IRCloggy #git 2009-05-04

Provider of IRC logs since 2005.

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

	2009-05-04
	← marcus__1 left	00:00
	→ shingo3 joined	00:05
	← blinton left	00:05
	texel → kent\r	00:06
	← mw left	00:06
	kent\r → texel	00:06
	← merlin83 left	00:08
	← aziz_ left	00:08
	← aspotashev_afk_ left	00:09
	→ mw joined	00:09
	← mw left	00:10
	← steveire left	00:12
elliottcable	oi all	00:15
	can I submodule in a single file from another repository?	00:15
	sdiz-afk → sdiz	00:16
	← thiago_home left	00:16
	→ wshimmy1 joined	00:20
	← shingo left	00:20
cehteh	elliottcable: no, but you can symlink a file into another repository (maybe submodule)	00:20
	→ Midnight joined	00:20
	← gotgenes left	00:21
	← ferdy left	00:22
Midnight	I'm trying to build git on OS X, and I get this error:	00:23
	http-push.c:15:19: error: expat.h: No such file or directory	00:23
	← dodo_the_last left	00:23
	→ bremner_ joined	00:25
Midnight	(Then a bunch more, but you know how useless gcc is at recovering)	00:25
snogglethorpe	is expat normally installed on osx?	00:26
slowriot	I'm having problems with empty commit messages in git.	00:26
	I add lines to the text file that do not begin with #, but it still complains.	00:26
	← bremner_ left	00:27
snogglethorpe	since git has non-adapative build configuration by default, it's probably not very good at dealing with unexpected issues	00:27
	cpg\|away → cpg	00:30
	← bonch left	00:30
context	midnight: why not install it from macports	00:31
	midnight: and that error is saying you are missing expat, or at least the dev files for it...	00:31
	midnight: gcc error messages most the time are very informative	00:32
slowriot	could the encoding I'm saving .git to be wrong?	00:32
	not .git, COMMIT_EDITMSG	00:32
snogglethorpe	slowriot: well what encoding are you using?	00:32
	also, "complains"?	00:33
	← catdude left	00:33
slowriot	whatever the default gedit encoding is	00:33
Midnight	context: Because I don't have macports installed and I find it rather a pain	00:33
	→ bremner_ joined	00:33
context	midnight: apparently. you are doing better without it	00:33
slowriot	by complains, I mean it prints "Aborting commit due to empty commit message."	00:34
	I just managed to commit	00:34
Midnight	context: Well, make didn't trash my system when it failed :P	00:34
	← awx left	00:34
slowriot	was it because there was no leading whitespace?	00:34
snogglethorpe	slowriot: i dunno how gedit works; maybe it's not saving to the right filename?	00:34
	i think gnome stuff always uses utf8 by default, which should be fine for git	00:35
context	midnight: macports has never "trashed" any part of my system in anyway	00:35
	← wshimmy left	00:36
doener	slowriot: I guess that gedit forks and detaches	00:36
Midnight	context: Doesn't mean it never happens.	00:36
doener	slowriot: so "git commit" continues its work before you even get to do anything within gedit	00:36
context	midnight: id really like to know how it "trashed" considering the only part of the system it really touches is /opt/local	00:37
Midnight	I'd like to know that tool	00:37
	*too	00:37
context	midnight: and make has nothing to do with macports. yell at gnu for that nifty PoC tool	00:37
	meh, whatever	00:37
Midnight	context: Er, I said make didn't	00:38
context	enjoy the errors	00:38
Midnight	As in... >_>	00:38
	→ scarabx joined	00:38
slowriot	doener: I am sometimes able to commit though. I think adding lines which are purely whitespace was messing it up.	00:38
	← bremner_ left	00:38
	← shingo3 left	00:39
doener	slowriot: google says that gedit does that only when another gedit instance is already running	00:39
	slowriot: so probably: no gedit running -> works. Some other gedit already running -> fail	00:39
slowriot	doener: oh, that could explain it	00:39
	→ leanucci joined	00:40
snogglethorpe	doener: ugh, that sounds awful	00:41
	doener: surely there must be a command-line option to tell it not to do that?	00:41
doener	snogglethorpe: google just finds people complaining that there isn't one. And apparently someone started working on a _plugin_ to make it support such an option	00:42
	snogglethorpe: I'm so glad I get to use vim...	00:42
	← MisterN left	00:42
doener	snogglethorpe: those "desktop" guys seem to ruin everything... goes off to work before he starts to lament about the HAL/DeviceKit/foobar .fdi XML hell	00:43
snogglethorpe	well i suppose gedit was probably made with a windows mindset, where using apps as cooperating tools is an alien concept	00:43
doener	snogglethorpe: probably... those guys seem to copy every bad idea the windows' developers had	00:45
snogglethorpe	yup	00:46
	it's kind of sad, because I like the gtk+ toolkit ...	00:47
doener	about:config, gconf, .fdi files, ... I switched from Windows because I could understand what's going on on my linux box... Nowadays, it's getting more and more of a blackbox. All things tend to turn to be machine-oriented, not humanly readable. And the documentation sucks (if there is any, I have just found "quirks" and config snippets...)	00:47
	→ Modius joined	00:49
	← leanucci left	00:50
	→ Radar joined	00:54
	→ qrush joined	00:55
	← eddyp left	00:55
	→ fujin joined	00:55
	← Modius_ left	00:58
	← slowriot left	01:00
	→ commondream joined	01:00
	→ Modius_ joined	01:00
	← Modius left	01:00
	→ mw joined	01:00
	→ didi joined	01:05
	→ whaley joined	01:06
didi	I started a branch and now I want this branch to become master and drop the old master altogether. Maybe saving the old master by tagging it. But I don't want to merge. I want just the new master. Is it possible?	01:08
	← jhubert left	01:08
fujin	Yup.	01:08
doener	didi: if there are no commits in "master" that aren't in the other branch, you can just use "git merge otherbranch" and it'll end up as a fast-forward	01:09
fujin	git branch -m/-M	01:09
snogglethorpe	didi: yeah, just go to master and do: git reset --hard OTHERBRANCH	01:09
fujin	or rename your master and rename your new master to master? =D	01:09
doener	didi: if there are commits in master that aren't in the other branch, then such a rename would qualify as rewriting history	01:09
	didi: others that already got the old version of master would still keep those commits, and just merge your new idea of master	01:09
didi	There are no commits at master. It is just freeze since I started the new branch.	01:10
doener	didi: then just merge, no need to play tricks	01:10
didi	Actually the brach was kind of a complete rewrite. I don't think it would be prudent to merge.	01:10
snogglethorpe	didi: reset --hard	01:11
fujin	git branch -m master oldmaster && git branch -m newmaster master	01:11
	=D	01:11
	← hobodave left	01:11
didi	Thank you all. This is a very nice community. I will do it.	01:11
doener	didi: you created a branch from master and added commits there. Of course the commits from master are also in that branch.	01:11
	didi: as you described it, master just lags behind the other branch	01:12
	→ hobodave joined	01:12
didi	doener: But in the new brach I deleted most of the files and severally changed those who remained.	01:13
	doener: the branch has actually little to nothing trace of master.	01:14
doener	didi: so? "git merge" will see that there are no commits in master that aren't in otherbranch, so it'll just fast-forward master. (Boils down to a "reset", but stops you from accidently losing stuff, in case that you _did_ have commits in master, that might be missing in otherbranch)	01:14
	didi: just do "git log otherbranch", it'll have the whole history of master	01:14
fujin	I've got loads of refs of remote branches, of remotes that I haven't fetched from for months - I often do code review of multiple peers' branches. Can I prune them in any way?	01:14
doener	didi: it's not about files, it's about the commit history	01:15
	fujin: "git remote prune" to prune remote tracking branches for which the remote branches don't exist anymore	01:15
	← Modius_ left	01:15
fujin	hmm	01:15
doener	fujin: or "git remote rm" to get rid of a remote incl. the remote tracking branches	01:15
didi	doener: Ok. I will try the merge then.	01:16
	→ Modius_ joined	01:16
	→ javatexan joined	01:17
	← schlort left	01:17
	→ Midnight_ joined	01:21
didi	doener: It worked! Nice. Which kind make me wonder... I though that the files that were in master will remain there, so I would end up with a big mess. Nice to know it is the opposite of that.	01:21
	← seangrove left	01:22
doener	didi: well, even if it wasn't a fast-forward, git would still see that "otherbranch" deleted the files, so that's a change to be considered while merging	01:22
	→ seangrove joined	01:23
	← rettub_ left	01:24
	→ rettub_ joined	01:24
fujin	doener: heh, these remote commands need builtin loops =D	01:27
	← Midnight left	01:27
	→ nis joined	01:27
	← didi left	01:29
	bryanl_ → bryanl	01:31
	→ wsc joined	01:33
	→ gotgenes joined	01:33
	← jrmuizel left	01:38
	← ashleyw_ left	01:38
	← whaley left	01:40
	→ Flathead joined	01:50
	→ wwalker joined	01:53
wwalker	given that file foo is under git control, I run git mv foo bar, then git commit -m yup, should foo be gone or there? should git status report the existence of a file foo?	01:54
	→ awx joined	01:55
	→ celling joined	01:55
doener	wwalker: gone and no	01:56
	→ wycats joined	01:56
wwalker	doener: foo isn't gone, and I just edited foo and git status shows it as modified	01:56
doener	wwalker: and is "bar" around?	01:57
	fujin always uses plain mv for this	01:58
wwalker	doener: yes, foo and bar exist	01:59
doener	wwalker: hm, interesting... git version?	02:00
wwalker	1.5.4.7	02:00
	http://rafb.net/p/6R8mTn73.html - summary from history \| egrep '(vim\|git)'	02:01
doener	ok, that's ancient	02:01
	→ jrmuizel joined	02:02
wwalker	I'm okay with the bug since I'm a few versions behind, just wanted to know I understood what was supposed to have happend.	02:02
	← scarabx left	02:02
doener	wwalker: hm, I don't see that bug being mentioned as fixed in git.git's log, but I only had a quick peek, and can't reproduce here	02:04
	← texel left	02:06
wwalker	I've used git mv before. I see now that I did a git pull between the 'git mv' and the 'git commit'. nothing in the pull was even in the directory of the foo/bar stuff, but apparently that triggered a bug.	02:08
	→ kuadrosx joined	02:11
	→ awx_ joined	02:12
	← qrush left	02:12
	← awx left	02:13
	← parasti left	02:13
	awx_ → awx	02:13
tony_	how can i copy one file from a different branch to the current branch?	02:19
	← drewr left	02:20
tony_	just copy them to another dir?	02:20
doener	git checkout <commit-ish> -- <path>	02:22
	e.g. "git checkout foo-branch -- file-bla"	02:22
	note that that gets just the file as it is in that commit, files have no history of their own in git	02:23
	(actually, s/commit-ish/tree-ish/)	02:23
tony_	doener: thanks!	02:23
	← montylounge left	02:24
	→ cytrinox joined	02:25
	← commondream left	02:26
	→ girishr joined	02:30
	← Midnight_ left	02:30
	→ paulboone joined	02:34
	← wsc left	02:36
	← cytrinox_ left	02:38
	→ CardinalNumber joined	02:39
	→ wsc joined	02:40
	← scientes_ left	02:41
	→ macrocat joined	02:42
	← neunon left	02:42
	cpg → cpg\|away	02:43
	→ AkumaStreak joined	02:44
	→ tairno joined	02:45
	→ scientes joined	02:45
	→ tjafk2 joined	02:51
	← LiamH left	02:51
	← ProperNoun left	02:54
	→ rovalent joined	02:55
	→ [1]intripoon joined	02:55
	← qhoxie_ left	02:58
	→ bonch joined	02:58
	→ Jinzhu joined	02:59
	← nis left	02:59
	← reprore left	03:00
	→ reprore_ joined	03:00
	← markelikalderon left	03:03
	← reprore_ left	03:03
	← wagle left	03:03
	← wsc left	03:03
	→ reprore_ joined	03:04
	← tjafk1 left	03:07
	← AAA_awright left	03:07
	→ AAA_awright joined	03:08
	Modius_ → Modius	03:11
	← intripoon left	03:11
	← AAA_awright left	03:12
	→ AAA_awright joined	03:13
	← AAA_awright left	03:13
	→ intripoon joined	03:13
	→ AAA_awright joined	03:13
	← AAA_awright left	03:13
	→ AAA_awright joined	03:13
	← AAA_awright left	03:13
	→ AAA_awright joined	03:14
	← AAA_awright left	03:14
	→ AAA_awright joined	03:14
	← AAA_awright left	03:14
	cpg\|away → cpg	03:14
	→ AAA_awright joined	03:14
	← AAA_awright left	03:14
	← snogglethorpe left	03:15
	→ AAA_awright joined	03:15
	← AAA_awright left	03:15
	→ AAA_awright joined	03:15
	← AAA_awright left	03:15
	→ AAA_awright joined	03:15
	← AAA_awright left	03:15
	← AkumaStreak left	03:15
	→ AAA_awright joined	03:15
	← AAA_awright left	03:15
	→ AAA_awright joined	03:16
	← AAA_awright left	03:16
	→ reprore__ joined	03:16
	→ AAA_awright joined	03:16
	← AAA_awright left	03:16
	← reprore_ left	03:16
	→ AAA_awright joined	03:16
	← AAA_awright left	03:16
	→ AAA_awright joined	03:16
	← AAA_awright left	03:16
	← Radar left	03:17
	→ AAA_awright joined	03:17
	← AAA_awright left	03:17
	→ AAA_awright joined	03:17
	← AAA_awright left	03:17
	→ AAA_awright joined	03:17
	→ Radar joined	03:18
	← rovalent left	03:19
	→ altrux joined	03:20
	← AAA_awright left	03:20
	→ AAA_awright joined	03:21
	→ Leemp joined	03:21
mugwump	darn, pre-commit can't stop the commit can it	03:22
	strange, it's supposed to	03:24
	← celling left	03:24
RandalSchwartz	is it enabled?	03:24
cehteh	yes so i thought ..	03:24
mugwump	I'm seeing the error message my script is printing	03:24
RandalSchwartz	ok	03:24
	.... Exiting with non-zero status from this script causes the git-commit to abort.	03:25
cehteh	are you sure it passes an exit code up .. or maybe it forks something but returns success by itself?	03:25
RandalSchwartz	are you?	03:25
mugwump	Well, if 1 is non-zero, then yes	03:25
cehteh	nah 1 is zero :)	03:25
	← Modius left	03:25
mugwump	oh bugger	03:26
RandalSchwartz	maybe this - This hook is invoked by git-commit, and can be bypassed with --no-verify option.	03:26
cehteh	almost .. :P	03:26
mugwump	aha ... it is a subshell	03:26
cehteh	so i thought	03:26
mugwump	\| while ( ... )	03:26
cehteh	yes	03:26
RandalSchwartz	substandard shell	03:26
	→ Modius joined	03:26
cehteh	exit $? or something like that	03:26
mugwump	annoying that ... and you can't pass variables back up either	03:26
	← paulboone left	03:26
RandalSchwartz	code it in perl :)	03:27
cehteh	perl wizards must feel retarded when using shell :)	03:27
RandalSchwartz	just limited	03:27
	→ travisjeffery joined	03:27
RandalSchwartz	like wearing heavy boots	03:27
	→ esparkman joined	03:27
mugwump	wahey	03:28
cehteh	hehe	03:28
	← [1]intripoon left	03:29
	← kukks left	03:30
	→ Modius_ joined	03:32
	← tairno left	03:32
	→ iron joined	03:33
cehteh	mugwump: i dont know your code, but maybe you can use {} instead () ... that doesnt fork a subshell	03:34
mugwump	ah, I just used $? after the while loop	03:35
	just submitted a patch including that to the list	03:36
	← Radar left	03:37
	→ bobmcw joined	03:37
	→ Radar joined	03:37
	← __iron left	03:39
	← javatexan left	03:40
	← DeafT0n3 left	03:41
	← Leemp left	03:43
	→ sergiodj joined	03:43
	← bobmcw left	03:45
	→ qhoxie joined	03:45
	← Modius left	03:47
	cpg → cpg\|biab	03:47
	→ rubydiamond joined	03:48
	← mw left	03:49
	→ Leemp joined	03:49
	→ mw joined	03:50
	← mw left	03:52
	← rubydiamond left	03:54
	→ Modius joined	03:56
	← Leemp left	03:59
	← LotR left	04:00
	merlin83_ → merlin83	04:02
	← Modius_ left	04:03
	← travisjeffery left	04:07
macrocat	I could never stand looking at perl code	04:10
RandalSchwartz	I have a hard time with cyrillic	04:10
	I don't know how all those russians stand it	04:10
	same thing with greek	04:10
	all those weird shapes	04:10
	those greek people must be weird.	04:11
	→ savonarola joined	04:11
	← ericindc_ left	04:11
mugwump	yeah. and how do the chinese remember all those thousands upon thousands of glyphs?	04:11
RandalSchwartz	just pointing out, your distaste of Perl cannot be universal. :)	04:11
	it's a matter of familarity	04:11
	and it's not only possible, it's trainable.	04:11
	→ JEEBcz joined	04:11
	← savonarola left	04:12
	← scientes left	04:13
	→ BugsBunnyBR joined	04:17
	← macrocat left	04:18
	mithro_ → mithro	04:27
	← BugsBunnyBR left	04:28
	ReAn` → ReAn	04:33
	→ nis joined	04:33
	→ neunon joined	04:36
	← JEEB left	04:38
	← releod left	04:40
	→ scientes joined	04:42
	→ paulboone joined	04:44
	→ bobmcw joined	04:45
	→ Gebolze joined	04:56
	← Gebolze left	04:56
	→ postmodern joined	04:56
postmodern	where does one report bugs for git?	04:57
mugwump	the list	04:57
nis	postmodern: /topic	04:59
mugwump	but if you like we can triage your issue here for you	05:00
postmodern	nis, the topic appears to not contain the word bug	05:00
mugwump	git does not use a bugtracker	05:01
postmodern	a friend just ran into an issue where git is using 100% cpu for over 45 minutes	05:01
mugwump	doing what?	05:01
nis	:-)	05:01
postmodern	wish there was a bug tracker i could send him to	05:01
nis	postmodern: but it has the mailing list address	05:01
	← jrmuizel left	05:01
postmodern	trying to commit a 100mb text file of course :)	05:01
mugwump	sending the message to the mailing list is a bit like a bug tracker	05:02
postmodern	of course 100mb is huge	05:02
nis	Ah.	05:02
postmodern	but 45 minutes of 100% cpu sounds like a bug	05:02
mugwump	you get a ticket which is your message on the list, and the follow-ups are the ticket updates	05:02
	→ Tuomas joined	05:02
mugwump	there are three statuses: unanswered, answered, and murky	05:02
nis	postmodern: not sure. 100mb is really huge.	05:02
	← curvature left	05:02
postmodern	nis, 100mbs is small 1gig is huge	05:03
mugwump	is this running 'git add' ?	05:03
	it shouldn't take significantly longer than piping the file to gzip	05:03
	unless you were, say, doing it over NFS	05:03
nis	postmodern: we are talking about _tracked_ file content.	05:03
	postmodern: post (you or your friend) an email to the mailing list.	05:05
mugwump	http://git.or.cz/gitwiki/SoC2009Ideas#head-2cdf2f7bd7667427d1e20c714ca33bd92aaa4905 might be relevant too	05:05
	→ curvature joined	05:05
	← reprore__ left	05:05
mugwump	in particular the linked talk on using git for media repositories (making some assumptions about what your 100mb text file is)	05:06
postmodern	mugwump, ah he says it's a new file	05:06
	→ reprore_ joined	05:06
mugwump	ok... but is this at 'git add' or 'git commit' time? Or some other time like push?	05:06
	→ travisjeffery joined	05:07
mugwump	also, which OS and filesystem?	05:07
nis	and hardware?	05:07
	← ia left	05:08
mugwump	there are lots of operations in git which don't work awfully well when the working set doesn't fit in memory	05:08
	'pack-objects' being one of those	05:08
postmodern	mugwump, git commit	05:08
	mugwump, on debian-stable / ext3	05:08
mugwump	that's not git 1.4.4.4 is it?	05:09
	oh, wait, stable ... ie lenny	05:09
doener	that's old-stable IIRC	05:09
	→ ia joined	05:09
mugwump	no I think you're good to go with your bug report ... that does sound weird	05:09
postmodern	mugwump, alright	05:09
	mugwump, so i should send my friend to the gmame URLs and subscribe that way?	05:10
mugwump	I'd also include something like output of 'vmstat 5 5' from while it is running	05:10
	I guess so	05:10
doener	hm, only thing that comes to mind would be the rename detection bug that caused objects to be read too often	05:10
	IIRC commit always runs status, which does the rename stuff	05:10
mugwump	still... that's a lot of time	05:10
doener	yeah, would only make sense if lots and lots of files were changed, causing the 100MB blob to be read again and again	05:11
mugwump	a listing of 'ps fvx' of the running processes might help	05:11
	→ scientes_ joined	05:12
postmodern	apparently he said it stopped finally	05:12
	→ jony joined	05:12
	← scientes_ left	05:12
postmodern	so it's not an infinite loop	05:12
mugwump	maybe doener has it then	05:12
doener	postmodern: just as a quick test, could you ask him whether "git show -C" takes a long time now?	05:12
	→ scientes_ joined	05:13
doener	postmodern: (assuming that he didn't already do yet another commit)	05:13
	otherwise, that needs to be "git show -C $bad_commit", where $bad_commit is the one that took so long to create	05:13
nis	postmodern: may I ask how he did a so big text file?	05:15
