IRCloggy #git 2008-11-09

Provider of IRC logs since 2005.

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

	2008-11-09
werdnus	141M .git	00:00
wereHamster	tobstarr: yes, see git-clone --depth	00:00
werdnus	wereHamster: There's the vn-remote "svn"]	00:01
tobstarr	wereHamster: thanx a lot	00:01
werdnus	wereHamster: is there some black magic I can use to tell it where trunk, branches and tags are?	00:01
	← nice_burger left	00:01
	→ pygi joined	00:02
wereHamster	werdnus: when git-svn checks out the repository, it checks out every single revision and recreates the history in git. I'm not aware of any way to avoid this, but then again, I know very few about git-svn	00:02
werdnus	wereHamster: I don't have a problem with it doing that.	00:03
	wereHamster: I have a problem with it redoing it.	00:04
	because it's already done it once for trunk, so I just want it to do the same for branches and tags, but not trunk.	00:04
wereHamster	so you already have a git directory with the svn source imported into a git project?	00:04
werdnus	correct.	00:04
	but I imported /mediawiki/trunk/phase3	00:04
	I want /mediawiki/trunk/phase3, and I also want to import mediawiki/branches//phase3 and mediawiki/tags//phase3	00:05
	(as branches and tags)	00:05
wereHamster	then add [svn-remote] entries to your .git/config file and git-svn fetch the svn branches/tags	00:06
	← parasti left	00:07
	→ bieneff joined	00:07
	← tobstarr left	00:08
	← jsimonss_ left	00:08
	← dwmw2_gone left	00:09
	← \ask- left	00:09
	← softdrink left	00:09
	therrg → therrg\|away	00:09
	← thedoctor left	00:10
pygi	bobesponja, ping	00:11
	→ softdrink joined	00:13
pygi	it's not nice to ignore people bobesponja xD	00:14
fynn	why does the following script hang:	00:16
	for fn in $( bzr ls --versioned ); do git add -f $fn; done	00:16
	← Weasel[DK] left	00:17
fynn	nvm, found it.	00:18
werdnus	wereHamster: git-svn fetch?	00:19
	werdnus looks up	00:19
fynn	hm, I can't commit an empty directory?	00:20
	→ jsimonss joined	00:20
Mikachu	no	00:20
pygi	fynn, you can "touch .gitignore"	00:21
fynn	isn't that something people sometimes miss?	00:21
pygi	and that will work	00:21
werdnus	Is there some reference for the .git/config file?	00:21
wereHamster	werdnus: you set up your svn-remote as [svn-remote "mediawiki-branches-XXX-phase3"], point it to the svn branch, and set it up to fetch into :refs/remotes/mediawiki-branches-XXX-phase3. Then you `git-svn fetch mediawiki-branches-XXX-phase3` to get the latest bits from the svn repository	00:21
Mikachu	werdnus: man git-config	00:21
Gitbot	werdnus: the git-config manpage can be found at http://git.or.cz/man/git-config	00:21
fynn	pygi: inside the empty directory you mean... and then of course I'd have .gitignore inside it.	00:21
wereHamster	use the existing [svn-remote] as a reference	00:21
pygi	fynn, true, but heh	00:22
	only until you add something, you can remove it afterwards	00:22
fynn	pygi: k, thanks.	00:22
	→ stuffcorpse joined	00:23
pygi	fynn, you could probably remove it from history altogether later on, but that's just evil :)	00:25
	← softdrink left	00:25
	← advany left	00:26
	→ foca joined	00:26
fynn	pygi: yeah, it's not a huge deal	00:27
	just something we got used to doing in other SCMs	00:27
pygi	fynn, I guess you know why you can't add empty dirs, right?	00:27
	→ softdrink joined	00:28
fynn	pygi: not exactly. something about git only tracking files?	00:28
pygi	fynn, no, git doesn't track files	00:28
	git tracks content	00:28
werdnus	wereHamster: okay I got it fetching, but it seems to insist on fetching from r0 again, which would mean 45000 revs	00:28
	→ bdiego joined	00:28
pygi	fynn, that's a better "definition" :)	00:28
fynn	OK...	00:28
pygi	and dirs are not content per-se	00:28
	:)	00:28
werdnus	wereHamster: can't I make it source some of the data from the original fetch?	00:28
	getha → thijso	00:29
wereHamster	werdnus: it will do that automatically as far as it can	00:29
	← Zimm left	00:29
	← stuffcorpse left	00:30
werdnus	wereHamster: well, it doesn't seem to be. There was an svn copy in the deep past, so it should use the data previously fetched, right?	00:30
wereHamster	if the branch has some commits common with the trunk you already have, it won't fetch those changesets. But of course git-svn still has to walk the whole history	00:30
werdnus	hmm	00:31
	← Yuuhi left	00:33
	← jackdempsey_ left	00:33
werdnus	hmm, it seems to have fetched fully in like five seconds	00:33
	← icwiener left	00:33
werdnus	maybe I misunderstood :)	00:33
	thanks for your help, wereHamster	00:33
wereHamster	glad it worked	00:37
werdnus	hehe, now to try and sync my local branch with the remote one :)	00:38
wereHamster	a simple `git-svn fetch` for each of the svn-remotes should do the trick	00:39
werdnus	yeah, I did that, but I have to apply the work on the local branch to the remote svn	00:39
	what I've done is this:	00:39
	andrew@gcentral_router:~/mediawiki$ git diff svn/group_rights group_rights \| git apply	00:40
	or should I do a git merge?	00:40
wereHamster	you want to commit your local changes to the remote svn repository?	00:40
werdnus	yeah	00:40
wereHamster	use git-svn dcommit	00:40
werdnus	Not quite that simple	00:40
	see, the local branch doesn't track the remote, for starters.	00:40
wereHamster	ah. you only track one subdirectory, right?	00:41
werdnus	and my local branch has already been rebased to trunk, whereas my remote branch hasn't.	00:41
	→ jackdempsey_ joined	00:41
werdnus	yes, I track /trunk/phase3, and I've made changes to a local branch which is based from that.	00:42
	However, I need to commit this branch to svn for review (this is /branches/group_rights)	00:42
	and that branch already exists.	00:42
	← jackdempsey_ left	00:42
werdnus	so I want to upload my changes on my local branch based on /trunk/phase3 up to /branches/group_rights	00:42
wereHamster	so create a svn-remote for /branches/group_rights, fetch that branch, rebase your work on top of it, dcommit the changes	00:43
werdnus	to accomplish this so far I've used the svn-fetch stuff you talked about to retrieve svn/group_rights as a remote branch.	00:43
	yeah, except my local branch has been updated to reflect trunk, whereas the remote branch hasn't.	00:44
	→ stuffcorpse joined	00:44
werdnus	andrew@gcentral_router:~/mediawiki$ git diff svn/group_rights group_rights \| git apply	00:44
	That worked okay, except for a binary file.	00:44
	I can either git checkout that file (how do I specify to check it out from a different branch?), or git reset and use git-merge.	00:44
wereHamster	create a temporary branch where you rebase your changes on top of the /branches/group_rights branch, dcommit, throw the temporary branch away	00:44
werdnus	andrew@gcentral_router:~/mediawiki$ git checkout -b temp_group_rights svn/group_rights	00:45
wereHamster	I assume someone else will then review your changes, and copy your work over to /trunk/phase3, right?	00:45
werdnus	yeah	00:45
	It's for a job I'm starting at Wikimedia tomorrow :)	00:46
	→ quoin joined	00:46
wereHamster	then the temporary branch thingy should work :)	00:46
	→ jchris joined	00:46
	werdnus waits for the git rebase	00:46
	jchris is about to attempt the rebase	00:47
jchris	maybe...	00:47
	I have committed some stuff to branchA, then branched to B and kept working	00:47
	now I realize that the last few commits to branchA should only be in branchB	00:47
	→ cannonball joined	00:48
jchris	can I use rebase to obliterate them from branchA?	00:48
wereHamster	jchris: yes	00:48
jchris	will they get obliterated from branchB later should I decide to merge A into B?	00:48
wereHamster	jchris: the git-rebase manpage has examples	00:48
	merge never removes commits	00:49
	→ david_koontz joined	00:49
jchris	cool thank wereHamster	00:49
wereHamster	the worst thing you can have are conflicts	00:49
	→ G_SabinoMullane1 joined	00:53
	← G_SabinoMullane1 left	00:57
	← markelikalderon left	00:57
	← glommer left	00:57
	← cbreak left	00:57
	← litage left	00:57
	← piggybox left	00:57
	← eternaleye left	00:57
	← G_SabinoMullane left	00:57
	← sgrimm left	00:57
	← madewokherd left	00:57
	← LiamH left	00:57
	← bilboed left	00:57
	← ferdy left	00:57
	← reuss left	00:57
	← joshdavey left	00:57
	← tjafk2 left	00:57
	← Jacolyte left	00:57
	← ruphy left	00:57
	← mpr left	00:57
	← SRabbelier left	00:57
	← Thumper_ left	00:57
	← chrizz_ left	00:57
	← dkr left	00:57
	← precision left	00:57
	← sugi left	00:57
	← GarethAdams left	00:57
	← RandalSchwartz left	00:57
	← lamont left	00:57
	← AlexanderS left	00:57
	← resmo left	00:57
	← jmspeex left	00:57
	← jackdempsey left	00:57
	← cardioid left	00:57
	← jchris left	01:00
	→ ferdy joined	01:02
	→ abbe joined	01:02
	→ Thumper__ joined	01:03
	→ markelikalderon joined	01:03
	→ glommer joined	01:03
	→ cbreak joined	01:03
	→ piggybox joined	01:03
	→ eternaleye joined	01:03
	→ bilboed joined	01:03
	→ sgrimm joined	01:03
	→ madewokherd joined	01:03
	→ LiamH joined	01:03
	→ reuss joined	01:03
	→ joshdavey joined	01:03
	→ tjafk2 joined	01:03
	→ Jacolyte joined	01:03
	→ ruphy joined	01:03
	→ mpr joined	01:03
	→ SRabbelier joined	01:03
	→ chrizz_ joined	01:03
	→ sugi joined	01:03
	→ resmo joined	01:03
	→ lamont joined	01:03
	→ jackdempsey joined	01:03
	→ GarethAdams joined	01:03
	→ jmspeex joined	01:03
	→ RandalSchwartz joined	01:03
	→ precision joined	01:03
	→ dkr joined	01:03
	→ cardioid joined	01:03
	→ AlexanderS joined	01:03
	← piggybox left	01:03
	→ piggybox joined	01:05
	← piggybox left	01:08
	→ jackdempsey_ joined	01:10
	therrg\|away → therrg	01:12
	← bilboed left	01:12
	← therrg left	01:12
	← pcapriotti left	01:14
	← david_koontz left	01:17
	← jackdempsey_ left	01:18
	← SRabbelier left	01:22
	← digitalhobbit left	01:23
