IRCloggy #git 2009-05-23

Provider of IRC logs since 2005.

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

	2009-05-23
	← aslakhellesoy left	00:01
	→ agib joined	00:03
	→ memiux joined	00:05
	→ Dashkal joined	00:06
	→ MisterN_ joined	00:12
	← LiamH left	00:13
	← Yuuhi left	00:13
	jefftriplett → jefftriplett-afk	00:15
	← d0k left	00:17
	→ jtaby joined	00:18
jtaby	Hey, I have two branches. When trying to checkout into another, I get the following message:	00:18
	error: pathspec 'jtaby/master' did not match any file(s) known to git.	00:19
	→ qrush joined	00:20
	→ knitt1 joined	00:20
	← Shadow_Master left	00:22
	→ hoohah joined	00:22
sitaram	full command please...	00:22
	← jtaby left	00:23
hoohah	hello. I'm trying to wrap my head around tagging. Is it to be used when new features are implemented, and I would like to categorize all commits prior to a push to this one tag?	00:23
djgera	Hi, I can do a "reset --hard" to last commit that are pushed in the origin (github) ?	00:23
sitaram	djgera: git reset --hard origin/yourbranchname	00:24
djgera	i will try... one second...	00:24
sitaram	djgera: (assuming your current branch is "yourbranchname"	00:24
	← j416 left	00:24
djgera	mm i need to do a hard reset in my local branch first?	00:25
sitaram	djgera: are you trying to rollback the commit that already went to GH?	00:26
djgera	yes	00:26
sitaram	aaah, ok	00:26
	where is your local branch with respect to the origin?	00:27
djgera	but, without doing a revert, so for no logs keep in github	00:27
	← paulboone left	00:27
	→ Shadow_Master joined	00:27
djgera	my branch (master) is now at the same point that in origin	00:27
sitaram	djgera: the basic idea is you get your local branch to look like what you want, then "git push -f origin"	00:27
	hoohah: a tag is just a label or a sticky note on a commit	00:28
djgera	ok, then do a hard reset in mi master, then, put this command.	00:28
	i will try	00:28
sitaram	djgera: yes	00:28
	← MisterN left	00:29
	← fridim_ left	00:30
	← poseidon left	00:30
	← Shadow_Master left	00:31
	→ Shadow_Master joined	00:31
	← Shadow_Master left	00:31
djgera	awesome!!, many thanks sitaram ! i just do only a --soft on my local branch, then force push	00:31
	→ arrenlex joined	00:33
arrenlex	When you fetch stuff, where does it go?	00:33
hoohah	sitaram: okay thanks	00:36
	← hoohah left	00:36
	← knittl left	00:36
	→ plasticine_ joined	00:37
	← Dashkal left	00:38
sitaram	arrenlex: if you fetch from a named remote, it all goes to branches called remotes/name_of_remote/heads/whatever	00:41
	← plasticine left	00:42
sitaram	otherwise FETCH_HEAD, I think	00:42
arrenlex	sitaram: What is a 'named remote'? You mean one with a definition in my ~/.git/config ?	00:43
	sitaram: What exactly is FETCH_HEAD? Is it a branch?	00:43
zuez	heh what's the best way to revert one file to a previous commit?	00:43
	meaning, it's already committed, I just want to check it out one commit previous to HEAD	00:43
norc	arrenlex: It is my understanding that FETCH_HEAD is used as a buffer for inspection before you merge the requested branch	00:45
arrenlex	norc: For whose inspection?	00:45
	→ Flathead joined	00:45
norc	arrenlex: your own	00:45
arrenlex	norc: How do I inspect it without merging it?	00:45
norc	arrenlex: git fetch git://foo/bar.git blah; git log -o ..FETCH_HEAD	00:45
sitaram	arrenlex: git checkout -b inspect-me FETCH_HEAD	00:46
norc	arrenlex: and then just "git merge FETCH_HEAD"	00:46
sitaram	arrenlex: or gitk FETCH_HEAD	00:46
arrenlex	sitaram: Ah, I see. Thank you. If I don't specify a branch, it pulls master?	00:46
	← carrus85 left	00:47
norc	arrenlex: Oh, and I think I may have misexpressed myself. FETCH_HEAD itself contains rather information and is used to reference the branch you just pulled.	00:47
arrenlex	norc: Is it the only way to reference the branch I fetched?	00:47
sitaram	arrenlex: not sure what it fetches by default; master sounds right	00:48
norc	arrenlex: Well, I suppose so. There are probably other ways I do not know about.	00:49
arrenlex	norc: Okay, thank you for the advice.	00:49
	sitaram: Thank you too!	00:49
norc	arrenlex: Ugh, this is not my day. Of course it is a reference to what you fetched	00:50
	A pull basically fetches and then merges.	00:50
arrenlex	norc: Alright. And when you push to a remote repository, where does it go on that repository?	00:52
	Does it go to master?	00:52
	Probably not, you might need to merge...	00:52
norc	arrenlex: You can push to master or any branch	00:53
arrenlex	norc: What happens if there would have to be a merge?	00:54
	← sitaram left	00:55
	← knitt1 left	00:56
	← MisterN_ left	00:56
hachi	is there any way I can modify the behavior of an inbound ssh git connection, so I can have it find a users git directory automatically?	00:58
	and by automatically I mean, in a configurable way on the server side?	00:58
	IE	00:58
	is there any way to get /~hachi/ to expand to something other than $HOME ?	00:59
norc	arrenlex: I do not understand	00:59
	hachi: I think the first problem could be better dealt with in #ssh, and the second one in #bash probably	01:00
	although I think that #ssh would be the right place	01:00
arrenlex	norc: Suppose Fred and Bob made conflicting changes in their separate repositories. Fred pushes up to master. Then Bob tries to push up to master. What happens?	01:00
	norc: By pushes up to master, I mean origin.	01:00
hachi	I'm pretty sure the git client is expanding /~hachi/....	01:00
	it doesn't say the path is subject to shell expansion... just this one case that looks like shell expansion	01:01
	→ Murr joined	01:01
	← ia left	01:02
	→ ia joined	01:03
	← gb\|work left	01:04
	← naeu left	01:04
	← qhoxie left	01:04
	→ ben_h joined	01:05
	→ melvinram joined	01:05
	→ solydzajs joined	01:05
	← solydzajs left	01:07
	→ releod joined	01:07
norc	arrenlex: Uhm, good question.	01:10
	arrenlex: My best guess is that the second user has to rebase, fix the conflict and then push	01:11
	arrenlex: like you push, it fails. you pull or fetch and rebase, update your patch and push again	01:12
	← willb1 left	01:14
	→ jchris joined	01:14
	→ willb1 joined	01:15
davi	How to git push a commit which is not in the master branch?	01:16
madewokherd	git push nameofremote nameoflocalbranch:nameofremotebranch	01:18
	← markelikalderon left	01:18
	→ AAA_awright joined	01:19
	← marvil07 left	01:20
	← Arafangion left	01:20
davi	madewokherd, So, if the remote branch is not created I must create it first before execute that command? Is not the remote branch created automatically if it is the first time that something is committed to a branch with such name?	01:21
arrenlex	norc: Ah, you're right. em@sam:/tmp/bob$ git push origin master	01:22
	To /tmp/bob/../origin/	01:22
	! [rejected] master -> master (non-fast forward)	01:22
	error: failed to push some refs to '/tmp/bob/../origin/'	01:22
madewokherd	davi: I'm not sure; that might have changed recently	01:22
	← flaguy48 left	01:23
davi	I am going to try if the remote branch is automatically created, thanks MadCoder	01:23
	thanks madewokherd	01:23
madewokherd	it is or used to be that on the initial push to an empty repository you'd have to do master:refs/heads/master	01:23
davi	It is not empty repo, It has a master branch	01:23
	but I want to push to the stable-001 branch which I have	01:24
	created in my local copy, and	01:24
	committed just its first patch	01:24
	The stable-001 branch is not yet in the remote repo.	01:24
	I am going to try git push stable-001 to see if the new branch is created in the remote repo	01:25
	$ git push stable-001	01:25
	fatal: 'stable-001': unable to chdir or not a git archive	01:25
	:)	01:25
Arrowmaster	you need to add the remote	01:26
	git push origin stable-001	01:26
	dansan\|afk → dan_san	01:28
	cpg\|biab → cpg	01:28
davi	thanks Arrowmaster	01:29
	It is great with thing works :)	01:30
	← rettub_ left	01:30
	→ rettub_ joined	01:31
	← mp_ left	01:32
	→ sgrimm joined	01:36
	← mbroeker left	01:38
	← willb1 left	01:42
	← dt_101 left	01:42
	→ willb1 joined	01:43
	→ hpoydar joined	01:44
	→ summerbug joined	01:44
	← parasti left	01:45
	← arrenlex left	01:45
	← dreiss left	01:45
	jefftriplett-afk → jefftriplett	01:47
	← ben_h left	01:49
	← djgera left	01:51
	→ mike_m joined	01:52
	← aziz left	01:53
	← norc left	01:53
	← eletuchy left	01:56
	jefftriplett → jefftriplett-afk	01:56
	← romao left	02:00
	← kelnos\|laptop left	02:01
	← lois[pac] left	02:04
Chani	say I create several repositories. possibly large ones. imported from svn. if, someday, months or years down the road, I decide that one of the subdirs in one of the repositories really belongs in one of the other repositories, can I move it over with all its history?	02:04
	sebleier\|away → sebleier	02:05
offby1	sure	02:05
	not sure _how_ exactly :)	02:05
Chani	offby1: but you're sure it's prossible?	02:05
	git doesn't seem to like dealing with only part of a repo	02:07
offby1	oh oh	02:07
	I suspect you'd want "git-filter-branch"	02:07
Chani	oh?	02:07
offby1	I also suspect you'd import the _whole_ repository, then filter out the stuff you don't want.	02:07
	never done anything like that, so I dunno	02:07
	← sekimura left	02:09
Chani	"WARNING! The rewritten history will have different object names for all the objects and will not converge with the original branch. You will not be able to easily push and distribute the rewritten branch on top of the original branch. Please do not use this command if you do not know the full implications, and avoid using it anyway, if a simple single commit would suffice to fix your problem.	02:09
	"	02:09
	heh. I suspect I wouldn't want that :)	02:09
	maybe clone the repo, delet all the other stuff, mv the folder to be kept up to the root, delete the folders it was in, commit all that and then somehow import that into the other repo	02:10
	→ danielnewman joined	02:12
	→ hagna joined	02:16
hagna	how do I get a remote branch?	02:16
	→ eletuchy joined	02:18
	→ ChanSevr joined	02:20
	→ cytrinox joined	02:20
Chani	ohh, I see how filter-branch might be useful, now	02:21
offby1	hagna: by cloning or fetching	02:22
	by whining and kvetching	02:23
hagna	offby1: so not git checkout -b newbranch origin/newbranch ?	02:24
offby1	hagna: well, that assumes you've already got origin/newbranch	02:24
	← Murr left	02:24
offby1	what you just said is how you make a local branch that tracks the remote branch.	02:24
	ChanSevr → chansevr\|watchin	02:26
	← donri left	02:27
hagna	offby1: yes I stick with the first question	02:30
	hmmm git fetch origin/newbranch says unable to chdir or not a git archive	02:33
	← hpoydar left	02:33
offby1	you need to invoke "git fetch" properly.	02:34
	→ cYmen joined	02:34
offby1	what does "git remote" say?	02:34
hagna	mybox	02:35
	it's the remote alias	02:35
offby1	and what does "git remote show mybox" say?	02:35
	← cytrinox_ left	02:36
hagna	offby1: one of the branches that it says it's tracking is the one I want	02:37
	→ willb2 joined	02:38
offby1	so try "git fetch mybox"	02:39
	or, check out that branch, and just do "git fetch".	02:39
hagna	when I git checkout the branch I get files in there that aren't supposed to be on that branch	02:40
	← Radar left	02:40
	→ Radar joined	02:41
offby1	shrug	02:41
	→ qhoxie joined	02:41
offby1	the perhaps you need to remove them	02:41
hagna	fine could I start over and clone it then	02:42
	← kef left	02:42
hagna	how do you specify a branch with clone?	02:42
offby1	you don't -- you clone the whole repos	02:43
	cpg → cpg\|out\|w\|the\|ki	02:43
cYmen	could somebody give me a quick explanation how to share code	02:43
	cpg\|out\|w\|the\|ki → cpg\|biab	02:44
cYmen	what I want to do is a simple setup with two computers and the user on one having ssh access to the other	02:44
	is that clone or push or what do I want to use on the remote machine?	02:44
	I read the tutorial and can't figure this out	02:44
	sorry	02:44
	← orafu left	02:46
offby1	heh	02:46
	→ orafu joined	02:46
offby1	I'd put the repository on one machine, then clone it on the other	02:46
	are there two users, or just one?	02:46
cYmen	two	02:47
offby1	y'know it sounds complicated, but it might be easiest to have _three_ repository.	02:47
	repositories.	02:47
cYmen	yeah I wanted it that way	02:47
offby1	One "bare" one, and the others clones of it.	02:47
cYmen	one master and a local checkout for me and a checkout for him	02:47
offby1	The two of you communicate by pulling and pushing from that bare one. (It's OK to pull directly from each other, too)	02:47
	the thing is it's best to avoid _pushing_ to a non-bare repository.	02:48
	(It's not a disaster if you do, but it can get quite confusing)	02:48
cYmen	ok, maybe I should ask a different question first	02:49
	what's a clone/commit vs push/pull?	02:49
offby1	well, those are four different terms.	02:49
	"clone" means to make a new repository that is more or less a copy of an existing one. You don't do that terribly often.	02:50
	← kirillm left	02:50
offby1	"commit" is how you make a change to a repository. You do it quite frequently: every time you've done some work that's worth saving.	02:50
	← elliottcable left	02:50
offby1	"push" means to send stuff from one repository to another -- typically to update the remote repo's version of a branch that both repositories have.	02:50
	"pull" means to get stuff from the remote repository, and to _merge_ it into the current branch.	02:51
	("pull" is not the exact opposite of "push", but "fetch" is)	02:51
	→ tjafk1 joined	02:51
	← radostyle left	02:53
	← willb1 left	02:54
	← chocolate_jesus left	02:54
	→ ben_h joined	02:54
cYmen	hm...I think it works maybe :)	02:56
	I'll get back to you with further problems ;)	02:56
	thanks a lot!	02:56
	→ dt_101 joined	02:56
	→ JNRowe joined	02:59
	← Radar left	02:59
	→ AkumaStreak joined	03:01
	→ Suraia joined	03:05
	→ qhoxie_ joined	03:06
	← jed left	03:07
	→ radostyle joined	03:08
	← tjafk left	03:08
	→ elliottcable joined	03:08
	→ hyperair joined	03:11
	→ altrux_ joined	03:11
	← elliottcable left	03:12
	← wsc left	03:13
	← dt_101 left	03:15
	chansevr\|watchin → ChanSevr	03:15
	← jchris left	03:17
	→ wsc joined	03:18
	← jfkw left	03:18
	← AkumaStreak_ left	03:19
	← ChanSevr left	03:19
	← hagna left	03:19
	→ bharat joined	03:20
	→ elliottcable joined	03:21
bharat	does anybody here have experience setting up vendor branches for git? I'm attempting to migrate a mature project with vendor branches from svn to git and can't find an approach that will satisfy our needs	03:22
	← altrux left	03:25
	← ToxicFrog`` left	03:25
	← qhoxie left	03:26
offby1	I don't have any experience, but I assume you'd do it analogously to how svn does it	03:27
	← beantmt left	03:27
offby1	you make a branch that tracks the official releases, and every now and then you get a new tarball and plop it on top of the old one, and commit that ...	03:27
