IRCloggy #git 2008-08-11

Provider of IRC logs since 2005.

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

	2008-08-11
	← chuckr left	00:01
	→ carrus85 joined	00:05
	← rmh3093 left	00:05
	← bts- left	00:06
	→ rmh3093 joined	00:07
	← sgrimm left	00:12
Mikachu	i haven't watched this yet but http://free-electrons.com/pub/video/2008/ols/ols2008-james-bottomley-git.ogg	00:13
j_engelh	Bottomley is very good at explaining things	00:14
	I still have his FreedomHEC2007 presentation	00:15
	"Practical Introduction to quilt and git"	00:15
	← mtodd_ left	00:16
Mikachu	yeah i have seen that one	00:17
bremner	j_engelh: is it about using them together?	00:19
j_engelh	no I do not think so. (Even if it were, you should be using stgit.)	00:20
	← BarryCarlyon left	00:21
bremner	j_engelh: yeah, I'm not too sure yet. I didn't get a smooth workflow (debian packaging). But maybe I should give it another try.	00:22
	^it=stgit	00:22
j_engelh	no premade packages?	00:22
bremner	j_engelh: no I mean, I was messing with stgit to make debian packages	00:23
	← perezd88 left	00:25
	→ doener_ joined	00:29
	→ ricky_ joined	00:32
	← ricky left	00:32
	ricky_ → ricky	00:32
Mikachu	i don't know where he gets his numbers from though	00:33
	he claims linux-2.6.git is 2GB fully packed, which is an order of magnitude too much :)	00:33
	← dragonball_ left	00:34
	← ereslibre left	00:35
	← swtaarrs left	00:40
	foca → foca\|postre	00:41
	← doener left	00:41
	→ ereslibre joined	00:42
	→ tswicegood joined	00:47
	→ _eric joined	00:47
	← NemesisD left	00:48
	→ igorgue joined	00:48
	← rmh3093 left	00:50
_eric	how would I find the latest revision that a file existing in?	00:50
	(assuming I had rm'd it)	00:50
johnw	i'd git log --stat \| less and search for the filename	00:51
_eric	heh	00:51
Mikachu	you can git log -- thefile	00:53
	if it doesn't work, try adding --diff-filter=D before the --	00:53
_eric	neat	00:53
	→ offby1 joined	00:54
	← Oompa left	00:56
	← kukks left	00:56
johnw	ooh, I like diff-filter	00:57
	→ rmh3093 joined	00:59
	← juliano left	01:00
	→ pvh joined	01:01
	→ Oompa joined	01:01
	← Gitzilla left	01:01
	→ offby1` joined	01:02
	← offby1 left	01:03
	offby1` → offby1	01:03
	→ offby1` joined	01:06
	← rmh3093 left	01:06
	← zawir left	01:07
	foca\|postre → foca	01:08
	→ rmh3093 joined	01:09
	→ Gitzilla joined	01:09
	→ offby1`` joined	01:09
	← offby1`` left	01:09
	← markkalderon left	01:10
	→ Oompa_ joined	01:10
	← tswicegood left	01:10
	← pvh left	01:10
johnw	i'm trying to figure out external diff helpers	01:11
	→ pvh joined	01:11
johnw	I have a [diff "araxis"] cmd = ... in my .gitconfig, and in .gitattributes I have "* diff=araxis", and I'm using git diff --ext-diff FILE, but still it uses the normal internal diff	01:11
	what piece am I missing?	01:11
	foca → foca\|sebazzzgae	01:13
	foca\|sebazzzgae → foca	01:13
	← Oompa left	01:13
	Oompa_ → Oompa	01:13
	→ swtaarrs joined	01:13
	→ giom joined	01:16
	← ramontayag left	01:17
	→ mw joined	01:17
	← hpa left	01:18
	→ supercali joined	01:18
	→ ramontayag__ joined	01:20
	ramontayag__ → ramontayag	01:20
	← offby1 left	01:21
	← Ademan left	01:23
	← offby1` left	01:24
	→ david_koontz joined	01:24
	→ k4r1m joined	01:27
	← jackdempsey left	01:28
	→ ereslibre_laptop joined	01:29
	→ perezd88 joined	01:30
	← ereslibre left	01:31
	← kanru left	01:32
	→ bongiojp joined	01:33
	→ fhobia joined	01:37
	→ patroy joined	01:39
patroy	Hi, how can you display what remote branch a local branch is tracking?	01:39
cbreak	maybe it's somewhere in git config -l	01:40
Rhonda	patroy: cat .git/config ducks :)	01:41
	→ jackdempsey joined	01:42
cbreak	why does git use human readable config files instead of xml?	01:42
	I thought human readable is out of fashion	01:42
Mikachu	probably because linus kills anyone in the same room who ever used xml	01:42
	→ advany joined	01:42
Rhonda	xml does require much more for parsing for the tools, too.	01:42
patroy	thx	01:42
Rhonda	It's neither human readable nor really helpful ressource wise.	01:43
	← swtaarrs left	01:44
	→ Modius joined	01:44
Mikachu	"Yeah, I know. I love the config file format (quite frankly, anybody who thinks XML and friends are sane is a total moron and should be shot before they reproduce), but the whole parsing code was a really quick hack. "	01:45
	← lll_ left	01:45
Rhonda	Like I said. ;)	01:46
	... or you, right.	01:47
	→ pvh_ joined	01:48
	← pvh left	01:48
	← aroben left	01:48
	← codeshot left	01:52
	← ereslibre_laptop left	01:54
	← flaguy left	01:55
	← pvh_ left	01:57
	→ pvh joined	01:57
	← patroy left	01:57
	→ dragonball_ joined	01:59
	→ kukks joined	02:00
	← perezd88 left	02:07
	→ kanru joined	02:07
	← giom left	02:09
	← david_koontz left	02:09
	→ NemesisD joined	02:09
	→ david_koontz joined	02:10
	→ giom joined	02:10
	← NemesisD left	02:13
	→ perezd88 joined	02:14
	→ garrin joined	02:14
	→ nikbin joined	02:15
	← nikbin left	02:15
	← advany left	02:15
	→ cilly joined	02:16
	← mw left	02:18
	← cbreak left	02:18
	← pjhyett_ left	02:22
	→ krawek joined	02:23
	→ david_koontz_ joined	02:23
	→ spearce joined	02:26
	← dragonball_ left	02:28
	→ crab joined	02:31
	→ chrisb joined	02:33
	→ arun joined	02:35
johnw	in several ways, XML feels like the Java of data formats	02:40
crab	bright, happy, and beautiful?	02:40
	DrNick2 → DrNick	02:40
johnw	haha	02:40
	← david_koontz left	02:40
johnw	exactly	02:40
	← pvh left	02:45
	→ pvh joined	02:45
	→ ggapol joined	02:48
	→ tswicegood joined	02:50
	→ imyousuf joined	02:53
	← imyousuf left	02:53
	→ LostInTheWorld joined	02:53
	→ david_koontz joined	02:58
	→ morii_ joined	02:58
	← Doska left	03:01
	→ pvh_ joined	03:02
	← pvh_ left	03:02
	← pvh left	03:03
	→ astubbs joined	03:06
	← LiamH left	03:07
	← morii left	03:10
	morii_ → morii	03:10
	← thannoy left	03:10
	← david_koontz_ left	03:13
	← lea_w left	03:15
	← david_koontz left	03:19
	← perezd88 left	03:20
	← carrus85 left	03:21
	← igorgue left	03:24
spearce	arrrgh. i _hate_ debugging transport related code.	03:27
	quickfetch always screws me up.	03:27
	← giom left	03:29
	→ igorgue joined	03:30
	→ aroben joined	03:33
	← LostInTheWorld left	03:36
	← vbgunz_ left	03:36
	→ bcarlyon\|laptop joined	03:38
	← pifel left	03:39
	→ cskaterun joined	03:43
	← bcarlyon\|laptop left	03:43
	← cskaterun left	03:43
	→ bcarlyon\|laptop joined	03:44
	← tlyng left	03:46
	← kukks left	03:46
	→ spearce` joined	03:48
	← spearce left	03:48
	← sofar left	03:49
	← spearce` left	03:50
	→ LostInTheWorld joined	03:51
	→ spearce joined	03:51
	← spearce left	03:52
	→ spearce joined	03:52
	← spearce left	03:52
	→ spearce joined	03:52
	← aroben left	03:54
	→ pygi joined	03:55
	→ makeworkeasy joined	03:56
	← kenpratt left	04:00
	← madewokherd left	04:04
	← tswicegood left	04:05
	← cilly left	04:09
	→ cilly joined	04:10
	→ carrus85 joined	04:11
	→ flask- joined	04:12
	→ jlilly joined	04:12
flask-	Okay, question: I've got a repository that contains the base functionality of a web CMS I've built. I forked a branch for a new site I'm developing. I've made a series of commits to that branch, representing some site-specific development but also some changes to the core functionality. I want to get the core functionality changes back into master, leaving out the changes to site-specific files. Whats the best course of action?	04:21
jlilly	flask-: I would imagine you would use 'chunks' to pick which bits stay and which bits go.	04:22
flask-	okay, where can I learn more about 'chunks' ?	04:22
jlilly	... which, iirc, is commiting interactively.	04:22
flask-	would that be --rebase interactive ?	04:23
jlilly	I don't think its rebase... as that would revert things iirc.	04:23
	I've only done it a few times and my mind is really clouded with HG at the moment.	04:23
	sorry :-/	04:24
flask-	;)	04:25
spearce	flask-: are the commits isolated to "stuff for core" and "stuff for site"?	04:25
	can you just cherry-pick the core commits onto the master branch?	04:26
	← tjafk1 left	04:26
	→ tjafk1 joined	04:26
flask-	no, the commits are mixed	04:27
spearce	fun. are the changes isolated to "site specific" and "core" files at least?	04:30
flask-	yeah	04:34
spearce	so assuming you have master checked out:	04:34
	for each commit you want to cherry-pick:	04:34
	git cherry-pick --no-commit $commit_sha1	04:35
	git status	04:35
	git checkout HEAD -- $site_specific_paths	04:35
	git commit -c $commit_sha1 -e	04:35
	→ mario_ joined	04:35
	← pygi left	04:35
spearce	basically what i am suggesting is to go through the changes one by one, apply them with cherry-pick, then back out the site specific files you don't want.	04:36
flask-	i don't recognize the syntax on the 'git checkout HEAD' command	04:36
	in fact I never quite figured out the whole "HEAD" thing	04:36
spearce	you may need to use git rm at the same time as git checkout to clear newly added site specific files.	04:36
	so go read the manpage for git checkout.	04:37
	but the summary of that command is "copy from commit HEAD paths $site_specific_paths into the index and the working directory"	04:37
	now HEAD is just an alias for the last commit on the current branch.	04:37
	→ alenoosh joined	04:37
spearce	so you are effectively copying the last committed state for a given file path back into the index and working directory, overwriting anything cherry-pick may have done.	04:38
flask-	and $site_specific_paths is just a space-delimited set of paths to site-specific files?	04:38
spearce	yes	04:38
flask-	is there a way to set that variable in .gitconfig so that i don't have to enumerate it manually each time?	04:39
spearce	or directories, it operates recursively on a directory if you give it a directory	04:39
	no	04:39
	what you are trying to do here is not something git can magically figure out for you.	04:39
flask-	it's not magic if it's the same files each time	04:40
spearce	oh, then just store them in a shell variable and use that as the argument to git checkout	04:40
	passing a file which isn't modified won't cause any errors	04:40
	but passing a "core" file path would be a bad idea, as the core change would be lost	04:41
flask-	why would it be lost? passing it would be including it in the commit, right?	04:41
	oh, n/m	04:42
	the "--" is for exclusion	04:42
spearce	uhm, no. but good thought.	04:42
flask-	erm ok...	04:43
	lost	04:43
	→ lenards joined	04:43
flask-	you had me until "passing a 'core' file path would be a bad idea"	04:43
lenards	question - how do you remove a remote branch? Is this github link wrong: http://github.com/guides/remove-a-remote-branch	04:44
spearce	lenards: that link is actually correct	04:44
	← rmh3093 left	04:44
spearce	flask-: in the loop i showed you we are using git checkout HEAD -- $site_path to toss away any changes made to $site_path by the commit we had previously cherry-picked.	04:45
	mario_ → pygi	04:45
flask-	OHHHHH	04:45
spearce	since the point of this is to keep the changes made to a core path, passing a core path to git checkout would toss the changes, and uh, isn't what you wanted here, yes?	04:45
lenards	spearce: I don't understand why it works... why would `git push` be how you remove a remote branch?	04:45
flask-	spearce: okay so lets say the current state of my topic branch is the result of 10 commits. do I need to cherry-pick in every commit?	04:46
spearce	lenards: git push is the only command that can make changes to a remote repository. it can create branches, update branches, and it was overloaded to delete branches, since its the only outgoing communication command.	04:46
	→ rmh3093 joined	04:46
spearce	flask-: yes, and in reverse order. meaning the order shown by `git log --reverse master..topic`	04:47
	← makeworkeasy left	04:47
flask-	Ouch	04:47
pygi	spearce, greetings	04:47
spearce	hi pygi	04:47
lenards	spearce: thank you... okay - but the syntax shown looks like the same way I'd push changes to that remote branch	04:47
	← foca left	04:47
spearce	flask-: you can try to automate this, if it is always the same paths:	04:48
	← pygi left	04:48
	→ pygi joined	04:48
flask-	spearce: okay, so indulge me a moment if you would. Knowing what I want to accomplish, if I were to start over, is there a workflow you can see that makes more sense?	04:48
spearce	for c in $(git rev-list --reverse master..topic); do git cherry-pick --no-commit $c && git checkout HEAD -- $site_paths && git commit -c $c; done	04:48
	but i'd do that on a temporary branch (git checkout -b tmp master) just in case it screws up badly. :)	04:48
	so moral of the story here is don't make core changes intermixed with topic branch specific changes.	04:49
	make the core branch changes on the core branch, and the site specific changes on a site specific topic branch.	04:49
Mikachu	\|\| break; done if you want to be neat :)	04:49
spearce	use git rebase on the topic branch to update it with the latest-and-greatest from the core/master branch.	04:49
	lenards: if you look carefully at the syntax the source side of the argument is empty (the stuff before the :). we are technically pushing _nothing_ to the remote branch to cause its deletion.	04:50
flask-	spearce: okay that does make sense	04:51
lenards	spearce: hmmm... okay so if I was pushing to a remote branch to update it... how would it look? `git push origin localbranch:remotebranch` ?	04:51
spearce	lenards: correct	04:52
	and its common for localbranch and remotebranch to be the same name, so "git push origin A" is an shorthand for "git push origin A:A"	04:52
lenards	spearce: excellent - thank you	04:52
	← chrisb left	05:04
	← spearce left	05:05
	← krawek left	05:05
	← jackdempsey left	05:05
	← johnw left	05:07
	→ imyousuf joined	05:07
	← jerbear left	05:10
	→ digitalhobbit joined	05:14
hacim	~.	05:14
	→ ismell joined	05:17
	→ DavidBoman joined	05:20
	→ cedricv joined	05:26
	← LostInTheWorld left	05:26
	→ x77686d joined	05:29
	← rmh3093 left	05:31
	→ rmh3093 joined	05:34
godlygeek	how do you show what branches exist in a remote repository that you just cloned?	05:35
	→ ph^ joined	05:35
godlygeek	ooh, nevermind - "git branch -r" does that.	05:38
	← Boman left	05:38
	fargiolas\|afk → fargiolas	05:40
	← xenoterracide left	05:41
	→ DrFrasierCrane joined	05:43
	← ramontayag left	05:50
astubbs	godlygeek, -a is also usefull	05:50
	also gitk --all	05:50
	→ ramontayag__ joined	05:54
	ramontayag__ → ramontayag	05:54
	→ vbgunz joined	05:56
	→ LostInTheWorld joined	05:58
	→ xenoterracide joined	06:04
	← Oompa left	06:06
	← rmh3093 left	06:08
	← DrFrasierCrane left	06:08
lenards	is there a way to get a post-pull hook?	06:08
	I know there isn't one in .git/hooks	06:09
	← mithro left	06:11
	→ ehc joined	06:13
	→ mithro joined	06:13
	← parasti left	06:13
	← ehc left	06:16
	→ kixxx joined	06:16
	← lenards left	06:16
	← kixxx left	06:17
	→ ehc joined	06:17
	→ DrFrasierCrane joined	06:20
	→ charon joined	06:20
	→ kixx joined	06:23
	→ ramontayag__ joined	06:23
	← carrus85 left	06:28
	→ drizzd joined	06:28
	→ rmh3093 joined	06:29
	← x77686d left	06:33
	← DrFrasierCrane left	06:37
	← bobesponja left	06:39
	← ramontayag left	06:39
	→ reel_ joined	06:39
	← ciaran left	06:40
	← drizzd_ left	06:40
	← charon left	06:44
	← igorgue left	06:47
	← bdiego left	06:53
	→ Eludias joined	06:55
	← normalperson left	06:56
	→ danten joined	06:56
	← Fullmoon left	07:03
	← fhobia left	07:03
	→ Fullmoon joined	07:03
	→ ben_h_ joined	07:06
	← digitalhobbit left	07:06
	→ normalperson joined	07:07
	→ rraasch joined	07:07
	← k4r1m left	07:07
	→ charon joined	07:10
	← bugfux left	07:11
	→ lea_w joined	07:15
	→ diask joined	07:16
	← towski left	07:17
	metze_away → metze	07:19
	← EgS left	07:19
	← Sput left	07:19
	→ EgS joined	07:19
	← pdherbemont left	07:20
	← bcarlyon\|laptop left	07:20
	← kixx left	07:21
	→ jackdempsey joined	07:22
	← skrit left	07:22
	→ towski joined	07:24
	→ _kid_ joined	07:25
	→ Marmouset joined	07:27
	→ lhunath joined	07:28
lhunath	is there a way to see the branches layout as shown by gitk and others in the console?	07:29
	I've no X server running on the box.	07:29
tango_	lhunath, git-forest http://dev.computergmbh.de/gitweb.cgi?p=hxtools;a=blob_plain;f=bin/git-forest	07:30
	→ kenpratt joined	07:30
lhunath	tango_: lovely.	07:31
tango_	lhunath, you're lucky, I just came across this yesterday. I haven't tried it either, so let me know if it works 8-)	07:32
lhunath	Heh.	07:32
	will do.	07:32
charon	lhunath: also git log --graph	07:32
	← Fullmoon left	07:32
lhunath	charon: interesting.	07:33
charon	(in 1.5.6-rc2 afaics)	07:33
	ah, even 1.5.6-rc0.	07:34
	→ bartocc joined	07:37
	← Pupeno-G left	07:40
	← towski left	07:40
	→ Fullmoon joined	07:42
	→ shadowfax joined	07:42
	dwmw2_gone → dwmw2	07:44
	→ evan_ joined	07:44
	→ BabelO joined	07:44
	← kenpratt left	07:45
	← lhunath left	07:47
	→ giom joined	07:49
	← Signum left	07:50
	→ deavid joined	07:55
	← reel_ left	07:55
	→ jmesnil joined	07:56
	← sverrej left	07:58
	← ehc left	08:01
	← LostInTheWorld left	08:01
	→ ereslibre joined	08:02
	→ towski joined	08:04
	← ereslibre left	08:04
	→ thiago joined	08:07
	← supercali left	08:10
	→ statim_ joined	08:16
	← statim left	08:16
	← Fullmoon left	08:16
	← priidu left	08:17
	→ dstuxo joined	08:20
	→ priidu joined	08:22
	← pygi left	08:26
	← alenoosh left	08:26
	→ alenoosh joined	08:27
	→ ciaran29d joined	08:28
	← shadowfax left	08:30
	← jackdempsey left	08:30
	→ ereslibre joined	08:33
	← towski left	08:35
	→ pygi joined	08:37
Araneidae	I'm confused. I was tracking the head of a repository, but now I don't know where I am anymore. I'm not on a branch, and I don't know how to get back! I have no branches...	08:38
	Araneidae still hasn't grasped some basics	08:38
doener_	"git branch" shows what?	08:38
Araneidae	Just says "* (no branch)"	08:38
doener_	nothing else?	08:38
Araneidae	That's it	08:39
doener_	heh, seems you dropped your local branches	08:39
	and "git branch -r"	08:39
Araneidae	origin/HEAD, master, stable	08:39
	I guess I want one of those...	08:39
doener_	did you have any local commits?	08:39
Araneidae	No	08:39
	Long merged	08:39
doener_	that's origin/HEAD, origin/master and origin/stable, right?	08:40
	ie. all have the origin/ prefix	08:40
Araneidae	Yes, I was being a ilittle lazy	08:40
	← Marmouset left	08:40
