IRCloggy #git 2008-10-21

Provider of IRC logs since 2005.

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

	2008-10-21
	→ bbuss joined	00:01
	→ quoin joined	00:01
	→ Flathead joined	00:05
	→ ciaran joined	00:06
	← eternaleye left	00:06
	← withanx left	00:10
	→ diask joined	00:12
	← diiask left	00:12
	← carrus85 left	00:15
	→ LiamH joined	00:17
	← deskinm left	00:18
	→ deskin joined	00:19
	→ qrush joined	00:19
	← akahn left	00:21
	← jburd left	00:21
	→ tiglionabbit joined	00:25
	← mw\|out left	00:28
	→ akahn joined	00:30
	← Tv left	00:33
	→ akahn_ joined	00:38
akahn_	Is there a way to overwrite my local uncommited changes when using `git pull`?	00:38
	→ sgrimm joined	00:38
	← joeyh left	00:39
	← akahn left	00:39
	akahn_ → akahn	00:39
	→ akahn_ joined	00:40
deskin	akahn: sure: git reset --hard HEAD; git pull	00:40
mashmore	disclaimer: I'm still a bit of a git newb too, but I'm pretty sure that - if you wanted to discard local uncommitted changes - you'd do that before you "git pull'	00:40
akahn	deskin: ah, makes sense, thx	00:40
	→ eternaleye joined	00:41
	→ Dythzer joined	00:46
leo2007	spearce: thank you so much for your kind help	00:47
	← bbuss left	00:48
akahn	How do I use gvimdiff with git mergetool? Where do I move the chunks to/from?	00:50
	With the three columns, where the middle one is the final file.	00:50
	→ timcharper joined	00:52
	← timcharper left	00:56
	← Dythzer left	01:12
	→ shenie joined	01:15
	→ jtoy joined	01:18
	→ gretch_ joined	01:22
	← schlort left	01:24
	← kef left	01:27
	← ciaran left	01:28
	→ pillowfactory joined	01:29
	← dkr left	01:30
	← CelticSoul left	01:30
	← pcapriotti_ left	01:33
	← mashmore left	01:34
	→ schlort joined	01:35
	→ warlock_mza joined	01:39
	→ roadt joined	01:41
	← roadt left	01:41
	→ mw\|out joined	01:44
	→ timcharper joined	01:48
	← timcharper left	01:49
	← mw\|out left	01:53
	← warlock_mza left	01:56
	→ drobbins joined	01:56
drobbins	hiya	01:56
	← BarryCarlyon left	01:57
	← statim left	02:01
	→ Tv joined	02:02
	← whee left	02:02
	← dmlloyd left	02:02
	→ tiglionabbit_ joined	02:04
offby1	aw, g'wan	02:04
	← offby1 left	02:04
	→ FunkeeMonk joined	02:10
	← markelikalderon left	02:10
	→ offby1 joined	02:12
	← tiglionabbit left	02:15
	← sgrimm left	02:15
	← akahn_ left	02:19
	→ imyousuf joined	02:20
leo2007	how to list all tags in a work tree?	02:21
offby1	"git tag"	02:22
	maybe "git tag -a"	02:22
MaddHatter	is there a way to "hide" branches from 'git branch'? I have some old branches I want to keep around; I just don't want to look at them all the time.	02:23
j_engelh	then just don't look at it	02:24
	→ dmlloyd joined	02:24
offby1	"Don't Do That, Then"™	02:25
MaddHatter	it's nicer when the branch list doesn't scroll off-screen :-/	02:27
leo2007	will tags be preserved in a repo clone from SVN?	02:28
	→ kyrus1 joined	02:29
offby1	Probably, at least if you pass the right command-line option	02:29
	I seem to recall branches getting preserved	02:29
	why not Try It And See™?	02:29
leo2007	actually I tried and couldn't see the tags	02:31
offby1	pay attention to the command-line flags; the first (few) time(s) I tried converting a repository, I forgot those flags, and didn't see any branches or tags.	02:32
	← tiglionabbit_ left	02:33
	→ tiglionabbit joined	02:35
leo2007	offby1: thx	02:38
	← diask left	02:42
	→ diask joined	02:43
	→ statim joined	02:43
	→ sgrimm joined	02:44
	← Slashbunny left	02:48
	→ sudoer joined	02:49
	→ nikolasc joined	02:50
	← shenie left	02:51
	← schlort left	02:52
	← neurodamage left	02:52
	← david_koontz left	02:55
	← jtoy left	02:56
	→ eternaleye_ joined	02:58
	← LiamH left	03:00
	← statim left	03:01
	← FunkeeMonk left	03:02
	→ tiglionabbit_ joined	03:04
	← madduck left	03:05
	← eternaleye left	03:12
	← tiglionabbit_ left	03:14
	← tiglionabbit left	03:14
	→ WALoeIII joined	03:21
	← flaguy left	03:22
	→ r0bby joined	03:23
	← jackdempsey left	03:25
	→ fhobia joined	03:26
	→ mithro joined	03:27
	→ \ask- joined	03:28
	→ willb joined	03:28
	→ zachinglis joined	03:31
deskin	leo2007: not sure if you're still trying svn stuff; since svn tags are actually mutable, git svn saves them as branches, so git branch would show them, not git tag	03:32
	→ schlort joined	03:34
	→ jtoy joined	03:35
	→ DGMurdockIII joined	03:38
DGMurdockIII	dose anyone now any free software that can turn a pocket pc in to a remote for a tv?	03:38
	hi	03:38
	← sudoer left	03:38
deskin	DGMurdockIII: tried #handhelds?	03:39
DGMurdockIII	ok	03:39
	← DGMurdockIII left	03:40
litage	i checked out an old commit with ``git checkout A_COMMIT_HASH''. how do i now go back to the latest commit?	03:40
	well, i guess "forward", not "back" =P	03:40
offby1	check it out!	03:41
j_engelh	checkout what you had before	03:41
offby1	use "git reflog" if you can't remember its hash	03:41
litage	j_engelh: so i have to use the commit's hash?	03:41
	→ paltman joined	03:41
j_engelh	anything that resolves to a commit	03:41
offby1	only if there is no ref pointing to it	03:41
	← zachinglis left	03:42
litage	thanks guys	03:43
j_engelh	no named ref ;)	03:43
	a commit is a ref, sort of :p	03:43
offby1	what would an unnamed ref be?	03:44
j_engelh	a commit SHA has no name	03:44
offby1	those count as refs?	03:45
j_engelh	dunno	03:45
offby1	I thought refs were "little files under .git/refs"	03:45
j_engelh	but git-push certainly accepts them!	03:45
	well ref is short for reference (sic: pointer)	03:45
	and well, I'd argue that a SHA is a ref to itself at least :p	03:45
	or rather	03:46
	the string you hand to checkout is a semi-named ref that will be resolved to the particular commit	03:46
	inconceivably confusing!	03:46
	offby1: also by your argument,	03:46
	"only if there is no ref pointing to it"	03:46
	but you can say git checkout HEAD@{1}	03:46
	now what's that...	03:47
	it's neither a named ref, nor a commit SHA	03:47
	oh well HEAD may be. But HEAD@{n}, HEAD^n, HEAD^{n} and HEAD~n probably are not :p	03:47
	offby1 rolls eyes	03:48
deskin	maybe refs should be thought of as things which keep commits from being gc'd, or changed with git update-ref/symbolic-ref	03:52
	← mithro left	03:52
	deskin shrugs	03:52
parasti	man gittutorial, of course	03:53
Gitbot	parasti: the gittutorial manpage can be found at http://git.or.cz/man/gittutorial	03:53
parasti	err, did I say "tutorial"? anyway, scroll right to the bottom and click "gitglossary(7)" :P	03:54
	→ mithro joined	03:55
	→ Ademan joined	03:57
	← Ademan left	03:58
	→ jony joined	03:59
	→ jhelwig joined	03:59
	→ paltman_ joined	04:00
	← parasti left	04:00
	→ johnw joined	04:00
	← jhelwig left	04:01
	→ jhelwig joined	04:01
	← fhobia left	04:01
	← uau left	04:02
	→ ben_h joined	04:02
	→ offby1` joined	04:06
	→ perezd88 joined	04:06
	→ jesselucas joined	04:06
	→ uau joined	04:07
	← SRabbelier left	04:09
	→ tjafk2 joined	04:10
	→ SRabbelier joined	04:12
	← tjafk1 left	04:13
	← bremner left	04:14
	← paltman left	04:16
	→ bremner joined	04:18
	→ fhobia joined	04:19
leo2007	deskin: git branch shows only one branch master	04:20
	← jtoy left	04:22
	← willb left	04:22
	← offby1 left	04:22
	→ paltman joined	04:23
	→ Jacolyte joined	04:28
	offby1` → offby1	04:29
	← paltman_ left	04:30
	← jony left	04:34
SRabbelier	pasky: dude?	04:37
	← iAmTheDave left	04:37
SRabbelier	pasky: you know what would be a sweet gitweb feature? optionally turn of line-numbering in blob view... so you can actually copy/paste from it	04:39
qrush	isn't that what the raw view is for?	04:41
	← bts- left	04:42
tpope	it's also quite possible to structure html in a way that web browsers won't grab the line numbers	04:42
	→ bts- joined	04:43
	← niki left	04:43
	← kyrus1 left	04:43
	→ rtomayko joined	04:43
	→ asp_ joined	04:46
	→ cedricv joined	04:51
SRabbelier	tpope: that would be ven better, probably	04:55
	qrush: which one?	04:55
qrush	the raw view. it's just the straight contents of teh file	04:56
	← Tv left	04:57
	← schlort left	05:01
	→ Tv joined	05:02
	→ mediogre joined	05:06
	→ schlort joined	05:07
	→ dkr joined	05:08
	← up_the_irons2 left	05:10
	← dkr left	05:10
	← madewokherd left	05:12
	→ up_the_irons2 joined	05:13
	← jesselucas left	05:13
	→ paltman_ joined	05:18
	→ dkr joined	05:18
	→ jony joined	05:21
	Arelius → AreliusFreeNode	05:23
	AreliusFreeNode → Arelius	05:23
	← diask left	05:23
	→ diask joined	05:24
	Arelius → AreliusFreeNode	05:24
	AreliusFreeNode → Arelius	05:25
	← paltman left	05:25
	← perezd88 left	05:29
	→ paltman joined	05:31
	← jony left	05:31
	→ ruskie joined	05:32
	→ jony joined	05:33
	← paltman_ left	05:38
	← qrush left	05:48
	← WALoeIII left	05:55
	← CodeOfficer left	05:56
	→ c0sin joined	05:56
	← pillowfactory left	06:05
	← uau left	06:06
	← carllerche left	06:08
	→ kef joined	06:08
	← Aikawa left	06:09
	← paltman left	06:15
	→ elmex joined	06:17
	← felipec left	06:18
	← Flathead left	06:19
	→ uau joined	06:20
	→ carllerche joined	06:20
	→ harinath joined	06:20
	→ priidu joined	06:21
	← SRabbelier left	06:23
	→ pio_ joined	06:29
	← jony left	06:31
	→ jony joined	06:33
	← pio_ left	06:34
	→ thorny_sun joined	06:35
thorny_sun	i'm very confused by git log manpage. how can i get git-log to show diffs on a particular file?	06:35
	git log -- filename isn't giving me diffs	06:36
	and i can't find the switch to do it	06:36
doener	-p	06:36
thorny_sun	oh-- so patch format is the same as diff format?	06:36
doener	diff generates patches ;-)	06:36
	← litage left	06:37
	← priidu left	06:39
	→ reel_ joined	06:41
	→ ph^ joined	06:41
	← ben_h left	06:44
	← bronson left	06:44
thiago_home	thorny_sun: what other format would you expect for a patch? :-)	06:45
	diff and patch are the two sides of the same coin	06:45
	→ bentob0x joined	06:45
	→ pantsman joined	06:54
	← thorny_sun left	06:57
	→ sbeyer joined	06:58
	→ madduck joined	06:59
	← mithro left	07:00
	← Ramblurr left	07:00
	← rtomayko left	07:02
	→ Weaselweb joined	07:03
	→ bronson joined	07:05
	→ koke joined	07:08
	→ priidu joined	07:13
	→ nud joined	07:17
	← asp_ left	07:18
	→ lucsky joined	07:22
lucsky	'morning	07:23
	← c0sin left	07:23
	→ Dythzer joined	07:25
	→ charon joined	07:26
	← carllerche left	07:29
	← tchan left	07:30
	→ ciaran joined	07:33
	← ph^ left	07:40
	→ ph^ joined	07:41
	← tango_ left	07:41
	← regulate left	07:46
	→ ben_h joined	07:51
	← uau left	07:52
	→ CelticSoul joined	07:53
	→ regulate joined	07:53
	→ digitalpardoe joined	07:54
	← digitalpardoe left	07:54
	→ zachinglis joined	07:55
	→ uau joined	07:57
	→ Hemebond joined	07:57
	← johnw left	07:58
	→ truls joined	07:58
truls	hey	07:58
	← fhobia left	08:00
	→ jonalv joined	08:00
	→ johnw joined	08:02
truls	playing with git blame <path>, and it only seems to be able to show correct info when I provide a direct relative pathname (like a/b/c.txt)	08:03
	→ markelikalderon joined	08:04
	→ advany joined	08:04
truls	anything else (like absolute path or ./a/b/c.txt) will still show the file, but with only info from the first commit	08:04
