IRCloggy #git 2009-06-23

Provider of IRC logs since 2005.

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

	2009-06-23
	← eseidelDesk left	00:01
	→ w\|shimmy joined	00:04
	→ scarabx joined	00:04
	← ctusar left	00:08
	→ jdavis_ joined	00:10
	← demio left	00:10
jdavis_	I removed a file from my repo. How do I see the last version that existed?	00:10
Pieter	git show --follow HEAD -- path/to/file or so	00:11
	← julm left	00:11
	→ aspotashev\|eeepc joined	00:12
	← solydzajs left	00:12
jdavis_	Pieter: I get no output from that.	00:12
	→ fynn joined	00:12
Pieter	hmm, it's something similar	00:13
	you can at least try git log --stat	00:13
fynn	What's the most reliable way to get all commits where a file called foo.py was updated/created/deleted?	00:13
Pieter	and then search for the filename	00:13
	:P	00:13
	← gberdyshev left	00:13
_Vi	fynn, May be "git log -M -- /path/to/file" or something like?	00:13
Ilari	fynn: 'git log -- foo.py'.	00:14
	→ gberdyshev joined	00:14
fynn	_Vi, Ilari: thanks, but foo.py can be at any path	00:14
	that's the trick part of the question ;)	00:14
jdavis_	Ok, I see the commit that removed it, what's the easiest way to get the actual file?	00:14
Pieter	git show sha:path/to/file	00:14
	→ fujin joined	00:14
jdavis_	Pieter: hmm... says "not in the working tree"	00:15
	← texel left	00:16
_Vi	fynn, You probably have either to traverse all commits.	00:16
	s/either//	00:16
	← ereslibre left	00:16
Pieter	jdavis_: that doesn't make sense :)	00:16
fynn	_Vi: possibly I need to use log -S	00:16
	→ ereslibre joined	00:16
	← wshimmy left	00:16
_Vi	fynn, It will show in which commit your string was added/removed. This string is not place for filename.	00:17
	fynn, Try "git log --summary" and grep for you filename.	00:19
	fynn, Or better "git log --stat".	00:19
fynn	_Vi: even better: --name-status	00:20
	grepping 'git log --name-status' has been my best result so far.	00:20
_Vi	fynn, --stat shows how much file was changed.	00:21
	→ archis joined	00:22
fynn	_Vi: yeah, but it's width-limited	00:22
	gberdyshev → gberdyshev_	00:22
	→ wsc joined	00:22
	→ danielnewman joined	00:22
jdavis_	It looks like I had to check it out, and then reset it to remove it from my WD.	00:23
	thanks.	00:23
fynn	so parts of the path can be cut off.	00:23
	← alezandro left	00:24
_Vi	fynn, --stat=0 ?	00:25
	← justin-george left	00:26
fynn	_Vi: thanks, but long paths still get cut off...	00:28
	← Pistahh left	00:32
	← whaley left	00:33
	← Alinermedia left	00:33
	→ kssddy joined	00:35
fynn	seems like --name-status is the canonical way of doing it.	00:36
	→ sgrimm joined	00:37
	cpg\|away → cpg	00:39
	← altrux left	00:42
	→ altrux joined	00:42
	→ pflanze joined	00:43
	→ joshsdc joined	00:43
pflanze	Hm, I've created a fork on repo.or.cz, but already closed the window with the instructions; what exactly is the username for git+ssh?	00:44
	→ bobmcw_ joined	00:44
	← cmarcelo left	00:44
pflanze	I've called the project picobit/chj, but obviously you can't give slashes in ssh user names, right?	00:44
_Vi	pflanze, It is the SSH username.	00:44
pflanze	and picobit or chj alone gives 'permission denied'	00:45
_Vi	pflanze, "bicobit/cgj" too?	00:46
pflanze	no, that gives ssh: picobit: Name or service not known	00:47
	→ sitaram joined	00:48
	→ jrmuizel_ joined	00:48
	← jrmuizel left	00:48
_Vi	pflanze, How have you cloned you repository?	00:48
pflanze	hm. Project admins might only be for the http interface, and ssh users created separately,	00:48
	and then added to the project by means of the admin interface.	00:49
	ha.	00:49
	I guess I got it (and I went through that 4 times already, that makes me smaaaart)	00:49
	→ Fullmoon joined	00:50
pflanze	pasky should make "don't forget to assign yourself as a user as well if you want to push! Everyone who wants to push must register himself as a user first." bold face.	00:51
	Even if it's only for me.	00:51
	→ tonetheman joined	00:52
	→ agile joined	00:52
	← hummingkid left	00:52
	← tonetheman left	00:54
	→ ben_h joined	00:56
	→ justin-george joined	01:01
	← ben_h left	01:02
	→ flazz joined	01:04
sitaram	hey guys; when I do "git log -M --name-status" I see renamed files prefixed with "R100". What does the 100 mean?	01:05
	aah, percentage similarity!	01:06
	(I think)	01:06
	yup	01:07
	sitaram talks to himself; this is a sign if advanced intelligence. (He hopes)	01:07
sitaram	of*	01:08
Ian00	how can I see how many total commits i have in my repository?	01:08
sitaram	git log --oneline --all\|wc -l maybe	01:08
	← WALoeIII left	01:09
	← bobmcw_ left	01:09
Ilari	IIRC 100% is special (only for rename without edit).	01:11
smoofra	damnit. stupid email client	01:12
sitaram	possibly. I wasnt even sure what it meant at first, so knowing that it is % is a step :)	01:12
_Vi	smoofra, line wrapping?	01:12
smoofra	it included all the headers from git-format-patch in the body	01:13
	← sitaram left	01:16
	→ julm joined	01:18
	← bc left	01:21
	→ dgfitch joined	01:21
	→ jerbear joined	01:21
	→ bc joined	01:22
	← pflanze left	01:22
	← jrmuizel_ left	01:24
	→ solydzajs joined	01:28
	← ereslibre left	01:29
	← jayallen left	01:30
	→ jayallen joined	01:31
	→ bcardarella joined	01:32
	→ tanuki joined	01:32
	← maxb left	01:32
tanuki	Is there a generally accepted way to deploy Rails sites using git?	01:32
	I'm almost where I want to be, but I still have to log into the deployment server and force reset	01:32
_Vi	tanuki, Setup a hook that will do it automatically.	01:33
	← panpainter left	01:33
diegoviola	why not just use apache and mod_rails?	01:34
tanuki	diegoviola: ?	01:34
	← danielnewman left	01:34
diegoviola	tanuki: what do you mean deploying rails with git?	01:35
tanuki	diegoviola: using git-push to copy files from my workstation to my webhost	01:35
diegoviola	oh	01:35
tanuki	basically, the working copy on my repository on the server is the live site	01:35
smoofra	tanuki: i have some scripts you might like	01:35
	→ maxb joined	01:36
	→ FunkeeMonk joined	01:36
diegoviola	tanuki: i guess you could set up apache with mod_rails... then have a repo on your web root directory, and pull/push from that	01:36
tanuki	diegoviola: Right, and that's what I'm doing.	01:36
	diegoviola: The problem is, when I push, I still have to log into the web host and reset in order for the changes to show up	01:36
smoofra	tanuki: clone http://elder-gods.org/~larry/repos/bin and look in git-pushable-hooks	01:37
diegoviola	with mod_rack (or mod_rails) you wont need to restart the web server when you pull/push	01:37
	you will see your changes on the fly	01:37
	→ cedricv joined	01:37
tanuki	diegoviola: the web server isn't the problem	01:37
	diegoviola: the problem is that git doesn't auto-update the working copy on the other end	01:37
	smoofra: Thanks, I'll check that out	01:37
smoofra	tanuki: you'll have to transplant them out of my motley collection of scripts, but i think they bascially do what you want	01:38
tanuki	I'm confused: does the pre-receive hook on the pushing repo run on the receiving repo, or does the receiving repo use its own copy?	01:39
smoofra	both hooks go into the receiving repo	01:40
tanuki	s/pre/post/	01:40
smoofra	they will prevent you from pushing if the receiving repo has any uncommited changes	01:40
	and if you are able to push, they'll update the index and the working tree after you push	01:40
	← bcardarella left	01:41
tomoj	tanuki: capistrano :)	01:42
	← gberdyshev_ left	01:43
	sebleier\|away → sebleier	01:43
	→ georgyberdyshev joined	01:43
	georgyberdyshev → gberdyshev_	01:43
	← archis left	01:44
tomoj	e.g. say you push some commits with migrations. now your live site is broken until you ssh in and run the migrations. with capistrano you can just do 'cap deploy:migrations' and it will fetch the latest commits and run the migrations. though with capistrano deploying new code takes two steps, since you have to cap deploy after every git push	01:44
smoofra	oh thats pretty cool	01:45
	→ djanowski joined	01:45
tanuki	tomoj: ... point well made.	01:46
tomoj	and it's crazy extensible so you can pretty much do anything with it	01:46
djanowski	hey all. i just committed something, but i'd like that commit to go to a new branch. any way to amend that?	01:46
	← altrux left	01:46
Ilari	djanowski: 'git branch newbranchname', 'git reset --merge HEAD^'?	01:46
smoofra	djanowski: git checkout -b newbranch	01:46
	djanowski: git checkout oldbranch	01:47
	djanowski: git reset --hard HEAD^	01:47
	← tbf left	01:47
djanowski	Ilari: awesome, thanks :)	01:48
Ilari	djanowski: I assumed you have git version new enough for 'reset --merge'....	01:49
tomoj	Ilari: why --merge in that case?	01:49
djanowski	Ilari: yeah i'm pretty up to date, thanks	01:49
smoofra	yea i don't think --merge does what he wants	01:50
Ilari	tomoj: Likely gets the working tree change effect without the destructivity of --hard...	01:50
tomoj	what will --hard destroy in a brand new branch?	01:50
	oh, anything that got carried over,huh	01:51
Ilari	tomoj: Uncommitted changes?	01:51
	→ ericindc joined	01:51
tomoj	I had never seen --merge before, that's pretty cool	01:51
smoofra	"git branch newbranchnaem" makes the new branch, "git reset --hard HEAD^" puts the current branch back to where it was before you commit	01:51
	← ericindc left	01:51
	← joshsdc left	01:52
	→ jrmuizel joined	01:53
fynn	tomoj: the funny thing is, remember the problem we solved for that guy yesterday?	01:53
	the "merge just 3 commits" problem?	01:53
tomoj	fynn: yeah, with rebase reordering -> fast-forward?	01:54
fynn	tomoj: yeah exactly	01:54
	when I thought about it later, I said "there must be something like --merge for reset" :D	01:54
tomoj	but wouldn't that drag along all the other stuff he didn't want?	01:55
fynn	yeah, in this case possibly he actually prefers --hard to --merge	01:55
Ilari	AFAIK, if you don't have any uncommitted changes, the end result of reset --hard and reset --merge is the same.	01:56
fynn	yup, I think so too.	01:56
	→ ben_h joined	01:56
tomoj	but doesn't --merge leave you without the commit you reset to? or am I reading it wrong?	01:57
	→ johndoigiii joined	01:57
	→ ianmcorvidae\|alt joined	01:57
johndoigiii	aghhh, I checked out into a branch called "--track" how in the hell can I delete the branch without git-branch thinking its a flag and not <branchname>?	01:58
tomoj	oh, I guess it gives you the commit but the difference is it leaves your uncommitted changes on top of that..	01:58
	← shaunalynn left	01:58
fynn	johndoigiii: git branch -D -- --track	01:58
	→ spaceonline joined	01:58
johndoigiii	awesome, thanks!	01:58
fynn	welcome.	01:58
johndoigiii	is that a common question?	01:59
fynn	no ;)	01:59
	→ ben_h_ joined	01:59
johndoigiii	ahh good then :) I feel less dumb now	01:59
Ilari	Wonder if those issues with illegal branch names have been resolved (like inability to rename them to something legal).	01:59
	← kukks left	01:59
	jayallen → jayallen-afk	02:00
tomoj	hehe.. I just made a branch named (no branch)	02:01
Ilari	tomoj: Spaces aren't legal in refnames. Unless that is that infamous U+00A0.	02:02
johndoigiii	ha what happens if you name it "" then try to `-D `	02:02
tomoj	illegal indeed, but if you go in and manually create the ref, it appears to work	02:02
	that would be a fun way to confuse someone	02:03
Ilari	johndoigiii: '*' is not valid branch names. You would need to manually create it.	02:03
johndoigiii	ahh okay	02:03
	← djanowski left	02:03
tomoj	they think they have a detached head and then switch out, but it mysteriously stays in the branch list	02:03
Ilari	tomoj: Most probably breaks stuff as well.	02:03
	johndoigiii: And not even update-ref can create it. Manually means editing the repository files directly.	02:04
	← ben_h left	02:05
johndoigiii	ohh okay	02:05
Ilari	tomoj: That refname can really confuse parsing code if some place uses space-delimited refnames.	02:05
	← Chillance left	02:05
johndoigiii	Ilari: awesome information to know	02:05
tomoj	I got rid of it :)	02:05
johndoigiii	I should start hanging out in here more often :)	02:06
Ilari	"(no branch)" (with that U+00A0) IS valid branch name (checked with check-ref-format).	02:07
fynn	Ilari: have you worked with Bazaar or Mercurial at all?	02:08
Ilari	fynn: No.	02:08
	→ dasil003 joined	02:08
fynn	Just wondering how they compare to git. Some of the stuff I've seen lately is so fancy.	02:08
	← aziz left	02:08
tomoj	I looked at them a couple days ago, and read that they don't have indexes, and stopped reading about them	02:08
	← Bass2 left	02:09
	← julm left	02:09
fynn	hehe. Mercurial does have an index, although it's some extension or something.	02:10
dasil003	I have an interesting scenario... Suppose you have a master branch representing mainline development, and a shared branch, we'll call it `staging` that represents a collection of topic branches from various developers	02:10
	The purpose of staging is for deploying topic branches to a staging server for testing	02:10
	Sometimes we deploy topic branches directly	02:10
	but sometimes multiple developers need to stage things at the same time	02:10
	in which case we merge down to 'staging' first	02:10
	→ metrix joined	02:11
	ben_h_ → ben_h	02:11
dasil003	we have a policy that staging never gets merged back up to anything, because no one has the whole story about all the topics contained there... each user is responsible for their own topic	02:11
	the question is...	02:11
	when the topics are complete, or we have a conflict and we need to reset the staging branch back to master state	02:11
	what is the best way to do that?	02:11
Ilari	Experimenting with "(no branch)", current branch display says (unknown)...	02:11
tomoj	Ilari: with spaces? it worked for me on 1.6.3.3	02:12
mugwump	dasil003, your question seems strange because the answer seems so simple, yet the question shows a lot of knowledge about git	02:12
	cpg → cpg\|biab	02:12
fynn	dasil003: reset...?	02:12
dasil003	heh	02:12
mugwump	I mean, superficially, git reset --hard origin/master / git push -f etc	02:12
fynn	yeah, it's funny.	02:12
mugwump	but is that all you're asking?	02:13
Ilari	Yeah, oops.	02:13
fynn	mugwump: maybe he's an easy guy to please?	02:13
dasil003	well, what happens to the downstream devs when I push a hard reset like that?	02:13
	← ianmcorvidae left	02:13
mugwump	they need to be aware of it	02:13
	ie, they need to not try to merge 'staging' into their topics or master	02:13
	which you already said :)	02:13
fynn	dasil003: and there's not much to do about it either way; either you reset staging, or you don't	02:13
mugwump	'pu	02:13
	' is developed like this	02:14
dasil003	well, when they `checkout staging` and then `git pull` what happens?	02:14
fynn	pu?	02:14
Ilari	Heh, trying to push it: -> [remote failure]	02:14
mugwump	pu = proposed updates, a branch of git.git	02:14
Ilari	(remote failed to report status)	02:14
mugwump	dasil003: if they have an old staging branch, they need to delete it first	02:14
	→ ctusar joined	02:15
mugwump	eg git branch -D staging; git checkout -b staging origin/staging	02:15
	← spaceonline left	02:15
mugwump	otherwise madness will ensue as you're trying to merge the latest random combination of branches with the last one	02:15
dasil003	so presumably there will be some history error "unknown ancestor" or something	02:15
	← solydzajs left	02:15
mugwump	no, it will just go ahead and merge it	02:15
	which will either be a big confusing merge conflict ... or silently leave you with EVIL history :->	02:16
dasil003	that's what I was worried about	02:16
Ilari	Invalid branchnames can not be fetched from, and can't be pushed without having symref point to them and remapping the name.	02:17
dasil003	that's why I was thinking what if I did a diff and just applied a patch that brought staging back in sync with master?	02:17
mugwump	not a good plan really, resetting is much better	02:17
	so here's what they should do:	02:18
Ilari	Or actually, having symref inside refs should allow fetching them...	02:18
mugwump	git checkout staging; git fetch; git reset --hard origin/master; git merge mytopic1; git merge mytopic2	02:18
	so that staging always gets re-written; it's what you described as how you work	02:19
	← sekimura_ left	02:19
mugwump	does that make sense?	02:19
dasil003	yeah, we're sort of feeling it out as we go because we just added 3 devs and the history is getting kind of unmaintainable with too many "mainline" branches	02:19
Ilari	tomoj: For causing confusion, "(no branch)" would be even better than the invalid "(no branch)".	02:19
tomoj	Ilari: the first one is with the unicode character?	02:20
Ilari	tomoj: Yes.	02:20
dasil003	what you said makes sense	02:20
metrix	We are using Subversion for 3 development teams at work and use branches for development and Trunk for our stable release where Team leaders were the only ones with commit access to trunk. This has was working great up to the point we realised SVN branch/merging didnt work if you tried to branch a branch. How would I do something similar with GIT?	02:20
Ilari	tomoj: Its not even hard to type. ALT+SPACE here.	02:20
dasil003	I guess I was secretly hoping that there would be a way to at least consistently raise an error rather than ending up with messed up history	02:21
	that scares me a little	02:21
	→ cytrinox_ joined	02:21
dasil003	wrt debugging issues remotely for junior devs ;)	02:21
	we're spread all over the world	02:21
	hard enough to sync up temporally	02:21
	but as long as everyone is aware of the issue it should be a problem	02:21
	← _grepper left	02:22
Ilari	metrix: If whatever authentication you use exports user identity (standard multi-user setup does, shared account and gitosis doesn't without some hacking), hooks can be used to veto branch updates on branch-by-branch basis.	02:22
	metrix: Gitosis requires one-line change for that.	02:23
metrix	I see	02:24
	I will read more about it	02:24
	Thank you!	02:24
	→ cream_o_wheat joined	02:24
	← cytrinox left	02:24
mugwump	what didn't work about branching branches in svn, out of curiosity?	02:25
	→ Sho_ joined	02:25
	← xckpd7 left	02:25
	← dreiss left	02:25
	cpg\|biab → cpg	02:26
Ilari	metrix: And since repos aren't that expensive, there could be seperate repo with just the master branch, other branches being in seperate repo. Then one can control access on repo, which is easier.	02:26
metrix	We would create a branch on a big project, test half way through and wait for paperwork to get done/QA to finish.. In the meantime we would create a branch off of that branch and continue developing. We would merge the first branch with no problems but when the second branch gets merged there would be conflicts with all the changes from the first branch. if you created files etc.	02:27
	→ bcardarella joined	02:28
	← aspotashev\|eeepc left	02:28
metrix	Maybe I was using SVN in the wrong way... But it seems it just seems that merging branches like that should just work.	02:29
mugwump	merging the second branch into trunk would fail?	02:30
	that actually surprises me, because I would have expected that to work, too	02:30
smoofra	merging anything into anything fails in svn	02:30
	→ drizzd_ joined	02:34
	← dantrell left	02:34
Ilari	metrix: In git, it would even be possible to tag/branch the branch to merge to mainline and keep the name of development branch. Only affects commit message in merge.	02:34
metrix	cool	02:35
	This is what we should have used in the first place	02:35
	← kennethlove left	02:35
mugwump	it's a hurdle to learn it all really. be sure to read the sections of the user manual on the underlying design	02:36
metrix	will do	02:36
mugwump	also there's a good tutorial called 'understanding git conceptually' which tries to go through it	02:36
	once you see how it all fits together and can poke at it with 'git cat-file' etc, you go through a phase of wondering how something simple could possibly work at all, then understanding usually follows after that :)	02:37
Ilari	metrix: The problem with using commit IDs directly is that while erge would happily merge one, one can't fetch/pull by commit ID.	02:37
	*merge	02:37
	← jrmuizel left	02:37
	← stephenjudkins left	02:38
	→ hax0r1337 joined	02:39
metrix	what would you suggest to use? the two requirements I have: 1. developers can somehow update each others branches and 2. some sort of restriction so only team leaders can merge to trunk	02:40
	→ boto_ joined	02:40
mugwump	My feeling is that if you're used to using access controls for trunk, carry that over to using git	02:41
	← qrush left	02:41
Ilari	metrix: Two repos, one for developer branches and one for team leaders. Developers and team leaders can read both and write the developer repo. Team leaders can also write to repo with the master branch.	02:41
mugwump	Sure, most people and systems will access control the entire repository as a single unit, but you can do whatever you like in an update hook	02:41
metrix	cool	02:42
	Ilari I will setup a test system and try it that way	02:42
Ilari	metrix: Except that two-repo setup can't be done without an ACL system (gitosis or POSIX ACLs).	02:42
	← bcardarella left	02:42
Ilari	metrix: At least if you want to restict readabilty of that repo with master.	02:43
	metrix: Or it can if team leaders use shared account...	02:44
metrix	I see	02:45
Ilari	metrix: Shared account doesn't mean "shared password". You can authorize multiple keypairs to authenticate to same user account.	02:45
