IRCloggy #git 2008-10-03

Provider of IRC logs since 2005.

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

	2008-10-03
	← pdarnows left	00:02
	← BabelO left	00:03
	→ ashleyw joined	00:03
	← drewr left	00:05
	← kanru left	00:05
	→ andersbr joined	00:07
	← withanx left	00:10
	← patmaddox left	00:10
	← ashleyw left	00:12
	← ben_h left	00:13
	→ ashleyw_ joined	00:13
	ashleyw_ → ashleyw	00:13
	← elmex left	00:13
	→ ben_h joined	00:13
	← ashleyw left	00:13
	← tvw left	00:14
	← tcurdt left	00:16
	→ ashleyw_ joined	00:16
	← ashleyw_ left	00:16
	→ ashleyw joined	00:17
	← ashleyw left	00:18
	→ ashleyw joined	00:19
	← gilimanjaro left	00:20
	← baggend left	00:22
	← ashleyw left	00:23
	→ ashleyw joined	00:23
	not-xjjk → xjjk	00:25
	← ashleyw left	00:27
	→ ashleyw joined	00:28
	→ foca joined	00:29
	→ rmh3093 joined	00:29
	→ flaguy joined	00:31
	mw → mw\|out	00:31
	← christophsturm left	00:32
	← softdrink left	00:32
	→ eternaleye joined	00:35
	→ softdrink joined	00:37
	← sergio left	00:38
	← Fullmoon left	00:38
	→ warreng joined	00:39
	← eternaleye_ left	00:40
	← ben_h left	00:42
	→ foca_ joined	00:43
	← foca left	00:44
	→ gretch_ joined	00:47
	→ Fullmoon joined	00:49
	← sgrimm left	00:51
	→ foca joined	00:54
	→ qrush joined	00:56
	← foca_ left	00:56
	← kef left	01:00
	← ciaran left	01:03
	→ foca_ joined	01:06
	← foca left	01:06
	← thechuck left	01:06
	→ ben_h joined	01:09
	→ foca joined	01:11
	← chase_ left	01:14
	← ereslibre left	01:15
	← bts- left	01:16
	→ khc joined	01:18
khc	is there a way to git clone to a remote machine, if the remote machine doesn't have git?	01:19
	← MethoS- left	01:19
Shadou	rsync	01:20
khc	well rsync doesn't work for me because I wanted to do a clone with a small depth	01:20
Shadou	sorry then I don't know.	01:21
	→ doener_ joined	01:21
Shadou	Install git into your ~/ on the remote machine?	01:21
	← WALoeIII left	01:23
uau	khc: what answer did you expect? did you really think you could create a shallow clone of a git repository without any use of git? that sounds silly	01:23
khc	Well I can't connect from that remote machine to the local machine	01:23
	→ foca__ joined	01:23
khc	uau: I have git locally	01:23
	I guess I can clone locally, and the scp over	01:24
	I just wonder if there's a one step thing I can do	01:24
	→ Flathead joined	01:24
	← foca left	01:25
	← foca_ left	01:28
	→ thechuck joined	01:31
	← doener left	01:33
	← Tv left	01:34
	← kanru3 left	01:34
	→ foca joined	01:37
	← foca__ left	01:37
	← thechuck left	01:38
	← lsk left	01:38
	→ offby1` joined	01:39
	← offby1 left	01:39
	← Fullmoon left	01:43
	→ EmilMedve joined	01:50
	→ stigrett joined	01:52
	→ JesseL627 joined	01:55
	← JesseL627 left	01:58
	→ foca_ joined	01:59
	→ kef joined	02:01
	→ KevBurnsJr joined	02:01
KevBurnsJr	Question : trying to git pull, and getting an error message "error : Entry 'filename' not uptodate. Cannot merge.	02:02
	tried git fetch -f, still getting the same error.	02:03
	What else can I do to force a merge?	02:04
	→ kyrus1 joined	02:05
khc	I am not sure if you can, usually I git stash, pull, then apply the stash	02:05
vmiklos	commit or revert your local changes.	02:05
	stash, commit or reset --hard	02:05
	offby1` → offby1	02:05
	→ sergiodj joined	02:06
	← foca left	02:06
	← kef left	02:07
	→ kanru joined	02:08
KevBurnsJr	thanks	02:10
	we just accidentally git rm -fr .	02:11
	← eternaleye left	02:11
	→ kef joined	02:11
KevBurnsJr	Any way to recover? :/	02:11
offby1	git reset	02:11
	any uncommitted changes are probably gone	02:12
	← mvanholstyn_ left	02:13
	→ eternaleye joined	02:13
	→ qhox joined	02:15
vmiklos	s/probably/certenly/	02:17
khc	s/certenly/certainly/ :-)	02:17
vmiklos	that's it, thx:)	02:18
	→ willb joined	02:24
	← foca_ left	02:24
	← kef left	02:26
	← eddyp left	02:26
	→ bct joined	02:27
bct	is there some way to record a change but never push it?	02:28
	i've got local database settings that i don't want in the main repo, but i don't want them to appear in git diff either	02:29
	→ foca joined	02:29
	→ Ryan52 joined	02:29
	→ quoin joined	02:30
offby1	maybe commit them to some branch that you don't plan on pushing	02:31
	or perhaps don't have those files in the repository at all	02:31
	and then add their names to .gitignore	02:32
bct	then i'd have to switch to that branch every time i wanted to use those files, right?	02:34
	i guess gitignore is the obvious answer	02:35
cbreak	gitignore doesn't work on files that are tracked	02:36
offby1	so un-track 'em then	02:37
	→ thechuck joined	02:37
	→ krawek joined	02:37
	→ mmiika joined	02:37
mmiika	in windows bash, how do I exit after "git log"	02:38
khc	q?	02:38
	← spearce left	02:39
	← thechuck left	02:39
mmiika	khc, ah, cheers	02:40
	← joydivider left	02:41
	← Ryan52 left	02:43
	← hmetz left	02:47
	→ thechuck joined	02:50
	→ quoin- joined	03:00
	→ Tv joined	03:01
	← mw\|out left	03:01
	← tongueroo left	03:02
johnw	if I'm using git+ssh with a host that has git installed on a non-standard path, which file do I need to add a PATH to for it to find git-upload-pack?	03:04
mutex	yeah	03:05
johnw	ah, the -u flag	03:05
mutex	I think you can even put it in the 'remote' config section	03:06
	I have receivepack = sudo -u <user> git-receive-pack	03:06
	in my config file	03:06
johnw	hmm... -u didn't do it, now it says git-pack-objects dies, which of course it would if git-upload-pack can't find it...	03:06
	which config file?	03:06
	on which server?	03:07
	i only need to clone/fetch btw, not push	03:07
	ls	03:07
krawek	johnw: I think you can use ~/.ssh/environment	03:07
johnw	cool, thanks	03:07
	→ spearce` joined	03:07
	← foca left	03:09
	← eternaleye left	03:12
	← spearce` left	03:16
	← quoin left	03:16
	← thechuck left	03:16
	→ eternaleye joined	03:21
	→ sgrimm joined	03:21
	← markkalderon left	03:21
	← mmiika left	03:26
	← jerem_y left	03:30
	← kyrus1 left	03:31
	→ thechuck joined	03:36
	← thiago left	03:39
	← Gof left	03:39
	← bobesponja left	03:39
	← fenglich_ left	03:40
	← mstormo left	03:40
	→ dmwit joined	03:40
	→ bobesponja joined	03:41
dmwit	I'm reading through Git From the Bottom Up.	03:41
	I think I grok what blobs, commits, and tags represent; but what is a tree (conceptually)?	03:41
offby1	oh that's easy	03:42
	it's basically like a Unix directory: a table of rows	03:42
	each row is a reference to another object (a tree or a blob), plus its name in the filesystem, plus maybe some permission bits.	03:42
	that's all	03:42
	→ hidek joined	03:43
dmwit	Okay. Suppose I have a tree t that references another tree t'. What does that mean?	03:43
	(i.e. why are we keeping these references?)	03:43
	← bct left	03:44
	→ stouset joined	03:46
offby1	it means that when you check out the commit that points to that tree, you get one directory that contains another directory.	03:47
	we keep the references so that we can create the right directories and files on your disk when you check the thing out.	03:47
	for some value of "we"	03:47
	:)	03:47
dmwit	ah, ah	03:48
	So the important bit in the tree is the filename, not the references.	03:48
offby1	_both_ are crucial	03:48
	else they wouldn't be in there	03:48
dmwit	(Though the references are, I guess, similar to the directory hierarchy.)	03:48
	errr... where "important" means "the bit I was overlooking" =P	03:48
offby1	without the references, git would have no idea what _content_ to put in a file, nor what _entries_ to put in a directory	03:49
	→ ajonat joined	03:49
dmwit	Thanks!	03:50
	→ Gof joined	03:50
	→ fenglich_ joined	03:50
	→ thiago joined	03:51
	→ FunkeeMonk joined	03:51
	← stouset left	03:55
	→ fenglich joined	04:00
	← CodeOfficer left	04:01
	→ spearce joined	04:03
	→ thiago_ joined	04:03
	→ spearce` joined	04:04
	→ fenglich__ joined	04:05
	← stigrett left	04:08
	← thechuck left	04:08
	onitunes → onitunes\|away	04:08
	→ tjafk2 joined	04:10
	→ foca joined	04:11
	← thiago_ left	04:11
	← tjafk1 left	04:12
	→ thechuck joined	04:13
	← fenglich_ left	04:13
	← thiago left	04:15
	← k776 left	04:20
	← fenglich__ left	04:20
	← spearce left	04:20
	→ fenglich__ joined	04:21
dmwit	This tutorial mentions git add --patch, but that doesn't seem to exist here.	04:22
	Did it get replaced?	04:22
	← fenglich left	04:22
dmwit	(It sounds like just the thing I want, coming from darcs.)	04:22
mutex	what do you mean by 'exist here' ?	04:22
	gretch_ → gretchen	04:22
mutex	what version of git are you using	04:23
dmwit	1.5.2.5	04:23
	→ thiago_ joined	04:23
machrider	my git (1.5.4.3) has it	04:23
dmwit	Okay, I'll look at upgrading then.	04:23
mutex	is it just not in the man page ?	04:23
	or when you run the command nothing happens	04:23
	← spearce` left	04:24
dmwit	When I run the command, I get a usage message.	04:24
machrider	i want to create a new git command... is there a document i can read to get started?	04:25
mutex	well it was introduced in 2007	04:25
	you are kind of using a fairly old version of git	04:25
	→ danten joined	04:26
	← danten left	04:26
	← Bass10 left	04:26
	← Gof left	04:28
machrider	wow, git's code is like 99% undocumented	04:29
mutex	its self documenting ;-)	04:30
	← qhox left	04:30
machrider	ahahahah	04:30
	commit.h: unsigned long date;	04:30
	doesn't specify units... so should i assume seconds since 1970?	04:30
mutex	actually I'm working on a patch right now that changes that value to time_t	04:31
	and time_t is seconds since epoch	04:31
	It isn't documented with docbook	04:32
	but there is a lot in /Documentation	04:32
	→ fenglich joined	04:33
	→ mariusSO joined	04:35
	mariusSO → mstormo	04:35
	← thiago_ left	04:35
	→ thiago__ joined	04:35
machrider	it's surprising, but at the same time, i'm used to it.	04:43
	← thiago__ left	04:43
	→ thiago joined	04:44
mutex	machrider: if one is a good coder the less a need for documentation	04:46
	← fenglich__ left	04:47
	← thiago left	04:51
	→ fenglich_ joined	04:52
	→ thiago joined	04:52
mutex	how do I get git to split my patch into segments rather than one large email ?	04:54
	git-format-patch doesn't seem to want to split the patches	04:54
	← thechuck left	04:56
	← warreng left	04:57
mutex	hmm maybe git does it automagicly	04:59
	when a threshold of changes occurs	04:59
	← quoin- left	04:59
	← niki left	05:01
	→ dec_ joined	05:02
	← fenglich left	05:03
khc	git-format-patch makes one file per commit	05:05
mutex	ah, ok	05:06
	→ girishr joined	05:07
	→ lacostej joined	05:08
	← dec left	05:09
	← ashleyw left	05:10
	→ nickh joined	05:15
	← qrush left	05:22
	→ neurodam1ge joined	05:22
	→ qhox joined	05:24
	→ Dreamer3 joined	05:26
Dreamer3	so i have a ton of origin/blah branches listed that are no longer really on origin... is there an easy way to make them go away?	05:26
	→ trochala joined	05:28
DrNick	Dreamer3: git remote prune origin	05:28
Dreamer3	that looks right	05:29
	sweet :-)	05:29
	DrNick: thanks :) perfect	05:29
	→ tongueroo joined	05:30
	← trochala left	05:31
	← lacostej left	05:31
	← neurodamage left	05:34
	→ iulian joined	05:35
	→ johan-s joined	05:35
	← softdrink left	05:36
	← dmwit left	05:38
	→ dmwit joined	05:38
	← tongueroo left	05:40
	→ tongueroo joined	05:40
	→ mtodd_ joined	05:41
	← mikearr left	05:44
	→ mikearr_ joined	05:44
	→ _slacker_ joined	05:46
	← mtodd left	05:47
	← ricky left	05:48
	→ ricky joined	05:49
	← ricky left	05:50
Flathead	I have a backup of my git repo that I want to restore, but git-clone doesn't work. I've tried git-clone --mirror, but I cant checkout on the resulting repo. What can I do to get a proper copy?	05:50
khc	Hmm I've never used --mirror, but --bare works for me	05:52
Flathead	khc, ok I'll give --bare a shot	05:53
	→ ricky joined	05:55
Flathead	khc, doesn't work either. When I try to git-checkout, I get this error: ``fatal: This operation must be run in a work tree''	05:56
khc	oh, you don't git checkout directly from a repo	05:57
	you clone it	05:57
Flathead	khc, first I did a git-clone --bare into .git, then tried to git-checkout	05:57
	← bobesponja left	05:57
khc	what are you trying to do exactly?	05:58
	← tongueroo left	05:58
khc	oh I see, you want to restore from a backup? Don't you just directly clone the backup, without any options?	05:59
Flathead	khc, if I do that, it doesn't preserve branches (they become remotes), origin, etc. I want an exact copy	05:59
khc	oh, just fix up your .git/config manually ;-)	06:01
Flathead	khc, that's what I've been doing, but I was hoping there was a simpler way	06:04
khc	try doing a git init before git checkout	06:05
	→ namenlos joined	06:07
	→ parasti joined	06:08
Flathead	khc, that did the trick! Thanks :)	06:08
	→ robinr joined	06:08
	← FunkeeMonk left	06:09
	← resmo left	06:14
	→ SRabbelier joined	06:14
	→ resmo joined	06:15
	→ priidu joined	06:16
	← mstormo left	06:16
	→ tongueroo joined	06:18
	← KevBurnsJr left	06:18
	← vuf left	06:18
	→ mstormo joined	06:20
	← nickh left	06:22
	→ Tricon joined	06:23
	← Tricon left	06:23
	→ thoroth joined	06:23
thoroth	if i want to convert a svn repo to a git repo and i have external libraries via svn externals - what is the recomended way to tie in those libraries?	06:25
	→ anttih joined	06:25
thiago_home	are they switching to git too?	06:26
up_the_irons	thoroth: i use symlinks, personally	06:26
