwartz	they treat it as a shiny object	05:12
	create a repo of your own?	05:12
	or clone an existing repo?	05:12
sitaram	RandalSchwartz: :)	05:12
disappearedng	like hm.. http://eclim.sourceforge.net/guides/development.html\	05:12
RandalSchwartz	git init, or git clone	05:12
	those are your choices	05:12
	are you the originator of the first tree?	05:12
	→ wezyde joined	05:12
disappearedng	I just want to download the new patches from the external repo	05:13
	no	05:13
	→ xtc joined	05:13
RandalSchwartz	or are you coming into the middle of existing project?	05:13
disappearedng	yeah	05:13
RandalSchwartz	ahh. you want to clone the repo	05:13
	not "new" patches	05:13
	but all patches	05:13
disappearedng	I have cloned it yesterday	05:13
RandalSchwartz	clone builts a local .git that is a copy of the remote .git	05:13
	AUgh	05:13
	so you already have a repo	05:13
	you want fetch then	05:13
	not clone	05:13
	RandalSchwartz sighs	05:13
disappearedng	ok	05:13
	disappearedng apologises	05:13
disappearedng	i thought repo meant like SVN repository	05:13
	← mediogre left	05:15
disappearedng	git fetch git://eclim.git.sourceforge.net/gitroot/eclim that complains that it's not a git repository	05:15
RandalSchwartz	No you already cloned	05:15
	the clone contains pointers to where it came from	05:15
disappearedng	ok	05:15
	so	05:16
	i actually have no idea what you are talking about btw	05:16
	→ priidu joined	05:16
	← travisjeffery left	05:16
	← jrmuizel left	05:18
sitaram	disappearedng: when you clone, you get a copy of the repository	05:18
disappearedng	ok	05:18
sitaram	within your copy, the parent from which you cloned is referenced as a "remote" called "origin"	05:18
	← wezyde left	05:18
sitaram	so you now do "git fetch origin" to get the latest changes from there	05:19
disappearedng	ok	05:19
	I am getting a fatal: Not a git repository	05:19
sitaram	disappearedng: then you did not clone yet? or you're in the wrong directory	05:20
deskin	or the remote disappeared	05:20
disappearedng	ok I was in the parent directory	05:20
sitaram	that wont get "not a git repo..."	05:20
deskin	er, my bad	05:20
disappearedng	when I go into the source directory everything worked thx	05:20
	← xenoterracide left	05:20
disappearedng	so why git over SVN?	05:20
sitaram	you're welcome	05:20
	xtc → xenoterracide	05:20
	sitaram gracefully bows out and lets other take the stage; the magic S word came up	05:21
disappearedng	lol	05:21
Ilari	disappearedng: And after fetching changes, you presumably want to rebase your changes on top of upstream or merge your changes with upstream...	05:22
	→ alanhaggai_ joined	05:22
disappearedng	Ilari: and how do I do that/	05:22
RandalSchwartz	I try not to run git over svn. :)	05:22
	← alanhaggai left	05:22
	alanhaggai_ → alanhaggai	05:23
disappearedng	cause I realize that after I did a git fetch origin the code remained the same	05:23
Ilari	disappearedng: 'git rebase origin/master' (assuming its master of remote you are following) or 'git merge origin/master'.	05:23
	disappearedng: Really depends on nature of changes you have and kind of access to upstream...	05:24
	← student0x77 left	05:24
	← baetis-fly left	05:24
	→ offbytwo joined	05:24
	← offbytwo left	05:24
	← ben_h left	05:25
	← giallu left	05:27
	→ Paf joined	05:27
	→ schacon joined	05:28
	← aziz left	05:28
	→ sea-gul1 joined	05:31
	← reithi left	05:32
Ilari	disappearedng: Basically, rebasing gives nicer history and is compatible with submitting changes as patches. But merging doesn't change commits on either side...	05:33
	← fujin left	05:34
	→ aisbaa joined	05:36
	← bambule left	05:37
	→ ReAn` joined	05:37
	→ Tuomas joined	05:38
	→ tim_jenssen joined	05:38
	← xenoterracide left	05:39
	→ xenoterracide joined	05:41
	← dedmajor left	05:42
	→ dedmajor joined	05:42
	← reprore__ left	05:43
	AkumaStreak → AkumaStreak\|zzz	05:46
	← schacon left	05:47
	← jpld left	05:47
	← AkumaStreak\|zzz left	05:47
	← km2 left	05:51
	← cydork left	05:55
	→ reprore joined	05:56
	ReAn` → ReAn	05:56
	→ mndoci joined	05:58
	← hax0r1337 left	05:59
	→ solar_ant joined	06:01
	← jceb left	06:02
	← reprore left	06:02
	→ johan-s joined	06:06
	→ cpg joined	06:07
	← amahi left	06:08
	← kalle_ left	06:10
	→ kalle_ joined	06:11
	← bcardarella left	06:12
	← robinr left	06:13
	→ alanhaggai_ joined	06:13
	→ ph^ joined	06:15
	→ fujin joined	06:15
	← aspotashev\|eeepc left	06:16
	← altrux left	06:16
	→ aspotashev\|eeepc joined	06:17
	← alanhaggai left	06:17
	alanhaggai_ → alanhaggai	06:17
	← solar_ant left	06:18
	→ xtc joined	06:19
	← xenoterracide left	06:19
	← Paf left	06:21
	← xtc left	06:21
	→ xenoterracide joined	06:22
	→ solar_ant joined	06:26
	← solar_ant left	06:27
	→ srid` joined	06:29
	→ rittyan joined	06:30
	← deskin left	06:30
	→ deskin joined	06:32
	→ ben_h joined	06:35
	← mndoci left	06:35
	→ amahi joined	06:36
	← amahi left	06:36
	→ cilly joined	06:36
	→ amahi joined	06:36
	→ marc_kdab joined	06:37
	→ Grahack joined	06:37
	→ solar_ant joined	06:37
	← aisbaa left	06:38
	← stuffcorpse left	06:38
	← greyface left	06:38
	← ngng left	06:38
	→ ngng joined	06:38
	← xenoterracide left	06:39
	→ giallu joined	06:40
	→ xenoterracide joined	06:40
	→ stuffcorpse joined	06:41
	→ bobmcw_ joined	06:43
	← bobmcw left	06:43
	→ tal67 joined	06:43
astrocub	how do you keep track of submitted changes in git?	06:44
	→ Weaselweb joined	06:44
voker57	git log?	06:45
tclineks	astrocub: log?	06:45
drizzd_	astrocub: do you mean what got merged upstream and what didn't?	06:45
astrocub	yeah, everything	06:46
	→ WALoeIII joined	06:47
	← xenoterracide left	06:47
	← krawek_ left	06:48
	→ fridim_ joined	06:48
	← tstellar left	06:49
	→ xenoterracide joined	06:49
	← cpg left	06:51
	amahi → cpg	06:51
	→ octagon joined	06:51
	→ khmarbaise joined	07:00
	→ aspotashev_ joined	07:02
	← johan-s left	07:04
	← fujin left	07:09
	← kalle_ left	07:10
	← tal67 left	07:11
	→ fujin joined	07:12
	← aspotashev\|eeepc left	07:13
	→ Bennid joined	07:13
	← cilly left	07:14
	← GyrosGeier left	07:15
	→ kalle_ joined	07:16
	← deskin left	07:16
	→ meyering joined	07:17
	→ tal67 joined	07:17
madduck	in trying to import one remote namespace from one repo to another, e.g. copy refs/remotes/one/* to refs/remotes/two/*, I have hit the point where the refs do not point to valid objects	07:21
	e.g.	07:21
	error: refs/remotes/upstream/tags/release-0.9 does not point to a valid object!	07:21
	jayallen → jayallen-afk	07:22
madduck	is there a good way for me to import all the objects from the first remote that I need into the current repo?	07:22
	(this is a revived git-svn repo, thus the remote refs don't really exist, but I need them)	07:22
	→ tatsuyao joined	07:23
	← jkp left	07:27
	→ jkp joined	07:27
	→ drizzd joined	07:29
	← dedmajor left	07:30
	→ thumper joined	07:32
	← fujin left	07:32
	← srid` left	07:33
	← bentob0x left	07:34
	→ galderz joined	07:36
	→ aspotashev joined	07:36
	→ johan-s joined	07:37
xenoterracide	is there a way to say use auto-commit message	07:38
	← ToxicFrog left	07:38
xenoterracide	for example merges have a default message	07:38
	→ bentob0x joined	07:38
xenoterracide	and instead of actually viewing it is there a way to just commit	07:38
	kind like I can do with -m but without it	07:38
madduck	you could use --no-commit and then commit with -m yourself	07:39
	← hgimenez left	07:39
xenoterracide	erm that makes no sense for what I want	07:39
	I want it to just commit	07:39
	→ harinath joined	07:39
xenoterracide	I want to run commit without specifying a -m	07:39
	on a merge	07:39
	and just have it commit no editor	07:40
	← harinath left	07:40
madduck	sed editor to cat?	07:40
	or somesuch	07:40
xenoterracide	nope	07:41
	bad idea	07:41
	→ ashleyw_ joined	07:41
xenoterracide	since it's only 1 merge that I have to do all the time that I want this to do	07:41
	→ mbuf joined	07:41
	← ashleyw left	07:43
	← damainman856 left	07:45
	→ tatsuyao_ joined	07:47
	→ zpoley joined	07:47
	← Predominant left	07:50
	← tatsuyao_ left	07:51
Grahack	hi all, suppose I'd like to contribute to a project. There is the SourceForge git repo, my local git repo where I work, and a hosted git repo from where I want the maintainer (or anyone) to track my changes. So I clone the SF repo to my laptop to create my local repo but how should I create the hosted one ? cloning from SF or my laptop ?	07:52
	→ jmesnil joined	07:53
	→ chalky joined	07:53
xenoterracide	since when did SF get git?	07:53
	I must have missed that	07:53
sitaram	both should be the same, assuming your work is in a named branch that is unique to you. You can clone from SF and push just your branch too	07:53
	xenoterracide: about 3 weeks ago :)	07:53
	maybe a tad more	07:53
xenoterracide	sitaram: huh	07:53
	I missed that news	07:53
Grahack	thanks sitaram	07:53
sitaram	didn't make waves; I can understand the reluctance of people who own SVN not to publicise it too much that they now support git :)	07:54
xenoterracide	now google needs to get it ;)	07:54
	→ harinath joined	07:54
	← harinath left	07:54
sitaram	I would have thought that would happen first (a) they dont have a vested interest in a competing legacy product and (b) they'd want to unseat SF	07:55
	butit didn't	07:55
	← porcnick left	07:55
voker57	they support hg and bzr too now afaik	07:55
	← kuadrosx left	07:56
xenoterracide	nice	07:56
	anyone know what the max upload (outside of repos) is for SF?	07:57
	← drizzd_ left	07:59
	→ fujin joined	07:59
Grahack	anyone familiar with gitosis here ?	08:00
	← tatsuyao left	08:00
	→ jfkw joined	08:01
Grahack	I was wondering about the purpose to create a fake user: security ?	08:01
	← darrob left	08:01
	→ Predominant joined	08:01
	→ shruggar joined	08:02
	→ Paf joined	08:04
	← thraxil left	08:05
	← stuffcorpse left	08:05
	→ thraxil joined	08:05
	→ AlexB joined	08:05
	→ Leonidas joined	08:05
	→ Helios joined	08:05
	→ Dreamer3 joined	08:05
	→ nick_h[litage] joined	08:05
	→ ArjenL joined	08:05
	→ osfameron joined	08:05
	→ BooK joined	08:05
	→ orospakr joined	08:05
	→ stuffcorpse joined	08:05
	→ HG` joined	08:05
	→ darrob joined	08:06
sitaram	what fake user?	08:07
	oh the "git" one?	08:08
	well it removes the need to create actual users on the system for all devs, which is a good thing, so yes: security	08:08
	← kalle_ left	08:10
	→ kalle_ joined	08:11
	→ iws joined	08:11
	→ Fullmoon joined	08:13
	← ned left	08:13
	← orafu left	08:13
	→ orafu joined	08:13
	→ mikl joined	08:16
	← tag left	08:17
	← jkp left	08:20
	→ jkp joined	08:20
	→ GyrosGeier joined	08:20
	→ naeu joined	08:22
Grahack	sitaram: because make gitosis use our user would work too I guess ?	08:23
	→ lucsky joined	08:23
	→ Chaser joined	08:27
	→ egoist joined	08:27
Ilari	Grahack: Making gitosis share account with something else isn't recommended.	08:29
martins	touch vendor/gems/see_me; git status # see_me and the entire gems directory is invisible. How can I make git keep track of changes in this directory? git add vendor # returns nothing.	08:29
	→ ned joined	08:29
	→ plasticine joined	08:34
	→ R_Dday joined	08:35
	← R_Dday left	08:35
	→ ceyusa joined	08:35
	→ R_Dday joined	08:35
	← R_Dday left	08:35
	→ samjam joined	08:37
	→ jkp_ joined	08:39
	← jkp left	08:39
	← mithro left	08:40
	→ Tricon joined	08:43
	← egoist left	08:44
	→ tono joined	08:45
	→ sejo joined	08:45
	→ dedmajor joined	08:47
	← agenteo left	08:48
	← Grahack left	08:49
sitaram	is vendor or vendor/gems in .gitignore or similar? otherwise git status should point to one of those things as "new, untracked"	08:51
martins	sitaram: no, I´ve checked! :)	08:51
	what´s similar, btw?	08:51
sitaram	.git/info/exclude	08:51
	→ ceyusa_ joined	08:52
	→ botanicus joined	08:52
sitaram	also rememebr .gitignore can be in any subdrectory also -- you have to check each directory in the chain	08:52
	→ nis joined	08:54
	← ashleyw_ left	08:54
	← ceyusa left	08:55
	← octagon left	08:57
	← ajonat left	08:58
	→ simmel1 joined	08:58
simmel1	hi, i got the following error from git, do i need to worry?	08:59
	error: Object 4b5127df968616dee2f4775d795198878ef1638b is a commit, not a tag	08:59
	fatal: Invalid symmetric difference expression 3fa8749e584b55f1180411ab1b51117190bac1e5...4b5127df968616dee2f4775d795198878ef1638b	08:59
martins	find . -name .gitignore -exec rm {} \; # And vendor/gems still won´t show up	09:00
sitaram	simmel1: what were you doing to trigger it? (Not that I know what this is, I don't...)	09:01
	→ MisterN joined	09:02
sitaram	martins: make sure you're in the right directory and not in some subdirectory elsewhere :) (can happen!)	09:02
	sdiz → sdiz-afk	09:03
	→ solar_ant_ joined	09:04
	← solar_ant left	09:04
martins	sitaram: hehe, I´ll try a few more things before I do an rm -rf vendors && mkdir -p !$/{gems,vendor} # and hopes I can rebuild it. :)	09:04
simmel1	sitaram: i just recognized i setup a tracking branch on a tag ... story goes like this:	09:05
martins	mkdir -p !$/{gems,plugins}	09:05
simmel1	git branch --track foo-2.6.27-patchesx v2.6.27	09:05
	git checkout foo-2.6.27-patchesx	09:05
	maybe thats the problem	09:06
	→ aser joined	09:07
aser	hi, is there a way to see a diff between a commit and current working directory including untracked files?	09:07
	→ SRabbelier joined	09:08
wereHamster	aser: git-add the untracked files :)	09:08
nis	aser: untracked files are not known by the git repo	09:09
	← bobesponja left	09:09
aser	wereHamster: well, I didn't really want to add them	09:10
	for example I wanted to sync my working dir with a commit, but see what will happen if I do a reset	09:10
	← kalle_ left	09:10
wereHamster	aser: why would you want to see the diff then?	09:10
	→ kalle_ joined	09:10
aser	I guess it would show the conflicts anyway, but...	09:11
	← Bennid left	09:11
	← solar_ant_ left	09:11
	→ solar_ant joined	09:11
aser	would be nice to have git-reset --dry-run	09:12
	← mikl left	09:12
wereHamster	you can always reset and then use the reflog to go back..	09:12
aser	and another thing... reset wouldn't really delete anything, would it?	09:13
wereHamster	I believe it would refuse to overwrite untracked files, but I'm not sure	09:13
	and tracked files you can always recover	09:13
sitaram	aser: git reset does nothing to untracked files -- that is git clean	09:15
	and a diff for untracked files does not make sense because there is "before" image (or is it /dev/null)	09:15
aser	sitaram: "before" image?	09:16
sitaram	yeah, a diff is .. well.. a difference between "before" and "after" isn't it?	09:17
aser	sitaram: how about a diff between a path and a tree? would it not be possible?	09:17
	a tree object to be precise	09:17
sitaram	sure...	09:17
	I think so (lemme check)	09:17
	yeah works fine	09:19
aser	sitaram: touch file; echo "x" > file; git diff HEAD..<what_here>?	09:22
sitaram	aser: oh that... I thought you meant a tracked directory/files	09:22
	that won't show, I think	09:23
aser	hm, what if I extracted a tree object from a commit object and tried that...	09:23
sitaram	aser: as long as the actual file you want to see in the diff is untracked, nothing will work	09:24
aser	sitaram: ah ok... oh well, I'll have to live with that. thank you!	09:24
	and thank you wereHamster and nis	09:24
sitaram	because the left side of the diff is... nothing. Zilch. Nada. Zip. Shunyam. Sunna (I snuck in a couple of Indian words there ;-)	09:24
	→ gilimanjaro joined	09:24
aser	hehe	09:24
	→ dwt joined	09:27
	← Octalot left	09:27
dwt	Hey there, I've fuddled my local history a bit, and am now wondering how to get out of that situation. Specifically I have a local commit that I want to get rid of, but I also have remote changes that I accidentally pulled. So now I'm in a bad merge and git won't let me do a git revert HEAD^ - which should be the right command from what I understand from the manual.	09:28
	So... what to do?	09:28
	→ mnemo joined	09:28
mnemo	I would like to make git always use the --color parameter for "git diff".... my first attempt was to define a bash alias as alias 'git diff'='git diff --color' but apparently aliases cant contain spaces... any other way to do this?	09:29
aser	dwt: git revert would create a merge... since you are in the middle of one...	09:30
	mnemo: ~/.gitconfig	09:30
	mnemo: and use [color] options	09:30
dwt	damn. Is there something like "git undo the last two steps" I'd like that	09:30
mnemo	aser: ah.. thanks	09:30
aser	mnemo: man git-config will give you the details	09:31
Gitbot	mnemo: the git-config manpage can be found at http://git.or.cz/man/git-config	09:31
mnemo	right found it	09:32
sitaram	mnemo: man git-config and look for color.ui	09:32
Gitbot	mnemo: the git-config manpage can be found at http://git.or.cz/man/git-config	09:32
sitaram	there's a whole bunch of color options, but color.ui is like a catchall, I think	09:32
simmel1	so i think i got it so far	09:32
	git allows to create tracking branches on lightweight and annotated branches	09:33
	arrr tags i meant	09:33
	git allows to create tracking branches on lightweight and annotated tags	09:33
	but when you checkout such a branch on an annotated tag, you get something like	09:34
	fatal: Invalid symmetric difference expression efc968d450e013049a662d22727cf132618dcb2f...e73a9936a2c2bd8f8f408417ac7af7c94043857b	09:35
	→ FreakGuard joined	09:35
	← deavid left	09:35
	← alanhaggai left	09:36
simmel1	efc... is the commit, e73... is the tag	09:36
	→ alanhaggai joined	09:36
simmel1	you also get error: Object e73a9936a2c2bd8f8f408417ac7af7c94043857b is a commit, not a tag	09:37
	but	09:37
	git-cat-file -t e73a9936a2c2bd8f8f408417ac7af7c94043857b	09:37
	→ botanicus_ joined	09:38
Grum	dwt: just as a quick sidenote, you know git revert doesnt 'undo' a commit but simply creates a new one with the 'negative patch applied' ?	09:38
aser	dwt: may you want to git-reset?	09:38
simmel1	any ideas? not that it was a good idea anyways to track a tag, just the error messages are confusing	09:38
	← NoirSoldats left	09:38
dwt	aser: possibly	09:38
	I think I managed to get out of the merge with a git reset --hard HEAD	09:39
	← cedricv left	09:39
dwt	so I probably need to go back with git reset --hard HEAD^^ (because I've now created the undo commit)	09:39
aser	dwt: yes, as long as you don't care about the current state of your working dir, reset is your friend	09:39
dwt	well, I do	09:39
	but there's nothing in there that can't be readded with two minutes of work	09:39
	I think I've got it	09:40
	thanks	09:40
	← botanicus left	09:42
	→ cedricv joined	09:42
	→ yawn joined	09:43
aser	dwt: if I were you, I'd create a backup of the directory, do a reset and recopy manually... :)	09:43
	← cedricv left	09:43
