IRCloggy #git 2009-09-28

Provider of IRC logs since 2005.

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

	2009-09-28
	← frede left	00:00
	← vvsh left	00:01
	→ Bass10 joined	00:03
	← Bass10 left	00:04
	→ g0bl1n joined	00:04
	→ Bass10 joined	00:05
	← j416 left	00:05
	← g0bl1n left	00:05
	→ krunk- joined	00:09
krunk-	can you dcommit a single directory? like git svn dcommit <directory_name> rather than a whole project?	00:09
	→ ajonat joined	00:10
offby1	doubt it	00:10
bremner	what does man git-svn-dcommit say?	00:10
krunk-	looking at the man	00:11
	→ jsquared joined	00:12
	→ warlock_mza joined	00:12
offby1	Hugh deMann	00:12
krunk-	I was a little unclear from it, but it seems to imply it's an all or nothing	00:13
engla	create a commit for only that subdirectory, and dcommit that; would the effect not be the same?	00:14
	→ Fragsworth joined	00:14
Fragsworth	I just deleted a file by accident, how do I restore it?	00:14
impulze	did you remove it without it being commited?	00:14
Fragsworth	yes	00:15
impulze	then study your filesystem documentation	00:15
Fragsworth	well, I want to restore it to the last commited state	00:15
impulze	nothing that git can take care of	00:15
	ah ok	00:15
	git checkout HEAD -- file	00:15
krunk-	engla: I'm workign on portfiles, I have 3 that are ready to be pushed and 2 that are still in progress. they all have multiple commit histories	00:15
impulze	git checkout <commitish> -- file (more generally)	00:15
	→ akume joined	00:17
	→ mrlemao joined	00:17
	← botanicus left	00:17
	← hashbangperl left	00:17
akume	hey i upgraded to the lastest git and i'm getting an error	00:17
	error: Protocol https not supported or disabled in libcurl while accessing https://secure2.svnrepository.com/g_ctiadmin/impressions//info/refs	00:17
	not sure what could be causing this	00:18
aeruder	akume: what platform?	00:18
akume	i'm on OSX	00:18
	→ nosatalian joined	00:18
akume	snow leopard	00:18
segher	the libcurl you used doesn't support https	00:18
impulze	what he said	00:18
akume	is that something that i have to upgrade?	00:19
	or a setting that i have to change?	00:19
segher	that's what the error says, at least, and i have no reason to doubt it :-)	00:19
akume	because i didn't have this problem before	00:19
	or is it a matter of me getting a different version of libcurl	00:19
	→ ^Einstein joined	00:20
	← magcius left	00:20
nosatalian	I have certain large chunks of related files (repos) which I do not need version history for, since they are monotonically increasing, but rsync is too slow, since it is stateless and must reconstruct all hashes each time. Is it possible to use git to have quick delta-propagation without having to use essentially double the disk space needlessly duplicating all the content in the .git/ dir? Basically what I want are semantics where	00:20
	blobs are sourced from the working directory (which is assumed/required not to be modified)	00:20
segher	akume: the curl that comes with osx supports https fine -- do you use macports perhaps?	00:21
akume	i did download macports around the time i was upgrading my git version	00:21
segher	get rid of it, unless you need it. or make sure it isn't used by your git build, anyway	00:22
	→ magcius joined	00:23
segher	macports has lots of compatibility problems with lots of packages, way more than the native osx things it replaces for exactly that reason :-P	00:23
	← tfarina left	00:23
akume	ok	00:25
	i'll try that	00:25
	← justin-george left	00:25
segher	good luck, let me know how it went :-)	00:25
	→ justin-george joined	00:27
	← rektide left	00:28
	→ eletuchy joined	00:29
	PuffTheMagic → CheebaHawk	00:29
akume	segher: that was it. a high five and a beer for you	00:31
segher	i'll be satisfied with the knowledge i helped a fellow human today. or, erm, wait, did you say beer? :-)	00:31
akume	=P	00:32
	← engla left	00:32
	← sergiodj left	00:33
	→ rektide joined	00:39
	→ sergiodj joined	00:41
	→ sdboyer_ joined	00:41
	← intripoon left	00:42
	→ cxvii joined	00:43
	→ adamtanner joined	00:45
	← magcius left	00:45
	→ madewokherd joined	00:45
	→ magcius joined	00:46
	← adamtanner left	00:52
	← sdboyer left	00:53
	← akume left	00:54
	← cxvii left	01:02
	← freerobby left	01:14
	← marvil07 left	01:17
	← corecode left	01:18
	← markelikalderon left	01:18
	← pecc left	01:22
	→ nothingHappens joined	01:23
	→ abms1116 joined	01:23
	→ VVSiz_ joined	01:24
	← VVSiz left	01:29
	← mrlemao left	01:32
	← Bass10 left	01:34
	→ freerobby joined	01:34
	→ patcito joined	01:39
	→ j416 joined	01:45
	→ jrmuizel joined	01:46
JakeSays	how can i figure out which git revision a branch was created on?	01:47
offby1	I doubt there's any wya.	01:48
	way.	01:48
cehteh	you cant (at least not reliably), branches only track heads	01:48
JakeSays	hmm.	01:48
offby1	Do you mean "which version of git did the guy use when he ran 'git branch'"?	01:48
	or are you talking about git's own git repository?	01:48
JakeSays	offby1: uh.. i'm assuming i mean the revision # of HEAD when the branch was created	01:50
	i think thats what i mean anyway	01:50
cehteh	thats nowhere recorded	01:51
	you can look at the log and try to figure out when it branched away, not very accurate, merges make this unreliable	01:51
	for what do you need this information?	01:52
	← mnick left	01:53
JakeSays	i'm working with someone who is trying to commit to svn a git branch so i can get to it (i only have access to their external svn server)	01:53
cehteh	i have no idea about svn, but svn branches and git branches are certainly quite different kinds of things	01:53
JakeSays	yeah i know. the whole thing is stupid if you ask me. wish they'd just open up their git repo.	01:54
cehteh	heh they use git and dont serve it?	01:54
	cehteh agrees, thats stupid	01:55
JakeSays	they use it internally	01:55
cehteh	thats really wtf :)	01:55
	→ Bass10 joined	01:56
kevlarman	JakeSays: why do you need the point where it branched for that though?	01:56
jjuran	"trying to commit to svn a git branch so i can get to it"	01:56
JakeSays	kevlarman: i have no idea.	01:56
	← aziz left	01:57
kevlarman	git log -1 --grep=git-svn-id would probably give you what you're asking for	01:57
jjuran	That's like displaying it in the bottom drawer of a locked filing cabinet in a disused lavatory in the cellar so you can get to it. :-P	01:57
	← Bass10 left	01:57
kevlarman	but it still makes no sense why you're asking for it	01:57
	→ Bass10 joined	01:58
jjuran	Seriously, I'd suggest trying to get access to the Git repo.	01:59
	← freerobby left	02:00
	→ gko joined	02:04
	→ bcardarella joined	02:04
JakeSays	oh geez. ok. they only have an svn repo, with each dev having a local git repo	02:07
offby1	weird	02:07
JakeSays	yeah	02:07
rapha	oh my god	02:07
offby1	I bet some silly corporate IT rule prevents them from ditching the svn repo	02:07
rapha	please tell me "git checkout ." only gets missing files and does not overwrite others	02:07
offby1	dunno,Try It And See™.	02:07
aeruder	rapha: i think i have bad news for you	02:08
offby1	easy enough test to run	02:08
aeruder	well, i think...	02:08
	actually, i think that one requires a -f to overwrite	02:08
	→ Pewpewarrows joined	02:08
rapha	aeruder: in that case it definately does not require an -f	02:08
	this is dangerous	02:08
	i need to strangle the appropriate developer	02:08
	argl	02:09
	and it didnt even tell me which ones it overwrote!!!	02:09
JakeSays	offby1: yeah, i bet you're right.	02:09
aeruder	rapha: yea, that is	02:09
rapha	HAHAH! I set vim to have backups in ~/.backups this morning!	02:10
	→ hashbangperl joined	02:11
	→ ben_h joined	02:13
aeruder	rapha: hm, i almost think it should require -f there, pretty dangerous	02:15
	→ montylounge joined	02:19
	→ cytrinox_ joined	02:20
	← bcardarella left	02:22
	← ajonat left	02:27
	→ upul joined	02:30
	sdboyer_ → sdboyer	02:31
	← warlock_mza left	02:36
	← cytrinox left	02:37
	cytrinox_ → cytrinox	02:37
	→ tro\|\| joined	02:38
	← rudi_s_ left	02:40
	→ Aszarsha_ joined	02:42
	← jaysern_ left	02:43
	→ jlf` joined	02:44
	→ cqpx joined	02:45
	→ aspotashev_ joined	02:48
	← tro left	02:49
	tro\|\| → tro	02:49
	← eletuchy left	02:51
	← scarabx left	02:52
	→ tjafk1 joined	02:52
	← xckpd7 left	02:55
	← Aszarsha left	02:59
	← tro left	03:00
	→ tro joined	03:00
	→ bcardarella joined	03:01
	→ amerine joined	03:01
	← aspotashev left	03:03
	← hugoxrosa left	03:04
	← coolcat left	03:04
	← montylounge left	03:05
	Aszarsha_ → Aszarsha	03:07
	← voker57 left	03:08
	Aszarsha → Aszarsha_	03:08
	Aszarsha_ → Aszarsha	03:08
	→ voker57 joined	03:08
	← tjafk left	03:09
	← mw__ left	03:12
	← abms1116 left	03:15
	← pheaver left	03:16
	← goncalo left	03:20
	← tro left	03:22
	→ tro joined	03:23
	→ j2labs joined	03:23
j2labs	I just cloned a project that has multiple branches, but "git branch" doesn't show them. How can I use them?	03:23
aeruder	j2labs: git branch shows your "local branches"	03:24
	git branch -a shows all, git branch -r just shows remote tracking branches	03:24
j2labs	aeruder: ah, great. thanks	03:24
	← j416 left	03:25
	→ qhoxie joined	03:30
j2labs	if i cloned a project on github, my project is seen as origin, right? if so, what is the project i forked referred to as?	03:31
aeruder	well, you didn't clone that, you cloned your project...	03:32
	so what you'd want to here is setup a new remote	03:32
	git remote add nonforked <url>	03:33
	git fetch nonforked	03:33
	and now you'd have nonforked/* and origin/*	03:33
sitaram	j2labs: I think thats why github uses the word "fork" for that action. So lets say you fork my project gitolite, github/j2labs/gitolite would be a fork of github/sitaramc/gitolite, and then you clone yours. You could clone mine and not fork, if you didnt want to push anything back or maintain some differences	03:34
	(I do that for many projects where I just want to use it)	03:34
j2labs	ok. the advice from both makes sense and i've now merged the official branch for a project into mine	03:35
	thanks! git is really neat in theory but i don't have it in practice yt.	03:35
aeruder	j2labs: gitk is good for figuring things out visually, you can do a gitk --all and see how everything relates to each other	03:35
	→ pheaver joined	03:35
j2labs	is github having issues right now? i'm getting errors just from cloning. i know they have a big move today	03:37
	aeruder: thanks for the suggestion about gitk	03:39
	← j2labs left	03:42
sitaram	j2labs: github is probably having some issues -- at the best of times they're a little slow; hopefully this move will make things better	03:43
	→ eletuchy joined	03:45
	→ SaCu_ joined	03:45
	← CheebaHawk left	03:46
schacon	j2labs: what is the error?	03:47
	oh, they asked the question and took off	03:47
	nevermind, then	03:47
sitaram	oh hi scott -- is the move done?	03:52
	did it go well? I got a couple of "bad gateway" type errors looking at something... retry worked so I didnt bother asking	03:53
	← dreiss left	03:53
	sitaram is not even sure what "bad gateway" means... caching proxy malfunction?	03:54
	→ gaveen joined	03:54
	→ eno__ joined	03:55
	← eno left	03:55
	← voker57 left	04:00
	← SaCu left	04:00
	→ voker57 joined	04:00
	← bsdvodsky left	04:01
	→ bsdvodsky joined	04:04
	← jspiros left	04:04
	→ raichoo joined	04:06
TheUni	am i correct in assuming that a git svn fetch of a remote repo, and that same repo but copied and cloned locally... different git id hashes will be generated?	04:07
	that seems to be what i'm observing	04:07
	← jrmuizel left	04:11
	← djszapi left	04:11
	qrush\|away → qrush_	04:13
kevlarman	TheUni: afaik they should be the same	04:13
	→ Theravadan joined	04:13
TheUni	kevlarman: from what i've seen they're not. and from what i can tell git-hash-object uses path as a filter. if that's the case, i'm wondering if there's a way to spoof the path for a git svn clone	04:14
	← cqpx left	04:14
	← qrush left	04:15
TheUni	problem is, after cloning from my local svn repo and pushing, a resulting clone from that and git svn fetch fails, and wants to redownload all history	04:15
	→ jason237 joined	04:15
	→ cqpx joined	04:15
kevlarman	that's different	04:15
	i've never seen it not want to redownload all history (i've tried)	04:16
	but the resulting hashes after you do seem to match up	04:16
TheUni	kevlarman: i've managed in the past i think	04:16
aeruder	that's due to the clone not copying all the metadata in .git/svn i think	04:16
TheUni	aeruder: understandable, but it's still in the commit messages, and it regens SOME of the history ok	04:17
	so i am assuming it's an error on my part	04:17
kevlarman	there's something in the git-svn man page about how to do that	04:17
	→ jrmuizel joined	04:19
TheUni	ah, rewriteRoot is what i'm looking for	04:21
	thanks kevlarman	04:23
	→ hyperair joined	04:25
	→ roop joined	04:26
jason237	is there any way to query what remote branch corresponds to the refspec that git pull will merge with?	04:26
	aeruder wishes people would just stop using git-pull and fetch+merge	04:26
aeruder	you can always just do the git pull and then reset --hard it away	04:27
jason237	I don't think that would help me; I would like that for things like git log -p remote.. or git rebase -i remote	04:29
aeruder	jason237: um, then i think you aren't wanting git pull	04:30
	and you just want git fetch	04:30
	which is pull without the merging it into your branch step	04:31
	i.e. it updates your remote tracking branches	04:31
jason237	I don't want to fetch at all	04:31
aeruder	why not?	04:31
jason237	I just want to know which remote branch my current branch is based on	04:31
	well, I want to fetch sometimes	04:31
	but I also want to be able to look at my local changes relative to the appropriate remote branch	04:31
	← jrmuizel left	04:32
jason237	and git pull knows which remote branch that is	04:32
kevlarman	jason237: branch.<branchname>.remote and branch.<branchname>.merge	04:32
jason237	kevlarman: well, yes, that's what it's based on, but it's non-trivial to map that onto a refs/remotes/ name	04:33
	→ jrmuizel joined	04:33
aeruder	jason237: i think the answer is that there isn't	04:34
	i wouldn't be surprised if they'd accept a patch to fix that	04:34
	but right now i think you gotta use kevlarman's approach (chop /refs/heads off of remote.<branchname>.merge and branch.<branchname>.remote)	04:35
jason237	ok, maybe I'll poke at that	04:35
	rkirti[zZz] → rkirti	04:35
	← qhoxie left	04:36
	→ jspiros joined	04:37
	← raichoo left	04:37
	← ereslibre left	04:38
	→ tedoc2000 joined	04:39
	→ jceb joined	04:39
	→ dreiss joined	04:42
	← segher left	04:46
	← amerine left	04:47
	← meanburrito920_ left	04:50
	← tedoc2000 left	04:51
	jlf` → jlf_away	04:57
	→ imranh joined	04:57
	← imranh left	05:00
	→ imranh joined	05:05
	→ Cesario joined	05:06
	→ octal joined	05:07
	← imranh left	05:08
	← wolog left	05:15
	← Pewpewarrows left	05:16
	→ amerine joined	05:16
	← jrmuizel left	05:17
	← Rolenun left	05:17
	→ krawek joined	05:17
	→ jrmuizel joined	05:21
	← jrmuizel left	05:21
	→ ajonat joined	05:22
	→ Tuomas joined	05:23
	← pheaver left	05:26
	→ pheaver joined	05:26
	← giallu left	05:26
	→ dirk2 joined	05:28
	← Bass10 left	05:28
	← flippo left	05:28
	qrush_ → qrush\|away	05:38
	→ fualo joined	05:38
	← marc_m left	05:39
fualo	How can I start tracking a remote branch if I didn't clone it initially?	05:39
	can I just add it to .git/config	05:40
	→ thiago_home joined	05:41
	← hyperair left	05:42
aeruder	fualo: git fetch should pick it up....	05:43
	git branch -r	05:43
	← onigiri left	05:45
	← slackstation left	05:46
fualo	thanks aeruder	05:51
wereHamster	aeruder: there's thn %{upstream} format	05:51
fualo	is there also a way to make a patch for a few line numbers?	05:51
	← fualo left	05:52
	← justin-george left	05:54
	→ marvil07 joined	05:57
wereHamster	aeruder: git for-each-ref --format='%(upstream)' $(git rev-parse --symbolic-full-name HEAD)	05:57
	→ fualo joined	05:59
	→ qhoxie joined	05:59
	← qhoxie left	06:00
	← jceb left	06:01
	← Guest67226 left	06:11
fualo	What's the most git way to take a chunk of a function from one branch, and move it to another branch	06:11
sitaram	vim	06:12
fualo	sitaram: copy and paste?	06:12
sitaram	seriously; git has nothing to do with until after you do it :)	06:12
fualo	sitaram: well I thought a selective patch	06:13
	or something	06:13
sitaram	but then, git is the only VCS that will retroactively figure out you refactored that code and show you the history from its previsou life on another planet	06:13
	er.. file	06:13
	now if you'd said "commit" that would be cherry pick	06:13
fualo	can I cherry pick across branches?	06:13
sitaram	sure	06:14
fualo	and for a single file?	06:14