	is this expected behavior, am i missing something?	08:04
	→ trochala joined	08:05
	→ shodges joined	08:07
Hemebond	truls: You asking me?	08:10
	→ olly joined	08:11
	→ pcapriotti joined	08:11
	← Jacolyte left	08:15
truls	Hemebond: you too	08:15
Hemebond	I have no idea. I just started using Git.	08:15
truls	heh, then dont worry about it	08:16
charon	truls: works here. which git version? any symlinks in the path?	08:17
doener	truls: "git blame foo" and "git blame ./foo" make no difference here	08:17
truls	charon: 1.5.3 on ubuntu, probably ancient	08:17
	1.5.4.3	08:17
charon	097971f5f56	08:18
Gitbot	[git 097971f5f]: http://tinyurl.com/644y33 -- Make blame accept absolute paths	08:18
charon	appears to be in 1.5.5	08:18
truls	ah, thanks. where should i have searched for that? google didnt help out	08:19
charon	i just did an educated-guess-search through git.git	08:19
	('gitk builtin-blame.c' and search for 'path')	08:19
truls	ah	08:19
doener	git log --grep=blame.*absolute :-)	08:20
truls	didnt really think about getting the git source, keep forgetting how useful that is	08:21
	(using git to look for changes that is)	08:21
	→ tango_ joined	08:21
truls	thanks for checking into my problem	08:21
	← pantsman left	08:22
	→ pantsman joined	08:22
	← bentob0x left	08:25
	→ Arafangion joined	08:26
Arafangion	Hey, is it possible to copy a remote branch from one git repo and put it into another git repo (with a different name)?	08:27
	→ offby1` joined	08:27
	← markelikalderon left	08:29
Arafangion	The situation is that I can do a git p4 clone of two branches into two git repositories, and I want to keep the same behaviour. If I do a git clone, I can't merge them properly, and if I create a remote tracking branch of the original repos, it works except that only the original repos (made with git p4 clone) can properly sync to perforce.	08:29
	← spearce left	08:29
	← zachinglis left	08:30
	→ carllerche joined	08:30
	← ben_h left	08:30
DrNick	Arafangion: git fetch is fundamentally about taking a branch in a remote repo and copying it into a branch in the local repo	08:33
doener	Arafangion: maybe there's just metadata for git p4 missing?	08:34
	→ Fullmoon joined	08:35
Arafangion	DrNick: I was trying to do it with git remote, which if I'm correct, sets the local repo such that fetch works propelry?	08:37
DrNick	that's not necessary	08:37
	→ jacobat joined	08:37
Arafangion	DrNick: Please enlighten me.	08:37
	Also, is it possible to do a merge even if there are no common objects between teh branches?	08:38
DrNick	git remote is for setting up convenient names for remote repositories that you plan on fetching from frequently	08:38
	git fetch can fetch from any old url	08:38
Arafangion	DrNick: I plan on fetching extremely frequently.	08:38
mugwump	fonseca: do you know who is behind the snarky comments on the gittorrent googlecode project?	08:39
tarbo	you may be eligible for frequent fetcher points	08:39
DrNick	and, yes, you can merge anything together, although it won't be easy unless they share history	08:39
	→ tchan joined	08:39
Arafangion	DrNick: So, what exactly does git fetch do?	08:39
DrNick	sets a local branch to the head of a remote branch, and copies all the necessary objects	08:40
Arafangion	Wow, simple.	08:43
	→ pehlert joined	08:43
	← offby1 left	08:43
Arafangion	How would I ensure that the metadata is also copied?	08:43
	Or do I have to do that manually?	08:43
DrNick	what metadata?	08:44
Arafangion	Whatever git-p4 likes to use.	08:44
	← Dythzer left	08:44
DrNick	no idea	08:45
Arafangion	Basically, what I have is a perforce repo containing two branches, A, and B.	08:45
doener	_if_ git p4 requires additional metadata/configuration settings, you'll most likely need to manually copy them, or have "git p4" rebuild that metadata from the commits (git svn can do that, because the meta-data is (partially) also in the commit messages)	08:45
	← kanru left	08:45
	→ kanru joined	08:45
Arafangion	I use git-p4 to generate git repositories for branch A and branch B, which I'm unfortunately making as separate repositories.	08:45
	Hmm, now that I think of it, this is fundamentally a git-p4 issue.	08:46
DrNick	you probably don't want A and B in separate git repositories	08:47
Arafangion	Maybe a better solution would be to write up a hook script in the respective repos, so that if I use git to push to it, it will automatically take that commit and put it into perforce.	08:47
mugwump	Arafangion: see also http://utsl.gen.nz/gitweb/?p=git-p4raw	08:47
Arafangion	DrNick: You're right, but even for this, I have to hack git-p4.	08:47
mugwump	if you've got direct access to the perforce repository it might work	08:47
	Arafangion actually has git-p4 running for a third attempt for a complete, "proper" attempt, which will hopefully work, but isn't hopeful.	08:47
Arafangion	mugwump: What do you mean by "direct"?	08:48
mugwump	as in, you can see a whole load of db.* files and a depot/ directory	08:48
	also a checkpoint and journal	08:48
Arafangion	mugwump: Ok, I definetly don't, and won't be able to get, direct access.	08:48
mugwump	mmm. It's hardly surprising that git-p4 doesn't support bringing across perforce merge information	08:50
	→ cyberpear joined	08:50
Arafangion	mugwump: I agree.	08:50
mugwump	it's hard enough for the users to figure out what happened	08:50
Arafangion	The issue I've actually got is that I can do a "proper" git-p4 clone @all, but that only works up to a certain revision, at which git-p4 starts referring to sha1 sums that don't exist, and falls apart.	08:51
	However, I could clone just the tip of the revision, but I can either do it as one branch at a time into it's /own/ repository, and I don't think I have enough resources to copy several branches at the same time.	08:52
	And even if I were to copy several at teh same time, I suspect that --detect-branches might not work satisfactory.	08:52
mugwump	well you could potentially graft it in git land on its way through	08:53
Arafangion	That's what I'm attempting to do.	08:53
	Well, worst comes to worst, I'll just write hooks to have the 'master' git repos send up commits for their respective branches.	08:56
	→ hurikhan\|Work joined	08:56
doener	Arafangion: hm, hashes that don't exist? Are the CRLF files? Windows? Is autocrlf enabled?	08:57
	Arafangion: autocrlf busts git-svn as well when there are crlf files in the svn repo	08:57
Arafangion	doener: It is windows, and yes, autocrlf is enabled by default.	08:57
doener	or at least it used to, there were some patches floating around that were supposed to fix that problem, but I don't know what became of them	08:58
Arafangion	doener: The issue is that git-p4 clone actually /creates/ the repo. Maybe if this fails (and it's again because of hashes that don't exist), I'll try modifying git-p4 yet again to disable it.	08:58
doener	if git p4 gets a crlf file, then the original file (with crlf) will have a different hash then the blob that actually ends up in the repo (with just lf)	09:00
Arjen	Is it possible to 'explode' a pack into its individual contents?	09:00
Arafangion	doener: I was under the impression that git had to be told which files were crlf ones.	09:00
doener	Arafangion: no, that's why it is called _auto_crlf ;-)	09:01
Arafangion	Damn.	09:01
	A useful feature to be sure, IF it isn't on by default. sigh, well I'll give disabling it a go.	09:01
doener	seems that the majority of windows git users couldn't handle lf files at all...	09:03
Arafangion	Only those who use notepad.exe	09:03
doener	but I didn't follow the respective discussions, simply because I don't care about the whole crlf/lf issue. Line endings are LF, period. :-p	09:03
Arafangion	Right now I don't care as long as the damn editors keep it consistent.	09:04
	Which, seriously, is a trivial issue.	09:04
	→ shenie joined	09:05
	← carllerche left	09:06
Arafangion	I don't want to get worked up about the state of text files, already worked up about the state of version control. ;)	09:07
wereHamster	is there a way to find out whether git fetch fetched new data or whether the tree was already up to date?	09:08
	(other then parsing the output)	09:08
Arafangion	wereHamster: du? ;)	09:10
	(And parsing the output, there's output to be parsed regardless of the approach)	09:10
wereHamster	ah well, then it's ` \| grep 'Current branch master is up to date.'`	09:11
Arafangion	Good man. :)	09:15
	→ huzaifas joined	09:17
mugwump	you can get that information from git-show-ref \| grep remotename	09:17
huzaifas	does any one know how i can import a cvs repo into git?	09:18
Arafangion	People still use cvs?	09:18
huzaifas	yep they want to use git now :)	09:18
Arafangion	Heck, subversion's nicer than cvs.	09:18
	huzaifas: I'm a perforce user, at the moment, btw.	09:19
	huzaifas: That said, git and cvs is probably the easiest conversion, I'd reckon, that's how it was tested with Linux, afaik.	09:20
	huzaifas: Google for git and cvs.	09:20
mugwump	lol, go watch the linus google tech talk vid Arafangion	09:20
DrNick	converting CVS to anything is hard	09:20
	because CVS doesn't have the notion of a changeset or commit	09:21
Arafangion	mugwump: I've read the transcript.	09:21
DrNick	and people had the strange habit of editing the ,v files directly, making coherency even more difficult to establish	09:21
pasky	huzaifas: depends on the size	09:21
DrNick	also, Linux has never been stored in CVS	09:21
Arafangion	DrNick: Indeed, but it was in bitkeeper, right?	09:21
pasky	huzaifas: for basic conversion, git cvsimport could do, but iirc it used to be not 100% reliable on stranger repositories	09:21
	huzaifas: there are more advanced solutions, e.g. cvs2git	09:21
Arafangion	DrNick: And guess what alternative RCS system bitkeeper is compatible with.	09:22
DrNick	SCCS	09:22
pasky	bitkeeper uses the sccs file format but it's not clear what do you mean by "compatible"	09:23
Arafangion	DrNick: I was under the impression that one can get a CVS repository out of BK.	09:23
DrNick	what's getting CVS out of BK have to do with getting CVS into git?	09:24
pasky	that you can use that to get bk into git :)	09:24
Arafangion	And hence it must've been one of the first other->git conversions.	09:25
	Arafangion is seriously thinking of writing his own git-p4 hack, which uses a staging area.	09:25
DrNick	no, Linus constructed the first git repo by hand	09:26
	I suppose the Xorg cvs import was early, though	09:26
Arafangion	As a series of patches, that'd make sense too, actually.	09:26
	→ malkomalko joined	09:26
Arafangion	Hmm, turns out Linus doesn't actually talk about using the CVS export from BK at all.	09:27
	Not according to the transcript.	09:28
	dwmw2_gone → dwmw2	09:28
pasky	the cvs export from bk was used for one of the original conversions, iirc	09:28
	but a custom one was created later	09:28
Arafangion	That's what I heard.	09:28
	← synsol left	09:29
