IRCloggy #git 2010-11-18

Provider of IRC logs since 2005.

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

	2010-11-18
	→ Kenjin joined	00:00
	← ProperNoun left	00:00
	← cannonball left	00:01
	← schlortoo left	00:01
	← chrischambers left	00:03
	→ alex__c2022 joined	00:04
	→ lgbr joined	00:05
lgbr	I can ssh one-way, and I want to effectively do a reverse clone from a repository on an ssh client to a new repository on an ssh server	00:06
cbreak	if you have a new repo on the server, just push	00:07
lgbr	cbreak: it complains if I do that on a non-bare remote repo	00:07
	← lucasvo left	00:07
cbreak	of course	00:07
	push it to a new branch	00:07
	one that does not exist on the remote yet	00:07
lgbr	can i move that new branch to master?	00:08
	← Sigma left	00:08
cbreak	yes...	00:08
	once you are sitting in front of the repo :)	00:08
	you can merge, reset master, just kill master and rename the other...	00:08
	the possibilities are only limited by your knowledge of git	00:09
	→ chrischambers joined	00:10
	→ kanru joined	00:10
	→ Sigma joined	00:11
	→ dTx joined	00:11
lgbr	cbreak: that worked. Thanks!	00:12
	→ schlortoo joined	00:12
	← bauruine left	00:13
cbreak	np :)	00:13
	← lgbr left	00:14
dTx	Git rulez. It's restored my faith in scm, and made my Software Engineering teamwork actually enjoyable. Just thought I'd say thanks for the work.	00:14
	→ amason joined	00:14
	→ bauruine joined	00:15
	← dTx left	00:15
	← Guest34333 left	00:16
	← letas left	00:16
	→ archis joined	00:16
	← archis left	00:17
	← darkredandyellow left	00:19
	← mcdebugger left	00:19
	← Utkarsh left	00:20
	→ mkramer joined	00:23
	→ normanrichards joined	00:24
	← mbroeker left	00:24
	→ leroux joined	00:24
	→ coderdad joined	00:25
	→ Utkarsh joined	00:25
	← tewecske left	00:25
	← adamm left	00:26
	← mkramer left	00:29
	→ xiangfu joined	00:29
	← aspotashev left	00:30
	← Utkarsh left	00:31
	← Aleric left	00:31
	← simplechat left	00:33
	← enherit left	00:33
	← patmaddox left	00:34
	← itrekkie left	00:36
	→ Utkarsh joined	00:37
	← mmoya left	00:38
	← Chaser left	00:40
	→ chrisshattuck joined	00:40
	→ rchavik joined	00:40
	→ jogla joined	00:41
	← Utkarsh left	00:41
	→ xiong joined	00:41
	→ dgtlmoon joined	00:42
	← sjl left	00:43
	→ mmoya joined	00:44
	→ itamarjp joined	00:45
	← constantc left	00:46
	← spaceonline left	00:46
	→ Utkarsh joined	00:47
	→ flazz joined	00:48
	← ricky left	00:49
	← cylence left	00:50
	← NickyP left	00:50
	→ guardian joined	00:50
	← guardian left	00:51
	← Kenjin left	00:53
	← nakkor left	00:55
	← tj left	00:58
	→ dgbaley27 joined	00:59
	← Utkarsh left	00:59
	→ tj joined	00:59
	← tj left	00:59
	← chrisshattuck left	01:02
	← bentob0x left	01:02
	→ tj joined	01:03
	→ ricky joined	01:04
	→ Utkarsh joined	01:05
	→ jcdnyc joined	01:05
	→ lemonchicken joined	01:05
	← noktoborus_ left	01:05
	← jcdny left	01:06
	→ noktoborus_ joined	01:06
	← Anti-X left	01:07
	→ NotreDev_ joined	01:08
	→ krzysz00 joined	01:08
	← NotreDev left	01:09
	NotreDev_ → NotreDev	01:09
	← abetaha left	01:09
	← alvaro_o left	01:09
	← kanru left	01:09
	← jcdnyc left	01:10
	→ _Utkarsh joined	01:10
	→ pedrobelo joined	01:10
	→ jcdnyc joined	01:11
	← Utkarsh left	01:12
	→ Anti-X joined	01:12
dgbaley27	I'm trying to do this: git archive -o linux-2.6.37-rc2.tar --remote=git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git v2.6.37-rc2	01:12
	fatal: The remote end hung up unexpectedly	01:13
	Is the command not correct, or is the feature disabled on their end...?	01:13
	← _Utkarsh left	01:14
	← jds left	01:16
	← kenichi left	01:18
	→ jds joined	01:18
	→ Utkarsh joined	01:19
	← Yuuhi left	01:20
	→ abetaha joined	01:22
	← Adys left	01:23
frogonwheels	dgbaley27: possibly they've disable that feature to stop people from pulling the whole kerenl tree each time?	01:23
	dgbaley27: clone it and archive locally I'd say.	01:24
	→ warlock_mza joined	01:24
dgbaley27	That is to prevent me from download the whole tree	01:24
	← m104 left	01:24
	← xiangfu left	01:26
	← pedrobelo left	01:26
	sshc_ → sshc	01:27
frogonwheels	dgbaley27: my guess only,..	01:29
	→ NickyP joined	01:30
	→ Adys joined	01:30
	← ricky left	01:30
frogonwheels	dgbaley27: it's to preven you from always downloading the whole tree like that (once off owuld be fine)	01:30
	→ pedrobelo joined	01:31
	← banjiewen left	01:31
	← sunil left	01:31
	→ ricky joined	01:31
	← kenneth__ left	01:32
	→ kenneth_reitz joined	01:33
	← aah left	01:33
	← envi left	01:33
	→ mdel joined	01:34
	← beatak left	01:34
	→ amh345_ joined	01:35
	← ricky left	01:36
	← amh345 left	01:38
	amh345_ → amh345	01:38
	→ envi joined	01:39
	← Evious left	01:41
	← jtaby left	01:46
	→ xiangfu joined	01:46
	← bitkiller left	01:46
	→ amh345_ joined	01:46
	← amh345 left	01:47
	amh345_ → amh345	01:47
krzysz00	if you're using, say github, and you need to go hack on your project for a while in a place with no Internet, can you still commit to git and just push up the changes when you get a connection?	01:47
SethRobertson	Of course	01:47
	It doesn't matter what your upstream is. That is a core feature of git	01:47
	← warlock_mza left	01:48
	← davidfetter_vmw left	01:50
	← sdwrage left	01:53
	← dgbaley27 left	01:54
	← abetaha left	01:56
	→ kanru joined	01:58
	← mikeric left	02:00
	← jfkw left	02:02
	← Utkarsh left	02:03
	→ jfkw joined	02:03
	→ Yuffster joined	02:04
apetrescu	krzysz00, you don't lose any of git's functionality in that situation. You can even branch, merge, etc. When you get connectivity next, pushing will sync all of that over.	02:04
	← kukks left	02:07
	← Elfe left	02:07
	→ russkey joined	02:07
russkey	is there a way to git commit --verbose by default when committing?	02:08
	← ooooPsss left	02:08
russkey	because "-v" is just too much to remember and type =P	02:08
	→ Utkarsh joined	02:08
	→ ponto_ joined	02:10
SethRobertson	russkey: You can use man git-config command to define an alias. vcommit or perhaps ci. Then `git vcommit` could actually run `git commit -v`	02:12
jast	russkey: the 'git-config' manpage can be found at http://git.or.cz/man/git-config [automatic message]	02:12
	← Utkarsh left	02:13
	→ nakkor joined	02:13
	→ chadh joined	02:13
	← nakkor left	02:13
	→ BiggFREE joined	02:14
	← ponto left	02:14
russkey	well, at that point i think i would rather make a gitc bash alias to go along with my gits alias (git status), SethRobertson. thanks	02:15
	← BiggFREE left	02:16
	→ BiggFREE joined	02:16
	→ gretch joined	02:16
	→ lightcap_ joined	02:17
	← lightcap_ left	02:18
	→ stliu joined	02:19
	← amerine left	02:20
	← steph021 left	02:22
	→ amerine joined	02:22
	→ sdwrage joined	02:23
	→ phantomcircuit joined	02:23
	→ ricky joined	02:23
	→ Elfe joined	02:25