postmodern	nis, i want to know the same thing	05:16
doener	ipsum lorem ...	05:16
mugwump	it's the sort of bizarre use case you have to expect to end up patching core git a bit to deal with imho	05:16
doener	ehrm, lorem ipsum... that was around	05:17
	s/was/way/ gives up	05:17
	← nurey left	05:18
	← dreiss left	05:18
nis	i don't see any interesting use case for so big saved ipsum lorem	05:18
mugwump	well it might be a database dump I guess	05:19
nis	sounds much interesting	05:19
	but as a dump is a dump; it should be compressed _before_ git add.	05:20
doener	uhm, no	05:20
	→ SRabbelier joined	05:21
	← agib left	05:21
mugwump	in general you don't want to stream compress anything going in to git	05:21
	it gets in the way of the delta compression	05:21
nis	I know. But the content is not expected to change.	05:23
	← scientes_ left	05:25
mugwump	until you update the dump in a new revision	05:25
nis	any classical archiving system will be more efficient to an archive goal, IMHO.	05:28
mugwump	yeah, it's a silly use case	05:29
Ilari	Doesn't commit also do summary (after actually committing), which includes diffstat?	05:31
	← paulboone left	05:32
	← giallu left	05:33
	→ tim_jenssen joined	05:35
	→ neuro_damage joined	05:35
neuro_damage	wow this is a new thing ...	05:35
	← altrux left	05:35
neuro_damage	I have a .git and just checked out this code modified it	05:35
	and I'm getting fatal: Not a git repository (or any of the parent directories): .git	05:35
	that's very different	05:35
	← octagon left	05:35
	→ Modius_ joined	05:35
	← wmoxam left	05:36
	← Modius left	05:36
nis	neuro_damage: the files under .git are not tracked.	05:38
	→ Weaselweb joined	05:39
	→ davidfetter joined	05:39
Ilari	neuro_damage: In fact, there is code which prevents adding any files under directory named '.git'...	05:40
nis	what files do you want to track?	05:41
	→ Weems joined	05:45
	cpg\|biab → cpg	05:49
	→ priidu joined	05:49
	→ eno_ joined	05:49
	← eno left	05:50
	← johan-s left	05:55
	← kuadrosx left	05:57
	→ ankit9 joined	06:00
	← ankit9 left	06:00
	→ Deiz joined	06:06
	→ aersd joined	06:07
aersd	I run git-log --after="2009-03-31" but I'm getting commits with dates in December and older! what's wrong?	06:08
Deiz	I've had a recurring issue where cloning a repo will just hang at "Initialized empty Git repository in ...", no DNS issues or anything, it just sits there forever until I terminate it.	06:09
	→ psankar joined	06:10
nis	Deiz: is the repo public?	06:15
Deiz	nis: Yes. I've had this issue with both the kernel and VLC repos	06:15
	← samjam left	06:15
	← broonie left	06:15
	→ broonie joined	06:15
	← dkr21 left	06:15
	→ dkr21 joined	06:15
nis	Deiz: both? I have no problem with any of them.	06:15
	Deiz: what command line do you type?	06:16
	→ wsc joined	06:16
	← bonch left	06:17
	← PugMajere left	06:17
	→ PugMajere joined	06:17
	← numbah left	06:17
	→ Stravy joined	06:17
	← luke-jr left	06:17
	→ numbah joined	06:17
Deiz	nis: git clone --mirror repo-path save-name	06:17
nis	what is the repo-path?	06:18
	→ luke-jr joined	06:18
Deiz	For the kernel, "git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git"	06:18
	→ samjam joined	06:19
Ilari	Deiz: How long you did wait? There should be some kinds of timeouts...	06:19
nis	Deiz: works as expected here.	06:19
Deiz	There is a timeout, if I wait about half an hour.	06:20
nis	Deiz: do you have a firewall?	06:20
Deiz	Nope.	06:20
nis	Deiz: a router?	06:20
Deiz	Nope, connected directly to a dumb switch	06:20
nis	git --version?	06:21
	→ pnkfelix joined	06:21
Deiz	1.6.2.4	06:21
	← davidfetter left	06:21
Deiz	I tried the same line on this system, and it's doing what it should. Counting objects, compressing, receiving.	06:21
	This issue has cropped up on both my desktop and laptop, with varying versions of git.	06:22
nis	sounds like you have a LAN weird issue.	06:22
	→ johan-s joined	06:23
Ilari	Deiz: Could it be that due to long response time, something on transmission path timeouts (of course, there shouldn't be anything stateful in the path) and as result, the reply packets don't make it back?	06:23
Deiz	Well	06:23
pnkfelix	hopefully quick question: I am trying to use git-svn to work with an SVN repository with branches	06:23
	→ khmarbaise joined	06:24
	→ galderz joined	06:24
Deiz	That could be. The timeout that happens after around half an hour is from the remote end.	06:24
	Remote hung up, or something to that effect.	06:24
	It could also be that my copy of git isn't properly transmitting requests, and just sits idle with a connection open.	06:24
pnkfelix	is the statement made by Thomas Zander in response to 'n0ha' in the comments here: http://labs.trolltech.com/blogs/2009/04/03/two-kde-svn-branches-and-git/ correct?	06:24
	← gotgenes left	06:25
	← travisjeffery left	06:25
Ilari	Deiz: That error could also come from connection timing out due to packets from remote end getting lost.	06:25
mugwump	pnkfelix, not at all	06:25
doener	pnkfelix: no	06:25
mugwump	however if you are using non-standard layout you'll need to use multiple git-svn remotes	06:26
	using repositories as deeply intermingled as kde's SVN repo, where subpaths are moved between projects regularly, is quite a difficult topic though	06:26
pnkfelix	ah. I am using a non-standard layout; I used -b branches// to deal with it	06:26
mugwump	that worked? wow	06:27
	cool :)	06:27
pnkfelix	maybe that is why I am having such difficulties	06:27
Ilari	Deiz: Does it give the OS-level error that occurs, like 'Connection Reset By Peer' or 'Timeout'?	06:27
doener	since some 1.6.x version or so	06:27
pnkfelix	I am not sure how well it "worked" :(	06:27
	→ thiago_home joined	06:27
doener	pnkfelix: well, what kind of problems do you see?	06:27
pnkfelix	My attempts to use "git-svn dcommit -n" and "git-svn rebase" while on a local branch off of the remote svn branch consistently yields the error msg:	06:28
Deiz	Ilari: I'm not sure. I've only seen it once	06:28
pnkfelix	Unable to determine upstream SVN information from working tree history	06:28
Deiz	Starting up another clone attempt, I'll report back when it errors	06:29
doener	pnkfelix: did you use "git merge"? Or filter-branch? Or some funky rebase -i?	06:29
pnkfelix	But my invocations of "git-svn rebase" and "dcommit -n" work when i'm on the master	06:29
	→ reuss joined	06:29
Ilari	Deiz: Can you dump the protocol exchange?	06:29
pnkfelix	I do not think I have used merge, and I certainly have not used filter-branch or rebase -i; I am git-newbie.	06:29
	the git log for the remote branch reflects the state of the svn branch, so I have been assuming that it is still in whatever pristine state it was in at the outset of the clone	06:30
Ilari	Deiz: At least the packets sent and received. The remote end should always eventually reply, even if just by closing connection/resetting it if it timeouts.	06:30
doener	pnkfelix: I gues "git svn info" also fails?	06:31
	← nis left	06:31
pnkfelix	while on the branch, yes	06:32
	→ aspotashev\|eeepc joined	06:32
	→ giallu joined	06:32
doener	pnkfelix: ok, "git log -1 --grep=git-svn-id", please pastebin the output	06:32
pnkfelix	(of course, its a remote branch so I'm not really on it, but rather on an unnamed local branch of the remote branch)	06:32
	← postmodern left	06:33
pnkfelix	OH	06:33
	that helped	06:33
doener	how?	06:34
pnkfelix	I did the initial import off of a local copy of the svn repository, generated via svnadmin dump and load	06:34
doener	ah, so the svn-id-info lines are all wrong now?	06:34
pnkfelix	and there's a trick to then get git-svn to redirect its view so that its pointing at the right url	06:34
	and apparently the trick worked for the master	06:34
doener	s/svn-id-info/git-svn-id/	06:34
pnkfelix	but the git log output has in its output:	06:35
	git-svn-id: file:///Users/pnkfelix/Dev/larcenydev/import2/orig/svn-archives/branches/pnkfelix/incrsumz@6180 7bdee66d-2101-0410-b94e-df48c4b400c7	06:35
doener	pnkfelix: which trick? I know of a few of them	06:35
pnkfelix	I think it was this one, I remember the caveat about having to get at least one update from svn: http://theadmin.org/articles/2008/9/30/git-svn-switch-to-a-different-a-svn-url	06:37
	← robinr left	06:37
	→ tal67 joined	06:37
pnkfelix	but my .git/config does not have separate url for the branches... so I guess this trick was not as thorough as I had hoped. Shoot.	06:37
doener	pnkfelix: that's one of the more hackish, less reliable ways	06:38
	joshthecoder → joshthecoder_afk	06:38
doener	pnkfelix: the easiest way (which unfortunately seems to fail sometimes) is to set rewriteRoot in the config to the old url, and change the url entry to the new url	06:38
pnkfelix	that's the second suggewstion in the git.or.cz page that the above linked to, right?	06:39
doener	pnkfelix: bad luck that the trick you used would mean that master breaks then...	06:39
pnkfelix	http://git.or.cz/gitwiki/GitSvnSwitch	06:39
	There's the caveat that you can only apply this once, right? But that's okay, if the repository moves then I'll cross that bridge then I guess	06:39
	← xinming left	06:40
doener	pnkfelix: that takes the problem into consideration right away, and does it at clone time. That prevents the "wrong" git-svn-id lines	06:40
	pnkfelix: I meant it the other way around, to get it to work when the git-svn-id stuff is already wrong	06:41
pnkfelix	Oh doing it will break my use of the other trick? I did not infer that from the text at git.or.cz...	06:41
mugwump	I wonder how thiago_home is getting on with the kde.git conversion. if it was anything like perl.git it can easily eat a year of free time..	06:41
	→ ph^ joined	06:41
doener	pnkfelix: if you look through master's history, you'll see the old and the new url	06:42
	pnkfelix: and adding the rewrite root stuff now makes the old url work and breaks the new one. And you can't use the old trick to go back to the old url (nothing to fetch from there)	06:42
pnkfelix	hmm.	06:43
	→ lohrun joined	06:43
pnkfelix	so the dumb obvious thing is to throw away my clone and make a new one	06:43
doener	yeah, that's the brute force approach ;-)	06:43
	→ meyering joined	06:44
pnkfelix	(and either do the --rewrite-root thing at the outset as described on git.or.cz or just suck it up and grab the repos from the original svn+ssh:// source url)	06:44
	are there other options like writing a script to edit the git logs to change all the urls?	06:44
doener	pnkfelix: hmhm, if you do "git log --no-walk --remotes", how many of the commits that are shown have the new url, and how many have the old url?	06:45
pnkfelix	or are the urls embedded in places that I won't see in textual output?	06:45
doener	pnkfelix: they're just part of the commit messages	06:45
	→ bentob0x joined	06:45
pnkfelix	on the remote branch, there are 207 occurrences of file:/// url in the log, and 0 occurrences of svn+ssh:/// ...	06:46
doener	pnkfelix: you have 207 branches?	06:46
pnkfelix	no I must have misread my output, sorry	06:46
	% git log --no-walk --remotes \| cat \| grep file:/// \| wc	06:47
	207 621 30785	06:47
	% git log --no-walk --remotes \| cat \| grep svn+ssh:/// \| wc	06:47
	0 0 0	06:47
	% git branch	06:47
	* (no branch)	06:47
	cavos-glut-play	06:47
	doc-mods	06:47
	hax0rz	06:47
	incrsumz	06:47
	master	06:47
	plt-4.x-support-via-r5rs-program	06:47
	pnkfelix/incrsumz/maybe-bogus	06:47
	% git checkout master	06:47
	Previous HEAD position was 5cf7ce1... Implementation of code to optionally bias the major gc region selection policy.	06:47
	→ dohzya joined	06:47
pnkfelix	Switched to branch "master"	06:47
	% git log --no-walk --remotes \| cat \| grep file:/// \| wc	06:47
	207 621 30785	06:47
	% git log --no-walk --remotes \| cat \| grep svn+ssh:/// \| wc	06:47
doener	:-(	06:47
pnkfelix	0 0 0	06:47
	% that was what I was basing my report on	06:47
	wait, is 207 branches believable? It migt be, let me check trac	06:48
doener	hm, that would indeed mean 207 remote tracking branches... "git branch -r \| wc -l"	06:48
mugwump	are you importing from svnsync or svk clones?	06:48
	RandalSchwartz wipes the paste from his creen	06:48
mugwump	there are switches for those	06:48
	→ aspotashev joined	06:48
pnkfelix	I'm just importing from an svn repository with a slightly non-standard layout	06:48
mugwump	oh right. was just wondering why they were file:/// urls	06:49
pnkfelix	the file://// urls are from a local copy of the repository	06:49
	I was getting frustrated by the time it was taking to clone	06:50
	and wanted to remove teh possibility that network traffic was part of the problem	06:50
mugwump	you could use filter-branch	06:50
pnkfelix	(and I have svnadmin access to repository)	06:50
mugwump	there's thiago's svn-fast-import-all or whatever	06:50
	it's tons faster... but difficult to get working	06:50
pnkfelix	I am not familiar with filter-branch. I infer the idea is to focus attention on a subset of the branches/ structure... is filter-branch a git-svn thing, or a git thing?	06:50
mugwump	it's a git thing	06:51
	basically it would let you rewrite all those file:// bits in the commitlog to whatever	06:51
doener	pnkfelix: http://git.pastebin.com/m319fcd36	06:51
pnkfelix	ah yes I think that would be an approach I could digest	06:51
mugwump	eg git filter-branch --all --commit-filter="sed 's!file:/home/foo!svn://host/foo!	06:51
doener	pnkfelix: run on a copy of your repo! just replace old-url and new-url in the msg filter thing	06:52
	i.e. in the sed call	06:52
mugwump	yeah that looks more sophisticated	06:52
	you need to throw away .git/svn afterwards	06:52
	mediaslave → evalsaidem	06:52
mugwump	but that's fine - re-run git-svn init, first fetch will rebuild databases	06:52
	← evalsaidem left	06:53
doener	mugwump: thought of that, there's a find+xargs to kill the rev_map files (that's enough)	06:53
mugwump	aha	06:53
pnkfelix	you can rerun git-svn on an already present repository?	06:53
	rerun "git-svn init", I mean>?	06:53
doener	pnkfelix: sure, but there's no need to.	06:53
mugwump	not with doener's script	06:54
aersd	could anyone help me please? I run git-log --after="2009-03-31" but I'm getting commits with dates in December and older! what's wrong? Could rebasing have destroyed something?	06:54
doener	pnkfelix: one use case for using "git svn init" more than once is to have multiple svn remotes, e.g. when you have hairy svn setups	06:54
pnkfelix	okay I will try doener's script.	06:54
	(also sorry for flooding your screens; I did not understand what pastebin was when it was mentioned before)	06:55
doener	aersd: is the commit date from December, or maybe the author date?	06:55
pnkfelix	(but I infer now that it is this online communal thingie)	06:55
	← scientes left	06:56
	→ Bennid joined	06:59
	→ ferdy joined	07:02
pnkfelix	this won't be so bad	07:03
	so while this runs, would someone mind explaining to me what the role of .rev_map is?	07:03
	(connecting up git sha-ids to svn rev#'s?)	07:04
doener	pnkfelix: right	07:04
pnkfelix	so where does the script restore that state?	07:04
doener	pnkfelix: not at all, the next rebase/fetch will do that automatically on demand	07:05
	→ agib joined	07:05
doener	pnkfelix: the information is available in the git-svn-id lines, the .rev_map files just provide faster access	07:05
pnkfelix	okay. I assume you mean the next "git-svn rebase/fetch", right?	07:05
doener	yep	07:05
pnkfelix	Ah. I had not figured that out, but it makes sense now	07:06
doener	pnkfelix: hm, you still have "git-svn"? Which git version is that?	07:06
pnkfelix	no, Its actually "git svn"	07:06
doener	pnkfelix: maybe your version still uses the .rev_db format	07:06
pnkfelix	but I've gotten used to reading "git-svn" everywhere online	07:06
	→ Radar_ joined	07:06
pnkfelix	so while I'm used to typing "git svn" as a command, it seems like "git-svn" will remain a better thing to google for	07:06
doener	pnkfelix: yeah, or "git svn" (including the quotes)	07:07
pnkfelix	well clearly I go overboard in all directions there.	07:07
	okay so I understand the first two "lines" of the script	07:09
	what is the for-each-ref line and the reflog line doing?	07:09
	← bobesponja left	07:09
doener	pnkfelix: filter-branch creates backups of the references it rewrites	07:09
	pnkfelix: e.g. "refs/heads/master" is backed up as (by default) "refs/original/refs/heads/master"	07:09
pnkfelix	ah so those lines are pruning the backups	07:10
doener	pnkfelix: the for-each-ref command prints all refs in the refs/original namespace, i.e. all backups, the xargs command then passes them to "git update-ref -d" to delete them	07:10
pnkfelix	but I should be able to try out the effect of the rewrite without doing the last three commands then, right?	07:10
	or will the backups disrupt git-svn, confuse it in some way?	07:11
doener	pnkfelix: yep, everything after the "find" is cleanup stuff	07:11
	pnkfelix: the reflogs are logs that have an entry for each update changed the commit that the ref resolved to	07:12
	→ total joined	07:12
doener	pnkfelix: e.g. when you have "foo" checked out and commit, then "foo" and "HEAD" will get a new reflog entry	07:12
	← Radar left	07:13
doener	pnkfelix: using the reflog, you can go back to older states of the refs. Interesting when you need to undo (fast-forward) merges, or rebases. Or just want to look at the past	07:13
	→ smtms joined	07:13
	← bwalk\|uni left	07:13
total	hi guys, how to open/explode .git file, i tryed in bash, but i get error	07:13
pnkfelix	and the contents of the reflog are set up to expire based on some notion of elapsed time?	07:13
doener	the reflog command from that script empties all reflogs, because the existing entries reference the pre-filter-branch states, which are not to be used anymore	07:13
pnkfelix	and then become collectable by "git gc"	07:13
	?	07:13
doener	pnkfelix: yep, reflog entries expire after 90 days by default	07:14
	← thiago_home left	07:14
doener	pnkfelix: and when a reflog entry expires, some objects might becomes unreachable and then gc may garbage collect them	07:14
	pnkfelix: e.g. when you do "git commit", realizes that something is wrong and do "git commit --amend", then the first commit object (the bad one) is still around, and still referenced by a reflog entry, but not reachable through the branch head (it got replaced in the branch head's history)	07:15
	pnkfelix: so once the reflog entry for that old commit object expires, gc may get rid of it	07:15
	pnkfelix: IOW, by default, you have 90 days to discover your faults	07:16
	← Radar_ left	07:16
pnkfelix	I see. That is good to know; there has been discussion in my lab about how dangerous git's history rewriting operations are	07:16
	→ Radar joined	07:16
pnkfelix	if everything sticks around in the reflog for 90 days, that will counter some misconceptions.	07:17
doener	pnkfelix: http://colabti.org/irclogger/irclogger_log/git?date=2009-05-03#l2701	07:17
pnkfelix	what about the svn-fast-import script by thiago that mugwump mentioned before?	07:18
	Is its main improvement that its faster than git-svn ?	07:19
	← Radar left	07:19
mugwump	normally when you rewrite history you'll save the old history somewhere, like a separate repo or some refs which aren't normally cloned	07:19
pnkfelix	(I've been curious if there are tools that deal better with, mmmm, "strange" project histories	07:19
mugwump	well, that's a custom tool which converts svnadmin dump output into git-fast-import format	07:19
	last report I heard it was difficult to build, unless you knew the KDE development process well	07:20
	→ Radar joined	07:20
	← xkr47 left	07:21
total	please help guys, i have git file, how i should use it? i running on windows and 1.6.1.9 git version, please	07:22
	← hendry left	07:24
	→ RodP2 joined	07:24
	→ tazle___ joined	07:24
pnkfelix	total: what are you trying to do? clone a public project? Or make a repository of your own?	07:24
	→ jmesnil joined	07:25
	→ thiago joined	07:26
	→ bwalk joined	07:27
	→ reprore joined	07:27
	← reprore_ left	07:27
total	pnkfelix: clone	07:28
	← tazle_ left	07:28
total	oh i have found, but there is problem, i have explode all files in folder now	07:29