wereHamster	I settled for if [ "$(git show-ref HEAD \| head -c7)" = "$(git show-ref origin/master \| head -c7)" ]; then	09:29
	→ mecha joined	09:30
	→ synsol joined	09:36
	← malkomalko left	09:38
	← \ask- left	09:39
Pieter	can anybody remember a patch to git-for-each-ref allowing natural sorting?	09:42
	that is, that 1000a is sorted after 100a	09:42
	← huzaifas left	09:43
	→ Kbyte joined	09:45
	← truls left	09:49
	→ thiago_ joined	09:51
	← shenie left	09:51
	← harinath left	09:58
	← mariusSO left	10:01
	← jonalv left	10:01
	← thiago left	10:03
	← kanru left	10:05
	→ kanru joined	10:06
	→ not-xjjk joined	10:06
	← kanru left	10:06
	→ kanru joined	10:06
	← bdowning left	10:09
	← akahn left	10:09
	← samurai left	10:10
	→ bdowning joined	10:11
	← sdboyer left	10:11
	→ sdboyer joined	10:12
	→ samurai joined	10:13
	→ akahn joined	10:15
	← jony left	10:16
	→ zockerkuddel_ joined	10:17
	← zockerkuddel_ left	10:17
	→ ankit9 joined	10:18
	← thiago_ left	10:21
Arafangion	Hmm. Maybe I could use p4 diff2 to manage my git-to-p4 setup ala "set of patches" style.	10:23
	And use a staging area to apply git's changes as desired when "pushing".	10:23
	← xjjk left	10:27
	Arafangion really hopes he gets git-p4 working.	10:32
	→ jony joined	10:34
Arafangion	What exactly is the origin?	10:43
	← priidu left	10:43
doener	"the" origin? "origin" is the default name of the remote for the repo you cloned from	10:44
Arafangion	How would I make it if I didn't do a clone.	10:44
	(Yes, that's the one I'm talking about)	10:44
doener	clone is kind of: git init; git remote add -f origin <repo-url>" + branch creation + checkout	10:45
Arafangion	makes sense, thanks. That reminds me, when I did git remote add foo	10:45
	git remote foo /c/otherrepo	10:45
	It didn't seem to work, the remote repository kept unexpectidly closing!	10:46
	It worked when I did -f, though.	10:46
doener	-f just means that "git fetch" will immediately be done for the new remote	10:47
Arafangion	That's what's confusing me, somehow doing -f is different to simply calling git fetch yourself.	10:47
doener	ie. "git remote add -f foo /path/to/foo" == "git remote add foo /path/to/foo && git fetch foo"	10:47
Arafangion	It's possible I didn't add that final 'foo' to the fetch command. Hmm.	10:47
doener	just "git fetch" is equal to "git fetch origin"	10:48
	so if you don't have an "origin" remote, that would fail	10:48
Arafangion	Now that makes sense, as I don't generally have an origin.	10:48
doener	well, probably unless you happen to have a subdirectory "origin" with another git repo in it	10:48
	the plain "git fetch" should also have said: fatal: 'origin': unable to chdir or not a git archive	10:49
Arafangion	It didn't.	10:50
	← Hemebond left	10:50
	→ kukks joined	10:52
	→ thiago joined	10:55
Arafangion	thiago: Welcome.	10:59
thiago	thanks	11:00
Arafangion	thiago: I'm hoping that when I get to work, it'll be third time lucky. :(	11:01
thiago	sorry?	11:01
	thiago has to leave for two meetings now	11:01
Arafangion	thiago: Not blaming you! Don't let me keep you.	11:01
	Arafangion is being conversational, he thinks.	11:02
	→ priidu joined	11:06
	→ emilsedgh_ joined	11:07
	→ cmarcelo joined	11:10
	→ jfontan joined	11:13
	← koke left	11:17
	emilsedgh_ → emilsedgh	11:19
jfontan	is there a way to clone an specific directory inside a repo?	11:21
Arafangion	jfontan: I don't think git has a concept of that.	11:22
jfontan	I don't want to loose the history of those files	11:22
	ouch	11:22
	→ EmilMedve joined	11:22
	→ mcella joined	11:23
	→ kishore joined	11:25
	← imyousuf left	11:25
	→ leethal joined	11:26
leethal	greetings! Are there any tacticts to completely remove all traces of a file from a repository?	11:26
kishore	I know there are plenty of blogs out there, but im still looking for a good svn to git migration guide for an svn repo that hosts multiple projects	11:26
leethal	it has never changed, just been added, in case that matters	11:27
Arafangion	leethal: best you could do is rewrite all the history that refers to that file, utterly messing up everybody else's repository if they are required to merge up to you.	11:27
leethal	I see. It's not a big deal, so I guess I'll leave it ; )	11:28
kishore	is there a good guide that implements hierarchy of projects and subprojects (submodules? what's the diff?)	11:28
broonie	Though if you're doing a svn->git conversion that's the best time to do it since nobody will have references to the history before it's converted.	11:28
	→ zachinglis joined	11:28
emilsedgh	i cannot seem to find how to add a new mergetool.i tried config --global mergetool.kompare.path /home/kde-devel/kde/bin/kompare but kompare is not showing up in my mergetool candidates	11:29
	→ p0w3r3d joined	11:30
	→ lorandi joined	11:31
Arafangion	emilsedgh: it's mergetool.kompare.cmd	11:35
emilsedgh	thanks Arafangion, let me try it	11:35
Arafangion	emilsedgh: See git-config help	11:35
	git help config, rather.	11:35
emilsedgh	git config --global mergetool.kompare.cmd /home/kde-devel/kde/bin/kompare didnt work, but i'll read them.thanks.	11:37
	→ marze joined	11:39
	→ bcarlyon\|laptop joined	11:39
	bcarlyon\|laptop → BarryCarlyon	11:41
	Mario_ → pygi	11:41
	← cyberpear left	11:43
Arafangion	As a curiosity, how does git behave with respect to file renames/moves which differ only by case, on different filesystems that handle case in different ways.	11:45
thiago	git doesn't track file names	11:46
	it tracks content	11:46
Arafangion	I'm aware of that, having said that, I know that git-p4 has issues if it is used on a case insensitive FS while there are case sensitive file movements.	11:46
thiago	Git won't like it either	11:47
Arafangion	I suspect that as far as 'native' git handles it, that is, git without the use of git-p4 or other tools, it'd be fine until you actually check out the offending revision.	11:47
	thiago: btw, my latest theory with git-p4 dying is that, unless my current sync actually succeeds, is that autocrlf might be messing things up.	11:49
thiago	we imported on Linux with autocrlf off	11:49
	Git doesn't care about what the file names are	11:50
	→ whee joined	11:50
thiago	the problem is when you change the case of a file in a case-insensitive FS	11:50
	when git stats the old name, it's there. But when it lists the dir, not only is it not there anymore, a new name appeared.	11:50
	→ earcar joined	11:51
	← gretch_ left	11:51
	← earcar left	11:51
	→ earcar joined	11:51
	← earcar left	11:51
Arafangion	I sometimes wonder if git should be case insensitive but case preserving on such OS's.	11:52
	→ Ramblurr joined	11:52
	→ makkalot joined	11:52
thiago	it should	11:54
	but it's not an easy question	11:54
makkalot	hi all i used command to have patches of all my commits : git-format-patch -C -M -o patches/ --author=makkalot --root , and i want to apply those on another machine which doesnt have git installed,so i can use only patch command i do patch -p0 < file.patch but alway got errors?	11:54
thiago	it's not about the OS, but the FS	11:54
	← p0w3r3d left	11:54
Arafangion	thiago: Trivial to test.	11:54
thiago	for example, MacOS X can run on HFS (insensitive) or ZFS (sensitive)	11:54
	good point (for an application)	11:54
	my experience is from a library perspective (Qt): I can't just create a new file	11:55
	I had to reject the task that requested that	11:55
deskin	makkalot: patch -p1, not -p0	11:55
makkalot	deskin: so it is not about the git format	11:56
	i will give a try	11:56
deskin	git puts fake a/path/to/file and b/path/to/file as leading the diff header	11:56
	nope, the diff is compatible with patch	11:56
Arafangion	thiago: Well, in Qt's case, if you had to do it, the best you could do would be to provide a hook.	11:57
	I must admit that I've never actually developed using Qt. :(	11:57
	Unfortunate, as I'm told it's the best, generally speaking.	11:58
	Arafangion is currently doing Gui development with Delphi.	11:58
bmalee	Arafangion: are you in therapy for that?	11:59
Arafangion	bmalee: It could be worse.	11:59
Cozminsky	could be worse, could be vanilla pascal	11:59
Arafangion	And this is remarkably good, clean Delphi.	12:00
	It's better than C++ in some ways, though I seriously miss RAII. :(	12:00
	Seriously.	12:00
	And not having templates is an annoyance.	12:01
	→ qrush joined	12:02
Arafangion	Delphi 2009 has closures, lambdas and generics, btw.	12:03
makkalot	deskin: thanks :)	12:03
deskin	glad to help	12:03
	→ grahal joined	12:04
	← marze left	12:06
	→ buckett joined	12:07
	→ tcurdt joined	12:08
	← Fullmoon left	12:10
	← EmilMedve left	12:10
tcurdt	hey folks ... quick question: moving head to a tag I used 'git checkout mytag'. I tells me though that I might want branching instead - but it does seem to move head to the tag. So what's the proper way of doing that then?	12:11
	← kishore left	12:11
	→ Jacolyte joined	12:24
Arafangion	tcurdt: You could make a branch from the tag.	12:24
	← deskin left	12:24
	→ bobmcw joined	12:24
Arafangion	tcurdt: Or ignore the warning, it's a tag, after all, used to label a particular commit sha1, which you're free to use as a branch.	12:25
	(I think)	12:25
	→ Ryback_ joined	12:25
	→ d0k joined	12:32
	← advany left	12:33
	→ koke joined	12:36
	→ girishr joined	12:38
doener	tcurdt: you just can't checkout a tag like a branch. You'll get a detached HEAD. Basically that just means that you have a commit checked out, and when you eg. create a new commit, HEAD will be updated to reference the new commit, while "normally" the branch head is updated	12:39
Arafangion	doener: The implication here is that he wants to check out a particular tag, perhaps to build that version.	12:41
doener	yeah, and then it's fine to just have a commit checked out.	12:42
Arafangion	Perhaps the warning should be emitted when he commits.	12:42
doener	well, it's not just commit. It's also reset, rebase, am, ...	12:42
Arafangion	Surely they share code?	12:44
	← emilsedgh left	12:44
	→ d0k_ joined	12:45
	← dantrell left	12:45
	→ davidfetter joined	12:45
	→ malkomalko joined	12:47
	→ christophsturm joined	12:47
doener	Likely the update-ref part. But checkout is the right place for that warning. It's when you switch from a branch to a detached HEAD. There are legitimate reasons to use a detached HEAD even when creating new commits	12:49
Arafangion	Such as?	12:49
	→ advany joined	12:49
doener	For example, for temporary integration tests of a bunch of topic branches. Quite often I just detach my HEAD merge a bunch of branches, see how that works out and then switch back	12:50
	Or I create a branch for it _after_ I see that it worked out just as I expected	12:50
	Or I start hacking around and don't know yet what I'm actually about to create and don't want yet another funky_idea branch	12:51
	→ jm joined	12:51
Arafangion	Can't you delete the funky_idea branch when you're done with it?	12:51
	→ mw\|out joined	12:52
Arafangion	Or rebase.	12:52
	→ Fullmoon joined	12:52
	← d0k left	12:53
