IRCloggy #git 2010-04-21

Provider of IRC logs since 2005.

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

	2010-04-21
	→ jmcantrell joined	00:02
	← agnathan left	00:02
	→ saccade joined	00:06
	← avuton left	00:06
	← meatmanek left	00:06
	← wriver left	00:07
	→ Remixman joined	00:10
	→ gsan joined	00:11
	← kenneth_reitz left	00:11
	→ pheaver joined	00:11
	→ juan_arandaalvar joined	00:12
	← justin-george left	00:12
	← dreiss left	00:12
	← Remixman left	00:14
	← altogether left	00:14
	→ Quevin joined	00:14
	← Quevin left	00:14
	← mjf left	00:15
	← webchick left	00:15
	← tildeequals left	00:18
	← frakturfreak left	00:20
	→ Remixman joined	00:21
	→ meatmanek joined	00:22
	→ jsilver joined	00:22
	← AAABeef left	00:24
	← timofonic left	00:25
	→ frakturfreak joined	00:26
jsilver	i just merged a branch with master branch, why doesn't it get pushed after i commit and push?	00:26
	← meatmanek left	00:26
	← Remixman left	00:28
wereHamster	jsilver: man git-config -> push.defaults	00:28
Gitbot	jsilver: the git-config manpage can be found at http://git.or.cz/man/git-config	00:28
wereHamster	push.default	00:29
	← a2ron left	00:29
	← loincloth left	00:29
	→ a2ron joined	00:31
	→ Fullmoon joined	00:31
	→ stringo0 joined	00:31
	← frakturfreak left	00:32
	→ magn3ts joined	00:33
	→ scarabx joined	00:35
	→ enherit_ joined	00:38
	→ altogether joined	00:40
	← enherit left	00:41
	enherit_ → enherit	00:41
	→ webchick joined	00:41
	← altogether left	00:42
	← mrlarner left	00:43
jsilver	wereHamster, thanks	00:44
	i was getting confused about how merging works when pushing a merged branch	00:45
	the commit didnt how up	00:45
	the merge commit	00:45
	→ timvisher joined	00:45
jsilver	apparently git just told the remote "merge the branches"	00:45
	or something	00:45
	← webchick left	00:46
	→ Rolenun joined	00:46
	← anair_84 left	00:46
	← lightcap left	00:47
wereHamster	fast-forward merge	00:48
	and merges happen in the local repo	00:48
	git doens't (can't) tell the remote repo to perform a merge	00:48
jsilver	hmm	00:50
	what's the proper procedure?	00:50
	i checked out master, merged, that worked, then commited and pushed	00:51
	the push worked but 0 deltas	00:51
	← dharrigan left	00:51
jsilver	so i got confused	00:51
wereHamster	was the branch you merged into master already part of the remote repository?	00:51
jsilver	yes	00:51
	we had been pushing it there	00:51
	to sync	00:51
	← MrPunkin left	00:52
wereHamster	then it was a fast-forward merge, in which case there are no new objects to push, git only updates the ref	00:52
jsilver	ahh	00:52
	thought so	00:52
	thanks !	00:52
	→ ilteris- joined	00:53
	← ilteris_ left	00:55
	← christophsturm left	00:55
	→ christophsturm joined	00:56
	← Bass10 left	00:56
	→ mithro joined	00:59
	← christophsturm left	01:00
	← Alagar left	01:03
	→ pklingem joined	01:05
	← Mooloo left	01:06
	← psynaptic left	01:07
	→ jafl joined	01:07
	← Ryback_ left	01:07
	→ dharrigan joined	01:10
	← magn3ts left	01:10
	← dharrigan left	01:10
	→ tildeequals joined	01:10
	→ Vampire0 joined	01:11
	← Vampire0_ left	01:13
	← pklingem left	01:14
	← enherit left	01:15
	→ hkweb joined	01:16
	→ timofonic joined	01:16
	→ royalty joined	01:17
	← snuxoll left	01:21
	→ VVSiz_ joined	01:24
	→ magn3ts joined	01:24
	→ jrmuizel joined	01:24
	← tedoc2000 left	01:24
timvisher	any emacs users around?	01:26
	I've got a question about replacing with \,() syntax	01:26
	#emacs is dead right now	01:26
	← VVSiz left	01:27
	← Gulug left	01:28
avar	C-h r m REGEX RET ?	01:29
	→ Gulug joined	01:29
	→ webchick joined	01:29
timvisher	Specifically about regex replacement in a fontified buffer.	01:29
	I just figured it out.	01:29
	M-x facemenu-remove-all	01:30
	took an hour to find that though…	01:30
	:(	01:30
	→ zaius joined	01:31
	← webchick left	01:31
	→ iruediger joined	01:31
	→ altogether joined	01:32
zaius	Hey all. I have a bunch of binaries that I no longer need day to day, but I would like to keep in the repository. Is it possible to keep them in a branch, and not have them taking up space in master?	01:32
timvisher	and now mysteriously it's not working again.	01:33
	timvisher pulls some hair out absent mindedly…	01:33
	→ AAA_awright_ joined	01:33
timvisher	I need to be able to remove it from \1 in the replacement string	01:34
	but I have no idea how to do that.	01:34
	← parasti left	01:35
	← AAA_awright left	01:35
	← warthog9 left	01:36
nDuff	zaius, if you just want to save the space in the working tree, absolutely -- but every cloned repo will still include those objects, so I don't know if it'll actually save as much space as you're hoping.	01:36
	← altogether left	01:37
nDuff	zaius, ...if these really are static binary files, as opposed to sources, you might consider incorporating them by reference -- having a makefile which downloads them when appropriate and contains only a checksum to compare the downloaded blob against. That doesn't help you much with the size of your repos, though, since they're already there in your history taking up space.	01:37
	← joevano left	01:38
	→ lightcap joined	01:40
	→ joevano joined	01:40
	← lightcap left	01:40
	← WALoeIII left	01:40
	xer0x → xer0x\|away	01:43
	→ dmg joined	01:45
	← timofonic left	01:46
zaius	yeah.. there's actually no need for the binaries any more, but i want to be able to go back to them if i need	01:47
	← kukks left	01:47
zaius	clone pulls the whole repo, right? including all branches?	01:48
	← deadguys left	01:50
	→ deadguys joined	01:50
	← juan_arandaalvar left	01:54
	← jrmuizel left	01:55
	→ stantont joined	01:55
	→ Chani joined	01:56
	→ gdey joined	01:56
Chani	so... I spent a while writing code in a new folder in one gitsvn repo. now I've realised I actually need the whole folder moved to a different location.. which happens to be under a different gitsvn repo. how much trouble would it be to preserve history?	01:58
	I'll probably squash it before I dcommit anyways, so I'm not overly attached to that history, but it would be nice to have	01:58
nevyn	Chani: add it as a remote and cherry pick the bits?	01:59
	Chani: how many commits are we talking about?	01:59
Chani	nevyn: I can add a completely different codebase as a remote?	01:59
nevyn	as long as the commits apply cleanly to both trees you should be able to transport individual commits I think....	02:00
	→ webchick joined	02:00
Chani	nevyn: 11 commits, and iirc only one touches a file outside that folder	02:00
nevyn	is the path within both projects to the directory the same?	02:00
	if so you're golden	02:00
Chani	er.. the same? it's a brand new folder	02:01
nevyn	do it in a branch for justin	02:01
Chani	although \where I created it, it's a folder in the gitsvn root. where I need to move it to is acutally a few levels down	02:01
nevyn	right but if you have projectagitroot/libs/directoryofinterest	02:01
	so cherrypick the changes first then git mv the directory and commit	02:02
	if it's new it's even less of an issue	02:02
Chani	hmm, yeah, prolly easier	02:02
	although... I should probably double-check that this actually is necessary before I do it	02:04
	← dys left	02:04
	→ camonz joined	02:04
	→ dys joined	02:05
	→ treak007 joined	02:05
	← hkweb left	02:05
	← magn3ts left	02:07
	← jafl left	02:08
	→ timofonic joined	02:08
	→ Sonic joined	02:09
	← Gulug left	02:11
	← sh1mmer left	02:12
	← jmcantrell left	02:12
	→ jmcantrell joined	02:13
	→ sh1mmer joined	02:20
	← webchick left	02:20
	← dmg left	02:20
	→ snuxoll joined	02:21
	→ altogether joined	02:22
	→ sh1m joined	02:23
	← cytrinox left	02:23
	← sh1mmer left	02:24
	sh1m → sh1mmer	02:24
	→ cytrinox joined	02:25
	→ Remixman joined	02:26
	← gdey left	02:28
	→ dmg joined	02:28
	← spearce left	02:30
	← darwin_ left	02:31
	→ darwin_ joined	02:31
	← novas0x2a left	02:36
	→ mefesto joined	02:37
	→ magn3ts joined	02:41
	→ tstclair joined	02:43
	← sagsousuke left	02:47
	→ warthog9 joined	02:50
	→ novas0x2a joined	02:51
	← eek812 left	02:52
	→ timj__ joined	02:52
	← Aaaarg left	02:53
	→ Aaaarg joined	02:56
	← Meow`` left	02:56
	← timj_ left	02:56
	← flaguy48 left	02:59
	→ dys` joined	03:00
	→ Meow`` joined	03:01
	← dys left	03:02
	→ joeconyers joined	03:03
	→ npouillard joined	03:04
	← Guthur left	03:06
	← sh1mmer left	03:08
eli	In a project that moves from svn to git, I'm now at the anticipated stage of people who are surprised by the staging area, and who look for a way to avoid it.	03:14
	Is there a quick way to make `git commit' with no paths behave as if it got "."?	03:15
	← Meow`` left	03:15
	← tstclair left	03:15
	nDuff shudders -- he has coworkers who do that, and they have no end of cases where they check in more changes than they want to by mistake.	03:16
eli	nDuff: How do they do it?	03:16
avar	eli: Write a shellscript? Or make the silly buggers learn git?	03:16
eli	avar: I'm leaning towards a shell script --	03:17
nDuff	eli, shrug -- they might use a shell function, or somesuch; different department, I only get called in to clean up their messes.	03:17
eli	what I want to avoid is them getting used to adding `-a' which is the wrong thing that leads to such problems.	03:17
	(And to make things worse, I've seen lots of svn->git pages mention `-a' as the way to go.)	03:18
	→ petdance joined	03:18
nDuff	'git add .' is the wrong way to go too; unless you're careful in your .gitignore you end up checking vim swapfiles and such cruft.	03:18
	→ Meow`` joined	03:18
eli	nDuff: I'm not talking about that,	03:20
	→ Belna_ joined	03:20
eli	What I think should work nicely is a quick script that checks if any of the arguments looks like a path -- if there is one, run `git commit' as is, if there isn't, add a `.' argument.	03:21
	(And do so at the beginning, so `git ci -m' doesn't end up being confusing.)	03:21
	← tanoku left	03:22
nDuff	probably should also check whether there's already anything staged	03:22
	and only change behavior if there isn't	03:22
eli	Good point.	03:23
	Actually, it looks like what I descibed works fine -- adding "." makes all changes that were `git add'ed in this directory as well as changes that were not be committed; and changes elsewhere that were added are not included.	03:31
	So this looks close enough to the svn thing.	03:31
godlygeek	eli: make them learn git, definitely. but, also - teach them about git commit -a	03:31
	→ Khisanth joined	03:31
nDuff	eli, well, my point was that it's probably a good idea to let folks who want to do things the Right Way do so, ie. make sure your script doesn't obstruct the preferred workflow.	03:32
eli	godlygeek: See above -- I think that getting them into the habit of using `-a' is bad -- it will make them commit stuff they didn't intend to.	03:32
godlygeek	adding an implicit . is worse...	03:32
nDuff	eli, ...if you're not shadowing 'git commit' but rather providing a separate command, though, that's fine.	03:32
godlygeek	at least with -a they can only commit changes to files that have already been added.	03:32
	→ jrmorrisnc joined	03:33
eli	nDuff: Yes, the point is of course to provide a migration path -- there's no objection to learning how to use git, but I'm not going to blame anyone for not becoming a fluent git user overnight.	03:33
nDuff	godlygeek, ...actually, reading the man page, it looks like that's the case with 'git commit' as well, so it's not as bad as an implicit 'git add .'	03:33
eli	godlygeek: not implicit -- check if there's no paths specified as an argument, and if none are -- include a ".".	03:34
nDuff	eli, nod; that's actually one of my objections to git -- it's not all that user-friendly to those who want to use the concepts they already know, as opposed to Bazaar and such.	03:34
godlygeek	what advantage does that have over using git commit -a ? i'm still not seeing one...	03:34
nDuff	godlygeek, git commit -a includes things not under the current directory	03:34
	→ cloudhead joined	03:34
nDuff	godlygeek, ...if their workflow is such that they're accustomed to committing subtrees...	03:35
eli	nDuff: Yeah -- one of the thing I practically demanded is that they install 1.7.x; I get the impression that things are way better now.	03:35
	→ Theaxiom joined	03:35
godlygeek	ok, but - "git commit -a" will commit all files that are tracked, "git add ." will start tracking new files...	03:35
Theaxiom	question, let's say I have a folder inside my repo that I want to be it's own repo, how would I accomplish that? I don't want it to be a submodule, because I would want it to essentially be the main repo location. Does that make sense?	03:35
eli	godlygeek: What nDuff said -- being surprised by committing stuff out of the current directory can lead to more shouting.	03:35
nDuff	godlygeek, he's not doing an implicit git add ., but rather a git commit .	03:35
godlygeek	i definitely think the risk of starting to track new files accidentally is a bad idea...	03:36
eli	godlygeek: It's not tracking new files -- I checked that.	03:36
nDuff	godlygeek, ...no risk of tracking new files with "git commit ."	03:36
eli	Here's the script, FWIW:	03:36
	#!/bin/sh	03:36
	seen_path=no; for p; do if [ -e "$p" ]; then seen_path=yes; fi; done	03:36
	if [ $seen_path = yes ]; then git commit "$@"; else git commit . "$@"; fi	03:36
godlygeek	oh, indeed you're right - i thought "git commit ." would add new files as well, my bad.	03:37
eli	In any case, this looks to me much better than the often recommended `ci = commit -a',	03:38
	not only in being more svn-line, but in not making people stick to it and leaving the door open for them to get used to git.	03:38
	← Fusco left	03:39
	← dmg left	03:39
eli	Theaxiom: perhaps this works: http://progit.org/book/ch6-7.html	03:40
Theaxiom	eli, thanks but I think I found what I needed @ http://help.github.com/splitting-a-subpath-to-a-new-repo/	03:40
eli	Theaxiom: Ah, I thought you wanted to go the other way.	03:42
	(Which is something I was looking into earlier today.)	03:42
Theaxiom	eli, ahh pulling another repo into your source tree?	03:42
eli	Yes	03:42
Theaxiom	eli, that will be good to know later on down the road, thanks :)	03:42
	eli, I actually think I can make use of that right now for a different folder, heh	03:43
	→ lresende joined	03:43
Theaxiom	eli, thanks	03:43
	← lresende left	03:43
	← Remixman left	03:45
	→ jtauber joined	03:48
	← aresnick left	03:50
	→ Remixman joined	03:50
	← s\|k left	03:50
	→ sh1mmer joined	03:58
	→ dmg joined	03:58
	← mythos left	04:00
	→ juan_arandaalvar joined	04:00
jrmorrisnc	so my boss wants me to present why git is better than CVS to my team.	04:00
	← iruediger left	04:01
	← juan_arandaalvar left	04:01
	← Remixman left	04:01
jrmorrisnc	I never used CVS so I'm not sure where to begin	04:01
	→ juan_arandaalvar joined	04:01
	← manyoso left	04:02
jrmorrisnc	anyone got a good link comparing cvs and git?	04:02
Rolenun	...	04:03
	http://www.youtube.com/watch?v=4XpnKHJAok8 =D	04:04
coppro	git does everything cvs does except it's better. It also does other stuff	04:04
jrmorrisnc	ok, I've watched the talk, besides merges and efficiency of moving things around, what other aspects could I play up, like distributed workflows, etc that ought to turn heads. Really want to make a good impression	04:05
	cherrypicking maybe?	04:06
	← ilteris- left	04:06
	→ Remixman joined	04:07
	→ ilteris- joined	04:07
Rolenun	http://progit.org/book/	04:07
jrmorrisnc	ty	04:07
Rolenun	np :)	04:08
nDuff	jrmorrisnc, the big thing I'd focus on is data integrity.	04:08
	jrmorrisnc, I maintained cscvs for a long time, a tool for parsing through CSV history and building patchsets out of it. CVS is very much prone to data corruption; if part of a ,v file gets truncated off, you don't notice until you try to check out part of the history from before the truncation happened.	04:09
jrmorrisnc	data integrity :D	04:09
	tyvm	04:09
	← jonshea left	04:09
jrmorrisnc	i use git and I never used anything else but SVN for a little while so I would have just blown past that	04:09
	→ jonshea joined	04:10
jtauber	offline usage has always been a huge one for me	04:10
Rolenun	I think my favorite part of git is that I can run a server from a usb drive just as easily as from a hard drive.	04:10
jtauber	and no single point of failure has saved my team a couple of times	04:10
Rolenun	exactly jtauber :D	04:10
nDuff	jrmorrisnc, ...also, CVS throws away information on which changes got committed together, as it doesn't have a concept of changesets. It's relatively easy to move your data from any changeset-oriented revision control system to a different one, but coming from something where the only storage layer is individual files, there's guesswork involved in the conversion. When your history is in the balance, you don't want guesswork.	04:10
	→ silence joined	04:11
	← kingfishr left	04:11
	silence → Guest2601	04:11
	→ kingfishr joined	04:11
jtauber	it's funny because I see SVN as so much better than CVS and git as so much better than SVN that the thought of comparing CVS and git makes my brain explode :-)	04:11
	← Remixman left	04:12
	xer0x\|away → xer0x	04:12
	Guest2601 → ajpiano	04:12
Rolenun	yeah, everytime I hear the comparison I tend to refer ppl to linus' google talk. ie "Git takes the approach: What would CVS never ever do..." :D	04:12
mugwump	SVN is better than CVS?	04:12
nDuff	mugwump, absolutely.	04:12
Rolenun	yes	04:12
mugwump	slightly better interoperability perhaps	04:12
	slightly less fragile	04:13
nDuff	mugwump, a better data model	04:13
mugwump	no branches	04:13
	no tags	04:13
nDuff	certainly there's branching and tag support, if the tool is used as intended	04:13
mugwump	they're not first class features	04:13
	it's not user servicable like CVS	04:13
Rolenun	but another of git's great abilities is to be able to merge in under 2 seconds. try doing that with cvs or svn :D	04:13
nDuff	mugwump, and diffs aren't first-class objects in git, but diff-related commands work just fine.	04:14
jtauber	svn just treats branching and tagging as lightweight copying, it's quite an elegant solution compared with cvs	04:14
nDuff	mugwump, ...I care what actually works in practice, and how well, with users following the preferred workflow more than I care about the underlying design.	04:14
mugwump	well, it treats lightweight copying as lightweight copying, and nothing else exists	04:14
	← Yuffster left	04:14
coppro	not true	04:15
	they added mergeinfo	04:15
mugwump	by the time mergeinfo was core svn was already legacy	04:15
	→ mythos joined	04:15
	→ Remixman joined	04:17
Rolenun	hehe, I think svn was legacy when git 1.0 was released =D	04:18
mugwump	It certainly was by June 2008	04:19
coppro	git 1.0 was legacy by the time it was released	04:19
mugwump	(1.5.0 release date)	04:19
	← tgunr left	04:20
	← ilteris- left	04:20
	← jfkw left	04:20
	→ chaiz joined	04:21
	→ ilteris_ joined	04:22
eli	nDuff: I'll +9 that -- we moved around 2004 from cvs to svn, and the horror that cvs was made me decide that it's better to dump that history rather than try to port it. A main factor in that decision is exactly that kind of corruption that was impossible to recover from.	04:22
mugwump	I've seen several svn repositories with unrecoverable revisions too	04:23
nDuff	mugwump, yes, but is that before or after they dumped bdb?	04:24
	← mefesto left	04:24
mugwump	pass, it was just some real svn repos	04:24
nDuff	mugwump, ...and at least svn has a fsck command built-in, so you can make that part of your nightly pre-backup cron job.	04:24
	← jmcantrell left	04:25
nDuff	mugwump, ...so you're still vastly better off than the cvs case where you only discover you lost history years later.	04:25
eli	mugwump: We used fsfs almost exclusively, and there was not even one corruption.	04:26
mugwump	but that doesn't really happen	04:26
	→ jony joined	04:27
nDuff	mugwump, which value of "that"?	04:27
	xer0x → xer0x\|away	04:27
eli	OTOH, the cvs corruption that nDuff talks about most certainly happened. In multiple places.	04:27
mugwump	I mean, we had a CVS repository here which was used for much longer than svn has been around and no corruption when we converted to git	04:27
	but you know, anecodotes and all that	04:27
nDuff	mugwump, as I said, I used to be cscvs maintainer. At least half my "customers" had corruption.	04:28
	mugwump, ...and given that role, my sample size was much larger than 1.,	04:28
	← juan_arandaalvar left	04:28
nDuff	s/,$//	04:28
	← segher left	04:28
	← Remixman left	04:28
Rolenun	yeah nDuff, I saw similiar corruption among multiple clients and farms	04:29
	← Semikolon left	04:29
mugwump	well, in hindsight it seems like snake oil which made everyone change systems for only a marginal gain	04:29
nDuff	mugwump, ...I strongly disagree, but don't have time to argue it further.	04:30
	→ jakubzalas joined	04:30
Rolenun	+1	04:30
	→ munichlinux joined	04:30
mugwump	all I'm saying is that the "branching is just copying" idea, which they stole from Perforce (badly), was a backwards step, and depending on what you rate as important, makes it worse for some people	04:33
	→ Remixman joined	04:33
	← Belna_ left	04:33
	← jony left	04:33
offby1	mugwump: I thought it was elegant when I first encountered it	04:34
mugwump	sure, I swallowed it initially too	04:34
nDuff	mugwump, ...if you buy git's proposition that it's perfectly reasonable to work backwards to deltas from snapshots, I see no reason at all why svn's concept isn't workable.	04:34
	→ jony joined	04:34
