IRCloggy #git 2009-05-11

Provider of IRC logs since 2005.

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

	2009-05-11
killerchicken__	so what you can do is gitk, and find a commit without sensitive information in it.	00:00
	note down its sha	00:00
	→ bryanray joined	00:00
killerchicken__	sha1, that is	00:00
	then do a git rebase -i sha1	00:00
Emachman	The history is as follows:	00:00
	Today: Commit 2 - Removing personal info	00:00
killerchicken__	this will bring up an editor, where you can pick commits. Remove all commits that have offending data in it, then save.	00:00
Emachman	April 28: Commit 1 - Initial import	00:01
killerchicken__	ah. so you only have two commits in total?	00:01
Emachman	And that's it	00:01
	Only 2 to that file	00:01
killerchicken__	doesn't matter	00:01
	just pick any commit before that commit 1	00:01
Emachman	Ok	00:01
killerchicken__	do you want to keep parts of commit 1, or get rid of it entirely?	00:02
spenser	przemoc: Thank you. Hopefully this gets fixed. I'm going to get back to work.	00:03
	przemoc: bye	00:03
	← spenser left	00:03
Emachman	Commit 1 is adding a bunch of files, not just the one with personal info	00:04
killerchicken__	so you want to edit that commit?	00:04
	then, when the text editor comes up, don't delete the line with the commit, but change the "pick" to "edit"	00:05
	then you can edit that commit	00:05
	→ qhoxie joined	00:06
struberg	spearce if I like to detect git-added files, I have to compare the ObjectId of this file in the DirCache with the one from the HEAD, isn't? Is there an iterator for that?	00:07
	or is there an much easier way?	00:08
spearce	struberg: yes. CanaonicalTreeIterator.	00:08
	there also are methods on TreeWalk to add an ObjectId; it creates the CTI for you	00:08
Emachman	Well... it's out of my git repo... but it's still accessbale from github by going to http://github.com/soxred93/<repo>/commits/master/	00:08
	→ paulboone joined	00:08
	→ WALoeIII joined	00:08
spearce	struberg: note it must be a tree objectid, not a commit objectid like HEAD often is; so you need to resolve HEAD to a tree. e.g. RevWalk.parseTree(HEAD).	00:09
struberg	spearce: txs!	00:09
	← Emachman left	00:09
struberg	spearce: thought about addint it as 3rd iterator to the treewalk	00:09
spearce	struberg: yup	00:10
	that's why treewwalk takes any number of iterators	00:10
struberg	so I can compare fielsystem/index/repo	00:10
	yea, took me a bit to figure it out ;)	00:10
	spearce: CanonicalTreeParser isn't?	00:11
spearce	struberg: uh, yea, that's it. :-)	00:11
	→ seangrove joined	00:14
	← stuffcorpse left	00:15
	→ stuffcorpse joined	00:15
	← ceej left	00:17
	← Yuuhi left	00:18
	→ ashleyw joined	00:18
	← voker57__ left	00:19
	→ voker57__ joined	00:22
	← bnferguson left	00:22
	← struberg left	00:23
cxreg	"Last week at eurocrypt, a small group of researchers announced a fairly serious attack against the SHA-1 digest algorithm"	00:24
Ilari	2^52 attack reportedly.	00:25
	And its collision attack, not second preimage attack.	00:26
	← voker57__ left	00:26
cxreg	52 out of 160?	00:26
	oh, I see. hmm.	00:27
	← JohnnyL left	00:28
Ilari	cxreg: Brute force would be 2^80.	00:29
killerchicken__	cxreg: The way I see it, this is not a problem today. But everyone using sha1 better had a migration plan to another hash for when sha1 is broken, which will most likely happen soon. I'm not sure if there is any development in that region. If not, there probably should be.	00:29
Ilari	Is there faster-than-generic 2nd preimage attack for MD5 yet?	00:29
killerchicken__	(with broken, I mean not just being able to create collisions, but preimage attacks at least)	00:30
cxreg	development in that region, in git you mean?	00:30
killerchicken__	Ilari: not that I know of	00:31
	→ agenteo joined	00:32
	→ voker57__ joined	00:33
RandalSchwartz	sha1 isn't used cryptographically in git, though	00:33
Ilari	In git, its kinda problematic. The namespace from different hashes is totally distinct from SHA-1 namespace.	00:33
killerchicken__	RandalSchwartz: it is, when you sign a tag.	00:33
Ilari	RandalSchwartz: It is.	00:33
RandalSchwartz	sure, but any faking would be quickly exposed	00:33
killerchicken__	and exactly how?	00:34
RandalSchwartz	and almost certainly result in a tree that doesn't compile	00:34
	the odds of finding a collision to a published hash that also represents a good tree are not worth considering	00:34
	this has been discussed time and again on the list	00:34
	→ Anjo_Malvado joined	00:34
RandalSchwartz	it's not worth rearguing it	00:34
Anjo_Malvado	error: insufficient permission for adding an object to repository database ./objects	00:34
	google didn't help. Could anyone?	00:35
Ilari	RandalSchwartz: There are two VALID postscript files that collide in MD5.	00:35
	← elad` left	00:35
RandalSchwartz	MD5 is far smaller	00:35
	Anjo_Malvado - pushing to a shared repo?	00:35
	was it set up as a shared repo?	00:35
killerchicken__	RandalSchwartz: nonetheless, it's very possible that a preimage attack on sha1 is discovered. git better had a plan to migrate away from sha1 in that case.	00:35
Anjo_Malvado	RandalSchwartz: yup	00:35
RandalSchwartz	right permissions, etc?	00:35
	you a member of the group that owns the repo?	00:35
Anjo_Malvado	RandalSchwartz: well, I followed the instructions...	00:35
	RandalSchwartz: oh, hummm	00:35
RandalSchwartz	and the files all mode 66x in there	00:35
	shared repos need group sharing	00:36
Ilari	RandalSchwartz: Only files in logs need to be that. The rest tend to be 444.	00:36
	RandalSchwartz wanders off... can't keep typing	00:36
	← gaveen left	00:36
Anjo_Malvado	RandalSchwartz: you mean in the bare repo, .git/*	00:36
	Ilari: .git/objects/* ?	00:37
Ilari	Anjo_Malvado: If you have '.git' directory (as oppsoed to 'foo.git'), it isn't bare.	00:37
	← voker57__ left	00:37
Anjo_Malvado	Ilari: yeah yeah, my mistake. I mean: /my/repo/software.git/objects	00:38
	sorry :P	00:38
	← arohner left	00:38
Ilari	Anjo_Malvado: All DIRECTORIES there should be 2770/2775 (for multi-user repo). All the regular files should be 444.	00:38
Anjo_Malvado	Ilari: okie, thanks. Will try now :)	00:39
Ilari	Anjo_Malvado: Note, Directories only. So 'chmod -R' WILL NOT WORK right.	00:40
Anjo_Malvado	Ilari: yup.	00:40
	Ilari: you mean: chmod 2775 branches hooks info objects refs	00:41
	Ilari: or do you mean: find . -type d -exec chmod 2775 {} \;	00:42
	→ voker57__ joined	00:42
Anjo_Malvado	Ilari: and: find . -type f -exec chmod 444 {} \;	00:42
	← ashleyw left	00:42
	Anjo_Malvado better backup his bare repo	00:43
Ilari	Anjo_Malvado: The latter. Exception to that "regular files only need 444" is stuff inside 'logs' directory.	00:43
	← nabilelisa left	00:43
Ilari	Anjo_Malvado: Those files (in case they exist) need 660/664.	00:43
Anjo_Malvado	Ilari: oh, I see. trying right now.	00:43
Ilari	Anjo_Malvado: Plus everything has to have right user group.	00:44
	Anjo_Malvado: And also do 'git config core.sharedrepository world' (2775 so 'world').	00:44
	← __iron left	00:45
	← voker57__ left	00:46
Anjo_Malvado	Ilari: there is no logs directory in my bare repo	00:46
	← aziz left	00:47
	→ voker57__ joined	00:48
	→ gaveen joined	00:48
	→ jamescarr joined	00:49
jamescarr	git add * works recursively right?	00:49
	→ bnferguson joined	00:50
Anjo_Malvado	Ilari: it worked, thanks :)	00:51
	← voker57__ left	00:52
	→ qhoxie_ joined	00:52
	→ ashleyw joined	00:52
	← ashleyw left	00:53
	→ voker57__ joined	00:54
	eno__ → eno	00:54
	→ RayMorgan joined	00:55
	→ rettub_ joined	00:55
Ilari	jamescarr: 'git add <directory>' is recursive.	00:56
	jamescarr: In 'git add ', the shell expands the '' before handing it in.	00:57
	→ neoeinstein_ joined	00:59
	← neoeinstein left	01:00
	neoeinstein_ → neoeinstein	01:00
	→ qhoxie__ joined	01:02
	→ ashleyw joined	01:08
Anjo_Malvado	Ilari: my bare repo doesn't have a logs directory. Did you mean that my local repo should have right permissions for its .git/logs directory?	01:09
	→ arohner joined	01:09
Ilari	Anjo_Malvado: logs directory isn't necressary on bare repo.	01:10
	Anjo_Malvado: If it isn't present, it doesn't contain any regular files.	01:10
Anjo_Malvado	Ilari: right, so that's why I guessed you were talking about the local repo	01:10
	Ilari: in the bare repo?	01:10
	→ ceej joined	01:11
Anjo_Malvado	Ilari: I have regular files... but still the bare repo doesn't have a logs directory	01:11
Ilari	Anjo_Malvado: Nope. The bare repo can have logs directory. And if there are regular files there, all pushing users need write perms to those files.	01:11
Anjo_Malvado	Ilari: oh, I see.	01:11
Ilari	Anjo_Malvado: In fact, those are the only files in repo that aren't WORM. That's why they need write perms while all others don't (instead needing write perms to their parent directories).	01:11
Anjo_Malvado	Ilari: so I guess some other operation will trigger the creation of a logs directory in the bare repo because so far it doesn't and everything is working now (pulls,pushes)	01:12
Ilari	Anjo_Malvado: In default configuration, nothing will trigger it.	01:12
Anjo_Malvado	Ilari: what does WORM stands for? o.O	01:12
	Ilari: what would?	01:12
Ilari	Anjo_Malvado: Write Once Read Many	01:12
	→ jgalvez joined	01:12
Anjo_Malvado	humm	01:12
Ilari	Anjo_Malvado: If reflog creation would be enabled, that would create files under it.	01:13
jgalvez	how do I replace a file in the current commit with the same file from the last commit? kind of like a rollback on a single file	01:13
Anjo_Malvado	Ilari: why would I use reflog? (what is it :)	01:13
Ilari	Anjo_Malvado: Reflogs record past values of branch pointer. Really useful in working repos, not so much in others.	01:14
	← marvil07 left	01:14
Ilari	jgalvez: 'git checkout <commit> -- file'.	01:15
Anjo_Malvado	Ilari: okay, thanks for your patience :]	01:15
jgalvez	Ilari: like, git commit 08f01131fa2adbc90f121f1a504b6d4b6271dbcc -- file ?	01:16
	← jfkw left	01:16
	← qhoxie left	01:17
	→ zuez joined	01:18
Ilari	jgalvez: Something like that.	01:19
	jgalvez: That rolls back file called 'file'.	01:19
	← parasti left	01:20
	← voker57__ left	01:22
	→ voker57__ joined	01:22
	cpg → cpg\|away	01:27
	→ gberdyshev joined	01:29
	← bryanray left	01:29
	cpg\|away → cpg	01:30
	← qhoxie_ left	01:30
	← Kbyte left	01:34
	→ hpoydar joined	01:39
	→ elik joined	01:39
	→ leachim6 joined	01:39
	← paulboone left	01:40
	← mbroeker left	01:46
	← smashwilson left	01:46
	→ Alien_Freak joined	01:46
	← ceej left	01:47
	← agenteo left	01:48
	→ ashleyw_ joined	01:50
	← Sho_ left	01:51
	→ Sho_ joined	01:52
	← Paf left	01:54
	← Sho_ left	01:55
	→ paulboone joined	01:55
	← kukks left	01:55
	← Pewpewarrows left	02:00
	← ashleyw left	02:05
	← arohner left	02:07
	→ [1]intripoon joined	02:07
	→ kuadrosx joined	02:09
	→ mithro joined	02:09
	→ travisjeffery joined	02:11
	← spearce left	02:12
	← WALoeIII left	02:14
	← Anjo_Malvado left	02:17
	← poseidon left	02:18
	← hpoydar left	02:21
	→ Radar joined	02:22
	→ theCarpenter joined	02:23
	← intripoon left	02:24
	[1]intripoon → intripoon	02:24
	→ cytrinox_ joined	02:25
	← drewr left	02:25
	← ethercrow_ left	02:26
	→ scarabx joined	02:28
	→ xenoterracide joined	02:29
s1n	does github support submodules?	02:30
killerchicken__	they have a guide for it, so I'd assume yes.	02:30
	http://github.com/guides/using-git-submodules-to-track-plugins	02:30
Arrowmaster	theres not much to really support though	02:31
s1n	killerchicken__: yeah, i just saw that	02:31
	← killerchicken__ left	02:31
	→ killerchicken__ joined	02:31
killerchicken__	ups	02:31
xenoterracide	/topic is out of date... 1.6.3 is out	02:32
killerchicken__	yeah, there's not much to support. Only thing I could think of is a special way of listing a submodule in the tree browser	02:32
	→ djgera joined	02:32
s1n	killerchicken__: the guide i saw was http://perlcabal.org/syn/S06.html#Macros	02:33
	doh	02:33
	stupid copy/paste	02:33
	http://github.com/guides/developing-with-submodules	02:33
killerchicken__	s1n: I was actually just looking for any indication of "omfg github breaks my submodules"	02:34
	and there's none of that.	02:34
s1n	killerchicken__: do i actually do a 'push push master'?	02:34
	← wooli left	02:35
s1n	or is that a typo	02:35
killerchicken__	s1n: I'm not sure what that gh thing is supposed to mean	02:35
	is that github specific?	02:36
s1n	i suppose	02:36
killerchicken__	then no idea. I would assume typo, but unsure.	02:36
s1n	is there a github channel i should be asking this in?	02:36
	← ericindc left	02:36
killerchicken__	/who #github says "yes"	02:36
	← gotgenes left	02:37
	← cytrinox left	02:38
	← theCarpenter left	02:39
	← cpg left	02:41
	← gaveen left	02:42
	→ cpg joined	02:46
	→ gotgenes joined	02:47
	→ paltman94 joined	02:50
	← leachim6 left	02:50
	← xenoterracide left	02:50
	← wmoxam left	02:51
	→ tjafk1 joined	02:52
	→ xenoterracide joined	02:55
	→ wmoxam joined	02:55
	→ fujin joined	02:56
	→ Modius joined	02:58
	→ gaveen joined	02:59
	← paltman94 left	03:00
	← JPohlmann left	03:01
	→ JPohlmann1 joined	03:01
	← djszapi left	03:06
	← tjafk2 left	03:09
	→ scientes joined	03:11
s1n	is it possible to have 2 files from 2 repos in the same directory? (i'm thinking a file in master and a file in a submodule)...	03:11
offby1	don't think so.	03:11
	you could put in a symlink though.	03:12
	→ [1]intripoon joined	03:13
s1n	symlink isn't overly portable...	03:13
offby1	true	03:15
	← k0001 left	03:17
	→ k0001 joined	03:18
	→ harinath__ joined	03:19
	← JPohlmann1 left	03:22
	← harinath_ left	03:23
xenoterracide	s1n: not really, not in any git-ish way	03:23
	s1n: obviously you could just copy them around	03:23
	even in a scripted way	03:24
	← qrush left	03:24
	→ xeno joined	03:26
	→ dreiss joined	03:26
s1n	xenoterracide: yeah, i'm trying to figure out how to structure a plugin type setup using submodules	03:27
xenoterracide	the only real problem you'd have is 'in the same directory'	03:28
	← intripoon left	03:30
	[1]intripoon → intripoon	03:30
	→ bcardarella joined	03:32
	← paulboone left	03:33
	← jamescarr left	03:39
	← orafu left	03:40
	→ orafu joined	03:40
	← xeno__ left	03:42
	→ jsbmsu joined	03:45
	← jsbmsu left	03:45
	← scarabx left	03:46
	← dan left	03:46
	→ rovalent joined	03:49
	→ jrmuizel joined	03:51
	→ schacon joined	03:52
	→ ethercrow joined	03:54
	→ paltman94 joined	03:54
	→ AkumaStreak joined	04:02
	→ patmaddox joined	04:04
	→ bentob0x joined	04:08
	← jrmuizel left	04:10
	← ashleyw_ left	04:13
	→ kate21de joined	04:14
	← travisjeffery left	04:16
	← gaveen left	04:18
Fissure	hm, it seems using fast-export for archiving repos gives smaller files than using bundles (after compression)	04:19
	is there a canonical file extension for fast-export output?	04:19
	→ jrmuizel joined	04:20
	← bcardarella left	04:20
	← dreiss left	04:20
mugwump	I'd use .gfi	04:20
	→ gaveen joined	04:20
Fissure	yeah, that's what i came up with too	04:21
mugwump	Also I think your result that it's smaller will vary vastly on the repository being compressed	04:22
	← paltman94 left	04:22
	→ jony joined	04:22
mugwump	sounds like you're making a small win from a better stream compressor by forsaking the benefits of delta compression	04:23
	→ paulboone joined	04:25
	→ priidu joined	04:27
Fissure	yeah; it may only help on smaller repos	04:29
	these are for school project so they only have a semester of history	04:29
	s/project/projects/	04:29
mugwump	eg, running git fast-export --all on perl.git gives 7.6GB of output. Doubt any stream compressor's going to get that down to less than the pack size of ~100MB	04:31
	mugwump sets bzip2 going, just to see ;)	04:32
	← paulboone left	04:33
	→ dreiss joined	04:34
Fissure	bzip2? lzma, man	04:38
	← twas left	04:38
	→ twas joined	04:38
	← schacon left	04:38
mugwump	I'll try that next	04:41
	spending much more time on stream compression is an outdated notion anyway, as you'll see from the figures	04:42
	→ girishr joined	04:42
Fissure	the increase in dictionary size by more than an order of magnitude helps, though	04:46
	→ ashleyw joined	04:46
Fissure	like i said, shallow history :)	04:46
mugwump	right, well with git the window is a sliding window, which can happily slide over any other object in the repository	04:47
Fissure	i meant for lzma vs bzip2	04:47
mugwump	I'm just glad Linus went with gzip ... good choice of compressed size vs time spent in compression/decompression	04:48
	← twas left	04:49
cehteh	a bundle is much smaller than a fast export here	04:50
	mugwump: lzo would have been a nice option too	04:51
Fissure	oh, it's definitely the right move for usage... i wouldn't use anything but gzip for real-time access	04:51
	cehteh: what did you compress the fast export with?	04:51
cehteh	not compressed :P	04:51
	btw squashfs can be used a nice archiving tool	04:52
	→ paulboone joined	04:52
	← paulboone left	04:53
cehteh	you get all filesystem metadata archived, it compresses very well, much better than tar.bz for example .. it works in parallel on multicore and is very fast	04:53
Fissure	well of course the fast export is bigger if you don't compress it :P	04:53
cehteh	bzip2 has a 900k compression window at best	04:53
mugwump	or, "at worst" if you have less L3 than that	04:53
Fissure	what does squashfs use?	04:53
cehteh	for the kind of data which fast import produces thats quite suboptimal	04:54
	Fissure: dunno lzop maybe, but the biggiest wins are because metadata is very well kept small	04:55
Fissure	that's why i said bzip2 was a poor choice	04:55
cehteh	$ du -sh .git*	04:57
	13M .git	04:57
	12M .git.squash	04:57
	mhm after git gc :P	04:57
	← FreakGuard left	04:57
	← orafu left	04:58
	→ orafu joined	04:58
cehteh	well you want to archive only one branch right?	04:58
	← curvature left	05:00
	→ curvature joined	05:00
Fissure	most of these only have one branch	05:02
	← rovalent left	05:02
	→ hyperair joined	05:03
	→ Jinzhu joined	05:05
	→ bobmcw_ joined	05:06
	← bobmcw left	05:06
	← qhoxie__ left	05:08
	→ Weaselweb joined	05:10
	← joshsdc left	05:12
	← gberdyshev left	05:13
xenoterracide	git commit -a is like svn commit isn't it?	05:15
Fissure	no	05:17
	git commit -a doesn't hit the server	05:17
xenoterracide	ok so it's like it except it doesn't hit the server	05:17
RandalSchwartz	most things in git don't "hit the server"	05:18
	at least, the things we use hourly	05:18
	xenoterracide understands git	05:18
	→ ivanoats joined	05:18
xenoterracide	but I'm doing a presentation to people who are probably more familliar with svn	05:18
	besides git doesn't require a server ;)	05:19
	RandalSchwartz understands git too	05:19
	RandalSchwartz wishes people would forget SVN exists	05:19
xenoterracide	well wish me luck in making converts ;)	05:19
mugwump	Fissure: ok after 40 minutes, bzip2 got my perl.git down to 1.45GB	05:20
	still 14.5 times bigger than 20 minutes of getting git to repack the data	05:22
	← ivanoats left	05:25
Fissure	again, try with lzma -9	05:25
