IRCloggy #git 2008-12-16

Provider of IRC logs since 2005.

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

	2008-12-16
	← __ash__ left	00:03
	→ jackdempsey joined	00:08
kevlarman	can someone help me figure out why git-filter-branch --env-filter won't change the author?	00:09
volante	anyone used any perl libraries for git?	00:10
johnw	i think you need to use commit-filter for that, don't you?	00:10
parasti	kevlarman: what's your script	00:10
kevlarman	essentially [ "$GIT_COMMIT_AUTHOR" = "A" ] && GIT_COMMIT_AUTHOR="B"; export GIT_COMMIT_AUTHOR	00:11
parasti	GIT_COMMIT_AUTHOR doesn't sound like a valid env var, see man git-commit-tree	00:12
Gitbot	parasti: the git-commit-tree manpage can be found at http://git.or.cz/man/git-commit-tree	00:12
	← scientes left	00:13
kevlarman	yeah i'm mixing my variables up	00:13
	i meant GIT_AUTHOR_NAME	00:13
schlurchz	kevlarman: might be helpful to paste your command line	00:14
	not here of course	00:14
	← joydivider left	00:14
Pieter	git	00:15
parasti	that's what everyone's thinking, isn't it.	00:15
Pieter	I don't know what they're thinking	00:16
	← k0001 left	00:17
kevlarman	http://pastebin.com/de3e7fe3	00:17
	← Thralas left	00:17
	← sping left	00:18
DrNick	you want something like git-filter-branch --env-filter "source $(pwd)/fixcommitter.sh" 1d4936d88f232c642b45fb913f2ac5c7bf0b9ad7..HEAD	00:19
	assuming that works	00:19
kevlarman	DrNick: it worked	00:22
	(why did it work?)	00:23
DrNick	filter-branch takes script fragments on the command line	00:23
	→ joydivider joined	00:27
	→ loincloth joined	00:27
Ilari	kevlarman: The reason original didn't work is that the script ran in subprocess of filter-branch, so it couldn't change environment variables in filter-branch process. In contrast, with source, it ran in the same process and thus could alter environment variables.	00:27
	← joydivider left	00:27
schlurchz	Ilari: the `hidden' subshell trap strikes again	00:28
	← ceej left	00:29
	← zerok left	00:29
	← branstrom left	00:30
Ilari	Hmm... Does the shell invoke subshell to execute the script or does the shell execute the file? :->	00:35
	← deskinm left	00:36
schlurchz	Ilari: well I guess it just executes the file, but it is a shell script, so it invokes a subshell :-)	00:37
	Ilari: but you're right, it was not the usual subshell trap.	00:38
Tv	if you tell a shell to run a command, it will check if it's an internal command (like "cd").. if not, it'll fork+exec it	00:39
	anything forked can't change parent process environment	00:39
	← Bass10 left	00:39
	→ deskinm joined	00:39
Ilari	At least bash can run stuff that clearly can't be execed (lacks shellbang; they run as bash shell scripts).	00:40
	→ bugfux joined	00:41
Tv	Ilari: that's not a bash specialty, i think	00:41
	→ therrg joined	00:41
Tv	Ilari: you know, original unix didn't have shebang..	00:42
bugfux	I have to define custom paths to get the git commands working on my server and then when I try to go git clone from a remote computer it is saying "git-upload-pack command not recognized"	00:42
	anyone know how to get git to recognize my .bash_profile file and load it	00:42
	or how can I tell git to know where the binaries are	00:43
Ilari	Trying to strace such things blows up real quick (ENOEXEC). Whereas stracing shellbang scripts works.	00:43
mugwump_	bugfux: did you go through the FAQ answer for that?	00:43
	← PloenK left	00:43
bugfux	nah I didnt know there was one, lemme check now	00:43
Ilari	bugfux: .bash_profile is not loaded on noninteractive login. You need to set up paths in .bashrc.	00:44
bugfux	ohhh bashrc cool thanks	00:44
Tv	Ilari: hmm yeah it seems it's special cased these days	00:45
bugfux	that worked, thanks guys	00:45
Tv	Ilari: but the special case handling is just fork+execing /bin/sh with the filename as arg	00:46
	→ bdiego joined	00:46
	← perezd88 left	00:46
	→ warthog19 joined	00:47
	← scrooloose_ left	00:48
	← warthog9 left	00:50
	→ bryanl joined	00:50
	← moccuo left	00:56
	← chris2 left	00:56
	→ bobmcw joined	00:59
	→ scrooloose_ joined	01:01
	← joevandyk left	01:01
	← mbroeker left	01:03
	← ebzzry left	01:04
	→ ebzzry joined	01:05
	← drewr left	01:07
	→ x_or joined	01:09
	← johnw left	01:13
	→ ashleyw joined	01:18
	→ Marmouset joined	01:18
	→ bobmcw_ joined	01:20
	→ pjhyett joined	01:20
	← therrg left	01:20
	→ branstrom joined	01:22
	← cgardner left	01:22
	→ ivanz joined	01:23
	→ stevenjenkins joined	01:24
	← david_koontz left	01:26
	→ moccuo joined	01:27
	→ k0001 joined	01:29
	→ h0nk joined	01:31
	← faithfulgeek left	01:32
	→ advany joined	01:33
	← loincloth left	01:35
	← bobmcw left	01:36
	← dragos-stoenica left	01:37
	→ madewokherd joined	01:39
	← CodeOfficer left	01:42
	← k0001 left	01:43
	← stevenjenkins left	01:43
	→ ben_h joined	01:44
	→ k0001 joined	01:47
	← schlurchz left	01:48
	→ cnk joined	01:48
	← pcapriotti_ left	01:49
	→ marvil07 joined	01:49
	← bbuss left	01:49
	→ CodeOfficer joined	01:55
	← carllerche left	01:55
	→ bjeanes joined	01:59
	zwu\|gone → zwu	02:00
	← bugfux left	02:00
	→ danopia_ joined	02:02
	← h0nk left	02:02
	→ faithfulgeek joined	02:04
	→ gretch joined	02:06
	← parasti left	02:08
	← RayMorgan left	02:09
	← bosie left	02:10
	→ ajonat joined	02:15
	→ loincloth joined	02:16
	danopia_ → danopia	02:17
	← loincloth left	02:19
	→ kbiddlecomb joined	02:23
	← alikins left	02:23
	→ clamothe joined	02:23
clamothe	I'm using git-svn. A branch I am working on got renamed.	02:24
	How can I point my local branch at the renamed branch, without losing any merge history?	02:24
	← bobmcw_ left	02:25
	← aziz left	02:29
	← __Zimm left	02:29
kbiddlecomb	sorry about the noob-like question, but I made some changes to my fork of a project. I tried to merge with upstream today, and got a conflict with one of the files..., i tried to do a "git stash", but got the error "fatal: git-write-tree: error building trees; the index is unmerged?"... what does this mean?	02:29
	→ Flathead joined	02:31
	→ nothingHappens_ joined	02:31
	→ Ori_B joined	02:32
kbiddlecomb	clamothe: can't you just move your local branch?	02:32
Ori_B	hi	02:32
	I'm having a problem with git rebase	02:33
	it tells me that it can't detach head	02:33
	→ webmat joined	02:35
	→ johnw joined	02:37
	→ snitko joined	02:37
	→ PloenK joined	02:41
	← nofxx left	02:41
	→ alikins joined	02:41
Ori_B	ah, figured it out. had an untracked file in my working directory that was named the same as a file I had in the repository	02:41
	← moccuo left	02:42
	← flaguy left	02:42
Ori_B	ah, nice. and new versions of git have the same change I put into the rebase script to get it to tell me that	02:43
	← Ori_B left	02:43
	→ jesselucas joined	02:44
	scrooloose_ → scrooloose	02:45
	← ahupp\|fb left	02:45
	→ zjason_ joined	02:47
	← advany left	02:47
	→ ahupp\|fb joined	02:48
	→ danopia_ joined	02:52
	← danopia left	02:52
	← foutrelis left	02:53
	→ RayMorgan joined	02:54
	→ nofxx joined	02:56
kbiddlecomb	can someone help me do a merge with my modified fork, and upstream?	02:56
	← snitko left	02:57
	← bryanl left	02:57
kbiddlecomb	i get "! [rejected] funtoo.org -> kbiddlecomb (non fast forward)"	02:57
jast	kbiddlecomb, please see faq non-ff	02:58
Gitbot	kbiddlecomb: Your push would lose changes on the remote. See http://git.or.cz/gitwiki/GitFaq#non-ff	02:58
	← webmat left	02:58
	← markelikalderon left	02:59
kbiddlecomb	jast, thats for trying to do a 'push'... i get that error trying to do a pull.	03:00
jast	what's the pull command you used?	03:01
	← zjason left	03:01
kbiddlecomb	git pull upstream funtoo.org:kbiddlecomb	03:01
jast	yeah	03:01
	that's how not to do it, generally	03:01
	what were you intending to do with this command?	03:02
kbiddlecomb	forgive me.... i'm brand new to git... i was trying to pull from the upstream/funtoo.org, and merge with my local branch of kbiddlecomb.	03:02
	(or even better, origin/kbiddlecomb)	03:02
jast	here's how to do that:	03:02
	1) checkout the target branch (kbiddlecomb here)	03:03
	2) git pull upstream funtoo.org	03:03
kbiddlecomb	did that.	03:03
	→ EmilMedve joined	03:03
jast	remember: pull simply runs fetch followed by merge	03:03
	and the foo:bar tells fetch what to do, not merge	03:04
kbiddlecomb	ah	03:04
jast	so what happens is that fetch attempts to get the stuff from the remote funtoo.org branch and write that into kbiddlecomb, which obviously doesn't work without merging	03:04
kbiddlecomb	fair enough... i need to spend some time learning the details of git over the holidays... obviously skimming the manpages / github guides wasnt enough :)	03:05
	now im getting "error: Entry '...' not uptodate. Cannot merge."... not up to date with what? the remote branch? I would think thats exactly when a merge would be needed :/	03:06
	← bentob0x left	03:06
jast	this message is a bit confusing, i'll freely admit that	03:07
	it means that merge cannot do its magic if there are uncommitted changes in one of the affected files	03:07
kbiddlecomb	hm... didnt think i touched this file though... so i guess i need a "git stash'? before the pull?	03:08
jast	you can use git diff <filename> to see what changed (git diff --cached <filename> to see what changed and got staged already)	03:08
kbiddlecomb	... not changed i made... i wonder if its from when I was trying random commands trying to get the merge to work :)	03:10
	← dramsay left	03:10
kbiddlecomb	s/not changed/not a change/	03:10
jast	so both of those commands give you no output?	03:10
	← ijcd_ left	03:10
kbiddlecomb	without --cached gave some changes, but they're not changed I made.	03:11
jast	well, at this point I don't think I can tell you where they came from	03:11
	→ scientes joined	03:11
jast	if you think you can live without them, do a git checkout HEAD -- filename to get rid of them	03:12
kbiddlecomb	shouldnt git stash work?	03:12
jast	otherwise figure something out. stash is a solution but it just delays the point at which you'll have to make the decision. ;)	03:12
	← johnw left	03:14
kbiddlecomb	hm... I just tried a "git checkout HEAD" to revert to origin (i don't care about any local changes), but now the pull is giving errors about untracked files that would be overwritten... how do I dump all local changes, including untracked files?	03:15
	zjason_ → zjason	03:15
jast	a checkout without filenames is different than one with	03:15
	→ ben_h_ joined	03:15
jast	to throw away all uncommitted changes, use git reset --hard	03:16
kbiddlecomb	jast: isnt a checkout without filenames a checkout of all files?	03:16
jast	there is a bit of a difference	03:16
	checkout without filenames switches to a different branch	03:16
	when you effectively tell it to switch to where you already are, it does nothing	03:16
kbiddlecomb	ok, makes sense.	03:17
jast	checkout with filenames gets the specified versions of those files, which may or may not be different from the current version	03:17
kbiddlecomb	"git reset --hard; git pull upstream funtoo.org" still gives errors about untracked files.	03:18
jast	I think it's a bit unfortunate that checkout and reset grew that way but that's life	03:18
	git clean removes all untracked files (this is, of course, a rather dangerous situation unless you know what you're doing)	03:18
kbiddlecomb	well, I know that the project was working last time I did a push, so i think it should be fine.	03:18
	(I pulled down from another computer, and it was working)	03:19
	← pastie left	03:19
jast	just saying :)	03:19
kbiddlecomb	jast: and I appriciate you making suggestions to keep me from doing something stupid :)	03:19
jast	here's a theory about what happened earlier, by the way	03:19
	→ pastie joined	03:20
jast	you probably did a merge of some kind and it conflicted, but instead of fixing it or cleanly aborting it you just sort of ignored it	03:20
	so the things you're cleaning up now may be the remains of an incomplete merge	03:20
kbiddlecomb	thats probably the case, since I got frustrated earlier and just started trying different commands :/	03:21
jast	for reference, you can abort a conflicted merge by simply using git reset --hard	03:21
	so I'm really tired and I guess I'll call it a day. good luck with the rest of your surgery.	03:22
kbiddlecomb	ok, i did a "git reset --hard; git clean -d -f; git pull upstream funtoo.org", and that seemed to work... now can I just do a simple "git push" to push to my origin?	03:22
	ok, well thanks for your help jast... night :)	03:22
jast	if there are any changes left that can be pushed, yes	03:22
kbiddlecomb	(I assume after a merge I don't need to do a commit before pushing)	03:23
jast	a clean merge automatically commits itself	03:23
kbiddlecomb	thats what I would have assumed.	03:23
jast	a conflicted merge needs a manual commit after fixing things	03:23
kbiddlecomb	well, thanks for all your help!	03:23
jast	you're welcome. see you around, perhaps.	03:23
kbiddlecomb	hopefully I'll be able to help noobs like me at some point :)	03:24
	← ben_h_ left	03:24
jast	oh, don't worry, this channel has a tendency of making regular lurkers super-smart by osmosis of wisdom	03:24
	it's how I am in a position to help now	03:24
kbiddlecomb	:)	03:25
lordpil	i think i've had reverse osmosis	03:27
	← scrooloose left	03:28
	→ scrooloose joined	03:29
	← aspotashev left	03:29
	jesselucas → lobolife	03:29
	← esden left	03:30
	→ ashleyw_ joined	03:30
	← ashleyw_ left	03:30
	← ahupp\|fb left	03:30
	→ threeve_ joined	03:31
	← lobolife left	03:31
	→ lobolife joined	03:32
	lobolife → jesselucas	03:32
	← jesselucas left	03:32
	→ cgardner joined	03:34
	→ PackageLost joined	03:36
	kbiddlecomb → kbiddlecomb\|away	03:38
	→ ben_h_ joined	03:40
elliottcable	Hey y'all	03:40
	I hear a lot about git's much-lauded cheap local branching	03:40
	and some people claim to develop every bugfix and minor feature and everything in it's own branch	03:40
	my problem is that, when I try something like that	03:40
	I end up having nearly as many "merging" commits in the resulting commit list as "meaningful" commits.	03:41
	← ben_h left	03:41
elliottcable	Is this considered acceptable, or is there a way to do this whole "branch-heavy" workflow without generating dozens of "Merging feature-foo into master" commits?	03:41
	ben_h_ → ben_h	03:43
	← ashleyw left	03:46
Toad	elliottcable: rebase	03:47
elliottcable	Toad: example of use in this setting?	03:47
	← ajonat left	03:47
Toad	elliottcable: don't merge your bug fixes in, just rebase them on top of master	03:48
	then the merge becomes fast-forward and you don't have an ugly merge commit	03:48
	our workflow is rebase for everything except for large changes, and merge for large ones to preserve history	03:49
	kbiddlecomb\|away → kbiddlecomb	03:51
	→ daxelrod joined	03:52
	← edwardam left	03:52
	← daxelrod left	03:52
	→ drewr joined	03:54
	→ edwardam joined	03:54
	← edwardam left	03:55
	→ edwarda__ joined	03:55
	← clamothe left	03:55
	→ elmex_ joined	04:02
	← ibuffy left	04:02
	→ scientes_ joined	04:03
	→ ibuffy joined	04:05
	← scientes left	04:07
	← drewr left	04:08
	→ daxelrod joined	04:08
	← daxelrod left	04:09