bharat	that works for keeping the vendor branch up to date	03:28
Chani	I don't suppose there's a way for an svn external to actually point to a git repo? :/	03:28
bharat	but the problem comes when I try to merge the vendor code into our main repo	03:28
	→ dreiss joined	03:28
offby1	and you have another branch that includes your patches, and you merge from the official releases into it	03:28
	bharat: what's the problem?	03:28
madewokherd	merging the vendor code into the main branch sounds like a bad idea	03:28
	Chani just discovered that all kde's documentation is kept as svn externals for easy access by translators...	03:28
bharat	ok hang on let me provide you with a full use case	03:28
	we have a repo (gallery3) that contains a subdir (gallery3/kohana) that comes from an upstream vendor	03:29
	http://github.com/bharat/gallery3/tree/bd88bba94dbeae136b570f0ece60c08c04f514b6/kohana	03:29
	our vendor branch for it is here: http://github.com/bharat/gallery3-vendor/tree/2ba65ca8bdb2bb9f45089d4aefda9c0cb6c69554/kohana	03:29
	we regularly get snapshots from the kohana folks and put them into the "upstream" dir	03:30
	but that's not exactly the way the code needs to be formed to put it into our project, so we massage it into a new form in the "modified" dir	03:30
	so that's a commit like this: http://github.com/bharat/gallery3-vendor/commit/2ba65ca8bdb2bb9f45089d4aefda9c0cb6c69554	03:30
	now with svn, I can just merge that commit from gallery3/vendor/kohana/modified into gallery3/trunk/kohana	03:31
	but git doesn't appear to allow me to merge part of one repo into part of another repo	03:31
	it wants me to do all these merges at the whole-repo level	03:31
offby1	er, yeah	03:31
	what's wrong with that? Are there conflicts?	03:32
bharat	oh absolutely	03:32
offby1	ah	03:32
bharat	we have about 10 different vendors	03:32
	many of them require some amount of reformatting before they fit into our hierarchy properly	03:32
	← memiux left	03:32
bharat	our vendor hierarchy: http://github.com/bharat/gallery3-vendor/tree/master	03:32
	with svn, we can turn vendor tarballs into diffs, and apply the diffs to the right point in our code and svn tracks it all	03:33
	I would have thought that git could do this easily	03:33
	and plus, we make our own local modifications to the code which we then file upstream tickets for and resolve when we pull the latest code from the upstream vendor after they fix the ticket	03:34
Arrowmaster	theres a subtree merge strategy	03:35
	bharat plays with git-format-patch	03:35
Arrowmaster	"This is a modified recursive strategy. When merging trees A and B, if B corresponds to a subtree of A, B is first adjusted to match the tree structure of A, instead of reading the trees at the same level. This adjustment is also done to the common ancestor tree. "	03:35
bharat	I've been trying to use that with "git pull -s subtree"	03:35
Arrowmaster	no clue if that helps	03:35
	→ rkirti joined	03:35
bharat	but it tries to merge the entire vendor hierarchy	03:35
madewokherd	you might want to look into submodules..	03:36
bharat	I have..	03:36
	the problem with submodules is that we can't then make our own local modifications to the code	03:36
	we can only pull from the upstream version	03:36
madewokherd	huh?	03:36
bharat	theoretically we could make those changes entirely upstream in our vendor code	03:36
	I'll give you an example	03:37
	we get file foo.php from vendor/kohana/upstream and then we move it into vendor/kohana/modified/bar.php and we apply it to gallery3/trunk/kohana/bar.php	03:37
madewokherd	what do the /'s mean here, are these paths?	03:38
bharat	but there's a bug, so we fix it in gallery3/trunk/kohana/bar.php and we file a ticket. they fix the ticket	03:38
	those are paths based on our svn hierarchy	03:38
	we carry the fix in our trunk because we need to use it. but eventually it shows up in the upstream code, which then gets applied and merged over and reconciled	03:39
almostautomated	LMAO, totally off-topic:: License Limitation for Coom Free License => 2. Is not allowed to use the CFL products to any life destruction purpose. This means that CFL products can not be used as products or as parts of other products for any activity which has as effect the destruction of life in any form of that life.	03:39
	← JNRowe left	03:40
	→ JNRowe joined	03:40
	← orafu left	03:41
madewokherd	bharat: so you keep another kohana repository with your changes, what's the problem?	03:41
	← smashwilson left	03:41
	→ orafu joined	03:42
bharat	I don't know how to explain it any better than I just did	03:42
madewokherd	do it by not referring to svn	03:42
	I have no idea how svn works	03:42
bharat	we need to merge part of one repository into part of another repository	03:42
madewokherd	so you need one repository that has both of those as remotes	03:43
bharat	ugh	03:44
	that's going to suck for 10+ different vendor libraries	03:44
	oh well. don't worry about it. from a lot of searhcing, it appears that nobody has solved this for git yet	03:44
	I was hoping against hope that somebody here had direct experience with it	03:44
madewokherd	yeah, we pretty much only have 2 of those libraries at work and I only deal with one of them	03:45
almostautomated	bharat: have you looked at submodules and super project setups ?	03:45
madewokherd	that makes it an objectively different case that I have no direct experience with	03:45
	← AAA_awright left	03:46
bharat	almostautomated: yes, I've been playing with submodules but I don't think they're going to work	03:46
almostautomated	ah, well; worth a shot :)	03:47
bharat	:-)	03:47
	http://stackoverflow.com/questions/769786/vendor-branches-in-git is a pretty good description of the problem	03:47
	← JNRowe left	03:49
	← tchan left	03:49
	→ tchan joined	03:50
	→ Arafangion joined	03:53
almostautomated	bharat: you said:: 'the problem with submodules is that we can't then make our own local modifications to the code'	03:54
	← up_the_irons left	03:54
	→ AAA_awright joined	03:55
bharat	yes. the docs say "submodules are meant for different projects you would like to make part of your source tree, while the history of the two projects still stays completely independent and you cannot modify the contents of the submodule from within the main project."	03:55
	http://www.kernel.org/pub/software/scm/git/docs/git-submodule.html	03:55
	→ girishr joined	03:55
offby1	_from within_	03:55
bharat	am I misunderstanding how that works?	03:55
offby1	try 'em!	03:56
	cpg\|biab → cpg	03:56
offby1	it just means you have to "cd" into the subproject, and commit ther	03:56
	there	03:56
bharat	can submodules operate on part of a repo?	03:57
	ie I have a gallery3 repo that has paths "gallery3/kohana" and I have a gallery3-vendor repo that has paths "kohana/modified" (as well as many other sets of vendor code)	03:57
	I'll experiment with that to see	03:57
Arafangion	bharat: It means that, eg, you could have your project, and then have a super-project that includes all of the dependencies.	03:58
	bharat: For example, freetype.	03:58
almostautomated	I'd imaging two repo's would be the ideal solution; a superproject of pristine sources, and one of custom sources and your project	03:58
cYmen	how do I fix a "your branch is ahead by 1 commit"? :/	03:58
almostautomated	s/imaging/inagine	03:58
	err imagine	03:59
bharat	I'll give that a shot. thanks for the advice	03:59
	← Sho_ left	04:01
	← danielnewman left	04:02
	← Alinermedia left	04:02
	← quoin left	04:03
	→ memiux joined	04:05
	→ Sho_ joined	04:05
	← squentin_ left	04:05
	→ bobmcw_ joined	04:10
	← bobmcw left	04:10
	→ up_the_irons joined	04:11
	← bharat left	04:11
	← Suraia left	04:17
	→ summerbug_ joined	04:17
	sebleier → sebleier\|away	04:19
	→ scarabx joined	04:20
	← drewr left	04:22
	← Sho_ left	04:22
	← melvinram left	04:28
	← eletuchy left	04:34
	← summerbug left	04:34
	→ eletuchy joined	04:35
	→ toxx__ joined	04:43
	← dan_san left	04:46
	→ sdiz joined	04:46
	fr → fr\|	04:49
	→ tvw joined	04:51
	← stuffcorpse left	04:52
	→ stuffcorpse joined	04:53
	← toxx_ left	04:59
	← curvature left	05:00
	→ curvature joined	05:00
	→ eno__ joined	05:08
	← rkirti left	05:08
	→ gruff joined	05:11
gruff	What git gui could you recommend. I want UI brilliance no gnarly gnuified cryptical shit	05:12
	...like gitk	05:12
	sdiz → sdiz-afk	05:14
	→ [AndrewR] joined	05:15
[AndrewR]	hey, I'm trying to revert a commit that added some files, but they've since been modified.. but I want to just get rid of them	05:16
	but they conflicted, so git is expecting me to git add them	05:16
	how can I totally wipe them?	05:16
	← k0001 left	05:18
	← eno left	05:19
	eno__ → eno	05:19
henr_k	[AndrewR]: unless you've pushed the commits elsewhere, you can do a git filter-branch	05:20
[AndrewR]	yeah, they're all pushed already, unfortunately	05:20
	basically this guy made two commits, the second one modified files in the first (and the first added a bunch of files)	05:21
	and I want to wipe both commits so there's no trace	05:21
	but there's been a bunch of dev since then	05:21
	so any file that conflicts I want to just tell git to delete the file, but to resolve a conflict you have to git add?	05:22
henr_k	well, when you do a git filter-branch, you're going to change all the id's after that first commit, so you're going to break a lot of repos	05:22
	← sergiodj left	05:22
	→ sergiodj joined	05:22
[AndrewR]	I'm fine with making a new commit that reverses those two	05:22
henr_k	git rm filename ?	05:23
	you will still have the content in the repo - is that OK?	05:23
	→ davidad joined	05:23
	→ kef joined	05:25
	← wsc left	05:27
	→ reprore joined	05:27
	← memiux left	05:30
	→ sitaram joined	05:40
	joshthecoder → joshthecoder_afk	05:40
	← plasticine_ left	05:40
	← altrux_ left	05:40
	→ altrux joined	05:41
	← scientes left	05:41
	← Bass10 left	05:42
[AndrewR]	turns out once they're added I can git rm them	05:46
	→ Swimming_Bird joined	05:47
	→ sea-gull joined	05:48
	← reprore left	05:48
	→ reprore joined	05:48
	→ fujin joined	05:49
	→ reprore__ joined	05:50
	← sergiodj left	05:54
	→ sako joined	05:57
sako	hey guys i wanna start using git to track my development.. is this a good setup: http://joemaller.com/2008/11/25/a-web-focused-git-workflow/	05:57
	gonna be for web development	05:58
	wondering if this is the best way and what advantages this method has	05:59
	→ repror___ joined	06:01
	← radostyle left	06:01
	← reprore left	06:01
sako	anyone?	06:02
	← summerbug_ left	06:03
	→ DavidKlein joined	06:04
Arrowmaster	sako: it doesnt look perfect but i dont see anything blatently wrong	06:05
sako	Arrowmaster: is there a better solution for using git for websites?	06:07
	right now i just have a git repo of the live folder	06:07
	and i work like that commiting changes..	06:07
	whats the advantage of the method above?	06:07
Arrowmaster	i know of some more advanced solutions	06:07
	→ summerbug joined	06:08
Arrowmaster	i dont know about better, i dont do web development	06:08
runasand	sako: the advantages of a hub and a prime is that you get to test stuff better before going live with it	06:08
	sako: it's better than just pushing right to the live site, I guess. Depends on the types of changes you do.	06:08
Arrowmaster	in that setup anything pushed to 'hub' automatically goes live on 'prime'	06:08
runasand	ah ok, I got it wrong then :o	06:09
	shit happens when you read techincal stuff at 8am ;p	06:09
sako	hmm actually i want a setup like that runasand	06:10
	i wanna be able to test before i go live.. so i would just skip the hook part right?	06:10
	in that example	06:10
Arrowmaster	what kinda web dev are you doing?	06:10
sako	its an e-commerce site	06:10
	php/mysql	06:10
	→ porcnick joined	06:11
	← scarabx left	06:11
	→ g3d joined	06:15
runasand	Arrowmaster: what's the difference between pushing directly to prime and going through hub?	06:16
Arrowmaster	sako: ive never used and have zero experience with it but i know theres some stuff like that but much more advanced, one is Capistrano for example and it could be overkill for you since its got stuff for dependancies and such (tends to be used by rails people i think)	06:16
	you could probably create two vhosts in your webserver and have a live and testing deployment	06:16
sako	yea i was thinking that..	06:16
runasand	sako: but to answer your question, yeah, just skipping the hook part would do it. I think :)	06:17
sako	so Arrowmaster do you know what the advantage of having that hub is??	06:17
	doesnt make sense to me why its necessary	06:17
Arrowmaster	runasand: well pushing directly to prime wouldnt update the working directory, usually the solution is to use a post-update hook to update the working directory when somebody pushes instead of having a post-update hook just run a git pull in another repo	06:17
	sako: its not necessary, most people would probably use a post-update hook to update the working directory than jump through hoops with two repos and have a hook in one repo run git pull in another repo	06:18
runasand	ah :)	06:18
	← reprore__ left	06:19
Arrowmaster	sako: do you plan to actually do any editing on the server and commiting to the repo or will everything come in via push?	06:20
sako	well i clone the git repo on my laptop	06:21
	and i do the editing then do git commit -a	06:21
Arrowmaster	a simpler way would be to skip the adding of a remote to the prime and just make a post-update hook that runs `git reset --hard master`	06:22
cYmen	how do I debug the remote host hanging unexpectedly?	06:22
	could it be because i'm trying to clone an empty rep?	06:22
Arrowmaster	anytime you push to the repo it will force the working dir to be updated to master	06:23
	cYmen: what version of git? client and server	06:23
	→ seangrove joined	06:23
cYmen	Arrowmaster: i didn't install an explicit server	06:23
Arrowmaster	if you dont know the server its fine	06:23
cYmen	the machine i'm cloning from is 1.6.0.2	06:24
	← tvw left	06:24
Arrowmaster	i mean from where you are trying to pull from	06:24
	ok	06:24
cYmen	the machine cloning is 1.6.3	06:24
Arrowmaster	empty repo shouldnt effect it	06:24
cYmen	uh nice...there's an xkcd comic about your use of effect ;)	06:25
sako	sorry im trying to understand this.. what does git reset --hard master do?	06:25
cYmen	anyway are there logfiles or something I could check?	06:25
sako	so i did git init , git add . , git commit -a on the server	06:25
	then i went to my laptop and did git clone .git	06:26
Arrowmaster	unless its managed by something like gitosis, i know a site that you cant clone empty repos from because the repos dont actually exist until they are pushed to	06:26
sako	and i make my changes on the lapto p and i do git commit -a when done..	06:26
	is that right>	06:26
Arrowmaster	sako: `git reset --hard master` just makes it update the index and working tree to master and forget anything else	06:26
