IRCloggy #git 2009-03-12

Provider of IRC logs since 2005.

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

	2009-03-12
yurinotigor	Ilari: git reset <dir> doesnt work. No messages	00:01
	Ilari: oops, my bad.... Thanks for your help!	00:02
parasti	yurinotigor: have a look at git status output next time	00:02
	← ekidd left	00:02
	→ hiwd joined	00:02
	→ jayallen_ joined	00:03
ckemmler	I have checked the contents of my id_rsa.pub file and it's the same of what's configured on my github account... WTF???	00:03
	it keeps telling me I'm palacehotel	00:03
	I don't understand how it works...	00:04
	→ cemerick joined	00:04
	→ mgarriss joined	00:04
	← jayallen left	00:04
	→ qrush_ joined	00:06
	→ Flathead joined	00:08
Ilari	ckemmler: What files does '.ssh/id*' match? Only '.ssh/id_rsa' and '.ssh/id_rsa.pub'?	00:10
ckemmler	no there are other id* files	00:10
	is that it then??	00:10
Ilari	ckemmler: Names of those files? Stuff like 'identity' or 'id_dsa'?	00:11
ckemmler	id_ckemmler_rsa, id_palacehotel_rsa, id_rsa	00:11
Ilari	ckemmler: md5sum those and see which matches the id_rsa one?	00:12
	← idahoev1 left	00:13
ckemmler	Ilari: how do I do that?	00:13
Ilari	ckemmler: 'md5sum .ssh/id_*rsa' or something like that.	00:14
ckemmler	Ilari; and, btw, I just created a new id_rsa with ssh-keygen and pasted it on my github account, retried and... no effect - ok I'll try that	00:14
	← Jzalae left	00:14
	→ bobmcw joined	00:14
	→ Jzalae joined	00:14
Ilari	ckemmler: Pasted id_rsa? You mean id_rsa.pub?	00:15
ckemmler	yes	00:15
	I don't have md5sum (mac os x leopard)	00:17
Ilari	ckemmler: Or maybe .ssh/config has some special private key for that host?	00:17
	← merlin1983 left	00:17
	← Phlogi left	00:17
ckemmler	there's nothing in config	00:17
	← Simetrical left	00:17
	← cgardner left	00:18
	← qrush left	00:18
Ilari	ckemmler: Do you use ssh-agent?	00:18
ckemmler	I don't remember what got me the ssh functionality	00:18
	yes apparently	00:18
	ssh-agent rings a bell to my computer	00:19
Ilari	ckemmler: If that id_palacehotel_rsa has been added to ssh-agent, it could very well try it as part of authentication attempt...	00:19
	→ dsal joined	00:20
ckemmler	Ilari: any idea how I can change that behaviour?	00:20
Ilari	Stuff like this is why I consider that there should be only one ssh keypair for interactive use per source account...	00:20
	ckemmler: No idea. I don't use ssh-agent.	00:20
	← lois[pac] left	00:21
ckemmler	Ilari: IT WORKED!!!	00:22
	thans so much - had to do ssh-add id_rsa	00:22
Ilari	ckemmler: ssh-add is command for manipulating keyring of ssh-agent.	00:23
	ckemmler: Also, if you add too many keys to ssh-agent, logins to remote systems can start failing...	00:23
ckemmler	so it's going to try every one of my identites thus added, or what?	00:23
	→ pjhyett joined	00:24
	→ scientes joined	00:24
spearce	ckemmler: yes.	00:24
	← boto left	00:24
ckemmler	spearce: OK - thanks	00:24
spearce	you can set specific keys for specific hosts in your ~/.ssh/config, IdentityFile keyword. I usually do that, even if i only have 2 keys on the client.	00:24
	← Khisanth left	00:25
	→ sitaramcell joined	00:25
Ilari	Why multiple keys on client? One for interactive use, others for special-purpose applications?	00:25
	→ Khisanth joined	00:25
	→ jayallen joined	00:26
spearce	Ilari: i use different keys on different hosts. my desktop has 3 keys on it. my home laptop has 8.	00:26
	← glommer left	00:26
spearce	kind of insane, i know	00:26
	← jayallen_ left	00:26
spearce	bad habit more than anything.	00:26
	→ glommer joined	00:26
Ilari	I only have two SSH keypairs, one per normal user account I have...	00:27
spearce	Ilari: i really should consolidate mine. its nuts that i have as many as i do.	00:28
	← brendon_work left	00:29
Ilari	Darn that SSH doesn't seem to support the more secure (112 or 128 bit strength) DSA2 and only supports the less secure (80 bit strength) DSA1...	00:30
	← cemerick left	00:30
	← qrush_ left	00:30
	→ jayallen_ joined	00:33
	→ qrush joined	00:33
	→ nighthwk1 joined	00:35
alvarezp	Hello	00:36
	← broken_ladder left	00:36
nighthwk1	is there a way to ignore permissions/ownership changes? I'm running into a bunch of "modified" files on cygwin git	00:37
	← Suprano left	00:37
Ilari	nighthwk1: There's config option for it to ignore those.	00:37
	nighthwk1: man git-config has pretty good list of those.	00:37
	→ broken_ladder joined	00:37
	→ Simetrical joined	00:38
	← sekimura left	00:38
	← broken_ladder left	00:39
	← johnw left	00:40
	→ johnw joined	00:40
	← f0rkit left	00:43
	← comawhite left	00:43
	← hobodave left	00:43
	→ LiamH joined	00:45
	← jayallen left	00:45
	← parasti left	00:47
	→ sitaram joined	00:47
	← sitaram left	00:47
	→ sitaram joined	00:48
	← altrux left	00:48
	← drewolson left	00:49
	→ samsoffes joined	00:50
	← sitaramcell left	00:52
	← nighthwk1 left	00:53
	→ cao joined	00:56
	← DrNick left	00:56
	← kate21de1 left	00:56
	→ DrNick joined	00:56
	→ merlin83 joined	01:00
	← hiwd left	01:01
	← greyface left	01:02
	← chris2 left	01:02
	← jackdempsey left	01:03
	← sitaram left	01:03
	→ troter joined	01:05
	→ greyface joined	01:08
	← bryanl left	01:08
	← glommer left	01:08
	← ProperNoun left	01:09
	← ckemmler left	01:10
	→ theCarpenter joined	01:10
	← theCarpenter left	01:10
	→ angerman joined	01:10
	← SRabbelier left	01:10
	→ theCarpenter joined	01:10
	→ lresende joined	01:11
	← _slacker_ left	01:11
	← Dashkal left	01:11
	← robinr left	01:16
	← osmosis left	01:16
	→ vbabiy joined	01:16
	→ gretch joined	01:18
	← jerome187 left	01:18
	← WALoeIII left	01:21
	→ EmilMedve joined	01:21
	→ comawhite joined	01:25
	← comawhite left	01:31
	→ bdimcheff joined	01:33
	→ jrmuizel joined	01:34
	→ seangrove joined	01:35
	→ cemerick joined	01:36
	← cemerick left	01:36
	← mbroeker left	01:36
	→ feanil joined	01:37
	jayallen_ → jayallen	01:42
	→ AAA_awright joined	01:42
	→ BartVB_ joined	01:42
	→ eurdk_ joined	01:42
	→ bcarlyon\|laptop joined	01:43
	→ ianmcorvidae\|alt joined	01:43
	→ webmat joined	01:45
	← kms2 left	01:45
	← madduck left	01:45
	← eletuchy left	01:45
	← nadim_ left	01:45
	← bdowning left	01:45
	← jnl left	01:45
	← f0rked left	01:45
	← hegge left	01:45
	← jpieper left	01:45
	← Weiss left	01:45
	← fonseca left	01:45
	← avtobiff left	01:45
	← jds left	01:45
	← sam left	01:45
	← ndim left	01:45
	← robtaylor left	01:45
	← dwmw2_gone left	01:45
	← feanil left	01:45
	← bobmcw left	01:45
	← univate left	01:45
	← ashleyw_ left	01:45
	← jfkw left	01:45
	← BarryCarlyon left	01:45
	← curvature left	01:45
	← unreal left	01:45
	← steveb left	01:45
	← koki left	01:45
	← Pistahh left	01:45
	← eurdk left	01:45
	← reto_ left	01:45
	← TML left	01:45
	← raz left	01:45
	← hiffy left	01:45
	← canidae left	01:45
	← lack left	01:45
	← AlexanderS left	01:45
	← jonmasters left	01:45
	← wereHamster left	01:45
	← shad left	01:45
	← sejo left	01:45
	← ianmcorvidae left	01:45
	← bronson left	01:45
	← _maks left	01:45
	← mike_m left	01:45
	← uau left	01:45
	← BartVB left	01:45
	← noth left	01:45
	← mgarriss left	01:45
	bcarlyon\|laptop → BarryCarlyon	01:45
	→ eletuchy joined	01:46
	→ nadim_ joined	01:46
	→ bdowning joined	01:46
	→ fonseca joined	01:46
	→ jds joined	01:46
	→ jpieper joined	01:46
	→ robtaylor joined	01:46
	→ ndim joined	01:46
	→ kms2 joined	01:46
	→ Weiss joined	01:46
	→ sam joined	01:46
	→ f0rked joined	01:46
	→ madduck joined	01:46
	→ avtobiff joined	01:46
	→ hegge joined	01:46
	→ jnl joined	01:46
	→ TML__ joined	01:46
	→ steveb_ joined	01:46
	→ raz joined	01:47
	→ curvature joined	01:47
	← cyphactor left	01:49
	→ feanil joined	01:50
	→ bobmcw joined	01:50
	→ univate joined	01:50
	→ ashleyw_ joined	01:50
	→ jfkw joined	01:50
	→ unreal joined	01:50
	→ shad joined	01:50
	→ reto_ joined	01:50
	→ bronson joined	01:50
	→ noth joined	01:50
	→ eurdk joined	01:50
	→ _maks joined	01:50
	→ dwmw2_gone joined	01:50
	→ mike_m joined	01:50
	→ uau joined	01:50
	→ Pistahh joined	01:50
	→ AlexanderS joined	01:50
	→ canidae joined	01:50
	→ wereHamster joined	01:50
	→ koki joined	01:50
	→ lack joined	01:50
	→ jonmasters joined	01:50
	→ hiffy joined	01:50
	→ sejo joined	01:50
	← qrush left	01:50
	→ dsal` joined	01:50
	← feanil left	01:51
	← ashleyw_ left	01:51
	← johnw left	01:51
	→ feanil joined	01:52
	→ canidae_ joined	01:53
	→ qrush joined	01:53
	→ catdude joined	01:54
	← vbabiy left	01:54
	→ mgarriss joined	01:55
	→ hobodave joined	01:56
	→ uau` joined	01:56
	→ ashleyw joined	01:57
	← seangrove left	01:57
	→ hiwd joined	01:57
	→ vbabiy joined	01:58
	← dsal left	01:59
	← mgarriss left	01:59
	← eurdk left	02:00
	← rubydiamond left	02:00
	← jchris left	02:02
	← canidae left	02:03
	canidae_ → canidae	02:03
	← uau left	02:04
	→ decomp joined	02:04
	→ facedownH joined	02:05
	← Octalot left	02:05
	← webmat left	02:05
facedownH	Can anyone recommend a tutorial/guide on git?	02:06
	← catdude left	02:06
	← feanil left	02:06
	← tongueroo left	02:07
	← Tv left	02:07
facedownH	preferrably a more recent version, I'm a current CVS user ( sadly, I know I know )	02:07
context	cvs still?	02:08
	i thought people stopped using that	02:08
	uau` → uau	02:09
facedownH	context: sadly, we at work still use it and i'm trying to find a more modern vcs	02:09
	← jonmasters left	02:09
facedownH	the majority of users are on Windows, i know there's a tortoisegit available but I'm just looking for a guide on creating a repository, etc	02:09
	→ unreal_ joined	02:09
	← sgrimm left	02:10
	← unreal left	02:11
	→ jcm joined	02:11
Ilari	facedownH: There's 'http://git-scm.com/documentation'. Hmm... It lists 'Git for Computer Scientists' but not 'Git from Bottom Up'...	02:12
	→ pjhyett_ joined	02:12
	→ hiffy_ joined	02:13
	← paulboone_ left	02:14
	← hiffy left	02:14
	hiffy_ → hiffy	02:14
	→ idahoev1 joined	02:14
offby1	facedownH: the latest msysgit works pretty well, if that helps any.	02:14
	→ bieneff joined	02:15
	← bdimcheff left	02:15
	→ jchris joined	02:15
	← spearce left	02:15
	→ ProperNoun joined	02:16
	→ qrush_ joined	02:18
	→ bdiego joined	02:18
	→ seangrove joined	02:19
	← benn_d left	02:20
	dsal` → dsal	02:21
	→ bobmcw_ joined	02:21
	→ anathematic joined	02:22
anathematic	how do I delete a branch in git?	02:22
	dw got it =)	02:22
	← tek0 left	02:23
	← stephenjudkins__ left	02:24
offby1	.oO("dw"?)	02:25
	← qhoxie left	02:26
	← pjhyett left	02:26
	→ feanil joined	02:28
	← binjured left	02:28
	← bobmcw left	02:28
	← felipe left	02:29
	→ cyphactor joined	02:29
bremner	dwbh?	02:29
	← bdiego left	02:30
dsal	Huh. git master isn't building for me on OS X.	02:30
Ilari	dsal: What error?	02:31
dsal	I just started a 1.6.2 build. I didn't see an error that made any sense or told any info. Something about perl.mk	02:31
	If 1.6.2 works, I'll bisect or something.	02:31
	→ tek joined	02:32
dsal	Yeah, 1.6.2 is good. Let me get this installed. I want to try out commit notes. :)	02:32
	← xyc left	02:32
ThaDon	dsal: commit notes?	02:33
	← qrush left	02:34
dsal	ThaDon: You can sort of append more info onto existing commits without invalidating hashes.	02:34
	← TML__ left	02:36
	← feanil left	02:36
ThaDon	dsal: interesting....	02:36
dsal	Yeah. Actually trying to find it now. I don't think it's actually in 1.6.2. :/	02:36
ThaDon	now, if I could only figure out why the second line of my commit message turns red when I'm using vim to type it in...	02:36
	← cyphactor left	02:37
dsal	Oh. It's been reverted. Damn. Looked like a good idea.	02:37
	→ tek0 joined	02:39
	← tek0 left	02:39
Ilari	dsal: It should be in pu...	02:40
	← anathematic left	02:40
	← ashleyw left	02:41
	→ cyphactor joined	02:41
dsal	Ilari: Ah, yeah, I see it there.	02:41
	← alvarezp left	02:41
	← idahoev1 left	02:41
	→ tek0 joined	02:41
	→ mithro joined	02:42
	→ CardinalNumber joined	02:43
	← ice799 left	02:43
	→ hpatwork joined	02:43
dsal	I must've had a partial build or something. I'm not getting the same build failure after a clean.	02:43
hpatwork	Is there a way to remote files deleted in the working directory from the index?	02:43
	← tek0 left	02:44
	→ Tv joined	02:44
RandalSchwartz	when you commit, it'll notice they're gone	02:45
	← flaguy48 left	02:45
RandalSchwartz	although a git rm with the right switches might hep	02:45
	help	02:45
	← univate left	02:45
	← tek left	02:46
	← aziz left	02:46
	→ univate joined	02:46
	→ Eridius_ joined	02:48
	← Eridius left	02:48
	Eridius_ → Eridius	02:48
	→ reprore joined	02:49
	← cyphactor left	02:50
	← markelikalderon left	02:50
	→ drewr joined	02:51
	→ the_undefined joined	02:52
	← qrush_ left	02:53
	← kukks left	02:55
	← brett_h left	02:55
dsal	pu looks pretty good. Keeping up with git is hard. :)	02:57
	← pjhyett_ left	02:57
	← the_undefined_ left	03:00
	← ProperNoun left	03:02
	← bieneff left	03:02
	→ felipe joined	03:14
	→ feanil joined	03:17
	← samsoffes left	03:17
	→ cyphactor joined	03:17
	→ _hrrld joined	03:20
	→ cytrinox_ joined	03:20
	→ samsoffes joined	03:20
	→ [1]intripoon joined	03:21
_hrrld	I was wondering how tools like git-gui and gitk are architechted. Do they typically shell out to the normal porcelan and parse the results or do they link with git lib-wise?	03:21
	← samsoffes left	03:21
	→ tek joined	03:22
	→ tongueroo joined	03:22
	← darwin left	03:23
dsal	_hrrld: git log --pretty=format:[stuff]	03:25
_hrrld	o.0?	03:26
	The shell out, but do custom pretty formatting to make parsing easier?	03:30
	They*	03:30
	← reprore left	03:30
	← jchris left	03:36
	← cytrinox left	03:36
	→ dbindner joined	03:36
	← intripoon left	03:38
	[1]intripoon → intripoon	03:38
	→ drewr` joined	03:40
	← merlin83 left	03:42
_hrrld	For the benefit of future searchers: quick grepping of the git-gui source does indicate lots of shelling out going on, nothing too tricky seeming.	03:42
	← _hrrld left	03:42
dsal	_hrrld: You can read it. I'm mostly speculating, though I've made tools to do visualization and find it quite easy. (and there's no libgit)	03:42
	← drewr left	03:43
	drewr` → drewr	03:43
	→ sgrimm joined	03:44
	→ slums joined	03:45
	→ dsal` joined	03:46
dbindner	So, I using git 1.5.6.3 on an Ubuntu system, and I'm noticing it takes me 11s to checkout different branches instead of the 0.2s it should take. At first I thought it was caused by lack of running git-gc (who knew?) but that didn't resolve it. So I used strace to try to see what's going on, and I believe a DNS query is failing, but I don't really know why a lookup would even be happening. Setup: The "main repository" is on a system calle	03:46
slums	i have a file which is tracked by git, and is in modified state. i run git update-index --remove --assume-unchanged file, but it still reported as modified	03:47
	how can i remove it from git status?	03:48
	← AkumaStreak left	03:51
	→ AkumaStreak joined	03:52
	→ mgarriss joined	03:55
	← dsal left	03:55
	dsal` → dsal	03:58
	← feanil left	03:58
	→ gotgenes joined	04:00
	→ ankit9 joined	04:02
	← mithro left	04:04
	→ mithro joined	04:05
	← ankit9 left	04:07
	← drewr left	04:08
	→ ankit9 joined	04:08
	← intripoon left	04:08
	→ ben_h joined	04:13
	→ merlin83 joined	04:13
offby1	doesn't "git status" itself print a helpful message explaining just that?	04:14
	I thought it did	04:14
	← tek left	04:15
	← mgarriss left	04:15
Fissure	he wants to have files with changes, but not have them show in status	04:15
	→ tek joined	04:15
	→ jony joined	04:15
offby1	oh	04:15
Fissure	slums: i think you might need to get it back to a clean state before doing --assume-unchanged	04:16
	also, unless it's a speed issue, using that setting means you're Doing It Wrong	04:16
	← ankit9 left	04:16
	→ ankit9 joined	04:16
	← AkumaStreak left	04:17
offby1	gasp	04:19
	→ theCarpenter1 joined	04:19
	→ sol1dus joined	04:20
	← theCarpenter1 left	04:21
sol1dus	is it possible to undo git-clean?	04:21
	ianmcorvidae\|alt → ianmcorvidae	04:21
offby1	heh	04:21
	sure -- restore from your backup.	04:21
sol1dus	or is it the same as undoing rm -rf	04:21
offby1	exactly.	04:21
sol1dus	ok, i figured :P	04:22
	← jony left	04:24
	← sol1dus left	04:24
	← ankit9 left	04:26
	→ ankit9 joined	04:26
	→ hiwd_ joined	04:27
	→ srid joined	04:28
	← hiwd_ left	04:28
	→ jchris joined	04:29
	→ intripoon joined	04:30
	← hiwd left	04:30
	← cyphactor left	04:31
	← dsal left	04:31
	← LiamH left	04:32
	← scientes left	04:35
	→ scientes joined	04:35
	← lresende left	04:37
	→ theCarpenter1 joined	04:37
segfaux	Reading about packfiles from http://book.git-scm.com. Is the fanout table + index just a two-level skiplist?	04:38
	← slums left	04:38
	← greyface left	04:39
Fissure	pretty much	04:39
	← theCarpenter left	04:39
Fissure	fanout just saves 8 binary search iterations	04:39
	→ jony joined	04:45
	→ jtaby joined	04:46
jtaby	when i push in git, i get a "you can't push to git"	04:46
	i'm using github	04:46
	← seangrove left	04:47
	← jrmuizel left	04:49
Fissure	be more exact with commands/error messages	04:51
	→ eno__ joined	04:53
	← gotgenes left	04:54
	← krawek left	04:56
	← ankit9 left	04:56
	← jony left	04:56
	→ ankit9 joined	04:56
	← jfkw left	05:02
	← ankit9 left	05:03
	← yurinotigor left	05:03
	→ ankit9 joined	05:03
	← tongueroo left	05:04
	← natlus left	05:04
	← gorozco left	05:05
	← eno left	05:05
	← ankit9 left	05:07
	← dbindner left	05:07
	← doener left	05:08
	→ doener joined	05:08
	→ Paf joined	05:09
	→ tjafk1 joined	05:10
	→ natlus joined	05:11
	← theCarpenter1 left	05:14
	← jtaby left	05:15
	→ mgarriss joined	05:15
	→ jony joined	05:16
	← vbabiy left	05:16
	← mgarriss left	05:18
	→ nick_h[litage] joined	05:21
	→ seangrove joined	05:21
	← jony left	05:26
	→ ankit9 joined	05:26
	← scientes left	05:27
	→ scientes joined	05:27
	← tjafk2 left	05:27
	← angerman left	05:28
	→ cao1 joined	05:29
	→ broken_ladder joined	05:29
	→ AAA_awright_ joined	05:29
	← cao left	05:29
telmich	good morning	05:30
	→ altrux joined	05:31
telmich	I'm trying to git fetch a https-url (https://brezn.muc.ccc.de/~apic/git/faq.eof.name) and get download error - SSL: certificate subject name (brezn.muc.ccc.de) does not match target host name 'brezn.muc.ccc.de	05:31
	← altrux left	05:31
telmich	I'm wondering what was the right location to tweak curl?	05:31
	← decomp left	05:32
	← seangrove left	05:35
	→ johnw joined	05:39
	← ia left	05:39
