IRCloggy #git 2008-11-20

Provider of IRC logs since 2005.

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

	2008-11-20
	→ EmilMedve joined	00:00
	← chris2 left	00:02
	← willb left	00:05
	← Wyk3d1 left	00:06
	→ jmatthews_ joined	00:06
	← jmatthews_ left	00:07
	← gottesmm left	00:08
	← Lemurnomicon left	00:09
	← loincloth left	00:10
	→ loincloth joined	00:11
	← jstad left	00:12
	partha → partha-bbl	00:13
	→ coderdad joined	00:14
	→ litage joined	00:15
	← Gitzilla left	00:15
	→ Gitzilla joined	00:16
	← cgardner left	00:17
	← radarek left	00:19
	← jm left	00:19
	aroben\|away → aroben	00:24
	→ jmatthews_ joined	00:25
	← kanru left	00:25
	← jmatthews_ left	00:25
	← loincloth left	00:28
bentob0x	ok, I'm trying to move a folder containing files into another folder containing files too eg: git mv images/ media/images/	00:29
	but I get: fatal: cannot move directory over file	00:29
	← rubbish left	00:29
	→ Gitzilla_ joined	00:29
bentob0x	sorry, I should specify that media/ already has an image directory and I use: git mv images/ media/	00:30
	git mv images/ media/images/ gives media/images/images/ as a result	00:31
vmiklos	git mv images/* media/images?	00:31
bentob0x	yes but images/ will still be there	00:31
vmiklos	rmdir images/	00:31
	← Gitzilla left	00:31
bentob0x	although git doesn't track empty directories	00:31
vmiklos	git does not track empty dirs, so that does not matter	00:31
	exactly	00:32
bentob0x	yea, I need to clean up that afterwards	00:32
	but technically, directories are files ...	00:32
	:)	00:32
	← ahupp\|fb left	00:32
vmiklos	git tracks "content"	00:32
bentob0x	touché	00:32
	thx vmiklos	00:32
vmiklos	np	00:33
	→ ahupp\|fb joined	00:33
numillumi	i hope i don't get flamed for asking: how do I list all revisions so i know the string to use for git show	00:34
	← elpargo left	00:35
	→ cgardner joined	00:35
vmiklos	git log --all?	00:35
	→ elpargo joined	00:35
vmiklos	flaming off people for asking is not a common habit in this channel ;)	00:36
numillumi	thx vmik	00:37
	→ johnw_ joined	00:37
	mw → mw\|out	00:40
	→ bieneff joined	00:41
	→ jmatthews_ joined	00:42
	← Loof left	00:43
	← aroben left	00:44
	← johnw left	00:45
	→ jesselucas joined	00:47
	→ madewokherd joined	00:48
	← jesselucas left	00:51
	→ kalikiana1 joined	00:53
kalikiana1	Is there any way git can give me a continuous number, ie. as if all existing commits were sitting in a long list?	00:54
	I basically need an indicator that says "this version is newer than that version"	00:55
jogla	Thats the date	00:55
	← tango_ left	00:55
	← mw\|out left	00:56
kalikiana1	Oh, stupid me, that's the obvious solution	00:56
	Thanks for the hint	00:56
	→ jesselucas joined	00:57
cbreak	newer date doesn't mean that it's newer	00:58
	the most recent commit could be based on one a year ago and just change a spelling mistake in a readme file	00:58
	← priidu left	00:59
kalikiana1	cbreak: That commit would still have a new date, that's all I need. Basically I want someone writing an extension for my application to be able to target a particular build	01:00
	→ stuffcorpse joined	01:00
kalikiana1	Which is mainly because I won't guarantee api for the moment	01:00
	→ rubbish joined	01:00
cbreak	date is not a good ID for that	01:00
	→ bobmcw_ joined	01:00
bremner	kalikiana1: make tags?	01:01
kalikiana1	Well, it ensures that the state of the api is known without me having to change a value with every second commit or so	01:02
	bremner: I want to avoid keeping track of every single, possibly unimportant change	01:02
	← jmatthews_ left	01:02
cbreak	how about using a branch for stable api and one for dev stuff?	01:03
kalikiana1	The idea is coming from experience, where a project introduces "unstable" api, which is there for quite a while, even over versions, but never appears publicly - and yet it's too crucial to not use it	01:04
	s/never/newer	01:04
cbreak	well, better that than having it not work :)	01:04
	sstangl → rkh	01:05
kalikiana1	If I make a branch it means virtually nobody will test it. It's like patches in a bug tracker not being tried out.	01:05
	And only once it's stable people suddenly notice it's totally unusable :P	01:06
cbreak	well, so? :)	01:06
	if you use the newest by date, you will get a random selection of commits from various branches	01:07
	← jesselucas left	01:07
	kalikiana1 → kalikiana	01:07
	→ tango_ joined	01:07
	→ jesselucas joined	01:08
	← jesselucas left	01:08
johnw_	is there a clever way to untabify just the changes introduced by a commit?	01:09
	johnw_ → johnw	01:09
kalikiana	^^ Let a script look at staged files and fix them?	01:10
	→ peeja joined	01:10
johnw	i don't want it to untabify all the code	01:10
	just the parts I changed	01:10
	rkh → sstangl	01:10
cbreak	what do you mean with untabify?	01:10
	→ joydivider joined	01:10
johnw	convert tabs to spaces	01:11
	i can just untabify the patch file, because it introduces an extra character at the beginning of the line	01:11
	s/can/can't	01:11
	→ bobmcw__ joined	01:12
johnw	ooh, maybe I can do this with an Emacs macro	01:12
	yes, that works nicely	01:13
	← ceej left	01:14
	→ jstad joined	01:16
	← bobmcw left	01:17
	← felipe left	01:17
	← tango_ left	01:18
bentob0x	I want to delete one commit out of a series of commit but I know that the commit has no incidences on the others	01:19
	git rebase -i and ignore that patch?	01:19
	→ jmatthews_ joined	01:19
	→ beilabs joined	01:19
	← jmatthews_ left	01:20
vmiklos	yes, just delete the given line from the "todo"	01:20
	→ tenpaiyomi joined	01:20
tenpaiyomi	Need some git help, if possible. I have a commit, like 5 or 6 commits back, that I need to extract out and move it to it's own branch. How would I go about doing that?	01:20
bentob0x	nice, thx vmiklos	01:20
	git is a beauty really	01:20
vmiklos	tenpaiyomi: git branch newname HEAD~5?	01:21
tenpaiyomi	vmiklos: Will that remove that commit from the branch it's in?	01:21
vmiklos	of course not	01:21
	you can't remove a commit without rewriting history	01:21
tenpaiyomi	Ok. Is there a way to.....make a new commit, that is the reversal of an old commit?	01:22
vmiklos	see man git-revert	01:22
Gitbot	vmiklos: the git-revert manpage can be found at http://git.or.cz/man/git-revert	01:22
tenpaiyomi	Righto, thank you	01:22
	← rubbish left	01:22
	← EmilMedve left	01:23
	← bobmcw_ left	01:25
	← nice_burger left	01:27
	← joydivider_ left	01:28
beilabs	Hey guys, new to git. Trying to merge one file from a branch to another. Getting the following error when I try and checkout the master branch though.	01:28
	app/models/topic_centre.rb: needs merge	01:28
	error: you need to resolve your current index first	01:28
	that's from the production branch.	01:28
	← tenpaiyomi left	01:28
	← sverrej left	01:30
	← ben_h left	01:32
cehteh	beilabs: you should only merge when you have commited all pending changes .. or stashed them away	01:34
beilabs	cehteh, got it working, thanks though	01:34
	← jeffmoss left	01:36
	← pete__c left	01:36
	→ jmatthews_ joined	01:36
	→ gretch_ joined	01:38
	← jmatthews_ left	01:38
	→ jesselucas joined	01:38
	→ not-xjjk joined	01:40
	→ warthog19 joined	01:43
	← warthog9 left	01:43
	← rphillips left	01:46
	→ flaguy joined	01:47
	← jesselucas left	01:47
	→ rphillips joined	01:47
	→ EmilMedve joined	01:49
	→ tango_ joined	01:49
	→ jesselucas joined	01:49
	← johnw left	01:50
	← david_koontz left	01:51
	← xjjk left	01:53
	→ bjeanes joined	01:54
	→ jmatthews_ joined	01:56
	← EmilMedve left	01:58
	→ threeve joined	01:58
	→ sverrej joined	02:00
	→ ben_h joined	02:02
	→ EmilMedve joined	02:03
	→ ben_h_ joined	02:03
	← brendon_work left	02:05
	→ priidu joined	02:05
	→ chaines joined	02:08
chaines	hello?	02:09
	← tvw left	02:10
	→ tvw joined	02:10
	← bentob0x left	02:11
	→ phpcurious joined	02:12
	← chaines left	02:12
	→ rubbish joined	02:12
	→ damog joined	02:12
deskin	chaines: if you've a question, best to just ask	02:12
	→ Qmoto joined	02:13
	→ xjjk joined	02:17
	← warthog19 left	02:17
	→ warthog9 joined	02:17
	← cgardner left	02:17
	→ bts- joined	02:19
	← rtomayko left	02:21
Qmoto	hey guys. Anyone willing to help me figure out a git-svn problem? :)	02:21
deskin	Qmoto: ask and we'll see	02:21
Qmoto	I can't seem to dcommit anymore. It's complaining about a specific commit in svn, which only consisted of svn:prop changes.	02:22
	i can still git svn rebase and pull in new ones, though	02:23
deskin	define 'complaining'? have a paste you can show us?	02:23
Qmoto	sure, just a sec...	02:23
	$ git svn dcommit	02:24
	Committing to https://ipg-cit-eutils.csd200a.com/svn/ipg-cit-eutils/trunk/dev ...	02:24
	No changes	02:24
	68af7eb26504e324cd6e3a2d50cd0562852a2dca~1 == 68af7eb26504e324cd6e3a2d50cd0562852a2dca	02:24
	W: 68af7eb26504e324cd6e3a2d50cd0562852a2dca and refs/remotes/trunk differ, using rebase:	02:24
	:040000 040000 7d76021c10cebf4a0b5c8fdf614bec8e66734a28 197442c7bd8c29cdd11f52ffb18fbd882d1c7f50 M eu	02:24
	Current branch master is up to date.	02:24
	# of revisions changed	02:24
	before:	02:24
	683a4ed53d8859f184733eaeebc6fcb904aa1e3b	02:24
	10e85dc1c3b03b77dcabb53b26893b2d30bd56c5	02:24
	07c1f476b6a5feae410e315cae57d878e898896d	02:24
	27aad097ace7ec5db128032f29d775791463ddc7	02:24
	after:	02:24
	68af7eb26504e324cd6e3a2d50cd0562852a2dca	02:24
	683a4ed53d8859f184733eaeebc6fcb904aa1e3b	02:24
	10e85dc1c3b03b77dcabb53b26893b2d30bd56c5	02:24
	07c1f476b6a5feae410e315cae57d878e898896d	02:24
	27aad097ace7ec5db128032f29d775791463ddc7	02:24
	If you are attempting to commit merges, try running:	02:24
	git rebase --interactive --preserve-merges refs/remotes/trunk	02:25
	Before dcommitting	02:25
	and sure enough, there are no tree changes in 68af7eb26504e324cd6e3a2d50cd0562852a2dca, only prop changes	02:25
deskin	hmm	02:25
	→ ferd joined	02:26
Qmoto	i mean, it looks like it first decides, "I need to rebase, because head doesn't match the remote"	02:26
	then it rebases,	02:26
	which is fine, because it didn't need to.	02:26
	but the list of commits differs...	02:26
deskin	what git version are you running?	02:26
Qmoto	git version 1.6.0.4.761.g47577	02:27
	however, I turned off things like the git-svn id messages,	02:27
	because I'm the only one on my team using git.	02:27
	dunno if that matters	02:27
	← jeckel left	02:28
deskin	you mean you're using --no-metadata?	02:28
	← not-xjjk left	02:28
Qmoto	right	02:29
	← joydivider left	02:29
Qmoto	(btw, I have a harder time helping myself with git-svn, my perl-fu is weak. I can't figure out how and where it tracks remotes, so it's hard for me to see what it's usually trying to do)	02:31
deskin	have you tried dcommitting again? It looks like it did rebase to include that one commit...	02:31
	← jstad left	02:31
	→ pyb2503 joined	02:31
Qmoto	yes, it's done this all day.	02:32
	← damog left	02:33
Qmoto	always complains of the same commit.	02:33
deskin	perhaps try dcommit --no-rebase? Not sure what it's supposed to do, but since you seem to have trouble involving rebases...	02:34
	is that commit one of your local ones which needs to end up in svn, or one from svn?	02:34
	← jesselucas left	02:35
	→ gottesmm joined	02:35
	→ jeckel joined	02:35
Qmoto	it's one from svn (i think, let me check)	02:35
	← xenoterracide left	02:35
	← rubbish left	02:35
	→ alikins joined	02:35
Qmoto	heh, new error msg (when trying --no-rebase):	02:36
	→ xenoterracide joined	02:36
Qmoto	Committing to https://ipg-cit-eutils.csd200a.com/svn/ipg-cit-eutils/trunk/dev ...	02:36
	Attempting to commit more than one change while --no-rebase is enabled.	02:36
	If these changes depend on each other, re-running without --no-rebase may be required. at /usr/local/libexec/git-core/git-svn line 447.	02:36
	No changes	02:36
	68af7eb26504e324cd6e3a2d50cd0562852a2dca~1 == 68af7eb26504e324cd6e3a2d50cd0562852a2dca	02:36
	Unable to extract revision information from commit 683a4ed53d8859f184733eaeebc6fcb904aa1e3b~1	02:36
ferd	git-svn gurus: I have to play with an SVN repo with a non-standard layout (ie, no "branches" directory). How should I tell git where each of the branches are and how to track them? (I'm fairly new to Git, although I read pretty much all the docs I've found)	02:36
	← gma left	02:37
ferd	do it need multiple "svn remotes" ?	02:37
	bjeanes → bjeanes\|away	02:37
Qmoto	@deskin: the one it's complaining about is definitely one it's already pulled in, not one I need to check in.	02:37
	bjeanes\|away → bjeanes	02:37
deskin	ferd: init one with --stdlayout, then look at the .git/config file and hack the fetch= lines to point wherever you want	02:37
Qmoto	@deskin: think it's a weird enough corner case to go to the mailing list?	02:37
deskin	you can have several fetch lines for a single remote	02:38
	→ jesselucas joined	02:38
deskin	Qmoto: well, I'm not seeing any progress	02:38
	you are rebased onto remotes/trunk?	02:38
Qmoto	yes	02:38
	← coderdad left	02:38
Qmoto	i have exactly one commit I'm trying to get in, and the one which had "no changes" was several commits ago.	02:39
	anyway, I'll try to put together an email then. Thanks for taking a look!	02:40
	← bieneff left	02:40
ferd	@deskin: Thanks. I tried that trying to infer from it, but I got a simplistic fetch line	02:41
	didn't know I could have multiple fetch= lines for an svn-remote	02:41
deskin	gotcha	02:41
	← scientes left	02:42
	← Qmoto left	02:42
	← spearce` left	02:43
	← ankit9 left	02:43
	→ scientes joined	02:44
	← foca left	02:46
	← kalikiana left	02:49
	→ lady_bulgari joined	02:49
	← stephens left	02:51
	→ foca joined	02:52
	← gottesmm left	02:52
	← Tv left	02:53
	→ digitalhobbit joined	02:53
	→ nice_burger joined	02:53
	→ antijava joined	02:55
	lady_bulgari → WWW-Veezy-NET	02:55
antijava	Since 1.6.0.4 I haven't been able to successfully 'make test'. t9100-git-svn-basic.sh fails with a bunch of errors. This is on OS X 10.5.5. Anyone else see this?	02:57
	← WWW-Veezy-NET left	02:57
	← pigeon left	02:57
	→ WWW-Veezy-NET joined	02:57
	→ pete__c joined	02:57
	→ pigeon joined	02:58
	← digitalhobbit left	02:58
	← WWW-Veezy-NET left	02:58
	→ WWW-Veezy-NET joined	02:58
	← WWW-Veezy-NET left	02:58
beilabs	hey guys, there is a file in the base dir called revisions.log. it contains info such as this 2008-09-15 21:08:02 test-deploy 2102 20080915122239	02:59
deskin	antijava: on linux, haven't had any errors; what tests are failing?	02:59
antijava	deskin: This is what I see: http://gist.github.com/26915	02:59
beilabs	Now I assume it's a log just noting the latest deployments, how can I continue updating this log?	02:59
	← jesselucas left	02:59
antijava	I've been using git-svn successfully up to 1.6.0.3. I'm afraid to install 1.6.0.4 until I can pass these tests though	03:00
	→ digitalhobbit joined	03:00
	→ Tv joined	03:00
deskin	huh, so many failures seems really strange	03:00
	do things pass with 1.6.0.3?	03:00
	are you building from git.git, or from tarballs?	03:00
antijava	I rolled back to 1.6.0.3 and they didn't pass either. Something must have changed in my environment...but no idea what. My installed version of git-svn still works.	03:01
	← unreal left	03:01
antijava	I'm building from the repo	03:01
	→ coderdad joined	03:01
deskin	try running t9100-git-svn-basic.sh -v, from within the t/ subdir?	03:01
	that'll show you the script output, so you'll see the error messages etc.	03:02