	← tvw left	01:27
	→ tvw joined	01:27
	→ emss joined	01:40
	← drizzd left	01:41
fynn	argh	01:43
	I keep getting an error: "svn-remote.svn.url already set"	01:43
	when I try to clone a remote repo.	01:44
	anyone knows how to fix this?	01:44
Pieter	http://groups.google.com/group/Google-Code-Search/browse_thread/thread/c03baa114348e0b7?hl=en	01:46
	hehe	01:46
fynn	you're implying the guy wants to hax?	01:47
werdnus	that makes me giggle.	01:49
	← jelmer left	01:51
	→ brettryan joined	01:51
brettryan	hello there	01:51
	→ jelmer joined	01:52
brettryan	could someone tell me how I would undo a `git rm'?	01:52
	I haven't performed a `git commit' yet.	01:52
	→ loinclot_ joined	01:52
fynn	you may be screwed.	01:52
bremner	brettryan: man git-reset	01:52
Gitbot	brettryan: the git-reset manpage can be found at http://git.or.cz/man/git-reset	01:52
fynn	bremner: would that help if he didn't commit?	01:52
	→ jelmer_ joined	01:53
bremner	I thought he meant he didn't commit the delete	01:53
brettryan	I was trying git reset, though I couldn't figure out how to simply restore the rm.	01:53
tvw	I am a single developer and new to version control: could git be helpful to me too, or is it only useful for groups of developers?	01:53
	jelmer_ → Guest29580	01:53
fynn	tvw: sure it could.	01:53
	even as a single developer, you need version control.	01:53
bremner	tvw: I find it very useful. But it will take a little time to be get used to	01:53
	← jelmer left	01:53
	→ jelmer joined	01:54
bremner	brettryan: is the file uncommited or the delete	01:54
brettryan	I did "git rm <file>" but then want to undo that.	01:55
	I have other changes that I want comitted.	01:55
tvw	ok, one last question or two: is there a good book for my type of a developer? Where do I best store my repository?	01:55
	← jelmer left	01:56
fynn	tvw: just read the git tutorial.	01:56
	← loincloth left	01:56
	← Guest29580 left	01:56
fynn	you can store it locally, and backup regularly, or remotely.	01:56
bremner	brettryan: did you previously commit <file>	01:56
fynn	you can even work against an SVN remote.	01:56
brettryan	<file> is in the repo yeah	01:56
tvw	thanks and good night!	01:57
fynn	good night.	01:57
	OK, there's an SVN repo I'd like to store my .git changes on.	01:57
brettryan	I did a `git reset HEAD file' but that just removed it from the files to be comitted.	01:57
fynn	how do I do that?	01:57
bremner	brettryan: ok, now try git checkout HEAD file	01:58
brettryan	ah fantastic, thank you very much	01:58
bremner	brettryan: cool, you're welcome	01:58
brettryan	bremner: Thankyou, ah geez, I feel a little silly, I didn't realise you could actually give checkout a filespec	01:59
	Up until now I thought checkout was just for switching branches	01:59
fynn	iow, I don't want to do 'git svn clone', but rather to commit directly to the remote SVN from my local git branch.	01:59
	← Weiss left	02:00
bremner	fynn: I don't really understand. Is git svn dcommit what you want?	02:02
	← lemonodor left	02:02
brettryan	bremner: If I may just clarify one thing, if I do git rm, then do I first need to do a git reset prior to performing the git checkout?	02:03
	just for future reference :)	02:03
fynn	bremner: yeah, except it seems that 'git svn dcommit' expects the local branch to be a clone of some remote repo.	02:03
bremner	brettryan: I'm not sure. Make a temporary branch and try it out (git checkout -b tmp)	02:04
	← eternaleye left	02:05
bremner	fynn: as far as I understand it, that might be the only option.	02:05
	← flaguy left	02:06
fynn	bremner: hm, too bad.	02:06
brettryan	bremner: okay, just did, it turns out you don't need to do a git reset, git is clever enough to update the cache correctly	02:06
bremner	brettryan: now we both learned something :-)	02:06
	← bieneff left	02:07
brettryan	bremner: hehe, thank you very much once again.	02:08
	← emss left	02:08
	← dkr left	02:08
	← sgrimm left	02:08
	← LiamH left	02:08
	← cbreak left	02:08
	← markelikalderon left	02:08
	← precision left	02:08
	← tjafk2 left	02:08
	← sugi left	02:08
	← Thumper__ left	02:08
	← GarethAdams left	02:08
	← RandalSchwartz left	02:08
	← lamont left	02:08
	← AlexanderS left	02:08
	← resmo left	02:08
	← jmspeex left	02:08
	← jackdempsey left	02:08
	← cardioid left	02:08
	← Jacolyte left	02:08
	← joshdavey left	02:08
	← chrizz_ left	02:08
	← mpr left	02:08
	← reuss left	02:08
	← glommer left	02:08
	← madewokherd left	02:08
	← ruphy left	02:08
brettryan	I'm coming from a CVS background so the concepts are similar but at the same time very different.	02:08
bremner	brettryan: yeah, the index is the biggest conceptual difference IMHO	02:09
	→ emss joined	02:09
	→ Thumper__ joined	02:09
	→ markelikalderon joined	02:09
	→ glommer joined	02:09
	→ cbreak joined	02:09
	→ sgrimm joined	02:09
	→ madewokherd joined	02:09
	→ LiamH joined	02:09
	→ reuss joined	02:09
	→ joshdavey joined	02:09
	→ tjafk2 joined	02:09
	→ Jacolyte joined	02:09
	→ ruphy joined	02:09
	→ mpr joined	02:09
	→ chrizz_ joined	02:09
	→ sugi joined	02:09
	→ resmo joined	02:09
	→ lamont joined	02:09
	→ jackdempsey joined	02:09
	→ GarethAdams joined	02:09
	→ jmspeex joined	02:09
	→ RandalSchwartz joined	02:09
	→ precision joined	02:09
	→ dkr joined	02:09
	→ cardioid joined	02:09
	→ AlexanderS joined	02:09
	← aspotashev left	02:09
brettryan	bremner: I would have to agree, at firse it seemed foreign, but now I'm liking it a lot	02:10
	the trick is going to be convincing my dev team to switch	02:10
	they're all used to working in the same working tree together. Like an RCS style.	02:10
	← emss left	02:10
bremner	brettryan: you can do that with a central bare repo to push and pull from	02:11
brettryan	yep, I'll end up doing that, but .. they all work on the same working tree currently.	02:11
mugwump_	you can still do that :)	02:12
	→ eternaleye joined	02:12
mugwump_	in fact git is a very good rcs replacement... no server required, easy to use	02:12
	you only need a handful of commands for rcs-like usage	02:12
brettryan	IMHO it's a nightmare, though they seem to feel that if they call out to the person sitting next to them that they want a file that someone else is working on they rename the file to <file>.<login> lol	02:12
	mugwump_: actually, I think you could be right there	02:13
	though currently we're using CVS, with a management script that copies from one machine to another based on state	02:13
bremner	brettryan: most people find working on laptops pretty convincing	02:13
mugwump_	http://utsl.gen.nz/talks/lightning/git-for-sysadmins/ # a little write-up of it	02:13
	→ ben_h joined	02:13
brettryan	we basically have three AIX machine (development, qa, production), developers checkout on the dev machine, then perform a `d2q' (devel to qa), then a q2p (qa to production)...	02:14
	so code is always shared between the developers	02:15
bremner	brettryan: just change the scripts to do pushes without telling them :-)	02:15
	here in the 21st century, we are using as many as two computers per developer :-)	02:15
	→ dduncan joined	02:15
brettryan	lol. Yeah I know :)	02:17
	We're in the process of migrating to Linux, so I'm preparing to change the VCS at the same time.	02:17
mugwump_	no need to be all elitist about modern largess though. the old workflow patterns are still possible	02:17
	and if you can show that, it's usually easier to switch ... you gradually learn the new things	02:18
brettryan	I was thinking more that developers might benefit from using a `pull' of a centralised repo in their home directories.	02:18
	that's true	02:18
	I think you're right, making changes to the scripts to not interfere with the current workflow might be a good start.	02:18
mugwump_	yeah, show it working how they're used to, then at the end show, "and you can ALSO do this..." and show the history in gitk	02:19
brettryan	can't do gitk, we basically ssh from windows machines.	02:20
	well actually, I have been using gitk from my macbook	02:20
bremner	tig is worth a look	02:20
brettryan	so I guess cygwin or something coul do the trick.	02:21
	tig?	02:21
bremner	tig = gitk in terminal, kinda	02:21
	← dduncan left	02:21
	← sgrimm left	02:21
brettryan	ah, nice.	02:21
	got a url?	02:22
	just found it, nvm :)	02:22
	→ sgrimm joined	02:23
	← tvw left	02:24
brettryan	The one thing I like the most about git is how it stages.	02:26
	I used to hate having to do cvs commit file1 file2 file3 file4...	02:26
	now I just git add each file as I need to and then worry about the commit when I'm ready.	02:26
fynn	how do I see just the untracked files in the current directory?	02:28
ben_h	mm	02:30
	the beauty of it is it's always at least as good as the other method	02:31
brettryan	fynn: I'm not too sure you can, I just tried looking in the man page, aparently git-status takes the same arguments.	02:31
	but if I were to `git status .' it still tells me about parent folder changes.	02:32
	same arguments as git-commit btw.	02:32
fynn	brettryan: right, the same steps that I've taken...	02:32
	← ruphy left	02:33
	→ bieneff joined	02:34
	→ ruphy joined	02:34
jnl_	fynn: git ls-files --others ?	02:34
ben_h	git ls-files -o -v	02:34
	is one way	02:34
	brettryan: always at least as good as in, git add -u .	02:34
	→ dkr_ joined	02:35
fynn	jnl_, ben_h: thanks, that's it.	02:35
	← dkr left	02:36
	← bobmcw left	02:39
brettryan	That git ls-files is interesting	02:41
	how do you get it to ignore files that are in .gitignore?	02:41
	werdnus → werdna	02:41
	← eternaleye left	02:42
brettryan	`git ls-files -o -v -X .gitignore`?	02:43
werdna	--standard-excludes	02:43
	andrew@gcentral_router:~$ git ls-files --help	02:43
aFlag	it doesn't list the ignored files here	02:43
werdna	--exclude-standard	02:43
