IRCloggy #git 2008-08-04

Provider of IRC logs since 2005.

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

	2008-08-04
doener	open(".git/index.lock", O_RDWR\|O_CREAT\|O_EXCL, 0666) = 3	00:00
	open(".git/index", O_RDONLY) = 5	00:00
	rename(".git/index.lock", ".git/index") = 0	00:00
	it does that	00:00
	← danten left	00:00
Mikachu	i wonder what vuf did then	00:00
	← d0k left	00:00
vuf	I wonder, too	00:00
doener	vuf: which command did you trace?	00:00
vuf	no rename here	00:01
	add	00:01
doener	well, you probably didn't really change the index	00:01
	that ends up with unlink() being called here as well	00:01
	→ Cory joined	00:01
vuf	aah, right ... I repeated the test a few times with the same file	00:02
	so it actually writes the index from scratch all the time ... didn't know that	00:03
Mikachu	you have to do that for any file that's not append-only	00:03
vuf	can't you seek and write?	00:04
Mikachu	not if you want to insert more data in a spot	00:04
	← Sonderblade left	00:04
	→ pjhyett_ joined	00:05
	→ mithro joined	00:06
	→ russells joined	00:07
	→ morii_ joined	00:07
vuf	so this was enlightening, but I am not sure it answered my question :)	00:07
Mikachu	you had a question?	00:08
vuf	the excl ensures the locking, I guess?	00:08
	→ frq1_dave joined	00:08
Mikachu	touch .git/index.lock; git add myfile	00:08
	→ rphillips joined	00:09
vuf	thanks ... I should go to sleep	00:10
	→ astubbs joined	00:10
	→ jamie_alm joined	00:12
	→ bryanl joined	00:13
	← jonesy left	00:14
	← peeja_ left	00:17
	→ lll joined	00:17
	morii_ → morii	00:18
	← bugfux left	00:21
	← pjhyett left	00:22
	→ bugfux joined	00:25
	← Kenjin left	00:25
	→ codeshot joined	00:25
	→ doener_ joined	00:28
	→ ben_h joined	00:33
	← ben_h left	00:34
	→ ben_h joined	00:34
	← codeshot left	00:35
	← rdavila_ left	00:38
	← zerogate left	00:38
	← zawir left	00:40
blix_	is there an easy way to tell if a directory is explicitly ignored by git?	00:41
johnw	git clean -d -x -n?	00:41
blix_	er, i would prefer not to nuke my working dir	00:42
johnw	that's the -n	00:42
	it just tells you if it's something being ignored	00:42
blix_	oh'	00:42
	ah, thats good then, thanks	00:42
johnw	oh, you want -X	00:43
	-x says to ignore ignored files	00:43
	→ iojdg joined	00:43
blix_	yeah, im reading through the git clean man page now, i never thought to look there	00:43
	thanks	00:43
johnw	sure thing	00:43
uau	i looked at speeding up 'git log -S' again	00:49
	the delta base cache behaviour looks interesting	00:49
	i added a more efficient hash implementation for it	00:49
	which does give much better results with somewhat larger cache	00:50
	but actually performs worse at 16m	00:50
	my current guess is that the somewhat "random" behaviour of the current hash implementation (only 256 hash slots, and no collision handling) might be beneficial when the size is limited	00:51
	if you have a long delta chain it's good to have a cached value from _some_ point of it	00:52
	← Jurko left	00:52
uau	so if you randomly drop hash values then with some luck (parameters that happen to fall into a good area) you might have SOME samples from more chains	00:54
	instead of all the intermediate values from the most recently used chains	00:54
	anyone have a better explanation?	00:54
	the speedup you can get is pretty significant	00:58
	stock git uses 132 seconds	00:59
	with better string search part in diffcore-pickaxe.c 70 seconds	00:59
	with another hash implementation for delta base cache in sha1-file.c the the actually increases to 80 seconds with default 16m delta base cache	01:00
	but then increasing the cache size drops it to 17 seconds	01:00
	→ weazen joined	01:00
uau	so you can get a factor of 7 speedup	01:01
	← ajonat left	01:01
	→ bryanl_ joined	01:02
	← bryanl_ left	01:02
	← bryanl left	01:02
johnw	i would suggest summarizing your results to the mailing list	01:03
	← frq1_dave left	01:06
bugfux	how can i show commits taht effected a directory?	01:12
RandalSchwartz	git log path/to/top	01:12
vmiklos	git log <directory>?	01:12
bugfux	oh that easy	01:12
RandalSchwartz	yeah that easy	01:12
	→ digitalhobbit joined	01:13
johnw	it's sad when people expect things to be hard	01:13
	← botanicus left	01:16
	→ morphir joined	01:17
	→ huyx joined	01:21
	← huyx left	01:22
	→ huyx_ joined	01:22
	← weazen left	01:23
	← pjhyett_ left	01:23
	→ weazen joined	01:23
	→ chrisb joined	01:24
	→ foca_ joined	01:25
	→ weazen_ joined	01:25
	← weazen left	01:25
	← weazen_ left	01:25
	→ jackdempsey joined	01:31
	→ crab joined	01:32
	← johnw left	01:39
	→ ggapol joined	01:41
	← foca left	01:42
	→ krawek joined	01:44
	← crab left	01:47
	→ lordscarlet joined	01:47
lordscarlet	can someone point me toward a resource that helps me figure out how to deal with merge conflicts/rejections?	01:48
	→ kewark joined	01:48
	→ jerbear joined	01:50
	← jamie_alm left	01:52
	← cbreak left	01:53
	→ LostInTheWorld joined	01:53
	← kukks left	01:54
	← bugfux left	01:54
	← foca_ left	01:54
	→ bdiego joined	01:56
	← Cory left	01:58
	→ johnw joined	02:02
	← krawek left	02:03
	→ isohead joined	02:03
jmspeex	lordscarlet: No idea about resources but generally, "git mergetool" solves the problem	02:04
isohead	Does "git push" not upload added files?	02:05
johnw	of course it does, if you've committed them	02:05
doener_	isohead: http://git.or.cz/gitwiki/GitFaq#head-b96f48bc9c925074be9f95c0fce69bcece5f6e73	02:05
	I guess that applies...	02:05
johnw	you're probably not using a bare repository	02:05
jmspeex	anyone knows how to cat a specific version of a file without checking it out?	02:06
doener_	git show revision:file	02:06
	→ jast joined	02:06
	→ mxey_ joined	02:07
	← lordscarlet left	02:08
	→ [dmp]_ joined	02:10
isohead	doener_: thanks, i didn't know what I was looking for. And didn't understand why status says added files were deleted. Well, that's solved now. I'll buy you a beer if I ever see you.	02:11
doener_	I'm just the bot to hand out the url, the one who wrote the faq entry deserves the beer ;-)	02:11
	← [dmp] left	02:12
	← jargan left	02:12
	→ FunkeeMonk joined	02:12
	→ garrin joined	02:12
gitster	heh, I don't drink beer ;-)	02:13
doener_	gitster: btw, I wondered if looking up the commit is really required. Shouldn't a plain "git reset --hard" work as well?	02:14
isohead	btw, is push the smart way to put stuff upstream?	02:15
	or is it the only way?	02:16
jmspeex	doener_: fatal: ambiguous argument '400d76e5e24618b:libspeex/quant_bands.c': unknown revision or path not in the working tree.	02:16
	I typed:	02:16
	git show 400d76e5e24618b:libspeex/quant_bands.c	02:16
doener_	jmspeex: and that revision actually had that file?	02:17
jmspeex	doener_: it seems like if I add -- and remove the :, it works	02:18
uau	jmspeex: i think that syntax would show any change made to the file in that revision	02:19
doener_	yep	02:19
	jmspeex: I'd bet that that file doesn't exist in that revision	02:19
jmspeex	doener_: it does exist (didn't in the path I wrote above, but fixing it doesn't change anything)	02:22
doener_	jmspeex: try "git ls-tree -r --name-only 400d76e5e24618b \| grep quant_bands" please	02:23
jmspeex	quant_bands.c	02:24
	quant_bands.h	02:24
doener_	and "git show 400d76e5e24618b:quant_bands.c" fails?	02:24
jmspeex	fatal: ambiguous argument '400d76e5e24618b:quant_bands.c': unknown revision or p	02:24
	Use '--' to separate paths from revisions	02:24
	Seems like I can get the files that are in the root directory, but not the others	02:26
doener_	jmspeex: (cd $(git rev-parse --show-cdup); git ls-tree -r --name-only HEAD \| grep quant_bands.s)	02:26
	and then use that path	02:27
jmspeex	If I do "git show 400d76e5e24618b:README", it works	02:27
doener_	the revision:filename syntax expects paths relative to the repo root	02:27
jmspeex	libcelt/quant_bands.c	02:27
doener_	then "git show 400d76e5e24618b:libcelt/quant_bands.c" should do	02:28
	oh lol I wrote .s above... interesting typo :-)	02:29
jmspeex	I figured that one	02:29
	Turns out I was typing .cc instead of .c	02:29
	Sorry about that	02:29
doener_	no problem	02:29
	← jerbear left	02:30
blix_	If i wanted to look up branches, remotes and tags without running git commands, would it be safe to just look under .git/refs/<whatever> or might there be refs listed elsewhere that arent under refs/?	02:30
doener_	blix_: .git/packed-refs	02:31
blix_	ah, ok	02:31
pasky	acpked-refs is invention from hell	02:31
blix_	out of curiosity, why are those not listed under .git/refs?	02:32
	pasky can't sleep :/	02:32
doener_	blix_: why have it twice? The point of packed-refs is to have a single file for many refs	02:32
	(although sometimes, there _are_ entries in .git/refs/ that also appear in packed-refs, don't ask me why)	02:33
blix_	doener_: in fact, all of mine are	02:33
doener_	pasky: too hot over at your place as well?	02:33
pasky	yeah, quite hot, but that's not quite it	02:33
	i can't tell why	02:33
RandalSchwartz	64 here in melon park	02:34
	menlo park too	02:34
	← spearce left	02:34
doener_	it's kinda funny, during the day, it's almost "cold" in here, and at night, it suddenly gets annoyingly warm	02:34
pasky	18 here	02:35
	;)	02:35
	today was really hot day in basel	02:35
doener_	sweating-while-dressed-in-shorts-and-tshirt here	02:35
pasky	not-sweating-but-cant-sleep-anyway	02:35
blix_	is the packed-refs file documented anywhere? seems simple enough, but i dont know what i dont know	02:36
	→ polpak joined	02:48
gitster	It is "packed" representation pack-refs (and callers like gc) creates, so that you do not have to readdir and open hundreds of files under refs/remotes/ and refs/tags/ that may rarely change. refs that exist under refs/ take precedence over it. If you have both, that means you haven't gc'ed lately.	02:50
blix_	ok, i need to handle both situations though	02:51
doener_	why do you want to avoid calling git?	02:52
gitster	everybody wants to reinvent a wheel, because it is fun.	02:52
doener_	seems I've gotten old... I actually enjoy not reinventing the wheel by now	02:53
gitster	it's not getting older, but getting wiser ;-)	02:53
	→ voodootikigod joined	02:54
doener_	I	02:56
	'm not so sure about that yet	02:56
	→ fhobia joined	02:57
	→ garrin_ joined	03:01
	→ bogolisk joined	03:02
bogolisk	Hi! what is the porcelain to show the sha1 of a tag? or should I just use git-rev-parse?	03:03
johnw	git show?	03:03
	← garrin left	03:04
	→ chieh joined	03:04
bogolisk	johnw: how to tell git show to restrict the output to just the sha1 of the commit?	03:04
	I guess I should just use git-rev-parse (for this purpose)	03:05
johnw	i think so	03:05
	that's kind of what it's there for	03:05
bogolisk	can some1 points to a article/text/email where Linus explained the choice of "content tracking" vs "change tracking"?	03:06
blix_	its not that i want to reinvent this particular wheel, its that I seem to save significant amounts of time by doing it manually rather than exec-ing	03:07
bogolisk	I've watched the google-talk but he went over that too quickly	03:07
johnw	did you really just say some1 to save two keystrokes?	03:07
blix_	dont asy me why but i trim about 20ms off my program time by doing it myself rather than fork/exec/read	03:07
johnw	20ms?	03:07
	← voodootikigod left	03:07
bogolisk	johnw: to save the length of the line!	03:08
johnw	i don't think some1 vs. someone even saves 20ms	03:08
bogolisk	johnw: I typed 60 per... hour! it does make a difference	03:08
	← chrisb left	03:08
blix_	was referring to parsing git-branch vs looking in .git/refs/heads + .git/packed-refs	03:09
johnw	do you need to do it 1,000,000 times or something?	03:10
	i don't even notice 20ms if I'm looking for it	03:10
blix_	no, but i do have to do it a few times if i am also looking for tags and remotes	03:10
	i am also on decent hardware, so i assume it would be worse for ppl on older hardware	03:11
johnw	why not look to see what git-branch does in the source	03:11
bogolisk	what's the plumbing to list the branches? git-rev-list --branches?	03:11
vmiklos	git for-each-ref?	03:12
bogolisk	vmiklos: thanx	03:12
vmiklos	np	03:12
blix_	for-each-ref and branch take virtually the same amount of time	03:13
pasky	gitster: (calling git to deal with refs is awfully slow)	03:13
	tg summary takes 1.5s calling all these rev-parses	03:14
	for my feeble 8 branches or so	03:14
bogolisk	what is the plumbing to see what's the current branch? (cat .git/HEAD?)	03:15
	← igorgue left	03:16
	kewark → krawek	03:16
vmiklos	git rev-parse?	03:16
	→ igorgue joined	03:16
	← CodeOfficer left	03:16
vmiklos	ah, no. git symbolic-ref	03:16
bogolisk	vmiklos: symbolic-ref is exactly what I need! thanx	03:17
	← bdiego left	03:18
vmiklos	it may worth to read the plumbing part of 'man git' before you ask ;)	03:18
bogolisk	is there a document describing the index other than the source code?	03:18
	← kanru left	03:19
	→ CodeOfficer joined	03:19
blix_	bogolisk: there is a git-cast that i found explained it really well	03:19
	http://gitcasts.com/posts/railsconf-git-talk	03:21
	its a long presentation, but its pretty well done	03:21
pasky	I have a suspicion that bogolisk wants more lowlevel description	03:21
	unless that's in the talk too	03:21
	and i think there's only the source code	03:22
	keep in mind that due to legacy reasons, git uses terms 'index', 'cache' and 'directory cache' interchangeably	03:23
	and actually the source code refers mostly to cache instead of index, due to unfortunate turn of events	03:23
blix_	yes, he goes into detail as to how the index file is composed	03:24
	in fact, that is a really good git internals presentation (IM limeted O)	03:25
bogolisk	blix_: does it talk about the differente "layers" (how are they called?) of the index	03:31
	the index is easy to understand except in a merge!	03:31
blix_	he just talked about the structure of the index, not how it was used during merging or anything	03:32
bogolisk	I'll take a look	03:32
jackdempsey	hey all......doing some deployment with capistrano and git	03:34
	← MK_FG left	03:34
jackdempsey	somehow i'm getting some permission denied errors	03:34
	things work ok when i run it manually on the deployment box	03:34
	anyone have a sec output is here	03:34
	http://pastie.org/246824	03:34
	→ ggapol_ joined	03:34
jackdempsey	the password is ok.....the place i'm deploying to is owned by the git user	03:34
	← LiamH left	03:34
jackdempsey	and again, when i run it on the box things are ok	03:35
	anyone know what this error often refers to?	03:35
	← ggapol_ left	03:35
	→ ggapol_ joined	03:35