doener	Rebase? Makes no sense to me in that context. And sure, I could delete the branch later, but I just don't like to bother coming up with a sane name until I know I want to keep the stuff in it.	12:54
	If it's already referenced by a branch and I decide that I _do_ want to keep it, it's just too tempting for me to just keep the crappy foobar branch name.	12:55
	→ parasti joined	12:56
doener	if it's just referenced by HEAD (and the reflog), it's way more likely for me to think hard and come up with an actually useful name	12:56
	→ digitalpardoe joined	12:56
	← digitalpardoe left	12:57
	← charon left	12:57
Arafangion	As a question, is it possible to 'hide' branch names?	12:59
tcurdt	@doner, @Arafangion: Yeah, indeed - just wanted to make a build from that tag.	12:59
	← d0k_ left	13:00
	← akahn left	13:00
	← Arafangion left	13:01
	→ willb joined	13:01
	← bobmcw left	13:03
	← shodges left	13:03
	→ d0k joined	13:04
	← reel_ left	13:05
	→ Beket joined	13:09
	→ reel_ joined	13:09
	→ threeve joined	13:09
	← tcurdt left	13:09
	→ Zimi joined	13:10
	→ loiseau joined	13:10
	← Zimm left	13:15
	← zachinglis left	13:15
	← lorandi left	13:16
	→ zachinglis joined	13:17
	→ lorandi joined	13:20
	→ imyousuf joined	13:22
	← advany left	13:26
	← not-xjjk left	13:31
	→ billybubba joined	13:35
	→ boto joined	13:41
	→ rothwell_ joined	13:43
	← ankit9 left	13:44
	← BubbaT left	13:45
Pieter	git cherry-pick f6fdf6	13:45
	nice hash	13:45
	:)	13:45
	→ rwaite joined	13:48
	→ rothwell1 joined	13:48
	→ snorkel joined	13:48
	→ ben_h joined	13:50
snorkel	how can I tar or zip just the files listed by a "git ls-files -m" ?	13:50
	mw\|out → mw	13:50
	← priidu left	13:52
	← rothwell left	13:53
	← cedricv left	13:59
	snorkel notes the deafening silence	13:59
	← qrush left	13:59
davidfetter	hello	14:00
snorkel	hello	14:00
davidfetter	snorkel, that looks like a tar question more than a git one. the man page may help	14:00
	davidfetter unclear about git-shell. is there a way for somebody with a spankin' new git-shell account to create repos, or can they only manipulate existing ones?	14:01
	← rothwell_ left	14:01
snorkel	thought there might be a shorthand for the task	14:02
	→ gilimanjaro joined	14:02
davidfetter	no idea	14:02
	→ spearce joined	14:02
davidfetter	<-- git n00b	14:02
snorkel	<-- git n00b and unix/linux n00b	14:03
davidfetter	ah	14:03
snorkel	a.k.a windows user	14:03
davidfetter	well, the tar man page should help, then :)	14:03
PerlJam	snorkel: What's the output of "git ls-files -m"	14:03
	?	14:03
snorkel	a list of files	14:03
	→ cyberpear__ joined	14:04
snorkel	...list of files that have been modified	14:04
PerlJam	then that's more of a shell question than anything :-)	14:05
Lemurnomicon	tar jcvf modified.tar.bz2 --files-from <( git ls-files -m )	14:05
PerlJam	tar czvf file.tgz $(git ls-files -m)	14:05
	Or Lemurnomicon's way.	14:05
Lemurnomicon	tomayto, tomahto . . .	14:05
snorkel	Perl, Lem: thanks	14:05
PerlJam	Lemurnomicon: yours is better in that it doesn't suffer from command line length problems (if he has bunches of files)	14:06
Lemurnomicon	yeah, depends on how many potentially modified files and if your tar likes reading from a descriptor; of course if you're using zsh then change it to =( ... ) to have it create a real file	14:06
	← leethal left	14:06
	← spearce left	14:07
	→ spearce joined	14:07
	→ kishore joined	14:10
	→ paltman joined	14:13
	→ stouset joined	14:14
kishore	thiago: Are you the thiago from Qt software?	14:15
thiago	yes	14:15
kishore	If yes, im curious to know how qt migrated to git? As reference for my migration	14:15
thiago	we turned perforce read-only	14:16
	then everybody had to use Git	14:16
	:-)	14:16
kishore	hehe	14:16
	Ok i was more like how you organized the modules	14:16
thiago	there are no modules	14:16
	there's one big repository called "qt"	14:16
kishore	single tree?	14:16
thiago	it's now 940 MB	14:16
kishore	oh	14:16
thiago	yes	14:16
	27k files, checkout of over 300 MB	14:17
	it contains a few extra tools that are not part of the Qt release (like the Qt license key generator), but those are really minor	14:17
kishore	and externals like phonon? which are maitained in KDE svn?	14:17
thiago	they're inside our tree. No extra submodule.	14:17
kishore	a copy?	14:18
thiago	yes	14:18
kishore	just curious.. once this "big" tree is created, is is possible to split it on a later date?	14:19
thiago	yes	14:19
	we're investigating whether we should	14:19
kishore	but it is possible! thats a relief	14:19
thiago	the repository won't shrink though	14:20
	→ withanx joined	14:21
kishore	The problem i have is that some people do not want to checkout the stuff that they do not work on...	14:21
	and with git partial checkout is not possible AFAIK	14:21
	← snorkel left	14:23
PerlJam	kishore: sounds like someone should be using submodules.	14:23
	→ priidu joined	14:24
	→ jony1 joined	14:24
	← jony left	14:24
kishore	PerlJam: thats right! I just want to refer a to a good document that helps me migrate from SVN	14:24
	→ p0w3r3d joined	14:24
kishore	PerlJam: can you refer me something?	14:25
	hopefully something that highlights the possible pitfalls! :)	14:26
	← eternaleye_ left	14:29
	→ rdavila joined	14:29
	→ EmilMedve joined	14:31
	← willb left	14:32
kishore	when setting up submodules as described in http://www.kernel.org/pub/software/scm/git/docs/user-manual.html#submodules	14:37
	the commands refer to operations that should be done server side.. right?	14:37
	← mattf left	14:38
	← imyousuf left	14:41
	← cyberpear__ left	14:42
	→ cyberpear__ joined	14:43
	→ aroben joined	14:44
	← lorandi left	14:44
	→ lorandi joined	14:44
thiago	kishore: in our case, everyone always checked out this much	14:45
	so it's not a change for them	14:45
	what is a change is that now they can't do partial updates	14:45
	if they pull, they get all changes	14:45
kishore	yeah	14:45
	→ jackdempsey joined	14:46
	→ bobmcw joined	14:49
	← jony1 left	14:50
	→ djanowski joined	14:50
	← Beket left	14:50
djanowski	hi everyone. i tend to use git pull --rebase quite a bit. so i was thinking of making it the default. now i see the warning on the man page stating that --rebase is potentially dangerous if i already published my history. what does publishing mean? that somebody pulled from me?	14:51
dsop	djanowski: yes	14:51
	djanowski: if someone else has a copy of your repository.	14:52
djanowski	dsop: for this project we use git in a centralized way. we all push and pull to and from a main repository. so i shouldn't worry about --rebase then. that right?	14:53
dsop	rebase might result in rewritten commits, which means that they have different sha1s. If someone pushed a commit that's commit Id is later changed due to a rebase, a second pull from the repository will result in duplicated commits.	14:53
	→ rdavila_ joined	14:53
	← malkomalko left	14:54
dsop	djanowski: if you just rebase your local changes on top of what you pulled, it's fine.	14:55
djanowski	dsop: thanks a lot!	14:55
	→ asp joined	14:55
	→ p0w3r3d_ joined	14:56
dsop	djanowski: but if someone alreadyd have a commit from you, or if you rebase the pulled changes on your local changes, you will get in trouble. I never used git pull --rebase, but I guess it rebsaes your local branch on top of what you pulled, so if the local branch is not yet published it's fine.	14:56
	← zachinglis left	14:56
	→ abbe joined	14:56
djanowski	hm, i wonder if rebase will mask the real commit timestamps	14:59
	← mediogre left	14:59
	← jhelwig left	14:59
	→ cedricv joined	14:59
	→ icwiener joined	15:00
j_engelh	yes	15:00
	→ Juice10 joined	15:00
	← lorandi left	15:01
	→ lorandi joined	15:01
	← gebi left	15:01
	← cyberpear__ left	15:02
	← lorandi left	15:02
	← rmh3093 left	15:02
	→ lorandi joined	15:02
djanowski	k thanks	15:03
	← p0w3r3d left	15:05
	→ mattf joined	15:06
	← reel_ left	15:06
Juice10	hi, how do I import a remote branch into a local git repository?	15:07
offby1`	"git remote add some-name git://url/to/remote/branch"; then "git fetch some-name"	15:08
	← djanowski left	15:08
	→ djanowski joined	15:09
j_engelh	nada...	15:09
	you can't add a branch as a remote, can ya	15:09
	always the whole repo	15:10
	→ charon joined	15:10
Juice10	j_engelh, how can I add it?	15:10
doener	git fetch remote-repo refs/heads/remote-branch-name:refs/some/local/ref	15:10
	if it's just a one shot, single branch fetch	15:10
	← mcella left	15:10
	← sgrimm left	15:11
doener	otherwise just "git remote add some-name remote-repo" and "git fetch some-name" to get all branches from the remote as remote tracking branches	15:11
	→ juanjoc joined	15:11
j_engelh	look at the faq..	15:11
Juice10	j_engelh, I was looking trough the documentation but I couldn't find anything that makes any sense	15:12
j_engelh	hm	15:13
Juice10	I asked the same question a week ago here, and I got an answer that worked only I can't find any log files so I had to ask it again	15:13
	→ bdiego joined	15:13
Juice10	I can remeber it was something with `git branch refs/heads/master` but what exactly I don't know	15:14
	→ zachinglis joined	15:15
	← rdavila left	15:15
	← tango_ left	15:16
	→ willb joined	15:17
j_engelh	remote add + fetch + branch, so yeah so what :p	15:17
Juice10	ah I did find it after all, (my spotlight search is not working): `git checkout -b new-branch-name remote-name/branch-name`	15:18
	→ gebi joined	15:20
	→ cyberpear joined	15:20
	→ sgrimm joined	15:23
	→ alley_cat joined	15:23
	← quoin left	15:25
	→ jesselucas joined	15:26
Juice10	what does "git checkout: updating paths is incompatible with switching branches/forcing" mean?	15:27
jast	probably that checkout got confused over the arguments you gave it	15:28
	this might happen if, for example, you say something like git checkout -b foo this-is-not/a-branch	15:29
	→ woeye joined	15:30
Pieter	I like how you can push a merge in 700 bytes :)	15:32
j_engelh	yes, if the sets of files are distinct.	15:33
Juice10	jast, that right, thats what happend, I was looking in the wrong .git/config (the remote was named server in the "wrong" config and "origin" in the other)	15:33
	← bronson left	15:34
	→ etgryphon joined	15:35
	← trochala left	15:38
	← thecatwasnot left	15:40
	→ CodeOfficer joined	15:40
	→ bronson joined	15:40
	offby1` → offby1	15:41
	→ LiamH joined	15:42
	← schlort left	15:45
	→ schlort joined	15:45
	→ thecatwasnot joined	15:49
	← girishr left	15:51
	← threeve left	15:55
	← sgrimm left	15:56
	→ fhobia joined	15:56
	→ jclinton joined	15:56
jclinton	it seems like git-svn gets confused too easily or i'm always doing something wrong.	15:56
	i get this consistently: Last fetched revision of refs/remotes/origin/svn/trunk was r8124, but we are about to fetch: r7854!	15:56
	when trying to do svn-rebase	15:57
	← gilimanjaro left	15:57
	← cedricv left	15:57
jclinton	the only solution i've found is to recheckout all of the work and re-svn-init	15:57
	is there a better way? what is going on?	15:57