jrockway	i noticed github added a feature that displays all commits that are not merged into the current branch	04:09
	is there a way to do that from the command line without naming each remote?	04:09
	→ daxelrod joined	04:10
	→ ijcd_ joined	04:10
	← daxelrod left	04:11
	← ijcd_ left	04:14
schacon	not that i know of - you have to do something like 'git rev-list remote1/master remote2/master ^HEAD'	04:14
jrockway	yeah	04:15
schacon	you can't do "refs/remotes/* ^HEAD"	04:15
jrockway	i guess i will hack something together then	04:15
	it seems like a useful feature	04:15
	but i don't use github :	04:15
	s/:/:)/	04:15
	→ vbabiy joined	04:17
	← elmex left	04:17
	→ esden joined	04:17
	elmex_ → elmex	04:17
	← esden left	04:18
	← rtomayko left	04:19
	→ shweazel joined	04:23
	← harinath left	04:23
	← k0001 left	04:25
	→ clamothe joined	04:26
schacon	jrockway: you really, really should :)	04:26
	← kukks left	04:26
	← jelmer left	04:28
	← ibuffy left	04:28
	← alexyk left	04:28
	← supercali_ left	04:30
	→ ijcd joined	04:32
	← cartucho left	04:32
	← hobodave left	04:35
	→ markelikalderon joined	04:37
	← markelikalderon left	04:38
	→ pygi joined	04:39
	← ben_h left	04:42
	← pygi left	04:43
	→ alexyk joined	04:43
	← stephens left	04:43
	→ carllerche joined	04:43
	→ ahupp\|fb joined	04:44
	→ girishr joined	04:45
	→ ankit9 joined	04:46
	→ pygi joined	04:47
	→ jony joined	04:49
	→ j416 joined	04:50
	→ markelikalderon joined	04:50
	← douthat left	04:51
	→ Mario__ joined	04:53
	← pygi left	04:53
jrockway	ok	04:53
	Mario__ → pygi	04:53
	← schacon left	04:54
	← dreiss left	04:55
	→ carllerche_ joined	04:57
	← carllerche left	04:57
	→ psankar joined	04:57
	← vbabiy left	04:59
	← marvil07 left	05:00
	← brett_h left	05:00
	← bjeanes left	05:00
	→ awarde joined	05:02
	→ bugfux joined	05:02
	→ carllerche joined	05:03
	→ OlofS_ joined	05:04
	← k776 left	05:05
	→ ceej joined	05:09
	→ tjafk1 joined	05:10
	→ girishr_ joined	05:11
	← tjafk left	05:11
	← threeve_ left	05:12
	← clamothe left	05:15
	→ hobodave joined	05:18
	→ ben_h joined	05:18
	← carllerche_ left	05:19
	← qrush left	05:20
	→ clamothe joined	05:21
	→ aspotashev joined	05:25
	→ agenteo joined	05:26
	→ douthat joined	05:28
	← PloenK left	05:28
	← girishr left	05:28
	← scientes_ left	05:29
	→ scientes joined	05:30
	← huangjs left	05:34
	← douthat left	05:45
	← jony left	05:47
	→ jony joined	05:47
	← moconnor left	05:48
	→ moconnor joined	05:50
	← clamothe left	05:51
	← aspotashev left	05:55
	← kbiddlecomb left	05:56
	→ clamothe joined	05:57
	→ rtomayko joined	05:59
scientes	how do you commit to your last commit again?	06:02
	ie redo it	06:02
doener	--amend	06:02
	← markelikalderon left	06:03
	← nothingHappens_ left	06:05
scientes	how do i reset 1 file?	06:05
	← oudeis left	06:06
	→ svend_ joined	06:06
doener	git reset -- file # That resets the file in the index	06:07
	j_engelh resets doener	06:07
	← resmo left	06:07
doener	git checkout HEAD -- file # get "file" from HEAD, into the index and working tree	06:07
	← bugfux left	06:08
	← carllerche left	06:10
	→ reithi joined	06:11
	→ carllerche joined	06:11
	← OlofS_ left	06:13
	→ svend__ joined	06:17
	→ OlofS joined	06:18
	← pjhyett left	06:18
	→ stuffcorpse joined	06:19
	← stuffcorpse left	06:20
	← svend left	06:20
	← bdiego left	06:21
	← alexyk left	06:21
	← svend_ left	06:21
	← altrux left	06:22
	psankar → psankar_brb	06:31
	← carllerche left	06:31
	← psankar_brb left	06:32
	brain_in_vat → biv_afk	06:37
	→ peritus_ joined	06:38
	→ sammcd joined	06:41
	← jony left	06:43
	→ jony joined	06:45
	← sammcd left	06:46
Loof	Anyone here know why git seems to never call must custom merge 'driver'?	06:48
doener	fast-forward case? Bad setup?	06:49
Loof	Hmm, I tested with the 'binary' driver... and got VERY close to what I want	06:50
doener	well, the binary driver is just like marking the file as binary (or having it auto-detected). Don't do anything but mark as unmerged.	06:51
Loof	Well, it pulls over the branch version rather than trying to merge the two.	06:52
	I'm still getting the same number of merges that 'fail to merge' but I'm not entirely clear why... nor do I see my script getting called (when I add logging)	06:53
	→ pjhyett joined	06:53
doener	your script? I thought you were using the "binary" driver? confused	06:53
Loof	I've got this in my ~/.gitconfig:	06:53
	[merge "branch"] name = Aggressively choose the branch driver = /root/git_hackathon/aggressive_driver %B %A	06:53
	Ok, 1st I tried just 'binary'...so I know that it is using my attributes	06:54
	because that changed how the merge functioned	06:54
	← RayMorgan left	06:54
Loof	I also tried: driver = cat %B > %A	06:55
doener	ok, and in your .gitattributes?	06:55
Loof	* merge=branch	06:55
	→ sammcd joined	06:56
	← sammcd left	06:56
Loof	Does that look about right?	06:57
doener	yep, and using the "cat %B > %A" works just fine for me	06:59
Loof	Hmm	06:59
	← plediii left	06:59
Loof	I don't get it	06:59
	I added a log entry to the script.. just to be sure it's being called...and I'm not seeing it being called, ever	07:00
	git merge -s recursive --squash 50	07:00
	error: Entry 'src/WINNT/eventlog/lang/de_DE/event.mc' not uptodate. Cannot merge.	07:00
	← robinr left	07:00
Loof	I'm wondering if that error is something other than merge returning an error	07:01
	(or, well, merge returning an 'conflict' exit code)	07:01
doener	not up to date means that you have uncommitted changes	07:01
	and the "-s recursive" might of course override your attributes	07:02
Loof	Well, I don't have uncomitted changes before the merge	07:02
doener	(not sure about that really overriding the attributes, but it would make sense to me)	07:02
	→ plediii joined	07:02
	← OlofS left	07:02
	→ zcopley_ joined	07:02
	← zcopley_ left	07:04
	→ OlofS joined	07:04
Loof	ok, I started over from a 'working' branch	07:04
	and it IS calling the driver when the merge doesn't fail	07:04
	→ resmo joined	07:04
	← doener left	07:06
Loof	But I'm still getting oddball failure cases	07:06
	→ ceej_ joined	07:07
Loof	Hmm, since doener left... I wonder if anyone else can help (or is even awake)	07:07
	→ namenlos joined	07:08
	→ doener joined	07:10
Loof	doener: For some reason when I redid the merges from scratch...everything seems to work well	07:11
	← peritus_ left	07:14
	→ quoin joined	07:17
	→ posixJ1 joined	07:19
	→ alexyk joined	07:19
posixJ1	When I test a applying patch I wrote on a copy of an original tree I get "patch1.txt:34: trailing whitespace"	07:20
	what shold I do to safely remove it ?	07:20
	→ oudeis joined	07:20
posixJ1	is there some script / command for it ?	07:20
doener	--whitespace=fix	07:20
	← bobesponja left	07:21
posixJ1	when preparing the patch you mean ?	07:21
doener	when using "git apply"	07:21
posixJ1	can I take a written patch and remove from it the white space ?	07:22
	by some command I mean ?	07:22
doener	dunno... I'd just apply it and then create a new one using format-patch again	07:22
	← pygi left	07:23
	← nice_burger left	07:23
posixJ1	doener, thnks	07:23
	and while preparing a patch - does the --whitespace=fix also holds	07:24
	?	07:24
doener	I don't see such an option in the format-patch man page.	07:25
	← ceej left	07:25
	→ priidu joined	07:26
	dwmw2_yoyo → dwmw2	07:26
	← nighthwk1 left	07:27
	ceej_ → ceej	07:27
	← OlofS left	07:28
	← madewokherd left	07:30
tokkee	http://www.google.com/trends?q=git%2C+subversion Git wins! scnr	07:37
	→ reel_ joined	07:41
	→ reel__ joined	07:41
	← reel__ left	07:41
	→ not-xjjk joined	07:46
	→ bentob0x joined	07:47
	→ quoin- joined	07:49
	→ pygi joined	07:51
	← PackageLost left	07:52
	← alexyk left	07:55
	← xjjk left	07:57
resmo	tokkee: hmm almost http://www.google.com/trends?q=git%2C+svn	07:57
tokkee	resmo: svn doesn't count ;-)	07:57
resmo	tokkee: hrhr	07:58
	← moconnor left	07:59
tokkee	Hrm ... "git" seems to be some turkisch word judging from the _huge_ majority of request that came from Turkey.	07:59
doener	haha, subversion gets hits like "A Chinese Internet dissident has been formally arrested and charged with subversion."	07:59
	← pjhyett left	07:59
tokkee	s/sch/sh/	07:59
doener	well, ok, the topic is not that funny :-/ dammit :-(	07:59
	→ moconnor joined	08:01
	→ mcella joined	08:01
	← posixJ1 left	08:02
	→ bjeanes joined	08:03
	→ markwaters1 joined	08:04
	← statim left	08:06
	← quoin left	08:07
	→ statim joined	08:07
	→ ffgeek200 joined	08:07
	← agenteo left	08:09
	→ gitster joined	08:13
bentob0x	how do you pull a remote branch in to a new local branch?	08:14
gitster	git fetch origin that-branch:my-branch && git checkout my-branch	08:14
	or if you already have that remote defined, just "git checkout -b my-branch origin/that-branch"	08:15
	← ben_h left	08:15
	← jmd left	08:15
gitster	Hmm, the /topic says 1.6.0.4 but that number does not mean much to me...	08:16
	→ naeu joined	08:16
	→ ph^_ joined	08:16
	← bjeanes left	08:16
	→ Weaselweb joined	08:18
	→ charon joined	08:22
	→ GarethAdams_ joined	08:24
jdq	gitster: that number gives you the current stable version... but its slightly outdated, as 1.6.0.5 is the current version I think	08:25
	→ bjeanes joined	08:25
	← bentob0x left	08:26
doener	jdq: I'm pretty sure he wanted to express something similar to that last part. After all, he is the one that makes the releases ;-)	08:26
	→ lucsky joined	08:26
	→ plediii_ joined	08:27
lucsky	'morning	08:27
	→ psankar joined	08:27
jdq	doener: ok, my bad ;)	08:27
	← oudeis left	08:27
	← psankar left	08:28
	← jackdempsey left	08:29
	→ nighthwk1 joined	08:29
	← spool left	08:30
	→ advany joined	08:33
	→ zerok joined	08:33
	→ scudco joined	08:33
	← advany left	08:34
scudco	if i've got a git patch file generated by git format-patch then when i apply it using git apply is is supposed to automatically add the commits to my branch or do i then need to add the files manually and commit locally?	08:34
	← nighthwk1 left	08:35
doener	apply only applies the patch. You should be using "git am"	08:35
charon	scudco: git-apply only puts it in the index (iirc). use git-am to "import" the commits from format-patch	08:35
	damn, doener wins.	08:35
scudco	charon, doener: thanks!	08:36
	→ jmesnil joined	08:36
scudco	perfect! that's great	08:37
	→ hurikhan\|Work joined	08:38
	← plediii left	08:42
ash_	anyone know if there's any scripts available which can check a bunch of git repositories and report on untracked changes?	08:43
	i want it to watch about 20 git repositories, and report nightly, if there are any untracked changes it emails a report with the patches	08:43
	→ OlofS joined	08:46
wereHamster	ash_: is maybe `git status` what you're looking for?	08:46
	→ harinath joined	08:46
	← OlofS left	08:46
wereHamster	or what do you mean by 'untracked changes'?	08:47
Ilari	ash_: At least with one repo, you can do it by looking at 'git diff HEAD' output (is it empty or nor). Now, run this for multiple repositories...	08:47
ash_	it is, but rather than emailing a big list of git-status output, i want it to only report on repos with uncommited changes and send the full diff	08:47
	i just need to script it up - just was wondering if there was a tool out there already	08:47
	on our production servers we do a git clone when deploying web apps.. there's been occasions where some php apps get exploited and someone will upload some malicious code or modify a file	08:48
	so i want to use it for security, and also to remind programmers if they've made changes that they didnt commit	08:49
	→ solydzajs joined	08:49
Ilari	ash_: So you also need list of files that aren't tracked...	08:49
wereHamster	ash_: git status; if [ "$?" = "0" ]; then git diff HEAD \| sendmail; done	08:49
ash_	not just new files, but any untracked changes	08:49
	any working dir modifications	08:50
wereHamster	this however will only work with git from git.git as the git-status exit value was fixed recently.	08:50
	→ peritus_ joined	08:50
Ilari	wereHamster: That's not enough. Firstly, 'status -a' is better here (catches unstaged modifications). And you need list of untracked files too.	08:51
wereHamster	maybe some magic with the output of 'git status' or 'git clean' to add the files temporarily and then make the diff	08:52
Ilari	ash_: Here's one: 'rm .git/modr.index.tmp ; export=GIT_INDEX_FILE=.git/modr.index.tmp ; git add -f . ; currentstate=`git write-tree` ; oldstate=`git rev-parse HEAD^{tree}'	08:53
ash_	it would also be useful when we give out ftp access to someone's webroot. they can write to the web root but not to the .git dir. so whenever they upload changes, a patch will be emailed out and so i can review it and commit	08:53
Ilari	ash_: (oops, it's missing one '`'. If currentstate and oldstate wind up different, then mail a git diff between them.	08:54
	ash_: That works by first doing force-add (override ignores) of everything, then writing it out and comparing tree object IDs.	08:56
	ash_: And it uses new temporary index, instead of the main index, so it won't interfere with normal repo work.	08:56
	ash_: Also, you could want to nuke .git/modr.index.tmp after that write-tree.	08:57
	← hobodave left	08:57
ash_	interesting, thanks i'll play around with that	08:58
	→ gregeh joined	08:58
Ilari	ash_: If the repo has no local uncommitted modifications, then that sequence won't even write anything into object database.	08:59
	← ahupp\|fb left	09:02
Ilari	ash_: However, with that sequence often encountering uncommitted changes, it creates some number of tree objects (one of which is dangling) per run, which adds up.	09:02
	→ Acryler joined	09:02
	→ pjhyett joined	09:08
	← Acry left	09:08
	← zumbrunn left	09:09
ash_	i think if i get the latest git with the git-status exit code working, that could make this pretty simple to do	09:10
	zwu → zwu\|gone	09:10
	← peritus_ left	09:12
	← moconnor left	09:13
Ilari	ash_: IIRC, the exit code won't cover untracked files...	09:14
wereHamster	ash_: I think Ilaris solution is much better ;)	09:14
	→ mbroeker joined	09:16
	← solydzajs left	09:18
	← mbroeker left	09:19
	→ mbroeker joined	09:20
	→ sgrimm joined	09:20
	→ davidw joined	09:20
	→ shri joined	09:21
davidw	hi - do you guys know how to get git to make patches without the a/ b/ stuff in the pathnames? Apparently my colleague's tortoise subversion thing is choking on those	09:21
Weaselweb	can't you specify a -p 1 on tortoise svn?	09:22
shri	i get the fatal error (fatal: unable to create '.git/index.lock': File exists) when i do git pull	09:22
	what do i do?	09:22
davidw	Weaselweb, not sure, she's not that much of an expert with it - it'd probably be easier to edit my patch by hand if needs be	09:23
Weaselweb	davidw: well, IMHO any patch program should provide a -p option	09:24
davidw	Weaselweb, agreed, but since she's remote, I can't even go over to her computer and fiddle around to see if I can get it to work	09:24
	→ Darkhost joined	09:25
Weaselweb	mh, tortoise... so, i'm expecting you can't use nx, just rdp	09:25
Darkhost	http://imperiumreklamy.com/pages/ptp.php?refid=krzysztof9222 Look at:) <lol>	09:25
Weaselweb	davidw: check tortise svn website, if there is such an option	09:25
	← Darkhost left	09:25
