IRCloggy #git 2008-01-12

Provider of IRC logs since 2005.

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

	2008-01-12
	← toxx_ left	00:01
	→ toxx_ joined	00:01
	← toxx_ left	00:01
Ilari	giraffeman: Yes, you can use your own versions of submodules. But don't loose the commits that are used by superrep by rebasing.	00:02
	*superrepo	00:02
giraffeman	Ilari: so, if I have a submodule.... and I git add it, and modify it, and commit it, I can still update it?	00:04
	how would I lose the commits?	00:04
Ilari	giraffeman: If you use your own commit as subrepo state and subsequently rebase it, it can get lost.	00:05
giraffeman	oh boy...	00:05
	so, how do i rebase?	00:05
Ilari	giraffeman: Merging never causes commits to be lost.	00:05
giraffeman	Ilari: is there a better way of doing this?	00:06
Ilari	giraffeman: Yes, after you have committed it, you can add it, and next superrepo commit records it.	00:07
	giraffeman: I'd say that you should use merging than rebasing in that subrepo.	00:07
	s/than/rather than/	00:08
giraffeman	okay... I will play around... still new to using git on it's own, but have been playing with it using git-svn for a while	00:09
Ilari	giraffeman: Rebasing is done using 'git rebase'. Merging is done using 'git pull' or 'git merge'.	00:11
giraffeman	here's what I don't understand	00:11
	I just added a submodule, using git-submodule add	00:11
	and, then I commited .gitmodule	00:11
	or whatever	00:11
	and then I added the submodule to the repo using git add .	00:11
	and committed it	00:11
	but now, if I cd in to that dir, and type git log, I get the log for the super repo	00:11
	so, how do I update the sub	00:12
	← p4tux left	00:12
	← mithro left	00:12
	→ CelticSoul joined	00:12
	← rkaway left	00:14
Ilari	giraffeman: Is there directory named '.git' in that subrepo?	00:14
	→ rkaway joined	00:14
giraffeman	no	00:14
	actually... there is in the original	00:15
	but not one I clone it	00:15
	once*	00:15
	→ kanru joined	00:17
Ilari	giraffeman: Hmm... Is there even known submodule in that repo? ('git-ls-files -s \| egrep ^16')	00:17
giraffeman	nope	00:18
	maybe I messed up adding it	00:18
Ilari	Anyway, I have played around with submodules, but not with git submodule...	00:19
giraffeman	yeah, I added it again, and still the .git directory doesn't end up in the repo	00:20
	that must be the problem	00:20
Ilari	Hmm... 'git submodule add <URL> <NAME>' should work...	00:20
	→ Gitzilla joined	00:21
giraffeman	$ git add vendor/plugins/resource_controller/.git/	00:21
	fatal: unable to add vendor/plugins/resource_controller/.git/HEAD to index	00:21
Ilari	giraffeman: So it exists... Does that ls-files and grep thingy show anything?	00:23
	giraffeman: When run from the superrepo.	00:23
giraffeman	it only exists in the original directory, where I added the submodule	00:23
	not in the cloned repo	00:23
	→ cartman joined	00:24
	← cartman left	00:25
giraffeman	nope, still not seeing anything from that	00:25
	→ cartman joined	00:25
giraffeman	I deleted the version that I had checked in	00:26
	and now I get something from the egreg	00:27
	egrep	00:27
Ilari	giraffeman: Here's the testcase I did: 'git init' and 'git submodule add ../git gitrepo'. Seems to work fine.	00:27
	And it does show up in output from that pipeline.	00:28
	Anyway, I haven't played with it enough to really be able to say what's wrong...	00:30
	→ EmilMedve joined	00:30
	→ toxx_ joined	00:36
	← ferdy left	00:38
	← slava left	00:40
giraffeman	Ilari: I just found a doc on it... apparently what I wanted to do isn't possible	00:42
	you can't check in a version of the changes	00:42
Ilari	giraffeman: One can use 'git add' on directory where submodule resides to stage the update for next commit.	00:44
giraffeman	yes	00:45
	but I can't keep a local copy of another git repo in my tree	00:45
	which is really what I need if I don't have commit access to the other repo	00:45
	or I don't want to host a local copy of it	00:45
	it's a vendor branch management thing... if my project depends on code that comes from somewhere else, I want a local copy of it..	00:46
loops	giraffeman, you always have a local copy with git..	00:46
giraffeman	loops: my understanding with subprojects is that that isn't exactly true	00:46
loops	giraffeman, when you clone a repo, you get everything local, not just the current version	00:46
giraffeman	right...	00:46
	but.. if I have project A, that depends on project B	00:47
	← glommer left	00:47
giraffeman	when I add project B as a submodule, I get a copy of it in project A... if I hack project B, and can't commit it back upstream, due to lack of access, then when somebody else I'm working with clones the repo, they won't see my changes	00:47
	→ glommer joined	00:48
loops	giraffeman, true enough.	00:48
giraffeman	even if I don't hack it... if another member of my project decides to commit from their git repo, and project B's repo is down for whatever reason... i'm screwed	00:48
	maybe I can script my away around this problem... will have to think about it	00:49
	→ Kebianizao joined	00:49
Kebianizao	Hello	00:49
loops	giraffeman, i think you can just enter into the submodule B directory and treat it as a regular git repo... ie.. fetch directly from your peer, instead of upstream.	00:49
Kebianizao	how could I revert changes on a single file?	00:49
	git reset filename doesn not work	00:50
loops	git checkout -- filename	00:50
Kebianizao	loops: would that modify any other file?	00:50
loops	Kebianizao, no, just the file you specify	00:50
Kebianizao	loops: thanks loops	00:50
	would I would expect reset to do that	00:51
loops	Kebianizao, many people do. it's a FAQ. there are some discussions on the mailing list if you care to read about the rationale	00:51
Kebianizao	loops: thanks I probably do	00:51
	→ aref_ joined	00:53
aref_	I'm trying to clone a repository via ssh. when I do "git clone ssh://ip/dir", after login it fails, claiming "bash: git-upload-pack: command not found" - I know for a fact it's there, anyone clued in as to what's going on?	00:54
Ilari	aref_: Does 'ssh <ip> git --version' work?	00:54
loops	aref_, are you sure it's in the PATH ?	00:55
	← devogon left	00:55
Ilari	aref_: Some shell startup scripts are not executed if the shell isn't interactive...	00:55
aref_	quite	00:55
	yes indeed, that must be it. thanks	00:55
Eridius	.bashrc is sourced for non-interactive shells	00:58
Kebianizao	bye	00:58
Eridius	you probably did what I did, which was only change the path in .bash_profile	00:58
	← Kebianizao left	00:58
	→ kampasky joined	01:03
	← madewokherd left	01:06
aref_	yeah	01:06
	having some trouble getting ssh to actually do what I want, though :p	01:07
	as per usual	01:07
	tried adding a modified path to ~/.ssh/rc, then to "environment" in same directory, after changing AllowUserEnvironment in /etc/sshd_config	01:07
	still no cookie, but with time I shall hopefully prevail	01:07
Eridius	aref_: just edit .bashrc	01:08
aref_	oh :p	01:08
	that was so simple I just skipped right over it	01:09
	thanks :P	01:09
Eridius	heh	01:09
aref_	wandaful!	01:09
	straight success	01:09
	← EmilMedve left	01:10
Eridius	when ssh is called to run a command directly, it triggers an interactive non-login shell. those shells source .bashrc	01:10
	interactive login shells, or non-interactive shells with --login, source .bash_profile or .profile	01:10
