IRCloggy #git 2009-05-06

Provider of IRC logs since 2005.

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

	2009-05-06
	→ ia joined	00:03
	→ Modius joined	00:06
	→ jerbear joined	00:07
	← lightcap left	00:09
	→ scarabx joined	00:10
	→ markelikalderon joined	00:12
	→ jrmuizel joined	00:13
	→ static^ joined	00:14
	→ wwalker joined	00:16
wwalker	How do I find out how far behind the master I am? git status tells me how far ahead of master my local work is, but doesn't tell me anything about the master (or any remote target)	00:17
	→ seangrove joined	00:18
	← felipec left	00:19
	← boto_ left	00:19
bdrewery	git log master..HEAD	00:20
	I think	00:20
	← WALoeIII left	00:21
	← wycats left	00:21
	← qrush left	00:22
	→ sgarza joined	00:23
	→ seangrove_ joined	00:24
	← sgarza left	00:26
	→ sgarza joined	00:26
	→ ericindc joined	00:26
	← seangrove left	00:27
	← sfwc left	00:29
	← jerbear left	00:29
	→ eshear joined	00:31
	← wshimmy left	00:33
	→ qrush joined	00:35
	← jrmuizel left	00:37
	→ faithfulgeek joined	00:39
	→ wycats joined	00:40
	← eshear_ left	00:40
	← faithfulgeek left	00:41
	→ jrmuizel joined	00:41
	→ faithfulgeek joined	00:45
	← sgarza left	00:46
	← Kbyte left	00:47
	← seangrove_ left	00:50
	→ seangrove joined	00:50
	→ MachX joined	00:51
	← seangrove left	00:54
	→ DrNick joined	00:57
	← wycats left	01:01
	← Dashkal left	01:02
	→ rettub_ joined	01:03
	← MachX left	01:03
	→ MachX joined	01:03
	→ Leemp joined	01:05
	← paulboone left	01:06
	← komplex left	01:07
	← poseidon left	01:10
	→ paltman94 joined	01:13
	→ catalis joined	01:14
	→ merlin83 joined	01:15
	← catalis left	01:16
	→ shadowhand joined	01:17
	← Tv left	01:17
	← markelikalderon left	01:17
	← lois[pac] left	01:20
	→ haesbaert joined	01:20
	→ seangrove joined	01:21
	← haesbaert left	01:26
	→ rworkman_ joined	01:27
	← shadowhand left	01:29
	rworkman_ → rworkman	01:30
	→ shadowhand joined	01:32
	→ JNRowe joined	01:33
	← dodo_the_last left	01:33
	→ montylounge joined	01:34
	→ naeu joined	01:37
	→ jony joined	01:37
	→ alezandro joined	01:40
	← naeu left	01:40
	← shadowhand left	01:42
	← rettub_ left	01:44
	→ rettub_ joined	01:44
	→ shadowhand joined	01:45
	→ okidogi joined	01:45
	← dreiss left	01:47
	→ arohner joined	01:47
	← MachX left	01:47
	→ gotgenes joined	01:48
bleything	so say I've got a branch that I've pushed upstream... and I've totally boned it locally with a rebase gone awry. how would I reset it to what's in the upstream repo?	01:51
	← ericindc left	01:51
	→ Tv joined	01:51
	← ivanoats left	01:53
	← jrmuizel left	01:53
	→ jepler joined	01:54
	← rmg- left	01:55
ThaDon	bleything: is it only one commit back you want to go?	01:57
	← texel left	01:57
bleything	well... I just want to undo the whole rebase.	01:58
	→ ericindc joined	01:58
	← qrush left	01:59
	→ octagon joined	02:00
	← faithfulgeek left	02:01
octagon	how do you remove an empty directory from git?	02:01
	← aziz left	02:02
Ilari	bleything: git reflog	02:03
ThaDon	bleything: well, I would say, assuming your remote is named origin: git reset --soft origin	02:03
Ilari	octagon: Just rmdir it?	02:03
bleything	Ilari: oh jeez. really? can I just delete the refs?	02:03
ThaDon	bleything: they anything left in the index that you want to go back to untracked status, just execute git reset (after you do the command above).	02:04
	bleything: backup before attempting, Idon't want you to screw anything up, but it sounds like what you want.	02:04
octagon	Ilari: it shows up as missing when i do "git status" if i rmdir it	02:04
Ilari	bleything: If rebase is still in progress abort it first (git rebase --abort).	02:04
	octagon: Probably its a submodule then, not a ordinary directory.	02:05
	→ gebi_ joined	02:05
	→ cemerick joined	02:05
	→ gretch joined	02:06
octagon	Ilari: i deleted the .gitsubmodules file with "git rm"	02:06
Ilari	octagon: Firstly, that doesn't delete the submodules. Secondly, there might be other submodules you just broke.	02:07
octagon	oh dear	02:07
bleything	cool, thanks for the tips! I'll try in a sec.	02:08
	→ AkumaStreak joined	02:08
	← cemerick left	02:08
	← shadowhand left	02:09
	← astrocub2 left	02:12
	← octagon left	02:12
	→ astrocub joined	02:13
	→ hobodave joined	02:15
	← FFForever left	02:17
	→ mw joined	02:18
bleything	hm okay. so I backed up my .git dir and then did git reflog delete HEAD@{0} until I was back at the point before I started the rebase.	02:19
	but that did not seem to help	02:19
Ilari	bleything: You just deleted the entries from log of past values. That doesn't do anything else.	02:19
bleything	oh.	02:19
	haha.	02:19
Ilari	bleything: Restore the backup...	02:20
	bleything: Specifically logs/HEAD	02:20
bleything	check.	02:20
Ilari	bleything: First, check that you aren't on (no branch). If you are, try git rebase --abort	02:20
	bleything: (git branch)	02:20
	→ jrmuizel joined	02:21
bleything	yeah, I'm in the branch in question.	02:21
	→ cytrinox joined	02:21
Ilari	bleything: Then find the correct entry from 'git reflog' and then 'git reset --hard <entry>' (destroys all uncommitted changes).	02:21
bleything	there we go.	02:22
Ilari	bleything: Where <entry> is 'HEAD@{<something>}'.	02:22
	← paltman94 left	02:22
bleything	is that it then?	02:23
	cpg → cpg\|away	02:24
Ilari	bleything: That should have undone the rebase.	02:24
	→ paltman94 joined	02:25
bleything	Ilari: it did indeed. thank you for your help... I feel like I should have been able to work that out on my own.	02:25
	Ilari: but I appreciate your patience :)	02:25
Ilari	bleything: Search 'Git for Computer Scientists' and 'Git from the Bottom Up'.	02:25
bleything	Ilari: will do.	02:25
	← qhoxie left	02:26
Ilari	bleything: They explain the low-level details that are useful to know... Really explains why user interface does things some way...	02:27
bleything	that's good. so far my reading has been most about how than why, and I think that I've reached the limit of what that can give me.	02:28
	er, how in terms of UI.	02:28
	← JNRowe left	02:30
	← paltman94 left	02:30
Ilari	bleything: And if you want to explore what's really in repo, start with 'git for-each-ref' and then 'git cat-file -p <ID>' the various object IDs (they appear as 40 hexdigits) you get...	02:31
	→ JNRowe joined	02:31
Ilari	bleything: Also, 'git ls-files -s' to dump contents of the index (a.k.a. staging area).	02:33
	→ dodo_the_last joined	02:36
bleything	awesome, thanks for the tips	02:37
	→ WALoeIII joined	02:39
	← flaguy48 left	02:39
	→ jay joined	02:39
	jay → Guest57928	02:40
	← JNRowe left	02:41
	→ jpeeler joined	02:46
	Guest57928 → JNRowe	02:47
	→ Jinzhu joined	02:48
	← jspiros left	02:49
	→ eno__ joined	02:49
	→ tjafk1 joined	02:52
	→ slestak\|1fk joined	02:52
	→ jspiros joined	02:52
	→ whaley joined	02:53
	← okidogi left	02:54
	→ zackattack joined	02:55
zackattack	how do I add /usr/local/git/bin/ to my PATH in mac os x?	02:55
	→ serfurj joined	02:57
Risar	zackattack, did you use the osx installer from code.google.com ?	02:57
zackattack	Risar: yes	02:58
	← slestak\|afk left	02:58
Risar	http://code.google.com/p/git-osx-installer/issues/detail?id=24	02:59
	check that out.	02:59
	← WALoeIII left	03:00
Risar	I'm not a mac user, but the jorb requires it and IIRC I had a similar issue that I believe that fixed.	03:00
zackattack	exactly what i wanted thanks Risar	03:02
Risar	np	03:03
	→ gwoo_ joined	03:06
	→ context joined	03:07
	→ nis joined	03:08
	→ faithfulgeek joined	03:08
	cpg\|away → cpg	03:15
context	can i add a branch to track on an already created branch	03:16
patrikf	context: what do you mean by tracking?	03:16
context	for git merge and pull	03:17
	→ paltman94 joined	03:18
context	nm i jsut removed them and readded them :x	03:18
	hmm nm	03:19
	not what i wanted	03:19
nis	context: git checkout -t -b new_branch tracked_branch	03:20
	→ zackattack_ joined	03:21
context	yeah	03:21
Ilari	context: The settings are 'branch.foo.remote' and 'branch.foo.merge'.	03:21
	← Travis-42 left	03:23
	← zackattack_ left	03:23
	← montylounge left	03:24
	← faithfulgeek left	03:27
	← mutex left	03:28
	← porcnick left	03:28
	← rworkman left	03:34
	← zackattack left	03:36
	→ sgrimm joined	03:39
	→ orafu joined	03:40
	→ schlort joined	03:40
	← sgrimm left	03:41
	← scarabx left	03:41
	← JNRowe left	03:41
	← whaley left	03:49
	→ arohner_ joined	03:49
	→ cedricv joined	03:50
	→ whaley joined	03:53
	→ nis_ joined	03:55
	← patrikf left	04:01
	→ ankit9 joined	04:01
	→ ivanoats joined	04:02
	→ jrmuizel_ joined	04:03
	← jrmuizel left	04:03
	→ fujin joined	04:04
	← arohner left	04:06
	← nis left	04:06
	→ fridim_ joined	04:07
	→ qhoxie joined	04:08
	← Modius left	04:21
	→ mfilipe joined	04:23
	← fridim_ left	04:23
	← whaley left	04:23
	→ eno joined	04:26
	← paltman94 left	04:29
	→ bdiego joined	04:29
	→ justatheory joined	04:31
	→ Alinermedia joined	04:31
justatheory	How do I pull a commit from one repo to another? I just want a single commit.	04:32
mugwump	fetch, then cherry-pick	04:32
justatheory	mugwump: How do I fetch a single commit?	04:32
mugwump	you can only fetch refs advertised by 'git ls-remote URL'	04:32
	justatheory wonders how GitHub does it for its merge queue…	04:33
	→ Modius joined	04:33
mugwump	well, it doesn't need to fetch - all the repos are on the same system	04:33
justatheory	I have repos on the same box.	04:33
mugwump	there's a guy at github who's had his hand in 3 git pure implementations in other languages by my count	04:34
justatheory	ah	04:34
	So I did the fetch	04:34
	Now how do I cherry-pick it?	04:34
	(Sorry to be so dense0	04:34
mugwump	git cherry-pick commitid	04:35
	or use gitk, right-click the commit you want and 'cherry pick this commit'	04:35
	you'd need gitk --all to see the remote refs	04:35
	→ paulboone joined	04:35
justatheory	Got it.	04:35
	Kind of mysterious	04:35
	That is, where the fetch puts stuff.	04:35
mugwump	pre-1.5 they were all intermingled in the same branch namespace	04:36
justatheory	So now that I've cherry-picked it, I can push it; but what about anything else that the fetch, um, fetched?	04:36
mugwump	well, push only pushes what you specify	04:36
justatheory	Yeah, I just cherry-picked one commit	04:36
	and then pushed it.	04:36
	So where is the stuff that fetch pulled down?	04:36
mugwump	use 'git branch -a'	04:36
	← eno__ left	04:37
mugwump	which is really just a pretty form of 'git show-ref'	04:37
justatheory	oh. Okay.	04:37
	Git is nice, but it's not as easy to learn as svn was, I tell you.	04:37
	thanks for the help mugwump	04:37
	← ericindc left	04:37
mugwump	:P enjoy	04:38
	bet you didn't do the same with svn	04:38
justatheory	no	04:38
	I could think of SVN as a kind of big file system	04:38
	And then use that to work around its…deficiencies.	04:38
mugwump	git is probably just as simple	04:38
justatheory	And I know that I can always get pointers here.	04:39
mugwump	there's a couple of essays describing it, 'git for computer scientists', and 'git from the bottom up'	04:39
justatheory	Now to get my jQuery question answered…	04:39
mugwump	also the user manual.	04:39
justatheory	yeah yeah	04:39
mugwump	worth reading ... at least one of those ;)	04:39
justatheory	I've read a fair bit, and will read more as time goes on.	04:39
	ta	04:39
	← jrmuizel_ left	04:45
	→ tbf\|anniversary- joined	04:45
	← mw left	04:47
	← dec left	04:47
	→ bentob0x joined	04:52
	← arohner_ left	04:52
	← Leemp left	04:53
	← bdiego left	04:55
	← mstrl left	04:56
	→ meyering joined	04:57
	→ Dreamer3 joined	05:00
Dreamer3	anyone using p4merge with git?	05:00
	→ paltman94 joined	05:00
	→ curvature joined	05:01
	← altrux left	05:01
	→ girishr joined	05:03
	← cedricv left	05:04
	→ wshimmy joined	05:04
	→ pastie joined	05:04
	← paltman94 left	05:07
	→ cedricv joined	05:08
	→ paltman94_ joined	05:08
	offby1 watches the tumbleweeds skitter down Main St	05:11
	→ Weaselweb joined	05:11
	← Alinermedia left	05:26
	← girishr left	05:26
	→ Alinermedia joined	05:26
	← justatheory left	05:28
	→ jceb joined	05:29
	tbf\|anniversary- → tbf	05:30
	← paulboone left	05:32
	→ vincent joined	05:34
	vincent → Guest82580	05:34
	→ SRabbelier joined	05:39
	→ texel joined	05:40
	← mfilipe left	05:41
	→ mutex joined	05:42
	→ struberg joined	05:43
	→ mikem joined	05:46
	← cedricv left	05:47
mikem	is there a command that simply spits out the sha1 of the current HEAD? I can do `cat .git/$(cat .git/HEAD \| sed "s/ref: //")` but maybe there's a built-in command?	05:47
nis_	mikem: git rev-parse HEAD ?	05:49
mikem	nis_: great, thanks :)	05:50
nis_	np	05:50
	← Alinermedia left	05:50
	← stuffcorpse left	05:51
	→ porcnick joined	05:51
	→ stuffcorpse joined	05:51
	→ naeu joined	05:52
	← johan-s left	05:54
	→ skyl joined	05:56
skyl	removing the .git dir and I can start from git init again?	05:57
nis_	skyl: yes.	05:57
	← paltman94_ left	05:58
	→ ljungk joined	05:59
	← joss left	06:00
	← Guest82580 left	06:02
	← ankit9 left	06:03
	← ivanoats left	06:05
	→ Ziban\|afk joined	06:09
	→ reprore joined	06:09
Ziban\|afk	http://change.menelgame.pl/change_please/2688554/	06:11
	← Ziban\|afk left	06:12
	→ priidu joined	06:14
	← texel left	06:15
	← stuffcorpse left	06:15
	→ stuffcorpse joined	06:15
	→ cydork joined	06:16
	→ psankar joined	06:17
	← seangrove left	06:19
	← naeu left	06:19
	→ agenteo joined	06:19
	← agenteo left	06:19
	→ johan-s joined	06:21
	→ Grahack joined	06:24
	→ mtkd joined	06:24
	← gilimanjaro left	06:26
	→ mediogre joined	06:26
	→ doener joined	06:26
	← FreakGuard left	06:28
	← gotgenes left	06:31
	→ kadoban joined	06:32
	→ eddyp joined	06:33
	→ giallu joined	06:37
	→ tal67 joined	06:38
	→ galderz joined	06:39
	→ bschindler joined	06:43
	→ josesito joined	06:47
	→ ankit9 joined	06:47
josesito	hi all, if i wanted to force push to a remote repo (although the repo has some commited or uncommited changes) how could I do it?	06:47
	(i just want my remote repo to take my local repo as the HEAD i don't want to pull first)	06:48
thiago_home	git push -f remotename what:you want:to push	06:49
	but don't push to a non-bare repository	06:49
josesito	non-bare?	06:50
Ilari	josesito: $faq non-bare	06:51
Gitbot	josesito: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare	06:51
josesito	also what:you want:to push is supposed to be path to a file or can it be a branch?	06:51
thiago_home	it's supposed to be source-ref:destination-ref	06:51
	as in: master:master	06:51
	or HEAD~:master	06:51
Ilari	josesito: Or just 'source-ref' if both sides are the same.	06:51
thiago_home	the standard syntax for push, which you should familiarise yourself with anyways	06:52
josesito	so if i'm currently on branch testing a git push -f would testing --> testing right?	06:52
	(yeah, i'm still learning)	06:52
doener	no	06:52
	the default refspec for push is ":", which means "push branch heads that exist in this and in the remote repo"	06:53
josesito	oh, so it will push all local branches to remote matching branches?	06:53
	← reprore left	06:53
	→ reprore joined	06:54
Ilari	josesito: Only those that have matching ones on remote side...	06:54
josesito	oh ok	06:54
	thank you	06:55
thiago_home	if you've just created "testing", then it won't be pushed	06:55
	because it won't exist on the other side	06:55
josesito	how can i create it on the other side then?	06:56
thiago_home	push it:	06:56
	git push remotename testing	06:56
	→ lohrun joined	06:57
josesito	oh ok	06:58
	so git push won't create a branch unless i explicitly tell it to	06:59
thiago_home	by default	06:59
Ilari	josesito: That's pretty much the idea... In case you have branches you want to push (and that comes up often).	06:59
thiago_home	you can change the default too	06:59
	→ reel_ joined	07:00
josesito	nah the default is great	07:00
	more control	07:00
Ilari	Yeah, do the right thing by default 99 times out of 100 and offer manual overrides for the 1 time out of 100...	07:03
	thiago_home prefers the default to be "HEAD" instead of ":"	07:06
	← struberg left	07:06
	→ struberg joined	07:07
Ilari	The usual forms of refspecs: ':' (push matching branches), 'HEAD' (push current branch), 'branchname' (push branchname), 'tagname' (push tagname), 'refname' (push refname), ':branchname' (DELETE branchname from remote), ':tagname' (DELETE tagname from remote), ':refname' (DELETE refname from remote), 'branchname1:branchname2' (push branchname1 as branchname2), 'tagname1:tagname2' (push tagname1 as tagname2, don't use for annotated/signed tags)	07:09
	'tagname1:tagname2' (push tagname1 as tagname2, don't use for annotated/signed tags), 'refname1:refname2' (push refname1 as refname2), 'committish:refname' (push committish as refname).	07:09
doener	and "tag tagname" push tagname, the "tag" prefix disambiguates in case that you have a branch head with the same name	07:10
	s/push tagname/(push tagname)/	07:10
Ilari	And all of those may be prefixed with '+' to force that update. Also in refnames, the last components on both sides may be '*' to push entiere namespace.	07:11
	→ _slacker_ joined	07:15
Ilari	For push, it has that 'default right side to left side'-behaviour. That's not true for fetch. Fetch defaults right side to FETCH_HEAD.	07:17
	Also, fetch doesn't allow left side to be comittish.	07:17
	← jceb left	07:19
	← eddyp left	07:20
	→ hyperair joined	07:22
	→ Stravy joined	07:24
	→ ph^ joined	07:25
	→ jmesnil joined	07:26
	→ naeu joined	07:28
	→ drizzd joined	07:30
	← Radar left	07:31
	← krawek left	07:31
	← TheDJACR left	07:32
	← kuadrosx left	07:33
	← thiago_home left	07:33
	→ Zaba_ joined	07:34
	→ RodP2 joined	07:36
	← cydork left	07:37
	→ seangrove joined	07:38
skyl	empty dirs are not tracked?	07:38
Stravy	skyl: no	07:38
	→ bizhat2 joined	07:40
	→ wycats joined	07:41
	→ priidu_ joined	07:42
	→ knitt1 joined	07:43
	→ JNRowe joined	07:43
	← priidu left	07:44
	→ gilimanjaro joined	07:44
	← JasonWoof left	07:45
	← bobesponja left	07:45
	← reel_ left	07:47
	→ agathon23-5 joined	07:49
	← bizhat2 left	07:50
	→ shruggar joined	07:50
agathon23-5	hello i would like to commit to the remote server with "git push". It tells me i am already up to date which can't be... in the past i had to checkout an older revision so is this maybe the problem?	07:51
	→ khmarbaise joined	07:54
	→ flavien joined	07:57
flavien	Hi. I have a patch that I would like to split into several files named hunk-0.patch, hunk-1.patch, etc.	07:57
	I'm sure a perl guru has written something like that already. Can't find it though.	07:57
	→ pan1nx joined	07:59
	→ fujin_ joined	07:59
	→ smeevil joined	08:00
	← ljungk left	08:00
	← fujin left	08:01
killerchicken__	flavien: what about doing an interactive rebase, editing the one commit, using git add -i (or -p) to selectively add some hunks, commit, add some more hunks, commit again, etc until you have it split the way you like it?	08:01
	← naeu left	08:01
josesito	what does "No submodule mapping found in .gitmodules for path foo/bar" mean? and how can i resolve it?	08:02
flavien	killerchicken__: in this context, I'm not using git. :-(	08:03
killerchicken__	flavien: heh. no idea, then.	08:03
wereHamster	agathon23-5 on which branch are you right now (git branch)?	08:03
flavien	Foudn -U option for patch, and it does the job.	08:04
	bye	08:04
	← flavien left	08:04
agathon23-5	i-wlan369:src nils$ git branch	08:04
	107924a3492ef30bc5649ddfe72384377d8d8bcd	08:04
	1fd69c8de6dd816a5e1f1bf4c975564ac301bdef	08:04
	* master	08:04
	→ Bleupomme joined	08:06
pan1nx	hi, anybody knowing the insights of gitosis?	08:06
agathon23-5	before i entered the command "git checkout master" was was part of my "not knowing what to do, fix it yourself" commands :(	08:06
wereHamster	agathon23-5: does the branch 'master' already exist on the remote repo?	08:06
Bleupomme	hello, could someone tell me how to do an equivalent of the svn export? Just remove the .git directory?	08:06
agathon23-5	how can i determine that?	08:07
	← tbf left	08:07
Arrowmaster	Bleupomme: git archive	08:07
	→ tbf joined	08:07
Bleupomme	Arrowmaster: thx	08:07
wereHamster	agathon23-5: have you already pushed master in the past?	08:07
