IRCloggy #git 2010-11-29

Provider of IRC logs since 2005.

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

	2010-11-29
	← cordoval_ left	00:03
	← banisterfiend left	00:04
	→ banisterfiend joined	00:07
	← shiba_yu36 left	00:07
	→ chrischambers joined	00:09
	← avena left	00:09
	→ ArseneRe1 joined	00:10
banisterfiend	man git-cherry	00:10
jast	the 'git-cherry' manpage can be found at http://git.or.cz/man/git-cherry [automatic message]	00:10
theoros	you know you can type these into your terminal.	00:11
frogonwheels	banisterfiend: or use git help cherry to be a little more platform agnostic	00:12
	banisterfiend: errm git help cherry-pick	00:12
theoros	cherry is its own command	00:12
frogonwheels	theoros: (man git-cherry-pick isn't quite so useful in windoze )	00:13
jast	theoros: the 'git-cherry-pick' manpage can be found at http://git.or.cz/man/git-cherry-pick [automatic message]	00:13
frogonwheels	theoros: oh!	00:13
	← foocraft left	00:13
banisterfiend	im trying to apply a patch, but i get this error: C:\john\ruby\gitlearn>git am 0001-added-love.patch	00:14
	previous rebase directory /c/john/ruby/gitlearn/.git/rebase-apply still exists b	00:14
	ut mbox given.	00:14
	what's does it mean?	00:14
frogonwheels	theoros: damn, I'd missed that one!	00:14
	← cylence left	00:14
banisterfiend	theoros: im on windows :/ im relying on the man pages on the web :)	00:14
frogonwheels	banisterfiend: you're in the middle of a rebase-like command still	00:15
	banisterfiend: git help <cmd>	00:15
theoros	__git_ps1 is so useful, i don't know if you can get it on windows	00:15
banisterfiend	frogonwheels: but im not in the middle of a rebase like command, why does it think i am, and wtf do i do to break out of it?	00:15
frogonwheels	banisterfiend: (I'm mostly on windows as well, it should show them in your favourite browser)	00:15
	← banisterfiend left	00:15
	← ArseneRe1 left	00:15
thiago_home	theoros: sure you can. Just use the shell __git_ps1 was written for.	00:15
frogonwheels	banisterfiend: remove the directory	00:15
thiago_home	theoros: it can be as useful or useless as it is on Linux	00:15
	theoros: __git_ps1 is very useless for me, for example.	00:16
theoros	thiago_home: yes well i wasn't commenting on whether it's universally useful. i was saying i don't know if you can get it on windows.	00:16
thiago_home	and I answered	00:16
theoros	however i find it quite useful to know when HEAD is detached	00:16
thiago_home	just use the shell it was written for	00:16
	if you go down to details, you're asking "how can I use this bash script" and the answer is "in bash"	00:18
abstrakt	what's the latest version of gitolite	00:18
frogonwheels	theoros: if you're using cmd (which I do) you can't really do anything like that... though I use gvim mostly, which helps a lot	00:18
abstrakt	I don't see a version number on the github page	00:18
	→ Yuffster joined	00:18
frogonwheels	abstrakt: 2bbcc8239cda81a5a0a0 ;)	00:19
theoros	i haven't used windows in a while, i wasn't suggesting it wasn't possible to get an equivalent or even to get it working, not sure where all these comments are coming from	00:19
abstrakt	frogonwheels, lol, ok... I guess I'm wondering what "version" that ubuntu 10.10 has	00:19
frogonwheels	abstrakt: also look at Commits\|Switch Tags the dropdown contains all the tagged commits	00:19
abstrakt	because ubuntu 10.10 has one	00:19
frogonwheels	so 1.5.7 looks like the latest	00:20
abstrakt	frogonwheels, ah ok cool thanks	00:20
frogonwheels	abstrakt: oh yay! finally.	00:20
	→ banisterfiend joined	00:20
frogonwheels	abstrakt: anyway - just clone it, not sure there's much of an advantage using a package ;)	00:20
banisterfiend	also how come reflog doesnt display everything? it doesnt show my 'initial commit' anymore for exmaple. How do i get reflog to show everything?	00:20
abstrakt	hmm, ubuntu has 1.5.4... wonder how much has changed between .4 and .7	00:20
	frogonwheels, ok, might do that	00:21
	frogonwheels, the "advantage" is that it's "easy"	00:21
frogonwheels	banisterfiend: reflog has a time limit	00:21
abstrakt	frogonwheels, the other advantage is that it's easy to remove	00:21
banisterfiend	frogonwheels: is there permanent version of reflog though? or a way to get it to show everything?	00:21
	it's very useful	00:22
frogonwheels	banisterfiend: don't use reflog. just use log	00:22
abstrakt	generally I'm under the impression that it's best to build packages for things	00:22
	→ gusnan joined	00:22
abstrakt	I'm just a bit loathe to really read the "ubuntu packaging guide"	00:22
banisterfiend	frogonwheels: what if im trying to access an orphaned treeish though	00:22
	→ rchavik joined	00:22
frogonwheels	abstrakt: yeah - though gitolite works by installing into a specific user though .. so kinda different.	00:22
banisterfiend	frogonwheels: will that sitll appear in logs? or only in reflog?	00:22
frogonwheels	banisterfiend: once the reflog has released it, it's up for garbage collection	00:23
	banisterfiend: so it may, in fact, have disappeared properly	00:23
banisterfiend	frogonwheels: so once the reflog has released it there's no other way i can find it?	00:23
	frogonwheels: hwat is the timelimit for reflog	00:23
frogonwheels	banisterfiend: look at the help	00:23
	→ tbf joined	00:23
	→ bitkiller joined	00:24
banisterfiend	man git-reflog	00:24
jast	the 'git-reflog' manpage can be found at http://git.or.cz/man/git-reflog [automatic message]	00:24
	← p3rg2z left	00:26
	→ Titosemi joined	00:27
	→ orafu joined	00:30
	→ meatflag_ joined	00:30
	← SianaGearz left	00:30
	← rolfb left	00:31
	→ Wyverald joined	00:32
	→ aresnick joined	00:33
	→ xiangfu joined	00:34
	→ spdr joined	00:34
	← meatflag_ left	00:34
	→ allaire joined	00:34
	← spdr left	00:34
allaire	hey, if I want to unstange all the staged files, should I do "git reset HEAD" or "git reset HEAD . "	00:35
	→ meatflag_ joined	00:38
	→ shiba_yu36 joined	00:38
	→ welly joined	00:40
	← banisterfiend left	00:42
	← Buglouse left	00:42
	→ Hasnf joined	00:44
	← Tuller left	00:44
	→ Kobaz joined	00:46
Kobaz	# Your branch is ahead of 'origin/master' by 3 commits.	00:46
	how do i see what commits those are?	00:46
	→ MikhX joined	00:46
abstrakt	does anyone have any idea why I might not be able to log in with pub/priv keys?	00:47
	I can log in with my password no problem	00:47
	I've tried using ssh-copy-id	00:47
allaire	Kobaz: git status?	00:47
abstrakt	there used to be an .ssh folder, but I moved that to ssh.bak and recreated my .ssh folder and then ran ssh-copy-id	00:47
	I looked in the authorized_keys file, the key is there, still asks for my password	00:47
	looks like the right key	00:47
Kobaz	allaire: that just shows you what stuff is modified etc	00:48
abstrakt	originally the .ssh folder and the authorized keys file were chmod 777 for some messed up reason	00:48
Kobaz	i would like to know which specific commit ids are missing from origin/master	00:48
abstrakt	so I did a chmod 755 to .ssh and I did a chmod 600 to authorized_keys, still asks for the password, I tried chmod 644 on authorized_keys, still asks for the password	00:48
	→ banisterfiend joined	00:50
Kobaz	allaire: ideally it should show which three commits are not present	00:50
	→ pfifo joined	00:51
abstrakt	whoops. probably wrong channel :)	00:51
pfifo	hi	00:51
allaire	Kobaz: git diff origin/master	00:52
Kobaz	ah okay... is there a short version, that will just show commit ids and not text diff	00:52
mase_wk	Morn all. I am trying to remove a directory and contents from the a git repo. I have performed a brand new clone of the a git repo then a git filter-branch --tree-filter 'rm -rf locale' HEAD and then tried to git push origin --force	00:52
	however i'm getting a ! [remote rejected] translation -> translation (non-fast-forward)	00:53
	i don't understand why this isn't allowing me to push because a) i'm passing the --force option b) no one else has pushed to that repo since	00:53
pfifo	when i try to run git, jim starts sucking cock and interferes with gits ability to track things	00:54
	← tbf left	00:54
pfifo	wat do?	00:54
mase_wk	pfifo: i think that's a discussion between you and jim	00:54
	← thiago_home left	00:54
mase_wk	and is not meant for this channel	00:55
pfifo	iwas 14 once, i can do it againg	00:55
	← Hasnf left	00:55
pfifo	what should i do when jim stars sucking dick to make git work correctly	00:55
	?!?!?!!?	00:55
	← SGulseth left	00:56
	pfifo will enjoy trolling, but would prefer to get actual help regading issues	00:56
	pfifo has realized thaat actual help is not posible and has decided to troll for entertainment instead	00:57
pfifo	sssssso	00:57
	frogonwheels wonders if pfifo realises that this channel is logged.	00:58
pfifo	i have my onw logs, i will willingly tell anyone	00:58
	...	00:58
mase_wk	Anyhoo, do I need to filter each branch before i can push ?	00:58
pfifo	you areall elitests and like to make fun of new comers	00:59
mase_wk	i am really only wanting to remove the locale directory from the one branch	00:59
	→ xtagon joined	00:59
allaire	hey, if I want to unstange all the staged files, should I do "git reset HEAD" or "git reset HEAD . "	00:59
pfifo	mase_wk, thee guys will lead you down the wrong path, corrupt your git repo and than make fun of you	00:59
	← rgrig left	01:00
frogonwheels	mase_wk: you want to look at where arguments go relative to repository & refspec in man git-push	01:00
jast	mase_wk: the 'git-push' manpage can be found at http://git.or.cz/man/git-push [automatic message]	01:00
mase_wk	pfifo: i'll take my chances	01:00
	← shiba_yu36 left	01:01
pfifo	a i did, i always considered myself an upsstanding member of free node but thi channel made me reconsider	01:01
	← rchavik left	01:01
	→ shiba_yu36 joined	01:01
	→ rchavik joined	01:01
pfifo	despite all the +modes you can give to a chan, they welcomed anyone and if you ask questions or try to increase your knoldege about omething the have a broblem	01:02
	→ sh1mmer joined	01:03
	→ Buglouse joined	01:03
	→ sako joined	01:03
pfifo	so, finding nothing but insults in this chan	01:03
	← darkredandyellow left	01:04
mase_wk	frogonwheels: perhaps i'm not understanding you correctly or the git terminology. Are you saying the --force option is in the wrong spot. I have looked at the man page which lists the -f before the repository and refspec . I tried so git push origin -f translation:translation	01:04
pfifo	i have decided to ... trolll? no, im better than that	01:04
mase_wk	but it too failed to push	01:04
pfifo	#git is my laugh factory	01:04
	bremner wonders if pfifo knows about /ignore	01:05
pfifo	bremner, i ignored jim... howerver i think you are completly underestumating it	01:06
	→ smik`` joined	01:07
pfifo	no one here in this room will actually HELP someone	01:07
frogonwheels	pfifo: actually they do all the time.	01:07
pfifo	th\is chat room is a compititin between elitist to determin who can make who cry	01:08
frogonwheels	mase_wk: forced pushes can be disallowed at the server	01:08
pfifo	fragkjrehor. ,.zwan ,dzzsa	01:08
	ed	01:08
	fd	01:08
	fdf	01:08
	df	01:08
	fd	01:08
	fd	01:08
frogonwheels	pfifo: not in my experience.	01:08
pfifo	dfd	01:08
	fd	01:08
	df	01:08
	d	01:08
	dv	01:08
	f	01:09
	wtf?	01:09
	frgnwhl	01:09
	frgnw	01:09
	i cant highlight you	01:09
	← __name__ left	01:09
pfifo	wyh	01:09
	why	01:09
	why	01:09
Kobaz	help... dir... ls.... eat flaming death	01:09
pfifo	why	01:09
mase_wk	frogonwheels: denyNonFastforwards = true be the setting which disallows it ?	01:09
	→ ted- joined	01:10
	→ carutsu_ joined	01:10
	← ted- left	01:10
	← smik` left	01:10
frogonwheels	mase_wk: for man git-config that appears to be the case	01:10
jast	mase_wk: the 'git-config' manpage can be found at http://git.or.cz/man/git-config [automatic message]	01:10
pfifo	fr0ggie, your the closessssssssssst thing to the guy i am actually trying to talk to	01:10
	← aresnick left	01:10
pfifo	fr0ggie, your the closessssssssssst thing to the guy i am actually trying to talk to	01:10
	fr0ggie WAKE UP	01:11
	fr0ggie WAKE UP	01:11
	fr0ggie WAKE UP	01:11
	← icwiener left	01:11
pfifo	fr0ggie WAKE UP	01:11
	fr0ggie WAKE UP	01:11
	fr0ggie WAKE UP	01:11
	fr0ggie WAKE UP	01:11
	fr0ggie WAKE UP	01:11
	fr0ggie WAKE UP	01:11
	fr0ggie WAKE UP	01:11
mase_wk	frogonwheels: cool thank you. i would never have found that the option had been set.	01:11
frogonwheels	mase_wk: :) np	01:11
pfifo	frogonwheels, i your woman blonde or brunette?	01:12
	→ aresnick joined	01:12
pfifo	ohh no woman	01:13
	kk	01:13
	seroiusly	01:13
	comon	01:13
	lilo is not klining me, and no one is baning me	01:14
	lilo is not klining me, and no one is baning me	01:14
	← allaire left	01:15
andres	Well. There could be the problem of communicating with irc from the other side...	01:17
	← nadim_ left	01:17
	← gusnan left	01:18
	→ nd___ joined	01:18
	→ ReleaseX joined	01:21
	← aresnick left	01:23
	→ aresnick joined	01:24
	← bcardarella left	01:24
fr0ggie	pfifo: Stop	01:25
	← bitkiller left	01:25
fr0ggie	That nonsense	01:25
	← sivy left	01:28
	← itamarjp1 left	01:33
	← Titosemi left	01:34
mase_wk	frogonwheels: now that i've removed the locale directory done a forced push. what will happen to the locale directory when people who have cloned the repo next do a git pull ? Will it remove the locale directory from their repos ?	01:36
	→ itamarjp joined	01:36
frogonwheels	mase_wk: Forced pushes are always a bit yicky for downstream, and hence discouraged. you will need to get everyone to rebase any work they've done onto the updated branch	01:37
mase_wk	frogonwheels: what about if they don't have any work. i requested that everyone commit /push their changes before undertaking this.	01:39
	will a rebase remove the locale dir ?	01:40
	→ Scott___ joined	01:40
frogonwheels	mase_wk: well done. just get them to reset --hard their tracking branch onto the newly fetched branch	01:40
	← aresnick left	01:41
frogonwheels	mase_wk: I _believe_ the reset --hard will cause the files in the locale dir that were tracked to be removed	01:41
mase_wk	frogonwheels: you rock!	01:42
frogonwheels	mase_wk: (Any files that _are_ in the commit they are leaving, and are _not_ in the commit they are going to, should be removed)	01:42
mase_wk	that all worked well. thank you	01:42
	couldn't have done that without your help	01:42
frogonwheels	np mase_wk :)	01:42
	← Scott___ left	01:43
	→ baccenfutter joined	01:43
	frogonwheels enjoys helping people who are quick on the up-take.	01:43
jim	you mean people unlike John Allen's proverbial stone...	01:45
	→ kanru1 joined	01:45
jim	I'm given to understand I have an undying fan on the channel :)	01:45
	← sako left	01:46
	→ mithro joined	01:46
frogonwheels	jim: yep. No helping some people.	01:46
pfifo	:)	01:48
	← sh1mmer left	01:48
pfifo	no helping som people	01:48
	→ stliu joined	01:49
	← stliu left	01:49
	→ stliu joined	01:49
jim	well the most positive that can be said, is pfifo needs to be encouraged to do the work necessary to learn what he wants to use (speaking in 3rd person because as I understand it, he's /ignoring me)... if he doesn't, he has no choice but to either complain or not complain	01:49
pfifo	no i never /ignore	01:50
jim	ahh, so you only claim to :)	01:50
frogonwheels	jim: I think most people will be with you on this one. Especially after that little tirade.	01:50
pfifo	yeah well if somene is really asking for it, yeah	01:50
	← xiangfu left	01:50
jim	if you were to watch gitcasts and read man pages (they move -fast-) in between viewings, you might actually gain some ability	01:50
pfifo	jim stop	01:51
	please	01:51
	im a noobie	01:51
jim	I don't bear him any ill will either, he just needs to do his work	01:51
	→ Lgb joined	01:51
abstrakt	sitaram, so I'll be the first to admit I probably just don't know ssh well enough	01:51
jim	pfifo: have you ever stopped to consider the following question: what is the HIGHEST, BEST use of my time right now	01:52
abstrakt	sitaram, maybe you can suggest an alternative approach to accomplish my goals then, which are...	01:52
pfifo	damnit jim (ohh that felt so good) do you want me to read something	01:52
	→ sh1mmer joined	01:52
jim	pfifo: I have no idea what your issue is... so try gitcasts, try git from the bottom up, try the man pages	01:53
	← flaguy48 left	01:53
jim	and PRACTICE	01:53
abstrakt	sitaram, I have a VPS, I am playing with different setups, I've setup gitolite as its own user, works fine... I've setup redmine, basically works fine, thing is I would like the redmine user (aka the user under which httpd) is running, to have read access to the gitolite repositories	01:53
jim	if you do these things, you'll get something out of it.	01:53
	→ bcardarella joined	01:53
jim	if you don't, you won't.	01:53
	← bcardarella left	01:53
abstrakt	sitaram, I have a deploy user, I ran gl-setup for the gitolite user, no problems, ran gl-setup for the deploy user and the issue probably has something to do with ssh-add I'm guessing, because there are now the two different proper distinct keys in .ssh/authorized_keys and one of them is wrapped with the command= parameter but when I ssh I get asked for my password... I do have two keys living in ssh-add... ssh-add -L reports my .ssh/id_rsa.pub fil	01:55
	e and in addition it also reports the other one I created at .ssh/gitolite/admin.pub	01:55
	I added the latter key using	01:55
	ssh-add .ssh/gitolite/admin	01:55
pfifo	jim, to argue your statement, if IRCis costing you money, costing you hours at work, and making you relationship between you and youe bosses wife turn sou, the seriously leave. BUT its not, you seem tp enjoy doing nothing but making fun of newbies inbetween filling out captchas on /b/	01:55
abstrakt	still being asked for my password	01:55
	← dbbnj left	01:55
abstrakt	let me try using ~/.ssh/gitolite/admin as you recommend in your troubleshooting docs	01:55
sitaram	abstrakt: slow down	01:56
	abstrakt: first, you're supposed to run gl-setup only once per gitolite "hosting user"	01:56
pfifo	jim...	01:56
	im pissed at git, not you	01:56
abstrakt	sitaram, indeed, I did only run it once for the deploy user	01:56
	→ aresnick joined	01:56
sitaram	abstrakt: so -- lets be clear what a "gitolite hosting user" is. It's (a) a real unix user on the box , say "foo"(b) had a bunch of repos under gitolite control (c) which are accessed by the "virtual" users as "foo@server:reponame"	01:57
	ok os far?	01:57
	s/os/so	01:57
ice799	Hi. I have two branches, origin/master and origin/B. origin/B diverged from master a long time ago and many rebases have occurred. Merging the two branches is going to be painful. Is there some way I can just tell git to make origin/B the new master? Of course, people who have the code checked out will need to git pull -f, but I'm OK with that.	01:58
sitaram	if you only want the redmine user to have read-only access to the repos, you have 2 choices. You can set the umask to 0027 and make the redmine user have a supplementary group of git (usermod -G git redmine)	01:59
	abstrakt: ^^, or you can give the redmine user a pubkey and add that key to gitolite	01:59
jim	pfifo: I recognize this (I have no idea why you'd be pissed at me; all I'm doing is putting what you need to know right in your face; I can't imagine why that would piss anyone off... :)	01:59
	pfifo: consider this thought:	02:00
pfifo	ok	02:00
jim	each thought you have continues to run inside your head, producing results, until it is unthought, or rethought	02:01
	→ wliao joined	02:01
abstrakt	sitaram, well, the redmine user is www-data, I'm pretty sure	02:01
	sitaram, because I'm running redmine under passenger on apache	02:01
	not sure if that will work to try to add a priv key for www-data	02:01
	← mfacenet left	02:01
	→ boombatower joined	02:01
	← boombatower left	02:01
	→ boombatower joined	02:01
sitaram	so use unix perms then, I guess...	02:02
abstrakt	sitaram, actually though, I think I may have figured it out, I just made a key pair for my deploy user and added the pub key to gitolite-admin and now I can do checkouts by using git clone gitolite@localhost:repo-name	02:02
pfifo	jim, i understand, as a developer Something that claims todo the job of git is obivously well thought out	02:02