ash_	Ilari: would it work if i put the temporary index file in /tmp and gave the script no write access to the git repo?	09:26
	→ oudeis joined	09:27
	→ solydzajs joined	09:28
	← bjeanes left	09:29
Ilari	ash_: It would need write access to object database (create new objects) if the trees differ...	09:29
davidw	Weaselweb, damn, it wasn't even the levels... : https://kerneltrap.org/mailarchive/git/2008/1/15/571079/thread	09:29
	← nikolasco left	09:30
Ilari	shri: You have some git process running in some other shell or gotten stopped for some reason?	09:31
Weaselweb	davidw: oh, that's too bad	09:39
davidw	yeah... have to look and see if anyone did anything	09:39
	→ zumbrunn joined	09:42
davidw	doesn't seem like it, although I'd have to check out git itself to find it	09:43
	← ph^_ left	09:46
	← segher left	09:48
	→ segher joined	09:48
Ilari	ash_: Additionally, AFAIK, git diff won't include new files unless you make a tree (or something) out of them (which needs ODB write access).	09:48
	← pygi left	09:49
Ilari	ash_: And to reliably check repo, you need to inpersonate the main repo user anyway (which also impiles needed write privs)...	09:50
	→ pcapriotti joined	09:50
	← scudco left	09:52
	← bronson left	09:55
	← mbroeker left	09:59
	→ ph^_ joined	10:04
	← ceej left	10:07
	→ solydzajs_ joined	10:12
	← solydzajs left	10:14
	← univate left	10:15
	← tango_ left	10:16
	→ univate joined	10:16
	→ mithro joined	10:19
j416	hi guys	10:24
	← CodeOfficer left	10:25
j416	I just started exploring git today, so I am a total newbie. For testing, I cloned the git repos to my computer, added some files, messed around. Now if I want to revert all changes to the latest commit in the remote repos, how would I do that?	10:25
Ilari	j416: 'git reset --hard origin/master'?	10:26
j416	Ilari: thanks, it works great (I think). Files not added to my index are not in any way affected by git actions right?	10:27
	(I noticed my unadded files were left as they are)	10:28
	→ cwillisf joined	10:28
	← ia left	10:28
cwillisf	hello... if I have a git repository created with "git --bare svn init", is there a way I can make it "forget" its connection to the SVN repository? I heard that there's a way to do it with clone, but I can't seem to find instructions	10:29
	→ parasti joined	10:29
	→ ia joined	10:29
Weaselweb	j416: git reset just restores the files included in the repo, as unadded file don't belong to the repo	10:30
	→ schlurchz joined	10:30
j416	Weaselweb: thank you! So - in practice - reset does exactly what clone does, but without the need to download the whole repos again?	10:31
	(removing my directory and running clone, is the same as running a hard reset?)	10:31
Weaselweb	j416: no, reset set your HEAD to the given commit. clone is just for import a git repo, later you will be using git fetch and/or git pull	10:32
j416	hm	10:32
	thank you. I will have to read some more.	10:33
	→ peritus_ joined	10:34
	→ markelikalderon joined	10:36
	← gretch left	10:37
	→ jds joined	10:38
	→ tango_ joined	10:42
	← dwmw2 left	10:42
	← cwillisf left	10:42
	→ advany joined	10:43
	→ quoin joined	10:47
	← alikins left	10:49
	→ macgerhard joined	10:50
	← Eridius left	10:51
	→ nbrosnahan joined	10:52
	→ pygi joined	10:53
	← nbrosnahan left	10:53
	← naeu left	10:57
	→ Eridius joined	10:58
	→ Eridius_ joined	11:00
	← Eridius left	11:00
	Eridius_ → Eridius	11:00
	← quoin- left	11:03
	← mtkd left	11:04
	→ mtkd joined	11:06
	← iulian left	11:07
	← Marmouset left	11:08
Ilari	j416: Reset is used to alter "branch pointer" of current branch. It has four modes: --soft (leave index and working tree alone), --mixed (default; reset index but leave working tree alone), --hard (reset both index and working tree) and --merge (very recent addition; reset index and merge changes into working tree).	11:10
	→ xjjk joined	11:10
	→ grahal joined	11:11
	← peritus_ left	11:11
	→ peritus_ joined	11:12
Ilari	j416: The mirror of remote repository 'master' that git keeps in 'origin/master' (really 'refs/remotes/origin/master') can be updated by either 'git fetch origin' or 'git remote update'.	11:15
	→ nbrosnahan joined	11:15
	→ Voker57 joined	11:16
	← JackTheRabbit left	11:18
	→ bbuss joined	11:18
	← gregeh left	11:20
Ilari	j416: Also, 'git reset' won't touch other branches besides the currently checked out one.	11:21
	j416: And if the local mirror of remote branch is out of date, then resetting to that state will result reset to mirrored state, not the state remote repository currently is in. Git reset does not do network traffic.	11:23
	→ sohum joined	11:23
	← not-xjjk left	11:23
	← mithro left	11:23
	→ guille_ joined	11:23
guille_	hi	11:23
	→ lorandi joined	11:23
guille_	how do would you find commits without branch?	11:24
Ilari	j416: Pretty much only high-level commands that do network traffic are 'clone', 'fetch', 'push', 'pull' and 'ls-remote'.	11:24
sohum	what precisely does git branch --track new_branch master do, assuming master is a local branch?	11:24
Ilari	sohum: Marks 'new_branch' having local branch 'master' as default pull source...	11:24
sohum	hmmm.	11:25
Ilari	guille_: Not sure if this works: First use 'git fsck --full' to find all 'dangling commit's and 'dangling tag's. Then 'git log <ids of dangling commits and tags> --not --all'.	11:26
sohum	ok, so here's the problem. I work mostly by myself, but occasionally I'll use git to back things up. I want to use branches to separate the different bits of work I'm doing, as well as maintaining a sort of local environment in master. but I don't want to have to constantly keep checking out the master branch to work on said environment, nor having to keep merging it into the subproject branches.	11:28
	Am I barking up the wrong tree here, or is there a way to do this?	11:28
	→ statim_ joined	11:30
	→ boto joined	11:30
	← nofxx left	11:32
guille_	hm. i have a bunch of dangling blobs and trees	11:33
	not sure what's the difference between dangling blobs, trees, and commits	11:34
	← univate left	11:34
guille_	well, commits i suppose they're commits not attached; but the rest?	11:35
	→ univate joined	11:35
	← statim left	11:37
macgerhard	when I do a git status, I get this:	11:38
	error: Could not read 40e42a504d10d78170bb89012b5a38fa3554f807	11:38
	error: unable to read tree object HEAD	11:38
	← jony left	11:38
macgerhard	fixing the permissions and chowning all files didn't help	11:38
sohum	guille_: a blob is actual data that's stored	11:41
	guille_: a tree I'm not so sure	11:41
	→ drewr joined	11:42
guille_	so in order to fix my dangling things, what should i do? sohum and Ilari	11:43
sohum	guille_: dangling stuff don't hurt, they just take up extra space, and could potentially be useful for emergency data salvage	11:45
	→ Sho_ joined	11:47
cbreak	you can do git fsck and git gc. but I would only do the latter if everything else works correctly.	11:48
guille_	oh ok thanks :) what i need to do now is get the dangling commit and merge it with the correct branch, isn't it?	11:48
	← quoin left	11:48
Ilari	macgerhard: 'git fsck --full'.	11:49
	→ bryanl joined	11:49
	← oudeis left	11:49
Ilari	guille_: Also maybe see 'git reflog' output...	11:49
shri	i get the fatal error (fatal: unable to create '.git/index.lock': File exists) when i do git pull	11:55
	i dont have any background processes running. may be i might have stopped old processs	11:55
	is there any way i can recover my rep?	11:55
	← HttpErrors left	11:55
Pieter	rm .git/index.lock	11:55
	→ gilimanjaro joined	11:56
	→ dwmw2 joined	11:56
	→ netoman joined	11:57
	→ Ademan joined	12:00
	← ke4qqq left	12:01
j416	Ilari: sorry I was away getting breakfast	12:01
	thank you for your explanation!	12:02
guille_	Ilari: reflog shown me the true! hehe thanks	12:02
	→ ffgeek200b joined	12:02
shri	what does git reflog do?	12:02
	← ffgeek200 left	12:02
guille_	man git-reflog but it basically shows you your recent changes in the repo (i think they have an expire time)	12:03
Gitbot	guille_: the git-reflog manpage can be found at http://git.or.cz/man/git-reflog	12:03
shri	Pieter: i tried removing the lock. when i do git pull again. i get the same error. i have no other git processes running	12:05
	→ cmarcelo joined	12:05
	→ mithro joined	12:05
guille_	shri: i don't remember the name, but there's a command to show which processes holds a descriptor of a file	12:06
shri	i get the error /usr/bin/git-merge: line 278: 31369 Bus error git-update-index --refresh 2>/dev/null fatal: unable to create '.git/index.lock': File exists	12:06
Pieter	shri: what OS are you on?	12:07
shri	linux	12:08
Pieter	what version of git?	12:08
shri	git version 1.5.0.6	12:09
sohum	may be irrelevant, but lsof <path> will show you what processes are using it	12:09
shri	it used to work well. its just that this rep is giving me issues	12:09
sohum	sorry, lsof \| grep <part of path>	12:10
	← charon left	12:10
	→ dave_levin joined	12:12
	→ Galaxor joined	12:12
Galaxor	Hi. I'm trapped. I did git svn rebase. There was some sort of problem, and it's asking me to fix "the problem", and suggests git rebase --continue, --skip, etc. But git status does not show me what files need fixing.	12:13
	Usually, that's how I figure out what to do.	12:13
shri	what could be wrong?	12:15
	← LotR left	12:15
	→ intripoon joined	12:15
Galaxor	Well, it's a merge, so it looks like some file or other doesn't merge cleanly.	12:16
intripoon	hi	12:16
	how do I undo a commit?	12:16
Galaxor	Usually, git status will show me which files didn't merge cleanly. I have to fix them and then say "git add".	12:16
shri	is there any tool which adds files on its own and commits?	12:18
	→ LotR joined	12:18
shri	i want to commit some 100 patches. each patch adds some new files. is there any tool which can do this automatically	12:18
	→ bosie joined	12:19
	← guille_ left	12:19
	→ ben_h joined	12:21
parasti	intripoon: if you haven't pushed and it's the last commit, use git reset. if you haven't pushed and it's not the last commit, use git rebase -i. otherwise use git revert. see the manpages for each	12:21
	← gambler__ left	12:21
	→ esden joined	12:23
bsod	yey uploading to github:)	12:24
	git rocks	12:24
resmo	bsod: we all know :)	12:24
bsod	heh :)	12:25
	→ Ryback_ joined	12:25
	→ Varon joined	12:25
	← shri left	12:26
	→ cartucho joined	12:26
macgerhard	Ilari: ok, I've done the fsck, it spotted a broken link. what do I do next? git gc?	12:26
	← Galaxor left	12:27
Varon	hello. i have make a big mistake in a subdirectory. how can i clone/checkout only this directory again?	12:28
intripoon	is git reset atomic as well?	12:29
	← esden left	12:29
intripoon	so when I git reset and ctrl-c, everything is like nothing has happened?	12:30
Varon	i think no ^^	12:30
	but dont know	12:30
	431 idling users -.-	12:33
parasti	Varon: if you didn't commit, you can just check it out again	12:33
	← bryanl left	12:34
Varon	how to checkout only a subdirectory in the whole git ?	12:34
timonator	i think checkout can do that, lemme check the manpage for you...	12:34
parasti	intripoon: control+c? why do you think you need that?	12:35
timonator	git checkout [<tree-ish>] [--] <paths>... - that sounds helpful	12:35
intripoon	I didn't	12:35
	but I typed in git reset to see what happenes	12:35
	because I thought, it'll tell me about some missing parameteres I might have to add	12:35
	but I guess it just started resetting	12:35
Varon	in manpage is git checkout hello.c ... but hello.c is just a file	12:35
intripoon	one commit	12:35
Varon	ok. thnx. i will try	12:36
intripoon	then I ctrl-c-ed it	12:36
parasti	intripoon: a plain git reset probably doesn't do anything	12:36
intripoon	because I don't want it	12:36
	→ oudeis joined	12:36
	← joshdavey left	12:36
parasti	intripoon: so I understand it's your last commit?	12:36
	intripoon: in that case git reset --hard HEAD^	12:36
	intripoon: btw, by running random commands without consulting a manpage you might end up shooting yourself in the foot :P	12:37
timonator	in the foot or in other places...	12:37
Varon	^^	12:37
intripoon	I think --mixed is what i need	12:38
	← pjhyett left	12:38
parasti	intripoon: the options might be different, yeah	12:38
intripoon	it shouldn't be so dangerous	12:38
Varon	just a shoot through a toe ^^	12:39
parasti	intripoon: the deal with --mixed and --soft is that it leaves your index/working tree in a "non-intuitive" state	12:39
intripoon	the thing is	12:39
	a friend of mine	12:40
	accidentally commits files to his repository that don't belong there	12:40
	and are hugh	12:40
	huge	12:40
	so he wants to keep his working tree like it is now	12:40
	but just undo its last commit	12:40
	I think that's what --mixed does	12:41
	no?	12:41
	then he could commit again without those huge files	12:41
parasti	intripoon: yeah, that sounds like the best approach	12:41
macgerhard	ok, so the real problem is fatal: bad tree object	12:41
parasti	intripoon: --hard would get rid of those files, so that's not what you want	12:41
macgerhard	I've tried doing a git repack -A -d, still no luck :(	12:41
	→ madsdyd_work joined	12:42
parasti	intripoon: get rid of them in the working tree, I mean	12:42
	→ aziz joined	12:42
intripoon	ok	12:42
	→ naeu joined	12:43
Varon	okay. git checkout ./* works for me fine.. if i am in the right directory	12:43
parasti	Varon: why not git checkout $dir?	12:46
	← dave_levin left	12:47
	← Flathead left	12:50
	← jmspeex_ left	12:51
	→ rolfb joined	12:53
macgerhard	how can i rewrite the tree so that it wouldn't reference a missing object?	12:53
	→ bobmcw joined	12:54
	→ nud joined	12:57
j416	If I do a reset --hard, will then all commits after the commit i am resetting to be lost forever?	12:58
timonator	no	12:58
cbreak	they will be gone out from history	12:58
timonator	commits that have nothing point at them are kept for some time, look at git reflog	12:58
	→ Beket joined	13:02
	← EmilMedve left	13:04
Ilari	macgerhard: Maybe first try 'git write-tree' (if you don't have any staged changes, that should recreate tree object of HEAD...	13:08
	→ foutrelis joined	13:09
	→ jelmer joined	13:10
	→ vbabiy joined	13:11
	→ muthu_ joined	13:14
intripoon	hmm, when I do git reset	13:15
	and check in git log	13:15
	nothing has changed	13:15
	it does not remove the newest commit for me	13:15
Ilari	intripoon: 'git reset' is the same as 'git reset HEAD' (move 'branch pointer' from HEAD to HEAD, i.e. write value it already has)	13:16
intripoon	so how do I undo the latest commit then?	13:16
	someone said git reset undoes the latest commit	13:17
Ilari	intripoon: 'git reset --soft HEAD^'?	13:19
	→ sping joined	13:19
	← madsdyd_work left	13:19
intripoon	I don't know	13:19
	should I try that?	13:20
Ilari	intripoon: Or replace --soft with '--hard' or '--merge' (not in any release) if these modes are more suitable.	13:20
Weaselweb	intripoon: git reset just sets HEAD to the given commit	13:20
intripoon	I don't want my working tree changed	13:20
	only the last commit should be undone	13:20
	→ bryanl joined	13:20
Ilari	intripoon: reset with --soft does not touch the index or working tree.	13:20
