IRCloggy #git 2008-09-09

Provider of IRC logs since 2005.

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

	2008-09-09
	→ rmh3093 joined	00:00
	← glguy left	00:01
	← statim left	00:02
	→ lu_zero_ joined	00:02
	← rmh3093 left	00:05
	→ rtomyako joined	00:05
	← rtomayko_ left	00:05
	→ rtomayko joined	00:05
	→ rmh3093 joined	00:07
	← [k]omplex left	00:07
	→ brockp_ joined	00:07
	← mithro left	00:07
brockp_	Quick question on something that stumped me, I have a repo thats really just a clone of my local, its how i publish my traning documents to the world,	00:08
Eridius	what's a traning document?	00:08
brockp_	Is there a way to delete branches on this remote repo?	00:08
Eridius	git push remote :branch	00:08
brockp_	just latex source,	00:09
	Eridius was making fun of you for misspelling "training" as "traning"	00:09
brockp_	Its a common problem I have,	00:09
	← dejanpan left	00:09
	→ dejanpan joined	00:09
brockp_	Thanks for the delete, I assume that users the next time they pull will have an updated list in git branch -r	00:10
	← WALoeIII left	00:10
Eridius	they'll have to use git remote prune	00:13
	or git remote update	00:13
brockp_	Ok Thanks,	00:13
	→ eternaleye joined	00:17
	← brockp_ left	00:17
	← Lash\| left	00:18
	← yofel left	00:21
	→ igorgue joined	00:23
	→ mithro joined	00:27
	→ dejanpan1 joined	00:30
	← dejanpan left	00:30
	→ up_the_irons joined	00:33
	← drewr left	00:37
	← bronson left	00:39
	→ drewr joined	00:39
	← eternaleye left	00:40
	← statim_ left	00:42
	← Yuuhi left	00:44
	→ ben_h joined	00:44
	→ mpj joined	00:44
mpj	is it possible to commit a change to a specific revision?	00:46
MaddHatter	you can create a branch from any sha1 and commit to it	00:46
	← dejanpan1 left	00:47
mpj	i just noticed that i stomped another user's change via a bad merge a while back and wanted to restore it somehow - if i were using monotone i'd commit my change as a child of the original merge, then merge that with the tip fixing it as close to the error as possible	00:47
	this via git svn as well… i haven't pushed my changes back upstream yet, wanted to see if i could somehow clean up that 1 revision…	00:48
	j_engelh substracts 100 points from mpj's account	00:48
doener_	mpj: you didn't dcommit yet and have a merge commit in there?	00:48
mpj	that likely puts me at -100 then	00:48
	nope, no dcommit yet	00:48
j_engelh	50 for using svn, 50 for abusing git to use svn	00:48
mpj	i've rebase'd several times and in the process poorly merged an upstream change into my own	00:48
doener_	mpj: is that merge commit the first not-yet-dcommitted commit?	00:48
	→ kewark joined	00:48
doener_	mpj: ah, so you didn't really merge as in "git merge"?	00:49
	← schacon left	00:49
	→ dejanpan joined	00:49
mpj	doener_: yeah i might be mis-using terminology here, just meant that i rebase'd quite a while ago and seem to have been rather brutish	00:50
doener_	if you didn't dcommit yet, you can use "git rebase -i" to fix things up	00:50
	→ lea_w joined	00:50
mpj	though this was now several rebase's ago	00:50
doener_	as long as it's not dcommitted, that doesn't matter at all	00:50
	you can rebase as much as you want	00:50
	just don't touch any stuff you have already published	00:51
mpj	ahh good fun	00:51
	i'll stash my current changes first too	00:51
	the joys of poorly tracking other's svn repo	00:51
	cheers for the help	00:51
	→ ajonat joined	00:52
	→ eternaleye joined	00:54
mpj	doener_: oh did i mention i'd made several local commits since as well? :)	00:54
	its almost as if i'd like to reset a specific revision, but nothing afterwards	00:55
doener_	mpj: as I said, use "git rebase -i" ;-)	00:56
	mpj: basically "git rebase -i remotes/$svn_branch $local_branch" and follow the on-screen instructions	00:56
mpj	intense, i'll give it a go	00:57
	somehow this is what i thought i needed to do: http://www.venge.net/mtn-wiki/DaggyFixes	00:59
	having used monotone…	00:59
doener_	you _could_ do that, but merges and git-svn don't go well together (they're broken apart most of the time, due to the rebasing that is involved)	01:00
	rebase -i allows you to just create a new history where you never even made that mistake	01:01
mpj	magical	01:01
	foca_ → foca	01:02
	→ nuncanada joined	01:02
	← parasti left	01:06
	← ereslibre left	01:09
	→ dbtid joined	01:10
dbtid	hello git experts	01:10
	← dejanpan left	01:10
	→ dejanpan joined	01:11
dbtid	i tried to do a git pull, and got an error message concerning files needed merging.	01:11
	how am i supposed to do the merge?	01:11
	← stouset left	01:12
dbtid	git -s resolve filename?	01:12
	no, that complains and says "You are in the middle of a conflicted merge."	01:13
ciaranm	fix stuff, git add it	01:14
	also, you might want to git pull --rebase. or you might not.	01:14
dbtid	well, it was already part of the project.	01:15
	i did the git add.	01:15
	now the i've edited the file, i see all of the <<<<<< and >>>>> lines	01:15
	thanks	01:15
	i wasn't expecting it to modify the file that way	01:15
ciaranm	git add doesn't add files	01:15
dbtid	right	01:16
	it marks them as being to be committed	01:16
	← codeshot left	01:17
dbtid	ok, so i fixed it.	01:17
	→ bronson joined	01:17
	← joydivider left	01:18
	→ nbrosnahan joined	01:18
	← nbrosnahan left	01:18
	→ aiur joined	01:18
	→ nbrosnahan joined	01:18
	← mithro left	01:19
	← warthog9 left	01:20
	← tongueroo left	01:21
	← ebel left	01:22
	→ mithro joined	01:22
	→ VVSiz_ joined	01:24
	→ dejanpan1 joined	01:31
	← dejanpan left	01:31
	→ jtoy joined	01:33
jtoy	whole can I see a whole file from a certain revision?	01:33
	how I mean	01:33
doener_	git show rev:file	01:33
	→ gretch_ joined	01:34
RandalSchwartz	... 45. What is the language you want computer communicate with you?	01:37
	A. english, unlike this question. :)	01:37
Eridius	what's that from?	01:38
RandalSchwartz	2008 git survey	01:38
Eridius	ah	01:38
RandalSchwartz	there were a few... odd questions.	01:38
	Perhaps a native speaker could have sorted that out.	01:38
Eridius	I did the survey, don't remember that question	01:38
	→ LostInTheWorld joined	01:38
Eridius	heh, google for "2008 git survey", I get the github blog entry that links to it	01:39
RandalSchwartz	heh	01:39
Eridius	ah, I skipped the "Translating Git" section	01:39
jtoy	ah, i need to keep a cheratsheet or something	01:39
RandalSchwartz	Maybe it makes sense only if you're not a native speaker. :)	01:39
Eridius	heh	01:39
RandalSchwartz	"this question in code! no tell secret!"	01:39
	Eridius chuckles	01:39
RandalSchwartz	OK - that's bad. I'll stop. :)	01:40
jtoy	thanks doener_	01:40
	← lorandi left	01:41
	→ johnw joined	01:43
	← spearce left	01:46
	→ bogolisk joined	01:46
	← dejanpan1 left	01:48
	← nbrosnahan left	01:50
	→ dejanpan joined	01:51
	← johnw left	01:52
	→ johnw joined	01:53
	→ warthog9 joined	02:03
	gretch_ → gretchen	02:03
	← drewr left	02:07
	→ crab joined	02:07
	← cbreak left	02:08
	← eternaleye left	02:08
	→ eternaleye joined	02:08
	← bogolisk left	02:09
	← ijcd_ left	02:10
crab	do people really think it's a good idea to s/git-(.*)/git \1/g everything?	02:11
j_engelh	crab: no.	02:11
	for one because that substitution is broken.	02:11
	← dejanpan left	02:11
	→ dejanpan joined	02:11
crab	i mean, the commands got renamed, that's fine. but obfuscating the tests and everything because "people might grep for 'git-'" seems amazingly insane.	02:11
j_engelh	s/git-/git / is ok	02:12
crab	well, that'll only change one occurrence. yeah, i could live with that.	02:13
Eridius	one occurrence per line	02:17
	← rdavila_ left	02:18
	← sgrimm left	02:18
	← nuncanada left	02:21
	← kewark left	02:22
crab	i'm happy to get such useful answers to my question.	02:24
offby1	crab: it didn't bother me, and I'm a curmudgeon who hates change of any kind	02:30
	← eternaleye left	02:31
	→ imyousuf joined	02:31
	← dejanpan left	02:31
	→ dejanpan joined	02:31
imyousuf	robinr: hi, wanted to discuss about deploying jgit artifacts to public maven servers	02:32
	→ univate joined	02:32
	→ mib_y83nto joined	02:32
imyousuf	Do you have some time for it now?	02:32
	← johnw left	02:33
	→ eternaleye joined	02:34
	← mib_y83nto left	02:34
	→ johnw joined	02:36
	← clarby left	02:37
	← crab left	02:37
	← rtomayko left	02:38
	← harinath left	02:39
	← johnw left	02:39
	← rtomyako left	02:39
offby1	how can I do a "fetch" that includes the remote stash?	02:40
	→ johnw joined	02:40
offby1	"git stash list" on the upstream shows a particular stash, but even after "git fetch server:repo master stash", "git stash list" locally doesn't show it.	02:41
MaddHatter	I don't think stashes are mobile that way. You'd have to convert it to a throw-away branch.	02:42
offby1	huh	02:43
doener_	there's a stash ref for the top most entry, but the stash itself is a reflog	02:43
	→ perezd88 joined	02:43
MaddHatter	offby1: you'd have to create a branch, apply the stash to the branch, then transfer the branch (optionally do a mixed? soft? reset and re-stash locally) then delete the branch you used for the transfer	02:45
offby1	huh	02:45
MaddHatter	There's probably some pluming way to just transfer the objects you want, but I haven't figured that out yet	02:45
offby1	well I seem to have done it by hand.	02:45
	I wonder why it doesn't work	02:45
	oh well	02:45
	not important, really	02:45
MaddHatter	it doesn't seem like it'd be hard to implement, but what do I know	02:46
	← kukks left	02:47
	← bronson left	02:49
volante_	hey, just wondering how to reset a single file to its last committed state?	02:49
	← igorgue left	02:50
offby1	git checkout file	02:50
	took me a while to figure that out	02:50
jast	not exactly	02:50
	that returns the file to its indexed state	02:50
	the foolproof way is git checkout HEAD -- file	02:50
	→ dejanpan1 joined	02:51
	← dejanpan left	02:51
	← qrush left	02:52
jast	MaddHatter, offby1, stashes are meant to be really temporary so it seems like a good idea to keep them local to each repository. as such i suggest not to work in a way that requires you to move stashes around.	02:53
offby1	jast: I figured.	02:54
MaddHatter	I usually end up creating a throw-away branch instead of stashing	02:58
	or making a throw-away commit on whatever branch I'm on and --ammending it later	02:58
	if I later decide I didn't want the branch/commit it's easy enough to undo	02:59
	← lea_w left	03:02
	→ _kid_ joined	03:04
	→ Oompa joined	03:06
	→ bdiego joined	03:06
	← dejanpan1 left	03:08
	→ harinath joined	03:09
	→ dejanpan joined	03:10
	← mw left	03:11
	→ igorgue joined	03:11
	← perezd88 left	03:16
	← krawek left	03:16
	→ statim joined	03:17
	← bobesponja left	03:18
	→ thom_ joined	03:23
	→ schacon joined	03:25
	← loincloth left	03:26
	→ moelee joined	03:28
moelee	hi i am trying to do a git add, but when i check the changes, it does not add the files under the vendors/plugins/open_id_authentication folder	03:29
	any ideas why this is happening?	03:29
offby1	what did you type?	03:29
moelee	git add .	03:30
	only way i can commit those files is if i force add it using	03:30
	← dejanpan left	03:30
moelee	git add ./vendor/plugins/open_id_authentication/*	03:30
offby1	what makes you think files haven't been added?	03:30
	→ dejanpan joined	03:30
moelee	but when i commit again, it deletes those files	03:30
	i checked using git diff HEAD	03:31
	and the files don't show up	03:31
Eridius	maybe they're in the ignore list?	03:31
offby1	try "git add -f ."	03:31
moelee	hmm nope not in .gitignore	03:32
	← pi-meson left	03:33
moelee	HMM	03:33
	sorry	03:33
	hmm	03:33
	this commits logs too	03:33
offby1	"git add -f ." will, as far as I can tell, add everything.	03:34
	you can fiddle the index afterwards, though.	03:34
	moelee: you said "but when i commit again, it deletes those files". Did it _really_ delete files, or are you just speaking casually?	03:34
	(and if the latter, just what did you mean?)	03:34
moelee	hm let me rephrase	03:35
	so what when i do git add .	03:35
	and i check the head, it shows in the head to remove all the files under open_id_authentiation	03:35
	open_id_authentication*	03:35
offby1	what does "check the head" mean?	03:35
moelee	with - sign preceding all those files	03:36
	git diff HEAD	03:36
offby1	it'd be easiest if you simply pasted a transcript somewhere	03:36
	then there'd be no misunderstanding of what you did and what you saw.	03:36
	→ JimmyB__ joined	03:36
offby1	(we'd still need to establish what you _expected_ to see, but we might be able to guess at that)	03:36
moelee	basically, git thinks that my i have deleted all my files under open_id_authentication	03:37
	even though i haven't	03:38
	so when i do git add .	03:38
offby1	transcript transcript transcript	03:38
	when you say things like "git thinks", I don't know what you mean.	03:38
	But if you show me a transcript, I'll know.	03:38
moelee	http://gist.github.com/9604	03:39
	← jerbear left	03:39
moelee	so that is what i see when i run git diff HEAD	03:40
offby1	that's just a diff.	03:40
	I want to know what you typed.	03:40
	"transcript" means _both_ what you typed, _and_ what you saw.	03:41
	I notice you're using subprojects.	03:41
moelee	http://gist.github.com/9604	03:41
	→ priidu joined	03:41
offby1	thx	03:41
	(I didn't know you could update gists)	03:41
moelee	edit link	03:42
offby1	yeah.	03:42
moelee	must be a new feature	03:42
	← bdiego left	03:42
offby1	anyway: after all the hoops I've made you jump through ... I'm going to say "I can't tell what's going on" ;-(	03:42
moelee	so when i do git commit, followed by git push	03:42
offby1	Apparently you're using subprojects, and I don't understand those.	03:42
moelee	i fear it will delete open_id_authentication files	03:42
	lol ok	03:42
	well, thx anyways	03:43
offby1	I wonder if you've in effect told it to overwrite the open_id_authentication directory with a subproject.	03:43
moelee	hmm	03:44
	← david_koontz left	03:44
moelee	ok i notice a .git file in that directory	03:44
	argh that might be it	03:44
	...	03:44
offby1	I hope that's an encouraging "hmm". I feel bad having made you jump through all those "transcript" hoops, and then not being able to help.	03:44
moelee	note to self, don't check out code into a git project	03:45
offby1	:)	03:45
moelee	sweet	03:46
	that was the problem	03:46
	good call offby1	03:46
offby1	hope I helped some :-\	03:46
moelee	thanks	03:46
offby1	never woulda guessed there were subprojects if you hadn't pasted a transcript :)	03:46
moelee	right on	03:46
	← Zarin left	03:47
	→ Zarin joined	03:50
	← patmaddox left	03:50
	← dejanpan left	03:50
	→ dejanpan joined	03:51
	→ joydivider joined	03:51
	→ xhanjian joined	03:51
	← aiur left	03:53
	← BarryCarlyon left	03:56
	→ BarryCarlyon joined	03:57
	← threeve left	04:01
	→ GregHouston joined	04:03
	← JimmyB__ left	04:07
GregHouston	Hi, I am a new Git user, and am using Git Gui. I also have the command line open. I added two files that I should not have. Reset removed one of them, but not the other. It did not remove the one that was not originally in the repo. Right now it is sitting in the unstaged changes box of Git Gui. How can I remove it?	04:08
ptman	GregHouston: try git rm --cached file	04:09
Eridius	unstaged? that's what you want, isn't it?	04:09
thom_	so can git track files that diverge off a common parent and other files that merge other multiple parents?	04:09
	→ perezd88 joined	04:09
	→ bronson joined	04:10
	← dejanpan left	04:10
	→ dejanpan joined	04:10
GregHouston	Well, with something in the unstaged box fetch doesn't work. Going to try ptman's suggestion.	04:10
	→ kopa joined	04:10
	→ alenoosh joined	04:12
	← univate left	04:12
	← madewokherd left	04:12
	← perezd88 left	04:16
offby1	thom_: I didn't understand your question	04:16
	← k776 left	04:16
	→ huperniketes joined	04:21
GregHouston	When I try $ git rm --cached .project It says: fatal: pathspec '.project' did not match any files. I am in the same directory as the file. May it be the filename starting with a period that is making it not recognize the file?	04:22