	← ggapol_ left	03:35
jackdempsey	seems simple enough and yet i can't figure out what permissions aren't right....	03:35
	→ ggapol_ joined	03:36
	→ bugfux joined	03:39
	→ cedricv joined	03:39
	→ spearce joined	03:40
bogolisk	what is the 0 in "git show :0:/foo/bar.c" called?	03:43
spearce	the index stage. 0 is the normal stage that most files are in.	03:43
	stages 1-3 are used during a merge (1 = base, 2 = yours, 3 = theirs)	03:43
bogolisk	spearce: thanx!	03:43
johnw	cool, i never new about index stages	03:43
	← ggapol_ left	03:44
	→ ggapol_ joined	03:44
bogolisk	spearce: during the merge, 0 is the merge-result, right?!	03:44
spearce	yes	03:44
	← ggapol_ left	03:44
bogolisk	tk	03:44
	ty	03:44
spearce	0 and 1-3 are mutually exclusive; you cannot have both in the index at the same time	03:44
bogolisk	oh!	03:45
	→ ggapol_ joined	03:45
spearce	if a file is unmerged and has entries in stages 1-3, when you record the resolution at stage 0 the 1-3 entries are removed	03:45
	← ggapol_ left	03:45
bogolisk	spearce: you mean I can't do "git diff :{0,1}:/foo/bar/.c" ?!	03:45
spearce	yes	03:45
bogolisk	spearce: I think "git diff :{0,3}:/foo/bar.c" would be nice don't you think?	03:46
spearce	why not `git diff MERGE_HEAD foo/bar.c` then in that case?	03:47
	it gets you the same result; diff the file in the working directory (which should be the resolved file) against their file	03:47
	bogolisk googles for MERGE_HEAD... whtat's dat!	03:47
spearce	the index stages are sort of a black magic implementation detail that often is not the best thing to show to an end-user.	03:48
bogolisk	spearce: ok!	03:49
	← johnw left	03:49
	→ kanru joined	03:49
bogolisk	spearce: are there :4:,:5:,:6:, ....>?	03:49
spearce	no. the space in the index is only 2 bits wide. :)	03:49
	0, 1, 2, 3 ... and you better not use anything else. :)	03:50
bogolisk	spearce: ok! thanx	03:50
	← ggapol left	03:52
bogolisk	spearce: does :ours:/foo/bar.c works or it has to be :2:?	03:52
spearce	ours and theirs may be an alias in there, i don't remember if the patch for that was actually applied. it may have been rejected under the point that users probably should be using MERGE_HEAD and HEAD rather than the obtuse :1:, :2:.	03:53
	← chieh left	03:53
bogolisk	HEAD=:2:, MERGE_HEAD=:3:,????=:1:	03:54
spearce	1 is the merge base of HEAD and MERGE_HEAD; aka $(git merge-base HEAD MERGE_HEAD)	03:54
bogolisk	spearce: ok	03:54
spearce	but in certain types of crisscross merges the merge base of those two commits may be a virtual merge base that was generated on the fly by the merge program, and this is not actually available to you from git merge-base. but its result is stored in :1:	03:55
bogolisk	spearce: uh.... ok!	03:56
blix_	ok, so i did a quick bench on reading the refs manually vs getting from git branch and git branch -r	03:59
	with calls 20x: 0.132226, avg: 0.007	03:59
	without calls 20x: 0.016777, avg: 0.001	03:59
	so, not a bad improvement	03:59
spearce	blix_: what are you trying to do? and why is for-each-ref not doing it for you?	03:59
	→ imyousuf joined	04:00
blix_	spearce: i am making some very verbose command output for pyrite and i noticed that i was calling a lot of git commands	04:00
	→ foca joined	04:01
blix_	spearce: i felt that some things the overhead of calling git was too much for what it was doing	04:01
	→ ijcd joined	04:04
doener_	blix_: how much overhead does "git for-each-ref refs/heads refs/remotes" actually have? It's twice as fast as "git branch && git branch -r" here	04:05
	blix_: and even for some git-svn monstrosity with 200+ branches, it only takes 0.02 seconds	04:05
	← ijcd left	04:05
spearce	its also pretty quick to get data from if your scripting language has a decent eval system and git for-each-ref has an output quoting format that matches your language.	04:05
doener_	and given the name "py"rite, I guess that's the case	04:06
blix_	yeah, i thought so too, but i have gotten decent improvements by bypassing git calls.	04:07
	for instance i do things like get the git_dir and work_dir manually rather than call git, and now i am getting heads and remotes	04:08
	tags should be easy as well	04:08
	spearce hates it when he starts new projects while others are unfinished.	04:08
spearce	sigh damn git over http	04:08
doener_	blix_: you fail to detect bare repos	04:10
blix_	er, i do? it certainly possible	04:11
doener_	at least as of f157ecb	04:12
blix_	how are you testing it?	04:13
doener_	"pyt status" and "pyt diff" show all files as being removed	04:13
blix_	ah, i see	04:13
doener_	which is, well, a bit "misleading" in a bare repo	04:13
spearce	well, git-gui also doesn't like a bare repository, except in `git gui blame ...` mode	04:13
blix_	yeah, i havent done much testing in there.	04:14
	with bare repos i mean	04:14
doener_	yeah, but that complains instead of showing bogus about ;-)	04:14
	← bogolisk left	04:16
	→ voodootikigod joined	04:24
	spearce cries. he opened http-push.c in an editor.	04:25
doener_	hm, 2.5kloc, sweet	04:26
	→ tjafk2 joined	04:27
blix_	ok, i pushed something out	04:30
	← voodootikigod left	04:44
	→ davidfetter joined	04:49
	→ iojdg1 joined	04:54
	← statim left	04:59
	→ stubblechin joined	05:00
stubblechin	does anyone have any best practices for getting my git hooks under version control? some of them represent a significant amount of work, and there appears to be no way to make git add them. I get "fatal: unable to add .git/hooks/applypatch-msg to index"	05:02
	← iojdg left	05:03
stubblechin	i'm thinking about making a separate, outside repository in which i'll keep all my other repositories' git hooks. layout would look like: githooks/repo1.git/pre-commit, githooks/repo2.git/post-receive, etc...	05:04
aeruder	stubblechin: that'd work, or perhaps just a hooks branch in your current repo	05:04
stubblechin	aeruder: smart! i like it because it keeps the hooks with their repo	05:05
aeruder	but you can't really (well, as easily) setup a hook to update the hooks directory	05:06
	kind of got yourself a bit of a chicken and the egg problem	05:06
stubblechin	you mean, when i commit something in app.git/hooks, it gets copied into app.git/.git/hooks?	05:07
	good idea. i think that can be done easily. a manual copy from /hooks into .git/hooks the first time, and then i won't ever have to do it manually again	05:08
	← jackdempsey left	05:09
	foca → blah	05:11
	blah → foca	05:11
	→ iojdg joined	05:12
	→ VVSiz joined	05:12
stubblechin	how can i create a branch that has an empty working directory, instead of having all my project's files and folders in it?	05:15
Eridius	git symbolic-ref HEAD refs/heads/newbranch	05:16
	then you can blow away the index (rm .git/index) and commit	05:16
	and it'll create the new branch	05:16
	with no history	05:16
blix_	what is .git/info/refs, it looks a lot like .git/packed-refs	05:17
	← mithro left	05:19
	← sverrej left	05:19
Eridius	blix_: it's unpacked refs	05:19
spearce	actually its _all_ refs (except HEAD). and is there for the http clients to see what refs exist in the repository.	05:20
	← iojdg1 left	05:20
spearce	the http clients don't read packed-refs or refs/heads/*, they read only info/refs	05:20
	→ sverrej joined	05:20
blix_	I was asking earlier if there were any branches/remotes/tags not found under refs	05:20
	→ ggapol joined	05:21
blix_	is that file updated by a hook or should it be there always	05:21
	because that sounds simpler than rummaging though the refs tree	05:21
spearce	its usually updated by the post-update hook during push, and its usually only done on a repository that is published over http	05:21
	← huyx_ left	05:21
blix_	ah, ok	05:22
	i have no idea how i got one on my local repo then	05:22
	→ statim joined	05:23
	← spearce left	05:25
	← foca left	05:26
	→ foca joined	05:27
	→ mithro joined	05:29
morii	I have to work on a remote cvsroot. Localy I would like to use git, at least just to learn it. Is it a stupid idea to bring the files created by a `pserver checkout` under a local git control, where the master branch should be kept in sync with cvs update. All my changes will be done on git branches then? Cause I have to send patches instead off doing cvs commits, for me this sounds practicable. (git newnewbe:)	05:38
gitster	sensible.	05:39
	→ ajonat joined	05:40
morii	oh I missed at least one question mark :)	05:43
gitster	no, yes.	05:44
morii	Is that common way to do that?	05:44
gitster	that's for "is it a stupid idea?" and "my changes will be on branches?"	05:44
morii	:) thx	05:45
	→ mithro_ joined	05:45
gitster	I would not say it is common. Many people instead do cvsimport to slurp the CVS history into your local repository and then create cvs commits back using cvsexportcommit, but that is _not_ the only way. I've actually used what you described in real projects.	05:45
	→ crack_head joined	05:45
crack_head	I'm using git with svn and I've never created a patch...can someone enlighten me?	05:46
	← mithro left	05:47
	mithro_ → mithro	05:47
crack_head	I just want to make a patch from my changes that haven't been committed to the index	05:47
	→ meyering joined	05:48
	→ bts- joined	05:49
	← krawek left	05:50
	→ Aikawa joined	05:53
gitster	"committed to the index"???	05:53
	"git diff" will show the changes you have in the working tree relative to the index.	05:54
	"git diff HEAD" will show the change you have in the working tree relative to the current commit.	05:54
	← igorgue left	05:55
	← madewokherd left	05:55
morii	gitster: I'll would like to have the cvs history too. But that's a mozdev.org repro. I tried a git-cvsimport - no chance. The only way to get the whole cvsroot would be to fetch it from a Mercurial mirror. This one isn't updated as I need it. And I dono 'hg' and dono if it's possible to convert this 'hg clone' into git. Learning git is enough for me now.	05:56
	→ jonesy joined	05:57
	→ kanzure joined	05:58
kanzure	I have a developer on the other side of the room that has git cloned my repo, did some updates, committed, and now how do I pull, assuming he has no web server?	05:59
crack_head	gitster : so can I do 'git diff >> mydiff.patch' and send that to someone to apply?	06:03
morphir	can one not just e-mail from the git command line?	06:04
	kanzure mumbles something about http://www.kernel.org/pub/software/scm/git/docs/user-manual.html#setting-up-a-public-repository	06:05
russells	A TopGit question (for pasky...) Is there intended to be an equivalent to "stg commit" in topgit? Or an equivalent to "git checkout t/newstuff ; tg patch > ../patchfile ; git checkout master ; git am ../patchfile" ?	06:07
	→ deavid joined	06:11
	→ MK_FG joined	06:22
	→ jfdsjfods joined	06:23
	→ leifkb joined	06:23
	← jfdsjfods left	06:23
	→ ilogger2 joined	06:29
	→ Airwulf joined	06:29
Airwulf	hey	06:29
	→ priidu joined	06:29
Airwulf	In our lab we have got a library that every project uses. How can I hold this library in sync over many projects (git repositories) ?	06:32
vmiklos	submodules?	06:33
	http://www.kernel.org/pub/software/scm/git/docs/user-manual.html#submodules	06:33
	and man git-submodule	06:33
	→ bugfux joined	06:34
	→ selckin joined	06:35
Airwulf	vmiklos, cool that's exactly what I'm looking for. thx for the hint	06:36
vmiklos	np :)	06:36
	← bugfux left	06:41
	→ bugfux joined	06:41
	→ _zack joined	06:42
	→ Ademan joined	06:56
	→ rompus joined	06:58
	→ hein2 joined	06:58
	hein2 → Sho_	06:58
rompus	I'm new git so bare with me. I just finished a version of my software. I want to tag it as '1.0 Beta' but i don't want to copy all the files into the tag. How do I do this?	06:58
doener_	"copy files into the tag"? That makes no sense.	07:00
vmiklos	probably he used svn before git ;)	07:00
doener_	A tag just points to an object, usually a commit	07:00
rompus	i did use svn before git, vmiklos	07:00
	oh, i see, thanks doener_	07:01
	so what do i do if i want a folder to contain the files of the "1.0 Beta" version? just create the folder?	07:01
	then tag it?	07:01
vmiklos	make your changes, commit, then 'git tag yourversion', that's it.	07:02
	no need to copy anything.	07:02
	git tag -l will list your tags.	07:02
	→ ph^ joined	07:02
vmiklos	see man git-tag, you can make annotated tags as well (where you can specify an optional tag message)	07:03
rompus	vmiklos: what if i want a folder to contain the files for the people to grab (not via git, just via download)	07:03
doener_	rompus: look into "git archive"	07:03
vmiklos	something like git archive --format=tar --prefix=project-version/ version > project-version.tar or so	07:04
	where 'version' is a tag you previously created.	07:04
rompus	awesome!	07:05
	now can i tag multiple commits?	07:05
MK_FG	git archive --format=tar --prefix=project-version/ version > project-version.tar \| bzip2 > project-version.tar.bz2 should do even better	07:05
doener_	s/> project-version.tar//	07:05
	← ph^ left	07:06
MK_FG	Yeah, right, sorry	07:06
rompus	okay, so i can tag files or commits?	07:07
MK_FG	Commits	07:07
doener_	commits, trees, blobs	07:07
MK_FG	(like 'set of files' frozen in time :)	07:07
rompus	what if some of the files are from way older commits?	07:07
vmiklos	but in most cases, you just want to tag commits :)	07:07
	→ boink joined	07:08
vmiklos	rompus: http://eagain.net/articles/git-for-computer-scientists/	07:08
	boink → raz	07:08
rompus	for instance, i uploaded a directory 5 commits back, 2 files 2 commits back and 1 files this last comment. I want all them tagged	07:08
vmiklos	ah	07:09
	man git-tag? :]	07:09
rompus	tried, did that first before coming here ;)	07:09
vmiklos	you can do git tag myolderversion HEAD~10 and so.	07:09
	if you want to mark HEAD~10 (see man git-rev-parse) as myoldversion	07:09
doener_	rompus: you cannot do mixed tags like in svn, if that is your concern	07:09
rompus	doener_: naw.. i just want to tag just a list of files, a tree, i guess, but not everything in the last commit	07:10
	if someone's willing to help me the first time, i think i could get it after that	07:11
	sorry to be difficult	07:11
doener_	you _could_ tag a tree, that is essentially a directory in your repo, but you generally do not want to do that. And having a tag that points to a random collection of blobs is impossible	07:12
rompus	http://github.com/karbassi/nerdstream/tree/master/client/trunk is where my files lie. I want to tag 5 files and 1 directory.	07:12
	→ ph^_ joined	07:12
rompus	then i want to do archive, like you all suggested	07:12
vmiklos	uh :)	07:13
	that is not a proper git svn import	07:13