	(actually, it would be possible to alias user accounts, and each alias has independent password).	02:46
	← jerbear left	02:46
smoofra	why not just make a unix group for it? or use filesystem acls?	02:46
	← FlyingSquirrel31 left	02:48
metrix	that's not a bad idea	02:48
	→ jony joined	02:48
metrix	it looks like I have quite a bit of reading to do	02:49
	Thank you for all the help	02:49
	→ eno_ joined	02:49
	→ bcardarella joined	02:50
Ilari	metrix: git has policy of dumping authentication problems elsewhere (filesystem permissions, sshd, etc...)	02:50
	→ amerine_ joined	02:50
	→ abdelhalim joined	02:50
Ilari	metrix: Also authorization problems.	02:51
	→ tjafk2 joined	02:53
	← drizzd left	02:53
	← tanuki left	02:53
	→ julm joined	02:54
	→ jrmuizel joined	02:55
	→ offby1` joined	02:56
	← offby1 left	02:56
	← boto left	02:56
	offby1` → offby1	02:57
	← jrmuizel left	02:58
hax0r1337	does git reset --hard leaves any tmp or unnecessary files hidden in .git dirs that I can safely delete?	02:58
	→ sitaram joined	02:59
mugwump	hax0r1337: go read man git-prune	02:59
Gitbot	mugwump: the git-prune manpage can be found at http://git.or.cz/man/git-prune	02:59
mugwump	hax0r1337: also, man git-reflog	02:59
Gitbot	mugwump: the git-reflog manpage can be found at http://git.or.cz/man/git-reflog	02:59
mugwump	Gitbot: tell hax0r1337, not me!	02:59
	and git repack -A	03:00
sitaram	mugwump: gitbot has a problem when the nick has numbers in it	03:00
mugwump	assuming, of course, you accidentally committed content you want to purge completely	03:00
hax0r1337	mugwump: thanks chief	03:00
	→ jrmuizel joined	03:01
hax0r1337	mugwump: I merged something that resulted in lots of errors/conflicts	03:01
mugwump	oh, well, in that case, just leave it - git reset --hard clears that state	03:01
	← eno left	03:01
	← mw left	03:02
mugwump	the only 'merge in progress' state is in the index (see man git-read-tree) or .git/MERGE_HEAD	03:02
Gitbot	mugwump: the git-read-tree manpage can be found at http://git.or.cz/man/git-read-tree	03:02
hax0r1337	mugwump: good to know, git wouldn't let me delete that branch or switch back to master, so I had to reset it	03:02
	← hax0r1337 left	03:06
	← tjafk1 left	03:09
	→ bugfux joined	03:10
	← abdelhalim left	03:10
	← porcnick left	03:11
	← jony left	03:11
	→ dreiss joined	03:11
	← Casan left	03:13
	→ jony joined	03:13
	← bugfux left	03:13
	jayallen-afk → jayallen	03:15
	← ben_h left	03:16
	→ MetaCosm joined	03:16
	→ abdelhalim joined	03:18
	← amerine_ left	03:18
	→ joshsdc joined	03:19
	← Fullmoon left	03:23
	← dodo_the_last left	03:24
	→ mugwump_ joined	03:24
	eno_ → eno	03:26
	→ makeworkeasy joined	03:26
	← jony left	03:26
	→ mw joined	03:26
	← jrmuizel left	03:28
	← mugwump left	03:31
	← mw left	03:31
	← madewokherd left	03:32
	→ mw joined	03:33
	→ MetaCosm_ joined	03:35
	→ joshsdc_ joined	03:36
	← joshsdc left	03:37
	joshsdc_ → joshsdc	03:37
	→ reprore joined	03:39
	→ mturner joined	03:40
	← MetaCosm left	03:41
	← mturner left	03:42
	→ Ori_B joined	03:43
Ori_B	Hi.	03:43
	I'm having a bit of trouble with git-daemon	03:43
	← joshsdc left	03:44
	→ joshsdc joined	03:44
Ilari	Ori_B: What kind of trouble? Anything in server logs?	03:45
Ori_B	well, I'm trying manually to start it from the command line, and;	03:45
	fatal: unable to allocate any listen sockets on host 208.78.101.172 port 9418	03:45
Ilari	Ori_B: Likely you already have one running.	03:46
smoofra	Ori_B: to see what's got your port: netstat --ip -lpn	03:46
Ori_B	...oh. it looks like git-daemon-run is competing with inetd... ugh.	03:47
	no wonder.	03:47
	← ctusar left	03:48
	← julm left	03:48
	← john3909 left	03:49
	→ DrAk0 joined	03:50
	→ ghio joined	03:51
Ori_B	yeah, that was the issue.	03:51
	a broken inetd.conf (probably automatically generated by debian...) that conflicted with my setup.	03:51
	← rettub_ left	03:51
	sebleier → sebleier\|away	03:53
	← mw left	03:53
Fissure	wow, repo.or.cz is really really slow	03:53
	6 k/s	03:54
	→ rettub_ joined	03:54
	← reprore left	03:56
	→ ccaputo joined	03:58
	← ccaputo left	03:58
	→ ccaputo joined	03:58
	← ccaputo left	03:58
	→ sea-gull joined	03:58
	→ john3909 joined	03:59
	→ ccaputo joined	03:59
	← ccaputo left	03:59
	← DrNick left	03:59
	→ julm joined	04:01
	→ hax0r1337 joined	04:02
	→ DrNick joined	04:02
	→ candybar joined	04:03
	→ AkumaStreak_ joined	04:04
	→ ankit9 joined	04:04
	→ mithro joined	04:06
docgnome	is there a way to list what repos git daemon is serving?	04:07
wereHamster	Fissure: does your tracepath to repo.or.cz go through level3?	04:08
offby1	docgnome: if you're on the same box it's running on, I'd poke around in /proc	04:08
mugwump_	/proc?	04:09
	mugwump_ → mugwump	04:09
mugwump	surely, just check the commandline git-daemon was started with, or /etc/inetd	04:10
	.conf	04:10
	← ankit9 left	04:12
	← memiux left	04:12
	→ ankit9 joined	04:12
offby1	that too.	04:12
	In fact, checking the commandline is effectively the same as poking around in /proc (after all, that's all that "ps" does :-)	04:12
	← julm left	04:18
	← ankit9 left	04:18
	→ tazle__ joined	04:18
	→ brizly joined	04:18
	← AkumaStreak left	04:18
	→ MetaCosm joined	04:19
	→ mturner joined	04:19
	→ klmann_ joined	04:19
	← diegoviola left	04:19
	← kadoban left	04:19
	← gaveen left	04:20
	→ kadoban joined	04:20
	→ Tuomas joined	04:20
abdelhalim	hello all	04:20
	→ gaveen joined	04:20
abdelhalim	I have a question regarding git over HTTP	04:21
	anyone using that ??	04:21
smoofra	sure why not	04:21
	makies_bot → makies	04:21
smoofra	it's a very easy way to make a public repo	04:21
	→ Derander_ joined	04:21
	→ MadCoder joined	04:21
abdelhalim	I saw that jgit implements also git with Amazon s3	04:22
	cannot this be done using the regular git over HTTP ?	04:22
	what is the difference from git perspective ?	04:22
	← joshsdc left	04:23
smoofra	i don't think s3 lets you create the url hierarchies that git needs	04:23
	← MetaCosm_ left	04:23
abdelhalim	I'm not expert, I'm trying to understand both git and S3, but isn't the bucket is just a container that may contain hierarchy ?	04:24
	→ reprore joined	04:25
smoofra	hrm looks like you're right	04:26
	← reprore left	04:26
	← maxb left	04:26
smoofra	s3 lets you name a file foo/bar/baz	04:26
abdelhalim	and I see that s3 allows you to use HTTP POST/PUT requests	04:27
Ilari	abdelhalim: Regular git HTTP push requires WebDAV.	04:27
smoofra	yea you probably could just put a git repo up on s3 and people would be able to read it	04:27
	→ maxb joined	04:28
abdelhalim	Ok, that might be a start	04:28
	← jim777 left	04:28
abdelhalim	so the difference is that git HTTP requires WebDAV	04:28
	and S3 doesn't support that ?	04:28
smoofra	Ilari: regualr git HTTP push exists?	04:29
	that's pretty nifty	04:29
	Derander_ → Derander	04:30
	→ resmo_ joined	04:31
	← mturner left	04:31
Ilari	smoofra: It exists, but nope, it isn't nifty.	04:33
smoofra	oh?	04:33
	← yann left	04:34
	→ yann joined	04:34
	← paltman94 left	04:37
	→ ben_h joined	04:37
smoofra	don't most people just ssh for repos they want to push to? i've never bothered setting up anything else.	04:38
	→ axius joined	04:38
	→ Flathead_ joined	04:39
	→ jceb joined	04:39
	Flathead_ → Flathead	04:41
	← candybar left	04:42
	← dreiss left	04:44
	→ candybar joined	04:44
mugwump	git push over webdav is deferring all smarts to the client	04:44
	→ sunoano joined	04:45
	← candybar left	04:46
	← marc_kdab left	04:46
	→ candybar joined	04:47
	← resmo left	04:47
	← MetaCosm left	04:51
	← ben_h left	04:54
	→ ben_h joined	04:55
	→ toxx__ joined	04:55
	→ memiux joined	04:57
	→ stephenjudkins joined	04:57
	← hax0r1337 left	04:58
	→ julm joined	04:58
smoofra	ah i see	04:58
	you can't put any hooks on it	04:58
	because webdav has no way to execute them	04:59
	← jonmasters left	05:01
	→ DavidKlein joined	05:03
abdelhalim	isn't this the same as S3 ?	05:04
	→ bugfux joined	05:06
	→ robinr joined	05:07
	← boto_ left	05:09
	→ Voight-Kampff joined	05:09
	← memiux left	05:09
	→ girishr joined	05:11
	← toxx_ left	05:11
	→ toxx_ joined	05:12
	← toxx___ left	05:12
	← axius left	05:14
	→ Cromulent joined	05:14
	→ jonmasters joined	05:14
	← julm left	05:16
	→ axius joined	05:17
	← justin-george left	05:21
	← Derander left	05:21
	← bcardarella left	05:22
	← cloudhead left	05:22
	← jceb left	05:24
	← candybar left	05:26
	→ roop joined	05:26
	← flippo left	05:27
	← fujin left	05:27
	→ candybar joined	05:29
	← candybar left	05:30
	← axius left	05:31
	→ cloudhead joined	05:31
	← flazz left	05:35
	→ vshriram joined	05:39
vshriram	when i do a 'git branch -a' i see lots of origin/something listing. what are these?	05:39
	and why do we need those?	05:40
mugwump	those are 'remote tracking branches'	05:40
vshriram	what exactly is a remote tracking branch?	05:40
	and whats its purpose	05:40
mugwump	they are branches which track the state of the remote repository	05:40
	→ MetaCosm joined	05:41
mugwump	normally, the remote refs/heads/* are copied to refs/remotes/origin/*	05:41
mstormo	vshriram: Those are the branches which are updated when you do 'git fetch' and also when you do 'git pull', and then their content is merged into your local branch	05:41
	(basically speaking)	05:41
vshriram	okay - i have access to a git tree. I am sure that all branches reside in that git tree and yet i see those?	05:41
mstormo	well, you cloned that repo from somewhere, right?	05:42
mugwump	"all branches reside in that git tree" suggests to me you're expecting in-tree branching a la Perforce/SVN	05:42
	← robinr left	05:42
	→ scientes joined	05:43
vshriram	yeah from the git server in our company	05:44
	← thiago_home left	05:46
	→ marc_m joined	05:47
	→ WinterWeaver joined	05:49
	→ psankar joined	05:50
	← vshriram left	05:50
	→ vshriram joined	05:52
	→ thiago joined	05:52
	→ Derander joined	05:55
sitaram	vshriram: since your local branches can get out of step from the "origin" due to your local commits, git keeps "track" (hence "tracking branches") of where the origin currently is. That is how, when you later do "git status" it can tell you stuff like "your branch is ahead of origin by 4 commits" or whatever	05:56
vshriram	is that the only reason why we have origin/xxx branches because they know where the origin is	05:58
	i thought when somebody said remote branch - it is somewhat connected to a different tree someone on the internet	05:59
sitaram	it just says "on this remote, called origin (whose URL you can see by running git remote show origin), the branch called XXX, is at this commit"	05:59
	→ rpag joined	06:00
	← ilteris__ left	06:02
	← krawek_ left	06:02
fynn	hm	06:02
	I made a stupid :(	06:02
	$ git push origin origin/master	06:02
	* [new branch] origin/master -> origin/master	06:03
vshriram	sitaram: are you saying that - origin/xx branches are needed just to keep track of where the origin of the present branch is?	06:03
fynn	how do I fix? :(	06:03
sitaram	vshriram: yes	06:03
fynn	I'm not sure what I even did there	06:03
sitaram	fynn: git push origin :origin/master	06:03
	fynn: but first "git ls-remote origin"	06:03
	to make sure	06:03
fynn	sitaram: hm, what am I looking for?	06:04
vshriram	what is the advantage of this - you always know where you started? say - last commit id of the branch when you created it	06:04
fynn	sitaram: I get a bnuch of hashes.	06:04
sitaram	fynn: you're looking for (a) a branch called master that is in good shape and has the right SHA for your correct origin/master,	06:04
vshriram	you tag it as baseline and start adding your stuff. isnt that the case	06:04
sitaram	fynn: (b) a new one called origin/master that has some SHA that you don't care about too much	06:05
	fynn: and if you see both, "kill the spare" :-)	06:05
	sitaram wonders if there are any Harry Potter fans here...	06:05
sitaram	vshriram: remote can move ahead due to other people's commits. You update your "notion" of where that is by "fetch" or "remote update"	06:06
fynn	sitaram: OK, there are 3 branches with the ID of my local HEAD:	06:06
	HEAD, refs/heads/master, and refs/remotes/origin/master	06:06
sitaram	then the push command I gave will work and undo what you did	06:07
fynn	could you perchance explain how it works :>	06:07
	← cloudhead left	06:07
sitaram	fynn: sure, the push command syntax is "remotename localbranchname:remotebranchname". Now this is a little obscure but by leaving out the localbarnchname, you're saying "delete the remotebranchname"	06:08
	s/barnch/branch	06:08
	← brizly left	06:08
vshriram	if remote has moved ahead (in the normal case - non origin branches) wont a git pull work?	06:08
fynn	sitaram: aha, this is a bit dangerousy	06:08
vshriram	probably all i need to do is to push the changes to the gitserver and update it by cloning it again	06:09
sitaram	vshriram: a git pull is a fetch+merge, no more and no less. Sometimes you DONT want a merge happening without the chance to look at what is coming in...	06:09
fynn	sitaram: thanks though, that definitely fixes the issue.	06:09
sitaram	vshriram: clone is only needed once. Why would you need to clone again?	06:09
	vshriram: push and pull as often as you please to keep things in sync, but if other people are also doing the same, you end up with more merge commits than real commits :-)	06:10
fynn	sitaram: one last question about push: is there a shortcut to push from whatever local branch I'm currently on?	06:10
sitaram	fynn: the latest git versions have a config var called "push.default" you may want to look at	06:10
fynn	vshriram: an alternative, if nobody is pulling from you, is to rebase	06:10
sitaram	fynn: otherwise, "git push origin HEAD" usually works	06:10
fynn	sitaram: yeah, I already have that configed actually :>	06:11
	what do you mean "usually"?	06:11
sitaram	fynn: just that I haven't used it too often. Don't let me scare you :-)	06:11
	I should have said "worked whenever I tried it" ;-) but I normally am explicit about my pushes	06:12
fynn	I couldn't possibly be more scared than what I am right now ;)	06:12
	btw, isn't there a more explicit option to delete a remote branch?	06:12
sitaram	fynn: not as far as I know	06:12
fynn	sitaram: weird... thanks!	06:12
sitaram	fynn: short of logging into the remote machine, cd to the repo, and "git branch -d ..."	06:12
	← Predominant left	06:13
sitaram	fynn: that weird syntax is useful; kids don't know how to delete a remote branch :-) Security through obscurity!	06:13
	← stephenjudkins left	06:13
fynn	sitaram: heheh	06:14
sitaram	fynn: but I believe there has been some debate about it, not sure what resulted. I haven't been able to spend as much time on the list lately as I would like to have... :-(	06:15
	← scarabx left	06:15
vshriram	okay, i will slowly understanding this - one unrelated question - if i have a git tree on my server and one out on the web totally different, i can merge the differences between the remote git tree (on the web) and the git tree on my server	06:15
	→ candybar joined	06:15
sitaram	vshriram: what do you mean totally different? no common commits at all?	06:15
vshriram	yeah common commits exist, after that initial point - that remote tree is diverging	06:16
	that is commits are getting added to the remote tree	06:17
sitaram	well either way you can merge. though having a common ancestor helps :-)	06:17
	← bugfux left	06:17
sitaram	if they have diverged too far, don't just "pull". Do a fetch, then a test merge into a temp branch that you can throw away if needed	06:17
vshriram	what do you mean by ancestor. suppose i do the following lets talk in terms of examples	06:18
sitaram	fynn: re "scared" -- 2 golden rules in git help me. (1) never fiddle inside ".git". (2) never attempt any push/pull/fetch/merge/rebase/reset on a dirty work tree	06:18
	→ hax0r1337 joined	06:19
vshriram	i have a linux-2.6.29 tree. I also have a android-2.6.29 tree (in google server). Now i checkout android-2.6.29 tree (from google server) and extract all the patches from it using git format-patch and apply them on my linux-2.6.29	06:19
	now google android 2.6.29 (google server) has changed. cant i merge it at a later date	06:20
sitaram	no	06:20
	the SHAs are all different, and git wont know which is which	06:21
	→ solofight joined	06:21
vshriram	so what do you suggest?	06:21
sitaram	why cant you merge the android-2.6.29 tree instead of extracting and applying?	06:21
solofight	when i do a git pull it says You are in the middle of a conflicted merge.	06:21
	← ghio left	06:21
solofight	how can i resolve this ?	06:21
	please help	06:21
sitaram	solofight: git status will tell you which files are in conflict. Edit them, resolve the conflicts, "git add" them, "git commit" and then retry your pull	06:22
	← kuadrosxx left	06:22
fynn	sitaram: surely 1 doesn't apply to .git/config	06:23
sitaram	fynn: agreed. Though even that I try doing with "git config ..."	06:23
solofight	sitaram: http://pastebin.com/m12dc9c34	06:23
	it says this	06:24
sitaram	solofight: look at line 2 of your paste	06:24
	→ galderz joined	06:24
solofight	how do i merge it ?	06:24
sitaram	solofight: do you know what a conflict is?	06:24
solofight	yeah i have an idea	06:25
	if they edit the same page same line	06:25
	conflict occurs right	06:25
	?	06:25
	multiple people i mean	06:25
sitaram	solofight: yes. Now do you know what a conflict looks like? (when you open such a conflicted file in a text editor)?	06:25
solofight	yeah >>>>>>>>>>> some text	06:26
	<<<<<<<<<<<<some text	06:26
	right ?	06:26
sitaram	solofight: yes	06:27
	solofight: except there's an ====== in between to separate the two sides	06:27
solofight	oh	06:27
sitaram	solofight: so you edit this file, fix up the final text as you like, remove the markers, and save. then "git add filename", commit	06:28
	solofight: you're essentially doing something manually because it could not be done automatically, that's all. Other than that this is just a normal merge	06:29
vshriram	sitaram: can i do <linux-2.6.29>+ my changes +<merge with android-2.6.29> ; where android-2.6.29 = linux-2.6.29+android changes	06:30
	i compared the commit ids - they are different	06:30
	→ saurabhverma joined	06:32
Derander	Hi. I'm trying to convert an svn repository (~/test/test) to a git repository (test_git_checkout) When I run the command I believe is appropriate, I get this error: http://gist.github.com/134396	06:32
	→ Fullmoon joined	06:32
Derander	The SVN repository is freshly svnadmin load'ed from a dump.	06:32
sitaram	vshriram: if I were you, I'd do it like this:	06:32
solofight	sitaram: thank you , will do and ask questions	06:32
sitaram	vshriram: (1) treat everything I get from outside as pristine in some sense. Keep them both updated in their own branches if needed, but make no changes to them locally	06:33
	vshriram: (2) create a merge branch for merging just your two remote sources since you seem to have two of them.	06:33
	vshriram: (3) keep all my changes in a separate branch, which -- assuming I haven't pushed to anyone -- I rebase off of whichever of the remotes is older	06:34
	vshriram: (4) create a throwaway integration branch for all the testing, where I merge whatever I want in whatever order and combination I want	06:34
	vshriram: make sense?	06:34
	→ cilly joined	06:35
	← bobesponja left	06:35
vshriram	what do you mean by "keep them updated in thier own branches" ? and whichever of remotes is older?	06:36
	can you take an example please i dont really follow	06:36
	what i do here is 1) Take patches from ouside trees using git format patches and apply them on top of my changes	06:37
	these patches are whatever i want to say android changes i clone the git tree of android - create patches from that git tree which apply on top of my 2.6.29 kernel	06:37
	→ ankit9 joined	06:39
	→ MetaCosm_ joined	06:39
	mina86\|aw → mina86	06:41
vshriram	sitaram: did i put you off?	06:42
	sorry	06:42
sitaram	vshriram: the problem is, I don't really know what the relationship is between linux and android trees? That is why I was a bit vague. Does the android tree track the linux tree faithfully? It seems like they are both out of step, and you need both of them in your work	06:42
	vshriram: no.. was away from keyboard :-) I dont get put off so easily; too old...!	06:42
vshriram	okay its is like this:	06:43
sitaram	so when I say "whichever is older", if you're not sure which then just choose one	06:43
vshriram	can we take this offline in case it is disturbing others	06:43
sitaram	vshriram: why should it? I want others to pitch in if they wish to; this is quite on-topic	06:44
vshriram	okay then	06:44
sitaram	not like we wandered off into vi versus emacs ;-)	06:44
Derander	oh dear.	06:44
	here we go	06:44
vshriram	android tree on google: <linux-2.6.29> + <android patches>	06:44
sitaram	but I do have to go for a meeting though; just remembered -- that is a dampener. I'll try and come back asap...	06:44
vshriram	okay np	06:45
sitaram	sorry vshriram (should plan better!)	06:45
vshriram	no problem sir	06:45
lut4rp	vshriram: just a passing thought, your company is working on Android?	06:46
vshriram	yes why?	06:46
	← JasonWoof left	06:48
vshriram	okay i will be back later	06:48
	← vshriram left	06:48
	← hax0r1337 left	06:49
	→ Pe3k joined	06:53
	← cilly left	06:54
	← MetaCosm left	06:56
solofight	sitaram: i did the same and even now	06:56
	it says same	06:56
	→ alexyk joined	06:57
Pe3k	hello, by merging branches I have to merge in kdiff3 two files, each has on specific line different text, but I want to have there both (first after second) is it possible in kdiff3?	06:57
