IRCloggy #git 2009-04-24

Provider of IRC logs since 2005.

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

	2009-04-24
	← haesbaert left	00:00
	← dotsintacks left	00:00
	→ [intra]lanman joined	00:01
	← RodP2 left	00:02
quentusrex	Ilari: I don't see an error anywhere... :(	00:05
Ilari	quentusrex: Oh, and you also need to +x the hook to enable it.	00:06
quentusrex	yup, did that...	00:06
	I have: /home/git/repositories/cgit-admin.git	00:06
	only cgit.conf and .git in there.	00:06
	I just did a git reset --hard and it didn't pull anyting... :(	00:07
Ilari	quentusrex: The git reset is run in post-update hook of the server repo.	00:07
quentusrex	sec	00:07
	I might have found it	00:07
	error: unable to create temporary sha1 filename ./objects/61: Permission denied	00:08
Ilari	quentusrex: Permissions problems.	00:09
	→ shf10105_ joined	00:09
Ilari	quentusrex: On remote side.	00:09
quentusrex	win!!!	00:09
	thanks Ilari	00:09
shf10105_	hi everybody, im quite new to git, is there a good toturial for setting up a private git server?	00:10
Ilari	quentusrex: Basically, if you are using file:///, the receive-pack on remote side will execute with priviledges of user invoking push.	00:10
quentusrex	ok, now I'm seeing an issue with cgit about 'path' is not a git repo	00:11
Ilari	shf10105_: Here's one tutorial: http://scie.nti.st/2007/11/14/hosting-git-repositories-the-easy-and-secure-way	00:11
	← kalle_ left	00:11
	→ kalle_ joined	00:11
shf10105_	Ilari: Thanks	00:11
	sdiz-afk → sdiz	00:11
	← shruggar left	00:11
Ilari	shf10105_: One thing it doesn't mention (and I have seen it cause trouble) is that filesystem contaiing the git user home directory must be mouted with execute permissions.	00:13
	shf10105_: Not as 'noexec'.	00:13
	iratiku → iratik	00:14
Ilari	shf10105_: It doesn't need to have permissions to have devices nor SUID/SGID.	00:14
quentusrex	nvm, I found it.	00:14
Ilari	shf10105_: That's because gitosis uses hook script in one place and if partition is noexec, executing that hook will fail, resulting problems.	00:15
	→ mithro joined	00:15
	→ agenteo joined	00:16
	→ esparkman joined	00:19
	← fujin left	00:22
	→ jackdempsey joined	00:22
	← halfmouse left	00:23
quentusrex	Ilari: how do I do a test to see if something is a git repo?	00:24
	→ faithfulgeek joined	00:24
quentusrex	cgit keeps telling me that a folder isn't a git repository	00:24
Ilari	quentusrex: IIRC, it uses presence of 'HEAD' 'refs' and 'objects' inside the .git dir.	00:25
quentusrex	but I hmm	00:25
	hmm....	00:25
Ilari	quentusrex: Probably one can also cause that error by having wrong permissions somewhere.	00:26
quentusrex	but I have one working,	00:26
	and another one that isn't...	00:26
	I have one with the project.git/.git	00:26
	and all the HEAD, etc are in the .git	00:26
	but the other one has everything in the project.git/ rather than in the /.git/ folder.	00:27
	→ k0001 joined	00:27
Ilari	quentusrex: git status should tell wheither it can find repo there...	00:27
	quentusrex: That is a bare repo...	00:27
quentusrex	which is a bare repo?	00:27
Ilari	quentusrex: The one without .git directory and instead with directories like refs and objects	00:28
quentusrex	ok,	00:28
	so, can I convert a non bare repo to a bare repo?	00:29
	→ CardinalNumber joined	00:30
	← antoranz_ left	00:31
Ilari	quentusrex: Conversions in both directions are possible. And note that bare repo doesn't have working tree.	00:31
quentusrex	basically, how do I get cgit to show the gitosis repo ?	00:32
	:(	00:32
Ilari	quentusrex: Just the repository files, in form that nothing but git or one of its reimplementations can get anything meaningful out.	00:32
	← tchan left	00:32
	← sakyamuni left	00:33
Ilari	quentusrex: No idea about configuring cgit...	00:33
	← jfkw left	00:33
offby1	So ... what's the _nice_ way to check out "the branch I last had checked out"? I _could_ do it like this, but it's kludgeaceous: git checkout $(git reflog \| head -1 \| awk '{print $6}')	00:34
	... is there a better way?	00:34
Ilari	offby1: 'git checkout @{-1}'? Requires quite new git...	00:34
	→ tchan joined	00:35
	← jackdempsey left	00:35
offby1	slik!	00:36
	er, "slick" :-)	00:36
	Ilari: that's fine; I live on the bleeding edge :)	00:36
	→ jackdempsey joined	00:36
offby1	Ilari: you never let me down, y'know that?	00:36
	hmm, 'git checkout -', even.	00:37
	inspired by "cd -", no doubt.	00:37
	next -- stacks of branches ... :-)	00:37
	← dgs__ left	00:37
	→ dgs__ joined	00:38
Ilari	Hmm... 1.6.3.rc0.1.gf800b... That's too old. :-)	00:38
offby1	f800b	00:39
	too old for what?	00:40
Ilari	Well, 1.6.3.rc1.51.gea0b7 is already out... :-)	00:40
	offby1 stares blankly	00:40
Ilari	Look up "heatseeker".	00:41
	← glommer left	00:41
Ilari	The comp slang term, not as in "heatseeker missile".	00:41
	← jackdempsey left	00:44
	← catalis left	00:45
	→ hax0r1337 joined	00:45
	← KwikkSilva left	00:45
	← bryanl left	00:46
	→ yec joined	00:46
	← Dashkal left	00:49
	→ glommer joined	00:50
	→ javatexan joined	00:51
	← cgardner left	00:51
	← ProperNoun left	00:52
	→ eletuchy_ joined	00:53
	← eletuchy_ left	00:53
	→ eletuchy_ joined	00:54
	← eletuchy left	00:54
	← dreiss left	00:58
	← fynn left	00:59
	→ fynn joined	01:00
	→ seangrove joined	01:00
	← andrewkasper left	01:01
	→ catalis joined	01:02
	← madduck left	01:03
	← bmalee left	01:03
	← GyrosGeier left	01:03
	← impulze left	01:03
	← vmiklos left	01:03
	← Tali left	01:03
	← incd^ left	01:03
	← daleglass left	01:03
	← hegge left	01:03
	← avtobiff left	01:03
	← janl left	01:03
	← xl0 left	01:03
	← replaca left	01:03
	← darjeeling left	01:03
	← hgb left	01:03
	← piroko left	01:03
	← majoh left	01:03
	← natlus left	01:03
	← xkr47 left	01:03
	← dionoea left	01:03
	← frsk left	01:03
	← OgMaciel left	01:03
	← IRSeekBot left	01:03
	← Ingmar left	01:03
	← pcgod left	01:03
	← maxb left	01:03
	← mithraic left	01:03
	← seb42 left	01:03
	← yinkei left	01:03
jjuran	http://web.bilkent.edu.tr/Online/Jargon30/JARGON_H/HEATSEEK.HTML	01:03
	→ GyrosGeier joined	01:04
	→ vmiklos joined	01:04
	→ OgMaciel joined	01:04
	→ xl0 joined	01:04
	→ hegge joined	01:04
	→ Ingmar joined	01:04
	→ daleglass joined	01:04
	→ natlus joined	01:04
	→ pcgod joined	01:04
	→ maxb joined	01:04
	→ replaca joined	01:04
	→ IRSeekBot joined	01:04
	→ incd^ joined	01:04
	→ piroko joined	01:04
	→ frsk joined	01:04
	→ yinkei joined	01:04
	→ bmalee joined	01:04
	→ xkr47 joined	01:04
	→ hgb joined	01:04
	→ majoh joined	01:04
	→ mithraic joined	01:04
	→ dionoea joined	01:04
	→ Tali joined	01:04
	→ impulze joined	01:04
	→ seb42 joined	01:04
	→ darjeeling joined	01:04
	→ madduck joined	01:04
	→ avtobiff joined	01:04
	→ janl joined	01:04
	→ navik joined	01:04
justatheory	Ilari: I've almost got this thing.	01:06
	→ hugoxrosa joined	01:08
arohner	how do I generate a tarball from a git tag?	01:08
justatheory	Now I just need to be able to fetch remote branches from the cvs export with into the svn export without replacing the branches with the same names in the svn export.	01:09
	Starting to drive me crazy.	01:09
	← loincloth left	01:09
nis	arohner: man git-archive ?	01:09
Gitbot	arohner: the git-archive manpage can be found at http://git.or.cz/man/git-archive	01:09
arohner	nis: yes, thanks	01:10
Fissure	justatheory: something like git fetch ~/cvs-import refs/remotes/:refs/remotes/cvs/	01:11
	→ haesbaert joined	01:11
	← kalle_ left	01:11
justatheory	Fissure: Yeah, but I don't want the "cvs/" bit there.	01:11
	So I have the branch foo in cvs and the branch foo in svn.	01:11
	→ kalle_ joined	01:11
justatheory	I want to merge them into one git repo, and have the earliest commit in foo in the svn git repo to be the parent of the latest commit in the foo branch in the cvs git repo.	01:12
	I have to do this with master, frankly	01:12
Fissure	i don't think you can automate that very easily	01:16
justatheory	Crap.	01:16
Fissure	just have to manually graft them together	01:16
justatheory	I can, but some of the commits are missing.	01:16
	If i use `git fetch` to get those commits, they replace those in the existing branch.	01:16
	→ iso42 joined	01:17
Fissure	so fetch them into another branch	01:17
	like i said	01:17
justatheory	Grrr. Was hoping to avoid that.	01:17
Fissure	why? you can just delete them later	01:18
	→ fujin joined	01:18
	→ ceej joined	01:19
Fissure	you can try to automate it by using git for-each-ref to iterate over branches, then doing some string operations combined with rev-list\|head and rev-list\|tail to link the histories together	01:20
	← hax0r1337 left	01:20
	→ devicenull_ joined	01:21
	← mw left	01:21
Fissure	i.e., echo `git rev-list svn/$branch \| tail -n1` `git rev-list cvs/$branch \| head -n1` > .git/info/grafts	01:21
	and loop that over each branch	01:21
	→ mw joined	01:22
	← spearce` left	01:24
	→ bremner_ joined	01:24
	→ halfmouse joined	01:25
	← catalis left	01:26
	← devicenull left	01:26
	← hugoxrosa left	01:27
	← WALoeIII left	01:28
	→ hugoxrosa joined	01:28
	← haesbaert left	01:28
	→ jackdempsey joined	01:30
justatheory	There are only three of them, master, rev_1_6, and rev_1_8.	01:33
	I already have the SHA1's to write to .git/info/grafts	01:33
	But if I put those branches into cvs/, graft them to their children, and then delete them won't they be gone?	01:34
	justatheory tries that.	01:34
	← iratik left	01:34
Fissure	no, though the objects will be up for garbage collection if no ref points to them	01:34
justatheory	okay	01:35
	trying	01:35
Fissure	so you'll need to filter-branch to solidify the grafts	01:35
	remember that branches are just pointers to commits	01:35
justatheory	okay, trying that	01:36
	→ merlin83 joined	01:37
	← bremner_ left	01:39
	→ KwikkSilva joined	01:39
	← gaveen left	01:44
	→ gaveen joined	01:45
	← jackdempsey left	01:45
	← sgrimm left	01:47
	← esparkman left	01:48
	← mooseish left	01:50
	→ mooseish joined	01:51
justatheory	Fissure++ # Worked, thanks!	01:51
	← lois[pac] left	01:52
	← halfmouse left	01:53
	← poseidon left	01:55
	← hobodave left	01:56
Fissure	great! enjoy using a sane version control system :)	01:57
	→ porcnick_ joined	02:00
	← ocornu left	02:00
	← hugoxrosa left	02:01
	→ disappearedng joined	02:02
	→ hax0r1337 joined	02:02
justatheory	thanks.	02:02
	Just a little bit more and i'll be done.	02:02
	→ scarabx joined	02:03
	← KwikkSilva left	02:03
	← ceej left	02:05
	← qhoxie left	02:08
	→ KwikkSilva joined	02:10
	→ qrush_ joined	02:11
	→ porcnic__ joined	02:12
	← Ademan left	02:13
	← hax0r1337 left	02:14
	← porcnick left	02:16
	← porcnick_ left	02:17
	→ Ademan joined	02:17
	→ KwikkSilva_ joined	02:18
	→ mankind_tweezer joined	02:18
	→ agile joined	02:19
	← KwikkSilva left	02:19
	← Tv left	02:19
	→ sgrimm joined	02:20
	→ cytrinox_ joined	02:20
	→ [1]intripoon joined	02:21
	→ Aiatek joined	02:23
	→ sfwc joined	02:23
	← Aiatek left	02:24
	← qrush_ left	02:24
	→ flaguy48 joined	02:27
	← nis left	02:27
	→ hyperair joined	02:28
	→ qrush_ joined	02:28
	→ blinton joined	02:31
	→ hiwd joined	02:35
	← cytrinox left	02:36
	← intripoon left	02:37
	[1]intripoon → intripoon	02:37
	← flaguy48 left	02:37
	→ qrush__ joined	02:39
	← qrush left	02:40
	→ manguy joined	02:42
	← markelikalderon left	02:45
	→ rphillips joined	02:51
	→ tjafk2 joined	02:51
	→ hax0r1337 joined	02:51
	→ baetis-fly joined	02:53
	← qrush_ left	02:54
	→ jackdempsey joined	02:54
	→ Paf joined	02:55
	← hiwd left	02:56
	→ aspotashev\|eeepc joined	03:02
	← arohner left	03:03
	← nurey left	03:05
	→ aspotashev joined	03:05
	← tjafk1 left	03:07
	→ Tv joined	03:09
	→ ajonat joined	03:14
	→ egoist joined	03:15
	→ girishr joined	03:16
	← girishr left	03:16
	← xyz left	03:16
	→ ben_h joined	03:20
	← kukks left	03:20
	→ arohner joined	03:21
	← shf10105_ left	03:25
alus	I get:	03:25
	error: Entry 'blah/foo.rb' would be overwritten by merge. Cannot merge.	03:25
	fatal: merging of trees 4	03:25
	how do I fix that?	03:25
	← hyperair left	03:30
	→ hyperair joined	03:30
Fissure	uh, don't have untracked files that the remote side committed?	03:31
	← benn_d left	03:32
	→ davidad joined	03:35
	← manguy left	03:35
	← blinton left	03:35
	→ blinton joined	03:35
	→ kabus joined	03:37
	← bratsche_ left	03:38
	→ bratsche_ joined	03:39
alus	Fissure: I don't know what you're asking. I previously git added, committed and pushed that file	03:39
	→ eno__ joined	03:40
	← CardinalNumber left	03:40
	→ alanhaggai joined	03:40
	← ajonat left	03:42
	← eno left	03:44
	→ cytrinox joined	03:45
	← mw left	03:45
	→ mw joined	03:45
	← hax0r1337 left	03:48
	→ jmspeex joined	03:48
	← alanhaggai left	03:48
	→ alanhaggai joined	03:48
	→ gwoo joined	03:49
	→ rphillip- joined	03:50
Fissure	alus: oh, right, my bad... it's tracked, but you have local changes to it	03:50
	so it won't let you trash your working copy by trying to merge the other side's changes	03:51
	← octagon left	03:51
	← arohner left	03:52
	← rettub_ left	03:52
	→ rettub_ joined	03:53
	← blinton left	03:53
	→ blinton joined	03:53
	← rphillips left	03:55
	← rphillip- left	03:55
alus	Fissure: I made no additional changes	03:56
	Fissure: I also tried deleting the local file, and it gives the same error	03:56
	→ rphillips joined	03:56
	← cytrinox_ left	03:57
	→ Tricon joined	03:59
	← iso42 left	04:02
Fissure	deleting the file may count as a change :)	04:03
	what does status say?	04:03
	← mw left	04:04
	→ petercoulton_ joined	04:06
	← intripoon left	04:06
	← petercoulton_ left	04:06
	→ mw joined	04:06
	← alanhaggai left	04:07
	→ Zenopus_ joined	04:11
	→ aspotashev_afk joined	04:12
	← aspotashev left	04:12
	← jrmuizel left	04:13
e_6	i have a repo on a webserver that just gets updated via pull, and i keep getting a file is not uptodate error on pull. tried git update-index --refresh, it tells me the file needs update...what exactly does this mean?	04:14
	← gaveen left	04:14
	← blinton left	04:14
	→ blinton joined	04:15
	sdiz → sdiz-afk	04:15
	sdiz-afk → sdiz	04:15
	→ gaveen joined	04:15
	← blinton left	04:15
	← ben_h left	04:15
	← rphillips left	04:16
	→ rphillips joined	04:16
	→ blinton joined	04:17
	← porcnic__ left	04:17
	→ psyq joined	04:17
e_6	if i remove the file it works, but its kind of annoying having to do that every time on a repo that never gets edited except via pulls	04:18
	→ davidfetter joined	04:18
Fissure	sounds like an autocrlf issue	04:19
	← greyface left	04:19
e_6	what's that all about?	04:20
	← scarabx left	04:20
e_6	nm	04:20
	→ jrmuizel joined	04:21
e_6	thank you sir	04:22
	← jackdempsey left	04:23
	← e_6 left	04:24
	← mw left	04:25
	← dgs__ left	04:26
	← gaveen left	04:27
	← Zenopus left	04:28
	Zenopus_ → Zenopus	04:28
	→ jackdempsey joined	04:28
	→ intripoon joined	04:29
	← seangrove left	04:31
alus	Fissure: I did a git add then a git commit and now I can pull. wee	04:33
	git is weird	04:34
	← psyq left	04:35
	→ greyface joined	04:35
	→ jceb joined	04:37
	→ joelmichael joined	04:40
	← jackdempsey left	04:41
	→ porcnick joined	04:42
	← joelmichael left	04:42
	→ raydeo joined	04:43
	→ gaveen joined	04:44
	→ defvayne23 joined	04:44
	eno__ → eno	04:44
	→ girishr joined	04:45
	← kabus left	04:45
	→ chrisrb joined	04:46
Fissure	alus: git prevents you from destroying data	04:47
	← eletuchy_ left	04:48
	← girishr left	04:48
	→ eventualbuddha joined	04:50
	→ girishr joined	04:51
	← mooseish left	04:51
	← girishr left	04:51
	→ girishr joined	04:52
	← defvayne23 left	04:54
	→ error404notfound joined	04:55
error404notfound	How can i save the output of git push/clone etc commands in a custom log file?	04:56
alus	Fissure: there was no data destruction operation that should have been possible here. I added a file, and committed it. git should know how to do that without panicing	04:56
	→ ceej joined	04:56
	→ aisbaa joined	04:56
	→ savonarola joined	04:57
	→ ben_h joined	04:57
	→ priidu joined	04:57
	AkumaStreak → AkumaStreak_	05:00
	← curvature left	05:00
	→ curvature joined	05:00
	← chrisrb left	05:01
	← justatheory left	05:02
	→ kate21de joined	05:02
	→ wshimmy joined	05:03
Ilari	alus: What if the merge goes wrong?	05:03
	error404notfound: Shell redirections or even tee(1)?	05:03
alus	merge of what? there's nothing to merge	05:04
	I only did one thing	05:04
	← AkumaStreak_ left	05:04
Ilari	alus: Pull does merge.	05:04
	alus: If file A would be change by merge and A has local uncommitted changes, merge will abort.	05:06
alus	but there were no uncommitted local changes	05:07
	and more importantly, there were no remote changes.	05:07
Ilari	alus: Some caches where out of whack? 'git status' should fix those...	05:07
	alus: Then the merge should quickly say it has nothing to do.	05:08
	→ mr_chapendi joined	05:08
	← blinton left	05:08
	→ alester joined	05:08
error404notfound	tried git clone git@localhost:gitosis-admin \| tee log.txt and the log.txt just contains: "Initialized empty Git repository in /Users/shoaibi/tmp/gitosis-admin/.git", not the progress, or stats about how manuy files were transfered	05:09
mr_chapendi	hey guys, i'd appreciate it if you could spare some time to answer a question for me	05:09
	→ Enselic joined	05:10
wereHamster	error404notfound: ... 2>&1 \| tee ?	05:10
baetis-fly	I didn't get a response in #github, so I hope no one minds if I ask here. Do you think this is good advice for someone who is a regular contributor?	05:10
	http://blog.mhartl.com/2008/10/14/setting-up-your-git-repositories-for-open-source-projects-at-github/	05:10
	← kalle_ left	05:11
error404notfound	wereHamster, nope, no use, even though cloning goes just smooth, log contains just one line..	05:11
	← savonarola left	05:11
Ilari	error404notfound: Ah yeah, terminal detection.	05:11
mr_chapendi	i guess i'm a bit confused about the concept of a central repository in git - i'm aware that such a thing is kind of an artificial distinction, as git is decentralized	05:12
	but say i dsignate some repository as the central repository, and i push to it	05:12
Ilari	error404notfound: Well, screen has some form of logging (<action> H)	05:12
mr_chapendi	and then someone else pushed to it	05:12
	does it merge the two, and who makes the decisions on any conflicts?	05:13
	or am i just so far off base that i'm not making sense?	05:13
	→ Grahack joined	05:13
Ilari	mr_chapendi: It rejects push on conflict.	05:13