thiago_home	if so, you can use git submodule	06:26
thoroth	where are the symlinked files located? within the git repo? are they still under subversion?	06:27
	→ Ademan joined	06:27
	→ mtodd joined	06:33
thoroth	up_the_irons: Do it like so? http://blog.alieniloquent.com/2008/03/08/git-svn-with-svnexternals/	06:33
up_the_irons	thoroth: you have your svn externals outside of the git tree	06:34
	thoroth: then just symlink to it	06:34
	← qhox left	06:35
up_the_irons	thoroth: i haven't used submodules at all, and it's been a while since i've used git-svn, so defer to that article, i guess :)	06:35
	← SRabbelier left	06:36
	→ SRabbelier joined	06:36
thoroth	I use git submoduled on current projects, But here I am trying to revive an older project and I am totally out of it it. I tried http://effectif.com/2008/4/24/easy-git-svn-for-rails but this threw some errors. So I guess I will have to do it step by step ... But I want to find a way that does not get me into dependency hell at a later stage... :-)	06:38
up_the_irons	yeah, just use the simplest solution possible	06:39
	← mtodd_ left	06:40
	← dmwit left	06:45
	→ dmwit joined	06:46
	dwmw2_gone → dwmw2	06:50
	← bronson left	06:51
	← statim left	06:54
	→ mtodd_ joined	06:54
	→ statim joined	06:54
	→ ph^ joined	06:55
	← johan-s left	06:55
	← _slacker_ left	06:55
	← krawek left	07:02
	→ bentob0x joined	07:04
	← robinr left	07:04
	→ BabelO joined	07:07
	→ Gof joined	07:09
	→ Weaselweb joined	07:10
	← marcoe left	07:10
	← mtodd left	07:10
	→ drizzd joined	07:11
	→ bts- joined	07:12
	← kanru left	07:13
	← thoroth left	07:15
	← resolve left	07:15
	→ Aikawa joined	07:18
	← tongueroo left	07:23
	→ johan-s joined	07:27
	← bartocc left	07:29
	→ nud joined	07:29
	← Flathead left	07:29
	→ cilly joined	07:30
pasky	fwoosh fwoosh	07:31
	← mstormo left	07:39
	→ ph^_ joined	07:44
	← sergiodj left	07:45
	← dmwit left	07:45
	← bts- left	07:47
	← werdnativ left	07:48
	← cilly left	07:50
	→ robinr joined	07:53
	← temoto left	07:55
uggabooh	before I try installing git to a rhel 3 server, is there some advice against using git pull for deployment?	07:58
	→ kanru joined	07:58
paakku	How do I check which pack contains a given object?	08:02
	in particular, I'm consindering adding a branch in ELinks, but perhaps I shouldn't start the branch from a commit in elinks-history, which is only grafted in.	08:03
	as then pushing that branch to the main repository would require also adding most of the history there.	08:04
uggabooh	paakku: git verify-pack -v will print the content of packs	08:05
paakku	uggabooh: yes but it's kind of slow	08:06
	← iulian left	08:06
	→ ciaran joined	08:06
paakku	uggabooh: the index already exists so can I use that somehow?	08:06
	→ ciskje joined	08:07
paakku	uggabooh: e.g. when I do git show some_commit_id, it chooses the right pack based on the index, I think, and does it fast	08:07
	→ d0k joined	08:07
uggabooh	paakku: I don't know of any way to get that information out of git	08:07
	← jchris left	08:09
	→ peritus__ joined	08:09
	← lordpil left	08:09
	→ lordpil joined	08:09
	← Eridius left	08:11
	← floppyears left	08:11
	→ floppyears joined	08:13
	← tmz left	08:14
	→ tmz joined	08:14
	→ Eridius joined	08:14
	→ charon joined	08:17
	→ cilly joined	08:20
	→ mstormo joined	08:20
	→ ebel joined	08:22
	→ Flathead joined	08:22
	← khc left	08:25
	→ offby1` joined	08:27
	→ _dave_h_d joined	08:28
uggabooh	why is it that git rebase prints out whitespace warnings by default?	08:28
_dave_h_d	hi when i do a git pull it tells me that needs update for all the files?	08:28
	→ marcoecc joined	08:30
	← mstormo left	08:30
thiago	I'm having problems with git diff and $GIT_WORK_TREE	08:32
	← offby1 left	08:33
uggabooh	_dave_h_d: "git status" should tell you that you have local changes. They must be committed before you can merge.	08:33
	→ _doomster_ joined	08:35
	→ shodges joined	08:35
	→ mstormo joined	08:35
	← univate left	08:37
paakku	thiago: last time I attempted that, it plainly did not work, so I ended up doing cd and setting GIT_DIR instead.	08:37
thiago	yeah, that's what I did now	08:37
paakku	thiago: I don't remember which version of git that was.	08:37
thiago	of course, cd with paths with spaces won't work...	08:38
	→ univate joined	08:38
_doomster_	how do I revert changes I made locally? I couldn't find an option for that...	08:38
	← Jacolyte left	08:38
parasti	_doomster_: git checkout	08:39
	_doomster_: or more accurately, man git-checkout :P	08:39
Gitbot	_doomster_: the git-checkout manpage can be found at http://git.or.cz/man/git-checkout	08:39
_doomster_	ah, -f it is....	08:40
	thanks!	08:41
	← paakku left	08:41
	→ GarethAdams joined	08:43
parasti	_doomster_: actually I was thinking "git checkout -- paths", but I guess -f works if you want to throw away _all_ changes	08:45
_doomster_	okay ;)	08:46
	→ tcurdt joined	08:50
tcurdt	hey guys ... I got a rebase question: is it possible to inline a branch like here http://imagebin.org/27904 ?	08:52
	(the blue dot is a tag)	08:52
	→ bcarlyon\|laptop joined	08:53
	→ gilimanjaro joined	08:54
	→ christophsturm joined	08:57
	← mstormo left	08:57
tokkee	tcurdt: git rebase branchA master # this will change master	08:58
	tcurdt: git rebase --onto branchA branchA master # this will change branchA	08:59
	← namenlos left	08:59
	→ tvw joined	08:59
tcurdt	tokkee: I thought doing git rebase branchA master I will end up with 1>2>3>7>8>9>4>5>6 ...no?	09:00
	← BarryCarlyon left	09:02
	bcarlyon\|laptop → BarryCarlyon	09:02
uggabooh	ah, so rebase is using apply, which is printing out whitespace warnings ... (?)	09:03
tokkee	tcurdt: No, that's what you'd get with "git rebase master branchA" ;-)	09:03
	← cilly left	09:04
tokkee	uggabooh: Yes.	09:04
uggabooh	that seems a bit wrong, for rebase to print warnings that you did not see during commit	09:05
tcurdt	tokkee: aaah! :) cool!	09:05
	← schlort left	09:05
tcurdt	one more question ... is there a way of getting a diff of all changes in a branch? (I know I could do a diff between 3 and 6 ...but)	09:05
tokkee	but?	09:07
parasti	compare the branch to your upstream (the branch you branched off)	09:07
	→ markkalderon joined	09:07
tcurdt	well ... I would need to figure out the sha where the branch has been branches off and then use the sha	09:07
parasti	why? did you delete it?	09:08
tcurdt	I mean something like ... "give me the diff from the root"	09:08
parasti	when you rebase, all your changes are applied on top of the "upstream" branch, so a diff between upstream and yourbranch will get you all the changes in the branch	09:09
tcurdt	parasti: no no ... forget the rebase for a sec :)	09:09
	(sorry)	09:09
parasti	hmm :)	09:09
	→ _slacker_ joined	09:09
tcurdt	http://imagebin.org/27904 first graph	09:09
	I want to get a diff of the changes on branch A	09:10
	preferably without looking up the sha for 3	09:10
	→ lordpil- joined	09:10
parasti	okay, I haven't actually used this, but the "git diff master...branchA" syntax is probably what you're after	09:11
	also see git merge-base	09:11
	→ DrFrasierCrane joined	09:12
	← peritus__ left	09:12
tcurdt	parasti: ah ... cool ... will give that a try!	09:13
	as always - thanks a lot guys! :)	09:13
	later	09:13
	← ajonat left	09:14
	← ph^ left	09:18
	→ schlort joined	09:19
	→ peritus_ joined	09:22
	← schlort left	09:23
	→ mstormo joined	09:23
	← d0k left	09:24
	→ Sho_ joined	09:25
	→ cilly joined	09:27
	← lordpil left	09:27
	→ ph^ joined	09:28
	→ chris2 joined	09:31
	← mstormo left	09:33
	→ mstormo joined	09:34
	→ flukebox joined	09:36
	→ FunkeeMonk joined	09:38
	→ Yuuhi joined	09:39
	← CelticSoul left	09:42
	← cilly left	09:46
	→ me joined	09:48
me	hi, newbie question, I went back in time with checkout HASH and changed a few files, I now want to go back to the last commit with checkout master, but it won't let me because I changed some files in my working directory. How do I tell git to ignore the changes I made and bring me back to the master branch?	09:49
	I don;t really want to create a new branch by commiting the changed files, I just want to lose them and get back to my old state	09:50
	→ drizzd_ joined	09:51
rvsjoen	checkout -f ?	09:51
	← ag left	09:51
	→ Beket joined	09:52
me	rvsjoen: what does the f flag do?	09:53
_doomster_	discard all changes	09:53
me	ok thanks guys	09:54
	← peritus_ left	09:54
me	i got error:unknown switch 'f'	09:55
	← markkalderon left	09:56
me	sorry, I made a typo, worked thanks	09:56
rvsjoen	-f "Proceed even if the index or the working tree differs from HEAD. This is used to throw away local changes." <- its in the man page	09:57
	→ bryanwb joined	09:58
	→ ag joined	09:59
bryanwb	git seems to change permissions on some of my files when I fetch, then checkout a version. The root user and group own the file on the machine. Any way around this?	09:59
me	yeah thanks, I notice this throws away all local changes. Is there a way to checkout master just throwing away the changes I made since I checked out an older commit? For example, if master is the HEAD branch and I've changed some files in the working directory, I checkout an earlier commit and change a file there. Now it won't let me checkout master because of that changed file, I want to...	10:00
	...get back to master just losing changes to that single file and not the uncommited changes I made whilst on master.	10:00
	is that possible?	10:00
Arrowmaster	i would have thought the uncommited changes would have been lost when you checked out an earlyer commit	10:02
bryanwb	me: to save uncommitted changes	10:02
	me: use git stash before the checkout	10:02
	me: and "git stash apply" to reapply your uncommitted changes	10:02
	→ elmex joined	10:03
me	braynwb: thanks	10:03
	← Gof left	10:04
bryanwb	bryanwb: if anyone could advise me on how to retain my file permissions across checkouts, I would really, really appreciate it	10:05
parasti	bryanwb: plain git does not try hard to preserve permissions and ownership -- it's not an archiver, it's a vcs	10:05
bryanwb	parasti: damn, any way to retain permissions besides manually resetting them across after checkouts?	10:06
parasti	bryanwb: somebody might have written something to achieve that -- don't remember any names though	10:07
	← drizzd left	10:07
Arrowmaster	etckeeper	10:07
	and theres another one i cant remember	10:07
bryanwb	Arrowmaster: thanks! will check that out	10:07
parasti	Arrowmaster: yeah, that's what I was thinking too	10:07
	→ schlort joined	10:08
Arrowmaster	the other one i cant remember claimed that etckeeper tracked permissions on untracked files and claimed to be better but like i said i cant remember it	10:08
	→ not-xjjk joined	10:08
bryanwb	Arrowmaster: parasti: tks guys, the OLPC project in Nepal tks u. Over and out http://blog.olenepal.org	10:09
	← bryanwb left	10:09
	→ Gof joined	10:09
	→ mariusSO joined	10:10
	→ fenglich joined	10:10
	← fenglich_ left	10:10
	← thiago left	10:11
	← mstormo left	10:11
	mariusSO → mstormo	10:11
	→ thiago joined	10:13
	← BarryCarlyon left	10:15
	← Beket left	10:16
	→ Beket joined	10:17
	→ kbingham joined	10:19
	← xjjk left	10:20
	← fenglich left	10:25
	→ mariusSO joined	10:26
	→ fenglich joined	10:26
	← thiago left	10:26
	→ thiago joined	10:26
	← mstormo left	10:26
	mariusSO → mstormo	10:26
	← Gof left	10:26
	→ Gof joined	10:27
	← me left	10:32
	← chris2 left	10:35
	← mstormo left	10:36
	← fenglich left	10:36
	→ fenglich joined	10:36
	→ mariusSO joined	10:37
	mariusSO → mstormo	10:37
	→ Vortex35 joined	10:39
	→ boto joined	10:40
	← ag left	10:45
	← FunkeeMonk left	10:45
	→ RaceCondition joined	10:45
	→ ag joined	10:47
	← floppyears left	10:47
	→ floppyears joined	10:47
	→ cilly joined	10:48
	← Gof left	10:49
flukebox	Any simple way to fix the confilcts while pulling ???	10:49
	→ Gof joined	10:49
	→ boto_ joined	10:50
	← boto_ left	10:51
	← Beket left	10:52
bremner	flukebox: use git mergetool ?	10:52
Vortex35	In git there isn't implicit branching by cloning like in mercurial. Is it?	10:53
flukebox	any one more thing .. when i push my updates to repo .. and changes are visible to another clone but not to the repo itself	10:54
bremner	fluxbox: faq non-bare	10:55
Gitbot	fluxbox: You shouldn't push to a non-bare repository. See http://git.or.cz/gitwiki/GitFaq#non-bare	10:55
flukebox	Gitbot: yeah i am suing non-bare .. files which were already there are visible.. but changes are not	10:55
	*using	10:56
bremner	flukebox: read the article	10:56
flukebox	okkk	10:56
	→ BarryCarlyon joined	11:00
	← DrFrasierCrane left	11:03
	← boto left	11:04
	← bremner left	11:06
	→ shriram joined	11:06
pasky	anyone got tips on student git hackers, except miklos, lea and sverre?	11:06
	hmm, isn't jakub actually a student?	11:07
shriram	how do i print history of changes of one particular file?	11:07
Mikachu	shriram: git log -- thefile	11:07
shriram	thanks	11:08
	← gretchen left	11:11
	← ebel left	11:11
	← shriram left	11:17
	← cilly left	11:17
	→ koke joined	11:18
	→ hein2 joined	11:18
	← Sho_ left	11:18
	hein2 → Sho_	11:18
tcurdt	I am confused. I am trying to migrated from git to svn but when I use git-svn the branches do not show up as would think. git branch -a shows them but when I look at gitk they do not seem to branch off anywhere. On the git-svn I do get "Successfully followed parent" though.	11:18