	→ ph^ joined	06:57
	← tango_ left	06:58
alexyk	which port is used by git clone git:// ? I can get my own repos with git@github via ssh but not public ones with git://github -- which port needs opening??	06:58
	← _rane left	06:58
	→ _rane joined	06:58
	→ bentob0x joined	06:58
	→ bugfux joined	06:59
solofight	thank you sitaram	07:00
	← disjunctor left	07:00
	← dermoth left	07:00
	→ disjunctor joined	07:01
	← Cromulent left	07:02
	→ st__ joined	07:02
	← johndoigiii left	07:02
	← st__ left	07:03
Pe3k	ok, I found out how to do it ...	07:03
	→ vshriram joined	07:03
vshriram	i have clone a git tree from a repository available on the net - can i find out how it was constructured. what command were issued. whether it was rebased/merged during its construction?	07:04
	sitaram has to go away again but can't resist #git... :-)	07:04
sitaram	vshriram: not easy	07:05
	→ emias joined	07:05
	← Fullmoon left	07:05
sitaram	vshriram: some hints can be in the commit messages, but it is the reflog that contains all those secrets. And a reflog does not travel :(	07:05
	vshriram: however, the idea is that if the SHAs are the same, this does not matter	07:05
vshriram	i tried reflog , but it gives very less info say 4/5 commits	07:05
sitaram	vshriram: a reflog is local, so it will only show you your stuff	07:06
vshriram	sitaram: i tried applying git am after formatting patches it gives me the same commit id	07:06
	as i do a merge	07:06
sitaram	vshriram: also, don't forget the "git cherry" commmand (not cherry pick) which can match commits where the actual change is the same but the SHA changed (due to author/committer change or maybe some whitespace or whatever)	07:06
	→ Grahack joined	07:07
sitaram	vshriram: if it gives you the same commit id then that's good; I had not expected that. I guess I need to learn something now :-)	07:07
vshriram	when you say sha is it commid	07:08
	← ben_h left	07:08
	← fynn left	07:08
sitaram	yes; that 40-digit thing	07:08
	→ lohrun joined	07:08
	sitaram has to go again; be back in 20-30 min, sorry again vshriram	07:09
vshriram	okay, if i want sources from different repositories located in different servers which have a common base, then i can do merges correct?	07:10
	okay no problem. i dont have much to do here - i better hang around and learn more	07:10
	← agenteo left	07:12
	→ Cromulent joined	07:13
	→ julm joined	07:16
	← abdelhalim left	07:17
	→ Juan_ joined	07:18
	← vshriram left	07:20
	→ vshriram joined	07:21
	← julm left	07:22
	← Grahack left	07:23
	→ julm joined	07:24
	→ patrikf joined	07:26
	→ JasonWoof joined	07:26
	← bentob0x left	07:29
	→ parasti joined	07:30
	→ jackylee joined	07:30
	→ khmarbaise joined	07:30
	ianmcorvidae\|alt → ianmcorvidae	07:32
	→ Stravy joined	07:32
	→ ceyusa joined	07:33
	← Derander left	07:34
	→ bentob0x joined	07:35
	→ giallu joined	07:35
	→ dermoth joined	07:37
	→ Modius joined	07:37
	→ Weaselweb joined	07:37
	← guitsaru left	07:37
	← rpag left	07:38
	→ ianmcorvidae\|alt joined	07:39
	← ianmcorvidae left	07:39
	marc_m → marc_kdab	07:40
	← wagle left	07:41
	← johan-s left	07:42
sitaram	vshriram: yes you can	07:44
	→ warthog9 joined	07:44
	← patrikf left	07:46
Pe3k	how can I get list of staged files?	07:47
	→ asksol joined	07:47
	→ harinath joined	07:48
	← eddyp left	07:48
	→ ilteris__ joined	07:48
	→ khmarbaise_ joined	07:49
	→ tango_ joined	07:49
	→ AAA_awright_ joined	07:50
sitaram	git status	07:50
	→ rpag joined	07:50
	→ ciskje joined	07:51
	→ Blastur joined	07:53
	→ johan-s joined	07:53
Blastur	hey. i tried to revert a commit, but accidentially reverted too much (not commited yet though) .. how can I undo a git revert command? :)	07:54
	→ tbf joined	07:55
	→ jmesnil joined	07:55
	← khmarbaise left	07:56
	→ asksol_ joined	07:57
	→ Grahack joined	07:57
sitaram	a revert creates a commit. what exactly is the command you used?	08:00
	→ lbt joined	08:00
	← sunoano left	08:00
	→ solydzajs joined	08:00
	→ cytrinox joined	08:01
	→ shruggar joined	08:05
	← asksol left	08:05
	← AAA_awright left	08:07
sgronblo	Ok, I now noticed I'd like to revert a change in an old commit to a file. I found the sha1 for the commit in the log. How do I first checkout the patch for that commit to make sure it has only what I want and then reverse the patch on the file?	08:07
shruggar	you can "git log -p hash" to see the patch	08:08
	git log -p hash filename (I think) to see the patch for that file	08:08
	← Cromulent left	08:09
sgronblo	Ok, I think I found the patch with git diff sha1 sha1^	08:09
Blastur	sitaram: i did git revert HEAD~2, but i realized it was "git revert HEAD^" i wanted to do .. so (in panic), i did "git reset --hard", and now im not sure what the state of my working copy is	08:10
	← jackylee left	08:10
sgronblo	Now it would be nice to be able to reverse apply this commit's patch	08:10
Blastur	but i realized i wanted to do "git revert HEAD^" even*	08:10
shruggar	sgronblo: git revert sha1 ?	08:11
sitaram	Blastur: first -- dont panic!	08:11
	Blastur: is this windows? or do you have a GUI available?	08:12
	→ cytrinox1 joined	08:12
shruggar	Blastur "git reflog" sounds like what you need	08:13
	→ Fullmoon joined	08:13
sitaram	Blastur: git revert is not what you think it does on other systems; in git it adds a new commit	08:13
	shruggar: not necessarily; he "revert"ed which means he added a new commit on top	08:13
Blastur	sitaram: im on ubuntu, no gui	08:13
	makies → makies_bot	08:14
sitaram	Blastur: no gui, ok don't worry. try this: git log --pretty=oneline --abbrev-commit -9	08:14
shruggar	sitaram: but "git reflog" should be able to tell him where he was originally, so he can "git reset" to that	08:14
	← alexyk left	08:14
Blastur	sitaram: i would basically just like to forget that the two last commits of http://pastebin.com/m5bdd2dc7 never happened	08:14
sitaram	shruggar: yes...	08:15
Blastur	or actually, all 3	08:15
	← julm left	08:15
	→ skyl joined	08:15
shruggar	git rebase -i HEAD~4 and delete what you don't want :)	08:15
sitaram	Blastur: is that all you have in reflog? if you want to forget all 3 you need one more below them	08:15
	← cytrinox_ left	08:15
sitaram	shruggar++	08:15
	→ julm joined	08:15
	← mnick left	08:16
Blastur	ok, in this case http://pastebin.com/m5c656846	08:16
skyl	can I put .svn in .gitignore and have anything that has a dir/sub_dir .svn ignored?	08:16
Blastur	i want to back to the latest "pull"	08:16
shruggar	skyl: just ".svn" should be enough, no need for asterisks	08:16
	← AkumaStreak_ left	08:16
	← bugfux left	08:17
sgronblo	Hmm, I did git revert commit with a dirty tree and I got some problems though I'm not sure it really was because of that.	08:17
	← afief left	08:17
skyl	and that is good for anything in the path ig here/.svn/there/filename.ext	08:17
	→ sunoano joined	08:18
	→ Pistahh joined	08:18
Blastur	shruggar: what is rebase suppose to do? :)	08:18
	→ ankit9_ joined	08:19
sitaram	Blastur: git reset --hard 9471485	08:19
	→ vshriram_ joined	08:19
shruggar	Blastur: rebase -i opens up an editor showing what amounts to a short log. You can move around lines, delete them, etc, and history will be rewritten to match what you have done. It is the most powerful weapon in the universe, and only you can stop the invading armies of the dead	08:20
sitaram	Blastur: rebase is for finer control over what you want to keep and not keep: http://sitaramc.github.com/other-stuff/gittalk.html#(61)	08:20
	← cytrinox left	08:21
sitaram	(although that example could do with a written version of my voiceover; it's still understandable I think)	08:21
	shruggar: just read what you said... LOL!!!	08:22
	→ ignas joined	08:22
	→ cytrinox joined	08:22
sitaram	Blastur: he is not joking; it is powerful :-)	08:22
osfameron	sitaram: what tool do you use to make the slides?	08:22
	← girishr left	08:22
osfameron	ah, Slidy	08:23
sitaram	osfameron: HTML slidy and plain text	08:23
	yup :-)	08:23
shruggar	sitaram, do you have mp3s with beeps? :)	08:23
	→ psoo joined	08:23
sitaram	shruggar: no mp3s at all. I meant "voiceover" in the literal, live, sense :-)	08:23
shruggar	ah	08:23
	osfameron occasionally thinks about using an html markup... but always goes back to OO Impress or similar	08:23
Blastur	sitaram: cool :) did not know git could do that!	08:23
	← Grahack left	08:23
Blastur	now i have managed to restore order using the git reset command though	08:23
sitaram	Blastur: see shruggar's description above :)	08:23
	← JasonWoof left	08:24
shruggar	you always need to worry about time zombies if you alter history :)	08:24
Blastur	new problem: when i push stuff to my colleague, it says "error: unable to create temporary sha1 filename ./objects/0c: File exists" .. even though the files in our "bare" repo is owned by a common group	08:24
	→ naeu joined	08:25
sitaram	you don't have +s on the group id of the directory	08:25
Blastur	oh, ok	08:25
	what does that mean? :)	08:25
vshriram_	can i set how many lines reflog displays	08:26
sitaram	that's a pure unix thing (meaning not git). Without that flag, files get owner and group as per the user who created them	08:26
	← jds left	08:26
vshriram_	it displays only 12 lines for me. can it display the entire log	08:27
shruggar	Blastur, it's the "sticky" bit, it basically means (in directory context) "files in this directory inherit the group"	08:27
sitaram	Blastur: but with that flag (chmod -R g+s your.repo), unix makes the files group owned, regardless of who wrote them	08:27
	(and as usual shruggar has a better way of saying it ;-) I need to hire him as my ghostwriter or something...!)	08:27
	vshriram_: by default, it displays everything. So you only have 12 lines	08:28
Blastur	ok .. but shouldn't it work if i have setup my machine to always create new files in that group?	08:28
shruggar	I'll shut up now, I do _not_ want a reputation as someone who is /good/ at saying things. That would go against years of effort	08:28
sitaram	shruggar: LOL! you did it again :(	08:28
	Blastur: explain how you did what you just said...	08:28
	meaning what command	08:29
shruggar	sitaram, slide 10 mentions 3 gui tools available. I know of "gitk" and "git gui", what else is there?	08:29
sitaram	shruggar: git gui blame	08:29
Blastur	sitaram: by setting the primary group in /etc/passwd for my user	08:30
	→ chuck joined	08:30
chuck	why doesn't git add empty directories?	08:30
	← cytrinox1 left	08:31
shruggar	sitaram, ah, I think I use vim for that. Also, it choked when I gave it an absolute file path o.o	08:31
sitaram	Blastur: hmm not sure what that would do. Can you paste "ls -al $REPO/objects/0c"? and I will explain better...	08:32
shruggar	chuck: the short answer is probably "because empty directories don't get output by "diff -u""	08:32
sitaram	shruggar: git gui blame is a little sensitive. Not too many people know it, so it's a little shy. :-)	08:32
	→ cytrinox_ joined	08:32
sitaram	chuck: git tracks content, not files, so empty directories don't get tracked. Often debated, but that's the way it is.	08:33
	chuck: workaround: "touch .gitignore" inside your empty directory	08:33
Blastur	sitaram: http://pastebin.com/m4c1c603c	08:33
chuck	hrm.	08:33
sitaram	Blastur: who tried to push and got an error then?	08:34
parasti	Git is also not a deployment tool, but a VCS, and empty directories don't make much sense in that context	08:34
Blastur	i tried to push, and i got the error	08:34
shruggar	chuck, basically the empty-directory thing isn't "interesting" enough for anybody to fix, and goes against some traditions of how patches are sent. And yeah, as sitaram said, it's difficult to give a directory an "identity" which can be usefully understood. The accepted work-around is to put .gitignore into the empty directory	08:34
Blastur	the ls-al was on my colleagues machine, who i try to push to	08:34
	maybe you meant i should ls -al in my own $REPO ?	08:34
sitaram	wait, I was unclear. I mean, I need that "ls -al" on the server side (the server hosting the repo to which you are pushing)	08:34
Blastur	yeah, thats it	08:35
sitaram	well what is the userid you're using to do the push?	08:35
	← ankit9 left	08:35
sitaram	Blastur: in other words, the userid in the git Remote URL	08:35
	← harinath left	08:35
	→ harinath joined	08:37
	← julm left	08:38
	→ jds joined	08:38
	→ julm joined	08:38
	← cytrinox left	08:38
	→ mbroeker joined	08:39
	→ cytrinox joined	08:41
	← vshriram_ left	08:42
	← sfwc left	08:42
yhager	how do I "unadd" a file?	08:42
sgronblo	weird, the git revert commit worked for 2 out of 3 files that are affected by the commit, but one of them was not updated or something so I get nothing to merge there.	08:43
	→ tono joined	08:43
parasti	yhager: read "git status" output	08:43
yhager	parasti: whoops. sorry.	08:43
parasti	yhager: no problem. this is kind of a FAQ :)	08:44
yhager	parasti: I urge everybody to read man page and program outputs, but on the trivial stuff I miss it myself..	08:45
	→ ppawel joined	08:46
sgronblo	Oh I think I found it...	08:46
	→ rindolf joined	08:47
	→ intripoon joined	08:48
rindolf	Hi all. "git status" gives me "Your branch is ahead of 'origin/master' by 2 commits." - what can I do about it?	08:48
yhager	rindolf: if you want to update origin/master, push your changes with 'git push'	08:48
rindolf	yhager: ok.	08:48
	yhager: did that. Still the same problem.	08:49
yhager	rindolf: any errors from git push?	08:49
rindolf	yhager: no.	08:50
	no errors.	08:50
Blastur	sitaram: well, my user is magnusol, im in group axusers. my colleague is joneib, and he is also in group axusers	08:50
	but i (magnusol) cannot push to his repo	08:50
rindolf	"No shrubberies."	08:50
	→ estan joined	08:51
yhager	rindolf: let's see if someone else picks this up, I'm quite a n00b in git, but push/pull I know :)	08:51
	→ girishr joined	08:51
volante	l/exit	08:53
	← cytrinox_ left	08:53
volante	oops	08:53
	← volante left	08:53
	dwmw2_gone → dwmw2	08:54
sitaram	Blastur: can you paste ls -al, not ls -l? The "a" is important	08:54
	rindolf: what is your current branch? and is your version very recent (enough to have "push.default" set to "nothing"?)	08:55
rindolf	sitaram: my current branch is "master".	08:57
	sitaram: and which version? Of git? Of the repository?	08:58
	← blaenk left	08:58
sitaram	rindolf: git	08:58
	rindolf: git --version	08:58
rindolf	sitaram: git version 1.6.3.2	08:58
sitaram	anyway try "git push origin master" and paste the output if it doesn't work	08:58
	rindolf: aaha! also try "git config push.default"	08:59
	→ gpampara joined	08:59
sitaram	and tell me what it says	08:59
rindolf	fatal: protocol error: expected sha/ref, got '	08:59
	*********'	08:59
	OK.	08:59
sitaram	rindolf: your ssh session is not clean	08:59
rindolf	sitaram: ah.	08:59
	sitaram: WDYM?	09:01
Blastur	sitaram: i did the g+s stuff on my colleagues machine now, and now i can push :)	09:01
	happy days	09:01
	→ kobani joined	09:01
_Vi	rindolf, Extra characters appear in SSH connection. Try "ssh ... ... echo qqq" - you should see only "qqq", nothing else.	09:02
	→ pcapriotti joined	09:03
	← vshriram left	09:03
sitaram	rindolf: sorry was away	09:03
rindolf	error: Malformed value for push.default: git://git@github.com:shlomif/cpanhq.git	09:04
	→ vshriram joined	09:04
	sitaram nods at what _Vi said	09:04
	← estan left	09:04
_Vi	rindolf, It may be caused by ".bashrc" or similar script that output time, fortune or something.	09:05
	rindolf, (on that side).	09:05
	← KiBi left	09:05
vshriram	i checkout a git repository from the net. i checkedout a branch 'git checkout orgin/xx' now i do a git reflog. it shows me only 2 entries. what about all the activty done at the remote server where git was located. why reflog does not display that	09:06
_Vi	vshriram, Reflog is your local one.	09:06
	vshriram, It is not cloned.	09:06
	vshriram, Use usual "git log".	09:07
vshriram	cant i use reflog thing. i dont want the log of commits done. i want to know what commands were used/whether they were rebased/merged things like that	09:07
_Vi	vshriram, In "git log" or similar things you will see when they was merged (not fast-forward).	09:08
	vshriram, Public branches just should not be rebased or filtered.	09:08
	→ BigMack83 joined	09:09
vshriram	vi: how do i know/make out if a merge/rebase has taken place. until you get a merge coflict you wont know	09:10
BigMack83	how would I use this git command if my server operates on a different port than normal? git clone ssh://username:xx.xx.xx.xx/home/public_html/gits/project.git	09:10
Blastur	thanks for all your help sitaram & shruggar	09:11
wereHamster	BigMack83: ssh://username@address:port/home/..	09:11
BigMack83	normally with ssh command i would just add the "-p xxxx" option to imply a certain port	09:11
sitaram	Blastur: you're welcome	09:11
BigMack83	wereHamster: thats what i tried, didnt work. maybe its another issue then	09:11
wereHamster	BigMack83: it must be another issue then	09:12
sitaram	BigMack83: another way is to create a stanza in $HOME/.ssh/config that has all the right values and just use that in your git url	09:12
	BigMack83: if you need details ask	09:12
BigMack83	might be a key thing. but i wouldnt think so since i dont have any setup	09:12
_Vi	vshriram, If do no changes and just pull from remote and it shows that it is not "fast forward", it is sign of remote rebase.	09:13
BigMack83	sitaramyou have a link showing how to do that?	09:13
sitaram	BigMack83: also, "didn't work" is not very useful ;-)	09:13
wereHamster	BigMack83: does normal ssh to that port work?	09:13
BigMack83	yes thats how im in the server now	09:13
_Vi	vshriram, Just don't use rebase on branches that can be fetched by someone.	09:13
BigMack83	sitaram: lol i know im tryign to get more info.	09:13
rindolf	OK, working now.	09:14
BigMack83	ssh: Could not resolve hostname bigmack83:209.20.89.68:16487: Name or service not known	09:14
	fatal: The remote end hung up unexpectedly	09:14
	→ struberg joined	09:14
BigMack83	^^ thats the error	09:14
sitaram	BigMack83: your URL looks weird. Shouldn't there be an @ sign between "bigmack83" and 209 instead of ":"	09:15
	?	09:15
vshriram	_Vi: can you elaborate?	09:16
	i didnt understand what you told	09:16
_Vi	BigMack83, I sometimes start something like "socat tcp-l:22,bind=127.0.0.2,fork,reuseaddr tcp:remote_host:remote_port" and use ssh://127.0.0.2/	09:16
	BigMack83 feels retarded	09:16
BigMack83	yea seems i was using ' : ' instead of ' @ ' for whatever reason and didnt see it. its connecting now. i need coffee	09:16
_Vi	vshriram, "rebase" is not a command that should be applied to public branches. Do you understand what is a public branch?	09:17
sitaram	_Vi: not needed, just use a stanza in ~/.ssh/config and then say "git clone github:projectname.git" or whatever (the stanza is called "github" -- and contains everything you need to connect!)	09:17
	BigMack83: :)	09:17
	← julm left	09:17
_Vi	sitaram, (Or similar thing with "iptables -A ... -j DNAT")	09:17
	→ jkacur joined	09:18
BigMack83	sitaram: thx. so can you elaborate on the stanza thing?	09:18
jkacur	should I be using git-apply directly or is that just low-level git?	09:18
	→ julm joined	09:18
sitaram	BigMack83: sure. wait	09:19
jkacur	if I have a patch that is not in a mail box, git-am -3 doesn't work, because you get a msg that "Patch does not have a valid e-mail address."	09:19
	ideally what I want is to apply the patch and then fix the conflicts, but not quite sure how to accomplish that	09:19
shruggar	sitaram: your words intrigue me. What manpage and keyword would I need to learn more about "stanzas"?	09:20
vshriram	Vi: public branch = that is accessible to public	09:20
	Vi: other developers pull, clone from this branch	09:20
	correct?	09:20
	→ cilly joined	09:21
	← struberg left	09:21
vshriram	now what if i have to create a repo using the contents of 2 or more public branches + my own contents. should i use rebase/merge. i am talking of linux here. so the base is same for all the 3.	09:22
sitaram	shruggar: hmm I've been calling them stanzas God knows why :-) Paragraphs maybe?	09:22
_Vi	vshriram, You want to create a repo just for you, or it will be clonable/pushable too?	09:23
vshriram	it will clonable and pushable	09:23
sitaram	BigMack83, shruggar, here is mine:	09:24
	http://pastebin.com/d7d6d3672	09:24
	← Flathead left	09:24
_Vi	vshriram, Calling "rebase" every time is bad idea. Using "merge" every time is better idea.	09:24
	→ cytrinox_ joined	09:25
sitaram	BigMack83, shruggar, in fact you can use them really well when you're behing proxies etc: http://sitaramc.github.com/tips/git-over-proxy.html	09:25
shruggar	sitaram, ah, misread, I'm dumb. I was thinking you meant a way to specify psuedo-protocols for git the way you can in svn	09:25
BigMack83	oh ok. thats cool. i see. but how do i use it in action? how would i tell git to use the file for that info	09:25
sitaram	shruggar: oh... sorry! I did say ~/.ssh/config, in my defense!	09:26
_Vi	vshriram, Once you call "rebase", you will "lose connection" with one of that "2 or more public remote branches".	09:26
	→ gitte joined	09:26
sitaram	BigMack83: "git clone gh:path/to/repo.git"	09:26
BigMack83	oooh, thats cool	09:26
shruggar	sitaram, yes, I said /I/ misread and was stupid :)	09:26
sitaram	BigMack83: where "gh" is the name of the "stanza" heh	09:26
BigMack83	thanks. will try it out	09:26
shruggar	also, why does "git instaweb" complain that I need to be in the top of my work tree instead of just, like.. going there itself?	09:26