alus	I think git has problems with file permissions on windows	05:13
wereHamster	mr_chapendi: the second person will have to pull first, then push	05:13
	→ sitaram joined	05:13
alus	that's my guess for what it thought changed	05:13
Ilari	mr_chapendi: And if you force it, it overwrites other's changes.	05:13
allagator	heh	05:13
	← jrmuizel left	05:13
	← raydeo left	05:14
Ilari	mr_chapendi: Any merging happens on local end. Push never creates any new states.	05:14
error404notfound	i am using "script" utility	05:15
mr_chapendi	so, if theres a conflict, you pull the central server's state, merge it locally, making all the decisions, and then push again?	05:15
	→ WALoeIII joined	05:17
	→ kalle_ joined	05:17
	→ travisjeffery joined	05:19
	← aisbaa left	05:20
	→ jony joined	05:20
	→ seangrove joined	05:21
	→ aisbaa joined	05:21
	→ jrmuizel joined	05:24
	← mr_chapendi left	05:24
	← eventualbuddha left	05:24
	← porcnick left	05:25
	→ gretch joined	05:25
	← ceej left	05:28
	← giallu left	05:28
	← fujin left	05:31
	← johan-s left	05:31
	→ gaveen_ joined	05:31
	← gaveen left	05:31
	← alester left	05:33
	→ ankit9 joined	05:35
Modius	Git going mega-slow doing checkins - "calling pre-commit hook", commit-msg hook etc taking ages for one measly file. Okay, this is on MSYGIT; but it's usually fast for me. Anything I should look at?	05:36
	← davidad left	05:36
Modius	No HD action or anything like that	05:36
Enselic	what kind of hooks are invoked?	05:37
	→ davidad joined	05:38
	← mankind_tweezer left	05:38
	→ bcardarella joined	05:39
	← egoist left	05:39
	← cao1 left	05:40
	← jrmuizel left	05:41
	← wshimmy left	05:41
Modius	I don't know what hooks are :) Pre-commit and commit-msg	05:42
	← Evanlec left	05:42
	→ porcnick joined	05:43
	→ solar_ant joined	05:43
	← sitaram left	05:44
	→ sitaram joined	05:45
	→ peteinlux joined	05:46
Fissure	look at .git/hooks/*	05:46
	apparently they're doing some really expensive operations	05:46
Modius	Weird that they just suddenly get slow	05:49
sergiodj	there's always a chance that you find a "sleep 1000" there, though :-P	05:49
	please, disconsider the last comment... :)	05:49
	← priidu left	05:52
	→ jceb81 joined	05:53
	← kate21de left	05:55
	→ priidu joined	05:55
	← priidu left	05:57
	→ priidu joined	05:58
	→ j416 joined	06:00
	← gaveen_ left	06:01
	→ Weaselweb joined	06:01
	← Grahack left	06:02
	→ wshimmy joined	06:05
	← jceb left	06:05
	← Enselic left	06:06
	→ gauge joined	06:09
	← WALoeIII left	06:09
	← aspotashev\|eeepc left	06:09
	→ WALoeIII joined	06:09
	→ aspotashev\|eeepc joined	06:10
	← kalle_ left	06:11
	← porcnick left	06:11
	→ kalle_ joined	06:11
	← davidad left	06:13
	→ tubastian joined	06:13
tubastian	hi everyone	06:13
	→ binjured joined	06:15
	← aspotashev_afk left	06:15
	→ aspotashev joined	06:15
	→ Grahack joined	06:16
	← ben_h left	06:16
	→ amahi joined	06:17
	← aspotashev left	06:18
	→ aspotashev joined	06:19
tubastian	i am quite new to git and have this question: i use git-svn to work on subversion repositories. my main repository is A, that i checked out via "git svn clone", nothing special about this, but there is a second repository B which I want "to import" in my A repository, e.g. A/other/B... on the one hand I want to change B in my own branch in A and commit them to A, on the other hand I want to keep the connection between A/other/B and the orig	06:20
	← fynn left	06:22
	→ fynn joined	06:22
	← cpg left	06:23
wereHamster	tubastian: you want to import B into a subdirectory of your A clone?	06:23
tubastian	yes	06:23
	the B svn repository is readonly	06:23
	so all changes i make should go into A	06:23
wereHamster	and you want to commit the result into A..	06:23
tubastian	yes	06:24
wereHamster	subtree merge comes into my mind...	06:24
	← WALoeIII left	06:24
wereHamster	eg. create a git-svn clone of both svn repos, and do a subtree merge of B into the appropriate subdirectory of A	06:25
	→ Tuomas joined	06:25
	← hyperair left	06:27
	→ BooK_ joined	06:29
	← BooK left	06:29
	← davidfetter left	06:30
	→ davidad joined	06:32
	← bratsche_ left	06:33
	→ bratsche_ joined	06:35
	→ dwave joined	06:37
tubastian	wereHamster: thanks, i found a manual at http://www.kernel.org/pub/software/scm/git/docs/howto/using-merge-subtree.html , but it fails at step 2:	06:37
	repositories A is checked out at ~/git/A, B in ~/git/B and i am currently in a sub-directory of a:	06:37
	git remote add -f 'B' ~/git/B	06:37
	git merge -s ours --no-commit B/master	06:37
	Merge with strategy ours failed.	06:37
	the	06:37
	the -v option does not give further information	06:37
	→ naeu_ joined	06:38
	→ tal67 joined	06:39
	← aisbaa left	06:39
	→ srid joined	06:40
	→ naeu joined	06:40
	→ jfkw joined	06:42
	→ giallu joined	06:44
	→ hyperair joined	06:47
	→ voker57 joined	06:49
	← k0001 left	06:50
tubastian	:/ g2g, thank you so far	06:50
	← tubastian left	06:50
	← chalky left	06:51
	← stuffcorpse left	06:53
	→ stuffcorpse joined	06:53
	← jony left	06:56
	← naeu_ left	06:57
	← Aides left	06:59
	→ rraasch joined	07:00
	← davidad left	07:00
	← altrux left	07:02
	→ davidad joined	07:02
	→ Aides joined	07:03
	→ khmarbaise joined	07:03
	← jceb81 left	07:03
	← petercoulton left	07:03
	← hyperair left	07:06
	→ hyperair joined	07:06
	← robinr left	07:08
	→ gilimanjaro joined	07:10
	← darrob left	07:11
	← kalle_ left	07:11
	→ galderz joined	07:12
	→ kalle_ joined	07:12
	← Paf left	07:14
	← stuffcorpse left	07:15
	→ tatsuyao joined	07:15
	→ stuffcorpse joined	07:16
	→ darrob joined	07:16
	← hyperair left	07:18
	→ jony joined	07:18
	← davidad left	07:19
	→ hyperair joined	07:19
	→ jmesnil joined	07:20
	→ davidad joined	07:20
	← stuffcorpse left	07:21
	→ stuffcorpse joined	07:21
	→ lack_ joined	07:22
	← priidu left	07:23
	→ thumper joined	07:24
	→ chalky joined	07:28
	← lack left	07:28
	lack_ → lack	07:28
	→ ph^ joined	07:29
	→ drizzd_ joined	07:30
	← srid left	07:32
hgb	I get this:	07:33
	[09:33:00][hgb@rdbuild2:~/src/dev/git (revert)]172 git svn dcommit -n	07:33
	Cannot dcommit with a dirty index. Commit your changes first, or stash them with `git stash'.	07:33
	[09:33:09][hgb@rdbuild2:~/src/dev/git (revert)]173 git status	07:33
	# On branch revert	07:33
	nothing to commit (working directory clean)	07:33
	Any ideas?	07:33
	Hmm.	07:34
	A 'git svn dcommit' worked...	07:34
	Weird.	07:34
	krawek → marciano_cocoman	07:34
	→ ocornu joined	07:35
	← Grahack left	07:36
	→ johan-s joined	07:36
	← naeu left	07:37
	→ vuf joined	07:42
	← j416 left	07:42
vuf	it's getting to the point where even stash, checkout, make fix, checkout back, rebase, stash apply is getting annoying ... will that git-new-workdir thing help me, and are there some gotchas in using it?	07:43
	← drizzd left	07:46
	← dedmajor left	07:55
	→ Grahack joined	07:56
	→ gaveen joined	07:56
	→ dedmajor joined	07:58
	→ dwt joined	07:58
	→ plasticine joined	07:59
	→ shruggar joined	08:00
	← travisjeffery left	08:01
	← plasticine left	08:02
	→ kbingham joined	08:02
	→ lucsky joined	08:03
	→ naeu joined	08:03
	← tango_ left	08:04
	→ whalesalad joined	08:05
	→ aspotashev_afk joined	08:05
	← greyface left	08:06
	→ Paf joined	08:06
	→ vienova joined	08:06
	← aspotashev left	08:07
whalesalad	Hey guys I just moved a remote svn repo to a new url. I'm using git-svn, I opened my .git/config and changed the url but none of my git-svn commands work any longer. Was there something else I needed to do?	08:07
	← FreakGuard left	08:08
	← g3d left	08:08
wereHamster	whalesalad: yep, there is. I don't know how to do that, but there was a blog post or tutorial about it. Try google	08:10
	← kalle_ left	08:11
	→ kalle_ joined	08:12
	→ Chaser_ joined	08:12
	← bgerber left	08:13
	← mithro left	08:13
	→ g3d joined	08:14
	marciano_cocoman → krwk	08:17
	→ robinr joined	08:17
whalesalad	Dear god git-svn is disgusting. Time to get these other developers on the git bandwagon	08:17
	← Predominant left	08:17
	← krwk left	08:19
	← robinr left	08:22
	← peteinlux left	08:22
	→ airtonix joined	08:23
	← mapreduce left	08:24
	← agenteo left	08:24
	→ priidu joined	08:25
	→ travisjeffery joined	08:25
airtonix	suppose i have three projects (a,b,c) and project-b & project-c require project-a as a dependancy. is it possible to get the git repos of project-b & project-c to include a certain branch or the master head of project-a ?	08:26
	like some kind of ln -s ?	08:26
doener	checkout submodules or subtree merges	08:26
	→ sampled joined	08:26
airtonix	cheers	08:27
	→ aspotashev_ joined	08:29
	← kuadrosx left	08:29
	← bobesponja left	08:31
airtonix	ok so inside ~/project-b , i do : git pull -s subtree project-a master; git add *; git commit -am "some more changes"; git push --tags origin master;	08:31
	→ cilly joined	08:31
airtonix	after doing all the other prep steps	08:31
	→ cedricv joined	08:32
airtonix	is this done at every commit on project-b?	08:32
	to ensure latest version of project-a ?	08:32
	refering from here : http://www.kernel.org/pub/software/scm/git/docs/howto/using-merge-subtree.html	08:33
doener	that's the subtree way, and no, I'd not do that all the time. Only when it's "interesting" to get a new version of A merged in	08:34
airtonix	ok	08:34
	→ RodP2 joined	08:35
	← Grahack left	08:35
doener	and it depends on how the two things relate and how you want the thing to work out whether you'll want submodules or subtree merges	08:35
airtonix	well i know that every new pull of project-a wont conflict with anything thats unique about project b or c	08:36
	since all unique files for b & c are seperate directories	08:36
doener	the subtree merge really combines the two repos. So when you clone B, you get the merged parts of A, too, automatically, as a part of B	08:36
airtonix	and project-a is yet its own directory	08:36
	← bcardarella left	08:36
doener	but with submodules, you just have a link to A in B. You then get to have a clone of A in a subdirectory of B, as its own repo.	08:37
	→ eletuchy joined	08:37
doener	B just records where A is, and which commit should be checked out	08:37
	← eletuchy left	08:38
airtonix	but i still need to manually pull down project-a while im within project-b if i want the new changes from project-a before commiting project-b?	08:39
doener	sure. If B should have its link to A updated, you need to get the new commits for A first	08:40
airtonix	ok	08:40
	← travisjeffery left	08:40
doener	if the directory in which A is was always at the tip of some branch, that would easily lead to trouble when you need an old version of B	08:40
	say you had version 1.0 of A and B, and they worked together. Later you go for v2.0 of A, which is incompatible to v1.0	08:41
wshimmy	is there a way to add patchwise an untracked file?	08:41
airtonix	maybe i ought to look at joining them at distribution instead?	08:41
	← priidu left	08:41
doener	Then you couldn't checkout an old version of B, if A was always taken to be the latest version	08:42
	but as submodules are recorded to be a specific commit, checking out an old version of B also gets you an old version of A	08:42
	← aspotashev_afk left	08:43
	← aspotashev\|eeepc left	08:43
	→ HG` joined	08:44
	← Paf left	08:44
	→ savonarola joined	08:46
	→ aspotashev joined	08:46
	→ aspotashev\|eeepc joined	08:46
	→ markelikalderon joined	08:47
	→ nickjohnson joined	08:48
nickjohnson	How do I revert a 'git rebase'?	08:48
	← davidad left	08:48
doener	nickjohnson: "git reflog show <your_branch>", look for the pre-rebase entry and reset your branch back to it	08:49
nickjohnson	thanks!	08:50
	→ bobmcw joined	08:51
	← bobmcw_ left	08:51
	→ eletuchy joined	08:52
	← tal67 left	08:53
	→ tal67 joined	08:55
	← whalesalad left	08:58
	→ fridim_ joined	08:58
	← nickjohnson left	08:58
	← aspotashev_ left	08:59
	→ offby1` joined	09:01
doener	wshimmy: you mean "git add -p" on an untracked file, picking only parts of it to be added?	09:02
wshimmy	doener: yep	09:02
doener	wshimmy: "git add -N <file>", then you can use "git add -p" (though only the "e"dit thing will make sense) or "git gui" to stage parts of it	09:05
	→ priidu joined	09:06
wshimmy	doener: perfect, thank you (again!)	09:07
	i didn't know -N existed...its right in the man page, sorry :-/	09:08
	sdiz → sdiz-afk	09:08
doener	well, it's not quite obvious that you need it. I only found that solution because I could make some sense out of the "no changes" message from "git add -p" and prior knowledge about -N	09:10
	→ travisjeffery joined	09:10
doener	one could argue that "git add -p" should automatically do that add -N... up for a patch? :-)	09:11
	← kalle_ left	09:12
	← hyperair left	09:13
wshimmy	doener: sure thing, but now that i can commit properly i need to finish something else this morning first :P	09:14
doener	heh :-)	09:14
	← cilly left	09:14
	→ cilly joined	09:15
	→ kalle_ joined	09:17
	← RodP2 left	09:23
	→ AMorozov joined	09:23
AMorozov	Hi!	09:24
	Is it possible to setup a shared git repo so that pushed commits would automatically be rebased if it's possible?	09:25
	← tal67 left	09:27
	→ Bonaparte joined	09:29
sitaram	not likely	09:29
AMorozov	consider the following workflow: Joe and Jane work on completely different tasks (subdirectories) in their copies of a single repo. When Joe decides to push his changes, he can't do it w/o previous previous/rebasing because Jane pushed her changes before. Is it possible to setup central repository (hooks or whatever) to perform server-side rebasing at least in certain cases?	09:29
Bonaparte	Is it possible to use svn:externals with git?	09:29
AMorozov	Bonaparte: no.	09:30
	You may use git submodules though.	09:30
sitaram	AMorozov: a rebase is the same as merge in the sense of possibly having conflicts, so it cannot be done on the remote side	09:30
AMorozov	sitaram: yes, I understand this.	09:30
	sitaram: but I'd like to handle certain cases where I know exactly that conflicts are simply impossible.	09:31
	← thumper left	09:31
AMorozov	w/o any "human" interaction.	09:31
sitaram	I'm not very familiar with the various hooks but you may be able to trick one of them...	09:32
	← aspotashev left	09:32
Bonaparte	AMorozov, got a link to a tutor that describes how to use an external subversion repo with git?	09:32
	→ SRabbelier joined	09:32
	← savonarola left	09:34
AMorozov	Bonaparte: also you may check a Google Code project called `repo`, if I remember correctly, a script wrapper over plain git. It implements semantics similar to svn:externals (i.e. no binding to an exact version of a software by default).	09:34
	→ porcnick joined	09:34
AMorozov	Bonaparte: look at this http://git.or.cz/gitwiki/GitSubmoduleTutorial	09:35
	→ petercoulton joined	09:35
	→ aspotashev joined	09:36
Bonaparte	AMorozov, thanks. But it is going over my head at the moment	09:36
	I will search for a simpler tutorials	09:36
	amahi → cpg	09:36
Bonaparte	So, submodule is the way to go to manage external dependencies?	09:36
	→ gilimanjaro_ joined	09:37
AMorozov	Bonaparte: the basic idea of git submodules is pretty simple :) In fact, git-submodules itself is just a shell script, I would call it a proof of concept :)	09:37
Bonaparte	Nice :)	09:37
	→ aziz joined	09:38
AMorozov	Internally git-managed filesystem trees can have a thing called "GitLink", similar to well-known unix symlinks, but instead of keeping a path to a target file, a gitlink keeps a SHA1 of an external git repository.	09:39
Bonaparte	Does it natively interface with svn repos?	09:40
AMorozov	Bonaparte: no, certainly no.	09:41
	→ thumper joined	09:41
AMorozov	Bonaparte: you need to convert an SVN repo to a git repo (using git svn) and then link git repos using git submodules.	09:41
Bonaparte	I get the idea now	09:42
	← solar_ant left	09:42
	→ tango_ joined	09:42
AMorozov	Just remember that gitlinks, the base of git submodules, link to an exact commit in external repo, so it differs from svn:external where you usually link to a branch, rather than a single commit.	09:43
	← vuf left	09:44
AMorozov	If you need the latter semantic you may check Google repo mentioned above.	09:44
	→ fujin joined	09:44
	→ mpoullet\|work joined	09:45
mpoullet\|work	good morning	09:45
	how to add a bunch of untracked files all at a time?	09:46
	→ solar_ant joined	09:46
	→ fujin_ joined	09:46
Bonaparte	I am okay with a particular svn commit	09:46
doener	for example, "git add ." would add all untracked files in the current directory and all subdirectories	09:46
	mpoullet\|work: whether that matches the bunch of files that you want to add is a different question ;-)	09:47
mpoullet\|work	doener: thanks, exactly what I wanted!	09:47
	← gilimanjaro left	09:51
	→ jrydberg joined	09:52
	← mpoullet\|work left	09:55
	→ neoeinstein_ joined	09:57
	← AMorozov left	09:57
	→ egoist joined	10:00
	← fujin left	10:02
	→ sea-gull joined	10:04
	→ tal67 joined	10:05
	← girishr left	10:09
	← aeruder left	10:09
	→ literal_ joined	10:10
	→ markoa joined	10:13
	← aspotashev left	10:14
	← neoeinstein left	10:15
	← markelikalderon left	10:17
	→ parasti joined	10:21
	← Flathead left	10:23
	← cilly left	10:24
	← ocornu left	10:25
	→ aeruder joined	10:26
	→ robtaylor joined	10:26
	→ jrockway joined	10:27
	→ girishr joined	10:28
	← priidu left	10:31
	→ bryanl joined	10:32
	← porcnick left	10:33
	→ priidu joined	10:33
	→ [1]intripoon joined	10:35
	sdiz-afk → sdiz	10:35
	← priidu left	10:35
	← egoist left	10:36
	→ priidu joined	10:36
	→ g3d_ joined	10:38
	← g3d left	10:38
	← priidu left	10:39
	← sea-gull left	10:39
	→ Predominant joined	10:39
	→ bgerber joined	10:39
	← bryanl left	10:39
	→ priidu joined	10:40
	→ sea-gull joined	10:41
	→ ceyusa joined	10:45
	→ Kimmie joined	10:48
	→ alanhaggai joined	10:50
	← fynn left	10:51
	← tango_ left	10:51
	← intripoon left	10:52
	[1]intripoon → intripoon	10:52
	← travisjeffery left	10:54
	← zpoley left	10:55
	← markoa left	10:56
	→ tango_ joined	10:57
andreaja	I'm using git svn and due to a snafu one of the SVN revisions came down wrong, is there any way to force git to redownload some revisions?	10:58
	→ psankar joined	10:58
	→ botanicus joined	10:59
	← psankar left	11:00
	← aspotashev\|eeepc left	11:00
	→ bryanl joined	11:00
	← solar_ant left	11:02
	→ kadoban joined	11:02
	← samjam left	11:02
	→ iListenU joined	11:03
	→ bratsche__ joined	11:03
	→ solar_ant joined	11:03
	→ samjam joined	11:03