	tazle___ → tazle_	07:29
total	but i have no change history	07:29
pnkfelix	so you have run "git clone", given it a url, and it has now downloaded the files you want to see, but you want to inspect the history?	07:30
total	yes	07:30
	git file is in my pc, i have already cloned it	07:31
	i have all exploded files now	07:31
	← felipec left	07:31
total	but there is no documents at "Unstaged Changes" and "Stages Changes"	07:31
	hmz maybe i can create new repository with old this project files, than write new files?	07:32
pnkfelix	Are you using some sort of GUI interface to git? I have oonly worked with the command line tool (on Windows, Mac OS X, and Linux, but always the command line...)	07:33
	so I am a little confused by some of your references	07:33
total	yes, i am on interface	07:33
Radar	"exploded" files?	07:33
total	pnkfelix, thats allright	07:33
Radar	How many pieces would you say they are in?	07:33
	just a rough estimate	07:33
	please don't count them all, it may take too long.	07:34
total	i going to new repository with old files	07:34
	than i will overwrite new files on it	07:34
	and i will have all changes history	07:34
	thanks ;)	07:34
pnkfelix	...um you're welcome i guess	07:35
	→ dwave joined	07:35
	← ricky left	07:36
	← hyperair left	07:36
	→ ngng joined	07:36
	pnkfelix now wants to try out the Windows git GUI just to see if it says something about explosions	07:36
	→ xyz joined	07:37
pnkfelix	total: to see the history, you could select Repository..Visualize master's History	07:40
	it makes sense that your initial clone would not have anything listed in "Unstaged Changes" or "Staged Changes"	07:40
	dwave → asksol	07:43
pnkfelix	doener: when "git svn rebase" rebuilds the _revmap data, does it do it for all branches? Or only for a relevant subset being manipulated by the rebase command? (I guess I'm asking if the _revmap's are lazily constructed caches, or if they're all built monolithically/eagerly)	07:46
	← ph^ left	07:46
pnkfelix	hmm. My version of xargs does not like the -r option that doener's script included. What was that for?	07:47
total	pnkfelix, thanks!	07:48
pnkfelix	(or rather, the -0r option; it specifically complains about 'r' in particular...)	07:48
tokkee	pnkfelix: That's the same as --no-run-if-empty ...	07:48
	← gaveen left	07:48
tokkee	"This option is a GNU extension."	07:48
	→ ph^ joined	07:49
pnkfelix	hmm. sounds like it might be an important part of the script	07:49
	I'll see if I can get GNU's xargs via fink...	07:49
reaVer	hi	07:50
	← zerok left	07:51
reaVer	I got 1 or 2 commit objects I want to remove	07:51
	→ tbf joined	07:51
jjuran	reaVer: git rebase?	07:51
reaVer	jjuran?	07:52
pnkfelix	reaVer: remove all trace of? Or would it suffice to check in a reversion of them?	07:52
	→ Lothrik joined	07:52
reaVer	pnkfelix: all trace of	07:52
	← reprore left	07:53
reaVer	it's too large to submit to another repo(hangs my machine in the process) and it now has to be sent for git commit object integrity	07:53
pnkfelix	well it certainly seems like filter-branch could be a howitzer for the job.	07:53
	(based on my shallow reading of the help info for filter-branch)	07:54
	← agib left	07:54
	→ _daniel_ joined	07:54
_daniel_	how can i clone a remote repo if i'm under a proxy	07:55
jjuran	Won't removing that commit from the history of any pushed or fetched branch be sufficient to avoid the hang?	07:55
Lothrik	Anyone in here gotten msysGit to work under Windows 7 (Build 7100)? I had it working under 7000, but now it just crashes whenever I try to clone.	07:55
	→ gebi joined	07:55
	→ Lgb joined	07:55
	→ reprore joined	07:55
	← reprore left	07:56
reaVer	jjuran: I'm unable to push it in the first place, but yeah, removing those objects should avoid the hang	07:57
	→ ricky joined	07:57
jjuran	reaVer: man git rebase, and check out interactive mode.	07:57
	→ marcus_ joined	07:58
jjuran	man git-rebase, rather.	07:58
Gitbot	jjuran: the git-rebase manpage can be found at http://git.or.cz/man/git-rebase	07:58
reaVer	jjuran: doesn't that break integrity on the repo I push too?	07:59
jjuran	reaVer: But the push never succeeded, right?	08:00
reaVer	yes	08:00
jjuran	reaVer: You either need to fix your git so it can handle the push, or abandon the commit and replace it with smaller ones that have the same effect.	08:01
_daniel_	please, does anyone knows what to do if i'm behind a proxy and want to clone a remote repo	08:01
jjuran	reaVer: And the issue with rewriting published history doesn't apply, since you haven't been able to publish it in this case.	08:02
	← Lgb left	08:02
pnkfelix	_daniel_: find someone willing to clone it and host on http:// for you?	08:02
	← sergiodj left	08:02
	→ plasticine joined	08:02
_daniel_	pnkfelix: yes this would be a solution	08:02
	→ Lgb joined	08:02
_daniel_	pnkfelix: but , it is possible to use git + proxy :) , because i want also to do push's onto the remote repo	08:03
	→ naeu joined	08:03
pnkfelix	i figured you were probably already aware of that one; the docs I have read indicate that firewalls are a reason that http:// support is important	08:03
	→ gaveen joined	08:03
	← asksol left	08:04
	← jony left	08:06
	marcus_ → imarcusthis	08:06
	→ ilor joined	08:07
	→ hendry joined	08:07
hendry	in gitweb, how do i should the diff between html5 and master? http://git.webconverger.org/?p=ikiwiki	08:08
	→ dedmajor joined	08:08
reaVer	filter-branch --tree-filter is giving me file not found	08:11
	→ zaius joined	08:12
zaius	how do I go back to an earlier commit and break it up?	08:13
teuf	zaius: you can do that with git rebase -i (assuming you haven't pushed your work to a public repo)	08:14
wereHamster	zaius: how far back in the history is the commit?	08:14
	← Weems left	08:14
Lothrik	Where can I get msysGit 1.6.1? I can't seem to find anything older than 1.6.2.2 on http://code.google.com/p/msysgit/	08:14
zaius	wereHamster: it's not the most recent one	08:14
	teuf: i've been playing with rebase, and it scares me :)	08:14
	teuf: so i go back to the commit, reset to HEAD^, then commit the changes in separate commits?	08:15
	(and I haven't pushed it anywhere yet)	08:15
teuf	zaius: git rebase -i COMMITSHA1^ , you set the commit to 'edit', then you reset, edit, commit, ...	08:15
	and when you're done, git rebase --continue	08:16
	zaius: start by doing the work in a temporary branch if you don't feel comfortable enough with rebase	08:16
	this way you can just drop the branch if things go wrong	08:16
zaius	cool, thanks	08:17
	never done any branching before either - how would I do that?	08:17
	(i've just been copying my whole working directory whenever I try something that could break things) :)	08:17
pnkfelix	doener: thanks very much for the help. The script fixed my repository and I can finally commit. :)	08:17
teuf	zaius: git checkout -b somename	08:17
	will create a somename branch and moves your working directory to this branch	08:18
zaius	ok, so rebasing on the brach will only affect that branch?	08:19
	even though the commit that i'm editing was before the branch?	08:19
teuf	yeah, the other branches won't be modified, they'll keep references to the "old" commit	08:20
zaius	ok. and how do i merge that branch back in?	08:20
teuf	once you're comfortable with the changes, either you do the same thing on your master branch, or you rename the temporary branch to master	08:21
	→ jony joined	08:21
zaius	right. i think i need to do some more reading on how git branching works	08:21
	← pnkfelix left	08:22
	→ dwave joined	08:22
reaVer	nvm, got it all done	08:22
	tnx pnkfelix, who just left	08:23
	and thanks for your time jjuran	08:23
	→ j416 joined	08:24
zaius	teuf: thanks for the help	08:24
	→ gaveen_ joined	08:28
	→ eddyp joined	08:29
	← gaveen left	08:29
	← ndim left	08:32
	← naeu left	08:33
	→ gressie joined	08:35
	gressie → Guest47481	08:35
	← Guest47481 left	08:36
zaius	is there a way i can stash untracked files as well?	08:38
	→ naeu joined	08:38
aersd	doener: it's just "Date" on log. How do I check it?	08:39
drizzd	zaius: git add	08:39
zaius	drizzd: so, git add on a separate branch? and then reset back to it once i'm done?	08:40
tokkee	zaius: git-add does not add to any branch but to the index ...	08:41
	zaius: git add <file>; git stash -> <file> ends up in the stash	08:41
zaius	ah i see	08:42
	zaius is learning	08:42
	→ trepca joined	08:42
Knirch	is there a way of abort a cherry pick that resulted in a merge conflict, or do I have to commit it and then reset/rebase/...	08:43
	sdiz → sdiz-afk	08:44
tokkee	Knirch: git reset --hard?	08:44
Knirch	tokkee: that's possible without committing? hmm, I'll try	08:45
	doh, nice, thanks	08:47
	I'm so caught up in bk, there you have to finish the current action before doing anything else	08:48
zaius	is there a way to apply a commit backwards? i.e. to roll back a feature	08:49
wereHamster	zaius: git revert $commit	08:50
zaius	and then commit that?	08:50
wereHamster	no, revert creates a new commit that undoes $commit	08:50
zaius	perfect, thanks	08:50
	→ solydzajs joined	08:51
	← tango_ left	08:58
	→ Pe3k joined	08:58
	← voker57 left	08:59
Pe3k	hello, is there one simple command to get state of working directory which was after 1st commit?	09:00
smtms	git checkout <commitID>	09:01
Pe3k	smtms:thanks	09:02
	→ smeevil joined	09:02
	→ nachocab joined	09:03
nachocab	hi, is there a way to see a listing of all the changes that took place in a file in each of the commits that modified it?	09:04
smtms	nachocab, in diff format?	09:05
nachocab	just a printout	09:05
smtms	nachocab, printout of what?	09:05
nachocab	well, the thing with the +++ and ---. Is that the diff format?	09:05
	→ jaeckel joined	09:06
	← dwave left	09:06
	← Lothrik left	09:07
	→ lolmaus^_^ joined	09:07
smtms	nachocab, I think it is, although I'm not sure what you are describing	09:07
nachocab	you know, like the 'blame' screen you get on github	09:07
	all the changes that happened to a file	09:08
	in each of the commits	09:08
lolmaus^_^	I've installed Git via my package manager on Arch Linux. I've done a 'git clone'. But where do files go? I would like to set up my Trac to work with Git but i dunno where to point it.	09:09
smtms	nachocab, try "git log -p"	09:09
	nachocab, or "git blame"	09:09
	nachocab, or whatever it is that you want :-)	09:09
	lolmaus^_^, when you did "git clone", git printed some informative messages	09:10
	← gaveen_ left	09:10
smtms	lolmaus^_^, for example, where is the newly created repository	09:10
	→ gaveen_ joined	09:11
nachocab	smtms: thanks! :)	09:12
	← dedmajor left	09:13
lolmaus^_^	smtms, oh i see it. Can i change the default path?	09:13
smtms	lolmaus^_^, you can say "git clone <repoURL> <some path>" and it will go there	09:15
lolmaus^_^	smtms, thx	09:15
	Does anyone know of any Git+Trac tutorials?	09:16
	← nachocab left	09:16
	← tbf left	09:16
	← Bennid left	09:17
	→ ankit9 joined	09:18
	→ cedricv joined	09:19
	→ tbf joined	09:19
	→ dwave joined	09:19
	→ voker57 joined	09:20
	→ dedmajor joined	09:20
	→ ph^_ joined	09:23
	← gilimanjaro left	09:27
	→ ph^__ joined	09:27
	→ gilimanjaro joined	09:29
	→ beni- joined	09:29
beni-	hi	09:29
	→ ph^___ joined	09:29
	← voker57 left	09:30
	← gilimanjaro left	09:30
eMBee	good afternoon	09:30
	← ph^_ left	09:30
	← psankar left	09:32
eMBee	is is possible to show if files are new when looking at the --stat output?	09:32
beni-	I was following this tutorial here <http://scie.nti.st/2007/11/14/hosting-git-repositories-the-easy-and-secure-way>. Can I set the shell to /usr/bin/git-shell instead of /bin/sh for user 'git'?	09:32
Gitbot	beni-: I'm sorry, there's no such object: http://scie.nti.st/2007/11/14/hosting-git-repositories-the-easy-and-secure-way.	09:32
beni-	...?	09:32
	ah, just answered my question myself ^^	09:34
	(there is command="gitosis-serve beni" in authorized_keys)	09:34
	bye	09:34
	→ LotR joined	09:34
	→ dwave_ joined	09:35
	→ voker57 joined	09:35
	← beni- left	09:38
	→ ia_ joined	09:39
	→ sitaram joined	09:40
	← ia left	09:40
	→ agib joined	09:40
	→ dwave- joined	09:44
	← galderz left	09:45
	→ galderz joined	09:45
j416	How can I prune all unreferenced objects, both packed and unpacked, directly without expiration time? would git prune ; git prune-packed suffice?	09:45
	that is, I did git reset --hard to a previous commit, and I want to remove all traces	09:46
	← ph^ left	09:47
j416	(I want to find out how large the repo was a that time, this is a copy so I don't care about reflog or anything)	09:47
	← JasonWoof left	09:47
	← stuffcorpse left	09:47
	→ stuffcorpse joined	09:47
Grum	git reflog expire --expire=0 --all; git gc --prune=now i think	09:48
	requires a fairly new git to have the git gc --prune	09:48
j416	thanks (I'm running 1.6.2.4)	09:49
Grum	that should purge all dangling things, repack the complete packfile	09:49
j416	seems to work! thanks	09:49
	→ jfkw joined	09:50
	← aersd left	09:50
j416	I have been trying out git to do incremental backups of all my email	09:50
	← piroko left	09:50
	→ piroko joined	09:50
eMBee	j416: interesting: how is that going?	09:51
j416	my email box has grown from 663 mb (16689 files) to 668 mb (17573) during a period of about 8 weeks	09:52
	← ph^__ left	09:53
j416	I set it to commit changes every 24 hours since 2009-03-08	09:53
	the (packed) git repo at the start was 398 mb. Now, 8 weeks later, it's 399 mb.	09:53
	Pretty impressive	09:53
	→ psankar joined	09:53
	← voker57 left	09:54
	← piroko left	09:56
sitaram	in older gits, the sequence is 'git repack -adf; git prune', IIRC	09:56
	→ voker57 joined	09:57
eMBee	and how much mail did you get in that time?	09:57
	oh, never mins	09:57
j416	:)	09:57
sitaram	prune only removes loose objects, not stuff that got packed before you did the refloge expire	09:57
	← naeu left	09:58
	eMBee didn't scroll up far enough (with his eyes on the screen that is)	09:58
j416	sitaram: running that now, let's see what it ends at	09:58
	→ spacepluk joined	09:58
sitaram	one way to test this is repack, then reflog expire, and then prune	09:58
	each time running "git count-objects -v; du -sk .git" as a check	09:58
	each time => between each step	09:59
j416	this takes time heh	09:59
	writing objects now	09:59
	still 398 mb	10:00
	i think that was the size after the first commit though, i remember that number	10:00
	→ riovena joined	10:00
	← tony_ left	10:01
sitaram	and that's the other thing... I suggest a temp branch and a few gratuitious git-filter-branches with some changes; otherwise nothing happens	10:02
	→ tango_ joined	10:03
	→ naeu joined	10:03
	← thiago left	10:03
spacepluk	hi, Is there any limitation to the files I can have inside .git? I'd like to have a cvs working copy there to use with cvsexportcommmit.	10:03
j416	sitaram: or I could just hard reset, kill the .git directory, and remake :)	10:03
sitaram	yeah but that wouldn't be educational :-)	10:04
Grum	sitaram: hmm i thought prune=now actually did repack the whole thing	10:04
sitaram	and boring on irc :-)	10:04
	j416 doing	10:04
dirker	Heya, does git archive --remote work with git://? Does it need some extra settings in .git/config?	10:04
sitaram	Grum: I meant in older gits...	10:04
eMBee	spacepluk: shouldn't be a problem, just make sure you don't have anything in gitignore	10:05
Grum	dirker: yes it should i think	10:05
dirker	ah, git daemon complains [15129] 'upload-archive': service not enabled	10:07
	→ rolfb joined	10:08
	ph^___ → ph^	10:09
	→ thiago joined	10:09
spacepluk	eMBee: thanks, It's a bare repository but it has some stuff in .gitignore (mostly compiled binaries, and tmp files), is this a problem?	10:11
	← dwave_ left	10:13
dirker	ok, got it, its the daemon.uploadarch configuration option (see git-daemon manpage)	10:14
	→ piroko joined	10:16
eMBee	spacepluk: should be fine, as long asyou don't have anything cvs specific in there. some people put CVS in gitignore to mke it easier to share a repo between git and cvs	10:18
	→ gilimanjaro joined	10:18
	→ javatexan joined	10:20
	→ piroko_ joined	10:21
eMBee	how can i get a list of files that are added in a commit?	10:22
	eMBee can't figure out which command or option is needed for that	10:22
	← deavid left	10:23
	→ deavid joined	10:23
spacepluk	eMBee: thanks :)	10:25
	← mike_m left	10:26
	→ piroko__ joined	10:26
	→ mike_m joined	10:27
charon	eMBee: git show --diff-filter=A --name-only; or if you want _only_ the filenames, something like 'git diff --name-only --diff-filter=A cmt^ cmt'	10:28
	charon briefly wonders why show/diff --diff-filter=M turn up different lists for a merge commit	10:28
	eMBee tries that, thanks	10:29
	gaveen_ → gaveen	10:29
	← eddyp left	10:30
	← khmarbaise left	10:30
	→ piroko___ joined	10:31
	← stuffcorpse left	10:31
	→ stuffcorpse joined	10:32
	→ markelikalderon joined	10:32
	→ Modius joined	10:33
	→ ndim joined	10:33
eMBee	charon: thank you, that works, nd it lead me to what i was actually looking for: --name-status	10:34
	← piroko left	10:34
	→ Modius__ joined	10:34
eMBee	though in reality --diff-filter is really what i need	10:36
	← Modius_ left	10:38
	→ Yuuhi joined	10:38
	← piroko_ left	10:38
	← gabeodess_ left	10:42
	← Stravy left	10:43
	→ gitte joined	10:43
	← piroko__ left	10:43
	← Modius__ left	10:45
	→ Modius__ joined	10:45
	→ aziz_ joined	10:45
	← gebi left	10:48
	← Modius left	10:51
	→ ashleyw joined	10:52
	← markelikalderon left	10:52
j416	sitaram: I tried killing .git and remaking it, and now, the size is 399 mb after git gc, and after the repack you suggested	10:53
	o_O	10:53
sitaram	but there wasn't any garbage to be collected so why would it change?	10:54
j416	running Grums line once again..	10:54
	sitaram: because the other one went down to 398mb	10:54
	with the exact same content	10:54
	(apart from those commits and additional objects that had been pruned	10:54
sitaram	try a different --window and --depth to repack -adf?	10:54
j416	hm, still no chanfe	10:54
	change**	10:54
	hm	10:55
	nah	10:55
	the conclusion is however that the size of my mail backup hardly changed over an 8 week period	10:55
	:)	10:55
	which is very neat	10:55
sitaram	you're not a manager, yeahhh!!	10:56
j416	what? heh	10:56
	sitaram knows managers who measure their effectiveness by number of emails sent and received :)	10:56
j416	lol	10:56
sitaram	or at least boast about it as if it were a measure :)	10:57
j416	hehe	10:57
	sdiz-afk → sdiz	10:57
	← dwave- left	10:58
	→ markelikalderon joined	10:59
	← dwave left	10:59
	← Flathead left	10:59
	→ knitt1 joined	11:02
	← jmspeex left	11:05
	→ bersace_nerim joined	11:07
bersace_nerim	Salut !	11:07
	Comment controller la syntaxe des fichiers sources d'un dépôt côté serveur ?	11:08
	← thiago left	11:08
	← knittl left	11:10
rvsjoen	is cherry-picking the way to go when I want to apply only a single commit from another branch than the one i'm currently working on ?	11:10
	← smeevil left	11:10
	→ thiago joined	11:12
charon	bersace_nerim: note that this is an english-speaking channel. not sure i understand your request right, but if you are indeed asking for a way to check the syntax of files committed, look into man githooks (either 'pre-receive' or 'update' will do)	11:14
Gitbot	bersace_nerim: the githooks manpage can be found at http://git.or.cz/man/githooks	11:14
	← _daniel_ left	11:14
bersace_nerim	charon: oops, i though i was on #git-fr sorry	11:15
	thx	11:15
sitaram	rvsjoen: yes	11:16
	you could also format-patch \| am, but that is essentially the same thing	11:17
	← markelikalderon left	11:17
	← tal67 left	11:21
bersace_nerim	Hi	11:22
sitaram	hi	11:22
bersace_nerim	how to check source code syntax on server side ?	11:22
GyrosGeier	hmm	11:22
	GyrosGeier has fun	11:22
sitaram	bersace_nerim: charon answered this above (man githooks)	11:22
Gitbot	bersace_nerim: the githooks manpage can be found at http://git.or.cz/man/githooks	11:22
GyrosGeier	mission: merge two trees without common ancestry	11:23
	← Jinzhu left	11:23
	← total left	11:23
sitaram	GyrosGeier: not difficult at all, I think...	11:23
	← dedmajor left	11:24
	→ gabeodess_ joined	11:24
	→ dwave joined	11:25
	dwave → asksol	11:25
	← Aides left	11:26