	← pedrobelo left	02:26
	→ pedrobelo joined	02:26
	← jspiros left	02:27
	← pedrobelo left	02:27
	→ hebz0rl_ joined	02:28
	→ karptonite joined	02:29
digitalfiz	is there a easy way to remove/add files all at once?	02:29
	← alex__c2022 left	02:30
digitalfiz	i moved a ton of images from 1 folder to another but in the repo it shows them in both the old directory and the new	02:30
	← chrischambers left	02:30
	→ Utkarsh joined	02:31
	→ yatiohi joined	02:31
	← trochala left	02:31
	← hebz0rl left	02:31
	← sdwrage left	02:31
	← tj left	02:33
	→ sdwrage joined	02:36
frogonwheels	digitalfiz: git mv	02:37
	← jarrod_ left	02:37
	→ mikeric joined	02:38
	← hobodave left	02:38
frogonwheels	digitalfiz: if you use git gui it shows up the removed/added and you can just stage both the delete and the add :)	02:38
	← zombor left	02:40
	← zomgbie left	02:41
	← phantomcircuit left	02:41
	→ andrewa2 joined	02:43
	← Unknown[NF] left	02:43
	← Utkarsh left	02:43
digitalfiz	frogonwheels, no single command to do it i already moved them all id like to not have to do that per file	02:43
	→ jmorris joined	02:44
SethRobertson	git add -A	02:44
andrewa2	i could use some help recovering from a corrupted object. It's currently where HEAD is pointing. I can clone the repo again from my remote, but how can I unpack the same object and replace it in my corrupted, original local repo? (or am i thinking about this the wrong way)	02:44
	(i'd like to uncorrupt this repo as I have a local topic branch i need to recover)	02:44
	→ jogla_ joined	02:47
	→ Utkarsh joined	02:47
	← danhealy left	02:49
	← jogla left	02:49
	jogla_ → jogla	02:49
	→ Murr_ joined	02:52
	→ _habnabit joined	02:53
_habnabit	Is there a way to use rebase to merge the second commit into the initial commit?	02:53
	← yhager left	02:53
	← krzysz00 left	02:53
frogonwheels	_habnabit: git rebase -i HEAD~2 (or is it 3) and then use 'squish' or 'bugfix'	02:54
	← saysjonathan left	02:54
_habnabit	Well, no, that's the most recent.	02:54
	I mean the absolute first.	02:55
frogonwheels	ooohhh	02:55
	_habnabit: sorry	02:55
_habnabit	Usually I do `git rebase -i <preceeding commit id>` but there is no parent commit.	02:55
	And if I pass the commit id of the first revision, I can't squish it.	02:55
frogonwheels	_habnabit: you essentially have to make a whole new tree!	02:56
andrewa2	_habnabit: http://stackoverflow.com/questions/598672/git-how-to-squash-the-first-two-commits	02:56
_habnabit	Aha.	02:56
	thx.	02:56
frogonwheels	andrewa2: cool	02:56
	← Utkarsh left	02:56
	← aisipos left	02:57
andrewa2	_habnabit: thank stackoverflow.com! ;-)	02:57
_habnabit	Oh man I didn't know you could detach HEAD.	02:57
	This is awesome.	02:57
	I made temporary branches for this before.	02:57
	→ phantomcircuit joined	02:58
	→ pedrobelo joined	02:58
andrewa2	not to keep pestering everyone, but does anyone have advice on my issue? i want to replace a corrupted git object. I can get a fresh clone of the repo, but how do I unpack that object and replace it in the corrupted repo?	02:59
	← pedrobelo left	03:00
rohanpm	andrewa2: what does `git fsck' say for your corrupt repo?	03:01
frogonwheels	andrewa2: http://git.661346.n2.nabble.com/How-to-replace-a-single-corrupt-packed-object-td680795.html ... unpack the packfile (-r ?) and then replace the corrupt object before packing it?	03:02
andrewa2	rohanpm: running `git fsck` just tells me about the corrupted object again	03:02
	fatal: object cc6df31dfbc5231441c4b815975d576420517d8a is corrupted	03:02
	(note, HEAD is pointing to that object)	03:02
rohanpm	well, first backup the entire corrupt repo if you haven't already	03:03
andrewa2	k	03:03
	→ normanrichards_ joined	03:03
rohanpm	then check that your clean repo actually has that object - e.g. `git show cc6df31dfbc5231441c4b815975d576420517d8a' should show something	03:03
	→ fruight1 joined	03:03
andrewa2	cool, i did that, it's a commit	03:04
	← fruight left	03:04
	← alberto56 left	03:05
chadh	Can anyone help with a gitolite wildcard repo question? I have been banging my head overthis for a couple hours, and I'm stuck	03:05
rohanpm	andrewa2: then check if the clean repo has the file .git/objects/cc/6df31dfbc5231441c4b815975d576420517d8a	03:05
frogonwheels	andrewa2: is the object unpacked anyway .. like in .git/objects/...	03:05
	← phantomcircuit left	03:05
	→ steph021 joined	03:06
	← steph021 left	03:06
	→ steph021 joined	03:06
	← normanrichards left	03:06
	← jmorris left	03:06
andrewa2	rohanpm frogonwheels: no, my clean repo has the objects all packed up. when i try `git unpack-objects -r < .git/objects/pack/*.pack` nothing happens...	03:06
	https://gist.github.com/07f9a42a72f5260787ac	03:08
	← russkey left	03:08
	← vitorpc left	03:08
frogonwheels	andrewa2: http://thread.gmane.org/gmane.comp.version-control.git/40893/focus=40896	03:09
	andrewa2: see comment about moving the pack away (from linus)	03:09
andrewa2	frogonwheels: aaaah, it's doing something....	03:10
	I suppose that's what this text means:	03:11
	"Objects that already exist in the repository will not be unpacked from the pack-file. Therefore, nothing will be unpacked if you use this command on a pack-file that exists within the target repository."	03:11
	http://www.kernel.org/pub/software/scm/git/docs/git-unpack-objects.html	03:11
	→ schambers joined	03:12
	→ Utkarsh joined	03:13
andrewa2	huzzah! after the unpack, i was able to copy the object from the fresh repo into the corrupted repo and initially everything looks good (just a few dangling commits upon `git fsck`)	03:15
	thanks frogonwheels and rohanpm!	03:15
	← Bombe left	03:16
frogonwheels	andrewa2: np - that was quite instructive for me too :)	03:17
	→ Bombe joined	03:17
andrewa2	frogonwheels: well good call on your part with actually reading git-unpack-objects's man page and understanding it ;-)	03:17
	← Utkarsh left	03:18
	→ normanrichards joined	03:18
	→ kukks joined	03:19
	← schambers left	03:20
	← bronson left	03:20
	← mmoya left	03:21
	← friskd left	03:21
	← normanrichards_ left	03:22
	→ friskd joined	03:22
	← friskd left	03:22
sitaram	chadh: what's the question?	03:22
	→ friskd joined	03:22
	→ sunil joined	03:23
chadh	sitaram: see http://pastie.org/1307431	03:23
	→ daleackerman joined	03:23
	→ rayners joined	03:23
	← _habnabit left	03:23
	← offby1` left	03:24
sitaram	chadh: can you paste output of "ssh gitolite@myserver info; ssh gitolite@myserver expand"	03:24
	→ offby1 joined	03:24
chadh	sitaram: heh. refresh. I just added them. thought that was interesting	03:25
offby1	this channel has some weird setting that prevents me from changing my nick back from the automatically-assigned one to my normal one.	03:25
	→ saysjonathan joined	03:26