	← bratsche_ left	11:03
	→ tatsuyao_ joined	11:05
	← samjam left	11:05
	← gretch left	11:05
	← tatsuyao left	11:05
	← gilimanjaro_ left	11:05
iListenU	Where can be problem? here's screen: http://parduoduv.puslapiai.lt/ (in top)? What i need to do to solve it??	11:06
	Where can be problem? here's screen: http://parduoduv.puslapiai.lt/ (in top)? What i need to do to solve it??	11:06
	← iListenU left	11:06
	→ psankar joined	11:06
	← psankar left	11:06
	← faithfulgeek left	11:08
	← qrush__ left	11:08
	→ aspotashev joined	11:10
	← kalle_ left	11:11
	→ kalle_ joined	11:12
	→ botanicus_ joined	11:13
	→ ehamberg joined	11:14
	← kadoban left	11:14
	cpg → cpg\|away	11:15
ehamberg	is there a way to mark a file as "NEVER commit changes to this file?". i often do stupid things for testing, and want to have a more reliable guarantee for those files to be committed than me being awake. :)	11:15
Weiss	ehamberg: .gitignore? (note that your .gitignore files don't have to be committed themselves (afaik))	11:18
	← stianhj left	11:19
parasti	can't ignore a tracked file	11:19
Weiss	hmm, ok	11:19
	→ kadoban joined	11:19
dohzya	I think you must put a test in your hooks	11:20
sitaram	ehamberg: commit when asleep is retractable, as long as you don't push while asleep :)	11:20
ehamberg	dohzya: yeah, i guess that would work.	11:21
	sitaram: hehe, good point.	11:22
	→ cilly joined	11:22
Zaba	and no, you need to commit the .gitignore itself	11:22
	← KwikkSilva_ left	11:24
Weiss	blah, ok	11:24
	Weiss shuts up :)	11:24
ehamberg	.git/info/exclude would work if one should choose to go with the "ignore" approach, i guess	11:25
sitaram	Zaba: you don't need to commit the .gitignore	11:25
	→ Yuuhi joined	11:25
sitaram	it won't propagate of course but locally, it'll work	11:25
	→ corni joined	11:25
dohzya	depending : you want to prevents YOUR mistakes (not commited .gitignore / info/exclude / local hook), or ANY CHANGES (commited .gitignore / hook on public repo) ?	11:27
	← merlin83 left	11:29
	← botanicus left	11:30
dohzya	must a freeze feature could be nice I think :)	11:30
	s/must/but/	11:30
	→ aspotashev_afk joined	11:30
	← aspotashev left	11:31
	→ faithfulgeek joined	11:31
ehamberg	i want to prevent my mistakes to be committed locally and ultimally pushed.	11:32
	"git freeze login.py" would be nice :)	11:32
dohzya	so a big warning should work and be painless :)	11:33
alus	git says "error: path 'db/schema.rb' is unmerged"	11:33
dohzya	(because I guess git-freeze should be really hard to design...)	11:33
alus	but I deleted that file. I want the one from the remote repo	11:33
	how do I check it out?	11:33
	→ haesbaert joined	11:34
dohzya	git checkout {repo}/{remote branch} -- {file}	11:35
alus	that says "error: path 'db/schema.rb' is unmerged"	11:35
	oh, with the origin/master it works	11:36
	I thought that was default	11:36
	→ solar_ant_ joined	11:36
	← ankit9 left	11:36
alus	ok, now git pull says "error: Entry 'db/schema.rb' would be overwritten by merge. Cannot merge."	11:36
	I just want to pull ;_;	11:36
	there are clearly no local changes - I just got a fresh copy	11:37
dohzya	did you commit after deleting the file ?	11:37
alus	no	11:37
	git status says "modified: db/schema.rb"	11:37
dohzya	git reset --hard <<-- you will loose all local changes	11:37
alus	git diff says nothing	11:38
	so	11:38
	that's ok with me, since there are no local changes	11:38
	why is git so confused	11:38
	→ rashman joined	11:38
dohzya	because it is fixable :)	11:38
alus	ok, now git pull says "CONFLICT (content): Merge conflict in db/schema.rb"	11:39
	I "resolved" the "conflict" in the file. now what?	11:39
dohzya	git commit	11:39
alus	"db/schema.rb: needs merge"	11:39
dohzya	did you resolved the conflict by hand or using git tools ?	11:40
alus	by hand	11:40
dohzya	so it doesn't know it's resolved :)	11:40
alus	ok, where is git resolve? ;)	11:40
dohzya	I'm really noob with all merge features, so I always use git mergetool	11:41
	← sampled left	11:41
	← solar_ant left	11:41
alus	"No known merge resolution program available."	11:41
dohzya	I'm sure there is a way to say "it's resolved", but I don't know it	11:41
	oO	11:41
alus	windows	11:41
dohzya	-t {the name of your tool}	11:42
wshimmy	you need to remove the conflicts by hand if you aren't using a mergetool	11:42
dohzya	wshimmy: how to say "ok it's resolved" ?	11:42
alus	-t does not understand paths?	11:43
	← priidu left	11:43
alus	it doesn't understand the command, either	11:43
	→ rashman00 joined	11:43
wshimmy	ach, sorry, not scrolled all the way down :P	11:43
	← rraasch left	11:43
	← rashman left	11:43
wshimmy	dohzya: once you resolve it, just commit?	11:43
	← rashman00 left	11:43
alus	commit says "fatal: cannot do a partial commit during a merge."	11:43
dohzya	I think you must to tell git that it's resolved	11:43
	but I don't know how	11:43
	→ rraasch joined	11:44
alus	uh, I just did git commit -a	11:44
	that committed something	11:44
	now I can pull again. woo!	11:44
dohzya	oO	11:44
alus	it says "Merge branch 'master' of [email@hidden.address]	11:45
	I have no idea what that means. I'm not merging any branches	11:45
dohzya	pull = fetch + merge	11:45
	→ g3d joined	11:45
	→ boto joined	11:45
wshimmy	dohzya, alus: you need to add it first and then commit it, just like any other commit	11:45
dohzya	haha ok, a simple git add	11:46
	← g3d_ left	11:46
alus	ok so every commit is a merge then?	11:46
	the file was already added, long ago	11:46
wshimmy	thats why commit -a "worked" - but if you had local changes you didn't want committed in your working dir, they were just committed	11:46
alus	there were no local changes	11:46
wshimmy	alus: then you're fine	11:46
alus	except the merge resolution	11:46
	er	11:46
	conflict resolution	11:46
wshimmy	alus: but you know you need to add the file before each commit	11:46
alus	that doesn't make sense to me	11:47
	don't you only need to add the file once?	11:47
wshimmy	alus: once you've added it once, its tracked by git	11:47
alus	it was added ages ago	11:47
wshimmy	alus: but every time you change it, you need to add it to the index so that its part of the next commit	11:47
alus	what	11:47
wshimmy	alus: unless you only commit by using git commit -a	11:47
alus	sometimes I commit like this: git commit file1 file2 -m "blah"	11:47
	that always works, except in this case	11:48
wshimmy	alus: thats equivalent to git add file1 file2 && git commit -m "blah"	11:48
	→ d0k joined	11:48
alus	except in this case.	11:48
	→ priidu joined	11:49
alus	git commit db/schema.rb said "fatal: cannot do a partial commit during a merge."	11:49
	→ qrush joined	11:49
	← bryanl left	11:49
wshimmy	what was part of your original commit? more than just db/schema.rb?	11:49
alus	I don't understand what you mean by original commit	11:49
wshimmy	alus: your attempted merge commit	11:50
alus	I had committed lots of things. then I got internet access, so I did git pull. then I got a conflict in db/schema.rb	11:50
	← cilly left	11:50
wshimmy	alus: okay. that's because git pull == git fetch && git merge	11:50
dohzya	alus : if you just want to see where the remote is, just do a fetch	11:51
	← faithfulgeek left	11:51
alus	I wanted to push, but I couldn't push until I pulled	11:51
	→ j416 joined	11:51
dohzya	:)	11:52
wshimmy	alus: so you needed to merge your local branch with the remote branch. in your case it was not a fast-forward, so git needs to make an additional 'merge' commit with both commits (tips of the local and remote branches) as parents	11:52
	→ gilimanjaro joined	11:52
wshimmy	alus: you couldn't push because your history didn't match that of the remote	11:53
alus	sure	11:53
	so what's confusing to me is what happened after I resolved the conflict	11:53
	by hand	11:53
	everything up to that point makes sense, and is what I wanted	11:53
wshimmy	alus: then you needed to manually make the 'merge' commit that git was trying to do when you used git pull	11:54
alus	I see.	11:54
	← jony left	11:56
	→ hdoria joined	11:58
dohzya	just to avoid possible problems : put a "--" before putting name in your command lines ;)	11:59
	git commit -m "truc" -- file1 file2	12:00
	← sunblush left	12:03
hdoria	i am trying to import a patch sent on mail body. i save the content on a text file and used "git am -s saved-email" but i am getting this error: Patch does not have a valid e-mail address. i am sure it does. any tips?	12:03
wshimmy	hdoria: because git am acts on a mbox/maildir	12:05
	hdoria: try git apply for a single patch	12:05
hdoria	i want to keep the original mail (from the person that sent the patch)	12:05
	is it possible?	12:06
	→ faithfulgeek joined	12:06
wshimmy	hdoria: neither git apply nor git am will modify your mail at all AFAIK	12:06
	← g3d left	12:06
	→ g3d_ joined	12:07
	→ cilly joined	12:07
	literal_ → literal	12:08
	→ appiappi joined	12:10
	← hummingkid left	12:10
	← appiappi left	12:10
	→ appiappi joined	12:11
	← kalle_ left	12:11
appiappi	hi, i'm trying to build a static git binary	12:12
	anybody have experience with it	12:12
	i'd like to upload this binary via ftp to a virtual host and execute it with php to ease the deployment of updates	12:12
teuf	:wi n17	12:12
	← appiappi left	12:13
	→ appiappi joined	12:14
	← appiappi left	12:15
	→ kalle_ joined	12:17
	← j416 left	12:20
	← LotR left	12:22
	← dedmajor left	12:23
	→ dedmajor joined	12:24
	→ lorandi joined	12:24
	→ gitte joined	12:24
	← Bonaparte left	12:25
	→ j416 joined	12:25
	→ bcardarella joined	12:26
	← haesbaert left	12:26
	→ juanjoc joined	12:26
	→ bryanl joined	12:29
	→ sikanda joined	12:31
	← priidu left	12:31
	→ krh_ joined	12:33
	→ jackdempsey joined	12:34
	→ LotR joined	12:35
	→ gitte_ joined	12:35
	← gitte left	12:35
	gitte_ → gitte	12:36
	→ voker57_ joined	12:36
	← faithfulgeek left	12:36
	← jackdempsey left	12:37
	← cilly left	12:37
	→ daKmoR joined	12:37
	← qrush left	12:38
daKmoR	howto create a pach file for unstaged/uncommited changes?	12:38
Ilari	daKmoR: 'git diff' with suitable switches?	12:38
	← javatexan left	12:39
Ilari	daKmoR: But you should probably commit and creat patch out of the commit.	12:39
	daKmoR: After all, this isn't SVN.	12:39
daKmoR	ok...	12:39
	thx	12:39
Ilari	daKmoR: Also, for keeping uncommitted changes, there's 'git stash'.	12:39
	→ RaySl joined	12:40
Grum	mmmm why would my git branch -a on one box list: remotes/origin/XXX and on the other origin/XXX (without the remote/) -- config seems to be identical enough	12:40
	and the versions only differ from 1.6.2.1 on the 'remotes/' and 1.6.2.3 on the other o.O	12:41
Ilari	Grum: Does it list 'heads/origin/XXX' as well?	12:41
	← bcardarella left	12:41
Grum	remotes/origin/HEAD -> origin/master	12:41
Ilari	Grum: Is there 'origin/HEAD' as well?	12:41
Grum	and * master and remotes/origin/master it lists	12:42
	not that it explicitly lists Ilari	12:42
Ilari	Grum: I just guessed it would insert that prefix in order to refer to correct ref in case two refs with the same shorthand name existed...	12:43
Grum	i dont understand it, both are fresh clones o.O	12:43
	← error404notfound left	12:44
Grum	oh Ilari, on the one which looks ok, it does have origin/HEAD next to origin/master	12:45
dohzya	refs/remotes/origin <=> remotes/origin <=> origin	12:46
	as refs/heads/master <=> heads/master <=> master	12:46
Grum	yes but why do i get 2 different outputs?	12:47
	it cant just be the 1.6.2.1 vs 1.6.2.3 version? O.o	12:47
	→ FreakGuard joined	12:47
dohzya	it can if you have a branch named origin/XXX	12:47
Ilari	Grum: MD5sum the repo files from both repositories and compare the lists?	12:48
	→ mfilipe joined	12:48
Grum	dohzya: both are completely fresh clones	12:48
	→ leanucci joined	12:48
Grum	from the same machine, just on another machine o.O	12:48
dohzya	humm.... 1 or 2 repo ?	12:49
Grum	i have a repo somewhere; i clone it to two different machines	12:49
	and for some reason the output of 'git branch -a' is different on each of those machines	12:50
dohzya	is it possible the original repo changed between your 2 clones ?	12:50
Grum	no	12:51
	100% impossible	12:51
	→ faithfulgeek joined	12:51
	← seangrove left	12:51
	→ hiwd joined	12:52
	← voker57 left	12:52
	← daKmoR left	12:52
dohzya	so the 2 repo are clean, have the same .git/config, and are up-to-date.... hum.... interesting...	12:52
	← faithfulgeek left	12:53
Grum	well i have some ~/.gitconfig differences	12:53
	but nothing that could remotely influence it afaik	12:53
	i'll move it out of the way to make sure	12:53
	→ faithfulgeek joined	12:54
	→ R_Dday joined	12:55
	← [intra]lanman left	12:55
R_Dday	hello	12:55
dohzya	just to make sure : is there something wierd with "git branch -v" ?	12:55
	→ scarabx joined	12:55
R_Dday	I am doing this	12:56
	ERROR:gitosis.serve.main:Need SSH_ORIGINAL_COMMAND in environment.	12:56
	sorry	12:56
	but this	12:56
	git clone [email@hidden.address]	12:56
	→ cemerick joined	12:56
jast	try leaving out the "repositories/"	12:56
	← corni left	12:57
R_Dday	and I get this error	12:57
	ERROR:gitosis.serve.main:Repository read access denied	12:57
	fatal: The remote end hung up unexpectedly	12:57
	I am on Ubuntu ...do you know why ?	12:58
	← dedmajor left	12:58
Ilari	R_Dday: You don't have read permission to that repository.	12:58
jast	145654 < jast> try leaving out the "repositories/"	12:58
R_Dday	but its been set up already	12:58
	I have givven my pubblic key	12:58
Grum	dohzya / Ilari: http://git.pastebin.com/m2fc6d2ec	12:59
	branch -v gives 100% identical output	12:59
	→ cmarcelo joined	12:59
	→ qrush joined	12:59
Grum	R_Dday: have you given yourself read or write access to that repo?	12:59
R_Dday	How do I do that ?	13:00
	← faithfulgeek left	13:00
R_Dday	@Grum	13:00
Ilari	R_Dday: Edit the gitosis.conf, commit changes and push.	13:00
	→ trmanco joined	13:00
Grum	checkout the gitosis-admin repo, edit gitosis.conf; commit; push	13:00
	← girishr left	13:00
R_Dday	under which folder @Grum	13:01
Grum	R_Dday: you wouldnt have to ask that if you did a clone of the gitosis-admin repo :)	13:01
jast	seriously. if you use the repositories/ prefix in the clone URL it CAN'T WORK.	13:02
	→ dedmajor joined	13:02
Grum	jast: lies	13:02
dohzya	jast : why ?	13:02
jast	because I just tested it	13:02
Grum	you can have a repositories folder in your repositories dir :)	13:02
dohzya	jast: it's works for me	13:02
jast	perhaps newer gitosis versions check for this and fix it transparently	13:03
dohzya	mmmmh	13:03
jast	I get this: http://git.pastebin.com/d192db448	13:03
dohzya	I said nothing (I don't use gitosis)	13:03
Grum	the problem is really simply, E_NO_READ_ACCESS (as said in the original error: ERROR:gitosis.serve.main:Repository read access denied)	13:03
R_Dday	Well does the permission need to be set on my end ...or on the remote git server ? @Grum	13:03
jast	R_Dday: remote server. also, try removing the "repositories/" from the URL you use in clone.	13:04
Grum	R_Dday: who setup your gitosis? and omg you must twitter a lot to write @xxx the whole time :D	13:04
	jast: that will not solve anything	13:04
	it finds the repo and decides you have no access; that is the error	13:04
jast	Grum: please read my paste	13:04
dohzya	Grum: maybe your git config --global or --system is not the same	13:04
	→ g3d joined	13:04
	← g3d_ left	13:05
jast	notice how it says "read access denied" the first time and doesn't the second time	13:05
Grum	you are right; but that is strange though	13:05
wshimmy	Grum, R_Dday: gitosis does not give you access if you do not go through gitosis to get to the repo	13:05
jast	I guess it's a security thing	13:05
Grum	guess so	13:05
	→ catalis joined	13:05
jast	this way you can't get gitosis to tell you which repos exist and which don't	13:05
Grum	so if you'd have ~gitosisuser/repositories/repositories/repo.git you could do it :D	13:05
jast	probably.	13:06
Grum	wshimmy: if you have setup gitosis you do not have a choice but going through it	13:06
jast	unless you have a broken authorized_keys setup. we had that lots of times. in this case the gitosis setup seems to be fine, though.	13:06
wshimmy	Grum: that's my point. you cannot use the absolute path to the repository (which does not have the proper permissions set)	13:07
Grum	you can use subdirectories wshimmy	13:07
jast	wshimmy: it always goes through gitosis, no matter the path. problem is that gitosis auto-prepends the path to the repos, so the full result path is simply invalid	13:07
wshimmy	Grum: yes, from the gitosis root	13:07
	← hdoria left	13:07
Grum	no, from ~/repositories/*	13:07
wshimmy	Grum: sorry, bad phrasing. ~/repositories is what I was calling the gitosis root	13:08
Grum	kk =D	13:08
	← j416 left	13:08
Grum	R_Dday: does: git clone [email@hidden.address] work?	13:08
	→ markelikalderon joined	13:08
R_Dday	Grum: yes thax	13:09
Grum	thank jast ;)	13:09
R_Dday	I am having difficultie understunding the git concept	13:09
	I am beginner do you know any info on basics	13:10
	→ kukks joined	13:10
	→ jrmuizel joined	13:11
	→ mw___ joined	13:11
	← kalle_ left	13:11
	→ LiamH joined	13:12
	← hiwd left	13:12
	→ faithfulgeek joined	13:13
	→ kalle_ joined	13:13
	→ Sho_ joined	13:13
	→ Ymir joined	13:15
	← Ymir left	13:15
	← kbingham left	13:15
	→ priidu joined	13:17
	← solar_ant_ left	13:18
	← khmarbaise left	13:18
	→ esparkman joined	13:20
	← g3d left	13:20
	→ Ryback_ joined	13:20
	→ bcardarella joined	13:24
	→ khmarbaise joined	13:24
	→ jackdempsey joined	13:25
	→ davidad joined	13:28
	→ iratik joined	13:29
	→ solar_ant joined	13:32
	← jackdempsey left	13:32
	← baetis-fly left	13:33
	→ spearce joined	13:36
	← krh left	13:37
	← spearce left	13:37
	→ spearce joined	13:37
	→ arohner joined	13:37
	← ceyusa left	13:40
	← johan-s left	13:41
	→ khear joined	13:41
	← fujin_ left	13:42
	→ helo joined	13:42
	→ fujin joined	13:42
helo	is this document up to date for setting up a get repo for http access: http://www.kernel.org/pub/software/scm/git/docs/howto/setup-git-server-over-http.txt	13:42
	the dav part seems to not be working	13:43
	→ dodo_the_last joined	13:43
Ilari	helo: DAV push is best avoided if possible. The recomended way to push is via SSH.	13:44
	→ mattions joined	13:45
helo	hmmm	13:45
	so i'll have to give the people who push to the repo SSH accounts? :/	13:45
Ilari	helo: Or install gitosis.	13:46