BigMack83	yea. thats pretty cool. does that work the same for things like ssh/scp/sftp and such?	09:27
vshriram	_Vi: i have to create a repo - <merge public repo 1> + <merge public repo 2> + Add my contents + <merge changes in public repo1 that taken place since> + add my changes + <merge changes taken place in public repo 2 that might have taken place>	09:27
sitaram	BigMack83: all of 'em	09:27
BigMack83	nice	09:27
vshriram	is that what you say?	09:27
BigMack83	see i like this room already, and ive only been in here for 10 minutes	09:27
_Vi	vshriram, Yes, it is the right way used in many repositories.	09:28
	vshriram, When you merge second time, it will automatically do "that taken place since".	09:28
sitaram	BigMack83: not to forget rsync :-)	09:28
BigMack83	yea. im still somewhat new to admin stuff, i havent even tried rsync yet.	09:29
	its on a long list of stuff i need to try out	09:29
sitaram	BigMack83: rsync will make you forget scp :)	09:29
BigMack83	ok	09:30
vshriram	Yeah	09:30
	shruggar clings on to his scp with his cold dead hands	09:30
	shruggar considers seeing a doctor about those	09:30
	sitaram breathes new life into shruggar's hands using rsync	09:30
shruggar	thanks! I - wait!	09:30
	sitaram believes scp killed shruggar's hands :-)	09:30
sitaram	LOL!	09:31
_Vi	vshriram, Are pubrepo1 and pubrepo2 related? If they come from common ancestor merge will work even better.	09:31
BigMack83	yea, but most of the others in my group are all windows users. i was too until 2 months ago. (ubuntu now) so i need to be able to use stuff they can use easily as well.	09:31
	← cpg left	09:31
BigMack83	hah	09:31
_Vi	vshriram, If they are completely unrelated, it may be better to put them in separate subdirectories in working tree (submodules).	09:32
BigMack83	sitaram: for hostname ssh.github.com would i replace ssh.github.com with just the IP if its on a personal server?	09:34
	← vshriram left	09:34
	→ cytrinox1 joined	09:35
	→ Juan__ joined	09:36
	→ vshriram joined	09:36
vshriram	_Vi: they come from a common ancestor but they are different features, which are not related	09:37
	← cytrinox left	09:38
vshriram	Vi: When i merged the first tree, i got all the commits of that tree < i mean 200 or so commits> + there was a merge conflict, which i resolved. so far okay. now i try to merge the second tree	09:38
sitaram	BigMack83: yes	09:38
vshriram	i dont get any conflicts but i get only one single commit	09:38
BigMack83	ok cool. thanks for the help and tips	09:39
vshriram	but when i use git format-patch to see - it in turn creates 100 commits out of that single commit which reads as merge....	09:39
	why is this?	09:39
	← DavidKlein left	09:40
	→ cytrinox joined	09:40
sitaram	BigMack83: you're welcome	09:40
wereHamster	vshriram: how did you invoke format-patch? did you limit it to a number of patches or a rev range?	09:41
	← cytrinox_ left	09:42
	← univate_ left	09:43
vshriram	i used that last <commit id> before the merge as the starting point	09:43
	← julm left	09:44
vshriram	i done the following: <git checkout mytree> + <git pull repo1 - which creates 200 commits> + <resolve merge conflicts> + <git pull repo 2 - which creates only one commit - commit message is merge..>	09:45
	now if i do git format-patch <commit id of merge conflict> .. -o ./patches i get 200 patches	09:45
	but for these 200 patches i get only one commit in my tree	09:45
	→ julm joined	09:46
wereHamster	vshriram: no, there is only on merge commit, but it has two parents and one of the parents is pointing to those 200 cmmit	09:46
	commits	09:46
	makies_bot → makies	09:47
	→ Cromulent joined	09:48
	→ DavidKlein joined	09:49
	← vshriram left	09:50
	→ cytrinox_ joined	09:50
	→ jaeckel joined	09:51
	← Juan_ left	09:51
	← cytrinox1 left	09:51
	→ vshriram_ joined	09:54
vshriram_	wereHamster: i did not understand what you told. am i doing something wrong?	09:55
	← Stravy left	09:56
wereHamster	vshriram_: git-format-patch will create one patch for each commit listed in git log --pretty=oneline <commit id of merge commit>..	09:57
	→ KingBug_ joined	09:57
	← cytrinox left	09:58
Blastur	what is the recommended way of pulling while having uncommited stuff in your working dir? git stash and then pull?	09:58
	← vshriram_ left	09:58
wereHamster	Blastur: yes	09:58
	← _Vi left	09:58
	→ univate_ joined	09:59
	KingBug → Guest4758	09:59
	KingBug_ → KingBug	10:00
	→ eddyp joined	10:02
	← ilteris__ left	10:02
	← cilly left	10:06
	→ cilly joined	10:06
	← Fullmoon left	10:07
	→ cytrinox joined	10:07
	← Cromulent left	10:08
	← k0001 left	10:09
	← cytrinox_ left	10:09
	→ Cromulent joined	10:09
	→ vshriram joined	10:09
vshriram	wereHamster: i thought so - one entry in git log = one patch using git format patch. when i did a merge i got a single commit saying <merge from so and so git tree and branch>. now i did a git format-patch <commit id of the last commit before the merge>.. now i get 200 patches out of it	10:10
	← Guest4758 left	10:11
	mstormo → mstormo_	10:12
	← cytrinox left	10:12
	mstormo_ → mstormo__	10:12
	mstormo__ → mstormo___	10:13
	→ cytrinox joined	10:13
	mstormo___ → mstormo	10:13
	← sitaram left	10:14
wereHamster	vshriram: git-merge creates one new commit, but that commit references a long history.	10:14
jkacur	Can someone explain what to do when a git apply patch fails?	10:15
	← cao left	10:15
	← vshriram left	10:15
jkacur	What I want it to do is apply the patch with conflict markers that I can fix.	10:15
	is that possible?	10:15
	(like what I think git-am -3 does)	10:15
	→ ShadowBot joined	10:16
	← robinsmidsrod1 left	10:17
	← ShadowBot left	10:18
	→ ntoll joined	10:18
	→ vshriram joined	10:19
	→ cao joined	10:19
	← sergiodj left	10:19
vshriram	sorry, getting disconnected too often	10:19
jkacur	:)	10:19
	freenode woes	10:19
vshriram	probably it does not like users who ask lots of questions ;-)	10:20
jkacur	freenode --eject 3_or_more_questions	10:21
	→ Fullmoon joined	10:23
	← vshriram left	10:25
	→ sitaram joined	10:26
	→ fujin joined	10:26
	→ cytrinox_ joined	10:26
	→ ShKoDrAnI joined	10:27
	← tbf left	10:28
	← galderz left	10:28
	→ girishr_ joined	10:29
	→ tbf joined	10:29
gitte	pasky: I had no idea what the talk of the missed ping packets was about yesterday, but I think I see it now: 1kB/sec. That is awful!	10:31
	← girishr_ left	10:31
	→ girishr_ joined	10:32
	→ Fullmoon_ joined	10:32
	← Cromulent left	10:33
pasky	gitte: i've sent a mail to the isp, no reply yet	10:33
gitte	pasky: I thought rover was in your university?	10:33
pasky	no	10:33
	← girishr left	10:33
gitte	Okay, but this issue is definitely new.	10:33
pasky	it's at UPC/Chello, fairly large household provider in *.cz, so hopefully someone notices this soon	10:34
	→ corni joined	10:35
gitte	pasky: my experience with such people: they do not read email, or delete them after reading without further action. You need to phone them, I guess...	10:35
	→ UUncia joined	10:36
pasky	well, i'm sorta friends with them, so I'd hope not :)	10:37
	→ girishr joined	10:37
	← cytrinox left	10:39
	→ janmejay joined	10:41
	→ cytrinox joined	10:41
gitte	pasky: now I have 0kB/sec	10:41
	pasky: maybe you want to pay them a visit?	10:42
janmejay	for a tracking branch, is there any way to lookup the branch it tracks?	10:42
sitaram	git remote show origin -- should do it	10:43
	→ jerome joined	10:45
	← emias left	10:45
janmejay	yes it does, but the repository i have is a git-svn repo. there is this local branch called 'master' which is tracking 'remote/trunk' and i have a script in which i want to discover this relation (so that it works for other tracking branches too)	10:45
	→ vermila joined	10:46
	← cytrinox_ left	10:48
	← Fullmoon left	10:48
	→ emias joined	10:49
	→ aziz joined	10:49
	→ ereslibre joined	10:50
	jkacur gets his work done with a bastardized use of quilt and git together	10:51
	→ wxr483 joined	10:52
	← girishr_ left	10:52
	→ vshriram joined	10:52
vshriram	weremaster: this does not happen often. sometimes it (merge) creates only one commit. sometimes merge creates 100's of commits	10:52
	→ ankit9__ joined	10:54
drizzd_	a merge always creates one and only one commit. It's the one saying "Merge branch '...'".	10:54
vshriram	hmm. not happening for me. does a git pull <repo ip address> <branch name in the repo> do the merge?	10:57
	→ ankit9 joined	10:58
drizzd_	yes, unless you configured to use rebase instead	10:59
wxr483	Hey people, I'm pretty new to branches and stuff, so I need some adwise... I'm working on a project where I download nightly builds every morning. The changes I made comes in two categories: in-house and public. The public patches are sent upstream and gets integrated in the next nightly. In-house patches are not.	11:00
	How should I set up my stuff to get working without to many conflicts?	11:01
drizzd_	I'm assuming you do understand what a merge is in git. A merge essentially ties two histories together. So the resulting history will have all commits from both histories.	11:01
	→ vshriram_ joined	11:02
gpampara	Well, two or more	11:02
	← ankit9__ left	11:03
drizzd_	gpampara: from the description vshriram gave above I inferred that he's only merging one branch at a time	11:03
gpampara	drizzd_: no worries, just mentioning it	11:04
vshriram	drizzd_: does a git pull do a merge?	11:05
drizzd_	vshriram: 13:00 < drizzd_> yes, unless you configured to use rebase instead	11:05
vshriram	how do i make sure i have not configured to use rebase?	11:06
drizzd_	wxr483: to avoid conflicts make sure you don't touch the same code in different branches	11:06
	makies → makies_bot	11:07
drizzd_	vshriram: read the man page for git pull	11:07
wxr483	drizzd_, yea, but if I add (or remove) large chunks at once, can't the offset go wrong?	11:08
drizzd_	wxr483: always apply patches to the version they apply to, _then_ merge if necessary	11:09
	(or rebase)	11:09
vshriram	drizzd_: i think git pull is working for me. i thought it will pull the patches on top of the first merged tree. but the patches of both the trees have been applied interchanging	11:10
	i did a git checkout baseline_branch + merge public repo 1 + merge public repo 2	11:11
	← ankit9_ left	11:12
vshriram	now i thought what will happen is <baseline branch> + patches of repo 1 + patches of repo 2	11:12
	→ murph_ joined	11:12
vshriram	what is happening is <base line branch> < some patches of repo 2> + < some patches of repo 1> + < some more patches of repo 2> + < some more patches of repo 1> and so on	11:13
	probably it is doing that based on commit date and something else	11:13
drizzd_	vshriram: take a look at gitk --all	11:13
	vshriram: the git log manpage says "by default commits are ordered in reverse chronological order"	11:14
	← harinath left	11:14
drizzd_	since git has a non-linear history you can't represent it as a simple sequence of patches	11:14
	← julm left	11:15
	→ Arafangion joined	11:16
	→ julm joined	11:16
	← UUncia left	11:16
	→ harinath joined	11:16
	← ignas left	11:17
	← solydzajs left	11:17
	→ ngirard joined	11:18
vshriram_	gitk --all does not display things neatly icons are very very big	11:19
ngirard	Hi all. Hopefully my question is not off-topic: when i try to push using git-gui, I get "Enter passphrase for key '/home/ngirard/.ssh/id_rsa':" printed to STDOUT, and i don't know how to enter it.	11:19
	Any thoughts ?	11:19
Arafangion	ngirard: The standard answer is to use a key manager, but I've never used it.	11:21
	→ d0k joined	11:21
ngirard	Arafangion: aha. Ok, I get it. Thanks for your answer.	11:21
	→ spaceonline joined	11:22
	→ saurabhverma1 joined	11:23
	→ WebDragon joined	11:24
	→ includao joined	11:25
	→ aspotashev\|eeepc joined	11:25
	← includao left	11:25
	→ dodo_the_last joined	11:26
	← murph left	11:29
	← Fullmoon_ left	11:34
	→ Sigma joined	11:34
	makies_bot → makies	11:35
	← lbt left	11:37
	← torbenh3 left	11:37
	← cilly left	11:39
	→ vhallac joined	11:40
saurabhverma1	hi all , can sombody help me with git granular permission handling ?	11:40
	what i want to do is , creating git repos , and assigning group based permission for write , the problem is some repos needs more than one AD group	11:41
	so what i was thinnking is to make single local group composing many AD group and give access	11:41
	is there any better way todo it ?	11:41
	← saurabhverma left	11:41
	→ bcardarella joined	11:42
	→ torbenh3 joined	11:42
Arafangion	AD?	11:42
saurabhverma1	active directory	11:44
Arafangion	What does AD have to do with it?	11:44
	→ ijin joined	11:44
	← vshriram left	11:45
	← bcardarella left	11:45
saurabhverma1	ok , We need to have git repos , all our users and groups are maintened in active directory , So each repo has to be accessed by users of particular AD group ,	11:45
	the problem is some repos needs to be accessed by multiple groups , and i dont know how can i map two or more domain groups to a local group	11:46
Arafangion	Users can easily belong to more than one group.	11:47
	← ereslibre left	11:47
saurabhverma1	Arafangion: true , but its not necessary two groups have same users , so two different set of groups have different set of users which can't be common	11:50
	→ Fullmoon joined	11:50
saurabhverma1	git init --shared can give group based access but i doubt it can have more than one group	11:50
Arafangion	saurabhverma1: Why couldn't they?	11:51
	saurabhverma1: You're overcomplicating things.	11:51
	saurabhverma1: If you're concerned, just give each git repository its own group. Add the users you care about to that particular group.	11:51
saurabhverma1	Arafangion: sounds good . so you suggests take all the users from all the Domain groups and add that to local group and give that group access to the repo ?	11:52
	→ ph^_ joined	11:52
	← ph^ left	11:53
Arafangion	My goodness, no.	11:53
saurabhverma1	may be i'm little saying with little complication , :( . I am brand new to all this Active directory and linux integration	11:54
Arafangion	Just... 1) Create a branch, call it wazoo-git-group. 2) Create the wazoo git repository. 3) Add the users who need to interact with the git repository to the wazoo-git-group.	11:54
saurabhverma1	Arafangion: ok i got what you are saying now , but these users are not local users , these are users defined in Active directory	11:55
Arafangion	saurabhverma1: AD is beyond the scope of this channel. If I recall correctly, it is basically an LDAP server with some Microsoft modifications... Consider investigating how to configure Linux to authenticate against AD instead.	11:56
	If possible.	11:56
	← WinterWeaver left	11:56
saurabhverma1	Arafangion: i have already done it , cool let me do some more investigation . thanks a ton :)	11:57
	→ ereslibre joined	11:57
Arafangion	saurabhverma1: I would personally advise you to consider using gitosis instead.	11:58
	saurabhverma1: And consider the need to have more than just a few people directly work on a project.	11:58
	saurabhverma1: I can't imagine you having more than a few dozen people work on a single project.	11:58
	mina86 → mina86\|aw	11:58
Arafangion	Directly, that is.	11:58
saurabhverma1	Arafangion: yeah thats a good option too . I'll consider gitosis too , I was going through it yesterday	11:59
	← candybar left	12:00
	→ rraasch joined	12:01
	mina86\|aw → mina86	12:01
	→ joshsdc joined	12:04
	← jkacur left	12:06
	← sitaram left	12:07
	→ Julien__ joined	12:07
	← Julien__ left	12:08
	→ cannonball joined	12:08
	→ candybar joined	12:09
	← kssddy left	12:10
	→ Ryback_ joined	12:13
	← gpampara left	12:16
	→ WinterWeaver joined	12:18
	→ ph^ joined	12:20
	→ guitsaru joined	12:21
	→ corni_ joined	12:22
	→ name joined	12:22
	→ memiux joined	12:24
	← ijin left	12:24
	← memiux left	12:24
	← guitsaru left	12:25
	makies → makies_bot	12:26
	→ rudi_s joined	12:28
	→ jmou joined	12:28
	makies_bot → makies	12:28
	← Fullmoon left	12:29
	→ hummingkid joined	12:30
	← julm left	12:32
	→ ignas joined	12:32
	← saurabhverma1 left	12:33
	← girishr left	12:34
	→ corni__ joined	12:36
	← ph^_ left	12:37
	← candybar left	12:37
	← gitte left	12:38
	← jmou left	12:38
	← joshsdc left	12:38
	WebDragon → WebDragon\|away	12:38
	→ jmou joined	12:38
	← corni left	12:39
	→ M_J_G joined	12:41
	← aspotashev\|eeepc left	12:41
	kobani → leehambley	12:41
	← pulpfiction left	12:41
	→ cilly joined	12:42
	← khmarbaise_ left	12:42
	→ Fullmoon joined	12:42
	corni__ → corni	12:42
	makies → makies_bot	12:44
	→ khmarbaise joined	12:45
	→ Grahack joined	12:46
	← kpreid_ left	12:46
	→ kpreid_ joined	12:47
	→ shadowhand joined	12:48
	→ saurabhverma joined	12:49
	← jefferai left	12:50
	→ jefferai joined	12:50
	M_J_G → gitigit	12:51
	gitigit → M_J_G	12:51
	← corni_ left	12:53
	→ corni_ joined	12:53
	→ scarabx joined	12:53
	→ lorandi joined	12:54
	→ DaveIngram joined	12:55
	→ xinming_ joined	12:55
xinming_	How do I do a svn copy for a directory within git-svn?	12:56
	→ mw joined	12:56
	← WinterDumb left	12:56
	→ Julien___ joined	12:56
	→ WinterDumb joined	12:56
	← Julien___ left	12:58
	← rindolf left	12:59
	→ qrush joined	13:00
	WebDragon\|away → WebDragon	13:01
	→ donvinzk joined	13:02
	← eddyp left	13:02
donvinzk	hello	13:02
drizzd_	xinming_: I don't think you can. Maybe git-svn detects that after the fact, but I doubt it.	13:03
	→ Bass2 joined	13:04
drizzd_	xinming_: the same goes for svn rename	13:04
xinming_	drizzd_: Ok, Since I don't like the branch feature within git-svn, I'd really like to totally checkout the repo, and do copy myself there.	13:04
doener	on a file-basis, git-svn at least correctly translates renames, not sure about copies	13:04
	← wsc left	13:05
xinming_	doener: I mean the directory copy. :-)	13:05
	← Pe3k left	13:05
xinming_	Now, What I do is copy within svn, and goes to the git-svn repository, and edit there. >_<	13:05
donvinzk	hi, everytime I apply modifications on files, I have to git add them manually before commiting, because git status reports me that they have changed but are not tracked. Why is that ?	13:06
drizzd_	donvinzk: git status only reports new files as untracked, not changed files	13:07
doener	xinming_: no, it doesn't do copy detection on a directory basis, just tried	13:07
xinming_	doener: So, the safe bet should be something like svn copy, and then git-svn >_<	13:08
donvinzk	it does on my machine	13:08
	I mean no	13:08
	it reports them as changed, but not tracked. The untracked is yet another subcategory in a git status	13:08
	→ boto joined	13:09
	← saurabhverma left	13:09
drizzd_	donvinzk: if you want to add all changed files, use git add -u, or git commit -a	13:09
donvinzk	ok sorry, it says "Changed but not updated", my mistake	13:09
	← corni left	13:11
drizzd_	I usually do git add -u, workwork, git diff, git add -u, workwork, git add -u, git diff --cached, git commit	13:11
donvinzk	right, the git add -u works, but still, I thought I didn't had to do git -u when I started using git	13:13
	→ _Vi joined	13:13
drizzd_	donvinzk: you don't have to if you use git commit -a, which is equivalent to git add -u; git commit	13:14
	→ hydrogen joined	13:14
	→ ericindc joined	13:15
	→ ben_h joined	13:15
	→ LiamH joined	13:15
	← scarabx left	13:16
	→ ben_h_ joined	13:16
	→ memiux joined	13:16
	→ corni__ joined	13:16
	→ joshsdc joined	13:17
	→ Cromulent joined	13:18
	→ kukks joined	13:20
	→ candybar joined	13:21
	← ben_h left	13:21
	← WebDragon left	13:22
	← WinterWeaver left	13:23
	← donvinzk left	13:23
	→ smerp joined	13:26
	← shadowhand left	13:27
	→ saurabhverma joined	13:31
	← FunkeeMonk left	13:31
	→ corni joined	13:32
	← corni_ left	13:33
	→ jrmuizel joined	13:33
	← ereslibre left	13:34
	→ flazz joined	13:34
	← Fullmoon left	13:35
	→ Fullmoon joined	13:35
	← Blastur left	13:35
	→ flazz_ joined	13:35
	→ samphippen joined	13:35
samphippen	http://dpaste.com/58712/ what does this mean?	13:36
	← jrmuizel left	13:36
	→ nurey joined	13:37
	← nurey left	13:38
jester2	Newbie question: I created a new branch, worked on it, committed it, and pushed the commits to my upstream server. Then from a different computer, I did a "git pull". On this computer, "git branch" still shows only the master, though "git branch -a" does show remotes/origin/my-new-branch. How do I get this branch onto this computer so that I can switch into it and edit? And is there something like "git pull all" that will just grab everything from the upstream	13:39
_Vi	jester2, May be try "git checkout origin/that_branch -b that_branch" on second computer?	13:41
jester2	Yup, that works.	13:42
	Thanks. But is there any way to do things like this automatically? That is, I don't know what branches someone might have created, and I'd like to just be able to have everything show up...	13:43
	← jmou left	13:43
	→ jmou joined	13:43
	← rraasch left	13:45
	→ ctusar joined	13:46
	← BigMack83 left	13:46
	← DavidKlein left	13:46
	jmou → julm	13:46
	← khmarbaise left	13:46
_Vi	jester2, When you do "git pull" it shows you them.	13:47
	← ankit9 left	13:48
_Vi	jester2, You can also do "git show-ref" to enumerate branches and tags.	13:48
	→ hobodave joined	13:48
	← corni__ left	13:48
jester2	_Vi: true, but I guess I'd just like them to _be_ there--for my local repo to just contain everything in the main one.	13:49
	I'll look into this more, though--I'm new to branching in general (too much effort on CVS), so I'm learning... Thanks	13:49
w\|shimmy	jester2: whenever you fetch from a remote you automatically get all of the branches	13:49
	jester2: you just don't have a local branch that corresponds to it yet	13:49