	← envi left	03:27
	← kadoban left	03:27
	→ kadoban joined	03:28
	← justin-george left	03:28
	← AndrewK left	03:28
	← airborn left	03:29
sitaram	chadh: the info command should look like: http://pastebin.com/7PuswXbh (notice the "C" part?)	03:30
	chadh: I used the exact config you gave me to get that output	03:30
	→ envi joined	03:30
sitaram	chadh: so... can you paste "cat ~/.gitolite/conf/gitolite.conf" from the server	03:31
chadh	sitaram: sorry, that info and expand command were run by the user I am having problems with. if I run it as chadh, it looks like yours	03:31
sitaram	chadh: who's trying to create the repo? You or him?	03:32
chadh	sitaram: I created the repo, but I want him to be able to read it. so I used setperms to add him to the READERS group	03:33
	→ Utkarsh joined	03:33
sitaram	chadh: can you paste the same commands output for yourself?	03:33
chadh	sitaram: I left out the initial clone and the setperms command.	03:34
	→ aah joined	03:34
chadh	sitaram: I added the info and expand to the bottom of the paste. I'll do a clone now	03:34
	← Bass10 left	03:35
	→ hobodave joined	03:36
	← programble left	03:39
	→ hobodave_ joined	03:39
chadh	sitaram: okay, maybe it is a little clearer now? I show a clone, update, commit, and push.	03:39
sitaram	chadh: it works fine here; same gitolite version... you'll have to paste me both the files in "~/.gitolite/conf" on the server I guess... but I'm not sure whats happening here	03:40
chadh	sitaram: you are able to check it out as the other user?	03:41
	← fr0sty__ left	03:42
sitaram	chadh: sure... want me to paste the whole thing like you did?	03:42
	← hobodave left	03:42
chadh	sitaram: I trust you.	03:42
sitaram	:)	03:42
	→ fr0sty joined	03:43
chadh	sitaram: the only place user randy is defined is in teh setperms stuff (his key is installed, though) could that be the problem?	03:43
	→ jgautier joined	03:43
sitaram	chadh: besides, ever since I created a proper test suite, I dont push till it passes anyway!	03:43
chadh	his name doesn't show up in the config file	03:43
	← fruight1 left	03:43
	→ fruight joined	03:44
sitaram	chadh: shouldn't matter... I just used your config as is, and my "randy" (actually "u1" in my test) doesnt appear either	03:44
	ChanServ set mode: +v	03:44
chadh	ok. well my whole config is at http://pastie.org/1307503	03:44
sitaram	chadh: erm... you dont have two servers with similar configs do you?	03:44
jgautier	hey i have some changes in my repository that have not been committed and i want to revert all my changes and get the exact copy from the remote repository	03:44
	whats the command to do that	03:45
chadh	sitaram: nope. just the one	03:45
	sitaram: btw, the install script didn't like a [email@hidden.address] key, but is it okay for me to use them for the gitolite-admin repo? You can see my "hack" in that pastie	03:46
	sitaram: so it's okay that the info and expand info for randy don't match up? How can that happen?	03:47
	← Utkarsh left	03:47
	→ Textmode joined	03:49
	← mdel left	03:49
sitaram	chadh: install script won't like it because of complications in dealing with it. Keep one key as "chadh.pub" and the others as "chadh@foo.pub"	03:49
	→ mdel joined	03:49
sitaram	chadh: info and expand are matching for randy -- if you imagine the repo was not created at all	03:49
	chadh: which is why I asked if you had two servers :-) I'm puzzled...	03:50
	chadh: what TZ are you in? I just got in to work and after about an hour I can spend time on a /query debugging this	03:50
chadh	sitaram: EST, but I'll be here for a while	03:51
sitaram	chadh: I'll try and make that 30 min then... I'll /query you when I'm done with the morning rush of emails etc	03:51
chadh	great. thanks!	03:51
	→ timj_ joined	03:52
	← Murr_ left	03:54
	← ignatp left	03:54
	← Elfe left	03:54
	→ Elfe_ joined	03:54
	→ chrislerum joined	03:55
	→ Utkarsh joined	03:55
	← timj__ left	03:55
	← jogla left	03:58
	→ dquestio2s joined	03:59
	→ shiba_yu36 joined	04:00
	→ mmoya joined	04:00
	→ dquestions joined	04:02
	← dquestions left	04:02
	← dquestio2s left	04:03
	← Utkarsh left	04:03
	→ justin-george joined	04:03
	← dreiss left	04:03
	→ dquestions joined	04:04
dquestions	hi, how do i force merge my changes (ignore conflicts and just take my current files as the current head)	04:04
	← karptonite left	04:04
dquestions	the problem is there is some conflicts with "project" files that just breaks stuff (not actual source code)	04:05
	so i'm not sure how to fix those merge conflicts	04:05
	?	04:05
	→ fr0sty__ joined	04:05
	→ anirudhr joined	04:05
	← saysjonathan left	04:06
	← fr0sty left	04:06
	← sunil left	04:06
dquestions	anyone?	04:06
	← anirudhr left	04:07
frogonwheels	dquestions: git checkout --ours -- <filename> iirc look at man git-checkout	04:08
jast	dquestions: the 'git-checkout' manpage can be found at http://git.or.cz/man/git-checkout [automatic message]	04:08
	← andrewa2 left	04:08
frogonwheels	or is it --mine	04:08
	← fruight left	04:09
	→ hohoho joined	04:09
dquestions	i already did git checkout	04:09
	i'm at the stage i like	04:09
	i just need to do a git merge --strategy=ours master?	04:09
	→ mmc joined	04:11
	→ Utkarsh joined	04:12
	← mklappstuhl left	04:14
	← hebz0rl_ left	04:14
	← Textmode left	04:15
dquestions	how do i do force revert without mergeing	04:15
	git reset	04:16
	i see	04:16
	← Utkarsh left	04:17
	← TeckniX left	04:17
dquestions	god damn it :(	04:17
	← wuj left	04:19
	→ Utkarsh joined	04:22
	→ cylence joined	04:22
	← cylence left	04:22
	→ lightcap_ joined	04:23
	lightcap_ → lightcap	04:27
	→ prs joined	04:27
	→ edude03 joined	04:27
edude03	Hello everyone	04:28
	← Utkarsh left	04:28
	→ phantomcircuit joined	04:30
	→ newchromiumuser joined	04:32
	← newchromiumuser left	04:32
	← phantomcircuit left	04:33
	→ sec^nd joined	04:33
sec^nd	hi	04:33
	hi	04:33
	→ phantomcircuit joined	04:33
	→ Utkarsh joined	04:33
sec^nd	I have a remote repo that I checkout from my local repo (ssh), I want to push changes from my local repo to the remote repo, but the remote repo wont allow it because of this error "remote: error: refusing to update checked out branch: refs/heads/master"	04:34
	+ some more text	04:34
	I tried push --all but got the same error	04:34
SethRobertson	faq non-bare	04:35
jast	SethRobertson: Pushing to non-bare repositories is discouraged: please see https://git.wiki.kernel.org/index.php/GitFaq#non-bare [automatic message]	04:35
	→ jarrod_ joined	04:35
sec^nd	what is the correct way to have repos that can be pushed and pulled from, meaning I can work on both and merge changes ?	04:35
offby1	sec^nd: like the man says: faq non-bare	04:35
sec^nd	what is git merge for then ?	04:37