jengelh	.bashrc is always sourced when bash is run :)	01:10
	it's bashrc after all	01:10
	← KirinDav left	01:10
Eridius	actually, .bashrc isn't called if it's a login shell	01:10
	you get .bash_profile, .bash_login, or .profile instead (the first one that exists)	01:11
jengelh	then i blame it on /etc/profile, which reads bashrc :p	01:11
Eridius	heh	01:11
arw_	jengelh: yes, but thats the distribution configuring weird things :)	01:11
	← fobos left	01:12
Eridius	you should see my .bash_profile. I have a ridiculous setup that lets me throw files into ~/Library/init/bash/** and it will source them, along with a require function that lets one file declare a dependency on another	01:12
jengelh	sounds like profile.d	01:14
Eridius	never heard of it	01:15
jengelh	uhm	01:17
	/etc/cron.d, perhaps?	01:17
	anything with a .d in it	01:18
Ori_B	Eridius: it's the usual Linux setup	01:18
	at least for a number of distros.	01:18
Eridius	ah	01:19
	jengelh: I understand the concept, but google couldn't tell me if it was actually in use by any distros, or where it came from	01:19
	← Yuuhi left	01:19
jengelh	distros usually do for i in directory.d/*; do action $i; done	01:20
	it is a convenient way to append to the crontab for example without having to edit the user's stuff in /etc/rontab	01:20
Eridius	well, my solution lets files declare dependencies on other files, and it's per-user instead of global. It also supports one level of subdirectories	01:20
johannes	I have a "fork" of a project consisting of a few patches, in my local development I rebase that from time to time to have the patches nice on top and to be in sync with the project.	01:21
	Now I'd like to publish it, but re-pushing it after a rebase won' work, what's the good way to handle such things?	01:22
jengelh	doing it like Andrew Morton - not using git? :)	01:22
johannes	:-)	01:22
Eridius	johannes: if you've pushed, then the remote repo should contain your patches and rebase should no longer be necessary	01:23
johannes	Eridius: but the base project gets updated so that tree is outdated one day, then merging around later gives conflicts ... if i merge all the time it's hard to see the differences between my tree and the original tree, and rebase looked nice for that	01:24
Eridius	if you don't make any more local commits, merge and rebase will do the same thing (rebase will find no commits to save and just set the local branch to the remote, and merge will just fast-forward the local branch to the remote)	01:26
	I guess what I meant to say was: why won't it work?	01:26
johannes	I did a rebase of my branch, pushed that, fetched the latest changes from upstream, rebased, tried to push my branch again and got a message telling me that my local branch is no strict subset of the remote	01:29
	← cartman left	01:31
	→ cartman joined	01:32
Eridius	perhaps you should run gitk and see where your master and your remotes/whatever/master refs are?	01:32
johannes	maybe I go to bed and take a fresh look tomorrow :-)	01:40
robinr	that happens after rebase	01:48
	you'll need to add -f to push to bypass a check	01:49
	→ jp_n9 joined	01:49
robinr	push normally checks that the new ref is a direct descendant of the old ref	01:49
Eridius	and why wouldn't the remote be a direct descendent in his case?	01:50
jp_n9	trying to install git from source on osx 10.4. anyone know what version of docbook it requires?	01:50
	can't find a dependency list on the site	01:50
Eridius	jp_n9: skip building documentation, just install the manpages from the distribution, or check out the git repo and use the stuff in the man branc	01:50
	*branch	01:50
jp_n9	Eridius: how do you skip the doc build?	01:51
Eridius	make all doesn't build documentation	01:52
	what are you running?	01:52
robinr	Eridius: because he rebased. The new "version" of the branch is by git's definition another branch	01:52
	← nashdj left	01:53
Eridius	robinr: I would expect rebase with no local commits to make the branch equal to upstream	01:53
jp_n9	Eridius: ok, tried make prefix=/usr/local all let me see if that works.	01:53
	Eridius: worked. thanks	01:57
	*thanks!!	01:57
	← Ademan left	01:58
	→ Ademan joined	01:58
Eridius	glad to hear it	01:59
robinr	rebase produces a new branch or nothing happens (i.e. no rebase)	02:00
Eridius	yeah, he said he pushed his changes, then fetched and rebased. That last rebase should have done nothing	02:01
robinr	he said he fetched the latest changes	02:01
Eridius	ok, I'll try again: that last rebase should have made his branch point to the same commit as the remote, no?	02:02
robinr	no. he pushes his branch to some ref, fetches "master" or whatever and rebases his branch on top of that	02:03
	he doesn't fetch his own branch	02:03
Eridius	he didn't say he was on a branch	02:04
	from what he said, I gathered that he was working on master, just continually rebasing to keep his changes on the tip of the remote tree	02:05
	err, remote history	02:05
	→ theCarpenter joined	02:05
Eridius	actually, he did say "my branch" later, but I assume he meant local branch. He certainly never said that he was trying to change branches	02:05
	of course, given that he's gone to bed this is pointless ;)	02:06
	← saintdev left	02:06
	← jp_n9 left	02:10
	← cedric_ left	02:12
	→ macournoyer joined	02:12
	→ priidu_ joined	02:14
	→ zed_ joined	02:15
	← rhalff left	02:17
	← tongueroo left	02:18
	← bdiego left	02:18
	→ EmilMedve joined	02:20
	← pombreda left	02:25
	bigon → bigon`	02:29
	← sgrimm left	02:30
	← bobesponja left	02:30
	← theCarpenter left	02:33
	← macournoyer left	02:33
	→ KirinDav joined	02:37
	← KirinDav left	02:39
	→ TuxCM- joined	02:43
	← santhosh_ left	02:44
	→ KirinDav joined	02:45
	← TuxCM left	02:58
	→ mithro joined	03:02
	← ironfroggy left	03:02
	← jwb left	03:04
	← robinr left	03:06
	← Nazcafan left	03:11
	← toxx_ left	03:11
	→ Qarl joined	03:16
tarbo	i am trying to understand the difference when git-rev-list lists some commits when comparing my master and upstream, but git cherry gives nothing	03:19
	this means that everything in my repo is reflected in upstream, even if the merging/committing happened in different orders?	03:20
	(and vice versa - everything in upstream is in master)	03:20
	← timocratic left	03:21
Randal	so your master sha1 is the same as the upstream sha1?	03:21
	as in - the very same commit?	03:21
	or - let's say this differently... how big is git-diff your_master...upstream_master	03:23
tarbo	'git-diff master upstream' gives nothing	03:24
	it appears that i dont know how to check the sha1 thing :-/	03:24
Randal	but they're not the same SHA1?	03:24
	git-branch -a -v	03:24
	compare your master and upstream	03:24
	same number or different number?	03:24
tarbo	different	03:24
	interesting, hadnt seen that before.	03:24
Randal	weird - so you have the same content arrived at via different paths	03:24
tarbo	right - i think its because i merged instead of rebased?	03:25
Randal	maybe	03:25
	but what's your goal now?	03:25
	you have a commit. do you like your history>?	03:25
	or the history of upstreasm?	03:25
	I'd suggest "history of upstream"	03:25
	nudge nudge	03:26
	Randal waits for the OBVIOUS light to go on overhead	03:26
	→ Tux__ joined	03:27