	jester2: you can git checkout origin/branch if you'd like just fine	13:49
jester2	OK. I guess I need to read up more on the whole concept of local stuff.	13:50
	Thank you.	13:50
w\|shimmy	jester2: just in a situation called a 'detached HEAD' - so if you commit while on that branch, you aren't committing it to any branch, since origin/branch is not local. if you plan to edit, go with a git checkout -t origin/branch or similar.	13:51
	← hydrogen left	13:51
	→ bobmcw joined	13:51
	← flazz left	13:51
	← psankar left	13:54
	→ nurey joined	13:56
	→ Julien___ joined	13:57
	→ bdiego joined	13:57
	Julien___ → Juju	13:58
	← Juju left	13:58
	← xinming_ left	13:59
	flazz_ → flazz	14:00
	→ sitaram joined	14:01
	← DrAk0 left	14:02
	← ericindc left	14:02
	→ sfwc joined	14:02
	← sitaram left	14:05
	← Fullmoon left	14:05
	→ Jujulien joined	14:08
	Jujulien → Jule	14:08
	← Arafangion left	14:08
shruggar	I've just accidentally committed to a detached HEAD. I don't care about what I committed (the work exists elsewhere), how can I just make sure it won't be sitting there taking up space forever, and will be garbage-collected or something?	14:09
	→ deanh joined	14:09
	Jule → Juliano	14:09
	← janmejay left	14:10
	← Juliano left	14:10
Ilari	shruggar: Check out some other branch to retach HEAD. It will be gc'd eventually.	14:11
Pieter	shruggar: do you really care about those 10KB or something?	14:11
	if you want to GC now, you'll have to checkout another branch and remove all your reflogs	14:11
	→ alezandro joined	14:16
	→ Yuuhi joined	14:17
	← hobodave left	14:17
	← memiux left	14:18
	← gaveen left	14:20
	→ withanx joined	14:20
withanx	there isn't a way to recursively update submodules?	14:20
	ben_h_ → ben_h	14:21
	→ gaveen joined	14:21
shruggar	afaik submodule support is very basic and somewhat broken in terms of UI	14:21
withanx	eh that's annoying	14:22
bobmcw	I anti-submoduled my project, and life got much better	14:22
	submodules are just pain	14:22
	← nurey left	14:22
withanx	they would be useful if they worked =/	14:23
offby1	I use 'em a little, and they're fine	14:23
shruggar	if submodules in git were as transparent and painless as they are in SVN, I'd call them useful, but they are somehow worse :)	14:24
offby1	they did take a while to figure out	14:24
	→ nurey joined	14:24
	→ Julien__ joined	14:24
w\|shimmy	I think they're actually pretty intuitive if you're developing with them. A pain for people who are checking out your source though.	14:25
	→ jrmuizel joined	14:25
	← Julien__ left	14:25
	← julm left	14:28
	→ julm joined	14:28
	→ JasonWoof joined	14:29
	→ tvw joined	14:30
	→ stephenjudkins joined	14:30
	← murph_ left	14:31
	→ whaley_ joined	14:31
bobmcw	w\|shimmy: I found them obnoxious as developer	14:32
	folks forgetting to update the root project to update to new submodule HEADs	14:32
	tagging being painful upon releases	14:32
	ie, submoduleA#HEAD was correct, but root#HEAD was pointing to submoduleA~15 still	14:32
	→ Fullmoon joined	14:34
	← marcelo left	14:35
	→ suman joined	14:35
	sebleier\|away → sebleier	14:35
	→ whaley- joined	14:36
suman	hi all... i have a remote branches and i want create another remote branch which branches off from a remote branch... what is the best way?	14:36
	← whaley_ left	14:36
w\|shimmy	suman: you want the branch to be on the remote?	14:36
suman	yeah	14:36
	i want to create another branch on the remote, branching off from a branch in the remote	14:37
	i know a couple of ways.. but am looking for the best way	14:37
	is there a direct way to create remote branch off another remote branch	14:37
w\|shimmy	suman: I suppose, git checkout -b new-branch remote/branch && git push remote new-branch seems straightforward enough, but I've not seen/used something like that before	14:38
suman	w\|shimmy: but the remote branch needs to be created rite.. do you thing 'git push remote new-branch' createa new branch, if one doesnot exist??	14:39
	→ lymeca joined	14:40
	← roop left	14:40
w\|shimmy	suman: yes, of course	14:41
suman	alrite... it does..	14:41
	does the new branch on the local machine track a new branch on the remote machine??	14:42
	→ shadowhand joined	14:42
	→ jrmuizel_ joined	14:42
	← ph^ left	14:44
	→ memiux joined	14:45
coldboot	If I have commits like this: 1-2-3-4-5-x, what's the easiest way to get 1-2-3-x-4-5 ?	14:47
	← ben_h left	14:47
offby1	rebase -i!	14:48
	"git rebase -i 1", for some value of "1".	14:48
M_J_G	coldboot: I guess you rebase -i 2, specify edit for 3 and make a new commit instead of amending 3.	14:48
offby1	opens an editor showing those commits ... just take the line for X, and move it up two lines.	14:48
	→ ben_h joined	14:48
offby1	Save and exit.	14:48
	Voilà.	14:48
coldboot	Woah, cool	14:49
	→ Vinnie joined	14:49
offby1	M_J_G: I don't think he wants to _edit_ the commit; I think he just wants to move it around.	14:49
coldboot	Yeah just moving it around.	14:49
	So what's the '1' for?	14:49
M_J_G	Yes, I thought he doesn't have x yet -my bad.	14:49
	→ mattikus joined	14:49
	→ nothingHappens joined	14:49
metellius	so I did my work in a local topic branch that was branched off a remote branch. what is the best practice for merging this back now? I'd rather like to avoid keeping a local copy of the parent branch, and somehow just merge the topic branch back and then end up with no local branches at all	14:50
	→ Octalot_ joined	14:50
	→ donvinzk joined	14:50
M_J_G	coldboot: '1' is the name of the ancestor of the first commit that you want to rewrite. Could be 3 here ;)	14:51
	← Cromulent left	14:51
	→ rlorandi joined	14:51
coldboot	M_J_G: Thanks	14:51
	Oh, the new "base" that you're rebasing to, of course.	14:52
	→ corybantic joined	14:52
	← aser left	14:52
donvinzk	hi, I did a git reset to revert changes on my local repo, and would like the remote to reflect this. However, right now, it is still pointing to the old location, which prevent me to fast forward merges. Graphical description here: http://imagebin.org/53423 . Could you help me revert the remote too ?	14:53
	← jrmuizel left	14:53
	→ altrux joined	14:53
offby1	metellius: check out (a local branch based on) the parent; "git merge your-local-topic-branch".	14:54
	Then "git push", or send patches, or whatever it is you do to get your work upstream.	14:54
coldboot	Can you un-add files?	14:54
	Without deleting them.	14:55
offby1	donvinzk: so you want to essentially destroy the commit labeled "Added dummy IObject for dev version"?	14:55
M_J_G	coldboot: git rm --cached file	14:55
knitt1	good afternoon	14:55
offby1	<dracula>Good Evening</dracula>	14:56
	← Fullmoon left	14:56
knitt1	i want to import an svn repo to git. what i'm not sure about is this: the toplevel structure is btT + 3 other folders. how can i import that intelligently?	14:56
	← agile left	14:57
suman	thanks w\|shimmy	14:57
offby1	knitt1: hmm.	14:57
	→ bugfux joined	14:57
	← _Vi left	14:57
donvinzk	offby1: yes, the remote/origin one. And I want the remote/master to point to 'git command added'	14:57
coldboot	Are the git developers working on making git faster for huge (# of files) repositories?	14:58
offby1	donvinzk: ok, I think you can do "git push origin +refs/heads/foamDev:remotes/origin/foamDev"	14:58
	the + is key	14:58
knitt1	coldboot: i think the bottleneck is hdd speed here	14:58
offby1	also, know that if that works, people who've already pulled that commit will get merge conflicts and/or other unpleasantness.	14:59
	They'll have to "force" update their local branches.	14:59
knitt1	do i have to import the folders other than btT manually into their own branches?	14:59
offby1	knitt1: I suspect so.	14:59
	but I doubt it's hard.	14:59
coldboot	Argh! git rm --cached removed my files, git status says they're deleted, I just wanted to un-add them from the index after I added them.	14:59
knitt1	if i have problems, i'll just come back :)	14:59
	→ kishore_ joined	15:00
M_J_G	coldboot: ls will tell you they're not deleted.	15:00
coldboot	M_J_G: I know that, but they're to be deleted from the repository.	15:00
knitt1	coldboot: git reset -- file	15:01
	← lorandi left	15:01
M_J_G	coldboot: So you only want to un-add the recently staged changes, not the whole file? knitt1 just answered that.	15:01
	→ robinsmidsrod joined	15:02
coldboot	Okay thanks	15:02
kishore_	I have a git project and in one of its branch i have added a submodule, but the folder of the submodule remains even when I checkout master again	15:02
	Is there some whay to avoid that?	15:02
	whaley- → whaley	15:03
kishore_	I would want that folder to disappear when I go to master as it does not belong there	15:03
	← julm left	15:03
uau	kishore_: IIRC not - git's handling of submodules is rather limited	15:03
	→ julm joined	15:03
kishore_	Right now, it is only clutter but i could imagine that it could mess with some makefiles that do file/folder globbing	15:04
	← wxr483 left	15:04
kishore_	uau: I know.. but is there a way to work with it? or is subtree merge the way to go?	15:04
	← psoo left	15:04
	← orospakr left	15:04
w\|shimmy	kishore_: when you checkout master, git just sees the folder as an untracked (empty) folder afaik	15:05
uau	"way to work with it" depends on what you need - git's functionality is limited, whether it's just too bad or still usable depends on your requirements	15:06
M_J_G	kishore_: You can try git clean	15:06
w\|shimmy	kishore_: ie, just delete it on master and git won't even notice it	15:07
	kishore_: or git clean, as M_J_G said. probably a safer move, in case I'm full of shit.	15:07
uau	however if you delete the submodule when switching to another branch then you need to check it out again when returning	15:07
coldboot	knitt1: Hard drive speed is the bottlekneck because it has to hash every file?	15:08
kishore_	uau: the folders are empty and yes i need to check them out again which i cant do now as that remote is not accessible from here at home for me	15:09
	→ hobodave joined	15:09
kishore_	so submodule is a bit limted it seems	15:09
uau	why are the folders empty?	15:09
	← Vinnie left	15:10
	← Grahack left	15:10
kishore_	i might have better luck with the subtree	15:10
uau	git shouldn't empty them even when switching to a branch without the submodule IIRC	15:10
w\|shimmy	afaik, git doesn't touch the submodule directory at all	15:11
	→ saurabhverma1 joined	15:11
uau	rather it'd leave the submodule checkout there even if it shouldn't be in the next branch	15:11
kishore_	uau: I cloned a local copy of the repo to experiment and it seems that is the issue as the remote server at office needed by the submodules is not available from here at home	15:11
uau	submodules are never initialized by normal clone (even if available)	15:11
	→ Fullmoon joined	15:12
uau	you always need to manually initialize them	15:12
kishore_	uau: Yes the folders are not emptied when i checnge branches	15:12
	→ mm_202 joined	15:13
kishore_	I think i would have better and true "distributed" behaviour with using subtrees i guess	15:13
	Is there a good guide for subtree merge? A tutorial for dummies? :D	15:14
	→ saurabhverma2 joined	15:15
	← vhallac left	15:15
uau	the 'subtree merge' is a rather limited special tool only	15:17
	← solofight left	15:17
uau	subtrees behave much better in checkouts, but i wouldn't call them 'true distributed behavior'	15:18
	i don't think they're suited for actual distributed development	15:18
	merlin1983 → merlin83_	15:18
	hiffy_ → hiffy	15:19
uau	(unless the subtree is the canonical copy)	15:19
mm_202	Hey guys, Im sure you'll tell me to RTFM or the changelogs, but could someone just quickly tell me what the major differences between git 1.6.0 and git 1.6.3 are?	15:19
kishore_	but with submodules, i cannot share my super project branch with my colleagues without much work	15:19
	like sharing each of the submodules and getting them to be cloned manually	15:19
uau	yes in practice embedding a copy of the other repository works much better for checkouts than git submodules	15:20
	→ saurabhverma3 joined	15:20
kishore_	ie while i am away from the sentral server	15:20
uau	but such embedded copies are clumsy too if you need to do any development affecting them	15:21
kishore_	uau: what then is the down side of the subtree merge? keeping it updated?	15:21
uau	and not at most occasionally update them to the latest version of the "upstream" module	15:21
	as i said above the "subtree merge" is just a limited special tool	15:21
kishore_	uau: Yeah i certainly want to be able to modify the subproject and send those changes to the original project too	15:22
	→ mkemp joined	15:22
uau	what you actually do is add a copy of the software somewhere in the tree (and perhaps try to update it)	15:22
	you can try to use the subtree merge for those tasks, but it's just a detail	15:23
	→ krawek joined	15:23
	→ DavidKlein joined	15:23
	→ knittl joined	15:23
kishore_	uau: I dont understand? detail?	15:23
	← ia left	15:24
uau	don't understand what?	15:24
kishore_	"but it's just a detail"	15:24
	you mean it is not too easy to manage?	15:24
	AAA_awright_ → AAA_awright	15:25
uau	i mean the "subtree merge" operation in git is just one (rather limited) tool that can perhaps help in doing the actual task	15:25
	← eldragon left	15:25
	→ ia joined	15:25
	← mkemp left	15:25
uau	so describing the actual alternative itself as "subtree merge" is misleading IMO	15:25
kishore_	Ah ok. So there are more opeartation to do to manage it well	15:25
	→ mkemp joined	15:25
kishore_	Can the detail be scripted?	15:26
	→ eldragon joined	15:26
kishore_	Maybe i can look through that detail if i first learn how to manage with subtree merge	15:26
	is there a tutorial?	15:27
	→ orospakr joined	15:27
	← donvinzk left	15:28
	← saurabhverma left	15:28
	→ tbf_ joined	15:29
	← shadowhand left	15:31
	← cilly left	15:32
	← saurabhverma1 left	15:32
hi3	what's a good way to commit current changes to a new branch?	15:33
drizzd_	hi3: git checkout -b my-new-branch	15:34
hi3	thanks drizzd.	15:35
	← knittl left	15:35
	← willb left	15:35
	→ _Vi joined	15:35
drizzd_	kishore_: yes, there is a tutorial	15:36
	kishore_: it's on kernel.org I believe	15:36
	← saurabhverma2 left	15:36
drizzd_	search "git subtree merge" and press "I'm feeling lucky"	15:36
	← mattikus left	15:37
	← tbf left	15:37
kishore_	drizzd_: Thanks. I looking through http://www.kernel.org/pub/software/scm/git/docs/howto/using-merge-subtree.html	15:37
drizzd_	that's the one	15:37
	← alvarezp left	15:37
	→ WinterWeaver joined	15:38
	→ lbt joined	15:38
	→ reprore_ joined	15:39
	→ apeiros_ joined	15:39
	← whaley left	15:41
	→ lighthead joined	15:41
	→ BlindOracle joined	15:42
	← vshriram_ left	15:42
	→ whaley joined	15:42
	→ agile joined	15:42
	← Moult left	15:43
	→ girishr joined	15:44
	→ cemerick joined	15:44
	→ Moult joined	15:44
	← girishr left	15:45
	← stephenjudkins left	15:45
drizzd_	kishore_: there is also git-subtree, which helps you with that. It's not in mainline git (yet) though: http://github.com/apenwarr/git-subtree	15:46
	sebleier → sebleier\|away	15:46
	← nurey left	15:46
	→ Cromulent joined	15:46
	← krynnotaur left	15:47
	→ Krynnotaur joined	15:47
	→ Chillance joined	15:47
	← tango_ left	15:48
	rylar_off → rylar	15:49
	sebleier\|away → sebleier	15:49
	← WinterWeaver left	15:50
	→ ericindc joined	15:50
kishore_	drizzd_: i stumbled upon that post and it was interesting. Any clues if it is likely to be merged with mainline?	15:51
	← julm left	15:52
kishore_	I dont want to end up supporting the tool for ever once i introduce it to my colleagues! :)	15:52
drizzd_	kishore_: I'm not really sure. IIRC there was not much response. It certainly makes handling subtree merges easier, but it doesn't really solve the problem submodules try to solve either	15:52
	→ julm joined	15:52
drizzd_	(which is that you may not want the entire project in one tree)	15:52
	← whaley left	15:53
	→ voker57 joined	15:53
	← ppawel left	15:54
	← gaveen left	15:54
drizzd_	otherwise, I think the tool should be fairly safe to use because it only builds on basic git functionality. You can achieve the same thing using git merge -s subtree and filter-branch manually.	15:54
	→ rafael_lorandi joined	15:54
	→ carrus85 joined	15:55
kishore_	drizzd_: ok	15:55
	How i wish git mainline tried addressing this issue :(	15:56
	← stuffcorpse left	15:56
	→ stuffcorpse joined	15:56
M_J_G	kishore_: Describing the actual issue concisely (on the mailing list) might a first step ;)	15:57
	→ krh joined	15:57
drizzd_	kishore_: the perfect solution would be narrow and sparse clone support: http://git.or.cz/gitwiki/SoC2009Ideas#NarrowandSparseclonesupport	15:57
M_J_G	kishore_: I.e.: What's the problem, where do subtree and submodule go "wrong"?	15:58
drizzd_	there have been a few attempts/approaches	15:58
	→ CardinalN joined	15:58
kishore_	M_J_G: I sure would but it seems to me that im not the first to feel so and others have expressed this	15:58
drizzd_	The above would render submodules entirely obsolete	15:59
kishore_	M_J_G: But i'll try anyway and maybe add to the momentum	15:59
	← marc_kdab left	15:59
	→ marc_kdab joined	16:00
	← leehambley left	16:01
	→ sitaram joined	16:01
	→ WinterWeaver joined	16:01
	← rlorandi left	16:01
	→ lightcap joined	16:01
	← Cromulent left	16:01
	← sunoano left	16:02
lightcap	So, I have a branch that I'm working on on my laptop and I need to push it to my desktop to continue working on it while the laptop goes off for repairs. How do I push a new branch to the desktop?	16:02
	→ lavamind joined	16:03
lavamind	hello, using Git, is it possible to do a merge that does everything except merge the history from the remote branch into the current branch?	16:03
	← giallu left	16:03
_Vi	lavamind, Yes, it is.	16:03
Pieter	git merge --squash	16:03
kishore_	drizzd_: how would that eliminate submodules? for that it would need to be able to maintain the blobs along with the super project blobs	16:04
sitaram	lightcap: do you already have a repo on the desktop? if so: git push ssh:you@deskop:/path/to/repo localbranch:branch-from-laptop	16:04
_Vi	lavamind, Yes, git merge --squash.	16:04
lavamind	Pieter, I find that --squash doesn't actually commit the changes, using it I would have to specify a log message manually, I'd like it to be automatic	16:04
lightcap	sitaram: yeah, I do. though it doesn't have the "orders" branch...	16:05
	← saurabhverma3 left	16:05
_Vi	kishore_, Use subtree merge strategy?	16:05
drizzd_	kishore_: you don't need any subprojects any more. You put everything into one project and if you want to work on a subdirectory you simply checkout the subdirectory and nothing else.	16:05
	← reprore_ left	16:05
sitaram	lightcap: so just do what I said; I was a little verbose in naming stuff to make it clear but feel free to ask for details	16:05
	← M_J_G left	16:06
drizzd_	checkout/clone, etc. It would be just like CVS in that respect.	16:06
	← naeu left	16:06
	→ reprore joined	16:06
	→ shadowhand joined	16:06
	← voker57__ left	16:06
	← lbt left	16:07
	← lighthead left	16:07
	← Octalot left	16:07
lightcap	sitaram: I thought that would work too; I tried it earlier. but I get error: Bad port ''	16:07
	even though I can ssh with no problem to that machine	16:07
kishore_	drizzd_: So yes i will be able to checkout a part of the repo but that still does not provide for something like svn:external when ProjectA depends on ProjectB	16:07
	← ntoll left	16:07
lightcap	oops	16:08
	the ssh url was bad. wierd.	16:08
	→ galderz joined	16:08
	→ gaveen joined	16:08
sitaram	lightcap: bad port means your ssh is not working; you'll have to get that going before worrying about git	16:08
lightcap	thanks sitaram, evidently it doesn't like a colon before the path	16:08
	no, it works, but the url ssh://host:/path/to/repo suggests a port :-)	16:09
	→ spearce joined	16:09
sitaram	lightcap: man git-clone has lots of examples; I can never remember the syntax and often look there	16:09
Gitbot	lightcap: the git-clone manpage can be found at http://git.or.cz/man/git-clone	16:09
lightcap	thanks again, it's working now.	16:09
sitaram	lightcap: also, do you know about "remotes"?	16:09
lightcap	yeah, I use github mostly :-)	16:10
	except for these weird edge cases.	16:10
sitaram	lightcap: so just set a remote for your desktop once; never worry about remembering the syntax	16:10
	← Fullmoon left	16:11
sitaram	I'm a great fan of putting all user+host+port in ~/.ssh/config, and then using the remote URL just for the actual repo path	16:11
Pieter	I'm a huge fan of hazelnuts	16:11
kishore_	drizzd_: narrow and sparse clones do help to keep the checkout of only those files that are needed but for true distributed nature i would imagine that somehow the blobs of the subproject should be alongside the superproject blobs	16:11
	← ProperNoun left	16:11
drizzd_	kishore_: I don't know what svn:external does. But if you simply want to add project A to project B, then subtree merge is the perfect solution.	16:12
kishore_	and intelligently cloning out the right checkout in the right path as needed when changing branches in the superproject	16:12
	← sgrimm left	16:12
lavamind	how can I do a git merge --squash that still commits the changes automatically after the merge	16:13
	← galderz left	16:13
lavamind	using git merge --squash --commit doesn't actually commit	16:13
drizzd_	kishore_: I guess I don't understand the problem you're having after all.	16:13
	← name left	16:13
	→ MisterN joined	16:14
kishore_	drizzd_: subtree seems to be the best way for me currently but i could only hope there was something better	16:14