	← jony left	05:25
	→ jony joined	05:25
Fissure	it does much better in places where window size matters	05:26
mugwump	what's the default with lzma?	05:26
	→ FreakGuard joined	05:29
xenoterracide	there a man page explaing refspec	05:31
	?	05:31
	→ Helios joined	05:32
	← jrmuizel left	05:32
xenoterracide	how is git push origin --all different from git push origin?	05:34
	→ jceb joined	05:34
	← Helios left	05:34
JasonWoof	xenoterracide: I think it makes all your branches exist in the remote	05:35
	normally push just pushes whatever branch you're in	05:35
	I _think_	05:35
mugwump	why not read the man page	05:36
xenoterracide	git push origin will push all that are already on the remote	05:36
	mugwump: I'm reading it	05:36
	it's not clear to me	05:36
JasonWoof	I just do "git push" and it pushes my current branch	05:36
xenoterracide	if manpages weren't written in alpha kernel geek	05:36
mugwump	ok so --all is like a pushspec of refs/heads/:refs/heads/	05:36
	unadorned 'git push' pushes all matching branches	05:36
	without an explicit pushspec	05:36
	→ tim_jenssen joined	05:37
JasonWoof	mugwump: branches matching what?	05:37
mugwump	I have long advocated the default behaviour to be more what JasonWoof just expected thoguh	05:37
	JasonWoof: all heads with the same name	05:37
xenoterracide	actually the default is now to yell at you :P	05:37
mugwump	so, like refs/heads/:refs/heads/	05:37
xenoterracide	1.6.3 changes a bit	05:37
mugwump	actually --all is like +refs/heads/:refs/heads/	05:37
xenoterracide	mugwump: so it will push branches not already on the remote?	05:38
mugwump	--all probably will, due to the +	05:38
xenoterracide	ok	05:38
JasonWoof	mugwump: same as what? sorry, I don't really understand refs. I think it terms of branches, and occationally tags	05:38
	→ robinr joined	05:38
	xenoterracide still wants something that pushes to all remotes all matching branches	05:38
JasonWoof	xenoterracide: sounds like that's what "git push" does	05:39
	→ Helios joined	05:39
xenoterracide	JasonWoof: no	05:39
	don't think so	05:39
	never has for me	05:39
	but maybe I'm mistaken	05:39
JasonWoof	I didn't either, but that's my best guess at what mugwump is saying	05:39
	I haven't worked with a remote that has more than one branch, so I haven't had to know about this stuff	05:40
	or at least haven't pushed to one	05:40
mugwump	xenoterracide: there's a 'git remote update'	05:40
xenoterracide	well I have	05:40
mugwump	but that's about it	05:40
xenoterracide	mugwump: that's pulls not pushes	05:40
mugwump	right	05:40
xenoterracide	I always use that to update	05:40
	because I don't like it auto merging stuff for me	05:41
	mugwump: I want a push equivilant of git remote update	05:41
	→ Eridius joined	05:41
JasonWoof	mugwump: when I do "git push" will it ever matter what I have in local branches other than the one I currently have checked out?	05:41
mugwump	JasonWoof: when in doubt, use 'git push origin HEAD'	05:43
	or set push = HEAD in your [remote ...] section	05:43
	← drizzd left	05:43
mugwump	xenoterracide: free ponies are down the corridor	05:43
xenoterracide	mugwump: hah	05:44
JasonWoof	I like just typing "git push" but I'd like to know if/when this might get me in trouble	05:44
	← stuffcorpse left	05:44
xenoterracide	I don't think I ever noticed git push working by itself	05:44
mugwump	ok here's where it gets you in trouble. Say you've got a 'master' and a 'release-3' branch	05:46
xenoterracide	oh, and I really wish that git would start writing man pages in something other than alpha geek	05:46
mugwump	you merge in 'master' to 'release-3'	05:46
	to see how well it merged	05:46
xenoterracide	or publish a clear explaination of some of this somewhere	05:47
mugwump	you forget about this, then switch to 'master' and do some work	05:47
	finally, 'git push' will push both branches	05:47
	→ andy7534211 joined	05:47
JasonWoof	oh	05:47
mugwump	xenoterracide: why? people will just complain anyway	05:47
	they're in unix english: concise	05:48
JasonWoof	ok, so "git push" will push all branches, that have a matching branch (by name) in the remote repo?	05:48
mugwump	exactly	05:48
xenoterracide	mugwump: there in git speak	05:48
	refspec still means nothing to me	05:48
	and I've still never found an explaination of it	05:48
mugwump	where are you reading 'refspec' ?	05:49
andy7534211	can I have two `remote.origin.url's? so when I do git pull it will pull from two repositories?	05:49
xenoterracide	mugwump: git help push	05:49
	the entire secion on it is still reading kinda alien	05:49
	I know how to use push and I still don't understand what it's saying	05:50
mugwump	so, <refspec> is described in the second paragraph	05:50
	of OPTIONS	05:50
xenoterracide	yes I've read it	05:50
mugwump	which part is difficult?	05:51
	xenoterracide doesn't really want to go into tearing it apart now	05:51
xenoterracide	I'm just griping	05:51
	→ khmarbaise joined	05:51
xenoterracide	and trying to figure out what to put on this slide	05:51
mugwump	ok well why not suggest improvements to the list as a patch or send your critique there	05:51
	→ stuffcorpse joined	05:51
xenoterracide	so I can explain it to other people	05:51
JasonWoof	xenoterracide: looks to me like the refspec thing for push is a way of pushing specific commits to a specific place on the remote	05:52
	hopefully you'll never need to do that	05:53
xenoterracide	JasonWoof: git push simply pushes to the remote of the current branch	05:53
	although I think all remotes would make more sense	05:53
mugwump	you could submit an enhancement to 'git remote' maybe	05:54
JasonWoof	at least with the colon. if you have a branch name that matches the name of the remote branch you want to push to, you can just give the name of the branch to push	05:54
mugwump	I actually have already submitted a change to that section of the manual based on IRC comments	05:54
JasonWoof	xenoterracide: mugwump just told me that "git push" pushes all branches that have remote branches with matching names	05:54
mugwump	I guess the man pages could do with more glossaries	05:55
xenoterracide	JasonWoof: right	05:55
	but that's not the same as pushing all remotes	05:55
mugwump	would that help?	05:55
	git remote \| xargs -n 1 git push	05:55
	:)	05:55
JasonWoof	:)	05:55
xenoterracide	heh	05:55
JasonWoof	mugwump: that's cheating ;)	05:55
xenoterracide	I don't actually need it these days	05:55
	I quit working on the project where I could have used it	05:56
JasonWoof	I don't think I'd want to push more than the current branch	05:56
xenoterracide	JasonWoof: I did it all the time	05:58
	you've worked on 2 or 3 branches then push	05:58
	you shouldn't necessarily be pushing after each commit anyways	05:58
	it cut down on bandwith I'm sure, because I'm sure some objects were shared from merges	05:58
mugwump	JasonWoof: there's a new setting push.default	05:58
xenoterracide	yeah it's very annoying	05:59
	because now git yells at you if it's not set	05:59
mugwump	JasonWoof: in git 1.6.3	05:59
xenoterracide	and you do a git push origin	05:59
	or whatever	05:59
mugwump	JasonWoof: use git config push.default current	06:00
	→ Tuomas joined	06:00
	← gotgenes left	06:01
mugwump	JasonWoof: see man git-config	06:01
Gitbot	JasonWoof: the git-config manpage can be found at http://git.or.cz/man/git-config	06:01
JasonWoof	cool	06:01
	I look forward to that feature, when I have a newer git	06:01
	→ aspotashev joined	06:02
JasonWoof	I've got 1.6.0.6 right now	06:02
mugwump	dark ages! :)	06:02
JasonWoof	I bet I'll have that version of git before I need the feature	06:02
	← aspotashev left	06:02
	→ aspotashev joined	06:02
mugwump	that's almost 5 months old!	06:03
	← FreakGuard left	06:03
	← hyperair left	06:03
JasonWoof	omg ;)	06:03
	I must be using a distro with a 6-month release cycle	06:04
xenoterracide	lol	06:04
	or cygwin	06:04
JasonWoof	I'll have to wait another 11 days before I'll get a new version	06:04
mugwump	no need to wait! build from source! :D	06:04
	xenoterracide had to wait like 3 days for git 1.6.3	06:05
	← girishr left	06:05
JasonWoof	yeah, yeah. first I'm going to learn how to use it	06:05
xenoterracide	it was excruciatingly long	06:05
	← RayMorgan left	06:05
JasonWoof	git development does seem to go very fast	06:05
xenoterracide	JasonWoof: so does kernel development	06:05
mugwump	Actually tbh I'm happy with even the version in lenny	06:06
xenoterracide	most distro's just don't patch like they should	06:06
JasonWoof	this is not the first feature that has come up as a solution to my questions/concerns that has been created since 1.6.0	06:06
xenoterracide	yeah, git's updates aren't so important imo	06:06
JasonWoof	git 1.6.0.6 works great. no complaints here	06:06
mugwump	there will be a version that will change all the defaults to produce fewer unexpected results at some point	06:07
JasonWoof	good	06:07
mugwump	then that will probably be considered a minimum version for most people, just like 1.5 was	06:07
	→ SRabbelier joined	06:07
mugwump	that push.default is one example of that	06:07
JasonWoof	I think it is wise to have the defaults be newbie friendly even at the cost of being powerful	06:07
mugwump	well, you need a deprecation cycle too	06:08
JasonWoof	wazzat?	06:08
	→ ashleyw_ joined	06:08
mugwump	if the defaults change, you need to warn people	06:08
JasonWoof	default -> an option -> an option you're not supposed to use anymore -> can't do it	06:08
	oh	06:08
	oic	06:09
	so it takes a couple versions	06:09
	→ agenteo joined	06:09
JasonWoof	one in which it warns people that if they want to continue to have it work the way it currently does, they have to add something to their config	06:09
mugwump	sure	06:09
JasonWoof	and if they don't, they'll get the new default next version	06:09
	yep, gotta change slow	06:10
mugwump	not slow. but not fast.	06:10
JasonWoof	guess I should be more specific there... when you're making changes that change the effects of commands people use a lot... you have to change slow	06:11
	grrr... still can't get it right	06:11
	eh, came out ok. I'm losing the ability to read though. should go to bed	06:12
mugwump	heh, sleep well	06:14
	→ hyperair joined	06:16
	← robinr left	06:18
	← aspotashev\|eeepc left	06:18
JasonWoof	_should_ doesn't seem to work on my for going to bed :)	06:20
	ok, push.default looks awesome. I'm trying to figure out whether I should have it set to "tracking" or "current"	06:20
	← jceb left	06:21
JasonWoof	I imagine in most cases it wouldn't make any difference	06:21
	"tracking" makes it so I can have a local branch correspond to a remote branch, and have that local branch have a different name than the remote one?	06:22
	← johan-s left	06:22
mugwump	yes, tracking does that	06:22
JasonWoof	and "current" just does all local<-->remote branch matching by branch names	06:22
	→ Beket joined	06:22
JasonWoof	?	06:22
	→ wmoxam_ joined	06:23
	← ashleyw left	06:23
mugwump	check the man pages .. experiment :)	06:23
JasonWoof	sounds like a good idea	06:23
	today I made myself a test repo, made some stupid commits, and massed around with rebase a lot	06:24
	→ girishr joined	06:24
JasonWoof	I feel much more confident with it now	06:24
	practiced branching off and doing my commit rewriting with rebase in the branch, and (once I got it right) learned how to make my master branch match that (reset --hard rewritten-branch)	06:26
	→ FreakGuard joined	06:26
JasonWoof	dunno why it makes me so much more comfortable to do all my experimenting with rewriting in a branch	06:27
	I'm afraid that if I do it on master I won't be able to get it back	06:27
	but that's silly. I could just make a branch or tag at the current master's HEAD, then go nuts, and I can always reset --hard master back to the way it was	06:27
	← wmoxam left	06:28
Beket	why not just branch, develop/rebase and then merge to master ?	06:29
	← ashleyw_ left	06:31
	→ giallu joined	06:36
	← gaveen left	06:36
	→ lohrun joined	06:37
	rjc_ → rjc	06:38
	→ tal67 joined	06:39
	← kate21de left	06:41
	← knittl left	06:43
	→ angerman joined	06:44
	→ RodP2 joined	06:46
	→ galderz joined	06:47
	→ knittl joined	06:48
	→ przemoc joined	06:50
	→ Stravy joined	06:55
	→ johan-s joined	06:58
	→ dohzya joined	07:01
	← kuadrosx left	07:01
	→ ceyusa joined	07:04
	→ solydzajs joined	07:05
JasonWoof	Beket: I was working on the situation where master had a commit in it that I didn't like, and didn't want to make public ever	07:05
Beket	JasonWoof, cool. AFAIK it's possible to exclude a commit (or range of commits) with rebase, but never tried it actually.	07:06
	← Radar left	07:07
	→ ciskje joined	07:07
	→ solitar joined	07:08
JasonWoof	yeah, you can mark them as "skip" with rebase -i, or use rebase's "--onto" option	07:08
	did the --onto thing once, still confused by it	07:09
	← angerman left	07:10
Ilari	JasonWoof: --onto controls the point rebuilding the branch starts from.	07:13
Beket	it's basically where the result of rebase is appended, right ?	07:14
	really roughly speaking*	07:14
	← nighthwk1 left	07:15
Ilari	Beket: Pretty much. Non-interactive rebase is bit like format-patch <upstream> + reset --hard <onto-commit> + am. <onto-commit> defaults to <upstream> if not specified.	07:17
	← AkumaStreak left	07:18
Beket	nicely said, thanks Ilari	07:18
	→ darkskiez_ joined	07:18
Ilari	Beket: And the second branch (if specified) acts like 'checkout <branch>' at the beginning.	07:19
	→ jmesnil joined	07:20
	→ ljungk joined	07:21
	→ kalle_ joined	07:22
	← FreakGuard left	07:23
	← kalle_ left	07:26
	← ljungk left	07:26
	→ ljungk joined	07:27
	→ gilimanjaro joined	07:27
	→ mugwump joined	07:33
	→ merlin83 joined	07:34
	→ mtkd joined	07:35
	→ FreakGuard joined	07:36
	← mugwump left	07:37
	→ mugwump joined	07:37
	→ dwave joined	07:39
	→ webcoder joined	07:39
	← FreakGuard left	07:41
	→ psankar joined	07:41
	→ ph^ joined	07:43
	→ eb_ joined	07:44
	→ JEEB__ joined	07:46
mugwump	Fissure: 108minutes of processing time, and lzma has reduced the 7.6GB input to 484MB	07:49
	I'll set lzma -9 going overnight :)	07:49
	→ eno__ joined	07:49
Arrowmaster	....	07:50
	← bobesponja left	07:50
Arrowmaster	thats nuts	07:50
	→ aphexer_ joined	07:50
eMBee	good afternoon	07:50
Fissure	like i said, it beat bzip2 pretty spectacularly :)	07:50
cehteh	"longer is better" :)	07:51
eMBee	is this article confusing or am i just not imersed in git myself yet? http://thecarlhall-dev.blogspot.com/2009/04/rename-git-remote-branch.html	07:51
	← ceyusa left	07:51
Fissure	7-zip uses the same algorithm and if you have 64-bit and enough ram you can make the window size up to 1 gig :)	07:51
eMBee	what does a larger window size do? allow duplication between the beginning and the end of the window be discovered?	07:52
	→ ia_ joined	07:53
eb_	hey, I checked out a tag without creating a new branch locally. so it was on (no branch). then I accidentally added a new file to (no branch). after I noticed it I switched to the master branch, but then that newly added file was gone. any way I can retrieve it again?	07:53
	→ Leemp2 joined	07:53
	→ j416 joined	07:53
eMBee	eb_: check your reflog	07:54
Fissure	git reflog HEAD, find commit id of old no-branch commit, use checkout $commit -- $file	07:54
eMBee	the commit you made should still be there	07:54
eb_	thx guys	07:54
	→ webcoder_ joined	07:54
	← ia left	07:54
	→ shruggar joined	07:55
eMBee	or cherrypick the commit into the branch that you want, or create a new branch at that poit (git checkout -b newbranch $commit	07:55
	→ psoo joined	07:55
	← psoo left	07:55
	← fujin left	07:55
	→ naeu joined	07:56
eb_	I will try the cherrypick into the master...	07:56
	cheers	07:56
	→ fujin joined	07:56
	← solydzajs left	07:59
eb_	lovely ;]	07:59
	coming from svn and only know a handful of commands in git	08:00
	but already i like it	08:00
	← bdowning left	08:00
	← ljungk left	08:01
	→ gretch joined	08:01
	← eno left	08:01
	← Leemp left	08:02
	← Ademan left	08:02
	→ Radar joined	08:02
	→ Ademan joined	08:03
	→ bdowning joined	08:03
	← webcoder left	08:03
	← gilimanjaro left	08:04
	→ ljungk joined	08:04
	← towski_ left	08:07
	→ jfkw joined	08:07
	→ Kbyte joined	08:09
shruggar	how do I show the log, but only for commits which effected a file in the current directory or lower?	08:11
eMBee	should be git log -- path	08:12
shruggar	thanks	08:12
	andy7534211 → luglog	08:12
	→ ceyusa joined	08:13
	→ shikamar1 joined	08:15
	← RodP2 left	08:15
	luglog → andy753421	08:16
Ilari	eMBee: Looks confused. That command looks like it should rename the remote (purely local thing).	08:16
	← shikamar1 left	08:16
eMBee	Ilari: exactly	08:17
	if you read his previous post it is more clear that the comand does what he intends to do, he just uses a confusing way to explain it	08:18
	→ marc_m joined	08:18
	← ljungk left	08:19
	→ RodP2 joined	08:20
	→ FreakGuard joined	08:21
	→ tbf joined	08:21
	← marc_m left	08:24
	→ marc_m joined	08:25
	→ ljungk joined	08:26
	← j416 left	08:31
	→ smeevil joined	08:31
	→ obeattie joined	08:33
	→ FlaPer87 joined	08:33
	→ abstrusething joined	08:33
	← aspotashev left	08:34
FlaPer87	hey guys, I've a wuestion... I've read around that keywords are not supported in git, so, Is there a way/workaround to keep using things like http://pastebin.com/m7914292b in my files?	08:34
	← andy753421 left	08:34
FlaPer87	I would like those values to be updated each time a user does a checkout	08:35
	← obeattie left	08:35
	→ cedricv joined	08:37
	← Modius left	08:37
	→ tono joined	08:38
Ilari	FlaPer87: Why won't you use VCS services to find the version while source is in version control and only expand keywords at export time?	08:39
	→ sunoano joined	08:39
	→ un\|ogic joined	08:40
Ilari	FlaPer87: git archive can expand all types of format strings --pretty=format: of git log accepts.	08:40
	FlaPer87: See man gitattributes and man git-log	08:41
Gitbot	Ilari: the gitattributes manpage can be found at http://git.or.cz/man/gitattributes	08:41
FlaPer87	Ilari: Ooook, I'll take a look to gitattributes thanks a lot	08:41
	=D	08:41
Ilari	FlaPer87: In practicular search for 'export-subst'.	08:42
	FlaPer87: Also look at how git.git does getting its version number.	08:43
	→ Weaselweb_ joined	08:44
	← Weaselweb left	08:44
	Weaselweb_ → Weaselweb	08:44
	→ FlaPer87_ joined	08:50
	← FlaPer87 left	08:52
	FlaPer87_ → FlaPer87	08:53
	→ ph^_ joined	08:53
	→ fluxusps joined	08:53
	← fluxusps left	08:53
	← FlaPer87 left	08:54
Knirch	In my project I have v1 v2 v3, v3 was created by converting from another scm, v1 and v2 were then reset to their respective heads. my problem now is that my v3 specific tags exist in v2. any creative ideas on how to get rid of the tags that aren't really part of the history?	08:57
wereHamster	what are v1,v2,v3? tags? branches?	09:00
	→ JPohlmann1 joined	09:00
wereHamster	and what do you mean with 'v3 exists in v2'?	09:00