	→ ia joined	05:40
	→ ice799 joined	05:42
	→ darwin joined	05:43
	→ ilogger2_ joined	05:52
	→ error404notfound joined	05:55
	→ curvature joined	06:01
	→ saurabh1403 joined	06:01
	→ josesito joined	06:02
	→ angerman joined	06:02
	← mika left	06:02
	→ oleg\|log joined	06:02
	→ cyphactor joined	06:02
	→ chalky joined	06:02
	oleg\|log → Guest3939	06:02
	→ merlin83 joined	06:03
josesito	if i had this folder: foo/ which is under version control, and it contains foo/bar with some files in it, and i wanted to get some other files from another repository to foo/ but those files also contain a bar directory, would it be safe to do a git fetch?	06:04
	→ dsop joined	06:05
johnw	josesito: git fetch will never change anything	06:06
	it just fetches remote objects into your repository	06:06
	git pull might not be a good idea	06:06
josesito	johnw: what could i do? i want to import the remote dir to my current dir having both under version control	06:07
johnw	you want them to both be named bar?	06:08
josesito	mmm...not exactly	06:08
	→ reithi joined	06:08
	→ andres joined	06:08
josesito	suppose this scenario, I have my .vim directory under version control and i want to install a plugin which is also version controlled with git	06:09
	→ mariusSO joined	06:09
	→ Sput joined	06:09
johnw	i think you want to use submodules	06:09
josesito	i think i do (i read on that somewhere just can't remember right now), how would i use them?	06:10
johnw	see man git-submodule	06:10
josesito	ty	06:11
	→ JohnDoy joined	06:11
	← mariusSO left	06:11
	→ tongueroo joined	06:14
	→ cedricv joined	06:14
	→ ankit9 joined	06:15
	← ankit9 left	06:15
	← josesito left	06:16
reithi	gm	06:17
	→ Squt joined	06:18
	→ ankit9 joined	06:21
	→ seangrove joined	06:21
	→ galderz joined	06:23
	→ mstormo joined	06:25
	→ idahoev1 joined	06:28
	→ scrogson joined	06:30
	→ feanil joined	06:30
	→ psankar joined	06:34
	→ bschindler joined	06:35
	← feanil left	06:36
	→ ijcd joined	06:40
	← idahoev1 left	06:40
	→ hobodave joined	06:41
	→ ljungk joined	06:42
	→ robinr joined	06:42
	→ jceb joined	06:50
	← jceb left	06:50
	→ ben_h joined	06:50
	→ jceb joined	06:51
	→ djpig joined	06:52
	→ plasticine joined	06:52
	← tongueroo left	06:53
	→ jceb81 joined	06:54
	← saurabh1403 left	06:55
	← jceb left	06:55
	→ comawhite joined	06:56
	← robinr left	06:56
comawhite	I'm wanting to create my own tree model. with qstyleditemdelegate, etc. And I was wonder (about MVC) if I'm returning a icon based on different roles (custom ones). How do I know which one of the delegate, model, view to return the icon so it's viewable in the tree?	06:58
	the is the only thing confusing me about MVC	06:59
	that*	06:59
DrNick	I suspect you're in the wrong channel	06:59
comawhite	oops sorry	06:59
	I thought I was in #qt	06:59
	my bad V_V	06:59
	comawhite hides	07:00
	→ p_masho joined	07:01
p_masho	can one have directionand advice please...	07:02
	I am surrentyl using svn, so to update my website,, i do and svn up on remote wesite.....	07:03
	→ jmworx joined	07:04
	← Guest3939 left	07:04
	→ lfoppiano joined	07:08
	← bschindler left	07:12
	→ porcnick joined	07:12
	→ Guest3939 joined	07:12
	← dsop left	07:13
	→ dsop joined	07:15
	→ gaveen joined	07:15
	→ vals_ joined	07:17
	→ ProperNoun joined	07:17
	→ peritus_ joined	07:19
	→ travisjeffery joined	07:20
	→ __doc__ joined	07:26
	← ankit9 left	07:27
	→ ankit9 joined	07:27
	→ eurdk joined	07:27
	← Guest3939 left	07:27
	→ shruggar joined	07:27
	→ giallu joined	07:30
	→ Weaselweb joined	07:31
	→ nick125 joined	07:35
	← cyphactor left	07:36
	→ ilogger2 joined	07:45
	→ dsop joined	07:45
	→ ph^ joined	07:46
	→ aisbaa joined	07:47
	→ CardinalNumber joined	07:47
	→ Arafangion joined	07:48
	→ ntoll joined	07:48
	→ Flathead joined	07:50
	→ f0rkit joined	07:52
	→ priidu_ joined	07:55
	→ dsop_ joined	07:56
	→ Guest3939 joined	07:56
	← dsop left	07:57
	→ elmex joined	08:03
	→ ankit9 joined	08:03
	→ ProperNoun joined	08:05
	→ GyrosGeier joined	08:06
	← ankit9 left	08:09
	→ johan-s joined	08:10
	→ tango_ joined	08:16
	→ zinou joined	08:19
	→ vuf joined	08:21
	← f0rkit left	08:22
	← CardinalNumber left	08:22
	→ gaveen joined	08:22
	→ cilly joined	08:25
	→ sids joined	08:28
	→ ankit9 joined	08:28
sids	is there anyway of recovering a file/directory that I have removed using git-rm?	08:29
vuf	sids: yes. did you just add, or also commit it?	08:31
	→ kiva joined	08:31
sids	vuf: I have committed it. But I don't want to revert the commit, it contains other changes too.	08:32
	→ ceyusa joined	08:34
	→ solydzajs joined	08:34
	← solydzajs left	08:39
	→ gilligan- joined	08:40
gilligan-	hi	08:41
	← ankit9 left	08:42
	→ lfoppiano joined	08:42
	Guest3939 → elgi	08:43
gilligan-	bit of a newbie question .. suppose i've made some changes to the working directory on the mater branch and then realize i'd much rather commit them to a different/new branch -- how should I usually do that ?	08:43
	guess i could stash the changes away.. create a new branch and apply them to that ?	08:43
charon	sids: git checkout $commit^ -- $file	08:45
	gets the contents of $file from the (first) parent of $commit, so if $commit deleted it, it will appear to be restored and staged for commit	08:46
sids	charon: that worked, thanks a ton!	08:47
	→ CardinalNumber joined	08:50
	→ felipec joined	08:51
felipec	how can I update the remote HEAD?	08:51
	← priidu_ left	08:53
vuf	sids: use "git checkout HEAD^ filename"	08:54
felipec	is it possible?	08:55
sids	vuf: thanks! charon had suggested "git checkout $commit^ -- $file" and that worked too.	08:56
vuf	sids: ah, didn't see his response ... it's the same thing :)	08:57
	gilligan-: no need to stash, just create the branch and commit	08:58
	→ parasti joined	09:00
angerman	can I compare a file to it's state at a tag?	09:00
Pieter	yes	09:00
angerman	how?	09:01
	I know how I can compare against a commit. but the help doesn't say about tags	09:01
mugwump	angerman: see man git-diff	09:01
	git diff revision -- filename	09:01
	→ ankit9 joined	09:01
	← felipec left	09:01
mugwump	on man git it explains that anything that accepts a revision argument accepts a tag, commit, or anything described on man git-rev-parse	09:02
Pieter	poort gitbot isn't here :(	09:02
angerman	mugwump: ahh, so I can just give it the tag name	09:02
	→ Gitbot joined	09:03
Pieter	man git-diff	09:03
	well that works	09:03
	→ SRabbelier joined	09:04
	→ henux joined	09:04
henux	can git handle automatic updating of timestamps which i keep in source files?	09:04
	← ProperNoun left	09:04
henux	"last modified" kind of timestamps	09:05
	→ priidu joined	09:06
charon	henux: short answer: no, long answer: read about export-subst in man gitattributes	09:07
Gitbot	henux: the gitattributes manpage can be found at http://git.or.cz/man/gitattributes	09:07
	← ankit9 left	09:12
	→ cedricv joined	09:12
	→ ankit9 joined	09:12
	→ mash_ joined	09:15
	→ jmesnil joined	09:16
	→ peritus_ joined	09:18
	← ankit9 left	09:18
	→ ankit9 joined	09:19
	→ j416 joined	09:20
	→ ciskje joined	09:21
	← gilligan- left	09:24
	← ankit9 left	09:26
	→ BartVB joined	09:28
	→ eurdk joined	09:28
schweizer	how can i do a cvs export in git?	09:30
	→ thumper_laptop joined	09:32
	→ shruggar joined	09:35
	→ Whtiger joined	09:35
	← sids left	09:40
	← cilly left	09:41
	→ mithro joined	09:44
henux	I have a remote git repo on ssh, and a cloned git repo in my local comp. The remote repo rejects my `git commit --amend` with this error message: ! [rejected] master -> master (non-fast forward)	09:44
	→ arcame joined	09:44
henux	Can I fix that?	09:44
	→ pi_ joined	09:47
pi_	hello	09:47
	any knows how to run git test suite?	09:48
Arjen	make test?	09:49
	→ _slacker_ joined	09:49
wereHamster	henux: faq non-ff	09:49
Gitbot	henux: Your push would lose changes on the remote. See http://git.or.cz/gitwiki/GitFaq#non-ff	09:49
pi_	thx	09:51
	← kiva left	09:54
Grum	wereHamster: can we teach gitbot new stuff as well? :D	09:57
wereHamster	Grum: no idea what kind of bot it is. And I don't even know who she blongs to. Try asking pasky	09:59
	Grum: what kind of things do you want to teach it?	10:01
	← shruggar left	10:01
	→ shruggar joined	10:02
wereHamster	I think that she loads the faq list straight from the gitfaq wiki	10:02
	→ segfaux joined	10:02
	→ cemerick joined	10:06
	→ shodges joined	10:08
	← shodges left	10:08
	→ gilligan- joined	10:09
	← gilligan- left	10:09
pi_	anyone knows how to send patch using pop3 in git	10:10
	just don't remember the command name	10:10
	→ roccoblues joined	10:11
Arjen	Sending using pop3 seems unlikely. Do you mean 'git send-email' ?	10:12
	→ JohnFlux joined	10:12
	→ justvisiting joined	10:12
	→ shadowhand joined	10:12
JohnFlux	what's the difference between "remotes/origin/mybranch" and "origin/mybranch" ?	10:12
	→ saurabh1403 joined	10:13
	← roccoblues left	10:14
JohnFlux	when a friend does "git branch" it says "* origin/mybranch" which doesn't make sense to me	10:14
	how can they have origin/mybranch currently checked out?	10:15
	→ gilimanjaro joined	10:15
broonie	JohnFlux: You can check out remote branches, you'll get a warnign when you do it.	10:15
JohnFlux	broonie: what does it mean though? what if he commits?	10:15
broonie	It means he shouldn't commit; IIRC he'll end up on a deteched HEAD but ICBW.	10:16
JohnFlux	hmm, I run the latest version of git, and can't get into that state	10:19
Pieter	JohnFlux: they probably created a local branch with the same name as the remote	10:19
	git checkout -b origin/branch origin/branch	10:20
	← segfaux left	10:20
JohnFlux	Pieter: oh, wow, git lets you do that?	10:20
Pieter	yes	10:20
	http://pastie.org/413985	10:20
broonie	It's a common way of naming topic branches.	10:21
JohnFlux	okay how do I fix this messup ? ;-)	10:21
	(I still can't see why you'd want to do that)	10:21
broonie	You wouldn't want specifically the same name as a remote - what's common is 'area/topic' style branch anming.	10:22
Pieter	git branch -m newbranchname	10:22
JohnFlux	Pieter: thanks	10:23
	should git prevent you from doing -b origin/something ?	10:23
	→ psychoschlumpf joined	10:23
JohnFlux	or at least require a --force-it option :-D	10:23
Pieter	I think so, I brought it up on the mailinglist once, but people didn't agree with me	10:24
psychoschlumpf	is there a way to get git to fill out version fields in sourcefiles like cvs does?	10:24
Pieter	arguing it's the users own fault, we're not going to protect them from that	10:24
JohnFlux	Pieter: urgh, this is why git is so unfriendly	10:24
	→ sunoano joined	10:24
JohnFlux	Pieter: to a newbie, doing "git checkout origin/mybranch" gives a warning saying to use -b. So the user does the 'obvous' "git checkout -b origin/mybranch origin/mybranch" and bingo, screwed eveything up	10:25
Pieter	yes, I know	10:25
	you don't have to tell me :)	10:26
JohnFlux	is there any reason that you want to do: git checkout -b origin/anything ?	10:26
	even if anything is different from mybranch	10:26
pasky	wereHamster: Piter's	10:27
JohnFlux	psychoschlumpf: there's a git faq that answers that	10:27
psychoschlumpf	JohnFlux: you have a link to it?	10:32
	there are a few when i google for "git faq"	10:32
wereHamster	psychoschlumpf: http://git.or.cz/gitwiki/GitFaq#head-4a0afe71a2bb7734777a8b0b345e0308aefdbd40	10:33
psychoschlumpf	thanks	10:33
	→ cao joined	10:35
	→ orafu joined	10:36
	→ ankit9 joined	10:36
	← ankit9 left	10:37
	→ ankit9 joined	10:37
	→ KIAaze joined	10:41
	← justvisiting left	10:44
	→ kraymer joined	10:47
	→ reprore joined	10:48
	→ sgrimm joined	10:49
	→ Fullmoon joined	10:49
KIAaze	hi, I'm having a problem with git clone from an http repository	10:49
	when I use git clone ssh://URL, I get the latest changes	10:50
	but not with http://URL	10:50
	why is that?	10:50
	→ cmarcelo joined	10:50
	← psychoschlumpf left	10:50
KIAaze	also, when I use git pull after having used "git clone http://URL" for the "checkout", I get the following error message:	10:51
	→ cilly joined	10:51
KIAaze	Warning: No merge candidate found because value of config option	10:51
	"branch.master.merge" does not match any remote branch fetched.	10:51
	No changes.	10:51
	→ angerman joined	10:52
	← ankit9 left	10:52
	← reprore left	10:52
	← angerman left	10:52
	← kraymer left	10:52
	→ ankit9 joined	10:52
	→ reprore joined	10:52
	← ankit9 left	10:52
	→ qrush joined	10:53
	→ maxb joined	10:56
	→ frosch joined	10:57
	→ tjafk joined	10:58
	← saurabh1403 left	10:58
	← priidu left	10:58
	→ ruphy joined	11:01
	→ priidu joined	11:01
	→ ekidd joined	11:04
	← j416 left	11:05
	← cilly left	11:07
	→ kraymer joined	11:07
	→ cilly joined	11:12
	→ dmathieu joined	11:14
	← eurdk left	11:14
	→ mika joined	11:14
	→ bryanl joined	11:16
	← qrush left	11:17
	← pi_ left	11:17
offby1	KIAaze: when a web server serves a git repository, someone needs to set up some hook or else it gets stale.	11:18
KIAaze	I did	11:18
offby1	I don't understand that warning	11:18
KIAaze	but I noticed that the post-update script wasn't executable (the official doc only said to copy the template, not make it executable)	11:18
	so I fixed that and now the http clone i up to date	11:19
	but the pull still gives me that error	11:19
JohnFlux	why do I have "remotes/origin/mybranch" but other people have "origin/mybranch" ?	11:19
KIAaze	what git version do you have?	11:20
JohnFlux	latest	11:20
	(I run from git)	11:21
	git.git	11:21
KIAaze	I use git version 1.5.2.4 and I have "origin/branch"	11:21
JohnFlux	KIAaze: that's pretty old	11:22
KIAaze	opensuse 10.3	11:22
	→ ngulden joined	11:25
	← Fullmoon left	11:29
	→ abbe joined	11:29
	→ chris2 joined	11:34
	→ ricky_clarkson joined	11:35
ricky_clarkson	Given that I committed to branch 1234 as if I had commit diarrhoea, how can I apply those commits to master as one commit with a new commit message?	11:36
KIAaze	If I upgrade to the latest git version from 1.5.2.4, might there be some problems with the old git repositories?	11:36
ricky_clarkson	I'm looking at the man pages, but could do with some direction.	11:37
JohnFlux	ricky_clarkson: git rebase -i HEAD..origin and rearrange/squash commits	11:37
	ricky_clarkson: when you are happy, then push upstream	11:37
ricky_clarkson	HEAD? When using git svn, will that do a 'real' commit?	11:37
	Ah, no.	11:37
	So master and HEAD don't mean the same thing?	11:37
	I'll give it a go, thanks.	11:38
JohnFlux	ricky_clarkson: HEAD is where you are currently. that would be the same as master if you're currently on master	11:38
	so git rebase -i master..origin would do the same	11:38
	ricky_clarkson: no 'real' commits are made to svn until you push	11:39
ricky_clarkson	Can I do a fake push to see what would be sent?	11:41
	← cilly left	11:41
vmiklos	git push --dry-run?	11:42
JohnFlux	ricky_clarkson: git push --help says yes you can :-)	11:42
ricky_clarkson	git rebase -i HEAD..origin - fatal: Needed a single revision	11:42
	ricky_clarkson -> manpages	11:42
JohnFlux	ricky_clarkson: maybe three dots? or maybe just git rebase -i origin	11:43
	every command seems to want a different format. I've never figured it out	11:43
ricky_clarkson	heh	11:43
JohnFlux	i just try random things until it works :-D	11:43
	→ cannonball joined	11:43
	→ MisterN joined	11:44
	← tjafk left	11:45
	→ saurabh1403 joined	11:46
	→ qrush joined	11:49
	→ Octalot joined	11:50
	→ radarek joined	11:53
	→ cilly joined	11:56
ricky_clarkson	I tried git svn dcommit --dry-run, but it only says it committed. Can I see the commit message?	11:57
	Yep, definitely confused.	11:58
	git checkout master	11:59
	git rebase -i branchname	11:59
	"Nothing to do"	11:59
	← qrush left	11:59
ricky_clarkson	but git log shows there's an extra commit in branchname.	11:59
vmiklos	git svn dcommit --dry-run may not be as smart as git push --dry-run :)	12:00
	→ beket joined	12:00
ricky_clarkson	Swapping things around lets me edit the message but no change appears to occur.	12:00
	git checkout branchname	12:00
	git rebase -i master	12:00
	git checkout master; git log \| less shows nothing new.	12:01
	→ blockowitsch joined	12:02
	→ Ryback_ joined	12:02
	→ ProperNoun joined	12:05
	← CardinalNumber left	12:06
offby1	"every command seems to want a different format." -- it's not _quite_ that bad, but I see what you mean	12:08
	→ bdiego joined	12:10
ricky_clarkson	I keep typing cd instead of git checkout	12:12
	→ qrush joined	12:13
	→ xinming_ joined	12:14
offby1	interesting	12:15
	have you got your bash prompt set up to display your current branch?	12:16
	→ girishr joined	12:16
JohnFlux	ricky_clarkson: you know about "cd -" ?	12:16
	ricky_clarkson: useful if you type "cd" accidently	12:16
	→ glommer joined	12:17
	← Flathead left	12:18
	→ CardinalNumber joined	12:18
	→ EmilMedve joined	12:18
ricky_clarkson	JohnFlux: No. What's it supposed to do? It just announces that OLDPWD isn't set.	12:19
	I guess it 'cds back'.	12:19
JohnFlux	ricky_clarkson: right, that's what it's supposed to do	12:19
	go to some directory, then do "cd; cd -" you should be back where you start	12:20
ricky_clarkson	That works.	12:20
	Seems only to support one level of undo.	12:21
JohnFlux	yeah it's not perfect, but it's great for accidental "cd"'s	12:21
	→ peper joined	12:22
peper	hello	12:22
	can i recover a lost tree somehow?	12:23
offby1	PROMPT_COMMAND="pwd > ~/dirs; git commit -m $pwd"	12:23
ricky_clarkson	Crap, I seem to have lost some changes in all this fiddling.	12:23
peper	i have overwritten current dir and some files in .git	12:23
offby1	do you know which files?	12:24
JohnFlux	offby1: PS1='\[\033[01;34m\]\w\[\033[00m\]$(__git_ps1 " (%s)")\$ '	12:24
	offby1: that's much better	12:24
ricky_clarkson	I had a diff left.	12:25
JohnFlux	ricky_clarkson: git reflog will tell you where you used to be	12:25
peper	those that were in my old .git	12:25
offby1	JohnFlux: well, I was joking; I meant "Hey; let's keep track of all our current directories in a git repository!"	12:25
ricky_clarkson	Er, I had happened to run git diff earlier in my shell and have now saved the output to a file.	12:25
offby1	peper: _all_ of them?	12:25
peper	i have found the last commit with fsck	12:26
	offby1: yes, old .git files overwritten new .git files	12:26
	but those that weren't present weren't deleted	12:26