	→ Textmode joined	04:37
sec^nd	can I merge changes between oooh nvm	04:37
offby1	er ... for creatig merge commits?	04:37
sec^nd	I guess merge merges commits	04:37
	lag :P	04:37
edude03	What is "no changes added to commit"?	04:38
	when I git commit	04:38
	← Utkarsh left	04:38
offby1	hmm, duno	04:38
	maybe you didn't "git add" anything	04:39
	does it say "no changes added to commit (use "git add" and/or "git commit -a")"?	04:39
	If so, then my guess is probably right	04:39
	or maybe "nothing to commit (working directory clean)"	04:40
	If so, then my guess is (again) probably right :-)	04:40
edude03	Yeah that's the error	04:40
sec^nd	-.-	04:40
	I did a reset hard and it still didn't work	04:40
edude03	the first one offby1	04:40
offby1	edude03: remember, git wants you to put all your changes into the "index" first, with "git add".	04:40
edude03	But to commit changes on stuff that's already added?	04:40
offby1	that _should_ work.	04:40
	if it doesn't, paste a transcript; either you've found a bug or (far more likely) are doing something wrong	04:41
sec^nd	both repos are on different commits because one is outdatated	04:41
	→ boombatower joined	04:41
offby1	sec^nd: as I said to edude03: paste a transcript, so we can see what's going on	04:41
sec^nd	I could probably update the other repo with git pull but I prefer push the changes so I dont have to log into the server	04:41
offby1	trying to describe it in prose is a waste of time.	04:41
	You're not knowledgeable enough to describe it precisely, and we (or at least I) am too lazy to try to decipher your description	04:42
	← steph021 left	04:42
	→ steph021 joined	04:43
	← daleackerman left	04:43
sec^nd	http://pastebin.com/RC1N5DF7	04:43
	offby1: ^	04:43
edude03	offby1: http://pastie.org/1307592	04:43
offby1	hole on folks!	04:44
	one at a time, no shoving.	04:44
	edude03 shoves sec^nd	04:44
	sec^nd stabs edude03	04:44
offby1	hey you!	04:44
	you must be _this_ high to enter.	04:44
	offby1 holds hand four feet off the ground.	04:44
	→ Utkarsh joined	04:44
offby1	sec^nd: anyway, your problem is simple: your remote is not bare (boo, hiss), and even worse, it has the same branch checked out that you're trying to push.	04:45
	If it were to let you do the push, then the checked out branch would (if someone were to do "git status") appear to have lots of changes, which is _super_ confusing.	04:45
sec^nd	-.-	04:45
offby1	So git is preventing you from doing that.	04:45
	Since it's super-confusing.	04:45
sec^nd	so how do I super fix it ?	04:45
offby1	so to get around this: maybe you can "ssh" into the remote box, and check out some other branch.	04:46
	or: ssh over there, and set the config setting that the message suggests.	04:46
sec^nd	offby1: there is only one branch (master)	04:46
	← rayners left	04:46
sec^nd	offby1: I just want to be able to push and pull from repos	04:46
offby1	or (ideally) ssh over there and convert the repo to a bare repo, and live in peace and harmony hereafter.	04:46
	sec^nd: is there a reason why you didn't make that repo bare to begin with?	04:46
	offby1 is guessing that it's a working tree where a developer does, like, y'know, actual hacking	04:47
sec^nd	offby1: I just git clone'd the repo from my local repo on the server	04:47
offby1	not sure I understand that	04:48
	do you control both repositories?	04:48
	→ FireFoxIXI joined	04:48
sec^nd	offby1: I'm working on both repos	04:48
	sometimes when I'm away I work on the remote repo	04:48
offby1	sure, but: do you _control_ both repositories?	04:48
sec^nd	when I'm local I work on the local repo	04:48
offby1	sigh	04:48
	just a simple "yes", "no", or "I don't know what you mean"	04:49
sec^nd	yeah I guess so ...	04:49
offby1	OK. Would it be a pain in the butt to make that remote repo bare?	04:49
sec^nd	LAG	04:49
	← Utkarsh left	04:49
sec^nd	[Lag: 14.782]	04:49
offby1	i.e., have no working tree?	04:49
	If not, just do it.	04:49
sec^nd	offby1: no	04:49
	how do I make it bare	04:49
offby1	pretty easy:	04:49
	"move" the .git directory up one level in the directory hierarchy.	04:50
	So: if it's currently (say) /usr/local/src/project/.git	04:50
	rename it to be /usr/local/src/project	04:50
	naturally you'll have to rename /usr/local/src/project _first_	04:50
sec^nd	thats confusing	04:50
offby1	once you've done that, then edit /usr/local/src/project/config and add "bare=true" to ... uh ... one of the stanzas	04:50
sec^nd	I know what you mean somewhat	04:50
	but how would I see the files ?	04:51
offby1	you wouldn't.	04:51
	if you need to see the files, then you shouldn't make it bare.	04:51
sec^nd	isn't git supposed to support multiple users ?	04:51
offby1	sure	04:51
sec^nd	meaning I can clone a repo and push changes back to that repo (with permission)	04:51
offby1	sure sure	04:51
sec^nd	how do I do that ?	04:51
	offby1 stares blankly	04:51
sec^nd	you seem off by 1 -.-	04:52
offby1	I suspect your question is eminently reasonable, but I don't see how it relates to what we've been discussing.	04:52
sec^nd	or more ...	04:52
	offby1: because I work on both servers	04:52
	I want to be able to sync to those repos	04:52
offby1	yeah, and you want a working tree on both, I'm guessing.	04:52
	well.	04:52
	you can (I think) give a "--force" argument to git push, which means "yeah yeah, I know, but push anyway".	04:53
	in which case you need to be aware of the potential confusion. But if you're OK with it, then fine.	04:53
	another option is to have a _third_ repo (gasp!) which is indeed bare, and which you use as a central clearinghouse kinda place.	04:54
	that's not as unusual as it might sound.	04:54
	github is a popular choice.	04:54
jgautier	hey whats the best git book for newbies?	04:54
offby1	jgautier: can't tell you, really.	04:54
sec^nd	google	04:54
offby1	jgautier: but my generic git advice is: general-purpose git advice: 1) Read "Git For Computer Scientists" (http://eagain.net/articles/git-for-computer-scientists/); 2) Run "gitk --all"; 3) Learn about "git reflog"	04:54
jgautier	cool i can get by using google but i really have no idea what im actually doing	04:55
offby1	edude03: I haven't forgotten about you ... but to be honest, I _have_ forgotten your question :-\|	04:55
jgautier	just running the commands that i find on google	04:55
	:P	04:55
edude03	offby1: how do I commit :\|	04:55
offby1	jgautier: I'd start with git-scm.com	04:55
sec^nd	edude03: you need to git add those files	04:55
	git add ...	04:55
	then git commit	04:55
	→ Utkarsh joined	04:55
sec^nd	you have to git add modified files too	04:56
offby1	(god I hate pastie.org)	04:56
edude03	I don't really care to track tohose files	04:56
	What?	04:56
	I have to add files that are modified?	04:56