	sako: yeah	06:27
sitaram	http://xkcd.com/326/	06:27
	← altrux left	06:33
	← eletuchy left	06:34
	→ eletuchy joined	06:34
	← willb2 left	06:38
	→ willb2 joined	06:38
Nugget	Grammar Police: to serve and correct	06:43
	← tango_ left	06:43
gruff	is there a way to do a clean checkout?	06:44
	overriding any changes ive got in the currenct WC?	06:45
	→ angerman joined	06:45
Chani	gruff: git checkout doesn't do that?	06:46
gruff	the -f option seems to have done it for me	06:46
	→ joaovitor joined	06:48
gruff	god... git seems unstableish for me	06:50
	maybe im doing something wrong	06:50
g3d	gruff: git-clean might also help for untracked files	06:50
	→ tango_ joined	06:50
	← repror___ left	06:52
	← eletuchy left	06:53
gruff	fatal: pathspec 'src/main/java/com/<path....>/variations' did not match any files	06:53
	what does that mean?	06:53
	→ reprore_ joined	06:53
	← doener left	06:53
g3d	gruff: that there is no such file	06:53
	→ doener joined	06:53
gruff	its on the disk anyway	06:53
g3d	so 'ls src/...(etc)' does return it?	06:54
	I mean find?	06:54
gruff	yeap	06:54
g3d	and you are in the right directory, the one containing the 'src' dir?	06:55
	← reprore_ left	06:55
gruff	i get that error when trying to do a "git rm" on the "variations" dir	06:55
	→ reprore_ joined	06:55
g3d	well, variations is a dir not a file	06:56
	so you should use the -r option	06:56
	as the manual says	06:56
	→ struberg joined	06:57
	→ avaja joined	06:58
gruff	oh.. seems it was a file locking problem	06:58
	← bobesponja left	07:00
	→ sergiodj joined	07:01
	← reprore_ left	07:08
	→ \|kukks\| joined	07:11
	← porcnick left	07:14
	→ wsc joined	07:14
	← joaovitor left	07:16
	krawek__ → krawek	07:19
gruff	how does one revert the WC?	07:22
	← summerbug left	07:23
	← neoeinstein_ left	07:23
	← cloudhead left	07:25
	← Swimming_Bird left	07:27
	← [AndrewR] left	07:27
	← kukks left	07:27
	→ brizly joined	07:28
gruff	ahh reset	07:30
	← brizly left	07:30
	→ ben_h_ joined	07:30
	→ brizly joined	07:31
	← Modius_ left	07:37
	← agile left	07:37
	← krawek left	07:37
	← jayallen left	07:37
	← ph^ left	07:37
	← PuffTheMagic left	07:37
	← reaVer left	07:37
	← CarlFK left	07:37
	← piroko left	07:37
	← Nugget left	07:37
	← tro left	07:37
	← nomadium left	07:37
	← Garoth left	07:37
	← Rhonda left	07:37
	← kelvie left	07:37
	← ph8 left	07:37
	← Gitbot left	07:37
	← orospakr left	07:37
	→ krawek_ joined	07:38
	→ Modius_ joined	07:38
	→ agile joined	07:38
	→ krawek joined	07:38
	→ jayallen joined	07:38
	→ ph^ joined	07:38
	→ PuffTheMagic joined	07:38
	→ reaVer joined	07:38
	→ CarlFK joined	07:38
	→ piroko joined	07:38
	→ Nugget joined	07:38
	→ ph8 joined	07:38
	→ nomadium joined	07:38
	→ kelvie joined	07:38
	→ Garoth joined	07:38
	→ orospakr joined	07:38
	→ tro joined	07:38
	→ Rhonda joined	07:38
	→ Gitbot joined	07:38
	← krawek left	07:38
	← ben_h_ left	07:41
	→ jony joined	07:42
	→ wgaffa joined	07:43
	→ kate21de joined	07:45
	← sergiodj left	07:45
	→ kumbayo joined	07:45
wgaffa	I have a submodule directory (restful_authentication) and when I do a git pull from another git repository and later do a git submodule update it reverts back to the "old" submodule? How can I tell submodule to not revert back to the old repository?	07:46
	← avaja left	07:46
	← ben_h left	07:47
	→ Radar joined	07:49
	→ jchris joined	07:58
	← brizly left	07:58
	← wsc left	07:59
	← glommer left	08:01
	→ drizzd joined	08:02
	← hyperair left	08:08
	→ bnjmn--_ joined	08:08
	← bnjmn-- left	08:08
	← struberg left	08:12
	← robinr left	08:13
	→ khmarbaise joined	08:14
	← AkumaStreak left	08:15
	← psankar left	08:18
	← jony left	08:19
	→ Tuomas joined	08:20
	→ priidu joined	08:23
	← mtvee left	08:24
	→ xyz joined	08:24
	→ Weaselweb joined	08:26
	→ thiago_home joined	08:27
	← davidad left	08:28
Ademan	so I ran git fetch /other/repos/thats/not/origin and it seemed to work, how can I now use that information in git merge ?	08:29
	→ abstrusething joined	08:31
	→ davidad joined	08:33
wereHamster	Ademan: git merge FETCH_HEAD	08:33
Ademan	wereHamster: really? huh, I was trying something similar, but having no luck, thanks, I'll give that a shot	08:33
	I definitely need to have a better understanding of what's going on, distinctions between tags and refs and what a remote is , a head etc	08:34
sea-gull	have anybody seen call graph for git?	08:35
Ademan	wereHamster: hrm, same issue as before: fatal: 'FETCH_HEAD' does not point to a commit does that mean my fetch failed silently?	08:35
sr	Ademan, http://book.git-scm.com/ :)	08:36
wereHamster	Ademan: did fetch print any output?	08:36
Ademan	wereHamster: none	08:36
wereHamster	usually it prints which heads it fetched (and where to)	08:36
Ademan	hrm, alright, if it helps the repos was a local directory	08:37
	thanks sr	08:37
	oh wow, I was thinking i'd end up 20 USD poorer, but no, awesome!	08:37
	→ kahmalo joined	08:39
kahmalo	What is a good way to replace a conflict resolution in .git/rr-cache?	08:40
	If I run git merge, and it finds and uses an old resolution, but I then notice the resolution is wrong and change the file and do git add and git commit, git doesn't seem to save the new resolution.	08:41
omega	Ademan: http://tom.preston-werner.com/2009/05/19/the-git-parable.html provides a good read to understand some of the concepts	08:41
	→ wsc joined	08:41
Ademan	thanks as well omega	08:41
kahmalo	So then I grepped for some keywords in rr-cache and removed the appropriate subdirectory. But is there some way that doesn't require grepping?	08:42
	→ fanf joined	08:43
kahmalo	There is git rerere clear, but it looks a bit too effective.	08:44
	→ jceb joined	08:46
	← Octalot left	08:47
kahmalo	The name of the subdirectory in rr-cache appears to be a hash of only the conflicts, not the rest of the file. I don't see any easy way to ask git what the hash is for a given conflicted file, so that I could remove the subdirectory.	08:50
	hm, in fact cmd_rerere() for "git rerere clear" does "if (!has_rerere_resolution(name)) unlink_rr_item(name);", so I think it wouldn't even clear the resolutions that have already been saved.	08:51
	→ the_zeno joined	08:54
	→ aslakhellesoy joined	09:01
	← sitaram left	09:06
	← aslakhellesoy left	09:09
	← cpg left	09:14
	→ cpg joined	09:14
	→ radarek joined	09:18
	→ hyperair joined	09:20
	→ aslakhellesoy joined	09:20
	→ reprore joined	09:21
	← cpg left	09:24
	← priidu left	09:27
	← eb_ left	09:28
	→ eb_ joined	09:28
	← fanf left	09:31
	← kuadrosx left	09:34
	← krawek_ left	09:38
	→ parasti joined	09:39
	→ knittl joined	09:44
	→ legnaleurc joined	09:45
	→ d0k joined	09:50
	→ sitaram joined	09:51
	→ CardinalNumber joined	09:51
	→ hoohah joined	09:54
hoohah	hello. is it possible to keep .git/hooks/post-commit hook itself under version control?	09:54
	I get fatal: This operation must be run in a work tree	09:54
Ilari	hoohah: Nothing under .git can be added.	09:54
	hoohah: Put a copy into working tree.	09:55
hoohah	Ilari: do you know a workaround for that?	09:55
	hm okay	09:55
	perhaps i should just cat > to the file in my shells login script?	09:56
Aides	symbolic link?	09:56
Ilari	hoohah: Hooks could be dangerous if clone transferred them.	09:56
	Aides: If you put symbolic link to it, it gets stored as symbolic link.	09:56
hoohah	Ilari: the hook itself only does a crontab -l > <fileundercontrol>	09:56
Aides	Ilari, no, i mean put a symbolic link in .git/hooks/, the hook itself in the tree	09:56
hoohah	something i had to address when crontab file was outside wrking tree	09:56
Ilari	hoohah: Still it could do 'rm -rf /' (or something even more nasty).	09:57
	Aides: That might work. But its quite unsafe...	09:57
hoohah	clones will have ,git/hooks/post-commit anyway, right?	09:58
	or will only the samples be supplied?	09:58
Ilari	hoohah: Clones have hooks initialized from local templates, IIRC.	09:59
hoohah	Ilari: local as in root dirs? or user dir?	10:00
Ilari	hoohah: The templates are part of git installation.	10:01
	hoohah: So most of the time they come from somewhere under '/usr', but if you use git installed to $HOME, then those templates are stored somewhere under $HOME.	10:03
hoohah	true	10:04
	Aides: your symlink suggestion.. won't be a deployable option, right?	10:05
	as it's still under .git	10:05
Aides	yeah	10:05
	i thought you needed a version control for it, not to be cloneable	10:05
Ilari	hoohah: Anyway, hooks being cloned would be extremely unsafe.	10:06
	← ProperNoun left	10:06
	← abstrusething left	10:09
	← reprore left	10:13
Ademan	is it possible to make git track empty directories?	10:15
hoohah	thanks	10:15
	← hoohah left	10:15
kahmalo	Ademan: IIRC the standard workaround is to put an empty .gitignore there	10:15
	→ ProperNoun joined	10:16
Ademan	kahmalo: thanks, is that just a git shortcomming?	10:16
	→ Murr joined	10:16
kahmalo	http://git.or.cz/gitwiki/GitFaq#head-1fbd4a018d45259c197b169e87dafce2a3c6b5f9	10:16
	← Murr left	10:16
kahmalo	tracking of truly empty directories been discussed, but it was not clear what should then happen in some merge or cherry-pick or revert situations	10:17
Ademan	kahmalo: ah, so it's more of a design decision than anything else?	10:21
Ilari	The way I would do it (unfortunately this way has backwards compatiblity problems), is to set sticky bit in directory to mean "manually tracked". Such directories are not autoremoved when becoming empty and don't need any files under them to be trackable. Then handle the +t bit on dirs like +x bit on files.	10:21
	→ memiux joined	10:21
kahmalo	Ilari: do you mean the sticky bit in the tree objects in Git, not in the directories in the working tree?	10:22
Ilari	kahmalo: Tree objects.	10:22
	→ xorg62 joined	10:23
xorg62	hi :)	10:23
Ilari	The index format would also need to be modified so it could store directory entries.	10:23
	The problem one runs with current versions is that git tree-to-index conversion just plain ignores mode bits in type 4 (Directory) entries.	10:24
	← g3d left	10:24
	→ cemerick joined	10:25
	← CardinalNumber left	10:25
Ilari	And as result, the written-back trees would convert it to ordinary (auto-tracked) directory entry (or just strip it off entierely if its empty).	10:25
kahmalo	sounds like you'd need a multi-year transition plan	10:26
	→ Fullmoon_ joined	10:26
	← Modius_ left	10:28
	→ Modius joined	10:28
Ilari	Submodules could be added in one X.Y.Z release because unknown entry types are errors (submodules are type 14, directories type 4, regular files type 8 and symlinks type 10).	10:29
	Maybe define type 20 (outside the normal valid type range which goes up to 15) for such entries. Should definitely cause tree reading to blow up...	10:32
	→ plasticine joined	10:32
Ilari	Fun: I told --cacheinfo to add mode 240000 entry. It added mode 160000 entry...	10:35
	→ chris2 joined	10:36
thiago_home	there are more bits left	10:36
	→ Fullmoon__ joined	10:36
thiago_home	besides, this is not a set of flags	10:36
	← Ecco left	10:38
	← khmarbaise left	10:39
sea-gull	Ilari: where can I read more about index format? I can't understand why gitlink is needed?	10:39
	← memiux left	10:40
Ilari	Except that mode 240000 (type 20) entry doesn't make it blow up enough.	10:40
	sea-gull: gitlink is another name for submodule.	10:40
	→ PuffTheMagic_ joined	10:40
	→ CardinalNumber joined	10:41
	← ProperNoun left	10:41
sea-gull	Ilari: Linus says: it's a directory link.	10:42
	Ilari: but I don't understand why it's needed	10:42
Ilari	sea-gull: Entries of that type store the revision submodule is at.	10:42
	Trying to add mode 150000 entries results mode 100644...	10:43
	← Fullmoon left	10:43
	→ reprore joined	10:44
	← PuffTheMagic left	10:45
drizzd	gitlinks are mode 160000	10:46
	→ g3d joined	10:48
drizzd	which is not a filetype currently in use AFAICS	10:48
Ilari	Eww... It seems that invalid entry types don't currently cause git to blow up. I tried with mode 13, which is definitely invalid.	10:49
drizzd	what does the file system say about that?	10:50
	← Fullmoon_ left	10:51
toxx__	kahmalo: the rr-cache problem with old wrong merges is also of interest to me. did you find any solution?	10:52
	← hyperair left	10:52
	← gruff left	10:52
Ilari	drizzd: Apparently checking it out, git removes it from working tree. Index considers it to be regular file.	10:53
sea-gull	Ilari: what does comment "more" mean?	10:54
	← wgaffa left	10:54
drizzd	odd. I wonder if one could cause something nasty by tampering with a repo like that	10:54
Ilari	drizzd: Tree entry: "150000 blob 543a4f23aea1b9c603e40954b2b07d080f648440 foog", index entry: "100644 543a4f23aea1b9c603e40954b2b07d080f648440 0 foog"	10:54
	sea-gull: "more"?	10:54
	→ ProperNoun joined	10:55
sea-gull	Ilari: it's after -> char name[FLEX_ARRAY] /* more */	10:55
Ilari	sea-gull: FLEX_ARRAY is presumably 1. That comment means that more space can be allocated for the name than FLEX_ARRAY.	10:56
	Fun message from 'git fsck --full': "broken link from blob unknown".	10:57
	And it doesn't complain about mode 240000.	10:57
kahmalo	toxx__: so far, grepping seems the easiest way.	10:59
toxx__	kahmalo: mkay... thanks :)	10:59
	← legnaleurc left	11:00
Ilari	The reason that message is odd is because blobs can't be sources of references.	11:00
drizzd	Ilari: what you're seeing probably happens because of create_ce_mode. If it's not a symlink, a directory or a gitlink, it assumes it's a regular file	11:02
	← sitaram left	11:04
	→ aziz joined	11:04
	→ DrFrasierCrane joined	11:11
	→ jony joined	11:12
	← CardinalNumber left	11:16
doener	Ilari: IIRC FLEX_ARRAY should be (empty) for C99, 0 as a GNU extension and 1 otherwise	11:19
	→ fridim_ joined	11:19
	→ MisterN joined	11:23
	← ProperNoun left	11:24
	→ felipec joined	11:25
	→ mbroeker joined	11:27
	→ JEEB joined	11:35
	← aslakhellesoy left	11:37
	← girishr left	11:38
	→ gitte joined	11:42
	← ashH left	11:46
	← mbroeker left	11:49
	→ mithro joined	11:51
	→ Yuuhi joined	11:55
	← cemerick left	11:56
	→ elpargo_ joined	11:58
	← Flathead left	12:02
	← elpargo left	12:04
	→ glommer joined	12:09
	→ cemerick joined	12:13
	← kahmalo left	12:15
	← reprore left	12:16
	→ farbing joined	12:16
	← MisterN left	12:17
	→ flaguy48 joined	12:17
	→ MisterN joined	12:18
	→ memiux joined	12:24
	→ j416 joined	12:25
	← Radar left	12:25
	← cemerick left	12:27
	→ mbroeker joined	12:28
	→ igli joined	12:30
igli	what would be the use-case for a tracking branch which didn't autorebase?	12:30
	s/the/a	12:31
	hmm guess a user can always set that up themselves if they want it	12:31
	→ bremner_ joined	12:32
	← moonese1 left	12:33
	→ patrikf joined	12:37
	← farbing left	12:38
Ilari	igli: Firstly tracking is from time before git pull got its rebase mode IIRC. And in case the branch is published, rebasing it is really inapporiate.	12:39
igli	ah ok. yeah i get you on the published bit, thanks Ilari	12:40