xl0	helo: You can give them accounts with git-shell as the shell. They won't be able to do anything besides git operations.	13:47
dohzya	why not using the git protocol ?	13:47
xl0	dohzya: Push over git?	13:47
dohzya	yes	13:47
	→ dreamcat4 joined	13:47
xl0	Never seem anyone use it. How does it work?	13:48
	How does it authorize the user?	13:48
Ilari	xl0: It doesn't.	13:48
fujin	doesn't :P	13:48
dohzya	ok :)	13:48
fujin	that's kinda the idea of git over ssh :D	13:48
xl0	Then it's useless.	13:48
	← ehamberg left	13:48
	→ azumanga joined	13:49
	→ hpoydar joined	13:49
Ilari	xl0: Well, if you hate your users, you can do all kinds of trickery with hooks (push over git:// runs normal push hooks on remote side):	13:49
	← scarabx left	13:49
	→ sunblush joined	13:49
xl0	Ilari: You need not only hate the users, but also hate yourself, setting it up and maintaining.	13:49
	← solar_ant left	13:50
Ilari	xl0: Maintiaing such system (if done properly) is probably on same order of difficulty as maintaining gitosis (if the initial setup is good). Its however extremely nasty to clients...	13:51
dohzya	how works github ?	13:52
fujin	ssh	13:52
dreamcat4	hi	13:52
	when i do "git-checkout origin/branch file1 file2 file3"	13:52
	it pulls in all of the files of the branch called "branch"	13:53
fujin	-- file1 file2 file3	13:53
dreamcat4	thanks ill try that now	13:53
	← agile left	13:53
azumanga	I have a big git repository which contains multiple projects. In retrospect this was a mistake! Is there any way to split out one directory (with history) into a separate repository?	13:53
Grum	dreamcat4: you are using a really old git if you still use git-checkout (mind the dash!)	13:53
	azumanga: yes look into man git-filter-branch with subdirfilter	13:54
Gitbot	azumanga: the git-filter-branch manpage can be found at http://git.or.cz/man/git-filter-branch	13:54
dreamcat4	@fujin thank you that works perfectly	13:54
fujin	np	13:54
dreamcat4	@Grum what alternative command would i use instead ?	13:55
Grum	just git checkout without the - :)	13:55
dreamcat4	(should i use?)	13:55
Grum	if that doesnt work, you have a way to old version of git (like 1.4 orso?)	13:55
dreamcat4	i have macports git latest which supports both command syntax	13:56
	→ km2 joined	13:56
dreamcat4	@Grum if you hyphenate the command, then its fairly trivial to "man git-last-command"	13:56
Grum	you should use the 'without dash' -- the files with a dash shouldnt be in your path :)	13:56
dreamcat4	huh ??	13:56
Grum	yes, or just git command --help or git help command :)	13:56
	(both alternatives give you the manpage)	13:57
	→ justatheory joined	13:57
dreamcat4	@Grum well thats fine but i prefer the old way ;)	13:57
Grum	dreamcat4: it will break in the future so you better adjust while you can =D	13:58
dreamcat4	@Grum oh right	13:58
Grum	the libexec/git-core directory shouldn't be in the path	13:58
	→ DavidKlein joined	13:58
Grum	only git (and there are some exceptions that are also in your path for backwards compat)	13:58
azumanga	Grum: Thanks, reading	13:58
	→ alester joined	13:59
dreamcat4	@Grum this is actually very interesting to me because i'm about to write a commandline tool which mimics the git syntax	13:59
	← gitte left	13:59
Grum	the exceptions are (on my box) git-cvsserver git-receive-pack git-shell git-upload-archive git-upload-pack	14:00
	← tatsuyao_ left	14:00
dreamcat4	@Grum i was going to support "gitnest command" initially, then both later	14:00
	→ netoman joined	14:00
dreamcat4	@Grum i was not going to go near those exception command	14:01
	@Grum but thanks for looking	14:01
	← azumanga left	14:01
Grum	you can make your own 'git command' by putting a git-command in the libexec/git-core folder :)	14:01
Ilari	Those "exceptions" are all, except for git-shell, subprotocol servers. Git-shell is there because it can't be invoked properly otherwise.	14:02
Grum	yup	14:02
	→ schambers joined	14:03
	→ tatsuyao joined	14:03
dreamcat4	@Grum i guess optionally create a symlink to libexec/git-core, if its a unique command name	14:03
	→ kevwil joined	14:04
	← gdelfino left	14:04
	← bratsche__ left	14:04
dreamcat4	@Grum although in my implementation its "gitnest add" "gitnest update" etc	14:04
dohzya	I think I would better to create a alias than a symlink	14:04
	→ bratsche__ joined	14:04
Grum	yup	14:04
dreamcat4	@Grum if only there was a sensible way to 'install' aliases in bash	14:05
	in a cross-platform compatible way	14:06
Grum	depends how cross platform	14:06
	← schambers left	14:06
Grum	windows? :D	14:06
dreamcat4	haha	14:06
Grum	but dohzya was talking about git config aliases	14:07
dohzya	mmmmh	14:07
	no :p	14:07
dreamcat4	do you mean cygwin of the microsoft shell tools	14:07
Grum	not?	14:07
	SLACK!	14:07
	man git-config	14:07
Gitbot	Grum: the git-config manpage can be found at http://git.or.cz/man/git-config	14:07
dreamcat4	oh	14:07
	← sikanda left	14:07
dohzya	no no	14:07
	alias can't be dashed	14:07
Grum	'be dashed' ?	14:07
dohzya	hum..... I think I'm lost :)	14:08
	→ ckm joined	14:08
dreamcat4	alias.*	14:08
	Command aliases for the git(1) command wrapper - e.g. after	14:08
	defining "alias.last = cat-file commit HEAD", the invocation "git	14:08
	confusion and troubles with script usage, aliases that hide	14:08
	If the alias expansion is prefixed with an exclamation point, it	14:08
	"alias.new = !gitk --all --not ORIG_HEAD", the invocation "git new"	14:08
	← lorandi left	14:08
	→ lorandi joined	14:09
	← Kimmie left	14:09
	→ cilly joined	14:10
	← ckm left	14:10
dohzya	I thought you wanted to create a dashed version of git commands :p	14:10
dreamcat4	@dohzya if you mean git-dash-command they dont need to be anymore as dashed will go away	14:10
Grum	actually it should already be away	14:11
dohzya	yes but I thought you want to create a dashed version, because you prefer it	14:11
	(that is a very respectable reason)	14:11
	← kalle_ left	14:11
Grum	bad form ;)	14:12
	→ kalle_ joined	14:12
dreamcat4	@dohzya yes i was interested, but i should made my git addon it follow the future git convention as pointed to by @Grum	14:12
	← lorandi left	14:12
dohzya	yes	14:12
dreamcat4	(should follow)	14:12
dohzya	but you still can create bash alias for dasher them :p (alias git-add='git add' for instance)	14:13
dreamcat4	@Grum i have macports git	14:13
	git-core @1.6.2.4 devel/git-core	14:13
	it seems to have the dashes still	14:13
Grum	if you do: git-[tab] you get a giant list?	14:13
	+ic	14:13
dreamcat4	yes	14:13
	→ jackdempsey joined	14:13
Grum	and in your $PATH there is nothing resembling git-core? :)	14:14
dreamcat4	oh yeah	14:14
	:/Users/id/.gem/ruby/1.8/bin:/opt/local/libexec/git-core:/Users/id/Library/bin:/opt/local/bin:/opt/local/sbin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/X11/bin	14:14
dohzya	dreamcat4: macport install in /usr/local/bin, maybe you have yet an installation in /usr/bin	14:14
Grum	you added that manually ;)	14:14
	remove it =D	14:14
	→ sikanda joined	14:14
dreamcat4	its there, maybe macports added it? i dont remember doing so	14:15
Grum	could be, depending on how long you use git hehe	14:15
dreamcat4	hey i don't want to though !	14:15
Grum	why not?	14:15
dreamcat4	dashes a fun, i like them :)	14:15
	← arohner left	14:15
Grum	it will break if the structure of git changes	14:15
dreamcat4	i mean, just for my personal command-line fun	14:15
	→ loincloth joined	14:15
Grum	you can still do all the things without dashes?	14:16
dreamcat4	i know im lazy	14:16
Grum	if you do: git xxx, it will just find git-xxx in libexec/git-core and execute it :D	14:16
	you know generally people find typing spaces easier than dashes ;)	14:16
	→ hummingkid joined	14:16
Grum	(and again -- it will break in the future)	14:16
dreamcat4	well i won't promote my misbehaviour if you say nothing either	14:17
	lets keep it our little secret	14:17
	;)	14:17
Grum	i dont care hehe, you will have to rewrite all of your scripts and cant work with git on a box with a sane setup ;D	14:17
dreamcat4	(laughs)	14:17
	yep	14:18
	(more laughing)	14:18
	have to go now	14:18
	bye	14:18
	← dreamcat4 left	14:18
	bratsche__ → bratsche	14:19
	← resmo left	14:20
	→ cydork joined	14:21
	← tim_jenssen left	14:21
	→ resmo joined	14:21
	→ altrux joined	14:21
	→ schambers joined	14:23
schambers	how can i push changes to ALL git repositories?	14:23
	→ arohner joined	14:23
doener	I don't want your stuff in my repo!	14:23
	← arohner left	14:23
doener	:-p	14:23
schambers	heh	14:23
	→ lorandi joined	14:24
dohzya	for i in $(git remote) ; do git push ; done	14:24
	?	14:24
doener	so you'll have to tell us at least what "ALL" means	14:24
schambers	i have 2 remote repositories defined	14:24
	in my config for a local repos	14:24
	ones at unfuddle, another at heroku	14:24
	i want to do a git push to both at the same time	14:25
	is that possible?	14:25
	← ia left	14:25
dohzya	maybe "git remote \| xargs git push" (I don't master the xargs command :/)	14:25
	← meyering left	14:26
schambers	so it would probably just be easier to push one at a time?	14:26
dohzya	yes	14:26
PerlJam	dohzya: that looks find to me.	14:26
	er, fine	14:26
schambers	ty	14:26
	→ tim_jenssen joined	14:26
dohzya	schambers: the good news is that it's easily aliasable	14:26
schambers	hmm?	14:27
	→ ia joined	14:27
dohzya	git config --global alias.push-all "!git remote \| xargs git push"	14:27
	so you will just need to use your new fresh git push-all command	14:28
	→ wmoxam joined	14:28
schambers	ah	14:28
	how can i make it so it pushes to master by defualt?	14:28
	→ kuadrosx joined	14:28
schambers	ie: git push unfuddle	14:28
	actually does git push unfuddle master	14:28
	← juanjoc left	14:29
dohzya	I think "for i in $(git remote) ; do git push $i master; done"	14:29
	← jrmuizel left	14:29
doener	or set remote.<name>.push to "master"	14:30
dohzya	maybe create a function which test if there is a refspec given, and put master if not	14:30
	hmmm.... too easy...	14:30
	→ meyering joined	14:30
	→ mithro joined	14:30
schambers	doener: where is that set?	14:30
	in config?	14:30
	← aspotashev_afk left	14:31
doener	yep	14:31
	→ jrmuizel joined	14:31
doener	e.g. "git config remote.origin.push master"	14:31
	→ hyperair joined	14:31
doener	the default is ":", which updates all branches that exist locally and in the remote repo	14:31
	→ PerlPilot joined	14:33
	← aziz left	14:34
	→ bdiego joined	14:34
	→ RaySl_ joined	14:34
	→ agile joined	14:35
	← hyperair left	14:37
	→ hyperair joined	14:38
	← kevwil left	14:38
	← bobmcw left	14:38
	→ RaySl__ joined	14:40
	← Strogg left	14:40
	← RaySl left	14:40
	→ Strogg joined	14:41
	← sikanda left	14:42
	← cilly left	14:44
	← jrmuizel left	14:45
	← faithfulgeek left	14:45
	→ chris2 joined	14:46
	← hyperair left	14:46
	→ hyperair joined	14:47
soul9	hello all	14:48
	→ benn_d joined	14:48
soul9	does anyone know, if i do a cherry picking of commits from one branch to an other, and maybe a patch is already half-applied, but the merge detects it correctly and applies the remaining part of the patch (commit), is the commitid still left unchanged?	14:49
doener	soul9: the sha1? That is _never_ unchanged from the commit you cherry-pick	14:50
	← disappearedng left	14:50
doener	(unless you manage to cause a sha-1 collision)	14:50
soul9	ok	14:50
	heheh	14:50
	→ sikanda joined	14:52
	← Ingmar left	14:52
doener	for the commit object hash, the changes don't matter at all, only the end result, as the commit references a tree, which in turn references blobs and other trees	14:52
	all diff stuff is generated on the fly	14:52
	→ Ingmar joined	14:52
doener	and for the cherry-pick stuff, besides other things, at least the parent commit will also be different, thus the commit id changes	14:53
	the only id that might be unchanged is the patch id, but that would also change in the scenario you described	14:53
	→ mbuf joined	14:53
	→ bobmcw joined	14:54
	← fujin left	14:55
mbuf	other than doing cat .git/logs/HEAD .git/refs/heads/master, how can I tell where the current HEAD or master is pointing to?	14:55
doener	git rev-parse HEAD master, git show HEAD master, git log HEAD master, ...	14:56
mbuf	doener: thanks	14:56
	← hyperair left	14:57
	← cedricv left	14:59
	→ Kimmie joined	14:59
	→ woei joined	14:59
	← RaySl_ left	14:59
	→ hyperair joined	15:01
	→ cilly joined	15:01
	→ willb joined	15:02
	← sikanda left	15:04
	→ javatexan joined	15:05
R_Dday	hello	15:05
	I am doing this	15:05
	→ fujin joined	15:06
leo2007	after knowing how to use rebase, it seems very hard to live without it	15:06
R_Dday	git checkout -t origin/master	15:06
	in order to swich branch	15:06
	but I get this	15:06
	fatal: git checkout: --track and --no-track require -b	15:06
	→ reprore joined	15:06
jast	that only works in git 1.6.1 and newer	15:07
	in older versions you'll have to use: git checkout -b master origin/master (though don't you already have a master branch?)	15:07
wereHamster	R_Dday: origin/master is a 'remote tracking branch'. You should not work on that directly	15:07
jast	wereHamster: look up what -t does in 1.6.1 ;)	15:08
doener	wereHamster: with the new -t shortcut, that's the same as "git checkout --track -b master origin/master"	15:08
	→ sikanda joined	15:09
doener	R_Dday: for just switching branches, that seems wrong though. You usually already have a branch head called "master", to just switch to that you'd just use "git checkout master"	15:09
	→ lois[pac] joined	15:09
	→ psankar joined	15:09
	→ harinath joined	15:10
	← kalle_ left	15:11
	→ kalle_ joined	15:12
	← harinath left	15:12
	← gilimanjaro left	15:12
	→ faithfulgeek joined	15:14
	← loincloth left	15:15
	→ raydeo joined	15:15
	→ seangrove joined	15:16
	→ sekimura_ joined	15:19
	→ aaron1 joined	15:20
	← sekimura left	15:20
aaron1	hello, I have a question about git on windows	15:21
dsop	don't ask to ask, just ask	15:21
aaron1	k	15:21
	when I clone a repo, I get files marked as modified	15:21
	← cilly left	15:21
aaron1	I am aware of the cr-lf issue	15:22
	but the file marked modified has cr-lf eols	15:22
	on the repo	15:22
doener	and that's the problem	15:22
Grum	you get all files marked as modified?	15:22
aaron1	no, just a handful	15:22
	there seems to be no difference between files marked modified and files not modified	15:23
doener	the file in the filesystem has crlf, gets converted to lf and compared to the repo contents which have crlf -> whole file differs	15:23
aaron1	in terms of eol	15:23
	← deavid left	15:23
	→ voker57 joined	15:23
Grum	doener: doesn't core.autocrlf true fix that?	15:23
doener	aaron1: not in the working tree, but in the repo	15:23
	→ aisbaa joined	15:23
doener	Grum: no, that breaks it	15:23
aaron1	no it doesn't	15:23
	I tried various settings for core.autocrlf	15:23
	→ deavid joined	15:23
aaron1	in gitconfig	15:23
doener	Grum: core.autocrlf true assumes that the repo holds files with just LF	15:23
	aaron1: set it to false, then "git reset --hard"	15:24
aaron1	ok	15:24
	but, I tried setting it to false before clone, and same problem	15:24
Grum	well you want a checkout with 'LF' then commit that, and then use core.autocrlf right?	15:24
aaron1	all files on repo seem to use cr lf as EOL	15:24
doener	aaron1: how did you check?	15:25
aaron1	hexedit	15:25
doener	aaron1: using "git show"/"git cat-file"?	15:25
Grum	euh CR is oldskool macs (pre 10.x)	15:25
	← tango_ left	15:25
doener	Grum: well, that you can achieve by having autocrlf enabled and then just doing "git reset --hard; git add -u; git commit -m Convert\ to\ LF"	15:26
	← tim_jenssen left	15:26
doener	Grum: but that adds a pretty annoying EOL-change commit	15:26
	→ jrmuizel joined	15:26
	→ loincloth joined	15:26
doener	Grum: no fun when you get to merge some LF with a CRLF version	15:26
Grum	doener: beats having 'half with/half without' by miles :D	15:26
aaron1	why are only some files marked modified	15:26
	if all files seem to have CRLF	15:27
	I haven't checked all of them	15:27
	← mbuf left	15:27
	← Tricon left	15:27
Grum	the checkout will be consistent aaron1, the files in the repo arent	15:27
doener	aaron1: again, what did you check? The files in the working tree? Of course they are all CRLF, autocrlf makes them so	15:27
	→ benn_d_ joined	15:27
aaron1	more info: this was on my home system yesterday, on windows	15:27
doener	aaron1: only checking the stuff that "git show" or "git cat-file" shows is a valid test	15:28
aaron1	today, at work, on Ubuntu, I did hexedit on checked out files	15:28
	and saw CRLF	15:28
	doener: thanks, I will try that	15:28
doener	then that files probably actually has CRLF EOLs (unless you enabled autocrlf on that box, too)	15:28
	and that causes the "modified" behaviour	15:29
aaron1	is CRLF on or off on unix?	15:29
doener	core.autocrlf you mean? Off by default. Only msysgit enables that by default	15:29
Grum	doener: why git add -u and not just commit -am ?	15:29
doener	Grum: I just don't do "commit -a" ;-) Good old finger memory	15:30
Grum	=D	15:30
	→ Zenopus_ joined	15:30
doener	Grum: and I prefer a mix of CRLF and LF files when the alternative means that I have to deal with a shitload of useless merge conflicts	15:31
	→ greyface joined	15:31
Grum	well you can just turn all into LF and have the people on windows use core.autocrlf true right?	15:31
	← netoman left	15:32
Grum	just punish people for putting CRLF in your repo =D	15:32
	doener mumbles something about git-svn	15:32
Grum	svnequalsfail	15:32
aaron1	so, does this make sense that of two files that seem to use the same EOL convention, only one is marked modified?	15:32
Grum	doener: simplest thing would be to just ignore the existence of windows users :D	15:33
	← gaveen left	15:33
aaron1	lol	15:33
doener	aaron1: if both of them use the same eol-style _in_the_repo_, then no	15:33
	← dwt left	15:33
doener	aaron1: if you just look at the working tree, that's useless information if core.autocrlf is turned on	15:34
aaron1	my hexedit was done on a unix system	15:34
alester	So much discussion on sha-1	15:34
	and nobody cares about my poor little patch. Oh woe!	15:34
aaron1	so, that info should be correct, yes?	15:35
	so, only about 10 files are marked modified	15:35
	a quick fix would be to add and check them in, I suppose	15:36
	→ nabilelisa joined	15:36