doener_	to be frank and honest, that sounds pretty broken	07:13
rompus	clean me up then.. i'm pretty lost	07:13
doener_	why would you want to tag only a part of it to begin with?	07:14
	git repos are supposed to hold one project, nothing more, nothing less	07:14
rompus	true, but some of the files are for testing	07:15
	→ boink joined	07:15
doener_	and tagging only part of a project sounds broken, people would miss part of the project	07:15
rompus	i tried to clean it up and separate the project into sections but i still failed	07:15
pasky	russells: yes, tg collapse, I plan to implement it soon	07:15
rompus	there is a server and client side	07:15
	=\.. guess i'll continue in the broken way	07:16
morii	gitster: got git-cvsimport to work :) I made a mistake using '-r pserver:balh' instead od just -d ... Thanks a lot!	07:16
rompus	thanks for trying guys/gals.	07:17
	← bugfux left	07:17
russells	pasky: thanks	07:21
	← Ademan left	07:22
	→ Ademan joined	07:23
	→ GarethAdams joined	07:25
	→ iojdg joined	07:26
	→ tvw joined	07:29
	← ph^_ left	07:32
	← GarethAdams left	07:33
astubbs	i'm trying to rebase a branch against master, and I keep getting the error "could not detach HEAD"	07:35
	any suggestions?	07:35
polpak	detaching ones head sounds painful	07:36
astubbs	lol	07:37
	← raz left	07:37
MK_FG	rofl	07:37
polpak	though I suppose it would only hurt for a few moments	07:37
doener_	astubbs: you have files that would be overwritten	07:37
astubbs	argh	07:37
	windows has issues with git	07:37
	it keeps showing funny errors like file not up to date, and claiming i have dirty state	07:38
	when status shows nothing	07:38
	and gitk lists uncomiited changes with empty diffs	07:38
	i've just done a hard reset ad i still get the same error	07:38
MK_FG	Erm.. try linux?)	07:39
astubbs	i wish	07:39
doener_	astubbs: _untracked_ files, to be more specific	07:39
astubbs	like they say - "change your company, or change your company" right?	07:39
	→ reel_ joined	07:39
astubbs	but git status shows nothing	07:39
doener_	you might have those in your .gitignore or .git/info/exlude	07:40
astubbs	i just did git clean -f and then tried again - same thing	07:40
doener_	+/- typos	07:40
	astubbs: "git clean -nx"	07:40
astubbs	nope, same thing	07:41
doener_	that just lists the files it would remove, it doesn't remove them	07:41
	You could have killed me for telling you to "git clean -fx"...	07:42
astubbs	i took out the n	07:42
	← rompus left	07:42
	→ ph^ joined	07:42
astubbs	git clean -nxd comes up blank	07:42
	→ charon joined	07:42
doener_	then I'm inclined to just say "Windows" :-/	07:43
astubbs	doh	07:43
doener_	though I cannot remember having heard something similar	07:43
astubbs	there needs to be an option for more verbosity in cases like this	07:45
	astubbs misses java exception stack traces	07:45
doener_	actually, that case got more verboseness lately, should be in 1.6.0rcX IIRC	07:45
	→ botanicus joined	07:46
astubbs	oh well at least that's good news	07:46
	i could do a merge	07:46
	maybe	07:46
	but i just went through git-svn hell	07:46
	learnt lot's about HEAD's and git itnernals though :)	07:46
	← boink left	07:47
	→ boink joined	07:47
	boink → raz	07:47
	→ ben_h joined	07:49
	← _zack left	07:50
astubbs	hmm	07:50
	i think it may have something to do with case sensitivity	07:50
	on git-merge master, it complains of "untracked working tree file Sms.java" would be overwritten by merge. Now - what happeed was at one point i renamed that from SMS.java to Sms.java ... in the currently checked out branch is is SMS.java	07:51
doener_	yeah, that sounds like it is the problem. Good job Redmond	07:53
astubbs	argh	07:53
	how about i commit, in the current branch, a rename to SMS_redmond.java	07:53
	→ GarethAdams joined	07:53
	→ BabelO joined	07:53
astubbs	then try the rebase?	07:54
	→ nud joined	07:54
doener_	that might do the trick, or just back to Sms.java	07:54
	although I wonder if that even works... maybe with git mv	07:54
	anyway, you can just drop that "bonus" commit after the rebase	07:55
astubbs	bingo	07:55
	← iojdg left	07:56
	→ hobodave joined	07:58
hobodave	hello	07:58
	Is it possible to define a local user.name and user.email for a git repo?	07:58
	I tried using git config --local but it doesn't seem to work	07:58
astubbs	anyone got any good recommendations for git compatible windows merge too?	07:59
	~~	07:59
	omgosh how on earth do you use vimdiff??	08:01
tango_	vimdiff firstfile secondfile	08:02
	then you move up and down as usual	08:02
	→ ciaran joined	08:02
tango_	switch pane with ctrl+w h/l	08:02
	and get/put diff with dg/dp	08:02
	← GarethAdams left	08:02
	→ Terminus joined	08:03
astubbs	whats dg dp?	08:04
	→ mcgrof joined	08:05
tango_	astubbs, the key combinations to get and put diffs. do you know how vim works?	08:05
astubbs	i'm a serious vim noob	08:06
hobodave	#vim	08:06
astubbs	yeah	08:06
tango_	astubbs, you should get comfortable with vim before going to vimdiff 8-D	08:06
	→ peritus_ joined	08:06
astubbs	how do you quit save and quit vimdiff all at once? :wq only does one pane at a time.	08:07
	tango_, i don't have a choice - have to resolve this now :)	08:07
tango_	astubbs, :wqa does it all	08:07
	write quit all 8-D	08:07
astubbs	yup got that	08:08
	← peritus_ left	08:08
astubbs	mergetool finished, now i got git-rebase --continue, and it's complaning "you must edit all merge conflicts... " doh	08:09
tango_	grep -R ===== *	08:09
	then git add the changed files	08:09
	→ GarethAdams joined	08:10
morii	tango_: btw, are you using vimdiff for gitk too? If you'r using gitk at all	08:11
astubbs	git status shows the 2 files that i edited using git-mergetool under changes to be committed, and only one other file - the .dotest/ dir as untracked. nothing else	08:11
tango_	morii, I don't use gitk	08:11
morii	k	08:12
	dwmw2_gone → dwmw2	08:12
tango_	but I use vim (and thus vimdiff) for most my editing task	08:12
	astubbs, sorry no idea	08:12
morii	me too	08:12
	→ _slacker_ joined	08:14
	← Terminus left	08:16
	→ \ask joined	08:20
	→ iojdg joined	08:23
	→ anathematic joined	08:24
	→ bcarlyon\|laptop joined	08:27
	→ mcella joined	08:30
	→ whee joined	08:31
	→ crab joined	08:41
crab	pasky?	08:42
pasky	just on my way out	08:43
	what's up?	08:43
crab	i'll tag git.pm patches in future, if i ever send any. but is there some way i can know which areas need tags and which not?	08:43
pasky	ah, that's you :)	08:43
	dunno, I tag all my patches in all areas	08:44
	can't see which areas would not need tags :)	08:44
	it's also good to thread your patches with the mails they actualy reply to	08:44
crab	so like [PATCH] builtin-foo.c: xyz?	08:44
	my patch did have in-reply-to: the bug report	08:45
pasky	oops	08:45
	how could i not notice that	08:45
	then it's all entirely my fault, sorry :)	08:45
	will you have a look at the git-svn breakage?	08:45
	i didn't actually test git-svn	08:45
crab	pasky: yes, i'm looking at it now	08:46
crack_head	is there an easy way to find the history of a particular directory?	08:46
pasky	'k	08:46
	crab: git log dir	08:46
	afk	08:46
	→ saidin joined	08:46
crab	s/crab/crack_head/	08:46
	← jds left	08:48
	→ Catfish joined	08:48
crack_head	so if I do 'git log --stat dir' ...that will show me what files changed...but is there a way to actually see the diffs?	08:49
charon	log -p	08:50
	you can also combine -p --stat	08:50
	→ peritus_ joined	08:51
	← saidin left	08:51
	→ aurax joined	08:52
	→ clairvy joined	08:54
	← tvw left	08:59
	→ vuf joined	09:02
	→ sbahra joined	09:04
	← peritus_ left	09:05
	← crab left	09:06
	← clairvy left	09:06
	→ saidin joined	09:08
	← GarethAdams left	09:08
	→ peritus_ joined	09:09
	← peritus_ left	09:14
	→ clairvy joined	09:16
	← aurax left	09:17
	← iojdg left	09:17
	→ iojdg1 joined	09:17
	→ peti joined	09:17
	→ tvw joined	09:18
	← saidin left	09:18
peti	Hi. My copy of gitweb.cgi writes the error message "key does not contain a section: -z" into the apache logfile every time the web site is accessed. Does anyone have a clue where that error might be coming from?	09:19
	← tvw left	09:19
pasky	whoa	09:20
	hmm, seems like git config error maybe?	09:20
	I assume no line numbers?	09:20
peti	pasky: No line number, unfortunately. This is all I get.	09:20
	→ _zack joined	09:21
pasky	hmm	09:22
	ah	09:22
	my guess: your gitweb.cgi is newer than your git binary	09:22
peti	pasky: Yes, that's possible.	09:22
pasky	specifically, your git is older or equal to v1.5.2.2	09:23
	sorry, you need to upgrade git or downgrade gitweb.cgi :)	09:23
	or revert b201927ac8a4544baa067f5c5c7c7f2d6780e29a in your gitweb.cgi	09:24
Gitbot	[git b201927ac]: http://tinyurl.com/6j66ky -- gitweb: Read repo config using 'git config -z -l'	09:24
peti	pasky: I have git v1.5.2.4. gitweb.cgi comes from the current HEAD. I upgraded because I hoped that might fix the problem. :-)	09:24
pasky	that's weird	09:24
	b201927ac8a4544baa067f5c5c7c7f2d6780e29a	09:24
	er	09:24
	pasky slaps his clipboard	09:24
pasky	pasky@pixie:~/src/girocco/gitweb> git describe 2275d502114c71045af991697048191fed88aac4	09:24
Gitbot	[git 2275d5021]: http://tinyurl.com/5rlgmv -- config: Add --null/-z option for null-delimted output	09:24
pasky	v1.5.2.2-580-g2275d50	09:24
telmich	when I am in a merge-conflict, is there a command that displays only the unmerged files? so I can use them in a script.	09:24
peti	pasky: I'll upgrade git and see what happens. First of all, thanks a lot for your help!	09:25
pasky	np	09:25
	telmich: git ls-files --unmerged	09:25
telmich	pasky: cool, thanks!	09:25
	→ saidin joined	09:31
	← saidin left	09:35
	→ russells joined	09:36
	→ chris2 joined	09:40
	→ saidin joined	09:43
	→ Jurko joined	09:43
	→ adamt joined	09:45
adamt	hi	09:45
pasky	ho	09:46
adamt	looking at the manpage of git-log, i can't make out how to use custom formats -- i get as far as git-log --pretty, and then i'm stuck. --pretty=oneline and such works, but how am i supposed to insert my own format instead?	09:47
pasky	--pretty=format:... ?	09:47
adamt	thanks	09:47
	writing the format without "format:" first obviously didn't work :-)	09:48
	→ MarcWeber joined	09:51
MarcWeber	Is there an easy way making gitk show the whole file touched by a patch?	09:51
	:-) sure there is : lines of context = 99999..	09:52
pasky	lol	09:52
	evil	09:52
	← saidin left	09:53
MarcWeber	But it's not suited for copy paste :-(	09:53
telmich	describe 5	09:54
	→ sverrej joined	09:55
charon	MarcWeber: if you select "tree" in the right-hand pane, and pick the file from there (manually, unfortunately) you'll see the contents	10:03
	← Ademan left	10:03
	← iojdg1 left	10:04
MarcWeber	charon manually is some clicks too much :) qgit has a nice feature for this. Maybe I find the time to write a patch	10:09
	→ danten joined	10:09
	← anathematic left	10:10
	→ d0k joined	10:16
	← adamt left	10:20
pasky	so i'm in windows explorer inside a git repository	10:22
	er, sorry, git-controlled working copy	10:23
	what's the canonical way to open git gui on the working copy?	10:23
	→ fonseca joined	10:23
vuf	pasky: get out of the explorer :)	10:24
pasky	and?	10:24
	← _zack left	10:25
vuf	don't know, but that's always a good start	10:25
dsop	pasky: right click on the directory tree and then GIT GUI (Here) ?	10:26
pasky	on what directory tree?	10:26
	I'm _in_ the directory tree	10:26
	→ saidin joined	10:26
pasky	why would I want to run git gui on a tree I'm not in?	10:26
vuf	don't you have a tree view?	10:27
pasky	what is a tree view?	10:27
dsop	well i talk about the directory tree on the left side of the explorer, not sure how it is called in tec terms	10:27
pasky	I have file and folder tasks, other places and details tabs on the left side	10:27
	well, "tabs"	10:27
	oh	10:28
	pasky clicked on the Folders button	10:28
pasky	ok	10:28
	that might fly	10:28
	though I stlil think the design is dubious	10:28
doener_	that of windows? no doubt	10:29
eMBee	good afternoon	10:29
pasky	no	10:29
	that of git-gui msysgit integration	10:29
eMBee	does anyone have a tested script for git-bisect run that applies a patch, runs a test and then reverts?	10:30
	→ iojdg joined	10:32
	← saidin left	10:33
pasky	so, I try to push via git ugui	10:34
	and get stuck on "Store key in cahche? (y/n"	10:34
	what now? :)	10:34
	and why the hell does it use plink? :/	10:34
	I selected openssh during installation	10:34
	how can I fix that up?	10:34
	→ saidin joined	10:35
	← nud left	10:36
	← saidin left	10:38
pasky	zomg	10:42
	msysgit is putting this in registry	10:43
	BLEH	10:43
	→ grahal joined	10:44
	→ professionalnerd joined	10:44
	→ saidin joined	10:45
	→ dwave joined	10:48
morii	e.g git merge --help will result in 'man git-merge' I suppose. Is it possible to teach git using 'pinfo' instead of 'man'?	10:49
	→ cbreak joined	10:50
doener_	morii: does "git help --info merge" work?	10:52
	hm, probably only if "info" is "pinfo"	10:54
morii	nope this will biring up pure 'info' - but I'll must have a look for env-vars then. Thanks!	10:54
doener_	no, "info" is hardcoded	10:54
morii	argh!	10:54
	then i have to pinfo git-merge :)	10:55
	+use	10:55
	← reel_ left	10:59
	→ d0k_ joined	10:59
doener_	morii: could you try "git config --global man.pinfo.cmd pinfo && git config --global man.viewer pinfo"	10:59
	morii: that's not exactly clean, but the code paths seem to be similar enough for this to work	11:00
	morii: or maybe with man.pinfo.cmd set to 'pinfo gitman'. I'm pretty clueless about the whole info stuff	11:01
	→ astubbs joined	11:02
	← is_null left	11:04
	→ nud joined	11:06
	← d0k left	11:07
	→ icwiener joined	11:11
morii	doener_:"git config --global man.pinfo.cmd pinfo && git config --global man.viewer pinfo"	11:11
	13:00 doener_: morii: that's not exactly clean, but the code paths seem to be similar enough for this to work	11:11
	does it!	11:11
doener_	:-)	11:12
morii	its starnge, setting an alias didn't helped at all	11:12
	s/starnge/strange/	11:12
doener_	shell alias?	11:13
morii	yep	11:13
	← saidin left	11:13
morii	i fi did a normal 'man something' pinfo was exec - but not for git	11:13
	→ joekarma joined	11:14
doener_	yeah, because git execs man, not your shell	11:14
	and git naturally doesn't care about aliases set in your shell	11:14
	→ reel_ joined	11:15
	→ lu_zero joined	11:20
	→ morii_ joined	11:20
vuf	if I set up a central repo that everybody pushes to, and have it "git reset --hard" in post-receive, that should give me a file system view of the current HEAD	11:23
	but won't locking kill that plan, if two people push at the same time?	11:23