MaddHatter	that means that file is not part of the index	04:23
	← schacon left	04:23
MaddHatter	you're trying to remove from the index a file that's not in it. The --cached option says rm "operate on the index instead of the working tree"	04:24
GregHouston	I get the same error without the --cached option.	04:25
MaddHatter	that doesn't seem surprising... (again, it's not in the index)	04:26
	what are you wanting git to do?	04:26
thom_	if I split two functions in one file into two new files will git track the original file as the parent?	04:26
GregHouston	Well I added the file, but have not committed it. I want to unadd it.	04:26
thom_	or can you make it do that?	04:26
MaddHatter	GregHouston: what does git status say?	04:26
	← tjafk left	04:27
	→ tjafk1 joined	04:27
GregHouston	Untracked files: .project So I guess it isn't added. Maybe I just need to add it to .gitignore.	04:28
MaddHatter	you can leave it untracked of that doesn't bother you, or you can tell git to specifically ignore it	04:28
GregHouston	Cool. Thanks for the help. Still in the confused-with-the-new-thing stage. There are a couple more steps to committing than with SVN that I am not use to yet.	04:30
MaddHatter	People don't seem used to the idea of the index, but it's quite useful once you get the hang of it (I think)	04:30
offby1	thom_: I think it will. Try It And See™.	04:31
	→ loincloth joined	04:34
thom_	LOL not proficient enough to pull off that test yet. just curious	04:38
	← dejanpan left	04:38
	→ sgrimm joined	04:39
offby1	you should experiment! It's good for what ails you.	04:39
	Make a toy repo (elapsed time: 0.2 seconds), create your file, make your edits and commits.	04:40
thom_	been doing that, still figuring out how to not munge repos between machines	04:41
	← ferdy left	04:48
	→ perezd88 joined	04:49
	→ Ratler_ joined	04:50
	→ meyering joined	04:51
	← jtoy left	04:52
	← loincloth left	04:56
	→ dummeto joined	04:57
dummeto	hi i need copy all log commit for send in email how to?	04:57
machrider	git log?	04:58
	← rmh3093 left	04:59
dummeto	:) i used git log -p	04:59
	sorry	04:59
	thanks .	05:00
	→ d3ntaku joined	05:00
	→ rmh3093 joined	05:00
Eridius	dummeto: I think you need a book on english grammar	05:01
dummeto	yes	05:02
	sorry i living in ARG	05:02
	thom_ guesses Eridius is American	05:02
	← GregHouston left	05:02
dummeto	english grammar :D yer now	05:02
	← Ratler_ left	05:03
	→ SRabbelier joined	05:03
	← dummeto left	05:03
	→ Ratler joined	05:05
	← kopa left	05:05
	← igorgue left	05:06
	← SRabbelier left	05:07
	→ SRabbelier1 joined	05:07
	SRabbelier1 → SRabbelier	05:08
	→ jitu3485 joined	05:12
	→ majoh_ joined	05:13
	→ sverrej joined	05:21
	→ bentob0x joined	05:23
	→ dejanpan joined	05:24
	← jackdempsey left	05:25
	→ mczepiel joined	05:26
	← majoh_ left	05:27
	→ majoh_ joined	05:28
	← Ademan left	05:28
	→ Ademan joined	05:29
	← mpj left	05:31
	← foca left	05:33
	→ khaled joined	05:39
khaled	I've created a shared reposity at a server. Then, I cloned, modified, commited, and pushed the edits from a different client. git status on the server shows that the files are modified, but when I open the files, they aren't modified. Any thoughts on what I am missing?	05:40
RandalSchwartz	faq bare	05:41
	!faq bare	05:41
Mikachu	khaled: see faq-bare	05:41
RandalSchwartz	faq-bare	05:41
Mikachu	hm	05:41
RandalSchwartz	darn it. what's the magic	05:41
Mikachu	$gitfaq list	05:41
Gitbot	Mikachu: Valid topics: branch, fetch-vs-pull, git-name, non-bare, non-ff, pager-color, pager-default, push-f, ssh-config	05:41
Mikachu	khaled: see non-bare	05:41
RandalSchwartz	faq non-bare	05:42
Gitbot	RandalSchwartz: You shouldn't push to a non-bare repository. See http://git.or.cz/gitwiki/GitFaq#non-bare	05:42
Mikachu	aha	05:42
RandalSchwartz	there we go	05:42
Mikachu	:)	05:42
	it would have been faster to open the browser	05:42
jitu3485	hi wher can i found the documentation to setup up a git repository ,so that all my colleagues can access that repo?	05:42
RandalSchwartz	"no wait... it's THIS button!"	05:42
	jitu3485 - have you read the basic git tutorial?	05:42
	there are discussions at the end of that about how to set up various repos	05:43
jitu3485	i am going throu it	05:43
RandalSchwartz	"access" can mean many things	05:43
	→ SjB joined	05:46
khaled	Thanks guys .. git reset --hard works	05:46
	← SjB left	05:48
	→ rtomayko joined	05:49
	→ tritz joined	05:50
	→ kopa joined	05:50
	← jitu3485 left	05:50
RandalSchwartz	for some meaning of "works"	05:52
	you should move to a non-bare repo to solve it for good	05:52
	or understand how you should never push to a possibly checked out branch	05:52
	→ hardkrash joined	05:53
hardkrash	evening folks, i got a question that is eluding me, i pushed the svn mirror to a bare remote git repo, initalized the svn repo on the remote server	05:56
	now i svn fetch on the remote server, but my local copy doesn't pull the branches from the mirror	05:57
	anyideas?	05:57
	RandalSchwartz averts his eyes at the mention of git-svn	05:57
	hardkrash chuckles at the psudo responce	06:01
hardkrash	this is git not tracking the remote refs	06:01
	the the --mirror pushed up	06:01
	i don't know how to tell git to pull them all back	06:02
	← kyofel left	06:05
	← softdrink left	06:06
	← Ratler left	06:09
	→ lamont joined	06:11
	→ jitu3485 joined	06:15
	→ deavid joined	06:16
	← kopa left	06:18
jitu3485	i am going through http://www.kernel.org/pub/software/scm/git/docs/gittutorial.html, but having a error after git commit	06:19
	fatal: no commit message? aborting commit.	06:19
	← johnw left	06:20
	← robinr left	06:20
tritz	jitu3485: try git commit -m "message"	06:22
Bombe	jitu3485, that error message is pretty self-explanatory, don't you think?	06:22
hardkrash	have $EDITOR set?	06:22
khaled	I have created a bare shared repository, modified files from other clients, pushed the modifications. When I checkout the master branch in another machine, I see the modifications. However, when I try to see the modifications on the server itself (it doesn't show up)	06:22
hardkrash	in your enviroment?	06:22
jitu3485	ok	06:22
khaled	Can someone pls help me with this issue?	06:22
tritz	jitu3485: i had this problem once, the $EDITOR always had an return code $? = 1, therefore git thought I want do abort commit	06:23
Mikachu	khaled: didn't you already ask, get an answer, and say "thanks, it worked"?	06:23
khaled	Yup when I had a non-bare folder. Now I created a bare folder .. things are still working on the clients, but the server itself doesn't show the modifications.	06:24
Mikachu	it's bare, it doesn't show any files at all...	06:24
	→ univate joined	06:24
hardkrash	bare repositorities have no files	06:24
	→ Beket joined	06:24
	→ schacon joined	06:25
khaled	Well, I mean .. what do I need to do so that I can see the modified files in the repository on the server?	06:25
	← schacon left	06:25
hardkrash	clone it?	06:25
	on the server	06:25
jitu3485	tritz, now done with git commit -m "message", but in cvs we can message	06:25
	'skip'	06:25
hardkrash	or have it not be bare	06:25
khaled	Gotcha .. so, the bare one acts as a history DB?	06:26
tritz	jitu3485: it is good practice to specify a message. else git log would have nothing to show.	06:26
jitu3485	ok	06:27
hardkrash	how do i pull all branches from my git remote mirror	06:29
	something like a git fetch --all	06:29
	← khaled left	06:32
	← Beket left	06:32
	→ BabelO joined	06:35
rtomayko	hardkrash: git should pull all of the branches on a remote automatically	06:37
	→ charon joined	06:37
hardkrash	it's not pulling the ones refs/remotes/tags/	06:38
	it not pulling any refs/remotes/*	06:38
thiago_home	that's normal	06:38
rtomayko	ahh, I see	06:38
thiago_home	it pulls refs/heads and refs/tags	06:39
Mikachu	refs/heads/* on the remote "foo" goes into your refs/remotes/foo/*	06:39
	← bronson_ left	06:39
rtomayko	hardkrash: you probably want rsync :)	06:39
	← bronson left	06:39
hardkrash	i want my remote server to fetch from my svn server for me	06:39
	so i can pull down from git	06:40
thiago_home	not a really good idea	06:40
	if you use git-svn, you shouldn't be using pull/push	06:40
hardkrash	reason?	06:40
	→ bronson joined	06:40
thiago_home	it wasn't written with that in mind	06:41
	← mithro left	06:41
thiago_home	it keeps also some local configuration files that aren't transmitted with pull/push	06:41
	you should use git-svn locally only	06:41
hardkrash	i didn't think that push and pull on the git portion would affect the svn portion	06:41
thiago_home	it wouldn't	06:42
hardkrash	that unfortunate, i wanted a server mirror.....	06:42
charon	you should be fine as long as you only 'pull --rebase'	06:42
thiago_home	pulling from git works fine, pushing back to it too	06:42
	the problem is committing back to SVN	06:42
Mikachu	if you're not going to commit to svn it should be fine	06:42
	→ drizzd joined	06:43
hardkrash	fortunatly i don't have write access	06:43
	← bronson left	06:43
hardkrash	:-)	06:43
	← imyousuf left	06:43
thiago_home	then it's fine	06:43
	you want to change the way you use git-svn, though	06:43
hardkrash	how do i pull all refs/remotes to my local repo?	06:43
thiago_home	it's storing in refs/remotes	06:43
	you should make git-svn store in local branches	06:44
hardkrash	on the server?	06:44
thiago_home	yes	06:44
	since that is a bare repository (it is, right?), that wouldn't be a problem	06:44
hardkrash	ahh i see, is there a convent way to switch it	06:44
	it is bate	06:44
	bare	06:44
thiago_home	edit the config file	06:45
	git-svn keeps its syncing instructions there	06:45
	→ namenlos joined	06:46
	→ Ratler joined	06:47
hardkrash	branches = scottie/branches/:refs/remotes/	06:51
	tags = scottie/releases/:refs/remotes/tags/	06:51
	become:	06:51
	→ girishr joined	06:51
hardkrash	branches = scottie/branches/:refs/heads/	06:51
	tags = scottie/releases/:refs/heads/tags/	06:51
	?	06:51
thiago_home	not refs/heads/tags, but refs/tags	06:51
	→ kopa joined	06:51
hardkrash	ok	06:51
thiago_home	then again, maybe you were right	06:51
	SVN tags can be moved, so they should be treated as branches instead	06:52
hardkrash	now the config can be edited, how do i move the data	06:52
	i rather treat the tags as tags	06:52
thiago_home	git branch -m	06:52
	hardkrash: SVN tags aren't Git tags	06:52
	hardkrash: SVN tags are branches	06:52
hardkrash	yes	06:52
	→ thom_logn joined	06:53
	← thom_logn left	06:53
hardkrash	but that is errornous imho, but i couls see hos that would be an issue	06:53
	could see how	06:53
	the error is on SVN's side	06:53
	→ psankar joined	06:54
psankar	Are there any tools to show statistics from git ?	06:54
hardkrash	well it is time to get some sleep	06:55
	thanks for the help, i'll tey and finish it tomorrow	06:55
thiago_home	psankar: lots of them	06:55
	→ mithro joined	06:55
	← hardkrash left	06:56
	← tritz left	06:56
	→ bronson_ joined	06:56
	→ adamm joined	06:56
psankar	thiago_home, When I googled it , I saw only one soc project whose binaries i was unable to find... you have any urls ?	06:56
	← bentob0x left	06:56
thiago_home	no	06:57
	first you have to say what kind of statistics you're looking for	06:57
	← adamm left	06:57
	→ adamm joined	06:57
	→ fujin joined	06:57
	← Ademan left	06:58
psankar	thiago_home, something like , top contributors, lines changed etc.	06:58
	→ takayaW joined	06:59
	→ sf_ joined	07:00
adamm	When I merge from another branch, I seem to pull in the entire history. But Linus' kernel branch, all that I see for merges is "Merge branch 'master' of git://git./linux/kernel".	07:00
	How can I merge without pulling in the history? Or it is a gitweb configuration thing?	07:00
	← LostInTheWorld left	07:01
thiago_home	psankar: there's a command to create a changelog-like log from the commit history	07:01
	I can't find it anymore, but it's part of the standard git tools	07:01
	→ imyousuf joined	07:02
thiago_home	probably git-log itself	07:02
psankar	thiago_home, git shorlotg ?	07:02
	← perezd88 left	07:02
thiago_home	psankar: yep	07:02
psankar	thiago_home, i wanted something graphical with charts and bars etc.	07:02
thiago_home	psankar: write it :-)	07:02
	adamm: you're seeing merges pulling in the history	07:02
	adamm: that's what Linus is using	07:03
psankar	thiago_home, I thought it could be already available .. :) Shall search more and if I dont find any i will write	07:03
	← psankar left	07:03
	→ psankar joined	07:03
adamm	thiago_home: see http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=summary vs http://git.debian.org/?p=collab-maint/rails.git;a=summary	07:04
thiago_home	adamm: what do you want me to see there?	07:06
	both have merges at the top	07:06
	→ bronson joined	07:07
adamm	Linus' merge is just "Merge branch" while the rails one has the entire upstream rails developer history in the merge. Is it possible to just have a note about a merge but not to pull in the entire patch history from another branch?	07:07
thiago_home	adamm: Linus's merge is the entire history	07:08
	→ majoh__ joined	07:08
	← majoh__ left	07:09
	→ \|mako\| joined	07:09
charon	adamm: it's possible, but usually you don't want to, since you'd lose lots of useful information _and_ the option to later merge again (easily).	07:10
	→ LostInTheWorld joined	07:10
thiago_home	adamm: like I said, I'm seeing the exact same thing in both gitwebs	07:10
	← Flathead left	07:10
	→ ph^ joined	07:11
adamm	thiago_home: ok... I understand.	07:11
	charon: Do you mean the only way would be to squash the merge into a giant patch then apply it?	07:12
	→ morphir joined	07:12
charon	yes. don't do that. it's almost always a terrible idea	07:14
	→ johnjosephbachir joined	07:14
adamm	yes, I can see why!	07:15
	→ bartocc joined	07:17
adamm	In that case, is there some way to configure gitweb to only display the stuff on the branch, and ignore commits from other branches? That is, only follow the "left side" of the merges?	07:17
sf_	is there a gitweb around that has categorization of projects ?	07:22
	something like what can be seen in cgit	07:22
	one title "Foo" with projects in it, and another "Bar" with stuff unrelated to Foo	07:22
	→ thiago_h1me joined	07:23
	← thiago_home left	07:23
	→ GarethAdams joined	07:23
adamm	?	07:23
	← mczepiel left	07:24
	→ ehc joined	07:25
	thiago_h1me → thiago_home	07:25
	← ehc left	07:27
	→ ehc joined	07:28
adamm	AHA!! found it! git log --first-parent	07:30
	← johnjosephbachir left	07:32
	← sverrej left	07:33
	← joydivider left	07:34
	→ Ademan joined	07:35
	← davidfetter left	07:40
	→ elmex joined	07:45
	← thiago_home left	07:49
	← sf_ left	07:56
	← ehc left	07:59
	→ lanwin joined	08:01
lanwin	hi	08:01
	→ sverrej joined	08:01
	Ademan → foobarbaz	08:01
lanwin	what is the default way to compleatly replace the master brunch with another brunch insted of merging it?	08:01
	foobarbaz → __foobarbazspam	08:01
	← __foobarbazspam left	08:02
Eridius	git reset --hard otherbranch	08:02
Mikachu	otherbrunch*	08:02
ciaran	mm brunch	08:03
	← hurikhan\|Work left	08:03
	← adamm left	08:05
	→ hurikhan\|Work joined	08:06
	→ jkr joined	08:10
jkr	Ahoy	08:10
	→ Ademan joined	08:11
jkr	I'm trying to git-am some mails, but I'm getting the message "Patch is empty. Was it split wrong?". What does this mean?	08:11
	And when I'm trying again I'm getting the message "previous rebase directory /home/jkramer/source/shell-fm/.git/rebase-apply still exists but mbox given." until I remove that directory (rebase-apply). Isn't there a nicer way to clean up the mess of the previous command?	08:13
	Alright, I solved the first problem. One of the mails in the mbox didn't contain a patch.	08:13
	← CodeOfficer left	08:13
	← harinath left	08:23
	→ harinath joined	08:25
	→ ph^_ joined	08:25
	→ sciz joined	08:28
	← ph^ left	08:29
Mikachu	jkr: maybe git am --skip or so	08:29
	jkr: try --continue first	08:29