nDuff	mugwump, ...yes, you need history-tracking metadata. Even if you started using git before they added that upstream, there were 3rd-party tools for doing that, and it came with a quite elegant way for storing both revision and subtree metadata.	04:35
mugwump	you typed "git" where you meant "svn"	04:35
	your fingers have learned	04:35
nDuff	err, right	04:35
	:)	04:35
	(it is late, and I am tired)	04:36
mugwump	10 years before svn existed, the tool they stole that from had such tracking	04:36
nDuff	...actually, I should probably go home while I can still pretend to almost be able to drive safely.	04:36
	→ maleknet joined	04:36
mugwump	but somebody on the internet is WRONG!	04:36
	;-)	04:36
nDuff	mugwump, yes, I know. I'm also PO'd that nobody but darcs and monotone do merging as well as Tom Lord's GNU Arch did, even years later. Sometimes state-of-the-art moves backwards in some respect or another. Too bad, it happens.	04:37
	← seanmccann left	04:37
mugwump	Hmm. I'm not sure what arch does that git rebase doesn't, but ok	04:37
	→ seanmccann joined	04:37
nDuff	mugwump, corner cases involving cherry picking and remerging	04:38
	you could trivially mark a changeset as not to be merged into a given branch, and that designation stuck	04:38
	← scarabx left	04:38
	→ meatmanek joined	04:39
	← cloudhead left	04:39
	← meatmanek left	04:39
	← magn3ts left	04:40
	← Remixman left	04:44
	→ KDII joined	04:44
	← KDII left	04:44
	→ jjuran joined	04:45
	← madewokherd left	04:45
	→ codetroll joined	04:49
	→ Remixman joined	04:50
	← gnufied left	04:55
	← Textmode left	04:55
	→ fool_ joined	04:56
	← fool_ left	04:56
	→ fool_ joined	04:56
	→ magn3ts joined	04:56
	→ roop joined	04:58
	← fool__ left	04:59
	← Remixman left	05:01
	→ jceb joined	05:01
	← Twisted_Mentat left	05:03
keyvan	anyone here to help me get gitbot working?	05:03
	looks pretty straightforward.. but i must be screwing something up	05:04
	eh ill just read the source.	05:04
	← kingfishr left	05:04
	→ kingfishr joined	05:05
	← jakubzalas left	05:05
	→ dreiss joined	05:07
	→ Remixman joined	05:08
	→ bdimcheff joined	05:09
	← ScottO_ left	05:11
	← FauxFaux left	05:12
	← royalty left	05:13
	→ FauxFaux joined	05:14
	← roop left	05:14
	→ roop joined	05:15
	→ sde_antrix1 joined	05:16
	← eletuchy left	05:16
	← wolog left	05:16
	← Meow`` left	05:16
	← friskd left	05:16
	→ gnufied joined	05:17
	← Remixman left	05:17
	← dominikh left	05:19
	→ codeshepherd joined	05:22
	← bdimcheff left	05:23
	← codeshepherd left	05:24
	→ Remixman joined	05:24
	← AAA_awright_ left	05:26
	→ AAA_awright joined	05:26
	→ seanmccann_ joined	05:27
	← magn3ts left	05:28
	→ jesterKing joined	05:29
	← seanmccann left	05:30
	← Remixman left	05:31
	→ thiago_home joined	05:32
	→ psankar joined	05:39
	→ ilogger2 joined	05:44
	→ Arelius joined	05:45
	→ tschundeee joined	05:45
	→ SunilThaha joined	05:49
	→ khaase joined	05:50
	→ Guest24555 joined	05:57
	→ Weasel[DK] joined	05:59
	→ lucasvo joined	06:02
	→ munichlinux joined	06:02
	→ tetha joined	06:06
	→ priidu joined	06:07
	→ hyperair joined	06:08
	→ jony joined	06:12
	→ Belna joined	06:15
	→ drizzd joined	06:16
	→ friskd joined	06:16
	→ roop_ joined	06:16
	→ cbreak_work joined	06:17
	→ ph^ joined	06:19
	← Arelius left	06:22
	→ Arelius joined	06:22
knittl	http://paste2.org/p/785036 opinions/testing? :)	06:23
	→ steffkes joined	06:24
	→ rseifert joined	06:25
	← tetha left	06:26
	→ WALoeIII joined	06:26
knittl	and i now see an hilight from yesterday, gotta fix that too	06:26
	← ph^ left	06:26
wereHamster	knittl: the if (s->branch) is useless, because of the assert right before that	06:27
	or the assert is wrong	06:28
	→ kenneth_reitz joined	06:29
knittl	wereHamster: yeah, i can probably remove the if	06:29
	→ wolog joined	06:29
	→ magn3ts joined	06:30
	→ y10 joined	06:31
	← y10 left	06:31
	→ eletuchy joined	06:33
	← munichlinux left	06:35
	→ ntoll joined	06:35
	→ carl- joined	06:38
	← Guest24555 left	06:39
knittl	but it's not hurting now	06:40
	and all my tests were successful	06:40
	i should probably look into the existing unit tests	06:40
	← drizzd left	06:40
	→ Alagar joined	06:41
Alagar	good morning all.	06:41
	any one can help me. what is the command get the latest code from server and get the specific version from server	06:42
	→ drizzd joined	06:42
	← friskd left	06:42
knittl	git fetch <remote> gets all code from the server	06:44
	git checkout/pull/merge to have it in your working tree	06:44
	→ Remixman joined	06:47
	← magn3ts left	06:47
	→ webchick joined	06:48
Alagar	knittl: thanks is any graphical mode git client available for working from windows machine?	06:49
	← priidu left	06:49
	→ nor3 joined	06:49
knittl	Alagar: tortoisegit/git gui/gitk	06:49
	→ chittoor joined	06:50
nor3	anyone with git-osx-installer experience? i expected 'git' to be in my path after running it, but it isn't	06:50
	→ vu3rdd joined	06:51
	← tschundeee left	06:51
	→ dmg joined	06:52
	→ tango_ joined	06:52
	→ skoop joined	06:56
staaleu	How can I run mergetool on unresolved conflicts in a single directory instead off all unresolved files at once?	06:58
	→ giallu joined	06:59
	→ LeLutin joined	06:59
	→ galderz joined	07:00
	← Remixman left	07:02
	→ d0ugal joined	07:03
	← nor3 left	07:05
	→ j416 joined	07:08
	← j416 left	07:08
	→ akamaus joined	07:08
	→ timofonic joined	07:08
	→ j416 joined	07:08
	→ Remixman joined	07:08
Alagar	knittl: thanks. could you please help me. how to create a new git repository in server machine.	07:09
akamaus	hello, how can I clone the git repository to the remote host?	07:09
	→ trivol joined	07:09
	→ killerchicken_ joined	07:10
	← hyperair left	07:10
cbreak_work	akamaus: run it on the remote host	07:10
	→ priidu joined	07:10
cbreak_work	or make a local clone and scp it over	07:11
akamaus	cbreak, so there is no way to 'push' the repo?	07:11
	→ raichoo joined	07:11
cbreak_work	only if there's a repo	07:12
	git's an SCM, not a file transfer program	07:12
	→ magn3ts joined	07:13
akamaus	cbreak_work, got it. I thought about something like darcs put.	07:14
	scp is too slow for my case, will try git init and git push	07:14
cbreak_work	too slow? :)	07:16
	I am not sure if you'll be happy with git push then	07:16
crab	if you're pushing over ssh, then git push will be just as slow.	07:16
	← Remixman left	07:16
	← WALoeIII left	07:16
cbreak_work	pushing over SSH is the only solution for the paranoid	07:17
	← dmg left	07:17
	→ loxs joined	07:18
staaleu	How can I run mergetool on unresolved conflicts in a single directory instead off all unresolved files at once?	07:18
akamaus	cbreak_work, scp transfers a big number of small files very slowly	07:18
	→ LRN joined	07:19
cbreak_work	you can zip it before transfering	07:19
	if you have means to unzip it on the destination at least :)	07:19
	but pusing to a newly init --bare'd repo is also an option	07:20
	→ adaro joined	07:20
	← adaro left	07:20
akamaus	yeah, that works )	07:21
loxs	I'm reading about merge, and I can't find if there is something like "merge --pretend" so that I can see what would really happen if I issued merge. Generally could you recommend me some article on "howto code review using git"?	07:22
	→ Remixman joined	07:23
Bombe	loxs, just merge. Either the merge works (then you can undo it by reset HEAD^) or it doesnt (in which case no commit is created).	07:23
j416	loxs: if you are very paranoid about it, you can create a new branch and merge into that, then just scrap that branch if you don't want it	07:24
Bombe	loxs, the --no-commit option will perform the merge but will not create a commit. Its closest to pretending a merge.	07:24
j416	but, merge then reset is faster :)	07:24
loxs	hmm if I use --no-commit, where do the changes go?	07:25
	→ cilly joined	07:25
loxs	to the working tree?	07:25
j416	index	07:25
	and working tree	07:25
loxs	so I can merge and then use diff to see what was changed?	07:25
	that's nice	07:25
	thanks	07:25
j416	perhaps you need git diff --cached	07:26
	I have never used --no-commit	07:26
	but you'll find out.	07:26
loxs	yeah, thanks	07:26
j416	loxs: but as Bombe said, it's just as easy without --no-commit	07:27
	→ khaase_ joined	07:27
j416	git merge otherbranch;	07:27
	if that succeeds	07:27
	you can do: 'git reset --hard HEAD^' to get back to where you were before the merge	07:28
	(given that your work dir was clean of course)	07:28
	→ nevyn^ joined	07:28
loxs	I'll check both ways. But I guess I'd prefer --no-commit	07:28
	if I merge with --no-commit, will the log still know who wrote that piece of code that I'm merging?	07:29
j416	you won't see anything in your log, because you haven't committed anything	07:29
loxs	I mean, after I commit	07:29
j416	if you commit, you will see that history,yes.	07:29
	it should record its parents	07:30
	← khaase left	07:30
loxs	I mean, git merge --no-commit >> code review (and some changes as he is still quite newbie programmer) >> commit	07:30
	→ ghe joined	07:30
	→ dmg joined	07:31
	→ khelll joined	07:31
loxs	ok, I'll test it	07:31
cbreak_work	do NOT make changes in a merge commit	07:31
	→ dj_tjerk joined	07:31
cbreak_work	that are not part of the merge	07:31
	if you fix errors, do so in a normal commit	07:32
loxs	cbreak_work, why so?	07:33
	→ ereslibre joined	07:34
	← ereslibre left	07:34
	→ ereslibre joined	07:34
cbreak_work	because it's a merge commit	07:35
	→ geshan joined	07:35
cbreak_work	no one reads what happens in them	07:35
	one of an SCM's main purposes imho is to document development	07:35
	so it's good to make the development clear	07:36
	→ amcsi joined	07:36
	← Remixman left	07:36
cbreak_work	well, it's your repo :)	07:39
	there's no technical problem when you change unrelated things in a merge commit	07:39
	roop_ → roop	07:39
	← drizzd left	07:40
	→ frakturfreak joined	07:40
	→ drizzd joined	07:41
	→ hyperair joined	07:41
	→ warthog9 joined	07:42
	→ mastro joined	07:43
	→ gebi joined	07:43
	→ Remixman joined	07:43
	→ ixti joined	07:43
	→ NET\|\|abuse joined	07:44
	→ bauruine joined	07:46
	→ tbf joined	07:48
loxs	cbreak_work, I see what you mean. thanks	07:48
	← d0ugal left	07:49
	← ereslibre left	07:49
	← Remixman left	07:49
	→ d0ugal joined	07:50
	→ ph^ joined	07:50
sgronblo	How can I reset a file to its initial merge-conflicted state if I messed up with my manual merging?	07:51
wereHamster	checkout -m <file>	07:52
j416	'git checkout -m -- <file>' would be safer, no?	07:53
	→ airborn joined	07:53
j416	(question, not a suggestion)	07:53
sgronblo	thanks	07:53
	→ lucsky joined	07:54
	→ pantsman joined	07:55
	→ Remixman joined	07:56
	← wolog left	07:56
	→ kar8nga joined	07:57
	→ zoniq joined	07:58
	← Remixman left	08:01
	Rhonda hugs git add -p and <e>dit tightly. That's soooo convenient and great. :)	08:02
crab	git add -p and <e>dit gasp and choke as the air is squeezed from their lungs.	08:03
	→ shruggar joined	08:05
	→ simplechat joined	08:06
	→ batsf joined	08:06
	→ dajero joined	08:07
	← batsf left	08:07
	→ bdowning joined	08:07
	→ Remixman joined	08:08
	→ wolog joined	08:09
	→ Bennid joined	08:09
	→ kenyon joined	08:09
	→ ryanakca joined	08:10
	→ Paraselene__ joined	08:11
	→ roop_ joined	08:11
	← roop left	08:12
	← Arelius left	08:13
	← lucasvo left	08:13
	→ Arelius joined	08:13
	→ tatsuya joined	08:14
	khaase_ → khaase	08:18
	→ CompanionCube joined	08:18
knittl	Alagar: you don't need a server with git	08:19
	← carl- left	08:19
	← khaase left	08:20
	→ khaase joined	08:20
	→ gaveen joined	08:20
	← Remixman left	08:21
	← pantsman left	08:27
	→ ZGirl joined	08:27
ZGirl	What does git got that SVN don't?	08:27
divVerent	speed - space efficient checkout - branching and merging that actually works	08:28
wereHamster	proper merging support	08:28
	→ Remixman joined	08:28
khaase	ZGirl: https://git.wiki.kernel.org/index.php/GitSvnComparison	08:29
crab	righteousness.	08:29
divVerent	you know, svn the "OMG I store every file twice, and don't even compress the second copy" SCM :P	08:29
crab	;)	08:29
wereHamster	ZGirl: http://whygitisbetterthanx.com/#svn	08:29
ZGirl	I like git because it sounds western	08:29
shruggar	good merging, sane history rewriting, distributed use	08:29
divVerent	history rewriting is never sane ;)	08:29
	unless locally	08:30
ZGirl	http://www.howtopersuadeyouremployertoswitchfromsvntogitovernight.com/	08:30
shruggar	divVerent: you've just committed the nuclear launch codes to svn, what do you do? :)	08:30
	← dj_tjerk left	08:30
cbreak_work	ZGirl: git has branches and tags	08:30
ZGirl	Like a diseased tree	08:31
divVerent	shruggar: IIRC there are tools for undoing the last committed rev	08:31
cbreak_work	no	08:31
	not a tree	08:31
	a DAG	08:31
	git also supports merging, yo you don't get only trees	08:31
divVerent	shruggar: well, REALLY bad example	08:31
wereHamster	svn is like a tree: you can branch but not merge ;)	08:31
divVerent	in git you cannot eradicate it once someone has pulled	08:31
shruggar	SVN has branches and tags. Git has an interface to branches and tags which not only will /not/ make your fingers, eyes, and brain bleed, but also (bonus!) is simple enough that you'll actually use them	08:32
divVerent	even if you filter-branch to remove it - too late, next time he pushes the code is back oin	08:32
cbreak_work	divVerent: in git you can rewrite history as much as you want, you just have to do it everywhere where it is	08:32
	and unlike svn, git has lots of potential redundancy	08:32
divVerent	cbreak_work: yes, and this is why it is not possible	08:32
	there is no secure way to FORCE every user to perform that command :P	08:32
	in THIS SPECIFIC case, svn is actually much better	08:33
shruggar	in svn, once someone has updated, there is no way to get rid of it. That's not an svn vs git thing, that's a laws of reality thing	08:33
divVerent	dump the repo, edit the dump to kill that file, undump it	08:33
	shruggar: sure, but same thing with git :P	08:33
cbreak_work	it's called "I know where you live and where your kids go to school... so ... run this command" method.	08:33
divVerent	git's automatic superhappy merging is normally a good thing	08:33
	in this specific case it's a bad thing :P	08:33
wereHamster	ZGirl: see what you started... ;)	08:34
divVerent	I would BTW like to implement a system for automatically downloading such a filter-branch command from a server	08:34
	in a git frontend script	08:34
	but until now I found no way for that that isn't a gaping security hole	08:34
cbreak_work	make it involve jails	08:34
divVerent	thing is, in a decentral system you often do not KNOW who has your repo, or may commit	08:35
shruggar	divVerent, no. With svn it's "Someone could already have the change, but how do I get rid of it so that future users won't get it? Oh right, take the server offline and run a 36-hour import/export cycle" with git it's "Someone could already have the change, but how do I get rid of it so that future users don't get it? Oh right, run this one command. Gee, that was easy."	08:35
divVerent	in svn, it's quite controlled who can commit	08:35
	in git, often changes are pulled from elsewhere, and also may merge in the forbidden history	08:35
ZGirl	I just wish you could ignore directories and files easier in SVN	08:36
divVerent	shruggar: except this does not work for git :P	08:36
	→ carl- joined	08:36
divVerent	well, if you use gitolite, you can also add a rule to forbid pushing that file back in	08:36
ZGirl	svn propedit svn:ignore stupid_directory	08:36
divVerent	that at least is SOMETHING	08:36
cbreak_work	git has pattern based ignores	08:37
divVerent	just a git update-ref won't fix anything, as that's the first change the next "git push" will reupload	08:37
cbreak_work	(based on the file name and parts of it's path)	08:37
	also, git doesn't track directories	08:37
	not sure if subversion does	08:37
shruggar	wouldn't making a single commit on top of the "revised history" make the whole problem go away? No more successful push	08:38
	← bauruine left	08:39
cbreak_work	user can still rebase/merge	08:40
shruggar	svn users can still commit non-sanely, if the revision numberis the same, afaik	08:41
	→ flazz joined	08:43
shruggar	either way, I think "not having the server down for a day" is a good thing	08:43
divVerent	true, at least once I have a good automated syswtem for forcing the user to do the same filter command	08:44
	→ Madsy joined	08:45
	→ Pupeno joined	08:45
ZGirl	I am a true versioning freak	08:46
Madsy	What is the correct method of moving source from one source file into another, and then delete the file? Can git track such merges?	08:46
shruggar	Madsy, the correct method is to just do it, and git notices automatically	08:47
Madsy	Say if I have two files a and b. I move all the function implementations from a into b, and delete a.	08:47
	shruggar: Okay, thanks.	08:47
shruggar	at least, that's the theory	08:47
	in practice, it could use some work for the more-obscure cases, since the less-obscure ones are generally considered "good enough"	08:48
	but in git, the best option is usually "try it, see what happens", and "don't change your workflow to suit git. If git does the wrong thing, complain on the mailing list"	08:49
Madsy	That's nice to hear.	08:49
	I'm damaged from heavy SVN use, so..	08:50
	Trying to unlearn all the svn'isms.	08:50
shruggar	That said, I don't actually know what will happen if you try to merge changes to file a in, once you've made the change. I've never done that, but I /think/ that's the kind of thing that git was designed specifically to be able to handle. I'm really not sure, though.	08:51
	→ loincloth joined	08:52
	→ sylr joined	08:52
	→ JEEBcz joined	08:53
	→ saLOUt joined	08:54
saLOUt	how do i checkout exactly one file from a special commit in my current master branch?	08:55
shruggar	saLOUt, what do you mean by "checkout"?	08:55
saLOUt	create a copy from history to the working dir	08:55
j416	saLOUt: git checkout <commit> -- <file>	08:56
	iirc	08:56
	→ mtkd joined	08:56
	→ aziz joined	08:56
saLOUt	for what do i need the --	08:56
j416	everything after -- is treated as a path	08:56
	it's for not confusing it for a branch name etc.	08:56
shruggar	the -- separates revisions from files, in case the file you want to check out is called "master"	08:56
	→ bx2 joined	08:57
knittl	shruggar: also if you want to check out a file called "-f"	08:57
shruggar	of course	08:57
j416	saLOUt: if you're sure there is no ref with the same name as your file, you can omit --, but it's good practice to use it :)	08:58
	s/ref/ref or other command etc./	08:58
	/flag	08:58
	→ Khisanth joined	08:58
	→ Yuuhi joined	08:58
	→ psynaptic joined	09:04
	→ Brodde joined	09:04
	→ fai\|safe joined	09:05
	→ bluenovember joined	09:06
	→ project2501a joined	09:07
	→ snitko joined	09:13
	← kenneth_reitz left	09:15
	← snitko left	09:18
crab	what umask should users of an init --shared repository have? or doesn't it matter?	09:20
	← fai\|safe left	09:21
doener	crab: testing with "umask 0077; git init --bare --shared" gives e.g. "drwxrws--- 4 doener doener 4096 21. Apr 11:21 refs". So I'd say: doesn't matter	09:22
	→ fai\|safe joined	09:22
	← ZGirl left	09:22
crab	i meant when they were pushing, not for the init. but it seems it doesn't matter, indeed.	09:24
	← fai\|safe left	09:26
	← jony left	09:28
	← loxs left	09:29
	→ ramlev joined	09:31
	← ramlev left	09:31
	← Remixman left	09:32
	→ petrux joined	09:33
	→ jony joined	09:34
	→ jovianjake joined	09:35
hachi	hi all	09:38
	I'm trying to debug an ssh issue that only happens when I'm using git	09:38
	if I say git fetch, my ssh client starts asking me for a password, doesn't seem to let me access the remote machine	09:38
	but if I copy and paste the hostname from the git config file to the same shell invoking it as an argument for 'ssh', I'm able to log into the box, either with a proper password	09:39
	or if I have my key loaded into an agent, no password	09:39
	if I invoke git fetch with -v it doesn't show any output before invoking ssh	09:39
	← jovianjake left	09:40
doener	hachi: set GIT_TRACE=1 to see the ssh command it executes	09:40
	→ jovianjake joined	09:40
hachi	if I invoke the same ssh command it is running, no issue	09:41
	→ tg joined	09:41
	→ fai\|safe joined	09:42
hachi	trace: run_command: 'ssh' 'git.dev.company.com' 'git-upload-pack '\''git/rpmtools.git'\'''	09:43
	if I run that command manually, less the excessive quoting, it works fine	09:43
	← tg left	09:44
hachi	strace shows the same thing... running it within the context of git fetch either prompts me for a password, or the far end refuses my key	09:44
	running it standalone works fine	09:44
	anyone know how I can debug this further?	09:45
	→ mum-n-dad joined	09:45
	→ tg joined	09:46
	← j416 left	09:48
mum-n-dad	Hello	09:48
	I just cloned this repository a few days ago. Haven't touched it at all. Tried pulling the latest version just now, and got "error: Your local changes to 'app/controllers/blog_controller.rb' would be overwritten by merge. Aborting. Please, commit your changes or stash them before you can merge." - I get this message way too often. Is there any way to force a pull?	09:48
	→ j416 joined	09:48
divVerent	did you modify blog_controller.rb?	09:49
	or did, maybe, some build script?	09:49
mum-n-dad	No	09:49
	I haven't touched it at all	09:49
divVerent	if YOU modified that file, you should commit it	09:49
mum-n-dad	I haven't man!	09:49
divVerent	if you did not, and don't care - use git reset --hard	09:49
	and then complain to the owner of the repo as apparently they put some generated or modified-by-build file in the repository :P	09:50