offby1	yikes!	15:57
	no idea :-\|	15:57
	never heard of that	15:58
	jclinton: I don't suppose your admin is doing something unusual with the svn repository -- restoring it from backup, e.g. ?	15:58
jclinton	this is with git-mirror.gnome.org	15:58
offby1	I assume that's a git repository, and your its administrator?	15:58
	s/your/you're/	15:58
jclinton	no, i'm just the consumer	15:58
	afaik, very few people are using the git mirror so far	15:59
	when you git clone git-mirror.g.o you get all the git-svn branches and such	16:00
offby1	sure	16:00
	so it's a _git_ repository?	16:00
jclinton	and it seems that as soon as you do anything locally, you get out of sync	16:00
	yea	16:00
offby1	If that's the case, I'm surprised you're using git svn at all, or seeing error messages from it.	16:00
jclinton	git-mirror is a git-svn mirror of svn.gnome.org	16:00
offby1	sure	16:01
jast	it's a bit non-intuitive to use git-svn mirrors properly	16:01
offby1	oh really?	16:01
	I'd have expected you could just pretend they were regular git repository.	16:01
	repositories.	16:01
jast	did you follow a howto or something like that?	16:02
	offby1, sure, as long as you don't want to commit back to svn directly, that's fine	16:02
jclinton	jast: http://live.gnome.org/GitForGnomeDevelopers	16:02
parasti	I thought the only use for git svn mirrors was to ease setting up your own git svn repo	16:02
	← bronson left	16:03
djanowski	how would i see the commits in one branch that are not in another?	16:03
jclinton	jast: what has been happening is that i have been doing git pull to stay up to date	16:03
	jast: and now that i try to do git svn rebase, it fails	16:03
jast	you can't use pull, sorry	16:03
parasti	djanowski: git log $another..$one	16:03
jast	git-svn needs to update its local data during updates	16:03
jclinton	jast: because i'm about to do git svn dcommit	16:03
	jast: so basically i should nuke the origin refs from .git/config ?	16:04
jast	what you can do is fetch	16:04
djanowski	parasti: thanks :-)	16:04
jast	but it won't always make things faster	16:04
jclinton	jast: as long as i don't merge the fetched?	16:04
jast	particularly if you fetch before their git mirror updates to the most recent version (the one containing your new commit)	16:04
	in fact i'm not sure if fetch gives you any advantage at all	16:05
	stouset → stouset\|away	16:05
offby1	jclinton: I wonder if the git-mirror is somehow out of sync	16:05
jclinton	offby1: it updates every 15 minutes	16:06
jast	well, it's unlikely it updates instantly	16:06
	→ tango_ joined	16:06
jclinton	offby1: the release i made yesterday showed up after about 5 minutes	16:06
jast	anyway, mixing git svn rebase and git pull (from a git-svn mirror) is evil :)	16:06
jclinton	ok, how do i get my git CWD back to a usable state?	16:06
	git svn fetch appears to be checking out the entire svn history again	16:07
jast	i suppose git-svn could be extended to deal with things in a smarter way	16:07
	did you do your own commits before or after the pull?	16:07
parasti	jclinton: it's probably confused by the history of the current branch	16:07
jast	(or even both?)	16:08
jclinton	jast: i have my own commits in a completely sep. branch now	16:08
	jast: master is clean	16:08
jast	so master only has some pull stuff in it?	16:08
	→ rmh3093 joined	16:08
jclinton	jast: correct	16:08
jast	okay	16:08
	see if you can spot the state before the pull(s) in git reflog master	16:09
parasti	aren't the remote tracking branches from the mirror intact, though	16:09
jclinton	jast: found it; very first entry in the reflog	16:10
	→ mithro joined	16:10
jclinton	jast: 63ffb60... HEAD@{51}: clone: from git://git-mirror.gnome.org/git/gnome-games	16:10
jast	just to be sure: the only things showing up in git reflog show master are that clone and a couple of pulls?	16:11
jclinton	jast: the only thing that concerns me is : 5b1aab4... HEAD@{40}: checkout: moving from master to aisleriot-clutter	16:12
	jast: if i lose that branch i will be in pain	16:12
jast	you used git reflog, right? that shows changes of HEAD, i.e. checkouts too	16:12
jclinton	yea	16:12
jast	when you use git reflog show master, you get the history of that branch only	16:12
	well, history is the wrong word, sort of, but you know what i mean :)	16:13
jclinton	oh i see	16:13
	ok so i have the correct reflog nw and it's fine to blow away everything in there	16:13
	jast: 63ffb60... master@{15}: clone: from git://git-mirror.gnome.org/git/gnome-games	16:13
	(instead of 51 commits back)	16:13
jast	okay, make sure master is checked out and do git reset --hard master@{thatentry}	16:13
	and then see if git svn rebase works now	16:14
	→ kilowattage joined	16:14
jclinton	warning: Log .git/logs/refs/heads/master has gap after Fri, 8 Aug 2008 10:27:51 -0500.	16:14
	otherwise done	16:14
	now git svn rebase?	16:14
	→ dgrazier joined	16:14
jast	i don't think that matters	16:14
	yeah	16:14
jclinton	cool, thanks	16:14
	→ niki joined	16:15
jast	oh, that side branch of yours, was it based on the fresh clone back then, or did you branch off after one of the pulls?	16:15
jclinton	the later case	16:15
jast	okay, so we may need to do a bit of fixing up there too	16:15
	but we'll see about that once the svn rebase is done	16:16
jclinton	rebase fails in the exact same way	16:16
	Last fetched revision of refs/remotes/origin/svn/trunk was r8124, but we are about to fetch: r7854!	16:16
	re-init svn?	16:17
jast	oh snap	16:17
	let's try something else first	16:17
jclinton	there appears to be some metadata lurking around	16:17
	→ Aikawa joined	16:17
	stouset\|away → stouset	16:17
jast	rename the file .git/svn/trunk/.revmap_something, optimally in a way that you can undo it later on	16:18
	← thecatwasnot left	16:18
	← kef left	16:19
jast	then try again	16:19
jclinton	jast: no revmap just an index	16:19
jast	hmm	16:19
jclinton	jast: boss wants to go to lunch right now	16:19
jast	what git version are you using?	16:19
jclinton	jast: so i'll have to continue later	16:20
	→ kef joined	16:20
jclinton	jast: 1.6.0.2	16:20
jast	highlight me when you get back	16:20
	hmm	16:20
jclinton	jast: cool, thanks	16:20
jast	well, we'll see about that later	16:20
jclinton	bbiab	16:20
	← Kbyte left	16:21
	→ carllerche joined	16:21
	→ leethal joined	16:22
	→ olauzon joined	16:23
	→ radarek joined	16:26
	→ c0sin joined	16:29
	← olly left	16:31
Pieter	http://source.android.com/submit-patches	16:32
	bt	16:32
	w	16:32
	Android's source is in Git	16:32
Jc2k	looks like it wasnt until a few hours ago tho	16:32
	their dbus repo looks kinda screwed	16:33
	apparently dbus is from 1970 or something..	16:33
	and their initial contribution is a series of mode changes...	16:33
Pieter	:)	16:34
	http://android.kernel.org/	16:34
	yeah, all 2 hours ago	16:34
Jc2k	http://android.kernel.org/?p=platform/external/dbus.git;a=summary	16:34
	1970-01-12 Upstream	16:34
	:P	16:34
Pieter	yeah, dbus took a while to get popular ;)	16:34
Jc2k	same for freetype	16:35
Pieter	I wonder what this _Gerrit_ tool is	16:35
	→ shenie joined	16:35
Jc2k	some kind of review board for patch submissions	16:35
	→ thecatwasnot joined	16:35
	← c0sin left	16:35
Pieter	http://git.source.android.com/?p=platform/manifest.git;a=commitdiff;h=fd8b112f12abb9dd18bb22fb6f9ca7f2a7633246 it's our spearce!	16:36
spearce	sssh.	16:36
	spearce is busy fixing some disaster related to that	16:36
spearce	but yes, me	16:36
Pieter	spearce: been a hectic day then?	16:37
spearce	:-)	16:38
	→ robtaylor joined	16:38
	← BarryCarlyon left	16:38
robtaylor	spearce: please tell me that its just a machine error that i can't see what changes have been made to dbus? :)	16:38
	→ twas joined	16:39
	← christophsturm left	16:40
	← asp left	16:40
Pieter	Recommit your edits using the --amend flag, for example:	16:40
	$ git commit -a --append	16:40
	heh	16:40
	→ jbailey joined	16:40
	→ iulian joined	16:42
	→ neuro_damage joined	16:43
	→ bronson joined	16:44
	← olauzon left	16:44
	← kilowattage left	16:44
	→ mithro_ joined	16:45
jast	Pieter, close but no cookie ;)	16:45
	← mithro left	16:45
	→ vmiklos joined	16:45
	mithro_ → mithro	16:46
	← fhobia left	16:46
	← nud left	16:47
	← hurikhan\|Work left	16:49
	→ kilowattage joined	16:50
	← kilowattage left	16:50
	← cyberpear left	16:51
	← niki left	16:51
	→ kilowattage joined	16:52
	→ mashmore joined	16:52
	← mashmore left	16:52
	→ mashmore joined	16:53
	← neuro_damage left	16:53
	→ chris2 joined	16:54
	→ neuro_damage joined	16:55
neuro_damage	when reverting back to a branch there's a conflict in the merge, how do I fix that?	16:55
	→ bentob0x joined	16:55
	← rdavila_ left	16:56
	→ xjjk joined	16:57
	← leo2007 left	16:58
	→ rdavila_ joined	16:59
	→ Beket joined	16:59
	aroben → aroben\|meeting	17:01
	← dgrazier left	17:01
	← makkalot left	17:04
	← koke left	17:05
	← Juice10 left	17:10
	← radarek left	17:10
	← p0w3r3d_ left	17:11
	← doener left	17:12
	→ doener joined	17:13
	→ niki joined	17:15
	→ litage joined	17:17
	← jfontan left	17:18
	→ Yuuhi joined	17:19
	→ joydivider joined	17:21
	← Jacolyte left	17:22
	← vmiklos left	17:23
	→ schacon joined	17:23
	→ vmiklos joined	17:23
	← d0k left	17:24
	→ ben_h_ joined	17:25
	stouset → stouset\|away	17:25
	→ d0k joined	17:26
	→ tewk joined	17:26
	stouset\|away → stouset	17:27
	← buckett left	17:27
schacon	@spearce: congrats on the release	17:28
spearce	thanks scott	17:28
	http://android.kernel.org/?p=tools/repo.git;a=summary	17:28
	repo	17:28
	← shenie left	17:28
spearce	you wanted it, you got it	17:28
tewk	git reset --hard HEAD doesn't reset "changed but not updated files" which just have \r\n differences, is that to be expectd?	17:28
Arjen	I'm gonna take a look at that code review tool	17:28
	→ d0k_ joined	17:30
	← ben_h left	17:30
	ben_h_ → ben_h	17:30
mithro	how does one reorder patches easily?	17:31
spearce	mithro: git rebase -i is your friend	17:31
mithro	spearce: I basically want to push a single patch without pushing all unpushed patches	17:32
	but it's not the next one	17:32
	← chris2 left	17:32
spearce	mithro: use rebase -i to reorder them so the one you want to push is earliest in the series	17:32
mithro	am I better creating a branch?	17:32
spearce	or make a branch and cherry-pick it into there	17:33
	→ p0w3r3d joined	17:35
	← zachinglis left	17:35
	← p0w3r3d left	17:35
	→ thewaltz joined	17:40
	→ perezd88 joined	17:41
	← perezd88 left	17:41
	→ perezd88 joined	17:42
	→ tongueroo joined	17:42
	→ threeve joined	17:45
	← d0k left	17:45
	→ shayarnett joined	17:45
	→ olly joined	17:46
	← niki left	17:46
shayarnett	is there anyway to get the filename and or path from a blob?	17:46
	or anywhere one can start digging online to find out about git disaster recovery :D	17:47
j_engelh	!gitforcomputerscientists	17:47
	but the reflog may be enough for you..	17:48
doener	shayarnett: git rev-list --all --objects \| grep $blob_hash	17:48
j_engelh	git fsck	17:48
	=> find your missing object	17:48