agathon23-5	yes	08:07
	git push origin master and then always git push	08:08
wereHamster	agathon23-5: check with git log whether the commits you did are indeed in your master branch\	08:08
	← tbf left	08:08
	→ tbf joined	08:09
agathon23-5	wereHamster: git log shows me the remote stuff which is not "up-to-date" with my local commits	08:10
wereHamster	'git log' shows your current branch (master) in your case.	08:11
eMBee	good afternoon	08:12
wereHamster	you probably commited on a detached HEAD. Are 107924 and 1fd69c the commits you did and want to push to the remote?	08:12
eMBee	git push fails with "fatal: Unable to create temporary file: Permission denied", seems something wrong whith the setup on the server. but how can i find out what?	08:13
	→ Mjollnir` joined	08:13
agathon23-5	wereHamster: yes i think the 1fd69c is the latest in my gitx	08:13
	→ tjh238 joined	08:14
madduck	is there a way to identify linenumbers of lines matching a regexp which I have edited between two refs?	08:14
	Mjollnir`: hello! ;)	08:14
Mjollnir`	madduck: :)	08:14
doener	madduck: so you have two commits and a regexp, and you want what?	08:15
	tjh238 pricks up ears	08:15
wereHamster	agathon23-5: git checkout master; git merge 1fd69c; then check the history with 'git log' to make sure the master branch indeed contains everything you want; git push	08:15
	→ naeu joined	08:15
madduck	doener: match the regexp against all lines edited between the two commits and output the file+line number pairs	08:15
Mjollnir`	doener: to be able to replace , for example, all instances of TODO in the codebase with TODO bugnumber , but only in lines changed between two commits	08:16
	doener: line numbers seems like a sensible starting point	08:16
pan1nx	tv, are you there?	08:16
	tv, I was asking if there is anybody knowing the insights of the gitosis, i guess you should be the answer? right?	08:17
	Tv?	08:18
	→ j416 joined	08:18
	← Tricon left	08:18
	→ DavidKlein joined	08:20
	→ FreakGuard joined	08:20
	← johan-s left	08:21
eMBee	ah, pebkac, the screen session i was running in was started before i joined the git group. restarting screen fixed the problem	08:22
agathon23-5	wereHamster: hmm the remote has only one revision updated and skipped all my others, that's bad	08:24
	← Grahack left	08:25
agathon23-5	how can i include all my other revisions on the remote machine since i already skipped them ... ?	08:27
madduck	patience...	08:28
	oops	08:28
	← tjh238 left	08:29
	→ agathon23-5_ joined	08:29
	← nis_ left	08:29
	→ cilly joined	08:30
wereHamster	agathon23-5: did you check with git-log or gitk whether your local master branch contains everything you want?	08:30
agathon23-5_	no	08:31
	stupid me	08:32
	→ gb\|work joined	08:32
gb\|work	hi there, how can i revert a bunch of files from the master repository if i accidentally deleted local files?	08:33
	please help	08:33
wereHamster	alright. Another idea: use the reflog (git log -g) and find the very last commit you made and want to be your HEAD. git checkout master; git reset --hard $shaofsaidcommit; [check git-log/gitk]; git push	08:33
	gb\|work: did you use git-rm or just rm?	08:33
	← fujin_ left	08:35
killerchicken__	Hey, I've set up a gitweb install to do some load test. Doing 10 requests in parallel is already a big thing for the machine, and since I only installed the available lenny package of gitweb, I wonder if newer versions do any better?	08:35
	Or is there another thing I can consider?	08:35
wereHamster	killerchicken__: cgit	08:35
gb\|work	wereHamster: just rm	08:35
wereHamster	gb\|work: git checkout -- path/to/files	08:36
killerchicken__	wereHamster: hm. I hoped to avoid that, I think the interface doesn't look as nice as gitweb	08:36
gb\|work	that's what i'd do normally, but if there is a ton of files gone	08:36
	→ tango_ joined	08:36
gb\|work	can i give it a wildcard?	08:36
wereHamster	gb\|work: have you any uncommited changes in the working tree?	08:36
	→ huyx joined	08:38
	→ girishr joined	08:39
gb\|work	i don't care about uncommited changes, i just want my code back :)	08:39
wereHamster	gb\|work: try git checkout -- .	08:40
	← huyx left	08:40
gb\|work	ok, will do, thx	08:41
	→ fujin joined	08:42
	← Bleupomme left	08:43
	← pan1nx left	08:44
	← agathon23-5 left	08:45
	→ pan1nx joined	08:49
	→ tono joined	08:49
	→ vals_ joined	08:49
pan1nx	Tv, can I ask you a question about gitosis?	08:50
	← tango_ left	08:50
	vals_ → tango_	08:50
	→ tttssstt` joined	08:51
	→ solydzajs joined	08:52
	← yec left	08:54
	→ cdleonard joined	08:58
Ilari	pan1nx: Ask anyway, in case somebody else could also answer it...	08:58
	← josesito left	08:58
pan1nx	is there a way to get the ssh-key comment (what is basically the user in the gitosis) out to be treated in the git-hooks?	08:59
	basically, I want also to compare the user name from the gitosis with the commiter name from the commit, if possible...	08:59
	← cdleonard left	09:02
Ilari	pan1nx: 1) You need to modify gitosis code (it is as python variable 'user' in one point of code in serve.py). 2) That comparision is probably bad idea unless you have some very restrictive environment.	09:04
	→ tttsssttt joined	09:04
	→ gitte joined	09:04
Ilari	pan1nx: The reason why that comparsion shouldn't be done is that in Git there are ways (especially in public projects) that pretty much arbitiary names may wind up as comitters (of even the head commit pushed) without anybody misrepresenting anything.	09:04
	← Dreamer3 left	09:04
pan1nx	Ilari, I need that restrictive environment	09:05
	→ dwave_ joined	09:05
	← JNRowe left	09:06
pan1nx	Ilari, it is only when a special action is triggered, I need to check that the gitosis user is the one that does the action...	09:06
Ilari	pan1nx: Basically, such environment needs to restrict that everybody that can fetch the repo can also push to it.	09:06
pan1nx	well, not necessary. You can restrict only one action, let's say, pushing a certain tag, or inside a special branch...	09:07
Ilari	pan1nx: Well, you can authenticate ref ops just fine. What is pretty much no-no is checking log comitters...	09:08
pan1nx	I mean, the current way, if you have right permissions, and you don't sign your commits, there is no way that you can guarantee who did what, because users can just type git config user.email/name and change their identity	09:08
	Ilari what do you mean by ref ops?	09:09
Ilari	pan1nx: That's not the only way strange comitters can apper.	09:09
	pan1nx: Creating ref, deleting ref, updating ref.	09:09
	pan1nx: And tags and branches are refs, so...	09:09
	pan1nx: E.g. Some contributor makes a change. Sends pull request to dev. Dev fetches the commit and looks at it. It looks fine, so the dev merges it. Now the dev doesn't have any unpushed changes and the contributor based the change on latest version. So the merge winds up being fast forward. Now the dev is up to date w.r.t. main repo so push suceeeds. The pushed commit (correctly) claims to come from contributor.	09:10
pan1nx	Ilari yes, but how to find a way to identify the gitosis user vs the ref ops	09:10
Ilari	pan1nx: You need to modify gitosis to export that info.	09:10
	→ Bleupomme joined	09:10
	← tttssstt` left	09:11
pan1nx	Ilari that is fine, the tags are treated different in the push, so you push them with --tags	09:11
Bleupomme	hello. I would like to checkout (clone) a branch and then do an equivalent of svn export to a new directory. Could someone help me?	09:11
pan1nx	the dev can push only the tags that it wants	09:11
	→ smtms joined	09:12
Ilari	pan1nx: Well, tags with odd taggers are usually rarer than commits with odd comitters.	09:12
pan1nx	Bleupomme, did you check the stack overflow? http://stackoverflow.com/questions/160608/how-to-do-a-git-export-like-svn-export	09:13
	→ moonflux joined	09:13
pan1nx	Ilari, except for me, the commits don't matter, but the tags are handled in the server as actions	09:13
Bleupomme	pan1nx: yes, but the problem is I don't understand how to do it with a branch	09:14
Ilari	pan1nx: That is, who does the tag is usually who pushes it too...	09:14
pan1nx	Ilari, hence my need to get the gitosis user	09:14
Ilari	Bleupomme: See 'git branch -r'. You can pass those to git archive	09:14
pan1nx	Bleupomme, just checkout that branch	09:14
	→ dwave- joined	09:14
Ilari	pan1nx: In the latest version from mainline repo, looks like suitable place to insert variable export is between lines 202 and 203 of serve.py	09:15
pan1nx	Ilari, but other than changing gitosis nobody needed this?	09:16
	← dwave- left	09:16
	← dwave left	09:16
	→ markelikalderon joined	09:17
Ilari	pan1nx: There have been few requests... Unfortunately it seems Tv is quite busy...	09:17
pan1nx	at some point there was something like GITOSIS_USER	09:17
Ilari	pan1nx: Google doesn't find any reference to it...	09:18
shruggar	why does git log --stat not work with --relative?	09:19
pan1nx	Ilari, there is even a question in http://stackoverflow.com/questions/824990/git-gitosis-how-to-check-validity-of-user-name-and-email	09:20
	similar...	09:20
	→ jasiek joined	09:21
Ilari	pan1nx: There's also one fork of gitosis. Maybe they could accept that change and eventually it would end up being merged back?	09:22
	→ aspotashev joined	09:22
	→ ceyusa joined	09:23
	← tbf left	09:23
	→ tbf joined	09:23
	→ AAA_awright joined	09:24
	→ jasiek_ joined	09:24
	← dwave_ left	09:24
	→ Radar joined	09:25
jasiek_	hey, im trying to share a repo via ssh. will git automatically manage file ownership and permissions on files within the repo?	09:25
	← tono left	09:26
Ilari	jasiek_: There's option to override umasks for repository files. Then there is Unix feature of setgid directories...	09:26
	→ tono joined	09:26
	← struberg left	09:26
pan1nx	Ilari, thanks, I will look into it	09:26
	→ struberg joined	09:28
Ilari	pan1nx: In Gentooo Gitosis fork, the corresponding lines are 223 and 224.	09:28
	← markelikalderon left	09:28
jasiek_	thanks	09:29
drizzd	jasiek_: mkdir project.git; cd project.git; chown .mygroup .; chmod g+s .; git init --bare --shared=group	09:30
	← AkumaStreak left	09:30
	→ Octalot joined	09:30
	← reprore left	09:30
Ilari	drizzd: Does that work? I remember trying that once and it didn't seem to give sane results. Of course it might have been fixed since...	09:32
	→ Krintus joined	09:32
Krintus	íà ôîðóì http://forum.forok.ru òðåáóþòñÿ admin è ìîäåðàòîðû ðàçäåëîâ. Âñå êòî çàèíòåðåñîâàí ïèøèòå â ëè÷êó	09:32
	ChanServ set mode: +o	09:32
	gitte set mode: +b	09:32
	gitte kicked Krintus (User terminated!)	09:33
	← bschindler left	09:33
	← Radar left	09:33
Ilari	drizzd: Appears to work in latest version...	09:33
	→ Radar joined	09:34
	← wycats left	09:34
pan1nx	Ilari, i just added os.environ['GITOSIS_USER'] = user and it works :D	09:35
	→ Kbyte joined	09:35
	← j416 left	09:36
	← aspotashev left	09:36
	→ bschindler joined	09:36
Ilari	pan1nx: Maybe make a patch out of it and send it upstream? :-)	09:36
	← jasiek left	09:37
	← ia left	09:39
Ilari	pan1nx: Regarding checking tags. Be aware that simple tags don't have tag objects and thus no tagger name. Also, somebody pushing tags they got from third party (after checking of course) is very rare...	09:40
	→ ia joined	09:40
pan1nx	yes... Ilari, you are right...	09:42
	→ marc_m joined	09:42
	← Stravy left	09:44
Ilari	pan1nx: 3rd party commits winding up into push is quite common. 3rd party commits winding up as pushed branch heads sometimes happens... git.git has 24 people marked as comitter of at least one commit. AFAIK, only 3 of them have held push priviledges to the "main" repo.	09:45
	pan1nx: Oops, aparently 23, not 24...	09:47
	→ wycats joined	09:48
drizzd	Ilari: what do you mean doesn't work? The --shared option?	09:48
	Ilari: the fact that you need to do "chmod g+s ." is a bug IMO	09:48
	I have to admit I never actually use shared repos, so I don't know if it really works.	09:49
	← girishr left	09:49
Ilari	drizzd: IIRC, when I tried to first give . proper permissions and then invoke init --bare --shared (I don't remember what version), some stuff there got wrong groups... No idea how that happened.	09:50
	drizzd: With latest version, it works.	09:50
drizzd	Ilari: ok, good to know	09:52
Ilari	drizzd: That is, some stuff there where owned by user primary group (wrong) and some by directory group (correct).	09:52
drizzd	It's one of those rarely used features I suppose	09:52
	Ilari: that's exactly what happens now if you do not set g+s on the repository root directory.	09:53
	← Kbyte left	09:56
	→ aziz joined	09:56
Ilari	drizzd: Scanning the logs, I found one change that might have to do with it. First release it is in is 1.6.2.3...	09:57
	← priidu_ left	09:58
Bleupomme	when I try to archive a repository I get following error http://gist.github.com/107464 . Please consider I'm a git newbie	09:58
	← cilly left	09:59
Ilari	Bleupomme: Specify the branch you want on command line too.	09:59
	← mikem left	09:59
Ilari	Bleupomme: 'git archive --remote=... foobranch'.	09:59
Bleupomme	Ilari: thx	10:00
drizzd	Ilari: do you mean	10:00
	5a688fe4	10:00
Gitbot	[git 5a688fe4]: http://tinyurl.com/ca5fz3 -- "core.sharedrepository = 0mode" should set, not loosen	10:00
Ilari	Bleupomme: Not very nice that it can send invalid requests (complete with error messages from those) if you get the syntax wrong...	10:00
	drizzd: Yes.	10:00
	drizzd: Running --shared without g+s on parent dir produces quite trippy results. objects gets parent dir owning group but refs doesn't.	10:02
	drizzd: Wonder how that happened...	10:02
	→ priidu joined	10:04
	→ Zaba joined	10:04
	→ mikl joined	10:04
	→ quoin joined	10:05
	→ intripoon joined	10:06
drizzd	Ilari: right, probably because refs/heads is created before shared_repository is configured	10:07
	because git does set g+s . at some point	10:07
	→ harinath joined	10:11
	→ bizhat joined	10:12
	← harinath left	10:12
	→ harinath joined	10:13
	← Jinzhu left	10:13
Ilari	drizzd: Maybe it should set it before creating any stuff there so one wouldn't need to manually g+s it?	10:14
	← Zaba_ left	10:16
	→ riovena joined	10:17
	← bizhat left	10:17
drizzd	Ilari: sounds reasonable	10:18
	→ bizhat joined	10:18
	← harinath left	10:19
	→ ssD joined	10:24
	← ssD left	10:25
	← gretch left	10:26
	→ kn1ttl joined	10:28
	← knitt1 left	10:30
	→ johan-s joined	10:33
	→ cilly joined	10:37
	→ harinath joined	10:37
	← jasiek_ left	10:38
	→ ericindc joined	10:39
	→ bizhat1 joined	10:39
	← bizhat1 left	10:40
	→ foutrelis joined	10:40
	→ kn1tt1 joined	10:43
	→ dwave joined	10:45
	← bizhat left	10:45
	→ sunoano joined	10:45
	← ericindc left	10:46
	→ auxbuss joined	10:48
	→ j416 joined	10:48
	→ Weaselweb_ joined	10:48
j416	if I tag something in a temporary branch, and then remove the branch	10:49
	what is tagged will never get removed, right?	10:50
	as long as its tagged	10:50
Grum	is there an easier way to make any commit (thus creating master) in a bare repo other than: git --work-tree=. --git-dir=/path/to/bare/repo.git --allow-empty -m "bogus commit" ?	10:50
	eerrrm add 'commit' after 'repo.git ' :)	10:50
	← cilly left	10:52
	→ commondream joined	10:52
	→ cilly joined	10:52
	→ darrob joined	10:53
	→ djinni joined	10:53
tango_	Grum, clone the bare repo, commit, push	10:55
Grum	which i dont want to do :P	10:55
killerchicken__	"I want to drive, but I don't want to use a car. Please help"	10:57
	← Weaselweb left	10:57
Grum	no, i want to know if there is an easier way	10:57
	'clone the repo commit push' is not easier	10:58
	Weaselweb_ → Weaselweb	10:58
	→ Kbyte joined	10:59
	← kn1ttl left	10:59
	→ cedricv joined	11:01
	→ ringlej joined	11:07
	← mithro left	11:07
	→ parasti joined	11:09
	→ jasiek joined	11:10
	← ronny left	11:12
	→ drewr joined	11:12
	← auxbuss left	11:12
ringlej	I have a merge commit with conflicts resolved as part of that commit. I can't seem to get 'git format-patch' to generate a patch of that commit though.	11:12
	→ auxbuss joined	11:13
	→ GyrosGeier joined	11:17
	→ TomK32 joined	11:19
TomK32	how can i change the timestamp for a commit?	11:20
	oh wait, google knows	11:21
eMBee	good evening	11:22
TomK32	hi eMBee	11:22
eMBee	when i create a bare cline, is there an easy way to upgrade it to a mirror?	11:23
	← tbf left	11:24
	→ Jinzhu joined	11:24
	← ringlej left	11:25
	← Jinzhu left	11:25
	← struberg left	11:25
eMBee	clone	11:26
	→ tbf joined	11:26
	← porcnick left	11:27
	→ ustedes joined	11:27
	→ paltman94 joined	11:28
	→ ringlej joined	11:30
	← jony left	11:30
Ilari	eMBee: There are some settings for it. See man git-config	11:31
Gitbot	eMBee: the git-config manpage can be found at http://git.or.cz/man/git-config	11:31
	← ankit9 left	11:31
eMBee	well, i know how to set the config manually, just wondered if there was a shortcut	11:31
	since its a local clone i can just copy from the other repos config file	11:32
	← darrob left	11:33
	→ ustedes_ joined	11:35
	← Radar left	11:35
	← paltman94 left	11:37
	→ cemerick joined	11:40
	→ ustedes- joined	11:41
	→ struberg joined	11:42
	ustedes- → whaley	11:42
eMBee	actually, i don't need a mirror, i am trying to create a shared repository that everyone pushes their work into, but the work is based upstream stuff. what is the best way to get upstream changes into the shared repo? have any user get upstream directly and then push it?	11:43
	→ juanjoc joined	11:44
Rhonda	eMBee: I think there are pull triggers possible.	11:45
	← marc_m left	11:45
	→ eshear_ joined	11:46
Rhonda	eMBee: CosmicRay does something along that path for his bacula git packaging and pulls from my backports branch regularly. Propably through cron.	11:47
eMBee	you mean triggers that will initiate an update? (can't really be pull because there is no workdir	11:48
	← eshear left	11:49
	→ reprore_ joined	11:50
Rhonda	Something along that lines propably. I just know he fetches my commits to my backports branch into his bacula stuff automatically from what I understand.	11:51
eMBee	hmm	11:51
	→ aspotashev\|eeepc joined	11:51
Rhonda	Need to prepare a patch for him in his own branch it seems, he forgot that dpkg can't change a directory to a symlink and needs maintainer script magic there. %-/	11:52
	→ Leemp joined	11:52
	→ mbroeker joined	11:54
	← reprore_ left	11:55
	← auxbuss left	11:57
	→ frando joined	11:57
	→ auxbuss joined	11:58
	→ cannonball joined	11:58
frando	hey. I need to move one branch of a git repo into a svn repo (I so far developed on a project using git only, but now my organization requires me to move the code incl. revisions into a central svn repo). What is so far in the git repos root folder has to be in one branch of the svn repo, say, dev/project1. I would like to continue using git with git-svn and git-svn-dcommit. My question is - how would I move the existing git repo (one branch only	12:00
	for now) including all revisions into a specific branch of the (already existing, but mostly empty) svn repo?	12:00
	← ustedes left	12:00
	← ustedes_ left	12:00
	→ aspotashev joined	12:00
	← tal67 left	12:02
drizzd	frando: I don't know exactly. Does this branch have a linear history?	12:03
frando	drizzd: yes. or, it includes some merges, but I don't have to track these	12:04
	linear history is enough. svn doesn't have anything else, anyways	12:04
eMBee	well, you need to remove the merges before git-svn can handle the branch. the merges need to appear as simple commits	12:05
	← aspotashev left	12:05
drizzd	frando: if you have merges, the history is not linear. Linearizing may not be trivial.	12:05
	→ aspotashev joined	12:05
frando	so i was thinking in the lines of git svn init on the svn repo, then have an "svn" branch in the newly created git repo point onto the right svn branch, then fetch my existing git repo into the new git repo, and then rebase the "svn" branch on the branch I want to move to svn and then svn dcommit. does that sound more or less logical?	12:06
	how do you remove merges?	12:06
eMBee	i think git-rebase ignores merges. but what does it do instead? treat the merge commit as a plain commit? or skip it completely?	12:07