dwt	Well, too late for that now	09:44
	→ cedricv joined	09:44
dwt	:)	09:44
	← Paf left	09:44
dwt	Well, still learning to cope with git. It's got quite a twisted sense of humor sometimes. :/	09:44
yawn	i merged a big piece of development (A) into an unrelated repository (B). now i have a couple of conflicts and would like to resolve them in favor of (A). do i need to do this manually with mergetool?	09:45
sitaram	simmel1: this looks like something serious enough to warrant a mail to the list; it's not even clear to me whether tracking a tag is reasonable (going by the usual meaning of a tag as being a fixed point)	09:45
	→ naeu_ joined	09:46
	← naeu left	09:47
	sejo → masterhumper	09:48
	masterhumper → OO11	09:49
	← ben_h left	09:50
	→ solydzajs joined	09:52
	→ ntoll joined	09:56
	← orafu left	09:57
simmel1	sitaram, i have written a mail to the list	10:00
sitaram	cool!	10:01
	→ orafu joined	10:02
	← MisterN left	10:02
	← orafu left	10:03
	→ Silvurial[Ag] joined	10:06
	← celling left	10:06
	← aspotashev left	10:06
	← simmel1 left	10:09
	jmesnil → jmesnil_away	10:10
yawn	in other words - i would like the opposite of "ours"	10:10
	as pull strategy	10:10
	← kalle_ left	10:10
	→ kalle_ joined	10:11
	→ bratsche_ joined	10:13
sitaram	yawn: you can do something like 'git checkout --theirs -- filename' for each such file	10:14
Ilari	yawn: There isn't one. You need to create temp branch, merge your branch with -s ours and then merge the temp branch with default merge strategy	10:15
yawn	sitaram: thanks! right now i'm trying to use "ours" on the other end and manually update the config on that	10:15
	→ majeru joined	10:16
yawn	which is not so nice ... bad order of commits :-)	10:16
	i'll try your suggestions	10:16
majeru	hello, is there a way to join two svn repos in a single git repo, and be able to keep in sync with them?	10:16
	and also track local changes to them	10:17
gebi	yes	10:18
	majeru: man git-svn (--prefix)	10:19
Gitbot	majeru: the git-svn manpage can be found at http://git.or.cz/man/git-svn	10:19
majeru	thanks	10:19
	i have read the manual page, but it's not that explicitly suited for my use case, or maybe I'm not understanding it good enough	10:21
yawn	llari: can you elaborate? i need to create a temp branch where?	10:21
	← nis left	10:23
	aser → g3d	10:24
	← mnemo left	10:28
majeru	gebi: the repos are different, and have nothing in common	10:30
yawn	basically i am just interested in attaching the old history to the new repository w/o checking out anything	10:31
majeru	as i can see it, --prefix seems to be for similar repos	10:31
hgb	Hi. I have a git repo using git-svn. It's in ~/src/dev/git, as a full clone thingy, with repo and working copy. I use several branches in there, routinely doing 'git svn rebase' to pick up changes from svn.	10:31
	However, when I work for a while in a branch, the other branches, including 'master' gets hopelessly out of date (we have a very active svn repo).	10:32
	What I normally do is 'git checkout master; git svn rebase; git checkout <my current branch>'	10:32
	The problem with this is that thousands of files are written to, although nothing is really changed (after the second checkout).	10:33
	So I'd like to have a working copy of master, where I can do git svn rebase and no one notices.	10:33
	And then next time I do git checkout master in ~/src/dev/git, it'll be fairly up-to-date.	10:33
	Is that a simple matter?	10:33
	→ Grahack joined	10:33
	→ j416 joined	10:34
	→ diadema joined	10:34
	sdiz-afk → sdiz	10:39
	→ ProperNoun joined	10:43
	→ KwikkSilva joined	10:43
	← Silvurial[Ag] left	10:44
	→ aspotashev joined	10:45
	← HG` left	10:48
	← aspotashev left	10:51
	→ hobodave joined	10:51
hobodave	hey guys	10:51
	← mtl left	10:52
	→ aspotashev joined	10:52
	→ jschoolcraft joined	10:53
	← seangrove_ left	10:53
	→ dwave joined	10:53
	→ tp1 joined	10:54
jschoolcraft	Not sure if this is the appropriate channel, if not please point me in the right direction. Has anyone set up atom/rss feeds for commit messages from gitosis hosted repos?	10:54
	→ Amenthes joined	10:56
	→ girishr joined	10:56
Amenthes	Hi, I have a git repository which gives me a "fatal: Out of memory, malloc failed" when gc-ing. What solutions do I have for this problem? I'm on Windows with the latest msysgit version.	10:57
	← CardinalNumber left	10:59
	→ grahal joined	10:59
	→ cemerick joined	11:02
Ilari	Amenthes: Large files in repo?	11:03
	→ knitt1 joined	11:03
	← tal67 left	11:04
Amenthes	Ilari, nope. The largest seems to be a git file which has 529 MB. Does it matter that I seldom do gc?	11:07
Ilari	Amenthes: .pack file?	11:07
Amenthes	Ilari, no. It has no extension. The second one, which is indeed a pack file has 75 MB.	11:08
	→ simmelpp joined	11:09
Ilari	Amenthes: What's the name of the 529MB file?	11:09
Amenthes	Ilari, it's a hash. Aparently from a previous commit. It's in the objects directory	11:10
gebi	majeru: doesn't matter, to add a second svn repo to a current one eg.: git svn -s --prefix foo/ -Rfoo http://svn.foo.org; git svn fetch foo	11:10
	← girishr_ left	11:10
	← kalle_ left	11:10
	→ kalle_ joined	11:11
	← solar_ant left	11:11
	← voker57 left	11:12
	← KwikkSilva left	11:15
	← jschoolcraft left	11:15
	← cpg left	11:18
	→ solar_ant joined	11:19
	→ egoist joined	11:20
	← dreiss left	11:20
Ilari	Amenthes: 38 hex digit named file inside objects/<2 hex digits>'?	11:20
	→ CardinalNumber joined	11:20
Amenthes	Ilari, exactly	11:20
	→ KwikkSilva joined	11:21
	← davidad left	11:22
sitaram	that's an actual object; 529MB zlib compressed! wow...	11:23
Ilari	Amenthes: Sounds like you added some large file...	11:23
sitaram	ISTR that git requires approx 2.5x the largest file in RAM to pack...?	11:23
	← Gitzilla left	11:24
Amenthes	How do I see the commit message for that hash?	11:24
wereHamster	Amenthes: that hash doesn't have to be a commit. See with: git cat-file -t $hash	11:25
	→ Gitzilla joined	11:25
sitaram	wereHamster: it won't be a commit! it'll be a file	11:26
	(that big, a commit? naah!)	11:26
wereHamster	sitaram: pipe /dev/random into git-commit and voila, a 500MB commit message :)	11:26
	← egoist left	11:27
Amenthes	What if git cat-file -t $hash gives me "fatal: Not a valid object name 51e2cf335373ebc0a2629a8cc8d17fb3705d0a" ?	11:28
	← giallu left	11:28
wereHamster	Amenthes: prepend the <2 hex digits> in front of the filename	11:28
majeru	gebi: i want to have them in different subdirectories, if possible	11:30
Amenthes	wereHamster, thanks. It says blob. Now I should see what large file I commited...	11:30
sitaram	Amenthes: git rev-list --all \| while read r; do git ls-tree -r -t $r \| grep d74c && echo $r; done	11:31
	where d74c is the hash I am looking for -- change it to the first 4 characters or so of yours	11:31
gebi	majeru: then why not just make 2 repositories?	11:31
sitaram	then gitk <the result SHA> (or one of them)	11:31
majeru	gebi: well, i'd like to unify them as sub-modules	11:32
	nad to be able to commit stuff to both of them, while keeping track of their each upstream	11:32
	and*	11:32
gebi	then make them git submodules	11:32
	← ProperNoun left	11:33
	→ Octalot joined	11:33
	→ aziz joined	11:33
	→ ben_h joined	11:34
fdv	hgb: or, if possible, you could rebase the master branch without checking it out? but I guess that's not really possible	11:35
hgb	fdv: That would be ideal, but I don't think it's possible.	11:36
sitaram	wereHamster: just saw your comment... hehe :)	11:36
fdv	does anybody know it that's possible (to rebase a non-checked-out branch)?	11:37
wereHamster	nope, not really possible	11:38
hgb	wereHamster: Do you know about my proposed scenario (above)? Can I make a working copy of a certain branch from a repo, rebase in it, and then next time I checkout in the other wc (which is located with the repo), the branch is uptodate?	11:39
wereHamster	don't know, sorry	11:41
	jmesnil_away → jmesnil	11:41
sitaram	Amenthes: did you run the command I gave you?	11:42
Amenthes	sitara, yes. No ouput	11:44
	I meant output	11:44
sitaram	did you remember to substitute some 4 consecutive characters of the 40-digit filename in place of my d74c?	11:45
	if so, I would (a) take a backup of the whole thing and then (b) remove that file and try "git fsck --full"	11:46
Amenthes	sitaram, :) Of course. I used the hash exactly as in the git cat-file command from above.	11:46
sitaram	:) no offense -- we all get tired sometimes that's all	11:46
Amenthes	sitaram, it's ok :)	11:47
	sitaram, is there a (b) plan?	11:47
	→ cannonball joined	11:47
bentob0x	question: repository1 => merged from repository2:branch_1, then repository2 => merge self:branch_1 into self:master, then delete branch_1 => can I safely pull repository2:master into repository1?	11:47
Amenthes	sorry, now I was tired :P	11:47
	→ tal67 joined	11:47
sitaram	the (b) plan would be similar. Backup the rest of the repo, just move this file out of the way (some other name, in the higher level directory -- a "move" on the same "drive" should be instantaneous) and then try the fsck	11:48
Amenthes	sitaram, thanks. I'll try these and tell you what happened.	11:50
	← alester left	11:52
sitaram	the other thing you can do, is something like this: git show large-40digit-hash \| head > junk; then take the junk file to a Unix machine and run "file" on it to see what it is (or just open it in wordpad or something). Might give you an idea what it is...	11:52
bremner	Is there an easy way to configure git web to show "clone this" url's for http:// and git:// for each project	11:53
gebi	bremner: for git:// it's @git_base_url_list = "git://foo.org" in gitweb.conf	11:55
	→ MisterN joined	11:55
	→ d0k joined	11:57
	→ qrush joined	11:58
bremner	hmm. ancient gitweb here, I wonder if that is a problem	12:00
	← esparkman left	12:00
	→ qrush_ joined	12:01
	→ la_mer joined	12:01
bremner	added git_base_url_list=("git://pivot.cs.unb.ca/git"); but no url shows up. I don't need to restart anything, right, its just CGI?	12:01
	← qrush left	12:02
	bremner decides to upgrade gitweb, just for the hell of it	12:03
	→ fluxusps joined	12:04
sitaram	bremner: the README says something about cloen urls; sorry I'm not in a position to test etc or say anything more useful	12:05
	→ Pe3k joined	12:06
	← fluxusps left	12:07
	→ bcardarella joined	12:07
Pe3k	how to show single file history dialog window in gitextensions? dialog is described in User Manual, but can't find where to start it...	12:09
	← ben_h left	12:09
	→ ben_h joined	12:09
	← hummingkid left	12:10
	→ ben_h_ joined	12:10
	← kalle_ left	12:11
Grum	bremner: did you put that exactly on that line? because you need to write perl, so at least it is missing a '@' :)	12:11
	→ kalle_ joined	12:11
	→ gitte joined	12:11
bremner	ahh, it shows up now. Not sure if I was blind before or the upgrade fixed it.	12:11
	← solar_ant left	12:12
Grum	blind i bet =D	12:12
bremner	Grum: thanks for the vote of confidence :-)	12:13
gitte	Grum: just because you have one eye. Hah!	12:13
Grum	i had three last time i checked	12:13
	← cemerick left	12:14
gitte	Grum: oh, no! Give me back my two eyes!	12:15
	Grum stabs wildly	12:15
Grum	no!	12:15
	← alanhaggai left	12:16
	→ solar_ant joined	12:16
	← tim_jenssen left	12:21
	← solar_ant left	12:22
	→ mib_454get joined	12:25
	→ hulo joined	12:26
	→ leanucci joined	12:26
hulo	holu. i'm fairly new at revision control. I basically want to clone a git repo of the app i like (ncmpcpp), and be able work on it without sending anything upstream	12:26
	← Amenthes left	12:27
hulo	upon the next upstream sync, i want to be able to patch my work dir appropriately with the changes i made	12:27
	again, not sending anything upstream	12:27
	is this possible?	12:27
j416	hulo: git pull	12:28
	→ aspotashev\|eeepc joined	12:28
hulo	so initially, i'd git clone the upstream	12:28
j416	hulo: git is distributed. Nothing will be sent upstream unless you say so.	12:28
hulo	what's the next step to setup my own verson	12:28
	← Pe3k left	12:28
j416	hulo: the clone is your version.	12:29
	you can branch if you like.	12:29
	← ben_h left	12:29
	gitte mumbles something about rebase possibly being a better option than pull...	12:29
j416	gitte: of course..	12:30
hulo	all suggestions welcome :)	12:30
j416	i was thinking of pulling in master then rebasing on that	12:30
	(in a separate branch)	12:30
doener	if master is just a "copy" of origin/master, you don't need master	12:30
j416	I mostly work locally so I have no good workflow when it comes to clones of remote repos.	12:31
gitte	j416: that's my gripe with it: why pull if you rebase anyway? (pull = fetch + merge)	12:31
	→ ben_h joined	12:31
	← ben_h_ left	12:32
j416	gitte: ah, rebase onto origin/master ?	12:32
	makes sense, thanks	12:32
doener	maybe we can agree on "git pull --rebase"? ;-)	12:32
hulo	so after the initial clone, i can work on it as i please, then simply $git rebase?	12:32
	→ ben_h_ joined	12:32
j416	:)	12:32
hulo	I found http://jbowes.wordpress.com/2007/01/26/git-rebase-keeping-your-branches-current	12:32
	it helped me understand rebase somewhat	12:32
j416	hulo: you should read the git book	12:32
	→ aspotashev__ joined	12:35
j416	it should answer most or all of your questions	12:35
gitte	doener: ah, thanks for reminding me... I always wanted to write a patch to support git pull --rebase=i	12:35
	← fujin left	12:35
hulo	j416: i'm a dive-head-first kind of guy :)	12:35
j416	hulo: then dive right into the book :)	12:35
hulo	and i think for such a small experiment, all i'd need would be snippets from the book	12:35
	not planning on overseeing corporate projects any time soon :D	12:35
j416	you should get familiar with the basic concepts of git	12:35
	→ \|chalky\| joined	12:35
	← aspotashev\|eeepc left	12:36
	→ jackdempsey joined	12:36
	→ lorandi joined	12:36
	→ ajonat joined	12:36
	← chalky left	12:37
	→ mfilipe joined	12:38
	→ yawn_ joined	12:38
	← aspotashev left	12:38
gitte	j416: it cannot hurt to get a basic understanding what a merge is (it is a commit), what a fast-forward and a direct ancestor is, and how the commit graph is represented.	12:39
	Ah, sorry.	12:39
	hulo: this was for you.	12:39
j416	hehe	12:39
hulo	gitte: he doesn't mind the reinforcement :D	12:39
gitte	hulo: because you will not really understand the difference between a rebase and a merge otherwise.	12:39
	hulo: how do you know?	12:39
	gitte is always open to learn some mind-reader tricks	12:39
hulo	you just missed your first lesson	12:40
	→ aspotashev joined	12:40
	← aspotashev_ left	12:40
hulo	2nd	12:40
	→ gwoo_ joined	12:40
hulo	;)	12:40
	→ brtzsnr joined	12:41
	→ deavid joined	12:41
brtzsnr	hi. how do I revert a locally modified file?	12:41
j416	brtzsnr: check it out from your repo	12:42
brtzsnr	never mind.. git checkout -- file	12:42
j416	yep	12:42
gitte	brtzsnr: if you already staged changes, you might want to insert a "HEAD" before the "--".	12:42
	→ cmarcelo joined	12:42
doener	brtzsnr: note that that gets the file out of the index. So e.g. a file deleted by "git rm" won't be restored, you need to use HEAD then (as gitte said)	12:43
	brtzsnr: OTOH, without HEAD is the right way if you want the version back that you already "git add"ed	12:43
	← qrush_ left	12:43
	→ tim_jenssen joined	12:45
	← j416 left	12:47
	← hobodave left	12:47
	→ j416 joined	12:48
	→ berndj joined	12:48
	← faithfulgeek left	12:49
brtzsnr	10x for the info... none of my modifications were staged so... first version was enough	12:49
	← [intra]lanman left	12:50
	← ben_h left	12:50
	→ parasti joined	12:52
	→ qrush joined	12:52
	→ alester joined	12:53
	← alester left	12:53
	la_mer → cemerick	12:54
	← yawn left	12:55
	→ km2 joined	12:55
	← knitt1 left	12:55
	→ RaySl joined	12:55
	→ hummingkid joined	12:56
	→ trmanco joined	12:57
	→ gwoo__ joined	12:57
	← km2 left	12:58
	→ jaeckel joined	12:59
	← gwoo left	12:59
	→ hcl2 joined	13:02
hcl2	i have this file "config.git-hash", what does it do? do I need to save it?	13:02
gitte	hcl2: never heard of that file name.	13:02
	← priidu left	13:03
hcl2	oh... nm, my bad	13:04
	git-ref-list HEAD \| sort > config.git-hash <-- from a mailing list...trying to duplicate svn rev numbers	13:05
	:/	13:05
	← h0nk1 left	13:05
Ilari	hcl2: ref-list?	13:05
hcl2	rev	13:05
	maybe	13:05
	i still can't get over how easy it is to organize/use branches in git vs svn	13:06
	→ catalis joined	13:07
	→ ThaDon joined	13:07
	← stuffcorpse left	13:08
	→ stuffcorpse joined	13:08
doener	I'm constantly amazed how the same parts/concepts of git manage to get "wow this is great/easy/simple/intuitive" and "OMFG this is hard/confusing/bullsh*t" opinions at the same time from different people	13:08
hcl2	and i've been using it for about 3 months	13:08
	→ kukks joined	13:09
hcl2	i just started doing the git-svn to hook into old svn projects because I have too many "idea" changes lying around un-committed	13:09
	→ ajonat_ joined	13:09
hcl2	so, basically i make a new co of trunk on the same machine	13:09
doener	it happened to me more than just once that someone complained how hard X is, and while I've been still thinking about how to explain that better, someone else came and told how great and easy X is...	13:09
	← ajonat left	13:10
hcl2	it's pretty hard to get used to, the syntax still doesn't make sense, but it still makes stuff easier	13:10
	if you just forget about trying to grok the syntax	13:10
doener	which syntax?	13:11
	→ khmarbaise_ joined	13:11
	→ esparkman joined	13:11
hcl2	hypothetical syntax	13:11
	→ Sho_ joined	13:11
	← ntoll left	13:11
hcl2	of git commands which don't use normal words, like git checkout somebranch vs git checkout dir/somefile.txt	13:11
	← khmarbaise left	13:11
hcl2	luckily i don't have any dirs called "master"	13:11
doener	you'd use "--" to disambiguate the argument then	13:12
	git checkout -- master ==> get the file "master" out of the index	13:12
hcl2	^ that syntax	13:12
doener	git chechout master -- ==> checkout the branch head master	13:12
j416	doener: maybe those who are very used to another system have a hard time re-learning, and people (as myself) who never really got to grips with any other system find it easier, because they don't really have any other references	13:12
hcl2	well, i used both cvs and svn at work everyday for a number of years each	13:13
j416	hcl2: then maybe you're just a quick learner hehe	13:13
hcl2	it's not that I don't "get" the syntax, it's that I can't put all that effort into memorizing esoteric thigns that I never need to do	13:13
	← gwoo_ left	13:13
	← aspotashev left	13:14