	→ unreal joined	03:03
antijava	ok, more detail: http://gist.github.com/26915	03:03
deskin	are you on some strange filesystem?	03:03
antijava	No, normal HFS	03:03
	HFS+ that is	03:03
	→ wrobbie joined	03:04
antijava	I did install Subversion 1.5.4 from the dmg on the tigris site...not using the Apple install. But I renamed /usr/local/bin/svn out of the way and the tests still failed	03:04
deskin	what's your svn version? that first 'git svn fetch' failure looks like the svn repo format changed	03:05
	← wrobbie left	03:05
antijava	http://www.collab.net/downloads/community/	03:05
	← digitalhobbit left	03:05
antijava	svn, version 1.5.4 (r33841)	03:05
	→ wrobbie joined	03:05
	→ jesselucas joined	03:06
	← ben_h_ left	03:06
	→ [1]intripoon joined	03:06
	→ ben_h_ joined	03:06
antijava	the svnrepo/db/format shows version 3	03:07
deskin	I'd try downgrading svn, or backing out to whatever previous version you had	03:07
	I think the error is coming from the SVN.pm module, not git-svn itself	03:07
antijava	Hmm. if I do /usr/bin/svnadmin create /tmp/foo it is version 2. So it seems to not like the newest release of svn.	03:08
	→ digitalhobbit_ joined	03:09
deskin	perhaps the latest SVN perl module didn't get installed, or the old one is being found first?	03:09
antijava	Not sure. The dmg installs in /opt/subversion and creates symlinks to it in /usr/local	03:09
wrobbie	newbie question: I forgot from which branch I created another (local) branch. Which command can show me?	03:09
antijava	Ok, passed all tests when I renamed /opt/subversion to /opt/subversion.old.	03:10
	How does git-svn or the Perl libraries find the svn libraries?	03:10
	← ben_h left	03:10
	ben_h_ → ben_h	03:10
	← priidu left	03:10
	→ loincloth joined	03:11
antijava	seems to be loading from /usr/lib, so getting the Apple version (according to running with DYLD_PRINT_LIBRARIES=1)	03:12
	← ruthd left	03:12
	← digitalhobbit_ left	03:12
deskin	wouldn't surprise me if the .dmg didn't have the perl modules	03:13
	sorry, my internet is flaking out	03:13
antijava	deskin: Seems to have something in /opt/subversion/lib/svn-perl	03:14
deskin	perl searches in standard locations; try 'perl -e 'use foobar;'' and see what paths it prints	03:14
	well, that's fine, but if perl doesn't include /opt/subversion/lib/svn-perl, then it won't care	03:15
	→ netoman_ joined	03:15
antijava	that perl -e command just prints an error message...no search path	03:16
deskin	that's the idea	03:16
antijava	http://gist.github.com/26915	03:16
	nevermind...I'm stupid....	03:17
	typed 'user', not 'use'.	03:17
	yep, not in the search path. Guess I need to make some symlinks to put it in place	03:17
	← ferd left	03:18
	→ Cody joined	03:18
	← Cody left	03:19
deskin	I get 'can't locate foobar.pm in @INC (@INC contains: /etc/perl	03:19
	sorry, internet again :)	03:20
	that'd be a start; I'd hope the 1.5.4 perl bindings can handle the 1.5.4 repository layout :)	03:20
antijava	deskin: Seems to pass the test now. I symlinked /opt/subversion/lib/svn-perl/* to /Library/Perl/5.8.8 and it seems to be happy now.	03:20
deskin	excellent	03:21
antijava	thanks for the help! I don't know my way around Perl.	03:21
deskin	neither do I :)	03:21
	but hacking on git-svn is making me learn ;)	03:21
antijava	I should file a bug with the collab.net people to get them to put the symlink in place when they install	03:21
	I prefer Ruby...no real desire to learn Perl	03:21
deskin	not a bad idea	03:22
	→ felipe joined	03:22
deskin	symlink or just install the .pm files in the standard perl include path	03:22
	← intripoon left	03:22
	[1]intripoon → intripoon	03:22
	→ Cody joined	03:23
antijava	I'd rather symlink so that when I install new versions of Subversion they get picked up automatically	03:23
	→ not-xjjk joined	03:23
deskin	depends how you look at it; for someone installing a package, I think SOP is to put the files under e.g. /usr/lib/perl/	03:24
	but either way, the user shouldn't have to hack things to make it find those files	03:24
antijava	I know. But for some reason their philosopy is to install in /opt and then make links. They make /usr/local/bin/svn a symlink to /opt/subversion/bin/svn	03:24
deskin	which works fine, I guess	03:25
	Cody → chaines	03:25
	netoman_ → netoman	03:25
	← offby1 left	03:27
	← unreal left	03:27
deskin	chaines: did you have a question earlier?	03:27
chaines	yeah, OK. On windows, I can't seem to connect to my github repositories	03:28
	→ unreal joined	03:28
chaines	I'm using msysgit	03:28
deskin	paste an error?	03:28
chaines	Permission denied (publickey) fatal: the remote end hung up unexpectedly	03:29
	but, the problem is, if I go from the bash prompt that came with msysgit, it works	03:29
	but, from actual windows command prompt, it does not	03:30
	and, I need to be able to get it working from windows command prompt, in order to use EGit, a git plugin for eclipse	03:30
deskin	well, it's using some ssh keypair to connect; I'd guess cmd.exe doesn't set your home correctly to find the ssh key	03:30
	→ gottesmm joined	03:31
deskin	try looking at $HOME from msys shell, vs %home% I think from cmd.exe?	03:32
	deskin can't remember windows environment variable syntax	03:32
jast	windows doesn't have a HOME variable by default (the syntax is right though)	03:33
	← jesselucas left	03:33
chaines	yeah.	03:33
jast	it has HOMEDRIVE, HOMEPATH and USERPROFILE (the last one being the only useful one)	03:34
deskin	well, that doesn't stop git from querying it if it wants to	03:34
	though perhaps msysgit uses the windows analogue	03:34
chaines	yeah... let me try something.	03:34
jast	msysgit should provide the standard environment variables you have in UNIX, i think	03:35
deskin	try setting %HOME% to whatever $HOME is and see if it works from cmd.exe	03:35
chaines	hmm... setting a %HOME% enviroment variable to the same thing as $HOME didn't fix it	03:36
deskin	ach	03:36
chaines	but...	03:36
	one sec	03:36
	← xjjk left	03:36
	→ brendon_work joined	03:37
chaines	nope.	03:38
	generating it in C:\.ssh\id_rsa doesn't work either	03:38
	← antijava left	03:39
deskin	generating it? Or copying the one you're using from msys?	03:41
	are the permissions correct? (though maybe I'm thinking too much of linux here)	03:41
	and maybe it doesn't matter :/	03:41
	← coderdad left	03:41
chaines	yeah, permissions are read and write	03:42
	it shouldn't need anything more than that	03:42
	→ ceej joined	03:42
chaines	and, I regenerated, and then updated my key on gitHub	03:43
deskin	gotcha	03:43
	← jmatthews_ left	03:44
deskin	this may be something to ask in #github	03:44
	→ jmatthews_ joined	03:44
deskin	though it really is a (msys)git thing I suppose	03:44
chaines	yeah... hmm.. I'll keep trying random locations :P	03:44
deskin	that, or the mailing list	03:45
	which couldn't hurt anyway	03:45
chaines	yeah.	03:46
	← gottesmm left	03:47
	→ offby1 joined	03:47
	→ johnw joined	03:50
	← careo left	03:51
	← loincloth left	03:54
	→ loincloth joined	03:55
	← hobodave left	03:56
	→ muthu_ joined	03:59
	← dreiss left	04:00
	← unreal left	04:03
	→ coderdad joined	04:04
	→ unreal joined	04:04
	→ ben_h_ joined	04:11
	cbreak → cbreak\|zzz	04:14
	→ mediogre joined	04:15
	→ joydivider joined	04:17
	← ceej left	04:19
	← Tv left	04:21
	→ johnw_ joined	04:22
	← gretch_ left	04:24
	← foca left	04:25
	→ foca joined	04:25
	→ hobodave joined	04:27
	← johnw_ left	04:27
	← paggas left	04:28
	→ rubbish joined	04:28
	← johnw left	04:29
	← hobodave left	04:29
	→ jesselucas joined	04:29
	→ jamesish joined	04:31
	← flaguy left	04:31
jamesish	Can I generate a diff between the most recent commit and the local changes I have in my repo?	04:31
	→ gretch_ joined	04:33
	gretch_ → gretch	04:33
	→ hobodave joined	04:33
cardioid	jamesish: git diff ?	04:34
jamesish	Jeeeeezus.	04:34
	Well, colour me foolish.	04:34
bdrewery	^	04:34
jamesish	Thanks, cardioid. I just didn't even imagine it'd do it without a commit.	04:35
	offby1 searches for "foolish" in his Crayola box	04:36
jamesish	I'd offer you mine, but I already scribbled all over myself with it.	04:37
	← ben_h left	04:39
	ben_h_ → ben_h	04:39
	← jesselucas left	04:40
bdrewery	oo found a git bug, is there a place describing how to send in a patch?	04:42
	nm found it	04:43
	→ vbabiy joined	04:46
	→ jeffmoss joined	04:49
	← tvw left	04:49
	← vbabiy left	04:52
	← rubbish left	04:52
	← netoman left	04:54
	→ ashleyw_ joined	04:56
	→ gottesmm joined	04:57
	← jamesish left	04:58
	← schlort left	05:03
	← bts- left	05:05
	→ lordpil joined	05:05
	← scientes left	05:06
	← joydivider left	05:07
	→ tjafk1 joined	05:10
	← loincloth left	05:12
	← rmh3093 left	05:12
	→ schlort joined	05:12
	→ rmh3093 joined	05:15
	→ jkkramer joined	05:17
	→ scientes joined	05:18
jkkramer	i have a git-svn repo which is a clone of an svn repo. i will be converting the svn repo to git in the near future, but have some big, in-development branches in my git-svn repo that i'd like to continue to use. anyone know if there's a way i could take the git-svn repo's branches and graft them onto the new git repo? the hashes won't match so i'm not sure how to go about it	05:21
	← litage left	05:22
	← ahupp\|fb left	05:22
	→ mikeando_ joined	05:23
	← mikeando_ left	05:24
	← tjafk2 left	05:26
	← mediogre left	05:28
rtyler	is there an easy way to revert a fast-forward merge?	05:33
	despite not having a merge commit? :/	05:33
	← threeve left	05:33
	← gottesmm left	05:37
	← kukks left	05:37
	→ JensB joined	05:37
JensB	Hi everybody	05:37
	is it possible to merge two Git repositories?	05:38
chaines	hmm... if anybody is curious as to my previous problem, the only real way to get cmd.exe and the msysgit bash to point to the same location is to use plink when installing msysgit	05:39
rtyler	yes	05:39
	→ fhobia joined	05:39
chaines	or at least, that is the only way i could get it to work	05:39
JensB	s/(is it possible)/How $1/	05:40
	→ eno__ joined	05:40
	→ c0sin joined	05:42
rtyler	hm,reverting a fast-forward merge is harder than I though	05:42
JensB	rtyler, was the "yes" direected at me? if so, could you give me some starter points to STFM?	05:42
	← c0sin left	05:43
	← jkkramer left	05:45
rtyler	JensB: git pull <repo2> <ref>	05:45
JensB	d*oh	05:45
	:)	05:45
	if it's that simple I owe you a coffee	05:45
	maybe after years of CVS and SVN I just think in too complicated ways	05:46
rtyler	JensB: I'm assuming the two repositories are of the same lineage?	05:46
JensB	rtyler, they started as different repos but they contain similar data. (similar, not identical)	05:47
	one is connected to a SVN uplink (legacy). that's the one I want to keep.	05:47
	the other was a private pet project of mine which is now replacing the main project (in SVN).	05:47
	→ drobbins_ joined	05:48
JensB	I could just copy all the files to the main Git but then I'd lose my pet project's history.	05:48
	← drobbins left	05:48
	← madewokherd left	05:49
JensB	since the timeline is linear (commits on both projects don't overlap) a merge is perhaps not too complicated and I wouldn't mind a single huge change in the old repo when replacing it with the initial stage of the new project.	05:49
	(how) would this be possible?	05:51
	← niki left	05:51
	← eno left	05:51
offby1	JensB: I think I did something like that recently. I did "git commit-tree", I think, specifying the SHA1s of the two existing commits, and the existing tree that I wanted to be the result.	05:52
	In other words, I created a merge commit "by hand", without running "git merge"	05:53
	I felt like I'd advanced to the black belt of git-fu.	05:53
JensB	:-)	05:54
	hm	05:54
	would this be identical to appending all commits from repo2 to repo1? I think this is what I'd need	05:55
offby1	"appending" sounds like "rebase" to me.	05:55
JensB	yes	05:56
	← Flathead left	05:56
JensB	after reading the rebase manpage, to me too - only that I'm talking about two completely unrelated repos	05:56
	← chaines left	05:56
JensB	not brances in one repo	05:56
lordpil	jeckel what exactly do you want to combine? do they have identical files with different history, or distinct files?	05:56
	by unrelated i'll assume you mean the latter	05:56
JensB	mostly(!) identical files with different history	05:56
offby1	if they're unrelated, I'd expect a shitload of conflicts when you rebase	05:56
JensB	but the histories don't overlap	05:57
offby1	oh	05:57
jeckel	lordpil: huh?	05:57
rtyler	hmm, with git-rev-list I can get the list of fast-forward commits, but is there no way to revert, or back-merge them all?	05:57
lordpil	jeckel by jeckel i meant mr hyde	05:57
	my apologies	05:57
jeckel	lordpil: heh, np	05:58
JensB	basically I want to append the history of repo2 to repo1 - I#m ok if this results in a huge negative diff (removing lots of stuff) at first	05:58
lordpil	i think by append you really mean combine?	05:59
	← Zimm left	05:59
JensB	lordpil, yes, only that the histories never overlap (the last commit on repo1 is earlier than the first commit on repo2).	05:59
lordpil	that comment confuses me a bit. i assume you want all the development from both branches to ultimately be merged together, right? so if foo is in repo X and bar in Y (as of the last commit in each), you want both foo and bar to appear in the new repo?	06:00
JensB	yes. let me elaborate. maybe this is far simpler than I think.	06:01
lordpil	you could try this: http://search.cpan.org/~book/Git-FastExport-0.05/script/git-stitch-repo	06:01
	rtyler jumps up and down	06:01
rtyler	anybody?	06:01
JensB	I started a project in Git half a year ago.	06:02
rtyler	reverting a fast-forward merge, or reverting multiple commits in once revert	06:02
	bueller?	06:02
	→ Jacolyte joined	06:02
JensB	this project contained was what my employer wanted. Two weeks ago I stopped working on repo1 and started a _separate_ Git repo (repo2) containing how I thought it would work best. Now I want to combine the two, so that when I do "git svn dcommit" the next time on repo1, the most current revision in the uplink SVN is repo2.	06:04
	the git-stitch-repo seems to tackle a similar problem. looking at it now.	06:05
lordpil	yes, i think it's what you want, though your explanation still mystifies me :)	06:06
JensB	heh	06:06
	do ask.	06:06
lordpil	using fast-export \| fast-import you can do pretty much anything you want	06:06
	drobbins_ → drobbins	06:06
	→ harinath joined	06:07
	→ Pryon joined	06:08
	← vmiklos left	06:09
	rtyler sighs	06:09
rtyler	the one time I actually need help :p	06:09
	← nice_burger left	06:10
lordpil	i don't know the answer to your question :) i'd say no, but i always seem to be wrong when i think something doesn't exist	06:11
bobesponja	rtyler: what help do you need?	06:12
JensB	lordpil, here's what I would like: repo1 contains commits 1 2 3 4 5; repo2 contains commits 1 2 3 4. I want repo1 to contain commits 1 2 3 4 5 6 7 8 9 where 6..9 equal repo2(1..4).	06:13
rtyler	bobesponja: trying to revert a fast-forward merge (no merge commit) or just revert multiple commits at once	06:14
JensB	git pull complains about files being overwritten, so doesn't work right out of the box.	06:14