JohnFlux	ricky_clarkson: might be worth running "git reflog" anyway so you know for next time	12:26
peper	so the commit should still be there, right?	12:26
offby1	peper: I dunno. Depends if the relevant files were overwritten :-\|	12:27
peper	*commits	12:27
	i have found the last one	12:27
Aides	ghftyrrthfhgfgh	12:27
ricky_clarkson	http://pastebin.com/f719bc369 is the output from git reflog	12:27
	→ esparkman joined	12:27
ricky_clarkson	I guess I can revert some of those until I have what I want.	12:28
peper	offby1: mm, just git reset --hard last_commit worked	12:28
	all recovered :)	12:28
offby1	huh!	12:28
	good	12:28
JohnFlux	ricky_clarkson: just do: git show 6240c73 etc	12:29
	ricky_clarkson: looking at the reflog for the commit that you've lost	12:29
	ricky_clarkson: when you find it, add it with " git cherry-pick 6240c73" or whatever	12:29
	← bryanl left	12:30
ricky_clarkson	hrm	12:30
	git show HEAD shows it.	12:30
	→ kbingham joined	12:31
ricky_clarkson	scrap that	12:31
JohnFlux	:-D	12:32
	← ProperNoun left	12:33
	→ lucsky joined	12:34
	→ lorandi joined	12:34
ricky_clarkson	Ok, found it.	12:36
	git cherry-picked it.	12:36
	Can you see how I lost it in the first place from that reflog output?	12:37
	← qrush left	12:38
offby1	See if you can find my car keys while you're in there	12:38
JohnFlux	ricky_clarkson: maybe because of the rebase?	12:38
	ricky_clarkson: if you delete a line there, it removes the commit	12:38
ricky_clarkson	I didn't delete a line, afaik.	12:39
JohnFlux	dunno. sometimes commits get lost	12:39
	:-D	12:39
	→ bwalk_ joined	12:41
	→ botanicus joined	12:41
	← ceyusa left	12:42
botanicus	Hi. I've created empty repo on my server (mkdir repo ; cd repo ; git init), pushed to it successfully, but when I try git log here, I get fatal: bad default revision 'HEAD'	12:42
offby1	well, they don't just wander off. If you did a rebase, then there'll be a series of commits (the ones that HEAD originally pointed to) that may now be unreachable.	12:42
botanicus	What does it mean?	12:42
	← frosch left	12:43
offby1	botanicus: it means you've never done "git checkout"	12:43
	→ Suprano joined	12:43
offby1	faq nonbare	12:43
	faq non-bare	12:43
Gitbot	offby1: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare	12:43
bwalk_	botanicus: ... to slow :)	12:43
offby1	botanicus: read that ^^	12:43
ricky_clarkson	offby1: I suppose gitk is supposed to help me see that.	12:43
botanicus	offby1, thanks	12:43
offby1	ricky_clarkson: not really.	12:44
	If there's some existing branch that refers to them, then "gitk --all" or "gitk branchname" will show 'em. If not, though, you gotta use the reflog to find 'em	12:44
ricky_clarkson	I see.	12:44
	← EmilMedve left	12:45
offby1	has everyone here read "Git For Computer Scientists"? If not, put down your forks and read it now.	12:46
	It's short.	12:46
	Yes, put down chopsticks too, if you're using chopsticks.	12:46
bwalk_	but I am scared by the word 'Directed Acyclic Graph'	12:47
	← dmathieu left	12:48
	← parasti left	12:48
botanicus	offby1, well, but git log works for committed changes, so I should see then even when the working tree is empty, not?	12:49
	→ batiestuta joined	12:51
	← ekidd left	12:52
	→ roccoblues joined	12:53
botanicus	But OK, I've created bare repo, pushed commits here (successfuly), and then I've got:	12:53
	Initialized empty Git repository in /webs/static/101ideas.cz/alpha/.git/	12:53
	warning: remote HEAD refers to nonexistent ref, unable to checkout.	12:53
	Uh, I've got when trying to do clone from the server repo	12:55
offby1	bwalk_: it's ok; we're here	12:55
bwalk_	:)	12:55
offby1	botanicus: well, "git log" follows the ancestry of some ref or other; HEAD by default. But in your case, you don't _have_ a HEAD (or perhaps you do but it points nowhere, or something)	12:56
ricky_clarkson	I really need to set EDITOR. Editing commit messages in vim inside an emacs term is confusing.	12:56
offby1	only because your mind is weak!	12:56
JohnFlux	is there a way to clean whitespace before I commit?	12:57
	ricky_clarkson: haha	12:57
offby1	ricky_clarkson: (server-start) (setenv "EDITOR" (concat exec-directory "emacsclient"))	12:57
botanicus	offby1, how is it possible? I thought HEAD exists in all cases.	12:57
offby1	botanicus: apparently not :)	12:57
JohnFlux	if you have an empty repo, the presumably HEAD points nowhere?	12:58
	*then	12:58
offby1	I guess.	12:58
botanicus	But the repo should not be empty, push worked well	12:58
offby1	surely if you do "git checkout 1234", you've got a "detached HEAD".	12:58
	botanicus: it's not a question of the repository being empty; it's a question of not having a ref named HEAD.	12:58
	botanicus: if you do "git log 1234" (for some value of 1234) you'll see a perfectly fine log.	12:59
	→ ceyusa joined	12:59
Pieter	HEAD points to heads/master, which in turn doesn't exist	12:59
offby1	I wouldn't be surprised if _all_ bare repositories had no HEADs.	12:59
botanicus	I see!	12:59
	I have alpha branch, not master one	12:59
ricky_clarkson	Should I avoid numeric branch names for any technical reason?	13:00
botanicus	Funny :) When I edit bare_repo/HEAD to points to good ref, it works	13:00
ricky_clarkson	E.g., I've been naming them after the ticket they fix.	13:01
	→ bels joined	13:01
offby1	ricky_clarkson: I suspect those are OK	13:01
	botanicus: sure, why not?	13:01
	← BartVB left	13:01
offby1	I just made a bare repository, and pushed to it; my HEAD file contains "ref: refs/heads/master", and yet I have no "refs/heads/master".	13:02
	→ qrush joined	13:02
	→ BartVB joined	13:02
offby1	Pretty much as I'd have expected.	13:02
	now, why I even have that HEAD file in the first place ... I dunno.	13:02
ricky_clarkson	So after a rebase, I'm apparently not on any branch.	13:03
	I didn't know that was possible. :)	13:03
bels	hi all - I am running git on windows via msysGit and having trouble getting kdiff3 to work with 'git mergetool'	13:03
	i had kdiff3 previously installed and I can bring it up independtly	13:03
	← reprore left	13:04
	→ reprore joined	13:04
bels	I checked out the man page for mergetool and tried various combinations of setting the options - including mergetool to kdiff3, mergetool.kdiff.path to the full path to the exe (case sensitive)	13:04
	→ annodomini joined	13:04
bels	the behavior is i run 'git mergetool', it prints out correct information about using kdiff3, and when I hit <enter> to continue to the merge, nothign happens	13:05
	i have git running as a process and I have to ctrl-c to kill it	13:05
	any hints?	13:05
	→ prlawrence joined	13:05
	→ RaySl joined	13:05
	→ annodomini_ joined	13:06
bels	I thought it could be a windows path problem with the git bash shell, so I added the path to kdiff3.exe to the windows path and verified (via echo) that hte path in the bash shell does include the kdiff3 directory	13:06
	and i can run it from the shell direclty just by typing kdiff3.exe, as expected	13:07
	→ hdl2 joined	13:07
	→ mfilipe joined	13:09
offby1	ricky_clarkson: no, after a rebase, you should be on the same branch as when you started. What makes you think otherwise?	13:10
ricky_clarkson	git status	13:10
offby1	bels: yeah, setting that up was a pain in the butt. I'll see if I can dig up how I didit	13:10
	bels: oops, hold on: kdiff3 was the easy one :-\|	13:11
	[mergetool "kdiff3"] path = c:/Program Files/KDiff3/kdiff3.exe	13:11
	that's all I did.	13:11
	Didn't need any quote marks, even, which surprised me	13:11
bels	offby1: weird	13:11
	offby1: so you don't need to specify the -t option with that right? you just run it and have it setup as your default	13:12
ricky_clarkson	So I managed to git svn dcommit some changes from some branch, and this shows up properly in the SVN repo.	13:12
	All well and good, but now master doesn't have those changes.	13:12
	→ bryanl joined	13:12
ricky_clarkson	I suppose I should rebase master to head.	13:13
bels	offby1: how did you specify the path to 'git config' ? I am using the following:	13:13
botanicus	I have also problem with my post-receive hook. I'd like to use it for deployment. It just should go to ../branchname and do git pull. Unfortunatelly It does not work. The hook: http://pastie.org/414083 and the message: http://pastie.org/414084	13:13
	→ krh joined	13:14
offby1	bels: I didn't specify any path at all :-\|	13:14
	afaik	13:14
	→ rubydiamond joined	13:14
rubydiamond	guys	13:14
	update-index --refresh: command returned error: 1	13:14
	getting above error	13:14
	on	13:14
bels	offby1: oh it was setup automatically in your config file	13:14
rubydiamond	git svn rebase	13:14
ricky_clarkson	crappity.	13:14
shruggar	ricky_clarkson, you can run git svn rebase from any branch, as far as I am aware, and it will act sanely. However, I have heard the running git svn dcommit from just any branch may act "strangely"	13:14
rubydiamond	I have modified some files..	13:14
	yet not added but	13:14
ricky_clarkson	shruggar: Thanks. Where did you hear that about git svn dcommit?	13:15
	← annodomini_ left	13:15
	→ jrmuizel joined	13:15
ricky_clarkson	I've been trying to do it from master always but managed not to last time.	13:15
offby1	bels: it might have been; I can't remember.	13:15
bels	offby1: i read that kdiff3 is installed the msysGit installer should recognize it and add it as one of the supported tools	13:15
shruggar	ricky_clarkson, In the manpage for git-svn, I believe. Somewhere towards the bottom, I expect	13:15
ricky_clarkson	Normally it's straightforward but for one project I need to control the number of commits and use a particular commit message template, which is why I'm having to learn new stuff.	13:16
offby1	shruggar: yes, "git svn dcommit" might not commit to the svn branch you expect. Run "git svn info \| egrep URL:" first to check; and if it's wrong, do "git reset --hard remotes/whatever" to fix it	13:16
	bels: sounds reasonable	13:16
	→ dedmajor joined	13:16
ricky_clarkson	I'm checking with svn log from a normal svn checkout elsewhere.	13:16
botanicus	any ideas?	13:16
shruggar	offby1, if I'm only using git-svn to talk to a single svn branch at all times, will it always act as expected?	13:17
offby1	shruggar: I'd think so, yes.	13:17
	→ dohzya joined	13:18
shruggar	offby1, I'd think so too.. but then I'd tend to expect it to /always/ commit to "the branch I expect", in all circumstances...	13:18
	→ tanek joined	13:18
ricky_clarkson	Ah, yes, I'm using one svn branch, trunk.	13:18
	So I'm safe for now I guess.	13:18
offby1	shruggar, ricky_clarkson: http://gist.github.com/78059	13:19
	that's a script that demonstrates git-svn properly committing to separate svn branches.	13:19
	← tanek left	13:19
prlawrence	offby: bels: `which kdiff3` gives no joy in msysgit bash, latest release...	13:19
offby1	however, if you delete the "reset" on line 45, it does The Wrong Thing	13:19
	prlawrence: not surprising; msysgit doesn't include kdiff3.	13:20
	It -does- include vim, but as far as I can tell, vim doesn't work as a 3-way mergetool	13:20
	kinda ironic	13:20
bels	y the default merge tools it lists are : merge tool candidates: opendiff emerge vimdiff	13:20
	i set both the tool and path properties:	13:21
ricky_clarkson	I have only ever used git branches, never svn ones. So when I have to use svn branches I'll probably do it from svn at first.	13:21
bels	mergetool.kdiff3.path=c:/Applications/KDiff3/kdiff3.exemerge.tool=kdiff3	13:21
	→ mw-home joined	13:21
JohnFlux	woah	13:21
	just found a strange bug	13:21
bels	(should be a space there after the .exe and before 'merge.tool'	13:21
JohnFlux	modify a file, then do: git commit --cleanup=whitespace filename	13:21
	→ parasti joined	13:21
JohnFlux	the commit message includes all the comments!	13:21
	all the comments in the commit message	13:22
	← annodomini left	13:22
bels	offby1 / prlawrence : basically it looks like it finds the property in my configuration file OK b/c i get a prompt to hit <enter> to use kdiff3	13:22
	offby1 / prlawrence : i hit enter, and then nothing happens - git.exe just blocks/waits (I can see in task manager) and no kdiff3 is shown	13:22
bwalk_	JohnFlux: confirmed	13:23
offby1	bels: I can't explain that :-\|	13:23
	bels: seems to me you're doing everything right.	13:23
bels	offby1 / prlawrence : if I do a 'which kdiff3', it shows the full path	13:23
offby1	bels: I don't think that matters.	13:23
	→ repror___ joined	13:23
JohnFlux	bwalk_: shall I report it to the mailing list?	13:23
offby1	My kdiff3 isn't on my PATH but it works anyway (presumably because my .gitconfig file contains its full path)	13:23
bels	offby1: yea I am like totally baffled. *nix based systems are not my forte so I am just assuming I'm overlooking something obvious	13:24
bwalk_	JohnFlux: sure, why not. does not look so intended	13:24
offby1	bels: wait. Surely you're not using *nix?	13:24
	I thought we were talking about msys	13:24
	← reprore left	13:24
bels	yes we are	13:24
bwalk_	JohnFlux: maybe test with upstream	13:24
offby1	oh.	13:24
bels	i am saying the commands / config etc are in the shell	13:24
offby1	0h.	13:24
	→ ekidd joined	13:24
bels	yes all on msysGit	13:24
JohnFlux	bwalk_: my version of git is only a few days old	13:25
	bwalk_: yours?	13:25
	→ willb joined	13:25
bwalk_	JohnFlux: no wait, what do you acutally want to do?	13:25
	← ekidd left	13:25
JohnFlux	bwalk_: get the bug fixed...	13:25
	→ kwatz joined	13:26
bwalk_	JohnFlux: what whitespaces do you want to be cleaned. --cleanup acts on the commit message, and --cleanup=whitespace _only_ cleans up whitespaces in the commit message, so it is a acutally correct result	13:26
bels	offby1 / prlawrence: are the other diff tools installed by default on msysGit? perhaps I can use those for the time being and sweat kdiff3 later?	13:27
offby1	I'm sure there's _something#!/bin/sh	13:28
	# See if git-svn really does let me dcommit to more than one branch.	13:28
	#set -x	13:28
	set -e	13:28
bwalk_	JohnFlux: you have read the manpage to git commit?	13:28
JohnFlux	bwalk_: oh, wait..	13:28
offby1	base=/tmp/experiment	13:28
	rm -rf $base	13:28
	mkdir -p $base	13:28
	svnrepo=$base/svnrepo # $(mktemp -d)	13:28
JohnFlux	bwalk_: hmm okay uh nevermind :-D	13:28
offby1	svnURL=file://$svnrepo	13:28
	svnwc=$base/svnwc # $(mktemp -d)	13:28
	gitrepo=$base/gitrepo # $(mktemp -d)	13:28
	#trap "for d in $svnwc $svnrepo $gitrepo; do rm -rf \$d; done" EXIT	13:28
	# First make a tiny svn repository.	13:28
	svnadmin create $svnrepo	13:28
	→ aziz joined	13:28
offby1	for d in branches tags trunk; do svn mkdir $svnURL/$d -m $d; done	13:28
	svn co $svnURL/trunk $svnwc	13:28
	cd $svnwc	13:28
	touch file	13:28
	svn add file	13:28
	svn commit -m "Create empty file in svn"	13:28
rubydiamond	git svn rebase gives me update-index --refresh: command returned error: 1	13:28
offby1	svn cp $svnURL/trunk $svnURL/branches/ted -m "ooh, a branch"	13:28
	echo "Trunk data from svn" > file	13:29
	svn commit -m "$(tail -1 file)"	13:29
	svn switch $svnURL/branches/ted .	13:29
	echo "Branch data from svn" >> file	13:29
	svn commit -m "$(tail -1 file)"	13:29
	→ EmilMedve joined	13:29
offby1	# Now dump out the entire svn repository.	13:29
	svn switch $svnURL .	13:29
	sigh	13:29
	sorry, accidental paste.	13:29
	→ brett_h joined	13:29
bwalk_	at least no passwords :)	13:29
Pieter	that's quite a paste	13:30
offby1	Pieter: what's ironic is that I added "accidental paste prevention" to my IRC client months ago, but recently disabled it by accident :-\|	13:32
Pieter	1:)	13:33
	eh, :)	13:33
offby1	with these cats walking across my desk, it's pretty useful	13:33
Pieter	sometimes irssi tells me "Are you sure you want to paste these 200000 lines?"	13:33
offby1	thanks for not riding me out of town on a rail.	13:33
Pieter	and tehn I press ^k instead of ^c	13:33
offby1	Pieter: excellent!	13:33
prlawrence	bels: offby: This looks like a nice and very recent HOWTO: http://davesquared.blogspot.com/2009/02/setting-up-diff-and-merge-tools-for-git.html	13:35
	→ FylGood joined	13:35
	→ boto joined	13:36
offby1	prlawrence: I'd seen that. I don't think I needed anything that fancy, though.	13:37
bels	prlawrence: yea I saw that earlier; from the content it looks like it was being setup for a non-standard / supported diff tool	13:37
	→ kukks joined	13:37
	→ annodomini joined	13:37
FylGood	I'm a single developer (though I am married), and was wondering if people feel that it is still a good idea to use a version control system rather than simply creating zip archives of my app. I assume yes, but wanted to ask before jumping into Git.	13:37
	→ countd joined	13:37
bels	prlawrence: about 1/4 of the down the page there are the prefes for kdiff3, which I tried and that failed too :)	13:37
offby1	FylGood: hate to be dull, but: it's up to you	13:38
FylGood	If I were part of a team, I understand that version control is a must, but what about the lone developer?	13:38
bels	prlawrence: but perhaps that extra script is necessary but I am not sure why	13:38
FylGood	offby1: Not dull at all. Just wanted to get a bit of perspective before jumping in	13:39
offby1	I (and I suspect many here) am a version-control geek, so naturally I keep my solo stuff under revision control. But am I really benefiting from it? Who knows.	13:39
prlawrence	FylGood: (lol) you are on a team and you don;t know it ...	13:39
	← sunoano left	13:39
prlawrence	FylGood: you and you in 6 months	13:39
bels	FylGood: I would recommend you use some level of VC just so you have some history to your project	13:39
offby1	FylGood: despite my admission of geekhood, I'd say: give it a try. It's certainly a good idea to know the concepts.	13:39
FylGood	prlawrence: lol Good point	13:39
bels	FylGood: What if you decide to back out some feature and need it back later or vice versa	13:39
	← girishr left	13:40
prlawrence	FylGood: and he is not as smart as you are :-)	13:40
offby1	bels: actually, I've kept my home directory in Subversion for _years_ now, and I almost never look at history. The main benefit that I get from it is that it's an easy way to replicate the files onto many other machines.	13:40
	of course most of that stuff doesn't quite qualify as "software" (but my .emacs comes close)	13:41