drizzd	frando: I think dcommit will try to do that, but I don't see how to dcommit to a different branch. hmm	12:07
	→ tal67 joined	12:08
drizzd	eMBee: I think it tries to do a rebase instead of the merge.	12:08
	← dwave left	12:08
eMBee	well, the doc says -p: Instead of ignoring merges, try to recreate them.	12:09
drizzd	the manpage says it "ignores" them, but which strain of the history does it follow then?	12:09
eMBee	which doesn't sound like it would be rebasing the merged branch	12:09
drizzd	maybe it just does ^1^1^1^1...	12:09
	eMBee: you're right, my bad	12:09
eMBee	that is a good question	12:09
frando	hmm.. so what would be the proper way to "linearize" the branch? I mean, it ought to be simple, right? the history of the branch is linear when not caring about other branches, I would think? what's the fundamental difference between a normal commit and a merge+commit, when only caring about one branch?	12:10
eMBee	how many merges do you have?	12:10
frando	eMBee: dunno, a couple. mostly merges that just add new code, very few that modify exisitng code	12:10
drizzd	frando: I think if you ask again in a few hours maybe there are git-svn proficient people around	12:10
	frando: the linearizing part is probably not the issue, you do that with dcommit all the time.	12:11
	eMBee has used git-svn and has an idea how to selve the problem manually	12:11
eMBee	but it depends on how many merges there are	12:11
drizzd	frando: I was just making a fuzz before because I could not imagine how it handled merges.	12:11
frando	IIRC, it should work just fine. I used git-svn in the past, and didn't have problems with merges. Back then, I started from an existing svn repo, though, and not vice versa.	12:11
drizzd	eMBee: but how do you tell git-svn which branch to commit to?	12:12
frando	drizzd: IIRC that's configurable	12:12
	→ vals_ joined	12:12
eMBee	drizzd: that's in the git-svn setup	12:12
	← tango_ left	12:12
	→ eevar_ joined	12:13
	→ aukjan joined	12:13
	→ Ryback_ joined	12:13
	← eevar_ left	12:14
	→ eevar_ joined	12:14
eMBee	frando: one way to remove the merges is to rebase and at every merge point stop and manually do the merge with squash instead	12:15
	← harinath left	12:16
	← kn1tt1 left	12:16
srnt	Is it true, that when I will do a merge, Git will automatically commit changes, without allowing me to check result of the merge?	12:17
eMBee	another way that i managed to do once is to just go ahead and run svn dcommit, until it fails, (it will fail after commiting the merge) and then run dcommit again. at the end do svn fetch to update your local tree. it's been 2 years that i did that, so i don't remember the details. (also git-svn may have been changed)	12:17
Ilari	srnt: There's --no-commit. But you can test it and fix it before pushing.	12:17
srnt	Ilari: ie. using --no-commit I can test the changes? Or using something else?	12:18
	Is it possible to configure git in such way, that it will commit all changes made to a repo, not only excplictly selected?	12:18
	→ lucsky joined	12:19
	→ Grahack joined	12:20
	← aspotashev left	12:20
	← aspotashev\|eeepc left	12:20
	→ harinath joined	12:20
Ilari	srnt: You can test them anyway even if it auto-commits.	12:21
srnt	Ilari: how?	12:22
Ilari	srnt: There's '-a' to commit, but you can't make it default (aliases can't override builtin or extenral commands).	12:22
	→ aspotashev joined	12:22
	→ aspotashev\|eeepc joined	12:22
Ilari	srnt: Test it. If it fails, fix and amend the last commit. When it passes, you can push (if you want to).	12:22
	← cpg left	12:23
	← harinath left	12:23
drizzd	srnt: see the chapter about merging in the manual	12:23
	If the result of an auto-merge fails the problem is generally not the merge, but one of the individual branches needs to be fixed.	12:24
frando	hm, another question. I'm configuring the svn-remote and am unclear on the difference between the "fetch = .." and "branches = ..." lines. if I configure the branches line like "branches = branches/project1/:refs/remotes/project1-svn/", to what would I set the fetch line?	12:27
	→ LiamH joined	12:27
	→ harinath joined	12:28
	← aukjan left	12:28
	← astrocub left	12:28
srnt	What would you say about anti-git argument: "Git is complicated. Everything is done in non-intuitive way." ?	12:29
spb	when it comes to any computer system 'intuitive' means nothing more than "what i'm used to"	12:30
LotR	"your intuition sucks"? :)	12:30
	→ bremner_ joined	12:31
	← Grahack left	12:31
Ilari	srnt: If you think everything in Git is done non-intuitive way, you clearly don't have your models right...	12:32
srnt	Ilari: ahm. I'm just trying to rationally direct decision of making SCM change in my work. I'm on the git side	12:34
	→ scarabx joined	12:34
	→ dwave joined	12:35
	← solydzajs left	12:38
	← hobodave left	12:38
parasti	srnt: Git is actually very simple. all those dozens of non-intuitive commands suddenly all make sense once you realise they all manipulate just a handful of objects	12:38
srnt	Is it true, that when I do "git add file", then change the file, and then commit it, file will be commited without the changes?	12:38
	parasti: object = thing identified by SHA1	12:39
	?	12:39
parasti	srnt: basically yes + index	12:39
	srnt: and that's true about git add. you have to add your new changes again	12:39
	→ mfilipe joined	12:40
parasti	srnt: btw, index is actually a horrible name. it's better called the staging area	12:40
srnt	are there any other differences like the git/cvs add difference?	12:41
	→ knittl joined	12:41
Ilari	srnt: ... And represents what would be in NEXT commit.	12:41
	→ codebrulee joined	12:42
parasti	srnt: it's best if you don't try to figure out the differences. Git and CVS/SVN are nothing alike	12:42
	(at least, as far as I'm concerned...)	12:42
	← alezandro left	12:42
	→ bremner__ joined	12:43
srnt	Ilari: (with which the "... And repre.." sentence should I connect with?)	12:43
	ahm	12:43
	so this is a kind of argument: "git is completly different"	12:43
	→ darkskiez joined	12:43
	→ marc_m joined	12:44
darkskiez	would it be possible to write a git backend to work with an svn frontend	12:44
Ilari	srnt: The staging area represents what would be in next commit.	12:44
parasti	srnt: that's not really an argument for anything. "different" doesn't mean "impossible/hard to learn"	12:44
Ilari	darkskiez: Sounds like what git-svnserver would be.	12:44
	darkskiez: So yes, it would be possible. But not necressarily easy.	12:45
	← voker57_ left	12:45
darkskiez	yeh, i was just wondering if the paradigms were so detached to make it unrealistic	12:45
	→ bremner___ joined	12:46
Ilari	darkskiez: There's lots of impedance mismatch, but it should be possible to make it work...	12:46
darkskiez	my inital thoughts were to make svn hook scripts to proxy the changesets to git	12:48
	im not sure about git hooks for reverse integration	12:48
	→ mstormo joined	12:48
Ilari	darkskiez: That would only free from having to interpret the protocol. And besides, backing it using Git ODB would likely be way faster.	12:49
	darkskiez: git cvsserver is way faster than the official CVS server.	12:50
darkskiez	yes, but one	12:50
	benefit, could be a rudimentary solution in a couple of scripts	12:50
Ilari	darkskiez: Its not simple to perform impedance matching with two so dissimilar systems. You essentially would need database just to keep track what corresponds to what...	12:51
	← knittl left	12:53
darkskiez	what if you didnt bother and committed changes on one branch to the matching branch on the other repo and did them in lock step	12:54
	like the commit wouldnt complete till the pre-commit hooks had comitted it on the other repo branch	12:54
	→ knittl joined	12:54
darkskiez	is there a use-case that would make that explode for work on a given branch	12:55
Ilari	darkskiez: You would need to be capable of locking refs/branches for update, and not even git exposes that operation.	12:55
	darkskiez: Worse yet, Git wouldn't probably react too well to hook updating a ref.	12:57
	→ fluxusps joined	12:58
	← fluxusps left	12:58
darkskiez	I found http://git.q42.co.uk/w/git_svn_server.git tho it looks a bit stale	12:58
Ilari	darkskiez: The way out of that 'synchronous update' problem is to back it with single repository. But then you are outside scope of hooks...	12:59
	→ wuzle joined	12:59
darkskiez	yeh, single repo seems most problem-free in the long run	12:59
Ilari	darkskiez: And then there's real fun stuff, like forced pushes. SVN probably wouldn't respond too kindly to that...	13:00
	← jasiek left	13:00
	→ paulboone joined	13:00
	← auxbuss left	13:00
	← Leemp left	13:01
	→ lorandi joined	13:01
darkskiez	i dont even know what that means right now.	13:01
	← bremner__ left	13:01
	← lorandi left	13:01
	→ lorandi joined	13:02
	← dwave left	13:02
wuzle	Can someone point me in the right direction as howto send notification emails when some pushes to a central git repository?	13:02
	← agathon23-5_ left	13:02
	→ gotgenes joined	13:04
wuzle	BTW, I am using gitosis, if that matters.	13:04
Ilari	darkskiez: If you try to bridge SVN and Git together using hooks, you are essentially dealing with a distributed system (with two parts), while backing frontends for both with single repository isn't a distributed system (it has only one part)...	13:04
	← bremner___ left	13:04
Ilari	wuzle: post-receive hook? (man githooks)	13:04
Gitbot	wuzle: the githooks manpage can be found at http://git.or.cz/man/githooks	13:04
wuzle	and this would be done on the central repo, not for each client?	13:05
Ilari	wuzle: Central repo only.	13:05
	→ dwave joined	13:06
	← bremner_ left	13:06
wuzle	Ilari: thanks	13:06
	→ Stravy joined	13:07
	→ ArSeN joined	13:07
Ilari	darkskiez: And programming distributed systems is harder than centralized ones, because you get all sort of new possiblities for race conditions...	13:08
	← wuzle left	13:09
	← ArSeN left	13:09
Ilari	darkskiez: Not to mention that some things that can be relied upon on centralized systems break when dealing with distibuted ones. Such as presence of global time.	13:09
	→ jrmuizel joined	13:11
	→ krh joined	13:13
Ilari	darkskiez: And how would you handle "simultaneous" commit to both Git and SVN sides? What if the SVN one is out-of-date, so it does that merge-on-commit thing?	13:15
	→ henster joined	13:15
darkskiez	well svn has pre-commit hooks, so it wouldnt accept the commit until it had committed to git ok	13:16
	and it serializes all checkins	13:16
henster	How come when running "git fetch origin" I get asked for my unix user password?	13:16
mugwump	henster: look at the url shown by 'git remote show -n origin'	13:17
	← gotgenes left	13:17
Ilari	darkskiez: Problem is, you need the "post-molestation" commit, not the "pre-molestation" one.	13:18
	→ yoann joined	13:18
Ilari	darkskiez: And you also need to rollback the other commit if committing to SVN fails.	13:19
henster	mugwump: Thanks. I see.. The url is actually currentusername@localhost:/path/to/url It looks like specifying the user and server forces git to ask for a password..	13:19
	← cedricv left	13:20
	← scarabx left	13:21
	→ jlnr__ joined	13:21
Ilari	darkskiez: Well, you could lock both repos for duration of commit/push. But there's no natural place to put aquiring/releasing that lock...	13:22
yoann	anyone experienced with migration from svn to git ? I'm hitting some git-svn problem, and would need some help solving help	13:22
Ilari	yoann: Error messages?	13:23
darkskiez	Ilari: thanks =]	13:23
	Is there a way to see where all of your space is taken up in your git repo, mine is 700Mb, and i'm not sure what all it taking it up.	13:23
Ilari	darkskiez: And even if you ignore the race possiblities (not a good idea), its still hard.	13:23
yoann	Ilari: no error, but the import contain the imported module, but also unrelated data that belong to other module. I think git-svn has problem handling some of the big directory move there was in the svn repository	13:24
Ilari	darkskiez: You can get listing of objects and their sizes with 'git rev-list --all --objects \| cut -b 1-40 \| git cat-file --batch-check'. Maybe look if there are large objects there?	13:25
	darkskiez: Be aware that sum of those object sizes is likely WAY larger than 700MB...	13:27
	← pan1nx left	13:29
	← foutrelis left	13:31
	→ g3d joined	13:31
	← henster left	13:31
	→ andrewkasper joined	13:32
	→ cmarcelo joined	13:34
Ilari	On my git.git copy, it gives packed size at ~34MB. The sum of object sizes is 1 239MB...	13:34
frando	Hm. So I have a branch with several merges etc etc, and now I do, there, a git-format-patch -o /tmp/git <my_root_commit>, and create a new empty branch, and do git am /tmp/git/*, and it fails in the middle, telling my a patch doesnt apply. How can this happen?	13:34
	→ rolfb joined	13:34
Ilari	frando: format-patch can't deal with merge commits.	13:34
frando	Ilari: Hm. But why not? I can diff between the commit before and after the merge, can't I?	13:35
jlnr__	I want to mirror my svn repos to a git repos, but after the first push, git-svn cannot "determine upstream SVN information" anymore (= no rebase/find-rev for me). Is there a common solution to this	13:35
frando	Ilari: or, in other words, how can I "linearize" a git branch? I have to dcommit it into a svn repo	13:35
jlnr__	? (the old repository didn't move or anything)	13:35
Ilari	frando: Linearization impiles data loss.	13:36
frando	Ilari: Well, I don't care losing the information when a merge happened and when not. I mostly want to preserve my commit messages.	13:36
	← smtms left	13:36
	← doener left	13:37
	← jrmuizel left	13:37
frando	Ilari: Also, I'd still have the original git repo lying around somewhere if I'd have to track down merges. But for now I just have to import the repo into an svn repo, and for that, I need a linear history. That has to be possible - after all, each commit, be it a merge or not, is just changing around source files, right, so it can be expressed in a patch file + commit message.. at least that's what I tought.	13:38
	and that's why I am wondering why git format-patch / git am can fail, and wondering even more what's the proper way to this. As I said, I don't care losing exact merge information. but I do care about my commit messages.	13:39
	→ name joined	13:41
	→ bdiego joined	13:44
	← wwalker left	13:44
sgronblo	Hmm, how do you display the staged changes again?	13:45
Zaba	git diff --cached?	13:45
sgronblo	Yes, thank you.	13:45
	→ ckm joined	13:45
jlnr__	ok, it seems that git-svn can't figure out my repository because i used "--no-metadata" (kind of regret that), but it should still find its stuff in the revision mapping. is there any way out of this?	13:45
	→ ekacnet joined	13:46
ekacnet	hello	13:46
	← ckm left	13:46
ekacnet	I might be stupid but I'm unable to clone a repo shared over rsync	13:46
	git is complaining	13:47
	especialy: warning: remote HEAD refers to nonexistent ref, unable to checkout.	13:47
	← stuffcorpse left	13:48
	→ stuffcorpse joined	13:48
	← esparkman left	13:49
	→ hpoydar joined	13:49
	→ bdimcheff joined	13:50
	→ mw___ joined	13:50
	→ spearce joined	13:51
	→ d0k joined	13:52
	→ agathon23-5 joined	13:53
	← marc_m left	13:56
	→ spearce` joined	13:58
	← spearce` left	13:59
	→ spearce` joined	13:59
	← ph^ left	14:02
	→ ph^ joined	14:03
	→ Kenjin joined	14:06
soul9	ekacnet, are you sure rsync doesn't exclude dotfiles or such?	14:09
	→ smtms joined	14:09
soul9	you can do git clone rsync://foo.baz/proj ???	14:09
	→ voker57 joined	14:10
	→ hugoxrosa joined	14:10
	← paulboone left	14:12
	→ scarabx joined	14:12
	← agathon23-5 left	14:12
	← spearce left	14:15
	← ph^ left	14:16
	→ javatexan joined	14:16
	→ jds joined	14:17
	→ jrmuizel joined	14:18
	→ hobodave joined	14:19
	→ alezandro joined	14:21
	→ Yuuhi joined	14:23
	→ mase_x200 joined	14:24
mase_x200	is it possible to be in love with version control? If so I think i am now.	14:25
	git is awesome	14:25
ekacnet	soul9: I don't think so	14:25
mase_x200	i can't believe i used svn for so long	14:26
soul9	ekacnet, so how you do it?	14:26
	you just rsync the git repo?	14:26
	then i would guess you are excluding files important to git..	14:26
ekacnet	i tried to do git rsync://myserver/repo foo	14:26
	← lorandi left	14:26
ekacnet	it failed	14:27
	→ lorandi joined	14:27
ekacnet	of course if I try rsync rsync://myserver/repo bar it works	14:27
soul9	ekacnet, looks like it works.	14:27
	git clone rsync://host.fqdn/project	14:27
ekacnet	soul9: I have a clone of samba4 and wanted to clone it (I suppose it's possible ...)	14:28
	so I did git clone git://samba.org/samba.git pouet	14:28
soul9	yes	14:29
	son	14:29
ekacnet	pouet is shared as pouet in rsyncd.conf so from other computer	14:29
soul9	SO?	14:29
	AH	14:29
	oops, sorry, caps :)	14:29
	ok	14:29
	i see....	14:29
	← ia left	14:29
soul9	well, git clone rsync://host/pouet should work	14:30
ekacnet	git rsync://myserver/pouet truc should work but i got	14:30
	Initialized empty Git repository in /usr/local/src/truc/.git/	14:30
	warning: remote HEAD refers to nonexistent ref, unable to checkout.	14:30
	→ ia joined	14:30
	→ catalis joined	14:30
ekacnet	git rsync://myserver/pouet truc should work but i got git-pack-refs to create pack-ref file (quite intuitive ...)	14:31
soul9	ekacnet, can you try and clone it locally on that machine?	14:31
	like: git clone /path/to/proj	14:31
ekacnet	sure I can try	14:31
	→ wmoxam joined	14:32
	→ camwest joined	14:32
soul9	that'll make sure that the git repo on that machine isn't broken	14:32
ekacnet	it works	14:32
	← khmarbaise left	14:32
soul9	ok	14:33
	ekacnet, and the repository you checked out via rsync, it works too?	14:34
	git operations work in it?	14:34
ekacnet	sort of	14:35
	git log is ok	14:35
	← struberg left	14:35
	← jlnr__ left	14:35
soul9	what does sort of mean?	14:36
	→ Travis-42 joined	14:36
soul9	:)	14:36
	is there a .git directory, and git branches works, etc?	14:36
ekacnet	yes	14:37
	I didn't test every commands but a quite few work !	14:37
	soul9: btw do you know how to clone a repo and get all the branch declared in it ?	14:38
soul9	yes	14:39
	git clone	14:39
ekacnet	because for instance with this test clone I get only I branch : the one where I am working on	14:39
soul9	git clone fetches all the branches	14:39
	git branch -a will show all the branches	14:39
	but i don't really understand why you would want to go through rsync to get the repo...	14:39
ekacnet	git branch master origin	14:40
	* prod_ecv v3-devel v4-0-test	14:40
	→ arohner joined	14:40
charon	ekacnet: i'm not 100% sure but i thought the rsync transport was deprecated, and old list mails seem to support this, such as http://article.gmane.org/gmane.comp.version-control.git/9307	14:41
	so problems are sort of expected	14:41
soul9	there you go..	14:41
ekacnet	http://pastebin.com/m6ce155cc	14:41
	charon: but in the same time it's in the man pages :	14:42
	→ tvw joined	14:42
charon	here's a newer mail: http://article.gmane.org/gmane.comp.version-control.git/103426	14:42
	(december 08)	14:42
ekacnet	soul9: see my pastebin about branches : when I cloned locally my master repo it just cloned prod_ecv (not the other branchs)	14:43
	← camwest left	14:43
	→ bobmcw joined	14:43
	→ DoctorPepper joined	14:43
ekacnet	charon: ok let's say that it's deprecated (in my git it's not said .... )	14:44
soul9	ekacnet, what version of git are you using anyways?	14:44
ekacnet	git version 1.5.6.5	14:44
soul9	ok, should be fine	14:45
	← whaley left	14:45
ekacnet	from debian 5.0.1	14:45
soul9	ekacnet, the repo with only one branch, it's the one you cloned using rsync?	14:45
	→ willb joined	14:45
charon	ekacnet: actually it does say so in gitcore-tutorial, but i doubt anyone reads that (for beginners it's too low level and for the rest it's boring)	14:45
ekacnet	not the one I cloned on local : git clone /usr/local/samba4 truc	14:45
	→ hobodave_ joined	14:45
soul9	hmm	14:45
ekacnet	charon: I read this tutorial	14:46