	← memiux left	12:40
igli	hmm makes things a bit more awkward if we can't rebase from upstream	12:41
Ilari	igli: If you have two remotes, all the more reason to be much more careful about using rebase.	12:42
igli	hehe fair enough :) I guess we'll soon see what happens (will test in closed env first)	12:42
	→ drewr joined	12:42
Weiss	two remotes + git-svn + rebase makes my head hurt (but it great for making yourself REALLY understand how git works..)	12:43
	→ ereslibre joined	12:43
igli	idk if i could cope with git-svn on top of all the other stuff ;)	12:44
Ilari	igli: If you use rebase in any sort of nonconventional ways, one needs to investigate "not fast forward" errors VERY carefully.	12:44
igli	ok will bear that in mind, thanks Ilari	12:44
Ilari	igli: Useful way is to first fetch (not pull!) the remote and then do 'gitk foo...bar/foo' (note the three dots).	12:45
	← mike_m left	12:45
igli	hmm yeah i know to use fetch. i tend to use gitk --all and let it give me a diff	12:46
	for automation, tho, that sounds useful for git-diff	12:47
Ilari	igli: If you want to try to automatically detect errors, its more like log-type than diff-type operation.	12:48
igli	hmm i was just thinking to present to user first, but that'd be more useful, Ilari	12:49
Ilari	igli: --left-right to rev-list is useful for extracting the commits involved.	12:50
	igli takes notes	12:50
Ilari	igli: For auto-detection, grab all commits user has but remote doesn't and try to match them against commits remote has but user don't. There shouldn't be matches.	12:52
igli	ah that's lovely, Ilari :-)	12:53
Ilari	igli: commit messages matching is quite bad. Patch ids matching is even worse.	12:54
igli	hehe well it can't be worse than some of the situations i've had to awk my way out of ;) i should get some time to play with this properly later in the week	12:54
	→ rittyan joined	12:55
	← jony left	12:57
	→ bx2 joined	12:58
	← aziz left	12:58
Ilari	igli: Patch IDs matching would mean user is trying to push change that already exists in upstream, but with new commit ID.	12:58
igli	hmm yeah, idea is that they'd rebase from upstream themselves before pushing, Ilari. not sure how it'll work out in practise (was thinking might need to ensure upstream and origin/upstream are at same point)	13:00
	→ memiux joined	13:00
Ilari	igli: One problematic thing is that you can't alias push, it has no local-side hooks, and AFAIK non-fast-forwards are rejected on client side.	13:01
	← __iron left	13:01
igli	hah	13:01
	→ __iron joined	13:02
igli	hmm well i'm happy to use shell scriptlets and aliases where appropriate. ultimately if others don't want to use same setup (assuming i get it working ok) then it's up to them	13:02
Ilari	igli: Rebasing from upstream is safe if 1) The branch current branch is rebased on is the only one what current branch is pushed to or 2) the branch isn't published.	13:02
	igli: Otherwise, one needs to be a lot more careful.	13:03
igli	ok.	13:03
	yeah i think we'll have to try and mandate 1) for people with rw	13:03
	← jceb left	13:03
Ilari	igli: Also, rebasing branch that has been merged somewhere screws things up.	13:04
	→ jceb joined	13:04
	← jceb left	13:04
igli	hmm not merged from, just if i merge in that branch? hmm	13:05
Ilari	igli: Isn't there local-side hook for rebasing?	13:05
igli	dunno yet lemme see	13:06
Ilari	igli: Yes, pre-rebase.	13:06
	→ ProperNoun joined	13:06
igli	yes	13:06
	interesting	13:06
xorg62	is there an website applet to read git CSS and show it ?	13:06
Ilari	xorg62: 'git CSS'?	13:07
igli	ew perl. should be able to awkify that	13:07
xorg62	RSS	13:07
	sorryu	13:07
	:)	13:07
	← gitte left	13:08
xorg62	or just new commit	13:08
	→ snitko joined	13:09
igli	man that's lovely functionality	13:09
	This sample hook safeguards topic branches that have been published from being rewound.	13:10
	→ bremner__ joined	13:10
	j416 → j416-test	13:12
xorg62	Oo	13:12
	j416-test → j416	13:12
	→ squentin joined	13:14
	← sfwc left	13:14
igli	xorg62: http://git.or.cz/gitwiki/GitTips see Web	13:16
	sdiz-afk → sdiz	13:16
xorg62	thanks	13:17
igli	and next time try google like i just did ;)	13:17
	yw	13:17
xorg62	i already tried	13:17
igli	git rss feed howto # wfm	13:17
	have to admit we do ours via trac.	13:19
	→ Bass10 joined	13:19
xorg62	Hmm	13:21
	not i what i meant	13:21
	← igli left	13:22
	← dodo_the_last left	13:23
	→ dodo_the_last joined	13:26
	→ sfwc joined	13:27
	j416 → Takechan_	13:27
	Takechan_ → Takechan	13:27
	Takechan → j416	13:27
	← bremner__ left	13:28
	← kumbayo left	13:29
	← radarek left	13:33
	← bremner_ left	13:33
	← sfwc left	13:37
	→ romao joined	13:41
	← fujin left	13:42
	→ sfwc joined	13:44
	→ fujin joined	13:45
	→ icwiener joined	13:45
	→ Octalot joined	13:46
	→ Leemp joined	13:55
	← MisterN left	13:57
	→ WinterWeaver joined	13:58
	→ markelikalderon joined	14:01
thiago_home	how can I find the date of reflog entries?	14:01
Ilari	thiago_home: 'git reflog show HEAD@{now}' shows it.	14:02
	sdiz → sdiz-afk	14:03
	→ jrmuizel joined	14:04
thiago_home	how can I make it show actual dates?	14:05
	← cods left	14:05
Ilari	thiago_home: 'git log --walk-reflogs <something>'?	14:11
	← markelikalderon left	14:12
	→ rudi_s joined	14:13
	← tokkee left	14:14
	← jrmuizel left	14:14
	→ bdiego joined	14:15
thiago_home	Ilari: no dates	14:16
	I have the dates of the commits, not the dates of the reflog	14:16
	→ jrmuizel joined	14:17
	→ name joined	14:18
	→ ereslibre_desk joined	14:18
	→ wirt joined	14:21
	→ laj_ joined	14:21
	→ altrux joined	14:22
	← wirt left	14:25
	→ lhz joined	14:26
	← laj left	14:26
	→ wirt joined	14:26
	← wirt left	14:26
	← sfwc left	14:27
	→ sfwc joined	14:27
	→ Serg_penguin joined	14:29
Ilari	thiago_home: For me, that displays dates like 'Date: Thu May 21 19:32:44 2009 +1000'. And timezone Kilo definitely doesn't look right for local time...	14:29
	thiago_home: Whereas that reflog show showed dates with timezones Bravo and Charlie, which are much more reasoable for local ops.	14:30
thiago_home	exactly	14:30
	it shows the date of the commit, not the date of the action that created the log entry	14:31
	I know it keeps the date in the log (your earlier command showed that)	14:31
	I can parse it out of .git/logs/refs/heads/branchname, but I was wondering if there's anything in Git already	14:31
Ilari	thiago_home: Probably no reliable way to parse it out using standard tools...	14:32
thiago_home	it's just a Unix time	14:32
	$ tail -1 .git/logs/refs/heads/last-make-install	14:33
	a697385cd7929976808072d218ac2abe92fc3537 eb04785f8ac2a033bd32594be2fb037541c5dbf9 Thiago Macieira [email@hidden.address] 1242909289 +0200 make install in /home/tmacieir/src/kde4/KDE/kdelibs	14:33
Ilari	thiago_home: Yeah, but no reliable programmatic way to get it into say ISO standard time using standard git tools.	14:33
thiago_home	ah, I see what you mean	14:34
	← ereslibre left	14:34
	→ moonese joined	14:38
	→ scientes joined	14:39
	laj_ → laj	14:42
	→ aziz joined	14:42
davi	ciabot_git.pl e7b8da210fcf2f6462e87db5f474b23455dbba9c stable-001	14:44
	Executing ciabot_git.pl by hand works.	14:44
	However placing the below line in the post-commit hook create log of processes when I execute git-commit, which even hang.	14:45
	post-commit hook:	14:45
	/var/www/gnuherds-app/.git/hooks/post-commit $commit_sha1 $branch_name	14:45
	→ mintux joined	14:45
	→ dodo_the_last1 joined	14:46
	← ereslibre_desk left	14:47
	→ elpargo joined	14:47
	← knittl left	14:49
Ilari	davi: Usually you announce pushed commits not when they are made...	14:50
	← sfwc left	14:51
	← __iron left	14:52
	→ sfwc joined	14:53
	→ __iron joined	14:55
	→ jatt joined	14:56
Ilari	davi: Commit is purely local operation.	14:56
	← jatt left	14:56
	→ jatt joined	14:57
	→ targion joined	14:58
davi	Ilari, So, what hook should I use?	14:59
	post-receive?	14:59
Ilari	davi: Probably.	14:59
jatt	I'm starting git-daemon but I do get:	15:00
	fatal: unable to allocate any listen sockets on host (null) port 9418	15:00
Ilari	davi: And that gets invoked in repository that's pushed to.	15:00
davi	Is there a event directly associated to git-push?	15:00
jatt	why does git think my host is (null)?	15:00
Ilari	jatt: Git-daemon already running?	15:00
	← jrmuizel left	15:00
jatt	Ilari: yes :), thanks that was it.	15:00
Ilari	jatt: (null) in practice means IP 0.	15:00
davi	Ilari, The problem is that Savannah does not allow it. That is because we have to do it in our local development repositories	15:00
	← dodo_the_last left	15:01
Ilari	davi: There's no hook that runs on push on local side. Also, you can't alias push subcommand.	15:01
	← snitko left	15:01
davi	Ilari, so you advise try with post-receive hook in the Savannah central repository?	15:02
	← elpargo_ left	15:02
	← seangrove left	15:03
	→ jrmuizel joined	15:03
Ilari	davi: I don't remember if push is bultin or not. If it is, it isn't even possible to replace the subcommand executable.	15:03
	davi: Yes, it is....	15:04
	davi: So if you need hook for local side, you need to hack the git source.	15:04
davi	Is it developed in C, Python?	15:05
Ilari	davi: That part is in C.	15:05
davi	I would just need be able to execute ciabot_git.pl with the sha1 and branch name	15:05
cYmen	I already asked last night but is there any way to debug a fatal: The remote end hung up unexpectedly? I have full access to the machine I'm cloning from I just don't know where to look or turn it on	15:06
	→ kosh[0] joined	15:08
	→ memiux_ joined	15:08
Ilari	davi: Or make your own command that invokes git push.	15:08
kosh[0]	hi, I am just about to pull my hair out here	15:08
	and I aint sure why this is happening, so I was wondering to ask someone before I do it :D	15:08
Ilari	cYmen: What protocol? git://, ssh://?	15:08
	cYmen: And note that git@foo is NOT git://	15:09
kosh[0]	if I git checkout master, it changes all the files to master, exactly the way I expected it	15:09
	I want to clone a branch from a git repo I have on a usb, which originally was a clone of this local git repo	15:09
cYmen	Ilari: ssh://	15:10
kosh[0]	but the local git repo doesnt have this branch, only the cloned usb repo does	15:10
davi	Ilari, I see, just an script which be able to get the sha1 and branch name and execute the ciabot_git.pl script with such parameter for each commit done. Such commits done will be the ones which are going to be pushed.	15:10
Ilari	cYmen: Any other errors?	15:10
cYmen	Ilari: just a warning that I'm cloning an empty repo	15:11
kosh[0]	now, whenever I create a new branch and check it out, I have a copy of what was in master at the time	15:11
	\|kukks\| → kukks	15:11
kosh[0]	so I created a branch of master into original	15:11
cYmen	Ilari: is ssh://user@pc/path/to/file the right syntax? I'm suprised there's no colon in it.	15:11
	→ scarabx joined	15:11
kosh[0]	but I want to create original as a clone of a external repo's branch, not the local branch I currently have checked out	15:11
Ilari	davi: You should do it AFTER pushing them out.	15:11
	cYmen: Watch out, IIRC, that syntax is relative to ROOT directory (/).	15:12
	→ cilly joined	15:12
parasti	kosh[0]: man git-fetch	15:12
Gitbot	parasti: the git-fetch manpage can be found at http://git.or.cz/man/git-fetch	15:12
kosh[0]	aha, so for this we use git fetch, not git pull	15:12
	I was using git pul	15:12
	pull	15:12
davi	Ilari, But If I do it after how I know what I have pushed? If I do it before I just to do a look for the commit-already-done but not pushed?	15:12
cYmen	Ilari: I know that's how I wanted it and the repo is indeed empty I just created it with git init --bare	15:12
	← memiux left	15:12
parasti	kosh[0]: git pull is git fetch + git merge	15:13
	kosh[0]: I guess you want the fetch part, but not merge	15:13
	← xyz left	15:13
Ilari	davi: Look up the ranges before but do the actual notification after?	15:14
kosh[0]	parasti: ah okkk, thanks	15:14
	I didnt really get that until now	15:14
	thx	15:14
davi	Ilari, why do it after? To avoid problem if it fails?	15:14
	To avoid problem if git-push fails?	15:14
Ilari	davi: For instance for that reason. And the notifications only make sense after the commits are available.	15:15
davi	I see.	15:15
	Is there any example which I can look to know how to get such sha1 and branch names to committed but not pushed?, so I can pass the parameter to the Perl script?	15:16
	← jrmuizel left	15:16
	← Serg_penguin left	15:16
Ilari	davi: Parse the refspecs, update remote mirrors and use these two to find out what commits are pushed?	15:17
	← cilly left	15:18
	→ soulinx joined	15:21
cYmen	Is there any way to get verbosity or debug output?	15:21
soulinx	hello..How can I go back to a previous revision in a repo?	15:21
	I want to revert all the repo to commit commit 2662783e9ca60f1bcba7d6efaba814da4764d8d2	15:23
Ilari	soulinx: man git-reset	15:23
davi	Ilari, Thanks for the Hint. I think I am going to delay this hack	15:23
Gitbot	soulinx: the git-reset manpage can be found at http://git.or.cz/man/git-reset	15:23
Ilari	cYmen: The standard way of populating bare repo is by pushing to it.	15:24
	→ elpargo_ joined	15:24
Ilari	cYmen: Does 'git ls-remote ssh://user@pc/path/to/file' complain or just no output?	15:25
soulinx	hmm I am confused..What's the argument?git reset 2662783e9ca60f1bcba7d6efaba814da4764d8d2 ?	15:26
Ilari	soulinx: Also see what --soft/--mixed/--hard mean.	15:26
	← altrux left	15:28
cYmen	Ilari: it does not complain; no output	15:29
	← elpargo left	15:30
Ilari	cYmen: That's valid repo URL then. So make some commits (if you don't have already) and push them there.	15:30
	cYmen: One would think that cloning empty repo is easy thing. Its not. Its quite of an special case.	15:32
	cYmen: And ls-remote on empty repo produces no output.	15:33
cYmen	Ilari: apparently...because now that there is something in the repo the error message is gone	15:33
	→ rshann joined	15:34
	→ cemerick joined	15:35
Ilari	cYmen: Cloning repo that doesn't have on any commits on current/default branch is bad enough.	15:35
	← cemerick left	15:35
rshann	Can anyone tell me if there is an ftp server for git repositories on Savannah?	15:36
	← stuffcorpse left	15:39
	→ stuffcorpse joined	15:39
	→ aspotashev_ joined	15:39
cYmen	Ilari: then how am I supposed to set it up?	15:41
Ilari	cYmen: Create repo, push commits to it?	15:41
cYmen	What I wanted to do what just set up a quick repo where me and a friend could share some code	15:42
	Ilari: so I don't need to clone?	15:42
	memiux_ → memiux	15:42
	→ summerbug joined	15:42
	← the_zeno left	15:43
Ilari	cYmen: Multi-user repos are bit more involved than single-user as permissions/access control need to be right.	15:43
	← Arafangion left	15:44
Ilari	cYmen: And clone does stuff like setting up remotes, fetching remote and creating local branch for it and checking it out. Nothing special.	15:45
	→ rshann_ joined	15:45