	there actually is one other case when this might happen	09:50
	did you set autocrlf?	09:51
mum-n-dad	Nope	09:51
	Not that either	09:51
divVerent	hm... very odd	09:51
mum-n-dad	I'm using this script to pull - http://pastie.org/927501	09:51
	so I just add git reset --hard to that one yeah?	09:51
	I have a ton of repos I mirror on my harddrive you see	09:51
divVerent	that script is certainly fine	09:52
	but yes, reset --hard should fix it	09:52
mum-n-dad	Any shell programmers know how I could only apply git reset --hard to that script if there's an error?	09:52
	thanks a lot divVerent - I really appreciate it	09:52
divVerent	Morasique: you REALLY only want to do that if you never have any local changes in these repos	09:53
	i.e. if you are a "end user"	09:53
	← fai\|safe left	09:53
divVerent	but then:	09:53
	before the git pull, put "git reset --hard" and also "git clean -f"	09:53
	sorry, mum-n-dad :P	09:53
mum-n-dad	first reset then clean?	09:53
divVerent	order does not matter	09:54
j416	maybe also: git clean -dfx	09:54
divVerent	reset handles modified files, clean gets rid of additional files	09:54
	← webchick left	09:54
mum-n-dad	awesome!	09:54
divVerent	yes, possibly	09:54
mum-n-dad	j416: what does that do?	09:54
divVerent	-d gets rid of empty dirs, -x gets rid of .gitignore'd files	09:54
j416	mum-n-dad: man git-clean explains it pretty well! :)	09:55
Gitbot	mum-n-dad: the git-clean manpage can be found at http://git.or.cz/man/git-clean	09:55
mum-n-dad	super neat	09:55
divVerent	but both should usually not stand in the way of pull	09:55
mum-n-dad	:D	09:55
divVerent	but maybe of a build run done later	09:55
	→ curtana joined	09:55
curtana	what is the meaning of a remote's HEAD? I understand that a branch has a HEAD which is the commit that the branch points to ... but what is a repository's HEAD ?	09:56
divVerent	usually a symbolic ref pointing to master	09:56
j416	curtana: 'HEAD' is not 'head'	09:56
divVerent	IIRC it also is the branch the user checks out by default on a git clone	09:56
hachi	divVerent: correct	09:56
curtana	aah	09:57
divVerent	so if you happen to have the default branch with another name, you can edit that ref	09:57
j416	curtana: every branch has a head, but HEAD is the pointer to the currently checked-out commit	09:57
hachi	the HEAD on a remote becomes the HEAD name, which is a tracking branch pointing to the remote of the same name... during a clone	09:57
curtana	j416, oh that's odd. gitg shows that my remote repositories have a branch, with a name such as: origin/HEAD	09:57
divVerent	but normally, HEAD points to thwe checked out commit	09:57
	on a remote (i.e. bare repo), however, there is no such thing, and then it's just the default branch for clone	09:57
	← bx2 left	09:57
divVerent	curtana: sure. It's the "default branch" and usually identical to master :P	09:58
curtana	ok	09:58
	so changing origin/HEAD would change the default branch that would be checked out following a clone of the repository?	09:58
divVerent	yes	09:58
curtana	thanks	09:58
divVerent	it's one of the things you should never do unless you have a GOOD reason :P	09:58
curtana	:)	09:59
divVerent	a good reason may be switching HEAD to maint instead of master, so end users using git get something known stable :P	09:59
hachi	AUGH	10:00
mum-n-dad	thanks again divVerent	10:00
	and j416	10:00
	you're cool people	10:00
doener	curtana: not origin/HEAD, but HEAD in the repo origin refers to	10:00
hachi	so, git-upload-pack apparently throws an error code if asked to upload-pack a repo that is empty	10:01
divVerent	doener: git branch -r typically shows it as origin/HEAD -> origin/master :	10:01
	P	10:01
doener	divVerent: that's a local symref, and won't change if origin's HEAD changes	10:01
hachi	and git-fetch doesn't discern between auth errors or upload-pack returning 'nothing'	10:01
divVerent	sure	10:01
	it is created on clone, and matches the origin's HEAD then	10:01
doener	divVerent: it's just a convenience thing, so you can use "origin" as a shortcut for the "primary" remote tracking branch of that remote	10:02
divVerent	sure	10:02
	just like it also is a convenience thing so you already get a working tree after clone :P	10:02
doener	divVerent: point being: changing origin/HEAD won't affect origin's HEAD, and thus not affect new clones at all	10:02
divVerent	sure	10:03
cbreak_work	why would you not get a working tree?	10:03
	people actually commit non-working trees? :/	10:03
divVerent	you would get no working tree if the remote had no HEAD	10:04
	you'd then be forced to explicitly checkout a branch first	10:04
	→ tvw joined	10:04
mum-n-dad	Curious about something	10:04
	At http://pastie.org/927513 - how do I set it to re-run the "git pull" command if I get "fatal: The remote end hung up unexpectedly"?	10:04
cbreak_work	hmm. sounds like a minor thing.	10:04
mum-n-dad	* rerun	10:04
divVerent	mum-n-dad: not sure if that can be caught easily	10:04
	a simple way would be:	10:04
	git pull \|\| git pull \|\| git pull	10:05
mum-n-dad	heheh	10:05
divVerent	but that retries on any error	10:05
doener	divVerent: hm, IIRC HEAD is used to identify some directory as a git repo. So not having HEAD should mean that you just get an error instead of a clone	10:05
divVerent	doener: oh :P	10:05
mum-n-dad	Neat actually	10:05
divVerent	well... the case I actually had, was HEAD pointing into void	10:05
	like in freshly init'd repos	10:05
j416	mum-n-dad: if you're doing this every 2 seconds anyway, I wouldn't worry if git pull fails once :P	10:05
divVerent	first commit then actually creates what HEAD points to :P	10:06
doener	yeah, that's a possible case :-)	10:06
j416	oh, sorry, misread your script. well well. :)	10:06
mum-n-dad	j416 :)	10:06
j416	mum-n-dad: why are you sleeping there though?	10:07
	doesn't it work otherwise?	10:07
mum-n-dad	Actually I was trying to avoid the fatal hangups	10:08
	← akamaus left	10:08
j416	ah..	10:08
mum-n-dad	<ft> well, you need to cd back...	10:08
	<ft> or do stuff in a subshell, which is probably better.	10:08
	<ft> also: while ! somecommand; do :; done	10:08
	→ amcsi_ joined	10:08
mum-n-dad	did that make any sense? its from #zsh	10:09
	<ft> oh, you're working with absoluute path-names... so that's probably not needed.	10:09
j416	mum-n-dad: yeah, that'd work I guess	10:09
cbreak_work	you could outsource the script to india	10:09
j416	mum-n-dad: while [git pull returns fatal, and tries less than max limit] do [increment tries]; done	10:10
	s/tries/'tries'/	10:10
	maybe.	10:10
mum-n-dad	while ! git pull; do :; done	10:10
j416	with a sleep in that perhaps	10:10
mum-n-dad	err	10:10
	<ft> while ! git pull; do :; done	10:10
	→ masterkorp joined	10:11
mum-n-dad	<ft> that'll try forever until is works.	10:11
j416	I don't know zsh	10:11
	← amcsi left	10:11
	→ codeshepherd joined	10:12
	→ mcepl joined	10:14
	→ bx2 joined	10:14
	→ opalka joined	10:14
	← galderz left	10:14
mcepl	can anybody help me with this? http://www.fpaste.org/AIwE/ ... there should be a full SVN repostitory there, but git svn seems to be unable to get it.	10:15
	← opalka left	10:15
	→ opalka joined	10:15
	→ bentob0x joined	10:16
	→ christophsturm joined	10:17
	→ ciaranm joined	10:18
	← gaveen left	10:18
	← j416 left	10:19
crab	doener/sitaram: i finished the first draft. comments welcome: http://toroid.org/ams/git-central-repo-howto	10:19
	→ galderz joined	10:20
	← codeshepherd left	10:20
	→ codeshepherd joined	10:20
	← codeshepherd left	10:21
opalka	Is there some option how to enable 'svn git fetch' in verbose mode? It fails for me locally and I would like to know the reason of failure.	10:21
	s/svn git fetch/git svn fetch/	10:21
	→ codeshepherd joined	10:22
	← mtkd left	10:22
	→ mtkd joined	10:23
	← codeshepherd left	10:24
	→ codeshepherd joined	10:25
	← NET\|\|abuse left	10:26
	← codeshepherd left	10:27
	→ locklace joined	10:28
	→ gsan1 joined	10:28
	→ codeshepherd joined	10:28
	← gsan1 left	10:28
	← codeshepherd left	10:28
	→ codeshepherd joined	10:29
	→ urkud joined	10:31
	← tg left	10:31
	→ tg joined	10:31
	→ gaveen joined	10:31
	← airborn left	10:32
	→ bushwakko joined	10:32
	← codeshepherd left	10:32
	→ parasti joined	10:33
	→ codeshepherd joined	10:33
bushwakko	hey, I created a new branch from my existing branch to test some changes. but when i checkout the original branch the changes follow	10:33
	→ pk4r joined	10:34
bushwakko	I did checkout -b testbranch then edited the files, did my test and did a checkout origbranch	10:34
	all the modified files follwed	10:34
	← codeshepherd left	10:34
cbreak_work	of course	10:34
	the working directory is the same all the time	10:34
	so until you commit your changes, they aren't part of the branch	10:34
bushwakko	ok	10:35
	→ codeshepherd joined	10:35
	← gebi left	10:35
	→ Garen joined	10:35
bushwakko	so I should have commited, then created branch, done changes and ocmmited, change branch again and then delete newbranch?	10:35
cbreak_work	depends	10:36
	you commit when the branch you want to commit to is checked out	10:36
	← codeshepherd left	10:36
crab	bushwakko: you can use "git stash" to store uncommitted local changes temporarily, try something out (in a new branch or not), then switch back to the original branch and git stash pop to resume with the changes.	10:37
cbreak_work	if you just want to undo some local changes, you can reset them away without making a temp branch	10:37
	→ codeshepherd joined	10:37
cbreak_work	git reset --hard HEAD kills all uncommited changes	10:37
crab	and yes, use reset to throw away local changes rather than commit and delete the branch.	10:37
cbreak_work	kill might not be the right word.	10:38
shruggar	can one specify multiple refspecs in git config remote.origin.push ? I'd like to push refs/heads/:refs/work/ AND refs/remotes/git-svn:refs/git-svn	10:38
cbreak_work	annihilate. exterminate.	10:38
	← codeshepherd left	10:38
	→ codeshepherd joined	10:39
crab	obliterate.	10:39
cbreak_work	fitting, yes	10:40
mum-n-dad	Anybody good with zsh here? On my git_pull_all.sh at http://pastie.org/927513 - how do I incorporate "retries" into the "while ! git pull; do:; done" (which now retries infinitely)?	10:40
	* do :;	10:41
cbreak_work	with a function?	10:41
	cbreak_work doesn't know this zsh thingie.	10:41
	← codeshepherd left	10:42
doener	shruggar: just add a second push line	10:43
shruggar	doener: I thought config keys need to be unique	10:43
doener	shruggar: or use "git config --add remote.origin.push blabla", the --add takes care of adding a second entry instead of overwriting the old one	10:43
shruggar	add	10:43
	ah	10:43
doener	shruggar: that would make --add pretty useless ;-)	10:44
shruggar	I didn't know about --add :)	10:44
	← geshan left	10:44
	→ codeshepherd joined	10:46
	→ Aides joined	10:47
	← codeshepherd left	10:48
mcepl	can anybody help me with this? http://www.fpaste.org/AIwE/ ... there should be a full SVN repostitory there, but git svn seems to be unable to get it.	10:48
	→ codeshepherd joined	10:48
	← urkud left	10:48
	roop_ → roop	10:50
sitaram	crab: looks nice; do you want to add a link to nathanj's GUI tutorial? just in case?	10:50
	← codeshepherd left	10:50
	→ codeshepherd joined	10:52
	→ jmut joined	10:52
	→ g0bl1n joined	10:53
doener	mcepl: drop the /trunk from the url	10:53
	mcepl: or don't use -s... but there's no standard layout _within_ the trunk/ directory	10:53
	→ scarabx joined	10:54
	← codeshepherd left	10:54
shruggar	is there any effort to get git-svn to use git-notes instead of its horrible revision-map of doom?	10:54
	→ codeshepherd joined	10:54
	← codeshepherd left	10:55
jast	is there actually any standard on how multiple key-value pairs are supposed to be stored in a note?	10:55
mcepl	doener: I don't think it's the problem ... git svn always freezes on r4000 (there should be like 12000 or so revisions) and that's it.	10:56
	→ codeshepherd joined	10:56
	→ Guthur joined	10:56
mcepl	hold on	10:56
shruggar	I seriously doubt it. I've just been using a different ref whenever I want a different namespace, and within each namespace I know the format I expect	10:56
mcepl	yeah, it actually looks like working	10:57
	let's see	10:57
jast	oh, right	10:57
	← codeshepherd left	10:58
zomg	I felt like headdesking hard today when a bunch of git newbies at work were refusing to use command line git, and instead tried doing everything via git-gui	10:58
	:D	10:58
	→ codeshepherd joined	10:59
	→ Remixman joined	10:59
gsan	zomg: consider yourself lucky to use git at work... i have to use svn	11:00
	← codeshepherd left	11:00
	→ codeshepherd joined	11:01
doener	I felt like harddesking today when looking through job offers and seeing "Good Photoshop Skills" as a requirement for a job as a software engineer.	11:01
	s/hard/head/	11:02
Ilari	Ability to make interface mockups using Photoshop?	11:03
doener	hm, now maybe I should use photoshop to adjust a screenshot of my irssi window and pretend I never made that typo	11:03
	← parasti left	11:03
zomg	gsan: well previous project used svn but I used git-svn :)	11:03
	doener: heh	11:04
	← codeshepherd left	11:04
gsan	zomg: trying to myself but running into problems...missing the SVN perl module	11:04
C_sun	so.. install it?	11:05
zomg	I was on windows and msysgit had everything so it worked out of the box	11:05
doener	Ilari: maybe... though I'm not sure that photoshop is the right tool for that job... But I'm not a GUI designer, so I have no clue and do my sketches using pen and paper ;-)	11:05
C_sun	Ilari: ability to draw code, not write it :p	11:06
Ilari	And making programmers design the UI is asking for trouble anyway.	11:06
	→ fai\|safe joined	11:06
shruggar	ask a programmer to design a UI, and you'll get their favorite programming language :)	11:07
posciak	can I see the actual message bodies during --dry-run with git-send-email?	11:07
gsan	C_sun: working on it	11:07
posciak	I also tried just sending the messages to myself only, even managed to supress most of ccs with --no-signed-off-by-cc but patch author (From:) still gets added to cc. Even --suppress-fro didn't turn that off... is this a bug btw?	11:09
	s/--supress-fro/--suppress-from	11:09
	← gaveen left	11:09
	→ roop_ joined	11:10
	← roop left	11:10
	← hyperair left	11:13
	→ parasti joined	11:16
	→ mstormo joined	11:17
jast	posciak: --suppress-cc=author (author doesn't have to be the same person as sender)	11:17
	→ mbrumbelow joined	11:18
posciak	jast: thank you	11:18
jast	and there's always --suppress-cc=all	11:18
posciak	that's exactly what I tried	11:18
	→ froschi joined	11:18
posciak	but it's strange that suppress-from doesn't work	11:19
	the doc says " If this is set, do not add the From: address to the cc: list."	11:19
	so...	11:19
jast	well, from is the sender... even if you suppress that, it still sends to the author, who might incidentally be the same person	11:19
	this is all conjecture, btw... I have no idea what send-email does internally	11:20
posciak	ah right, now I get it	11:20
	so it's not From: from the actual patch, but From as in the sender	11:20
	→ mithro joined	11:21
	← fai\|safe left	11:22
jast	I suppose	11:23
	→ Remixman_ joined	11:25
opalka	doener: Thanks for help related to /trunk at the and and standard layout. It was really the reason of failure	11:25
	← froschi left	11:25
	← Remixman left	11:25
	← mcepl left	11:26
	→ iruediger joined	11:26
	← iruediger left	11:26
	→ iruediger_ joined	11:26
	Remixman_ → Remixman	11:26
	← masterkorp left	11:28
	→ d0k joined	11:28
stepnem	ugh... is there no equivalent to the -F option (--show-function-line) in `git diff'??	11:29
shruggar	I believe there is	11:30
	← mtkd left	11:30
	→ mtkd joined	11:30
stepnem	well, then I'd love to hear about it	11:31
	← mstormo left	11:31
sylr	Hi	11:32
	→ mstormo joined	11:32
sylr	git grep '\s' matches all lines with a space	11:32
	why does git grep '\t' matches all lines with a 't' and not a tab character ?	11:32
	→ badboy_ joined	11:33
	→ TheNewAndy joined	11:34
	→ stamina joined	11:34
	← mstormo left	11:35
	→ drizzd_ joined	11:36
	→ mstormo joined	11:36
	← mbrumbelow left	11:36
joel____	autoban for referencing ipad!	11:38
shruggar	stepnem, look for "diff" in the manual for gitattributes	11:38
	← frakturfreak left	11:38
	← drizzd left	11:39
	→ froschi joined	11:39
	→ gebi joined	11:40
	→ roger_padactor joined	11:42
	→ hyperair joined	11:42
	→ sagsousuke joined	11:44
	→ psoo joined	11:44
	← mstormo left	11:46
avar	Is it possible to make git format-patch attach the the equivalent of 'git format-patch --stdout > a_patch.patch' ? There's --attach but that still leaves the commit message outside the attachment.	11:46
	→ urkud joined	11:49
	← mtkd left	11:50
	→ frakturfreak joined	11:50
	← psoo left	11:55
avar	doesn't look like it, suckery!	11:55
	← iruediger_ left	11:55
	→ ponto_ joined	11:57
	→ domcat joined	11:57
	→ pantsman joined	11:57
	← ntoll left	11:57
ponto_	hi, how does git-svn handle empty directories?	11:57
	→ iruediger joined	11:57
	← loincloth left	11:58
	← bushwakko left	11:58
	← simplechat left	11:58
	→ bushwakko joined	12:00
avar	I can't even find the source for git-format-patch, hrm	12:01
	→ masterkorp joined	12:01
	→ dumb joined	12:02
	→ SlickT10 joined	12:02
dumb	please need help	12:03
	anyone familiar with github	12:03
	→ fai\|safe joined	12:05
ArjenL	avar: format-patch is in builtin/log.c	12:06
avar	thanks	12:06
mum-n-dad	divVerent: So I ended up with http://pastie.org/927642	12:06
	For those of you who don't know what I'm talking about - it's this shell script to pull all my repos	12:07
	When I get fatal: The remote end hung up unexpectedly - it's suppose to retry pulling 5 times	12:07
	looks like it's trying infinitely	12:08
	but why am I getting fatal: The remote end hung up unexpectedly in the first place?	12:08
divVerent	ask the server admin	12:08
	in case the repo is a git:// and hosted via inetd, it may be inetd's connections-per-second limit	12:08
doener	mum-n-dad: you need to set retries to 5 in the loop, not outside it	12:09
mum-n-dad	divVerent: It's github - do you know what its limit is?	12:10
divVerent	no	12:10
mum-n-dad	doener: Excellent excellent man!	12:10
doener	mum-n-dad: first repo: retries goes from 5 to -1	12:10
	→ fachschaft joined	12:10
	← fachschaft left	12:10
	→ jfkw joined	12:10
	→ radarek joined	12:10
doener	mum-n-dad: second repo, retries stays < 0, forever	12:10
	→ mariusSO joined	12:11
	→ codetroll joined	12:11
avar	how do I amend the content of the last commit? Do I need to reset HEAD^ and commit again (copy/pasting the commit message) or is there something like --amend?	12:11
doener	avar: yes, there is: --amend	12:12
divVerent	you mean, like git commit --amend?	12:12
	;)	12:12
avar	The content, not the commit message. At least the commit message is the only thing in my editor.	12:12
divVerent	oh	12:13
	you first do the content changes you want	12:13
mum-n-dad	doener: while ! git pull; do ((retries--)) && :; done	12:13
divVerent	and then you git add the files	12:13
mum-n-dad	like that mybe?	12:13
divVerent	and then git commit -a	12:13
	*--amend	12:13
	avar: there is also an alternate way: git rebase -i HEAD^	12:13
	and then change the "pick" to "edit"	12:14
avar	Ah right, then I can take in the current index during --amend, nice	12:14
divVerent	obviously, that way makes most sense for ediitng more than one commit :P	12:14
	→ afrasi joined	12:14
divVerent	right	12:14
	and I bet there is also a git-reset command that undoes the last commit, and keeps the workign tree, and lets you commit again ;)	12:14
mum-n-dad	doener: Any clue?	12:15
	← SlickT10 left	12:18
doener	mum-n-dad: http://pastie.org/927661	12:19
	→ liye joined	12:20
doener	divVerent: reset HEAD^. The working tree is only affected with --hard. And --soft makes it keep the index as well.	12:21
divVerent	ah, nice	12:21
	→ tPl0ch joined	12:21
divVerent	would be nicer if the options would be easier to understand	12:21
	→ ntoll joined	12:21
doener	and rebase -i HEAD^ to edit the commit referenced by HEAD is pure overkill	12:21
divVerent	like git reset --both, --index, --refonly ;)	12:21
	or even --work, --index, --ref	12:21
	← froschi left	12:22
	← fai\|safe left	12:22
doener	well, --both would have to be --all-three ;-) (ref, index, worktree)	12:22
divVerent	right :P	12:22
mum-n-dad	doener: Awesome dude..	12:22
divVerent	but it ALWAYS resets the ref ;)	12:22
knittl	git commit --amend …?	12:23
divVerent	actually... what if it did not?	12:23
	→ dominikh joined	12:23
divVerent	why shouldn't I be able to JUST reset the index to a specific rev :P	12:23
jast	the second form of reset does not reset the ref	12:23
divVerent	odd... doesn't it duplicate git checkout COMMIT -- paths?	12:24
jast	there are different commands for affecting only the index	12:24
	no. reset in second form doesn't touch the working tree.	12:24
tPl0ch	Hi, I have a repository that has some submodules in it, and they work flawlessly. Now I want to clone a bare repo from that and put that on a server. If I do git clone --bare project project.git and send it via scp to the server and try to clone the repo from the server, it gives me following error: http://pastebin.com/TMt8Azjt . I am using git latest (1.7.0.5)	12:24
	← dajero left	12:24
mum-n-dad	doener, divVerent: http://www.zshare.net/download/552763591f4802ce/	12:24
divVerent	ah	12:24
	→ Bass10 joined	12:24
mum-n-dad	thanks guys	12:24