	→ _graham_ joined	17:48
doener	shayarnett: that might show you various filenames for the blob	17:48
j_engelh	well at least these are the start points	17:48
	doener: they won't	17:48
	blobs don't have filenames	17:49
doener	_if_ there are trees referencing the blob, of course	17:49
shayarnett	yeah i don't see any trees or commits	17:49
	only thing i can dig up in the repo is dangling/unreachable blobs :(	17:49
doener	j_engelh: thus _various_ names. rev-list walks from the commits to the trees and then to the blobs. And the blobs get their filename shown as recorded in the tree from which rev-list found them...	17:49
j_engelh	oh --objects is that magic flag :)	17:50
	← olly left	17:51
	→ tonguero_ joined	17:51
spearce	schacon: oh, you guys may also be interested in gerrit since its the code review server	17:55
Arjen	spearce: Can you run Gerrit outside of Google app engine?	17:56
	→ vbabiy joined	17:56
spearce	Arjen: no. its tied to the GAE APIs, though folks have been trying to build their own replacement for the APIs based on the open source GAE SDK.	17:56
	Gerrit is actually the Rietveld code review tool that already exists for GAE, but heavily patched	17:56
	so much so that it almost doesn't resemble it anymore. but that's why its on GAE, its what we started from	17:57
Arjen	Right, that immediately clears up the strange name :-)	17:57
	→ niki joined	17:57
tango_	android uses git: http://android.kernel.org/	17:58
	← tongueroo left	17:58
	← lorandi left	17:58
	→ lorandi joined	17:58
	← loiseau left	17:58
	← Ramblurr left	17:59
	← ia left	18:00
	→ ia joined	18:01
jclinton	jast: back.... we were at reset --hard for master head to the clone point	18:08
	jast: but that didn't fix git-svn	18:08
	→ scientus_ joined	18:08
	← mithro left	18:08
scientus_	how do i convert from bare to working ?	18:09
jast	so what do you have in .git/svn/trunk?	18:09
scientus_	i converted but now i cant push to it	18:09
	i moved everything to .git	18:09
	changed config so bare= false	18:09
jclinton	jast: i have .git/svn/origin/trunk/index and a log	18:09
jast	scientus_, edit .git/config and set bare to false	18:09
	oh :)	18:09
scientus_	and did git reset --hard	18:09
jast	hmm.	18:09
scientus_	and am getting fatal: '/home/git/fp.git': unable to chdir or not a git archive	18:09
	→ alexross joined	18:10
scientus_	does the config ignore #?	18:11
	cause its bare = false#true	18:11
	nope	18:11
	if you push to a working directory does the working copy update automatically?	18:12
	thats what i want	18:12
	aroben\|meeting → aroben	18:12
	← djanowski left	18:12
scientus_	so it can have a running server on it	18:12
	→ sgrimm joined	18:15
jast	scientus_, no, it doesn't. see faq non-bare	18:15
Gitbot	scientus_: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare	18:15
scientus_	than how should i do that?	18:15
	have a running server allways on head?	18:16
jast	read the faq entry	18:16
	jclinton, what does the svn section of your .git/config look like?	18:17
scientus_	where do iput the binary file?	18:17
	sh file	18:18
	← shayarnett left	18:18
scientus_	ok iread it sry	18:18
jast	:)	18:18
scientus_	i understand why it is not automatic	18:19
	and then i put the sticky bit on the server script (its rails), should i instead jsut restart it when there is a commit?	18:19
jast	what has this got to do with rails?	18:20
	→ statim joined	18:21
scientus_	just the control of restarting the server	18:21
	from anyone who can push to it	18:21
jast	what server?	18:21
scientus_	the script/server in the working directory	18:21
	which is mongrel	18:21
jast	oh, so you're trying to use the repository contents as a rails application?	18:22
scientus_	yeah	18:22
	so the working head is running and can be tested	18:22
jast	well, i don't know if it's a good idea to automatically restart it for each commit but yeah, i guess that's the most straightforward way to do it	18:23
jclinton	jast: sorry was afk, looking	18:23
jast	though i do think you can configure the rails thing to automatically pick up changes in files	18:23
scientus_	yeah i would like to avoid that	18:23
jast	i haven't looked at it for a long time	18:23
scientus_	i guess i could right a few sticky-bit scripts	18:23
jclinton	jast: url = svn+ssh://svn.gnome.org/svn/gnome-games, fetch = trunk:refs/remotes/origin/svn/trunk	18:23
	etc	18:23
	→ cedricv joined	18:23
jast	i wonder where the origin dir comes from, i don't have that	18:24
jclinton	jast: git svn init -s svn+ssh://svn.gnome.org/svn/modulename --prefix=origin/svn/	18:24
jast	ah	18:24
jclinton	jast: from the howtoo	18:25
	jast: and you already have those remote branches after cloning the git-mirror.g.o	18:25
jast	jclinton, something else that might help is to update .git/svn/.metadata (backup recommended)	18:27
	so see what the highest revision number in git log origin/svn/trunk is and try putting that in there	18:28
jclinton	jast: ok	18:30
	→ magnus joined	18:31
doener	jclinton: I just read up. So you cloned the git-mirror.g.o repo, and used both, "git svn rebase/fetch" as well as "git pull" to update?	18:31
scientus_	im actually not sure where to put the post-update sh file, in .git?	18:32
jast	doener, yep	18:32
jclinton	doener: yes	18:32
doener	jclinton: which git version?	18:32
jclinton	doener: latest	18:32
jast	scientus_, .git/hooks. server side.	18:32
	and make it executable.	18:32
jclinton	doener: so trying to fix it so i can get back to using only svn rebase	18:33
doener	jclinton: "latest" == git.git master? or next? or v1.6.0.2?	18:33
jclinton	1.6.0.2, doener	18:33
doener	jclinton: ok. git-svn is probably confused because you have commits (from pull) that are newer than what it thinks it has	18:34
	jclinton: in git.git master, there's support for partial revmap rebuilds which, I guess, also helps in your situation	18:34
jclinton	doener: yea, so we tried resetting head using --hard to the commit at initial clone	18:34
jast	we already killed those, from the branch at least	18:34
	*local branch	18:34
doener	jclinton: that fixes neither the metadata nor the remote branches. At least one needs to be fixed	18:35
jclinton	doener: i see	18:35
jast	oh	18:35
doener	jclinton: if you cannot update, removing the revmap files from .git/svn/* might be enough to cause a full rebuild	18:35
jast	jclinton, when you looked for the rev_map file, did you consider it's a dotfile, i.e. hidden? (grasping at straws here)	18:35
jclinton	doener: well at this point i have 1 branch that i need to save; any way to put all those commits aside and just reclone git-mirror.g.o and the merge that branch back in?	18:35
	jast: i'll use find this time, a moment	18:36
doener	jclinton: yeah, you can do that with rebase. But I'd start out with removing the .revmap files	18:36
jast	shell completion should pick it up if you type the dot yourself	18:36
	we'll still have to fix the topic branch, probably, because it's partially based on pulls	18:37
jclinton	i have some rev maps	18:37
doener	jast: as long as they were fast-forwards, there's no problem. And otherwise a "git svn rebase" should fix it	18:37
jclinton	i'll nuke them	18:37
jast	good point	18:37
doener	jclinton: after nuking them, use "git svn fetch", _not_ rebase.	18:38
	rebase might get confused again ;-)	18:38
jclinton	ok doing it	18:38
	hum it did a whole bunch of stuff	18:39
doener	once the fetch completed, rebase should work again	18:39
ben_h	hi all	18:39
jclinton	and now it's appears to be repulling the entire revision history	18:39
jast	a complete fetch? that'll take ages	18:39
jclinton	which will take 4-5 hours	18:39
doener	jclinton: the revmap rebuilding is pretty verbore	18:39
jast	revmap rebuilding is pretty fast, actually	18:40
ben_h	i'm doing a git rebase -i <tag name> to condense several commits into one. when i edit the commit message to change 'pick' to 's' for each commit and save, I get the following error: grep: /Users/ben/projects/twippet/current/.git/rebase-merge/done: No such file or directory	18:40
jast	and it has different messages than fetching	18:40
jclinton	r7477 = 388493aba3fde81fe8f0fddd32d8c4069724331d	18:40
ben_h	any idea what's causing that?	18:40
doener	s/bore/bose/	18:40
jclinton	was the revmap rebuild	18:40
	lots of lines like tha	18:40
	t	18:40
	→ mcella joined	18:40
jclinton	now i have lines like this: Done rebuilding .git/svn/origin/svn/gnome-2-4/.rev_map.1eca851f-d525-0410-b55f-d2baa8e04955	18:40
	Found possible branch point: svn+ssh://svn.gnome.org/svn/gnome-games/trunk => svn+ssh://svn.gnome.org/svn/gnome-games/tags/GNOME_GAMES_2_3_8, 3338	18:40
	which appears to be a complete fetch in progress	18:40
scientus_	how do i list all remote reops?	18:41
jast	i doubt it	18:41
doener	jclinton: lots of "rXXXX = YYYYYY" lines without anything in between == revmap rebuild	18:41
jast	scientus_, git remote	18:41
scientus_	doesnt list anythgin	18:41
	→ rdavila__ joined	18:41
jclinton	doener: yea, that was at the very beginning	18:41
jast	jclinton, if it was a complete refetch, it'd have to have been pretty damn quick if you're already at 3338	18:41
jclinton	doener: appears to be done	18:41
	but now it's refetching all	18:41
scientus_	and i have a repote repo added via git remote add orgin ssh://blah	18:41
jast	you can always look at the config file if all else fails	18:42
doener	jclinton: is it refetching stuff you already had, or maybe just new stuff?	18:42
	→ flaguy joined	18:42
	→ loiseau joined	18:43
	→ asp joined	18:43
doener	wow, the g-m.g.o gitweb is fast as a glacier	18:44
	→ BubbaT joined	18:45
doener	ok, I admit it, glaciers are a lot faster than that one...	18:46
Pieter	works fine here	18:49
doener	took 6 minutes for me	18:49
	← billybubba left	18:51
	→ zachinglis joined	18:51
	← mcella left	18:52
	← alley_cat left	18:54
doener	jclinton: hm, the "git svn init" command from that page gives you a setup that doesn't match the one for the git mirror	18:54
	← rmh3093 left	18:55
doener	jclinton: as long as you only use "git svn rebase", that doesn't matter. But "git svn fetch" thinks that you actually don't have any of the tags yet... I think...	18:55
	→ alley_cat joined	18:56
doener	jclinton: can't test that from here though, don't have any ssh access tehre	18:57
	s/tehre/there/	18:57
	← kilowattage left	18:58
	→ markelikalderon joined	19:02
jclinton	doener: sorry i had to answer phone	19:02
	← rdavila_ left	19:02
	→ Sigma joined	19:02
jclinton	doener: i'm not sure but w/e it was doing with those old revs it's finished	19:02
	doener: so i think i'm set	19:02
	jast: i need to fix my broken branch though	19:03
	→ david_koontz joined	19:03
jast	try if git svn rebase inside it works	19:03
doener	jclinton: running "git svn rebase" on the "broken" branch should work now. You might need to skip a bunch of commits (there might be a few duplicates from the pull), but it's unlikely IIRC	19:04
	unlikely that you need to do the skipping that is	19:04