jogla	I use git-svn, can i diff the changes in the svn repository against my head before pulling them?	08:29
sciz	I'm pretty new to git, so this might be trivial. Anyhow, when I switch between branches with different folder hierarchies git leaves empty folders that doesn't exist on the current branch. Any way of avoiding this?	08:33
charon	jogla: you can use 'git svn fetch' to get the changes, then inspect them with gitk or git-log as usual	08:35
	→ koke joined	08:36
jogla	do i have to make a branch first, so that they don't get in my master?	08:37
	← \|mako\| left	08:40
charon	fetch only touches the git-svn/trunk/whatever remotes, your local branches are only updated with 'git svn rebase'	08:40
jogla	Ahh, cool	08:42
	Thank you	08:42
	← ajonat left	08:43
	← koke left	08:43
	→ fdr- joined	08:45
	← tommorris left	08:45
	← imyousuf left	08:46
	← fdr- left	08:46
	→ botanicus joined	08:46
	ph^_ → ph^	08:48
jogla	Do i always have to commit all my changes, before i can rebase?	08:49
	→ soul9 joined	08:49
soul9	hi all	08:49
Mikachu	jogla: yes, you can uncommit them after	08:51
jogla	Hehe, ok	08:51
	→ clarby joined	08:51
jogla	git reset --soft HEAD^1?	08:52
Mikachu	if you want them to still be in the index	08:52
	→ nud joined	08:52
jogla	and git reset --mixed HEAD^1 cleans also the index?	08:52
	But they are still in my working tree?	08:52
Mikachu	yes	08:53
	and --mixed is the default, as is 1	08:53
	so git reset HEAD^ is enough	08:53
jogla	Ahh, ok	08:53
	And if i did not commited them yet, but want to get rid of them?	08:53
Mikachu	--hard, but be careful with that	08:53
jogla	Ahh, ok	08:54
	→ imyousuf joined	08:54
jogla	But when they aren't committed i need to commit them before i can do --heard, right?	08:54
Mikachu	no	08:54
	or um	08:54
	if you commit them, then --hard won't remove them	08:54
	→ GNUix joined	08:55
jogla	Ahh, ok, seems to have worked :)	08:56
	Thank you	08:56
	← Fullmoon left	08:57
	→ Fullmoon joined	09:00
resmo	i have 2 workstations and wondering remote origin was already added in config on one of them, why?	09:02
	or why not on the second?	09:02
	→ shodges joined	09:04
charon	is there a way to get some delta statistics on a pack? for example, how many objects were deltified, average chain length, max chain length, etc	09:04
soul9	I am working on a project using git	09:06
	and i'd like to have my own git tree, to be able to work and pull my changes from my own server	09:06
	→ takayaW_ joined	09:06
soul9	but I want the changes to show up like a sinle commit on the project repo	09:06
	how is this possible?	09:07
resmo	single commit?	09:07
soul9	yeah, like i work on a project	09:07
	and commit as i go	09:07
	and when the project is done, i want to commit it to the main repo	09:07
	and i want that commit to be one single commit	09:08
resmo	so you like to forget the hole history of the commits?	09:08
	s/hole/whole	09:08
soul9	well, my own local repo can remember, but yeah	09:09
	← BarryCarlyon left	09:10
resmo	so i would say i can import a tarball into your repo and then commit each time	09:10
soul9	hm	09:10
	so i work and commit in my own repo	09:10
mugwump	see the git-import-tars contrib script	09:11
	it does just that, just very fast :)	09:11
soul9	then export a trarball, unpack it on top of the original git tree and that's it?	09:11
	s/trarball/tarball/	09:11
mugwump	soul9: you'll need to update the index before you commit	09:12
	ie, 'git rm \' first, then untar, then 'git-add \'	09:12
	→ Sho_ joined	09:12
	← bronson_ left	09:12
soul9	what is \*?	09:12
mugwump	well, git will expand globs like the shell does	09:13
soul9	but why do i need to do this?	09:13
	← takayaW left	09:13
soul9	i get the tarball, extract it on top of the original repo	09:13
mugwump	because that's the way the software works. :)	09:13
soul9	that's not enough?	09:13
	heh	09:13
mugwump	no, you need to commit it	09:13
soul9	yeah	09:13
	so i do	09:14
mugwump	and to commit it you have to add it to the index	09:14
	→ chris2 joined	09:14
soul9	i don't remember, but something like git export ...	09:14
mugwump	(aka "staging area")	09:14
soul9	i get a tar.gz	09:14
	unpack it on top of the original repo	09:14
	and then git commit	09:14
	man git-archive	09:14
Gitbot	soul9: the git-archive manpage can be found at http://git.or.cz/man/git-archive	09:14
soul9	oops ☺	09:15
mugwump	ok step back... what are you trying to achieve?	09:15
soul9	☺	09:15
	so:	09:15
	i am working on an opensource project	09:15
	that uses git	09:15
mugwump	actually, I just backlogged a bit	09:16
soul9	i want to have my own git tree, that I can commit to and push to my server, so i can work from multiple locations	09:16
	but have the commit to the original tree look like one big commit	09:16
mugwump	The command to take your own work, and 'squash' it, possibly amending it along the way, is 'git rebase -i'	09:16
soul9	what's 'squash'?	09:17
mugwump	or, there are more sophisticated programs like stgit, topgit, etc	09:17
	squash means combine multiple commits into a single one	09:17
soul9	tthat's perfect	09:17
	so i work, do my commits	09:17
	then git rebase -i?	09:18
	and it'll combine the commits into one, and maybe even push it?	09:18
mugwump	well, take a look at the man page for the details, but normally it would be something like 'git rebase -i origin/master'	09:18
	assuming 'origin/master' is where you want to put your commits (not pushing yet)	09:19
	then it pops open an editor, and you cut and paste lines until it makes sense to the people who are going to be reviewing/accepting the changes	09:19
	(each "line" represents one of the changes you made)	09:19
	but instructions to how that works appear as you go	09:20
soul9	ok	09:20
	so i have to branch first	09:20
	?	09:20
mugwump	then you can check it all looks right with 'git-log -p', push it to your github fork (or whatever) with git-push, or submit to a ML with 'git-format-patch'	09:20
	sure, make a branch, go for it. in fact I think you probably want to find the right section of the user manual to help you out along the way first :)	09:21
	→ trochala joined	09:21
mugwump	eg http://www.kernel.org/pub/software/scm/git/docs/user-manual.html	09:21
soul9	o	09:22
	k	09:22
reuss	if i do git reset origin/HEAD, it shouldn't EVER touch my local working tree, right? ...	09:24
Mikachu	as long as you don't say --hard, reset never will	09:24
reuss	i think i'm experiencing otherwise - but it's run from one of my scripts, so it could be i'm messing up somehow there...	09:25
mugwump	it touches the index of course	09:25
	use --soft if you don't want that	09:25
reuss	but it looks like it's messing with my working dir, because when i do git status, files are listed under modified ...	09:26
Mikachu	stop and think about what you're doing :)	09:26
	if you change the index and not the work tree, of course you will end up with modifications	09:27
	if it did change the worktree, status would say everything was clean	09:27
	→ imyousu1 joined	09:27
	← morphir left	09:27
reuss	this is confusing - i'll do some more tests and see what i can find out	09:29
	→ ekneuss joined	09:30
	→ koke joined	09:32
	→ ferdy joined	09:32
	→ _slacker_ joined	09:32
	← Fullmoon left	09:33
	→ ereslibre joined	09:37
	← imyousuf left	09:38
	← imyousu1 left	09:39
	← luks left	09:42
	→ cilly joined	09:45
	VVSiz_ → VVSiz	09:45
	→ kbingham joined	09:47
	→ tritz joined	09:48
	→ aack joined	09:50
	← thom_ left	09:55
	→ bcarlyon\|laptop joined	09:58
	bcarlyon\|laptop → WhereIsBarry	09:59
	WhereIsBarry → BarryCarlyon	09:59
	→ JakubNarebski joined	10:06
	→ mnemoc joined	10:06
	JakubNarebski is jnareb is ShadeHawk is Jakub Narebski	10:06
	← xhanjian left	10:07
JakubNarebski	anyone here can tell me how a past discussion on Git vs. Bazaar design of repository/branches concluded?	10:07
mnemoc	hi, how can I output exactly what git filter-branch --msg-filter's filter will filter? (for testing the patterns)	10:07
	sciz → sciz\|away	10:08
mnemoc	I can't find the way to make `git log` less human friendly :(	10:10
Rhonda	mnemoc: Run it without and with, store both outputs, diff them?	10:10
JakubNarebski	git-rev-list? git log --pretty=format:<format string>?	10:11
	I don't think that git-filter-branch has -n / --no-commit option, but IIRC it leaves original branch(es) in refs/original/ namespace or something like that	10:12
mnemoc	I really fear to run it :( I just change to change title lines "style" from using [foo] blah blah to foo: blah blah	10:14
JakubNarebski	mnemoc: see man git-filter-branch	10:15
Gitbot	mnemoc: the git-filter-branch manpage can be found at http://git.or.cz/man/git-filter-branch	10:15
JakubNarebski	Always verify that the rewritten version is correct: The original refs, if different from the rewritten ones, will be stored in the `namespace refs/original/`	10:16
	and you can always clone repository first	10:16
	→ tzafrir_laptop joined	10:16
mnemoc	I'm on it... but it seems it will filter the whole history, and I just want what is between the current HEAD and origin/master	10:16
	good idea, a test clone	10:17
tzafrir_laptop	my repos at repo.or.cz are suddenly empty. Those two repos mirrored, and the original is on-line	10:17
JakubNarebski	why not use "git filter-branch [...] HEAD..origin/master"?	10:17
tzafrir_laptop	anybody else noticed a similar problem?	10:17
JakubNarebski	tzafrir_laptop: hav you contacted pasky (Petr Baudis)?	10:18
mnemoc	JakubNarebski: doh, I was using the oposite range	10:18
JakubNarebski	mnemoc: mnemonic is I think rather easy: a..b is (for linear history) from a to b, in the increasing order	10:19
mnemoc	HEAD..origin/master is decreasing order in my mental model :(	10:20
JakubNarebski	tzafrir_laptop: nope, I have no such problem with my single repo.or.cz repository...	10:20
Tv	a..b is always b ^a to me	10:20
tzafrir_laptop	JakubNarebski, thanks	10:20
Tv	set operations	10:20
tzafrir_laptop	will ask him	10:21
JakubNarebski	like a..b is <from> <to> like in "cp a b", or "diff a b"	10:21
pasky	so which repos?	10:23
	JakubNarebski nudges tzafrir_laptop: pasky here...	10:23
mnemoc	JakubNarebski: sure, I just have to redefine my mental model where origin/master is below (and comes first) HEAD	10:23
tzafrir_laptop	http://repo.or.cz/w/asterisk-bristuff.git http://repo.or.cz/w/libpri-bristuff.git	10:24
	pasky, just double-checked I can clone the original of libpri	10:24
Tv	mnemoc: well isn't that just gitk.. time flows up	10:24
mnemoc	JakubNarebski: I just see the tree starting from root and not from the leaves :-\ and that is a mistake	10:25
JakubNarebski	mnemoc: a..b agrees with convebtion of ASCII-art of history, like ...<---a<---.<---.<---b, where '<---' denotes parent links, and "time" flows from left to right like in all timelines (natural for all writing left-to-right)	10:25
	convention	10:25
mnemoc	Found nothing to rewrite	10:28
	mnemoc cries	10:28
mnemoc	does title lines belong to --msg-filter ?	10:28
tzafrir_laptop	JakubNarebski, the unicode bidirectionality algorithm includes mirroring of some character	10:29
	characters, that is	10:29
	→ peritus_ joined	10:30
pasky	tzafrir_laptop: looks like UPC nameservers are b0rked	10:30
tzafrir_laptop	hence '>' is not 'left to right', but rather 'closing'	10:30
pasky	i've added nameserver 127.0.0.1 to /etc/resolv.conf, dunno why wasn't it there before ;)	10:30
	i don't understand how the refs disappeared, though	10:30
	next update round should re-mirror it	10:31
tzafrir_laptop	pasky, thanks	10:31
	JakubNarebski, using the same '>' characters you used, but different letters:	10:31
	← jitu3485 left	10:32
tzafrir_laptop	..<---א<---.<---.<---ב	10:32
	(that's a whole different type of mirroring. I'm done with my off-topic)	10:33
	← cilly left	10:34
	→ cilly joined	10:34
JakubNarebski	tzafrir_laptop: I don't quite understand. In the languages where you write left-to-right, you have what you wrote earlier on the left, and what you wrote later on the right	10:34
	so ...1<---2<---3<---4<---... ordering mimics that	10:34
tzafrir_laptop	JakubNarebski, do you use a graphic IRC client?	10:35
JakubNarebski	probably would be different if Git was written by someone who use for example hebrew, sanscrit, or one of Arabic languages	10:35
	no, I don't	10:35
	rcirc, with iso-8859-2 encoding, due to legacy configuration of my GNU emacs 21	10:36
tzafrir_laptop	ok. copy that line to any text editor that supports bidirectional rendering	10:36
	e.g. gedit	10:36
	or run your session in mlterm or xiterm :-)	10:36
mnemoc	git filter-branch --msg-filter 'tee -a filter.out' HEAD..origin/master <---- where is filter.out supposed to be created? I can't find it	10:37
Mikachu	mnemoc: give an absolute name then	10:38
JakubNarebski	I'm not that interested in this to try to fight sometimes quirku Emacs <-> X clipboard interface	10:38
mnemoc	Mikachu: good idea =)	10:38
	Mikachu: ehm, no file.... wth	10:39
	→ cbreak joined	10:39
mnemoc	.oO	10:39
	origin/master..HEAD worked :D	10:40
	117 commits rewritten, life is good, life is good	10:41
	thanks all for your support/help :D	10:41
	← nud left	10:45
	→ Lash\| joined	10:47
	→ cypher23 joined	10:48
bremner	JakubNarebski: if you run KDE, the klipper thingy is good at making emacs and x play nice	10:55
	← mithro left	10:56
	→ mediogre joined	10:56
JakubNarebski	first, it might matter that it is GNU Emacs 21, not GNU Emacs 22 (old distribution)	10:56
	second, I run Windows Maker	10:56
bremner	JakubNarebski: as someone who teaches logic, I'm happy to say what I said is none-the-less true :-)	10:57
JakubNarebski	third, I have long-lasting (suspend/hibernate) Emacs session, and I think it is something I do in Emacs that _sometimes_ screws up Emacs <-> X clipboard integration	10:57
	→ jmd joined	11:00
	→ warlock_mza joined	11:01
	← jmd left	11:10
	→ advany joined	11:11
	→ jmd joined	11:12
jmd	How do I set the default branch?	11:12
	← Lemurnomicon left	11:12
Eridius	define "default"	11:12
	→ pygi joined	11:13
jmd	The one upon which future git commands for this repo will operate (unless told otherwise).	11:13
Eridius	you mean, uh, the current branch?	11:13
	git checkout branchname	11:13
JakubNarebski	you mean current or checked out branch?	11:14
	git checkout <branchname>	11:14
	← tzafrir_laptop left	11:14
JakubNarebski	or if you want to create it, "git checkout -b <branchname>"	11:14
Eridius	the only "default branch"-ness in git is when you init a new repo, it starts on the branch "master"	11:14
	→ simmel joined	11:14
	← ferdy left	11:14
jmd	So I can't arrange things so that "git checkout" will grab another branch?	11:15
JakubNarebski	I don't understand you, jmd	11:16
	→ parasti joined	11:16
	→ jkp joined	11:16
JakubNarebski	"git checkout <branchname>" changes which branch is current	11:16
jkp	hi, can anyone tell me how to use git-svn to clone just a single branch of a project?	11:16
	URL for branch is https://swig.svn.sourceforge.net/svnroot/swig/branches/gsoc2008-cherylfoil	11:16
jmd	JakubNarebski, Is it really that complicated? How do I arrange my repo, so that "git checkout" will checkout a branch other than "master" ?	11:16
tritz	jmd: do you mean the branch you get on 'git clone'?	11:17
	→ serp joined	11:17
Mikachu	jkp: git svn clone theurl	11:17
JakubNarebski	jmd: "git checkout" (without further arguments) IIRC checks out state of index	11:17
Eridius	jmd: you don't call "git checkout", you call "git checkout branchname". Try reading the documentation	11:17
	← girishr left	11:17
jmd	tritz, I thought that "git clone" always cloned the entire repo? At least that's what I was told here last week.	11:18
	→ girishr joined	11:18
	→ dmq joined	11:18
jkp	Mikachu: it tells me thats not an SVN repo	11:18
tritz	jmd: yeah, but it the checks out a branch	11:18
jkp	i think its looking for the standard layout of an SVN repo	11:18
Eridius	jmd: yeah, but it keeps all the branches as remote refs, and only produces a local branch for the remote repo's HEAD	11:18
jkp	Eridius: hey :)	11:18