	→ Bonaparte joined	11:27
	→ Aides joined	11:28
Bonaparte	How can I print a list of untracked files?	11:28
	→ gebi joined	11:28
	→ tal67 joined	11:29
sitaram	git ls-files -o?	11:29
	← gabeodess_ left	11:29
	← Radar left	11:29
Bonaparte	Thanks	11:30
sitaram	git ls-files --exclude-standard -o if you want the list to exclude "ignored" files	11:30
	→ DavidKlein joined	11:30
Bonaparte	I was thinking git --help would print all available commands	11:30
	One more question	11:31
	→ jmspeex joined	11:31
Bonaparte	When I edit a file, do I have to explicitly type git add <filename> before commiting?	11:31
jast	yes, unless you use git commit -a	11:32
Bonaparte	I see	11:32
	→ dedmajor joined	11:32
Bonaparte	I am unable to figure out how to add a file to the ignore list	11:33
sitaram	cho filename >> .gitignore	11:33
	echo*	11:33
Bonaparte	git ignore <filename> doesn't seem to be the correct usage	11:33
sitaram	it's just a plain text file, edit it)	11:33
	there's not git ignore command I think	11:33
	→ MisterN joined	11:33
	→ commondream joined	11:34
	→ mikl joined	11:34
	← ankit9 left	11:35
	→ Ryushin joined	11:35
	→ d0k joined	11:36
Ryushin	I'm trying to learn how to use git to pull Linus tree and two other branches. Can someone look over my set up commands and let me know if this is going to do what I want: http://www.pastebin.ca/1412036	11:37
charon	bersace_nerim: to be a bit more specific, in both hooks you get 'post' revisions which you can then use to look at a list of filename-blob pairs within using 'git ls-tree $post', and then finally get at the content of the blobs using 'git cat-file blob $sha'	11:39
	← voker57 left	11:40
sitaram	Ryushin: I'd create a throw-away merge branch instead of merging on the master	11:40
charon	bersace_nerim: or for a really thorough job, you'll need to check the whole pre..post range	11:40
Ryushin	sitaram: What is the difference?	11:41
	Or if there are docs that describe it, even better.	11:41
sitaram	Ryushin: http://www.pastebin.ca/1412038	11:42
	Ryushin: if the merge result is going to be used only for testing, and not coding and feeding changes back, it is simpler to make it temporary. You just track the 3 branches (one big one and the 2 small ones) you want independently	11:43
	→ voker57 joined	11:43
Ryushin	sitaram: Thanks. I'll read up on the changes you made.	11:43
	← fynn left	11:43
Ryushin	Thanks sitaram	11:43
sitaram	charon: you want to chip in on this; I'm guessing you'd know more than I do about why a temp merge branch is a better idea :)	11:44
	?	11:44
	← asksol left	11:46
charon	sitaram, Ryushin: from a workflow POV, because you're supposed to make a branch for every "activity" (topic, use) and this temporary use is different from the main reason master exists	11:46
	from a practical POV, because you otherwise have to reset --hard back the master, which is a loaded gun waiting for a foot to pass by	11:46
Ryushin	Well, no point in using a gun anywhere near my feet. So the temp merge branch is the way to go. Thanks much.	11:48
sitaram	charon: -- thanks; I particularly like "...pass by" joke; intend to use it somewhere!	11:48
charon	:)	11:48
	→ gabeodess_ joined	11:48
	← gabeodess_ left	11:51
	→ dwave joined	11:52
seangrove	hey all, I'm working on going through some big merges right now, and there's one very long file in particular that has hundreds of conflicts	11:57
	I actually just want to take the newer file - is there a way to just grab it wholesale from the other branch and overwrite the conflict-ridden one?	11:58
	← jony left	11:58
Pieter	git checkout --theirs -- path/to/file	11:58
seangrove	Pieter: it's a local branch,	11:58
	so would --theirs just be the local branch I want to pull from?	11:58
Pieter	--theirs is a command-line option	11:59
seangrove	hmm, not sure I follow	12:00
	I'm merging branch b into branch a - branch b's file should completely overwrite branch a's	12:00
Pieter	oh, so you don't want to take the version that gets merged in, but the current version	12:00
	oh, no	12:01
	hah, I keep redaing it wrong :(	12:01
seangrove	heh	12:01
Pieter	--theirs will default to check out stuff from the branch you're merging in	12:01
seangrove	no problem, my git terminology's probably all off anyway	12:01
	oh, great	12:01
	I'll give that a try	12:02
	"error: path 'public/javascripts/controls.js' is unmerged"	12:03
	→ scarabx joined	12:03
seangrove	I'm pretty sure writing a regex to fix this will take quite some time...	12:05
	seems like because I'm in the middle of resolving a conflict, it's not going to work so easily...	12:06
	← xckpd7 left	12:07
seangrove	ah, got it	12:07
	just had to specify the branch name explicitly	12:07
	thanks pieter!	12:07
	→ ignotus joined	12:08
	← ashleyw left	12:09
	→ dodo_the_last joined	12:09
ignotus	hello, where do I find an explanation of "git diff tag1 tag2" "git diff tag1..tag2" "git diff tag1...tag2" and any other syntax that is possible?	12:09
	→ quoin joined	12:11
tokkee	ignotus: git-diff manpage ...	12:11
literal	also "SPECIFYING RANGES" in the git-rev-parse man page	12:12
tokkee	Nope - git-diff does not use ranges ...	12:13
charon	they're not ranges to git-diff however, it parses them specially	12:13
ignotus	ahh I see, thanks tokkee	12:15
	← riovena left	12:19
	→ mobhawk joined	12:19
	→ grahal joined	12:19
	→ aspotashev__ joined	12:22
	→ aspotashev___ joined	12:22
	← aspotashev__ left	12:23
	→ aspotashev__ joined	12:23
	← aspotashev__ left	12:23
	→ aspotashev__ joined	12:24
	→ rudi_s joined	12:24
	← aspotashev left	12:25
	← aspotashev\|eeepc left	12:26
	← aspotashev___ left	12:26
	→ aspotashev\|eeepc joined	12:26
	← dodo_the_last left	12:26
	← aspotashev__ left	12:26
	→ aspotashev joined	12:26
	→ dodo_the_last joined	12:27
	→ Ryback_ joined	12:28
	→ elad joined	12:28
elad	How do I check the diff from my stash?	12:28
Bonaparte	I accidentally deleted a file on my local system. I have not commited yet. How can I pull the file from the repository?	12:29
	git pull is not actually pulling the file	12:29
Grum	you didnt commit therefor it isnt stored?	12:29
rudi_s	elad: git stash show -p	12:31
Bonaparte	Grum, I haven't commited after I deleted the file	12:31
	→ cmarcelo joined	12:31
Grum	Bonaparte: what did you do? cloned a repo, deleted a file in the checkout and now you want that file back?	12:32
wshimmy1	Bonaparte: if the file was present in your last commit; git checkout HEAD -- path/to/file	12:32
	wshimmy1 → wshimmy	12:32
Bonaparte	Grum, yes	12:32
	wshimmy, the file was present in the last commit	12:33
	→ Jinzhu joined	12:33
Grum	if you read what: 'git status' tells you	12:34
	then you would have gotten the answer wshimmy just gave you :)	12:34
	← ph^ left	12:35
Bonaparte	Thanks grahal and wshimmy	12:37
	Grum, too	12:37
	→ _grepper joined	12:37
	→ drewr joined	12:38
	→ ericindc joined	12:38
	→ ph^ joined	12:39
	← Pe3k left	12:39
	← voker57 left	12:39
	→ voker57 joined	12:40
RurouniJones	I don't suppose some kind soul here could provide me with a chunk of code to change an email in a commit history using git filter-branch --commit-filter. Something along the lines of http://pastie.org/467574 which is something nofunctional I tried to put together after reading manpges.	12:41
	← piroko___ left	12:42
	→ ashleyw joined	12:42
	→ khmarbaise joined	12:43
Grum	RurouniJones: you are using variables with upper/lower case in it	12:43
	you are assining variables to a string	12:43
	there is lots of stuff wrong with that :D	12:43
RurouniJones	I know, I have no knowledge of shell scripting	12:44
charon	also remember to export	12:44
RurouniJones	Hence my plea	12:44
	→ piroko joined	12:44
Grum	man bash :(	12:44
	→ struberg joined	12:44
RurouniJones	Ah, the usual geek reply.	12:45
sdiz	RurouniJones: http://tldp.org/HOWTO/Bash-Prog-Intro-HOWTO.html ?	12:46
Grum	i dont know it either :( ... and i actually use man bash :(	12:46
	← ignotus left	12:46
	→ Bennid joined	12:47
RurouniJones	Yes yes, I know all about how I can go and learn shell scripting so that I can use it one time to make this change, I am fully aware of the usual responses.	12:47
	← ashleyw left	12:47
RurouniJones	Which is why I phrased my original question in that manner.	12:47
charon	RurouniJones: http://pastie.org/467581	12:47
	RurouniJones: in addition to what we said already, note that it's --env-filter if you just want to change the GIT_AUTHOR stuff, and [ is an actual command and must be followed by a space	12:48
RurouniJones	Thank you charon. I don't think I would have realised I had to use export like that for a while.	12:48
charon	RurouniJones: oh, you'll probably want to say 'export GIT_COMMITTER_EMAIL="$GIT_AUTHOR_EMAIL"' too...	12:48
	← luke-jr left	12:48
RurouniJones	Luckily I know in this case that it is only a few commits where I was the author only.	12:49
charon	(guarded by similar logic, if you have any commits with committer != author)	12:49
	← Bennid left	12:49
	← dodo_the_last left	12:49
RurouniJones	Hehe, quite.	12:49
	← ericindc left	12:49
RurouniJones	Perfect, that did it. Thank you very much.	12:50
charon	np	12:50
	now you can use the time i just saved you to learn bash ;)	12:50
RurouniJones	I think I need to go and get a book about using ruby for system admin work	12:51
	→ rraasch joined	12:52
	→ _daniel_ joined	12:53
	→ altrux joined	12:54
	→ netoman joined	12:56
	→ lorandi joined	12:56
	→ ashleyw joined	12:57
	← dwave left	13:00
	→ ashleyw_ joined	13:00
	← mikl left	13:01
	← scarabx left	13:02
	→ AkumaStreak joined	13:02
	→ _slacker_ joined	13:04
	← rraasch left	13:04
	→ ashleyw__ joined	13:05
	→ rionave joined	13:05
	→ azzie joined	13:05
	→ LiamH joined	13:07
	→ dwave joined	13:07
	→ nick_e joined	13:07
azzie	Hi everyone, I am using git on an old sun machine and ancountered a funny thing	13:07
	I am short on diskspace there and my /usr/libexec/git-core eats op something like 120MB	13:08
	I looked inside, and most files from git-add to git-write-tree have exactly the same filesize of something like 1MB each	13:10
	First I thought it is some workaround specific for this sun workstation, but then I checked my git installation at home (Gentoo Linux) and... same problem there	13:10
	cpg → cpg\|away	13:11
	→ catalis joined	13:11
charon	azzie: that's entirely expected, they're all hard links to the same file	13:11
	(unless your build process messed up)	13:11
_daniel_	Hello guys, I'm trying to setup a remote repository as in this link http://toolmantim.com/articles/setting_up_a_new_remote_git_repository	13:11
	but when I do git push origin master, i get fatal: '/userhome/userdir': unable to chdir or not a git archive	13:12
	fatal: The remote end hung up unexpectedly	13:12
azzie	charon: That would be a perfect reason for the same filesize, but if the installation takes 120MB, something must have gone wrong	13:12
	charon: let me check this in detail from hardlinking perspective	13:13
	← ashleyw left	13:14
	→ tpr joined	13:15
tpr	Hi. Is it true, that with git it is possible to overwrite data in repository, using push force?	13:15
	← knitt1 left	13:16
azzie	charon: OK, things start to get explainable - my Gentoo installation is correctly hardlinked, but the sun installation is just wasting space	13:16
	charon: Thanks for the hint. Do you know if there is a reason for using hard links and not symlinks like most packages I've seen do?	13:17
	← wshimmy left	13:18
	← ilor left	13:18
	→ reprore joined	13:19
	→ okidogi joined	13:19
charon	azzie: that structure is older than my participation in git, but if i had to guess... because every symlink needs an extra inode, while a hardlink doesn't?	13:20
	← ashleyw_ left	13:20
okidogi	hi, how can I remove commits between two commits, say 0afcdcc and 0bcdefc	13:20
	← ph^ left	13:20
	→ aser joined	13:21
charon	azzie: oldest match i can find is f94fbbee90, which does not justify the choice (as it probably would if it were done today)	13:21
Gitbot	[git f94fbbee9]: http://tinyurl.com/dce8bl -- Retire git-log.sh (take #3)	13:21
tpr	Is it true, that using git push force it is possible to "corrupt" main repository?	13:22
	→ AAA_awright_ joined	13:23
charon	okidogi: using git rebase --onto, see man git-rebase	13:23
Gitbot	okidogi: the git-rebase manpage can be found at http://git.or.cz/man/git-rebase	13:23
	← mithro left	13:23
azzie	charon: Yep, I know. But interestingly the busybox project (which is exploiting the idea of a shared binary to its limits) seems to be using symlinks by default	13:24
charon	tpr: no, you can use a forced push to "lose" data, but that can be disabled. consistency is always checked by the receiving end before updating any refs	13:24
okidogi	charon: thanks for your reminding. I'll just delete the pick *** line when rebase.	13:24
charon	or that.	13:24
	← Bonaparte left	13:25
okidogi	charon: ah, yes.	13:25
	→ krh joined	13:26
	← piroko left	13:26
	← grahal left	13:26
	← ndim left	13:26
	← stuffcorpse left	13:26
	← javatexan left	13:26
	← naeu left	13:26
	← spacepluk left	13:26
	← solydzajs left	13:26
	← qhoxie left	13:26
	← AAA_awright left	13:26
	← cpg\|away left	13:26
	← yzlin left	13:26
	← ahall left	13:26
	← da5id left	13:26
	← dna left	13:26
	← jeek left	13:26
	← Leonidas left	13:26
	← Fissure left	13:26
	← mkching left	13:26
	← masterdriverz left	13:26
	← dirker left	13:26
	← teuf left	13:26
	← lll left	13:26
	← pigeon left	13:26
	← thijso left	13:26
	← Nafai left	13:26
	← qzio left	13:26
	← stick left	13:26
	← cj left	13:26
	→ piroko joined	13:27
	→ grahal joined	13:27
	→ ndim joined	13:27
	→ stuffcorpse joined	13:27
	→ javatexan joined	13:27
	→ naeu joined	13:27
	→ spacepluk joined	13:27
	→ solydzajs joined	13:27
	→ qhoxie joined	13:27
	→ cpg\|away joined	13:27
	→ qzio joined	13:27
	→ yzlin joined	13:27
	→ ahall joined	13:27
	→ da5id joined	13:27
	→ dna joined	13:27
	→ jeek joined	13:27
	→ Leonidas joined	13:27
	→ cj joined	13:27
	→ Fissure joined	13:27
	→ lll joined	13:27
	→ stick joined	13:27
	→ thijso joined	13:27
	→ pigeon joined	13:27
	→ dirker joined	13:27
	→ mkching joined	13:27
	→ masterdriverz joined	13:27
	→ Nafai joined	13:27
	→ teuf joined	13:27
azzie	charon: They support hardlinks as well, but choose symlinks by default for some reason. Anyway, I think I could replace hardlinks with symlinks if I really needed/wanted, right? :)	13:27
	→ mfilipe joined	13:27
	← stuffcorpse left	13:27
charon	azzie: sure	13:27
	← masterdriverz left	13:27
azzie	charon: Thanks for your help	13:27
	daniel: please provide more details, if you managed to clone a repository you should be able to push to it	13:28
	→ Bennid joined	13:28
charon	azzie: in fact the infrastructure is such that you can place an arbitrary git-foo in $PATH (extended, at that point, to include libexec/git-core) and git will attempt to exec() it when you say 'git foo ...args...'	13:28
	→ alip joined	13:29
azzie	charon: This is reasonable, I have my own git-latexdiff :-)	13:30
	charon: Do you know by the way when the "difftool" infrastructure is going into some release?\	13:30
	← nothingHappens_ left	13:30
	→ stuffcorpse joined	13:31
tpr	charon: thank you. How to disable this?	13:31
charon	azzie: haven't followed closely, but last "what's in" says it's going to be in -rc4, thus in 1.6.3	13:31
	→ voker57_ joined	13:32
charon	tpr: receive.denyNonFastForwards man git-config	13:32
Gitbot	tpr: the git-config manpage can be found at http://git.or.cz/man/git-config	13:32
	charon </faqbot> <coffeebreak> ...	13:32
tpr	charon: generally it's the picture in which one of the repositories is "main", ie: no-overwritable, no-corruptable safe place	13:32
	thanks	13:33
	← reprore left	13:33
	→ ph^ joined	13:34
azzie	charon: Cool, cat't wait for it then. Setting an external diff wrapper through git-attributes is cumbersome. And it seems that I cannot have it on-demand only. An --ext-diff option seems to exist, but it's always on for me.	13:34
	→ reprore_ joined	13:34
	→ ryanew joined	13:35
	← ryanew left	13:35
tpr	are there any pitfalls when one should be aware of when upgrading from CVS ?	13:35
_daniel_	azzie: i solved the problem , it seems that the path was not complete	13:36
azzie	daniel: good :)	13:36
PerlJam	tpr: don't expect anything to work like it does in CVS? :)	13:36
	→ redduck666 joined	13:36
redduck666	how can i merge from a (remote) branch and tell git "in case of conflict take whatever they have" ?	13:36
rudi_s	Hi. Is there a list of all git options (for gitconfig) with descriptions? man git-config says this is not a complete list. Thanks.	13:37
Gitbot	rudi_s: the git-config manpage can be found at http://git.or.cz/man/git-config	13:37
tpr	PerlJam: anything else? For example: tags?	13:37
	are they similar ?	13:37
	→ Dreamr3 joined	13:38
	→ ProperNoun joined	13:40
	← _grepper left	13:40
azzie	redduck666: how about starting a new development branch from the remote and merging your old development branch with "ours" strategy - never tried it myself, but it seem this is what it is there for	13:40
	← voker57_ left	13:41
	→ bschindler joined	13:41
	← aziz_ left	13:41
	→ voker57_ joined	13:41
sdiz	redduck666: -s ours, do it in reverse direction?	13:42
	→ Sho_ joined	13:42
redduck666	azzie: that is what i have done :-), but as the development goes on the upstream slightly modifies my patches and when i try to pull them again i have to correct each and every lil change they made	13:42
tpr	are tags in cvs and git similar ?	13:42
	→ spearce joined	13:42
redduck666	"correct" meaning open the file, search for the <<<< part, fix it and git add file	13:43
	→ boto joined	13:43
	← Dreamer3 left	13:43
redduck666	sdiz: can that deal with upstream slighttly modifying my patches?	13:43
sdiz	redduck666: no.	13:44
	(unless you means whitespace changes)	13:44
azzie	redduck666: sdiz proposed the same thing I did, I can't see why it wouldn't work with upstream conflicting with your changes	13:44
redduck666	sdiz: what is the reccommended way to deal with that?	13:44
	← voker57 left	13:44
redduck666	azzie: just a second, lemme give you an example	13:45
sdiz	redduck666: don't be lazy, use merge and then git-mergetools ?	13:45
	← ashleyw__ left	13:45
	→ spearce` joined	13:45
	→ bdimcheff joined	13:46
redduck666	azzie: http://dpaste.com/40786/ this is staff i end up with after the pull (containing my changes)	13:46
azzie	redduck666: to do it properly I would recommend git-mergetool -t kdiff3 - otherwise you always rish loosing some work	13:46
redduck666	hrm, i'll look up mergetools, thanks :-)	13:46
	→ leanucci joined	13:47
azzie	redduck666: OK, I understand the reason for the conflict, but "ours" merge strategy should always automatically choose local version	13:47
	redduck666: does this what you are showing remain valid for merging with "ours" strategy?	13:48
	redduck666: since you want the opposite effect, you should create a new devel branch from recent upstream and merge in your old branch with "ours" strategy	13:49
	→ jrmuizel joined	13:49
redduck666	aha, i see	13:49
azzie	sdiz: correct me if I'm wrong, I never needed to do it myself ;-)	13:49
	← neunon left	13:50
	← tal67 left	13:50
	→ neunon joined	13:50
	lolmaus^_^ → lolmaus	13:51
	← neunon left	13:51
	→ cemerick joined	13:51
	→ dodo_the_last joined	13:51
	→ tal67 joined	13:53
	← psankar left	13:53
	← esparkman left	13:54
	→ juanjoc joined	13:54
	← CardinalNumber left	13:56
	→ repror___ joined	13:57
	← reprore_ left	13:57
	→ mbroeker joined	13:58
	← repror___ left	13:58
	→ davidfetter joined	14:00
	← plasticine left	14:00
	← spearce left	14:01
	← drewr left	14:01
	→ elmob joined	14:01
	← hobodave left	14:03
	→ nurey joined	14:04
	→ mw___ joined	14:06
	→ parth-vader joined	14:06
	→ kukks joined	14:08
	← _daniel_ left	14:09
	→ aziz_ joined	14:09
	← ph^ left	14:10
	→ ashleyw joined	14:12
	→ dwave_ joined	14:13
	→ reprore joined	14:16
	← jrmuizel left	14:17
	← azzie left	14:17
	← lolmaus left	14:17
	→ plasticine joined	14:18
	← aser left	14:18
	← j416 left	14:18
	← plasticine left	14:18
	← ilteris_ left	14:20
	→ ilteris_ joined	14:20
	← mfilipe left	14:21
	→ xkr47 joined	14:21
	sebleier\|away → sebleier	14:21
	← Dreamr3 left	14:23