	this is my gift to you	12:24
	→ Adaptee joined	12:26
	→ ericlee joined	12:28
	← scarabx left	12:29
jast	tPl0ch: why the detour via a bare repository?	12:29
	couldn't you just init an empty repository on the server and directly push to that?	12:30
Alagar	knittl: thank for your time. i need to maintain all the code base in centralized server.	12:30
	→ kukks joined	12:30
	→ rolfb joined	12:31
	→ mythos joined	12:32
	← mythos left	12:32
	→ mythos joined	12:32
	→ jonalv joined	12:33
	← christophsturm left	12:33
	→ christophsturm joined	12:33
knittl	Alagar: then either set up a git server or use ssh	12:34
tPl0ch	jast: well, ssh ing into the server and doing an init will take the same time as clone bare and send via scp, but I will try	12:34
	← christophsturm left	12:34
	→ christop_ joined	12:34
jonalv	Okey guys call me crazy or missinformed (but please tell me why) but I want to do a merge using opendiff. IT is awesome for resoving conflicts but I want to do the complete merge using it no matter if the merge was without conflicts. I guess you could compare it to doing a interactive merge but using open diff. Can this be done? And if so, how?	12:34
	→ tstclair joined	12:35
jast	tPl0ch: I was asking mainly because there might have been some hidden pitfalls triggered by doing the manual copy thing. it usually works out pretty well, but well. can't hurt to try. :)	12:35
knittl	jonalv: where's the problem with git's internal diffing algorithm?	12:35
	← saLOUt left	12:36
jast	jonalv: problem is that recursive merge often does several merges in the background and sort of combines the result... that might give you quite a lot of manual work if you did all of that merging interactively	12:36
Alagar	i got the following documentation for the git server seting up. http://engineeredweb.com/blog/10/2/building-your-own-git-server	12:37
	which one is the best for the server setup	12:37
jonalv	knittl: firstly I have one big commit so I would have to split it up a zillion times. Second I just love the way open diff has one version to the left one version to the right and the end result in the bottom. Instead of each version as two rows over each other. I guess I am just picky with my visualisations... :)	12:37
Alagar	knittl: could you please advice me.	12:37
	← skoop left	12:37
knittl	Alagar: there's ssh missing	12:38
	jonalv: i don't know how, but you can tell git to use a different conflict visualizer	12:38
	← codetroll left	12:38
	→ Meow`` joined	12:38
	← mithro left	12:38
knittl	diff.* or difftool in git config iirc	12:38
jonalv	knittl: yea, but the thing is there is no conflict. I just want to visualze the diffs and "cerry-pick" the parts I want...	12:39
Alagar	knittl: iam able to view for the following methods for git integration Gitosis, gitolite, gitweb	12:39
	knittl: which one is the best for our centralized codebase management. ?	12:39
knittl	Alagar: that's why i said it is missing	12:39
	Alagar: i don't know i only use git locally on or servers not administered by me	12:40
tPl0ch	jast: nope, no luck, still the same error	12:40
	→ mithro joined	12:41
knittl	jonalv: how do you want to cherry pick when using another program …?	12:41
jonalv	knittl: manually, sry maybe cherry+pick isn't exactly what I want to do. IT's just sort of what I want to do	12:41
	knittl: What I want to do is apply some parts of a commit but not all of them	12:42
Alagar	knittl: thanks, could you please give me any good documentation for the git centralized server setup ?	12:42
jonalv	knittl: And I want a fancy user frienldy interface for doing that. (and opendiff seems like a good candidate)	12:42
knittl	so you want to do a git add --patch?	12:43
	Alagar: http://www.google.com/search?ie=UTF-8&oe=UTF-8&sourceid=navclient&gfns=1&q=man+git+server	12:43
jonalv	knittl: yea	12:43
knittl	look into diff.* or difftool	12:44
	jonalv: http://www.kernel.org/pub/software/scm/git/docs/git-config.html	12:44
	→ liye_ joined	12:44
	← liye left	12:45
jonalv	knittl: was there any particular option you had in mind on that page?	12:45
	liye_ → liye	12:45
jonalv	knittl: oh nm (I missed you line before)	12:46
knittl	jonalv: diff.* and difftool	12:46
curtana	git can already use opendiff for a merge tool. but i don't think the merge resolution process works in the way you want it to jonalv	12:46
jonalv	curtana: oh	12:46
	curtana: how so_	12:47
	?	12:47
curtana	git mergetool -t opendiff	12:47
jonalv	curtana: yea but that is for resolving conflicts	12:47
knittl	just read the manpage about git config dp	12:48
	* :P	12:48
	→ auslander joined	12:48
	← auslander left	12:48
	→ auslander joined	12:48
	→ DavidKlein joined	12:50
	→ Zephyrus joined	12:51
curtana	jonalv, yes, exactly. i don't think you can do what you want with git	12:51
Alagar	knittl: thanks. do you know how to setup ssh auth using ssh keys. ?	12:51
jonalv	curtana: no I am beginning to fear the same... :)	12:51
curtana	though maybe...	12:51
	← dumb left	12:52
curtana	you could use git-difftool -t opendiff <commit that you want to pull bits from>	12:52
	then use the difftool to apply hunks to the local tree	12:52
	it would not be a merge or cherry pick in the git meaning of the term. but it would perhaps do what you want	12:52
jonalv	curtana: just to clearify the lingo, with " use the difftool to apply hunks to the local tree" do you mean to stage?	12:53
knittl	Alagar: man ssh ;)	12:54
	Alagar: http://pkeck.myweb.uga.edu/ssh/	12:54
	← Zephyrus left	12:54
	→ Remixman_ joined	12:56
	← Remixman left	12:57
	← mithro left	12:57
	← d0ugal left	12:58
	→ d0ugal joined	13:00
jonalv	curtana: that worked and it was exactly what I wanted to do! Thank you! curtana++	13:01
	→ psankar joined	13:01
	← urkud left	13:01
	→ mdr22 joined	13:02
Alagar	knittl: the git server installation is very hard for me. iam not able to view any good manual	13:02
knittl	which system are you on? @ Alagar	13:02
Alagar	knittl: iam using CentOS 5.3	13:02
	knittl: i have done the git installation using the following commands( #yum install git	13:03
	knittl: iam not able to findout how to config then next	13:03
knittl	Alagar: have you read the link about ssh, i provided?	13:04
	do that, then you have a working ssh server with keys	13:04
	← masterkorp left	13:05
Alagar	knittl: thanks for your effort and time. i have done the ssh key generation but then iam not able to findout next steps	13:05
knittl	copy the public keys to the server	13:06
	try logging in	13:06
	← Adaptee left	13:07
	← Remixman_ left	13:07
	→ Remixman joined	13:07
	← g0bl1n left	13:08
	→ loincloth joined	13:08
	→ Adaptee joined	13:08
	← psankar left	13:10
	→ przemoc joined	13:11
	→ mithro joined	13:13
	→ urkud joined	13:13
	→ Guest24555 joined	13:14
	→ Semikolon joined	13:14
cbreak_work	anyone got an URL to the bash script that provides auto completion of git commands and their parameters?	13:15
knittl	cbreak_work: it's in the git repository	13:16
	in contrib/completion	13:16
cbreak_work	neat, thanks	13:16
shruggar	545bd4b38368e3c2a3958133bbeef6a19e831fff	13:16
Gitbot	[git 545bd4b38]: http://tinyurl.com/y37pyga [blob]	13:17
shruggar	I wonder if gitbot would pick up on... HEAD:contrib/completion/git-completion.bash	13:17
	← Remixman left	13:17
shruggar	hm, nope	13:17
	→ Remixman joined	13:17
shruggar	oh, probably the HEAD it doesn't like, I /am/ rather dumb	13:18
	→ royalty joined	13:18
knittl	master	13:18
	refs/heads/master	13:18
	hmm…	13:19
	→ gaveen joined	13:19
	Guest24555 → robinr	13:19
shruggar	well 545bd4b got the link, even if it was completely unnamed :)	13:19
	→ tanoku joined	13:19
	← urkud left	13:20
robinr	anyone with the same problem as I. make barfs on xxdiff	13:20
	← Meow`` left	13:20
robinr	it's not git version, but my kubuntu I guess	13:20
	← frakturfreak left	13:22
	→ _iron joined	13:23
	← royalty left	13:23
	← christop_ left	13:26
	→ christophsturm joined	13:27
	→ y10 joined	13:27
	← psynaptic left	13:28
	→ psynaptic joined	13:28
	← psynaptic left	13:28
	→ psynaptic joined	13:28
	← psynaptic left	13:29
	→ frerich joined	13:30
	→ frakturfreak joined	13:30
	← y10 left	13:30
	← liye left	13:30
cbreak_work	nice, git-completion can even make my prompt show the branch I am on	13:30
frerich	Hi! Is anybody aware of some web pages about different models for doing software releases of source code which is managed with git? I'm interested in different approaches to make sure that bug fixes from a stable branch end up in the branch for the next major release, and how to do so without having developers forget about it.	13:31
	→ g0bl1n joined	13:31
	→ liye joined	13:32
	→ Remixman_ joined	13:34
	← Remixman left	13:34
	Remixman_ → Remixman	13:34
RandalSchwartz	you mean, besides the text that comes with git itself?	13:34
	frerich?	13:34
	→ camonz joined	13:35
RandalSchwartz	"man gitworkflows"	13:35
Gitbot	RandalSchwartz: the gitworkflows manpage can be found at http://git.or.cz/man/gitworkflows	13:35
RandalSchwartz	you want something more than that?	13:35
	→ aspotashev\|eeepc joined	13:35
	→ mefesto joined	13:35
frerich	RandalSchwartz: Yes, exactly - I saw that already but I wondered how other people do it. I've been talking to the Amarok people (a KDE media player) and learned that they just tag 'master' every now and then and release that - and all features (and bugfixes) are done in separate branches and eventually merged into master, so that 'master' never gets and commits but merge commits.	13:36
	→ nbaum joined	13:36
	→ [Po]lentino joined	13:36
frerich	Another model I learned was that bugfixes are always done in a 'patchlevel' branch, from where they occasionally get merged into a 'minor release' branch, from where they eventually get merged into a 'major release' branch.	13:36
RandalSchwartz	Yeah, I've seen that done by past clients	13:37
frerich	So for a software with x.y.z releases, there's one branch for X releases, one for Y and one for Z.	13:37
	← mithro left	13:37
frerich	We'll need to do a new major release of our (git hosted) software real soon now, after which we can start a new branching workflow. So I'm researching a bit about different approaches and how much human work they require.	13:38
crab	frerich: nvie.com/git-model	13:38
	← vu3rdd left	13:38
frerich	crab: Interesting link, thanks a ton! That's exactly the kind of thing I'm looking for.	13:39
nbaum	Is there a way to add only changed lines matching a particular regex to the index?	13:40
frerich	nbaum: You could use 'git add --patch' and then use the 'find next hunk to stage given a regexp' feature.	13:41
	→ Remixman_ joined	13:42
	← Remixman left	13:42
	→ level09 joined	13:43
level09	if i have a team of many developers, should I assign different ssh logins for each dev ?	13:43
	or can i just use one ssh login for all ?	13:43
	→ simNIX joined	13:43
	← liye left	13:43
shruggar	level09: see gitolite	13:43
level09	does that mess with the development workflow ?	13:43
	level09 looking	13:44
	← dominikh left	13:44
	→ psoo joined	13:44
	→ liye joined	13:44
	→ dominikh joined	13:44
crab	frerich: the merge-into-master model works too, you can even commit stuff directly to master without any ill-effects.	13:44
nbaum	frerich: There's a few hundred files, with one of these changes per file. / seems to only search to the end of the current file.	13:45
	→ LiamH joined	13:45
crab	frerich: just don't commit things to master and cherry-pick to backport branches. git is positively awful at that. better to merge up than trickle down.	13:45
	→ spearce joined	13:46
crab	though personally i find it very hard to do that (work in the old branches)	13:46
	→ TeckniX joined	13:46
	← spearce left	13:46
frerich	Yes, it certainly requires some discipline.	13:47
	→ spearce joined	13:47
	→ cannonball joined	13:47
	→ Remixman joined	13:47
	← opalka left	13:47
	← psoo left	13:47
	← Remixman_ left	13:48
	← liye left	13:48
crab	i think it means, in practice, that your older versions become stale sooner	13:50
	← auslander left	13:51
	← Remixman left	13:52
	→ Remixman_ joined	13:52
	← robinr left	13:53
	→ mithro joined	13:54
	→ manyoso joined	13:54
	→ petercoulton joined	13:56
	→ psoo joined	13:57
	→ smuf joined	13:57
	→ Remixman joined	13:57
	← Remixman_ left	13:58
	← giallu left	13:58
	→ t0rc joined	13:59
	← zoniq left	13:59
	→ zoniq joined	14:00
	← ph^ left	14:01
	← domcat left	14:03
	← Remixman left	14:03
	→ nevyn^_ joined	14:04
	← nevyn^ left	14:04
	nevyn^_ → nevyn^	14:04
curtana	jonalv, you're welcome ;)	14:04
	→ sepen joined	14:05
	← petercoulton left	14:05
	← nevyn^ left	14:05
	→ nevyn^ joined	14:06
	→ petercoulton joined	14:06
sepen	hi! could be possible to 'hack' my working copy to modify every date appeared in 'git log'?	14:06
	could be possible in the repository too?	14:06
curtana	jonalv, but get into the habit of commiting small, individual changes. a good rule of thumb is: if you would want to cherry-pick this change, then it should be in its own commit	14:06
	→ [Po]lentino_ joined	14:07
	← _iron left	14:07
	→ auslander joined	14:08
	→ jrmuizel joined	14:08
jonalv	curtana: yes, now this was for some text that I got sent to me...	14:08
	← [Po]lentino left	14:09
	→ roop__ joined	14:09
	← roop_ left	14:09
	→ unix_remote joined	14:09
	← dmg left	14:09
curtana	sepen, depends, what do you want to do to the dates	14:09
sepen	curtana, replace year 2010 by 2009	14:10
tty1	hi guys	14:10
curtana	don't think you can do that easily	14:10
	semi-easily	14:10
tty1	when i do git status it shows a file which i removed as being copied to a file to which it was not copied (it is a new file but it is unreleated to the file i deleted). is there a way i correct this misconception on git's part before i commit?	14:11
curtana	you could run git-format-patch, edit the files that are spat out, and re-apply them with git-am perhaps	14:11
sepen	hmmm	14:11
	← jrmuizel left	14:12
RandalSchwartz	does the new file have roughly the same contents as the old file?	14:13
	it must, or git wouldn't think it's a copy	14:13
sepen	curtana, can't be possible directly on the repo?	14:14
tty1	RandalSchwartz, well the old file was very minimal, only had <10 ines in it.. so despite being an entirely new class with entirely new methods the header ont he file (which is larger) i guess may make it think its similar when its not.	14:15
	RandalSchwartz, the license header is the majority of it	14:15
RandalSchwartz	one way would be to have two separte commites	14:15
	commit first without the new file, then commit with	14:15
	← roop__ left	14:15
RandalSchwartz	that'd ensure no relationship	14:15
tty1	RandalSchwartz, there has got to be a better way	14:16
	→ psankar joined	14:16
RandalSchwartz	well - if the license gets edited by someone, don't you want that to apply there?	14:16
tty1	RandalSchwartz, if i do it in two seperate commits then the intermediate commit will be broken, i dont want that	14:16
	RandalSchwartz, sure	14:16
RandalSchwartz	then don't worry about it	14:16
	that's the only way that "copy" info will be used	14:16
	← tstclair left	14:16
tty1	RandalSchwartz, i just dont want the logs to show it as the same file	14:16
RandalSchwartz	the logs are fiction :)	14:16
tty1	bah	14:17
RandalSchwartz	git isn't "remembering" the copy	14:17
	← jony left	14:17
RandalSchwartz	it's observing the copy	14:17
	every time :)	14:17
	if you don't pass -C, it won't find that	14:17
	and as you say, it's mostly the same file, so I would want to see it as a copy anyway	14:17
divVerent	there is a reason why not	14:18
	"git blame" :P	14:18
	→ Yuffster joined	14:18
	→ jboom joined	14:18
tty1	sounds reasonable	14:19
RandalSchwartz	even git blame doesn't follow semi-copies without -C	14:19
	so again, if you're seeing it as a copy, it's because you asked for it	14:20
	→ mtkd joined	14:20
divVerent	what are "semi-copies"?	14:20
RandalSchwartz	copies detected by "similarity"	14:21
	as opposed to having identical content	14:21
	→ GarethAdams\|Work joined	14:21
	← TheNewAndy left	14:21
	→ tstclair joined	14:21
divVerent	RandalSchwartz: ah, okay	14:22
	so when I move a file and hten change it	14:22
	I should do a separate commit	14:22
	so git blame works wihtout -C :P	14:22
RandalSchwartz	if you can	14:22
divVerent	sure	14:22
	this may cause a broken commit of course	14:22
RandalSchwartz	sometimes, it breaks the build	14:22
	yeay	14:22
divVerent	like with java classes, where the file name must be in the file	14:22
RandalSchwartz	I just flag those in the commit message	14:22
	"BISECT IGNORE"	14:22
divVerent	does git-bisect know about that? ;)	14:23
	→ kpreid_ joined	14:23
	← kpreid_ left	14:23
RandalSchwartz	so if someone is bisecting, they'll know.	14:23
divVerent	ah, only for the user :P	14:23
RandalSchwartz	no - but there should be some sort of hook like that :)	14:23
	← aspotashev\|eeepc left	14:23
divVerent	a bisect script could sure return a compile-fail status immediately :P	14:23
	instead of trying	14:24
	← jmut left	14:24
RandalSchwartz	I'm not sure a script can return "skip"	14:24
	→ kpreid_ joined	14:24
	← frakturfreak left	14:24
RandalSchwartz	and you don't want to return "bad", because that would imply an endpoint	14:24
divVerent	script can return three states	14:25
	good, bad , compile fail	14:25
RandalSchwartz	ahh - it can	14:25
divVerent	latter is skipped	14:25
RandalSchwartz	exit 125 = skip	14:25
	nice	14:25
divVerent	so it wouldn't be hard to check the commit message and immediately exit 125	14:26
	for faster bisecting :P	14:26
RandalSchwartz	indeed	14:26
C_sun	So. How do I search for a string (a la git log -S) in all unreachable commits?	14:26
	RandalSchwartz wanders off	14:26
	→ tgunr joined	14:27
	← mtkd left	14:27
	→ Mpenz joined	14:28
shruggar	C_sun: I expect something involving "git fsck --unreachable" and a 'for' loop	14:29
C_sun	shruggar: I fear that that might only show the unreachable heads, not all unreachable commits.	14:29
	→ RobertLaptop joined	14:29
	← tgunr left	14:29
shruggar	C_sun: but if you have all unreachable heads, "git log -S" on each of them would do the job	14:31
crab	what is an unreachable head?	14:31
	oh, i get it.	14:31
C_sun	a head is a commit who is not the parent of another commit.	14:31
shruggar	I'm pretty sure that something only counts as "unreachable" if nothing at all points to it. So /all/ unreachable commits would be heads	14:32
	← psankar left	14:32
shruggar	git log -SwhatIwant "$(git fsck --unreachable\|sed -n 's/unreachable commit //p')"	14:32
	→ mtkd joined	14:32
	← jovianjake left	14:33
shruggar	wait, cancel those quotes	14:33
	force of habit :)	14:33
	→ psankar joined	14:34
C_sun	Great ... looks like git-fsck has an endless loop	14:34
shruggar	it really just takes a long while. You'll probably want to redirect output to somewhere	14:35
C_sun	No I mean when I run git-fsck --unreachable -v, it just stops halfway through	14:35
mum-n-dad	doener: while ! git pull && sleep 5 && ((retries--)); do :; done	14:35
	doener: Is that the right way to add the sleep?	14:35
	I still haven't figured out what causes the "fatal: The remote end hung up unexpectedly" though.	14:36
	As I just wrote in #github:	14:36
	→ froschi joined	14:36
mum-n-dad	I just wrote this zsh script to mass pull all my GitHub repos - http://pastie.org/927642 - but I keep getting "fatal: The remote end hung up unexpectedly" for certain repos. What's GitHub's connections-per-second limit, i.e. must I wait a number of seconds in between my pulls? Could it be a firewall limitation?	14:36
C_sun	mum-n-dad: it certainly looks ugly from a sh perspective	14:36
	← t0rc left	14:36
	→ dmg joined	14:36
mum-n-dad	C_sun: is there a better / neater way to do it?	14:36
	C_sun: "while ! git pull; do :; done" will keep retrying "git pull" if it fails	14:37
	← LRN left	14:37
	← psankar left	14:37
C_sun	for ((i = 5; i >= 0; --i)); do if git pull; then break; fi; sleep 5; done;	14:38
shruggar	git pull && break \|\| sleep 5? ;)	14:38
	→ psankar joined	14:40
	→ dj_tjerk joined	14:40
mum-n-dad	shruggar: what does that do? looks neat..	14:40
	C_sun: oh heeeeeeeelllllll no :D	14:40
C_sun	heel no?	14:40
	heh	14:40
	← amcsi_ left	14:40
	→ giallu joined	14:40
	→ pklingem joined	14:40
shruggar	foo && bar \|\| baz is a shorter way to type if foo; then bar; else baz; fi	14:41
	← dmg left	14:42
C_sun	shruggar: perhaps. But it reads like strcmp(foo, bar) == 0, printf("that matches") && a=5, b=4 \|\| break;	14:43
	→ marshall joined	14:43
C_sun	(you'd never do that in C)	14:43
mum-n-dad	ok how about this - http://pastie.org/927926 - will it break the script if "git pull" gets an error i.e. the "fatal: The remote end hung up unexpectedly"?	14:43
PerlJam	C_sun: you might for an obfuscation contest	14:43
thiago	shruggar: it's not exactly the same thing	14:44
mum-n-dad	or will the loop just continue?	14:44
thiago	shruggar: it's equivalent to: if foo; then if ! bar; then baz; fi; else baz; fi	14:44
	← mdr22 left	14:45
shruggar	thiago: I suppose that's true, yes. In this case "bar" was "break", which I don't actually expect to fail :)	14:45
thiago	ok, then it makes sense	14:45
shruggar	but yes, as a general case it's not equivalent	14:46
	→ tedoc2000 joined	14:46
	← magn3ts left	14:46
mum-n-dad	Anyone? At http://pastie.org/927926 - will "while ! git pull; do break; done" stop the script if "git pull" fails, or will the loop keep on running?	14:47
wereHamster	mum-n-dad: that will run git pull one time, it's a long way to write 'git pull'	14:48
	→ jrmuizel joined	14:49
mum-n-dad	Sorry?	14:49