tarbo	Randal: not really a goal, other than better understanding. i wanted to make sure that i wasnt leaving anything out in either direction before i played around with it. but yes, i think history of upstream would be preferable. is there an easy way to do that? i probably would have deleted and recloned once i figured out that i hadnt lost anything, but im guessing thats ugly :-)	03:27
Randal	if you apply patches A B C to get D	03:27
	or apply patches B C A to get D'	03:27
	but D' and D have the same result	03:27
	they will still be DIFFERENT sha1s	03:27
	this is a given	03:27
tarbo	makes sense now	03:27
Randal	if you want to play well with others, abandon your sha1	03:28
	and play with the upstream from here out	03:28
	git branch -D master	03:28
	git checkout -b master origin/master	03:28
tarbo	ah, lovely	03:28
Randal	PRESMUING...	03:28
	git diff master origin/master is EMPTY EMPTY EMPTY	03:28
	don't do it unless	03:29
	← zed_ left	03:29
	← jmspeex left	03:29
Ilari	Or maybe even 'git reset --soft origin/master'...	03:29
Randal	that won't help	03:29
	the parent will still be the old master	03:29
	bad parentage	03:29
	→ jmspeex joined	03:30
Randal	you want a new commit to be based on the sha1 of origin/master	03:30
	hence, make a local branch based on the tracking branch	03:30
	which is always the rule :)	03:30
	→ zed_ joined	03:31
tarbo	Randal: at what point do i need git diff master origin/master to be empty? and what if it isnt?	03:31
Randal	right now	03:32
	before you type those two commands	03:32
	you TOLD me it was empty	03:32
	I am giving you advice based on that	03:32
tarbo	ohhhhh	03:32
	i se	03:32
Randal	if you were lying to me, you get bad advice	03:32
	please don't lie to me	03:32
tarbo	mine is named upstream	03:32
	i thought origin/master was something special	03:32
Randal	no	03:32
	sorry	03:32
tarbo	my fault	03:32
Randal	I have no idea how you're tracking upstream	03:32
	same rules apply though	03:33
tarbo	so what _is_ origin/master? because i do have it...	03:33
Randal	git diff BRANCH-YOU-ARE-ON YOUR-INTERESTING-UPSTREAM # if empty...	03:33
	git branch -D BRANCY-YOU-ARE-ON	03:33
	git checkout -b BRANCH-YOU-ARE-ON YOUR-INTERESTING-UPSTREAM	03:33
	see the pattern	03:33
	?	03:33
tarbo	Randal: indeed	03:34
Randal	so that's what I thought you were doing	03:34
Ilari	tarbo: origin/master is mirror of branch master of repository nicknamed origin.	03:34
Randal	why do you have "upstream" and origin/master different	03:34
	where did you get "upstream"?	03:34
	wait - is upstream a local branch?	03:34
	as opposed to a tracking branch?	03:35
	literally "u p s t r e a m"	03:35
tarbo	Randal: thats what i dont understand. i created it myself. clearly i did a lot of things wrong in the run up to opening my mouth here :-)	03:35
Randal	without a / ?	03:35
	→ madewokherd joined	03:35
tarbo	Randal: yes, i created it with git pull <remote repo> +:upstream	03:35
Ilari	tarbo: Or actually it is refs/remotes/origin/master, but usually one can omit the 'refs/remotes' part.	03:35
Randal	so go back to "git branch -a -v"	03:35
	see which of those actually represents an interesting upstream commit	03:36
	and try "git fetch origin" first. :)	03:36
	or after, and again. :)	03:36
tarbo	origin/master was crusty, i did the fetch and now it matches 'upstream'	03:38
Randal	aha	03:38
tarbo	which presumably i should have done initially instead of making upstream.	03:38
Randal	so stop that with your "upstream" thing then	03:38
	get rid of it. it confuses you	03:39
	git branch -D upstream	03:39
tarbo	well, i think i confuse me. but it is certainly unnecessary	03:39
Randal	and now...	03:39
	git diff master origin/master	03:39
	IF EMPTY...	03:39
	git branch -D master # may fail...	03:39
	git checkout -b master origin/master	03:39
	if the branch fails, it's because you're on it	03:40
	in that case, you'll have to go headless for a second. :)	03:40
	git checkout origin/master # will say "going headless"	03:40
	then the -D and the -b	03:40
Ilari	Another way to compare trees for exact equality is: 'git rev-parse master: origin/master:'. Note the two ':'s.	03:40
Randal	Yeah - that compares the filetree instead of the history producing the filetre	03:41
Ilari	It should dump two hex digit sequences. If they are identical, the snapshots are identical.	03:41
	→ dsaxena joined	03:42
Randal	although I could imagine the diff being empty without the trees being identical	03:42
	maybe it's not possible. :)	03:42
	← TuxCM- left	03:42
	Randal waits for tarbo to come back	03:45
	Randal cues up the "jeopardy countdown music"	03:45
tarbo	Randal: i am here, that definitely worked (and git-rev-parse was also empty). tyvm for your patience :-)	03:46
	→ jp_n9 joined	03:48
	Randal bows	03:48
	Randal hits head on desk	03:48
Randal	ouch!	03:48
tarbo	Randal: incidentally, we met a few times at pdx-pm when i was living in portland (and i attended a session of yours at a tpc) - dont think you learned my name, but just wanted to say love the show	03:49
Randal	"the show"?	03:49
tarbo	you know, like radio call in guys	03:49
	i could have said "megadittoes"	03:49
Randal	Oh heh	03:49
	I was thinking you had listened to one of my three podcasts	03:49
	or seen my git tech-talk	03:50
tarbo	no, but perhaps i will now :-)	03:50
Randal	video.google.com search "randal schwartz git"	03:50
	or podcast.geekcruises.com	03:50
	or twit.tv/floss	03:50
	or perlcast.com	03:50
	there you go. more randal than you can handle. :)	03:50
	if you're interested in my explorations of smalltalk, methodsandmessages.vox.com	03:51
	tarbo rolls up his sleeves and prepares to use the schwartz	03:51
Randal	or following me minute by minute, merlyn.jaiku.com	03:51
tarbo	jaiku is super interesting, but your laptop story sucks	03:54
	← tvachon left	03:56
	← KirinDav left	03:58
Randal	indeed	03:58
	I'm typing this on what you could compute as a $8000 laptop. :)	03:58
	→ tvachon joined	03:58
Ilari	What, no $25k super high end system obtained using money from HELOC? :->	04:01
Randal	what is HELOC?	04:02
	← jmspeex left	04:03
Ilari	Randal: Home Equity Line Of Credit. Using stuff like that isn't good idea for obivious reasons...	04:03
Randal	Oh - never seen it called that.	04:03
	→ jmspeex joined	04:03
Randal	just googled for it	04:03
	anyway - ansewr no. :)	04:04
	at this point, in fact, all three of my properties have only firsts on them, with 30-year fixed. I'm quite happy	04:04
	and I'm debt free, with credit cards being paid off month-to-month	04:04
	no car loan even. :)	04:05
	so, I'm not rich, but I'm not a debtor	04:05
	→ FunkeeMonk joined	04:06
Ilari	That's the old-generation stuff. New generation is pumping out whatever value increase of home using HELOC and using negative-amortization ARMs. Of course when that boom ends, one is screwed.	04:07
	Of course, if too many do that, when the boom ends, the whole nation is screwed.	04:07
Randal	Yeah	04:09
ivazquez	It'll be 1929 all over again.	04:09
Randal	I'm already seeing that. my primary residence has actually topped out for a few years	04:10
	but that's mostly because I'm not "updating" it	04:10
	it needs to have the kitchen-diningroom wall knocked out to be "more modern"	04:10
	and create an intimate large cooking and eating area	04:10
	in any event, this is not $current_channel topic. enjoy.	04:11
Ilari	And BTW, that topic does not look right, as IIRC 1.5.3.8 has been released alrady....	04:12
	← jp_n9 left	04:13
tarbo	/join #mortgagecrisis /* no one here! */	04:13