doener_	ok, then just "git checkout -b master origin/master" to recreate a local branch called "master" that starts off from "origin/master"	08:40
	(and to check it out at the same time)	08:41
Araneidae	Cheers. `git status` now tells me about three files that are untracked. I guess I need one of the extra checkout flags to lose those	08:41
	(They're files I've never touched)	08:42
doener_	no, you need "git clean" to get rid of untracked files	08:42
charon	are junio's topic branches available somewhere?	08:42
	i.e. the branch heads, i have the commits.	08:42
Araneidae	doener_, thanks very much.	08:43
doener_	no	08:43
	problem	08:43
	oops :-)	08:43
	→ resmo joined	08:44
Araneidae	How do I tell git to remove a large block of files? I can use `git add .` to add files, but `git rm` needs to be told which files to remove. I've ended up in the past processing the output from `git status` to figure out what to pass to `git rm` which doesn't seem right.	08:45
doener_	"git add -u" updates all tracked files in the index	08:45
	so if you deleted a tracked file in the working tree, "git add -u" will delete it in the index	08:46
Araneidae	Oh	08:46
	I didn't understand that from the man page.	08:46
	That's exactly what I want!	08:46
	Have to say the man page doesn't say that at all	08:47
doener_	Araneidae: hm, what's unclear about "Update only files that git already knows about, ... and marking deleted files for removal"?	08:47
Araneidae	Hmm. My copy doesn't have the last half of that.	08:48
	git version 1.5.4.3	08:48
doener_	it's in there since 1.5.4.6	08:48
	Araneidae evidently needs to update	08:48
doener_	that's true anyway, we're almost at 1.6.0 ;-)	08:49
Araneidae	I'm running the Ubuntu distribution here, guess I could usefully build a local copy.	08:49
	← dstuxo left	08:50
	ciaran29d → ciaran	08:50
doener_	Araneidae: yeah, not such a bad idea. or maybe the package from backports.org, if Ubuntu likes those	08:51
	→ dstuxo joined	08:51
Pieter	8~.	08:51
doener_	Araneidae: btw if you build from a git.git clone, use "make quick-install-doc", not "make install-doc", the latter takes ages	08:51
	← dstuxo left	08:51
	→ draco_ joined	08:51
draco_	hi	08:52
	I'm adding my untracked files with git-add but then git-diff does not show them	08:52
	→ dstuxo joined	08:52
Araneidae	backports.org looks interesting	08:52
	→ shingara joined	08:52
shingara	Hi	08:52
draco_	what shuld I do to add files preperly?	08:52
shingara	I have a problem with git-svn	08:52
Pieter	git add file?	08:52
doener_	draco_: "git diff" shows the diff of the working tree against the index	08:53
Pieter	draco_: that's desired behaviour	08:53
draco_	Pieter, i'm doing that, and git-status shows that file as well	08:53
shingara	git-svn want commit my master update in a tags and not in a trunk	08:53
Pieter	try git diff HEAD or git diff --cached	08:53
doener_	draco_: "git add" already added the file to the index, so you want a diff of the index against HEAD	08:53
shingara	the branch remote/trunk exist and it's in good place	08:53
draco_	hmm	08:53
	doener_, that worked	08:53
	so HEAD is _real_ repo and index is mine?	08:54
	I'm a little confused	08:54
doener_	HEAD is a (symbolic) ref pointing to the commit or branch that you have checked out	08:54
	the index lies between your working tree and whatever you have checked out.	08:55
draco_	does git-pull changes index or my working tree ?	08:55
doener_	"git add" "stages" changes in the index, and on "git commit" the new commit in created from the contents of the index	08:55
Araneidae	Interesting: I have a regression. `git svn clone <url>` works just fine with 1.5.4.3 (my Ubuntu version), but fails miserably with 1.5.5 (built from source on RHEL)	08:56
	(Two completely different machines being used here, on different networks.)	08:56
doener_	draco_: "git pull" is fetch+merge. The merge part of course affects your working tree and index	08:56
charon	Araneidae: why build 1.5.5 manually when there's 1.5.6.5 already?	08:57
Araneidae	Did that a few months ago	08:57
	Araneidae has too many machines to deal with	08:57
doener_	Araneidae: IIRC the early 1.5.5 had some breakage from git-svn optimizations	08:57
draco_	thanks	08:57
Araneidae	Guess I'd better update that and see if a new build fixes it. Good to know it's a known issue	08:58
doener_	Araneidae: same is true for 1.5.6 and 1.5.6.1	08:58
Araneidae	ahhh	08:58
doener_	but the only sane choice at the moment is 1.5.6.5 (or higher) anyway	08:58
Araneidae	right. I see why Ubuntu sticks with 1.5.4.3!	08:59
doener_	no, that's because the froze the versions of all programs in the release you're running at some point and that happened to pick 1.5.4.3	08:59
Araneidae	Could be -- except they're constantly publishing updates to so many components	09:00
	The Ubuntu update process really annoys me: wait a few weeks and it can be a couple of 100MB for an update!	09:00
	Araneidae prefers the Debian stick in the mud approach	09:01
doener_	heh, with debian sid, you can wait a few days and have several 100mb of updates ;-)	09:02
	← evan_ left	09:04
	← lu_zero left	09:04
	← ramontayag__ left	09:09
	→ sverrej joined	09:11
	← draco_ left	09:14
	→ ramontayag__ joined	09:15
	ramontayag__ → ramontayag	09:15
	→ chris2 joined	09:22
	← ramontayag left	09:22
	→ Boman joined	09:23
	← ben_h left	09:24
	← ben_h_ left	09:24
	fargiolas → fargiolas\|afk	09:25
	← dmiles_afk left	09:26
	→ ebel joined	09:27
	→ tvw joined	09:28
	→ MK_FG joined	09:28
	→ warthog9 joined	09:29
	fargiolas\|afk → fargiolas	09:34
	→ Zyroth joined	09:35
	← niv_ left	09:35
	← lea_w left	09:35
	→ shadowfax joined	09:35
Zyroth	If I run `$ git clone . ssh:// ...`it actually makes a new directory "ssh:" in my current repository with a clone in it... how but I obviously want to be the clone on a server... what am I doing wrong?	09:36
	→ abizern joined	09:37
charon	Zyroth: clone makes a local copy of a remote repository. 'man git-push' for the other direction	09:37
	→ bcarlyon\|laptop joined	09:37
Zyroth	thanks!	09:37
Magnus`	does anyone know a way to find a commit which contain a given tree hash?	09:39
	→ reel_ joined	09:39
	→ ben_h joined	09:40
	→ ben_h_ joined	09:40
charon	Magnus`: git rev-list HEAD \| while read commit; do git ls-tree $commit \| grep -q <hash> && echo $commit; done	09:41
	add 'break' if you're only interested in the first one, i guess	09:41
	will also find blobs, not just trees	09:42
Magnus`	thanks, I can use something like that	09:42
charon	ah, damn, that's nonrecursive, let me see	09:42
doener_	maybe with --all instead of HEAD, if the tree is not known to be in your current branch	09:43
charon	yeah	09:43
	also, make it 'git ls-tree -r -t ...'	09:43
	that is recursive and still shows the 'tree' entries	09:43
	aha, seems 'git ls-tree -r -d' shows only the trees, not the other entries. that might be faster.	09:44
	← shingara left	09:44
Magnus`	i am just interested in the tree of a specific path, so it's a bit simpler	09:45
doener_	if the tree is the toplevel one, this is faster: git rev-list --pretty=format:%T --all \| grep -B1 $tree_hash	09:45
Magnus`	it's not the toplevel...	09:45
	→ icwiener joined	09:46
	← DavidBoman left	09:48
	→ DrFrasierCrane joined	09:54
	→ markkalderon joined	09:57
	→ DavidBoman joined	09:59
doener_	Magnus`: do you know which pathname the tree belongs to?	10:01
	Magnus`: anyway: git rev-list --all \| git diff-tree -t --stdin \| grep '^$[^:]\\|:04$'	10:02
	Magnus`: that lists all commits along with the tree changes that happened in those commits	10:03
	you can add, for example, " -- somedir" to the rev-list command to limit the output to commits that had changes in "somedir"	10:04
	and it's reasonably fast. About 1.5s to list all tree changes from git.git	10:05
	ehrm, 5s. 1.5s was for Documentation/ only	10:07
	for comparison, the shell scripting loop took 2m34s ;-)	10:09
charon	diff-tree -t is a nice trick, i'll write that down somewhere	10:10
	← mithro left	10:13
doener_	oh, log just needs --raw to make -t work.	10:14
	→ tlyng joined	10:14
doener_	git log --raw -t --pretty=oneline \| grep '^$[^:]\\|:04$'	10:14
	that's even 20% faster. just a little less than 4s for the whole of git.git	10:15
	fargiolas → fargiolas\|afk	10:15
	← thiago left	10:15
	← alenoosh left	10:16
doener_	shows abbreviated tree hashes though, and I'm too lazy to look up how to get those in full. So I'll leave that as a lesson for the reader ;-)	10:18
	→ compubomb joined	10:20
	← MK_FG left	10:21
compubomb	how do you get started setting up a git:// once you have your git repo setup ?	10:21
	i mean, what do you have to do to be able to use the git_url ?	10:21
	← Boman left	10:21
doener_	git-daemon	10:21
	← schlort left	10:22
	→ schlort joined	10:25
	← ebel left	10:26
	→ sbeyer joined	10:27
	crab wonders wtf to say about the cherry-pick/rerere change.	10:28
	← ereslibre left	10:30
	← giom left	10:30
pasky	I like the commit message :)	10:33
	→ snitko joined	10:34
	← pygi left	10:34
	→ pygi joined	10:34
	→ mithro joined	10:37
crab	thanks.	10:46
	← markkalderon left	10:48
	← CodeOfficer left	10:49
	→ bartocc_ joined	10:50
	← chris2 left	10:50
	→ lea_w joined	10:52
	→ Sput joined	10:58
	→ wildfire joined	10:58
wildfire	hi	10:58
	git commit is telling me I have 'some suspicious patch lines'	10:58
	but those lines are okay, how can I force git commit to commit this stuff?	10:59
	→ cbreak joined	10:59
doener_	you probably enabled the pre-commit hook, which checks for whitespace noise	11:00
	→ thiago joined	11:00
doener_	either disable that hook, fix the whitespace noise, or use --no-verify (IIRC, check git-commit(1))	11:00
wildfire	doener_, ahh - yes; thanks '-n' to git commit was it	11:00
	doener_ would have fixed the whitespace breakage instead :-)	11:01
cbreak	a whitespace noise checker? excelent!	11:01
	where?	11:01
doener_	.git/hooks/pre-commit	11:01
	just chmod +x it	11:01
	→ jacobat joined	11:01
doener_	the example hook only contains the whitespace thing	11:02
avu	wildfire, using git on windows by any chance?	11:02
doener_	oh, maybe rename from pre-commit.sample to pre-commit, that was changed lately	11:02
avu	doener_, ah, very good. so the hooks won't be all executed by default on windows anymore	11:02
doener_	right	11:03
	→ hml joined	11:03
doener_	not sure if that's 1.6.0 stuff or already in 1.5.6.x	11:03
hml	maybe this question itself is wrong; i want to setup a git repo for the sole purpose of having data pushed to it on a 10 minute base; like a time stamp of a dir; anyone else doing this? I find that i often don't git commit often enough; and if i had this time stamped autocommit, it may help me	11:04
wildfire	doener_, these are sample input data --- I want it to have the appropriate problem so that the unit tests work	11:04
cbreak	nice	11:04
	← schlort left	11:04
doener_	cbreak: that hook used to have a perl script in it, the new one uses "git diff --check" instead, which respects the core.whitespace settings	11:04
	again, not idea in which version that was added/changed	11:05
	← imyousuf left	11:05
cbreak	both seem to work anyway.	11:07
	← bartocc left	11:10
	→ zawir joined	11:10
	→ Zoup joined	11:10
	→ Aerpe joined	11:11
Zoup	i did create a patch of files ( both modified and new added ) and git-applied them , but only modified files are visible in git-status and new files are not created at all	11:11
	→ dennister joined	11:11
	← bartocc_ left	11:13
dennister	any ubuntu hardy users in here? having real trouble using git and documentation to compile a custom kernel with speakup patch(es)	11:13
	i'd really appreciate some help...used to compiling gentoo kernels with /usr/src/linux instead	11:14
	→ schlort joined	11:15
avu	dennister, what's your actual git related prblem?	11:15
	← hml left	11:16
	← reel_ left	11:16
dennister	git pull...did the git init successfully, i may have the source wrong in /etc/apt/sources.list, but "git pull always gives me errors	11:16
	→ reel_ joined	11:16
avu	dennister, git has nothing to do with the sources.list file	11:17
dennister	okie, will remove it then	11:17
avu	dennister, also, if you want a copy of an existing directory, you don't do init/pull but clone	11:17
dennister	ty...very helpful so far :-)	11:17
	← ben_h left	11:20
Mikachu	does ubuntu even come with gcc? :)	11:20
dennister	ok, seems I did do a git clone last night, following this (i believe flawed) howto: https://wiki.ubuntu.com/KernelGitGuide	11:20
albertito	dennister: on a quick check it seems sane, at least for a user who just wants the latest Ubuntu-patched kernel	11:22
dennister	after doing the long download, did the 'git pull' as per howto, got error, took a look at the git tutorial, did the 'git init' and that worked, then tried to do another git pull, as per howto, and that's where i got more errors	11:22
albertito	dennister: just git clone git://kernel.ubuntu.com/ubuntu/ubuntu-hardy.git; cd ubuntu-hardy; make menuconfig and so on	11:22
avu	dennister, read the howto again. it's fine but you missread it.	11:23
albertito	dennister: then, when you want to update your copy of the kernel, do "cd ubuntu-hardy; git pull", and that's just it	11:23
dennister	small wrinkle: i need to install the speakup patches...this is for a blind user who needs speakup screen reader	11:23
avu	dennister, you run the git pull command in the directory created by git clone only when you want to update what you got	11:23
	→ pieter__ joined	11:24
pieter__	ah, internet via gprs	11:24
dennister	o i c...so i'll unpack the speakup tarball as per those instructions, and do the symlink from /usr/src/linux to the .git, then do the make menuconfig? i figure i'll be all right once i get to the menuconfig portion...being on familar ground by then	11:26
Mikachu	is there any particular reason you want the latest unstable linux version?	11:27
	← lea_w left	11:27
dennister	Mikachu: if ur talking to me, I'd actually not prefer an unstable version...hardy is up to 2.6.19.36 i believe...i just need to compile the custom kernel with the speakup patches	11:28
Mikachu	why not just download 2.6.26.2 from kernel.org?	11:29
avu	or install a linux-source package from ubuntu	11:29
astubbs	ok - need a little help here. i just made a massive commit and the realised i was on HEAD^ and of course no i'm on "no branch". how do I make branch master's HEAD be where I am now? - it's been a long day... but i'm sure glad i didn't loose anything :)	11:30
	← Zoup left	11:30
albertito	dennister: you should check what kernel the speakup patches are against. I googled and it seems there were packages for older ubuntu releases, maybe they are already shipping it?	11:30
	astubbs: git show, copy the sha1 of the commit. git checkout master; git reset --hard <sha1 of the commit>	11:31
dennister	wiat, my kernel now is 2.6.24.19...this kernelgitguide in ubuntu community documentation says that doesn't work anymore...believe me...i'd prefer to use methods I'm already familiar with	11:31
albertito	astubbs: but you will loose the current "HEAD" commit, you are aware of that, aren't you?	11:31
Mikachu	astubbs: git checkout -b oopsie; gitk oopsie...master, see if there's anything in master you accidentally threw away from this branch	11:31
avu	dennister, this KernelGitGuide is for people who want the latest ubuntu development kernel.	11:31
astubbs	albertito,yeah that's cool	11:32
dennister	albertito: the speakup patches are for 2.6.10 and later...and no, there are no speakup packages for hardy in the repositories...they and some other necesary stuff like kernel-patch-?? have been dropped from hardy repositories	11:32
albertito	astubbs: you can also: git branch tmp; git rebase master; git checkout master; git merge tmp if you don't want to loose that commit	11:32
avu	dennister, the packages are now calles linux-patch-foo	11:33
	dennister, you should really head over to #ubuntu with this :)	11:33
albertito	dennister: I found this on launchpad: https://launchpad.net/ubuntu/+source/speakup, maybe you can use that? have you asked in #ubuntu?	11:33
wildfire	dennister, why are you not just searching for the speakup-modules as avu indicates? Basically you seem to be taking the longest, most difficult path to get a kernel with those patches compiled in	11:33
dennister	avu: i always try #kubuntu and #ubuntu channels first...gwas very patient but no one could help me	11:33
avu	dennister, well, you clearly have no git problem. in fact, using the kernelgitguide method is most likely not what you want, which means you are quite offtopic here	11:34
dennister	posted it to ubuntu forums, too, lots of views, no one could help...got to kernelgitguide from another ubuntu page, which said i had to use the git guide, as other methods didn't work anymore...	11:35
	← Aerpe left	11:36
albertito	dennister: have you tried using the mailing list? there is a ubuntu kernel mailing list where this is probably on topic (since they did ship packages before)	11:36
avu	dennister, didn't work? of course you can just install a linux-source-version package	11:36
dennister	soory if i'm offtopic, but as you can see, i have tried many other channels first, before coming here...and u guys have been the only ones who've been able to offer me any useful advice	11:36
avu	dennister, really, this git repository is only for the latest development kernels.	11:36
	dennister, see https://help.ubuntu.com/community/Kernel/Compile	11:37
albertito	avu: his problem doesn't seem to be the compilation on itself, but the speakup patches	11:37
astubbs	yikes this is freaky -getting some big old (cygwin) git-read-tree**** fatal error could nt load shell32, win#2 error. Hangup.	11:37
	doin a back up .git now before i screw anything	11:38
	← morphir left	11:38
avu	albertito, yeah but from what I have agthered, those patches exist for older kernel versions as well, so there is no reason he should use an unstable development kernel	11:38
	gathered even	11:38
	← ggapol left	11:39
avu	wow, reading further seems to suggest that the ubuntu folks really did manage to break the in-repository kernel source packages?	11:40
dennister	avu: yes, the link u gave me was the exact doc that told me to use kernelgitguide...due to this: NOTE: This method has been broken since at least June 14th, 2007. Details are here. Use method #1 above ("git") instead.	11:40