bels	offby1: I am coming at it from a coder's perspective. I just started some side project at home for my wife (text manipulation utilities, nothing fancy) and am using VC	13:41
offby1	I would too.	13:41
	FylGood: I also recommend GitHub. They've done an excellent job.	13:41
bels	offby1: but I have heard of doing it with config files etc as well - really anything that is (ideally) text and is likely to change a lot I would use VC	13:41
	→ girishr joined	13:42
	→ marvil07 joined	13:42
offby1	sure	13:42
bels	FylGood : You can also look at it from the perspective of why not do it? There is a slight overhead / configuration, but a lot of the GUI and plug-in IDE tools makes that less and less	13:42
FylGood	offby1: GitHub looks great. Definitely a big selling point on why to use Git.	13:42
offby1	I also use a tool called "etckeeper" which (/me glances around nervously to make sure RandalSchwartz isn't listening) keeps one's /etc directory in a git repository. _That_ has saved my ass a time or two.	13:43
bels	offby1: us windows users just blast and re-install the OS every 3 years :)	13:43
	→ markelikalderon joined	13:44
offby1	only 3 years?	13:44
	I do likewise with *nix, too.	13:44
prlawrence	bels: 3 years!?	13:44
FylGood	bels: I was every 4-6 months when I was a Windows user	13:44
	→ jfkw joined	13:44
offby1	Spring Cleaning, I call it.	13:44
bels	aha yes def - I started using VMs for this now	13:44
offby1	"Stop and Copy".	13:44
FylGood	bels: But I'm a happy Mac user now.	13:44
offby1	Or "Mark and Sweep"	13:44
FylGood	bels: ;)	13:45
prlawrence	offby: I agree with the 6mos assesment. you must run a tight ship	13:45
bels	I got this tool called nLite (I think) that basically packages up a windows CD and strips out what you don't want	13:45
	when I need to reinstall i pop that CD in,boot up and everything is entered/done automatically	13:45
offby1	I dint say nothin' 'bout six months. Did I?	13:45
	bels: that looks interesting.	13:46
prlawrence	offby1: sorry, that was FylGood.	13:46
bels	offby1: yea. you can also add other random stuff to the CD image ifyou want - common tools (7-zip, etc) and what not; i dig it	13:46
	← kwatz left	13:46
	← willb left	13:46
	← parasti left	13:46
	← cilly left	13:46
	← arcame left	13:46
	← mash_ left	13:46
	← lfoppiano left	13:46
	← JohnFlux left	13:46
FylGood	bels: If nLite strips out everything you don't want in Windows, then what's left. (Sorry, I couldn't help myself)	13:46
schweizer	is it possible to clone directly a certain branch?	13:47
prlawrence	offby1: do you git init your $HOME or do you have a Makefile or shell script stage your .files from a local git repo?	13:47
FylGood	Thanks everyone very much for the great advice. I'm gonna go ahead and setup Git with my current project and see where it takes me. Thanks again!	13:48
bels	FylGood: ahah good point :)	13:48
offby1	schweizer: not exactly, but you can "git init; git remote add ...; git fetch ..."	13:48
	schweizer: which will get just that branch.	13:48
	→ JohnFlux joined	13:48
	← annodomini left	13:48
	← abbe left	13:48
	→ boto_ joined	13:48
offby1	prlawrence: well, neither, since I use Subversion, not git, for my home directory. But in subversion terms: yes, I did the equivalent of "git init"	13:49
	→ kwatz joined	13:49
	→ willb joined	13:49
	→ parasti joined	13:49
	→ cilly joined	13:49
	→ arcame joined	13:49
	→ mash_ joined	13:49
	→ lfoppiano joined	13:49
	→ giallu joined	13:49
	→ shad joined	13:49
	→ reto_ joined	13:49
	→ noth joined	13:49
	→ _maks joined	13:49
	→ dwmw2_gone joined	13:49
	→ mike_m joined	13:49
	→ Pistahh joined	13:49
	→ AlexanderS joined	13:49
	→ wereHamster joined	13:49
	→ koki joined	13:49
	→ lack joined	13:49
	→ sejo joined	13:49
	← parasti left	13:49
	→ rovalent joined	13:49
schweizer	offby1: thx	13:49
bels	offby1 / prlawrence : thanks for the help, I am going to see what I can do to try and get this to work	13:49
	→ Fullmoon joined	13:49
bels	offby1 / prlawrence : maybe try that script on that dave^2 site...	13:49
prlawrence	offby1: sorry, you said you use Subversion. noneth... ah, you beat me to it	13:49
	← boto_ left	13:50
	→ boto_ joined	13:50
	← Arafangion left	13:50
bels	offby1 / prlawrence : actually one last thing: is there a verbose option on mergetool? I tried --verbose and -v and got errors so I assume not...	13:52
	→ brasko joined	13:53
offby1	beats me	13:53
	you know as much as I do	13:54
bels	offby1: ok np - thanks again for the help	13:54
brasko	Hi, I'm dealing with a large binary file in git. I've heard that another distributed vc system allows you to only clone the most recent binary file, instead of the entire history to make the clones small. Does git support that?	13:54
wwalker	anyone know of an article talking about using git to manage you dotfiles etc? like http://joey.kitenet.net/svnhome/ but with git? I can probably just use this URL but thought I'd ask about a git specific implementation though they should be quite similar.	13:54
	← FylGood left	13:54
	← roccoblues left	13:55
	← bels left	13:56
prlawrence	bels: git-mergetool is a shell script... feel free to read the source and throw in whatever debugging statements woudl help you grok its problem	13:56
bremner	wwalker: http://vcs-home.madduck.net/	13:57
wwalker	bremner: thank you	13:57
prlawrence	Yikes, speaking of reading the source, I see lots of TABs mixed in with spaces-only indentation. Does git.git have a standard, or anythign goes?	13:58
	← markelikalderon left	13:59
	→ parasti joined	13:59
	← mithro left	13:59
Pieter	prlawrence: use a tab instead of 8 spaces	14:00
	→ markelikalderon joined	14:00
	← marvil07 left	14:01
	→ annodomini joined	14:02
prlawrence	Ah, just found a post by Linus stating that the two can and will mix, so use 8 spaces to represent any tabs. http://kerneltrap.org/mailarchive/git/2007/10/16/344968	14:03
	→ marvil07 joined	14:03
	← boto left	14:03
prlawrence	Pieter: thanks!	14:03
offby1	brasko: I don't think so. On the other hand, git's compression is surprisingly good.	14:04
	wwalker: actually I think that article says, at the very bottom, that he's using git now anyway :)	14:04
	← jrmuizel left	14:04
rubydiamond	git svn rebase does not work.. if you made some changes in files	14:06
offby1	uncommitted changes?	14:07
	← markelikalderon left	14:08
rubydiamond	Octalot, yeah	14:08
	offby1, yeah	14:08
brasko	offby1: but, my binary files are already compressed	14:08
offby1	.oO("Octalot"?)	14:08
	brasko: oh well.	14:08
rubydiamond	offby1, what is the reason	14:09
offby1	rubydiamond: that, I don't know. Probably "if I let you rebase when you've got uncommitted changes, you'll later get very very confused".	14:09
	rubydiamond: so either commit; or (if you _really_ don't want to) "git stash".	14:09
rubydiamond	hmm	14:09
offby1	then after rebasing, do "git stash apply" and your changes are back.	14:10
	→ spearce joined	14:10
rubydiamond	offby1, that looks like nice trick	14:10
	→ bantic joined	14:10
offby1	I have a coworker who recently learned about "git stash" ... for weeks after, he would blurt out "git stash is COOL!"	14:10
	← spearce left	14:10
	→ spearce joined	14:10
offby1	at random intervals	14:10
rubydiamond	haha	14:11
	:)	14:11
brasko	offby1: http://4.bp.blogspot.com/_i3HweI5bST8/RwCBda6kWKI/AAAAAAAAAVI/Zv1QOb4r1y8/s400/ohWELL_loRES.jpg :(	14:12
	offby1 scratches head	14:12
rubydiamond	ha ha	14:12
	→ Finkregh joined	14:13
offby1	"But don't ask me what I think of you; I might not give the answer that you want me to"	14:13
JohnFlux	offby1: I always wonder why git stash is needed	14:13
offby1	JohnFlux: well, there's one example	14:13
	→ moonese joined	14:13
JohnFlux	all the commands that you need it for should be 'fixed' to just do it automatically :-)	14:13
offby1	technically, it's not _needed_; you could create a branch; check it out; commit; check the original branch back out.	14:13
	it's just a convenience.	14:13
	but then, a car is just a convenience; your horse could take you just as far, with suitable care and feeding.	14:14
	and time.	14:14
	← marvil07 left	14:14
	← dwmw2_gone left	14:14
	← koki left	14:14
	← Pistahh left	14:14
	← kwatz left	14:14
	← reto_ left	14:14
	← lack left	14:14
	← sejo left	14:14
	← AlexanderS left	14:14
	← shad left	14:14
	← wereHamster left	14:14
	← _maks left	14:14
	← mike_m left	14:14
	← noth left	14:14
	← willb left	14:14
	← lfoppiano left	14:14
	← giallu left	14:14
	← mash_ left	14:14
	← arcame left	14:14
	← cilly left	14:14
	→ kwatz joined	14:15
	→ willb joined	14:15
	→ cilly joined	14:15
	→ arcame joined	14:15
	→ mash_ joined	14:15
	→ lfoppiano joined	14:15
	→ giallu joined	14:15
	→ shad joined	14:15
	→ reto_ joined	14:15
	→ noth joined	14:15
	→ _maks joined	14:15
	→ dwmw2_gone joined	14:15
	→ mike_m joined	14:15
	→ Pistahh joined	14:15
	→ AlexanderS joined	14:15
	→ wereHamster joined	14:15
	→ koki joined	14:15
	→ lack joined	14:15
	→ sejo joined	14:15
	→ dmathieu joined	14:16
	→ ekidd joined	14:19
	→ jackdempsey joined	14:19
	→ moonese1 joined	14:23
	→ Frando joined	14:24
	→ Yuuhi joined	14:25
	→ tjafk joined	14:27
	→ j416 joined	14:29
j416	NoirSoldats: are you here?	14:30
	→ wmoxam joined	14:30
	→ abbe joined	14:31
	→ idahoev1 joined	14:31
	→ hdevalence joined	14:32
	← peritus_ left	14:33
	→ FylGood joined	14:33
	← moonese left	14:39
	← FylGood left	14:39
	→ EmilMedve1 joined	14:40
	→ markoa joined	14:41
	→ jrmuizel joined	14:41
	← markoa left	14:42
	← EmilMedve left	14:42
com4	How do i compare my repository with the origin?	14:43
offby1	I'm not sure you do, exactly	14:44
	You can compare commits, though	14:44
	→ EmilMedve joined	14:45
com4	I want to see which commits aren't on origin if any. there isn't a way to do that?	14:46
	← EmilMedve1 left	14:47
joevano	com4: im pretty sure git diff can do it with the correct options	14:48
com4	joevano: k, i'll look around. Thanks	14:48
	→ jfactor joined	14:49
spearce	com4: git log origin/master..master	14:50
offby1	ha, spearce beat me to it	14:50
com4	spearce: :D thanks	14:50
	← repror___ left	14:51
	← ekidd left	14:51
	→ jayallen joined	14:53
	→ roccoblues joined	14:53
bryanl	with git, how do i know i have changes to push?	14:54
offby1	git log origin/master..master :)	14:54
	or similar, depending on the branch.	14:55
JohnFlux	bryanl: git status	14:55
	bryanl: git cherry -v	14:55
offby1	actually in all those cases it might be worth doing "git fetch" first.	14:55
	it's conceivable that upstream has your changes even though you didn't put them there ... isn't it?	14:55
	... or does that word not mean what I think it means?	14:55
bryanl	What i'm trying to do is have my zsh prompt show that i have changes to push	14:56
	i already have it showing that i have changes to commit	14:56
	→ loincloth joined	14:57
	← dmathieu left	14:57
offby1	huh	14:57
	→ harinath joined	14:58
JohnFlux	bryanl: I'd like to have that too :-)	14:58
offby1	simply comparing the current branch with what it's tracking might suffice ...	14:58
bryanl	i think i just figured it :)	14:58
	let me go investigate	14:59
	i need to figure out how to get utf8 characters in my zsh prompt now	14:59
	→ FylGood joined	14:59
	offby1 flees	14:59
	→ jelmer joined	15:02
	← roccoblues left	15:02
	→ camwest joined	15:03
	→ offby1 joined	15:04
	→ lois[pac] joined	15:05
	← zinou left	15:05
	→ reprore joined	15:05
	← jelmer left	15:09
wwalker	I just ran "git add ." it added all the directories in the current directory, but only some of the sub directories and none of the files. wth???	15:09
	→ jelmer joined	15:10
	← saurabh1403 left	15:10
Grum	you have a .gitignore file	15:10
	and how do you 'check' if 'git add' actually added files in a subdirectory?	15:10
wwalker	Grum: yes: *~ .svn .DS_Store none of which match any of the skipped file	15:11
	← botanicus left	15:11
wwalker	Grum: git status	15:11
	doh	15:11
Pieter	can you show that output?	15:11
wwalker	nope, git commit, git push and then git clone confirm that most of the files are not there but some are	15:12
	→ peritus_ joined	15:12
Grum	so which files arent there?	15:12
Pieter	JohnFlux: having a hard time with the git checkout -b stuff? :)	15:13
Grum	and what happens if you git add a missing directory manually?	15:13
JohnFlux	Pieter: back me up here!	15:13
Pieter	:)	15:13
wwalker	Pieter: Grum: http://rafb.net/p/RqjpRl39.html	15:13
	← jrmuizel left	15:14
JohnFlux	Pieter: the trouble is that the developers are so used to git now, they don't understand that it's complex and frightning to new userss	15:14
Pieter	wwalker: well, it says 'changes to be committed' looks right to me?	15:14
	→ ceej joined	15:14
wwalker	Grum: all of the sub dirs and files in .vim were skipped	15:14
	→ jrmuizel joined	15:14
	→ rue joined	15:15
Grum	and you are sure they are there? :) -- got a find orso to show them?	15:15
JohnFlux	they can't conceive that a new user might use the commands in not quite the right way :-/	15:15
Pieter	wwalker: are they listed in 'git ls-files --others'?	15:15
wwalker	Pieter: I've don git commit..; git push; cd elsewhere; git clone; and the new dir has .vim so that part worked, but no contents of .vim.	15:15
Grum	what on earth is --others? :D	15:15
	man git-ls-files	15:16
Pieter	wwalker: that's unlikely, git doesn't do empty directories	15:16
Gitbot	Grum: the git-ls-files manpage can be found at http://git.or.cz/man/git-ls-files	15:16
wwalker	Pieter: No output from --others	15:16
	← hdevalence left	15:16
Grum	what does ls -al for dotfiles/.vim say?	15:16
wwalker	Grum: in the original, a LOT, in the clone . ..	15:17
Grum	you are adding files in dotfiles/ -- so what does it say about the .vim?	15:18
rue	Morning-ish! A slightly broader question: does anyone know of strategies devised for handling bugfixes and other backport-type stuff in a project? (I can elaborate if anyone cares:)	15:18
Grum	it recognizes it as a file .. i am sure git doesnt get that wrong :D	15:18
wwalker	for clarity... http://rafb.net/p/tPqI2926.html	15:19
Grum	whoha	15:19
	Pieter: he made git clone an empty dir o.O	15:19
wwalker	and http://rafb.net/p/F2lDQ771.html	15:20
	wwalker did not want an empty dir ...	15:20
Grum	it is not even possible to have an empty dir wwalker	15:20
	something is really fucked up :)	15:20
wwalker	that is what I thought....	15:20
	→ sbfaulkner joined	15:20
Grum	you are using the latest stable git right?	15:21
wwalker	no	15:21
	hang on	15:21
Grum	then you shall soon and try again =D	15:21
wwalker	git version 1.6.0.6	15:21
Grum	apt system i'm sure of it ;)	15:21
wwalker	Grum: ACK!!! sorry, I have to manage large numbers of systems, no apt here. Fedcora 10 is the client, CentOS 4 with the EPEL repo (for git only, then disabled) is the server	15:22
	git version 1.5.3.6 on the server	15:22
Grum	ah :D most stable apt repos have 1.6.0.6 :D	15:23
	but still	15:23
	this is rather odd o.O	15:23
	← willb left	15:24
wwalker	so, I don't appear to have done anything wrong at the command line, but rather have found a bug (possibly resolvable by upgrading git	15:24
Grum	could be .. but git in its design cannot handle empty folders	15:25
	i doubt there is any diskspace problems that would cause that	15:25
rue	Well, not so much 'cannot' as 'does not' handle	15:26
Grum	its cannot if it does not tbh	15:27
JohnFlux	Pieter: ;-)	15:27
	→ ekidd joined	15:29
wwalker	my wifes tupperware does not handle dog food, but it can....	15:30
	→ tal67 joined	15:30
offby1	with software, "cannot" == "does not".	15:31
wwalker	OK, I'll update both ends from the vendors and try again. Hopefully it'll just work. I really hate non-vendor code on servers	15:31
Modius	Does anyone here run a Git that can be pushed to on a windows server? What is considered optimal? The cygwin Git or get the the non-cygwin git? I currently have a cygwin/openssh one working but can't get it working on another server, want to revisit the problem.	15:31
rue	offby1: There is the distinction that "does not" implies knowledge	15:33
	→ vbabiy joined	15:35
	→ ijcd joined	15:36
	→ paulboone joined	15:37
	→ krawek joined	15:38
	→ marvil07 joined	15:38
	← FylGood left	15:39
	← boto_ left	15:39
shruggar	didn't Linus say that git /can/ handle empty folders, it would just require an extra type of entity for tracking folders, and would have a couple edge-cases when merging? There was a thread on it, which I think basically was resolved as "okay, we could do it, but nobody here wants to bother"	15:39
Grum	there was a thread on it -- tbh it should be implemented :D	15:40
	→ moonese joined	15:40
shruggar	though Linus' suggestion of how it would be implemented didn't make a lick of sense to me, adding a special type for "empty directory", rather than just adding a type for "directory"	15:41
	← moonese left	15:41
	→ jpic joined	15:42
rue	Non-empty directories are Tree nodes?	15:42
jpic	hello, how to stop tracking a file but not remove it?	15:42
	← moonese1 left	15:42
offby1	git rm --cached	15:43
jpic	thanks, nice	15:43
Grum	i dont see why you cannot reference an empty tree; trees are objects themselves right?	15:44
	← j416 left	15:44
	→ boto joined	15:44
	→ broken_ladder joined	15:45
rue	But empty directories are not trees?	15:45
	→ mbroeker joined	15:45
	← broken_ladder left	15:48
	← paulboone left	15:53
	→ gorozco joined	15:53
	→ hobodave joined	15:53
	CardinalNumber → ProperNoun	15:54
	→ tanek joined	15:54
	→ ruediger joined	15:56
ruediger	hi	15:56
	→ ashleyw joined	15:56
offby1	so you say.	15:56
	But time will tell.	15:56
t3d0	heh	15:57
ruediger	I made some modifications in a branch which I want to commit in another branch (master branch) and not in the current branch. When I try to checkout the master branch I get "error: You have local changes to '<file>' ...". How can I commit these changes to the master branch?	15:57
	→ ijcd_ joined	15:59
	← rue left	16:00
	→ roccoblues joined	16:01
	← lorandi left	16:01
Pieter	ruediger: git checkout -m, but you might get merge conflicts	16:01
	→ lorandi joined	16:01
Pieter	ruediger: because that file is different between the two branches	16:01
mfilipe	hi! I'm at crud branch and I want execute "git pull" to master branch and execute "git rebase" to crud. how do I do this?	16:02