	or, rather, I think that I need locking in the hook as well?	11:27
telmich	I am wondering, why I don't have a commit that reverts an issue, though I see it in git-log	11:27
doener_	telmich: hm?	11:28
telmich	I am talking about http://unix.schottelius.org/git/foreign/linux-2.6	11:28
	the commit 5577e9a124deeb9a4336fcaf347465cfd1ed251b reverted a change, which I merged in f610c0cb5f48e3ead74703b9a082eb9d099278c1	11:29
doener_	is that available via git://?	11:29
telmich	I am now on 744c7deb5645b36b295b9e6c3326fd012be2272e, which merged other stuff from linux-next and the checked out files do not have the reverted commit applied	11:30
	doener_: no, sorry, only http	11:30
	git:// will come on the new server	11:30
	maybe it is a problem, because I merged from a rebased tree	11:31
doener_	hm, ls-remote doesn't show anything for that url	11:31
	morii_ → morii	11:32
doener_	and fetching doesn't work either	11:32
telmich	hmm, it should; gitweb is below cgi-bin/gitweb.cgi	11:32
	/cgi-bin/gitweb.cgi even	11:33
doener_	and clone comes with the infamous "remote HEAD refers to nonexistent ref"	11:33
telmich	strane	11:33
	strange	11:33
doener_	telmich: but yeah, the rebasing might be guilty, you have the commit that introduces the changes twice	11:36
	03bac96fae0efdb25e2059e5accbe4f3ee6328dd and 0571c5d20aca71c735222132b02aebddf593045c	11:37
	→ crab joined	11:37
crab	pasky?	11:37
telmich	doener_: hmm, can I somehow detect that I've a commit twice when merging from a rebased branch?	11:38
doener_	"git cherry" or "git log branch1...branch2 --left-right --cherry-pick"	11:39
	(rev-list instead of log also works)	11:39
	hm, the "branch1...branch2" part should be at the end... I need to get some sleep	11:40
	← professionalnerd left	11:40
	→ professionalnerd joined	11:40
crab	pasky: well, i fixed the test failure too.	11:41
	→ saidin joined	11:42
	← saidin left	11:43
telmich	is there already a shorter way for: git-checkout -b temp origin/master; git-merge -s ours; git-checkout master; git-merge temp; ?	11:52
	(like git-merge theirs?)	11:53
	-s even	11:53
	← crab left	11:53
	→ MK_FG joined	11:55
	→ mcella_ joined	11:56
	→ advany joined	11:57
	→ janm joined	12:00
	→ snitko joined	12:00
	← professionalnerd left	12:01
	→ professionalnerd joined	12:01
	← mcella left	12:02
	→ warlock_mza joined	12:03
	→ glauber_ joined	12:03
	→ Zoup joined	12:10
niv	win 12	12:13
whee	how can I get .git/info/exclude to match based on complete paths?	12:17
	e.g., I want to exclude the "foo/" directory at top-level, but not bar/baz/foo/	12:18
Pieter	/foo	12:18
	ie, start with a slash	12:18
whee	that doesn't appear to work :\|	12:18
Mikachu	git version?	12:19
whee	git version 1.6.0.rc1.64.g61192	12:19
	→ domg472 joined	12:19
whee	bah, probably a cygwin problem	12:20
	→ saidin joined	12:20
whee	works fine in bsd	12:20
	or I'm stupid and not realizing I'm missing one part of the path :D	12:23
	← hobodave left	12:25
	→ cilly joined	12:26
	→ Lemurnomicon joined	12:26
	← professionalnerd left	12:27
	→ professionalnerd joined	12:27
	→ giom joined	12:30
	← saidin left	12:30
whee	semi-related: as I'm building a gitignore, is there anything that will tell me if I'm ignoring tracked files?	12:33
	I know the gitignore won't affect tracked files, but that case usually means my patterns match too much	12:34
vuf	I think not. There were a few question the last few days about people not realizing that they already tracked the ignored files.	12:35
Mikachu	you can compare git ls-files with git ls-files -i -X .gitignore	12:36
whee	0% git ls-files -i -X ../.git/info/exclude	12:37
	fatal: cannot use ../.git/info/exclude as an exclude file	12:37
Mikachu	hrm	12:37
whee	if I use .git/info/exclude, it works, but that's not relative to PWD	12:37
Mikachu	actually, git ls-files -i only shows ignored tracked files in the first place	12:37
	→ deavid joined	12:37
	→ radarek joined	12:38
vuf	Mikachu: I should just wait for your answer rather than going and investigating :). But yeah, that's what I found as well.	12:38
Mikachu	heh	12:39
whee	would you expect -X to be relative to the current directory or the repository root?	12:39
	that seems odd :\|	12:39
Mikachu	whee: try using the absolute path	12:39
	i know you can use ~/.gitignore	12:40
telmich	hmm, strange, why is that git-repo not valid	12:40
	http://unix.schottelius.org/cgi-bin/gitweb.cgi?p=foreign/linux-2.6;a=summary looks good	12:40
whee	Mikachu: that works too	12:40
telmich	and I can push to it without problems	12:40
vuf	-X ../.gitignore does not work	12:42
	→ vbabiy joined	12:42
whee	vuf: it doesn't like relative paths for me	12:42
	→ cmarcelo joined	12:42
Mikachu	$PWD/../.gitignore should be fine	12:42
	← deavid left	12:43
	→ deavid joined	12:43
Mikachu	% strace git ls-files -i -X ../.gitignore\|&grep chdir	12:44
	chdir("..") = 0	12:44
	open("../.gitignore", O_RDONLY\|O_LARGEFILE) = -1 ENOENT (No such file or directory)	12:44
	← astubbs left	12:44
Mikachu	so i'd say it's relative to the repo root	12:44
	you could also use --exclude-standard	12:45
	i just didn't bother to read so much in the man page earlier :)	12:45
whee	neat	12:46
	thanks	12:46
	→ tjafk joined	12:47
	→ bdowning_ joined	12:50
	→ _bdowning joined	12:55
	← glauber_ left	12:55
	← bdowning_ left	12:56
	→ madewokherd joined	12:58
	← warlock_mza left	12:58
	→ glommer joined	13:00
	← _bdowning left	13:02
	→ bdowning joined	13:02
	→ sergiodj joined	13:05
	← nud left	13:05
	→ bogolisk joined	13:07
	→ plxotl joined	13:08
	→ EmilMedve joined	13:08
bogolisk	I've googled and RTFMed but just cannot figure out the simplest way to print the full path of the root directory	13:11
	pwd \| sed 's=^$.*$'"$(git ls-files --full-name foo_bar.c)"'$=\1=g'	13:11
	is it the right way	13:11
Mikachu	bogolisk: you mean like git rev-parse --git-dir ?	13:12
	→ zawir joined	13:12
Mikachu	there's also --show-cdup and --show-prefix	13:12
bogolisk	Mikachu: huh....I think so! thanx!	13:13
	Mikachu: thanx, that's exactly what I'm looking for!	13:15
Mikachu	yw	13:15
	→ LiamH joined	13:17
	→ urkud joined	13:18
	→ mw joined	13:19
	→ Beket joined	13:24
	→ krh joined	13:33
	← Zoup left	13:37
	→ lamont joined	13:37
	→ nud joined	13:41
	→ earcar joined	13:42
	→ Ryback_ joined	13:42
	→ ysis joined	13:43
	→ danp joined	13:45
	→ bongiojp joined	13:47
danp	hi	13:47
	→ markkalderon joined	13:48
	← Ryback_ left	13:48
danp	i'm trying to track down where in git history a file in a CentOS SRPM came from. i generated its sha1 with git-hash-object and i can see it wih 'git show <sha1>'. how can i see what commit involved that id?	13:48
	→ Ryback_ joined	13:48
	← Sho_ left	13:50
	→ swombat joined	13:50
swombat	how do I undo the last commit (and return all the files contained therein to their "modified but not saved" state)?	13:50
Pieter	git reset HEAD^	13:51
swombat	er i mean "changed but not updated"	13:51
	← cilly left	13:51
swombat	Pieter: i don't want to lose those changes	13:51
	← urkud left	13:51
swombat	just uncommit them	13:51
	won't git reset HEAD just discard all the changes?	13:51
	also, I only want to undo the last commit - not all the ones before	13:52
	(which haven't been pushed yet)	13:52
Mikachu	swombat: git reset HEAD^ won't discard changes, git reset --hard HEAD^ will	13:52
Pieter	git reset HEAD^ will remove last commit, but keeps the changes in yoru working copy	13:52
	→ Sho_ joined	13:52
swombat	ah cool	13:52
	ok	13:52
Pieter	if you do git reset --soft HEAD^, the index will be kept the same too	13:52
swombat	thanks :-)	13:52
	ok, that worked	13:53
	thank you very much :-)	13:53
	→ Xyri0 joined	13:53
	→ cilly joined	13:54
	← earcar left	13:55
vuf	when would you actually use --soft ?	13:55
pasky	danp: I'd git log --raw --abbrev=40 concerned_filename	13:56
	danp: and then just search for your id	13:56
Mikachu	vuf: before git-commit --amend was available	13:57
Pieter	vuf: if you want your index to stay the same.. eg git reset --soft HEAD^ && git commit --amend -C HEAD will do a squash of your last two commits	13:57
pasky	will it?	13:58
	← ph^ left	13:58
pasky	git reset will change your HEAD too, won't it?	13:58
danp	pasky: that did it. thanks!	13:58
Pieter	yeah	13:58
pasky	danp: welcome	13:58
	→ mabs29 joined	13:59
pasky	Pieter: so you'd git commit --amend -C HEAD@{1} instead?	13:59
Pieter	pasky: nah, with -C HEAD you keep the first commit message	14:00
	which is often what I want ;)	14:00
pasky	uh	14:00
Mikachu	you would most likely want -c and change something?	14:00
pasky	which is first?	14:00
	oh nice, -c is shorthand for -e -C	14:00
	didn't know that :)	14:00
Mikachu	the second commit is probably just "squash this into 3234cd"	14:00
pasky	uh	14:00
	well	14:00
	you mean you want to keep HEAD^'s commit message?	14:01
Pieter	yeah	14:01
pasky	then why not just git commit --amend?	14:01
Pieter	if you have commit A and then B with A as a parent, and you want to squash them to keep A's message	14:01
	because you had already committed?	14:01
	I dunno :)	14:01
pasky	so?	14:01
	--amend won't magically reset commit message of what you are amending	14:01
Mikachu	ah yes, the default is -c HEAD	14:01
pasky	(by itself)	14:01
	→ foca joined	14:01
Mikachu	but Pieter wanted to suppress the editor i guess?	14:01
Pieter	Mikachu: right	14:01
pasky	</dev/null ? ;)))	14:02
ciaran	I use git commit --amend -C HEAD a lot	14:02
Mikachu	i have that aliased	14:02
	→ chrisb joined	14:02
Mikachu	and git commit --amend is just git amend too :)	14:02
Pieter	does -C have a default?	14:02
pasky	hmm	14:02
	pasky never wants to avoid the editor that badly	14:03
Mikachu	error: switch `C' requires a value	14:03
	i do git symbolic-ref h HEAD in some repos :P	14:03
pasky	gee :)	14:03
	you lazy typers	14:03
	Xyri0 → Xyrio	14:03
Pieter	Mikachu: might be a nice patch for that	14:04
	→ xasd joined	14:05
	→ m00l joined	14:06
vuf	I think that my conclusion is that I don't want reset --soft :)	14:06
	← xasd left	14:07
telmich	hmm, is there a better way to find out commits that are only in the other branch named $cherry than that:	14:07
	git-cherry $cherry \| grep -v '^+ '	14:07
pasky	is this one not good enough? ;)	14:08
telmich	pasky: well, I thought there's something already included in git ;-)	14:08
	→ pflanze joined	14:08
	→ imyousuf joined	14:08
telmich	pasky: and it seems to be pretty slow (doing that for all Next/Trees right now)	14:09
pasky	well, git cherry :)	14:09
	but I don't think I've actually ever used it myself	14:09
Mikachu	telmich: git log a...b --cherry-pick?	14:09
pasky	so can't advise more	14:09
Mikachu	telmich: --left-right	14:09
	s/git log/gitk/ is also nice	14:10
pasky	I always have to laugh when I see the --left-right option	14:10
	it so much embodies what all is wrong with git CLI	14:10
	and is hilarious at the same time	14:10
Mikachu	i would be happy if you could select only one of them, instead of marking both	14:10
pasky	well maybe it's not such a good example of CLI wrongness... but it's hilarious anyway :)	14:11
Mikachu	i'm not sure why it's funny ;)	14:11
peti	pasky: A git update remedied the 'does not contain a section: -z' error. Thanks again for the hint.	14:12
pflanze	Hello. It seems git (v1.5.6.2) doesn't accept multiple --reference options with "clone"? Only the last one ends up being in objects/info/alternates. But I think multiple alternates are ok. So, what to do?	14:12
	→ kukks joined	14:13
Pieter	hmm	14:13
	← selckin left	14:13
pasky	meneither, actually	14:13
	but it is	14:13
	(to me, anyway)	14:13
	pasky gives knetworkmanager very annoyed look	14:13
pasky	it could have at least some no-detach option so I could run it in an infinite loop	14:13
Pieter	if you specify -C to have an optional argument you run into problems	14:13
	← peti left	14:13
Pieter	because then "git commit --amend -C HEAD" will be expanded to "git commit --amend -CHEAD -- HEAD"	14:13
pflanze	pasky: fghack from daemontools might work (I've never tried it)	14:14
pasky		14:14
	daemontools are evilinherently evil	14:15
	ok, I give up	14:15
	I guess this wifi connection wants me to do some actual local work :)	14:15
imyousuf	I have a rather strange scenario - I have 2 git repositories and for one reason or the other now i need to merge them. For example lets name them A and B and lets suppose B's history and content will now have to be part of A. Now I can add it as a remote repo and merge it and it will work fine since Objects_A and Objects_B are mutually exclusive. Now what I want to do is merge B into A but at a path, e.g. path_2_b/ of A. Is there any way I can do it? The bad	14:15
Mikachu	imyousuf: your message got cut off, but you want to read documentation/howto/using-subtree-merge.txt	14:16
	<HEAD:Documentation/howto/using-subtree-merge.txt> maybe works	14:16
Gitbot	Mikachu: I'm sorry, there's no such object: HEAD:Documentation/howto/using-subtree-merge.txt.	14:16
Mikachu	doh	14:16
pasky	I always get it wrong too	14:16
Pieter	howtos?	14:16
Mikachu	hah! merge-subtree	14:16
imyousuf	Mikachu: thanks, thats a great direction :)	14:17
Mikachu	<HEAD:Documentation/howto/using-merge-subtree.txt>	14:17
Gitbot	[git using-merge-subtree.txt]: http://tinyurl.com/6o88by [blob]	14:17
Pieter	I can add some keywords for often needed docs	14:17
	like this, and the push -f, and push to bare repo faq things	14:17
pasky	would be cool	14:18
Pieter	something that responds to "Gitbot: keyword?"? shouldn't be too intrusive	14:18
	→ hobodave joined	14:18
	← dwave left	14:19
pasky	I'd make it [keyword] anywhere on the line instead	14:19
	so I can say imyousuf: [subtree]	14:19
Pieter	ok.. we should be sparse with the keywords then, otherwise we'll get spammend ;)	14:19
pasky	and you could create git wiki links from unknown keywords (after basic sanity check), maybe?	14:19
	I say be liberal and tone it down when it turns out to be too annoying	14:20
imyousuf	thanks pasky ;) (assuming you are the bot)	14:20
pasky	:))	14:20
imyousuf	in fact there is a similar bot in #spring	14:20
	Mikachu adds git log --only-left, mohaha	14:21
Pieter	isn't that just a..b \| sed s/^/</ ? :)	14:21
Mikachu	not with git-log or gitk	14:22
	and specifically, not with --cherry-pick	14:22
	pasky shakes head silently	14:22
pasky	this is so wrong :))	14:23
vuf	Gitbot: tell me about the push to bare repo faq thing please	14:23
Mikachu	pasky: that was not silently ;)	14:23
pasky	ok ok	14:23
	pasky goes back to girocco hacking	14:23
grahal	is there a config variable to enable -s on commit by default?	14:23