doener	hcl2: that "--" separator is quite a common thing... e.g. "grep -- -- <somefile>" to have grep searching for "--"	13:14
	→ mndoci joined	13:14
hcl2	but, i'm not doing squashing, or rewriting (although I did fix a spelling mistake in a commit message yesterday)	13:14
	i woudl just do grep "\-\-" or whatever	13:14
	→ Pieter joined	13:14
hcl2	only time I use -- is when passing :1 or :0 to xinit	13:15
j416	hcl2: I have found it easiest to learn 'from the inside out', starting by learning how git works, to understand what _can_ be done, and then learn the command that actually does that. Instead of just learning a command and then what it does.	13:16
	this way I can be sure there is a way to do something, I just have to find the syntax	13:17
hcl2	j416: it took a while, but there was an "ahah" moment where i did learn the inside from the out (at least i like to think so)	13:17
	→ krh joined	13:17
hcl2	there's no way to describe branches though, to someone who has used cvs/svn	13:17
	→ halfline joined	13:17
hcl2	the only thing to say is "you will actually use branches"	13:18
j416	haha	13:18
hcl2	if you already do, you will use them the actual way you wanted to	13:18
doener	hcl2: heh, interesting that you mention that, gitte raised that topic on the mailing list a few days ago. Maybe you have some input?	13:18
hcl2	explaining?	13:18
doener	hcl2: yeah, what's hard about that? Any idea how the docs could be improved or so?	13:19
	← bcardarella left	13:19
hcl2	well, i didn't really read the docs, i watched the linus video, got it, played, googled	13:20
doener	ok	13:20
hcl2	it depends on what you're coming from... the mindset change cannot be underestimated though	13:20
	coming from cvs/svn	13:20
	there's just no good way to write it out other than, it's 100% different, but 99% the same	13:21
	← Grahack left	13:21
doener	I came from svn (I used cvs, but never really much more than update/commit), but git just happened to work how I expected it to work	13:21
	→ djpig joined	13:22
doener	and branching is totally natural, you clone, you commit ==> you have a new branch, automatically, by the virtues of having a DVCS	13:22
	→ jasmin joined	13:22
	← ben_h_ left	13:23
	→ ben_h joined	13:23
doener	hm, I think "virtues" is not the word I meant to use there...	13:23
	→ ben_h_ joined	13:23
	→ coopsh joined	13:23
osfameron	"by virtue of" is fine I think	13:23
	← ben_h_ left	13:24
j416	looks fine	13:24
	← mbuf left	13:24
j416	http://81.233.1.70/1h/virtue.txt	13:25
	erm.	13:25
doener	j416: ah, thanks, so s/the virtues/virtue/ and I'm fine :-)	13:26
	← aspotashev__ left	13:26
	doener has a bad habit of using phrases he picked up somewhere without knowing their exact meaning (and realizing that only after using them)	13:26
j416	hehe	13:27
	I think that is a good thing in a way. It helps you learn.	13:27
	I do that too sometimes.	13:27
	→ aspotashev joined	13:27
	→ aspotashev\|eeepc joined	13:27
	→ petercoulton joined	13:27
osfameron	my mum used to say "miffed" to mean confused (it means "annoyed" in UK slang)	13:28
j416	crazy	13:28
	it would certainly miff me if you used it	13:29
doener	j416: yeah, it's not bad per se, but I should change the order from "hear, use, lookup in dictionary" to "hear, lookup, use" ;-)	13:29
j416	hehe	13:29
	yeah	13:29
	"hear, use, get corrected" might be faster though :)	13:29
drizzd	doener: That's the most important improvement of DVCS IMO	13:29
doener	j416: then I'd be in line with folks that need to get told to RTFM though ;-)	13:30
j416	haha, true	13:30
drizzd	You can make branches with most conventional CVSs, but you have to _think_ about it.	13:30
	→ alanhaggai joined	13:30
	← simmelpp left	13:30
drizzd	*VCSs	13:30
doener	drizzd: CVSs is fine, it's Crippled Versioning Systems ;-)	13:31
	← ben_h left	13:31
drizzd	hehe, I'll remember that	13:31
	→ Grahack joined	13:31
ciaranm	all you people abusing cvs clearly never had to use any of its predecessors	13:32
Zathrus	what's wrong with branching in CVS? Now pardon me while I spend the rest of the day merging our latest branch back to HEAD.	13:32
	→ PeakerWork joined	13:32
Zathrus	ciaranm: I did. And still do, to a limited extent.	13:32
Rhonda	ciaranm: I love RCS big time, and still use it on regular basis.	13:32
	← jackdempsey left	13:33
Zathrus	since the previous team didn't move from RCS until 2006.	13:33
PeakerWork	does anyone know a C comparison thing that knows to ignore meaningless changes? "git diff -w" helps, but still plenty of changes that don't affect anything are shown	13:33
	→ Bass10 joined	13:34
drizzd	PeakerWork: you could diff the generated assembler code	13:34
dedmajor	guys, could you help me a bit? i have a clone of remote git repo with two local checkouts - branch master and branch v1.0 (for example). let say i'm in branch master now. i make changes and commit some code. now, what is the optimal way to cherry-pick this commit to branch v1.0? i do checkout v1.0, then git pull (to avoid non-fast forward push), then git cherry-pick, then git push (to push both commits), then git checkout master, and continue	13:35
	developement. So, i thinks it's to obscure and i need some help to figure out what i'm doing wrong )	13:35
	→ scarabx joined	13:35
drizzd	If you just want to ignore all whitespace changes (including new lines and removed line breaks), I suppose you could simply compare an auto-formatted version.	13:35
	← plasticine left	13:35
PeakerWork	drizzd: that's even more difficult :)	13:35
	drizzd: Nicer to compare a canonized AST or such, perhaps even one that canonizes name	13:35
ciaranm	dedmajor: there was a cherry-pick --onto proposal at one point, but the powers that be didn't like it	13:36
hcl2	doener: I used git/svn with only commit update and nothing fancy only because I realized that branching/merging was just not worth the problems. I never did anything fancy	13:36
drizzd	well, the names part only works for static variables	13:36
	or local variables	13:36
hcl2	doener: and git did not work anything like I expected, in fact Ithought I did some commands wrong because they were too quick	13:36
	→ swombat joined	13:37
swombat	hello...	13:37
hcl2	like git co -b newbranch .... too quick	13:37
	did not think it worked	13:37
swombat	"Your branch is ahead of 'origin/master' by 1 commit." << I want to get rid of that commit (I made it on a clone which doesn't have remote access)	13:37
	how do i do that?	13:37
dedmajor	ciaranm: so, anyone is cherrypicking commits now in the same way as me?	13:37
cannonball	hcl2: You should add some sleep commands so that it feels more like other products...	13:37
ciaranm	swombat: git reset --hard	13:37
	→ loincloth joined	13:37
hcl2	cannonball: yeah, that's what I do with some of my php code ;)	13:37
swombat	ciaranm: still the same	13:37
dedmajor	ciaranm: err/everyone	13:37
ciaranm	dedmajor: i do more or less what you said, except i don't pull	13:38
	snuxoll: what command did you run exactly?	13:38
	← Fullmoon left	13:38
doener	dedmajor: the cherry-pick vs. pull order doesn't matter much, it might in some cases allow a fast-forward _locally_ instead of a merge, but for the push part, cherry-picking first and then merging is just fine	13:38
j416	hcl2: the only thing that happens when you create a new branch is really git writing a SHA1 to file, so it should be quick :)	13:38
hcl2	alias "git co -b"="sleep 2; git co -b"	13:38
cannonball	hcl2: We're still using git 1.5.x. I can't wait to get it up to 1.6.x, it has some tunings that is supposed to make it much faster over nfs which is what our central home directories are.	13:38
swombat	ciaranm: i've already committed the stuff, I but couldn't push it (since it's on a read-only clone), and so now i want to uncommit it	13:38
ciaranm	swombat: uh huh. git reset --hard.	13:39
doener	dedmajor: what you should consider is doing things the other way around. Put bugfixes into 1.0 first, and then merge that to master (I assume that 1.0 is the maintenance branch, and master is for the next version)	13:39
hcl2	but... when you first playing with git, and you come from svn+https over at SF... you honestly think something might be wrong	13:39
j416	hehe	13:39
swombat	ciaranm: http://pastie.org/private/bb5kklhp5jkucu01i3chmq	13:39
	→ Fullmoon joined	13:39
doener	dedmajor: that way, you get all accumulated bugfixes from 1.0 to master in one swoop, and don't have "duplicated" commits, like you do with cherry-picking	13:39
	wshimmy1 → wshimmy	13:39
hcl2	you think that you screwed up the command..... or that it's just getting ready to do stuff and that it will take 4 seconds later	13:39
ciaranm	swombat: you need to read man git-reset	13:39
	swombat: you need to tell it what to reset to	13:40
swombat	ah	13:40
doener	dedmajor: also makes it a lot harder to forget some commits. The merge always gets all the stuff from 1.0 to master	13:40
	→ jschoolcraft joined	13:40
	→ markelikalderon joined	13:40
hcl2	also, the fact that all branches can get access to the history of all other branches is a little wierd	13:40
swombat	ciaranm: great, that worked	13:40
doener	dedmajor: while the cherry-picking means that you need to remember to pick each interesting commit	13:40
swombat	(specifying which commit)	13:40
dedmajor	doener: i thought master is a developement unstable branch, which should be tested before changes goes into stable branches (like 1.0)	13:40
swombat	thanks for the tip!	13:40
hcl2	that is not like svn where branches are just copies	13:40
	i still don't understand git tags, haven't used them yet	13:41
	→ gberdyshev joined	13:42
	→ ckm joined	13:42
doener	dedmajor: that's your choice. For example git.git has 4 branches. "maint" which gets bugfixes and is merged to "master". "master" in turn is meant to be pretty much stable and becomes the next release. And "next" and "pu" are the "unstable" branches	13:42
j416	hcl2: a tag is just something that points to a commit, often as its own object with a message etc.	13:42
	used for giving things a name	13:42
	← gberdyshev left	13:43
doener	dedmajor: of course things that go to "maint" should be fine and tested, before you let them get out, but that doesn't mean that you must cherry-pick them from master	13:43
hcl2	i just haven't figured out the syntax for doing them. again, i know what they are in svn, they are copies	13:43
dedmajor	doener: oh, thanx ) i'm very appreciating your explanations )	13:43
	→ gberdyshev joined	13:43
j416	in git they are just references	13:43
	or, contain references that is	13:43
doener	dedmajor: that could even be "bad" in some way. "master" might have changed even to make the bugfix work on it, but the cherry-picked version in 1.0/maint fails in a subtle way	13:43
	s/even/enough/	13:43
	→ faithfulgeek joined	13:44
doener	dedmajor: but going the other way around, that would mean that it works on 1.0 (for which it was written) and then subtly fails on master (which isn't _that_ bad, as master isn't as "critical")	13:44
dedmajor	doener: i'm still very new to git, so i need a bit of time to perceive all that you are saying now ) wait 0	13:45
	→ Ryback_ joined	13:45
	→ alanhaggai_ joined	13:47
	← Ryback_ left	13:47
doener	j416: tags don't have to reference commits, they can reference any object. For example there's the junio-gpg-pub tag in git.git that tags a blob containing Junio's GPG key	13:47
	→ Ryback_ joined	13:47
j416	doener: wow, I didn't know that.	13:47
	(I did think though: doener will say something about what I said)	13:48
	:P	13:48
	→ priidu joined	13:49
doener	hm, is that good or bad? :-\|	13:49
hcl2	haha	13:50
	← petercoulton left	13:51
sitaram	the only bad thing is for me: I can't even read as fast as doener can type :(	13:52
	→ petercoulton joined	13:52
	→ giallu joined	13:52
	sitaram thanks God for backlogs	13:52
	← gilimanjaro left	13:53
	→ spearce joined	13:54
	→ loinclot_ joined	13:55
	→ [intra]lanman joined	13:55
	← loincloth left	13:56
dedmajor	doener: well, the way you have explained seems to be very smart, you helped me very much )	13:57
	← alanhaggai left	13:57
	alanhaggai_ → alanhaggai	13:58
	← Grahack left	13:58
doener	dedmajor: you're welcome	13:59
	← markelikalderon left	14:00
	→ bcardarella joined	14:00
	→ hpoydar joined	14:01
	→ gnublade joined	14:01
j416	doener: always a good thing :) Keep on correcting hehe	14:02
	→ petercoulton_ joined	14:02
	← petercoulton left	14:03
	← petercoulton_ left	14:03
	→ spearce` joined	14:03
drizzd	j416: doener is like a safety net. I don't even think about what I'm saying any more.	14:03
j416	drizzd: :D	14:03
	→ boto joined	14:03
	→ petercoulton joined	14:04
	← dwt left	14:07
	→ bdiego joined	14:07
	→ solar_ant joined	14:07
	→ bcardarella_ joined	14:07
	→ markelikalderon joined	14:08
	→ kraymer joined	14:08
	← solar_ant left	14:08
	→ alester joined	14:09
	→ km2 joined	14:10
	→ jackdempsey joined	14:11
	→ ToxicFrog joined	14:12
	→ bcardarella__ joined	14:12
	← j416 left	14:13
	→ DavidKlein joined	14:13
	← bcardarella left	14:14
	→ andres joined	14:14
	← yawn_ left	14:16
	→ troyt joined	14:17
	← bcardarella__ left	14:17
	← spearce left	14:18
	← priidu left	14:18
	← mndoci left	14:20
	← gnublade left	14:20
	← scarabx left	14:21
	← aspotashev left	14:23
	→ eletuchy_ joined	14:24
	→ bryanl joined	14:24
	← ia left	14:24
	→ ia joined	14:25
	→ petercoulton_ joined	14:26
	← petercoulton left	14:26
	← petercoulton_ left	14:26
	→ petercoulton joined	14:26
	→ jchris joined	14:26
	← ahf left	14:27
	← Pewpewarrows left	14:27
	→ bcardarella joined	14:27
	← girishr left	14:28
	← bcardarella_ left	14:30
	← markelikalderon left	14:30
	→ ahf joined	14:33
	← mib_454get left	14:33
	→ bewst joined	14:37
	→ bdimcheff joined	14:37
bewst	can someone explain to me what remotes/origin/master and remotes/public/master mean? In particular, what's the significance of that middle element (origin or public)?	14:38
	← swombat left	14:38
Grum	it is the name you gave for the remote	14:38
hcl2	the name of the remote location, i think	14:38
Grum	the default name is 'origin'	14:38
doener	bewst: if you use "git remote add foo git://host/repo" that sets a default fetch refspec for that remote	14:38
	bewst: in that case, the config would have remote.foo.fetch set to "refs/heads/:refs/remotes/foo/"	14:39
	bewst: so "git fetch foo" becomes equal to "git fetch foo refs/heads/:refs/remotes/foo/"	14:39
hcl2	^ all that stuff is still confusing to me	14:39
doener	bewst: and that fetches all branch heads from the "foo" remote, and stores them in the refs/remotes/foo/ namespace	14:39
	→ Zenopus_ joined	14:39
bewst	doener: ...yeah... but... how does public/master and origin/master line up with the examples you just gave?	14:40
doener	bewst: so if it fetches "refs/heads/master" (the "master" branch head) it's stored in refs/remotes/foo/master	14:40
	bewst: those are shortnames	14:40
bewst	I am seeing both public and origin in this repo, and I only ever cloned or pulled from one place.	14:40
	doener: shortnames for what?	14:41
doener	bewst: when you use e.g. "master", git tries things like refs/heads/master, refs/remotes/master etc.	14:41
bewst	doener: according to your examples, they would be servers.	14:41
	→ broonie joined	14:41
	→ priidu joined	14:41
bewst	doener: yes, I know what master is.	14:41
	doener: I'm asking about the element before master.	14:41
	← jschoolcraft left	14:41
doener	bewst: that's just one element in the refname, and it's just a default, chosen so that it's the same as the name of the remote	14:42
	bewst: there's no "implicit" meaning to it, it's all just defaults	14:42
	← bewst left	14:43
Grum	doener: if you are merging multiple branches what is the 'sane' thing to do? just merge from origin/xxxx or make a local branch for each and then merge?	14:43
	seems all of our merge commits will contain origin/ :)	14:43
	→ Pewpewarrows joined	14:43
	→ bewst joined	14:43
	→ dodo_the_last joined	14:43
bewst	doener: sorry, ERC just locked up emacs on me	14:44
	doener: so I missed anything you may have typed recently	14:44
Grum	o.O use a real ircclient =D	14:44
doener	bewst: what's the last thing you saw/remember?	14:44
Grum	he just missed:	14:44
bewst	doener: you were telling me what "master" means or so I thought	14:44
Grum	16:42:20 < doener> bewst: that's just one element in the refname, and it's just a default, chosen so that it's the same as the name of the remote	14:44
	16:42:35 < doener> bewst: there's no "implicit" meaning to it, it's all just defaults	14:45
	→ justatheory joined	14:45
bewst	doener: and I was saying that I don't know where those names came from... thx, Grum	14:45
	← gwoo__ left	14:45
doener	bewst: no, I meant to explain that "master" is just a shortname, the full refname is refs/heads/master (if were talking about a local branch head)	14:45
bewst	doener: OK, let me try my question this way	14:45
doener	bewst: and similarly, that "origin/master" is a shortname, usually for refs/remotes/origin/master	14:45
	and those names are generated by the means of that config setting from "git remote add"	14:46
Grum	but you can have a branch called origin/master as well -- in which case it would refer to: refs/heads/origin/master	14:46
doener	so it all boils down to the fact that "git remote add" sets a default refspec that puts the fetched branch heads into the refs/remotes/<remote-name>/ namespace	14:46
bewst	doener, Grum: I'm looking at a gitk window and trying to find out what these things mean that I'm seeing there. Is there a place I could post a screenshot?	14:47
Ilari	bewst: IIRC, the expansion rules are (not in order of percedence, X is the entered shorthand: X, refs/X, refs/heads/X, refs/tags/X, refs/remotes/X, refs/remotes/X/HEAD .	14:47
doener	so the "origin" and "public" parts "normally" mean that the stuff comes from the "origin" or "public" remote respectively	14:47
	"normally" because it's all just defaults, you can do "git fetch origin refs/heads/foo:refs/remotes/public/foo" and "inject" origin's foo into the refs/remotes/public namespace	14:48
bewst	doener: I'm trying to tell you that there's only one repo that could possibly ever have been used as a remote	14:48
	doener: so I don't understand where these distinctions are coming from	14:48
	→ schacon joined	14:49
Grum	bewst: you added the 'public' remote manually	14:49
	→ hdl2 joined	14:49
Grum	and if you didn't the you copied the complete git directory from did	14:49
	err +person	14:49
	→ wmoxam joined	14:49
bewst	Grum: from did err +person?	14:49
doener	bewst: in gitk, do these names read as just "origin/master", "public/master" or "remotes/origin/master" and "remotes/public/master"?	14:50
bewst	The latter	14:50
doener	ok	14:50