intripoon	I don't want to set HEAD to the given commit	13:21
	I want to undo the last commit	13:21
	so it disappears completely	13:21
	in the git log	13:21
Weaselweb	where should HEAD stay, when you want to remove the last commit?	13:21
broonie	overwriting the head will do that.	13:21
intripoon	HEAD should be where it was before the last commit	13:22
macgerhard	Ilari: I have a lot of unstaged changes	13:23
	I think what I'll do is just clone again and then copy the files across & take it from there	13:23
	→ dramsay joined	13:23
macgerhard	what I would really like to know is how this happened...	13:23
Ilari	macgerhard: In this case staged changes, not unstaged changes.	13:23
Weaselweb	intripoon: you said you don't want to set HEAD...	13:24
intripoon	I don't know how you call what I want	13:24
	I want my git repository to be in exactly the same state like it was before the last commit	13:25
Weaselweb	as already mentioned: git reset --soft HEAD^ will set HEAD to the commit before the current one, leaving working dir and index alone	13:25
intripoon	what is the index?	13:25
macgerhard	Ilari: nope, they're still unstaged. if I do a git commit it won't commit anything. those are unstaged changes if I remember correctly.	13:25
Ilari	macgerhard: Did you try that 'git write-tree' thing? What does it report back? '40e42a504d10d78170bb89012b5a38fa3554f807'?	13:26
	← advany left	13:26
macgerhard	nope, it gives me another hash	13:26
Ilari	macgerhard: Darn...	13:27
macgerhard	I've tried it earlier on, gave me a different hash	13:27
	← nano- left	13:27
	→ name joined	13:27
macgerhard	just cloning from master and copying all files to it took care of it ;). From personal experience, when something is too difficult to do in git, think outside it. 9/10 works a treat.	13:29
Ilari	macgerhard: The first thing to do when encountering repository corruption is to make backup copy before doing anything potentially destructive. When recovering data, its easy to make mistakes that really lost data...	13:30
macgerhard	yeah, got burned a few times	13:30
	mainly when messing with reflog...	13:30
	or purges	13:31
Ilari	macgerhard: BTW: does 'git cat-file -t 40e42a504d10d78170bb89012b5a38fa3554f807' error out in the new clone?	13:32
macgerhard	unable to find	13:33
	that file	13:33
	that appeared as a tree in my errors	13:33
Ilari	macgerhard: Yeah... I expected that... Its not in upstream repo...	13:34
	← markelikalderon left	13:34
macgerhard	I used find on the upstream, it wasn't there so it was something screwy locally	13:35
Ilari	macgerhard: The way object database errors are fixed is to obtain good copy from somewhere (or recreate) the corrupted object.	13:35
macgerhard	Ilari: yet again providing me with invaluable insight into git :). cheers mate!	13:36
	I'm back on track now, it should be cool once I've finished commiting the unstaged changes	13:36
	← vbabiy left	13:37
Ilari	macgerhard: What happened could very well be silent data corruption. Fsck should have said if the object is corrupted or missing...	13:40
	→ EmilMedve joined	13:41
macgerhard	yes, it did say that the object is missing	13:42
	← oudeis left	13:42
	← muthu_ left	13:44
	← faithfulgeek left	13:46
	← girishr_ left	13:46
	→ girishr joined	13:46
wereHamster	wow, after a repack the repo size went down from 55MB to 3.1MB :)	13:48
Ilari	Altough very unlikely to be issue: Git integerity checking mostly only covers faults in lower layers. It does not cover Git itself very well. There are ways Git can write corrupt objects that pass SHA-1 check (altough that is very unlikely)...	13:49
	cbrake_away → cbrake	13:49
	→ bugfux joined	13:49
Ilari	If one sees two object ID that are almost the same (say, at least 128 bits match), one can suspect that one of the IDs itself is corrupted...	13:51
	← peritus_ left	13:52
	→ loincloth joined	13:54
	→ leachim6_ joined	13:54
	← leachim6 left	13:55
	← ankit9 left	13:55
	← girishr left	13:56
	→ guille_ joined	13:57
	→ sdboyer-laptop joined	13:58
	sdboyer-laptop → sdboyer\|sprint	13:59
	→ jm joined	13:59
	→ threeve_ joined	14:02
	→ nighthwk1 joined	14:14
	→ __Zimm joined	14:14
	→ calin joined	14:14
	← davidw left	14:14
	→ faithfulgeek joined	14:14
	→ krh joined	14:15
	→ markkalderon joined	14:16
	→ bnovc joined	14:16
	← agile left	14:16
	→ willb joined	14:18
	→ nofxx joined	14:19
	→ esden joined	14:20
	→ gambler joined	14:21
	→ knobo joined	14:23
knobo	Is subversion first or second generation version control system, and is git third or fourth?	14:23
	→ vbabiy joined	14:25
	→ juanjoc joined	14:25
timonator	define "generation"?	14:25
	→ peritus_ joined	14:25
	→ agile joined	14:26
	→ slainer68 joined	14:26
	← threeve_ left	14:26
	→ threeve_ joined	14:26
knobo	I'm not an expert in version control systems. I just wander if we can group types of version control systems somehow based on old vs. modern systems.	14:27
Beket	knobo, "new" vcs are distributed, such as git or hg. svn ain't one of those.	14:27
knobo	so if distributed vcs is one major step in vcs, are there more features that we could categorize as "steps"?	14:29
	if not, we have two generations.	14:29
slainer68	hi! i'm a git newbie. if i've got 2 locals branches, one branch master and one branch to implement a new big feature. if I commit an important bugfix on master, how can I merge it to the other branch if I need to ? thanks for your answers.	14:29
	← threeve_ left	14:29
Weaselweb	slainer68: that's really simple: checkout the develop branch and do "git merge master"	14:30
slainer68	thanks Weaselweb. and i think there is a parameter to only merge a single commit ?	14:31
knobo	how about rcs vs. svn?	14:31
	→ __ash__ joined	14:31
Weaselweb	slainer68: merging a single commit makes no sense, you merge branches	14:31
	← bugfux left	14:32
slainer68	Weaselweb: but what if there are 4 fixes bugs on master and I only want to merge one of these 4 in my dev branch ?	14:33
Voker57	slainer68: git cherry-pick	14:33
	→ nikolasco joined	14:33
Weaselweb	but that's not a merge, IMHO	14:33
	→ alikins joined	14:34
Voker57	true	14:34
	→ jackdempsey joined	14:34
	→ webmat joined	14:35
slainer68	thanks Voker57 & Weaselweb	14:35
cebewee	knobo: atomic commits	14:36
	← Beket left	14:37
	→ gitte joined	14:37
Ilari	knobo: I say Subversion is third-gen and Git fourth-gen... But that's just my opinion...	14:40
	← pygi left	14:40
	← nikolasco left	14:40
	← clamothe left	14:40
	→ sujith_h joined	14:42
	← parasti left	14:43
	→ sh12 joined	14:44
	← peritus_ left	14:46
	← ben_h left	14:47
sh12	i have been working on git for a long time now. i started my work at 2.6.25 branch. i have been making changes to git tree at 2.6.25 onwards and using 2.6.25 head still	14:47
	now kernel is at 2.6.28-rc8, there are plenty of changes in the kernel between 2.6.28-rc8 and 2.6.25. how do i merge only my changes to 2.6.28-rc8 tree	14:48
	please help	14:48
	← jackdempsey left	14:49
	← markwaters1 left	14:49
	redinger_away → redinger	14:49
Weiss	sh12: are you sure you don't mean "merge all the changes between 2.5.25 and 2.6.28-rc8 to your own branch"?	14:49
sh12	i want to merge my changes only to 2.6.28-rc8 tree	14:50
Weiss	alternatively, you can use 'git rebase' to transplant your branch onto a different base	14:50
	→ oudeis joined	14:50
sh12	how do i do it?	14:50
	what does it do?	14:50
Weiss	man git-rebase has explanations with diagrams	14:51
Gitbot	Weiss: the git-rebase manpage can be found at http://git.or.cz/man/git-rebase	14:51
Weiss	better than i could explain..	14:51
Weaselweb	sh12: why would you want to merge YOUR changes into master (respectivles origin/master)	14:51
	→ cannonball joined	14:51
sh12	okay i dont want to work on 2.6.25 anymore. i want to work out of the latest kernel sources	14:51
	← przemoc left	14:52
Weiss	merging either way should work, because you have a nice common ancestor between the two heads you want to join	14:52
Weaselweb	then, rebase is the thing you want	14:52
	→ przemoc joined	14:52
Weiss	but beware: rebasing involves 'changing history', so if anyone else is working on your topic branch then you'll 'pull the carpet out from under their feet'	14:53
	← sujith_h left	14:53
sh12	i get lots of confilicts in the files i havent modified	14:53
	how is it possible?	14:53
Weiss	sh12: that can happen if you get the syntax wrong and ask it to rebase the wrong changes	14:53
	is that possible?	14:54
sh12	weiss: im still trying to do git merge	14:54
	i will try git rebase	14:54
Weaselweb	upstream changes and yours merge cleanly	14:54
	rebase won't change anything to that	14:54
sh12	i think the git tree has become corrupted after merging	14:54
	do i need to download the git all over again?	14:54
	← EmilMedve left	14:55
	← j416 left	14:55
Weiss	it's unlikely it's corrupted. you need to set your branch back to where it was before	14:55
Weaselweb	or fix the merge conflicts	14:56
sh12	can i get a list of merge conflicts	14:56
	i think the list was too long. i cant see it on the console now	14:56
Weiss	if you generate loads of spurious conflicts in files you haven't touched, it's not even remotely worth trying to resolve them..	14:56
	← mithraic left	14:57
sh12	what is the best way of doing this	14:57
Weiss	'git reflog' will tell you which commit your branch pointed at before you started. use 'git reset --hard <sha1>' to put things back how they were before	14:57
	← drewr left	14:57
sh12	can i specify a tag. since i dont remeber the commit number	14:58
Weiss	sh12: you don't have to remember the number, because git keeps a record for you (run git reflog to see what i mean)	14:58
	→ oudeis_ joined	15:00
	← webmat left	15:00
	→ webmat joined	15:00
	← ffgeek200b left	15:00
	→ k0001 joined	15:00
	→ ffgeek200b joined	15:00
	← naeu left	15:01
	→ brett_h joined	15:04
	→ sujith_h joined	15:04
	← Sho_ left	15:06
	→ retybok joined	15:08
	← vbabiy left	15:08
	← oudeis left	15:09
sujith_h	I am using git for the first time. let me describe the things I did and the trouble I am facing. I tried to checkout anonymously the code from a savannah site using the command:`git clone git://git.savannah.nongnu.org/smc.git`. I added the folder I needed to add with my local repository. And executed the command "git add .". The newly created repository got added. Now did a commit. That also worked. Now issue lies in the 'git push' command. I had given my	15:09
	ssh public key to the savannah server. But its not allowing me to push. Can anyone help me regarding this. The error I am getting when I execute `git push` is : fatal: The remote end hung up unexpectedly	15:09
retybok	How would you handle the distinction between files that are used only internally in a project and files that are included in a source release?	15:09
	← zjason left	15:10
retybok	sujith_h: are you allowed to push on the remote repository?	15:10
sujith_h	retybok: Can you help me to test that?	15:11
	I am totally new to git	15:11
Weiss	sujith_h: almost certainly you need to use a different URL for pushing than for your initial (anonymous) clone. probably something like ssh://username@git.savannah.nongnu.org/smc.git. should be details on the site somewhere..	15:11
	→ jackdempsey joined	15:12
	→ zjason joined	15:15
Ilari	sujith_h: If you really do have push access for that project on Savannah, just edit the remote repo URL in '.git/config' to be apporiate ssh:// or git+ssh:// URL...	15:15
	→ naeu joined	15:16
	→ eevar2 joined	15:17
	← eevar2 left	15:18
	→ pi-meson joined	15:18
	→ EmilMedve joined	15:20
	→ pygi joined	15:21
	← rangaparma left	15:22
	← bbuss left	15:23
	← naeu left	15:25
sujith_h	Ilari:I changed in the config file to ssh://myusername@git.savannah.nongnu.org/smc.git. But i got the error: "Permission denied (publickey).fatal: The remote end hung up unexpectedly". But when I execute command "sftp [email@hidden.address] to test my ssh acces i am able to execute the command without any problem	15:25
	← Voker57 left	15:26
Ilari	sujith_h: dl.sv.gnu.org and git.savannah.nongnu.org are presumably not the same machine...	15:26
	← jackdempsey left	15:27
sujith_h	Ilari: Ooh	15:28
pi-meson	should git svn clone correctly interpret the svn branches/ dir and give me branches?	15:29
vmiklos	yes, if you use the '-s' option	15:30
	see the manpage	15:30
pi-meson	vmiklos: so "git branch" should then list the svn repo's branches?	15:31
Ilari	pi-meson: 'git branch -r'.	15:31
pi-meson	Ilari: that was the problem, thanks!	15:32
	← dramsay left	15:32
pi-meson	Ilari: does that mean that the clone that I do with git svn does not actually grab/convert those branches into my local git repository?	15:32
Ilari	pi-meson: It keeps them as remote branches. You could make local branches based on those remote ones.	15:33
	→ threeve_ joined	15:33
	← lorandi left	15:34
	→ gonzzor joined	15:34
pi-meson	Ilari: I apologize, i'm somewhat new to git -- does this mean that they are physically part of my repo or not? i.e. if the thing hosting the svn repo went away, would I lose any data?	15:34
	→ lorandi joined	15:34
gonzzor	According to the man git-daemon, git://host/alice/foo should make git-daemon use ~alice/foo, but it doesn't. Is it a type in man page?	15:35
Gitbot	gonzzor: the git-daemon manpage can be found at http://git.or.cz/man/git-daemon	15:35
vmiklos	gonzzor: that is git://host/~alice/foo	15:36
gonzzor	Yeah..	15:36
	Just saw it on the manpage gitbot linked to..	15:36
	Explains a lot of my problems..	15:36
Weaselweb	pi-meson: no, you wouldn't lose anything as git is a decentralized version controll system, where everybody has the full repository	15:36
vmiklos	gonzzor: though you have reason, the manpage seem to have somehow missing that tilde	15:37
gonzzor	vmiklos: Yes.. But I don't have the newest version of git..	15:37
vmiklos	i can reproduce it with 1.6.1-rc2.something as well	15:38
gonzzor	Ok..	15:38
	What should I put in my whitelist if I want --user-path=public_git work for ~100 users? I can get it to work if I manually add ~user1 ~user2 to whitelist... All users are under /home. But /home doesn't work in whiltelist..	15:38
vmiklos	isn't --export-all a solution for you?	15:40
	→ naeu joined	15:41
	→ drewr joined	15:41
sh12	im trying git rebase, i think it has deleted all my changes	15:41
	i have done all my changes (2.6.25) in a branch called devel	15:41
	im sitting in devel right now	15:42
	i gave a git rebase linux-2.6.28-rc8	15:42
	→ xlq joined	15:42
sh12	i dont see my changes in devel	15:42
Ilari	pi-meson: If you want to create local branches based on them, one way is 'for i in `git for-each-ref --format=%$refname$ refs/remotes` ; do git update-ref `echo $i \| sed -e s@^refs/remotes/@refs/heads/@g` $i ; done'	15:42
sh12	the head is moved to 2.6.28-rc8	15:42
	but without my changes	15:42
vmiklos	sh12: what was the output from the git rebase command?	15:42
pi-meson	Ilari: Okay, thank you so much. I think I also need to go and reread the part on branches in my git book again.	15:43
sh12	i got fail to merge changes	15:43
	← xlq left	15:44
sh12	http://pastebin.ca/1286887	15:44
jast	yeah, you've got conflicts in that rebase that need to be resolved	15:45
