IRCloggy #git 2009-01-28

Provider of IRC logs since 2005.

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

	2009-01-28
kansan	how do i get the git repo info from command line?	00:01
	specifically address for the repo?	00:01
doener	s/repo/remote ?	00:01
	that'd be "git remote -v" or "git remote show <remote>"	00:01
dsop	wasn't there a special merger for ChangeLog/NEWS files somehwere?	00:01
vmiklos	man gitattributes:)	00:02
doener	dsop: .gitattributes supports a "union" merge setting	00:02
	→ jmspeex joined	00:03
vmiklos	you probably want the union merge driver:)	00:03
doener	and I think GNU had its own thing	00:03
	http://git.savannah.gnu.org/gitweb/?p=gnulib.git;a=blob;f=lib/git-merge-changelog.c;h=524d5d34939242a658fbd989ce57cb05de2fdb03;hb=283dd26efdfbeeff672cbcb0e6811f54e032364f	00:04
	uuuhh... the indentation... it hurts...	00:05
	;-)	00:05
	← alezandro left	00:05
	← ebel left	00:07
	← LiamH left	00:09
	→ EmilMedve joined	00:11
	→ CardinalNumber joined	00:15
	← jm left	00:21
	→ matt joined	00:23
matt	hey guys, what webinterfaces for git are there	00:23
	im not too familiar	00:23
	← CardinalNumber left	00:24
matt	something run locally? like this http://code.suckless.org/hg/	00:24
	← ProperNoun left	00:24
doener	matt: http://git.or.cz/gitwiki/InterfacesFrontendsAndTools#head-e5a6762d6aed31c5a2034d52c1733ead46402c31	00:24
matt	doener: thanks	00:24
kansan	how do i compare a branch'ed version of one file to the one in master? also is there a gui that makes this easy? (on hardy)	00:25
	→ tav_ joined	00:25
doener	git diff branchA branchB -- <file>	00:25
	and I doubt that a gui can make that any easier...	00:26
	← korpios left	00:26
kansan	fatal: bad revision "Master'	00:26
doener	s/M/m/	00:27
generalBordeaux	anybody having any insight of probable reason for git-ls-remote not working?	00:27
kansan	doener, ok so no diffs will happen till i actually commit something to the branch	00:27
doener	generalBordeaux: crystal ball is in repair	00:27
kansan	how do i diff against the most recent remote copy of a file	00:27
doener	kansan: just use the appropriate remote tracking branch as the second argument...	00:28
kansan	ah k	00:28
doener	eg. "git diff master origin/master -- some_file"	00:28
	→ nuncanada joined	00:29
generalBordeaux	http://git.cachevideos.com/videocache.git , git version 1.5.5.6	00:29
	doener, ^	00:30
	it doesn't give any output .. not even error :(	00:30
	→ jasber joined	00:31
	← patmaddox left	00:31
doener	generalBordeaux: what's the command that you use? That'd be more interesting than some gitweb	00:31
generalBordeaux	git ls-remote http://git.cachevideos.com/videocache.git	00:31
	git ls-remote http://git.cachevideos.com/videocache.git master	00:31
	and git ls-remote --tags http://git.cachevideos.com/videocache.git	00:31
	doener, ^	00:32
doener	well, that's the gitweb url, not the url to the repo	00:32
	→ cartucho joined	00:32
	→ troter joined	00:32
	← tav left	00:33
generalBordeaux	doener, is it not supposed to be the same ... if everything is default?	00:33
doener	no	00:33
generalBordeaux	then how do i really find out the repo url ?	00:33
	i use ssh for pushing ...	00:34
doener	to that repo? Then that's one url you can also pass to ls-remote	00:35
	→ scientes joined	00:36
generalBordeaux	doener, that works ... but thats password locked ... how can we access it without ssh ...	00:36
	i mean via http or git://	00:36
	here i am talking about anonymous access!	00:37
	→ jeckel joined	00:37
doener	for git:// you'll need a git-daemon running, for http, you need the repo in a place that's served by your webserver, and you need to run "git update-server-info" after each push (for http)	00:37
	there's a sample hook to do the update-server-info stuff for you	00:37
generalBordeaux	repo is in /some/dir/videocache.git and git.cachevideos.com is rooted at /some/dir/ ...	00:39
	doener, awesome man!!! it works :)	00:39
	→ jwormy joined	00:39
	← k0001 left	00:40
generalBordeaux	doener, you rock!! thanks a millions ... i was banging my head on this since last few hours!	00:40
jwormy	r0bby: lo'	00:40
doener	generalBordeaux: so the update-server-info was missing?	00:40
generalBordeaux	doener, yup :D	00:40
doener	generalBordeaux: a more recent (but maybe not yet released?) version would've told you to do that, when you ran the ls-remote ;-)	00:40
generalBordeaux	doener, anyways .. thanks again .. i need to run update-info every time i push?	00:41
doener	yep, otherwise, the client gets old information	00:41
	but as I said, there's a sample hook in the hooks/ directory	00:41
generalBordeaux	doener, sure ... i'll setup a cron for that ...	00:41
doener	you can just enable that	00:42
	← willb left	00:42
doener	nah, cron is crap... you'll always have a window with outdated information	00:42
	just enable the hook so it's run automatically after the push	00:42
generalBordeaux	doener, ok	00:43
	← whee left	00:44
generalBordeaux	i am enabling the post-update hook ... i think that would be safe	00:44
	← joshdavey left	00:44
doener	yeah, that does nothing but the update-server-info by default	00:44
	→ PieterMobile joined	00:45
generalBordeaux	yeah..	00:45
	← jwormy left	00:46
	← mbroeker left	00:46
	drobbins → drobbins\|away	00:46
	← PieterMobile left	00:48
	→ tvoet_ joined	00:48
	→ kennethlove joined	00:49
	← johan left	00:50
	← Ariens left	00:53
	→ RayMorgan joined	00:55
	← dsal left	00:56
danderson	So, I need to cherry-pick some CLs from one repo to another. The repos have no history in common (artefact of hg conversion gone wrong). I have both repos locally, how could I do this?	00:56
	→ nothingHappens_ joined	00:57
RayMorgan	hey, I was wondering if there is anyway to search a range of revisions for some content (kind of like git grep ... maybe it can do that?). Is there anyway to do this?	00:57
vmiklos	git log -S'foo bar' commit1..commit2?	00:58
parasti	RayMorgan: man git-grep?	01:00
	→ Phoenixfire159 joined	01:00
doener	danderson: CL = commit list? or what?	01:00
	← _slacker_ left	01:00
danderson	changesets, sorry	01:01
	perforce terminology, leakage from work	01:01
doener	EFOREIGN_TERM	01:01
piroko	danderson: You should convice your work to use git ;)	01:01
RayMorgan	vmiklos: thanks!	01:01
danderson	I want to cherry-pick a list of commits from one repo to the other, with the added fun that they share no common history	01:01
	the commits will apply, but I can't just pull the repo in as a remote and have everything blend	01:02
doener	danderson: add the other repo as a remote, set remote.<name>.tagopt to --no-tags in the config	01:02
	danderson: then "git fetch <remote>"	01:02
	danderson: and you'll have all the stuff in the new remote tracking branches	01:02
	danderson: then you're free to cherry-pick, merge whatever	01:02
danderson	piroko: some of us already do, with adequate glue	01:02
doener	danderson: afterwards, just delete the remote	01:03
danderson	and there are other considerations that make a full move a practical impossibility :(	01:03
	doener: what does the --no-tags do? My remote-fu is weak	01:03
doener	danderson: passed to "git fetch", stops it from fetching tags (which you probably don't want, and which aren't removed when you drop the remote)	01:04
	just to prevent your repo from being cluttered with unrelated tags	01:04
	← kennethlove left	01:04
danderson	aah, I see.	01:04
	I don't think I have any tags in this repo, but I get the general idea	01:04
piroko	danderson: Hehe. Understood	01:04
doener	danderson: git doesn't care whether the branches have a common histoy	01:05
	s/toy/tory/	01:05
piroko	HISTOY	01:05
	Sorry, tired :P	01:05
danderson	doener: I suspected as much, but I wasn't sure if there was some higher level porcelain that made assumptions, and what chicken to sacrifice to disable them :P	01:06
	drobbins\|away → drobbins	01:09
	← francois___ left	01:10
	← nbrunson left	01:10
	argnel returns	01:13
	← paulboone_ left	01:15
tazle_	hm, when I switch to a branch and get a message "Your branch is ahead of the tracked remote branch 'origin/master' by 1 commit." - how do I see what commit it is?	01:15
	← robinr left	01:15
	→ reprore joined	01:17
	→ paulboone joined	01:17
	← paulboone left	01:17
	← cgardner left	01:18
foutrelis	tazle_, I'm guessing it's the last one	01:19
	← jony_ left	01:20
	→ jony_ joined	01:21
tazle_	foutrelis: I would guess so too - but I would like to diff to be sure	01:21
	← bradhaydon left	01:22
	← johnw left	01:22
	← jony_ left	01:22
	→ jony_ joined	01:23
	→ joshdavey joined	01:26
foutrelis	tazle_, I can only think of gitk as a quick way to get a clear picture of the relation between local and remote master	01:26
	drobbins → drobbins\|away	01:27
doener	argnel: hunting some bug right now...	01:27
	→ richcollins joined	01:27
doener	argnel: but you could try what I suggested right before you left	01:27
richcollins	Why do I get this message when the remote repo and the location repo only have one branch?	01:27
	You asked me to pull without telling me which branch you	01:27
	want to merge with	01:27
	→ ketralnis_ joined	01:27
	← ketralnis_ left	01:27
argnel	doener: yup, just trying to figure it out :)	01:28
	← ketralnis left	01:29
	← gaveen left	01:30
	← ankit9_ left	01:30
	← jony_ left	01:30
argnel	doener: so the fetch = :refs/remotes/git-svn before rewrite.. i want to take the trunk/test out of the url, and change the fetch to branches/test:refs/remotes/git-svn ?	01:31
	→ gaveen joined	01:31
	→ jony_ joined	01:31
	→ ankit9_ joined	01:31
	→ madewokherd joined	01:31
	← foutrelis left	01:37
doener	argnel: no	01:41
	← tvakah left	01:41
	→ tvakah joined	01:42
doener	argnel: you drop "/trunk/test" from the url, add trunk/test before the : in the fetch line	01:42
argnel	doener: in the existing fetch line?	01:42
doener	argnel: and add _another_ fetch line: fetch = branches/test:refs/remotes/test	01:42
	yep	01:42
argnel	ok	01:43
doener	that way, you just start tracking the "test" branch in addition to the test directory in trunk	01:43
	and git-svn should figure out the branch point, and not refetch all the stuff	01:43
	so then you have two tracking branches, git-svn (as before) and test (the new stuff)	01:45
	gitk --all should already show the stuff just fine	01:45
	but you might need to manually rebase your local branches, eg. "git rebase --onto test git-svn"	01:45
argnel	so fetch = :refs/remotes/git-svn becomes fetch = trunk/test:refs/remotes/git-svn - adding fetch = branches/test:refs/remotes/newbranch	01:46
	← david_koontz left	01:47
	← tvoet_ left	01:48
doener	yep	01:48
	← adam[pac] left	01:48
argnel	it failed because it can't find a git repo. wtf.	01:48
	→ davidfetter joined	01:49
argnel	ok, typo	01:49
	doener: but now it fails "Unable to determine upstream SVN information from HEAD history."	01:49
	→ gretch joined	01:50
doener	argnel: could you pastebin the current config?	01:50
argnel	doener: http://paste.lisp.org/display/74388	01:51
	← seangrove left	01:51
	→ happybob joined	01:52
doener	argnel: you removed "trunk/test" from the url, not "/trunk/test"	01:52
argnel	gar	01:52
	doener: ok, no failure, but unchanged results	01:53
doener	argnel: after the change, "git svn fetch" should fetch stuff into the new remote tracking branch	01:53
	argnel: and "gitk --all" should show that	01:53
	as a child of the git-svn stuff	01:54
	→ marvil07 joined	01:55
argnel	doener: ok, i see what was wrong, i'd thrown a --no-rebase in there and forgotten about it	01:55
	→ Elazar joined	01:58
doener	argnel: and as I said, it might still try to dcommit to trunk/test initially, you'll probably need the manual rebase	01:59
Elazar	I'm trying to overwrite a local file with what's currently in the remote repo and I'm getting this -- error: X does not look like a v2 bundle file / fatal: Could not read bundle. Suggestions?	01:59
	← ijcd left	01:59
argnel	doener: which i'm trying to do now, however its not actually doing anything	01:59
doener	argnel: or you could try to remote the old fetch line after the "git svn fetch", then a plain "git svn rebase" might also do the trick	01:59
	s/remote/remove/	01:59
	argnel: git rebase --onto newbranch git-svn	02:00
	argnel: that doesn't do anything?	02:00
argnel	doener: error, wants a range	02:00
doener	for the rebase?	02:00
	oh wait... wants a range... that rings a bell...	02:01
argnel	$ git rebase --onto newbranch git-svn First, rewinding head to replay your work on top of it... fatal: Not a range. Nothing to do.	02:01
doener	ancient git version?	02:01
	← lois[pac] left	02:01
doener	but that should actually have done something already IIRC	02:01
	try "git rev-parse HEAD newbranch git-svn"	02:01
	→ akitada joined	02:01
doener	the first and the second value should be identical	02:02
argnel	doener: its git.git 9a01387	02:02
	(my git version)	02:02
	← softdrink left	02:02
doener	hm, pretty recent	02:02
	thought the "wants a range" thing was fixed at some point	02:03
	→ Leefmc joined	02:03
doener	anyway, what does the rev-parse tell?	02:03
argnel	doener: first and third the same, middle different	02:03
doener	ok, so it seems to have gone back	02:03
	well, for this simple case, reset --hard newbranch works as well	02:03
	and when you actually _have_ commits to rebase (ie. not yet dcommitted), then the rebase should also do	02:04
	and rebase -i might work in any case	02:04
argnel	i have commits to dcommit in this test :)	02:04
doener	not on your current branch	02:05
	otherwise the commits for HEAD and git-svn wouldn't be equal	02:06
argnel	i keep forgetting i'm working on a different version of the script :)	02:08
	doener: i removed the git svn dcommit, now all three are (properly) different	02:08
doener	argnel: then the rebase should work now, too	02:09
	(and as I said, with -i, it works anyway)	02:09
argnel	mais oui!	02:09
	so whats it going to dcommit to?	02:09
	← drwilly left	02:10
doener	should dcommit to branches/test	02:10
	dcommit -n will tell	02:10
argnel	indeed, it does	02:11
	← ahupp\|fb left	02:11
	← EmilMedve left	02:11
	→ EmilMedve joined	02:12
argnel	doener: why does it select branches over trunk?	02:12
doener	because the latest commit (in first-parent history) comes from branches/test	02:13
	← Tv left	02:14
doener	argnel: to determine its upstream, git-svn looks at the git-svn-id lines in "git log --first-parent"	02:14
	→ kandjar joined	02:14
doener	and then uses the first one that's consistent with its .revmap stuff	02:14
kandjar	Hi there,	02:14
	Is there a way using git to have sensitive information and public information in the same repository (let say, you have a project, you want the header files to be public but the source files to be private)	02:15
	← RayMorgan left	02:15
	→ bdimcheff_ joined	02:16
	← bartman left	02:16
context	kandjar: no. use submodule's	02:17
kandjar	submodules will manage different git repository	02:18
argnel	doener: well, thanks very much for explaining this, the git-filter-branch method was becoming a nightmare from which there was no escape :)	02:18
DrNick	alternately, why even make your source control public if you don't want most of it to be public?	02:18
kandjar	your sensitive files and public files may bein the same folder, how to you manage that?	02:18
	that sounds painful	02:18
context	drnick: for linking to it	02:19
DrNick	linking what to it?	02:19
context	actually. yeah	02:19
	why make the repo accessible if the public can only see the headers	02:19
	wanna go 'here are the new headers, but you cant use them yet'	02:20
	← nuncanada left	02:20
	← bdimcheff left	02:22
kandjar	context: dont you think they would want to see the history too?	02:28
	← david455 left	02:29
	← SRabbelier left	02:30
	← akitada left	02:30
	← perezd88 left	02:31
	→ cookiecaper joined	02:33
cookiecaper	Hi. I want git to monitor my PostgreSQL database as well as my workspace files, but the files PGSQL uses to comprise its database are not in my project's working tree and I don't want to change that. I see that git doesn't follow symlinks. Anyone have any recommended	02:34
	methods for getting git to cover my database too?	02:35
	→ seangrove joined	02:36
cehteh	dump the database as text into your working dir regulary	02:36
	(pre-commit hook?)	02:36
	git wont handle a constantly changeing binary blob too well either	02:37
	← Elazar left	02:37
	→ ToxicFrog joined	02:37
cookiecaper	eh, I guess that's true. OK, I'll look into that, thanks.	02:37
	← halfdan left	02:38
	→ s0ber_ joined	02:43
	→ aresnick joined	02:44
aresnick	Is there a way to re-checkout a single file (I've deleted a file I'd like to undelete, but I don't want to undo the other changes I've made since the last push)	02:45
doener	git checkout HEAD -- file	02:45
aresnick	doener Ah, OK--thanks!	02:45
	← gwoo left	02:46
	← s0ber left	02:51
	← ankit9_ left	02:51
	→ ankit9_ joined	02:51
aresnick	If a large, binary file is added to a git repo, then a long time later removed, does the size of that initial file put a lower bound on how small the repo can get (since presumably, you can always work back to it?)	02:52
doener	yep	02:53
	← seangrove left	02:53
aresnick	doener: OK; I inherited such a repo =) Is there any way to remove it from the history?	02:53
context	kandjar: no.	02:53
	just keep a changelog	02:54
aresnick	doener: That is, to stop tracking, and to tell git to delete the related history of, a given blob?	02:54
doener	aresnick: filter-branch	02:54
	← cartucho left	02:54
	← richcollins left	02:54
Nafai	When I'm trying to push to a remote repo, it failed. Is there any way to find out more of why it failed?	02:54
	-v doesn't say why	02:55
doener	Nafai: what _does_ it say?	02:55
aresnick	doener: Jesus--that's awesome	02:55
	doener: Thanks!	02:55
piroko	aresnick: Keep in mind that will rewrite your history.	02:56
Nafai	! [rejected] master -> master (non-fast forward)	02:56
context	nafai: -f	02:56
doener	Nafai: well, there you have the reason	02:56
	← ice799 left	02:56
context	nafai: did you change history after pushing ?	02:56
	dont -f unless you know what you're doing	02:56
doener	Nafai: http://git.or.cz/gitwiki/GitFaq#head-a3da5acfa0f5411f1211b43e4907e844639971f1	02:57
Nafai	context: I haven't changed the history at all	02:57
doener	Nafai: probably someone else pushed in the meantime	02:57
Nafai	I'm the only one pushing to this repo	02:57
aresnick	piroko: Yep	02:58