sitaram	you could do it with format-patch, (or git diff), edit the patch file, and then apply it to the other branch, but thats not a lot better than yank/put in vim (or whatever is your poison in terms of text editors)	06:15
fualo	emacs	06:15
	ok	06:15
	copy paste itis	06:15
	:-)	06:15
	→ ntoll joined	06:16
sitaram	fualo: see the thing is if its's not the entire commit, you have to use an editor somewhere, for best results :)	06:16
fualo	yea, true	06:16
	I see	06:16
	→ cao joined	06:17
	cao → Guest44078	06:17
	← Arild left	06:18
	→ priidu joined	06:19
	← delsvr left	06:20
	→ jin joined	06:22
	← bcardarella left	06:23
	← Thell left	06:26
	← loincloth left	06:31
	VVSiz_ → VVSiz	06:32
	→ mastro joined	06:35
	← ^Einstein left	06:35
	→ bousquet joined	06:36
	→ bentob0x joined	06:37
	← amerine left	06:40
	→ LiHong joined	06:41
	← LiHong left	06:41
	jin → jin72	06:41
	→ LiHong joined	06:41
	→ kprav33n joined	06:43
	← dreiss left	06:43
kprav33n	I have a personal git repository of org files that I share between computers	06:43
	By org, I mean Emacs org-mode files.	06:43
	I want to set the file permission 0600 on these files so that other users on the machine don't read my personal notes.	06:44
	Is there a way to change these file permissions in the repository?	06:45
	I changed the file permissions locally, but they are lost when I do a git pull.	06:45
	→ Unregistered joined	06:45
sitaram	git doesn't track or keep permissions except the first "x" in rwxrwxrwx	06:49
	→ frede joined	06:49
sitaram	you'll have to do it externally	06:49
	→ PointMan joined	06:50
Unregistered	http://bit.ly/26ax7L ->if possible need help	06:50
sitaram	maybe look at the post-merge hook (but it's not portable)	06:50
	→ ^Einstein joined	06:50
sitaram	kprav33n: and why is your $HOME readable anyway... seems to me you should stop people right there :)	06:51
Unregistered	http://bit.ly/26ax7L ->if possible need help	06:51
	← Unregistered left	06:51
kprav33n	sitaram: I have a lot of stuff in my home directory that I share with others	06:51
	← PointMan left	06:51
	→ ph^ joined	06:52
	← aspotashev_ left	06:52
sitaram	^Einstein: hi... got a minute?	06:53
^Einstein	sitaram: sure!	06:53
sitaram	^Einstein: you can download the latest master of gitolite; it will give better error messages when you dont match up the pubkeys and the config file properly: see commit http://github.com/sitaramc/gitolite/commit/c66e1ad73286e1a193373234142daab44e5f7dd6	06:53
	→ cedricv joined	06:54
^Einstein	Cool, I'll git pull right now :)	06:54
fualo	when I do git push origin master, is git pushing the current checkout branch to master, or local master --> master?	06:54
^Einstein	Thanks, sitaram.	06:54
	← disappearedng left	06:55
	← dirk2 left	06:55
	→ johan-s joined	06:56
sitaram	you're welcome!	06:56
	← octal left	06:57
jlf_away	hi #git, what's the best way to host a repository on my machine and allow a few other people to push changes into it without giving them shells on my machine?	06:57
	→ tim_jenssen joined	06:57
sitaram	gitolite	06:58
wereHamster	gitolite	06:58
	← ph^ left	06:58
sitaram	jinx!	06:58
	http://github.com/sitaramc/gitolite	06:58
jjuran	kprav33n: You can set the umask before running git checkout, but that affect's all created files.	06:58
	affects	06:58
	→ dpino joined	06:59
	→ ph^ joined	06:59
jlf_away	thanks gents, i'll have a look	07:00
	← agenteo left	07:02
	→ roop_ joined	07:02
	→ Unregistered joined	07:03
Unregistered	http://bit.ly/Tp2F1 Is this good?need advice from pros	07:03
	← Unregistered left	07:03
	roop → Guest31146	07:04
	roop_ → roop	07:04
	→ giallu joined	07:04
kprav33n	jjuran: Is it possible to set this somewhere in the config file?	07:05
	I don't want to do this everytime.	07:05
	I saw the man page of git config	07:05
	→ j416 joined	07:06
kprav33n	I found this variable, core.sharedRepository and I have set it to 0600.	07:06
	But that doesn't help.	07:06
sitaram	kprav33n: no but you can set an alias that does both and use that instead of your regular push I guess; the alias page in the git wiki has a few examples	07:06
kprav33n	sitaram: Okay, thanks!	07:07
sitaram	kprav33n: and you may get a bonus that the umask will only apply within that function (test it...) thus not affecting the rest of your work	07:09
kprav33n	$ git config core.sharedRepository 0600	07:10
	That solves it.	07:10
	Thanks, everyone!	07:11
	→ freezed joined	07:12
	→ ciscbrain joined	07:12
freezed	Hello	07:12
kprav33n	I thought of a bash function first. But I wanted to see if there is a git config item and well, there was one!	07:13
	← kprav33n left	07:13
freezed	If I want to manage multiple projects with redmine or trac for example, should I create I repository and a new directory for each project OR one repository for each projects ?	07:14
sitaram	kpreid_: check if that affects only the files under ".git" (the actual repo) or the working tree as well...	07:14
	kprav33n	07:15
	(sorry kpreid_ ! bad tab completion)	07:15
	← Guest31146 left	07:16
^Einstein	Does anyone here have experience with either the hg-git plugin for mercurial or the deprecated hg-to-git.py script, or know of another way to inherit/convert and push hg changesets accurately?	07:17
	← schacon left	07:18
	→ gsuveg joined	07:19
gsuveg	re	07:19
^Einstein	hg gpush fails for me no matter what I try - I think its use of ssh on the client side is improper, but not sure - and for whatever reason I lose some of my history, or it is very skewed, with hg-to-py	07:19
	er, hg-to-git.py*	07:19
gsuveg	ist have a git server via ssh, and ive local 2 branch. its possible to send all branches to server ? how?	07:20
	→ wolog joined	07:22
sitaram	git push remotename --all	07:22
	→ galderz joined	07:23
	→ mattcodes joined	07:24
gsuveg	sitaram: thanks!	07:24
	← bousquet left	07:25
mattcodes	okay two directories c:\gitexp\test1 and c:\gitexp\test2. Test2 is a clone of test1. I make a change to story.txt in test2, commit it, and then git push origin master.	07:25
	i change directory to c:\test1	07:25
	what do I need (or should do) to see the change in test1	07:25
	as opening story in test does not have the changes I pushed from test2	07:25
	← hobodave left	07:26
mattcodes	hmmm "git checkout -f" seems to have worked and now shows updated story.txt in test1, but I dont Im following the right practice? I think i might be missing something...	07:27
Mikachu	mattcodes: faq non-bare	07:27
Gitbot	mattcodes: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare	07:27
freezed	what is a bare repository ?	07:28
j416	freezed: a repo where there is no .git dir, but instead the contents of .git is in the main dir itself, and there is no working copy.	07:29
	used for pushing to.	07:29
	→ tbf joined	07:30
	fpcfan-away → fpcfan	07:30
kupsch	I want to track/checkout a single file from a remote repository. How can I achieve that?	07:32
mattcodes	hmmm read got ya (I think)... removing simplification. in my example I have c:\workspaces\CRMProject and c:\workspaces\MattCommons I have on the CRMProject done "git submodule add MattCommonsPath ThirdParty\MattCommons" to bring the commons in to the proejct. in theory i dont need a working copy of MattCommons on its own do I? just bring it in and send it back and forth as commons is never deployed alone... I think I get the idea..	07:32
	(thinking out loud)	07:32
	→ kent\n joined	07:33
drizzd_	kupsch: you can do that using git-archive --remote	07:33
	kupsch: otherwise $faq clone-sudir	07:33
	kupsch: faq clone-subdir	07:33
Gitbot	kupsch: Git cannot clone subdirectories. See http://git.or.cz/gitwiki/GitFaq#clone-subdir	07:33
kupsch	well it's not a subdir, it's just a single file	07:33
drizzd_	This includes cloning individual files, of course	07:34
kupsch	k, thanks	07:34
gsuveg	bye	07:38
	← gsuveg left	07:38
	→ jds joined	07:42
	← marvil07 left	07:42
freezed	j416: thanks	07:47
	← squentin left	07:47
	→ PointMan joined	07:48
	→ [Po]lentino joined	07:48
	← mattcodes left	07:49
	← pheaver left	07:52
	→ SRabbelier joined	07:54
	→ ixti joined	07:54
ixti	Hello everyone	07:54
	can someone help me a little?	07:54
	what is the difference between: `git reset --hard` and `git checkout -R .`	07:55
teuf	ixti: just ask your questions, you'll see if someone answers or not :)	07:55
	ah sorry	07:55
ixti	:))	07:55
	git-checkout without -R	07:56
sitaram	git checkout -R??? cant find it in the manpage...	07:56
ixti	don't know why i placed it :))	07:56
	:))	07:56
	sorry, seems like, i need to have a little coffee	07:56
	:))	07:56
	so the final version of my question is: what is the difference between: `git reset --hard` and `git checkout .`	07:57
sitaram	wait	07:57
Mikachu	ixti: without arguments, not much	07:58
sitaram	ixti: http://permalink.gmane.org/gmane.comp.version-control.git/104457	07:58
freezed	Have a nice day	08:00
ixti	Mikachu: sitaram: thank you	08:00
	→ mattcodes joined	08:00
	← freezed left	08:00
mattcodes	when in a super project that has various sub projects using git submodules, if there a way to call Update on the superproject and the sub projects update as well, ala SVN Update on a super project with svn:externals, I guess its probably a sort of recursive git pull?	08:01
	→ icwiener joined	08:05
	→ shruggar joined	08:07
	→ intripoon joined	08:07
	→ lucsky joined	08:08
drizzd_	mattcodes: my git version has a --recursive option	08:10
	mattcodes: apparently it's a new feature in 1.6.5-rc0	08:12
mattcodes	ah thanks, im running windows for .net dev at the moment so I guess I'll wait, no probs, good to know its on its way	08:13
fualo	are there any git plotters? i.e. size/lines per day?	08:13
Adlai	fualo: github has an "impact" graph	08:13
drizzd_	fualo: there are lots of those. maybe you can find some in the wiki	08:13
fualo	ok, Ill check the wiki	08:14
	← JakeSays left	08:15
	← LRN left	08:15
	→ JakeSays joined	08:15
	→ jinzhu joined	08:15
	← ajonat left	08:16
	← fualo left	08:17
	→ pheaver joined	08:17
	← patcito left	08:19
	← Dashkal left	08:19
	→ solydzajs joined	08:19
	← pheaver left	08:21
	← dancor left	08:21
	← mattcodes left	08:26
	← wolog left	08:29
	→ mnick joined	08:31
	→ warthog9 joined	08:33
	→ abms1116 joined	08:33
	→ fdv joined	08:34
	→ struberg joined	08:34
	→ adeodatus joined	08:37
	← fdv left	08:37
	→ maattd_ joined	08:40
	→ loops joined	08:43
	← lucsky left	08:44
	← krawek left	08:45
	→ gitte joined	08:46
	← adeodatus left	08:47
	→ lucsky joined	08:47
	→ aatanasov joined	08:49
	→ romfert joined	08:51
	→ engla joined	08:51
	← maattd left	08:55
	maattd_ → maattd	08:55
	← ia left	08:56
	← abms1116 left	08:57
	→ noginn joined	08:59
	← j416 left	09:01
	← patmaddox left	09:01
	→ abms1116 joined	09:04
	→ sergeykish joined	09:05
	← bsdvodsky left	09:05
	→ ArdRigh joined	09:06
	→ ia joined	09:08
sergeykish	Hello, I need a way to list files by status - staged, modified, untraced. I've found `git ls-files -m` for modified and `git ls-files --exclude standard -o` for untraced. How can I list cached?	09:10
	→ tbuehlmann joined	09:10
sitaram	git ls-files without any options is supposed to do just that, I think	09:10
Pieter	git diff-index	09:11
sergeykish	* files added to index	09:11
Pieter	git diff-index --cached HEAD	09:12
sergeykish	Pieter, thanks `git diff-index --name-only HEAD` looks good	09:14
	→ ccherret1 joined	09:16
	← ccherrett left	09:16
	ccherret1 → ccherrett	09:17
	→ jinzhu_ joined	09:18
	← cqpx left	09:21
	← gretch_ left	09:22
	← struberg left	09:23
	← ccherrett left	09:24
	→ JohnFlux joined	09:26
JohnFlux	Doing "git status" locks the repository. Does it really need to lock the repository?	09:27
	→ davido joined	09:27
	→ gretch joined	09:27
	→ markelikalderon joined	09:28
	→ ccherrett joined	09:28
	→ relix joined	09:29
	← tono left	09:30
sergeykish	`git diff-index HEAD` show items even if it is not added to index, I need `git diff --cached` filelist	09:31
	Sorry, found --cached flag :)	09:31
	← impulze left	09:34
	← voker57 left	09:34
	→ voker57 joined	09:34
	→ j416 joined	09:34
	→ impulze joined	09:35
	← impulze left	09:36
j416	does git fully support unicode filenames?	09:36
engla	JohnFlux: what do you mean with locks?	09:36
	j416: git treats filenames as byte strings, you can encode them however you want	09:37
	→ impulze joined	09:37
j416	engla: ok! thank you.	09:37
engla	that's the principle at least	09:37
dbrock	can I unpack all refs somehow?	09:37
j416	seems to work, the only thing is that it displays my utf-8 text as escape sequences	09:37
	I wonder why.... hm.	09:37
JohnFlux	j416: so in other words, you have to guess what a repository has encoded its filenames as	09:37
j416	ok	09:37
charon	some commands deliberately escape filenames in the output	09:38
JohnFlux	j416: and if its anything except utf8, it will not work when you check it out in windows	09:38
Ilari	j416: On Mac OS X, IIRC, you need to use UFS as filesystem (I don't recall wheither ZFS works, HFS+ definitely doesn't work).	09:38
j416	oh	09:38
engla	Ilari: really?	09:38
	← ixti left	09:38
j416	Ilari: I'm using HFS+	09:38
JohnFlux	j416: Windows version git assumes that git is using utf8	09:38
j416	why doesn't it work?	09:38
	JohnFlux: good.	09:38
	good to know	09:38
engla	well there is an issue due to OS X normalization of unicode	09:38
j416	oh	09:39
engla	I thought it was resolved	09:39
j416	normalization?	09:39
Ilari	HFS+ corrupts unicode (at least characters like ä and ö and others like those).	09:39
engla	well git treats it as byte strings	09:39
	Ilari: normalizes, not corrupts	09:39
j416	Ilari: corrupts?	09:39
	in what way?	09:39
engla	j416: unicode can have different representations	09:39
j416	yes	09:39
engla	j416: but there are canonical and non-canonical composition of characters, such as ö	09:39
j416	non-canonical composition of 'ö'? how can that be	09:40
	i thought that was clearly defined	09:40
engla	j416: and OS X always normalizes using one of the standard algorithms	09:40
j416	'¨' + 'o', or 'ö'	09:40
	→ tono joined	09:40
j416	ok	09:40
Ilari	HFS+ uses NFD-like form (it isn't even standard NFD). Would cause massive interoperability problems.	09:40
engla	Python tells me this	09:40
j416	hmm	09:41
engla	ö is '\xf6', but the NFD form is 'o\u0308'	09:41
charon	engla: well, i do call it a corruption on the grounds that it "normalized" an umlaut on our local SVN (i know...) to the NFD, resulting in two different directories both called "übungen"... except at the byte level of course.	09:41
engla	that is an o with some combining diaresis	09:41
	charon: yes, it's corruption in the Linus Torvalds and the UNIX perspective	09:41
j416	crazy	09:41
	← markelikalderon left	09:41
j416	I wonder what that would be good for..	09:42
	the file system should accept the name I give to a file and not mess with it. Scary..	09:42
engla	j416: on OS X, you can guarantee that if two folders are independently named ö, they have the same encoding	09:42
	→ kraymer joined	09:42
engla	in the name	09:42
	j416: well then you have the UNIX philosophy that git uses	09:43
charon	engla: where the "unix perspective" is that it should make communication with other systems possible? :-) </troll>	09:43
	← _Vi left	09:43
engla	well no	09:43
j416	engla: yeah... either that, or Unicode have to fix their definition	09:43
Ilari	NFD/NFC insensitive but preserving would be one thing. Corrupting the stuff is whole another.	09:43
j416	s/have/has/	09:43
engla	the unix perspective is the least influence possible so it's the lowest level	09:43
	you can have a mix of filesystem encodings, it is not defined	09:44
	it leads to other problems	09:44
	← jinzhu left	09:44
engla	j416: well NFD is explicitly a decomposition, not much to fix there. they should have used NFC at least	09:44
	linux must always be incompatible since it accepts anything as a file name	09:45
j416	I guess.	09:45
engla	Windows? Requires UTF-8?	09:45
j416	lucky CJK doesn't have these problems	09:45
	no umlauts..	09:46
	→ jmibanez joined	09:46
j416	(why isn't umlaut called ümlaut?)	09:46
engla	ok. but you can decompose to radicals or something?	09:46
j416	engla: korean yes, japanese/chinese, no such mapping afaik.	09:46
	← ryoma left	09:47
	→ Yuuhi joined	09:47
	→ T2000-lap joined	09:47
	→ ixti joined	09:47
Ilari	I have table of fixups to do for "NFD"... 925 entries in that table.	09:47
engla	plural for umlaut must be umläute(?) :-)	09:47
	→ ikrabbe joined	09:48
T2000-lap	engla: no its "Umlaute"	09:48
engla	thanks	09:48
charon	ungeläutert ;-)	09:48
	→ psoo joined	09:48
Ilari	Fortunately, all those fixups are non-expanding in UTF-8.	09:49
ikrabbe	Hi, is there somewhere a comprehensive definition of all those short-names used in the git manual like: <remote-ref>, <refspec>, <ommit-ish> etc.	09:49
jjuran	I don't speak German, but 'u' and 'ü' are different vowels.	09:49
charon	ikrabbe: man gitglossary?	09:49