Mikachu	git config --global alias.sommit 'commit -s' :)	14:24
	→ ph^ joined	14:24
pasky	how many aliases do you have, mikachu?	14:24
Pieter	mostly you add a signoff when exporting patches or so	14:24
grahal	ok thanks	14:24
	pasky is just in chatty mood and can't help it	14:24
Pieter	(I did create a patch to add a signoff configoption to format-patch :))	14:24
Mikachu	pasky: 25	14:24
bremner	pasky: get back to work on topgit	14:24
pasky	uf	14:24
Mikachu	my latest is vount-objects for count-objects -v ;)	14:25
pasky	bremner: I'll do that later, I'm at work now :)	14:25
	and they don't pay me for topgit	14:25
	sadly ;)	14:25
bremner	pasky: they pay you for IRC though :-)	14:25
pasky	um	14:25
	I ask things here that I need for my work!	14:25
	;)	14:25
	like the sad state of git-gui on windows	14:26
telmich	Mikachu: thanks, that git-log version is "MUCH" faster	14:26
Pieter	what's wrong with git-gui on windows?	14:26
pasky	(the saddest part is that I'll have to fix that)	14:26
	Pieter: you can't run it on your current folder directly and ssh pushing just doesn't work without some kind of ssh agent	14:26
telmich	Mikachu: I am using it now, to find out things missing in linux-next: http://unix.schottelius.org/cgi-bin/gitweb.cgi?p=nsbin;a=blob;f=linux/cherry-branches.sh;h=15006458d51d049f6453cb8f9459a068baf8f0dc;hb=3801d72e51a1c72dd10bac6c715b532ed8e17a4d	14:27
	→ rdavila joined	14:27
Pieter	hmm, that isn't very nice	14:27
pasky	(actually, the latter is problem on linux too)	14:28
	(besides the fact that git-gui still misses a lot of features... I got really depressed when I realized there's no choice to me than to code in my programming language nemesis TCL)	14:28
imyousuf	Mikachu: thanks once again for the direction, but there is a small problem :) and that is once I perform the steps mentioned in (http://tinyurl.com/6o88by) the doc I get the same content in the root directory and in the --prefix specified directory. Is that normal?	14:29
Mikachu	imyousuf: no idea, i've never used it ;)	14:29
	← ftehw left	14:30
	→ ftehw joined	14:30
imyousuf	Mikachu: I got the answer I skipped the -s ours thats why it was landing in the root as well :)	14:31
Mikachu	okay	14:31
	→ Weasel[DK] joined	14:32
	← priidu left	14:33
	→ rgk joined	14:34
	← Beket left	14:35
	→ swombat_ joined	14:37
	→ kiba joined	14:40
kiba	hello	14:40
	how do you excatly move a directory to the upper level?	14:40
	← advany left	14:40
	→ kanru joined	14:40
kiba	for example, I want to move the content of said directory trunk to upper directory	14:41
	instead of blah/trunk..it will be blah	14:41
imyousuf	kiba: mv the directory to wherever you want and then	14:41
	← m00l left	14:41
imyousuf	do - git rm old-path-to-dir/	14:41
	→ m00l joined	14:41
imyousuf	git add new-path-dir/	14:41
	and git commit	14:41
	git will detect it as a rename of objects	14:42
kiba	?	14:42
	← ysis left	14:42
imyousuf	mv trunk /* ./	14:43
	git rm -rf trunk/	14:43
	git add .	14:43
	git commit	14:43
	← sbahra left	14:43
Mikachu	kiba: git mv blah/trunk trunk	14:43
kiba	mikachu: um..I don't want it to move to trunk	14:44
	I want to get rid of trunk and move it to blah	14:44
Mikachu	kiba: surely you can change the name in the command yourself?	14:45
imyousuf	:-D	14:45
	→ kanzure joined	14:45
kiba	I cannot move trunk up :/	14:45
	→ Cacheaway joined	14:45
kiba	mv: `trunk' and `./trunk' are the same file	14:46
	why it said that, imyousuf?	14:46
	← danp left	14:46
imyousuf	kiba: in the command there is a space after trunk by mistake	14:46
	mv trunk/* ./	14:47
	its late here got to get going, catch you all tomorrow, thanks to all for the help, it was a fruitful day :)	14:47
	BTW Mikachu was referring to - git mv blah/trunk blah	14:47
	thanks Mikachu I did not know the existence of the git mv command :)	14:48
	→ warlock_mza joined	14:48
	→ Beket joined	14:48
	← imyousuf left	14:48
	← LiamH left	14:49
	← swombat left	14:52
	← Cacheaway left	14:57
	→ saidin joined	15:00
telmich	how can I list remote branches from git://git.linux-nfs.org/~bfields/linux.git without importing them?	15:03
	→ fy___ joined	15:04
jdq	maybe git branch -r	15:05
	← saidin left	15:06
telmich	no, I am just fetching one branch, due to git-remote add -t name ...	15:06
	← fy___ left	15:06
	→ glasser joined	15:11
	← glasser left	15:12
	→ aroben joined	15:12
	→ saidin joined	15:13
	← kanzure left	15:13
	→ kanzure joined	15:15
	→ JD joined	15:17
	← saidin left	15:19
	← Ryback_ left	15:21
	→ UVSoft joined	15:22
	→ saidin joined	15:25
	← snitko left	15:26
	← saidin left	15:26
	← Airwulf left	15:29
Pieter	telmich: git peek-remote	15:29
UVSoft	hi	15:30
	→ bugfux joined	15:30
UVSoft	me and my friend have a shared repo, both of us cloned the repo, i made a commit, created a bundle (git-bundle), and sent it to the friend via email (because at that moment i didnt have direct access to the repo). when he received my bundle, he already had made several commits. than he made some more commits and decided to fetch my commit from the bundle and push it to the repo. my question is what will happen to the log, what will be the order of commits a	15:31
	nd how cant i get my friend's commits that were made before my commit and after (on the time line)?	15:31
	← m00l left	15:32
	→ m00l joined	15:33
telmich	Pieter: KISS	15:33
	→ Ryback_ joined	15:34
telmich	UVSoft: he can also create a bundle with git-bundle; or if now possible, you can pull from his shared dir	15:35
	UVSoft: the order is non-linear, you have to branches that develop in parallel; tell your friend to use gitk to display them	15:35
UVSoft	telmish, when i receive his commits, my log will be different from his one?	15:37
	sorry, telmich	15:37
telmich	UVSoft: yes; you'll have a different merge commit and different order	15:37
UVSoft	will it make any problems when i come back and be able to push/pull from our shared repo directly?	15:39
	→ jerbear joined	15:39
Pieter	it'll work itself out	15:41
	he can send you his new commits by creating a bundle of yourlastcommitsha..HEAD and send that to you	15:41
	→ davi__ joined	15:41
Pieter	that'll include his commits before he added yours and also the commits he added after yours	15:41
	but not those you made yourself, thus saving space	15:41
vuf	telmich: I don't understand about the different merge commit?	15:41
	→ advany joined	15:44
UVSoft	Pieter, so he must create to bundles: before and after my commit?	15:44
	davi__ → davi	15:44
Pieter	UVSoft: no, just after	15:45
	it'll combine all his commits and send them to you	15:45
UVSoft	or he just cant create one big bundle and git on my side will skip existing commit (my one)	15:45
	how git will know that i already have that comimt?	15:45
	→ jackdempsey joined	15:45
Pieter	because he uses the "git bundle yourhead..HEAD" syntax	15:45
	which tells bundle not to include any commits you have	15:46
	→ catdude joined	15:46
	→ swombat joined	15:47
vuf	then the history will look the same, no?	15:47
UVSoft	suppose this log looks like: ...<commits before i left> <his commits> <my commit> <his commits>, what will be `yourhead`	15:48
	?	15:48
Pieter	UVSoft: he can do "git bundle show-heads" to get your last commit sha	15:48
	UVSoft: he should use that	15:48
	UVSoft: or he can manually look up the <my commit> part in his history, and use that	15:48
telmich	vuf: well, if I do git-merge abc and the other one does git-merge dcf the merge commit contains the name of the other branch -> other commit	15:48
vuf	telmich: why wouldn't you move the abc merge back along with the rest of the history?	15:49
	→ sbahra joined	15:49
	→ rdavila_ joined	15:50
	metze → metze_away	15:50
rgk	when i try to clone the git repo on my server it gives me an odd error time out error, any ideas? git clone git://madgizmo.org/git/MGE	15:50
telmich	vuf: you also got other commits, you did not have before	15:50
	← rdavila_ left	15:50
Pieter	rgk: anything in your logs?	15:51
	rgk: I get an unexpected hangup	15:51
UVSoft	Pieter, but my last commit will be the first at his side, and `yourhead`..HEAD will be empty, wont it?	15:51
rgk	ill check the logs right now	15:51
UVSoft	*not the first, but the last	15:51
Pieter	UVSoft: you said yourself: <his commits> <my commit> <his commits>,	15:52
rgk	http://paste2.org/p/55306	15:52
	it says its not exported?	15:52
UVSoft	so he will omit first <his commits>	15:52
Pieter	UVSoft: no, because he has done a merge commit	15:52
	UVSoft: so you can't look at it like something linear	15:52
	UVSoft: I can draw you a graph if you want	15:53
UVSoft	Pieter, it would be great)	15:53
	telmich gives Pieter a fresh version of dia ;-)	15:53
	← charon left	15:56
	→ Zdra joined	15:56
Pieter	UVSoft: http://pastie.textmate.org/private/ic2e42fvs9pkmf59desfsw	15:57
Zdra	Hi, is there a simple command to do that kind of thing:	15:57
	for remote in `git remote`; do echo "prune $remote";git remote prune $remote;done	15:57
	→ onitunes joined	15:57
Pieter	Zdra: no.. you can add an alias for that though	15:58
onitunes	I moved to a new machine, installed git from the tarball, and now normal git commands always "fill the screen"	15:58
	I'm on Mac OS X	15:58
	10.5.4	15:58
vuf	"you get the same history as he has" wohoo	15:58
onitunes	How can I get back the previous behavior where it just printed some lines, without taking over the erminal	15:58
	erminal/terminal	15:58
	my terminal is bash	15:58
	← reel_ left	15:59
	← JD left	15:59
Pieter	onitunes: do you have some custom LESS settings or so?	15:59
vuf	onitunes: what does "echo $LESS, $PAGER, $GIT_PAGER" say?	15:59
onitunes	not that I know of; it's a completely stock, new MacBook Pro	15:59
	, ,	15:59
	that's what it printed...	15:59
vuf	ok	16:00
Zdra	Pieter: ok, thanks	16:00
rgk	Pieter: i just added the file git-daemon-export-ok to .git	16:00
	can you clone it?	16:00
Pieter	rgk: yes, works fine	16:00
Zdra	"git remote prune" should default to prune everything	16:00
rgk	Pieter: nice thanks	16:01
Pieter	Zdra: I'm not sure, that sounds like a change some people might not expect	16:01
vuf	onitunes: I don't know OS X, so I am not going to stab further	16:01
Pieter	Zdra: you can send it to the mailinglist though, see what people think	16:01
	→ digitalhobbit joined	16:01
	← swombat_ left	16:02
	← zawir left	16:02
onitunes	vuf: k, thanks...	16:02
	← bugfux left	16:02
onitunes	so the consensus is that this is specific to my setup and not a change in git?	16:02
	for instance, it used to never use pager,IIRC	16:02
	now it does...	16:03
	→ bts- joined	16:03
Pieter	onitunes: it's new that status uses a pager	16:03
	onitunes: but most other commands used a pager already	16:03
onitunes	how can I turn it off	16:03
	?	16:03
Pieter	export GIT_PAGER=cat	16:03
onitunes	will that turn it off everywhere?	16:04
Pieter	yeah	16:04
onitunes	I do like it, e.g. in git log	16:04
vuf	yes, but that won't help if you want a pager for more than a screenful	16:04
onitunes	git diff, etc.	16:04
	okay, well, where in git can I change	16:04
	it?	16:04
	git-status?	16:04
	I just need to revert it	16:04
	I run git status all the time	16:04
Pieter	I'm not sure, in git.c itself perhaps	16:05
onitunes	really? that blows...	16:05
	okay, I'll look	16:05
Pieter	anyway, I'm not seeing your problems on the same setup	16:05
	perhaps you can try a snapshot and see if it works on that?	16:05
onitunes	git status isn't using a pager on your machine?	16:05
Pieter	right	16:05
	→ npgoog joined	16:05
vuf	are you sure that you don't have a funny less?	16:05
	→ snitko joined	16:06
	← ben_h left	16:06
onitunes	it's the stock Mac OS X less at /usr/bin/less	16:06
vuf	git uses less -F which should not page less than a screenful	16:06
	→ amfwrk joined	16:06
	dwmw2 → dwmw2_gone	16:06
onitunes	right, it doesn't page less than a screenful, but it does move all the text to the bottom of my terminal	16:07
	← rdavila left	16:07
onitunes	so I have to commands-k to clean up my terminal again...	16:07
vuf	oh, think I completely misunderstood	16:07
onitunes	I just want it to print the status to stdout, without any gimmicks	16:08
	→ tarbo joined	16:08
	onitunes thinks that sounds reasonable :-)	16:08
onitunes	as it is, i end up with terminal-height sized blank lines in my terminal logs....	16:09
	every time I do git status, which is often since I switch between lots of projects....	16:09
	→ aksyn joined	16:13
aksyn	greetings	16:13
	→ bartman joined	16:13
aksyn	new to git, can someone give me a hand understanding how to svn import a branch (only) from a big subversion project?	16:14
	i want to git svn import the 1.4 branch of asterisk, so that i can work on it locally.. but i'd rather not clone the whole repository (as it'll take days/weeks) and can't get fetch -rXXX to actually do anything useful	16:14
	git svn clone http://svn.digium.com/svn/asterisk/branches/1.4 -r135457	16:15
	that's the main route i've been going down.. which seems to init the repository, but neither that, nor a fetch -r135457 return anything	16:15
	(and git doesn't seem to be very verbose in its output... is there a debug switch i'm not aware of?)	16:15
UVSoft	Pieter, thanks	16:16
	→ raven79 joined	16:18
raven79	hi all	16:18
	← nud left	16:19
	→ tvachon joined	16:20
	→ dotsintacks joined	16:23
	← snitko left	16:23
	← bcarlyon\|laptop left	16:25
onitunes	Pieter: is the pager stuff newer than the git 1.5.6.4 release?	16:28
	in git.c, I see a --no-pager flag, but that doesn't seem to work...	16:28
	I was hoping I could alias git status to git status --no-pager	16:28
	→ vbgunz joined	16:29
	← sverrej left	16:30
	← kanzure left	16:31
	← kiba left	16:32
	→ crab joined	16:33
	→ bennyk_ joined	16:33
	→ kanzure joined	16:34
	→ ddollar joined	16:35
bennyk_	Hi, do you gues know of any good replacement for trac (tickets, timeline, wiki etc) with good git support? I don't really like trac and maybe there is any other systems like it that you could recommend?	16:35
Pieter	onitunes: there were some complaints about the pager after 1.5.6.. not sure what happened with that though	16:36
onitunes	Pieter: git --no-pager status	16:37
	seems to work...	16:37
Mikachu	it's been restored to no paging in 1.6.0 iirc	16:37
	and configurable per command maybe even	16:37
	→ DiscoMur1 joined	16:38
	→ agib joined	16:38
onitunes	anyway to use the alias file in the latest edge release to make git status do git --no-pager status?	16:38
	anyway/any way	16:38
	→ charon joined	16:38
onitunes	VICTORY!	16:39
	[alias]	16:39
	st = !git --no-pager status	16:39
vuf	congrats	16:40
onitunes	status = !git --no-pager status	16:40