	bewst: could you pastebin your config?	14:50
bewst	doener: and they point to different commits.	14:50
	loinclot_ → loincloth	14:50
bewst	doener: sure...	14:50
	loincloth → cloth_of_loin_va	14:50
	hdl2 → hdl	14:50
bewst	doener: do you mean my .git/config ?	14:50
doener	yep	14:50
	cloth_of_loin_va → cloth_of_loins	14:51
	← Sho_ left	14:51
	→ antoranz joined	14:51
	← ahf left	14:51
Grum	doener: that is default behaviour afaik, the [remotes/origin] is in 'light orange' and the [master] part is green	14:52
	→ juanjoc joined	14:52
	→ ahf joined	14:52
bewst	doener: OK, I see it in http://gist.github.com/100525	14:52
Grum	i'm trying to understand where the confusion comes from	14:52
doener	bewst: yeah, two remotes, but they reference the same repo	14:52
bewst	doener: yeah, but different commits, somehow?	14:53
Grum	ofc!	14:53
doener	bewst: "remotes" are a configurable entity, they're _not_ the same as the repo	14:53
	bewst: probably because one of them didn't get fetched from recently	14:53
Grum	if you don't update one where you do update the other	14:53
doener	bewst: if you do "git fetch public" that will update the remote tracking branches for "public", i.e. refs/remotes/public/*	14:53
	bewst: but "git fetch origin" will update the refs/remotes/origin/* stuff	14:54
bewst	doener: I think I get it. Hang on...	14:54
doener	bewst: if you fetch only one of them, the other will "lag behind"	14:54
	→ Sho_ joined	14:54
bewst	doener: thanks, now all is clear	14:55
	doener: just one other question. How do I delete one of these remotes? as they both point to the same repo it just adds confusion	14:55
doener	git remote rm <name>	14:55
Ilari	bewst: And btw, why there are two remotes, one for git:// and one for ssh://? Those should be in sync.	14:55
doener	though your config is a bit interesting	14:55
	← Sho_ left	14:56
bewst	Ilari: oh, I think I can explain that... brb	14:56
doener	bewst: "master" is setup with remote "public", but that's apparently the one that "git clone" created	14:56
	→ Sho_ joined	14:56
	← DavidKlein left	14:56
	bobmcw_ → bobmcw	14:57
	← schacon left	14:57
	← Sho_ left	14:57
doener	dropping the "public" remote seems like it's less "work" (as you don't need to adjust the config for "master")	14:57
	→ Sho_ joined	14:57
	tedoc2000_afk → tedoc2000	14:57
alester	Ilari: Why is it back to have one git: and one ssh: ?	14:58
bewst	Ilari, doener: what happened was that on /this/ machine, I mistakenly cloned from github's "public" (I think meaning read-only) URL and then later switched to the ssh one so I could commit	14:58
doener	but I wonder why "public" is there in the first place... submodule stuff? Or just "misuse" or "clone -o <name>"	14:58
	← Zenopus left	14:58
doener	alester: it's pointless (if it's the same repo), and leads to confusion, as we've just seen, cause things get out of sync	14:58
	Zenopus_ → Zenopus	14:58
bewst	doener: I'm using submodules, but I don't think it has anything to do with that	14:58
	→ knitt1 joined	14:58
	→ ferdy_ joined	14:59
doener	bewst: me neither, but I'm totally clueless about submodules, so I'm just throwing around my thoughts ;-)	14:59
jast	if it's the same repo, why not just remove one remote and prune away its remote tracking branches?	14:59
bewst	doener: I think it's just because I muddled my way through pushing to the ssh:// URI without really knowing what I was doing.	14:59
jast	submodules don't touch refs/remotes	14:59
	→ antoranz_ joined	14:59
bewst	doener: wow, submodules are one thing I (sort of) understand.	15:00
	→ neoeinstein joined	15:00
doener	jast: I just recalled some old discussion on the mailing list where someone had a need for $something_related_to_remotes for submodules	15:00
bewst	jast: I planned to remove that remote. How do I prune away the remote tracking branches?	15:00
jast	bewst: look at git remote prune	15:00
doener	bewst: well, I get the "how to use them" part, but have no clue about the implementation	15:00
jast	the implementation is a bit weird to wrap your head around but it actually makes a lot of sense	15:01
	→ schacon joined	15:01
doener	jast: "remote rm" deletes the remote tracking branches IIRC	15:01
jast	the only thing git stores about the submodules in the supermodule is the commit ID in the gitlink	15:01
bewst	Do I want to prune then remove, vice versa, or doesn't it matter?	15:01
jast	oh, that's possible	15:01
	→ octagon joined	15:01
jast	just remove the remote and see if the extra tracking branches are gone	15:02
doener	yep, it does, just tried :-)	15:02
	→ altrux joined	15:02
	→ DavidKlein joined	15:02
bewst	thanks, all!	15:03
jast	so if you update a submodule "foo" at ./foo, new objects are fetched into ./foo/.git	15:03
	← schacon left	15:03
	→ schacon joined	15:04
jast	I guess prune would have been the wrong thing anyway, because it enumerates over remotes that still exists and removes remote tracking branches that are no longer needed in those	15:04
	← schacon left	15:04
jast	I'm not actually sure that it removes branches that have no corresponding remote	15:04
	← knitt1 left	15:04
bewst	is there any easy way to "repoint" all my submodules to a different URL? They are referencing the URL of the remote I want to remove.	15:05
jast	I suggest a search&replace in .gitmodules and .git/config	15:06
	→ straszheim joined	15:06
jast	well, actually you may want to leave them as they are in .gitmodules	15:06
bewst	yo, troy	15:06
	→ knittl joined	15:06
straszheim	hey hey	15:06
bewst	jast; OK, it doesn't matter that much	15:06
straszheim	who's that?	15:07
	→ ben_h joined	15:07
bewst	dave	15:07
straszheim	hey!	15:07
bewst	ahoy	15:08
straszheim	http://sodium.resophonic.com/git/boost/	15:08
	trying to get a git mirror up	15:08
bewst	nice! What's the problem?	15:09
straszheim	have a zillion tags/branches imported into the repo	15:09
	they dont' appear here	15:09
	trying to see if there is a way for people to browse, say, git-svn cloned release branch	15:09
	→ aspotashev joined	15:10
straszheim	locally, git branch -a shows them, and i can switch to them locally	15:10
jast	git-svn branches are typically kept in the remotes/ namespace	15:10
bewst	Looks like they may have become non-public remotes, but you should get help from someone who knows what he's talking about ;-)	15:10
jast	typically, clone only clones refs/heads and refs/tags	15:10
	so you'd have to change the refspecs used by fetch	15:10
	← kalle_ left	15:11
	→ kuadrosx joined	15:11
	→ Modius joined	15:11
	→ kalle_ joined	15:12
	← mfilipe left	15:12
jast	ah, actually I don't even think refs/tags is cloned	15:12
straszheim	i appear to have got everything... this ran for two days ;)	15:12
bewst	straszheim: we should start using mirrors and personal git repos for development. Eventually Boost will just fall over and switch :-)	15:13
	← antoranz left	15:13
straszheim	bewst: that's the diea	15:13
	idea	15:13
bewst	either spelling works	15:13
straszheim	git svn clone http://svn.boost.org/svn/boost -Ttrunk -ttags/release -bbranches	15:13
	there was my clone	15:13
jast	straszheim: oh, so that repository you linked to is a direct git-svn clone?	15:13
doener	straszheim: I guess you see the missing stuff with "git branch -r" but not "git branch", right?	15:13
straszheim	jast: yeah, i git-svn cloned and then put it up under cgit	15:13
jast	okay	15:14
straszheim	doener: exactly	15:14
jast	most tools don't show remote branches	15:14
	so there's two things you can do	15:14
	either you move all the branches into the local namespace yourself	15:14
	or you create a clone of the original repo with a special fetch command	15:14
	uhm, in this case, by original I mean the git-svn repo you created	15:15
	straszheim is following	15:15
doener	there's a patch on the mailing list that would allow git-svn to operate on bare repos, that'd be helpful here	15:15
	← rittyan left	15:15
	← lottadot left	15:15
straszheim	doener: I just pulled the .git directory out of the nonbare git-svn clone	15:15
jast	so I guess I'd recommend keeping the git-svn repo elsewhere and mirroring the branches/tags into that public repository	15:16
straszheim	ah, but i cant' fecth	15:16
jast	how about push?	15:16
straszheim	eh, can't fetch to the .git/ that i just pulled out of the git-svn clone	15:16
jast	but can you push to it?	15:16
straszheim	with the full repo I can	15:16
	let's see	15:17
doener	straszheim: basically, "git push <remote> 'refs/remotes/:refs/heads/'" should do for most purposes	15:17
jast	oh, you may want to edit the config file of that .git dir you pulled out... make it say bare = true in the [core] section	15:17
doener	straszheim: as all the "pushable"/"fetchable" stuff from git-svn lives in the refs/remotes namespace	15:17
jast	otherwise most everything will fail	15:17
doener	jast: hm, unless the directory is still called .git, the autodetection _should_ work, I think	15:18
jast	unless it explicitly says bare = false	15:18
doener	true	15:18
jast	which it does in my clones	15:18
doener	oh, and that's the default it seems... d'oh	15:18
straszheim	ok, rename .git -> something, set bare=true, can svn fetch	15:19
	→ netoman joined	15:19
	→ aspotashev_ joined	15:19
	← aspotashev\|eeepc left	15:19
jast	svn fetch only works if the working tree exists	15:19
	unless you apply the git-svn patch doener talked about	15:19
	→ dotsintacks joined	15:19
doener	jast: hm, really? I don't see any "bare" checks in the current git-svn	15:19
dotsintacks	anyone ever run into "Not a git repository" for all your repos when setting up cgit?	15:19
straszheim	ah. hm. there was nothing to fetch, so no failure. was getting complaints from perl before.	15:19
jast	you were the one to mention it	15:19
dotsintacks	my repos are definitely readable by apache...	15:20
doener	jast: I just recall that it refuses to fetch to refs/heads/* at all	15:20
jast	oh.	15:20
doener	jast: and that	15:20
	's what the patch is about, I think	15:20
	oh, oops, stupid enter key :-)	15:20
	← k0001 left	15:20
jast	happens to me a lot lately, that stupid enter thing	15:20
straszheim	okay, so i can keep the svn clone private, with a cronjob, and mirror that to the cgit directory	15:21
jast	do you happen to know if git-svn triggers any core git hooks?	15:21
	because then we could even automate the remapping thing without needing two repos	15:21
doener	jast: not sure. But instead of the remapping, we could also just wrap the fetch + some symref setup	15:22
jast	I never used any symrefs (apart from HEAD)	15:23
	← sdiz left	15:23
	→ jds joined	15:23
	→ sdiz joined	15:23
jds	Heya	15:24
jast	doener: how would you "wrap the fetch"?	15:24
doener	jast: a shell script?	15:24
jds	If I'm reading "git help ignore" right, .gitignore containing /* should only ignore files in the current directory. It actually seems to ignore everything - why?	15:24
jast	oh, that kind of wrap	15:25
	→ gp5st joined	15:25
jast	jds: because that includes ignoring each direct subdir of the current dir :)	15:25
	and if git never looks at subdir foo, of course it won't look at foo/bar, either	15:25
gp5st	hello, I'm on a bsd machine and when I do `git init` i get "fatal: cannot tell cwd" It's version 1.6.1.3 of git; any ideas?	15:26
	→ gwoo joined	15:26
	→ lois[pac] joined	15:26
	cloth_of_loins → loincloth	15:27
gp5st	there was a thread at kernel trap, but i didn't get any answers from it	15:27
	the other difference from what I find in searches is that this is local, not a remote problem	15:27
jds	jast: Hmm. Any way of negating that ignore?	15:27
jast	gp5st: do you have a pwd util (sorry, I'm one of those linux fanboys who never use anything else) and does it work in that same dir?	15:27
doener	jds: !/*/ might do to unignore all directories	15:28
jast	jds: you can prefix patterns with ! to negate them. there was some gotcha with subdirs and negated catch-alls, though...	15:28
	I don't remember the specifics	15:29
gp5st	jast: yes, but pwd and getcwd() are slightly different	15:29
doener	jast: there's a catch with things like !foo/bar/goo when "foo" is ignored	15:29
jast	well, I don't know the implementation of your pwd	15:29
	doener: oh, yeah, that was it	15:29
gp5st	jast: it's the same on linux too, i believe	15:29
jast	gp5st: still, I don't have the source code of pwd handy	15:29
gp5st	jast: i'm reading the getcwd man page and it says it'll fail if any component of the path fails acccess it errors:(	15:29
	← diadema left	15:30
jast	that's one reason why it may fail	15:30
doener	strace says that _my_ pwd uses getcwd :-)	15:30
gp5st	i'm on a shared box, so i don't have acces to / or /home, :-\	15:30
	hmm	15:30
	← johan-s left	15:30
doener	I'm also on Linux though	15:30
jast	if you don't have access to / and /home you can't access /home/yourself in the first place	15:30
gp5st	well, this usually works on linux, i think it's the bsd ppl being strict	15:30
	← Pewpewarrows left	15:30
gp5st	you're right, i can't access /home	15:31
jast	hmm.	15:31
gp5st	but i do have acces to /home/me	15:31
jast	I think you need executable rights to /home at least	15:31
	let me see if I can duplicate this situation on linux	15:31
	→ HG` joined	15:31
	→ vienova joined	15:31
gp5st	thanks	15:31
	→ gotgenes joined	15:32
gp5st	the thing is pwd works, but i think pwd works even if you can't see the current dir's contents	15:33
jast	are you using a pwd shell builtin or the external program?	15:33
gp5st	...good question	15:34
	haha	15:34
jast	try using an explicit path to pwd	15:34
	should be /bin/pwd I guess	15:34
doener	even with a leading up directory having 0000 permissions, getcwd works here	15:34
gp5st	if i do /bin/pwd it gived me my home	15:34
	where i am	15:34
	:-(	15:34
	→ Pewpewarrows joined	15:34
gp5st	doener: you're on linux, right?	15:35
doener	gp5st: I guess there's something like strace available that could show the error returned by getcwd, right?	15:35
	→ ben_h_ joined	15:35
doener	gp5st: yep, just saying that this is not a problem on this platform	15:35
gp5st	yeah	15:35
	i have strace on the box	15:35
	i've never used it	15:35
jast	well, just do this:	15:35
doener	gp5st: strace -e getcwd -f git init	15:35
gp5st	invalide system call getcwd	15:36
	let me see if it's a bsd thing and it's called different	15:36
	i'm usually a linux person:-\	15:36
doener	ah, the syscall is linux specific, just the C lib function is portable	15:36
	gp5st: just drop the "-e getcwd" and get the full output, shouldn't be much harder to read	15:37
	→ aisbaa joined	15:37
	← bryanl left	15:38
gp5st	strace: open("/proc/...", ...): Permission denied	15:38
	trouble opening proc file	15:38
	guess i need root? but i don't have that	15:39
doener	hm, interesting, looks like they don't like you to use strace or so...	15:39
jast	I don't think you need access to /proc	15:39
emias	getcwd(3) is portable (across POSIX), but strace(1) is Linux-specific :-)	15:39
jast	oh.	15:39
gp5st	thanks for your help; i may just give them a call since they can figure out what's up on this specific machine; i've just never ran into this issue	15:39
	→ jmesnil_ joined	15:39
gp5st	haha, i wonder why i have strace on here then	15:39
jast	I wonder if it's worth making error handling more verbose in that spot	15:40
doener	anyway, found some BSD getcwd implementation and that seems to call opendir() and fstat() on each path component	15:40
jast	that would explain things...	15:40
gp5st	ooh	15:40
	← altrux left	15:41
emias	gp5st: On FreeBSD, you'd normally use truss(1). (IIRC, there's some port of strace, but that requires a mounted Linux-/proc ...)	15:41
gp5st	why? i know they are ?ber conscience of security, but would getcwd be an issue if i know where i am if i'm allowed to be there?	15:41
	and i guess I should have looked for the source of getcwd:-p	15:41
	emias: thanks	15:42
	← Modius left	15:42
gp5st	i'm new to bsd, so still getting my bearings on what's different	15:42
jast	question is if this can be solved in some way	15:42
gp5st	i'm more than willing to recomile it and use a local version (I'm using there's)	15:43
	← jmesnil left	15:43
gp5st	or maybe this has been fixed, the current is 1.6.2 and i have 1.6.1	15:43
	→ Modius joined	15:44
jast	we can easily check in current master...	15:44
	<master:builtin-init-db.c>	15:44
Gitbot	[git builtin-init-db.c]: http://tinyurl.com/cstooe [blob]	15:44
jast	no change in git	15:44
gp5st	so is there anyway around this?	15:45
	← bcardarella left	15:45
	← khmarbaise_ left	15:45
jast	you could patch git, I guess	15:45
	but we'll have to cheat to avoid using getcwd	15:46
	← \|chalky\| left	15:46
jast	doener: have you got a link to that getcwd implementation?	15:46
doener	http://lists.freebsd.org/pipermail/freebsd-bugs/2008-March/029672.html	15:46
	http://www.koders.com/c/fid62A3A6B5FEE22D0690D4CB56A912B777DCD0E014.aspx?s=mdef%3Amd5	15:46
	→ Yuuhi joined	15:46
gp5st	:( /bin/pwd works for this; i wonder why	15:47
jast	I'd love to be able to strace that pwd binary	15:47
doener	jast: hm, actually, that's just BSD licensed and the google result text confused me... d'oh	15:48
jast	oh.	15:48
	← aspotashev left	15:48
jast	ht://Dig. I'll just close that tab and pretend I never opened it.	15:48
	gp5st: do you know which bsd (and version) that is?	15:48
gp5st	jast: FreeBSD flure.pair.com 6.2-RELEASE-p12 FreeBSD 6.2-RELEASE-p12 #1: Wed Nov 26 15:06:45 EST 2008	15:49
	whoops, meant to get rid of the server name	15:49
	← hulo left	15:50
	→ sunblush joined	15:50
	→ hobodave joined	15:52
	← dwave left	15:52
	→ aspotashev joined	15:53
	← darrob left	15:53
jast	doener: the official getcwd in fbsd 6.2 does this too: http://svn.freebsd.org/viewvc/base/release/6.2.0/lib/libc/gen/getcwd.c?revision=165962&view=markup	15:55
	← ben_h left	15:55
gp5st	i've been trying to find which dir it was in	15:55
	thanks	15:55
jast	me too, took some time :)	15:56
	→ EvanCarroll joined	15:56
EvanCarroll	is there anyway to checkout foo as bar	15:56
jast	it actually uses some __getcwd extern but I don't know where that comes from	15:56
	← stuffcorpse left	15:57
jast	EvanCarroll: git show <commit>:foo >bar	15:57
	→ stuffcorpse joined	15:57
doener	jast: ah, and I just found the note in my getcwd(3) man page, that that EACCES can only happen on Linux with an old kernel that doesn't have the getcwd syscall, nor is /proc/self/cwd available	15:57
	→ cgardner joined	15:57
	← giallu left	15:57
doener	gp5st: at this point, I'd ask on the mailing list, but there are quite a few getcwd() calls, so I'd not expect a "oh, we can change that easily" response	15:59
gp5st	i wouldn't expect that:-p	15:59
	http://it.toolbox.com/blogs/bsd-guru/using-git-on-freebsd-13680	15:59