	→ wagle joined	14:24
	← leanucci left	14:25
	→ leanucci joined	14:25
	→ patrikf joined	14:26
	→ alezandro joined	14:27
	→ knitt1 joined	14:29
	← stuffcorpse left	14:30
	→ stuffcorpse joined	14:31
	→ hyperair joined	14:31
	→ gotgenes joined	14:31
	→ paulboone joined	14:31
	→ wmoxam joined	14:37
	← lohrun left	14:37
	← reprore left	14:38
	→ plasticine joined	14:39
ReAn	is there an easy way to checkout the latest tag?	14:40
	→ Fullmoon joined	14:41
	← Fullmoon left	14:42
	→ hummingkid joined	14:43
gilimanjaro	define latest...	14:45
ReAn	most recently added	14:45
gilimanjaro	you mean the closest commit that also has a tag?	14:45
ReAn	yea	14:45
gilimanjaro	i don't think that's very simple	14:46
	because as far as i know tags point to commits	14:46
	← rolfb left	14:46
gilimanjaro	and not the other way around	14:46
	← seangrove left	14:46
ReAn	ill have to write a script i guess	14:47
gilimanjaro	i think so	14:48
	→ scientes joined	14:48
	→ MauL^ joined	14:50
MauL^	when I say checkout, it tells me M with a file .. but the file is not updated on the working dir	14:50
	→ jrmuizel joined	14:53
	← solydzajs left	14:54
	bwalk → bwalk\|uni	14:55
	← bryanl left	14:55
	→ drewr joined	14:55
	← Jinzhu left	14:55
	← MauL^ left	14:57
reaVer	how can I tell git under what group it should add objects in a shared repository?	14:57
	← ashleyw left	14:57
	→ bschindler\|lapto joined	14:57
	→ ReAn` joined	14:57
	← Ryushin left	14:58
	→ Ryushin joined	14:59
	← ReAn left	15:00
	→ hobodave joined	15:00
	← bschindler left	15:00
	→ lohrun joined	15:00
	→ shikamaru joined	15:03
shikamaru	hello there	15:03
	← parth-vader left	15:04
shikamaru	I'm looking for a web interface for git, written in ruby	15:04
	I found wit	15:04
	but it timeouts here :s	15:04
	← qhoxie left	15:04
shikamaru	do you know another one ?	15:05
	→ Bass10 joined	15:05
	→ qhoxie joined	15:05
	sdiz → sdiz-afk	15:05
	→ ashleyw joined	15:06
	← Bass10 left	15:06
	→ Bass10 joined	15:07
	← ashleyw left	15:07
	← naeu left	15:07
	→ name joined	15:08
	→ naeu joined	15:08
	→ ashleyw joined	15:10
	→ icwiener joined	15:10
	← paulboone left	15:10
	← ashleyw left	15:11
	← johan-s left	15:11
	→ kn1ttl joined	15:11
	→ hugoxrosa joined	15:13
	→ sergiodj joined	15:14
	← hendry left	15:15
	→ ashleyw joined	15:15
spacepluk	hi, someone is tracking/commiting to a CVS repository with cvsimport/cvsexport? I'm getting the commits twice when I re-import the exported commits...	15:15
	← okidogi left	15:15
	→ jtong_ joined	15:16
	← ashleyw left	15:16
	← coldboot left	15:17
	→ romao joined	15:18
	← gilimanjaro left	15:19
	← wycats left	15:19
	→ redduck676 joined	15:19
	→ ashleyw joined	15:20
	← qhoxie left	15:20
	← stuffcorpse left	15:20
	← redduck666 left	15:20
	→ stuffcorpse joined	15:21
drizzd	ReAn`: git describe	15:21
	← jtong left	15:21
	jtong_ → jtong	15:21
	→ j416 joined	15:21
	→ octagon joined	15:22
	← j416 left	15:22
	→ j416 joined	15:23
drizzd	reaVer: AFAIK the existing group is used	15:23
	reaVer: that's just standard file system behavior	15:24
	← stuffcorpse left	15:27
	→ stuffcorpse joined	15:27
ReAn`	btw, what's the package name for it under gentoo?	15:27
	mw___ → mw	15:29
	→ wycats joined	15:30
drizzd	ReAn`: probably git-core	15:30
	→ paulboone joined	15:30
	→ johan-s joined	15:30
	→ qhoxie joined	15:30
drizzd	make that maybe	15:30
	← plasticine left	15:31
	← davidfetter left	15:31
	← knitt1 left	15:31
	Shadow_Master → Shadow_Melon	15:33
	→ davidfetter joined	15:33
	→ bryanl joined	15:33
	→ spaceninja joined	15:34
	← khmarbaise left	15:34
	← bschindler\|lapto left	15:34
spaceninja	does git work like rsync?	15:34
	can it both save changes and copy it to a server?	15:35
Ilari	spaceninja: Git and rsync are made for different purposes....	15:35
spaceninja	i know	15:36
	→ [t0rc] joined	15:36
	← agile left	15:37
	← catalis left	15:37
spaceninja	so git should be on the server	15:38
[t0rc]	Just for the sake of stating things - the git download link on the site is a bit broken.	15:38
	→ BlindOracle joined	15:38
	→ kumbayo joined	15:38
	→ mrbig joined	15:38
Ilari	spaceninja: Pushing commits via SSH requires git to be installed on remote side.	15:39
spaceninja	ok	15:39
Ilari	spaceninja: The only git-specific server program is git daemon, which is daemon for git:// (anonymous repo access).	15:40
	→ vipul joined	15:40
	→ catalis joined	15:41
	→ faithfulgeek joined	15:41
	→ WALoeIII joined	15:42
mrbig	Hi all! I use git to track website codes. The web frontends share a common NFS export, and are set to slow synch (about 5 min). If I do some git merge, for this 5 mins of synch period I get "Stale NFS file handle" errors on the other frontends.	15:43
	← stuffcorpse left	15:43
mrbig	I guess this is because of some git "magic" with filehandles. Is there a way to turn it off?	15:43
	← qhoxie left	15:44
	→ qhoxie joined	15:46
	← giallu left	15:47
	← j416 left	15:47
	← doener left	15:47
[t0rc]	is there a place I can download the latest git?	15:48
	→ unixluser joined	15:49
jast	source code: the official website (http://git-scm.com/). binaries: depends on the OS.	15:49
	Shadow_Melon → Shadow_Master	15:49
unixluser	hi guys. i created my first branch on remote server. Now, locally, i only want to track that one branch, and not master. I don't even want master branch on my local machine at all	15:50
	can this be done	15:50
	→ lightcap joined	15:50
	← yec left	15:50
[t0rc]	jast, the links are broken on the site for the bz2 source	15:50
Ilari	mrbig: The filehandles shouldn't be exportable between processes. So no, git doesn't do any magic with filehandles.	15:51
	→ stuffcorpse joined	15:51
	← bobmcw left	15:51
	→ bobmcw joined	15:51
jast	[t0rc]: indeed. here's a direct link: http://kernel.org/pub/software/scm/git/git-manpages-1.6.2.4.tar.bz2	15:51
	whoops, sorry	15:51
	[t0rc]: http://kernel.org/pub/software/scm/git/git-1.6.2.4.tar.bz2	15:51
drizzd	unixluser: yes, you can configure which branch to track using "git remote add"	15:52
mrbig	llari: ok, might not the filehandles. All I see that git is extremely fast on doing anything related to files. So it shure must do some tricks (and it's also mentioned in the docs)	15:52
	but when I use it on an NFS filesystem, this leads to these errors.	15:52
shikamaru	if someone is interested, I've found that redmine can handle git repositories through grit	15:52
	(if you want a ruby web interface for git)	15:53
[t0rc]	jast, thats an old version isn't it? (.5 is newest yerrs?)	15:53
drizzd	mrbig: it is fast mainly because if it's design, not because of any magic	15:53
	s/if/of	15:53
[t0rc]	exit	15:53
	← [t0rc] left	15:53
mrbig	drizzd: when I use other programs/scms to manipulate files I never get these errors. I either see the old or the new state of the file.	15:54
jast	[t0rc]: the source archive for .5 doesn't seem to have been uploaded yet, so if you wait a while, the official link will probably work later on	15:54
mrbig	But when git is touching it, it takes them into some kind of intermediate state	15:54
	← sgrimm left	15:54
drizzd	mrbig: all I can say is that I'm doing the same thing -- accessing files on NFS from several machines at the same time -- and I've never had any problems with git	15:55
	→ saurabh1403 joined	15:55
jast	drizzd: probably because you don't have sync delay set to 5 minutes? ;)	15:55
drizzd	probably	15:55
mrbig	We need this synch time, to reduce the traffic between the web frontends, and the nfs server.	15:56
	(For the static files, and codes of course)	15:56
spearce`	git creates updated files usually by writing them to a new temporary name in the directory, and then linking the temporary file over the real file, then deleting the temporary file name. that's probably causing the real name to become invalid until the temporary file syncs, 5 minutes later. yay nfs.	15:57
	spearce` → spearce	15:57
	→ lhz joined	15:57
mrbig	spearce: that's what I'm thinking about. Do you know any way to turn it off, and to use traditional open/close?	15:57
spearce	nope, its a fundamental way git handles files. patch git, or use rsync to maintain your frontend servers	15:58
	← patrikf left	15:58
spearce	or, patch your nfs server to not be so damn lazy.	15:58
	its clearly exporting the directory update, but not the new inode.	15:58
unixluser	drizzd: how do i locally delete master branch (leave remote intact)	15:59
Ilari	unixluser: 'git branch -d master' (or -D)	15:59
unixluser	drizzd: atm, my .git/config has both branches	15:59
csgeek	anyone know if git supports git svn checkouts of externals. I found a reference to a parameter in 1.4 of git, but I can't seem to find the option in a modern build	15:59
unixluser	Ilari: all that does is modify my config, right?	16:00
drizzd	unixluser: right, if you clone master will be tracked automatically. You can simply delete it as Ilari suggested, without influencing the remote branch.	16:00
	← altrux left	16:00
mrbig	spearce: thanks. I'll find some other ways. I could use two repositories, and just symlink them over when things change.	16:00
Ilari	unixluser: It actually deletes the branch from local repo.	16:01
unixluser	error: The branch 'master' is not an ancestor of your current HEAD	16:01
	If you are sure you want to delete it, run 'git branch -D master'	16:01
	← struberg left	16:01
Ilari	unixluser: Read the error message.	16:02
lhz	how can origin pull commits from a remote branch into origin/master? (now it is creating that branch plus a blank one).	16:02
unixluser	not sure what ancestry means in this case	16:02
Ilari	unixluser: Especially the second line.	16:02
	→ struberg joined	16:02
	→ d0k_ joined	16:02
unixluser	Ilari: but i'm a bit unsure	16:02
drizzd	unixluser: master is not entirely included in the history of your current branch	16:02
	see the output of git log ..master	16:03
	→ dwave- joined	16:03
	→ dtillberg joined	16:03
unixluser	welp i went ahead and did -D	16:03
	drizzd: will that pose a problem?	16:03
	i pretty much did: git push origin #{current_branch}:refs/heads/#{branch_name} to create the branch	16:04
drizzd	unixluser: if you didn't do any work on master locally, it's simply a copy of your remote master, so you can recreate it any time	16:04
unixluser	okay	16:05
	← bentob0x left	16:05
drizzd	unixluser: normally you want to create branches using "git branch <branchname>" and then push that using "git push origin <branchname>"	16:06
	the latter is only necessary if you want to copy the branch to origin for whatever reason	16:07
	← faithfulgeek left	16:08
	← d0k left	16:09
	← Bass10 left	16:09
	→ agile joined	16:10
	→ Alinermedia joined	16:10
	← jmesnil left	16:10
	→ kuadrosx joined	16:11
	→ JasonWoof joined	16:11
	→ faithfulgeek joined	16:11
	→ releod joined	16:12
unixluser	thanks again guys	16:12
	← unixluser left	16:12
	← bryanl left	16:13
	→ boto_ joined	16:13
	eno_ → eno	16:14
	→ nothingHappens joined	16:15
	← boto_ left	16:15
	← boto left	16:16
	← cedricv left	16:20
	→ maik_ joined	16:23
	→ bryanl joined	16:23
	→ sgrimm joined	16:23
	→ rubydiamond joined	16:24
	← WALoeIII left	16:25
	← hummingkid left	16:25
	← qhoxie left	16:26
	← cods left	16:26
	← nick_e left	16:27
	← mrbig left	16:29
	→ cods joined	16:30
	→ max_ep_ joined	16:30
	→ MauL^ joined	16:32
MauL^	hey guys. on remote server, I did a checkout. then on my local machine, I pushed changes to remote server. then on remote, how can I update the working dir ?	16:33
	← max_ep left	16:34
wwalker	MauL^: sounds like you just need a git pull	16:34
	→ ph^ joined	16:35
MauL^	wwalker, I made a push from local to remote. what will pull do additionally	16:35
	← dtillberg left	16:36
	← tal67 left	16:37
	← trepca left	16:37
	→ dtillberg_ joined	16:37
drizzd	MauL^: faq non-bare	16:37
Gitbot	drizzd: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare	16:37
	← fujin left	16:39
	← javatexan left	16:39
	← Modius__ left	16:39
	→ Modius__ joined	16:40
	→ bryanray joined	16:40
	→ javatexan joined	16:40
bryanray	Hey, guys ... there is a "term" for "local branches" ... but for the life of me I can't remember what it is?	16:41
jast	how about "local branches"? ;)	16:41
bryanray	clever :p	16:41
	← Modius__ left	16:41
jast	I don't know any other term	16:41
	dtillberg_ → dtillberg	16:41
	→ boto joined	16:41
bryanray	nah, some people call them "conversation" or something like that ... :/	16:41
	grrr. it's driving me nuts	16:41
jast	I have never, ever heard anyone call them that in here	16:41
	or anything like it	16:42
	→ Modius__ joined	16:42
bryanray	topic branch perhaps?	16:42
	→ neunon joined	16:42
	← tango_ left	16:43
bryanray	yah. topic branches	16:43
	← javatexan left	16:43
eMBee	topic branches can be remote too	16:43
bleything	yeah, topic branch just means "a branch in which you work on a specific thing"	16:44
wwalker	MauL^: never mind, you are not setup the way I am.	16:44
	← lohrun left	16:44
	wwalker is still quite new to git.	16:44
	← Ryback_ left	16:44
bryanray	yeah. i just new there was a term for it that i couldn't remember.	16:45
MauL^	then what is the command to merge what has been pushed and the working dir	16:45
	← boto left	16:45
	→ boto joined	16:45
	← sergiodj left	16:47
	→ lois[pac] joined	16:47
	→ tal67 joined	16:48
	← boto left	16:50
	← _slacker_ left	16:50
drizzd	MauL^: you cannot merge remotely. To merge locally use "git merge"	16:50
	← jaeckel left	16:50
	joshthecoder_afk → joshthecoder	16:50
	→ boto joined	16:51
	→ qhoxie joined	16:53
	→ tag joined	16:53
	← gotgenes left	16:53
	→ mfilipe joined	16:54
tag	When I remove a folder in git, is there any way to instruct git to prune the folder in the upstream repository? I'm having an annoying problem with git svn where the tree of parent folders is kept in the repository after restructing a repository, but the content in the folders is all removed.	16:54
	→ patrikf joined	16:55
wereHamster	tag: git doens't track empty directories	16:57
	so when you remove all files from a folder, git will 'forget' the folder and it won't exist when somebody clones you repository	16:58
	← icwiener left	16:59
	→ mshadle joined	16:59
mshadle	anyone here use gitosis on jaunty?	16:59
drizzd	Looks like git does not remove empty parents. This is a bug I suppose.	16:59
	← ashleyw left	17:00
drizzd	wereHamster: try git rm asdf/bsdf/csdf, where asdf contains only bsdf and bsdf contains only csdf	17:00
	asdf will not be removed	17:00
	→ khmarbaise joined	17:00
	→ ashleyw joined	17:01
	← ashleyw left	17:01
wereHamster	ah, well, git rm is a wrapper around rm, and you wouldn't want rm to remove parent directories, would you?	17:02
	if you want to clean untracked files, git-clean is the way to go	17:02
	← elad left	17:02
drizzd	wereHamster: but it does remove bsdf	17:02
	wereHamster: would you expect it to leave the directory if you use "git mv"?	17:03
wereHamster	ah, I didn't think git-rm would remove bsdf	17:03
	← saurabh1403 left	17:04
	← frsk left	17:04
lhz	would I use git pull repopath HEAD:mybranch at origin to keep track of branch history (using gitk). And if I'm not interested in that (throw away branches) I would use git pull repopath mybranch?	17:04
drizzd	tag: git clean -d	17:05
	→ neunon_ joined	17:05
	← neunon_ left	17:05
	← neunon left	17:05
	Shadow_Master → Wikiepdia	17:05
	→ sergiodj joined	17:05
	Wikiepdia → Shadow_Master	17:05
	→ andrewkasper joined	17:05
wereHamster	drizzd: shall I bring up this issue on the mailing list and ask the developers why git behaves that way?	17:05
	← mshadle left	17:06
drizzd	wereHamster: sure, should be easy to fix	17:08
struberg	spearce: how can I get a Tree from GitIndex so I can commit it?	17:09
spearce	beats me. i've never used GitIndex. :-)	17:09
	→ sergio_ joined	17:09
wereHamster	drizzd: I'm not sure what the correct fix is. Would you expect git to remove all untracked (parent) directories?	17:10
spearce	look at usage in egit projects?	17:10
	← bryanl left	17:10
struberg	ah kk, because I simply used index.add(db.getWorkDir(), fileToAdd); and the .git/object/.. looks the same like I get with a manual git-add	17:11
	← faithfulgeek left	17:11
struberg	spearce didn't find any usage in egit till now	17:11
	→ faithfulgeek joined	17:11
	→ faithfulgeek_ joined	17:13
	← faithfulgeek left	17:13
struberg	spearce: I've searched for usages of Commit#commit() and only found test cases with hardcoded ids, isn't this used in egit?	17:13
drizzd	wereHamster: yes, all empty parent directories	17:14
	← rudi_s left	17:14
	→ tango_ joined	17:15
	→ bizhat2 joined	17:15
bizhat2	Anything like TortoiseGit (change color of modified file to RED, non altered GREEN) in Linux ? What everyone use on linux ? command line ?	17:15
spearce	struberg: oh, ick.	17:16
	egit is using the Tree and TreeEntry APIs, not GitIndex, to prepare the top level tree for commit.	17:16
	ugh.	17:16
drizzd	bizhat2: command line, git gui, gitk, eclipse, editor plugins; see the wiki for many more	17:16
spearce	(see CommitAction.java)	17:16
struberg	spearce: txs will look at it.	17:16
spearce	here's one reason to use DirCache. DirCache.writeTree(ObjectWriter).	17:17
	→ bcardarella joined	17:17
literal	bizhat2: the command line tools can use colors, which is what I use	17:17
struberg	does this mean GitIndex is subject to be dropped out?	17:17
spearce	yes, i want to get rid of the class, its slow, doesn't handle the 'TREE' extension, and doesn't support something simple like writeTree. :-)	17:18
struberg	spearce: or does this only mean that egit doesn't use the 'Index' mechanism yet but commits directly?	17:18
	Shadow_Master → NotShadowMaster	17:18
spearce	GitIndex came about to support EGit, but based on what I see in CommitAction.java, that support is still woefully incomplete.	17:18
	← maik_ left	17:18
spearce	some time ago i looked at GitIndex, ran away screaming, and created DirCache.	17:19
struberg	hehe :)	17:19
	NotShadowMaster → Shadow_Master	17:19
spearce	we haven't yet finished moving to DirCache. so there is still code using GitIndex. and Tree and TreeEntry. sigh	17:19
	← spacepluk left	17:19
	← ngng left	17:22
	→ giallu joined	17:22
	← cpg\|away left	17:23
	← Modius__ left	17:23
	← sergiodj left	17:23
	→ beni- joined	17:23
	→ tefflox joined	17:23
beni-	hi!	17:24
tefflox	hello, I have a public key that I want to use to allow a second dev to my local repo. how do i do that?	17:24
	I mean the other dev sent me her public key	17:24
	→ bryanl joined	17:25
beni-	i'm new to git, and trying to import my SVN repo into my new git repo. i have a git repository and want to export the stuff from SVN to my (existing!) git repo. does this work with 'git svn clone...'? i think no, because man page sais "clone: Runs init and fetch" -> this would re-initialize my repo. how can I do it otherwise?	17:25
	← gitte left	17:26
	← ProperNoun left	17:26
beni-	tefflox: hm well, don't know much about git, but I think that it depens on your setup. Do you have your repository on a server? or just on your local machine?	17:26
tefflox	local	17:26
patrikf	tefflox: you'd have to set up an account for her, allow her to ssh to your machine (read man authorized_keys), put your and her user into a group and make a group-read/writable bare repository	17:26
	→ ashleyw joined	17:26
patrikf	tefflox: the better workflow would be for you both to publish your repositories somewhere	17:27
	→ rylar joined	17:27
patrikf	and pull from each other	17:27
	→ Modius joined	17:27
tefflox	is that hard to do?	17:28
	you mean i send her a copy of the app, ask her to make a repo that i can pull from? and somehow allow her to pull from mine?	17:28
beni-	tefflox: nope, its easy	17:29
tefflox	that sounds complicated	17:29
beni-	tefflox: look here: http://scie.nti.st/2007/11/14/hosting-git-repositories-the-easy-and-secure-way	17:29
	tefflox: if you have a server :D. my link describes everything very well, worked for me in the first try :D	17:29
	→ jackdempsey joined	17:29
tefflox	what if i don't have a server?	17:30
drizzd	tefflox: if you do not have a direct network connection to your co-worker, you have to use e-mail	17:31
	← MauL^ left	17:31
drizzd	or a public hosting service	17:31
	→ Dashkal joined	17:31