doener	Nafai: did you use rebase?	02:58
	→ muthu_ joined	02:58
Nafai	I don't think so	02:59
doener	Nafai: try "git fetch" and then "gitk --left-right origin/master...master"	02:59
Nafai	Hrm, it is slightly different	02:59
doener	that shows you those commits that exist only on in origin/master _or_ in master	02:59
	→ alus joined	02:59
doener	i.e. how they differ	03:00
alus	I have local changes which have been committed but not pushed	03:00
doener	might give you a clue what happened	03:00
alus	is there a way to branch and push the commits there instead?	03:00
doener	alus: git branch <new_name>	03:00
Nafai	doener: Is there an easy way do get the same info with gitk, this is on a remote machine	03:00
doener	Nafai: git log --graph ... should do, but not quite as nice	03:01
Nafai	Thanks	03:01
doener	Nafai: but maybe with --pretty=oneline it's ok	03:01
alus	doener: ok. after someone else checks that out and we're done, how do I land the branch?	03:01
doener	alus: after creating the new branch, you can reset your current branch	03:01
alus	how?	03:01
doener	eg. to origin/master or whatever	03:01
Nafai	doener: Thanks	03:02
doener	git reset --hard origin/master (bye bye uncommitted changes)	03:02
alus	uh, how do I tell what it is now?	03:02
	ah, just git branch	03:03
	so again, how do I land the branch?	03:03
doener	to push a new branch, you need to explicitly name it in a refspec to push	03:03
	e.g. "git push origin my_new_hot_branch"	03:03
	→ TokyoDan joined	03:03
alus	got it	03:03
doener	that's because "git push <remote>" defaults to "git push <remote> :"	03:04
	and the ":" refspec means: update branches that exist in both, the local and the remote, repo	03:04
	s/,//2	03:04
	→ offby1` joined	03:04
alus	uh	03:04
	I did "git branch mybranch"	03:04
	then push seems to have pushed to master?	03:04
doener	push without a refspec?	03:05
alus	yes :/	03:05
doener	yeah, expected, see above.	03:05
	← davidfetter left	03:05
alus	dang.	03:05
	undo?	03:05
doener	git show origin/master@{1}	03:06
	is that the commit you want to go back to?	03:06
alus	yes	03:06
doener	if so: git push <remote> +origin/master@{1}:master	03:06
	the "+" forces the push (as it's not fast-forward)	03:06
alus	what is "<remote>" there?	03:06
doener	and you better hope that nobody fetched in while the "bad" stuff was up	03:06
	s/in while/while/	03:06
	ok, so you did the "git branch" to create the new branch	03:07
alus	yes	03:07
doener	then you still have the old branch checked out	03:07
alus	I guess so, sure	03:07
doener	now you want to reset that old branch back	03:07
	most likely to origin/master	03:07
alus	right	03:08
	← aziz left	03:08
doener	so: git reset --hard origin/master (make sure there are no uncommitted changes)	03:08
alus	uhm	03:08
doener	... or that you don't care about them	03:08
	the reset will drop them	03:08
alus	sorry. "git push <remote> +origin/master@{1}:master" will roll back master to the revision I want, but what is "<remote>" in that command?	03:08
doener	for example, "origin"	03:09
mugwump	anything from the list 'git remote' returns	03:09
doener	"git remote" gives a list of remotes that you have	03:09
	← Niamor left	03:09
alus	ok	03:09
	that seemed to... happen	03:09
	so now I have to check out the mybranch branch seperately? or can I push commits to it directly from this checkout?	03:10
	→ johnw joined	03:10
	→ Niamor joined	03:10
doener	you can push right away	03:10
	oh wait, push to it?	03:10
	→ Tv joined	03:10
alus	I have committed but now, uh, unpushed? changes	03:11
doener	no, if you want "git commit" to affect "mybranch", you'll have to check it out first	03:11
alus	ok. I have already committed some stuff, that I accidentally pushed earlier	03:11
doener	yeah, and mybranch references that	03:11
	when you use "git branch <name>" that branch will start at the current commit	03:11
alus	ok..	03:12
doener	so "mybranch" is currently at the commit at which master was when you created the branch	03:12
alus	so how can I push those to mybranch?	03:12
doener	you can only push to repos	03:12
alus	wow. ok. terminology mismatch	03:13
doener	what is it that you want to do? Push the mybranch branch to the other repo?	03:13
alus	what is a branch?	03:13
	I mean to create something which my friend can check out from another pc	03:13
doener	let's go back a bit...	03:14
alus	ok	03:14
doener	so you made commits on master	03:14
alus	right	03:14
doener	and then you did "git branch mybranch"	03:14
alus	right	03:14
doener	ok, so "mybranch" has all the commits that were in master	03:14
	try "git log mybranch"	03:14
	that should show them	03:15
alus	it does	03:15
	but so does "git log"	03:15
doener	sure, because "master" still contains those commits as well	03:16
alus	ok	03:16
doener	you probably didn't do the reset yet, so master is unchanged	03:16
	does "git status" tell you anything about uncommitted changes?	03:16
	← tango_ left	03:16
alus	just some untracked files, as expected	03:16
doener	ok, and "git branch" says "* master" (besides other things), right?	03:17
	i.e. you still have "master" checked otu	03:17
	s/otu/out/	03:17
alus	it says "* master\nmybranch"	03:17
doener	ok	03:17
	then "git reset --hard origin/master" to reset the checked out branch to origin/master	03:17
alus	HEAD is now at d356acd Merge branch 'master' of blah	03:18
doener	ok, now "git log" should be different	03:18
alus	yes	03:18
doener	and "git log master..mybranch" should show those other commits	03:18
	→ libc_ joined	03:19
alus	yup	03:19
doener	ok, and now you can push your mybranch to the remote	03:19
	git push origin mybranch	03:19
alus	wait, what does that mean?	03:20
	landing the branch?	03:20
doener	that expands to "git push origin refs/heads/mybranch:refs/heads/mybranch"	03:20
	← bdiego_away left	03:20
alus	ok, so just making it available at the server	03:21
doener	so it pushes the branch "mybranch" from this repository to the branch "mybranch" in the "origin" repository	03:21
alus	ok	03:21
	← jeckel left	03:21
alus	that seems to have worked	03:21
	great! everything seems to be in order	03:22
	← yonkeltron left	03:22
alus	doener: thanks :)	03:22
doener	np	03:22
alus	so. now how does my friend get a copy of mybranch?	03:23
doener	assuming that he cloned the other repo, he should also have a remote called "origin"	03:23
	so he can do "git fetch origin"	03:23
	that will fetch the new branch and store it as origin/mybranch	03:23
	and then he can create a local branch from it using "git branch mybrach origin/mybranch2	03:24
	s/2/"/	03:24
	→ tango_ joined	03:24
doener	or "git checkout -b mybranch origin/mybranch" if he wants to check it out at the same time	03:24
	alus: try "git branch -r"	03:24
	→ yonkeltron joined	03:24
doener	alus: that should show origin/mybranch for you, too	03:24
alus	yes	03:24
doener	the extra argument to "git branch" just sets the starting point for the new branch	03:25
alus	sorry, that git checkout command will create a directory?	03:25
	→ cytrinox_ joined	03:25
doener	so instead of starting at the current commit, it starts at origin/mybranch	03:25
	no, "git checkout -b mybranch origin/mybranch" is the same as "git branch mybranch origin/mybranch && git checkout mybranch"	03:26
	git uses in-tree branch switching	03:26
alus	ok	03:26
	← libc left	03:26
alus	ok, pulling from this state	03:28
	"git pull" complains, as does "git pull nokogiri"	03:29
doener	pull is fetch + merge	03:30
	"git pull <remote> <refspec>" is the full command	03:30
	for example, "git pull origin master" will fetch the master branch from origin and merge it into your current branch	03:30
alus	ah, I see	03:31
doener	"git pull" without arguments works only when the branch has configured defaults	03:31
alus	oo, how do I configure those?	03:31
doener	if you open .git/config, you'll see a [branch "master"] section in there	03:31
	with two entries, remote and merge	03:31
alus	I see	03:31
doener	remote should be abvious, and merge is the name of the branch on the remote	03:31
alus	makes sense	03:32
doener	for the master branch, it's likely refs/heads/master	03:32
alus	yup	03:32
doener	so you can either copy and adapt that section for your new branch, or use "git config"	03:32
	→ Elazar joined	03:32
doener	git config --add branch.nokogiri.remote origin	03:32
	or whatever	03:32
alus	right	03:33
doener	that stuff gets setup automatically (by default), when you use "git branch <name> <remote-tracking-branch>"	03:33
	where <remote-tracking-branch> is one of the branch shown by "git branch -r"	03:33
	as you had no remote tracking branch yet, when you created the branch, no such setup was done	03:34
Ilari	--add should only be used with settings that make sense when duplicated (such as remote fetch/push settings).	03:34
doener	ah right... finger memory...	03:34
	← jony_ left	03:35
Elazar	Running 1.6.1, having a problem trying to fetch. Tried doing a clean clone and repeating the fetch, but get the same error. Error message + .git/config: http://www.pasteninja.com/paste/2161. Suggestions?	03:35
	→ gwoo joined	03:35
doener	Elazar: path/to/file? What's that supposed to be?	03:35
Elazar	doener: The path to the file I'm trying to fetch from the remote repo to overwrite the local copy?	03:36
doener	Elazar: that's not how it works	03:36
	Elazar: you can use "git fetch origin" to update the remote tracking branches for that remote	03:36
	Elazar: and then you can use, for example, "git checkout origin/master -- path/to/file"	03:36
	to check that file out of the origin/master branch	03:37
Elazar	doener: Ah. Yup, that did it. Thanks then. :)	03:37
	← cytrinox left	03:37
	→ muthu joined	03:44
	→ foca joined	03:47
	← aresnick left	03:51
	← parasti left	03:55
	→ ivanoats joined	03:57
	← jarrettgreen_ left	03:58
	← muthu_ left	03:59
	→ jim_c_ joined	04:00
jim_c_	Evening all.	04:00
	git diff HEAD..origin will show me all the diff hunks if I push to origin. Is there a way to see what changesets will be pushed?	04:00
RandalSchwartz	that's git log	04:00
	log lists commits	04:00
	what you're calling "changesets"	04:01
	← kukks left	04:01
jim_c_	is there a command which will show me which commits differ between a local branch and a remote branch?	04:01
RandalSchwartz	git log localcommit...remotecommit	04:02
jim_c_	In mercurial, I'd use `hg outgoing`	04:02
	→ girishr joined	04:03
jim_c_	without trying to be obtuse :-), I'm not getting the syntax right	04:04
	`git log master...origin/master1 gives me no output	04:04
RandalSchwartz	then those are identical, likely	04:05
jim_c_	They aren't though, because I cloned a project from github, and made 1/2 dozen local commits.	04:05
rtyler	spearce: ping	04:06
jim_c_	!@#$	04:06
rtyler	spearce: is JGit usable outside of the context of just the Eclipse plugin	04:06
jim_c_	Too many terminal windows open, and I was in the wrong clone.	04:06
	→ Kathrin-26^away joined	04:06
spearce	rtyler: yes. Gerrit2 (review.source.android.com) is entirely based on it	04:06
jim_c_	That does do what I want. Thanks RandalSchwartz.	04:06
rtyler	spearce: I'm pitching it for use as part of a rewrite of the Hudson git plugin	04:06
	spearce: these poor souls are spoiled on the idea because of SVNKit suck	04:07
	:/	04:07
	spearce: are there no code level docs running around? :/	04:07
spearce	not that many, no. jonas started some docs in the code.google.com/p/egit wiki but nobody has done anything to finish the sections out more.	04:07
	mostly its "look at existing applications; e.g. netbeans, egit, gerrit2, and the .pgm package that come with"	04:08
	← girishr left	04:10
	← exg left	04:10
	→ paulboone joined	04:10
jim_c_	Sometimes git is infuriating to use. `git log master..origin/master` I dropped one of the ".". status is 0, no output, no error message. </rant>	04:10
	→ exg joined	04:10
rtyler	spearce: http://www.nabble.com/Re%3A-Re%3A-git2-plugin-and-scm-polling-in-a-multi-configuration-project-p21681183.html	04:12
	this is what I'm up against >_<	04:13
	rtyler is crafting a reply	04:13
gouki	How is the 'owner' field on gitweb populated?	04:13
	→ jarrettgreen_ joined	04:13
	← RaySl left	04:14
	→ bryanl joined	04:16
	← ivanoats left	04:17
	← muthu left	04:17
spearce	wtf is hudson processing gigs of data through git for?	04:17
	← Phoenixfire159 left	04:17
rtyler	spearce: I don't know wtf he's referring to there	04:17
	I can tell you firmly that my Git repo is bigger than his :P	04:17
	rtyler giggles	04:17
rtyler	I can CC you on the reply if you'd like :P	04:18
	there's fracturing with the git plugin(s), insofar that there's "git" and "git2"	04:18
spearce	oh, cute. so they can't even agree on a plugin code base, let alone how it interacts with the object store.	04:18
rtyler	git2 was written because "git" does nasty stuff currently with tags and git-rev-list(1)/git-rev-parse(1)	04:18
	funny thing is	04:19
	we switched to git2 because the "cli client" had a stupid stdout parsing issue when it was built on top of porcilean and we switched to 1.6	04:19
	(I used the porcilean term right?)	04:19
spearce	yea, you shouldn't ever build on top of porcelain... bad bad idea	04:20
	plumbing is your friend if you are invoking the cli tools	04:20
	← simosx left	04:20
	← jim_c_ left	04:22
gouki	Just to let know that I found the answer to my question: http://lists-archives.org/git/658697-where-does-gitweb-take-the-owner-name-from.html	04:23
rtyler	do you want to be included in the conversation? or shall I be a puppet JGit evangelist? :P	04:23
	← Kathrin-26^away left	04:23
	→ girishr joined	04:23
	← jarrettgreen_ left	04:23
spearce	yea, cc me	04:23
rtyler	spearce at spearce org?	04:24
spearce	yup	04:24
rtyler	nothing like a good beer and mailing lists	04:25
	that guy's from redhat too	04:27
	he should know better	04:27
	→ vbabiy joined	04:27
rtyler	spearce: sent	04:27
	← alus left	04:29
	→ Kathrin-26^away joined	04:29
	← krawek left	04:30
	← Kathrin-26^away left	04:32
rtyler	spearce: if i've misrepresented JGit at all, feel free to publicly correct me, I'm far more concerned with having a good Hudson plugin than appearances :P	04:32
	→ rgk joined	04:33
	→ Kathrin-26^away joined	04:34
rgk	hi, im hosting a git repo on my server but when i go to clone it i get a connection refused, is there any configuring i need to do to make it work? i remember i had this problem in the past but i fixed it (on an other server) just can't figure it out again.	04:34
spearce	rtyler: no, you didn't misrepresent it at all	04:37
	aside from slugging around 7 GB pack files over the network, it keeps up with C git very well performance wise, and there's no compatibility issues. nor would there really be, as like you pointed out, we're rather close to the C git people	04:38
	← jpeeler left	04:38
	→ jpeeler joined	04:39
ftehw	rgk, i get the same issue when trying to connect under ssh (connection terminated remotely).	04:43
	→ volante joined	04:44
ftehw	...trying to diagnose the issue now	04:44
	drobbins\|away → drobbins	04:44
	→ bradhaydon joined	04:44
bradhaydon	can anyone tell me if it's possible to push back to a subtree's external repo?	04:47
	← TokyoDan left	04:49
	← girishr left	04:50
rgk	ftehw: sorry it was actually a problem with git-daemon but now i have the error fatal: The remote end hung up unexpectedly	04:50
	→ girishr joined	04:51
	← gwoo left	04:51
	→ jony_ joined	04:51
	→ ezyang joined	04:54
	→ esparkman joined	04:54
ezyang	Hello all; I'm currently planning on using git-svn as an interface between an SVN repos and a local git working copy. I wonder if there were any documents on good "Git sanitization", such that git and svn play nice	04:55
	(If I had the choice, I'd switch the SVN repos over to git, but I don't)	04:55
	→ achiang joined	04:56
rgk	my problem is i can clone the git repo through ssh but with the git:// url it won't work and i get the error fatal: The remote end hung up unexpectedly any ideas?	04:57
	← nothingHappens_ left	04:58
ezyang	rgk: Who's running the server?	04:59
	bdimcheff_ → bdimcheff	04:59
ezyang	It's probably a server problem, but if you're trying to clone off of Github we can rule that out	05:00
achiang	i have a file foo.c, the contents of which i've completely replaced (but kept the same filename foo.c) -- how do i convince git to produce a more readable diff? it seems to be keying in on things like curly braces, etc. and replacing lines around those...	05:01
	example: http://pastebin.com/m20537e60	05:01
	← jony_ left	05:01
	← Beket left	05:01
ezyang	achiang: You can tell Git to use a different diff utility, but really, it	05:01
	's expected behavior	05:01
	I assume you want to submit a patch?	05:02
achiang	ezyang: hrm, if you say so... it just produces a diff that's really hard for reviewers to read	05:02
ezyang	Yeah	05:02
	→ jony_ joined	05:02
achiang	ezyang: yeah, i'm working on some pci patches	05:02
	← spearce left	05:03
	← ceej_ left	05:03
achiang	guess i'm out of luck?	05:03
	← reprore left	05:04
ezyang	There are things you can do	05:04
achiang	i was thinking of jumping through some hoops, like git-rm the file as one patch and then adding the file back in as a second patch	05:05
ezyang	You can do that, assuming removing the file doesn't break the system	05:05
	← paulboone left	05:05
ezyang	(people don't like it when you do that)	05:05
	You could also just simply give the file a new name, if names aren't important	05:05
achiang	well, i'd have to remove it from the makefiles, Kconfig stuff, etc. to keep bisectability	05:05
ezyang	Yeah	05:05
achiang	hrm, i could rename the source file, but yes, keep the name of the kernel module it produces the same	05:06
	maybe.	05:06
	that would involve digging into the Kconfig stuff a bit more to figure out how feasible that would be	05:06
ezyang	Hmm	05:10
	→ tjafk2 joined	05:10
	→ neurodamage joined	05:10
	→ bdiego joined	05:10
neurodamage	hey what's up folks longtime, I forget when using git, how do I force the current version I'm commiting that's getting rejected to be the master on the remote repo?	05:10
	I tried git -f push	05:10
ezyang	neurodamage: try git push --force	05:11
neurodamage	heh, yeah I just saw that in the man page, darn, thanks	05:11
ezyang	also, please remember the problems associated with doin gthis	05:11
neurodamage	ezyang: ?	05:11
	I'm the only developer	05:12
ezyang	neurodamage: Ah, ok	05:12