	also works...	16:40
	:-)	16:40
	onitunes will update the git wiki...	16:40
cardioid	Is status the only command where this is a problem?	16:40
Mikachu	onitunes: you can't override existing commands with aliases, i guess you discovered that already	16:40
onitunes	Mikachu: hmm, it works for me	16:41
Mikachu	i mean you can't name the alias "status"	16:41
onitunes	hmm	16:41
	well, the pager isn't running....	16:41
	so I'm assuming it's using my alias	16:41
Mikachu	if you name it "st" it works of course	16:41
onitunes	(I'm on the latest edge, just a couple of minutes ago, release)	16:41
	hmm	16:42
	I think you're right	16:42
	it doesn't work	16:42
	what does work is that they fixed the status command	16:42
	...even better...	16:42
	:-)	16:42
	→ LostInTheWorld joined	16:42
onitunes	wait, spoke too soon :/	16:42
Mikachu	it's fairly easy to do this too http://comm.it.cx/?p=git.git;a=commitdiff;h=d1a745e235	16:42
onitunes	Mikachu: you're right, can't overload built in alias....	16:43
	"st" it is...	16:43
	→ timocratic joined	16:43
	→ ajonat joined	16:49
	→ krawek joined	16:49
	→ bennyk joined	16:49
Pieter	onitunes: for your info, that's the default in the upcoming git	16:49
onitunes	I guess I'm not the only one then :-D	16:50
	→ bobesponja joined	16:50
onitunes	thx for the help Pieter; looking at git.c got me a solution that works today	16:50
Pieter	(no rc2 last night?)	16:50
	→ dancor joined	16:52
dancor	what is the easiest way to dole out repo commit access	16:52
	← onitunes left	16:52
Pieter	dancor: gitosis	16:52
	→ rwaite joined	16:53
	→ GarethAdams joined	16:55
	→ kumbayo joined	16:56
cardioid	gitosis is the sweetness	16:57
Pieter	what does --boundary really do?	16:58
Mikachu	it doesn't do what the documentation says? :)	16:58
Pieter	I don't know where I should find doc for it	16:59
	other than --boundary	16:59
	Output uninteresting commits at the boundary, which are usually not shown.	16:59
	which I don't really get :(	16:59
Mikachu	there's a graph under the --left-right option (sorry pasky)	16:59
	→ fhobia joined	16:59
Mikachu	i'm not sure it helps either :)	16:59
	← warlock_mza left	17:00
	→ bdiego joined	17:00
Pieter	I was trying to figure out what git-bundle does	17:01
	it writes the --boundary commits as a header to the bundle file, so I wondered why	17:01
	← aksyn left	17:03
pasky	:))	17:04
	← bennyk_ left	17:04
crab	hi pasky.	17:05
	→ freaky[t] joined	17:06
	→ ijcd joined	17:10
	← cilly left	17:11
	← icwiener left	17:11
	→ swombat_ joined	17:13
	← swombat left	17:13
	→ urkud joined	17:15
	← tarbo left	17:18
	→ svend joined	17:20
	← bennyk left	17:20
	→ bennyk joined	17:21
	→ rmh3093 joined	17:22
crab	justatheory: er, what? \df+ will already show you the source for non-C functions	17:22
	urk	17:23
	← crab left	17:23
	→ zawir joined	17:23
	← urkud left	17:23
Pieter	pasky: http://gist.github.com/3930	17:23
	(git/pieter.git)	17:23
	→ spearce joined	17:23
	→ statim joined	17:24
zawir	spearce: have you already seen git annuncement?	17:24
spearce	yea, i just read it	17:24
Mikachu	javagit?	17:24
spearce	yea	17:25
	their faq is interesting. "How is this different from egit/jgit?" "Who/What?!"	17:25
zawir	spearce: yeah, i can understand their need for fast java api. although it seems sad that they haven't tried to contribute to jgit.. (?)	17:25
Pieter	link?	17:26
spearce	www.java.git.com	17:26
Pieter	java.git?	17:26
spearce	err, www.javagit.com	17:26
	sorry, one too many . in the linky	17:26
Pieter	ah, I'll read that after dinnen	17:26
	interesting	17:26
	(dinner(	17:26
	→ advany_ joined	17:27
	← advany left	17:28
dmilith	anyone using git with netbeans?	17:28
spearce	probably. nobody has started a plugin for it that i know of, though several people have asked about using jgit under netbeans. we want that to happen. we just haven't seen anyone produce netbeans code to call it. :)	17:28
zawir	;)	17:29
dmilith	You're wrong	17:29
	there's one.	17:29
spearce	yay!	17:30
dmilith	nbgit	17:31
	→ Aikawa joined	17:31
zawir	dmilith: it seems to be abandonned/inactive project	17:33
	dmilith: using similar approach to javagit...	17:33
dmilith	javagit is capable to be used in nb?	17:34
	zawir: it's active.. but sloooow	17:34
	it's developed by one person	17:34
	iisp	17:34
cardioid	I know Tv isn't here right now, but I was hoping someone else could help me understand something with gitosis (and maybe git in general)	17:34
	→ Daenyth\|Work joined	17:35
Daenyth\|Work	Is there a way I can tell git that I want to use an ssh key other than ~/.ssh/id_{r,d}sa?	17:35
zawir	dmilith: probably it could be somehow, i'm not a javagit developer.	17:35
Daenyth\|Work	Right now I'm using ssh-agent, but I'd rather just have git use the correct key directly	17:35
cardioid	When I edit the gitosis-admin project and add another repo to the writable = line in a group (and then push), that doesn't create a repository	17:35
	(or at least not in ~git/repositories)	17:36
	When I then make the repo on my local machine, add the remote, and push, THAT'S when I see the repository show up on the gitosis machine	17:36
zawir	dmilith: afair i've read a post somewhere that the author now has not enough time for nbgit development	17:37
dmilith	zawir: temporarely i rather use console "normal" way ;} and it's ok. but it could be nice if netbeans >= 6.1 could understand git repos..	17:37
zawir	dmilith: sure, it'd be;)	17:37
cardioid	Daenyth\|Work: afaik git just uses ssh, so the same way you set it up there. I have an .ssh/config with IdentityFile set up for different hosts	17:38
dmilith	zawir: oh well ;} i'll wait. thanks	17:38
Daenyth\|Work	cardioid: would you mind giving me an example from your .ssh/config? I'm using github, so if you have one for that, it would be perrfect	17:38
	perfect^	17:38
	So there's no .git config thing I can do though?	17:39
	→ dwave joined	17:39
cardioid	Daenyth\|Work: http://gist.github.com/3934	17:41
	→ CodeOfficer joined	17:41
Daenyth\|Work	thanks!	17:41
cardioid	I've seen weird behavior with some ssh agents, though.	17:41
	If I had both of those keys loaded, it would always serve id_dsa and always have me going to github as ymendel	17:42
	It would only work if I unloaded the keys and then loaded only the github_flogic key	17:42
	in the end I worked out a mirroring scheme, and a benefit is that I never need to push to github direectly as flogic	17:42
Daenyth\|Work	so for 'user', I would put my github name?	17:42
cardioid	No, you always connect to github as 'git'	17:43
	It's the key that identifies you	17:43
Daenyth\|Work	ah ok	17:43
cardioid	Does anyone know about gitosis (or git in general) and the lifecycle of repository creation?	17:43
Pieter	hah, and Gitclipse too	17:43
Daenyth\|Work	so just Host github.com\n\tUser git\n\tIdentityFile ~/.ssh/foo?	17:44
cardioid	Daenyth\|Work: Yeah, that should work	17:44
	You can see what's going on by trying to ssh there with -v	17:44
	or -vv even	17:44
Daenyth\|Work	I didn't know I could ssh to git actually	17:44
cardioid	Oh, you can't actually get in	17:45
Daenyth\|Work	ah	17:45
cardioid	but you can see what ssh is trying to do (like which key it's giving), and the message you get will tell you what's going on	17:45
	→ joydivider joined	17:46
	← statim left	17:46
bogolisk	Is "git branch new_b && git-symbolic-ref HEAD refs/heads/new_b" a "good" way to switch branch while preserving the index and the wdir?	17:47
	→ david_koontz joined	17:47
Daenyth\|Work	cardioid: in your config file, you have ~/ for some lines, but /home/ for IdentityFile. Any specific reason?	17:47
	← Beket left	17:48
cardioid	Oh, I think someone was teling me it didn't do any expansion, requiring an absolute path, and I believed them	17:48
Daenyth\|Work	aha	17:48
cardioid	I'm pretty sure I had it as ~/.ssh/id_dsa once upon a time	17:48
	(and it worked just fine, liars)	17:48
Daenyth\|Work	heh	17:48
cardioid	With the absolute path, though, it's made more clear to others why I have a User directive in the Hosts * section	17:49
	→ arwed joined	17:49
	Daenyth\|Work shrugs	17:50
Daenyth\|Work	generally I don't let other people in my ~/.ssh anyway ;p	17:50
cardioid	I mean when I paste it	17:51
	← _slacker_ left	17:51
cardioid	Because I've pasted this a lot	17:51
Daenyth\|Work	thanks for the tips :)	17:51
	kemme kill ssh-agent and push..	17:51
cardioid	Actually, I'm glad it's a gist now because I can keep track of it and show the same thing over and over instead of making lots of pasties	17:51
Daenyth\|Work	woo	17:52
	push went fine	17:52
	thanks for freeing me from keyring dependency :D	17:52
cardioid	np	17:52
	← Daenyth\|Work left	17:52
	→ pjhyett joined	17:54
cardioid	I guess I'll try my gitosis questions again later	17:54
	← m00l left	17:55
	→ Daenyth\|Work joined	17:56
Daenyth\|Work	mm, speakig of gist, what's the best way to make a new gist 'repo' from CLI, instead of using the web interface. Something like nopaste to get the repo started	17:56
Pieter	gistie	17:56
	← MarcWeber left	17:57
Daenyth\|Work	thanks!	17:57
Pieter	http://gist.github.com/1059	17:57
	→ Tv joined	17:58
	→ drizzd joined	17:58
	← cmarcelo left	17:58
Daenyth\|Work	looks like a perl script written in ruby :P	17:58
cardioid	Tv!	17:58
Daenyth\|Work	thanks	17:58
Tv	hello	17:58
	← ciaran left	17:58
cardioid	I have yet to use any pastie/gistie CLI. Maybe I should look into that. I think it's more common that I copy/paste sections of a file.	17:59
	→ cmarcelo joined	17:59
cardioid	Tv: I'm glad you're here because I am trying to figure out the gitosis repo-creation lifecycle	17:59
Daenyth\|Work	well, nopaste is handy to me	18:00
	logfiles or output	18:00
	/exec -o foo \| nopaste - or so	18:00
	puts the url in channel, very simple	18:00
cardioid	I added another repo took a group's 'writeable =' line in the gitosis-admin repo and pushed it. Then I looked on the server in ~git/repositories and it wasn't there	18:00
	Then I locally set up the repo (made a directory, init, remote add, commit, push). Then I looked on the server in ~git/repositories, and it was there	18:01
	What is it that actually creates the repository?	18:01
Pieter	the first push	18:02
cardioid	The action of pushing a commit creates the repository, yes, but what on the server is creating it?	18:03
	Is there a hook I can access?	18:03
Pieter	it's the gitosis script that's called	18:03
	it checks if the repo exists before calling receive-pack	18:03
cardioid	My goal is to ensure that any repo created in a particular group has a particular post-receive hook	18:03
Pieter	look at the gitosis code; it's fairly simple	18:04
	and easy to adjust	18:04
	→ tarbo joined	18:04
cardioid	I purport that it's not all that simple to look at the gitosis code, but that's just my feelings about Python getting in the way ;)	18:06
bogolisk	Is "git branch new_b && git-symbolic-ref HEAD refs/heads/new_b" a "good" way to switch branch while preserving the index and the wdir?	18:07
	← radarek left	18:08
Daenyth\|Work	icky python :p	18:08
	→ bcarlyon\|laptop joined	18:09
Pieter	bogolisk: no, use git checkout -b new_b	18:09
	← tarbo left	18:10
	→ warthog9 joined	18:11
Tv	cardioid: can't do custom hooks with current gitosis, sorry	18:11
	cardioid: well you can but i won't support your setup ;)	18:11
bogolisk	Pieter: does git-checkout leave the index alone?	18:11
Pieter	bogolisk: yes	18:12
	← npgoog left	18:13
	← swombat_ left	18:13
cardioid	Tv: That's understandable. Is there a reason you don't create the repo in the gitosis-admin post-update hook? Problems with having a completely empty bare repo or just not trying to clutter the directory with repos that may never be used?	18:13
	IIRC, you can't clone an empty repo, right?	18:13
	→ ciaran joined	18:14
bogolisk	Pieter: ty	18:14
Pieter	cardioid: you also can't pull a repo without any commits in it	18:15
	→ tarbo joined	18:15
Tv	yeah that wouldn't get you anything useful apart from manual customization -- and i hope to make the essential setup non-manual later	18:16
	→ sverrej joined	18:16
	← david_koontz left	18:16
cardioid	One of my gitosis annoyances is that I can't just update the config in gitosis-admin, push that, and then clone the repo. The mkdir/init/remote add thing is a pain, but that's just the way it has to be, eh?	18:17
	Unless you made an 'empty' repo with content, like a 0-byte .gitignore file as the initial commit	18:17
Pieter	you can make an empty commit in git	18:18
	you don't need any files or so	18:18
cardioid	Tv: So there's some plan for eventually allowing repo customization in gitosis?	18:18
	Pieter: Oh yeah?	18:18
Pieter	sure	18:18
	git commit --allow-empty	18:18
cardioid	Oh snap	18:19
Pieter	I think most gitosis things work by first creating a local repo, committing a bit, then thinking "I want to share this"	18:19
	in that case, it's not very useful to have the empty commit thing there	18:20
cardioid	Yeah, it makes sense	18:20
	Gitosis is definitely doing it the right way. I was just having trouble finding where the repo was created.	18:20
Pieter	thiago: heh, TT is completely switching to git?	18:21
	must be a lot of source	18:21
	← ciaran left	18:21
vuf	tt?	18:22
EgS	trolltech	18:23
vuf	oh	18:23
rwaite	oh, i thought you were talking about template toolkit	18:23
	→ Bitruder joined	18:24
Bitruder	When I try to push, it asks for the password of my linux user, but my username on the git server is different. Where do I specify that?	18:25
Pieter	Bitruder: use user@host in your url	18:25
Bitruder	thanks	18:26
	→ cilly joined	18:26
Tv	cardioid: for some definition of customization, yes	18:26
	← freqmod_qu left	18:28
	→ ciaran joined	18:28
	→ jonesy joined	18:29
cardioid	Tv: For what I want, which is to include a specific post-update hook if the repo is in a given group	18:29
jonesy	Running OS X Leopard 10	18:29
cardioid	Tv: and the post-update hook is dependent on the repo name, not completely static	18:30
Daenyth\|Work	jonesy: you poor thing	18:30
jonesy	Running OS X Leopard 10.5.4. Just installed Git from source. Trying to clone a repo and I get the error "fatal: could not create work tree dir <dir>	18:30
	any idea why I would be getting this?	18:31
	installed v1.5.6.4	18:31
Daenyth\|Work	do you have write permission on pwd	18:31
	→ joekarma_ joined	18:31
jonesy	yes	18:31
Daenyth\|Work	hm	18:32
	no clue	18:32
jonesy	even tried sudoing just in case	18:32
Daenyth\|Work	--verbose?	18:32
vuf	that error is if mkdir fails	18:33
jonesy	don't believe I can use verbose on the clone command	18:33
	→ sgrimm joined	18:33