Mikachu	lol at "Reasons for NOT compiling a custom kernel" "You have no idea what you are doing, and if you break something, you'll need help fixing it."	11:41
avu	dennister, well, get ubuntu support	11:41
Mikachu	i would have put one more "if" in that sentence	11:41
dennister	Mikachu: like i said, i have built/compiled many a gentoo kernel before	11:41
avu	dennister, this is a git channel. most people in here don't use ubuntu and our discussion about it steals screen real estate from legitimate git questions.	11:42
dennister	fine. i get the hint: i'm buggering off...thanks for the help, and the hurt feelings	11:42
	← dennister left	11:43
	→ d0k joined	11:44
	← X-Blade left	11:44
astubbs	crap - git-read-tree crashes checking out my oppsie at about 30%	11:44
	i can see the commit in gitk though	11:44
	and the entire patch	11:44
	just can't seem to reset to it	11:45
	→ mithro_ joined	11:45
	← mithro left	11:46
	mithro_ → mithro	11:46
	← jlilly left	11:46
astubbs	ok making progress... there are four commits in the chain after i went to head^. i've reset to the first one ok	11:48
	(6403 files btw)	11:48
	i'll try climbing up the ladder, resetting to each one	11:48
	1	11:48
	2	11:49
	3	11:49
	now this is the last commit... gulp	11:49
	ok this is strange	11:51
	dir/file currently doesn't exist	11:51
	now i got up one more step	11:51
	crap. got that git-read-tree crash again	11:52
	perhaps i could create a patch and reapply it?	11:52
	→ dragonball_ joined	11:53
	astubbs googles git cherry pick	11:53
	→ mithro_ joined	11:53
	← mithro left	11:54
	mithro_ → mithro	11:54
	← priidu left	11:55
	→ Pakorrerme joined	11:55
Pakorrerme	Hi every body	11:55
	I've just installed git + gitosis in my debian distro	11:56
	→ ereslibre joined	11:56
	→ lu_zero joined	11:56
Pakorrerme	but i when i try to connect with ssh i get some error like that: "fatal: destination directory 'gitosis-admin' already exists."	11:57
	what's matter?	11:57
	can anybody help me?	11:58
charon	astubbs: maybe you can 'git fsck' to rule out a corrupt object in that last commit?	11:59
Pakorrerme	charon, it's me	12:00
	I need to install git urgently	12:00
astubbs	but the last commit is the really big important one	12:00
	will that delete anything?	12:01
	→ MK_FG joined	12:01
charon	astubbs: no, fsck just checks consistency	12:01
Pakorrerme	What's is the function of gitosis, thank you?	12:01
charon	note that dangling objects aren't an issue, but still reported	12:01
	→ unreal_ joined	12:01
	← unreal left	12:01
astubbs	just reported a whole lot of dangling	12:01
	→ cannonball joined	12:03
charon	Pakorrerme: i doubt you'll get help if you vaguely describe an error without saying with what exact commands you got it, then being impatient and prodding people that just so happen to be awake. try pasting some of the shell session, including the error, on pastebin.com or so	12:03
Magnus`	running 'git checkout $rev -- dirA' does not remove files which where added after $rev. Is that a bug or a feature?	12:04
astubbs	ok i can't seem to reset to it	12:04
	what's the easiest way to make a patch from it? (i suppse that wont work either anyway)	12:04
Mikachu	Magnus`: if you want to remove files, you already have git rm	12:04
astubbs	git format-patch <rev> looked like a no-op	12:04
Magnus`	Mikachu: i just want to get dirA at the state it was in at $rev	12:04
	← pieter__ left	12:05
Mikachu	Magnus`: git rm dirA; git checkout rev dirA	12:05
	→ frq1_dave joined	12:05
	← tlyng left	12:05
astubbs	ok tried git reset --hard HEAD@{4} - that's the commit. and i get the unable to create symlin dir/file that i was alluding to earlier	12:05
	error is (File exists)	12:05
Magnus`	say that HEAD:dirA contains file1 file2 file3 and HEAD^:dirA contains file1 file2. Doing git checkout HEAD^ -- dirA will not remove file3 from dirA	12:06
astubbs	hmm	12:07
Mikachu	Magnus`: yes, it won't, that's intended	12:07
astubbs	git status shows that there's actually heaps of changes now - looks like my commit	12:07
	→ epic_ joined	12:07
Magnus`	Mikachu: why is that? and is there a way around it?	12:08
epic_	how do i use git-svn init against a non-standard repository structure?	12:08
Mikachu	Magnus`: i already said what to do	12:08
	Magnus`: git-checkout by definition checks file out, it doesn't delete them	12:08
	files*	12:08
Magnus`	ok, thanks!	12:09
astubbs	ah-ha! i think git reset --soft HEAD{5} after that has done it!	12:09
	→ grahal joined	12:10
astubbs	oh thank god! my master is now pointing at my large important commit	12:10
	shesh what a nightmare	12:10
	now to dcommit before i loose it!	12:10
Pakorrerme	Hello when i connect to git from shh witch "sudo git clone git@localhost:gitosis-admin.git" i get this message "Initialized empty Git repository in /home/gitosis-admin/.git/" but after It ask me for a password, what password? thank you	12:10
	→ janm joined	12:11
charon	astubbs: a soft reset keeps the index and working tree, so it never reads the tree you're pointing it at	12:11
Pakorrerme	hello!!!	12:11
charon	Pakorrerme: google for gitosis-init.	12:12
	→ Ademan joined	12:12
Pakorrerme	anybody can help me --> when i try to connect to git from shh witch "sudo git clone git@localhost:gitosis-admin.git" i get this message "Initialized empty Git repository in /home/gitosis-admin/.git/" but after It ask me for a password, what password? thank you	12:12
	charon thank you	12:12
	→ warlock_mza joined	12:13
charon	for example, http://gentoo-wiki.com/Gitosis#Configuring has a nice explanation.	12:13
Pakorrerme	charon thank you	12:13
	i'm using debian	12:13
thiago	Pakorrerme: "sudo"? Why?	12:14
Pakorrerme	thiago, it¡s because i use debian, and i want to execute that command like root	12:15
thiago	so... why?	12:15
Pakorrerme	¿?	12:15
	K	12:15
thiago	why do you want to use root?	12:15
Pakorrerme	I don't know if it is necessary or not	12:15
	ahh	12:15
	i understand your answer	12:15
	← _kid_ left	12:16
Pakorrerme	maybe you are thinking about I should have to execute this command from client	12:16
thiago	why do you want to use the superuser for a git repository?	12:16
	← Zyroth left	12:19
astubbs	ok - here's an easy question. you know when you have rebase conflicts - does it always apply an entire commit at once, and show you all the conflicts? or does it stop at the first conflict?	12:20
	bcarlyon\|laptop → BCarlyon\|Laptop	12:20
thiago	it applies everything	12:21
astubbs	oh ok good - so i only have one conflict :)	12:21
	what i find weird is that, it says there's a conflict, then i do git-ermgetool and it says there's nothing to merge!	12:22
	and stauts only shows .dotest	12:22
	so then i try and continue - and it says - "no changes", did you forget to git add?	12:23
Pakorrerme	Are all my repositories saved in /var/spool ?? thank you	12:25
	← AlexB left	12:26
	→ ben_h joined	12:26
	→ _kid_ joined	12:28
	→ lorandi joined	12:28
	→ jmalonzo joined	12:31
epic_	how do i change a git svn svn+ssh username? it keeps trying to log on to ssh with my local username..	12:31
doener_	add username@ to the url in the config, or just setup your ssh config to use the correct username for that host	12:32
epic_	got it :)	12:32
	thanks	12:32
	→ mw joined	12:36
	→ morphir joined	12:37
	← frq1_dave left	12:38
	← cannonball left	12:40
charon	not really a git question: do mmap() areas of real files count towards RSS? DRS? VSIZE?	12:41
Mikachu	i think it depends on if they're paged in or not ;)	12:42
	→ markkalderon joined	12:42
	→ madewokherd joined	12:44
	← janm left	12:44
	→ cannonball joined	12:45
	→ vbabiy joined	12:48
Pakorrerme	Hi	12:50
	→ parasti joined	12:50
	← Leonidas left	12:52
	→ lhunath joined	12:52
lhunath	is it a bad idea to want to put my GIT hooks in the repository?	12:52
	← astubbs left	12:53
	→ irotas joined	12:55
Pakorrerme	if I have created git user with the flag "disabled-password" then, why does git user ask me for a password?	12:55
cupsh	I want to cherry-pick a bunch of commits from another branch. rebase is the tool of choice, right?	12:56
Mikachu	disabled-password means you can't log in at all	12:56
	→ robinr joined	12:56
Pakorrerme	Mikachu, what?	12:57
Mikachu	disabled-password means you can't log in at all	12:57
Pakorrerme	ahh ok	12:57
Mikachu	lol	12:57
	→ skarab joined	12:58
Pakorrerme	And because when I type "git clone git@YOUR_SERVER_HOSTNAME:gitosis-admin.git" it ask me for a password	12:58
	← mithro left	12:58
	← lhunath left	12:59
	→ Aerpe joined	13:01
	← ben_h_ left	13:01
	← ben_h left	13:01
	← skarab left	13:01
zawir	cupsh: right, you can use "pick" in rebase	13:02
	← crab left	13:06
	→ sergiodj joined	13:08
	→ QuercusMax joined	13:08
	→ X-Blade joined	13:09
	→ Ryback_ joined	13:09
	← ph^ left	13:10
	← dragonball_ left	13:11
	← BabelO left	13:14
	→ BabelO joined	13:15
	→ jeukku joined	13:16
	→ AlexB_ joined	13:16
	← AlexB_ left	13:16
	→ aroben joined	13:17
	← aroben left	13:17
	→ giom joined	13:18
Pakorrerme	Hi	13:18
	← shadowfax left	13:19
	fargiolas\|afk → fargiolas	13:19
	→ LiamH joined	13:20
	← pygi left	13:20
	← bongiojp left	13:21
	→ cmarcelo joined	13:21
	→ rlorandi joined	13:23
	← lorandi left	13:23
	→ dragonball_ joined	13:29
	← Sho_ left	13:29
	← cmarcelo left	13:30
	→ SjB joined	13:30
Pakorrerme	Command "git clone git@SERVERHOST:gitosis-admit" doesn't create the file gitosis.conf	13:30
	→ ph^ joined	13:30
	→ bongiojp joined	13:31
	→ ben_h joined	13:32
	→ ben_h_ joined	13:32
	→ priidu joined	13:32
	→ Sho_ joined	13:33
	← ben_h_ left	13:34
	→ jerbear joined	13:36
	→ pygi joined	13:36
	→ rdavila joined	13:37
rmh3093	is there a way to change the message on old commit	13:38
doener_	commit --amend/rebase -i/filter-branch	13:39
	and don't do that for commits you already published	13:39
charon	Pakorrerme: did you gitosis-init like the link i posted earlier says?	13:40
	→ mithro joined	13:40
rmh3093	yeah its already pushed	13:40
Pakorrerme	charon, yes i did that	13:42
	i don't know why gitosis.conf don't appeared	13:42
	← reel_ left	13:44
	→ reel_ joined	13:44
	← lu_zero left	13:44
	← giom left	13:44
	→ giom joined	13:46
	← priidu left	13:47
	← giom left	13:47
	→ Cacheaway joined	13:49
	→ lea_w joined	13:52
	← Aerpe left	13:52
	→ zdennis joined	13:53
zdennis	Anyone know of a way to get a log of revisions on a particular date (or date range) ?	13:54
	← rraasch left	13:54
zdennis	I am looking through docs on git-log, but I don't see anything	13:54
doener_	--since/--until, see git-rev-list(1)	13:54
zdennis	thx doener_	13:55
doener_	also understands things like --since=3.days.ago	13:55
	→ priidu joined	13:55
	epic_ → epic	13:55
	→ kukks joined	13:55
	→ giom joined	13:56
	→ juanjoc joined	13:57
	← epic left	13:57
	← jeukku left	13:57
	unreal_ → unreal	13:58
	→ foca joined	13:59
	← priidu left	14:01
Pakorrerme	Hi	14:01
	→ botanicus joined	14:03
	→ priidu joined	14:03
Pakorrerme	charon, I followed the tutorial that you recommend me but It doesn't work well, when I make a connection using git clone git@MYHOST:gitosis-admin.git it doesn't create a gitosis.conf file	14:04
	← snitko left	14:04
	→ Boman joined	14:04
charon	can you paste the shell sessoin where you do that clone, plus 'cd gitosis-admin; ls -al' on pastebin.com?	14:04
	← rmh3093 left	14:04
	← mithro left	14:04
Pakorrerme	ok	14:05
	→ mithro joined	14:06
	→ rmh3093 joined	14:06
	→ atom joined	14:07
	← mcgrof left	14:07
	← icwiener left	14:07
	→ spearce joined	14:08
atom	hi peeps. Got a wee bit of a problem. I've renamed a lot of files and then modified them, and now when I try git-status (before a commit) git shows a lot of those as file deleted/added instead of renamed. Is there a way to inform git of the rename+edit so viewing file history will be more useful in the long run?	14:09
lea_w	atom: You don't. http://git.or.cz/gitwiki/GitFaq#head-557ab0dbe84efb4e6a091f43c05b326a52b8fc03	14:10
doener_	git always detects renames after the fact	14:10
atom	lea_w: thought so...	14:10
doener_	atom: how much did you edit those files, and how many files are we talking about?	14:10
atom	doener_: it's not such a big deal, it's about 12 files.	14:11
	doener_: and most of the edits are capitalization changes and a few small bugfixes.	14:11
doener_	did you already add the files to the index?	14:11
atom	doener_: yes, unfortunately.	14:11
doener_	the old ones as well as the new ones?	14:12
	→ ddollar joined	14:12
	jmalonzo → janm	14:12
atom	doener_: um... now I'm not sure if I understand the "index" correctly. You mean if I added the new filenames and removed the old ones? yes, I did, I did a git-add .	14:13
Pakorrerme	charon, http://pastebin.com/m64d65c63 thank you	14:13
doener_	ok, so they're all listed under "Changes to be committed", right?	14:13
atom	doener_: yeah, that.	14:14
doener_	atom: I guess your changes account for more than about 33% of the lines of those files then. That's the threshold git uses for renames	14:15
atom	doener_: yeah, that's quite possible.	14:15
	← QuercusMax left	14:15
atom	doener_: but reading the FAQ entry and related pages that lea_w posted, I see that it really doesn't matter.	14:15
Pakorrerme	charon Have you seen anything wrong?	14:15
doener_	atom: well, for files in the index, the rename detection already kicks in	14:16
	atom: so if that doesn't see them as renamed, probably nothing else will	14:16
atom	doener_: I can live with that. I just thought there was a simple fix and that I would have done, but it's not that important.	14:17
doener_	rename first, modify in a later commit	14:17
atom	doener_: that would mean reverting all those files, commiting and copying in the new ones.	14:18
doener_	not necessarily	14:18
atom	?	14:18
doener_	http://git.pastebin.com/m3427c45d	14:19
	a somewhat terse description ;-)	14:19
charon	Pakorrerme: i can't see any mistake there. maybe update to newest gitosis, nuke the gitosis-admin.git directory and redo the setup. if that doesn't work, you'll have to find someone more competent than me	14:19
atom	doener_: interesting.	14:20
	→ aroben joined	14:20
Pakorrerme	charon, oohhh thanks	14:20
doener_	atom: if you're lucky, that might work.	14:20
	← cedricv left	14:20
	← kanru left	14:21
atom	doener_: learning git properly will do me some good someday :)	14:21
	← DavidBoman left	14:21
atom	'till then, I'm cp -r'ing everything just to be sure :)	14:21
doener_	atom: and if that doesn't work, you could still ask git for a diff like this "git diff $old_branch^:oldfilename $old_branch:newfilename" and just apply those using patch	14:23
atom	doener_: I'm still reading the manpage of git-cherry-pick	14:24
	→ botanicus_ joined	14:27
	→ rmh3093_ joined	14:28
	← rmh3093 left	14:29
atom	doener_: thanks.	14:29
doener_	atom: just tried that, cherry-pick didn't work for me. http://git.pastebin.com/m14f72062 -- that's the "somewhat manual" way	14:29
	cherry-pick caused a bunch of conflicts, as the rename detection "failed" and it tried to patch the wrong file	14:30
	the paste has a complete example for a single file, using "git diff" and patch	14:30
atom	yeah, saw that. I didn't really go through with it.	14:31
	I just committed the changes. It's not like anyone's gonna go browsing through this history.	14:31
	doener_ should stop doing lenghty paste noone needs :-/	14:32
atom	doener_: I read it through though	14:32
	→ zjshang_1 joined	14:34
	← zjshang_1 left	14:34
	→ zjshang_1 joined	14:34
	→ chrisb joined	14:35
	← giom left	14:36
	→ giom joined	14:39
	← mithro left	14:41
	→ mithro joined	14:42
	→ kanru joined	14:45
	← charon left	14:48
	→ chris2 joined	14:53
	← MK_FG left	14:54
	→ MK_FG joined	14:55
	← zjshang_1 left	14:55
	← botanicus left	14:56
	→ frq1_dave joined	14:56
	→ Oejet joined	14:59
	→ jackdempsey joined	15:03
	← priidu left	15:06
	→ cedricv joined	15:07
Pakorrerme	Hi	15:10
	← leachim6 left	15:10
	← jdahm left	15:10
Pakorrerme	gitosis doesn't export to me the file gitosis.conf why?	15:11
	I'm using debian	15:11
	→ yoh joined	15:11
	→ leachim6 joined	15:11
yoh	is there any tool to draw 'collapsed' tree of branches -- ie avoiding regular commits but just merges -- to see what is the 'interaction' between the branches? (I didn't find anthing like that in gitk unfortunately)	15:12
Arrowmaster	Pakorrerme: the post update (i think thats it?) hook might not be +x	15:15
	→ priidu joined	15:15
spearce	yoh: no. nobody has created such a tool to show only merges.	15:15
	← thiago left	15:16
pasky	shouldn't that be pretty easy?	15:16
	← uau left	15:16
pasky	1. git rev-list --parents \| grep '.* .* .*' \| cut -f ' ' -d 1	15:16
	← jackdempsey left	15:16
pasky	2. ???	15:17
	3. PROFIT	15:17
spearce	true, but uh, no fancy graph.	15:17
pasky	well	15:17
spearce	and you've cut away edges that may be needed to show the graph.	15:17
pasky	the ??? involves feeding that to gitk	15:17
	hmm	15:17
	that's a good point ;)	15:18
	→ charon joined	15:18
	→ uau joined	15:18
Mikachu	find the code for --no-merges and reverse the condition :)	15:18
spearce	well, that gives us what pasky said. which won't feed gitk nicely.	15:22
	it'll be as bad as doing `gitk --reverse`. it plots horribly confused graphs, assuming it can even plot it at all	15:22
pasky	I could prefix non-merges with -	15:22
	→ vbabiy_ joined	15:22
pasky	since it's kind of like ouboundary	15:22
	and gitk could have a way to	15:22
	get the list of commits on stdin	15:22
	would that work?	15:22
	I'm no tfamiliar how does gitk actually use the mysterious left-right, boundary etc. stuff	15:22
	(actually, I don't know what the hell --boundary actually is since the documentation is so totally unhelpful as usual	15:22
	mikachu: that's the same problem	15:22
	→ powr-toc joined	15:23
Pakorrerme	Arrowmaster, What do you refer to??	15:23
	→ jackdempsey joined	15:23
powr-toc	can you have a git submodule which is a project managed by git-svn?	15:23
Pakorrerme	Arrowmaster, please	15:23
Arrowmaster	sudo chmod 755 gitosis-admin.git/hooks/post-update	15:25
Pakorrerme	Does Anybody know why gitosis doesn't export to me the file gitosis.conf when I do a clone, thank you very much I'm hopeless	15:26
Arrowmaster	then try commiting to gitosis-admin again, or just amending the previous commit	15:26
	← ereslibre left	15:26
Arrowmaster	wait you cloned gitosis-admin and it doesnt contain gitosis.conf?	15:27