neurodamage	:)	05:12
	not normally something I would do otherwise, heh	05:12
	← Kathrin-26^away left	05:13
	← Leefmc left	05:14
	→ Kathrin-26^away joined	05:14
rgk	ezyang: me	05:15
	its a linode server	05:15
ezyang	Ok. Check if git-daemon is running	05:16
	→ ProperNoun joined	05:16
	← vbabiy left	05:17
ezyang	Check if the port's open	05:17
ftehw	curious about the .git suffix. examples online are typically `git clone git://whatever.com/foo.git`. oddly the dir that is created locally is 'foo'. what's up with that?	05:17
bradhaydon	anyone familiar with subtrees?	05:18
ezyang	ftehw: git clone will automatically munge out the suffix .git, since that indicates a bare repository and you usually clone a working copy	05:18
	bradhaydon: Ask your question	05:18
	→ dbb joined	05:18
ftehw	OIC. that's where the 'bare repo' comes in. thanks.	05:18
bradhaydon	ezyang: How do I commit and push back to the subtree's external repo?	05:19
	→ seangrove joined	05:20
ezyang	bradhaydon: You don't mean submodules, do you?	05:20
bradhaydon	ezyang	05:20
	http://www.kernel.org/pub/software/scm/git/docs/howto/using-merge-subtree.html	05:20
	← resmo left	05:20
bradhaydon	whoops :)	05:20
ezyang	Yeah, it says right there that it's kind of hard to push your changes to the subtree	05:22
	You should use submodules	05:22
	← drewr left	05:22
rgk	ezyang: 9418/tcp open unknown	05:22
	it seems to be open	05:22
kevlarman	rgk: do you have a file named git-daemon-export-ok in your .git?	05:23
rgk	thats it!	05:23
	← marvil07 left	05:24
rgk	nvm i have that file	05:24
	← seangrove left	05:25
	← tjafk1 left	05:27
	→ halfdan joined	05:27
ezyang	Hmmm	05:28
	Are there any other error messages?	05:28
	Also, what command line are you using to start git-daemon?	05:28
	drobbins → drobbins\|away	05:32
	→ dgrant joined	05:34
	← achiang left	05:35
	← beachsurfin left	05:38
	tav_ → tav	05:39
rgk	ezyang: i fixed it, all i did was restart git-daemon and it seems to be working now	05:39
	thanks for your help	05:39
	→ dbb_ joined	05:39
ezyang	great :-)	05:40
	(next time someone asks this question, I'll tell them to powercycle) :-)	05:40
	→ ahupp\|fb joined	05:42
rgk	haha	05:43
	← niki left	05:43
	→ peritus_ joined	05:43
	← esparkman left	05:45
	→ spearce joined	05:48
	→ beachsurfin joined	05:48
	← ahupp\|fb left	05:49
	→ cartucho joined	05:51
	→ muthu_ joined	05:52
	← cartucho left	05:52
	← Niamor left	05:52
	← stephenlb left	05:53
	→ Niamor joined	05:53
	← dbb left	05:55
	→ RayMorgan joined	05:56
	← hobodave left	05:56
	← zpoley left	05:57
	← spearce left	05:57
	← dbb_ left	05:58
	→ bdiego_ joined	06:00
	← bdiego left	06:00
	← peritus_ left	06:01
	← BugeyeD left	06:02
	→ manguy joined	06:02
	→ BugeyeD joined	06:02
manguy	I'm looking to stop using svn and completely switch to git. If I use git svn to convert an svn project to git, after that, the project just becomes like a regular git project right?	06:03
	→ mvanholstyn joined	06:03
	→ gwoo joined	06:03
johnw	manguy: you still need to transfer the tags manually	06:04
	git-svn will have made them all into branches	06:04
	don't know if there's an option for that or not	06:04
	also, you don't need the metadata	06:04
	so use the clone option which omits it	06:05
	← curvature left	06:05
manguy	johnw: Ah, I don't have any tags or anything like that. I'm using svn at the most basic level	06:05
	→ curvature joined	06:05
johnw	then in that case you're fine	06:05
	→ zpoley joined	06:07
	← dreiss left	06:08
	→ zpoley_ joined	06:08
manguy	johnw: Also, perhaps you might not know, but how can I get git-svn on Mac OSX? It doesn't seem to recognize the command. Is that part of the swik package?	06:08
	→ Tuomas joined	06:08
	→ albertito joined	06:08
	← lordpil left	06:09
	→ lordpil joined	06:09
	← zpoley_ left	06:09
johnw	how did you install git?	06:10
	→ zpoley_ joined	06:10
	← albertito left	06:10
manguy	johnw: Compiled from source	06:10
johnw	ah, I'm not sure then	06:10
manguy	But perhaps I should have used mac ports	06:10
johnw	I used MacPorts: port install git +svn	06:11
	← mvanholstyn left	06:12
	→ mvanholstyn joined	06:12
	← mvanholstyn left	06:12
kevlarman	manguy: if you're using recent git, then git-svn (as opposed to git svn) won't work	06:12
johnw	ah, true	06:13
	→ almery joined	06:13
manguy	Should I just do a "port install git +svn" then?	06:14
	→ TokyoDan joined	06:14
	→ mithro joined	06:16
johnw	or try "git svn"	06:16
	but MacPorts will also keep you up-to-date easier :)	06:17
	→ melvinram joined	06:18
	→ fynn joined	06:18
fynn	uhm hi!	06:18
	I have a diff that looks like this:	06:18
	→ paisal joined	06:19
fynn	- foo = bar	06:19
	\ No newline at end of file	06:19
	+ foo = bar	06:19
	I don't see what changed here. Can anyone explain?	06:19
	← melvinram left	06:19
johnw	just read what you pasted again, closely :)	06:20
	fynn reads closely and still doesn't understand	06:20
johnw	"No newline at end of file" is true of one file, but not the other	06:20
	← bradhaydon left	06:20
	drobbins\|away → drobbins	06:21
fynn	right, but I can't see how it is; I deleted everything in the new file as well.	06:21
	→ binjured joined	06:21
fynn	the new file literally ends with "foo = bar".	06:21
johnw	how are you testing that assertion?	06:21
fynn	using trusty vim editor.	06:21
	→ hobodave joined	06:21
	← xenoterracide left	06:22
johnw	i didn't know vi could distinguish those two cases	06:23
	emacs can	06:23
	also try "cat -v"	06:23
fynn	I thought if the file ends with a newline then there would be a literal empty line at the bottom of the vi buffer?	06:24
	← zpoley left	06:24
johnw	no	06:24
	vi shows you the same thing either way	06:24
	in Emacs, the only way you can try is typing M->	06:24
	if it has a newline at the end, point will go to its own line	06:25
	otherwise, it will stop at the end of the last line	06:25
	here's another way to test:	06:25
	wc -c foo1.txt foo2.txt	06:25
	you'll see that the byte count differs	06:25
	or, look at the size of the file :)	06:25
fynn	Interesting... why wouldn't vi / emacs show a blank line at the end of a file that ends with a \n?	06:27
	→ reithi joined	06:27
	→ j416 joined	06:28
j416	"Namespace refs/original/ not empty"	06:28
	→ cartucho joined	06:28
j416	what do i need to do to get rid of this?	06:28
	.git/refs/original doesn't exist.	06:28
	o_O	06:28
	← qrush left	06:28
j416	(i removed it already)	06:28
fynn	johnw: thanks in any case, that was interesting :)	06:29
	→ qrush joined	06:29
	→ NSaibot joined	06:29
	← cartucho left	06:30
johnw	refs/original is left after doing a filter-branch	06:31
	you can say filter-branch -f to ignore it	06:31
j416	johnw: that's the thing	06:31
	-f shouldn't be needed	06:31
	→ mczepiel joined	06:32
j416	since there is no refs/original	06:32
	this must be a bug	06:32
johnw	again: you're sure?	06:33
	git doesn't usually say stuff unless it means it for some reason	06:33
j416	i will try to reproduce it.	06:34
johnw	you aren't using a different GIT_DIR than you think, are you?	06:34
	→ priidu joined	06:35
j416	ls ./git/refs/ shows "heads" and "tags". git status is clean. git filter-branch --tree-filter 'rm -rf mydirectory/something' gives "Namespace refs/original not empty"	06:36
johnw	ls .git/refs, you mean?	06:36
	→ Beket joined	06:37
	← paisal left	06:37
johnw	also, refs/original might be packed	06:37
	grep original .git/packed-refs	06:37
	← NSaibot left	06:37
fynn	OK, I get this weird error message trying to "git push"	06:38
	! [rejected] master -> master (non-fast forward)	06:38
	What does it mean and how do I heroically overcome it?	06:39
johnw	fynn: because you use commit --amend, or filter-branch, or rebase	06:39
fynn	wut wut	06:39
	I do no such thing.	06:39
johnw	you rewrote a commit, so now the HEAD of your remote branch is not an ancestor of your current master's HEAD	06:39
	← mithro left	06:39
johnw	you did something	06:39
fynn	Dear lord. Somebody else must have done it. Anyways, what do I do to push?	06:39
johnw	origin/master is not in the history of master	06:39
	push -f	06:39
	but beware	06:39
	that's only safe if you're the only user	06:40
fynn	I am not :(	06:40
	Maybe I should git pull?	06:40
johnw	then you shouldn't do it	06:40
	if you git pull, it's going to merge your master with remote's master	06:40
j416	johnw: yes, ls .git/refs/ sorry	06:40
johnw	which may also not be what you want	06:40
	→ ph^ joined	06:40
fynn	can I somehow preserve my modifications?	06:40
	before I do that?	06:40
johnw	fynn: my recommendation is this	06:40
j416	aaaaah	06:40
	that's it	06:40
	thank you	06:41
johnw	type "git diff HEAD origin/master"	06:41
j416	it was packed	06:41
johnw	see what your local has	06:41
	then "git checkout -b local" to create a branch	06:41
j416	how do i remove the packed commit??	06:41
johnw	then "git checkout master ; git reset --hard origin/master", to get your master branch back to tracking the head	06:41
	and now, decide if you want to start cherry picking in commits from "local"	06:41
	j416: perl -i -ne 'print unless /original/;' .git/packed-refs; git gc	06:41
fynn	johnw: OK, dthe diff returns nothing	06:42
johnw	fynn: reall	06:42
fynn	did I mention I didn't commit yet?	06:42
johnw	really	06:42
fynn	yes sirry bob	06:42
johnw	but your commits MUST have changed	06:42
	anyway, do git checkout -b local	06:42
fynn	I haven't even committed	06:42
johnw	save your commits on a branch	06:42
	→ mithro joined	06:42
johnw	then "git checkout master"	06:42
	then "git reset --soft origin/master"	06:42
	← Niamor left	06:42
johnw	that'll get you where you want to be	06:43
	→ Niamor joined	06:43
fynn	OK, what 'git checkout -b local' does is branch off my current working tree to a local branch called "local"?	06:43
johnw	your HEAD will match remote HEAD, but your working tree will remain as it is	06:43
	fynn: it branches off your HEAD commits to a branch called "local", it doesn't do anything with your working tree	06:43
fynn	Hm, OK, should I commit my changes first?	06:43
	→ peritus_ joined	06:44
fynn	because I didn't commit, just made changes to files.	06:44
johnw	fynn: you don't need to	06:44
j416	johnw: your script doesn't make sense to me, did you miss a ">" ?	06:44
johnw	reset --soft only changes HEAD, it doesn't alter working tree flies	06:44
	j416: no, it edits the file in place	06:44
	that's what -i does	06:44
fynn	OK, Imma try	06:44
	wish me luck!	06:44
j416	ah sorry. I am no perl-oneline-guru :)	06:44
	so in plain english, remove the line from packed-refs and do a git gc?	06:45
johnw	yep	06:45
	line or lines	06:45
fynn	johnw: I did everything you said	06:46
	now 'git diff' shows nothing	06:46
j416	great, thank you!	06:46
	← Kathrin-26^away left	06:46
fynn	what should I do?	06:46
johnw	fynn: what does "git status" say?	06:46
	→ Kathrin-26^away joined	06:46
johnw	j416: ur welcome	06:46
	← bdimcheff left	06:46
fynn	nothing added to commit but untracked files present	06:46
j416	hm, git status does something more than showing status right?	06:47
	→ robinr joined	06:47
johnw	fynn: do those untracked files contain the work you did?	06:47
	→ Weaselweb joined	06:47
fynn	johnw: nope.	06:47
	my changes were on tracked files.	06:47
johnw	you said "reset --soft", right?	06:47
j416	i unpacked a zipped backup of my repo and work dir, and git reports the working dir is dirty	06:47
fynn	I c&pied verbatim	06:47
	→ ahupp\|fb joined	06:47
j416	but after doing git status, it's ok again	06:47
	o_O	06:47
fynn	bah I lost all the changes	06:47
johnw	that's very odd	06:47
	fynn: wait, you may not have	06:48
	what does "git diff local master" say?	06:48
fynn	nothing :(	06:48
	All the changes on the working tree were reverted.	06:48
johnw	i feel very badly, fynn, but I can't think of any command I told you that would have changed your working tree	06:49
fynn	Luckily I saved a diff	06:49
johnw	can you type "history" and show me the commands you typed?	06:49
fynn	johnw: it's OK, I saved a diff	06:49
	sure:	06:49
johnw	i'd like to know where I went wrong	06:50
	i should have had you stash the changes first :(	06:50
	i.e., "git stash && git stash apply"	06:50
fynn	johnw: http://dpaste.com/113764/	06:50
johnw	it's great you have the diff	06:50
	patch -p0 < foo.diff	06:50
	ah, there's the problem	06:51
	you did reset --hard, not reset --soft	06:51
fynn	--hard	06:51
	johnw> then "git checkout master ; git reset --hard origin/master", to get your master branch back to tracking the head	06:51
	":)	06:51
johnw	omg	06:51
fynn	I c&pied verbatim like the dolt I am	06:52
johnw	my deepest apologies	06:52
fynn	no worries	06:52
johnw	i was thinking --soft, and said it 3 other times	06:52
	but finger memory screwed me	06:52
	next time I talk to anyone, i'll have them stash first, no matter what	06:52
fynn	yeah, or at least save a diff	06:53
	anyways, is there a better way to apply than patch?	06:53
	it keeps asking me for the filename since it can't parse the git notation.	06:53
	not a huge deal, just curious	06:53
johnw	git-apply?	06:53
	git apply foo.diff	06:53
manguy	If I just delete the .git directory, will that take it out of git basically?	06:55
	Is everything stored in that directory?	06:55
fynn	johnw: thanks	06:56
	but after all the hassle, I get the exact same error:	06:56
	error: failed to push some refs to 'storage.counsyl.com:/git/base.git'	06:56
	→ palam joined	06:56
fynn	manguy: yes.	06:56
manguy	fynn: Thanks	06:56
johnw	manguy: there, and in ~/.gitconfig	06:56
fynn	manguy: sure. sometimes there's also .gitignore and .gitconfig	06:56
palam	i just ran install and make install. now when i try git<return> i get command not found... help please?	06:56
fynn	it can be in any nested directory.	06:56
palam	i'm on macosx 10.5 PPC	06:56
johnw	fynn: can you show me the output from "git branch -va"?	06:57
	→ paisal joined	06:57
johnw	palam: I recommend installing git using MacPorts	06:57
fynn	johnw: yeah, sec	06:57
palam	johnw: oh. well, will my installation attempt have screwed up anything?	06:58
johnw	palam: just "make uninstall"	06:58
	you may not have /usr/local/bin on your pathbtw	06:58
	← ashleyw left	06:58
palam	johnw: how do i check that?	06:59
johnw	echo $PATH	06:59
fynn	local 1234567 Commit comment	06:59
	* master 1234567 Commit comment	06:59
	origin/HEAD 1234567 Commit comment	06:59
	origin/master 1234567 Commit comment	06:59
palam	:/usr/local/bin: is there...	06:59
johnw	they are all really 1234567?	07:00
palam	johnw: make uninstall gives me make: *** No rule to make target `uninstall'. Stop.	07:00
fynn	johnw: nope, but they're all the same :)	07:00
	it's the same patch in all 4 branches.	07:00
johnw	show me your push command	07:00
	then, look in your .git/config file and search for [branch "master"]	07:01
	and tell me what it says there too	07:01
	something is not adding up	07:01
	→ cilly joined	07:02
vmiklos	palam: default make install installs git to ~/bin, do you have it in your PATH?	07:03
fynn	johnw: ~/repo$ git push	07:04
	[email@hidden.address] password:	07:04
	To git.example.com:/git/repo.git	07:04
	! [rejected] master -> master (non-fast forward)	07:04
	error: failed to push some refs to 'git.example.com:/git/repo.git'	07:04
johnw	can I see your whole .git/config?	07:04
	the answer lies there	07:04
fynn	yeah, sec	07:04
palam	vmiklos: my $PATH is /Users/palam/Tools/airsdk/bin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/X11/bin	07:04
	→ Quantumplation joined	07:04
johnw	palam: export PATH=$PATH:$HOME/bin	07:05
	← robinr left	07:05
johnw	see if that command helps you	07:05
palam	johnw: YES!	07:05
	thanks johnw, vmiklos!	07:05
fynn	johnw: here's the "master" section:	07:05
	[branch "master"]	07:05
	remote = origin	07:05
	merge = refs/heads/master	07:05
vmiklos	np	07:05
johnw	then add that command to your .bashrc	07:05
	← RayMorgan left	07:05
fynn	I can anonymize and paste the entire config file if you want.	07:05
johnw	fynn: yes please	07:06
	cuz now I need to see another section of it	07:06
fynn	sure, sec.	07:06
	← dgrant left	07:06
Quantumplation	Ello =) So, i recently watched the linus tech-talks about git and started wondering why i've never heard about this before ^^ so i'm starting to convert some of m projects to git (mainly for backup purposes, but also some collaboration) and i was wondering what the best method for storing a repositor on a USB memory stick would be? Should i clon the repository to the flashdriv and just push it a couple of times a day,	07:06
	← chalky left	07:07
fynn	johnw: http://dpaste.com/113773/	07:07
johnw	Quantumplation: if it's for backup purposes, I'd just rsync the whole working tree + git directory	07:08
	← agenteo left	07:08
Quantumplation	rsync?	07:08
johnw	fynn: you're using ssh to talk to this repo?	07:08
	Quantumplation: if you've not heard of rsync yet, let me introduce you to a lifelong friend	07:08
Quantumplation	=) Please do	07:09
johnw	http://samba.anu.edu.au/rsync/	07:09
fynn	johnw: I don't know, really	07:09
	→ ashleyw joined	07:09
johnw	fynn: if you're using http, that's our problem	07:09
fynn	johnw: OK, what should I try then?	07:09
johnw	well, I do this:	07:10
	git+ssh://host/path	07:10
	just change the url in your [remote "origin"] section	07:10
Quantumplation	john: should i keep the rsync utility on the thumb drive? (i'm working in a primarily windows environment right now)	07:10
johnw	git+ssh://vcs.example.com/git/base.git	07:10
	Quantumplation: i'm not sure if there's a tiny, statically bound .exe for windows or not	07:11
	there ought to be!	07:11
	→ xenoterracide joined	07:11
	← ankit9_ left	07:12
johnw	fynn: because based on what you showed me, push shouldn't be actually doing anything at all	07:12
xenoterracide	is it possible to pass the password for git imap-send on the command line?	07:12
	→ ijcd joined	07:12
xenoterracide	I see nothing about doing that in the man page... but maybe there's something I'm missing	07:12
	→ ashH joined	07:12
	→ kermit` joined	07:13
fynn	johnw: I got the same error :/	07:13
	johnw: you mean, it shouldn't be doing anything, cause I didn't commit anything?	07:13
johnw	yep	07:14
	if master == origin/master, push is a null op	07:14
	which points to a protocol problem	07:14
	or server config problem	07:14
	fynn tries a few things	07:14
johnw	haha	07:14
	← almery left	07:15
fynn	how do I unstash?	07:17
	I stashed so I could git pull	07:17