	← dmlloyd left	16:14
kishore_	drizzd_: shall i explain?	16:14
lightcap	hmm. confused again. So I just committed the changes on the "order" branch. Then I pushed to my desktop "order" branch and a git status on the desktop shows the same files I just committed as needing to be committed?	16:16
Ilari	lightcap: $faq non-bare	16:16
Gitbot	lightcap: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare	16:16
lightcap	ah. well shit.	16:17
kishore_	Lets say i am working on ProjectA in the master branch and i now want to add feature to create a featureX branch...	16:17
Ilari	Why would narrow clone obsolete submodules? Submodules can be used for tree-to-commit linkage too.	16:17
	→ Fullmoon_ joined	16:17
lightcap	Ilari: well, what should I have done then to share the changes? pull?	16:18
	← doener left	16:18
kishore_	now featureX depends on a certain library (libraryForX) that is another project	16:18
	→ doener joined	16:18
drizzd_	kishore_: I'm a little busy right now, so it may take a while for me to respond.	16:18
kishore_	drizzd_: thats fine. Should go on or just leave it at this? I could send a mail to the lists explaining the limitations of subtree and submodule	16:19
	drizzd_: thanks for your time though :)	16:19
Ilari	lightcap: Or seperate bare repo and push to that.	16:20
	→ sunoano joined	16:20
	← _Vi left	16:20
lightcap	ok, how come I can always push to github though, isn't that essentially what I was trying to do (but going to my own server, not github)?	16:21
tomoj	lightcap: github has a bare repo	16:21
gebi	some form of tracking submodule would be nice, not tied to a specific sha1d but e.g to master/HEAD of the repo	16:22
	→ ianmcorvidae joined	16:22
drizzd_	kishore_: if you have an interesting problem, I'm sure you will get responses on the list.	16:22
lightcap	okay, I'm clearly lost on terminology.. time to google :-)	16:22
	← asksol_ left	16:22
	← ciskje left	16:23
	→ kuadrosx joined	16:26
	← rafael_lorandi left	16:26
	→ lorandi joined	16:27
	← sitaram left	16:27
	← ericindc left	16:27
	→ dmlloyd joined	16:28
	→ stephenjudkins joined	16:29
	→ kn1ttl joined	16:29
	→ cmarcelo joined	16:29
	← boto left	16:30
	→ Alinermedia joined	16:30
	→ Cromulent joined	16:32
	→ willb joined	16:33
	← kishore_ left	16:34
	← shruggar left	16:34
	← shadowhand left	16:34
	→ bleything joined	16:35
	← Juan__ left	16:35
	→ machrider joined	16:35
	← WinterWeaver left	16:36
machrider	is there any way to customize the behavior of the 'clone' operation that is done via git submodule add/update?	16:36
	for example, clone supports a --depth option, can i somehow get that passed in?	16:37
	← Fullmoon_ left	16:38
	← ianmcorvidae\|alt left	16:39
	→ icwiener joined	16:43
	← Chillance left	16:43
	← knitt1 left	16:45
	→ ericindc joined	16:47
samphippen	can git ignore the local repository in any way?	16:49
	like svn commit	16:49
wereHamster	samphippen: no. why would you want to do that?	16:49
samphippen	to reduce the number of steps required to get code onto the internet	16:50
	→ WinterWeaver joined	16:50
	← marc_kdab left	16:51
wereHamster	oh come on, typing in 'git push' once a while isn't that hard, is it?	16:51
samphippen	git commit -m and git push is two steps though	16:51
	my friend is complaining that svn has less steps	16:51
	I was wondering if there was a way to reduce the number of steps	16:51
wereHamster	usually you do more then one commit before you pus	16:51
tomoj	two steps is better than one in this case	16:51
	→ __iron joined	16:52
	→ cpg joined	16:53
drizzd_	samphippen: you can certainly introduce a new command which does "git commit && git push", but that won't solve your problem, because a push can be rejected if something else was committed in between	16:53
	in which case you have to merge	16:53
	→ Derander joined	16:54
	← aziz left	16:55
cardioid	samphippen: Tell your friend that git is not svn	16:56
	← ericindc left	16:56
	→ mattikus joined	16:56
	← mkemp left	16:56
	→ DrAk0 joined	16:58
drizzd_	well, he has a point	16:58
	if you have a really simple development scheme, which goes like	16:58
	edit; commit; edit; commit	16:58
	with some parallel development, but very modularized, so there are never any conflicts or even dependencies	16:59
	then you will have to do some more work with git with little gain	16:59
	you have edit; commit; push; -> rejected, oh no, git pull; git push; edit; commit; ...	16:59
	← bugfux left	17:00
	→ qhoxie joined	17:00
	→ nurey joined	17:01
drizzd_	samphippen: so I suppose svn commit in git would be "git commit -a && git pull && git push" -- something like that	17:01
	← reprore left	17:01
	← ceyusa left	17:01
drizzd_	although if you do that, I would recommend using git pull --rebase instead	17:02
	→ reprore_ joined	17:02
Ilari	Actually, its more conservative in merging (like git checkout without -m). But still can result something broken.	17:02
	→ icwiener_ joined	17:04
	→ texel joined	17:04
drizzd_	Ilari: what do you mean? you can always abort if rebase goes wrong	17:05
Ilari	drizzd_: You can't abort SVN commit after the automatic merge results something really broken.	17:06
drizzd_	Otherwise I always considered rebase to do pretty much the same as a merge, especially if there is only one commit involved downstream.	17:06
	Ilari: oh, right. I didn't understand what you meant with "it" :-)	17:07
smoofra	so what are junios diffirent branches? (next, maint, pu)?	17:07
drizzd_	there should be some documentation in the todo branch	17:08
	← Alinermedia left	17:08
drizzd_	smoofra: git show todo:MaintNotes	17:09
	git show origin/todo:MaintNotes, rather	17:09
	but the short version is maint: bugfixes, master: features for the current release, next: features for the next release, pu: everything else	17:10
	→ bobesponja joined	17:11
	→ whaley joined	17:12
	→ naeu joined	17:16
	← icwiener left	17:18
	← reprore_ left	17:18
	→ reprore joined	17:19
	→ aziz joined	17:21
	← DrAk0 left	17:22
lavamind	Anyone using Gitorious care to comment about it? I'm using Github right now, but moving to an Floss-based service is appealing	17:22
	← Cromulent left	17:23
JasonWoof	gitorious is great	17:23
	I've only done basic stuff so far, but I've been quite happy	17:24
	→ AkumaStreak joined	17:24
JasonWoof	they added most of the bling people were wining	17:24
	and they seem responsive to feedback	17:24
	← lohrun left	17:24
lavamind	yeah I see they offer a project wiki	17:24
JasonWoof	I sent in a patch to display the year in dates on gitorious.org and it was merged within a day, and the website was updated in about a week	17:25
	→ tanuki joined	17:25
	← cedricv left	17:26
lavamind	interesting	17:26
tanuki	I have stashed changes that I want to reapply. However, there are some (trivial) conflicts in some of the files; I want to have the existing version take precedence in the merge rather than conflicting. Is there a way to do this easily, or do I have to manually resolve the conflicts?	17:26
lavamind	what do you think about the fast the gitorious uses wrappers instead of a full implementation	17:26
JasonWoof	I like the presentation of projects on gitorious much better. there's clear links to the main author, website, wiki, etc	17:26
lavamind	(as github)	17:26
	→ cedricv joined	17:27
JasonWoof	huh?	17:27
	← ignas left	17:27
	→ sakyamuni joined	17:27
lavamind	« There's also the fact that they rewrote the Git server in Erlang and the Git client in Ruby, which gives them much more stability than the typical commandline wrappers that other Rails-based Git repository interfaces (including Gitorious) use. »	17:28
	taken from http://stackoverflow.com/questions/78991/why-is-github-more-popular-than-gitorious	17:29
	I'm not sure what's meant by "stability" though, codebase stability or runtime stability ... ?	17:29
JasonWoof	no idea	17:30
	never heard of gitorious crashing or anything	17:30
	← stephenjudkins left	17:30
JasonWoof	I don't see how it really matters what languages were used to implement github, since they won't release the code	17:31
tomoj	I think the point was just that they wrote custom shit to deal with the large volume	17:32
rektide	they've learned that posting about how awesome the github server code is doesnt earn them any medals	17:32
	took em long enough	17:32
	→ cloudhead joined	17:34
JasonWoof	I'd be happy to get all woogly googly about the code if they release it under the agpl or something	17:34
lavamind	right, it's true any claim Github could make about its closed server software is basically unverifiable...	17:34
	→ cilly joined	17:35
	← whaley left	17:35
	← harinath left	17:36
JasonWoof	gitorious looks nice, works well, isn't so cluttered, and is responsive to feature requests (well, at least feature requests with nice patches anyway)	17:36
smoofra	projects that aren't responsive to feature requests with nice patches seriously annoy me	17:37
	i wrote a few bugfixes and a nifty feature patch for sbcl a year ago	17:37
	and they pretty much just ignored me	17:37
	not even a "no, i don't like your patch because of X"	17:37
JasonWoof	bugger. least they can do is say "thank you for your patch, I think blah blah blah"	17:38
	→ WALoeIII joined	17:38
	→ tango_ joined	17:39
MisterN	lol, github and stable in one sentence?	17:39
wereHamster	MisterN: why? Have you had problems with github in the past?	17:40
MisterN	wereHamster: yeah it had some instability	17:40
	granted, it works most of the time, but unstable enough for me to be happy to also host the project on sourceforge	17:40
	← jaeckel left	17:40
MisterN	github's graphs are also constantly broken	17:40
	→ ankit9 joined	17:41
	→ montylounge joined	17:41
JasonWoof	gitorious is a little rough around the edges, but I think it has a nice feature set and good overall usability. So I'm happy to use it now. I trust it'll be fixed over time, especially now that they have trolltec invested in its success	17:41
	← deanh left	17:41
smoofra	i'll take rough around the edges and open source over polished software-as-a-service any day	17:42
	← ankit9 left	17:42
	→ ankit9 joined	17:42
smoofra	if only there was a open source replacement for google reader	17:43
	i'd be happy	17:43
	→ deanh joined	17:45
	→ tbf joined	17:45
	→ rvhia joined	17:45
	→ bugfux joined	17:46
	← tbf_ left	17:46
rvhia	how do i get a file at certain date?	17:46
lavamind	thanks for sharing your thoughts about Gitourious/Github, I think it's fairly obvious we should support Gitorious instead of paying for Github	17:46
	← ankit9 left	17:46
	→ ankit9 joined	17:47
tomoj	"we should"?	17:47
lavamind	take it as a must, if you wish :)	17:47
	← julm left	17:49
charon	rvhia: that's not particularly well defined in the face of nonlinear history; there may be many commits around the given time with many different versions of the file	17:49
	→ julm joined	17:49
	← nurey left	17:50
	→ bcarlyon\|laptop joined	17:51
	← glommer left	17:53
	bcarlyon\|laptop → BarryCarlyon	17:53
JasonWoof	rvhia: you might run gitk (make sure you edit the view so it's sorting by date) and scroll down to the date, and choose a commit. You can grab the sha1 for that commit, or tag it, and grab the file with the commandline	17:53
	→ boto joined	17:53
	← reprore left	17:54
	← DaveIngram left	17:55
	← hi3 left	17:56
rvhia	the issue i have now is that one file seems to screw up, i want to find out when it was changed	17:56
	in svn, i can just pull a certain date	17:56
	→ nurey joined	17:56
	← ankit9 left	17:57
	→ ankit9 joined	17:57
	→ whaley joined	17:58
	→ reprore_ joined	17:58
	← ankit9 left	17:58
	← jpieper left	17:58
	→ marc_m joined	17:59
	← vermila left	18:00
JasonWoof	rvhia: use gitk, you can have it hilight all commits that changed a certain file	18:00
	marc_m → marc_kdab	18:00
JasonWoof	also, I think you can do "git log [filename]" to get a log for just commits that changed that file	18:00
	→ k0001 joined	18:00
smoofra	rvhia: git rev-list --since='Jun 1' --until='Jun 15' master	18:00
JasonWoof	rvhia: or you can use git-bisect	18:01
smoofra	rvhia: that'll list the commits in the first half of june that are ancestors of master	18:01
	← jmesnil left	18:01
	→ kumbayo joined	18:01
	→ robinr joined	18:01
	← cilly left	18:07
	→ ghostrunners joined	18:09
	→ Stravy joined	18:11
	→ docwhat joined	18:14
ghostrunners	so im pushing a new repo onto gitosis over my lan (git push origin master:refs/heads/master), and my windows git.exe keeps crashing, how do I find out why?	18:15
	windows flames accepted.	18:15
charon	ghostrunners: GIT_TRACE=2 ?	18:16
Ilari	ghostrunners: Possible log messages? Messages from client side?	18:18
ghostrunners	ack, sorry fellers - dumb error on my part - i forgot to add and commit my files prior to pushing, doh!	18:19
	idi0t and windows, what a combo! :)	18:19
	rylar → rylar_off	18:19
Ilari	ghostrunners: But that shouldn't cause git.exe to crash...	18:20
	→ _Vi joined	18:20
	rylar_off → rylar	18:20
corni	lol	18:20
ghostrunners	Ilari: I'll try to recreate the error by pushing remotely without add/commit files first	18:22
	← reprore_ left	18:23
	→ reprore joined	18:23
	← tvw left	18:24
	← _Vi left	18:25
	← tono left	18:26
	→ bcarlyon\|laptop joined	18:27
ghostrunners	Ilari: ok, easily, recreated it - created directory on localhost, added some files, ran >git init, ran >git remote add origin git@somehost:someproject.git, then ran >git push origin master:refs/heads/master and it crashes on windows	18:27
	→ saurabhverma joined	18:27
	← reprore left	18:30
	→ mefesto joined	18:30
ghostrunners	I have a log file produced by the error, if you think it will help, i'll pastebin it for	18:30
	→ reprore joined	18:31
	← montylounge left	18:32
	← sakyamuni left	18:33
	← reprore left	18:33
	→ reprore_ joined	18:34
	→ _Vi joined	18:34
	rylar → rylar_off	18:34
	← lavamind left	18:35
	→ sdboyer joined	18:36
	→ designs703 joined	18:37
	← nurey left	18:37
	← julm left	18:38
designs703	How can I force a pull to write? I edited a config file for this particular app, and I want to pull and overwrite the changes I made to that file	18:38
	I get a fatal cannot merge error as a result right now	18:38
	→ kobani joined	18:39
	→ julm joined	18:39
	← ghostrunners left	18:40
uau	designs703: i'm not sure whether you can get the pull command to do that, but you can run 'git reset --hard' first to destroy any changes you have in the local tree (in all files, not just conflicting ones)	18:40
designs703	uau, sounds hot, thanks	18:40
smoofra	designs703: do you want to keep the history of the changes you made?	18:41
designs703	smoofra, no, I don't need them	18:41
	it was just because the dev had root with no password as the DB credentials	18:42
smoofra	yea then just git reset --hard to thow them out	18:42
designs703	got it	18:42
	→ johndoigiii joined	18:43
	← designs703 left	18:43
	← reprore_ left	18:44
	→ jonshea joined	18:45
	← julm left	18:45
	→ julm joined	18:46
	← docwhat left	18:46
	→ reprore_ joined	18:46
	← reprore_ left	18:46
	→ reprore joined	18:47
	→ hacim joined	18:48
hacim	if I do: git checkout --track -b werise origin/werise shouldn't that branch automatically pull from there?	18:48
	← BarryCarlyon left	18:49
	→ loincloth joined	18:49
	→ saurabhverma1 joined	18:50
_Vi	hacim, Yes, it probably should.	18:50
hacim	_Vi: thats why i'm confused, because when I do a git pull it complains, "You asked me to pull without telling me which branch you..."	18:50
_Vi	hacim, Is this branch currently checked out?	18:51
	→ DrAk0 joined	18:51
	→ justin-george joined	18:52
	→ uau joined	18:52
hacim	_Vi: yes, I'm currently in it	18:52
_Vi	hacim, What is in .git/config?	18:53
	← saurabhverma left	18:54
	← DrAk0 left	18:54
hacim	_Vi: remote = origin merge = refs/heads/werise	18:55
	→ camwest joined	18:56
	← bcarlyon\|laptop left	18:57
	→ BarryCarlyon joined	18:58
	→ maxb_ joined	18:58
_Vi	hacim, Is git recent (1.6) version?	19:00
	→ bx2 joined	19:02
	→ corni_ joined	19:04
	→ abizern joined	19:05
	← naeu left	19:06
	← julm left	19:07
	→ julm joined	19:07
	→ c0m1 joined	19:08
	← marc_kdab left	19:09
c0m1	i've been using git for a few weeks now. I heard it was possible to track a subversion repo from a git repo. Could someone give me rough idea of how that works before I go try to figure out the details	19:09
mefesto	c0m1: git-svn	19:09
_Vi	c0m1, If starting from scratch, it is "git svn clone [-s] URL".	19:09
c0m1	_Vi: and if I cloned it, then I could push my master branch back to the subversion repository?	19:11
	← johndoigiii left	19:11
_Vi	c0m1, Yes, you may push (dcommit) current branch back.	19:12
	CardinalN → ProperNoun	19:12
	← dasil003 left	19:12
_Vi	c0m1, But your svn branch will be rebased every pull from (and push to) svn.	19:13
	→ btanaka joined	19:13
c0m1	_Vi: rebased?	19:13
	→ jnareb joined	19:13
_Vi	c0m1, Yes, recommended way of working with git-svn is using "git svn rebase" to fetch changes and 'git svn dcommit' to publish changes.	19:13
	← Derander left	19:14
	→ sjzzalx joined	19:14
_Vi	c0m1, I mean your local git-svn branch will be rebased, not that in public svn repository.	19:14
c0m1	_Vi: thx. I'll start looking through the help page now	19:16
	← ProperNoun left	19:17
jnareb	What do you think about running Git User's Survey 2009?	19:18
	My last email about proposed questions didn't receive any answers... :-(	19:18
	← texel left	19:18
_Vi	jnareb, If it will be, I will answer questions in it.	19:18
	← reprore left	19:19
jnareb	_Vi: do you have any suggestions about proposed answers? http://mid.gmane.org/200906171959.13051.jnareb@gmail.com	19:19
_Vi	jnareb, Are questions will be the same as in Git User's Survey 2008?	19:20
jnareb	And where announce it (besides git mailing list, #git channel and #git channel topic, Git Wiki FrontPage, Git Homepage (http://git-scm.com), repo.or.cz, GitHub / GitHub Blog and Gitorious(?	19:21
	_Vi: no, it will be much shorter (less than 30 questions last count, only three free-form essay questions; compared to 60 in 2008 survey)	19:21
	← corni left	19:22
	corni_ → corni	19:22
_Vi	jnareb, It is no problem for me to answer even 120 questions.	19:22
	→ texel joined	19:23
	→ nurey joined	19:23
_Vi	jnareb, In previous survey the number of non-answers correlated with number of the question.	19:23
	jnareb, May be include all necessary questions, but leave user ability to leave it prematurely? Important questions first, optional last.	19:24
	→ ereslibre joined	19:25
	← maxb left	19:26
	← plediii_ left	19:26
jnareb	_Vi: actually if you use survey from a JavaScript-capable browser, with JavaScript and cookies enabled, Survs.com uses cookies to identify your individual response, and you can always come back later to correct or finish survey questions	19:26
	_Vi: which wasn't unfortunately mentioned	19:27
	_Vi: if you use version for non-JavaScript, you unfortunately loose this capability...	19:27
	people complained about number of questions, so I went to reduce number of questions in survey (and number of answers in questions)	19:28
_Vi	jnareb, I mean not to count "I'm tired of survey, I'm escaping" as non-answer of all remaining questions.	19:28
	← saurabhverma1 left	19:29
	→ naeu joined	19:29
	← abizern left	19:29
jnareb	_Vi: yes, I understand	19:30
_Vi	jnareb, BTW GUS2008 results was the source from which I've discovered some previously unused features of git (like "git add -p").	19:31
jnareb	_Vi: although if people have known that you can submit incomplete survey, and later return to it and finish it at later time, there would be less "abandomenets"	19:31
	← boto left	19:31
_Vi	jnareb, I think if people can stop any time without any harm to survey, it may be almost infinite.	19:31
jnareb	_Vi: for now, the question about "What git commands do you use?" is not present in the survey	19:31
_Vi	jnareb, This question I considered The Main Question Of Survey that time.	19:32
	← fujin left	19:33
	→ dasil003 joined	19:33
	→ jceb joined	19:33
	← WALoeIII left	19:35
	→ Cromulent joined	19:38
	→ mtvee joined	19:38
	← flazz left	19:39
	→ flazz joined	19:39
	→ psoo joined	19:40
	→ Weaselweb_ joined	19:42
	→ corni_ joined	19:42
	← jdavis left	19:43
	← Weaselweb left	19:43
	→ thiago_home joined	19:43
	← gambler_ left	19:44
	→ rleigh joined	19:44
	← jdavis_ left	19:44
	← texel left	19:44
	Weaselweb_ → Weaselweb	19:44
	← sjzzalx left	19:44
	→ boto joined	19:45
	→ ciupicri joined	19:45
	→ gambler_ joined	19:46
jnareb	Well, I can always add it, and yet fit within 30 questions (even counting two parts of question separately)	19:47
	BTW. I think that e.g. difftool is the new invention...	19:47
	→ jdavis joined	19:47
ciupicri	is there some guide regarding storing configuration files in git? I was for example thinking of having at least 2 branches: one branch with the original configuration files distributed with the packages (upstream) and one with my customization. So when the files changes upstream, I could apply those changes to my files. Is this a bad idea?	19:47
	→ jdavis_ joined	19:48
jnareb	there is a problem storing configuration files in version control (for code), period. I tink most commonly used solution is to store skeletons / templates.	19:49
ciupicri	jnareb, so git shouldn't be used for this? Is this what you're trying to tell me?	19:50
Nugget	what's the problem with using version control for config files? I've happily done that for over a decade and wouldn't have it any other way	19:51
	← samphippen left	19:52