jim	I encourage you to run an internal inquiry to determine the thought or thoughts that are blocking you from understanding git	02:02
abstrakt	sitaram, and that works fine	02:02
sitaram	ok	02:02
abstrakt	sitaram, just looking at the fact that I'll probably have to git pull constantly in order to keep redmine up to date	02:02
	→ gretch joined	02:02
abstrakt	i dunno maybe not	02:02
	sitaram, I usually set up a user called "webdeveloper" and I put that user in the www-data group (the user under which apache httpd is running)	02:03
jim	a -BIG- part of the reason we have git now, is Linus must have been -super- pissed at Larry McVoy for trying to trap him into continuing to use bitkeeper	02:04
abstrakt	and then I chown webdev.www-data for all my actual public_html folders, for example, for all my vhosts, and then chmod g+s and I umask apache over to 002	02:04
	and I also umask my webdev user over to 002	02:04
	so that if either the shell user or apache creates files in a webroot, they can both read/write/access those files	02:04
pfifo	jim, In actual practice, git dose the oppisite of what I expect. it makes me upset, and there is nothing that i can reference to explain it, it simpley says, "you fail" and i have no cchoice but to try to prove it wrong	02:04
	← meatflag_ left	02:05
jim	you're not the first to say that	02:05
	that's why you should practice	02:05
abstrakt	pfifo, are you coming from subversion?	02:05
	pfifo, git is fairly different from subversion	02:05
pfifo	YES	02:05
	FUKIN YES	02:05
	no	02:05
abstrakt	pfifo, yeah git is a whole new paradigm	02:05
mase_wk	frogonwheels: now that i've removed that directory from the translations branch. what will happen if i want to merge that branch with another which still has the locale directory ? should i go through each branch and remove the locale directory first before merging ?	02:05
pfifo	No i am not	02:05
	→ meatflag joined	02:05
abstrakt	ok, so you're coming from...	02:05
	nowhere?	02:05
pfifo	i am coming from cvs	02:06
abstrakt	you've never used version control before?	02:06
	ohhhh, lol	02:06
jim	yeah, git is just totalluy different	02:06
	so am I actually	02:06
abstrakt	pfifo, wow, yeah, try to throw out everything you thought you knew about version control	02:06
pfifo	i know cvs inside and out and havemigrated to svn	02:06
frogonwheels	mase_wk: if you merge that branch into another, then it will remove the locale directory.	02:06
jim	and there are things I don't understand yet too	02:06
	but I have tools... git init, gitk	02:06
	I can play with it	02:06
frogonwheels	mase_wk: but if you merge another branch into your one without the locale directory, you will get conflicts if any files in that directory changed.	02:07
pfifo	ok look, im not here trying to know everything, i want to know enough to DO MY JOB	02:07
	← dquestions left	02:07
pfifo	dose that make sense?	02:07
jim	this will probably require you to get good enough to improvise	02:08
ice799	Hi. I have two branches, origin/master and origin/B. origin/B diverged from master a long time ago and many rebases have occurred. Merging the two branches is going to be painful. Is there some way I can just tell git to make origin/B the new master? Of course, people who have the code checked out will need to git pull -f, but I'm OK with that.	02:08
mase_wk	frogonwheels: ok thanks	02:08
pfifo	i honsetly do not want to know how to admin a repo	02:08
	pfifo cant admin an svn or even a cvs	02:08
jim	but you do admin a repo, and you start admining another one every time you type git init	02:08
frogonwheels	ice799: you can always do a speculative merge to start with - see how painful it's actually going to be	02:09
jim	or git clone	02:09
ice799	frogonwheels: I started to do the merge and then gave up	02:09
pfifo	jim: thats more than any tutorial has told me	02:09
	→ xiangfu joined	02:09
ice799	frogonwheels: I ended up just git reset --hard origin/master on my local master branch	02:09
	→ dquestions joined	02:09
frogonwheels	ice799: You could also rebase from the point of departure and --skip any conflicts.	02:10
ice799	Hmmm.	02:10
	I'm not sure if I know how to do that.	02:10
frogonwheels	ice799: And then git reset --hard origin/master on your B	02:10
	ice799: git rebase -i {commitid}	02:10
	← flazz left	02:10
frogonwheels	ice799: oh no.. um	02:10
	ice799: look at man git-rebase and search for --onto	02:10
jast	ice799: the 'git-rebase' manpage can be found at http://git.or.cz/man/git-rebase [automatic message]	02:10
	→ flazz joined	02:11
ice799	so the strange thing is that	02:11
frogonwheels	ice799: git rebase --onto origin/master {fromcommit} B	02:11
	ice799: that will minimise that stuff you lose.	02:11
jim	pfifo, when you git init, you create a .git which git assumes is inside a project dir	02:11
frogonwheels	ice799: and the rebase will automatically skip anything that is already applied ( ie where a reverse patch would a apply cleanly I guess)	02:11
jim	every time you git add, or git commit, you're managing the repo for the purpose of saving the state of your project	02:12
ice799	frogonwheels: i have a bunch of cherry-pick commits in branch B and as I was merging it seemed like git didnt know about these commits. The merge would stop and I'd be on (no branch) with the message "Finished one cherry-pick"	02:12
	→ SvenDowideit_ joined	02:12
	SvenDowideit_ → SvenDowideit	02:12
ice799	and if I tried to git cherry-pick SHA on "(no branch)", it would output nothing and cherry-pick nothing.	02:12
frogonwheels	ice799: Oh .. weird.. possibly pastebin your exact commands/output	02:13
ice799	OK, I'll give it another shot with rebase onto.	02:13
pfifo	jim, i got extremly upset this morning, im sorry, but I do NOT understand why, after me sleeping 8 hours it is asking me to take my 8 hour old code and mergetool it with 20 minute old code from the repo	02:14
	it dosent understand versioning imho	02:14
ice799	Another question - what is the easiest way to get rid of merge commits? Should I just rebase and resolve conflicts over again? Kind of painful.	02:14
frogonwheels	ice799: that way you'll lose less I believe. it will still involve some force pushing and stuff.	02:14
	← aresnick left	02:14
frogonwheels	ice799: you can put a -i on that rebase --onto command I guess?	02:15
jim	I saw the aftermath of your upsetness... you turned all the furnature upside down and nailed it to the ceiling :)	02:15
	← dogmatic69 left	02:15
pfifo	damnit its still there >:D	02:15
	→ aresnick joined	02:15
pfifo	i deleted my fork on github	02:16
	← tbrock left	02:16
pfifo	im not asking for much, in fact, i want git todo it the easy way, OVERWRITE my code and make my local repository the same as the newest code	02:17
	but git seems to be using me as its patch command	02:18
abstrakt	sitaram, so I'm sure this isn't news to you, but I've isolated the problem down to that quirk in ssh-add/ssh-agent	02:18
	sitaram, from your ssh-troubleshooting.mkd document "In that case, add the key you want using ssh-add ~/.ssh/mykey and try the access again."	02:18
	sitaram, however that doesn't seem to change anything, unless the implication is that I still must manually select the key I want to use	02:19
jim	you might have to reset to do that	02:19
	← Mage left	02:19
jim	Are you pushing into a nonbare repo?	02:20
pfifo	jim, who?	02:20
abstrakt	sitaram, I have two keys, one for the deploy user, and another I created specifically for gitolite access	02:20
jim	you... but you might have a few usage confusions around this part	02:20
	→ Vampire0_ joined	02:20
abstrakt	sitaram, I've discovered that I can clone only if I have a single key in either ssh-agent or in .ssh/authorized_keys	02:20
	sitaram, if I have both keys in both places, I can't log in	02:21
	sitaram, and as I said, specifically using ssh-add ~/.ssh/gitolite/mykey doesn't actually change this behavior	02:21
pfifo	jim, unexplainable confusion is not good, but something that can direct me to learnable info is welcomed	02:21
abstrakt	pfifo, how much of the git docs have you read?	02:21
jim	remember I said "might"... I still don't know what you;re doing exactly	02:22
	and even if I did, remember I'm new too	02:22
pfifo	abstrakt, I have read everything provided by github's help section.	02:22
abstrakt	pfifo, http://git-scm.com/documentation <- seems to be a fair bit of stuff on there like the "everyday git" and "svn crashcourse" commands	02:22
	→ Mage joined	02:23
abstrakt	pfifo, um... github? how bout read the git docs	02:23
	pfifo, this isn't #github, this is #git	02:23
	no wonder you're frustrated :P	02:23
pfifo	lol	02:23
	yeah, !!!	02:23
	no	02:23
abstrakt	pfifo, so what, exactly is wrong with http://www.kernel.org/pub/software/scm/git/docs/user-manual.html	02:24
pfifo	i am truly frustrated because I cannot understand why git is doing the opposite or a tangent of what i expect	02:24
jim	take a look at this... can you see/hear it?	02:25
abstrakt	pfifo, what you expect based on your usage of svn? or what you expect based on what some docs are telling you, and if so what docs?	02:25
jim	http://blip.tv/file/4094595	02:25
sitaram	abstrakt: git in ssh mode will always choose ~/.ssh/id_[rd]sa, none of the others. You need a "host foo" para in ~/.ssh/config to force that, with at least the "identityfile" directive in it. man ssh_config	02:27
pfifo	what i expect based on svn... I mean cvs and svn are so similiar, i expect git to have the same puropse, but its flexibility is so freat that it excceds my expetations, and provides me with results that are un imaginiable, its beove and beyond attitude are interfering with my normal work process	02:27
abstrakt	sitaram, ok thanks	02:27
banisterfiend	jim: you're the 'jim' who pfifo said started sucking cock and so broke his git setup?	02:27
abstrakt	lol	02:28
	I love you guys	02:28
	← dquestions left	02:28
pfifo	yes i will hold to that, jim enjoys it, but he seems to know his stuff, so i might not break his nose yet	02:29
	;)	02:29
	still not 100 liking him yet	02:29
	→ kpreid joined	02:30
jim	banisterfiend: and went on some wild rampage while I was at my bass lesson	02:30
	→ dquestions joined	02:30
pfifo	giutar?	02:30
banisterfiend	u r a wild child	02:30
jim	I play elec 5 string fretless bass	02:31
pfifo	ahh that explains it	02:31
jim	we were working on one of the Bach 2part inventions	02:31
pfifo	ok yeah im trolling now	02:32
	← boombatower left	02:33
	← banisterfiend left	02:34
	→ cylence joined	02:34
	→ boombatower joined	02:34
	← boombatower left	02:34
	→ boombatower joined	02:34
pfifo	ok everyone here, and the guy im working with says git > svn. they are so different that it is making me frustrated, but i have bee provided with a ource of documentation from kernel.org	02:35
	i can troll for sure, and usually enjoy doing so, perhaps there was a reverse troll here, but none the less with the domain kernel.org, im going to learn that info	02:36
	→ MattDiPasquale joined	02:37
	→ hyperair joined	02:37
abstrakt	sitaram, ok, so yeah I see how that works	02:37
	sitaram, however, how would I differentiate between logging in for git purposes and logging in as a regular shell user?	02:38
	sitaram, actually I think I remember you having docs about giving the git user shell access..	02:38
	→ petercoulton_ joined	02:38
	← petercoulton left	02:38
jim	http://a62.video2.blip.tv/8780006144972/Schacon-GitIn30Minutes520.flv?bri=19.0&brs=808	02:38
	pfifo: look at that	02:38
	→ banisterfiend joined	02:38
pfifo	ok	02:39
	← aresnick left	02:40
jim	like I said, it goes -fast-	02:40
cylence	if a conflict occurs between two branches, is there a command line way to say "just take what the other branch (the incoming stuff) has?	02:40
jim	so you'll want to view it several times, or pause and read man pages, etc	02:41
abstrakt	sitaram, btw "system-install followed by user-setup" <- no such named method in the INSTALL-1.mkd doc	02:42
	sitaram, I can only presume you mean "from DEB" method?	02:42
	← noktoborus_ left	02:42
	← petercoulton_ left	02:43
	→ nb joined	02:43
cylence	any thoughts on between-branch conflicts?	02:45
abstrakt	sitaram, hmm, I have no gl-tool available, or so it seems	02:46
frogonwheels	cylence: you can use git checkout --ours and --theirs to resolve individual files..	02:48
	cylence: if the first branch is local, and you want to throw it away (which is what you're doing) you can just git reset --hard origin/{incoming}	02:49
	where {incoming} is your branch name	02:49
cylence	whoa, really? there are --ours and --theirs flags? how did I not know that?	02:49
	that's pretty rad	02:49
	→ bcardarella joined	02:49
frogonwheels	cylence: is the first branch just a local branch or a pushed one?	02:49
cylence	so mine is local, and the other one (master) is a tracking branch, but also local from where I'm merging	02:50
	frogonwheels: pardon my lack of highlighting :/	02:50
	← gebi left	02:50
frogonwheels	cylence: so if you want to restart your 'mine' branch at where the master is at now, then just reset it to point to there.	02:52
	cylence: ie if you're not going to bother merging, then don't :)	02:52
sitaram	abstrakt: yeah it seems ssh-troubleshooting.mkd went out of synch a little. That whole para should instead use the names referred in 1-INSTALL ("from-client" versus the other 3)	02:53
	abstrakt: will fix; later	02:53
cylence	sure, but I have one conflicting file that I'd rather just yield to what they have	02:53
	frogonwheels: to you, of course	02:53
	← justin-george left	02:54
frogonwheels	cylence: :) ah yes - well that's the --theirs thing :)	02:54
abstrakt	ah nevermind it's in /usr/share/gitolite	02:55
cylence	and just specify the file in question?	02:55
frogonwheels	cylence: exactly. git checkout --theirs -- pathto/myfile.txt	02:55
cylence	frogonwheels: beautiful!! thank you. :D	02:55
	→ bitslip joined	02:55
	→ noktoborus_ joined	02:57
abstrakt	sitaram, sweet... gl-tool shell-add is what I have been looking for this whole time	02:58
	→ tarikk joined	03:01
tarikk	hey there.. pretty new with git .. not sure what to do with this error.. fatal: Not a git repository (or any of the parent directories): .git . im trying to compile android source and after everything builds i get this error at the end and wont generate the system image	03:02
	i think my git dir is corrupt or something.. what can i do?	03:02
SethRobertson	tarikk: what files are in .git? How did you get that git repository?	03:04
tarikk	sorry i dont have a .git.. there is a .repo dir.. i used repo tool to retrieve the git repo	03:06
	→ hebz0rl joined	03:14
	← chrischambers left	03:14
	→ bnovc joined	03:16
	→ Thumper_ joined	03:17
	→ fowlduck joined	03:19
	← Elfe left	03:19
	abstrakt hugs sitaram	03:22
abstrakt	hooray! I think I have accomplished what I wanted! wheeee!	03:22
	→ envi joined	03:22
abstrakt	"think" being the operative word here :)	03:23
	← kukks left	03:23
	← zombor left	03:25
	← MetaCosm_ left	03:26
	← dquestions left	03:27
	→ dquestions joined	03:27
	→ Elfe joined	03:31
	← meatflag left	03:37
	→ meatflag joined	03:37
	→ mdpatrick joined	03:37
	→ zombor joined	03:38
mdpatrick	Hi, kind of new to git.. how do I revert a single file from my commited copy?	03:38
	← programble left	03:40
	← smik`` left	03:46
	→ consolers joined	03:46
	← consolers left	03:47
	← flaggy left	03:49
jim	something like git reset file commit? actually never mind that... man git-reset	03:51
jast	the 'git-reset' manpage can be found at http://git.or.cz/man/git-reset [automatic message]	03:51
jim	you mean you have a commit that has a version of a file you want, and in your project tree you've modified that file?	03:52
	→ towski joined	03:52
	→ gnufied joined	03:54
	← pfifo left	03:55
	← towski left	03:56
	ChanServ set mode: +v	03:57
Fangorn_	mdpatrick, look at the first example in man git-checkout	03:57
jast	mdpatrick: the 'git-checkout' manpage can be found at http://git.or.cz/man/git-checkout [automatic message]	03:57
	← itamarjp left	03:58
	← boombatower left	03:59
	→ jsquared joined	04:01
mdpatrick	thank you Fangorn, and jast	04:03
Fangorn	if you don't mind me asking, how did you solve your problem?	04:03
	← MattDiPasquale left	04:03
	← xiangfu left	04:09
	← flazz left	04:11
	← mmc left	04:12
mdpatrick	Fangorn: Was just reading over gitref.org on the checkout section. It looks like I just have to check out my project over again.. kind of weird, though, because that means it would've overwrote all the work i'd done in other files (if said work existed)	04:13
	Fangorn: Though I guess, I could just stage all of the files except the one I want to restore... commit, then checkout?	04:14
DrNick	you can checkout a single file	04:14
	from an arbitrary commit	04:14
	→ Bruce_Wayne joined	04:15
Fangorn_	mdpatrick, see step 2 in example one in man git-checkout	04:15
jast	mdpatrick: the 'git-checkout' manpage can be found at http://git.or.cz/man/git-checkout [automatic message]	04:15
	→ flazz joined	04:17
mdpatrick	Fangorn_: Thats it!!! Thank you.	04:17
	mdpatrick <3 git	04:17
Fangorn_	glad ot help	04:17
	← Bruce_Wayne left	04:18
	→ Bruce_Wayne joined	04:19
	← Bruce_Wayne left	04:20
	← dquestions left	04:22
	→ Belna joined	04:23
	→ dquestions joined	04:23
	→ Eridius joined	04:25
	→ abetaha joined	04:25
	← hyperair left	04:25
	← zombor left	04:25
	→ lightcap joined	04:27
	→ fr0sty-away joined	04:28
	← banisterfiend left	04:28
	→ hobodave joined	04:30
	→ Mmike joined	04:31
	→ tansell joined	04:33
tansell	any idea why a "git reset --hard" still leaves stuff in a changed state? (ie still appearing in git status)	04:34
	→ Cromulent joined	04:34
	→ banisterfiend joined	04:36
	← bcardarella left	04:39
SethRobertson	tansell: what os?	04:39
tansell	ubuntu	04:39
	git version 1.7.3.1	04:40
SethRobertson	What does `git diff` say? Compare with `git diff -b`	04:40
tansell	git diff and git diff -b output quite different things	04:41
SethRobertson	Does `git diff -b` show anything?	04:41
	→ hyperair joined	04:41
tansell	SethRobertson, yes- it has a bunch of warnings about CRLF stuff and two of the binary files differ	04:42
SethRobertson	What filesystem is this? ext4? fat? smb?	04:42
	→ OOPMan joined	04:43
SethRobertson	tansell: Ah, well then it sounds like you need to tweak the eol and crlf settings. See man git-config	04:43
jast	tansell: the 'git-config' manpage can be found at http://git.or.cz/man/git-config [automatic message]	04:43
SethRobertson	You can convert to the new linux eol style (LF) or set your repo to use the dos style (CRLF) or let your editors fight it out	04:44
	→ Rockj joined	04:44
tansell	SethRobertson, I just want to reset this git repository back to a state where git-status returns nothing	04:44
	← xtagon left	04:45
SethRobertson	Try setting core.autocrlf to input. `git config core.autocrlf input` and then re-running reset	04:46
	← Slix` left	04:46
SethRobertson	(p.s. that's the "let your editors fight it out" option	04:46
	→ xtagon joined	04:47
	← xtagon left	04:48
	→ mndoci joined	04:49
tansell	SethRobertson, it doesn't seem to help...	04:49
SethRobertson	Strange. Let me check.	04:50
	→ justin-george joined	04:50
	← mndoci left	04:50
	→ mndoci joined	04:50
	← MikhX left	04:51
tansell	SethRobertson, there appears to be a .gitattributes in this repository which contains "* crlf=input"	04:51
	← banisterfiend left	04:51
SethRobertson	Ah, .gitattributes is going to override, but...that looks like it might be sane.	04:51
	Well, there is a .git file which will override the .gitattributes file, so you could use that to kill the "* crlf=input" setting	04:54
	→ napster_123 joined	04:55
tansell	yeah	04:55
	http://www.spinics.net/lists/git/msg138133.html	04:55
SethRobertson	and crlf is a legacy setting	04:56
	→ aah joined	04:57
	→ morganick joined	04:57
	→ brntbeer joined	04:58
jjore	Hi, while using plumbing I noted there's support for multiple roots. Is there any porcelain support for multiple roots?	04:58
	→ banisterfiend joined	04:59
SethRobertson	What do you mean by multiple roots? Unconnected branches?	04:59
jjore	I've checked my $HOME into a repo and would like to clone just a single directory	05:00
	I could just create a commit to point at the tree I'd like to use.	05:01
	→ MikhX joined	05:02
SethRobertson	You can do shallow clones, and you can do sparse checkouts, but you cannot do sparse clones	05:06