	← gilimanjaro left	16:02
	← hdl2 left	16:03
	→ saurabh1403 joined	16:05
	→ robomalo joined	16:05
	← rubydiamond left	16:05
	→ Frando_ joined	16:06
bwalk_	mfilipe: the way you said	16:07
	← ashleyw left	16:07
	← brasko left	16:07
	← jfactor left	16:07
	gorozco → p4tux	16:07
	← vuf left	16:07
	→ ashleyw joined	16:08
	← johan-s left	16:08
	← roccoblues left	16:09
mfilipe	bwalk: I'm working in a company that use git. It has a team working on this repository. I create a branch called "crud" to work. I want get patches of remote repository to this branch (crud) and I think that I have give a "git pull" in master and execute a rebase on "crud" branch. right?	16:09
	← rovalent left	16:10
bwalk_	mfilipe: what do you mean by patches of the remote to this branch?	16:10
facedownH	is git ideal over say, svn and mercurial for a small size team ( 20-30ish ) whereas 70-80% are familiar only with CVS/Windows as of now? Yeah, we're still using CVS.	16:10
offby1	facedownH: depends how willing they are to learn, and how much they enjoy using the command line	16:11
	git will be more of a brain-stretcher for them than svn	16:11
mfilipe	bwalk_: origin/master branch is a remote repository with new patches (commits)	16:11
offby1	and the GUI stuff on Windows is way more primitive than say TortoiseSVN	16:11
mfilipe	I want get this commits to my crud branch	16:12
bwalk_	mfilipe: why not to your master branch?	16:12
facedownH	offby1: you mean to say TortoiseGIT wont be sufficient?	16:12
	offby1: Ive heard Mercurial and its commands are somewhat more like cvs/svn	16:13
mfilipe	bwalk_: so, it is that I want say... I want get commits of remote repository to master branch and, with new commits, I want execute a rebase in crud branch	16:13
	→ C-Otto joined	16:13
C-Otto	hi there	16:13
offby1	facedownH: tortoisegit isn't close to Ready For Prime Time in my humble opinion.	16:13
	→ jerome187 joined	16:13
	→ qhoxie joined	16:14
offby1	facedownH: it's true, mercurial's commands are more similar to svn's than are git's.	16:14
bwalk_	mfilipe: you want to put your work in crud on top of the upstream master?	16:14
	→ cgardner joined	16:14
offby1	facedownH: I think you're going to find it's _really_ hard to decide which to switch to.	16:14
C-Otto	i'd like to ignore a file which nevertheless is included in the repository. the indended meaning is: the committed file defines default values, the local file (which is ignored!) contains configuration settings	16:14
offby1	Because you can't really judge one of these tools without having used it for a while for real work	16:15
	and that's expensive :-\|	16:15
C-Otto	that does not work with .gitignore as soon as i add the file (using -f)	16:15
	← jackdempsey left	16:15
	→ sitaramcell joined	16:15
offby1	C-Otto: this was a super-frequently-asked question in #svn. The answer here is likely the same as the answer there: check in a _template_ instead, and have a build process of some sort generate the real config file from the template.	16:15
mfilipe	bwalk_: no, I want get the new commits to my branch, because I need them.	16:15
offby1	Thus the template needn't change much	16:15
	← giallu left	16:16
C-Otto	offby1: i never used SVN, thanks though :)	16:16
	→ sitaram joined	16:16
mfilipe	bwalk_: Will I need execute "git checkout master" > "git pull" > "git checkout crud" > "git rebase master"?	16:17
	← ijcd left	16:17
bwalk_	mfilipe: that would merge your crud branch onto of your master	16:18
	→ stephenjudkins joined	16:18
bwalk_	mfilipe: which is one way of doing it, don't know, if it is the best	16:18
	← krawek left	16:18
	← sitaram left	16:20
beket	Hi people:) Is there a "theis" merge term ? I need to pull from remote & if there are any conflicts at all, discard my version & keep their changes.	16:20
mfilipe	bwalk_: thanks	16:22
	:)	16:22
bwalk_	no problem, if that's what you wanted	16:22
	→ bergwolf_ joined	16:22
	← Frando left	16:23
	→ tongueroo joined	16:23
	← bergwolf_ left	16:23
	→ cyphactor joined	16:23
beket	hm, how do i abort a merge ?	16:24
JohnFlux	beket: yeah there really should be a --abort option imho	16:24
	beket: I think the best way is git reset --hard IF you had no uncommitted changes	16:25
beket	thanks JohnFlux	16:26
	→ broken_ladder joined	16:28
	→ willb joined	16:31
	← sitaramcell left	16:32
	→ LiamH joined	16:32
	← cyphactor left	16:32
	← countd left	16:33
	← sgrimm left	16:33
	→ paulboone joined	16:33
	← vbabiy left	16:35
	→ JohnnyL joined	16:35
	← paulboone left	16:35
	← JohnnyL left	16:35
	→ dsal joined	16:36
	→ mw-home_ joined	16:39
	← broken_ladder left	16:40
	→ jfactor joined	16:41
	→ b_jonas joined	16:42
	→ mgarriss joined	16:44
	→ Orangebat joined	16:45
	→ roccoblues joined	16:45
	← C-Otto left	16:45
	→ alley_cat joined	16:46
	→ jceb81 joined	16:47
	→ mrball joined	16:47
	← cannonball left	16:47
	→ binjured joined	16:48
	mrball → cannonball	16:49
	← idahoev1 left	16:49
	← roccoblues left	16:49
	→ tvw joined	16:50
	← mw-home left	16:50
	← mbroeker left	16:51
	← jceb81 left	16:51
Modius	Can't get away from the "remote end has ended connection" with my new Git/OpenSSH/Cygwin Windows server - is there any place I can check the log for this failure?	16:52
	← ngulden left	16:52
	→ jceb joined	16:52
Ilari	Modius: Maybe the URL is wrong? Or it can't connect to remote host?	16:54
b_jonas	okay... so which program should I use to migrate part of an svn repo to a git repo? git-svn or something else?	16:55
Ilari	Modius: Does 'ssh user@host git --version' work?	16:55
	b_jonas: Git-svn plus some fixing up...	16:55
Modius	Ilari: Yes, that works	16:56
	→ jdq joined	16:56
Ilari	Modius: Only outputs version number?	16:56
Modius	Ilari: "git version 1.6.1.2"	16:56
	→ xyc joined	16:57
Modius	Are there any server-side logs I can check for OpenSSH?	16:57
Ilari	Modius: What messages does the failing operation give?	16:57
Modius	fatal: The remote end hung up unexpectedly	16:58
Ilari	Modius: Anything else?	16:58
Modius	No	16:58
	→ johan-s joined	16:58
Ilari	Wonder if it even connects, because IIRC using wrong URL with ssh:// should give some other error message also...	17:00
	← ntoll left	17:00
	→ broken_ladder joined	17:01
schweizer	is it possible to push a branch to the server?	17:01
dsal	schweizer: Of course.	17:01
schweizer	but i dont want to merge my branch with the master on the server	17:01
	i just want to add my branch to the available branches on the server	17:02
	dsal: how would i do that?	17:02
	← broken_ladder left	17:02
	→ cyphactor joined	17:02
Ilari	schweizer: 'git push <remote-nick-or-URL> <branch-to-push>'.	17:02
schweizer	thx	17:02
	← Yuuhi left	17:03
	xinming_ → xinming	17:03
schweizer	and how can i checkout that branch from the server in another/new repo?	17:04
	← jfactor left	17:04
schweizer	git clone only gets me the master from server	17:04
	...i need a good cheat sheet :)	17:05
jast	schweizer: you sure about that? try git branch -r :)	17:05
	← sbfaulkner left	17:05
PerlJam	schweizer: clone gets you the branches too, they just aren't local branches	17:05
jast	(also, git checkout -b localbranchname origin/remotebranchname to get a local branch from a remote tracking branch)	17:05
schweizer	jast: hey thx	17:05
jast	(or, starting from git 1.6.1, git checkout --track origin/remotebranchname to save keystrokes)	17:06
b_jonas	hmm, I'll have a problem here…	17:06
Modius	I have the same client able to connect to one windows server cygwin repo and not a second - both have the repo in the same directory, I am only changing the machinename in the full ssh://blahblah connection string. This leads me to believe that "The remote end has hung up" is server-relsated. Is there any way I can check the log on the server side to find out what went wrong?	17:06
schweizer	cant i just do: git checkout origin/branchname ?	17:06
Modius	Server is cygwin/openssh	17:06
b_jonas	no, I don't. it somehow works automatically. cool	17:06
jast	schweizer: yes, but then commits you do on top of that will be lost as soon as you switch back to some other branch	17:06
b_jonas	uh huh, I can see why	17:07
schweizer	ah ok	17:07
	i think i go with the --track version	17:07
jast	schweizer: basically, remote tracking branches mirror the remote branches, so changing them isn't allowed	17:07
	(remotes branches = those in the upstream repo)	17:07
b_jonas	git svn fetch is running now…	17:08
schweizer	jast: so the only possiblity is to use checkout -b (which gives the branch a new name) ?	17:08
	← garo left	17:08
	→ roccoblues joined	17:08
schweizer	...because i want to change the branch	17:08
	→ garo joined	17:08
	← shadowhand left	17:08
jast	schweizer: yeah, you can use the same name actually, without the origin/ prefix of course (I was just trying to show you don't have to). or --track if your git is recent enough.	17:09
schweizer	jast: ok thx	17:09
jast	basically that "copies" the branch and creates the appropriate setup to make fetch/pull without arguments work	17:10
	→ porcnick joined	17:11
	→ rlorandi joined	17:11
	→ sekimura joined	17:12
ricky_clarkson	I edited two files, one on purpose, one accidentally. How can I restore the accidental one to its last committed?	17:12
	← roccoblues left	17:13
	→ bronson joined	17:13
bwalk_	ricky_clarkson: git checkout -- file	17:13
	← Suprano left	17:14
	→ Suprano joined	17:15
	→ rubydiamond joined	17:15
b_jonas	ok, git svn seems to have worked. thanks	17:15
ricky_clarkson	Thanks.	17:16
	→ vbabiy joined	17:16
rubydiamond	b_jonas: hey..	17:16
	← mgarriss left	17:17
rubydiamond	I am getting this error	17:17
	http://pastie.org/414341	17:17
	when I try to git svn rebase	17:17
	← orafu left	17:18
	← lorandi left	17:19
	→ sgrimm joined	17:19
	→ jfactor joined	17:20
	← ciskje left	17:20
b_jonas	git diff -M is not finding my renames	17:21
	→ paulboone joined	17:21
	→ boto_ joined	17:21
rubydiamond	anybody here for helping	17:21
	http://pastie.org/414341	17:21
	→ sunoano joined	17:22
Orangebat	can anyone tell me how git decides if it's going to open and read a file or not? I'm having trouble with a repo on ubuntu that now takes 9 seconds to 'git stat'	17:22
b_jonas	all I did was move all files to a new sub directory yet git diff -M does not find it	17:22
Aides	rubydiamond: you need to stash it or commit it	17:22
rubydiamond	Aides: I just don't want to commit	17:22
Aides	then stash	17:22
rubydiamond	as I want to svn update first	17:22
	then after that commit the files	17:22
Orangebat	an strace shows that it's opening a lot of the files, and the lstat/open/mmap/close is taking too long	17:22
Aides	git makes it easy to alter commits	17:22
	if you need to	17:23
rubydiamond	Aides: if I do git stash	17:23
dsal	rubydiamond: I'll often commit and reset --hard HEAD^ (just so	17:23
	that I have it if I change my mind later)	17:23
rubydiamond	I would have to make changes again	17:23
Aides	rubydiamond: why would there be stash then?	17:23
rubydiamond	dsal: but everybody in team will come to know.. what bad commit you did	17:24
Aides	you can then 'git stash apply' your changes back	17:24
dsal	rubydiamond: No, nobody will see it.	17:24
rubydiamond	Aides: I want only some of files to be committed..	17:24
dsal	rubydiamond: But it will be in your reflog if you want it back.	17:24
rubydiamond	those might be updated by some other svn using developer	17:24
	← paulboone left	17:24
	← ashleyw left	17:24
	→ bmalee joined	17:24
rubydiamond	I am not sure while committing to git.. that file is latest or not	17:24
ricky_clarkson	Can git log tell me which files have changed, been deleted, removed etc.?	17:25
dsal	rubydiamond: Always commit before brining in new code.	17:25
	→ PeakerWork joined	17:25
	→ Fink_x joined	17:25
dsal	ricky_clarkson: It does in many ways. git log --stat for example	17:25
b_jonas	ah, but git commit does find them in the status message. I wonder why git diff doesn't then. or maybe it does but does not show them in the output.	17:25
rubydiamond	suppose I git commit and found that that file was changed by another developer.. so .. would git svn dcommit fail ?	17:25
PeakerWork	bzr has a "stash"-like feature that lets me interactively go through each hunk in the diffs. Does "git stash" or something else exist to do something similar?	17:25
dsal	rubydiamond: rebase before you dcommit	17:26
	dcommit ~= push	17:26
rubydiamond	dsal: but it says... http://pastie.org/414341	17:26
	before doing rebase	17:27
dsal	PeakerWork: When I want to do something like that, I just commit.	17:27
rubydiamond	I just dont want to commit as my changes are so local to my machine	17:27
dsal	rubydiamond: Yes, that's why I said commit first.	17:27
PeakerWork	dsal: but I have several different stuff in there	17:27
rubydiamond	I have changed some constants to suit my machine..	17:27
PeakerWork	dsal: on the same files	17:27
rubydiamond	I want to git svn rebase ..	17:27
dsal	PeakerWork: git add -p	17:27
	← tal67 left	17:27
dsal	rubydiamond: You need to fix your source tree so you don't modify	17:28
	stuff you don't want to commit. If you want the changes back, stash	17:28
	is a fine way to do it.	17:28
	→ tal67 joined	17:28
dsal	Is my client actually sending three messages when it wraps?	17:28
Aides	dsal: yep	17:28
rubydiamond	dsal: what will happen, if I stash	17:28
	dsal ugh. When did that happen?	17:28
	← tal67 left	17:28
rubydiamond	and the do git svn rebase	17:28
	→ d0k joined	17:29
dsal	rubydiamond: Stash will temporarily set your changes aside. You can get them back later.	17:29
rubydiamond	need I do git svn rebase again when I apply stash	17:29
	dsal ah, auto-fill mode in an erc buffer does gross stuff	17:29
ricky_clarkson	dsal: Thanks.	17:30
dsal	rubydiamond: Most importantly, fix your build so that you aren't changing revisioned files. What you haven't committed won't be pushed.	17:30
ricky_clarkson	Is there a way of having that more human-friendly?	17:30
	← tango_ left	17:30
Modius	Is there a way for me to manually SSH in to a bash and replicate the instructions that git-clone or git-push runs to see why they're failing and I'm getting a "the remote end hung up unexpectedly"?	17:30
	→ altrux joined	17:30
dsal	ricky_clarkson: I think that's quite human friendly. What are you looking for?	17:30
	→ perplexes joined	17:30
rubydiamond	dsal: If I do git svn rebase after doing stash	17:31
	would it be available	17:31
	when I do .. git stash apply	17:31
dsal	rubydiamond: git stash pop	17:31
ricky_clarkson	dsal: svn log gives lines like A newfile, M oldfile, D deletedfile	17:31
dsal	rubydiamond: Please experiment with it.	17:31
ricky_clarkson	which I normally pass through sed to construct a daily report.	17:31
PeakerWork	does "git diff" not have an option to ignore some regexp pattern? Like diff's -I ?	17:31
	← boto left	17:32
dsal	ricky_clarkson: git log --name-status	17:32
	→ Khisanth joined	17:32
dsal	(though I don't think that's more friendly)	17:32
ricky_clarkson	Amazing. Thanks.	17:32
	← jfactor left	17:32
ricky_clarkson	I can agree with you, I personally prefer reading --stat's output.	17:32
dsal	ricky_clarkson: daily report: git diff --stat '@{yesterday}'..	17:32
ricky_clarkson	heh	17:33
	→ bcarlyon\|laptop joined	17:33
Orangebat	I'm seeing git lstat/open/mmap2/close/munmap the same large files 4 times during a status, any idea of the cause?	17:33
	← cyphactor left	17:35
	← peritus_ left	17:35
	← jfkw left	17:35
	bcarlyon\|laptop → BarryCarlyon	17:35
b_jonas	Orangebat: stupid question, but is there an execve between them?	17:35
	→ cyphactor joined	17:36
	→ peritus_ joined	17:36
	→ jfkw joined	17:36
	← cyphactor left	17:36
	← jfkw left	17:36
	→ cyphactor joined	17:36
	→ jfkw joined	17:36
ricky_clarkson	Can '@{yesterday}' work with git log too?	17:36
	→ rue joined	17:37
	→ paltman94 joined	17:37
ricky_clarkson	--since='@{yesterday}' :)	17:37
Orangebat	b_jonas: nope. It seems git has for some reason decided to re-check my files every time I run status now	17:38
	b_jonas: instead of just lstat'ing them	17:38
rubydiamond	what are other awesome commands, I could use with git svn	17:39
	← Finkregh left	17:40
	→ markelikalderon joined	17:41
	← shruggar left	17:41
	← reprore left	17:42
	→ reprore__ joined	17:42
	← jmesnil left	17:44
seanius_	is there an easy way to edit an initial commit to remove a directory and re-add it as a second commit?	17:44
	→ tek joined	17:44
	← tek left	17:44
	→ tek joined	17:44
	→ wsc joined	17:46
	→ broken_ladder joined	17:46
	← kbingham left	17:47
	← girishr left	17:48
	→ jfactor joined	17:51
	→ RodP joined	17:53
	← lfoppiano left	17:53
	tek → tek0	17:54
	← markelikalderon left	17:55
	→ jadams joined	17:57
jadams	I have a remote that has a commit I missed from back in september. How do I merge in a single commit from a remote? It will apply cleanly...	17:58
jast	jadams: it's not technically a merge, but fetch and then use git cherry-pick	18:00
seanius_	cherry-pick?	18:01
	→ RaySl_ joined	18:01
jadams	jast, seanius_: thanks	18:02
	← arcame left	18:02
	→ ice799 joined	18:04
jadams	if the commit is 51ea50 on a remote, I can't just git cherry-pick 51ea50, I have to specify the remote...right?	18:05
jast	after you've fetched the commit ID is enough	18:06
jadams	ok thanks, didn't notice this machine was missing the remote I was concerned with	18:06
	thanks :)	18:06
	→ dmlloyd_ joined	18:06
	← RaySl left	18:07
	→ jerbear joined	18:07
	← kraymer left	18:07
	← aisbaa left	18:08
	→ markelikalderon joined	18:09
	→ csgeek joined	18:10
csgeek	trying to do a merge..and it's saying I have a conflict... how do I tell it to take the version from a particular branch	18:10
PerlJam	just checkout that particular file from that particular branch	18:12
csgeek	thank you	18:14
	← csgeek left	18:15
	← peritus_ left	18:15
	dmlloyd_ → dmlloyd	18:17
	→ peritus_ joined	18:20
	← gaveen left	18:21
	→ paulboone joined	18:22
	→ gaveen joined	18:22
	← jceb left	18:23
	→ defn joined	18:23
	→ jceb joined	18:24
defn	Hey everyone. I am trying to set up Git on a VM at work. The VM is behind RSA. In order to access the box from the outside I need to use my RSA keyfob to ssh into a gateway, and then into my VM.	18:24
	→ netoman joined	18:24