	any hints?	11:19
	→ Absolute0 joined	11:20
	← ciaran left	11:20
Absolute0	fatal: http://github.com/ramin32/tandemgraph/info/refs not found: did you run git update-server-info on the server?	11:20
	→ ciaran joined	11:20
Absolute0	Why am I getting this?	11:21
	→ dancor joined	11:21
	← clairvy left	11:21
dancor	is it possible to nuke history on a branch only	11:22
anttih	Absolute0: there is no such file	11:22
	← uggabooh left	11:22
Absolute0	anttih: that means that the repo wasn't properly initialized?	11:22
anttih	Absolute0: no, not neccessarily, what are you trying to do?	11:23
	why are you trying to access that file?	11:23
Absolute0	anttih: I am not	11:23
	I am doing a clone of the repo	11:24
	→ clairvy joined	11:24
Absolute0	git clone http://github.com/ramin32/tandemgraph	11:24
Mikachu	Absolute0: why do you think that's the correct url to clone from?	11:24
	← clairvy left	11:24
anttih	Absolute0: hmm, then I don't know. ask the github guys	11:24
Absolute0	what would be the correct url?	11:24
	→ clairvy joined	11:24
parasti	Absolute0: if you visit that URL, you'll find out	11:25
	← clairvy left	11:25
Absolute0	ah :)	11:26
	thanks	11:26
	← Absolute0 left	11:26
	← Sho_ left	11:27
	→ Sho_ joined	11:28
	→ clairvy_ joined	11:28
	← yofel left	11:28
	→ yofel joined	11:28
tcurdt	doh ... ok .... gitk --all shows them ... but still not all parents are branching off the correct sha	11:28
dancor	how do i modify the commit message if it was only the first commit in my repo	11:30
	← clairvy_ left	11:30
dancor	i normally git-reset --soft HEAD^ && git-commit -c ORIG_HEAD -a	11:30
	→ bremner joined	11:31
	→ clairvy_ joined	11:33
	← SRabbelier left	11:35
	→ Beket joined	11:35
	→ alexross joined	11:41
	← johan-s left	11:43
	← tcurdt left	11:44
	→ johan-s joined	11:44
	→ bobmcw joined	11:49
dancor	is it possible to have different branches not share the same first commit	11:50
vmiklos	sure.	11:51
	just initialize a separate repo, then you can push the other branch to the same bare repo on the server.	11:52
	like git push origin master:myotherbranch	11:52
gebi	you can also use git symbolic-ref HEAD refs/heads/<new parrentless branch>	11:53
	→ chris2 joined	11:54
	→ v_silent joined	11:57
vmiklos	yes, but in practice in most cases you want to have both branches checked out so you'll need two repos ;)	12:00
	← emiel_ left	12:00
dancor	what do people normally do if they want to publicize a project repo but still have like a private dir for private configuration files	12:01
	→ j_engelh joined	12:01
dancor	i could have the private repo include the public one in a subdir?	12:01
gebi	dancor: make the programm behave like all other programs (reading config from /etc/foo/ then ~/.foo.cfg or ~/config/foo)	12:02
	→ peritus_ joined	12:02
dancor	gebi: i need some files for people using it here at my company	12:04
	gebi: i have the README.mycompany, a company-specific installer wrapper, and our default config file	12:05
gebi	oh ok, maybe give submodules a trie	12:07
	try	12:07
	→ cypher23 joined	12:07
v_silent	Hi guys, Im new in git, and here is my question i created a new repository on local home directory and cloned it by git clone to another temporary directory after i made changes in temporary repository files and commited them but how can i get changes in original repository	12:07
bremner	gebi: you could also maintain your private stuff as a patch using topgit or just as a branch that you rebase onto the public branch	12:08
gebi	if it's really private stuff i'd _NEVER_ put it in the same repo	12:09
	v_silent: mercurial user :)?	12:10
	→ boto joined	12:10
v_silent	gebi: no im using CVS but i want to migrate to git	12:10
	→ cilly joined	12:11
	← reaVer left	12:11
gebi	v_silent: normaly you would just create a branch in the repo you want to work on	12:11
	you can push your changes back	12:11
	(but it's not recommend to push into a non-bare repo in git)	12:11
	→ reaVer joined	12:12
	→ mw\|out joined	12:12
v_silent	first of all i created a main repository and imported files in it	12:12
	→ bcarlyon\|laptop joined	12:13
v_silent	then i did git-clone is that like cvs checkout ?	12:13
gebi	no	12:13
v_silent	oh	12:13
gebi	it's like rsync the whole cvs repo from the server	12:13
	← ia left	12:13
	← cilly left	12:13
v_silent	so is there something like cvs checkout ?	12:14
	→ ia joined	12:14
	← nud left	12:14
gebi	v_silent: http://git.or.cz/course/svn.html	12:14
	→ emiel_ joined	12:15
	→ deskin joined	12:15
	← Beket left	12:17
	→ Beket joined	12:17
	→ cizra joined	12:20
cizra	Hi	12:20
	→ cannonball joined	12:21
	→ cypher256 joined	12:21
cizra	About git-svn: I have a SVN repo that has the # sign in pathnames. git-svn chokes on that. I don't know perl, so I can't really modify it either. Has this deficiency maybe been fixed in git-svn? My version is 1.5.6.5-1~bpo40+1 (Debian package. git-svn --version balks)	12:22
	The error message is	12:23
	Filesystem has no item: 'http://myserver.com/myproject/tags/Release_#1' was not found in the repository at revision 225 at /usr/bin/git-svn line 3812	12:23
Mikachu	cizra: try git-svn from <git.git next:git-svn.perl>	12:25
	did i get the name wrong maybe	12:25
cizra	The proper fix would be to replace # with %23 in URLs.	12:25
	um. Does that refer to the Git source repository of Git?	12:25
Mikachu	maybe just <next:git-svn.perl>	12:25
Gitbot	[git git-svn.perl]: http://tinyurl.com/4rrlwa [blob]	12:25
Mikachu	there we go	12:25
cizra	OK, thanks	12:26
Mikachu	the "raw" link at the top	12:26
cizra	OK, thanks	12:26
deskin	cizra: there is a patch, 'git-svn: fix handling of even funkier branch names' since your version, so it might be worth trying next as Mikachu suggests	12:26
Beket	What is faster? Clone a git repository or import the same repository from cvs ?	12:26
Mikachu	Beket: uh, the first	12:27
Beket	Mikachu, thanks	12:27
	bcarlyon\|laptop → BarryCarlyon	12:29
cizra	I think I lack some Perl Git bindings or something...	12:29
Mikachu	i'm not sure if you can actually run the source script directly, or if the makefile changes some variables in it?	12:30
cizra	Well, seems like I can't.	12:30
	270 my $fh = Git::temp_acquire('svn_delta');	12:30
	Git:: hints like there's a Git Perl module or something.	12:30
	→ temoto joined	12:30
deskin	Mikachu: I've been able to run the source script directly, the makefile just changes a version number from +GIT_VERSION+	12:31
Mikachu	okay	12:31
	i dunno about the Git:: thing	12:31
pasky	you should get tht all by make install	12:32
cizra	OK	12:32
deskin	there's a Git.pm which is installed; it's in git.git/perl/	12:34
	<master:perl/Git.pm>	12:34
Gitbot	[git Git.pm]: http://tinyurl.com/3om35w [blob]	12:34
	→ Bass10 joined	12:35
	← cypher23 left	12:35
	cypher256 → cypher23	12:35
cizra	deskin: Can I just install the CPAN module Git::Wrapper?	12:36
	→ lorandi joined	12:36
deskin	idk; I'd build and install git from source using make, as that's how almost everyone does it	12:36
	and Git::Wrapper certainly isn't needed to do that	12:36
cizra	OK	12:37
	← peritus_ left	12:37
Mikachu	deskin: look at #17 http://www.survs.com/WO/WebObjects/Survs.woa/wa/shareResults?survey=M3PIVU72&rndm=OKJQ45LAG8	12:38
deskin	what specifically?	12:38
	← BarryCarlyon left	12:38
deskin	oh; I meant to say, almost nobody installs Git by going and grabbing Git::Wrapper from CPAN	12:39
jogla	I think i found something that git is not able to do: working in two different branches at the some time ;)	12:41
	s/some/same/	12:41
	→ stouset joined	12:41
	← ph^ left	12:41
cizra	jogla: Ever seen "git stash"/	12:41
	?	12:41
	ph^_ → ph^	12:41
	→ deavid joined	12:42
cizra	Hm, are git repository formats stabilized? That is, are different versions interoperable?	12:42
tpope	there's only one repository version	12:43
cizra	OK	12:43
pasky	jogla: git new-workdir?	12:43
	tpope: that is somewhat simplifying	12:43
	but true for practical matters, I guess	12:43
	cizra: you might get in trouble if you use git-1.4.4.4 or older	12:44
jogla	cizra: Have fun, i want you to use that when i really have to switch between them often	12:44
	how often in one minute do you want to use git stash?	12:44
pasky	jogla: what are you really trying to do?	12:45
cizra	pasky: OK, thanks	12:45
	→ Ryback_ joined	12:45
tpope	I was oversimplifying from repositoryformatversion = 0 :)	12:46
	what changed after 1.4.4.4?	12:46
Mikachu	jogla: you know about git clone?	12:46
jogla	git clone could work	12:47
pasky	tpope: see 1.6.0 release notes	12:48
jogla	im working in branch 'foo'	12:48
pasky	technically, you might get into problems even with some 1.5 versions if you are using newer features like submodules	12:48
jogla	and now someone is reporting a bug, over Jabber	12:48
	and part by part i get all the inofrmation to fix the bug	12:49
	→ drewr joined	12:49
jogla	but i have enough time in between to continue working in branch 'foo'	12:49
cizra	OK, about git-svn and weird filenames: I fetched and compiled the latest git and it still fails.	12:50
tpope	ah	12:50
pasky	jogla: so why don't you create another clone, or another workdir?	12:50
deskin	cizra: cool, thanks for pursuing this; I'd suggest sending a quick bugreport to [email@hidden.address]	12:51
	← EmilMedve left	12:51
flukebox	why git give when changes are made in same file ??	12:51
	and any way to resolve that ?	12:52
	→ RaceKondition joined	12:52
	← temoto left	12:52
	← RaceKondition left	12:52
cizra	deskin: Weird. Reading the patch it seems that # is escaped...	12:53
	+scary_ref='Abo-Uebernahme%20(Bug%20#994)'	12:53
	(it's a test case)	12:53
	I can't read Perl, does this line escape #s? I suppose not.	12:54
	if ($url =~ m#^([^:]+)://([^/])(.)$#) {	12:54
	→ tomtt joined	12:54
deskin	no, just testing if it matches	12:54
	→ temoto joined	12:54
cizra	Well, and does it?	12:54
deskin	depends what $url is	12:55
cizra	Let's say, something that contains the # character	12:55
tomtt	I know i can commit all deletions of files using git-commit -a, but how can I only commit the deletions in a specific subdir?	12:55
cizra	Hm, wait, no	12:55
	s/([^\w.%+-]\|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;	12:56
	This seems like the escaper.	12:56
deskin	cizra: if you still have the git source around, you can cd t ; ./t9118<tab> and see if the testcase passes	12:56
cizra	OK	12:57
	← meyering left	12:57
deskin	FWIW it looks like the URI sent to SVN has # expanded as %23 already	12:57
cizra	deskin: passes	12:57
	→ meyering joined	12:57
	← robinr left	12:58
	← temoto left	12:58
tomtt	or alternatively, how can i git-add all changes in a specific directory, including any deletions?	12:58
	→ aroben joined	12:58
	→ robinr joined	12:58
	← RaceCondition left	12:58
cizra	grumble	13:03
pasky	tomtt: git add -u ?	13:04
	or something like that	13:04
deskin	cizra: it looks like the escaping should work, so I'd think the problem is that the URI isn't actually being escaped. Are you getting the error when trying git svn fetch?	13:04
cizra	Yes, creating the initial checkout or whatever it's called.	13:05
	No, clone	13:05
deskin	ok; I see, Release_#1 is a tag in your URL	13:06
cizra	yep	13:06
	Found possible branch point: <stuff>	13:06
	Found branch parent: (tags/Release_#1) 7a399366c95e6a139ec35a704e5176330642fbe2	13:06
	← koke left	13:09
	→ peritus_ joined	13:09
cizra	And actually I believe the case should be covered by the test, but...	13:10
	← foca left	13:12
	→ foca joined	13:13
	← charon left	13:16
	→ tongueroo joined	13:17
gilimanjaro	any egit users here that know why my commit dialog shows changed files that i haven't touched, and which git-status on the command line also doesn't show?	13:20
	← tongueroo left	13:23
	→ juanjoc joined	13:26
	→ gzy joined	13:26
flukebox	anyhelp for concurrently made changes ?	13:28
	git always giving me confilts	13:28
spb	hrm?	13:28
flukebox	*confilcts	13:28
	hrm ???	13:28
spb	what exactly are you trying to do	13:29
gzy	hi, would you have some advice on why git-svn clone segfaults when trying a googlecode project (any rep i tried)?	13:29
flukebox	i have two clones for same repo .. and i am making changes in both clones in same file and then commint from one clone and then trying the same from second clone	13:29
	but that gives me confilcts	13:30
	and if and first pull from the main repo then my changes in 2nd are lost	13:30
	*if I first pull	13:30
	→ koke joined	13:31
flukebox	any solun ? or am i missing something ? .. what i want is .. to set up a repo and then 2-3 person would be contributing for the repo	13:31
cizra	gzy: How did you make a Perl script segfault?	13:32
	flukebox: You should merge them	13:32
	→ jerbear joined	13:32
flukebox	cizra: how ?	13:32
cizra	flukebox: with "git merge", I suppose	13:32
gzy	cizra, i have no idea :) i'm on a school box now.. i'll try it again from home but i wanted to make sure i'm doing everything right before mailing admins	13:33
	(actually git-svn segfaults on me without any parameters...)	13:33
	Arjen enjoys some Abba	13:34
Arjen	Whoops, sorry :-)	13:34
	→ LiamH joined	13:35
cizra	gzy: Man, git-svn is a freakin' Perl script. Those can't segfault, unless your Perl is broken.	13:35