soul9	that should be fine thogh, try git branch -a	14:46
	→ Strogg joined	14:47
ekacnet	I __have__ real problems to understand git it seams	14:47
	http://pastebin.com/m33bc9cd3	14:47
	for the git branch -a	14:47
	→ cocksr8 joined	14:47
	→ camwest joined	14:48
	← camwest left	14:48
	← andrewkasper left	14:49
ekacnet	charon: and thanks to samba introduction (http://wiki.samba.org/index.php/Using_Git_for_Samba_Development) it's a bit better	14:49
Ilari	ekacnet: branch -a shows to different kinds of refs...	14:49
	→ Bass10 joined	14:49
ekacnet	so ?	14:49
	my need is pretty simple : I have a repo with branches I want to clone it somewhere else and have this branchs there as well ...	14:50
	← Bass10 left	14:50
cocksr8	Hi. I did a git fetch /path/repo.git refs/tags/NAME_OF_TAG but that hasn't pulled the tag name in... and I have since blatted over FETCH_HEAD... is there any way I can checkout NAME_OF_TAG or find it's SHA?	14:50
ekacnet	if I can avoid to have to setup a web server just for this it could be good ...	14:50
Ilari	ekacnet: After cloning, the local branches of original repo appear as remote tracking branches in cloned repo.	14:50
	→ Bass10 joined	14:51
	→ kukks joined	14:51
ekacnet	Ilari: not always: if i do git clone git://git.samba.org/samba.git samba	14:51
	I get already some branchs !	14:51
	← juanjoc left	14:51
Ilari	ekacnet: And clone creates local branch for one of the branches.	14:51
ekacnet	Ilari: so I just have to do git checkout mybranch (if mybranch already existed in my source repo) ?	14:52
Ilari	cocksr8: 'git fetch <repo> refs/tags/NAME-OF-TAG:refs/tags/NAME-OF-TAG'?	14:52
	→ mobhawk joined	14:53
ekacnet	or git checkout -b mybranch origin/mybranch	14:53
Ilari	ekacnet: 'git checkout --track origin/foo' (if you have recent enough git).	14:53
	ekacnet: That checkout -b also works (even with older versions).	14:53
ekacnet	rather complicated all this stuff .....	14:54
	→ __iron joined	14:54
	← iron left	14:54
ekacnet	mercurial miss me :-)	14:54
Ilari	ekacnet: Checkout can only really check out local branches. 'git checkout -b foo bar' means the same as 'git branch foo bar', 'git checkout foo'.	14:55
	→ willb1 joined	14:55
Travis-42	is there a way to change the width of tab indents for things like git-diff? I work with a project where most developers display tabs as 4 spaces, and so git-diff usually fails to display a single line in an 80-character width (but would if a tab was shown as 4 spaces)	14:56
	→ marc_m joined	14:56
soul9	ekacnet, you can just use ssh.	14:57
cocksr8	Ilari: I don't have access to the remote repo at the moment	14:59
	I just have my clone	14:59
	→ will___ joined	15:00
	vals_ → tango_	15:00
	← stuffcorpse left	15:00
Ilari	cocksr8: 'git fsck --full --unreachable' and look for 'dangling tag'?	15:00
	will___ → throughnothing	15:01
	→ altrux joined	15:01
	→ willb2 joined	15:01
	← hobodave left	15:02
darkskiez	I have a git blob id, how can i find what refers to it	15:02
	← TomK32 left	15:02
sgronblo	I screwed up a bit. I thought I had just staged a few lines with add -p on a file and then did a commit and ended up committing all the changes in the file. Is it possible to change the last commit I made?	15:02
	→ ilteris__ joined	15:02
Ilari	darkskiez: No fast way exists. You basically need to follow references and find path to it.	15:03
	→ daryn joined	15:03
darkskiez	how would i start to approach that	15:03
Ilari	sgronblo: 'git reset -- file', 'git add -p file', 'git commit --amend'?	15:03
	darkskiez: One useful thing would likely be grepping the ID from 'git rev-list --all --objects'. That should give you (one) path for it.	15:04
	darkskiez: But that doesn't give what commit has it.	15:04
	← willb left	15:05
	← tvw left	15:05
	→ tvw joined	15:05
	willb2 → willb	15:05
sgronblo	Ilari: after i did git reset -- file nothing seemed to happen. git status doesn't show the file as changed so add -p also reports No changes.	15:06
Ilari	sgronblo: Oops... 'git reset HEAD^ -- file'.	15:07
	→ ceej joined	15:08
ekacnet	soul9: i tried git-daemon it seems to works pretty good now !	15:09
Ilari	ekacnet: Except that git daemon virtually can't do pushes.	15:09
darkskiez	ok, making progress, thanks Ilari	15:10
ekacnet	Ilari: yes	15:11
	I'll survive for the moment :-)	15:11
	Ilari: only ssh allow push ?	15:12
soul9	Ilari, what does virtually mean for you? :)	15:12
	virtual = reality? :D	15:12
	because you can't write with git-daemon, for that you need ssh	15:12
	→ resmo_ joined	15:12
	← Kenjin left	15:12
	→ agathon23-5 joined	15:13
	← agathon23-5 left	15:13
Ilari	ekacnet: Well, some other stuff supports push as well, but its either local-only or has some nasty downside.	15:13
	← willb1 left	15:14
	→ qrush joined	15:15
Ilari	soul9: There IS option to enable writing with git daemon. It isn't secure.	15:15
	← catalis left	15:16
srnt	It is generally bad idea to switch to remote branch ?	15:18
j416	srid: you will just detach HEAD, no harm done.	15:18
	→ MisterN joined	15:18
	← __iron left	15:19
jast	sure, decapitation wouldn't ever do any harm ;)	15:19
srnt	j416: me?	15:20
j416	ops	15:20
	sorry, yes	15:20
srnt	hm	15:20
	what do you mean by saying "detach HEAD"	15:20
	← name left	15:21
j416	you will be on "no branch"	15:21
	you can continue to commit and do what you want, but you will want to create a branch from it if you want to keep your changes	15:22
	← lohrun left	15:22
darkskiez	is there a way i can delete blobs from my git repo	15:22
soul9	Ilari, i see, didn't know that..	15:23
Ilari	darkskiez: They will eventually be garbage-collected if they aren't part of any branch, tag or other ref...	15:25
	→ dwave_ joined	15:25
	→ bartocc joined	15:26
darkskiez	so how can i track the refs	15:26
	→ vipul joined	15:26
darkskiez	100's of megs of files have been commited by accident over time and deleted	15:26
	leading to a big repo	15:26
	← Octalot left	15:27
j416	darkskiez: man git-gc	15:31
Gitbot	darkskiez: the git-gc manpage can be found at http://git.or.cz/man/git-gc	15:31
darkskiez	it still has references tho	15:31
	so it wont get gc'ed	15:31
	← johan-s left	15:32
	→ jasiek joined	15:32
	← smeevil left	15:32
	← g3d left	15:32
	→ perlmonkey2 joined	15:33
perlmonkey2	What does this mean: "Forward-port local commits to the updated upstream head"?	15:33
	→ bcarlyon\|laptop joined	15:33
	→ jackdempsey joined	15:35
	← gilimanjaro left	15:37
offby1	heh	15:38
Ilari	perlmonkey2: Sounds like rebasing...	15:38
	← psankar left	15:38
Ilari	perlmonkey2: Eergh... ERECURSIVEDEF likely.	15:39
rolfb	perlmonkey2: it sounds dirty	15:39
	if you want it to	15:39
offby1	Didn't Richard Pryor burn himself badly while rebasing?	15:39
	← jasiek left	15:39
	→ releod joined	15:40
	← bobmcw left	15:40
perlmonkey2	That is from the git man pages on 'rebase'	15:40
	← seangrove left	15:40
Ilari	perlmonkey2: Essentially means rebase will tear down the branch and rebuild it on top of new upstream branch.	15:40
perlmonkey2	so that is how you commit changes to your local repo to another repo?	15:41
Ilari	perlmonkey2: Nope. Push is for uploading commits to another repo.	15:41
	← dodo_the_last left	15:41
	→ kana_ joined	15:42
	← scientes left	15:42
	← cilly left	15:42
perlmonkey2	Ilari: yeah, see I never would have drawn that conculsion from the man entry.	15:42
	forward port local commits to an upstream head sounds like pushing your commits to the non-local repo.	15:43
Ilari	perlmonkey2: Git doesn't have hierarchial branches. All branches are local and indepedent.	15:43
	kana_ → kana	15:44
perlmonkey2	Ilari: okay, say you have a team. They at some point want to merge all their local repos to build a release. who's repo acts as the clearing house for all the merges?	15:44
	← ekacnet left	15:45
perlmonkey2	ie, how does everyone who has a local repo cloned from github put back all their changes to github?	15:45
rolfb	perlmonkey2: by using a fork	15:46
	and not a clone	15:46
	← qrush left	15:46
rolfb	and requesting the original repo to pull from your fork after you updated it	15:46
	→ struberg joined	15:46
Ilari	perlmonkey2: Usually they who made some feature branch merge it to mainline.	15:47
cthompson	yep, that's how I pull patches from people who clone my repo	15:47
	I'd pull something you did into a branch called perlmonkey2	15:47
	test and verify, then merge into my devel branch, delete the perlmonkey2 branch	15:48
Ilari	perlmonkey2: Sometimes in Linux work some maintainer rejects pull request by essentially "merge this to mainline and resend pull request" if the branch doesn't merge cleanly.	15:48
	← Mjollnir` left	15:49
	← rolfb left	15:49
Ilari	perlmonkey2: As they who did the feature presumably know the code best and are qualified to fix the conflicts.	15:49
	→ MauL^ joined	15:50
	→ girishr joined	15:50
	← Stravy left	15:50
	perlmonkey2 goes back to reading documenation	15:51
MauL^	hi. I've done a lots of change to my working dir. Today, I've made a commit -a. only some of the changes are commited! then I did commit -a again.. the other parts are commited. is there a logic here? is there a limit for committing each time ?	15:51
offby1	nope	15:51
	→ saurabh1403 joined	15:51
offby1	that's confusing	15:51
Ilari	perlmonkey2: The idea is that if there is some difficult merge, if one can do it, the rest can pull the merge.	15:51
offby1	MauL^: it doesn't surprise me that the first "commit -a" didn't commit all changes, if those changes were to files that git didn't "know about"	15:51
	← dwave left	15:51
	→ qrush joined	15:52
MauL^	offby1, yes thats exaclty like that	15:52
	I've added many new files	15:52
	← Bleupomme left	15:52
MauL^	so I need to make 2 commit ?	15:52
	← kana left	15:52
perlmonkey2	MauL^: git status	15:52
MauL^	now it says clean.	15:52
	but I did twice already	15:53
	← dwave_ left	15:53
drizzd	after git commit -a, a second git commit -a is a no-op	15:53
offby1	MauL^: if you did "git commit -a", then "git add", then did some more editing, then yes, you must do another "git commit"	15:53
Ilari	MauL^: If you want to combine those two commits (and haven't pushed the yet), look at interactive rebase (man git-rebase)	15:53
drizzd	the same goes for git commit, for that matter	15:53
Gitbot	Ilari: the git-rebase manpage can be found at http://git.or.cz/man/git-rebase	15:53
	← wycats left	15:54
	→ justatheory joined	15:54
	→ TomK32 joined	15:54
MauL^	umm ok	15:55
	gwoo_ → gwoo	15:55
Ilari	MauL^: Also, if the two commits are the two last ones, 'git reset --soft HEAD^', 'git commit --amend' would also work.	15:56
perlmonkey2	okay, since I'm looking at an SVN style repo managment, I think this is all I need to do. git clone, git fetch/pull to keep up to date, git commit, then git push.	15:57
	← tango_ left	15:57
perlmonkey2	although git fetch I think requires a merge, while git pull does a fetch then merge.	15:57
Ilari	perlmonkey2: pull = fetch + merge	15:57
perlmonkey2	cool	15:57
	← qhoxie left	15:57
	← ringlej left	15:58
Ilari	perlmonkey2: And note that pull/merge doesn't require commit afterwards unless it conflicts.	15:58
perlmonkey2	and since that remote repo is a fork of the master, and only 2 devs are working on it, that should be fine.	15:58
	Ilari: groovy, thanks.	15:58
Ilari	perlmonkey2: As by default it does commit after successful automerge.	15:58
perlmonkey2	Ilari: that makes sense. I would have assumed that was what happened as I'm merging the repo, not hte working copy.	15:59
	although I would hope the working copy is updated too?	16:00
	→ dwave joined	16:00
Ilari	perlmonkey2: One thing about merging: Doing it with local uncommitted changes is playing russian roulette with your changes.	16:00
	→ lois[pac] joined	16:00
Ilari	perlmonkey2: Yes, merge does update it.	16:00
perlmonkey2	Ilari: hmmm, so no updating without first committing all changes....that sucks.	16:00
	← bschindler left	16:01
Ilari	perlmonkey2: look at 'git stash'.	16:01
perlmonkey2	okay	16:01
	hah, nice	16:01
smtms	is there a way to see which commits would be merged by "git merge"?	16:01
	brennen\|out → brennen	16:01
Ilari	smtms: Fetch those first and then 'git log ..branch-to-be-merged' (note the two dots).	16:02
perlmonkey2	wow, that kind of forces a test driven development style. You need to keep your changes committed often, which means they need to pass their tests often. I'm not sure I like being forced to do that, but I can't argue with yet another reason to write tests.	16:02
Ilari	smtms: Ah, its git merge, so nothing to fetch.	16:02
	← mediogre left	16:02
	← frando left	16:02
Ilari	perlmonkey2: Also read about topic branches.	16:02
	→ camwest joined	16:03
	← MauL^ left	16:03
	→ nighthwk1 joined	16:04
	← camwest left	16:04
	→ Dreamer3 joined	16:05
	→ LotR joined	16:06
	→ wycats joined	16:06
	← bcarlyon\|laptop left	16:06
	→ WALoeIII joined	16:07
	← giallu left	16:07
	← mase_x200 left	16:07
	→ Alinermedia joined	16:08
	→ bcardarella joined	16:08
Ilari	perlmonkey2: In Git, "branch" is not a dirty word. :-)	16:08
perlmonkey2	okay, if I'm polluting this channel just tell me to shut up. But I'm still trying to 'get' git rebase.	16:08
	← altrux left	16:08
perlmonkey2	it updates from the upstream branch and then tries to merge back your local commits?	16:09
jast	it applies them one by one	16:09
Ilari	perlmonkey2: Its quite close to format-patch+reset+am combo...	16:10
	→ paulboone joined	16:10
smtms	sounds scary :-)	16:10
	→ JasonWoof joined	16:10
	← mikl left	16:11
	→ faithfulgeek joined	16:11
Grum	its not that scary :)	16:11
zuez	If the remote branch I want to merge/fetch from is origin/foo and my local branch is foo, isn't it just "git config branch.foo.merge 'refs/origin/foo'" ?	16:12
	→ AkumaStreak joined	16:12
j416	If I have a file X (from someplace else, but which I guess is in the repo), which I don't know anything about, how can I search my git repo to get the commit where the file was introduced?	16:13
bcardarella	What the command to show where a remote branch is tracking back to? (address)	16:14
JasonWoof	j416: you want to search for the file name or contents?	16:14
j416	JasonWoof: contents	16:14
	← bartocc left	16:14
	← gb\|work left	16:14
	→ eletuchy_ joined	16:15
bcardarella	nm, found it in one of the files	16:15
perlmonkey2	oh, just when I think I've got it understood, the local repos I have are origin and crypt. I'm not even sure why I have the origin.	16:15
	→ dodo_the_last joined	16:15
perlmonkey2	the remote repo branch I wanted to work on was just the crypt one.	16:16
	← eletuchy left	16:17
	→ rubydiamond joined	16:17
	→ Odd_Bloke joined	16:18
JasonWoof	j416: gitk has a feature to search for commits that add/remove a string, but I don't know how to search for the whole file contents	16:18
	← marc_m left	16:18
Odd_Bloke	Hello all. Is there any way to stop 'git archive --format tar' from creating the 'pax_global_header' file in the tar it produces?	16:18
j416	hmm	16:21
	→ kumbayo joined	16:21
DrNick	did you try reading the man page?	16:21
	→ johan-s joined	16:22
	→ juanjoc joined	16:23
	→ mczepiel joined	16:23
	← Kbyte left	16:23
drizzd	Odd_Bloke: afaik it doesn't do that. Maybe it's a file from your repo?	16:24
	→ boto joined	16:25
	← tal67 left	16:25
Odd_Bloke	drizzd: I'm going on what I'm seeing and http://archives.free.net.ph/message/20060301.180552.3c03b2d3.en.html to suggest that it's not.	16:25
	← jmesnil left	16:26
j416	JasonWoof: I solved it using something like: git whatchanged \| grep -B 10 <abbr. sha1>	16:26
	which found the commit	16:26
	but it's not very neat	16:26
	← fujin left	16:26
j416	perhaps a useful feature to add? git find	16:26
peper	hey	16:26
JasonWoof	j416: cool	16:26
j416	or git locate	16:27
	hm	16:27
Odd_Bloke	Hmm, now I can't reproduce.	16:27
peper	what was the easiest way to fix an author name e-mail in the whole history?	16:27
	filter-branch?	16:27
j416	peper: yep	16:27
drizzd	Odd_Bloke: git ls-files -t pax_global_header	16:27
	← jspiros left	16:27
	→ kuadrosx joined	16:28
JasonWoof	j416: don't think many people will need that automated. Most people identify files by name, and that's easy to find with gitk	16:28
j416	JasonWoof: I was thinking if you get a file from someone, or you send a file to someone, and you want to, from that file data, find what version it came from	16:29
drizzd	Odd_Bloke: well I can't reproduce	16:29
j416	but it may not be that common.	16:29
	→ rubydiam_ joined	16:29
JasonWoof	j416: hmm... that does sound like a more common usage than I could think of	16:30
Odd_Bloke	drizzd: Yeah, my bad, it's happening somewhere else entirely.	16:30
peper	j416: with --env-filter, right?	16:30
drizzd	Odd_Bloke: but I have tar-1.22 and it's only supposed to happen pre 1.14	16:30
j416	peper: yep	16:30
Odd_Bloke	drizzd: I have >1.14 as well. I think it's happening when bzr-buildpackage unpacks the tarball using Python's tar libraries incorrectly.	16:30
yoann	I'm following the manual for failed rename detection (http://git.or.cz/gitwiki/GitFaq#head-67dfccf2c616ec2f7e7343ee0bd1bdb08c2e5e35), after following the instruction, further merge doesn't detect the rename either. Is this expected ?	16:30
Odd_Bloke	So I was miles off asking in here. :p	16:31
JasonWoof	j416: dunno how hard it is to get the developers to code something that you can do now with a one-liner	16:31
	git whatchanged \| grep -B 10 `sha1 FILE`	16:32
peper	j416: Namespace refs/original/ not empty	16:32
	what does it mean?	16:32
	→ countd joined	16:32
	→ simosx joined	16:32
drizzd	yoann: once you resolved the rename in a merge, it's gone. what is there to detect?	16:33
JasonWoof	j416: hmm... I see my one-liner is far from working, and that there's quite a fudge factor	16:34
yoann	drizzd: my other branch still contain the unrenamed version, further change in that branch pulled to the other branche where the file has been renamed fail	16:34
j416	peper: you have run filter-branch before. It saves the original stuff to refs/original in case you mess up	16:34
drizzd	yoann: if you continue working on the unmerged branch without the rename, it's like doing the same rename again	16:34
j416	just remove .git/refs/original	16:34
	JasonWoof: it needs to be the abbreviated sha1	16:34
JasonWoof	j416: such a feature would also make a quick way to found out how long it's been since files were changed	16:34
j416	i guess.	16:34
	it wouldn't be very hard to implement.	16:35
JasonWoof	maybe that's easy already, git log foo.txt will tell you pretty quick	16:35
yoann	drizzd: I don't understand your last statement, could you rephrase?	16:35
drizzd	yoann: rename detection only works if a significant amount of the file stays intact. otherwise it's like a complete rewrite	16:35
j416	JasonWoof: ah, yes. indeed.	16:35
drizzd	yoann: ok, so you have a file A, in master you renamed it to B and in topic you edited A.	16:36
	← Odd_Bloke left	16:36
yoann	drizzd: the FAQ entry I copied earlier seems to imply it is possible to tell git about a rename (when the rename detection failed). Or am I missing something ?	16:36
	→ tango_ joined	16:36
drizzd	But you changed A so drastically that it is not recognized as a rename.	16:36
	→ tillberg__ joined	16:36
drizzd	"How to manually resolve conflicts when Git failed to detect rename?" means just that	16:37
	→ kraymer joined	16:37
drizzd	git never knows about renames (although the rerere mechanism might help with that, but I don't think that applies to you)	16:37
	continuing with your previous question...	16:38
	you now merge topic into master	16:38
	you have to resolve manually and you decide to take the changes made to A and rename it to B.	16:38
	→ Hendu joined	16:38
drizzd	then you continue to make changes to A on topic	16:38
	now if you merge again git will assume the previously merged tip of topic to be your starting point	16:39
	from this point it looks like A was renamed to B in master	16:39
	and A was modified in topic	16:39
	← willb left	16:39
drizzd	if again A changed significantly enough to fool rename detection, you will again have to resolve the merge manually	16:39