Daenyth\|Work	perhaps the repo name has unicode characters not supported by your filesystem?	18:34
	can that happen?	18:34
	→ resmo___ joined	18:35
Tv	cardioid: there will be a way for the system admin (root) provide hooks that the gitosis admin (non-root) can enable for various repos	18:37
	← kanzure left	18:38
cardioid	Tv: I think I can wait for that. I'm loath to start maintaining my own fork of gitosis just to save a couple steps when I create a repo in a certain group.	18:39
	→ _bernie joined	18:39
	← Daenyth\|Work left	18:40
	→ Airwulf joined	18:40
Airwulf	hello	18:40
	I have got some problems with submodules. How do I correclty push changes of a submodule I did inside a super project to the public submodul repository?	18:41
	← Bitruder left	18:41
	→ freqmod_qu joined	18:41
Airwulf	when I do 'cd submodule;git commit -a; git push' I must enter a commit message but push tells me everything is uptodate	18:43
	← ajonat left	18:44
	mw → mw\|food	18:44
	← tvachon left	18:45
	← joekarma left	18:47
	→ netoman joined	18:47
netoman	Hello! Is there a way to git-cvsimport into a bare repository? It allways creates the .git/ subdirectory and working copy. Is there a reliable workaround?	18:48
	→ kevins joined	18:50
	→ ajonat joined	18:51
kevins	any ideas?	18:51
	$ git svn rebase	18:51
	First, rewinding head to replay your work on top of it...	18:51
	could not detach HEAD	18:51
	rebase refs/remotes/trunk: command returned error: 1	18:51
Pieter	what git verison do you have?	18:52
kevins	I'll be happy with blowing away any differences between where I am on my branch and the latest SVN revision.	18:52
	$ git --version	18:52
	git version 1.5.6.4	18:52
	→ tvachon joined	18:52
kevins	I just did a git fsck and a git gc and nothing looked too bad.	18:53
Pieter	is your working tree dirty?	18:53
kevins	there are a few untracked files only.	18:53
	should I nuke them?	18:54
Pieter	perhaps one of those files _is_ tracked in the last svn commit?	18:54
kevins	$ git status	18:55
	# On branch mainline	18:55
	nothing to commit (working directory clean)	18:55
	$ git svn rebase	18:55
	First, rewinding head to replay your work on top of it...	18:55
	could not detach HEAD	18:55
	rebase refs/remotes/trunk: command returned error: 1	18:55
	← timocratic left	18:55
Pieter	what happens if you try "git checkout remotes/trunk" ?	18:56
	→ bugfux joined	18:56
kevins	will that do anything bad to my SVN repo :-P	18:56
Pieter	no :)	18:56
kevins	I get an error : error: Untracked working tree file ' somefile' would be overwritten by merge.	18:57
	where did that come from?	18:57
	I'll do whatever to get rid of that local problem...	18:57
Pieter	:)	18:57
	apparently an ignored file has been checked in on the svn side	18:57
pasky	* svn checkout https://subversive.cims.nyu.edu/osp/javagit/trunk/	18:58
	rofl	18:58
	these guys are just joking, right?	18:58
	→ Daenyth\|Work joined	18:58
kevins	really now....	18:58
Daenyth\|Work	Is there a way I can use git-svn to check out a single file from an svn repo? Possibly with proper commit info, like authoer of the last commit on the file..	18:58
	author^	18:58
	← UVSoft left	18:59
whee	pasky: but some of the tools they are using don't interface with git repsitories	18:59
Daenyth\|Work	I have a bundle of things managed by git, but one thing in the bundle is managed by another person using svn	18:59
_bernie	any git-svn expert around?	18:59
Daenyth\|Work	(on a separate repo)	18:59
_bernie	is git-svn known to work well with bare repositories?	18:59
	→ david_koontz joined	19:00
_bernie	I'm having trouble with "git svn fetch" not updating the master branch, and "git svn rebase" complaining a lot	19:00
pasky	my git-svn does not work with bare repositories at all	19:00
	it just spews some cdup errors	19:00
	→ Eridius joined	19:00
pasky	and I actually plan to fix that tomorrow ;)	19:00
_bernie	git-svn version 1.5.5.1 (svn 1.4.6)	19:00
pasky	(but I have busy day tomorrow)	19:00
	← bugfux left	19:00
_bernie	pasky: did it break at some point? I used git svn some time ago to mirror the whole gcc repository and it was a charm.	19:01
	→ bugfux joined	19:01
pasky	I'm not regular git svn user	19:01
	← lu_zero left	19:01
pasky	I somehow thought this was never supported, but I have actually no basis for that	19:02
	→ nud joined	19:02
_bernie	pasky: the gcc mirror on git.infradead.org is a bare repository...	19:02
	pasky: it's 1.5.4.1, while the one that does not want to work sanely is 1.5.5.1	19:03
	→ Fullmoon joined	19:04
	← raz left	19:06
	→ boink joined	19:06
	boink → raz	19:06
	← GarethAdams left	19:07
	← ijcd left	19:08
	→ ijcd joined	19:08
	← giom left	19:09
kevins	Pieter: I looked through my .gitignore files and I can't imagine what it would be having trouble with....I removed the directory that git checkout remotes/trunk was complaing about, but it left me in (no branch) instead of my normal mainline.	19:09
Pieter	yeah, that's good	19:09
	no, try "git checkout mainline" and "git svn rebase"	19:10
kevins	so I did : $ git checkout remotes/trunk and I got some joy :	19:11
	Note: moving to "remotes/trunk" which isn't a local branch	19:11
	If you want to create a new branch from this checkout, you may do so	19:12
	(now or later) by using -b with the checkout command again. Example:	19:12
	git checkout -b <new_branch_name>	19:12
	But now: git checkout mainline	19:12
	error: Untracked working tree file 'somefile' would be overwritten by merge.	19:12
	→ crab joined	19:12
	→ statim joined	19:12
kevins	I can't get back to mainline. Should I just delete mainline and make where I am at the new mainline?	19:12
Pieter	and you don't have any local changes?	19:12
	try git checkout -f mainline	19:13
kevins	well, the local changes are the directories I had to remove to get the remote/trunk.	19:13
	← pjhyett left	19:13
kevins	-f is working.	19:13
	got bless you... -f worked to get me from (no brach) back onto mainline.	19:14
	god	19:14
	→ timocratic joined	19:14
Mikachu	you weren't in the middle of a rebase?	19:14
kevins	rrrgh...no!.... :	19:15
	$ git svn rebase	19:15
	M somefile	19:15
	r2953 = ca89e70a4e3285d78af6e7e23e3734b9c91386ad (trunk)	19:15
	First, rewinding head to replay your work on top of it...	19:15
	could not detach HEAD	19:15
	rebase refs/remotes/trunk: command returned error: 1	19:15
	→ perezd88 joined	19:15
kevins	I was not in the middle of a rebase. I just need to get my mainline back to the head of the SVN trunk....no matter what it takes...	19:16
	← Airwulf left	19:18
Pieter	kevins: try to remove those files you removed first again	19:19
kevins	ok. I removed the whole directorty that seems to be giving me problems.	19:20
	I do git svn rebase and I get lots of	19:21
	resources/somefile: needs update	19:21
	update-index --refresh: command returned error: 1	19:21
	I feel like I should just do a git reset --hard HEAD sha1 to get that branch to the right place.	19:22
	maybe just abandon the broken mainline and "start over" somehow....it doesn't provide anything helpful at this point.	19:23
	← spearce left	19:24
	rwaite → rwaite\|afk	19:24
	→ spearce joined	19:25
	← MK_FG left	19:27
kevins	thanks for the help everyone.....just did a checkout to a different (where svn rebase was working) branch, branch -d mainline, checkout -b mainline and now svn rebase works.	19:28
	git svn had been working flawlessly up to then....I have no idea where it was confused.	19:29
	← kevins left	19:29
	→ advany joined	19:29
	← advany_ left	19:30
	← chrisb left	19:36
	← crab left	19:36
	← bogolisk left	19:37
	→ chrisb joined	19:37
	→ _zack joined	19:38
	← markkalderon left	19:43
	← hobodave left	19:43
	→ bts-_ joined	19:43
vuf	what is this: .dotest/patch:108: trailing whitespace.	19:47
	no +x hooks	19:48
Pieter	windows?	19:49
vuf	no Linux	19:51
	during a rebase	19:51
Pieter	do you have a .gitattributes?	19:51
	→ thannoy joined	19:51
vuf	no	19:52
Pieter	.git/info/gitattributes then?	19:52
vuf	no attribute files	19:53
Pieter	core.whitespace set?	19:54
	or apply.whitespace	19:54
vuf	nope ... 1.5.6.4, btw	19:54
	← bts- left	19:54
	→ warlock_mza joined	19:56
pasky	hmmm	19:57
	oh, pieter	19:57
	repo working now?	19:57
	or still broken?	19:57
Pieter	let me check	19:58
	still broken	19:58
pasky	you need to give yourself push access	19:58
	at the edit project page ;)	19:58
	but we need better error reporting	19:58
Pieter	I thought I did that	19:58
pasky	you didn't, it seems	19:58
Pieter	perhaps I don't know my username	19:58
pasky	hm, I should try to squeeze the better error reporting to 1.6.0 yet maybe	19:59
	mw\|food → mw	19:59
pasky	but first I need to get my git-repo branchcloud to some sensible shape	19:59
	and for that I need to do some tg hacking	19:59
	sigh	19:59
	so much to do :)	19:59
	→ juliano joined	19:59
pasky	Pieter: there are no users with push access whatsoever	19:59
Pieter	ah	19:59
	if you don't fill in the password, the error is at the top	20:00
	but the user is still "added" at the bottom	20:00
	so I didn't notice the error :)	20:00
	works now, thanks	20:00
	→ ruphy joined	20:01
pasky	yes this is something I plan to fix ;)	20:02
	there's a lot of usability improvements brewing right now	20:02
	I just need to fix up the damn tg to support remotes so I can get it to repo.or.cz :)	20:02
Pieter	:D	20:03
pasky	BTW, any tg-interested people around? what do you think would be the best mode for tg vs remote branches? instantiate all topic branches locally right away (and automatically fast-forward on "pulls" whenever possible), or fall back to origin topic branches if there is no local one?	20:03
	← advany left	20:06
	← Daenyth\|Work left	20:07
	← bts-_ left	20:09
whee	I'd rather not have local branches cluttered up	20:13
	aroben → aroben\|meeting	20:14
	← nud left	20:15
	← Jurko left	20:16
whee	pasky: does git have any problems with a large number of branches?	20:17
	I know I have problems sorting through git branch output when there are a lot :\|	20:17
	← arwed left	20:20
	→ lu_zero joined	20:21
pasky	the trouble is, doing the fall back is (i) quite more complex (ii) much slower	20:23
	about handling large number of branches, frankly, I don't know :)	20:23
whee	I suppose I don't really understand tg :)	20:23
pasky	I didn't yet get over 20 branches myself	20:23
whee	I haven't used it, but I like it from what I've read in the announcement	20:23
pasky	good :)	20:24
	the lack for pull/push support is the biggest shortcoming so far I think	20:24
	after that gets fixed it should get qiute practically usable I think	20:24
whee	I haven't used stgit because it seems like it does things outside of the usual git methods, while tg sounds less zany	20:24
	→ Ademan joined	20:26
	→ ikitat joined	20:26
	→ saidin joined	20:27
	→ ml joined	20:28
ml	hi everyone, I have a question...	20:29
jengelh	that does not look like a question.	20:29
ml	It's coming :-)	20:29
jengelh	fail.	20:30
	← elmarco left	20:30
pasky	ssshht jengelh :)	20:30
	it takes time to compose good question!	20:30
	i'm sure that the more time, the better question!	20:31
	→ dfischer joined	20:31
dfischer	Let's say you have a plaintext password visible in a file that has	20:31
	been comitted over time. Is there a way to remove that file from every	20:31
	commit in the git repo?	20:31
pasky	you can use git filter-branch, but it will rewrite history, with all its caveats	20:32
Pieter	use git rebase -i to edit the commit that introduced it	20:32
dfischer	thanks =]	20:32
pasky	isn't that conflict-prone?	20:32
jengelh	pasky: the time wasted to post a metaintroducer could have been better spent on the question too...	20:33
	→ peritus_ joined	20:34
pasky	as well as time wasted reading and replying to your comment ;)	20:34
ml	I have a git clone of a git-svn repository. Note, while the git-svn reference does not recommend this, one of the Apache committers has figured out a way to make it work. This is only relevant to my question in that it results in some remote branches that look like 'origin/tags/whatever' (because in Subversion, "tags" are really branches). So then, I am in the project-relative path "parent/", and I need to check this path out to be like "/origi	20:34
	git checkout /origin/tags/cocoon-2.2/cocoon/cocoon-6 .	20:34
	but it says	20:34
	error: '/origin/tags/cocoon-2.2/cocoon/cocoon-6' is outside repository	20:34
	ok that's my question, sorry it took so long to compose :-)	20:35
	← mcella_ left	20:35
whee	ml: perhaps you don't want the leading slash?	20:35
	→ elmarco joined	20:35
	← catdude left	20:36
	→ markkalderon joined	20:36
	← tcoppi left	20:36
ml	@whee: bah, I copy/pasted the wrong line. That was my second attempt, the first was without the leading '/', it yielded:	20:36
	error: pathspec 'parent/origin/tags/cocoon-2.2/cocoon/cocoon-6' did not match any file(s) known to git.	20:36
Pieter	pasky: I'd rather not have it autosetup those branches	20:36
drizzd	ml: how is that a path?	20:37
	it's a mixture of path, remote, and tag names	20:37
Pieter	ml: is that listed in "git show-ref"?	20:38
ml	no, there's no "tag name" in there.	20:38
pasky	Pieter: care to elaborate?	20:38
Pieter	pasky: you want to let it create local branches for every remote topgit branch?	20:38
pasky	hmm	20:39
ml	drizzd: my command was "git checkout origin/tags/cocoon-2.2/cocoon/cocoon-6 ." The path is '.'. The 1st argument is supposed to be the "<tree-ish">" from `man git-checkout`, but obviously I know not what I am doing...	20:40
pasky	you would like to have multiple topgit remotes per repository?	20:40
	travesty	20:40
	ponders	20:40
	damn :)	20:40
Pieter	yes :)	20:40
	← ikitat left	20:40
	← _zack left	20:40
drizzd	ml: git show-ref should give you tree-ishes to choose from	20:42
	→ nud joined	20:43
drizzd	ml: and you probably don't want the '.', unless you know you want it	20:43
ml	Well, I want this checkout to apply only to the tree rooted at my current directory.	20:44
drizzd	ok then I guess you really do want it	20:44
ml	Okay, so 'git show-ref' gives me a bunch of stuff... I tried:	20:44
	git checkout refs/remotes/origin/tags/cocoon-2.2/cocoon/cocoon-6 .	20:44
	and get:	20:44
	error: pathspec 'parent/refs/remotes/origin/tags/cocoon-2.2/cocoon/cocoon-6' did not match any file(s) known to git.	20:45
	so I must not be specifying a "tree-ish" in the right way...	20:45
	it's still interpreting it like a local path	20:45
	(relative path)	20:45
drizzd	ml: can you do git cat-file -t refs/remotes/origin/tags/cocoon-2.2/cocoon/cocoon-6 please?	20:46
	← spearce left	20:46