jast	tefflox: what kind of system are you running on your machine?	17:31
drizzd	or a private hosting service (for free usually)	17:31
	← elmob left	17:33
tefflox	maybe i'll just buy a github account	17:33
	they're pretty	17:33
	oh so pretty	17:33
	← Modius left	17:33
	→ tclineks joined	17:34
	→ ngng joined	17:34
	→ Modius joined	17:34
	→ bryanl_ joined	17:38
	→ bonch joined	17:38
	→ paltman94 joined	17:39
	→ smeevil joined	17:40
beni-	hm, no one an idea how I can check out an existing SVN into a already existing git-repository?	17:43
	← bryanl left	17:43
	← naeu left	17:44
jast	beni-: well, what would you want to happen to the history that you already have?	17:44
tefflox	The guy who was helping me is hosting the "origin" on his machine. Now how do I change that origin to github?	17:44
	can I push to two repos at once?	17:44
jast	tefflox: the simplest way is to edit the URL in the config file (.git/config)	17:45
	(first question)	17:45
	← jonny142 left	17:45
beni-	jast: append all SVN stuff onto the existing git stuff?	17:45
	← spaceninja left	17:45
	→ aisbaa joined	17:45
jast	beni-: but does that actually make sense? I mean, if there are completely different things in your git history and the svn history, the resulting concatenation would be mostly meaningless	17:46
beni-	jast: hm well yeah, that's true	17:46
	→ gotgenes joined	17:46
beni-	jast: does that mean, that there is no possibility to do that?	17:46
jast	beni-: there is, but it involves a bit of manual tweaking, and it will probably prevent you from using git-svn to keep up to date with the original svn repository later on	17:47
	particularly if the git history is supposed to be "before" the svn history in the structure	17:47
	→ javatexan joined	17:47
beni-	jast: hm well, I want to move over from the existing SVN to git. SVN will not change any more later	17:47
	sergio_ → sergiodj	17:48
beni-	jast: ok, basically i can just remove the new stuff that I put into my new git, do a svn clone, and then put back all the actual git stuff.	17:48
jast	well, that should be doable	17:51
beni-	ok	17:52
	i'll do that :D	17:52
	← girishr left	17:53
jast	actually you can just use git svn clone to create a new repo and then "import" the existing stuff on top of that later on	17:53
	→ Tricon joined	17:55
	→ KalElias joined	17:55
	← smeevil left	17:55
KalElias	hi...what is a save way to disable a ssh git repo?	17:56
jast	what's an "ssh git repo"?	17:57
	and what do you mean by "disable"?	17:57
	← jpieper_ left	17:57
KalElias	1. a git repository that is accessed via ssh 2. disable = turn something off	17:57
jast	yeah, but turn what off? access via ssh?	17:57
	← sgrimm left	17:58
KalElias	every access :)	17:58
jast	because, you know, one way to disable a git repo is to delete it	17:58
	actually that's about the only way... most other ways could be undone by logging in via ssh and simply doing the opposite of whatever you did to "disable" it	17:58
KalElias	or shutdown the server...i'm asking for a SAFE (not save) way	17:58
jast	if there is unrestricted ssh access and filesystem access to the repo, you can simply change any config change that would forbid you from using it	17:59
KalElias	the real question is...can break something if i move the repo while one is using it?	17:59
jast	so to completely forbid all access you'd have to change permissions so that no user can access the files	17:59
	← ashleyw left	18:00
	→ WALoeIII joined	18:00
jast	generally not, because git operations are usually atomic.. but if you move the working tree, too, that's a different story	18:00
	→ jpieper joined	18:01
jast	for example, some editors save the new version of the file by first truncating the file and then writing the new contents to it	18:01
	→ codebrulee joined	18:01
jast	if you move the entire tree in the wrong moment, the editor might have truncated the file but not have written the new contents yet...	18:02
	and there are some git operations that don't like being interrupted either, e.g. git gc	18:03
KalElias	how is git cg contolled? is there a daemon?	18:03
jast	a safe way might be to temporarily close down ssh access, wait for active sessions to end and then move the repo	18:03
	→ john3909 joined	18:04
jast	cg?	18:04
KalElias	sry. gc	18:04
jast	gc is a local operation that does optimizations and cleanups	18:04
	← aspotashev\|eeepc left	18:04
uau	what would break if you interrupt git gc? IIRC it writes a new pack, then deletes old files	18:04
jast	I'm not sure, actually	18:04
	I'm just advising caution	18:04
uau	so wouldn't it just leave multiple copies of data there, thus wasting diskspace?	18:05
	← sitaram left	18:05
jast	actually I think the only problem is if git gc runs simultaneously with other git operations	18:05
	← ned left	18:06
jast	hmm, actually I can't find any reference to it being dangerous	18:06
uau	i think that should not be a problem either now that there is a time delay on deleting unreferenced data	18:06
	← faithfulgeek_ left	18:06
	→ altrux joined	18:06
	→ Cress joined	18:07
	← bcardarella left	18:07
uau	IIRC it could cause problems before; some other git commands created unreferenced objects and then added references to them, which could be a race condition with gc deleting unreferenced objects	18:07
emias	The User's Manual says that "running "git prune" while somebody is actively changing the repository is a BAD idea."	18:07
	gc calls prune, of course.	18:07
	→ MetaMucil joined	18:07
uau	emias: wrong	18:07
	it does NOT execute the equivalent of "git prune"	18:08
	→ parasti joined	18:08
emias	Well, git-prune(1) tells me that "users should run git-gc, which calls git-prune."	18:08
uau	that's what i referred to above - that IIRC it once did, but now there is a time delay based on the timestamps of the objects	18:09
emias	Ah sorry, didn't read the backlog :-)	18:09
	→ jpieper_ joined	18:10
	← beni- left	18:11
	← lightcap left	18:11
	→ lightcap joined	18:11
	→ dwave-_ joined	18:12
	→ GothAlice joined	18:13
	→ andre_pl_ joined	18:13
emias	Heh, sorry #2, just missed that message.	18:13
	→ sekimura joined	18:14
	→ bobesponja joined	18:14
andre_pl_	I checked out some code with git, and now I'm trying to push my changes back to the main repository but they are being rejected, how can I find out why, and how to get around it?	18:14
jast	andre_pl_: the message mentions something about a non-fast forward, doesn't it?	18:15
emias	uau: Is it safe to push into a repository during a gc in that repository?	18:15
andre_pl_	it does	18:15
	← Cress left	18:15
drizzd	faq non-ff	18:15
Gitbot	drizzd: Your push would lose changes on the remote. See http://git.or.cz/gitwiki/GitFaq#non-ff	18:15
	davidfetter → all	18:15
andre_pl_	jast: ! [rejected] master -> master (non-fast forward)	18:15
	all → davidfetter	18:16
jast	andre_pl_: have a look at that FAQ link, it might help	18:16
emias	That is, git-repack(1) won't bite git-receive-pack(1), either?	18:16
jast	it shouldn't	18:16
	new packs, while being created, usually have a temporary filename	18:16
	→ Ryback_ joined	18:17
	← altrux left	18:18
emias	Sounds good. I wasn't completely sure and therefore refrained from running gc on our shared repositories so far.	18:18
jurg`	mhm	18:20
	→ scientes_ joined	18:20
andre_pl_	jast: Ok, I did the git fetch origin and viewed the log, it looks safe to commit my work, but now if I run git commit -a again it tells me there is nothing but untracked files present, did I just wipe out my log	18:20
	?	18:21
jast	does git status say anything different?	18:22
andre_pl_	same message	18:22
	my code seems to still be there	18:22
jast	so did you change existing files or did you create new files?	18:23
	→ perlmonkey2 joined	18:23
andre_pl_	jast: mostly created, but I did modify 1 or 2.	18:23
jast	actually didn't you already commit your changes before?	18:23
perlmonkey2	Is this a place to ask dumb questions about using git, or is it a git dev channel?	18:23
jast	perlmonkey2: any questions welcome	18:23
andre_pl_	jast: d'oh, you're right I did.	18:24
	I just need to get them to the sever now,	18:24
	← bonch left	18:24
andre_pl_	so pull, then push?	18:24
jast	pretty much, yeah	18:24
	fixing any conflicts that may pop up	18:24
	but pull tells you all about that anyway	18:24
	← jpieper left	18:24
	→ drizzd_ joined	18:26
perlmonkey2	jast: wonderful. Okay, clone'd a project from github. Made a trivial change (add'd space to css file). commit -a to the local branch/fork. Then tried to push back to github. Fail. Can't push to git://github..., use git@github... Why would my new branch have a bad url? Did I do this wrong? Is there a configuration issue? Am I too retarded to use git (most likely).	18:26
	→ ashleyw joined	18:31
	→ naeu joined	18:31
	← mw left	18:31
	← KalElias left	18:31
	→ mw joined	18:31
	← Tv left	18:33
	→ markelikalderon joined	18:35
	→ Kimmie joined	18:35
literal	perlmonkey2: I guess you used the public clone url rather than the private one	18:35
	you can change it in .git/config	18:35
	← Kimmie left	18:35
perlmonkey2	literal: thank you thank you and thank you	18:35
andre_pl_	jast: Thanks a ton, everything worked fine after doing some manual merging	18:36
	→ dreiss joined	18:36
	← deavid left	18:36
	→ Kimmie joined	18:36
	← krh left	18:36
	← galderz left	18:39
	→ ilor joined	18:39
jast	perlmonkey2: you cloned from a git:// URL; those are virtually always read-only	18:41
	(the only way to make the git:// writable is to allow anonymous push... nobody really wants that)	18:41
	*the git:// transport	18:42
	perlmonkey2: if you have write access to that github repository, you can, for example, edit .git/config in your local repo and change the URL there	18:42
perlmonkey2	jast: well I think I made a mistake and blew it away to redownload it. When what I wanted to do was switch to a different branch hosted on the repo. So I guess I wanted to make my local branch the same as the github branch. I should have perms to push there.	18:43
	→ deavid joined	18:43
jast	I'm not sure I understand. did you want to work on a different branch? what did you want to do about the changes you made? what do you think happened instead?	18:44
	← struberg left	18:44
perlmonkey2	jast: my tests were just superficial tests so I could test push.	18:44
jast	I'm not quite sure what situation you've got and what your goal is	18:45
perlmonkey2	jast: there is a repo, with like 10 branches. I want to push to just one of those. I think I'm pretty close to figurin gthis out, I just need to wait on the redownload.	18:45
jast	"the redownload"?	18:45
perlmonkey2	I deleted it on local disk	18:45
	← andre_pl_ left	18:45
perlmonkey2	so I'm waiting on a new clone	18:45
	→ struberg joined	18:45
	← drizzd left	18:46
jast	hmm. well, anyway, it's easiest to use the git@ style URL for cloning, then you don't have to edit the URL after the clone finishes (which isn't too complicated, though, so it's not that big a deal)	18:46
perlmonkey2	I just don't understand why I culdn't clone the branch I want to work on.	18:46
jast	git generally clones all branches you see in the remote repo but it doesn't create local branches for all of them automatically	18:46
RandalSchwartz	you clone a repo, not a branch	18:46
perlmonkey2	ic....	18:47
	so there is some on disk db with huge numbers of diffs then.......	18:47
	← bryanl_ left	18:47
jast	so if you want to work on a certain branch, you create a local branch based on it	18:47
perlmonkey2	okay, I'm slow, but I'm think I'm getting it :)	18:47
jast	in git 1.6.1 and newer you use, for example, git checkout --track origin/somebranch	18:47
	→ bnovc joined	18:48
perlmonkey2	jast: after you've cloned the project you would do that?	18:48
jast	that will create a local branch called somebranch that's based on 'somebranch' from where you cloned	18:48
perlmonkey2	okay	18:48
jast	you can then work in that local branch and git knows where to take new stuff from if you use pull	18:48
perlmonkey2	that's pretty sweet	18:48
jast	that's not automatic because you can work with an arbitrary number of remote repositories at the same time, and git doesn't want to assume you want pull/fetch/push to use a particular remote by default	18:49
bnovc	whats the equivalent of "hg export <changeset>" to get a patch file for a given log entry	18:49
	"git show" is similar but it pops up in less and doesnt go to a file	18:49
perlmonkey2	jast: okay....I think I'm on the right track (thanks to you). I'll see if I can make something happen now :)	18:50
jast	bnovc: git format-patch for a series of patch files with metadata, or git show with shell redirect to get just a diff and a bit of information for whoever looks at the resulting diff file	18:50
	perlmonkey2: good luck :)	18:50
	← tjafk2 left	18:50
jast	basically, git format-patch outputs files that can be processed by git am (including commit message and author info)	18:51
bnovc	oh, i guess everything above the first "diff" is ignored w/patch	18:51
	→ ericindc joined	18:51
bnovc	i was thinking it requierd a hash	18:51
	→ quoin- joined	18:52
jast	patch deals with nothing but diffs, so it ignores everything else	18:52
	git apply works very similarly to patch; git am uses an extended format that is roughly like an e-mail	18:53
	← ericindc left	18:54
jast	in fact, git am can use mbox files if the e-mails are structured in a particular file, and format-patch can create mbox files or individual patch files that have this structure	18:54
bizhat2	i am working on branch master, when i release software i tag it with "git tag v1.2", after doing this, what is best practce to do ? Create a new branch and do work for new version ? I want to do in a way that if needed, i will be able to add bug fix for v1.2	18:54
	← quoin left	18:54
	← paltman94 left	18:55
	joshthecoder → joshthecoder_afk	18:55
	→ racerx joined	18:55
jast	bizhat2: that depends a bit on how many old versions you want to provide bugfixes for. the git repository, for example, has a master branch which the next release will be eventually tagged from, and a maint branch that's based on the current release and which bugfixes are committed to. maint has bugfix release tags derived from it.	18:55
	← hugoxrosa left	18:56
jast	in more detail, suppose 1.6.2 is the current series of releases, then maint will contain bugfixes and spawn tags like 1.6.2.1 and 1.6.2.2, whereas master contains new stuff that will go into 1.6.3	18:56
racerx	how do I generate an ssh key if my 'git' user on my git repo has it's password disable?	18:57
	→ makkalot joined	18:57
jast	the good thing here is that you can merge maint into master occasionally and thus get the bugfixes forward-ported with almost no effort	18:57
	→ ashleyw_ joined	18:57
	← dwave left	18:57
bizhat2	jast: thanks, let me try to understand :-)	18:58
jast	racerx: you use ssh-keygen locally and if you don't use anything fancy like gitosis on the target git account, you just add the public key part as a new line to the .ssh/authorized_keys file in the target account (create it if it doesn't exist, make sure both the dir and the file are readable by the target user only)	18:58
	→ cmarques joined	18:58
makkalot	hi i'm trying to pull my svn repo via git-svn clone but got that error and pulling is interrupted ".git/refs/heads/master was not found in commit" any ideas howto fix ?	18:58
	← lightcap left	18:59
jast	makkalot: can you provide the full output of git svn clone? e.g. by pasting it to http://git.pastebin.com/. if there is any sensitive information in it, feel free to edit it out.	18:59
	← nurey left	18:59
racerx	jast: actually I did use gitosis. I'm trying to use capistrano to deploy to my deployment server. I think that it needs the ssh key from my git repo. I could be wrong.	18:59
jast	that message on its own doesn't tell me anything. perhaps that's my fault, but it may just mean that I need more information.	19:00
	← wycats left	19:00
racerx	jast: were u referring to me?	19:01
jast	racerx: I don't know about capistrano. in gitosis, you generate a ssh key pair and add the <something>.pub file that it generates to the keys/ dir of the gitosis-admin repo, under a name like username.pub	19:01
	racerx: no :)	19:01
	racerx: sorry, the dir is called keydir/	19:01
	← scientes_ left	19:02
	→ ash joined	19:02
	joshthecoder_afk → joshthecoder	19:03
	← lhz left	19:03
racerx	my git repo is running fine. I can push and pull because I was able to give the git repo my local ssh key. I think and I could be wrong that when I tried to write to the deployment server I may have to put the git ssh key on the deployment server.	19:03
	← ash_ left	19:04
jast	I don't know. perhaps there's a channel for capistrano or something...	19:04
makkalot	jast, http://git.pastebin.com/m6c514901	19:04
jast	makkalot: did someone add a git repository to a svn repository there?	19:04
racerx	But if you are not familiar with capistrano+git it is alright. Thank you very much. I've been trying capistrano for a week and no one was responding. Again thanks.	19:05
jast	racerx: there are two more things you can do: 1) come back another time and hope you find someone who knows about it; 2) post to the git mailing list and ask for help or a direction to turn into	19:05
makkalot	jast, yes and i deleted it later via svn ,so these are old commits .git directory is not there actually	19:05
	→ ereslibre joined	19:06
	→ gigo joined	19:10
	→ henr_k joined	19:11
	→ paltman94 joined	19:11
	← gaveen left	19:11
	← ashleyw left	19:12
	→ shadowhand joined	19:12
	→ gaveen joined	19:13
	→ ash_ joined	19:14
	← gaveen left	19:14
	→ tillberg joined	19:15
	→ ericindc joined	19:15
	→ altrux joined	19:15
	← dtillberg left	19:15
	→ nurey joined	19:16
	← nurey left	19:16
	← racerx left	19:16
	← john3909 left	19:16
	→ coldboot joined	19:16
coldboot	Does git add go across different filesystems? Is there a way to make it do that or not do that, depending on the answer?	19:17
	← dwave_ left	19:17
	→ ednice joined	19:18
	← ash left	19:18
coldboot	I'm using it to manage chroots for a fragile cross-compiler toolkit, so when it breaks itself I can revert the changes.	19:18
ednice	Was pep new to the chat.anyone from by	19:19
	Ny	19:19
	→ aaron joined	19:19
aaron	is there an easy way to get a diff between a branch's head and the commit from which it branched off of the master?	19:20
	sorry for convolution	19:21
	→ Tv joined	19:21
	→ bcardarella_ joined	19:21
	→ fridim_ joined	19:21
	→ brizly joined	19:21
	→ larryp joined	19:21
	← ednice left	19:21
	→ montylounge joined	19:22
	→ rudi_s joined	19:24
	→ robinr joined	19:24
	← aaron left	19:25
struberg	spearce: maybe a dumb question, just to avoid misinterpretation. A call to Tree#addFile(f) doesn't add the content of a file, but only the 'name'?	19:27
	→ Bass10 joined	19:27
spearce	correct	19:27
	→ faithfulgeek joined	19:27
spearce	i think the id is set to null or ObjectId.zeroId()	19:28
	one or the other	19:28
struberg	I'm trying to get the jgit commit done	19:28
	← Bass10 left	19:28
	← ashleyw_ left	19:28
struberg	so this must also work if one did git-add myFile.txt	19:28
	→ xxx__ joined	19:28
struberg	and then use jgit commit to commit the things in the index	19:28
spearce	right	19:28
bnovc	jast: thanks	19:28
	→ ashleyw joined	19:29
	→ eetfunk joined	19:29
	→ Bass10 joined	19:29
eetfunk	I'm trying to use git-p4 but i always get an error on git-p4 submit. Here's the error I'm getting: http://www.spinics.net/lists/git/msg102536.html Anybody has an idea?	19:30
struberg	spearce: I didn't find a Tree function to add already existing objectIds, any hints 4 me?	19:31
spearce	struberg: FileTreeEntry f = tree.addFile("foo.txt"); f.setId(theId);	19:32
	no, its not pretty to work with.	19:32
struberg	spearce, txs!	19:33
	← rionave left	19:34
	← iron left	19:34
jast	whoever invented "txs" as an abbreviation? it makes no sense at all. ;)	19:34
	← markelikalderon left	19:35
	← deavid left	19:39
	← larryp left	19:40
	→ deavid joined	19:40
	→ tjafk joined	19:40
	← dreiss left	19:41
	← cmarques left	19:41
	← fridim_ left	19:42
	← javatexan left	19:43
	← Bass10 left	19:43