Eridius	hi jkp	11:18
dmq	anybody know of a nice tool for doing blame mining in git?	11:18
JakubNarebski	if you don't want the default branch (before any commit is born) to be named branch, either "git checkout -b <branchname>" would be enough, or you would have to fiddle with git-symbolic-ref	11:18
dmq	a visual git-blame?	11:18
Eridius	jkp: if you have a URL to an svn repo, you can just add /branches/branchname to get just a branch (assuming standard layout)	11:19
jmd	Eridius,"git checkout" has always worked for me in the past. Are you sure you don't need to check the documentation yourself?	11:19
JakubNarebski	dmq: "git gui blame <file>"	11:19
Eridius	every folder in svn is its own repo	11:19
	→ mithro joined	11:19
Eridius	jmd: seeing as how you're the one asking the trivial question, yes I'm sure	11:19
jkp	Eridius: i tried this: git svn clone https://swig.svn.sourceforge.net/svnroot/swig/branches/gsoc2008-cherylfoil swig-doxygen.git	11:19
dmq	wow. color me impressed. :-)	11:19
Eridius	and what's it say?	11:19
jkp	it didnt like it, told me it wasnt an SVN repo	11:20
JakubNarebski	Eridius: git-svn has some options denoting where branches and tags are, and which one to select, see man git-svn	11:20
Gitbot	Eridius: the git-svn manpage can be found at http://git.or.cz/man/git-svn	11:20
dmq	thanks Jakub	11:20
Eridius	jkp: what version of git do you have?	11:20
	← gretchen left	11:20
Eridius	maybe there's something weird about an svn 1.5 repo?	11:20
jkp	ah, maybe not actually	11:20
tritz	jmd: git checkout defaults to HEAD	11:20
	→ grahal joined	11:20
jkp	i think i misread the message	11:20
Eridius	JakubNarebski: yeah, but he wants to clone a single branch, not denote the folder where all branches live	11:20
jkp	there is some kind of inconsistency in that repo i think	11:20
JakubNarebski	I guess that other such tools, like QGit, Giggle or (if you are MacOS X user) GitNub have similar things	11:20
	jkp tries again	11:20
Eridius	jkp: oh, I bet it was complaining about the path not existing?	11:20
jkp	y	11:20
Eridius	the message should have also told you to ignore it ;)	11:21
jkp	ive not used git-svn before, only git-p4	11:21
Eridius	it's just trying to find the first revision where the branch existed	11:21
jkp	which i know pretty well now	11:21
jmd	tritz, But HEAD is just an alias for something else isn't it?	11:21
Eridius	so it's saying "oops, I've found a revision where this branch doesn't exist"	11:21
jkp	right	11:21
Eridius	jmd: HEAD is the "current branch"	11:21
	well, unless it's detached. But you can ignore that for now	11:21
	what "git checkout" actually does is just spit out the message that you get when your local branch doesn't match the remote tracking branch, e.g. "Your branch is ahead of 'origin/next' by 1 commit."	11:22
JakubNarebski	HEAD -> refs/heads/<branchname> -> <commit = branch tip>	11:22
Eridius	you normally get that message when switching branches, assuming they're out-of-sync with the remote ref that they track	11:22
jmd	So my question is how can I configure git so that either a) HEAD is realiased to a branch of my choosing; or b) "git checkout" checks out a different branch instead of HEAD	11:22
Eridius	so "git checkout" doesn't actually checkout HEAD, it just spits out that message. It's effectively the same as checking out HEAD, though	11:23
	jmd: read the very first response you get	11:23
	*got	11:23
	seriously, if you're not going to bother reading the documentation, at least read the responses you get here	11:23
JakubNarebski	jmd: "git checkout <branchname>" changes checked out branch, and then HEAD is alias (in fact: symref) to <branchname>	11:23
	git usually uses HEAD for missing ref parameter (for example ..b is HEAD..b), or in some cases index	11:24
tritz	jmd: cat .git/HEAD gives something like 'ref: refs/heads/...'. after git checkout anotherbranch repeat cat .git/HEAD and see the difference	11:24
jmd	Hmm. If I say "git checkout foo" how does it know that foo is a branch and not a file that I want checked out?	11:24
Mikachu	because you don't have a file called foo	11:25
jmd	And if I happened to have a file with that name ?	11:25
Tv	jmd: that's why you use git checkout -- filename when you mean it	11:25
	→ sf_ joined	11:25
JakubNarebski	jmd: and if you have both, it would tell you so. If you want for example to check out from index file named HEAD, use "git checkout -- HEAD"	11:25
segher	sounds like it knows it's a branch because a branch with that name exists, not because there is no file with that name, then	11:26
JakubNarebski	jmd: the '--' is used thorough git to distinguish between ref parameters, and filename parameters	11:26
mnemoc	been able to add elements from subdirectories in .gitignore is a bug a or a feature? I mean, may this change in future?	11:29
jmd	So now I've done "git checkout <branch>" and I get a message saying the branch is not local.	11:29
Eridius	jmd: ok, it's telling you that you've now created a detached HEAD	11:29
	it sounds like you need to read some tutorials	11:29
	← alenoosh left	11:29
Eridius	or, hell, read the manpages	11:29
	I mean, it's a travesty that you might have to read the documentation supplied with git, but maybe you should just buckle down and do it	11:30
mnemoc	.oO( is that to me or to jmd? )o	11:30
Eridius	to jmd	11:30
mnemoc	:)	11:30
JakubNarebski	mnemoc: you can add ignored files. What ignoring files does is that they doesn't show up, and are not added by globbing and wholedir operations	11:30
jmd	It seems to have checked it out 'though.	11:30
Eridius	mnemoc: a .gitignore file controls the ignore status of all files rooted in that directory, including in subdirectories. The path used for comparing against the .gitignore entires all start from that directory	11:31
	so yes, a .gitignore can ignore files in subdirectories	11:31
	Eridius frequently uses a single .gitignore in the project root for simplicity	11:31
Eridius	jmd: yeah, but you have a detached HEAD	11:31
	which will cause you problems if you want to make any commits	11:31
JakubNarebski	jmd: git doesn't allow you to change remote-tracking branches, as they are used to track branches in remote repository, and they change on fetch (or "git remote update")	11:31
Eridius	or pull	11:31
JakubNarebski	jmd: git now allows to check them out to view using detached HEAD (HEAD -> <commit> directly, and it is not symref: check out "cat .git/HEAD"	11:32
mnemoc	I'm aware I can, I actually use it this wa, my doubt was if this may change in future releases because it's considered misusage... :) but It's clear now that it is considered a feature =) thanks ;-)	11:33
jmd	.git/HEAD just contains an md5 hash	11:33
JakubNarebski	you could have instead fo "git chackout -b <branch> --track origin/<bramch>", or with new git "git checkout --track origin/<branch>"	11:33
	jmd: BTW "git checkout <not a local branch>" should be quite chatty and tell you what you can do	11:33
mnemoc	uh, this is even nicer than git checkout -b foo origin/foo =)	11:33
parasti	it's exactly the same, except with an ancient option	11:34
JakubNarebski	(but IIRC "git checkout --track <remote>/<branch>" is in master, but not yet in released version	11:34
jmd	I really just want to tell git that I'm working on a branch, not the trunk. Surely that's possible?	11:34
Mikachu	that is a rhetorical question?	11:35
	since the answer is on the line above	11:35
JakubNarebski	jmd: please read Git User's Manual about remote tracking branches, then got back	11:35
	jmd: think about what happens when someone updates branch you base your work on in remote repository	11:36
jmd	I really would be quite happy dealing only with a local repository for now.	11:37
Eridius	mnemoc: you should read the gitignore manpage. It clearly outlines how the .gitignore files behave, including the very question you asked	11:37
mnemoc	no asciidoc, not manpages :'( .... Eridius: I'll search for an online version, thanks for the hint	11:38
Eridius	how did you install git?	11:38
JakubNarebski	if you checkout branches which are listed in "git branch" output, you would have no problems	11:38
	if you want to create new branch, use "git checkout -b <newbranch>"	11:38
Eridius	mnemoc: ↑?	11:38
jmd	What about those in "git branch -r"	11:39
mnemoc	Eridius: from source	11:39
	→ pi-meson joined	11:39
Eridius	mnemoc: you have a clone of git.git?	11:39
JakubNarebski	mnemoc: you can use Gitbot; see me saying 'man gitignore'	11:39
Gitbot	mnemoc: the gitignore manpage can be found at http://git.or.cz/man/gitignore	11:39
mnemoc	Eridius: no, tarball	11:39
tritz	jmd: that are remote branches and need to be checked out with: git checkout -b mylocalbranch theremotebranch	11:39
mnemoc	JakubNarebski: thanks =) I'll misuse Gitbot :p	11:40
Eridius	mnemoc: source tarball doesn't include manpages? huh	11:40
JakubNarebski	jmd: those are remote-tracking branches. as they are outside your control, they cannot be checked out directly. you can base your own local branches on them, e.g. "git checkout -b <local name> --track <remote>/<remote-tracking branch>"	11:40
Eridius	if you clone git.git you can install manpages with `git archive origin/man \| sudo tar xvC /usr/local/share/man`	11:40
jmd	How can I make those branches local ?	11:40
Eridius	git.git has a 'man' branch which includes built copies of the lateset manpages from master	11:40
	jmd: by reading documentation	11:40
JakubNarebski	jmd: see above	11:40
Eridius	or by, yeah, reading what tritz said	11:41
mnemoc	Eridius: cool	11:41
JakubNarebski	mnemoc: you can also get tarball of manpages and HTML documentation from kernel.org	11:41
Eridius	roughly once a day I pull the latest from git.git and then install the next branch	11:41
	I have a little alias which does a make -j5 and make install (with the right prefix), then runs that git archive \| tar	11:42
	so I can just say `git pull`, and assuming there were changes, `git install` and voilá	11:42
	→ foca joined	11:43
	← psankar left	11:43
JakubNarebski	Eridius: nice	11:43
Eridius	git config alias.install '!sudo -v && make -j5 all prefix=/usr/local && sudo make install prefix=/usr/local && git archive origin/man \| sudo tar xvC /usr/local/share/man'	11:43
	I also have an alias 'lcrev' which shows me all the commits that were just pulled, in reverse order, with no merges	11:44
	← botanicus left	11:44
Eridius	so I use that to see what's changed	11:44
	git config --global alias.lcrev 'log @{1}.. --stat --no-merges --reverse'	11:44
	→ Yuuhi joined	11:44
jkp	Eridius: so its super slow doing the original import: im making a local copy of the source repo via rsync and cloning that because by all accounts its a lot quickier	11:45
	question though, if i want to repoint the origin of my svn clone (git) after i have cloned my local svn copy, i can do that right?	11:46
Eridius	jkp: it is definitely quicker to pull from locally. the svn protocol isn't particularly efficient ;)	11:46
	yeah, though don't ask me for specifics. I've never done it	11:46
	I think there's something like svn.rewriteUrl	11:46
jkp	ie, i want it to point to the source repo not my local copy	11:46
	kk	11:46
	man that was quick!	11:49
	← lu_zero_ left	11:49
	→ kyofel joined	11:50
jkp	and so much quicker for the git svn clone as well	11:50
	that shoudl be on the FAQ for git-svn	11:51
	→ Nido joined	11:51
	→ whee joined	11:54
	← pi-meson left	11:58
	← sverrej left	11:59
	→ jitu3485 joined	11:59
	← deavid left	12:00
	← jitu3485 left	12:02
	→ mithro_ joined	12:03
	→ sverrej joined	12:03
	← mithro left	12:03
	→ kukks joined	12:03
	sciz\|away → sciz	12:04
	→ cannonball joined	12:08
	→ tmz joined	12:09
	← sciz left	12:10
	→ tmz_ joined	12:10
	mithro_ → mithro	12:10
	← JakubNarebski left	12:11
	→ lorandi joined	12:12
	sf_ → nud	12:13
	→ morphir joined	12:13
	← tmz_ left	12:16
	→ Lemurnomicon joined	12:18
	→ lanfeust21 joined	12:19
	← charon left	12:22
	→ Gof joined	12:23
	← jmd left	12:23
	← lanfeust21 left	12:27
	→ juanjoc joined	12:27
	← takayaW_ left	12:29
	→ flopenner95_ joined	12:30
flopenner95_	ich suche leute die in meine bande mommen wollen	12:31
mnemoc	flopenner95_: auf Englisch bitte	12:33
	← BarryCarlyon left	12:34
	→ deavid joined	12:35
	← lorandi left	12:38
	→ lorandi joined	12:39
	→ kanru joined	12:42
pasky	damn	12:43
	what's going on with repo.or.cz's dns?	12:43
	i have no idea where these temporary failures in resolution are coming from	12:44
	→ charon joined	12:44
	→ qrush joined	12:44
cehteh	hackers try to poison it and inject malware?	12:46
	→ bcarlyon\|laptop joined	12:47
	→ jitu3485 joined	12:48
	→ z3ro_ joined	12:48
	bcarlyon\|laptop → BarryCarlyon	12:49
	→ mw joined	12:51
lanwin	is there a way to disable the "WARNING: terminal is not fully functional" message in lastest msysgit?	12:51
	→ koke_ joined	12:55
	← flopenner95_ left	12:57
	→ antares joined	12:59
antares	is there a way to list remotes with their urls?	13:01
	→ mattetti joined	13:01
	→ djanowski_ joined	13:02
	← seanius left	13:04
tmz	antares: does 'git remote -v' do what you want?	13:04
	→ seanius joined	13:04
	→ ferdy joined	13:07
	← djanowski_ left	13:07
jitu3485	i have setup a apache server, and created a bare repository as per http://www.kernel.org/pub/software/scm/git/docs/howto/setup-git-server-over-http.txt : still i am not able to clone from the repository	13:07
mattetti	tmz: git remote -v is awesome, thanks	13:08
	→ Mikachu joined	13:09
mnemoc	there is a way to get the current branch more elegantly than $(git branch \| grep '^*' \| cut -c3-) ?	13:12
Mikachu	yes	13:12
mnemoc	Mikachu: which? :)	13:13
Mikachu	git symbolic-ref HEAD 2> /dev/null \|\| git rev-parse HEAD	13:13
	← koke left	13:13
mnemoc	ow	13:13
pasky	just git symbolic-ref HEAD	13:13
	if you want to know the current branch	13:13
	not current head ;)	13:13
Mikachu	it will bomb out on detached head	13:13
pasky	yes	13:13
	which is correct	13:14
	since then there's no current branch	13:14
mnemoc	G	13:14
	git symbolic-ref HEAD \| cut -d/ -f3 =) thanks!	13:14
pasky	I'd rather sed 's#^refs/heads/##'	13:14
Mikachu	or -f3- at least	13:15
	you can have a branch called foo/bar	13:15
mnemoc	oh	13:15
	← mattetti left	13:15
pasky	though support for these is inferior in git	13:15
Mikachu	yeah, if you had a branch called foo before, i think it will stumble a bit on the reflogs	13:15
albertito	jitu3485: you probably need to run "git update-server-info" inside the bare repository	13:16
	← BarryCarlyon left	13:16
jitu3485	albertito, k	13:16
pasky	and there's no way to write refspec matching all heads if you have heads containing slashes	13:16
albertito	jitu3485: in fact, you need to do that every time your repository changes, and you can use a hook for that. It's so common, you will find it by default in the .git/hooks/post-update	13:17
Mikachu	pasky: are you sure?	13:17
pasky	well	13:17
	at least I know noone who knows how to write such a refspec ;)	13:17
Mikachu	i have at least some refspecs doing it, maybe it's in a different place	13:18
	and you can for example fetch refs/:refs/ and it will match multiple slashes	13:18
j_engelh	jitu3485: just use git://	13:18
	→ ebel joined	13:18
	← mithro left	13:18
pasky	sorry, you're right	13:19
	since git show-branch doesn't actually take refspecs	13:19
	but plain globs	13:19
Mikachu	there's a command i've never used	13:19
pasky	for whatever reason	13:19
	it's ratheer useful	13:19
	if you have a lot of topic branches	13:19
	→ threeve joined	13:20
	→ EmilMedve joined	13:20
jitu3485	albertito, after running update error is fatal: : not a valid SHA1 fatal: Not a valid object name HEAD	13:20
Mikachu	the "graph" it shows is confusing at best	13:20
	→ bcarlyon\|laptop joined	13:21
	→ swhitt joined	13:21
	→ jm joined	13:22
	→ mithro joined	13:24
	← nud left	13:25
	← koke_ left	13:26
	→ koke joined	13:27
	→ rjent joined	13:27
	→ stouset joined	13:28
	→ rip747 joined	13:28
	← charon left	13:29
	→ rdavila joined	13:29
rip747	really quick question	13:29