	→ CodeOfficer joined	13:36
PerlJam	cizra: not true. There are ways to make perl segfault (but usually they involve hairy regex and/or deeply magical features)	13:37
Arjen	Maybe the Subversion bindings are broken	13:37
flukebox	cizra:http://rafb.net/p/CPMDus82.html	13:37
	→ rdavila joined	13:37
	→ withanx joined	13:37
cizra	flukebox: I'm a Git newbie. Note that your changes have to be at least in different parts of the file.	13:38
flukebox	cizra: okk ...	13:39
cizra	flukebox: (otherwise git can't know which one to use or how to merge then)	13:39
	them*	13:39
gzy	well... since it crashes without any arguments, i wrote the admins... i'll stick around for a few minutes and if the're fast enough i'll let you all know what it was...	13:40
flukebox	cizra: any good tutorial ? i had wasted my last 10+ hours for a solution	13:40
cizra	flukebox: Lemme test it first	13:40
flukebox	cizra:okk	13:41
cizra	flukebox: Otherwise, http://justfuckinggoogleit.com or ask others, not me	13:41
flukebox	okk	13:41
	cizra: no probs	13:41
	→ RaceCondition joined	13:42
	→ jnareb joined	13:43
cizra	flukebox: Works for me. What I did was:	13:43
	flukebox: create master and one file with 3 lines	13:43
	flukebox: create 2 branches, a and b	13:44
	flukebox: in a, modify the first line. In b modify the third line	13:44
	flukebox: checkout master and "git merge a" then "git merge b"	13:44
	flukebox: no errors, the merge imported both modifications.	13:44
flukebox	cizra: thanks ;-) let me try it out	13:44
	BTW... if i am working in same branch ... so i guess that may be the prob .. and it doesn't matter where in the file i made the changes	13:45
jnareb	pasky: did you send your gitweb patches ("hex, not octal" and "ids for summary table") also to git mailing list, or only directly to maintainer? I haven't seen them via GMane NNTP interface...	13:47
cizra	flukebox: Well, you shouldn't, I guess...	13:47
	flukebox: dunno, I don't know much about Git	13:48
flukebox	okies	13:48
	;-)	13:48
	← stouset left	13:49
	cizra is struggling to publish a Git repo with either lighttpd or git-daemon and fails utterly	13:49
cizra	Help?	13:49
	[11755] '/circmon-trunk.git': unable to chdir or not a git archive	13:49
	→ thechuck joined	13:49
cizra	It freakin' IS a git archive!	13:50
	→ SEJeff joined	13:50
	← clairvy_ left	13:50
SEJeff	How can I checkout a remote branch and track it locally?	13:50
lack	cizra: But can the git-daemon user or lighttpd user actually CD into that directory? What are the permissions?	13:51
	← gzy left	13:51
cizra	lack: Permissions aren't a problem	13:51
	→ Phillemann joined	13:52
jnareb	SEJeff: "git checkout <remote-branch>" would checkout <remote-branch> _state_ and detach HEAD	13:53
lack	SEJeff: 'git checkout -b mybranch origin/remotebranch' ? You may need to specify '--track' depending on your config options.	13:54
	→ spearce joined	13:54
	→ softdrink joined	13:54
jnareb	SEJeff: "git checkout --track <remote-branch>" (or in older git "git checkout -b <localname> <remote-branch>" would set up local branch which you can then modify based on and to follow <remote-branch>	13:54
SEJeff	ok, how does this config look: http://git.pastebin.com/m2ac11910	13:55
	For tracking the next branch on origin to a local branch named next	13:55
	jnareb, lack Looking at the gitweb that seems to have done it. Thanks!	13:56
jnareb	SEJeff: good. that setup would make "git pull" on branch 'next' fetch from origin and merge 'origin/next' branch	13:57
SEJeff	perfect, that is exactly what I wanted	13:57
lack	cizra: Then is the path right? is it really at /circmon-trunk.git	13:58
flukebox	http://rafb.net/p/3uSCE850.html	13:58
cizra	lack: no, I hoped it'll take the relative path	13:58
	→ SRabbelier joined	13:59
	← willb left	13:59
	→ Freundlich joined	13:59
	→ lionr joined	14:00
Freundlich	Hi. I created a new repository with gitosis using this guide: http://scie.nti.st/2007/11/14/hosting-git-repositories-the-easy-and-secure-way. The new repository can be cloned via ssh but it doesn't have any branches (not even master). I pushed to the repository with: git push origin master:refs/heads/master like the tutorial says.	14:01
lionr	hi - i am trying to git push some changes out to a repository and i am getting the following error: http://rafb.net/p/YjLwVP73.html -- any ideas? i am "already up to date" according to "git pull".	14:01
	← Sho_ left	14:01
	← deavid left	14:02
	← Magnus` left	14:02
pasky	jnareb: all is sent to [email@hidden.address] and I'm getting it back	14:02
offby1`	lionr: look for a file or directory named .git/objects/95 on both sides; if it exists (and is empty!!) delete it	14:03
lack	flukebox: Edit prac/off.c manually to resolve the conflict. Then 'git commit -a' (probably -a, check 'git status' first) just like it says at the bottom of 'man git-merge'	14:03
Gitbot	flukebox: the git-merge manpage can be found at http://git.or.cz/man/git-merge	14:03
pasky	jnareb: I have sent 7 gitweb patches in the last two days	14:04
	offby1` → offby1	14:04
pasky	jnareb: has any made it to gmane? :)	14:04
lack	Freundlich: If you go onto the server and look in refs/heads/ in that repo, do you see master?	14:05
lionr	offby1`: On the shared side (i.e,, to where I am pushing) it has a file in the ./objects/95/ directory.	14:05
offby1	grr	14:05
	← marcoecc left	14:05
offby1	so much for the easy fix	14:05
	I really don't know what that error is. I'd have expected git to not care if that directory already existed.	14:05
jnareb	pasky: doesn't look like it	14:06
	pasky: I wonder if it is GMane anti-SPAM filter, or vger anti-SPAM filter that caused trouble...	14:06
	→ nille joined	14:07
pasky	I can't quite imagine how could the vger anti-SPAM filter play any role in that	14:08
	→ powerfox joined	14:08
pasky	and if GMane is filtering what it gets from vger, that's silly :)	14:08
	← foca left	14:08
Freundlich	lack: Ow. There is refs/heads/refs/heads/master. What did I do wrong?	14:08
	emmanuel_away → emmanuel	14:08
powerfox	Hi! If I apply a patch, should I specify author's email in --author? Or just a name?	14:08
gilimanjaro	just name	14:10
	email in --email :)	14:10
lack	Freundlich: Probably you should have just pushed 'master:master'	14:10
	→ bcarlyon\|laptop joined	14:11
powerfox	gilimanjaro: oh, didn't see it :) Thanks	14:11
cizra	What's the difference between git pull and git fetch?	14:11
lack	Freundlich: Take a look at your config file, look at the refs section of your origin remote.	14:11
	bcarlyon\|laptop → BarryCarlyonb	14:11
	BarryCarlyonb → BarryCarlyon	14:11
Freundlich	lack: Sorry, I don't understand why the tutorial wants me to push master:refs/heads/master but pushing master:master does the trick.	14:11
	Thanks! :)	14:11
	This worked in the past... but that was with 1.5.something.	14:12
offby1	Freundlich: all's well that ends well	14:12
	wow, that guy sure is friendly	14:12
lack	cizra: pull does 'fetch and merge' Check out 'man git-pull' for details.	14:12
Gitbot	cizra: the git-pull manpage can be found at http://git.or.cz/man/git-pull	14:12
cizra	Thanks.	14:12
powerfox	gilimanjaro: are you sure there is --email in git-commit?	14:13
gilimanjaro	ehr	14:13
	possibly not	14:13
	i was thinking of git-config mostly	14:13
	which doesn't have --email as such, but i use it to set it	14:13
	← tango_ left	14:13
	→ tango_ joined	14:14
powerfox	gilimanjaro: yeh, there is in config. Will use email in author as suggested in man then	14:14
gilimanjaro	in fact, git-commit sais to use --author="John Doe [email@hidden.address] syntax	14:14
	→ kumbayo joined	14:14
	→ stouset joined	14:15
powerfox	yeh :)	14:16
	gilimanjaro: thanks for help	14:16
gilimanjaro	anyone here have suggestion on how to handle symlinks in a git repo when using jgit/egit?	14:17
	gilimanjaro hears crickets	14:18
	← thechuck left	14:18
	→ thewaltz joined	14:18
	offby1 watches a tumbleweed	14:19
	→ warreng joined	14:19
	→ willb joined	14:19
pasky	hmm	14:20
	seems like some of my other mails did not make it to gmane too	14:20
	I wonder why :(	14:20
	gmane does not like me	14:20
spearce	gilimanjaro: handle symlinks like track symlinks in the repository?	14:20
	pasky: your rename patch fails the test suite. :-\	14:20
	and your gitweb tag patches yesterday (only 2?) won't apply to gitweb.	14:21
gilimanjaro	spearce: not quite; we have symlinks in our repo (which is don't work on and don't care about), but egit keeps seeing them as Addex, index diff files	14:21
	spearce: and wants to commit them each time i commit something else	14:21
spearce	gilimanjaro: symlinks are a problem in java, the jre doesn't know about them so it does stat() and not lstat(). so egit is seeing the link target, not the link. :-\|	14:22
	there's probably some bugs in egit around that too. jgit tries to handle the symlink as much as it can as a symlink in git data structures, but when its looking at the filesystem it can't tell a symlink apart from its target.	14:22
	← SEJeff left	14:23
gilimanjaro	spearce: ok; and there is no way to just tell egit to consistently ignore that file?	14:23
	← shodges left	14:23
spearce	not currently. patches are welcome.	14:23
gilimanjaro	spearce: i wouldn't mind clicking them once each and thereby not having to deselect the files for each commit	14:23
spearce	i avoid projects with symlinks in egit.	14:23
gilimanjaro	ok	14:23
	← Phillemann left	14:24
gilimanjaro	spearce: i'd like to, but our dev-team all is fond of them	14:24
	spearce: and me and my minion are the only ones trying to use eclipse	14:24
	→ camwest joined	14:24
gilimanjaro	spearce: i'll try to build something for it though, and submit a patch. any place or method you'd recommend to fix this?	14:24
pasky	spearce: hmm, I will check the testsuite... I thought I ran it through	14:24
	spearce: are you applying the tag patches on top of the CSS patch?	14:25
	→ kevwil joined	14:25
pasky	oh my, could the duplicate To be a spam criterium?	14:26
spearce	gilimanjaro: sorry, nothing's coming to mind right now. there are more than a few code paths that may need to understand the "avoid list" or whatever it is.	14:26
pasky	isn't that strange	14:26
spearce	pasky: possibly, don't duplicate to.	14:26
pasky	i don't, git-send-email does! ;)	14:26
	sigh	14:26
camwest	hi there, i'm having some trouble, when i go git commit -v it loads the commit message into "vi" but i'd rather it load textmate, when i go: echo $GIT_EDITOR it shows: "mate -w" but it doesn't load in textmate, any ideas?	14:26
pasky	i'll look into it	14:26
jnareb	pasky: don't you use TopGit?	14:26
spearce	i tried putting the gitweb stuff onto your pb/gitweb branch, which has "add support for extending the action bar with custom links" as its tip. oh, and its missing the css id changes. doh.	14:27
	← thewaltz left	14:27
	→ thewaltz joined	14:27
gilimanjaro	spearce: ok; i'll read up on the git-index tree and stuff. maybe it should ignore diff's if it sees the file that's different is listed as a symlink in the git repo? (or index; i'm not familiar with the terminology yet)	14:27
jnareb	pasky: By the way, which branch I should look for your patches at git://repo.or.cz/git/gitweb.git (gitweb-repo remote)?	14:27
pasky	jnareb: I do, but topgit calls git send-email	14:27
	← eternaleye left	14:27
pasky	but probably in a way that confuses it	14:27
	jnareb: I'm sorry, I can't parse your last question	14:28
spearce	gilimanjaro: yea, that's possibly a reasonable work around for now. assume symlinks are not dirty if the index or HEAD says its a symlink.	14:28
	→ EmilMedve joined	14:28
jnareb	pasky: let me rephrase: which branch (if any) at git://repo.or.cz/git/gitweb.git does contain the patches you send to git mailing list?	14:28
pasky	next and pu :)	14:28
	but I'm not sure if it contains all of them at this point	14:28
	if you aren't gonna use topgit, I'll make an exported branches for you	14:29
	containing purely the patches	14:29
gilimanjaro	spearce: ok, i'll work in that direction then. putting the files in .git/info/exclude doesn't seem to work for me, so I'll have to patch to be happy I guess... :)	14:29
jnareb	Hmmm... 'gitweb-repo/next' has single-line commit messages (summary only)	14:30
	→ qrush joined	14:30
jnareb	or I will se them as they hit git.git ;-)	14:30
	s/se/see/	14:30
pasky	well they are topgit bracnhes	14:31
	so they are kinda raw	14:31
	if you use git to look at them :)	14:31
	I'll do the export	14:31
	if tg export actually works ;)	14:31
camwest	i'm having some trouble, when i go git commit -v it loads the commit message into "vi" but i'd rather it load textmate, when i go: echo $GIT_EDITOR it shows: "mate -w" but it doesn't load in textmate, any ideas?	14:31
jnareb	camwest: declare -x GIT_EDITOR="mate -w" ?	14:32
gilimanjaro	export $GIT_EDITOR?	14:32
cizra	I'm confused. "git status" tells me "Changed but not updated: modified: lots of files", but git reset --hard HEAD doesn't unmodify them.	14:32
gilimanjaro	" The editor used to edit the commit log message will be chosen from the GIT_EDITOR	14:32
	environment variable, the core.editor configuration variable, the VISUAL	14:32