	← codebrulee left	16:40
	← aziz left	16:40
	← RodP2 left	16:40
	← schlort left	16:40
	← DrNick left	16:40
	← ccherrett left	16:40
	← paulboone left	16:40
	← wycats left	16:40
	← Dreamer3 left	16:40
	← ilteris__ left	16:40
	← perlmonkey2 left	16:40
	← mw___ left	16:40
	← drewr left	16:40
	← shruggar left	16:40
	← hyperair left	16:40
	← jpeeler left	16:40
	← Pistahh left	16:40
	← troyt left	16:40
	← brennen left	16:40
	← GyrosGeier left	16:40
	← kuadrosx left	16:40
	← girishr left	16:40
	← daryn left	16:40
	← jds left	16:40
	← hpoydar left	16:40
	← tbf left	16:40
	← AAA_awright left	16:40
	← gitte left	16:40
	← curvature left	16:40
	← Modius left	16:40
	← rettub_ left	16:40
	← boto left	16:40
	← AkumaStreak left	16:40
	← JasonWoof left	16:40
	← TomK32 left	16:40
	← cocksr8 left	16:40
	← Yuuhi left	16:40
	← javatexan left	16:40
	← scarabx left	16:40
	← d0k left	16:40
	← bdimcheff left	16:40
	← Travis-42 left	16:40
	← LiamH left	16:40
	← cannonball left	16:40
	← cemerick left	16:40
	← tttsssttt left	16:40
	← naeu left	16:40
	← _slacker_ left	16:40
	← pastie left	16:40
	← tjafk1 left	16:40
	← mika left	16:40
	← murph_ left	16:40
	← Whtiger left	16:40
	← jepler left	16:40
	← EvanR left	16:40
	← nighthwk1 left	16:40
	← ceej left	16:40
	← throughnothing left	16:40
	← alezandro left	16:40
	← djinni left	16:40
	← Tv left	16:40
	← warthog9 left	16:40
	← releod left	16:40
	← quoin left	16:40
	← simosx left	16:40
	← dodo_the_last left	16:40
	← LotR left	16:40
	← Strogg left	16:40
	← DoctorPepper left	16:40
	← moonflux left	16:40
	← meyering left	16:40
	← playya left	16:40
	← juanjoc left	16:40
	← MisterN left	16:40
	← wmoxam left	16:40
	← hugoxrosa left	16:40
	← cmarcelo left	16:40
	← mbroeker left	16:40
	← FreakGuard left	16:40
	← gwoo left	16:40
	← cytrinox left	16:40
	← tillberg__ left	16:40
	← bcardarella left	16:40
	← galderz left	16:40
	← samjam left	16:40
	← skyl left	16:40
	→ KingBug joined	16:41
yoann	drizzd: thanks for the explanation	16:41
	drizzd: is there something to check the similarity of both file, as git can compute it ?	16:41
drizzd	yoann: hmm, not sure. I think git log --name-status -M gives you some similarity stats	16:42
	→ jurg` joined	16:42
	→ josh^^ joined	16:42
	→ siyb joined	16:42
	→ Cap_J_L_Picard joined	16:42
	→ axiom joined	16:42
	axiom → majoh	16:42
	→ jast joined	16:42
	→ jkp joined	16:43
	→ reaVer joined	16:43
	→ literal joined	16:43
	→ simosx joined	16:43
	→ dodo_the_last joined	16:43
	→ LotR joined	16:43
	→ releod joined	16:43
	→ Strogg joined	16:43
	→ DoctorPepper joined	16:43
	→ quoin joined	16:43
	→ moonflux joined	16:43
	→ meyering joined	16:43
	→ playya joined	16:43
	→ hatseflats joined	16:43
	→ peper joined	16:43
	→ MarcWeber joined	16:43
	→ numbah joined	16:43
	→ impulse9 joined	16:43
	→ eMBee joined	16:43
	→ bartman joined	16:43
	→ jegade joined	16:43
	→ dwmw2_gone joined	16:43
	→ janl joined	16:43
	→ avtobiff joined	16:43
	→ madduck joined	16:43
	→ darjeeling joined	16:43
	→ seb42 joined	16:43
	→ Tali joined	16:43
	→ dionoea joined	16:43
	→ mithraic joined	16:43
	→ hgb joined	16:43
	→ bmalee joined	16:43
	→ incd^ joined	16:43
	→ maxb joined	16:43
	→ pcgod joined	16:43
	→ natlus joined	16:43
	→ daleglass joined	16:43
	→ hegge joined	16:43
	→ Ingmar joined	16:43
	→ impulze joined	16:43
	→ Blafasel joined	16:43
	→ rleigh joined	16:43
	→ shana joined	16:43
	→ jasonmay joined	16:43
	→ krig joined	16:43
	→ metze joined	16:43
	→ masa- joined	16:43
	→ milli joined	16:43
	→ dgfitch joined	16:43
	→ PsychoTrahe joined	16:43
	→ sgronblo joined	16:43
	→ fred joined	16:43
	→ jpic_pro1 joined	16:43
	→ koki joined	16:43
	← mfilipe left	16:43
drizzd	yoann: it will say something like R077 file.orig file.renamed (where 77 means 77% similarity)	16:43
yoann	drizzd: I do not see that	16:44
	drizzd: it provide me with a standard log	16:44
	→ bdrewery joined	16:44
	→ mpr joined	16:45
	→ jepler joined	16:45
	→ EvanR joined	16:45
	→ Whtiger joined	16:45
	→ Tekerson joined	16:45
	→ mlins joined	16:45
	→ bryanray joined	16:45
	→ thiago joined	16:45
	→ tim_jenssen joined	16:45
	→ nil joined	16:45
	→ Louie` joined	16:45
	→ ft joined	16:45
	→ nadim joined	16:45
	→ bla joined	16:45
	→ andreaja joined	16:45
	→ jjuran joined	16:45
	→ ice799 joined	16:45
	→ bgamari joined	16:45
	→ Zenopus joined	16:45
	→ Jc2k joined	16:45
	→ precision joined	16:45
	→ mugwump joined	16:45
	→ sdiz-afk joined	16:45
	→ andres joined	16:45
	→ coopsh joined	16:45
	→ Pieter joined	16:45
	→ lamont joined	16:45
	→ rholden joined	16:45
	→ nud joined	16:45
	→ jefftriplett-afk joined	16:45
	→ zwu joined	16:45
	→ alus joined	16:45
	← rubydiamond left	16:45
	→ mrg05 joined	16:45
	← mrg05 left	16:45
drizzd	yoann: well it should at least say D file.orig A file.renamed (if rename detection fails)	16:45
	→ djinni joined	16:46
	→ segher joined	16:46
	→ tazle_ joined	16:46
	→ nomadium joined	16:46
	→ vmiklos joined	16:46
	→ kelvie joined	16:46
	→ lu_zero joined	16:46
	→ KwikkSilva joined	16:46
	→ Garoth joined	16:46
	→ seanius_ joined	16:46
	→ orospakr joined	16:46
	→ tro joined	16:46
	→ Rhonda joined	16:46
	→ leo2007 joined	16:46
	→ Gitbot joined	16:46
	→ jettero joined	16:46
	→ Woody2143 joined	16:46
	← mpr left	16:46
	→ klmann joined	16:46
	→ bcardarella joined	16:46
	→ galderz joined	16:46
	→ broonie joined	16:46
	→ pasky joined	16:46
	→ deavid joined	16:46
	→ KiBi joined	16:46
	→ wsc joined	16:46
	→ paparent joined	16:46
	→ dominikh joined	16:46
	→ cxreg joined	16:46
	→ shad joined	16:46
	→ hiffy joined	16:46
	→ jarh joined	16:46
	→ gitte joined	16:46
	→ joevano joined	16:46
	→ djpig joined	16:46
	→ rektide joined	16:46
	→ tek0 joined	16:46
	→ schlort joined	16:46
	→ Kudos joined	16:46
	→ ruskie joined	16:46
	→ tpope joined	16:46
	→ eikke joined	16:46
	→ BooK_ joined	16:46
	→ a-atwood joined	16:46
	→ EvanCarroll joined	16:46
	→ decklin joined	16:46
	→ radii joined	16:46
	→ tokkee joined	16:46
	→ zuez joined	16:46
	→ sr joined	16:46
	→ univate joined	16:46
	→ PugMajere joined	16:46
	→ agib joined	16:46
	→ Tsion joined	16:46
	→ sergiodj joined	16:46
	→ aser joined	16:46
	→ imarcusthis joined	16:46
	→ rettub_ joined	16:46
	→ Modius joined	16:46
	→ curvature joined	16:46
	→ AAA_awright joined	16:46
	→ GyrosGeier joined	16:46
	→ tbf joined	16:46
	→ hpoydar joined	16:46
	→ jds joined	16:46
	→ daryn joined	16:46
	→ kuadrosx joined	16:46
	→ Gitzilla joined	16:46
	→ Weiss joined	16:46
	→ rphillips joined	16:46
	→ Arrowmaster joined	16:46
	→ cousin_it joined	16:46
	→ droidcor1 joined	16:46
	→ Ilari joined	16:46
	→ Nugget joined	16:46
	→ apeiron joined	16:46
	→ Derega joined	16:46
	→ wagle joined	16:46
	→ jtong joined	16:46
	→ brennen joined	16:46
	→ troyt joined	16:46
	→ Pistahh joined	16:46
	→ jpeeler joined	16:46
	irc.freenode.net set mode: +o	16:46
	→ hyperair joined	16:46
	→ shruggar joined	16:46
	→ drewr joined	16:46
	→ ilteris__ joined	16:46
	→ perlmonkey2 joined	16:46
	→ Dreamer3 joined	16:46
	→ wycats joined	16:46
	→ paulboone joined	16:46
	→ codebrulee joined	16:46
	→ aziz joined	16:46
	→ RodP2 joined	16:46
	→ DrNick joined	16:46
	→ ccherrett joined	16:46
	→ tonyj joined	16:46
	→ hummingkid joined	16:46
	→ jchris joined	16:46
	→ max_ep_ joined	16:46
	→ dermoth joined	16:46
	→ dedmajor joined	16:47
	→ FFighter joined	16:47
	→ thumper joined	16:47
	→ qhoxie joined	16:47
	→ ustedes joined	16:47
	→ boto joined	16:47
	→ AkumaStreak joined	16:47
	→ JasonWoof joined	16:47
	→ TomK32 joined	16:47
	→ cocksr8 joined	16:47
	→ Travis-42 joined	16:47
	→ Yuuhi joined	16:47
	→ javatexan joined	16:47
	→ scarabx joined	16:47
	→ d0k joined	16:47
	→ bdimcheff joined	16:47
	→ LiamH joined	16:47
	→ cannonball joined	16:47
	→ cemerick joined	16:47
	→ tttsssttt joined	16:47
	→ naeu joined	16:47
	→ _slacker_ joined	16:47
	→ pastie joined	16:47
	→ tjafk1 joined	16:47
	→ mika joined	16:47
	→ murph_ joined	16:47
	→ sebleier joined	16:47
	→ elvedin joined	16:47
	→ Helios joined	16:47
	→ nick_h[litage] joined	16:47
	→ ArjenL joined	16:47
	→ osfameron joined	16:47
yoann	drizzd: I got: M file.orig	16:47
	→ mpr joined	16:47
	→ meebey joined	16:47
yoann	drizzd: nothing about the renamed file	16:47
	← niki left	16:47
	← Pistahh left	16:47
	← troyt left	16:47
	← brennen left	16:47
	← wagle left	16:47
	← Nugget left	16:47
	← Weiss left	16:47
	← Gitzilla left	16:47
	← droidcor1 left	16:47
	← apeiron left	16:47
	← shruggar left	16:47
	← perlmonkey2 left	16:47
	← jtong left	16:47
	← joevano left	16:47
	← cousin_it left	16:47
	← drewr left	16:47
	← wycats left	16:47
	← Ilari left	16:47
	← Derega left	16:47
	← ilteris__ left	16:47
	← jpeeler left	16:47
	← Arrowmaster left	16:47
	← rphillips left	16:47
	← Dreamer3 left	16:47
	← paulboone left	16:47
	← hyperair left	16:47
	← _slacker_ left	16:47
	→ joevano joined	16:47
	→ paulboone joined	16:47
	→ wycats joined	16:47
	→ Dreamer3 joined	16:47
	→ perlmonkey2 joined	16:47
	→ ilteris__ joined	16:47
	→ drewr joined	16:47
	→ shruggar joined	16:47
	→ hyperair joined	16:47
	→ jpeeler joined	16:47
	→ Pistahh joined	16:47
	→ troyt joined	16:47
	→ brennen joined	16:47
	→ jtong joined	16:47
	→ wagle joined	16:47
	→ Derega joined	16:47
	→ apeiron joined	16:47
	→ Nugget joined	16:47
	→ Ilari joined	16:47
	→ droidcor1 joined	16:47
	→ cousin_it joined	16:47
	→ Arrowmaster joined	16:47
	→ rphillips joined	16:47
	→ Weiss joined	16:47
	→ Gitzilla joined	16:47
	← zuez left	16:47
	→ fujin joined	16:47
	← bryanray left	16:47
	→ Grum joined	16:47
	→ tchan joined	16:47
	→ john3909 joined	16:47
	→ ash_ joined	16:47
	→ ricky joined	16:47
	→ zuez joined	16:47
	→ offby1` joined	16:47
	→ juanjoc joined	16:47
	→ MisterN joined	16:47
	→ wmoxam joined	16:47
	→ hugoxrosa joined	16:47
	→ cmarcelo joined	16:47
	→ mbroeker joined	16:47
	→ FreakGuard joined	16:47
	→ gwoo joined	16:47
	→ cytrinox joined	16:47
	← zuez left	16:47
	→ zuez_ joined	16:47
drizzd	yoann: M file.orig is _not_ a rename	16:47
	← mczepiel left	16:48
	← faithfulgeek left	16:48
	→ niki joined	16:48
	← wycats left	16:48
	→ ahf joined	16:48
yoann	drizzd: yes, I wonder why it say that...	16:48
	← shruggar left	16:49
	← TomK32 left	16:49
yoann	drizzd: at git mv time, I can see:	16:49
drizzd	yoann: if you had renamed the file, then the renamed file would be in the logs	16:49
	← saurabh1403 left	16:50
yoann	drizzd: R087 prewikka/IDMEFDatabase.py prewikkapro/IDMEFDatabase.py	16:50
	← daryn left	16:50
	→ nighthwk1 joined	16:50
yoann	drizzd: but it appear earlier as D prewikka/IDMEFDatabase.py, A prewikkapro/IDMEFDatabase.py	16:50
	→ mw joined	16:50
drizzd	yoann: how about git log --name-status -M1	16:51
	→ throughnothing joined	16:51
peper	j416: stupid split: ah, thanks. also, i'm a bit confused, what quoting am i supposed to use for <command>?	16:52
yoann	drizzd: so we got R087	16:52
	drizzd: if that's 87%, why isn't the rename detected later on merge ?	16:52
	← juanjoc left	16:52
j416	peper: ' for example	16:52
drizzd	yoann: because you said you had a rename earlier that was not detected	16:52
	yoann: see if it's detected with -M1	16:52
	→ juanjoc joined	16:53
yoann	drizzd: I get it as R087 prewikka/IDMEFDatabase.py prewikkapro/IDMEFDatabase.py	16:53
	drizzd: (this is M1)	16:53
drizzd	the same as the rename that was detected with just -M? that seems odd	16:53
	ustedes → whaley	16:54
	→ skyl joined	16:54
drizzd	yoann: what exactly did you mean by "earlier"?	16:54
	→ tansell joined	16:54
	→ gb\|work joined	16:54
	← gb\|work left	16:55
	→ alezandro joined	16:55
	→ samjam joined	16:55
drizzd	yoann: in any case, 87% similarity should be enough for merge to detect the rename	16:55
j416	peper: sorry for not helping you out more, I'm busy with something. The filter-branch man page is very comprehensive, though.	16:55
	→ gb\|work joined	16:56
	← gb\|work left	16:56
	→ ferdy joined	16:56
j416	peper: basically, just check where GIT_AUTHOR_NAME (or what it is you're checking) is what you are after, and then set it in your command.	16:56
	← ferdy left	16:56
j416	maybe you need to export it as well, I don't remember.	16:57
	→ ferdy joined	16:57
peper	j416: the manpage seems to be using ` ` for quoting	16:57
	→ gb\|work joined	16:57
peper	yeah, i need to export it	16:57
	just was confused b/c shell expands ` `	16:57
	← bcardarella left	16:57
	→ markelikalderon joined	16:57
	→ IRSeekBot joined	16:57
yoann	drizzd: is it possible that the detection fail because I modified the file in the same commit I moved it ?	16:57
j416	peper: not ``, ' '	16:57
drizzd	yoann: I just tried a merge with 77% similarity and it worked.	16:57
	← fujin left	16:57
peper	j416: exactly	16:58
j416	peper: my man pages use ' '	16:58
	maybe you are looking at some malformatted version	16:58
	→ jcm_ joined	16:58
	j416 off to dinner	16:58
drizzd	yoann: no, if you had not modified it you would have 100% similarity and the rename is guaranteed to be detected. But 87% is more than enough also.	16:58
	← qrush left	16:59
	→ s0ber joined	16:59
	← Hendu left	16:59
peper	j416: hmm, git filter-branch --help	16:59
	git filter-branch --index-filter ´git rm --cached filename´ HEAD	16:59
	oh that's not `` ;]	16:59
	→ dohzya joined	16:59
drizzd	yoann: can you share those changes? I only need the merge-base version and the tips	16:59
	to get the merge-base do "git merge-base branchA branchB"	17:00
	← Dreamer3 left	17:00
	← aspotashev left	17:00
	→ aspotashev joined	17:00
yoann	drizzd: have to go right now, but I'd be happy too	17:01
	→ bcardarella joined	17:01
	← moonflux left	17:01
	← tvw left	17:01
yoann	drizzd: what is the tips ?	17:01
drizzd	yoann: the heads, the branches	17:01
	← WALoeIII left	17:01
drizzd	the current version in each branch	17:01
	ciao	17:02
yoann	drizzd: the merged branche contain some private code I unfortunately can't share :/	17:02
	← kraymer left	17:02
drizzd	yoann: I only need the file of interest	17:02
yoann	drizzd: I'll try to provide a reproducable example	17:02
drizzd	ok, great	17:02
	→ kraymer joined	17:02
yoann	drizzd: thank you very much for your help	17:02
drizzd	np	17:03
	← riovena left	17:03
	→ killerchicken_ joined	17:03
	← javatexan left	17:04
	→ Octalot joined	17:07
	← voker57 left	17:07
	→ cilly joined	17:09
	→ voker57 joined	17:10
	→ k0001 joined	17:10
	→ jspiros joined	17:11
	← bcardarella left	17:12
	jefftriplett-afk → jefftriplett	17:13
	→ argonel joined	17:16
	→ mfilipe joined	17:16
peper	hmm	17:16
	← tono left	17:16
peper	can i somehow rewrite commiter too?	17:16
	ah, committer	17:17
	→ aspotashev_ joined	17:17
	← aspotashev left	17:18
	← zuez_ left	17:19
	→ stuffcorpse joined	17:19
	→ zuez joined	17:19
	→ boto_ joined	17:19
	→ faithfulgeek joined	17:21
	→ jonshea joined	17:21
	← agib left	17:21
	→ agib joined	17:22
	→ lightcap joined	17:22
	→ Shadow_Master joined	17:23
	← gitte left	17:23
	→ tvw joined	17:24
j416	peper: man git-commit-tree	17:24
Gitbot	peper: the git-commit-tree manpage can be found at http://git.or.cz/man/git-commit-tree	17:24
j416	shows a list of variables	17:24
	← cilly left	17:25
	→ Dashkal joined	17:25
peper	j416: i did it with filter-branch --env-filter too. just had a typo ;]	17:25
j416	peper: great :)	17:26
	filter-branch is awesome	17:26
	→ Andurin joined	17:26
	← countd left	17:27
	← boto left	17:30
	→ doener joined	17:30
	→ Zaba_ joined	17:30
	→ abstrusething joined	17:32
	→ sgrimm joined	17:33
	← galderz left	17:33
	← gwoo left	17:33
	→ willb joined	17:36
	→ radarek joined	17:37
	← nighthwk1 left	17:39
	→ qrush joined	17:39
	→ javatexan joined	17:39
	→ nighthwk1 joined	17:40
	← literal left	17:40
	→ bobesponja joined	17:41
	← Zaba left	17:43
	← Andurin left	17:44
	← lightcap left	17:44
	→ bcardarella joined	17:48
	← Alinermedia left	17:48
	→ lightcap joined	17:50
	→ tedoc2000 joined	17:51
	← faithfulgeek left	17:51
	← lightcap left	17:51
	← cocksr8 left	17:51
	← markelikalderon left	17:53
	→ ceej joined	17:53
	→ mczepiel joined	17:53
	→ Dave2 joined	17:53
	→ jayne joined	17:53
	→ Tv joined	17:53
	→ warthog9 joined	17:53
	→ bleything\|away joined	17:53
	→ BlindOracle joined	17:53
	← doener left	17:54
	→ doener joined	17:54
	Dave2 → Guest59324	17:55
	jayne → Guest46827	17:55
	→ towski joined	17:56
	← aziz left	17:57
	← nighthwk1 left	17:57
	← bentob0x left	17:57
	→ j416_ joined	17:58
	← mczepiel left	17:58
	→ markelikalderon joined	17:59
	→ thiago_home joined	17:59
	← lorandi left	18:00
	← j416 left	18:00
	→ lorandi joined	18:00
	j416_ → j416	18:01
	← perlmonkey2 left	18:01
	→ khmarbaise joined	18:01
	← lorandi left	18:01
	→ lorandi joined	18:04
	Guest59324 → Dave2	18:04
	→ ph^ joined	18:07
	← abstrusething left	18:07
struberg	spearce I did run mvn clean install and got an assertion in testCache_TooSmallLimit(org.spearce.jgit.lib.WindowCacheGetTest) in .checkLimits(WindowCacheGetTest.java:112)	18:08
	tried to reproduce but all built well the 2nd time ^^	18:08
	→ robinr joined	18:09
spearce`	struberg: i saw that once too. but only once. there's some sort of randomness in there. :-\	18:10
	spearce` → spearce	18:10
	→ gwoo joined	18:11
	→ amystrat joined	18:11
	→ faithfulgeek joined	18:11
amystrat	How can I find the commit where a certain file was removed?	18:12
doener	amystrat: git log --diff-filter=D -- file	18:12
struberg	spearce yea looks like it's stochastic, some timing issue maybe. such things are hard to detect, but usually hits you even harder if they appear ;)	18:12
doener	amystrat: add other options or committish arguments as needed	18:13
amystrat	awesome thx!	18:13
	← quoin left	18:14
spearce	struberg: oh, right. so this test is checking that the window cache didn't exceed the memory usage limit set by packedGitLimit. but the assert fail means we did in fact exceed it.	18:15
	the window cache can temporarily exceed the limit for a short time between when a new entry is inserted and when the expire is supposed to start flushing older stuff.	18:15
	but this is a single threaded junit test; we should have by the end of the test expired enough to remain under the limit.	18:16
	← bcardarella left	18:16
spearce	a bug. and it worries me a bit in that we may be able to exceed the packedGitLimit at other times.	18:16
struberg	spearce: can this 'expire' code be triggered from the outside?	18:16
mutex	hrm	18:16
spearce	no; its not exposed - its all private. in theory yes, we could expose the method, but callers shouldn't have to manage it.	18:16
mutex	I see remote tags, but git fetch --tags does not seem to fetch them	18:17
	am I missing something ?	18:17
	→ cpg joined	18:18
robinr	spearce: we could perhaps make it package private so unit tests can access it	18:18
spearce	robinr: but my point still holds, the unit test in this case shouldn't have overshot.	18:18
	explicitly running the expire is just a bandaid.	18:19
robinr	ah	18:19
struberg	robinr spearce imho we shouldn't need to change the package structure for testing	18:19
	so maybe a sleep(exceedtime) or something?	18:19
	→ bcardarella joined	18:19
context	this refers to msysgit (so you prolly dont care) but i check something out and it still says its different	18:20
struberg	robinr before I change interfaces and expose internalish stuff, I go the reflextion way ;)	18:20
context	i imagine this is lineendings, but how can i fix it	18:20
spearce	the only reason the cache is allowed to temporarily overshoot is for a multi-threaded case where the are concurrent cache insertions. they can temporarily overshoot while the disk io is occurring. but they should eventually settle back out to within the limit.	18:20
robinr	struberg: it's not really exposed if it's package level. Besides it's a common idiom for providing access to unit tests	18:21
	→ dreiss joined	18:21
	← scarabx left	18:22
spearce	sigh. today i found out jgit Patch can't parse a --unified=0 diff correctly; the EditList comes out wrong. my unit tests aren't complete enough for that. (but i think its 100% coverage in there, so, yea, stupid)	18:22
robinr	well 100% coverage is a proof that it's fully tested	18:22
	<100 is proof that it isn't	18:23
	brennen → brennen\|lunch	18:23