Grum	alester: what did you patch? :D	15:36
doener	aaron1: not if you have core.autocrlf turned on. Just because the *nix versions usually don't turn that on by default doesn't mean that you cannot have turned it on yourself, or maybe you're using a package with a retarded maintainer or whatever	15:36
alester	Oh, heck, it was just some refactoring of redundnat code.	15:36
	nothing big.	15:36
	I just don't know what the git dev cycle is like.	15:37
	← fujin left	15:37
	← reprore left	15:37
	→ whaley joined	15:37
nabilelisa	hi all - can anybody talk to me about git server options? i'd like to be able to setup something like github for private use within my office - i understand that's not the case but maybe gitosis is sufficient? if anybody has any experience in those i'd appreciate some advice	15:37
	→ reprore_ joined	15:38
	← rraasch left	15:38
	→ drewolson joined	15:38
Grum	ask your questions nabilelisa :)	15:38
doener	alester: I don't see any patches from you on the mailing list	15:38
Grum	and yes you can use gitosis and gitweb to make something usable quite easily	15:38
nabilelisa	Grum: well basically i installed gitosis and was only able to read the repo - not write. is that a limitation or was i doing it wrong?	15:38
	→ gaveen joined	15:38
alester	here's an example	15:38
	http://n2.nabble.com/-PATCH-1-2--Localizing-temporary-variables-td2674500.html	15:38
leo2007	I just clone a git repo that has a .git directory > 40 M	15:39
	← javatexan left	15:39
Grum	nabilelisa: that is because you didn't tell gitosis the repo was writable for the user you were authenticating with	15:39
leo2007	how can I cut down its size. I am only interested in it latest 100 or so commits	15:39
nabilelisa	Grum: ok, so then if i spend more time i can probably find the appropriate config file. thanks. do you know what the difference is between using the git:// protocol vs. the http://?	15:39
	← voker57_ left	15:40
aaron1	I have another question about submodules: my user name on the remote repo differs from my login, so I clone a repo using the following: git clone user@gitrepo/project	15:40
	then I run git submodule init, and all seems fine	15:40
Grum	nabilelisa: git:// is smart and faster than http://	15:40
aaron1	but when I run git submodule update, it asks me for the ssh password for the login user	15:41
	→ javatexan joined	15:41
aaron1	not the user I used to clone	15:41
Grum	nabilelisa: if you are using gitosis you normally use ssh for authentication	15:41
Ilari	nabilelisa: Also both can do fetch and push, but push with either is bad idea (for git:// its insecure, and too hard to set up for http://).	15:41
Grum	which is why people usually use gitosis, as it provides a 'secure' way to push with some basic ACLs	15:42
doener	alester: ah, had been looking for your surname	15:42
nabilelisa	Grum: ok, aaaah, ok cool	15:42
alester	that's what I searched on.	15:42
	← benn_d left	15:42
nabilelisa	Grum: so if i use gitosis, then i get ssh and git:// then it should be safe/secure?	15:42
doener	alester: I search for "From" in mutt ;-)	15:42
Grum	nabilelisa: we've set it up that people just push/pull from git:project	15:42
	→ loinclot_ joined	15:42
doener	s/search/searched/	15:42
Grum	nabilelisa: gitosis by default provides read/write via ssh (that is the reason of it existing)	15:43
	you can give anonymous read access through git:// but why would you do that? :D	15:43
nabilelisa	Grum: ok great. no im not interested in annoymous read :)	15:43
	Grum: so is it enough to just install and configure gitosis? or do you recommend that i do anything else too?	15:44
Ilari	nabilelisa: Also, git daemon is for git://, so if you don't need it, you don't need to run git daemon.	15:44
Grum	nabilelisa: while using ssh you can severly shorten the 'clone url' by making a simple .ssh/config entry like: Host git\n\tHostname host.of.gitosis.box\n\tUser gitosis-user	15:44
nabilelisa	Ilari: thanks - i thought so but it's good to have confirmation :)	15:45
Grum	nabilelisa: hence we clone from git:project now (lovable short urls) which is the same as cloning from: git@gitbox:project.git or ssh://git@gitbox/project.git	15:45
nabilelisa	Grum: thats very interesting, is there a page i can read up on that? does it have a name?	15:46
Grum	nabilelisa: that is just default ssh configuration	15:46
nabilelisa	oh right	15:46
	→ marvil07 joined	15:46
	→ tango_ joined	15:46
nabilelisa	Grum: so basically i should just go ahead and use gitosis - the defaults sound like they are pretty much close to what i need	15:46
Grum	gitosis only provides multiuser accesses through a single user using ssh authenticated by public/private keys and adding a basic ACL	15:47
	err multiuser access ;)	15:47
nabilelisa	Grum: yes, i noticed the single ssh user - but my commits will still have my name on them, right? because ive configured git on my local machine to know my email/name?	15:47
Grum	yes	15:48
	and gitosis knows who you are because you authenticate with your private key	15:48
	against your public key in the gitosis repo	15:48
nabilelisa	Grum: i see	15:48
Grum	so it can deny access for certain repos if you please	15:48
nabilelisa	Grum: very nice. i think i'll go and give it a try now. thanks for all your help!	15:49
Grum	yw :)	15:49
nabilelisa	Ilari: thanks for your help - much appreciated!	15:49
Grum	there are some nice guides for gitosis out there btw, first hit on google works like a charm (english that is ^^)	15:49
Ilari	nabilelisa: One problem with gitosis is that one can't log pushes without modifying gitosis (one line change).	15:49
	→ ceej joined	15:50
	← Zenopus left	15:50
	Zenopus_ → Zenopus	15:50
nabilelisa	Ilari: hrm, what line is that? :P	15:50
Grum	gitosis is quite the raw product, it was made to suit a single person's wishes	15:50
	← benn_d_ left	15:50
	← deavid left	15:50
	→ benn_d joined	15:51
nabilelisa	Grum: i get that feeling too - do you know of anything more refined? something as slick as github? :P	15:51
Ilari	nabilelisa: There's place (in serve.py?) where user's name is in variable 'user'. You need to export that as environment variable.	15:51
	nabilelisa: There's gitorious. I don't know much about it.	15:51
	← altrux left	15:52
Grum	nabilelisa: just use gitweb next to it :)	15:52
	point it at the repo folder and pew!	15:52
	← bratsche left	15:52
Ilari	nabilelisa: And gitosis is not that complicated. One could quickly do very crude replacement (without remote admin features) if one understands the construction principles involved.	15:53
	→ bratsche joined	15:53
	→ reprore__ joined	15:53
	← reprore_ left	15:53
Grum	doener: do you have any idea why the output of git branch -a is different in 2 freshly cloned repos ? (see: http://git.pastebin.com/m2fc6d2ec )	15:53
emias	In most cases, a Unix user per Git user and SSH access (plus gitweb) should do just fine. No idea why everybody seems to be stingy with creating Unix users these days (just to re-implement user management on top of a single Unix user afterwards).	15:54
	← kadoban left	15:54
	← chalky left	15:54
	loinclot_ → loincloth_	15:54
	→ deavid joined	15:54
Ilari	emias: Gitosis can manage more flexible ACLs. There are POSIX ACLs, but they have usability problems...	15:55
Grum	doener: the git version is 1.6.2.1 vs 1.6.2.3 .. but that cant be giving such odd results? there are no global or system configs..	15:55
emias	Ilari: Well, for our site, managing access through plain Unix groups is enough and I would've assumed that this is true for most (though admittedly not all) sites.	15:57
Ilari	emias: Especially changing permissions implemented using POSIX ACLs... Requires owner override permissions (root).	15:57
	← alanhaggai left	15:57
	← loincloth left	15:58
Grum	managing access through unix groups is a pain and severe overkill if you just want a git repo :)	15:58
	mostly because you have to add a user to the system if you jsut want to give someone access	15:58
	instead of simply adding a keydir/$user.key file :)	15:59
	sunblush → sunblush\|away	15:59
	← drewolson left	15:59
	← psankar left	16:00
	→ MisterN joined	16:00
	→ juanjoc joined	16:00
straszheim	Grum: which setup just uses keydir/$user.key ?	16:02
doener	Grum: 209d336ae3923f2c1783a42b12fca50f3bee0802	16:02
Gitbot	[git 209d336ae]: http://tinyurl.com/cw9kcu -- builtin-branch: improve output when displaying remote branches	16:02
doener	Grum: that's the responsible commit, but that's in 1.6.3-rc0, not 1.6.2.3	16:02
Grum	straszheim: gitosis	16:03
	→ rylar joined	16:03
Grum	doener: hmmm so why on earth am i having this problem?	16:03
	→ romao joined	16:03
	→ patrikf joined	16:03
	→ cilly joined	16:04
Grum	and the 1.6.2.3 is doing it 'normally' from my PoV	16:04
	← Kimmie left	16:04
doener	Grum: in the pastebin, the first one shows the new behaviour	16:04
	Grum: are you 100% sure that that box isn't running something newer?	16:05
	← aaron1 left	16:07
	← sikanda left	16:07
straszheim	hey, that gitosis looks nice and clean. good tip. thanks.	16:08
bremner	any gotchas for always using pull --rebase?	16:09
	→ jceb joined	16:10
Ilari	bremner: If the branch is pushed elsewhere, that's bad thing to do.	16:10
	→ dwave_ joined	16:10
Ilari	bremner: Also, if branch has merges since last pull, that's also bad idea.	16:11
	← kalle_ left	16:11
	→ kalle_ joined	16:12
	← naeu left	16:12
bremner	hmm, ok the merges since last pull I didn't think about. thanks	16:12
	→ corni joined	16:12
	← faithfulgeek left	16:12
	loincloth_ → loincloth	16:12
	← catalis left	16:14
	← bobmcw left	16:14
	→ zpoley joined	16:15
	← yec left	16:16
	→ ceyusa joined	16:17
	← jayallen-afk left	16:17
	→ Alinermedia joined	16:20
	neoeinstein_ → neoeinstein	16:21
	← khmarbaise left	16:21
	→ leanucci_ joined	16:21
	← mattions left	16:23
	→ WALoeIII joined	16:23
	→ faithfulgeek joined	16:24
	← airtonix left	16:25
	← cilly left	16:26
	→ argonel joined	16:26
	← dwave_ left	16:27
	→ jnareb joined	16:27
	← dwave left	16:27
jnareb	Hello	16:28
jacobat	Hi	16:28
jnareb	Was 'Interactive graph GUI' project announced somewhere beside GSoC page, like git mailing list or SoC2009Ideas wiki page?	16:29
	http://socghop.appspot.com/student_project/show/google/gsoc2009/git/t124022376285	16:29
	← jceb left	16:29
	← shruggar left	16:31
	→ ashleyw joined	16:32
	← nabilelisa left	16:32
	→ jceb joined	16:34
	← bgerber left	16:34
wereHamster	jnareb: My understanding is this: It shows you the graph and you can interactively reorder commits, change commit messages etc and it will issue rebase -i or whatever is needed in the background	16:34
	← tal67 left	16:35
jnareb	I was not asking what is it about. I was asking whether the idea was presented to the public before being accepted as GSoC 2009 project.	16:35
	where Git Wiki and git mailing list counts as 'public'	16:36
	#git IRC channel or #gsoc IRC channel also can count as public because they are archived	16:36
	sunblush\|away → sunblush	16:36
	→ bobmcw joined	16:37
	wereHamster fails at reading, is wondering why	16:38
	← aisbaa left	16:38
	→ qhoxie joined	16:38
alester	I think this weekend I"m going to take the plunge and start running git past splint.	16:38
	← jmesnil left	16:39
	→ jmesnil joined	16:40
	← leanucci left	16:40
jnareb	wereHamster: by the way, if I remember correctly in old DOS XTree file manager you were able to move and delete directories in a tree view	16:41
	→ jceb81 joined	16:41
	→ simosx joined	16:41
	→ elvedin joined	16:42
	← whaley left	16:43
	→ alanhaggai joined	16:43
	→ jzellman joined	16:43
elvedin	How do I do username expansion (pushing to remote site - ssh://$USER@site.hostname/dir/path ) if I'm sharing my local repo with other developers?	16:44
	→ dwave joined	16:44
	← reprore__ left	16:44
	← alanhaggai left	16:44
ThaDon	was there a port of midnight commander to DOS? i seem to recall using it	16:44
	→ reprore joined	16:44
	← sunblush left	16:45
Ilari	elvedin: 'ssh://USER@site/~username/path/' or something like that?	16:45
jnareb	ThaDon: No, XTree was not NC-like file manager; it was "Explorer like" (but it was earlier than Explorer) which operated on tree view of directories + view of files in a directory	16:45
elvedin	Ilari: yeah	16:46
	We want to user the $USER shell variable to determine which user to SSH as	16:46
jnareb	elvedin: usually it makes more sense to use key-based authentication for git+ssh	16:46
elvedin	Yep, but we can't do that	16:46
jnareb	elvedin: currecntly you cannot	16:47
	elvedin: (at least not for shortcut names). remote.<remotename>.url is not explanded WRT environmental variables and similar stuff	16:47
Ilari	elvedin: Doesn't ssh by default use current user id as remote user id?	16:47
	← sitaram left	16:48
jzellman	Hi, I'm having a problem when i do subsequent git pulls on a clean master it tells me that my branch is ahead of master by an ungodly amount of commits. http://gist.github.com/101204 Anyone see this before?	16:48
	← priidu left	16:48
jnareb	elvedin: athough there was talk about adding '--path' type to git-config, which would do envvar expansion (and tilde expansion)	16:48
elvedin	Ilari: It does, but we sudo into a common development user	16:48
	→ hobodave joined	16:48
	→ altrux joined	16:49
jnareb	elvedin: and you cannot use key-based authentication (ssh-add/ssh-agent and perhaps keychain) authentication why?	16:49
elvedin	jnareb: Security issues	16:49
	← reprore left	16:49
	→ reprore joined	16:50
elvedin	We try not to subvert our sysadmin/security overlords too much as they'll completely lock us out if we do this	16:50
	← leanucci_ left	16:50
	→ leanucci joined	16:50
leo2007	any possibility to reduce the size of .git dir?	16:51
	→ sitaram joined	16:51
elvedin	What about the GIT_SSH variable?	16:51
jnareb	leo2007: "git gc"	16:51
leo2007	i run that and it is still 41M	16:51
	the rest of the project files combined is < 2 M	16:52
jnareb	leo2007: you might need to expire reflogs	16:52
Ilari	leo2007: 'git repack -a -d -f --window=250 --depth=250', but thats VERY slow.	16:52
jzellman	will git reset orgin/master work for this?	16:52
	→ octagon joined	16:52
jnareb	leo2007: and/or use "git prune" (assuming that there is not other process that access this repository)	16:52
	← giallu left	16:52
Ilari	jzellman: Try 'git fetch origin'?	16:53
leo2007	no, basically I am cloning a repo so that I can use it but I'm not developing it.	16:53
Ilari	jzellman: Ah, it already ran it.	16:53
	← devicenull_ left	16:54
jnareb	elvedin: you don't need admin rights for key-based authentication for SSH. only public key on remote host, and private key added to ssh-agent on local host, IIRC	16:54
	← jceb left	16:54
jnareb	leo2007: "git clone --depth=1" if you don't need history	16:54
	leo2007: .git contains full history of a project	16:54
	leo2007: (which might include large files currently deleted)	16:55
	→ schlort joined	16:55
jnareb	What nick Dscho uses here?	16:55
	← rylar left	16:55
leo2007	nice let me try it out	16:55
	→ octagon_ joined	16:56
	→ AkumaStreak joined	16:56
	→ a-atwood joined	16:56
	← octagon left	16:57
jnareb	About 'foreign' git projects at GSoC 2009: did anyone find something more than:	16:57
	* #git (Mono project)	16:57
	* git backend, Mercurial API/CLI (Mercurial)	16:57
	* git backend for Build Service (openSUSE)	16:58
	?	16:58
leo2007	jnareb: reduce the size a bit, from 41M -> 33M	16:58
	how to expire anything before 2009-02-28	16:58
jnareb	leo2007: what did reduce size?	16:58
sitaram	leo2007: is this a fresh clone?	16:59
leo2007	git clone --depth=1 did	16:59
	yes	16:59
sitaram	if so, you shouldn't have any reflogs...	16:59
leo2007	I tried 'git clone --depth=1 git://repo.or.cz/org-mode.git'	16:59
sitaram	so when you say expire before 2009-02-28, it doesn't help if you already did a depth=1 clone,	16:59
Ilari	jnareb: Someone on mailing list slammed Mono for picking that project...	17:00
sitaram	nice :)	17:00
	sitaram is way behind on reading the list	17:00
sitaram	leo2007: maybe you want to find out what file is taking all that space?	17:00
leo2007	it is .pack file under .git/objects/pack	17:01
	← schambers left	17:01
jzellman	Ilari: git fetch origin did not work	17:01
sitaram	leo2007: I didn't mean that... I meant what file in the repo	17:02
Ilari	jzellman: It still displays ungodly amount of local commits?	17:02
jzellman	llari: this all began when my coworker ran git push --mirror origin which totally f'ed things up.	17:02
	llari: yeah it does	17:02
leo2007	sitaram: files in the repo are relatively small, all of them combined is less than 2 M	17:02
Ilari	jzellman: And "didn't work" means "didn't chnage things" or "failed with error"?	17:02
jzellman	llari: sorry, didnt change things	17:03
Ilari	leo2007: Did last commit remove those, IIRC, depth 1 really gets 2 commits.	17:03
	sdiz → sdiz-afk	17:03
Ilari	leo2007: Also, other branches?	17:03
sitaram	leo2007: then you have many many files perhaps?	17:03
Ilari	jzellman: Maybe that 322 commits is accurate? Maybe try 'git push origin master'?	17:03
jzellman	llari: new paste: http://gist.github.com/101211	17:03
sitaram	leo2007: in the top of the work tree: 'du -sk . .git'? can you tell us what this says?	17:04
	→ sikanda joined	17:04
	sdiz-afk → sdiz	17:04
leo2007	sitaram: 41M.git	17:04
	cpg\|away → cpg	17:04
jzellman	what about git reset origin/master? http://colabti.org/irclogger/irclogger_log/git?date=2009-04-22#l1228	17:05
	→ d0k_ joined	17:05
	← jceb81 left	17:05
leo2007	the project has less than 100 files	17:05
sitaram	leo2007: and "."? (i.e., "du -sk ."?)	17:05
	→ krawek joined	17:06
	→ btanaka joined	17:06
leo2007	sitaram: 45M	17:07
Ilari	jzellman: I think that wouldn't be a good idea. Unless you want data loss.	17:07
sitaram	ouch	17:07
Ilari	leo2007: How many branches 'git branch -r' lists?	17:08
	→ benn_d_ joined	17:08
leo2007	Ilari: 7	17:08
Ilari	jzellman: That --mirror might have destroyed commits from upstream repo, and part of those 322 commits might be those destroyed commits.	17:08
	← benn_d left	17:09
leo2007	is everything normal?	17:10
	I thought git is usually more efficient than svn	17:10
	← Sho_ left	17:11
Ilari	jzellman: pull not grabbing anything and your branch being 322 commits ahead would indicate that you have history that continues from where the github repo leaves off...	17:11