Pakorrerme	Arrowmaster, yes	15:27
	It doesn't	15:27
yoh	spearce, thanks for the information	15:27
	pasky, thanks for quick attempt ;-)	15:27
Arrowmaster	does it have keydir?	15:27
Pakorrerme	neither	15:28
yoh	may be you would tolerate 1 more question from me: for cherry picked commits -- what is the easiest way to see all places where it was cherry picked to?	15:28
	cbrake_away → cbrake	15:28
Arrowmaster	i dont know maybe a messed up install?	15:28
yoh	in idial world gitk could draw an edge upon request from original commit to the places where it was cherry picked	15:29
Pakorrerme	I don't know	15:29
Arrowmaster	how did you install it?	15:29
	← spearce left	15:29
	→ jwinkler81 joined	15:29
Pakorrerme	I installed git from repos	15:29
	and gitosis using git clone	15:29
yoh	and then if ever graph with 'no commits' comes up into reality -- adding "dashed arrows" to show from which branch to which some commits were cherry picked would be really great -- it would in 1 image summarize the general development workflow people of that git repository use	15:30
Pakorrerme	I have been all the morning trying to configure it without success :(	15:30
	→ mtodd joined	15:30
	← ismell left	15:30
Arrowmaster	did you follow something like http://scie.nti.st/2007/11/14/hosting-git-repositories-the-easy-and-secure-way	15:30
Pakorrerme	Yes it is	15:30
	I followed that link	15:31
	← mountie left	15:31
	← zawir left	15:32
	→ glennfu joined	15:32
	→ danlucraf1 joined	15:32
glennfu	Hello! I'm a noob to Git and I need a hand... I'm working in a single branch, and someone just made some changes and now I can't catch up to his changes	15:33
pasky	yoh: yes, that would be nice - are you planning to implement that? :)	15:33
	see also git cherry, git show-branch	15:33
glennfu	I have 2 files in the .gitignore file that Git is complaining need to be updated, and 1 file that I want to simply force-update to the newest version	15:33
	What do I need to do?	15:33
yoh	pasky, ;-) shouldn't be too hard I guess if combined with graphviz... but not sure if I am to accomplish the mision	15:34
	→ ebel joined	15:34
	← rmh3093_ left	15:34
pasky	graphviz?	15:34
	→ spearce joined	15:34
pasky	i thought you wanted it in gitk?	15:34
yoh	in gitk it would be 'idial'	15:34
	for quick overview to build some clumsy graph graphviz might be the easier choice	15:34
jaalto	Is is possible to take of certain file 5 revisions backward in current branch?	15:34
pasky	spearce: oh, you are working at google now? or just visiting? :)	15:34
	jaalto: yes, using git checkout	15:35
	← vbabiy left	15:35
Pakorrerme	Arrowmaster my .gitosis.conf inside my /home/git directory is empty	15:35
jwinkler81	I have several files changed locally, and simply want to reset one file back to where it was at the last commit. I've checked docs and examples and nothing seems to work (in particular, using reset <file>, reset -- <file>, reset HEAD <file>. Any help would be much appreciated :)	15:35
spearce	pasky: actually working here. i've been here two weeks now.	15:35
Mikachu	jwinkler81: s/reset/checkout/	15:35
jaalto	pasky: Err. I don't want to lose current file	15:35
pasky	spearce: oh, cool	15:35
	→ rmh3093 joined	15:35
Pakorrerme	Arrowmaster, http://pastebin.com/m64d65c63	15:35
Arrowmaster	Pakorrerme: i have no clue and cant really help then	15:36
	← DrFrasierCrane left	15:36
Pakorrerme	:(	15:36
pasky	jaalto: so, either you want to keep current file or take it 5 revisions back	15:36
	needa decide	15:36
Pakorrerme	maybe	15:36
pasky	spearce: what are you working on there, unless it's secret? :)	15:36
doener_	jaalto: "git show HEAD~5:filename"	15:36
jwinkler81	Mikachu: lol, thank you :)	15:36
jaalto	pasky: I explained poorly. I want to take a diff <current>..<current>~5 of certain file	15:36
spearce	pasky: we're really secretive. but i'm working on android. that's at least a known product. beyond that i probably cannot say anything.	15:37
Pakorrerme	Maybe the gitossis instalation was unclear	15:37
pasky	jaalto: then do? :)	15:37
	→ schacon joined	15:37
jaalto	doener_: Tahnks	15:37
pjz	spearce: bah, you should be working on jgit & egit!	15:37
pasky	spearce: I see, ok - have fun there :)	15:37
yoh	spearce, are you working to port it to ARMv4? ;)	15:37
doener_	jaalto: that show command doesn't give a diff, but the file itself...	15:37
pasky	I've seen some android stuff pop up at repo.or.cz just few days ago ;)	15:37
	but I suppose thtat's just a coincidence	15:37
Pakorrerme	Arrowmaster, How Can i uninstall gitosis	15:37
doener_	jaalto: to diff, just diff those two revisions and limit the output to that file using " -- filename"	15:38
Arrowmaster	delete it	15:38
pjz	spearce: oh yeah, count me in on wanting android on armv4 - I'd like android on my FreeRunner, please :)	15:38
Pakorrerme	I installed it from tar ball	15:38
	Arrowmaster, where?	15:38
Arrowmaster	i dont know ive only messed around with gitosis once or twice	15:39
	→ ereslibre joined	15:39
jaalto	doener_: Got it, thank you.	15:39
Pakorrerme	pfff	15:40
	← glennfu left	15:40
	← jwinkler81 left	15:40
	→ jackdempsey_ joined	15:40
	← jackdempsey left	15:41
spearce	pjz: some of my time does go to jgit/egit development. google has this "20%" bucket.	15:42
	← bongiojp left	15:43
	→ krh joined	15:43
pjz	spearce: ah, yeah, didn't know you were using it for that.	15:44
Pakorrerme	Arrowmaster, with who user i should have to generate de public key	15:44
	→ jschoolc joined	15:45
	← Pakorrerme left	15:45
powr-toc	Sorry to repeat the question, but can git-svn be used to manage submodules?	15:45
	← danlucraf1 left	15:47
	← spearce left	15:47
	→ mountie joined	15:47
	→ LostInTheWorld joined	15:48
	→ docwhat joined	15:48
	→ spearce joined	15:48
docwhat	Hello! Is there some way to add in a different diff mechanism for .xml files?	15:48
	← mithro left	15:49
docwhat	The current diff mechanism means that there are always conflicts when trivial changes are made in XML.	15:49
Mikachu	docwhat: see GIT_EXTERNAL_DIFF in man git	15:49
doener_	docwhat: you can specify a custom merge driver	15:49
Pieter	and 'man gitattributes'	15:49
Mikachu	i don't think the diff mechanism has anything to do with merges though	15:49
docwhat	Mikachu: That's the diff viewer. I meant the merging process.	15:49
Pieter	docwhat: what you're really looking for is a custom merge driver though	15:50
Mikachu	you asked for a different diff mechanism	15:50
spearce	pasky, pjz: i'm not doing hardware support for android, and i know nothing about it, and besides, even if i did, i couldn't comment about whether or not we support (or will support) any particular hardware. at least not until after android launches and its code goes 100% open source.	15:50
docwhat	Sorry. I didn't know what term to use.	15:50
	How can I find out about custom merge drivers?	15:50
doener_	docwhat: man gitattributes, as Pieter said	15:50
Pieter	spearce is busy converting the android code from perforce to git ;)	15:51
	→ babm joined	15:51
	← vbabiy_ left	15:51
spearce	hah.	15:51
	← babm left	15:51
spearce	there are some technical challenges behind open sourcing the code tree.	15:52
	→ digitalhobbit joined	15:52
	→ vbabiy joined	15:52
spearce	most of my 80% time is around those challenges, yes.	15:52
yoh	btw -- if anyone (like me) starts playing with that funky graph creation -- there is an example to draw full graph: http://git.or.cz/gitwiki/ExampleScripts#head-8f06066d9d57a06128491c0aef0d62f210c8b5cd	15:52
Pieter	and the 20% is jgit? :)	15:52
Mikachu	can you use your 20% time to sit at home and do nothing? :	15:52
	)	15:52
docwhat	doener_, Pieter : Thanks.	15:52
	→ netoman joined	15:52
docwhat	It looks like since the merge driver is specified in .config, it won't be passed through a clone... Is there a way to allow that?	15:53
spearce	no, you cannot use your 20% time to sit at home and do nothing. that might actually be a fireable offense. you have to use that 20% time to better yourself and/or better google.	15:53
netoman	Hello! Is there a way to cvsimport into a bare repo?	15:53
	← botanicus_ left	15:53
Mikachu	okay, that is what i thought	15:53
	→ offby1 joined	15:53
	← abizern left	15:55
	→ bobesponja joined	15:55
docwhat	How do you specify that .xml files should use a specific merge driver, then?	15:55
spearce	netoman: do the import using the cvsimport tools (or the fast-export.git based tools) and then mv .git ../proj.git to get a bare repository.	15:56
docwhat	Or will git try it automatically?	15:56
robinr	howdy Shawn	15:57
	I'm back.	15:57
doener_	docwhat: same man page	15:57
	← nutrimatt left	15:57
	→ offby1` joined	15:57
spearce	Pieter: actually more than 20% of my time is allowed to go egit/jgit. for various reasons we have decided to try and accelerate egit development. so some of my newer egit work is now copyrighted by my employer, as its on their time and equipment. very fuzzy lines.	15:58
	welcome back robinr.	15:58
	← jschoolc left	15:58
	→ bartocc joined	15:58
robinr	spearce: that's good news	15:58
Pieter	spearce: ah, cool.. I don't care much for egit, but I like jgit :)	15:59
spearce	yea. given that egit is very dependent upon jgit, this push for egit improvements does trickle down into jgit.	16:00
docwhat	doener_: Ah-hah. I see. The page confused me. I set the attribute "merge." Are gitattributes passed via clone? If so, what happens when you use a custom merge driver, since the .config isn't passed via clone....	16:00
	→ offby1`` joined	16:01
	→ supercali joined	16:01
	→ sdiz joined	16:01
netoman	spearce: Great! Didn't know I could get rid of the working dir so easily... Thanks.	16:01
	→ kazim59 joined	16:02
doener_	docwhat: you can track the .gitattributes file, just like you can do that with .gitignore	16:02
docwhat	Ah. But the .config would have to be set up some other way, though, right?	16:03
	It seems like there should be something like a .git-policy directory that contains things like default .config values, hooks, etc. So that they will survive a clone.	16:03
kazim59	schacon: hey, what software do you use to create those awesome slides at RailsConf Git Talk?	16:04
	← wildfire left	16:04
Mikachu	you don't want config and hooks to be on by default after a clone...	16:04
schacon	Keynote	16:04
Mikachu	echo rm -rf ~ >> hooks/pre-commit	16:04
schacon	or, the graphics?	16:04
docwhat	Mikachu: For an open source project, I agree...but when you trust your upsource, why not?	16:04
kazim59	schacon: thanks! love your slides, and git :)	16:04
schacon	Omnigraffle	16:04
kazim59	schacon: I don't know... both I guess.	16:05
	schacon: oh oh.. they are for OS X, only :(	16:05
schacon	yeah, sorry.	16:05
	you linux or win?	16:06
kazim59	schacon: linux.	16:06
schacon	Dia and OO can probably get the same thing done	16:06
	← garrin left	16:06
kazim59	schacon: i wish to give a talk like that at my college about git.	16:06
madduck	anyone working on packaging topgit for debian?	16:06
schacon	I can export it to PP for you if you'd like	16:07
kazim59	schacon: thanks a lot, but I guess I can extract slides from your presentation for my college, if your license allows that..	16:08
	← charon left	16:08
schacon	yeah, np	16:08
	→ hobodave joined	16:08
hobodave	hi everyone, when using github, do 'forks' of a repository automatically stay in sync?	16:08
kazim59	schacon: thanks!	16:08
schacon	hobodave: no	16:09
hobodave	thanks	16:09
	how do you sync it up then?	16:09
	→ offby1``` joined	16:09
	← offby1``` left	16:10
	→ bugfux joined	16:16
	→ bts- joined	16:16
schacon	hobodave: it's just a clone - you can either remove it and add it again, or you can clone it, add the other as a remote and merge them yourself. and auto-fast-forward would be an interesting feature, though	16:16
	← offby1 left	16:16
	→ towski joined	16:16
	→ cached joined	16:16
schacon	hobodave: I created a ticket for auto-ff to look into doing a bit later, if you want to follow it or comment on it : http://logicalawesome.lighthouseapp.com/projects/8570-github/tickets/805-auto-fast-forward	16:16
hobodave	cool, thanks :)	16:16
	← awx left	16:16
	← offby1` left	16:16
	→ tarbo_ joined	16:16
	→ wildfire joined	16:16
	→ awx joined	16:16
	→ enkrypt_ joined	16:16
	← enkrypt left	16:17
	→ mranostay_work joined	16:18
mranostay_work	hello	16:18
	anyone know of a web frontend to gitosis?	16:18
yoh	is there easy way to get --children effect instead of --parents for rev-list?	16:18
	→ zawir joined	16:18
	→ kumbayo joined	16:20
	← unreal left	16:20
Tv	mranostay_work: for what part of it? it integrates with gitweb	16:20
mranostay_work	sorry i mean for administration as well	16:20
	→ perezd88 joined	16:21
Tv	mranostay_work: that is against the project goals	16:21
Pieter	http://img.thedailywtf.com/images/200808/errord/packagingmadness.jpg	16:21
	heh	16:21
Tv	i mean, gitosis itself will not include a web user interface	16:21
	and any web ui anyone writes is going to be "interesting", as it's going to need to understand merges	16:22
madduck	does Petr Baudis hang out on IRC?	16:22
Pieter	yes, as pasky	16:22
madduck	pasky: hi! :)	16:22
pasky	hi ;)	16:22
madduck	fwiw, I think I will give an introduction to topgit to the debconf crowd in 2.5 hours	16:22
pasky	cool :)	16:22
madduck	http://debconf8.debconf.org/video	16:22
	if you want to watch and participate	16:23
	pasky: so i am going to try it out now and also create a debian package.	16:23
	2 hours is ample time to understand it, right? :)	16:23
pasky	I'm afraid I won't be online from now+1h to now+4h :( but I'll be sure to at least replay the talk when I come back	16:23
	madduck: well, hopefully ;)	16:23
	BTW please use HEAD instead of 0.1	16:24
madduck	https://penta.debconf.org/dc8_schedule/events/233.en.html	16:24
	pasky: of course. :)	16:24
	← offby1`` left	16:24
madduck	so i better get going so i have time to ask you questions within the next 55 minutes. :)	16:24
	pasky: btw, do you know bzr loom? does it compare?	16:25
spearce	i hate perforce	16:25
	→ bdiego joined	16:25
	← tarbo left	16:26
pasky	madduck: I've heard about bzr loom few days ago from another debconf guy ;)	16:26
	but I'm not sure about its feature set, I didn't get around to try it myself	16:26
madduck	yeah, me neither	16:26
	→ Aikawa joined	16:27
pasky	I don't know how does bzr loom work in distributed environment and it seems to me it does not allow free-form dependencies	16:27
madduck	http://lists.alioth.debian.org/pipermail/vcs-pkg-discuss/2008-August/000271.html	16:27
pasky	but only linear ones	16:27
	← foca left	16:27
	← perezd88 left	16:27
madduck	yeah	16:27
	so pasky++ for doing it right. :)	16:27
pasky	;)	16:27
	→ foca joined	16:28
Pieter	pasky: I don't think it works distributed (looms)?	16:28
pasky	oh, it doesn't?	16:28
	in that case loom would be more on par with stgit?	16:28
Pieter	I think so	16:28
pasky	in that case loom doesn't sound interesting at all, I think	16:28
	← reel_ left	16:28
pasky	I thought the point was it is possible to have it distributed	16:29
madduck	i am pretty sure bzr loom is distributed	16:29
	madduck glares at $PREFIX/libexec/topgit	16:29
Pieter	ah, they are	16:30
wildfire	pasky, as far as I understand bzr loom is similiar to your 'topgit'	16:31
	it allows multiple people to work on a topic branch	16:31
	→ ilogger2 joined	16:36
pasky	madduck: I think it would be nice	16:36
powr-toc	madduck: so I don't need to do anything special?	16:36
	sweet.	16:36
wildfire	pasky, I do not known how it weaves those two things together	16:36
machrider	the only time you need to do anything special is when you interact with subversion	16:36
	the rest of the time, it's plain old git	16:37
powr-toc	machrider: cool	16:37
pasky	Pieter: ok	16:37
	Pieter: still don't understand your original sentence though :)	16:37
Pieter	pasky: then, when switching up, the changes in the threads down (the dependencies in topgit) are merged into the current thread	16:37
pasky	what is "switching"?	16:37
	probably not just like git checkout?	16:37
Pieter	actually, yes	16:38
pasky	oh	16:38
	→ Sho_ joined	16:38
pasky	so it's like automatically calling tg update on git checkout?	16:38
Pieter	yes	16:38
pasky	hmm	16:38
	curious	16:38
madduck	pasky: done.	16:39
Pieter	pasky: Integrating the changes from upstream into your loom is straightforward now. When you go up a thread bzr will automatically perform a merge, and you simply need to check that the merge came out correctly and then commit it.	16:40
	so it's like a merge --no-commit on checkout	16:40
madduck	thread == topic branch with one patch?	16:40
	(but many commits?)	16:40
Pieter	yes, exactly	16:40
	so a thread is really like a tg branch	16:41
Oejet	spearce: You sound old, and bitter. :P	16:41
pasky	I see	16:41
	well, I can't say I would identify with such a behaviour	16:42
spearce	Oejet: old, not yet. bitter, probably. but i'm not sure what comment of mine you are basing this on. its been a while since i said anything. :)	16:42
pasky	I prefer the user having to update explicitly	16:42
	maybe Oejet likes perforce	16:42
j_engelh	How could you like perforce??!!	16:43
Oejet	I have no reason to hate Perforce, because I never tried it. :)	16:43
j_engelh	perforce is slow like svn, to begin with.	16:43
madduck	pasky: bzr and git differ on commit/merging/updating/auto-doing stuff all over the place in this way	16:43
Pieter	I thought perforce was based for speed? :)	16:43
	*build	16:43
j_engelh	if it crashes twice a day?	16:43
spearce	perforce doesn't have fancy things like "git log --grep=foo -- path" like git. i miss it.	16:44
madduck	pasky: so assume I have master->A->B and master->C topic branches	16:44
	and now I create D to depend on A and C	16:44
	there will be conflicts in .topmsg	16:44
	is that by design? I don't see whyt .topmsg shouldn't just be recreated from scratch, rather than merged.	16:45
	→ ferdy joined	16:45
Oejet	Well, what does Perforce then have, that Git does not?	16:46
vmiklos	commercial support? (just guessing)	16:46
madduck	Oejet: better visualisation	16:47
	→ perezd88 joined	16:47
Pieter	I guess that's also why bazaar has been getting so much attention	16:47
schacon	single point of failure?	16:47
	:)	16:47
Pieter	the Canonical guys seem really keen on helping projects convert	16:47
madduck	you should be at debconf. :)	16:47