lordpil	well that's why i'm confused. if files are being overwritten, then you DO have files with the same name, right? if there are no file name overlaps you should be able to do this easily even if they overlap in time	06:15
bobesponja	rtyler: you want to erase those commits from history totally?	06:15
JensB	yes, there are lots of identical files - this is why i said the diff between repo1(5) and repo2(1) can be large	06:16
rtyler	bobesponja: they've been "shared" with origin, so we're in a bit of a pickle	06:16
	← ebzzry left	06:16
rtyler	normally I'd just reset --hard back to prior to those commits	06:17
	→ reithi joined	06:17
rtyler	`for r in `git rev-list master...master-media --since="16:00" --no-merges`; do git revert -s $r; done`	06:17
	that's what I've got right now to accomplish this	06:18
	but that's crap IMHO	06:18
	→ NemesisD joined	06:18
NemesisD	if in my master branch i do git rm filename, will that remove it from the working tree or just prevent it from being tracked (which is what i want)	06:18
bobesponja	rtyler: how about you just checkout the commit before the merge that happened, that way you keep your other commits	06:18
lordpil	if you want the history for file foo which exists in both to be the full history, you should be able to do it with fast-export/import. i don't know if stich-repo will let you or not, but it should work in theory	06:19
rtyler	bobesponja: that doesn't change anything in the "origin" repository	06:19
lordpil	another option would be to do a git format-patch for the entire second repository and apply it on top of the first	06:20
	← xenoterracide left	06:21
	→ gottesmm joined	06:23
	→ girishr joined	06:24
	→ dreiss joined	06:25
JensB	lordpil, git fast-import doesn't work - I'll try different options ... thanks	06:26
lordpil	format-patch ought to work	06:26
	→ cedricv joined	06:31
	→ chaines joined	06:31
JensB	even against a different repo?	06:31
doener	NemesisD: "git rm --cached" will only delete it from the index. Without --cached, it's also deleted from the working tree	06:31
NemesisD	thx	06:32
chaines	you forgot the k, and the bye	06:33
	or, bai*	06:33
	← ashleyw_ left	06:33
chaines	:P	06:33
	← chaines left	06:34
	→ ebzzry joined	06:36
	← girishr left	06:39
	← ajonat left	06:40
gittish	JensB: git merge -s ours is an option too	06:41
	i.e. merge repo1's version to your new version	06:42
	stitching will not show close to a "true" story	06:42
	-s ours isn't a merge per se, just connecting the histories	06:43
	← doener left	06:44
	← NemesisD left	06:50
JensB	gittish, will try that, thanks	06:51
	→ joydivider joined	06:52
	→ tritz joined	06:53
	→ mikeando joined	06:55
	mikeando → mikeando_	06:56
	← JensB left	06:58
	mikeando_ → mikeando	06:58
	→ Beket joined	06:59
	→ Arrowmaster` joined	07:00
	← Arrowmaster left	07:01
	Arrowmaster` → Arrowmaster	07:01
	← mikeando left	07:03
	← lordpil left	07:03
	→ ebzzry_ joined	07:09
	→ ashleyw joined	07:11
	← ebzzry_ left	07:11
	← ebzzry left	07:12
	psychosc1lumpf → psychoschlumpf	07:12
	← gittish left	07:13
	← joydivider left	07:19
	→ aspotashev joined	07:21
	→ mediogre joined	07:22
	← jeckel left	07:23
	← pyb2503 left	07:26
	→ morphir joined	07:27
	← tritz left	07:29
	→ Weaselweb joined	07:32
	→ elmex joined	07:33
	← Beket left	07:41
	→ girishr joined	07:43
TML	how can I diff two git trees which were pulled from the same origin but have since diverged?	07:48
shd	git diff branch1..branch2	07:49
	← jeffmoss left	07:49
shd	TML: i assume the other branch is in a separate working tree, right?	07:50
	TML: git remote add nickname PATH_TO_OTHER_DIR ; git diff nickname..HEAD	07:51
	also, add "git remote update" in between the two commands	07:51
	correction, git diff nickname/master..HEAD	07:52
	→ eipipuz joined	07:53
	→ mcella joined	07:53
	← bronson left	07:53
TML	yes, two working trees - I'll take a peek at " git remote add " in the docs, thanks	07:53
shd	TML: remotes are the preferred way of tracking other repositories	07:53
	TML: the remote syntax is tightly integrated into all commands	07:53
	TML: you can for example say: "git pull nickname master"	07:53
eipipuz	hi, Anyone that can help me setup github with git for windows? It's my first time with this technology, I come from hg.	07:54
	redinger → redinger_away	07:54
	← girishr left	07:54
TML	shd: I started with a "git-svn" tree, am trying to leave SVN behind, but want to bring my pending changes in the svn-based working tree into my shiny new git-tree	07:55
thiago_home	TML: commit then cherry-pick	07:56
	→ girishr joined	07:56
	→ mczepiel joined	07:56
	← dreiss left	07:57
TML	thiago_home: Won't that give me a bunch of git-svn-id stuff in my new tree, though?	07:57
	→ foca_ joined	07:58
thiago_home	commit to Git	07:58
	not to SVN	07:58
	then you can cherry-pick from one Git checkout to the other	07:58
	← aspotashev left	07:59
	madduck waves to thiago_home	07:59
madduck	or I am mistaking you for someone else actually. :)	08:00
TML	thiago_home: Right, but won't cherry-picking from the SVN tree "corrupt" my new git tree with associated git-svn-id's that I cherry-picked?	08:00
	→ girishr_ joined	08:05
	→ bentob0x joined	08:06
	← girishr left	08:11
	→ drizzd_ joined	08:12
	← girishr_ left	08:12
	→ Flathead joined	08:13
	← rvsjoen left	08:13
	→ girishr_ joined	08:15
	← foca left	08:15
thiago	TML: don't cherry-pick from SVN	08:17
	first you update SVN	08:17
	← muthu_ left	08:17
thiago	then you cherry-pick what you had in your other repository but didn't send to SVN	08:17
	then you forget SVN	08:17
	→ bsnyder joined	08:17
	→ tritz joined	08:21
	← coderdad left	08:21
	→ ben_h_ joined	08:23
	→ girishr____ joined	08:26
	← drizzd left	08:27
	← bobesponja left	08:29
	→ lucsky joined	08:29
	← a-atwood left	08:29
lucsky	'morning	08:29
	→ a-atwood joined	08:29
	→ exon joined	08:32
	← eipipuz left	08:32
	← scientes left	08:32
	← girishr_ left	08:33
	← hobodave left	08:33
	← bjeanes left	08:35
	→ girishr_ joined	08:36
	→ arekm joined	08:36
phpcurious	hi, newbie here. where am i supposed to put gitignore files ? is it inside my local repository or inside .git folder?	08:39
	→ _kid_ joined	08:40
lucsky	inside your repository	08:40
	← girishr____ left	08:41
lucsky	phpcurious: http://www.kernel.org/pub/software/scm/git/docs/gitignore.html	08:41
	← ben_h left	08:41
phpcurious	lucsky: thanks.	08:41
	← girishr_ left	08:43
	← ben_h_ left	08:43
	→ scientes joined	08:43
	← mczepiel left	08:43
	→ girishr_ joined	08:44
drizzd_	phpcurious: you can put ignores which you don't want to commit into .git/info/exclude	08:45
	← gambler left	08:49
	schme_ → schme	08:56
	← tango_ left	09:00
	→ gambler joined	09:04
	→ twas joined	09:05
	← girishr_ left	09:05
	← EmilMedve left	09:08
ashleyw	hey, I've just started using git again, and I'm unsure of something — I created a new repo at github, and pushed it, worked fine and my name was correct. Pushed a second commit later, and the name was incorrect?	09:09
	git config --list doesn't even show the name sent in the 2nd commit?	09:09
madduck	the "name"?	09:09
	→ hurikhan\|Work joined	09:10
ashleyw	the author	09:10
madduck	set user.name and user.email then	09:10
ashleyw	I have, thats correct	09:11
	and the name in the log is correct	09:11
	but github must be getting it from somewhere?	09:11
	ahh! I think I know whats wron	09:11
	*g	09:12
	tokkee is thinking about something like "git rebase --stash" - any thoughts about that?	09:12
madduck	tokkee: describe	09:12
	← moh left	09:13
tokkee	That would basically be a convencience shorthand for "stash; rebase; stash pop".	09:14
madduck	sounds easy. where's the patch? :)	09:14
tokkee	Well, it's not _that_ easy ;-)	09:14
	→ moh joined	09:16
tokkee	A rebase might be interrupted (e.g. by a merge conflict) and the user might do all kind of stuff then.	09:16
	That should be solved by introducing a separate namespace for stashes managed by rebase but I'm not 100% sure about that.	09:17
	→ Voker57 joined	09:17
	← scientes left	09:18
tokkee	Also, there are several "entry" and "exit" points into rebase which would have to be handled sanely.	09:18
	→ ashleyw_ joined	09:19
	← ashleyw left	09:22
	→ tvw joined	09:24
	→ pygi joined	09:24
	← phpcurious left	09:26
	← Jacolyte left	09:29
	→ phpcurious joined	09:29
madduck	true	09:31
	is there a convention/dictionary for the Signed-Of-By/Acked-By/etc. tags?	09:31
	i mean, which ones are there, and what do they actually mean?	09:31
	http://www.mjmwired.net/kernel/Documentation/SubmittingPatches#320 has some info	09:36
	for the kernel anyway	09:36
	← twas left	09:39
	← bsnyder left	09:42
bentob0x	out topic: what's the command to recursively list all files that aren't of a given type (like not .txt or not .php for instance)?	09:43
schme	bentob0x: You could use find	09:44
madduck	or zsh	09:44
	find . ! -type \*.txt	09:45
	echo */^*.txt	09:45
	echo */~*.txt even	09:45
	→ bsnyder joined	09:46
bentob0x	yes find seems to be the way to go	09:47
	I really should learn that command properly	09:47
	thx guys	09:47
	→ priidu joined	09:48
	dwmw2_gone → dwmw2	09:50
	→ botanicus joined	09:54
	→ ebzzry joined	09:58
	→ pcapriotti joined	10:00
	← sverrej left	10:01
	→ rvsjoen joined	10:04
	→ tango_ joined	10:12
	← wrobbie left	10:13
	→ nud joined	10:13
	← ia left	10:17
	→ Kbyte joined	10:17
	← fhobia left	10:17
	→ ia joined	10:17
	← iulian left	10:23
	→ sverrej joined	10:26
	← ashleyw_ left	10:28
	→ ashleyw joined	10:28
	← priidu left	10:30
	→ tjogin joined	10:33
tjogin	is it possible to make a git repo ignore a certain file, but only in a particular repo? i want just this particular repo to ignore the file, while other repos which synch with it to not ignore it?	10:34
Rhonda	tjogin: Create a .gitignore file within that particular repo	10:35
	Add .gitignore to the .gitignore file to make it ignore itself, too	10:35
	→ _joshua joined	10:35
	→ naeu joined	10:35
tjogin	hmm	10:36
	← joevandyk left	10:36
lucsky	funny, it's the second time I'm pasting this url today :)	10:36
	http://www.kernel.org/pub/software/scm/git/docs/gitignore.html	10:36
tjogin	i want to pull changes from one repo, but have it not overwrite or merge two files in my repo. thats it.	10:36
madduck	lucsky: you're a bot. :)	10:37
lucsky	tjogin: i guess you need $GIT_DIR/info/exclude	10:37
	"Patterns which are specific to a particular repository but which do not need to be shared with other related repositories (e.g., auxiliary files that live inside the repository but are specific to one user's workflow) should go into the $GIT_DIR/info/exclude file."	10:37
_joshua	so, I committed something before I was ready (my bad), is there a way to un-commit? I don't mean revert, but something where I go back to the state just before the commit happened?	10:37
tjogin	lucsky: ah, i had forgotten about what that was called	10:37
	→ koke joined	10:37
	→ joevandyk joined	10:39
tjogin	lucsky: it appears git status still considers a file listed in the exclude file as modified?	10:39
	ok, how about this instead: when i pull it says it cannot merge because two files are not up to date. is it anyway to make it just overwrite it with whatever is in the other repo?	10:41
	← tango_ left	10:42
	← scook0 left	10:42
	← glondu` left	10:43
	→ scook0 joined	10:43
telmich	tjogin: if it was added once, git keeps on tracking it	10:43
tjogin	how do i un-add it, but only in this particular repo?	10:44
telmich	tjogin: git rm --cached <file>	10:44
	→ dimsuz joined	10:44
dimsuz	hi guys! git-svn fails to dcommit newly added files with the following output: http://rafb.net/p/LKlPIJ18.html	10:46
tjogin	i get "fatal: Untracked working tree file 'filename' would be overwritten by merge."	10:46
dimsuz	is there anything i can to about it?	10:46
	tjogin: remove it	10:46
tjogin	dimsuz: uh, it needs to exist	10:46
dimsuz	tjogin: are you sure it doesnt? :)	10:47
	→ mithro_ joined	10:49
	→ parasti joined	10:50
	← gottesmm left	10:57
tjogin	wtf would cause "error: Entry 'filename' not uptodate. Cannot merge."?	10:57
	it not being up to date is the entire reason why i want to update my repo!	10:57
	← Strogg left	10:59
	→ Strogg joined	10:59
nud	tjogin: it means you have a file modified locally, and you can't have that if you want to do a merge	11:01
	git stash is probaly what you want	11:01
	→ gottesmm joined	11:01
	→ tango_ joined	11:01
	← not-xjjk left	11:01
tjogin	nud: ill look it up, thank you	11:03
	→ xjjk joined	11:03
nud	the error message should probably be changed	11:03
	→ lordpil joined	11:05
	← CodeOfficer left	11:05
	→ not-xjjk joined	11:09
	← schlort left	11:10
	→ schlort joined	11:10
	→ ph^ joined	11:11
	← phpcurious left	11:12
	→ lordpil- joined	11:12
	→ aspotashev joined	11:14
	→ kanru joined	11:19
yoka187	hmm I have merge issue with sqlite database file and I would like to tell git to replace my version of it with the one in master.. how can I do it? (db file is included in git because of our dev environment..)	11:20
	← xjjk left	11:22
	← unreal left	11:22
yoka187	so the question is, how to ditch my local file and use the one in remote master branch?	11:23
	← tritz left	11:23
	← arekm left	11:23
yoka187	anyone?	11:23
	→ arekm joined	11:24
arekm	how to see list of tracked files? (I have few tracked files in a directory with tons of files)	11:24
	→ hendry joined	11:28
hendry	how do you 'svn revert' or 'git reset' a single file? I'm getting a fatal: Cannot do hard reset with paths.	11:29
robtaylor	hendry: git checkout -- filename	11:29
	← lordpil left	11:30
hendry	robtaylor: thanks!	11:30
	← ToxicFrog left	11:32
	→ ToxicFrog joined	11:32
	→ Sho_ joined	11:33
	→ Ryback_ joined	11:35
	← aspotashev left	11:38
	← esden left	11:41
	→ Fullmoon joined	11:42
	→ priidu joined	11:42
	← bsnyder left	11:43
	← Flathead left	11:45
	← samjam left	11:46
	→ Arachnid joined	11:57
Arachnid	Is it possible to merge in only specific commits from a remote branch of my repository, rather than the whole branch?	11:57
	← tjogin left	11:57
	cbreak\|zzz → cbreak	11:59
	→ xenoterracide joined	12:00
bremner	Arachnid: clumsy way: fetch, make a new local tracking branch, rebase -i on that branch, and then merge from it	12:03
pflanze	pasky: I'm wondering about Git.pm (the ml thread), whether I should try to help the effort; do you have a pointer to Lea's suggestion or code?	12:03
Arachnid	You can merge specific CLs? (If so, can't I just merge straight from the tracking branch without a rebase?)	12:03
bremner	Arachnid: you can cherrypick individual commits	12:04
Arachnid	That's all I need. I just couldn't see that option in 'git merge' anywhere.	12:04
	→ bsnyder joined	12:04
bremner	Arachnid: man git cherry-pick	12:04
	Arachnid: or is it man git cherrypick	12:05
pflanze	pasky: also, what about people requiring parsed Git datastructures (commits etc), any consensus here? Have you noticed my intergit code?	12:05
bremner	Gitbot: why do you hate my freedom?	12:05
Arachnid	Aha, thanks. :)	12:06
	→ boto joined	12:06
	→ ankit9 joined	12:07
	← Arachnid left	12:10
	→ esden joined	12:15
	← _joshua left	12:16
	← eMBee left	12:19
	← esden left	12:19
	cbrake_away → cbrake	12:23
gitte	bremner: because Gitbot sucks	12:23
	→ esden joined	12:23
	← ankit9 left	12:23
	→ ankit9 joined	12:25
	→ ilogger2 joined	13:25
trochala	But the email address seems correct	13:25
	Also a .git/rebase-apply dir is created	13:26