jzellman	llari: ok thanks, yeah that's what i was thinking too	17:11
Grum	doener: git version 1.6.2.3.1.gfb039 ... that is what is running on the remote box, no idea what can be causing this though :)	17:11
Ilari	jzellman: Maybe try to push it back: 'git push origin master'?	17:11
	→ Sho_ joined	17:11
	← kalle_ left	17:11
Ilari	Grum: Ah, that can be pretty much anything... :-)	17:12
	→ Sigma joined	17:12
Ilari	Grum: Probably some version out of master.	17:12
Grum	no, the 1.xxx is my commit	17:12
	→ kalle_ joined	17:12
Grum	1.6.2.3 + 1 commit is what that says	17:12
	← d0k left	17:13
	← Chaser_ left	17:13
	← loincloth left	17:14
	→ jceb joined	17:14
	→ ocornu joined	17:15
Grum	humz, the local one was the one being odd for its version, so i recompiled it and it is now doing sane stuff	17:15
	might as well upgrade to 1.6.2.4 now .. :D	17:15
	← jnareb left	17:15
	→ aspotashev\|eeepc joined	17:15
	← mfilipe left	17:16
sitaram	leo2007: just for kicks: git ls-tree -lrt HEAD\|sort -k4n\|tail	17:16
	→ mfilipe joined	17:16
	→ j416 joined	17:16
	→ aziz joined	17:17
jzellman	llari: just tried git push origin master: http://gist.github.com/101218	17:17
	→ girishr joined	17:17
	← marvil07 left	17:18
	sdiz → sdiz-afk	17:19
	→ giallu joined	17:19
jzellman	llari: you'll notice the forced update I did a git show on b27a181 and it was the last commit before things got hosed by --mirror	17:19
leo2007	sitaram: http://paste.lisp.org/display/79143	17:20
	sdiz-afk → sdiz	17:21
	→ solar_ant joined	17:21
	← jmesnil left	17:22
	→ Dashkal joined	17:23
	→ reprore__ joined	17:24
Ilari	jzellman: Try 'git ls-remote origin' and see what it says as hash for refs/heads/master and compare to 'git for-each-ref' output.	17:24
	← reprore left	17:24
Ilari	jzellman: I have feeling that that HEAD branch is screwing things up.	17:24
	jzellman: Does 'refs/heads/HEAD' exist on remote repo (git ls-remote origin)?	17:25
	← vienova left	17:25
sitaram	leo2007: that doesn't look right for "sort -k4n"... It should have sorted by the 4th field, numerically	17:26
leo2007	sitaram: I am on mac os x. Would that matter?	17:27
jzellman	llari: yes refs/heads/HEAD does exist	17:27
leo2007	man the manual of sort says -k, --key=POS1[,POS2]	17:28
Ilari	jzellman: Maybe nuke it: 'git push origin :refs/heads/HEAD'.	17:28
jzellman	llari: refs/heads/HEAD is different than refs/heads/master (remote and locally). Also refs/heads/master is the same locally and remote	17:28
sitaram	leo2007: probably... sorry I don't know the syntax for BSD sort; I was hoping to get a name for the 10 largest blobs in your 40MB of repo	17:28
leo2007	s/man/and/	17:28
jzellman	llari: nuke sounds scary, haha	17:29
sitaram	leo2007: try with a space afer the k then, like "-k 4n"	17:29
Ilari	jzellman: It seems to be in past of master anyway. And is probably the source of the problems.	17:29
jzellman	llari: I think you are correct, is refs/heads/HEAD needed for anything?	17:30
Ilari	jzellman: Its local branch HEAD. Usually shadowed by ref HEAD.	17:31
	jzellman: And tends to be problematic anyway. Can it even be checked out?	17:31
leo2007	sitaram: I just did git ls-tree -lrt HEAD and the output is http://paste.lisp.org/display/79144	17:31
	it is about 10 lines only	17:31
Ilari	jzellman: Qualifying it causes checkout to detach. If its not qualified, checkout might mix it up with ref HEAD...	17:32
	→ kraut joined	17:33
	sdiz → sdiz-afk	17:33
	→ smtms joined	17:34
	← kraut left	17:34
	→ stephenjudkins joined	17:34
	← altrux left	17:34
jzellman	llari: I can check it out: git branch --track my_head origin/HEAD	17:34
	Branch my_head set up to track remote branch refs/remotes/origin/master.	17:34
	sdiz-afk → sdiz	17:34
	→ ilteris_ joined	17:34
	→ loincloth joined	17:34
Ilari	jzellman: I mean check out local branch HEAD (not to be tried in any production repo).	17:35
sitaram	leo2007: my apologies: you need "-l -r -t" -- all 3 options separated out; I did not realise git was sensitive that way :)	17:35
	it should be far longer :)	17:35
	→ altrux joined	17:36
doener	Ilari: works fine, refs/heads/HEAD takes precedence. http://git.pastebin.com/m71edf9b	17:36
leo2007	sitaram: http://paste.lisp.org/display/79146	17:36
	yes it is longer	17:36
	← solar_ant left	17:36
	← reprore__ left	17:36
	→ nis joined	17:37
	→ kraymer joined	17:37
	→ reprore_ joined	17:37
	← aspotashev\|eeepc left	17:37
Ilari	jzellman: Well, maybe just get rid of that branch. I guess its the source of the problems.	17:37
sitaram	leo2007: you seem to have got the sort working also; this does seem to be sorted the right way	17:38
leo2007	excellent	17:38
	← thraxil left	17:38
sitaram	leo2007: the bad thing is the largest file appears to be just over 600kb, which is peanuts. How about 'git ls-tree -l -r -t \| wc -l' (no need to paste; will be just a number)	17:39
	meaning just type it in the channel	17:39
	← Sho_ left	17:39
	← jceb left	17:40
	→ Sho_ joined	17:40
leo2007	sitaram: 136	17:40
justatheory	Ilari, doener, Fissure: Thanks! http://justatheory.com/computers/vcs/git/bricolage-to-git.html	17:40
	→ bleything joined	17:40
sitaram	and also git rev-list --all \| wc -l	17:41
	→ jceb joined	17:41
	← dodo_the_last left	17:41
	→ dodo_the_last joined	17:41
bleything	I'm having some troubles with git-svn. when I do a rebase, I'm getting conflicts (which I totally don't understand, but that's a different issue)... it tells me to resolve the merge and run git rebase --continue.	17:42
	← bratsche left	17:42
leo2007	sitaram: 2124	17:42
	→ aspotashev\|eeepc joined	17:42
jzellman	llari: ok im confused now. I was under the impression that HEAD should be the same as refs/heads/master. What is the point of refs/headsHEAD? Is this a typical ref or is this due to how we got back up and running after --mirror?	17:42
bleything	so I do that, and it says "No changes - did you forget to use git add". No, I did not. I resolved the commit and added it.	17:42
	→ bratsche joined	17:43
Ilari	jzellman: I have no idea where it came from. The way it causes problems is: On one fetch, the remote mirror is up to date w.r.t. master. Now, it has HEAD symlinked to master, so it isn't up to date w.r.t. HEAD. Updating head updates master due to symlink. Now master is out of date. Next fetch brings master back up to date, but cause HEAD now be out of date.	17:44
	→ solar_ant joined	17:44
Ilari	bleything: Does 'git status' show any changes?	17:45
bleything	after the git add, no.	17:45
Ilari	bleything: 'git rebase --skip'.	17:45
bleything	interesting.	17:45
	← leanucci left	17:46
Ilari	jzellman: Its presumably safe to just delete it (git push origin :refs/heads/HEAD).	17:46
	→ leanucci joined	17:46
bleything	Ilari: neat, that worked. thanks :D	17:46
sitaram	leo2007: I'm trying to figure out why a "depth=1" clone will have to many revisions; give me a minute or two	17:46
leo2007	sitaram: that's in a repo without the depth=1	17:47
	do you want me to give you the results with depth=1	17:47
	the repo is still more than 33M	17:48
Ilari	leo2007: If 'git branch -r' gives 7 branches, there are more branches than just one involved.	17:49
	← altrux left	17:50
	→ saurabh1403 joined	17:50
	← saurabh1403 left	17:51
	RaySl__ → RaySl	17:51
	← thumper left	17:52
sitaram	leo2007: oh	17:52
leo2007	Ilari: what does that mean?	17:52
	← doener left	17:52
	→ doener joined	17:52
	→ sunblush joined	17:52
Ilari	leo2007: It means that there can be more files involved than just 2M worth...	17:52
sitaram	leo2007: I just tried depth=1 on a repo with 4 branches, and I got 8 revisions. The depth always adds one (it is a known fact/bug/issue/whatever), so 2*4	17:53
leo2007	ok	17:53
sitaram	leo2007: what was the exact clone command you used in the depth=1?	17:53
leo2007	git clone --depth=1 git://repo.or.cz/org-mode.git	17:54
sitaram	leo2007: I'm stumped... but that seems to be a public repo, so let me try...	17:56
leo2007	sitaram: yes, it is public	17:56
	← jceb left	17:58
sitaram	leo2007: give me a few min to take a look around...	17:58
leo2007	sitaram: no problem, i'll be around in the next 1 hour	17:58
	← ashleyw left	17:59
	→ antoranz joined	18:00
	→ hulo joined	18:00
	→ jnareb joined	18:00
	→ benn_d joined	18:01
hulo	hellu! Im following http://toolmantim.com/articles/setting_up_a_new_remote_git_repository I created a repo in my ~ (for my config files) but there has been a problem. Everything is okay until the part where "git checkout --track -b dev/rails-2 origin/dev/rails-2	18:01
	this appears to remove all of the git-tracked files in ~/	18:01
	this is bad news for me :( as they were mostly conf files	18:01
	→ dreamcat4 joined	18:02
cpg	hi all, can someone help me clean up this merge issue? http://gist.github.com/101243	18:02
	← benn_d_ left	18:02
Ilari	hulo: Git isn't meant to track home directory.	18:02
	← intripoon left	18:02
jnareb	hulo: "git checkout -"	18:02
	hulo: git would not remove files it didn't save _somewhere_	18:03
	hulo: (at least it shouldn't)	18:03
hulo	Ilari: i know. it's for bigger projects that that	18:03
	following the instructions: i sought to create ~/scratch as the workdir, where i can make changes	18:03
Ilari	hulo: In case where thre's no existing checkout and one needs to checkout something without overwriting local files, you need to decompose the checkout operation.	18:04
	jnareb: This is one of those exceptional cases. :-(	18:04
hulo	i don't quite understand what you meant Ilari	18:04
sitaram	leo2007: well to start with there are 203 tags -- and so you have 302 revisions even in the shallow repo :)	18:04
	(203+7=210, 210*2=420, but there were some overlaps)	18:05
Ilari	hulo: Don't use checkout if you don't have existing checkout and the contents of worktree are important.	18:05
hulo	Ilari: oh	18:05
	→ khmarbaise joined	18:05
hulo	Ilari: i thought that command pretty much meant :i'll be working in branch tree named dev/rails-2 now	18:06
Ilari	hulo: If you don't have checkout yet, it will imply --force, which destroys local changes.	18:06
leo2007	sitaram: is there a way to just keep the last two revision?	18:06
hulo	Ilari: not just changes right? the original committed files too?	18:07
	i had to basically clone from the remote repo to get my files back	18:07
	← faithfulgeek left	18:07
Ilari	hulo: Destroying commits is more difficult thing to do (without nuking or corrupting the whole repo).	18:07
	→ altrux joined	18:07
Ilari	hulo: Checkout won't destroy commits.	18:08
	← jnareb left	18:08
	← sikanda left	18:09
hulo	can i ask what may be a very silly question: so in my case, ~/.git/config shouldn't mention scratch at all (only master-as branch) and ~/scratch/.git/config would have scratch-as branch with no mention of master, right?	18:09
sitaram	leo2007: yes	18:09
Ilari	hulo: There are two independent repos, each with their own contents.	18:10
sitaram	leo2007: create a blank repo (mkdir x; cd x; git init) than fetch just master from the original (not the shallow, you can't fetch from there). After that, make a shallow clone from there to some other dir	18:10
hulo	Ilari: i understood a branch to mean essentiall the same files, but with the branch containing modifications that could later be merged with the original	18:11
	s/understood/misunderstood/ ?	18:11
leo2007	sitaram: then I won't be able to git pull for updates, right?	18:11
Ilari	hulo: Pretty much. In git, branches can have multiple roots and two branches can be independent (do not share any commits).	18:12
	← esparkman left	18:13
sitaram	leo2007: no	18:14
	→ esparkman joined	18:14
sitaram	leo2007: I just checked the 10 largest blobs from all 300 revisions, removed duplicates, and their total came to 400+ MB!	18:14
	→ bobesponja joined	18:15
	← ceyusa left	18:16
sitaram	this repo has had a lot of churn; it's rare to see so much (I think) but now the 40MB .git size doesn't seem so bad :)	18:16
leo2007	sitaram: did they use git correctly?	18:17
	→ davidfetter joined	18:17
sitaram	I don't know enough git to answer that, and anyway it's subjective	18:18
	→ e_6 joined	18:20
	→ sikanda joined	18:20
leo2007	anyway I am deleting the repo now. I think I will just use the released version.	18:21
jzellman	llari: git push origin ;refs/heads/HEAD did the trick. thanks for all your help	18:22
	→ iamwilhelm joined	18:24
	← bratsche left	18:25
sitaram	leo2007: ok; good luck!	18:25
	sitaram has to sleep	18:25
	← mw___ left	18:26
leo2007	sitaram: thank you sitaram. Sleep well.	18:27
	→ bratsche joined	18:27
	← ambriento left	18:28
	→ kadoban joined	18:28
	→ ashleyw joined	18:28
	← khear left	18:29
	→ fynn joined	18:31
	→ catalis joined	18:32
	← tag left	18:32
	→ faithfulgeek joined	18:35
	← aspotashev\|eeepc left	18:37
	→ Phil_Ewert joined	18:38
	← hyperair left	18:41
	← spearce left	18:42
	← Sigma left	18:44
	→ jnareb joined	18:46
	← dwave left	18:47
	→ mikl joined	18:47
jnareb	spearce (and Dscho): would you be creating http://git.or.cz/gitwiki/SoC209Projects wiki page with this year 2 git projects?	18:47
	→ AAA_awright joined	18:48
jnareb	I do not have direct access to git.or.cz thanks to my ISP (TP S.A.) stupidity of null-route blocking of git.or.cz...	18:48
	→ hugoxrosa joined	18:49
jnareb	there are three 'foreign' git project that I found: Mono's #git, Mercurial git backend, openSUSE build git backend (perhaps there are more)	18:49
	← dodo_the_last left	18:49
	→ hyperair joined	18:50
	← catalis left	18:51
	→ catalis joined	18:51
Pieter	jnareb: what's the last one?	18:51
hulo	Ilari: can i bug you for a while longer?	18:52
	*short-while	18:52
	← ned left	18:52
	← sikanda left	18:52
	→ Jc2k joined	18:53
	→ tcurdt joined	18:54
	← solar_ant left	18:54
	→ dodo_the_last joined	18:54
tcurdt	argh ... help! folks I did a "git-svn rebase" ... all good ... then a "git-svn dcommit" which then failed with error 1 "cannot rebase: you have unstaged changes"	18:55
	what's going on? I did not have any changes ...otherwise the git-svn rebase would not have worked	18:55
	and in what state am I now?	18:55
Pieter	rebase might work with unstaged changes	18:55
	try 'git status'?	18:55
	← hyperair left	18:56
	→ hyperair joined	18:56
tcurdt	well ... now shows a lot of modified and deleted ... essentially my changes that I already have committed to git and wanted to commit to svn now	18:56
	→ kabus joined	18:57
	← chris2 left	18:57
Pieter	looks like something went wrong then	18:57
tcurdt	I did "git stash" "git-svn rebase" "git-svn dcommit" ...and the last command failed	18:58
	but what went wrong? :) ...and more importantly how can I come back to a sane state	18:58
	← hulo left	18:59
	→ jethro_ts joined	19:00
Ilari	tcurdt: Did the rebase finish?	19:00
	tcurdt: And not just fail, telling you to fix conflicts and continue rebasing...	19:00
tcurdt	Ilari: yepp ... the "git-svn rebase" went through fine	19:00
	→ mw joined	19:01
	→ lhz joined	19:01
tcurdt	apparently some of the svn commits made it	19:02
	some didn't	19:02
	seems there was an in-flight commit from someone else	19:02
	← jethro_ts left	19:02
tcurdt	but that should not break it, right?	19:02
	→ rylar joined	19:02
Ilari	tcurdt: Maybe use git reflog to find the result of last rebase, reset to that and retry rebase followed by dcommit?	19:02
	← antoranz left	19:03
tcurdt	Ilari: sounds good ... but I have never done that before :)	19:03
	← iratik left	19:03
	→ sikanda joined	19:04
	tcurdt is starring at the man page of git-reflog but has no clue	19:05
Ilari	tcurdt: Essentially, undo what git svn dcommit did to branch.	19:05
	tcurdt: Just 'git reflog'.	19:05
jnareb	Pieter: http://stick.gk2.sk/blog/2009/04/prototype-git-backend-for-opensuse-buildservice/	19:07
	Pieter: or http://socghop.appspot.com/student_project/show/google/gsoc2009/opensuse/t124023150939	19:08
tcurdt	hm ... Ilari, now what ... I see the checkout, rebase and so on ... you mean finding the sha1 that reflects the current state in svn and go back to that?	19:08
bcardarella	Is there a way to edit the commit message of a locally commited revision before pushing back to a remote?	19:08
jnareb	definitely UGFWINII ;-)	19:08
	← dreamcat4 left	19:08
jnareb	bcardarella: "git commit --amend"	19:08
	→ ned joined	19:08
bcardarella	jnareb:	19:08
jnareb	bcardarella: or GitFaq	19:08
Ilari	tcurdt: The state that's the result of last rebase.	19:08
bcardarella	jnareb: thanks	19:08
Ilari	tcurdt: And it produces helpful HEAD@{...} thingies. Those are valid SHA-1 expressions.	19:09
	→ piratepete joined	19:10
	← piratepete left	19:11
tcurdt	Ilari: hm ... I see those ... but I still have no clue how to make use of this helpful information :) "moving from master to" "updateing HEAD" "merge work: fast forward"	19:11
	no idea	19:11
Ilari	tcurdt: Can you pastebin the output?	19:12
tcurdt	man ... I am using git for a while now and still feel like a stupid newbie	19:13
	sure	19:13
	Ilari: http://rafb.net/p/g3Vmqd39.html	19:14
	→ dotsintacks joined	19:14
Ilari	tcurdt: HEAD@{2} is the last entry connected to rebasing.	19:14
tcurdt	Ilari: going from bottom to top?	19:15
Ilari	tcurdt: Actually, what branch you are currently on (git branch)?	19:15
tcurdt	on master	19:16
	← sikanda left	19:16
dotsintacks	so I was trying to add a call out to source-highlight to cgit to format my tree views nicely...but I need popen() to redirect stdout to a different file descriptor	19:16
Ilari	tcurdt: The first entry is the most recent.	19:16
dotsintacks	anyone familiar with run-command.c enough to know if it has anything that can help me accomplish this?	19:16
Ilari	tcurdt: And third entry (HEAD@{2}) is the last connected to rebasing.	19:17
	→ leanucci_ joined	19:17
tcurdt	Ilari: OK ... I see that	19:17
	← AAA_awright left	19:17
Ilari	dotsintacks: IIRC, run-command can redirect both stdin and stdout.	19:17
	→ porcnick joined	19:17
Ilari	tcurdt: After resetting to that, redo the rebase.	19:18
tcurdt	Ilari: so now I go back to HEAD@{2} and try a git-svn rebase again?	19:18
	→ AAA_awright joined	19:18
	← hyperair left	19:19
Ilari	tcurdt: Yes. Don't use 'checkout' for going back to that, use reset.	19:19
	→ sikanda joined	19:19
tcurdt	hard/soft?	19:20
Ilari	tcurdt: Hard or mixed.	19:20
	→ jtong joined	19:20
	← davidfetter left	19:23
	→ arohner joined	19:25
krh_	tcurdt: runny	19:25
	→ radarek joined	19:27
	→ loinclot_ joined	19:28