	Add the standard git exclusions: .git/info/exclude, .gitignore in each directory, and the user´s global exclusion file.	02:43
brettryan	ahh, I didn't see that in the man page :)	02:44
	sorry, I should open my eyes wider.	02:44
	← tcoppi left	02:46
	→ tcoppi joined	02:46
ben_h	there are also some system-dependent files that are implicitly excluded like .DS_Store on OS X	02:46
	that --standard-exclude is required for.	02:47
	← tcoppi left	02:47
	→ Cowmoo joined	02:48
	← bdiego left	02:49
brettryan	Thanks guys, you're all so helpfull :)	02:50
Cowmoo	say I have a remote repo in the dir remotehost:/blah	02:50
aFlag	hm, just downloaded jdk1.6.0_10 and I ran jdk1.6.0_10/bin/jconsole and, once again, only a blank screen shows up :-(	02:50
Cowmoo	I want to clone into my local home dir, and want all the files to be in /home, not /home/blah	02:51
	so then I'd push say /home/file, and it would end up in remotehost:/blah/file	02:51
	is that possible? As executing 'git clone remotedir . ' from my local homedir gives me an error saying '.' already exists (obviously)	02:51
	← ruphy left	02:52
	← ben_h left	02:52
	← LiamH left	02:52
	← cbreak left	02:52
	← markelikalderon left	02:52
	← precision left	02:52
	← tjafk2 left	02:52
	← sugi left	02:52
	← Thumper__ left	02:52
	← GarethAdams left	02:52
	← RandalSchwartz left	02:52
	← lamont left	02:52
	← AlexanderS left	02:52
	← resmo left	02:52
	← jmspeex left	02:52
	← jackdempsey left	02:52
	← cardioid left	02:52
	← Jacolyte left	02:52
	← joshdavey left	02:52
	← chrizz_ left	02:52
	← mpr left	02:52
	← reuss left	02:52
	← glommer left	02:52
	← madewokherd left	02:52
Cowmoo	test	02:52
	→ tcoppi joined	02:53
	← brettryan left	02:54
	→ chrizz joined	02:54
	→ ruphy joined	02:56
	→ ben_h joined	02:56
	→ Thumper__ joined	02:56
	→ markelikalderon joined	02:56
	→ glommer joined	02:56
	→ cbreak joined	02:56
	→ madewokherd joined	02:56
	→ LiamH joined	02:56
	→ reuss joined	02:56
	→ joshdavey joined	02:56
	→ tjafk2 joined	02:56
	→ Jacolyte joined	02:56
	→ mpr joined	02:56
	→ chrizz_ joined	02:56
	→ sugi joined	02:56
	→ resmo joined	02:56
	→ lamont joined	02:56
	→ jackdempsey joined	02:56
	→ GarethAdams joined	02:56
	→ jmspeex joined	02:56
	→ RandalSchwartz joined	02:56
	→ precision joined	02:56
	→ cardioid joined	02:56
	→ AlexanderS joined	02:56
	← chrizz_ left	02:56
	← Thumper__ left	02:56
	→ Thumper_ joined	02:56
	← Jacolyte left	02:57
	← joshdavey left	02:57
	← mpr left	02:57
fynn	is there an easy way to make it so that all 'git diff' operations have --color by default?	02:57
	→ Jacolyte joined	02:58
	→ joshdavey joined	02:58
	← sgrimm left	02:58
	→ sgrimm joined	02:59
mugwump_	fynn: see man git-config	02:59
Gitbot	fynn: the git-config manpage can be found at http://git.or.cz/man/git-config	02:59
	→ Rh0nda joined	02:59
fynn	mugwump_: thanks	03:00
	← Rhonda left	03:00
fynn	mugwump_: hmm	03:03
	is there a way to enable color for _all_ commands that support the --color switch?	03:03
	ah, the .ui switch	03:07
	found it.	03:07
	← foca left	03:09
	← jefferai left	03:21
	← ahupp\|fb left	03:22
	→ aFlag_ joined	03:23
	← aFlag_ left	03:23
	→ aFlag_ joined	03:23
	← aFlag left	03:23
	aFlag_ → aFlag	03:24
	→ jefferai joined	03:25
	→ qrush joined	03:31
	← qrush left	03:31
	→ piclez_ joined	03:37
	→ nice_burger joined	03:38
	← rtomayko_ left	03:39
	→ rtomayko joined	03:43
	→ spearce joined	03:50
	→ mithro joined	03:54
	← rtomayko left	03:58
	→ rtomayko joined	03:58
	→ qrush joined	03:59
	← ppicazo left	04:00
	← Arafangion left	04:02
	← spearce left	04:02
	→ elmex_ joined	04:02
	→ dcolish joined	04:06
	← aFlag left	04:09
	← tarbo left	04:10
fynn	I like git.	04:10
	is nice.	04:10
nice_burger	me too	04:10
offby1	isn't.	04:10
nice_burger	is	04:10
	→ spearce joined	04:12
	← elmex left	04:16
	elmex_ → elmex	04:16
	← spearce left	04:18
	← LiamH left	04:20
	→ levander joined	04:23
levander	Why bother to setup git-daemon or http and not just use ssh for a bare public repository?	04:23
jayne	because then you have to create users and such	04:25
levander	ah	04:25
	← luke-jr left	04:25
levander	thanks	04:25
	← bieneff left	04:31
	→ Arafangion joined	04:32
	→ loincloth joined	04:33
	→ tarbo joined	04:36
	← loinclot_ left	04:38
	→ cartucho joined	04:45
	← cartucho left	04:45
	→ DrFrasierCrane joined	04:49
	← rtomayko left	04:49
	→ rtomayko joined	04:49
	→ robinr joined	04:50
	← quoin left	04:51
	← markelikalderon left	04:53
	→ quoin joined	04:53
	→ loinclot_ joined	04:58
	← Cowmoo left	04:58
	← loincloth left	05:02
	→ psilo2 joined	05:03
	→ ahupp\|fb joined	05:05
	→ quoin- joined	05:08
	← quoin left	05:12
	← madewokherd left	05:16
	→ dreiss joined	05:23
	← ashleyw left	05:23
	← paltman left	05:26
	← tjafk2 left	05:27
	→ tjafk2 joined	05:27
	→ markelikalderon joined	05:31
	→ tewk_ joined	05:39
	→ aspotashev joined	05:44
	← stuffcorpse left	05:44
	← dcolish left	05:45
	← tewk left	05:45
	→ stuffcorpse joined	05:45
	→ the_gittr joined	05:48
	← rtomayko left	05:49
	→ scientes joined	05:54
	→ dcolish joined	05:54
	→ muthu joined	05:55
	← dcolish left	05:56
	→ scientes_ joined	05:57
	← muthu left	05:57
	← carllerche left	05:58
	← markelikalderon left	05:59
	← nice_burger left	06:04
	→ dcolish joined	06:12
	← tcoppi left	06:12
	→ tcoppi joined	06:14
	← robinr left	06:17
	→ jackdempsey_ joined	06:18
	← aspotashev left	06:19
	→ markelikalderon joined	06:19
	← dcolish left	06:21
	← piclez_ left	06:24
ben_h	PICs	06:33
	as in the microcontrollers	06:34
	← jm left	06:36
	→ towski joined	06:41
	← the_gittr left	06:43
	← dreiss left	06:44
	→ litage joined	06:44
	← jackdempsey_ left	06:45
	← xjjk left	06:46
	→ xjjk joined	06:46
	← warthog9 left	06:50
	← markelikalderon left	06:52
	→ markelikalderon joined	06:53
	← scientes_ left	06:56
	→ warthog9 joined	07:05
	← bobesponja left	07:09
	← abbe left	07:10
	← qrush left	07:11
	→ Marmouset joined	07:11
	→ osbert joined	07:11
	→ digitalhobbit joined	07:19
	← selckin left	07:29
	→ tvw joined	07:31
	→ foca joined	07:34
	→ bwwww joined	07:40
ben_h	anyone here today?	07:44
	→ yaroslav joined	07:47
	→ d00dlez joined	07:47
yaroslav	Is there any way to rename a tag if I already pushed?	07:48
	* make that "delete a tag on remote"	07:50
	← litage left	07:52
Tv	yaroslav: git push theremote :thetag	07:53
yaroslav	Tv: big thanks	07:54
	→ cilly joined	07:56
	← stuffcorpse left	07:56
	→ stuffcorpse joined	07:57
	← G left	07:58
	← d00dlez left	08:04
	→ d00dlez joined	08:04
	← foca left	08:06
	→ ft joined	08:12
	← markelikalderon left	08:12
	← d00dlez left	08:13
	→ d00dlez joined	08:14
	← cilly left	08:16
	→ daniel-w_ joined	08:19
	← d00dlez left	08:23
	→ d00dlez joined	08:24
	← daniel-w left	08:31
	← jefferai left	08:32
	← pygi left	08:32
	→ foca joined	08:33
	→ kumbayo joined	08:34
	→ jefferai joined	08:35
	← d00dlez left	08:42
	→ Weaselweb joined	08:42
	→ d00dlez joined	08:44
	→ ferai joined	08:47
	← digitalhobbit left	08:52
	→ thrope joined	08:54
	← Jacolyte left	08:54
	→ Jacolyte joined	08:54
	← jefferai left	09:00
	← Tv left	09:07
	← osbert left	09:07
	→ cilly joined	09:09
	← bwwww left	09:22
	→ ppicazo joined	09:24
	← yaroslav left	09:29
	← ben_h left	09:32
	→ ben_h joined	09:33
	← thrope left	09:44
	→ clairvy joined	09:54
	← cilly left	09:58
	→ ph^ joined	09:58
	→ loiseau joined	09:59
	→ joshdavey_ joined	10:00
	→ priidu joined	10:01
	← towski left	10:01
	→ pcapriotti joined	10:02
	← joshdavey left	10:02
	→ pygi joined	10:06
	→ hidek_ joined	10:07
	← hidek left	10:21
	← cedric_ left	10:25
	→ cedric_ joined	10:28
	→ DavidMJ joined	10:30
DavidMJ	what is the best way to create a git repository and allow some close friends to push while have a global pull?	10:31
pygi	gitosis	10:32
	I guess ... :)	10:32
	→ Yuuhi joined	10:32
Arafangion	Wouldn't file permissions be sufficient?	10:35
	→ thrope joined	10:37
	← Marmouset left	10:39
	← foca left	10:42
	→ parasti joined	10:50
	← ppicazo left	10:52
	← thrope left	10:53
	← CodeOfficer left	10:56
	→ david_koontz joined	10:56
	→ drizzd joined	10:57
	→ charon joined	11:00
Ilari	Arafangion: Yes, it would. Set shared flag to world (group read/write, world read), set git daemon export flag and start git daemon.	11:01
	← godlygeek left	11:05
	→ thrope joined	11:06
	→ drizzd_ joined	11:08
	→ not-xjjk joined	11:08
	→ SRabbelier joined	11:10
	← SRabbelier left	11:11
	→ SRabbelier joined	11:12
	← drizzd left	11:14
	← xjjk left	11:22
	→ chris2 joined	11:22
	← DavidMJ left	11:35
	→ Zyroth joined	11:35
Zyroth	I am trying to use git-instaweb, but it tells me that it cannot bind to port 80 - although the default seems to be port 1234. There is a different web service running on that port, so port 80 is not free. Any ideas?	11:36
drizzd_	have you tried with --port=1234 ?	11:36
Pieter	or used a different web engine?	11:37
drizzd_	the default can be overwritten using git config instaweb.port	11:37
	← thrope left	11:38
	→ thrope_ joined	11:38
	→ w00t_ joined	11:38
werdna	hey w00t_	11:38
	← thrope_ left	11:46
	← Aphelion left	11:50
	← w00t left	11:51
	← pygi left	11:54
	← nud left	11:57
	→ nud joined	11:57
	→ gour joined	12:07
gour	vmiklos: hi, i'd like to use LP for project hosting and use darcs for tracking. do you think that your darcs-fast-export could be used to 'sync' local darcs branches and provide public bzr repo at LP?	12:10
wereHamster	Zyroth: only root can bind to port 80. Normal users have only permissions to bind to port >1024	12:10
	← charon left	12:11
	← david_koontz left	12:12
	→ thrope joined	12:13
	← thrope left	12:13
	← clairvy left	12:15
	← d00dlez left	12:20
	→ vbgunz_ joined	12:21
	→ name joined	12:28
	← vbgunz__ left	12:34
	← Flathead left	12:38
	→ jnareb joined	12:42
	← Dreamer3 left	12:44