samjam	madduck: are you a vampire? You're not showing up in the channel log lately http://colabti.org/irclogger/irclogger_log/git?date=2008-11-20	13:27
madduck	samjam: i am very busy and need to log off when i need to get work done	13:27
	since this cahnnel is too addictive. :)	13:27
samjam	I mean your last few comments that I see here are merely not on the channel log	13:28
	I think the logger must have "left"	13:28
madduck	no idea	13:28
	→ Niamor joined	13:30
	→ rubbish joined	13:36
numillumi	what are the implications of being "not currently in any branch"	13:38
gitte	RandalSchwartz: oh yeah, but the original goes like this: "To a hammer, everything looks like a nail"	13:39
	→ mw\|out joined	13:41
madduck	numillumi: you cannot commit, your HEAD is detached.	13:43
	pasky: can I get your permission to add vim modelines to all the tg files to ensure proper shifting/tabbing?	13:43
	pasky: vim:sw=8:noet basically	13:44
numillumi	but i just committed	13:45
	thats why i was curious	13:45
parasti	you can commit, but your commits will be "lost" when you switch to an actual branch	13:45
	→ EmilMedve joined	13:45
madduck	numillumi: well, sure, but there is no ref, so if you switch away without remembering the ID or without using reflog, the work is gone.	13:46
	→ attdn joined	13:47
numillumi	aye, i see sorry for being a newb--can I commit a file to a particular branch? i.e. since a file is locally modified while i was not on a specific branch	13:48
	→ coderdad joined	13:49
doener	if you didn't commit the changes, you can just checkout another branch, the changes are kept in your working tree/index	13:49
attdn	hi all, I'm on windows, wondering if someone can point me in the direction of a tutorial on how to merge using Git GUI.	13:50
	→ coderdad_ joined	13:50
	← coderdad left	13:51
numillumi	and if i did?	13:51
	i will rtfm	13:51
attdn	eh, nevermind, there is such a thing as a dumb question.	13:51
	← attdn left	13:52
	mw\|out → mw	13:52
	→ atheken joined	13:53
	→ wrobbie joined	13:57
	← rubbish left	14:04
	metze_away → metze	14:07
	→ FabianB_ joined	14:09
	→ bdiego joined	14:10
FabianB_	hi, when fetching via git svn fetch I get: Last fetched revision of refs/remotes/trunk was r9031, but we are about to fetch: r9023!	14:10
	can someone give me a hint on how to fix this?	14:11
	→ Zimm joined	14:11
	← atheken left	14:11
doener	hm, you're the second one to report such a problem within two days...	14:12
	which git version?	14:12
FabianB_	1.6.0.4	14:13
	→ bronson joined	14:13
FabianB_	I was on 1.5.4.5 when I initially created this repo	14:14
doener	what does "svn info" say about the latest revision for trunk?	14:15
deskin	FabianB_: did you ever update the svn remotes by fetching from a git repo directly?	14:15
doener	and did you dcommit lately?	14:15
FabianB_	deskin: yes	14:16
doener	deskin: IIRC the patch that updates the revmap incrementally is in 1.6.0.4, isn't it?	14:16
deskin	no	14:16
	it's only in master	14:16
FabianB_	doener: 9003, which is older than the above mentioned revisions	14:17
doener	deskin: ah, it's even from you :-)	14:17
deskin	doener: that's why I'm eerily good at diagnosing this situation ;)	14:17
FabianB_	hmm.. so should I clone git and build the master branch?	14:17
deskin	FabianB_: you need at least 2beec897; master or next have it	14:18
Gitbot	[git 2beec897]: http://tinyurl.com/5ltlq5 -- git-svn: do a partial rebuild if rev_map is out-of-date	14:18
deskin	or, you need to not update svn remotes directly via git	14:18
doener	FabianB_: either that, or just delete .git/svn/trunk/.rev_map.$somecrap	14:18
	(as a workaround to fix things right now)	14:18
deskin	right, that'll work too	14:18
FabianB_	then it'll recreate it	14:19
	ok, thank you, will do both	14:19
doener	yep, with the git-fetched revisions in there as well	14:19
	if you update to master, you don't need to delete	14:19
	deskin: funnily the guy yesterday actually didn't even have enough revisions in his svn repo to explain the error	14:19
deskin	but deleting .rev_map won't break things in any case, whether you upgrade git or not	14:20
	hmm	14:20
doener	so I didn't even think of this being a different bug	14:20
	well, "bug"	14:20
FabianB_	I was upgrading from 1.5 b/c I though it might be fixed in 1.6, when since I'm at it I can install from master as well (deps are built already) :)	14:23
	→ moh joined	14:23
	→ webmat joined	14:26
	→ petergunz joined	14:26
	coderdad_ → coderdad	14:27
	→ stathis_ joined	14:28
petergunz	hi, can i just checkout master/HEAD for example without cloning before remotely?	14:29
PerlJam	petergunz: sounds like you want a shallow cloen	14:30
	er, clone	14:30
	← petergunz left	14:30
	→ petergunz joined	14:32
	→ pygi joined	14:33
	← petergunz left	14:34
	→ petergunz joined	14:34
	→ ashleyw joined	14:36
	→ vbabiy joined	14:36
	← wrobbie left	14:37
petergunz	PerlJam: clone --depth,this should do it, thanks	14:37
deskin	petergunz: a possible alternative is git archive --remote, if you don't want a git repository at all but just the tree	14:38
	but that may not be enabled on the remote	14:38
	← petergunz left	14:42
	→ cannonball joined	14:43
	→ iulian joined	14:47
	→ ToxicFrog joined	14:49
	→ elpargo joined	14:57
	→ elpargo_ joined	15:03
	→ juanjoc joined	15:03
	→ xinming joined	15:03
	→ mediaslave joined	15:07
pasky	what is law of demeter?	15:08
moh	http://en.wikipedia.org/wiki/Law_Of_Demeter	15:10
madduck	pasky: i just tried to explain in the mail	15:11
	pasky: it basically means that should_do_help would have to be renamed to maybe_do_help	15:12
	pasky: and that we wouldn't be able to test it separately without mocking out do_help	15:12
	pasky: arguably, this is taking things a bit too far, I know.	15:12
	→ elpargo__ joined	15:12
	← mediaslave left	15:12
madduck	anyway, I think using POSIX getopt, albeit far from perfect, will at least make things a bit easier for us.	15:13
	→ jorgevargas joined	15:14
madduck	I will look into (a) parsing CLI with getopt (b) reworking the help infrastructure to sport short help messages for when there was an error, and long messages when explicitly asked for, and (c) standardise option parsing in the subcommands and add a standard option table to the help output	15:14
	in fact, maybe I will just break README up into .txt files and generate manpages from them, as well as extract synopsis for short help output	15:15
	this will have to wait a bit though. :/	15:15
	maybe I can find a minion in debian :)	15:16
	← elpargo__ left	15:17
	→ drewr joined	15:17
	stick → stick[R6]	15:17
madduck	and possibly move help into tg-help...	15:17
	← elpargo left	15:18
	← elpargo_ left	15:19
	→ webmat_ joined	15:20
pasky	madduck: i agree :)	15:20
	about taking things a bit too far ;)	15:21
	and about getopt too	15:21
	and README too	15:21
madduck	and the minion? :)	15:21
	and tg-help?	15:21
	→ jeckel joined	15:21
	→ aroben joined	15:23
	→ _kid_ joined	15:24
	→ willb joined	15:29
	→ mediaslave joined	15:29
	→ d0k joined	15:29
	→ FabianB joined	15:33
lack	Mildly complicated rebase question with pretty diagrams: http://rafb.net/p/ERNOOp95.html	15:34
	← jeckel left	15:34
spb	you want to rebase both D and H on top of C	15:35
PerlJam	lack: isn't what you want identical to the first example in the rebase manpage?	15:36
	← webmat left	15:36
	lack checks the manual	15:36
lack	PerlJam: The problem is that I have a branch head at both 'A' and 'C', whereas the manual just has a branch at 'C' called "topic"	15:37
	spb: And keep D and H's position relative to eachother, yes.	15:38
	→ mithraic joined	15:38
	← FabianB left	15:41
	→ jeckel joined	15:41
spb	rebase D on C, then H on D?	15:41
lack	spb: What would the second command look like? 'rebase --onto D C H'?	15:42
spb	or you could just rebase H, then recreate D's branch pointer	15:42
lack	true that.	15:42
	→ swhitt joined	15:43
	→ pillowfactory joined	15:43
	← FabianB_ left	15:47
	partha-bbl → partha	15:47
	stick[R6] → stick	15:50
	→ lucs joined	15:51
	→ cory joined	15:52
	← mithraic left	15:53
	cory → Guest69896	15:53
	Guest69896 → bantic	15:53
	→ kanru joined	15:54
	← stathis_ left	16:01
	→ jmrodri joined	16:03
	← jmrodri left	16:04
	→ rubydiamond joined	16:04
	→ loincloth joined	16:05
	→ aknowles joined	16:06
	→ spuk- joined	16:06
	→ bieneff joined	16:08
	← bieneff left	16:08
	→ bieneff joined	16:08
bieneff	PRIVMSG nickserv :identify florble	16:08
	→ earcar joined	16:08
swhitt	bieneff: oops	16:08
	aroben → aroben\|away	16:08
bieneff	duh	16:08
	heh	16:08
	← bieneff left	16:09
	→ bieneff joined	16:10
	← bieneff left	16:11
	→ bieneff joined	16:12
bieneff	clearly, i need coffee	16:12
	→ mithraic joined	16:13
	← bieneff left	16:13
	← earcar left	16:13
	→ jackdempsey joined	16:17
	← spuk- left	16:19
	→ hacim joined	16:19
hacim	i can't believe I cannot figure this out, but how do I search for any commits touching a particular file in gitk?	16:20
	i can search for paths, and I can search for commit messages that mention words, but I want to know all the changes for a certain file	16:20
deskin	hacim: gitk -- path/to/file?	16:20
hacim	deskin: heh, thats one way... i was hoping you would just say 'oh click on the whatever and pull it down to file'	16:21
deskin	I don't use gitk much, far more likely to find a command-line way for you :)	16:22
	→ spuk- joined	16:22
hacim	deskin: that would be git-log -- path/to/file?	16:22
deskin	that's the analogue	16:22
	→ stephens joined	16:23
deskin	for me, 'View -> New View' has a section where you can put in path names (file names) that sems to work fine	16:23
	→ jackdempsey_ joined	16:24
	→ Yuuhi joined	16:25
	→ SRabbelier joined	16:26
doener	hacim: you can right click a file in the lower right pane and select "highlight this only" to have the commits touching that file in a bold font face	16:26
	→ akitada joined	16:26
	→ xanonus joined	16:26
	→ dgrazier joined	16:26
doener	hacim: that also sets to search bar to "touching paths" and fills in the file name, so you can use the next/prev buttons to jump to the relevant commits	16:26
	but real filtering seems not to be doable like that	16:27
	← dgrazier left	16:27
	→ aspotashev joined	16:27
	→ bsnyder joined	16:27
hacim	thanks doener, thats closer to what I expected to happen	16:27
	→ hobodave joined	16:28
	← jackdempsey left	16:29
	→ doener joined	16:30
	→ harryjr joined	16:36
harryjr	git submodule add -b 2-2-stable git://github.com/rails/rails.git gui/vender/rails should add the 2-2-stable branch, right? but in fact it's adding master branch	16:37
	→ whee joined	16:37
	→ coderdad_ joined	16:38
	→ nice_burger joined	16:40
gitte	harryjr: probably a bug (except if 2-2-stable is identical to master).	16:44
harryjr	i'm on 1.6.0.2 ... will try to upgrade to latest	16:44
	→ alley_cat joined	16:45
	gitte does not think that would help.	16:45
	→ moccuo joined	16:45
moccuo	i was wondering if anyone does this. if its possible to make your entire drive under a git repository, would it be wise to make have your server under git? that way if something goes wrong, or you mess something up. you can revert back to a certain state of the server?	16:47
harryjr	:/	16:47
gitte	moccuo: has been tried before. We kinda do something like that with msysGit.	16:48
	→ litage joined	16:48
moccuo	cause im new, and still learning. and quiet often i screw shit up on the server. and i actually just reinstall the distro and redeploy everything.	16:48
PerlJam	moccuo: you generally want to do that sort of thing at a much lower level than git though.	16:48
moccuo	PerlJam: ah, kk	16:48
gitte	The thing is: you usually do not need a complete history.	16:49
	→ ashleyw_ joined	16:49
gitte	moccuo: normally you just need the last few changes, maybe daily snapshots for the last week, and then a few more weekly snapshots.	16:49
	→ luciddream joined	16:49
gitte	That's what this new-fangled Apple "backup" system does.	16:49
	→ bsnyder_ joined	16:50
	→ LiamH joined	16:50
	← bsnyder_ left	16:50
	→ tango_ joined	16:50
	→ marvil07 joined	16:51
	→ radarek joined	16:51
	← bsnyder left	16:51
	← harryjr left	16:52
PerlJam	gitte: are you familiar with the 10-tape backup method (used with back tapes obviously)	16:52
gitte	Nope.	16:54
PerlJam	you get one tape a day, everyday and then there's an additional friday tape (one per week) for each week in the month, and then there's a monthly tape for each month in the quarter. You rotate through these pulling off the quarterly tapes for archival	16:54
gitte	PerlJam: Meaning that the everyday-tapes wear out the quickest, right?	16:55
PerlJam	yep	16:55
	toss those after N uses.	16:55
	← coderdad left	16:56
bdrewery	not sure how to word this, but how can I easily get the commit range given 2 branch names, ie log master..branch -> log 32e312..2en12u?	16:56
	really just trying to figure out how to mark an item as resolved in my bug tracker and give a range of commits without branch names	16:56
doener	git rev-parse master..branch	16:57
	→ fhobia joined	16:57
bdrewery	thanks	16:57
doener	the output does not use the .. notation though	16:57
bdrewery	well that's fine	16:58
	→ PloenK joined	16:58
	← tango_ left	17:00
	→ careo joined	17:02
	→ tango_ joined	17:02
	← ashleyw left	17:04
	→ eipipuz joined	17:07
	← eipipuz left	17:07
	→ siprbaum_ joined	17:12
	→ kumbayo joined	17:21
	→ CodeOfficer joined	17:22
	gitte sighs about a certain poster who is too married to his setup to see where the light is.	17:25
	→ bobesponja joined	17:26
	← d0k left	17:30
vuf	OTOH, one can be so used to git that he fails to see any problems with it	17:30
	→ carllerche joined	17:32
	gitte doubts that; there are too many problems with Git -- all of them fixable, mind you.	17:34
	← kumbayo left	17:34
	→ bosie joined	17:35
vuf	s/any/some/	17:36
	→ carpeliam joined	17:36
	← coderdad_ left	17:37
carpeliam	how do i push a branch to a remote? i tried "git push" to try to push the active branch to the remote, it just says "Everything up-to-date" without actually doing anything	17:37
Voker57	--all	17:38
	or tell it branch exclusively	17:38
	→ com4 joined	17:38
carpeliam	Voker57, thanks, i saw the "--all" option but i didn't see anything for pushing a specific branch.	17:39
vuf	about the time stamps, why not just extract the tar to get them?	17:39
Voker57	carpeliam: git push <ref>	17:39
parasti	I don't think that's valid syntax... it's git push $remote $refspec	17:40
carpeliam	parasti, so if my remote was called "origin" and my branch was called "foo", it'd be 'git push origin foo'?	17:41
vuf	bingo	17:41
parasti	carpeliam: I believe so	17:41
carpeliam	that looks good, thanks	17:41
	→ askentasken joined	17:43
	← fhobia left	17:45
	→ tjafk2 joined	17:45
	→ coderdad joined	17:46
	← tango_ left	17:51
	← coderdad left	17:51
	→ Cacheaway joined	17:55
	→ botanicus joined	17:56
rtyler	quick question, I'm trying to find out the merge commit responsible for bringing a commit into a branch	17:57
	→ ia joined	17:57
aspotashev	Hi all! is it possible to run git-am without commiting? (I want to edit author/subject before commit)	18:00
cbreak	hmm... apply?	18:00
	→ tango_ joined	18:00
cbreak	aspotashev: check man git-apply	18:00
Gitbot	aspotashev: the git-apply manpage can be found at http://git.or.cz/man/git-apply	18:00
	→ ankit9 joined	18:00
	← ankit9 left	18:01