	every time someone says git, a canonical guy is around to argue	16:48
	well, that's immensely exaggerated. :)	16:48
schacon	Bazzar? Really? I was under the impression that bazaar was similar to git except much, much slower	16:49
madduck	apparently the speed issue has been fixed, but it's not similar to git at all	16:49
Pieter	Bazaar has some really weird conventions that differ from git	16:49
madduck	they both do the same thing (source control), but completely different	16:49
Pieter	and Bazaar is still pretty slow in my testing	16:50
madduck	pasky: another thing... after I created D to depend on A and C, there is no branch D created	16:50
pasky	madduck: that is a bug	16:50
schacon	oh, i thought bz used snapshot storage similar to git, not delta based	16:50
pasky	madduck: this should work	16:50
	(the merging)	16:50
madduck	instead, I am left on a detached head	16:50
pasky	does this happen with HEAD?	16:50
madduck	yes	16:50
pasky	that's strange	16:50
	will have a look on that	16:50
Pieter	schacon: yeah, but that's not really a pro or con for speed	16:50
schacon	i meant, in it being 'similar' to git - i figured it was slow because it's in python instead of C	16:51
Pieter	well, one of the storage models is based on the same model as git.. but they used to have delta storage and some other things	16:51
madduck	pasky: http://scratch.madduck.net/.tmp__vit.LSx11383 shell snapshot and http://scratch.madduck.net/.tmp__cdt.CkWx9455.tar.gz git repo	16:52
Pieter	storage model really doesn't matter at all for the end user	16:52
	→ ciaran29_ joined	16:52
	→ jas4711 joined	16:52
	→ selckin joined	16:53
	← selckin left	16:53
	← ciaran29_ left	16:54
pasky	madduck: what git version?	16:54
madduck	1.5.6.3	16:54
	pasky: ^	16:54
	→ selckin joined	16:54
	→ AhtiK joined	16:55
	→ lolage0 joined	16:56
pasky	tg: Please commit merge resolution and call: tg create	16:56
	I think you forgot this after git commit?	16:56
	→ shingo joined	16:57
madduck	pasky: yeah, that was it.	16:58
pasky	oh	16:58
	oh...	16:58
	what _shell_ do you use?! :)	16:58
madduck	zsh	16:58
pasky	I mean, your /bin/sh	16:58
madduck	dash	16:58
pasky	hm	16:58
	and it, um, doesn't know echo -e? ;)	16:59
madduck	i don't think so	16:59
	-e	16:59
pasky	pasky@pixie:~/src/topgit> git grep 'echo -e'	16:59
	tg.sh: echo -e ".topmsg\tmerge=ours"	16:59
	tg.sh: echo -e ".topdeps\tmerge=ours"	16:59
	:)	16:59
	that's the problem	16:59
	ok	16:59
madduck	ouch	16:59
	bashism. :)	16:59
dato	trivial to fix :)	16:59
madduck	pasky: do you have time to commit that fix still? :)	17:00
	hint hint :)	17:00
pasky	sure	17:00
	just pushed it out :)	17:00
Pieter	what's the fix then?	17:00
pasky	you'll need to rm .git/info/attributes	17:00
Pieter	ah	17:00
	→ foca_ joined	17:00
madduck	oh, I see how this works.	17:00
	yeah, that fixes things	17:01
	pasky: if you want I can also make a patch	17:01
pasky	what patch?	17:02
madduck	for the echo -e thing	17:02
	Pieter: echo ".topmsg<09>merge=ours"	17:02
pasky	as I said, it's already pushed out :)	17:02
	http://repo.or.cz/w/topgit.git	17:02
madduck	ah, missed that line	17:02
	thank you!	17:02
pasky	well, thanks for the report ;)	17:02
madduck	i love under-pressure talk preparation when the people related to the subject are on IRC	17:03
Oejet	madduck: How long until your presentation begins?	17:03
Pieter	how'd you fix a \n with an echo -e then?	17:03
madduck	Oejet: just under 2 hours	17:03
	Pieter: echo "bla bla	17:03
	"	17:03
	just include the newline	17:03
Pieter	that's not very nice	17:04
madduck	POSIX is not very nice. :)	17:04
	Pieter: or you use cat <<_eof	17:04
	heredocs	17:04
	or multiple echo calls.	17:04
Oejet	madduck: I cannot wait. :)	17:05
madduck	i hope you don't have too high expectations. :)	17:05
	→ chris2_ joined	17:06
	→ BabelO_ joined	17:08
pasky	I think I'll finally implement remote topic branches support tonight	17:08
	← BabelO_ left	17:08
madduck	pasky: when i do tg export, the goal is a flattened series, right?	17:09
	for upstream. but the patches include .topdeps and .topmsg... :/	17:10
pasky	do they?	17:10
	they shouldn't	17:10
madduck	they do. :)	17:11
pasky	hmm	17:11
	they indeed do :)	17:11
madduck	hey, and how does one pronounce your name?	17:12
	Petr seems straight forward.	17:12
	→ npgoog joined	17:12
pasky	bah, stupid git-update-index	17:12
madduck	Baudis is that Bohdis or Bowdis?	17:12
pasky	Bowdis is closest, but it's "ow" as in "wow", not "bow" ;)	17:12
madduck	yeah, what I thought.	17:13
	→ adante_ joined	17:13
madduck	cool, this will be a fun, impromptu live demo. :)	17:14
	adante_ → adante	17:14
pasky	madduck: anyway, fixed, pushed out :)	17:14
	→ vbabiy joined	17:14
madduck	pasky: can you explain one last thing to me? with one cleanup only when you push upstream, do you mean tg export?	17:15
	so upstream never sees your tg branches	17:15
pasky	yes	17:15
madduck	but you export them to a patch series and then pull, cheryr-pick, format-patch, quilt etc.	17:15
	→ Zao joined	17:15
madduck	sweet. this is awesome	17:15
pasky	well it's not patch _series_, since it's still non-linear - but you can linearize that easily using git log's --topo-order	17:16
madduck	this seems to be exactly what was lacking from the debian git packaging workflow that I have been harping on and on about for almost two years now.	17:16
pasky	cool :)	17:16
	another thing is how well will this scale with many tens of patches; but we can always rewrite it in C ;)	17:17
madduck	sure thing	17:17
	what, the software is now 10 days old? :)	17:18
	see it first at debconf! :)	17:18
	← npgoog left	17:18
madduck	where on the Rhine were you sitting when you hacked this up?	17:19
	pasky: ^	17:19
pasky	Basel	17:19
	madduck is from Zurich :)	17:19
pasky	cool :)	17:19
telmich	pasky: wtf? why are you so near?	17:19
	← adante left	17:19
telmich	madduck: gossau ZH	17:19
pasky	lol	17:19
	→ adante joined	17:19
pasky	we gotta meet sometime ;)	17:19
telmich	ack	17:19
madduck	pasky: you live there?	17:19
yoh	gy gy -- I am progressing a bit with my "graphviz based git tree collapsed graph" -- now I just need to figure out (besides parsing comment) on to which branch that merge belonged at that moment -- could it be done easily?	17:19
pasky	I'm here from July to start of October	17:19
	summer internship	17:19
telmich	madduck: zurich city?	17:20
madduck	maybe in the context of http://openexpo.ch/	17:20
	telmich: oerlikon	17:20
telmich	madduck: that's in the midst of the week :-(	17:20
madduck	pasky: i am gone from zurich until sep 23	17:20
	pasky was in zurich two days ago ;)	17:20
madduck	i know, telmich.	17:20
pasky	I've been thinking about openexpo but it's in the middle of the week	17:20
telmich	pasky: and why did't you tell? ;-)	17:20
pasky	so that's probably not going to fly for me	17:20
madduck	i might come to basel on 26 september with a friend.	17:21
pasky	telmich: didn't expect many people from here to be from CH :)	17:21
	oh, cool	17:21
telmich	pasky, madduck: if you want to, we can meet here in gossau for an evening, there's plenty of space available at my home	17:21
pasky	sounds nice	17:21
	I also have some more friends in zurich I found out about only after I've returned from there ;)	17:22
	so I'll be probably going there at least once again yet	17:22
telmich	lol	17:22
madduck	lapse:~/.tmp/cdt.CkWx9455\|test4\|% tg update #8259	17:22
	/home/madduck/.tmp/cdt.HLSx9918/libexec/topgit/tg-update: 90: Bad substitution	17:22
telmich	madduck: are you related to the ccc zurich or the lugs?	17:22
madduck	neither	17:22
pasky	anyway, plenty of time to arrange that yet - will be great to meet you guys though	17:22
madduck	not anymore at least	17:22
pasky	I really have to le...	17:22
	wha, bad substitution?	17:22
telmich	are you guys doing bashism? ;-)	17:22
madduck	pasky: I think action="${depline:0:1}"	17:23
telmich	yep, bashism	17:23
pasky	there is "fi" at line 90 of my tg-update	17:23
	hmm	17:23
	that's one downside of rhine	17:23
	I don't have offline SUS reference	17:23
telmich	pasky: hehe	17:23
madduck	is that substring extraction?	17:23
telmich	pasky: you've internet near the rhine	17:23
	← adante left	17:23
pasky	not really	17:24
madduck	telmich: http://kerneltrap.org/mailarchive/git/2008/8/3/2795494, first para	17:24
	→ ijcd joined	17:24
	→ adante joined	17:24
pasky	there are some paid hotspots, but at points I don't use to sit anyway	17:24
telmich	pasky: depends on where you are; if you stay near the restaurant and a little bit into the east you've	17:24
pasky	"the" restaurant?	17:25
	→ LostInTheWorld joined	17:25
madduck	pasky: make sure to go to schaffhausen to see the rhine falls!	17:25
	pasky notes that :)	17:26
Oejet	madduck, pasky, telmich: I am also in ZH!	17:26
madduck	"Rheinfelden"	17:26
pasky	:)	17:26
	lol	17:26
	→ ijcd_ joined	17:26
	→ DrFrasierCrane joined	17:26
madduck	pasky: what does ${depline:0:1} and ${depline:1} do?	17:26
pasky	first one takes the first character	17:27
madduck	yeah, i just tried	17:27
pasky	second one trims the first character	17:27
	action="$(echo "$depline" \| cut -c 1)"	17:27
	dep="$(echo "$depline" \| cut -c 2-)"	17:27
	so this one should be equivalent	17:27
madduck	trimming is ${depline#*?}	17:27
pasky	can you check if that fixes it?	17:27
	huh	17:27
Oejet	pasky: I am looking for an internship.	17:28
pasky	why not just #? ?	17:28
madduck	don't know. :)	17:28
pasky	madduck: anywya, does it work for you?	17:29
	pasky thinks so far all topgit users were using bash ;)	17:29
madduck	the cut solution does, yes	17:29
pasky	("those debian guys..." ;)	17:29
madduck	the subst one does not	17:29
pasky	ok	17:29
madduck	weirdlly.	17:29
pasky	hmm	17:29
	curious	17:29
	I'll just commit the cut one now	17:30
	can worry about performance later :)	17:30
madduck	yeah	17:30
pasky	ok	17:31
	pushed out	17:31
	and I'm really off now ;)	17:31
	good luck with the talk :)	17:31
madduck	thx	17:32
	→ amystrat joined	17:34
	→ ph^_ joined	17:36
BugeyeD	hi all. any way to imbed tags such as can be done with rcs (eg. $Id$ and $Log$) ?	17:37
doener_	gitattributes(5)	17:38
	but for version numbers "git describe" is the better choice	17:38
BugeyeD	excellent, thanks! my google-fu is severely lacking ...	17:39
doener_	shrugs To me, the $Id$ stuff is totally useless	17:39
	→ CodeOfficer joined	17:40
BugeyeD	i'm accustomed to 'head script' showing the relative version of a script, i'm just trying to replicate that with git. in my case, git won't be available on every box running said scripts.	17:40
	← ijcd left	17:42
	→ bryanray joined	17:42
telmich	Oejet: where are you exactly?	17:42
Oejet	telmich: I watched your presentation at the Google Open Source Jam, although I did not get a chance to say hi. :(	17:42
telmich	Oejet: ahh, ok :)-	17:43
Oejet	telmich: 8006	17:43
telmich	lol, so near	17:43
	hmm, did we already have a git talk at OSSJ in zurich? ;-)	17:43
Oejet	telmich: Hm, I don't know, but it could be very nice.	17:45
Modius	In a (solo) development mode where one tries to check in frequently upon incremental improvement, using GIT, is it idiomatically "done" that you just check these in as separate commits, or try to consolidate them into lower granularity commits through amending a prior?	17:47
	→ BabelO_ joined	17:47
	→ richcollins joined	17:48
	← shingo left	17:48
richcollins	How do I show which files where modified for each commit when doing a git log	17:48
	??	17:48
	← ph^_ left	17:48
	→ shingo joined	17:49
vmiklos	git log --name-status?	17:49
	man git-log :)	17:49
	chris2_ → chris2	17:49
	← BabelO_ left	17:52
	→ BabelO joined	17:54
	→ emlprime joined	17:55
selckin	i have a semi public repo at a certain url, if i replace it with a new one (i know, bad), and people try to pull from the new one with an old clone, will git complain graceful, or is there a chance to break their local repo to a point they can't recover their own changes?	17:55
cypher23	selckin, git will complain	17:56
	but that's basically it	17:56
	git won't overwrite any local changes	17:56
drizzd	might still be confusing though. It's like changing all your history	17:57
cypher23	yes	17:57
	rebasing their changes on your new history might also be a PITA	17:58
selckin	yea pita is fine ;), thanks	17:59
	→ rwaite joined	17:59
	← adante left	18:00
amystrat	I created a remote branch and a tracking branch. If I make changes to local tracking branch and push, it states "everything up to date" and does not push changes to remote branch. Any ideas?	18:01
	→ adante joined	18:01
	→ eikonos joined	18:02
richcollins	Why would I get this error when pushing to a bare repo?	18:03
	fatal: hard reset makes no sense in a bare repository	18:03
	← lolage0 left	18:04
	← adante left	18:04
	← rwaite left	18:05
	→ blowery joined	18:06
	→ adante joined	18:06
	→ ajonat joined	18:07
albertito	richcollins: do you have any hook activated?	18:07
Mikachu	rickrolling: you must have a hook enabled in there that tries to do a hard reset, presumably to update the working tree on push, which doesn't make sense on bare repos	18:07
richcollins	Hrm I don't remember adding one	18:08
cypher23	amystrat, how did you push? simply "git push"?	18:08
amystrat	cypher23, yes	18:08
albertito	richcollins: is the bare on a windows machine, or are you accesing it through samba?	18:08
richcollins	no	18:08
	It is a linux machine	18:08
	I'll check for hooks	18:09
	→ johnw joined	18:09
cypher23	amystrat, try "git push origin branch" (replacing origin and branch as needed)	18:09
	← blowery left	18:10
	→ blowery joined	18:10
	← blowery left	18:10
	← adante left	18:10
richcollins	albertito: Ah looks like someone added an email hook	18:10
amystrat	cypher23, thanks that worked. I don't remember having to do that before	18:10
Mikachu	having reset --hard in an email hook seems... odd	18:11
	→ ehc joined	18:11
cypher23	amystrat, how did you create the tracking branch?	18:12
amystrat	git --track localbranch origin/remotebranch	18:13
	cypher23 ^	18:13
	← ehc left	18:13
	→ bcarlyon\|laptop joined	18:13
albertito	amystrat: assuming you said git checkout --track -b localbranch origin/remotebranch; actually, --track is the default when the other one is a remote AFAIR	18:13
	→ lolage0 joined	18:14
yoh	hey guys -- so that collapsed tree drawing... here is an example of my endavour http://www.onerussian.com/tmp/graph_example.png	18:14
	→ Sjaq joined	18:14
	← AhtiK left	18:14
	→ ehc joined	18:14
yoh	those dashed lines stay where multiple commits without merges (ie collapsed)	18:14
cypher23	amystrat, ok. But I think push only pushes the branches you specify (or you can use --all)	18:14
yoh	also trying to color merges appropriately -- ie depending into which branch we merge	18:15
amystrat	thanks guys	18:15
	→ adante_ joined	18:15
	← vbabiy left	18:16
cypher23	amystrat, according to the manpage (for Git 1.6.0.rc2) you can use "git push :" to push all local branches that also exist remotely. haven't tried that one though	18:16
amystrat	cypher23: interesting, I will have to give that a try	18:17
	→ Oejet joined	18:17
	→ vbabiy joined	18:17
	→ charon joined	18:18
Mikachu	cypher23: isn't that the default?	18:18
cypher23	amystrat, you can also do something in config for the remotes, to specify what gets pushed. see the man page	18:18
	Mikachu, ah. Yeah, you're right.	18:18
	← adante_ left	18:18
	← Sjaq left	18:19
	→ chrisb joined	18:19
drizzd	yoh: that would be a very nice feature	18:19
yoh	it would be a nice feature for gitk... but not sure if I am ready to jump in to patch it :-/	18:21
	→ adante_ joined	18:21
	→ david_koontz joined	18:22
yoh	and unfortunately building such a graph for a reasonably active project makes it quite huge	18:22
drizzd	you will probably have to understand gitk pretty well to do it	18:22
	→ warthog9 joined	18:22
yoh	drizzd, that too... and unfortunately have no time to do so heh heh	18:22
drizzd	hmm, maybe I'll look into it this weekend	18:23
yoh	drizzd, that would be cool	18:23
drizzd	I'm thinking eventually we might even want to leave out the merge commits and just show the branches/tags	18:24
	after all, it's not about individual commits any more	18:24
	→ joelr1 joined	18:25
	→ supercali joined	18:25
drizzd	but I like your idea because I think it should be doable with minimum changes to gitk	18:25
yoh	yeap -- that is what I was trying to grasp --- whenever you get into somebody's repository with a reasonable amont of branching -- hard to figure out workflow	18:25
	having such a 'summary' might help	18:25
drizzd	exactly	18:25
joelr1	good day! i have shared access set up via git-shell but the files created keep acquiring g+r permissions instead of g+w. any way to fix that so that two people can commit?	18:25
	← adante_ left	18:26
yoh	so it could be not even those blobs for merges but rather 1 timeline per each branch with arrows from to whenever merge occured (or cherry picking ;-))	18:26
	→ adante_ joined	18:26
yoh	do you know any efficient image viewer? I generated jpg for a real project and it is hard to view it -- image magic is difficult to scroll, gimp is slow, xzgv is crashing	18:27
drizzd	I use gimv (gimageview)	18:27
	or ghostview	18:27
Mikachu	i don't think anyone has bothered writing a partial jpeg decoder for huge images	18:28
Oejet	yoh: A browser might also work.	18:28
drizzd	or imagemagick	18:28
	← adante_ left	18:29
yoh	Oejet, doh... sure ;-)	18:29
drizzd	Mikachu: what does that mean, 'partial decoder'?	18:29
Mikachu	drizzd: only decode what's shown on screen	18:29
yoh	firefox considers that jpg broken ... heh heh... lets see png	18:29
Mikachu	drizzd: a 100000x100000 jpg at 10MB will hardly fit in ram decoded	18:30
	rockbox decodes on demand when scrolling in the image	18:30
	← ijcd_ left	18:30
drizzd	I see your point. But I'm sure you can also reduce the resolution of your camera ;-P	18:30
Mikachu	sure but we're talking about big generated graphs here	18:31
yoh	same for png... so the only choice is imagemagic but it is inconvinient.... -- that scroll 'preview' is too narrow to navigate in X direction	18:31
	file is 2977 x 38555	18:31
	;-)	18:31
	you are welcome to look http://www.onerussian.com/tmp/graph.jpg	18:32
	→ adante joined	18:32
	→ sgrimm joined	18:32