Knirch	you could seem them as branches, but currently they're seperate repositories	09:00
	uhm, no you can't see them as branches, my bad :)	09:01
wereHamster	so they are separate repositories...	09:01
Knirch	yes	09:01
	← _Vi left	09:02
	→ killerchicken_ joined	09:03
Knirch	in hindsight I should have converted the tags into v1, then the missing into v2 and then v3.. but the conversion will take ~3 hours, so not too keen on redoing it	09:03
wereHamster	how can you identify v3 tags that don't belong in v2?	09:05
	→ struberg joined	09:05
	← JPohlmann1 left	09:06
	→ Bogh joined	09:06
Knirch	build_v3.<buildnr> is one type of tag I have, those I could just "manually" delete in the v1/v2 trees	09:06
	hmm, guess I could do something creative if I manage to get the same output as tig/gitk/etc, ie list the tags in git log output	09:07
wereHamster	for r in `git for-each-ref refs/tags/build_v3.*`; do git tag -d ...; done	09:08
	JEEB__ → JEEB	09:08
	← ph^ left	09:09
Knirch	wereHamster: well yes, that takes care of the special log tags, then I have 5 years worth of developers adding random tags :)	09:09
wereHamster	is there any way you can automatically identify those? By looking at their history for example?	09:10
Knirch	is it possible to list the tags aswell in git log output?	09:10
Ilari	Knirch: --decorate?	09:12
Knirch	Ilari: sweet, thanks	09:13
	found one silly way	09:16
	$ git log --pretty=format:%H >/tmp/taglist	09:16
	$ for tag in $(git tag -l); do if ! grep -q $(git log -n1 $tag --pretty=format:%H) /tmp/taglist; then echo $tag REMOVE ; fi ; done	09:16
	nice, v1 went from 16M -> 6.5M	09:17
	thanks for the talk	09:17
	→ solydzajs joined	09:18
	← killerchicken__ left	09:19
	← dreiss left	09:20
	→ j416 joined	09:23
	← Stravy left	09:27
	→ peritus_ joined	09:27
	→ ph^ joined	09:29
	← j416 left	09:30
	→ gressie joined	09:31
	← gressie left	09:31
	→ aspotashev joined	09:31
	→ tango_ joined	09:33
	→ webcoder joined	09:35
	← dodo_the_last left	09:35
	← Bogh left	09:38
	→ jaeckel joined	09:41
	← ljungk left	09:42
	← webcoder_ left	09:43
	→ MisterN joined	09:44
	→ mib_f6i623 joined	09:44
	← darkskiez_ left	09:45
	← ph^_ left	09:46
	← khmarbaise left	09:53
	← JasonWoof left	09:54
	← aspotashev left	09:55
	→ ignas joined	09:57
	→ khmarbaise joined	09:57
ignas	hi	09:57
	→ Ryback_ joined	09:58
	← Beket left	10:00
	→ __iron joined	10:00
ignas	is there a way to connect additional parent to a revision?	10:01
	→ Bogh joined	10:01
	← yoann left	10:02
AAA_awright	ignas: git merge with the ours strategy	10:03
	← mib_f6i623 left	10:03
	← peritus_ left	10:03
ignas	ours strategy?	10:03
	also - the parent is way back in the history	10:03
	(it)	10:04
AAA_awright	--strategy ours	10:04
ignas	it's a result of import from other VCS	10:04
doener	ignas: grafts	10:04
AAA_awright	It won't work if it is an ancestor	10:04
doener	ignas: look for "grafts" in the git-filter-branch man page	10:04
	→ Lenary joined	10:05
	← Kbyte left	10:05
doener	ignas: filter-branch is also what will make the grafts permanent (by creating new commits with the changed parent information)	10:05
	→ Kbyte joined	10:05
	→ njouini joined	10:05
ignas	hmm	10:05
	→ miib_e6i623 joined	10:06
	njouini → nisse	10:06
nisse	Hi!	10:06
	I have a git-svn problem.	10:06
	→ seb42 joined	10:06
Ilari	...	10:07
doener	nisse: crystal ball is in repair	10:07
nisse	:d	10:07
	Well, we do not need that.	10:07
	The problem is a bit obscure.	10:07
	You see, I get lot's of conflicts when I try to dcommit	10:08
	Maybe I've just missed something important.	10:08
	Well, the question is: Is local branching/merging bad when working with git-svn	10:09
	?	10:09
AAA_awright	Avoid it, that's all I know from the git-svn man page	10:10
nisse	But: Then there's no use of git.	10:10
	I have to do rebase all the time.	10:10
	And keep my developing linear	10:10
killerchicken_	no. you just have to linearize it before you commit to svn.	10:11
Ilari	nisse: Topic branches work right if you keep them up to date by rebasing.	10:11
killerchicken_	there are still huge advantages over regular svn	10:12
Ilari	nisse: When topic branch is finished, do final (interactive) rebase it on top of master and then do something like 'git push . HEAD:master' ("reverse-merge" current branch to master).	10:12
	→ peritus_ joined	10:12
Ilari	nisse: Then check out master and dcommit.	10:13
nisse	Hmm, that seems more complicated that it needs to be.	10:13
Ilari	nisse: Also, git svn has 'git svn rebase'. It rebases current branch on top of correct one.	10:13
nisse	Maybe I have wrong workflow here.	10:14
	My current is:	10:14
	(on master) git checkout -b new_feature	10:15
	(on new_feature) git svn rebase	10:15
	everyting is well here	10:15
	then I do some further branching and merging	10:15
	and merge everything to new_feature	10:16
	and when I do git svn dcommit everyting blows up!	10:16
Ilari	nisse: 'git merge' and 'git svn' don't mix (at least if you don't really know what you are doing).	10:16
nisse	Okay, I guess both are true in my case ;)	10:17
	Ilari what my workflow should be then?	10:17
	If I cannot branc/merge as I like and then dcommit	10:18
	→ bizhat2 joined	10:18
nisse	svn is like disease	10:19
Ilari	nisse: When beginning each feature, make a branch for it. Make some commits there and keep it up to date with 'git svn rebase'. When finished, use interactive rebase (or whatever to clean it up) and update for one final time. Then merge topic branch into master and dcommit it.	10:20
nisse	It propagates itself even into my local git repos	10:20
	Okay, a sec, I'll digest what you've written.	10:20
	← abstrusething left	10:20
	← ccherrett left	10:21
nisse	Ah!	10:22
	Ilari: okay I get that	10:23
	What about squashes?	10:23
Ilari	nisse: Squashes are effectively the SVN way of merging.	10:23
nisse	When I do commit's to svn I want them to be squashes of my development history	10:24
	I have no interes in publishing my minor commits to svn	10:24
	So before I do dcommit I've been doing merge --squash another_branch	10:25
	Could that result in problems?	10:25
	← bizhat2 left	10:25
Ilari	nisse: Nope.	10:25
nisse	Hmmm...	10:25
Ilari	nisse: As squash is effectively about redoing the whole side branch as "unrelated" change.	10:25
nisse	Okay	10:25
	← mithro left	10:26
nisse	Hmm.	10:26
	Ilari: okay. Now I do understand bits from here and there.	10:27
Ilari	nisse: The reason why git-svn is so limited is that SVN is so limited.	10:27
nisse	Ilari: yeah, it was my first scm	10:28
	Crappy	10:28
	Ilari: Hmm, so I should always dcommit from master?	10:28
	→ Modius joined	10:29
nisse	and merge branches to it	10:29
Ilari	nisse: If dcommitting to trunk yes. Other SVN branches should have their own "primary" branches.	10:29
nisse	Okay	10:29
doener	hm, I usually don't go that far. I just dcommit on the topic branch and then rebase master	10:29
nisse	And it doesn't matter if I do (on master) merge --squash foo_brach and then dcommit?	10:29
	doener: Hum?	10:30
Ilari	nisse: Note that git-svn only pays attention to line through history that passes through first parents.	10:30
	nisse: Since all SVN "branches" are linear.	10:30
	→ mithro joined	10:32
doener	nisse: When you keep "master" clean, work on "foo" and keep "foo" up-to-date, then the "git merge foo" would be a fast-forward	10:32
	nisse: i.e. master is just updated to the state of "foo", there's no merge commit being created	10:32
	nisse: and then dcommit will send the individual commits to svn	10:32
nisse	Hmm, I'm even more confused that I was when we begun :)	10:32
	→ ccherrett joined	10:33
doener	nisse: the same thing happens when you do the dcommit on foo, and then update master from svn	10:33
Ilari	nisse: Well, just don't try feeding git svn true merge commits.	10:33
nisse	Okay, maybe if I tell the naive thing what I want you guys could just tell me the right workflow	10:33
	I want to have feature branches and then commit to svn when feature is finished	10:34
doener	nisse: if you do "merge --squash", or if master is ahead of "foo" when you merge and a merge commit is created, svn won't see the individual commits, but just a single big set of changes in a single revision	10:34
Ilari	nisse: True merge commits show up on git log as having 'merge xxxxxx... yyyyy...' line	10:34
nisse	Okay	10:34
	And I do not want my teammates to see my individual "fixed typo in comments" - type of commits here	10:35
	→ gitte joined	10:35
nisse	So should I :	10:35
Ilari	nisse: Eliminating those should belong to final interactive rebase pass (to clean up stuff before sending forward).	10:36
	← solydzajs left	10:36
nisse	Ilari: could you be more verbose	10:36
Ilari	nisse: When topic branch is "finished", it needs to be cleaned up (commits split, cobined, rearranged) before submitting it upstream.	10:37
nisse	Yes	10:37
Ilari	nisse: In that process, one can combine those "oops, fix" commits into commits that they fix.	10:38
	→ Lenary_ joined	10:38
nisse	Okay	10:38
un\|ogic	how can you move the branch HEAD to a different commit?	10:38
	→ dodo_the_last joined	10:38
un\|ogic	nvmind, got it now	10:40
Ilari	un\|ogic: 'git checkout <hash>' (read-only) or 'git reset --hard <commit>' (kill uncommitted changes and delete commits after that from current branch).	10:40
un\|ogic	thnx	10:40
Ilari	un\|ogic: Sometimes those get mixed up. Not a good thing.	10:40
nisse	http://pastie.org/474304	10:41
	← girishr left	10:41
un\|ogic	i know. I need to do it for illustrative purposes :)	10:41
	→ Yuuhi joined	10:41
nisse	Ilari: Okay my workflow is there	10:41
	Is there somethign wrong	10:41
un\|ogic	doing: "git checkout <SHA>" followed by "git branch -f master" does the job	10:41
doener	nisse: that'll leave just a single commit in svn, making its history less useful	10:41
nisse	doener: This is what I want	10:41
	→ brizly1 joined	10:42
	→ sitaram joined	10:43
nisse	But anyways. Doing that results in horrible conflicts	10:43
	What's wrong there?	10:43
Ilari	nisse: Seems reasonable (apart from commit message from that dcommit). And 'git add .' is mostly for importing stuff...	10:43
	nisse: Maybe master isn't up to date with SVN at beginning? Or SVN gets commits at the middle?	10:44
	s/dcommit/merge/	10:44
	← tbf left	10:45
nisse	Ilari: no svn does not get any commits	10:46
	→ _Vi joined	10:46
FreakGuard	hm... is there a small script about branch merging?	10:46
	copying stuff from one to another.. rather complicated	10:47
ignas	why are my tags not appearing in the remote repository?	10:49
Ilari	nisse: Note that 'git svn dcommit' first does equivalent of 'git svn rebase' before the actual committing takes place.	10:49
	ignas: 'git push <remote-repo> <tagname>'?	10:49
ignas	I am tagging things on my master, then comminting + pushing, but the tags do not show up on the remote	10:49
	ahh, I have to push tags explicitly	10:49
	← Lenary left	10:49
	→ bizhat2 joined	10:50
sitaram	yes because tags are usually local (I making notes to myself...)	10:51
Ilari	ignas: And no need to commit after making tag for it to somehow "be part of branch". Tags never are part of some branch (they are in effect branches of their own).	10:51
	→ bx2 joined	10:51
Ilari	ignas: The "tags need to be committed" is one type of SVN braindamage.	10:52
ignas	Ilari: ahh, makes sense, thanks	10:52
FreakGuard	I've got a remote on github... how to push two branches?	10:53
	→ v0lksman joined	10:53
Ilari	FreakGuard: 'git push <remote-repo> branch1 branch2'	10:53
	← bizhat2 left	10:54
	← brizly left	10:56
	→ struberg1 joined	10:56
FreakGuard	I need to refork from a branch and make that branch my new master while renaming the other	10:57
	but my knowledge of git is somehow limited :)	10:57
	hmmmm	10:58
	→ PupenoG joined	10:58
	← Lenary_ left	10:58
PupenoG	How do I change branches when I get the "error: You have local changes to 'ueoaueaou'; cannot switch branches."?	10:58
FreakGuard	PupenoG: git stash maybe?	10:59
PupenoG	FreakGuard: I've changed three files, I need to switch branches, commit one, come back, merge that other branch, commit two.	10:59
	← knittl left	10:59
FreakGuard	PupenoG: git stash saves your current changes	10:59
	git stash apply applies them back	10:59
	→ aziz joined	11:00
PupenoG	FreakGuard: if I git stash save, when I git stash apply on the other branch, I'll get a merge conflict I'll have solve, and after merging them I'll get back and get another conflict.	11:00
FreakGuard	uh-oh	11:01
	ask someone else	11:01
	← struberg left	11:01
	← miib_e6i623 left	11:02
PupenoG	FreakGuard: ok, thanks.	11:02
	← hyperair left	11:02
PupenoG	So, anybody else know how to do it?	11:02
	→ hyperair joined	11:02
	→ tbf joined	11:02
PupenoG	I remember git stash save used to only pick those files added (the index?), that was useful. Now it's less useful (by picking everything).	11:03
	← gretch left	11:04
	← sitaram left	11:05
	→ knittl joined	11:06
	← solitar left	11:09
	→ girishr joined	11:10
	← Jinzhu left	11:11
	← peritus_ left	11:14
	→ mbroeker joined	11:14
webcoder	hello all, does anyone knows solution to these messages:	11:15
	error: src refspec master does not match any	11:16
	i am able to clone and push gitosis-admin with no problems	11:16
	but followed three tutorials for creating new repos and this is all i get	11:17
FreakGuard	how to copy a file from another branch to the actual one?	11:18
webcoder	after initialising git init in my local folder even added in .git/config some extra lines i.e. remote "origin" with correct URL and branch "master"	11:18
	→ boto joined	11:22
	→ masak joined	11:22
	→ grahal joined	11:23
wereHamster	FreakGuard: git show otherbranch:path/to/file > path/to/file; git add path/to/file; git commit	11:23
	but if you want to preserve history, it will get much more complicated	11:23
FreakGuard	wereHamster: hm... I somehow crashed something on merging the two branches... how to remerge? :)	11:24
jaeckel	PupenoG: try 'git help stash'	11:24
	← struberg1 left	11:25
	→ mikl joined	11:25
wereHamster	FreakGuard: look at the reflog, reset --hard	11:25
	→ bob2 joined	11:26
	→ struberg joined	11:26
FreakGuard	wereHamster: git merge <remote> right?	11:27
	→ sdboyer joined	11:27
wereHamster	git merge <nameofbranchorsha>	11:27
	← priidu left	11:28
FreakGuard	wereHamster: or fetch	11:28
	→ angerman joined	11:28
wereHamster	depends on what you want to do...	11:28
FreakGuard	worked out, thanks ;)	11:30
	→ priidu joined	11:30
	→ jstein joined	11:31
	→ Lenary joined	11:31
	← dwave left	11:32
	← jony left	11:32
jstein	by accident i copied a large directory into my source tree and added it with git. How can i get rid of it? My .git is now 280 MB instead of 8 MB. The wrong dir is deleted in the sourcetree now. and i tried git clean -f and git add -u but	11:33
	→ dwave joined	11:33
jstein	git status shows still so many "# deleted: " lines	11:34
wereHamster	jstein: did you commit the new files?	11:34
jstein	wereHamster, the wrong files? yes	11:34
PupenoG	jaeckel: I've already read the man page, couldn't find any argument it'll make it save only the added files.	11:34
jaeckel	PupenoG: look @ git stash --keep-index	11:35
	you take the first file in the index	11:36
PupenoG	jaeckel: I don't have that option.	11:36
wereHamster	jstein: two things you can do: look at the reflog and then reset --hard, that will remove the files from the history but they will remail for ~30 days in the .git directory (due to being referenced by the reflog). This is only recommended if you haven't already pushed the new commits to a public repo	11:36
jaeckel	humm	11:36
	you're using an old version?	11:37
	it's an option to git stash save	11:37
PupenoG	1.5.6.3	11:37
jstein	wereHamster, its a stand alone local repo.	11:37
wereHamster	jstein: reset --hard will reset _all_ files in your working tree, so if you have any additional changes you'll have to stash them	11:38
	← angerman left	11:39
jstein	wereHamster, well, i dont want to loose my changelogs of existing files. i fear i should work with two repos. One for working, one for storing, right?	11:40
jaeckel	hmm	11:40
	← bobmcw_ left	11:40
jaeckel	the --keep-index option was first mentioned with 1.5.6.1.95.ge713	11:41
	→ bobmcw joined	11:41
wereHamster	jstein: 'changelog of existing files'?	11:41
	← Leemp2 left	11:42
v0lksman	anyone have any good docs for using Meld for conflict resolution? trying to learn...Any tips, tricks and docs appreciated....	11:42
nisse	Okay.	11:43
	Next problem.	11:43
	It seems that I'm buried it conflicts with git-svn	11:43
jaeckel	sorry, but i don't have another solution	11:45
	how about updating? :)	11:45
jstein	wereHamster, i am working on the source tree and mage a git commit -a after each edit. But one day i hat wrong files in my source, added them. Them deleted the files. Now i want that git only knows about files, wich are currently in my source dir. All history about deleted files should be wipedout	11:45
jaeckel	i think this won't work, since it's a version control system that wants to track changes like this ;)	11:46
wereHamster	jstein: how far in the past is the 'bad' commit?	11:46
jstein	wereHamster, i think 3 commits in the past.	11:48
wereHamster	jstein: does that 'bad' commit contains anything else then the files you want to have removed?	11:49
jstein	wereHamster, some small changes on good files. But it may get lost, if necessary	11:50
wereHamster	jstein: git rebase -i $shaofthebadcommit^; then select 'edit' in the first line, git rm the files; git commit --amend; git rebase --continue	11:51
	→ eevar2 joined	11:53
	← RodP2 left	11:54
eevar2	how do I list all commits to a given remote repo? and which branch they're on?	11:54
	→ glommer joined	11:54
wereHamster	eevar2: all commits in a given repo?	11:55
eevar2	i guess	11:55
	do i have to clone it first, or?	11:56
wereHamster	eevar2: you need to clone it, then you can list all commits with git rev-list	11:56
eevar2	ok, thanks	11:56
	→ peritus_ joined	11:57
	→ bobmcw_ joined	12:01
	← bobmcw left	12:01
	← GeertB left	12:03
	sfaci_ → csgeek	12:05
	← ignas left	12:06
jstein	wereHamster, thank you, i will try that now.	12:07
	← khmarbaise left	12:09
	→ parasti joined	12:09
	← Radar left	12:09
	← hummingkid left	12:10
	← hyperair left	12:10
eevar2	hmm.. can't figure out how to coax rev-list / log into showing branches. does that mean every commit is done to 'master'?	12:12
	or my currently active branch	12:12
	or whatever	12:12
	guess i should sit down and read up on how push/pull/fetch and everything remote actually works ;)	12:13
ndim	eevar2: If you need a graphical overview of what is in your repo, run "gitk --all".	12:14
	I am not a GUI person myself, but I do think graphically, so that is very helpful to me for getting an overview of branches and commits.	12:14
	→ khmarbaise joined	12:16
	→ [1]intripoon joined	12:17
	← intripoon left	12:17
	[1]intripoon → intripoon	12:17
	→ saLOUt joined	12:19