Ilari	ikrabbe: man gitglossary?	09:49
Gitbot	ikrabbe: the gitglossary manpage can be found at http://git.or.cz/man/gitglossary	09:49
	engla will nicht sein deutsch vergessen	09:49
jjuran	j416: ^	09:49
j416	jjuran: it was a joke :P	09:49
	(and besides, wouldn't it make more sense to call it ümlaut to clarify its meaning? :)	09:50
jjuran	But then you'd pronounce it (nearly) 'eemlaut', which sounds funny. :-P	09:51
j416	eemlaut.. o_O	09:51
	no	09:51
	ümlaut. :)	09:51
	→ LRN joined	09:52
j416	if it was ümlaut from the start no one would have thought it sounded funny, though. :P	09:52
jjuran	I'm told that ü and ie are considered close enough to rhyme.	09:52
	→ kukks joined	09:52
j416	ü and 'ie' ?	09:52
charon	i wanted to spell it with a diacrisis on the 'm', but apparently that's not a character... sigh	09:53
jjuran	As in Brüder and diesen	09:53
j416	ah	09:53
charon	er, diaeresis.	09:53
ikrabbe	its enough to rhyme, yes	09:53
Ilari	charon: Unicode combining diacrisis?	09:53
j416	very lazy rhyme.	09:53
charon	Ilari: dunno, i just tried to spell it with the deadkey for umlaut and it refuses to put it on an 'm'	09:54
j416	charon: I don't think m with umlaut is proper, is it?	09:54
	does unicode define ¨+m ?	09:54
jjuran	The more umlauts, the more you rock.	09:55
j416	o_O	09:55
Ilari	m̈	09:55
ikrabbe	an umlaut is a vocal by (laut->vocal)	09:55
	← ArdRigh left	09:55
j416	ooh.	09:55
charon	j416: of course there's no m-umlaut, but then it wouldn't be funny	09:55
j416	:D	09:55
charon	so lets call them um̈lauts	09:55
	→ Ozzy_ joined	09:55
j416	hahahaha	09:55
charon	(thanks Ilari)	09:55
j416	um̈laut	09:56
	← mapreduce left	09:56
j416	well well	09:56
	sorry for bringing up this OT. :)	09:56
	← Aszarsha left	09:56
j416	</OT>	09:56
	← Ozzy_ left	09:56
j416	thanks for the info.	09:56
jjuran	O Freunde, nicht diese Töne!	09:57
	→ Ozzy_ joined	09:58
engla	ikrabbe: laut <=> vocal? I would have thought laut = phoneme	09:58
j416	entschuldigung!	09:58
jjuran	No problemo	10:00
j416	ah-ri-ga-toh	10:00
	← Ozzy_ left	10:00
ikrabbe	engla, you're right, I just guessed laut<->Vokal<->vocal, but Vokal<->vowel and laut<->phoneme, though all those words are connected together quite tight I think.	10:01
	← m-takagi left	10:02
tbuehlmann	I have an umlaut in my name!	10:03
charon	JohnFlux: about the locks... did you try with a git from the next branch? it has a new git-status that is no longer 'git commit --dry-run'	10:03
	→ wolog joined	10:04
JohnFlux	charon: ah cool	10:04
	← radii left	10:04
charon	so if it doesn't lock everything there, good; if it does, at least we can fix it in the right place	10:05
	→ MisterN joined	10:05
	→ radii joined	10:06
^Einstein	the all-mighty Google has provided me very little with regard to using git with mercurial repositories. The hg-git plugin for mercurial isn't authing to ssh properly, and git-to-hg.py (a standalone script to convert a .hg into a .git) seems to not be functioning properly. Any ideas?	10:12
	→ _Vi joined	10:13
	← jmibanez left	10:14
	→ khinester joined	10:14
khinester	hello	10:14
	→ sde_antrix joined	10:15
	← j416 left	10:18
	→ hyperair joined	10:18
khinester	i am using git as a datastore for an application built in python. is there a method in which i can have a persistant sync between two remote repositories?	10:18
	← radii left	10:18
	→ benjo2 joined	10:18
	← JackWinter left	10:20
	← gko left	10:21
	← lu_zero left	10:21
	← nothingHappens left	10:22
	→ benc1 joined	10:23
benc1	how do I show what branch I'm currently on?	10:23
	← LiHong left	10:23
teuf	benc1: git branch	10:23
	→ WinterWeaver joined	10:24
	→ JackWinter joined	10:24
benc1	teuf: thanks	10:24
	→ trochala joined	10:25
trochala	Hello, how can I grep commit messages?	10:25
JohnFlux	trochala: git log \| grep blah ?	10:26
	benc1: much better solution is to modify your bash prompt to always tell you what branch you are on	10:26
engla	trochala: git log --grep=	10:27
JohnFlux	benc1: shall I tel you how?	10:27
	engla: what was wrong with what I said/	10:27
	engla: I mean, is there a difference?	10:27
engla	JohnFlux: perhaps nothing in practice. But I suppose git log's version is going to be more efficient	10:27
Adlai	JohnFlux: git log --grep probably searches as it goes	10:27
trochala	engla: thanks :)	10:28
benc1	JohnFlux: sure	10:28
doener	JohnFlux: your command shows lines from the commit messages that contain blah	10:28
benc1	how do I delete the recent commit?	10:28
JohnFlux	doener: ah yes of course	10:28
doener	JohnFlux: the log command shows commits that contain 'blah' somewhere in the commit message	10:28
	JohnFlux: I'd dare to say that the results of your command are somewhat useless ;-)	10:28
Adlai	doener: ah true, that's the real difference	10:28
JohnFlux	doener: yeah	10:28
benc1	I've commited to a branch and now I want to go back to the branch before the commit	10:29
engla	JohnFlux: git log --grep also has autopaging behaviour	10:29
	which is a nice git feature	10:29
JohnFlux	benc1: have you pushed the branch?	10:29
	benc1: I mean, have you already pushed your bad commit upstream	10:29
benc1	JohnFlux: no. I created a branch and did one commit -a but it didn't include a new file	10:30
doener	benc1: then you can simply amend the commit	10:30
benc1	JohnFlux: no. just commit -a -m "comment"	10:30
doener	benc1: git add $missing_file; git commit --amend	10:30
benc1	doener: will it overwrite the last commit?	10:30
JohnFlux	yes	10:31
benc1	cool	10:31
doener	benc1: it will be replaced in the branch head's history	10:31
JohnFlux	benc1: it helps to say what you're actually trying to do :)	10:31
benc1	JohnFlux: I'm trying to figure it out myself :)	10:32
	→ nothingHappens joined	10:33
	→ ddvlad_ joined	10:34
	→ froschi joined	10:34
	→ vvsh joined	10:35
ddvlad_	hey, how can i revert a single file (a sort of reset --hard, but for a single file)? a colleague suggested rm + checkout that file, but said it's probably not the standard way to do it	10:36
	← wolog left	10:36
	← bentob0x left	10:36
	→ stamina joined	10:37
doener	ddvlad_: git checkout HEAD -- file	10:37
	ddvlad_: or "git checkout -- file", if you want the version that's in the index	10:37
	→ jfkw joined	10:38
	← s0ber left	10:40
engla	I think this alias is nice: it discards all modifications not staged:	10:41
	alias.discardchanges=!git checkout -- $(git ls-files -m)	10:41
doener	engla: hm, shouldn't "git checkout -- ." do the same, just with less processes?	10:42
	s/less/fewer/	10:42
engla	I didn't know checkout -- . would work. And it might do the same work, but all in the checkout process	10:43
doener	engla: another possibility would be "git stash save --keep-index", to stash away the unstaged changes	10:44
engla	yeah but this is for explicitly just clearing them	10:45
	you might stage any changes and then discard any debugging output-related changes	10:45
	→ s0ber joined	10:46
	← MisterN left	10:46
doener	yeah, I primarily mentioned that for others, that might have other changes around that should go into another commit, but want to test what they're actually going to commit now	10:47
_rane	I'm looking for a way to make git automatically remove trailing whitespace from my commits, any tips?	10:47
	← ikrabbe left	10:47
dbrock	if I want to rebase a whole branch onto another branch and the branches share no common ancestor, what do I specify as "upstream"?	10:48
	← LRN left	10:48
GarethAdams\|Work	dbrock: if the branches are in the same repository, they have a common ancestor	10:48
doener	dbrock: a "whole branch" as in "including the root commit"? Then use "git rebase --root --onto <otherbranch>"	10:48
	GarethAdams\|Work: not necessarily	10:48
dbrock	doener: ah	10:49
doener	GarethAdams\|Work: clone git.git, fetch linux-2.6.git. Same repo, different root commits	10:49
	GarethAdams\|Work: and even in git.git alone, there are multiple root commits, as independent things got merged in (git-gui, gitk, ...)	10:49
GarethAdams\|Work	oh	10:49
	← benc1 left	10:50
	→ ph^_ joined	10:50
doener	GarethAdams\|Work: and there are even unconnected branches in git.git, like the "man" and "html" branches, that contain the pre-built docs	10:50
dbrock	hmm, my git-rebase doesn't support the --root option; I guess the cygwin version is a bit old	10:50
	← hyperair left	10:50
	→ hyperair joined	10:51
dbrock	does the "root commit" have a name?	10:51
	is the root commit the first normal commit or is it the parent of the first normal commit?	10:51
doener	dbrock: it's just a commit that has no parent	10:52
	dbrock: e.g.: git init; git add .; git commit # Creates a root commit	10:52
dbrock	(the first normal commit doesn't have a parent as far as I can tell)	10:52
jjuran	doener: I have a question about that. Isn't it disruptive to switch from master to html and have all the code disappear? Or are these disjoint branches only checked out in separate working trees?	10:52
engla	a root commit has 0 parents, normal commit 1 and merge commit has 2 or more. (What's the maximum?)	10:52
doener	engla: IIRC there was a limit for octopus merges to have 16 parents, but that has been increased or lifted, not sure	10:53
dbrock	but if I give the root commit as the upstream for rebase, it won't include that commit, will it?	10:53
	→ capndiesel joined	10:54
engla	jjuran: merged-in subtrees are especially disruptive with bisect.. if your bisect goes out on a subtree branch. Bisect can't jump to 'html' or any disconnected branch of course	10:54
doener	dbrock: it won't. "git rebase --onto <onto> <upstream> <branch>" creates patch from <upstream>..<branch>, which excludes <upstream>, thus there's the --root switch	10:54
jjuran	engla: I use temporary octopus merges to tie my topic branches together -- a couple dozen at most, so far.	10:54
engla	ok	10:54
doener	dbrock: with --root, you don't provide <upstream> at all	10:55
dbrock	doener: ah, yes, okay, thanks, so... hmm...	10:55
ddvlad_	doener: thanks for the answer, i was awol for a bit :)	10:55
doener	jjuran: well, if you work on the html branch, you don't want the code around, it has nothing to do with the html stuff	10:55
jjuran	engla: To clarify: I have only tried to merge a couple dozen branches at once, and Git handled it just fine.	10:56
dbrock	doener: yeah, I gathered, but I wonder what to do when my git-rebase has no support for the --root option	10:56
doener	jjuran: but as that's just holding the build results, Junio probably never switches to it, but has a script to do all the work of building the html docs and updating the branch	10:56
engla	jjuran: at that point it's more surprising there aren't any explicit or implicit conflicts	10:56
doener	dbrock: hm, not sure whether you can cherry-pick root commits, but if so, that would be an option	10:57
	← ustunozgur left	10:57
jjuran	doener: I wouldn't want to work on the code and then switch to html. My editor would see the files disappear and close their windows. And then upon checkout I'd have to recompile everything.	10:57
doener	dbrock: cherry-pick the root commit, then do "git rebase --onto other_branch_with_cherry_picked_root root	10:57
dbrock	doener: ah, so cherry-pick it and THEN rebase	10:57
	yeah, good idea, thanks!	10:57
doener	jjuran: then do things in a separate working tree, no problem with that	10:58
jjuran	doener: So those branches aren't really for human consumption anyway. Accurate?	10:58
doener	jjuran: In this case, yes. I assume that Junio has scripts to update them, and as a user you do "make quick-install-man" (for the man branch, I guess the target for the html docs is quick-install-html, but I don't know, as I don't use it)	10:59
	→ Escafane joined	10:59
	← _Vi left	11:00
	→ boto joined	11:00
	→ _Vi joined	11:00
	→ botanicus joined	11:01
	← khinester left	11:01
	← tbuehlmann left	11:02
	← ^Einstein left	11:02
	← magcius left	11:06
jjuran	engla: The workflow doesn't allow for conflicts. If two branches conflict then one of them can't be a topic branch.	11:08
	← ixti left	11:09
	→ jds_ joined	11:10
doener	jjuran: well, that's a bit too general to be said that way. I can have "fix bug in XYZ" and "improve performance of XYZ", two topics, and they can easily conflict	11:10
	← jds left	11:11
	← mrtazz left	11:12
	→ mrtazz joined	11:12
	→ brizly joined	11:14
	→ romankrv joined	11:14
	→ psynaptic joined	11:16
	→ jkp joined	11:19
	→ corni joined	11:21
_rane	how do I see what's currently staged to be committed?	11:23
	on diff level	11:23
teuf	git diff --cached	11:24
	← jinzhu_ left	11:24
_rane	thanks	11:24
	← gretch left	11:25
	← jds_ left	11:28
	← brizly1 left	11:29
	→ solsTiCe joined	11:31
	← upul left	11:31
solsTiCe	hi. how do i get the lsit of files tracked by git ? git status show the one not tracked	11:31
	← abms1116 left	11:32
	→ aspotashev joined	11:32
wereHamster	git ls-files	11:32
	← ruphy left	11:33
solsTiCe	ok. thanks	11:34
	→ Scytale joined	11:34
	← hyperair left	11:35
Scytale	Hi. I want to move a subdirectory D of repo A into a new repo B, keeping history. "filter-branch --subdirectory-filter" looks pretty much like the way to go, but there is one problem:	11:36
	← relix left	11:36
	→ xymbol joined	11:37
Scytale	The subdirectory D was recently renamed, and I want to include the commits before its rename as well. filter-branch apparently doesn't do that. What are your suggestions?	11:37
	→ spaceonline joined	11:37
	→ Chaser__ joined	11:38
	← Theravadan left	11:39
Chaser__	fired up firefox today to notice that all my history book marks is gone the left right stop and refresh buttons dont work !	11:39
BC-bd	Scytale: no idea if it works, but checkout the dir just up to the rename, export that, and then checkout to head and export the delta starting from the rename	11:39
Chaser__	oops sorry wrong channel	11:40
	→ marc_m joined	11:41
	→ Cynos joined	11:41
	→ f00li5h joined	11:41
	f00li5h meows	11:41
	marc_m → marc_kdab	11:42
Scytale	BC-bd: Could you please tell me how you would "export that"? I also thought about using cherry-pick on all the commits to form a new branch...	11:42
Cynos	hi all, I've got myself in a pickle that I'm unsure how to undo. There's a rather bad commit on a branch on my remote bare repo, and I want to get rid of it before anyone pulls it down...	11:42
	Can I rebase it?	11:42
BC-bd	Scytale: I'd guess with filter-branch	11:43
Chaser__	Cynos, you can and do push -f but thats not recommended instead revert the particular commit using git-revert	11:45
	→ dodo_the_last1 joined	11:53
	→ sswam joined	11:54
sswam	error: git-checkout-index: unable to read sha1 file of common (fedaaee6fa5c0006f5b7264732cb4e29584ef100) - while doing git clone git://git.collabora.co.uk/git/farsight2.git any idea what's the trouble there?	11:55
	my git too old?	11:55
	← doener left	11:55
	→ doener joined	11:55
jjuran	doener: Agreed, but only one can be an 'active' topic branch, by definition. One of the conflicting topics must be demoted to 'inactive' until it gets rebased on top of the other one.	11:57
	→ abms1116 joined	11:57
doener	jjuran: hu? Why that? You can just merge one after the other to your "main" branch, fixing conflicts if they arise	11:58
sswam	so I did a git fsck-objects, it says there's a missing blob. How do I ask it to fetch that off the server?	12:00
	→ univate joined	12:00
	← dodo_the_last left	12:00
jjuran	doener: The point is to keep them unmerged while I'm still working on them.	12:01
	→ freerobby joined	12:02
doener	jjuran: sure, but you can work on both on them "at the same time", switching between them	12:02
	jjuran: and then, at some point, you merge each one, and that can cause conflicts	12:03
	jjuran: and there's no need to rebase anything (which is even more likely to cause conflicts anyway)	12:03
jjuran	doener: The goal is to have a build which incorporates the topics, either because they have important fixes or new features I'm testing out.	12:03
	doener: Hence the ever-regenerated octopus merge called 'tip'.	12:04
Cynos	Chaser__ thanks for that. For some reason I never looked at git revert >_<	12:05
Chaser__	no prob	12:05
doener	jjuran: so? That doesn't explain why you said "if they conflict, they're no topic branches"	12:05
	jjuran: topic branches may very well conflict, that's my whole point	12:05
	→ qrush joined	12:07
jjuran	doener: I'd rather discuss workflow than semantics.	12:08
	← hummingkid left	12:10
	→ jepler_ joined	12:10
	→ keystr0k joined	12:11
jjuran	doener: I see the elegance in working on one topic at a time, but sometimes they're interdependent. Topic foo might be a bugfix required for feature bar, so I need to have them both active to verify that the bugfix works, or that it's still necessary at all.	12:11