	environment variable, or the EDITOR environment variable (in that order)."	14:32
	so you could also use the core.editor config var	14:32
jnareb	camwest: why can't you use core.editor in ~/.gitconfig?	14:32
camwest	jnareb: i guess i can, i didn't know about it though	14:33
spearce	pasky: why the hell does your taqcould stuff conflict with the pb/gitweb branch already in next? :-)	14:33
	→ pygi joined	14:33
	→ sabayonlive-2236 joined	14:33
sabayonlive-2236	is there any gui for git ? is it recommended ?	14:34
cbreak	tried git-gui?	14:34
	(I don't like it)	14:34
sabayonlive-2236	nope	14:34
	is it cool ?	14:34
	cbreak: did you use a gui ?	14:34
pasky	spearce: well, I currently integrate a lot of patches in parallel, so I try to base them only on minimal dependencies and if you already applied something I submitted, it might conflict	14:35
cbreak	I use cli-ui normally	14:35
seezer	there are also gitk and qgit	14:35
	← tomtt left	14:35
spearce	pasky: i think it was two trivial hunks that i was able to work out by hand.	14:35
sabayonlive-2236	which is the best ?	14:35
pasky	ok	14:35
jnareb	sabayonlive-2236: see http://git.or.cz/gitwiki/InterfacesFrontendsAndTools page	14:36
pasky	spearce: just apply what is easy for you, in few days I will look what made it and update + resend the rest :)	14:36
sabayonlive-2236	jnareb: thanks	14:36
	→ sveajobb joined	14:36
jnareb	sabayonlive-2236: gitk and git-gui, qgit, giggle, GitNub or GitX on MacOS X, git-cola (formerly ugit)	14:36
sveajobb	Ooops.. I accidently remove one stash-refs too many :\	14:36
jnareb	sabayonlive-2236: or you can try Eclipse Git plugin	14:36
sveajobb	is there a way to re-add it the last one?	14:36
	← johan-s left	14:36
spearce	pasky: your "unify shortlog/log links" won't apply to anything: master, pb/gitweb-tagcloud, or pb/gitweb (in next). wtf does it go to?	14:37
jnareb	unfortunately one of sign of email being spam is its lack of adherence to email RFC, as spam sending engines are very sloppy; so legitimate email send from imperfect tools also gets marked as spam	14:38
pasky	spearce: :))	14:38
cbreak	for some reason, git gui is german here :/	14:38
pasky	spearce: it's master + the extra actions patch	14:38
flukebox	how to use git hooks ??	14:38
spearce	cbreak: LANG=en_US git gui; its translated to a few languages, german among them.	14:38
pasky	spearce: I tried to use mail threading to designate the dependencies	14:39
	spearce: is that too confusing?	14:39
camwest	gilimanjaro: I set core.editor to mate -w and it still is trying to use VI for git commit -v	14:39
cbreak	hmm... ok...	14:39
spearce	pasky: yea, i don't see the email threading in my email client. :-)	14:39
cizra	Hey, how can I drop whatever I've been doing and reset everything to the same state as the upstream? I'm quite confused.	14:39
pasky	oh, really?	14:39
	oops	14:39
	→ nickh joined	14:39
	→ tongueroo joined	14:40
jnareb	cizra: git-stash (quite safe) or "git reset --hard HEAD" (abandon your changes)	14:40
spearce	pasky: nope, i can't get it applied over the action bar with custom links patch. even after cherry-picking that onto master. to try and get a temporary state.	14:41
pasky	that's strange	14:41
cizra	jnareb: The latter doesn't work. It still shows that I have uncommitted modifications and that I'm ahead of the origin/master by 2 commits.	14:41
Mikachu	cizra: git reset --hard origin/master	14:42
gilimanjaro	camwest: what does git-config --get core.editor say?	14:42
camwest	Cameron-Laptop:trunk camwest$ git-config --get core.editor	14:42
	mate -w	14:42
cizra	Mikachu: That got rid of one set of modifications, one remains.	14:43
spearce	pasky: sorry. i'm going to ask you for a resend after i publish my tree today and get you to merge it for me. ;-)	14:43
	→ sergiodj joined	14:43
pasky	sure :)	14:43
Mikachu	cizra: git clean -n lists that set?	14:44
gilimanjaro	camwest: that's weird... only GIT_EDITOR has higher prio	14:45
sabayonlive-2236	jnareb: that is your favorite gui ?	14:45
	jnareb: whats*	14:45
camwest	gilimanjaro: GIT_EDITOR is set to mate -w too	14:45
cizra	Mikachu: No, it shows nothing.	14:45
Mikachu	then i have no idea what changes you are talking about	14:45
jnareb	sabayonlive-2236: Emacs ;-)	14:46
mutex	the internet ate a patch I sent in last night	14:46
	← warreng left	14:46
mutex	strange!	14:46
camwest	gilimanjaro: i can go git commit -v \| mate -w and that works fine, but it's a big hassle =)	14:46
lack	mutex: Some one has to feed the internet to keep it alive. Your turn was apparently last night.	14:46
jnareb	sabayonlive-2236: to be more serious, I use gitk from time to time, qgit from time to time, and tig from time to time, but all rarely	14:46
mutex	:-(	14:46
gilimanjaro	camwest: GIT_EDITOR works for me	14:46
	camwest: what OS?	14:46
camwest	gilimanjaro: it was working for me fine up until about a day ago	14:46
	gilimanjaro: OSX leopard	14:47
	→ athought joined	14:47
	→ bts- joined	14:47
gilimanjaro	gilimanjaro: well so am I, so that's not the problem :) (doh! that textmate)	14:47
camwest	haha yeah that textmate hehe	14:47
cizra	Mikachu: git commit -a -m foo (at this point the "modified" file list disappears); git reset --hard origin/master (at this point it reappears)	14:47
gilimanjaro	camwest: what shell?	14:47
camwest	i dunno, how do i check? the default (bash i think?)	14:47
	i can go echo $GIT_EDITOR from the command line and it points to mate	14:48
Mikachu	cizra: that sounds unlikely	14:48
cizra	Mikachu: I'm trying to have Windwoes git mirror a Linux-hosted repository.	14:48
	Mikachu: Might confuse something.	14:48
gilimanjaro	camwest: i have no clue; i can imagine git has an issue with a graphical editor for some reason	14:49
	k-s[AWAY] → k-s	14:49
	← tongueroo left	14:49
camwest	gilimanjaro: yeah just weird i'm not getting any errors, /shrug	14:49
sabayonlive-2236	jnareb: does emacs has a git visualizer or you just hack your emacs ?	14:49
camwest	gilimanjaro: thanks for your help	14:49
	→ tongueroo joined	14:49
gilimanjaro	camwest: possibly because it returns to the command prompt immediately	14:49
camwest	gilimanjaro: that's what -w is for.	14:49
	gilimanjaro: " -w, --wait Wait for file to be closed by TextMate."	14:50
pasky	spearce: oh, but none of the patches have hit master yet, right?	14:50
	← pygi left	14:50
pasky	hmh..	14:50
gilimanjaro	camwest: ah ok. i don't use textmate myself... i'll see if i can get hold of it	14:50
pasky	now I wonder how to handle this best	14:50
	next never converges to master, right?	14:51
	and rebases occassionally	14:51
	→ RaceKondition joined	14:51
spearce	pasky: right. your current pb/gitweb topic is likely to merge soon. but next itself will never merge to master.	14:51
	→ Garoth joined	14:51
	→ pygi joined	14:52
	→ thechuck joined	14:52
pasky	hmm, so I can stick my head in ground and wait until pb/gitweb gets to master	14:52
gilimanjaro	camwest: hold on... you're doing export GIT_EDITOR="mate -w"	14:52
camwest	yeah	14:52
pasky	or do some hacky stuff, but putting my development to branch that gets rebased basically destroys all my properties	14:52
	I can't really do that :(	14:53
spearce	pasky: pick out the branch tip from within next. its unlikely to rebase, usually branches in next merge to master.	14:53
	mw\|out → mw	14:54
pasky	"branch tip from within next"?	14:54
	→ jpeeler joined	14:54
pasky	what is that?	14:54
lack	Can I specify multiple lines in .git/objects/info/alternates?	14:54
spearce	find the merge pb/gitweb into next commit, use its 2nd parent as your branch base.	14:54
	lack: yes	14:54
lack	spearce: That's exciting, thanks :)	14:54
pasky	oh	14:55
	ok	14:55
	← nille left	14:55
	→ Ace_NoOne joined	14:56
	← bts- left	14:57
	← RaceCondition left	14:58
	→ brasko joined	14:59
brasko	git status says 'modified: MarathonHistory.h', but git diff is empty	15:00
jnareb	sabayonlive-2236: GNU Emacs is just editor, and I use vc-git.el, i.e. VC interface to Emacs (on level of individual files), so I can see for example if file is modified wrt HEAD, and on which branch I am. I have used git.el occasionally...	15:00
brasko	I did 'git rm foo.c' 'git mv bar.c foo.c', git diff returns nothing	15:00
	bug in git? or unsupported use case?	15:00
	or dumb user?	15:00
jnareb	sabayonlive-2236: currently there is no graphical interface in Emacs to Git; I use command line tools: git-commit, git-add, stg-new, stg-refresh, etc.	15:00
	→ girishr_ joined	15:01
	→ cbreak_ joined	15:02
	← boto left	15:03
	← cbreak left	15:03
	cbreak_ → cbreak	15:03
	→ Ilari joined	15:07
	← emiel_ left	15:07
	→ mlins joined	15:07
gilimanjaro	spearce: I'm looking at GitIndex.java, and looking at lines 518 & 519 it seems like symlinks are currently always considered modified... or am I misunderstanding something?	15:07
	← mlins left	15:07
	→ mlins joined	15:07
	← mlins left	15:08
	← cizra left	15:08
	→ cizra joined	15:08
spearce	gilimanjaro: yup. but you may be able to mark the index entry "assume unchanged" and have it hit line 492-493, which bypasses that code.	15:09
	that assumedValid is i think called assume unchanged in command line git	15:09
	→ emiel_ joined	15:09
	← emiel_ left	15:10
	→ davidfetter joined	15:10
spearce	yup, git update-index --assume-unchanged	15:10
gilimanjaro	spearce: yes, i saw that in my context-menu	15:10
	← rdavila left	15:10
	→ emiel_ joined	15:10
spearce	maybe that's the clicky thing you were looking for earlier to mark those and ignore them?	15:10
gilimanjaro	spearce: but what's the logic begin assuming every symlink is always modified...	15:10
	→ rdavila joined	15:11
	← koke left	15:12
gilimanjaro	spearce: if i do the clicky, the commit dialog changes. it is now 'Added' instead of 'Added, index diff'	15:13
	→ werdnativ joined	15:14
	← sveajobb left	15:15
	← girishr left	15:16
jnareb	pasky: would you be resending your gitweb patches (those which didn't got accepted, that is)?	15:17
spearce	gilimanjaro: yea, you are right, that logic doesn't seem correct to me neither. it shouldn't be dirty just because its a symlink.	15:17
gilimanjaro	spearce: but even if it's not dirty, it still shows up as Added	15:18
	spearce: I think the problem is that headTree.findBlobMember(repoPath); isn't finding the file	15:18
pasky	jnareb: I'm not really sure yet, since in the end I would prefer to wait for the accepted things to pop up in master, it turns to be rather hacky to update it against spearce's branch tip, and also I'm still afraid of that operation	15:19
	jnareb: but I'm still working on creating "clean" branches for you	15:19
	← ia left	15:19
jnareb	pasky: thanks a lot	15:19
	By the way, I think the whole 'switch maintainership for a short while' thing clearly shows superiority of distributed version control systems, and having central repository by social convention, not by technical reasons	15:20
gilimanjaro	spearce: i don't know much about git internals, but i assume a symlink is a blob in a repo just like a regular file?	15:21
spearce	gilimanjaro: correct	15:21
jnareb	gilimanjaro: the only difference is in 'tree' object holding reference to it	15:21
	← sgrimm left	15:21
gilimanjaro	gilimanjaro: so the actual bug is that that method isn't finding the symlink blob	15:22
	→ warreng joined	15:22
	→ paggas joined	15:22
	← Flathead left	15:22
	← brasko left	15:23
gilimanjaro	spearce: though even if it did find it, it would mark it as changed because java wouldn't be able to compare the contents of the symlink on filesystem against the blob content	15:23
spearce	yup. :)	15:24
pasky	jnareb: yes, it's working out great (not that I would be unhpapy with Junio or anything :)	15:24
spearce	that code is why i've been trying to move away from the Tree class and instead use TreeWalk and NameConflictTreeWalk.	15:24
gilimanjaro	spearce: but that could be fixed with assume unchanged. first thing is to find the blob though, so it's no longer considered 'Added'	15:25
spearce	f'ing linus and his tree sorting algorithm.	15:25
Ilari	gilimanjaro: Symlinks have type field '12' whereas regular files have type field value '10' (directories have '4' and subprojects have '16'). The type field is in the tree object that points to object itself.	15:25
	stouset → stouset\|away	15:27
jnareb	spearce: didn't Linus agreed that the_index is purely local matter, and migh be changed?	15:27
Ilari	jnareb: The problem is that the sorting order is part of tree object syntax...	15:28
spearce	jnareb: yes. but we can't do anything about the #@*#@!(@!(! stupid sorting in a tree	15:28
	← Aleph_One left	15:28
jnareb	ah, sorry for the noise then	15:29
Ilari	... At least nothing backward-compatible...	15:29
	pasky headaches on tg export	15:29
jnareb	I assume that here were some reasons for that... but I seem to remember Linus admitting error there (and now backwards compatibility...)	15:29
Ace_NoOne	question: is there a reason auto-completion doesn't work for "git log remotes/..."?	15:30
jnareb	Ace_NoOne: probably becuse it assumes you want to use "git log <remote>/..."	15:30
	but it looks like it doesn't work either	15:31
Ace_NoOne	jnareb: not sure I understand	15:31
pasky	jnareb: ok, I've created clean/* branches - unfortunately, they don't yet share history, so e.g. clean/next includes all of clean/master history, but independently	15:31