	That script will run git pull in all my repos - I have a lot of them	14:49
	Trying to do a mass pull script	14:49
wereHamster	if git pull succeeds, the loop body is not executed, if pull succeeds, the loop body executes which breaks out of it right away..	14:50
	→ adaro joined	14:50
mum-n-dad	I keep getting "fatal: The remote end hung up unexpectedly" though - so I want the script to terminate whenever that happens	14:50
thiago	mum-n-dad: while ! git pull; do sleep 10s; done	14:50
mum-n-dad	wereHamster: Did you see the whole script though?	14:50
	thiago: But that won't terminate the script if it fails, will it?	14:51
	that'll just make it sleep 10 seconds if it fails, and then the whole loop will run again?	14:51
thiago	if it fails, it will sleep for 10 seconds	14:51
mum-n-dad	(when I say loop, I refer to http://pastie.org/927926)	14:51
	thiago: Not quite what I need but its cool	14:51
thiago	do you want to keep on trying to pull?	14:52
mum-n-dad	no, if it fails i want the whole script to stop	14:52
adaro	there is this repository in svn that contains a number of projects, two of which i would like to fork, ideally i want to copy just those projects to a git repository and keep up to date with changes on the svn trunk	14:53
	is there an easy way to do this ? i thought about setting up a cron job that uses git svn and pushes the changes to a remote repostory	14:54
	→ shennyg joined	14:54
adaro	and branch of that, but i am wondering if that will allow me to easily pull in the changes and if its a problem that i am not copying the entire repository	14:54
thiago	mum-n-dad: git pull \|\| break;	14:55
	← frerich left	14:55
adaro	but just two projects (out of 20)	14:55
mum-n-dad	thiago: will that break the whole script, including the parent for folder in $PWD/**/.git(:h); do loop?	14:56
	→ jonshea joined	14:56
	→ lanthan joined	14:56
stepnem	shruggar: great, that works, thanks! (the diff -F <=> .gitattributes thing)	14:56
thiago	mum-n-dad: yes	14:56
	mum-n-dad: I see you're using zsh, so yes	14:57
mum-n-dad	cool	14:57
	awesome thiago.. thanks!	14:57
	thiago: http://pastie.org/927926	14:58
	what do you think?	14:58
	→ balgarath joined	14:58
mum-n-dad	my git_pull_all.sh	14:58
knittl	isn't there a git rebase -i master...branch?	14:58
shruggar	stepnem, oh, good to know. I was just skimming the manpage and it looked relevant :)	14:58
knittl	to rebase onto the merge base of the two?	14:58
balgarath	hi all. trying to checkout a remote branch and getting this: http://pastie.org/927958	14:59
	tried a few solutions on google, but nothing working yet	14:59
cYmen	Is it possibly to remotely create a repository?	14:59
	Say I just want to create a bare repository for backups at a machine I have ssh access to..	14:59
balgarath	lol, nm	15:00
	im retarded	15:00
	→ hakunin joined	15:00
balgarath	and have been up waaay too late :)	15:00
	← bushwakko left	15:00
	← mtkd left	15:00
cbreak_work	knittl: does that make sense? if you're on branch, then rebasing onto the merge base is a no op	15:01
	cYmen: log in via ssh, and git init --bare	15:02
	→ magn3ts joined	15:02
mum-n-dad	Thiago: At http://pastie.org/927926 - this script should: if "git pull" fails, retry it 5 times (but sleep 5 seconds prior to each attempt) - and if that too fails, stop the whole script. Am I doing it right?	15:02
	→ altrux joined	15:02
	← psoo left	15:03
tty1	anyone ever have any problem with git causing their network to become unstable every couple of days?	15:03
cbreak_work	git doesn't do anything on the network on it's own	15:04
cYmen	cbreak_work: alright	15:04
	→ brodyberg joined	15:04
cYmen	cbreak_work: Now what's the easiest way to attach the two repositories?	15:04
cbreak_work	you don't need to do anything on the remote	15:05
	on the one you want to push from, you can git remote add the other	15:05
jonalv	tty1: Have you ever noticed that it often rains when you are watching television?	15:05
cYmen	cbreak_work: should I use --mirror? I don't understand the stuff in the man page :/	15:05
	→ masak joined	15:06
cbreak_work	no	15:06
	← stamina left	15:06
cbreak_work	--mirror isn't a valid flag for git remote	15:06
cYmen	oh...	15:07
	← kar8nga left	15:07
cYmen	my man page says otherwise	15:07
	git remote add [-t <branch>] [-m <master>] [-f] [--mirror] <name> <url>	15:07
cbreak_work	hmm... you're right	15:08
	← radarek left	15:08
cbreak_work	but don't do it anyway	15:08
cYmen	kk	15:08
	is there a way to specify the repository with relative paths?	15:08
	so cymen@server:foo for cymen@server/home/cymen/foo ?	15:09
	→ jeevan_ullas joined	15:09
wereHamster	user@host:path/relative/to/users/home	15:09
knittl	cbreak_work: i am doing a rebase -i to sqash, fixup and reorder commits	15:09
cbreak_work	man git-clone has a ton of possible url syntaxies	15:09
Gitbot	cbreak_work: the git-clone manpage can be found at http://git.or.cz/man/git-clone	15:09
	→ Zephyrus joined	15:09
	← psankar left	15:10
	← hakunin left	15:10
	→ hakunin joined	15:10
	→ fcoury joined	15:11
cYmen	wereHamster: I tried that and it didn't work. Let me double check.	15:11
	→ icwiener joined	15:11
	→ darwin_ joined	15:12
cYmen	wereHamster: It tries to add the :foo part to the hostname O_o	15:12
knittl	cYmen: it's user@host/path/…	15:13
	← sagsousuke left	15:13
wereHamster	[user@]host.xz:path/to/repo.git/ (straight from the man page)	15:14
cYmen	maybe it's a problem with ssh	15:14
	← cbreak_work left	15:15
wereHamster	cYmen: are you using gitolite?	15:15
	← carl- left	15:15
cYmen	no	15:15
	I think..at least it doesn't ring any bells.	15:16
wereHamster	so you do have shell access to the server, right?	15:16
cYmen	yup	15:16
	→ akamaus joined	15:16
wereHamster	what does 'ssh user@host ls path/to/repo.git' show?	15:16
	JEEBcz → JEEB	15:17
mum-n-dad	I think I'm done with my git_pull_all.sh: http://pastie.org/927926	15:17
	Thanks guys for your help	15:17
cYmen	wereHamster: HEAD config ...	15:17
	← lanthan left	15:17
wereHamster	cYmen: that looks correct. How about 'git ls-remote user@host:path/to/repo.git' ?	15:18
	→ stamina joined	15:18
cYmen	bleagh	15:18
	I added ssh:// in front and that breaks it.	15:18
wereHamster	yes, it does, as per man page	15:19
cYmen	"An alternative scp-like syntax may also be used with the ssh protocol:"	15:19
level09	how do I deal with @ in the password ?	15:19
cYmen	sounds to me like it should be used with the same ssh:// prefix	15:19
level09	tried escaping	15:19
	and %40	15:19
	but did not work	15:20
cYmen	wereHamster: Thanks!	15:20
knittl	so, let's grep for the commit i am pretty sure i saw a while back	15:20
wereHamster	cYmen: scp-like syntax doesn't use the ssh:// prefix!	15:20
	→ engla joined	15:21
cYmen	Yeah, I was only justifying my error...	15:21
wereHamster	if in doubt, follow the man pages ;)	15:21
	→ peritus- joined	15:21
peritus-	hey. what's the opposite of git-format-patch ? I want to add 12 .patch files to my HEAD (complete with date/author/commit message/etc)	15:22
crab	peritus: git-am	15:22
knittl	peritus-: git am	15:22
peritus-	crab, knittl: but they are not in a mailbox :(	15:22
	but i'll read man git-am again	15:23
Gitbot	peritus-: the git-am manpage can be found at http://git.or.cz/man/git-am	15:23
knittl	ah, 9f21e97 has to do with rebase and ... syntax	15:23
Gitbot	[git 9f21e97]: http://tinyurl.com/y82b324 -- rebase: fix --onto A...B parsing and add tests	15:23
knittl	so it only works with --onto, i see	15:23
	← kpreid_ left	15:24
knittl	peritus-: just give it a file which looks like a mailbox	15:24
tek0	can anyone imagine a solution for working with GIT_DIR=foo/.git-foo GIT_WORK_TREE=foo git submodule update --init? git says "fatal: working tree 'foo' already exists"	15:26
peritus-	knittl: :*	15:26
tek0	updating after manually cloning works	15:26
knittl	:>	15:26
	→ seanmccann joined	15:26
	← seanmccann left	15:26
	→ disappearedng joined	15:26
	← raichoo left	15:27
	→ disappearedng_ joined	15:27
shruggar	is there a tool for easily deleting branches from a remote which do not exist locally? I would like to delete all refs/heads/work/* from the remote "zaonce" which do not exist in my local refs/heads/*	15:29
	← tPl0ch left	15:29
	→ frakturfreak joined	15:30
	→ ProperNoun joined	15:30
	→ Nihathrael joined	15:30
	← ProperNoun left	15:30
	→ ProperNoun joined	15:30
offby1	hmm	15:31
	I guess "git remote prune zaonce" isn't it.	15:31
shruggar	yeah, that's the opposite	15:31
	→ kpreid_ joined	15:31
crab	wereHamster: did you find it?	15:32
cYmen	hm..I want to version control some html pages and easily push or whatever them to the apache documentroot. What would be a good way to do that? Should I just use a shell script and copy everything?	15:32
Nihathrael	Hi all, i'm having a problem with non http pushes. Git will always hang after the following message: http://gist.github.com/373818 . I've tried fiddeling with my bashrc because i read it's not supposed to contain any echos and stuff, otherwise ssh might not work, but even without a bashrc it will not work. Firewall also doesn't seem to be a problem, since even when running in DMZ the problem accures unchanged. normal ssh to github test server works ...	15:32
	... fine, pulling also. Just the push fails	15:33
	any ideas?	15:33
	→ MattDeKrey joined	15:33
	← disappearedng_ left	15:33
	← disappearedng left	15:33
ThunderChicken	Check your .profile in addition to your .bashrc	15:34
mum-n-dad	How come I can "git clone git://github.com/railsdog/spree-active-shipping.git" but not "git pull" an existing clone of the repo? (fatal: The remote end hung up unexpectedly)	15:34
wereHamster	crab: yes. I read it only quickly but it looks ok	15:34
	mum-n-dad: check the url (git config remote.origin.url or git remote show origin)	15:35
Nihathrael	ThunderChicken: i don't have a .profile	15:35
	← ixti left	15:35
mum-n-dad	wereHamster: Wrong URL!	15:36
	Thanks man!	15:36
	→ lanthan joined	15:38
crab	wereHamster: thanks.	15:39
mum-n-dad	im out guys, thanks again!	15:39
	← mum-n-dad left	15:39
	→ ajpiano joined	15:41
	← mefesto left	15:41
	← bentob0x left	15:42
	→ ph^ joined	15:45
	← ajpiano left	15:45
	→ ph^_ joined	15:47
	← auslander left	15:47
	→ ajpiano joined	15:48
	→ auslander joined	15:48
	← auslander left	15:48
	→ auslander joined	15:48
	→ ProperN[out] joined	15:50
	← petrux left	15:50
	→ disappearedng_ joined	15:50
	→ disappearedng joined	15:50
	← ph^ left	15:50
	→ raichoo joined	15:50
	← ProperNoun left	15:53
	→ florianb joined	15:55
	← spearce left	15:56
	→ thiago_home joined	15:56
	← Pupeno left	15:57
	← lanthan left	15:57
	← ph^_ left	15:57
	→ friskd joined	15:58
	→ jstemmer joined	15:58
	→ ph^ joined	15:58
	← petercoulton left	15:58
	← loincloth left	15:58
	→ sh1mmer joined	15:59
stepnem	shruggar: too bad there doesn't seem to be a way to have a global .gitattributes file like .gitconfig	15:59
wereHamster	there is	15:59
stepnem	\o/	16:00
cYmen	If I want to remote push to a non-bare repository. How do I set it up?	16:00
stepnem	wereHamster: pleas do tell!	16:00
shruggar	info/attributes, I assume	16:00
stepnem	please* (sorry, I'm so excited ;))	16:00
wereHamster	oh, .gitattributes. I misread that as .gitignore -.-	16:00
stepnem	ah :\|	16:00
	shruggar: that's repo-local	16:00
	→ setuid joined	16:02
	→ alley_cat joined	16:02
shruggar	really I think those diff-options don't belong in gitattributes, really should be part of gitconfig	16:02
setuid	If a user is running "git clone ssh://user@remote/git/repo", and the 'git' binary is not on the standard path on the remote, is there some way to get that working? IOW, is it parsing the environment of "user" on the remote first?	16:03
	Can I set up $HOME/.bash_profile with some PATHs that point to the proper 'git' binary location?	16:03
	← sh1mmer left	16:03
	← kpreid_ left	16:03
wereHamster	~/.bashrc	16:04
setuid	So it does parse remote-user's profile?	16:04
wereHamster	where 'it' is the user's shell	16:05
setuid	Right	16:05
	→ ajpi joined	16:06
	← jonalv left	16:07
	← ajpiano left	16:07
cYmen	Damn. Google throws a ton of howtos at me that explain how to make repositories available with apache but not a single one that explains how to track websites served with apache.	16:07
setuid	cYmen, netcraft?	16:07
ekontsevoy	I have a bare repository which we access via git-shell, and files in that repo somehow get chmod 444 (read-only) and git-push starts to fail. Do you guys have an idea where to look?	16:08
cYmen	setuid: huh?	16:08
	← tvw left	16:08
knittl	ekontsevoy: un-chmod it	16:08
cYmen	I could just push to a non-bare repository and put "git reset --hard" in a post receive hook. Would that work?	16:09
bremner	ekontsevoy: did you init --shared? see man git-init	16:09
Gitbot	ekontsevoy: the git-init manpage can be found at http://git.or.cz/man/git-init	16:09
	← jboom left	16:09
	→ lessthanzero joined	16:10
lessthanzero	I've been googling for an hour, and I can't find any help for my question, which is...	16:10
knittl	lessthanzero: 42 is more than zero ;)	16:11
lessthanzero	I want to push to my gitosis based bare git repo - I can clone from it just find but when I push back, I get error "unable to create thread: Resource temporarily unavailable"	16:11
	← ponto_ left	16:11
lessthanzero	knittl: how many roads must a man walk down..	16:11
knittl	7	16:12
lessthanzero	ERR - 42.	16:12
bremner	lessthanzero: sounds like a github specific issue	16:12
lessthanzero	bremner: *gitosis ?	16:12
	← setuid left	16:12
bremner	oh, gitosis. sorry.	16:12
lessthanzero	(you mean, gitosis. I'm not using github)	16:12
	;) hrrm.. ok.	16:12
	I can't find much of ANYTHING.	16:13
bremner	everybody has switched to gitolite	16:13
	← tg left	16:13
bremner	can you ssh onto that machine?	16:13
	→ tg joined	16:13
	→ Theravadan joined	16:14
	← tg left	16:14
	→ lordi joined	16:16
lordi	http://www.managerleague.com/index.pl?ref=284582 - The Best Online Football Manager, Sign Up Please!:)	16:17
	→ mw joined	16:17
	← lordi left	16:17
	→ binjured joined	16:17
	→ ProperNoun joined	16:17
	→ lanthan joined	16:17
	→ tg joined	16:18
jay-mccarthy	how do i do something like: svn export -r rev repo/path?	16:18
ekontsevoy	bremner: yes, it is --shared and I determined that they become read-only after I run git-gc	16:18
	→ jboom joined	16:19
	← ProperN[out] left	16:20
bremner	ekontsevoy: what user do you run git-gc as? are they in the right group?	16:20
cYmen	If I git clone/pull what determines the access rights of the file? My umask?	16:20
thiago_home	yes, or your group-sharing settings	16:20
	cYmen: see man git-config	16:21
Gitbot	cYmen: the git-config manpage can be found at http://git.or.cz/man/git-config	16:21
ekontsevoy	bremner: yes	16:21
bremner	ekontsevoy: hmm, no more ideas from me	16:21
	→ sh1mmer joined	16:21
ekontsevoy	bremner: yeah, I have another repo sitting side by side, where this isn't happening. I'll just examine both very closely and hopefully will find the difference! :)	16:22
	→ gabeodess joined	16:22
	← [Po]lentino_ left	16:22
	→ lhz joined	16:23
ekontsevoy	By the way, for future reference, what's the best way to take a repo from a person who's been working on it for a while and make a bare shared clone of it?	16:23
gabeodess	wereHamster: Thank you for your guidance yesterday, it worked just as you said	16:23
	you saved my ass.	16:23
	→ vu3rdd joined	16:23
	← rolfb left	16:23
cYmen	thiago_home: so I simply add core.sharedRepository = true to the bare repository?	16:23
	or to the checkout of that user?	16:23
	← jeevan_ullas left	16:24
thiago_home	cYmen: after you clone	16:24
	→ [Po]lentino_ joined	16:24
	→ Belna_ joined	16:24
	→ defectiv joined	16:29
	→ BugeyeD joined	16:30
	→ ixti joined	16:30
	← trivol left	16:32
BugeyeD	hi all. newbie-ish question. i cloned a repository, created a branch, committed changes to the branch, then cloned the branch to another box. on the second box i committed changes, then used 'git push' to send the changes to the first clone. the command completed successfully, but i don't see the changes on the first box. what am i missing?	16:32
esr	Is there a git-format-patch incantation I can use to make a patch from the diff between HEAD and the working state? If so, what is it?	16:32
Nihathrael	BugeyeD: git checkout possibly	16:32
BugeyeD	Nihathrael: so the push updated the .git but not the working files?	16:33
Nihathrael	yep i think that's the case	16:33
	not absolutely sure though, haven't done that in some time	16:33
BugeyeD	Nihathrael: that's so simple, i hope it's correct. thanks. :)	16:33
Nihathrael	esr: git diff > file.patch or is it something else you are looking for?	16:34
	← khaase left	16:35
esr	Nihathrael: I'm looking specifically for a git-format-path invocation; I want to email the result.	16:36
	s/path/patch/	16:36
shruggar	is there any way to get "git status" to note that a file was copied?	16:37
	← galderz left	16:37
knittl	shruggar: it usually automatically does	16:37
	maybe try -C and -M	16:37
shruggar	I've seen it detect moves, but not copies	16:37
cYmen	hm..I'm trying to cd to a dir and run git pull in a post-receive hook but it somehow doesn't work can I not change directories in a hook or something?	16:38
shruggar	if it does at all, it should here. These are just literal "cp foo bar", as I'm going to remove foo later, but need it for now during migration	16:38
knittl	shruggar: it should notice it	16:38
	shruggar: but there's no difference if a file is copied or not	16:39
BugeyeD	Nihathrael: hmmm, git reset --hard <commit> seemed to be needed. does this sound wrong?	16:39
	← chittoor left	16:39
knittl	shruggar: AND finally, you have to add the new file, for git status to recognize it as copied	16:39
crab	esr: git-format-patch can't do that.	16:39
Nihathrael	BugeyeD: yep, that doesn't sound right	16:40
BugeyeD	grr	16:40
	← level09 left	16:40
shruggar	yeah, all been added, and I am using git-svn, so I want them to be written to svn as copies	16:41
	→ jasonmog joined	16:41
jasonmog	how do i reset a deleted file?	16:41
	← mastro left	16:42
cYmen	thiago_home: somehow it doesn't work. is this line in .git/config right: sharedRepository = true	16:42
thiago_home	cYmen: which section is it in?	16:42
jasonmog	i have not committed yet	16:42
knittl	jasonmog: git checkout -- file	16:43
jasonmog	werd	16:43
knittl	shruggar: puh, i don't know if git-svn will handle copies correctly	16:43
jasonmog	zomg u rawk	16:43
	→ kpreid_ joined	16:43
Nihathrael	BugeyeD: did you do "git checkout <branchname>?	16:44
zomg	jasonmog: I know but thanks	16:44
BugeyeD	Nihathrael: i tried that first, but was already in that branch.	16:44
cYmen	core	16:44
	thiago_home: core	16:44
	→ ereslibre joined	16:45
	← jasonmog left	16:45
Nihathrael	BugeyeD: any reason you can not pull instead of push?	16:45
thiago_home	cYmen: that should be how...	16:45
	→ aresnick joined	16:46
cYmen	hm..whenever I fetch the rights are always rw-------	16:46
BugeyeD	Nihathrael: wouldn't the pull come from the original clone source?	16:46
	→ dreiss joined	16:47
BugeyeD	Nihathrael: honestly, i'd rather push - so i know the "source" has been updated with my local work	16:47
Nihathrael	you said you cloned from ORIGIN, now you have pc1 and pc2 right? now you cloned pc1 to pc2, made some work on a branch. Now the right way to make pc1 have your changes is by using pull on pc1 i think	16:47
	usually if you use a central repo you want to push into, you would use a bare repo	16:48
	without a checkout	16:48
	→ virhilo joined	16:48
BugeyeD	Nihathrael: i made changes on pc2. was hopiong i could push to pc1, then later from pc1 push to ORIGIN. am i barking up the wrong tree?	16:48
virhilo	git reset --hard b05da413030a26de51430d429cae072389ac7275 and now how to apply it on remote repo?	16:48
knittl	BugeyeD: you should only push to bare repositories	16:49
	i assume pc1 also has a working tree/checkout	16:49
	you will only get problems with pushing that way	16:49
BugeyeD	okay, i see. i just need to go back and read up on the underlying concepts. thanks all.	16:49
knittl	yw	16:51
Nihathrael	no problem	16:51
	→ codetroll joined	16:51
	← afrasi left	16:52
	← virhilo left	16:53
	← kpreid_ left	16:53
	← florianb left	16:54
cYmen	How are hooks executed? I have this simple script: http://paste.pocoo.org/show/204566/ it works fine if executed in a terminal but fails as post-receive hook.	16:55
	→ riz2 joined	16:55
shruggar	cYmen: most common problem is with GIT_DIR being set	16:56
cYmen	shruggar: what dat?	16:56
	→ jakubzalas joined	16:56
shruggar	an environment variable	16:56
cYmen	Yes, I get that but what does it do, how is it used, etc etc :)	16:56
	→ slonopotamus joined	16:57
shruggar	it tells git where to look for .git	16:57
	I assume you're trying to do an "update my website when I push" script	16:57
cYmen	Most definitely, yes. :)	16:57
	I tried googling for one but only found howtos on setting up web access to repositories...	16:58
shruggar	unset GIT_DIR, and it will probably work as expected. Though for the love of dog, don't use "git pull", use "git fetch" followed by "git reset --hard"	16:58
	"git pull" is just asking for trouble, as it will attempt to merge when it feels it needs to	16:59
cYmen	um..oookay	16:59
	← sh1mmer left	16:59
	→ kpreid_ joined	16:59
	← wolog left	16:59
	→ khaase joined	17:00
shruggar	that is "git fetch your-remote-name", followed by "git reset --hard refs/remotes/your-remote-name/your-branch-name"	17:00