	i'm gonna see what one of the ports does, and then see if i can compile it	16:00
jast	that post is very old and also doesn't cover your issue, I think	16:00
gp5st	yeah, i was reading more through it	16:01
	← martins left	16:02
	← stuffcorpse left	16:02
	→ stuffcorpse joined	16:03
	← bentob0x left	16:03
	→ gafmtt joined	16:03
emias	gp5st: But you said that /bin/pwd works, right?	16:03
	← leanucci left	16:03
	→ leanucci joined	16:03
emias	That one simply calls getcwd(3), too: http://svn.freebsd.org/viewvc/base/release/6.2.0/bin/pwd/pwd.c?revision=165962&view=markup	16:04
doener	emias: hm, might be the $PWD fallback that works	16:04
ilteris_	hey guys, looks like I messed up my app, I already committed a working version and pushed it to the server, should I just delete everything from my system and clone a clean copy?	16:05
	or should I just pull the version and merge it?	16:05
	→ fmdb joined	16:05
emias	doener: That's only read if /bin/pwd was called with the "-L" option.	16:05
gwoo	gp5st: if it helps im running 6.3-RELEASE-p2 with git 1.6.2.2 from ports	16:05
jast	okay, freebsd actually uses a getcwd syscall but it may not be present on all architectures or something	16:05
gp5st	hmm	16:06
jast	ilteris_: define "messed up"	16:06
EvanCarroll	jast: cool	16:06
doener	emias: oh, oops, got confused by the comment above the if... made me think that -L is the default	16:06
	ferdy_ → ferdy	16:06
ilteris_	jast, well basically my app is not working as it was working yesterday, I messed up some code I think :\|	16:06
jast	you could now use bisect to find the exact commit where things went wrong (if you made several new commits since then)	16:07
ilteris_	now I am not entirely sure if I pushed a working copy before this, it should the commit before or the one before that. So I need to get back a certain commit, how should I go with that?	16:07
	oh exactly what I was looking for	16:08
	how do I use bisect?	16:08
emias	gp5st: Just to make sure, could you try "git init" and "/bin/pwd" once again (from within the same shell and directory)?	16:08
jast	ilteris_: man git-bisect :)	16:08
Gitbot	ilteris_: the git-bisect manpage can be found at http://git.or.cz/man/git-bisect	16:08
jast	it has examples, I think	16:08
gp5st	/bin/pwd works	16:08
	← bewst left	16:08
emias	gp5st: And "git init" still spits out "cannot tell cwd"?	16:08
gp5st	that's super weird	16:09
	so if i do git init after i relog in, it works	16:09
	→ bradly joined	16:09
gp5st	but then if i rm -r .git and then git init it works	16:09
	→ Alinermedia joined	16:09
gp5st	hmm, let me see if it's the same pattern with checkout	16:09
doener	gp5st: "relog in" == ?	16:10
	← kalle_ left	16:11
doener	gp5st: log out, log in, and do "git init" in $HOME? Or something else?	16:11
	← gotgenes left	16:11
gp5st	yeah, sorry	16:12
	i meant log out, then log in	16:12
	why would that work?	16:12
doener	gp5st: and you do that "git init" in $HOME?	16:12
gp5st	well, ~/repos; but yes, if i log out and log in it works, then if i rm -r .git and git init it doesn't	16:13
doener	gp5st: to get into ~/repos, do you "cd repos" or "cd ~/repos"?	16:14
	← Alinermedia left	16:14
gp5st	i start in $HOME and then cd repos	16:14
doener	hm, if you log out, log in and then "mkdir foo; cd foo; git init", does that work?	16:15
	← bradly left	16:15
ilteris_	hey jast is there any other way than using bisect in order to get another commit form the server rather than the last one?	16:15
	← vienova left	16:15
gp5st	well i can mkdir foo; cd foo; git init w/o logging out then in	16:16
	→ kalle_ joined	16:16
	← scientes left	16:16
doener	gp5st: you can come over and watch my head explode now	16:17
jast	ilteris_: you have local copies of all commits. you can easily return to any old version without ever talking to the server, actually. bisect is just a diagnosis tool that makes use of this.	16:17
MarcWeber	Do same tree contents always get the same hash? So is comparing $HASH^{tree} hashs the way to test for equality?	16:17
emias	gp5st: So you're not able to reproduce the problem within your new ~/foo directory?	16:18
jast	for example, you can throw away all newer commits by getting the commit ID of an older commit (e.g. using git log) and using git reset --hard <commit id> (this will throw away all uncommitted changes).	16:18
ilteris_	jast: yup I am checking out my old commits through gitx now, I have no clue how to "return" to those commits tho, bisect looks complicated to me :\|	16:18
sitaram	MarcWeber: yes	16:18
gp5st	doener: there is a huge redmark on my forehead form problems like this	16:18
jast	I can't stay to give you any additional details. if you need something else than git reset, please ask everyone else about it.	16:19
doener	gp5st: hm, maybe, could you do "pwd -L" and "pwd -P" before the working "git init" and after the failing one?	16:19
	→ AkumaStreak joined	16:19
sitaram	MarcWeber: internally that's what git diff does I think, when you compare two tree-ish objects	16:19
ilteris_	jast thanks	16:19
gp5st	emias: if i rm -r .git then git init	16:19
	→ cody-somerville joined	16:19
gp5st	i can't	16:19
	→ javatexan joined	16:19
gp5st	i can't git init	16:19
ilteris_	man git-reset	16:19
Gitbot	ilteris_: the git-reset manpage can be found at http://git.or.cz/man/git-reset	16:19
cody-somerville	Whats the proper way to handle renames in git?	16:19
	sdiz → sdiz-afk	16:20
straszheim	sorry, i have to reask here, i think i'm missing something dumb. I've got my git svn cloned repository, and I'm going to move the remote branches into the local namespase, so's I can put it up where people can clone and pull from it to track svn w/o using git-svn	16:20
gp5st	that hapens in $HOME and any subdir	16:20
straszheim	not worried about dcommitting from it	16:20
	("move the remote branches into the local namespace" is what I'm trying to grok)	16:21
	← jchris left	16:21
jast	straszheim: if you actually move them in the git-svn repo, git svn fetch will stop working	16:21
doener	straszheim: you can't "move" it, as git-svn would complain about writing to refs/heads/ stuff	16:21
	→ esparkma_ joined	16:21
gp5st	[me@compy ~]$ mkdir foo	16:21
	[me@compy ~]$ cd foo	16:21
	[me@compy ~/foo]$ ls -la	16:21
	total 4	16:21
	drwxr-xr-x 2 me users 512 Apr 23 12:19 .	16:21
	drwxr-xr-x 9 me users 512 Apr 23 12:19 ..	16:21
	[me@compy ~/foo]$ git init	16:22
	Initialized empty Git repository in /usr/home/me/foo/.git/	16:22
straszheim	jast: ok, so i clone the git-svn repo	16:22
gp5st	[me@compy ~/foo]$ ls -la	16:22
	total 6	16:22
	drwxr-xr-x 3 me users 512 Apr 23 12:19 .	16:22
straszheim	in the clone, i do the move	16:22
gp5st	drwxr-xr-x 9 me users 512 Apr 23 12:19 ..	16:22
	drwxr-xr-x 6 me users 512 Apr 23 12:19 .git	16:22
	[me@compy ~/foo]$ rm -r git	16:22
	rm: git: No such file or directory	16:22
	← WALoeIII left	16:22
jast	gp5st: please paste lengthy things to a pastebin next time, e.g. http://git.pastebin.com/	16:22
gp5st	[me@compy ~/foo]$ rm -r .git	16:22
	[me@compy ~/foo]$ git init	16:22
	← sitaram left	16:22
	← knittl left	16:22
gp5st	fatal: cannot tell cwd	16:22
	[me@compy ~/foo]$	16:22
	ugh	16:22
	i copied the pastebin address	16:22
	http://pastebin.com/m3b6e1d04	16:22
	sorry about that	16:22
jast	anyway, gotta go	16:22
gp5st	jast: i know:( i hit cmd-c and ...	16:22
	jast: cya, thanks for your help	16:22
	→ knittl joined	16:22
straszheim	jast: k, thx	16:22
emias	gp5st: And then, /bin/pwd still works?	16:23
doener	gp5st: that misses the "pwd -L/-P" calls I asked for :-/	16:23
	→ romao joined	16:24
gp5st	sorry	16:24
	h/o	16:24
	http://pastebin.com/m1f6e1df4 they're at the bottom	16:25
doener	straszheim: http://pastebin.com/m2089dc0b	16:26
gp5st	err, sorry, /bin/pwd does the same thing	16:26
doener	straszheim: that's basically it. The push does the renaming	16:26
	straszheim: as you already have the svn clone, you can of course skip the "git svn clone" step	16:26
emias	gp5st: What's the output of "truss git init 2>&1 \| grep getcwd"?	16:26
doener	gp5st: and before the working "git init"?	16:27
	straszheim give it a go	16:27
	→ vienova joined	16:27
	→ mfilipe joined	16:28
	straszheim would never have thought about creatively using 'push'	16:28
	→ willb joined	16:28
gp5st	emias: downer: so i made a ~/baz dir and then ran that comment, it took a few seconds, then returned a prompt, then an ls -la showed no .git dir	16:29
	truss: cannot open /proc/curproc/mem: Permission denied	16:29
	truss: cannot open1 /proc/84354/mem: Permission denied	16:29
	i get these idk if they're an issue	16:29
emias	gp5st: Next try: ktrace git init	16:30
gp5st	-bash: /usr/bin/ktrace: Permission denied	16:30
emias	Ugh.	16:30
gp5st	i love being an unprivlaged user:(	16:30
emias	Should work just fine as an unprivileged user (normall) ...	16:30
	(normally)	16:31
	← fmdb left	16:31
doener	gp5st: what does "pwd -L", "pwd -P", "echo $PWD" tell right after login?	16:31
gp5st	my home dir	16:32
	→ Alinermedia joined	16:32
	← tal67 left	16:32
gp5st	grrrrr, i keep forgetting to sanatize those:-\	16:32
doener	/usr/home/... or /home/... ?	16:32
	→ esparkm__ joined	16:32
	← shruggar left	16:33
	← Alinermedia left	16:33
gp5st	http://pastebin.com/m318a82f4 that's right after login	16:33
doener	ok	16:34
	→ vbabiy joined	16:34
telmich	gp	16:34
	← esparkman left	16:34
	← sgrimm left	16:34
	→ mfilipe_ joined	16:35
	→ bcardarella joined	16:35
gp5st	i'm sorry:( I swear i'm not making this up ::bangs head against all::	16:35
	→ aresnick joined	16:36
	← Chaser left	16:36
	← raimo_ left	16:36
	→ WALoeIII joined	16:36
	← aresnick left	16:36
	→ sekimura joined	16:37
	→ leanucci_ joined	16:38
EvanCarroll	jast: doesn't work says to use '--' to seperate path from revisions	16:38
	← leanucci left	16:38
EvanCarroll	fatal: ambiguous argument 'HEAD:Reference.pm': unknown revision or path not in the working tree.	16:38
	← gafmtt left	16:39
	→ solofight joined	16:39
	→ leanucci joined	16:39
	→ sergiodj joined	16:42
	→ Alinermedia joined	16:42
doener	EvanCarroll: which command? And is Reference.pm really in the root directory of the repo?	16:42
	← naeu_ left	16:43
gp5st	i just can't win:-\ even once i have the repo there, i can't clone them, it hangs up on me right after/when it sends the objects	16:43
	→ rinogo joined	16:43
rinogo	Hey all!	16:44
	← esparkma_ left	16:44
rinogo	Can someone please help me understand the difference between 'git clone ...' and 'git add remote ...'?	16:44
gp5st	git clone makes a copy of the repo	16:44
	← cody-somerville left	16:45
RandalSchwartz	git clone - make a local repo pulled from a remote	16:45
	git add remote, just add a remote that you can fetch from	16:45
	or push to	16:45
gp5st	git add remote makes an entry for a server in teh config a short name for it if you will	16:45
RandalSchwartz	no objects are transferred	16:45
gp5st	ditto, git add remote doesn't download or send anything	16:45
doener	git clone ~= git init + git remote add + git fetch + git checkout	16:46
rinogo	Hmmm... So, I have a git repo already set up on my VPS (Centos 5) that I can use without incident locally from the system. However, I'd like to be able to work locally on my OS X laptop and check in to the git repo on the server... Which command is right for me?	16:46
	← sergiodj left	16:46
doener	oops, add "git branch" in between fetch and checkout	16:46
	← ajonat_ left	16:47
Zathrus	doener: can't that be done as part of the checkout?	16:47
	→ RodP2 joined	16:50
doener	Zathrus: sure, the fetch can also be done by "remote add" if you use -f	16:50
	→ peti` joined	16:50
doener	Zathrus: but the idea was to show the involved "high level" steps in the clone command, having convienience stuff in that list seems wrong ;-)	16:51
	← marc_kdab left	16:51
	→ btanaka joined	16:51
peti`	Is it possible set an "Author" attribute on a commit in the history? When I made the commit, I didn't specify an author (other than the committer, me), but now I'd like the commit to show who sent the patch. How can I do that?	16:52
rinogo	So, would you guys say I want to do a git clone or a git add repo?	16:52
wereHamster	peti`: how far back in the history is the commit?	16:53
gp5st	when i do a git clone, aorund halfway through the receiving objects stage, the remote hangs up. is there anyway to figure out why? can ssh be set up to hang up if there is no login shell started after so long? could it be a problems with the remotes? I did git fsck on them and thena repack	16:53
rinogo	(" locally from the system. However, I'd like to be able to work locally on my OS X laptop and check in to the git repo on the server... Which command is right for me?")	16:53
peti`	wereHamster: Very far, unfortunately.	16:53
wereHamster	peti`: changing the author is not possible without rewriting history. Are you aware of the implications?	16:53
	← mfilipe left	16:53
	→ qhoxie joined	16:54
peti`	wereHamster: A little. I know the basic repository format, etc., and I can use rebase --interactive and such.	16:54
gp5st	ringo: clone the repo, then you can push your changes back to the remote after you make them	16:54
	rinogo:	16:54
rinogo	gp5st: Thanks :)	16:54
	→ ttelford joined	16:55
wereHamster	peti`: if other people cloned your repo and you rewrite the history, it will cause them some problems when pulling from you again.	16:55
peti`	wereHamster: I misread you. I read "implementation". :-) Sorry. No, what are the implications of attaching an Author afterwards?	16:55
	wereHamster: I see. That's fine with me though, that wouldn't be a problem.	16:56
	← ttelford left	16:56
wereHamster	peti`: I wouldn't suggest you interactive rebase, instead look at git-filter-branch	16:56
	→ jceb joined	16:56
peti`	wereHamster: Very nice! Thank you for the tip.	16:57
wereHamster	peti`: with --env-filter and set GIT_AUTHOR_NAME/EMAIL	16:57
	→ hugoxrosa joined	16:57
	→ woei joined	16:58
	← leanucci_ left	16:59
peti`	wereHamster: This is exactly what I was looking for. Thanks for your help.	17:00
	← troyt left	17:00
	← galderz left	17:00
	→ LiamH joined	17:01
	← jmesnil_ left	17:02
	→ gberdyshev_ joined	17:02
	→ troyt joined	17:04
	→ d0k_ joined	17:05
	→ static^_ joined	17:05
hobodave	is there a shortcut to add [branch "master"]\n remote = origin \n merge = refs/heads/master into my .git/config ?	17:05
	it's annoying to have to edit that every time I create a repo	17:05
wereHamster	hobodave: man git-config	17:06
Gitbot	hobodave: the git-config manpage can be found at http://git.or.cz/man/git-config	17:06
peti`	hobodave: "git config" might solve that?	17:06
	→ khmarbaise joined	17:06
hobodave	that's still me doing the same work	17:07
	is this a fairly recent change in git?	17:07
wereHamster	also, which version are you using? I never had to add these lines to the config file, git always created them for me	17:07
peti`	hobodave: In a cloned repository, git creates those entries automatically. At lesst, it does when I say "git checkout -b foo origin/foo".	17:07
hobodave	peti`: I'm creating the repo locally	17:07
	then pushing it to the remote origin	17:07
	so it only happens on 'new' ones	17:08
	so git-config is unavoidable?	17:08
	→ joevano joined	17:08
wereHamster	well, then how should git know which remote to track? :)	17:08
peti`	hobodave: Why can't you clone the external repo, work in there, and say "git push"? Where is the difference to the setup you use now?	17:08
	→ markelikalderon joined	17:08
hobodave	peti`: this is a NEW repo	17:09
	there is no external repo	17:09
wereHamster	hobodave: then you have to do it manually, as git can't magically know whcih remote you want to track	17:09
peti`	hobodave: Well, than Git can't know about any external repos when you create a branch. You'll have to do it manually, I guess.	17:09
	→ kbingham joined	17:10
hobodave	yep	17:10
	I understand now	17:10
	thanks guys	17:10
	→ giallu joined	17:10
	→ ben_h joined	17:10
	← stuffcorpse left	17:11
	← kalle_ left	17:11
	← peti` left	17:11
	→ stuffcorpse joined	17:11
	← jaeckel left	17:12
	← d0k left	17:12
	← bdimcheff left	17:13
	← troyt left	17:14
ilteris_	what's the command for resetting a specific file from a specific commit?	17:15
PerlJam	ilteris_: git checkout <commit> -- <path to file> # I think. (too hungry to be sure)	17:16
	→ kalle_ joined	17:17
ilteris_	thanks PerlJam I'll check it now	17:17
	← gp5st left	17:17
	→ Dashkal joined	17:19
	→ solar_ant joined	17:19
	← gberdyshev left	17:19
	→ dreiss joined	17:20
	→ sgrimm joined	17:21
	→ troyt joined	17:22
	→ naeu joined	17:22
	→ schacon joined	17:22
	← vbabiy left	17:23
	← solar_ant left	17:24
	← ben_h left	17:25
	← kbingham left	17:25
	← Pewpewarrows left	17:27
	← g3d left	17:27
	← gitte left	17:28
hobodave	is this the appropriate forum to ask about gitosis?	17:28
	← brtzsnr left	17:29
hacim	doener: forgot to say thanks for the help yesterday with the git branch renaming :)	17:29
	that was exactly what I was missing	17:29
Dashkal	hobodave: There are people in here familiar with it. Ask. If anybody can help they likely will answer.	17:30
	→ Pewpewarrows joined	17:30
hobodave	can I add multiple keys for a user to their keydir/user.pub ?	17:30
hacim	hobodave: yes you can	17:31
hobodave	nice!	17:31
hacim	hobodave: in the same way you would just concatenate them in an authorized_keys file	17:31
hobodave	I was hoping so, but was afraid to try in case it locked me out	17:31
	→ solar_ant joined	17:31
	← grahal left	17:32
	→ mooseish joined	17:32
	← solar_ant left	17:32
	→ jas4711 joined	17:34
	← ben_h_ left	17:34
	← jas4711 left	17:35
	→ cpg joined	17:36
	→ solar_ant joined	17:36
	← yec left	17:36
	→ gaveen joined	17:36
	← hdl left	17:36
	→ bobesponja joined	17:36
	→ jas4711 joined	17:36
	→ aresnick joined	17:37
	← vienova left	17:37
	← thumper left	17:37
	← aresnick left	17:37
	← markelikalderon left	17:37
	→ hdl2 joined	17:38
	← willb left	17:40
	→ ttelford_ joined	17:42
	→ g3d joined	17:42
	→ chalky joined	17:43
	aspotashev_ → aspotashev\|eeepc	17:43
	→ Chommik joined	17:46
Chommik	hello	17:46
	← troyt left	17:46
	← ttelford_ left	17:47
	← bcardarella left	17:48
	← RaySl left	17:48
	→ troyt joined	17:49
	→ hyperair joined	17:49
btanaka	is it possible to sort the branches listed via "git branch" in chronological order (based on date the branches were created)?	17:49
	→ RaySl joined	17:49
	← g3d left	17:50
	← srid left	17:52
	← qrush left	17:52
	← doener left	17:52
	→ doener joined	17:52
	→ saurabh1403 joined	17:53
	→ reprore_ joined	17:54
	← tono left	17:54
	← solar_ant left	17:55
	→ g3d joined	17:55
khmarbaise	What is the best way to migrate a subversion repository to Git?	17:55