cYmen	Ilari: branches I haven't completely figured out yet	15:46
	→ kosh[0]_ joined	15:46
kosh[0]_	ok, I have read the git fetch man page	15:47
	→ hyperair joined	15:48
Ilari	cYmen: In git, pretty much the only thing git needs to know about a branch is its name and ID of newest commit on it.	15:48
	→ robinr joined	15:48
kosh[0]_	but I can't figure out how to clone a remote branch into my local repo	15:48
	→ summerbug_ joined	15:48
kosh[0]_	git fetch /Volumes/N85/Documents/affilaites.git original-branch doesnt work	15:48
Ilari	kosh[0]_: 'git fetch <URL> original-branch:save-branch-as'.	15:49
kosh[0]_	I dont end up with a branch called original-branch (it's not actually original-branch) thats what I call it so you know what it is	15:49
	ahhhh ok	15:49
	← __iron left	15:50
	← stuffcorpse left	15:50
	→ __iron joined	15:50
	→ stuffcorpse joined	15:50
Ilari	kosh[0]_: push has that replicate left-side-to-right-side, but fetch doesn't.	15:50
thiago_home	it does	15:51
	oops, I see what you mean	15:51
	if you just name "original-branch", it saves as FETCH_HEAD	15:51
kosh[0]_	yeah, it seems to fetch into the current branch, not create a new one	15:52
	← summerbug left	15:52
	← Weaselweb left	15:53
kosh[0]_	AHA!!!	15:53
	ok, bloody hell, at the end	15:53
	← precision left	15:53
	← targion left	15:53
	→ precision joined	15:54
	← aspotashev left	15:55
	← rshann left	15:57
	← kosh[0] left	15:59
	→ aslakhellesoy joined	16:01
	← ericindc left	16:02
	← glommer left	16:05
	← WinterWeaver left	16:07
	← bobmcw_ left	16:10
	→ bobmcw joined	16:10
	← qhoxie_ left	16:10
	→ laj_ joined	16:12
	→ knittl joined	16:14
	← laj left	16:20
	→ elpargo joined	16:23
	joshthecoder_afk → joshthecoder	16:23
	← sfwc left	16:24
	→ sfwc joined	16:25
	← knittl left	16:26
cYmen	is there a way for me to have a local patch that will be applied automatically but never checked in?	16:27
	I have a different include path then all the others... :)	16:27
Ilari	cYmen: Create local branch, make the change on it and don't push this branch (and don't merge it into any branch either)?	16:29
thiago_home	that's called "rebase"	16:29
cYmen	hm..anybody happen to have step-by-step instructions? :)	16:29
thiago_home	git pull --rebase	16:29
	and be careful about what you push	16:30
	→ gitte joined	16:32
	← dodo_the_last1 left	16:33
cYmen	hm...I need to fix this first...so I checked it out and changed my include...the author changed something else now obviously I can't check it out again because it's not up-to-date...how do I fix this?	16:33
Ilari	cYmen: And be carefult what branch you do it on. And install git command completion and configure prompt to show the current branch.	16:34
cYmen	Ilari: yeah my zsh config already has that	16:35
Ilari	cYmen: 'git checkout -b localchanges', commit your changes, fetch (not pull!) and rebase your changes on top of new upstream.	16:35
cYmen	uh...what? :)	16:36
	→ dodo_the_last joined	16:36
	→ priidu joined	16:36
cYmen	ok...I'll try to figure out what that means ;)	16:36
Ilari	cYmen: First creates new branch and checks it out.	16:36
cYmen	thanks though!	16:36
Ilari	cYmen: The second step is to commit your changes for local system to new branch.	16:36
	→ Handcrafted joined	16:36
Ilari	cYmen: That branch is for local config changes ONLY. For changes that are applicable for others, use other branches.	16:37
	→ abstrusething joined	16:38
	← elpargo_ left	16:38
	→ akitada joined	16:39
Ilari	cYmen: In git, the standard workflows tend to involve many branches.	16:40
Handcrafted	I don't know if this is a stupid question, but: I'm going to use Git for keeping track of some of my personal projects, I got it recommended by a friend. So now I'm reading the Git Community book and I got to the part on the difference between Git and SVN. I was confused when I read "Git does not use delta storage". Instead Git uses a snapshot of the tree.	16:40
	My question is, doesn't this mean that the size of the repository will increase a lot everytime I commit a change? If it has to store both the version before and after?	16:40
	→ hagna joined	16:41
	gitte thinks it is "who can write the longest lines" day again.	16:41
g3d	Handcrafted: this is a different thing. git can store things in both ways.	16:41
cYmen	Ilari: but I build from the localchanges branch then?	16:41
Ilari	cYmen: Yes.	16:42
gitte	Handcrafted: the thing is: after a "git gc" (which is also automatically started from time to time), you have a _more_ efficient packing than your common delta packing.	16:42
hagna	so what do you do when you can't pull in changes and you have to push them but the repo you push to is not bare. seems like I read you can push to a branch on that repo and then merge is that so?	16:42
g3d	Handcrafted: to switch to "delta based storage" use git-igc	16:42
Ilari	Handcrafted: On object layer, it uses snapshotting, but lower layers use delta compression.	16:42
g3d	s/git-igc/git-gc	16:43
Handcrafted	Okay, thanmks everyone	16:43
	:)	16:43
gitte	Handcrafted: would be nice if you provided a patch to the community book.	16:43
Ilari	hagna: Yes, push is safe as long as the branch isn't checked out.	16:43
cYmen	Ilari: ok, I did "git checkout -b localchanges", made the change, ran "git commit -a"...now what's next? my other branch was master...	16:43
Handcrafted	gitte: Might do that when I understand Git better :)	16:43
gitte	Handcrafted: don't wait too long, or you will no longer be able to do it.	16:44
	Handcrafted: and doing it would be a nice way of saying "thank you".	16:44
Ilari	cYmen: Maybe check out master, 'git pull --rebase', then come back to localchanges and 'git rebase master'.	16:44
hagna	Ilari: so i just need to create a branch on this remote machine's repo or can I even do that from the client I will push from?	16:44
Handcrafted	gitte: Okay, I'll be looking into this git-igc thingy	16:45
cYmen	Ilari: "maybe"?	16:45
Ilari	hagna: Push will create new branch if the target doesn't exist.	16:45
g3d	Handcrafted: sorry, I meant git-gc	16:45
Ilari	cYmen: Or 'git pull --rebase origin master'...	16:46
Handcrafted	g3d: thanks	16:46
Ilari	cYmen: Directly from localchanges.	16:46
gitte	Handcrafted: I said "git gc".	16:46
cYmen	Ilari: what is origin in this case?	16:46
Ilari	cYmen: Ah yeah, it probably isn't a clone, so do 'git remote add origin <URL>'.	16:47
hagna	Ilari: ok so I have a remote named foobar then I do git push foobar mybranch?	16:47
Ilari	hagna: No.	16:47
hagna	dang	16:47
Ilari	hagna: 'git push foobar mybranch:incoming' (name the branch on remote side 'incoming') or something like that.	16:48
hagna	ahh ok	16:48
	I dno't use the : much	16:48
	if the remote happens to be origin can I do git push mybranch:incoming?	16:50
Ilari	hagna: No.	16:51
	hagna: If you specify refspecs, then specifying target is required.	16:51
	→ knittl joined	16:51
Ilari	hagna: Target can only be omitted if one uses the default specs for default repository.	16:51
hagna	Ilari: uh oh error: dst refspec incoming does not match any existing ref on the remote and does not start with refs/.	16:52
cYmen	Ilari: isn't it strange that i'm supposed to work on the master branch but build the localchanges branch?	16:53
	→ markelikalderon joined	16:53
	← stuffcorpse left	16:53
Ilari	hagna: Old version? Then maybe 'mybranch:refs/incoming' (that's not local branch 'incoming', but close enough).	16:53
	→ stuffcorpse joined	16:54
Ilari	hagna: local branch foo is refs/heads/foo.	16:54
hagna	old version on the local or the remote	16:54
	ahh it's the local	16:54
	remote is 1.5.3.7	16:54
	← davi left	16:55
hagna	local is 1.5.5.1	16:55
Ilari	cYmen: The localchanges is supposed to hold the non-distributable local configuration changes.	16:55
	cYmen: While master contains the distributable stuff.	16:55
	→ voker57__ joined	16:56
cYmen	so what does git pull --rebase origin master do exaclty?	16:56
Ilari	cYmen: Rebasing localchanges on top of master ('git rebase master' on localchanges) will rebuild your local changes on top of updated master.	16:56
cYmen	I mean there's three branches involved now right? origin, master and localchanges	16:56
Ilari	cYmen: Fetch branch 'master' from remote 'origin' and rebase current branch on top of that.	16:57
	← abstrusething left	16:57
cYmen	Ilari: I get a lot of 'trailing whitespace' messages....mean anything?	16:57
	And what do you do with a branch if you rebase it on top of another?	16:58
Ilari	cYmen: Probably just warnings by default. Trailing whitespace means there are whitespace characters on line between last-non-ws character and the LF that ends the line.	16:59
	cYmen: Rebase essentially tear down the branch and rebuilds it from patches.	16:59
	*tears	16:59
cYmen	so I should basically make a script now that checksout the localchanges branch now and builds?	17:00
hagna	Ilari: ! [remote rejected] masvbotlappy -> refs/incoming (funny refname)	17:00
	anyway I guess I'll make a patch and apply it on the remote	17:01
Ilari	hagna: 'mybranch:refs/heads/incoming' then.	17:01
	hagna: That's actual local branch 'incoming'.	17:04
cYmen	Ilari: now after all that rebasing do I still pull as usual?	17:04
Ilari	cYmen: Master is updated as usual, localchanges is updated by rebasing it on master.	17:06
	→ galderz joined	17:07
cYmen	so my build script should git checkout localchanges; git rebase master; ...build...; git checkout master ?	17:07
	← akitada left	17:07
Ilari	cYmen: Except that rebase can fail due to conflicts...	17:07
cYmen	well I only added one include so chances are slim	17:07
Ilari	cYmen: Then It exits waiting for conflicts to be resolved.	17:07
cYmen	mh...can I get in trouble if I try to checkout again or build?	17:08
	→ the_zeno joined	17:09
Ilari	cYmen: Best to check wheither the rebase succeded so one can intervene before the rest of script does any damage by running stuff with faulty assumption (in practicular, running checkout <branch> while rebase is in progress is no-no).	17:09
	→ ashleyw joined	17:10
	← ashleyw left	17:10
cYmen	hm...how would I check for that?	17:11
	does it return 1 or something?	17:11
	← rudi_s left	17:11
Ilari	cYmen: At least the HEAD is detached on conflict exit.	17:13
	← voker57_ left	17:13
Ilari	cYmen: Can be checked by using symboilic-ref to query HEAD.	17:13
	→ robomalo joined	17:14
cYmen	Ilari: hm...problem...I change something in master nothing major and would like to try it out by building but it won't let me change branches because I have changes -> am I supposed to commit everything not knowing if it even compiles?	17:14
	jefftriplett-afk → jefftriplett	17:14
	→ ashleyw joined	17:15
Ilari	cYmen: Stash the changes?	17:15
cYmen	Ilari: mh...can I build from a stash? or what are you suggesting?	17:16
	← sfwc left	17:17
cYmen	simple scenario: I change a parameter, want to try how it works...but my build script doesn't work because it's not allowed to change branches :)	17:17
	← memiux left	17:17
Ilari	cYmen: Well, maybe you could commit directly on localchanges. But you need to cherry-pick (not merge!) the change to master.	17:18
	→ memiux joined	17:18
	→ Sho_ joined	17:18
	→ sergiodj joined	17:18
cYmen	any other way?	17:18
	→ moonese1 joined	17:19
	→ ericindc joined	17:19
	→ kumbayo joined	17:19
Ilari	cYmen: On next rebase, the localchanges branch should drop its version of the change and use the one from master instead.	17:19
cYmen	could I use a third branch for local editing?	17:19
Ilari	cYmen: The main limitiation is that you can't merge anything that contains those local config fixes to anything else or your changes "escape".	17:20
	← fujin left	17:20
	→ tokkee joined	17:21
	← ashleyw left	17:21
cYmen	I don't understnad	17:21
	oh I think I do now	17:22
	← moonese left	17:22
cYmen	well that's why I suggest using that branch only for building	17:22
	and having one for local editing	17:22
Ilari	cYmen: This follows from branch model. Branches are essentially sets of commits, and merge corresponds to set union (may make extra commit in order to bind the histories).	17:22
cYmen	the question is if I have an editing branch where I commit every bullshit...can I later only merge the final change to the master branch?	17:22
Ilari	cYmen: Except that unles you clean it up before merging, your bullshit will become part of the master branch.	17:23
cYmen	exactly so is there a good way to clean it up or to only commit the diff between the current version and the last commit to master?	17:24
	← rshann_ left	17:24
	→ legnaleurc joined	17:24
Ilari	cYmen: The standard way to clean up commits is 'git rebase -i'.	17:24
	← jatt left	17:24
	→ ilteris__ joined	17:25
cYmen	now what does that do now?	17:25
Ilari	cYmen: Its interactive rebase.	17:25
cYmen	so what do you suggest?	17:25
	create another branch? edit in localchanges?	17:25
Ilari	cYmen: Allows splitting, editing, dropping and combining of changes.	17:25
	cYmen: Also, 'git commit --amend' is handy if you note you need to edit the last commit.	17:26
	← the_zeno left	17:27
Ilari	cYmen: Maybe read 'Git for Computer Scientists' and 'Git from the Bottom Up'.	17:28
	cYmen: Those two are really useful understanding how git really works.	17:29
	← gitte left	17:29
cYmen	ok...well thanks so far	17:31
	It's nice to be tutored	17:31
Ilari	cYmen: Basically, when you merge in a branch, it "pulls in" ALL commits from that branch.	17:32
	← galderz left	17:32
	→ willb1 joined	17:33
cYmen	hm...what does "file needs update" mean?	17:33
Ilari	cYmen: The file has uncommitted changes.	17:34
	→ WinterWeaver joined	17:36
cYmen	"rewinding head to replay your changes on top of it"?	17:36
Ilari	cYmen: That's just progress message.	17:37
cYmen	mkay :)	17:37
Ilari	cYmen: Also, if you didn't use commit -a, it would be possible to keep that include line uncommitted.	17:39
	← redondos left	17:39
Ilari	cYmen: But rebase definitely doesn't like such uncommitted lines.	17:39
	→ redondos joined	17:39
Ilari	cYmen: Maybe stash such lines for duration of rebase?	17:40
cYmen	hm...I could have a stash with local build patches	17:41
	sounds like it's worth trying out	17:41
	→ altrux joined	17:42
Ilari	cYmen: Hint: git gui	17:42
g3d	cYmen: I'd say better have a branch, branching is cheap	17:42
Ilari	cYmen: From RMB context menu for diff pane: "Stage hunk for commit", "Stage line for commit".	17:42
	cYmen: Yes, selecting changes to commit with line granularity.	17:42
g3d	or git add -i, it works for hunks too :)	17:42
	← willb1 left	17:42
cYmen	uh...don't know what a hunk is :)	17:43
g3d	cYmen: a fragment	17:43
	→ willb1 joined	17:43
Ilari	cYmen: It is (part of) changes. Hunks are delimited by sufficient number of non-changed lines.	17:44
cYmen	ok...	17:45
	g3d: so what do you mean by git add -i?	17:45
	← komplex left	17:45
	→ naeu joined	17:45
g3d	cYmen: it's a console-based "gui"	17:46
	→ moonese joined	17:46
g3d	cYmen: some people (including me) prefer it to the real gui	17:46