johnw	git stash apply	07:17
	← peritus_ left	07:17
fynn	cool thanks.	07:17
Quantumplation	hm, the windows rsync seems to rely on cygwin. that should be fine	07:18
fynn	johnw: hurrrrrray, push workzzzz!	07:19
johnw	yay!	07:19
	congrats	07:19
	fynn is a super 1338 hacker staggadish	07:19
fynn	with the help of his hacker mentor johnw	07:19
	← Kathrin-26^away left	07:19
johnw	and i finally got the new cgit working! http://git.newartisans.com/	07:19
fynn	a lesser achievement, I should think	07:20
	but still, worth mentioning.	07:20
johnw	hahah, indeed!	07:20
	→ Kathrin-26^away joined	07:20
fynn	dude, I just, like, committed.	07:20
johnw	feels good, doesn't it?	07:20
	until the end of time, that sha1 will always reference those changes	07:20
	well, in theory	07:20
fynn	yeah, my masculinity has been affirmed. my gf will finally let me off the couch now and into the bedroom	07:21
	← palam left	07:21
fynn	possibly even into the bed!	07:22
johnw	now, don't get cocky	07:22
fynn	yeah, probably not :(	07:22
kermit`	hello,I clone my project from the server,then I modify it and commit many times but haven't push.Now, I want to push it, but I don't want leave the interim commits.How can I do? Now I always copy all of the current files, and then git-reset --hand, and recover the files, then git-commit. Is there any simplier way? Thanks.	07:22
johnw	kermit`: git rebase -i	07:23
	use the "squash" command to merge all your commits down into one	07:23
	you'll see what I mean by squash after you type "git rebase -i origin/master"	07:23
fynn	johnw: ah, CGI and git... I see what you did there	07:23
johnw	fynn: those are some of my personal mirrors of other projects, most of which do not use git	07:23
kermit`	Sorry,johnw,what is squash?	07:25
johnw	kermit`: in the rebasing file, you change the word "pick" to "squash" if you want to collapse one commit into another	07:25
kermit`	Thanks, johnw.I'd like to test it at once;p	07:27
	→ chalky joined	07:27
	→ meyering joined	07:28
Quantumplation	Can rsync be used without a "server" running?	07:29
	→ lresende joined	07:33
	← ahupp\|fb left	07:33
johnw	sure	07:33
	→ tal67 joined	07:33
	← rphillips left	07:37
	← paisal left	07:38
	→ ludde joined	07:38
Quantumplation	hrm... all git commands are based on the repo of the current directory right?	07:38
	← cilly left	07:38
	→ psankar joined	07:38
thiago_home	the repo that the current directory is part of, yes	07:39
Quantumplation	so if i have c:/proj/Project1 and c:/proj/Project2, i just cd to "change repo"?	07:39
	rather, cd (in order to* change repo	07:39
	→ _eric joined	07:40
_eric	Is there a way to do the equivelant of a "git add -p" into a stash?	07:40
fynn	johnw: thanks for all your help, looks like there was something weird in the repo.	07:40
_eric	I have some debug code that I want to stash (and check everything else in)	07:40
	← gaveen left	07:40
johnw	glad I could help somewhat	07:41
	reithi is away: Gone away for now	07:41
	← johan-s left	07:43
	→ cilly joined	07:44
	→ ijcd_ joined	07:46
Quantumplation	hm... msysGit is having trouble with e:/ and h:/ drive specifications...	07:46
	← mczepiel left	07:47
	← muthu_ left	07:47
Quantumplation	Got it working =P	07:48
	← chalky left	07:49
	← Beket left	07:50
	← tal67 left	07:50
SeJo	someone using reviewboard here?	07:53
	→ pierre- joined	07:54
	→ tal67 joined	07:56
	← johnw left	07:57
	← thiago_home left	07:58
	→ paisal joined	07:59
	← Quantumplation left	07:59
	→ peritus_ joined	08:00
	→ martinpaul joined	08:00
	← ijcd left	08:02
	← flazz_ left	08:04
	← fynn left	08:04
	→ giallu joined	08:05
	→ ilogger2 joined	08:07
SeJo	so once it is decides we pick from their branches?	08:07
	→ EvilGuru joined	08:08
	→ flazz joined	08:08
	→ chalky joined	08:08
	→ johan-s joined	08:10
	→ ceej joined	08:12
	→ Beeet joined	08:13
	→ hdl2 joined	08:13
ndim	SeJo: Get rid of the "one central repository for everyone" idea.	08:13
	SeJo: Have "a few" get commit access to the repository with the "master" branch.	08:14
	SeJo: And have everyone else publish their own repos, and have "the few" pull from there, and push to maste.r	08:14
	→ peritus_ joined	08:17
	→ Casan joined	08:19
	→ ceyusa joined	08:19
	→ hurikhan\|Work joined	08:21
	→ mithro joined	08:26
	→ dandre joined	08:27
	← dandre left	08:27
	→ palam joined	08:28
	→ beachsurfin joined	08:29
	→ jonalv joined	08:30
	→ galderz joined	08:34
manguy	I've added one file to .gitignore, specifically 'lib/settings.rb' But git is still including it in the modified changes. How can I keep a blank template of the file in the repository, but have my own copy to change the settings as I wish?	08:34
SeJo	ndim: but i want everyone to push to the same server	08:35
	ndim: so i need to create repos for everyone?	08:35
	→ HG` joined	08:35
SeJo	ndim: won't those repos differ too much from the master?	08:36
	→ Arafangion joined	08:36
ndim	SeJo: Well, you can put many people's repos onto the same server.	08:37
	SeJo: Well, if people don't maintain the branches in their repos, it will eventually differ. But that problem is not specific to people publishing their own repos.	08:38
	← jonalv left	08:39
	→ jmesnil joined	08:39
	→ saimon joined	08:40
	→ harinath1 joined	08:42
ezyang	SeJo: Tell people to merge/rebase frequently	08:43
	git pull --rebase is your friend	08:43
	→ nud joined	08:44
	← palam left	08:45
	← HG` left	08:47
	→ plasticine joined	08:48
	→ Niamor joined	08:53
	→ naeu joined	08:58
	← peritus_ left	09:00
	→ palam joined	09:03
	← chalky left	09:06
	hdl2 → hdl	09:06
	→ chalky joined	09:06
	→ lucsky joined	09:09
lucsky	'morning	09:09
	→ tav_ joined	09:12
	→ eevar2 joined	09:14
	→ schlort joined	09:14
eevar2	best way to merge changes from svn into my current work, w/o using git-svn?	09:15
	→ HG` joined	09:16
Ilari	eevar2: Checkout (some) commit(s) using svn, copy the files into git repository add and commit (into seperate branch). Then merge that branch in.	09:17
	← Niamor left	09:17
	→ Niamor joined	09:18
	→ charon joined	09:19
eevar2	does it matter what branch i'm on when creating that new branch? -- do i want to merge the changes into master first, and then merge master into my current branch?	09:19
	← Beeet left	09:20
eevar2	or am i just complicating things?	09:20
	← HG` left	09:23
	→ loiseau joined	09:23
	← palam left	09:23
	→ skiv02 joined	09:25
	→ palam joined	09:25
	→ Marmouset joined	09:28
	→ rraasch joined	09:29
	← nud left	09:29
	→ WALoeIII joined	09:33
	→ ijcd joined	09:33
	→ ankit9 joined	09:35
	→ z3om joined	09:36
	z3om → zeom	09:37
	→ nud joined	09:37
	→ antirobotrobot joined	09:37
	← palam left	09:38
	← ankit9 left	09:40
eevar2	hmm.. possible to undu git stash clear..?	09:40
	*undo	09:40
	reithi is back.	09:41
tokkee	eevar2: I'm not sure if there are reflogs for stashes but you can surely find that out ;-)	09:41
	eevar2: Else, look thru your unconnected objects.	09:42
charon	tokkee, eevar2: stash is implemented using a reflog, so that's no help here	09:42
	try 'git fsck --lost-found'	09:42
	→ peritus_ joined	09:43
	→ palam joined	09:46
eevar2	yay, fsck --lost-found worked	09:47
	guess i should be reading man pages before issuing random git commands ;)	09:47
	← palam left	09:49
	← WALoeIII left	09:54
	→ cilly joined	09:58
	→ ankit9 joined	10:00
	→ seangrove joined	10:01
	→ Kathrin-26^away joined	10:02
	→ greyface joined	10:07
	→ ashleyw_ joined	10:08
	→ AAA_awright_ joined	10:10
	← Marmouset left	10:13
	← cilly left	10:14
	→ cilly joined	10:15
	← charon left	10:17
	→ jaeckel joined	10:18
gouki	I'm trying to change the 'Owner' field on GitWeb. I tried creating a file called owner on the repository root (as mentioned on the README). No luck. Creating a new section [gitweb] on the config doesn't work either. Any other ideas?	10:24
	→ SRabbelier joined	10:24
	→ warlock_mza joined	10:24
	← loiseau left	10:24
	→ charon joined	10:25
	→ _slacker_ joined	10:28
Aides	gouki: by repository root you mean the place where hooks/ objects/ config and other files reside, right?	10:29
gouki	Aides, yes. Exactly.	10:29
	→ mcella joined	10:29
Aides	what's your git version?	10:31
	← chalky left	10:35
	→ beket joined	10:35
	beket → Beket	10:35
	→ chalky joined	10:38
	→ ThoMe joined	10:39
ThoMe	hello :-)	10:39
steveb	hello	10:39
ThoMe	is a good web-tool for git avaialable? for search in my code (source code), or search scriptname, or "search when change the file X" ... ?	10:39
	hm?	10:43
steveb	i hear good things about gitorious	10:43
	← peritus_ left	10:44
ThoMe	steveb: is it free?	10:44
johan-s	no codesearch though... yet.	10:44
ThoMe	johan-s: codesearch ? is sa tool?	10:44
ndim	My first thought is that all that functionality is quite easy on a command line.	10:45
johan-s	ThoMe: no, you just said "for search in my code (source code)" and gitorious doesn't search in the git repositories themselves currently	10:45
ThoMe	hm	10:46
ndim	And my second thought is that many people can't use or don't have a command line, and thus need such specialized tools.	10:46
ThoMe	johan-s: hm. other ideas?	10:46
	→ _Vi joined	10:46
ndim	I.e. the Windows(TM) way: Buy and install and learn a separate program for every command line one liner you can think of.	10:46
	→ peritus_ joined	10:46
	→ shaftyy joined	10:47
steveb	can gitorious look at previous revisions of a source file?	10:48
	← ceej left	10:49
	→ sverrej_ joined	10:50
steveb	im converting my work svn repo to git as a kind of proof of concept thing. so fat the svn2git has taken 19 hours. and still going	10:50
	so far*	10:51
	← Niamor left	10:52
ned	how do i revert a specific file from a commit?	10:53
	→ Niamor joined	10:53
skiv02	im trying to set up a central git repository on a linux machine, which will be accessed by windows users. do you guys recommend using ssh or GIT daemon for pushing/pulling? if SSH, is there a way of binding SSH user authentication to ActiveDirectory?	10:54
	← EvilGuru left	10:55
Aides	ned: what do you mean by "revert from commit"?	10:55
	i can think of at least three conflicting meanings of that :)	10:56
Tuomas	ned: if you haven't yet pushed, just "git reset --soft HEAD^" and commit again without that file (assuming the commit with extra file is the last one)	10:56
ned	Tuomas, thanks :)	10:56
	→ grahal joined	10:58
	← grahal left	10:58
gouki	Aides, I'm sorry I took so long to reply. Version: 1.5.4.3	10:59
Aides	gouki: have you considered updating? i can't be sure it will help, but it may	11:00
gouki	Aides, I don't have much control over it (it's on a colleague server) but I'll mention it.	11:01
	← petaflot left	11:01
gouki	From what I found on KernelTrap archives, the file owner or an entry owner = name in config should work with this version ... :S	11:02
	→ cemerick joined	11:04
	→ pantsman joined	11:04
	→ JackTheRabbit joined	11:06
	← peritus_ left	11:07
	← sverrej_ left	11:09
	→ grahal joined	11:10
	→ xjjk joined	11:12
	→ EvilGuru joined	11:13
	← hurikhan\|Work left	11:15
	← AAA_awright_ left	11:15
	→ harryjr joined	11:17
	→ AAA_awright joined	11:17
	harinath1 → harinath	11:17
harryjr	svn co --username=harry https://svn.turntool.com/svn/turntool/plantool/trunk works fine but git svn init --username=harry -s https://svn.turntool.com/svn/turntool/plantool keeps asking for password	11:17
	what would be the issue?	11:17
	→ DRMacIver joined	11:18
	→ hurikhan\|Work joined	11:20
	← ankit9 left	11:20
	← Kathrin-26^away left	11:21
	→ Kathrin-26^away joined	11:22
	→ bcarlyon\|laptop joined	11:31
	← EvilGuru left	11:31
	← hdl left	11:32
	← Niamor left	11:32
	→ Niamor joined	11:33
	→ Aides joined	11:33
	→ ankit9 joined	11:35
	deizel is back	11:35
	bcarlyon\|laptop → BCarlyon\|Laptop	11:35
	→ lorandi joined	11:36
	→ xlq joined	11:43
xlq	I'm doing rebase -i --preserve-merges	11:43
	and it's 'refusing to squash a merge', giving me the sha1 of a commit which is quite blatantly not a merge :(	11:44
	→ jriga joined	11:44
	→ ashleyw joined	11:45
	← Kathrin-26^away left	11:46
	→ Kathrin-26^away joined	11:46
	→ sverrej joined	11:46
	← ashleyw left	11:47
	→ Voker57 joined	11:47
sgronblo	I made the directory non-dirty and then stash applied some old stuff. Found out that the changes applied from the stash were unnecessary so i removed them all. Now status tells me these files are unmerged. How do I get rid of the unmerged status for them?	11:48
Pieter__	git add -u	11:48
	→ tango_ joined	11:49
Pieter	weird nickserv stuff, I never understand it	11:49
	→ priidu_ joined	11:50
sgronblo	ok, now it still says a file is modified but git diff doesn't give any output at all for that file	11:51
	← mithro left	11:52
xlq	git diff --cached	11:53
steveb	can anyone tell me how clever svn2git (or git-svn) is? seems my companies svn structure is less than standard... we have trunk/ branches/ tags/, but also tags/<some_product>/<more_tags>	11:53
	will git-svn pick up all the tags in a tag, or just the original tag?	11:53
	← Aides left	11:54
Voker57	you can use --tags=tags/*/	11:54
	since 1.6.0 afair	11:55
steveb	i'll look at the docs a little more. im only really doing this as a proof of concept anyway. our svn repo is a mess	11:55
	but it's already been importing for 20 hours	11:56
ndim	Or --tags/*/projname	11:56
tokkee	But that won't include tags in tags/ _and_ tags/*/, I suppose.	11:56
steveb	i wonder if it will be finished by the end of the day	11:57
sgronblo	Can't git stash show show me a true diff instead of just a summary?	11:57
ndim	A layout where --trunk=trunk/foobar --branches=branches//foobar --tags=tags//foobar makes sense was quite popular some time.	11:57
steveb	we have a mish-mash of lots of different schemes	11:58
	→ Sho_ joined	11:58
steveb	very arbitary...	11:58
	← ashleyw_ left	12:00
	→ peritus_ joined	12:03
	→ bentob0x joined	12:06
	← JackTheRabbit left	12:08
	→ Hendu joined	12:11
	→ cmarques joined	12:13
Hendu	Hi. Have a wee problem. I did a "git add ." and didn't realize that I included files which shouldn't be commit. How do I undone this?	12:13
harryjr	git reset HEAD	12:13
	← ankit9 left	12:13
Hendu	nice one	12:13
	thanks	12:14
	→ simosx joined	12:15
nothingmuch	how do i run a specific test? i can't find anything useful in the docs	12:15
	← Niamor left	12:15
	→ Niamor joined	12:15
harryjr	nothingmuch: ?	12:15
	→ gebi_ joined	12:15
nothingmuch	i am trying to make a patch	12:15
Thumper_	Hendu: git reset FILE # works too	12:15
nothingmuch	i don't want to run the whole test suite	12:16
harryjr	nothingmuch: wrong channel, i assume.	12:16
nothingmuch	nope	12:16
Hendu	Thumper_: thanks got that	12:16
	← jaeckel left	12:16
nothingmuch	i've futzed with t/t7600-merge.sh and I'd like to run it in debug mode, and also to be sure i have the env setup properly	12:17
Thumper_	nothingmuch: cd t; t/7600-merge.sh or bash -x t/7600-merge.sh	12:18
	→ nedko joined	12:18
nedko	someone please help me: http://pastie.org/private/jspjrunp07chvii9x08fa	12:19
nothingmuch	aha, tests take options	12:19
	--verbose	12:19
Thumper_	nedko: you need to merge the upstream changes before you push	12:20
	nedko: pull first	12:20
	→ clairvy__ joined	12:23
xlq	With 'git commit --amend', how do I add a new file to commit?	12:24
	→ Owner joined	12:25
tokkee	xlq: git add <file>; git commit --amend	12:25
	xlq: I.e. just like adding a file to any commit.	12:26
xlq	tokkee: it doesn't seem to notice the new file	12:26
	→ boto joined	12:26
	← boto left	12:26
tokkee	xlq: Any error messages? What does "git status" say?	12:26
	→ codebrulee joined	12:26
	→ boto joined	12:26
xlq	Oh! I see	12:26
	the file hasn't actually changed!	12:27
	← nud left	12:27
ndim	Interesting. It is always the 414th fork in "git svn" which fails to fork.	12:27
tokkee	xlq: :-)	12:27
	ndim: Even with different SVN repos?	12:28
ndim	tokkee: Untested so far.	12:28
	Hmm. And it is the 414th for in that specific place.	12:28
	Perhaps it IS the ulimit -u of 1024?	12:28
	→ Acry joined	12:28
	← Acry left	12:29
	→ fgomez joined	12:30
	→ mikl joined	12:30
	→ Ryback_ joined	12:31
harryjr	svn co --username=harry https://svn.turntool.com/svn/turntool/plantool/trunk works fine but git svn init --username=harry -s https://svn.turntool.com/svn/turntool/plantool keeps asking for password (Authentication realm: <https://svn.turntool.com:443> TurnTool Subversion Repository \ Password for 'harry)	12:32
nedko	Thumper_: i've pulled and nothing changed, this is why i checked parent sha1, there is only one commit to push	12:33
steveb	harryjr: what does svn co --username=harry https://svn.turntool.com/svn/turntool/plantool do?	12:33
	← plasticine left	12:33
harryjr	steveb: checks out trunk	12:34
	→ ankit9 joined	12:35
nothingmuch	jast, doener: http://gist.github.com/53926	12:35
steveb	harryjr: has svn maybe cached your password somewhere? is it password protected?	12:35
harryjr	svn co doesn't ask for password, so, it's cached yes.	12:36
steveb	ok, that's why git svn is asking for a password then	12:36
	← cilly left	12:36
	→ cilly joined	12:36
harryjr	well, that should be fine - but entering the correct password doesn't do anything	12:36
	→ gaveen joined	12:37
xlq	Grr	12:37
	git rebase doesn't preserve merges	12:37
	When I specify --preserve-merges, it grumbles about squashing a merge, which I'm not doing.	12:37
harryjr	except asking for the password again	12:37
	← Niamor left	12:37
	→ Niamor joined	12:38
	→ nud joined	12:38
xlq	hmm, looks like a known bug	12:39
	I'll have to avoid squashing :(	12:40
	← nedko left	12:41
harryjr	steveb: just tried removing the subversion configs, the password works fine with svn co, but not git svn. :/	12:43
steveb	i've got no idea im afraid	12:44
	→ name joined	12:49
ndim	OK. "Can't fork" at the 414th fork, again - this time with "ulimit -u 2048", same SVN repo.	12:50
	→ foutrelis joined	12:54
	deizel is gone	12:55
	← cilly left	12:55
	→ cilly joined	12:55
	← bentob0x left	12:57
	← Niamor left	12:57
	→ Niamor joined	12:58
Beket	Can I stash my uncomitted work, change branch and commit there ?	12:59
ThoMe	rake db:migrate RAILS_ENV=production	13:00
	(in /usr/gitorious/mainline)	13:00
	rake aborted!	13:00
	no such file to load -- spec/rake/spectask	13:00
	/usr/gitorious/mainline/Rakefile:10	13:00
	(See full trace by running task with --trace)	13:00
	hm, why an error?	13:00
	→ vuf_ joined	13:01
wereHamster	Beket: yes, you can.	13:03
Beket	thank you wereHamster	13:03