saLOUt	i changed my .git/info/exclude file and also the one on my git-server managed via gitosis. what can i do to let people using git clone have the same exclude file?	12:20
	→ eMBee joined	12:20
	→ dgfitch joined	12:20
wereHamster	saLOUt: create a .gitignore file and commit it	12:22
saLOUt	wereHamster: in wich folder? parent of .git?	12:23
wereHamster	root folder of the working tree or any subfolder	12:23
	← nisse left	12:23
	→ quoin joined	12:23
	→ argonel joined	12:24
	→ cannonball joined	12:27
	→ UUncia joined	12:28
	bobmcw_ → bobmcw	12:28
UUncia	If I'd like to store my repository on a remote computer, is there a need for a git server or something?	12:29
bob2	no	12:29
	→ RaceCondition joined	12:31
RaceCondition	how do I set up the origin remote so that when I push to origin, I push to it's myname branch instead of master?	12:31
	so that each developer can have a branch by their name in the central repo	12:32
	without having to specify branch name each time when pushing	12:32
	→ gh34 joined	12:33
	← _Vi left	12:33
	→ _Vi joined	12:33
wereHamster	RaceCondition: set a push refspec in the cofig file, something like push = master:refs/heads/yourname	12:34
	← wsc left	12:34
RaceCondition	wereHamster: what about simply push = master:erik?	12:34
	is that somehow different to the example you gave?	12:34
wereHamster	I think that will work, too	12:34
	→ aspotashev joined	12:35
RaceCondition	nice	12:35
saLOUt	if i want exclude alle hiden files in this and child directories. Would be .* in .gitignore correct?	12:36
un\|ogic	saLOUt: yes	12:36
saLOUt	un\|ogic: thx	12:36
	← mithro left	12:36
	→ lorandi joined	12:37
	→ paltman94 joined	12:37
	← quoin left	12:38
	← peritus_ left	12:38
	→ jrmuizel joined	12:39
	← wmoxam_ left	12:40
	← parasti left	12:40
	→ cilly joined	12:42
	→ struberg1 joined	12:43
	← struberg1 left	12:44
	← _Vi left	12:44
	→ tobiassjosten joined	12:45
	→ _Vi joined	12:46
tobiassjosten	Is there a single command I can run to both do "git reset --hard" and "git pull", or something similar in effect?	12:47
	Basically, I want to SSH to a remote machine and have it replace its workdir with the latest from origin.	12:47
wereHamster	sh -c 'git reset --hard; git pull' <- there, a single command	12:49
	← knittl left	12:49
	→ knittl joined	12:49
	← lorandi left	12:49
tobiassjosten	wereHamster: Heh, brilliant. :)	12:49
	wereHamster: Would this work? sh -c 'cd /git/dir; git reset --hard; git pull'	12:52
	→ seezer joined	12:52
tobiassjosten	Or would the cd be lost after the semi colon?	12:52
wereHamster	that should work	12:52
RaceCondition	is it possible to have git checkout a svn repo as an external repo	12:52
	?	12:52
tobiassjosten	wereHamster: Awesome. Thanks a lot!	12:53
wereHamster	RaceCondition: not directly, you'll have to import the svn repo into git and add it as submodule	12:53
tobiassjosten	Hooks are repository specific, right? They are not synced with pull/push?	12:55
	← Simon- left	12:55
	→ parasti joined	12:55
wereHamster	tobiassjosten: correct	12:56
	→ Simon- joined	12:56
	← struberg left	12:56
	→ mithro joined	12:57
	→ ameyp joined	12:58
	← ameyp left	12:59
	→ tnuic joined	13:00
	→ solydzajs joined	13:00
	← bnferguson left	13:00
	← tnuic left	13:00
	→ qrush joined	13:02
	← jrmuizel left	13:03
	← naeu left	13:04
	→ nighthwk1 joined	13:04
	← lohrun left	13:05
	→ lohrun joined	13:06
Ilari	UUncia: Note that if you want ssh push to work, you need to install git on remote side. If 'ssh user@host git --version' works, you should be good to go.	13:08
UUncia	ok	13:08
	thx	13:08
	what about git://?	13:09
Ilari	UUncia: That needs git daemon to be running.	13:09
UUncia	on the server side, right?	13:09
Ilari	UUncia: Yes.	13:09
UUncia	ok	13:09
	thx	13:09
	→ mobhawk joined	13:09
Ilari	UUncia: file:/// doesn't need any kind of server (git internally invokes required stuff).	13:10
UUncia	i c	13:10
Rhonda	Ilari: Hmm, shouldn't a local git push to a bare + rsync be able to do the trick?	13:10
Ilari	Rhonda: rsyncing live git repos (even bare ones) isn't recomended.	13:11
	→ bschindler joined	13:11
	← __iron left	13:12
Ilari	Rhonda: Like, object database needs to be updated before refs, but with rsync there's no guarantee about that.	13:12
Rhonda	two-step rsync could do it?	13:12
	One with specific exclude patterns, the second with all.	13:13
Ilari	Rhonda: If you use rsync, better to be only process that updates the target repo.	13:13
Rhonda	Sure	13:13
Ilari	Rhonda: As rsync probably won't handle races gracefully.	13:13
	Rhonda: But objects and refs(+packed-refs) are pretty much all that is needed. Plus	13:14
	→ drewr joined	13:14
	→ brennen joined	13:15
hgb	Hi, guys. I'm working with git-svn with a very busy subversion repo. I've been working on my git branch for a while now, and have just commited and 'git svn dcommit'ed, so I'm done with the current (git) branch.	13:15
	I want to return to master, eventually to create new git branches.	13:16
	But, if I just do 'git checkout master', thousands of files will be changed, only to be restored when I do 'git svn rebase' (in master)	13:16
	I'd like to do 'git branch -D master; git branch master; git checkout master' instead.	13:16
	Are there any problems related to deleting master that way?	13:17
Ilari	hgb: 'git push -f . HEAD:master' (overwrite master with contents of current branch)?	13:17
hgb	Ilari: That sounds like an excellent idea!	13:17
Ilari	hgb: Sightly shorter but equivalent: 'git push . +HEAD:master'.	13:18
	hgb: That doesn't switch branches (so you need checkout master after that)...	13:18
hgb	Ilari: right. Sounds like exactly what I need.	13:19
	Hm	13:19
Ilari	hgb: That '.' there means "Target is this repo"	13:19
	→ ameyp joined	13:19
	→ quoin joined	13:20
hgb	Ilari: Thanks. Worked like a charm.	13:22
	→ Jinzhu joined	13:22
	hgb notes this in his git quickref.	13:22
	→ d0k joined	13:23
	→ GeertB joined	13:23
	→ catalis joined	13:24
	→ cmarcelo joined	13:24
	← Tuomas left	13:24
	→ Tuomas joined	13:24
	← brennen left	13:25
	→ jrmuizel joined	13:26
	→ Stravy joined	13:26
	← ameyp left	13:26
	← _Vi left	13:27
	→ _Vi joined	13:27
	→ LiamH joined	13:29
	→ voker57 joined	13:30
	→ gwoo joined	13:30
	← mikl left	13:32
	→ bnferguson joined	13:32
	→ joshsdc joined	13:33
	← bnferguson left	13:35
hgb	Ilari: If I drop the +, it would only happen if it could use fast forward?	13:36
	← voker57__ left	13:36
	→ keystr0k joined	13:37
	← sunoano left	13:37
keystr0k	I'm going a 'git add .' but nothing is being added... the ignore file isn't ignoring anything but .DS_Store files, and the folders aren't empty. ???	13:37
	I've used git for quite sometime and never seen this happen.	13:37
	← ceyusa left	13:39
wereHamster	keystr0k: what does git-status say?	13:39
keystr0k	wereHamster: it shows a lot of untracked folders.	13:39
	→ Bass10 joined	13:40
Ilari	hgb: Yes.	13:40
wereHamster	does git-add work if you explicitely mention the files/folders (instead of a simple .)?	13:40
keystr0k	wereHamster: nope... hmmm	13:40
	wereHamster: tried this: git add sites/all/themes/* and this: git add sites/all/themes/	13:41
Lenary	guys	13:41
	← paltman94 left	13:41
Lenary	i just released Ginatra - a gitweb clone in Sinatra and Grit - http://github.com/lenary/ginatra	13:42
	i'd be grateful if anyone would take a look	13:42
	→ ignas joined	13:42
Ilari	keystr0k: Are there any regular files not covered by ignore rules in those directories?	13:42
keystr0k	wereHamster: it isn't imperitive that I get this fixed. In fact, I only have 4 or 5 commits in the history of the repo... I can just start again from scratch. I moved from a Windows VM to a Mac...maybe that is the problem?	13:42
	Ilari: yes... lots of them.	13:42
	→ fanf2 joined	13:42
thiago	for those interested: the Git repository of Qt is now available on qt.gitorious.org	13:45
Ilari	keystr0k: Under those untracked directories?	13:45
thiago	[/shameless plug]	13:45
keystr0k	Ilari: definitely, yeah... lots of php files. I am only ignoring .ds_store	13:45
	→ sunoano joined	13:46
Ilari	keystr0k: Try finding untracked directory from status output, then what 'git add <regular-file-under-untracked-dir> ; echo $?' prints?	13:47
	← jrmuizel left	13:47
	← rjc left	13:48
keystr0k	Ilari: I apologize, but I've wiped the repo and started fresh. I only had a few commits, and they were not of any importance to me. I will add what you've suggested to my git notes. :)	13:48
	→ bwalk joined	13:48
	→ spearce joined	13:48
Ilari	keystr0k: That sort of brute-force solutions aren't often taken kindly...	13:50
keystr0k	Ilari: I understand.	13:50
Ilari	keystr0k: Especially in middle of troubleshooting.	13:50
	→ abstrusething joined	13:51
	→ amoeba joined	13:51
keystr0k	Ilari: again, I understand and apologize. I do appreciate the help.	13:51
	→ ProperNoun joined	13:51
	← psankar left	13:52
	← keystr0k left	13:55
	← knittl left	13:55
	→ knittl joined	13:56
	← shruggar left	13:56
	→ aspotashev\|eeepc joined	13:56
	→ ceyusa joined	13:58
	← mobhawk left	13:59
	→ shruggar joined	13:59
	→ mobhawk joined	13:59
	→ bryanray joined	14:02
	→ legnaleurc joined	14:06
	← Jinzhu left	14:08
	← girishr left	14:08
	→ dfd joined	14:08
	→ girishr joined	14:08
	→ wmoxam joined	14:09
	→ javatexan joined	14:10
dfd	hello, can anyone give me a suggestion as to how to save a completely different codebase under the same project namespace? That is, how to best wipe all files in a git repo and replace with a completely different set of files?	14:10
	← girishr left	14:10
	→ bdimcheff joined	14:11
Ilari	dfd: Perhaps easiest is to create another repo, make at least one commit there and then fetch that commmit in?	14:12
bmalee	dfd: why not just create a new repository?	14:12
	→ parth-vader joined	14:12
	← abstrusething left	14:13
	← bx2 left	14:13
dfd	Ilari: thanks, I'll look into that now. bmalee: well, we're basically resurrecting an old project with a new codebase, and want to keep the codebase under the same namespace. So, I'm tagging the last old commit and then starting from there	14:13
	→ krh joined	14:14
dfd	I know it's a bit unorthodox, but makes the most sense for us organizationally	14:14
	← amoeba left	14:16
	→ fargiola` joined	14:18
	→ jackdempsey joined	14:19
	→ bdiego joined	14:19
	← solydzajs left	14:20
	→ jrmuizel joined	14:20
	← javatexan left	14:20
	→ alezandro joined	14:21
	← aspotashev\|eeepc left	14:21
	← Lenary left	14:22
	→ paulboone joined	14:22
	→ cemerick joined	14:23
	→ legnaleurc_ joined	14:24
	→ javatexan joined	14:24
	→ agile joined	14:25
	→ nighthwk\| joined	14:26
	← javatexan left	14:26
	→ jerojasro joined	14:27
	← bobmcw left	14:27
	→ Sho_ joined	14:27
	→ bobmcw joined	14:28
jerojasro	do anyone know how to enable rss feeds on gitweb? I can't find any info about that	14:28
	← nighthwk\| left	14:28
Ilari	jerojasro: Isn't some optional feature like snapshots?	14:29
	→ nighthwk\| joined	14:29
	→ name joined	14:29
jerojasro	Ilari: not according to the wiki page for it http://git.or.cz/gitwiki/Gitweb	14:29
crankharder	hwo can I see a list of files change between now and X days ago?	14:31
bmalee	crankharder: git log @{X.days.ago}..	14:33
	→ DavidKlein joined	14:34
	→ rjc joined	14:34
crankharder	that is showing me the entire log up until X days ago	14:35
	← fargiolas left	14:35
jackdempsey	there's a --name-only flag in git log	14:35
	← ph^ left	14:35
	← paulboone left	14:35
	→ netoman joined	14:35
jackdempsey	look through there a bit and play with the options	14:35
	bmalee: is there a good place to read up on using @, i haven't found one	14:35
	hard to google it :-)	14:35
RandalSchwartz	have you tried the fine documentation?	14:36
	← cilly left	14:36
RandalSchwartz	it comes with git, you know.	14:36
bmalee	jackdempsey: man git-rev-parse	14:36
Gitbot	jackdempsey: the git-rev-parse manpage can be found at http://git.or.cz/man/git-rev-parse	14:36
	→ j416 joined	14:36
jackdempsey	ah nice, thanks!	14:36
crankharder	and --name-only is adding the list of chagned files, not removing the commit messages/author/date etc	14:36
jackdempsey	yep, i noticed that as well...there should be a way to take those things away	14:37
alezandro	hi all, do I have to clone a new git tree when going from kernel 2.6.28.y stable to 2.6.29.y stable ? i'd like to keep just one kernel source tree in /usr/src....	14:37
xblaster	Hello	14:37
	I have a little question	14:38
	Some dirty developers has commit some crap in a branch	14:38
	fargiola` → fargiolas	14:38
xblaster	like binary of around 200MB	14:38
	Can I "proper" this branch this just merging old commits with just one commit	14:38
Ilari	alezandro: 'git remote add -f 2.6.29.y <URL>'.	14:39
	alezandro: Should be few megs of transfer.	14:39
j416	xblaster: as long as there is anything at all referencing that 200MB object, it will remain.	14:39
xblaster	j416: I know that	14:39
	and if I remove it, anyway git will keep a trace of it	14:40
alezandro	Ilari: cool, ok. i'll try it	14:40
xblaster	it was I ask if I can do a sort of "squash"	14:40
Ilari	xblaster: git rebase -i?	14:40
xblaster	Ilari: maybe this could be a good idea :)	14:40
j416	xblaster: if you squash in such a way that the reference disappears, sure	14:40
	← bschindler left	14:41
Ilari	alezandro: You can have multiple remotes in single repository.	14:41
wereHamster	alezandro: clone git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-2.6-stable.git and you'll never have to worry about adding the proper remotes anymore ;)	14:41
alezandro	wereHamster: what about when linux-2.8 comes out?	14:42
wereHamster	never have to worry in this lifetime	14:42
alezandro	LOL, ok	14:42
	→ njero joined	14:42
	← legnaleurc left	14:43
alezandro	and i'm assuming pub/scm/linux/kernel/git/stable/linux-2.6-stable.git will have all the tags for every minor version, like 2.6.28.6, 2.6.28.9, etc....	14:43
	right?	14:43
wereHamster	alezandro: yes, that's correct	14:43
alezandro	cool, that's exactly what I need...	14:44
	← UUncia left	14:44
	→ cilly joined	14:44
	← Sho_ left	14:45
	← nighthwk\| left	14:45
	→ Sho_ joined	14:46
	→ DMG_ joined	14:46
	→ quentusrex joined	14:47
quentusrex	I have a checkout of a remote repo. How do I dump all local changes and update to exactly the remote repo?	14:47
	For some reason I keep getting told that there are local changes to the git repo, when I have only been using it as a read only reference	14:48
	← DMG_ left	14:48
quentusrex	I have done a git reset --hard	14:48
	and did a git pull	14:48
Ilari	quentusrex: What OS?	14:48
quentusrex	Ubuntu	14:48
Ilari	quentusrex: 'git reset --hard origin/master'?	14:49
quentusrex	error: Untracked working tree file 'src/Node.cpp' would be overwritten by merge	14:49
	→ DMG_ joined	14:49
	← DMG_ left	14:49
Ilari	quentusrex: That is some new file that has gotten created.	14:49
	→ rubydiamond joined	14:49
quentusrex	ok, awesome	14:49
	that was it	14:49
	just needed to add the origin/master on there.	14:50
	→ nighthwk\| joined	14:50
quentusrex	thanks.	14:50
	→ legnaleurc__ joined	14:51
	← nighthwk\| left	14:51
	→ nighthwk\| joined	14:51
	← mobhawk left	14:52
	→ schacon joined	14:53
	→ mobhawk joined	14:53
	→ incorrect joined	14:55
	← neoeinstein left	14:57
	← eevar2 left	14:57
	← drewr left	14:57
	→ releod joined	15:00
	→ s0ber_ joined	15:04
	→ jelly-bean joined	15:04
jelly-bean	somehow i have an empty directory in my git repo. how do i remove it? git rm says: "fatal: pathspec '.gitk-tmp.27468/' did not match any files"	15:05
	← cilly left	15:05
Grum	is it tracked? what does git status say?	15:06
	← njero left	15:06
jelly-bean	nothing to commit (working directory clean)	15:06
Grum	git cannot track empty dirs, so it will prolly be listed under 'untracked', so just rm -rf if you want to get rid of it (or git clean -d)	15:06
	if its not under untracked it means that it is in a gitignore somewhere :)	15:06
	→ fridim joined	15:06
	← legnaleurc_ left	15:07
Grum	or it might be considered 'special' and 'always ignored' (like .git folders)	15:07
jelly-bean	it must be special because it's not in any .gitignore	15:07
	i think you're right that it's untracked though	15:07
	← parth-vader left	15:07
jelly-bean	i just rm -rdf'ed it	15:07
	seems to be gone. thnx :)	15:08
	→ paulboone joined	15:08
	← Ademan left	15:09
	← jstein left	15:10
	← nighthwk\| left	15:11
	→ nighthwk\| joined	15:11
	→ jsbmsu joined	15:13
	← Stravy left	15:13
	← schacon left	15:13
	← s0ber left	15:15
	→ Hendu joined	15:16
tobiassjosten	I am working in a team against a bare git hub. We want to tag our current HEAD but doing so locally doesn't seem to transfer. Am I misinterpeting the usage of tags in Git?	15:16
	← grahal left	15:17
Ilari	tobiassjosten: 'git push <repo> <tag>'.	15:17
	→ juanjoc joined	15:17
	← jelly-bean left	15:18
tobiassjosten	Ilari: That seemed to work. Thanks!	15:18
	← knittl left	15:20
	→ angerman joined	15:22
	→ cilly joined	15:25
	→ bnferguson joined	15:26
	← jsbmsu left	15:27
Ilari	tobiassjosten: Regarding git tags: Git tags are not on any branch (they are in effect special type of branch themselves) and tag creation completes immediately without needing commit.	15:28
	← nighthwk\| left	15:28
	→ grahal joined	15:28
	← scientes left	15:29
	→ neoeinstein joined	15:31
	→ tclineks joined	15:33
	→ schacon joined	15:34
	→ drewr joined	15:36
	← name left	15:40
	→ imarcusthis joined	15:43
	← j416 left	15:44
	→ pity16_ joined	15:44
	← quoin left	15:45
	← pity16_ left	15:45
	→ mw_ joined	15:45
	→ RayMorgan joined	15:47
	→ souldier joined	15:50
	← jerojasro left	15:50
souldier	hey	15:51
	for some reason i keep getting ssh: connect to host port 22: bad file number fatal: the remote end hung up unexpectedly	15:51
	I can ssh into my server an all	15:51
	→ radarek joined	15:52
	→ hugoxrosa joined	15:53
	← bentob0x left	15:53
Ilari	souldier: Clone, fetch, pull or push?	15:54
	← marc_m left	15:54
	→ marvil07 joined	15:54
souldier	push	15:55
Ilari	souldier: Does 'ssh user@host git-receive-pack foo' print anything from remote?	15:56
	souldier: That's how git push invokes ssh...	15:56
	← giallu left	15:57
	→ elmob joined	15:58
	← bdowning left	15:58
	← _Vi left	15:58
souldier	ok i did that and instead of foo wrote where mytest.git is and i got something like capabilities^{} report-status delete-refs etc	15:59
	→ bdowning joined	15:59
	← Modius left	16:00
	→ hyperair joined	16:00
tobiassjosten	Ilari: One more tag question, if you don't mind. :) If I delete the tag (git tag -d tag), then how do I push that change upwards?	16:00
Ilari	souldier: That looks sane. Wonder why that 'git push' won't work...	16:00
	→ Modius joined	16:00
Ilari	tobiassjosten: 'git push <repo> :<tag>' (the same kind of stuff as branch delete).	16:01
souldier	Ilari: how am i supposed to do the push correctly?	16:01
	→ StyleWarz joined	16:01
StyleWarz	heyho	16:01
Ilari	souldier: Was that SSH error transistent or persistent?	16:02
souldier	well maybe its that im doing somethign wrong in the process	16:02
	can you help me do a test?	16:03
	maybe throuh PM or something	16:03