doener	jjuran: this whole thing started by you saying that two branches with conflicting changes can't be topic branches. I said that that is too general. If your workflow doesn't allow branches to conflict, that's a totally different story, but doesn't justify the redefinition of a topic branch	12:11
jjuran	Okay, I withdraw the term 'topic branch' and substitute 'branch which shall be merged into tip'. :-P	12:12
doener	that, I can live with ("tip" is weakly defined, if at all ;-))	12:13
	(in general)	12:13
	and it's specific to your workflow	12:13
	→ lu_zero joined	12:14
	→ parasti joined	12:15
	→ rudi_s joined	12:15
	← benjo2 left	12:15
	→ PhilK joined	12:16
	→ bs joined	12:16
	→ mw__ joined	12:16
bs	hi	12:16
	← ddvlad_ left	12:17
	← PhilK left	12:18
jjuran	sswam: If you have that object in another repo you can copy it manually.	12:18
	→ PhilK joined	12:18
sswam	I think I'll try cloning again with my mangled repo as a reference.	12:19
	← rwadkins1 left	12:20
	← galderz left	12:26
	→ benjo2 joined	12:28
	← solydzajs left	12:29
	→ LRN joined	12:30
	→ galderz joined	12:30
	← jepler left	12:31
	→ Aszarsha joined	12:31
	← Cynos left	12:32
	→ tstclair joined	12:32
	← [Po]lentino left	12:34
	← gaveen left	12:36
dsop	hmm i wonder why I cannot clone just one branch from a remote repository using git clone.	12:36
	→ lottadot joined	12:36
dsop	or at least if I'm the only one who want to clone just a specific branch	12:37
	→ Cynos joined	12:37
	← Cynos left	12:37
doener	dsop: the bandwidth/space savings are (usually) negligible. And the "main" repo of a project usually has few branches. For additional remotes, you can specify that you want just a single branch.	12:39
	dsop: But I kind of doubt that a patch would be rejected ;-)	12:39
dsop	yes. I'll look into the code, shouldn't be that hard to implement it	12:39
	→ ixti joined	12:39
doener	dsop: Actually, I found the clone code to be somewhat cryptic ;-) But I guess the limiting should be easier than the "adjust HEAD" stuff, for which I had to look at it	12:40
	→ gh34 joined	12:40
	→ Ryback_ joined	12:41
	← freerobby left	12:43
	→ alienbrain joined	12:43
	→ mm_work joined	12:44
alienbrain	In git status I see: "typechange: <file>" what does typechange mean?	12:44
jjuran	alienbrain: I imagine it means the x bit of the permissions is toggled.	12:44
ringlej	how can I generate a list of all files being tracked by git so that it is usable as input to another command that accepts a list of file names from stdin?	12:45
teuf	git ls-files	12:45
doener	jjuran: that would be "mode change" IIRC	12:45
jjuran	alienbrain: Or it became a symlink, or vice versa. etc.	12:45
	ok, never mind	12:45
	← LRN left	12:45
ringlej	teuf: thanks	12:46
doener	the symlink thing should actually be it	12:46
alienbrain	jjuran, there is indeed a permissions change on the file, but there are also changed contents and the --- is /dev/null; I think it was a symlink indeed! Thanks	12:46
sitaram	doener: could it also be git's internal view of whether the file is a text file or a binary file?	12:47
doener	sitaram: I don't think so. It's still a regular file. The text/binary thing should only be interesting for the diff/merge machinery	12:47
sitaram	ok	12:48
	just wondering...	12:48
doener	haha, great way to make me look stupid. The "typechange" is obviously produced by the diff machinery...	12:48
	← abms1116 left	12:48
	→ lorandi joined	12:48
doener	but I hope you get what I actually meant, I fail to find words to explain it	12:48
	typechange is when the upper bits of the mode change, the file type. Regular file, symlink, submodule, ...	12:50
jjuran	doener: I think you're saying that Git plumbing doesn't interpret file contents as anything more than a bytestream.	12:50
doener	jjuran: yeah, that's pretty close to what I have in mind.	12:51
	→ JackWinter2 joined	12:53
	→ m-takagi joined	12:53
	← _Vi left	12:54
	→ freerobby joined	12:54
	→ ruphy joined	12:55
	→ hakunin joined	12:56
	← freerobby left	12:56
	→ Juice10 joined	12:57
	← ntoll left	13:00
	→ hummingkid joined	13:00
	← hummingkid left	13:00
	→ augustl joined	13:01
	→ ereslibre joined	13:01
augustl	hi! Is there a way to push to all of the remotes in .git/config with one command?	13:01
	← vvsh left	13:01
wereHamster	you can have several urj's in one remote IIRC	13:01
	url's	13:02
augustl	ah. Can you recall the syntax to do this via a git-* command? Or does it require manual editing of .git/config?	13:02
doener	git remote \| xargs -n1 git push	13:02
wereHamster	just edit the config file	13:02
	→ scarabx joined	13:03
	→ bnovc joined	13:04
	← peper left	13:04
augustl	I see, thanks	13:05
bnovc	is there a way to have git update to a given time? e.g. update to whatever commit was HEAD on 01/01/09 or something	13:05
	→ ferdy joined	13:06
bnovc	git checkout `git rev-list -n 1 --before="2009-07-27 13:37" master`, apparently	13:06
	← benjo2 left	13:07
	← JackWinter left	13:08
	→ flazz joined	13:08
teuf	git checkout master@{2009-07-27 13:37} might work	13:08
	→ mdwright joined	13:08
thiago	it's not the same thing	13:08
	→ AnAnt joined	13:08
thiago	teuf's answer is more correct; a strict answer to the question	13:09
	→ j416 joined	13:09
mdwright	Does anyone remember what the earliest version of git that allows cloning to an empty dir is? Or at least approximately when it was introduced?	13:09
engla	teuf's answer works on the reflog, the local repository history only	13:09
	teuf: and you need to replace one space with a period (.)	13:09
	master@{two.days.ago} works without the shell splitting it	13:10
doener	mdwright: v1.6.2-rc0~137^2	13:10
mdwright	doener: Thank you	13:10
doener	mdwright: git log --grep='empty directory' builtin-clone.c	13:11
	← Scytale left	13:11
AnAnt	Hello, I am trying to do the a merge from a branch called 'upstream', and I get a lot of conflicts, can't I do the merge such that I get all changes from 'upstream', isn't there a "theirs" strategy ?	13:11
teuf	engla: do you mean that if I haven't pulled up to date stuff to master in years and if I just did a git pull master which gets me lots of commits, master@{1 week ago} will refer to the master from years ago, not master + commits older than 1 week that I just pulled?	13:12
	→ binjured joined	13:13
	→ maravillas joined	13:13
	← augustl left	13:14
engla	teuf: I don't know the inner workings, but I think @{} works with the reflog. Anyway, my repo history goes back to 1. Dec 2007. And when I say git checkout master@{1.dec.2007} it says warning: Log for 'master' only goes back to Sun, 6 Sep 2009 01:55:21 +0200	13:15
	teuf: on Sept 6 I deleted my copy and cloned it anew.	13:16
teuf	ah ok, interesting	13:16
engla	I agree that the same function, but working with different semantics would be great	13:17
	newest commit on branch now called master, not newer than 1.dec.2007	13:17
	← mdwright left	13:17
	← rudi_s left	13:20
	← ereslibre left	13:23
	← scarabx left	13:24
	→ rudi_s joined	13:24
	→ roop_ joined	13:24
	→ ereslibre joined	13:24
	→ MisterN joined	13:25
	← sergeykish left	13:26
shruggar	teuf, check the help for git rev-parse regarding --since and --until	13:28
	→ drubies joined	13:30
	← proofek left	13:33
	← zenwryly left	13:33
	→ zenwryly joined	13:34
	← j416 left	13:34
	→ jnveqv joined	13:35
	← romfert left	13:37
	← roop left	13:38
	← jdahm left	13:39
AnAnt	Hello ?	13:42
	→ romfert joined	13:43
rapha	cheep	13:43
sitaram	tweet	13:43
Rhonda	dent	13:43
jjuran	O HAI	13:43
AnAnt	I am trying to do the a merge from a branch called 'upstream', and I get a lot of conflicts, can't I do the merge such that I get all changes from 'upstream', isn't there a "theirs" strategy ?	13:44
	← intripoon left	13:44
Rhonda	Is it possible to extract a git binary patch into a real file without applying it to some git repository?	13:44
	→ spearce joined	13:45
jjuran	AnAnt: Do you want to start over from upstream, or merge your changes?	13:45
	To start over, git reset --hard origin/upstream	13:45
AnAnt	merge my changes	13:46
	← agile left	13:46
	← tango_ left	13:46
jjuran	I recommend against blindly accepting 'their' changes. I would try merging in upstream a bit at a time.	13:46
AnAnt	but is there a way to blindly accept their changes ?	13:47
	my changes are in files that are not in 'upstream'	13:47
	→ calin joined	13:47
AnAnt	so 'upstream' changes files x y z ...., and my changes are in a folder called 'debian/'	13:48
jjuran	AnAnt: Oh, that's easy, then.	13:48
AnAnt	jjuran: how ?	13:48
	→ aziz joined	13:48
jjuran	AnAnt: git checkout upstream; git checkout mybranch files/i/changed; git commit	13:48
dsop	how do I get a list of changed files between two tags?	13:49
jjuran	dsop: I heard it's git diff --name-only foo bar	13:49
	AnAnt: You probably want to create a new branch to do the merging, but you get the idea.	13:50
	→ ereslibre_laptop joined	13:51
	→ squentin joined	13:52
	→ cjkvqo joined	13:53
	← jnveqv left	13:55
	← ereslibre left	13:55
	← ph^_ left	13:56
	→ GNUtoo joined	13:58
	← Sho_ left	13:58
	→ jerbear joined	14:00
	← marc_kdab left	14:01
	→ ctusar joined	14:01
	→ TeckniX joined	14:02
	← qrush left	14:03
	→ qrush joined	14:03
	→ suman_ joined	14:04
	← sh1mmer left	14:04
	← romfert left	14:05
	→ cmarcelo joined	14:06
	← jlf_away left	14:06
doki_pen	how do I export a set of commits to file?	14:07
Rhonda	doki_pen: git format-patch	14:08
doki_pen	Rhonda: ah, that's right! ty	14:08
Pieter	or git bundle	14:09
doki_pen	Pieter: ah, I never used that one	14:10
	→ segher joined	14:12
	→ InTheWings joined	14:13
InTheWings	Is that normal to have stashed changes pulled from master ?	14:13
doener	that question makes no sense to me	14:14
	← drubies left	14:15
drizzd_	same here	14:15
	← AnAnt left	14:15
doener	the "master" branch shouldn't have stash entries in its history, and "pull" is "fetch + merge/rebase", so that makes even less sense	14:15
	← parasti left	14:16
InTheWings	Well, i have some stashed changed in my stash stack, and that's not mine	14:17
	so I guess these have been 'pulled' from repo	14:17
	s/stashed changed in my stash stack, and that's not mine/stashed changes in my stash stack, and these arent mine/	14:18
	→ jlf` joined	14:18
	→ peper joined	14:19
	→ romfert joined	14:19
	→ ProperNoun joined	14:20
	← mastro left	14:20
	→ icwiener_ joined	14:21
	→ nh2 joined	14:22
	→ tango_ joined	14:22
nh2	is Git able of showing me the git log of a given commit or do I have to check it out first?	14:23
	→ mivok joined	14:24
Ilari	nh2: 'git show <commit>'?	14:24
	nh2: Or 'git log <commit>'?	14:24
	→ Bass10 joined	14:25
	→ freerobby joined	14:26
nh2	Ilari: indeed, thanks. I could not make that out of the syntax of man git-log, but it seems that is <since>	14:26
Gitbot	Ilari: the git-log manpage can be found at http://git.or.cz/man/git-log	14:26
	← calin left	14:26
drizzd_	InTheWings: no, that's not normal. The stash is always local.	14:27
	← icwiener left	14:27
	→ torarne joined	14:27
InTheWings	drizzd_: yes, that's what i guess, something is wrong	14:27
drizzd_	Maybe a hook is doing this?	14:28
	Or someone else had write access to your repo?	14:28
	→ hugoxrosa joined	14:28
	← johan-s left	14:29
	← ferdy left	14:30
	→ btcnlf joined	14:30
	← romfert left	14:32
InTheWings	drizzd_: I don't have write access to repo. I only contribute through patchs	14:32
	← cjkvqo left	14:33
	← flazz left	14:34
	→ loincloth joined	14:35
	→ flazz joined	14:35
	→ jrmuizel joined	14:38
	→ Swimming_Bird joined	14:39
	← maattd left	14:42
	→ johan-s joined	14:42
	← brendon_work left	14:43
	→ mbroeker joined	14:45
xymbol	can someone point me to a good tutorial about managing multiple remotes? I need two local branches pointing to master on different repositories	14:45
	← flazz left	14:45
	← johan-s left	14:45
	→ bcardarella joined	14:45
	→ romfert joined	14:46
	→ altrux joined	14:46
	← przemoc left	14:46
	← btcnlf left	14:47
sitaram	xymbol: just make two "remotes". git remote add remoteA <url of remote A>, then the other one	14:48
Mikachu	git remote add otherplace git://woo.com; git checkout -b othermaster otherplace/master	14:48
sitaram	then "git fetch remoteA" and "git fetch remoteB"	14:48
doener	remote add -f ....	14:49
sitaram	aah! even better	14:49
doener	or a fetch inbetween remote add and checkout -b	14:49
Mikachu	oopsies	14:49
xymbol	sitaram: did that already; though I'm not sure if each branch is correctly set up to pull/merge back to each remote	14:49
	doener: thanks!	14:49
sitaram	xymbol: if unsure, you can fetch and push explicitly, without worrying about the defailts etc	14:50
doener	IIRC "git remote show <name>" shows which local branch heads have a "pull" configuration that uses this remote	14:50
xymbol	sitaram: good point :-)	14:50
	→ Turias joined	14:51
	← ciscbrain left	14:51
xymbol	doener: tried that, I see correct configuration for git pull; I understand git push is broken though (master pushes to master (local out of date))	14:51
	→ tedoc2000 joined	14:51
xymbol	sitaram: explicit push does work	14:52
	→ ereslibre_desk joined	14:52
sitaram	xymbol: and for what it is worth, that is what I do mostly now; rarely use the defaults anymore...	14:52
xymbol	sitaram: maybe you're right, less magic	14:53
sitaram	there's been a fair amount of churn in that area, new options, new messages... so I sorta backed off :)	14:53
Mikachu	shouldn't git push -v --dry-run tell you what happens too?	14:53
sitaram	yess, less magic. Plus I also teach so it's useful to give people a "baseline" that is easy to understand	14:54
	Mikachu: I guess I never tried... /me tries	14:54
xymbol	Mikachu: yes, and push is failing: git push doesn't push; git push remotename localbranch:remotebranch does)	14:54
Mikachu	xymbol: ah, i doubt automagic push can 'rename' branches	14:55
	xymbol: you can set it up in your .git/config though	14:55
	just put in push = localbranch:remotebranch lines	14:55
xymbol	Mikachu: that's probably what I missed	14:55
	→ Neo_The_User joined	14:56
	→ ruffyen joined	14:56
Neo_The_User	how do you undo a merge that you just did and has 200 conflicts?	14:56
Mikachu	Neo_The_User: git reset --hard	14:56
	Neo_The_User: assuming you didn't have anything you want to save that you haven't committed	14:56
Neo_The_User	correct	14:56
	thanks!!	14:56
Mikachu	i should probably have said that in the other order	14:57
Neo_The_User	worked too XD	14:57
	← Neo_The_User left	14:57
ruffyen	im curious if i have two computers using git and one repo, and I create a branch on computer(1) commit then do git push origin master....if i do a git pull origin master from computer(2) will i get the branch as well?	14:57
Mikachu	ruffyen: no, pull only merges one branch	14:58
xymbol	Mikachu: push.default or remote.name.push?	14:58
Mikachu	xymbol: the latter	14:58
ruffyen	Mikachu: so how would i get teh branch in the repo	14:58
Mikachu	and "name" is a placeholder there	14:58
	ruffyen: git fetch	14:58
xymbol	Mikachu: thanks! ;-)	14:58
ruffyen	so git fetch <branchName>?	14:59
	→ killerchicken_ joined	14:59
	→ agile joined	14:59
Mikachu	ruffyen: no, just git fetch	14:59
killerchicken_	how can I view a diffstat for a random commit in my history?	14:59
Mikachu	killerchicken_: git show --stat anyrandomcommit	15:00
ruffyen	how does it get from computer(1) into the repo though...i dont have to do anything special?	15:00
Mikachu	ruffyen: oh you did 'push origin master' yeah then it won't be in the remote repo	15:00
	ruffyen: you have to git push origin thenewbranch	15:00
killerchicken_	Mikachu: great, thanks!	15:00
ruffyen	ah ok	15:00
	Mikachu: thanks	15:00
	thats what i need	15:00
xymbol	Mikachu: awesome, that was it!	15:01
doener	Mikachu, xymbol: push.default = tracking can do renaming	15:01
	← sde_antrix left	15:01
doener	it will push the checked out branch to the branch from which it would pull	15:01
xymbol	doener: yes, I thought so; that makes the push config follow fetch's, right?	15:01
Mikachu	o	15:01
xymbol	doener: cool, thanks!	15:02
	← sswam left	15:03
	→ xyz joined	15:03
	← jrmuizel left	15:04
Swimming_Bird	i'm getting this strange thing where there are some head errors with git svn	15:04
	i cant rebase for the life of me	15:04
	→ kraymer_ joined	15:05
	← rkirti left	15:06
	→ sh1mmer joined	15:08
	→ Thell joined	15:08
	→ LiamH joined	15:08
	→ evildmp joined	15:09
	→ PuffTheMagic joined	15:09
	→ hobodave joined	15:09
	← ereslibre_laptop left	15:09
evildmp	how should I roll back to <some_commit>?	15:09
	git revert <some_commit>?	15:09
sitaram	reset	15:10
wereHamster	roll back is reset --hard	15:10
sitaram	revert creates a new commit on top that is the inverse of <commit>	15:10
evildmp	maybe I am using the wrong term	15:10
Thell	Quick question: I seem to remember reading somewhere how to have git echo out a value it is using for an env variable. like 'git --exec-path', but I'm wanting to see the GIT_SSH value it is using. Any idea or is my memory wack?	15:10