jjore	Oh really? So I make a commit+branch then I can't clone /just/ thta?	05:07
	→ dbbnj joined	05:07
jjore	Dang.	05:07
	← abstrakt left	05:07
SethRobertson	However, do a `du` of .git. It is likely to be pretty small	05:08
jjore	Lol. It's 140G.	05:08
SethRobertson	That is a large home directory. Have you commited a google earth cache or something?	05:08
jjore	Nah. It's got my source + home movies + vacation pix + mp3s.	05:09
	Among the problems I'm solving are de-duping content.	05:10
kadoban	do you really gain anything for having pics/movies/mp3s in git? i mean, you're not changing those very often, right?	05:10
SethRobertson	git will only store one copy in the repo	05:10
	← abetaha left	05:10
jjore	Incidentally, it turns out that directly editing a file list in a fast-import input is a wicked nice way to push files around.	05:10
SethRobertson	Also note that deleting stuff out of git will not shrink the repo. You would need to git-filter-branch for that	05:11
	← tarikk left	05:11
jjore	SethRobertson: yes. I'm considering keeping a rolling re-rooted tree to discard history older than N days.	05:11
SethRobertson	I	05:12
	← flazz left	05:12
	→ tarikk joined	05:12
SethRobertson	I'm still not sure what you mean by root in the git context.	05:12
jjore	but honestly, the first order goals I /just/ accomplished were de-duping things (I dropped some 300GB just by having content-addressed files) and having a very nice way to edit the trees.	05:13
	It's very nice to edit file names and paths by just editing a text file.	05:13
SethRobertson	Only filter-branch and `git rebase -i` when you squash commits are really going to be saving space while preserving everything else	05:13
	Of course you will need to git-gc git-prune and/or git-repack to actually drop the space	05:14
jjore	Yes, I'm mostly aware of the plumbing required.	05:14
SethRobertson	OK, go forth and commit	05:14
banisterfiend	SethRobertson: you should stay away from those commands, they'r too complicated/hard to understand	05:14
jjore	Wha?	05:14
	rebase -i? I know several companies making excellent use of that.	05:15
SethRobertson	A newbe should avoid filter-branch and wild rebasing until they understand git.	05:15
	← Anti-X left	05:15
	← leroux left	05:15
SethRobertson	However, once you know, go crazy	05:15
jjore	SethRobertson: I just wondered if there was any porcelain for multiple roots because it's quite obvious the plumbing "supports" it.	05:15
	→ leroux joined	05:15
jjore	So thanks.	05:15
SethRobertson	(and take backup clones beforehand, until you are an expert)	05:15
	← cylence left	05:16
	→ simplechat joined	05:16
	← simplechat left	05:16
	→ simplechat joined	05:16
	← dbbnj left	05:16
SethRobertson	I still don't know what you mean by multiple roots. Different content trees?	05:16
jjore	Well, you know how a commit points to a tree as a root directory?	05:17
	If I had multiple, unrelated commits which pointed to sub-trees then I thought I could request only the content required to satisfy that single commit/branch.	05:18
	→ tty2 joined	05:18
SethRobertson	You can do sparse checkouts, but the repo will be full sized.	05:19
wereHamster	jjore: putting $HOME into git is a bad idea. Git wasn't designed for that	05:19
	→ Anti-X joined	05:20
banisterfiend	man git-checkout	05:20
jast	the 'git-checkout' manpage can be found at http://git.or.cz/man/git-checkout [automatic message]	05:20
SethRobertson	I put my $HOME into git. What I don't do is put unrelated bits into the same repo (home directory, source code, mp3s, etc)	05:21
	← tty1 left	05:21
SethRobertson	I used to use CVS for it, and as with most things, transitioning to git has only made it better	05:21
jjore	SethRobertson: my original problem was 3-4 separate dueling partially done backups with mildly redundant paths.	05:21
	So you know, I've been using this as a step to normalize everything.	05:22
wereHamster	jjore: take a look at bup, or git-annex	05:22
SethRobertson	By home directory I of course mean config dot files and ~/bin, not normal subdirectories	05:22
jjore	I don't yet know why bup is considered "not quite there" and as yet, I think annex isn't what I'm looking for.	05:22
	→ abetaha joined	05:23
	← Eridius left	05:24
jjore	I suppose clone is porcelain and maybe there's somehting more low-level. It occurs to me that I could manually provision a repo with all the right objects by copying the various object files along. I don't yet know the pack file format though.	05:24
wereHamster	there is no low-level command to clone only a subtree	05:25
	in fact, there is no plumbing command behind clone. Clone uses fetch internally, and fetch is as low as you can get	05:26
	→ kenneth_reitz joined	05:26
	→ bcardarella joined	05:27
	← Anti-X left	05:27
	← bcardarella left	05:27
	← MetaCosm left	05:27
	← kenneth_reitz left	05:27
	← Buglouse left	05:28
jjore	Bummer.	05:28
	→ MetaCosm joined	05:28
	← hyperair left	05:29
	→ Buglouse joined	05:29
	← abetaha left	05:29
jjore	This really sounds like either I'm going to get to use git a convenient stopping point to somewhere else or learning just what it is that bup is doing	05:29
	→ dbbnj joined	05:32
	→ Anti-X joined	05:32
wereHamster	with but you can't clone a subtree either though	05:33
	← pedrobelo left	05:34
jjore	That really seems arbitrary.	05:34
	→ abetaha joined	05:34
jjore	But oh well.	05:34
wereHamster	no, it's not	05:34
	→ cytrinox_ joined	05:35
wereHamster	because you are supposed to create separate repositories for parts which belong together, and not put everything into one big repo	05:35
jjore	"supposed to?"	05:35
	The plumbing is quite clear about just being a series of trees, all the way down.	05:36
wereHamster	again, you are UGFWIINIF	05:36
jjore	UGFWIINIF?	05:37
wereHamster	using git for what it is not intended for	05:37
jjore	Well, cough.	05:37
	← dquestions left	05:37
jjore	On several dimensions.	05:37
	→ khmarbaise joined	05:37
	→ Jonathan__ joined	05:37
wereHamster	it's not a general purpose, system for everything. It's a SCM (where the S stands for source)	05:37
jjore	Might be nice if there were a `libgit` which catered to just the plumbing.	05:38
	→ dquestions joined	05:38
wereHamster	you mean like libgit2 ?	05:38
jjore	I haven't looked inside that.	05:38
	← Anti-X left	05:39
wereHamster	but what you are suggesting is kindof what bup is doing - building on top of that plmbing, but with focus on big files, backup etc...	05:39
jjore	Yes.	05:39
	cytrinox_ → cytrinox	05:39
	→ cylence joined	05:39
wereHamster	I have to go...	05:39
jjore	Thanks.	05:40
	→ maleknet joined	05:41
	→ sbronson joined	05:41
ice799	Hi. I just created a tag like this: git tag v0.6.0 SHA. How do I push tags to remote branches?	05:41
	← abetaha left	05:41
ice799	ah. git push --tags	05:42
	← brntbeer left	05:42
	→ towski joined	05:42
	→ Anti-X joined	05:43
	← Yuffster left	05:44
	← sbronson left	05:45
	→ abetaha joined	05:46
	→ bronson joined	05:46
	← Belna left	05:46
	→ c_nick joined	05:47
c_nick	is there a way i can see the files last committed.. ?	05:48
fr0sty-away	git show?	05:48
bronson	git log --stat	05:48
	← dquestions left	05:49
banisterfiend	git whatchanged	05:49
	→ dquestions joined	05:49
banisterfiend	git whatchanged -1 is the coolest	05:49
	← Anti-X left	05:50
	← dbbnj left	05:50
	← rchavik left	05:51
c_nick	thanks bronson	05:51
	← abetaha left	05:52
	→ dytrivedi joined	05:53
	→ Phrogz joined	05:53
c_nick	thanks banisterfiend	05:54
	git whatchanged is good -1 gives nothing	05:54
	→ nohk joined	05:54
	← dytrivedi left	05:54
	→ Anti-X joined	05:55
Phrogz	Is there a way to have git diff display \t as specific # of spaces, or is that the job of the terminal?	05:55
	← khmarbaise left	05:55
	← aah left	05:55
	→ flaguy48 joined	05:56
	← c_nick left	05:56
	← Anti-X left	05:59
Phrogz	Sure, I can get diff \| expand -t 2 but then I lose the lovely coloration.	06:00
	→ eletuchy joined	06:03
	→ sivy joined	06:05
	→ Anti-X joined	06:05
	← sivy left	06:06
	→ sambarino1 joined	06:06
	← Phrogz left	06:07
	→ shachaf joined	06:09
	← Anti-X left	06:10
	→ xiangfu joined	06:11
	→ Yuffster joined	06:11
	← Rene1 left	06:11
	← madewokherd left	06:12
	← morganick left	06:12
	→ Aaaarg joined	06:15
	→ evancharlton joined	06:16
evancharlton	hey guys, I messed up with git-filter-branch (applied something I shouldn't have to the entire history)--is there a way to undo?	06:17
	→ doener joined	06:17
frogonwheels	evancharlton: many branches? just one branch?	06:18
evancharlton	frogonwheels, just one branch	06:18
frogonwheels	evancharlton: git reset --hard HEAD@{1} should do the trick (use the reflog)	06:18
	→ hyperair joined	06:18
evancharlton	frogonwheels, hmm, didn't seem to do the trick. I was trying to change the author info on HEAD and accidentally changed every commit, if that helps	06:19
frogonwheels	frogonwheels: you sure? look at git reflog show master (assuming that's the branch)	06:21
	→ amerine joined	06:22
	← amerine left	06:22
frogonwheels	frogonwheels: you should be able to see the original master commit - just need to reset --hard to that commit	06:22
evancharlton	frogonwheels, OH	06:22
	I see now	06:22
	thanks	06:22
	recovered! \o/	06:23
frogonwheels	evancharlton: yay :)	06:23
	evancharlton: I think you can use git commit --ammend with some variable to change just the last commit author	06:24
	--amend	06:24
	evancharlton: huh.. look at --reset-author :)	06:25
evancharlton	frogonwheels, that's what I did, but it still has the committer as the wrong email	06:25
	hm	06:25
	author is correct	06:25
	← fowlduck left	06:26
	→ nadim_ joined	06:26
	→ amerine joined	06:27
	← amerine left	06:28
	→ amerine joined	06:28
	← krang left	06:29
	→ ciupicri joined	06:29
	→ offby2 joined	06:29
	→ freedrull joined	06:31
	← leroux left	06:31
	→ WolfSpider joined	06:34
	← WolfSpider left	06:34
	→ WolfSpider joined	06:36
	← WolfSpider left	06:36
jim	is there git stuff for emacs? for eclipse or netbeans?	06:36
selckin	yes yes yes	06:37
	← ciupicri left	06:37
	→ WolfSpider joined	06:37
	← WolfSpider left	06:37
	→ j0ran joined	06:38
jim	I seem to have netbeans 6.1	06:39
mase_wk	does anyone have a recommendation for a comparison tool for when there is a merge conflict. with SVN i used to use vimdiff and give it the two files however git only has 1 file with a <<<<< ===== >>>>> to separate the diffs	06:40
	→ sako joined	06:40
	← MikhX left	06:41
selckin	git mergetool	06:41
	git mergetool -t vimdiff	06:41
	← evancharlton left	06:41
	→ dc5ala joined	06:42
	→ parasti joined	06:46
	← offby2 left	06:47
	← dquestions left	06:48
	→ dquestions joined	06:48
	→ WolfSpider joined	06:49
	← WolfSpider left	06:49
	← banisterfiend left	06:49
	→ MikhX joined	06:50
	→ WolfSpider joined	06:51
	← WolfSpider left	06:51
	← Textmode left	06:53
	→ Titosemi joined	06:53
Titosemi	morning	06:54
	→ freimatz joined	06:55
	→ WolfSpider joined	06:55
	← WolfSpider left	06:55
	← Cromulent left	06:57
	→ jjcli joined	06:59
_ikke_	morning	06:59
	→ kmap joined	07:01
	→ Shango joined	07:02
	→ krang joined	07:07
	← Guest62552 left	07:09
	← Shango left	07:13
	→ mishok13 joined	07:16
	→ gebi joined	07:18
	← gebi left	07:18
	→ gebi joined	07:18
	→ Zhonghua joined	07:18
frogonwheels	mase_wk: actually git has the 3 files (ours, theirs and base) stored in indexes (like the staging index)	07:19
mase_wk	ah ok	07:20
	i didn't realise that.	07:20
	→ ntoll joined	07:20
	← towski left	07:22
	→ towski joined	07:22
	→ iaj joined	07:24
	← towski left	07:26
	→ SGulseth joined	07:27
	→ Cromulent joined	07:28
	→ cbreak_work joined	07:28
	→ pratz joined	07:28
	← pratz left	07:29
	← nohk left	07:32
	→ tewecske joined	07:33
	→ dvaske joined	07:35
	→ Schmallon joined	07:36
	← OOPMan left	07:37
	→ Fatal joined	07:37
	→ DrNick2 joined	07:37
	→ nohk joined	07:37
	← Cromulent left	07:38
	← DrNick left	07:38
	DrNick2 → DrNick	07:38
	→ nohk_ joined	07:40
	→ saschpe joined	07:41
	→ nothingmuch joined	07:42
	→ siprbaum joined	07:42
	← nohk left	07:43
	nohk_ → nohk	07:43
	→ stuffcorpse joined	07:43
	→ goshawk joined	07:43
	→ thiago_home joined	07:45
	→ kenyon joined	07:45
	→ jsquared1 joined	07:46
	← mikepack left	07:46
	← jsquared left	07:46
	← kenyon left	07:48
	→ kenyon joined	07:48
	→ guardian joined	07:49
	→ priidu joined	07:50
	→ mmc joined	07:52
	→ ph^ joined	07:52
	← guardian left	07:56
	← nohk left	07:56
	→ galderz joined	07:57
	→ nohk joined	08:01
	→ opalka joined	08:02
	→ teweWork joined	08:03
	→ Cromulent joined	08:03
	→ chrissbx joined	08:04
	← hobodave left	08:05
	← amerine left	08:05
	← dc5ala left	08:06
	→ p3rg2z joined	08:06
	← Yuffster left	08:07
	← p3rg2z left	08:09
	← ntoll left	08:11
	→ ch077179 joined	08:11
	→ msieradzki joined	08:13
	→ _ollie joined	08:15
	← j0ran left	08:15
	← _ollie left	08:15
	→ _ollie1 joined	08:15
	← _ollie1 left	08:16
	→ xrn joined	08:17
	→ Schoentoon joined	08:17
	→ OOPMan joined	08:17
	← Schmallon left	08:18
	→ giallu joined	08:22
	← eletuchy left	08:22
	← thiago_home left	08:23
	→ psankar joined	08:23
	← napster_123 left	08:23
	← xrn left	08:24
	← lightcap left	08:24
	→ tvw joined	08:27
	← bronson left	08:27
	← msieradzki left	08:27
	← jfkw left	08:30
	→ thiago_home joined	08:30
	→ pantsman joined	08:31
	→ SIGBUS joined	08:32
	dl9pf_ → dl9pf	08:33
	→ codespectator joined	08:33
	← dl9pf left	08:33
	→ dl9pf joined	08:33
	← JMurray left	08:33
	→ eletuchy joined	08:35
	← steph021 left	08:36
	→ banisterfiend joined	08:36
	→ joeytwiddle joined	08:37
	← nohk left	08:38
	→ luka joined	08:39
	→ skoop joined	08:40
	→ xrn joined	08:40
	→ tasslehoff joined	08:41
	← luka left	08:41
	→ luka joined	08:41
	→ sunn joined	08:41
	← luka left	08:41
	→ david_brent2 joined	08:42
	xrn → zemacsh	08:42
	← thiago_home left	08:42
	→ rraasch joined	08:43
	→ nohk joined	08:44
tasslehoff	Can I make git output the date of a certain commit in a custom format? I need to make a YYMMDDHHMMSS-version number for some fpga code :)	08:45
	→ ayanes joined	08:45
	← ayanes left	08:45
	→ ayanes joined	08:45
	→ pasky joined	08:46
sitaram	tasslehoff: you could use "--date=iso" and massage the output I guess. Or even --date=raw	08:47
	← dissidentss left	08:47
	→ psoo joined	08:48
	← david_brent2 left	08:48
	→ j0ran joined	08:48
	← cylence left	08:50
	→ guardian joined	08:52
	→ nevyn^_ joined	08:52
	→ simplexio^work joined	08:55
	→ nohk_ joined	08:56
	→ acs joined	08:57
simplexio^work	could some one point me to some howto/tutorial which would help me "cleaning" git repo which is made from svn repo, what i eant to do is remove useless commits, and make patch file from latest situation vs initial commit	08:57
	→ luke-- joined	08:58
luke--	hi there	08:58
	I have a question	08:59
tasslehoff	sitaram: thx. I need to emulate subwcrev.. Find the mentioned version number, and a status saying if I have local modifications/commits. I guess maybe a python-script parsing the output from "git status" and "git log -1".	08:59
luke--	how can you remove a branch from github when you've already removed it in your local repo?	08:59
	← pantsman left	09:00
wereHamster	push --delete origin branch	09:00
	ChanServ set mode: +v	09:00
acs	hi	09:00
	I did `git fetch`, then `git merge origin/master`, and I got this error:	09:01
	Automatic merge failed; fix conflicts and then commit the result.	09:01
	why did I get this err ? And how can I update the local repository ?	09:01
	← Tommy[D] left	09:01
wereHamster	you did get that error because there are conflicts which git can't resolve on its own	09:02
acs	I saw, but I did run git reset --hard, and the error still persist	09:02
luke--	wereHamster: thanks!	09:02
	← nevyn^_ left	09:03
wereHamster	acs: did upstream rewrite history?	09:03
	→ Tommy[D] joined	09:04
	← nohk_ left	09:05
	← nohk left	09:06
	→ nevyn^_ joined	09:06
acs	wereHamster: ha?	09:07
	→ nohk joined	09:07
acs	wereHamster: what does that mean ?	09:07
	→ abetaha joined	09:07
	→ nohk_ joined	09:08
wereHamster	acs: do you have any local changes?	09:09
	(which you want to keep)?	09:09
	← abetaha left	09:11
	→ thiago joined	09:13
	← nohk_ left	09:14
	thiago → Guest71226	09:14
	← Guest71226 left	09:14
	→ thiago joined	09:14
	→ gzy joined	09:21
	← guardian left	09:21
gzy	hi, my origins head points to tmp, i now pushed a master branch and would like to point origins head to it, how?	09:22
	→ MetaCosm_ joined	09:22
	← tarikk left	09:23
acs	wereHamster: I had, but I did run git reset --hard	09:23
	wereHamster: now, I do not have any more	09:23
wereHamster	acs: I mean, do you have any local commits which you want to keep	09:23
acs	wereHamster: no, I have no local commit that interests me	09:23
wereHamster	then use reset --hard oriign/mastre	09:24
acs	wereHamster: hm. I did not know that option. let me see	09:24
	→ d0k joined	09:24
acs	wereHamster: great. thanks :)	09:25
	← phantomcircuit left	09:26
tasslehoff	any commands I can use to check 1. do I have local changes to version controlled files, 2. am I ahead of upstream. I can parse the output from git status, but that seems a bit frail.	09:26
	→ lanthan_afh joined	09:27
	→ Zeelot joined	09:28
	→ guardian joined	09:28
wereHamster	git diff --exit-code --quiet	09:28
	→ Schmallon joined	09:28
	→ plantian joined	09:29
	← justin-george left	09:29
wereHamster	and for 2, you can use log, rev-list, cherry.. and parse those	09:29
Arafangion	Why would cherry help?	09:30
	→ doki_pen joined	09:30
tasslehoff	wereHamster: excellent. thanks	09:31
wereHamster	it starts lines with +/- depending on which side has the commit	09:31
	so you can count the number of +'s and -'s and see how far ahead/behind you are	09:31
	→ vitor-br joined	09:31
wereHamster	but rev-list is probably better here	09:32
	→ justin-george joined	09:32
	→ pratz joined	09:35
	← gzy left	09:35
	← Heimidal left	09:35
	→ phantomcircuit joined	09:36
	→ dpino joined	09:38
	← gebi left	09:41
	→ riffraff joined	09:41
	→ dpy joined	09:41
tasslehoff	why is rev-list better? cherry seems like exactly what I need	09:42
	← carutsu_ left	09:44
wereHamster	cherry is not plumbing	09:44
	← phantomcircuit left	09:44
	→ engla joined	09:45
	← engla left	09:45
	→ engla joined	09:45
	→ akamike joined	09:45
	← Jonathan__ left	09:46
tasslehoff	if porcelain works for me, I want to stay away from plumbing :)	09:46
	→ Zenopus joined	09:47
wereHamster	depends for what purpose you need the commands	09:47
	→ shentino joined	09:47
riffraff	I believe I may have found a bug in git: doing a diff between two commits consistently sends the git process to 100% cpu and it stays like that for more than fifteen minutes	09:47
wereHamster	if you are writing a script you should use plumbing	09:47
riffraff	dong --name-only works fine	09:47
	s/dong/doing/	09:47
wereHamster	riffraff: ist hat a private repo? Or can you share it?	09:47
doener	riffraff: did you try with --patience?	09:48
riffraff	doener: no never heard of it I'm sorry	09:48
doener	riffraff: there were reports before that for certain files, the default diff algorithm takes ages, while patience diff finishes quickly	09:48
riffraff	wereHamster, private i was hoping there is a way I can help a diagnosis	09:48
	ok trying it now	09:49
wereHamster	run it in gdb and look what it's doing	09:49
riffraff	confirmed, patience takes one moment	09:49
tasslehoff	wereHamster: for speed/efficiency, or because porcelain-output may change (or for another reason)?	09:50
	← lanthan_afh left	09:50
wereHamster	output may change	09:50
Arafangion	Certain options can finetune what git's diff is doing. For example, how eagerly it should look for changes that have come from other files, how many levels it should look, etc.	09:51
riffraff	so the bug is known: shall I try to further investigate it or is it useless?	09:51
Arafangion	riffraff: It'd be a useful comp-sci project.	09:51
	→ tatsuya joined	09:52
	← stliu left	09:53
riffraff	I take that as "you are hitting an algorithmic edge case and is not a code bug"	09:54
	→ anderse joined	09:55
Arafangion	riffraff: All diff algorithms are heuristics, indeed.	09:55
	← meatflag left	09:57
	→ airborn joined	09:58
	→ curtana joined	10:00
	→ Xeross joined	10:00
	← vitor-br left	10:00
	→ KB1JWQ joined	10:00
	← Schoentoon left	10:01
Xeross	Is there a way to get git-format-patch to only generate patches for 1 branch, and not for the merges from other branches that happen inbetween ?	10:01