	→ fridim_ joined	19:43
	→ bcardarella joined	19:44
	→ jwrjared joined	19:45
	← fridim_ left	19:45
	← eetfunk left	19:47
	← bcardarella left	19:50
	→ Kbyte joined	19:50
	← jwrjared left	19:50
	← bcardarella_ left	19:51
	→ bcardarella joined	19:54
	← leanucci left	19:56
	← paltman94 left	19:58
	→ jchris joined	20:00
	→ gitte joined	20:00
	ChanServ set mode: +o	20:02
	gitte changed the topic to: 1.6.2.5 \| Homepage: git.or.cz \| Everyone asleep or clueless? Try [email@hidden.address] \| Channel log http://tinyurl.com/gitlog \| Mailing list archives: http://tinyurl.com/gitml \| Gits on git: http://tinyurl.com/gittalks \| Pastebin: http://gist.github.com/ \| GSoC '09: http://socghop.appspot.com/org/home/google/gsoc2009/git	20:02
	← aisbaa left	20:02
	→ spacepluk joined	20:03
	→ bonch joined	20:03
	→ krh joined	20:03
spacepluk	hi, is there any option to parse the author line added by cvsexportcommit in the commits generated by cvsimport?	20:04
	→ ttilley joined	20:04
	← ttilley left	20:04
gitte	spacepluk: they are not recorded as Authors?	20:05
	← mw left	20:05
	→ mw joined	20:05
	→ wwwald joined	20:05
robinr	gitte: there's on option that adds them to the message	20:05
	cvs uses "unix" style user id's	20:06
gitte	Ah, okay. Did not know that.	20:06
zuez	When a git clone happens... I'm only bringing down one branch from a remote repo. How do I clone all of the branches in that repo?	20:06
	→ thiago_home joined	20:06
gitte	robinr, spacepluk: I thought the mechanism was to use .mailmap...	20:06
robinr	the user authenticated to the server and not the one in the message	20:07
	← rudi_s left	20:07
robinr	gitte: you lose the information when exporting to cvs	20:07
spacepluk	gitte: when I use cvsimport the commiter appears also in the author field of the commit, even for commits that have a different author in the message body put there by cvsexportcommit	20:07
gitte	robinr: ah... but I cannot see why that matters (only to make commits different which should not be different...)	20:07
	← brizly left	20:07
charon	zuez: a clone by default gets all local (wrt the original repo) branches, and sets up 'master' to track 'origin/master' (which wrt the server is 'master'). see the remotes/origin/ branches for the rest.	20:07
	→ frsk joined	20:08
gitte	spacepluk: now I get it: cvsexportcommit.	20:08
zuez	charon: I have two branches on a shared repository, master/experimental... and when folks are git cloning it, they're only bringing down the experimental one	20:08
gitte	spacepluk: how about parsing it with a very simple sed script?	20:08
charon	zuez: so 'git branch -a' only shows origin/experimental, not origin/master?	20:09
robinr	maybe there should be an option to "sudo" during cvsexportcommit	20:09
zuez	charon: It shows both, but only a local experimental branch	20:09
spacepluk	gitte: you mean, after performing cvsimport ?	20:10
	← faithfulgeek left	20:10
charon	zuez: well then 'git checkout -b master origin/master' to create a tracking branch	20:10
gitte	spacepluk: you wanted to parse what is there, without modification, no?	20:10
Ilari	zuez: But its odd that it shows 'experimential' not 'master' (unless you have intentionally changed it). Leads me to suspect that something's wrong...	20:11
zuez	charon: any idea why a git clone is only tracking a branch I created off of master?	20:11
	→ babbage joined	20:11
zuez	llari: There used to be a working directory in the 'shared' repo... I might have done a git checkout experimental there at one point...	20:11
	llari: Maybe that's what's wrong	20:11
charon	zuez: the shared repo has a HEAD different from 'master'	20:11
	← lois[pac] left	20:11
charon	probably. i thought the default cloned branch would always be called master, but apparently i'm wrong.	20:12
Ilari	zuez: Shared repo with working dir? That's not correct.	20:12
babbage	Which hook gets called when "git fetch" succeeds?	20:12
Ilari	babbage: AFAIK, none.	20:12
babbage	Ilari, thanks. Maybe that was the wrong question.	20:12
spacepluk	gitte: not really. I've a team that commits to a git repo, then I export all the commits to a cvs repository, and import them back. But after importing I appear as commiter/author for all the commits.	20:12
zuez	llari: I got rid of the working directory after I realized it wasn't a bare repo...	20:12
Ilari	zuez: And it got 'locked in' state where 'experimential' is the default?	20:12
gitte	spacepluk: why on earth do you import back?	20:13
zuez	llari: I think that's the case, but I don't know all of the intriciate details of git	20:13
babbage	I actually want to send a change notification to buildbot when a repo changes. For example when a commit is made or a merge or rebase etc. happens. In fact more specifically, any time the head revision of any branch changes. How would I do that?	20:13
Ilari	zuez: Both "currently checked out branch" and "default branch" use HEAD as pointer.	20:14
zuez	llari: I'm not sure if I can make the 'default branch' on the shared repo back to master	20:14
spacepluk	gitte: the boss wants everything to be in the central cvs repository... and sometimes other people commits some patches to cvs directly... I'm open to suggestions :)	20:14
zuez	llari: Best bet will be to dupe it and just push out a local .git directory I think	20:15
Ilari	babbage: Such date isn't interesting until it hits public repos, and such are usually updated by push (which does call hooks).	20:15
gitte	spacepluk: "kill the boss" does not cut it, right?	20:15
Ilari	zuez: Its one command if you have shell access to it.	20:15
spacepluk	gitte: mmm, maybe I should try that... :P	20:15
gitte	spacepluk: can you control the central cvs server?	20:15
babbage	Ilari, not so. I'm running buildbot to test changes in my private repo.	20:15
zuez	llari: I do, but 'git checkout master' requires having a working directory, but it's a bare repo now	20:15
Ilari	zuez: 'git symbolic-ref HEAD refs/heads/master'.	20:16
gitte	spacepluk: I could imagine that cvsserver could help your problems. No more exporting, no more importing either.	20:16
	→ catalis_ joined	20:16
zuez	llari: Thanks, I'll do that	20:16
spacepluk	gitte: I guess I can talk to the admin to get full access to the server.	20:16
gitte	spacepluk: I know that back in 2006, I was able to switch over from cvs to git-cvsserver with a few users not noticing.	20:17
	→ tvw joined	20:17
gitte	spacepluk: those who noticed asked why the server is so fast all of a sudden.	20:17
Ilari	babbage: If operations are invoked by script, have that script call hooks?	20:17
	→ dreiss joined	20:17
zuez	llari: beautiful, that did it.	20:17
Ilari	zuez: It would also have been possible to use 'ln' tool (yes, actual symbolic links).	20:17
	zuez: Its one command change too...	20:18
babbage	Ilari, why would there be a script? I'm writing the code. I just use "git commit" or "git merge" like anybosy else does	20:18
	→ bdiego joined	20:18
	→ e_6 joined	20:18
zuez	llari: nod I need to brush up on what HEAD/ORIG_HEAD/refs etc are in the world of Git, that would have helped.	20:18
	← tillberg left	20:18
spacepluk	gitte: there's some "process" in the company to migrate to a real VCS. But we (onsite workers) need some solutions to work right now. This is not optimal but helps a lot :P	20:19
	→ neoeinstein joined	20:19
jast	zuez: the nickname actually starts with a capital "i" ;)	20:19
babbage	Ilari, I can't simply invoke the buildbot from pushes to the public repo, because the whole point of the testing with buildbot is to avoid pushing buggy changes!	20:19
zuez	woops. My fault	20:20
	I need a different font for my tty.	20:20
gitte	spacepluk: AFAIR I had to provide some hook or .bashrc which set the GIT_AUTHOR_NAME and GIT_AUTHOR_EMAIL according to the login, but that worked nicely.	20:20
jast	I'm sure we can forego punishing you just this once!	20:20
	→ faithfulgeek joined	20:20
jast	but next time there will be consequences. we'll force you to listen to really bad music for two hours.	20:21
	← WALoeIII left	20:21
gitte	jast: don't start a big I-know-the-worst-music contest!	20:21
Ilari	zuez: HEAD is (indirect) pointer to current branch head. ORIG_HEAD is previous value of HEAD as seen by some commands (its mostly obsoleted by reflog). refs/* are the various true refs, like branches, tags and remote tracking branches.	20:21
gitte	jast: that's even worse than the emacs-vs-vi and all the language wars!	20:22
jast	it's all a matter of opinion, isn't it, so fighting about it doesn't make sense	20:22
	taste, too	20:22
spacepluk	gitte: the actual problem is that we have some people (myself included) that are working on the client's house without internet	20:22
jast	and since we're all incredibly smart people in here, no such thing will ever happen	20:22
gitte	spacepluk: in that case, cvsserver should _really_ help.	20:22
	→ lightcap joined	20:23
gitte	spacepluk: you just work via Git, the others who really want to can work via cvs.	20:23
zuez	Ilari: thanks again, I appreciate it	20:24
babbage	gitte, I think you misunderstood spacepluk's problem which is, I assume: (s)he cannot change the master repo from CVS to git (yet)	20:24
gitte	babbage: because the boss wants a central cvs repository, yes.	20:24
	babbage: changing that to a cvsserver-served Git repository fulfills that mission.	20:25
	← montylounge left	20:25
spacepluk	gitte: mmmm, now I get it :) thanks!	20:25
	→ WALoeIII joined	20:25
	→ helo joined	20:25
gitte	spacepluk: yw	20:25
babbage	gitte, not really, because "cvs commit" stops working	20:25
	← ph^ left	20:25
gitte	babbage: ah, okay. So all those "committed via cvsserver" commits in my history do not exist... ;-)	20:26
Ilari	Hmm... Reminds me of, is that git cvsserver still too fast for CVS clients? :-)	20:26
	← WALoeIII left	20:26
babbage	gitte, I had no idea it supported commit now.	20:27
gitte	babbage: this was in 2006.	20:27
jast	"supporting commit" is the whole point of cvsserver, isn't it?	20:27
	otherwise you could have simply written a tiny script to mirror git stuff to a cvs repo and be done with it	20:28
gitte	babbage: from 3fda8c4: ... and add/remove/commit are supported.	20:28
Gitbot	[git 3fda8c4]: http://tinyurl.com/c5fsqp -- Introducing git-cvsserver -- a CVS emulator for git.	20:28
	← catalis left	20:28
gitte	babbage: it is from Feb 22 2006.	20:28
	→ WALoeIII joined	20:28
babbage	s/ now//	20:29
	← jmspeex left	20:30
helo	after i git reset --soft HEAD^ to 'undo' a bad commit, none of the changes i've made show up with "git diff"... what do i need to do to see what has changed?	20:30
babbage	So, I still need a way of triggering buildbot when I commit/merge/rebase.	20:30
	Any ideas?	20:31
helo	rather, what is going to be comitted with 'git commit -a -m ..."	20:31
	← coldboot left	20:31
babbage	helo, git diff --cached ?	20:31
	→ camwest joined	20:31
gitte	babbage: I have no idea if the update hook is enough for you, but I'd guess so.	20:32
	← lorandi left	20:32
	→ suman_gurung joined	20:32
	← faithfulgeek left	20:32
Ilari	helo: Also see 'git status -a'.	20:33
babbage	gitte, I assume not, since nobody is running "git push".	20:33
Ilari	babbage: There's post-commit, but merge might not invoke it.	20:34
babbage	Quite.	20:34
Ilari	babbage: At least delayed merge commit should invoke it...	20:34
gitte	babbage: indeed, you're correct, update is only called by receive-pack.	20:34
babbage	I'm surprised tbh, because this seems to me like a reasonably common use case.	20:35
gitte	babbage: I think there was some discussion a while ago on the list, and the conclusion was that if you call the programs locally anyway that you could install aliases to do the job.	20:35
	← catalis_ left	20:35
babbage	gitte, hmm, icky but possible.	20:35
gitte	babbage: you should be able to find the discussion on gmane by searching for "hook" and "merge", author "junio".	20:35
	babbage: in the alternative, you can write a patch and push for inclusion.	20:36
babbage	I'm already working on patches to buildbot, the GNU C manual, faketime, gnulib and findutils. So the idea that I could also work on changes to git is at this point largely theoretical.	20:37
	→ lorandi joined	20:37
Ilari	babbage: Alternatively, you might have another repo that buildbot monitors. Then just push there to trigger it.	20:37
babbage	Ilari, yes, I was just thinking about that.	20:37
	→ faithfulgeek joined	20:37
babbage	Ilari, if I do that with "git push --mirror" does it also work for non-fast-forward changes?	20:38
	From the manpage it looks like the answer is yes, now that I re-read it	20:39
	← d0k_ left	20:39
	← Bennid left	20:40
Ilari	babbage: Better to use repo with default refspecs like '+refs/heads/:refs/heads/' and '+refs/tags/:refs/tags/'. + means that git will always try to force this update.	20:40
	← Kimmie left	20:40
Ilari	babbage: Since you probably don't needs stuff like refs/remotes/...	20:41
	← bcardarella left	20:42
	→ lois[pac] joined	20:42
Ilari	babbage: Yes, multiple remote.foo.push lines are supported.	20:43
	← gotgenes left	20:44
	→ Radar joined	20:44
	← faithfulgeek left	20:44
	→ whaley joined	20:45
	→ VidiVici joined	20:45
VidiVici	Hi,	20:45
babbage	Ilari: remote.foo.push?	20:45
	→ earcar joined	20:45
Ilari	babbage: Sets what 'git push foo' actually pushes.	20:46
	babbage: a config option(s).	20:46
	babbage: One of the few that make sense to appear multiple times.	20:47
babbage	Oh I see, you mean I need one entry for +refs/heads/:refs/heads/ and another for +refs/tags/:refs/tags/ ?	20:47
Ilari	babbage: Yes.	20:49
	babbage: See --add of git config so it lets you add second value without overwriting the first.	20:50
	← DavidKlein left	20:50
	→ faithfulgeek joined	20:50
	← lightcap left	20:50
	→ mjf joined	20:50
	← Ryushin left	20:50
Ilari	babbage: Or directly edit .git/config	20:51
	→ Lothrik joined	20:52
	← rubydiamond left	20:54
	← spacepluk left	20:54
Lothrik	Two questions.. one, is there any way to get Git to work on Windows 7 (build 7100).. and two, is there any way to apply a .patch to a repository downloaded from github.com? It doesn't include any .git folder, so git am isn't letting me patch it :-/	20:55
	→ foutrelis joined	20:56
Ilari	Lothrik: Its snapshot then, not a repository. And git apply should still work.	20:56
	Lothrik: No idea about Windos 7... :->	20:56
jast	Lothrik: to the first question, have you tried msysgit? I doubt anyone has built anything much more special than that yet	20:57
VidiVici	Hi I am working on a project that would use as a default file	20:57
	sorry	20:57
Lothrik	mysgit worked on build 7000	20:57
	but stopped working in 7100	20:57
	→ coldboot joined	20:57
jast	go shout at microsoft, they'll love you for it ;)	20:57
Lothrik	haha	20:57
	← ashleyw left	20:58
Lothrik	I think it has something to do with msvcrt.dll, it got updated in build 7100 :-/	20:58
jast	that's a visual c library	20:59
	or visual c++ or whatever	20:59
Ilari	Lothrik: What error messages?	20:59
Lothrik	it just crashes git.exe when I go git clone	20:59
	or rather, it hangs	20:59
	→ wycats joined	21:00
	← khmarbaise left	21:00
	← jzellman left	21:00
babbage	Ilari, when I add the remote to my working repo, should I use "git add remote --mirror" or not use --mirror?	21:00
coldboot	Is there a way to get git to delete all untracked files?	21:01
	Excluding ignored patterns?	21:01
	← jrmuizel left	21:01
vmiklos	hmmm	21:01
Ilari	babbage: Probably not use mirror.	21:01
VidiVici	Hi, I'm new to git and have a few questions	21:01
babbage	Hmm, ok	21:01
vmiklos	i remember there was a good mailing list post about why merging too often is a bad idea.	21:02
Ilari	coldboot: git clean with some args?	21:02
VidiVici	1) Can I merge commits on a same branch?	21:02
vmiklos	(i mean pointless merges)	21:02
	→ jzellman joined	21:02
coldboot	Ilari: Thanks	21:02
gitte	jast: msvcrt.dll is just the c runtime.	21:02
	Lothrik: the issue is known, and hopefully fixed.	21:02
Lothrik	:o	21:03
	→ jzellman_ joined	21:03
gitte	Lothrik: or better, "worked around".	21:03
Lothrik	well that's nice to know :3	21:03
gitte	Lothrik: it would have been nice to know, too... IRC messages will be missed by most msysGit guys.	21:03
	→ sjzzalx joined	21:03
coldboot	Ilari: So if I'm using git to track the state of some chroot, I could checkout an old version, then run "git clean" to completely revert it back to the previous state?	21:03
	← struberg left	21:03
sjzzalx	Can I git log to origin? It doesn't seem to do anything except show local changes for me :(	21:04
Lothrik	I tried to fix it myself, had a friend send me the msvcrt.dll from Vista.. I put that in my \Git\bin\ folder, and added \Git\bin to my PATH.. but it still crashed :x	21:04
jast	sjzzalx: what do you mean by "log to origin"?	21:04
sjzzalx	I mean, can I do "git log origin" and have it display origin's changes?	21:04
Ilari	coldboot: Checkout should delete them if they become untracked and have no local changes.	21:04
sjzzalx	the commits etc on origin	21:04
	before I pull	21:04
Ilari	sjzzalx: Use fetch to just copy the commits.	21:05
	← VidiVici left	21:05
Ilari	sjzzalx: pull = fetch + merge	21:05
jast	sjzzalx: fetch and then try, for example, gitk localbranch...origin/somebranch	21:05
coldboot	What if they were never tracked at all by git?	21:05
Ilari	coldboot: Then they are left as is.	21:05
jast	coldboot: then clean will remove them	21:05
	→ jrmuizel joined	21:05
coldboot	Okay that's what I thought, just checking. Thanks guys.	21:05
jast	possibly with -d	21:05
	to be really thorough	21:05
coldboot	I'm using it to track scratchbox the maemo cross-compiler toolkit, because it's such a fragile piece of crap.	21:06
	← boto left	21:06
jast	hmm, my first approach would have been to use some virtualization app with differencing disk images	21:06
thiago_home	coldboot: I feel your pain	21:06
coldboot	thiago: Isn't it awful?	21:06
Lothrik	O_O..	21:06
thiago_home	yep	21:07
Lothrik	I just ran git clone from the command prompt.. and it's working	21:07
coldboot	thiago: I tried scratchbox 2 to make it better, but it's even more broken.	21:07
Lothrik	I have no idea why	21:07
	→ lightcap joined	21:08
	← commondream left	21:08
Ilari	coldboot: Can't UML mount image files plus virtualize a lot of the "system"? Of course, that SW is probably too broken to run under UML...	21:08
	→ boto joined	21:08
coldboot	Ilari: What's UML?	21:09
Ilari	coldboot: Version of Linux kernel that runs as user-mode process.	21:09
coldboot	Ilari: Ah.	21:09
	← mbroeker left	21:09
coldboot	Ilari: It's too late for that approach now, they've hacked it together with a bunch of shitty bind mounts.	21:09
Ilari	coldboot: Also, loop device plus namespaces?	21:10
	← faithfulgeek left	21:10
coldboot	Ilari: All probably better solutions than what they've got so far, but I'm not sure.	21:10
	→ bcardarella joined	21:10
Ilari	coldboot: Bad thing that namespace creation is priviledged operation (but so is populating a namespace).	21:11
	← cemerick left	21:11
Ilari	coldboot: each namespace has independent set of mounts.	21:12
	← bcardarella left	21:13
	← aspotashev left	21:13
	→ adam joined	21:14
	→ gilimanjaro joined	21:14
	← patrikf left	21:14
	adam → Guest55190	21:14
	← quoin- left	21:14
Guest55190	Hi. Would you say, that "Git is more 'arbitrary', less rules are protected so it's general less safe than CVS"	21:15
	Guest55190 → srnt	21:15
coldboot	What rules?	21:15
	← priidu left	21:15
babbage	I'm not sure that statement actually refgers to anything.	21:16
	rerfers	21:16
	dammit.	21:16
coldboot	haha	21:16
cehteh	you have 3 tries	21:16
coldboot	You get an alligator in spelling.	21:16
gitte	srnt: yes, Git is less safe than CVS. Maybe you wanted to go to #subversion?	21:16
cxreg	cvs is less safe than anything. you can change history with no proof that you did it.	21:16
perlmonkey2	wow, this is driving me nuts. So I did 'git clone project'. But I need the 'crypt' version of project. So in the downloaded directory, I do 'git checkout project/crypt', but that complains with a pathsec error. What could I be doing wrong?	21:17
	I mean, I need the crypt fork/branch of project.	21:17
	← Radar left	21:17