jnareb	Nugget: I'm not talking about using version control for config files (sorry ciupicri if it is what you wanted to do; take a look at etckeeper and IsiSetup), but about storing config files in "code" repositories.	19:53
	→ ntoll joined	19:53
Nugget	ah, gotcha. yeah, that can be a real headache	19:53
	← julm left	19:53
	→ julm joined	19:54
jnareb	ciupicri: and take a look what Conary and/or vcs-pkg tries to do; it is from what I understand automatic rebasing of your changes (not merging, rebasing) on top of package changes.	19:54
	ciupicri: see the likes of StGit, Guilt (patch management), and TopGit (topic branch management)... although that can be overkill.	19:54
	→ rlb3 joined	19:54
jnareb	(IIRC Conary uses bazaar, or own VCS)	19:55
dasil003	What's the command to just output the current commit hash?	19:55
ciupicri	jnareb, btw this idea came to me from mergemaster http://www.freebsd.org/cgi/man.cgi?query=mergemaster&apropos=0&sektion=0&manpath=FreeBSD+7.2-RELEASE&format=html and etcupdate http://netbsd.gw.com/cgi-bin/man-cgi?etcupdate++NetBSD-current	19:55
	jnareb, I'll check those projects, thank you	19:55
madduck	jnareb: rebasing is unlikely to be useful in a team context (re: vcs-pkg)	19:56
	← cedricv left	19:56
madduck	dasil003: git show-ref HEAD /	19:56
dasil003	madduck: thanks	19:56
jnareb	ciupicri: ask madduc, IIRC he works on vcs-pkg (which includes problem of config files in packages)	19:58
smoofra	oh wow	19:58
	topgit may be exactly what i need	19:58
	does it have a homepage somewhere or just the repo.or.cz page?	19:59
madduck	smoofra: no homepage	19:59
	smoofra: best documentation is probably in the Debian package	19:59
	smoofra: there is #topgit though	19:59
	ciupicri: i am about to head out, so maybe #vcs-pkg/oftc or the mailing list would be best	19:59
	← corni left	19:59
ciupicri	madduck, ok	20:00
	cpg → cpg\|biab	20:00
	← Cromulent left	20:00
ciupicri	madduck, I'll read about the project and ask if I have any questions/ideas	20:00
	← tbf left	20:00
jnareb	madduck: oftc?	20:04
smoofra	oh man it is. it its exactly, exactly what i need	20:04
	← Moult left	20:05
	→ s0ber_ joined	20:05
madduck	jnareb: irc.oftc.net	20:05
	← nurey left	20:07
	→ corni__ joined	20:09
	→ jsbmsu joined	20:09
	← _Vi left	20:11
jnareb	Ah	20:11
	→ _Vi joined	20:11
	← mefesto left	20:12
	→ tatsuyao joined	20:13
	→ jelly-bean joined	20:13
	→ juanjoc joined	20:14
jelly-bean	when specifying ranges how can i say the range from 6f0088 to aff0c52	20:15
	← Tuomas left	20:17
	corni__ → corni	20:17
	← s0ber left	20:18
	→ cilly joined	20:19
drizzd_	jelly-bean: commit..commit	20:19
	← juanjoc left	20:19
	← cilly left	20:19
	→ khmarbaise joined	20:20
	← bobmcw left	20:21
	← ben_h left	20:22
	→ stephenjudkins joined	20:24
	← rlb3 left	20:24
	→ rlb3 joined	20:24
	→ KiBi joined	20:24
	→ xyz joined	20:25
	dansan\|afk → dansan	20:25
	→ corni__ joined	20:25
jelly-bean	drizzd_: thnx	20:26
	← corni_ left	20:26
kn1ttl	i just imported a project with git-svn, and now i want to delete some branches, but can't get it to work	20:27
	there's a branch remotes/hacking … how do i delete it?	20:27
	→ tvw joined	20:28
	← jelly-bean left	20:28
kn1ttl	git branch -D hacking (or remotes/hacking) results in a: branch not found	20:28
	resmo_ → resmo	20:30
	← neoeinstein left	20:30
	→ Grahack joined	20:31
	→ neoeinstein joined	20:31
	← tchan left	20:32
	→ mefesto joined	20:32
	← psoo left	20:32
	→ tchan joined	20:32
	← neoeinstein left	20:32
	→ psoo joined	20:33
	→ tbf joined	20:33
	← bugfux left	20:33
Ilari	Why are you trying to delete remote tracking branch?	20:34
	kn1ttl: ^^	20:34
jnareb	kn1ttl: try "git branch -r -D remotes/hacking"	20:34
	→ jmou joined	20:34
kn1ttl	Ilari: migrating to git. and project owners want to remove old cruft	20:34
smoofra	so i've noticed that a lot of things that used to be shell scripts have been rewritten in c	20:34
kn1ttl	unmaintained branches and such	20:34
jnareb	kn1ttl: or "git remote <remote> prune" if you want to delete branches which vanished (got deleted) on <remote>	20:34
smoofra	is there some sort of campaign to purge all the shell scripts or something?	20:35
	← gaveen left	20:35
kn1ttl	jnareb: it's without remotes/ ;) but it worked. thanks a lot	20:35
Ilari	kn1ttl: You constructed authors file?	20:35
kn1ttl	hm?	20:35
	← LylePerl left	20:36
jnareb	smoofra: it is called builtin-ification... and yes, there is one. Shell scripts are less portable (e.g. MS Windows) and have worse performance. Same is tru to lesser extent with Perl scripts.	20:36
Ilari	kn1ttl: Otherwise the usernames look like foobaz <foobaz@538395836> or some junk like that.	20:36
kn1ttl	ok	20:36
	no, i haven't	20:36
	how can i do that?	20:36
Ilari	kn1ttl: Construct one (man git-svn should document the format) and then rerun clone.	20:37
Gitbot	Ilari: the git-svn manpage can be found at http://git.or.cz/man/git-svn	20:37
Ilari	stupid bot.	20:37
	← Grahack left	20:37
kn1ttl	xD	20:37
	rerun clone?	20:37
	aaach	20:37
ciupicri	jnareb, I've looked a bit tat vcs-pkg, but I don't think that it's for me. I'm not creating distros :-)	20:38
	← mw left	20:38
Ilari	kn1ttl: Also, you might want to fix the merges if any. For that, metadata might be useful (but it still needs to be removed).	20:38
kn1ttl	there are merges, but i guess fixing them would be too complicated g	20:39
Ilari	kn1ttl: Actually, the most complicated part of fixing a merge is finding what commit was actually merged in.	20:39
kn1ttl	will that involve rebase a lot?	20:40
Ilari	kn1ttl: No rebase. grafts and filter-branch.	20:40
	← bentob0x left	20:40
	→ brizly joined	20:40
	→ neoeinstein joined	20:41
kn1ttl	i think i will just leave _dirty_ merges in the repo	20:41
	← rlb3 left	20:41
	← stephenjudkins left	20:41
	← ciupicri left	20:42
Ilari	kn1ttl: Not an option if devevelopment on those branches continues. You need to fix at least their last merges or merging them isn't going to work.	20:42
jnareb	ciupicri: I was thinking more about how vcs-pkg deals with configuration files when present in package	20:42
kn1ttl	[22:42] <Tronic> I don't care all that much about the history. xD	20:42
	← corni left	20:42
kn1ttl	Ilari: i can still do that at a later point, right?	20:42
Ilari	kn1ttl: No, it can't be done later. And without it, deleting branches causes commits to be lost.	20:43
	jnareb wonders if git-svn would support svn:mergeinfo from Subversion 1.5 (and svnmerge / SVK merge markers)	20:43
	→ rlb3 joined	20:43
	← corni__ left	20:43
kn1ttl	Ilari: loosing commits is not a problem. as i said … i'm removing old cruft	20:44
Ilari	AFAIK, svn:mergeinfo isn't suitable. SVK merge markers at least have suitable sematicts.	20:44
	← julm left	20:44
	← khmarbaise left	20:44
Ilari	kn1ttl: Whereas with SVN you can manually specify what range to merge (get it wrong and you get horrible conflicts). With git, that isn't an option. git merge always autodetects what to merge.	20:45
	→ David-T_ joined	20:45
kn1ttl	ok, so i really have to play a lot around	20:46
Ilari	kn1ttl: And without proper history, that autodetection goes wrong -> horrible amount of conflicts.	20:46
	← David-T left	20:46
kn1ttl	ok	20:46
	my only option then …	20:46
Ilari	kn1ttl: It probably isn't difficult to construct a script that takes in pairs of commits and spits out graft data for fixing the merges.	20:46
kn1ttl	for me it is ;)	20:47
Ilari	kn1ttl: Probably most difficult thing in it is how to parse lines containing two items seperated by space or something.	20:47
kn1ttl	http://www.simplisticcomplexity.com/2008/03/05/cleanly-migrate-your-subversion-repository-to-a-git-repository/ that looks good	20:48
	→ susen joined	20:48
	← jceb left	20:49
susen	Enter text here...h 0131	20:49
	halloooooooooooo	20:49
	→ jceb joined	20:49
Ilari	...	20:49
	David-T_ → David-T	20:49
kn1ttl	ok, but nothing mentioned about grafing merges	20:49
susen	0131j	20:49
Ilari	kn1ttl: Doesn't get branches besides trunk.	20:50
	← susen left	20:50
kn1ttl	ya ok. i'll add -s option	20:50
jnareb	<Ilari> AFAIK, svn:mergeinfo isn't suitable. SVK merge markers at least have suitable sematicts. <--- Why the new shiny feature of Subversion 1.5 is not suitable? How can one screw up something so simple?	20:50
madduck	jnareb: I think the problem is inherent deep down in the design of SVN	20:53
	anyway, I understand you don't want to discuss this; neither do I.	20:54
	In fact, I've been meaning to go to bed for over an hour!	20:54
	madduck swear	20:54
madduck	s	20:54
Ilari	Simpler than I expected: "while read commit merged ; do echo `git rev-parse $commit` `git rev-parse ${commit}^` `git rev-parse ${merged}` ; done".	20:54
	→ shadowhand joined	20:55
kn1ttl	thanks … no way i could've come up with that myself ^^	20:55
Ilari	That eats commit and merged in pairs and spits out data suitable for grafts data.	20:55
	← sea-gull left	20:56
kn1ttl	so ok. i think i have my authors file	20:56
	← loincloth left	20:56
Ilari	kn1ttl: But that leaves the harder step: Actually finding what what was merged and where.	20:57
jnareb	good night madduck	20:57
	← apeiros_ left	20:57
kn1ttl	luckily there were only a few merges	20:57
jnareb	I have thought that in Subversion 1.5 they finally solved (part) of problem by adding merge tracking to SVN core...	20:57
kn1ttl	ok, importing again	20:57
Ilari	kn1ttl: The metadata is helpful in finding the merged in commits. But it must then be stripped later. Fortunately, "burning in" the grafts requires filter-branch anyway.	20:58
	← mtvee left	20:58
Ilari	jnareb: Apparently it didn't...	20:58
	→ bugfux joined	20:58
kn1ttl	Ilari: so, when first cloning i should pull metadata too. because in this far too simplified example they use --no-metadata	20:58
	← cannonball left	20:59
	→ julm joined	20:59
Ilari	kn1ttl: Metadata is useful for figuring what commit corresponds to what SVN revision. And merge messages might refer to SVN revisions.	21:00
jnareb	Ilari: what are probmes with it? (I have only read RelNotes for SVN 1.5, and noticed only that you have explicitely use it, and that trunk is treated special)	21:00
	→ nurey joined	21:00
kn1ttl	hm. when inspecting history with gitk i can see merges	21:01
	how come?	21:01
	← boto left	21:01
dasil003	you can see everything in gitk	21:01
kn1ttl	dasil003: git repo imported with git-svn	21:02
Ilari	kn1ttl: Eh? Two lines going downwards from commit? Can you put a screenshot somewhere+	21:02
dasil003	sorry, i just love gitk	21:02
kn1ttl	even two parent commits	21:02
	→ nurey_ joined	21:03
	→ boto joined	21:03
kn1ttl	http://knittl.is-a-geek.net/public/gitk.svn.png	21:03
	→ pulpfiction joined	21:04
	← bdiego left	21:04
	→ sjzzalx joined	21:04
sjzzalx	is it reckless to use git reflog a lot?	21:05
	← brizly left	21:05
sjzzalx	and reset to refs listed there?	21:05
Ilari	kn1ttl: It is merge. But I noticed something: The first parent seems to point duplicate of commit in the second chain. That's not good.	21:05
kn1ttl	LOLZ? there are different initial commits	21:06
	← jmou left	21:06
kn1ttl	ah ok. for the other folders beside btT	21:07
Ilari	kn1ttl: Not good.	21:07
kn1ttl	Initial revision	21:07
		21:07
		21:07
	git-svn-id: https://ultrastar-ng.svn.sourceforge.net/svnroot/ultrastar-ng/trunk/UltraStar-ng@2 a3f66644-493e-0410-89de-92bcaa846536	21:07
smoofra	i don't really understand what git-reflog is supposed to accomplish	21:08
	← nurey left	21:08
	← deanh left	21:08
smoofra	i mean, i can read the man page, i see what it does, i just don't understand why you would ever want something that did that	21:08
kn1ttl	smoofra: delete/reset branches	21:08
	with reflog you can go back	21:09
wereHamster	smoofra: one often used feature is git reset --hard HEAD@{1}, eg reset to the HEAD to where it pointed before the last command	21:10
	← rlb3 left	21:10
smoofra	why not just HEAD^ ?	21:10
Ilari	kn1ttl: What the point of all this is: One can make the history like it occured when migrating (also some parts might be required by future merges). But these changes are almost impossible to do after migrating.	21:11
	→ rlb3 joined	21:11
wereHamster	smoofra: for example after a rebase, HEAD^ will not be the same as HEAD@{1}	21:11
	cpg\|biab → cpg	21:11
	← jceb left	21:11
	← eletuchy left	21:12
kn1ttl	Ilari: what do you mean?	21:13
	← camwest left	21:13
	← _Vi left	21:13
Ilari	kn1ttl: Altering published history is very nasty operation.	21:13
kn1ttl	yeah, all operations are done locally	21:13
	← qrush left	21:14
	→ wsc joined	21:14
	→ qrush joined	21:14
	→ _Vi joined	21:16
Ilari	kn1ttl: All migration operations are local except the final push to public repo... And that's the final step.	21:16
	← boto left	21:16
smoofra	oh i guess i get it. reflog is just for undoing local mistakes. actually that is easier than searching for the commits you want with gitk or git-lost-found	21:17
wereHamster	smoofra: exactly	21:17
jnareb	smoofra: reflog is local history of where branch tips pointed to (e.g. if you do "git reset --hard HEAD^, then HEAD^ != HEAD@{1})	21:17
	smoofra: it is generalized ORIG_HEAD (to get out of mistakes)	21:18
	← Sigma left	21:18
sjzzalx	Yeah. Is that a bad idea, should I be really careful, or not?	21:19
jnareb	unfortunately the idea of Attic fo reflog for deleted branches didn't take (there is reflog for HEAD, but it might be not enough if one uses "git branch -D").	21:19
sjzzalx	that = resetting to things in reflgo	21:19
jnareb	sjzzalx: if you don't use it when you have uncomitted changes -- not so much (resetting to item in reflog is saved in reflog; think of reflog as a "journal" of operations on given branch, in the sense of journalled filesystem)	21:20
	← tbf left	21:23
	← nurey_ left	21:24
	→ sakyamuni joined	21:25
	→ break joined	21:28
	← mefesto left	21:29
	← tanuki left	21:29
	← rlb3 left	21:29
	→ Cromulent joined	21:29
	← doener left	21:30
	← jayallen left	21:31
	← qrush left	21:32
	← julm left	21:34
	→ julm joined	21:35
	→ bobmcw joined	21:35
	→ kennethlove joined	21:35
	→ jayallen joined	21:35
	→ bobmcw_ joined	21:36
	→ eddyp joined	21:38
	← LiamH left	21:40
	→ bartocc joined	21:44
	← icwiener_ left	21:44
	→ bremner_ joined	21:46
wereHamster	smoofra: exactly	21:46
	sorry.	21:46
	← julm left	21:47
	→ julm joined	21:48
	← ctusar left	21:48
	← btanaka left	21:49
	→ milluminu joined	21:50
milluminu	hihi	21:50
	← _Vi left	21:52
	← bobmcw left	21:53
milluminu	i just bought version control with git (oreilly) so i'm trying to rtfm	21:53
	so i created a git repo with remote origin to git://someurl/base.git	21:54
	and git ls-remote origin shows branches	21:54
	now how do I populate my directory with a specific branch from among those?	21:54
	← mm_202 left	21:55
dansan	Hello wonderful people!!! If I screwed up because I typed "git reset HEAD~18" for some reason and meant "HEAD~2", how do I fast-forward 16 commits?	21:55
	→ bcardarella joined	21:55
dansan	I currently have "Your branch is behind 'origin/master' by 16 commits, and can be fast-forwarded." in git statys	21:55
	*git status	21:55
Pieter	milluminu: git fetch origin; git checkout --track origin/branch	21:55
	if your git is new enough	21:55
	→ ereslibre_desk joined	21:55
sjzzalx	dansan: I would just git pull/merge again and reset HEAD~2	21:56
milluminu	thanks pieter	21:56
sjzzalx	if it's simple like that	21:56
dansan	sjzzalx: I have lots of local changes	21:56
	dansan is still a git newbie :(	21:56
sjzzalx	dansan, you have them now?	21:56
	or you reset them away?	21:57
dansan	oh yes, plus all of the changes from the previous commits to the actual branch	21:57
	Sorry, I did git reset --soft HEAD~18	21:57
	when I meant 2	21:57
sjzzalx	oh, then that doesn't matter. --soft just checks for consistency, iirc	21:57
dansan	I wanted to edit a commit message, and somebody showed me that command yesterday	21:57
	← cemerick left	21:57
sjzzalx	dansan, use git rebase -i instead	21:58
	← bartocc left	21:58
	← bugfux left	21:58
sjzzalx	or plain git rebase	21:58
dansan	thanks	21:58
	→ camwest joined	21:58
	→ eletuchy joined	21:58
jnareb	dansan: "git commit --amend"	21:58
dansan	what arguments for git rebase? 'origin'?	21:58
	njareb: thanks! ;)	21:58
jnareb	dansan: might be what you want	21:59
	→ eseidel_ joined	21:59
dansan	jnareb: somebody showed me a batch way to do it as well, that lets you choose "pick", "edit" etc, in vim, it's in my IRC log :)	21:59
kn1ttl	dansan: git rebase -i	21:59
eseidel_	is this the proper place to ask git user questions? I'm trying to find the right git command to determine if has staged a delete (either in the index, or simply locally). I could parse git status, but I expect there is a better way to see if an individual file is staged for deletion	22:00
	→ johndoigiii joined	22:00
dansan	kn1ttl: "git rebase -i" along just gives me usage information	22:00
	→ ccaputo joined	22:00
	← ccaputo left	22:00
tomoj	dansan: you need to get back to the right HEAD before you do anything else, I think	22:01
jnareb	eseidel_: for scripts you should use either git-ls-files or git-diff / git-diff-files	22:01
	← jonshea left	22:01
dansan	tomoj: correct, and that will solve my problems	22:01
	← hyperair left	22:01
dansan	tomoj: how do I do that? I'm 16 commits behind it	22:01
tomoj	I'm thinking something like 'git reset --soft HEAD@{1}' would work, but don't do that unless someone else says it's safe :)	22:02
kn1ttl	dansan: then git rebase -i HEAD~16 or something like that. depends on how many commits you want to edit	22:03
dansan	sounds like a job for tar czf ../sadhfoasjfw.tgz $(find -iname '.c' -or -iname '.h") loL!	22:03
	ahh!	22:03
	I only need to edit the 17th commit, commits 1-16 are from the repository	22:03
	← dasil003 left	22:04
tomoj	dansan: don't do that	22:04
sjzzalx	dansan	22:04
	dansan, git pull	22:04
	→ stephenjudkins joined	22:04
sjzzalx	and yeah, don't edit commit already pushed	22:04
dansan	what does git pull do?	22:04
	← kpreid_ left	22:04
sjzzalx	git pull fetches changes from origin and merges them with your local copy	22:04
	← WinterWeaver left	22:04
eseidel_	jnareb: git ls-files seems like the right diretion. thank you. "git ls-files -d deleted_file" doesn't seem to show me anything though. "git ls-files existing_file" works fine though	22:04
dansan	hmm, isn't there a way to simply "fast-forward" or re-commit?	22:05
sjzzalx	dansan: git pull will fast-forward you completely	22:05
	and then you can reset back to what you want	22:05
dansan	hah! git pull fails because a file has local changes	22:05
sjzzalx	but don't edit someone else's commit, that's a bad idea	22:05
dansan	error: Entry 'ANNOUNCE' not uptodate. Cannot merge.	22:05
	→ rlb3 joined	22:05
dansan	I never edit other people's commits!	22:05
sjzzalx	dansan: ok, dude, let's talk about this	22:05
dansan	hehe	22:05
sjzzalx	dansan: Right now, you're 16 commits behind. Do you or do you not have a bunch of changes you want to keep?	22:06
dansan	git status: http://www.pastebin.ca/1471710	22:06
	I have a lot of changes i want to keep	22:06
	almost all are backed up, some are not	22:06
	(only about 20 minutes work not backed up)	22:06
jnareb	See you later	22:07
	← jnareb left	22:07
tomoj	dansan: why does it say 156 commits?	22:07
	→ hyperair joined	22:07
	→ jerome__ joined	22:07
dansan	Jesus! after doing git pull it's saying 156!	22:07
	← hatseflats left	22:07
sjzzalx	dansan: OK. So, tell me what you want to do.	22:07
dansan	ahh, probably because there are more commits that I haven't merged, crap! :(	22:07
	← jdavis_ left	22:07
sjzzalx	You want to keep all of the changes you have?	22:07
	→ stephenjudkins_ joined	22:07
dansan	Yea	22:07
sjzzalx	and then get back to the tip on origin/master?	22:07
dansan	But only MY changes	22:08
	Except that after git pull, origin/master changed	22:08
sjzzalx	right, git pull will do that	22:08
dansan	because it updated with the repository.	22:08
sjzzalx	Yeah	22:08
	so, do you want to get back to the tip of origin/master with your changes?	22:08
tomoj	sjzzalx: won't 'git reset --soft HEAD@{...}' undo the soft reset from before?	22:08
dansan	I wasn't ready for that	22:08
	→ _Vi joined	22:09
sjzzalx	dansan: If you want to keep things exactly the way they are, don't do anything. If you want to get back to tip of origin, you have to do different stuff. Do you want to? And do you want your changes to go there with you?	22:09
dansan	I'm thinking about manually copying the files I modified, blowing everything away, etc :)	22:09
kn1ttl	there's git stash?	22:09
	← ereslibre left	22:10
sjzzalx	tomoj: Yeah, if he checks the reflog he should be able to reset to any place he wants.	22:10
	→ whaley_ joined	22:10
sjzzalx	dansan: Why?	22:10
dansan	Well, I have to deal with any potential merge conflicts now, hopefully, there are none	22:10
	← psoo left	22:10
	← tatsuyao left	22:10
sjzzalx	dansan: Just tell me what you want to happen	22:10
	← johndoigiii left	22:10
	→ ph^ joined	22:10
sjzzalx	Do you want to get to the tip of origin/master or not?	22:10
dansan	I would ideally like to get back to the condition I was in prior to typing "git reset --soft HEAD~18" when I meant "git reset --soft HEAD~2"	22:10
tomoj	dansan: so check the reflog and 'git reset --soft HEAD@{..}'	22:11
sjzzalx	dansan: ok. do git reflog, find what you're missing	22:11
	and then run the command tomoj pasted	22:11
	and is that all you want?	22:11
	← whaley left	22:11
	whaley_ → whaley	22:11
dansan	what do I replace '{..}' with?	22:11
	← bx2 left	22:11
tomoj	dansan: whichever entry in the reflog is the one you want	22:12
sjzzalx	dansan: You can replace HEAD@ with the real SHA, it's right to the left of it	22:12
dansan	ok :)	22:12