ml	drizzd: fatal: Not a valid object name refs/remotes/origin/tags/cocoon-2.2/cocoon/cocoon-6	20:47
drizzd	that's weird	20:48
ml	If I work my way up, I get the same thing up to here:	20:48
	git cat-file -t refs/remotes/origin/tags/cocoon-2.2	20:48
	which says:	20:48
	commit	20:48
drizzd	what happens if you do the same thing to the sha1 listed next to it in git show-ref?	20:48
	→ spearce joined	20:48
chris2	tg looks great	20:49
ml	drizzd: same thing: one-line output, 'commit'	20:49
drizzd	well if refs/remotes/origin/tags/cocoon-2.2 is a already a ref, I don't see how it can also be a path to another ref	20:49
	→ wacky_ joined	20:49
drizzd	ml: refs/remotes/origin/tags/cocoon-2.2/cocoon/cocoon-6 obviously doesn't exist and also does NOT show up in git show-ref as you claimed it does	20:50
wacky_	are you sure the "Find commits adding/removing strings" feature in `gitk` works well ?!	20:50
	→ boink joined	20:50
ml	drizzed:	20:51
	[codeshack:cocoon/git/parent] ml% git show-ref refs/remotes/origin/tags/cocoon-2.2	20:51
	de2e80516dc1c6b1dc02deb5a635ffe676dbe129 refs/remotes/origin/tags/cocoon-2.2	20:51
	← kumbayo left	20:51
	→ tcoppi joined	20:52
	→ bts- joined	20:52
ml	drizzd: and then:	20:53
	[codeshack:cocoon/git/parent] ml% git cat-file -t refs/remotes/origin/tags/cocoon-2.2	20:53
	commit	20:53
	→ schlort joined	20:54
drizzd	ml: If you're waiting for me to say something, I got nothing	20:54
ml	drizzd: well that was something ;-)	20:55
	anyway, gitweb running on the original git-svn repository is able to figure this out... so I must've fubar'd something, I just don't know what	20:56
	→ JonathanT joined	20:56
	→ advany joined	20:56
vuf	hm .git/packed-refs can I edit that by hand?	20:57
	→ Sonderblade joined	20:58
vuf	well, seems like it	20:58
	← JonathanT left	20:58
	→ LostInTheWorl1 joined	20:59
	→ markkald_ joined	20:59
	← LostInTheWorld left	20:59
	← markkalderon left	21:00
	→ hobodave joined	21:01
	← raz left	21:02
	boink → raz	21:02
	→ haggis joined	21:03
	← ph^ left	21:06
	← Ademan left	21:07
	← janm left	21:08
	← dwave left	21:10
hobodave	hmm	21:11
	how do I use --no-ignore-externals with git-svn?	21:11
Pieter	pasky: chris2 blogged you! ;)	21:11
hobodave	I'm getting 'Unknown option: no-ignore-externals	21:11
	→ ajonat_ joined	21:11
haggis	Does anyone in here use msysGit with Plink on Windows? I'm getting a strange error "bash: [email@hidden.address] command not found" when performing a git clone	21:11
	chris2 is amazed by who reads his blog	21:11
	← ml left	21:12
	← ajonat left	21:12
haggis	cloning my public repos works fine though	21:13
	→ Arrowmaster` joined	21:14
	← drizzd left	21:16
	← ajonat_ left	21:16
	← cilly left	21:16
	← sverrej left	21:16
	← krh left	21:16
	← madewokherd left	21:16
	← chris2 left	21:16
	← shad left	21:16
	← dwmw2_gone left	21:16
	← explody left	21:16
	← Arrowmaster left	21:16
	Arrowmaster` → Arrowmaster	21:16
	← danten left	21:16
	→ ajonat_ joined	21:16
	→ cilly joined	21:16
	→ sverrej joined	21:16
	→ krh joined	21:16
	→ madewokherd joined	21:16
	→ chris2 joined	21:16
	→ davidfetter joined	21:16
	→ nominolo joined	21:16
	→ jmspeex joined	21:16
	→ uau joined	21:16
	→ explody joined	21:16
	→ dwmw2_gone joined	21:16
	→ shad joined	21:16
	→ pete__c joined	21:16
	→ zeenix joined	21:16
	→ worch joined	21:16
	→ cout joined	21:16
	→ corecode joined	21:16
	→ mountie joined	21:16
	→ sr joined	21:16
	→ ehamberg joined	21:16
	→ Arjen joined	21:16
hobodave	how do I use --no-ignore-externals with git-svn?	21:17
	I'm getting 'Unknown option: no-ignore-externals	21:17
	← ajonat_ left	21:17
pasky	Pieter: woot, great	21:17
	now who is chris2 and what is his blog? ;)	21:17
	← sverrej left	21:17
	→ sverrej joined	21:18
chris2	pasky: anarchaia.org	21:18
pasky	\o/	21:19
	→ ajonat_ joined	21:19
hobodave	how do I use svn:externals with git-svn?	21:20
Pieter	not	21:20
	you can create submodules for them, if you really want, but it's a bit of work	21:21
hobodave	how would you recommend doing it?	21:21
	does git have its own externals mechanism?	21:21
Pieter	yes, submodules	21:21
hobodave	that I can use separately, without having to worry about checking it into my svn repo?	21:21
	they have to be git repos?	21:22
chris2	hmm	21:22
	several people could work on the same topgit branches, no?	21:22
	→ unreal_ joined	21:22
pasky	yes	21:22
	once i figure out how to support topgit remotes :)	21:22
chris2	pasky: that would rock	21:22
	← Fullmoon left	21:23
chris2	then we "just" need a patch annotation system, and voila, code reviews :-)	21:23
	pasky is not really familiar with the modern code review systems	21:24
pasky	I'm mailinglistish traditionalist ;)	21:24
	← BabelO left	21:25
spearce	i heard rietveld is getting git support soon.	21:25
	← schlort left	21:26
pasky	yes, I've just read in some link from chris2's blog	21:26
	← pete__c left	21:26
	← unreal_ left	21:26
	← Arjen left	21:26
	← zeenix left	21:26
	← corecode left	21:26
	← ehamberg left	21:26
	← dwmw2_gone left	21:26
	← mountie left	21:26
	← uau left	21:26
	← worch left	21:26
	← cout left	21:26
	← explody left	21:26
	← nominolo left	21:26
	← shad left	21:26
	← madewokherd left	21:26
	← chris2 left	21:26
	← krh left	21:26
	← sr left	21:26
	← jmspeex left	21:26
	← davidfetter left	21:26
	← cilly left	21:26
pasky	too bad I have no clue what rietveld is ;)	21:26
	In 1916, Rietveld started his own furniture factory, while studying architecture.	21:27
spearce	its the code review system google built, but ported to be open source and run outside of their environment.	21:27
pasky	hmm	21:27
Pieter	but it's based on the google app thing, right?	21:27
	is that any good?	21:27
spearce	the internal one is based on perforce; rietveld is based on svn.	21:27
	i have no idea if it is based on google app engine, or runs standalone	21:27
	← saidin left	21:27
	→ bryanray joined	21:27
pasky	This project shows how to create a somewhat substantial web application using Django on Google App Engine.	21:28
	→ voodootikigo joined	21:28
Pieter	too bad they downgraded it to svn before releasing then	21:28
spearce	heh. so yea, it requires the app engine. which makes sense actually as the internal tool it came from was based upon the stuff that backs app engine.	21:28
	→ ehamberg joined	21:28
spearce	not that it may be the best thing for the open source community, but it was the easiest way to port the code and open it.	21:29
	→ cout joined	21:29
	→ nominolo joined	21:29
	→ krh joined	21:29
	→ corecode joined	21:29
	← ruphy left	21:30
pasky	hmm	21:30
	so it's something like bazaar's PQM?	21:30
	but with more web-oriented controls?	21:30
spearce	maybe. i know nothing about bazaar's PQM. :)	21:30
pasky	it's a nice gadget	21:30
	→ mountie joined	21:30
	→ zeenix joined	21:31
pasky	https://launchpad.net/pqm	21:31
	what's not obvious from the description is how it actually works :)	21:31
	someone sends a patch to the mailing list, PQM grabs it, puts it into the queue and watches for replies	21:32
	if other developers ack it, it gets merged	21:32
	→ chris2 joined	21:32
pasky	or at least that's my idea from watching the bazaar mailing list for a while	21:32
	→ uau joined	21:32
	→ cilly joined	21:33
	→ eternaleye joined	21:33
	← nud left	21:33
chris2	something like that	21:35
	my idea would be to make a mob repository	21:35
	→ mutex joined	21:35
chris2	and add some comment interface for every topgit branch	21:35
	← mutex left	21:36
pasky	that's interesting	21:37
	→ shad joined	21:37
chris2	that way, people can easier collaborate on the same branches too	21:38
	→ flaguy joined	21:39
	→ BabelO joined	21:41
	→ unreal_ joined	21:42
	→ madewokherd joined	21:42
	→ explody joined	21:42
	→ pete__c joined	21:42
	→ worch joined	21:42
	→ sr joined	21:42
	→ Arjen joined	21:42
	→ mutex joined	21:43
	← vbabiy left	21:44
	← netoman left	21:48
	→ jmspeex joined	21:49
	→ GarethAdams joined	21:50
	→ davidfetter joined	21:53
	→ nadim joined	21:54
	← warlock_mza left	21:55
	→ icwiener joined	21:55
	← krh left	21:56
	← sergiodj left	21:58
	← jackdempsey left	21:58
	← spearce left	21:59
	← tarbo left	22:01
	→ Hobbes joined	22:03
	→ nexus10 joined	22:03
	← nexus10 left	22:04
	← chris2 left	22:05
	→ tarbo joined	22:06
	← markkald_ left	22:07
	← ciaran left	22:08
nominolo	can someone give me a quick hint which man page i should read to find out why pulling over http does not get the latest revision, while ssh works just fine?	22:08
	has it something to do with packing?	22:08
Mikachu	nominolo: git update-server-info	22:10
nominolo	Mikachu: ah, thanks	22:10
	← BabelO left	22:13
	→ ijcd_ joined	22:14
	← voodootikigo left	22:16
	→ markkalderon joined	22:17
	← EmilMedve left	22:18
	→ Khisanth joined	22:20
	← vuf left	22:21
	← advany left	22:22
	→ voodootikigo joined	22:22
	→ ereslibre joined	22:23
	← voodootikigo left	22:26
	← fhobia left	22:27
	bcarlyon\|laptop → BarryCarlyon	22:30
	← ijcd left	22:31
	→ nexus10 joined	22:34
	← icwiener left	22:34
	← Weasel[DK] left	22:36
nexus10	hi -- is there any way I can checkout just a subdir from a git repo? There's a project.git and I need project/foo	22:37
Mikachu	cp -r project/foo .	22:38
nexus10	Mikachu: I'd love to -- but it's part of a capistrano deployment recipe	22:39
	Mikachu: and I can't see how to hook in a `cp -r` before cap starts screaming that it can't find stuff	22:40
	← jerbear left	22:40
	aroben\|meeting → aroben	22:42
	→ kaizoku joined	22:43
kaizoku	I can't figure out how to commit part of a file rather than the whole thing	22:44
	Anyone know off the top of their head?	22:45
	→ sebell joined	22:46
nexus10	kaizoku: is git --patch what you want? I think it uses --interactive...	22:47
cardioid	kaizoku: add --patch or --interactive	22:47
	That'll let you put some changes in the index, and then committing is normal after that	22:47
kaizoku	Ah, thanks	22:48
	← chrisb left	22:49
cardioid	Depending on which changes you want to add, you may have to make some edits anyway	22:52
	← aroben left	22:52
Mikachu	in very new git versions, there's the 'e' option in --patch mode	22:53
	→ mithro joined	22:53
dfischer	Why would .gitignore not be ignoring something? shouldn't log/* ignore everything in log?	22:54
	→ eikonos joined	22:54
Mikachu	dfischer: only untracked files	22:55
dfischer	how do you untrack something?	22:55
Mikachu	git rm --cached	22:55
dfischer	thanks	22:55
	← kaizoku left	22:55
	→ Fullmoon joined	22:56
	→ priidu joined	22:56
cardioid	Mikachu: What's 'e'?	22:59
Mikachu	you'll see when 1.6 is out ;)	22:59
cardioid	Tease!	22:59
	Is it like s but better?	22:59
Mikachu	you could say that	22:59
cardioid	is it a four-letter word?	23:00
	Can I buy a vowel?	23:00
Mikachu	yes :)	23:00
BarryCarlyon	£50	23:00
cardioid	Interesting	23:00
	← d0k_ left	23:00
	← peritus_ left	23:01
BarryCarlyon	But for you its £12.50	23:01
cardioid	1/4 the price!	23:03
	cardioid &	23:03
	→ rofrol joined	23:05
Mikachu	fg %cardioid	23:05
	→ towski joined	23:08
towski	how do I revert a single file?	23:08
Mikachu	do you want to revert changes to a file, or revert the file to a specific revision?	23:08
towski	revert to the last revision	23:09
Mikachu	ie you made changes in the working tree and you want to put it back to HEAD?	23:09
towski	(revert changes)	23:09
	yes	23:09
Mikachu	git checkout HEAD file	23:09
	it won't ask for confirmation before throwing your changes away	23:09
towski	ok cool thanks	23:10
	← rmh3093 left	23:13
	→ rmh3093 joined	23:14
	← Fullmoon left	23:14
	← charon left	23:14
	← rofrol left	23:15
	← Sonderblade left	23:20
	← bugfux left	23:20
	→ raffa joined	23:25
cehteh	git complains that a file has unresolved conflicts because it contains '======='	23:25
	← markkalderon left	23:25
cehteh	how to circumvent that .. these are legally there, asciidoc header	23:25
	→ markkalderon joined	23:26
nexus10	if I cannot checkout just a portion of a repo, is it possible to create a new repo which contains just this portion (i.e. a subdir) and which keeps the history?	23:26
cehteh	git-filter-branch	23:27
nexus10	ah... dunno that, will go read. Tx a lot.	23:28
	← botanicus left	23:29
	→ collothm joined	23:29
	← collothm left	23:30
doener_	cehteh: that's your pre-commit hook complaining I guess. --no-verify stops commit from calling it	23:30
cehteh	donno .. not my problem, i only relayed for a project member	23:31
	ah	23:31
	he	23:31
	cehteh points to raffa :P	23:31
cehteh	didnt seen you here :P	23:32
raffa	:-)	23:32
doener_	If you ask, I just pretend it's your problem, crystal ball is in repair :-p	23:34
raffa	doener_: Mine too. That's why I asked cehteh. ;-)	23:39
	← deavid left	23:39
cehteh	/nick gitrelay	23:39
	everyone here wants a cool git* nick :)	23:40
	→ kewark joined	23:44
	sebell → sb-away	23:45
	← thannoy left	23:45
	→ vbabiy joined	23:46
	← raven79 left	23:46
d-snp	hey guys, I'm a bit new to git	23:49
	I added a lot of stuff to my commit buffer	23:49
	but instead of committing it to head I want to commit it to a branch	23:49
	if I do git branch branchname	23:49
	will it commit what I have in the buffer to the branch?	23:49
doener_	you'll also have to "git checkout branchname"	23:50
	or just "git checkout -b branchname" as a shorthand	23:50
aeruder	probably git checkout -m branchname, no?	23:50
doener_	aeruder: hm? He just created the branch from HEAD, how should that produce any source for conflicts?	23:51
aeruder	doener_: sounds to me like he has local modifications, i thought checkout refused to change with local mods	23:51
d-snp	yeah I have local modifications	23:51
aeruder	maybe there's a special case when they are the same refs, i'd assume so	23:51
doener_	only when the file differs between the branches	23:51
	← raffa left	23:51
aeruder	doener_: ah, ok	23:52
	d-snp: don't use -m then :)	23:52
d-snp	ok	23:52
doener_	no need for a three way merge is the bases are the same	23:52
	s/is/if/	23:52
	← markkalderon left	23:53
	← mabs29 left	23:54
	← krawek left	23:54
	→ mabs29 joined	23:55
d-snp	I think it worked, thanks :)	23:55
doener_	d-snp: the thing to realize here (and which many don't) is that your changes are just on your working tree and/or index until you commit. So they'll "follow" you when you switch branches	23:57
d-snp	ok cool	23:57
doener_	Pretty often folks ask why "uncommitted changes on a branch" (no such thing), don't disappear when they checkout another branch	23:57
d-snp	it's exactly the behaviour I was hoping for :P	23:58
	→ morii_ joined	23:59
doener_	yeah, it's great :-)	23:59

Logs Search ←Prev date Next date→

Channels Documentation