cYmen	I was asking earlier but nobody responded: Would it be ok to just push to the repository in question and do a "git reset --hard" in the post-receive?	17:00
crab	howto	17:00
	urk	17:00
shruggar	cYmen: not exactly, but yes	17:00
	→ krawek joined	17:01
wereHamster	cYmen: google 'git website howto'	17:01
	→ saccade joined	17:01
	→ v0n joined	17:01
cYmen	wereHamster: :)	17:03
	← slonopotamus left	17:04
	← shruggar left	17:04
	← pklingem left	17:05
	← flazz left	17:06
cYmen	hm..I tried git fetch and git reset --hard without the names but it didn't work. What exactly is the difference here when I supply remote-name and branch-name?	17:06
	← Zephyrus left	17:07
	→ ShooterMG joined	17:07
Ilari	cYmen: 'git reset --hard' is the same as 'git reset --hard HEAD'.	17:07
cYmen	but shouldn't that work after a fetch?	17:08
	→ flazz joined	17:08
Ilari	cYmen: Fetch does not change HEAD	17:09
	← flazz left	17:09
	→ flazz joined	17:09
	← bluenovember left	17:09
	← marshall left	17:09
	← flazz left	17:09
	→ flazz joined	17:09
	→ bentob0x joined	17:09
	→ WALoeIII joined	17:10
	← mw left	17:10
	← eletuchy left	17:11
cYmen	sigh	17:11
	→ brizly joined	17:13
jay-mccarthy	how do i clone a particular commit?	17:13
cYmen	I need to read something. I have used git a little but haven't understood the underlying concepts. Any suggestions?	17:13
jay-mccarthy	i find http://eagain.net/articles/git-for-computer-scientists/ useful	17:13
rvsjoen	http://git-scm.com/documentation	17:13
crab	cymen: http://www.newartisans.com/2008/04/git-from-the-bottom-up.html	17:14
	← vu3rdd left	17:14
	← christophsturm left	17:14
cYmen	Thanks for the suggestions.	17:14
	→ christophsturm joined	17:14
	fcoury → fcoury\|away	17:16
	→ sh1mmer joined	17:17
	← curtana left	17:18
	→ xperience joined	17:18
	← christophsturm left	17:19
	→ petercoulton joined	17:19
	→ LRN joined	17:22
	← sh1mmer left	17:23
	→ kipras joined	17:24
	← steffkes left	17:25
	← camonz left	17:26
	→ AAABeef joined	17:27
	→ jaysern joined	17:28
	→ sh1mmer joined	17:28
	→ aep joined	17:28
	→ airborn joined	17:29
aep	can i push some changes from a repo, to a completely other repo, which does not share the same commit base, but the same code?	17:29
	← xperience left	17:29
	→ wolog joined	17:29
wereHamster	sure	17:30
	← ixti left	17:31
aep	well, got a hint for me how?	17:31
	← masak left	17:32
wereHamster	git push <theotherrepourl> <whateverbranchyouwanttopushthere>	17:32
	→ ixti joined	17:33
	← LRN left	17:33
aep	uh well that fails with ! [rejected] master -> master (non-fast-forward)	17:33
	→ LRN joined	17:33
aep	which i guess is because its not the same base	17:34
wereHamster	aep: faq non-ff	17:34
Gitbot	aep: Your push would lose changes on the remote. See https://git.wiki.kernel.org/index.php/GitFaq#non-ff	17:34
aep	yeah, i know, since its not the same base >_<	17:34
	← tatsuya left	17:34
wereHamster	so push the branch to a different ref on the other side	17:35
	← kpreid_ left	17:35
aep	there is no ref that has the same base	17:35
wereHamster	git does't care	17:35
aep	they're untirely unrelated repos	17:35
	hm	17:35
	→ [Po]lentino__ joined	17:35
wereHamster	git push <theotherrepourl> master:anothermaster	17:35
aep	maybe i'm doing that test wrong then. i have two repos a and b. in a there is a file a and in b there is a file b. each commited. now i want to push the change that added b to b, to the repo a	17:36
	← [Po]lentino_ left	17:36
aep	err let me pastebin that.	17:36
wereHamster	cd a; git push ../b master:masterfroma; cd ../b; git merge masterfroma	17:36
	→ spearce joined	17:37
aep	interesting	17:38
	now it makes sense, thanks alot wereHamster	17:39
	→ aheinecke joined	17:39
	→ dmg joined	17:39
	→ sixteneighty joined	17:40
	→ webchick joined	17:41
cxreg	hrm	17:42
	commands that take a pathspec, you can't negate that typically, right?	17:42
	like you can't say log for all directories but this one	17:42
	← Alagar left	17:43
	← nevyn^ left	17:44
	→ eletuchy joined	17:45
	← LRN left	17:46
	→ pkahle joined	17:46
cxreg	hrm: https://kerneltrap.org/mailarchive/git/2008/10/17/3701944/thread	17:47
DrNick	POSIX globs don't have a negation, so, no	17:47
cxreg	shrug	17:48
	you could say --invert-path or something	17:48
	are you certain that git paths are even posix globs?	17:48
	→ LRN joined	17:48
DrNick	either glob() or fnmatch(), iirc	17:48
	→ Axius joined	17:49
	→ psoo joined	17:51
	← jrmuizel left	17:52
	← magn3ts left	17:53
	→ pklingem joined	17:54
	→ jrmuizel joined	17:54
	→ tremby joined	17:56
tremby	what's the easiest way to reverse a commit (which has gone to the remote already, it was faulty)? there have been other commits afterwards which should stay.	17:56
tango_	tremby: git revert	17:56
doener	git revert $commit	17:56
tremby	that does just the one?	17:57
	oh nice. i always assumed that went back to that point in time, ie the later commits disappeared too	17:57
cxreg	no, that's reset	17:57
	→ Shyde joined	17:57
tremby	of course. that is indeed what i was thinking of.	17:57
	← flazz left	17:58
	→ morelli joined	17:58
	→ flazz joined	17:59
	ajpi → ajpiano	17:59
	← flazz left	17:59
	→ flazz joined	17:59
	→ marshall joined	17:59
	→ warlock_mza joined	18:00
	← zoniq left	18:02
	→ simplecoder joined	18:02
	← peritus- left	18:03
	← Axius left	18:03
	← jonshea left	18:03
	→ macek joined	18:03
	→ mefesto joined	18:03
	→ kpreid_ joined	18:04
	← simplecoder left	18:05
	← disappearedng_ left	18:06
	← disappearedng left	18:06
	→ jonshea joined	18:06
	→ gnufied joined	18:07
	← GarethAdams\|Work left	18:07
	← rseifert left	18:08
	→ Heimidal joined	18:08
	← Heimidal left	18:08
	→ Heimidal joined	18:08
	← bx2 left	18:09
	→ fisch_ joined	18:10
	→ enjo451 joined	18:12
	← marshall left	18:12
enjo451	What is the proper way to create a remote branch at a specific tag?	18:12
morelli	hi is anybody up and familiar with gitolite?	18:12
	→ camonz joined	18:13
	→ tildeequals joined	18:14
	← Belna_ left	18:14
morelli	enjo451 what do you mean by "proper", one way git checkout -b <br_name> tag; git push origin br_name	18:15
	→ adiabatic joined	18:15
	→ bx2 joined	18:15
	← bx2 left	18:15
	← smuf left	18:16
bremner	morelli: hard to know what "familar" means to you. Try asking...	18:16
enjo451	thanks morelli:)	18:17
morelli	Use of uninitialized value $ENV{"GL_RC"} in do "file" at hooks/update line 36.	18:17
	→ jceb joined	18:17
morelli	when pushing	18:17
spearce	enjo451: another way is git push origin tag^0:refs/heads/br_name	18:17
	the ^0 and refs/heads/ bits are really important	18:18
bremner	morelli: most gitolite issues are ssh issues in disguise. What happens if you run "ssh git@gitolite-host"	18:18
morelli	i logon	18:19
bremner	that is not supposed to happen :)	18:19
adiabatic	Newbie question. A buddy of mine forked my repo on github and made some changes over multiple pushes. What would be the best way to get a diff of what he's done? git-pull into another directory on disk and then git diff, or...?	18:19
bremner	morelli: did you try the SSH troubleshooting hints in the gitolite docs?	18:19
	→ rseifert joined	18:20
	← rseifert left	18:20
	← tremby left	18:20
wereHamster	adiabatic: why (a diff)? Why not just merge his changes?	18:20
morelli	but i could clone the repo so why problem with push ?	18:20
	i will look now	18:20
	→ elmob joined	18:21
	→ kar8nga joined	18:21
	→ disappearedng joined	18:21
	→ disappearedng_ joined	18:21
	← webchick left	18:21
adiabatic	wereHamster: well, I'd like to look at what he changed and try and figure out what he did before his stuff goes into my repo	18:21
	but that shouldn't be much of a problem as long as I don't commit anything, right?	18:22
bremner	morelli: I suspect your clone bypassed gitolite. Did you give "repositories" as part of the path?	18:22
wereHamster	adiabatic: fetch his changes, then inspect them with git log -p for example	18:23
enjo451	Is there a nice way to figure out the commit at which a particular file was deleted?	18:23
morelli	yes git@gitserver:repositories/testing.git	18:23
bremner	yeah, that is bypassing gitolite. You are using the admin key I guess	18:24
	→ webchick joined	18:24
wereHamster	enjo451: git log -- <file which was deleted>	18:25
morelli	hmm frankly speaking I really don't know witch one ;)	18:25
Ilari	morelli: Try 'ssh gitolite'.	18:25
	→ brez- joined	18:26
	→ segher joined	18:26
Ilari	morelli: IIRC, the hostname really was "gitolite".	18:26
	→ aspotashev\|eeepc joined	18:26
morelli	gitolite said hello	18:27
	and gave me my permisions	18:27
Ilari	morelli: 'git clone gitolite:<repository>'.	18:27
	→ tgunr joined	18:27
Ilari	morelli: E.g. 'git clone gitolite:testing'.	18:27
	→ corni joined	18:28
morelli	that way have cloned the admin repo but wanted to test out way for the user	18:28
Ilari	morelli: Gitolite identifies the user by SSH key.	18:29
	← enjo451 left	18:29
	→ meanburrito920 joined	18:29
	→ JayM joined	18:29
Ilari	morelli: So to change what user it identifies as, you need to change what key SSH sends.	18:29
	← d0ugal left	18:30
morelli	in .ssh/config?	18:30
	← Nihathrael left	18:31
Ilari	morelli: Easiest way is to create new alias in .ssh/config	18:31
	← ph^ left	18:31
	→ tvw joined	18:31
	→ Alagar joined	18:31
	→ ph^ joined	18:31
	← khaase left	18:33
adiabatic	wereHamster: thanks	18:34
	← eletuchy left	18:35
tgunr	I have a repo here with an anomomly i can't figure out, one of the folders does not show up if i use `git ls-files' but if I cd to the folder and use `git status .' the status indicates it is up to date. Why doesn;t it show up in the ls-files. Also, I went to an empty folder and cloned the repo and the folder is missing in the clone. The folder is not being ignored and a `git add missingfolder' doesn't do anything.	18:35
	→ eletuchy joined	18:35
tgunr	er anomaly	18:36
	→ d0ugal joined	18:36
	← kar8nga left	18:36
	→ justin-george joined	18:37
tgunr	the folder in question does have a wierd path name: /Volumes/leopard/Projects/XMacVPL/( Xcode Projects )/( xcbuild_ffi )	18:37
	← d0ugal left	18:37
davr	You're on a mac, so I can forgive spaces in a path, but parens?	18:38
adiabatic	Mac or no, "(blah blah blah)" is an odd path name	18:38
tgunr	don't ask, i can't control that part	18:39
	→ khmarbaise joined	18:39
engla	there shouldn't be any restriction on pathnames	18:39
tgunr	at least yet, after I get it into VCS with git I can change it	18:39
	→ urkud joined	18:39
engla	it's just a string of bytes to a unix tool, right?	18:39
	← airborn left	18:39
tgunr	there is another similar folder which did get added ok :/Volumes/leopard/Projects/XMacVPL/( Xcode Projects )/( xcbuild )	18:40
davr	You'd think...but I've seen apps that break even with just spaces in the paths	18:40
tgunr	eng: correct if escaped properly it should not matter	18:40
	← Shyde left	18:41
adiabatic	Apache won't build properly if it's in ~/Desktop/New Stuff, but it will in ~/Desktop/Incoming.	18:41
tgunr	question right now is why is the folder not in the repo and how do I get it in?	18:41
	→ rseifert joined	18:41
davr	yeah, linux people have trouble with spaces sometimes. I see it occasionally with linux apps ported to windows as well. Windows apps generally always work with spaces, since both $HOME and default program install dirs have spaces in them. Maybe MS did that on purpose to force devs to make it work	18:42
Ilari	tgunr: 'git status .' is what 'git commit .' would do.	18:42
DrNick	is it empty?	18:42
	← stamina left	18:42
Ilari	tgunr: And yes, you can't have tracked empty directories.	18:42
tgunr	`git status .` shows on branch master, nothing to commit	18:43
	not empty	18:43
	← froschi left	18:43
	→ khaase joined	18:43
	→ SimManiac joined	18:44
SimManiac	hi	18:45
	→ Tali joined	18:45
morelli	Ilari now server asks me for the password and i should know what is wrong, but i don't	18:46
	← Tali left	18:46
Ilari	morelli: You haven't added that key to keydir (and committed & pushed)?	18:46
morelli	it is added & pushed but , on push there was a warning that the name is not in config but with testing repo and @all i ignored it	18:47
Ilari	morelli: login to gitolite account and check if the key is in .ssh/authorized_keys?	18:48
tgunr	found the problem, tried to rename the folder with git and discovered it contains folders which are empty, anomaly explained :)	18:49
	← rseifert left	18:49
	spearce → git\|spearce	18:49
morelli	yes it is	18:50
	→ andreaa joined	18:50
	→ airborn joined	18:51
	→ Gitzilla joined	18:51
	→ rseifert joined	18:52
	→ schacon joined	18:53
	← dmg left	18:53
morelli	i'm an idiot, the name in ssh/config had ".pub" at the end ;)	18:53
	→ dmg joined	18:53
	mithro → thousandparsec\|m	18:54
	← kpreid_ left	18:54
morelli	but the same error when pushing Use of uninitialized value $ENV{"GL_RC"} in do "file" at hooks/update line 36.	18:55
	← hyperair left	18:55
	→ frakturfreak_ joined	18:55
	← webchick left	18:55
	→ arian1 joined	18:58
	← frakturfreak left	18:59
	← arian1 left	19:00
	→ slonopotamus joined	19:01
	→ SqueeDee joined	19:02
	→ d0ugal joined	19:03
	→ sgh joined	19:03
	← lessthanzero left	19:03
	→ kpreid_ joined	19:03
jay-mccarthy	how do i get the commit messages for commits between X and Y, not including X?	19:04
	← d0ugal left	19:05
wereHamster	just the commit message?	19:05
jay-mccarthy	i'm really curious about the range syntax	19:05
	"X Y" includes X	19:05
	← khaase left	19:05
wereHamster	X Y includes everything reachable from X or Y	19:05
jay-mccarthy	so i don't understand then	19:06
	→ TheUni joined	19:06
SqueeDee	Hey all. We use github as our 'origin' or forced centralised repository, and nearly any time i make a feature branch i want it on there immediately so that I can tell our CI tool to track, build and test the feature branch. Is there something more elegant than: "git branch <branch name>; git checkout <branchname>; git push origin <branchname>; git checkout <mainline>; git branch -D <branchname>; git checkout -t origin/<branchname>" ?	19:06
wereHamster	don't understand what?	19:06
jay-mccarthy	how do use git show to do what i wan	19:06
wereHamster	jay-mccarthy: git show X; git show Y;	19:07
jay-mccarthy	that's not right either...	19:07
	I know that commit X happened, and there may have been A, B, C, etc, then I know that Z happened	19:08
	→ lll joined	19:08
	→ peritus- joined	19:08
jay-mccarthy	is there a way to get the messages for A, B, C, and Z, only knowing X and Z?	19:08
SqueeDee	isnt that what git log does for you?	19:08
wereHamster	how are X, A, B, C and Z related?	19:09
SqueeDee	< total newb so tell me to shut up if i make things worse.	19:09
jay-mccarthy	X was the head at the last push, Z is the head at the current push	19:09
	A B and C were part of the push that Z was the head of	19:09
	basically i want to find all the commit messages that were part of a push	19:09
wereHamster	git log X..Z ?	19:09
	← corni left	19:10
	← khelll left	19:10
	→ rolfb joined	19:10
jay-mccarthy	that looks right	19:11
wereHamster	SqueeDee: git --version	19:11
SqueeDee	you can use relative magic too cant you? like +1 etc?	19:11
wereHamster	+1 where?	19:11
SqueeDee	sec, I'll research before i make bold claims	19:11
	← pkahle left	19:11
jay-mccarthy	is there an option for more parseable output?	19:12
	← elmob left	19:12
wereHamster	jay-mccarthy: man git-log --format=...	19:12
Gitbot	jay-mccarthy: the git-log manpage can be found at http://git.or.cz/man/git-log	19:12
SqueeDee	oh sorry it was ^ as an exclusion, thats cool.	19:12
	→ Textmode joined	19:12
SqueeDee	done by x..z anyway	19:12
	← sylr left	19:13
Textmode	Morning all	19:13
	→ maddog_ joined	19:13
maddog_	hiho	19:13
SqueeDee	Morning	19:13
wereHamster	SqueeDee: your git version pls	19:14
SqueeDee	a moment	19:14
	1.6.3.3	19:14
maddog_	I have currently a state on a branch on my workstation which is same sommits behind the status on the server. Is it possible to force my version to be the newest and pushing that to the server?	19:14
wereHamster	SqueeDee: too bad for you, 1.7 has push --set-upstream, which would shorten your commands to: git checkout -b <branchname>; git push --set-upstream origin <branchname>	19:15
	→ cbrinke1 joined	19:15
SqueeDee	im happy to update	19:15
cbrinke1	I have accidentally merged two repos together, what is the easiest way to remove one of the repos?	19:16
wereHamster	maddog_: so you want to effectively throw away the additional commits on the server?	19:16
	→ fgiraldeau joined	19:16
wereHamster	cbrinke1: how did you merge them?	19:16
	← camonz left	19:16
	← mefesto left	19:17
	→ mefesto joined	19:17
cbrinke1	history: I had a .git repo for my .dotfiles, I also had a ~/projects/.git repo as well. Apparently somehow the projects repo imported all of the dot files	19:17
SqueeDee	wereHamster that looks sooo much better. and now i'm aware of at least one cause of confusion in the office.	19:17
morelli	Ilari here http://gist.github.com/374278 how it looks	19:17
	← aspotashev\|eeepc left	19:17
SqueeDee	nearly everyone is looking at the kernel.org docs and I'm sure no-one is using 1.7	19:17
	→ aspotashev\|eeepc joined	19:17
wereHamster	cbrinke1: git rm --cached .dotfiles; git commit ?	19:18
Ilari	morelli: You are using wrong type of key (admin key, not gitolite user key).	19:18
cbrinke1	The strange bit is that it also imported the whole history from the other repo.	19:19
	→ rgr joined	19:19
cbrinke1	wereHamster: will that eliminate the dotfiles history from the repo?	19:19
wereHamster	cbrinke1: where is the .git for the dotfiles? Inside the dotfiles directory?	19:19
cbrinke1	~/.git = dotfiles repo ; ~/project/.git = project repo	19:20
wereHamster	and the actual dotfiles are where?	19:20
cbrinke1	~/{.vimrc,.bashrc} (etc)	19:21
	I was unaware of the dangers of having a nested git repo	19:21
wereHamster	cbrinke1: in that setup, I would understand if you somehow managed to add the project repo as a submodule of the dotfiles repo..	19:21
	← pklingem left	19:22
wereHamster	eg. cd ~/; git add project; git commit;	19:22
cbrinke1	thats the strange bit, the project repo has ended up with the dotfiles history.	19:22
	I am not sure how this happened however, and the log is essentially worthless as it shows both repos worth of history	19:23
wereHamster	and where in the tree of the project repo are the dotfiles?	19:23
cbrinke1	~/project/{.bashrc,.vimrc}	19:23
morelli	Ilari not here http://gist.github.com/374285 admin.pub =/= id_rsa.pub and both are in keydir in gitolite-admin repo	19:23
wereHamster	cbrinke1: gitk --all might help you understand what happened	19:24
cbrinke1	does gitk have a "sort by timeline" ?	19:24
adiabatic	I just tagged a release with `git tag -a 1.70`, but when I tried to push to github, it said that everything's up-to-date…and github doesn't have my tag. What do I need to do to get the tag published?	19:24
	adiabatic waves to Ilari	19:25
	→ pklingem joined	19:25
morelli	git push origin tag_name	19:25
	→ Plouj joined	19:25
Plouj	hi	19:25
	← pklingem left	19:25
Plouj	do I have to do something special when rebasing with submodules?	19:25
wereHamster	hi plouj :)	19:25
	← justin-george left	19:26
Plouj	for some reason git keeps saying a submodule dir is modified even after I do git checkout <submodule-dir> on it	19:26
	hi wereHamster	19:26
Ilari	morelli: id_rsa is presumably your admin key. You can't use that key as user key (create new keypair)?	19:26
adiabatic	morelli: thanks. Why wouldn't tags go out with the normal commits, though?	19:26
	→ pklingem joined	19:27
	→ simplecoder joined	19:28
	→ webus joined	19:28
morelli	no id_rsa is not the admin key	19:30
	→ erikvold joined	19:31
	→ icefox2 joined	19:31
morelli	adiabatic because that how git works, you can push all tgs by git push origin --tags or one ref	19:31
	← cbrinke1 left	19:31
morelli	like ref name tag or branch	19:32
	→ Nemurenai joined	19:32
	← jaysern left	19:32
	→ navetz joined	19:32
Plouj	ah, I forgot git submodules update, I think	19:33
navetz	hey guys I am new to git and need some help, I want to download the tooltip widget from jqueryui's github branch. I don't really know hwo to do this.	19:33
morelli	by default from what i know git push origin pushes only master btw	19:33
	← [Po]lentino__ left	19:33
navetz	is it some kind of submodule?	19:34
	icefox2 → icefox	19:34
Ilari	morelli: What key is the default key SSH sends then? You have .ssh/identity or .ssh/id_dsa?	19:36
	→ webchick joined	19:36
	→ hugojosefson_ joined	19:37
	→ joeconyers joined	19:37
morelli	Ilari from what i know default is id_rsa and i don't have identity or id_dsa	19:37
	→ agib joined	19:37
	→ ShaunR222 joined	19:38
morelli	maybe msysgit is doing something wrong ?	19:38
Ilari	morelli: Then id_rsa is key that has shell access to gitolite account. You can not use that key as gitolite user key.	19:38
	morelli: At least not over SSH.	19:38
ShaunR222	i'm very new to git and using a versioning system like this. I want to use it on a small code project i have. I already did a git init; git add .; git commit in my project dir. I'm curious how creating versions works or is suppose to work. Should i use tags for each version/release?	19:39
morelli	Ilari yes it has, and why not?	19:39
	hugojosefson_ → hugojosefson	19:39
Ilari	ShaunR222: Each release should probably be tagged. Revisions are created by commit command.	19:39
	morelli: Because that key bypasses gitolite.	19:40
	→ malte_ joined	19:40
morelli	aha	19:40
ShaunR222	ok, for example git version 1.7.0.5, thats a tag?	19:40
	← ereslibre left	19:40
	→ adaro_ joined	19:41
morelli	so new key for the tests i must generate :)	19:41