	→ ankit9 joined	18:02
	← ankit9 left	18:02
	→ ankit9 joined	18:03
	← carllerche left	18:03
	→ carllerche joined	18:03
	→ carpeliam_ joined	18:03
	← carpeliam left	18:04
	carpeliam_ → carpeliam	18:04
aspotashev	cbreak: yes, I can "git apply --index 0001-README-s-encoding-changed-to-UTF8.patch" and then commit the changes	18:05
	→ Tv joined	18:05
aspotashev	cbreak: but then I should edit everything manually - author/date/subject	18:06
	← Cacheaway left	18:06
aspotashev	cbreak: I want to extract the majority of info from the e-mail'ed patch	18:06
cbreak	you can set them in the environment vars?	18:06
	→ Cacheaway joined	18:06
aspotashev	cbreak: what vars?	18:07
cbreak	I can not recall	18:07
	hmm...	18:07
	git commit has --author and other flags	18:07
	← PloenK left	18:08
aspotashev	cbreak: yes, but it's dull work	18:08
	cbreak: (to write it manually)	18:08
cbreak	well, no other idea here	18:08
	I work solo, so I never had such problems :D	18:08
deskin	aspotashev: man git-commit-tree will list the env vars	18:09
Gitbot	aspotashev: the git-commit-tree manpage can be found at http://git.or.cz/man/git-commit-tree	18:09
vuf	rtyler: now, what is a branch ...	18:09
rtyler	vuf: magic? :)	18:10
	← ankit9 left	18:10
	→ ankit9 joined	18:11
rtyler	vuf: I'm more concerned with finding a merge commit that contained commit X	18:11
aspotashev	deskin: nonetheless I'll need to re-write the date each time I want to apply a patch	18:11
deskin	yes; merely letting you know how, if that's the route you choose	18:12
cbreak	you could randomize the date	18:12
	or just take average dates	18:12
vuf	rtyler: so, given your current tip, you want to find a merge commit that has a parent with and a parent without X	18:13
cbreak	maybe a script	18:13
rtyler	vuf: sounds about right	18:13
aspotashev	ok, but if I run git-am and "patch doesn't apply", then I get a ".git/rebase-apply" directory where I can edit the necessary info	18:14
	may be it's possible to "stop" git-am when everything's ok?	18:14
	← tango_ left	18:15
	→ rubbish joined	18:15
vuf	rtyler: okay, I don't know how to do that :). I think that even for simple cases, it is not so obvious which commit is the one you want ... but I may be wrong	18:15
	← Cacheaway left	18:16
	→ joydivider joined	18:16
	→ coldwaterlover joined	18:17
	→ dreiss joined	18:18
	rtyler nods	18:19
	← vbabiy left	18:20
	→ kukks joined	18:21
	→ pissedasken joined	18:22
	← ankit9 left	18:22
pissedasken	if anyone answered i missed it so please repost	18:22
	→ ankit9 joined	18:22
	pissedasken → pissed-at-glocal	18:23
PerlJam	Why is it that "git svn" doesn't pipe through a pager like all of the other git commands?	18:23
	→ Jacolyte joined	18:23
	← webmat_ left	18:23
	← askentasken left	18:24
	→ x_or joined	18:25
	→ cgardner joined	18:25
thiago_home	PerlJam: git svn what?	18:26
PerlJam	and why does it go to STDERR if you just say "git svn" but STDOUT if you say "git svn -h"?	18:26
	that seems weird. Why output to STDERR at all?	18:26
	→ grahal joined	18:27
	→ TGM joined	18:27
TGM	Hello	18:27
rtyler	PerlJam: because git-svn sucks :)	18:27
TGM	What's the revert for "git pul"?	18:27
deskin	PerlJam: git svn log pages for me	18:27
	← carllerche left	18:28
vuf	TGM: it might involve git reset	18:28
PerlJam	deskin: not "git svn log", just "git svn"	18:28
	deskin: the usage is more than one page, but there's not pager	18:29
	deskin: also, I just noticed that there really is no context-sensitive help in git-svn.	18:29
	(though it looks like someone thought ahead enough to make it easyish)	18:29
	→ niki joined	18:30
TGM	How can I revert a this change "git pull git://github.com/Blaymoira/mangos.git master" ?	18:33
pissed-at-glocal	im new to version control in general(but god knows i need it), how do i just create a repository?	18:33
vuf	git init	18:33
kandjar	hi there	18:33
	→ gitte joined	18:35
kandjar	I would like to dupplicate a git repository removing a folder and a set of files (every *.blah files). I heard it was possible through git filter-branch but I can't manage to find the correct command line	18:36
	can someone give me some hint?	18:36
vuf	TGM: use reset to chop off history, like "git reset --hard ORIG_HEAD" to get back to pre-pull on your branch	18:36
	→ scientes joined	18:36
	→ Cacheaway joined	18:37
Tv	kandjar: it's shown in the manpage	18:37
pissed-at-glocal	is git good for personal projects?	18:37
Tv	pissed-at-glocal: yes	18:37
	← Jacolyte left	18:38
gebi	hell yes :)	18:38
	← moccuo left	18:38
	← rubbish left	18:39
TGM	vuf: that will reset evertying else	18:39
kandjar	tv: I couldnt find a way to get rid of a specific folder accross the whole history (it didnt like --all)	18:39
Tv	--all?	18:40
	oh as in all branches	18:40
kandjar	yes	18:40
vuf	TGM: did you do work after the pull?	18:40
Tv	huh, that should work, based on the docs	18:40
	kandjar: you can always just list them manually	18:40
kandjar	it didnt like the cmd line I pass	18:41
	→ moccuo joined	18:41
Tv	kandjar: well how about you show it to us?	18:41
kandjar	let me redo it, and I ll paste the cmd	18:41
pissed-at-glocal	i have a file in /progs/ that i want to copy/add to /progs/.git/, how?	18:42
	i want it to be a part of the project	18:42
jast	git add filename, and eventually commit	18:43
	→ tarbo joined	18:43
	→ ajsharp joined	18:44
	→ Arachnid joined	18:45
kandjar	tv: just to make sure I understood the concept: git filter-branch modify the git repository and not create a new one based on the filter, right?	18:45
thiago_home	right	18:45
	→ bschindler joined	18:45
pissed-at-glocal	fatal: this operation must be run in a worktree	18:45
Arachnid	If I create and then delete a branch, will the creation and deletion be recorded in my history (eg, will they show up if I push to somewhere like github)?	18:45
pissed-at-glocal	jast^^	18:45
kandjar	tv: also the git filter-branch ... HEAD will apply the command to every changelist from the root to the current HEAD, right?	18:45
	← dreiss left	18:46
bschindler	Hi... I have created a stash with git stash save, then I did a git pull. When doing git stash apply I got a conflict. I resolved the conflict. How can I proceed with git stash	18:46
jast	Arachnid, nope. push only transmits matching refs (those existing on both sides) by default, anyway.	18:46
	bschindler, there's nothing to proceed. that's it already.	18:47
	you can drop the stash entry now if you want to	18:47
bschindler	ah, git stash doesn't do that by itself?	18:47
jast	bschindler, only when you use drop instead of apply	18:47
	pissed-at-glocal, did you read one of the tutorials? you need to create a repository. see man git-init.	18:48
Gitbot	jast: the git-init manpage can be found at http://git.or.cz/man/git-init	18:48
thiago_home	Arachnid: no, branch creation and deletion isn't part of the history	18:48
jast	uhm. that nick detection does need a bit of work.	18:48
parasti	doesn't "drop" just remove the entry? and "pop" is "apply" + "drop", no?	18:48
Arachnid	thiago_home: cool, thanks	18:48
jast	err, yes, sorry	18:48
	← ashleyw_ left	18:48
jast	bschindler, i meant pop, not drop. pop is apply + drop.	18:48
bschindler	oops... too late ^^	18:49
pissed-at-glocal	i did git init	18:49
	im in it now	18:49
	then add path-tp-file which isn not inside init	18:49
jast	well, if you've already applied the stash, drop is okay. you don't need that stash anymore, after all.	18:50
	← spuk- left	18:50
	← ankit9 left	18:50
	→ ankit9 joined	18:50
bschindler	good to know for next time - thanks!	18:50
	→ precision joined	18:51
	← Tv left	18:51
	→ askentask joined	18:52
	← pissed-at-glocal left	18:52
kandjar	wow....	18:53
	so I think I did something wrong ;)	18:53
	→ elpargo_ joined	18:53
	→ spuk- joined	18:53
kandjar	here is what I did: git clone my-repo.git	18:53
	then: git filter-branch --tree-filter 'rm -rf folder-to-delete' HEAD	18:54
	→ tango_ joined	18:54
jast	well, for one thing, tree-filter is a lot slower than index-filter for this kind of operation	18:54
	← scientes left	18:54
kandjar	theissue is:	18:54
	→ Tv joined	18:55
	→ scientes joined	18:55
kandjar	1 - I cant do it for anothewr folder (I got this error: Namespace refs/original/ not empty)	18:55
	← moccuo left	18:55
kandjar	2 - Looking at the history thgouth gitk --all	18:56
	→ spearce joined	18:56
kandjar	I have the old history + the rewritten one on top of it!	18:56
	→ atheken joined	18:57
	← ankit9 left	18:57
	→ ankit9 joined	18:58
	→ Sigma joined	18:58
	→ david_koontz joined	18:58
	→ EiNZTEiN joined	18:58
atheken	hi all, on Git for Windows, after a merge, the BEFORE, AFTER, etc. files are sticking around, what can I do to have these go away after the merge completes (both in success and failure), I'm using "DiffMerge" with the following command: "DiffMerge.exe" --nosplash --result="$PWD/$MERGED" "$PWD/$BASE" "$PWD/$REMOTE" "$PWD/$LOCAL"	18:59
	← ankit9 left	18:59
	→ gittish joined	18:59
	→ eno joined	18:59
atheken	sorry, MERGED, BASE, LOCAL, etc files.	18:59
	→ blp joined	19:00
	→ ari-_-e joined	19:01
	→ ThaDon joined	19:02
blp	When I do "git clone", I get a clone of the remote HEAD. If I want a clone of a different branch, I can then "git fetch" and "git checkout origin/<branch>". Is there a way to avoid those two extra steps if I just want to clone a given remote branch?	19:02
	← rubydiamond left	19:02
Tv	blp: the fetch is unnecessary	19:02
	blp: and you might actually want something like git checkout -b foo origin/foo	19:02
	mw → mw\|food	19:03
	← swhitt left	19:03
blp	tv: It's for an automatic build script that doesn't actually ever commit anything, so creating a local branch is superfluous.	19:03
kandjar	can someone tell me what I did wrong? or did I misunderstood something?	19:05
drizzd_	I thought there used to be an option to clone which controlled which remote branch would be checked out by default	19:05
blp	tv: Of course you're right that the git fetch is unnecessary. I don't know how I missed that.	19:05
drizzd_	oh, that's git remote add -m	19:06
jast	kandjar, filter-branch keeps a copy of all filtered branches before the filtering, for safety purposes. once you have verified the filtered branches are okay, rm -rf .git/refs/original (or move it somewhere else), then filter-branch will run again.	19:07
kandjar	move it somewhere else?	19:08
	← jorgevargas left	19:09
jast	well, rename it to .git/refs/foo or something, if you think you'll want to go back to before the filtering later on. otherwise, just remove it.	19:09
blp	tv: thanks	19:11
	← blp left	19:11
kandjar	oh I see, I just realize the two tree were not connected	19:11
	weird...	19:12
drizzd_	AFAIK filter-branch should not be trusted anyways	19:12
kandjar	I dont get how this works... I can have multiple tree of changelist without connection between each others???	19:12
jast	drizzd_, hmm? why not?	19:13
	kandjar, yeah, that's what branches do :)	19:13
kandjar	jast?	19:14
	→ ijcd joined	19:14
jast	yes?	19:15
kandjar	when I create a new branch it always attach the branch to the current changelist or to the tag/changelist number/branch number I specify	19:15
	it doesnt create a non connected brnach	19:15
drizzd_	kandjar: last I herad filter-branch was buggy and produces bogus results if you're unlucky	19:16
jast	what do you mean by "attach to the current changelist"?	19:16
	hmm, i can't say i've heard that about filter-branch	19:16
kandjar	doh...	19:17
	→ webmat joined	19:17
	→ ankit9 joined	19:17
jast	can't say i have experienced any problems with filter-branch either	19:17
kandjar	drizzd_: ok here is what I wanna do: at some point, I put my emacs script in a git repo (works great); but before that I was making the backup by hand... SO what I wanted to do it to insert these backup in the git repo	19:18
	to do so, I first created another git repo with all the backup put in it	19:18
Rhonda	While we are at it, is it possible to create an completely empty branch with git branch? One that doesn't have a specific start-point?	19:18
vuf	Rhonda: yes	19:19
jast	Rhonda, yes, though it's not recommended. the procedure is explained at http://book.git-scm.com/5_creating_new_empty_branches.html	19:19
kandjar	then what I was thinking was" create a patch list from the old repo and apply it to the old one	19:19
	→ coderdad joined	19:19
	← ankit9 left	19:19
kandjar	but at the same time, I want to clean the old repo (I used to put the compiled script in it and some non emacs related folder)	19:19
	that s why I thought about using filter-branch	19:19
	→ tritz joined	19:19
Rhonda	jast: Well, if one wants to create something that shares files but in itself is different enough, it helps.	19:20
	→ ankit9 joined	19:20
	→ ijcd_ joined	19:20
jast	Rhonda, it's your choice anyway. what does it matter if i'd never use it. :)	19:20
Rhonda	:)	19:20
	← ankit9 left	19:21
	→ ankit9 joined	19:22
	→ sdboyer joined	19:22
	← gitte left	19:23
kandjar	does that make sens?	19:24
	→ vuf joined	19:24
	← ijcd_ left	19:24
kandjar	the only thing I need to verify is that: the filtered tree matches the non-filtered tree minus the folders	19:24
jast	there shouldn't be any reason for filter-branch to have done anything else	19:25
	← ankit9 left	19:25
jast	you could write a script to verify it	19:25
	→ ankit9 joined	19:25
	→ bieneff joined	19:26
	→ nud joined	19:29
	→ ijcd_ joined	19:29
	← ankit9 left	19:33
	→ ankit9 joined	19:33
	→ ceej joined	19:34
	← ijcd left	19:35
	← ankit9 left	19:35
	→ LuboC joined	19:36
	← bosie left	19:37
	→ markelik_ joined	19:38
	← TGM left	19:39
arekm	anyone with knowledge about github.com crap? (no sane link to download tarballs it seems)	19:40
	→ olly_ joined	19:40
	← spuk- left	19:41
	→ jesselucas joined	19:43
	← LuboC left	19:43
ari-_-e	alright, so I've got stuff in the index, and I want my working directory to match it	19:44
	is there something less like plumbing that I should be using that git checkout-index?	19:44
	s/that/than/	19:44
	→ alikins joined	19:45
	→ earcar joined	19:45
	← earcar left	19:45
markelik_	arekm: you could ask over at #github	19:45
arekm	thanks	19:45
	→ earcar joined	19:45
	→ spuk- joined	19:46
	→ Sonderblade joined	19:47
	← hacim left	19:49
	→ mcella joined	19:51
	→ priidu_ joined	19:51
	→ priidu__ joined	19:52
	← priidu_ left	19:53
	← priidu__ left	19:53
	→ priidu joined	19:53
	← earcar left	19:53
	→ ankit9 joined	19:54
	← botanicus left	19:55
	← ankit9 left	19:55
	← alley_cat left	19:57
	→ earcar joined	19:59
	→ senjin joined	20:03
senjin	Hi, how can I push a branch I made locally to a remote repository so other people can track it?	20:04
	→ gilimanjaro joined	20:04
	mw\|food → mw	20:04
	← atheken left	20:04
jast	senjin, git push someremote thebranch	20:07
	→ foca_ joined	20:07
jast	ari-_-e, the second form of git checkout (e.g. git checkout .)	20:07
senjin	jast: thanks!	20:09
	← radarek left	20:09
	← cgardner left	20:09
	← vuf left	20:11
	→ smtlaissezfaire joined	20:12
smtlaissezfaire	Where's the official bug tracker? I can't seem to find anything on the community page of the wiki	20:13
jast	smtlaissezfaire, bugs are discussed on the mailing list (and it's very high-volume)	20:13
	if you want to report a bug, just send a mail to the list; people will include cc their replies to you	20:13