wereHamster	however I don't think a stash is necessary, as git-checkout doesn't touch changed files	13:03
	→ JackTheRabbit joined	13:03
wereHamster	so a simple git checkout my-other-branch; git commit should suffice	13:03
xlq	I am doing a rebase.	13:06
	One commit introduces a file that I don't want at all. So I amended that commit and removed that file.	13:07
	Now, a later commit modifies that file, and causes a conflict.	13:07
	What do I have to do now, to make sure that file never gets into the rebased commits?	13:07
wereHamster	ammend that commit as well	13:07
xlq	Right. But it's conflicted on that commit, and it wants me to 'git add' then 'git rebase --continue	13:08
wereHamster	does 'git rm the-file.txt; git rebase --continue' do the right thing?	13:08
xlq	no :(	13:09
	that seems to re-introduce the file into the commit I amended to remove it	13:09
ThoMe	can anybody help me?	13:09
	i try: rake db:migrate RAILS_ENV=production	13:10
	no such file to load -- spec/rake/spectask	13:10
	/usr/gitorious/mainline/Rakefile:10	13:10
	(See full trace by running task with --trace)	13:10
	i need a file?	13:10
wereHamster	ThoMe: does that have anything to do with git?	13:10
ThoMe	rake != git ?	13:10
	sorry	13:10
wereHamster	xlq: what about git rm --cached ?	13:10
xlq	ThoMe: rake is a build system, git is a DCVS. Completely different.	13:11
ThoMe	ups ok	13:11
wereHamster	ThoMe: try in #ruby	13:11
xlq	wereHamster: no :(	13:12
wereHamster	xlq: after git-rm, what does git-status say?	13:13
xlq	wereHamster: all the changes that it should commit. It doesn't mention the file I want to get rid of.	13:13
	→ hdl2 joined	13:13
	→ aziz joined	13:13
wereHamster	well, that's what you want, right?	13:14
xlq	wereHamster: yeah	13:14
wereHamster	you don't want the file to be in the next commit, right?	13:14
xlq	correct	13:14
	I don't want the file anywhere in the repo	13:14
	but as soon as I do git rebase --continue	13:14
	hdl2 → hdl	13:14
xlq	it re-appears, in the commit it creates, and in the commit before!	13:14
	→ bentob0x joined	13:14
wereHamster	there's something else wrong with your workflow since git cannot change prior commits. The file cannot reappear in the 'commit before' just like that	13:15
xlq	yeah	13:16
	O_o	13:16
	but it is	13:16
	→ rolfb joined	13:16
xlq	Maybe gitk is showing me the original commits?	13:17
	Ahh, that is so!	13:17
	Where does rebase store the commits it makes, then?	13:17
wereHamster	rebase rewrites the history, it will throw away the old commits and replace them with the new ones	13:18
xlq	So how come I can still see the old commits if I run hgk during the rebase?	13:18
	← charon left	13:19
wereHamster	what's hgk?	13:19
	→ Kbyte joined	13:19
xlq	*gitk, sorry	13:19
ndim	On git-svn's 414th fork, strace says the following happens: clone(child_stack=0, flags=CLONE_CHILD_CLEARTID\|CLONE_CHILD_SETTID\|SIGCHLD, child_tidptr=0xb809d708) = -1 ENOMEM (Cannot allocate memory)	13:19
wereHamster	xlq: what about git-log? Does it also show the old commits?	13:20
xlq	looks like it, yeah	13:21
	← ankit9 left	13:21
	→ ankit9 joined	13:25
	← Kbyte left	13:25
	→ statim joined	13:25
	→ Kbyte joined	13:26
	→ joshdavey joined	13:27
xlq	OK, I've finished the rebase	13:27
	it achieved absolutely nothing!	13:27
	← Owner left	13:27
xlq	Maybe git rebase --preserve-merge is just completely bugged	13:27
wereHamster	could you paste the commands you use so I can see what you are doing?	13:28
xlq	I have the branch 'first' pointing at the first ever commit, and the branch 'master' at the last commit	13:28
	git checkout master && git rebase -i -p first	13:28
	← codebrulee left	13:29
steveb	fantastic. after 21.5 hours my git-svn import failed.	13:30
	← rraasch left	13:30
xlq	steveb: :(	13:30
wereHamster	then what? You 'edit' the commit that introduced the unwanted file, amend and continue?	13:31
spb	steveb: it's git punishing you for disappearing from the internet	13:31
xlq	wereHamster: exactly	13:31
	→ zepolen joined	13:31
steveb	spb: apparently so	13:31
wereHamster	steveb: I'm not sure but I think git-svn can resume the checkout	13:31
	just go into the git repo and 'git svn fetch'	13:31
steveb	spb: i havnt vanished from the internet anyway. i've just... relocated	13:31
xlq	oh, um	13:32
steveb	wereHamster: it did something funky. the git repo doesnt seem to be valid	13:32
wereHamster	steveb: any errors in particular?	13:32
	→ codebrulee joined	13:32
	→ cmarcelo joined	13:33
wereHamster	xlq: oh, then at some point another commit fails and git tells you to resolve the conflict, right?	13:33
xlq	wereHamster: yeah	13:33
spb	steveb: relocated where, anyway	13:33
	it had better be somewhere more interesting	13:34
wereHamster	xlq: have you already published the branch?	13:34
xlq	wereHamster: that's not a problem	13:34
	wereHamster: no, I haven't	13:34
wereHamster	xlq: then take a look at git-filter-branch	13:34
	→ gambler joined	13:35
wereHamster	I think --preserve-merges causes the file to re-appear	13:35
steveb	spb: more interesting than what? ##uk? no one speaks there any more!	13:36
spb	more interesting than ##uk was before you disappeared from it!	13:36
steveb	are you saying i was the life and soul of it?	13:36
	how sweet	13:36
spb	no, just that you killed it	13:36
	← lorandi left	13:37
	→ peritus__ joined	13:37
	→ lorandi joined	13:37
steveb	i didnt dissappear from it. i went to america for a summer and everyone stopped talking	13:37
spb	ergo you killed it	13:37
	→ deavid joined	13:38
	← peritus_ left	13:38
steveb	it's not my fault if no one else has the social skills to keep a channel going when i leave	13:38
	→ comcor_ joined	13:39
xlq	Do I need to specify -p every time I do git rebase --continue ?	13:40
	→ steltho joined	13:41
	← greyface left	13:41
steltho	After I tag a commit whit git tag is there a way to view the tag using git log ?	13:41
	← foutrelis left	13:44
heipei	steltho: git tag -l	13:45
xlq	after a git-filter-branch, master is unchanged	13:45
nothingmuch	that means none of your filters had any effect	13:46
tokkee	steltho: "tig" will show you tag names as well.	13:46
xlq	Ahh, I did it wrong	13:46
	← Niamor left	13:47
	→ Niamor joined	13:48
steltho	Ok thanks, but is there some way I can get the tag to show up as a commit that is viewable with git log. For example ...	13:48
nothingmuch	i know gitx does it	13:48
	trying to figure out if it passes some useful option to git log	13:49
steltho	When I pull the latest version of the kernel, the latest commit always has a message like: Linux 2.6.29-rc2	13:49
	← comcor left	13:49
	→ evanpro joined	13:50
evanpro	Howdy	13:50
	I've got a problem	13:50
steltho	Is the Linux 2.6.29-rc2 commit created by git tag, or is it done another way?	13:50
evanpro	Periodically in a dev directory, I'll run something like "find -name *~ \| xargs rm" to clean up those little ~ files that accumulate over time.	13:51
	Unfortunately, I did this in a (local) git repo recently	13:51
	And apparently there are some meaningful files in .git that end in ~	13:51
	← gaveen left	13:52
	→ gaveen joined	13:52
	→ esparkman joined	13:55
nothingmuch	steltho: sorry, found nothing useful	13:55
	→ foutrelis joined	13:56
xlq	Ok, after converting a repository, how would one add a .gitignore for all the commits?	13:57
	(it doesn't really matter about the very first commit)	13:57
	← chalky left	13:58
evanpro	OK, somehow my .git/index is now length 0	13:59
	→ rjbs joined	14:00
rjbs	Hm. I am trying to import svn->git. The svn repository started as "everything in root" and was reorged to trunk/branches/tags later.	14:01
	Anyone know of a tool that will do the import nicely?	14:01
tokkee	evanpro: Did your shell expand *~ to something "dangerous"?	14:02
	xlq: What do you want to ignore?	14:02
	→ ivan_ joined	14:02
xlq	tokkee: does that matter?	14:02
evanpro	tokkee: no, actually it's escaped, find -name \*~	14:02
tokkee	rjbs: Uh! That does not sound like a simple task.	14:02
rjbs	tokkee: I know, that's why I'm looking to see if someone else has solved it!	14:03
evanpro	But when I run git gc, I get: "error: cannot lock ref 'HEAD.~1~'"	14:03
	← steltho left	14:03
Pieter	steltho: git log --decorate	14:03
tokkee	rjbs: You might be able to import that by importing everything into a single (master) branch (i.e. ignoring trunk/branches/etc.) and then doing some magic using git-filter-branch (--tree-filter).	14:03
	→ dedmajor joined	14:03
	→ bdimcheff joined	14:04
rjbs	I thought about that... I think if it comes down to me having to do anything (because I have like 50 of these to do) I will just punt and fix it in the final imported rev, so there's history, even if it's mangled.	14:04
tokkee	xlq: I don't really get what you're trying to achieve.	14:04
xlq	tokkee: add a file to all commits, as if it had been there all along	14:05
	I thought I'd be able to create a commit off the first commit, and rebase everything else on to it, but I get conflicts.	14:05
	← ivan_ left	14:05
tokkee	xlq: Oh, now I get it - I thought you would want to ignore commits ;-)	14:06
xlq	tokkee: no :)	14:06
tokkee	xlq: Hrm ... I guess you want the first commit to include that as well, right?	14:06
xlq	tokkee: yeah, but I don't mind so much about that one.	14:06
tokkee	xlq: If that's not the case you could use git-rebase -i.	14:06
xlq	But I get conflicts!	14:07
	→ chalky joined	14:07
tokkee	What kind of conflicts?	14:07
xlq	with loads of files that shouldn't be affected.	14:07
tokkee	rjbs: If the layout of all repos is mostly identical, you might be able to automate it after doing it once.	14:08
	→ [1]bcarlyon\|lapt joined	14:08
rjbs	yeah, I know. just bitching :)	14:08
tokkee	xlq: How did you do it?	14:08
xlq	git checkout -b first <sha1-of-first-commit>	14:08
	vim foofile	14:08
	git add foofile && git commit	14:08
	git checkout master	14:09
	git rebase -p -i first	14:09
	→ bnovc joined	14:09
bnovc	Is there anything like git-svn for mercurial? My office uses mercurial, but I wanted to try out git ... searching for things like "git mercurial" just brings up a lot of debates about which is better	14:10
	→ mbroeker joined	14:11
xlq	bnovc: there's 'hg convert', not sure if it can do more than just convert the repo though	14:11
tokkee	xlq: Either do "git rebase -i <sha1-of-first-commit>; chose "edit" for the first commit; apply your changes; amend that commit; git rebase --continue".	14:11
	xlq: Or: git checkout -b first <sha1-of-first-commit>; edit; git add && git commit --amend; git checkout master; git rebase -p --onto first <sha1-of-first-commit> master	14:12
	(i.e. what you've already done but note the different call to git-rebase)	14:13
bnovc	xlq: well is there an easy way to get changes back to the hg repo after that? beside exporting and importing a bunch of patches	14:13
	and updating from the hg repo for that matter	14:13
xlq	bnovc: no clue	14:13
tokkee	xlq: Did you amend the first commit or did you add another?	14:14
xlq	tokkee: I added another	14:14
	tokkee: is that bad?	14:14
tokkee	xlq: Should not ...	14:15
xlq	tokkee: what?	14:15
tokkee	xlq: That should not be bad.	14:15
xlq	tokkee: also, -i is needed with -p	14:15
	tokkee: it seems to get so far and then says: "Already uptodate!"	14:15
	"Finished one cherry-pick."	14:15
	then displays status	14:15
	Ahh, there are some empty commits	14:16
	that do nothing	14:16
tokkee	xlq: Well, if you rebase all of master, you will try to reapply the first commit which is already there though.	14:17
	→ kukks joined	14:17
tokkee	xlq: So, that commit will not introduce any changes. Using git rebase --skip when "Already uptodate" pops up should fix that.	14:18
xlq	ahh, k	14:18
tokkee	xlq: So, you didn't have any merge conflicts but only that "Already uptodate" message?	14:20
xlq	tokkee: that time, yeah	14:20
	but .gitignore still isn't there	14:20
	← gaveen left	14:23
tokkee	xlq: Did you make sure that the new commit is still there? (the one that introduces .gitignore)	14:23
xlq	I think I got the branch names mudled	14:23
	telmich just had a look at http://bazaar-vcs.org/Workflows, which looks really nice (especially with the fancy user images)	14:23
	← BCarlyon\|Laptop left	14:23
	[1]bcarlyon\|lapt → bcarlyon\|laptop	14:23
telmich	do we have something like that for git already? ;-)	14:24
	→ EvilGuru joined	14:24
	→ bartman joined	14:25
xlq	I just went through the rebase again	14:26
	it says "Successfully rebased and updated refs/heads/master"	14:26
	It has done NOTHING!	14:26
	→ drewr joined	14:26
	→ gaveen joined	14:27
	→ akitada joined	14:28
	← akitada left	14:28
	→ Acry joined	14:29
	← Acry left	14:30
wereHamster	xlq: wait. You want to _add_ a file to all commits? I thought you'd want to _remove_ a file	14:31
xlq	Both!	14:31
	But I've already removed the file now, with git-filter-branch	14:31
	→ EmilMedve joined	14:31
	→ softdrink joined	14:32
wereHamster	you should also be able to add a file with filter-brnach	14:32
xlq	I should also be able to rebase.	14:32
tokkee	xlq: What does "git log master..first" say?	14:33
xlq	It's the very first commit	14:33
tokkee	Only that?	14:34
xlq	no it's not	14:34
	blank output	14:34
tokkee	Then your new commit disappeared ...	14:34
xlq	O_o	14:34
	oh, um	14:34
tokkee	Don't ask me why / how though ...	14:34
	;-)	14:34
xlq	I have too many branches now	14:34
	bar first first2 foo gitig master old2master oldmaster show	14:34
	I don't remember creating 'show' though :\	14:35
tokkee	Did you backup your repo before doing anything? ;-)	14:35
xlq	yep	14:35
tokkee	Then starting over might be a good idea ;-)	14:35
wereHamster	you don't need to backup, there's always the reflog	14:35
	→ jm joined	14:35
tokkee	wereHamster: Yeah - but restoring / copying a backup might be easier sometimes ;-)	14:35
	→ akitada joined	14:36
wereHamster	yep, true	14:36
dedmajor	hi guys, i'm very new to git, and need some help hint. i have some branches in my project, for example, master and dev, i checked out dev version, made some changes and committed it. now i want this changes to be merged in master branch. how should i properly do that?	14:36
xlq	dedmajor: git checkout master && git merge dev ?	14:37
	← akitada left	14:37
wereHamster	dedmajor: git-merge or git-cherry-pick	14:37
	depends on what you want/need	14:37
	→ docelic joined	14:37
	→ rwaite joined	14:38
docelic	Hey folks, if I call 'git pull' and it imports some stuff from the origin repository, and there are conflicts... When I open the conflicting files, does the "<<<<<<< HEAD:src/Makefile" mean I'm looking at the local head, or what would be the head in origin?	14:39
	← JackTheRabbit left	14:40
	→ Savago joined	14:41
	→ plasticine joined	14:41
docelic	I see, the top part of the conflict is local, bottom is the "pulled" one	14:42
	← plasticine left	14:42
ndim	OK, I have found the problem why "git svn clone" "Can't fork": A simple OOM issue. Apparently, "git svn clone" easily uses something on the order of a GB of memory.	14:43
	→ fdv joined	14:43
xlq	This is absolutely ridiculous.	14:44
	→ lll joined	14:44
wereHamster	xlq: are you now trying to introduce a new file?	14:45
Savago	Hello there! I have a very-user-level question about git. What could create this message: adenilson@t60ade:/media/secure3/gcal$ git diff	14:45
	fatal: index file smaller than expected	14:45
xlq	I'm trying to do all sorts of things. Nothing seems to work.	14:45
Savago	and how to fix it?	14:45
	→ Aides joined	14:46
	← statim left	14:46
wereHamster	xlq: the only thing that comes to my mind is: you're doing it wrong ;)	14:46
	← Kathrin-26^away left	14:46
xlq	groan	14:47
	→ Kathrin-26^away joined	14:47
	→ imarcust1is joined	14:47
	→ davidfetter joined	14:47
	→ LiamH joined	14:47
	← imarcusthis left	14:47
	← chalky left	14:47
	→ mikearr joined	14:52
	→ akitada joined	14:54
	→ statim joined	14:55
	→ thorkilnaur joined	14:55
xlq	If I've done a merge in the past, and already resolved conflicts in the merge	14:55
	git rebase -i -p seems to want me to resolve them again!	14:56
wereHamster	xlq: git-rerere	14:56
xlq	whoa, wha?	14:56
Aides	man git-rerere	14:56
wereHamster	git-rerere - Reuse recorded resolution of conflicted merges	14:56
	→ schacon joined	14:57
	→ nothinghappens joined	14:57
	→ [1]bcarlyon\|lapt joined	14:58
xlq	wereHamster: can that actually help me now? Since the merges were originally done ages ago :(	14:59
wereHamster	I've never used it myself, I only know that the command exists	14:59
	→ Almad joined	15:00
jast	only if you enabled rerere caching before	15:00
	← antirobotrobot left	15:00
xlq	It's too late for that :(	15:00
	→ spearce joined	15:01
xlq	So there's absolutely no sane way to rebase a section of the graph which contains merge conflict resolutions?	15:02
Almad	Hi, is there any way to set user.name and user.password via environment variable? Cannot google it :]	15:02
	→ resmo joined	15:02
	← Kathrin-26^away left	15:03
xlq	Almad: don't know about environment variable, but what's wrong with ~/.gitconfig?	15:03
wereHamster	Almad: GIT_AUTHOR_NAME, GIT_COMMITTER_NAME	15:03
	← peritus__ left	15:03
wereHamster	GIT_AUTHOR_EMAIL also seems to exist	15:03
	xlq: when you want to set it in filter-branch for example..	15:04
	→ brett_h joined	15:05
	→ chalky joined	15:06
	→ marvil07 joined	15:07
resmo	how much source code of git is not yet C or when (year?) will be a native windows support available?	15:10
Circuitsoft	resmo: msysgit is reasonably functional now.	15:10
	It doesn't support git-svn or several other commands yet,	15:10
	But it works well enough for general development use.	15:10
jast	git-svn actually works in some versions of msysgit	15:11
	→ iulian joined	15:11
resmo	I'd like a native and official supported git version on win32	15:11
xlq	msysgit is native	15:11
Circuitsoft	That is in process.	15:11
	I think the fact that msysgit is separate from git makes it appear as an unofficial project.	15:12
	unofficial == unsupported	15:12
jast	most (if not all) of the windows-specific C code is in the official git repository	15:12
	effectively, certain tools will always require at least parts of current msysgit, notably filter-branch which needs to have a shell	15:13
	← hurikhan\|Work left	15:13
Almad	wereHamster: Thank You very much ;)	15:13
	xlq: ~/.gitconfig is fine for usage, not so much for my automated test suite ;)	15:13
	→ krawek joined	15:14