	jnareb: but that shouldn't be big deal	15:31
jnareb	pasky: thanks a lot!	15:31
	git remote update	15:31
pasky	clean/next is basically all that is interesting, but you probably don't want to look at the blame and diff bits of next, on the other hand	15:31
jnareb	Oops, not here...	15:32
pasky	(I didn't submit any of these, they just linger there :)	15:32
	oh and two of the commits are empty, these were already merged upstream :)	15:32
	hmm, still a lot of work to do polishing topgit :)	15:33
Ilari	Ace_NoOne: You can usually shorten 'remotes/origin/master' as 'origin/master'.	15:33
jnareb	pasky: doesn't TopGit have "tg clean --applied" or something like that, like StGit has? or 'do not rewrite history' prevents that?	15:33
	← grahal left	15:34
	→ Vortex34 joined	15:34
jnareb	pasky: I see you (re)implemented project search form in gitweb	15:34
	→ dec joined	15:34
	→ ia joined	15:34
	→ sgrimm joined	15:34
pasky	jnareb: it doesn't prevent that, i just didn't implemented it yet	15:35
	← Vortex35 left	15:35
pasky	jnareb: oh, yes, today i rediscovered your patch from spring :)	15:35
	but this one is a lot simpler	15:35
	though with less fnacy features	15:35
	stouset\|away → stouset	15:35
pasky	afk	15:36
cizra	Is it true that git-svn repos should not be cloned?	15:37
	← paggas left	15:37
jnareb	pasky: Reading commit description in my StGit stack I see two such 'fancy' features: lazy loading of information (only this part of information that is needed for search is loaded / generated), and having sorting work with searching for projects	15:37
	cizra: git-svn information is local to repository, so if you clone it you would clone only git bits (as far as I understand it)	15:37
	→ etgryphon joined	15:38
cizra	jnareb: Assuming that this is OK with me, is it the only issue? I'd only use git to manage the clones.	15:39
	← _dave_h_d left	15:40
gilimanjaro	spearce: i'm trying to do this properly, so i want to write a test that fails. i'm guessing it should be in org.spearce.jgit.lib.ReadTreeTest?	15:41
jnareb	I don't know, as I don't use git-svn. That is one issue I know about.	15:42
deskin	cizra: if you aren't using noMetadata, git-svn can rebuild SVN metadata from the git-svn-id: lines in the commit messages; so, your clone can actually work with svn directly if you want (but doesn't need to)	15:42
spearce	gilimanjaro: yea, that seems like a reasonable test case class for this	15:42
gilimanjaro	spearce: i'm a java-n00b so this will get hairy... :)	15:43
deskin	at any rate, sorry I dropped off earlier; I think I'm close to diagnosing your '# in pathname' issue earlier, could you help by pasting or /msg me with the last several lines of output you see when the clone fails?	15:43
spearce	gilimanjaro: :-) java is just like c, but without the fun of segfaults, and with this weird thing called "java". :)	15:43
Ace_NoOne	Ilari: well, I'm dealing with SVN in this instance - so would that be remotes/trunk == trunk?	15:44
gilimanjaro	spearce: i'm a perl guy... so everything is about as alien as it could get	15:44
jnareb	spearce: mmap; lstat; was: license ;-)	15:44
cizra	deskin: Hmm, OK	15:44
spearce	jnareb: what are you listing what you don't like about java? it has mmap, but it sucks a** because you can't control when the munmap occurs.	15:44
	jnareb wonders why there is no git web interface in Python...	15:45
Pieter	all pythonists use mercurial ;)	15:46
	← dec_ left	15:46
	← thechuck left	15:47
jnareb	Pieter: git-cola and pyrite and git-p4 and yap/yagp contradicts it...	15:48
	← camwest left	15:48
	→ ashleyw joined	15:48
jnareb	Pieter: but it looks like people (like one of gitweb-XMMS2 fork) move from Python to Ruby for web applications... at least it seems that way to me.	15:48
spearce	sadly i've written thousands of lines of python in the last few weeks. :-\|	15:49
robinr	python is the google native language?	15:50
	stouset → stouset\|away	15:50
	stouset\|away → stouset	15:51
spearce	python is one of the chosen few. the others are C++ and Java. it depends on what you are doing as to which language is the chosen one.	15:51
	right now i'm writing tools meant to run on end-user systems that wrap around git. so python was chosen. :-\|	15:51
robinr	you mean "forced" :)	15:52
	there are worse languages out there. trust me.	15:52
jnareb	is Perl ever used for GUI?	15:53
andres	jnareb: there are relatively large perl-tk and perl-gtk applications around	15:53
jnareb	spearce: you could have write in C#, or VB.NET	15:53
robinr	push home robin	15:53
spearce	robinr: i have a strong distaste for python now, and i can actually justify it now that i've coded in the damn thing.	15:53
	→ Anon134 joined	15:53
	→ johan-s joined	15:53
robinr	jnareb: think worse. COBOL or RPG	15:53
cizra	APL!	15:54
Anon134	NICK/ STARWARSFAN	15:54
	Anon134 → STARWARSFAN	15:54
robinr	cizra: APL is quite nice.	15:54
STARWARSFAN	BYE	15:54
	← STARWARSFAN left	15:54
jnareb	robinr: COBOL always reminds me of Userfriendly.org strip, where Dust Puppy admits to writing Edwin A.I. in COBOL on dare ;-)	15:54
cizra	robinr: Dunno, I've never written it. But it must be a pain to use all those weird symbols.	15:54
robinr	I was impressed when one gui did in APL in 20 lines, basically what our 3000 lines of C++ did.	15:54
	it is just a very compact functional language	15:55
	my guess is those few APL people should learn to use more lines and comments	15:56
	← athought left	15:56
robinr	cizra: I've never used it either so it was fun to see a real-world example and someone to explain it.	15:57
jnareb	Let's reimplement Git in Smalltalk ;-)))))	15:57
robinr	I guess one can learn to read it.	15:57
	reading tip: you read APL right-left	15:57
	Cya	15:58
eMBee	spearce: could you elaborate yopur distaste? in personal message if you don't want to make it public?	15:58
spearce	not having an end of block marker is a huge annoyance to me. i don't care what it is, but I want _something_ there to end the indentation level.	15:58
	← lionr left	15:59
spearce	its lack of being able to scope variables to a block (like to the inside of a loop or the body of an if-then statement) is also really annoying, and has bitten me multiple times.	15:59
	→ mtodd__ joined	15:59
spearce	at least perl has my and/or local, and with my can verify your variable usages under "use strict"	15:59
	← juanjoc left	16:00
spearce	hmm. small update to git.git today, only next/pu.	16:00
	eMBee agrees with the scoping but not with the block end marker, but the later is a taste issue :-) anything else?	16:00
Mikachu	spearce: the thing that really confused me is that you have to declare global variables _inside_ the function that uses them	16:00
	spearce: otherwise it defaults to a local scoped shadowing variable	16:01
spearce	yea, that's really freaking annoying. its like tcl all over again.	16:01
	→ selckin joined	16:01
spearce	oh, and statements ending at LF, except when there's a paren or something holding the line still open. also annoying.	16:01
	its like half-TCL.	16:01
	i still haven't fully learned the object model either, so i'm still having issues figuring out how to reference what should be class statics from within the class instance members.	16:02
	oh, and needing to say f'ing "self." all over the place. perl also has that feature but whatever. i like not needing to type things out for the compiler. at least C++ and Java have "this." as implicit.	16:02
	← brendon_work left	16:02
lack	Is there any difference between specifying 'ssh://.../foo.git' and 'git+ssh://.../foo.git'?	16:03
	→ grahal joined	16:03
spearce	the latter takes more time to type?	16:03
	they are the same protocol.	16:03
	← peritus_ left	16:04
spearce	actually most people prefer "user@host:foo.git". its even shorter.	16:04
Mikachu	the only time you need ssh:// is when you have to specify the port, i think	16:04
lack	Ah, so the ssh:// is implied if you don't have a protocol on there?	16:04
	Nicke.	16:04
Mikachu	lack: sort of, but the syntax is different	16:04
	lack: host:path defaults to ~, while ssh:// starts in /	16:04
	→ patmaddox joined	16:05
	← pygi left	16:05
lack	Of course. Thanks :)	16:05
jnareb	lack: if you don't have protocol, it defaults to local filesystem	16:06
	← mtodd_ left	16:06
	← EmilMedve left	16:06
jnareb	lack: almost file:// (but tries hardlinking first)	16:06
Mikachu	if you don't have a protocol, and there's no : near the start, you mean :)	16:07
neurodam1ge	how do I in the post-receive use the two revision numbers to get a log between the revisiosn?	16:07
lack	Ah, git is too smart for its own good :)	16:07
j_engelh	git log number1..number2 ?	16:08
	← GarethAdams left	16:10
	← ciskje left	16:11
	← samjam left	16:12
	← ashleyw left	16:12
	mw → mw\|out	16:14
	← robinr left	16:16
	→ d0k joined	16:18
	deskin → deskin_away	16:18
jnareb	lack: actually file:// going as if it was remote clone is a bit of trick making use of the fact that git use cURL library	16:18
	→ samjam joined	16:19
	→ ashleyw joined	16:20
mutex	hmmm	16:21
	my last commit added a file to the index that I do not want added	16:21
	how can I edit the last commit ?	16:21
	I am familiar with --amend	16:21
	← ashleyw left	16:21
mutex	I suppose i could 'add' a git-rm for the file	16:21
	← emiel_ left	16:22
	← girishr_ left	16:22
lack	mutex: If you haven't pushed yet, git-revert may be your answer.	16:22
	Wait, I lie.	16:23
mutex	yeah revent commits, but I want to modify my top commit	16:23
	I have not pushed yet	16:23
	... I think...	16:23
Ilari	mutex: Then 'git rm --cached' it and then --amend?	16:23
mutex	that was my thought	16:24
	actually	16:24
	I tried that last night	16:24
	apparently git rm --cached with a file name is deprecated	16:24
lack	mutex: I think 'git-reset' is wath I was thinking about...	16:24
	→ charon joined	16:24
mutex	ah yeah git-reset	16:25
	my bad	16:25
	and yeah that did it	16:25
	→ zachinglis joined	16:26
	← bryanray left	16:29
	← v_silent left	16:31
	← mw\|out left	16:32
	← RaceKondition left	16:32
	dwmw2 → dwmw2_gone	16:36
	← willb left	16:37
	→ schacon joined	16:37
	→ iulian joined	16:38
	→ akt joined	16:38
	← akt left	16:39
	→ p4tux joined	16:39
	→ t_ joined	16:39
t_	while hunting a bug I'm trying to hack some printfs in builtin-pack-objects and failing to do so. any idea?	16:40
	s/hack som/hack in some debug/	16:41
	→ willb joined	16:41
schacon	it's probably being called from another command and the stderr/out is being redirected	16:41
mutex	t_: I tried that the other day too	16:41
	failed to work	16:41
schacon	what command are you running?	16:41
mutex	try another fd	16:41
t_	I'm trying to teach progress obey -q :D	16:42
	← _slacker_ left	16:42
jnareb	t_: GIT_TRACE could help you	16:42
	→ bdiego joined	16:42
t_	as there's on fprintf(stderr in builtin-pack-objects.c I assumed it would be fine just to see that it's not that easy and there's some reconfiguration going on	16:42
	jnareb: will try...	16:42
	← alexross left	16:43
	→ niki joined	16:43
	→ _jonesy_ joined	16:46
	← _jonesy_ left	16:46
t_	jnareb: it's more involved than that. I get some traces but not mine	16:48
schacon	t_: what command are you running? actually running 'git pack-objects' or is it being called from something else like 'push'?	16:49
t_	schacon: git clone with or wihout -q	16:49
	and as I see it progress_*() does not have a way to obey -q yet	16:50
	← kanru left	16:51
schacon	i started seeing the printf's when I added this to builtin-fetch-pack.c : cmd.stdout_to_stderr = 1;	16:51
	→ kanru joined	16:52
	→ aaa joined	16:52
schacon	right before the start_command(&cmd) line	16:54
	→ dkerschner joined	16:54
dkerschner	so I have a file I've commited but I want to see the changes that were made with the last commit. Is there a quick way to do that other than getting the commit number?	16:54
	→ WALoeIII joined	16:55
dkerschner	like git diff previous file or something	16:55
	→ Weasel[DK] joined	16:55
	aaa → Beket_	16:55
t_	schacon: still searching....	16:55
jnareb	git diff file?	16:55
	git diff HEAD file?	16:55
	git log -p -- file?	16:56
	← BarryCarlyon left	16:56
	← powerfox left	16:56
	← kbingham left	16:56
dkerschner	ah. git log -p is what I wanted. Thanks	16:57
	guess I could just use gitk too	16:57
	← Ace_NoOne left	16:58
	← kevwil left	16:58
	→ corecode joined	17:01
MarcWeber	Why does git say "Finished one cherry-pick" if the cherriy pick has already be done?	17:02
	→ juanjoc joined	17:02
cizra	Git loves picking cherries!	17:02
t_	schacon: heh. still searching and haven't found who call cmd_pack_objects(). it's either from builtin-clone.c or git.c or something else. maybe I can find it	17:03
	→ EmilMedve joined	17:03
	← nickh left	17:04
	→ Neo_The_User joined	17:04
schacon	clone should call fetch-pack if you're doing it over git:// or ssh	17:06
	← Beket left	17:08
	→ yofel_ joined	17:08
	← yofel_ left	17:09
	← EmilMedve left	17:13
	→ bobesponja joined	17:13
	← Neo_The_User left	17:13
tango_	jnareb, http://git.oblomov.eu/git?a=commitdiff;h=68fbb09605f0f3bc597a0b19ae1f47aae01c9971 <- how's this for the input parameter rework patch?	17:14
jnareb	or file://	17:14
	← jnareb left	17:14