robinr	besides the junit tests doesn't know what it's supposed to do	18:23
	← rubydiam_ left	18:23
	← amystrat left	18:23
robinr	hmm, I should have seen that	18:23
	do you mean the patches I did not apply yet?	18:24
spearce	robinr: no, i think you've applied everything of mine except that sun bug link	18:24
robinr	oh, yea, I did apply	18:25
spearce	there patch.EditListTest (to test the new toEditList() method) doesn't parse a unified=0 format patch. such patches have the starting line number -1 (due to the standard link gitster mentioned on list) so the edit list comes out as number - 2 instead of number - 1.	18:25
robinr	without reading as much as I had intended to do..	18:26
spearce	yea, its hard to keep up with me when i'm doing this at day-job. :)	18:26
	robinr goess off to see dirty jobs	18:27
robinr	best tv show there is	18:27
	→ drizzd_ joined	18:27
	→ jnareb joined	18:28
	← towski left	18:29
	← faithfulgeek left	18:30
struberg	spearce pushed SimpleRepository#push to my repo and no will start with maven-scm-providers-jgit finally :)	18:30
	← bcardarella left	18:30
	→ almostautomated joined	18:31
spearce	heh. nice.	18:31
	→ willb1 joined	18:32
	→ aziz joined	18:33
	→ bcardarella joined	18:33
almostautomated	Hi all! A curious question; I've initialized a git svn repo and then done git svn fetch and it is taking a long time compared to git svn clone; and by a long time I mean git svn clone takes just a couple of minutes and now I'm sitting on a good 15 minutes with fetch. I've never used the setup of git svn -s init then git fetch before. Is this normal?	18:34
	→ mjf joined	18:34
jnareb	Till 23 May (according to GSoC timeline, http://tinyurl.com/d6m3x6) there is community bonding period... but as far as I see there was no presentation of 'domestic' and 'foreign' GSoC students...	18:34
	→ davidfetter joined	18:35
	→ mountie joined	18:36
	→ Alinermedia joined	18:36
	← mountie left	18:36
cxreg	anyone know bzr and hg along with git?	18:38
	← willb left	18:38
cxreg	is it a pain keeping them all straight in your head, if so?	18:39
	← FFighter left	18:39
	→ saurabh1403 joined	18:41
	→ willb joined	18:41
	← drizzd left	18:43
	→ fanf2 joined	18:45
jnareb	cxreg: there are a bit different... why you need to use -all_ of them, anyway?	18:46
	cxreg: you can try to find a common wrapper around them	18:46
	→ dythzer joined	18:46
	← struberg left	18:46
jnareb	cxreg: but there is always a problem with it being lowest common denominator in features...	18:46
cxreg	I don't need them all. But various projects use of course just one of them, and all 3 are growing in popularity	18:48
	I'd rather not use wrappers, knowing your tools is a good idea	18:48
	I guess it's not that different from knowing multiple languages	18:48
dythzer	Hi, how do you make command abbreviations in git? for example "git co" instead of "git checkout"?	18:49
	→ astrocub joined	18:49
Grum	dythzer: look into aliases in man git-config	18:49
Gitbot	dythzer: the git-config manpage can be found at http://git.or.cz/man/git-config	18:49
	→ khmarbaise_ joined	18:49
dythzer	ok, thanks :)	18:49
Grum	it should be something like: git config alias co checkout (orso)	18:50
	alias.co checkout to be accurate	18:50
	→ nighthwk1 joined	18:51
	→ thirsteh joined	18:52
	→ gotgenes joined	18:52
	← Bass10 left	18:52
	→ bentob0x joined	18:52
	← aspotashev\|eeepc left	18:55
	→ lhz joined	18:55
	→ paltman94 joined	18:57
	← khmarbaise left	18:57
	← willb1 left	18:57
	→ _grepper joined	18:58
	→ tmz joined	18:58
_grepper	how can I get the contents of a single file on another branch?	18:58
	→ wwwald joined	18:59
_grepper	git checkout <branch> -- /path/to/file ?	18:59
	I think it worked :)	18:59
	this channel is so good that it makes us answer our own questions -- wow	18:59
	← bobesponja left	19:00
	→ Lo-lan-do joined	19:01
Lo-lan-do	Hi all. I'm trying to debug dulwich and/or bzr-git, and I'm getting a "protocol error".	19:02
	Apparently the (dulwich/bzr-git) server sends "e HEAD\0multi_ack side-band-64k thin-pack ofs-delta\n".	19:03
	→ mib_xdoycc joined	19:03
Lo-lan-do	I assume the "e" is a shortened sha, since I get other hex digits on other repositories.	19:03
	← lhz left	19:04
Lo-lan-do	Am I right in assuming these shortened shas are not supposed to go on the wire?	19:04
	→ eshear joined	19:04
	← Alinermedia left	19:05
	← paltman94 left	19:05
	→ Alinermedia joined	19:06
spearce	Lo-lan-do: correct.	19:08
	→ r0bby joined	19:08
spearce	the protocol says the SHA-1 must be a full SHA-1, all 40 digits	19:08
Lo-lan-do	spearce: Thanks.	19:08
	← DoctorPepper left	19:11
	← eshear_ left	19:13
Ilari	Hmm... It sends side-band-64k but not side-band... Has the server seen client caps yet at that point?	19:13
	→ Dreamer3 joined	19:14
spearce	Ilari: no. the server is only supporting the new style side-band format.	19:14
	you don't have to send both. if you are a new server implementation, only supporting side-band-64k may actually be reasonable.	19:14
	← j416 left	19:14
Ilari	spearce: Yeah. its probaby has been supported for a while... Isn't the only real difference that original one may only send 4Ki at time?	19:15
	→ ivenkys joined	19:15
spearce	1000 bytes at a time. but yes.	19:16
	→ sdboyer-laptop joined	19:16
spearce	side band 64k increases it to 65520 bytes per block	19:16
	older git's read into a 1000 byte long buffer in the stack i think, so larger blocks didn't work. ;-)	19:16
	← saurabh1403 left	19:17
	→ gitte joined	19:17
Ilari	Where does that 16 bytes come from? 4 bytes for length, 1 for largest encodeable value and 1 from subchannel ID, but where does that 10 more bytes come from?	19:17
	← Tv left	19:17
	← lorandi left	19:18
spearce	its only a 5 byte header for a packet: 4 bytes for the hex encoded length, 1 byte for the subchannel id.	19:18
	why the cap is 65520 and leaves 11 bytes of space, i don't know.	19:19
	→ hcl2 joined	19:19
hcl2	i'm using git svn for the first time, and i just did a dcommit, now i have a branch called "-". did I make that or is that part of the svn bridge?	19:19
	← doener left	19:20
spearce	from dulwich: "WTF: Why have the len in ASCII, but the channel in binary."	19:20
	spearce has wondered that one for years too	19:20
	→ tal67 joined	19:20
EvanR	can someone help me with repo.or.cz, cant seem to push	19:21
alezandro	all: fetched a branch from another developer, those changes went to FETCH_HEAD in my local branch, which has my own changes... i'd like to try those two changesets together without commiting.. what would be a good way to do that?	19:22
spearce	Ilari: the other thing that's weird there is the packet size is 65520, but that has to include the 5 byte header, so really the cap is 65520-5 = 65515 bytes/packet.	19:22
jnareb	spearce: this weird ASCII/binary is in other places too (c.f. tree format and object format)	19:23
	spearce: probably for easier debugging	19:24
spearce	jnareb: yea, i know. tree specially with the id in binary and mode/name in string form.	19:24
	→ seangrove joined	19:24
spearce	yet commit is all ascii	19:24
	→ catalis joined	19:24
Ilari	And then there's the special '0000' packet...	19:24
spearce	(well, critical headers anyway)	19:24
	yea, not to be confused with the "0004" packet which takes up the same network space but has different meaning. like that your sender is braindead.	19:25
Ilari	Isn't it actually ASCIIZ string?	19:25
spearce	oddly, no. the \0 isn't required in the string. but advertisement lines for refs end with \n for no reason other than to make it easier to read the slop when you run "git upload-pack ."	19:26
	← tal67 left	19:27
	← Dreamer3 left	19:28
Ilari	Well, there doesnt' seem to easy way to sanely encode trees in ASCII, since the only invalid character in names is NUL... And escaping some characters there would likely be too slow.	19:28
	brennen\|lunch → brennen	19:28
	→ scarabx joined	19:28
	→ nurey joined	19:29
	→ rolfb joined	19:29
	← rolfb left	19:29
spearce	oh, yea, the tree "mode name" string terminates in \0 as its the only sane delimiter at that point in the data.	19:29
Ilari	mode is terminated by space?	19:31
	Guest46827 → jayne	19:31
spearce	yes	19:31
	→ neoeinstein_ joined	19:31
gitte	Hey spearce!	19:32
spearce	hey gitte	19:32
gitte	How's life on the other ocean's shoreline?	19:32
spearce	nice. good weather.	19:32
	been swamping poor robinr in patches.	19:33
	:-)	19:33
gitte	spearce: it started to turn spring here, but now it reverted to rain.	19:33
	spearce is impressed with how far dulwich has gotten	19:33
gitte	spearce: I am working quite some on the diff cleanup.	19:33
	dulwich? that pure-python clone?	19:33
spearce	yea.	19:33
	its catching up to jgit in level of features implemented.	19:33
gitte	I was somewhat unimpressed that they seemed to distance themselves as far as possible from the Git mailing lit.	19:33
spearce	there's a good part of the network protocols done.	19:34
Ilari	Does it get v2 pack deltas vs. v3 pack deltas right? :-)	19:34
gitte	including pack generation?	19:34
spearce	pack generation is only starting. it can produce a basic pack from what i read in the code.	19:34
gitte	spearce: but without deltas, I assume.	19:34
spearce	no, and it also lacks delta reuse.	19:34
	or whole object reuse for that matter.	19:35
	→ ukleinek joined	19:35
	← khmarbaise_ left	19:35
gitte	the latter being in JGit already...	19:35
	(I meant the delta reuse)	19:35
spearce	right.	19:35
	→ khmarbaise joined	19:35
Ilari	Can JGit do full-object reuse from loose objects with packlike headers?	19:35
ukleinek	can anybody tell me the preconditions and expected results for merge stategies?	19:35
spearce	Ilari: no. new-style packlike loose object headers can't be written by any current git implementation, and reuse was removed from git-pack-objects in the c code before jgit really had reuse working.	19:36
gitte	ukleinek: yes ;-)	19:36
	ukleinek: I would imagine Documentation/merge-strategies.txt to take care of most of that.	19:36
ukleinek	oh, the man with the umlauts	19:36
spearce	Ilari: i also forget the difference between v2 and v3 deltas, and i can't see anything in JGit's BinaryDelta.apply logic about it, so i can't answer that question about dulwich's code. :-)	19:37
gitte	ukleinek: to be honest, I half-expected you to insist on umlauts in IRC.	19:37
cehteh	öh?	19:37
gitte	ukleinek: but you stopped short of one... scared?	19:37
	cehteh: ukleinek got married only so he could take on a surname with umlauts :-P	19:37
ukleinek	gitte: I tried /nick ukleinekönig, but it didn't work :-)	19:37
	← bcardarella left	19:38
cehteh	oh hi ukleinek :)	19:38
ukleinek	cehteh: in contrast to gitte I don't keep my umlauts secret	19:38
	cehteh: but shhhhh	19:38
gitte	ukleinek: but you make others' secret umlauts public. Shame on you!	19:39
	ukleinek: seriously again, does that document help you?	19:39
ukleinek	gitte: so I guess I don't get an explanation by you	19:39
Ilari	The difference involves interpretation of bit 6 in copy opcodes. V2 had source select there. V3 signals presence of copy size bits 16-23.	19:39
cehteh	nicks in unicode would be fun	19:40
	☠☠☠	19:40
spearce	Ilari: so yea, i went and looked it up myself. neither jgit nor dulwich support v2 copy-from-dest.	19:40
	→ emiltin joined	19:40
ukleinek	gitte: I already saw that, and no it didn't help.	19:40
	→ _Vi joined	19:40
ukleinek	gitte: I want to know how to use them	19:40
gitte	ukleinek: I will not give you a run-down of the complete merge story, no, I do not have enough time.	19:40
emiltin	hi. if i accidently deleted a file in my working tree, how do i restore it from the latest commit?	19:41
almostautomated	When an svn repo has had its sturcture changed at some point from a flat to standard layout, and a git clone is done on the (now new) trunk branch; is there way to to have the git repo get all the history for items in the trunk farther back the the 'move' to the 'trunk' branch?	19:41
Ilari	emiltin: 'git checkout HEAD -- file'?	19:41
gitte	ukleinek: but if you have specific question, I will answer them (remember: I did not try to start an umlaut war .-)	19:41
	← qrush left	19:41
ukleinek	things I found out are: exitcode=0 == success, 1 == conflicts, 2 == strategy cannot handle merge case	19:42
	gitte: do I need index==HEAD?	19:42
gitte	ukleinek: only for the files which are touched by the other side.	19:42
spearce	Ilari: hell, even git.git can't do V2 source-select.	19:42
	→ uroboros joined	19:42
	← uroboros left	19:42
emiltin	Ilari: ok thanks. --file is simply the nam of the file, or i need to pass that switch?	19:42
	← radarek left	19:43
ukleinek	emiltin: note the space between -- and file	19:43
brennen	speaking of return values from git commands, where in the docs should i look for them, generally?	19:43
Ilari	emiltin: Note the space there.	19:43
emiltin	aha ok!	19:43
brennen	exit codes. whatever the term i'm wanting is...	19:43
	→ struberg joined	19:44
ukleinek	gitte: and what is the state of index and wc after a merge strategy was run?	19:44
	index contains the merge result, and wc == index if ret==0?	19:45
gitte	ukleinek: modulo the changes wc had before, yes.	19:45
ukleinek	gitte: thanks	19:45
gitte	ukleinek: when there were conflicts, the wc files contain the conflict markers.	19:45
	ukleinek: except in the case of binary files.	19:45
	→ daryn joined	19:45
Ilari	ukleinek: And presumably index entries are collapsed to stage 0 for files that automerged successfully.	19:45
gitte	ukleinek: there, you will have file~HEAD and file~deadbeef versions in the wc.	19:46
jnareb	What does 'comminity bonding' period in GSoC consist (should consist) of? And should we expect current GSoC students presenting their projects and themselves on git mailing list? What about 'foreign' GSoC projects touching git?	19:46
gitte	jnareb: bonding is supposed to start after exams.	19:46
ciaranm	the 'community bonding' period in gsoc is mostly a farce	19:46
smtms	it is not	19:47
	→ paltman94 joined	19:47
gitte	jnareb: and it basically means that students get to know their mentors better.	19:47
smtms	their community better	19:47
ciaranm	it's part of google's "we're trying to get people involved in open source, not trying to improve open source" thing	19:47
jnareb	ah	19:47
gitte	jnareb: and only complete morons underestimate the importance of this bonding period :-)	19:47
ciaranm	the reality is, anyone who isn't already familiar with the whole thing isn't going to get anywhere with their project	19:47
smtms	ciaranm, the whole thing?	19:48
ciaranm	the code, the development process and whatever passes for a community	19:48
smtms	ciaranm, are you speaking from experience?	19:48
gitte	ciaranm: if I do not have a clue, I try to hide the fact from the public instead of blurting it out.	19:48
ciaranm	smtms: yes, i am. i've been involved in these for quite a few years.	19:48
jnareb	BTW. I tried to convince student from openSUSE build system (to be in git) to contact git mailing list if he has problems (on neighbour IRC channel). We will see if I succeded.	19:48
gitte	ciaranm: maybe you want to do the same.	19:48
	jnareb: that is a good thing!	19:49
ciaranm	gitte: if you believe someone who isn't already familiar with a project is going to come along and get something useful done, you're going to end up sorely disappointed	19:49
gitte	jnareb: FWIW the Mono student seems to be very responsive (unlike his mentor, or for that matter, last two years' mentors for Mono).	19:49
jnareb	although I guess that most help would need git# guys...	19:49
	gitte: nice!	19:50
ciaranm	with all the soc stuff i've been involved in, the people who ended up delivering something useful or even just sticking around afterwards were people who were already effectively part of the project	19:50
gitte	jnareb: it probably helped tremendously that spearce invested some of his scarce time to relate important lessons from JGit's history.	19:50
smtms	ciaranm, there are different levels of familiarity	19:50
	ciaranm, tell Google to discontinue this unsuccessful experiment	19:50
ukleinek	git-merge-recursive does some merging internally (for the bases). Can I access that somehow from sh?	19:50
gitte	ciaranm: did you live under a rock the last two years?	19:50
	ukleinek: no, you can only redo it yourself.	19:51
jnareb	ciaranm: I think that last year very succesfull GSoC students (JGit push, merge builtin) weren't active Git contributors before...	19:51
gitte	ukleinek: using "git merge-base -a"...	19:51
ciaranm	smtms: google are well aware that the people who deliver successful projects are those who were already involved. they're just trying to change that rather than accepting it and changing the process.	19:51
jnareb	ciaranm: ...and are now	19:51
gitte	jnareb: the merge builtin was done by someone we saw before, but basically all the others were Git newbies.	19:51
	ciaranm: it seems you did not get my message.	19:51
ukleinek	gitte: oh, I think you got my question wrong (probably my failure)	19:52
gitte	ciaranm: why don't you just say it directly? So far, your statements were provably wrong. And not only a bit.	19:52
ciaranm	it seems you're trying to find a way of thinking google gets it entirely right by looking at as small a sample as you can find	19:52
gitte	ukleinek: so, what type of access to the inner workings do you want?	19:52
	← harinath left	19:53
	← daryn left	19:53
ukleinek	gitte: I do many merges and expect nearly all of them to work without conflicts. So I want a command that does this internal merging for trees I specify	19:53
	gitte: is this clearer?	19:53
gitte	ukleinek: you mean you do not want to touch the working directory?	19:53
ukleinek	... and only if the merge fails I want to update the wc	19:53
	gitte: yes	19:54
gitte	ukleinek: you might be interested in "git merge-tree".	19:54
spearce	oh, ugly, dulwich doesn't complete a thin pack when it stores it to disk.	19:54
gitte	ukleinek: AFAIR it does not touch the working directory at all.	19:54
	spearce: uh oh.	19:54
	← gotgenes left	19:55
gitte	ukleinek: and if merge-tree fails you can still resort to calling merge-recursive, which will do the same stuff, but touching the working directory.	19:55
Ilari	Wonder how modern versions of git would react to such pack? Loud complaints for sure.	19:55
	ukleinek goes and does some experiments, thanks	19:55
gitte	ukleinek: yw.	19:55
	← emiltin left	19:55
spearce	Ilari: they will fail to resolve the object that is missing its delta base...	19:55
Ilari	spearce: Previous versions would totally fail. But newer ones have alternate-storage search, so I though that might make a difference...	19:57
	← khmarbaise left	19:58
jnareb	ciaranm: see http://git.or.cz/gitwiki/SoC2008Projects	19:58
Ilari	spearce: BTW, does the alternate representation search search through all possible paths when trying to reconstruct object with bad base?	19:58
ciaranm	looking back... every single successful gentoo summer of code project has been from someone who was already heavily involved in gentoo. every person who wasn't already well known ended up failing and not coming back, as opposed to only half for existing contributors	19:58
Ilari	spearce: I guess no...	19:58
spearce	Ilari: it should try everything. but i'm not that familiar with that particular code in git.git; nico wrote it and i haven't peeked in those areas in a while.	19:59
smtms	Gentoo is representative of GSoC as a whole?	19:59
ciaranm	smtms: more so than git, given its larger sample size	20:00
killerchicken_	I think people with experience working on the project have a way higher chance to succeed. But for my org last year, I was the only one who had experience with the project before, but nevertheless 4/7 students passed.	20:01
gitte	killerchicken_: my experience is that the likelihood of success depends very, very much on the enthusiasm of the student and the quality of the mentor.	20:02
	killerchicken_: if the mentor neglects her duties, chances of success are pretty low.	20:02