	← Bogh left	16:03
Ilari	souldier: Maybe do 'strace -f git push <other args> 2>strace.log' to get strace log of what its doing?	16:03
	← RaceCondition left	16:03
	← Hendu left	16:04
	→ kukks joined	16:04
StyleWarz	can anyone tell me how i can change my git repo to a specific tag?	16:05
	mw_ → mw	16:05
	← tango_ left	16:05
StyleWarz	like i need my repo to be somewhat other than head	16:05
	→ knittl joined	16:05
souldier	Ilari: what should be my syntax for git push	16:05
voker57	git checkout mytag	16:05
	→ Korren joined	16:06
StyleWarz	voker57: aaah ty	16:06
Ilari	souldier: The usual syntaxes are 'git push', 'git push <remote>' or 'git push <remote> <branches>...'.	16:06
StyleWarz	voker57: do i need to do anything else to update the files in the repo to be the "older" version?	16:06
Ilari	souldier: And whatever you specify, it shouldn't cause SSH to fail connect() with EBADF.	16:06
	← cilly left	16:07
	← galderz left	16:07
souldier	i was doing something like git push ssh://[user@]host.com/home/user/mytest.git master	16:07
Ilari	souldier: If you are pushing often there, you should create a remote for it (git remote add mytest <URL>), then you can refer to that repo just as 'mytest'.	16:08
souldier	but if that syntax is correct i should have problem should i?	16:09
	→ jchris joined	16:09
souldier	maybe i did something wrong while doing the add and commit shit	16:09
	→ SteveC_ joined	16:11
Ilari	souldier: It seems to be some sort of problem with SSH client.	16:11
souldier	hm maybe, im using my msisgit in windows	16:11
	mSysGit or some shit like that	16:12
	← ignas left	16:12
Ilari	souldier: Ergh... Msysgit doesn't have strace...	16:12
souldier	yea, i know, any other option?	16:12
SteveC_	Hey guys, I am trying to use git svn to keep my local git repository synced with my work's svn repository. I want to compact all of my git commits into 1 svn commit. I read online I can run 'git rebase -i git-svn' but i get the error "needed a single revision, invalid base"	16:13
Ilari	souldier: If you retry the failed operation, does it fail again?	16:13
souldier	yea	16:14
SteveC_	I am on Mac OS if that makes a difference w/Git 1.6.2.2	16:14
Ilari	souldier: Maybe file a bugreport to msysgit bug tracker?	16:14
souldier	is there another git client thing for windows?	16:14
Ilari	souldier: Or if you don't have newest version, try upgrading first.	16:14
	← k0001 left	16:15
souldier	i have 1.6.3	16:15
Ilari	souldier: Git also runs on top of Cygwin.	16:15
	← Kbyte left	16:15
StyleWarz	i did git checkout <mytag> but the contents of my actual repository are still the "newest" instead of the ones from the "old tag".. did i forget anything?	16:17
Ilari	StyleWarz: Did it even succeed?	16:17
	souldier: Cygwin doesn't fix the slowness of Git on Windows...	16:18
StyleWarz	Ilari: there was no error	16:18
Ilari	souldier: In fact, it might be even slower than msysgit.	16:18
StyleWarz	jruby /data/do # git checkout 0.9.11	16:18
	HEAD is now at c7fa1c3... Update LICENSE files and change maintainer in Rakefile's	16:18
	← souldier left	16:19
Ilari	StyleWarz: It checked out version c7a1c3.... Git log should show when the latest commit was made.	16:19
StyleWarz	Ilari: 19th january.. but the content of all files is still new	16:20
	← tobiassjosten left	16:20
Ilari	StyleWarz: 'git status' should show which files have uncommitted modifications.	16:20
StyleWarz	nothing to commit (working directory clean)	16:21
	← ciskje left	16:21
	← gitte left	16:22
	← masak left	16:22
Ilari	StyleWarz: Then all files should be in states that they where in that tagged commit...	16:22
StyleWarz	very strange	16:22
	but thanks	16:23
	← mtkd left	16:23
	← tal67 left	16:23
	→ bizhat2 joined	16:24
	→ carrus85 joined	16:25
	→ sgrimm joined	16:25
	← incorrect left	16:29
	← shruggar left	16:30
	→ lorandi joined	16:30
	→ bonch joined	16:31
	← bizhat2 left	16:31
	→ mitsuhiko joined	16:32
mitsuhiko	hi everybody	16:32
	what would be the easiest way to delete all remote branches except master?	16:32
	← bonch left	16:32
	→ willb joined	16:32
	← jackdempsey left	16:33
	← neoeinstein left	16:34
	→ neoeinstein joined	16:35
Ilari	mitsuhiko: Delete from remote repo?	16:35
mitsuhiko	yep	16:35
	i could of course do "git push origin :branch" for every single branch, but there are liek 15	16:36
	← SteveC_ left	16:36
Ilari	mitsuhiko: loop over what 'git ls-remote origin' returns?	16:37
mitsuhiko	oki. i'll do that :)	16:38
	→ giallu joined	16:39
	← johan-s left	16:40
	→ nighthwk\| joined	16:42
	→ gberdyshev joined	16:42
	→ qhoxie joined	16:43
	→ k0001 joined	16:44
	← StyleWarz left	16:45
	→ jceb joined	16:45
	→ ceej joined	16:45
	→ Ademan joined	16:46
	→ tango_ joined	16:48
	→ graywh joined	16:48
	→ altrux joined	16:49
graywh	is there a special filename that can be used to compose a commit message before doing the commit?	16:49
	→ mtkd joined	16:52
	→ UUncia joined	16:52
	← UUncia left	16:52
	→ UUncia joined	16:52
	→ bx2 joined	16:52
UUncia	Hi	16:52
	← grahal left	16:53
graywh	i guess -F is good enough	16:53
UUncia	I'd like to setup a git "server" on a windows machine. I installed the git installer. Do I need to setup a SSH server as well?	16:53
un\|ogic	graywh: or -	16:53
	→ Dashkal joined	16:53
graywh	un\|ogic, - is only valid as the filename for -F, isn't it?	16:54
un\|ogic	it should take it from stdin	16:55
	so you should be aple to cat commitmsgfile \| git commit -F -	16:55
graywh	but that's longer than git commit -F commitmsgfile	16:55
un\|ogic	true :)	16:56
	handy though if you need to do something to the file before passing it to git commit	16:56
	→ kuadrosx joined	16:56
	← khmarbaise left	16:57
	→ amoeba joined	16:57
	→ khmarbaise joined	16:58
	← jaeckel left	16:59
	← khmarbaise left	16:59
	→ kraymer joined	17:00
	→ __iron joined	17:01
	← Ademan left	17:01
	→ rlorandi joined	17:02
Ilari	UUncia: ssh push would require ssh server... IIRC, there is Java SSHD that can be used in combination with Jgit, but its not as performant as C Git (but if you cared about performance, you wouldn't be using Windows server).	17:02
	← lorandi left	17:03
	→ ilogger2 joined	17:04
	→ priidu joined	17:07
	→ parth-vader joined	17:08
	← parth-vader left	17:09
	→ Ademan joined	17:13
	→ Alinermedia joined	17:13
	→ jrmuizel joined	17:14
	→ malkomalko joined	17:15
	→ UukGoblin joined	17:15
UukGoblin	hi	17:16
	→ jackdempsey joined	17:17
	→ jtong joined	17:18
Ilari	UukGoblin: Hi.	17:18
UukGoblin	is there a place I can read about git's deep internals? Like, at the moment, I'm quite interested in why does it know what blobs to diff when doing a repack	17:19
	does it try to diff a file with every other blob in repository?	17:20
	→ xenoterracide joined	17:20
UukGoblin	it's quite interesting -- if I add a big fat file (45M one) on one branch, and then go to another branch a few commits back and add a very similar yet slightly different big file (45M), after git-gc, the repository will still only take about 45M...	17:20
	→ peritus_ joined	17:21
killerchicken_	Does anyone have experience with tracking some tex documents including graphics with git, and if yes, does it work well? The graphics are in the range of max 500kb, and I don't want to store the resulting pdf files - only source material.	17:22
Fissure	look in the Documentation/technical directory	17:22
	killerchicken_: yeah, i track tex all the time...	17:22
UukGoblin	Fissure, thanks, will have a look	17:23
killerchicken_	Fissure: do you have many graphics?	17:23
Ilari	killerchicken_: Maybe disable deltification of those graphics files? Should somewhat increase performance.	17:24
killerchicken_	Is that something I can do should I start noticing performance issues, or do I need to do it right away?	17:25
Ilari	killerchicken_: You can do either way.	17:25
UukGoblin	hahah	17:25
	OK, where do I find IRC logs from 10 feb 2006? :-D	17:25
Ilari	killerchicken_: As its storage layer thing, not object layer thing.	17:26
killerchicken_	good. I haven't started with that, but since I have quite image-heavy documents, I was afraid git might not be the perfect choice.	17:26
UukGoblin	ok, some of it seems to be pasted below ;-]	17:26
Ilari	UukGoblin: It tries to delta against files in window around the target file. The list is sorted by hash value using intentionally bad hash that biases last bytes in file name.	17:26
Fissure	killerchicken_: you won't notice unless you try a gc --aggressive with your aggressivewindow set really high	17:27
	← malkomalko left	17:27
killerchicken_	Fissure: cool	17:27
Ilari	Don't use 'gc --aggressive'.	17:27
	→ rlorandi joined	17:27
killerchicken_	Generally not, or not in my use case?	17:27
UukGoblin	Ilari, interesting :-) is it possible to give it hints?	17:27
Ilari	Generally not.	17:27
Fissure	Ilari: oh? why?	17:28
Ilari	UukGoblin: Well, git-pack-objects can accept hints.	17:28
	→ Lenary joined	17:28
	→ rylar joined	17:28
Ilari	Fissure: For those cases where one would want --aggressive, better to use repack.	17:28
	→ drizzd joined	17:28
UukGoblin	cool	17:28
Fissure	more fine-tuned control?	17:29
Ilari	Fissure: Yes. And the values --aggressive uses might not be sane.	17:29
UukGoblin	that's really awesome, /me looks around more ;-)	17:29
Ilari	UukGoblin: Normally it uses one pathname of file as hint.	17:30
	UukGoblin: And if you change that, you might break the delta attribute logic.	17:31
UukGoblin	Ilari, I did my trick-test with totally differently named files	17:31
	→ naeu joined	17:32
Ilari	UukGoblin: There's still window, so it may take differently named files.	17:32
UukGoblin	Ilari, well, basically, I'm planning on doing something nasty... I (maybe, not sure yet) want to store files with the filenames of a sha-sum of their content in git... pretty randomly, perhaps even just blobs, not commits... so I'm wondering whether there exist ways to tell git where to look for differences between files :-) and apparently there are some.	17:32
	don't worry if I sound crap, that's probably because I have too many crazy ideas ;-)	17:33
	← Ademan left	17:33
Ilari	UukGoblin: Why would you want to name them by SHAsum?	17:33
UukGoblin	Ilari, well... OK, the project I'm thinking of is to create a webproxy that will store all the traffic it proxies	17:34
quentusrex	Is there a way to setup default git settings?	17:34
	→ dwave joined	17:34
Ilari	UukGoblin: Git is quite unsuitable for that.	17:34
UukGoblin	Ilari, and then update the store with updates to the actual sites	17:34
quentusrex	Or is it possible to have a git repo with the 'standardized' default settings file that can easily be 'imported' into a repo?	17:35
Ilari	quentusrex: 'git config --global' or 'git config --system'?	17:35
RandalSchwartz	it never ceases to amaze me about how people want to abuse git. :)	17:35
	→ cilly joined	17:35
quentusrex	so that I can have color settings and other stuff standardized?	17:35
UukGoblin	Ilari, well, it worked quite well when I used it like git init, wget -r, git add ., git commit, wget -r...	17:35
	→ sekimura joined	17:35
UukGoblin	hey, it's cool! ;-) I'm just exploring possibilities here ;-)	17:35
Ilari	quentusrex: The --global stuff is defaults by user account. --system is default per system. User defaults override system defaults and repo-specific settings override user defaults.	17:36
	→ rjc joined	17:36
UukGoblin	pack-heuristics.txt is actually quite funny, too! :-D	17:37
	→ dreiss joined	17:39
	← jackdempsey left	17:43
	→ kumbayo joined	17:44
	→ RodP2 joined	17:46
	→ texel joined	17:46
	← peritus_ left	17:50
	→ wsc joined	17:51
	→ doener joined	17:55
	→ justatheory joined	17:55
justatheory	Is there a way to make `git pull origin master` completely quiet unless there is an error?	17:56
	It seems to output status messages to STDERR which is annoying.	17:56
	→ j416 joined	17:59
	→ ascarter joined	17:59
charon	justatheory: there's a -q flag	18:01
j416	I have three figures that fit on one page. But when I add more text, they will split up so that one is printed on its own page. I want to keep the three figures on one page, how can I do this? I have tried putting the figures in a separate file and using \include to get a page break, but it instead rendered the two on a separate page, and still, the third is on the next page. \newpage, \clearpage, \pagebreak gives the same results. Any ideas?	18:02
Ilari	j416: EWRONGCHAN?	18:02
j416	Ilari: very wrong. sorry about that :)	18:03
justatheory	charon: Yeah, that doesn't effect the messages sent to STDERR, such as:	18:06
	"* branch master -> FETCH_HEAD"	18:06
	Annoying.	18:06
	→ nighthwk\| joined	18:06
	→ kennethlove joined	18:06
justatheory	Oh, wait	18:06
	I put it in the wrong place.	18:06
	Duh	18:06
	→ vmiklos joined	18:10
	→ bryanray joined	18:12
	→ robinr joined	18:12
	→ k0001 joined	18:13
	→ abstrusething joined	18:14
quentusrex	ok, I have a git repo, that does svn integration. I have git cloned it to my local machine.	18:14
	I made a commit on the (middle) repo,	18:14
	but when I did a git pull on my local machine, I don't get that commit	18:14
	any ideas?	18:14
	→ johan-s joined	18:15
	→ gitte joined	18:17
	← Alinermedia left	18:18
	→ Alinermedia joined	18:18
	→ bizhat2 joined	18:18
	→ schmrz joined	18:20
	← cilly left	18:20
	→ tbf joined	18:21
schmrz	Is there any way that git-clone gives more output. Even if the host doesn't exist Git doesn't complain.	18:21
gitte	schmrz: it complains here...	18:22
	schmrz: fatal: Unable to look up non-existing.com (port 9418) (Name or service not known)	18:22
schmrz	schmrz@lappy:~/Projects$ git clone git://12.13.14.15/repo	18:23
	Initialized empty Git repository in /home/schmrz/Projects/repo/.git/	18:23
	← nighthwk\| left	18:23
schmrz	And it just sits like that.	18:23
	*stays	18:23
gitte	schmrz: then it will eventually time out, I guess.	18:24
schmrz	Is it possible to set the timeout period?	18:24
gitte	schmrz: dunno.	18:24
schmrz	Well the manual doesn't say anything about it, so I guess you can't.	18:25
	→ patmaddo_ joined	18:28
	→ bobesponja joined	18:28
	← wsc left	18:29
	→ peritus_ joined	18:30
	→ wwwald joined	18:33
	→ wsc joined	18:34
quentusrex	How do I switch to a different branch?	18:34
	→ gotgenes joined	18:34
	→ benn_d joined	18:35
kusma	quentusrex: git checkout branchname	18:35
quentusrex	master?	18:35
	so "git checkout master" returns a did not match any file(s) known to git. error	18:36
benn_d	when git-push is to a ssh server, what would the 'initial command' be on the server side?	18:36
	→ altrux joined	18:36
	→ ameyp joined	18:37
kusma	quentusrex: then master doesn't exist, I guess.	18:37
ameyp	is there any way of taking a snapshot of a git repository and then updating it with git-pull?	18:37
quentusrex	I have checked out a branch from the remote repo that is no longer developed. So Now I need to change to another branch on that remote repo.	18:38
	the branch I'm on is "testing"	18:38
	← justatheory left	18:38
quentusrex	and now the branch is "master"	18:38
kusma	quentusrex: "git branch -r" will list all remote branches.	18:38
	now, you might want to setup a remote-tracking branch called "master", tracking "origin/master" (or whatever the remote is called)	18:39
quentusrex	how would I do that?	18:39
	→ krh joined	18:40
kusma	quentusrex: "git branch master origin/master"	18:40
schmrz	If I want to share Git repository over HTTP do I just copy my repo to /var/www ?	18:40
	← ameyp left	18:42
	→ paulboone_ joined	18:42
schmrz	In other words do I just copy it to my HTTP document root?	18:42
	and then git clone http://example.com/repo	18:42
quentusrex	kusma: thanks a ton for the help so far. How do I get "git pull" to pull the right branch now?	18:43
kusma	quentusrex: "git pull origin master"	18:43
quentusrex	how do I set that in the config for that to be the default?	18:44
	→ tatsuyao joined	18:44
	→ jpnurmi joined	18:44
kusma	something like, I guess: git config origin.fetch "master"	18:44
	I haven't tried it myself, though ;)	18:45
	→ psoo joined	18:45
schmrz	Anyone?	18:45
	→ lovingwoman123 joined	18:46
quentusrex	kusma that didn't work	18:46
kusma	oh, sorry then :)	18:47
quentusrex	same error wanting the repo and refspec	18:47
	but that seems close :)	18:47
	→ ameyp joined	18:47
ameyp	is there any way of updating a snapshot of a repo using git-pull?	18:47
	← lovingwoman123 left	18:48
quentusrex	How would I setup a repo to keep an 'unclean' repo updated? and on commit to the repo I have to manage any conflicts?	18:48
	I have a workstation and the server. The server has the repo, and the server has a checkout of the repo for live configs.	18:49
	→ codebrulee joined	18:49
quentusrex	I'd like to have something that would check to see if there are any uncommitted files that would be overwritten	18:50
	if there are then don't allow the commit.	18:50
	→ yann joined	18:50
	→ gilimanjaro joined	18:54
	← gilimanjaro left	18:54
quentusrex	Ilari: Do you know of a good commit hook that would check a working copy on the server to see if there are conflicts, from both committed and uncommitted files? then if there are then fail the commit.	18:55
	→ _graham_ joined	18:55
	abstrusething → abstrusething\|aw	18:57
Ilari	quentusrex: That mixes up stuff that should be kept seperate real good.	18:58
quentusrex	well, I just need something to 'alert' that there are uncommitted or committed files that conflict	19:00
	← j416 left	19:00
quentusrex	and that the uncommitted ones would overwritten or something...	19:00
	→ javatexan joined	19:01
SamB	the latter would be handled locally, the former would cause the checkout to fail ...	19:01
quentusrex	right,	19:01
	does it to that already?	19:01
SamB	unless you pass -f	19:01
quentusrex	awesome	19:02
SamB	so just make the script that runs the checkout send an email when it fails	19:02
	→ bcardarella joined	19:03
quentusrex	! [rejected] master -> master (non-fast forward)	19:03
	?	19:03
SamB	quentusrex: did you do a rebase /	19:04
quentusrex	?	19:04
	← ameyp left	19:04
SamB	quentusrex: the new master wasn't descended from the old one	19:04
quentusrex	hmm	19:04
	how do I fix that?	19:04
	I'm feeling very git ignorant right now...	19:05
SamB	well, I'd be more worried about how that happened	19:05
	what have you done lately?	19:05
quentusrex	well, the remote repo branch that I checked out was "testing"	19:05
	then it changed to master	19:05
SamB	oh, it's a remote repo?	19:05
quentusrex	yes	19:06
	the remote repo is the 'more important' repo	19:06
SamB	okay, then I guess you didn't cause it ;-)	19:06
quentusrex	which has a local check out for the working copy	19:06
v0lksman	if anyone can help me figure out process and merging I'd much appreciate it...been reading all day but can't figure this last bit out... http://dpaste.com/43078/ any tips welcome to the way I'm doing things.	19:07
	→ dkr21 joined	19:07
	← dwave left	19:07
Ilari	v0lksman: Git should have left conflict markers in that file.	19:08