tango_	wow I killed him 8-D	17:14
	← flaguy left	17:16
	→ eternaleye joined	17:16
t_	schacon: hmm. after no success I just tried to disable all printfs in progress.c to no avail. I know that my changes are included in /usr/local/libexec/git-core/* as I can find my magic strings in the binaries. interesting...	17:16
	← gilimanjaro left	17:18
schacon	out of curiosity, is 'whatchanged' just an old 'log'?	17:21
	→ dec_ joined	17:21
	← Aikawa left	17:23
	→ Aikawa joined	17:23
	← bobesponja left	17:24
	onitunes\|away → onitunes	17:25
	→ mdalessio joined	17:25
spearce	schacon: yes	17:26
	→ bobesponja joined	17:29
Mikachu	looks like it's just git log --raw	17:29
	→ RaceCondition joined	17:29
	← bdiego left	17:30
schacon	i remember years ago it used to be more useful, but I couldn't remember what changed exactly	17:30
	→ nickh joined	17:31
johnw	I've having the hardest time cloning from a machine over ssh which puts Git in a place not on my PATH. The remote server won't source .bashrc either, and using -u to clone doesn't work at all	17:31
	any further suggestions? is there a way I can change the ssh command being used to login?	17:31
	→ deavid joined	17:32
	→ joydivider joined	17:32
	← cizra left	17:32
	→ cizra_ joined	17:32
johnw	gitProxy?	17:32
	→ bdiego joined	17:32
	→ blowery_ joined	17:33
	→ Jammer joined	17:34
	← dec left	17:34
Jammer	I have local repo and I just commited, but forgot to add few files. Now when I try to do git commit -a --amend I get error "fatal: empty ident <jammer@host.(none)> not allowed"	17:37
	→ jchris joined	17:37
	← rdavila left	17:39
j_engelh	set your host ident correctly	17:40
Jammer	what does that mean exactly?	17:40
j_engelh	like, your hostname.	17:40
Mikachu	even better set user.name and user.email	17:41
	→ _graham_ joined	17:42
Jammer	Mikachu, I set them with "git var" (dunno if there's something I need to do to actually update it?)	17:42
Mikachu	you can't set anything with git var...	17:42
	deskin_away → deskin	17:43
	← eternaleye left	17:43
Jammer	err sorry git config I mean :)	17:43
	and it shows up in git var	17:43
johnw	it apperas i'm running into an actual bug in 1.6.0	17:44
Jammer	some solutions seemed to suggest the lack of full name in shell could effect this? I don't have root access because it's shared hosting environment	17:44
Mikachu	johnw: try 1.6.0.2	17:44
johnw	ok	17:45
	i'm in a situation where I actually cannot change my PATH upon login via ssh	17:45
	→ kyrus1 joined	17:45
Mikachu	so?	17:45
johnw	so i can't tell it where "git" lives	17:45
Mikachu	set remote.<name>.{upload,receive}pack	17:46
	→ pureluck joined	17:46
	→ perezd88 joined	17:46
pureluck	hey. i can't receive objects from a repository. it freezes at 0%. any reason for this?	17:46
t_	schacon: ok, trace_printf works but not yet where I want it and I also found a thread about this problem and args.quiet and args.no_progress should have fixes this but apparently not quite or not anymore. http://marc.info/?l=git&m=121529226023180&w=2	17:47
	s/fixes/fixed/	17:47
	← blowery left	17:47
	← jds left	17:47
	→ digitalhobbit joined	17:48
	→ boto joined	17:51
	→ bronson joined	17:51
	→ EmilMedve joined	17:54
	← tvw left	17:56
	← thiago_home left	17:59
	← spearce left	17:59
	← dmq left	17:59
	stouset → stouset\|away	18:00
	→ tvw joined	18:02
	→ x_or joined	18:02
x_or	I'm working on a project where the team has decided to use SVN rather than git. Is it possible for me to use git and then git diff and patch into the svn repository? Is this a really stupid idea? They are not open to using git right now.	18:03
cbreak	there's git svn (not used it myself). man git-svn might help	18:04
Gitbot	cbreak: the git-svn manpage can be found at http://git.or.cz/man/git-svn	18:04
gebi	x_or: take a look at git-svn	18:04
	→ markkalderon joined	18:04
x_or	I did import the project and have a git repo now, but I need to basically commit my changes back into svn. Does git-svn allow me to go back and forth, or only one way (svn to git)?	18:04
pureluck	._.	18:04
gebi	x_or: man git-svn	18:05
Gitbot	x_or: the git-svn manpage can be found at http://git.or.cz/man/git-svn	18:05
x_or	gebi: Thanks, I will re-read this.	18:05
pasky	(both ways)	18:05
gebi	dcommit is what you want ;)	18:05
	→ bts- joined	18:06
	← perezd88 left	18:07
	→ johnw_ joined	18:08
	→ olly joined	18:09
	→ girishr joined	18:11
	← johnw left	18:15
t_	DOH! "remote: " is a string received fromthe remote site. /me bangs head on table until it bleeds	18:18
	← dsop left	18:18
	→ jesselucas joined	18:18
t_	ENGLIGHTENMENT!	18:19
	→ thiago_home joined	18:21
	→ dsop joined	18:22
	→ gilimanjaro joined	18:24
	→ felipec joined	18:27
Pieter	RFCs 2822 and 2821 are now officially obsoleted by RFCs 5322 and 5321	18:29
	ooh	18:29
	← thiago_home left	18:29
pasky	does returning content type application/xhtml+xml instead of text/html have any _positivE_ effect?	18:29
johnw_	the e-mail rfc?	18:29
Pieter	yes	18:30
johnw_	what's new?	18:30
Ilari	pasky: I think no (except if you want to use XHTML 1.1)...	18:30
Pieter	Using port 587 is (finally) recommended	18:30
felipec	is there some kind of tagged submodule?	18:30
	like . = upstream/1.0, debian = debian/1.0	18:31
	→ BarryCarlyon joined	18:31
pasky	Ilari: then why does gitweb do it? :)	18:34
Ilari	pasky: Dunno... What (X)HTML version does it use?	18:35
	→ GSaldana joined	18:35
	← GSaldana left	18:36
Pieter	if you send the xhtml header, you can use xml constructs that are not allowed in html :)	18:36
	j_engelh tries submitting RFC to obsolete Pieter and replace by Pieter~	18:37
j_engelh	You know what surprised me... the Internet Message Protocol always shared at least two digits with its predecessor. RFC 822 -> 2822 -> 5322.	18:38
	→ rwaite joined	18:38
j_engelh	This high occurrence of the digit 2 reminds me of ST:TNG episode "Yesterday's Enterprise"	18:39
	← rmh3093 left	18:40
	→ _jonesy_ joined	18:42
	← dotech left	18:43
	← _jonesy_ left	18:43
	→ bbs joined	18:44
bbs	i have a tag that i wanna push	18:44
	git push <tag>	18:44
	nvm i'm dumb	18:44
	got it	18:44
	← bbs left	18:44
Mikachu	yw	18:44
	← EmilMedve left	18:44
	← RaceCondition left	18:45
	← pureluck left	18:45
	← chris2 left	18:47
	→ thiago_home joined	18:47
	← grahal left	18:48
pasky	j_engelh: did you get my msg?	18:49
	Ilari: xhtml 1.0	18:49
	Pieter: like what constructs?	18:49
	→ grahal joined	18:49
pasky	Pieter: the problem is that users like to submit READMEs with weird html	18:50
Ilari	pasky: Isn't XHTML 1.0 just XML version of HTML 4.01 strict?	18:50
pasky	which doesn't bode well	18:50
	Ilari: i don't know :)	18:50
	← girishr left	18:50
j_engelh	Ilari: not at all	18:50
Pieter	pasky: like <script src="src.js" />	18:50
cbreak	GitX looks interesting.	18:50
	→ Beket joined	18:50
	← grahal left	18:51
Ilari	j_engelh: Well, some things are done differently, but aren't the elements available the same?	18:51
j_engelh	Phew...	18:51
	I am happy that all commmands I know are recognized by my browser ;-)	18:52
	Since it's based on 4.01 Strict, I would not expect XHTML to know frames	18:52
	softdrink pets xhtml 1.1 strict	18:53
pasky	Pieter: hmm	18:53
	→ _jonesy_ joined	18:53
pasky	Pieter: well that still seems to work for me ;)	18:53
	← statim left	18:54
Pieter	pasky: depends on the browser :)	18:54
	← Tv left	18:54
pasky	oh never mind	18:54
	i use </script>	18:54
Pieter	yeah, that's the easiest way :)	18:54
	→ EmilMedve joined	18:54
	→ statim joined	18:54
	stouset\|away → stouset	18:55
	→ jatt joined	18:55
	→ polpak joined	18:56
tango_	is there a particular reason why gitweb should spout xhtml instead of just html?	18:58
	tango_ hates xhtml	18:59
johnw_	would it even make a difference in this case?	18:59
softdrink	how come?	18:59
	← jatt left	19:00
	← olly left	19:00
	→ jatt joined	19:00
	← zachinglis left	19:00
	← Beket_ left	19:01
	→ anvimal joined	19:01
	→ gk joined	19:02
	→ lol_h joined	19:02
	← jatt left	19:04
	← _jonesy_ left	19:04
	→ timcharper joined	19:04
t_	got it	19:05
gk	hello, suppose I have local bare repository - can I have "detached" workdir on the same system without any cloning? for example some script checkouts some branch of that repo into some temp dir, does some work, commits and deletes the tmp dir	19:06
	→ CelticSoul joined	19:06
	← nickh left	19:07
	→ spearce joined	19:07
cbreak	you can just clone it	19:07
	I think it tries to use hard links	19:08
	maybe --shared works, but it's wrapped in warnings in man git-clone, so maybe not a good idea	19:08
Gitbot	cbreak: the git-clone manpage can be found at http://git.or.cz/man/git-clone	19:08
	→ popol joined	19:11
	← thiago_home left	19:11
Ilari	gk: git-new-workdir (in contrib)?	19:12
	gk: But creating multiple working dirs can lead to problems if you checkout the same branch twice...	19:14
	← bentob0x left	19:16
	← lol_h left	19:23
	← flukebox left	19:25
	→ Phillemann joined	19:28
	→ Jacolyte joined	19:28
	← Jammer left	19:29
	← bronson left	19:31
Phillemann	What's the .git/description file used for?	19:32
	← _doomster_ left	19:33
	→ _doomster_ joined	19:33
Phillemann	I'm looking for a way to print the name of a repository.	19:33
	→ rmh3093 joined	19:34
mxey	the name?	19:34
Freundlich	The thing is we want to get the repository name within a hook script.	19:35
Phillemann	Well...yeah. git has no notion of a "name" I guess, but maybe something that comes close to that. :)	19:35
Freundlich	So, maybe .git/description is the right thing to look at?	19:35
	← p4tux left	19:35
mxey	Phillemann: The closest thing is the directory path.	19:35
Phillemann	Hehe, _that's_ what we're currently using.	19:35
Freundlich	mxey: That's what we have been using...	19:35
Phillemann	Freundlich: Guess we have to stick to that...	19:36
Freundlich	But that feels kind of weird.	19:36
cbreak	you could give it a name in the form of a text file that is part of the repository :)	19:36
	→ zachinglis joined	19:37
	→ netoman joined	19:38
	← jerbear left	19:38
Mikachu	gitweb shows the contents of the .git/description	19:39
	→ kalikiana joined	19:45
	← qrush left	19:45
kalikiana	Heya, silly question: I have documentation k	19:45
	marked up as restructured text	19:46
	Now git thinks it's full of conflicts due to === functioning as markup	19:46
	Is there some trick to make an exception when committing that file?	19:46
	ie. so I don't always see the "conflicts"	19:46
Mikachu	kalikiana: are you using windows?	19:47
kalikiana	Nope, linux	19:47
	How is that relevant?	19:47
Mikachu	afaik there's only a hook that checks for conflict markers, git itself doesn't care	19:47
	→ icwiener joined	19:47
Mikachu	on semi-old windows versions of git, the hooks were enabled by default	19:47
	← Beket left	19:47
kalikiana	Well, I have nothing against the complaints with source files	19:48
	That's totally fine	19:48
	It just is irritating in this case	19:48
johnw_	Mikachu: upgrading everything to 1.6.0.2 fixed it, thanks	19:48
drizzd_	kalikiana: have you had a look at the hook then?	19:49
kalikiana	To be honest no, because I'm uneasy with perl :)	19:50
	kalikiana looks now	19:51
kalikiana	Oh, regex. Maybe I'll just tweak the expression	19:51
	← polpak left	19:51
kalikiana	Would it be correct to assume that a conflict with === does not contain any spaces between the =?	19:52
Mikachu	i think you can even assume what number of === there are	19:52
	and i think it might be safer to check for <<<<<< than =====?	19:53
j_engelh	yes	19:53
	in PHP at least you have === as an operator	19:53
	while <<< exists only in the whacky realm of java	19:53
	← Alien_Freak left	19:54
j_engelh	or was it >>> I dunno..	19:54
Mikachu	well, there are 7(?) in conflict markers	19:54
j_engelh	so just resolve them	19:54
Mikachu	?	19:54
kalikiana	The hook as it is checks for 7 <>=, ie /^(?:[<>=]){7}/	19:54
	So it checks all those characters you mentioned :P	19:55
Mikachu	you could check that there is a <<<<<<< then a ========= and then a >>>>>>>>	19:55
deskin	j_engelh: (ot) think it's >>>, signed vs. 0-extended shift	19:55
	→ thiago_home joined	19:55
j_engelh	people could have just used / 2	19:56
Mikachu	isn't <<< and >>> rotate in some languages?	19:56
	j_engelh rotates Mikachu	19:56
Mikachu	ie 1101 <<< 2 gives 0111	19:56
j_engelh	is not there any pseudo-spec as to what markers git/other vcs use	19:57
	like /^<{n}$/ replace n appropriately	19:57
Mikachu	git usually puts the commit message for the respective commits on the < and > lines too	19:57
	and then in newer git there is the \|\|\|\|\|\|\|\| marker too	19:57
j_engelh	what's that for	19:57
Mikachu	shows the base hunk	19:57
j_engelh	when do I get it?	19:58
Mikachu	git checkout --conflict=diff3 thefilewithaconflict	19:58
	or you can make it the default with some option	19:58
gk	cbreak: when I want to have separate workdir for script I am not doing this because of space savings, basically when the scripts runs nobody knows in what state the "real" workdir is	19:58
	so the sanest thing to do seems to be to use some other temp working dir	19:58
Mikachu	gk: just set GIT_WORK_TREE and GIT_DIR appropriately and i think it should work	19:59
cbreak	a normal clone should give you a new workdir too	19:59
	← zachinglis left	19:59