sjzzalx	i find that easier usually because it doesn't change	22:12
tomoj	dansan: and, again, don't edit a commit which comes before commits that are in the remote repository	22:12
	← jerome left	22:12
dansan	tomoj: sorry, I would never do that :)	22:13
	tomoj: I still have a boat load of changes that aren't mine as locally modified files	22:14
	I suspect due to git pull	22:14
	→ porcnick joined	22:15
eseidel_	I'm confused as to when "git ls-files -d" ever lists anything	22:15
dansan	git status says " Your branch and 'origin/master' have diverged, and have 2 and 140 different commit(s) each, respectively."	22:15
	eseidel_ → eseidel	22:15
	← camwest left	22:15
dansan	I'll just copy my files and blow it all away (I'm cloned from a local repository anyway, no big downloads to start over)	22:16
sjzzalx	dansan: Yeah, because 140 commits happened between your changes and your last pull	22:16
dansan	hmm	22:16
sjzzalx	dansan: Well, make a backup, that's a good idea, just cp -Rv folder/ bakfolder/	22:16
	dansan: and after that go back to your real folder and try git rebase	22:17
dansan	oh, I already did that one :)	22:17
	← joshsdc left	22:17
sjzzalx	you already rebased?	22:17
	fr\| → fr	22:17
	← kumbayo left	22:18
	← naeu left	22:19
tomoj	doesn't a merge that fails leave your index and working tree alone?	22:19
dansan	ok, I have a wine/git and a wine/git.dev. wine/git is where I originally cloned from the repository and my wine/git.dev I cloned using the wine/git so I could have two copies, one either unmodified or slightly modified and the other a playground...	22:19
	← suman left	22:20
dansan	And I don't have to download the files twice	22:20
	So when I update, I usually git fetch && git rebase origin in my wine/git and then do the same in my wine/git.dev and do any 3-way merging needed	22:20
sjzzalx	tomoj: A failed merge puts your working tree up to the failure and then you're expected to fix it before moving on, if I remember right	22:21
	← mattikus left	22:21
sjzzalx	but you can undo it easily with a git reset --hard HEAD~1	22:21
tomoj	sjzzalx: unless you have local changes	22:21
	then what, git reset --merge HEAD~1?	22:21
	→ mattikus joined	22:21
	← lymeca left	22:21
dansan	so actually, I did a "git format-patch --keep-subject -o out origin" earlier and I have the patch files for the two commits I made that appear to be preventing it from doing the rebase, so maybe I can revert those local commits and rebase will work	22:22
sjzzalx	tomoj: In that case I usually just stash, rebase, or branch, depending on what I need. From the man page, though, it looks like git reset --merge would work too	22:22
	← mattikus left	22:22
	→ Derander joined	22:23
sjzzalx	dansan: resolve the merge error and git rebase --continue?	22:23
	← pcapriotti left	22:23
	← stephenjudkins left	22:24
	← rlb3 left	22:24
	→ rlb3 joined	22:24
	rylar_off → rylar	22:24
	→ hendry joined	22:25
hendry	i have trouble git clone git:// repo(remote: aborting due to possible repository corruption on the remote side.), though ssh:// works. Any ideas?	22:25
dansan	ok, I think I got it now. I did "git reset --soft" and gave the SHA of before I made my two commits, then I did "git reset" and it got rid of the "divergence" so now I'm going to commit everything in a dummy commit and rebase origin and that should work	22:26
Ilari	hendry: You sure both URLs refer to the same repository?	22:26
	← Cromulent left	22:26
	← kennethlove left	22:26
dansan	w00t! conflicts :)	22:26
	← Sho_ left	22:26
_Vi	hendry, Network is OK? Remote git-daemon is OK?	22:26
	→ rlb3_ joined	22:26
	← rlb3 left	22:26
	rlb3_ → rlb3	22:26
	← Weaselweb left	22:27
	← MisterN left	22:27
Pieter	"Electronic Warfare Project Manager" -- I want to apply for that job just for the title	22:27
	← bremner_ left	22:27
	← eletuchy left	22:28
dansan	ok, thanks guys, I got it from here. This is just the part I didn't want to deal with just yet (the mergeing), but it doesn't look terribly bad	22:28
	→ eletuchy joined	22:28
	← flazz left	22:28
	← Ryback_ left	22:29
	← Yuuhi left	22:29
	→ kpreid_ joined	22:29
robinr	spearce: there was a problem with the EGit renamee patch. You got the mail from bugzilla?	22:29
	→ xymbol joined	22:30
rektide	smoofra: an open source google reader is on my todo list	22:30
	← ntoll left	22:30
	← lorandi left	22:30
	→ tro joined	22:31
spearce	robinr: oh, yea, but i didnt' have a chance yet to fix it.	22:31
	→ agenteo joined	22:34
	sebleier → sebleier\|away	22:35
hendry	Ilari: _Vi error: Could not read 38e87a084c1eb3859cff1305536e184648f4dc28 permission problem? what id is git-daemon-run supposed to work with	22:35
	← eseidel left	22:36
dansan	whew! Ok, I got my stuff where it needs to be (and I got my merge out of the way too :)	22:36
_Vi	hendry, "git fsck" on remote side OK?	22:36
sjzzalx	dansan: That's good news :)	22:36
	→ chriswarren joined	22:36
dansan	sjzzalx: thanks for the help! Although I wish I didn't do "git pull", lol!	22:36
_Vi	hendry, "git fsck" from user that runs "git-daemon".	22:36
dansan	SLOWLY, I'm learning to get around in git	22:36
sjzzalx	dansan: Yeah. You should know about git pull --rebase. That's what git fetch && git rebase origin/master does, anyway	22:37
	git pull --rebase is just a faster way to do it	22:37
dansan	ooh, fun :)	22:37
sjzzalx	git pull without the rebase switch merges instead of rebasing, that's the difference	22:37
Ilari	Also, pull --rebase can deal with nonlinear movement (at least in some cases).	22:37
dansan	And I got "git mergetool" working well for me with kdiff3, which I appreciate the hell out of	22:38
	← xyz left	22:38
dansan	Also, the merge with the repository may help me because somebody added test code to a file that affects me :)	22:38
chriswarren	hey all - I really hope someone can give me a hand with this. I committed and pushed a change that removed a directory, plus added a few files. I need to get the directory back. what's the best way to do this? I've manged to get the dir back, but I lose the other files in the process.	22:39
_Vi	chriswarren, "git checkout HEAD^ -- /path/to/your/dir"	22:39
hendry	_Vi: can't understand wtf is going on with the git-daemon setup in Debian squeeze. userid i think should be root, but instead it's 105. argh	22:41
_Vi	hendry, dit-daemon is root?	22:41
kn1ttl	Ilari: still around? :)	22:41
	i finished re-cloning the svn repository with the authors file	22:41
_Vi	hendry, "setfacl -Rm u:105:rwx /path/ro/repo" ?	22:42
kn1ttl	and what do i do with the "svn"-merges	22:42
_Vi	hendry, Better expect everything that can be run not as root runs not as root.	22:42
	→ guitsaru joined	22:42
robinr	spearce: ok, that's working then.	22:43
hendry	_Vi: oops, i mean git-daemon	22:43
	_Vi: git fsck on the suspect repo from git-daemon user seems fine	22:44
	← ph^ left	22:44
_Vi	hendry, "avahi:x:105:106:Avahi mDNS daemon"?	22:44
dansan	So does git have an fdisk option too?	22:44
sjzzalx	dansan: git fsck	22:44
_Vi	hendry, "git fsch --full" also?	22:44
chriswarren	_Vi: Thanks a lot	22:44
dansan	Maybe I can use it to mow my freakin lawn! I keep getting suprised by this thing :)	22:44
hendry	_Vi: no the user/group is OK. got confused as it takes time for the id to come in affect on my system for whatever reason	22:45
_Vi	hendry, What do you mean "it takes time"?	22:45
dansan	sjzzalx: Do you mean to tell me there is a git file system?	22:45
	← cpg left	22:46
dansan	to dedicate a partition?	22:46
	→ jpieper joined	22:46
hendry	_Vi: sorry. i am really tired. wasn't running fsck as the gitdaemon user. now i get broken link: http://static.natalian.org/2009-06-23/whoops.txt	22:47
	← chriswarren left	22:47
xymbol	hi there, is there a general consensus on what to use as push.default? tracking or current?	22:48
	→ mm_202 joined	22:48
sjzzalx	dansan: no	22:48
_Vi	hendry, As git-daemon user, "find /path/to/repo/.git \| cpio -o > /dev/null". No errors?	22:48
sjzzalx	dansan: There are writings to that effect, about how git is filesystem, how it's awesome, how it's more important than Unix, and all of this other stuff	22:49
	but git fsck just checks the consistency of the repository	22:49
	makes sure you have objects for commits, etc.	22:49
mm_202	Hey guys, do any of you use git to keep files (text files, openoffice, mp3s, avi's, etc) synced between computers?	22:50
	← cmarcelo left	22:50
hendry	_Vi: ah, find: `/srv/git/faq/objects/d6': Permission denied	22:50
_Vi	mm_202, What is the problem?	22:50
sjzzalx	mm_202: sure. That's the whole point of git, keeping files synced between computers.	22:50
dansan	sjzzalx: lol! Well no surprise, it was spawned by Torvold's ego	22:50
_Vi	hendry, Then "setfacl -Rm u:dit-daemon:r-x /path/to/repo/.git" or something liks that.	22:50
	← hpoydar left	22:51
sjzzalx	dansan: http://www.advogato.org/person/apenwarr/diary/371.html	22:51
dansan	sjzzalx: I'm a Stalman-ist you might say (and not Stalin as most people seem to hear)	22:51
	→ bremner_ joined	22:51
mm_202	_Vi: no problem. I just have about 3-5 computers that I would like to keep my media synced together, but Im talking around ~100GB, would the git overhead not make it good choice? (versus using something like rsync)	22:51
	→ glommer joined	22:52
hendry	_Vi: what about instead, sudo adduser gitdaemon www-data ?	22:52
_Vi	hendry, E.g. make sure git-daemon user have read access to ".git" directory. (and write access, if you want to push over git protocol).	22:52
dansan	sjzzalx: I try to use the term "GNU/Linux"	22:52
	Nugget snickers	22:52
hendry	_Vi: because my git stuff is generally under www-data group	22:52
_Vi	mm_202, May be rsync will be better solution (especially for non-text files).	22:52
sjzzalx	dansan: Yeah. I think that debate is long lost. Everyone knows GNU contributed large parts of the userland, but plenty of others have too.	22:52
_Vi	hendry, Include git-daemon user to www-data group.	22:52
	→ jmou joined	22:52
_Vi	hendry, (and restart git-daemon, of course).	22:53
sjzzalx	mm_202: Yeah, I would use rsync unless you need to keep track of any differences between the files.	22:53
	← tim left	22:53
	→ tim_jenssen joined	22:53
sjzzalx	mm_202: For instance, if you were updating a movie file a lot, if you're a movie editor or something	22:53
dansan	sjzzalx: Yea, I know, it's a lost argument. Almost nobody outside Stalman's circle says "GNU/Linux", but I still try to do it anyway,	22:53
Nugget	the "GNU prefix" debate is sort of offensive to the people who helped build all the software that GNU copied, too.	22:53
dansan	Nugget: agreed	22:53
	→ ph^ joined	22:54
sjzzalx	dansan: I think it's a little silly. If I use KDE, should I call it KDE/Linux? The KDE people wrote so much of my userland, if I happen to use KDE.	22:54
	← julm left	22:54
dansan	I just hate that it revolves around one person's name and to lay-persons, they appear to get the credit.	22:54
mm_202	sjzzalx: For avi's and mp3s, yeah, rsync works fine. But I'd like to keep track of changes in my photos and start adding meta-tags to them to keep track / organize them.	22:54
Nugget	but yeah, it totally devalues the hard work of millions of coders and elevates the GNU code to an undeserved level	22:54
	well, not millions, but you know what I mean	22:54
sjzzalx	dansan: Stallman just has a superiority/correctness complex and wants special credit, which he already gets from FSF stuff, so I think he's fine	22:54
krh	bzzzt, off topic!	22:54
mm_202	I was just curious if anyone else here has used git for something like that?	22:54
dansan	Nugget: point well taken	22:54
_Vi	mm_202, Use git if you 1. can give advantage of branching/merging or 2. Have versions of files that should be saved or 3. need SHA hashes that confirm both files and their histories.	22:55
dansan	Nugget: So let's just start calling it Moof instead of Linux!	22:55
	dansan brainstorms	22:55
Nugget	I don't even use linux :)	22:55
	dansan has a flood.	22:55
	dansan is wet now	22:55
dansan	crap, that didn't work :(	22:55
	Nugget, what's your OS?	22:55
Nugget	OS X on the desktop, FreeBSD in the rack, and OpenBSD as the firewall.	22:56
	→ bremner__ joined	22:56
dansan	ahh, gotcha. So BSD all of the way then essentially	22:56
	← thiago_home left	22:56
	→ qrush joined	22:56
Nugget	Unix is my dogma. BSD just does the best job at it.	22:56
mm_202	_Vi: what would be an approx estimate of the git overhead on a 1GB compressed file?	22:56
dansan	gotcha =)	22:56
	Isn't OS X the 1st one that was built on top of the BSD kernel?	22:56
	jmou → julm	22:57
Nugget	2 points. 1) there's no such thing as "the BSD kernel". 2) OS X is a Mach kernel, and just uses FreeBSD bits for the userland code.	22:57
_Vi	mm_202, By "compressed" you mean "uncompressible"?	22:57
sjzzalx	dansan: The first what? Certainly not the first OS or anything else. It's the first semi-successful mass-market consumer OS based on BSD.	22:57
mm_202	_Vi: nod.	22:57
_Vi	mm_202, "nod" means "Yes"?	22:57
	→ eseidel joined	22:57
mm_202	lol, yes.	22:57
dansan	sjzzalx: hah! I meant the first Apple OS based upon BSD, lol!!	22:58
	← jrmuizel_ left	22:58
dansan	sorry about that =)	22:58
Nugget	depends on if you count NeXT or not. :)	22:58
mm_202	_Vi: Im curious if the overhead if linear or constant, etc.	22:58
dansan	ahhh!	22:58
	→ MattRich joined	22:58
dansan	Wait, I thought that was a DEC product	22:58
sjzzalx	dansan: Yes, it's the first Apple OS that was based on free software, BSD, Mach, etc.	22:58
dansan	Shows you what I know	22:58
_Vi	mm_202, I think if you add 1GB avi file to git, the repository will grow about 1GB.	22:58
Ilari	mm_202: Something like ~100kB.	22:58
mm_202	NIce!	22:58
Ilari	mm_202: Plus size of file itself+	22:59
mm_202	lol, yeah, figured that :)	22:59
	← MattRich left	22:59
Nugget	dansan: http://en.wikipedia.org/wiki/NeXT	22:59
sjzzalx	mm_202: It shouldn't add 1GB every time you commit a new version, though, unless it's just totally chagned.	22:59
_Vi	mm_202, (But it will be a problem to remove that file from repository)	22:59
mm_202	git is great, but not magic	22:59
sjzzalx	because git uses deltas, much like rsync	22:59
Ilari	Compressed stuff doesn't delta.	22:59
mm_202	yeah, but the metatags on jpegs will most likely delta since its just text.	23:00
	But I guess I'll just play with it and see.	23:00
_Vi	mm_202, You will wait while it will be adding all your big files to index. And you will probably wait when you commit (not sure). You will wait when using "git fsck". Fetch/push should work be more or less well.	23:00
Ilari	mm_202: Unfortunately, large files do nasty things to performance if deltification is enabled.	23:00
	→ lightcap_ joined	23:01
mm_202	Yeah, I might just play with it for my home dirs's Documents and such across boxes and not try it with media.	23:01
_Vi	mm_202, If there is certain advantages of using git for this thing, you may use git for small files and just rsync large ones.	23:02
	merlin83 is interested to see a 1TB git repository storing the revision history of someone's homemade video ;-)	23:02
hendry	_Vi: thanks, you taught me a couple of new tricks, setfacl and that "find /path/to/repo/.git \| cpio -o > /dev/null" was clever	23:02
mugwump	someone the other day was talking about putting '/*' in the .gitignore so that files/directories had to be manually added to be tracked	23:02
	← hendry left	23:02
_Vi	hendry, setfacl trick may cause problems in future (new files created by non-git-daemon user will not have this ACL entry).	23:03
	hendry, Better add git-daemon to the group or make repository o+rx.	23:04
	→ scarabx joined	23:05
	← candybar left	23:06
	← dansan left	23:07
	← rlb3 left	23:07
	← willb left	23:07
Ilari	Just give directories apporiate default ACL?	23:07
	→ rlb3 joined	23:07
robinr	should work	23:08
	I use default ACL's for cvs and that works	23:08
	→ suman joined	23:09
	→ joshsdc joined	23:09
	← DavidKlein left	23:10
	← bremner_ left	23:10
	← Stravy left	23:10
	← ph^ left	23:12
	← w\|shimmy left	23:13
_Vi	hendry, Yes, default ACLs should also work: "setfacl -dRm u:git-daemon:r-x /path/to/repo/.git".	23:14
	← corybantic left	23:14
_Vi	Ilari, robinr, Thanks for advice about default ACLs existence.	23:14
mm_202	Completely separate question: how difficult is it to extract text from a bare git repo? (without cloning it or using any git tools)	23:14
	← stephenjudkins_ left	23:14
_Vi	mm_202, Probably simple.	23:14
	← sjzzalx left	23:14
mm_202	lol, damn	23:15
_Vi	mm_202, (testing)	23:15
mm_202	Does git have any soft of encryption module?	23:15
	k	23:15
	→ dansan joined	23:15
	← agile left	23:15
	mm_202 will just avoid storing sensitive text files in git..	23:15
_Vi	mm_202, Yes, simple.	23:16
	mm_202, "git show revision:path/to/file".	23:16
	→ dreiss joined	23:16
_Vi	mm_202, Git is good at storing sensitive text files.	23:17
cehteh	just gpg encrypt your data	23:17
	← bremner__ left	23:17
mm_202	_Vi: yes, but without using any git tools	23:17
	→ wshimmy joined	23:17
	→ bremner_ joined	23:17
	→ ph^ joined	23:17
_Vi	mm_202, Without git tools? No way.	23:17
mm_202	cehteh: yeah, Im curious if I could setup hooks to encrypt and decrypt data to/fro the repo on the fly.	23:18
Ilari	mm_202: You would need specialized tools to read even loose objects. Packs are even worse.	23:18
_Vi	mm_202, (any way will be reimplementing part of git).	23:18
cehteh	mm_202: if you dont use crypto, then anyonw who wants to see them can see them with no much effort	23:18
	dansan → dansan\|afk	23:18
	← ph^ left	23:18
cehteh	mm_202: should be possible	23:18
	either with gitattributes or with a hook	23:18
_Vi	mm_202, Just store repository on encrypted fs.	23:18
	← hobodave left	23:18
_Vi	mm_202, (dm-crypt, encfs).	23:18
cehteh	well against what do you want to protect it?	23:19
	mm_202 does some googling...	23:19
_Vi	mm_202, I have LUKS partition on my USB drive with git repository inside (with backup of my files).	23:19
	→ EGreg__ joined	23:19
mm_202	cehteh: just some coder 'friends'. Nothing serious like the FBI or anything	23:19
EGreg__	I think I'll be moving to git from svn soon	23:19
	but what will I use instead of trac/bugzilla?	23:19
cehteh	trac :)	23:19
mugwump	you can use trac	23:20
cehteh	well the trac git plugin sux	23:20
Ilari	In security, the always important question is "what I want to protect against?"	23:20
cehteh	or is it fixed meanwhile?	23:20
EGreg__	yeah but it's a hack, cehteh :)	23:20
	i mean, without the but	23:20
mugwump	which one? there are several	23:20
EGreg__	nope, i don't think it has been fixed.	23:20
mugwump	at least two anyway	23:20
cehteh	Ilari: exactly	23:20
EGreg__	I like to use trac, and it's the main reason for me to use svn :P	23:20
	in short: BAH ^2	23:20
	→ daum joined	23:21
cehteh	we just use trac with wiki and repository integration disabled intentionally	23:21
mm_202	_Vi / Ilari / cehteh: thanks guys for your help and answering my questions!	23:21
_Vi	mm_202, If you want to use Git for very serious nonstandard purpose, you should learn how it works internally.	23:22
	→ sArge-TF joined	23:22
	→ sgrimm joined	23:23
cehteh	my laptop uses encrypted dm anyways .. against when someone steals it	23:23
mm_202	_Vi: Im a C/C++ coder so that shoiuldnt be too difficult. Can you suggest a good starting point, or should I just goggle around?	23:24
cehteh	but if you want to protect single files committed on a server, while still providing public checkouts, then it gets bit more complicated	23:24
mm_202	nah, its all private.	23:24
daum	hey guys i have a branch called 'exp' on my origin, i tried git fetch origin/exp and now i have a branch called origin/exp in my repository, but when i try git push origin exp i get error: src refspec exp does not match any.	23:24
cehteh	just encrypt your disk then?	23:24
	← sArge-TF left	23:25
cehteh	well thats slightly complicated with a server because it needs attention to boot	23:25
Ilari	daum: 'git fetch origin/exp'? I would expect that to fail.	23:25
mm_202	cehteh: yeah, when I rebuild my computer I'll just do that and be done with it.	23:25
Derander	Hi. I'm trying to convert an SVN repository that I just freshly created from a dump. I'm using git svn clone to start the conversion, and I get this error: http://gist.github.com/134908	23:25
wshimmy	mm_202: git clone git://git.kernel.org/pub/scm/git/git.git	23:25
mm_202	wshimmy: danke	23:25
wshimmy	mm_202: Documentation/technical has some good stuff in there	23:25
	(not to mention, all the good *.[c,h] stuff in there ;P)	23:26
daum	Ilari, hm maybe it was a bit different(was a few days ago) but i have now: git branch	23:26
	master	23:26
	* origin/exp	23:26
Ilari	Derander: Upgrade SVN?	23:26
	← robinr left	23:26
Ilari	daum: Maybe you want to rename origin/exp to exp.	23:27
Derander	Ilari: I will see if I can do this.	23:27
	→ stephenjudkins joined	23:27
daum	Ilari, looks like that did it	23:28
	← bremner_ left	23:29
	← stephenjudkins left	23:30
	rylar → rylar_off	23:32
	→ bentob0x joined	23:32
	← bobmcw_ left	23:34
Derander	Ilari: thanks, the repository is cloning now.	23:35
	→ jrmuizel joined	23:37
	→ flazz joined	23:38
	← _Vi left	23:39
	→ _Vi joined	23:40
	← rlb3 left	23:41
	→ candybar joined	23:43
	→ ilteris__ joined	23:45
	← flazz left	23:46
	← bentob0x left	23:47
	→ flazz joined	23:48
	→ archis joined	23:48
	← c0m1 left	23:49
	← archis left	23:51
	← Bass2 left	23:51
	→ aspotashev\|eeepc joined	23:56
	→ w\|shimmy joined	23:57
	← carrus85 left	23:58
	← thommey left	23:59
	← scarabx left	23:59

Logs Search ←Prev date Next date→

Channels Documentation