quentusrex	SamB: how do I delete the whole local check out and start over?	19:08
	or would that not be needed?	19:08
SamB	quentusrex: shouldn't be ... but if you want to, you could just move the directory and try again	19:09
schmrz	Guys, please how do I setup a HTTP repository?	19:09
gitte	schmrz: you read the documentation ;-)	19:09
	→ henon joined	19:09
SamB	man git-update-server-info	19:09
Gitbot	SamB: the git-update-server-info manpage can be found at http://git.or.cz/man/git-update-server-info	19:09
Ilari	schmrz: Hopefully you don't have to set up HTTP push. :-)	19:10
gitte	schmrz: http://www.kernel.org/pub/software/scm/git/docs/howto/setup-git-server-over-http.txt	19:10
Ilari	quentusrex: Use git reset if you want to nuke changes?	19:10
schmrz	(in /var/www) mkdir repo; cd repo; git --bare init; git-update-server-info; Is that all? I have read the documentation but something is wrong...	19:10
	I don't need push	19:10
v0lksman	Ilari: yes it does...so I would manually update that chunk of code?	19:11
quentusrex	Ilari: would that fix : ! [rejected] master -> master (non-fast forward)	19:11
SamB	schmrz: what symptoms are you seeing?	19:11
Ilari	quentusrex: Maybe 'git checkout master', 'git pull origin master'?	19:11
schmrz	"warning: remote HEAD refers to nonexistent ref, unable to checkout." <--- This is what I get when I try to pull from that HTTP server.	19:12
Ilari	v0lksman: Edit the code to be as you want and then add the file.	19:12
	schmrz: You need commit there first.	19:12
	← rlorandi left	19:12
schmrz	Ilari: When I do that I get "fatal, this opetation must be run in a work tree"	19:12
	*operation	19:12
Ilari	schmrz: And the update-server-info must be run after ANY modification to that repo. So best to stick it into apporiate hook.	19:12
	schmrz: Push commit there.	19:12
	schmrz: IIRC, one of example hooks just needs enabling.	19:13
SamB	schmrz: bare repositories don't have working trees, no	19:13
	so don't try to commit into them directly ;-)	19:13
quentusrex	Ilari: I did the checkout, then tried to push	19:13
schmrz	SamB: That is what I'm trying to understand :\	19:13
quentusrex	and it failed with the same error.	19:13
Ilari	quentusrex: Pull first, then push.	19:14
schmrz	SamB: Where do I commit then?	19:14
Ilari	schmrz: Create local repo for it.	19:14
SamB	schmrz: in your working repository	19:14
v0lksman	Ilari: hrm...interesting...for some reason I was expecting more for some reason...	19:14
	sry...brain fried... :)	19:14
schmrz	Ilari: And then I push from local repo into that in /var/www ?	19:14
SamB	if you were using SVN, it would be called your "checkout"	19:14
quentusrex	Ilari: I just did a checkout, made a local change, did a git pull, then a git push, and it failed with the same error.	19:15
Ilari	schmrz: Yes, although you should create a remote to refer to that repo so you don't have to type long path every time.	19:15
	→ jefferai joined	19:16
schmrz	Ilari: Ok. Thank you. Just one more thing. How do I create a hook for update-server-info every time I push into that repo?	19:16
mitsuhiko	hmm. say i have a repo which does not merge properly with the upstream one where i want to pull from. can i somehow just "discard" all changes so that i can use the upstream version as the new head?	19:16
Ilari	quentusrex: Is that some HTTP remote?	19:16
SamB	it's something of a design flaw that git doesn't support empty repositories ...	19:16
quentusrex	it is over ssh for a remote repo, yes.	19:16
Ilari	quentusrex: Does 'git log master..origin/master' yield anything?	19:16
	mitsuhiko: Reset.	19:17
quentusrex	Ilari: it does have some lines, but nothing related to the error	19:17
Ilari	quentusrex: Lines there ARE related to the error.	19:17
quentusrex	ok... Then I don't understand why that is true.	19:17
mitsuhiko	Ilari: that does not work, i get an error after pulling then (not uptodate, cannot merge)	19:18
Ilari	quentusrex: master..origin/master is all commits that are in master of origin but not local master. Push would delete those so it complains.	19:18
quentusrex	ok	19:18
Ilari	mitsuhiko: It has --hard option, but that is destructive.	19:19
quentusrex	then why aren't they in the local? I did a git pull	19:19
	and a git reset --hard	19:19
mitsuhiko	Ilari: i don't have any local changes	19:19
Ilari	quentusrex: Reset???	19:19
	← bryanray left	19:19
schmrz	Ilari: Is it the post-commit hook?	19:19
mitsuhiko	basically i try to update http://github.com/mitsuhiko/django/tree/master with the latest changes from http://github.com/django/django/tree/master	19:20
	→ SjB joined	19:20
Ilari	schmrz: post-update or somesuch.	19:20
mitsuhiko	actually i would start over if people wouldn't have pulled from me already	19:20
	← tbf left	19:20
Ilari	mitsuhiko: 'git reset --hard foo/master', 'git pull bar master'?	19:21
	quentusrex: Note that if it conflicts, reset --hard will abort the merge.	19:21
	← SjB left	19:21
Ilari	quentusrex: I asked about HTTP because HTTP sometimes can suffer from false up-to-date conditions, but AFAIK, ssh:// doesn't.	19:23
mitsuhiko	Ilari: now git segfaults on me	19:23
	→ __iron joined	19:23
quentusrex	Ilari: I'm trying to push to a workstation from my workstation. How do I test what revision the remote repo is at?	19:24
	→ amoeba joined	19:24
	← jpnurmi left	19:24
Ilari	mitsuhiko: What git version?	19:24
mitsuhiko	1.6.1.2	19:25
Ilari	quentusrex: 'git ls-remote <remote>'.	19:25
	mitsuhiko: That's bit old. And regarding old versions, anyone with priviledges to fix /topic.	19:25
mitsuhiko	Ilari: i have the version that comes with cygwin	19:26
quentusrex	ok, so it's at a8e6f982618b6f04f2809aab34062e721143494c	19:26
	→ nighthwk\| joined	19:26
Ilari	quentusrex: Maybe try 'git merge a8e6f982618b6f04f2809aab34062e7'. What it outputs?	19:26
	mitsuhiko boots up his mac	19:27
quentusrex	Ilari: "already up to date"	19:27
	→ LotR joined	19:28
	← LotR left	19:29
Ilari	quentusrex: That was for refs/heads/master?	19:29
quentusrex	yup	19:29
	→ jceb joined	19:29
	→ LotR joined	19:29
	→ dwave joined	19:29
Ilari	quentusrex: What's the current branch on local side?	19:29
quentusrex	how do I check?	19:29
Ilari	quentusrex: 'git branch'.	19:29
quentusrex	hmm...	19:29
	no branch	19:29
	→ jhasse joined	19:30
Ilari	quentusrex: Install git bash completion and configure bash prompt. Would have saved lot of trouble in cases like this.	19:30
quentusrex	what's the issue?	19:31
	and I dont' see an ubuntu package for that.	19:31
Ilari	quentusrex: You where manipulating wrong branch. And thus it wouldn't resolve push problem on master.	19:31
jhasse	I got a repository A which pulls from B. There is still a file in B newer than in A although I did "git commit -a" in B. How is this possible?	19:31
	→ tvakah joined	19:31
quentusrex	Ilari: so, how do I 'fix' it?	19:32
	→ jjuran joined	19:32
frsk	jhasse: The files are not added? What does git status say about those files?	19:32
Ilari	quentusrex: Maybe something like 'git branch tmp ; git checkout master ; git merge tmp'...	19:33
jhasse	frsk, git status does not list these files at all (they are added)	19:33
	← kennethlove left	19:33
Ilari	quentusrex: That should also bring the commits from remote side with it.	19:33
	→ nsxdavid joined	19:33
quentusrex	ok	19:34
	that seems to have worked	19:34
nsxdavid	If I have a file that's unmerged in git status after a pull from upstream (github) and I want to just use what was upstream, do I use: git checkout --theirs <file>	19:34
mitsuhiko	Ilari: unlucky on my mac as well, the conflicts stay	19:35
	i just can't see any changes i made	19:35
	→ ciskje joined	19:35
quentusrex	ok, now on the remote side how do I update the checkout?	19:36
	I'm getting: fatal: 'master': unable to chdir or not a git archive	19:36
	fatal: The remote end hung up unexpectedly	19:36
	→ Kissaki joined	19:36
	← Kissaki left	19:36
cemerick	I had pieced together a command to list the N largest objects in a git repo, but now I've lost it. Anyone have something similar laying around?	19:37
quentusrex	and "git checkout text.txt" doesn't get the commit I just sent in	19:37
	→ gaveen joined	19:37
quentusrex	I'm now running the commands on the other workstation.	19:37
Ilari	quentusrex: If you get that sort of errors, you are probably using branch name as remote (often doesn't work).	19:37
	→ tbf joined	19:37
quentusrex	git branch only shows 'master' and 'with-im'	19:38
Ilari	quentusrex: Is 'git status' a mess now?	19:38
quentusrex	no	19:38
Ilari	quentusrex: Like lots of files modified, lots deleted.	19:38
quentusrex	there are a few, but they were there before hand.	19:39
	by lots I mean 4 modifies and 2 deleted.	19:39
Ilari	quentusrex: Maybe stash your changes 'git stash' then run 'git pull'?	19:39
quentusrex	git pull: fatal: 'origin': unable to chdir or not a git archive	19:40
	fatal: The remote end hung up unexpectedly	19:40
Ilari	quentusrex: Did you push to repo with working tree???	19:40
quentusrex	and how do I unstash?	19:40
	Ilari: I think it might have a working tree :(	19:41
Ilari	quentusrex: Maybe do 'git stash apply' to restore the stash. Then 'git reset --soft HEAD@{1}' to undo the push.	19:42
	quentusrex: Or actually, what branch that repo is on?	19:42
mitsuhiko	Ilari: any ideas what else i could check before throwing away the repo?	19:42
quentusrex	how do I undo a git reset --soft?	19:43
	:(	19:43
	→ bryanray joined	19:43
Ilari	quentusrex: 'git reset --soft HEAD@{1}'. But now you need to find the right commit from reflog next time.	19:44
	← altrux left	19:44
	← k0001 left	19:44
	→ k00011 joined	19:44
quentusrex	ok, so I have everything back to a 'stable' state.	19:44
	there is a working tree in the other repo	19:44
Ilari	quentusrex: If you pushed to checked out branch with local changes, you pretty much need to back off the push if you want to preserve your changes.	19:44
mitsuhiko	i can't see how that is caused. my latest revision (HEAD) exists in the repo i want to pull from as well	19:45
	and i don't have any changes there. why does git give me merging conflicts?	19:45
	k00011 → k0001	19:45
	→ jsbmsu joined	19:45
quentusrex	My end goal is to be able to push to a remote repo that has local changes, and if there are any issues at all for my push to fail.	19:45
	← jhasse left	19:45
	→ mjf joined	19:45
Ilari	quentusrex: That pretty much requires push to fail if there are any local changes.	19:46
quentusrex	what if none of the changes are in the same file?	19:46
Ilari	quentusrex: You would need to determine that without doing it in order to reject the push in time.	19:47
	→ ankit9 joined	19:47
quentusrex	precommit hook?	19:47
Ilari	quentusrex: No. pre-receive or update hook on remote side.	19:48
	→ webcoder joined	19:48
quentusrex	ok	19:48
Ilari	quentusrex: 'git commit' invokes pre-commit.	19:49
quentusrex	ok	19:49
	so how would pre-receive be able to tell if any of the committed files are changed and uncommitted on the remote repo?	19:49
Ilari	quentusrex: Request the names-only diff of update that would be appiled and compare with names-only diffs from HEAD to working tree and HEAD to index.	19:51
	→ _Vi joined	19:51
	→ cilly joined	19:53
	→ aleksei joined	19:54
	→ sergiodj joined	19:54
	→ struberg joined	19:58
	← peritus_ left	19:59
	← sergiodj left	20:00
zuez	Is there a compressed way to run a git diff to see how many lines have been modified from staging area to repo?	20:00
	eg: a diff of lines +/- staging vs what's in the actual repo.. to further clarify	20:00
	← RodP2 left	20:01
	abstrusething\|aw → abstrusething	20:01
	→ aspotashev\|eeepc joined	20:02
Fissure	git diff --stat --cached	20:02
	or s/stat/summary/ if you want it even shorter	20:02
	← cilly left	20:02
zuez	Fissure: Excellent, thanks mate.	20:03
	→ RodP2 joined	20:04
seanius_	is there a way to have a common system-wide or otherwise user-specified hooks dir?	20:04
Ilari	seanius_: AFAIK, no.	20:04
seanius_	:(	20:04
Ilari	seanius_: Its possible to use custom template or edit the system template, but that's only for new repos.	20:05
seanius_	right	20:05
	→ sergiodj joined	20:05
	seanius_ sends a bug to debian's bts and assumes it will trickle its way through the tubes	20:07
	→ muszek joined	20:09
	→ precision joined	20:10
muszek	hi... how do I do a "git reset --hard" on a single file?	20:10
Ilari	muszek: 'git checkout HEAD -- file'?	20:10
	→ faithfulgeek joined	20:10
muszek	Ilari: thank you	20:10
	→ smeevil joined	20:12
	→ rinogo joined	20:12
	→ paulboone joined	20:13
rinogo	Hi all - I'm using git for my web dev project, and it's getting very mundane having to type a commit message every time I want to check in, deploy my changes, and test. How can I enable empty commit messages for git?	20:13
	→ schacon joined	20:14
Fissure	wouldn't it make more sense to just test locally before you commit?	20:14
	→ radarek joined	20:15
	← sergiodj left	20:15
seanius_	rinogo: you're digging yourself into a hole if you do that	20:15
bleything	rinogo: empty commit messages defeat one of the major reasons for version control in the first place.	20:15
	→ ciskje_ joined	20:16
rinogo	@Fissure: I wish I could - my dev environment is on the server and requires the interaction of several components. The configuration of these elements on my local system in an identical manner is unfortunately quite difficult.	20:16
	← jsbmsu left	20:16
	→ hobodave joined	20:17
Fissure	so develop on the server...	20:17
rinogo	@bleything: Yes, I understand. The problem is that I have to commit to get my changes onto the server, deploy them, and test them. Empty commit messages on 'test' commits, and elaborate commit messages on important commits would be much better than what I'm forced to do now - type garbage chars for the commit message.	20:17
	← ciskje left	20:17
	abstrusething → abstrusething\|aw	20:18
	→ jsbmsu joined	20:18
UUncia	what's a bare repository?	20:18
rinogo	@Fissure: I am developing on the server. However, I'm using a local copy of the repository, along with Textmate and the Git plugin to streamline development.	20:18
	→ [intra]lanman joined	20:18
	seanius_ is glad he doesn't have to work on that project	20:18
	← nsxdavid left	20:18
Ilari	UUncia: Repository without working tree.	20:18
bleything	rinogo: well... I'd suggest fixing your workflow rather than hacking your tools. but, have you tried commit -m '' ?	20:18
UUncia	hmmmmm....	20:18
seanius_	rinogo: or set EDITOR=true	20:18
UUncia	you mean an empty repository?	20:19
seanius_	or actually maybe that won't work	20:19
rinogo	bleything: How would you suggest I fix my workflow given that I can't deploy on my local machine?	20:19
Fissure	UUncia: no, one without a working tree	20:19
Ilari	UUncia: Nope. Bare repo can have commits in it.	20:19
seanius_	rinogo: use hooks maybe	20:19
UUncia	so I probably don't know exactly what a working tree means.	20:19
Fissure	rinogo: don't make changes on your local machine, make them on the one where you can actually test them	20:20
bleything	rinogo: you could edit files directly on the server, for one. you said it would be difficult to set up a dev environment on your machine, but that's a one-time cost that I would suggest is worth the effort	20:20
	← paulboone_ left	20:20
rinogo	Fissure: But then I'd have to either 1) edit files directly on the server via vim or something similar, or 2) manually synch my files via sftp (zero fun)	20:21
Ilari	UUncia: The place you edit the files so you can then add the changes and commit.	20:21
Fissure	rinogo: i fail to see the problem with (1)	20:21
UUncia	i c.	20:21
	thx	20:21
seanius_	windows user maybe	20:21
bleything	rinogo: you could use unison or rsync to sync your tree.	20:21
	← amoeba left	20:21
rinogo	bleything: Good idea. Might be the best option	20:21
bleything	rinogo: or you could use sshfs to mount your server locally and edit your files in textmate that way	20:22
	→ kirkt joined	20:22
kirkt	is it possible to ask windows git related questions here?	20:22
	kirkt ducks	20:22
Fissure	kirkt: of course	20:22
bleything	rinogo: don't get me wrong, I'm not trying to be proscriptive here, but in my mind this is one of those things that's worth it to just do right in the first place.	20:22
rinogo	Fissure: For a project of this scope (it's going to require full-time work for at least four months), I'd hate to be editing all of my files on vim via ssh	20:22
kirkt	uhm, well it seems that when i'm using msysGit, doing a clone from github often gets stuck randomly	20:23
bleything	rinogo: out of curiosity, what makes it so difficult to set up a local dev environment?	20:23
	← ankit9 left	20:24
kirkt	to the point where i left it several hours to clone a 200kb repo and it didnt do anything beyong 86%	20:24
rinogo	everyone: thanks for all of the input. I understand I don't have the ideal workflow, but your suggestions should help. I think sshfs/rsync might be the best solution	20:24
	→ cs_student joined	20:25
bleything	rinogo: another option might be to use branches for development/pushing/testing, then squash all of those commits down to a reasonable commit message before merging into master	20:25
	but I'm one of those people who thinks SCM history is sacred so that's not somethign I'd do :)	20:25
Fissure	kirkt: no idea, never seen that happen	20:26
	bleything: shouldn't you be using hg instead then? ;)	20:26
bleything	Fissure: who said I'm not? :D	20:26
rinogo	bleything: There are a variety of issues, but the most frustrating is that the site is based on Drupal, deployed with custom change management scripts to aid in tracking/merging both code and DB changes across dev/staging/production. (Anyone who has worked with Drupal knows that it can be very frustrating that it stores so much configuration in the DB.. :( ). With the config files of the change management scripts, it's tou	20:27
bleything	Fissure: I inherited a project that's in git and I haven't had time to experiment with changing yet. thinking about it.	20:27
rinogo	bleything: Another nice option - thanks.	20:27
bleything	rinogo: that cut off at "it's tou..." but I get the idea. that does sound like a bummer.	20:27
	rinogo: it annoys me when deployment policy gets in the way of developer productivity	20:27
	→ ereslibre joined	20:28
rinogo	bleything: Thanks for your commiseration! :)	20:28
bleything	:D	20:28
	→ Radar joined	20:29
	→ _wwwald joined	20:30
	← wwwald left	20:31
	← kumbayo left	20:32
	← jceb left	20:35
	→ jackdempsey joined	20:40
gitte	jsbmsu: Hi! how are you?	20:41
	→ tag joined	20:43
tag	Is there an equivalent to cvs up -P for git?	20:43
	→ aspotashev joined	20:43
tag	ohh	20:44
	← struberg left	20:46
	→ ivanoats joined	20:47
Ilari	tag: pull with some parameters might come close.	20:48
tag	I found that the stragglers were actually left because I didn't clean my working copy with my build tool before switching branches	20:49
spearce	gitte: please post more without a subject line	20:49