	← benn_d left	19:28
	→ benn_d joined	19:29
jnareb	krh_: heh	19:29
	→ brink joined	19:29
	→ bcardarella_ joined	19:30
brink	After git svn init and git svn fetch I have an empty directory and git status says Intial commit. Shouldn't I see some stuff?	19:32
jnareb	a'propos git jargon: it is good that we axed the last ent (= tree-ish) some time ago...	19:33
	← loincloth left	19:33
Ilari	brink: 'git branch -r'. There's your "some stuff".	19:33
brink	Ilari: I see. So do I check out that branch?	19:34
Ilari	brink: To work on those, create a local branch for it first (git checkout -b newbranch remotebranch)	19:34
	→ wsc joined	19:34
krh_	jnareb: yeah, definitely	19:34
	← leanucci left	19:34
brink	Ilari: Something like 'git branch -b master git-svn' ?	19:35
Ilari	brink: Seems reasonable.	19:35
	brink: Arrgh... checkout not branch.	19:35
brink	Ilari: Of course. Checkout.	19:36
Ilari	brink: 'git checkout -b newbranch remotebranch' is the same as 'git branch newbranch remotebranch && git checkout newbranch'.	19:36
	→ bcardarella__ joined	19:37
Pieter	heh	19:38
	google code does hg	19:38
	← bcardarella left	19:39
	← SRabbelier left	19:40
corybantic	when unstaging a file using "git reset HEAD <file>", what does it mean when the terminal say "<file>: needs update" rather than "<file>: locally modified"?	19:40
	→ SRabbelier joined	19:41
Ilari	corybantic: Both refer to some kind of uncommitted changes to <file>	19:41
corybantic	Ilari: when it says "needs update" it doesn't seem to actually unstage it -- is there something else I should be doing to unstage it?	19:42
Ilari	corybantic: 'git rm --cached'?	19:42
	→ nekrad joined	19:43
nekrad	hi folks, in some of my repos, git-repack leaves lots of single objects, instead moving evrything to the pack files - is this usual ?	19:44
	→ RodP2 joined	19:44
corybantic	yeah that doesn't seem to work either, maybe there is something unusual w/ my env. this is what's happening: http://pastie.org/private/pm7sedjrjebvjp3payrng	19:44
	→ nighthwk1 joined	19:47
Ilari	nekrad: Those are dangling?	19:48
nekrad	and this means ?	19:48
	← khmarbaise left	19:48
Ilari	nekrad: Not referenced by any ref or reflog entry...	19:49
	corybantic: CRLF translation enabled?	19:49
nekrad	but wouldnt git-gc remove them ?	19:50
	→ dwave joined	19:50
Ilari	nekrad: Not if they are recent enough.	19:50
nekrad	hmm. how old do they have to be to get removed by git-gc ?	19:51
	← dwave left	19:51
corybantic	Ilari: nope, it's not set	19:52
	← reprore_ left	19:52
Ilari	corybantic: Hmm... that reset should have unstaged those changes... BTW, that rm unstaged the entiere file, marking it for deletion. Maybe you want to reset to bring it back.	19:52
corybantic	Ilari: how would I check that crlf translation is not on? I don't see it in my ~/.gitconfig, not sure if it might be somewhere else	19:53
	→ leanucci joined	19:53
tcurdt	krh_: runny?	19:54
	Ilari: thanks for your help!! got it working!	19:54
	← leanucci_ left	19:54
	→ jayallen joined	19:55
Ilari	corybantic: It could be also in .git/config	19:55
	corybantic: But that's not likely.	19:55
corybantic	yeah doesn't seem to be there	19:55
	hmm	19:55
	i'm running git v 1.5.4.3	19:56
	← bcardarella_ left	19:57
	← catalis left	19:57
	← dotsintacks left	19:57
Ilari	corybantic: That bit... old.	19:57
	← scientes left	19:59
	← jrmuizel left	19:59
	→ antoranz joined	19:59
Ilari	corybantic: Try retracking that file (using git reset HEAD <file>) and if it still has changes, maybe use git diff --cached on it?	19:59
corybantic	ok	20:00
	← tatsuyao left	20:01
	→ mjf joined	20:01
	← tcurdt left	20:05
	→ JohnnyL joined	20:06
	→ bcardarella joined	20:08
	→ hyperair joined	20:09
	→ bcardarella_ joined	20:09
	← bcardarella__ left	20:10
	← kabus left	20:11
	→ transplant joined	20:11
	← kalle_ left	20:12
	→ kalle_ joined	20:13
	loinclot_ → loincloth	20:15
	← nekrad left	20:15
	→ yann joined	20:15
	← mfilipe left	20:16
	← bcardarella left	20:17
	← helo left	20:17
	→ schacon joined	20:18
	← bcardarella_ left	20:18
	← mjf left	20:20
	→ hiwd joined	20:20
	← jnareb left	20:21
	→ mfilipe joined	20:21
	← hugoxrosa left	20:21
	→ ehird joined	20:22
	← mw left	20:25
	→ baetis-fly joined	20:25
	→ aspotashev\|eeepc joined	20:27
	← radarek left	20:27
	← javatexan left	20:28
	→ leanucci_ joined	20:32
	→ galderz_ joined	20:32
	← galderz_ left	20:32
	← neoeinstein left	20:34
	→ jerbear joined	20:35
	← lucsky left	20:35
	→ mw joined	20:37
	← aspotashev\|eeepc left	20:37
	← galderz left	20:37
	← octagon_ left	20:37
	→ octagon_ joined	20:38
	← jerbear left	20:38
	← benn_d left	20:39
	→ benn_d joined	20:39
	← baetis-fly left	20:40
	← stephenjudkins left	20:40
	→ Cap_J_L_Picard joined	20:40
	← transplant left	20:42
	→ stephenjudkins joined	20:43
	← Cap_J_L_Picard left	20:43
	→ Cap_J_L_Picard joined	20:45
	← fridim_ left	20:47
	← hiwd left	20:48
	← brink left	20:49
	← leanucci left	20:49
	→ brink joined	20:50
	← faithfulgeek left	20:51
	→ hiwd joined	20:52
	hiwd → montylounge	20:53
	← benn_d left	20:54
	→ benn_d joined	20:54
	← corni left	20:54
	← qrush left	20:55
	← jas4711 left	20:55
	→ kate21de joined	20:56
	← esparkman left	20:57
	← AAA_awright left	20:57
	→ tobbez joined	20:58
	→ AAA_awright joined	20:58
	→ poseidon joined	20:59
	→ andrewkasper joined	20:59
tobbez	why could it be that my local git repo thinks it's ahead of origin/master when it's _at_ head of origin/master?	20:59
	← jackdempsey left	21:01
	→ hugoxrosa joined	21:01
tobbez	i just cloned a remote repo, and when i pull changes made to that repo by another collaborator i get "Your branch is ahead of 'origin/master' by 2 commits." in git status (where 2 is the number of commits the other collaborator has made)	21:01
doener	you probably did "git pull <remote> <branch>" which doesn't update the remote tracking branch, which is used to calculate those numbers	21:02
	just "git fetch <remote>" to update them	21:02
	← cemerick left	21:04
tobbez	ah, did that by mistake instead of git pull <remote> after getting used to git push where you have to provide branch. thanks :)	21:04
	→ kalperin joined	21:04
	← lorandi left	21:05
doener	tobbez: "git pull <remote>" is probably one of the most, uhm, misunderstood commands	21:05
kalperin	hiya. is there a good way to roll back the latest bad commit (already pushed to remote repository).	21:06
bremner	kalperin: man git-revert	21:06
Gitbot	kalperin: the git-revert manpage can be found at http://git.or.cz/man/git-revert	21:06
	→ Casan joined	21:06
	→ dwave joined	21:06
doener	tobbez: especially when "<remote>" really is a remote (like "origin") and not just a repo url (like git://host/repo), it probably just happens to work for the same reason that makes the "<remote>" part unneeded ;-)	21:06
kalperin	thanks bremner!	21:07
doener	tobbez: if you care about "details" --> http://git.pastebin.com/m5f1ac3b1 -- if not, just ignore that	21:07
kalperin	i'm not sure that revert will work because there are two parents	21:07
	← dwave left	21:08
	← wsc left	21:08
	→ cilly joined	21:09
bremner	oh, bummer. try in the local repo? you can always reset	21:10
doener	revert -m<parent>	21:10
	→ jas4711 joined	21:11
	← benn_d left	21:11
doener	e.g. when you did something like "git checkout master; git merge foo" and want to "revert" the merge of foo into master, then "git revert -m1 $merge-commit" will do that	21:11
	→ benn_d joined	21:11
	← cilly left	21:11
	← JohnnyL left	21:11
tobbez	doener: thanks for the explaining, reading it now	21:11
	← kalle_ left	21:12
	→ kalle_ joined	21:12
doener	OTOH, using "git revert -m2" would in this case try to revert the changes relative to the second parent (foo)	21:12
	← bryanl left	21:13
doener	note that doesn't undo the fact that git sees the branch as merged, it just reverts the changes from the merge. So if you do "git merge foo" later, it would just say "Already up-to-date"	21:13
kalperin	i'm a little concerned about that because of this from the man page:	21:13
	Reverting a merge commit declares that you will never want the tree changes brought in by the merge. As a result,	21:13
	later merges will only bring in tree changes introduced by commits that are not ancestors of the previously reverted	21:13
	merge. This may or may not be what you want.	21:13
	→ sikanda_ joined	21:13
kalperin	it seems like this would tie our hands for later merges	21:14
doener	kalperin: well, you can bring the changes back in by reverting the revert	21:14
	→ Tricon joined	21:14
	→ qrush joined	21:14
doener	kalperin: but it does get a bit error-prone, as you might forget to do that...	21:14
kalperin	yeah, that doesn't see, like an optimal solution	21:15
	reading this now: http://www.kernel.org/pub/software/scm/git/docs/howto/revert-a-faulty-merge.txt	21:15
	→ ashleyw_ joined	21:15
cxreg	is removing $GIT_DIR/refs/original an intentionally manual step of filter-branch?	21:16
doener	cxreg: isn't that true for all kinds of backups? You don't want some tool to automatically destroy it, do you? ;-)	21:16
	→ s0ber_ joined	21:17
	→ qrush_ joined	21:18
	← qrush left	21:19
cxreg	doener: heh. fair enough. i'm just trying to form succinct instructions for people to fix their commit notes if we reject them using an update hook	21:19
	→ qrush joined	21:20
cxreg	because it's hard to make everyone remember to install pre-commit hooks :/	21:20
	← qrush_ left	21:20
cxreg	or PREPARE-COMMIT-MSG or whatnot	21:20
	→ univate joined	21:20
doener	cxreg: if there are no merges involved, I think rebase -i is better than filter-branch (unless you need to do basically the same change to every commit message, but that seems like a weird requirement)	21:21
	← qrush left	21:22
cxreg	doener: yeah but rebase -i is harder to teach people	21:22
doener	cxreg: and if you use filter-branch, there's an easy for-each-ref + xargs command in the man page to get rid of refs/original	21:22
	← sikanda left	21:22
	← ashleyw left	21:22
cxreg	well, maybe not	21:22
	it's just edit and ammend	21:22
	← Phil_Ewert left	21:23
	← Tricon left	21:23
	→ gilimanjaro joined	21:23
cxreg	doener: hm, ok thanks	21:23
	← montylounge left	21:23
cxreg	i didn't even know for-each-ref existed :-)	21:24
doener	cxreg: harder than filter-branch --msg-filter? No way... That filter needs a script that spits out the right commit message for each commit to be filtered	21:24
	→ catalis joined	21:24
	→ wsc joined	21:24
cxreg	yeah. hm.	21:24
	← sikanda_ left	21:25
	→ earcar joined	21:25
	→ sikanda joined	21:25
	← seangrove left	21:26
cxreg	maybe i should write a git fix-yer-damn-notes and distribute that	21:26
	← wsc left	21:26
	→ dodo_the_last1 joined	21:27
	← Weaselweb left	21:27
	← dodo_the_last left	21:27
	→ Tricon joined	21:28
	← markelikalderon left	21:29
	← earcar left	21:30
	← binjured left	21:30
	← gilimanjaro left	21:30
kalperin	so here's what i want to try for the revert:	21:31
	git revert -m 509e34fc458bae4c5af70558f82f2714fe8c27d3 6edff054e4920bc7d6de3f7d142207fa33ce542c	21:31
	← sunblush left	21:31
kalperin	i think i'm referring to those revs incorrectly	21:32
MarcWeber	Which is the easiest way to check wether the working directory is clean?	21:32
bremner	MarcWeber: git status?	21:32
MarcWeber	bremner I'd like to use it within a script. &> /dev/null and check exit status?	21:32
bremner	git status -a > /dev/null	21:32
	err, check the -a	21:33
	→ hax0r1337 joined	21:33
	← bdiego left	21:33
bremner	afaik, that's the way	21:33
	← leanucci_ left	21:33
	← LiamH left	21:34
	← hax0r1337 left	21:34
kalperin	how can i tell which parent is m1 and which is m2?	21:35
	← wmoxam left	21:35
doener	kalperin: -m takes a number to identify the parent, first parent, second parent and so on, not the actual parent	21:35
bremner	MarcWeber: I _think_ -a doesn't treat no changes as a an error	21:35
kalperin	thanks doener. how can i tell which is 1 and which is 2?	21:36
doener	kalperin: using --parents with "git show/log/rev-list/..." should show them in the right order	21:36
kalperin	excellent, thanks	21:36
doener	kalperin: or just do "git show $commit^1", "git show $commit^2"	21:36
	kalperin: the ^n syntax selects the nths parent of the commit	21:36
	← eletuchy left	21:37
	← Tuomas left	21:38
	← benn_d left	21:40
	← stephenjudkins left	21:41
	← RaySl left	21:42
	← mfilipe left	21:43
	← kraymer left	21:43
	← krh_ left	21:43
MarcWeber	doener: :-) Thank you! -a is fine	21:44
doener	MarcWeber: bremner	21:45
	→ neoeinstein joined	21:46
	→ RaySl joined	21:48
doener	MarcWeber: but as you pulled me into the discussion, I'd probably avoid using "git status" in a script, especially since there were some discussion about changing "git status" from being a "git commit" dry-run	21:49
	MarcWeber: using the "update-index -q --refresh" and "diff-files --quiet" or so is probably better/more future-proof (i.e. use plumbing for scripting)	21:51
	← antoranz left	21:52
	← raydeo left	21:53
	← loincloth left	21:53
	→ raydeo joined	21:53
MarcWeber	doener Thanks. I've taken a note. I'll dive into this later then	21:54
	← j416 left	21:54
doener	MarcWeber: There should be plenty of examples in the git commands that are shell scripts, e.g. git-rebase	21:55
	→ boto_ joined	21:56
	← ashleyw_ left	21:56
	→ bremner_ joined	21:58
tclineks	when i git svn rebase from a branch tracking an svn branch it tries to pull trunk commits onto my branch	22:00
doener	tclineks: you probably merged trunk	22:00
	... to your branch	22:00
tclineks	is git rebase (svn branch ref) the right thing to do?	22:00
	:/	22:01
	← tobbez left	22:01
doener	rebase linearizes history	22:01
tclineks	i thought it just looked for the most recent commit with a git-svn-id line	22:01
	← HG` left	22:01
doener	tclineks: let's say you have branch "foo", and remotes/foo is the corresponding remote tracking branch for the svn branch	22:01
	then "git svn rebase" fetches stuff from svn updating remotes/foo	22:02
	→ flaguy48 joined	22:02
doener	and then does "git rebase remotes/foo"	22:02
	the "full" form of that command is: "git rebase --onto remote/foo remotes/foo foo"	22:02
	→ leanucci joined	22:02
tclineks	hm	22:02
	← RodP2 left	22:02
doener	and it takes all commits from the range remotes/foo..foo, "replays" them onto remotes/foo and makes "foo" reference the result	22:03
	now, if you merged "trunk" to "foo" after your last dcommit, that merge commit will be in that range remotes/foo..foo	22:03
	and so are all the commits from trunk's history that you merged	22:04
tclineks	i don't think i have trunk merged	22:04
doener	and as rebase linearizes history, it doesn't keep that merge commit, but just the commits from the merged history	22:04
	tclineks: did you finish the rebase? or did you abort it?	22:05
tclineks	aborted	22:05
	← leanucci left	22:06
	→ leanucci joined	22:06
doener	tclineks: ok, what's the local branch head, and what's the corresponding remote tracking branch for the svn branch?	22:06
tclineks	foo and svn/releases/foo	22:07
	← hugoxrosa left	22:07
	→ ashleyw joined	22:07
doener	ok, then "git rev-list --parents svn/releases/foo..foo \| grep '.\{82\}'" should tell if there are any merge commits in the to-be-rebased history	22:08
	← Casan left	22:08
	→ jceb joined	22:08
	→ rektide_ joined	22:09
	← patrikf left	22:09
hachi	I'm still having some issues with my brain and bare repos... pardon if someone feels like they already explained it to me	22:09
	even on a bare repo git seems to create a 'master' for me... and have it pointing somewhere	22:10
	and it's not pointing at anything meaningful, but then if I use gitweb on the bare repo	22:10
	← boto left	22:10
hachi	it reports where master is... not where the latest commits are	22:10
	does anyone have advice for dealing with this sort thing... what I'm supposed to do?	22:10
tclineks	doener: no output	22:11
	→ hugoxrosa joined	22:11
tclineks	there are refs listed though (just no merge commits)	22:11
hachi	rather, is there a way I can get HEAD to point to the newest commit, regardless of branch ?	22:11
	← sfwc left	22:12
	← kalle_ left	22:12
doener	hachi: no, and you'd have to define "newest" anyway. The last pushed one? The one with the latest commit date? The one with the latest author date? Something else?	22:12
hachi	:\	22:12
	→ bremner__ joined	22:13
	→ kalle_ joined	22:13
	← bremner_ left	22:13
hachi	this is a bare repo purely to allow others to clone from it, so I don't think there's any 'push'	22:13
doener	hachi: and it would lead to "interesting" behaviour, as HEAD determines the "primary" branch of a bare repo	22:13
hachi	oh	22:13
	it does?	22:13
	okay, then... can I safely have HEAD refer to a remote?	22:13
	← agile left	22:13
hachi	and if not... how can I move master?	22:13
doener	ehrm, a remote? You can't reference a remote, that's a config entity	22:14
hachi	since I can't use 'reset', without a working tree	22:14
	no, I mean a remote branch	22:14
cxreg	update-ref	22:14
LotR	hachi: maybe you have to step back and explain why it is you think you want these things	22:14
	→ gilimanjaro joined	22:14
doener	hachi: HEAD is used e.g. by "git clone" to determine which branch to create	22:14
	hachi: if the bare repo has HEAD referencing refs/heads/master, the clone will have a local branch head "master"	22:14
hachi	okay	22:15
doener	hachi: if it references refs/heads/foo, "git clone" will create "foo" instead	22:15
hachi	LotR: I have these git-svn clones that I want to be able to clone out to places easily	22:15
	they're bare repos	22:15
drizzd_	doener: well, it's still master, but it points to origin/foo	22:15
doener	drizzd_: no	22:15
hachi	but when I do a fetch on them, the 'master' isn't updated... and I can't call 'rebase' because it has no working tree	22:15
doener	hachi: ah, git-svn, that creates "master" after each fetch, if it doesn't exist yet	22:16
hachi	yes, but now 'master' is stuck someplace in the past	22:16
	← wshimmy left	22:16
hachi	so I need to move master after I do each git-svn fetch	22:16
	IE, I need to rebase	22:16
	maybe that's all I should be doing	22:16
	→ leanucci_ joined	22:16
hachi	rebasing	22:16
drizzd_	doener: my bad	22:16
	← leanucci left	22:17
hachi	does git-rebase require a working tree?	22:17
	← km2 left	22:17
doener	hachi: no... if you don't need master, I'd rather drop it after each fetch, less confusion	22:17