babbage	just "git checkout crypt".	21:17
srnt	gitte, no. I'm on the git side. I just heard someone saying sentence above.	21:17
	→ ilor_ joined	21:17
babbage	perlmonkey2, you can use "git branch" to list the branches.	21:18
srnt	gitte, and I wanted to find some less irrational picture	21:18
gitte	perlmonkey2: does "git branch -r" list anything like "origin/crypt"?	21:18
perlmonkey2	babbage: okay, thanks	21:18
gitte	srnt: irrational, alright.	21:18
babbage	srnt, well, the best thing to do is ask them to back their statement up with some specific examples, then.	21:18
perlmonkey2	gitte: nope, it only shows "master".	21:18
	← jzellman left	21:18
perlmonkey2	gitte: but on github if I view the allbranches tab, I see it there.	21:19
babbage	"git pull origin crypt:crypt" should add the remote branch you need.	21:19
	→ moonflux joined	21:19
gitte	srnt: if you really need to respond to something as handwaving as that (and cannot ignore that idiot, e.g. when she happens to be your boss): we use a cryptographic hash. That makes sure that the whole history is integer.	21:19
	→ ashleyw joined	21:19
	← sergiodj left	21:19
	← name left	21:19
perlmonkey2	babbage: sweet!	21:19
babbage	I don't think you meant integer, there.	21:19
gitte	perlmonkey2: did you omit the "-r"?	21:19
	← drewr left	21:19
	← dwave- left	21:20
	→ henr_k_ joined	21:20
gitte	srnt: and once anything is tagged with a GPG key, it is tagged. You cannot sneakily exchange any file as you can do in many other SCMs, very much including CVS.	21:20
	← henr_k_ left	21:20
perlmonkey2	gitte: yes, somehow I screwed that up.	21:20
	gitte: hmm, I need to grok the pull command then.	21:20
srnt	gitte, huh, that's concrete argument	21:21
gitte	perlmonkey2: so, it shows "origin/crypt" or what?	21:21
	srnt: helps you?	21:21
perlmonkey2	first time I've even seen it.....so you "pull" a fork to your local repo and it becomes your working copy?	21:21
gitte	perlmonkey2: could you please answer my question first?	21:21
perlmonkey2	gitte: yes, it shows origin/crypt and like 10 others including the master.	21:21
	← codebrulee left	21:21
perlmonkey2	gitte: yes, it matches the allversions tab on github.	21:21
gitte	perlmonkey2: instead of trying to solve the riddle by "pull"ing your repository into destruction?	21:22
perlmonkey2	err, allbranches	21:22
	← bdiego left	21:22
gitte	perlmonkey2: so, first you have to check out a local branch, and that has _nothing_ to do with "pull".	21:22
	← xyz left	21:22
srnt	gitte, yes, thank you	21:22
perlmonkey2	I haven't successfully checked anything out yet. I've only cloned the master.	21:22
gitte	perlmonkey2: checking out a branch is done with "git checkout <branch>", but that branch does not exist yet, it is only a "remote branch".	21:22
perlmonkey2	gitte: but my pull did add the files associated with the crypt branch.	21:23
babbage	Ilari, so the post-receive hook is doing the right thing, thanks. The buildbots now have 7 builds queued up (one for each branch head).	21:23
gitte	perlmonkey2: so you have to say "git checkout -t origin/crypt". The "-t" says: "create a local branch for that remote branch".	21:23
perlmonkey2	gitte: so I had to pull the remote branch into my local repo?	21:23
	← Weaselweb left	21:23
	← camwest left	21:23
gitte	srnt: yw!	21:23
	perlmonkey2: I have no idea where your repository is right now, as you chose to pull wildly instead of listening and understanding first.	21:24
	perlmonkey2: so maybe you need help from somebody else than me.	21:24
	dfr\|work → dfr\|vegas	21:25
	dfr\|vegas → dfr\|work	21:25
	← gigo left	21:27
	→ eddyp joined	21:29
	→ sergiodj joined	21:29
	← ilor_ left	21:31
	← ilor left	21:35
	← mfilipe left	21:36
	→ bcardarella joined	21:36
	← tbf left	21:38
	← jzellman_ left	21:38
	→ jzellman joined	21:38
	→ poseidon joined	21:38
coldboot	How do you revert a single file back to the HEAD version? Equivalent to "svn revert foo.c"	21:39
joshk	git checkout foo.c	21:39
	← gitte left	21:39
coldboot	joshk: Thanks	21:39
jast	that's not entirely correct, actually	21:40
	it reverts back to the last thing you had staged	21:40
	to revert back to HEAD, use git checkout HEAD foo.c	21:40
	pilif\|away → pilif	21:40
coldboot	jast: Ah ok	21:40
	jast: If I had manually staged part of a change, I would probably want to revert back to the last staged thing anyway.	21:41
jast	often it makes no difference... sometimes (especially when you're not careful with what you stage) it does	21:41
	yeah, sure	21:41
	← pilif left	21:42
	← lorandi left	21:43
	← Ryback_ left	21:44
cj	hurm... how do I tell git I'm done doing the manual conflict resolution?	21:44
	← dfr\|work left	21:44
	← makkalot left	21:45
Ilari	cj: git add	21:45
	→ plasticine joined	21:45
cj	ah, thanks	21:46
jast	you may also want to finish a potential operation that caused the conflicts, e.g. git merge or git rebase	21:47
	(or continue, in case of rebase)	21:47
	→ RaySl_ joined	21:47
cj	how do I finish the merge?	21:47
jast	git commit	21:47
Ilari	cj: git commit?	21:47
jast	it will automatically make it a merge commit	21:47
RaySl_	I just did a checkin that clobbered someone elses changes for some reason after I pushed	21:48
cj	ah, that's what I did... just making sure :)	21:48
	← plasticine left	21:48
RaySl_	is there a way to resolve this? his changes are more important than mine	21:48
Ilari	RaySl_: You forced it?	21:49
RaySl_	Ilari: nope	21:49
	← naeu left	21:49
jast	perhaps you can describe the sequence of events a bit more	21:49
RaySl_	I commited 4 files, and one of them blew away his updates to the same file	21:50
Ilari	RaySl_: Without forcing, it will refuse to push branches that would cause commits to be lost on remote side.	21:50
jast	well, just by committing you can't really obliterate someone else's stuff, so something else must have happened after that	21:50
RaySl_	jast: I was working locally and doing local commits	21:50
	I did a pull then a push	21:50
cj	I'd like to see more context about the changes shown in 'git diff origin', such as committer, commit message, etc... is there a way to get that information easily?	21:51
RaySl_	afterwards, it gave me no errors	21:51
	→ up_the_irons joined	21:51
Ilari	cj: 'git log -p HEAD...origin'?	21:51
	cj: OOps, 'git log -p HEAD..origin'	21:52
jast	RaySl_: and then what happened that made you (or the other person?) notice that something disappeared? and where did it disappear?	21:52
RaySl_	jast: the app stopped working, he made some pretty big changes to that file	21:52
	I only did a 1 line change	21:52
bizhat2	what is equivalent of svn export in git ?	21:53
	← wwwald left	21:53
	← mjf left	21:53
jast	well, did he commit those changes? did he push them? did they disappear from his working tree or from a central repo or from where else?	21:53
Ilari	bizhat2: git archive	21:53
cj	thanks, Ilari	21:53
coldboot	If I havea bunch of modified files that haven't been staged, how do I revert them?	21:55
	← jason237 left	21:55
	→ mjf joined	21:55
perlmonkey2	holy crap, I was able to commit a change to the github fork.	21:56
	finally	21:56
	← bonch left	21:56
bizhat2	Ilari: i run git archive [email@hidden.address] getting error fatal:Not a git respo (orany of the parent directories): .git	21:56
	any idea	21:56
	← boto left	21:56
	→ boto joined	21:56
Ilari	bizhat2: see the manual page (man git-archive)	21:56
Gitbot	Ilari: the git-archive manpage can be found at http://git.or.cz/man/git-archive	21:56
perlmonkey2	and if I want to keep several local forks alive at the same time, I can just clone my local master and then checkout forks on the clones. But I'm not sure what it matters if I pull down a fork to a local copy. It just makes it available right? doesn't break anythign?	21:57
Ilari	bizhat2: It needed some option for designating some remot repository for snapshot.	21:57
	← mike_m left	21:57
	→ hacim joined	21:58
Ilari	perlmonkey2: Watch out. Cloning won't copy remote branches and pushing to repo with working tree is good way to get confused.	21:58
	← tvw left	21:59
perlmonkey2	Ilari: So I should clone my local repo. checkout the branch I need in the clone. Then push to my local master. Then push that to the remote mastere?	22:00
	← krh left	22:00
Ilari	perlmonkey2: That 'push to my local master' doesn't sound good idea... Pushing to currently checked out branch screws the repo.	22:02
perlmonkey2	What I need is one more tut showing me yet one more way to do this.	22:02
	→ bryanl joined	22:02
bizhat2	Ilari: i can just git cone and delete .git folder ? that will work ?	22:02
	→ Leemp joined	22:02
Ilari	perlmonkey2: You can have multiple remotes defined for one repo and you can have multiple local branches.	22:02
	← bdimcheff left	22:02
	→ chuck joined	22:03
	→ pilif joined	22:03
perlmonkey2	Ilari: I think i'm back to square one on understanding this.	22:03
chuck	Hey, every time I pull in a git repo, git always does a merge and it logs it in "git log"	22:03
	Is there any way I can avoid this?	22:03
	AAA_awright_ → AAA_awright	22:04
Ilari	bizhat2: Well, based on git@, it might be using gitosis, which doesn't allow remote snapshotting...	22:04
	→ mweltin1 joined	22:04
perlmonkey2	Ilari: in my local master, I checked out the branch I need to work on. When I commit and push, my changes go to the remote branch. There are only two people working on this branch, so hopefully this is good enough.	22:05
Ilari	chuck: pull rebase mode? But understand the implications first.	22:05
chuck	Ilari: But that's not normal behaviour of git	22:05
Ilari	perlmonkey2: Remotes are just nicknames for other repos.	22:05
chuck	I haven't made any changes to the repository, yet it always "merges" and adds an entry to git log	22:05
	→ Travis-42 joined	22:06
mweltin1	I have a local that tracks changes of a remote branch. Does the command "git push" push my changes to the remote master and branch, or just the remote branch?	22:06
Ilari	chuck: Well, to combine two divergent branches without creating extra merges, you need to rebase. The alternative is to create new commit to pull the histories together.	22:06
mweltin1	assume I have the local branch checked out.	22:06
Travis-42	Is there a way to give someone only partial access to a git repository?	22:06
	← RaySl_ left	22:06
perlmonkey2	Ilari: I think i get that part. The remote branch I'm working with has the same status as if I cloned the origin to local and then cloned it as a new branch. I think.	22:07
Ilari	perlmonkey2: If you just clone the clone (without doing anything else), you only get one branch into the clone of clone.	22:08
	→ Radar joined	22:08
	→ jason237 joined	22:08
Ilari	perlmonkey2: Because the clone has only one local branch and that is cloned.	22:08
perlmonkey2	Ilari: the master?	22:08
	← suman_gurung left	22:08
moonflux	Hi folks. Got a nice little task and am not sure whats the best way to achieve it: I'm trying to glue together two local git repos, lets call them git0 and git1. git1 starts where git0 ends (two svn imports). So I guess I'll somehow have to get the objects of the one into the other and then rebase git1 onto git0 or something. Any idea?	22:09
Ilari	perlmonkey2: Most of the time. Master is just the default.	22:10
	← LiamH left	22:10
	← wwalker left	22:10
Ilari	moonflux: Sound like grafts + filter-branch	22:10
perlmonkey2	Ilari: I don't understand why people create multiple local clones. In the directory I cloned the remote master, I can switch between branches just by checking them out.	22:11
Ilari	moonflux: And for getting the commits, git fetch.	22:11
chuck	Ilari: How do I fix it though :< I don't know all of the git terminology, I just want this repository to not add those log entries	22:11
moonflux	Ilari: Already worked with filter-branch to clean them up but not sure how to pull the one into the others. Will ask google about grafts...	22:11
Ilari	perlmonkey2: Because they find local branches confusing? Hint: Install git bash completion and configure branch display in prompt.	22:12
perlmonkey2	Ilari: yeah, did that and it is awesome.	22:12
moonflux	Ilari: So I just fetch eg. git0 into git1 and rebase then?	22:12
perlmonkey2	Ilari: the whole concept of having all my branches in a single directory is just brilliant. Just one question, if I switch branches without committing changes, will they be lost, or show up when I switch back?	22:13
Ilari	moonflux: Rebasing doesn't necressarily work. Grafts can be used to override parents of commit (and filter-branch uses those when filtering).	22:13
perlmonkey2	and if you branch a branch before it is merged back into the master.....what happens then?	22:13
Ilari	perlmonkey2: The changes will follow (assuming checkout succeeds).	22:13
moonflux	Ilari: Hmm... from what I found from a quick googling grafts sound like exactly what I need. Thanks, I'll have a try.	22:14
	→ temoto joined	22:14
perlmonkey2	Ilari: thanks a million. I'm at least somewhat able to make this work now :)	22:14
temoto	Hello. Workflow question. We're trying to adopt "dozen topic branches" style. How do decide - whether to squash topic branch into master or to merge it?	22:15
Ilari	perlmonkey2: All info git really stores about a branch is pointer to its newest commit (plus few last values of that pointer, but that's not required).	22:15
sjzzalx	If I've marked a couple of files as --assume-unchanged, but I would actually like to know when upstream updates these, but not have git automatically do anything about it except tell me, is there an easy way to do that?	22:15
	← kumbayo left	22:15
	← jrmuizel left	22:15
sjzzalx	For instance, I mark settings.php as unchanged, but want to know when there's an update on origin so that I can look and see if I need to address that change in my file. Can I do this?	22:16
	← tefflox left	22:16
Ilari	temoto: Rebase -i it to split/merge into logical commits and then merge it?	22:16
	temoto: It might become just 1 commit, or it might become 10.	22:16
	← SRabbelier left	22:16
Ilari	sjzzalx: 'git diff HEAD..origin/master -- settings.php'?	22:17
	sjzzalx: Or the version with '...' instead of '..'.	22:17
	← lightcap left	22:19
	← releod left	22:19
temoto	Ilari: that's not for long living branches, right?	22:20
	→ spaceonline joined	22:20
	← agile left	22:20
Ilari	temoto: If they are long-lived, merge them anyway.	22:21
	temoto: The ordinary kind, not the squash kind.	22:21
	→ sgrimm joined	22:21
	→ sofia1 joined	22:22
sofia1	Hi ... I am new to git	22:22
temoto	Ilari: understood, thanks.	22:23
sofia1	I made some changes to my git repo and now I want to undo my changes...how can I do that?	22:23
Ilari	temoto: I regard squash merge as obsoleted by interactive rebase / SVN braindamage.	22:23
	← Leemp left	22:23
sofia1	Any help is highly appreciated !!	22:23
temoto	sofia1: git reset HEAD^	22:24
	sofia1: that will make it like you did not commit last changes.	22:24
	← Tuomas left	22:24
Ilari	sofia1: look up man git-reset. Watch out, it has some destructive modes (especially --hard).	22:24
Gitbot	Ilari: the git-reset manpage can be found at http://git.or.cz/man/git-reset	22:24
temoto	sofia1: but did changed files. git reset --hard HEAD^ to undo file changes too.	22:24
sofia1	Thanks, let me try	22:24
	→ offby1 joined	22:25
	← Sho_ left	22:26
	← Travis-42 left	22:27
babbage	How do I choose some options to pas to ssh when doing a push? (the options I want to set are -o BatchMode=yes -i ~/.ssh/id_patches)	22:27
Ilari	babbage: ssh config file....	22:28
	← helo left	22:29
	← srnt left	22:29
babbage	Ilari, urgh, that only allows me to choose based on destination host. And the destination is in fact localhost :(	22:29
Ilari	babbage: Create an alias?	22:30
babbage	Ilari, are you really saying that git spawns a shell to launch ssh?	22:30
spearce	no, he means an alias in ~/.ssh/config	22:31
	← ericindc left	22:31
spearce	e.g. "Host myspeciallocalhost\nHostname localhost\nBatchMode yes\n..."	22:31
Ilari	babbage: See 'Host' and 'HostName' in ssh_config(5)	22:31
babbage	yes, that could work	22:32
	→ lightcap joined	22:32
babbage	ick, though	22:32
	→ Travis-42 joined	22:32
	← bizhat2 left	22:33
	→ gotgenes joined	22:36
	← mweltin1 left	22:36
	→ jrmuizel joined	22:39
	← cmarcelo left	22:39
babbage	Ilari, yes, that seems to work pretty well.	22:39
	Thanks.	22:39
	← juanjoc left	22:41
	→ jmspeex joined	22:42
	← intripoon left	22:42
	← jackdempsey left	22:44
	← Travis-42 left	22:45
	sebleier → sebleier\|away	22:45
sofia1	Ilari: I have lost the latest commit...this is not i was looking for	22:46
	← lightcap left	22:46
Ilari	sofia1: 'git reflog' and reset to suitable commit.	22:47
sofia1	I made some changes to my git repo but I had not committed those...I wanted to get back to the point I started from	22:47
	← AAA_awright left	22:48
sofia1	still unable to get to the point I started	22:49
	→ AAA_awright joined	22:50
	→ paltman94 joined	22:50
	← Yuuhi left	22:51
sofia1	llari: I am actually started with a version v10 ...then I made changes to the repo but I did not commit those changes. However I wanted to get back to original v10 version...What reset HARD^ does is it takes me to version v9 and not original v10...do u have any ideas ?	22:51
	← boto left	22:54
	← vipul left	22:54
	→ lightcap joined	22:54
sofia1	Does anyone know how to undo the changes that are not commited in git ?	22:54
	← Lothrik left	22:57
mugwump	'git reset' will unstage them, 'git checkout HEAD filename' get the last checked in version of a file	22:59
	← RodP2 left	22:59
	→ Voight-Kampff joined	23:00
	← wycats left	23:01
	← Deiz left	23:01
moonflux	Ilari: Got it grafted, worked like a charm. Thanks a lot :)	23:01
Ilari	moonflux: You need to filter-branch it as well to make it permanent.	23:02
	→ _ke joined	23:03
_ke	hello	23:03
moonflux	Ilari: Jepp, did so already.	23:03
_ke	is there any way to _exclude_ certain directories from git log output?	23:03
Ilari	_ke: AFAIK, no. Only filter by directory/file modified (there can be multiple).	23:04
_ke	Ilari, yep, thats the other way round, which i did not want ;)	23:04
	← madewokherd left	23:04
	→ jerbear joined	23:05
Ilari	_ke: Exclude is more difficult than include to implement as well.	23:05
_ke	Ilari, of course	23:05
	Ilari, hm.. i just want to create a NEWS file and exclude all the translators	23:06
	← lightcap left	23:06
Ilari	_ke: Since with include you need just to look up tree IDs corresponding to all included entries, whereas with exclude you would need to look up all the others.	23:07
_ke	Ilari, pardon?	23:07
Ilari	_ke: If directory list is inclusive, it only needs to look up those from revisions to decide wheither there's change. With exclusive list, it would need to look up entries for other directories (possibly much more).	23:09
_ke	Ilari, ok	23:09
	← Voight-Kampff left	23:10
	→ Voight-Kampff joined	23:10
	← shadowhand left	23:10
	← jrmuizel left	23:12
	← jerbear left	23:13
temoto	sofia1: git checkout . will undo all uncommited changes.	23:15
	← rylar left	23:15
mugwump	or 'git reset --hard'	23:15
	← thiago_home left	23:15
temoto	sofia1: let us know if it what you want.	23:15
	it is*	23:16
	→ ericindc joined	23:17
	← astrocub left	23:18
	← bcardarella left	23:18
	→ astrocub joined	23:19
	→ agile joined	23:19
	→ scarabx joined	23:23
	← paltman94 left	23:24
	← sjzzalx left	23:25
	← jzellman left	23:29
	← xxx__ left	23:29
	→ madewokherd joined	23:30
	← wmoxam left	23:32
	→ Travis-42 joined	23:34
	→ javatexan joined	23:34
	→ boto joined	23:35
	← romao left	23:37
	← aziz_ left	23:37
	→ hectorsq joined	23:39
	← sergiodj left	23:39
	← Travis-42 left	23:40
	← mjf left	23:40
	← hectorsq left	23:41
	← bobmcw left	23:42
	← bryanl left	23:44
	← coldboot left	23:45
	← earcar left	23:45
	← moonflux left	23:47
	← WALoeIII left	23:47
	← alezandro left	23:51
	← temoto left	23:54
	→ jwash joined	23:55
	← dodo_the_last left	23:55
	→ bryanl joined	23:56
jwash	http://www.Marie-gets-Deflowered.com/?id=43a43334	23:56
	← jwash left	23:57
	→ bremner_ joined	23:57
	→ JohnnyL joined	23:58
	← JohnnyL left	23:58
	→ jackdempsey joined	23:59

Logs Search ←Prev date Next date→

Channels Documentation