killerchicken_	gitte: true, too. I feel that a student who has worked with the project will usually have a certain level of enthusiasm to start with.	20:03
gitte	killerchicken_: a lot of students do not have time during their terms to get involved with Open Source projects.	20:03
	← paltman94 left	20:04
	→ auxbuss joined	20:04
smtms	a student that is already involved with a project can compensate for a lacking mentor	20:04
killerchicken_	hm, ok. For me, it was different anyways, my term continues until mid-july	20:04
gitte	smtms: only so far.	20:04
	→ porcnick joined	20:05
ciaranm	google's aim of bringing in new people to open source is largely unsuccessful. they don't offer enough money to get many people who aren't already open source people to join in, which means at best they can persuade some people who do open source anyway to put a bit more time into it and maybe look at a second project for a while.	20:05
	← dythzer left	20:05
killerchicken_	ciaranm: for me, they offer enough money so I can do what I'd love to do anyways (contribute to Tor), while not being forced to take another job during the summer	20:06
jnareb	we have some very good mentors (and happened on good students)	20:06
smtms	a student that is already involved compared to one that isn't is more likely to succeed with a low-quality mentor, that's what I'm saying	20:06
	→ laeos joined	20:06
ciaranm	killerchicken_: that's the point. they're not bringing in new people. they're motivating existing people. but that's not how the program is structured.	20:06
	boto_ → boto	20:06
gitte	smtms: I would not sign that statement.	20:07
ciaranm	half the questions on the org application form are about how you're going to get people to join, integrate them into the community and persuade them to stay on	20:07
killerchicken_	ciaranm: works for me ;)	20:07
gitte	smtms: every student needs guidance.	20:07
killerchicken_	Tor has a few people who were brought in through gsoc	20:07
	← bentob0x left	20:07
jnareb	ciaranm: btw that might depend on the organization and projects/ideas proposed: how steep is 'barrier to entry'; in particular do you have to have at least some knowledge of org code to be able to do a project in a summer	20:08
gitte	smtms: and that is actually part of the deal with Google: they pay a lot of money, the students do a lot of work, but the organization's responsibility is to provide fun and knowledge.	20:08
	smtms: a project that does not require a mentor is therefore not a good Google project; it is also more likely to be a crap project.	20:08
jnareb	not all SoC students were retained in GSoC 2008, unfortunately...	20:09
gitte	jnareb: a major goal of GSoC, yes.	20:09
	jnareb: some even had the guts to come back when there was a prospect of money.	20:09
spearce	jnareb: i wonder how our SoC retention compares to overall retention of people who send patches to gitster, me, paulus, eric wong, etc.	20:09
gitte	spearce: as we have currently 0 students from GSoC, I think it is clear...	20:10
	→ qrush joined	20:10
	→ jceb joined	20:11
	→ willb1 joined	20:12
	← drizzd_ left	20:12
jnareb	so the 'bonding' period is mainly about getting students and mentors to know each other, and to bring students knowledge up to point where they would feel comfortable to communicate with the rest of developers (in case of Git: post on git mailing list and send patches there), isn't it?	20:12
gitte	jnareb: that is my understanding, yes.	20:12
jnareb	(and create public repository somewhere: repo.or.cz, gitorious, github, kernel.org, ...)	20:13
ciaranm	anyone who's not already comfortable sending patches isn't going to get very far	20:13
gitte	jnareb: and as you probably know yourself, it is much easier to work with people you know and like, so that is also an important part of the bonding period.	20:13
	→ RayMorgan joined	20:13
jnareb	ciaranm: there are also SubmittingPatches and CodingGuidelines conventions for each project...	20:14
	→ gb\|ln joined	20:14
ciaranm	jnareb: and anyone who's going to be able to get anything useful done will either already know them or be skilled enough to be able to pick that sort of thing up straight away	20:15
	← willb left	20:15
	→ drizzd joined	20:15
RayMorgan	is there anyway when doing a merge to ignore DOS style carriage returns? It is causing the file to conflict on every line	20:15
	→ gotgenes joined	20:16
jnareb	RayMorgan: I thought that git ignores end-of-line differences during merge (although you can have mixed eoln file with conflicts), isn't it?	20:16
gitte	RayMorgan: best bet is to commit a DOS->Unix only commit on the offending side prior to the merge.	20:16
	← porcnick left	20:18
	← mib_xdoycc left	20:18
	→ piratepete joined	20:19
	→ Modius_ joined	20:19
	→ neoeinstein joined	20:20
	→ jerojasro joined	20:20
jnareb	me wonders a bit about 'Interactive graph GUI' project for Git GSoC	20:21
gitte	jnareb: what about?	20:21
jerojasro	hi all. I tried to cherry-pick a commit, but it caused a conflict. I don't want to solve the conflict, but to cancel the cherry-pick operation. how can I do that?	20:21
	← Weaselweb left	20:21
	← auxbuss left	20:22
	← gb\|work left	20:23
jnareb	gitte: is it known at least what programming language (and tooldkit, if appropriate) will be used for that?	20:23
	gitte: and what commands are to be supporter: cherry-pick, format-patch, history selection and simplification, rebase, interactive rebase?	20:23
gitte	jnareb: toolkit/language: same thing as I talked about in length when discussing the domain-specific merge helpers: wrong question.	20:24
jnareb	jerojasro: "git reset --hard HEAD" should do it (IIRC there is no "git cherry-pick --abort")	20:24
	← Modius left	20:24
gitte	jnareb: I am even quite sure that David is not aware that he is singularly responsible for Saurabh's failure to win a slot.	20:24
	→ xintron joined	20:25
jnareb	gitte: what do you mean: wrong question?	20:25
xintron	is it possible to clone a git repo into an existing folder somehow?	20:25
gitte	jnareb: the main focus now is not what the user will be able to do in the end, or what language to implement it in, but the _look_: how do we _present_ the data.	20:25
jnareb	ah.	20:25
gitte	jnareb: you have to climb a ladder from the bottom. Do not start with the uninteresting (because trivial) stuff.	20:26
jnareb	hmmm... it is easy to send patches or pull requests... it is harder to do whiteboard presentation via net...	20:26
	gitte: one interesting thing would be visual presnetation of "a --not b" rev-list specifier	20:27
	← Modius_ left	20:27
gitte	jnareb: the mentors also decided that the first round of mockups should be discussed in a small round (because we'd get too much bikeshedding otherwise).	20:27
	jnareb: but that is not something we need in the first steps, right?	20:28
	→ Modius_ joined	20:28
	← codebrulee left	20:28
jnareb	gitte: how do you plan to communicate mockups?	20:28
pasky	would that be something like git-viz?	20:28
gitte	jnareb: .png files. or .svg. Or whatever Jeff wants to provide.	20:28
pasky	merge by drag'n'drop!	20:29
almostautomated	Still trying to figure out how to have git svn get the history of a file across an svn move. Out svn repo had a flat layout and it was recently switched to trunk/branches/tag. I'd like to clone trunk, and get the history of objects prior to the move as well. Is that possible?	20:29
	Right now I'm trying a clone on the svnroot...	20:29
gitte	pasky: that is something I'd love to have by the end of the summer!	20:29
pasky	:)	20:29
	← Modius_ left	20:29
pasky	Add caching support to git-daemon !!!	20:29
	yaay	20:29
jnareb	switching branches and rewinding by double clicking?	20:29
	← astrocub left	20:29
gitte	pasky: Jeff even thought about using tcldot, and managed to learn enough Tcl in a few days to get an initial script going -- from any existing Git repository!	20:30
	→ Modius_ joined	20:30
gitte	jnareb: even interactive rebase.	20:30
pasky	cute	20:30
gitte	jnareb, pasky: but those things are relatively trivial compared to the major task.	20:30
jnareb	graphical rebase interactive (crossed with git-sequencer, which is unfortunately not finished)?	20:30
almostautomated	I'm guessing that after doing the clone of svnroot I'll have all commits, but won't have the trunk/branches/tags ?	20:30
	→ porcnick joined	20:31
gitte	jnareb, pasky: a sane and intuitive way to represent the history. gitk is not intuitive, as I had to realize the other day when I had to explain what everything means in that GUI.	20:31
	← lucsky left	20:31
gitte	jnareb: yes, sequencer is not finished. I am trying to get Chris to nudge his student, so far with little success.	20:31
jnareb	gitte: what major task? what is the _goadl_ of 'Interactive graph GUI' project? better understanding of operations, or how they work on DAG of commits?	20:31
gitte	jnareb: goal is an intuitive graphical interface.	20:32
	jnareb: and by intuitive I mean: even a complete moron does not need to read the manual.	20:32
jnareb	side note: did any of you perhance worked with AVS or AVS/Express (AVS = Advanced Visualisation System)?	20:33
	graphical interface for building visualisation pipeline...	20:33
gitte	jnareb: never heard of it.	20:33
	← Lo-lan-do left	20:33
	← qrush left	20:33
jnareb	(IIRC there is open source or free as in beer for students Khronos, which is similar but smaller)	20:34
	← xintron left	20:35
gitte	jnareb: you do not talk about Khoros, do you?	20:35
jnareb	in AVS and AVS/Express you in simplest case put on workspace 'source' block which e.g. read the file in some format, and 'sink' which displayed for example 3D graph or something, and connected them with a line; another source was simple widget providing integer selecting plane of cut or something	20:36
	← SRabbelier left	20:36
jnareb	Khoros, yes, probably	20:36
	→ s0ber_ joined	20:37
gitte	jnareb: I found that to be totally unintuitive, and my colleagues who have to work with a similar user interface completely agree.	20:37
jnareb	as an example of graphical interface to something programmatical, just like manipulation of DAG in git is (currently) CLI/programmatical	20:37
	→ docwhat joined	20:38
ciaranm	"the only intuitive interface is the nipple" etc. are you looking for something that involves as little learning as possible, or something that's as efficient and effective as possible for someone who knows what they're doing?	20:38
gitte	jnareb: the basic building block is a DAG, yes, but do we want to let the user adjust the layout? And if yes, how much?	20:39
	jnareb: those questions do not arise with Khoros.	20:39
	jnareb: also, in Khoros, the boxes have widely different purposes. There are data boxes, visualization boxes, action boxes...	20:39
	jnareb: in Git, we only have commits (well, if you really want, tags, too).	20:40
	← neoeinstein_ left	20:40
jnareb	gitte: one question for example with representing DAG is whether to mark edges of graph as directed	20:40
	→ qrush joined	20:40
gitte	jnareb: with an AG I would agree, but not with a DAG.	20:40
	← DavidKlein left	20:41
	→ ericindc joined	20:41
jnareb	I just hope to avoid WTF like the example from The Daily WTF where you programmed by editing UML-like diagram ....	20:44
robinr	depends on whether the graph is pack so one cannot know which way it runs	20:44
	s/packed	20:44
jnareb	gitte: but IIRC neither gitk nor qgit shows arrows...	20:44
gitte	jnareb: programming by editing the UML diagram is common practice.	20:45
jnareb	gitte: and github network graph neither	20:45
	→ DavidKlein joined	20:45
gitte	jnareb: just because gitk or qgit have that shortcoming does not mean we have to repeat the mistake.	20:45
jnareb	well, if you have strict ordering (top-to-bottom, or left-to-right) the direction of arrows is implied	20:46
gitte	jnareb: did I mention I want this thing to be _intuitive_?	20:47
	← wwwald left	20:47
gitte	jnareb: in such a case you do not do magical things. You lay out things plain and obvious.	20:47
	→ bcardarella joined	20:48
jnareb	well, direction of arrows is not obvious, be if (as it should be), from child to parent, or from parent to child	20:48
jepler	in gitk there's simply not enough space between revisions to have room for an arrow. the extra visual clutter would make it worse. In a graph produced by dotty there's typically a lot of space, and the arrow fits just fine	20:48
	look at both and see which is better	20:48
	in your app with your layou	20:49
	t	20:49
	gitte feels it was a wise choice to do the first round in a smaller circle ;-)	20:49
jnareb	top-to-bottom is nice because we are used to vertical scrollbars more	20:49
	left-to-right is nice because we are used to left-to-right axes and timelines	20:49
	← s0ber left	20:49
	← Modius_ left	20:49
	→ Radar joined	20:49
jepler	left-to-right may have cultural problems, not sure about top-to-bottom	20:50
ciaranm	left-to-right requires vertical labels, which gets tricky	20:50
	→ Modius joined	20:51
jnareb	example of top-to-bottom is gitk, qgit, and I suppose Giggle and GitX too	20:51
	→ petercoulton joined	20:51
	→ chmrr joined	20:52
jnareb	example of left-to-right is GitHub, somewhat Ohloh (pure timeline, no DAG), and most ASCII-art examples in manpages	20:52
ciaranm	top to bottom means you can fit the commit summary on the screen, something the manpage examples don't try to do	20:52
	← bcardarella left	20:52
jnareb	top-to-borrom is also tig, git log --graphs and git show-branch	20:53
ciaranm	the man pages deal with arbitrary 'A', 'B', 'C' commits, where you don't care what they are. that's not what happens when you're dealing with real data.	20:53
	→ v[V]\|per[X]x joined	20:53
jnareb	but in GUI they can be dots/boxes, expanding on mouseover (c.f. GitHub network interface)	20:53
ciaranm	i get the impression it'd be a lot more useful to have that information instantly there, rather than something you'd have to look for	20:54
	→ bcardarella joined	20:54
jnareb	I am not sure if we want/need in 'Interactive graph GUI' show summaries (i.e. be oneline log like)	20:54
	← priidu left	20:54
drizzd	if you want something intuitive you can't use text for 90% of the displayed area	20:54
ciaranm	how else would you identify commits?	20:54
drizzd	add detail on demand	20:55
	← bcardarella left	20:55
ciaranm	the other thing to bear in mind is that for git, you know a lot more about the shape of your data than is generally the case, which means highly generalised tools like graphviz aren't going to do a very good job of it	20:55
jepler	Recently I was trying to use 'git apply' to apply a non-git patch that removed some files. after 'git apply', how do I conveniently 'git rm' the right files? or is there a flag to 'git apply' to do this?	20:55
ciaranm	drizzd: you mean i'd have a pile of meaningless dots or boxes on screen, and i'm supposed to rebase / whatever based upon hunting around with tooltips?	20:55
drizzd	I think it would be much more useful if you have a simplified history at first. Gives you a better overview of the repository layout.	20:55
	ciaranm: if I rebase the thing I care about most is the branches, not the individual commits.	20:56
	→ Dreamer3 joined	20:56
ciaranm	so, what, you'd see a bunch of meaningless lines that sometimes join together?	20:56
	← d0k left	20:56
jnareb	drizzd: --simplify-by-decoration like?	20:56
drizzd	exactly, but they are not meaningless, because they correspond to branches/tags etc.	20:56
	and branch names can be much more easily fit on the screen than the log messages	20:57
	← sgrimm left	20:57
drizzd	jnareb: yes, only expandable on the fly	20:57
	← v[V]\|per[X]x left	20:57
ciaranm	this is sort of suggesting that a single view model isn't going to work, and having different displays depending upon what action you're trying to perform might be necessary	20:58
Ilari	jepler: --index?	20:58
ciaranm	one thing if you're working with branches, another for cherry picks, interactive rebases in progress and so on	20:58
drizzd	ciaranm: sure, I agree	20:58
jepler	Ilari: I'll try that next time. That last sentence makes it sound like it is what I am looking for.	20:59
	thanks	20:59
Ilari	Or worse, nonlinear interactive rebases... Not supported yet, but might in future...	21:01
gitte	Ilari: when I'm done with jgit diff, I will finish my rebase -i -p series.	21:01
	→ Modius_ joined	21:01
Ilari	gitte: I don't mean -i -p, but true nonlinear rebase...	21:02
Knirch	hmm, maybe I'm thinking incorrectly, but some people I work with are tracking a few versions of their software, 1.2, 1.3, 1.4. they keep cherry-picking commits from 1.2 into 1.3, won't that make the eventual merge really chaotic?	21:02
	← cannonball left	21:02
ciaranm	Knirch: not really. git doesn't do too bad a job of that.	21:02
gitte	Ilari: but -i -p _is_ nonlinear rebase.	21:03
	← commondream left	21:03
gitte	Ilari: and with my patches you _will_ be able to delinearize history.	21:03
	→ texel joined	21:04
Knirch	ciaranm: but it just feels relly illogical, and doesn't git take the whole chain into account when merging two branches, if you cherry picked stuff up the chain beforehand, won't that cause git to do a worse job at helping you merge?	21:04
almostautomated	Does this also pertain to 'svn move'? => From the git svn man page bugs section "...Renamed and copied directories are not detected by git and hence not tracked when committing to SVN..."	21:04
ciaranm	Knirch: nope. you're probably not clear on what merges are to git. try "git for computer scientists"	21:04
	← bdiego left	21:04
Knirch	ciaranm: will do	21:05
almostautomated	And does it pertain to checking out the svnroot when the svn repo layout changed from flat to standard t/b/t ?	21:05
	→ jonshea_ joined	21:07
	← jonshea_ left	21:07
	← Modius left	21:10
	← ph^ left	21:10
	← RayMorgan left	21:10
	→ RayMorgan joined	21:10
	← kumbayo left	21:11
Knirch	ciaranm: so if I grasp it correctly, git won't take the changes that led up to the current file into account when merging, so if two seperate branches would cause a conflict at the merge point, I'd be presented with the conflict as it is now, and not when the actual conflict appeared? (which kind of makes sense, I might have wrongly assumed it tried to help out in that aspect)	21:11
	← Aides left	21:11
	→ camwest joined	21:11
	← Modius_ left	21:11
ciaranm	Knirch: pretty much, yes.	21:11
	← eletuchy_ left	21:12
	→ eletuchy joined	21:13
Knirch	so basically, what my collegues are doing is only distrupting the visual history of the file, as they'll record a bunch of commits at two different points in time when they finally merge 1.2->1.3	21:13
	← tbf left	21:13
	← LiamH left	21:13
ciaranm	Knirch: correct. git doesn't really track cherry picks, so what you'll see is the same thing having been done in two different places in your history tree.	21:14
	← justatheory left	21:14
ciaranm	it's not really a big deal once you get used to it	21:14
	← ukleinek left	21:15
	→ tbf joined	21:16
	← cemerick left	21:17
EvanCarroll	I have a project that I've forked from tar ball, and now I want to merge with history into a pull of the official svn repository, so i essentially have two roots. How do I do this elegantly?	21:17
	presumably, the	21:18
	← Radar left	21:18
EvanCarroll	presumably, the files in the tarbal were at some point in the repo in the same fashion	21:18
	→ __iron joined	21:23
	→ JohnnyL joined	21:23
	← jonshea left	21:23
	← mbroeker left	21:24
gitte	EvanCarroll: you could import that tarball-based branch into your git-svn repository and then issue something like	21:24
jnareb	Knirch, ciaranm: well, git kinds of track^W detects cherry-pick in the form of patch-id during git-rebase...	21:24
gitte	EvanCarroll: git log --format:%H:%T \| grep :$(git rev-parse <root>:)	21:24
	EvanCarroll: where <root> is the initial commit of your tarball-based branch.	21:25
	EvanCarroll: this assumes, of course, that the HEAD is at the current svn/trunk or something close.	21:25
	→ LHC joined	21:25
EvanCarroll	it is, head is the current svn trunk	21:25