	← troyt left	17:55
	→ leanucci_ joined	17:56
	→ troyt joined	17:56
	← priidu left	17:57
	← leanucci left	17:57
bremner	man git-svn	17:59
	← dodo_the_last left	17:59
Gitbot	bremner: the git-svn manpage can be found at http://git.or.cz/man/git-svn	17:59
bremner	khmarbaise: ^^	17:59
	→ stianhj joined	18:00
	← reprore_ left	18:00
khmarbaise	bremner: If i understand the page correct it is intended to just synchronize a single branch/trunk with git and migrate the full history of a repository...	18:00
	← solofight left	18:00
	← aspotashev left	18:00
bremner	khmarbaise: what do you want to do?	18:01
	→ priidu joined	18:01
khmarbaise	bremmer: /and migrate/and not migrate/	18:01
	bremmer: I would like to migrate the full history with all branches to git without loosing branches, cause I'm working on different branches ....	18:01
	→ aspotashev joined	18:01
	← hyperair left	18:02
	→ solofight joined	18:02
khmarbaise	and of course the tags as well...	18:03
	← kuadrosx left	18:03
	← ngng left	18:03
bremner	khmarbaise: hmm, ok. it can be done, I guess. See --branches and --tags for start. I don't have any personal experience with svn branches, so I'll leave it for more informed folk	18:04
justatheory	doener++ # Thanks for the help yesterday, `git filter-branch --tag-filter cat -- --all` did the trick.	18:05
Nilbus	how do you reset 1 file to how it was at HEAD?	18:05
	← hugoxrosa left	18:06
	→ markelikalderon joined	18:06
	← troyt left	18:06
stianhj	i just did a git clone of my remote repo, and after successfully completing I run git status, and tells me a file is "Changed but not updated".. there were no files before, and I haven't changed the file after cloning. So I run a diff on the file, and git has removed every single line, and added the exact lines again (@@ -1,370 +1,370 @@). what's wrong here?	18:07
	→ earcar joined	18:07
	→ iso42 joined	18:07
	→ radarek joined	18:08
khmarbaise	bremner: I have a account on github.com and tried to migration there and it seemed to be working...but now i would like to do the same my own...some other hint ?	18:10
	→ aspotashev_ joined	18:10
	← aspotashev left	18:10
	← aspotashev\|eeepc left	18:10
	← cannonball left	18:11
	← kalle_ left	18:11
	→ kalle_ joined	18:11
bremner	khmarbaise: afaik, git-svn with --branch --tags or --stdlayout should work for you, _If_ you svn layout is not too creative	18:11
	→ cannonball joined	18:12
	→ jrmuizel joined	18:12
stianhj	s	18:12
	→ kuadrosx joined	18:13
	→ priidu_ joined	18:15
	→ solar_ant joined	18:15
	→ FreakGuard_ joined	18:16
	← priidu left	18:16
	← FreakGuard left	18:16
	← Pewpewarrows left	18:17
	→ troyt joined	18:18
PuffTheMagic	i cant seem to get git to track some latex files	18:18
	→ Grahack joined	18:18
	← Tv left	18:18
khmarbaise	Hm...i have tried to convert via git svn info URL --stdlayout (and --trunk etc.) but after git svn fetch is ready i do "git branch -l" and see only "* master" nothing else.....	18:18
PerlJam	git branch -a	18:19
	(I'm fairly sure it's not going to make local branches for you)	18:20
Chommik	is there any tutorial saying how to create a repo on apache server? i found something on google, but it don't work for me	18:20
	→ Pewpewarrows joined	18:21
	→ wwwald joined	18:21
bremner	khmarbaise: git branch -a ?	18:21
	← Fullmoon left	18:22
kusma	hahahhaha	18:22
	BTD er paa nrk2	18:22
	→ qrush joined	18:22
	→ hugoxrosa joined	18:22
kusma	whops	18:22
PuffTheMagic	how do I tell git to track a new file type	18:23
	it seems to be ignoring .tex files	18:23
stianhj	check your .gitignore file?	18:23
	tag_ → tag	18:23
stianhj	kusma: BTD? på nrk2?	18:24
kusma	stianhj: feil vindu ;)	18:24
	→ cilly joined	18:24
stianhj	hehe, ok	18:24
	→ dodo_the_last joined	18:24
stianhj	Chommik: http://tinyurl.com/fomkq	18:24
	→ jkp joined	18:25
	← jkp_ left	18:25
stianhj	Chommik: That's "Howto setup git server over http (apache)"	18:25
Chommik	stianhj: i was doing as it says, but I get an error from apache	18:25
rinogo	gp5st: BTW, thanks for the help - git clone seems to have done the trick.	18:26
Chommik	stianhj: http://rafb.net/p/THOqqT57.html	18:27
	i set DavLockDB in httpd.conf but it makes no difference	18:27
	→ davidad joined	18:29
	→ nurey joined	18:29
	← markelikalderon left	18:30
	→ altrux joined	18:31
	← rinogo left	18:32
	→ tasmani joined	18:32
	← tasmani left	18:33
	← solar_ant left	18:34
	FreakGuard_ → FreakGuard	18:34
	← cbreak\|evil left	18:35
	→ bryanl joined	18:37
	← ceyusa_ left	18:38
	→ markelikalderon joined	18:38
	→ bcardarella joined	18:38
	← bryanl left	18:39
	← DavidKlein left	18:39
	← solofight left	18:39
	→ cbreak joined	18:41
	→ darrob joined	18:43
	← alanhaggai left	18:43
	← catalis left	18:46
	← cilly left	18:48
	→ catalis joined	18:49
	→ Skiessi joined	18:49
Skiessi	is there anything which does what one would assume 'git log --branch foo' to do?	18:50
krh	Skiessi: what do you expect it to do?	18:52
	git log foo will show you the commit log for the branch foo	18:53
Skiessi	okay	18:53
	← Skiessi left	18:53
	→ willb joined	18:55
	→ bryanl joined	18:56
justatheory	So now I have a git repo migrated from an old cvs repo, and one migrated from a later svn repo.	18:58
	→ MateuszDraco joined	18:58
justatheory	There are some branches common between these two repos, and some that are not.	18:58
	How can I stitch them together into a single repo with complete history?	18:58
	→ tono joined	19:00
justatheory	I used the examples here: http://git.or.cz/gitwiki/GraftPoint?highlight=%28graft%29	19:00
	better link: http://git.or.cz/gitwiki/GraftPoint	19:01
	→ solofight joined	19:01
solofight	help \| how can i create a git repo which is accessible by anybody over http	19:02
justatheory	That seems to work, but branches in the cvs repo that arent in svn don't show up.	19:02
solofight	i read abt http auth	19:02
	but i dont know basics such as assigning git:// protocol to my repo	19:02
	any links	19:02
	plz guide me	19:02
Ilari	justatheory: Basically, pick canonical version and make all children of noncanoical ones children of the canonical one. Also, update all refs to noncanonical commits point to canonical equivalents.	19:02
justatheory	Ilari: Sorry, not able to parse that.	19:03
Ilari	justatheory: Also, maybe remote tracking branch behaviour is involved?	19:03
MateuszDraco	hi! is there any good windows git client for windows with nice, neat gui? ;)	19:03
justatheory	Ilari: Yeah, all the branches appear to be remotes.	19:04
	← aspotashev_ left	19:04
Ilari	justatheory: Remotes aren't cloned on git clone.	19:04
	← intripoon left	19:04
justatheory	Ilari: Hrm.	19:04
	→ cilly joined	19:05
justatheory	Okay, I created my cvs and svn migrations by doing a lot of grafting, running `git filter-branch`, and then cloning to drop the dupes.	19:05
	The branches had been both local and remote, but were remote only in the clone.	19:05
Ilari	solofight: For git://, you need git-daemon + mark the repo as exportable. For HTTP fetching, you need to put the repo files somewhere accessable and run git update-server-info after EVERY update to repo.	19:05
	← loincloth left	19:05
	→ loincloth joined	19:05
justatheory	Ilari: Does that sound right?	19:07
	← wshimmy left	19:07
	← stuffcorpse left	19:07
	← Pewpewarrows left	19:08
Ilari	justatheory: Yeah, about right.	19:08
justatheory	Okay	19:08
	So the trick is to make the old branches from cvs show up in the svn to git repo	19:08
	← catalis left	19:08
justatheory	I copied of the pack files, and the commits show up, but not the branches.	19:08
	Do I need to make them local first?	19:08
	← aisbaa left	19:09
Ilari	justatheory: Yeah, make them local before cloning... The local branch stuff is in packed-refs and refs/heads	19:09
	← priidu_ left	19:10
	← cedricv left	19:10
justatheory	and by "before cloning" you mean "before copying the files in .git/objects/pack"?	19:10
	← kalle_ left	19:11
	→ Pewpewarrows joined	19:11
justatheory	Ilari	19:11
Ilari	justatheory: Why are you copying those files between repos manually?	19:11
justatheory	Ilari: Was following http://git.or.cz/gitwiki/GraftPoint	19:11
	The two scripts at the end	19:11
	→ e_6 joined	19:11
	← bcardarella left	19:11
	→ kalle_ joined	19:12
Ilari	justatheory: And just copying packs might not suffice. There can also be loose objects.	19:12
justatheory	Ah	19:12
	So I can clone from the cvs repo to an existing svn repo?	19:12
	→ loinclot_ joined	19:12
justatheory	hrm, not according to the `git clone` docs…	19:13
Ilari	justatheory: Clone doesn't work if destination exists. Perhaps you want fetch instead?	19:13
	← loincloth left	19:13
	→ xyz joined	19:13
justatheory	Ilari: I dont' know. That's what I'm trying to figure out.	19:13
	→ earcar_ joined	19:13
Ilari	justatheory: fetch copies branches whole.	19:13
justatheory	Oh, that looks promising	19:14
	→ priidu joined	19:14
	justatheory reads `git help fetch`	19:14
stianhj	i just did a git clone of my remote repo, and after successfully completing I run git status, and tells me a file is "Changed but not updated".. there were no files before, and I haven't changed the file after cloning. So I run a diff on the file, and git has removed every single line, and added the exact lines again (@@ -1,370 +1,370 @@). what's wrong here?	19:15
Ilari	stianhj: Msysgit?	19:15
stianhj	Ilari: yes	19:15
Ilari	stianhj: Probably automatic CRLF translation screwing up for some reason...	19:16
stianhj	Ilari: good point.. didn't think of that. annoying..	19:16
	hdl2 → hdl	19:17
Ilari	stianhj: If your editors can load LF-only files, you could set CRLF translation to be input only.	19:17
	stianhj: 'git config --global core.autocrlf input'.	19:17
stianhj	this is a project developed on linux, but I cloned it at home to have a look at it..	19:18
	isn't there a line ending conversion thingy in msysgit?	19:18
	→ sergiodj joined	19:18
	← MateuszDraco left	19:18
Ilari	stianhj: You could try setting CRLF conversion to be input only (using above command) and then using git reset --hard...	19:19
stianhj	ok	19:19
Ilari	stianhj: Or better yet, instead of reset, 'git checkout -f'.	19:19
stianhj	the -f is?	19:19
Ilari	stianhj: Force.	19:19
stianhj	i tried git checkout filename earlier, didn't do anything	19:20
solofight	Ilari: can i pm you ?>	19:20
stianhj	git reset --hard, tells me HEAD is at the last commit, but git status gives a modified file	19:20
Ilari	solofight: man git-daemon talks about how to mark repo as exportable.	19:21
Gitbot	solofight: the git-daemon manpage can be found at http://git.or.cz/man/git-daemon	19:21
stianhj	Ilari: git checkout filename -f does nothing as well	19:21
	→ seangrove joined	19:22
Ilari	stianhj: And those files are still listed as modified?	19:22
	cpg → cpg\|away	19:22
stianhj	yes	19:22
	← gwoo left	19:22
justatheory	Ilari: If I use `git fetch`, should I fetch new commits from svn into the older repo from cvs, or the other way around? Or does it matter?	19:23
Ilari	justatheory: Doesn't really matter.	19:23
	→ Modius_ joined	19:23
justatheory	ok	19:23
	justatheory tries it	19:23
	→ ceej joined	19:23
Ilari	justatheory: Or actually, CVS to SVN, because CVS tends to be fixed and can be filter-branched before fetch.	19:23
	justatheory is confused again.	19:24
	← cannonball left	19:24
justatheory	Sorry, I'm quite the git newbie and am still trying to understand the huge panoply of commands.	19:24
Ilari	justatheory: Fetching commits that have active grafts rules might result something broken.	19:24
	LotR is reminded he owes justatheory a reply email	19:24
	stianhj is brb	19:24
justatheory	Ilari: Ah, yes. I've already run filter-branch in both repos.	19:24
Ilari	justatheory: Then it doesn't really matter.	19:25
justatheory	okay	19:25
	loinclot_ → loincloth	19:25
	← Chommik left	19:26
justatheory	That looks better.	19:26
	→ gitte joined	19:26
justatheory	All the tags were copied, but not the branches.	19:26
	So I guess I need to make the branches local first?	19:26
Ilari	justatheory: With suitable refspecs, it can copy remote branches as well.	19:27
justatheory	Oh, I didn't specify any.	19:27
	justatheory isn't sure what a refspec is	19:27
	→ robinr joined	19:27
	→ rudi_s joined	19:28
	→ dwave joined	19:28
solofight	Ilari: have a doubt , stuick in middle plz help me	19:28
	← earcar_ left	19:28
Ilari	justatheory: Basically, it specifies what branches to transfer and what names to save them as.	19:28
solofight	inside /version/ i typed git-init	19:28
	and git add 2009	19:29
	← earcar left	19:29
solofight	git commit	19:29
justatheory	Ilari: Ah, I want all branches and their named kept the same.	19:29
	→ bcardarella joined	19:29
Ilari	justatheory: Basic from is 'refs/heads/foo:refs/heads/bar', copy local branch 'foo' and save as local branch 'bar'.	19:29
justatheory	Though there are three branches that exist in both, so I use grafts to hook them up.	19:29
solofight	now should i be inside /version/and type git-daemon -export-all ?	19:29
Ilari	solofight: Probably 'git daemon --detach --verbose --export-all' (that pretty much the minimum)...	19:30
	solofight: But then repo paths when fetching are relative to _filesystem root_.	19:31
	→ mugwump joined	19:31
Ilari	solofight: And you shouldn't export working repos, as they can be too unstable to publish.	19:31
solofight	oh,...	19:31
	so what should i do ?	19:31
	← Dashkal left	19:32
	→ plasticine joined	19:32
stianhj	Ilari: any ideas?	19:33
Ilari	solofight: Create repo somewhere using 'git init --bare', then go to original repo and 'git remote add public <path of new repo>', 'git push public master' so it pushes your branch to that public repo. Them others can clone that.	19:33
	stianhj: Maybe try 'git checkout .'?	19:33
	stianhj: Maybe try 'git checkout HEAD .'?	19:33
	stianhj: You did change CRLF mode to input translation only?	19:34
stianhj	Ilari: git checkout . and git checkout HEAD . and git checkout HEAD filename, etc doesn't do anything	19:36
	Ilari: and yes I did the input translation thing	19:37
	← Modius left	19:37
Ilari	stianhj: What does 'git config core.autocrlf' yield?	19:39
	→ hyperair joined	19:39
stianhj	Ilari: input	19:39
Ilari	stianhj: Hmm... Wonder why it doesn't overwrite those files...	19:40
	→ chris2 joined	19:40
	← lorandi left	19:41
	← javatexan left	19:41
solofight	Ilari: i did all what you said	19:41
stianhj	Ilari: I did a clean clone as well, no difference	19:41
solofight	then i have to go to new repo and type git daemon --verbose --detach --export-all	19:42
	?	19:42
	→ intripoon joined	19:42
Ilari	solofight: No need to go to the repo. That git daemon command can be run from anywhere.	19:42
solofight	ok	19:42
	now i executed the git daemon	19:43
	now wat i do ?	19:43
	← woei left	19:43
	→ cannonball joined	19:43
solofight	how do i get that to my local system ?	19:43
Ilari	solofight: 'git clone git://server/path/to/repo.git' maybe. Note, the path is relative to filesystem root.	19:44
	→ Dashkal joined	19:44
	← hyperair left	19:46
SRabbelier	I'm getting " MSGFMT po/de.msg make[1]: *** [po/de.msg] Error 127" while making git.git on my new Ubuntu 9.04 VM, ring any bells with anyone?	19:46
	→ hyperair joined	19:46
Fissure	gettext not installed?	19:46
SRabbelier	Fissure: very well possible	19:47
	← mugwump left	19:47
Ilari	SRabbelier: Looks like msgfmt not found.	19:47
SRabbelier	Ilari: msgfmt == gettext?	19:47
	→ iListenU joined	19:48
	→ zpoley1 joined	19:48
SRabbelier	ah, msgfmat is provided by gettext	19:48
iListenU	Please help, i get problem in: http://parduoduv.puslapiai.lt at header error. (top). What can it be?	19:48
	SRabbelier installs gettext	19:48
	← iListenU left	19:48
zpoley1	hey guys - how do you revert a single file's changes that haven't even been git added yet?	19:48
Ilari	zpoley1: 'git checkout file'?	19:48
	← saurabh1403 left	19:48
zpoley1	Ilari: great! thank you	19:49
SRabbelier	zpoley1: or 'git checkout HEAD -- file' if you did add them ;)	19:49
	→ Kimmie joined	19:49
zpoley1	SRabbelier: very nice. thank you	19:50
	→ mw joined	19:51
	← ceej left	19:51
	SRabbelier wonders if git's makefile could be made more forgiving	19:51
	← Dashkal left	19:52
SRabbelier	then again, I guess that's what ./configure is for	19:52
	← dwave left	19:52
Grum	while in a rebase -i how can i actually edit the things the commit changed? it appears i can only do a commit --amend and when i ask for git status it tells me i it is all clean (i sortof expected files to be marked as 'modified')	19:53
	← btanaka left	19:53
doener	Grum: git reset HEAD^, and then use just "commit" instead of "commit --amend"	19:54
Grum	cheers	19:55
straszheim	okay, I think I've got this mostly worked out	19:55
	http://sodium.resophonic.com/git/boost/about/	19:55
	am I doing anything stupid?	19:55
	→ lorandi joined	19:56
solofight	Ilari: says http://pastebin.com/m5589c2e3	19:56
doener	straszheim: you should probably setup HEAD to reference something else than refs/heads/master	19:56
straszheim	ah!	19:56
	→ earcar joined	19:57
straszheim	yes that -n is a little annoying. 'trunk' will do.	19:57
	← dreiss left	19:57
Ilari	solofight: Firewall preventing connection (to port 9418)?	19:57
doener	straszheim: assuming that that repo is bare, just "git symbolic-ref HEAD refs/heads/trunk"	19:57
solofight	dontknow	19:58
	i dont have any firewalls	19:58
	nuvedacl.com should be accesible by browser is it ?	19:58
Ilari	solofight: Browser uses port 80. git:// uses 9418 by default.	19:58
solofight	cause i created a domain name called nuvedacl.com and created a vvirual host	19:58
	virtual hosts	19:59
Ilari	solofight: And it isn't case of nothing listening on remote side. That would cause "Connection Refused" instead of "Connection Timed Out".	19:59
solofight	oh..	19:59
straszheim	hah!	19:59
	thanks doener!	20:00
solofight	i dont have any fire walls	20:00
	← jceb left	20:00
Ilari	solofight: Well, either 174.129.251.48 is a blackhole or there's firewall somewhere.	20:01
SRabbelier	straszheim: so as I read this it says "the '-n' will...", but there's no -n in the commandline above it? :P	20:01
	straszheim is editing it right now to reflect doener's fix	20:02
justatheory	Stupid question: What is origin/HEAD and how does it relate to master?	20:02
straszheim	now it automagically checks out 'trunk' for you. :)	20:02
	→ Dashkal joined	20:02
straszheim	thanks for looking. I think we've got this almost locked down.	20:03
Ilari	solofight: And if its blackhole, there's nothing there.	20:03
	→ greyface joined	20:03
Ilari	solofight: That is, the address doesn't correspond to any host.	20:03
	→ wavethenavel joined	20:03
	→ Tv joined	20:03