	→ abetaha joined	04:56
	← steph021 left	04:56
offby1	edude03: you don't need to do anything about the untracked files.	04:56
	→ steph021 joined	04:56
offby1	but you do need to either "git add" ProgressLog.txt, or else just do "git commit -a"	04:56
edude03	OK so in future git commit -a always?	04:57
	← Textmode left	04:57
Cacheaway	edude03: "add" is not something you do one to add files to the project	04:57
offby1	edude03: that'll work, but you lose the (arguably) nice benefits of the index.	04:57
	some of us actually _like_ the index.	04:57
	offby1 glances around nervously.	04:57
offby1	really, we do.	04:57
edude03	I'm a git beginner	04:57
	offby1 backs away slowly	04:57
edude03	What's the index?	04:57
Cacheaway	you "add" changes to the staging area (unfortunately named the index) and then commit	04:57
offby1	edude03: it's the thing that sits between your working tree and the repo, and is why you gotta "git add" ;-\|	04:57
	edude03: you need to read at least a tutorial.	04:58
	← jgautier left	04:58
offby1	you _have_ to know about the index to use git. No getting around it.	04:58
	← kukks left	04:58
	→ Mizar-jp joined	04:58
Cacheaway	edude03: the "index" is the staging area for your commit. it consists of what you have "added" since the last commit	04:58
	edude03: are you coming from svn?	04:58
edude03	Yes :(	04:59
Cacheaway	ok	04:59
offby1	I have never quite understood why the index exists in the first place. I assume it was originally created as a performance hack, and later proved useful for other reasons.	04:59
	→ sh1mmer joined	04:59
Cacheaway	in svn, you "add" files to the project and they are now under source control, and when you commit, you are committing all the changes	04:59
	→ eletuchy_ joined	04:59
	← welly left	04:59
sec^nd	me too	04:59
	→ phantomcircuit_ joined	04:59
Cacheaway	in git, there is a middle ground between your working copy, and the commit. it's called the index.	05:00
offby1	yep	05:00
Cacheaway	the index is basically a staging area for the enxt commit	05:00
	→ hyperair joined	05:00
offby1	yep	05:00
Cacheaway	"git add" puts changes into the index.	05:00
offby1	and is often (confusingly) _called_ the "staging area".	05:00
	It's even also called the "cache".	05:00
	Three names for one thing.	05:00
	← Utkarsh left	05:00
edude03	What's the staging area?	05:00
offby1	index == cache == staging area	05:00
	three names, same thing.	05:01
edude03	What's the purpose?	05:01
offby1	I guess that's better than three separate concepts, but still.	05:01
bob2	http://vi-server.org/vi/gitbasic.txt	05:01
offby1	edude03: heh. As I said above: I've never known.	05:01
Cacheaway	git allows you to commit some changes and leave others for another commit, or never commit them at all	05:01
	for example	05:01
offby1	true true	05:01
Cacheaway	if you are on a fresh checkout, and you modify two files, you can then "git add file1"	05:02
	you are not adding the file to git	05:02
	you are adding your changes in the faile to the index	05:02
	→ cytrinox_ joined	05:02
	← eletuchy left	05:02
Cacheaway	at this point, if you do a commit, ONLY the changes in file1 will be actually committed,	05:02
	file2 will still be in a modified state	05:03
	← phantomcircuit left	05:03
Cacheaway	does that make sense?	05:03
edude03	Yeah that makes perfect sense :P	05:04
bob2	the above link is a pretty good explanation	05:04
	← dr_win left	05:04
edude03	So why do I lose my index if I git commit -a?	05:04
	Does it just restage everything?	05:04
Cacheaway	you don't lose anything, you just kinda skip it	05:04
edude03	bob2: Sorry I forgot to thank you I was just reading it	05:05
bob2	edude03: thank _Vi :)	05:05
mikeric	how can i rebase edit a commit that's just before a merge commit?	05:05
Cacheaway	git commit -a basically runs a git add on all modified files	05:05
	and then a commit	05:05
edude03	What's wrong with that?	05:05
	→ richard28530 joined	05:05
edude03	That's usually exactly what I want to do :p	05:05
mikeric	for some reason it doesn't work as i'd expect. it recommits them ontop of head.	05:05
	→ Utkarsh joined	05:06
Cacheaway	nothing is WRONG with it, you are just bypassing a powerful tool you have at your disposal	05:06
	← cytrinox left	05:06
	cytrinox_ → cytrinox	05:06
Cacheaway	I use it sometimes when i know I want to commit everything i've changed	05:06
edude03	Well for me	05:07
	Usually I'll set out to fix a specific bug	05:07
	which touches multiple files (like having to update a method signature)	05:07
	so lets say I have to update 4 files, I'd like to commit the bug fix as a whole (which would be all 4 files)	05:08
	← eletuchy_ left	05:09
edude03	But anyway I understand now so thanks guys	05:10
	One other thing though	05:10
	I'm working on my local machine and only my local machine	05:10
	but what do I do when I want to move the repo to a remote server?	05:10
	← Utkarsh left	05:10
Cacheaway	yes that's fine. But how about this scenario. There is a fifth file you changed which you do not want to commit. perhaps it is a config file in which you enabled debugging or somehthing while hunting down the bug. using the SVN way (commit -a) you would need to revert that file first. if you git add the 4 files you want to commit, and then do a regular commit, you don't need to revert first	05:11
edude03	True, I can definitely see how that can be useful	05:12
Cacheaway	you seeing it as useful is all I was hoping for ;) I'm not eveangelizing	05:12
	you have a few ways you can move it to another server. the easiest would be just to copy the whole folder	05:13
edude03	Evangelize all you want I left SVN willingly :P	05:13
Cacheaway	heh.	05:13
	Are you are going to be wanting to push / pull from the remote server?	05:14
offby1	edude03: ever notice how "svn switch" breaks your working copy about 30% of the time?	05:14
Cacheaway	like, you are setting it up there so you can work from 2 places?	05:14
edude03	hmm setting it up so multiple people can commit	05:14
offby1	edude03: you probably want gitolite	05:15
	unrelated to dynomite, in case you were wondering	05:15
edude03	I was think gitorious actually	05:16
offby1	oh, that too	05:16
	→ zombor joined	05:16
offby1	or github	05:16
Cacheaway	edude03: you get a pass as an svn convert, but surely you mean so multiple people can puish?	05:16
offby1	phish?	05:16
sec^nd	-.-	05:16
offby1	squish?	05:16
Cacheaway	push	05:16
sec^nd	how can users share a repo ?	05:16
edude03	Is that what I mean?	05:16
	I mean for team work purposes?	05:17
offby1	sec^nd: basically two ways: 1) gitolite; 2) wizardly Unix permissions	05:17
Cacheaway	edude03: every user will have his/her own copy of the entire repo, and be doing commits	05:17