	→ rylar joined	21:25
EvanCarroll	but the tarbal wasn't of HEAD.	21:25
	the tarbal is just something i pulled down with git-cpan-init, rather than doing it the right way and cloning svn	21:26
LHC	anyone else discover git by watching that youtube vid	21:26
gitte	EvanCarroll: the idea being to search for a commit in the svn history that has the same tree as the tarball import.	21:26
	LHC: which one?	21:26
	← ceej left	21:27
LHC	http://www.youtube.com/watch?v=4XpnKHJAok8	21:27
gitte	LHC: it would have been enough to say "Linus' Git talk".	21:28
	LHC: yes, there were some people.	21:28
LHC	its better to be more interactive and link to video :P	21:28
gitte	LHC: I had to follow the link to know what you're talking about.	21:29
	LHC: hardly friendly for me.	21:29
LHC	maybe you never heard of linus git talk	21:29
	and you wanted to link to the vid. so one less potential step	21:29
gitte	LHC: I asked "which one" because I know of _more than one_ youtube video about Git.	21:30
Knirch	best would be "Linus' Git talk (ref: http://...)" imho	21:30
gitte	Anyway, off to some real work.	21:30
LHC	and I linked you to it :P	21:30
almostautomated	more than likely everyone in here has heard of Linus's Git Talk	21:30
	→ solydzajs joined	21:30
LHC	would it be safe to assume that? haha jk im just being a git	21:30
almostautomated	hardy har har :P	21:30
LHC	lol	21:30
	→ agile joined	21:31
	→ bcardarella joined	21:32
	← willb1 left	21:32
	← bcardarella left	21:32
	← robinr left	21:33
	→ bobmcw joined	21:33
	→ elmob joined	21:33
	→ jonshea joined	21:37
	→ fedalto joined	21:37
EvanCarroll	gitte: is there anyway to merge, heuristically, based on two seperate git repos without a common root, with two different directory structures, like if gitrepoa, is found in gitrepob's foo folder, can i do something like merge gitrepoo:/ gitrepob:/foo/	21:38
	sorry i'm just totally lost with how to reconcile these two repos	21:38
	→ altrux joined	21:38
	→ el_isma joined	21:38
el_isma	hello! Panic ON! how do I revert a revert?	21:38
DrNick	revert the revert	21:38
gitte	EvanCarroll: you should definitely find the corresponding svn commit first.	21:38
el_isma	DrNick: I guessed that. How?	21:39
DrNick	the same way you revert any other commit	21:39
jnareb	el_isma: published or not published? if published, revert, if not published rewind (e.g. via git-reset)	21:39
	← solydzajs left	21:40
LHC	just curious what do u guys do, IT too?	21:40
	← docwhat left	21:40
el_isma	I stupidly did this: git reset HEAD, git reset HEAD^, git reset HEAD^. Git warned me about changed files everytime. gitk now shows I lost 2 commits. The repo is private.	21:40
struberg	spearce any tips where I can look for something like git-ls-files in jgit?	21:40
EvanCarroll	gitte: the problem is the svn commit has a lot more sutff.	21:40
	it just isn't what the tarbal was generated from.	21:41
gitte	LHC: no, I am trying to take over the world, Pinky.	21:41
LHC	gitte, ok brain remember to bring the sense of humour, its killer bro	21:41
DrNick	el_isma: take a look at the ref log (git log -g), reset to the commit you want	21:41
gitte	EvanCarroll: if it has lot more stuff, that's no problem.	21:41
	← bdimcheff left	21:41
	← cpg left	21:42
el_isma	DrNick: thank you!	21:42
	← naeu left	21:42
	← qrush left	21:42
	← Dreamer3 left	21:43
cehteh	RandalSchwartz: #l	21:43
	ups :) sry	21:43
spearce	struberg: pgm.debug.ShowDirCache or some stuch.	21:44
	→ faithfulgeek joined	21:44
struberg	spearce txs!	21:44
	el_isma kisses git	21:44
EvanCarroll	gitte: how do i find the ref behind a tag?	21:44
gitte	EvanCarroll:do you mean rev?	21:45
	← el_isma left	21:45
	← RayMorgan left	21:46
	→ tectonic joined	21:46
	→ qrush joined	21:47
EvanCarroll	no, i mean for that command you pasted, git rev-parse <root>: I want the <root> which should be at tag 1.03	21:47
	→ malibu joined	21:47
	← camwest left	21:47
gitte	EvanCarroll: by <root> I meant the initial commit of your tarball-based branch...	21:47
	→ Dreamer3 joined	21:48
gitte	EvanCarroll: and the colon makes rev-parse return the tree.	21:48
malibu	Hi guys.. sorry for the lame question but I gotget.. Does git commit actually upload changes? or do I need to execute something else	21:48
	← DavidKlein left	21:49
	← jceb left	21:49
EvanCarroll	malibu: git push uploads changes	21:49
	malibu: git is decentralized commited changes does not mean uploading them	21:49
malibu	Oh yeah! Thanks	21:49
EvanCarroll	gitte: totally lost. I need a book on this	21:49
EvanR	when i do git push ssh://repo.or.cz/foo, it asks for me a password... but its supposed to use the ssh public key stuff	21:50
	or is it private...	21:50
EvanCarroll	gitte: so you want me to type in that command, in my svn cloned repo, with the root of my tarball fork, after I fetch? or something	21:50
EvanR	yes public	21:51
EvanCarroll	EvanR: it is supposed to use the public key, but repo.or.cz has to have the public key, or it can ask you for a password, but i don't think repo.or.cz is set to do that	21:51
gitte	EvanCarroll: yes, but this will not work (as you said that the tarball does not contain the same files)	21:51
	→ gb\|work joined	21:51
EvanCarroll	gitte: so what is my plan of action/	21:51
	← LHC left	21:51
EvanCarroll	is there anyway, I can just apply the merge history and deltas manually for whole directories that i explicitly specify.	21:52
	I wonder if git-diff will accept a different directory location	21:53
	← malibu left	21:53
	→ bcardarella joined	21:53
gitte	EvanCarroll: I'm a little bit too tired to repeat what I showed another guy last week, but it boiled down to "echo $root $(($(git diff --numstat $commit..$root \| cut -f 1,2 \| tr -c 0-9 +)0)".	21:53
	EvanCarroll: this same thing inside a "git rev-list svn/trunk \| while read commit; do ... done".	21:54
	EvanCarroll: and after that, a sort -k 2 -n.	21:54
	→ j416 joined	21:54
EvanR	EvanCarroll: i set up the right public key, im pretty sure. usedto work, now doesnt. keep asking for pass rather than just working. git push origin master. git push ssh://EvanR@.... same thing	21:54
gitte	EvanCarroll: the idea being to compare all svn commits with your root, and order the commits by the total number of added + removed lines.	21:55
	← j416 left	21:55
EvanCarroll	EvanR: did you ssh-add the public key, or does it have no password?	21:55
EvanR	no password	21:55
gitte	EvanCarroll: and if you're certain that there are more files in svn, you can add the diff option "--diff-filter=AM".	21:55
	← jnareb left	21:55
alezandro	another developer added a new branch on the origin, but when I do "git branch -r" I only see origin/master. What am I missing ?	21:55
EvanR	hdd crashed, had to regenerate keys, thats why it isnt working anymore	21:55
	i must have done something wrong wrt that	21:55
EvanCarroll	alezandro: did he push that branch, or did you pull that branch	21:56
	alezandro: you don't have to pull down all of his branches., and he doesn't have to push them all up.	21:56
alezandro	he pushed, and I can pull it, but it still doesn't show up on the list	21:56
	→ synsol joined	21:56
alezandro	it shows up as origin/new_branch on his "git branch -r"	21:57
	← spearce left	21:57
alezandro	but I only have origin/master	21:57
	← srid left	21:57
	← intripoon left	21:58
	← synsol left	21:59
	offby1` → offby1	21:59
	← gb\|ln left	21:59
	→ camwest joined	21:59
	→ synsol joined	22:00
	→ bjackman42__ joined	22:00
	→ spearce joined	22:00
	→ seva joined	22:00
EvanCarroll	gitte: i'm still staring at that	22:00
	← krh left	22:00
seva	i have a need to keep most of my project private and part (as in subdirectory) synchronized with external svn repo, any suggestions?	22:01
gitte	EvanCarroll: just call "git diff --numstat HEAD^!"	22:01
	EvanCarroll: you will see a machine-readable stat of the diff: #added #removed file	22:01
	EvanCarroll: (or #removed #added file, I always confuse that)	22:02
	Dreamer3 → Dreamer3\|optmizi	22:02
	Dreamer3\|optmizi → Dreamer3\|optmize	22:02
	← ceyusa left	22:02
gitte	EvanCarroll: the cut takes only the first two columns (i.e. #added and #removed)	22:02
	EvanCarroll: the tr turns non-digits (basically tabs and newlines) into + signs, and the trailing 0 is needed so that $((...)) can evaluate the sum (the sum must not end with a plus sign).	22:03
	← texel left	22:03
gitte	EvanCarroll: the "git rev-list svn/trunk \| while read commit; do ... done" executes that for every $commit.	22:04
	← struberg left	22:04
EvanCarroll	gitte: I think this is just too complex for git, or me, I'm thinking the merge history is just simply not worth it, git-cpan-clone is really cool in that it does everythign for you, and it you re-pull it even applies the later cpan tarball changes	22:04
gitte	EvanCarroll: and the sort -n -k 2 sorts by the second key, numerically (as opposed to lexicographically, where 10 sorts between 1 and 2).	22:04
EvanCarroll	i'm just looking for a way to revert back to the original repo	22:05
gitte	EvanCarroll: without the proper branch point I can only wish you good luck.	22:05
EvanCarroll	there isn't a proper branch point because the cpan tarbal was generated from ./POE, not /, which also has dev stuff	22:06
	and i don't think github will permit me to pull just /POE	22:06
	← catalis left	22:07
	→ rroblak joined	22:07
	← juanjoc left	22:08
gitte	EvanCarroll: but then you could compare just the subtrees "POE".	22:08
EvanCarroll	right, if i knew how.	22:08
	and i could only hope that cpan packager, used the whole subtree	22:08
gitte	EvanCarroll: IOW: T=$(git rev-parse <root>:POE); git rev-list svn/trunk \| while read commit; do T1=$(git rev-parse $commit:POE 2> /dev/null) \|\| continue; test $T1 = $T && echo $commit; done	22:09
	EvanCarroll: of course, you will have to adjust the thing.	22:09
	EvanCarroll: the idea is to get the desired tree name of POE in your tarball import.	22:09
	EvanCarroll: and then to compare it to the tree names of POE in all svn commits (if they have POE, if not, "\|\| continue" will kick in)	22:10
	← tbf left	22:10
	← tvw left	22:10
EvanCarroll	so am i doing that inside of the tarbal repo, or inside of the POE-svn-clone'd repo	22:10
gitte	EvanCarroll: as I suggested earlier, just fetch that tarball branch into your git-svn clone.	22:11
	EvanCarroll: makes life much easier.	22:11
	→ commondream joined	22:11
EvanCarroll	done it already	22:11
	man i wish the github interfaced supported this =(	22:12
rroblak	Can somebody point me to some arguments as to why git merging is so much easier than svn merging?	22:14
spb	have you ever tried to do both?	22:14
rroblak	I'm trying to understand the reasoning behind people saying this	22:14
	spb, I've only done svn merging	22:14
	← tectonic left	22:15
gitte	rroblak: Git can determine merge bases much better, even if the other branch was merged at an earlier stage already.	22:17
	rroblak: and Git can work with multiple merge bases much better.	22:17
	rroblak: even if I hear that subversion stole^Wlearnt a lot of Git recently.	22:17
	← jrmuizel left	22:18
	← camwest left	22:19
	← troyt left	22:20
	→ troyt joined	22:20
rroblak	gitte, gotcha	22:21
cxreg	what features did svn steal?	22:21
	→ pigeon joined	22:21
cxreg	that can only be a good thing, one would guess	22:21
gitte	cxreg: better merging. But don't ask me, this is just handwaving hearsay.	22:22
	→ texel joined	22:22
	→ mfilipe_ joined	22:22
cxreg	I wouldn't think svn has enoug introspection for that	22:22
	but I'm no svn expert	22:22
	← jpeeler left	22:22
gitte	they even introduced some better revision coupling for svn:externals...	22:22
cxreg	based on submodules?	22:23
	whoa, 1.6.2.5? weren't we just at 1.6.2.2? /me must have been snoozing	22:23
	← mtkd left	22:24
	← mfilipe left	22:24
seva	i have a need to keep most of my project private and part (as in subdirectory) synchronized with external svn repo, any suggestions?	22:24
gitte	cxreg: Git for Windows is still at 1.6.2.2... we are kinda waiting for 1.6.3 ;-)	22:24
	← jerojasro left	22:25
	← nighthwk1 left	22:26
	→ nighthwk1 joined	22:26
	→ jbeard joined	22:27
	← hugoxrosa left	22:28
	→ jpeeler joined	22:29
jbeard	Hi, quick question. I've tried to use git-svn to import the GWT SVN repo. It captured many of the svn tags as git remote branches, but I seem to be missing some. I'm a git newb, so I'm wondering if I'm tripping over a simple gotcha.	22:29
gitte	spearce: in some respect, I am back where I was in November... it compiles, but does not work.	22:29
	spearce: however, the design is now much cleaner, and only my shtupid tyops prevent the thing from working.	22:29
JohnnyL	anyone have a suggestion for trying to remember all the keywords in git?	22:30
spearce	gitte: don't you hate it when that happens?	22:30
gitte	spearce: I knew it was coming.	22:30
	spearce: I hate it more when I stare at the screen for 2 hours trying to decide how to put things into code. That does not happen often to me...	22:31
	→ brendon_work joined	22:31
	← jpeeler left	22:31
spearce	gitte: happens often enough here.	22:31
	i wrote 3 or 4 different versions of that diff.Edit class before I just broke down and realized i should use yours.	22:32
gitte	spearce: only because you do more intelligent things than yours truly.	22:32
	spearce: heh ;-)	22:32
	← Ryback_ left	22:32
	← hyperair left	22:33
	← Dreamer3\|optmize left	22:34
EvanCarroll	gitte: is this a difficult task for you?	22:34
	→ jpeeler joined	22:34
gitte	EvanCarroll: writing a diff engine? Yes.	22:35
	→ mm_202 joined	22:35
EvanCarroll	gitte: as in, is this difficult because i have no idea what i'm doing, or is this difficult because of git.	22:35
	← arohner left	22:35
JohnnyL	anyone anyone?	22:35
EvanCarroll	I know, with a good ammount of certainty that the only chances in the files i've touched, are the ones i've made	22:35
	← jackdempsey left	22:36
gitte	EvanCarroll: ah, you mean the svn grafting?	22:36
	← playya left	22:36
	← boto left	22:37
	→ playya joined	22:37
EvanCarroll	yea.	22:37
	→ Dreamer3\|optmize joined	22:38
gitte	EvanCarroll: I'd really try to do the numstat thingie to determine the most similar commit.	22:38
	EvanCarroll: that should be pretty straightforward once you understood how I munge the output of --numstat.	22:38
EvanCarroll	not clicking at all. =(	22:40
	← qrush left	22:43
	← faithfulgeek left	22:43
	← nighthwk1 left	22:44
	← hobodave_ left	22:44
	← elmob left	22:44
	← Yuuhi left	22:45
	← bcardarella left	22:46
gitte	EvanCarroll: but you get what I try to do, right?	22:46
	← jbeard left	22:46
	← MisterN left	22:46
	← petercoulton left	22:47
	jefftriplett → jefftriplett-afk	22:47
EvanCarroll	well, no not really, I'm missing 50% of the git foundation to understand what you're even talking about, maybe I'm just fixating on how simple i think this should be.	22:48
	→ neclimdul joined	22:48
EvanCarroll	take each diff from one repo, apply it to another repo, in the cwd. with merge history.	22:48
	it doesn't seem nearly the hard =(	22:48
gitte	EvanCarroll: if you want to try to rebase your changes ont svn/trunk without actually knowing where the branch point was, you can do that.	22:49
	EvanCarroll: I would not, as it is much more likely to get nasty conflicts, but here you go:	22:50
EvanCarroll	that works, i mean it isn't nearly as good, but it is better than nothing	22:50
gitte	EvanCarroll: git checkout <tarball> && git rebase --onto svn/trunk <tarball-root>	22:50
EvanCarroll	I have both of these online if you can take a crack with your quick script	22:50
	the whole poe codebase is like 3 megs, and mine is probably not even 100k	22:50
gitte	EvanCarroll: be advised that with this recipe, you do not get merge conflicts; there cannot be any.	22:50
	EvanCarroll: if it fails, you will end up with a non-applying diff.	22:51
	EvanCarroll: but you asked for it, I guess ;-)	22:51
	→ cemerick joined	22:51
cxreg	wow, there's a ton of commits between 1.6.2.5 and 1.6.3-rc4	22:52
gitte	cxreg: yep. 1.6.2.5 is just on a maintenance branch.	22:52
	← vipul left	22:52
	→ faithfulgeek joined	22:54
EvanCarroll	gitte: the git checkout tarball in the svn directory pulls the stuff onto the svn repo, with bad directory mashup	22:54
	oh wait.	22:55
	I pulled	22:55
	← mw left	22:56
	← scarabx left	22:57
	→ dereine joined	22:57
dereine	hi	22:58
	is it possible to use a ftp as repo?	22:58
	git push ftp://...	22:58
	← Alinermedia left	22:58
	sebleier → sebleier\|away	22:58
zuez	Curious, why is git pushing to two branches whenever I do a 'git push', something like this: http://pastie.textmate.org/private/zeetpno7bttdturl6aj5oa	22:58
mm_202	Hey guys, I have a [noob?] question. I have a bare git repo, that is owned by git:git. My user is a member of the git group, and the git repo has file perms 775. The problem is when that new files created in the repo are 600. How do I change the default to 664?	22:58
	← cemerick left	22:58
zuez	It pushes to my branch and to master... I only want it to push to the branch...	22:59
	← bjackman42__ left	23:00
EvanCarroll	gitte: this isn't going to work anyway	23:00
	→ Alinermedia joined	23:00
EvanCarroll	because the files that i haven't touched have changed in the same way throughout both repos from when i did a git-cpan pull	23:00
dereine	anyone?	23:01
	← Alinermedia left	23:03
	→ Alinermedia joined	23:03
	← seva left	23:04
qhoxie	is there a way to make checkout -b overwrite the destination branch if it exists?	23:05
	→ bcardarella joined	23:06
	← thiago_home left	23:09
	← kraymer left	23:09
	→ jrmuizel joined	23:10
	→ Modius joined	23:10
	→ j416 joined	23:11
	← RodP2 left	23:12
	← troyt left	23:12
drizzd	qhoxie: no	23:12
qhoxie	drizzd: thanks	23:13
	→ jelly-bean joined	23:13
jelly-bean	when i try to commit git says "you have some suspicious lines" and cites trailing whitespace	23:13
	i want to commit anyway	23:13
	or have it auto-cleanup end-line whitespace	23:13
drizzd	--whitespace=ignore	23:13
jelly-bean	can i do that?	23:13
drizzd	or something like that	23:14
	→ troyt joined	23:14
	← faithfulgeek left	23:14
jelly-bean	not working	23:15
drizzd	actually, git commit ignores whitespaces for me	23:16
	← jelly-bean left	23:16
	← hpoydar left	23:17
	jefftriplett-afk → jefftriplett	23:17
	→ hpoydar joined	23:17
	← hpoydar left	23:18
	→ faithfulgeek joined	23:18
	← commondream left	23:19
drizzd	uh, probably was a pre-commit hook	23:19
	← mfilipe_ left	23:19
	→ catalis joined	23:20
	← bcardarella left	23:22
	→ khmarbaise joined	23:22
	→ qrush joined	23:23
	← qrush left	23:23
	← ericindc left	23:23
	→ bcardarella joined	23:24
	← wmoxam left	23:26
	← gitte left	23:26
JohnnyL	i guess the best way to remember all of the git commands is to really only write down the essentials for your project.	23:26
	→ qrush joined	23:27
	→ bremner_ joined	23:28
	dereine → dereine[OFF]	23:28
	→ towski joined	23:33
	→ Radar joined	23:34
	← nurey left	23:35
	→ nurey joined	23:35
	→ lemonade` joined	23:37
lemonade`	if you clone a repository, do you get the history too?	23:37
	if you clone a repository, do you get the history too?	23:38
bdrewery	yup, that's the point	23:38
	:)	23:38
	→ mw joined	23:38
lemonade`	ok thanks	23:38
	← mw left	23:38
lemonade`	btw, did I post twice? I wasn't identified the first time. did it show up once or twice?	23:39
	→ mw joined	23:39
lemonade`	k then	23:40
	← lemonade` left	23:40
	← alezandro left	23:41
	→ bdimcheff joined	23:41
	← jspiros left	23:43
	← towski left	23:45
	→ jspiros joined	23:48
	← stuffcorpse left	23:48
	→ stuffcorpse joined	23:48
	← paulboone left	23:48
	→ arohner joined	23:49
	← aziz left	23:51
	→ mw_ joined	23:52
	→ FFighter joined	23:54
FFighter	how do I subscribe to the git mailing list?	23:54
	← bremner_ left	23:56
FFighter	I can't find a way (address) for that :S	23:57
	← mjf left	23:57
qhoxie	FFighter: [email@hidden.address]	23:57
killerchicken_	FFighter: http://vger.kernel.org/vger-lists.html#git	23:57
	→ scientes joined	23:57
FFighter	thanks	23:58
	→ marvil07 joined	23:59

Logs Search ←Prev date Next date→

Channels Documentation