smtlaissezfaire	jast: No, I'm sure it's an old bug, regarding submodules.	20:14
	I just patched another program to get around it, and would like a link documenting the bug in git.	20:14
	I'll search the internets/ml for it, then	20:14
	→ anttih joined	20:14
smtlaissezfaire	jast: Thanks	20:14
jast	what's the bug?	20:14
smtlaissezfaire	well, if you have tracked files in one branch, say /foo/bar/baz, and then replace /foo/bar with a submodule	20:15
	Let me restart: You're in branch A, which has /foo/bar/baz.txt. You switch to branch B, delete /foo/bar	20:16
	then, you add a submodule in /foo/bar	20:16
	← grahal left	20:17
smtlaissezfaire	jast: Actually, I'll just send you a pastie	20:17
	← bschindler left	20:17
jast	okay	20:18
	← mcella left	20:19
smtlaissezfaire	Hmm. That's right. Seems to be fixed in 1.6.0.1, but not in 1.6.0.3. I'll try it there	20:20
	← senjin left	20:20
	→ Loof joined	20:21
	→ d0k joined	20:21
Loof	Heya, one thing I'm unclear on... I understand that filter-branch will break downstream repos, and you have to reclone... But what happens if a downstream source does a filter across the entire repo and then does a push?	20:22
jast	push will refuse to work	20:23
Loof	It will complain about non-ff	20:23
jast	and using --force will rid you of all changes in the target repo	20:23
Loof	but if you force it... what happens?	20:23
jast	simple: the target branch gets overwritten with the filtered source branch	20:23
	completely, that is	20:24
Loof	and what happens to children of that branch or people who cloned that repo?	20:24
jast	do you want to guess? :)	20:24
	→ IdahoEv joined	20:24
	foca_ → foca	20:24
Loof	I assume it's the same thing as if I ran filter on the upstream repo	20:24
jast	if by "children of the branch" you mean other branches based on that one, they don't automatically change, no. of course, those branches are now no longer connected and merging won't work properly anymore.	20:25
smtlaissezfaire	jast: I can't seem to replicate what I was experiencing before. I bet I had a dirty .git/config	20:25
jast	suffice to say that altering history will always break normal operation unless you do it in private or instruct all other involved parties how to remunge their local copies	20:25
	smtlaissezfaire, well, so are you happier or sadder now? ;)	20:26
	→ icwiener joined	20:26
smtlaissezfaire	jast: Happier. At least I know what was going on.	20:27
	jast: Although I still think submodules operate in sort of a strange way:	20:27
	http://gist.github.com/27179	20:27
	← ari-_-e left	20:28
Loof	So I guess the changing history will kill the other developers, regardless of how you go about it :)	20:28
	oh well, figured I'd see if there were a 'clean' way to do it	20:28
smtlaissezfaire	jast: check out the stuff at the end	20:28
wagle	is there a nice clean way to uninstall git?	20:28
jast	i'm looking at that now	20:28
Loof	wagle: Use your package manager? :)	20:29
jast	wagle, if you installed it via make, then no. it's not too hard to find and remove the files, though.	20:29
	Loof, what is it you're trying to do?	20:29
	→ moccuo joined	20:29
wagle	to stay up to date, yiou dont use a package manager, you make and make install	20:30
Loof	jast: Oh, someone messed up an automated process and introduced a bunch of temp files into the repo... and it didn't get caught right away	20:30
	They're across a number commits, and dev. has already proceeded	20:31
	so, I was thinking it'd be nice to get rid of them	20:31
wagle	jast, i have to clean up the mess that removing the dash forms of the commands left.. and there are several perl pod files that were editted in the install.. i have no idea how to unedit those	20:31
Loof	I don't want to make a new commit because they're across several branches	20:31
	→ bosie joined	20:31
jast	hmm.	20:31
Loof	so either a bunch of changes... or a rebase, which also will mean recloning	20:31
	so either way it's annoying	20:32
	filter or ignore seemed like the right answer	20:32
	(ignore as in actually ignore the issue, not git ignore)	20:32
	I don't know of any other real options	20:32
jast	Loof, are the new commits in the developers' local repos interspersed with upstream commits/merges yet?	20:33
	← askentask left	20:34
	→ askentask joined	20:34
wagle	getting to the moon is easy.. just jump	20:35
	← earcar left	20:36
	← markelik_ left	20:38
IdahoEv	What's the right way to go back to a previous state on the current branch, and move forward from there? Yesterday I did git checkout <commit hash>, thus reverting to that state, committed that, then continued developing. Two hours later I noticed I was on a detached head ("No Branch"), and had done 10 commits on that detached head, which has screwed up lots of things.	20:39
jast	okay	20:39
	→ nessundorma joined	20:40
jast	first thing, create a temporary branch from what you've currently got, to prevent losing it: git checkout -b foo	20:40
	next, make sure the number 10 is correct	20:41
IdahoEv	the number 10?	20:41
jast	the number of commits you created after detaching HEAD	20:41
IdahoEv	ah, ok	20:41
jast	because in the final step we'll transplant that many commits over	20:41
	next, checkout the actual branch	20:41
	this should return you to where you were originally	20:42
IdahoEv	ok, i used gitk to find the point where i diverged	20:42
jast	now, the transplantation	20:42
	git rebase --onto <target branch> foo~10 foo	20:43
IdahoEv	alrighty, fingers crossed	20:45
jast	don't worry, lots of possibilities exist to undo stuff like that	20:45
	← moccuo left	20:45
	→ gtyh joined	20:46
jast	now, the rebase may cause conflicts. that's normal and you should just fix them as described.	20:46
	← gtyh left	20:46
IdahoEv	no conflicts apparent	20:46
	this left me in the temporary ('foo') branch, is that expected?	20:46
jast	not really	20:46
IdahoEv	(i executed the command while in the master branch, did git rebase --onto master foo~4 foo	20:47
jast	it should leave you in whatever you put after the --onto	20:47
IdahoEv	and got "switched to branch "foo", current branch foo is up to date"	20:47
jast	hmm, let's see	20:47
uau	jast: your command line was wrong	20:47
Garoth	Hey, I vaguely recall syntax like, master@{last week}	20:48
	Could some one remind whe what that's all about?	20:48
uau	--onto is what the _contents_ of the result are based on	20:48
	not what branch is used to store it	20:48
jast	Garoth, you can use that to refer to the state of the branch at that point in time, e.g. foo@{1.week.ago}	20:48
	ah	20:48
Garoth	jast: so you need dots between each word?	20:49
jast	yes	20:49
Garoth	jast: and is that enabled at all times, or do you need to do something?	20:49
jast	no, that's automatic, unless you've got an ancient repository or an ancient version of git	20:49
	it uses the information you can see with git reflog show somebranch	20:50
uau	the default is to store reflog information for 30/90 days for the state of each branch and for the state of HEAD	20:51
	and prune it after that	20:51
	(30 days for unreferenced objects, 90 days if the objects would stay reachable anyway because of other references)	20:52
	→ vmiklos joined	20:52
	← drewr left	20:52
jast	okay, i guess i've been thinking in the wrong direction.	20:52
	so let's try again.	20:52
	IdahoEv, checkout foo and do a git rebase <target branch>. after that foo should contain the fixed sequence of commits.	20:53
IdahoEv	jast: and then i can merge it back into <target branch> to get those changes where i really want them	20:54
jast	yes, which will result in a fast-forward merge	20:54
	→ dreiss joined	20:55
IdahoEv	jast: yes, it worked, cool	20:55
jast	great	20:55
IdahoEv	so now that i know how to fix it ... how to avoid it in the first place?	20:55
jast	be careful about detaching HEAD :)	20:55
	→ tvw joined	20:56
jast	i've modified my shell prompt to show me the current branch at all times	20:56
IdahoEv	Say i realize the state two commits ago is where i really want to be. I don't need to erase the commits, I'm happy just applying those diffs in reverse to my working copy and committing that as a new change	20:56
	if i do git checkout HEAD~2, i will end up on no-branch	20:56
	← ajsharp left	20:56
smtlaissezfaire	jast: I know you're busy, but did you look at that pastie?	20:56
jast	so you want to get rid of two commits?	20:56
IdahoEv	no, i don't need to erase the commits form the history	20:56
	← CodeOfficer left	20:57
	→ defunkt joined	20:57
jast	smtlaissezfaire, i did, and to be honest i don't know if it's supposed to be that way	20:57
IdahoEv	often i can't safely because the've been pushed to a remote	20:57
	Say my state is A -- B --C --D --E and I decide I don't like the changes I made in E	20:57
jast	use revert	20:57
IdahoEv	and I'm happing with the history looking like A - B - C - D - E - D' where the files are in the same state in D' as in D.	20:58
jast	just so we're clear, this whole thing we did now didn't actually remove any changes, it only reordered them	20:58
IdahoEv	sure	20:58
jast	okay	20:58
smtlaissezfaire	IdahoEv: It all depends. if you've published that commit (meaning, if someone else has it), then you'll have to revert.	20:59
IdahoEv	okay, but revert does a single commit. what if i want to go back, say, 4 commits? Do I have to run revert four times, specifying each of those previous commits it reverse order?	20:59
smtlaissezfaire	If you haven't and want to pretend it was never there, you can reset --hard HEAD^	20:59
	IdahoEv: Just assume, though, that it's gone forever if you do that	20:59
defunkt	any gitosis users know how to access the name of the person doing the `git push` in the post-receive hook?	21:00
smtlaissezfaire	Yo - defunkt. Looks like you picked up a bunch of followers for me re: guillotine.	21:00
IdahoEv	smtlaissezfaire: and if I want to go back to the state several commits ago, but can't erase them with reset because they've been pushed to other people?	21:02
jast	IdahoEv, you can create one commit that undoes the effect of several others, if that's what you want	21:02
IdahoEv	jast: sure, i'm just asking how to do that	21:02
defunkt	smtlaissezfaire: heh nice	21:02
	smtlaissezfaire: looks interesting	21:02
jast	there are several ways	21:02
smtlaissezfaire	Yeah, the shell is the new thing.	21:02
	The new thing is the shell. Of course, I have no persistence, and the thing is pretty slow right now, so it won't be used for anything useful anytime soon.	21:03
	→ dmiles_afk joined	21:03
jast	for example: git checkout <whatever>; git reset --soft <target branch>	21:03
	→ CodeOfficer joined	21:03
jast	hmm, no, that doesn't actually do the trick	21:04
smtlaissezfaire	Well, you can do --soft <commit>, too	21:04
jast	soft will reset the branch though	21:05
	that's what we're trying to avoid	21:05
IdahoEv	i've never really understood soft resets anyway	21:05
jast	a soft reset basically changes what commit the current branch points to and nothing else	21:05
	it leaves the index and working tree completely unchanged	21:06
	so you reset the branch to something but keep all your stuff in the index/working tree	21:06
	nearly the opposite of what we want right now	21:06
IdahoEv	what would be ideal (for the situation i was in last night), is basically applying the last 4 commits in reverse to my working copy, so i can then commit them as a new change.	21:07
	in svn (which i switched away from only about 6 months ago), this would be done with a merge command	21:07
jast	yeah, you can use revert -n for that but you'd have to use it once per commit	21:07
IdahoEv	well, if that's what i have to do, i can do it. it's usually only a couple of changes anyway. but it seems like there ought to be an easy way to do this	21:08
jast	yeah, i just can't think of a reliable way right now	21:08
	i'd suggest using the second form of checkout but i don't think it would handle deleted files correctly	21:08
deskin	IdahoEv, jast: something like 'git rev-list --reverse HEAD~3..HEAD \| xargs git revert --no-commit; git commit -m "reverted last 3 commits"'?	21:09
jast	hmm, and i've got a low-level way	21:09
	don't tell anyone i suggested it but it should do the right thing	21:09
	git read-tree <whatever>; git commit	21:09
	this will make the next commit identical to <whatever> in content	21:10
	→ withanx joined	21:11
withanx	is it possible to do a diff only on staged files?	21:11
	→ paltman joined	21:12
	→ jfontan joined	21:12
madduck	git diff --cached	21:12
withanx	thanks	21:13
smtlaissezfaire	jast: I think I'm going to start documenting these submodule bugs, because they are really a time waster.	21:13
	I've probably spent half my day dealing with submodule issues	21:13
madduck	submodule ought to be able to handle both, svn:externals-style and the current way...	21:13
jast	i don't think the svn:externals style makes enough sense	21:14
smtlaissezfaire	what is the svn:externals-style - you mean staying current @ HEAD ?	21:14
dmiles_afk	your working on presenting a .git repo as a SVN repo?	21:14
smtlaissezfaire	I'm just looking for submodules + branches to work	21:14
jast	anyway, git can't easily remove a submodule just because it isn't present in some other branch	21:14
	doing that would remove all the config stuff in the submodule's .git dir	21:14
dmiles_afk	i need to give some people SVN access to my current GIT repo.. i been looking for a solution	21:15
jast	there's a pretty strict limit on what operations actually irrevocably destroy data, and the first form of checkout definitely isn't among those that are allowed to by default	21:15
madduck	smtlaissezfaire: yes	21:16
jast	oh, and here's my argument against svn:externals-style submodules	21:17
	suppose the external repo gets a new updated that breaks your application	21:18
smtlaissezfaire	jast: But what's the real danger of removing .git directory in the submodule? You'd be crazy to have a different branch which has specific commits in it that aren't publically accessible.	21:18
jast	suddenly you're in the position that some users of a specific revision of your project have a bug and some don't.	21:18
	← luciddream left	21:18
smtlaissezfaire	Well, if I remember correctly, svn externals had both - you could stick it to a revision, or you could run with HEAD.	21:18
jast	smtlaissezfaire, well, if you know there are no changes you can lose, feel free to rm -rf the submodule after switching to a branch that doesn't have it	21:19
doener	smtlaissezfaire: HEAD is not what you think	21:19
jast	yeah, you can bind it to a revision	21:19
doener	smtlaissezfaire: HEAD references what you have checked out	21:19
jast	which is the only sane thing to do	21:19
doener	smtlaissezfaire: what you mean is a branch head	21:19
	(in git terms)	21:20
dmiles_afk	master has a HEAD and each branch is bound to a HEAD?	21:20
jast	the design of submodules doesn't allow for an automatically updating submodule entry, by the way	21:20
magnus_	I think it makes sense to be able to both track a branch head and to lock a submodule to a specific commit	21:20
smtlaissezfaire	jast: For the most part I agree, and that's certainly true for production ready software. For alpha software, though, you may want to run with HEAD.	21:20
doener	dmiles_afk: no, there's only one HEAD in each repo.	21:20