jnareb	pasky: Could you please review my 'gitweb: Better processing format string in custom links in navbar' patch in response to your latest gitweb patches (if you have time)? TIA.	12:45
	← loiseau left	12:50
jnareb	So it looks like GitTogether'08 was a success, isn't it?	12:50
drizzd_	I'm sure it was enjoyable at least	12:53
	→ Voker57 joined	13:00
Ilari	Regarding that Git and media files thing, there are security issues in adding a facility for requesting blobs from repo...	13:02
	→ charon joined	13:03
drizzd_	Ilari: you mean you could dig for sha1s?	13:03
Ilari	drizzd_: Yup...	13:04
vmiklos	gour: at the moment darcs-fast-export does not support incremental conversion, so probably no.	13:04
drizzd_	so the server would have to verify that the sha1 is reachable from the published refs	13:05
	OTOH, it can only show what had been pushed an therefore published before	13:05
Ilari	drizzd_: If client suppiled base commit and path, it would only need to show reachability in commit graph, not in full object graph...	13:06
drizzd_	true. that would not even add a lot of overhead.	13:07
	If I understand correctly, it's currently possible to do either shallow (no history) or narrow (only subtree) checkouts with git, but not both simultaneously?	13:08
	→ d00dlez joined	13:08
Ilari	drizzd_: Its shallow fetching/narrow checkout. Remember while SVN couples the two, in git they are seperate operations.	13:09
jnareb	drizzd_: currently you can do shallow clone (cutting history), and that clone is a bit limited (but perhaps that will change)	13:09
drizzd_	right, so no "narrow clones"	13:09
jnareb	drizzd_: sparse checkout (or narrow/subtree checkout) is currently Work In Progress by pclouds	13:09
	drizzd_: and I think those two features are totally orthogonal: one affects repository database, the other affects working directory and index	13:10
drizzd_	for the media files thing, the "narrow database" is the more important thing, because it's potentially huge.	13:11
jnareb	drizzd_: so it should be possible to have shallow clone with sparse checkout	13:11
	→ markelikalderon joined	13:11
jnareb	drizzd_: currently you can workaround using submodules (to be able to not have them in main repo database, and not checked out)	13:11
	drizzd_: as to security of 'request for blob' that 'lazy clone' support would require: first, if you have gitweb enabled you can now access arbitrary sha1; second, support for fetching arbitrary sha1 (limited to blobs) can be required to have been enabled as extra feature;	13:13
Ilari	Another problem with media files is that some media files are inheritly undeltifiable but Git still tries to delta them, wasting much time for nothing...	13:14
	→ imyousuf joined	13:14
jnareb	Ilari: see undocumented `delta' gitattribute	13:14
	→ aspotashev joined	13:15
jnareb	Ilari: commit a74db82e15cd8a2c53a4a83e9a36dc7bf7a4c750	13:17
Gitbot	[git a74db82e1]: http://tinyurl.com/635w4h -- Teach "delta" attribute to pack-objects.	13:17
ben_h	anyone know a git command that can tell if a repo contains any commits or not, and exits with success either way?	13:17
	git log, git show, git status all exit with status 1 on empty.	13:18
LotR	git status \|\| true :)	13:18
jnareb	git show-ref --verify HEAD, perhaps?	13:18
	or "git rev-parse --verify HEAD 2>/dev/null"?	13:19
ben_h	hmm	13:19
	--verify --quiet looks good	13:19
doener	ben_h: define "contains any commits". You can have a bunch of commits, but no ref that points to them	13:20
ben_h	doener: in this case, it'll be a freshly inited repo	13:20
doener	ben_h: do you need to catch the "new unconnected branch" case? That is, right after a "git symbolic-ref HEAD refs/heads/new_branch"?	13:21
ben_h	sorry, you've lost me there	13:22
	:P)	13:22
	- P	13:22
doener	ben_h: after "git init", HEAD is a symbolic ref to refs/heads/master (which does not exist yet). Therefore, git rev-parse --verify HEAD fails	13:23
ben_h	what i'm doing is parsing the time since the last commit by calling	13:23
	git log -1 --pretty=format:%at	13:23
	and using that as input to Time.at in ruby.	13:23
	doener: ahh ok.	13:23
	→ koke joined	13:23
doener	ben_h: you can recreate that situation later using "git symbolic-ref HEAD refs/heads/new_branch". The next "git commit" will then create that new branch, and the commit won't have a parent.	13:23
ben_h	hm	13:23
	so that commit's diff would be the full contents of the files?	13:24
doener	but in that case, you can of course still have a lot of commits. They're just not reachable through HEAD (but probably through some other ref)	13:24
	ben_h: yep. It's just yet another root commit	13:24
	← reuss left	13:24
ben_h	right	13:24
jnareb	ben_h: diff --root would be (unless you configured diff.showroot)	13:24
	ben_h: just 2 eurocents	13:25
ben_h	ok, so the general case, which covers my original case and that one, is the HEAD not referencing a valid commit.	13:25
doener	ben_h: ok, so you're basically just interested in the author time of the latest commit on the current branch, right?	13:26
	← koke left	13:26
ben_h	yep that's exactly what i want	13:26
	or if it doesn't exist, the epoch	13:26
jnareb	ben_h: perhaps git-for-each-ref would be a better solution here than git-show or git-log? you can get all branches with git-for-each-ref...	13:27
	→ reuss joined	13:27
ben_h	hmm	13:28
	something like	13:29
	git for-each-ref --count=1 --sort -authordate	13:29
	?	13:29
jnareb	ben_h: that would find most recent ref, and not show HEAD time	13:30
ben_h	HEAD time?	13:30
	← markelikalderon left	13:30
Pieter	Mikachu: can you update the topic? :)	13:30
jnareb	ben_h: probably with some --format, and with --heads, or refs/heads/, or HEAD (see man git-for-each-ref)	13:30
Gitbot	ben_h: the git-for-each-ref manpage can be found at http://git.or.cz/man/git-for-each-ref	13:30
jnareb	authordate from current branch, as opposed to most recent authordate from all branches	13:31
	→ lucsky joined	13:32
jnareb	ahhh... 1.6.0.4 just tagged (but not announced on git mailing list, unless GMane has hiccups again)	13:32
	← reuss left	13:32
ben_h	Gitbot: maybe a post-commit hook for /topic is in order ;)	13:32
	→ reuss joined	13:32
	→ flaguy joined	13:32
gour	vmiklos: any plan to add it or i'm better to forget about using LP for hosting branches?	13:34
jnareb	ben_h, Pieter: but then Gitbot would have to have some of op rights (at least to change /topic)	13:34
ben_h	true	13:34
Pieter	jnareb: yeah, I'm fine with Mikachu or pasky just updating the topic :)	13:34
vmiklos	gour: do you work for darcs1 or darcs2 repos?	13:35
ben_h	nice. this seems to work jnareb	13:35
	git for-each-ref --count=1 --sort -authordate --format=%$authordate$	13:35
	but i don't see what you mean about current vs all branches sorry	13:35
vmiklos	gour: adding support for the darcs2 format is non-trivial, adding incremental support is probably easy (i just did not need it)	13:35
gour	vmiklos: i'd use darcs-2 cause it fixes some old problems (which brought me to bzr...)	13:36
vmiklos	gour: you mean the format? i don't mean the program version	13:36
gour	vmiklos: yes, format	13:37
jnareb	ben_h: 'git show --pretty="format:%at" HEAD' would always show authordate from current branch	13:37
	ChanServ set mode: +o	13:38
	Mikachu changed the topic to: 1.6.0.4 \| Homepage: git.or.cz \| Everyone asleep or clueless? Try [email@hidden.address] \| Channel log http://tinyurl.com/gitlog \| Mailing list archives: http://tinyurl.com/gitml \| Gits on git: http://tinyurl.com/gittalks \| Pastebin: http://gist.github.com/ \| pasky: "It can't be slow, its a shell script!"	13:38
	Mikachu set mode: -o	13:38
ben_h	jnareb: ahh, but for-each-ref is across all	13:38
vmiklos	gour: ok, then probably it's easier to try to use something else for now. or if you know python, you can try adding support for it, i would need to read a lot more docs on the hashed format to do so ;/	13:38
jnareb	ben_h: 'git for-each-ref --count=1 --sort=-authordate --format="%(authordate)" refs/heads' would show most recent authordate from all branches, not necessarily from current branch (there might be branch with newer commits)	13:38
ben_h	yeah	13:39
vmiklos	gour: or you can use a darcs->git converter and host it on repo.or.cz :)	13:39
	gour: see the 'Darcs + other VCS' section of http://wiki.darcs.net/index.html/RelatedSoftware	13:40
	(i would especially try git-darcs-import)	13:40
Ilari	'git for-each-ref --format=%$objecttype$\ %$*objecttype$ \| grep "commit" \| wc -l' :->	13:41
	jnareb wonders if tree walking refactoring required for packv4 would also help --follow to work not only on simplest histories, and work with --max-count (currently git log --follow --max-count=n doesn't work	13:41
nud	packv4 ?	13:42
	nud thought we were still at v2	13:42
jnareb	vmiklos: do you remember if git-darcs-import is mentioned on http://git.or.cz/gitwiki/InterfacesFrontendsAndTools in conversion section?	13:42
	→ chenk85 joined	13:43
gour	vmiklos: well, LP provides much more than tracker, otherwise i can host darcs branches directly (maybe at haskell.org)	13:43
jnareb	nud: future / planned packv4 (which was discussed on GitTogether'08)	13:43
Ilari	nud: There's v3 as well, which is little different from v2... Don't know if git tools produce packs that are not in the common subset of v2 and v3.	13:43
nud	jnareb: do you have a pointer ?	13:43
jnareb	nud: like with IPv4 and IPv6... ;-)	13:43
nud	I mean, an URL with some documentation :-)	13:44
vmiklos	jnareb: i haven't tried it myself, so i would it would be better not to list it till then, i think ;) (but maybe i'm too serious)	13:44
jnareb	nud: to packv4, or to GitTogether?	13:44
nud	jnareb: packv4	13:44
	gittogether seems to have too much material for me to read in the time I have :-)	13:44
Ilari	nud: Neither v2 or v3 is the superset of another...	13:45