divVerent	Is there a nice way to keep unversioned stuff as part of a git repo so it gets pulled?	19:41
Ilari	ShaunR222: Yes. And there are 27 commits in 1.7.0.5 that aren't in 1.7.0.4...	19:41
divVerent	(namely: scripts to use the git repo)	19:41
	currently, I simply put that script versioned in the repo	19:41
	→ cbrinke1 joined	19:41
divVerent	but the problem is, when the script is updated, and people work in branches, they still have the old version	19:42
ShaunR222	when commiting it opens vi and shows files i modified/added and wants me to comment. Should i just comment at the top or should i be commenting about each modified/newfile?	19:42
	← webchick left	19:42
wereHamster	at the top	19:42
	← hugojosefson left	19:42
PerlJam	ShaunR222: it's your commit message, put what's useful to you.	19:42
divVerent	one idea for this might be having "two workdir checkouts in one dir", if git supports that	19:42
cbrinke1	wereHamster: ok, I found where I made the mistake, is there anyway to undo a merge operation?	19:42
wereHamster	ShaunR222: if you really thin you should comment on each file individually, you should make individual commits	19:42
divVerent	so the script always comes from master	19:42
Ilari	ShaunR222: One (short) subject line describing what was changed, blank line and then longer description giving why the change was made.	19:43
adaro_	is it possible to specify svn locations when using git svn init , i basically want to get a git repo containing two svn modules that i can sync with a single command	19:43
	→ ereslibre joined	19:43
	← ereslibre left	19:43
	→ ereslibre joined	19:43
wereHamster	cbrinke1: reset --hard <commitbeforethemerge>, which is HEAD@{1} if you just did the merge and nothing else	19:43
ShaunR222	ok i see, so after each mod i should commit and then comment about that modification.. rather than do a bunch of changes and then commit.	19:43
wereHamster	yes	19:43
divVerent	another idea for this would be "virus-like": on every startup, the script gets the most current version of itself and places it in .git/, then runs that :P	19:43
Ilari	ShaunR222: The rule is to make each logical change as its own commit.	19:43
PerlJam	ShaunR222: After each logically complete modification.	19:43
adaro_	*multiple svn locations	19:43
	missed a word there :)	19:44
	← d0k left	19:44
	→ ehabkost joined	19:44
	→ icicicici joined	19:45
	→ hugojosefson joined	19:45
icicicici	I've created a new repo (blank directory), then done an svn clone into the repo, followed by a git update-server-info, and I find that I can't clone my repo over http	19:46
cbrinke1	wereHamster: that did it, thanks for the pointers	19:46
ShaunR222	i have a main, we'll say stable version and then a beta version normally... With git i assume stable would be the master branch and beta would be a beta branch?	19:46
icicicici	I have other repos in the folder I can clone.. it appears that git is looking for "info" which in this repo is in .git/info	19:46
	hugojosefson → hugojk	19:46
icicicici	can anyone help me out?	19:46
	ShaunR222 → ShaunR	19:46
	hugojk → hugoj	19:46
Ilari	ShaunR: Yes.	19:46
	← ShaunR left	19:46
	→ ShaunR joined	19:46
	hugoj → hugoj_	19:46
	hugoj_ → hugojoseson	19:46
	hugojoseson → hugojosedson	19:46
	→ hyperair joined	19:47
sitaram	morelli: when you ran easy-install, did you get a message at the end that looked like this: http://pastebin.com/Xegp94jh	19:47
Ilari	icicicici: Repos to be cloned over HTTP should be bare. And why are you using HTTP anyway (especially dumb HTTP)?	19:47
ShaunR	ok, thanks for the info!	19:47
	hugojosedson → hugojosefson	19:47
icicicici	Ilari: What ought I use instead of http? I rather like http, but I'm open	19:47
	→ bx2 joined	19:48
	sitaram updates doc/6 yet again...	19:48
Ilari	icicicici: The primary protocol for anonymous cloning is git:// (and if you use HTTP, make sure it is capable of smart HTTP).	19:48
icicicici	Ilari: it's not anonymous I want. I do http with apache auth	19:48
morelli	Ilari yes exactly the same	19:48
Ilari	icicicici: Then ssh.	19:48
icicicici	ugg.. I'd rather not do ssh (for many reasons)	19:49
	← g0bl1n left	19:49
	← Nemurenai left	19:49
icicicici	Ilari: so I can't clone an svn repo into an empty git tree (techinically I did an init to svn)	19:49
	then share over http?	19:49
	→ pheaver joined	19:49
	→ onigiri_ joined	19:50
	← pheaver left	19:50
Ilari	icicicici: Create repo in location where whatever is serving it finds it and push it there.	19:50
	icicicici: And that repo will be bare.	19:50
morelli	sitaram yes partly exactly the same	19:51
icicicici	Ilari: so I'll create a repo on my "server"	19:51
	and that one's init-ed bare	19:51
	← peritus- left	19:51
icicicici	Ilari: then do a git fetch from svn elsewhere, and push to the "server"?	19:51
Ilari	icicicici: Why not ssh? And as said, make sure smart HTTP works if you serve it over HTTP.	19:52
	icicicici: And yes, you need bare repo on server that gets served to clients.	19:52
divVerent	21:48:45 icicicici \| Ilari: it's not anonymous I want. I do http with apache auth	19:52
icicicici	Ilari: yup on the smart http side. as for ssh, I don't relish the idea of account mgmt with ssh	19:52
divVerent	problem is - IF you want to use that, you need SSL too for security	19:53
	and then you are probably better off just using SSH instead	19:53
icicicici	divVerent: I'm not worried about the SSL side <yet>	19:53
divVerent	icicicici: use gitolite then	19:53
icicicici	but I will be	19:53
SqueeDee	wereHamster thankyou for the advice earlier.	19:53
divVerent	that goes through ssh, but uses ONE unix account for all	19:53
	and does not give them shell access	19:53
sitaram	morelli: so please do as it says. For you, the url must be "gitolite:reponame". Your users (whose keys you add into the config, and do not have a direct shell key) can use the type of URL you're using	19:53
icicicici	divVerent: ooh shiny	19:53
divVerent	think of sshd with forced commands as "just another inetd" ;)	19:53
	← cbrinke1 left	19:53
divVerent	icicicici: the big disadvantage of gitolite/ssh is that it ONLY allows public key login	19:54
	and NOT password login	19:54
icicicici	divVerent: ah, then I can't do that either	19:54
divVerent	but one can get used to that :P	19:54
	why not?	19:54
icicicici	it's okay for me, but not my user set	19:54
	← hugojosefson left	19:54
divVerent	make a script that creates a ssh key with a passphrase, silently sends the key to you, and they won't know the difference	19:54
	:P	19:54
SqueeDee	You can educate people on keys. I've got designers, operations, management, all using keys, across platforms.	19:55
	← warlock_mza left	19:55
icicicici	lolz.. yeah, I'd rather do the http thing for now	19:55
divVerent	SqueeDee: oh, the M-word.	19:55
SqueeDee	it was worth the effort.	19:55
divVerent	I see :P	19:55
sitaram	morelli: I just added a new section at the top of doc/6; please see "the most common problems that an admin will see" section in http://github.com/sitaramc/gitolite/blob/pu/doc/6-ssh-troubleshooting.mkd	19:55
icicicici	SqueeDee: yeah, I've got that + different pain	19:55
	it's awesome	19:55
divVerent	sitaram: BTW... maybe there SHOULD be a way to use gitolite with password ;)	19:55
	idea: one shared ssh key, and it prompts for username/password on network operation	19:56
	I wonder if gitolite would actually be ABLE to do such a prompt	19:56
	stderr isn't a problem, but accessing the terminal for input might be	19:56
sitaram	divVerent: sure, when Ilari's git-daemon2 gets in, and SRP access can happen :) [I think...]	19:56
Ilari	icicicici: Smart HTTP has serverside compat mode for old clients (fetch only, but you DO NOT want to push over dumb HTTP).	19:56
divVerent	sitaram: oh, an alternate way :P	19:56
	sure, why not	19:56
	so basically, git:// will learn auth?	19:57
	and then can be used for ssh + userpass too?	19:57
icicicici	Ilari: why no? I'm somewhat new to this.. and do @ home	19:57
Ilari	divVerent: Yes, git-daemon2 supports passwords via TLS-SRP.	19:57
divVerent	oh, already TLS?	19:57
	even better	19:57
	doesn't that mean it'll be able to REPLACE ssh?	19:57
	but I'd still want gitolite support in that ;)	19:57
sitaram	divVerent: Ilari is the expert on that (and there is a small side-branch of gitolite called "pu-ssh-plus" to cater to that, but changes are minor; all the action is on his side)	19:57
	→ stringo0 joined	19:57
sitaram	divVerent: yes he uses gitolite, that special branch	19:57
divVerent	but sure, sounds good	19:57
	← eletuchy left	19:58
	→ peritus- joined	19:58
divVerent	sitaram: but that still will require the ssh:// protocol, right?	19:58
sitaram	divVerent: I always like to say "this feature exists and currently has a grand total of 1 user" :) But honestly, Ilari can explain far better than I can; I'm sure to mess it up!	19:58
divVerent	well... currently the only reason why gitolite doesn't work with u/p auth, is that ssh only allows forced commands on pubkeys	19:58
	→ Tobarja_n810 joined	19:58
divVerent	but not on user accounts	19:58
	otherwise, one could create multiple names for the same UID in /etc/passwd, give them different passwords, and gitolite could figure out the rest ;)	19:59
	sitaram: also... ONCE sshd gets such a feature... it wouldn't need the smallest change to gitolite to support it ;)	20:00
sitaram	divVerent: you can put a ForceCommand in a Match block in /etc/ssh/sshd_config but (a) that is too intrusive (b) it completely precludes direct shell access; you can only du "su - git" after that for CLI, (c) still cannot distinguish me from you	20:00
wereHamster	but ssh allows forcedcommands for all acconuts	20:00
divVerent	sitaram: oh... great	20:00
	but... where is the problem then?	20:00
	I create 10 alternate usernames in /etc/password for the gitolite UID	20:01
morelli	sitaram i have added fresh key to keydir and "R access for repositories/testing DENIED to me" but i assumed that @all means all keys in keydir even when added name of the new key this is this MUST use thing?	20:01
divVerent	and then use a Match block to ForceCommand them the gl-auth-command command	20:01
	shouldn't that do?	20:01
	the trick is having multiple user accounts of the same UID	20:01
	→ maks_ joined	20:01
maks_	~/src/linux-2.6$ git name-rev 15121c18a22ae483279f76dc9e554334b800d0f7	20:01
divVerent	like BSD people commonly do with the username "toor" at UID 0	20:01
maks_	15121c18a22ae483279f76dc9e554334b800d0f7 undefined	20:01
	← simplecoder left	20:01
sitaram	divVerent: maybe it will; and if you try it and write it up it will go into contrib/, but I will not be doing anything that requires root :)	20:02
maks_	where shall I bug report this?	20:02
divVerent	sitaram: no, no	20:02
	not YOU would do that	20:02
	→ simplecoder joined	20:02
divVerent	this is nothing that can, or should, be automated	20:02
	only documented that it CAN be done (if it works)	20:02
adaro_	if you have a git repo that contains lots of folders is it possible to take one folder from that repo, fork it and still merge changes from the main repo back in	20:02
	← simplecoder left	20:02
morelli	sorry	20:02
	→ warlock_mza joined	20:02
	→ hugojosefson joined	20:02
morelli	it now works as expected	20:03
sitaram	divVerent: excellent idea... sadly it is past 01:30 here and I am probably too sleepy to try it :) I'll try it tomorrow or if you try it let me know	20:03
	← maks_ left	20:03
sitaram	morelli: it always works as expected! If it doesn't seem to, change expectations ;-)	20:03
	← hugojosefson left	20:03
divVerent	sitaram: I'll try it right now	20:03
	I am not yet convinced that it actually will work	20:04
	as the Match block might instead match against the first username for the UID :P	20:04
	agib → agib\|away	20:04
divVerent	of course, I really wouldn't recommend such a setup	20:04
Ilari	divVerent: Yeah, the only reason it wouldn't work would be sshd.	20:04
divVerent	as it will give a security hole quite fast	20:04
	one would have to ALWAYS make sure to FIRST add the account to sshd_config, restart sshd, and only then actually create the user	20:05
sitaram	divVerent: you need one matchblock for each user though	20:05
divVerent	and opposite order for deleting	20:05
	exactly	20:05
	and THAT is a problem :P	20:05
sitaram	big one!	20:05
divVerent	once you forget the order, evil user may gain shell access	20:05
	← icicicici left	20:05
divVerent	actually... I think you may be able to do it with just one match block	20:05
	and a wildcard pattern	20:05
	if there is a way to get the login user name from the environment, that could be used in the gl-auth-command call	20:05
	checking that too :P	20:05
	one would name the users git-foo then :P	20:06
sitaram	divVerent: yeah that can be done; one line change in gitolite if needed	20:06
divVerent	I prefer not having explicit support for that hack in gitolite ;)	20:06
	← altrux left	20:06
divVerent	if possible, that is	20:06
sitaram	divVerent: agreed; I was only saying if you want to test it you can do that	20:06
	→ altrux joined	20:06
divVerent	that really shouldn't be the "usual setup" of gitolite	20:06
sitaram	divVerent: agreed	20:06
	→ khelll joined	20:06
divVerent	mainly because repeatedly asking the PW is annoying	20:06
	and pubkeys are SO MUCH better :P	20:07
	so even if it works and gets documented	20:07
	maybe the doc for it should NOT be with the usual gitolite doc	20:07
sitaram	frankly, I've got hardcore Windows folks using git happily, pubkeys and all...	20:07
divVerent	but also in a contrib dir :P	20:07
morelli	sitaram yes that is true but this whole _ssh_ is confusing sometimes, and I don't know what to expect ;)	20:07
	← petercoulton left	20:07
	← tg left	20:07
sitaram	01:32:07 <sitaram > divVerent: maybe it will; and if you try it and write it up it will go into contrib/, but	20:07
	← webus left	20:08
	← rseifert left	20:08
sitaram	morelli: that's a nice change for me. Usually people will say "I know all about ssh, it's gitolite that is not working" :) And graaaaadually, slooooowly, come round to realising they don't know ssh as well as they thought they did :)	20:08
	→ tg joined	20:09
sitaram	morelli: http://sitaramc.github.com/0-installing/9-gitolite-basics.html will help; there's an ssh overview section	20:09
divVerent	sitaram: BTW, a slightly less intrusive way to achieve the same:	20:09
	make a one-line shell script calling gl-auth-command	20:09
	and set it as login shell for the extra accounts :P	20:09
	THAT way BTW is known to work for root vs toor	20:09
	→ dabd joined	20:09
sitaram	divVerent: awesome... and that script can pass on $USER as the argument too... oh man this will work as is	20:09
divVerent	as typically, root and toor have different shells, as the whole POINT is to still have access if the "comfortable" login shell is broken	20:10
	sitaram slaps head	20:10
divVerent	still, trying in sshd first :P	20:10
	(and using $USER of course)	20:10
sitaram	divVerent: toor? what is that?	20:10
divVerent	typically a second root	20:10
	← altrux left	20:10
divVerent	on BSD systems, many set root to have /usr/local/bin/bash as shell	20:10
	→ altrux joined	20:10
divVerent	but bash depends on some libs, and if these break...	20:10
	they then can still login as "toor", which uses /bin/sh or /bin/tcsh	20:10
thiago_home	install a static bash	20:11
sitaram	but they have the same UID?	20:11
divVerent	thiago_home: sure, POSSIBLE	20:11
	→ BrianHV joined	20:11
divVerent	sitaram: yes	20:11
	both are 0	20:11
	thiago_home: alternate solution for the same	20:11
	← dabd left	20:11
divVerent	but BSD users prefer their ports to be managable using portupgrade :P	20:11
morelli	thanks for help Ilari and sitaram for the tool ;) good bye	20:11
Ilari	divVerent: As example: http://pastebin.com/0tybNDBm	20:11
	→ Wandrewvious joined	20:11
	→ pipegeek joined	20:11
sitaram	yeah that's the part that scares me somewhat though... from an auditability point of view, you'd have to think long and hard about this to be able to say "no, my users cannot get a shell"	20:11
	→ dabd joined	20:12
	← dabd left	20:12
	→ dabd joined	20:12
pipegeek	Is there a reason behind gitweb's naming convention for git repositories where every repository directory ends in .git? When setting up a repository server, is this an important convention to follow?	20:12
sitaram	for example, how does it react to a very quick Ctrl-C as it starts up	20:12
divVerent	damn... just got the "WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED" for ssh localhost	20:12
	did I now "hack myself"?	20:12
sitaram	divVerent: I have to go get some sleep; good luck...	20:12
	← morelli left	20:12
divVerent	sure, will try :P	20:12
	login shell way probably IS better, once other login ways are provided :P	20:13
	but then one still needs the Match block to turn off e.g. port forwarding	20:13
	← lhz left	20:13
sitaram	(all in all, game's not worth the candle...)	20:13
	good night all...	20:13
jay-mccarthy	how do i fetch on a bare repo?	20:13
divVerent	hehe :P	20:13
	warthog9 → korg\|warthog9	20:14
jay-mccarthy	it says: fatal: 'origin': unable to chdir or not a git archive	20:14
	→ zuez joined	20:14
	← Wandrewvious left	20:14
	thousandparsec\|m → mithro	20:14
zuez	how do I check all of the commits that a particular file was changed on?	20:14
divVerent	sitaram: working example:	20:14
	Match User git-*	20:14
	ForceCommand echo "You are $USER"	20:15
thiago_home	zuez: git log -- filename	20:15
zuez	thanks thiago	20:15
	← WALoeIII left	20:15
divVerent	sitaram: I'll write something up ;)	20:15
	as this POC seems to work	20:15
	← Tobarja_n810 left	20:15
Ilari	divVerent: Oh, nice. I didn't expect it to work so well. :-)	20:15
divVerent	already found the first catch	20:15
	does not work with /bin/true as login shell ;)	20:16
	even if it is in /etc/shells	20:16
	← peritus- left	20:16
divVerent	so locking down the account that way fails	20:16
	→ mjf joined	20:16
divVerent	need to go the "shell script as login shell" way instead then	20:16
Ilari	divVerent: Setting ForceCommand to "/.../gl-auth-command $USER" should do the trick.	20:16
divVerent	apparently ssh passes the forced command to the login shell	20:16
	so the login shell can't be a dummy	20:16
Ilari	divVerent: Yup.	20:17
divVerent	but with the login shell not a dummy, we get into trouble if a second way to login is provided	20:17
	← fisch_ left	20:17
divVerent	e.g. puke rsh :P	20:17
pipegeek	Is there any technical reason to name bare repositories something.git, or is it just a convention?	20:17
Ilari	divVerent: Or telnet. :-)	20:17
Nugget	telnet is eeeeeeevil!	20:17
vinse	is there a way to turn off the "Did you mean..." for misspelled commands? dont know about other platforms, but with msys it can cause like a 5 seconds freeze	20:17
divVerent	is that a bot?	20:17
	telnet	20:17
	no :P	20:17
	but, MORE seriously, X11 login may be allowed (although locally)	20:18
zuez	thiago_home: it's odd, I did that, but I don't see where that file got modified in the previous commit	20:18
	→ linx\| joined	20:19
	← mythos left	20:19
	← dmg left	20:19
	← unix_remote left	20:20
divVerent	sitaram: proof of concept of an actually secure way worked too :P	20:21
	starting to document it now	20:21
	→ pklingem1 joined	20:21
	→ Wicked joined	20:21
	← pklingem left	20:21
	→ WALoeIII joined	20:22
	← WALoeIII left	20:22
Wicked	hello all. i just updated git on my freebsd box and i went to restart it and its just hung.....been like close to 5 mins...not really sure where i should check to see whats going on.	20:22
thiago_home	zuez: explain	20:22
	← ryanakca left	20:23
	← Plouj left	20:23
Ilari	divVerent: Not a bot, but has some sort of autorespond set.	20:23
	→ tmske joined	20:24
	← fgiraldeau left	20:24
Ilari	divVerent: And once it triggers, it won't trigger again for some time.	20:24
	← cilly left	20:24
	→ Zenopus joined	20:24
tmske	Hi, how can I checkout a file from svn (I'm using git-svn but one file got merged completly wrong so I want to fetch that from svn)	20:24
Ilari	Wicked: What is hung?	20:24
	→ novas0x2a joined	20:25
	→ ryanakca joined	20:25
	→ webchick joined	20:25
Wicked	the rc.d start up script its using seems like its not forking git_daemon into the background or something. i issued a "/usr/local/etc/rc.d/git_daemon start" and normally it would start the daemon then in the background..now its not	20:26
	the console just says: Starting git_daemon.	20:26
Ilari	Wicked: What version you upgraded from and to what version?	20:26
	← webchick left	20:27
Wicked	Upgrade of git-1.7.0.4 to git-1.7.0.5	20:27
	← ereslibre left	20:27
	korg\|warthog9 → warthog9	20:28
Wicked	if i open a new window in screen and do a ps i can see git_daemon is running.	20:28
	→ peritus- joined	20:29
Ilari	Wicked: The real git daemon process is git-daemon, not git_daemon	20:29
Wicked	yes. thats just a rc script to start git	20:30
	← hyperair left	20:30
	← novas0x2a left	20:30
Wicked	well git-daemon	20:30
	→ eletuchy joined	20:30
	→ CSSnub joined	20:30
Ilari	Wicked: What is parent process ID of git-daemon? It should be 1 if it became daemon...	20:31
Wicked	im wondering if the freebsd guys changed something in the /usr/local/etc/rc.d/git_daemon script that doesnt fork it into the bg	20:31
	→ WALoeIII joined	20:31
	→ pkahle joined	20:31
	← WALoeIII left	20:31
	← agib\|away left	20:32
	→ mythos joined	20:32