	→ bolle joined	17:46
	→ willb joined	17:48
	→ summerbug joined	17:48
cYmen	ok	17:49
	I don't have the basics down enough to get picky. But it will happen at some point ;)	17:49
	← willb2 left	17:50
	→ cpg joined	17:50
	← willb1 left	17:51
	→ kuadrosx joined	17:53
	→ Dashkal joined	17:55
Ilari	cYmen: When you get them down, you probably understand how to do it properly.	17:56
	→ HG` joined	17:56
	→ ashleyw joined	17:57
	← moonese1 left	18:02
Chani	how do I undo a git add?	18:02
	I always had trouble with htis in svn...	18:02
Ilari	Chani: 'git reset -- path/to/file'?	18:03
	→ the_zeno joined	18:03
Chani	Ilari: will that delete things?	18:03
	I don't want to elete any files, just make git forget that I added them. I haven't committed anything	18:03
	Chani forgot to delete the .svn folders before adding	18:03
	→ AkumaStreak joined	18:03
	← bremner left	18:05
Ilari	Chani: Then there is 'git rm --cached -r .svn'.	18:05
	→ bremner joined	18:05
Chani	oh. hmmm.	18:05
	although hte man page for git reset shows how to undo add :)	18:05
g3d	Chani: git will never delete or otherwise make you loose data unless you add arguments like "--force or --hard". So no worries.	18:06
	forgot some "" in there but you get the point hopefully	18:06
Ilari	Except that git rm by default deletes stuff from working tree as well.	18:06
Chani	g3d: except git rm ? :)	18:06
g3d	well ok, you got me there	18:07
Ilari	rm also has some safety checks, but I don't know if it prevents deletion in current case.	18:07
	← summerbug_ left	18:07
Chani	huh.	18:08
g3d	I'd say this really should be changed...	18:08
Chani	Ilari: fatal: pathspec '.svn' did not match any files	18:08
g3d	Chani: -r ?	18:08
Chani	g3d: yep	18:08
Ilari	Chani: Was anything under it even added?	18:08
	→ willb1 joined	18:09
	← willb left	18:09
	← markelikalderon left	18:09
Chani	Ilari: yes. just not committed	18:09
Ilari	Chani: git staus should show what .svn dirs have tracked stuff under it.	18:09
Chani	yep.	18:09
	ohhh	18:09
	Ilari: it only works if I first cd into one of the folders that has a .svn	18:10
	maybe -r isn't working correctly	18:10
Ilari	Chani: It is.	18:10
	← hagna left	18:10
	← HG` left	18:11
Ilari	Chani: It means recurse from specified. -r to grep means the similar thing.	18:11
Chani	yabut, it's not recursing	18:11
	← memiux left	18:11
Chani	hmm. looks like I don't have many subfolders anyways. I can do them manually	18:12
	→ cloudhead joined	18:12
Ilari	Chani: Its recursing to that .svn.	18:12
Chani	ohhhh	18:12
	yeah, I just got it	18:13
	duh.	18:13
	it's not recursively looking for .svn folders :)	18:13
Ilari	Chani: Neither would 'grep -r foo .svn'.	18:14
Chani	yeah.	18:15
	→ HG` joined	18:17
	→ rittyan1 joined	18:18
	→ cemerick joined	18:18
	← ashleyw left	18:20
	← aslakhellesoy left	18:21
	← agib left	18:23
	→ sunoano joined	18:23
	← HG` left	18:23
	→ ashleyw joined	18:24
	→ bremner_ joined	18:24
	← AkumaStreak left	18:24
	← ProperNoun left	18:24
	→ markelikalderon joined	18:25
	← bremner_ left	18:25
	← willb1 left	18:26
	→ bremner_ joined	18:26
	← ashleyw left	18:26
	→ memiux joined	18:27
	→ radarek joined	18:29
	← bremner_ left	18:29
	← rittyan left	18:30
sako	anyone here using git for web development projects?	18:33
	→ bremner_ joined	18:34
	→ memiux_ joined	18:35
Ilari	sako: What's so special about them? Making deploy-on-push work?	18:36
	→ jatt joined	18:37
	rittyan1 → rittyan	18:37
	← DrFrasierCrane left	18:37
	← rittyan left	18:38
	→ rittyan joined	18:38
	→ __name__ joined	18:39
	← name left	18:39
	__name__ → name	18:40
sako	elliottcable: so i cloned my git repo on my laptop	18:40
	→ bdowning_ joined	18:40
sako	when i do changes and do commit -a	18:40
	that commits to my local repo?	18:40
	← angerman left	18:41
	→ bobesponja joined	18:41
Ilari	sako: Yes.	18:41
sako	hmmm	18:41
	so when i commit changes and i want them to go live	18:41
	do i have to push>	18:41
	← memiux_ left	18:41
	← sdboyer left	18:42
Ilari	sako: Push transfers commits between repositories. To make it live, you need something else (unless your webserver can read directly from git repos).	18:42
	← bdowning left	18:42
elliottcable	sako: if django is anything like Rails, you can easily set up a git remote that will reload when you push to it.	18:43
sako	i see	18:43
elliottcable	sako: I personally keep a "staging" remote, and push to that when I want to try some changes on the development server.	18:43
	→ sshc joined	18:43
	bdowning_ → bdowning	18:43
sshc	what's wrong with git pull?	18:43
	→ memiux_ joined	18:43
	← memiux_ left	18:44
sako	elliottcable: so the "staging	18:44
	is that like a development copy of your server? so you can test and then go live?	18:44
	→ moonese1 joined	18:45
Ilari	sshc: git pull is either fetch + merge or fetch + rebase...	18:45
sshc	Ilari: so I cane git fetch && git merge instea of git pull?	18:46
Ilari	sshc: Yes.	18:46
	← memiux left	18:46
	← the_zeno left	18:47
sako	http://joemaller.com/2008/11/25/a-web-focused-git-workflow/	18:47
	← patrikf left	18:47
	← impulze left	18:47
sako	Ilari: does that link look good?	18:47
	im gonna follow the setup for a repo like that..	18:47
	→ LCID_Fire joined	18:48
LCID_Fire	Hi	18:48
cYmen	What's the default way to solve a merge conflict anyway? I just replace the broken file with what I want to be the merge result?	18:48
	← Dashkal left	18:49
LCID_Fire	cYmen: IMO you should manually check the rejected changes!?	18:49
Ilari	sako: The update-server-info is only needed for fetching via protocols like http:// and ftp://, so presumably not necressary.	18:50
LCID_Fire	Is there a way to turn a fetched svn directory structure (branches, tags, trunk) into git branches?	18:51
Ilari	cYmen: 'git add <file>'.	18:51
	→ elpargo_ joined	18:51
Ilari	cYmen: Was it merge or rebase that conflicted?	18:52
	Chani sighs	18:52
Chani	I have a problem.	18:52
sako	Ilari: so if i just follow that and remove the exec update-server-info ill be fine?	18:52
cYmen	Ilari: rebase	18:52
	LCID_Fire has a bunch	18:52
Ilari	sako: Well, it isn't necressary but doesn't hurt that much (some CPU usage & disk usage).	18:52
	→ WALoeIII joined	18:52
Ilari	cYmen: Then after adding, 'git rebase --continue'.	18:53
	← kukks left	18:53
	→ ibolmo joined	18:53
	→ camwest joined	18:53
Chani	our translators need to read the doc/ subfolder of every single kde module. since kde is big, it's probably not feasible for them all to check out all of kde. but git doesn't seem to offer a way to download only a certain folder. --depth gives a way to only get part of the history, but that's not useful for this. :(	18:54
sako	hmmm i gotta read a good book or something :)	18:54
	any suggestion?	18:54
cYmen	Ilari: Is there any tool to resolve conflicts?	18:54
Chani	isn't there any way to download only part of a git repo?	18:54
	→ impulze joined	18:54
Ilari	sako: For some really insightful stuff, google "Git for Computer Scientists" and "Git from the Bottom Up".	18:55
LCID_Fire	Chani: AFAIK not	18:55
Chani	:(	18:55
Ilari	cYmen: There's git mergetool.	18:55
cYmen	Ok, right now I can't even see the new file so that's sort of weird.	18:55
LCID_Fire	Chani: Personally I'd say it's a bug not a feature ;)	18:55
Chani	this is kinda blocking KDE from switching to git.	18:56
	← WinterWeaver left	18:56
	← summerbug left	18:56
jatt	hi, how would someone do:	18:56
	scp -rp .git [email@hidden.address]	18:56
	git push --all	18:56
	? I found this in a guide about git, without explanation...	18:56
	It's possible to pull changes for such a repository?	18:56
LCID_Fire	Chani: Do they need to change the files?	18:56
Ilari	jatt: Don't use scp to copy git repositories.	18:58
g3d	jatt: a bare repository is just that, the contents of .git/ dir in normal repository	18:58
Chani	LCID_Fire: no.	18:58
LCID_Fire	Chani: why don't you then provide a web interface?	18:58
Chani	LCID_Fire: our translators hate web interfaces	18:58
	with a passion	18:58
	plus, when I say they need to read them, I mean they need to run a script which reads all of them	18:59
	→ khmarbaise_ joined	18:59
jatt	g3d: it's possible to pull changes for such a bare repository? I cloned the repostory and I want to have those changes back in the bare repository I created.	18:59
Ilari	jatt: Instead, init the repo on remote side and push needed branches there.	18:59
cYmen	Gna, I accidentally git add blafile.orig - what do I do? :)	19:00
Chani	if they had to read them directly I'd have an even bigger problem, because right now those folders are automagically fetched as svn externals, and with git there'll have to be a script to fetch them even if I find a way to do this	19:00
LCID_Fire	Chani: what about using git for normal development and using svn for translation stuff?	19:00
	← ibolmo left	19:00
jatt	Ilari: yes you are right, I think using scp is a pretty bad idea, I was following the following guide: http://vafer.org/blog/20080115011320	19:00
Chani	LCID_Fire: the translator are going to stay on svn, at least from now. but projects want to move to git. and each group of projects has its own doc/ folder.	19:01
Ilari	cYmen: 'git rm --cached blafile.orig'.	19:01
	← moonese left	19:01
Chani	LCID_Fire: I have to design a hybrid system here, and I almost had it until I came across this folder full of svn externals :(	19:01
g3d	jatt: what do you mean by "pull for"? pull to? if so, you can't, as it's bare repository. better do it right as Ilari said.	19:02
Ilari	jatt: There are lot of crap guides about git. Hopefully those guides that teach the use of update-index (git 1.4.X, really obsolete) have already disappeared.	19:02
LCID_Fire	Chani: you should probably talk to some of the git developers since your request is something I heard a lot of people complain about	19:02
	← sergiodj left	19:03
Chani	the only other solution I can see right now is to have someone run a server that checks out every kde git repo every day, copies the doc folders into one folder, and offers that folder up on the interwebs for the translators' script to download or rsync or whatever	19:03
jatt	as far as I understood the guide did	19:03
	scp -rp .git [email@hidden.address]	19:03
	because gitweb probably needs the .git file only. I don't know I think I will start from scratch again	19:03
	this time without scp...	19:03
	→ ibolmo joined	19:03
cYmen	Ilari: I called git add blafile and git rebase --continue but it says nothing new :/	19:03
ibolmo	are hooks committed to the repo or are they just local?	19:04
Ilari	ibolmo: Local.	19:04
	cYmen: Ah that. 'git rebase --skip'-	19:04
g3d	jatt: as I said before, a bare repository is basically contents of .git/ on normal repositories, i.e. repo without the working dir.	19:04
cYmen	Ilari: but won't that skip my patch?	19:04
Chani	LCID_Fire: what's the best way to contact the developers? irc, mailing list?	19:04
ibolmo	ah ok, so i don't have to worry about putting in sensitive information?	19:04
	→ abstrusething joined	19:04
LCID_Fire	Chani: I'd say mailing list	19:04
Ilari	cYmen: If the conflict resolution is identical to what existed before, that patch should be skipped, no?	19:05
	→ fujin joined	19:05
Chani	LCID_Fire: [email@hidden.address] ?	19:05
	→ aslakhellesoy joined	19:05
Ilari	ibolmo: Actually, hooks can be quite dangerous, so they aren't copied by any git tool.	19:05
ibolmo	hehe gotcha	19:05
Ilari	ibolmo: Hooks can cause enough security problems already in shared repos.	19:06
	← elpargo left	19:06
LCID_Fire	Chani: might be - I unsubscribed a long time ago ;)	19:06
cYmen	Ilari: Why? I mean I still want my local changes applied, right?	19:06
hyperair	sandsmarK: which represents 'undefined' better? NaN or inf?	19:06
	ugh shit wrong channel	19:07
jatt	g3d: I see, but how is the workflow when working with bare repositories? For instance I import a project to it and another user makes a clone of it. If that user wants to upload his changes, how can he do it? git push?	19:07
Ilari	cYmen: Rebase tries to detect already appiled patches, but when that fails, you need to manually drop the duplicate.	19:07
g3d	jatt: yes, by pushing usually. you could update it by fetching for example. just not pull, as it does merge.	19:08
	jatt: although normally you just push	19:08
jatt	g3d: thanks, I think I understand now	19:08
Ilari	jatt: Pushing to non-bare repos is possible, but one needs to be careful.	19:09
g3d	jatt: and yeah, your repository is still not bare, edit its config	19:10
hyperair	what are the implications of pushing to a non-bare repo?	19:10
ibolmo	Ilari: perhaps a hook is not something i'd want to do.. i'd like to have a %%variable%% in a .cfg file that i'd like to change depending on the branch name..	19:10
jatt	I expect to use the bare repo only for gitweb purposes. I have been using git for a while but only used git pull until now.	19:10
ibolmo	anyway to have the same file appear different on both branches? basically removing the need of the %%variable%%	19:10
jatt	g3d: how did you recognize that the repo wasn't bare yet?	19:10
ibolmo	without worrying about conflicts on merges	19:10
g3d	jatt: because you just copied your non-bare one	19:11
	jatt: at least that's what I understood	19:11
	jatt: see in config core.bare	19:11
Ilari	ibolmo: Ugh. Can't be sanely done.	19:11
ibolmo	haha	19:12
	← aslakhellesoy left	19:12
jatt	g3d:	19:12
	bare = false	19:12
		19:12
	→ npmccallum joined	19:12
jatt	hm. that crappy guide got all wrong then :)	19:12
Ilari	ibolmo: Its even worse issue than the one with keywords.	19:12
npmccallum	can I have a single git repo with other git repos inside of it?	19:13
Ilari	npmccallum: Its possible to put git repo inside other, either as untracked or submodule.	19:13
g3d	jatt: yep	19:13
ibolmo	what do you mean Ilari?	19:13
	← bx2 left	19:13
npmccallum	Ilari: basically, there are a 5 other git or svn repos out there I want to track in a single repo, along with some other local files	19:14
	Ilari: are there instructions anywhere to do this?	19:14
Ilari	ibolmo: Similare strategies at resolving it could work as in case of keywords.	19:14
	← bremner_ left	19:14
	→ ashleyw joined	19:14
	← ibolmo left	19:15
cYmen	Ilari: How does it detect already applied patches? I mean none of my patches are already applied...	19:15
	→ bremner_ joined	19:15
elliottcable	sako: sorry, had to run out	19:16
	sako: did you get it working?	19:16
sako	not yet	19:17
	→ icwiener_ joined	19:17
	← bremner_ left	19:17
sako	im trying to figure out if i should use the method described in that url..	19:17
	http://joemaller.com/2008/11/25/a-web-focused-git-workflow/	19:17
Ilari	cYmen: Compute patch IDs (distinct from commit IDs).	19:17
	→ archis joined	19:18
Ilari	cYmen: But conflict resolution winding up being identical to previous commit is good sign of patch that should be dropped.	19:18
	sako: Seems fair workflow.	19:18
	← WALoeIII left	19:18
	→ bremner_ joined	19:20