	nud: While IPv6 is the superset of IPv4...	13:45
	nud: At least in address space...	13:45
jnareb	nud: there are no slides from packv4 talk on GitTogether'08 by Shawn O. Pearce (spearce) from GitTogether.	13:45
	← zeenix left	13:46
vmiklos	gour: ah, i see. well, if you need it right now, you can still try tailor, and anyway it's on my todo to do these, but i can't promise it'll be ready for tomorrow ;)	13:46
jnareb	Ilari: and I think it is one of reasons why deployment of IPv6 lags (as compared to AAAA -> AAAAAA conversion/deployment)	13:46
Ilari	nud: The difference between those two versions is that bit 6 of delta copy opcode encodes source select bit in v2 but encodes presence of copy size byte 3 in v3.	13:46
jnareb	nud: so you would have to Google or search mailing list archives (marc, gmane) for packv4...	13:47
	Ilari: ???	13:47
gour	vmiklos: no rush...i was thinking to just use bzr, but seeing fresh interest for darcs, my old love is re-awaken ;)	13:47
jnareb	See you later...	13:47
	→ DaveCass joined	13:47
vmiklos	gour: it's priority is now a bit higher, given that i know someone would actually use these features:)	13:48
DaveCass	Hey everyone, does anybody have any idea why I can't connect to git.kernel.org? (fatal: The remote end hung up unexpectedly)	13:49
Ilari	nud: So if no delta copy opcode has bit 6 set, the pack will be interpretted the same in both versions...	13:49
	DaveCass: Wrong repository URL?	13:49
DaveCass	tripple checked :(	13:49
gour	vmiklos: thanks...how are you satisfied with git? (i still resist to go into it seeing it as too complex, at least, it was when i was evaluating it many months ago...)	13:49
Ilari	DaveCass: What URL are you using?	13:50
	DaveCass: Or is it some SSH url?	13:50
jnareb	gour: git UI improved significantly, but take a look a few months since, when 'use cases' based UI planned on GitTogether (as opposed to current 'do this' based UI) would get implemented	13:51
	...probably	13:51
nud	I've found emails about packv4	13:51
	from Feb 2007...	13:51
DaveCass	"git clone git://git.kernel.org/pub/scm/ linux/"	13:51
vmiklos	gour: where i was really not comportable with it, i sent patches :)	13:52
jnareb	DaveCass: certainly wrong URL	13:52
Ilari	DaveCass: Maybe "git clone git://git.kernel.org/pub/scm/linux"?	13:52
DaveCass	*"git clone git://git.kernel.org/pub/scm/ linux/kernel/git/avi/kvm-userspace.git"	13:52
	my bad	13:52
Ilari	DaveCass: Why is that space there?	13:52
DaveCass	0_0	13:53
vmiklos	gour: anyway, i wrote a darcs-like porcelain on top of git plumbing, it may help you till you don't understand the details of git. search for "darcs-git" (case-sensitive) at http://git.or.cz/gitwiki/InterfacesFrontendsAndTools	13:53
DaveCass	thanks Ilari... i need a break i guess	13:53
werdna	I love it and I started using it a week or so ago.	13:54
	← werdna left	13:54
	← DaveCass left	13:54
	→ chris2_ joined	13:54
	← r0bby left	13:54
jnareb	nud: see also http://repo.or.cz/w/git/packv4.git but I see it doesn't have any new commits also...	13:55
	→ bcarlyon\|laptop joined	13:56
nud	jnareb: yep I already saw this one	13:56
	too bad all the GT08 slides are not available	13:56
jnareb	yep	13:57
	→ werdnus joined	13:58
gour	vmiklos: darcs-git looks interesting	14:02
	→ lemonodor joined	14:03
	← lemonodor left	14:04
jnareb	vmiklos: how darcs-git compares speed-wise/performance-wise with ordinary Darcs (or with Git)?	14:04
vmiklos	it's a thin wrapper, so it's the same as git	14:05
	it's like darcs what -s -> git diff HEAD -M --name-status and such	14:06
jnareb	vmiklos: but if it uses patch commutation for merge like darcs... or does it implement subset of darcs UI?	14:06
vmiklos	or darcs changes --last=2 -> git log -2, etc.	14:06
	← chris2 left	14:06
vmiklos	yes, it's pretty much just about the ui	14:06
	← Simetrical left	14:07
vmiklos	and especially since there is git add -p.. :)	14:07
	ok, in git there is no opposite of git add -p, while darcs-git has 'dg rec' (like git add -p) and the opposite of it, 'dg rev'	14:08
	→ robinr joined	14:08
	→ markelikalderon joined	14:09
	bcarlyon\|laptop → BarryCarlyon	14:09
	← ferai left	14:09
	→ jefferai joined	14:10
	← Zyroth left	14:11
	gour is trying darcs-git add...	14:11
	→ nice_burger joined	14:12
Ilari	There exists delta data that's valid in both v2 and v3 but interpretted differently. For example '01 00 D0 01 80'.	14:13
	→ rmh3093 joined	14:13
Ilari	Oops, that likely doesn't work...	14:13
	Yup, doesn't work...	14:13
	← jaalto left	14:16
gour	vmiklos: hmm, darcs-git is dangerous tool..it might attract darcs user to the weird land of git and then one is lost ;)	14:16
	i might resist playing too much with it...	14:16
	→ thrope joined	14:17
vmiklos	:)	14:18
	→ Simetrical joined	14:20
jnareb	vmiklos: IIRC interactive unstaging is planned for git (and would probably go well together with requested git-unstage command)	14:23
	→ jm joined	14:23
	→ cilly joined	14:31
	→ rubydiamond joined	14:34
	← ben_h left	14:36
	← rubydiamond left	14:36
	← Ilari left	14:37
	← thrope left	14:38
	← imyousuf left	14:41
	→ rubydiamond joined	14:42
	→ chris2 joined	14:43
	→ thrope joined	14:44
	← chris2 left	14:46
	← chris2_ left	14:48
	→ madewokherd joined	14:49
xenoterracide	I need to merge 2 branches... the problem is that git thinks that branch b is a fast-forward of a. (merging b into a) however I can't deal with it like that because it really isn't.	14:51
jnareb	if git thinks it is, it probably is	14:52
thiago_home	in other words: why do you think git is wrong?	14:52
Mikachu	if you really want a merge commit, you can say --no-ff	14:52
jnareb	you can force creating merge commit anyway with --no-ff (--ff=never) option	14:52
xenoterracide	git is wrong because the file was added from outside the tree	14:52
thiago_home	"the file"?	14:52
	which file?	14:52
xenoterracide	I only changed one file in the other branch	14:53
jnareb	added, but was this addition comitted?	14:53
xenoterracide	yeah	14:53
thiago_home	do you see the change in git log?	14:53
xenoterracide	yeah	14:53
	but I haven't done the merge yet	14:54
thiago_home	how about in the other branch? Do you see the change as well?	14:54
jnareb	there is simply no way for git to be wrong wrt to fast forward. Either comit is in ancestry of other commit, or it isn't	14:54
xenoterracide	um... no	14:54
jnareb	this doesn't check what is different, only parentage	14:54
	→ chris2 joined	14:54
thiago_home	then we have two options: either you're missing the change that is there, or you're merging the wrong thing	14:54
jnareb	see for example result of "git merge-base --all a b"	14:54
xenoterracide	well with git it is ancestry... and there is some ansesty between msys's msys.bat and msysgit's msys.bat if you go back far enough... but I need to merge them so I get only the changes I want in the main tree	14:55
thiago_home	commit ancestry	14:56
	not file ancestry	14:56
xenoterracide	?	14:57
thiago_home	what we said	14:57
	git is not wrong	14:57
xenoterracide	basically I need to have the merge resolve in a conflict so I can fix it	14:57
doener	xenoterracide: fast-forward from a to b on "git merge b" happens when there are no changes in a that are not in b	14:57
Mikachu	xenoterracide: look at gitk a...b	14:57
thiago_home	why do you need to have a merge conflict?	14:58
Mikachu	it will just be a straight line	14:58
jnareb	"git merge b" when on branch 'a' says 'fast-forward' when commit 'a' is in ancestry of 'b' (or vice-versa, but that is more rare), i.e. if history didn't diverge at all	14:58
doener	xenoterracide: ie. there is simply nothing to "merge", you just have a few new commits on top of a	14:58
jnareb	xenoterracide: or "git log --graph --left-right a...b"	14:58
	→ Dreamer3 joined	14:58
Dreamer3	can i make a totally empty new branch?	14:58
doener	jnareb: in the vice-versa case, wouldn't it say "already up-to-date"?	14:58
Mikachu	Dreamer3: yes	14:59
Dreamer3	how?	14:59
doener	jnareb: would be a fast-backwards otherwise ;-)	14:59
Mikachu	Dreamer3: rm .git/index; git symbolic-ref HEAD refs/heads/newbranch	14:59
	Dreamer3: then add something and commit it	14:59
Dreamer3	interesting	14:59
	that's a bit of a hack, no?	14:59
thiago_home	Dreamer3: yes	14:59
Mikachu	well, you can use git rm -r instead of rm .git/index	15:00
thiago_home	the "clean way" is to go to a new dir and do git init all over again	15:00
	→ jackdempsey_ joined	15:00
Dreamer3	that wouldn't be a branch though, right?	15:01
Mikachu	it would be after you fetch it into the first repo	15:01
thiago_home	once you've done one commit, you can pull it to a branch	15:01
jnareb	and then pull from that repo (and delete temporary repo)	15:01
Mikachu	you don't want to pull it, just fetch	15:01
jnareb	that is how 'todo' branch and gitweb made it into git.git	15:01
Dreamer3	hmmmm	15:02
jnareb	fetch if you want independent unrelated branch	15:02
Mikachu	i think they were actually pushed, not fetched :)	15:02
jnareb	same thing :-)	15:02
Mikachu	Dreamer3: the way i said is easiest, that's why i said it :)	15:02
jnareb	but creating independent branch / additional root is not something for newbies (well, at least intentional creation of new root, not just joining two projects)	15:05
	← chris2 left	15:08
	← xkr47 left	15:10
	→ df00z joined	15:13
df00z	Hey, im in a branch called origin/for-review, how can I reset it to what is uploaded upstream?	15:13
	i changes some files	15:13
	i want to lose my changesw	15:14
jnareb	df00z: you could not have changed origin/for-review, as it is remote-tracking branch	15:14
	→ idletask joined	15:14
idletask	Hello	15:14
jnareb	are you on detached HEAD, or on tracking branch for-review?	15:14
wereHamster	see `git branch` and look for the branch with an asterisk	15:15