	→ saintdev joined	04:19
	← jmspeex left	04:27
	→ pombreda joined	04:28
	← giraffeman left	04:28
	→ fhobia joined	04:29
	← EmilMedve left	04:35
	← tvachon left	04:36
	← FunkeeMonk left	04:48
	← alley_cat left	04:57
	← stephenjudkins left	04:57
	→ sgrimm joined	04:58
	→ FatalError joined	05:01
	← fhobia left	05:04
	→ stick joined	05:07
	→ jmspeex joined	05:18
	← stick left	05:23
	→ vintrepid joined	05:31
	→ mediogre joined	05:48
	← Sho_ left	05:58
Qarl	What does "superproject" refer to in git?	05:58
	→ KirinDav joined	06:13
gitster	When a project has "another" project bound at one of its directory, it is called superproject while the other project is called "submodule".	06:13
	→ tvachon joined	06:17
	→ Pengwn joined	06:18
	← tvachon left	06:18
Ilari	Isn't latest stable git release 1.5.3.8 (altough topic says 1.5.3.7, and can't fix it myself because of +t)?	06:22
gitster	I can't either. I'll be pushing out 1.5.4-rc3 tonight.	06:23
	← KirinDav left	06:33
	→ tjafk2 joined	06:36
	← ivazquez left	06:52
	← rkaway left	06:53
	→ ivazquez joined	06:54
	→ dduncan joined	06:56
	→ FunkeeMonk joined	06:58
	← zealo left	07:03
	← pedro_mg left	07:03
	← pedro-mg left	07:03
	← vintrepid left	07:08
	← zed_ left	07:08
	→ vintrepid joined	07:15
	← dsop_ left	07:16
	→ dsop joined	07:17
	→ pedro_mg joined	07:25
	→ pedro-mg joined	07:25
	→ srid joined	07:30
	← pedro-mg left	07:30
	← pedro_mg left	07:30
	→ jerbear joined	07:34
	← dysinger left	07:43
	→ Ingvae joined	07:54
	← dduncan left	08:09
	← cartman left	08:15
	← bartman left	08:27
	← Fullmoon left	08:28
	→ devogon joined	08:29
	→ Eludias joined	08:29
	→ bartman joined	08:31
	← Mohjive left	08:32
	← FunkeeMonk left	08:34
	→ FunkeeMonk joined	08:36
	← bartman left	08:38
	→ kmap joined	08:38
	→ bartman joined	08:38
	← FunkeeMonk left	08:39
	gitster asks whoever is channel operator of #git to update the topic to 1.5.4-rc3	08:42
	← jerbear left	08:49
doener_	gitster: I think there was a thread on the ml about this, but can't find it. Anyway, "git checkout no_such_branch_nor_file", will ask you if you forgot to git-add.	08:50
	that's kinda misleading if you actually wanted to switch branches.	08:50
	← context left	08:52
	→ context joined	08:52
doener_	oh, checkout is a shell script. Actually wanted to ask if you know how hard it would be to improve that and eventually try to guess if the user meant a branch or a file	08:52
Pieter	it does that already?	08:52
doener_	for the error message? since when?	08:53
Pieter	oh, for the error message	08:53
	:)	08:53
doener_	but I guess for a shell script I can figure that out myself, just wanted to avoid reading through the C core g	08:54
	→ FunkeeMonk joined	08:56
doener_	heh, "git checkout maste -- file" gives a really ugly error message	08:56
	← loops left	08:57
	← sgrimm left	09:02
	→ sgrimm joined	09:07
	← GyrosGeier left	09:08
	→ cedricv joined	09:22
	→ kmap_ joined	09:31
	← kmap left	09:31
	→ octopod joined	09:35
	→ desegnis joined	09:36
	→ penguin394 joined	09:40
	→ stephenjudkins joined	09:48
	→ reval joined	09:51
	← FunkeeMonk left	09:52
	← stephenjudkins left	09:56
	→ HG` joined	10:01
	← penguin394 left	10:02
	→ penguin394 joined	10:04
	→ sverrej joined	10:06
	← sgrimm left	10:11
	← penguin394 left	10:16
	→ robinr joined	10:34
	← Ingvae left	10:37
	← reval left	10:38
	→ ferdy joined	10:46
	← srid left	10:59
reuss	is there an easy way of adding all "modified" files to the next commit?	11:00
numbah	git-add -u	11:01
reuss	kool, i'll try that	11:01
	← HG` left	11:02
reuss	worked like a charm - that helps a whole lot :)	11:04
	→ harinath joined	11:23
	← Pengwn left	11:25
	→ Phoop joined	11:28
	→ toxx_ joined	11:38
	→ kumbayo joined	11:45
	→ aref joined	11:51
	→ Yuuhi joined	11:52
	← captnmark_ left	11:52
	→ captnmark_ joined	11:53
	← aref_ left	12:00
	→ IanL joined	12:32
Eridius	when I use git-filter-branch it left behind refs/original/...	12:37
	is there a "correct" way to delete that, or should I just rm it?	12:38
	← mithro left	12:44
doener_	is there any magic in git (or git-svn) that re-creates a "master" branch when you have none?	12:45
	I renamed master to trunk in a git-svn repo, but somehow master keeps reappearing	12:46
	hm, ok, I basically answered the question myself... Let's try "Why does git do that?" instead ;-)	12:46
thiago_home	there's no magic I know of	12:47
	what was the operation you did that made it reappear?	12:47
doener_	not sure, I tried to reproduce it earlier, failed and guessed that I somehow did that myself. But now there's a new master branch again.	12:48
thiago_home	pay attention to when it comes back up	12:49
	→ dysinger joined	12:49
doener_	AFAICT it must be one of: checkout, log, diff, commit (--amend), svn rebase, svn dcommit	12:49
Eridius	post_fetch_checkout does actually poke the master branch (after making sure it exists), but that's the only reference I can see, and it should actually exit out if master doesn't exist	12:49
	hrm, maybe it's in Git.pm	12:50
	nope, only reference to master is in a comment	12:50
thiago_home	checkout, log and diff shouldn't cause any changes to existing branches	12:51
Eridius	doener_: I bet it's svn rebase - I'd imagine that takes your svn-remote.svn.branches arg and just copies all remote branches to local branch names, and the remote branch is named master	12:51
thiago_home	except for checkout -b	12:51
	commit should commit to your current branch only and not touch anything else	12:52
	git-svn... that I have no idea.	12:52
doener_	Eridius: nope, master points to the same commit that remotes/1.0 points to right now	12:52
	Eridius: and "svn rebase" only works on the current branch anyway	12:52
Eridius	hmm	12:53
	I guess I meant git-svn fetch	12:53
	← jas4711 left	12:53
doener_	Eridius: does verify_ref actually return "true" for invalid refs? Or do I just "read" that Perl code wrong?	12:53
	without knowing Perl at all, I'd say that it returns when refs/heads/master^0 is valid, and that would explain what I'm seeing	12:54
	→ alley_cat joined	12:54
Eridius	it must return true if it's not valid, because you can't call update-ref on a ref that doesn't exist, right?	12:55
	or maybe you can..	12:55
	huh, I guess you can. that's probably it	12:56
doener_	yep, you can	12:56
	normalperson: ping?	12:58
	→ mithro joined	13:00
	← soma- left	13:12
	→ GyrosGeier joined	13:14
	→ cods joined	13:17
	→ Sho_ joined	13:20
	→ jp_n9 joined	13:24
	→ bdiego joined	13:30
	← jp_n9 left	13:35
	← kumbayo left	13:36
vbgunz	I see this a lot in plenty of man pages, .ft C and .ft ... what do they mean?	13:37
doener_	vbgunz: IIRC someone said that it might be related to your asciidoc version (too old? too new?) IIRC there's some documented setting in the Makefile	13:39
vbgunz	I just see it in a lof man pages for Git, curious if it is something I should already know	13:43
doener_	vbgunz: would I meant to say is: they should not be there, might be due to your version of asciidoc that they show up	13:44
	s/would/what/	13:45
	weird typo...	13:45
vbgunz	oh	13:45
doener_	vbgunz: try to define ASCIIDOC8 when generating the docs	13:45
vbgunz	I didn't build Git, I got it with apt-get from my repos	13:46
doener_	then complain to your upstream ;-)	13:46
vbgunz	e.g., man git-show, scroll to DISCUSSION and you'll probably see some examples of the config all enclosed in .ft C and .ft	13:47