	← scarabx left	19:21
	← LCID_Fire left	19:22
	→ eletuchy joined	19:24
	← bgerber left	19:25
Chani	>.<	19:25
	I did a search on the mailing-list archive, and found some results, but I only get individual emails with no navigation links	19:26
	without a way to read the whole thread the messages I found aren't much use :(	19:26
	I can't figure out how to get back to the frames view that had navigation buttons	19:26
	← camwest left	19:28
Chani	n/m. figured it out.	19:28
Handcrafted	Okay, remeber me from before, the guy with the problem in the Git Community Book. Well, I just submitted my commit. Thank you all for telling me to do that. I don't think I could've learned to use Git in a better way :)	19:29
	→ camwest joined	19:30
	← radarek left	19:30
	← Handcrafted left	19:31
	→ aslakhellesoy joined	19:32
	← icwiener left	19:33
	→ djgera joined	19:34
cYmen	Ilari: I don't get how the patches are not applied but my changes are still gone	19:34
Chani	so, uh, stupid question. how do I subscribe to [email@hidden.address] ?	19:35
	← legnaleurc left	19:36
	← rjc left	19:36
	→ rudi_s joined	19:39
	→ moonese joined	19:39
	→ rjc joined	19:40
Ilari	cYmen: You screwed up the conflict resolution?	19:40
	← aslakhellesoy left	19:40
cYmen	hm...I don't know my change weren't in my local file either I mean they are in the patches...	19:41
	← ashleyw left	19:43
	→ kosh[0] joined	19:46
	← archis left	19:47
	← bremner_ left	19:47
	← soulinx left	19:51
	← kosh[0]_ left	19:51
	→ melvinram joined	19:55
	→ ashleyw joined	19:55
	→ lvh joined	19:55
lvh	hi :-)	19:55
Ilari	...	19:56
	← moonese1 left	19:56
	← sunoano left	19:56
	→ ashleyw_ joined	19:57
	← melvinram left	19:58
	→ melvinram joined	19:58
	← Sho_ left	20:00
	→ danielnewman joined	20:02
	← ashleyw_ left	20:05
	→ memiux joined	20:05
jatt	I've setup a bare repository and did a push of my local repository to populate it. gitweb shows the bare repository correctly and I can clone it with:	20:06
	git clone ssh://myserver.com/var/git/myapp.git	20:06
	however with	20:06
	git clone http://localhostgitweb/myapp.git	20:06
	the clone fails	20:06
	→ rolfb joined	20:06
jatt	with	20:06
	fatal: http://localhost/gitweb/ubnt-hal-0.7.379.git/info/refs not found: did you run git update-server-info on the server?	20:06
Ilari	jatt: Read that error message.	20:07
jatt	I did run git update-server-info on the server, but the same error happens. The strange thing is that git clone works (giving the path of the repository on the server0.	20:08
	← JEEB left	20:08
	→ ShadeHawk joined	20:08
Ilari	jatt: ssh://, git:// and file:// don't need that extra info.	20:08
	→ [AndrewR] joined	20:08
ShadeHawk	Today is (according to GSoC timeline) time for GSoC students to start working on projects. Shouldn't they present themselves on git mailing list?	20:09
	← ashleyw left	20:13
ShadeHawk	there is quie a bit of "foreign" GSoC projects which touch git more or less; I think we 'touched base' with SuSE Build System (to be on top of Git) and GitFS for Plan9	20:14
jatt	Ilari: sorry which extra info do you mean?	20:14
cYmen	is there a default "always remote" or something for merging?	20:15
jatt	now that I have gitweb working I wanted just to clone the repository using the URL, i.e.: git clone http://localhost/gitweb/ubnt-hal-0.7.379.git	20:16
	← rjc left	20:16
ShadeHawk	I wonder how it is with git# people from Mono, from PyVCAL VC Access Library from Python, hg-git (Mercurial over Git) from Mercurial	20:17
	→ rjc joined	20:17
jatt	it is possible to do it with a bare repository?	20:17
	← rudi_s left	20:17
	→ angerman joined	20:18
ShadeHawk	jatt: gitweb has nothing to do with cloning; moreover you need to do extra to have the same URL for gitweb and for HTTP fetch	20:18
	jatt: There is I think something in Documentation/howto about that	20:18
jatt	ShadeHawk: I understand. I thought it would happen automatically :).	20:19
	← mintux left	20:19
	← bobesponja left	20:20
ShadeHawk	jatt: and you shoulr run git-update-server-info (perhaps from hooks) to be able to serve via HTTP	20:22
	pasky: Something broken on Git Wiki with upgrade	20:23
jatt	ShadeHawk: ok I will take a look. I just want that once I say my colleagues to pull that they can use just the URL and that they don't need to ask for the path of the respository on the server. thanks!	20:23
pasky	ShadeHawk: what broke?	20:23
	← lvh left	20:23
	→ KuSuRSuz-yok joined	20:23
ShadeHawk	pasky: check out RecentChanges (I see two gray boxes, no changes lis, no images), and e.g. SoC2009Projects page, with excessive vertical whitespace for single {{BR}}	20:23
KuSuRSuz-yok	hii	20:24
	← KuSuRSuz-yok left	20:24
ShadeHawk	jatt: and gitweb has configuration which allow to state on projects summary page where to find repositories to fetch / clone from (see gitweb/README or gitweb/INSTALL)	20:25
pasky	ShadeHawk: hmm, i'm clueless on what broke :(	20:25
Ilari	jatt: HTTP is inferrior. And you can have clone URLs anyway on the gitweb page.	20:25
	← bdiego left	20:26
jatt	Ilari: really, wow that sounds like the feature I'm looking for. I don't see that option in the gitweb page I got per default... do I need to configure gitweb for it I guess?	20:27
ShadeHawk	pasky: what's more, it looks like RecentChanges page has box or iframe or frame or something, which has link to itself... which if you position scrollbars right lead to funny recursive page	20:27
	jatt: yes, you need	20:28
	← memiux left	20:29
ShadeHawk	pasky: By the way, were you thinking about replacing MoinMoin by something else? (ikiwiki, SocialText, MojoMojo, FosWiki <-- all in Perl; or it it is to be mved to git-scm.com, perhaps something in Ruby...)	20:29
	s/it it/if it/	20:30
jatt	I see now that the kernel.org gitweb page has the link and has a message informing the user how to clone the repository, so yes using HTTP to clone them is probably not the right thing.	20:30
	← cpg left	20:30
	← rjc left	20:31
ShadeHawk	jatt: I think gitweb/INSTALL or gitweb/README has it well explained; if not, tell me what were not clear enough, please...	20:31
	→ ashleyw joined	20:32
	ShadeHawk is still reading git wiki via proxy; damn TP S.A. and its blackholing / null route firewall (misconfigured MAPS?)	20:34
	→ wezyde joined	20:34
	→ rjc joined	20:36
	← ashleyw left	20:37
pasky	ShadeHawk: frankly, i have no time or resources to take care about the wiki at all - someone else must do all the work	20:37
	actually, i have no idea how to fix the totally borked wiki now	20:38
	i don't have another few hours to fiddle with that :(	20:38
ShadeHawk	pasky: Can you ask Scott Chacon (of git-scm.com) to move and take care of git wiki?	20:40
pasky	i think we talked briefly	20:44
	i will try to prod him	20:44
	→ memiux joined	20:44
	← icwiener_ left	20:45
	← rjc left	20:45
	→ icwiener joined	20:46
ShadeHawk	pasky: I wonder what wiki software would he prefer	20:46
	→ rjc joined	20:46
	→ ashleyw joined	20:46
	← Ademan left	20:48
	← ashleyw left	20:48
ShadeHawk	perhaps git-wiki or Jekyll (~= ikiwiki)?	20:48
	→ Aeroforce joined	20:50
	← Aeroforce left	20:51
	→ blink joined	20:54
blink	Anyone up for answering some questions relating to git push?	20:55
	→ ashleyw joined	20:58
	→ rshann_ joined	20:59
jatt	I tried to find the gitweb feature that turns the "git" links on (as in http://git.kernel.org) but couldn't find it. I even checked the gitweb.perl source code for clues ;) (git v1.6.3.1). Do you know how to enable this feature?	21:01
blink	I have an origin repo and an official repo, both with a master branch. I have official/master in a tracking branch canonical. When in my local master branch push will push to origin branches as expected, but when I am in canonical git push does not simply push it's objects to official/master. Is there a configuration option to prevent myself from accidentally pushing local master to official/master when pushing from inside canonical?	21:01
	→ cpg joined	21:02
	← sea-gull left	21:02
	→ loincloth joined	21:03
	← markelikalderon left	21:04
	← rphillips left	21:04
	→ rphillips joined	21:05
	→ aslakhellesoy joined	21:08
	→ kennethlove joined	21:09
	→ pano joined	21:09
	← rjc left	21:12
	→ rjc joined	21:12
	cpg → cpg\|lunch	21:17
	→ gitte joined	21:18
	← jatt left	21:19
	← melvinram left	21:24
pasky	now the goddamn thing has locked me out by surge protection and i don't know why	21:24
	← camwest left	21:25
	→ jrmuizel joined	21:27
ShadeHawk	Do you think it would be useful to have "Git User's Survey 2009" (this time without that many free-form questions)? How many responses would there be? The sequence 100 (2006), 600 (2007), 3000 (2008) hints at 15000,..	21:30
	← rshann_ left	21:32
Pieter	ShadeHawk: yes, but tone down the # of questions.. less than 20, I'd say, with 2-3 open max	21:33
ShadeHawk	Pieter: what questions would you like to see in survey?	21:34
	← aspotashev_ left	21:34
	→ aspotashev joined	21:34
Pieter	why people use git (artist/programmer/..), if it's for fun or profit, what system they run git on	21:34
	→ Ademan joined	21:34
Pieter	and GUI use, use of cgit,jgit,or something else	21:35
	→ Cowmoo joined	21:35
Cowmoo	How can I list all files in my git repo?	21:36
Pieter	git ls-tree -r HEAD	21:36
	→ nh2 joined	21:36
Cowmoo	thanks	21:37
nh2	how do I disapply some patches (e.g 3 out of 100)?	21:37
thiago_home	nh2: git show patch1 patch2 patch3	21:37
Pieter	pasky: is repo still b0rken?	21:37
gitte	pasky: you were talking about repo.or.cz, I guess.	21:37
ShadeHawk	well... implementation (git, jgit, grit, Dulwich, git#, ObjectiveGit,...) / GUI (git-gui, gitk, qgit, GitX, GitNub,...) / tools (gitweb, jgit, StGit, Guilt, TopGit, ...) -- that are three questions	21:37
pasky	Pieter: it seems so	21:38
gitte	pasky: can you connect from another IP?	21:38
ShadeHawk	nh2: disapply?	21:38
pasky	gitte: no, about gitwiki	21:38
	anyway	21:38
	i'm giving up on fixing gitwiki	21:38
nh2	ShadeHawk: I mean reverse them	21:38
pasky	i just don't know what's wrong with it, and i've just wasted another hour on it	21:38
gitte	pasky: gitweb is broken, too...	21:38
pasky	i know :/	21:38
	i wanted to fix gitwiki first	21:38
nh2	there is a feature I do not like in a current Branch	21:38
gitte	pasky: what is the symptom?	21:38
pasky	gitte: look at GitFAQ or RecentChanges	21:38
ShadeHawk	nh2: applied comits? man git-revert	21:38
nh2	and I want to take it out again	21:38
Gitbot	ShadeHawk: the git-revert manpage can be found at http://git.or.cz/man/git-revert	21:38
Pieter	ShadeHawk: you can probably dump them all together though.. most people won't know what implementation they use I guess	21:38
pano	is anyone here that knows about git? :-D	21:39
	I've got this strange problem	21:39
	setting up a personal clone on gitorious...	21:39
gitte	pasky: you mean the gray boxes?	21:39
pano	already tried to ask in #gitorious but got no response yet	21:39
pasky	gitte: yes, and missing toc and large spacings...	21:40
ShadeHawk	pasky, gitte: After upgrade problems... why it always hits git wiki?	21:40
pasky	gotta travel from work to home now	21:40
	ShadeHawk: because moinmoin is a piece of crap	21:40
thiago_home	pano: you haven't asked a question yet	21:40
pasky	awful one, too	21:40
	← Cowmoo left	21:40
gitte	pasky: looks like a css problem to me.	21:40
pano	can I just paste the stuff I wrote in #gitorious?	21:40
ShadeHawk	pasky: probably moin4wiki doesn't help, either...	21:41
gitte	pasky: are you sure that it reads from the correct place?	21:41
thiago_home	if it's less than 4 lines	21:41
pasky	gitte: i'm not sure about anything anymore, but it should	21:41
ShadeHawk	gitte: RecentChanges doesn't show recent changes	21:41
pano	thiago_home: this means I can't :-D	21:41
gitte	pasky: have you looked into the access.log?	21:41
pano	k	21:41
ShadeHawk	pasky, gitte: And SystemInfo didn't work even before upgrade	21:41
pano	I created an account on gitorious	21:41
gitte	pano: there is a wonderful link in our topic to a pastebin IIRC.	21:41
pano	gitte: I'll just describe my problem again^^ thanks though	21:42
	→ scientes_ joined	21:42
pano	I then added a newly created ssh key to it, and created a clone of a project	21:42
pasky	gitte: what should i look for there?	21:43
pano	if I try to do git clone ssh://pano@gitorious.org/and/so/on I always get some errors as output	21:43
	Permission denied (publickey).	21:43
	fatal: The remote end hung up unexpectedly	21:43
gitte	pasky: for the path of the css files, I guess.	21:44
pano	My guess is, that there is some ssh configuration issue on my side, but I can't figure it out	21:44
nh2	ShadeHawk: great, it seems to work	21:44
pasky	212.198.82.134 - - [23/May/2009:23:37:10 +0200] "GET /wiki/modern/css/print.css HTTP/1.1" 200 775 "http://git.or.cz/gitwiki/GitFaq" "Mozilla/5.0 (X11; U; Linux i686; fr; rv:1.9.0.7) Gecko/2009030814 Iceweasel/3.0.9 (Debian-3.0.9-1)"	21:44
	stuff like that	21:44
	afk back in 1 hour	21:44
thiago_home	pano: when you go to your project's page, what does it say under Push url?	21:44
nh2	but can you tell me why one has to git add the changed files manually after reverse?	21:44
ShadeHawk	pano: 1. Try "ssh [email@hidden.address] 2. do you have gitorious key in "ssh-add -l"?	21:45
gitte	pasky: hmm.	21:45
	pasky: it looks as if a text-only browser does not see recent changes, either.	21:45
	← British0zzy left	21:46
pano	pano@arch ~/Projekte/rekonq % ssh [email@hidden.address]	21:46
	Bad owner or permissions on /home/pano/.ssh/config	21:46
	wtf?^^	21:46
thiago_home	pano: how about my question?	21:46
	pano: when you go to your project's page, what does it say under Push url?	21:47
pano	thiago_home: hmm, can't seem to find what you mean	21:47
ShadeHawk	gitte: I read git wiki through proxy (because of my ISP blackholing repo.or.cz for some reason or other), and on RecentChanges I have nice, recursive gray boxes (i.e. there is box with scrollbar, in which is the same page with box with scrollbar, inside which...)	21:47
thiago_home	if you don't understand me, tell me. Don't ignore me.	21:47
	pano: go to http://gitorious.org/~pano/rekonq/pano-clone	21:47
	pano: what do you see under Push url	21:47
gitte	pasky: it looks as if the RecentChanges template got fscked up.	21:48
pano	thiago_home: [email@hidden.address]	21:48
thiago_home	pano: what URL did you try to clone?	21:48
pano	I entered "git clone ssh://pano@gitorious.org/~pano/rekonq/pano-clone.git"	21:49
thiago_home	do you see that they are not the same?	21:49
pano	thiago_home: yeah... sry It now asks me for my passphrase, I guess it'll work now. Thanks!	21:50
ShadeHawk	thiago_home, pano: I think they should be equivalent (there is scp/ssh like way of writing SSH URLS, and there is URL-like way of writing SSH URLs)... but might not be	21:50
	← jrmuizel left	21:51
thiago_home	ShadeHawk: the username is not the same	21:51
	ShadeHawk: that was pano's error	21:51
	→ ashleyw_ joined	21:52