defn	Is there a way to get access to that VM's git repo from the outside world in a way that doesn't suck(tm)?	18:24
	← tek0 left	18:25
	→ tek0 joined	18:25
JohnFlux	defn: you can do git over ssh	18:27
	defn: or maybe ssh in, run git instaweb, then ssh in again with port forwarding	18:27
defn	JohnFlux: yes, but is there a way that would allow me to seamlessly access that VM behind the gateway, or at least, make it seem that way	18:27
jast	defn: ssh supports port forwarding/tunneling, you could do that in the part where you ssh to the gateway, and just keep it open in the background	18:27
	← JohnFlux left	18:27
jast	unless they were mean and disabled it	18:28
	jayallen → jayallen-afk	18:29
fdv	considering git-filter-branch, is there a tool somewhere to either visualize or somehow list the sizes of objects in a repo? What I'm looking for are the largest objects committed throughout the history of all branches.	18:29
	→ plediii joined	18:29
fdv	I guess git-filter-branch could be used for that purpose as well, but it's probably not very efficient	18:30
	→ shruggar joined	18:30
dsal	fdv: This might help you: http://stackoverflow.com/questions/298314/find-files-in-git-repo-over-x-megabytes-that-dont-exist-in-head	18:30
fdv	dsal: thanks, that url formulated my question much better than I did :)	18:32
	→ csgeek joined	18:32
	→ FFighter joined	18:33
FFighter	how can I checkout a specific file from a specific commit?	18:34
jast	FFighter: well, how about, say, git checkout <commit> <file> ;)	18:34
	← csgeek left	18:34
FFighter	jast, that's easy uh?	18:35
	thanks	18:35
	:)	18:35
	→ csgeek joined	18:35
	← peritus_ left	18:36
	→ kumbayo joined	18:38
	→ Voker57 joined	18:38
	← tek0 left	18:38
	→ stuffcorpse joined	18:38
	→ tek0 joined	18:38
	→ bobesponja joined	18:40
rubydiamond	any git svn use in	18:43
	test	18:44
	← rubydiamond left	18:45
	← wsc left	18:47
	← Suprano left	18:51
	← mfilipe left	18:51
	→ Suprano joined	18:52
	→ johnw joined	18:53
	← sekimura left	18:56
	→ sekimura joined	18:57
	← radarek left	18:58
	→ Sigma joined	18:59
	→ dlisboa joined	19:00
dlisboa	I seem to have gone dumb all of a sudden, or maybe I never knew this, but: is there a way to unstage files that is not through add-interactive?	19:01
bmalee	dlisboa: git reset <file> ?	19:02
dlisboa	I swear I tried this...will do again, just to be sure	19:02
	no, I didn't, and it's not this actually	19:03
	→ sanity- joined	19:03
b_jonas	dlisboa: do you want the files appear deleted next commit, or appear unchanged?	19:03
dlisboa	I mean, you do some changes, add them to staging area, but realize you don't want that in this commit, but you still would like to keep the changes.	19:03
	→ galderz joined	19:04
b_jonas	the latter than	19:04
	I too think it's git reset. what's your problem with that command?	19:05
	← galderz left	19:06
dlisboa	b_jonas, syntax, it appears. Missed the '--'	19:07
	thanks	19:07
	→ galderz joined	19:07
	← galderz left	19:08
	→ Frando joined	19:09
	→ w00t joined	19:10
marvil07	a blob is always a file ?	19:10
	if not, where can I find an example?	19:10
	← dlisboa left	19:11
	→ scientes joined	19:11
	← FFighter left	19:15
	← ijcd_ left	19:15
	→ roccoblues joined	19:16
	← plediii left	19:19
	→ FFighter joined	19:20
tvw	I did some changes to my files, but I did not add or commit. How can I undo those changes with git?	19:20
	→ wsc joined	19:21
	← FFighter left	19:21
parasti	tvw: git checkout $file	19:21
	→ FFighter joined	19:21
	← wsc left	19:21
	→ wsc joined	19:21
	← roccoblues left	19:22
drizzd	marvil07: depends on what you define as a "file". A blob can also contain the target of a symbolic link, for example. It's just data.	19:22
	→ mfilipe joined	19:22
tvw	parasti: doesn't this just checkout the given file? I want to reset my project to the state of the last commit	19:23
	→ lightcap joined	19:23
	← mw-home_ left	19:23
parasti	tvw: it matches $file to what's in the index	19:24
	tvw: to match index and working tree to the currently checked out commit, git reset --hard	19:24
	← w00t_ left	19:24
	← Frando_ left	19:24
parasti	tvw: of course $file can be a directory name as well, including the current dir	19:24
	← Frando left	19:25
	→ Frando joined	19:25
tvw	parasti: Thanks. I will check the documentation for git reset. but mayby git checkout $dir could be enough.	19:26
marvil07	ok drizzd, I understand that, but I heard something like "git do not track files, it tracks contents" (like chunks of files, insted of files) is that true?	19:26
	→ jlec joined	19:27
jlec	hey guys	19:27
	I am fetch a svn repo like this:	19:27
	git --bare svn fetch	19:27
	→ seanius joined	19:27
parasti	tvw: if you only have unstaged changes, "git checkout ." and "git reset --hard" should have the exact same effect	19:28
jlec	but only the initial fetch is merged into master	19:28
	How can I get the following fetches also merged into the master	19:28
parasti	tvw: that is, given you're in the root dir of your repo	19:28
	→ marcelo_ joined	19:29
	← FFighter left	19:29
tvw	parasti: I just did the last commit yesterday. Today I did some changes, but decided to return to the state of yesterday.	19:29
parasti	tvw: if you're not sure you really want to drop your changes completely, have a look at man git-stash	19:31
Gitbot	tvw: the git-stash manpage can be found at http://git.or.cz/man/git-stash	19:31
	→ travisjeffery joined	19:31
	← marcelo_ left	19:31
	→ marcelo_ joined	19:32
	→ marcelo__ joined	19:33
	← marcelo_ left	19:33
	← reprore__ left	19:35
	← marcelo__ left	19:35
	→ marcelo__ joined	19:35
	→ marcelo_ joined	19:37
	← marcelo__ left	19:37
tvw	parasti: git reset --hard HEAD was what I needed: Thanks!!	19:38
	→ marcelo__ joined	19:39
	← marcelo_ left	19:39
	← ruediger left	19:39
	← marvil07 left	19:40
	→ marcelo_ joined	19:42
	← marcelo__ left	19:42
	← bdiego left	19:43
	← seanius left	19:45
	→ FreakGuard joined	19:46
FreakGuard	how to convert bzr to git?	19:46
	← alley_cat left	19:46
	→ drewr joined	19:46
	→ marcelo__ joined	19:48
	← marcelo_ left	19:48
	→ reprore_ joined	19:49
	→ tango_ joined	19:51
	→ segfaux joined	19:52
	← marcelo__ left	19:52
	→ marcelo__ joined	19:52
	→ mbroeker joined	19:53
	← jlec left	19:55
prlawrence	can anyone hit www.google.com?	19:58
Fatal	works fine here	19:58
prlawrence	nevermind, its back :-)	19:58
	← cilly left	19:58
	thumper_laptop → thumper	19:59
dsal	prlawrence: They go down a lot.	19:59
prlawrence	dsa1: lol? I've never before had a problem. Actually we're still hit or miss here.	20:00
dsal	prlawrence: I've got a third-party xmpp bot [email@hidden.address] -- you can ask it to monitor stuff for you, or just one-off requests (or to tell you when a broken site comes up)	20:01
	→ Aides joined	20:01
	→ rudi_s joined	20:02
	← blockowitsch left	20:03
	← jerbear left	20:04
	← reprore_ left	20:04
prlawrence	dsa1: cool. however, your bot like to tell me "no such command" and isn't replying to 'help'. my pidgin client just says the bot is typing :-)	20:10
dsal	prlawrence: bizarre. help is working for me.	20:11
	To be somewhat topical, the project is in a public git repo, so I'd welcome a fix. :) http://github.com/dustin/whatsup	20:12
prlawrence	dsa1: OK, thx for the tip anyway. I'll stop complaining about my obviously PEBKAC troubles for a while ;-)	20:12
	→ loic joined	20:13
prlawrence	dsa1: I was just trolling the repo looking for a command reference :-)	20:13
	→ mrbats joined	20:14
	← Sigma left	20:14
mrbats	hello	20:16
loic	hi, I've cloned a repository, created a local branch and now I'd like to push this branch remotely, how can I do this ?	20:16
mrbats	quick question - after I update my .gitignore file how do I get git to register the changes	20:16
	← Aides left	20:16
	→ pgregory joined	20:17
mrbats	i.e. i have added my development database to gitignore but it is still tracking it	20:17
	→ Aides joined	20:17
pgregory	Ilari: are you around?	20:17
mrbats	loic: this might get you started http://github.com/guides/setting-up-a-remote-repository-using-github-and-osx	20:18
pgregory	if not, is anyone else around who might be able to help me grafting SVN branch merges as part of a permanent switch from SVN to git?	20:18
	→ marvil07 joined	20:19
	MisterN → Foobar	20:19
	Foobar → zdujfjf	20:19
	zdujfjf → MisterN	20:19
	marvil07 is in doubt	20:21
marvil07	I heard something like "git do not track files, it tracks contents" (like chunks of files, insted of files) is that true?	20:21
	← noth left	20:22
jast	it doesn't track chunks of files but the entire file contents	20:22
	what people mean is probably that the contents are stored completely independently from the filename	20:22
ToxicFrog	Yes.	20:23
jast	so you can have two identical files in a repository and git stores the contents just once	20:23
mrbats	quick question - after I update my .gitignore file how do I get git to register the changes	20:24
	→ kumbayo_ joined	20:24
jast	mrbats: the changes are picked up right away, but note that .gitignore only applies to files that aren't already tracked by git	20:24
	← kumbayo left	20:25
mrbats	so how do I un-track files	20:25
	← tek0 left	20:25
mrbats	??	20:25
ToxicFrog	git rm --cached foo	20:25
	See also man git-rm	20:25
mrbats	ok off to try - thanks guys	20:25
Gitbot	ToxicFrog: the git-rm manpage can be found at http://git.or.cz/man/git-rm	20:25
jast	mrbats: by having them removed in the next commit. do what ToxicFrog said and then eventually commit that. presto, no more of that file afterwards.	20:25
mrbats	will it delete it	20:25
jast	as long as you use --cached, no	20:25
mrbats	ok cool	20:26
	← pgregory left	20:26
jast	only from the index, so it's no longer in the next commit	20:26
	→ plediii joined	20:26
	→ jerbear joined	20:26
	→ hpatwork joined	20:27
	→ tal67 joined	20:29
loic	mrbats: I've tried a git push origin my_local_branch	20:29
	mrbats: but it still barks that the branch does not exist on origin	20:29
	→ noth joined	20:29
	← sanity- left	20:29
loic	so I think the question is, how do I create the branch remotely :)	20:30
Voker57	git push :branch	20:30
loic	(I've write access on origin)	20:30
Voker57	oh, create	20:30
	just push it	20:30
ToxicFrog	git push origin branchname should work.	20:30
loic	well	20:31
	The destination refspec neither matches an existing ref on the remote nor ... blahblah	20:31
mrbats	if you followed the instructions on that page I linked to command should be git push 'github' master	20:31
	not 'origin'	20:31
	depends what you called remote in your config file	20:32
	← saurabh1403 left	20:32
loic	I've clone a repository of mine through ssh	20:32
	cloned	20:32
	it's not github	20:32
	there's already an "origin" entry	20:32
	→ mfwitten joined	20:32
	→ MisterN_ joined	20:34
	← MisterN_ left	20:34
mfwitten	I just wanted to throw this out there: If someone here well-versed with git could help get KDE moved to git fully, it would be a major boost to git's userbase; they are currently considering various tools (though it looks like they are leaning toward git)	20:34
	(from what I've read)	20:34
	Plus you might get some C++ contributsion ;-D	20:35
Voker57	mfilipe: what help exactly is needed?	20:35
dsal	mfwitten: What's the hold-up?	20:37
mrbats	loic: sorry mate probably past my knowledge now	20:37
loic	mrbats: ok, that was a PEBCAK	20:37
	my bad	20:37
	the local branch was created on another repository ...	20:38
Voker57	oops, that was to mfwitten	20:38
loic	now it works	20:38
mrbats	loic: cool glad you got it working	20:38
loic	so the answer was, git push origin local_branch_name	20:38
	← jerbear left	20:38
	→ seezer_ joined	20:39
	← perplexes left	20:39
	→ tvakah_ joined	20:39
	→ pgregory joined	20:41
pgregory	hi all	20:41
	→ shweazel_ joined	20:41
pgregory	if I've cloned an SVN repo with --no-metadata, how do I then do the necessary grafting of merges, as I don't seem to be able to do a "git-svn log r????"	20:42
Ilari	mfwitten: The git core devs seem to have fair amount of C++ haters... :-)	20:42
	→ snitko joined	20:42
	→ lenbust__ joined	20:42
	→ tmz joined	20:43
	→ Jc2k_ joined	20:43
Ilari	pgregory: Filter-branching git-svn repository really screws git svn...	20:43
pgregory	Ilari: Filter-branching?	20:44
	all I did was clone it.	20:44
shd	Ilari: not to mention how it screws up with svn	20:44
	← jceb left	20:44
Ilari	pgregory: --no-metadata might also screw with git svn operation just by itself.	20:44
	→ vanweerd joined	20:45
pgregory	Ilari: wasn't it you yesterday that said I must do --no-metadata?	20:45
	→ Andphe joined	20:45
	→ robinr joined	20:45
	← ceej left	20:45
Ilari	pgregory: It wasn't required, but its recomended for imports...	20:46
	→ ceej joined	20:46
pgregory	Ilari: so now, is there any way I can get the SVN information needed to do the merge grafting?	20:46
	→ jceb joined	20:46
	← mrbats left	20:47
mfwitten	Voker57: I'm not sure what's needed, I just know that various people are considering git, mercurial, bzr, etc. There is now a #kde-git and you can checkout http://mail.kde.org/pipermail/kde-scm-interest	20:47
	dsal: I assume you mean why don't I help out? Well... I don't know KDE or git well enough to provide excellent help.	20:47
dsal	mfwitten: Nah, I mean what's slowing them down from choosing.	20:48
	I just spent some more time with bzr. I still don't like it.	20:48
	← plediii left	20:48
mfwitten	dsal: What in particularly don't you like? (I've never used it).	20:48
	→ Adman65 joined	20:48
Ilari	pgregory: You probably can get it from dates of commits and what branches they are on...	20:48
mfwitten	dsal: What's slowing them down is culture.	20:48
pgregory	Ilari: I have a commit message like "Merge from trunk 2457:2550" and then later "Merge from trunk 2551:2590" and so on.	20:49
dsal	mfwitten: Giant, slow codebase. Some operations (e.g. clone) are avoided just because it's so slow. I really just wanted add -p today.	20:49
pgregory	is there any way I can make those merges make sense in git?	20:50
	→ ciskje joined	20:50
dsal	I was very much a mercurial user before git. I still don't mind hg, but git makes a lot of things easier -- and my repo sizes are likely to be much smaller.	20:50
mfwitten	ah	20:50
	← cannonball left	20:50
Andphe	Hi, sorry this noob question, but how can I grab a specific branch from a repository ?	20:51
Ilari	pgregory: Find next older commit on trunk and use it as commit merged?	20:51
dsal	Andphe: A branch is just a pointer to an edge of a graph. You likely already have them all. What are you wanting to do?	20:51
Andphe	there is piece of software, forked and the changes are in a branch	20:51
	I need to determine those changes	20:52
dsal	Actually, that's one thing I really didn't like in hg -- a branch was a name encoded into a commit, never to be removed, and confusing to reuse.	20:52
Andphe	so I have a copy of the original software	20:52
mfwitten	dsal: Wouldn't you say that it's a pointer to a node in the graph?	20:52
dsal	Andphe: You mean it's in someone else's repo? You can add it as a remote and fetch.	20:52
pgregory	Ilari: sorry, not with you.	20:52
dsal	mfwitten: I would. I think I might be kind of high right now. Weird tea mix.	20:52
mfwitten	haha	20:52
Ilari	pgregory: Find the date of the merge commit. Now, find newest commit on trunk thats older than that.	20:53
Andphe	dsal, I cloned from a repo a software , anyone else forked this software and have his own repo, but the changes that I need are in a branch of the second repo	20:53
	dsal, I want to compare both codes	20:54
	this is my first time with git	20:54
dsal	Andphe: git remote add someguy git://other/repo	20:54
	Andphe: git fetch someguy	20:54
	Now you'll have everything you need. :)	20:54
	Andphe trying	20:54
pgregory	Ilari: if I do "git log -1 --grep=Merge --parents <branch>" I get two sha1 codes, which I understand are the first and second values in the graft, the sha of the previous (older) commit on the trunk is the third?	20:55
Ilari	pgregory: Yes.	20:56
	→ fr\| joined	20:56
	fr\| → fr	20:57
Ilari	pgregory: Its useful to use --date-order when looking for newest older commit so listing is in order from newest to oldest.	20:57
	← qrush left	20:58
Ilari	pgregory: The most usable format for time could be unix time ('%at' format string)	20:59
pgregory	Ilari: I was just going to look in gitk, is there a better way?	21:00
	← FreakGuard left	21:00
	← snitko left	21:00
	→ FreakGuard joined	21:01
	→ peritus_ joined	21:01
	← esparkman left	21:02
marvil07	thanks jast	21:02
	marvil07 really late :p	21:02
	← peritus_ left	21:02
jast	I even forgot what this was about by now :)	21:02
	pgregory notes, this IMHO is one of the weakest parts of git, compared with svn, if I do git-<tab> I get 133 commands, how the hell are you supposed to know which one you want?	21:03
	RaySl_ → RaySl	21:03
spb	by reading the documentation	21:03
	→ sfwc joined	21:03
ToxicFrog	pgregory: how is svn better? You type 'svn<tab>' and get one command and no idea which subcommand you need...you have to hit the docs either way.	21:03
pgregory	spb: I have read the documentation, many times, it's great at describing what the tools do, but how do I know which one I want when someone says I need to seach commit history in --date-order?	21:04
	→ _arc_ joined	21:04
pgregory	ToxicFrog: svn is better in that it has a very concise number of tools, each of which is clear in its purpose.	21:05
Ilari	pgregory: log...	21:05
ToxicFrog	Well, if you've read the documentation, you know that "git log" is for viewing commit history.	21:05
	← camwest left	21:05
ToxicFrog	So man git log and look under "commit ordering".	21:05
	→ perplexes joined	21:05
	← ceyusa left	21:05
pgregory	ToxicFrog: yep, but I've also read that git-show is used for showing information about database, that could also do what I want surely, at least at first glance?	21:06
Ilari	pgregory: 'git log --date-order --pretty=format:%at\ %H <branchname>' outputs output like 1235305715 75c6c402a022937c2148060a14490bea45b96862 for each commit. First is timestamp and second is hash.	21:06
pgregory	and that's just one.	21:06
	← tanek left	21:06
pgregory	Ilari: thanks.	21:06
	→ alvarezp joined	21:06
	← RaySl left	21:07
	→ gerard_ joined	21:07