	gitte: or better yet, change the subject line to "Re: posting without a subject line"	20:49
gitte	spearce: will do ;-)	20:49
	→ matthieu joined	20:50
gitte	spearce: or: ", was Re:"	20:50
	← krh left	20:50
	→ solydzajs joined	20:50
jsbmsu	gitte: oh, hi there, I'm well, and you?	20:50
	→ fbond joined	20:50
Ilari	...	20:51
fbond	Hi. How can I pull and overwrite my local tree (I want to have a mirror of the remote repository, but it has been rebased)?	20:51
gitte	jsbmsu: heh, I am overworked as always.	20:51
Ilari	fbond: 'git pull --rebase'? It should handle the nonlinearity.	20:52
	← aspotashev\|eeepc left	20:52
fbond	Ilari: Thanks.	20:52
gitte	jsbmsu: after reading your response to my clueless response, I managed to see just _how_ clueless it was.	20:52
jsbmsu	gitte: uhh..which part of which response? :)	20:52
gitte	jsbmsu: the minimize button.	20:53
	jsbmsu: did I manage to be clueless somewhere else, too?	20:53
	→ mw joined	20:53
jsbmsu	gitte: no. I'm wondering if "clueless" is the word you're looking for though	20:54
	← jackdempsey left	20:54
	← _wwwald left	20:54
gitte	jsbmsu: I really mean it ;-)	20:54
	← fbond left	20:54
	→ igli joined	20:55
jsbmsu	gitte: well, I think the problem is we're trying to read each other's minds -- or see each other's mind's eye	20:56
gitte	jsbmsu: heh.	20:57
	jsbmsu: I am actually intrigued by the idea to let the object under the mouse cursor enlarge once the cursor hovers there for a fraction of a second.	20:58
	→ eventualbuddha joined	20:58
gitte	jsbmsu: I am just searching for an intuitive way to minimize it again, without the user having to click on some "minimize" button.	20:58
jsbmsu	gitte: well, I think auto-expanding could get really irritating..user holds mouse for a moment, then says "wtf, get out of the way"	20:59
gitte	jsbmsu: but you are right, in the meantime everybody -- and her dog, too -- spent so much time learning the ways of Microsoft that we might just as well have that button, too...	20:59
	← aspotashev left	20:59
gitte	jsbmsu: user wiggles the mouse out of the way, problem solved ;-)	20:59
jsbmsu	gitte: or more generally, I think there are some expectations you need good reasons for violating	21:00
	gitte: I guess it depends on how dramatic the expansion is	21:00
	gitte: but I don't think a single click is much work to open a commit, but experience says anything clickable should do something when moused over to tell you it's "alive"	21:01
	→ mikearr joined	21:01
Ilari	When it comes to "intuitive" user interfaces, there is NO substitute for usability testing.	21:02
jsbmsu	Ilari: can't disagree with you there	21:02
ciaranm	notepad is intuitive. vim isn't. thus, notepad > vim.	21:02
killerchicken_	In many ways, that is very true. In many other ways, that's very wrong...	21:03
xenoterracide	unless you're already used to vi, then notepad is not intuitive	21:03
ciaranm	walking is intuitive. flying a plane is not. thus, walking > plane.	21:04
killerchicken_	Assuming I want to edit a config file for the first time in my life, and I'm used to a Windows gui, using Notepad for the task makes much sense.	21:04
jsbmsu	jjjjjjj why am I typing so many j's?	21:04
	← yann left	21:04
xenoterracide	+ you can turn vim into notepad	21:04
	but notepad can never be vim	21:04
gitte	jsbmsu: I think that it is intuitive when the user sees something happen and understands why.	21:04
	jsbmsu: but I am not married to the idea.	21:05
jsbmsu	gitte: as opposed to? the user knows what is going to happen before she clicks?	21:05
xenoterracide	ciaranm: crashing a car is intuitive, driving it is not crashing > driving	21:05
igli	hover is fine, but it's not essential.	21:05
gitte	jsbmsu: I just positively hate when I have to click a bazillion times in Windows just to do the thing that should be quick and fast. And I want to Kill Bill.	21:05
	← matthieu left	21:06
gitte	jsbmsu: most users will not know what they should do, because they haven't read the documentation.	21:06
jsbmsu	gitte: fair enough. don't look now, but you just repeated my argument for the panes ;)	21:06
gitte	jsbmsu: my best definition of intuitive is: "usable without having read the documentation."	21:06
ciaranm	the nipple is intuitive. a cup is not. thus, nipple > cup.	21:06
jsbmsu	gitte: I like it	21:06
ciaranm	no, wait, that one's true.	21:06
gitte	jsbmsu: maybe I did.	21:06
	jsbmsu: hmmm ;-)	21:07
jsbmsu	gitte: but I also understand your motivation for avoiding it	21:07
	→ smashwilson joined	21:07
	xenoterracide wants to see how intuitive notepad is when I pull your mouse out of your computer	21:08
jsbmsu	gitte: I think another important thing is to think about what the user wants or expects to be able to do easily, and (if possible), make it so	21:09
	→ aspotashev\|eeepc joined	21:10
jsbmsu	gitte: that may sound too vague...an example: my dad uses flickr...it's great for a user who wants to view a slideshow	21:10
gitte	jsbmsu: yeah.	21:11
jsbmsu	gitte: but he also says "my photos are backed up"...well, if he wants to re-get all his photos, he has a few clicks per photo, one at a time	21:11
gitte	jsbmsu: right.	21:11
	→ RaceCondition joined	21:11
	← RaceCondition left	21:11
jsbmsu	gitte: so if users really wanted to grab a photo archive often, the tool is terrible	21:11
	← priidu left	21:11
gitte	jsbmsu: I think you can actually tell people what they want to do with a good UI.	21:11
xenoterracide	it's interesting... the mouse is not intuitive, ever seen a first time user try to use one...	21:11
	← rinogo left	21:12
gitte	jsbmsu: for example, inspecting changes without checking the files out. Once people find out that this is possible, they start using that feature.	21:12
igli	trackballs, xenoterracide ;)	21:12
gitte	xenoterracide: yes, they drop it as soon as the mouse bites them.	21:12
xenoterracide	igli: I don't think a tracball is any easier for most novices... but I could be wrong	21:13
	but everyone who's literate can use a keyboard	21:13
jsbmsu	gitte: for ex. we have assumed a user only wants to inspect one commit at a time...is that so accurate?	21:13
igli	it was for me. never used a mouse for years. still won't.	21:13
	← faithfulgeek left	21:14
jsbmsu	gitte: there's something to be said for a gitk log view...this tool may suck in that regard	21:14
xenoterracide	igli could be, I just love seeing people use 2 hands to put a mouse on a spot. move there hands go to click and then have the mouse move before the click signal is issued	21:14
igli	heh; I love watching people move the mouse, then lift it up and move it back ;)	21:15
	what a waste of energy ;)	21:15
gitte	jsbmsu: the thing is, gitk is pretty fixed in its output.	21:15
xenoterracide	igli: heh	21:15
gitte	jsbmsu: your tool could be taught to show number of changed files, or lines.	21:15
	jsbmsu: or lists of changed files.	21:16
xenoterracide	I want a trackpoint on my desktop keyboard	21:16
gitte	jsbmsu: or even diffs.	21:16
	jsbmsu: the important part is that you have a design to work with.	21:16
xenoterracide	then I never have to use anything but the keyboard	21:16
igli	meh i hate those. they're so fiddly after a trackball	21:16
gitte	jsbmsu: this information should be shown either in the same box, or in boxes that can expand just like the commit itself.	21:16
jsbmsu	gitte: so are you picturing something that's ultimately fairly user-configurable in terms of appearance?	21:16
xenoterracide	igli: they are better than a touchpad	21:17
gitte	jsbmsu: you could even have a "+" on the commit, and for users like me a feature that hovering already expands it.	21:17
igli	heh i hate both, xenoterracide ;) but yeah IBM used to have nice ones.	21:17
gitte	jsbmsu: I want it to be the revolution of usability in VCS UIs.	21:17
	jsbmsu: kind of the opposite of PVCS.	21:17
xenoterracide	http://xenoterracide.blogspot.com/2009/05/git-presentation.html doing a presentation tomorrow if anyone wants to look over my slides see if I forgot anything	21:17
jsbmsu	gitte: Pvcs?	21:18
xenoterracide	igli: I'm not buyin a laptop without a trackpoint ever again	21:18
	I can't stand touchpads	21:18
gitte	jsbmsu: I don't want to give you eye cancer, otherwise I'd post a link to a screenshot.	21:18
	jsbmsu: and it is slow like hell. And limited.	21:18
	jsbmsu: kind of reminds me of eclipse.	21:18
kirkt	i dont understand, i commit to a git repo, and push to github and the user is written as [unknown] any idea about this?	21:19
jsbmsu	gitte: by the way, is JGit (same as JavaGit?) a wrapper around part of Git, or a complete recode?	21:19
	→ hpoydar joined	21:19
gitte	jsbmsu: JGit is a complete recode, JavaGit is a set of wrappers.	21:20
	← nighthwk\| left	21:20
gitte	jsbmsu: JavaGit is also half-dead AFAICT.	21:20
	→ fridim joined	21:20
killerchicken_	xenoterracide: how about explaining concepts instead of commands? For example, someone not used to the idea of an index will not understand much, I'm afraid	21:20
jsbmsu	gitte: in the same vein: if we do the GUI in Java/Swing, do we have to use JGit or can we still access the official (or msys) Git?	21:20
Tommy[D]	kirkt: Did you use the same email and you have in your github account?	21:21
xenoterracide	killerchicken_: speech accompanies the slides ;)	21:21
kirkt	Tommy[D]: define 'use' ?	21:21
killerchicken_	xenoterracide: of course	21:21
gitte	jsbmsu: if you find it easier to access CGit, that's okay.	21:21
Tommy[D]	kirkt: you can set a username and an email in your git config	21:21
xenoterracide	killerchicken_: since I won't be reading the slides the concepts aren't there as much	21:21
gitte	jsbmsu: in some respects, JGit is not yet there. For example, I am in the middle of implementing jgit diff, as that does not exist yet.	21:22
Tommy[D]	kirkt: and you can add an email to your git account, if they are both the same, your accout is shown instead	21:22
kirkt	Tommy[D] git config user.email ?	21:22
	→ carrus85 joined	21:22
Tommy[D]	kirkt: yes	21:22
xenoterracide	killerchicken_: otherwise good suggestion	21:22
jsbmsu	gitte: at any rate, I'd like to decide on a language soon so I know if I need to be studying Tcl/Tk or brushing up on Swing	21:22
gitte	jsbmsu: what's your preference?	21:23
killerchicken_	xenoterracide: yeah, it just seems that the concepts will take the most explaining, and that's why I wondered why there's not a single slide about the	21:23
	m	21:23
	← eventualbuddha left	21:23
xenoterracide	killerchicken_: depends on which concepts you mean... I'll be explaining most of them as I go along	21:24
jsbmsu	gitte: I don't know that I can make a well-informed decision... about all I can say is I think I'd prefer to build against official Git	21:24
gitte	jsbmsu: well, JGit is as official as a non-C Git can be: it is actively developed by the guy who has the second most commits in C Git.	21:25
xenoterracide	killerchicken_: curious which concepts do you think will take a long time?	21:26
	← ivanoats left	21:27
jsbmsu	gitte: perhaps I should say the most popular version of Git	21:27
	→ Naib joined	21:27
	← __iron left	21:27
jsbmsu	gitte: do you think JGit will seek to maintain compatibility with CGit?	21:27
	→ dereine joined	21:28
Naib	hi, I had some issues with a repo so I have done a few test commits/pushes. issue is now fix (problem elsewhere). I would like to scrub these ~6 test commit's/pushes	21:28
	what is the best way?	21:28
killerchicken_	xenoterracide: the index, merging branches, reflog	21:28
dereine	is there a way to transfer a bzr repo to a git repo?	21:28
killerchicken_	xenoterracide: sorry, would love to talk more about this, but I gotta run. I hope your presentation goes well, good luck!	21:29
xenoterracide	killerchicken_: I'm not sure in all cases I fully understand all concepts, but it's going t be dissussed	21:29
	killerchicken_: later	21:29
	→ romao joined	21:29
kirkt	Tommy[D]: still doesnt work	21:29
Tommy[D]	kirkt: and you did add that email to your github account?	21:30
	→ __iron joined	21:30
kirkt	yep its the default	21:30
	← radarek left	21:31
gitte	jsbmsu: JGit thrives to be completely compatible with CGit.	21:31
jsbmsu	gitte: by compatible, I primarily mean what is stored in .git/ ...able to alternate accesses to a repo with JGit and CGit with no problems	21:31
	→ WALoeIII joined	21:31
	← Radar left	21:31
gitte	jsbmsu: and it is largely feature-compatible. A few notable exceptions are being worked on...	21:31
Tommy[D]	kirkt: and you did a new commit? This wont change already done commits since they dont have the email set	21:31
	← carrus85 left	21:31
gitte	jsbmsu: that is one of the most important things: JGit and CGit need to be able to share repositories _without_ problems.	21:32
	jsbmsu: they also need to be able to talk to each other.	21:32
jsbmsu	gitte: I guess it could result in much less mess cross-platform, compared with the msysGit situation	21:32
	← codebrulee left	21:32
	← gotgenes left	21:33
kirkt	Tommy[D] yes, git commit file -m ".."	21:34
Tommy[D]	kirkt: does "git show" contain your email?	21:34
jsbmsu	gitte: the other issue that I can't well assess, is which GUI toolkit (Tk or Swing) is going to serve us better	21:34
kirkt	Tommy[D] show about where?	21:36
henon	Swing is the superior toolkit compared to Tk IMHO	21:37
Tommy[D]	kirkt: the header part should show your email	21:37
kirkt	Tommy[D] sorry for the ignorance, i should put git show in the commandline?	21:38
Tommy[D]	kirkt: where else? it should show you the last commit	21:39
	→ ivanoats joined	21:40
	xenoterracide grumbles at the tardiness of people who schedule appointments	21:40
	← igli left	21:40
jsbmsu	henon: any particular (brief) reasons?	21:42
	→ eventualbuddha joined	21:42
kirkt	Tommy[D] yes it shows correct username and email	21:43
	← tbf left	21:44
	← RodP2 left	21:44
hachi	I've got a repo here, with HEAD set to refs/heads/trunk	21:45
	refs/heads/trunk exists and has a sha1 in it	21:45
	gitweb is refusing to show the 'trunk' branch now	21:45
	do I try to examine that revision with something to try and debug this?	21:45
	→ wycats joined	21:46
	→ dodo_the_last1 joined	21:46
henon	jsbmsu, this is not very brief but it ll do:	21:46
	A big difference between Tk and Java's AWT/Swing is that Java provides not just a set of widgets, but also the entire infrastructure for creating new widgets, drawing graphics (manipulating individual pixels), etc. This means Java is very, very complicated, but allows you to create new widgets without needing to code in C. Tk, on the other hand, will usually require you to code in C to create an entirely novel widget' (but, it is act	21:46
	ually quite rare that this is needed, and this is possibly why Tk is so much easier to use -- many, many so-called 'new widgets' can be implemented simply as mega-widgets, without any C).	21:46
hachi	according to `git log $sha` it does exist	21:46
wycats	Is there a way to tell git not to create a merge conflict if the only difference is whitespace?	21:46
	or rather, to ignore whitespace in the calculation of merges	21:47
gitte	jsbmsu: yes, JGit will eventually give us a much easier time to support obscure platforms such as Windows or VMS.	21:47
	jsbmsu: you might even argue that Windows is an evolution of VMS...	21:47
wycats	one side made whitespace changes and the other side made real changes	21:47
	we're getting a ton of conflicts	21:47
jsbmsu	gitte: if you knew what VMS was..	21:47
gitte	jsbmsu: just Cesar-encode it: WNT.	21:48
eventualbuddha	when a new branch is pushed to a remote, the post-receive hook gets 000...000 SHA refs/heads/my-new-branch. how do I get a list of all the pushed commits?	21:48
	→ hyperair joined	21:48
jsbmsu	gitte: well now you're just being silly	21:48
gitte	jsbmsu: of course I am...	21:49
	jsbmsu: I just spent an evening chasing an msysGit bug.	21:49
	← aspotashev\|eeepc left	21:49
gitte	jsbmsu: after roughly 3 hours, the solution is...	21:49
	jsbmsu: unset CYGWIN	21:49
jsbmsu	gitte: ouch	21:49
gitte	jsbmsu: indeed... so you may forgive me my little nastiness in the vague direction of Redmond.	21:50
	← _Vi left	21:50
hachi	anyone know why git would be setting permissions on refs/*/ to be 600 instead of 644 ?	21:51
jsbmsu	gitte: understood...I feel "stuck" to windows more than married	21:51
	gitte: just for reference, what time do you have?	21:51
	→ offbytwo joined	21:52
	← offbytwo left	21:53
eventualbuddha	wycats: I've noticed that too. seems silly	21:53
	(I know there are whitespace-sensitive files, but...)	21:53
Ilari	hachi: Umask?	21:53
hachi	ugh	21:53
	good call	21:53
	I bet someone changed the umask on these machines	21:54
	→ _Vi joined	21:54
hachi	(I will destroy them)	21:54
	← bcardarella left	21:54
	→ stuffcorpse joined	21:55
	→ dduncan joined	21:56
	→ Radar joined	21:56
	→ lightcap joined	21:56
UUncia	Is it possible to deliver/commit only some changes from one branch to another?	21:57
	→ Gitzilla joined	21:57
shd	UUncia: sha1 ids will have to change then.. use format patch in the branch, then the patches to another	21:57
kirkt	Tommy[D] on some places it recognize it was me, on the main page it writes [unknown]	21:58
shd	+then apply the patches to another..	21:58
UUncia	lol... I have no idea what that means... I'm still a beginner.	21:58
	in git, that is.	21:58
	→ Nugget joined	21:58
	← dduncan left	21:58
shd	UUncia: there are special cases, where it is possible, but not really a beginner feature :)	21:58
	← _graham_ left	21:59
shd	UUncia: you could do: branch A: git format-patch HEAD~4 (take 4 patches), then branch B: git am patch-files-you-want-to-apply	21:59
	← Naib left	21:59
	← ivanoats left	22:00
gitte	jsbmsu: sorry... Previously, my IRC client told me when there was a message for _me_ in that channel...	22:00
	jsbmsu: it is exactly midnight here.	22:00
	→ hummingkid joined	22:00
UUncia	shd - shouldn't something like that be really needed?	22:01
jsbmsu	gitte: so you're 7 hours ahead of me	22:01
shd	UUncia: it's very possible, but fixing conflicts will arise	22:01
UUncia	well, that's only logical.	22:02
shd	UUncia: it's easy to pick the commits (patches), but applying them might cause conflicts	22:02
gitte	jsbmsu: are you in Chicago, or at least its tz?	22:02
jsbmsu	gitte: same tz	22:02
	→ cmarcelo joined	22:02
	← smeevil left	22:03
	→ bremner_ joined	22:03
	→ memphismemosley joined	22:04
	← memphismemosley left	22:04
gitte	jsbmsu: out of curiosity: which town/state?	22:05
jsbmsu	gitte: to be more specific, I guess the concern is which GUI toolkit well be helpful in managing the layout of commits and floating a "commit window" above the graph	22:06
	gitte: the small college town of Starkville, Mississippi...at least until the end of summer	22:06
	gitte: you're in Dresden?	22:07
gitte	jsbmsu: I am pretty comfortable with both Tcl/Tk and Java/Swing, but henon is absolutely right, Java gives you much better control over the low-level stuff.	22:07
	jsbmsu: that's correct!	22:07