	as a general rule, you tend to get more conflicts with rebase than with merge	15:45
	edit the conflicted files, git add them and git rebase --continue	15:46
	the thing you pasted actually explains that	15:46
Weiss	sh12: did you properly sort out your previous situation, and start from a clean state?	15:46
jast	some of git's error message are quite helpful... if you read them :)	15:46
	← pigeon left	15:46
sh12	yes	15:47
	i download the git tree allover again	15:47
Weiss	sh12: muppet check: which branch did you have checked out before doing the rebase?	15:47
jast	devel	15:47
	as can be seen from the paste	15:47
vmiklos	gonzzor: http://article.gmane.org/gmane.comp.version-control.git/103283	15:47
Weiss	ah, ok	15:47
sh12	see i have devel branch which is 2.6.26	15:48
	i made all the changes to 2.6.26	15:48
	now git tree moved to 2.6.28-rc8	15:48
	→ Garito joined	15:48
sh12	i want to move my changes to 2.6.28-rc8	15:48
	→ andrewmaier joined	15:48
jast	yea, so follow the instructions	15:48
Weiss	do the conflicts concern files that you modified?	15:48
andrewmaier	Morning all	15:48
	→ pigeon joined	15:48
Garito	hi!	15:48
sh12	no	15:49
Garito	anyone who can help me to reconstruct a git repository?	15:49
	try to google without successs	15:49
jast	Garito, what seems to be the problem? :)	15:49
	→ withanx joined	15:49
sh12	Weiss, sorry to trouble you, i need to run for a meeting	15:49
Weiss	no problem	15:49
Garito	I change my ubuntu	15:49
sh12	im sorry to have you guys hanging	15:49
Garito	and I hope copying the folder will be enought	15:49
sh12	i will catch you later/tommorrow	15:50
andrewmaier	I'm trying to figure out how to not track a file. I'm using git to interface with svn and I've cloned the repo. Now I need to change a build.xml file but I don't want to ever commit changes to that file. I've tried dding it to .gitignore to no avail. Any Ideas?	15:50
sh12	bye	15:50
Garito	now I try to commit but don't work	15:50
Ilari	Garito: Define 'don't work'.	15:50
	← sh12 left	15:50
jast	andrewmaier, well, the easiest way is to not use git commit -a and never git add that particular file	15:50
Garito	can't commit anything with qgit (apair in gray)	15:50
andrewmaier	It's like, please clone this repo, save that one file. Is that possible?	15:50
vmiklos	andrewmaier: gitignore just controls what git-add adds to the repo, so that won't help for you	15:50
andrewmaier	Jast: Yeah, sad times.	15:50
jast	hmm?	15:50
andrewmaier	Re: "the easiest way is to not use git commit -a and never git add that particular file"	15:51
	>.<	15:51
jast	what's the problem with that?	15:51
andrewmaier	:shrug: Just defeats my normal qorkflow	15:51
	It's not really a problem per se	15:51
Garito	any idea?	15:51
gonzzor	vmiklos: Thanks	15:51
jast	as long as you don't switch branches there's an alternative	15:51
	← lamont left	15:51
jast	it's only semi-reliable though, I believe	15:51
	→ lamont joined	15:51
jast	git update-index --assume-unchanged build.xml	15:52
Garito	jast: this is for me?	15:52
jast	using stuff like git reset may defeat it	15:52
	Garito, nope :)	15:52
Garito	sorry ;)	15:52
Ilari	Garito: Does regular 'git commit' from commandline work? If not, what errors it prints?	15:52
Garito	I try, please, wait	15:52
andrewmaier	I'm going to try git update-index --assume-unchanged build.xml	15:52
	→ mbroeker joined	15:52
vmiklos	see 6259ac6628477aa5ebde9bd9e8daaeecca2a74ae	15:52
Gitbot	[git 6259ac662]: http://tinyurl.com/5zaxzc -- Documentation: How to ignore local changes in tracked files	15:52
cbreak	...	15:53
	would it not be faster to type the url than the hash? :)	15:53
	→ nice_burger joined	15:53
andrewmaier	Jast: I think that worked wonderfully	15:53
	Jast: You're my morning hero	15:54
jast	hooray :)	15:54
vmiklos	cbreak: not at all, to be honest	15:54
andrewmaier	Jast: 2 internets for thee.	15:54
jast	thanks	15:54
andrewmaier	:)	15:54
jast	to be on the safe side, check with git status -a before committing to see if it still works	15:54
	← thiago_home left	15:55
Garito	trying to commit from command line seems to work ok	15:57
	← scientes left	15:57
Garito	I try to work on it and, if I have any kind of question I will be back here to ask	15:57
	Thank you so much	15:57
	and see you!	15:57
	← Garito left	15:58
	← andrewmaier left	15:58
	← pigeon left	15:59
	← reithi left	16:00
	→ andrewmaier joined	16:00
	→ Beket joined	16:02
	← andrewmaier left	16:02
	← pgollucci left	16:04
	← cgardner left	16:04
	← x_or left	16:05
	→ zerok_ joined	16:06
	→ zobie joined	16:07
	→ nikolasco joined	16:08
	→ pigeon joined	16:10
	→ rlorandi joined	16:12
pi-meson	once i've pushed a branch to a remote repository with "git push origin foo" (for branch foo), simply deleting it locally won't delete it from the remote repo, correct?	16:12
	← sujith_h left	16:12
Pieter	yes	16:13
	→ PloenK joined	16:13
	→ rubydiamond joined	16:15
	→ jackdempsey joined	16:16
	→ moconnor joined	16:17
	→ marvil07 joined	16:17
	→ patrikf joined	16:18
	← HopsNBarley left	16:21
	← zerok left	16:22
	← reel_ left	16:24
	← schlurchz left	16:25
	← aziz left	16:26
	→ bieneff joined	16:27
	← lorandi left	16:27
	← willb left	16:28
	→ willb1 joined	16:28
	← bieneff left	16:29
	→ aziz joined	16:29
	→ bieneff joined	16:29
	→ rubydiam_ joined	16:29
	← Beket left	16:30
	→ stephens joined	16:31
	→ bobesponja joined	16:31
	→ alexyk joined	16:32
	→ willb joined	16:33
	← willb1 left	16:33
	← kanru left	16:34
	→ corybantic joined	16:35
pi-meson	is it possible to restrict "git status" output to a single directory?	16:36
	← moconnor left	16:36
	← metafollic left	16:36
vmiklos	git status <dir>	16:36
	← lfranchi left	16:37
pi-meson	vmiklos: I'm not sure that does what i'm looking for; running git status foo from within myrepo/bar/ (where myrepo/bar/foo exists) still shows me other things at the same level of directory hierarchy as foo	16:38
	→ macgerhard_ joined	16:41
	← macgerhard left	16:41
vmiklos	i haven't said 'cd <dir>; git status' but 'git status <dir>'	16:41
Pieter	vmiklos: that doesn't work	16:42
	vmiklos: that's the same as git commit --dry-run -- dir	16:42
	→ Jacolyte joined	16:42
	← rubydiamond left	16:43
	→ arekm joined	16:43
vmiklos	ok, then i don't see what you want to achieve	16:43
gonzzor	Why give "git add .", fatal: pathspec '' did not match any files, but git add gives, "Maybe you wanted to say 'git add .'?" Is my git broken?	16:43
	→ willb1 joined	16:43
arekm	if I've commited two patches and I wan't to merge these two into one, do I do rebase or something else?	16:43
PerlJam	arekm: aye, rebase -i is your friend there.	16:43
Pieter	vmiklos: something like 'git status \| grep ' sub/dir', I think	16:44
	vmiklos: the pointi s that you want to see what is staged already in those dirs	16:44
	vmiklos: but not see 'git status' output from other dirs	16:45
	vmiklos: I don't think that's possible	16:45
arekm	PerlJam: ok, looking for a howto then :)	16:45
	→ mulicheng joined	16:46
	→ lfranchi joined	16:47
arekm	PerlJam: thanks, worked :)	16:48
	← arekm left	16:50
	→ kumbayo joined	16:51
	→ amystrat joined	16:51
	→ altrux joined	16:51
amystrat	how can I add to my index just a part of a file	16:51
alexyk	I've looked at all the git guis -- gitk, qgit, gitx on mac -- and they all look the same, letting you see the history. But how do you do add -p with them? what else can you do in these guis?	16:51
amystrat	ie. lines 10-15, but not 20-22	16:51
	← gilimanjaro left	16:52
Pieter	alexyk: gitx kan do that	16:53
jast	git-gui too	16:53
Pieter	alexyk: and git gui to, by right clicking on the hunks	16:53
	→ Voker57 joined	16:53
	← svend__ left	16:53
alexyk	pieter -- ah, in diff view?	16:54
Pieter	alexyk: for gitx or git gui?	16:54
alexyk	gitx	16:54
Pieter	in gitx, in the commit view, there's a 'stage' button next to each hunk	16:54
alexyk	ok, thx!	16:54
	← willb left	16:57
	→ ceej joined	17:03
	→ gilimanjaro joined	17:03
	→ advany joined	17:03
manithree	alexyk: git-cola lets you stage hunks (and I think lines) also	17:04
	← tango_ left	17:07
alexyk	so what else are these guis good for?\	17:07
	can you do rebase?	17:07
neurodamage	oh boy a conflicted merge, how do I rectify this :)?	17:07
	← faithfulgeek left	17:09
	← nbrosnahan left	17:10
	→ nbrosnahan joined	17:10
	→ kek joined	17:11
	→ RayMorgan joined	17:11
	edwarda__ → edwardam	17:11
	→ ivanoats joined	17:11
	aziz → aziz\|away	17:11
	→ faithfulgeek joined	17:11
	→ iulian joined	17:12
	→ kevwil joined	17:12
	→ florian_bravo joined	17:12
	→ bentob0x joined	17:15
	→ tvw joined	17:16
	→ muthu_ joined	17:16
	→ gottesmm joined	17:17
	→ bod joined	17:17
jast	neurodamage, edit the conflicted files, git add them, commit	17:18
alexyk	neurodamage: git mergetool	17:18
	← pi-meson left	17:19
	← harinath left	17:20
	→ pi-meson joined	17:21
	← muthu_ left	17:21
	← gottesmm left	17:21
	→ LiamH joined	17:22
	→ muthu_ joined	17:22
bod	I've just checked out a branch (using -b local_branch_name) , made a change and pushed... my change is visible, but of course the repository branch does not reflect the change.	17:23
	while I can push tags, I can't see how to affect heads in the repository	17:23
	← kevwil left	17:24
bod	There is --all, but that seems a little more broad-spectrum than I'm after	17:24
	← namenlos left	17:25
	→ cgardner joined	17:26
	→ supercali joined	17:27
	← supercali left	17:27
	→ supercali joined	17:28
	← gilimanjaro left	17:28
	← faithfulgeek left	17:28
	→ zjason_ joined	17:29
	← __ash__ left	17:31
	← macgerhard_ left	17:32
	→ tango_ joined	17:32
	← manithree left	17:33
	rubydiam_ → rubydiamond	17:33
	← zjason left	17:34
	← amystrat left	17:34
	→ LeeQ joined	17:35
LeeQ	Hello git channel... noob here needs help getting started gitting	17:35
	→ faithfulgeek joined	17:36
	→ zjason joined	17:37
	→ __ash__ joined	17:37
jast	LeeQ, sure, but as long as you don't ask a question don't expect an answer ;)	17:37
	→ carllerche joined	17:37
patrikf	LeeQ: if you have specific questions, ask them :) - otherwise, start with reading the tutorial...	17:37
	← naeu left	17:38
	→ aroben joined	17:38
LeeQ	Sorry, had someone walk in just after I typed that... questions incoming	17:38
	aziz\|away → aziz	17:38
	← sgrimm left	17:38
jast	bod, what do you mean by "change is visble, but [...] the repository branch does not reflect the change"? isn't that sort of contradictory?	17:38
LeeQ	alright, My current method of deployment for my ROR apps is simply write the code locally, FTP it to my server, and hit refresh in my browser to see how it works...	17:39
	← markkalderon left	17:39
	← __ash__ left	17:39
LeeQ	I would like to stop being a loser and start doing things the right way	17:39
paroneayea	is there an equivalent of like, svn cat?	17:40
	what I'd like to do is	17:40
jast	paroneayea, git show commit:file	17:40
LeeQ	so I would like to develop locally (currently windows)and then use version control to move it to my server (linux)	17:40
paroneayea	jast: aha! thanks!	17:40
	← rtomayko left	17:40
paroneayea	perfect!	17:40
LeeQ	will I have a problem going between windows and linux?	17:41
Voker57	LeeQ: maybe not. but git is not too easy to install on windows	17:41
paroneayea	jast: oh, hm	17:41
	→ Yuuhi joined	17:41
paroneayea	does that only show the diff?	17:41
	I want to show the whole file at that revision	17:41
jast	leave out the ":file" and you get the commit diff. include it and you get the full file.	17:41
	→ coldwaterlover joined	17:41
Voker57	LeeQ: didn't you install it already?	17:42
LeeQ	no sir	17:42
	I have done nothing yet	17:42
Voker57	look here then http://www.google.com/url?sa=U&start=1&q=http://code.google.com/p/msysgit/&ei=aOhHSdmLBYbq0ASoqMDRDQ&sig2=WS9_2WU1o9UCRjGxmqu9YQ&usg=AFQjCNEYfLwmX-yP7p3gqngH4d8HehqTPw	17:42
jast	LeeQ, just make sure core.autocrlf (see man git-config) is enabled on the windows system, otherwise you may get mixed line endings	17:42
Gitbot	LeeQ: the git-config manpage can be found at http://git.or.cz/man/git-config	17:42
paroneayea	jast: aha :)	17:42
	awesome.	17:42
jast	one of msysgit and cygwin git has it enabled by default, i don't recall which one	17:42
Voker57	fuck, google is poisoning urls	17:42
jast	Voker57, only if you're logged in :}	17:42
	that's just because Google wants all your data, so don't worry	17:43
paroneayea	also only tracks you if you're logged in... all the more reason not to log in! :)	17:43
Voker57	jast: so this line has my password? :0	17:43
jast	session data, perhaps	17:43
	← zjason_ left	17:43
jast	it's probably secured in some additional way	17:43
paroneayea	it doesn't have your password.	17:43
	← gitte left	17:43
paroneayea	actually, the customizegoogle extension for firefox is pretty nice.	17:43
	→ powerfox joined	17:44
	Voker57 logs off	17:44
Voker57	nah, url's the same	17:44
jast	you can login but you can't ever logout completely ;)	17:45
	(just guessing)	17:45
	→ TwoPixelGrid joined	17:45
powerfox	Hi there. Is there anyway to join some commits in the repo?	17:45
	I can create another branch from Com1, then merge it with Com3 with squash to join Com2 and Com3 to new merge commit. But it painful.	17:45
	← calin left	17:46
	← gonzzor left	17:46
Voker57	ok i erased cookies and urls are nice	17:46
	also banned them	17:46
PerlJam	powerfox: git rebase -i ?	17:46
	→ naeu joined	17:47
	→ aspotashev joined	17:47
	→ corybantic_ joined	17:48
bod	jast: I checked out origin/t using -b t ...I now hove .git/refs/heads/t which is the same as .git/refs/remotes/origin/t	17:48
	jast: I then made my change, commited, tagged and pushed with --tags	17:49
	← faithfulgeek left	17:49
jast	and the tag got updated but the branch didn't?	17:50
bod	jast: so now, .git/refs/heads/t is no longer the same as .git/refs/remotes/origin/t. which is unchanged	17:50
powerfox	PerlJam: thanks, that is what I need. Didn't know rebase can do such cool things :)	17:51
jast	push doesn't automatically update the remote tracking branches, try a git fetch	17:51
bod	although the chnages, and the tag which are in the repository	17:51
	→ CodeOfficer joined	17:51
bod	OK	17:51
	→ snitko joined	17:51
bod	pdating local tracking ref '...'	17:54
	that seems to have done the trick, thanks	17:54
pi-meson	is it possible to select a subset of files from branch X commit Y and merge them with branch Z?	17:55
	← sping left	17:56
TwoPixelGrid	Hi all. First time in channel. Q: Is anyone using git for version controlling things other than source code? i.e would it be a suitable platform for version controlling media-heavy projects? I produce audio and trying to figure out how to bring version control into that world. Right now I'm stuck save-as-renaming song iterations etc...	17:56
	← corybantic left	17:56
	→ mithraic joined	17:56
offby1	I suspect it'll -work-, but not very well.	17:57
	The reason is: you can't really merge non-textual stuff	17:57
	and since one of git's big selling points is ease of merging ...	17:57
	you'd wind up not taking advantage of that selling point.	17:57
	← bod left	17:58
cnk	yes - but is there a better alternative? I don't know of anything that can identify the differences in big blobs of binary content	17:58