doener	justatheory: origin/HEAD is usually a symbolic ref, setup by "git clone" and references the same branch head that HEAD of the origin repo referenced when you cloned	20:03
	→ catalis joined	20:04
justatheory	doener: Okay then I can ignore it for now. Thanks.	20:04
	→ mugwump joined	20:04
doener	justatheory: it works as a shortcut, if origin/HEAD references origin/master, then "origin/master", "origin/HEAD" and "origin" are all the same, when used in a context where a ref is accepted	20:04
justatheory	okay	20:04
	← cilly left	20:05
	← eletuchy_ left	20:05
	← bcardarella left	20:05
	→ javatexan joined	20:06
	→ bobmcw_ joined	20:06
	esparkm__ → esparkman	20:09
	← kalle_ left	20:11
	→ Kebianizao joined	20:11
Kebianizao	Hello	20:11
	→ bgerber joined	20:11
	← rudi_s left	20:11
literal	hi	20:11
Kebianizao	could I export (in the svn sense) just a certain branch of a repo?	20:11
	I don't want the whole story	20:11
literal	git archive	20:12
Kebianizao	thanks	20:12
	← loincloth left	20:12
	→ loinclot_ joined	20:12
	← hacim left	20:12
	← schacon left	20:14
	→ dwave joined	20:15
	→ sakyamuni joined	20:16
	→ kalle_ joined	20:16
	← cannonball left	20:17
Kebianizao	git archive --remote git://git.kernel.org/pub/scm/linux/kernel/git/anholt/drm-intel.git (remote end hung unexpectedly) :/	20:17
	→ hgimenez joined	20:17
Kebianizao	I've tried ending with qa-branch which is the branch I want to export	20:17
	loinclot_ → loincloth	20:17
	→ johan-s joined	20:18
Kebianizao	is that the right way? could anyone please test?	20:18
	← priidu left	20:18
	← bobmcw left	20:19
	→ eletuchy joined	20:19
literal	hm	20:19
	← solofight left	20:19
	→ priidu joined	20:20
	← partha left	20:21
Fissure	Kebianizao: i don't know if kernel.org has archive sending enabled	20:25
	you can probably just get it through its gitweb...	20:25
Kebianizao	heh, that would explain it	20:25
literal	github doesn't seem to have it either, but they have Download links for that	20:25
Kebianizao	really? how? http://git.kernel.org/?p=linux/kernel/git/anholt/drm-intel.git;a=shortlog;h=qa-branch	20:25
	→ AlexMax joined	20:25
Kebianizao	I can't see anythin like that	20:25
AlexMax	I'm reading in a git faq that this sort of command is used to show a file at a particlar point in time	20:26
	→ bobmcw joined	20:26
AlexMax	$ git show HEAD^:path/to/file	20:26
literal	Kebianizao: the "snapshot" link	20:26
Ilari	http://git.kernel.org/?p=linux/kernel/git/anholt/drm-intel.git;a=snapshot;h=44ab43155e8071fbf037513e57de9a79044edf56	20:26
AlexMax	however, when i do this command myself, I get a message about needing a -- between the revision and the file name	20:26
Kebianizao	ah, I see it, but that's on tree right?	20:26
Ilari	Wrong branch.	20:26
	http://git.kernel.org/?p=linux/kernel/git/anholt/drm-intel.git;a=snapshot;h=ba1d2a9be507cda299c15740ff7e2bb3705a4792	20:27
AlexMax	and when i do, it does not display the entire file, just the changeset for that partiuclar file in that particular revision	20:27
literal	Kebianizao: what do you mean "on tree"?	20:27
Kebianizao	yess, the tree view	20:27
	→ WALoeIII_ joined	20:27
literal	Kebianizao: no, there are tons of snapshot links in the link you pasted	20:27
Kebianizao	ahh, I'll recheck	20:28
literal	on the right	20:28
Kebianizao	ahh :) silly me	20:28
	sorry	20:28
Fissure	AlexMax: that error message means HEAD^:path/to/file is invalid (no such object exists)	20:29
	← WALoeIII_ left	20:31
	← bratsche_ left	20:32
Kebianizao	thanks for all, bye	20:33
	← Kebianizao left	20:33
	← jkp left	20:33
	→ jkp joined	20:34
	→ WALoeIII_ joined	20:34
	→ bratsche_ joined	20:35
	← WALoeIII_ left	20:35
	→ gwoo joined	20:35
	← bobmcw_ left	20:36
	→ stuffcorpse joined	20:37
	→ halfmouse joined	20:37
	← Grahack left	20:39
	← dotsintacks left	20:39
	→ WALoeIII_ joined	20:39
	→ dotsintacks joined	20:39
	← AlexMax left	20:40
halfmouse	if i want to lose the changes i've made in a git repo, the correct command is git reset --soft or --hard?	20:40
	(changes not commited)	20:41
Ilari	halfmouse: --hard	20:41
	→ dreiss joined	20:43
	→ hax0r1337 joined	20:44
	← WALoeIII left	20:44
	← wavethenavel left	20:44
	← WALoeIII_ left	20:45
	← dwave left	20:49
	→ peteinlux joined	20:51
hobodave	hey guys	20:51
	← priidu left	20:52
hobodave	what would I use to clone a repository such that it could be updated via 'git fetch'	20:52
	→ bobmcw_ joined	20:52
	← cemerick left	20:53
Ilari	hobodave: What are you trying to do? git fetch does update remote tracking branches by default.	20:53
	← d0k_ left	20:53
hobodave	I have a redmine installation that needs a local copy of the repository	20:53
	http://gist.github.com/100634	20:53
	I was told to use that	20:53
	however, I dont know how to get the repository so that it works w/ that script	20:54
	← Weaselweb left	20:54
Ilari	hobodave: That doesn't look sane. Trying to update checkouts with git fetch? That doesn't work.	20:55
hobodave	it's not a checkout	20:55
	it requires the .git dir	20:55
	that's it	20:55
Ilari	hobodave: Presumably clone --mirror then. Assuming it sets origin.	20:56
	← neoeinstein left	20:57
	← alester left	20:58
	→ wezyde joined	20:58
	← wezyde left	20:58
	→ tony_ joined	20:58
	→ schacon joined	20:58
	→ porcnick joined	20:59
	← chris2 left	20:59
	→ chris2 joined	21:00
	← esparkman left	21:00
straszheim	ok, that rocks. here's a different question: if the upstream svn makes a new branch/tag, is there a way to make this availabe in the tracking git repo?	21:01
	← jasmin left	21:01
	← gitte left	21:01
	→ neoeinstein joined	21:01
	→ loinclot_ joined	21:01
	← faithfulgeek left	21:01
straszheim	oh, there is. sorry for the noise.	21:03
	← loincloth left	21:03
	→ lack joined	21:03
	← bobmcw left	21:08
	← loinclot_ left	21:08
	→ fujin joined	21:09
	← disappearedng left	21:09
	← peteinlux left	21:09
	← kalle_ left	21:11
	← earcar left	21:11
	← khmarbaise left	21:11
	→ kalle_ joined	21:12
	← sea-gul1 left	21:12
	→ offbytwo joined	21:14
	← offbytwo left	21:14
	← qrush left	21:15
	→ peteinlux joined	21:18
	← Tricon left	21:18
	→ dwave joined	21:19
peteinlux	hello	21:20
	← dwave left	21:20
	→ gotgenes joined	21:21
Ilari	peteinlux: hi	21:21
gotgenes	Is there a way I can git stash to just store a diff from the last committed state, and when appyling the stash, not try to get the index back to its previously committed state?	21:22
peteinlux	llari: just came in seems like everyone is away	21:22
	← altrux left	21:22
gotgenes	Basically, I want to stash some changes, merge from another branch, then put those changes back on top after the merge.	21:22
	Something like bzr shelve	21:22
	← leanucci_ left	21:22
Ilari	peteinlux: Proper IRC clients have channel activity monitoring...	21:23
	→ jeberly joined	21:24
	→ ceej_ joined	21:24
	→ poseidon joined	21:24
peteinlux	i have a question aswell, im tracking a project in svn, i do a 'git svn rebase' on top of that i have my own branch where I do git rebase master, now i went a little to far and fetched and also rebased to the head of svn repo instead of a tag, can i possibly tell git to fetch back?	21:24
	← hyperair left	21:25
	← ceej_ left	21:25
parasti	gotgenes: git stash does not restore index state by default	21:25
	→ ceej joined	21:25
Ilari	peteinlux: 'git reflog' and then use 'git reset --hard <committish>'?	21:25
	← jeberly left	21:25
	→ hyperair joined	21:26
	→ amystrat joined	21:27
amystrat	just started getting this error on gitosis -> error: hooks/post-receive exited with error code 1	21:27
	how can I troubleshoot the post-receive hook?	21:27
Ilari	amystrat: Eh, I thought git ignores exit status of post-receive...	21:29
	amystrat: On what repo? gitosis-admin.git or something else+	21:29
amystrat	this just started happening, we have been using gitosis 6+ months	21:29
	on a project repo, not the admin	21:29
peteinlux	llari: thanks	21:30
Ilari	amystrat: Should that repo even have post-receive hook?	21:30
amystrat	yeah, we have an email being sent out	21:30
Ilari	amystrat: Where did you get the hook script from?	21:31
amystrat	it was the default script modified a bit	21:32
	← hyperair left	21:33
	→ Beket joined	21:33
Beket	Hi all:) Is it possible to count how many commits are between A and B commit ?	21:33
Ilari	amystrat: Modifying scripts "a bit" can break them in subtle ways. Especially if rest of the script is not taken into account.	21:33
Beket	(assuming that there is only one branch of development)	21:34
amystrat	llari: indeed, but it was working fine for 6 months	21:34
	I would just like to be able to see the errors	21:34
	to try and isolate	21:34
	← peteinlux left	21:34
amystrat	what we modified was to have it send the entire diff, not just a summary on commit	21:34
	→ WALoeIII joined	21:34
	← WALoeIII left	21:34
	→ WALoeIII joined	21:35
Ilari	Beket: 'git rev-list A...B \| wc -l'?	21:35
	← gotgenes left	21:35
	→ bcardarella_ joined	21:36
	← trmanco left	21:36
Ilari	Beket: E.g. if B is direct parent (and only parent) of A, that returns 1.	21:37
Beket	Thanks Ilari :) I was stupid to checkout the old commit & I'm building world as we speak.	21:37
	→ krawek joined	21:37
amystrat	llari: is there a way to fire off post-receive manually?	21:37
Beket	So i'll test it later	21:37
Ilari	amystrat: It takes input from stdin, so it might be nontrivial.	21:38
amystrat	thx	21:38
	← davidad left	21:39
Ilari	amystrat: Or actually, capturing proper testcase is nontrivial. Once you have that, its quite easy to repeat.	21:39
	→ nekrad joined	21:40
amystrat	llari: can you expand?	21:40
	it is consitently failing	21:40
nekrad	hi folks, is it possible to cut off an old tail of the history (w/o loosing the recent commits) ?	21:40
	← Tuomas left	21:41
Ilari	amystrat: post-receive API transmits updated refs list via stdin. So to get proper testcase, you would need to capture that input and then replay it.	21:41
	amystrat: Replay is no problem (shell input redirection), but the capture is.	21:41
amystrat	I would not know how to capture that or where to look to find out how	21:42
	← RaySl left	21:42
	→ altrux joined	21:42
Ilari	amystrat: Another way would be to just make up correct data for some testcase.	21:43
	amystrat: Did you push any new branches when it started failing? That's an edge case, so it very well might cause failures...	21:44
	← ph^ left	21:44
amystrat	llari: can't be sure, we have 10+ devs and no one is sure when it started	21:45
	← bcardarella_ left	21:45
	← romao left	21:45
Ilari	amystrat: The format of the input is: '<oldsha> <newsha> <ref>', one line per ref. 40 0s means "NULL". Old ref is NULL on ref create, new ref is NULL on ref delete.	21:46
	amystrat: ref corresponding to branch X is refs/heads/X	21:46
amystrat	llari: cool, is there docs on that?	21:47
Ilari	amystrat: man githooks maybe?	21:47
Gitbot	amystrat: the githooks manpage can be found at http://git.or.cz/man/githooks	21:47
amystrat	thx all for your help, need to do some digging now	21:48
	← kraymer left	21:48
	← gberdyshev_ left	21:48
	← hugoxrosa left	21:48
	← lucsky left	21:49
	← naeu left	21:49
	← willb left	21:51
	← jackdempsey left	21:52
	→ hyperair joined	21:52
	→ ocornu joined	21:54
ocornu	hi	21:54
	← Beket left	21:54
	→ qrush joined	21:54
max_ep	Hmm, how do I delete a remote branch? I pushed out some changes and realized I'd brainlessly done a merge rather than rebase... I was unable to push -f it, so I made a new branch...	21:54
jimmyz2	can anyone point me in the right direction to get git to run a "php --syntax-check" <filename> on a push if the filename ends in .php, to verify no syntax errors get checked in?	21:55
max_ep	but how do I get rid of the junk one	21:55
qrush	git push <remote> :<branch>	21:55
	http://gitready.com/beginner/2009/02/02/push-and-delete-branches.html	21:55
max_ep	Oh, never would've thought to look in push :) Thanks	21:56
	Luckily not an issue I'd run into before	21:56
	→ lll joined	21:57
max_ep	Awesome, it's perfect now :)	21:57
	← bdiego left	21:58
ocornu	how can i switch to a branch/commit without changing the working tree?	21:58
max_ep	ocornu: git stash	21:58
qrush	you can switch using git checkout and your working tree will stay. or just do git stash	21:58
robinr	close. checkout will change you working tree. it merges you local changes to the other branch	22:00
qrush	er, yeah	22:00
	:)	22:00
	← Yuuhi left	22:00
qrush	just do a git stash	22:00
	← GyrosGeier left	22:01
	← vmiklos left	22:01
	← xl0 left	22:01
	← OgMaciel left	22:01
	← lorandi left	22:01
Ilari	jimmyz2: man githooks . Note that you probably want hook that can deny push (pre-receive or update) and note that multiple commits can be pushed at once.	22:01
Gitbot	Ilari: the githooks manpage can be found at http://git.or.cz/man/githooks	22:02
	→ minciue joined	22:02
	← gwoo left	22:02
	→ GyrosGeier joined	22:02
	→ vmiklos joined	22:02
	→ OgMaciel joined	22:02
	→ xl0 joined	22:02
	→ hegge joined	22:02
	→ Ingmar joined	22:02
	→ daleglass joined	22:02
	→ natlus joined	22:02
	→ pcgod joined	22:02
	→ maxb joined	22:02
	→ replaca joined	22:02
	→ IRSeekBot joined	22:02
	→ incd^ joined	22:02
	→ piroko joined	22:02
	→ frsk joined	22:02
	→ yinkei joined	22:02
	→ bmalee joined	22:02
	→ xkr47 joined	22:02
	→ hgb joined	22:02
	→ majoh joined	22:02
	→ mithraic joined	22:02
	→ dionoea joined	22:02
	→ Tali joined	22:02
	→ impulze joined	22:02
	→ seb42 joined	22:02
	→ darjeeling joined	22:02
	→ madduck joined	22:02
	→ avtobiff joined	22:02
	→ janl joined	22:02
ocornu	but git-stash only works when there are local changes, right?	22:03
Ilari	ocornu: Local uncommitted changes.	22:03
ocornu	yes	22:04
	i have none	22:04
	i have two clean branches	22:04
	→ leanucci joined	22:04
ocornu	i want to switch to one branch while keeping the working tree state of the second one	22:04
jimmyz2	sweet, thanks guys..	22:05
qrush	then just do git checkout	22:05
	git checkout branch2	22:05
ocornu	qrush: that updates the working tree	22:05
qrush	Well, of course it does.	22:05
Ilari	ocornu: If you REALLY want to do that: 'git symbolic-ref refs/heads/branch2'.	22:05
justatheory	What is a "dirty working directory"?	22:05
Ilari	ocornu: If you REALLY want to do that: 'git symbolic-ref HEAD refs/heads/branch2'.	22:05
qrush	so you want to look at the changes in another branch?	22:05
	because that's what git log, git show, etc are for	22:06
	→ jerbear joined	22:06
Ilari	ocornu: Yeah, 'git log' accepts commit IDs and branches to show the log of.	22:06
	justatheory: Means uncommitted changes.	22:07
	→ hugoxrosa joined	22:07
	→ kraymer joined	22:07
	← minciue left	22:07
justatheory	Ilari: Strange, I didn't have any.	22:08
	But realized that I had a typo in my filter-branch command, and now I don't get that error.	22:08
	So I'm happy.	22:08
Ilari	justatheory: 'git status' shows which files have uncommitted changes.	22:08
justatheory	Ilari: I know. I had none.	22:08
	But it was a typo somehow.	22:08
	← Predominant left	22:09
ocornu	what i've done is that i committed a big change in one new branch (i probably shoud have stashed it instead), now i want to apply/commit piece after piece this big change in a clean way	22:09
	← catalis left	22:09
Ilari	justatheory: And as hint, if you can get away using --index-filter instead of --tree-filter, use --index-filter, as it is by far faster.	22:09
justatheory	I'm using --msg-filter	22:09
thiago_home	index-filter faster?	22:09
justatheory	Stripping out git-svn-id: stuff	22:09
thiago_home	ah, checking out	22:10
	I see	22:10
Ilari	ocornu: Check out the branch, 'cherry-pick --no-commit' the big change, 'git reset --mixed' and start applying?	22:11
	← kalle_ left	22:11
	→ kalle_ joined	22:11
justatheory	Can I use more than one filter command in a call to filter-branch?	22:12
	like --msg-filter and --tag-name-filter cat?	22:12
ocornu	Ilari: well, i reset it the "big change commit", stashed it, and applied it where it belongs. it works	22:12
	thanks for your help :)	22:12
Ilari	justatheory: At least according to syntax, multiple are allowed. Don't know how it would react to some combos...	22:12
justatheory	Well, I want to filter all messages in all branches	22:13
	And have it update what's in grafts	22:13
	Maybe I need two calls?	22:13
	← seangrove left	22:13
	→ foutrelis joined	22:13
Ilari	justatheory: Some examples of suspicous combinations include stuff like using both --tree-filter and --index-filter in one call,	22:13
justatheory	sure	22:14
Ilari	justatheory: --tag-name-filter can be used with other filters.	22:14
justatheory	does --msg-filter hit commits in all branches, or just the current branch?	22:14
	I'm guessing only if I also use `--tag-name-filter cat`.	22:15
	← Kimmie left	22:15
Ilari	justatheory: Yes it does, but without --tag-name-filter, tags aren't updated (or something to that effect)	22:15
	justatheory: Perhaps --all to rewrite everything?	22:16
	← hax0r1337 left	22:16
justatheory	Yeah, that gave me errors.	22:16
Ilari	justatheory: What errors? "gave me errors" is not helpful.	22:16
justatheory	Ilari: http://rafb.net/p/YU0fnh11.html	22:17
	I know, i was working on the paste.	22:17
	Looks like it rewrote all the branches, though.	22:17
	grafts too, I think.	22:17
	Maybe I don't need the --all?	22:17
Ilari	justatheory: Those errors came from lack of --tag-name-filter.	22:18
	justatheory: Or actually warnings. They aren't errors.	22:18
justatheory	right, that's why I was wondering if I even needed it.	22:18
	right, sorry, my mistake	22:18
Ilari	justatheory: It even mentions it.	22:18
justatheory	I know.	22:18
	It's a good warning	22:18
	→ iratik joined	22:18
justatheory	I'm just trying to decide whether or not I actually need to filter tags since I'm filtering mesages.	22:18
Ilari	justatheory: Tags are not on any branch!	22:19
	justatheory: So if you don't filter them, they retain their old commit messages.	22:19
justatheory	Tags are just applied to particular commits, no?	22:19
Ilari	justatheory: Yes.	22:19
justatheory	Ah, I see.	22:19
	So a tag can be a commit in itself, or applied to an existing commit.	22:20
	So I'll throw in the tags stuff, just to be sure everything is covered.	22:20
	git filter-branch --msg-filter 'perl -0777 -pe "s/\r?\ngit-svn-id:.+\n//ms"' --tag-name-filter cat -- --all	22:20
	← wmoxam left	22:21
	← jerbear left	22:21
	justatheory tries that.	22:21
Ilari	justatheory: The commit is designated by its ID. And still no way to create two different commits with the same ID with feasible effort is known.	22:21
justatheory	I knew that.	22:21
Ilari	justatheory: Note that filter-branch doesn't like to run if backups from old run are still present.	22:21
justatheory	I was thinking of CVS, I guess, where you can tag existing commits.	22:21