sitaram	that would not be "rollback"	15:10
evildmp	I have a list of commits in git log	15:10
wereHamster	or not hard, depending on what you want to do	15:10
evildmp	these are commits I have got from the repo	15:11
	← PuffTheMagic left	15:11
	→ PN joined	15:11
evildmp	and I want to see if a problem goes away when I go back a couple of commits	15:11
	← sprsquish left	15:12
wereHamster	evildmp: man git-bisect	15:12
Gitbot	evildmp: the git-bisect manpage can be found at http://git.or.cz/man/git-bisect	15:12
evildmp	so I can see the one I want, and it has a commit id by it - how to I turn bakc the clock to that point	15:12
sitaram	Thell: I'm not sure what you mean; but git rev-parse does echo out some internal stufflike --git-dir etc	15:12
	← ruffyen left	15:12
sitaram	not env vars thought	15:12
	though*	15:12
Thell	sitaram: I must be mis-remembering	15:13
	→ wolog joined	15:13
Thell	Can't find anything grepping the help files that is what I was thinking either	15:13
	← kraymer left	15:13
sitaram	evildmp: if you want to temporarily go to a particular commit (two commits prior as you said) just to test, you can do "git checkout master^^" but do NOT make any commit on this until you come back with "git commit master"	15:14
	← romankrv left	15:14
evildmp	i am not commiting anything, just reading	15:14
sitaram	evildmp: if, however, you want to permanently lose those 2 top commits, git reset --hard master^^	15:14
	evildmp: then checkout is probably best	15:14
evildmp	sitaram: I can always git update to get them back, can't I?	15:14
Mikachu	they'll still be in your reflog too	15:14
sitaram	evildmp: I dont know any "git update" but "git checkout master" will	15:15
evildmp	sorry, I meant git pull	15:15
	I was getting confused with hg	15:15
sitaram	sure it would	15:15
jettero	what is the state of things for windows+git?	15:15
	I never tried it	15:15
sitaram	works fine...	15:16
	a lot of people in my teams use msysgit	15:16
	→ PuffTheMagic joined	15:16
Mikachu	bash is terrible in cmd.exe though	15:16
jlf`	hi #git, i'm having trouble getting gitolite set up. i've run src/gl-compile-conf which happily created the repos, and i installed my public key. i followed the instructions in 3-faq-tips-etc to populate an empty repo on a client and push, but i see "! [rejected] master -> master (non-fast forward)" any advice?	15:16
	→ fhr joined	15:17
jettero	is there a gui for it? my windows peeps aren't going to use it if they have to type things	15:17
sitaram	jlf`: if you got that far you're well past gitolite probles :)	15:17
Ilari	jlf`: That repo isn't empty...	15:17
sitaram	jlf`: what does "git ls-remote origin"say	15:17
	← botanicus left	15:17
	→ cj joined	15:18
jlf`	8dc3daa665ddc030993d10c8a35adfa9af901566 HEAD 8dc3daa665ddc030993d10c8a35adfa9af901566 refs/heads/master	15:18
		15:18
cj	hey all	15:18
	how do I get a diff between the local master and the origin master?	15:18
Mikachu	cj: git diff master origin/master	15:18
sitaram	jlf`: as Ilari said, your remote has commits in it already, so pullfirst	15:18
cj	Mikachu: thanks	15:19
ToxicFrog	jettero: there's git-gui (basic graphical interface) and gitk (for examining history); I believe there's also a TortoiseGit now, too.	15:20
	→ schacon joined	15:20
jettero	ToxicFrog: thanks. I'm reading the things on the git wiki too	15:20
sitaram	jettero: a few things will always need typing but none of the day-to-day stuff is in that category	15:21
jettero	it looks like there's a few choices	15:21
jlf`	sitaram: git pull origin -> You asked me to pull without telling me which branch you want to merge with, and 'branch.master.merge' in your configuration file does not tell me either.	15:21
sitaram	jettero: there'salso "gitextensions"	15:21
jettero	looks like regular old git-gui will do most everything	15:21
jlf`	(sorry, i'm new at using remote repos if it wasn't obvious already :)	15:21
sitaram	jlf`: did you do any stuff on your local side; perhaps you should first clone, since your remote seems to have content already	15:22
	jlf`: otherwise, 'git config branch.master.remote origin;git config branch.master.merge refs/heads/master'	15:22
jlf`	hmm, not sure why it would since i've only done gl-compile-conf	15:22
sitaram	jlf`: perhaps you're using some other repo than the one you set up with gitolite, because if your git ls-remote showd you that stuff, it's certainly not an empty repo	15:23
	→ justin-george joined	15:23
sitaram	showed	15:23
	← ProperNoun left	15:23
sitaram	jlf`: humor me and clone to some other (empty) directory... then run "git log" on that clone see what comes up	15:24
	← ilteris left	15:24
	ilteris_ → ilteris	15:24
	→ ilteris___ joined	15:24
	→ Arild joined	15:25
	→ maattd joined	15:25
jlf`	sitaram: that command line worked, and i see from the output that you are correct -- i believe i specified the url wrong and it was pulling from jlf@remote-server:~/repo.git instead of jlf@remote-server:~/repositories/repo.git	15:26
Ilari	jlf`: Ugh. Neither should be valid.	15:27
	jlf`: For gitolite.	15:27
jlf`	where repo.git in ~ was an earlier attempt before i learned that i need gitolite	15:27
	← bgerber left	15:27
sitaram	jlf`: so that repo is still around and you're still using it...! I suggest safely renaming it (like "repo.old.git.NOT" ;-) and then continuing with gitolite	15:28
	→ adeodatus joined	15:28
	→ marvil07 joined	15:28
	→ JNR___ joined	15:28
	→ roop__ joined	15:28
jlf`	sitaram: done	15:29
	i believe i entered the wrong origin in the line `git remote add origin git@gitolite-server:my-new-project.git' from the faq -- if i am [email@hidden.address] and i want the remote to my [email@hidden.address] , how should that line read?	15:30
sitaram	jlf`: ok next, your url on client should simply be "user@host:repo.git"	15:30
	jlf`: your gitolite repo is always <user>@<host>:<reponame as it appears in the config file>.git	15:30
	nothing more or less	15:31
jlf`	hmm.. that's what it was before. could the presence of ~/repo.git confused it?	15:31
sitaram	definitely not the repositories/ bit	15:31
	→ agib joined	15:31
sitaram	jlf`: no	15:31
	← aatanasov left	15:31
Ilari	sitaram: Is 'ssh://<user>@<host>:<port>/<reponame>.git' handled correctly (that appears as '/<reponame>.git' on other end)?	15:31
jlf`	the repositories bit was never part of a command, i was just saying which one i wanted to use	15:32
sitaram	Ilari: yes	15:32
	→ fualo joined	15:32
Ilari	jlf`: If jlf is account gitolite is using, specifying ~/ in URL should cause permission denied error.	15:32
sitaram	Ilari: it expects an optional leading slash	15:32
	expects => accepts I guess	15:32
agib	does anyone know the deal with the "error: object directory ... " errors on github now?	15:33
Ilari	agib: pushing or fetching?	15:33
	agib: And what's the exact error?	15:33
agib	Ilari: pushing	15:33
	→ [Po]lentino joined	15:33
agib	error: object directory /data/git/repositories/4/c8/56/e4/pluginaweek/state_machine.git/objects does not exist; check .git/objects/info/alternates.	15:34
	actually pushing and fetching	15:34
sitaram	jlf`: just type in "git ls-remote user@host:repo.git" and see what it says	15:35
Ilari	jlf`: try 'ssh user@host' (where user@host is exactly as it appears in the URL). You should get error message (it also displays what you authenticated as).	15:35
	agib: Err... .git/... WTF?	15:35
	agib: Does that '.git/objects/info/alternates' exist locally?	15:36
agib	no	15:36
	info/ is an empty dir	15:36
	← sh1mmer left	15:37
	→ meanburrito920_ joined	15:37
agib	weird.. lot of people bringing up the issue on github's tender... but can't find a resolution	15:37
	it doesn't seem to be a fatal error though... i.e. I think the push worked	15:37
	→ ntoll joined	15:37
jlf`	sitaram, Ilari: see http://paste.lisp.org/+1VRS	15:38
	→ jrmuizel joined	15:39
hobodave	hey guys	15:39
sitaram	jlf`: you cannot use the same pubkey for shell access and gitolite access	15:39
	→ flazz joined	15:39
hobodave	if I have uncommitted changes in my working directory, is it possible to create and switch to a new branch without losing these or committing them to my current branch?	15:40
	→ LOAST joined	15:40
LOAST	can some one help me with sony sound forge??????	15:40
hobodave	LOAST: are you serious?	15:40
LOAST	ya	15:40
jlf`	sitaram: oh ok.. sorry if i missed that in the docs, let me rectify	15:40
drizzd_	hobodave: git checkout -b newbranchname	15:40
sitaram	jlf`: if you have time please read http://sitaramc.github.com/0-installing/9-gitolite-basics.html -- it will explain all the ssh magic behind this and why this doesnt work, then see footnote #2 in the install document at http://github.com/sitaramc/gitolite/blob/master/doc/0-INSTALL.mkd	15:40
	→ LRN joined	15:41
hobodave	drizzd_: thanks drizzd_ I wasn't sure if that would trash y working directory	15:41
	→ Levenson joined	15:41
drizzd_	hobodave: it won't	15:41
LOAST	does any one one know where i could ask that q to???	15:41
	← JNR___ left	15:41
	← Levenson left	15:41
	→ JNR___ joined	15:42
	← roop_ left	15:42
hobodave	LOAST: http://superuser.com/	15:42
	→ bgerber joined	15:42
LOAST	thank you	15:43
jlf`	sitaram, Ilari: thanks for your help, sorry it came down to me not reading carefully enough :)	15:43
sitaram	jlf`: no problem; did it work now?	15:44
	← LOAST left	15:44
jlf`	i actually have to go for a meeting now but i'll try it in a bit	15:44
	it seems very likely that's the issue though	15:44
	→ hyperair joined	15:45
sitaram	ok	15:45
	be sure to read those 2 items if/when you have time	15:45
jlf`	will do, thx again	15:45
	← jlf` left	15:46
	→ drewolson joined	15:46
	← giallu left	15:46
	← Escafane left	15:48
	← xyz left	15:48
	← adeodatus left	15:49
	← pflanze left	15:50
	PN → ProperNoun	15:52
	← ndim left	15:55
	→ ablemike joined	15:55
	→ elmob joined	15:56
	← jkp left	15:56
	← justin-george left	15:56
	← jaysern left	15:58
	→ sh1mmer joined	15:58
	→ smuf joined	15:59
	→ alley_cat joined	16:00
	→ patmaddox joined	16:00
	→ kumbayo joined	16:01
	← Cesario left	16:01
	← elmob left	16:02
	← meanburrito920_ left	16:03
	← wolog left	16:04
	← schacon left	16:05
	→ evildmp_ joined	16:07
	→ Theravadan joined	16:07
	← drewolson left	16:07
	→ wolog joined	16:07
	← TeckniX left	16:10
	→ godsflaw joined	16:10
	→ TeckniX joined	16:11
	dwmw2_gone → dwmw2_OLM	16:11
	→ tbuehlmann joined	16:12
	← jepler_ left	16:13
	← evildmp_ left	16:14
	← evildmp left	16:14
	→ jepler joined	16:14
	← ereslibre_desk left	16:16
	← wolog left	16:18
	→ elmob joined	16:19
	← jchris left	16:19
	→ ndim joined	16:19
	→ jchris joined	16:20
	→ onigiri joined	16:21
	← JohnFlux left	16:21
	→ p_masho joined	16:21
jepler	can someone point me at a discussion of why 'git rebase --whitespace=fix -i' isn't accepted?	16:21
Thell	jepler: I can't point to a discussion, but I think that using 'apply.whitespace fix' might work in that situation	16:22
p_masho	newbie question. I'm using svn as a main repository so I want to be careful I dont mess this up. Can I add a gitignore to a subdirectory ? we got a main-repos and underthat around 4 "sub-project" directories.. or does it have to be top level. I want some developers to only checkkout the "subproject"	16:23
	p_masho trying to convert my company to git..	16:23
engla	you can have .gitignore files in any folder	16:23
	→ wolog joined	16:23
	→ tillberg joined	16:23
jepler	Thell: or doing 'git rebase --whitespace=fix' followed by 'git rebase -i'	16:23
engla	I think they stack with the closest first	16:24
	← tillberg left	16:24
Thell	jepler: nifty, I'll try to remember that one!	16:24
p_masho	brilliant...	16:24
	← galderz left	16:25
	← Teo` left	16:25
jepler	p_masho: unlike svn or cvs, you can't really check out only part of a git project (but git does have subprojects, which I haven't made myself familiar with yet)	16:26
	→ Teo` joined	16:26
p_masho	another newbie question.. I see that if I do commits locally with git, then "push" dcommit to svn, is there a way to make one en masse commit, instead of all the "local" git commits I made. The man page is a confusing me a bit really..	16:26
	→ tillberg joined	16:26
godsflaw	not sure if this is a late meme or not: http://xkcd.com/624/	16:27
p_masho	jepler: our companies using svn as their main repos atmo. Shoudl this "test" on thsi project work with git-svn then we might actually break stuff up, maybe..	16:27
jepler	p_masho: based purely on the docs of git-svn, it sounds like dcommit will always create 1 svn commit for 1 git commit	16:28
	→ jlf` joined	16:28
kevlarman	p_masho: you can use rebase -i to squash all the local commits into one	16:28
	← patmaddox left	16:29
engla	p_masho: or merge topics into the main git branch by merge --squash, then push that branch	16:29
	← GarethAdams\|Work left	16:30
	→ tongueroo joined	16:32
	← sergiodj left	16:33
	→ herlo joined	16:33
	← elmob left	16:34
	← shruggar left	16:34
	→ c15c8ra1n joined	16:35
	→ pflanze joined	16:36
	← froschi left	16:38
	← Juice10 left	16:39
p_masho	kevlarman: engla what I'm having a lot of options here.. :-)	16:40
	so can you advice me.. One of the problems we have is that with svn, every little commit is "logged", its actually a pain, henve git..	16:41
	So initially the idea it to >> git svn "pull", >> then create a git branch locally, then play in branch/commit, >> then when ready merge branch into master >> ?? rebase ?/ >> git svn dcommit as one "blob"	16:43
Mikachu	"blob" actually means something in git so call it something else, or you'll be confused :)	16:43
p_masho	I meant one commit of all teh local commits	16:43
	→ DaveIngram joined	16:44
	p_masho oops ...............................all the local commits.	16:44
	→ b0fh_ua joined	16:45
b0fh_ua	hello! is there any way to synchronize existing SVN repository with existing git repository?	16:45
	← qrush left	16:46
	← hyperair left	16:47
	→ hyperair joined	16:48
	dwmw2_OLM → dwmw2_gone	16:48
	← tafryn left	16:48
	← mnick left	16:48
	→ DavidKlein joined	16:49
	← kraymer_ left	16:50
	← noginn left	16:50
	← godsflaw left	16:50
	→ plediii joined	16:50
	← wolog left	16:51
	→ mrlemao joined	16:53
	← mrlemao left	16:54
	← sh1mmer left	16:54
	← b0fh_ua left	16:54
	→ fualo_ joined	16:55
	→ abms1116 joined	16:56
	→ krawek joined	16:56
	← abms1116 left	16:56
	→ qhoxie joined	16:57
	→ abms1116 joined	16:57
	→ sh1mmer joined	16:57
	→ godsflaw joined	16:58
	herlo is looking for any good pre-receive hooks which can check syntax of a particular subset of committed files.	16:58
	← abms1116 left	16:58
	→ pheaver joined	17:00
	→ mattcodes joined	17:01
	← lu_zero left	17:01
mattcodes	cloned a super project which contains some submodules, these weren't fetched initially, so i read you do git submodules init	17:01
	followed by git submodules update	17:01
	and voila	17:01
	→ abms1116 joined	17:01
mattcodes	I tried that but I enter the wrong private key twice	17:01
	I know what it is now (sorry I meant passphrase), but now it wont update	17:02
	the submodule directories are empty aside from .git	17:02
	← abms1116 left	17:02
mattcodes	I've rather not do a fresh clone... what the steps to manually retrieve submodules?	17:02
	→ abms1116 joined	17:03
	← stamina left	17:03
	← abms1116 left	17:04
	→ WALoeIII joined	17:05
	→ vvsh joined	17:06
mattcodes	done a fresh clone and all showing now, but would still like to know how to force update of submodule	17:08
	→ abms1116 joined	17:09
	← romfert left	17:09
	← abms1116 left	17:10
	← Theravadan left	17:10
	→ ^Einstein joined	17:10
	→ abms1116 joined	17:10
	← abms1116 left	17:11
	→ abms1116 joined	17:12
Thell	I'm getting a 'can't unlink' permission denied on a submodule just added to a topicBranch when I switch back to master. Checked google and found one thread mentioning permissions on object dir 16. Which I don't have. All the dirs I do have show the same permissions. Any direction on where to troubleshoot this?	17:13
	→ bugfux joined	17:15
	← Chaser__ left	17:15
	← [Po]lentino left	17:16
	→ reidab joined	17:16
	← fualo_ left	17:16
	→ fualo_ joined	17:16
	← T2000-lap left	17:19
	→ lhz joined	17:19
	← psoo left	17:20
	→ bentob0x joined	17:22
	→ til77 joined	17:22
	→ Theravadan joined	17:23
	← Adlai left	17:24
	→ Adlai joined	17:24
	→ qrush joined	17:25
	→ sergiodj joined	17:25
	← ablemike left	17:26
	← GNUtoo left	17:27
	→ patrikf joined	17:28
Thell	It seems that I need to reset and clean to get rid of the unlinkable submodule dir.	17:28
	→ \|Inthewings\| joined	17:29
	Thell wonders if this may be a windows issue.	17:30
Mikachu	you can't unlink open files on windows, that could be it?	17:30
	→ schacon joined	17:30
jlf`	sitaram: your suggestion did fix that issue, thanks again for your help	17:31