jsbmsu	gitte: and with something like this, do you think there's a good probability we will want that level of control?	22:07
henon	it also has the better look and feel, but your mileage may vary	22:07
	→ merlin1983 joined	22:08
jsbmsu	henon: vary by...hmm..OS? :)	22:08
gitte	jsbmsu: well, pulsating rectangles are probably easier using Java/Swing... ;-)	22:08
henon	=D	22:08
	→ sjzzalx joined	22:08
sjzzalx	Where should I configure push.default? I tried to put it under core in .git/config but thing cried to me	22:09
gitte	henon: with Tcl/Tk 8.5, I actually have to disagree slightly: the look and feel of Swing on Linux leaves to be desired.	22:09
	sjzzalx: I think you should put "default" into "[push]"...	22:09
jsbmsu	I thought Swing (at least by default) had an OS-independent look...is that not the case?	22:10
	← schmrz left	22:10
	→ bcarlyon\|laptop joined	22:11
sjzzalx	gitte: OK, thanks. I don't have a push section ... is that supposed to go locally or on the thing I am pushing to?	22:12
	→ drewr joined	22:12
	cs_student → poseidon	22:12
gitte	sjzzalx: the push setting goes locally IIRC.	22:12
sjzzalx	gitte: OK, thanks. :)	22:12
gitte	sjzzalx: and the notation you used ("push.default") suggests to use "git config" to set it, which creates the section by itself.	22:13
eventualbuddha	given a SHA, is there a way to list all commits not already in the tree of one of the named branches (e.g. master, etc)?	22:13
	← Nugget left	22:13
sjzzalx	ok	22:13
	thanks gitte	22:13
eventualbuddha	I'm trying to figure out how to generate a list of commits that were pushed in a post-receive	22:13
	→ obama joined	22:13
	obama → Guest83303	22:14
henon	gitte, whow, Tile looks really great	22:14
	bcarlyon\|laptop → BarryCarlyon	22:14
gitte	henon: doesn't it?	22:14
	← bremner_ left	22:14
	← Guest83303 left	22:14
	→ bush_78 joined	22:15
	← bush_78 left	22:15
	→ snuxoll joined	22:15
jsbmsu	gitte: something tells me you would prefer Java/Swing, anyway	22:16
	→ Nugget joined	22:16
	→ obama joined	22:17
gitte	jsbmsu: actually, I would prefer Tcl/Tk, because the development speed would be faster: it is interpreted, and readily available on all platforms where we run Git.	22:17
	obama → Guest68314	22:17
	← texel left	22:17
gitte	jsbmsu: but some people (for example a namesake of yours) have pretty serious concerns regarding Tcl/Tk and the number of developers able to speak this language.	22:18
	← Guest68314 left	22:18
jsbmsu	gitte: do you disagree? is it a legitimate concern?	22:19
henon	true, if you want to push the project a popular language helps alot. if you asked on ruby-lang who would join a git-gui project at least 10 would readily join	22:19
	→ hawken joined	22:19
	→ carrus85 joined	22:22
henon	tcl/tk is available for all of the popular high level scripting languages ...	22:23
	← carrus85 left	22:23
	→ obama joined	22:24
	obama → Guest13675	22:24
henon	ruby being the best of all (me thinks :D )	22:24
	→ carrus85 joined	22:24
gitte	jsbmsu: I really haven't made up my mind, but I think that your preference would be important.	22:24
	← Guest13675 left	22:24
jsbmsu	i wonder how well non-Tcls stay up-to-date with Tk	22:25
	→ qrush joined	22:26
jsbmsu	gitte: my gut votes for Tcl/Tk, but for something as logical and rational as programming, I'm not sure the gut is worth much	22:26
	→ bush_53 joined	22:26
	← hyperair left	22:26
gitte	jsbmsu: well, the thing is, you were very good at learning Tcl/Tk that quickly, I have no doubt that you can switch from Tcl/Tk to Java/Swing if need be.	22:27
	jsbmsu: and of course, I would help you there.	22:27
hawken	Hello	22:27
	all	22:27
	← bush_53 left	22:27
sjzzalx	Hello	22:28
	how are things	22:28
jsbmsu	gitte: if we went with Tk, any opinion on choice of scripting language?	22:28
	→ FFighter joined	22:28
FFighter	hello	22:28
sjzzalx	hello	22:28
FFighter	I removed two subdirs and added them as a submodule	22:28
	not git says that I can't pull because these files would be overwritten by merge	22:28
	← Alinermedia left	22:28
FFighter	I though on forcing it	22:28
	but I'd rather do something else	22:29
	any hints?	22:29
gitte	jsbmsu: well, I would go for Tcl, as we have that in msysGit already.	22:29
jsbmsu	gitte: ah, yes	22:29
gitte	jsbmsu: I cannot compile Python in msysGit, so that would not be my preference, but then, I'd not participate in this project using Windows, but Linux.	22:29
	→ KwikkSilva joined	22:30
	→ Guest91181 joined	22:30
jsbmsu	gitte: my guess is most of my peers I will want to use it with will be on Windows :)	22:30
gitte	jsbmsu: we do not have Tk bindings for our Perl, as Perl is MSys for the time being (nobody stepped forward doing the massive work of making Perl scripts Windows-compatible), and Ruby is not in msysGit at all...	22:30
FFighter	I really need a hand here :)	22:30
	we are about to push a new build	22:30
gitte	jsbmsu: yeah.	22:30
	jsbmsu: but with a bone-headed Open Sorcerer as me not being in the way, you could just install a binary for Python or Ruby.	22:31
	← Guest91181 left	22:31
jsbmsu	gitte: there's something to be said for ease of installation...if Tcl/Tk is already there for the msysGit users, that's one less tep	22:32
gitte	jsbmsu: right.	22:32
	→ Dashkal joined	22:32
	← dwave left	22:32
gitte	jsbmsu: You should experiment a little with drawing on a (potentially large) canvas before reaching a conclusion, though.	22:33
jsbmsu	i'm trying out git-svn on Ubuntu. I have a local copy of the SVN repo, but I can't figure out how to format the SVN URL in order to clone it	22:33
	→ struberg joined	22:33
jsbmsu	gitte: I think I'll dive into Tcl/Tk this week and play around with it	22:33
gitte	jsbmsu: great!	22:34
	← cmarcelo left	22:34
bremner	jsbmsu: file://svnrepopath ?	22:34
	→ faithfulgeek joined	22:34
jsbmsu	bremner: sometimes you have to ask a question to figure out the answer...typing the wrong user name after /home/. thanks though!	22:34
henon	gitte, please keep me up to date! i am very very interested. is there already a project site?	22:36
gitte	henon: for the graph-based UI? No, not yet.	22:36
	henon: but certainly, we'll keep you posted, and definitely come back to you with questions about intuitive design and usability, as that is your major subject.	22:37
henon	i also know a lot about efficient rendering of large data	22:37
	ok, so ... it is late for me. gn8 gitte, jsbmsu	22:38
gitte	henon: cool!	22:38
	henon: git night!	22:38
jsbmsu	henon: thanks, take care	22:38
gitte	henon: my fix for issue 138 seems to work... ;-)	22:38
	← jrmuizel left	22:38
henon	you make me very happy today!	22:38
gitte	;-)	22:39
	→ tobydeh joined	22:40
tobydeh	hi everyone	22:40
	← henon left	22:40
tobydeh	whats the quickest way to set password protection on a per project basis to repos served over http	22:41
thiago_home	.htaccess	22:41
tobydeh	ok, when i use the git client to clone a url with a .htpasswd i get a 401	22:43
	← k0001 left	22:43
thiago_home	401 = Authentication Required	22:44
	← sjzzalx left	22:44
tobydeh	i know that	22:44
	im asking how to tell the client the user and pass	22:44
	it doesnt ask for it	22:44
thiago_home	I don't know if it supports that	22:45
	← solydzajs left	22:45
zuez	odd, how can my master branch be ahead of origin/master by two commits if I don't touch master branch and only pull to it?	22:45
	← smashwilson left	22:45
thiago_home	I don't see "401" anywhere in the Git source code, so I doubt it handles password challenges	22:46
	zuez: you have two commits ahead of origin/master	22:46
	it could be that you pulled directly into it, bypassing the local origin/master copy	22:46
tobydeh	thiago_home you just need to create a .netrc with the login details	22:46
	because git uses curl	22:47
thiago_home	ah, didn't know that	22:47
tobydeh	now you do :)	22:47
	thiago_home has never used the HTTP protocol for Git transfers	22:47
thiago_home	it's a lot less efficient than the native protocols	22:47
tobydeh	yes thats true, but if im sharing source with the public (most of which are SVN users) i need the simplest possible method of allowing them to checkout and update	22:48
	what protocl do you use?	22:49
	→ texel joined	22:52
	→ josesito joined	22:54
	← jsbmsu left	22:54
josesito	how can I know what files changed between current branch and a specific commit of another branch?	22:54
	→ jsb119 joined	22:55
	← tobydeh left	22:55
	← jsb119 left	22:56
	← faithfulgeek left	22:57
	→ jsbmsu joined	22:57
	→ bnferguson joined	22:57
zuez	thiago_home: How does that work, where I bypass the local origin/master copy?	22:58
	thiago_home: considering I never work on the local master branch, I shouldn't be ahead of the origin/master	22:58
thiago_home	git pull remotename branchname	22:58
	that does the pull, but doesn't touch origin/master	22:58
zuez	What's the best way to fix this	22:58
	I prefer not to push	22:58
	← kirkt left	22:58
	← psoo left	23:00
	→ psoo joined	23:00
	→ hachi joined	23:01
	→ bobmcw_ joined	23:02
	← [intra]lanman left	23:02
	→ agile joined	23:04
	→ Blafasel_ joined	23:04
	→ joshsdc joined	23:05
robinr	spearce: anything obviously wrong when I get an NPE at repo = server.getRepositoryCache().get(repoName); com.google.gerrit.server.ProjectAdminServiceImpl$10.run(ProjectAdminServiceImpl.java:405) ?	23:07
bizhat2	when i run git remote rm orgin, i get error error: Could not remove config section 'remote.orgin', how do i remove orgin ? I need to add new orgin	23:07
	,	23:07
spearce	robinr: no git_base_path in system_config table. or, more accurately, its null. see the installation guide in the documentation	23:08
bizhat2	git remote rename also not working -> error: Could not rename config section 'remote.orgin' to 'remote.orgin_old'	23:08
bob2	did you remove config section remote.origin from .git/config?	23:09
	← mjf left	23:10
	← psoo left	23:10
	→ Run joined	23:11
bizhat2	bob2: no, it was working, i want to change my respo url as i moved to a different git hosting, so i run "git remore rm orgin" to delete current orgin, this is correct way to do ?	23:11
	can i remove [remote "origin"] from .git/config ?	23:12
	→ meyering joined	23:12
	→ jrmuizel joined	23:14
	← stuffcorpse left	23:14
	→ stuffcorpse joined	23:14
	→ cemerick joined	23:16
struberg	spearce where can I pull the fix for the ArrayIndexOutOfBoundsException from, or should I apply the patch from the mail?	23:17
spearce	struberg: have to apply from mail, i didn't publish it anywhere	23:18
robinr	spearce: nope, it's set	23:19
spearce	robinr: did you restart the server since you set it?	23:19
	→ arohner joined	23:19
	← muszek left	23:19
	← ciskje_ left	23:19
robinr	not, sure. trying	23:20
spearce	robinr: git_base_path is cached on startup and if null, RepositoryCache doesn't get constructed, leading to an NPE here.	23:20
	→ gotgenes joined	23:20
robinr	works :)	23:20
spearce	robinr: also, make sure your system_config only has 1 row in it; guy here the other day somehow got 2 rows in there and gerrit was reading the one with the NULL in it.	23:20
	ok	23:21
	→ ywarnier joined	23:21
	→ bcardarella joined	23:22
	← qrush left	23:24
	→ scarabx joined	23:24
	← bizhat2 left	23:25
	← ywarnier left	23:25
	→ seangrove joined	23:25
	→ neoeinstein joined	23:27
	← seangrove left	23:28
	→ seangrove joined	23:28
	← FFighter left	23:28
	→ RaceCondition joined	23:28
	← RaceCondition left	23:29
	← seangrove left	23:30
	→ jpeeler joined	23:31
	→ towski joined	23:31
	← __iron left	23:32
	→ seangrove joined	23:32
	→ JohnnyL joined	23:33
zuez	git log origin/master.. is ultimately show me the head of origin/master or?	23:33
struberg	grml, anyone applied patches from yahoo webmail yet? don't have a mbox on this laptop...	23:33
	→ SinJax joined	23:34
SinJax	hello, im having some conceptual problems with git sparked by some (possibly naughty) behaviour of some friends... basic question first:	23:34
	how do you diff two branches?	23:35
RandalSchwartz	git diff thisbranch...thatbranch	23:35
SinJax	or rather "Is it possible to diff two branches?"	23:35
	← webcoder left	23:35
RandalSchwartz	no. completely impossible. who would ever want to do that... :)	23:35
SinJax	RandalSchwartz: heh :P	23:35
	← Lenary left	23:36
RandalSchwartz	next question?	23:36
	dereine → dereine[OFF]	23:36
	RandalSchwartz waits	23:36
	← schacon left	23:37
SinJax	and now the naughty bit :)	23:37
	it would seem from the diff	23:37
	wait sorry ill start from the beggining :D	23:38
RandalSchwartz	"ill"? are you sick?	23:38
	or did you mean "I'll"?	23:38
JohnnyL	SinJax maybe you should read the short intro tutorial?	23:38
SinJax	yes i mean I'll :)	23:38
RandalSchwartz	are you paying for IRC by the character? :)	23:38
	← lightcap left	23:38
	← josesito left	23:38
SinJax	two branches, one is the master of one repo, and another branch master of another repo	23:38
RandalSchwartz	do they have a common ancestor?	23:39
SinJax	the problem is that the diff of these two branches seems to think the same lines were remove then added again	23:39
gitte	SinJax: you could have contradicted Merlyn and said "I wanted to say 'will'".	23:39
SinJax	RandalSchwartz: thats the problem ;)	23:39
RandalSchwartz	you mean "that's"?	23:39
	you seem to have a problem with your "'" key	23:39
gitte	SinJax: I guess the lines are not really identical, but have white-space issues.	23:39
RandalSchwartz	hey gitte	23:40
gitte	SinJax: you can test with "git diff -w <branch1>..<branch2>"	23:40
	hey RandalSchwartz!	23:40
	RandalSchwartz defers answering questions when Smarter People(tm) are around	23:40
gitte	RandalSchwartz: how are you?	23:40
RandalSchwartz	Busy	23:40
SinJax	oh randal, though its remarkably interesting to discuss the apostrophe and its many meanings, id argue its primarily a cultural throwback...exemplified by the fact that you can correct me perfectly well	23:40
	ehm	23:40
RandalSchwartz	finally have a full-time contract	23:40
gitte	RandalSchwartz: I do the same, so usually I let you do all the talk.	23:40
	RandalSchwartz: great news!	23:40
RandalSchwartz	after 8 months of non-employment	23:40
gitte	RandalSchwartz: was well needed, I know.	23:40
RandalSchwartz	so yes, I'm about to lose one of my properties	23:41
gitte	RandalSchwartz: that sux, 8 months of waiting.	23:41
RandalSchwartz	and take the "been repo'ed" hit	23:41
	→ rettub_ joined	23:41
gitte	RandalSchwartz: could you grace a poor non-native speaker with some insight as to what "been repo'ed" means?	23:41
RandalSchwartz	reposessed	23:42
	gone in default for lack of payment	23:42
	→ schacon joined	23:42
RandalSchwartz	foreclosed, in the realm of real estate	23:43
gitte	RandalSchwartz: oh, that's bad, I am very sorry!	23:43
	← romao left	23:43
SinJax	gitte: ok good good...now the problem! :D what i want to do is merge these two branches, but as though they were contemporaries... as it stands when a merge occurs one branch is assumed to be newer and just over writes	23:43
	id like to be asked about every change	23:44
RandalSchwartz	I look at it this way... for 30 years, from the time I was 16, I've paid every bill on time that was put in front of me	23:44
	← tatsuyao left	23:44
SinJax	does that make sense?	23:44
RandalSchwartz	for me to finally have to say "oops, screwed up", not bad.	23:44
gitte	RandalSchwartz: at least you have a contract now.	23:44
RandalSchwartz	and only in an era when millions of other americans are saying the same thing	23:44
	← cemerick left	23:44
RandalSchwartz	including many others with credit ratings like mine	23:44
gitte	RandalSchwartz: unfortunately, many people knowing your talents could not help you...	23:44
RandalSchwartz	well - I'm now working at a rate I have not seen since 1995	23:45
gitte	SinJax: define "change".	23:45
RandalSchwartz	but at least it's coming in :)	23:45
gitte	SinJax: oh, I just realize, that was asking for puns ;-	23:45
SinJax	gitte: ...every difference between the files	23:45
	gitte: merge with every difference treated as a conflict to be resolved	23:45
gitte	RandalSchwartz: you're still located in Portland?	23:45
SinJax	is that possible?	23:45
	← gaveen left	23:46
gitte	SinJax: everything's possible in Git.	23:46
	SinJax celebrates :D	23:46
SinJax	what should i read? :D	23:47
	→ altrux joined	23:47
gitte	SinJax: the question is if I can explain it in a comprehensible way before I drop into my bed.	23:47
	→ gaveen joined	23:47
SinJax	im already in my bed ;)...this might have to wait till tomorrow	23:47
gitte	SinJax: you can create a fake commit with the same tree as one of the both branch tips, then graft that commit as pretended merge base of both, and then merge.	23:48
	→ tatsuyao joined	23:48
gitte	SinJax: if that sounds like gibberish to you, I just proved my point... ;-)	23:48
SinJax	no no this sounds sane to me	23:48
	fake commit them both	23:48
	then fake merge them both	23:48
	then merge	23:48
	yes?	23:48
gitte	SinJax: well, the tricky parts are "graft" and "pretend"... ;-)	23:49
	SinJax: they are independent, you said, right?	23:49
SinJax	... they can be assumed to be yes	23:50
	im doing this all in branches so i don't mind experimenting :d	23:50
RandalSchwartz	I'm still in Portland, yes	23:51
	I'm working remotely for a company in Burbank though	23:51
	← jsbmsu left	23:51
	← hpoydar left	23:52
gitte	SinJax: the thing is: you'd have to make a _new_ commit (fully independent) to fake a merge base.	23:52
	← jrmuizel left	23:52
	→ qrush joined	23:52
SinJax	gitte: ok...ok that sounds fine, so how do i do that?	23:52
	i mean doing a git commit on either branch just says im up to date	23:53
	accurately!	23:53
gitte	SinJax: something like 'git symbolic-ref HEAD refs/heads/temporary-merge-base && git commit -m "faked merge base"'	23:53
	SinJax: it cannot be on either branch.	23:53
	SinJax: otherwise you would get into nasty loop problems, the "DAG" would be a "DCG" now.	23:53
SinJax	ah ok.... so make a new branch	23:54
	then	23:54
	hmm ill man page this symbolic-ref lark	23:54
gitte	SinJax: after creating that fake merge base, you'd need to add two lines "$COMMIT $FAKE_PARENT" with both branch tips as $COMMIT to .git/info/grafts.	23:54
	SinJax: after that, you'd have to check out the former branch (which was left with that cute and cunning symbolic-ref call)	23:55
	SinJax: and then you could "git merge <other-branch>" getting _a lot_ of merge conflicts.	23:55
	SinJax: you should remove the faked merge base as well as the grafts as soon as possible after resolving the merge conflicts.	23:55
	gitte is now conking out.	23:56
SinJax	ill	23:56
	look at this tomorrow	23:56
	thank you thought gitte	23:56
	i think im sleeping too ;)	23:56
	→ jrmuizel joined	23:56
SinJax	good night :)	23:56
	← SinJax left	23:56
	gitte is not a thought...	23:56
	← tatsuyao left	23:57
spearce	no, its an irc handle. :)	23:57
	← JohnnyL left	23:57
gitte	its a GMane bot.	23:58
	→ cemerick joined	23:58
	→ davi joined	23:58

Logs Search ←Prev date Next date→

Channels Documentation