jast	Almad: for future reference, there's a complete list of git environment variables in man git	15:14
	oh, the bot is gone again, so here's the url: http://git.or.cz/man/git	15:14
	→ hurikhan\|Work joined	15:14
	← bcarlyon\|laptop left	15:14
	[1]bcarlyon\|lapt → bcarlyon\|laptop	15:14
	→ tek0 joined	15:14
Almad	jast: ah, thanks, was blindly looking at man git-commit :)	15:15
	(and git-config)	15:15
	→ Kathrin-26^away joined	15:15
resmo	I feel like companies are scared using git because of the "not official supported on windows" thing...	15:15
bremner	resmo: and?	15:16
	→ peritus_ joined	15:17
bremner	i.e. what would	15:18
jast	where does it say that windows isn't officially supported?	15:18
bremner	official support look like?	15:18
jast	I mean, patches specific to windows are accepted, and there at least two competent regular developers who do contribute such patches	15:19
	→ Phoenixfire159 joined	15:20
fdv	jast: http://code.google.com/p/msysgit/ - 'Unfortunately, Git on Windows is only officially supported using Cygwin...'	15:21
resmo	sound great. Any differences between "git on Win" and "git on nonWin"?	15:22
fdv	(since you asked)	15:22
Thumper_	fdv: use it anyway. it works	15:22
jast	I'm not sure that's the current state of things	15:22
	and I don't know who maintains that site either, so perhaps ask about that on the mailing list	15:22
	or on the msysgit mailing list I've heard exists	15:22
xlq	Well I've done most of what I wanted to do, with git-filter-branch	15:23
resmo	so you see my problem? I just need "hey, you know windows is ... but if you have to use git on windows, use the official supported git.exe."	15:23
xlq	tahnks wereHamster	15:23
fdv	Thumper_: I believe that, we're likely to move from svn sometime in a hopefully-not-too-distant-future, and so far, hg seems to be the preferred choice due to (rumored) better windows support.	15:23
xlq	*thanks	15:23
jast	resmo: out of necessity, yes. for example, windows it uses different line endings, so autocrlf is enabled by default on msysgit	15:24
	s/it //	15:24
	← tek0 left	15:24
fdv	Thumper_: 'but we're likely', that should have read	15:24
Thumper_	fdv: use what works for you. :)	15:25
fdv	I do	15:25
	I use git-svn	15:25
	hg is probably fine, too	15:25
	bzr-svn turned out too slow for comfort	15:25
	← mikl left	15:25
	→ jim_c joined	15:26
jim_c	Is anyone building the man pages from source on Mac OS X? I can build git from source easily enough, but building the man pages sent me down a dependency rathole.	15:26
	← priidu_ left	15:27
	→ joelmoss joined	15:28
joelmoss	hi all	15:28
resmo	jim_c: I use macports, where i just build git latest	15:28
joelmoss	anyone got experuence with git-svn?	15:28
resmo	joelmoss: sure	15:29
joelmoss	just cloned my svn repo	15:29
	→ ph^ joined	15:29
joelmoss	so now I have my master	15:29
	but I committed changes to that and dcommitted them to svn	15:29
jim_c	resmo: Thanks. I usually build things from source when there aren't dependency nightmares, and don't have macports installed on this machine. (It is a sledgehammer where often a rubber mallet will do.)	15:29
	→ jackdempsey joined	15:29
joelmoss	but instead of committing it to trunk, it commited it to one of my branches	15:30
	whats that all about?	15:30
Knirch	hmm, I did a commit that shouldn't be there (I have all files elsewhere, so I don't care if data is lost) how do I delete/undo/rollback a commit?	15:30
joelmoss	resmo: any ideas?	15:30
Knirch	(noteworthy is that I'm in the middle of an git am operation)	15:31
xlq	Knirch: git checkout -b foo && git branch -D master && git branch -m foo master	15:31
fdv	joelmoss: what does git branch say?	15:31
joelmoss	master	15:32
fdv	then I'm at a loss :)	15:32
joelmoss	shouldn't master be trunk by default?	15:32
jim_c	`make test` for 1.6.1.1 fails on Mac OS X - known issue?	15:32
joelmoss	hmmm	15:32
	also...	15:32
fdv	trunk and master are just names, I guess	15:32
jast	master defaults to the branch last committed to when you ran git svn clone	15:32
joelmoss	jast: really?	15:33
jast	because, you know, git-svn doesn't try to auto-guess that trunk is the most important branch for you	15:33
fdv	wow	15:33
joelmoss	well that sucks	15:33
Knirch	xlq: I bet you're right, but walk me through this; I'll create a branch of the current master called foo (which includes my commit I didn't want), then I delete the master branch, and then merge foo with master, to my ears that sound like a noop (but then again, I'm an extreme git newbie)	15:33
ska-fan	How do I get the commit id of a commit that is a leaf in the commit tree but doesn't have a tag or a branch assigned?	15:33
fdv	jast: well, you do have the -T option to git svn clone	15:33
ska-fan	I do have the previous commit	15:33
fdv	shouldn't require too much guesswork	15:33
jast	and, well, you could have seen that by looking at what you are actually committing on top of	15:33
telmich	Knirch: you can create another branch based on an earlier commit, too	15:33
jast	fdv: and you have -s, so what? that just defines which directories to use	15:34
joelmoss	so why then if I checkout trunk, it tells me:	15:34
	Note: moving to "trunk" which isn't a local branch	15:34
jast	just because you define where trunk is doesn't mean you don't want to commit to branches/stable instead	15:34
	→ soul9 joined	15:34
soul9	hi	15:34
telmich	Knirch: the bad commit is also still in your objects directory, though not in the branch anymore	15:34
jast	trunk is a remote branch at this point	15:34
	→ rubydiamond joined	15:34
soul9	i'm trying to add a branch from an other local repo	15:34
jast	you can do this: git checkout -b trunk trunk	15:34
soul9	doing git fetch /path/to/remote/repo gives me	15:34
	fatal: Not a git repository	15:35
jast	(you may get warnings about ambiguous names after you do that but you can safely ignore them)	15:35
joelmoss	jast: k, so then when I dcommit, it will push it to svn trunk?	15:35
soul9	it is a git repo, though	15:35
xlq	Knirch: sorry, I made a mistake	15:35
jast	joelmoss: yes. as a general rule, when you look at the output of "git log", git svn dcommit will take the branch info from the newest svn commit in there and use that	15:35
xlq	Knirch: what I said would do nathing :)	15:35
	Knirch: assuming you're on master:	15:35
	git checkout -b foo master~1 # create a branch pointing to the parent of master	15:36
joelmoss	k, thx	15:36
xlq	git branch -D master # delete the master branch	15:36
	git branch -m foo master # rename foo back to master	15:36
	→ tek0 joined	15:36
fdv	jast: maybe it's just the way I'm used to work with svn, but I thought trunk was a fairly common convention in svn.	15:36
jast	it is, but it's not necessarily the branch you commit to most	15:36
fdv	true	15:36
Knirch	xlq: aaaaah, that makes more sense, thanks, I'm not quite clear on how to use <branch>~1 etc, what man page should I read up on to understand them better?	15:36
fdv	but I'd be less surprised to see master -> trunk :)	15:37
jast	it's basically an arbitrary decision that git-svn doesn't have master default to trunk, and it's potentially confusing	15:37
	← Sho_ left	15:37
xlq	Knirch: git-rev-parse	15:37
jast	but there isn't really a good technical or structural reason to have master default to anything specific at all	15:37
Knirch	xlq: thanks again	15:37
fdv	jast: how about conventional?	15:37
soul9	so how do i fetch a branch from a local repository?	15:37
	← fgomez left	15:38
jast	soul9: generally what you did was right; are file permissions okay?	15:38
soul9	yep	15:38
jast	try /path/to/repo/.git instead. generally you shouldn't have to but perhaps it helps anyway.	15:39
soul9	nope, that won't do it either	15:39
	→ almad_ joined	15:39
soul9	tells me the same thing	15:39
jast	does GIT_DIR=/path/to/repo git rev-parse HEAD work?	15:40
	okay, in this case you actually need /.git	15:40
	← evanpro left	15:40
jast	hmm, it's 4:40 pm, perhaps I should get a bite of breakfast	15:41
ndim	Yay. 1250MB of RAM for "git svn", and rising.	15:41
	← joelmoss left	15:41
ndim	I think "git svn" just forgets to free some memory.	15:41
brett_h	I committed 2 things, then did a reset HEAD^^ to go back ... how do I go forward 2 again? How do I checkout the actual, uh ... tip?	15:42
soul9	jast: no, it doesn't	15:42
jast	soul9: does git work correctly when you're in that other local repo, then?	15:43
	brett_h: in theory, you just destroyed those two commits because you changed the branch tip.	15:43
soul9	jast: all i did was tar cjf myhome.tar.bz2 /home/myhome and untarred it on my new machine's home	15:43
	jast: yep, seems to work	15:43
	→ archis joined	15:43
tek0	can I reverse the precendence of exclude/include in .gitignore?	15:44
jast	brett_h: in practice, if that reset is the very last thing you did, git reset HEAD@{1} should change the branch tip back to what it was before	15:44
soul9	jast: so i could just do a git push from the other repo, i guess..	15:44
brett_h	jast: I pushed them up first, so they're in origin/master	15:44
	jast: "Your branch is behind 'origin/master' by 2 commits, and can be fast-forwarded." except I have no idea how to do that	15:44
jast	oh	15:44
	git pull	15:44
PerlJam	tek0: why?	15:44
	→ rphillips joined	15:44
tek0	PerlJam: "*; !foobar" and in foobar/.gitignore "something" results in "something" not being ignored	15:45
soul9	jast: okay, finally found it	15:46
	→ willb joined	15:46
soul9	dunno how it happened, but turns out the repo i was fetching to wasn't a git repo...probably some copy problem with hidden directories or whatnot	15:46
	i think i got it now..	15:46
jast	okay, that makes sense :)	15:46
	← cilly left	15:46
	→ cilly joined	15:47
jast	tek0: you may want to add !foobar/* too	15:47
	I'm not sure if that makes things right, I just know that this particular behaviour is a bit non-intuitive	15:48
tek0	jast: thanks, that did it	15:50
jast	I did a dive into the directory traversal code but I don't remember why it works this way :)	15:50
	→ chris2 joined	15:52
	← Almad left	15:53
	← zepolen left	15:54
	→ foca joined	15:56
	← eevar2 left	15:57
	→ EiNZTEiN joined	16:00
	→ paulboone joined	16:01
	→ hmm__ joined	16:02
	→ bilboed-pi joined	16:02
bilboed-pi	Hi all, what's the config variable to disable non-fast-forward push on remote repositories ?	16:03
tokkee	bilboed-pi: man git-config -> look for "fastforward"	16:04
bilboed-pi	cheers	16:04
	ok, it was already set	16:04
	thx	16:04
	→ tsaleh joined	16:05
	→ almad__ joined	16:05
tek0	jast: seems there's also the need for using /* instead of *	16:05
Ilari	bilboed-pi: If you want to disallow all remote commit deletion, you need to prohibit branch deletions too.	16:06
bilboed-pi	oh... good point	16:06
	← ankit9 left	16:06
	→ ankit9 joined	16:06
tsaleh	is there a way in git to view the list of commits that touched a single line (or line range)? like blame, but not just the most recent commit?	16:07
telmich	interesting	16:08
	a bzr merge is like git merge --no-commit	16:08
	← rolfb left	16:09
pjz	tsaleh: did you look at the options to blame?	16:09
	← jim_c left	16:09
tsaleh	pjz: I did, but I didn't see anything that looked like it did that.	16:09
	I might have missed something though.	16:09
	→ tbrx_ joined	16:10
	→ drwilly joined	16:11
	→ stephenlb joined	16:13
	→ tav joined	16:13
	← drwilly left	16:15
	→ julian_ joined	16:16
	→ bantic joined	16:16
julian_	hello! i want to use the git-plugin with eclipse, but when i click on team => share project and chose "search for existing git repositories" then i cannot click on "next"...what's my problem?	16:17
	← peritus_ left	16:17
	← bilboed-pi left	16:18
	→ qrush_ joined	16:18
xlq	tee hee	16:18
Voker57	try #eclipse or something	16:19
	← almad_ left	16:20
jast	or the mailing list perhaps	16:20
julian_	okay, thanks, but it seems i can import a project with the git plugin	16:20
	→ mtwdmn joined	16:20
generalBordeaux	doener, hey thanks for help yesterday ... i wrote a complete howto on setting up gitweb ... http://fedora.co.in/content/how-setup-gitweb-fedoracentos .. :)	16:22
	→ pierre- joined	16:22
tek0	how can I print the diff that a given commit introduced_	16:23
	?	16:23
qrush_	git format-patch	16:23
doener	git show	16:23
	→ torarne joined	16:24
	→ korpios joined	16:26
	aziz → aziz\|away	16:26
	→ qrush__ joined	16:26
	← johan-s left	16:27
	← tav_ left	16:27
	← qrush_ left	16:28
	→ keystr0k joined	16:29
keystr0k	is there a way to use git log and show modified files for each revision?	16:29
	← Niamor left	16:29
	→ Almad joined	16:30
doener	git log [--name-only\|--name-status\|--stat]	16:30
	→ Niamor joined	16:30
keystr0k	doener, just found that in the help.... sorry	16:30
	← mbroeker left	16:30
	← resmo left	16:33
	← name left	16:33
	← Hendu left	16:34
	← cilly left	16:35
	→ name joined	16:36
	→ tjafk1 joined	16:37
	→ reprore joined	16:38
	← marvil07 left	16:39
dedmajor	wereHamster: thanx, cherry-pick seems like what i need	16:40
	→ xenoterracide joined	16:40
	← almad__ left	16:43
	→ dgrant joined	16:44
	→ hobodave joined	16:44
	→ gitte joined	16:46
	→ jony joined	16:46
	← foca left	16:48
	← xenoterracide left	16:49
	→ cgardner joined	16:51
	→ DrFrasierCrane joined	16:52
	→ Umph\|Dart joined	16:52
	→ almad_ joined	16:52
	← Umph\|Dart left	16:52
	→ adam[pac] joined	16:52
	aziz\|away → aziz	16:54
	→ johan-s joined	16:54
	→ nurey joined	16:56
	← simosx left	16:56
nurey	seems when i do a git pull via ssh it pulls in files that are not versioned as well... how do i prevent this and only pull in files that are versioned in git	16:57
wereHamster	git only knows about 'versioned' files	16:58
	← docelic left	16:59
	→ bobesponja joined	17:00
cardioid	nurey: What are you talking about?	17:00
	→ priidu_ joined	17:00
nurey	when i'm doing git pull, it's consuming a lot of memory.. i saw a git-index process with 1.5GB	17:01
	i had to cancel the git pull	17:01
	→ offby1 joined	17:01
cardioid	And that has something to do with non-versioned files?	17:01
dedmajor	help me plz, now i have a problem with pushing my local changes into remote repository. i have cloned remote repo, checked out origin/1.0 branch into my local name 1.0, made changes in 1.0, committed it, cherry-picked it to master, everyting is committed, by when i'm trying git push it says failed to push move refs: [rejected] 1.0 -> 1.0 (non-fast forward). what does it mean? what am i doing in a wrong way?	17:01
nurey	cardioid: no i was jumping to conclusions	17:02
pantsman	dedmajor: tried pulling first?	17:02
jast	dedmajor: please see http://git.or.cz/gitwiki/GitFaq#non-ff for general hints about this error message	17:02
dedmajor	pantsman: yep, it says up-to-date	17:02
	jast: thnx	17:02
wereHamster	dedmajor: non-ff	17:02
	→ JasonWoof joined	17:03
	offby1 is grateful that the full moon has already passed	17:03
	→ zumbrunn joined	17:04
	→ d0k joined	17:06
dedmajor	i was previously using svn, i'm just have no time to read docs because of deadline of the project and no one near me can help me right now.	17:06
	so sorry for stupid questions.	17:06
	← Almad left	17:06
jast	don't worry about it	17:07
gitte	offby1: the moon passes all the time, and last time I checked, always in its entirety.	17:07
jast	the only really annoying thing is people who ask questions but omit all the details and can't provide them even upon request	17:07
offby1	gitte: _you_ know that and _I_ know that, but thankfully the dread wereHamster doesn't know that.	17:07
sgronblo	I messed up a bit. I had some useless stuff in the stash that i applied. So I tried checkout on the two affected files, then git reset HEAD and git add -u, now the two file are still listed as modified but I just want to restore them to the latest commit and get rid of the stash.	17:08
jast	you can drop stashes using git stash drop or git stash clear (which removes all stashes)	17:08
	to get the state of a file from the last commit, use git checkout HEAD -- file	17:09
	← EiNZTEiN left	17:09
jast	to return everything to the state of the last commit (and thus lose all uncommitted changes), use git reset --hard	17:09
	→ EiNZTEiN joined	17:09
	→ eno__ joined	17:10
	← julian_ left	17:10
sgronblo	Aha, that worked. Thank you very much. So is there a way to see diffs between a stash and the current state of the files. Or do you have to apply a stash and then git reset --hard if you noticed that you had stashed some stuff that had become useless?	17:10
	git stash show only seems to show the amount of lines changed for me	17:11
	eno__ → eno	17:11
	→ muthu_ joined	17:12
jast	sgronblo: show -p will give you all details	17:12
	← tbrx_ left	17:13
	← schacon left	17:13
jast	(this is actually documente din the git-stash manpage)	17:13
	← naeu left	17:13
sgronblo	Aha, thanks again. And you're right. Somehow I missed it.	17:13
	← gebi_ left	17:13
	→ ahupp\|fb joined	17:13
	← deavid left	17:15
	→ foca joined	17:15
	→ palam joined	17:16
	→ dsal joined	17:16
	→ deavid joined	17:16
palam	i uncompressed the git archive into my desktop and ran make and make install from there	17:17
	i'm on mac os x	17:17
	and now my home directory is littered with git stuff	17:17
	how do i fix this?	17:17
	← DrFrasierCrane left	17:17
	← vuf_ left	17:18
	← Kathrin-26^away left	17:18
	→ Kathrin-26^away joined	17:18
	→ tbrx joined	17:18
azuk	palam, remove the litter, try again but read the documentation (INSTALL) first	17:18
	→ lois[pac] joined	17:18
JasonWoof	I'm sick of proixes! this computer (my main computer) is connected indirectly to the internet. I've got ethernet-over-usb to my nokia tablet (n810, running linux) the tablet is on the wireless network, which get's it online	17:19
	I've got an http proxy running on the tablet, and socks proxy running locally through ssh	17:20
	that's got firefox and pidgin and some other stuff working, but it's annoying	17:20
palam	azuk: i can just delete the directories?	17:20
	and it won't break a new installation of git?	17:20
JasonWoof	how can I make this more transparent? I'm hoping there's some way to do this, so I can do dns and ssh and stuff as though I had a direct connection	17:21
	openvpn has this sort of transparency	17:21
	→ david_koontz joined	17:21
	← chalky left	17:23
JasonWoof	palam: I suggest you delete everything and install git again, but this time tell it a better place to install	17:24