	← cemerick left	21:52
	→ Shinu joined	21:52
ShadeHawk	ah, sorry, I didn't notice	21:52
gitte	pasky: This is a wrong link, it seems: /gitwiki/TableOfContents%282%29?action=content	21:53
pano	thiago_home: thanks again for your help, and sry for my blindness :-)	21:53
ShadeHawk	I just copied the URL they gave to config file and forgot about it...	21:54
	for my gitorious project...	21:54
gitte	pasky: sorry, the linked file is empty...	21:57
	← kate21de left	21:58
	← scientes left	21:59
	← wezyde left	22:00
	← g3d left	22:00
	← hyperair left	22:02
	← schlort left	22:04
	← rjc left	22:05
	→ ashleyw__ joined	22:05
	→ rjc joined	22:05
nh2	what does git want to tell me saying "CONFLICT (delete/modify): [file] deleted in 8421f29... blabla and modified in HEAD. Version HEAD of [file] left in tree." after a git revert?	22:07
	sorry _in_ a git revert	22:07
	← ashleyw left	22:07
ShadeHawk	nh2: that the file was added in commit you want to revert, and was modified since (I think)	22:08
	← lhz left	22:09
nh2	ShadeHawk: so the revert wants to remove it again, right?	22:09
	← Ademan left	22:10
	← aslakhellesoy left	22:10
ShadeHawk	nh2: yep, revert of addition is removing... and meanwhile you have modified the file	22:12
	nh2: if you want to remove addition of file (i.e. remove a file), just "git rm <file>" and commit change	22:12
nh2	ShadeHawk: ok, then it is clear to me	22:13
	ShadeHawk: thank you	22:13
ShadeHawk	nh2: you are welcome	22:14
	→ jrmuizel joined	22:14
ShadeHawk	I see that somebody is asking on git mialing list about partial checkouts. I wonder what is happening on this front.	22:16
	There are a few issues which were started, and never finished...	22:16
	← Fullmoon__ left	22:16
	← khmarbaise_ left	22:16
ShadeHawk	* pack v4	22:16
	* "smart" HTTP protocol	22:16
	* partial checkout	22:16
	* detecting rename of directory as a whole	22:17
	→ dt_101 joined	22:17
ShadeHawk	Have I forgot anything?	22:17
	← ashleyw_ left	22:20
	← __iron left	22:20
	← loincloth left	22:23
	← knittl left	22:23
	← rittyan left	22:23
	← jspiros left	22:23
	→ loincloth joined	22:24
	← jrmuizel left	22:27
cYmen	after using mergetool and git add how do I finish the process?	22:29
	→ __iron joined	22:31
	→ melvinram joined	22:32
	← robomalo left	22:32
ShadeHawk	cYmen: git-commit I think	22:34
	→ bremner_ joined	22:34
	← d0k left	22:34
	→ AkumaStreak joined	22:36
	← blink left	22:36
	← __iron left	22:37
	→ __iron joined	22:38
nh2	I have just reverted all the commits I wanted to revert, but in my files there is still something like <<<<<<< HEAD:src/mumble/mumble_pch.hpp and after this line, somethin from the original state before reversion	22:40
	← dreiss left	22:40
thiago_home	edit and fix	22:41
nh2	thiago_home: you mean just per editor?	22:41
thiago_home	yes	22:42
nh2	OK	22:42
	← djgera left	22:42
ShadeHawk	or use git-mergetool (if you have some favorite GUI merge tools)	22:42
	→ bx2 joined	22:44
	← angerman left	22:44
	→ bremner__ joined	22:45
	← Octalot left	22:45
	← bremner__ left	22:46
	→ muhoo joined	22:47
	→ bremner__ joined	22:48
muhoo	i have two machines: A and B. both used to be tied to CVS but got out of sync. i converted machine A (and all its commit history from CVS) to git a while ago, and have been using git on it. i have a bunch of untracked changes on machine B that I need to merge in. BUT NOT ALL OF THEM! i need to cherry-pick them, but they are in the form of one massive patch.	22:50
	← ph^ left	22:50
muhoo	in other words, i've cloned machine A's git repo onto B, and branched to a "B" branch, and now a git diff shows a massive amount of changes, some i want to keep, and most that i don't.	22:51
	but they're not in the form of individual git commits since machine B has had no version control for a while.	22:51
	any advice on how to get out of this mess?	22:52
	→ cemerick joined	22:52
	← bremner_ left	22:52
	← cemerick left	22:52
muhoo	it should never happen again now that i have git :-)	22:53
offby1	muhoo: "git add -p", or perhaps "git add -i"	22:53
	← naeu left	22:54
	← __iron left	22:54
muhoo	nice!	22:55
	it'll let me pick hunk by hunk. awesome	22:55
	← jdahm left	22:55
muhoo	can i do that from a merge too?	22:55
pasky	i know what'sw the problem with the wiki	22:56
	macro syntax changed	22:56
	i overlooked that	22:56
	so i'll just mass convert {{...}} to <<...>>	22:56
	also i replaced underlay	22:56
	that fixed recentchanges	22:56
offby1	muhoo: I don't know what you mean by "can I do that from a merge too"	22:56
muhoo	offby1: ah. maybe: is there a -p or -i option to git merge?	22:56
	→ __iron joined	22:57
muhoo	so that i can pick my merges hunk by hunk as well?	22:57
offby1	gosh, I don't think so	22:57
gitte	pasky: gitfaq is still an issue...	22:58
	cpg\|lunch → cpg	22:58
pasky	gitte: i used future tense :)	22:58
muhoo	offby1: would it be sensible to commit all the current state of machine b to a branch first, then go back and use git add -i to cherry pick what i actually want to keep?	22:59
	or would that even be possible	22:59
offby1	it'd be possible, I think.	22:59
	But I don't think you'd use "git merge"	22:59
	→ ashleyw joined	22:59
offby1	instead I'd check out one branch, then plop the other branch's contents "on top of it", probably with git-archive, then do the "git add -p" thing again	23:00
muhoo	i'm really close to wrapping my brain around this, but not quite.	23:00
	← icwiener left	23:00
muhoo	thanks	23:00
	→ icwiener joined	23:00
muhoo	sorry for being dense, but i want to make sure i'm understanding this before hosing up my production system :-)	23:01
	i was going to first git clone machine a into a scratch directory, then branch to "b", rsync over machine b's files to it,	23:02
	→ dreiss joined	23:03
muhoo	then do {the part i'm not quite clear on yet, git add, git merge, etc},	23:03
	then copy the .git directory from machine B's scratch area, over to the production directory on machine B.	23:03
	← AkumaStreak left	23:04
muhoo	so it sounds like that {} section of the process would be to git add -p/i the changes, then commit it.	23:04
	← [AndrewR] left	23:04
muhoo	my question would then be: how would i save the current "untracked" state of machine B into a commit somewhere so i could reference it later?	23:06
	or a branch, mayb	23:06
	oh wait, i think i got it	23:06
	branch to "b-old", rsync over B's state into scratch directory, commit it,	23:07
	then branch back to master,	23:07
offby1	yep yep	23:07
muhoo	rsync over B again	23:07
ShadeHawk	pasky: thanks for work on it	23:07
offby1	no real need to rsync again	23:07
muhoo	then do the git add -p/i	23:07
offby1	you've already got those bits	23:07
	but otherwise, yes, that's what I was picturing	23:07
muhoo	offby1: don't they disappear once they're committed and i change branches?	23:07
ShadeHawk	still... incompatibile change if a a lightweight formatting language in a wiki ?!?	23:08
offby1	disappear?!	23:08
ShadeHawk	WTF they thought !?	23:08
offby1	of course ot	23:08
	not	23:08
ShadeHawk	see you later	23:08
	← ShadeHawk left	23:08
offby1	muhoo: now, if the rsync is cheap, go ahead and do it, since it's obviously clear	23:08
muhoo	offby1: yeah. i've noticed that. when i switch branches, uncommitted changes stay put, but any committed changes are now part of the branch they were committed to.	23:08
offby1	but just know that you don't really have to do it. You could run "git archive" to get the stuff back from the other branch	23:08
muhoo	offby1: oh, but would git add work then?	23:09
	→ fynn joined	23:09
muhoo	i'm having a brain-lock on the "git-like" way to do it, apparently.	23:09
	maybe i'll stick with rsync, but i'd like to understand how i'd do it with git-archive too	23:10
fynn	So we have a team using git, and an integration server. We also have a test suite that takes all-in-all a full hour to run.	23:10
offby1	muhoo: it's pretty easy -- "git-archive" just spews the content of your favorite commit to stdout, in the form of a tar file	23:11
	in your case, your favorite commit would be old-b	23:11
fynn	What's a good URL about getting a setup where each developer pushes his changes to the integration server, which runs the tests, and then merges the commits to a central server if the tests pass?	23:11
muhoo	offby1: ah, cool, so i could untar it into the working dir	23:11
offby1	sure	23:11
	→ bremner_ joined	23:11
muhoo	perfect	23:11
	this is a fucking awesome system	23:11
	it's like i've been hitting myself over the head for 10 years with CVS, and now the pain finally stopped.	23:12
offby1	well, not to put too fine a point on it, but: duh.	23:12
	← dt_101 left	23:12
muhoo	offby1: thanks, i'm excited to try this now.	23:14
	→ sping joined	23:14
	← ashleyw__ left	23:14
sping	any ideas where to find/contact repo.or.cz admins?	23:14
	← DavidKlein left	23:15
sping	the site is broken, webmaster@ doesn't work	23:15
gitte	sping: look through the irc log and you'll find out soon enough.	23:15
sping	gitte: irc log of what?	23:16
gitte	sping: IOW if you think he does not know, you're mistaken.	23:16
Fissure	he sent a mail to the list saying he's upgrading the machine to lenny	23:17
sping	Fissure: oh my :-)	23:17
	okay, thanks guys	23:17
	→ marvil07 joined	23:19
fynn	OK, I'll simplify the question:	23:20
	Joe pulls revision 1 from the central server, works on it, and creates revision 1'. Then he would like to commit it, but central is now on revision 2.	23:20
muhoo	offby1: -p is tha shizznit. awesome	23:20
	git add -p, just what i wanted.	23:20
fynn	is there an automatic way to merge the changeset 1 => 1' with revision 2?	23:20
offby1	muhoo: supposedly "-i" is even slicker, but I've never been able to figure it out	23:21
marvil07	muhoo, yeah, -p rocks	23:22
pasky	sping: what's broken?	23:23
sping	pasky: parts of the site deliver perl source code instead of executing it	23:23
pasky	ah	23:23
	indeed	23:23
muhoo	heh, and git-gui's staged-changes/unstaged changes section is even slicker still :-)!	23:25
	if only there were an emacs mode for this instead of using git add or git-gui...	23:25
	muhoo waits for it	23:25
	→ [AndrewR] joined	23:26
muhoo	hmm, git-gui kind of sucks, though, it doesn't let me pick hunk by hunk	23:26
	git add -p is much nicer. still poking around on emacswiki looking for a emacs mode for this	23:27
	→ idletask joined	23:27
	← bx2 left	23:27
fynn	OK, I'm pretty sure there's an answer to my question =)	23:28
sping	fynn: you want to merge the first two commits ever?	23:28
pasky	sping: fixed	23:29
fynn	sping: not exactly. basically, someone pulls from a server at revision 1, and makes a bunch of changes, that create branch 1'. now I want to merge 1' with master, which is at 2.	23:29
sping	pasky: sure? i tried reload on http://repo.or.cz/, same as before.	23:29
pasky	it looks fixed from here	23:30
sping	pasky: now it's working for me too	23:31
	← kumbayo left	23:31
	← bremner__ left	23:31
fynn	to put another way, we have revision X. from this revision, two child revisions were created: Y and Z. Z is the current HEAD. how do I merge Y and Z?	23:31
muhoo	now, i can push a branch too, and that'll create a branch on the remote side, right?	23:32
	→ jjuran joined	23:32
muhoo	i.e. git push fubarbranch remote fubarbranch ?	23:32
sping	fynn: i'm thinking of git fetch and git merge $remotename/$branchname, not sure if i understand you correctly	23:33
fynn	(i.e. how do I have the X => Y changes included in Z)	23:33
	← drizzd left	23:33
sping	muhoo: should be <git push remote fubarbranch>, branch name only once	23:33
fynn	sping: maybe my phrasing makes it seem too complicated, since reflecting on it, it seems like a classic merging scenario.	23:34
	I just need to merge two changesets, both based on the same revision.	23:35
sping	fynn: cherrypick also comes to my mind	23:35
fynn	right, but that's not automatic, right?	23:35
gitte	pasky: I still have the error with the empty table of contents.	23:35
sping	fynn: no, you pick cherries by hand :-)	23:35
	→ LiamH joined	23:35
	← sping left	23:36
	← aziz left	23:36
fynn	sping: yeah, I'll try to clarify: let's say rev X is an empty file foo.txt. rev Y adds the line 'bar', and rev Z adds 'baz'.	23:36
	what I'm asking is how I end up with foo.txt looking like 'bar\nbaz'	23:36
	I need it to be done automatically.	23:37
	→ blink joined	23:37
gitte	pasky: It seems that GitLink() is also not handled correctly.	23:38
muhoo	spuk-: thanks	23:38
	picking cherries is a young man's sport	23:39
pasky	please be patient:(	23:39
	perl -i -pe 's/(^\|[^{]){{($\|[^{])/$1<<$2/g; s/(^\|[^}])}}($\|[^}])/$1>>$2/g;' /revisions/	23:39
	theeere	23:39
gitte	pasky: basically, when I delete the "?action=content" part of the URLs, nothing is substituted, but the space also does not appear.	23:39
	pasky: another incompatibility in markup?	23:39
Pieter	pasky: yugh	23:39
muhoo	pasky: you have written a teco session	23:39
pasky	now wiki should work as usual	23:40
gitte	pasky: yes!!!	23:40
muhoo	http://en.wikipedia.org/wiki/Text_Editor_and_Corrector	23:40
	→ Radar joined	23:40
	← loincloth left	23:41
	← blink left	23:41
	← mbroeker left	23:42
	← kuadrosx left	23:43
	cpg → cpg\|biab	23:43
	→ TheGrudge_2 joined	23:44
	← fridim_ left	23:45
	→ djgera joined	23:45
	→ ashleyw_ joined	23:46
TheGrudge_2	Hi, I have a question about the bisect command. When I "land" on a commit that doesn't compile, can I go one commit forward or backward so that it compiles again? Using git bisect good / bad is no good idea, because I might miss the corrupting commit	23:46
pasky	/dev/md2 255G 234G 22G 92% /srv	23:46
	hmm	23:46
	← pano left	23:46
	← ashleyw left	23:46
uau	TheGrudge_2: see the manpage, that situation is explained there	23:46
FreakGuard	TheGrudge_2: hm. you don't get a binary, therefore bisect won't be a help?	23:47
	oh, sorry :)	23:47
gitte	pasky: how did _that_ happen?	23:47
TheGrudge_2	uau: Oh ok... I must have missed this	23:47
	uau: thanks!! :)	23:47
	→ schlort joined	23:47
pasky	gitte: i'm running a du -s now :)	23:47
	i will make use of the fact that no mirroring etc. jobs are running and do some grand mainenance on the repos	23:48
gitte	pasky: sounds good...	23:49
pasky	ideally, i would get rid of the raid5 there now, and instead do raid 0+1 with bigger disks	23:49
	i think the performance would be much better	23:49
gitte	pasky: did mugwump not have a mirror of repo.or.cz?	23:49
pasky	i think he did	23:50
	no clue about its status or even an url now, though	23:50
muhoo	so i'd use git reset to discard changes once i'm done, yes?	23:51
	← orafu left	23:51
TheGrudge_2	uau: so I guess it is git bisect skip... ok will try this	23:51
	← djgera left	23:54
	sshc → a	23:54
	→ Jinzhu joined	23:55
	← thiago_home left	23:55
	← Tuomas left	23:55
	← gitte left	23:55
	← TheGrudge_2 left	23:58
	→ ^Azerthoth joined	23:59
	← parasti left	23:59

Logs Search ←Prev date Next date→

Channels Documentation