Thell	Mikachu: I thought of that and closed every explorer/file browser type thing I have open	17:31
	→ dreiss joined	17:32
Thell	it doesn't _seem_ to be hurting anything	17:32
	← godsflaw left	17:33
sitaram	jlf`: you're welcome!	17:33
	→ tfarina joined	17:33
	← DaveIngram left	17:34
	← Adlai left	17:35
	→ Adlai joined	17:36
	← sh1mmer left	17:36
	← hakunin left	17:37
	→ Arild4 joined	17:37
	→ DaveIngram joined	17:37
	← ixti left	17:37
	→ godsflaw joined	17:38
	← trochala left	17:38
	→ sh1mmer joined	17:38
	← InTheWings left	17:39
	→ khmarbaise joined	17:41
	← hyperair left	17:41
	→ gitte_ joined	17:41
	→ jkp joined	17:43
	← tono left	17:44
	← smuf left	17:44
	← gitte left	17:44
	gitte_ → gitte	17:45
	→ ablemike joined	17:45
	← roop__ left	17:45
	→ rudi_s_ joined	17:46
	← mattcodes left	17:46
	→ frederga_ joined	17:48
	→ patmaddox joined	17:48
	→ justin-george joined	17:49
	→ azol joined	17:49
	← lorandi left	17:52
	→ lorandi joined	17:52
	→ rcy joined	17:52
	← Arild left	17:53
	Arild4 → Arild	17:53
	← frederga_ left	17:54
	→ frederga__ joined	17:55
	→ mattcodes joined	17:55
	→ amerine joined	17:55
mattcodes	How do I add tracking to an existing local branch (experiment) that already has a few commits. I've just pushed it now to my hosting provider, and I want to link it up so in future I can git push to the branch on the git server	17:57
	→ patcito joined	17:57
	rylar_off → rylar	17:58
	→ godsflaw_ joined	17:58
jason237	git config branch.<name>.remote/merge	17:58
	→ Grahack joined	17:59
	← godsflaw_ left	17:59
	→ godsflaw_ joined	17:59
	← EricInBNE left	18:00
mattcodes	jason where do I specify the branch on the server? on the server its origin/experiment	18:00
	← godsflaw left	18:00
jason237	branch.<name>.remote is the name of the remote for the server, branch.<name>.merge is the ref on the server, i.e. refs/heads/experiment	18:01
	← mw__ left	18:01
jettero	ok yeah, the webosinternals terminal app rocks	18:02
	dunno why I never tried it before	18:02
	wrong channel, please ignore me	18:02
mattcodes	sorry jason Im confused where I do the subsitution..	18:02
	← dpino left	18:02
jason237	I'm not sure what you mean. You need to add the information about the remote branch to your .git/config	18:03
	← rudi_s left	18:03
jason237	via git config branch.experiment.remote origin; git config branch.experiment.merge refs/heads/experiment	18:04
	← tjafk1 left	18:04
mattcodes	ah so I do "git config branch.experiment.remote origin" and then "git config branch.experiment.merge /refs/heads/experiment">	18:04
jason237	yep	18:04
mattcodes	yep you beat me to it	18:04
	→ tjafk joined	18:04
mattcodes	thanks! :)	18:04
thiago	without the leading slash	18:04
	→ drubies joined	18:04
	→ GarethAdams\|Work joined	18:05
	→ sprasanna joined	18:06
	→ Sigma joined	18:07
	← loops left	18:12
	→ krisfremen joined	18:13
krisfremen	hiya everyone	18:13
	← cantoma left	18:13
	← DaveIngram left	18:13
krisfremen	i was wondering if there's something.. like a script that will auto commit whenever there's a file change in the repo?	18:13
	→ notjohn joined	18:15
krisfremen	ohhkay.. i didn't just ask that...	18:16
	my brain isn't working	18:16
	→ orospakr joined	18:16
mattcodes	on my commit messages on branches the author is showing as unknown	18:17
	ignore	18:18
	was testing between PC and laptop and didnt set up globals on laptop	18:18
	← azol left	18:18
	→ drizzd joined	18:18
	→ DaveIngram joined	18:20
	← DaveIngram left	18:21
	→ ryoma joined	18:21
	→ Nazcafan joined	18:21
Nazcafan	hello	18:21
	← pheaver left	18:22
	← mattcodes left	18:22
	← loincloth left	18:23
	→ DaveIngram joined	18:24
	← DaveIngram left	18:26
	← alienbrain left	18:26
	→ Cesario joined	18:27
	← sprasanna left	18:27
	← sh1mmer left	18:28
	→ sh1m joined	18:28
	← jerbear left	18:29
	→ wolog joined	18:30
	→ pdherbemont joined	18:31
	← psynaptic left	18:32
	← pdherbemont left	18:32
	→ pdherbemont_ joined	18:32
	→ hydrogen joined	18:34
	← tillberg left	18:34
	→ tillberg joined	18:35
	← pdherbemont_ left	18:36
	← ablemike left	18:36
	→ pdherbemont joined	18:37
	→ azol joined	18:37
	← drizzd_ left	18:38
	→ DaveIngram joined	18:38
	→ loincloth joined	18:39
	← tango_ left	18:40
	→ magcius joined	18:41
	→ mw__ joined	18:41
	← ryoma left	18:42
	→ smotchkkiss joined	18:42
	→ ryoma joined	18:44
	→ pheaver joined	18:46
	→ simplecoder_ joined	18:51
	← hydrogen left	18:52
	← simplecoder_ left	18:52
	← cmarcelo left	18:52
	← pheaver left	18:52
	→ _Vi joined	18:53
	← c15c8ra1n left	18:54
	→ cantoma joined	18:54
	← abms1116 left	18:55
	← khmarbaise left	18:55
	→ lu_zero joined	18:56
	← mbroeker left	18:56
	← icwiener_ left	18:57
	→ pheaver joined	18:57
	→ j2labs joined	18:58
	→ jceb joined	18:58
	→ sun_ joined	18:59
	→ oven joined	19:00
	→ stribog joined	19:00
	← stribog left	19:00
Thell	any ideas on why in one repo a submodule add works perfectly, and in another repo on the same machine at the same dir level that same submodule add freezes at Initialized empty git repo ?	19:00
	← solsTiCe left	19:01
sun_	hi i am trying to merge something i fetched. there's a subtree which has a bunch of tempfiles which changed between the 2 commits, and so i put it in .gitignore, but git merge says unmerged files in that tmp dir. so then i just said git commit, and it says bstat/bstate/junk[etc].html needs merge, error building trees. i put this subtree in the .gitignore but its still trying to merge it, how do i get git to forget about that subtree?	19:01
j2labs	After I've added a remote, I'm not sure whether I should prefer fetch or pull. Can someone help me understand the difference?	19:01
Thell	pull == fetch + merge	19:01
	← alley_cat left	19:02
Thell	boind just fetch, you can check and see what was fetched prior to merging	19:02
	s/boind/doing	19:02
	→ tstclair_ joined	19:02
	→ markelikalderon joined	19:02
	← Nazcafan left	19:02
Thell	AHA! It didn't freeze, just took forever to do. Yippie.	19:02
sun_	j2labs the best advice i got was to use fetch and then merge until you understand what they do, then experiment with pull and see if it still does what you want, since it has a lot of defaults which do things you dont want and wont notice until much later	19:03
oven	I use git-svn. I have a release branch in svn called r12 and a local-r12 tracking that. I have master tracking svn trunk. I've done a bug fix on local-r12, and dcommit-ed it. Now I want that change in trunk as well. Should I cherry pick it or try something like merging local-r12 onto master?	19:03
sun_	hi anyone know how to get git merge to ignore a subtree of tmpfiles after they've been fetched? putting it in .gitignore is not working..	19:04
	← patmaddox left	19:05
oven	sun_: are the tmpfiles in git? Should they be? If not, why not delete them?	19:05
sun_	oven yes they are. how do i delete them?	19:06
oven	sun_: git rm <filename>, then git status to show what you've done, then git commit -m "deleted some tmpfiles"	19:06
	← jchris left	19:06
	← tstclair left	19:07
	→ meanburrito920_ joined	19:07
Thell	they were fetched though right? Not merged? So they aren't in your local branches?	19:07
	← tillberg left	19:07
	→ elmob joined	19:07
	→ tillberg joined	19:08
bnovc	git forced me to do a merge commit when i pulled in a change and i can't figure out how to correctly compact these commits	19:08
	i tried `git rebase -i HEAD~1` but it only lists the 2nd item and it says i need two things	19:08
	i tried manually listing my HEAD above it and that failed as well	19:08
	← patrikf left	19:09
	← cantoma left	19:10
	← magcius left	19:11
sun_	oven thanks! worked like a charm!!	19:11
	← sun_ left	19:11
j2labs	git fetch name_of_remote is the right syntax, correct?	19:14
	cuz i think github just hung up on me... :(	19:14
	→ ablemike joined	19:15
oven	anyone know a bit of workflow in svn with remote branches?	19:15
	← DaveIngram left	19:15
	impulze looks at the channel name	19:16
	→ magcius joined	19:16
j2labs	oven: this is #git	19:16
oven	impulze: well, I meant git-svn	19:16
	sorry	19:16
	← herlo left	19:16
	← fualo_ left	19:16
oven	is there a dedicated channel for git-svn?	19:17
	→ fualo_ joined	19:17
MisterN	oven: why should there be?	19:17
	← plediii left	19:18
oven	I've read about every blog entry about using git-svn with remote branches, but not a single one of them has mentioned how to merge or copy changes from one remote branch to another	19:19
	← vvsh left	19:19
oven	anyone have real experience doing this?	19:19
	or a pointer to someone who has written about it?	19:19
	← tedoc2000 left	19:19
	← Adlai left	19:20
oven	I could of course not do this with git-svn, but instead use my svn client directly against the repository to do the merge, but that would be a bit like admitting failure.	19:20
	← godsflaw_ left	19:20
	→ darrinholst joined	19:21
	→ mnick joined	19:22
darrinholst	Is a git push over ssh atomic? i.e. What happens when 2 pushes happen at the same time?	19:22
thiago_home	it's atomic per branch	19:23
	← ablemike left	19:23
thiago_home	either you update the branch, or you don't	19:23
	however, if you push 2 branches and someone else pushes to one of your branches, you may get one branch updating but not the other	19:24
	→ przemoc joined	19:24
oven	MisterN: because noone seems to want to say anything about git-svn here :)	19:24
thiago_home	oven: what other questions (besides yours about overflow, which I didn't understand) went unanswered?	19:25
Ilari	darrinholst: Without forcing, first push succeeds and the second gets either non-fast-forward error or locking failure. If second forces, it will overwrite first user's changes or get locking failure.	19:26
	darrinholst: (and this happens per ref/branch).	19:26
darrinholst	Ilari: ok, thanks? So that locking happens in git-receive-pack?	19:27
Ilari	darrinholst: Compare-and-swap operation on ref (branch).	19:27
	← jrmuizel left	19:28
	→ jrmuizel_ joined	19:28
Ilari	One gets locking failure if the compare and swap fails (incorrect old value).	19:28
	jrmuizel_ → jrmuizel	19:28
	→ godsflaw joined	19:28
	← _Vi left	19:28
darrinholst	cool, thanks	19:28
Ilari	darrinholst: All "smart" transports (file://, ssh:// and git://) have the same atomicity in pushes (since they share the layer 7 protocol).	19:29
	darrinholst: (not that git:// is usable for pushing due to lack of authentication).	19:30
oven	thiago_home: I've got a remote branch in svn called r12, and a local-r12 that tracks it. Then I've got master tracking svn trunk. Now, I've committed a bugfix to local-r12, and dcommit-ed it to r12. I want that bugfix on trunk as well. What to do?	19:30
	Would it be wise to do git checkout master ; git pull . local-r12, for instance?	19:30
Ilari	oven: checkout master, cherry-pick the fix and dcommit.	19:31
oven	or should I try cherry picking the fix?	19:31
	Ilari: right	19:31
Ilari	oven: Merging (pull is fetch + merge) and SVN don't mix.	19:31
oven	I suspected as much...	19:31
thiago_home	oven: cherry-pick, remove the git-svn-id lnie	19:31
	line	19:31
oven	thiago_home: oh?	19:31
	→ cantoma joined	19:31
oven	thiago_home: why is that?	19:32
thiago_home	git-svn needs to be told that it's a new commit, not one from SVN	19:32
	→ \|azol\| joined	19:32
oven	oh...	19:32
thiago_home	if it sees the git-svn-id line, it thinks the commit came from SVN	19:32
	then it won't send it again	19:32
oven	oops, that's scary	19:32
	I may have done that already	19:32
	← yann left	19:34
Ilari	oven: Wonder what it did...	19:34
oven	Ilari: it looks like it got a new git-svn-id, actually	19:36
Ilari	oven: Does that ID look reasonable?	19:37
	→ Adlai joined	19:37
	→ rfay joined	19:37
rfay	I'm afraid this is a stupid question, but I'm not finding it. I want git diff to always use --no-prefix. Can I set that with git-config? Or somehow?	19:38
	← darrinholst left	19:38
	← azol left	19:38
oven	Ilari: http://pastebin.com/m2f209517	19:38
	Ilari: Yes, I would say so	19:39
Adlai	rfay: if it's possible, it'll be under the man page for git-config	19:39
	man git-config	19:39
Gitbot	Adlai: the git-config manpage can be found at http://git.or.cz/man/git-config	19:39
	→ DaveIngram joined	19:39
oven	Ilari: this example is the other way around, though... a bug fix on trunk that needed to be on r12	19:39
rfay	Adlai, thanks. But I don't find any listing there. I figured this might be a more generic thing, like there's a way to set the default args for any command.	19:39
	← ^Einstein left	19:40
Adlai	rfay: doesn't look like it's possible to me, either.	19:41
rfay	Adlai, much appreciated.	19:41
oven	Ilari: the sha1 is of course different as well, as it's been rebased onto the other branch	19:41
	→ markelik_ joined	19:42
	← markelikalderon left	19:42
	← til77 left	19:42
oven	I've got a feeling that svn and git don't mix all that well :) There are some very fundamental differences that show up particularly well when you use multiple branches in svn :)	19:42
	→ yann joined	19:43
	← jceb left	19:45
	→ MarkAlanEvans joined	19:45
MarkAlanEvans	What do i do about an error like this: error: Untracked working tree file 'app/tmp/cache/cake_model_default__zuujit_item_types_list' would be overwritten by merge.	19:45
	→ carrus85 joined	19:45
	← pflanze left	19:46
	← kiva left	19:47
	→ kiva joined	19:47
	← meanburrito920_ left	19:48
	← Grahack left	19:49
	← cantoma left	19:49
	→ funkycat90210_ joined	19:49
Ilari	oven: Git has nonlinear branches. SVN "branches" are linear.	19:49
jsquared	Ilari: can you clarify what you mean by that?	19:50
	← funkycat90210_ left	19:50
	→ ^Einstein joined	19:50
	← boto left	19:50
	← marvil07 left	19:50
	← jforman left	19:51
Ilari	jsquared: Git branches can contain merges and thus fail to be linear. SVN doesn't have true branches and underlying revision model is linear. The "branching" layer doesn't implement any nonlinear behaviour.	19:51
	sh1m → sh1mmer	19:52
Ilari	jsquared: This holds true even in 1.5+ merge tracking.	19:52
gitte	jsquared: the main point being: it is not enough to track merges; you have to make _use_ of them, too.	19:52
	→ plediii joined	19:53
gitte	jsquared: for example in Git, if you have criss-cross merges (a clear feature of non-linear branching), Git will know about all merge bases and make use of them when you merge the next time.	19:53
	→ krisserver joined	19:53
	← DaveIngram left	19:53
	← godsflaw left	19:53
	← Fragsworth left	19:54
Ilari	And furthermore, not just are commits from other branches referenced, they are part of branch itself (this can be seen by requesting just a single branch in repository with merges).	19:55
	← elmob left	19:55
	→ kennethlove joined	19:55
	← \|azol\| left	19:55
jsquared	gitte, Ilari: by "nonlinear", do you just mean that two commits might have two common (and non-identical) ancestors?	19:56
gitte	Ilari: well, the SVN repository being purely remote, the same can be said about SVN. The repository trivially contains all referenced objects...	19:56
	jsquared: not only that.	19:57
jsquared	okay	19:57
gitte	jsquared: there can be _tons_ of parents.	19:57
	gitte looks for her favorite gitk screenshot	19:57
	→ ajonat joined	19:57
jsquared	is that a pathological case or something that actually happens?	19:57
Ilari	jsquared: Something that actually happens.	19:57
jsquared	hmm, okay	19:58
	why would one commit have tons of parents? I'm not understanding something here	19:58
Ilari	jsquared: Of course, if you overdo it, prepare get reputation of doing so... Like the ACPI guys (making 12-parent merge and AFAIK also 13-parent one).	19:58
gitte	jsquared: you are stating the obvious...	19:58
	→ Nazcafan joined	19:59
	→ ablemike joined	19:59
gitte	Ah!	19:59
	http://www.spearce.org/2007/07/difficult-gitk-graphs.html	19:59
jsquared	okay, so it's something that actually happens, but not something that's seen as a terribly good idea? :P	19:59
	haha, jesus	19:59
Nazcafan	hello	19:59
jsquared	gitte: that looks like a PCB wiring diagram ;)	20:00
Ilari	...	20:00
	→ eletuchy_ joined	20:00
gitte	jsquared: It is always a question of intelligent decisions...	20:00
	← krisserver left	20:00
gitte	jsquared: in some circumstances, octopodes are the _only_ thing that makes sense.	20:00
	→ krisserver joined	20:00
Ilari	jsquared: And even has four lines jumping to "hyperspace"... :-)	20:00
	← JNR___ left	20:01
	→ DaveIngram joined	20:01
Ilari	jsquared: (gitk does that when it has insufficent space to draw all the lines)...	20:01
Nazcafan	I am quite a newbie using git	20:01
	→ pflanze joined	20:02
Nazcafan	I pulled from a server, but it's ip address recently changed, how can I push to it, now?	20:03
Ilari	Nazcafan: Edit '.git/config'.	20:03
Nazcafan	ah thanks	20:03
	← smotchkkiss left	20:03
Ilari	Nazcafan: Should be pretty obivious what's the right thing to change there.	20:04