Mikachu	(jnareb: you can have a local branch called origin/for-review, but it is very confusing)	15:15
df00z	origin/for-review	15:15
idletask	Does anyone here know egit internals?	15:15
	← chenk85 left	15:15
jnareb	if the latter, "git reset --hard origin/for-review" when on for-review	15:15
idletask	I'd like to add stash support to the plugin, but don't know where to start	15:15
jnareb	df00z: but man git-reset anyway	15:15
Gitbot	jnareb: the git-reset manpage can be found at http://git.or.cz/man/git-reset	15:15
df00z	jnareb: when i originally switched branches, did I do it wrong? git checkout -b origin/for-review origin/for-review	15:16
jnareb	not wrong, but a bit confusing	15:16
	git reset --hard remotes/origin/for-review	15:16
	when on branch origin/for-review (git checkout origin/for-review)	15:17
	and it should be "git checkout --track origin/for-review" in cutting edge git ;-)	15:17
df00z	To sync I should be able to do a git-pull right	15:17
	if I want to get all the latest changes in the git repository	15:18
	i can do it if i just have head	15:18
jnareb	pull is fetch+merge (or fetch+rebase)	15:18
wereHamster	origin/foo usually points to remote branches. Local branches are named just 'foo' (as in git-checkout -b foo origin/foo)	15:18
jnareb	fetch only downloads latest changes, and updates remote-tracking branches	15:18
	pull also merges	15:18
	→ jelmer joined	15:20
df00z	git pull drm-intel/for-review :\	15:21
	http://git.kernel.org/?p=linux/kernel/git/anholt/drm-intel.git;a=summary	15:21
	What's the proper way to grab for-review, and sync changes	15:21
Mikachu	df00z: s,/, ,	15:21
df00z	er, lose changes	15:21
	i want to just stay in sync with the tree	15:21
Mikachu	df00z: git fetch drm-intel; git reset --hard drm-intel/for-review	15:21
	← jelmer left	15:23
df00z	git pull git://git.kernel.org/pub/scm/linux/kernel/git/anholt/drm-intel.git	15:23
jnareb	df00z: git fetch drm-intel; git reset --hard remotes/drm-intel/for-review (might be needed with your naming)	15:23
df00z	err git clone	15:23
	then i need to switch to for-review	15:24
	← jnareb left	15:24
wereHamster	git checkout -b for-review origin/for-review	15:24
df00z	and then to update? git pull for-review?	15:24
wereHamster	then when you want to get the latest changes: git fetch origin; git reset --hard origin/for-review	15:25
df00z	ah cool	15:25
	Thanks	15:25
	→ qrush joined	15:25
Mikachu	df00z: as long as you don't have any local changes you should be able to say just git pull	15:25
	df00z: but if you have changes, that will try to merge	15:25
df00z	Do you know of any idiot's guide to GIT?	15:30
	Haha	15:30
	the man pages are a bit too intense for me	15:30
	→ selckin joined	15:30
jackdempsey_	df00z: there're some good tutorials out there	15:30
	i'd just google a bit	15:30
Mikachu	df00z: did you read the introduction to git that comes with git?	15:31
jayne	I think the key is to just ignore whatever you don't need. Otherwise you'll be overwhelmed.	15:31
df00z	No, i'll take a look at the intro	15:31
fynn	OK, to make sure I understand:	15:34
	if I do 'git clone ~/foo --bare', the resulting foo.git is a bare repository, without a working tree.	15:34
	even if I rename it to .git, and try to 'git reset' from it, it wouldn't work.	15:35
	because bare repositories can't have working trees, ever. right?	15:35
wereHamster	see the .git/config file, there's a flag that says whether the repo is bare or not	15:35
	→ nud_ joined	15:35
jayne	bare repositories know they're bare (it's specified in the config)	15:35
fynn	so, they're ashamed of themselves? :>	15:36
	anyways, it means that unless I remove the flag from the config, I can't use it as a regular .git directory.	15:36
	but is that flag the only difference?	15:36
wereHamster	you can't use it to check out the tree from it. But you can still clone/fetch from it	15:37
jayne	well, bare repositories are usually .git instead of having a /.git	15:37
fynn	right, my question is: if I take a bare repository foo.git, rename it to .git, and remove the 'bare' flag from .git/config, is it exactly identical to a non-bare repo?	15:38
jayne	I've never tried it, but I think that would work	15:38
	fynn tries	15:39
fynn	works.	15:39
	thanks wereHamster, jayne.	15:39
	← Yuuhi left	15:41
fynn	OK, another question:	15:42
	suppose I move that .git directory to some new directory	15:43
	the .git directory contains two files: foo.py and bar.py	15:43
	the new directory may contain bar.py. what's the proper procedure to do a 'git reset' in that situation?	15:44
	← gour left	15:44
	← ke4qqq left	15:44
fynn	i.e. the best would be to be prompted for each clash: "file bar.py has local modifications, would you like to overwrite? y/n"	15:44
wereHamster	the .git directory shouldn't contain any files other then some internal git files.	15:44
fynn	wereHamster: no, I mean the original working tree may contain files that are in the new directory.	15:45
wereHamster	well, you started with a bare repo, right?	15:45
fynn	not in this case, but I can if it matters.	15:45
wereHamster	if you want to reset any local modifications, git reset --hard HEAD	15:46
fynn	e.g. the repo is ~/foo/.git, and tracks ~/foo/bar.py. I cp ~/foo/.git to ~/baz, and want to git reset.	15:46
	← cedric_ left	15:46
wereHamster	I sometimes even do: rm -rf *; git reset --hard HEAD; which would wipe any files created my the build process etc	15:46
fynn	wereHamster: right, but what if I'd like to be prompted for each change?	15:46
	some local modifications may be important.	15:47
wereHamster	fynn: see git status if there are any local modifications	15:47
Mikachu	fynn: git add -p	15:47
	← quoin- left	15:47
Mikachu	fynn: then git checkout -- .	15:47
wereHamster	then git checkout HEAD -- bar.py to 'reset' the file bar.py	15:47
Mikachu	(i think, backup first)	15:47
	→ Ilari joined	15:47
fynn	hm, /me tries	15:48
Ilari	Darn net connection...	15:48
	→ chris2 joined	15:48
	→ chenk85 joined	15:52
	← nud left	15:52
fynn	wereHamster, Mikachu: OK, 'add -p' is really cool in a darcs-like way.	15:52
	the only unfortunate thing is that I don't have an 'r' option in the prompt.	15:53
Mikachu	what is r?	15:53
fynn	"reset this file to HEAD".	15:53
Mikachu	fynn: but that's why i gave you the second step :)	15:53
	fynn: just say no, and afterwards you can throw away all un-added changes	15:53
fynn	Mikachu: right, 'add -p' + the checkout thing basically sort me out.	15:54
	Mikachu: thanks. how come you know so much about git?	15:54
Mikachu	fynn: too much free time	15:54
	← schlort left	15:54
fynn	I thought maybe you are a traveler, from a future in which git brought and end to wars and global hunger.	15:54
Mikachu	lol	15:55
	→ _graham_ joined	16:00
	→ jelmer joined	16:02
	← cilly left	16:02
	→ G_SabinoMullane joined	16:02
	← jelmer left	16:04
	← zoke left	16:05
	→ jelmer joined	16:05
	→ schlort joined	16:07
	← rubydiamond left	16:10
	← ia left	16:13
	→ ia joined	16:14
	← Dreamer3 left	16:18
	← Thumper_ left	16:22
	→ Thumper_ joined	16:22
	→ towski joined	16:25
	← doener left	16:29
	← towski left	16:30
	→ cedricv joined	16:32
	→ doener joined	16:33
	→ bieneff joined	16:33
	← Modius_ left	16:35
	→ deskin joined	16:38
rmh3093	is there a way to put a commit message on a tag	16:39
	← warthog9 left	16:40
wereHamster	rmh3093: yes	16:41
	rmh3093: see man git-tag	16:41
Gitbot	wereHamster: the git-tag manpage can be found at http://git.or.cz/man/git-tag	16:41
	→ jerbear joined	16:41
rmh3093	thank	16:41
	s	16:42
	← jas4711 left	16:44
	← jerbear left	16:46
Ilari	rmh3093: True tag objects always have freeform message field...	16:46
	rmh3093: Basically, you need either to make 'annotated' or 'signed' tag...	16:47
rmh3093	ok	16:48
	so 'git tag -a -F <tagmessagefile> [tag]'	16:48
Ilari	rmh3093: You can addionally put object ID of object (usually commit) to tag after that tag name...	16:49
	rmh3093: If you omit it, it defaults to 'HEAD'.	16:49
rmh3093	awesome	16:49
Ilari	rmh3093: Anything with object ID can be tagged.	16:49
rmh3093	thanks	16:50
	→ litage joined	16:50
	→ ashleyw joined	16:53
	← schlort left	16:53
	← unreal left	16:53
	→ Dreamer3 joined	16:54
rmh3093	Ilari, now what if I wanted to view only the commit message of a commit	16:54
	is there a command for that	16:54
fynn	what's the proper use-case for setting excludes in .git/info/exclude?	16:55
	← Dreamer3 left	16:56
drizzd_	fynn: stuff that is not generated but you want to ignore because you have it in your clone	16:56
deskin	fynn: there are a lot of them; for one, editor temporary files (~, .swp) which correspond to your $EDITOR of choice but don't belong in an in-project .gitignore for everyone	16:56
	→ unreal joined	16:56
fynn	deskin: in that case, wouldn't you set it up in your core.excludesfile?	16:57
	→ ppicazo joined	16:58
fynn	very likely, you want to ignore .swp on all your projects, not just one in particular...	16:58
deskin	fynn: that's fine too	16:58
drizzd_	fynn: if you don't have a use-case, why do you care? I certainly need it.	16:58
fynn	yeah, basically, .git/info/exclude is for project-specific excludes, that are not global across users.	16:58
	unlike .gitignore, which is global across users, and core.excludefile, which is global across projects.	16:59
deskin	or in my case, using git-svn I want to ignore certain helper scripts around that project, but a .gitignore is the wrong place because I'm not going to add that file into the SVN repo itself	16:59
fynn	drizzd_: I'm sure it has uses, though they may be rare. what's your use-case?	16:59
drizzd_	I have configuration files adapted to my use. Strictly speaking, I should have them in different directories, but I'm too lazy to change directories all the time.	17:02