yoh	my code is still missing reliable detection of the original branch where that merge/commit occured -- is there any git way to deduce what was the branch?	18:33
	→ FH` joined	18:33
	← amystrat left	18:33
Mikachu	you made my duck go upside down	18:33
	hm, it decoded but it only shows gray :)	18:34
albertito	my firefox hang himself with that image	18:34
yoh	gy gy -- firefox on my other box just crashed while opening	18:34
	so please be careful	18:34
drizzd	hmm, gimv fails to display anything but the top	18:35
	→ Daenyth\|Work joined	18:36
	→ alb joined	18:36
Daenyth\|Work	Does a git:// url always follow the form of "git://domain/projectname.git", or is it possible for it to be different	18:36
Mikachu	projectname can contain many slashes, but otherwise no :)	18:37
Daenyth\|Work	mm	18:37
vmiklos	also you may omit '.git' if you configure the server that way	18:37
Daenyth\|Work	ah	18:37
	damn	18:37
	hrm	18:37
netoman	I ammended a commit and now I can't push anymore... What happened?! Any clues?	18:37
yoh	inkscape is the winner!!	18:37
Mikachu	that's unexpected	18:37
yoh	it chooses too large font a bit but renders .svg nicely	18:38
	with nicely browsing ;-)	18:38
	alb → albertito	18:38
Mikachu	oh not the jpg file?	18:38
yoh	Mikachu, nope... sorry for the confusion	18:38
Oejet	yoh: How big is the SVG file?	18:38
yoh	I generated also svg for it from the same .dot	18:38
	600k ;-)	18:38
	← joelr1 left	18:39
Daenyth\|Work	for anyone familiar with archlinux, I'm trying to find a way to allow a git:// url in the source array of a PKGBUILD... Trying to make a script which will automate either a clone/fetch, but I can't figure out a good way to make it work with $_gitname	18:39
yoh	aaa... I opened it in mozilla... now it is thinking -- you can grab it from the same url but with ext .svg	18:39
Oejet	yoh: I would guess your JPG image decompresses to about 450MB in memory.	18:40
	→ tweakism joined	18:40
tweakism	I just did git reset --hard HEAD^	18:40
	is there a way to undo that?	18:40
Mikachu	yes and no	18:40
tweakism	I just successfully wiped out 300 lines of code that were in HEAD but not in HEAD^	18:40
Mikachu	if you had uncommitted working tree changes, those are gone	18:40
	getting back HEAD is just a matter of git reset --hard HEAD@{1}	18:40
	assuming you've done nothing else since	18:40
	→ AhtiK joined	18:41
tweakism	Mikachu: you're awesome	18:41
Daenyth\|Work	Mikachu: I'm not familiar with that @{} notation, would you mind explaining it?	18:41
Mikachu	man git-reflog :)	18:42
Daenyth\|Work	fair enough	18:42
tweakism	Mikachu: I guess the proper way to remove all WIP changes is git reset --hard HEAD, not HEAD^1, eh?	18:42
Daenyth\|Work	tweakism: yes	18:43
Mikachu	you mean uncommitted? yes	18:43
Daenyth\|Work	or you can use git stash if you just need to put them aside for a short while	18:43
	and then work on them later	18:43
albertito	netoman: you can use git push -f, but be aware that if somebody has pulled your tree with the un-amended commit, you will make a terrible mess for him/she	18:44
	← adante left	18:45
	→ adante joined	18:47
	→ sbahra joined	18:48
Oejet	madduck: Good luck. :)	18:49
yoh	ok... probably enough of experiments for me for today... python script I've used is now online http://git.onerussian.com/?p=code/snippets.git;a=blob;f=graphviz_plot.py;h=d16668fd9c27c0460b2f273a6991dd68ceb14851;hb=HEAD	18:50
netoman	albertito: Thanks. Nobody has pulled from it yet, so no problem.	18:50
	← adante left	18:50
yoh	may be some time I will add subgraphs in there after better detection of the owning branches will get implemented -- it would be nice to have each branch in its own (most of the time) vertical (if possible)	18:51
	foca_ → foca	18:51
	← DrFrasierCrane left	18:51
	→ adante joined	18:52
Daenyth\|Work	__revision__ = '$Revision: $'	18:52
	how do you use tags like that with git	18:52
	← adante left	18:52
yoh	nohow ;-)	18:52
Mikachu	Daenyth\|Work: /topic	18:53
albertito	Daenyth\|Work: man gitattributes	18:53
Daenyth\|Work	thanks!	18:53
yoh	just elderly leftovers in my .emacs for automatically generated headers ;-)	18:53
Daenyth\|Work	ew emacs ;P	18:53
Mikachu	oh wait, that url isn't in topic anymore	18:53
	Daenyth\|Work: sorry :)	18:53
Daenyth\|Work	hehe np	18:53
	thanks albertito	18:53
	→ Boman joined	18:54
yoh	albertito, hm -- thanks! I wasn't aware... not yet sure how/where to use but RTFM should help!	18:55
albertito	Daenyth\|Work, yoh: you're welcome	18:56
	bcarlyon\|laptop → BCarlyon\|Laptop	18:56
Mikachu	see http://tinyurl.com/yqpgv9 too	18:57
	→ ciaran joined	18:59
yoh	so it is a recent feature! cool... I started to panic that I didn't read manual / Linus' comments careful enough whenever I Started using git ;-)	18:59
selckin	is there an easy way to get a side-by-side diff from git diff ?	19:00
Mikachu	no	19:00
yoh	ah... what is wrong with me -- I mixed up 2008 and 2007 ;-)	19:01
	→ archis joined	19:03
Mikachu	also note the patch is rejected quite firmly :)	19:03
	← CodeOfficer left	19:03
	→ CodeOfficer joined	19:03
	→ towski joined	19:04
	← CodeOfficer left	19:04
	← tweakism left	19:09
	← vbabiy left	19:09
	→ alb joined	19:09
	→ vbabiy joined	19:09
	→ johnw- joined	19:10
	→ CodeOfficer joined	19:12
	→ CodeOffi_ joined	19:12
	→ ijcd joined	19:12
	→ CodeOff__ joined	19:15
	→ pygi joined	19:17
	← ajonat left	19:18
	→ icwiener joined	19:21
	→ statim joined	19:21
	→ xazmine joined	19:24
	→ lea_w1 joined	19:24
xazmine	Is there a way to combine a whole branch into just one commit?	19:24
	lea_w1 → lea_w	19:24
netoman	I do git cvsimport and it imports an old revision of a file (1.21 instead of 1.22 wich is the latest). Am I doing something wrong?	19:25
xazmine	I'm using git-svn, and I'd like to commit a whole series of commits as just one commit to git-svn...	19:25
	→ Eridius\|work joined	19:26
	→ ebel joined	19:26
	← johnw left	19:26
	← ciaran left	19:26
	→ ciaran joined	19:27
	← CodeOfficer left	19:29
	→ davi_ joined	19:29
	← ciaran left	19:29
	→ ciaran joined	19:30
	→ giom joined	19:30
	CodeOff__ → CodeOfficer	19:30
	← giom left	19:30
	← albertito left	19:31
	← CodeOffi_ left	19:31
yoh	xazmine, man git-rebase , search for squash	19:34
	not sure if there could be some negative side-effects due to git-svn	19:34
	→ polpak joined	19:34
	← ijcd left	19:34
	alb → albertito	19:36
	→ ijcd joined	19:42
	foca → foca\|machito	19:43
	foca\|machito → foca	19:44
	→ _bdiego joined	19:44
	_bdiego → bdiego	19:44
	→ PeterFA joined	19:46
	→ nutrimatt joined	19:48
	→ ph^ joined	19:48
	→ jaushman joined	19:48
jaushman	while setting up gitosis I run "sudo -H -u git gitosis-init < /tmp/id_dsa.pub" and I receive the error "IOError: [Errno 13] Permission denied: '/home/git/gitosis/projects.list.12603.tmp'"...anybody have any suggestions on solving this?	19:50
Tv	jaushman: sounds like user git can't write to /home/git	19:50
	well /home/git/gitosis but still	19:51
Daenyth\|Work	sounds like that to me also	19:51
jaushman	so, do I set that permission when I do "sudo adduser"? forgive me if my questions are remedial, but i'm new to Linux and Ubuntu	19:54
	→ k_dub joined	19:54
jaushman	the user is git and gitosis exists in /home/git	19:55
	→ Ademan joined	19:56
	→ rdavila_ joined	19:56
	← pygi left	19:57
Tv	jaushman: did you create /home/git yourself?	19:57
	→ CodeOffi_ joined	19:58
Daenyth\|Work	jaushman: can you getent passwd \| grep git	19:59
jaushman	Tv: yes, as part of adduser using "--home /home/git \"	19:59
	→ CodeOff__ joined	19:59
Tv	jaushman: ls -ld /home/git /home/git/gitosis	19:59
	davi_ → davi	20:00
	→ giom joined	20:00
	← k_dub left	20:00
jaushman	Daenyth: git:x:108:114:git version control,,,:/home/git:/bin/sh	20:00
	ls -ld /home/git /home/git/gitosis	20:00
Daenyth\|Work	mm	20:00
	← giom left	20:00
Daenyth\|Work	looks ok	20:00
jaushman	drwxr-xr-x 5 git git 4096 Aug 11 14:59 /home/git	20:01
	drwxr-xr-x 8 root root 4096 Aug 11 14:39 /home/git/gitosis	20:01
Daenyth\|Work	that's it	20:01
	chown -R git.git ~git	20:01
	as root	20:01
Tv	jaushman: how did that latter happen..	20:01
Daenyth\|Work	sudo mkdir?	20:01
Tv	jaushman: ok well why ;)	20:01
jaushman	wow...i've been staring at this for hours and didn't see that	20:02
	→ pjhyett joined	20:02
Tv	jaushman: i recommend a cardboard dummy	20:03
	← chris2 left	20:04
	← davi left	20:04
jaushman	:D thank you both	20:04
madduck	pasky: so topgit coverage was short but I think between that talk and the debian package, you'll soon have a few new users	20:11
	better keep that shell script stuff POSIX-compliant! :)	20:12
	← Daenyth\|Work left	20:12
Oejet	madduck: Nice talk.	20:12
	→ egn joined	20:13
madduck	thanks dude	20:13
	→ markkalderon joined	20:14
	← CodeOfficer left	20:15
	← CodeOffi_ left	20:16
	CodeOff__ → CodeOfficer	20:16
pasky	hehe :)	20:18
Oejet	madduck: There has been some discussion about using vcs for package patch management in Fedora.	20:21
madduck	Oejet: i know. please let anyone know about vcs-pkg.org	20:22
	we really ought to be working together.	20:22
	→ Auris-- joined	20:24
Oejet	madduck: So you read the "RFC: Exploded source repo layouts" thread by Doug Ledford?	20:25
	→ neoeinstein joined	20:26
madduck	nope, where?	20:26
Oejet	http://thread.gmane.org/gmane.linux.redhat.fedora.devel/89035	20:27
	→ neoeinstein_ joined	20:29
	← neoeinstein left	20:29
	← emlprime left	20:29
	→ synsol joined	20:31
	← neoeinstein_ left	20:31
	→ tvw joined	20:31
	→ neoeinstein joined	20:31
synsol	hey all, im almost scared to ask this here but is there a git for windows?	20:31
doener_	in fact, two	20:32
	on for cygwin, on using msysgit	20:32
Oejet	Also the see posts by the same author under the earlier thread: http://thread.gmane.org/gmane.linux.redhat.fedora.devel.announce/181/focus=88473	20:32
doener_	s/on/one/g	20:32
madduck	pasky: feature request #1 from the crowds: convert existing branches to tg branches	20:32
	← neoeinstein left	20:32
	→ neoeinstein joined	20:32
synsol	really? do u know if either r good?	20:32
	→ fixUp joined	20:33
	← neoeinstein left	20:33
madduck	Oejet: are you a fedora developer?	20:33
Oejet	madduck: No, I just lurk. :)	20:33
	→ _graham_ joined	20:34
madduck	ic, would you write a mail to the thread to let them know about vcs-pkg?	20:34
	i am on my way out now and will read the threads later	20:34
	→ ahti_ joined	20:34
	doener_ buys a "y", an "o", an "a" and an "e" for synsol	20:34
bremner	what is the state of tg? is it worth trying to use	20:34
doener_	synsol: no, I don't know, I don't use Windows.	20:34
	synsol: the msys one is called msysgit (ask Google about it). On cygwin, you can use the regular upstream stuff I think	20:35
	→ rdavila__ joined	20:35
Oejet	madduck: I do not dare to post to fedora-devel. :D	20:36
synsol	k, thx, btw i dont use it either, unfortunatly an ignorant/stupid company i do work for does	20:36
	but they pay well so ill forgive them	20:36
	← CodeOfficer left	20:36
	← rdavila__ left	20:37
madduck	Oejet: alright.	20:37
	bremner: yes	20:37
bremner	madduck: thanks	20:37
madduck	pasky: if you need a series of all patches, then you have to create a tg branch that depends on all of them, right?	20:37
	→ djinni joined	20:38
Oejet	madduck: A search on "vcs-pkg" on the fedora-devel archive turns up nothing relevant.	20:38
madduck	yeah, i am afraid it hasn't spread much yet	20:38
pasky	madduck: the feature request is too unclear	20:41
	madduck: "series" usually means that each of the patch depends on exactly one other patch, except the first one	20:42
jaushman	newb still looking for help...running git clone git@<server>:gitosis-admin.git on local machine results in ERROR:gitosis.serve.main:Repository read access denied and i'm unclear where to look to resolve this permissions issue	20:42
madduck	pasky: i asked about it on the mailing list the other day. squashing tg trees into quilt series	20:44
	tg export basically	20:44
	and if you create a dummy tg branch which depends on all feature branches	20:45
	and export that, you should get a series across all feature branches, no?	20:45
	← nonix4_ left	20:45
	→ redondos joined	20:47
pasky	ah	20:47
	yes	20:47
madduck	i wonder if it would make sense to provide a feature for this	20:48
	a squash-all feature	20:48
pasky	I call these dummy branches [STAGE]s (instead of [PATCH]es)	20:48
madduck	ok	20:48
	i would love a command	20:48
	tg make-series -o debian/patches	20:48
	which would (possibly using a file to limit) export all branches to files in debian/atches	20:48
	i suppose i could write thta. :)	20:48
pasky	what kind of files?	20:49
madduck	git format-patch output	20:49
	← AhtiK left	20:49
	← archis left	20:49
madduck	so 1. create stage branch, 2. export to tmp branch, 3. format-patch tmp branch to debian/patches/*	20:49
	maybe debian/patches/series could be used to determine the branches	20:50
	although no, that can't work	20:50
pasky	so something like tg export tmp; git rev-list master..tmp \| xargs -l tg patch	20:50
madduck	the series file has to be created	20:50
pasky	er	20:50
	not tg patch	20:51
	git format-patch	20:51
	so also without the -l ;)	20:51
madduck	git-format-patch master at the top of tmp would do it all	20:51
pasky	ah	20:51
	yeah	20:51
madduck	git checkout tmp; git-format-patch -o debian/patches master	20:51
pasky	but I suspect in fact you want quilt series?	20:51
madduck	i'll hack this up later, okay?	20:51
	yeah. but aren't git-format-patch files quilt-compatible?	20:51
pasky	but there is no series file... it would make sense to have something like tg export --quilt debian/patches	20:52
	that would just do it all	20:52
madduck	yes, absolutely	20:52
	← ph^ left	20:52
	← rdavila_ left	20:52
madduck	including writing the debian/patches/series file	20:52
pasky	yes	20:52
madduck	that's a nice approach	20:52
	i need to run out for a bit	20:52
	if you havent' done it by the time i come back, i will. :)	20:52
	ttfn	20:53
pasky	;)	20:53
	→ flask-- joined	20:53
flask--	Is there a way to make a pattern in .gitignore that will make git track a directory but none of the files or subdirectories underneath it?	20:54
Eridius\|work	flask--: git doesn't track directories	20:54
flask--	hmmm	20:55
doener_	flask--: just add a (tracked) .gitignore file in that directory	20:55
	that one can just contain *	20:55
flask--	ahhhh	20:55
	→ tcoppi_ joined	20:55
flask--	you can do per-directory .gitignores ?	20:55
doener_	yep	20:55
	flask-- does the git happy-dance	20:55
j_engelh	dance? http://www.youtube.com/watch?v=UX6e7sO1ss0	20:56
flask--	heh	20:58
	→ yann_ joined	20:59
	yann_ → yann	20:59
	→ cilly joined	21:00
flask--	http://photos1.blogger.com/blogger/6295/1280/320/dilbert.gif	21:01
	→ ckoehler joined	21:01
	→ glguy_ joined	21:02
	← chrisb left	21:02
glguy_	With a freshly git-gc'd repo, would one expect to see a performance difference in git-clone for http:// vs git:// urls?	21:02
ckoehler	how do you have two separate branches with features 1 and 2, if 2 depends on 1? i want code from both and use it together, by necessity, but want to keep them separate code wise	21:03
	hope that made sense...	21:03
	:)	21:03
vmiklos	glguy_: not really	21:04
	→ archis joined	21:04
vmiklos	the problem is that in most cases 1) you fetch, not clone 2) the repo is not freshly git-gc'd :)	21:04
	→ mw-home joined	21:04
mw-home	git seems hard to learn. where's a good place to start?	21:04
j_engelh	fetch not clone, shaken not stirred...	21:05
polpak	ckoehler: submodules possibly?	21:05
doener_	ckoehler: what do you mean by "separate"? Obviously, they can't be totally separated when 2 depends on 1	21:06
	→ dragonball_ joined	21:06
pasky	so, how does debian/patches look like?	21:06
	are the patches numbered?	21:06
	← mw-home left	21:07
	→ ereslibre joined	21:07
doener_	glguy_: clone shouldn't show any noteworthy differences. But fetch might. While git:// users only get a pack that contains what they need, http:// users will have to download the whole new pack, which can mean quite some overhead because they obviously download all the objects that you already have then	21:08
	s/you/they/	21:08
	→ pgregory joined	21:09
pgregory	hi all	21:09
	can anyone help me to understand a merging scenario I've got?	21:10
j_engelh	real questions please...	21:10
pgregory	I'm struggling with the basic concepts of git, and really want to grasp it.	21:10
doener_	only if you care to explain it...	21:10
pgregory	sorry, I'll get to the point	21:10
	→ elux joined	21:10
elux	hey guys	21:10
Oejet	mw-home: What other VCS's do you know?	21:11
ckoehler	doener_: no, not totally separate	21:11
	← fixUp left	21:11
elux	how can i convert a svn repository into git?	21:11
	→ fixUp joined	21:11
doener_	elux: git-svn	21:11
ckoehler	just two different branches, features that are being worked on at the same time	21:11
	but one depends on the other	21:11
pgregory	I have a local repo that has come from git-svn, I did git-svn init https://...., followed by git-fetch -r2368	21:11
ckoehler	so eventually they will end up in the same branch, or master	21:11
Eridius\|work	pgregory: not -r2368:HEAD ?	21:12
ckoehler	but for development, they are in two branches	21:12
pgregory	Eridius\|work: I then did git-svn fetch, so yes in the end it is.	21:12
ckoehler	i guess they don't have to be dependent on each other even	21:12
doener_	ckoehler: assuming that "2" doesn't involve any merges, you could rebase 2 when you changed 1. But that's only a choice if you keep 2 local and don't push it out	21:12
ckoehler	just two features, in two branches, that i want to see at the same time	21:12
pgregory	and then I have git-svn and master, so I branch for work, using git checkout -b refcount	21:12
	and now I have git-svn master and refcount, just as I want, all of which mirror the current HEAD of the SVN repo.	21:13
ckoehler	doener_: right, but then code from 1 is in branch 2	21:13
	→ QuercusMax joined	21:14
doener_	ckoehler: if 2 depends on 1, the code _must_ be there confused	21:14
ckoehler	doener_: ya, sorry, scrap the dependency	21:14
pgregory	now, elsewhere I have another repo, that was originally based off the same SVN repo, but an earlier revision, and has some changes on it in a branch called attributes.	21:14
ckoehler	i have two separate features that are awesome	21:14
pgregory	I want to get the changes from the attributes branch into the refcount branch.	21:14