CSSnub	Is it necessary for me to execute the commands 'exec ssh-agent bash' and 'ssh-add privatekeyfile' in order to get access to an unfuddle repository? I eventually did get access and I'm now trying to write a tutorial for the other devs but I'm not sure if that helped me solve my problem - I'm pretty sure setting up ~/.ssh/config was the bit that solved it.	20:32
	I guess I'm not exactly sure what adding something to ssh-agent even does	20:32
Ilari	CSSnub: This is pretty standard SSH. Without ssh-agent, you just have to type passphrases a lot.	20:34
	→ altogether joined	20:34
	→ kar8nga joined	20:34
Ilari	CSSnub: If key has been added to ssh-agent, there won't be passphrase prompt even if it is encrypted.	20:34
	→ kblin joined	20:34
kblin	hi folks	20:34
CSSnub	Oh OK - so that would be helpful to include in said tutorial I suppose. :) Thank you. Does it pose any increased security risk?	20:35
Ilari	CSSnub: Well, the ssh-agent exits (and destroys the unencrypted copies of keys) with the session.	20:35
	← Adaptee left	20:36
CSSnub	So everytime I need to update my local version of the repository I will need to execute those commands again?	20:36
	← kblin left	20:37
Ilari	CSSnub: No, every time you close the shell that has ssh-agent active.	20:37
	→ lanthan_ joined	20:37
	← parasti left	20:38
Ilari	CSSnub: If you don't do that, you just get passphrase prompt (that can get annoying).	20:38
engla	seahorse can be a good ssh agent (Gnome desktop)	20:38
	← jstemmer left	20:39
divVerent	sitaram: http://rm.endoftheinternet.org/~nexuiz/0001-first-draft-of-gl-auth-loginshell-and-documentation.patch	20:39
	have to go now	20:39
	will get back to it later	20:39
	← peritus- left	20:40
	← DavidKlein left	20:40
	← lanthan left	20:41
	← shennyg left	20:41
	→ shennyg joined	20:42
	← malte_ left	20:43
	→ justin-george joined	20:43
	→ unix_remote joined	20:44
	← brizly left	20:44
	← ntoll left	20:47
	← shennyg left	20:47
	→ nairb774 joined	20:47
	← justin-george left	20:47
	← slonopotamus left	20:48
	→ shennyg joined	20:49
	→ slonopotamus joined	20:50
	← shennyg left	20:50
	← git\|spearce left	20:51
	← jceb left	20:51
	→ agile joined	20:52
	→ pheaver joined	20:52
	→ jlehmann joined	20:53
jay-mccarthy	what is the difference between %s and %b in git log/show?	20:53
	→ eek812 joined	20:54
	→ justin-george joined	20:54
	→ madewokherd joined	20:55
	← onigiri_ left	20:55
	→ parasti joined	20:56
	← slonopotamus left	20:57
Ilari	jay-mccarthy: %s is subject line, %b is body (everything except subject line).	20:57
	→ Twisted_Mentat joined	20:57
jay-mccarthy	when i run "git commit" i don't write a "subject line"	20:58
	is %s the first line of what i type and %b everything else?	20:58
Wicked	indeed the freebsd guys did change that script...and they did forget to fork it into the background	20:58
doener	jay-mccarthy: the first paragraph	20:58
	→ Garen_ joined	20:59
jay-mccarthy	so if i want everything, is that %s%b or %s%n%b or %s%+b?	20:59
Ilari	jay-mccarthy: IIRC, %B (but that's pretty new)	20:59
jay-mccarthy	my git doesn't support that	21:00
	is there no other way?	21:00
	← Garen left	21:00
	→ jmcantrell joined	21:00
	→ novas0x2a joined	21:00
	← lucsky left	21:00
	← kpreid_ left	21:01
	← justin-george left	21:02
	→ shanker joined	21:03
	← kar8nga left	21:04
	← Brodde left	21:04
	→ archis joined	21:05
	→ icwiener_ joined	21:05
	← archis left	21:06
	← pheaver left	21:06
	← ixti left	21:07
	← auslander left	21:07
	← sgh left	21:07
	← joeconyers left	21:08
	← icwiener left	21:08
	← shanker left	21:09
	→ sgh joined	21:09
	→ Pupeno joined	21:11
	→ Dieterbe joined	21:11
Dieterbe	how does it come that git can apply diffs on changed source code? in my experience, using the 'patch' tool, it fails a lot if even simple things have changed in the target file	21:13
	but git seems to be able to apply patches just fine, even if the target file has been changed	21:13
crab	it just tries harder than patch(1) does.	21:15
engla	with revision control you can do things like three-way merge, even when applying patches	21:17
	and that's a big upgrade from two-way "merge"	21:17
Dieterbe	so basically git-apply is like patch on steroids	21:18
	→ pheaver joined	21:18
engla	absolutely. For format-patch patches you use 'git am' however	21:18
	← adaro_ left	21:19
	→ adam[pac] joined	21:19
Dieterbe	indeed. i use git-am to apply patches people send me by mail. and it amazes me how they always apply, even if the patches were written against older code and such	21:20
	→ justin-george joined	21:21
engla	I like to say that git's patch management is how it's supposed to be. Atomic by default, very smart merging	21:21
	← macek left	21:21
	→ Cesario joined	21:21
	← gebi left	21:21
	← mefesto left	21:22
	→ Garen joined	21:23
	← pklingem1 left	21:23
	← nairb774 left	21:24
	→ rado1 joined	21:24
	← Garen_ left	21:25
	→ gbacon joined	21:25
	← stringo0 left	21:26
	→ webchick joined	21:26
	→ ereslibre joined	21:27
	← ereslibre left	21:27
	→ ereslibre joined	21:27
	← icwiener_ left	21:31
	→ Garen_ joined	21:31
Dieterbe	engla: uhu	21:31
	anyway i'm watching http://developer.yahoo.com/yui/theater/video.php?v=prestonwerner-github, awesome talk schacon	21:32
schacon	ugh, i hate that one	21:32
	← tbf left	21:32
schacon	the first section of slides was all messed up	21:32
Dieterbe	oh. well it's your first video i see :)	21:32
schacon	it was embarrassing	21:33
	→ tty1 joined	21:33
	← Garen left	21:33
	← psoo left	21:33
schacon	that's the most i've ever f'd up a talk and i've done like 40 in the past year	21:33
	but thanks for the nice words! :)	21:34
	← Bennid left	21:34
	← raichoo left	21:35
Dieterbe	well, i'm not done yet watching it. surely you skipped quite some stuff but i find it very interesting nonetheless, as i imagine many others do	21:36
	← adam[pac] left	21:36
	→ keyvan joined	21:36
	← keyvan left	21:37
	→ k3yvn joined	21:37
	→ vasandgvd joined	21:37
	→ nairb774 joined	21:37
	← dj_tjerk left	21:38
	← Garen_ left	21:39
	→ pklingem joined	21:39
	→ Garen joined	21:40
	← bx2 left	21:41
	→ raichoo joined	21:41
	→ jrichardson joined	21:41
	← raichoo left	21:41
	→ doener joined	21:44
	→ TDJACR joined	21:45
	← khmarbaise left	21:45
	← pheaver left	21:46
	→ pheaver joined	21:46
	→ slonopotamus joined	21:47
	→ kpreid_ joined	21:48
	← saccade left	21:50
	→ binjured_ joined	21:51
tgunr	anyone know how to correct a "error: pack-objects died of signal"? google is not much help	21:51
	this was during a clone of a fresh --bare repo	21:52
	← binjured_ left	21:53
	→ g0bl1n joined	21:54
	→ icwiener joined	21:55
	← pk4r left	21:56
	← g0bl1n left	21:57
jrichardson	tgunr: have you tried redoing it?	21:57
	→ s4msung joined	21:57
tgunr	the clone?	21:57
	→ simplechat joined	21:58
	← alley_cat left	21:58
tgunr	error: git upload-pack: git-pack-objects died with error.	21:59
	fatal: git upload-pack: aborting due to possible repository corruption on the remote side.	21:59
	I suppose i can re-upload the bare repo to the server again but this is a bit disconcerting	21:59
jrichardson	tgunr: Yeah, possible corruption would make me nervous too!	22:00
	← slonopotamus left	22:00
	→ mikedub joined	22:01
	← kipras left	22:01
	← RobertLaptop left	22:01
	← ereslibre left	22:01
mikedub	so I just issued `fatal: Out of memory, malloc failed	22:02
	oops	22:02
	← sixteneighty left	22:02
mikedub	I issued `git add *` and got the error above.	22:02
cbreak	did you add a lot or huge files?	22:02
	← brodyberg left	22:03
mikedub	a lot of files, yeah	22:03
	inital commit	22:03
	← webchick left	22:03
mikedub	something like `find . -type f -name '*php' -exec git add {} \;` should work, though, eh?	22:04
cbreak	both should work	22:04
	do you have only little memory?	22:04
mikedub	2GB	22:05
	→ n3kl joined	22:05
mikedub	there are a ton of files though	22:05
tgunr	thats not little	22:05
mikedub	like, a silly amount	22:05
	← navetz left	22:05
	k3yvn → keyvan	22:05
tgunr	what os?	22:05
n3kl	Hello	22:05
mikedub	linux	22:05
n3kl	is this the git vcs channel?	22:05
	← Paraselene__ left	22:06
mikedub	the machine is pretty lean	22:06
	← aheinecke left	22:06
mikedub	I have to attribute it to there being too many files	22:06
mugwump	n3kl: yes	22:06
tgunr	i would think 2GB is way more than adequate	22:06
mikedub	me too :]	22:06
tgunr	try the incremental find and add	22:06
mikedub	yeah I've got that running now	22:07
	← nairb774 left	22:07
	← eek812 left	22:07
mikedub	git add * ran for ~5 mins before i got the out of mem error	22:07
mugwump	how big are you files?	22:07
	*your	22:07
mikedub	smallish	22:07
	← flazz left	22:08
mikedub	1-30kb	22:08
tgunr	thumbnails :)	22:08
mikedub	heh no I'm ignoring image files	22:08
	it's an ecom system	22:08
	Magento	22:08
	built kinda like the Zend framework.. loads of files buried in a sea of directories	22:09
n3kl	I am comming from years of subversion to <2 days of git and struggling with something. I created a local repository, added a remote and pushed the files to the remote. Works good when I clone from the remote hub also. Now when I clone from the original and push the changes back to the original, then attempt to push those changes to the git hub, the clones don't get the updated version.	22:09
	They can see the log entries, but I can't update to teh local filestructure of the latest revision	22:10
	what am I doing wrong?	22:10
	→ coppro joined	22:11
mikedub	6500+ files I guess	22:12
mugwump	n3kl: the answer is to be found in the first few paragraphs of man git-push (the refspec stuff), and looking at 'git show-ref' and 'git ls-remote origin'	22:12
Gitbot	n3kl: the git-push manpage can be found at http://git.or.cz/man/git-push	22:12
	← rgr left	22:13
cbreak	n3kl: did you push to a non-bare?	22:13
mugwump	what you are trying should work, you just need to make sure you are pushing the right refs to the right places	22:13
n3kl	the first push to the hub was a bare	22:13
cbreak	do you know what bare means?	22:14
	← binjured left	22:14
tgunr	n3kl: are you expecting all the clones to be updated after a push?	22:14
n3kl	cbreak: maybe, it looks like it only contains the repository and not the filesystem. Though, to answer your question the repo was created with git init --bare	22:15
	← LiamH left	22:15
cbreak	ok, then pushing to it should not be a problem	22:15
mikedub	it created 9046 objects in the git repo	22:15
n3kl	tgunr: yes. I expect that once I push to the hub, that when I do a fetch from the clones of that hub, the local file structure is of the latest revision	22:15
cbreak	fetch won't do that	22:16
n3kl	though, when I did push changes back to the original, the original is not a bare repo	22:16
cbreak	fetch will update remote tracking branches	22:16
	not the local branches	22:16
	use pull or rebase for that	22:16
	pull or pull --rebase I mean	22:16
	← sgh left	22:17
n3kl	so fetch only updates the metadata, so to speak?	22:17
	→ Meow`` joined	22:18
	← pheaver left	22:18
n3kl	am what I attempting common?	22:18
	← coppro left	22:18
	← jlehmann left	22:19
	→ jovianjake joined	22:19
	← engla left	22:20
	→ blaines joined	22:20
	← blaines left	22:20
	→ nairb774 joined	22:20
	→ blaines joined	22:21
	→ webchick joined	22:23
	← unix_remote left	22:24
n3kl	Alright, so when I do a git pull --rebase on the clone of the original that I originally made the changes on, I loose my changes to the filestructure.	22:24
	← aspotashev\|eeepc left	22:27
	← pklingem left	22:27
tgunr	does git version 1.7.1.rc1 support post-checkout hook?	22:28
	← webchick left	22:29
	→ coppro joined	22:29
	← jboom left	22:29
tgunr	neve mind, brain fart	22:30
	→ jboom joined	22:30
	→ Garen_ joined	22:31
	→ tazle_ joined	22:31
jrichardson	once I have changed a file, and want to change it back, in the local repo (without having commited the original changes) how do I do that? I tried rm file, and then git pull to get fresh copy, to no avail. What am missing?	22:32
tgunr	i think git rm --cached, double check the option	22:32
	← khelll left	22:32
	← Garen left	22:33
	← SimManiac left	22:33
tgunr	nope	22:33
	git reset HEAD file ?	22:33
	→ ipatrol joined	22:33
ipatrol	Why won't git://github.com/zvoase/baseconv.git work?	22:34
	→ bousquet joined	22:34
tgunr	define work :)	22:34
	← tg left	22:35
mugwump	Look buddy, doesn't work is a strong statement. Does it sit on the couch all day? Is it making faces at you? Does it want more money? Is it sleeping with your girlfriend? Please be specific!	22:35
	← jboom left	22:35
ipatrol	I can't download the content for the tree	22:35
	← andreaa left	22:35
	→ spaceonline joined	22:35
	→ tg joined	22:36
tgunr	works here	22:36
nDuff	ipatrol, what does it do when you try?	22:36
	← Weasel[DK] left	22:36
tgunr	as in: I got the repo	22:36
ipatrol	nDuff: All I get is the .git folder	22:37
nDuff	ipatrol, and if you check out a branch with code on it?	22:38
ipatrol	nDuff: I use Git GUI	22:38
	→ corto joined	22:38
	← nairb774 left	22:38
	← cannonball left	22:39
	nDuff is not in a position to support arbitrary GUIs.	22:39
	→ camonz joined	22:39
corto	hi guys, i'm just wondering if there is a simple way to copy a snapshot of current working tree to another dir... like when preparing my kernel, i would copy a snapshot in /usr/src/linux, but continue working in my git dir...	22:39
ipatrol	nDuff: It comes with git for Windows	22:39
n3kl	can I wipe out my last commit. I have taken things from bad to worse	22:41
ipatrol	nDuff: what's a revision expression?	22:42
	← brez- left	22:42
mugwump	ipatrol: see man git-rev-parse	22:42
Gitbot	ipatrol: the git-rev-parse manpage can be found at http://git.or.cz/man/git-rev-parse	22:42
	← jonshea left	22:43
n3kl	how do I know which commit I am working on now?	22:43
tgunr	n3kl: look at git commit --amend	22:43
ipatrol	All I want to do is download http://github.com/zvoase/baseconv	22:45
tgunr	given a sh1 reference, what is the quickest way to find the file associated with it? I need to use it in a script.	22:45
ipatrol	Yay, github archive downloads!	22:46
tgunr	ipatrol: git clone git://github.com/zvoase/baseconv.git works for me	22:46
	→ novas0x2a1 joined	22:46
	← thiago_home left	22:46
n3kl	ug, thanks for the tips	22:46
ipatrol	This gui probably has issues	22:47
	← novas0x2a left	22:47
	→ MrPunkin joined	22:48
	← altrux left	22:50
MrPunkin	I stashed away some changes, changed branches, patched some code, changed back to master and merged in the patch, but now am having issues popping the stash back onto master.	22:50
	→ nairb774 joined	22:51
	→ RobertLaptop joined	22:52
	→ aadster joined	22:52
aadster	ns identify santa572june	22:53
	← coppro left	22:53
mugwump	aadster: attack localhost	22:54
aadster	hm	22:54
	whats that	22:54
mugwump	er	22:54
	sounded like you were a zombie	22:54
	10:53 < aadster> ns identify santa572june	22:54
	→ ereslibre joined	22:54
	← ereslibre left	22:54
	→ ereslibre joined	22:54
mugwump	if that's some password of yours, I'd change it ;)	22:54
aadster	fk	22:54
	thats my pass i use for everything	22:54
mugwump	orly	22:55
aadster	ns identify santa572june [email@hidden.address]	22:55
	hm	22:55
	why itsn ot working	22:55
mugwump	ooo	22:55
	now I can log into your gmail	22:55
	← mikedub left	22:55
mugwump	and start assuming your identity	22:55
aadster	hm	22:55
	i doubt that	22:55
	you dont have my ip	22:55
	→ cdr- joined	22:55
aadster	u think u know abut computers	22:56
	my dad teaches them	22:56
mugwump	yeah? my dad is a computer	22:56
aadster	ns identify santa572june [email@hidden.address] register	22:56
corto	my dad is a moron, he's proud of me	22:57
aadster	ns identify santa572june [email@hidden.address] aadster	22:57
rolfb	aadster: considering trying /msg nickserv instead ?	22:57
	mugwump chuckles	22:57
	← Arelius left	22:57
	→ Arelius joined	22:58
aadster	msg nickserv identify santa572june [email@hidden.address]	22:58
rolfb	hmm	22:58
	.. / not working?	22:58
aadster	no	22:58
	→ coppro joined	22:58
tgunr	ok, twice now i have uploaded a 1.2GB repo created locally with git clone --bare and both times a subsequent clone of the repo on the server is reuslting in git-pack-objects died with error. Is there any way to check the objects on the server? I see git-fsck but that seems to be for a local cloned repo	22:59
crab	git fsck is for any old repo.	22:59
cbreak	how much ram is on the server?	22:59
	do you have shell access?	22:59
tgunr	i shell access	23:00
	don;t know how much ram, let me see	23:00
cbreak	can you do git repack on the server?	23:00
tgunr	crab: include a bare?	23:00
crab	tgunr: sure.	23:00
	← coppro left	23:00
	→ eighty4 joined	23:00
MrPunkin	Can anyone tell me if you can apply stashed changes to an updated commit of the branch the stash originated from?	23:01
cbreak	you can attempt to apply them to anything	23:01
	← aadster left	23:01
tgunr	ah ha: Yikes! One of your processes (git, pid 1479) was just killed because your	23:01
	processes are, as a whole, consuming too much memory. If you believe you've	23:01
	received this message in error, please contact Support.	23:01
	← nairb774 left	23:02
	← TeckniX left	23:02
cbreak	git repack uses a lot of memory	23:02
	I read somewhere about twice as much as the biggest blob or something	23:02
	but I have no idea	23:02
	some people said they ran out of 32bit memory space :)	23:03
	← schacon left	23:03
	← giallu left	23:03
	← warlock_mza left	23:03
	← corto left	23:04
	← spaceonline left	23:05
	← priidu left	23:05
	← jrichardson left	23:10
AlericInglewood	um, how to rename a branch?	23:11
	← bentob0x left	23:11
	→ jonshea joined	23:12
ThiefMaster	git branch -m from to	23:12
AlericInglewood	thanks	23:12
	← jonshea left	23:13
	← Arelius left	23:14
	→ ilteris- joined	23:14
	→ Arelius joined	23:14
	→ orafu joined	23:15
	→ hayden92 joined	23:15
	→ nairb774 joined	23:18
	← hayden92 left	23:18
	→ christophsturm joined	23:18
	→ byteframe joined	23:20
	→ djbpython joined	23:20
byteframe	How can I list the last tag created?	23:20
	← MrPunkin left	23:20
djbpython	new to git, i set up a repo add everything then commit -a, trying to archive and its just printing everything to the terminal...	23:21
	← CSSnub left	23:21
djbpython	--format=tar and i've verified it does support tar	23:22
	following everything i've found on the internet, seems straight forward	23:22
	← tvw left	23:23
tgunr	given the following reference in .git/objects how can i find out which file this is?	23:24
	-rw-r--r-- 1 tgunr 194103351 2010-04-11 09:01 9a462c0fe7e5ede65d0c2a77f0606ba11148d9	23:24
	→ jsilver joined	23:24
davr	tgunr: what are you using git archive for?	23:24
	err, sorry, that should be djbpython:	23:24
	→ TeckniX joined	23:24
tgunr	this file is really huge and casuing me greif on the server	23:24
	← pantsman left	23:24
tgunr	grief too	23:25
djbpython	davr to send a django project to my webhost	23:25
davr	tgunr: look under "Removing Objects" here: http://progit.org/book/ch9-7.html - it tells how to find what the largest files in your repo are	23:27
	→ onigiri_ joined	23:27
davr	djbpython: according to the man page for 'git archive', writing to std out is what its supposed to do by default. did you forget to put "--output=<file>" ?	23:28
djbpython	doh	23:28
davr	i guess you could also do "git archive (blah) > outputfile"	23:28
tgunr	cc, reading	23:28
Ilari	tgunr: 'git rev-list --all --objects \| grep 9a462c0fe7e5ede65d0c2a77f0606ba11148d9' should print one name for it.	23:29
	← codetroll left	23:29
djbpython	davr, that option is not in the usage at the command line which is how i missed it i guess	23:29
	thanks	23:29
	← Arelius left	23:31
	→ Arelius joined	23:31
	← TeckniX left	23:31
	→ schacon_ joined	23:32
	← jrmuizel left	23:32
	← aziz left	23:33
	→ t0rc joined	23:36
	← Yuuhi left	23:36
t0rc	If you have git and ssh running on a few computers, do you have to do anything special to pull/push between them?	23:36
	← mithro left	23:37
RandalSchwartz	the git command has to be in your path	23:37
	or maybe it's git-receive-pack	23:37
	→ spearce joined	23:37
qfr	t0rc nope not really	23:38
	I used to do it frequently	23:38
	← v0n left	23:38
RandalSchwartz	git remote add othername hostname:/path/to/repo	23:38
	git fetch othername	23:38
qfr	I just started out with git init --bare or whatever it was called and then I could push to it	23:38
RandalSchwartz	git push othername	23:38
	etc etc	23:38
t0rc	nice nice	23:38
	→ ajonat joined	23:39
RandalSchwartz	just be sure your workflow doesn't require pushing into a checked out branch	23:39
	that'll be a world of hurt	23:39
t0rc	it'll just be my work. :)	23:40
RandalSchwartz	well - even still	23:40
	doesn't matter whose it is	23:40
	don't push into a checked out branch	23:40
	RandalSchwartz wanders off to get dinner	23:40
	→ juan_arandaalvar joined	23:40
t0rc	what do I do if I'm only using one branch though?	23:41
	← t0rc left	23:43
	→ bx2 joined	23:43
	→ fgiraldeau joined	23:45
	← spearce left	23:45
	→ enherit joined	23:45
	← dreiss left	23:46
	→ Titan8990 joined	23:46
	← Arelius left	23:47
	← simplechat left	23:48
	→ Arelius joined	23:48
schacon_	btw, if anyone is interested - github now supports smart http protocol	23:48
	if you want to test it out and you have a newer git, have at it	23:48
mugwump	win on toast	23:49
	← altogether left	23:49
schacon_	let me know if anything asplodes	23:49
	→ slonopotamus joined	23:50
	← icwiener left	23:51
	→ enherit_ joined	23:52
	← bousquet left	23:53
	← Heimidal left	23:55
	← enherit left	23:56
	enherit_ → enherit	23:56
	→ navetz joined	23:57
	→ enherit_ joined	23:57
	→ jomofo joined	23:58
	← AAABeef left	23:59

Logs Search ←Prev date Next date→

Channels Documentation