Mikachu	doesn't seem to be that quick	13:29
rip747	i'm trying to edit a commit message and i'm doing something wrong	13:29
	here is the syntax i used: git commit --reedit-message "<commit sha>"	13:30
Mikachu	that's not what that option does	13:30
	it's for making a new commit with the same message as an old commit	13:30
rip747	what would be the syntax then?	13:31
	→ sf_ joined	13:31
	→ CIA-15 joined	13:32
albertito	rip747: if it's the last commit, git commit --amend. If it's before the last commit, then git rebase -i, mark the one you want to edit as "edit", and then follow the instructions	13:32
	rip747: note that this rewrites history, and if you have already published the commit in question, it will make a mess for people who cloned it and you shouldn't do it =)	13:32
Mikachu	and it also won't work if you've done a merge after that commit	13:33
rip747	trying it out now	13:33
albertito	jitu3485: do you have any commits in that repository, or is it empty?	13:34
	← antares left	13:34
jitu3485	albertito, it is empty	13:34
	← koke left	13:34
	→ koke joined	13:35
jkp	can anyone tell me how to repoint the origin of a git-svn mirror?	13:36
	← rdavila left	13:37
	← namenlos left	13:37
jkp	i created the repostiory using a local mirror of the origianl svn repo but now i want to repoint it to use the source repo on the web	13:37
	→ krh joined	13:37
	→ loincloth joined	13:38
	← advany left	13:38
Tv	jkp: same uuid? just edit .git/config i think	13:38
	← morphir left	13:39
	← stouset left	13:39
jkp	Tv: :) thanks dude	13:39
	→ morphir joined	13:40
	→ namenlos joined	13:40
parasti	I could swear I've seen guides of doing this -- can't find a single one of them anymore	13:41
	← EmilMedve left	13:42
jkp	parasti: doing what im trying to do?	13:42
parasti	yeah, it seems like a common scenario	13:42
	ben_h <3 git	13:43
Tv	dunno, git-svn(1) has example of the config	13:43
	i think it's kinda implied in all unix software that you edit configuration to change the, well, configuration	13:44
ben_h	you know what's awesome? git stash && git svn rebase && mate merge fixes && git rebase --continue && git stash pop	13:44
Tv	pop?	13:44
ben_h	like apply but it removes it from the stash stack	13:45
Tv	ah apply and throw away	13:45
	funky	13:45
ben_h	git stash list shows what's in the stack	13:45
parasti	Tv: hmm, haven't looked at git-svn(1) after realising it's kinda outdated	13:45
	← mithro left	13:45
ben_h	Tv: i assumed apply was doing that, until i listed one day and found about 20 items :)	13:45
Tv	i've never really used stashes as a stack, seems kinda messy to me	13:45
jkp	Tv: changign the config doesnt really work	13:46
	KidA% git svn rebase	13:46
	Unable to determine upstream SVN information from working tree history	13:46
Tv	i guess i just don't work in a LIFO manner in cases like that	13:46
ben_h	yeah me either	13:46
	i generally only have one item on the stash stack	13:46
Tv	ben_h: yeah i clear explicitly	13:46
	→ mithro joined	13:47
Tv	jkp: it probably has piles of assumptions; git-svn is crummy that way	13:47
	huh, funny, git-svn actually seems to use both urls and svn repo uuids.. it should know to compare just uuids	13:48
ben_h	the crumminess is mainly there for svn compatability	13:49
	→ tomsw joined	13:49
Tv	ben_h: not nearly all of it	13:49
ben_h	:)	13:49
Tv	jkp: that message is probably from the code that tries to decide what svn branch you're on -- it would have to compare urls	13:49
	jkp: try providing that info explicitly	13:49
	jkp: maybe it'll start being automagic after a commit	13:49
	jkp: hopefully needless to say, back up first	13:49
jitu3485	can you suggest link for setting up a git server?	13:50
wereHamster	jitu3485: your own server or on one of the git hosting services?	13:50
jitu3485	own server	13:51
	← moelee left	13:51
	→ jerbear joined	13:51
jkp	Tv: i dont have the commit bit	13:51
Tv	jitu3485: maybe you want gitosis	13:51
jitu3485	wereHamster, my own server	13:51
Tv	jkp: well a fetch then	13:51
wereHamster	jitu3485: your distribution should provide a init script for the git daemon	13:51
j_engelh	no... xinetd is just fine	13:51
jitu3485	mine is ubuntu hardy 8.04	13:51
	→ advany joined	13:52
	→ aiur joined	13:52
	→ EmilMedve joined	13:54
jitu3485	wereHamster, there is no init script for git-deamon in /etc/init.d/	13:55
wereHamster	jitu3485: google 'ubuntu hardy git-daemon-run'	13:55
jitu3485	ok	13:56
jkp	hmmm	13:56
	not getting anywhere fast here :/	13:56
	→ markkald_ joined	13:57
	→ softdrink joined	13:57
	→ nuncanada joined	13:58
	→ bf_ joined	13:59
	→ qrush_ joined	14:01
bf_	i was wondering if anyone could explain why 'git-whatchanged --pretty=format:"foo %H" ' outputs an additional line per commit? Is there a way to limit whatchanged to only the format?	14:01
	→ icwiener joined	14:01
	→ juanjo joined	14:02
Mikachu	bf_: what do you mean 'to only the format'? do you want git-log?	14:03
	→ spearce joined	14:04
ben_h	how can i discover which commit a file disappeared in?	14:07
	it was there before a rebase, and now it's gone :)	14:07
	← mithro left	14:08
rip747	albertito: thank you, it worked!	14:08
jwb	ben_h, git log -- <path of where file used to be>	14:08
ben_h	ahh yeah. thanks	14:08
	← rip747 left	14:09
	→ stouset joined	14:10
	→ mithro joined	14:10
jkp	anyone able to tell me how i might use this as the basis for my source repository relocation? http://ertza.murgilduta.net/2008/07/18/fixing-git-svn-after-sourceforge-update/en/	14:11
ben_h	anyone know of a way to batch-strip whitespace from a path?	14:11
parasti	bf_: as fas as I know, when using an explicit format, the commit SHA1 will always be printed	14:11
	→ bschindler joined	14:11
jkp	in that case hes only changign http to https	14:11
	i need to go from file to https	14:11
ben_h	i've got the precommit hook turned on that catches it, but it's quite laborious to strip it all file by file sometimes when people make large changes that introduce mess	14:11
jitu3485	Tv, when i execute git clone git@YOUR_SERVER_HOSTNAME:gitosis-admin.git it prompts for password , but i have not given any password	14:12
bschindler	Hi... I have a repo which has two branches and a remote, which has the same branches. Now when I do a git push, to which branch is being pushed to? I'm asking because git remote show origin shows both branches in both local branches	14:12
ben_h	bschindler: afaik, when you push to a remote, all the branches that exist locally and on the remote will be pushed	14:12
Tv	jitu3485: you're not using the ssh key, then	14:12
	jitu3485: or your ssh server is refusing the login, and not telling you because it's paranoid about secutiry	14:13
bschindler	okay... let me try	14:13
jitu3485	Tv, i am going as with http://scie.nti.st/2007/11/14/hosting-git-repositories-the-easy-and-secure-way	14:13
juanjo	I am trying to do a dcommit via svn+ssh and it doesn't work. With only svn I can do things like svn-git rebase, but I need svn+ssh to do commits. It looks like git is not able to parse the url from .config, but my git version is 1.5.4.3, which should support this. Any clue why this can happen?	14:13
Tv	jitu3485: that doesn't change anything in what i said	14:13
	→ davidfetter joined	14:13
jitu3485	ok i try again	14:14
	← qrush left	14:15
Simetrical	When I do git commit --author=..., how do I specify a timestamp other than the present?	14:15
Tv	Simetrical: env var, if nothing else	14:16
Simetrical	Which env var?	14:16
Tv	Simetrical: man git	14:17
ben_h	hmm just had a read of that blog post	14:18
	Tv: the gitosis config process looks really nice	14:18
	edit files and push and the post-commit hook takes care of it	14:18
	what happens if you make a syntax error in the config?	14:19
Tv	ben_h: the whole reason gitosis exist is to make it sweet to edit it.. i had the same setup before with hand-edited authorized_keys, that was horrible	14:19
ben_h	heh	14:19
Tv	ben_h: syntax errors lead to bad suck	14:19
	→ JimmyB__ joined	14:19
ben_h	:\|	14:19
parasti	Simetrical: maybe try man git-commit-tree instead	14:19
Gitbot	Simetrical: the git-commit-tree manpage can be found at http://git.or.cz/man/git-commit-tree	14:19
Simetrical	Oh, GIT_AUTHOR_DATE. Thanks.	14:19
Tv	ben_h: but the syntax is very hard to make invalid	14:19
ben_h	is there a way the commit can be like a transaction, where post-commit rolls back the whole commit?	14:19
Tv	ben_h: then again, locking yourself out is perfectly doable with valid syntax	14:19
Simetrical	It's in man git, but I almost missed it, since it only says "see git-commit-tree (1)". :)	14:19
Tv	ben_h: post-commit is too late to abort	14:19
ben_h	thought so	14:19
	sanity checks in pre-commit might be nice	14:20
Tv	ben_h: but yeah some day gitosis will get a rescue mode	14:20
	ben_h: i disagree with sanity checks	14:20
	→ cmarcelo joined	14:20
doener_	Simetrical: and an example for the date format is in git-tag(1)	14:20
Tv	ben_h: sanity checks mean writing the same code twice, and hoping they match (in the future, too)	14:20
Simetrical	Thanks.	14:20
Tv	ben_h: but a rescue mode would be good	14:20
	ben_h: something like ssh server sudo -H -u git gitosis-rescue HEAD~1	14:21
Simetrical	It would be nice if this were in the git-commit man page.	14:21
doener_	(hint: send patches!)	14:21
Tv	Simetrical: it's kinda low-level enough that most people wouldn't want it there	14:21
ben_h	Tv: i guess that's all you need	14:21
Tv	Simetrical: i mean, how often do normal people lie to their computers about what time it is	14:21
Simetrical	I guess so.	14:22
	Thanks, though.	14:22
Tv	there's also the aspect of duplicating information	14:22
ben_h	Tv: the last time i heard of that happening was a couple of years ago, in 1973	14:22
	qrush_ → qrush	14:22
jitu3485	Tv, id_rsa.pub key should be created on same system running gitosis or another system?	14:23
Tv	jitu3485: at the client	14:23
doener_	maybe just add an example for GIT_{AUTHOR,COMMITTER}_{DATE,NAME,EMAIL} to git-commit-tree(1) as well, and add a smallish note to git-commit(1)	14:23
Tv	jitu3485: don't transfer keys between machines	14:23
	doener_: yeah, that would be better	14:23
	jitu3485: also, you're asking ssh questions, not gitosis/git questions	14:24
	jitu3485: you should probably learn how ssh works first, or you'll be confused about the problems you see	14:24
jitu3485	ok	14:24
	foca → foca\|hola	14:25
	foca\|hola → foca	14:25
	→ zachinglis joined	14:26
	→ bryanl joined	14:27
	→ rdavila joined	14:29
Tv	jkp: oh btw you need the rewriteroot option	14:29
	apparently git-svn can't really handle svn urls changing	14:30
	← cilly left	14:30
Tv	even if it claims to be tracking via repo uuid	14:30
	stupid code	14:30
	→ cilly joined	14:30
	→ selckin joined	14:31
wereHamster	Tv: git svn clone $newurl newsrcdir; cd newsrcdir; git fetch ../oldsrcdir master:oldsrc; git merge oldsrc	14:32
doener_	wereHamster: git svn and merge? no good idea	14:33
	→ peeja joined	14:33
	← juanjo left	14:34
wereHamster	or, hm. Maybe git rebase HEAD oldsrc	14:35
bf_	sorry, stepped away. yeah, i want a single line per commit of only the particular fields I'm interested in. Is that possible?	14:35
	← ruphy left	14:35
Tv	bf_: git log --pretty=format:...	14:35
	→ ruphy joined	14:35
doener_	wereHamster: no, you'll need something like "git rebase --onto newsrc oldsrc~N oldsrc", where N depends on which is the first commit you want to rebase (obviously)	14:36
bf_	ok. and considering i'm not changing branches, I can use oldsha...newsha to limit the results?	14:36
doener_	wereHamster: using HEAD as upstream will fail miserably, as the commits from oldsrc and newsrc aren't the same (at least the git-svn-id lines _will_ be different, and that's enough...)	14:37
wereHamster	doener_: I assumed he want's all the commits from the old source copy over to the new source	14:37
	oh, right. I didn't realize that	14:38
doener_	wereHamster: that would start with r1 from svn from both clones, so all the svn revs get duplicated (and cause a shitload of conflicts ;-))	14:38
wereHamster	when I did that I used the same repo URL, so all tho commits were the same	14:38
doener_	ah yeah, trivial case, cloning the same repo again. Using --rewrite-root might even work to "simulate" that, but I'd not rely on it, and would want the correct new url in there anyway	14:40
	and there's the authors file that maps the svn names to real names and email addresses. If you use that and had your authors file out of date at some point for the old clone, the new clone of the same repo would still differ	14:41
	way too much stuff to look out for when using git-svn :-/	14:42
	→ lanfeust21 joined	14:46
	← lanfeust21 left	14:47
jkp	Tv: ahh	14:49
	ok	14:49
	→ Kbyte joined	14:49
jkp	does anyone know how to make git ignore changes to files that are already stored?	14:49
	← cilly left	14:50
jkp	ive cloned an autotools based project and running the build updates some autoconf related stuff	14:50
	im not interested in those changes though	14:50
Mikachu	adding generated files to the repo is not very smart	14:50
	→ cilly joined	14:51
	→ markoa joined	14:51
jkp	Mikachu: i dont have a choice when the repo was created using git svn	14:51
	it clones everything, and i dont want to delete them because then when i push back to the repo in the end it will delete them	14:51
Mikachu	i didn't say you were the not very smart person :)	14:52
	there are two answers to your question, one is "you can't", the other is "git update-index --assume-unchanged files"	14:53
jkp	Mikachu: thanks	14:53
	← jitu3485 left	14:53
	→ RandalSchwartz joined	14:53
	← ereslibre left	14:54
	← namenlos left	14:56
mnemoc	does "Lars Hjemli" (cgit guy) live here?	14:58
pasky	i think not, but i'm not quite sure	14:59
spearce	i don't think he lives on #git. i'm fairly certain he isn't an icrd process given his creativity with cgit. but best way to reach most people in the git world is to just email the mailing list, or the individual. most people around here are email-centric	15:00
mnemoc	I just realized there is no way to get a raw patch or file from cgit :-\	15:00
	spearce: yes, i'm writting him a mail currently	15:01
Mikachu	mnemoc: hm? http://hjemli.net/git/cgit/plain/README	15:01
	→ ikitat joined	15:01
mnemoc	oh	15:01
Mikachu	http://hjemli.net/git/cgit/patch/?id=b28765135dd6f52635977454eaf95d0e6c7e7271	15:01
	mnemoc looks at his installation	15:02
ikitat	hi, I'm trying to clone a git svn clone following the directions at the bottom of the git-svn man page	15:02
wereHamster	why are people leaving gitweb for cgit? Is gitweb slow or not being maintained anymore?	15:02
mnemoc	wereHamster: slow	15:02
wereHamster	is cgit written in C? (I know gitweb is perl)	15:03
ikitat	when I get to the last step `git svn rebase` it fails	15:03
mnemoc	Mikachu: this has to be new, I updated one week ago and I don't have that	15:03
Mikachu	are people leaving gitweb for cgit?	15:03
ikitat	log --no-color --first-parent --pretty=medium HEAD: command returned error: 128	15:03
mnemoc	Mikachu: freedesktop did	15:03
	← rdavila left	15:03
mnemoc	(I did, but I'm noone)	15:03
ikitat	i'm using cygwin on vista :( version 1.6.0	15:03
wereHamster	git.kernel.org seems reasonably fast, but gitweb.fd.o is _very_ slow	15:03
	← Sho_ left	15:04
ikitat	here are the steps I'm following http://dpaste.com/76937/	15:04
Jc2k	wereHamster: gitweb.fd.o doesnt have the caching patch that i heard rumors of	15:05
mnemoc	Mikachu: 8 days Merge branch 'lh/plain' :)	15:05
ikitat	here is the full error I'm getting http://dpaste.com/76938/	15:05
Mikachu	wereHamster: kernel.org has a super fast version of gitweb	15:05
Jc2k	Mikachu: well, a super cached version of gitweb..	15:05
ikitat	the svn repository took over 8 hours to clone, and i really don't want to use svn directly	15:05
mnemoc	I guess git.kernel.org has monster hardware behind	15:06
Mikachu	that too	15:06
	← rtomayko left	15:06
ikitat	my original git svn clone seems to work just peachy though	15:06