ckoehler	and i want to see them in a dev build at the same time	21:14
doener_	ckoehler: and you can easily limit eg. log output with "git log 1..2", or diff output with "git diff 1...2"	21:14
pgregory	and I can't get my head round how.	21:14
ckoehler	doener_: hm yeah...	21:15
	isn't there some kind of patch management app for git?	21:15
	so you can stack patches around?	21:15
	not sure if it would help here	21:15
pgregory	anyone got any advice?	21:15
doener_	ckoehler: just use a scratch branch for testing. "git checkout -b tmp master; git merge 1; git merge 2; #Do some testing#; git checkout master; git branch -D tmp"	21:16
pasky	ckoehler: plenty of them	21:16
	ckoehler: you can use just git with rebase -i for that	21:16
	ckoehler: or you can use stgit or guilt	21:16
	ckoehler: or if you ahve special needs, you can try for topgit, but it's still in very early development	21:16
ckoehler	doener_: good point	21:16
	pasky: that's a lot of options, thanks!	21:16
doener_	pgregory: you repos are "incompatible" in so far that the two git-svn imports created different commits	21:17
pgregory	doener_: right, so am I stuffed then?	21:17
doener_	pgregory: no. You only want the stuff in "attributes" that is not in "git-svn" in the same repo, right?	21:17
pgregory	doener_: yes	21:18
doener_	pgregory: btw, you were the one with the home/laptop/work git-svn setup, right?	21:18
pgregory	doener_: yes	21:18
doener_	osiris or something :-)	21:18
pgregory	doener_: good memory :)	21:18
doener_	ok, do you remember the git rebase command I gave you back then?	21:19
	← glguy_ left	21:19
pgregory	doener_: unfortunately, after much reading I still don't feel much closer to proper understanding :(	21:19
	→ glguy_ joined	21:19
pgregory	doener_: yes, roughly, although I have to admit I didn't fully understand "why" that worked, and reading and re-reading the git-rebase manpage hasn't helped at all.	21:20
doener_	ok, we'll need basically the same thing again	21:20
	→ digitalhobbit_ joined	21:20
	→ loincloth joined	21:21
doener_	let's try to understand rebase once more then. Gimme a moment to prepare something	21:21
pgregory	doener_: really appreciate your help, I have a feeling that proper understanding of git would be a real benefit.	21:21
	→ Ryan52 joined	21:21
	← Ryan52 left	21:22
	→ flaguy joined	21:23
	← lolage0 left	21:24
doener_	pgregory: http://git.pastebin.com/m4299a0dc -- that creates a simple git repo with two branches and shows the result in gitk. I'll use that for some explanations	21:24
	pgregory: let me know when you got the gitk running	21:24
	→ dramsay joined	21:25
Modius	Matthew Lamari	21:28
	→ skyweb joined	21:29
	→ SjB joined	21:29
	← derRichard left	21:33
	← vbabiy left	21:33
	← polpak left	21:34
pgregory	doener_: sorry, got disconnected	21:35
doener_	np	21:35
	← redondos left	21:35
	← BabelO left	21:36
	→ BabelO joined	21:37
doener_	pgregory: but you got the paste url, right?	21:37
pgregory	yep, done	21:37
	→ thannoy joined	21:38
doener_	ok, so in gitk, you should see that the side branch forked off at 2	21:38
pgregory	yep	21:38
j_engelh	fork at 2, news at 11?	21:38
doener_	pgregory: so master and side-branch have two commits in common (1 and 2)	21:38
pgregory	right	21:38
	← xazmine left	21:39
doener_	pgregory: and each of the branches has two commits that are only in one of the branch, the "a" ones in master and the "b" ones in side-branch	21:39
pgregory	yep	21:39
doener_	now, when specifying revisions, you can say, for example, that you want commits that are in side-branch, but not in master. That is "side-branch ^master" or just "master..side-branch"	21:40
	← elux left	21:40
doener_	in this case, that range would contain 3b and 4b	21:40
pgregory	doener_: what does master..side-branch actually mean?	21:41
doener_	just what I said above. "master..side-branch" == "side-branch ^master". The ^ prefix means "not". So that's the commits that are reachable from side-branch, but not from master.	21:41
	where "reachable" means that you can walk the graph that you see in gitk downwards. That is, you follow the edges in the graph from the child to the parent, but not the other way around (it's a directed graph)	21:42
	→ Sylia joined	21:43
doener_	so from "side-branch" you can walk to 4b, 3b, 2 and 1, but not to 3a and 4a	21:43
	similarly, from master you can reach 4a, 3a, 2 and 1, but not 3b and 4b	21:44
	← BabelO left	21:44
	→ tlyng joined	21:44
	→ BabelO joined	21:45
doener_	pgregory: still unclear?	21:46
	← BabelO left	21:46
pgregory	doener_: sorry, just trying to digest	21:48
doener_	no problem, take your time.	21:48
pgregory	right clear	21:48
	the syntax seems a little odd to me, .. is used as a 'range' operator in Ruby that I'm working with at the moment.	21:49
	← ckoehler left	21:49
	← richcollins left	21:50
Eludias	pgregory: In that case, be prepared that '...' is also something else.	21:50
doener_	right, that's what comes next :-)	21:50
pgregory	Eludias: cheers, I'll be sure to read up	21:50
Eludias	pgregory: '..' is also a kind of range operator (depending on the git command)	21:50
doener_	"master...side-branch" is similar. It means "all commits that are reachable from either master or side-branch, but not from both"	21:52
Eludias	pgregory: For example, 'git log a..b' shows _all_ revisions from a to b, while 'git diff a..b' shows _only_ the difference between a and b (so is independent on the intermediate revisions).	21:52
doener_	(note that with diff the meaning of .. and ... is different, because you don't specify ranges there, but let's ignore that for now)	21:52
	pgregory: for some practice, which commits are in the range master...side-branch?	21:53
	← dragonball_ left	21:53
	→ kinection joined	21:54
kinection	how do you update a submodule? it seems like git submodule update is now doing anything.	21:55
	*not	21:55
	the item is registered in .gitmodules	21:55
selckin	did you init it?	21:56
kinection	there are files in there	21:56
	i'm fairly sure i did in the past	21:56
	i did it again	21:57
wildfire	Eludias, btw both of those commands will work the same without the '..' in them (as I learnt a few weeks ago)	21:58
doener_	wildfire: no, for log "a b" and "a..b" are quite different. For diff they're the same though	21:59
	← ereslibre left	21:59
pgregory	doener_: 3a,3b, 4a, 4b	21:59
doener_	pgregory: right	21:59
	pgregory: ok, so let's move on to the rebase. You got the required basics now	22:00
	cbrake → cbrake_away	22:00
doener_	the "full" format for rebase is: git rebase --onto <onto> <upstream> <branch>	22:00
	pgregory: what rebase does, is that it tries to create new commits that introduce the same changes as some existing commits, but on top of a different base commit	22:01
	← glguy_ left	22:02
	→ jdarius joined	22:02
doener_	pgregory: say you have two patches that apply to version 1 of your program. Now version 2 is done and you want to update your patches.	22:02
	→ jackdempsey joined	22:02
doener_	pgregory: so instead of using version 1 as the base for your patches, you now want to use version 2	22:03
	→ tona joined	22:03
	← tona left	22:03
doener_	pgregory: and that's just what rebase helps you to do.	22:03
	pgregory: let's first look at the commits that rebase will try to "replay".	22:04
	pgregory: and luckily, that's quite easy, because that is just <upstream>..<branch>	22:04
	pgregory: let's say we want to move 3b and 4b, so that instead of 2, 3b has some different parent	22:07
pasky	madduck: (it's implemented)	22:07
	← tcoppi_ left	22:07
doener_	pgregory: the rebase command would then look like this "git rebase --onto <onto> master side-branch"	22:07
	pgregory: because as we have seen earlier, "master..side-branch" contains just those two commits, 3b and 4b	22:08
	pgregory: let me know when you digested that or if there are any questions	22:09
	← _graham_ left	22:09
	→ supercali_ joined	22:09
jdarius	i'm trying to git push origin master ssh:path/to/rep ... Q: how do I specify my username on the remote machine? (eg, "--user jdarius")	22:09
	→ bugfux joined	22:10
	← kinection left	22:10
	→ \ask- joined	22:10
	→ rmh3093 joined	22:11
	→ ereslibre joined	22:12
jdarius	is there a flag for that? I'm not too clear on git syntax conventions	22:13
DrNick	use normal ssh syntax	22:13
	i.e. user@host:path	22:13
doener_	jdarius: that push command makes no sense. You're trying to push a branch called ssh	22:13
DrNick	and there's also that, too	22:14
	→ cbreak joined	22:14
doener_	jdarius: it's either "git push $remote_name $refspecs" or "git push $url $refspecs"	22:14
	← jaushman left	22:15
pgregory	doener_: did you mean "git --onto <onto> master..side-branch" ?	22:15
	oh no, sorry just read	22:15
jdarius	reading up on this. thanks DrNick and doener_	22:15
doener_	pgregory: :-)	22:16
pgregory	the rebase takes the two last parameters and effectively does the ..	22:16
	confusing	22:16
	→ joydivider joined	22:16
doener_	pgregory: not necessarily. The thing is, you always rebase a branch, thus the name <branch>.	22:17
pgregory	so it considers all commits in that branch up to the fork?	22:18
doener_	pgregory: The <upstream> parameter defines your, well, upstream. You obviously don't want to alter things that are in your upstream	22:18
	pgregory: right, in this case it simply walks up to the forking point. With a more complex history, the walking might not be as easy, but the idea stays the same.	22:19
pgregory	I'm trying to work out in my head what git will actually do, will it reapply the diffs for all thos commits to the HEAD of the current branch?	22:20
doener_	we didn't define that yet. So far, we only talked about the set of commits that it will try to re-apply	22:20
	Auris-- → Auris-	22:21
	→ WALoeIII joined	22:21
doener_	Let's look at the upstream once more. Say you have "git rebase --onto <onto> origin/master master".	22:21
	In that case "origin/master" is your upstream. Stuff that comes from some other repository on which you based your work.	22:22
	So that rebase command would re-apply your commits in master that are not in origin/master. In other words: Your local commits.	22:23
	→ tusho joined	22:23
doener_	ok. Next stop: <onto>. That specifies the commit onto which you want to re-apply.	22:24
	That's simple, right? ;-)	22:24
	so: git rebase --onto master master side-branch	22:25
	pgregory: what do you expect that to do?	22:25
	→ comex joined	22:25
	← supercali left	22:25
	→ eternaleye joined	22:27
pasky	hm	22:27
	for the last 1.5 months, t/t5540-http-push.sh is broken	22:28
	→ tcoppi joined	22:28
pasky	but noone cares	22:28
	pasky wonders if there's anyone in the world ever running this test ;)	22:28
madduck	pasky: :))	22:29
pgregory	doener_: errm, reapply the changes on sidebranch to master at teh HEAD? i.e. 3b forks from 4a?	22:29
	← ijcd left	22:29
vmiklos	pasky: given that you need to do more than a 'make test' to run it, not too much people ;)	22:30
doener_	pgregory: try to avoid using HEAD or so for now (looks like you misunderstand that, we'll look at that later). But the general idea is right	22:30
	pgregory: it takes the changes from 3b and 4b (ie. master..side-branch) and replays them on top of 4a	22:31
pgregory	yep, that's what I thought	22:32
doener_	pgregory: ok, so that's done for now. Now <branch> has another purpose other than just specifying the right hand side of the range.	22:33
pasky	madduck: is tg export --quilt satisfactory for you now?	22:33
doener_	After the rebase, <branch> is also updated to point to the result of the rebase.	22:33
pasky	the code turned out to be amazingly simple	22:33
doener_	pgregory: so in this case, when the rebase is done, side-branch is updated to point to the new 4b commit	22:33
pgregory	doener_: would it much up the lesson if I did that now?	22:34
	just to see it visually in gitk.	22:34
doener_	pgregory: it would get you a bunch of conflicts due to the setup of the example	22:34
pgregory	oh	22:34
doener_	I'll quickly adjust that	22:34
pgregory	right, of course, silly me	22:35
	→ codeshot joined	22:35
	pasky hopes gitster's back is alright again	22:36
doener_	pgregory: http://git.pastebin.com/m3a5c9bcc	22:36
	pgregory: with that one, you can try the rebase	22:36
	← supercali_ left	22:37
madduck	pasky: about to try it out.	22:37
	→ lu_zero joined	22:37
	← Oejet left	22:37
	→ supercali joined	22:37
madduck	as i am packaging my first topgit-using package: topgit itself. :)	22:37
jdarius	ok, drnick and doener_ , I got it. For posterity, I was confused about what my 'git remote add' command in the previous step was doing.	22:39
jaalto	Is it normal that "git fetch -v" does not display a thing? I would expect to seen "Connecting...done" at leat if nothing is fetched.	22:39
pasky	madduck: :)	22:39
doener_	jdarius: -v makes it quite talkative here, even if there's nothing new to fetch: http://git.pastebin.com/m6b88d7ef	22:41
jaalto	I have 1.5.6.3	22:41
doener_	sorry, I meant jaalto	22:41
	1.6.0rc2 here	22:41
jaalto	doener_: That's more like it (pastebin). I wonder if it makes any difference that the repo is under superrepository.. But still.	22:42
doener_	pgregory: oh, oops, the new example has a bug in it that causes it to miss the "1" commit. But that doesn't make much of a difference	22:43
jaalto	I also tried "git ls-remote origin" and it came empty. But I have repos downloaded from there as seen by "git branch -b"	22:43
doener_	branch -b?	22:44
jaalto	-a	22:44
doener_	maybe the remote repository got wiped or so?	22:44
pgregory	doener_: done, makes sense now	22:45
doener_	jaalto: your remote branches get only deleted by "git remote prune", not by plain fetching	22:45
	← cilly left	22:45
pgregory	doener_: in gitk I get effectively one long chain of commits.	22:45
doener_	pgregory: right	22:45
pgregory	presumably if I changes something and committed on master now, I'd see the familiar fork?	22:45
doener_	pgregory: exactly, because master still points to 4a	22:46
jaalto	doener_: Hm, I tried to clone th repository again (not mine; 3rd party). and this time it reads: "warning: remote HEAD refers to nonexistent ref, unable to checkout". So something is fishy or the repository is no longer there.	22:46
pgregory	right	22:46
doener_	jaalto: yeah, looks like the remote repo got wiped or so	22:46
pgregory	doener_: it's starting to gel now, and I think I can see the solution to my problem in the details.	22:47
doener_	pgregory: ok, you know about the commit^ and commit~ syntax, right?	22:47
pgregory	doener_: no	22:47
jaalto	doener_: Shouldn't the fetch say with -v something about that?	22:47
doener_	jaalto: the repo is just empty, it's still a repo. clone complains because it actually needs to use HEAD, fetch just sees that there is nothing and thus does nothing	22:48
	jaalto: Might be a welcome addition to have some diagnostic message though	22:49
jaalto	doener_: Indeed. Plain <empty> is a little exotic.	22:49
doener_	pgregory: ok, first, the ^n syntax. With that one, you select a parent of a commit	22:50
	← bugfux left	22:50
	→ rdavila joined	22:50
jaalto	doener_: Thanks.	22:50
doener_	pgregory: so "master" points to the commit 4a, master^1 selects the first parent of master	22:50
	pgregory: so in the example, that's the 3a commit	22:50
pgregory	doener_: oh, yes, sorry misread your comment, I thought you meant "git commit^"	22:51
doener_	pgregory: as a shorthand, master^ is the same as master^1	22:51
pgregory	doener_: yep, read up on most of that.	22:51
doener_	ok	22:51
	then for some more practice, recreate the second example, and then come up with a rebase command that creates a graph where 4b is a child of 3a	22:52
	→ ivarv joined	22:52
	← tlyng left	22:53
pgregory	doener_: ok, give me a sec.	22:53
doener_	side-branch should point to the new 4b after the rebase, and 3b should not be in side-branch anymore	22:53
	→ tlyng joined	22:53
jaalto	doener_: It looks like the project is dead. How do I remote this subproject from the superproject?	22:53
doener_	jaalto: no idea, didn't use submodules yet	22:54
ivarv	newbie q: i've got a local repo with a bunch of changes. I would like to reset some of those changes (ie: bring back a directory I deleted on the filesystem, but not via git). Is this possible ?	22:54
doener_	ivarv: so you didn't commit that deletion yet?	22:54
ivarv	doener_, nope	22:54
jaalto	doener_: ok, I'll skim the docs to refresh	22:54
doener_	ivarv: then "git checkout HEAD -- directory"	22:54
	← icwiener left	22:55
ivarv	doener_, awesome ! thanks so much	22:55
	doener_, i would never have thought of checkout.. I was looking at revert and reset..	22:55
	→ ft joined	22:56
johnw-	in I want my server's repo to contain an up-to-date working tree, it's enough to put "git reset --hard HEAD" into hooks/post-update, right?	22:56
charon	so i'm preparing a patch series that depends on jc/post-simplify in the middle of it (i.e. i want 2 patches in master, and 1 kept queued until later), but that dependency will inevitably result in conflicts	22:57
	should i somehow send along the merge diff?	22:57
	if so, can format-patch do that for me?	22:57
	→ _jarod joined	23:00
pgregory	doener_: nope, can't do it	23:03
doener_	pgregory: ok, let's start with the --onto parameter. Which argument would that require?	23:03
pgregory	I can get all of side-branch to fork off 3a using "git rebase --onto master^ master side-branch" but can't work out how to select only 4b to rebase.	23:03
doener_	ok, the --onto is correct, the side-branch as <branch> is also (obviously) correct	23:04
	pgregory: now remember what we said about which commits get replayed. There was a simple rule to get the range.	23:05
	→ polpak joined	23:05
	← jdarius left	23:05
	dwmw2 → dwmw2_gone	23:05
doener_	pgregory: your command uses the range master..side-branch	23:05
pgregory	doener_: <upstream>..<branch>	23:06
doener_	pgregory: but you need a range that only contains 4b	23:06
pgregory	hmm	23:06
	← skyweb left	23:07
pgregory	side-branch^ side-branch?	23:07
flaguy	doener_: in your running example with pgregeroy, second paste, line 9: s/file/file1/ Yes?	23:07
doener_	exactly, just that the range needs ".." in between (but not the rebase command)	23:07
	flaguy: yep	23:07
	→ rdavila_ joined	23:08