	→ FunkeeMonk joined	13:47
doener_	no, my manpages are alright :-)	13:48
vbgunz	maybe it means file type config...	13:48
	heh, no sweat, more curious than concerned :)	13:48
doener_	vbgunz: no, it means that your manpages are broken	13:48
robinr	vbgunz: it's a bug.	13:48
vbgunz	:O	13:48
robinr	fixed in some asciidoc/docbook version	13:49
vbgunz	man, I never noticed... I see most man pages without a problem, I didn't even think it could be a problem... so, I can basically ignore it? otherwise, man pages look fine	13:49
robinr	you manpages are built the wrong way.	13:49
	I don't think you'll miss any information, it just looks wierd	13:50
	← stni left	13:50
doener_	vbgunz: yeah, you can ignore it, or better yet, complain to your upstream and get a package with correct man pages :-)	13:50
vbgunz	cool, comparing it to the HTML version in got-docs, they're all the same	13:50
	→ hendry joined	13:51
vbgunz	s/got-docs/git-docs	13:51
	→ chris2 joined	13:53
vbgunz	heres a question of convention. I can put exlude patterns in one of 3 places. .gitignore, .git/info/exclude or ~/.gitconfig ... I am learning Git, not only for version control but to hopefully one day open source a project and hopefully work with others. when this is the case, shouldn't I always use .gitignore instead of anything else? as a matter of style, shouldn't others know what I am ignoring?	14:06
	→ Man-erg joined	14:06
	← cedricv left	14:06
doener_	vbgunz: ~/.gitconfig? How would you add exclude patterns there?	14:11
	vbgunz: .gitignore is for files that make sense to ignore for the project	14:11
vbgunz	core.excludesfile, tells where to find more exlude patterns	14:11
doener_	vbgunz: .git/info/exclude is for files that you want to ignore _locally_, ie. those that are specific to your local environment	14:12
	vbgunz: ok, yeah, but that's not an exclude pattern on its own right.	14:12
	;-)	14:12
vbgunz	yeah, I know, but ignoring files locally is a bad thing is it not? I mean, I wish to ignore files that end with a ~ ... some developers may not know what I am ignoring and may violate it, right?	14:13
	btw, I put this in my .gitignore file ^[^~]+~$ and ran git status and still see those files ... is my regex off?	14:15
robinr	it certainly helps others, especially newbies to include such information explicitly in the repo	14:15
	→ bobesponja joined	14:15
doener_	vbgunz: IIRC .gitignore doesn't take regex, just globs	14:16
vbgunz	not sure what a glob is... I'll figure it out	14:18
robinr	wildcards	14:18
	*	14:18
	*~	14:18
	*.exe	14:18
doener_	vbgunz: files that are already tracked are not ignored IIRC, so if you ignore *~ but someone else adds a file foo~, you'll be able to work with that one normally	14:18
vbgunz	hmm	14:19
	→ Temaris joined	14:20
doener_	oh well, if you explicitly say "git add foo~" to stage new changes, it'll force you to use -f, but "git add -u" will happily stage any changes in foo~	14:21
	personally, I think it should force -f only if "foo~" is not tracked yet... hm	14:21
	← mediogre left	14:22
doener_	vbgunz: it's basically your choice whether you put them into .gitignore or .git/info/exclude, but .gitignore might get pretty big if you add rules to catch e.g. the backup files of every editor known to mankind	14:23
vbgunz	robinr: thanks for those glob hints.	14:25
	doener_: not sure what it is, must be my coffee, am trying to follow what you say without prematurely busting a response into the channel...	14:26
	let me start off with a simple glob and check the status	14:26
	→ doener joined	14:27
doener	vbgunz: might be my lack of coffee (or that I'm unable to express myself due to English not being my native language)	14:27
vbgunz	yeah, thats out of the way, .gitignore accepts globs and not regex. I'll put glob patterns on the my todo	14:28
	doener: I am going to understand, I think I need a few minutes. I find just reading something and trying to remember it is harder than actually going through it, I'll get it	14:29
doener	vbgunz: heh, yeah, that's actually the reason I hang around here. 80% of the time when I try to help, I actually have to figure it out myself first. Way more effective than reading the man pages without a real use for all the features.	14:31
	I'm just collecting use cases here :-)	14:31
vbgunz	heh, experimenting with a couple *~ files now	14:32
	→ srid joined	14:34
dysinger	pardon my intrusion on the conversation - need to ask a dumb question - how can I merge a single commit from master into a branch ?	14:37
doener	dysinger: git cherry-pick	14:37
dysinger	sweet	14:37
	thanks	14:37
vbgunz	doener: hmm, added file~ and used add -f. got it commited, went to a clone and pulled and though the clone had .gitignore, it did pull file~ ... so, although I am personally ignoring certain patterns, if I explicitly add -f an ignored file, it will act like a normal file and no longer be exluded	14:40
	ok, got that	14:40
	← simmel left	14:43
	← hendry left	14:43
vbgunz	doener: not too sure what you mean about using -u ... commit -a picked it up... a new clone shows the modified file~	14:44
	← doener_ left	14:44
	→ simmel joined	14:44
doener	vbgunz: commit -a does an implicit add -u	14:52
vbgunz	yes	14:52
doener	vbgunz: but instead of using commit -a, you can also do "git add foo~; git commit" (useful if you don't want to commit all changes in your working tree)	14:53
	and then "git add" will force you to use -f again, although the file is already being tracked	14:53
vbgunz	git add -u, updates files in which say git add file~ will require the force option. in other words, -f on a new file makes sense but -f on a file that is already tracked is a hurdle, this is what you mean, correct?	14:54
	yes, I just got what you meant... heh, kept experimenting and saw that behavior	14:54
	does sound like a bug but that is debatable I suppose... I mean Git is supposed to be stupid I think :)	14:55
doener	yeah, "git add -u" and "git add foo~" should both just accept changes to any tracked file. (or both force you to use -f)	14:55
	→ theCarpenter joined	14:56
vbgunz	makes sense	14:57
	← xaiki__ left	15:05
	→ xaiki_ joined	15:06
	→ nashdj joined	15:06
vbgunz	am going to watch the video by randal, been meaning to do that for a while :)	15:06
Eridius	"the video"?	15:08
PerlJam	vbgunz: it's a good presentation.	15:09
	bigon` → bigon	15:09
	← theCarpenter left	15:09
PerlJam	Eridius: Randal did a presentation at google that was recorded	15:09
Eridius	ah	15:09
	→ theCarpenter joined	15:09
	← desegnis left	15:09
vbgunz	I saw the one with Linus, I heard there was another one by someone else, anyone knows?	15:10
theCarpenter	Randal's =]	15:10
PerlJam	vbgunz: Just Linus and Randal. Those are the only ones I know of.	15:10
theCarpenter	google "Randal Schwartz git"	15:10
johannes	first hit there ;-)	15:11
	→ kukks joined	15:11
	← Sho_ left	15:19
	→ Sho_ joined	15:21
theCarpenter	can git export an SVN log?	15:28
reuss	if you have gitweb setup - and you can list projects, and see changes - is there a URL you can point to for cloning a project over HTTP? - or do you have to have full access to the tree on a URL to clone it?	15:28
	← ivazquez left	15:33
	→ geoff2k_ joined	15:34
	← theCarpenter left	15:35
	→ theCarpenter joined	15:35
	← theCarpenter left	15:38
	→ cartman joined	15:40
	→ stni joined	15:42
	← wayneeseguin left	15:43