offby1	beats me.	17:59
pasky	offby1: yes, i don't think git is _worse_ than anything else	17:59
cnk	git is certainly better than versioning by renaming	17:59
offby1	yeah, certainly no other revision control system that I know of will do any better.	17:59
	Perhaps zillion-dollar proprietary systems like ClearCase would, but something tells me you're not considering them.	17:59
cnk	TwoPixelGrid: there is a git example of using it to manage binary content - I think it is in the Peepcode PDF, but am not 100% sure that's where I saw it	18:00
	← muthu_ left	18:00
pasky	git is still probably better than many others	18:00
	since you can define custom merge drivers for particular files	18:00
TwoPixelGrid	cool thx..ill check it out..	18:00
pasky	so if you have a tool that can merge doc files or whatever...	18:00
cnk	one selling point for git is that it would help you id corruption of the binary files	18:00
TwoPixelGrid	cnk : the only other direction i know to look is in the 'content management' world - i.e. AlienBrain -- used by game/movie production pipelines	18:01
cnk	(assuming what you checked in was OK, that is)	18:01
	that is probably worth looking at - but I don't know enough about that realm to comment	18:01
	← jmesnil left	18:02
cnk	certainly they are trying to solve the same problem you are	18:02
TwoPixelGrid	http://www.alienbrain.com/ fyi	18:02
	yeah..the audio production world has no concept of version control.. so we resort to convoluted file/folder structures etc..	18:03
	← naeu left	18:04
	← cannonball left	18:04
	← slainer68 left	18:05
	→ ahupp\|fb joined	18:06
	← neurodamage left	18:06
	→ gilimanjaro joined	18:07
	→ faithfulgeek joined	18:07
	→ smitty519 joined	18:07
	← smitty519 left	18:08
	→ markelikalderon joined	18:10
	← gilimanjaro left	18:11
	← EmilMedve left	18:11
	→ cannonball joined	18:11
	→ dreiss joined	18:12
	→ bugfux joined	18:13
	→ EmilMedve joined	18:14
	← bugfux left	18:18
	← metajack left	18:19
	→ metajack joined	18:19
	→ gilimanjaro joined	18:20
	← unreal left	18:20
	← oudeis_ left	18:21
	→ oudeis_ joined	18:21
	→ joydivider joined	18:21
	→ thiago_home joined	18:28
	→ svend joined	18:28
	→ drizzd_ joined	18:29
	← faithfulgeek left	18:29
SoopmaLuva84	i am having some issues getting gitweb to function properly	18:30
	i seems to be ignoring my owner files	18:30
	→ sgrimm joined	18:30
SoopmaLuva84	and base_url	18:30
	i cant seem to enable snapshot support either	18:31
	→ rtomayko joined	18:32
SoopmaLuva84	any ideas how to fix this?	18:33
	→ naeu joined	18:33
	← rolfb left	18:35
	← Yuuhi left	18:36
	→ faithfulgeek joined	18:36
	← webmat left	18:38
	← ijcd left	18:38
	→ SgtPepperKSU joined	18:40
	← SgtPepperKSU left	18:40
	← naeu left	18:43
	← drizzd left	18:45
	← ahupp\|fb left	18:47
	← alexyk left	18:50
	← zobie left	18:52
	← Ratler left	18:52
	→ Ratler joined	18:52
	← marvil07 left	18:54
	← EmilMedve left	18:54
	→ seed__ joined	18:56
	→ alexyk joined	18:56
	→ EmilMedve joined	18:57
seed__	Hi, I have my branch master and a branch dev , I just pull from origin into master, and I would like to apply changes on my dev branch is there a better way to do this than pulling origin from dev?	18:57
mulicheng	seed__: on the dev branch, won't "git merge master" work?	18:57
kevlarman	probably rebase makes more sense	18:58
	→ SgtPepperKSU joined	18:58
	→ hobodave joined	18:58
seed__	mulicheng: does merge delete one of the both branches afterward?	18:58
	→ flaguy joined	18:59
Ilari	seed__: Nope, merge doesn't delete branches...	18:59
	← ivanoats left	19:00
mulicheng	seed__: yeah, you still have both branches there, just that after the merge, your dev branch contains the changes made on master after you created the branch.	19:00
	rebase can work just fine too.	19:00
	→ berndj joined	19:01
	→ david_koontz joined	19:02
seed__	yes merge worked welll :), any way i'll take a look at git rebase :)	19:02
jast	generally I wouldn't recommend rebase	19:02
	← alexyk left	19:02
SgtPepperKSU	I have what is hopefully a simple question: I have two branches, BranchA and BranchB, that have merged a third branch, BranchC. BranchA later reverts the merge of BranchC as those changes aren't needed. Now, how can BranchB merge BranchA without it automatically reverting the merge of BranchC (which is what I'm seeing happen)?	19:02
	So I want BranchB to effectively have merged BranchA and BranchC.	19:03
	But, whenever it merges BranchA, BranchC's changes go away.	19:03
	I understand why it is happening (it is replaying the revert from BranchA), but I was wondering how I can get around this.	19:04
seed__	git told me my change could be fast forward so should I assume I won't get any trouble by using rebase ?	19:04
SgtPepperKSU	If I try to redo the merge of BranchC, it says it is already up-to-date	19:04
berndj	SgtPepperKSU, can you use git-merge --ours [blah blah]?	19:04
seed__	jast: why don't you recommend rebase?	19:05
SgtPepperKSU	when I merge A into B?	19:05
	→ unreal joined	19:06
	→ OlofS joined	19:06
	← berndj left	19:06
	→ bryanl_ joined	19:07
	← bryanl_ left	19:08
jast	seed__, it tends to cause more conflicts. generally I'd only recommend rebase if a) you want to make your history "prettier" or b) you have to use git-svn	19:09
	← florian_bravo left	19:10
	→ alexyk joined	19:10
seed__	jast: ok	19:11
SgtPepperKSU	berndj, --ours doesn't appear to be a valid option for git-merge	19:11
	→ bryanl_ joined	19:11
Ilari	SgtPepperKSU: '-s ours'?	19:12
SgtPepperKSU	right, just realized that. Will try	19:12
kevlarman	is there an easier way to get a YYYYMMddhhmm date than piping git show into grep and sed?	19:13
	→ docwhat joined	19:14
	← alexyk left	19:16
	→ ahupp\|fb joined	19:18
	← willb1 left	19:19
SgtPepperKSU	using '-s ours' I don't even pick up the changes from BranchA	19:19
	← nud left	19:19
	→ nud joined	19:19
	→ PhilK joined	19:19
	→ Sigma joined	19:20
jast	kevlarman, exactly the date or is a bit of additional info okay too?	19:20
kevlarman	just the date	19:20
	← ivanz left	19:21
jast	--pretty=format:%aD (or %ai, depending on what format you want)	19:21
kevlarman	%ai is closer	19:21
	since it's going into a filename i'd prefer to keep the spaces and weird characters out	19:22
	→ QBall joined	19:22
QBall	every time I want to push to a remote git repo (initial import) I get this error: error: unable to create temporary sha1 filename ./objects/45: Permission denied	19:23
	the permission of that file is! correct	19:23
	← bryanl left	19:23
	→ snorkel joined	19:23
timonator	QBall: on the remote side?	19:23
	or locally?	19:24
kevlarman	i guess i'm stuck with sed	19:24
	← powerfox left	19:24
QBall	locally	19:24
	pushing to repo.or.cz	19:24
	I just made	19:24
	I had this before, then I got around it, but can't remember how	19:24
timonator	oh, strange. repo.or.cz would hardly have a problem with permissions	19:24
QBall	every time it happens	19:24
	← Ravatar- left	19:24
QBall	timonator: full output git push pushrepo masterCounting objects: 8, done.	19:25
	Compressing objects: 100% (7/7), done.	19:25
	Writing objects: 100% (8/8), 6.07 KiB, done.	19:25
	Total 8 (delta 0), reused 0 (delta 0)	19:25
	error: unable to create temporary sha1 filename ./objects/45: Permission denied	19:25
Ilari	QBall: If its just made: 1) Did you add your user as authorized to push? 2) Wait a bit, there seems to be some cronjob involved...	19:25
SoopmaLuva84	can someone help me with gitweb	19:25
QBall	fatal: failed to write object	19:25
	error: unpack failed: unpacker exited with error code	19:25
	To ssh://repo.or.cz/srv/git/gmpc-discogs.git ! [remote rejected] master -> master (n/a (unpacker error))	19:25
	error: failed to push some refs to 'ssh://repo.or.cz/srv/git/gmpc-discogs.git'	19:25
	aaah	19:25
	shit	19:25
	http://mpd.pastebin.ca/1287019	19:25
	wrong clipboard	19:25
	sorry	19:25
timonator	heh	19:25
QBall	:(	19:26
	→ alexyk joined	19:26
QBall	ctrl-c and middle mouse button	19:26
	this is getting me so pissed off	19:26
	the error is frigging idiocity	19:26
timonator	so never use ctrl-c?	19:26
QBall	ctrl-v I ment.. a well	19:26
	how am I gonna fish this pushing error	19:27
timonator	it appears from the log, that the remote side outputs the error	19:27
	especially since it speaks about an "unpacker"	19:27
Ilari	QBall: Firstly, have you authorized any user to push to remote repo? AFAIK, by default nobody is authorized (not even project admins).	19:28
QBall	damn	19:28
	QBall grabs a gun	19:28
QBall	shoots himself	19:28
timonator	that's kinda contraproductive, man	19:29
QBall	so is the error message	19:29
	← alexyk left	19:29
	← rtomayko left	19:31
	→ ijcd joined	19:31
	← oudeis_ left	19:32
	← ijcd left	19:32
snorkel	I've been using this command to add untracked files to the index...	19:32
	git ls-files --exclude-standard -o \| git update-index --add --stdin	19:32
	Is there a shorter way?	19:32
Ilari	QBall: That kind of errors from repo.or.cz either indicate that user is not authorized to push, or the authorization has not yet propagated to actual repo...	19:32
	→ ijcd joined	19:32
	→ willb joined	19:33
timonator	snorkel: does it have to ignore files already under revision control?	19:33
Ilari	snorkel: 'git add .'?	19:33
timonator	Ilari: that will also add changes in already controlled files	19:33
Ilari	Yeah. Oops.	19:33
timonator	of course... you could do:	19:33
	git stash ; git add . ; git stash apply	19:33
QBall	Ilari: now I know	19:33
timonator	while not dramatically shorter, it's less complex... missing all the switches and all	19:33
QBall	an object/45 permission error is vague	19:33
SgtPepperKSU	So, from the lack of further response, I take it I'll have to due a rebase -i to get rid of the merge altogether, along with all the nastiness that comes along with rebasing a pushed branch?	19:34
snorkel	timonator: that might be more pleasant to use	19:34
jast	QBall, well, it's difficult for git to figure out what causes the problem, that's why	19:35
	it basically just means that for some reason or another, the directory can't be written to	19:35
Ilari	IMO, the new error message (in later versions) is even worse.	19:36
QBall	jast: stop haunting me :D	19:36
	jast: didn't realize it	19:36
jast	QBall, bah, I've been here for months, continuously	19:36
	:)	19:36
timonator	Ilari: what does the new one say?	19:36
	→ mw-home joined	19:37
jast	SgtPepperKSU, one other course of action I can think of right now is to do something other than merging to bring over the history from that other branch, though of course that will destroy further chances of merging the two branches properly	19:38
	→ ahupp\|fb_ joined	19:39
	danopia_ → danopia	19:39
jast	hmm, I think the hook scripts of repo.or.cz might be improvable to give a clearer error message	19:39
	→ snorkel_cs joined	19:39
jast	I'll look into that, perhaps	19:39
	← snorkel left	19:40
jast	but first I'll go get something to eat	19:40
	→ dominikh joined	19:41
	→ Ratler_ joined	19:41
Ilari	timonator: Well, newest versions have seemingly yet again reworded those errors...	19:41
	→ noob2000 joined	19:41
snorkel_cs	timonator: stashing; adding ;applying is slower ... I guess I can add an alias for the long command...right?	19:42
	← LeeQ left	19:42
noob2000	Hi, I have a local repository, which has been cloned from a remote one, but does not have remote tracking branches. Is there any way I can tell the latest shared commit with that repository (without fetching first)?	19:42
	← snitko left	19:43
	→ snitko joined	19:43
timonator	snorkel_cs: i guess, yeah	19:43
	marcelo → FFighter	19:44
timonator	snorkel_cs: maybe more like this:	19:44
jast	noob2000, look at git peek-remote, perhaps	19:44
timonator	git diff > patch ; patch -R < patch ; git add . ; patch < patch	19:44
	no, wait	19:44
	that won't clean the index	19:44
	anyway, whatever floats your boat...	19:45
jast	what was the goal here?	19:45
	adding untracked files and nothing else?	19:45
timonator	right	19:45
Ilari	noob2000: Well, at least fetch shouldn't transfer the shared part...	19:45
	← snitko left	19:45
snorkel_cs	jast: shorter command line for .. git ls-files --exclude-standard -o \| git update-index --add --stdin	19:46
noob2000	Ilari: I cannot perform fetch, from various reasons	19:46
	jast: peek-remote is the old version of ls-remote	19:46
timonator	noob2000: maybe you can gather the info from git reflog?	19:46
jast	why do you need a shorter command?	19:46
timonator	it won't of course tell you the right data, if the remote one pulled from you in the mean time...	19:47
jast	anyway, a slightly shorter version is to use "\| xargs git add" at the end	19:47
	and now I really have to go	19:47
timonator	bye jast!	19:47
	← pygi left	19:47
	→ rootPosixXombie joined	19:48
noob2000	timonator: I hope I get what you're trying to do right, but what about git-ls-files -o?	19:48
	← nice_burger left	19:48
snorkel_cs	noob2000: that doesn't use ignore/exclude	19:48
Ilari	noob2000: Look for first commit in 'git log' that doesn't have you as committer?	19:48
timonator	noob2000: don't know	19:49
noob2000	ok	19:49
	thanks	19:49
	← noob2000 left	19:49
rootPosixXombie	git log of a certain file shows that there were many commits: commit1, commit2 , commit3. I want to see the file before commit 3 was done. How can I do it?	19:49
timonator	git checkout it?	19:49
	snorkel_cs → snorkel	19:50
DrNick	assuming commit2 is commit3's predecessor, git show commit2:path/to/file	19:50
timonator	watch out for changes you made between commits	19:50
	that's way more elegant	19:50
rootPosixXombie	timonator, these is not my commits	19:51
	I am talking about existing kernel git tree	19:51
	← Ratler left	19:52
	← ahupp\|fb left	19:52
	← bryanl_ left	19:53
	→ peeja joined	19:53
	← SgtPepperKSU left	19:53
timonator	just use what DrNick told you and you'll be fine	19:53
hobodave	hey guys	19:53
	I've made changes in my working copy, that I want to move to a separate branch, how do I do that?	19:54
	I don't want to commit them to master (my current branch)	19:54
rootPosixXombie	timonator, that does not work	19:54
	git show commitNUm simply show the change	19:54
	→ oudeis joined	19:54
rootPosixXombie	I want to see all the file	19:54
	as it was before commit1, commit2, commit 3, ........ comit N were done	19:55
	is there a way?	19:55
timonator	did you add :path/to/file, too?	19:55
	→ zjason_ joined	19:56
timonator	sure works for me.	19:56
snorkel	hobodave: git checkout -b NewBranch ... somebody correct me if I'm wrong	19:56
rootPosixXombie	it gives an error	19:57
hobodave	that preserves my working copy?	19:57
	I have a ton of changes I can't afford to lose	19:57
timonator	rootPosixXombie: can you no-paste that error?	19:57
rootPosixXombie	git show d34f3015c5f:a.a	19:57
	or fatal: ambiguous argument ':a.a':	19:57
snorkel	hobodave: hang on one sec	19:57
timonator	that's certainly strange; a.a is a regular file?	19:58
wereHamster	hobodave: commit the changes, then git checkout -b newbranch; then go back to the original branch with git checkout master; and remove the commit with 'git reset --hard HEAD~1'	19:58