	to their own copy	05:17
offby1	edude03: that's metaphysically absurd, man! How can _I_ know what _you_ mean!	05:17
sec^nd	offby1: I'm still having the same problem, -.-	05:17
offby1	sec^nd: which?	05:17
sec^nd	I want these repos to have a way to be in sync	05:17
Cacheaway	pushing and pulling are how two copies of the same repo sync up	05:17
sec^nd	the push error	05:17
	Cacheaway: its giving me errors :/	05:18
	→ Utkarsh joined	05:18
edude03	LOL offby1 I mean whatever the term I mean that would allow me to work with other people on the same code	05:18
mikeric	is there a way to merge a single commit in?	05:18
sec^nd	hmm	05:18
Cacheaway	edude03: there are two mechanisms, push and pull	05:18
mikeric	forget everything before and after it.	05:19
offby1	sec^nd: like I say: 1) make a third, bare repo, and have both existing repositories push to that instead of pushing to each other; or 2) pass "--force" when you push; or 3) configure the remote to allow this sort of push; or 4) check out some other branch on the remote	05:19
	I _think_ those are the only options	05:19
sec^nd	I'm going to try and create a seperate repo for them and then have a working repo	05:19
Cacheaway	if a user has write access to a remote repo, they can push to it	05:19
offby1	mikeric: you probably want "git cherry-pick"	05:19
	it doesn't, strictly speaking, "merge" the commit; but it applies that patch that the commit represents.	05:19
Cacheaway	otherwise, they can give read access to the owner of the remote, and the owner can pull from them	05:19
	so, there are a few options	05:20
edude03	Hmm that makes sense	05:20
	→ yhager joined	05:20
edude03	But you can't pull from another users repo if they don't have a public IP right?	05:20
Cacheaway	you could setup one certral location, and it's the master repo, and everybody can push to it and pull from it	05:20
edude03	Or otherwise aren't set up to serve	05:20
Cacheaway	correct	05:20
edude03	Yeah so a central location is what I'm trying to set up	05:21
	or will be soon	05:21
	← CrypTom left	05:21
Cacheaway	well, this can be as easy as sharing a common login on a ssh server somewhere	05:21
sec^nd	how do I pull from a repo overwriting all changes ?	05:21
offby1	yeah ... although gitolite is probably a better choice if there's any chance you'll have more than (say) three collaborators	05:22
Cacheaway	git speaks ssh and can use it for pushing and pulling	05:22
offby1	sec^nd: first, you could "git reset --hard HEAD", which will destroy all uncommitted changes.	05:22
	Cacheaway defers to offby1 as he knows nothing of gitolite	05:22
offby1	sec^nd: then, you can "git fetch; git merge --strategy theirs". I think. (Haven't actually tried it.)	05:22
	Cacheaway: I've only played with it, but I know that's what it's for, and I know the author is generally here on #git and is very helpful.	05:23
Cacheaway	offby1: you need to pay to have a private project on github, right?	05:23
	← Utkarsh left	05:24
	Cacheaway is going away to catch up on Dexter TTFN #git, gl, edude03	05:25
edude03	LAter Cacheaway thanks again	05:25
	YEah you have to pay for github	05:26
	I'd love to use it but I'm not sure it's worth it for me :/	05:26
	Besides	05:26
	← Eridius left	05:26
edude03	Just recently github crashed and lost a bunch of data	05:26
bob2	lost a tiny amount of data at worst	05:27
edude03	I don't know the magnitude to be honest	05:27
	but 1 byte = too much :P	05:27
sitaram	edude03: not much, if at all. Only some pull requests. Actual repo data was never lost -- it's not even part of the "database" that they accidentally broke	05:27
	→ enherit joined	05:28
sitaram	edude03: 1 byte == too much if this was git repo data, I agree. It wasn't. It was ancillary stuff; most people get by ok without that :) YMMV	05:28
edude03	yes two == signs	05:28
	http://twitter.com/marcan42	05:29
	Was talking about it the other day	05:29
	← madewokherd left	05:29
sitaram	edude03: it's definitely worth it if you don't want to roll your own. As for that twitter page, I doubt the status is still the same (but I am surprised he said the repo was gone... not accessible while the super-structure went AWOL, I think)	05:30
offby1	edude03: coincidentally, just when github was down, I wanted to get some commits from my laptop to a hosting site. It took me maybe ten minutes longer than it would have if github were up, but I did it with no trouble.	05:30
	← hyperair left	05:30
offby1	(most of those ten minutes were spent reading the "git bundle" manpage)	05:31
edude03	I'd also love to roll my own github	05:31
	← zombor left	05:31
	← sebastorama left	05:31
edude03	Gitorious should be close enough though	05:31
	→ Utkarsh joined	05:31
	← mdel left	05:31
edude03	I just need some CI and I'm good	05:31
offby1	edude03: your uptime would probably be about 85% of theirs.	05:31
edude03	Why do you assume that?	05:31
sitaram	edude03: gitolite comes close; see what KDE does with it at http://community.kde.org/Sysadmin/GitKdeOrgManual (section 4 and 5 are pretty much gitolite, with some custom code on their side)	05:32
	→ CrypTom joined	05:33
offby1	edude03: because I'm guessing you're not a professional git-hosting company with years of experience.	05:33
edude03	You'd be right about that.	05:33
	offby1 slaps self on back	05:34
edude03	Though I do run a webhosting company	05:34
offby1	that's a start.	05:34
edude03	mhm	05:34
offby1	it's entirely possible that you'd do as well hosting your own repositories as you would by using github. But for (I'd guess) 95% of hackers, github is a good choice.	05:34
edude03	We run a VMWare cluster of 5 servers + 2 drobos for storage	05:34
mikeric	offby1: doesn't that make me the commiter at that point?	05:35
offby1	mikeric: I'm not sure, actually.	05:35
	Try It And See™.	05:35
mikeric	offby1: yes, it does.	05:35
	so does a rebase, actually.	05:35
offby1	It might make you the committer, but not the author (or vice-veras)	05:35
mikeric	right	05:36
	← Utkarsh left	05:36
mikeric	i want to leave the history untouched.	05:36
offby1	can't do it, then.	05:36
	without doing a simple merge.	05:36
mikeric	i'm trying ot prevent something that i seem to do a lot.	05:36
offby1	go on	05:37
mikeric	i make a bunch of commits, and then i pull from origin. at this point i have all my new commits, mixed in with some new commits from co-developers.	05:37
offby1	I suspect that, whatever your problem is, there's a reasonably simple solution	05:37
mikeric	and a merge commit at the top.	05:37
	then i push back to origin.	05:37
offby1	sure sure	05:37
	totally typical	05:37
mikeric	only at this point do i realize a mistake in a commit i had, before the merge.	05:37