Adman65	when i check out a branch, the contents of the working directory are automatically updated to reflect that checkout?	21:07
Ilari	Adman65: Yes.	21:08
drizzd	that's the idea	21:08
	← Frando left	21:08
	Jc2k_ → Jc2k	21:08
Adman65	so whatever directory you're in is like the desk, and the papers are kept inside .git, moving papers back and forth in between	21:09
Ilari	pgregory: Actually, add --first-parent there too...	21:09
Adman65	i am trying to familiarize myself with git, never used any sort of version control	21:09
shruggar	Adman65: I think that metaphor broke my brain :)	21:09
dsal	Huh. Isn't there a log option when doing path limits that lets you see all changes in each change outside of the limits?	21:09
	That is, I want to see what came with change I'm finding.	21:10
Adman65	shruggar: not good?	21:10
	← mfwitten left	21:10
jast	pgregory: it's at least partially related to the architecture of git. git is very scriptable so there are heaps of low-level tools. if you use the "git foo" form instead and proper shell completion, some of the shell completion scripts reduce the list of commands shown to you.	21:10
Ilari	pgregory: So 'git log --date-order --first-parent --pretty=format:%at\ %H <branchname>'. That makes difference when there are grafts crosslinking branches.	21:10
jast	Adman65: ah, don't worry, I'm sure we can fix his brain. let's see...	21:10
	→ mjf joined	21:10
shruggar	Adman65, have you used /any/ version control method at all? Such as occasionally making an extra copy of a set of files on a separate backup disk?	21:10
pgregory	Ilari: right, thanks.	21:10
	→ seanius joined	21:10
jast	GIT_DIR="shruggar" git reset --hard	21:10
Adman65	shruggar: ya just that kinda thing. First time using a system	21:11
	shruggar forgot to commit recently, no longer remembers what he was doing :)	21:11
jast	Adman65: your metaphor is not entirely accurate, of course, but we can fix it	21:11
	Adman65: we need to make the metaphor a bit yucky by making VERY literal use of a copying machine	21:12
shruggar	Adman65, okay, so git is like a collection of those backup disks, and "checking out" is like replacing the copy you're working on with one of those backups	21:12
jast	err, liberal use :)	21:12
pgregory	hmmm	21:13
Adman65	i see	21:13
jast	Adman65: so you've got papers on your desk and if you commit them, a copy of all of them gets stored in .git. checkout does another copy of one of the bundles of papers you stored in .git and puts it on your desk. sort of like that. :)	21:13
Adman65	i see	21:13
	well i shall continue reading the user guide	21:13
jast	only checkout is careful about making sure any uncommitted work on your desk doesn't suddenly disappear	21:13
Andphe	dsal, great now I have two directories that I can compare	21:13
pgregory	right, so I get a list of encoded datestamps on one branch, and their sha1's.	21:13
Andphe	thanks a lot	21:14
dsal	Andphe: You shouldn't have two directories -- do it all in one place.	21:14
pgregory	how on earth do I match that to a commit date, but one older, on a different branch?	21:14
dsal	Andphe: git log ..someguy/branchname --- or --- git diff someguy/branchname	21:14
Andphe	is this the best reference for git http://www.kernel.org/pub/software/scm/git/docs/user-manual.html ?	21:14
jast	it's not really a reference	21:15
	→ sanity- joined	21:16
jast	for reference the manpages are rather excellent, especially if you know which command you're looking for :)	21:16
Andphe	today is my first day with git	21:16
	w00t → evilw00t	21:16
jast	so you need an introduction	21:16
	you certainly can't go too wrong with the user's manual	21:17
gerard_	if you know what you are looking for then most of the problem is already solved ;)	21:17
	→ telmich joined	21:17
Andphe	I'm a svn guy	21:17
jast	gerard_: well, for some people it seems to be different ;)	21:17
pgregory	Andphe: me too	21:17
Andphe	so I need something that helps me to start with git	21:17
jast	Andphe: don't worry, it'll pass ;)	21:17
Andphe	:)	21:17
	← sanity- left	21:17
Andphe	well, but that link is a good place ?	21:18
jast	we also have http://git.or.cz/man/gittutorial, which explains how to get started in a few specific kinds of situations	21:18
	"that link" is the official user's manual	21:18
pgregory	jast: :) I'm not sure it will, I've been trying to switch for a week or so, and all I've managed to do so far is waste 3x6 hours cloning the SVN repo.	21:18
jast	so, if nothing else, it will be correct :)	21:18
	pgregory: pure git is a lot more lovely than anything that still involves svn, anyway :)	21:18
	→ sanity- joined	21:18
jast	Andphe: I think your link is a decent start, and if anything seems unclear to you, just ask here and we'll clear it up in a jiffy	21:19
pgregory	jast: sure, starting from scratch it is probably fine, but I have a project with 8 years history, I don't particularly want to use it, nor maintain multiple systems.	21:19
Andphe	thanks guys	21:19
pgregory	s/use/lose/g	21:19
	→ cilly joined	21:19
loic	I've already had a bad time trying to explain svn to my coworkers	21:19
	← cilly left	21:19
loic	no way I'll make them use git	21:20
pgregory	loic: :) odd, I've had the same bad time trying to persuade team members of the benefits of git :(	21:20
gerard_	I think git is actually easier to understand if you never used svn	21:21
pgregory	pehaps	21:21
jast	I think the problem is that if you aren't aware of even the existence of advantages, you're just a lot less willing to be evangelized about it	21:21
Andphe	have a nice day, thanks a lot for your help :)	21:21
	← bwalk_ left	21:22
	← Andphe left	21:22
pgregory	I'm still completely at a loss about how to get my SVN stuff across, if it's even possible.	21:22
	→ qrush joined	21:22
jast	what's the problem?	21:22
Modius	Is there any strategy or practice used by gitters, or in general, to maintain projects where a subset (e.g. a sub-library) may be tracked independently of the full wider project? E.g. a company's app where some sub-library is open-sourced.	21:22
jast	Modius: have a look at git's submodules feature. there's a chapter about it in the user's manual, for example.	21:23
annodomini	Modius: submodules	21:23
Modius	Is there a Git repo-in-repo model that anyone uses?	21:23
pgregory	jast: I have a repo that I've clones using "git svn clone -s --no-metadata https://...."	21:23
jast	yeah, that's basically what submodules are	21:23
pgregory	and now I have a messy git repo.	21:23
jast	what kind of mess?	21:23
pgregory	apparently, I need to 'graft' the merges that have happened over time in SVN within the new git repo to clean it up, but doing so is proving to be....well, for me impossible.	21:24
	→ mfwitten joined	21:24
annodomini	Is there any way to use git to test before doing a "fetch" if the repo you are fetching from contains any revisions in common with your current repo?	21:24
	← mfwitten left	21:24
pgregory	jast: Ilari has been very helpful in trying to get me going, but I'm afraid he's (as are most people here, and the docs for that matter) talking a foreign language to me.	21:25
jast	oh, merge stuff. I see.	21:25
	← kwatz left	21:25
jast	pgregory: well, let's try again, just for the fun of it :)	21:25
pgregory	happy to	21:25
Ilari	pgregory: The git-svn imported the data... What's left is to fix up the errors in history (those errors could blow up real bad on merging)...	21:25
annodomini	I want to be able to avoid doing a fetch from a completely unrelated repo in a script; but I'm not sure how to make sure that doesn't happen.	21:25
segfaux	Anyone working on moving builtin-grep.c to parse-options?	21:25
	pgregory just wishes there was an online tutorial for switching a large SVN repo with lots of history to git.	21:26
jast	annodomini: not really, because fetch is a pretty non-destructive operation anyway. there's git ls-remote to see remote refs, though, if you feel adventurous.	21:26
	→ supercali joined	21:26
jast	pgregory: well, it's a tedious thing to do, particularly if you did many merges, that's the trouble :)	21:26
drizzd	annodomini: git ls-remote I believe	21:26
pgregory	jast: If I ever get this done, I'll write one up, I'm sure it would be useful.	21:27
annodomini	jast, drizzd: Hm. That tells me what the refs on the remote refer to, but not if they share common history with my local repo.	21:27
jast	pgregory: basically, git's merges contain information about which part of the history got merged in. your subversion repository likely doesn't contain that information (unless you did all of the merges with svn 1.5), so you'd have to manually figure it out	21:28
	annodomini: git branch --contains, among others	21:28
pgregory	jast: right, I have merge messages of the form "Merge from trunk @2550:2568"	21:28
	← lucsky left	21:28
jast	oh, that's helpful, I guess	21:28
	but since you used --no-metadata you don't have svn revision number info in git commits, do you?	21:29
pgregory	jast: not as helpful as it might be it seems, as the --no-metadata means I can't do git-svn log.	21:29
	→ idahoev1 joined	21:30
annodomini	jast: Yeah, but if local and remote have diverged, it doesn't tell you. I'm working with some pretty big repos here, so I want to avoid accidentally doing a fetch of the wrong repo	21:30
Ilari	Its possible to search for lines that have less numeric value than specified, probably with awk, but I can't figure it out...	21:30
	← ceej left	21:31
	→ nick108 joined	21:31
jast	pgregory: yeah, so one thing you could do is clone yet again, this time with metadata... you probably won't be all that happy about this approach, though	21:31
	← tal67 left	21:31
	→ ivarv joined	21:32
pgregory	jast: I have a "with metadata" copy on my work machine, from a previous clone.	21:32
	I can get that.	21:32
jast	ah, right	21:32
	well, I think the manual work will be easier if you have the metadata	21:32
	we can always get rid of it after we have made things nice and clean	21:32
pgregory	ok, I'll have to pop away for a mo', as the VPN connection will force me to lose IRC.	21:33
	5mins...	21:33
jast	(an alternative, of course, is to just live with the non-merges and do a bit of trickery the next time you do an actual merge in git)	21:33
	okay, enjoy the ride :)	21:33
	← priidu left	21:34
	→ ceej joined	21:34
	← drewr left	21:34
nick108	I'm trying to follow the guide on http://github.com/guides/fork-a-project-and-submit-your-modifications	21:34
	but when I'm doing a git remote add upstream I get this error: git: 'remote' is not a git-command	21:35
jast	which version of git are you using?	21:35
nick108	1.4.4.4	21:35
	→ schlurchz joined	21:36
nick108	I'm on debian etch and used apt-get to install git	21:36
cehteh	outch	21:36
	get the one from backports	21:36
nick108	ok	21:36
jast	1.4.4.4 is very old	21:36
cehteh	1.4 is prehistoric	21:36
jast	we didn't in fact have "git remote" back then	21:36
nick108	:-)	21:37
cehteh	any 1.5.2 and up should be fine i tihnk	21:37
	← bryanl left	21:38
Ilari	pgregory: The commit IDs aren't interchangeable... The timestamps would...	21:39
schlurchz	Hi, how can I track down an "ambigous ref"? When I do a "git checkout master", it succeeds with "warning: refname 'master' is ambiguous."	21:39
jast	schlurchz: probably in the remotes/ namespaces or something. look at git branch -a.	21:39
	→ Yuuhi joined	21:40
Ilari	schlurchz: 'git for-each-ref \| grep master'?	21:40
schlurchz	Looks all natural. refs/heads/master and refs/remotes/origin/master, nothing else.	21:40
jast	Ilari: teaching plumbing commands considered harmful! ;)	21:40
schlurchz	I did: git branch -a, git ls-remote ., and the proposed git for-each-ref. All looks normal...	21:41
	→ la_mer joined	21:41
	← ivarv left	21:42
	← prlawrence left	21:42
	← la_mer left	21:42
	schlurchz is confused	21:42
schlurchz	Is there no nice -v switch somewhere that I could try to debug this?	21:43
	marcelo__ → FFighter	21:43
	← KIAaze left	21:44
schlurchz	Oh boy, I think I've got it.	21:44
	→ FylGood joined	21:45
	← qrush left	21:46
	← cemerick left	21:46
schlurchz	That comes when one messes with plumbing :-) A spurious ".git/master" ref, leftover from a manual git update-ref :-)	21:46
parasti	jast: the sooner one learns plumbing, the sooner one stops looking at Git as "a bunch of commands that don't mean what they meant in SVN"	21:46
	→ Modius joined	21:46
Ilari	Figured it out: git log --date-order --first-parent --pretty=format:%at\ %H <branch> \| awk -- '$1 < <timestamp>' \| head -1 \| cut -d \ -f 2	21:46
jast	parasti: there is absolutely no use in using a plumbing command for something that can be done with porcelain too	21:46
	→ jwheare joined	21:47
	← jrmuizel left	21:48
	← paltman94 left	21:48
	→ cemerick joined	21:48
	← jceb left	21:48
pgregory	back	21:48
jast	parasti: also, by teaching plumbing, you're creating the perception that git is this insanely complicated beast with weird commands and scripting needed for every simple operation	21:48
parasti	jast: you mean git for-each-ref, right? what's the porcelain for that? git tag \| grep and git branch -a \| grep?	21:48
jast	parasti: in this case git branch -a was sufficient	21:49
	← loincloth left	21:49
jast	oh, yeah, git tag too, fine	21:49
pgregory	jast: I have a repo with metadata now.	21:49
jast	parasti: also, as you see, the problem we were trying to fix was actually caused by misuse of plumbing. :}	21:50
	← vbabiy left	21:50
	→ Korren joined	21:51
jast	pgregory: well, we can do two things now. one is the "proper" way where we fix all merges to become actual merge commits. the other is to fix the minimum amount of merges necessary to make git merge work smartly everywhere.	21:51
Ilari	pgregory: Be aware that the presence of metadata changes the commit IDs...	21:51
	Korren → korren	21:51
pgregory	Ilari: right	21:52
jast	Ilari: I propose we worry about commit IDs much later. :)	21:52
pgregory	jast: well, as I'd like eventually for this repo to form the future for the project, I'd say the best way is the 'proper' way.	21:52
parasti	jast: that's right, _misuse_. that's something you can do with porcelain, too ;)	21:52
	→ lfoppiano joined	21:52
jast	parasti: sure, but plumbing has a LOT more potential for misuse	21:52
Ilari	pgregory: Agreed.	21:52
jast	pgregory: it's your choice, of course. you'll be doing a lot more work, that's all. :)	21:53
	← robomalo left	21:53
pgregory	jast: I don't mind the work, as long as I understand what I'm doing.	21:53
Ilari	pgregory: Its much easier to do all kinds of cleanups to make history reflect what really happened when importing than after-the-fact.	21:53
jast	pgregory: so basically now you pick a merge to fix. I would suggest one of the earliest.	21:53
pgregory	if I can get a handle on doing it once.	21:53
jast	pgregory: by the way, do you want to do any other cleanup work?	21:54
pgregory	jast: what other cleanup work is there to be done?	21:54
jast	well, I don't know your repo :)	21:54
pgregory	I want the best repo for the project.	21:54
	it's quite a complex repo, and apart from sorting out the branches, I want to change the layout.	21:54
jast	if there's anything special about it, now's the time to mention it	21:54
	okay.	21:54
	what kind of change?	21:55
pgregory	currently under 'trunk' we have "aqsis", "win32libs" and "testing"	21:55
	← rlorandi left	21:55
	→ gwoo joined	21:55
pgregory	at least "win32libs" should be moved out ideall, and probably "testing" too, as they are not necessary for normal developers, and quite heavy.	21:55
Ilari	pgregory: Since you got repo with metadata, you want to get rid of it... But it can be done in the same go as filter-branch that is used to "burn in" the grafts.	21:55
pgregory	Ilari: yeah, I remember you mentioning that before.	21:56
jast	and for now we do need the metadata	21:56
	so get rid of a couple of directories. easy.	21:56
Ilari	pgregory: For that sort of moving out, probably one should get topology right first (means fixing merges).	21:56
pgregory	jast: does the fact that the merges are incremental complicate things at all.	21:57
jast	pgregory: not really. if they are properly incremental (i.e. not skipping anything from the source branch) the result will even make complete sense.	21:57
pgregory	i.e. we branch a feature branch, do some work, trunk continue, at some point we merge a set of changes from trunk. Then do some more work, trunk again continues, and we merge another set of changes from trunk.	21:57
jast	otherwise the merges might look a little wonky in retrospect	21:57
	→ hyperair joined	21:58
jast	yeah, that sort of incremental work should be fine	21:58
	there's one important difference between svn and git here, by the way	21:58
	← b_jonas left	21:58
jast	in git, the existence of branches/tags is not versioned	21:58
Ilari	pgregory: No more complicated than merging when source branch continues from merge point (i.e. you can't just use newest commit on the source branch).	21:58
hyperair	it seems git-svn moves the latest tag along with each commit. is there any way to unhook it somehow?	21:59
	→ drewr joined	21:59
	→ JHilgeman joined	21:59
jast	so if you temporarily had a branch that was used for merging, we may have a bit of a harder time figuring out the right thing to do, because that branch name just isn't there anymore.	21:59
pgregory	jast: never done that to my knowledge.	21:59
JHilgeman	Okay, I'm fed up with SVN and someone suggested giving git a try.	22:00
	→ ehamberg joined	22:00
jast	hyperair: no. git-svn follows what the svn repo does. people there shouldn't commit to tags, period. you can manually create arbitrary git tags, though.	22:00
pgregory	all branches are feature or release branches.	22:00
jast	pgregory: yeah, in git we tend to throw away feature branches once they get merged into mainline	22:00
	because the history is all there anyway, just not the name itself	22:00
hyperair	jast: right, so people shouldn't commit to tags, but that's EXACTLY what git-svn did!	22:00
JHilgeman	Can someone point me to some sort of best practices for git?	22:00
pgregory	jast: yeah	22:00
	← krh left	22:01
	→ flazz joined	22:01
JHilgeman	We have a setup where we need a trunk and then a branch for each developer, who will be merging updates to the trunk and pulling updates from the trunk a lot.	22:01