	you might get some conflicts though, if you had merge commits (don't remember if you used "git pull" or "git pull --rebase")	19:05
jclinton	ok trying it	19:05
	i created a new branch off broken just in case	19:05
	called "fix"	19:05
	→ WaW joined	19:06
	→ shep-werk joined	19:06
shep-werk	howdy all	19:07
	← Weaselweb left	19:07
shep-werk	is there a way to define a custom diff, but in relation to the git repository?	19:07
	→ ebel joined	19:07
jclinton	like change the default formatting of git diff, shep-werk?	19:07
	← Fullmoon left	19:08
	← alley_cat left	19:08
jclinton	doener: hop damn, it worked	19:09
doener	great :-)	19:09
jclinton	doener: your recomendation would be to delete the origin from .git/config so i don't accidentally git pull ever again?	19:10
j_engelh	no	19:10
	just don't use pull, hmk?	19:10
jclinton	well that's easier said than done when i have purely git projects that i work on	19:10
	and once you git pull, you're fucked	19:10
	like i was today	19:10
doener	jclinton: if you could upgrade to git.git master (or maybe just wait for 1.6.0.3? dunno if it will have the git-svn fix)	19:10
	oops	19:10
jclinton	doener: ok	19:10
doener	anyway, if you can upgrade, do that, then setup "git pull" to use rebase (there's some branch.<name>.whatever config setting)	19:11
jclinton	doener: awesome	19:11
	doener, jast: thanks for the help	19:11
doener	jclinton: to get updates for your current branch "git pull" should be fine then, and git-svn should not get confused anymore	19:12
jast	it doesn't use svn rebase though, does it :)	19:12
jclinton	doener: that will be awesome	19:12
	looking forward to that	19:12
shep-werk	jclinton: not quite - i want to preprocess the files, then do a diff	19:12
	so I tested my script with the external-diff variables	19:13
doener	jast: no. Still using the git-mirror thing, but with the git-svn fix, that should no longer break "git svn rebase". At least that's how I read the commit message	19:13
jast	would be great	19:13
doener	jclinton: you might also want to try to change the "tags" line in your svn-remote config section, so that it stores the tags in refs/tags/svn/* instead of refs/remotes/origin/svn/tags, because the git-mirror repo does it like that	19:15
	jclinton: avoids the "double fetch" you have seen	19:15
jclinton	doener: k	19:16
scientus_	how do i make users commiting break the owner situation in my git repo	19:20
jclinton	scientus_: like file permissions?	19:20
shep-werk	jclinton: but I would like to put my diff program into the repository, so it goes along with it	19:20
scientus_	yes	19:20
jclinton	scientus_: there's a group mode that uses GID's	19:20
	scientus_: that's for a case where you don't have one person responsible for a release, like a release manager	19:21
	scientus_: not really what git was designed to make possible but still usable that way	19:21
scientus_	for whoever pushes it should be the same user, git takes care of the security and who commited to what right?	19:22
jclinton	security is just file ownership	19:22
	and ssh	19:22
scientus_	but shouldnt git init --shared do it right?	19:22
jclinton	yes	19:23
	as well as setting the directory perms correctly	19:23
	→ daniel-w_ joined	19:23
jclinton	shep-werk: there might be a hook, that would be the place to look	19:24
	shep-werk: what you want is highly custom, though	19:24
	→ tjafk1 joined	19:25
	← ben_h left	19:27
	stouset → stouset\|away	19:28
	→ gittish joined	19:29
	→ Fullmoon joined	19:29
shep-werk	jclinton: its not that custom - all the support for custom diffs via gitattributes is there	19:30
	its just the ability to say "use the program relative from the root of the git checkout" that is odd	19:30
	→ radarek joined	19:30
MaddHatter	If I do a commit --amend and realize after starting the editor that I didn't mean to do that, is there a way to abort?	19:34
	usually exiting the editor without savings causes an abort, but not with commit --ammend	19:35
	← woeye left	19:35
	→ eternaleye joined	19:36
	→ onitunes joined	19:36
jclinton	MaddHatter: ctrl-z, kill	19:36
	← daniel-w left	19:36
magnus	shep-werk: how about using $(git rev-parse --git-dir)/path/to/diff	19:37
	stouset\|away → stouset	19:38
magnus	shep-werk: not sure if it is evaluated by the shell, but might be	19:38
	shep-werk: or --show-cdup	19:38
	← robin_ left	19:39
	← niki left	19:39
	← tjafk2 left	19:41
	→ ereslibre joined	19:43
	← cedricv left	19:44
	→ mcella joined	19:44
	← mcella left	19:44
	→ alb joined	19:46
	→ Sham joined	19:46
	← Sham left	19:46
	← priidu left	19:46
	→ ben_h joined	19:47
	→ releod joined	19:48
	→ RobotDeathSquad_ joined	19:48
	→ theutogon joined	19:48
RobotDeathSquad_	Is it possible to "undo" a 'git commit --amend' ?	19:49
	→ priidu joined	19:49
releod	Does anyone have the command handy to take an existing SVN repo, and import it into a new GIT repo? I want to get rid of the SVN repos, and convert them all to GIT.	19:49
	→ niki joined	19:49
shep-werk	magnus: thanks for the suggestion, but it doesn't seem to expand	19:50
	magnus: what does --cdup go with?	19:50
	→ dgrazier joined	19:51
magnus	shep-werk: it is an option to git rev-parse as well	19:52
WaW	what is ORIG_HEAD?	19:53
shep-werk	magnus: what vers of git do you have?	19:53
	magnus: that just outputs "--cdup"	19:54
vuf	WaW: HEAD before you merged	19:54
	← iulian left	19:54
	← albertito left	19:54
	← CodeOfficer left	19:57
	→ CodeOfficer joined	19:57
	← bbs__ left	19:57
WaW	vuf: and how long it is usable? until next commit?	19:58
	→ qrush joined	19:58
vuf	WaW: I don't know, but would guess until next merge	19:58
WaW	ah, right	19:58
	thanks	19:59
	and what is MERGE_HEAD then?	20:00
	→ trotek joined	20:01
parasti	ORIG_HEAD is apparently also used by rebase and am	20:01
	a recent man git-rev-parse should have documentation for these	20:02
Gitbot	parasti: the git-rev-parse manpage can be found at http://git.or.cz/man/git-rev-parse	20:02
	→ Sonderblade joined	20:03
	← icwiener left	20:05
WaW	thanks, that is i searched for	20:05
	← theutogon left	20:09
	← etgryphon left	20:11
	onitunes → onitunes\|away	20:13
	← lorandi left	20:13
magnus	shep-werk: sorry, it is git rev-parse --show-cdup	20:15
	← Tv left	20:17
	→ icwiener joined	20:20
	← johnw left	20:20
gittish	RobotDeathSquad_: git reset --hard HEAD@{1}	20:21
	← Vortex34 left	20:26
	← Sigma left	20:26
	→ kahmalo joined	20:28
	← sbeyer left	20:29
kahmalo	I get an error when I try to convert gnus from CVS to git	20:29
	git cvsimport -v -kk -d :pserver:gnus@cvs.gnus.org:/usr/local/cvsroot -r cvs.gnus.org gnus	20:29
	Fetching lisp/gnus-uu.el v 1.1	20:29
	Unknown: error	20:29
	cvsps version 2.1	20:29
	git-cvsimport from 23abd3f48cb217d1558fa1984bfa8c502717c08f	20:30
Gitbot	[git 23abd3f48]: http://tinyurl.com/5mx2hv -- Merge branch 'js/objc-funchdr'	20:30
kahmalo	how do I fix the error?	20:30
	stouset → stouset\|away	20:31
kahmalo	I'll try running cvsps separately, to find out whether the error happens there	20:32
	← ben_h left	20:32
	→ cory joined	20:37
	← lamont left	20:37
	← ndim left	20:37
	cory → Guest15740	20:37
	← rwaite left	20:37
	tonguero_ → tongueroo	20:38
	→ lamont joined	20:38
	Guest15740 → bantic	20:39
kahmalo	"cvsps --norc -u -A --root :pserver:gnus@cvs.gnus.org:/usr/local/cvsroot gnus > cvsps.out" succeeds	20:39
	"git cvsimport -v -d :pserver:gnus@cvs.gnus.org:/usr/local/cvsroot -r cvs.gnus.org -P cvsps.out gnus" fails quickly	20:39
	← priidu left	20:40
	← d0k_ left	20:41
	← icwiener left	20:41
	stouset\|away → stouset	20:42
kahmalo	Apparently, "Unknown: error" means the CVS server responded "error" and git-cvsimport doesn't know why.	20:42
	→ priidu joined	20:42
kahmalo	ah there, strace shows the real error: E cvs [checkout aborted]: unexpected EOF reading RCS file /usr/local/cvsroot/gnus/lisp/gnus-uu.el,v	20:43
	dwmw2 → dwmw2_gone	20:44
kahmalo	cvs -d :pserver:gnus@cvs.gnus.org:/usr/local/cvsroot co -r 1.1 gnus/lisp/gnus-uu.el gives the same error.	20:44
	So does it mean the CVS repository is corrupt?	20:45
cradek	sure looks like	20:45
kahmalo	I wonder why cvs rlog doesn't complain.	20:46
	Perhaps I can edit the output of cvsps to remove the lost versions?	20:46
cradek	if the RCS file is truncated, who knows what will work and what won't	20:47
	should restore that file from backup	20:47
	← eternaleye left	20:47
	→ MarcWeber joined	20:48
	→ warlock_mza joined	20:48
MarcWeber	Is there an easy way to reverse the git history of changes? I mean if you have commits 1 -> 2 -> 3 get a repo having 3 -> 2 -> 1 ?	20:48
j_engelh	--reverse	20:48
	YOU'D NEVER BET! :p	20:49
MarcWeber	git what reverse? git rebase ?	20:49
	I'd like to get a fast darcs -> git converter and I know that removing the last patch is a fast operation when using darcs..	20:50
	→ eternaleye joined	20:50
j_engelh	MarcWeber: git log --reverse, git-rev-list --reverse, ...	20:50
	← ereslibre left	20:50
MarcWeber	But there is no git rebase -i (pipe buffer to tac) :w ?. I'll try that then. Thanks	20:51
gittish	EDITOR=tac?	20:52
j_engelh	lol	20:52
	→ ndim joined	20:52
	→ eternaleye_ joined	20:54
MarcWeber	Using that would result in some conflicts.. So I'll checkout HEAD^ each time while committing to another branch using --revere or such	20:55
drizzd	is there any way to `git grep' the current work tree if I alias it to git grep --cached?	20:57
kahmalo	I thought aliases could not override builtin commands.	20:58
drizzd	possibly, I didn't try	20:58
	so I guess my question is moot	20:58
	← tcoppi left	20:59
	onitunes\|away → onitunes	20:59
	→ tcoppi joined	20:59
	← pantsman left	20:59
drizzd	I just wish git grep --cached were the default	21:00
kahmalo	It seems all 1.* revisions of lisp/gnus-uu.el files are corrupt. I edited the cvsps output file and now the conversion seems to be proceeding.	21:01
	→ gitster joined	21:01
	→ cyberpear joined	21:01
kahmalo	Ohloh also saw the corruption: http://www.ohloh.net/topics/593/	21:01
MarcWeber	Is there an automatic addremove command in git?	21:02
gittish	git add -A	21:02
MarcWeber	:-) And I thought I knew git..	21:02
w00t_	can I have a git repository inside a git repository?	21:03
MarcWeber	w00t_: Yes. git submodule	21:03
	→ shenie joined	21:03
w00t_	(preferably .gitignored or similar so it doesn't get added to clutter up the first one)	21:03
gitster	or not submodule.	21:03
kahmalo	MarcWeber: git add -A was not in 1.5.6.2	21:03
hachi	so... how does git svn know where to dcommit back to?	21:04
w00t_	i'm not sure submodule would quite do it, the projects (repositories) while being loosely related as a dependancy are quite different	21:04
gitster	you can have a subdirectory that is independent repository that is not submodule.	21:05
hachi	does `git svn dcommit` always commit back to trunk?	21:05
MarcWeber	w00t_: Yes, you can have "untracked" files which in turn can be git repos	21:05
gitster	you can do e.g. "git init; echo sub >.gitignore; mkdir sub; cd sub; git init" and make sure you do not tell the toplevel about the "sub" directory.	21:05