	→ Schoentoon joined	10:01
	← tty2 left	10:02
Xeross	as I have about 5 commits from a different branch that I don't need patches for	10:02
	→ deepakk joined	10:02
	→ tty1 joined	10:02
	← tty1 left	10:02
	→ tty1 joined	10:02
	← Cromulent left	10:02
tasslehoff	'git diff --exit-code --quiet' doesn't give a diff for added files. is there a way to make it care about those as well?	10:02
	→ thierryp joined	10:04
	deepakk → dkannan	10:04
	← goshawk left	10:06
	→ thelazydeveoper joined	10:07
	← thelazydeveoper left	10:11
	→ rolfb joined	10:13
	← noktoborus_ left	10:13
	→ thelazydeveloper joined	10:13
jaalto	I'm getting "Connection refused" from "git clone git://repo.or.cz/anything-config.git". would someone briefly check this command. It it works, that may be my local problem, or configuration problem in repo.or.cz	10:13
Fatal	jaalto: works here<tm>	10:14
	→ spaceonline joined	10:14
jaalto	Fatal: Thanks. Hm, I'll have to contact repo.or.cz	10:15
	→ bluenovember joined	10:15
	→ Titosemi1 joined	10:16
Fatal	jaalto: I'd sooner expect some local issue :) tried git clone -vv git://repo.or.cz/anything-config.git ?	10:16
	← Titosemi left	10:16
	← shiba_yu36 left	10:18
	→ albel727_2 joined	10:20
	→ ExtraSpice joined	10:21
	← ph^ left	10:22
	← tatsuya left	10:25
	→ tatsuya joined	10:26
	← wliao left	10:27
	← eletuchy left	10:27
Cocytuz	moving files in git: doing "git add ." when there are several hundred moved and slightly modified files makes git not detect the renames. WHen adding fewer files at a time, git detects renames. Why?	10:28
	→ eletuchy joined	10:28
	← mtx_init left	10:29
Cocytuz	And: is git rename info stored in the repository in any way? Or is it just rm/add with the detection logic in the logg/diff tools?	10:29
	→ jacobat joined	10:32
	→ noktoborus_ joined	10:33
	→ Tabmow joined	10:33
	→ Rhonda joined	10:33
	→ ph^ joined	10:33
Rhonda	No, rename info isn't stored. git tracks content, not files	10:33
	← spaceonline left	10:35
	← mishok13 left	10:35
Cocytuz	Rhonda: ok, thanks. Do you know why the git behaves differently in the "git add ." and "git add somefile.c somefile_renamed.c" scenarios?	10:35
Rhonda	git add . only adds tracked files.	10:36
divVerent	11:33:46 Rhonda \| No, rename info isn't stored. git tracks content, not files	10:37
	it "sort of" is, if the commit contains JUST a rename without change	10:37
	due to the blob hashes, such a commit would perfectly record rename info)	10:37
Rhonda	divVerent: Well, no. Then the treeish only contains the new filename. It doesn't contain information about the old filename.	10:37
divVerent	Rhonda: sure, I was just saying, with SUCH a commit the info can be trivially reconstructed from the commit and its parent	10:38
	git doesn't store any diffs, just states of the repo at each commit	10:38
Rhonda	Everything else is comparison of blobs. If there is no difference it's easy, but it is still the same algorithm of "detecting" renames.	10:38
	→ artefon joined	10:39
divVerent	I was mainly saying that if one wants git to be able to reliably show renames, one should make sure renames occur in a commit with no other changes	10:39
engla	yes. but in the case of perfect renames you can detect renames and moves by diffing the tree objects themselves	10:39
	→ lanthan_afh joined	10:40
divVerent	the problem is that doing this would violate a possible "all commits ought to compile" policy	10:40
engla	divVerent: If one wants git to reliably show renames, use the -M option to diff and log :-)	10:40
divVerent	e.g. in Java where the source files contain their own path :P	10:40
Rhonda	rename a file and copy it to another new name. Now which one is the rename and which one the new copy, even in "perfect renames"? :)	10:40
divVerent	Rhonda: I don't think it makes sense in that situation to distinguish :P	10:40
	← tansell left	10:40
engla	divVerent: Java is a conspiracy to violate DRY in 100 ways	10:41
Rhonda	Cocytuz: Anyway, did my response on git add . help you? :)	10:41
divVerent	DRY?	10:41
	oh, googled it	10:41
Cocytuz	Rhonda: Very much. Me stupid, herp derp. :/ Thanks :)	10:41
	← justin-george left	10:41
divVerent	it's funny that google finds it for "dry", even tho0ugh it is an english word :P	10:41
	Rhonda doesn't find the acronym dry expansion.	10:41
	← psankar left	10:41
Rhonda	Cocytuz: No, it's not that obvious, actually. No need to feel stupid.	10:42
divVerent	engla: guess why I don't like Java :P	10:42
	← guardian left	10:42
engla	DRY = Don't repeat yourself. You shouldn't repeat yourself.	10:42
divVerent	but well... even in C, renaming a source file often has implications in other files	10:42
	→ bitkiller joined	10:42
engla	^^	10:42
Rhonda	TLAs are an EBD because of LNS.	10:42
divVerent	like makefiles, #include statements	10:42
crab	"herp derp"?	10:43
divVerent	but, these typically are OTHER files, so the commits can usually be cratfed in a way that they only do perfect renames or changes of other files	10:43
	evil case is if foo.c includes foo.h, and you want to rename both	10:43
Rhonda	bump the soname!	10:43
divVerent	probably the best way is to first rename foo.h to bar.h, and change all references (including the one in foo.c)	10:43
engla	divVerent: it's not worth it to try to craft those commits. Just do the minimal changes necessary for move and commit it.	10:43
divVerent	and only then rename foo.c to bar.c, and change all references	10:43
	engla: unless I want git to be able to track the rename properly :P	10:44
engla	divVerent: git does detect renames just fine, use -M and you will also get a nice diff showing the minimal changes that were done	10:44
	divVerent: what do you mean with properly here?	10:44
divVerent	engla: so it detects them at all	10:44
engla	divVerent: git doesn't track any rename properly if you're that way.	10:44
divVerent	had quite some failures in the past	10:44
	engla: it does if they are perfect renames	10:44
	then it even is reliable	10:44
	but sure, it depends on whether you even need it	10:45
engla	You are referring to merges etc?	10:45
divVerent	no	10:45
	more like git-blame across rename	10:45
	of course, in MOST cases, a rename is either just a blatant directory rename (typically without changes in the files)	10:46
	or it is a rename combined with huge refactoring	10:46
	the latter, git-blame has no chance at anyway	10:46
	the former is handled well	10:46
	what annoys me more sometimes is false positives in rename handling, especially for binary files	10:47
	like, if a new texture is added that just happens to be somewhat similar to a totally unrelated one	10:47
	like, if a new texture is added that just happens to be somewhat similar to a totally unrelated one	10:47
	git shows this as a copy in the diffstat	10:47
	(I hate hitting UpArrow when I press Enter... eeepc keyboard)	10:48
engla	hm true, that's just noise	10:48
	→ eagles0513875\|2 joined	10:48
eagles0513875\|2	hey guys :)	10:48
divVerent	of course, it just annoys me a bit when viewing the diffstat	10:48
eagles0513875\|2	getting ready to setup my own git repository :)	10:48
divVerent	it makes no sense to git-blame a bitmap file :P	10:48
eagles0513875\|2	looking forwrad to using it for a number of projects	10:48
banisterfiend	anyone here find the git tui confusing?	10:49
	→ mishok13 joined	10:49
engla	eagles0513875\|2: good. welcome to git.	10:49
banisterfiend	it doesnt seem so intuitive	10:49
divVerent	git tui? you mean the regular command line?	10:49
eagles0513875\|2	engla: its not restricted to any particular programming language right	10:49
banisterfiend	yeah	10:49
	i mean cli sorry	10:49
	→ guardian joined	10:49
divVerent	eagles0513875\|2: not really	10:49
	← Mage left	10:50
divVerent	git may have trouble with langauges like Whitespace :P	10:50
	as in, no useful diffs	10:50
eagles0513875\|2	lol	10:50
divVerent	but it works fine with about anything in the real world	10:50
engla	eagles0513875\|2: git can track any collection of files.	10:50
eagles0513875\|2	divVerent: i have a php project i want to put in a repo as well as a java project	10:50
	→ icwiener joined	10:50
eagles0513875\|2	ok engla	10:50
divVerent	if you need whitespace support, you can still write your own diff handler ;)	10:50
eagles0513875\|2	divVerent: i dont need white space support	10:50
divVerent	was just joking	10:50
	git works with about any programming language where the sources look somewhat like text	10:51
	it also works with other stuff, just not as well	10:51
eagles0513875\|2	ok	10:51
engla	banisterfiend: maybe it's not intuitive.. but there are general principles and option syntax that carry over from command to command	10:51
	→ Yuuhi joined	10:51
divVerent	basically, I think "if the diff of a change is human readable, git is the right tool"	10:52
banisterfiend	engla: well i find sometimes options in one command have a different meaning to options in another command, and sometimes the same option is expressed differently, like -n1 versus -1	10:52
	← SIGBUS left	10:52
banisterfiend	engla: and i remember one command (cant recall its name right now) used a different option than -m for specifying a message\\	10:52
	← akamike left	10:53
engla	banisterfiend: ah that's bad. Things that we longtime git users don't think about. Which command uses -n1 ?	10:53
banisterfiend	engla: holdon	10:53
	→ strk joined	10:54
	→ t_b joined	10:54
strk	I'm experimenting with git-svn and don't understand _when_ dcommit will create a commit for each git commit and when it would instead create only one for merge	10:54
engla	no wonder it ends up a bit confusing... git is conservative to preserve old options working, while adding new features and flags all the time	10:54
strk	both happened and didnt' understand why (multicommit before, single comit after)	10:54
banisterfiend	engla: git tag	10:55
	engla: git tag -1 doesn't work, you have to use git tag -n1	10:55
engla	strk: git merges don't transfer well to svn	10:55
strk	uhm	10:55
engla	banisterfiend: It has a different meaning than in git log though	10:56
tasslehoff	is there a way I can specify my own dateformat in rev-list?	10:56
	→ magnusol joined	10:56
banisterfiend	engla: but -n1 means the same as -1 in git log	10:56
strk	engla: how could it be that I got all commit and logs the first time ?	10:56
engla	banisterfiend: no? -<n> in git log is the number of commits to list. git tag -n<n> is the number of lines of the tag message to show, this is what I read from the man page.	10:57
strk	gah, git svn fetch made things even more complex:	10:58
	# Your branch and 'github/master' have diverged,	10:58
	# and have 1 and 1 different commit(s) each, respectively.	10:58
	diff doesn't show any	10:58
engla	strk: I don't know git-svn so well, but it can be that the first parent of the merge is treated as the branch, while changes from the second parent are just squashed in	10:58
banisterfiend	engla: they seem to give exactly the same output	10:58
	→ ereslibre joined	10:59
banisterfiend	engla: try it, git log -n2 versus git log -2 appear exactly the same, same for -n3 and -3 too, etc	10:59
	man git-log (that's for me btw)	11:00
jast	the 'git-log' manpage can be found at http://git.or.cz/man/git-log [automatic message]	11:00
engla	banisterfiend: ok, but -n in git tag is a different thing.	11:00
banisterfiend	oh ok	11:01
	well im as lost as an octopus in a garage	11:01
engla	then do an octupus merge and come back into one flow	11:02
strk	alright, no more merges for svn compat	11:02
banisterfiend	is flyspray dangerous for humans?	11:06
	i just squirted a fuck load of it in this little room	11:06
	to kill a huge fly	11:06
	do i have to open the windows now? (and risk letting more flies in :/)	11:06
engla	I avoid eating pesticides so I'd avoid breathing it as well	11:07
	→ cassielengel joined	11:08
	← cassielengel left	11:09
	← eagles0513875\|2 left	11:09
strk	why do I get a "diverged" status after git svn dcommit ?	11:09
	maybe I shouldn't have a remote when I have an svn-remote ?	11:09
	→ CassielEngel joined	11:10
	→ itamarjp joined	11:10
	← simplexio^work left	11:12
engla	strk: you get a new commit back with some git-svn annotations in the commit log I think.	11:13
	→ napster_123 joined	11:14
	ChanServ set mode: +v	11:14
engla	strk: also, merges are flattened when dcommitting, right	11:14
strk	https://github.com/strk/geos/commits/master	11:14
	← CassielEngel left	11:14
strk	5fb9 (second) is the same of 9e7f (4th)	11:14
	only the latter has the svn id annotation	11:15
crab	oh gh0d, git-svn	11:15
	→ zorzar_ joined	11:15
strk	eh	11:15
	help !	11:15
	:>	11:15
	maybe I should dcommit <my_dev_branch>	11:16
engla	you can't transfer the merges as merges anyway	11:17
	git-svn will limit you	11:17
	← psoo left	11:17
	→ psoo joined	11:18
tasslehoff	is http://pastebin.com/cwiGTnXr using plumbish enough commands that I can be called robust?	11:19
	← dpy left	11:20
	→ vitor-br joined	11:24
	→ g0bl1n joined	11:24
sgronblo	git-svn rulz	11:28
	without it you'd have to use plain svn	11:29
	→ ph^_ joined	11:36
	→ Yuffster joined	11:37
	← Yuffster left	11:38
	← ph^ left	11:38
	→ marcob joined	11:39
sitaram	hi all; any svn access control experts here?	11:40
	← Fatal left	11:40
sitaram	if so, question: in this example: http://pastebin.com/pSEQkSn9 on this page: http://svnbook.red-bean.com/en/1.5/svn-book.html#svn.serverconfig.pathbasedauthz , can someone tell me what would happen if you interchanged the 2 paragraphs?	11:41
	(in this example, I'm wondering if it's specificity that wins, or last match. Both interpretations are possible	11:41
	(I'm writing an article on gitolite's acl language versus svn... just in case someone wonders whay I'm asking)	11:41
	→ nonno joined	11:44
	→ ph^ joined	11:45
	← g0bl1n left	11:46
	→ Belna joined	11:46
	→ ToreadorVampire joined	11:46
	→ oriba joined	11:47
	→ roop joined	11:47
	← ph^_ left	11:47
ToreadorVampire	Hey all - a follow-on from a question last week about hiding/suppressing branches/tags from the output of git branch (which I now know is not possible) - does anyone know how the tab-autocomplete suggestions for git are created (using Debian Squeeze if it matters)? Is it possible to change some config in order to omit remote branches and omit tags from that autocomplete list? It occurs to me that that's mostly where I want to suppr	11:49
	ess things	11:49
	← Buglouse left	11:49
selckin	should be doable	11:49
ToreadorVampire	If I want to know the name of a remote branch then I'll use git branch -r or something like that. My main peeve is when I want to check out a local branch but I forgot what I named it. I type git checkout <tab> and I'm faced with a gigantic list of tags and remote branches that I don't care about	11:49
	If I could suppress those from the autocomplete that would be great	11:50
	In fact, just suppressing tags from that would go a long way - there aren't THAT many remote branches, most are tags	11:51
	→ chrischambers joined	11:52
bremner	ToreadorVampire: probably you could help by specifying which shell you use...	11:52
	→ paulweb515 joined	11:52
ToreadorVampire	Oh, bash	11:52
	ToreadorVampire is skimming over http://repo.or.cz/w/git.git/blob/2ca880fe54660869bc93a2302efced9ab64511d9:/contrib/completion/git-completion.bash atm	11:53
	→ linusoleander joined	11:54
ToreadorVampire	Seems I have a copy of that helpfully placed in /etc/bash/bash_completion.d/git ... now to figure out how to suppress tags from it	11:54
	→ chikei joined	11:55
	← zemacsh left	11:57
	← fr0sty-away left	11:59
	← Titosemi1 left	11:59
	→ fr0sty-away joined	12:00
	← EricInBNE left	12:01
	→ Ryback_ joined	12:04
	← linusoleander left	12:05
	→ philthno2 joined	12:05
	← Manifesto left	12:08
	→ ketin joined	12:10
	→ gebi joined	12:11
	← gebi left	12:11
	→ gebi joined	12:11
	→ gusnan joined	12:11
ketin	Ok, git noob here... having a small issue. Set up a git repo, did the init, add and commit, that worked fine.	12:11
	I'm trying to access it through a git:// address, what would that be? I can't seem to find it.	12:12
	→ Cromulent joined	12:12
ketin	I did already try the name of the folder the source is in. I was never asked to assign a repo name so I'm not sure how to hit it remotely.	12:13
ToreadorVampire	Usually I access my remote repos via ssh	12:13
ketin	I plan to eventually. Trying to set it up with an IDE that supports git:// for the time being.	12:14
ToreadorVampire	In which case you want: git clone "ssh://hostname.for.the.repos/path/on/that/host/to/the/repos"	12:14
	Oh	12:15
	I don't really know/use the git:// protocol - when I figured out I could use ssh I stopped looking for other solutions :)	12:15
ketin	Yeah, I have to figure this one out for the windows people I work with before we move to git.	12:15
zomg	ketin: is the repo on your local machine?	12:18
wereHamster	ketin: git:// is unauthenticated/anonymous protocol	12:18
	you can push through it, but it's not secured	12:18
	so it's disabled by default	12:18
ketin	It's on a server on the network.	12:18
	wereHamster: Yeah, I ran git-daemon manually.	12:18
wereHamster	you can push through ssh (yes, it works on windows) or http	12:18
zomg	You don't even need git-daemon if you can use ssh though :)	12:19
	(or samba since you mentioned windows)	12:19
	← kpreid left	12:19
ketin	Well I looked at the IDE and it doesn't really supprt ssh keys for git so...	12:19
ToreadorVampire	Not sure if you even need IDE support with git	12:20
zomg	Yeah	12:20
ToreadorVampire	When you install git for Windows you get a sort-of-bash commandline that you can use for git-related stuff	12:20
ketin	Yeah, the developers will want single command gui interfaces though.	12:20
zomg	git works best from cli imo but it does require a bit getting used to first	12:20
ketin	Atleast for commit and push.	12:21
zomg	Not real devs if they don't like cli	12:21
	:P	12:21
ketin	Tell me about it.	12:21
ToreadorVampire	Not too knowledgeable on that because I don't use Windows myself, I just know that you get the bash thing because I work with a Windows guy who uses my git repository (badly)	12:21
	:D	12:21
zomg	push doesn't even need a gui.. I mean you just say git push and bam	12:21
	← simplechat left	12:22
ketin	Yeah, they're going to want it built in to their GUI's though... fucking .NET and django guys.	12:22
zomg	Hey now let's not group Django with .NET	12:22
ketin	The php guys won't give a flip if it's cli.	12:22
zomg	;)	12:22
	→ genomega joined	12:22
zomg	You can just tell them to figure it out themselves for their IDE	12:22
ketin	zomg: Well the Django guy is also .NET.	12:22
Cromulent	PHP guys would want someone to do it for them	12:23
	← riffraff left	12:23
zomg	Cromulent: hey now, I'm a PHP guy	12:23
	:D	12:23
Cromulent	my commiserations :p	12:23
	← strk left	12:23
ketin	Well if I can get this working I can get them off of... source safe	12:24
zomg	Nah it's ok since we actually do it properly ;)	12:24
	ketin shudders	12:24
zomg	VSS... :x	12:24
	ToreadorVampire is a .NET guy, although uses MonoDevelop/Debian for all development - and generally deploys to Debian servers running mono	12:24
zomg	I hear even MS never used it on any of their projects...	12:24
	ketin: guess they're using Visual Studio or something?	12:25
ketin	Yeah.	12:25
zomg	http://stackoverflow.com/questions/507343/using-git-with-visual-studio	12:25
	Did you see that?	12:25
	← Eiler left	12:25
	→ Eiler joined	12:25
	→ Dave^\| joined	12:25
	→ stamina_ joined	12:25
ketin	Hmm.	12:26
	→ SIGBUS joined	12:26
	← t_b left	12:26
ketin	So ssh would just be ssh://hostname/var/www/localhost/htdocs/blah/, correct?	12:29
	← Cromulent left	12:29
	← Wyverald left	12:33
	→ _ollie joined	12:35
paulweb515	ketin: I see a lot of ssh://username@hostname/dir/to/repo	12:35
	← _ollie left	12:37
	→ Adaptee joined	12:38
	← ayanes left	12:38
	→ ayanes joined	12:38
	→ lubosz joined	12:42
	→ AlexC_ joined	12:43
AlexC_	morning	12:43
	→ Rickardo1 joined	12:44
Rockj	Anyone could assist me in helping me sharing a repo? I've made a directory under public_html and done git init --bare and git update-server-info and enabled the post hook script, but it doesnt let me clone the repo. it's empty :O	12:44