hachi	yeah, but I'm better off having it exist	22:17
	→ wshimmy joined	22:17
doener	hachi: how so?	22:18
hachi	that way people dont end up on a random branch when they clone	22:18
	I can force them to be on 'trunk' (the subversion concept) when they clone from me	22:18
	by default	22:18
doener	hachi: git symbolic-ref HEAD refs/heads/trunk	22:18
	← neoeinstein left	22:18
hachi	okay, yeah	22:18
	→ leanucci joined	22:19
hachi	except it's refs/remotes/trunk, isn't it?	22:19
	I'm trying this out now	22:19
doener	hachi: wait a second	22:19
hachi	oh wait	22:19
	→ nealinux joined	22:19
hachi	you're saying... make a local one called 'trunk' and have it follow remotes/trunk	22:19
	I think	22:19
	bleything → bleything\|away	22:20
cxreg	remotes/[remote name]/trunk	22:20
	← bleything\|away left	22:20
	→ bleything joined	22:20
drizzd_	I'm not sure what happens if you make HEAD a symbolic ref to something that's not a local branch.	22:20
	← Tricon left	22:21
cxreg	hilarity ensues?	22:21
hachi	what do you suspect might break?	22:21
drizzd_	I can imagine it will be treated like a local branch.	22:21
ciaranm	you can do it, but you can't commit iirc	22:21
drizzd_	but what if you point it to a tag?	22:21
	← juanjoc left	22:22
drizzd_	ciaranm: you actually tried? :-O	22:22
ciaranm	back in the good old days it was fairly easy to do that accidentally	22:22
	you could checkout a remote branch without getting screamed at	22:22
hachi	what happened in the old days if you committed to a remotes/whatever/branch ?	22:22
	did `pull` break?	22:22
	← botanicus_ left	22:23
ciaranm	can't remember. i think it stopped you before that point.	22:23
	try it. you can probably still git checkout remote/blah.	22:23
drizzd_	hachi: I'd say "undefined behavior". If you want to mess with the repo like that, please do so at your own risk.	22:23
hachi	you can	22:23
	it yells at you	22:23
	drizzd_: I'm... not too worried though	22:23
	since this is basically a read only repo	22:24
doener	hachi: assuming that that repo is a plain svn mirror: http://git.pastebin.com/m1838b28e	22:24
	← sikanda left	22:24
	→ absentia joined	22:25
doener	hachi: damn, I missed the symbolic-ref step, this should be complete: http://git.pastebin.com/m75455cf	22:25
absentia	does git just arbitrarily ignore files? I just can't get git to see/add www/run and www/logs	22:25
	I've tried erasing the dir	22:25
	adding it again	22:25
	adding the individual file	22:25
	removing the files	22:25
	renaing the files	22:25
doener	absentia: empty directories? git doesn't track directories, just files	22:26
absentia	yes, empty	22:26
	can I tell it to f'n track empty dirs?	22:26
	→ toad_ joined	22:26
hachi	where does gitweb read the 'Last Change' from?	22:26
	now, after making these changes to a repo	22:26
	it still says my Last Change was weeks ago	22:27
doener	absentia: I'd put a .gitignore containing "*" in them	22:27
toad_	what does this mean?: HEAD is now at 15dd668 doh! off-by-one error (with test case)	22:27
	← trmanco left	22:27
absentia	o, thanks!	22:27
hachi	oh, I bet... master	22:27
	← absentia left	22:27
hachi	okay, this makes sense	22:27
	→ hacim joined	22:27
hacim	i'm periodically getting this in my logs: git-daemon: [27289] 'receive-pack': service not enabled for '/srv/gitosis/repositories/crabgrass.git'	22:27
doener	hachi: what that "recipe" does is using a "proxy" repo for the svn fetching	22:27
	← kate21de left	22:27
hachi	doener: yeah... like the git-svn docs	22:27
hacim	what is causing that, what is receive-pack service do I want receive-pack service enabled?	22:28
hachi	that's a large part of what I'm doing	22:28
doener	hachi: and the "push" part makes the refs/remotes/* appear as refs/heads/* in the bare repo	22:28
hacim	i guess receive-pack is an anonymous push?	22:28
doener	hachi: so setting HEAD and cloning "just works", no hassle for the users	22:28
hachi	thanks doener, I'll poke at this more	22:28
	toad_ is trying to switch from svn to git, a dev built a script, it just quit out, i don't know at what point it quit out, but backing up the imported repo and running the commands it was supposed to run gives that error ...	22:28
	← raydeo left	22:28
	← LotR left	22:29
doener	hachi: no, the git-svn manpage puts the work on the users cloning your repo, they have to modify the fetch refspec for their remote	22:29
toad_	script for importing the repo from svn (this seems to have worked), and then splitting it up into separate repo's for each subproject (this has definitely not worked)	22:29
hachi	oh.... oh wow	22:29
	← willb left	22:29
hachi	you're stellar	22:29
doener	hachi: the setup in the pastebin does that work on the server side	22:29
hachi	thanks for that bit of info	22:29
	is there an easy way I can change a bare repo to be a... non-bare repo?	22:30
	hmm, maybe I should just keep it bare	22:30
	do you just mv the stuff into .git and `checkout` ?	22:30
doener	hachi: there's also a patch on the mailing list that allows git-svn to work on bare repos, putting stuff directly into refs/heads/ instead of refs/remotes, which will makes this stuff a lot easier (no proxy repo needed)	22:31
	→ qrush joined	22:31
hachi	YES	22:31
Grum	hachi: and remove the line: bare = true from the .git/config	22:31
hachi	I look forward to that	22:31
	or I suppose I could apply it right now	22:31
drizzd_	doener: what about your fetch refspec hack?	22:31
hachi	Grum: thank you	22:31
Grum	and git checkout master ;)	22:32
	to get a working dir as well ;)	22:32
doener	drizzd_: hm? Which one? Regarding git-svn, there's a fetch refspec thing in the man page, the recipe I pastebinned uses a push refspec "hack"	22:32
hachi	git is kinda painful	22:32
	cause every time I talk to you folks	22:33
doener	drizzd_: or do you mean something totally different?	22:33
	← bremner__ left	22:33
hachi	which... you are all awesome by the way	22:33
drizzd_	doener: what you just described could simply be achieved using git config remote.origin.fetch +refs/heads/:refs/heads/	22:33
hachi	I get an even better way to do things	22:33
drizzd_	or not?	22:33
doener	drizzd_: yeah, on the client side. But that means that each client has to do that. And it doesn't solve the "HEAD problem"	22:33
schacon	would a library injecting extra data into the commit object be horribly bad? 'encoding' is the only one i can think of that is quite variable. if a library adds extra custom stuff would that break anything in the core git tools?	22:34
doener	drizzd_: using the "proxy + push" or the git-svn+bare-repo patch, users can just clone as usual, less work for them	22:34
	→ Casan joined	22:34
toad_	do I have any options apart from starting from scratch without history, or re-running the script with more self-documentation (it will take at least 3-4 hours) ? and what does the error message mean anyway?: "HEAD is now at 15dd668 doh! off-by-one error (with test case)"	22:35
	← qrush left	22:35
	→ ikanobori joined	22:35
ikanobori	is it possible to have git push, push to two repo's?	22:35
	← lhz left	22:35
	← leanucci_ left	22:36
drizzd_	doener: sorry, I think I'm too tired for this. I'll get back to you if I still don't get it tomorrow. ;-)	22:36
tclineks	ikanobori: no, i don't think so	22:36
hachi	doener: how can I find this patchy patch patch?	22:36
	I'm... rubbish at finding things that I don't know the name of	22:36
	← Sho_ left	22:37
ikanobori	tclineks: mrm, sucks :)	22:37
tclineks	shrugh	22:38
	← yann left	22:38
tclineks	alias it	22:38
hachi	doener: this patch makes it so I can just change my fetch/branches/tags lines to say refs/heads instead of refs/remotes?	22:38
	→ soycamo joined	22:38
hachi	cause it feels like I should be able to change them right now	22:38
doener	hachi: http://www.spinics.net/lists/git/msg101188.html	22:38
ikanobori	oh wait	22:38
	found it	22:38
doener	hachi: not sure how that patch works in detail	22:38
	→ bgerber joined	22:38
hachi	ccheers	22:38
	← hugoxrosa left	22:39
toad_	if i post my error message again will i get kicked?	22:39
	← leanucci left	22:39
Grum	git-svn should get a proper 'one time export' function	22:39
	toad_: use a pastebin somewhere	22:39
doener	hachi: and you can't just change them to refs/heads right now, because git-svn doesn't check bareness at all, yet. So it prevents you from having fetch "killing" your local branch heads	22:39
toad_	for 62 characters?	22:40
Grum	how many lines? ;)	22:40
toad_	if you say so	22:40
	one line but i already asked twice and ppl ignored me	22:40
Grum	o.O	22:40
doener	toad_: that's not an error message, but a status message	22:40
Grum	just ask again	22:40
toad_	doener: even with the doh! ?	22:40
doener	toad_: the "doh! off-by-one error (with test case)" is the subject line of the commit message of that commit	22:40
	← cmarcelo left	22:41
toad_	ah okay thanks	22:41
doener	toad_: try "git show 15dd668"	22:41
toad_	okay ...	22:41
	well i've got these three commands, any of them could have quit ... git reset --hard HEAD, git checkout git-svn, git reset --hard, git clean -dfx	22:42
	is it safe to rerun them all? presumably ...	22:42
	→ leanucci joined	22:42
doener	basically, yes, the "reset --hard" drops uncommitted changes, the checkout checks that branch/commit out, and "clean -xdf" removes all untracked files	22:43
toad_	thanks	22:44
doener	so unless you added new changes/files, you already killed everything they might kill	22:44
	→ botanicus joined	22:44
toad_	okay now it's invoking git-filter-branch badly somehow ...	22:44
drizzd_	toad_: maybe you should be talking to the author of the script...	22:44
toad_	yeah but he's not around and he's in hong kong so won't be around ...	22:44
	→ gilimanjaro_ joined	22:45
	→ cyphactor joined	22:45
	← corybantic left	22:46
	← gilimanjaro left	22:46
doener	tclineks: sorry, got distracted. If that "rev-list + grep" thing produced any output, then those are merge commits (and their parents)	22:46
tclineks	it didn't	22:47
AAA_awright	git-gui no longer a command?	22:47
	I thought it was a seperate program, why is it part of git now?	22:47
tclineks	i think i must've reset my branch to a trunk commit or something, propbably operator error	22:47
doener	AAA_awright: "git gui"	22:47
tclineks	i just reset to the remote ref	22:47
AAA_awright	I just discovered that	22:47
doener	AAA_awright: the dashed forms got deprecated in 1.6.0	22:47
AAA_awright	I thought that git-gui was still seperate from the git biniary	22:48
	← nighthwk1 left	22:48
	→ scarabx joined	22:48
	← DavidKlein left	22:48
	← cyphactor left	22:49
	→ lritter joined	22:49
	→ flujan joined	22:50
lritter	Hello?	22:50
toad_	hmmm my git doesn't have filter-branch --prune-empty ...	22:50
flujan	hello guys, I hava repository with git... I wanna remove a file and checkout it again . how is it possible?	22:51
Grum	you mean you just want to do a checkout of 1 file? :D	22:51
toad_	i'm guessing running the git-filter-branch without --prune-empty will result in a lot of empty commits?	22:51
	→ qrush joined	22:52
	← nealinux left	22:52
lritter	Can anyone diagnose this when I am trying to install a plugin for Rails: sh: git: command not found	22:52
		22:52
tclineks	toad_: totally depends on what sort of filter you apply	22:52
	← gilimanjaro_ left	22:53
Grum	lritter: that you dont have 'git' in your path?	22:53
toad_	git filter-branch -f --subdirectory-filter trunk/website --msg-filter sed "s/$git-svn-id:.*$@/\1\/trunk\/website@/" --prune-empty website	22:53
parasti	lritter: install git...	22:53
lritter	Ohh, how do I get it in my path?	22:53
Grum	install it!	22:53
toad_	lritter: apt-get install git	22:53
lritter	I installed git	22:53
toad_	lritter: from source code?	22:53
lritter	yeah	22:54
Grum	did you install it systemwide? or just make; make install ?	22:54
toad_	did you install it or just compile it?	22:54
lritter	does it have to be in my rails dir or something?	22:54
Grum	(that is defaultly in the users homedir)	22:54
lritter	Oh humm, I think I installed it	22:54
parasti	recompile and reinstall with make prefix=/usr/local	22:54
Grum	if he has rights to do so	22:54
parasti	(of course, it's arguably easier to change PATH)	22:55
lritter	OK, should I delete the existing copy?	22:55
	← hpoydar left	22:56
	← rylar left	22:56
Grum	delete what from where exactly lritter? :)	22:57
lritter	Well, git did install but I think it went to /opt/git/bin/git	22:57
hachi	fatal: HEAD not found below refs/heads!	22:58
	hachi screams and runs around	22:58
soycamo	you've lost your head?	22:58
hachi	:]	22:58
	yes	22:58
nis	flujan: could you explain what you expect?	22:58
toad_	lritter: export PATH=/opt/git/bin/git:$PATH	22:58
	lritter: but for a more permanent solution reinstall with a proper prefix	22:58
flujan	nis: i deleted some files accidentally... need to restore them	22:58
hachi	I actually can't figure out what this thing is complaining about, but yeah	22:59
	← arohner left	22:59
	→ markelikalderon joined	22:59
lritter	OK, so reinstalling with make prefix=/usr/local will make it happy?	23:00
	→ arohner joined	23:00
	← Yuuhi left	23:01
	← hacim left	23:01
toad_	normally prefix is a configure option, no?	23:02
	./configure --prefix=/usr/local	23:02
	← schacon left	23:02
toad_	hmmm apparently not in git	23:03
nis	flujan: you could do 'git show HEAD:foo > foo'	23:03
	← leanucci left	23:04
lritter	Oh joy! It worked!	23:04
	→ Eudoxus joined	23:04
	→ agile joined	23:04
nomis	toad_: when using the source code from svn/git you need to pass all these options to autogen.sh.	23:04
Eudoxus	If I am a single developer working on a single project locally, is it even worth it for me to bother moving to Git from Svn?	23:04
lritter	Thanks beautiful IRC peoples!	23:05
nomis	argls.	23:05
	sorry. I was under the impression of being in #gimp...	23:05
	← lritter left	23:05
madewokherd	Eudoxus: I'd say so, just from the ease of setting up/manipulating repos	23:06
	no need to actually run a git server to use it	23:06
nis	tango_: --prefix exists here	23:07
	← arohner left	23:07
nis	rah	23:07
	toad_: --prefix exists here	23:07
toad_	fortunately the tarball is signed by the same key as the kernel, so i can get a clean copy with no trouble	23:08
parasti	nis: but configure might not exist	23:08
nis	parasti: create it, then. :-)	23:09
	parasti: make configure	23:09
parasti	never know if that's in sync with the Makefile...	23:10
	← joevandyk left	23:10
toad_	oooooh, the import script is really working	23:11
	i think i'll end up with all the subprojects being branches though, i want them as separate repo's, i guess that's trivial to fix	23:11
nis	flujan: Oh, i've forgotten: 'git checkout -- path' should do the trick too	23:12
	← kalle_ left	23:12
flujan	nis: thanks for the tip nis	23:12
	→ killerchicken joined	23:15
kalperin	still working on reverting a bad commit pushed to the repository	23:15
	anyone know what this means: Automatic revert failed. After resolving the conflicts,	23:16
	mark the corrected paths with 'git add <paths>' or 'git rm <paths>' and commit the result.	23:16
	there's actually not much information on the google	23:16
	→ airtonix joined	23:16
toad_	kalperin: it means there were conflicts?	23:16
nis	kalperin: git status should help you there.	23:16
kalperin	i can just resolve and commit, right?	23:17
killerchicken	hey, I have a bare repo that I pushed to my server, and I serve it via gitweb. Ideally, I'd like to use a cronjob to keep it up to date from upstream. I can guarantee that I will never push again, so every update will be a fast forward. How would I achieve that?	23:17
	Or are there other good ideas?	23:17
	→ kalle_ joined	23:17
nis	kalperin: Normally, yes.	23:17
	← parasti left	23:17
nis	killerchicken: using a trivial shell script?	23:19
	killerchicken: what's the problem?	23:19
killerchicken	I'm not sure how to keep the repo updates.	23:19
	→ amahi joined	23:19
killerchicken	updated*	23:19
	→ ajonat joined	23:20
	→ davidfetter joined	23:20
	← markelikalderon left	23:20
nis	killerchicken: add a script in your cron jobs sounds good.	23:20
killerchicken	erm. The problem is not adding a script. The problem is what to put in that script	23:20
	→ eletuchy joined	23:21
nis	killerchicken: exactly what you would do in an interactive session.	23:21
killerchicken	I don't know what I would do in an interactive session.	23:21
nis	(with the shebang)	23:21
killerchicken	Can't you just give me the git command?	23:22
	That's the only piece I'm missing	23:22
	← qrush left	23:22
	→ neoeinstein joined	23:22
nis	git pull ?	23:22
	git fetch ?	23:22
	← altrux left	23:22
killerchicken	git pull cannot work - it's a bare repo	23:22
	I thought about fetch, but I'm missing some options, I think.	23:23
	When I use it, my gitweb checkout isn't updated	23:23
	or rather, not checkout	23:23
	but the view on the website	23:23
	← davidad left	23:23
	→ leanucci joined	23:24
	→ qrush joined	23:24
nis	is 'git log' ok?	23:24
	(up-to-date)	23:24
	← d0k_ left	23:24
killerchicken	no	23:24
	When I fetch, it says " * branch HEAD -> FETCH_HEAD"	23:25
	I just want it to update whatever I have.	23:25
	→ peff joined	23:25
	← Eudoxus left	23:26
	← cpg left	23:27
	rektide_ → rektide	23:27
	→ KwikkSilva joined	23:28
	← leanucci left	23:29
	→ JohnnyL joined	23:29
	← btanaka left	23:30
	← ocornu left	23:30
	← Eridius left	23:30
	→ cpg joined	23:30
	→ leanucci joined	23:31
	← leanucci left	23:31
	→ stephenjudkins joined	23:31
	← woei left	23:33
	← stephenjudkins left	23:33
	← Ryback_ left	23:34
	octagon_ → octagon	23:35
	→ markelikalderon joined	23:36
	← amahi left	23:38
	→ seangrove joined	23:40
	→ jmatthews_ joined	23:40
	← gauge left	23:41
	← jmatthews_ left	23:41
	→ km2 joined	23:42
hachi	doener: whenever I do that 'git push bare' like you suggested for the svn bare stuff	23:44
	it claims to be creating a whole slew of new branches... roughly all of them	23:44
	dfr\|work → dfr\|noob	23:44
hachi	rerun it immediately, and it says it's creating new ones again	23:45
	something seems... off	23:45
	→ k0001 joined	23:45
	← nis left	23:45
	dfr\|noob → dfr\|work	23:47
	← SRabbelier left	23:49
	→ juanjoc joined	23:50
	→ stephenjudkins joined	23:51
	← RaySl left	23:54
	→ Leemp joined	23:56
	→ Eridius joined	23:58

Logs Search ←Prev date Next date→

Channels Documentation