hachi	the conceptual trunk in subversion	21:05
	← spearce left	21:06
MarcWeber	hachi: yes, I'm not sure how it does it, but it worked for me all the time	21:06
hachi	well, I'm wanting to commit to a branch :\	21:06
	→ spearce joined	21:07
hachi	a subversion branch	21:07
MarcWeber	hachi: Then you need to track that	21:07
	hachi: git branch mylocalbrach svnbranchname --track	21:07
hachi	okay	21:07
w00t_	gitster: oh, that's exactly what i'm after - assuming .git/config's ignore stuff can also handle it	21:07
hachi	roger roger, I understand... I had no clue that git svn followed that	21:08
w00t_	(that's just "what we use" unfortunately)	21:08
tewk	:q	21:08
MarcWeber	hachi: cat .git/config. There you'll find a line in section [svn-remote ...] looking like this:	21:09
	fetch = nixpkgs/trunk:refs/remotes/trunk	21:09
	→ Tv joined	21:09
hachi	yup	21:09
	oh!	21:09
	okay	21:09
MarcWeber	hachi: But .. how does it know that master tracks trunk? I don't know yet	21:09
hachi	I'll try it out on something	21:10
	I bet it just goes to trunk if it's not marked otherwise	21:10
	← cyberpear left	21:10
	→ rtomayko joined	21:10
	← bobmcw left	21:11
kahmalo	lisp/gnus-uu.el 4.20 seems to be the oldest preserved revision.	21:12
	→ cyberpear joined	21:12
	← eternaleye left	21:12
	← pehlert left	21:14
w00t_	gitster: hmm, 'git status' seems to then be picking up 'state' from the primary repository, not the one in the subdirectory	21:15
	← eternaleye_ left	21:16
w00t_	(whilst in the subdirectory, of course)	21:17
	oh, my mistake I think	21:17
	← spearce left	21:17
	← diask left	21:17
	→ diask joined	21:18
mugwump	hachi: it knows by looking back at the commit log	21:18
	ergo, you can't currently dcommit merges	21:18
	← parasti left	21:19
	← jackdempsey left	21:19
	→ spearce joined	21:20
	← lucsky left	21:20
	← _eric left	21:21
	→ thannoy joined	21:22
	→ felipec joined	21:22
hachi	mugwump: that's fine, I just want to commit back to a branch, not to trunk	21:22
	stouset → stouset\|away	21:27
	← shenie left	21:27
	← mattf left	21:27
	← flaguy left	21:30
dsop	sometimes i whish the man branch of git would have a decent makefile	21:32
scientus_	how do i make git push orgin master work immediately after git clone	21:32
	or whatever	21:32
	you still have to alias it i guess	21:32
	← vbabiy left	21:33
	→ _eric joined	21:35
felipec	scientus_: that's how it works right now	21:36
scientus_	hm mdidnt work, maybe cause the server hostname isnt in dns?	21:36
	← radarek left	21:37
felipec	scientus_: what url did you use?	21:37
	→ drizzd_ joined	21:37
scientus_	ssh://s/	21:38
	still didnt work if i pulled by ip	21:39
	i mean cloned	21:39
felipec	scientus_: what does git remote -v shows?	21:39
scientus_	fatal:not a git repository	21:40
	whoops	21:40
	git remote -v	21:40
	origin ssh://192.168.15.99/home/git/fp	21:40
	shawn@shawn-desktop:~/playground/fps/fp2/fp$ git push orgin master	21:40
	fatal: 'orgin': unable to chdir or not a git archive	21:40
	fatal: The remote end hung up unexpectedly	21:40
	hmm that looks right	21:40
felipec	schacon: s/orgin/origin/	21:41
vmiklos	scientus_: ^ :)	21:42
felipec	scientus_: er, that was for you	21:42
	← alexross left	21:42
scientus_	i think i spelled it wrong every time i manually entered it :D	21:42
	lol	21:42
	← Beket left	21:42
scientus_	thx vmiklos	21:43
	← boto left	21:45
	→ foca joined	21:45
	→ bbuss joined	21:47
	← Aikawa left	21:48
dsop	hmm I get a pack has bad object at offset 147800: failed to apply delta. index-pack failed.	21:48
	when pulling from linus tree	21:49
	→ deskin joined	21:53
	→ albertito joined	21:53
dsop	hmm just with next. mabe it's that index issue that is fixed with some patches on the mailinglist, but not yet pushe by junio	21:54
vuf	probably	21:55
	← juanjoc left	21:55
	← drizzd left	21:57
	→ mattf joined	21:58
	← mecha left	21:58
	← dgrazier left	22:00
	→ shenie joined	22:01
	← Sonderblade left	22:02
gitster	the fix's been there for the past 15 hours ;-)	22:02
vuf	probably not, then :)	22:03
	← bdiego left	22:03
	← EmilMedve left	22:07
	← alb left	22:09
	→ eternaleye_ joined	22:14
foca	why could it be that suddenly I started getting this after saving on vim when doing a rebase -i:	22:14
	".git/.dotest-merge/git-rebase-todo" 14L, 470C written	22:14
	Could not execute editor	22:15
	if I try with textmate I get this: mate: failed to establish connection with TextMate. but I guess that's more of a textmate issue than a git issue	22:16
	← thannoy left	22:17
	← zachinglis left	22:17
	→ paltman_ joined	22:19
	← rhaamo left	22:19
tango_	scientus_, btw if you're using bash I found that the command line completion does wonder to prevent misspellings	22:20
	← paltman_ left	22:20
deskin	foca: I take it git is refusing to use what vi claimed to save?	22:20
foca	deskin: yup, looks like it	22:20
	→ WALoeIII joined	22:21
deskin	someone had a problem like this a week ago	22:21
	try vi ; :q ; echo $?	22:22
	are you using macvim?	22:22
foca	exit status is 1	22:22
	← ph^ left	22:22
deskin	yeah, that's the symptom	22:22
	← bbs left	22:22
foca	nope, just from terminal	22:22
	← markelikalderon left	22:23
foca	hm, any ideas why is it exiting with a 1?	22:23
deskin	no :) http://colabti.org/irclogger/irclogger_log/git?date=2008-10-14,Tue#l38 has this other user's troubles: onigiri was the nick	22:23
	you can try setting a different editor, or playing with your .vimrc to see if vi will exit with 0	22:24
foca	it didn't do this a while ago, I think it started happening when I started using screen....	22:24
	← rdavila__ left	22:25
	→ rhaamo joined	22:26
deskin	maybe 'set nocompatible' in your .vimrc?	22:27
	doesn't seem to change anything for me	22:27
foca	heh, that's the first line	22:28
	deskin is on Ubuntu, though	22:28
	← Ryback_ left	22:28
deskin	googling 'vim exit code' is interesting	22:28
	← withanx left	22:28
	← _graham_ left	22:30
	← felipec left	22:30
	→ mashmore_ joined	22:32
	→ SRabbelier joined	22:33
	← paltman left	22:35
	→ olly joined	22:35
	← WALoeIII left	22:37
	→ WALoeIII joined	22:37
	← jclinton left	22:39
vmiklos	GIT_VERSION = 1.6.0.2.607.g21337.dirty	22:41
	hm, i have a 1337 version :)	22:41
j_engelh	a dirty one :p	22:41
vmiklos	sure:)	22:41
	don't worry, most of my local modifications and up on the list as patches ;)	22:42
	end up*	22:42
gitster	1.6.0.3 tomorrow, by the way.	22:43
	← leethal left	22:46
vmiklos	'git rev-list v1.6.0.2..junio/maint\|wc -l' says 112 (i guess most maint releases have less updates)	22:46
	→ cannonball joined	22:47
	→ markelikalderon joined	22:47
j_engelh	why so complex	22:47
	git describe junio/maint ;)	22:47
gitster	--no-merges would be more appropriate, though...	22:48
j_engelh	0 :p	22:48
	← mashmore left	22:48
	← jesselucas left	22:50
	← mashmore_ left	22:51
	← bentob0x left	22:56
	→ paltman joined	22:57
	→ cmacleod joined	22:58
	← cmacleod left	22:58
	← shenie left	22:59
	← LiamH left	23:00
	→ parasti joined	23:00
	← bantic left	23:00
	← markelikalderon left	23:01
	← threeve left	23:04
	← eternaleye_ left	23:04
	← carllerche left	23:06
	→ carllerche joined	23:06
	← Yuuhi left	23:07
	→ vbabiy joined	23:07
	→ markelikalderon joined	23:10
	→ jfontan joined	23:12
	→ daxelrod joined	23:17
	← jm left	23:18
	← asp left	23:18
	← malahal__ left	23:20
	← ebel left	23:21
	→ flaguy joined	23:22
	← daxelrod left	23:24
Pieter	anyone who knows how I can move a window in screen to a different position?	23:24
MarcWeber	What do you mean by position ? nr3 to nr2 ? or top to bottom when having split screens	23:25
	(In either case I don't know.. )	23:25
Pieter	nr3 to 2	23:25
	didn't know you could split the screet ;)	23:25
	screen, that is	23:25
MarcWeber	Try ctrl-a S	23:25
	ctrl-a tab cycles	23:25
Pieter	oh, I don't want that	23:26
j_engelh	tk?	23:28
	→ madewokherd joined	23:28
	← _kid_ left	23:29
Pieter	C-a :number N	23:30
	→ t3mp3st joined	23:31
t3mp3st	what is the easiest way to recover the last committed version of a file that has been accidentally deleted in the working copy (the file exists in HEAD)?	23:32
Pieter	git checkout HEAD -- path/to/file	23:32
t3mp3st	ah, cool; I was trying to use reset for some reason :(	23:32
Pieter	yeah, reset and checkout have very similar semantics	23:32
	← pygi left	23:33
t3mp3st	(wow, git tab completion is incredible)	23:33
	zsh expanded the tree object	23:33
	→ cedricv joined	23:34
	→ Flathead joined	23:34
	← t3mp3st left	23:37
	← nikolasc left	23:38
	← cmarcelo left	23:40
	← charon left	23:40
	← aroben left	23:42
	← whee left	23:42
	← bbuss left	23:43
	← olly left	23:45
	→ alb joined	23:45
	→ listrophy joined	23:45
	→ mikem joined	23:46
listrophy	git tag confuses me. it doesn't seem to do anything	23:47
dsop	listrophy: what is your problem?	23:49
listrophy	i do git tag [tagname] and nothing seems to happen, other than prevent me from doing git tag [tagname] again	23:49
dsop	listrophy: what did you expect?	23:50
listrophy	i dunno. i'm pretty new to scm	23:50
	i want to tag a version of my code	23:50
j_engelh	well you just did that then	23:50
listrophy	do i have to do something special when i commit?	23:51
dsop	listrophy: try git tag and see if you tag is there	23:51
	listrophy: no,you just tagged it	23:51
	which means, you put a marker on your current state of your repository.	23:51
listrophy	dsop: it's there, but when i clone it from somewhere else, the tag isn't there	23:51
mikem	hi, i've developed a feature in the branch new_feature. I'd like to port this feature to release_1_0, so i made a branch port_new_feature from release_1_0. both new_feature and release_1_0 share an ancestor way back. part of new_feature is a new subdirectory, which is what I want to copy from new_feature into port_new_feature. however, new_feature also has changes to other files which I don't want. what's the best way to do this? i realize i may not be exact	23:51
dsop	listrophy: clone it from the repository where you tagged it, or from an upstream repository where you first oushed to?	23:52
listrophy	i tag, commit, push, then clone into a different directory	23:52
dsop	listrophy: ifyou push, you have to say explicitly git push --tags	23:53
listrophy	ah bugger	23:53
dsop	...I even wonder why it behaves that way	23:53
j_engelh	or git push v1.0	23:54
listrophy	ok, so in my 'cloned' directory, how do i switch to the tag i made?	23:54
dsop	mikem: try to use git cherry-pick -n and than unstage the already staged files taht you don't want to have in and than commit	23:54
listrophy	just check it out?	23:55
dsop	mikem: but beware, you than have 2 different commits on your feature branches.	23:55
	listrophy: git checkout <tagname>	23:55
listrophy	dsop: alright. thanks	23:55
mikem	dsop: ok. well, the alternative is i just copy the directory and `git add` the new files and go from there	23:55
	← listrophy left	23:56
	→ onitunes_ joined	23:56
	→ pjhyett joined	23:59

Logs Search ←Prev date Next date→

Channels Documentation