offby1	aha	05:37
	ze plotz, she's a-thicken	05:37
mikeric	a mistake that would better be amended than a new commit.	05:37
offby1	not really an option.	05:38
	once you've pushed.	05:38
	unless every one of your collaborators is willing to jump through some hoops	05:38
mikeric	i've done a git push -f a few times	05:38
	they don't seem to care	05:38
	i guess they just reset back a few and re-pull	05:39
offby1	shrug then that's a reasonable fix.	05:39
mikeric	but that doesn't fix it.	05:39
offby1	it doesn't?	05:39
mikeric	that just duplicates the commits.	05:39
	which is worse, lol.	05:39
offby1	well, the old value of the branch will still have the old, bad commit.	05:39
mikeric	the problem is with the merge and rebasing.	05:39
offby1	But everyone is expected to just forget about that old value.	05:39
	→ dreiss joined	05:40
offby1	mikeric: you may need to cook up a tiny demonstration, to explain; I'm starting to have trouble following you.	05:40
mikeric	sure, one sec.	05:40
	← sivy left	05:41
	→ Alexander_ joined	05:42
	→ Utkarsh joined	05:42
	← jfkw left	05:43
mikeric	offby1: probably not any better.. http://pastie.org/1307667	05:47
	← nullvariable left	05:47
offby1	gimme a sec	05:47
	gotta put on my anti-pastie goggles	05:47
	← Alexander_ left	05:47
mikeric	sorry about the pseudo git logs there	05:47
	:)	05:47
offby1	yeah ... lemme ponder it some more	05:48
	there's something weird	05:48
	you're showing that _after_ you pull, you get new commits -interspersed- with yours.	05:48
	That can't be right.	05:48
mikeric	yea, that's definitely right.	05:48
	→ girishr joined	05:48
offby1	have you tried "gitk --all"?	05:48
mikeric	isn't that how git merge works?	05:49
offby1	nope	05:49
mikeric	yes	05:49
offby1	I'm stumped, then.	05:49
	unless you painstakingly construct like a shell script that reproduces the problem, I don't think I'm going to understand what you're asking.	05:49
	← Utkarsh left	05:49
	← tampakrap left	05:49
	→ burhan joined	05:50
offby1	now, in general, if you realize there's a problem in some commit that isn't the tip, then the simplest course of action is probably to just create a commit that fixes the problem, and warn people not to build anything before that fix.	05:50
burhan	what's the name of the widget that displays the source code on github.com?	05:50
offby1	no idea.	05:50
	→ mention joined	05:51
burhan	is the source of github.com available?	05:51
offby1	it's probably specific to github, and proprietary, and You Can't Have It Nyaa Nyaa.	05:51
	burhan: heh, hell no.	05:51
mikeric	offby1: i guess it just comes down to me not wanting my silly mistakes present in the history..	05:51
burhan	its quite good	05:51
offby1	burhan: I wouldn't be surprised if they've released a fair amount of open-source, but the guts of the service are how they make money.	05:51
mikeric	i'm ocd like that, i like to keep the commit history curated.	05:51
offby1	burhan: if you really want to know, ask schacon;he works there	05:52
	mikeric: I'm the same way.	05:52
	← girishr left	05:52
	← MikhX left	05:52
offby1	mikeric: I pile up a ton of local (more or less private) commits, and when it's time to publish 'em, I do a flurry of rebasing first.	05:52
mikeric	i hate little commits here and there like "whoops fixing blah", "another fix", etc.	05:52
offby1	Most times I squash 'em all down to just one big patch-bomb commit.	05:52
	yep yep	05:52
	← spatulasnout left	05:53
offby1	mikeric: you've got the right attitude, in my humble opinion: your commits should be coherent.	05:53
	→ spatulasnout joined	05:53
offby1	tell a story, etc.	05:53
	← enherit left	05:53
mikeric	yea	05:53
offby1	mikeric: once you've pushed, though, all bets are off.	05:54
	it's out there and you gotta deal with it.	05:54
	offby1 strains for an analogy	05:54
mikeric	well	05:54
	you could always reset back and push -f	05:54
	:)	05:55
	→ Utkarsh joined	05:55
mikeric	unless you merged and then pushed.	05:55
offby1	sure, but that requires a bunch of patient collaborators	05:55
	← chrislerum left	05:55
mikeric	yea	05:55
offby1	at some point, you're down to just using "unison"	05:56
	← Utkarsh left	05:59
	→ everyman_ joined	05:59
	→ ichat_system joined	06:00
	← sdwrage left	06:01
	→ devanjan joined	06:01
	→ dieagram joined	06:02
mikeric	offby1: hmm, i just figured it out.	06:03
	← aresnick left	06:04
mikeric	if i specify a commit sha to rebase to (instead of head^^^^^) (even if they are the same commit), then it doesn't take the merge commits into account.	06:04
	← ichat_system left	06:05
mikeric	and then when i rebase, it keeps all the existing commits, but removes/ignores the merge commits, in which case it actually applies cleanly.	06:05
	at least it seems like it did :)	06:05
	→ Utkarsh joined	06:05
mikeric	man, i'm horrible at explaining this.	06:06
blast_hardcheese	If I'm absolutely positive that nobody has checked out from me and I haven't pushed anywhere yet, is checking out something in the past, then git commit --amend kosher? Will that break time?	06:06
	→ OOPMan joined	06:06
	→ hasrb joined	06:07
offby1	blast_hardcheese: if you're sure, then go ahead	06:08
	→ doener joined	06:08
blast_hardcheese	Alright; I just did it, just seemed really scary, since I had a bunch of uncommitted changes that I was stashing/popping then re-stashing	06:09
	→ eletuchy joined	06:10
	→ roop_ joined	06:10
	→ MikhX joined	06:11
	← Utkarsh left	06:12
	→ saysjonathan joined	06:12
	→ freimatz joined	06:13
	→ jspiros joined	06:13
blast_hardcheese	Oh crap, I lost a file	06:15
	Yeah, that was a terrible idea.	06:16
	→ Utkarsh joined	06:17
	← richard28530 left	06:17
selckin	there is notting wrong amend, and it won't make you lose files	06:17
	→ alex__c2022 joined	06:18
	← saysjonathan left	06:18
blast_hardcheese	git stash; git checkout <hash>; git stash pop; git add -p; git commit -m; git stash; git checkout master; git stash pop	06:18
	→ saysjonathan joined	06:18
blast_hardcheese	I lost two untracked files	06:18
selckin	the problem is git checkout <hash>	06:18
	→ Guest94165 joined	06:18
selckin	that puts you on a detached head, where you definitly shouldn't commit	06:18
	→ Guest97882 joined	06:19
blast_hardcheese	oh.	06:19
theoros	blast_hardcheese: when you did git checkout master	06:19
	blast_hardcheese: what does it say for "previous commit was"	06:19
blast_hardcheese	previous head position?	06:20
theoros	yes	06:20
blast_hardcheese	aaae, the hash I checked out	06:20
selckin	blast_hardcheese: if you want to amend to a random commit in the past, make a new commit, do git rebase -i <hash before it>, and then move the commit behind it, and pick squash/fix	06:20
blast_hardcheese	crap.	06:21
	← krosenvold left	06:21