	→ wayneeseguin joined	15:44
	→ jp_n9 joined	15:47
	← IanL left	15:47
	← geoff2k left	15:49
	← jp_n9 left	15:57
vbgunz	41:34 seconds into randals presentation he mentions something I find very very interesting. "git merge --squash --no-commit t; git commit" ... everything about it makes sense but what does t do? t should be the name of the branch, correct?	15:57
	→ Oejet joined	15:59
	→ \|kukks\| joined	16:01
	← kukks left	16:01
	→ theCarpenter joined	16:05
Eridius	vbgunz: where's the video?	16:06
	nevermind, I found it	16:07
vbgunz	cool	16:07
	I really dislike all the single commit messages that can sort of pollute your log. the above command seems to alleviate that in a sense. I am going to try it when the vid is done but if it does what I think it does, thats going to be sweet	16:08
Eridius	ok, I'm assuming t is the name of the topic branch	16:08
	he probably just condensed "topic-name" into "t" for space	16:08
vbgunz	so am I, I don't see a branch referenced so that has to be it	16:08
	yeah	16:08
	← cartman left	16:12
	\|kukks\| → kukks	16:16
	→ Dodji joined	16:16
	→ z3om joined	16:17
	← aref left	16:18
	← zeom left	16:33
	z3om → zeom	16:33
	← Phoop left	16:37
	→ zed_ joined	16:38
	← pcapriotti left	16:39
	→ nessundorma joined	16:44
	← nashdj left	16:45
	→ neosun joined	16:46
	← a-atwood left	16:47
	← theCarpenter left	16:48
	← srid left	16:49
	→ a-atwood joined	16:54
	→ vbgunz_ joined	16:58
	← vbgunz left	17:02
	→ norc joined	17:03
	← Oejet left	17:03
	← a-atwood left	17:04
norc	Does git init-db do the same thing as git init?	17:07
	(The docs are talking about git init where I can find only init-db)	17:07
Ilari	norc: Sounds like you should upgrade.	17:08
norc	Ilari: debian stable	17:08
	nuff said? ;)	17:08
Ilari	norc: Lemme guess: 1.4.<something>?	17:08
tpope	even debian stable should have git init	17:08
norc	Ilari: 1.4.4.4	17:08
tpope	though yeah, on debian you should grab the version on backports.org	17:08
	norc doesnt like backports	17:09
	→ aref joined	17:09
thiago_home	upgrade to 1.5	17:09
norc	what do I need to upgrade, git-core?	17:09
	or is git itself a bundle?	17:10
thiago_home	git-core	17:10
	that's everything	17:10
norc	ok	17:10
	→ a-atwood joined	17:10
tpope	well, it's missing gitk, git-svn, git-cvs, and a handful of other things	17:12
norc	tpope: that can be installed additionally	17:12
tpope	of course	17:12
	but it's not "everything"	17:12
	→ theCarpenter joined	17:14
norc	git works like charm :)	17:15
	setting up is a piece of cake :)	17:15
thiago_home	ah, right, git-svn is extra	17:15
Ilari	norc: gitk is highly recommended install through (unless it is some sort of server without graphical libs). Also, probably git-gui and docs are worth install.	17:18
norc	Although I have to get used to git (never used it before) I think it's the right thing to do for my project.	17:18
	Ilari: I'll do	17:18
	(its my home machine currently)	17:19
	thiago_home uses qgit	17:20
thiago_home	graphs look better	17:21
Ilari	norc: Also, you should set up username and email for author/committer info.	17:22
norc	Ilari: already had to	17:22
	Ilari: cant use git without	17:22
Ilari	norc: Ah, good.	17:24
	← bdiego left	17:25
norc	Do I have to pull repeadtly to keep my git tree up-to-date?	17:27
	← octopod left	17:27
norc	(Like in CVS)	17:28
	err, well, like not in cvs ;)	17:28
	s/like not/unlike/	17:28
	→ HG` joined	17:28
tpope	depends	17:29
	if you're using a central server setup with a bunch of other people, yes	17:29
norc	tpope: thats what I want :)	17:29
tpope	although pull isn't always the best choice	17:29
	sometimes you want to rebase	17:29
norc	What other choice do I have	17:30
	mhm	17:30
	→ morphir joined	17:30
norc	Isn't there a way that my people can just submit a change which gets automatically accepted into the tree?	17:30
	(I trust my developers)	17:30
tpope	that's the opposite, pushing	17:30
	and yes, just give them ssh with the appropriate file permissions	17:31
	→ bdiego joined	17:31
Ilari	norc: Fetching and rebasing is bit more like CVS/SVN workflow, but one is able to keep multiple changes (even SVN allows just 1).	17:31
norc	mhmm	17:31
Ilari	norc: There's also 'gitosis' for controlling repository access... Never used it though.	17:32
norc	this is more complicated than I thought initially	17:32
	looks like its not going to be ready tonight.. ;)	17:32
	I'm just using git because I was told that it's better than SVN or CVS	17:33
tpope	if you trust your developers you won't need anything like that	17:33
	just give them ssh	17:33
norc	tpope: Uh.. first I wouldn't trust them that far. Secondly I'd prefer some control system which keeps track of all changes	17:34
theCarpenter	is there any way i can export an SVN log if im running git?	17:34
morphir	can anyone with admin privileges delete my account at http://repo.or.cz? (I submitted a incorrect ssh key)	17:34
norc	At some point I might want to introduce new developers	17:34
tpope	norc: there's a hump at the beginning of the learning curve but after that it's smooth sailing	17:34
	well then change their shell to git-shell	17:34
Ilari	norc: There's 'git-shell' which is restricted shell for git remote access.	17:34
Tv	theCarpenter: svn log is linear, git history isn't; though have you looked at "git log"?	17:34
	norc: or just use gitosis	17:34
	norc: gitosis means developers don't need shell accounts	17:35
norc	Ok.	17:35
	For some reason I still feel that I want a software that keeps track of changes with comments in a standardized way.	17:35
Tv	norc: eh, that's what git stores about commits	17:35
norc	Tv: exactly.. thats why I dont want them to directly edit these files..	17:36
Tv	norc: like, not vi the repository direcly? well gitosis will give you that	17:36
norc	Tv: because some of them are on windows and 2 are using IDEs	17:36
	well, 2 of the linux guys. ;)	17:36
	gitosis	17:37
	It's been suggested so many times, I should look into it.	17:37
tpope	with git-shell they can't edit files either	17:38
norc	Ok, doesn't seem like it's what I need.	17:38
	→ Fullmoon joined	17:38
tpope	but really, if that is a concern you need to get some less retarded developers	17:38
	also, since every body has a full copy of the repo, it's not like editing it directly can destroy the history	17:39
	← theCarpenter left	17:39
norc	(It's not urgent, my project is in its early alpha stages, so at this point I dont need it yet, but I want some system rdy when my project is beta)	17:40
	→ theCarpenter joined	17:40
norc	Just fyi. :)	17:40
	tpope: perhabs I have the wrong impression of what GIT is exactly for.	17:40
tpope	it's highly resistant to any sort of mucking with the history	17:40
	if someone tries to change it manually, people will notice	17:41
norc	tpope: Can it be that CVS is more what I need here?	17:41
	← johan-s left	17:41
norc	If it's supposed to be on a central server	17:41
	or SVN	17:42
tpope	cvs may be closer to what you have in mind, but that's not what your concerns suggest	17:42
	it's way easier to muck with history in cvs	17:42
norc	I dont think they'll screw around with it	17:42
	→ johan-s joined	17:42
tpope	svn is harder but only because it's so opaque, not because it's inherently secure	17:43
	← sverrej left	17:43
norc	tpope: what do you mean with opaque here?	17:43
	Are you referring to the administration or the usage?	17:44