	and I have a 'myTODO' in git.git	17:03
	← bieneff left	17:03
drizzd_	and I ignore /*.patch	17:03
	→ koke joined	17:04
	→ bieneff joined	17:05
	← jackdempsey_ left	17:05
	→ Zyroth joined	17:05
	← bieneff left	17:06
fynn	OK, another thing I don't understand:	17:08
	in git, revision IDs are long-ass SHA sums, instead of ordinal numbers.	17:09
	so how do you refer to specific revisions/changesets?	17:09
	(since you can't use r123 or c123)	17:09
	→ noz joined	17:10
	→ jackdempsey_ joined	17:11
wereHamster	fynn: using the sha1 hashes	17:11
idletask	fynn: you have the option to only use _part_ of an SHA1	17:11
	There are many, many other forms too	17:11
Ilari	fynn: Revision ID of commit is the object ID of the commit object.	17:11
idletask	http://git.or.cz/gitwiki/GitCheatSheet <-- here's a list	17:12
Mikachu	fynn: if you and i both clone the same repo and make a commit each, they can't both be r122 if the original repo was r121, and there's no way to synchronize them, so that's why git doesn't use numbers	17:12
	→ LiamH joined	17:13
fynn	Mikachu: hm, that actually makes sense!	17:13
	idletask: thanks for the link, though I don't see any specific reference to what you can use as an ID on that sheet...	17:14
	Ilari: and the object ID of the commit object is that SHA sum?	17:14
Ilari	fynn: And those revision numbers are globally signaficant. commit 7b7a3f3df384194e3531d7c709c7b02bb130d1f6 means the same thing in any repository.	17:14
	fynn: 'the commit object' being the commit object for revision (there can be only one for each revision by definition).	17:15
fynn	Ilari: until there's a clash... ;)	17:15
	btw, that hash is a hash of what? a diff file?	17:16
idletask	Well, SHA1 was chosen for a reason	17:16
	fynn: it depends on the object type	17:16
	A commit object is one type of object	17:16
	It affects a tree object, and one or more blob objects	17:17
Mikachu	fynn: it has the date and name of the committer and author, the commit message, and the tree	17:17
Ilari	fynn: The hash is computed from object type and object contents.	17:17
Mikachu	fynn: git cat-file commit HEAD will show you	17:17
	fynn sees	17:17
fynn	this "object type" thing is way over my head for now.	17:17
	none of the previous SCMs I used had "objects".	17:18
idletask	"Git tracks contents, not files"	17:18
Ilari	fynn: There are four object types in Git: 'blob', 'tree', 'commit' and 'tag'.	17:18
	fynn 's head explodes	17:19
parasti	mandatory link... http://eagain.net/articles/git-for-computer-scientists/	17:19
bremner	fynn: sure they did, they just hid them	17:19
Ilari	fynn: blobs store file contents, trees bind them into directories, commits build trees into history and tags introduce trust into system.	17:20
deskin	parasti: haha, I was pasting that same link :)	17:20
	→ schlort joined	17:22
	→ bieneff joined	17:23
Ilari	fynn: You can explore by starting from 'git cat-file -p HEAD' and then following SHA sums you find (git cat-file -p <shasum>)...	17:23
fynn	Ilari: thanks. is there a good URL for explaining the objects types stuff?	17:24
	I think your nutshell explanation actually makes some sense.	17:24
Ilari	fynn: That above link parasti gave?	17:24
fynn	ah, OK, didn't think it was for me :)	17:24
	I thought Linus and the rest of the kernel elite hate computer scientists? ;)	17:25
	Ilari: thanks, that explanation is pretty elegant,.	17:26
Ilari	fynn: Usually all the objects users directly deal with are tags or commits... Altough trees and blobs are usually even more numerious in repositories...	17:28
	← bieneff left	17:29
fynn	right, but trees and blobs would be handled behind the scenes by commands like git-commit	17:29
Ilari	fynn: Actually, blobs are usually created by 'git add' and trees are constructed by 'git commit' (previously it invoked 'git write-tree' to do that).	17:30
	← ruphy left	17:33
Ilari	fynn: There's just not usually need to designate trees or blobs...	17:34
fynn	thanks, I think I'll read parasti's link, then I'll be smarter	17:35
	btw, who are you guys?	17:35
	I mean, are you kernel developers? git developers? random C programmers?	17:36
idletask	"Random systems engineer" would rather be it for me	17:36
	I want to switch my company from CVS (ewww) to git	17:37
	I have the plumbing covered, now I need the porcelain covered	17:37
	The Eclipse git plugin exists but is still rough around the edges, and I don't know Eclipse development at all :(	17:37
fynn	CVS? the mind cringes. supposedly, that blows even harder than SVN.	17:38
idletask	Ooh yes	17:38
	Especially since it has been mishandled for years	17:38
	But cvs2svn, which can convert to git too, has done wonders	17:39
Mikachu	there are people who prefer cvs over svn :)	17:39
reuss	CVS is horrible. I've created a tool that enables me to use git at work, where we're stuck with CVS so far. The other day a snag came up where i couldn't import a cvs repostory, and i've blissfully forgotten everything i knew about CVS, but got damn, it's useless in modern development styles.	17:39
idletask	I usually cvs checkout the module, echo CVS >.gitignore, git init, work and then submit to cvs	17:39
	deskin keeps meaning to look at cvs2git, hopefully it'll convert openbsd-src where cvsps did not	17:39
	→ carllerche joined	17:39
idletask	deskin: cvs2git needs physical CVS repos access...	17:40
	← ashleyw left	17:42
	→ foca joined	17:42
deskin	idletask: ya, I have a mirror of the entire repository so that's not a problem	17:43
idletask	Then it should go smoothly	17:44
	Take the trunk version, though, and run contrib/git-move-tags.py afterwards	17:45
	cvs2svn creates tag objects by default, but CVS has no such notion	17:45
deskin	ok	17:45
idletask	This tool converts what tags _can_ be converted this way (some can't)	17:45
	It imported my 53-modules, 8-years-history, 1.5 GB repos in a little less than 3 hours	17:46
deskin	that's pretty good	17:46
	better than git-svn :)	17:46
idletask	That's with git repo creation and master checkout, too :p	17:47
	It creates a git-fast-import file that you have to, well, git fast-import in a bare repo	17:47
deskin	easy enough	17:47
	← aspotashev left	17:49
	→ bcarlyon\|laptop joined	17:50
	→ aspotashev joined	17:51
	→ dreiss joined	17:53
	→ abbe joined	17:54
	← ToxicFrog left	17:55
	→ ToxicFrog joined	17:56
	→ ashleyw joined	17:56
	← ashleyw left	17:59
	→ koke_ joined	18:00
	← koke left	18:00
	→ koke joined	18:01
	← koke_ left	18:01
	→ d00dlez_ joined	18:05
	→ warthog9 joined	18:06
	→ xkr47 joined	18:07
	← BarryCarlyon left	18:07
	bcarlyon\|laptop → BarryCarlyon	18:07
	← d00dlez_ left	18:14
	→ ashleyw joined	18:15
	← d00dlez left	18:17
	← ppicazo left	18:17
	← ToxicFrog left	18:25
	→ loincloth joined	18:26
	→ ToxicFrog joined	18:26
	← Jacolyte left	18:26
	→ d00dlez joined	18:28
	→ warthog19 joined	18:28
	← warthog9 left	18:29
	→ advany joined	18:29
	← df00z left	18:31
	← name left	18:32
	→ idle_task joined	18:33
	→ qrush_ joined	18:35
	→ radarek joined	18:37
	← d00dlez left	18:39
	→ loinclo__ joined	18:40
	← loinclot_ left	18:41
	← loincloth left	18:46
	→ CodeOfficer joined	18:46
	← idletask left	18:47
	← cedricv left	18:47
	← ashleyw left	18:48
	← sergiodj left	18:48
	→ cedricv joined	18:48
	← Hallin left	18:49
	→ ppicazo joined	18:49
	→ ashleyw joined	18:50
HopsNBarley	here's an easy one - how do i delete a ref in a bare repo?	18:51
doener	git update-ref -d <ref>	18:51
	or via push: git push <remote> :<ref>	18:52
	← loinclo__ left	18:53
	→ loincloth joined	18:54
	→ Tv joined	18:55
	→ name joined	18:56
HopsNBarley	doener, thanks!	18:58
	→ spearce joined	19:02
	→ bcarlyon\|laptop joined	19:03
	← advany left	19:05
	→ ShadeHawk joined	19:05
	→ ashleyw_ joined	19:07
	→ sergiodj joined	19:08
	→ eternaleye joined	19:10
	→ bobesponja joined	19:10
Ilari	HopsNBarley: Or 'git push . :<ref>'? That at least gets the git push DWIM guessing (unlike the update-ref-one)...	19:13
ShadeHawk	Ilari: what this strange construct is meant to do (what "git branch -D <ref>" cannot do)?	19:16
	← BarryCarlyon left	19:18
	bcarlyon\|laptop → BarryCarlyon	19:18
doener	ShadeHawk: git branch -D takes a branch, not a general ref	19:19
ShadeHawk	well, branch or remote-tracking branch (with -r)	19:19
	and "git tag -d" to delete tag	19:20
	← deskin left	19:20
	← ashleyw left	19:21
	→ swhitt joined	19:23
	→ daddius joined	19:24
	→ loinclot_ joined	19:28
	← mojombo left	19:28
	← loincloth left	19:32
	← abbe left	19:34
	→ Weasel[DK] joined	19:37
	stick → stick[S3]	19:38
	→ ceej joined	19:40
	→ daaku joined	19:42
	← G_SabinoMullane left	19:44
	→ advany joined	19:46
	→ cilly joined	19:46
	← daddius left	19:48
	← drobbins left	19:48
	→ Zimm joined	19:49
	← nadim left	19:51
	→ nadim joined	19:52
	→ eno joined	19:57
	← ashleyw_ left	20:02
	→ deskin joined	20:07
	← chenk85 left	20:07
xenoterracide	is it normal for a git merge -s ours to not have a diff?	20:09
	when I try git diff HEAD HEAD~1 I don't get anything	20:10
doener	well, that's the point of -s ours	20:10
deskin	xenoterracide: yes; think about it	20:10
xenoterracide	yeah but it still changed things	20:10
	doesn't it just prefer local changes to remote	20:11
	→ k776 joined	20:11
deskin	no	20:11
	→ daddius joined	20:11
xenoterracide	well then I'm not thinking about it right... how does it work	20:11
	given that it changes things	20:11
Mikachu	it doesn't	20:11
wereHamster	see the man page: This resolves any number of heads, but the result of the merge is always the current branch head.	20:11
doener	it doesn't just "prefer" them. It just ignores anything it merges except for adding that as parents to the merge commit	20:11
deskin	^^^ that's where you're wrong	20:12
xenoterracide	I read the man page	20:12
corecode	hey	20:12