AlexC_	we have all of our Git repos in /srv/gitosis/repositories and I'd like to use the Git daemon to serve some repositories for when clients would like access to the code, but read only. How can I setup for example, git.their-domain.com and have Git Daemon only possible to serve their repo?	12:44
	→ _peterz joined	12:45
Rockj	(I'am pushing my code from home to publicweb. git remote add publicweb username@hostnanme:fullpath_to_gitdir_under_public_html , git push publicweb master )	12:45
	→ petercoulton joined	12:47
	→ xiong joined	12:50
	→ stevepsharpe joined	12:51
	→ malfy_ joined	12:55
	← malfy_ left	12:55
	→ charon joined	12:56
	→ kpreid joined	12:58
	← philthno2 left	12:58
	→ Wyverald joined	12:59
	→ unreal joined	12:59
	→ dr_lepper joined	12:59
	← dr_lepper left	12:59
	→ dr_lepper joined	12:59
	← thelazydeveloper left	13:02
	→ auslander_ joined	13:04
FauxFaux	Rockj: So you're cloning an inited repository, and it's empty? Shocker.	13:04
Rockj	FauxFaux: I've been tricking myself somehow	13:05
	after the clone it says it is "empty"	13:06
	so I thought it didn't fetch any files...	13:06
	hides	13:06
	← auslander_ left	13:06
	← ToreadorVampire left	13:06
FauxFaux	I still don't understand, but you sound happy, so all is good. ¬_¬	13:06
	← emdete left	13:06
Rockj	if I remember correctly, git fetch tells me some stats after I've pulled some files, right?	13:07
	← teweWork left	13:07
Rockj	it doesn't do on the initial git clone, but it just says "Initialized empty Git repository in /tmp/ticketdroid/.git/" - I thought it didn't get anything then. (of course I never double checked with a ls :p)	13:08
	anyway, time to get some breakfast and wake up sighs	13:08
	→ teweWork joined	13:09
FauxFaux	Ah yes, clone just shows the init message if there's not much to do.	13:09
	← teweWork left	13:10
	← d0k left	13:10
	← Rickardo1 left	13:10
	← arun_ left	13:12
	→ smik joined	13:12
	→ arun_ joined	13:12
	→ teweWork joined	13:14
	← lubosz left	13:14
	← _peterz left	13:15
	→ jjuran joined	13:15
	→ LRN joined	13:16
	→ dels joined	13:16
	← Dave^\| left	13:16
	→ Dave^\| joined	13:16
	← Dave^\| left	13:17
	→ Dave^\| joined	13:17
	→ linusoleander joined	13:18
	← ReleaseX left	13:18
	← linusoleander left	13:19
	← jelmer left	13:20
	→ jaql joined	13:20
	→ SianaGearz joined	13:20
	← oriba left	13:21
	→ simosx joined	13:21
	← simosx left	13:21
	→ simosx joined	13:21
	← ereslibre left	13:22
	→ Wyverald1 joined	13:23
	← Wyverald left	13:23
	← SIGBUS left	13:24
	← rolfb left	13:24
	→ jstemmer joined	13:24
	→ coderdad joined	13:25
	→ jelmer joined	13:25
	→ Amorphous joined	13:25
	→ smik` joined	13:26
	← jelmer left	13:26
	→ Wyverald joined	13:27
	→ jelmer joined	13:27
	← smik left	13:27
	← Wyverald1 left	13:28
	→ zheng joined	13:31
	→ Rickardo1 joined	13:37
	→ EricInBNE joined	13:38
	→ guide joined	13:43
	← kpreid left	13:43
	← xiong left	13:43
	← LRN left	13:44
	← maleknet left	13:46
	→ sven^ joined	13:48
	→ g0bl1n joined	13:48
	→ LRN joined	13:50
	→ tstclair joined	13:50
	→ _dels joined	13:51
	← Dave^\| left	13:51
	→ q0rban joined	13:51
	→ marc_kdab joined	13:51
	← dels left	13:51
	← zheng left	13:55
	→ zheng joined	13:55
	← hyperair left	13:57
	→ paradigm joined	13:57
	→ philthno2 joined	14:02
	← avinashhm left	14:02
	→ avinashhm joined	14:03
	← artefon left	14:04
	→ hendry joined	14:04
	← krang left	14:04
hendry	i have a shared repo /srv/git/foo and I want to sync it up with a remote/srv/git/foo, how do I do that in one step? do I really need a working tree?	14:04
	→ tizbac joined	14:05
	← jelmer left	14:05
	→ smuf joined	14:05
	← tasslehoff left	14:06
	→ hyperair joined	14:06
	← gusnan left	14:08
wereHamster	hendry: are both bare?	14:08
	→ spuz joined	14:12
	← teweWork left	14:13
	← Adaptee left	14:13
	→ HG` joined	14:13
hendry	wereHamster: yah	14:13
wereHamster	cd /srv/git/foo && git fetch remote/srv/git/foo --all	14:13
	← napster_123 left	14:15
	→ teweWork joined	14:15
	→ zombor joined	14:16
	→ jcdny joined	14:16
	→ rchady joined	14:16
hendry	wereHamster: hmm, fatal: fetch --all does not take a repository argument	14:16
	← zheng left	14:16
doener	well, --all means "fetch from all remotes"	14:17
hendry	seemed to fetch something, but git log doesn't say the right things. gr	14:17
	← dkannan left	14:17
wereHamster	ah, I thought it meant fetch all refs	14:17
	like, push --all	14:17
doener	hendry: git fetch remote/srv/git/foo refs/:refs/	14:18
hendry	doener: thanks! that worked	14:18
	→ elmob joined	14:18
hendry	jeepers, git is a tad cryptic	14:18
doener	hendry: also: git remote add --mirror origin remote/srv/git/foo	14:18
	hendry: and from then on just "git fetch origin"	14:19
	hendry: just be careful not to push, as the mirroring gets setup in both directions	14:19
hendry	in my case, we're just migrating servers from one to another, and i need to ensure i'm in sync when dns changes over somehow	14:19
	_dels → dels	14:20
	→ NickyP joined	14:21
	← nadim_ left	14:21
	← kanru1 left	14:22
	→ chrisred joined	14:23
	→ Manifesto joined	14:23
	→ nadim_ joined	14:23
Eiler	is there a gui for git in linux (gentoo)?	14:25
	→ dbbnj joined	14:25
tango_	Eiler: git gui if you emerged with tcl/tk support	14:25
Eiler	aha i did not, what flag is that i need?	14:26
	← iaj left	14:27
	→ drecute joined	14:27
drecute	hello	14:27
	→ iaj joined	14:27
drecute	how do i resolve a conflict after git push -f	14:28
	→ lostern joined	14:28
wereHamster	what conflict? There is none if you use push -f	14:28
bremner	there is for the poor sod that pulls.	14:28
drecute	i got a conflict on the server end of the repository	14:29
	so auto merge failed	14:29
bremner	err, faq non-bare?	14:29
jast	err: Pushing to non-bare repositories is discouraged: please see https://git.wiki.kernel.org/index.php/GitFaq#non-bare [automatic message]	14:29
bremner	drecute: you know about this^	14:29
drecute	no	14:29
	← simosx left	14:29
	→ brntbeer joined	14:29
drecute	let me read up	14:29
	← roop left	14:30
	→ fowlduck joined	14:30
	→ jelmer joined	14:31
	ChanServ set mode: +v	14:31
lostern	I want to use git-svn instead of svn on a large repository which I have checked out in svn. Is there a way to git-svn clone the version I have checked out instead of over the network and then change the git-svn repo to point to the network remote?	14:31
wereHamster	lostern: no. what you can do is setup a svn mirror and clone that	14:31
yoh	rudi_s: /hg git/ thanks for your suggestion -- I have gone through README again and still do not see how I could have local git clone of remote hg repository (i.e. without local .hg)... or have I missed something?	14:32
drecute	so a possible solution is to git reset --hard on the server repository right?	14:32
lostern	wereHamster: Thanks	14:33
bremner	drecute: depends on your definition of "solution".	14:33
	← gebi left	14:33
drecute	i just need an answer damn it	14:34
	→ mfacenet joined	14:34
	← mfacenet left	14:34
	→ mfacenet joined	14:34
	← dels left	14:34
	→ gebi joined	14:34
	← gebi left	14:34
	→ gebi joined	14:34
	→ linusoleander joined	14:34
bremner	well, do you care about throwing away changes? do you plan on doing this every time you push? Shouldn't you really set up a bare repo to push to?	14:34
	oh, wait, those are questions, not answers.	14:35
	→ bcardarella joined	14:36
	→ pigdude joined	14:37
drecute	sometimes i just wonder what kind of support u guys provide	14:37
bmalee	The voluntary kind? :p	14:38
bremner	money cheerfully refunded.	14:38
wereHamster	the best you can get.. if you are able to articulate your problem..	14:38
sgronblo	rofl	14:38
	usually when you whine about how somebody who does not get paid to help you is not helping you, you are just likely to be seen as ridiculous	14:39
	→ Wyverald1 joined	14:39
	→ nairb774 joined	14:40
EvanR-work	sgronblo: what do you think free software means anyway	14:40
wereHamster	the level of support is directly proportional to the amount of information we get about the problem. The more we know the better our answers	14:40
bremner	whining?	14:40
EvanR-work	people are supposed to help you with any problem and do stuff for you at no cost	14:40
	package deal	14:40
bremner	drecute: care to try again? the laughter will stop pretty soon.	14:41
	→ Rai_ joined	14:41
sgronblo	you didnt think that sounded like whining?	14:41
bremner	I did, I meant to answer "what is Free Software about"	14:41
parasti	EvanR-work: well, nobody is really "supposed to", those who do are simply keeping with the spirit	14:42
EvanR-work	im not serious	14:42
	← Wyverald left	14:42
EvanR-work	in case you missed that	14:42
	← linusoleander left	14:42
sgronblo	EvanR-work: you forgot to include a smiley to clearly indicate that you were being ironic	14:44
EvanR-work	thats sort of like a laugh track in a sitcom, watering it down	14:45
sgronblo	well I hope you didn't take it seriously :) :) :)	14:46
parasti	I am ashamed in a corner, FYI	14:47
	← xiangfu left	14:47
	→ msieradzki joined	14:48
	→ LiamH joined	14:48
	← freimatz left	14:48
	← shentino left	14:50
	→ rolfb joined	14:51
	→ g0bl1n_ joined	14:51
	→ kogent joined	14:51
	← g0bl1n left	14:52
	g0bl1n_ → g0bl1n	14:52
	→ kanru joined	14:53
	→ sonnym joined	14:53
	← brntbeer left	14:55
	← OOPMan left	14:55
	→ ereslibre joined	14:59
	← ereslibre left	14:59
	→ ereslibre joined	14:59
	→ khmarbaise joined	14:59
	→ smik`` joined	15:00
	→ Wyverald joined	15:00
	→ kpreid joined	15:01
	← smik` left	15:02
	← parasti left	15:02
	→ parasti joined	15:03
	→ qfr joined	15:03
	← Wyverald1 left	15:03
	→ strk joined	15:03
strk	git-am: previous rebase directory ... still exists but mbox given	15:04
	→ simonj joined	15:04
strk	what does it mean ?	15:04
	← parasti left	15:05
	← ereslibre left	15:06
	← tizbac left	15:07
	← LiamH left	15:08
	← qfr left	15:09
	→ LiamH joined	15:09
_nil	i can use a git filter-branch to nuke a directory correct?	15:10
	another developer used an capitalized directory name	15:10
	← NickyP left	15:10
bremner	why not just move it?	15:10
_nil	so he did git commit Server/blah instead of server/blah	15:10
	bremner: because it created the directory on the server	15:11
	i can see Server/blah on github	15:11
	and the proper server/blah	15:11
	→ frakturfreak joined	15:12
bremner	my point is that filter-branch is probably un-neccesary. Just fix it and push again	15:12
	→ mastro joined	15:13
bremner	filter-branch is the tactical nuke of git. Effective, but with some side effects.	15:13
_nil	bremner: ok i did that already so i guess i'll just leave it	15:14
	→ rayners joined	15:15
	← ph^ left	15:16
	← iaj left	15:17
	← Wyverald left	15:19
	→ ereslibre joined	15:19
	→ jfkw joined	15:19
	→ iaj joined	15:20
	→ mabrand joined	15:21
_nil	bremner: thx	15:22
	→ artefon joined	15:22
bremner	_nil: welcome.	15:23
	→ brntbeer joined	15:24
	→ Juri_Agarin joined	15:25
	← welly left	15:26
	→ parasti joined	15:27
	→ berkes joined	15:29
	→ sftp joined	15:31
	→ WebDragon joined	15:32
spuz	hello, if git diff is telling me an entire file has changed but visually, the files look exactly the same, is there a way I can get git to tell me what the actual differences it has found are?	15:35
	this might happen if for example the line endings have changed (though that's not the case for me right now)	15:35
engla	it might be that line ending conversion settings have changed	15:36
	after all git diff shows the percieved changes, changes that would be if you did commit -- with some line ending settings I understand that the worktree file will be different from the stored file	15:37
	spuz: ^	15:37
	→ wanders joined	15:40
	→ stliu joined	15:40
	→ ceez joined	15:41
	← anderse left	15:41
	← giallu left	15:41
	← magnusol left	15:41
	→ magnusol joined	15:42
	← lostern left	15:42
	→ Blackguard joined	15:43
spuz	ok here's a related question, is it possible to open a file that is either in git's index or in HEAD in an external editor?	15:43
	→ Naikrovek joined	15:44
	→ ussy joined	15:46
	← ussy left	15:46
	→ Remixman joined	15:47
drecute	i cant seem to be able to get any file at http://utsl.gen.nz/git/post-update	15:47
_rane	git show HEAD:path/to/file \| $EDITOR	15:47
	or something	15:47
	→ beatak joined	15:47
	← magnusol left	15:47
_rane	if that's what you meant	15:47
drecute	_rane: u talking to me?	15:48
	← gnufied left	15:48
	→ SpookyET joined	15:48
SpookyET	Hello. Has anyone got a good zsh completion for git? The one bundled with zsh blows.	15:49
_rane	drecute: no	15:49
parasti	spuz: what _rane said and "git show :path" to show a file at index	15:49
	← banisterfiend left	15:49
	→ magnusol joined	15:50
spuz	parasti: thanks	15:50
	→ sivy joined	15:51
	← mndoci left	15:51
ft	SpookyET: Maybe you should start reporting bugs to the zsh developers so they know where the problems are.	15:51
	→ Belna_ joined	15:52
	→ blackxored joined	15:52
	→ alvaro_o joined	15:52
	← mfacenet left	15:52
blackxored	is there an easy way to push only several commits to origin without rebasing???	15:52
	specifically I want to omit the latest commit from being pushed	15:53
	→ nullvariable joined	15:53
	→ o_o joined	15:53
parasti	then push HEAD~1 with the full syntax	15:54
	o_o → Guest57261	15:54
	← magnusol left	15:54
	Guest57261 → robotarmy_	15:55
blackxored	fatal: remote part of refspec is not a valid name in HEAD~1	15:55
	i could use HEAD^	15:55
	as well right?	15:55
parasti	yes, but it won't change the fact that it is not full syntax :P	15:55
blackxored	heheheheh	15:55
parasti	full syntax is git push +<src>:<dst>	15:55
	sorry, actually git push origin +<src>:<dst>	15:56
blackxored	so assume it's github how would you do it	15:56
	← elmob left	15:56
	← wanders left	15:56
blackxored	git push origin +HEAD~1:master	15:56
parasti	I would use something like git push origin HEAD~1:refs/heads/mybranch	15:56
blackxored	???	15:56
	← strk left	15:56
parasti	or maybe just that works, I can never remember	15:56
	→ yrlnry joined	15:56
blackxored	i'll tell you in a min	15:57
parasti	the + might hurt actually, might be better to leave that out	15:57
	→ kukks joined	15:57
	→ banisterfiend joined	15:57
	← tlvb left	15:57
	← kpreid left	15:57
	← tchan1 left	15:57
blackxored	yep, that worked, god I love git, thanks parasti ;)	15:58
	for reference just	15:59
	git push origin <commitref>:<remote branch> :P	15:59
drecute	i have the conflict info here http://fpaste.org/Oxdm/	15:59
	how do i go about this pls	15:59
	→ spaceonline joined	16:00
bluenovember	drecute, git mergetool	16:01
engla	drecute: if you have a merge conflict, first check that you are mergin what you think you are.. unexpected conflicts come from that. else.. you edit the files to resolve the merge manually.	16:01
	→ iravid joined	16:03
	← chikei left	16:03
	→ SimonP86 joined	16:04
	→ ph^ joined	16:04
	→ kpreid joined	16:05
	← dquestions left	16:05
	→ dquestions joined	16:05
	→ stringoO joined	16:05
	← frakturfreak left	16:06
	→ frakturfreak joined	16:06
	← banisterfiend left	16:07
drecute	engla: so if i'm sure about what i'm merging, i can run git mergetool right?	16:07
	← thierryp left	16:07
	→ chrisshattuck joined	16:07
engla	drecute: you should first find out which branches you are merging.	16:07
	→ d1b joined	16:08
engla	drecute: git mergetool will help you edit the files to resolve the conflicts.	16:08
	drecute: ah now I see "I'm sure" and not I'm not sure.. sorry! yes go ahead	16:09
	← khmarbaise left	16:09
	→ wanders joined	16:10
	← simonj left	16:11
acs	I am on master branch, I wrote code, and I do not want to commit it on master. I want to commit it on a branch BBB. How can I do it ?	16:11
parasti	git checkout BBB	16:11
	(possibly with git checkout -m BBB, as it might refuse otherwise)	16:11
	← sh1mmer left	16:11
	← j0ran left	16:12
acs	let me see...	16:12
	→ magnusol joined	16:12
	→ skyl joined	16:13
acs	parasti: it worked without -m...	16:13
	thanks	16:13
skyl	can I use .gitignore or sth else to ignore symlinks?	16:13
	→ bartek joined	16:13
yrlnry	Git should treat symlinks the same as any other file.	16:14
bartek	Hi. What's the proper git command to archive my repo into a tar while preserving history? git archive seems to mainly be for simply exporting the code, not history -- right?	16:14
yrlnry	If you put the name of the symlink into .gitignore, git should ignore it.	16:14
	→ tj- joined	16:14
yrlnry	bartek: if the tar gile includes the .git directory, that is your history right there.	16:14
	→ cylence joined	16:14
	tj- → tjholowaychuk	16:15
bartek	touche	16:15
	← cbreak_work left	16:15
yrlnry	bartek: if you copy that .git directory anywhere, incuding out of a tar file, git will be able to reconstruct the history and check out the old revisions.	16:15
drecute	how did it know	16:15
	how does git find the right file	16:15
bartek	magic	16:16
	← pratz left	16:16
drecute	even when the files are not in the same directory and they dont have the same names	16:16
bartek	linus ssh's into your box	16:16
	→ Cromulent joined	16:16
	→ banisterfiend joined	16:16
bartek	stupid question i know :p	16:16
yrlnry	drecute: the file names are recorded in the respository.	16:16
	In the tree objects.	16:16
	bartek: try for example: mkdir /tmp/copy; cp -r .git /tmp/copy/.git; cd /tmp/copy; git blah blah	16:17
drecute	wow	16:17
	→ _psychic_ joined	16:17
bartek	yrlnry: hmm, what about the files above the .git repo? Don't I need those as well?	16:18
yrlnry	bartek: No, that's just the working directory. Git can recover them from the repository if the working directory is clean/	16:18
	← magnusol left	16:18
bartek	Ah, makes sense .. thanks for clarifying	16:18
yrlnry	bartek: try it, and after you cd /tmp/copy, do "git reset --hard".	16:18
bartek	So when I tar, I should just tar my .git directory, not the base working directory	16:19
yrlnry	Either one is fine.	16:19
bartek	ok	16:19
	→ napster_123 joined	16:19
yrlnry	If you have changes in the working directory that are not committed to the repository, they will not be in the .git subdirectory.	16:19
bartek	Right	16:19
	That's very good to know	16:19
	→ Adaptee joined	16:19
yrlnry	Although any changes you added with git-add will be in there.	16:19
	→ brntbeer_ joined	16:20
yrlnry	I suggest you try it a bit. It's all very predictable.	16:20
	→ ShooterMG joined	16:20
	← brntbeer_ left	16:20
skyl	one may explicitly add a file that is in .gitignore, yes?	16:21
crab	yes.	16:21
	← codespectator left	16:21
	← nohk left	16:21
	→ yairgo joined	16:21
yrlnry	Hi, Crab!	16:22
crab	hi yrlnry. i liked your git reset piece very much.	16:23
	← FYLGood left	16:23
drecute	is it really compulsory to do git-pull before git-push?	16:24
	→ warbee joined	16:24
_rane	no	16:24
yrlnry	drecute: No.	16:24
drecute	but i do have issues everytime i run git push without git-push -f	16:24
yrlnry	drecute: the potential problem is that if the remote head has moved forward, your push will fail if you don't somehow adjust your repository to match before trying the push.	16:25
drecute	git always want to be to git pull first	16:25
	→ juanjoc joined	16:25
	← Zhonghua left	16:25
crab	drecute: is someone else pushing to the same remote repository? if so, you're overwriting their changes every time you push -f	16:25
drecute	yrlnry: so doing git-pull is valid then	16:25
yrlnry	"valid"?	16:26
drecute	yrlnry: so doing git-pull before git-push is valid then	16:26
	→ FYLGood joined	16:26
drecute	crab: yes	16:26
yrlnry	I don't know what "valid" means in this context, sorry.	16:26
	← stringoO left	16:26
drecute	yrlnry: i mean it is right to do so	16:26
	← kpreid left	16:27
crab	drecute: it's one of the ways in which you can deal with the situation.	16:27
	→ nohk joined	16:27
drecute	crab: how do i?	16:27
	because i don't want to have issues again anytime i want to do git-push	16:27
yrlnry	Personally, I never use pull; I always use fetch, followed either by git merge --ff-only (if that will work) or git-rebase (if it won't).	16:28
	But theere is no "right" way; you need to come to an agreement with your collaborators about the best way to handle the situation,.	16:28
drecute	yrlnry: we are always trying to	16:29
	but we always come to a deadlock	16:29
	← hyperair left	16:29
crab	why deadlock?	16:29
PerlJam	drecute: probably because you are always overwriting each other's changes with push -f :)	16:29
	→ nevyn^__ joined	16:29
crab	drecute: have a look at the diagrams and push/pull explanation on http://toroid.org/ams/git-central-repo-howto	16:30