tpope	administration	17:44
	or more specifically, mucking with files inside	17:44
norc	tpope: what do you mean	17:44
tpope	you said you were concerned about some ass editing files with a text editor	17:45
	kmap_ → kmap	17:45
tpope	well you don't have to worry about that with svn, but only because it's database is binary	17:45
norc	tpope: Well, if all developers would make notes in the changelogs with dates and their name and what they did it would be fine	17:46
	but still, I'd like to be able to revert changes etc	17:46
tpope	well you can revert in git, but that means issuing a new commit that undoes the old one	17:46
norc	tpope: Let me give you a summary of what I need, perhabs you can give me an answer to what I actually wanna use	17:47
	I have a project which will have about 10-20 developers when I need some control system. I (and the other developers) want to be able to see changes with information like who, when and what	17:48
	The project is centralized on a server, from which the software will be distributed	17:49
	People shall be able to obtain an experimental tree, which shall be GIT, SVN or whatever	17:49
Tv	norc: have you guys ever used any version control?	17:49
	norc: that really sounds like you haven't	17:49
norc	Tv: I havent. :)	17:49
	I admit, I'm totally new to this area	17:49
Tv	norc: you're listing the definition of the word, really	17:49
norc	but Im fighting my way through :)	17:49
	Tv: please enlighten me	17:50
Tv	tracking who made what change when why is the basic functionality	17:50
norc	which is what I want :)	17:50
arw_	norc: i hereby declare you 'enlightened' :)	17:50
norc	arw_: aww.. thx. :)	17:50
	Furthermore I'd like to be able to revert changes if necessary to return to an old state of development	17:51
Ilari	Tv: Does gitosis support logging who did what push and when? If everybody has distinct ssh accounts, I think one can recover that from reflogs...	17:52
norc	and the ability to see code the way it was before some patches were applies.	17:52
Tv	Ilari: err, how would the reflog help there?	17:52
norc	Now what do I want? ;)	17:52
arw_	norc: even stuff as old as rcs can do that :)	17:52
Tv	Ilari: besides, reflogs are garbage collected, i wouldn't consider them reliable	17:52
norc	(some cute webinterface would be nice too btw)	17:52
Tv	Ilari: gitosis has no special logging	17:52
arw_	norc: but, as you are asking in #git, you want git :)	17:52
	norc: and yes, you can do all that with git	17:53
Tv	Ilari: that feature sounds more like "i don't trust my developers", and those features don't belong in gitosis, they belong in some extra hooks	17:53
norc	arw_: well, Im not asking if I can do that. I could do that with my own software if I modified it accordingly.	17:53
	arw_: I'm asking if it makes sense to use git, or if git aims for something different	17:54
Tv	norc: you're still only listing the very basic fundamentals	17:54
	← csc` left	17:54
Tv	norc: join the irc channel of any version control system and they'll say yes ;)	17:54
norc	Tv: I thought that if I'd ask here (or in SVN/CVS channel) I would get an easy answer, because you guys are experienced with this kind of software	17:55
Tv	norc: yes but you aren't asking for anything hard ;)	17:55
norc	I knew that there was a risk taht you were trying to sell me your software in a matter of speaking	17:55
arw_	norc: the easy answer is, any modern version control software will do what you listed easily.	17:55
norc	arw_: how do they differ then?	17:55
	← theCarpenter left	17:56
norc	simply in their features and design?	17:56
Tv	well design can imply huge differences	17:56
	→ theCarpenter joined	17:56
Tv	the biggest is probably centralized vs distributed	17:56
norc	Tv: I'm curious, what is a distributed system for especially?	17:57
	I can't imagine why someone would prefer that	17:57
arw_	norc: design like in "how does the data-storage-mechanism look" and derived from that things like how it will behave in certain environments.	17:57
Tv	norc: i think at this point you should be asking wikipedia..	17:57
tpope	norc: use case 1: work on an airplane and actually keep track of your changes	17:57
norc	tpope: didnt know there was an article, didnt see it at first	17:57
tpope	norc: use case 2: two developers want to work together on some experimental stuff they don't want to put in trunk	17:57
arw_	norc: theres plenty	17:57
tpope	norc: use case 3: I want to work on a patch for a project I don't have commit access too, and it's complex enough that I want to manage history on that patch	17:58
	← kmap left	17:58
	kmap_lab → kmap	17:58
norc	tpope, arw_, Tv: I think I get your points now	17:59
	Ok, won't use SVN as there is no good way to coop with lighttpd here.	18:02
	← theCarpenter left	18:03
	→ desegnis joined	18:03
	→ sverrej joined	18:06
	← FunkeeMonk left	18:08
	← nessundorma left	18:09
	→ wastrel joined	18:09
	← kukks left	18:11
	→ kukks joined	18:13
	→ ironfroggy joined	18:14
	→ Pommes joined	18:16
	← vbgunz_ left	18:25
	→ sgrimm joined	18:26
	→ orospakr joined	18:34
	→ cedricv joined	18:38
	← cedricv left	18:38
	← HG` left	18:48
	← priidu_ left	18:48
	→ ruphy joined	18:57
chris2	how could i checkout a subdirectory at a different revision than the rest of the repo?	18:57
Randal	for the purposes of merging?	18:58
chris2	for the purposes of debugging, e.g.	18:59
Randal	git checkout SomeRevision path/to/top/of/tree	19:00
	but you'll need to reset that before you commit	19:00
chris2	just read access	19:00
	easy, nice.	19:00
ruphy	hello	19:00
	chri2, have a look at git-bisect :-)	19:00
Randal	I think that shoould work	19:00
ruphy	I was updating a local repo with git-svn, but seems that I get stucked on a 'checksum mismatch'	19:01
	and no idea on how to solve that :\	19:01
	the repo doesn't seem to have problems	19:01
	and this is what I get:	19:01
thiago_home	backtrack one revision	19:01
ruphy	thiago_home: git-svn one revision?	19:01
thiago_home	no	19:01
	git reset --hard last-rev-that-worked	19:01
	then change the remote ref as well	19:01
	and erase the rev_map file	19:02
	git-svn will recover	19:02
ruphy	let me try	19:02
	thiago_home: I should to the reset on the git-svn branch, right?	19:02
	(I'm just trying with fetch for now)	19:02
thiago_home	no, reset your local branch	19:03
	then update-ref the refs/remotes/git-svn ref	19:03
ruphy	thiago_home: what should I update it to?	19:05
	git update-ref refs/remotes/git-svn [?] ?	19:05
thiago_home	to the last ref that worked	19:06
	one before the one that git-svn complains about	19:06
ruphy	ok	19:06
	→ sverrej_ joined	19:07
ruphy	thiago_home: seems to work man, thanks :-)	19:07
	thiago_home: I'll write a blog post, google tells nothing about that	19:07
thiago_home	ok :-)	19:08
ruphy	thiago_home: why is this needed? shouldn't just the update-ref work? [19:59:31] <thiago_home> git reset --hard last-rev-that-worked	19:09
thiago_home	you could do update-ref on refs/heads/master	19:10
ruphy	thiago_home: and, btw, I didn't erase the rev_map file because I couldn't find it :P	19:10
thiago_home	but that doesn't change your checkout	19:10
	the rev-map file is in .git/svn/git-svn/.rev_map*	19:10
	thiago_home is so glad git-svn switched to the map format instead of the old rev_db	19:10
ruphy	ah, here's still the old one... .rev_db	19:11
thiago_home	upgrade your git-svn	19:12
	your rev_db is 29MB in size.	19:12