pgregory	doener_: yeah, I was offering the arguments to the rebase command	23:08
	pgregory just tries this...	23:08
doener_	ok, just wanted to clarify. And yeah, trying that is a good idea	23:08
	→ Hobbes joined	23:09
pgregory	done, and all makes sense, but what's happened to 3b?	23:10
doener_	pgregory: that commit (as well as the old 4b) is still around. There is just currently no ref pointing to it, so gitk doesn't show it	23:10
pgregory	right, just to clarify in my head...	23:10
	rebase arguments, --onto defines the commit that the changes are grafted onto, and the other two define a range of commits by way of the '..' operator?	23:11
doener_	pgregory: try this: recreate the starting repo, then "git branch old-side-branch side-branch" ot create a second ref pointing to 4b, then do the rebase, and look at the results in gitk	23:11
	pgregory: yep, that's quite a valid way to remember how rebase works. Just that the last argument also specifies the branch that gets updated to point to the result of the rebase	23:12
	→ zdennis joined	23:13
pgregory	doener_: hmm done that, but the gitk graph is not easy to follow.	23:13
	← ereslibre left	23:14
	→ exg_ joined	23:14
	→ ereslibre joined	23:14
	→ Cap_J_L_Picard joined	23:14
pgregory	doener_: ahh right, gottit.	23:15
doener_	not as hard as it looked, right?	23:15
pgregory	right	23:15
doener_	ok phew ;-)	23:15
	→ Sput joined	23:15
	→ EgS joined	23:15
pgregory	just confusing with the crossing edges	23:15
	← polpak left	23:16
pgregory	so, now I think I have the tools to rebase the attributes changes onto refcount.	23:16
doener_	ok, there are a few shorthands for rebase. First, <onto> defaults to <upstream>	23:16
	so "git rebase master side-branch" is the same as "git rebase --onto master master side-branch"	23:17
tango_	damnit, svn is SLOW ...	23:17
	← ebel left	23:17
doener_	and <branch> defaults to the branch you have currently checked out	23:17
pgregory	doener_: so if I have side-branch, "git rebase master" would do the same?	23:18
doener_	so if you have side-branch already checked out, "git rebase --onto master master side-branch" is the same as "git rebase --onto master master"	23:18
	← rdavila_ left	23:18
doener_	yep, if side-branch is checked out, "git rebase master" is also the same as "git rebase --onto master master side-branch"	23:18
pgregory	cool	23:18
doener_	and that's probably also one of the most common modes of invocation	23:18
	I just started out with the full syntax, because I feel that it is easier to understand that way	23:19
selckin	if i run git commit in one terminal and have typed the commit message, then go change some code and git add it, then close my editor, will it pickup the changes?	23:19
	→ Oompa joined	23:19
segher	is there some config option (per branch) to say "git rebase" means "git rebase some-particular-branch"?	23:19
doener_	selckin: good question. try it and report? :-)	23:19
pgregory	doener_: ok, just to make sure I've got it, that "commond mode of invocation" would be to merge changes back from a branch to the master?	23:20
selckin	doener_: /lazy, gimme sec :)	23:20
doener_	segher: you can configure it so that "git pull" means "git rebase somebranch"	23:20
	pgregory: no. master is not affected.	23:20
segher	hrm... will that actually do some git fetch as well?	23:20
selckin	doener_: doesn't seem to .	23:20
doener_	segher: yep.	23:20
pgregory	doener_: nope, sorry, bring master changes up to the branch, by reapplying the branch changes further up the master?	23:21
segher	i'll have to look at that again then, thanks	23:21
doener_	segher: are we talking about a local or about a remote branch?	23:21
	→ cehteh joined	23:21
doener_	segher: ie. master or origin/master?	23:22
segher	i usually do smth like git checkout master ; git pull ; git checkout work ; git rebase master	23:22
doener_	segher: that's fine. Just specify the remote as . in the config. fetching from . doesn't do much as that's the repo itself	23:23
	you need to do that anyway as you rebase against a local branch	23:23
	→ devjj joined	23:23
doener_	pgregory: yeah, that's more like it.	23:23
devjj	Is it possible to edit/delete a remote origin once added?	23:23
doener_	devjj: git remote rm	23:24
segher	well, master really is just <some-remote>/master	23:24
devjj	doener_: Thanks.	23:24
segher	but i have it in a local branch "master" as well	23:24
	→ glguy_ joined	23:24
pgregory	doener_: thanks for all your help, much appreciated.	23:24
SamB	devjj: or you could edit your .git/config if you'd rather	23:24
	← rdavila left	23:24
	→ johnw_ joined	23:25
glguy_	is there a better way to revert specific uncommitted hunks than: git add -i (select the good code) ; git checkout FILENAME ; git reset FILENAME	23:25
doener_	pgregory: ok, now to your actual task. You would add a the old clone (the one with the attributes branch) as a remote to the new repo (git remote add old-repo /path/to/old/repo)	23:25
	pgregory: then fetch from that remote: git fetch old-repo	23:26
pgregory	doener_: yep, done that.	23:26
doener_	pgregory: then you have the attributes branch as old-repo/attributes	23:26
	then you create a local branch from that: git branch attributes old-repo/attributes	23:26
	and then rebase that branch	23:27
segher	glguy_: git add -i ; make sure you added everything you want ; git checkout .	23:27
	erm, commit first	23:27
	→ unreal_ joined	23:27
doener_	pgregory: as <upstream> you just use the first commit in the attributes branch that you do not want	23:27
glguy_	is: git checkout . like git reset --hard	23:27
segher	pretty much, yes	23:27
pgregory	doener_: all sounds good, I'll write that down and give it a got shortly.	23:28
doener_	glguy_: no. "git checkout ." puts the contents of the index into the working tree	23:28
Mikachu	git checkout HEAD -- . is like git reset --hard, yes	23:28
glguy_	while reset destroys the index	23:28
segher	doener_: well i did say commit first :-)	23:29
Mikachu	actually git checkout HEAD -- . and git reset --hard are not exactly the same	23:29
	segher is still scared of the index	23:29
Mikachu	if you have files in the index that aren't in HEAD, the checkout command will restore them too, if they're deleted in the working tree	23:29
	:)	23:29
	possibly	23:29
	← johnw- left	23:29
doener_	pgregory: good. Let me know how it works out	23:30
pgregory	doener_: will do	23:30
	pasky spends a minute trying to parse mikachu's statement	23:37
pasky	hum	23:37
	oh	23:37
	i get it \o/	23:37
Mikachu	pasky: it happens in some circumstances that i haven't determined exactly	23:37
	i didn't check "."	23:37
pasky	I just use these two intuitively	23:38
	and it feels that there are differences	23:38
doener_	http://git.pastebin.com/m757dc58c	23:38
pasky	but my head just starts hurting bad when I try to consciously define them	23:38
Mikachu	they're more different when the argument isn't HEAD of course :)	23:38
	← devjj left	23:38
pasky	well, that too :)	23:38
Mikachu	doener_: ah, but for full effect you should rm -rf a b first :)	23:39
doener_	"git checkout HEAD -- ." first updates the index to _also_ cotain the contents of HEAD, _not_ deleting any additional files	23:39
Mikachu	but it shows enough	23:39
	doener_: i think that it is not desired behaviour though	23:39
doener_	and then it updates the working tree in the same way, using the contents of the index	23:39
	Mikachu: it's described in the man page like that	23:39
Mikachu	yes, i asked junio and he said he would take a patch :)	23:39
	since them i'm optimistically bringing it up every chance i get hoping someone else will fix it :P	23:40
doener_	ah, yeah, I was around back then, didn't remember who was the other one	23:40
	Mikachu: I'm reluctant to fix it, because if I down, Murphy will have made someone use that behaviour who will badly flame me ;-)	23:40
	Mikachu: plus, I have no clue how to fix it :-)	23:41
	→ x77686d joined	23:41
Mikachu	it seems unlikely someone would depend on it	23:41
	yeah me either :)	23:41
madduck	pasky: for export --quilt, I still need to create a tg branch that depends on all the feature branches I want to include in the export	23:41
	the stage branch	23:41
pgregory	doener_: ok, got it to work (I think) with "git rebase --onto refcount attributes^^ attributes" which reapplied the top two commits from attributes.	23:41
pasky	madduck: yes	23:42
doener_	doesn't "unlikely" jsut beg to summon Murphy upon the one that fixes it? ;-)	23:42
madduck	unfortunately, both export and export --quilt includes a commit/patch/file for this branch	23:42
	which has "no changes"	23:42
doener_	pgregory: yep, if those two commits are what you wanted, and you wanted them on top of refcount, then that's exactly the right thing	23:42
pasky	madduck: I think (i) export -a might make sense (export all the branches) (ii) export -n might make sense (exclude empty patches; tg summary can detect them, you can reuse the code)	23:43
pgregory	doener_: strangely though it wouldn't accept "git rebase --onto refcount attributes^2 attributes"	23:43
	doener_: and from the git-rev-parse manpage I can't see why.	23:43
Mikachu	pgregory: the shorthand for a^^ is a~2, not a^2	23:43
doener_	pgregory: ^2 is the second parent of "attributes". Only merge commits have more than one commit	23:44
madduck	pasky: and maybe even export -f $FILE to read those branches to export from a file	23:44
	→ zdennis_ joined	23:44
pasky	I'm not sure about that one	23:44
pgregory	oh, oops	23:44
doener_	pgregory: attributes^^ is the first parent of the first parent of attributes. The shorthand for walking the first parents is ~	23:44
madduck	pasky: are you sure that "empty branches" can be ignored everywhere?	23:44
bts-	anyone here with git-svn experience know how to fetch externals now that	23:44
doener_	pgregory: so, as Mikachu said, xxx^^ == xxx~2	23:44
pasky	madduck: what do you mean?	23:44
bts-	.. now that --no-ignore-externals has been removed	23:44
	→ BOOKZ joined	23:44
pgregory	doener_: right, that makes it very clear now.	23:44
madduck	pasky: okay, then don't implement -f. This is something that can easily be done on top	23:44
	→ unreal joined	23:45
madduck	pasky: Imagine I have a tg set A->B->C and B happens to be empty	23:45
BOOKZ	I have an issue with Git.pm	23:45
madduck	so A->C actually	23:45
pgregory	that's great, thanks for all your help, not only have you helped to solve the problem, but more importantly, I understand how.	23:45
madduck	but I created it off B	23:45
	can B just be left out?	23:45
	I think so, just making sure	23:45
pasky	I can't imagine what the problem would be	23:46
	if not	23:46
	but, about -f...	23:46
	madduck neither	23:46
pasky	I'm not really happy with current tg export usage	23:46
	it's highly arbitrary that it always just exports current branch	23:46
	← unreal_ left	23:46
madduck	well, then maybe we should reolve this before I create and upload the debian package?	23:46
	because it will get used a lot	23:46
Mikachu	maybe you should clearly mark it as subject to change too	23:47
pasky	but the usage will be volatile during the 0.n releases (though mind you, I hope and expect 1.0 to be at most few weeks away)	23:47
madduck	volatile is fine	23:47
doener_	pgregory: about HEAD, I gave some explanations a few days ago. http://colabti.org/irclogger/irclogger_log/git?date=2008-08-08,Fri#l488	23:47
madduck	but if we already know that something will change, then we might just as well change it now and then upload. :)	23:47
pasky	well :)	23:48
BOOKZ	http://rafb.net/p/wkXPgf24.html	23:48
pasky	I mean...	23:48
doener_	pgregory: from that one to about 8:08 on the timeline.	23:48
	← WALoeIII left	23:48
pasky	tg export (--collapse BRANCH\|--quilt DIR) [TOPIC]...	23:48
	this would be the natural usage, wouldn't it?	23:48
	→ agib joined	23:48
	j_engelh collapses	23:48
BOOKZ	http://rafb.net/p/wkXPgf24.html # this shell script does what I want (creates a diamond)	23:48
pasky	lol	23:48
	and when I see it written now, this is still backwards compatible	23:49
	so we're fine?	23:49
j_engelh	if it does not do stacks like stgit i am not going to use it :p	23:49
BOOKZ	http://rafb.net/p/SqZFTY15.html # this perl script doesn't (A4 is not a merge)	23:49
	(using Git.pm)	23:49
pasky	j_engelh: it can do stacks	23:49
madduck	pasky: so tg export actually takes a list of branches?	23:49
	madduck tries	23:49
pasky	madduck: no	23:49
	madduck: that's just the suggested usage :P	23:49
j_engelh	pasky: yeah but the last thing you showed me was a bunch of branches	23:49
pasky	*new	23:49
	*proposed	23:49
	*etc...	23:50
	j_engelh: it can do stacks and _MORE_ :)	23:50
j_engelh	hah	23:50
madduck	pasky: okay, that would be nice	23:50
pasky	BOOKZ: so what does the script do instead?	23:50
j_engelh	is it at least written in C/C++?	23:50
	→ chrisb joined	23:50
	pasky gives j_engelh a Look	23:51
agib	does anyone know how to display all branches except 1 with gitk? is there some sort of --all-but=branch option?	23:51
pasky	madduck: okay, it would be nice if someone sent me a cute patch overnight ;)	23:51
	agib: maybe try ^branch ?	23:51
j_engelh	but but but	23:51
Mikachu	agib: --all ^foo ?	23:51
madduck	pasky: to honour enumerated branches?	23:51
pasky	j_engelh: you're welcome to rewrite tg summary in C	23:51
	it'd be great help, actually :P	23:51
j_engelh	^foo does not work	23:51
pasky	madduck: yes	23:51
	madduck: and another patch for -n ;)	23:52
vmiklos	pasky: do you plan to provide tarballs for topgit releases? it may be interesting if tg gets more popular.	23:52
Mikachu	j_engelh: ah, i guess it excludes too much	23:52
madduck	i already have a patch adding DESTDIR to Makefile, but I won't give it to you jsut yet because I'd like to have more testing i.e. more branches in the topgit source package at the moemnt. :)	23:52
pasky	vmiklos: http://repo.or.cz/w/topgit.git?a=snapshot	23:52
	madduck: sure :)	23:52
vmiklos	pasky: ah, i always forget snapshots are enabled on repo.or.cz :)	23:53
madduck	vmiklos: Debian will provide orig.tar.gz files for snapshots I package	23:53
j_engelh	madduck: you know what amazes me, that debian's mplayer fails	23:53
madduck	to play out streams?	23:53
	our	23:53
agib	Mikachu: hmm doesn't work	23:53
	something with revparse?	23:53
BOOKZ	pasky: the script does a commit, but not a merge	23:53
Mikachu	agib: yeah i take it back	23:53
j_engelh	madduck: whatever the page says, so yes :p	23:53
madduck	i believe debconf8.debconf.org/video has some information on that, j_engelh	23:53
BOOKZ	so A1-A2-A4 and no A3 in the history	23:53
SamB	madduck: what, you aren't going to use the repository directly in your debian packages?	23:53
Mikachu	j_engelh: my mplayer also failed to play it	23:54
	without the -cache trick	23:54
j_engelh	well mine does not.	23:54
pasky	BOOKZ: so what happens when it calls the merge?	23:54
Mikachu	good for you :)	23:54
j_engelh	Mikachu: which just means packman did a better job :p	23:54
pasky	i get unusable video with packman mplayer	23:54
madduck	SamB: ?	23:54
pasky	but vlc works fine	23:54
Mikachu	yeah, clearly it's up to the packager to make sure the ogg demuxer works :)	23:55
j_engelh	perhaps because there's nothing in the room atm?	23:55
pasky	oh	23:55
	you mean right now	23:55
	no idea :)	23:55
madduck	we are all eating dinner or drinking beer	23:55
	no more talks	23:55
	Mikachu (away)	23:55
BOOKZ	pasky: it says "Merge with strategy recursive failed." (which is somewhat expected)	23:55
	← david_koontz left	23:55
madduck	tomorrow again at 13UTC I think	23:55
	pasky wants video stream of madduck eating dinner and drinking beer	23:55
j_engelh	yeah I forgot to apply European time shift	23:55
madduck	pasky: you don't. :)	23:55
BOOKZ	and git status tells me that the file is unmerged and modified	23:56
madduck	pasky: I am going to release 0.2-1 now and then get hacking on that other stuff	23:56
BOOKZ	but when I change it and commit, it's not a merge , just a child of the HEAD	23:56
pasky	madduck: ok, cool	23:56
madduck	pasky: can I subscribe you to the package tracker so that you get bug reports sent directly?	23:56
pasky	sure	23:56
madduck	same email as author email?	23:56
	let's see if I can fasttrack this through Debian NEW. :)	23:57
pgregory	the only book I've found on Git is "Pragmatic Version Control Using Git", worthwhile or not?	23:57
	→ technel joined	23:57
pasky	not	23:57
technel	If dir/ is a Git repository and I copy dir/ to dir2/, is that a Git repository as well?	23:57
pasky	at least not in the form I saw it the last time	23:57
Eridius\|work	pgregory: http://git-scm.com/documentation	23:57
pasky	BOOKZ: sorry, gotta get some sleep	23:57
Eridius\|work	technel: yes	23:58
doener_	pgregory: I read part of an example chapter. not impressed	23:58
BOOKZ	pasky: thanks anyway	23:58
pasky	BOOKZ: at least I might tell you that it's more comfortable to use git rev-parse HEAD than that log incanation ;)	23:58
Eridius\|work	pgregory: there's also the in-progress http://book.git-scm.com/	23:58
technel	Eridius\|work: See, I did that (via rsync, actually), and the directory has .git/ in it, but when I run "git status" it says: git: /lib/libz.so.1: no version information available (required by git) -- any ideas?	23:58
	→ david_koontz joined	23:58
pasky	pgregory: or just stick with the User Manual?	23:58
j_engelh	technel: update libz	23:58
BOOKZ	what's weird is that the two merge commands do not give the same output	23:59
	← tvw left	23:59
	→ xenoterracide joined	23:59
j_engelh	some distributions are just to ... old!	23:59
Eridius\|work	technel: sounds like a problem with your git binary or libz, not with the git repo	23:59
pasky	pgregory: it's probably the most complete resource	23:59
	→ rswarbrick joined	23:59
technel	j_engelh: uh, is libz supposed to be a pkg?	23:59
j_engelh	i've seen that error too, and it happened before upgrading glibc + zlib	23:59
	technel: depends on which distroland you live in	23:59
SamB	hmm, my mplayer can't parse those playlists	23:59
j_engelh	SamB: uh	23:59
BOOKZ	one says: "Auto-merged A / CONFLICT (content): Merge conflict in A / Automatic merge failed; fix conflicts and then commit the result."	23:59
j_engelh	SamB: mplayer -playlist <url or file>	23:59
	SamB: works wonders btw	23:59

Logs Search ←Prev date Next date→

Channels Documentation