xenoterracide	if "x%COMSPEC%" == "x" set COMSPEC=command.com I know that line wasn't in my local branch	20:12
corecode	anybody have a pointer how to lock down ssh access for git repo pushes?	20:12
xenoterracide	so it does change thing	20:12
corecode	i.e. only to allow git push, nothing else	20:13
Mikachu	xenoterracide: your faulty memory is a bad argument against facts	20:13
	→ Marmouset joined	20:13
bremner	corecode: git-shell or gitosis	20:13
xenoterracide	could be	20:13
	← cilly left	20:14
	← cedricv left	20:16
corecode	bremner: thanks, marvellous	20:17
xenoterracide	ok my memory is bad I was looking for a different line... god I hate .bat files	20:17
	so what is the point of ours?	20:18
	s/ours/-s ours/	20:18
	if it doesn't change anythng why do it?	20:18
	← iulian left	20:18
Tv	xenoterracide: it marks the things as merged	20:18
xenoterracide	oh	20:19
	→ ashleyw joined	20:20
	← ashleyw left	20:21
	xenoterracide goes back to trying to resolve how I'm going to make this work	20:22
	→ melvinram joined	20:27
	← melvinram left	20:27
	→ alexs_ joined	20:31
	← dreiss left	20:31
	← alexs_ left	20:33
	→ dwmw2 joined	20:35
	→ bobmcw joined	20:37
	dwmw2 → dwmw2_gone	20:38
	→ alexis joined	20:39
alexis	hey there	20:41
Ilari	...	20:41
	stick[S3] → stick	20:41
alexis	I'm trying to clone a git-svn repo and I'd like to use git-svn from that clone	20:42
	I followed the instructions described in the last paragraph of the 'Basic examples' section in the man page	20:42
	but I can't manage to get it work	20:43
	do you guys know how to do that?	20:43
	← BarryCarlyon left	20:43
deskin	alexis: what isn't working?	20:43
	← alexis left	20:44
	← ia left	20:44
	← DrFrasierCrane left	20:47
	← priidu left	20:50
	loinclot_ → loincloth	20:50
	← lucsky left	20:52
	← daddius left	20:53
	← eternaleye left	20:55
jast	ah, good call, i wanted to submit a correction to that part of the git-svn manpage	20:57
	→ pygi joined	20:57
	← Zyroth left	20:58
	← aspotashev left	21:00
	← Voker57 left	21:01
	→ ashleyw joined	21:01
	← ashleyw left	21:04
	← CodeOfficer left	21:04
	← nice_burger left	21:06
	← advany left	21:08
	← bobmcw left	21:10
ShadeHawk	xenoterracide: "git diff HEAD^2 HEAD" or "git diff-tree -p -m HEAD"	21:12
	→ towski joined	21:16
	← tclineks left	21:17
	→ Juice10 joined	21:19
	← Weaselweb left	21:23
	→ ashleyw joined	21:24
	→ spool joined	21:28
	← ashleyw left	21:29
	← towski left	21:30
	→ archis joined	21:33
	← awarde left	21:35
	stick → stick[S3]	21:36
	← warthog19 left	21:36
	← carllerche left	21:38
	→ theking0 joined	21:39
	→ bcarlyon\|laptop joined	21:39
	→ ashleyw joined	21:41
	← scientes left	21:42
	← tvw left	21:43
	→ towski joined	21:44
	→ sergio_ joined	21:46
	← sergiodj left	21:47
	← Marmouset left	21:50
	→ ia joined	21:52
	sergio_ → sergiodj	21:53
noz	After some rebasing, I have a file listed in the "Changed but not updated:" section of status, as 'unmerged:', but 'git diff' shows no differences. How can I get rid of it from the list?	21:53
ShadeHawk	git add <file>	21:56
	← deskin left	22:04
	→ _Vi joined	22:08
	← towski left	22:09
	← Juice10 left	22:13
	→ awarde joined	22:13
	→ advany joined	22:13
	← advany left	22:14
	→ marcan joined	22:15
	← radarek left	22:16
thiago_home	probably a modified file that was erased	22:19
	git add or git rm	22:19
	you choose	22:19
marcan	I'm getting "error: failed to push some refs to 'repo'" after a push but the push seems to complete fine and there are no more error indicators. Any clue what's wrong?	22:20
	here's an example: http://pastie.org/private/8ijzg64woe9wbemjy82g	22:20
	I have another repo that should be set up similarly (both here and on the server) which isn't showing this error	22:20
pygi	bobesponja, ping	22:21
	→ nice_burger joined	22:22
	← sergiodj left	22:26
	→ ciskje joined	22:26
Ilari	marcan: I have seen reports of that issue before... The problem seems to be that something returns failure code when it should have exited with success...	22:26
	→ sergiodj joined	22:27
marcan	sounds about right from what I saw in the code	22:27
	I guess I should just ignore it for now?	22:27
Ilari	marcan: You can ignore it for now. Real push failures would also show the failed refs as '[rejected]' or some such.	22:28
	← noz left	22:28
marcan	Ilari: right, that's what I would have expected	22:29
	← charon left	22:30
marcan	thanks :)	22:30
Ilari	marcan: IIRC, someone reported that exact problem once and then found the issue... I don't remember what the actual issue was...	22:31
marcan	hm	22:32
Ilari	marcan: You have any hooks enabled on the remote side?	22:32
segher	ilari: some obscure configuration mistake, you remember?	22:32
marcan	don't think so	22:32
	I do have a symlink along the way to the server repo which I don't have with the other repo, but I doubt that's an issue unless git specifically checks for it	22:33
	i.e. my repo is ssh://remote/symlink_to_repo	22:33
segher	i use that a lot, doesn't hurt	22:33
Ilari	marcan: Are those repos on the same server?	22:33
marcan	yup	22:34
Ilari	marcan: Maybe run diff on files, excluding stuff under objects/, refs/ and logs/?	22:34
marcan	hmm	22:35
	oh, wait...	22:35
mugwump_	if you've got a symref from one branch to the other and you try to push both, you get that error	22:35
marcan	nope, not that	22:36
	(the wait thing)	22:36
	mugwump_: hmm	22:36
	mugwump_: I don't think I have any of those	22:36
mugwump_	if you use git push -n it should tell you what it's trying	22:37
	putting symlinks in some places, particularly under refs/, could cause that as well	22:37
	← Weasel[DK] left	22:37
mugwump_	use find . -type l or somethign	22:37
marcan	git push -n is invalid here	22:38
SRabbelier	is there a better way than 'git diff -U5000' to get a diff of the entire file (e.g., full context)	22:38
vmiklos	-U-1?	22:39
	(untested)	22:39
	or -U0	22:39
SRabbelier	vmiklos: -U0 gives no context	22:39
	vmiklos: -U-1 says invalid option	22:39
ShadeHawk	SRabbelier: git diff <file> /dev/null ?	22:40
	SRabbelier: git diff --root (in some cases) ?	22:40
SRabbelier	ShadeHawk: nono, I want to get the diff against HEAD	22:40
	ShadeHawk: I just want full context	22:40
ShadeHawk	ah. no, I don't know	22:40
vmiklos	i think -U-1 could be an option, but then obviously it isn't implemented atm	22:40
SRabbelier	vmiklos: ah, ok	22:41
	→ cbowns joined	22:41
mugwump_	marcan: sorry, git push --dry-run	22:41
	another possibility is that the update hook is returning an error code	22:41
	mugwump_ → mugwump	22:41
marcan	mugwump_: no symlinks that I can see either	22:42
	← cbowns left	22:42
	→ cbowns joined	22:42
	→ d00dlez joined	22:43
Ilari	Wouldn't update hook failing cause [rejected]?	22:44
	Or something like that?	22:44
	← idle_task left	22:44
	← Illocution left	22:44
	← Tekerson left	22:44
	→ idle_task joined	22:44
	→ Tekerson joined	22:44
	→ Illocution joined	22:44
crab	last crab	22:45
marcan	mugwump: --dry-run works	22:45
crab	urk	22:45
marcan	once I take it out it pushes fine but shows the message	22:45
mugwump	what does the dry run actually say?	22:45
	← tarbo left	22:45
	→ tarbo joined	22:46
	← name left	22:46
segher	ilari: there's a post-update hook as well	22:46
marcan	mugwump: http://pastie.org/311015	22:48
	← jm left	22:48
	→ tclineks joined	22:50
	→ careo joined	22:51
mugwump	that is very strange. if it's not just a bad hook, then try sending the report to the list	22:51
	← mithro left	22:55
Ilari	I guess SSH returns with 1 exit status...	22:56
	← d00dlez left	22:56
	← Ian00 left	22:57
	← ciaran left	22:58
	→ ciaran joined	22:58
	→ Ian00 joined	23:00
	stick[S3] → stick	23:01
	stick → stick[S3]	23:02
	← ceej left	23:06
	← parasti left	23:07
	← kumbayo left	23:08
	← elmex left	23:11
	← softdrink left	23:15
	→ cmacleod joined	23:16
	→ carllerche joined	23:18
	cmacleod → stick	23:18
	← stick left	23:19
	← _graham_ left	23:21
marcan	mugwump: ah, it was a bad hook in the end	23:22
	or rather some weird permissions mess in the hooks folder	23:22
	stick[S3] → stick	23:22
	stick → stick[S3]	23:22
marcan	I'm not entirely sure how that got that way, but it's fixed now	23:22
	stick[S3] → stick	23:24
	stick → stick[S3]	23:24
	→ softdrink joined	23:24
	← ciaran left	23:25
	→ ciaran joined	23:26
	werdnus waves	23:27
werdnus	andrew@gcentral_router:/var/www/wiki$ git svn dcommit -n -i refs/remotes/svn/group_rights	23:27
	That's the wrong URL	23:27
	Committing to svn+ssh://svn.wikimedia.org/svnroot/mediawiki/trunk/phase3 ...	23:27
	I've specified this in .git/config: url = svn+ssh://svn.wikimedia.org/svnroot/mediawiki/branches/on_wiki_configuration	23:27
	Any ideas on how to get it to commit to the correct place?	23:28
	← tclineks left	23:28
	→ pigeon_ joined	23:29
	← litage left	23:30
	← carllerche left	23:32
	→ carllerche joined	23:32
	stick[S3] → stick	23:32
	stick → stick[S3]	23:32
	→ quoin joined	23:33
	stick[S3] → stick	23:33
mugwump	werdnus: you need to branch in svn first	23:34
werdnus	I started from refs/remotes/svn/group_rights	23:35
	I did git checkout -b group_rights refs/remotes/svn/group_rights	23:35
	← ph^ left	23:36
	← cbowns left	23:36
	← ciaran left	23:37
	→ cbowns joined	23:37
	→ ciaran joined	23:37
	→ mithro joined	23:38
	← cbowns left	23:39
	→ Flathead joined	23:39
	← rmh3093 left	23:40
	← joshdavey_ left	23:41
	→ ciskje_ joined	23:41
	→ cbowns joined	23:41
	→ rmh3093 joined	23:43
	→ rtomayko joined	23:44
	→ qrush__ joined	23:46
	← rmh3093 left	23:46
	← carllerche left	23:48
	← qrush left	23:50
	→ arekm joined	23:51
	← Ademan left	23:53
	← stampp- left	23:53
	→ ciaran_ joined	23:55
	← ciaran left	23:55
	← ciskje left	23:55
	← pete__c left	23:55
	→ jesselucas joined	23:55
	← ciaran_ left	23:55
	→ ciaran joined	23:56
	← reuss left	23:57
	← careo left	23:57

Logs Search ←Prev date Next date→

Channels Documentation