Nazcafan	yeah, got it	20:04
	at my work, they are still using CVS	20:04
	→ tedoc2000 joined	20:05
	← schacon left	20:05
Nazcafan	they are planning to switch to something else, but unfortunately, they seem to have settled for bazaar	20:05
Ilari	Nazcafan: If the new repo is empty, one needs to push to populate it...	20:05
Nazcafan	mhh	20:06
Ilari	Nazcafan: Otherwise (it already has the stuff), fetching the new repo (git remote update) could be in order.	20:06
	← capndiesel left	20:06
	→ By-Gokhan joined	20:06
	← ablemike left	20:06
Nazcafan	Ilari, sorry I didn't catch your latest statement	20:07
	← tedoc2000 left	20:07
By-Gokhan	http://www.seslibitanem.com :) http://chatvebiz.tr.gg :) http://www.seslibitanem.com :) http://chatvebiz.tr.gg :) http://www.seslibitanem.com :) http://chatvebiz.tr.gg :) http://www.seslibitanem.com :) http://chatvebiz.tr.gg :) http://www.seslibitanem.com :) http://chatvebiz.tr.gg :) http://www.seslibitanem.com :) http://chatvebiz.tr.gg :) http://www.seslibitanem.com :) http://chatvebiz.tr.gg :) http://www.seslibitanem.com :) http://chatvebiz.tr.gg	20:07
	http://www.seslibitanem.com :) http://chatvebiz.tr.gg :)	20:07
Nazcafan	ilari ?	20:07
	← eletuchy left	20:07
	→ tedoc2000 joined	20:08
	← By-Gokhan left	20:08
	→ _Vi joined	20:08
	→ a3li joined	20:08
Ilari	Nazcafan: If the new repo already has stuff, running 'git remote update' could be in order...	20:08
	Nazcafan: updates remote tracking branches...	20:08
Nazcafan	ah	20:09
	→ radii joined	20:09
Nazcafan	I'll try that	20:09
	→ Octalot joined	20:09
	← eletuchy_ left	20:09
Nazcafan	I am the only to use the remote repo, actually	20:09
	it is kind of a private repo	20:09
	→ pstef joined	20:10
	← krisfremen left	20:10
	← plediii left	20:10
Nazcafan	(you could call it a backup)	20:10
impulze	well it teaches you to not fall into bad habits	20:11
	→ meanburrito920_ joined	20:11
Ilari	Nazcafan: When computer is going offline, its handy to have all relevant related repositories as remotes and to update all the remote tracking branches, so one can work on merging stuff without net connection.	20:11
Nazcafan	ah	20:12
	got it, nice, actually	20:12
Ilari	Nazcafan: remote tracking branches are mirrors of local branches of other repositories.	20:13
	← magcius left	20:15
	→ smotchkkiss joined	20:16
	→ Sent joined	20:16
	→ giallu joined	20:16
	← ntoll left	20:16
	← rcy left	20:17
Nazcafan	Ilari, what is the difference between git remote update and git push, when the repo already has stuff ? none ?	20:21
	→ smotchkkiss_ joined	20:22
	← frede left	20:23
	→ jchris joined	20:24
Ilari	Nazcafan: git remote update runs 'git fetch <remote>' in loop. It doesn't push.	20:25
Nazcafan	ah	20:25
	so right now the remote repo is unchanged, right ?	20:26
thiago_home	yes	20:26
Nazcafan	so if I saw no conflicts, I just need to push, now ?	20:27
	→ loinclot_ joined	20:27
	→ cxvii joined	20:27
	← corni left	20:27
	→ montylounge joined	20:28
thiago_home	update didn't pull	20:28
	you may not have the latest changes from upstream in your branch	20:28
	← frederga__ left	20:29
Nazcafan	thiago, If I remember, pull = fetch + merge, right?	20:30
impulze	yes	20:30
thiago_home	right	20:30
Nazcafan	so now I just have to merge, I guess	20:30
thiago_home	yes	20:31
	→ magcius joined	20:31
	← loincloth left	20:31
	→ froschi joined	20:31
Nazcafan	mhhh	20:32
wagle_home	how would you list commits (say with git-rev-list) where a file (path) is modified by a merge commit (from the second parent)?	20:32
Nazcafan	thiago_home, I am not sure of the proper syntax to proceed with the merge	20:33
	← sh1mmer left	20:33
	→ sh1m joined	20:33
	← gitte left	20:33
	← DaveIngram left	20:34
thiago_home	git merge <branchname>	20:34
	→ DaveIngram joined	20:35
Nazcafan	thiago_home, how do I know the branch name? git branch only yields "* master"	20:36
thiago_home	remotename/branchname works too	20:36
	see git branch -r	20:36
Nazcafan	aaah	20:36
	there I see the remote branches	20:37
	← smotchkkiss left	20:37
	← lucsky left	20:37
Nazcafan	I see origin/HEAD, and origin/master	20:37
	← sergiodj left	20:39
	→ adamtanner joined	20:39
	→ smotchkkiss joined	20:40
	→ godsflaw joined	20:40
	→ cenuij joined	20:41
	← krisserver left	20:41
	→ krisfremen joined	20:41
	← gh34 left	20:41
Ilari	Nazcafan: origin/HEAD is the default one (symbolic refernce, kind of like Unix symbolic link). 'origin/master' is copy of 'master' of 'origin'.	20:44
	← Thomas__ left	20:44
	→ godsflaw_ joined	20:44
	dwmw2_gone → dwmw2_OLM	20:44
	→ Thomas__ joined	20:44
seanius	is there an easy way to gpg sign a series of patches output from git format-patch?	20:45
	← kumbayo left	20:46
Fissure	seanius: don't think so	20:47
Nazcafan	Ilari, I am kind of lost, there, so where does origin/HEAD points to if it is similar to a simlink? origin/master?	20:47
Ilari	Nazcafan: Correct.	20:47
Nazcafan	ok	20:47
	so it is the same to merge	20:48
Fissure	seanius: you could in theory use a bundle instead, and include a signed tag in it	20:48
Ilari	Nazcafan: refs/remotes/origin/HEAD (remote tracking branch 'origin/HEAD') can be shortened as 'origin'.	20:48
	→ mmw joined	20:48
	← cxvii left	20:48
	adamtanner → cxvii	20:48
Ilari	Nazcafan: Merge also accepts commit IDs. How commits to merge are specified is irrelevant except for the default commit message...	20:49
	← x_or left	20:49
Ilari	Nazcafan: Note that commit IDs to merge are interpretted as anonymous branches tipped at designated commit.	20:49
	← froschi left	20:49
	← Sent left	20:49
	← cxvii left	20:50
Nazcafan	ok, I merged and it indicated "already up to date" which was predictable since noone had touched the remote repo since I had last pulled	20:50
	now am I supposely safe to push there?	20:51
	→ Pewpewarrows joined	20:51
	← smotchkkiss_ left	20:52
seanius	Fissure: okay, thanks. how do i output a bundle?	20:52
Fissure	seanius: man git-bundle	20:53
Gitbot	seanius: the git-bundle manpage can be found at http://git.or.cz/man/git-bundle	20:53
	← ^Einstein left	20:53
seanius	Fissure: oh, heh. zsh didn't tab complete it so i didn't realize it was there :)	20:53
	← keystr0k left	20:54
	← sh1m left	20:55
	→ sh1mmer joined	20:55
	engla wishes everyone in #git would use the 'git help foo' form when teaching	20:55
	engla takes the example 'git help bundle'	20:55
	← notjohn left	20:56
	← montylounge left	20:56
	← Swimming_Bird left	20:57
	fpcfan → fpcfan-away	20:59
	← godsflaw left	20:59
	→ doctorcode joined	21:00
impulze	or git foo --help	21:00
	← tedoc2000 left	21:01
	← zirpu left	21:01
doctorcode	if i add some files to branch A and switch to another branch B, those new files from A don't seem to be deleted. is that normal? what's the concept behind that?	21:01
	→ ^Einstein joined	21:01
Nafai	doctorcode: You have only done 'git add' but not 'git commit' before switching to B?	21:02
	→ kvanderw joined	21:02
doctorcode	Nafai: i commited of course	21:02
Nafai	Ok, just making sure.	21:02
	That does seem odd.	21:02
thiago_home	they should disappear if the other branch doesn't have them	21:02
doctorcode	it seems that when switching between branches those files that are not tracked in the other branches just remain in its place	21:03
	→ tedoc2000 joined	21:03
doctorcode	ok - i will recheck that	21:03
	thx	21:03
kvanderw	trying to maintain a local repository server and github.... while I have the local repository working.... linking to github (any secondary repository) seems like some kind of magic. I want to be able to have external programmers hitting github, and corporate programmers hitting the local repo server.... Any clues on where to look? jfgi hasn't got me much more than confused.	21:04
thiago_home	"linking" ?	21:06
	← lhz left	21:08
kvanderw	distributed repositories.... how to get the two 'repository' servers to update each other.	21:09
kevlarman	kvanderw: for the non-github one, you can use a post-update hook	21:10
	to push to github	21:10
kvanderw	I have seen the --mirror command, but that seems to fail the situation where external contractor is updating 'github' and we want our repository server to reflect the same.	21:10
	→ fynn joined	21:10
kvanderw	yes... the post-update hook is where I have the --mirror	21:11
	→ BeerSerc_ joined	21:11
	← BeerSerc left	21:11
kvanderw	but... how do changes to 'github' show up on our local hub?	21:11
kevlarman	i don't think you want --mirror	21:11
fynn	Hey. Suppose I want to reduce the size of the repo to the minimum, which is better: "repack -adf --window=100 --depth=250" or "repack -adf --window=250"	21:11
kvanderw	that is where I am landing as well. and the reason for this discussion.	21:11
	:)	21:11
kevlarman	i don't think you can make it work in 2 directions actually	21:12
kvanderw	can I get to 'post-update' on github?	21:12
kevlarman	you'll run into problems when something gets pushed to the same branch at the same time	21:12
kvanderw	so they update each other?	21:12
kevlarman	not afaik	21:12
kvanderw	ok... I see that	21:13
kevlarman	why can't you just use a single repo?	21:13
kvanderw	so, as many 'human' users can be distributed as I like, but multiple repos seems to be a problem.	21:13
	because I thought I could <shrugs>	21:14
	→ cantoma joined	21:14
kvanderw	I am suffering from growth.	21:14
	I thought I would use 'github' for external contractors.	21:14
	and our local repo for staff.	21:14
doener	kvanderw: I'd rather have your external guys have just the github repo as a remote, and the "internal" guys having both repos as remotes	21:14
kvanderw	it sounded good at the time	21:14
	so, you are referring to mulitple URL lines in the config?	21:15
doener	multiple remote sections even	21:15
	← Tuomas left	21:15
doener	so external guy has just "origin" (pointing to github), and internal guy has "origin" (internal repo) and "github" (github repo)	21:15
	or something like that	21:16
	→ godsflaw__ joined	21:16
kvanderw	doener: the question there is, if I have multipel remote sections, don't I need to push once for each remote? and with one remote but multiple URLs, it is a single push??	21:16
	← godsflaw__ left	21:16
	→ godsflaw joined	21:16
	← Theravadan left	21:16
doener	then internal devs can just fetch from/push to both repos	21:16
	← godsflaw_ left	21:16
	← engla left	21:16
doener	kvanderw: I'm not sure that a single remote can have multiple urls at all	21:16
	← fualo_ left	21:17
	→ fualo_ joined	21:17
doener	kvanderw: and you'd just hit the same "someone else pushed, thus my push gets rejected" issue anyway	21:17
fynn	Hm, to create an alias that runs a bash script, do I still need to use the !"<bach commands>" syntax?	21:17
kvanderw	doener: me neither... but I saw specific instructions to that effect during my googling today.	21:17
doener	kvanderw: unless one repo is a pure mirror (read-only for users)	21:18
doctorcode	i just did git checkout f47........... which was some older commit. are my newer changes not lost?	21:18
doener	doctorcode: no	21:18
	→ krisserver joined	21:18
doctorcode	i don't seem them in "git log"	21:18
	(its not dramatic, its just a test scenario)	21:18
doener	doctorcode: because "git log" is the same as "git log HEAD", and HEAD references what you have checked out	21:19
kvanderw	doener: right, I have gotten far enough to determine that I could do the 'r/o' thing... but that wasn't really what I wanted.	21:19
doctorcode	how can i get back?	21:19
	(thx)	21:19
doener	doctorcode: just do "git log master" or so, and see that your branch head wasn't affected by the checkout	21:19
	doctorcode: just checkout your branch head again	21:19
doctorcode	doener, thx!	21:20
kvanderw	doener: Thanks for the help... I'll go off and 'play' some more.... :)	21:21
	← WinterWeaver left	21:21
	→ delsvr joined	21:21
	← fhr left	21:23
doctorcode	Nafai: I just tested what you said in a nearly empty git repo ... seems to work fine. thank you	21:24
Nafai	np	21:25
	loinclot_ → loincloth	21:26
	→ tfarina_ joined	21:28
	← mm_work left	21:28
	→ joab joined	21:29
	← krisserver left	21:32
	← smotchkkiss left	21:32
	→ lightcap joined	21:33
	→ Seventoes joined	21:33
Seventoes	git-svnimport doesn't seem to be installed on my system, is there a place to download it?	21:34
	← krisfremen left	21:35
fynn	is there a single command that would not only commit --all, but also add any non-ignored untracked file before committing?	21:35
	basically, a single command for "git add -A; git commit"?	21:36
	← Nazcafan left	21:36
impulze	just create an alias	21:37
joab	I had to reformat my server so I did a clone --bare on my repo onto my work station. How do I get it back on the server? Can I just tar it and upload?	21:37
	← dodo_the_last1 left	21:37
	← tillberg left	21:38
	← _Vi left	21:38
	→ plediii joined	21:39
segher	joab: the easiest way is to just mkdir && git init --bare on the server, and git remote add origin && git push origin master on the client	21:39
joab	segher: From the bare repository or the one with the working directory?	21:40
segher	either	21:40
	you didn't really need the bare copy afaics	21:41
	← Seventoes left	21:41
segher	every clone is a full copy (of the branches and tags it cloned only, of course)	21:41
joab	segher: I already have origin and all that set up from earlier on so I should just be able to init --bare on the server and then push?	21:41
segher	yeah	21:41
joab	segher: Sweet. Thanks :)	21:42
	→ nils13 joined	21:42
	← tfarina left	21:43
	← nils13 left	21:43
	← flazz left	21:43
	← LiamH left	21:45
	← maravillas left	21:46
	→ Theravadan joined	21:47
	→ __iron joined	21:47
	← altrux left	21:47
	→ ablemike joined	21:48
	← loincloth left	21:48
	← freerobby left	21:48
	← pdherbemont left	21:49
	→ tango_ joined	21:49
	← lightcap left	21:49
	← Sigma left	21:50
	→ zmbmartin joined	21:50
	← jchris left	21:50
PointMan	joab: yep	21:50
zmbmartin	how do I get pull when it says that I am behind by 1 commit	21:51
	→ eletuchy joined	21:51
zmbmartin	sorry I meant git pull when I am behind by 1 commit	21:52
	← hugoxrosa left	21:53
	← lorandi left	21:54
	← eletuchy left	21:54
	→ eletuchy joined	21:55
fualo_	can I reset an interactive rebase?	21:55
impulze	reset as in abort?	21:55
fualo_	oh, that works? easy. thanks	21:55
	← ccherrett left	21:55
impulze	:S i was merely asking	21:55
	did not know what you mean with reset	21:55
	← m-takagi left	21:59
	← eletuchy left	22:01
	← priidu left	22:03
	← tbf left	22:04
zmbmartin	fatal: You have not concluded your merge. (MERGE_HEAD exists) <-- How can I fix this	22:04
	← ctusar left	22:04
	→ psynaptic joined	22:05
	← doctorcode left	22:05
	→ mybad joined	22:06
mybad	hello	22:06
	i'm using git for few days now (switching from subversion) and i'm enjoing using it	22:06
	← meanburrito920_ left	22:06
	→ indirect_ joined	22:07
mybad	i would like to know, what's the best alternative to have a "build versions"(like what we can do with subversion) using git please	22:07
thiago_home	what's a "build version" in SVN?	22:07
wereHamster	mybad: <HEAD:GIT-VERSION-GEN> and the corresponding rules in the Makefile	22:08
Gitbot	[git GIT-VERSION-GEN]: http://tinyurl.com/lowbne [blob]	22:08
mybad	thiago_home, I've meant that there are a few scripts floating around that perform this trick for subversion (geting incremental version numbre from subversion API)	22:08
	Gitbot, thank you	22:09
	← zmbmartin left	22:09
	← DavidKlein left	22:09
impulze	haha wereHamster, pwned ;)	22:09
	rfay → rfay_afk	22:09
wereHamster	what, why?	22:09
impulze	you weren't thanked :)	22:10
	→ boto joined	22:10
impulze	instead the bot got the honors :D	22:10
wereHamster	I'm ok witn that, I'm sure the bot appreciates it	22:10
	→ jerbear joined	22:10
fynn	wereHamster: Suppose I want to reduce the size of the repo to the minimum, which is better: "repack -adf --window=100 --depth=250" or "repack -adf --window=250"	22:10
wereHamster	fynn: ugh, no idea	22:11
thiago_home	fynn: YMMV	22:11
fynn	wereHamster: who should I ask?	22:11
thiago_home	fynn: yourself	22:11
	← cenuij left	22:11
thiago_home	fynn: run both and check which one produces the best result	22:11
fynn	thiago_home: is this one of those Zen answers	22:11
	I can't run it against all repos in the world :)	22:12
thiago_home	no, it's the actual answer	22:12
	the result will depend per repository	22:12
	← suman_ left	22:12
thiago_home	hence the YMMV	22:12
	indirect_ → indirect	22:12
	→ eletuchy joined	22:12
wereHamster	fynn: I think --window=100 will yield smaller packs	22:13
fynn	wereHamster: yeah, I sort of suspect the former is better too, and initial benchmarks suggest it is.	22:14
	wereHamster: I took the later from http://git.or.cz/gitwiki/GitBenchmarks#Git.2CMercurial.2CBazaarrepositorysizebenchmark	22:14
	which is why I'm asking.	22:14
	→ psoo joined	22:14