drecute	PerlJam: but git-pull always fails anytime i run it from my end	16:30
jast	yrlnry: you might as well always use pull --rebase, that amounts to the same thing :)	16:30
drecute	PerlJam: sometimes it is intentional	16:30
crab	why does it fail?	16:30
yrlnry	Maybe the other guy is push-f-ing also.	16:30
PerlJam	drecute: "git push -f" should be an extraordinary thing to type, not every day usage	16:31
_rane	depends	16:31
	→ krainboltgreene joined	16:31
jast	push -f is another expression for "please throw away all upstream that I don't already have locally"	16:31
	*changes	16:31
	→ saysjonathan joined	16:31
	← rolfb left	16:32
	← nevyn^_ left	16:33
_rane	if I work alone and have accidentally pushed a typoed commit message, I might push -f amended message just because I'm pedantic	16:33
	← sako left	16:33
yrlnry	crab: thanks, I am glad you liked it. What did you like about it?	16:33
	crab: I thought it was rather thin, myself.	16:34
PerlJam	_rane: sure, under those very specific circumstances.	16:34
	← avinashhm left	16:34
drecute	thanks crab	16:34
	that is a good resourc	16:34
	← opalka left	16:35
	→ WebDragon\|laptop joined	16:35
crab	yrlnry: what i liked most is that i can now point people to the three rules.	16:35
	→ Ruudjah joined	16:35
yrlnry	Aha!	16:35
	→ cannonball joined	16:35
yrlnry	crab, drecute : I saw this recently and thought it seemed very helpful: http://marklodato.github.com/visual-git-guide/index-en.html	16:36
	→ slonopotamus joined	16:36
	← gretch left	16:36
	→ codespectator joined	16:37
	→ GarethAdams\|Work joined	16:38
	← WebDragon left	16:38
	← marc_kdab left	16:41
	→ goshawk joined	16:42
	← wanders left	16:42
	→ aresnick joined	16:42
	PerlJam looks at yrlnry's text on git-reset	16:43
	← philthno2 left	16:44
	→ avinashhm joined	16:44
	→ aah joined	16:45
	→ tchan joined	16:45
	→ hyperair joined	16:45
	→ linusoleander joined	16:45
	← Ruudjah left	16:45
	← aah left	16:46
	← saschpe left	16:46
	← slonopotamus left	16:47
	→ slonopotamus joined	16:48
	Xeross → Xeross\|AFK	16:48
yoh	rudi_s: /hg git/ ha -- grepping through the logs of this chatroom -- found out about plans for git-remote-hg -- I guess that is what I need ;)	16:50
	→ Heimidal joined	16:51
mdpatrick	I really should log all of the channels I'm in and do exactly what you just suggested, yoh.	16:51
	That's kind of genius.	16:51
	Ooh... Nice. I've been logging all along. There's got to be some gold in there.	16:52
	→ abetaha joined	16:52
	→ hobodave joined	16:52
jast	there are public logfiles of this channel; see topic	16:53
	→ chikei joined	16:53
rudi_s	yoh: ;-) That would be interesting. At the moment I use a local hg repository and it works fine.	16:54
	← lanthan_afh left	16:55
	→ jrmuizel joined	16:55
	← mlq left	16:56
	WebDragon\|laptop → WebDragon	16:57
	← AlexC_ left	17:00
	→ rolfb joined	17:01
	← rolfb left	17:01
drecute	once there's a conflict and i run mergetool, what does it to?	17:01
	how does it resolve the conflict	17:01
	→ rolfb joined	17:01
drecute	does it copy one change to another based on some logic?	17:02
SethRobertson	A human (you) tells it what to do	17:02
bluenovember	It should bring up your merge resolution tool of choice, then you select which of the versions to keep	17:02
	quite likely you will have to do more than just chose between changes for a given conflict	17:03
	← blackxored left	17:03
SethRobertson	Personally, I usually just go and edit the files directly	17:03
	← alberto56 left	17:03
	← rolfb left	17:04
	→ xtagon joined	17:07
	→ WebDragon\|laptop joined	17:07
	→ Bruce_Wayne joined	17:08
	→ ericenns joined	17:08
	→ amerine joined	17:08
	← luke-- left	17:09
	→ thierryp joined	17:10
	→ shentino joined	17:10
drecute	SethRobertson: what of in a case where you are collaborating with someone	17:11
	← WebDragon left	17:11
drecute	so both of u have different versions of these conflicting files	17:11
	← Eiler left	17:12
	→ boombatower joined	17:12
	← boombatower left	17:12
	→ boombatower joined	17:12
	← MikhX left	17:12
	WebDragon\|laptop → WebDragon	17:12
	→ sterlo joined	17:13
	← sterlo left	17:13
	→ nakkor joined	17:13
	→ sterlo joined	17:13
	→ mw joined	17:13
yoh	is there anyone using git_remote_helpers python module? could you give me a hand on how to give it a try? (in particular Sverre's git-remote-hg.py)	17:13
	I just wonder how to make git aware of hg...	17:13
	→ thelazydeveloper joined	17:14
	← SGulseth left	17:15
	→ wanders joined	17:15
	→ Eiler joined	17:15
	→ philthno2 joined	17:16
crab	yrlnry: i guess he beat me to it.	17:17
	← thierryp left	17:18
	← stliu left	17:19
	← kanru left	17:20
	← frakturfreak left	17:20
	→ hex45_ joined	17:21
yrlnry	crab: I think the world has room for more than one of those./	17:21
	→ MikhX joined	17:21
	→ troyt joined	17:22
	← Eiler left	17:24
parasti	xmms2 list	17:24
	oops	17:25
	→ sako joined	17:25
	← bcardarella left	17:25
	← codespectator left	17:25
	← hex45_ left	17:26
	← bartek left	17:26
	← smuf left	17:28
	← genomega left	17:28
	← nohk left	17:29
	→ bcardarella joined	17:29
	← bcardarella left	17:29
	→ nohk joined	17:30
	← EricInBNE left	17:31
	→ Eiler joined	17:31
engla	SethRobertson: one person does the merge, and one person then resolves the merge in that place and pushes the result.. done.	17:31
	drecute: ^	17:31
	SethRobertson: oops wrong person	17:31
	→ ignatp joined	17:31
	→ frakturfreak joined	17:32
	→ joel_falcou joined	17:32
	← PeaMan left	17:33
	→ aah joined	17:33
drecute	engla: then both can pull the result to local wd	17:33
	→ mikepack joined	17:33
drecute	?	17:33
engla	drecute: why could they not? :-)	17:34
	→ davidfetter_vmw joined	17:34
	← banisterfiend left	17:34
drecute	tnx engla	17:35
	→ Modius joined	17:35
	→ PeaMan joined	17:35
	← MikhX left	17:35
	→ gusnan joined	17:36
	→ d0k joined	17:36
	→ MikhX joined	17:36
	→ genomega joined	17:36
	→ Tobias\| joined	17:38
Tobias\|	Hmm	17:38
	I made some changes to a file, then reverted them	17:38
	→ kelvie_ joined	17:38
Tobias\|	But git seems to be shouting about fast-forwards when I push to that repository still?	17:38
	← Eiler left	17:38
	→ kenichi joined	17:39
	← tstclair left	17:40
	← nevyn^__ left	17:41
yrlnry	if you used "git-revert", that does not revert to the old commit, but rather writes a new commit in which the changes are undone.	17:41
Tobias\|	How annoying	17:42
	Any way to actually revert?	17:42
	→ kpreid joined	17:42
yrlnry	Absolutely. You can use git reset --hard <<commit>>	17:42
fr0sty	Tobias\|: what did you do? (what git commands)	17:42
parasti	if you used git revert, then you should NOT be getting fast-forward errors	17:42
yrlnry	where <<commit>> is the old commit that has only those changes that you actually want to have in the history.	17:42
	→ Evious joined	17:42
jast	note that reset --hard will truncate all history after that commit	17:42
	→ __name__ joined	17:43
	→ banisterfiend joined	17:43
Tobias\|	edited a file, git committed that file	17:44
	git reverted	17:44
	Then tried to push to it	17:44
gebi	Tobias\|: gitk --all is usually a good tool to get a high-level overview about your repo and your upstreams	17:44
jast	a "git revert" doesn't cause fast-forward problems. perhaps someone else pushed stuff before you did?	17:44
gebi	including a git fetch before to sync lastest upstream changes to your local repo	17:44
	→ tizbac joined	17:44
	← banisterfiend left	17:45
	→ nohk_ joined	17:45
	→ tstclair joined	17:45
fr0sty	Tobias\|: git remote update; git status	17:45
	what does it say about the upstream (if anything)?	17:45
	← galderz left	17:47
	→ opalka joined	17:47
	→ steph021 joined	17:47
	← steph021 left	17:47
	→ steph021 joined	17:47
	← nohk left	17:48
	nohk_ → nohk	17:48
Tobias\|	I just used `git pull` and it worked	17:49
	How bizzare :(	17:49
jast	not unexpected, that	17:49
	← g0bl1n left	17:50
	→ giallu joined	17:50
	← guardian left	17:51
	→ bcardarella joined	17:52
	← davidfetter_vmw left	17:52
	→ banisterfiend joined	17:53
	← hyperair left	17:53
	→ jarrod_ joined	17:54
	← jarrod_ left	17:54
	→ jarrod_ joined	17:54
	← ayanes left	17:54
	→ kipras joined	17:54
	← KU0N left	17:54
	→ towski joined	17:55
	→ vdv joined	17:56
	Xeross\|AFK → Xeross	17:56
	→ bcardarella_ joined	17:57
	← artefon left	17:57
	← bcardarella left	17:57
	→ bcardarella__ joined	17:57
	← bcardarella__ left	17:58
	← Bruce_Wayne left	17:58
	← bcardarella_ left	17:58
	→ KU0N joined	17:58
	→ bcardarella joined	17:59
	← bcardarella left	18:00
	← ereslibre left	18:00
	← teweWork left	18:01
	→ NfNitLoop joined	18:01
	← chrisred left	18:04
	→ vmiklos joined	18:04
	→ justin-george joined	18:04
	← ShooterMG left	18:05
	← marcob left	18:06
	→ WebDragon\|laptop joined	18:10
	← rsewill left	18:10
	→ carutsu_ joined	18:11
	→ phantomcircuit joined	18:12
	→ smik``` joined	18:12
	← WebDragon left	18:13
	← smik`` left	18:14
	→ Eiler joined	18:14
	WebDragon\|laptop → WebDragon	18:15
	→ tlvb joined	18:15
	← iaj left	18:15
	← slonopotamus left	18:17
	← justin-george left	18:17
	→ bcardarella joined	18:18
	→ marvin24_DT joined	18:18
	→ Dave^\| joined	18:18
	← Schoentoon left	18:18
	→ lhz joined	18:19
	→ Schoentoon joined	18:19
	← itamarjp left	18:19
	← shentino left	18:20
	→ Ruudjah joined	18:22
	← Schmallon left	18:23
	← bcardarella left	18:24
	→ Naikrovek_ joined	18:24
	ChanServ set mode: +v	18:24
marvin24_DT	Hi!	18:24
	what is the fastest way to change a patch date?	18:24
	there is a commit in linux kernel tree (12ca45fea91cfbb09df828bea958b47348caee6d) which fell out of the futur	18:24
	← Naikrovek_ left	18:24
marvin24_DT	and I want to push it into past agai	18:24
	the problem is that a rebase takes hour	18:24
	→ bcardarella joined	18:24
	← bcardarella left	18:24
marvin24_DT	*hours	18:25
	← Naikrovek left	18:26
marvin24_DT	... and stops with:	18:26
	Rewrite af25e94d4dcfb9608846242fabdd4e6014e5c9f0 (4198/212361)fatal: empty ident <> not allowed	18:26
	could not write rewritten commit	18:26
	grrr	18:26
	← frakturfreak left	18:26
	← philthno2 left	18:27
	← gusnan left	18:27
	→ philthno2 joined	18:28
	← Schoentoon left	18:28
fr0sty	marvin24_DT: what are you trying to accomplish? (and why it is important?)	18:28
	→ Schoentoon joined	18:29
	← envi left	18:29
	← eletuchy left	18:29
marvin24_DT	answer one: change the patch date of commit	18:29
	answer two: don't ask	18:29
	or look at http://gitorious.org/ac100	18:30
	→ constantc joined	18:30
	→ sh1mmer joined	18:31
	→ envi joined	18:31
fr0sty	marvin24_DT: if this commit is already in the linux source tree (or has been published anywhere else 'authoritative' any work you do in your 'corrected' repository will be of little value.	18:31
	← Dave^\| left	18:32
fr0sty	you will have a parallel, but non-identical history which will make it very difficult to interact with other repositories.	18:32
	→ Dave^\| joined	18:32
fr0sty	that being said: you probably have an issue with your .mailmap file causing the empty ident.	18:33
	← Schoentoon left	18:34
	→ marc_kdab joined	18:34
marvin24_DT	I checked that the author line is empty for this commit	18:35
	→ codespectator joined	18:35
marvin24_DT	see af25e94d4dcfb9608846242fabdd4e6014e5c9f0 in kernel tree	18:35
	← nohk left	18:35
	← envi left	18:35
marvin24_DT	I know it is a bad idea to change the commit, but I don't know the correct solution	18:35
	does git store two kind of dates?	18:36
	on for patch creation and one for commit date?	18:36
	how to extract the commit date?	18:36
fr0sty	git log --format=..	18:39
	← Cromulent left	18:39
fr0sty	%ad is authordate	18:39
	%cd is committer date	18:39
	the commiter date of your funky commit seems only partially in the future...	18:39
	heh,	18:39
	nevermind	18:39
	wrong year...	18:40
	nov 30 2009	18:40
	← ignatp left	18:40
fr0sty	not 2010	18:40
	→ Schoentoon joined	18:41
marvin24_DT	the authordate is wrong	18:41
	and gitorious sort by authordate	18:41
fr0sty	and... ?	18:43
	you want the gitorious authors to change the behavior?	18:43
	→ Zhonghua joined	18:43
fr0sty	you want the linux kernel people to rewrite the last 12 months of history?	18:43
	→ xiong joined	18:44
	← curtana left	18:44
gebi	marvin24_DT: please post the link to your patch submission on LKML :)	18:45
	→ guardian joined	18:45
marvin24_DT	well, they told me to ask here ...	18:46
	ping ping game	18:46
	pong	18:46
	← bluenovember left	18:46
	← yairgo left	18:47
	← Eiler left	18:50
	← psoo left	18:50
	→ khmarbaise joined	18:51
	← khmarbaise left	18:51
	→ ReleaseX joined	18:51
	→ khmarbaise joined	18:51
	← kpreid left	18:52
	→ tomsdale joined	18:52
	→ bronson joined	18:52
	← avinashhm left	18:53
	← Manifesto left	18:54
fr0sty	they?	18:55
	→ saschpe joined	18:55
	→ Cromulent joined	18:55
	→ Eiler joined	18:56
	→ kpreid joined	18:56
	← gebi left	18:57
marvin24_DT	#gitorious	18:57
	→ fedesilva joined	18:57
fr0sty	marvin24_DT: how is gitorious getting its list of 'recent' commits?	18:58
	s/commits/Activities/	18:58
	→ j0ran joined	18:58
	→ PerlJam joined	18:59
	→ ppawel joined	18:59
cbreak	hooks?	18:59
marvin24_DT	maybe here: http://gitorious.org/gitorious/mainline/blobs/master/app/models/event.rb	18:59
	← nonno left	18:59
marvin24_DT	I'm a git newbe	18:59
	will try my luck on gitorious channel	19:00
ppawel	hey folks, I want to pull a rather big SVN repo to local git using git-svn but I don't want to download contents of src/tags/*; I used git svn fetch --ignore-paths=^src\/tags but it seems to be downloading it anyway?	19:00
bremner	marvin24_DT: the question is what you are trying to accomplish. If it is a purely cosmetic issue, then rewriting history is probably not the answer	19:00
engla	marvin24_DT: did you do the commit yourself that has the wrong date?	19:00
marvin24_DT	bremner: yes, that what I'm also thinking	19:01
engla	marvin24_DT: if it's not your recent commit then don't do it, it's pointless. as you said -- the rebase takes forever, you're writing a whole new history for the kernel and it's useless.	19:01
marvin24_DT	engla: no, someone else	19:01
SethRobertson	http://blog.patrickcrosby.com/2008/01/30/git-fatal-empty-ident-git-example-com-not-allowed.html	19:01
ppawel	so - does --ignore-paths exclude paths from downloading or exclude them from being imported to git?	19:01
	→ napster_123_ joined	19:02
marvin24_DT	the ident looked more like "<>"	19:02
	← j0ran left	19:03
marvin24_DT	commit af25e94d4dcfb9608846242fabdd4e6014e5c9f0	19:03
	Author: <>	19:03
	Date: Fri Jul 1 23:27:00 2005 -0700	19:03
SethRobertson	marvin24_DT: I have to imagine the problem is still the same. You should check it anyway.	19:03
	← codespectator left	19:03
fr0sty	marvin24_DT: those are separate commits...	19:03
	← amerine left	19:03
marvin24_DT	yes, lot of problems accumulated over the years	19:04
fr0sty	the future commit was from Daniel Vetter and Eric Anholt	19:04
	← napster_123 left	19:04
	napster_123_ → napster_123	19:04
	→ PerlPilot joined	19:04
jim	if you change any of that stuff, you end up changing the shas of the commits	19:04
SethRobertson	ppawel: what command are you using --ignore-paths with?	19:04
ppawel	SethRobertson, git svn fetch	19:04
	→ Eaven joined	19:04
	→ Manifesto joined	19:04
fr0sty	that 'blank' one involved [email@hidden.address]	19:04
marvin24_DT	fr0sty: maybe I'm the first one to rebase the kernel tree	19:04
jim	if others are already using the code, this is NOT something you should be doing	19:05
	→ yairgo joined	19:05
	← Xeross left	19:05
jim	if your rebase will affect others, that is	19:05
	→ Gitzilla joined	19:05
fr0sty	marvin24_DT: not unlikely. it is like affixing wheels to a tomato.	19:05
	→ artefon joined	19:05
fr0sty	tine consuming and completely unnecessary.	19:05
jim	some tomatoes roll pretty good already...	19:05
fr0sty	precicely...	19:06
	→ frogonwheels_ joined	19:06
	→ flazz joined	19:06
	← vitor-br left	19:06
	← PerlJam left	19:06
	→ Xeross joined	19:07
fr0sty	marvin24_DT: if all you want to do is clean up the gitorious view, get them to tell you what command they use to get a list of commits and we can go from there.	19:07
	← PerlPilot left	19:07
fr0sty	probably adding --before=<today> would be sufficient.	19:07
	→ shorg joined	19:07
	→ PerlJam joined	19:07
marvin24_DT	looking at the first lines of http://gitorious.org/gitorious/mainline/blobs/master/app/models/event.rb	19:07
	it seems there is no order at all	19:07
	← Eiler left	19:09
	→ WebDragon\|laptop joined	19:09
	→ eletuchy joined	19:09
fr0sty	you need to find where they get their list of events. (I know jack about ruby, btw)	19:09
	→ bartek joined	19:10
bartek	Hi. Assuming I just have the .git folder from a git repo, how can I get the current HEAD of the working code into my directory?	19:10
fr0sty	the latest* functions and their sql might be the place to start.	19:10
	trim out items from the mysterious future.	19:10
	→ fisch_ joined	19:10
fr0sty	bartek: git reset --hard	19:10
bartek	ah, thanks	19:11
	← WebDragon left	19:12
	→ PerlPilot joined	19:13
	← eletuchy left	19:13
	← Belna_ left	19:14
	← tjholowaychuk left	19:14
	→ itamarjp joined	19:15
jim	bartek: curious, how did you get just the .git?	19:15
	WebDragon\|laptop → WebDragon	19:16
	← fisch_ left	19:16
	← opalka left	19:16
	← PerlJam left	19:16
	→ jmotta joined	19:16
	← philthno2 left	19:16
jmotta	hi .. why when i type "git commit -a" i get lots of this "At EOF" messages?	19:16
	← dpino left	19:16
	→ frakturfreak joined	19:18
	→ vitor-br joined	19:18
bartek	jim: I had an archive of just the .git portion of the repo to move over to a new location, just want the working code now	19:18
	← PeaMan left	19:18
	→ NickyP joined	19:19
	→ andyburke joined	19:20
	← Gitzilla left	19:20
	→ Anti-X joined	19:21
	← khmarbaise left	19:21
	→ thierryp joined	19:21
jim	bartek, oh, ok... that's one way (incl. the reset--hard) to do it...	19:21
bartek	Yup, I got it.	19:21
	Also woah .. is it just me or is filter-branch way faster on a Linux machine?	19:22
	→ khmarbaise joined	19:22
jim	as opposed to a toaster?	19:22
bartek	On my Macbook Pro, it's about half a second for each rewrite (I'm doing the entire history), on a Linux box with comparable (even lesser specs) it's blazing	19:22
	→ j0ran joined	19:22
	← sh1mmer left	19:22
fr0sty	jmotta: can you be more specific? what does 'git status' tell you?	19:22
	← Cromulent left	19:24
	← khmarbaise left	19:25
jim	bartek: dunno :) interesting tho	19:26
bartek	heh, ya	19:26
	← smik``` left	19:27
	← msieradzki left	19:27
	→ smik``` joined	19:27
	→ mutex joined	19:29
	← rraasch left	19:29
	→ Gitzilla joined	19:30
	→ apetrescu joined	19:31
	← thelazydeveloper left	19:31
	→ psoo joined	19:32
	← constantc left	19:32
	→ kar8nga joined	19:33
	→ g0bl1n joined	19:36
	→ rsewill joined	19:36
	→ MattDiPasquale joined	19:36
	→ yngress joined	19:38
	→ Medo42 joined	19:39
	→ zenom joined	19:40
	→ sh1mmer joined	19:40
	→ denysonique joined	19:41
	ChanServ set mode: +v	19:41
	→ msieradzki joined	19:41
	→ gusnan joined	19:43
Medo42	Good evening. What's the usual convention for author information if you don't want to give an email address? I've been using "Authorname <>" so far but apparently some git tools (e.g. rebase) don't like that and become upset because they expect a valid email address.	19:44
	← doener left	19:44
	← kpreid left	19:45
	← yngress left	19:45
	→ woer joined	19:45
	← rsewill left	19:45
	← woer left	19:45
	← artefon left	19:46
	→ artefon joined	19:46
	→ dreiss joined	19:46
engla	that's bad.	19:47
	Medo42: just a word should work instead of email	19:47
	← jmotta left	19:47
engla	I mean like Authorname <medo>	19:47
	→ rsewill joined	19:47
Medo42	Let me try that.	19:47
	→ brosner joined	19:48
	← j0ran left	19:49
	→ anr78 joined	19:49
	← tatsuya left	19:49
	← brosner left	19:50
	← mishok13 left	19:50
anr78	I've converted an svn repo to git, and setup hosting on a server using gitolite. All is well, but I was wondering if there are any "common tasks" I should do on the server. Some hooks that should be enabled or something like that.	19:50
	← aah left	19:53
	→ thiago__ joined	19:53
	→ John___ joined	19:53
	← artefon left	19:54
wereHamster	anr78: no	19:54
FauxFaux	Uh, no non-fast-forward merges? Monthly repack -adf?	19:55
	FauxFaux wonders why wereHamster disagrees.	19:55
	← nairb774 left	19:55
Medo42	Ok, single word as email works, thanks.	19:55
	→ Eridius joined	19:55