cbreak	completely independent from the original	19:59
	← softdrink left	19:59
cbreak	(without the local changes though)	19:59
	→ olly joined	19:59
	← olly left	19:59
	← Phillemann left	19:59
gk	cbreak: exactly, the refs will be different and besides it seems odd to create clone just for 100ms of the script run :)	20:00
	Mikachu: thanks, I will try	20:00
	→ zachinglis joined	20:01
	← aroben left	20:03
	→ pygi joined	20:07
mutex	hmmm I tried to send a patch to [email@hidden.address] but it doesn't show up on the list	20:07
	is there some sort of length cutoff ?	20:07
	maybe I need to split my patch into two different sets ?	20:08
	→ thorny_sun joined	20:11
mutex	patch in the email is 200 lines	20:11
thorny_sun	if i make a branch and then later merge it into master-- is it standard practice to then delete the branch-- is the commit history on the branch still accessible after it is deleted?	20:12
	→ sanity_ joined	20:13
sanity_	I've added a file but not committed it, how do I "un-add" it?	20:13
Mikachu	thorny_sun: yes, the branch points to the commits, not the other way around	20:13
	sanity_: git reset thefile	20:13
sanity_	Mikachu: thanks	20:13
Garoth	Oh, hey. What do you recommend for this situation:	20:14
thorny_sun	Mikachu: so when i delete the branch i'm not deleting those commits?	20:14
Mikachu	thorny_sun: correct	20:14
	← kalikiana left	20:14
t_	thanks guys. patch created and sent off to mailinglist and accepted :). bye.	20:14
thorny_sun	but then how do you know where those commits come from?	20:15
Garoth	I'm on my branch, which is a copy of master. I write some changes. Those changes get merged into origin/master as patches. I want to sync my branch against origin/master (with git pull, I suppose). I get conflicts, right? What do you suggest?	20:15
	← t_ left	20:15
Mikachu	thorny_sun: the merge commit itself points to all the commits it is made from	20:15
thorny_sun	ah	20:15
	Garoth: sounds like a case for rebase?	20:16
Garoth	thorny_sun: and just delete my commits?	20:16
	(and pull the new ones?)	20:16
Mikachu	Garoth: rebase will tell you which commits conflict and let you skip them	20:16
Garoth	Mikachu: so that's the best solution then?	20:17
Mikachu	or if they conflict cleanly, it will silently drop them for you	20:17
thorny_sun	sorry- i just exhausted my knowledge with that comment :)	20:17
Mikachu	Garoth: assuming nobody is tracking your branch	20:17
Garoth	Right...	20:17
Mikachu	or alternatively, you don't care about the people who do :)	20:17
	→ Aleph_One joined	20:17
Garoth	:)	20:17
	Mikachu: okay, but if you have push access to origin, and push the changes in your branch to master	20:18
	Then it does a merge	20:18
	And then git won't conflict on further pulls from master?	20:18
	→ _slacker_ joined	20:18
Garoth	*in your branch to origin/master	20:18
Mikachu	that was too confusing to follow	20:19
Garoth	*won't conflict on further pulls from origin/master	20:19
tango_	push only work if it's a fast forward	20:19
Garoth	Yeah, I'll tyr again.	20:19
	I'll try again.	20:19
	Okay, so if I have my branch Garoth, with some changes, and I want to merge those changes into origin/master the right way (assuming I have full access), what is the best way to do that?	20:19
Mikachu	if there are no other commits in origin/master, just push	20:20
	if there are, you can either merge them, or rebase on top of them before pushing	20:20
Garoth	Right, and that'll ff the origin/master	20:20
	Okay, so.	20:20
	And then let's say someone else makes some commits to master	20:20
Mikachu	after you pushed your branch?	20:20
Garoth	Yes.	20:21
	And then I want to update my local garoth branch to origin/master's changes	20:22
	Will just a pull be intelligent enough to merge stuff properly?	20:22
	→ Tv joined	20:22
Mikachu	that depend on how you created your branch, but if origin/master fully contains your branch, then a merge/pull will just do a fast-forward	20:22
Garoth	Oh, that's true.	20:23
Mikachu	ie, it will work if you did 'git branch garoth origin/master'	20:23
Garoth	Okay then!	20:23
Mikachu	or if you set it up manually afterwards	20:23
	otherwise, git pull theremote master should do it	20:23
	of course, none of this applies if you send the commits as emails	20:23
Garoth	Yeah, of course.	20:23
	→ jatt joined	20:24
Garoth	I can see why the kernel does topic branches rather than user branches though	20:24
	It's kind of cleaner.	20:24
	Mikachu: oh, one last thing then -- is there a way to push your branch to the remote server and have your branch be created there automatically, or do you always have to ssh in and manually make branches?	20:24
Mikachu	uh, just pushing branches should create them	20:25
	if it doesn't, you have a hook that disallows it	20:25
Garoth	Oh, I see.	20:25
	Hmm.	20:25
	Really?	20:25
Mikachu	or misusing push	20:25
Garoth	I usually get "doesn't match dest ref" or something	20:25
	(from memory)	20:25
Mikachu	when you do what?	20:25
Garoth	git push origin garoth	20:25
Mikachu	the proper way is 'git push theremote thebranch'	20:25
Garoth	^	20:25
Mikachu	well, that should work	20:25
Garoth	Hmm. Well, I'll look into it, thanks.	20:26
	I didn't think I had any hooks on.	20:26
Mikachu	maybe old git versions could play a part too	20:26
Garoth	It'd be very nice for my commiters to be able to make their own branches :)	20:26
	← Vortex34 left	20:27
	→ Vortex34 joined	20:27
jatt	it's possible to get a diff between my master repository and a remote repository but not pull the changes (git pull shows the diff, but applies the changes to my master repository).	20:27
	?	20:27
Mikachu	jatt: no	20:27
	jatt: er, yes, fetch instead of pull	20:27
	→ lucsky joined	20:28
lucsky	hello	20:28
	→ danten joined	20:29
	← Weaselweb left	20:32
	→ drizzd joined	20:35
lucsky	what kind of reason could there be for a [PATCH] message sent to the git list to be completely ignored ?	20:35
Mikachu	either it didn't make it to the list, or the contents were stupid, or it was just missed :)	20:35
lucsky	well if I did something wrong i'd rather know :)	20:35
Mikachu	can you see it in the archives?	20:35
lucsky	it seems that it made it to the list	20:35
	http://article.gmane.org/gmane.comp.version-control.git/97012	20:35
Mikachu	i think most people are nervous about touching git-svn	20:36
Pieter	lucsky: git-svn patches usually wait for an ack from eric, did you cc him?	20:36
Mikachu	you might want to try and CC it to eric	20:36
Pieter	ninja	20:36
lucsky	well there you go, I didn't know that	20:36
	thanks everyone	20:36
	now	20:37
	who's eric ? :)	20:37
Mikachu	lucsky: git log -- git-svn.perl	20:37
lucsky	so that would be Eric Wong right ?	20:37
Mikachu	indeed	20:37
Pieter	I fixed GitX to hilight trailing whitespace.. Hope that makes the world a little better	20:37
	← zachinglis left	20:38
lucsky	ok, I'll cc eric then, thanks again	20:38
Mikachu	possibly you need to cc the current git maintainer too	20:38
	not need, but it usually helps anyway :)	20:39
	← etgryphon left	20:39
lucsky	and that would be ? Shawn ?	20:39
Mikachu	Documentation/SubmittingPatches has lots of fun details	20:39
	yeah	20:39
Pieter	with normal git patches it could happen that it is applied without any notice	20:39
	→ jerbear joined	20:42
lucsky	Pieter: might be yeah, I'll ask anyway, it's just a two char patch but I'd rather make sure I don't reproduce possible errors in my next patches in any	20:42
	*in+if	20:42
Pieter	I also think your commit message is a bit odd	20:43
lucsky	eh, that is totally possible :)	20:43
	← thorny_sun left	20:43
	← sbeyer left	20:44
	← drizzd_ left	20:45
	← rwaite left	20:45
	→ tcurdt joined	20:46
	← spearce left	20:47
tcurdt	guys, I want to push via ssh and the directory is not yet existing on the server. Will I have to create a bare repo and manually scp it in place first?	20:48
	← jerbear left	20:49
Mikachu	tcurdt: yes	20:49
	tcurdt: or ssh in and git init	20:49
	← anvimal left	20:50
	→ anvimal joined	20:51
	cizra_ → cizra	20:51
	← Aleph_One left	20:51
	→ spearce joined	20:51
tcurdt	Mikachu: thx! ... IMO there should be an option on git-push that takes care of this	20:51
lucsky	ok, sent a note to eric and shawn, thanks Mikachu and Pieter	20:52
Mikachu	you can use gitosis, i think it does that for you	20:52
tcurdt	Mikachu: I know .. currently moving away from it though :)	20:53
	github for public repos, ssh for private ones	20:53
	← danten left	20:54
	← Weasel[DK] left	20:56
Mikachu	tcurdt: uh, i didn't write github anywhere :)	20:57
tcurdt	Mikachu: wdym?	20:58
	← cannonball left	21:01
	← bobmcw left	21:01
	→ robinr joined	21:05
	→ softdrink joined	21:07
	← lucsky left	21:07
	→ bronson joined	21:10
Mikachu	tcurdt: i said gitosis, you started talking about github, i was confused	21:12
	← sabayonlive-2236 left	21:12
tcurdt	aaaah :) ok	21:12
	← patmaddox left	21:16
	← kyrus1 left	21:19
	k-s → k-s[AWAY]	21:20
	← bronson left	21:20
	← tcurdt left	21:22
	→ JimmyB_ joined	21:24
	→ tjafk1 joined	21:26
	← lorandi left	21:27
	← BabelO left	21:31
	← LiamH left	21:34
	← kumbayo left	21:37
	← ph^ left	21:37
	→ ph^ joined	21:41
	← CelticSoul left	21:41
	← frsk left	21:42
	→ ciskje joined	21:42
	← icwiener left	21:43
	← tjafk2 left	21:44
	← anvimal left	21:47
	→ patmaddox joined	21:49
	← cizra left	21:53
	→ frsk joined	21:55
	→ ciskje_ joined	21:58
	← rmh3093 left	21:59
	← deavid left	22:01
	← ciskje_ left	22:02
	→ ciskje_ joined	22:02
	← charon left	22:03
	→ bronson joined	22:04
	← warreng left	22:08
	→ Beket joined	22:10
	← ciskje left	22:10
	← willb left	22:11
	← Beket left	22:14
	→ Beket joined	22:18
	→ Eridius_ joined	22:20
	← Eridius left	22:21
	← jesselucas left	22:22
	← Aikawa left	22:22
	→ digitalhobbit_ joined	22:23
	← Ryback_ left	22:24
	← boto left	22:24
	← tsb left	22:27
	← digitalhobbit left	22:28
	→ tsb joined	22:30
	→ beowulf joined	22:31
	← E78C8EE4BABA left	22:32
beowulf	up_the_irons: reading your post on gitosis, trying to install on dreamhost, any advice for when you can't edit PATH for non-login, non-interactive shells?	22:32
	emmanuel → emmanuel_brb	22:33
	→ rmh3093 joined	22:35
	→ Adam12 joined	22:36
	← d0k left	22:36
	← juanjoc left	22:37
	← stouset left	22:38
	← felipec left	22:40
	← EmilMedve left	22:41
	→ angelixd joined	22:43
	→ NewMonarch joined	22:46
NewMonarch	Hey, I accidentally committed w/o adding first. Is there a way to undo my most recent commit? I don't want to lose the changes, I just want to remove the commit from the repo so I can redo it.	22:47
frsk	given that you haven't pushed it anywhere; git reset HEAD^	22:47
NewMonarch	The biggest thing is that I renamed several files. I want those to go in as renames instead of delete/add.	22:47
frsk	That will remove the last commit	22:47
NewMonarch	frsk: Will it reset my working directory?	22:48
	(I don't want that.)	22:48
j_engelh	git add foobar; git commit --amend	22:49
frsk	NewMonarch: No, afaik it won't. I'd consider making a backup of your directory first, though. :-P	22:50
	NewMonarch: Haven't coded in some time, that's why I'm not 100% sure. Sorry. =)	22:50
NewMonarch	j_engelh: Seems cleaner. Work ok?	22:51
j_engelh	either works	22:51
	← softdrink left	22:52
robinr	ammend is pretty safe	22:53
	← priidu left	22:53
parasti	NewMonarch: git reset only resets the working tree with --hard	22:53
	← ph^ left	22:54
NewMonarch	Yep, amend worked great. Even converted deletions to renames.	22:54
Mikachu	it didn't convert anything, deletions and renames are detected when you generate the diff	22:55
robinr	the "renames" are only gits observation that something looked like a rename. Git didn't "store" a rename	22:55
	← drizzd left	22:56
NewMonarch	robinr: Ah. Well, I'm just about to merge w/ another branch and wanted it to "realize" I'd renamed stuff instead of trying to add those files back in.	22:56
	→ david_koontz joined	22:58
	← markkalderon left	22:58
	← JimmyB_ left	23:01
	← dkerschner left	23:02
	→ digitalhobbit joined	23:04
	→ qrush joined	23:06
	→ Millow joined	23:08
	not-xjjk → xjjk	23:08
	→ eternaleye joined	23:08
Millow	hello, I ned help, getting this error when trying to do "git push"	23:08
	error: hooks/update exited with error code 1	23:08
	← drewr left	23:10
	← digitalhobbit_ left	23:10
	→ kbingham joined	23:12
	→ thorny_sun joined	23:15
	← ciskje_ left	23:21
	← thorny_sun left	23:23
	← timcharper left	23:24
	→ daaku joined	23:24
	← gilimanjaro left	23:26
Millow	in git, is there liek "svn up"	23:27
	so I can do rm -rf and then "git update" to get a clean	23:28
	not checkout	23:28
	just update the current repo	23:28
	← univate left	23:28
	→ univate joined	23:30
	→ codeshot joined	23:30
	← NewMonarch left	23:30
	← sgrimm left	23:37
	emmanuel_brb → emmanuel	23:37
	← netoman left	23:41
	← _graham_ left	23:41
	← Beket left	23:42
	→ shodges joined	23:42
	→ capisce_ joined	23:44
	← whee left	23:44
up_the_irons	Millow: to just update the current branch, it's "git pull"	23:45
Millow	k	23:45
	← capisce left	23:45
	← not_benh left	23:46
	← patmaddox left	23:51
	← _slacker_ left	23:51
	→ stigrett joined	23:52
	← jpeeler left	23:52
	→ markkalderon joined	23:55
	← eternaleye left	23:58
	→ zachinglis joined	23:59

Logs Search ←Prev date Next date→

Channels Documentation