jast	hyperair: well, if someone committed to an svn-side tags/foo dir, git-svn reproduces that on the git side	22:01
JHilgeman	So if anyone can point me in the right direction as to how best to use git...	22:01
pgregory	jast: right, I've picked a feature branch that has quite a few update merges from the trunk, as a starter.	22:01
JHilgeman	I'd appreciate the help.	22:01
hyperair	jast: how else do you create a tag on svn?	22:02
mugwump	JHilgeman: I usually refer people to http://kerneltrap.org/mailarchive/git/2008/6/19/2168434	22:02
	← ehamberg left	22:02
Ilari	JHilgeman: Branch for each developer... Each devel has whole local repository...	22:02
pgregory	jast: I presume I start with the first of those?	22:02
jast	hyperair: well, committing there once is fine, of course. it's if you change the stuff in a tags/ subdir that things get sucky	22:02
hyperair	jast: the stuff wasn't changed in tags.	22:02
pgregory	jast: the branche is soc2007-dsm, and the first merge is from trunk 1082:1103	22:02
hyperair	jast: the previous commit tagged 0.5.0-2	22:02
jast	pgregory: if first=oldest, then yeah, makes sense. it doesn't actually matter, really, but it fits with the actual history so it's kind of nice.	22:03
JHilgeman	Ilari: yes, that is the intention. Each developer should have a complete copy of the trunk at the beginning.	22:03
hyperair	jast: and then i dcommited, and the next thing i know, 0.5.0-2 was updated.	22:03
mugwump	best practices are also not to use the work "trunk" :)	22:03
Ilari	JHilgeman: Clone grabs complete copy of repository.	22:03
	← RodP left	22:04
JHilgeman	mugwump: that seems to be about maintaining the git project	22:04
jast	hyperair: that's, uh, very unusual, I think. perhaps you could ask on the mailing list, and cc the git-svn maintainer (Eric Wong, [email@hidden.address]	22:04
Ilari	pgregory: Ok, you need has of commit coresponding to trunk at r1103...	22:04
	*hash	22:04
hyperair	jast: alright, i'll give it a try	22:04
mugwump	JHilgeman: right, but the development practices for git.git are considered best practice	22:04
JHilgeman	ah	22:05
	got it	22:05
	thx	22:05
mugwump	in particular the section on "master" vs "maint" vs "pu", etc	22:05
	→ whalesalad joined	22:05
Ilari	hyperair: If you created local branch from that tag, just that sort of thing will happen on dcommit...	22:06
jast	pgregory: so now we do black magic to make the commit containing the merge refer to 1103. to do that, you need to know the git commit id of r1103, e.g. using git svn find-rev.	22:06
	→ brokenladder joined	22:06
hyperair	Ilari: i didn't create a local branch from that tag.	22:06
pgregory	hmm, if I do "git svn find-rev r1103 trunk" I get nothing :(	22:06
	pgregory notes 'trunk' is a remote if that means anything.	22:06
jast	pgregory: try without the "trunk"	22:07
hyperair	Ilari: i ran git-svn clone -T trunk -t tags -b branches svn://blahblah/ blah	22:07
whalesalad	Does anyone know of a way to get a list of all the authors on an svn project, to create a .svn_authors file to use with git-svn?	22:07
pgregory	nope, nothing.	22:07
hyperair	Ilari: it just happens that the previous commit was a tag.	22:07
jast	hurr.	22:07
Ilari	whalesalad: svn log?	22:08
jast	hyperair: dcommit figures out where to commit to by looking at the history of the current branch and finding the newest git-svn-ified commit. it then commits to the svn branch specified in there.	22:08
whalesalad	Ilari: I'm not the greatest grep programmer, any way to parse that file for just the authors? A place to get me started?	22:08
hyperair	jast: but it commited to a svn tag, not a branch.	22:08
Ilari	pgregory: 'git log --grep="@1103"'?	22:08
pgregory	I've looked at all the logs, and there's a lot missing.	22:09
jast	hyperair: git-svn doesn't really care. if the git-svn-id line refers to a .../tags/foo URL, it'll take that, too.	22:09
	pgregory: failing that, git log --all --grep="@1103"	22:09
pgregory	theres a 1104, and a 1095, but nothing in between.	22:10
	→ davidad joined	22:10
Ilari	pgregory: They belong to some other branch?	22:10
jast	and 1103 does positively exist on the svn side, right? and can be read from the repository, too?	22:10
	(i.e. no data corruption)	22:10
	← beket left	22:11
hyperair	jast: and how does one change what git-svn-id points to?	22:11
Ilari	pgregory: And that was with --all? Because commits to another branches show as gaps when looking at just a single branch...	22:11
pgregory	heh, no, it doesn't, the commit message must be wrong	22:11
hyperair	jast: previous git-svn-id was a /tags/foo URL, and all the commits after it seem to get tags/foo as well	22:12
pgregory	I have 1103 in a different branch.	22:12
	could be the message was wrong, and referred instead to 1104	22:12
jast	hyperair: you don't. if you want to commit to elsewhere, you base your local branch on something other than the tags/foo thing.	22:12
pgregory	in fact, that sort of holds up.	22:12
hyperair	jast: how?	22:12
	→ Beket joined	22:12
Ilari	pgregory: I think the message is right, referring to trunk at 1103, which is same state as trunk at 1095...	22:12
hyperair	jast: how do you see where your locak branch is based anyway	22:12
pgregory	Ilari: yeah	22:13
	← cemerick left	22:13
jast	hyperair: for example, git checkout -b newbranch trunk or git checkout -b release-1.0 or whatever you've got	22:13
	hyperair: you can see it in the git log messages and I think there's a git-svn command for it too	22:13
Ilari	pgregory: Would be real fun if one tried to automate...	22:13
pgregory	ok, so we go with 1095	22:13
hyperair	jast: okay i'll try that. but shouldn't git master be based on trunk directly after a clone?	22:14
pgregory	I now have the sha1 for 1095	22:14
	→ diegoe joined	22:14
jast	hyperair: sadly not. master is always based on the last thing that got committed to at the time you cloned.	22:14
diegoe	hey :-)	22:14
	← FreakGuard left	22:14
	← kumbayo_ left	22:14
hyperair	jast: blargh. no wonder.	22:14
jast	pgregory: and you also need the sha1 of the merge.	22:14
	now we get to the black magic.	22:15
pgregory	and the two from "git log --grep=Merge --parents soc2007-dsm"	22:15
Ilari	pgregory: Then stuck those three SHA-1 values into grafts and do reload in 'gitk --all'. Now the merge commit should show up differently.	22:15
	← jwheare left	22:15
jast	pgregory: edit the file .git/info/grafts (or create it) and insert the following line: <merge sha1 id> <current parent of merge sha1 id> <sha1 id of 1095>	22:16
	← seanius left	22:16
jast	this will tell git to pretend that the merge commit has now two parents: the original one and the one you merged in	22:16
	the order is relevant, by the way :)	22:16
	→ FreakGuard joined	22:17
jast	not technically, but you'll want to keep the original parent as the first parent to make the history look sane.	22:17
pgregory	jast: ok, what is the order, I have two sha1's from the log command	22:17
	and one from the find-rev	22:18
jast	pgregory: the one I indicated a couple of lines above :)	22:18
	← idahoev1 left	22:18
jast	after that, all you need to do is use the same kind of process many more times until .git/info/grafts is larger than the known universe	22:18
pgregory	right, I think I get that.	22:18
Ilari	pgregory: The two from --parents are <merge sha1 id> <current parent of merge sha1 id>, in that order...	22:18
pgregory	yeah, that's what I figured.	22:18
	← JHilgeman left	22:19
jast	so you just append the ID of the merged commit to that and presto	22:19
	→ jerbear joined	22:19
	← Beket left	22:19
pgregory	right, now if I go into gitk and search for the sha1 of the commit id	22:20
	I get a commit with an arrow coming off and down to the right, pointing into space.	22:20
jast	into space?	22:21
	follow it to see where it ends! :)	22:21
	← drewr left	22:21
	← broken_ladder left	22:21
pgregory	it goes to another commit that has trunk@82 in it.	22:22
Ilari	pgregory: Err... Does its ID match the ID from find-rev?	22:23
	→ mithraic joined	22:23
diegoe	hey guys, i was reading git's soc ideas page and I'm interested in the line-level browser idea, would you please point me to more info about it?	22:23
pgregory	doh, yep, my fault.	22:24
	→ beket joined	22:24
pgregory	now it makes more sense (when I put the right sha1 in)	22:26
	← perplexes left	22:26
pgregory	now it points back to the few commits on the branch prior to the trunk merge.	22:26
	← mfilipe left	22:27
pgregory	and up, it points to a few commits until it hits the next merge point, at which point it jumps over to the left.	22:27
Ilari	pgregory: Next merge should go in similar manner but probably goes faster... Good thing that first one also provoded example about handling gaps...	22:27
	→ perplexes joined	22:27
	← beket left	22:27
	pgregory goes to try...	22:27
	→ AAA_awright joined	22:27
	← FylGood left	22:30
pgregory	now, I have...	22:30
	← ruphy left	22:30
	← davidad left	22:30
	← d0k left	22:31
	→ up_the_irons joined	22:31
	← up_the_irons left	22:31
nick108	cehteh: got git version 1.5.6.5 from etch-backports, now it works fine :-)	22:31
	thx for the pointer!	22:32
	→ up_the_irons joined	22:32
	→ cemerick joined	22:32
pgregory	right, I'll continue with this branch and see where it leads me, hopefully I'll have a eureka moment while doing it.	22:32
	← up_the_irons left	22:33
pgregory	thanks jast and Ilari for all your help, I hope it's fruitful I'd like to switch permanently.	22:33
jast	pgregory: good luck with the cleanup. if you need anything else, I'm just a highlight away for the next two hours or so.	22:33
pgregory	I'll have to stop now, early start in the morning, but I think I have the basic idea now.	22:34
	I'll almost certainly have to pop back here later, but I'll try and see if I can do as much of this part of the process myself now.	22:34
jast	yeah, sounds like a plan :)	22:34
pgregory	I'm beginning to get an feel for the display in gitk too, so that should help.	22:34
	cheers, and goodnight.	22:34
	→ JohnnyL joined	22:35
	→ agenteo joined	22:35
	← JohnnyL left	22:35
	← cemerick left	22:35
	→ jwheare joined	22:35
	← AAA_awright left	22:36
jwheare	what's the best way to create a new project in github that's a subset of another project while retaining history?	22:36
	→ AAA_awright joined	22:36
	← wmoxam left	22:37
	→ up_the_irons joined	22:38
jast	jwheare: if by subset you mean something that doesn't include all of the files, you're basically breaking history in that merges back and forth won't really work anymore, but basically you can clone that original repository locally, do a creative git filter-branch on it and get the result up on github.	22:39
	→ nrook joined	22:39
	→ Beket joined	22:40
	→ Kudos joined	22:40
jwheare	i was afraid it would involve filter-branch	22:40
mugwump	come on, it's like on switch now :)	22:42
	one switch	22:42
	--subdir-filter or something like htat	22:42
jwheare	ah yeah, there's an example in the man	22:43
	i'm still pretty confused by git to be honest	22:43
	have been repeatedly findig myself in wtf just happened situations	22:44
	→ Gurpartap joined	22:44
	← aziz left	22:44
	→ aziz joined	22:44
Gurpartap	I'm cloning a github repo. I think it's downloading _all_ the history code as well. How do only fetch the latest code only?	22:44
	Eventually, I want to diff against it.	22:45
up_the_irons	Gurpartap: git clone --depth <N>, will only get you the last <N> items	22:46
Gurpartap	depth = 1 or 0 in my case?	22:46
mugwump	go for 10 or 20	22:46
up_the_irons	Gurpartap: but the repo will be limited in use since it doesn't have the full history	22:46
Ilari	jwheare: eagain.net/articles/git-for-computer-scientists/ and www.newartisans.com/blog_assets/git.from.bottom.up.pdf Understand what those are about and most operations make much more sense...	22:46
up_the_irons	Gurpartap: depth = 1, you get 1 commit	22:46
	← FreakGuard left	22:46
mugwump	jwheare: also read the user manual sections on design ... it helps	22:46
Gurpartap	up_the_irons: hm	22:47
jwheare	thanks!	22:47
Gurpartap	up_the_irons: does that also mean only 1 last committed file will be fetched?	22:47
	(i have cvs bg)	22:48
	← EmilMedve left	22:48
up_the_irons	Gurpartap: no, you'll get all the files	22:48
	Gurpartap tries	22:48
up_the_irons	Gurpartap: just only 1 piece (the latest) of the history	22:48
Gurpartap	Okay	22:48
	That's what I need. :)	22:49
	← rue left	22:49
Gurpartap	The latest snapshot.	22:49
	→ sakyamuni joined	22:50
	← lightcap left	22:50
	← brett_h left	22:50
	→ davidw joined	22:51
	← annodomini left	22:51
	← Modius left	22:51
davidw	ify	22:51
	oops... hey, I did an git svn rebase	22:51
Ilari	...	22:51
davidw	and it seems to have disappeared some of my git changes	22:51
	where did they go/	22:51
	→ FreakGuard joined	22:51
Gurpartap	Thank you up_the_irons, mugwump :)	22:51
	← AAA_awright left	22:52
jast	davidw: you probably haven't finished the rebase yet...	22:52
	→ Modius joined	22:52
davidw	jast, is there a way to just roll it back and start over? I haven't finished it yet but I want to try again, I guess	22:52
	err, I mean I did finish it	22:53
jast	davidw: git rebase --abort	22:53
davidw	no rebase in progress	22:53
jast	okay. did you do anything else after you finished the rebase?	22:53
	→ AAA_awright joined	22:53
davidw	nothing worth worrying about	22:53
jast	okay. note this will trash uncommitted changes: git reset --hard HEAD@{1}	22:54
	(in some shells you may need to escape the {})	22:54
davidw	what's the @{1} ?	22:54
jast	"state of HEAD before the previous operation"	22:54
davidw	ah	22:54
jast	hmm	22:54
davidw	so there's also a 2, 3, etc...counting backwards?	22:54
jast	just to make sure, have a look at git log -g	22:54
	this shows you the "history" of operations	22:54
	← FreakGuard left	22:55
jast	you can pick any of these	22:55
up_the_irons	Gurpartap: np :)	22:55
Gurpartap	Now I'm wondering how I should do the diff. :)	22:55
jast	what diff?	22:55
Gurpartap	I have a GIT repo copy, then a modified copy	22:56
	Need to prepare a patch	22:56
	"patch"	22:56
	:)	22:56
gerard_	can I push a remote branch to another repository? or do I have to make a local branch for it first?	22:56
	← Ryback_ left	22:56
	→ bryanl joined	22:56
	← brokenladder left	22:56
jast	gerard_: you can, you just need to be specific about it. for example: git push someremote refs/remotes/foo/bar:targetname	22:57
Ilari	gerard_: 'git push <repo-nick-or-URL> foo/bar:refs/heads/bar'.	22:57
jast	Gurpartap: so, the standard diff util, perhaps? alternatively copy the modified files over into the repo you cloned and run git diff.	22:58
	← LiamH left	22:59
	→ FreakGuard joined	23:00
gerard_	ah thanks guys, it was pushing the individual commits but not the branch itself	23:01
	that was really weird	23:01
jast	push shouldn't actually be able to push commits without refs	23:01
	← hyperair left	23:01
	→ ayaka_t joined	23:01
	← whalesalad left	23:03
	→ hyperair joined	23:05
	← bantic left	23:06
davidw	ok	23:06
	it says I have conflicts... is git capable of doing something like svn does where it just asks what I want to do - in this case, simply use 'theirs' ?	23:07
	← netoman left	23:09
	← nick108 left	23:09
	← boto_ left	23:10
Ilari	davidw: It leaves conflict markers... Also, to use their version 'git checkout --theirs -- file'.	23:12
Gurpartap	Cloned a git repo, and each of the file there had date in the $Id$ tag as 2008-03-09 instead of 2008/03/09 (viz cvs's format). Now, how can I avoid this change in git diff ?	23:13
davidw	Ilari, --theirs is an unknown option	23:13
	→ brendon_work joined	23:13
	→ kansan joined	23:14
	← rudi_s left	23:14
Ilari	davidw: What git version?	23:14
davidw	1.5.6.3	23:15
Gurpartap	nvm, I'll try not to change those heads..	23:16
	jayallen-afk → jayallen	23:16
	→ gitte joined	23:19
Ilari	davidw: Also, it leaves conflict markers to files...	23:19
davidw	I know... but I don't want them, they're generated files. I just want to say 'whatever, use whatever version, but get rid of the damn conflict'	23:20
	svn makes this easy, I can't even tell if git makes it possible without upgrading to some future version	23:20
	← abbe left	23:21
	→ kate21de joined	23:22
jast	davidw: git checkout MERGE_HEAD -- filename	23:22
Ilari	jast: Err... its git svn rebase...	23:22
jast	oh.	23:22
	← lfoppiano left	23:23
	→ ChrisAlbright joined	23:23
jast	I can't think of anything quite as nice then, off the top of my head	23:24
	→ ideogon joined	23:24
Ilari	'git show :3:path/to/file'?	23:25
jast	yeah, but then you need to save it as a file and stage it, so that's two operations	23:25
wagle	with git diff, how do you send the two files to your own diff command that provides the output?	23:26
jast	wagle: look at git diff --ext-diff; it's a bit of a hassle to set up though	23:27
	wagle finally finds the --ext-diff option	23:27
	← cmarcelo left	23:27
	← ciskje left	23:28
	← ekidd left	23:28
wagle	jast, thanks	23:29
	← bobesponja left	23:30
	→ bobesponja joined	23:30
	← FreakGuard left	23:31
	→ stouset joined	23:31
	← stouset left	23:31
wagle	wow.. i already wrote the interface a year ago! wotta guy I am!	23:31
	← jwheare left	23:32
	← ph^ left	23:33
	← mjf left	23:34
	→ FreakGuard joined	23:36
	→ Dashkal joined	23:37
	→ felipec joined	23:37
	→ nixblicker_ joined	23:37
Dashkal	Is there a way to do a fast-forward merge without a working tree? I'd like to make master point to FETCH_HEAD in a bare repos. Currently just deleting master and remaking it.	23:38
	← davidw left	23:40
Dashkal	I'd rather fast forward master to FETCH_HEAD and have git complain if master is not reachable from FETCH_HEAD. My current method leaves me a little vulnerable to mistakes.	23:40
	← hyperair left	23:40
	→ jwheare joined	23:42
	← kate21de left	23:44
	← marvil07 left	23:45
	← SRabbelier left	23:45
	← diegoe left	23:45
	← ideogon left	23:46
	→ k0001 joined	23:47
	→ galderz joined	23:49
	← nrook left	23:49
	← galderz left	23:49
	→ davidad joined	23:49
	← k0001 left	23:50
	← dsal left	23:52
	→ k0001 joined	23:52
	→ lresende joined	23:52
	← lresende left	23:53
	→ lresende joined	23:53

Logs Search ←Prev date Next date→

Channels Documentation