	(assuming the original branch was indeed 'master' ...)	19:59
Ilari	hobodave: 'git checkout -b newbranch' does not overwrite uncommitted modifications.	19:59
rootPosixXombie	timonator, ok, this was my mistake	19:59
	I though to create a file a.a	19:59
	← statim_ left	19:59
rootPosixXombie	it works - shows the file on screen	19:59
	can I redirect it to lets say a file named a.a	19:59
	?	19:59
snorkel	hobodave: i just tested git checkout -b newbranch and it worked fine	20:00
timonator	rootPosixXombie: yeah, with > a.a	20:00
rootPosixXombie	thnks	20:00
	great	20:00
	works	20:00
snorkel	hobodave: you can also zip up your changes before that if you want to be extra safe...	20:03
	← advany left	20:03
jast	no need to	20:03
	→ bryanl joined	20:03
jast	checkout -b foo doesn't change anything	20:03
	except HEAD, of course :)	20:04
snorkel	jast: trust no one...especially version control :)	20:04
jast	I know pretty well what to expect from common git commands	20:05
snorkel	jast: until some exception happens	20:05
jast	you don't have to trust me, of course	20:06
	feel free to do a bit of superfluous backup stuff if you don't	20:06
snorkel	jast: hobodave said he had a ton of changes that he didn't want to lose so i was merely suggesting that he could also zip them if it made him feel better	20:07
hobodave	heh	20:07
	snorkel: does git provide a method to do that?	20:07
jast	for committed stuff, yes	20:07
timonator	git diff > mychanges for uncommited stuff	20:07
snorkel	hobodave: what platform are you on?	20:07
jast	git diff won't include untracked files, of course	20:07
timonator	of course not	20:08
jast	but untracked files are left alone by virtually all commands anyway	20:08
	just like checkout -b leaves alone all files in the working tree ;)	20:08
timonator	what about a pull that pulls the creation of a file that locally exists as untracked? it will say "merge failed" or something, right?	20:08
jast	it will not even start merging	20:08
	it will say "untracked file would be overwritten by merge" and abort	20:08
	→ ajonat joined	20:09
timonator	nice	20:09
	← zjason left	20:09
hobodave	snorkel: osx	20:10
rootPosixXombie	now, a difficult question: how can I now which commit changes a certain file ?	20:10
hobodave	I can figure that out though	20:10
	git diff	20:10
rootPosixXombie	a certain field	20:10
hobodave	I'm not totally clueless :)	20:10
	just was nervous that I had over a thousand lins of changes w/ no commit (bad me)	20:10
timonator	a certain... field?	20:10
	← mcella left	20:10
rootPosixXombie	I know for sure that there was a change in a filed named y	20:10
	I mean:	20:10
	it was	20:10
	varA.interval	20:11
	and now it is varB.interval	20:11
	there were many many many commits	20:11
	between when it was varB.interval	20:11
timonator	i believe you can use pickaxe for that kind of thing	20:11
rootPosixXombie	what is this	20:11
	pickaxe?	20:11
	→ mdt joined	20:11
doener	git log -SvarB.interval	20:12
mdt	Hi everyone	20:12
rootPosixXombie	doener, trying...	20:12
snorkel	hobodave: i use cygwin and use something like 'tar jcvf changes.tar.bz2 -T <(git ls-files -m)	20:13
	edwardam → edwardam_lunch	20:13
snorkel	hobodave: or 'git diff --name-only HEAD^' for the files in the last commit	20:13
	→ k776 joined	20:13
	→ solydzajs joined	20:14
peeja	how might I search for a file/directory by name across a number of branches?	20:16
mdt	I would like to change git-gui to be able to, while standing on a superproject, see, commit and push, changes on every submodule. So automatically a commit about a single bug resolution that modified several files on different submodules gets done on each one and the superproject as well. Anybody interested on giving me a hand?	20:17
	snorkel claps	20:17
snorkel	mdt: sorry couldn't resist :)	20:17
mdt	:(	20:18
	did I said something stupid? :S	20:18
jast	no, someone's in a punny mood, that's all	20:18
snorkel	jast: ha!	20:18
mdt	:P	20:18
	→ LeeQ joined	20:19
mdt	so, is there a way already to do what I want? or will I have to go into the git-gui sourcecode?	20:19
jast	probably the latter	20:20
	← rlorandi left	20:20
	→ ijcd_ joined	20:21
mdt	I was afraid about that	20:21
	jast: do you think anyone else would be interested on having that feature?	20:22
jast	I wouldn't. I'm too interested in having meaningful commit messages in all repositories.	20:22
	← hurikhan\|Work left	20:23
mdt	But what do you do if you fix a bug that requires to touch at least tow files on different submodules?	20:23
	j_engelh tows mdt away	20:23
jast	i commit the changes and update the gitlinks	20:24
	← foutrelis left	20:24
mdt	but you have to make the commit on each submodule, repating the msg and do it again on the superproject if you want to update the submodule reference?	20:25
	← ahupp\|fb_ left	20:25
jast	why would I repeat the message?	20:25
	I would write different messages that are relevant for each repository	20:26
	→ tpiep joined	20:26
	→ statim joined	20:26
mdt	you would repeat the message if the change made on each repository is about the same thing.	20:27
	→ plediii joined	20:27
	← sstangl left	20:27
mdt	if two submodules communicate between. And you want to change the communication interfeace between them, you would have to make changes on both submodules, and the git message would be the same.	20:29
jast	because I tend to explain the changes	20:29
	and it's quite unlikely you have to make exactly the same change in each submodule to fix the bug	20:29
	→ yann joined	20:29
mdt	hm, don't know why I need it so often then =(	20:30
	→ calin joined	20:31
mdt	I have two submodules very connected between each other, and I often need to commit the same message like "Fixing bug #123, changing blabla" on both of them _and_ the superproject.	20:31
	→ ahupp\|fb joined	20:32
	→ robinr joined	20:32
	→ _graham_ joined	20:32
	← ijcd left	20:37
mdt	anyway, thanks for the help	20:38
	← mdt left	20:40
	← solydzajs_ left	20:40
	→ Darkhost joined	20:42
Darkhost	http://imperiumreklamy.com/pages/ptp.php?refid=krzysztof9222<---- Look at this!!	20:42
	← aspotashev left	20:42
	← plediii_ left	20:42
	← Darkhost left	20:43
	← rootPosixXombie left	20:43
	→ greyface joined	20:44
	← plediii left	20:44
	← snorkel left	20:45
	→ xyzzy joined	20:46
greyface	I'm trying to figure out this git thing, but it turns out that after a couple git resets, my local dir is empty of my source ;/	20:46
	→ geckosenator joined	20:48
geckosenator	if I type git checkout"	20:48
	and I get M file1	20:48
	← bobmcw left	20:48
geckosenator	and a bunch of files with M before them	20:49
	what is that supposed to mean?	20:49
jast	it means you have uncommitted changes in those files and they got carried across the checkout	20:49
geckosenator	oh	20:49
jast	in checkout without an argument it just means uncommitted changes :)	20:49
geckosenator	but if I specifically checkout the directory they are in, it undoes the changes	20:49
jast	checkout with a path specified works differently from the version without	20:50
geckosenator	ok	20:50
greyface	How can I reset my local git repo to how it was in the very beginning?	20:51
	← dwmw2 left	20:51
geckosenator	greyface: that's what I think I want to do :-P	20:51
damog	greyface: git reset 00hard	20:52
	greyface: git reset --hard	20:52
jast	--hard :)	20:52
greyface	my local dir is still empty ;/	20:52
jast	did you actually ever commit anything? what is the "very beginning" you spoke of?	20:53
greyface	jast: yea I commited to github	20:53
	very beginning being before I installed git	20:53
damog	lol	20:54
geckosenator	heh	20:54
jast	so you want to throw away all of your repositories? just delete them.	20:54
greyface	nono	20:54
QBall	lol.	20:54
greyface	I want to recover my files	20:54
jast	anything you didn't commit that is now gone is gone for good.	20:54
Ilari	greyface: Does newest commit contain anything (git ls-tree HEAD)?	20:55
greyface	that didnt return anything	20:55
QBall	uwh owh	20:55
greyface	thing is, I never deleted those files	20:55
	they deleted when I did a reset	20:56
jast	well, there are quite a few commands that remove stuff, and that are explicitly documented as doing such	20:56
	so does the repository on github look like what you want?	20:56
	← rubydiamond left	20:56
greyface	well, at least it has most of the files	20:57
	lol	20:57
	→ pjhyett joined	20:57
jast	how did you start out? by cloning from github?	20:57
	→ snorkel joined	20:57
greyface	no I had a local project	20:57
	and I pushed it	20:58
	added all my local files to git, pushed it, was working fine	20:58
jast	okay, let's first see of there are any local remains. what does "git ls-remote ." output?	20:58
greyface	e7aef9df053077fe1549f74034f99c515cd8a5c6 HEAD	20:59
	and 2 more lines	20:59
	e7aef9df053077fe1549f74034f99c515cd8a5c6 refs/heads/master	20:59
	1709e5078b221adc08f8a07614d6aa3caec18311 refs/remotes/origin/master	20:59
jast	okay	20:59
	and the output of "git show HEAD:" is empty?	21:00
Ilari	greyface: 'git ls-tree origin/master'. Does it show anything?	21:00
jast	if so, what about "git show HEAD^:"?	21:00
greyface	ok so	21:00
	first one is empty	21:00
	2nd one has my files	21:00
jast	2nd being Ilari's or mine?	21:01
greyface	git show HEAD; is empty	21:01
	git ls-tree orig... is full	21:01
	and HEAD^ returns an error	21:02
jast	what error?	21:02
greyface	unknown revision	21:02
jast	hmm, interesting	21:02
greyface	ambiguous argument 'HEAD^': unknown revision or path not in the working tree	21:02
jast	you must have accidentally done something rather funky to get into this situation	21:02
greyface	lol	21:02
	what I did is	21:02
jast	the quick fix is git reset --hard origin/master	21:02
greyface	git reset --hard HEAD^	21:02
	a bunch of times	21:03
xyzzy	hahahhaa	21:03
	i did that too once	21:03
jast	oh, so you removed all commits, one by one	21:03
greyface	lol >_>	21:03
	yea	21:03
	I wanted to go back to the inital state	21:03
timonator	not removed, per se...	21:03
jast	the very first initial state is "nothing" :)	21:03
greyface	LOL	21:03
	nice	21:03
xyzzy	luckily i had the original HEAD's sha1 written down on paper	21:03
	→ vbabiy joined	21:03
timonator	LOL :D	21:03
	the whole sha1?	21:03
jast	UI	21:04
	erm	21:04
	I accidentally the whole sha1!	21:04
Ilari	greyface: Maybe look at 'git reflog' to see if there's more stuff recoverable...	21:04
timonator	apparently you also accidentally the verb!	21:04
greyface	ok so I did git reset --hard origin/master	21:04
jast	quite true. it's an internet meme.	21:04
greyface	now I gotta pull it right	21:04
timonator	yeah, i thought that was the memonic response	21:04
greyface	oh no	21:05
	it's back already	21:05
	lol	21:05
	edwardam_lunch → edwardam	21:05
greyface	ok awsome jast	21:06
jast	for the future, if you want to go back to some time way in the past, find the right commit and use its sha1 with git reset	21:06
greyface	ok :)	21:06
	almost had a panic attack	21:06
	lol	21:06
jast	that will save you lots of steps too	21:07
	and perhaps don't reset away that many commits if you've already published them, depending on what you're trying to do	21:07
greyface	I was trying to remove files from git	21:07
	with git -rm, but it removed the actual files	21:07
	← Ramblurr left	21:07
greyface	thats when i tried to go back	21:07
jast	rewriting history tends to give other people who've got a copy of the repo headaches	21:07
	you can always undo changes to individual files	21:07
	in this case, git checkout HEAD -- file would have done the trick	21:08
greyface	right	21:08
jast	also, for the future, git rm --cached untracks files without physically deleting them	21:08
greyface	but how do I remove a file from git	21:08
	without deleting the file	21:08
	ah	21:08
jast	cool, I answered your question faster than you asked it ;)	21:08
greyface	yea lol	21:08
	→ ankit9 joined	21:09
	→ Beket joined	21:09
jast	to keep you from panicking in the future: it's actually very hard to lose committed stuff you lost rather recently unless you use certain cleanup commands	21:10
	← geckosenator left	21:11
	→ rtomayko joined	21:11
greyface	aha yea I thought it was pretty safe I just found it surprising that it went back to a version which didn't actually exist in my versionning system	21:12
	← unreal left	21:12
jast	yeah, I didn't know either that reset would step past the very first commit, and I don't quite understand how it can do so	21:12
	but let's not worry about that since we've got things fixed anyway	21:13
greyface	yea, thanks for the help!	21:13
	← drewr left	21:13
jast	you're welcome	21:13
Ilari	I'd say that the first version is probably empty. Reset can't step past the intial version.	21:13
	← EmilMedve left	21:13
greyface	yea	21:13
jast	yeah, quite possible, I just didn't think it were based on the descriptions	21:14
Ilari	Since reset always requires valid commit ID to write as "branch pointer". And state behind initial commit has no branch pointer value.	21:14
jast	that's what had me puzzled	21:14
greyface	yea me 2 even	21:14
	lol	21:14
	→ dwmw2 joined	21:14
Ilari	greyface: What does 'git rev-parse e7aef9df053077fe1549f74034f99c515cd8a5c6^{tree}' print?	21:15
greyface	maybe I did a commit before adding any files	21:15
	but I dont even remember	21:15
	543b9bebdc6bd5c4b22136034a95dd097a57d3dd	21:15
jast	you can now look at the history and find out :)	21:16
greyface	howdidothat?	21:16
	→ unreal joined	21:16
jast	i think the most intuitive way is to use gitk	21:16
	or one of the other GUI viewer tools	21:16
greyface	oh	21:17
	ok im in there...	21:17
Ilari	greyface: Hmm... That doesn't seem to be the empty tree... Puzzled...	21:17
	← netoman left	21:17
	← pjhyett left	21:17
greyface	ah theres a first commit	21:17
	before my first proper commit	21:17
	and its empty	21:17
	oh	21:18
	yea I remember now	21:18
	← coldwaterlover left	21:19
	→ rem7 joined	21:19
	← solydzajs left	21:19
	← oudeis left	21:19
greyface	Ilari, are you viewing my git repo?	21:19
j_engelh	"Are you vieweing my git repo? Are you, punk? That's mine! Hear me? Mine!"	21:20
greyface	lol	21:20
	→ chris2 joined	21:21
	← bentob0x left	21:23
	→ esdentem joined	21:24
rem7	what does it mean when a file changed mode? I'm pretty sure I didn't modified any files in my repo, but I did put it in my USB, and when I copied it in another machine, git status showed that all my files had changed. so when I did a commit it said files changed mode from 100644 -- 100755...	21:24
	← rtomayko left	21:25
rem7	does that have to do with permissions, users?	21:25
	← esden left	21:25
xyzzy	you probably mounted your usb memory stick with different umask	21:25
	← marcan left	21:25
timonator	permissions only	21:25
doener	rem7: usb stick -> likely FAT fs -> no unix permissions	21:25
	← ajonat left	21:25
	→ marcan joined	21:26
	← snorkel left	21:26
	esdentem → esden	21:26
rem7	is there a way to ignore permission changes?	21:26
doener	core.filemode or so, check the git config man page	21:26
rem7	thanks	21:26
	← OlofS left	21:26
doener	but I'd just fix up the permissions...	21:26
rem7	how so? like, chmod the file the same way in both machines before doing anything?	21:27