	→ ignatp joined	19:55
wereHamster	I never said I disagree...	19:56
	← banisterfiend left	19:56
anr78	FauxFaux: that first one was the reason I asked. I remember reading something about it :)	19:56
FauxFaux	non-ff-merges seems pretty critical to me. :)	19:56
	→ pantsman joined	19:57
bremner	but also the default	19:57
	→ friskd joined	19:57
bremner	well, maybe not. But in gitolite you say whether you want non-ff merges in the config file already	19:58
FauxFaux	Ah, cool.	19:58
	What other single commands are there destroy data? git reset --hard foo && git gc --prune=now # is two.	19:58
	rm -rf .git	19:59
parasti	non-ff ref updates, you mean? not merges	19:59
bremner	yeah, good point	19:59
	← _psychic_ left	19:59
	→ eletuchy joined	20:00
	→ nineteeneightd joined	20:00
	← MikhX left	20:00
	→ alester joined	20:03
	→ Pacman_Home joined	20:03
	→ banisterfiend joined	20:04
	→ _psychic_ joined	20:04
	← Zhonghua left	20:04
	← Medo42 left	20:05
	→ Textmode joined	20:05
	← John___ left	20:06
	→ foocraft joined	20:06
	→ jasonn1234 joined	20:06
kanzure	still doing an svn->git conversion.. what does "Moved remotely" mean in the context of svn?	20:07
FauxFaux	svn mv will operate on urls.	20:08
	← gusnan left	20:08
FauxFaux	svn mv svn://foo/bar/baz svn://foo/bar/quux # etc.	20:08
	← drecute left	20:09
kanzure	FauxFaux: so in foo/bar/baz will it leave a commit message saying "Moved remotely"?	20:09
FauxFaux	There isn't a foo/bar/baz as it's gone..	20:10
	→ nairb774 joined	20:11
	← jrmuizel left	20:11
	→ ggreer joined	20:11
	→ smolyn joined	20:12
ggreer	I've got a machine with git 1.6.3.3, which doesn't support a lot of git status options. I want to write a script that complains if there are any modified files in the repo	20:14
	with 1.7, I'd just do `git status --porcelain --ignore-submodules --untracked-files=no \| wc -l`	20:14
	but 1.6 doesn't support some of those options	20:14
	is there any way to get 1.6 to output similar to --porcelain?	20:15
	← ph^ left	20:16
	→ rajeshsr joined	20:16
rajeshsr	hi all	20:16
	← xiong left	20:16
	→ WebDragon\|laptop joined	20:17
rajeshsr	Just wondering why should i do git rebase at all. The main reason i use git is checkpointing. If I rewrite my history, I can't go to a previous state and check how a bug was introduced etc..	20:17
	Can anyone give some insight on this?	20:17
kanzure	FauxFaux: if that's true then why does this svn repo still have branches like that? for instance, in branch xyz there's a few hundred commits and then the latest one is "Moved remotely" where everything is deleted	20:18
	→ AviZ__ joined	20:18
	→ drizzd joined	20:18
	→ stringoO joined	20:19
	← AviZ__ left	20:20
	← WebDragon left	20:21
	→ AviZ__ joined	20:21
	← AviZ__ left	20:21
	← drizzd_ left	20:21
	→ AlexC_ joined	20:22
AlexC_	morning	20:22
	→ justin-george joined	20:22
engla	rajeshsr: with rebase -i you can treat your work in progress series of commits like a patch stack, you can reorder and combine or split patches. So (for example) it's a very convenient tool to prepare a patchset before you submit/review/merge it	20:22
AlexC_	how can I use 'git log' to search all branches?	20:22
ggreer	this is what I'm using for now: git status --untracked-files=no \| grep "modified:" \| wc -l	20:23
	:/	20:23
	→ AviZ_ joined	20:24
rajeshsr	englas: So rebase is the ugly way Mathematicians present their proofs, without giving any insight as to how they arrived their at all! And if I don't care about "cleanliness", I can live with merge?	20:24
fr0sty	AlexC_: glancing at man git-log seems to imply that --branches would show all branches.	20:24
jast	AlexC_: the 'git-log' manpage can be found at http://git.or.cz/man/git-log [automatic message]	20:24
engla	rajeshsr: it's one way to use rebase, and yes, you can make it look like you never make mistakes.	20:24
	rajeshsr: to its defense, a clean patch stack is much easier to review.	20:25
AlexC_	fr0sty thanks :)	20:25
rajeshsr	engla: In this case i happen to be the solel user of git, while the main project is using some other versioning system! So, thats not a main concern!	20:26
engla	rajeshsr: it still sounds like you are not the sole contributor to the project	20:26
	WebDragon\|laptop → WebDragon	20:27
	→ Eiler joined	20:27
rajeshsr	engla: yes, there are more people..	20:27
	→ lemonchicken joined	20:27
engla	I mean git is just a way to organize the changes you do. Reviewers (if any) benefit from easier to review patches and it doesn't matter which VCS they use.	20:27
rajeshsr	Just wondering if people don't care about preserving their "private" branch history and lose it to rebasing? Or you can do reabse on fresh branch which you will push upstream while keeping your private branch with you?	20:28
	← AviZ_ left	20:28
	→ AviZ_ joined	20:28
engla	rajeshsr: I rebase & forget :-)	20:29
rajeshsr	engla: well, when submitting diffs, it is an absolute diff between the current state of the code in the server and the one on HEAD of the branch I choose to submit. So I don't see much incentive in rebasing in my case!	20:29
engla	rajeshsr: then you are using a more primitive work flow(!)	20:29
rajeshsr	engla: So, what could be a better way?	20:30
engla	you create a new branch for the feature you are working on: git checkout -b featureX	20:30
rajeshsr	yes	20:30
engla	then you commit on that branch, then it's done and you use rebase -i to clean up and combine patches	20:30
	← hebz0rl left	20:30
	← ppawel left	20:30
rajeshsr	why not just a merge! :)	20:30
engla	then you use format-patch to generate a series of patches for the branch that you submit for inclusion	20:31
	then the reviewer reads the patches one by one in order and they will be easy to review	20:31
rajeshsr	oh, i think i miss format-patch!	20:31
	→ programble joined	20:31
	← programble left	20:31
	→ programble joined	20:31
engla	easy features are just one patch, complicated stuff is divided up	20:31
jpr5	I think people use rebase way too much, making more problems for themselves than necessary.	20:32
	→ Awww2 joined	20:32
rajeshsr	well, i have a master branch which always keeps with the upstream and merge from it to my current feature	20:32
wereHamster	jpr5: rebasing alone does not make problems	20:32
	← marvin24_DT left	20:32
rajeshsr	and finally generate the diff the hard way! :)	20:32
	← AviZ_ left	20:32
rajeshsr	engla: can format-patch help me?	20:33
	← Awww2 left	20:33
ggreer	are there any plans to move toward more straightforward commands in git? for example, as was mentioned above, to create a branch you have to run git checkout -b branchname and not git create-branch branchname	20:33
engla	rajeshsr: format-patch is gold in a git-only workflow. However its patches can be applied by standard tools such as patch too	20:33
ggreer	or something like that	20:33
bremner	ggreer: nothing prevents you from making a script	20:33
	→ aviz_ joined	20:33
jpr5	wereHamster: of course not. but it's easy to argue that it introduces the average user to such a wide range of possibilities to break things versus merging.	20:33
bremner	or an alias	20:33
parasti	ggreer: use git branch branchname if that bothers you	20:33
engla	ggreer: no, to create a branch you say "git branch branchname" :-)	20:34
wereHamster	jpr5: s/it/git/ ...	20:34
jpr5	wereHamster: no, it == rebasing	20:34
	→ vtts joined	20:34
	← acs left	20:34
jpr5	wereHamster: merging is easy, and that's supposed to be the point	20:34
ggreer	ah, that makes more sense. I've been bit by checkout quite a few times	20:34
rajeshsr	engla: hmm! So in this approach am not missing anything which rebase can give me, right? :)	20:34
parasti	ggreer: although I won't deny that git UI leaves much to be desired :P	20:34
	← ignatp left	20:34
	→ kpreid joined	20:34
ggreer	git checkout is a combination of svn revert and svn branch	20:34
wereHamster	jpr5: I got that. My point was that git as a whole can sometimes be overwhelming (if the user is not willing to invest some time to learn it)	20:35
	→ RobertMeta joined	20:35
	← RobertMeta left	20:35
	→ RobertMeta joined	20:35
engla	rajeshsr: well I don't know. I work the way I told above even if it's for my own project.	20:35
	→ juanjoc_ joined	20:35
parasti	I didn't even know svn branch existed	20:35
	isn't it just cp	20:36
	← aviz_ left	20:36
jpr5	wereHamster: right, but if they're going to invest that time, it should be invested in the easy stuff. I've watched an inordinate # of questions about rebasing clearly from people who haven't yet mastered basics.	20:36
ggreer	err, svn cp	20:36
engla	ggreer: except svn branches and git branches are very strange cousins	20:36
	← Ruudjah left	20:36
parasti	ggreer: well I honestly don't know, I wasn't trying to correct you :)	20:36
	→ aviz_ joined	20:37
rajeshsr	engla: strange how people are happy with letting their history of commits get erased! :) It is because am able to checkpoint I use git mainly, for I tracked many regressions very easily with it!	20:37
aviz_	Hi All	20:37
	← Tommy[D] left	20:37
ggreer	engla: yes. it reminds me of how cassandra uses relational database terms	20:37
	← g0bl1n left	20:37
aviz_	Anyone here is using git over ssh with kerberus auth ?	20:37
wereHamster	aviz_: does ssh alone work?	20:37
	← juanjoc left	20:38
	← MetaCosm_ left	20:38
engla	rajeshsr: it's interesting yes. I want to say that "history is important" -- and now I mean that it's easy to read later, that is, the history should be cleaned-up they way I do it. And on the other side you say you want the untweaked history	20:38
	← napster_123 left	20:39
rajeshsr	engla: Difference of perspectives, having same goal, leading to different decisions! :)	20:39
	→ makubi joined	20:39
	→ Tommy[D] joined	20:39
aviz_	wereHamster: It all works.. i am have issue with windows users who need to access the repository though. i am using plink+kerb for the ssh transport. but forsome reason plink is too slow for the initial authentication.. wondering if someone had the same problem and have a soloution for it.	20:40
ggreer	engla: if anything, the similar terminology has made it harder to learn git	20:40
	← Eiler left	20:40
	→ Eiler joined	20:41
anr78	no non-ff merges essentialy means that people are not allowed to rewrite history for already pushed stuff?	20:41
	→ kylehayes_ joined	20:41
bremner	anr78: non-ff pushes means that	20:41
	s/non-ff/no non-ff/	20:41
	→ darkredandyellow joined	20:42
anr78	bremner: ok, and that is something I want to turn on I suspect	20:42
bremner	yes	20:42
	← phantomcircuit left	20:43
	→ dl9pf_ joined	20:44
anr78	and no non-ff merges is something else I should also turn on? I've only used git-svn until now, so I feel a bit nervous about the job of setting this up at work :)	20:44
	← itamarjp left	20:44
bremner	non-ff mergers are fine, and sometimes needed.	20:44
	← dl9pf left	20:45
guardian	hello	20:45
	is git supposed to find p4merge automagically on mac osx?	20:45
bremner	non-ff merge just means a new commit is created, possibly to resolve some conflicts	20:45
guardian	i set difftool to p4merge but i get The diff tool p4merge is not available as 'p4merge' when trying to launch git difftool —tool=p4merge	20:45
	p4merge is in /Applications/p4merge.app/Contents/MacOS/	20:46
anr78	bremner: ah, yeah, I want that :)	20:47
	← Eiler left	20:48
bremner	anr78: some people go to lengths to avoid merge commits using rebase, but sometimes that is not possible/sensible	20:48
	← FabianB__ left	20:49
	→ mtx_init joined	20:50
	→ aah joined	20:51
	→ FabianB__ joined	20:51
	← bronson left	20:52
	→ theoros joined	20:53
	← theoros left	20:53
	→ theoros joined	20:53
	← RobertMeta left	20:53
anr78	ok. I just want to turn on common safety nets, and figure out the rest as we go along.	20:54
	→ amerine joined	20:54
	← rajeshsr left	20:54
	← jasonn1234 left	20:54
	← vitor-br left	20:57
	← stamina_ left	20:58
	→ vitor-br joined	21:01
	← nadim_ left	21:02
	← vitor-br left	21:03
	→ MikhX joined	21:03
	→ welly joined	21:04
	← jelmer left	21:05
	→ ignatp joined	21:05
	→ jelmer joined	21:05
	→ vitor-br joined	21:06
	→ thiago_home joined	21:07
	← hendry left	21:08
	← chrisshattuck left	21:10
	← darkredandyellow left	21:11
	← kar8nga left	21:11
	→ darkredandyellow joined	21:11
	→ Eiler joined	21:13
	← parasti left	21:14
	← stringoO left	21:14
	← aresnick left	21:17
	← eletuchy left	21:17
	← Eiler left	21:17
anr78	and external backup could be done by creating a bare clone on my usb drive?	21:18
NfNitLoop	yep.	21:19
	or, more easily, on some other networked machine.	21:19
	at which point you could automate it. :)	21:19
	→ shentino joined	21:19
	← jstemmer left	21:20
	→ Schmallon joined	21:20
	← alester left	21:22
anr78	:)	21:22
	← Xeross left	21:22
SethRobertson	A backup might be good to use --mirror with, though clone in general will not copy the .git/config settings of course	21:23
	→ parasti joined	21:24
kanzure	so, i'm converting an svn repository to git.. the svn repo was originally cvs until cvs2svn was used	21:24
	i've found a case where history is lost, but i can find it in the git repo	21:25
	http://diyhpl.us/cgit/nanoengineer-fixed/commit/?h=Distribution&id=d7a2ecb79487035ac542e4fcaecf2e63a2e14714	21:25
	http://diyhpl.us/cgit/nanoengineer-fixed/commit/?h=master&id=fdee9debedc807b3af5d3d3e8cbd780881fe8ee0	21:25
	how can i fix the git history so that it understands that this is the same commit? or something	21:25
	→ simplechat joined	21:25
	← simplechat left	21:25
	→ simplechat joined	21:25
SethRobertson	You could do a `git rebase -i` and squash it	21:26
	Only if no-one else is using the repo, and you have a backup	21:26
anr78	SethRobertson: --mirror so that if I ever need the backup it will be ready to push to the right upstream?	21:27
	← ericenns left	21:28
SethRobertson	anr78: No, as I said the .git/config is not mirrored. However, the remote heads will be saved.	21:28
anr78	ah, so that was what you said :) thanks	21:29
SethRobertson	You could of course manually copy the .git/config when you do the mirroring clone (or fetch update). However, there are other files which you might also want to copy (hooks, info/*, and possibly others)	21:29
	zorzar_ → zorzar__away	21:29
	← zorzar__away left	21:30
piotr_	please I need help to recover lost data, how do I apply lost stash when I have id?	21:30
SethRobertson	anr78: Another option is to just copy the entire directory tree, (best done if you can assure an exclusive lock, even though in theory git will not have problems if it does change)	21:30
	→ xiong joined	21:31
	→ RobertMeta joined	21:31
	← RobertMeta left	21:31
	→ RobertMeta joined	21:31
SethRobertson	piotr_: What "id" do you have?	21:31
anr78	SethRobertson: the reason I thought about copying was that I was nervous about locking.	21:31
piotr_	hash	21:32
SethRobertson	piotr_: And specifically, does `git show <id>` show your missing change?	21:32
piotr_	272a75e7a0e9629e6ef542497fc63276c5782554	21:32
	yes	21:32
	but I don't know how to apply it	21:32
	git statsh apply 272a75e7a0e9629e6ef542497fc63276c5782554 shows something like git status	21:33
SethRobertson	piotr_: `git show <id> \| patch -p1` should work, there are almost certainly other methods as well (you could probably merge it)	21:33
	→ aphelion joined	21:35
SethRobertson	Hmm. `git show` isn't producing patch-standard patch info. Odd.	21:35
	→ level09 joined	21:35
	← ggreer left	21:36
aphelion	is there an option or configuration that prevents people from committing files with conflicts? or pushing files with conflicts to remote? or a post-receive hook that can check for files with conflicts and tell the person doing the push to stop being a retard?	21:37
	→ stringoO joined	21:37
	→ nimrod10 joined	21:37
cbreak	aphelion: it is impossible to commit files with conflicts anyway	21:38
aphelion	a coworker of mine, every damn time there's a conflict in this css file, uses vimdiff and doesn't see/fix every conflict... so i pull and get a css file filled with >>>>> HEAD garbage	21:38
cbreak	because to commit something you have to add it	21:38
	← spaceonline left	21:38
	→ zorzar joined	21:38
cbreak	and add resolves the conflict	21:38
aphelion	cbreak: right. see above.	21:38
	← ignatp left	21:38
aphelion	i should have worded it differently	21:38
Eridius	cbreak: if you commit a file that has the commit markers in it, e.g. >>>>>, then I have no problem saying that you "commited a file with conflicts"	21:38
parasti	I thought your wording was clear :P	21:38
SethRobertson	piotr_: However, `git merge --squash <id>` will work	21:38
aphelion	i mean prevent you from adding/committing/pushing files that have the merge/diff3/etc format conflict resolution text in them	21:38
cbreak	hmm.	21:38
	maybe with a hook	21:39
	you could reject stuff that contains certain patterns	21:39
	do you control a central repo?	21:39
	or can you force him to instal the hooks?	21:39
aphelion	yes	21:39
Eridius	aphelion: you could just educate your coworker to stop commiting things that he hasn't actually compiled	21:39
aphelion	no	21:39
	→ Xeross joined	21:39
aphelion	Eridius: you dont think i've tried that?	21:39
	Eridius: several times in the past week?	21:39
	you'd think it's a pretty basic concept	21:40
	dont push broken shit	21:40
	dont commit broken shit	21:40
cbreak	do you have several coworkers?	21:40
aphelion	but this is a rant for another time	21:40
cbreak	if so, maybe naming&shaming helps :)	21:40
aphelion	i just want a tool that will gently point out the error to him so that he can fix it before pushing to remote	21:40
cbreak	aphelion: http://progit.org/book/ch7-3.html	21:41
	that's the general idea	21:41
aphelion	negative reinforcement never works	21:41
	→ nadim_ joined	21:41
parasti	next you know, he's in here asking what the problem is and how to disable it	21:41
	rayners → rayners_afk	21:41
aphelion	lol	21:41
	→ Yuffster joined	21:42
Eridius	you also don't want to train him to use `git push` as a way to validate his work	21:42
aphelion	Eridius: that's pretty depressing	21:43
	← zombor left	21:44
	← RobertMeta left	21:45
	← KU0N left	21:45
	← abetaha left	21:46
	→ KU0N joined	21:47
	← AlexC_ left	21:50
	→ mfacenet joined	21:50
	← brntbeer left	21:51
	← skoop left	21:51
	→ brntbeer joined	21:51
	→ Eiler joined	21:51
	← thierryp left	21:51
	← Remixman left	21:52
	← anr78 left	21:52
	← jaql left	21:53
cbreak	aphelion: pre-commit hook :)	21:53
	I am quite sure you can build something that rejects big successions of >>>>>> and <<<<<<<	21:53
	→ bronson joined	21:54
	→ RobertMeta joined	21:54
	← RobertMeta left	21:54
	→ RobertMeta joined	21:54
SethRobertson	I was of the impression that git rejected those by default, though	21:54
	← darkredandyellow left	21:55
	← Dave^\| left	21:55
	→ eletuchy joined	21:55
cbreak	I think the default hook only rejects based on whitespace, and it's not enabled	21:55
SethRobertson	Yeah, I was wrong. It was something we manually added as hooks	21:57
	→ alberto56 joined	21:58
	← _psychic_ left	21:58
	→ codespectator joined	22:00
	→ ppawel joined	22:00
	→ _psychic_ joined	22:00
	← Ryback_ left	22:00
ppawel	folks, I'm using git svn to clone a repo's trunk. can I fetch branches later?	22:01
	I mean will git add it to history properly?	22:01
	← PerlPilot left	22:01
cbreak	no	22:01
	→ PerlJam joined	22:01
	→ Buglouse joined	22:03
	→ yatiohi joined	22:04
	← Yuffster left	22:05
	→ linusoleander_ joined	22:05
	kylehayes_ → kylehayes	22:06
	→ tbf joined	22:07
	← flazz left	22:07
	← thiago_home left	22:07
	← psoo left	22:08
	q0rban → q0rban\|afk	22:08
	→ peterjmag joined	22:09
	← linusoleander left	22:09
	linusoleander_ → linusoleander	22:09
	→ gebi joined	22:10
	→ shwoodard joined	22:10
	→ cbreak joined	22:11
	← mastro left	22:11
guardian	when launching "git difftool" i can see that the diff tool command is being invoked for deleted files with $REMOTE being set to /dev/null — what's the point of invoking the difftool when a file's status is "deleted" or "new file" anyway?	22:11
	← kylehayes left	22:11
	← ppawel left	22:11
ndim	guardian: So that you can see what changes.	22:13
	→ jauler_ joined	22:13
ndim	Or what the difference is.	22:14
guardian	in fact i'm trying to setup p4merge correctly	22:14
	→ poseidon_ joined	22:14
	→ hemmecke joined	22:14
guardian	under windows, it chokes on /dev/null paths	22:15