wereHamster	Jc2k: was that patch already merged into upstream?	15:07
j_engelh	mnemoc: I don't guess ;)	15:07
mnemoc	j_engelh: :D	15:07
parasti	ikitat: maybe you just don't have a branch checked out	15:07
Jc2k	wereHamster: i dont know if it was merged in to 1.6 or not. it certainly wasnt before 1.6.	15:07
ikitat	parasti, this is true	15:07
	git branch returns nothing though	15:07
	if I try to do a git checkout, I get nothing	15:08
	er, i should say I get a failure	15:08
parasti	pick a branch from 'git branch -r' and do a 'git checkout -b master $thatbranch'	15:08
wereHamster	Jc2k: how can I find out?	15:08
ikitat	hey, that's a start	15:09
	nice!	15:09
	← sf_ left	15:09
ikitat	that step is missing from the man page	15:09
	I feel like I've spent hours trying to solve it ...	15:09
	parasti, thank you very much... you made my day	15:10
	← cmarcelo left	15:11
	← ikitat left	15:12
	← cilly left	15:13
	→ X-Blade joined	15:16
	→ schacon joined	15:16
	← statim left	15:17
	→ statim joined	15:17
	→ cmarcelo joined	15:18
Jc2k	wereHamster: i have no clue, this is just an observation from idling in here too much	15:19
	→ bdiego joined	15:20
	→ cilly joined	15:22
	→ jackdempsey joined	15:22
	→ deavid1 joined	15:23
	→ sf__ joined	15:25
	bcarlyon\|laptop → BarryCarlyon	15:26
	← aiur left	15:27
	← X-Blade left	15:29
	→ drewr joined	15:29
	→ X-Blade joined	15:32
	→ perezd88 joined	15:33
	→ LiamH joined	15:37
	→ vbabiy joined	15:40
	← lanwin left	15:42
	→ sbfaulkner joined	15:42
	← schacon left	15:44
tomsw	A newbie question: git blame shows me the last commit responsible for changing a particular line in a file. Is there any way I can see the entire commit history for that line in that file?	15:46
	→ neilio joined	15:47
	← neilio left	15:47
	→ neilio joined	15:47
	→ rip747 joined	15:49
	← sgrimm left	15:50
rip747	does anyone know anyway to resync a git with an svn repo. i tried just doing a git reset --hard <commit id> and then doing a git svn rebase but no dice	15:51
	← qrush left	15:51
	→ kumbayo joined	15:52
ciaran	git reset --hard trunk && git svn rebase	15:53
	← perezd88 left	15:53
	→ CodeOfficer joined	15:54
tritz	tomsw: I'd do 'git blame <file>', copy the sha1 from the beginning of the particular line, the 'git blame <sha1>^ <file>' repeatedly to limit the blame to <sha1>'s parent commit	15:56
	← sf__ left	15:56
ciaran	how is that different from git log <file>?	15:56
	← stouset left	15:57
tritz	ciaran: I guess that would also list the file even if that particular did not change	15:58
	particular line	15:58
ciaran	and yours won’t?	15:58
tritz	ciaran: blame shows the commit which changed this line	15:59
	← kopa left	15:59
ciaran	but you are just showing the blame info for file at the parent of that commit, no? So why is it localising to a given line?	15:59
	→ lea_w joined	16:00
rip747	ciaran: tried that but it didn't work.	16:00
ciaran	define didn’t work	16:00
	← neilio left	16:01
rip747	when i do the rebase it doesn't fetch anything from the svn server	16:01
cardioid	ciaran: You get the blame info, find that line again, and get the sha1 from that line	16:02
	← BarryCarlyon left	16:02
cardioid	It's a very interactive process	16:02
ciaran	I see	16:02
tritz	ciaran: left to the interesting line in blame there is the sha1 that last changed that line, if I tell blame to start at sha1^ then in this listing I get left to the line the previous commit that changed that line	16:02
	→ stouset joined	16:03
cardioid	It'd be interesting to make that automatic; give it a file and a line	16:03
	← stouset left	16:04
tritz	cardioid: how would you identfify the line along the history. It surely has changed. The line number also doesn't count.	16:04
	← amerine left	16:05
Mikachu	it would be nice if git gui blame had a link to sha1^ instead of sha1	16:05
	spearce: ^ :)	16:05
	← rmh3093 left	16:05
	← sverrej left	16:05
cardioid	tritz: It wouldn't be easy	16:05
	That's why it's a very interactive process. Making a computer figure it out would be hard.	16:05
tritz	cardioid: one could guess it from the diff	16:05
Mikachu	isn't that what blame does in the first place?	16:06
	← dejanpan left	16:06
	→ aroben joined	16:07
tritz	Mikachu: but you don't see how it was before. all you get is the new line content and who and when changed it.	16:07
	→ qrush joined	16:07
cardioid	tritz: Ah, true. Do a diff as well, apply it, look for that line now	16:07
tritz	exactly	16:07
	if multiple lines changed in the diff that makes it hard to determine which line became which	16:08
tomsw	I'm using git blame -L '/some pattern/,//' to get the line I want	16:09
cardioid	tomsw: That gives me an error	16:10
	-L '/some pattern',+1' works, though	16:11
dmq	has anybody encountered incorrect log entries in post-receive-mails generated when pushing to a branch?	16:11
	← jackdempsey left	16:11
	← bartocc left	16:11
	→ adamm joined	16:12
tomsw	cardioid, it works for me: git blame -L '/ABSENCE.TYPE/,//' models/data-models/resources-models/system.lisp	16:12
	as does the "git blame <sha1>^" process - thanks!	16:12
	← jkr left	16:12
cardioid	I get fatal: -L parameter '': empty (sub)expression	16:13
	→ sverrej joined	16:15
	← priidu left	16:15
	← tritz left	16:16
	→ perezd88 joined	16:16
	← ben_h left	16:17
	→ Ryback_ joined	16:17
	← advany left	16:20
	← LostInTheWorld left	16:20
	→ sanity_ joined	16:20
	→ sgrimm joined	16:21
	→ Eludias joined	16:21
sanity_	I've created a new branch locally, and would like to push it to a remote repository (github) - I did this before but have forgotten how :-/	16:21
	→ jackdempsey joined	16:21
	← aroben left	16:25
	→ neilio joined	16:28
cardioid	sanity_: give push the branch name	16:28
neilio	I used git-svn to pull down my subversion repo that has a trunk and branches/redesign set up. I'm now trying to move everything up to github	16:29
sanity_	cardioid: didn't work :	16:29
	Vegas:SenseArray(roc-for-vaclav) ian$ git push roc-for-vaclav	16:29
	fatal: 'roc-for-vaclav': unable to chdir or not a git archive	16:29
	fatal: The remote end hung up unexpectedly	16:29
neilio	I've figured out how to get the master up there - git push origin master	16:29
	→ schacon joined	16:29
neilio	but how do I get the branch/redesign also added as a remote git branch?	16:29
sanity_	ah: git push origin roc-for-vaclav - worked	16:30
	→ bobesponja joined	16:30
	← Kbyte left	16:31
	← rip747 left	16:32
	→ ijcd joined	16:32
	← ijcd left	16:33
	→ lon_ joined	16:33
	← peritus_ left	16:34
	→ dynamin joined	16:35
	→ advany joined	16:37
dmq	man	16:38
	contrib/post-receive-email is behaving very weird	16:39
	whenever we push to a branch it shows the log entry for the most recent thing on trunk	16:40
	er, master. :-)	16:40
	← X-Blade left	16:41
	→ Fullmoon joined	16:42
	← Fullmoon left	16:43
	→ bcarlyon\|laptop joined	16:43
	→ X-Blade joined	16:44
lon_	Hi guys, is there a book or website you can recommend me for learning git?	16:45
	→ joern_ joined	16:46
neilio	lon_ Check out the peepcode screencast	16:47
	← schacon left	16:47
albertito	lon_: http://git.or.cz/gitwiki/GitDocumentation has several pointers	16:47
	lon_: I personally like http://www-cs-students.stanford.edu/~blynn/gitmagic/ and http://eagain.net/articles/git-for-computer-scientists/	16:47
lon_	thanks! looks great!	16:48
	← neilio left	16:49
	→ joshbuddy joined	16:52
	← Catfish left	16:52
	← bobesponja left	16:53
	← kanru left	16:54
	← eternaleye left	16:55
	← adamm left	16:55
	→ ehc joined	16:55
	← Zarin left	16:56
	← dynamin left	16:56
	← ehc left	16:57
	← bts- left	16:57
	→ adamm joined	16:58
	bcarlyon\|laptop → BarryCarlyon	16:58
	→ neilio joined	16:58
	→ ehc joined	16:59
	← neilio left	16:59
	← adamm left	17:00
	← bf_ left	17:00
	→ rmh3093 joined	17:01
	dmq apologises for slandering post-receive-email	17:02
dmq	was my own damn fault.	17:02
	← ehc left	17:03
	← cbreak left	17:04
	→ anttih joined	17:07
	← Yuuhi left	17:08
	→ rtomayko joined	17:08
	← sergiodj left	17:09
	→ sergiodj joined	17:09
	← anttih left	17:10
	← lon_ left	17:11
	→ schacon joined	17:15
	← sverrej left	17:15
	← shodges left	17:20
	→ uau joined	17:20
	← deavid1 left	17:20
	→ neoeinstein joined	17:21
	← cilly left	17:22
	→ cilly joined	17:23
	→ joydivider joined	17:23
	→ mbroeker joined	17:23
	→ drizzd_ joined	17:25
	← hurikhan\|Work left	17:29
	← perezd88 left	17:29
	← mediogre left	17:29
	← vbabiy left	17:31
	→ vbabiy joined	17:31
	→ davi joined	17:33
	→ aiur joined	17:35
	← cilly left	17:38
	→ cilly joined	17:39
	→ f0rkit joined	17:40
	← krh left	17:40
	← qrush left	17:41
	→ perezd88 joined	17:42
	→ adamm joined	17:43
	← drizzd left	17:43
	← f0rkit left	17:44
	→ stouset joined	17:45
	← SRabbelier left	17:45
	→ SRabbelier joined	17:46
	← ebel left	17:47
	← advany left	17:47
	→ tritz joined	17:49
	← davi left	17:50
	→ ijcd joined	17:50
	← girishr left	17:54
	→ girishr joined	17:54
	← markoa left	17:54
	← joern_ left	17:55
	→ aroben joined	17:55
	← cilly left	17:57
	→ cilly joined	17:57
	→ sverrej joined	17:58
parasti	echo hmm	18:01
	→ robinr joined	18:02
	→ ahe joined	18:02
	→ thiago_home joined	18:03
wereHamster	git svn clone or fetch crashes with a sigsegv :(	18:03
	→ tongueroo joined	18:04
	← robinr left	18:04
	→ robinr joined	18:07
	← koke left	18:10
	→ koke joined	18:10
	→ lu_zero joined	18:11
	→ Pistos joined	18:12
	→ invalidrecord joined	18:12
Pistos	Is it generally a safe procedure to cherry pick commits out of a branch and apply/merge them to trunk?	18:12
invalidrecord	silly question but whats the git eqiv of svn log . -vr HEAD:BASE	18:12
Pistos	i.e. not going to confuse future merges to trunk, or rebases from trunk to branch?	18:12
wereHamster	I have this git log output: http://pastey.net/95714 And I'm confused that both git show HEAD and HEAD~1 show a commit with the following message: "Merge branch 'master' of ssh+git://192.168.0.11/home/tomc/devel/murrine" but different hashes	18:13
thiago_home	wereHamster: ok, so?	18:14
jast	invalidrecord, to answer that question we first need to know what that svn command does. ;)	18:14
doener_	wereHamster: merge, do more commits on the other branch, merge again	18:14
wereHamster	thiago_home: I would expect HEAD~1 to be 9bb7ec252a9999edfb519f2dc1d9ad2ac1d4b470	18:14
thiago_home	why?	18:14
wereHamster	as shown in the log, the second entry	18:14
thiago_home	how can you guess what the SHA-1 will be before it is done?	18:14
doener_	thiago_home: likely HEAD^2 is 9bb7ec...	18:15
invalidrecord	jast: it shows the logs from head to your copy	18:15
wereHamster	doener_: yes, that's true	18:15
jast	wereHamster, HEAD has two parents. HEAD~1 means the first parent.	18:15
	← drewr left	18:15
doener_	and HEAD~^ is HEAD^1, ie. an other commit	18:15
thiago_home	wereHamster: the first merge created a commit	18:15
invalidrecord	im porting a script to use git	18:15
thiago_home	the second merge created a new commit	18:16
	the problem is that 9bb7ec252a9999edfb519f2dc1d9ad2ac1d4b470 does not contain HEAD~	18:16
	so git had to create that second merge commit	18:16
	→ f0rkit joined	18:17
	→ Bryson joined	18:17
jast	invalidrecord, something like git cherry [-v] origin?	18:17
	→ xyzzy42 joined	18:17
wereHamster	thiago_home: I'm sure it makes sense to you, but I did not understand a single word you just said -.-	18:18
	but don't worry, it's not that important	18:18
thiago_home	wereHamster: first of all, your log is too short	18:18
Pistos	What option do I need on git log -Sfoobar to show the full diff of each matched commit?	18:18
thiago_home	Pistos: -p	18:18
Pistos	thiago_home: Thank you.	18:18
doener_	Pistos: --pickaxe-all -p	18:18
Pistos	Beauty.	18:18
invalidrecord	jast not sure ill try	18:18
doener_	thiago_home: doesn't -p alone just show the matching hunk?	18:18
xyzzy42	I want to get the abbreviated git commit sha of the checkout of source for a build script	18:19
	is there a good command for this?	18:19
doener_	xyzzy42: look at "git describe"	18:19
wereHamster	thiago_home: below the third commit there is a nice linear history as imported from a svn repository	18:19
doener_	xyzzy42: not an abbreviated hash, but better ;-)	18:19
thiago_home	xyzzy42: git log --pretty:format=%t HEAD^!	18:19
	→ bts- joined	18:19
jast	describe only works if you have tags	18:19
xyzzy42	doener_: that would nice, except the tree doesn't have tags	18:19
	I tried: git rev-list -n 1 --pretty=format:g%h HEAD but it gives an extra line out output	18:20
	← warthog9 left	18:20
jast	xyzzy42, a low-level command to do this is git rev-parse --short HEAD	18:20
doener_	git rev-parse --short HEAD	18:20
thiago_home	yeah, even better	18:21
	or HEAD^{tree} if you want the tree's SHA-1 instead of the commit's	18:22
xyzzy42	thanks, that's exactly what I needed	18:22
jast	or HEAD: if you want it to be shorter :)	18:22
xyzzy42	I wish git-describe would just default to that if there aren't any tags	18:23
	thiago_home learns something new too	18:23
thiago_home	HEAD:	18:23
	→ alley_cat joined	18:24
doener_	xyzzy42: "git describe --all --long" might also be interesting if you don't always package from the same branch	18:24
	← aiur left	18:25
	→ juanjo joined	18:25
thiago_home	the problem with git describe --all is that it matches the branch name	18:25
xyzzy42	--long must be too new for my get	18:25
juanjo	Hi, does anyone know where this message comes from when I try to do a rebase after a fetch? "Unable to determine upstream SVN information from working tree history"	18:25
thiago_home	juanjo: from git-svn	18:26
juanjo	thiago, yes	18:26
doener_	thiago_home: that was the idea ;-) He has no tags, and that way you now at least one branch in which the commit was	18:26
juanjo	thiago_home, yes, using git-svn	18:26
doener_	xyzzy42: "git describe --always" does fallback to the abbreviated commit hash	18:26
thiago_home	juanjo: that wasn't a question. I was answering your question.	18:26
juanjo	thiago_home, and how can one fix it?	18:26
jast	"unable to determine..." typically shows up when the URL information in the commit logs doesn't match the configured svn repository URL	18:27
	→ Method joined	18:27
xyzzy42	doener_: looks like git describe has been improved since 1.5.4	18:27
Method	there doesn't seem to be a git export, how do I export a repo into a non-revision controlled state for releases and such?	18:27
doener_	Method: git archive	18:27
Method	ta	18:27
thiago_home	juanjo: you probably did a merge of other SVN stuff. That's confusing git-svn.	18:27
parasti	juanjo: there's a dozen of different reasons why that message might show up; tell us what you actually did	18:27
thiago_home	juanjo: take a look at your git log and see where the last SVN commit is.	18:28
	juanjo: remove the rest of the commits.	18:28
	→ ehc joined	18:28
	← koke left	18:29
juanjo	parasti, The story is the following one: I created a repo in github based on a svn repository. I worked there during a couple of months with other people. Now we want to merge it back to SVN. We do a git clone of github, then a git reset to the state of the repo when we mirrored the svn, and I try to do a fetch and rebase from the svn after I edit .git/config. When I try that, fetch works but rebase answers with an error message as above	18:30
	← lea_w left	18:30
thiago_home	juanjo: that wasn't a good idea	18:30
juanjo	thiago_home, which one? :)	18:30
thiago_home	working on a git-svn repository with other people	18:31
	you have to rebase all of the changes on top of current SVN	18:31
juanjo	thiago, github is not a git-svn repository	18:31
thiago_home	I recommend first linearising the history	18:31
	remove all merges	18:31
	← kbingham left	18:31
juanjo	thiago_home, we worked against github, not against git-svn. I am trying to use git-svn just to get now the changes merged back to SVN from github remote repo	18:32
thiago_home	right, but you started with a git-svn clone	18:32
	and then you pushed that to github	18:32
juanjo	yep	18:32
thiago_home	and then many people pushed to github, right?	18:32