ruphy	yes	19:12
thiago_home	the rev-map is under a megabyte	19:12
ruphy	thiago_home: anyway, it should still work with a git-svn fetch... when git-svn fails with that error, it doesn't rebase the current branch	19:12
	woo, neat	19:12
	so just update-ref should work	19:13
	anyway ;-)	19:13
	thiago_home: you'll be in MV, right?	19:13
thiago_home	yep	19:13
	→ vbgunz_ joined	19:14
ruphy	thiago_home: ok, I'll draft it in the plane and I'll ask you to correct it there then ;-)	19:15
thiago_home	draft what?	19:15
	← MicahElliott left	19:17
	← sverrej left	19:22
	← sverrej_ left	19:25
	→ drizzd_ joined	19:25
	→ sverrej_ joined	19:25
	→ onnadi3 joined	19:29
onnadi3	Hello all. I'm new to the channell. Please tell me if there is anyone working on the one-click git port as I would like to ask some questions about it	19:31
	← tcoppi left	19:32
	→ reval joined	19:35
	robfitz_ → robfitz	19:36
	← kukks left	19:37
	→ kukks joined	19:38
	← drizzd left	19:40
	← bdiego left	19:53
	→ vfe4t67yg joined	19:55
	→ tcoppi joined	20:01
	← neosun left	20:05
	← reval left	20:05
	→ tongueroo joined	20:18
	← desegnis left	20:19
	→ cartman joined	20:22
	→ theCarpenter joined	20:30
	← cartman left	20:36
	← vfe4t67yg left	20:48
	← Man-erg left	20:59
	→ nud joined	21:04
	← pombreda left	21:06
	← semi_ left	21:15
	← sverrej_ left	21:15
	← Pommes left	21:20
	→ dmlb2000 joined	21:23
	→ HG` joined	21:26
	← morphir left	21:31
	→ reval joined	21:33
	← chris2 left	21:36
	→ cartman joined	21:38
	→ robfitz_ joined	21:42
	← vbgunz_ left	21:45
	← orospakr left	21:46
	← aref left	21:46
	← robfitz left	21:46
	→ flaguy joined	21:52
	← flaguy left	21:59
	→ flaguy joined	21:59
	← tcoppi left	22:06
	→ tcoppi joined	22:09
	← Qarl left	22:11
	→ Qarl joined	22:12
	→ sverrej joined	22:16
	← ruphy left	22:19
	→ pergesu joined	22:26
	→ ruphy joined	22:28
	→ macournoyer joined	22:30
	← nud left	22:35
	→ desegnis joined	22:38
	← ironfroggy left	22:48
	→ ironfroggy joined	22:49
	← desegnis left	22:52
	← vintrepid left	22:54
	← Spyr1 left	23:03
	→ pmax joined	23:07
pmax	how do I undo a remote branch deletion?	23:08
	→ alp joined	23:09
Tv	pmax: do you have the branch locally?	23:11
pmax	I think so	23:11
Tv	pmax: just push it again	23:12
pmax	but I really don't know	23:12
	how?	23:12
Tv	well git branch will tell	23:12
robtaylor	pmax: git branch to see	23:12
pmax	okay I see it here	23:12
	how do I put it back?	23:12
Tv	git push REMOTE BRANCH:refs/heads/BRANCH	23:13
pmax	hm I think I got it	23:13
	how do I check out a branch on the server?	23:14
	I don't seem to have it locally	23:14
	anything I've tried with git branch so far has failed	23:15
Tv	eh?	23:15
pmax	some guy has created a branch on the server	23:15
Tv	"check out a branch on the server" doesn't really mean anything	23:15
pmax	there is code on the git server that someone checked in	23:15
	with a branch name	23:15
	I want to get it	23:15
Tv	with default config, fetch will mirror remote branch foo to remotes/REMOTE/foo	23:16
	remotes/origin/foo usually	23:16
	if you want to create a local branch based on it, run git checkout -b foo remotes/origin/foo	23:16
pmax	actually I just want to do what I think is cherry pick	23:16
	I want to merge one of those checkins into my branch	23:16
	but it says I don't have that revision #	23:16
	so I figured I don't have his branch somehow	23:17
Tv	are you sure you want cherry pick?	23:17
	sounds like you just want git merge remotes/origin/foo	23:17
pmax	I want to put his fix into my branch	23:17
	but his branch is based very far away from mine	23:17
	I happen to know that his changes will merge cleanly however	23:17
	git merge says "not something we can merge"	23:18
Tv	well do you see his branch in git branch -r	23:18
pmax	no git branch -r is blank	23:18
Tv	run git fetch	23:18
pmax	lots of error: no such remote ref ....	23:19
	and then a fatal: Fetch failure	23:19
Tv	i wonder what you've done	23:19
pmax	me too	23:19
	← ruphy left	23:19
Tv	sounds like you've configured your git repo weirdly where defaults would have worked	23:20
	pmax didn't configure anything but I guess that could be it	23:20
pmax	okay I checked out the whole thing all over again	23:20
context	ok so, i have a branch i accidently merged , and have sense checked in, i need to rebase this branch to remove the merge, and then rebase on top of that branch again	23:20
pmax	and git branch -r shows things fine	23:20
	looks like my local checkout is screwed	23:21
	so back to where I started	23:23
	← devogon left	23:23
	→ ruphy joined	23:23
pmax	how do I take this checkin in branch A and merge it into branch B?	23:23
	I was told this is what cherry pick is	23:23
	branch A branches from the mainline 2 months later than branch B did	23:23
	so it's based on quite different code	23:23
	but the patch will apply cleanly	23:24
	→ redondos_ joined	23:24
	→ devogon joined	23:25
pmax	when I run git cherry-pick A it seems to find the checkin and merge it	23:26
	but then I can't push	23:26
	← zed_ left	23:26
pmax	it says "B is not a strict subset of local ref B. maybe you are not up-to-date and need to pull first?"	23:26
	← Qarl left	23:27
	→ Qarl joined	23:27
pmax	ah I have to pull and then push, but my old corrupt repo prevented it	23:27
context	hmm	23:27
	ahh nm i figured it out :D WIN !	23:28
jengelh	DOS.	23:28
	← Eludias left	23:29
pergesu	what's the difference between merge and rebase?	23:34
	← Pieter left	23:34
arw_	pergesu: i think that is a faq	23:35
Tv	pergesu: http://eagain.net/articles/git-for-computer-scientists/	23:35
jengelh	eagain haha	23:35
arw_	http://metku.net/index.html?path=mods/loginoki/index_eng	23:35
	ah, sorry	23:35
Tv	arw_: what are you, finnish?	23:35
arw_	http://git.or.cz/gitwiki/GitFaq#head-1168c3027a2b7060df8c5cf141756c8e0e33139c	23:35
	← redondos left	23:36
arw_	Tv: german	23:36
	redondos_ → redondos	23:36
Tv	arw_: that domain is finnish	23:36
context	damnit !	23:36
	yeah im confused :/	23:36
arw_	Tv: what domain? the first link? maybe, but thats just some random link from some random irc-channel :)	23:36
pergesu	arw_ and Tv: thanks, I'll check it out	23:37
Tv	arw_: metku = prank or something	23:37
context	YEY !	23:37
	nm i got it	23:37
	rebase --onto is nifty	23:37
	→ Pieter joined	23:37
	robfitz_ → robfitz	23:38
context	hmm	23:38
	rebase duplicated commits :x	23:41
	how boo	23:41
	← cartman left	23:41
Tv	context: try reading some documentation first next time	23:42
	← Dodji left	23:43
Tv	context: there is no way to edit git commits in any way without effectively creating new commits	23:43
	context: the sha1 is a cryptographic hash of everything related to the commit	23:43
	context: anything changes -> sha1 changes -> it's a new commit	23:43
context	tv: umm i know what im doing .	23:43
	ive just never used rebase --onto before	23:43
Tv	context: all rebases	23:44
context	tv: im using git-svn and kinda borked un-dcommit'ed stuff with a merge	23:44
	tv: i know!	23:44
	→ sverrej_ joined	23:44
	← sverrej left	23:44
	YGingras_ → YGingras	23:45
context	just need to do a cherry-pick and reset hard now and il all done :D	23:46
	there	23:48
	all set good and done	23:48
	tv: thats for the insperation and confidence	23:48
	guh	23:51
	git-svn is bugging me	23:51
	→ csc` joined	23:55
context	but i got everything fixed :D	23:58

Logs Search ←Prev date Next date→

Channels Documentation