fynn	they were aiming to absolutely minimize repo size there, and since they were suggeting the later, even though the former seems to work better, I assumed I was missing something :)	22:15
	→ Derander joined	22:15
	← mivok left	22:15
	← kpreid_ left	22:18
	→ telemachus joined	22:18
telemachus	A question about branches: If I have a branch that I expect to be short-lived, does it pay to create a remote version of it (and push commits to that remote branch) or is it easier to just merge (rebase?) the branch back into master and push that?	22:19
	→ wildwobby joined	22:19
wildwobby	I hate git.	22:19
qrush	...why?	22:20
wildwobby	I did a "git rm -rf" in attempt to drop track of all of the files in the repo... but it DELETED all of them	22:20
wereHamster	telemachus: only if you want to share it	22:20
wildwobby	I am used to hg.	22:20
	Now I lost my whole damn project	22:20
	Is there a way to get it back?	22:20
wereHamster	wildwobby: git checkout HEAD -- .	22:20
	← jrmuizel left	22:20
telemachus	wereHamster: you mean only create the remote if I want to share it? To be honest, it's a personal project (nobody here but me), but I'm curious about the implications for later history...	22:21
wildwobby	fatal: invalid reference: HEAD	22:21
	It was going to be an initial commit	22:21
	But I added the wrong files at first.	22:21
	FML.	22:21
telemachus	back in two	22:22
	telemachus → telemachus-laund	22:22
kvanderw	doener: FYI from our previous conversation about 2 urls.... http://stackoverflow.com/questions/165092/can-i-push-to-more-than-one-repository-in-a-single-command-in-git	22:22
wereHamster	telemachus: doesn't matter to the historu	22:22
kevlarman	wildwobby: i think you can still find the index	22:22
wildwobby	kevlarman: What do you mean?	22:23
	→ freerobby joined	22:23
kevlarman	err, but you removed all the files from it	22:23
	no nvm	22:23
wildwobby	the only thing that is left is the .git directory.	22:23
wereHamster	wildwobby: not the index, but amongst fsck --lost-found	22:23
d-snp	lol wildwobby that's a bold command to make :D	22:24
wildwobby	wereHamster: ??	22:24
kevlarman	wildwobby: --cached is the option you were looking for btw	22:24
fualo_	Does gitweb extract data for it's visualization via git commands? or deep internals?	22:24
	→ dodo_the_last joined	22:25
	wereHamster doesn't understand why someone would think that 'git rm -rf' behaved differently than 'rm -rf'	22:25
d-snp	I can sort of imagine making that mistake	22:25
	though I wouldn't dare issueing anything like that on untracked files...	22:26
	eh, or uncommitted	22:26
wildwobby	wereHamster: Well in hg they have an "hg remove"	22:26
	And that doesnt delete the file, just drops track	22:26
d-snp	goes to show how important it is to backup at all times ;)	22:26
kevlarman	wildwobby: git rm --cached does that	22:27
d-snp	but the git thing has -rf :\|	22:27
	at least drop the f :P	22:27
fualo_	er, I meant gitHUB, does github extract data for it's visualization via git commands? or deep internals?	22:27
wildwobby	I ran it without -rf at first and it said: error: 'build/WWGraph.build/WWGraph.pbxindex/imports.pbxbtree' has changes staged in the index	22:27
	Like a million times	22:27
	← joab left	22:28
	← dodo_the_last left	22:28
	→ dodo_the_last1 joined	22:28
kevlarman	that was git trying to keep you from shooting yourself in the foot	22:28
wereHamster	wildwobby: do yol have another backup of the files?	22:28
	→ loincloth joined	22:28
wildwobby	wereHamster: Nope. And luck has it I deleted my Time Machine backups last night!	22:28
d-snp	lol	22:28
	haha	22:28
wildwobby	I wanted to use the space for other things.	22:28
d-snp	it was sad, now it's becoming ridiculous :P	22:28
wereHamster	wildwobby: so prepare for a _long_ day (or night)	22:28
	wildwobby: git fsck --lost-found	22:29
	← agile left	22:29
wildwobby	wereHamster: Whats that do?	22:29
d-snp	you know there's manual pages..	22:30
wildwobby	I know....	22:30
wereHamster	will print lots of hashes. For each one: git show <hash> to see what file that is, then redirect that to its original location (git show <hash> > build/WWGraph.build/WWGraph.pbxindex/imports.pbxbtree)	22:31
	→ kvanderw_ joined	22:31
wildwobby	Oh sweet! So I just have to come up with the filenames?!?	22:31
wereHamster	or wait. Look for 'dangling tree' in the output	22:32
Mikachu	wereHamster: there won't be any trees if he never committed	22:32
wereHamster	do 'git show' on each one until you find the root tree	22:32
	telemachus-laund → telemachus	22:33
wereHamster	oh, right.	22:33
	← mmw left	22:34
	→ travisjeffery joined	22:34
wildwobby	So I have to come up with the file names?	22:34
wereHamster	yes, git doesn't save those until you commit	22:35
	← travisjeffery left	22:35
	→ Z3R0 joined	22:35
	→ doctorcode joined	22:35
wereHamster	(only in the index, which you wiped)	22:35
	← bentob0x left	22:35
doctorcode	i just perfomred a git commit - what does this mean? ... rename helpers/trace/{ => lib}/trace.c (93%)	22:35
	especially the 93%?	22:35
wildwobby	wereHamster: Ok... not as bad as I thought... there are only like 8 source files the rest are generated by x-code	22:35
	← robinsk left	22:36
wildwobby	Thanks guys!	22:36
wereHamster	doctorcode: similarity between the two files	22:36
	← fynn left	22:36
doctorcode	wereHamster, cool. so its not just a rename - i also made changes to the file	22:36
wereHamster	seems like it	22:37
doctorcode	and git is 93% sure that these files are the same?	22:37
Mikachu	it is 100% sure they are 93% the same	22:37
doctorcode	or is it just some user information, and even 0% wouldn't matter?!	22:37
	great... thx	22:37
	← davido left	22:37
Mikachu	it's just informational	22:37
	the info isn't stored anywhere	22:37
doctorcode	okay - i just thought git needs to be somehow smart to identify those two files... and it might even fail with this...	22:38
	but great! thx	22:38
	git is so great!!	22:38
	← kvanderw left	22:38
	kvanderw_ → kvanderw	22:38
doctorcode	it really keeps me on track... everytime i type "git status" i know what i'm realy doing. and i always keep on cleaning up	22:38
	← thiago_home left	22:38
doctorcode	hehe ;)	22:39
	← pflanze left	22:39
	← tango_ left	22:39
	→ tango_ joined	22:40
	→ maattd_ joined	22:40
	← TeckniX left	22:42
	← telemachus left	22:44
	→ [t0rc] joined	22:45
	→ __tcpip joined	22:45
__tcpip	Anyone know if there is a way to "git add" a folder that is within an larger excluded folder?	22:46
	right now it comes back and says... fatal: '/system/profiles/avatars/135/687/204' is outside repository	22:46
	→ indirect_ joined	22:47
d-snp	__tcpip: in the filesystem a directory contains all the pointers to the files	22:47
	if you exclude that, git can't know where the file is	22:47
	so you really would have to make your exclude more specific	22:47
__tcpip	d-snp: I want most of the avatars subfolders excluded	22:48
	d-snp: just 135 should be included	22:48
Ilari	__tcpip: Did you really mean to use absolute path?	22:48
d-snp	I'm not an expert on exclude patterns, I only know how directories work..	22:48
	dwmw2_OLM → dwmw2_gone	22:49
__tcpip	oh lord.... i'm embarrassed	22:49
Ilari	__tcpip: I am not entierely sure how to un-ignore only that directory. But one can definitely force-add files there.	22:49
	← maattd left	22:49
	maattd_ → maattd	22:49
__tcpip	I just did a git add . while in the folder and it seemed to work!	22:49
	← Turias left	22:50
Ilari	__tcpip: Do the files show up in 'git ls-files'?	22:50
	→ ereslibre joined	22:50
__tcpip	yep	22:50
	jerbear → jmcantrell	22:52
	→ telemachus joined	22:52
	→ kpreid_ joined	22:52
telemachus	Why would I get only an "Already up-to-date" message when I try to merge a branch back into master?	22:53
	← DaveIngram left	22:53
	→ agenteo joined	22:54
	← jmcantrell left	22:55
	→ meanburrito920_ joined	22:55
	→ jmcantrell joined	22:55
	← ablemike left	22:55
	→ Cheaterman joined	22:56
Ilari	telemachus: Are you sure you are on master? What does 'git log master..branch-to-merge' show?	22:57
xymbol	telemachus: make sure you're in the right place; git status is your friend	22:57
	→ KyleMac2 joined	22:57
Cheaterman	Wow, many people here. Hello and sorry for the disturbance (especially since I'm not sure weither if this is supported or not...); I'm trying to checkout a cvs repository over git, and it doesn't work well ; I get answered with a "unknown scrambling method". It's a sf.net cvs, and I'm supposed to log in as an anonymous user before checking it out, but dunno how to do this with git. One last thing, the command I used: git cvsimport -v -d:pserver:	22:57
telemachus	xymbol: thanks, but I have my branch in my prompt at all times	22:58
	Ilari: ditto	22:58
Cheaterman	And please keep in mind that I'm familiar with git, but not at all with cvs.	22:58
xymbol	(maybe you've not checked out the new branch and worked on master)	22:58
telemachus	I just realized that I fucked up and rebased backwards (thus wiping out a commit)	22:58
	← giallu left	22:58
Ilari	telemachus: 'git reflog'.	22:58
telemachus	Luckily, gvim was still open with both changed files, so I resaved and recommitted	22:58
	Ilari: ?	22:59
xymbol	telemachus: be cautious with rebase, it's a great tool but needs to be handled with care	22:59
	:-)	22:59
telemachus	xymbol: Yes, I think I just learned that.	22:59
	← ajonat left	22:59
Ilari	Well, rebase doesn't start with dirty tree, and rebase gone awry can be undone by looking up the commit from reflog and resetting.	22:59
telemachus	So, let's try this from the top. I create a short-lived branch (fix_schedule). I work on it. I make brilliant fixes. I commit.	22:59
	→ DaveIngram joined	23:00
xymbol	telemachus: google for "rebase versus merge", you'll find several articles on the topic	23:00
telemachus	At that point, I can either (a) merge the branch back into master or (b) rebase.	23:00
	Let's say I want to rebase, just for fun.	23:00
	I want to be in master when I do that, yes?	23:00
Ilari	telemachus: 'git checkout fix-schedule', 'git rebase master', 'git checkout master', 'git merge fix-schedule'.	23:00
	telemachus: rebase and merge alter current branch.	23:01
telemachus	Ilari: that's exactly what I did with no joy.	23:01
	Hmm.	23:01
Mikachu	(the first two steps can be git rebase master fix-schedule)	23:01
wildwobby	All fixed. Thanks so much I was SOOOO worried I had lost all of that code.	23:01
telemachus	Mikachu: ok, but where should I be when I run git rebase master fix_schedule?	23:01
	← maattd left	23:01
Mikachu	doesn't matter	23:01
	well, you should have a clean index and worktree	23:02
	→ Adlai` joined	23:02
	← mybad left	23:02
	← indirect left	23:02
	indirect_ → indirect	23:02
	← Adlai left	23:02
	← tedoc2000 left	23:02
	Adlai` → Adlai	23:02
telemachus	Mikachu: both branches have untracked files, but nothing in the index	23:02
Mikachu	telemachus: also, look into the reflog if you're going to mess up rebase a lot	23:03
telemachus	(the untracked files are simply things I can't decide about)	23:03
	→ tedoc2000 joined	23:03
Mikachu	telemachus: a branch can't have untracked files	23:03
xymbol	telemachus: untracked files are "nowhere" in Git	23:03
telemachus	Mikachu: master had them, so when I created the branch it got them too	23:03
Mikachu	telemachus: they just sit in your working tree, git does nothing with them at all	23:03
xymbol	they don't live in branches but your working directory	23:03
telemachus	Mikachu: right, that's cool	23:03
Ilari	telemachus: Untracked files are property of working directory, and that is shared between all branches.	23:04
Mikachu	until you accidentally overwrite them with git reset --hard or git checkout -f ;)	23:04
telemachus	I just meant that other than the two garbage items, all is clean - both branches	23:04
Ilari	... Except maybe sometimes complain about them being trashed in some operations...	23:04
	← psynaptic left	23:04
telemachus	Now I want to rebase - or merge	23:04
	Or explode	23:04
xymbol	telemachus: you may want to look at git-stash to have incomplete stuff somewhere else	23:04
telemachus	or something	23:04
	→ doctorcode_ joined	23:05
	→ anarcat joined	23:05
anarcat	hello!	23:05
	→ cenuij joined	23:05
Ilari	...	23:05
telemachus	Hmm. We got sidetracked. Never mind the untracked crud, please.	23:05
xymbol	telemachus: generally, you can live with merge which is less "dangerous";	23:05
anarcat	how do i "fold" multiple commits on branch A into a single commit on branch B?	23:05
	← qrush left	23:05
xymbol	OTOH rebase allows you to rewrite history which is a good thing sometimes	23:05
telemachus	xymbol: so that's, git checkout master; git merge branch; git -d branch, yes?	23:06
xymbol	exactly	23:06
Ilari	telemachus: That works to merge it...	23:06
	← ^Einstein left	23:07
Ilari	telemachus: Doesn't necressarily give linear history...	23:07
xymbol	telemachus: git checkout -b fix_issue; edit; edit; edit; git add changes.txt; git commit; git checkout master; git merge fix_issue; git branch -d fix_issue	23:07
telemachus	Ilari: right	23:07
	→ ^Einstein joined	23:07
telemachus	But rebase made my face hurt, so for tonight, it's merge for me.	23:07
xymbol	telemachus: as Ilari pointed out, (non) linear history eventually becomes a problem; esp. if you're collaborating	23:07
	← KyleMac left	23:08
Cheaterman	No one uses cvs over git. Right?	23:08
	KyleMac2 → KyleMac	23:08
telemachus	This is what I get for reading blogs about why rebase is so cool.	23:08
	;)	23:08
Mikachu	Cheaterman: i think dhcpcd moved from git to svn	23:08
	← psoo left	23:08
xymbol	Cheaterman: sorry, jumped one SCM at a time ;-)	23:08
Cheaterman	Mikachu: Strange idea. Why am I concerned?	23:08
Mikachu	Cheaterman: i must have misunderstood your question	23:08
Cheaterman	xymbol: No problem	23:08
	milli` → milli	23:09
Cheaterman	Mikachu: Well, my question is the 4 lines § a few pages above	23:09
xymbol	telemachus: take your time and look at git rebase -i (interactive); it may make more sense	23:09
telemachus	xymbol: cool. I will play with that next time. Thanks.	23:09
xymbol	gotta go, bye!	23:09
Ilari	Cheaterman: It cuts off in beginning of cvsimport command.	23:09
	← Yuuhi left	23:09
Cheaterman	"git cvsimport -v -d:pserver:anonymous:@qjackctl.cvs.sourceforge.net:/cvsroot/qjackctl qjackctl"	23:09
	Never mind.	23:10
	← ^Einstein left	23:10
	→ ^Einstein joined	23:11
	← kvanderw left	23:12
impulze	hm is gitweb broken for links2?	23:12
	all i get is "Generating..." when trying to view something	23:12
Ilari	impulze: Try reloading the page.	23:12
impulze	weird	23:13
	that worked	23:13
Ilari	The generating... page tries to auto-refresh, but auto-refresh can fail.	23:14
impulze	Ilari: i have to do this everytime i want to view an object? what's happening there?	23:14
	← [t0rc] left	23:14
	← tbuehlmann left	23:15
impulze	can? it's reproducable for every view	23:15
Ilari	impulze: The method it uses for auto-refresh fails for Links2.	23:15
	← carrus85 left	23:15
impulze	yes	23:16
	← nh2 left	23:16
impulze	i've found a menu option called "auto-refresh"	23:16
	it seems to do something that makes it more happy :P	23:16
	← ^Einstein left	23:16
impulze	thanks though	23:16
Ilari	<meta http-equiv="refresh" content="0"/>	23:16
	← fualo_ left	23:17
	→ fualo_ joined	23:17
	← plediii left	23:17
	→ marvil07 joined	23:18
	← __iron left	23:19
	→ __iron joined	23:20
	← doctorcode left	23:20
	doctorcode_ → doctorcode	23:20
	← aspotashev left	23:22
	← up_the_irons left	23:23
	→ up_the_irons joined	23:23
	→ adam[pac] joined	23:25
	→ cenuij- joined	23:27
adam[pac]	hi, i just discovered git mergetool. I've tried using diffuse to resolve some conflicts. After exiting the tool, git status shows me an untracked file with a .orig extension. The merged file seems to have my resolutions but is not showing as modified. How can I prevent the .orig file from showing up?	23:27
	← telemachus left	23:27
	→ Guest333 joined	23:28
	← godsflaw left	23:28
adam[pac]	And is there a way to automatically commit the merge and remove .git/MERGE_HEAD after a succesful merge resolution? It seems like I had to run "git commit" even though git status didn't detect changes.	23:28
	← jmcantrell left	23:28
	→ WALoeIII_ joined	23:30
	→ plediii joined	23:31
	→ agile joined	23:33
	← PointMan left	23:34
	← dreiss left	23:36
	→ godsflaw joined	23:36
	← pixelbeat left	23:37
	← \|Inthewings\| left	23:38
	← cenuij left	23:39
	mithro_ → mithro	23:39
	→ ^Einstein joined	23:39
	← boto left	23:45
	← Z3R0 left	23:47
	← doctorcode left	23:47
	→ ablemike joined	23:47
	← WALoeIII left	23:47
	WALoeIII_ → WALoeIII	23:47
	→ godsflaw_ joined	23:50
	→ dreiss joined	23:54
	← godsflaw left	23:58
	→ ab3 joined	23:59

Logs Search ←Prev date Next date→

Channels Documentation