parasti	juanjo: you could try a git-svn init in the cloned repo, then tweak .git/config	18:32
juanjo	exactly, and a few people to SVN meanwhile	18:32
thiago_home	that was the problem	18:33
	← aroben left	18:33
Method	doener_: do you know what fatal: Not a valid object name means with git archive?	18:33
thiago_home	I recommend re-cloning from SVN	18:33
	→ dec_ joined	18:33
thiago_home	then rebase the Git changes on top of that	18:33
juanjo	thiago_home, how do I rebase the github changes on top of a git-svn clone? which commands should I use?	18:34
Method	doener_: nm, i needed to put HEAD	18:34
thiago_home	juanjo: git rebase	18:34
	juanjo: the --onto option is probably useful too	18:34
juanjo	thiago_home, git rebase for sure :) But I was asking which exact commands cause I have tried a thousand combinations and I get all kind of errors, as the one I mentioned before	18:35
thiago_home	juanjo: git svn clone ....; git fetch github master:github; git checkout github; git rebase master	18:35
	make sure that none of the commits you're adding have git-svn-id lines	18:36
juanjo	parasti, not sure if I understand your suggestion, you mean git-svn init in the cloned github repo?	18:36
	← icwiener left	18:36
parasti	juanjo: that's what I said -- however, the effect would be the same as that thiago_home suggests	18:36
	s/that/what/	18:37
juanjo	ok, I will try that right now	18:37
	→ mczepiel joined	18:38
parasti	if you have the option of just running git-svn clone again, then use it	18:38
	→ hpa joined	18:38
thiago_home	in general, I'd expect you to already have a pristine git-svn clone somewhere	18:38
parasti	what you're trying to do is not trivial	18:38
thiago_home	so the first step shouldn't be needed	18:39
	mw → mw\|food	18:40
	← sgrimm left	18:40
	→ sgrimm joined	18:41
	→ advany joined	18:41
	→ koke joined	18:42
	← koke left	18:44
	→ robhudson joined	18:46
	→ etnt joined	18:46
	← robhudson left	18:46
	← etnt left	18:46
	→ __elihu__ joined	18:46
	→ bentob0x joined	18:46
Method	doener_: if i do --prefix=whatever-version HEAD whatever then the resulting tarball says whatever-versionwhatever/*	18:50
	← sgrimm left	18:50
	→ sgrimm joined	18:50
	← Lash\| left	18:50
	→ warthog9 joined	18:50
juanjo	thiago_home, in your suggestio, what's the difference between doing the fetch in github branch and then rebase and directly doing the fetch on top of the master?	18:51
	→ Lash\| joined	18:53
	← jackdempsey left	18:54
	→ JarrettGreen_ joined	18:55
JarrettGreen_	I have recently made a .gitignore to ignore some osx directoy file stuff. But those files are already in my repo on github- how can i remove them?	18:56
albertito	JarrettGreen_: remove them locally using git rm and git commit, then push the resulting commits	18:56
JarrettGreen_	I'm ignoring them after they've already been added.	18:56
	so if the file is .DS_Store I should git rm .DS_Store	18:57
	them commit and push that?	18:57
	← softdrink left	18:57
albertito	JarrettGreen_: you can use -r to make it recursive, and yes	18:57
	JarrettGreen_: verify that it does what you expect using git status prior to commiting	18:58
JarrettGreen_	great thanks!!!!	18:58
	→ david_koontz joined	18:58
	← Pistos left	19:01
	→ softdrink joined	19:01
	→ not_benh joined	19:04
	← tjafk1 left	19:04
	→ bobesponja joined	19:04
	← bobesponja left	19:05
	→ bobesponja joined	19:06
	← bobesponja left	19:06
	→ bobesponja joined	19:07
	→ piotrb joined	19:09
	← cedricv left	19:09
	← ahe left	19:10
piotrb	I'm importing a svn repos to git, and I'm wondering if its possible to do a multi-phase import of sorts. By that I mean import a directory tree, and then import another tree inside of it (I'd like to combine history of 2 seperate but related trees into one tree (tree2 being a subfolder of tree1)	19:11
	← advany left	19:11
jast	piotrb, look into man git-submodule	19:11
Gitbot	piotrb: the git-submodule manpage can be found at http://git.or.cz/man/git-submodule	19:12
jast	though that may not be exactly what you want	19:12
	→ rtomyako joined	19:12
piotrb	sounds like I want the exact opposite	19:12
not_benh	jast: are you looking to merge two seperate svn repos in to one git repo?	19:13
jast	i'm not looking for anything	19:13
piotrb	not_benh: I am yes	19:13
	not_benh: technically they're just 2 diffrent paths in the same repo .. but the idea's the same	19:13
not_benh	piotrb: oddly thats what I was going to ask... you just beat me to it.	19:14
piotrb	... you're no help then =)	19:14
not_benh	piotrb: if I find anything I'll let you know	19:14
piotrb	cool .. will lurk around here for a bit	19:14
not_benh	piotrb: sadly nothing good to report so far	19:14
	→ johnjosephbachir joined	19:14
	← bentob0x left	19:17
	→ Beket joined	19:17
	← jm left	19:19
	→ ebel joined	19:21
	← johnjosephbachir left	19:21
piotrb	not_benh: interesting line from top of git-submodule doc: If you want to merge the project histories and want to treat the aggregated whole as a single project from then on, you may want to add a remote for the other project and use the subtree merge strategy, instead of treating the other project as a submodule.	19:23
	← f0rkit left	19:24
not_benh	piotrb: interesting	19:25
	← ijcd left	19:27
	→ icwiener joined	19:27
seanius	out of curiosity, is there any way to change the ctime/mtime on a file after it has been entered into git?	19:27
	→ ijcd joined	19:27
seanius	specifically to set it back to a date before the time in the first commit?	19:28
vmiklos	iirc git does not store file modification dates, only the ones of the commit.	19:29
	so you can't have files with different dates in the first commit.	19:29
	← pygi left	19:30
	← sergiodj left	19:30
seanius	vmiklos: hrm.. so then wrt mtime, can you think of an easier way to create a commit that takes a file "back in time" than one of those date fudging utilities?	19:31
vmiklos	imho you can just create a commit back in time	19:32
	do a git ls-tree HEAD, the date is not stored.	19:32
	and for the blob, only it's type (blob) and the length is stored.	19:32
	← tomsw left	19:33
seanius	so it's the times in the patch itself that infer the time on the file then?	19:33
	→ GuyFromHell joined	19:34
seanius	er, s/patch/commit/	19:34
	← ijcd left	19:34
	→ ijcd joined	19:34
not_benh	is there a way to clone to a branch?	19:35
	→ _WINNIE_THE_BEAR joined	19:35
_WINNIE_THE_BEAR	where can i download git ?	19:35
	anyone knows ?	19:35
	→ jm joined	19:36
MaddHatter	did you try the home page listed in the topic?	19:36
albertito	_WINNIE_THE_BEAR: http://git.or.cz/ has instructions	19:36
MaddHatter	not_benh: clone the repository then checkout the branch you need	19:36
	mw\|food → mw	19:37
_WINNIE_THE_BEAR	thanks...	19:37
seanius	hrm... it seems that commit -c almost does what i want	19:37
	→ eternaleye joined	19:37
	→ sergiodj joined	19:38
GuyFromHell	Anyone have insight beyond the "it exists" that I'm finding online for consistent segfaults on git-svn cloning?	19:39
_WINNIE_THE_BEAR	ok http://code.google.com/p/msysgit/downloads/list hawe more then one files for download and i cofused about what to download	19:39
	→ lea_w joined	19:40
_WINNIE_THE_BEAR	can anyone help me finding which is for using to download files ?	19:40
	from svn's ?	19:40
	repositories or whatewer they called.	19:40
seanius	lolz	19:40
GuyFromHell	>_> I'm going to say that git is saved in a git repository...	19:41
MaddHatter	just download and install the latest version (@winnie)	19:41
_WINNIE_THE_BEAR	ok.. thanks ;)	19:42
MaddHatter	you don't need the net install	19:42
_WINNIE_THE_BEAR	yeah i dont like net installers ;)	19:42
	i downloaded some files using ftp.	19:45
	so they are a bit different.	19:45
	can git convert those comment codes from a cpp file ?	19:46
	convert ? i meant remove .	19:46
	→ LostInTheWorld joined	19:47
_WINNIE_THE_BEAR	and when you browsw some programmers files using nonstandard way you hawe to be ready to see strange things ;)	19:47
	→ joevandyk joined	19:48
	→ lolage0 joined	19:48
	→ abjdshfsd joined	19:48
	→ dave_h_d joined	19:49
MaddHatter	git doesn't modify your files (except possibly to convert line endings)	19:50
	← sverrej left	19:50
	← Method left	19:51
	→ sverrej joined	19:51
Dreamer3	so i have a merge that didn't go so well	19:51
	tons of conflicts	19:51
	how can i just undo it	19:51
	i haven't commited	19:51
	reset --hard HEAD?	19:52
	i'm still not sure when to use hard	19:52
	← Beket left	19:53
piotrb	not_benh: you there?	19:53
not_benh	piotrb: sure am	19:53
	→ Beket joined	19:53
piotrb	I seem to be on the right track	19:54
not_benh	piotrb: sweet	19:54
Dreamer3	anyone?	19:54
piotrb	I managed to merge 2 sample git repos into eachother	19:54
not_benh	piotrb: I'm currently trying to use git svn remote but it seems that things stay disconnected	19:54
piotrb	not_benh: first things first	19:54
	→ tjafk joined	19:54
not_benh	piotrb: nice, so are you planning on pulling two seperates and then merge from there	19:54
piotrb	just pull down the 2 projects seperately with git-svn as if you were to import them seperately	19:55
MaddHatter	Dreamer3: that will work; I'm not sure if there's a more canonical way to do it	19:55
piotrb	yea you not gonna do it in one step	19:55
	this way you have 2 git repos with full history	19:55
not_benh	piotrb: theres always room to dream	19:55
piotrb	not_benh: then check out http://www.kernel.org/pub/software/scm/git/docs/howto/using-merge-subtree.html	19:56
	not_benh: with that I was able to merge my sample (1 file) repos into eachother .. git2 as /g2 inside git1	19:56
_WINNIE_THE_BEAR	is there a web page like goodle which searches ftp svn like spesific pages ?	19:56
	without wasting all my bandwith ?	19:57
piotrb	not_benh: http://pastie.org/269149	19:58
	not_benh: each of the "foo" commits was on a seperate repos initially	19:59
not_benh	piotrb: ahh very cool	20:00
piotrb	and of course .. the 2 explicitly named git1/git2 commits were in their respective repos	20:00
Dreamer3	errr	20:01
	i'm getting lost int he middle of a rebase	20:01
	--continue does nothing because it says no changes	20:01
	and i'm stuck on "not currently on any branch"	20:02
	→ dduncan joined	20:02
	← dduncan left	20:03
	← warlock_mza left	20:05
tritz	Dreamer3: if lost, "git rebase --abort" takes you to right before the rebase	20:05
	← cilly left	20:05
	← rmh3093 left	20:07
	← JarrettGreen_ left	20:11
	← Lash\| left	20:11
Dreamer3	yeah	20:13
	that works	20:13
	ok i was doing it wrong	20:13
	now... once i've done a rebase... can you do a rebase the opposite way?	20:13
	→ krh joined	20:14
	→ pygi joined	20:16
	stouset → stouset\|away	20:16
parasti	"rebase the opposite way" doesn't make sense to me	20:16
	maybe you could explain what you mean exactly by "opposite way"	20:17
Dreamer3	ok, i branch from stable	20:17
	i make patch	20:17
	← bdiego left	20:17
Dreamer3	i rebase onto master with stable as upstream...	20:17
	i merge my branch into master	20:17
	it's banged on in master	20:17
	now i want to deploy it to stable	20:17
	← _WINNIE_THE_BEAR left	20:18
Dreamer3	can i rebase "back to" stable	20:18
	→ priidu joined	20:18
Dreamer3	originally the branch could have been merged with stable, but not now that's it's rebased onto master	20:18
parasti	oh... I guess you can, by just switching master and stable	20:18
Dreamer3	yeah, i can't seem to rebase backwards	20:19
	or could be because i merged it into master :)	20:19
	i'm trying to find the best way to apply groups of changes to both master and stable	20:19
	we test things as a group in amster, but then sometimes push things to production in bits and pieces	20:19
	→ mario_ joined	20:21
	← pygi left	20:22
	mario_ → pygi	20:22
tritz	Dreamer3: how about merging?	20:25
	Dreamer3: you may want to take a look in http://repo.or.cz/w/git.git?a=blob;f=Documentation/howto/maintain-git.txt	20:25
Dreamer3	thanks	20:26
	this could be interesting	20:26
	tritz: right we want to merge things... but i don't see how to merge form a branch into both master and stable	20:26
	you get tons of conflicts depending on what you branched from	20:26
	what i really want to merge to do is just apply the diff of my changes since i created the branch	20:26
	ie i branch at revision 12 (just to simplify)... i commit a few tiems... to revision 19	20:27
	i want to merge revisions 13-19 into both master and then barring any problems stable	20:27
tritz	Dreamer3: have you considered "git cherry-pick" to just select a single commit for merge?	20:28
	→ ToxicFrog joined	20:28
Dreamer3	does chery-picking prevent fast forwards?	20:32
	i'm not sure exactly what it does	20:32
	tritz: yes, but in my example i'd have to cherry-pick 6 times	20:32
tritz	Dreamer3: it merges a single commit into the current branch	20:32
	Dreamer3: i'd to it one by one	20:32
Dreamer3	yeah that sucks for long lists of commits :)	20:33
	so if i have master	20:33
	and stable	20:33
	and i cherry pick 3 of 20 commits... can stable still fast-forward merge with master?	20:33
	considering it now has 3 random commits in it	20:33
	→ icwiener_ joined	20:35
Dreamer3	If changes were made on only one of the branches since the last merge, they are simply replayed on your other branch (so-called fast-forward merge). If changes were made on both branches, they are merged intelligently (so-called three-way merge):	20:36
	o	20:36
	i'd assume it'd have to be a 3 way merge at that point	20:36
	which is why sometime's you'd see weird conflicts?	20:36
	→ dejanpan joined	20:37
	→ cbreak joined	20:37
tritz	Dreamer3: sorry, maybe someone else can help you there...	20:37
	stouset\|away → stouset	20:39
	pasky taught topgit remote branches	20:40
	→ sikanda joined	20:41
	← lea_w left	20:42
	← juanjo left	20:43
	→ flaguy joined	20:43
	← rtomayko left	20:48
	← icwiener left	20:50
sikanda	I'm getting a weird error when cloning a repo with submodules. I try to run "git submodule init" and "git submodule update" but get an "Unable to checkout" error saying the pathspec for the SHA doesn't match any files	20:51
	→ pillowfactory joined	20:52
sikanda	any ideas what's going on, no one else on the project gets this error, but we're all using the same version of git, built the same way (with macports)	20:52
	← priidu left	20:55
pillowfactory	How does git config work? I've got a my user config set with user.name = foo and user.email = [email@hidden.address] and I've got a repo with repo specific config where user.email = [email@hidden.address] Should I expect the [email@hidden.address] email to "override" the user config? When I issue git config --list from my repo it shows both user.email config entries, and when I push to github it is using the user-level email address of [email@hidden.address]	20:56
	→ dejanpan1 joined	20:58
	← dejanpan left	20:58
vuf	the repo config overrides the global config	20:58
	you add addresses when committing, not when pushing	20:58
sikanda	no takers on the submodule issue? or what might be contributing to it. we're all using v1.6.0.1	20:59
pillowfactory	ok... so when I push to github it says "(config.user.name) committed" instead of "(git username) committed" How can I get the latter?	21:00
	← spearce left	21:00
	← alley_cat left	21:02
	→ spearce joined	21:03
	← lorandi left	21:03
	→ peritus joined	21:03
jast	huh? what, in your opinion, is the difference between config.user.anem and git username?	21:04
	← cannonball left	21:05
	← spearce left	21:05
	← kukks left	21:05
	→ Snaury[away] joined	21:06
pillowfactory	jast - a little of my own nomenclature: config.user.name = git config --get user.name, git.username = my git username, pillowfactory	21:06
	← CodeOfficer left	21:06
	→ kukks joined	21:06
	← piotrb left	21:06
jast	uh. your github username or what?	21:06
pillowfactory	yes	21:07
	← joshbuddy left	21:07
jast	because git has no concept of a username apart from the user.name thing that gets used for author and committer names in commits	21:07