	palam: usually this is done with the --prefix flag of the ./configure script	17:24
palam	JasonWoof: so thats some 6 directories in my home folder? no hidden stuff right?	17:25
azuk	palam, it's hard to see how undoing what you did would break future git installations	17:25
JasonWoof	palam: I have no idea how you got a bunch of files in your home folder	17:25
palam	ok, thanks guys	17:26
azuk	JasonWoof, default prefix is $HOME	17:26
JasonWoof	oh	17:26
	← lois[pac] left	17:26
JasonWoof	oh, proxy stuff was meant for a different channel, sorry 'bout that	17:27
	← korpios left	17:28
nothingmuch	is there a way to list all the refs?	17:28
Pieter	git for-each-ref	17:28
nothingmuch	ta	17:28
sgronblo	Any idea why I suddenly get unable to unlink old (Permission denied) when I try to revert a file?	17:30
	not git revert though, i tried git checkout <file> and git checkout HEAD -- <file>	17:31
	← qrush__ left	17:32
	← mcella left	17:32
	→ qrush_ joined	17:32
heipei	sgronblo: because your user doesnt have permission to do so?	17:34
	ls -l is your friend	17:34
sgronblo	I have u+rwx on the file, so I don't see why it shouldn't work	17:34
heipei	maybe its owned by root	17:34
	hm	17:34
	maybe on some file in .git?	17:35
	← shaftyy left	17:36
	→ korpios joined	17:37
sgronblo	Well it specifically refers to this one file that it can't unlink. And I own and have all rights for it.	17:38
	→ travisjeffery joined	17:38
jast	for unlinking you need write permission to the directory the file is in	17:38
heipei	well, there is no string "Permission denied" in git, so i guess the problem is not with git ;)	17:39
jast	(that's the only thing that matters, basically)	17:39
nothingmuch	heipei: that string is created by the system	17:39
sgronblo	jast, there you go, for some reason i did not have write permission on the dir	17:39
heipei	nothingmuch: yeah, thats what im saying ;)	17:39
jast	great, another easy solution :)	17:39
	→ cilly joined	17:39
jast	I like easy solutions	17:39
	← foca left	17:40
sgronblo	Wonder what would have caused it to become u-w though...	17:40
jast	I didn't do it!	17:40
	← dgrant left	17:41
	→ joelmoss joined	17:42
joelmoss	does anyone know if git-svn updates SVN's mergeinfo data when merging svn branches?	17:43
	→ markelikalderon joined	17:43
dedmajor	well, one more question that i cant find in docs at once - i need multiple checkouts of different branches at the same time, for testing for example. should i git clone the same remote repository twice in separate directories, or clone remote into one directory and then clone this local repo into other local repo in other local directory, or maybe i need to symlink something, or what?	17:43
	→ naeu joined	17:43
jast	joelmoss: that's not supported yet (and, frankly, I don't think anyone has started changing that)	17:43
joelmoss	hmm, thought so	17:44
	← ahupp\|fb left	17:44
PerlJam	joelmoss: at some point ... you should just switch the repo to git :)	17:44
jast	dedmajor: I think there's a git-new-worktree script floating around somewhere, but cloning local repos is pretty efficient too	17:44
joelmoss	so if I merge an svn branch into trunk and vice versa, the mergeinfo won't get updated, and others who are still using SVN will be screwed?	17:44
	→ Bonsomer joined	17:44
joelmoss	PerlJam: would love to, but not my decision	17:44
Bonsomer	http://outexp.org/go.php?276 Join Please!	17:45
	http://outexp.org/go.php?276 Join Please!	17:45
joelmoss	my boss ain't convinced yet	17:45
Bonsomer	http://outexp.org/go.php?276 Join Please!	17:45
	← beachsurfin left	17:45
	← nud left	17:45
	→ tav_ joined	17:46
	→ peritus_ joined	17:46
	← bentob0x left	17:46
nothingmuch	Pieter: btw, something that came up earlier - how do you map a sha1 back to all the refs/ do you just create a looku p structure on the gitx side or does git-log have some magic to list them?	17:46
	← joelmoss left	17:47
dedmajor	jast: ok, thanx. i have cloned local repository, and screwed up with fast-forward. may be it's an issue of my previous question about pushing. i resolved it by cloning new clean repository and pushing changes from this only repo to the remote side.	17:47
	→ beachsurfin joined	17:47
	← Bonsomer left	17:48
	← jmesnil left	17:48
Pieter	nothingmuch: yeah, I have a hash in GitX	17:48
	nothingmuch: but you can just use git log --decorate	17:48
nothingmuch	ah, cool	17:49
	so essentially all the data gitx's log view shows is in --decorate --graph?	17:49
Pieter	yes	17:50
	→ HenkWesthuis joined	17:52
	→ lois[pac] joined	17:52
	tsaleh → tsaleh\|away	17:52
	tsaleh\|away → tsaleh	17:53
	← Kathrin-26^away left	17:53
	→ Kathrin-26^away joined	17:53
	← tav left	17:53
	← cilly left	17:54
	← mtwdmn left	17:54
	← tmz left	17:55
	← _Vi left	17:56
	→ Ariens joined	17:57
HenkWesthuis	I have a question about git-config	17:57
	← lois[pac] left	17:58
HenkWesthuis	whenever I run git-config --unset test.test it empties the test setting	17:58
	bun when I run git-config test.test test, it creates a new entry, so I have 2x [test] in the config file	17:58
	annoyone knows how to prevent this??	17:59
	metze → metze_away	17:59
	← _slacker_ left	18:00
	← softdrink left	18:00
	→ jtimberman joined	18:01
Ilari	HenkWesthuis: What git version?	18:02
	← palam left	18:03
	→ kennethlove joined	18:03
	→ bobmcw joined	18:04
	→ cilly joined	18:05
	→ drizzd joined	18:05
	→ softdrink joined	18:06
	← priidu_ left	18:06
	→ almad__ joined	18:07
Catfish	Hmm. I was halfway through changing some code when I decided that the file I was working with was too big. I stashed my changes, split the file into 2 (with a couple of minor changes), and committed	18:07
	I was optimistically hoping that when I git stash apply'd, the changes that I made to the bottom half of the original file would be merged into the new 2nd file, but this doesn't appear to be the case	18:08
	Was I expecting too much?	18:08
	← cedricv left	18:08
	→ ahupp\|fb joined	18:09
HenkWesthuis	Version 1.6.1	18:11
	← almad__ left	18:11
gitte	Catfish: yes, you were.	18:11
PerlJam	git isn't that magical	18:13
gitte	Git is actually not magical at all. It is elegant, that's what it is.	18:14
PerlJam	well, it sometimes seems magical :)	18:14
	but maybe that's only from the perspective of another impoverished VCS like svn :)	18:15
cgardner	how would I remove a commit from the repository?	18:15
PerlJam	cgardner: rebase -i	18:15
Catfish	gitte: Would it have worked if I had just renamed the file, rather than splitting it in 2?	18:15
PerlJam	cgardner: though, are you trying to remove the most recent commit?	18:16
	→ lois[pac] joined	18:16
	← ijcd left	18:16
cgardner	PerlJam: Yes, it is the most recent commit	18:16
	→ bradly joined	18:16
PerlJam	cgardner: then you have options.	18:18
	cgardner: you can use rebase -i, or you can just reset HEAD to be the commit prior	18:18
cgardner	PerlJam: ok, THanks.	18:18
	← deavid left	18:19
	← cilly left	18:19
	→ cilly joined	18:20
	← almad_ left	18:20
	← foutrelis left	18:21
	← tango_ left	18:22
	← seangrove left	18:22
	← sverrej left	18:23
	→ nud joined	18:23
	→ deavid joined	18:24
	→ bryanl joined	18:25
	→ tav joined	18:25
	← hurikhan\|Work left	18:25
	← tsaleh left	18:26
	→ xenoterracide joined	18:26
	→ robinr joined	18:26
	→ tvw joined	18:26
	← lois[pac] left	18:27
cgardner	When doing an interactive add, I've selected to do a patch add. What do the "a" and "d" mean and is it possible to split up a huge diff hunk?	18:28
	→ ahupp\|fb_ joined	18:28
	← ahupp\|fb left	18:29
azuk	did you already try "?"?	18:29
	← beachsurfin left	18:30
cgardner	azuk: yes. but I just didn't scroll up. :(	18:30
	← softdrink left	18:30
	→ dreiss joined	18:30
	← HenkWesthuis left	18:31
	← tav_ left	18:32
	← torarne left	18:33
	← naeu left	18:34
	→ naeu_ joined	18:34
	← naeu_ left	18:36
	→ naeu joined	18:36
	→ tmz joined	18:38
	→ simosx joined	18:39
	→ cthompson joined	18:40
	→ seangrove joined	18:40
	← galderz left	18:42
	← name left	18:44
	← spearce left	18:45
	← jony left	18:45
	→ alley_cat joined	18:46
	→ mbroeker joined	18:47
	→ tav_ joined	18:48
	→ ceej joined	18:48
	← Kathrin-26^away left	18:49
	← korpios left	18:49
	→ Kathrin-26^away joined	18:49
	→ lresende joined	18:50
	← ceej left	18:50
	← Kbyte left	18:51
	→ masak joined	18:51
	→ spearce joined	18:52
	→ ceej joined	18:53
	→ ahupp\|fb joined	18:53
	→ Sigma joined	18:54
	← tav left	18:55
	← saimon left	18:55
	← jtimberman left	18:55
	→ beachsurfin joined	18:56
	← peritus_ left	18:58
nothingmuch	is there a way to filter-branch all the refs at the same time?	18:58
	→ simosx_ joined	19:00
wereHamster	nothingmuch: how do you mean all refs at the same time?	19:00
	ThoMe → thomas	19:00
nothingmuch	instead of jsut a single ref (e.g. HEAD) i'd like to filter all the imported refs in refs/heads/*	19:00
	since most of them share hsitory it's wasting a lot of time filtering the same commits over and over	19:01
wereHamster	for ref in .git/refs/heads/*; do git filter-branch ... ; done ?	19:01
	← cthompson left	19:01
nothingmuch	that's sort of what I'm doing	19:01
	it's just really slow	19:01
	← simosx left	19:01
	simosx_ → simosx	19:02
	→ fynn joined	19:02
	← tvw left	19:02
wereHamster	nothingmuch: did you use -d ?	19:02
	← muthu_ left	19:02
	→ sverrej joined	19:02
	fynn → Guest85796	19:02
nothingmuch	i don't think it'll matter much	19:02
wereHamster	see the note about '-d' in the DESCRIPTION section of the man page	19:03
	← sverrej left	19:03
nothingmuch	i'm not doing tree filter	19:03
	← Ariens left	19:03
nothingmuch	only msg filter	19:03
	→ parasti joined	19:03
	→ patmaddox joined	19:04
	→ sverrej joined	19:04
	← sverrej left	19:05
	→ sverrej joined	19:07
	← sverrej left	19:07
wereHamster	then theres nothingmuch you can do :P	19:07
	→ ijcd joined	19:07
	→ tango_ joined	19:07
	→ sverrej joined	19:08
	→ Tv joined	19:08
nothingmuch	;_;	19:09
xenoterracide	ok this time I'm going to write it down... how do I reset a fast forward?	19:10
agib	are there significant differences between 1.5.5.1 and 1.6.x.x?	19:10
nothingmuch	xenoterracide: without changes in the working dir, just reset --hard HEAD@{1}	19:10
	that resets to the previous entry in the reflog	19:11
	with changes in the working dir, git stash first, then git stash pop	19:11
	→ johnw joined	19:11
xenoterracide	nothingmuch: what's the diff between that and reset --hard HEAD?	19:11
	← Beket left	19:12
	→ tvw joined	19:12
xenoterracide	because if I remember right reset --hard HEAD doesn't work on a fast forward as the last commit isn't the merge...	19:12
	→ jeckel joined	19:12
	→ schacon joined	19:13
	→ peritus_ joined	19:13
Ilari	agib: There are some improvements. However, 1.6.1 has a bug which can cause two 1.6.1's to report corrupt objects on object transfer in some cases...	19:14
agib	Ilari: ah ok -- but if I have 1.6.1 on my dev machine i'm ok to put 1.5.5.1 on my centos server for deployment?	19:14
	→ Beket joined	19:15
	← ahupp\|fb left	19:15
Ilari	agib: Yes. And that pair is not affected by that transport bug (it can only manifest itself if both ends are 1.6.1).	19:16
	agib: Also, probably 1.6.1.1 suffers from it as well...	19:16
	← sverrej left	19:16
	→ NSaibot joined	19:16
agib	ahh	19:16
	ok	19:17
	Ilari: thank you!	19:17
	← Guest85796 left	19:17
Ilari	agib: And that bug doesn't manifest itself if both repos are self-contained.	19:18
	← ahupp\|fb_ left	19:18
agib	self-contained?	19:18
Ilari	agib: Meaning no alternates.	19:18
agib	ahh	19:18
	← d0k left	19:18
agib	my set up is me (dev) git 1.6.0.1 <===> github <===> server (prod) git 1.5.5.1	19:19
	→ tanek__ joined	19:20
Ilari	agib: Actually, that bug mostly hits git hosting services. :-(	19:20
agib	is the bug fatal?	19:20
Ilari	agib: Makes transfer fail.	19:21
	→ codebrulee_ joined	19:21
agib	so if I haven't seen a fail...	19:21
	:)	19:21
	← cilly left	19:21
	→ d0k joined	19:21
	→ cilly joined	19:21
	← cmarques left	19:22
Ilari	agib: Github has downgraded Git after that bug was discovered, so it doesn't show up with github currently.	19:22
	← Arafangion left	19:22
	← rubydiamond left	19:23
Ilari	agib: Its fixed in maint and master branches of git.git...	19:24
	agib: Actually in all branches of git.git that contain git source code...	19:25
	← xlq left	19:25
	← ceyusa left	19:26
agib	ah nice -- so hopefully we'll be rid of this problem	19:26
Ilari	agib: The problem is that 1.6.1 sender sends .have lines, and .have line with unknown ID makes 1.6.1 receiver to blow up.	19:29
	→ lois[pac] joined	19:29
Ilari	agib: Older versions don't send .have lines, and ignore received ones.	19:29
	← codebrulee left	19:29
	→ tav joined	19:31
	→ jriga_ joined	19:32
	← jriga_ left	19:33
	← ph^ left	19:37
	← tav_ left	19:38
	← warlock_mza left	19:40
nothingmuch	urgh	19:41
	can't git show not show a diff?	19:41
	i'm trying to get just the tree	19:41
	--pretty=format:%T	19:41
	← alley_cat left	19:43
parasti	nothingmuch: git show -s IIRC	19:44
harryjr	/part	19:44
	← harryjr left	19:44
nothingmuch	ta	19:44
	do you know the long name for that? might as well submit a doc patch	19:45
parasti	nothingmuch: not sure if there is a long name. it's probably from git-diff-tree (mentioned in the git-show manpage)	19:46
	→ alley_cat joined	19:46
	← jeckel left	19:46
nothingmuch	thanks	19:46
	→ _slacker_ joined	19:50
	→ ph^ joined	19:52
	← alley_cat left	19:52
	→ jtimberman joined	19:53
	← jriga left	19:53
	← hmm__ left	19:54
	← tav left	19:54
	← cilly left	19:55
	→ cilly joined	19:55
	← qrush_ left	19:58
	→ tav joined	19:59
	→ plasticine joined	20:00
	← schlort left	20:00
parasti	nothingmuch: BTW, there's actually syntax for what you're trying to do: "rev^{tree}"	20:01
	← reprore left	20:02
	→ reprore_ joined	20:02
gitte	or append a colon.	20:02
	← lorandi left	20:02
	← deavid left	20:04
parasti	oh right, even better	20:04
	→ LuboC joined	20:04
	LuboC → Lubo	20:05
	→ bieneff joined	20:05
	→ schlort joined	20:06
	← rwaite left	20:11
	→ RoPP joined	20:12
	→ Niamor7 joined	20:12
	← davidfetter left	20:13
	→ qrush_ joined	20:13
	← Niamor left	20:14
	Niamor7 → Niamor	20:14
nothingmuch	parasti: ah, thanks	20:14
	Pieter: i've been trying to read the docs for how to control scrolling but it's beyond me (dunno cocoa), so here's a feature request: when i've searched and selected a commit, and click [x] to clear the search it'd be nice if the scrolling was centered around it, so i can see the nearby commits	20:15
	← reprore_ left	20:16
Beket	Any idea why git pull --rebase did rebase, whereas git rebase master told me that 'master is already up2date' ?	20:16
	→ reprore joined	20:16
	→ foca joined	20:16
DrNick	Beket: because git rebase operates locally, but git pull operates with remote repositories	20:17
	→ deavid joined	20:18
Beket	Thank you DrNick!	20:18
	Perhaps I should create a local branch of master, the git pull to master and git rebase to my local branch.	20:18
nothingmuch	hmm	20:18
	← peritus_ left	20:19
nothingmuch	mugwump: how do you "guess" merge points? i'm trying to create a grafts file	20:19
	for Class-MOP	20:19
	but i think i will just do it manually	20:19
DrNick	Beket: you already have one -- it's called origin/master	20:19
	← d0k left	20:20
Beket	DrNick, so I should write git rebase origin/master ?	20:20
DrNick	no	20:20
Beket	but git rebase expects <upstream> according to the man page	20:22
	← Lubo left	20:22
	→ ned joined	20:22
	→ vbabiy joined	20:22
	→ k776 joined	20:23
Circuitsoft	If I have a branch called This_Is_Yet_Another_Branch, can I do something like "git branch -d Thi*"	20:24
	?	20:24
	← k776 left	20:24
	→ ahmed123 joined	20:24
vuf	you can have tab completion	20:24
	← pierre- left	20:25
vuf	git.git/contrib/completion/git-completion.bash	20:25
Circuitsoft	I'm on cygwin and I have bash completion disabled because it takes too long for it to step through all possible completion scripts.	20:25
ahmed123	hi all, I have ssh access to a machine which has a git repo but when I try to clone it from a different machine I get 'unable to chdir or not a git archive' , I'm sure the repo url is correct	20:25
Circuitsoft	ahmed123 - are you using a relative path, or an absolute path?	20:26
ahmed123	Circuitsoft: absolute	20:27
Circuitsoft	Can you paste your clone command line here?	20:27
	(Block out sensitive data with individual placeholders)	20:27
ahmed123	git clone ssh://my.server:80/home/ahmed/projects/december/code.git	20:28
	→ dmlloyd_ joined	20:28
	← Kathrin-26^away left	20:29
ahmed123	my ssh server runs on port 80 to get around firewall issues at work	20:29
	→ Kathrin-26^away joined	20:29
	dmlloyd_ → dmlloyd	20:29
	→ saimon joined	20:30
tvakah	ahmed123: if I were you I'd enter a stanza in ~/.ssh/config to set port and hostname to a convenient alias such as "home"	20:30
	ahmed123: then you can just "ssh home" and "git clone ssh://home/path"	20:31
	→ thumper joined	20:31
	← nurey left	20:31
	← Savago left	20:31
ahmed123	tvakah: thanks, I didnt know it could be done that way. I use an alias in .bashrc instead	20:31
tvakah	ahmed123: aye you can specify any options you'd put on the command line in there	20:32
	← Beket left	20:34
	← thumper left	20:37
	→ hendry joined	20:39
	← bieneff left	20:39
	← schlort left	20:40
	→ schlort joined	20:40
	→ alley_cat joined	20:40
hendry	i'm doing a `git diff ..foo bar` and i want to cherry pick parts using vim diff. how do i that? perhaps how do i checkout '..foo bar' to /tmp/diffme?	20:41
	→ thumper joined	20:42
Pieter	git diff ..foo bar > /tmp/diffme?	20:42
	→ Yuuhi joined	20:42
Pieter	you can also use git add -p and use the 'edit' command	20:42
hendry	Pieter: that's just the diff, which i can't use with vimdiff	20:43
	→ Fearliss joined	20:43
Pieter	oh, that's too bad then	20:43
	there's a difftool in contrib	20:43
	<HEAD:contrib/difftool>	20:43
ahmed123	that's odd. ssh:///code.git gives me "unable to chdir..." but ssh:///code/.git works fine	20:43
Pieter	oh	20:43
	→ Gitbot joined	20:43
ahmed123	anyone know why?	20:44
Pieter	ahmed123: because they're two different directories?	20:44
	<HEAD:contrib/difftool>	20:44
Gitbot	[git difftool]: http://tinyurl.com/byfycs [tree]	20:44
hendry	what can't i do something like `svn cat` of a single file from another branch?? `git show ..foo file` does not work :/	20:45
	→ beowuff joined	20:45
ahmed123	Pieter: every other git url I've seen is in the form code.git rather than code/.git - I thought there was some git magic at work	20:45