j_engelh	more like, copy the file with proper permissions in the first place	21:27
	← ricky left	21:29
	← tvw left	21:31
	→ rtomayko joined	21:32
	→ bobmcw joined	21:32
	← rem7 left	21:33
	← patrikf left	21:33
	← yann left	21:35
	→ OlofS joined	21:36
	→ HttpErrors joined	21:36
	← docwhat left	21:37
	→ kukks joined	21:38
	→ drewr joined	21:39
	→ oudeis joined	21:40
	← vbabiy left	21:41
	← Weaselweb left	21:44
	→ bugfux joined	21:45
	→ snitko joined	21:52
	← Acryler left	21:53
	← OlofS left	21:54
	→ thiago_h1me joined	21:54
	← nikolasco left	21:54
	→ parasti joined	21:55
	← Varon left	21:57
wereHamster	is there a git command to check whether `pwd` is a git repository? Something to replace an 'if [ -d .svn ]; then' test to generate a version string	22:01
	→ livinded joined	22:01
	← PhilK left	22:01
	← pi-meson left	22:01
livinded	do all branches in a git repo share untracked files?	22:01
	← rtomayko left	22:01
	→ rtomayko joined	22:02
Ilari	livinded: Index and working tree (including untracked files) is shared by all branches. But of course files that are untracked in one branch might not be untracked in another...	22:03
livinded	I branched master to create a new branch, added some files, didn't commit, checked out master, and the files were still there	22:03
	is that supposed to happen or is something wrong?	22:04
	← bobmcw left	22:04
Arjen	That's normal	22:04
livinded	now if I commit in my new branch and then checkout master again will they be there?	22:05
Arjen	No	22:06
	← thiago_home left	22:06
	biv_afk → brain_in_vat	22:07
livinded	alright, I thought something weird was going on, but that makes sense	22:08
	← kumbayo left	22:11
	→ thiago_home joined	22:14
	cbrake → cbrake_away	22:15
	← seanius left	22:15
	← lucsky left	22:18
	faithfulgeek → fakefowlduck	22:18
	fakefowlduck → faithfulgeek	22:18
	← bryanl left	22:19
	→ ben_h joined	22:19
	← faithfulgeek left	22:20
	→ faithfulgeek joined	22:22
	→ seanius joined	22:22
	→ marcot joined	22:25
	→ __name__ joined	22:25
	← mw-home left	22:25
marcot	Hello, how do I get a list of all unreferenced commits?	22:25
	greyface → greyface\|away	22:25
	→ pygi joined	22:26
marcot	I just have a branch, and a commit is not being shown on git log, cause it's not referenced by HEAD. But when I do git log abc1 it's shown.	22:26
	→ daxelrod joined	22:27
Pieter	you can't do that in any easy way	22:27
	← gilimanjaro left	22:27
vmiklos	git prune -n -v?	22:27
Pieter	easiest is probably to have another branch point to it	22:27
	and then use git log --all	22:27
Ilari	marcot: 'git reflog' and 'git fsck --full'?	22:28
	← thiago_h1me left	22:28
	← SoopmaLuva84 left	22:28
	← nud left	22:29
	xyzzy → berndj	22:30
marcot	vmiklos: There's no -v option here for git-prune. Maybe it's cause I'm using version 1.5.6.5.	22:30
	Ilari: I think git fsck --full is what I want, thanks.	22:31
	Ilari: Or just git fsck is giving the same result.	22:31
vmiklos	marcot: ah, -v is not necessary.	22:31
	-n itself will print out the necessary info as well	22:31
	← daxelrod left	22:31
marcot	vmiklos: oh, yes, it gave me more things than git fsck	22:32
	→ daxelrod joined	22:32
	→ michaelboutros joined	22:32
	← daxelrod left	22:33
michaelboutros	hello.	22:33
vmiklos	probably you need something like 'git prune -n --expire now\|grep commit$', but you'll figure that out	22:33
michaelboutros	I have, for example, cloned a repo from Github	22:33
	to get updates, should i use git-fetch or git-pull? they seem to do the same thing...	22:33
	← name left	22:33
jast	git pull actually means git fetch followed by git merge <the remote branch>	22:34
marcot	Hum, none of them gave me what I want...	22:34
jast	so git fetch doesn't update your local branch	22:34
marcot	I couldn't git log on any of the results.	22:34
	→ svend_ joined	22:34
marcot	So I couldn't get to the commit I tested earlier.	22:34
vmiklos	are you sure the commit is not in any reflog?	22:35
marcot	There was no commit..	22:35
vmiklos	then you won't see it as a commit	22:35
marcot	It's in reflog	22:35
	→ agenteo joined	22:35
michaelboutros	jastL so git pull would be the way to go?	22:35
vmiklos	try git log -g	22:35
	-g reads from the reflog	22:36
jast	michaelboutros, generally yes. git fetch has the advantage that you can review the changes on each side (using, for example, git log master...origin/master) before merging	22:38
	but you don't really always need to do that	22:38
michaelboutros	ah, thanks a lot then :)	22:38
	yeah, git pull would do what i need	22:38
	thanks!	22:38
	← michaelboutros left	22:38
marcot	vmiklos: git log -g gives me the commit, but how can I filter only the unreferenced ones?	22:38
	← berndj left	22:38
	→ gottesmm joined	22:39
	← faithfulgeek left	22:39
	→ xyzzy joined	22:39
	← TwoPixelGrid left	22:41
vmiklos	marcot: if you see the commit _only_ in the reflog, then git prune --expire now -n should find it as well	22:42
	→ fg3 joined	22:42
	xyzzy → berndj	22:42
	← boto left	22:42
	deskinm → deskinm_away	22:42
fg3	can git import a perforce repo?	22:42
vmiklos	(you can also try git branch -a --contains <hash> to see which branch has a reference to it, if git prune does not list the commit)	22:43
	← Beket left	22:43
marcot	vmiklos: It's not finding, it. git prune --expire now -n \| grep commit returns me nothing.	22:43
vmiklos	then probably it's referenced	22:43
marcot	I just have a branch.	22:44
vmiklos	try the git branch -a --contains trick	22:44
	<HEAD:contrib/p4import/README>	22:44
Gitbot	[git README]: http://tinyurl.com/59buua [blob]	22:44
vmiklos	fg3: ^ have you read this one?	22:44
	← kukks left	22:44
marcot	vmiklos: git -a --contains abc1 returns me nothing.	22:44
	→ mcella joined	22:44
marcot	vmiklos: but it's in git log -g	22:44
	I'll do a shell script.	22:44
	← gottesmm left	22:45
	→ jbwiv joined	22:45
	← svend left	22:45
vmiklos	marcot: strange, the documentation does not talk about git-prune respecting reflogs.	22:46
jbwiv	guys, I have a remote branch I'm trying to checkout and modify, so that I can push it back. I'm trying to understand the proper way of doing this. This remote branch lags behind the local master. Is the proper way to do this: $ git branch -b my1_5; git pull origin origin/v1_5 ?	22:46
marcot	vmiklos: maybe it's a bug then.	22:46
vmiklos	marcot: does git fsck --unreachable\|grep commit gives you anything?	22:46
	→ bugfux_ joined	22:47
marcot	vmiklos: no.	22:47
	vmiklos: it gives the same as git-prune -n (which is the same with --expire now)	22:47
vmiklos	interesting.	22:47
jbwiv	or should I simply do $ git checkout -b -r origin/v1_5 ?	22:47
	← brett_h left	22:47
marcot	vmiklos: do you want to see the repo?	22:48
vmiklos	well, if it's a bug, then you can try to write a script that demonstrates your problem	22:48
marcot	I'm not sure how to reproduce this, sorry, I'm not very experienced in git.	22:48
	I know this is the situation of my repository now.	22:48
vmiklos	is the repo public?	22:49
	→ thiago_h1me joined	22:49
	← thiago_home left	22:50
vmiklos	jbwiv: isn't v1_5 a tag?	22:50
	← livinded left	22:50
jbwiv	vmiklos: I believe it's a remote branch...how would I tell?	22:50
vmiklos	git branch -a lists branches, git tag -l lists tags	22:51
jbwiv	it's not a tag	22:51
	because it's tracking a separate line of development	22:51
	→ aresnick joined	22:51
vmiklos	try 'git checkout -b v1_5 origin/v1_5'	22:51
	then you can work in your local v1_5 branch	22:51
jbwiv	origin/v1_5 shows up in git branch -a as a remote tag	22:51
j_engelh	argh svn!	22:51
vmiklos	git branch lists branches only :)	22:52
aresnick	Hi! I'm new to git, and I was wondering if it was possible to check out a subset of a repository or branch? That is, if I just wanted to check out a single file, how would I do that?	22:53
	← drewr left	22:53
jbwiv	vmiklos: here's what is confusing me. The new PragProg Git book says "You can see your local branches when you run git branch. If you add the -r parameter,Git shows you the remote branches"	22:53
	→ kukks joined	22:53
jbwiv	but then it turns around and says	22:53
Ilari	aresnick: Its not currently supported (there are patches for that)...	22:53
marcot	vmiklos: yes, it's public.	22:53
jbwiv	"You can checkout those branches just like a normal branch, but you shouldnt change them. If you want to make a change to them, create a local branch from them first, then make your change."	22:53
vmiklos	jbwiv: perfect	22:54
jbwiv	vmiklos: then it talks about using git pull to do this	22:54
	but with no example of how	22:54
	so I tried to create an empty branch off master, and then pull	22:54
aresnick	Ilari: OK, would you be able to point me to them, offhand?	22:54
jbwiv	but of course, many conflicts from that	22:54
aresnick	Ilari: [the patches]	22:54
	← cmarcelo left	22:54
vmiklos	jbwiv: git checkout -b local origin/remote is the way to create a local branch from a remote/tracking branch	22:55
jbwiv	vmiklos: so the key is to do 'git checkout -b v1_5 origin/v1_5' instead of 'git checkout origin/v1_5'?	22:55
vmiklos	yes. only the previous will create a new local branch	22:55
jast	aresnick, you can always get single files from the repository by using, for example, git show commit:file (or, for checking it out, git checkout commit -- file)	22:56
vmiklos	the other just updates the working dir, in case you want to compile-test or something	22:56
Ilari	aresnick: They are in branch 'next' of git.git...	22:56
vmiklos	(ie: you don't want to commit)	22:56
	marcot: what's the url?	22:56
jbwiv	vmiklos: aha...I see. Ok thanks. So is this the equivalent to doing a branch off of master and pulling the changes, or is there a difference?	22:56
	aside from the pull requiring you to merge?	22:56
vmiklos	marcot: in fact i think it won't be reproducible if i clone the repo, it's something related to your _local_ reflogs	22:57
	marcot: so if you really want to let somebody debug this, you probably will need to tar up your repo and upload it	22:57
	jbwiv: huh, yes. but that's really overcomplicating things :)	22:58
	if you want to do it 'manually' then do a git branch v1_5 origin/v1_5, then git checkout v1_5	22:58
marcot	vmiklos: ok, I'll upload a tarball.	22:58
	vmiklos: http://marcot.iaaeee.org/distro-creator.tar.gz	22:59
jbwiv	vmiklos: great....thank you!	22:59
vmiklos	marcot: great, i like small repos:)	23:00
Ilari	jbwiv: There is difference. The case it shows the most is if local master and remote branch have both diverged.	23:00
	→ tpiep_ joined	23:01
vmiklos	marcot: what is the hash of the commit which is in the reflog but now shown by git-prune/fsck?	23:01
	← krh left	23:02
jbwiv	Ilari: difference between my scenario or vmilkos?	23:02
	← juanjoc left	23:02
Ilari	jbwiv: Difference of creating new branch off master and pulling and creating new branch out of remote branch.	23:02
	← bugfux left	23:03
marcot	9ffcd249dddcea7f7c8914bba4ec833aab0ab58c	23:03
	← Eridius left	23:03
jbwiv	Ilari: ok, and in my case they have indeed diverged	23:03
Ilari	jbwiv: Then you might want to merge them. If you created branch out of remote one, then on it 'git merge master'.	23:04
marcot	vmiklos: http://marcot.iaaeee.org/unreferenced.txt	23:04
jbwiv	Ilari: i see....ok, thanks!	23:05
	← mcella left	23:05
vmiklos	marcot: thx	23:05
	← seed__ left	23:05
	← jbwiv left	23:05
	← brain_in_vat left	23:07
vmiklos	marcot: try 'git fsck --unreachable --no-reflogs'	23:08
marcot	vmiklos: it showed some commits, but the unreferenced.txt file contains more..	23:08
	← tpiep left	23:09
marcot	Gotta go now.	23:10
	There are some commits not in my script but in git fsck --unreachable --no-reflogs, and the other way around two.	23:10
vmiklos	yes, it is interesting ;)	23:10
	→ sohum_ joined	23:11
marcot	vmiklos: http://marcot.iaaeee.org/fsck.txt	23:11
	Bye, I'll come back here latter.	23:11
	← marcot left	23:11
	← sohum left	23:12
	sohum_ → sohum	23:12
	→ ebzzry_ joined	23:13
	← berndj left	23:13
	→ mcella joined	23:14
	← ebzzry left	23:15
	→ EmilMedve joined	23:18
	← ebzzry_ left	23:20
	← Ryback_ left	23:21
	← ph^_ left	23:22
	← __name__ left	23:22
	← tpiep_ left	23:28
	← snitko left	23:29
	← aroben left	23:29
	← corybantic_ left	23:32
	← Sigma left	23:33
	← elmex left	23:33
	→ scientes joined	23:34
	← _graham_ left	23:34
	← fg3 left	23:34
	→ Eridius joined	23:35
	→ jpschewe joined	23:35
	← loincloth left	23:39
	→ _Vi joined	23:39
	→ wwwald joined	23:40
	→ loincloth joined	23:40
	← jpeeler left	23:45
	→ PhilK joined	23:45
jpschewe	I'm looking for some help with git-pull. I'd like to use git-pull to "push" changes from my local repository to the remote repository because the remote repository isn't accessible from my local repository and the remote repository is git-svn.	23:45
cbreak	log in on the remote end and pull?	23:46
	→ rolfb joined	23:46
jpschewe	Except the remote end can't access the local side.	23:46
Ilari	jpschewe: Ouch. Pull is fetch + merge, and merges and SVN don't mix... :->	23:46
cbreak	so neither can access the other	23:47
QBall	local cannot access remote and remote cannot access local	23:47
cbreak	how do you expect anything to work?	23:47
QBall	that is gonna be iffy	23:47
	→ lamont` joined	23:47
	← ceej left	23:47
cbreak	you could use an intermediate bare repo	23:47
Ilari	jpschewe: 'git bundle' ('sneakernet transport')?	23:47
jpschewe	local can access remote, but remote cannot access local.	23:47
cbreak	then you can push?	23:48
	or is remote non-bare?	23:48
jpschewe	But the repo isn't bare.	23:48
cbreak	make a bare one then :)	23:48
	← rolfb left	23:48
cbreak	or do some post-update hook magic hackery	23:48
Ilari	Or remap the ref name and complete the transfer from remote side...	23:48
	→ loki_val joined	23:48
	← intripoon left	23:48
jpschewe	ilari that's what I was thinking of, how does one do this?	23:49
loki_val	What's the config switch to make git look extra hard for copied files?	23:49
	← lamont left	23:49
Ilari	jpschewe: 'git push <URL-or-nick> <branch>:<remote-newbranch>' and then on remote 'git merge <remote-newbranch>'...	23:49
	→ intripoon joined	23:49
	→ ceej joined	23:50
jpschewe	will that work with git-svn being the remote repo too?	23:50
Ilari	jpschewe: Ouch... There's git-svn involved... Then you probably need to rebase your changes on version fetched from remote...	23:51
jpschewe	Hmm, I think I'll just copy teh files back.	23:52
	thanks.	23:52
	← wwwald left	23:53
	lamont` → lamont	23:53
	metze → metze_away	23:53
	→ yonkeltron joined	23:54
	→ jpeeler joined	23:55
Ilari	vmiklos: Regarding the 'strange untracked file behaviour', perhaps local branch and remote branch have diverged; They both tracked that file at last common ancestor, but both sides have dropped it. Now when merging, the remote set of changes includes deletion of that file. That could explain why it tries to remove untracked file.	23:55
	← mithro left	23:55
yonkeltron	can someone explain git packs to me? i've tried to figure it out on the docs and such but have had less success than i would have liked	23:57
	← cannonball left	23:57
	← loincloth left	23:57

Logs Search ←Prev date Next date→

Channels Documentation