magnus_	smtlaissezfaire: exactly	21:20
jast	HEAD always points to the branch that's currently checked out (or an arbitrary commit in the case of a detached HEAD)	21:21
	→ ankit9 joined	21:21
jast	in fact, HEAD defines what branch is currently checked out	21:21
dmiles_afk	doener and jast, ah .. so it reflects mainly what the working dirrectory would supposed to be conatining	21:21
jast	in a way, yes.	21:22
doener	smtlaissezfaire, magnus_: That makes your superproject history kind of useless. You can't look back anymore because you don't just don't get a consistent result when you check out an old revision	21:22
jast	more importantly, it regulates which branch your next commit will end up in	21:22
doener	smtlaissezfaire, magnus_: So you could just not commit at all	21:22
jast	if you really want to automatically advance a submodule, you write a script that checks and actually records the changes. this gives you the best out of both worlds.	21:23
doener	jast: rather "which branch head gets updated to reference the new commit". Just to avoid giving the impression that there's a "strict" relation between branches and commits	21:23
jast	yeah, whatever, fine with me. ;)	21:23
	hmm, that would be a nice idea for a contrib hook script	21:24
	<HEAD:contrib>	21:24
Gitbot	[git contrib]: http://tinyurl.com/6eb552 [tree]	21:24
dmiles_afk	i have to give eaitehr CVS or SVN access to my javaCyc.git repo .. CVS is posible but not SVN right?	21:24
jast	i love Gitbot	21:24
	dmiles_afk, that's the way it is, currently, yes	21:25
doener	dmiles_afk: right	21:25
dmiles_afk	how sone until SVN?	21:25
	soon	21:25
doener	Once you've written it	21:25
	;-)	21:25
dmiles_afk	they are actually my vendor .. they dont want to learn git	21:25
	ah .. i was going to see if a reply system might work as some manyual sync	21:26
	reply/replay	21:26
	← ankit9 left	21:26
dmiles_afk	i was thinking smtlaissezfaire was writting it maybe	21:26
doener	do you need bidirectional interaction? Or just read access for svn?	21:27
smtlaissezfaire	writing what?	21:27
dmiles_afk	well i'd like to throw my current git with ll the versions into there just so they feel like i am actualyl doing stuff as well	21:27
jast	a git to svn gateway	21:27
	i suppose it would be possible if you threw away merge information (hmm, or supported svn 1.5's new-fangled half-decent merge support)	21:28
smtlaissezfaire	dmiles_afk: writing what?	21:28
dmiles_afk	bidirrectional would be nice	21:28
jast	eww	21:28
doener	dmiles_afk: you could look into git-svn's fetch/set-tree mode of operation. That's a bit harder to learn/use than the dcommit/rebase stuff, but for your use-case, it might be enough	21:28
dmiles_afk	smtlaissezfaire: a git to svn gateway	21:28
smtlaissezfaire	Oh, for submodules?	21:29
	Oh. no way. I'm looking to fix up regular submodule support for git.	21:29
dmiles_afk	smtlaissezfaire, i thought thats why submodules were a topic	21:29
doener	dmiles_afk: the nice thing about set-tree is that the svn history is separated from your git history so merges don't get fscked up	21:29
jast	git-svn doesn't support submodules, no	21:29
	→ johnw joined	21:30
jast	i'm not sure if there's anything to fix	21:30
dmiles_afk	doener, yeah git-svn just might be enough	21:30
smtlaissezfaire	jast: There certainly is.	21:30
jast	what, then? having git automatically delete submodules isn't a fix, for a start	21:31
	→ jackdempsey joined	21:31
smtlaissezfaire	I'd love to just send you a repository, unfortunately it's all closed source.	21:31
	jast: No, that's a feature request, not a bug	21:31
jast	hmm, didn't we have a repo data scrambler somewhere?	21:31
	apparently not	21:32
smtlaissezfaire	jast: I'm actually compiling a list of things.	21:32
jast	there's been a bit of discussion about submodules on the list iirc	21:32
smtlaissezfaire	Yeah, maybe I'll sign up.	21:33
	→ vivien joined	21:33
jast	it's extremely high-traffic though	21:33
smtlaissezfaire	I can tell you this much: A lot of people have been turned off by them. One is even thinking about swearing off git because of some submodule issues he ran into.	21:33
	And after today, I can't blame him.	21:33
jast	some people use the usenet gateway	21:33
	i guess it would help if i knew the use case and the problems in it	21:34
doener	jast: what? A year of git@vger is just about 2 month of linux-kernel@vger ;-)	21:34
	s/month/months/	21:34
jast	even so, the git list is high-volume	21:35
	← d0k left	21:35
smtlaissezfaire	jast: I've got an example. Just need to remove some names to protected the innocent	21:35
doener	hm, I'm not subscribed to that many lists, but I always felt like git@ is rather "quiet". But maybe the stuff there just comes in batches, that would explain my impression	21:36
smtlaissezfaire	jast: Here's one example: http://gist.github.com/27218	21:36
doener	(ok, it's a lot more active than the fvwm lists, but hey, it's fvwm :-))	21:36
	← jackdempsey_ left	21:37
jast	what about that? i see nothing related to submodules there	21:38
vivien	What is the best GUI for git?	21:38
doener	xterm	21:38
smtlaissezfaire	jast: Well, the stuff in vendor/plugins/calendar_helper used to be a tracked file, and now is a submodule.	21:38
	That's the 3rd to last line of the second pastie	21:38
jast	second pastie?	21:39
	all i see is a clone command and a checkout command	21:39
vivien	Imagine I want to get a graphical diff (say in kdiff3) between two commits (for one file, or for the whole repository). How can I do that?	21:39
doener	jast: the lower box	21:40
	← bronson left	21:40
doener	jast: at first I thought that was just an input text box, like on pastebin.com :-)	21:40
smtlaissezfaire	jast: There are two files in that pastie. It's the second file.	21:40
jast	oh.	21:40
	well, they looked similar. :)	21:40
vivien	In kdesvn, it is easy: you mark the revision to compare and it launches kdiff3 (or any other external diff).	21:40
jast	so what's the situation leading to this? calendar_helper became a submodule?	21:41
smtlaissezfaire	Yeah. Although the strange thing is that those are both fresh checkouts of the same repos.	21:42
	Albeit on different architectures, with different git versions	21:42
	→ ph^ joined	21:42
deskin	vivien: in the case of a conflicted merge which needs resolution, you can use git-mergetool which will use kdiff3 or something else to let you resolve the merge conflicts	21:43
vivien	deskin: Thanks. But that's just for a mere diff between to revisions in the past. No merging involved.	21:43
	← aspotashev left	21:45
vivien	deskin: I meant "between two revisions" of course	21:45
	→ ankit9 joined	21:45
jast	smtlaissezfaire, so which one of them behaves as you'd like it to?	21:45
smtlaissezfaire	Well, the first.	21:46
	bremner swore off submodules, rather than git	21:46
	← xanonus left	21:46
	← askentask left	21:46
deskin	vivien: I'm aware; I don't know of a good way to do what you want though. You could try defining a kdiff3 external diff driver, and using it with gitattributes	21:47
doener	vivien: http://kerneltrap.org/mailarchive/git/2008/2/13/843394	21:47
	→ askentask joined	21:47
doener	vivien: only thing I happen to know about	21:47
	← defunkt left	21:47
doener	vivien: I don't use kdiff3 myself, I'm happy with a plain "git diff" and an occassionaly run of vimdiff	21:48
vivien	doener: kdiff3 gives you more context. And lines side by side. The plain "git diff" is poor (in terms of information) compared to the graphical view. Otherwise, I use only command line. GUI is very useful for diffs, IMHO.	21:49
	→ defunkt joined	21:50
vivien	doener: Thanks for the link. I have actually wrote a similar script, with more options.	21:50
	→ zumbrunn joined	21:50
doener	Well, as I said, I'm happy with the plain unified diff. It's maybe two times a year that I need more context, and then vimdiff does the job just fine (and I can use vim to fix up the stuff! :-))	21:51
vivien	doener: But this kind of script is inefficient. You need to find one or two SHA1s.	21:51
	→ radarek joined	21:51
doener	vivien: hm? What do you mean by "find one or two SHA1s"?	21:52
vivien	doener: I do not mean only one file, but also a full repository. And with options! kdiff3 is really good.	21:52
	doener: you need to refer to the commits. So "git log", then select one or two commits (copy/paste their SHA1) for the script. Takes time. Inefficient.	21:53
	← marvil07 left	21:53
	← juanjoc left	21:54
	← ankit9 left	21:55
doener	hm. I just do "gitk&", locate commit, <alt>+<tab>, <shift>+<ins>, optionally repeat, <enter>	21:56
vivien	doener: If the history is more than one page, you paste the first-commit SHA1, which drives you to the command line. And then the other SHA1 you would like to select is several pages above. (If not missing because you quit the log too early). This is really annoying.	21:56
	→ ankit9 joined	21:56
	→ ciskje joined	21:56
vivien	doener: middle click instead of shift+ins, I guess. But this is slow.	21:56
doener	for me, that's a lot faster than moving my fingers to use the trackpoint (using a mouse is totally inacceptable when talking about efficiency, as far as I'm concerned ;-))	21:58
vivien	doener: in kdesvn, you select directly in the log. And you can change one of the commits just as fast. And you can choose one file, or the whole repository.	21:58
	how do you select the commit then?	21:58
	I mean get the SHA1	21:58
doener	gitk automatically puts the hash into the paste buffer when you select the commit	21:59
vivien	and even browse the logs. No mouse anywhere?	21:59
	→ ceecil joined	21:59
vivien	It does not seem to work for me.	21:59
doener	most of the time, I don't even have a mouse connected at all. The trackpoint on my keyboard is enough, all I need it for is moving a window every now and then	22:00
vivien	shift+ins does not give me the SHA1	22:00
doener	might be xterm (or even debian xterm) specific, I use it for too long to remember	22:01
vivien	( I am keyboard guy. But I am first searching for efficiency, which keyboard usually gives. )	22:01
	I use konsole	22:01
	C-y does not give the SHA1 either	22:02
	← gittish left	22:02
doener	never used konsole, can't help there	22:02
	my desktop basically consists of fvwm, xclock, firefox (vimperator) and lots of xterms	22:03
	→ bschindler joined	22:04
doener	anyway, we're getting off-topic, and I'm just saying that I'm happy with what I have. Not saying that you should feel the same.	22:05
	← nice_burger left	22:05
	← askentask left	22:05
bschindler	Hi... Was there a bug with git-format-patch? I created a patch using it, went over to another tree with exactly the same content wanted to apply it and it does not work - the line line numbers are wrong :o	22:05
	→ askentask joined	22:05
vivien	doener: don't you miss shortcuts with firefox? I use konqueror because it can be tuned a lot	22:06
doener	vivien: with the vimperator extension, it feels a lot like vim. no mouse required at all	22:06
	and lots of shortcuts, of course	22:07
vivien	doener: ok, good to know. Well firefox is quite heavy and slow (even 3.0) compared to konquerr, but it could be interesting to test that.	22:07
doener	might be superstition, but to my, it feels a bit faster with vimperator than without it. But it's still a memory hog, yeah.	22:09
vivien	doener: I do not think what you propose is as efficient as a kdesvn-like behavior, but thanks for your help! I learned a few things!	22:10
doener	But vimperator made me unable to switch. I occassionaly need to test some webstuff with opera, and I feel a lot like "epic fail" then.	22:10
jast	yeah, speed and compactness definitely isn't the prime reason for using firefox	22:10
vivien	deskin: Thanks to you too!	22:10
deskin	any time	22:10
	← tritz left	22:10
vivien	konqueror has a lot to propose	22:11
jast	it lacks the extensibility via addons	22:11
	the addons repository for firefox is just amazing	22:11
vivien	but maybe it is better under kde than without the libs shared with other applications	22:11
jast	and it's about the only reason i use firefox	22:11
	→ eshear joined	22:11
vivien	like what extensions?	22:11
jast	that and konqueror can't deal with cookies the way i like	22:12
schme	pfft	22:12
	just buy more RAM and a faster CPU.	22:12
vivien	:-)	22:12
jast	adblock with easylist, cookie editor, dom inspector/firebug, downthemall!, download statusbar, greasemonkey, switchproxy, web developer toolbar	22:12
	there are somewhat similar extensions for konqueror but none of them come close in terms of power	22:13
vivien	I see. I can understand that.	22:14
	Are there web shortcuts like in konqueror?	22:14
	E.g., you type "gg:word" and konqueror launches a google search on "word"	22:14
jast	yes	22:15
schme	Personally I find using the trackball for webbrowsing a very nice pause from using the keyboard.	22:15
vivien	or "dict:word" and it launches the dictionnary	22:15
jast	it's part of the bookmark system	22:15
	for browsing i randomly mix mouse and keyboard	22:15
	when i read longer things i use the mouse, otherwise i use the keyboard	22:15
schme	Hmmm..	22:15
	Still nice to get yer hands off the keyboard every now and then I feel.	22:16
jast	when i do that, i call it "have a break"	22:16
	← defunkt left	22:16
schme	What a strange concept.	22:16
vivien	rsibreak tells you when to stop a bit ;-)	22:16
jast	yeah, most people don't understand it when i try to explain it	22:16
	← nud left	22:16
jast	i used rsibreak for a while	22:16
schme	I'll google up rsibreak. Maybe it is nicer than workrave.	22:17
jast	it disrupts my concentration too much	22:17
schme	Can you not turn it off?	22:17
	← Sigma left	22:17
jast	of course you can	22:17
	i did by deinstalling it	22:17
vivien	C-q	22:17
schme	:)	22:17
jast	my workflow doesn't tend to cause rsi anyway, i'm quite sure	22:17
vivien	you can suspend it also	22:18
jast	i take my hands off the keyboard whenever i'm not typing, and i really don't type all the time	22:18
	← coldwaterlover left	22:18
schme	rsibreak needs kdelibs?	22:18
	→ cgardner joined	22:18
jast	it's integrated with kde	22:18
vivien	don't know. I am under kde, so...	22:18
schme	Ah.	22:18
vivien	Is your HD full?	22:19
jast	hmm	22:19
schme	Well I'm not going there :)	22:19
jast	what do you think of kde4?	22:19
vivien	It is still beta	22:19
schme	My HD is far from full.	22:19
vivien	but it will be great in the mid-term	22:19
jast	it isn't, officially	22:19
vivien	it is, practically	22:19
jast	yeah	22:20
	→ ben_h joined	22:20
	→ dhabersack joined	22:20
jast	i'm not really happy with the kubuntu folks for forcing it on me	22:20
schme	They are?	22:20
vivien	Officially, they are quite carefull about selling it. On Kubuntu webpage, there is that link: "Is KDE 4.1 for you?"	22:20
jast	well, unless you prefer to stay with kubuntu 8.04	22:20
	← eshear left	22:20
vivien	I do stay on 8.04 too!	22:20
schme	You could just prefer not to use kubuntu ?	22:20
	→ eshear joined	22:20
jast	what else would i use?	22:20
vivien	Debian is too slow at releasing stable versions	22:21
schme	Well whatever you like.	22:21
jast	gnome? no thanks. something not based on debian? no thanks	22:21
vivien	Kubuntu has a good release calendar	22:21
jast	debian stable? stone age. debian unstable? 400 megs of update per week.	22:21