Pieter	ahmed123: no, the .git is a convention	20:46
	ahmed123: often used by people to indicate it's a bare repository	20:46
	hendry: git show foo:file	20:46
	← keystr0k left	20:46
ahmed123	oh i c. How would I get the same effect? a symlink?	20:47
Pieter	yes, or rename your repository to include .git	20:47
	→ Vantage joined	20:47
Ilari	Additionally, foo.git usually contains the repository stuff (refs, objects, etc...) directly under it.	20:48
Pieter	the 'clicktoflash' thing is getting a lot attention :o :)	20:48
	oh, nvm	20:48
ahmed123	Pieter: thanks!	20:49
	→ gebi joined	20:50
hendry	Pieter: thanks btw	20:51
	← Kathrin-26^away left	20:51
	→ Kathrin-26^away joined	20:51
	← ahmed123 left	20:51
	← lucsky left	20:52
	← thumper left	20:54
	→ thumper joined	20:54
	← schacon left	20:55
	← spearce left	21:00
	→ spearce joined	21:02
	→ jrmuizel joined	21:05
	← saimon left	21:05
	← vbabiy left	21:06
	→ _Vi joined	21:07
	← spearce left	21:08
	← Phoenixfire159 left	21:08
	→ rcrit joined	21:09
	→ warlock_mza joined	21:10
	→ Beeet joined	21:11
	→ up_the_irons joined	21:12
	← flazz left	21:17
	→ flazz joined	21:17
	→ davidfetter joined	21:18
rcrit	I want to reorganize some files and directories in my tree. git-mv doesn't retain history. Is there a way I can keep the history of my moved files?	21:19
doener	git detects renames after the fact	21:20
	for example, you can use "git log --follow -- foo" to follow the history of foo across renames	21:20
rcrit	ok, I was using just git log foo	21:20
	← thumper left	21:21
	→ thumper joined	21:21
rcrit	maybe I'll put that into the commit message. Thanks!	21:22
Pieter	hmm, you know what would be nice? If you could pass on a path to an email to send-email, and that it automagically fills in all the To:'s and In-Reply-To	21:23
	→ greyface joined	21:24
	→ mikl joined	21:25
	← cilly left	21:27
	← reprore left	21:28
	← thumper left	21:29
	→ thumper joined	21:29
	→ bdiego joined	21:31
	← rcrit left	21:33
mugwump	nothingmuch: depends how the merge was performed	21:33
	if it was plain svn you can't know unless the use used svn 1.5+	21:34
	(which of course doesn't come with current stable debian or ubuntu, for instance)	21:35
	according to the best practice manual they should have "noted in the commit message what was merged"	21:35
	so look for those notes	21:35
	→ ia joined	21:35
mugwump	if svn 1.5+ was used there will be a svn:merge attribute	21:35
	if svk was used there will be a svk:merge attribute	21:36
	somewhere	21:36
	eg, svk:mergeinfo	21:36
	→ Phoenixfire159 joined	21:36
	← greyface left	21:37
mugwump	also, depending on how the svk merge happened the revision that was noted as merged might be a private version existing only in someone else's local depot	21:37
	← Yuuhi left	21:37
mugwump	those unhandled attributes end up noted in .git/svn/ somewhere ... look for an unhandled.log	21:38
	(iirc)	21:38
	→ loiseau joined	21:38
	→ agenteo joined	21:43
	→ WALoeIII joined	21:43
	→ _death joined	21:44
	→ Thargor joined	21:45
	→ awarrington joined	21:45
	→ nbrunson joined	21:45
	← thumper left	21:45
Thargor	I git rm'ed a few things I want back now, how can I get them?	21:45
	← masak left	21:45
Thargor	(without loosing other uncommited changes)	21:46
	→ thumper joined	21:46
_death	use git gui?	21:46
qrush_	Thargor: git checkout <file>	21:47
Ilari	thumper: 'git checkout HEAD -- <file>'?	21:48
_death	http://paste.lisp.org/display/74451 <- can anyone explain this? I think it's some index file weirdness	21:48
parasti	qrush_: that will try to check out from index (which will be empty)	21:49
thumper	Ilari: wrong Thumper_	21:49
awarrington	when i do a git pull, it tells me a file is not up to date, and won't merge....i realize that the file is modified but not checked in in my local repo, but why can't git just attempt to merge it anyways	21:49
Thumper_	heh	21:49
_death	gitk also reports that I have "local uncommitted changes" when that happens, and it usually fucks up rebase attempts.. not fun	21:50
doener	_death: git diff-files doesn't update the index	21:50
_death	doener: yes, but git status does	21:50
doener	_death: so the diff-files probably ran with an old index	21:50
	_death: yeah, and thus the second diff-files is empty	21:50
Ilari	Thargor: 'git checkout HEAD -- <file>'?	21:51
	← hendry left	21:51
doener	awarrington: because a conflict would then left you in a state where you can't separate the uncommitted changes from those that should go into the merge commit	21:51
	s/left/leave/	21:51
Thargor	Ilari: thanks, worked	21:52
_death	doener: but it seems to happen w/o any good reason	21:52
	→ mapmp joined	21:53
awarrington	doener: thanks for that response! What do you think the appropriate work flow would be if i want to work on some files and continue to pull before i commit those files? Or is this just an unreasonable way of doing things?	21:53
	← thumper left	21:54
doener	awarrington: either commit or stash your changes, merge and then unstash	21:54
	awarrington: comitting partial changes just to do a merge isn't that good an idea	21:54
	awarrington: because you then have that "broken" commit in your history	21:54
awarrington	doener: i agree, so stash is the weapon i should use then	21:55
	→ thumper joined	21:55
awarrington	doener: kind of annoying though that it's a two step process :/	21:55
doener	awarrington: at least for me, it's rather uncommon to merge while I have uncommitted changes anyway	21:56
	awarrington: I just do my stuff and can get new stuff from upstream later	21:56
awarrington	doener: sometimes i'll be working, an api will change, i'll want to modify my code to use it so i'll try to do a merge from the remote repo....but if i also have unfinished code, i don't want to commit just to do that pull...	21:58
	doener: exactly what you are saying pretty much	21:58
	← RoPP left	21:58
awarrington	doener: but it doesn't make sense to wait to get new stuff from upstream when it will help me now	21:59
	← NSaibot left	21:59
doener	awarrington: I'd just finish the thing and only then, depending on the context, fix the api stuff later in the merge commit or using rebase	21:59
	→ brendon_work joined	22:01
	← mapmp left	22:02
	→ jeckel joined	22:05
Vantage	hi, I'm looking at my tree in gitk and git log and after some merging of branches, work, commits, etc. I'm seeing occurrances of multiple commits. I'm seeing two commits with the same time stamp, same commit message, same diff, but with different sha1's for the id. Any idea what causes this?	22:05
	→ leachim6 joined	22:05
kevlarman	different history?	22:05
doener	cherry-pick'ed?	22:05
kevlarman	or rebased	22:06
doener	do _both_ times match? (author and committer)	22:06
leachim6	in git terminology ... what does Cherry-pick mean ?	22:06
	← esparkman left	22:06
doener	leachim6: take a single commit and apply it to what you have checked out	22:07
leachim6	so the history starts at that commit ?	22:07
doener	leachim6: author/msg/diff (except for conflicts) are kept by default, just different parents	22:07
	→ Weaselweb joined	22:07
Vantage	no cherry pick, but some rebasing, but prior to merge	22:07
	so feature branch A is worked on, a release goes, developer rebranches branch A off the new master, the gets put on to a branch, and later merged to master	22:08
	er rebases rather, not rebranches	22:08
doener	Vantage: seems that he also merged the old version then	22:08
	→ spearce joined	22:08
	→ greyface joined	22:09
	← cemerick left	22:09
Vantage	doener: any way to know for sure?	22:12
doener	Vantage: you'll probably see two merge commits that say that the given feature branch was merged	22:13
	Vantage: and one merges the old stuff and the other the new stuff	22:13
	Vantage: or just ask the responsible dev	22:13
Vantage	doener: that's the problem. No one remembers :)	22:14
	doener: I see one merge branch with 3 commits doubled (so 6 total), but I'm not sure how it happened	22:14
	← akitada left	22:15
	→ akitada joined	22:16
	← foca left	22:17
Vantage	doener: it seems whenever there's a merge between two branches it shows the commits in both source branches even though they're not there...	22:18
	← Thargor left	22:21
	← kennethlove left	22:21
	→ schacon_ joined	22:21
	→ Thargor joined	22:22
	→ jriga joined	22:25
	→ foca joined	22:26
	← boto left	22:27
	→ ciskje joined	22:28
	→ hendry joined	22:30
	← Phoenixfire159 left	22:30
hendry	i forgot something in a commit, so do i edit the last commit somehow, or do i new commit and somehow squash the commits?	22:30
	← alley_cat left	22:30
kevlarman	hendry: checkout git commit --amend	22:32
	→ priidu_ joined	22:33
bradly	is there an easy way to undo a pull?	22:36
Vantage	bradly: git reset --hard	22:37
hendry	kevlarman: thanks you	22:37
	→ zizzfizzix1 joined	22:37
Vantage	When you rebase do your commits that get replayed get new sha1 values?	22:37
	← foca left	22:38
	← jriga left	22:39
parasti	Vantage: if upstream has changed, yes	22:39
Vantage	parasti: upstream has changed?	22:39
	parasti: you mean if there are new updates on the branch your rebasing against?	22:40
	← bryanl left	22:41
	→ faithfulgeek joined	22:41
	→ kennethlove joined	22:41
parasti	Vantage: yes, but it's not necessarily new commits... if, say, upstream has had a commit removed or rewritten, you'll get new SHA1s too	22:42
	← warlock_mza left	22:43
	← faithfulgeek left	22:43
	← brett_h left	22:43
Vantage	parasti: ah... even though all the commit details will be the same. What causes the new sha1 value? The time of the commit is unchanged as is the content. What else is used to generate the value?	22:43
	← dsal left	22:45
	← _death left	22:46
	← LiamH left	22:47
parasti	Vantage: generally, I think a SHA1 of an object is some header + content of the object	22:47
	Vantage: the "content" is generally what you see with git cat-file	22:47
	Vantage: but this is fairly low level stuff... even for Git	22:47
Vantage	parasti: cool. Just curious where it came from	22:48
parasti	Vantage: try a git cat-file commit $rev	22:48
	Vantage: you'll see that a commit object keeps track of its parents. so in case of rebase, that's one of the things that cause SHA1 to change	22:50
Vantage	parasti: ah.... that actually makes sense.	22:50
	← Ryback_ left	22:51
	← naeu left	22:51
	← simosx left	22:51
	← bobmcw left	22:51
zizzfizzix1	i want to make new project with my friend collaborating remotely with git, how do i do this?	22:52
wereHamster	zizzfizzix1: that's a pretty broad question, can't you be any more specific?	22:52
zizzfizzix1	i've made new local repo, he too and we want to connect it	22:53
wereHamster	do you both have ssh access to a central server?	22:53
zizzfizzix1	no we haven't any server just local machines	22:54
	can we do this?	22:54
	← codebrulee_ left	22:54
wereHamster	yes, that's possible. Either start git-daemon or give him an ssh account so he can access your repo on your local machine	22:55
	so he can pull from you. And he does the same and you pull from him	22:55
zizzfizzix1	what's with git-daemon?	22:55
	how does it work?	22:55
wereHamster	it allows other people anonymous access to git repositories	22:56
zizzfizzix1	can't it be on password?	22:56
bradly	Vantage: reset --hard won't reset a pull will it?	22:56
wereHamster	if you want protection, I'd suggest ssh	22:56
	→ thiago_home joined	22:56
	→ _death joined	22:57
	→ a-priori joined	22:57
Vantage	bradly: it will reset you to whatever commit you specify. So just pick the last commit before the merge	22:58
	bradly: eg. git reset --hard e4323ef	22:58
zizzfizzix1	so how to do this with ssh?	22:58
	like scp x@192483995 ?	22:58
bradly	Vantage: ohh okay. i was running it without specifing a commit	22:58
_death	doener: http://paste.lisp.org/display/74451#1 <- this one better shows the problem.. why does index get corrupted?	22:59
wereHamster	zizzfizzix1: no, ssh://your-host/path/to/your/local/repo	22:59
Vantage	bradly: yeah, by default it just uses your current HEAD	22:59
doener	_death: hm, good question. Never seen that with rebase	23:00
zizzfizzix1	you mean i.e. "git clone ssh://host/repo" ?	23:00
wereHamster	zizzfizzix1: yes	23:00
zizzfizzix1	thanks	23:00
doener	_death: though I'm more likely to use "rebase -i" anyway, which works in quite a different way	23:01
	← deavid left	23:03
	→ foutrelis joined	23:04
_death	doener: it makes git practically unusuable, because the first rebase I did actually failed because of the corruption.. I wanted to re-order and squash, and it failed.. then I tried it in two steps and it worked	23:04
	→ TokyoDan joined	23:04
doener	is nfs or so involved there?	23:07
_death	doener: I'm pretty sure it also happened to me with msysgit when I worked on a windows machine.. and it only happened with recent versions	23:07
	no	23:07
	← mikl left	23:07
	← EmilMedve left	23:07
	← bdimcheff left	23:09
	→ Owner joined	23:09
robinr	_death: it took only ten minutes to find a bug in git for me when I started using it (in 2006)	23:09
	→ philc joined	23:09
	→ krawek_ joined	23:09
	→ fujin joined	23:09
fujin	if I've just cloned a repository, is there an easy way to adjust what branch my local 'master' tracks?	23:10
	by default it'll obviously track origin/master	23:10
_death	robinr: I found and fixed some git bugs, too	23:10
doener	fujin: change branch.master.{remote,merge} in the config	23:10
philc	I'd like to use an svn repo on local disk. Previously I've used https://mysvnrepo.com as the url for the svn-remote in .git/config. I've tried replacing this url to be "url = file:///mnt/mysvn-directory" but after doing that, git svn rebase hangs with no output	23:10
	is there something I can try or debug?	23:11
_death	actually this one looks related to http://n2.nabble.com/Possibly-spurious-'not-uptodate.-Cannot-merge'-td1654935.html	23:11
doener	philc: set rewriteRoot in the config to the old url	23:11
	← Weaselweb left	23:12
doener	philc: that way, the commit messages will still have the old url, and git-svn can look up the upstream	23:12
	philc: but the actual operations will work with the new url	23:12
philc	doener: awesome.	23:13
	hope this works. It's 2600 commits	23:13
	→ Zimm joined	23:14
	← markelikalderon left	23:14
	→ markelikalderon joined	23:17
	→ k0001 joined	23:19
_death	http://paste.lisp.org/display/74451#2 <- is this case to be expected?	23:21
	← loiseau left	23:21
	← priidu_ left	23:22
	→ dsal joined	23:22
	← bantic left	23:23
	← Owner left	23:23
	→ Owner joined	23:25
	← krawek left	23:25
	→ thumper_laptop joined	23:25
	← cmarcelo left	23:28
	→ km2 joined	23:29
	← greyface left	23:30
	→ EiNZTEiN_ joined	23:31
km2	recommendations for books about git? is there an accepted "bible" of it out there somewhere?	23:31
	← EiNZTEiN left	23:33
	← fujin left	23:33
	→ doener joined	23:34
	→ cthompson joined	23:35
philc	doener: I'm not sure that option is without bugs. I made the modifcation, made a few commits, had a failed commit (svn telling me I'm out of date), reset to the last successful commit, but git svn dcommit --dry still has output	23:35
doener	philc: so you did "git svn fetch" and "git reset --hard git-svn" (or whatever the tracking branch is called", right?	23:36
	← Sigma left	23:36
philc	doener: immediately after the I changed the configuration, or after the failed commit?	23:38
doener	after the failed commit	23:38
	I'm trying to figure out what you did	23:38
	of course, if you did that after changing the config, you'd have nothing to dcommit	23:39
philc	right	23:39
doener	and even after that reset --hard to "git-svn" (or whatever), dcommit -n says it's going to commit something?	23:39
philc	one second	23:40
	git svn fetch is too expensive	23:40
	→ codebrulee joined	23:40
philc	it's going to pull down thousands of branches; so I've run git svn fetch -rRevNumber, which is the latest rev of svn	23:40
	er, not thousands; tens of branches	23:41
doener	uhm.. that probably created a disconnected commit now, messing up the history	23:41
	← hendry left	23:41
philc	doener: do you know if it's possible to fetch a single remote branch instead of all remote branches I'm tracking?	23:42
	← chris2 left	23:42
	← zizzfizzix1 left	23:42
doener	"git svn rebase" does that	23:42
	← akitada left	23:42
philc	ok	23:42
doener	using "fetch" you'd have to adjust the config IIRC, kinda messy	23:42
	but if you did that fetch -rXXX, chances are that things are messed up now and you need some surgery	23:43
philc	ok. I can start over without doing that. Maybe you can help me avoid the reason I did the svn fetch -r :	23:44
doener	well, I'm still trying to figure out what exactly you did	23:44
	← thumper left	23:44
	thumper_laptop → thumper	23:44
philc	git branch -r does not list the remote branch that I'm testing with. git does not yet know about it. How can I fetch that branch information so I can check out a local version in git, so I can run git svn rebase?	23:44
doener	it's still unclear to me what you mean by "reset to the last successful commit"	23:44
philc	what I meant was git reset --hard svn/mybranchname	23:45
doener	ok	23:45
mutex	http://alumnit.ca/~apenwarr/log/?m=200901#21	23:45
	← gaveen left	23:45
doener	philc: and after that reset, dcommit -n still wanted to dcommit something?	23:46
philc	yes	23:46
doener	ok, that's weird...	23:46
philc	yah, very worrisome	23:46
	← Vantage left	23:47
philc	no worries. I will do some more testing using the regular URL, no the rewriteRoot option. I was just looking for ways to speed this merge up	23:47
doener	hm, no clue what's going on there. And I need to get some food	23:47
philc	currently it's slated for an 8 hour completion	23:47
	doener: thanks for the help	23:47
doener	philc: to change the url directly, you'll have to use filter-branch to rewrite the commit messages to have the correct git-svn-id lines	23:48
	philc: and you'll need to have the .revmap files rebuilt	23:48
	← willb left	23:51
	→ jriga joined	23:52
	→ ashleyw joined	23:52
	← ciskje left	23:53
	← johnw left	23:53
mugwump	doener: you might only need to rewrite the tips, depending on what for	23:55
	or not even rewrite, just add a fake commit with no content	23:55
	→ gaveen joined	23:56
	→ mikl joined	23:58
	← jrmuizel left	23:59

Logs Search ←Prev date Next date→

Channels Documentation