	← banisterfiend left	22:17
poseidon_	So I'm a freshman in computer science. I get many little one file projects, and I have my own mini projects. I generally organise these by seperating them into respective directories. However, I want a more robust way of keeping track/saving/acessing my work. I figure I should have a git repos on my laptop and on my vps so I can acess it anywhere.	22:17
	→ romy joined	22:18
poseidon_	Would it be best to have one git repository which contains all the different directories in which my projects are stored?	22:18
cbreak	one per project	22:18
	→ ka1ser joined	22:18
cbreak	so one folder per project	22:18
hemmecke	Assume I have to repos A and B. They have no common commit. B contains new code that I now want to put on top of the master branch of A but without losing the history in B. I expect the sha1's of the B to be rewritten, but I want to keep the history? How?	22:18
	→ thiago_home joined	22:18
	← codespectator left	22:18
cbreak	hemmecke: in B, fetch A's master branch	22:19
Eridius	hemmecke: why not just merge the two	22:19
cbreak	then switch to B's master branch	22:19
	and git rebase masterA	22:19
	← sonnym left	22:20
cbreak	a merge is probably better :)	22:20
hemmecke	cbreak: That's unconnected. Also grafts is not what I want.	22:20
cbreak	hemmecke: no connection needed	22:20
Eridius	hemmecke: you can rebase an entire history on top of another with `git rebase --root --onto ...`, but why do you want to rebase? Won't a merge do what you want?	22:20
cbreak	you can fetch the linux kernel into a git.git	22:20
Eridius	or was B created from a snapshot of A?	22:20
hemmecke	Cannot merge, because I will later have to dcommit to an svn repository. :-(	22:20
cbreak	... oh my...	22:20
	→ Scala_ joined	22:20
Eridius	:o	22:20
	cbreak walks backwards slowly	22:21
	← parasti left	22:21
romy	is git cherry-picking and then rebasing the best way to submit changes upstream to a github fork ?	22:21
	cbreak then turns around and runs screaming into an unspecified direction	22:21
hemmecke	Yes I know, but there are people who still keep svn. :-(	22:21
Scala_	I have a folder that says "modified" on git status, but it doesn't show up on github and I can't commit. Any idea?	22:21
Eridius	hemmecke: well, `git rebase --root --onto ...` is probably what you want	22:21
cbreak	Scala_: is it a submodule?	22:21
hemmecke	I'll try. Thanks a lot.	22:21
cbreak	(does it contain a .git folder?)	22:21
Scala_	cbreak: I wasn't indending it to be, let me check	22:21
	oh wow it does	22:22
	let me get rid of that	22:22
romy	specifically, if i'm working on a forked repo and only want the maintainer to pull _some_ of the commits i've made	22:22
SethRobertson	romy: It depends on the sensibilities of the upstream. However, I would imagine that a clean history would never be rejected while a dirty one would sometimes be rejected. However, you do not need to cherry pick and then rebase, you can do it all in one fell rebase	22:22
	← berkes left	22:22
SethRobertson	Ah, well if you only want to do some, then just cherry-pick. I don't see any need for a rebase afterwards	22:23
romy	SethRobertson: the rebase is so that, once it's merged upstream and i pull, there aren't duplicate commits	22:23
	→ banisterfiend joined	22:24
SethRobertson	(Assuming you cherry pick onto a copy of the upstream branch)	22:24
	romy: Unless upstream changes the commits, git will be aware that they are dups.	22:24
	Rebasing is not wrong, of course, just not necessary.	22:24
	→ mjf joined	22:24
romy	SethRobertson: it'll be aware how ?	22:25
	i thought merging didn't do contents-based comparisons, only rebasing	22:26
poseidon_	cbreak: did you mean one repo per project or one folder per project (all in one repository)?	22:26
SethRobertson	romy: well I do `git pull --rebase` when I get new data from upstream, so I suppose that is possible, but I didn't think so. I'll double-check	22:26
cbreak	poseidon_: one repository per project	22:27
	obviously, each repo requres it's own folder	22:27
romy	SethRobertson: yeah, that does the rebase instead of a merge though, and without it I don't think I could avoid dupes	22:27
	if i cherrypick, that is	22:27
	which makes me wonder what everyone else does when they contribute patches back upsteram on, say, github	22:28
cbreak	merging doesn't need content based comparisons	22:28
	all that merging does it one commit with merged content	22:28
	and two parents	22:28
	← yairgo left	22:28
	frogonwheels_ → frogonwheels	22:28
romy	i understand, but do you see my problem wrt cherry-picking commits back upstream ?	22:29
cbreak	rebasing has to apply patches from each commit on top of the previous one	22:29
	once you pull --rebase from upstream, dupes will vanish	22:29
SethRobertson	He doesn't want to pull --rebase from upstream	22:29
	← WebDragon left	22:29
SethRobertson	(or she)	22:29
cbreak	since rebasing doesn't do empty commits	22:29
romy	yes, so the question is... is that the 'best practice' workflow for contributing patches ?	22:29
cbreak	I think so	22:30
	→ rolfb joined	22:30
cbreak	as long as your branches are private	22:30
	← ka1ser left	22:30
	← level09 left	22:30
romy	and if they weren't ?	22:30
cbreak	then... tough look	22:31
	you can't rewrite history	22:31
	→ phantomcircuit joined	22:31
cbreak	so you have to merge	22:31
	← skyl left	22:31
	→ aresnick joined	22:31
cbreak	well... technically you can	22:31
	but make sure that everyone who knows about the branch uses pull --rebase :)	22:31
romy	so, everyone who forks a project (that other people then use) and submits patches upstream	22:31
	has duplicate commits when they merge upstream ?	22:31
	← dbbnj left	22:32
SethRobertson	romy: Yes, unless you `git pull --rebase`	22:32
	← Adaptee left	22:32
SethRobertson	Or you have direct commit privs onto upstream	22:32
	Of course, you can retroactively `git pull --rebase` to make history linear again	22:33
dmlloyd	rebase is the best thing ever	22:33
SethRobertson	When used with understanding and care	22:33
	← pigdude left	22:34
	← eletuchy left	22:34
	→ dbbnj joined	22:34
	← kipras left	22:34
romy	meh	22:34
	→ TeckniX joined	22:34
	Xeross → Xeross\|AFK	22:35
	← Scala_ left	22:38
cbreak	romy: how do you think you can eliminate dupes without rewriting history? :)	22:38
	just think about how git works	22:38
	and you'll see that it's impossible	22:38
	→ Scala_ joined	22:39
	← mfacenet left	22:39
	→ WebDragon joined	22:42
	← HG` left	22:43
	← lemonchicken left	22:43
	← bartek left	22:43
	← pantsman left	22:44
	← WebDragon left	22:46
	→ EricInBNE joined	22:47
ajpiano	hmm	22:49
	so i have a repo that had some bad line endings committed to it and it was screwing all sorts of stuff	22:49
	a fix for those line endings was just pushed	22:49
	but i can't pull the fix, cause the line endings are screwed up and i can't stash	22:49
	← Scala_ left	22:49
ajpiano	what should i do to be able to pull the changes?	22:50
SethRobertson	ajpiano: Do you have a .gitattributes file somewhere?	22:50
	What OS are you on?	22:50
ajpiano	i'm on osx	22:50
	→ Titosemi joined	22:51
ajpiano	i don't think i do, but i might not be looking in the wrong place	22:51
	→ lemonchicken joined	22:52
SethRobertson	find . -name .gitattributes; ls -l .git/info/atributes	22:52
	→ spaceonline joined	22:52
SethRobertson	Also, what are your core.eol, core.safecrlf, core.autocrlf settings?	22:53
dmlloyd	EOL handling is the worst thing ever :P	22:53
ajpiano	hm	22:54
FauxFaux	Indeed. They're binary files, dammit.	22:54
	← LiamH left	22:54
ajpiano	SethRobertson, there was * crlf=input in the .gitattributes file	22:54
FauxFaux	I'd kind of like a git make-files-in-wc-match-repo. ¬_¬	22:54
	→ constantc joined	22:54
	→ heavy joined	22:54
SethRobertson	ajpiano: You can use your .git/info/attributes file to override that setting.	22:54
	ajpiano: man gitattributes	22:55
jast	ajpiano: the 'gitattributes' manpage can be found at http://git.or.cz/man/gitattributes [automatic message]	22:55
SethRobertson	Or you can delete that file, commit, merge in upstream, and then rebase/cherry-pit that file deletion	22:56
ajpiano	i have it on input,ah	22:57
	ddin't think ofcommiting and then removing it after	22:57
	derp.	22:57
SethRobertson	derp?	22:57
ajpiano	i was trying to avoid a commit at all costs.	22:57
	derp @ me.	22:57
	→ ecksun joined	22:57
	← d0k left	22:58
	← mjf left	22:58
	← saschpe left	22:59
	← rolfb left	22:59
ajpiano	hm, modifying the .gitattributes, then stashing the change to .gitattributes worked fine....woo.	22:59
	thanks SethRobertson	22:59
kanzure	gitk --all seems to list more branches than git branch -a.. why is that? or how can i get that more complete list from a command?	23:00
Eridius	kanzure: what is gitk --all showing that git branch -a doesn't?	23:01
	→ schacon joined	23:01
kanzure	some of the older, deleted branches apparently	23:01
engla	kanzure: --all includes tags as well	23:01
kanzure	oh hm	23:01
	that might be it	23:02
engla	as well as HEAD and the stash	23:02
	← hemmecke left	23:02
	← fedesilva left	23:03
	← banisterfiend left	23:03
	→ eletuchy joined	23:03
	→ ph^ joined	23:06
	← bronson left	23:06
	← ph^ left	23:07
	→ axiom joined	23:07
	axiom → majoh	23:07
smolyn	having some overall process trouble-- wondering what people think is a good system: have a product that has a lot of branches (lots of in-the-field stuff we can't change). we're using git describe to automatically set the version in the build. biggest problem is that when merging changes from older branches to newer, this of course brings in commits that are attached to these tags, causing the tags in the other branches to revert. ie. we branch	23:08
	1.1 from master, tag master as 1.2, we then tag a commit on the 1.1 branch to 1.1.1, merge back to master and we get master as 1.1.1	23:08
	has anyone else had similar troubles, and what have you done about it?	23:08
	→ sonnym joined	23:08
	← spuz left	23:08
	← guardian left	23:09
rohanpm	smolyn: what do you mean by "tag master as 1.2" ... do you mean `git tag' or something else?	23:09
smolyn	git tag -a	23:09
	→ bronson joined	23:10
rohanpm	what do you mean you "get master as 1.1.1" ? a tag is a one-way label from name -> SHA1, and so is a branch	23:11
	→ gusnan joined	23:11
	→ banisterfiend joined	23:11
smolyn	right but the commit attached to that tag will make it's way into master when we do a merge	23:12
	← theoros left	23:12
SethRobertson	smolyn: Unless you are rebasing the branch, that isn't going to happen	23:12
	← __name__ left	23:13
	← fowlduck left	23:13
	→ theoros joined	23:13
rohanpm	smolyn: so why is it a problem to merge a commit pointed to by the 1.1.1 tag into the master branch?	23:13
SethRobertson	Or it could be a fast-forward merge	23:14
	smolyn: mkdir foo; cd foo; git init; echo A>A; git add A; git commit -m A; git checkout -b new; echo B>B; git add B; git commit -m B; git tag B -a -m Btag; git checkout master; echo AA>A; git commit -a -m AA; git merge new; gitk --all --date-order	23:14
smolyn	we're definitely not rebasing-- it's possible our developers are forgetting to not do fast-forwards.	23:14
SethRobertson	gitk	23:14
smolyn	rohanpm: git describe starts showing us "1.1.1" instead of "1.2"	23:14
rohanpm	oh, I see...	23:15
	smolyn: well, it seems like you're using the tags as two different things; in your example here, you seem to be using 1.1.1 to mean "the version we released as 1.1.1" and 1.2 to mean "the commit where we consider the 1.2 branch to start development"	23:16
	s/branch/version/	23:16
	→ Schmallon1 joined	23:16
	→ abetaha joined	23:16
smolyn	ya sort of it does seem that way... we tag when we start dev on a version #, but we also need to increment that number in branches sometimes so it gets tagged there too	23:18
	← Schmallon left	23:18
	← brntbeer left	23:18
	→ Scala_ joined	23:18
	← romy left	23:18
	→ Schmallon joined	23:19
	← coderdad left	23:19
	← Schmallon1 left	23:21
	← ceez left	23:21
	← iravid left	23:22
	← abetaha left	23:22
	→ iravid joined	23:22
	← vitor-br left	23:23
	← dbbnj left	23:23
	← friskd left	23:23
	→ Yuffster joined	23:24
	← ExtraSpice left	23:24
	→ dbbnj joined	23:24
	→ friskd joined	23:25
	← banisterfiend left	23:25
	→ vitor-br joined	23:27
	→ abetaha joined	23:28
	→ Smirnov_ joined	23:28
	← Smirnov_ left	23:29
	→ Smirnov_ joined	23:29
	Smirnov_ → Smirnov	23:29
Smirnov	anyone know if theres any graphical git tools like qgit except they let you look at the diffs in the context of the full file (rather than just the patch)	23:29
bremner	can't gitk do that?	23:30
	← lhz left	23:30
pasky	http://git.et.redhat.com/?p=ovirt-server.git;a=blob_plain;f=git-hook/update;hb=refs/heads/vcs-admin	23:30
	wow @ what crazy things people started doing with git too now ;)	23:30
bremner	Smirnov: do you have an example of the kind of view you are looking for?	23:31
	← dbbnj left	23:31
	← sterlo left	23:32
	← _psychic_ left	23:32
SethRobertson	bremner/Smirnov: It would probably have to be an n-way display (for octopus merge) of the new version and all of the previous versions, with colorization.	23:32
Eridius	Smirnov: I usually just use `git difftool` for that. I have it configured to use Kaleidescope (an OS X graphical diff viewer)	23:32
	← msieradzki left	23:32
Smirnov	yeah something like git difftool	23:32
	except its really annoying to use that to look at lots of changesets/files at once	23:32
	→ banisterfiend joined	23:33
	← abetaha left	23:33
Eridius	true. I wish there was a shorthand that meant $FOO^ $FOO	23:33
	→ dark joined	23:35
kanzure	does anyone know where i can find examples for git-commit-tree?	23:35
	← constantc left	23:35
Eridius	kanzure: what's confusing about it?	23:36
dark	I have write access to a remote, bare git server, through http. But I have access just to push. The server has HEAD pointing to refs/heads/master. But the name of my branches are devel and stable. Is there a way to change the HEAD of the remote server?	23:36
	→ _psychic_ joined	23:37
	← mmc left	23:37
dark	I'm not sure, but this is maybe just cosmetic.. I can do git checkout after cloning it	23:37
Smirnov	Eridius: well there is bash-isms like commit{^,} which expands to commit^ commit but thats still typing	23:37
	← linusoleander left	23:38
	← gebi left	23:38
kanzure	Eridius: i'm really trying to do something else entirely	23:38
	← plantian left	23:38
dark	changing _my_ head does not help to change the head of the remote server, after the push. (there is a --tags for git push, why not --head?)	23:38
Eridius	Smirnov: true	23:38
kanzure	Eridius: i have two commits in an svn-imported repo with different id values but they should be documented as being branched (to preserve history)	23:38
	Eridius: http://diyhpl.us/cgit/nanoengineer-fixed/commit/?h=Distribution&id=d7a2ecb79487035ac542e4fcaecf2e63a2e14714	23:38
	Eridius: http://diyhpl.us/cgit/nanoengineer-fixed/commit/?h=master&id=fdee9debedc807b3af5d3d3e8cbd780881fe8ee0	23:39
Smirnov	maybe theres some code review tool for git that isnt web based, that would be nice	23:39
kanzure	(it was moved from branch "Distribution" to master)	23:39
	→ phantomcircuit_ joined	23:39
kanzure	so i was thinking my plan of attack might have to involve git-commit-tree and git-rebase at some point	23:39
	← phantomcircuit_ left	23:40
	← phantomcircuit left	23:40
kanzure	examples for git-commit-tree might help me figure out if that's true or not :/	23:40
	→ phantomcircuit joined	23:40
SethRobertson	kanzure: Didn't I recommend `git rebase -i` to you before, and suggest squashing the two commits together?	23:40
	← ch077179 left	23:41
SethRobertson	I'm not sure what you mean by "documented as being branched" though	23:41
kanzure	SethRobertson: i thought you were talking to someone else. my apologies	23:41
SethRobertson	But, I see that they are committed on different branches. That makes everything rather more difficult to do	23:42
kanzure	SethRobertson: i really don't know what i'm doing. git rebase -i seems like it would come after a lot of other work?	23:42
	← tizbac left	23:42
	← sftp left	23:42
SethRobertson	kanzure: perhaps. But it is no longer clear to me what you are trying to do since the two commits are on different branches	23:43
	← Eiler left	23:43
kanzure	SethRobertson: basically under svn someone moved all of these files into trunk (which is now branch master)	23:43
SethRobertson	Is the "Distribution" branch one that you want to keep? Do you want to undo the "remove" on that branch? I'm not sure what your end goal is	23:44
kanzure	SethRobertson: so ultimately on branch master i'd like the history to follow	23:44
	→ mfacenet joined	23:44
	← xiong left	23:44
kanzure	SethRobertson: currently the head on Distribution is void of all files, true	23:44
	→ fedesilva joined	23:45
	→ Eiler joined	23:45
kanzure	SethRobertson: my goal is to preserve the revision history of these files despite the "jump" that svn decided to do	23:45
SethRobertson	kanzure: Do you have commits after this badness on master ?	23:46
kanzure	SethRobertson: oh yes, many	23:47
	but only on master not on "Distribution"	23:48
	← Schmallon left	23:48
	→ mrtazz joined	23:48
	→ madewokherd joined	23:49
	→ gebi joined	23:50
	→ leroux joined	23:50
SethRobertson	kanzure: Are the (non-deletion) changes that occurred in "Distribution" ones that occurred atomically from the perspective of master (eg. all useful distribution changes occurred between one commit and the next)	23:50
	→ tj_ joined	23:51
	tj_ → tjholowaychuk	23:52
	→ flazz joined	23:52
	← flazz left	23:52
SethRobertson	kanzure: or, for the purposes of your history unification, can we pretend that occurred	23:52
	→ IdahoEv joined	23:52
	→ flazz joined	23:52
kanzure	SethRobertson: i don't understand. Distribution's commit history is fairly vanilla.. i'm sure some commits in Distribution (prior to svn moving everything) involved file deletes	23:52
	← sh1mmer left	23:52
IdahoEv	suppose I have a directory in my repo that is currently a submodule. I want to just make it a plain directory and part of the repo. How should I go about it?	23:53
jpr5	git rm --cached dir/to/submodule	23:53
	mkdir dir/to/submodule && cp -r stuff/that/was/in/it dir/to/submodule/	23:53
	← saysjonathan left	23:54
SethRobertson	kanzure: Basically, I can only think of two ways of doing this. The first method is one I don't actually technically know how to do. Strip off the deletions in Distribution, squash the additions on master into one commit, and then forge a merge to tell git that DIstribution was merged into master during that one squashed commit (the latter is what I don't know how to do)	23:54
kanzure	SethRobertson: http://diyhpl.us/cgit/nanoengineer-fixed/log/?h=Distribution	23:54
jpr5	probably want to nuke the submodule from .gitmodules too; git rm --cached doesn't seem to nuke entries from there	23:54
	← boombatower left	23:54
kanzure	(that link was to show the commit history)	23:54
	→ schambers joined	23:54
	→ kanru joined	23:55
SethRobertson	kanzure: The other method is to rebase to move all of the commits in Distribution onto master branch and then rewrite all of the following master history on top of the new master history. Very very nasty stuff	23:55
kanzure	okay. now at least i have a starting point :)	23:55
SethRobertson	IdahoEv: See documentation on subtree merge. jpr5's info is just importing the content and not the history (as I understand it)	23:56
	Of course if you don't care about history, then win	23:56
	← Buglouse left	23:57
	→ sh1mmer joined	23:58
	← MattDiPasquale left	23:58
kanzure	SethRobertson: the "forge a merge" part is git-commit-tree, i'd just have to set two parent ids	23:59
	← warbee left	23:59
	← fr0sty-away left	23:59
smolyn	SethRobertson: rohanpm: so it looks like git has special handling for numeric tags. we're appending "v" on the number for using regexs with git --describe	23:59
	that seems to break it	23:59
SethRobertson	kanzure: yes, but you have to do it back in history and then rebase everything after it on master on top of the forgery	23:59
IdahoEv	SethRobertson: thanks - in this case i don't really care about the history for this purpose. just trying to get around a seriously annoying limitation of github	23:59

Logs Search ←Prev date Next date→

Channels Documentation