	that's a github-specific thing, i guess... some sort of mapping between your mail address and user account perhaps?	21:07
	i don't use github	21:07
	→ Ademan_ joined	21:07
	← Ademan left	21:08
	→ ajonat joined	21:08
pillowfactory	well... I just suspect that it's not using my repo-specific email addr and that's why it's defaulting to my name.	21:09
	→ rmh3093 joined	21:09
cardioid	pillowfactory: Add the e-mail used in those commits to your github account profile	21:10
pillowfactory	cardioid: Is it possible to use the repo-specific email addr in those commits instead of the global email?	21:11
cardioid	What?	21:11
vuf	you use addresses when committing, not when pushing	21:12
pillowfactory	right...	21:12
cardioid	The commit has the e-mail address used to commit. Your github account has one (or more) e-mail addresses associated with it	21:12
pillowfactory	let me make sure I'm saying this right	21:12
	← krh left	21:12
cardioid	When github sees a match, it uses your github username there and links to the account	21:12
	When github does not see a match, it just uses the name associated with the commit	21:12
	→ aroben joined	21:13
cardioid	If you want github to use your github username, you have to have commits with an e-mail address that is connected to your github account	21:13
	How you do that is up to you	21:13
pillowfactory	"git config --global --get user.email" = foo	21:13
	my-repo-root: "git config --get user.email" = bar	21:14
	when I commit to my-repo it's using foo as the email address in the commit; shouldn't it use bar since it's the repo-specific config?	21:15
	← dejanpan1 left	21:15
	→ tenpaiyomi joined	21:16
up_the_irons	pillowfactory: it should, do you have any environment variables that might be overriding that?	21:16
	pillowfactory: GIT_COMMIT_AUTHOR, etc...	21:16
vuf	pillowfactory: it should, and I just tested that it does	21:17
	→ Eridius joined	21:17
tenpaiyomi	Ok, I had some changes that I had made, and went to do a pull, but had changes. Thinking that I was on a different repo in which I had made previous, no longer needed changes, I stashed and dropped the changes. Later on, I realized that I was on the incorrect repo	21:17
	Is there a way to recover the changes?	21:17
	→ dejanpan joined	21:17
	← Beket left	21:17
	→ invalidrecord_ joined	21:18
	→ krh joined	21:18
vuf	tenpaiyomi: "dropped" how?	21:18
parasti	tenpaiyomi: git reflog stash, maybe?	21:18
tenpaiyomi	vuf: git stash drop	21:18
pillowfactory	up_the_iron: no I dont have any	21:18
parasti	tenpaiyomi: sorry, that should be git reflog show stash	21:18
vuf	tenpaiyomi: yeah, reflog then	21:19
	→ Signum joined	21:19
tenpaiyomi	Ok, it shows the stash, but how do I recover it?	21:19
neoeinstein	git stash apply	21:19
pillowfactory	up_the_irons: when you do a "git --config --list" in your repo does i list both email addresses?	21:20
	← invalidrecord left	21:20
	← glommer left	21:20
	← sikanda left	21:20
Signum	Q: What is the favorite way to allow pushing into a central repository? git-daemon doesn't seem to have security. I heard HTTP is unreliable and slow.	21:20
up_the_irons	pillowfactory: yea, it does	21:20
	← ph^ left	21:20
up_the_irons	pillowfactory: the repo specific one is lower on the list	21:21
vuf	Signum: ssh	21:21
pillowfactory	thanks up_the_irons. Mine does too, but I'm not sure why it's not working correctly	21:21
up_the_irons	pillowfactory: np	21:21
	i don't know either :)	21:21
pillowfactory	haha	21:21
	→ Beket joined	21:21
pillowfactory	not that big a deal, but one of those things...	21:22
Signum	vuf: Okay. So if I have several projects/repositories I'd set up distinct system users? Or is there a good proven working way to restrict pushing to certain repositories for certain contributors?	21:22
	vuf: Currently I'm just using git-daemon to publish everything under git.workaround.org.	21:22
	→ ph^ joined	21:22
	← ph^ left	21:23
tenpaiyomi	Awesome, thank you both	21:23
	→ rtomayko joined	21:23
	Signum just stumbled over gitosis and reads	21:24
	← trochala left	21:24
vuf	Signum: you use a custom shell, but I never tried it myself. Look up gitosis, I guess.	21:24
Signum	vuf: thanks	21:24
	I assume something like WebDAV isn't supported for pushing? I was used to use that with svn years ago.	21:25
	← invalidrecord_ left	21:25
Mikachu	supported is a strong word	21:26
	← chris2 left	21:27
	→ amf joined	21:27
Signum	Mikachu: Well, "git-push" would need to be able to know how to use WebDAV/PUT. :)	21:27
	Mikachu: And according to "man git-push" http is basically supported.	21:28
Mikachu	it does, but i don't think anyone would claim it works well	21:28
Gitbot	Mikachu: the git-push manpage can be found at http://git.or.cz/man/git-push	21:28
	← threeve left	21:28
Mikachu	heh	21:28
	→ invalidrecord joined	21:28
	← Bryson left	21:29
	→ _graham_ joined	21:29
mugwump	it works fine, web-dav is the broken part :)	21:30
	← davidfetter left	21:30
MaddHatter	has anyone encountered git push setting GIT_DIR to "." on the remote end?	21:30
vuf	MaddHatter: yes	21:30
mugwump	MaddHatter: that happens inside hooks/	21:30
MaddHatter	but apparently reset has a problem with that?	21:31
mugwump	right :)	21:31
	← bryanl left	21:31
mugwump	are you trying to make 'git-push' update a working copy? there's a script linked from teh FAQ	21:31
MaddHatter	so the solution is just to reset GIT_DIR to `pwd` or what?	21:32
	yeah, I've looked at the script	21:32
	← Ademan_ left	21:32
	← peritus left	21:33
	← foca left	21:34
	→ foca joined	21:35
	← _slacker_ left	21:35
	← JimmyB__ left	21:35
	← Eludias left	21:36
	→ qrush joined	21:36
	→ aroben_ joined	21:37
	← dec_ left	21:38
	← dejanpan left	21:38
	← bobesponja left	21:38
	→ dejanpan joined	21:39
vuf	MaddHatter: I had luck doing "unset GIT_DIR"	21:39
	cd .. ; unset GIT_DIR	21:39
MaddHatter	that's probably a better approach	21:39
	→ ph^ joined	21:39
MaddHatter	pwd can hang sometimes	21:40
	← girishr left	21:40
vuf	I think this is all just part of the test "don't push to non-bare repo unless you know what you are doing" :-)	21:40
	→ Bryson joined	21:40
MaddHatter	well, if git cvsexportcommit worked on Windows I wouldn't have to screw around with this :-/	21:41
vuf	cvs? ugh	21:41
	← tritz left	21:42
	→ lea_w joined	21:42
	→ admiralsausage joined	21:42
	← nuncanada left	21:43
	→ sexovivo joined	21:43
sexovivo	.www.prize.bg/sexsimace	21:43
	→ t joined	21:44
	→ mextli joined	21:45
	→ drewr joined	21:46
	← WarrenK left	21:47
	← bschindler left	21:47
	→ cilly joined	21:47
	→ WarrenK joined	21:47
	→ bdiego joined	21:47
	← sexovivo left	21:48
	→ stephenjudkins joined	21:48
	→ christel joined	21:49
stephenjudkins	i'd like to use 'git log' to show all commits after (datewise) a given commit. how do I do this?	21:49
	← mextli left	21:50
	← t left	21:50
vuf	stephenjudkins: git log --since=date	21:50
	→ nuncanada joined	21:51
	← pillowfactory left	21:51
	→ codeshot joined	21:52
	← kumbayo left	21:53
	← aroben left	21:53
	→ stephenjudkins_ joined	21:54
	← Beket left	21:54
	← stephenjudkins left	21:54
	→ stephenjudkins joined	21:55
admiralsausage	I sometimes like to put a small diff in my commit messages (eg to illustrate how a problem might be introduced, or whatever). I was surprised to find that git gets very confused by this - I had thought that commit messages would be "opaque". Are there guideline as to what you can & can't write in a commit message?	21:55
Tv	admiralsausage: confused how?	21:56
vuf	admiralsausage: for starters, don't swear :)	21:56
admiralsausage	Can't rebase a branch containing that commit - the reason given is that git can't find the file referenced in the diff within the message ...	21:56
	"fatal: sha1 information is lacking or useless (file.cpp)."	21:57
	Repository lacks necessary blobs to fall back on 3-way merge.	21:57
	If I change the commit message, the rebase succeeds	21:57
	← foca left	21:57
Tv	sounds kinda unlikely	21:57
	admiralsausage: can you provide a minimal failing example in a pastebin?	21:58
admiralsausage	Tv: sure	21:58
stephenjudkins	vuf: thank you	21:59
	← dejanpan left	21:59
	→ dejanpan joined	22:00
	→ doener joined	22:01
	← lea_w left	22:01
admiralsausage	If I use a commit message with the contents of http://pastebin.com/m6e2064fe then I get the error message in http://pastebin.com/m3d5e24f0 when trying to rebase that commit	22:01
	← krh left	22:01
	← SRabbelier left	22:02
admiralsausage	It seems (to a naive observer) like it's piping the output of git-show directly to patch, for example, and assuming that patch will strip out the comment	22:02
	I'm sure it's not that simple but it feels like something of the kind is going on	22:03
	it's easy enough to reproduce on a trivial new repository	22:03
	← vbabiy left	22:04
jast	something like that, yes	22:04
	i'm afraid that currently the only thing you can do is not put diffs into commit messages (why put one in there anyway?)	22:04
vuf	why not	22:05
	← rjent left	22:05
parasti	add a space to every line in the diff and git will be happy	22:05
	s/add/prepend/	22:05
	→ lea_w joined	22:05
admiralsausage	jast: I put one in a commit today as an example of a way to provoke a race condition that my patch was fixing. It was the first time I tried but I'm sure I could think of other reasons too.	22:06
jast	yeah, okay, of course parasti's suggestion solves this kind of thing	22:06
admiralsausage	But I will avoid it in the future. I just wondered if this was a case of "stupid user" or "bug"... maybe some of both	22:06
jast	perhaps more like a design limitation	22:06
admiralsausage	parasti: That's what I ended up doing! Thanks	22:06
parasti	I believe I've read about this on the mailing list, so it's probably a known problem	22:07
	← clairvy left	22:07
jast	hard to fix without breaking compatibility	22:07
vuf	admiralsausage: I don't think you're a stupid user for doing that	22:08
admiralsausage	parasti: even better, I guess I don't have to report it	22:08
	ok thanks folks	22:08
	← admiralsausage left	22:09
	← EmilMedve left	22:10
	← invalidrecord left	22:10
	← stephenjudkins_ left	22:10
	← doener_ left	22:12
	→ rjent joined	22:14
	← sbfaulkner left	22:14
	← Eridius left	22:14
	→ cliffstah joined	22:15
	← cliffstah left	22:15
	← GuyFromHell left	22:15
	← ferdy left	22:16
	← adamm left	22:17
	← zachinglis left	22:17
	→ sbfaulkner joined	22:17
	→ aroben joined	22:19
	← lolage0 left	22:19
	→ aroben__ joined	22:20
	← aroben_ left	22:20
	→ dejanpan1 joined	22:21
	← dejanpan left	22:21
	→ admiralsausage joined	22:21
	← admiralsausage left	22:22
	← _graham_ left	22:22
	→ sykopomp joined	22:23
sykopomp	is there a recommended source with a nice guide on how to use git when cooperating with several people with different branches, what good habits for using it comunally are, etc?	22:24
	← gebi left	22:24
	← bdiego left	22:25
	← ph^ left	22:26
	→ bdiego joined	22:26
	← jerbear left	22:26
	→ invalidrecord_ joined	22:27
mugwump	sykopomp: try http://www.kernel.org/pub/software/scm/git/docs/user-manual.html#sharing-development	22:27
	→ gebi joined	22:28
mugwump	might be a little basic	22:28
sykopomp	Thank you though, reading now.	22:29
	→ Eridius joined	22:31
	invalidrecord_ → invalidrecord	22:32
	← aroben left	22:35
	← dejanpan1 left	22:38
	← stouset left	22:38
	→ dejanpan joined	22:41
	← neoeinstein left	22:42
	← invalidrecord left	22:43
	→ kbingham joined	22:46
	→ madewokherd joined	22:46
	→ Ademan_ joined	22:51
	← rjent left	22:52
	← aroben__ left	22:52
	→ glommer joined	22:53
	→ aroben joined	22:54
	← sbfaulkner left	22:56
	← ehc left	22:57
	→ invalidrecord joined	22:59
	→ dejanpan1 joined	23:00
	← dejanpan left	23:00
	← __elihu__ left	23:01
	← mithro left	23:03
	← tenpaiyomi left	23:05
sykopomp	one more thing: I'm thinking of mirroring my main git repo on sourceforge (which only lets you choose between svn and cvs). Google isn't being helpful, sorry :(	23:05
	→ tenpaiyomi joined	23:05
	← LiamH left	23:05
	← peeja left	23:06
	→ davidfetter joined	23:06
	← jm left	23:06
	→ _bdiego joined	23:06
	← bdiego left	23:06
	_bdiego → bdiego	23:07
	→ bobesponja joined	23:08
	← cmarcelo left	23:09
pasky	[ANNOUNCE] TopGit-0.3	23:10
	← softdrink left	23:10
	← parasti left	23:10
Nido	sykopomp: i've heard of people using git to actually do merging on svn/cvs projects. I imagine it wouldn't be too hard to do something similar	23:10
	→ admiralsausage joined	23:12
Nido	git creates a directory .git to manage things (for as far as I can see with my limited understanding), you could just periodicly push your entire working repos onto sourceforge's subversion tree	23:12
Zao	sykopomp: There's ways of pulling from a svn repos into a local git one, and vice versa.	23:14
	http://www.kernel.org/pub/software/scm/git/docs/git-svn.html	23:14
mugwump	sykopomp: another idea, you can save to svn using git-svn set-tree	23:14
	or, just use a free git hosting service like github or repo :)	23:15
	← juanjoc left	23:15
mugwump	or gitosis	23:15
	← dejanpan1 left	23:17
j_engelh	gitosis sounds like a disease, from a medical standpoint ;-)	23:17
	← _kid_ left	23:17
admiralsausage	I'm using git-svn on top of my company's SVN server. It has branches in a few places, for example "branches/release4.0" and "branches/personal/pete/my_experiment" might both be copies of "trunk", and so I would like git to see them both as true branches. I'd like to tell git-svn that "anything copied from /trunk (directly or indirectly) is a branch". This seems more elegant than messing with globs (I can't express my s	23:18
	ituation with a glob!)	23:18
	→ _kid_ joined	23:18
	j_engelh subtracts 100 points from admiralsausage's account	23:18
admiralsausage	grr line lengths	23:18
	→ dejanpan joined	23:20
	→ xjjk joined	23:20
	→ Beket joined	23:21
	RandalSchwartz averts his eyes at the mention of git-svn again	23:21
	RandalSchwartz wonders when #git-svn will be created so we can declare this a "git-svn free zone"	23:21
MaddHatter	is there a quick way to git log --only-stuff-from-this-committer?	23:21
	oh there it is --author... nevermind :-)	23:22
RandalSchwartz	Yes - cleverly disguised	23:23
	← tenpaiyomi left	23:23
MaddHatter	but it's like halfway down the man page... surely you can't expect me to read _that_ _far_ can you?	23:23
	although it would be sorta nice if synopsis had more than just git-log <option>	23:24
	→ pigeon joined	23:27
j_engelh	ok	23:29
	let's try topgit	23:29
	→ ciaran_ joined	23:30
j_engelh	pasky: no homepage?	23:30
jast	admiralsausage, not supported, sorry	23:32
	→ Arbe joined	23:33
Arbe	http://oniushi.net/internet/olawl.php	23:33
	^figured you guys would need this	23:33
admiralsausage	jast: OK at least I can stop looking. Thanks again.	23:33
jast	Arbe, because we want to see endlessly repeating error messages? i don't think so.	23:34
	← Arbe left	23:34
	← ciaran left	23:34
	ciaran_ → ciaran	23:34
	→ sbfaulkner joined	23:35
	← pigeon left	23:36
	← loincloth left	23:37
	→ pigeon joined	23:37
	→ CodeOfficer joined	23:38
	← sbfaulkner left	23:38
	← seezer left	23:39
	→ jon___ joined	23:40
	← pigeon left	23:40
	← dejanpan left	23:40
	→ dejanpan joined	23:40
	← ebel left	23:40
	← qrush left	23:41
	← perezd88 left	23:41
	→ jackdempsey joined	23:42
	→ loincloth joined	23:43
	← Ryback_ left	23:44
	← loincloth left	23:45
	→ seezer joined	23:45
	→ pigeon joined	23:45
	← admiralsausage left	23:46
	→ ferdy joined	23:47
	← joydivider left	23:48
	→ joydivider joined	23:48
	← pigeon left	23:48
	← jon_ left	23:49
	→ pigeon joined	23:50
schacon	wanted to float this here just to get some initial feedback - what would you guys think about a C git shared library that just does basic object read/write, maybe some basic pack/server stuff?	23:51
	instead of trying to libify the existing C source	23:51
	or as a stopgap	23:51
	for scripted lang bindings and such	23:51
	stupid idea, or would others find that useful?	23:52
	← Ademan_ left	23:58

Logs Search ←Prev date Next date→

Channels Documentation