	Ilari: Yeah, I keep rm -rf'ing and cp -rf from backup as I figure stuff out.	22:22
	Thanks	22:22
	← nekrad left	22:22
	← botanicus_ left	22:23
amystrat	Ilari: fixed the error: hooks/post-receive exited with error code 1 error... it was failing because we have an extra bit of script that pings hudson our CI server... which was down	22:23
justatheory	Bah, getting dirty working directory error again…	22:23
Ilari	amystrat: Yeah, extra bit of code being the problem...	22:23
amystrat	:)	22:23
Ilari	justatheory: Did you run status last time after you got that error?	22:24
justatheory	yeah	22:24
Ilari	justatheory: run status and retry the command.	22:24
justatheory	it was a typo in my filter-branch script again	22:24
Ilari	justatheory: What kind of typo?	22:24
justatheory	Ilari: git filter-branch --msg-filter 'perl -0777 -pe "s/\r?\ngit-svn-id:.+\n//ms"'--tag-name-filter cat -- --all	22:25
	See how "--tag-name-filter" has no space before it?	22:25
	← kraymer left	22:25
	← amystrat left	22:25
Ilari	justatheory: Actually, you should run 'git status' (or 'git update-index --refresh') on copy after cp'ing repos.	22:26
justatheory	oh? Huh.	22:26
	I'm about finished with my migration project, and will soon be blowing all these things away.	22:27
	→ Tricon joined	22:27
	← Tricon left	22:27
justatheory	For those interested, I've been blogging my migration project here: http://justatheory.com/computers/vcs/git/	22:27
	→ seangrove joined	22:28
	← HG` left	22:28
Ilari	justatheory: Otherwise some cache data will be wrong, and it could cause problems. Nowadays it hides effects from diff, but it still could cause trouble with other commands (especially dirty worktree detection).	22:28
	*Nowadays git hides such effects from diff...	22:29
justatheory	Ilari: Good to know, thanks.	22:29
Ilari	justatheory: Wonder why that typoed command causes dirty working tree error?	22:29
	← kold left	22:30
justatheory	Ilari: Well, I had just cp -rf'd the directory from elsewhere, and it did start to work after I ran `git status`, as you said.	22:30
	→ fynn joined	22:30
	→ scarabx joined	22:30
justatheory	After `git-status`, I ran it again with the space missing and got:	22:30
	fatal: bad revision 'cat'	22:30
Ilari	justatheory: Yeah, I got the same error.	22:31
	→ iratiku joined	22:31
justatheory	So I think that it's exactly the issue you're describing.	22:31
	the dirty directory error just comes earlier	22:31
	and I didn't know that `git status` would fix the issue; I thought it was informational-only.	22:31
	← juanjoc left	22:32
Ilari	justatheory: And BTW, when dealing with SVN, perhaps 'tr _ -' would be better tag name filter than cat... At least when used once.	22:32
	← dodo_the_last left	22:32
fynn	how do I revert the working tree to match the index?	22:32
	← qrush left	22:33
Ilari	fynn: 'git checkout .'?	22:33
justatheory	Ilari: Why is that?	22:33
	I've renamed all of the tags by this point.	22:33
	or, rather, created new ones and killed off SVN's old ones.	22:33
fynn	Ilari: hm, I thought that would revert to HEAD	22:34
	anyways, thanks :)	22:34
Ilari	justatheory: Git can have '.' or '-' in tag names, so using those is more "Git style".	22:34
	← chris2 left	22:35
justatheory	I've changed the tag names from "rel_1_8_3" in svn to "v1.8.3" in git.	22:35
Ilari	justatheory: If you use dashes, be careful not to use stuff like em dashes. Git accepts those, but others are likely in for a nasty surprise.	22:35
	justatheory: Well, that works too.	22:35
justatheory	yeah, I'll stick to ASCII in tags and branch names.	22:36
	← iratik left	22:36
justatheory	Kept all the branched named things like "rev_1_10".	22:36
Ilari	justatheory: I once when testing did branch that had astral character in its name. Would be real fun to use.	22:36
justatheory	lol	22:36
	I love Unicode, but not so much in filee names. Or tags or branches.	22:37
	But look for smart quotes in my commit logs!	22:37
	Will have to watch for that when I port SVN::Notify…	22:37
Ilari	justatheory: Well, shouldn't merges be marked properly so you wouldn't need those tags anymore?	22:37
justatheory	Yeah, The merge tags are deleted.	22:38
	I kept release tags only.	22:38
	I'll get this on github in the next day or two and you can see how it ended up.	22:38
	← plasticine left	22:38
	← seangrove left	22:39
	← MisterN left	22:40
	← km2 left	22:41
	← jrmuizel left	22:42
	→ haesbaert joined	22:43
	→ Mr_Grieves\| joined	22:44
	← wwwald left	22:46
	← boto left	22:47
Mr_Grieves\|	I'm having trouble figuring out how to use branches. I have two branches, "master" and "branch".	22:47
	I'd like to put the changes in master to branch, but not the other way around.	22:47
jimmyz2	in case anyone is interested, i solved my php parsing check pre commit issue with this:	22:47
RandalSchwartz	a branch is an endpoint, not a line	22:48
jimmyz2	http://phpadvent.org/2008/dont-commit-that-error-by-travis-swicegood	22:48
RandalSchwartz	that's the first thing to understand	22:48
Ilari	justatheory: Characters unconditionally valid in git branch/tag names: !"#$%&'()+,-0123456789;<=>ABCDEFGHIJKLMNOPQRSTUVWXYZ\]_`abdefghijmnpqrstuvwxyz\|} characters conditionally valid: ./@lock{	22:48
RandalSchwartz	do master and branch have a common ancestor?	22:48
Ilari	justatheory: And all characters with bit 7 set are unconditionally valid.	22:48
	← Sho_ left	22:48
Mr_Grieves\|	RandalSchwartz -- yes, "branch" branched off of "master" a while ago.	22:49
RandalSchwartz	and so master now looks like A B C X Y Z	22:49
	and branch looks like A B C D E F	22:49
	← cmarcelo left	22:49
RandalSchwartz	what do you want to do?	22:49
	← leanucci left	22:49
RandalSchwartz	what do you want branch to look like?	22:49
Mr_Grieves\|	I want to make branch look like A B C D E F X Y Z, and leave master as it is	22:50
RandalSchwartz	well, it won't be X Y Z	22:50
Ilari	Mr_Grieves\|: checkout branch and merge master to it (but don't merge the other way, it would take changes from branch)?	22:50
	← radarek left	22:50
RandalSchwartz	since X is a child of C not F	22:50
	do you want A B C D E F X' Y' Z'	22:50
	hoping that X' Y' and Z' don't cause any conflicts?	22:50
	it'd be better to do:	22:51
	→ Predominant joined	22:51
	← ckm left	22:51
Ilari	Mr_Grieves\|: That makes history of branch look like A <- B <- C <- {D <- E <- F, X <- Y <- Z} <- M	22:51
RandalSchwartz	A B C X Y Z D' E' F'	22:51
	where your local changes are rebased ontop of the new master	22:51
	← Ryback_ left	22:51
RandalSchwartz	that'd be:	22:51
	git checkout branch	22:51
	git rebase master	22:51
	RandalSchwartz waits for Mr_Grieves\|	22:52
Mr_Grieves\|	Hang on, phone call :)	22:52
	→ qrush joined	22:52
	← hcl2 left	22:53
Mr_Grieves\|	Hrm, it sounds like I'm going about this wrong. I want to be able to work on multiple bugs/features at once, and use master to stay up to date with other peoples changes.	22:54
	← hyperair left	22:54
Ilari	Mr_Grieves\|: Note that if 'branch' has been pushed somewhere (or it has been fetched/pulled), rebasing would lead to errors on push (overriding those would cause problems for others).	22:54
	Mr_Grieves\|: Topic branches?	22:54
Mr_Grieves\|	Ilari -- That's what I'm worried about.	22:54
	← hpoydar left	22:55
Ilari	Mr_Grieves\|: Does that branch even exist in upstream?	22:55
Mr_Grieves\|	No, it is local only.	22:55
RandalSchwartz	keep rebasing your local changes on the remote fetches then	22:55
Mr_Grieves\|	Topic branches sound like what I'm looking for.	22:56
	CardinalNumber → ProperNoun	22:56
RandalSchwartz	that makes a linear history that will merge cleanly	22:56
Ilari	History procuced by merge will merge just as cleanly.	22:56
Mr_Grieves\|	RandalSchwartz -- I'm wary of rebasing, I've screwed that up too many times :)	22:56
RandalSchwartz	learn what it's doing first	22:56
	that's the key	22:56
Mr_Grieves\|	Indeed. I tend to be lazy about those things :)	22:57
RandalSchwartz	if the proper workflow requires you to learn, I suggest you learn. :)	22:57
	→ gdelfino joined	22:58
Ilari	Mr_Grieves\|: The rules are: Rebase either 1) Branches that do not exist anywhere else or 2) Branches on top of their only upstream.	22:58
mugwump	and only then if you didn't invest significant effort in testing the combined tree that might be invalidated by the rebase	22:59
Ilari	Mr_Grieves\|: And for high-quality work, you need BOTH merge and rebase and to know when to merge and when to rebase.	22:59
	→ damainman856 joined	23:00
	Mr_Grieves\| goes off to read some more docs :)	23:00
Ilari	Mr_Grieves\|: Another more subtle problem with rebase is that it can result many new states to test (if you care about testing at that level), whereas merge will only result one.	23:00
	← schacon left	23:00
	Modius_ → Modius	23:01
Ilari	Mr_Grieves\|: The check push does is that the version in target repo is in set of commits within branch to be pushed. If it is, push doesn't need forcing, otherwise it does.	23:02
	→ snowman856 joined	23:02
Mr_Grieves\|	I see -- Thanks for the tips. It'll probably take a bit of practice to get it to stick for me :)	23:03
	← hugoxrosa left	23:03
Ilari	Mr_Grieves\|: That is, if pushed branch includes tip of target, it updates without forcing.	23:03
	Mr_Grieves\|: The rule of thumb with rebasing is not to let it rebase any commits that exist anywhere else. And note that publishing patches out of commits doesn't count as publishing commits.	23:05
	← foutrelis left	23:05
	← ceej left	23:06
Mr_Grieves\|	Ok :)	23:06
	→ Ademan joined	23:06
	→ JohnnyL joined	23:07
	← JohnnyL left	23:07
	← LiamH left	23:07
	← Alinermedia left	23:08
justatheory	Ilari: I think you listed most of 7-bit ASCII there. :-)	23:08
	← Dashkal left	23:08
	→ shruggar joined	23:10
Ilari	justatheory: It is most of graphical characters of 7-bit ASCII. The unconditionally banned ones (out of ASCII graphical range of 32-126) are <space>:*?[^~	23:11
	← kalle_ left	23:11
Ilari	justatheory: So out of 95 standard ASCII characters, its 80 unconditionally valid, 8 conditionally valid and 7 unconditionally invalid.	23:12
	→ qrush_ joined	23:12
Ilari	justatheory: Lower case alphabet chacarters 'l', 'o', 'c' and 'k' are only conditionally valid. This comes from combination '.lock', which can't be at the end of branch/tag name.	23:13
	justatheory: So 'foo-lock' and 'foo.locks' is valid branch name, but 'foo.lock' isn't.	23:14
	*are	23:14
	→ qrush__ joined	23:15
justatheory	Interesting.	23:16
poseidon	Can someone help me figure out what I'm doing wrong here? http://rafb.net/p/e4t0rY19.html	23:16
justatheory	Not sure I need to learn what locks are just yet.	23:16
	← qrush_ left	23:16
Ilari	justatheory: .lock files are internally used by git. Branches with '.lock' suffix could collide with those.	23:16
	← qrush__ left	23:16
	→ kalle_ joined	23:17
justatheory	ah	23:17
Ilari	poseidon: You didn't commit before pushing.	23:17
	→ botanicus joined	23:17
	← javatexan left	23:17
poseidon	Ilari, but I added them. Do I have to add new files and commit them?	23:17
robinr	so we should use an otherwise invalud name for lock-files instead	23:18
Ilari	poseidon: Push only pushes commits, not files.	23:18
	poseidon: So you need to make commit to push first.	23:18
	← fridim_ left	23:19
Ilari	robinr: Current rules are not strict enough. You can cause problems with stuff like 'foo.lock/bar'.	23:19
poseidon	k, but for when I make new files, can I simply do git commit newfile or do I have to do add then commit?	23:19
	← damainman856 left	23:19
Ilari	robinr: Put 'foo' along with that...	23:19
Fissure	poseidon: you have to add first	23:19
	committing a file only works when it's already tracked	23:19
Ilari	poseidon: 'git commit newfile' doesn't work, but 'git commit oldfile' does....	23:20
	← snowman856 left	23:20
	← mfilipe_ left	23:20
dotsintacks	I need to track down the last time a line number changed, but it has gone through several renames.	23:21
Ilari	poseidon: IIRC, marking file as intent-to-add is enough to make 'git commit file' work.	23:21
gdelfino	Anyone knows gitosis required version of python? I have 2.4.3 and I'm having problems with "sudo -H -u git gitosis-init" under CentOS 5.3	23:21
dotsintacks	s/line number/line/	23:21
robinr	dotsintacks: blame -M or -C -C	23:21
	could could also add -w for ignoring whitespace changes	23:22
justatheory	Ilari: I suppose that everything you just told me is nicely documented somewhere, yes?	23:23
Ilari	justatheory: man git-check-ref-format	23:23
Gitbot	justatheory: the git-check-ref-format manpage can be found at http://git.or.cz/man/git-check-ref-format	23:23
	← botanicus left	23:23
dotsintacks	robinr: but doesn't git blame still just annotate each line with the last change?	23:23
	so that rename would be the last change.	23:23
Ilari	justatheory: Doesn't document stuff like 'foo.lock/bar' and 'foo' collding.	23:23
justatheory	Perfect	23:23
Ilari	*colliding	23:23
dotsintacks	unless -C /ignores/ renames	23:23
justatheory	Patch it?	23:23
Ilari	dotsintacks: Git few versions ago (1.6.2.1) had only ./ as conditionally valid, and @lock{ was in unconditionally valid list. Of course, one still could get into trouble with those (which was why those characters where made conditionally valid).	23:26
	→ loincloth joined	23:26
Ilari	Oops.	23:26
	justatheory: ^^	23:26
	← parasti left	23:27
Ilari	justatheory: '@' is only "dangerous" if immediately followed by '{' later followed by '}'. Its invalid in pair '@{'.	23:28
robinr	non-ascii could be fun too	23:28
justatheory	Ilari: I am unlikely ever to run into this stuff, because I've been trained to keep tags and branches simply named by CVS and Subversion. But a discussion like this in the docs would definitely be beneficial to all.	23:28
	I need not be the only recipient of your wisdom. ;-)	23:28
	← rettub_ left	23:30
Ilari	Of course, by meddling with repository internals, its possible to create branch names that include unconditionally invalid characters (and thus are invalid). On at least some Git versions, such names cause overt failure of some commands, and I wouldn't be surprised if some commands misbehaved on those.	23:30
	→ rettub_ joined	23:30
Ilari	E.g. some git version couldn't rename such branch.	23:30
	→ Dashkal joined	23:31
	← zpoley1 left	23:31
	→ catalis joined	23:32
Ilari	robinr: At least such characters don't cause misbehaviour.	23:32
	← majeru left	23:32
	→ majeru joined	23:32
	← [intra]lanman left	23:32
	← iso42 left	23:32
	← SRabbelier left	23:32
	→ nis joined	23:32
Ilari	OTOH, some of the chararacters in unconditionally invalid list are only there for shell safety (*?[).	23:33
	→ scientes joined	23:35
	← aziz left	23:36
robinr	. is for avoiding collision with the commit...range syntax	23:38
	← sekimura left	23:39
robinr	or was then end of the sentence?	23:39
	:)	23:39
quentusrex	I have a git repo on a remote server,	23:39
	← jimmyz2 left	23:40
quentusrex	how do I get the repo to have a copy of the files in the directory, and update the files on a git push?	23:40
	→ jrmuizel joined	23:40
robinr	anway, good night	23:40
	→ sekimura joined	23:40
	← Zathrus left	23:41
quentusrex	Ilari: any idea how to have the files 'checked out' in a git repo? and keep the server repo updated for each push?	23:41
Ilari	. is because of range syntax and '.lock', '/' is because it gets translated to directory seperator (and that has its own rules), '@' is because of reflog syntax, 'l', 'o', 'c', 'k' are because of '.lock' suffix, '{' is because of reflog syntax, <space> is because space is used as separator in many places, ':', '^' and '~' are for ref syntax, '*', '?' and '[' are for shell safety.	23:41
	quentusrex: post-update hook.	23:42
	quentusrex: man githooks	23:42
Gitbot	quentusrex: the githooks manpage can be found at http://git.or.cz/man/githooks	23:42
quentusrex	right, but what is the proper way to check out the repo into the repo directory?	23:42
Ilari	quentusrex: Unset GIT_DIR, move one directory up and 'git reset --hard'?	23:43
	quentusrex: Alternatively set GIT_WORK_DIR and 'git reset --hard'.	23:44
quentusrex	umm, I think I've not been clear. I have on my workstation a folder called apache-settings/ it is now a git repo( I did git init) I created the files, and committed. I added a remote repo, and pushed the local to the remote repo. But the remote repo doesn't have the files in the directory like mine does on my workstation.	23:44
	I'm trying to get the remote repo have teh files, and in the post-update hook, update the copies of the files,	23:45
Ilari	quentusrex: $faq non-bare	23:45
Gitbot	quentusrex: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare	23:45
quentusrex	then how do I check it out in another spot?	23:46
Ilari	quentusrex: I gave two ways of doing the update in post-update. But beware, both make the repository unsuitable for local changes.	23:46
	← scarabx left	23:46
quentusrex	Ilari: I don't want local changes at alll :)	23:47
	atleast no changes to be made on the server directly.	23:47
	→ Flathead joined	23:48
	→ nealinux joined	23:48
Ilari	quentusrex: As said, in post-update hook, either do 'unset GIT_DIR', 'cd ..', 'git reset --hard' or 'export GIT_WORK_DIR=/path/to/tree' 'git reset --hard'.	23:49
quentusrex	Ilari: since I'm new to git (or atleast I feel that way today) do you mind walking me through this?	23:49
	← netoman left	23:49
Ilari	quentusrex: One problem doing just that it does it even for push to other branches, slowing things down.	23:49
quentusrex	I have a repo exclusively for the settings	23:50
	one repo for each 'type' of settings.	23:50
	← mw left	23:50
	→ mw joined	23:50
Ilari	quentusrex: And then there's some setting you need to set, or you get nasty warnings in newest versions (and it will start failing in 1.7)...	23:50
quentusrex	where are the logs?	23:51
Ilari	quentusrex: What logs?	23:51
quentusrex	git logs.	23:51
Ilari	quentusrex: Reflogs are stored in .git/logs/	23:51
quentusrex	before I do the git reset --hard	23:52
	don't I need to check out the repo first?	23:52
	cpg\|away → cpg	23:52
Ilari	quentusrex: It is implictly already checked out (with wrong contents of course).	23:52
quentusrex	it says it is a bare repo	23:53
	on the server	23:53
	I'm pushing to the server repo	23:54
	and the server is a bare repo	23:54
Ilari	quentusrex: 'git init --bare'? Or you forgot to unset GIT_DIR?	23:54
	quentusrex: Plus move one level up before reset.	23:54
	← merlin83 left	23:55
Ilari	quentusrex: Can you pastebin the error message?	23:55
	→ arohner joined	23:57
	← Modius left	23:58
	← nealinux left	23:58
	→ Modius joined	23:58

Logs Search ←Prev date Next date→

Channels Documentation