	← paltman left	22:21
schme	I see xmonad works with kde, so kde should be ok.	22:21
	That's odd. I have some boxes running debian unstable, I don't think they've been updated for weeks.	22:21
vivien	Agreed jast	22:22
jast	perhaps i should go back to living without a desktop environment	22:22
	← dhabersack left	22:22
vivien	Debian unstable is unstable	22:22
schme	But seriously. What forces you to actually use KDE on kubuntu? You could just use something else :)	22:22
jefferai	anyone know why I might get a Segmentation fault at the end of every git svn command I run?	22:22
vivien	don't want to see my hardware broken every couple of months	22:22
jefferai	schme: apparently, the same force that compels Ubuntu users to use GNOME	22:22
deskin	jefferai: it's a known problem in svn's 1.5.x perl bindings	22:23
jast	i've used sid for years, i just got tired of all the updates	22:23
schme	Hmm.. haven't seen any hardware stopping working on any of my debian machines the last 8 years or so though :)	22:23
vivien	Me too jast!	22:23
parasti	Debian testing anyone?	22:23
vivien	I had problems on a regular basis	22:23
schme	jefferai: So lazyness combined with lack of reading skill to figure out how to change the VM? :)	22:23
jefferai	deskin: fark...is it actually a problem? or does the command finish properly and just segfaults at the end?	22:23
	→ jmatthews joined	22:23
deskin	if you built svn yourself, there's a config option to disable some runtime module check to fix it	22:23
jast	testing doesn't give you security updates, i think	22:23
deskin	I think the latter, but I've not dealt with it myself	22:24
parasti	it does, kinda semi-officially	22:24
schme	Don't get me wrong here. I'm not advocating debian at all. I just find it odd that you feel you need to update it, and amused by the breaking :)	22:24
vivien	jast & schme: I will switch to Kubuntu 8.10 when I am sure khotkeys can work with it! which does not seem to be the case. What a shame.	22:24
jast	khotkeys is horribly broken	22:24
schme	What's khotkeys then?	22:24
	I'm learning tons here today. Nice irc!	22:24
jast	kubuntu doesn't contain the kde4 version that supports panel autohiding and extended panel options either	22:24
jefferai	jast: khotkeys has not been ported to KDE4 yet	22:24
	I think it may be in 4.2	22:24
deskin	here it is: configure --disable-runtime-module-search when building svn	22:25
jast	many things haven't been ported to kde4 yet	22:25
	← ph^ left	22:25
jast	though i'd be happy to see a version of khotkeys that's less broken than the one in kde3	22:25
jefferai	deskin: thanks	22:25
jast	ever tried defining three-step shortcuts?	22:25
vivien	khotkeys manages shortcuts. It is pretty advanced. You can do cool stuff with it. Well... when it works! I use the khotkeys from KDE3 under my KDE4	22:25
schme	I see.	22:26
jast	khotkeys is better than some of the competition, granted	22:26
	but i liked fluxbox and openbox a lot better in that regard	22:26
schme	Seriously though. Even my 70+ year old dad runs kubuntu 8.10. If he can do it, I'm sure you can ;)	22:27
	Aha.	22:27
	→ tcurdt joined	22:27
schme	jast: Use openbox then ;)	22:27
jast	what i hate about openbox is that it insists on forcing xml down my throat	22:28
schme	:(	22:28
	I hate it being a pain in the arse :)	22:28
jast	xml is one of the worst things ever	22:28
	especially when people claim it's suitable for human consumption	22:28
withanx	what's wrong with xml?	22:29
jast	it's infuriatingly annoying to edit	22:29
schme	:(	22:29
withanx	you'd prefer something like JSON?	22:29
jast	and it's ambiguous in that you can define hundreds of different ways of storing the same data, more so than most other formats i can imagine	22:29
schme	I prefer sexps meself ;)	22:29
	→ eddyp joined	22:29
schme	jast: That's the point I think.	22:29
jast	i almost prefer sexps too	22:30
smtlaissezfaire	A guy named schme who likes s-exps?	22:30
jast	json is okay, yaml is okay (though a bit too powerful)	22:30
smtlaissezfaire	what?!	22:30
schme	smtlaissezfaire: Ya what are the chances!	22:30
	← gilimanjaro left	22:30
vivien	<comment><first_word>XML</first_word><second_word>is</second_word><third_word><italic>very</italic></third_word><fourth_word>readable.</fourth_word></comment>	22:30
jast	i wrote this piece: http://xml.mustdie.de/	22:30
schme	hahaha.	22:31
tcurdt	Folks, I have a local change that I want to get rid of "git reset --hard changedfile" does not work ... What am I missing?	22:32
schme	oh git stuff.	22:32
	That's my cue for going out for a fag :)	22:32
halfline	tcurdt: git-checkout file	22:32
withanx	tchan: checkout /file	22:32
jast	tcurdt, reset with a file parameter doesn't do that kind of thing. you can use git checkout HEAD -- file (to take the last committed version)	22:32
tcurdt	oh	22:32
withanx	oops	22:32
tcurdt	thx!	22:32
jast	wihout HEAD it uses the staged version	22:32
withanx	git reset file; git checkout file	22:33
	:-P	22:33
	er	22:33
	git reset HEAD file	22:33
jast	reset should default to HEAD there	22:34
withanx	ah	22:34
jast	see the last example in the manpage	22:34
	by the way, that's one of the more confusing things in git, the fact that reset and checkout both do two different things each, two of which being rather similar but in different commands	22:35
	i wonder if it wouldn't be better to scrap the second form of reset and instead add a flag to the second form of checkout	22:36
	but the proponents of maximally short commands for each tasks would probably complain loudly	22:37
	← Cacheaway left	22:37
	→ Cacheaway joined	22:38
	← olly_ left	22:38
	→ moccuo joined	22:38
	→ lresende joined	22:38
jefferai	alright, got another question for you gurus....I want to checkout something from svn using git-svn, and then store my working branches in a bare remote repository I have created...I seem to have added the remote ref successfully, but I can't seem to get the branches initialized properly	22:40
	so first I do something like "git remote add -f myremote https://myserver/project.git"	22:41
	and I get Updating myremote	22:42
	← jesselucas left	22:42
jefferai	then:	22:42
	git checkout -b myversion myremote/master	22:42
	fatal: git checkout: updating paths is incompatible with switching branches/forcing	22:42
	Did you intend to checkout 'myremote/master' which can not be resolved as commit?	22:42
	what am I doing wrong here?	22:42
	← anttih left	22:42
jast	do a fetch first	22:43
	oh	22:43
parasti	the -f apparently does that	22:43
jefferai	right	22:43
jast	so when exactly did you push your changes?	22:43
Arachnid	Anyone know how I push a new branch to github?	22:43
jefferai	I didn't, yet	22:43
jast	(and how?)	22:43
jefferai	I have a bare repo on the http server	22:44
jast	well, is there anything at all in the bare repo yet?	22:44
jefferai	no...that's what I'm trying to do, get the initial stuff in there	22:44
Arachnid	Oops, never mind, found the guide. :)	22:44
jast	yeah, so do a push before you fetch/checkout	22:44
jefferai	ah	22:44
	so, make a branch	22:44
	then push it up to there?	22:44
	jefferai tries	22:44
jast	note that http fetch and push are a bit limited (and slow)	22:45
jefferai	how limited?	22:45
	← tcurdt left	22:46
	← Sonderblade left	22:50
jast	most notably it will be slower	22:51
pillowfactory	anyone familiar with gitosis?	22:52
	← nessundorma left	22:52
	→ nbrosnahan joined	22:52
	← nbrosnahan left	22:52
jast	also i don't think you can delta-compress the repository via http, so it may be considerably larger than if autogc kicked in	22:52
	pillowfactory, somewhat. what's up?	22:52
	← bantic left	22:54
pillowfactory	I just installed gitosis and had it working and it had me as a member of gitosis-admit, but with my name as lpillow@my-macbook-local and I really didn't want that so I changed it to just lpillow and renamed the keyfile to lpillow and forgot the .pub.	22:54
	← bosie left	22:54
pillowfactory	I pushed my changes and now I'm locked out because the gitosis.conf member name is out o sync with the keyfile name	22:55
	and I can't push my changes to correct it	22:55
	it is now prompting for my git@mydomain's password	22:56
	but user git doesn't have a password	22:56
	← litage left	22:56
jast	pillowfactory, you'll have to fix that on the server, e.g. as root	22:56
pillowfactory	right... that's what I thought, but what do I need to change	22:56
	← webmat left	22:56
pillowfactory	do I just need to reappend my pub key to /home/git/.ssh/authorized_keys ?	22:57
jast	yes, taking care to include the various options	22:57
	here's a sample entry:	22:57
pillowfactory	thansk... the options are missing for sure	22:58
jast	command="gitosis-serve jast",no-port-forwarding,no-X11-forwarding,no-agent-forwarding,no-pty ssh-rsa <key data here>	22:58
	substitute your username, of course	22:58
	then fix things locally and commit/push	22:58
	jackdempsey → jack_afk	22:58
	← jack_afk left	22:58
	→ moccuo_ joined	22:58
	← askentask left	22:59
pillowfactory	thanks jast!	23:00
	that did it	23:00
	I was missing the options	23:00
	← moccuo left	23:02
jast	for future reference: the first bit is the important one	23:04
wagle	what do project _____ lines mean in a git generated diff file?	23:05
wereHamster	git commit --amend doesn't seem to modify my commiter email. Is there a special flag I have to pass it?	23:05
	I configured user.name/email in ~/.gitconfig, and indeed git config --get user.email returns the correct email	23:06
smtlaissezfaire	jast: Here you go. I've show you most of these examples already	23:06
	http://github.com/smtlaissezfaire/git_submodule_bugs/tree/master	23:06
shd	wagle: can you give us an example?	23:06
	← bdiego left	23:06
pillowfactory	thanks again jast	23:06
	← mediaslave left	23:06
jast	wereHamster, see the --author flag. --amend reuses what's already in the commit, by default.	23:06
	← pillowfactory left	23:06
wagle	http://benno.id.au/android/armv4t.diff	23:07
	shd,	23:07
shd	wagle: i don't see any underscore lines there "___"	23:07
	→ solydzajs joined	23:07
jast	ever heard of placeholders? ;)	23:08
shd	wagle: oh.. now i see "project bionic/"	23:08
jast	i can't say i've ever seen that before	23:08
shd	wagle: that's strange.. i haven't seen those, it's probably just a comment that shouldn't confuse the patching tool	23:08
jast	yeah, but it would be interesting to know what put it there	23:09
	and for what purpose	23:09
shd	jast: i would guess the person who created the patch put it there manually for extra information	23:09
parasti	only place where "project appears in git.git is apparently gitweb	23:09
drizzd_	just guessing: submodules?	23:09
jast	shd, wagle said the diff was generated by git diff	23:10
dmiles_afk	git-svn work with git svn clone svn+ssh:dmiles@localhost:/protocyc/trunk	23:10
	?	23:10
wagle	apparently, it confuses git.. someone else is getting the error, not me	23:10
drizzd_	maybe 'repo' does that? I think android uses that.	23:11
jast	dmiles_afk, i don't think that's the right URL syntax but in general it should work like that, yeah	23:11
	← EmilMedve left	23:12
	→ flaguy joined	23:12
jefferai	jast...so I've been trying to push up to my repo on the server, but every single time, I get an error 22	23:12
	both with http and https	23:12
	wagle looks into repo	23:12
wagle	thanks	23:13
drizzd_	"repo diff [ project-list ]"	23:13
jast	that's some kind of curl error number, i guess	23:13
jefferai	ah wait	23:13
	man	23:13
	jefferai is stuipd	23:13
	← SRabbelier left	23:13
jast	okay, better you than me ;)	23:13
	→ Fla99 joined	23:13
Fla99	is there a way to reverse a git pull?	23:14
	← loincloth left	23:14
	→ Jacolyte joined	23:14
jefferai	sometimes, the moment you ask for help...	23:14
jast	if by reverse you mean undo, sure	23:14
jefferai	ah, ok	23:14
smtlaissezfaire	Fla99: look @ git reflog	23:14
jast	was the pull the very last thing you did?	23:14
dmiles_afk	coolm this worked: git svn clone svn+ssh://dmiles@denton/protocyc/trunk	23:14
	i was justing thinking the ssh tunnel was confusing things.. i just needed my url correct	23:15
drizzd_	I thought a trunk is where you store stuff in your care	23:16
parasti	typing while afk? woah :)	23:16
jast	drizzd_, it's a metaphor. in subversion, trunk is what you just randomly throw all your stuff into. ;)	23:17
drizzd_	jast: hey, that may just make sense	23:18
dmiles_afk	i am thinking my .git lives at the level of trunk	23:18
	my working dirrectory corrisponds to trunk that is	23:18
drizzd_	and it's easy to spell too	23:19
jast	well, a git repository can manage an arbitrary number of branches, actually	23:19
drizzd_	unless you're drunk	23:19
jefferai	ahhhhhSWEET	23:19
drizzd_	and roots even	23:19
jefferai	finally	23:19
jast	perhaps that's the real reason: it's called trunk because you can only make sense of it when you're drunk, but someone was drunk while standardizing the name.	23:19
jefferai	git-svn -> branch -> http server repo done	23:19
wagle	drizzd_, i can't really find something called repo? you have any hints on how to find it?	23:20
drizzd_	wagle: type 'repo android' into google	23:21
wagle	i did that..	23:21
	wagle does it again	23:21
drizzd_	source.android.com/download/using-repo	23:21
wagle	ok.. got it.. sorry	23:22
	→ chris2 joined	23:22
drizzd_	wagle: use www.google.com/ncr if you're not american	23:22
	→ quoin joined	23:23
jast	ncr is absolutely essential	23:23
	sadly	23:23
	i've configured my browser to open that on start, along with my actual start page	23:24
	→ aspotashev joined	23:24
jefferai	what's ncr?	23:24
parasti	redirects to google.com? hmm?	23:24
jast	it disables country-specific behaviour	23:25
	normally google auto-guesses from your location which version of it you want to see	23:25
	and it's the only way to deter it from that	23:26
wagle	ah	23:26
jast	unless you like storing googles Cookies For Eternity with that setting	23:26
	i forgot a (tm) there, i think	23:26
parasti	oh, I guess I've actually visited that URL once, long ago...	23:26
jast	i wonder if i can have that stored without giving google a unique ID for tracking me	23:27
	← vivien left	23:28
	← ceej left	23:28
	← ciskje left	23:29
	← tango_ left	23:30
	→ ceej joined	23:30
	→ tango_ joined	23:32
	← coderdad left	23:32
	→ kek joined	23:35
kek	how do i add a empty directory?	23:36
jast	by adding some dummy file inside it, e.g. .keep or .gitignore	23:36
kek	thanks	23:37
thiago_home	kek: by making it non-empty	23:38
	← moccuo_ left	23:39
	→ pyb2505 joined	23:40
	→ bgerber joined	23:40
	← alikins left	23:40
	← pyb2505 left	23:40
	→ alikins joined	23:41
	← withanx left	23:42
tango_	eh, I just git-sent a couple of patches for the xkb mailing list and the maintainer sort of got confused, he's used to using bugzilla instead 8-)	23:43
	tango_ hopes to be able to teach him how much faster git+mailing list for these things	23:43
vmiklos	well, bugzilla is a kind of mailing list :)	23:46
	→ vuf joined	23:47
	← akitada left	23:47
	→ pyb2505 joined	23:49
	← solydzajs left	23:52
jast	yeah, but without the usability	23:53
	← aspotashev left	23:53
	→ ajonat joined	23:54
	→ aspotashev joined	23:54
	← aroben\|away left	23:54
	→ earcar joined	23:55
	← cannonball left	23:57
tango_	lol	23:58

Logs Search ←Prev date Next date→

Channels Documentation