theoros	blast_hardcheese: so you can just do git tag dont_lose_me aaae	06:21
	← Utkarsh left	06:21
blast_hardcheese	theoros: what would that get me?	06:21
theoros	that commit still exists, it's just not reachable	06:21
	→ krosenvold joined	06:21
	→ blahdeblah joined	06:21
blast_hardcheese	oh, no; I haven't lost any commits	06:21
	not that I can tell anyway	06:21
	I just lost files	06:21
theoros	oh you lost _untracked_ files	06:21
blast_hardcheese	they were untracked files.	06:22
theoros	how did that work	06:22
blast_hardcheese	Yeah I have no idea.	06:22
	Oh, wait a minute, I think I know how.	06:22
	← itamarjp left	06:22
blast_hardcheese	one sec	06:22
	← saysjonathan left	06:23
	→ saysjonathan joined	06:23
blast_hardcheese	Hm.	06:23
blahdeblah	Hi. Is there a git shorthand for diffing a remote repo's current state with its state at the last time i merged it? I want to view what has happened on the remote since my last pull.	06:23
	← Yuffster left	06:23
	→ albel727_2 joined	06:23
blast_hardcheese	theoros: Is there a way I can list commits by time?	06:24
	← alex__c2022 left	06:24
blast_hardcheese	the unlinked ones, that is	06:24
bob2	blast_hardcheese: git log master..blah/master	06:24
	assuming ytou have updated lately	06:24
blast_hardcheese	where blah is something like origin?	06:25
	← coderdad left	06:25
bob2	yes	06:25
	→ sambarino joined	06:26
	→ Utkarsh joined	06:26
	← albel727 left	06:27
	← saysjonathan left	06:27
	→ alex__c2022 joined	06:28
blast_hardcheese	I don't think I lost anything... Yeah, I lost the untracked files, but I got VERY lucky-- they were still open in my IDE	06:28
	→ saysjonathan joined	06:28
	→ hyperair joined	06:29
	← alex__c2022 left	06:29
	← Utkarsh left	06:29
	← xtagon left	06:30
kanzure	hmm this is odd..	06:30
	someone took a raw copy of a .git repo (not knowing what git was)	06:30
	and just edited the files, and now has a repo of his own starting from that state	06:30
	so i want to merge in the commits that he has in his fresh repo, into the mainline .git repo	06:31
	any ideas?	06:31
	← edude03 left	06:31
selckin	just do it?	06:31
kanzure	what?	06:31
	→ richard28530 joined	06:31
selckin	thats normal workflow	06:31
kanzure	no it'sn ot	06:31
	he didn't clone the repo	06:31
selckin	doesn't mather	06:32
kanzure	i want those commits- with those dates- in the history	06:32
	this is not normal workflow	06:32
selckin	add his repo as a remote to yours and pull from it, and add yours to his as a remote and push	06:32
	either way its pritty normal	06:32
kanzure	hm	06:32
	← dieagram left	06:33
kanzure	ok i'll try that	06:33
	← saysjonathan left	06:33
	→ petran joined	06:33
	→ saysjonathan joined	06:34
blahdeblah	So, any suggestions on how to diff a remote's current state against the last time i merged it? I have the commit id if that helps...	06:34
theoros	diff your_remote/branch last_merge_commit	06:35
doener	blahdeblah: git diff <your_branch>...<other_branch>	06:36
	blahdeblah: that's three dots, and makes git do the merge base calculation for you	06:37
digitalfiz	so when i get master to a state i want should i branch it to like a 1.0 branch? im a n00b git user just started really	06:37
	→ tgunr joined	06:37
theoros	git doesn't prescribe your workflow	06:37
	you have to come up with one yourself	06:37
	you could have master contain your finished releases	06:38
	→ Utkarsh joined	06:38
theoros	or you could have a separate branch to handle that	06:38
digitalfiz	well what i want is like a "snapshot" or the code in that current state for future reference	06:38
doener	blahdeblah: in general: git diff A...B == git diff $(git merge-base A B) B ==> diff B against the common ancester	06:38
	digitalfiz: for that use case, use a tag	06:38
	← saysjonathan left	06:38
theoros	you may find http://nvie.com/posts/a-successful-git-branching-model/ enlightening in some way	06:38
digitalfiz	that people can pull later if say 1.0 works for them but 1.2 doesnt	06:38
theoros	you could use a subset of that workflow or alter it in some way etc.	06:38
digitalfiz	awesome ty	06:39
	→ saysjonathan joined	06:39
theoros	where you tag, though, isn't really important	06:39
	if you get it to the state you want it and you tag it, that tag is reference-able	06:39
	→ goshawk joined	06:40
blahdeblah	doener: That doesn't seem like what i want, if i'm understanding you correctly. I want to see diff of - i.e. what's changed on the remote since i last i merged it. Both diff targets are on the remote.	06:40
	→ psankar joined	06:40
blahdeblah	s/- i.e.//	06:40
	Bad cut & paste editing by me	06:40
theoros	that seems like what you want	06:41
	take the remote, look at the closest common ancestor of your local branch, which will be the last time the two branches were merged	06:41
	then find the diff of the remote branch by comparison to that ancestor	06:42
	← eventualbuddha left	06:42
	blahdeblah tries	06:42
	← mmoya left	06:43
theoros	(i hope you aren't trying that manually)	06:43
	maybe it would be good to do it manually to see what A...B is doing	06:43
blahdeblah	no, with ...	06:43
Rickardo1	I have managed to setup git and using it locally. Now I want to setup a server to push my repos on. Is it github I´m looking for?	06:43
	→ eventualbuddha joined	06:43
theoros	Rickardo1: you can set up your own server if you don't want to use github	06:43
	← saysjonathan left	06:44
blahdeblah	theoros: in the example doener gave, A is origin/master and B is the commit id of the last time origin/master was merged locally, correct?	06:44
frogonwheels	Rickardo1: github is a public repository - (free is public anyway)	06:44
	→ saysjonathan joined	06:44
Rickardo1	theoros: Ok, is there different git server software?	06:44
frogonwheels	Rickardo1: use gitosis if you want to host it on your own machines.	06:44
	← Utkarsh left	06:44
	→ ss23 joined	06:44
frogonwheels	no	06:44
theoros	i can never remember, is gitolite > gitosis, or the other way around?	06:44
Rickardo1	frogonwheels: ok	06:44
frogonwheels	Rickardo1: use gitolite	06:44
ss23	I commited, haven't pushed, is there any way I can be like "Remove file X from that commit I just did"	06:44
frogonwheels	theoros: gitolite > gitosis yes (gitosis came first)	06:44
	← Tianon left	06:45
theoros	the names really don't help	06:45
ss23	(I would look in the documentations, but I don't know where to start :S)	06:45
theoros	ss23: type git status	06:45
	oh, sorry	06:45
blahdeblah	theoros: Because if that is correct, it's no giving me what i want: it's giving a diff of the local repo against origin/master at that commit id, which is not what i want	06:45
theoros	i read indexed instead of committed	06:45
	ss23: man git-reset	06:45
jast	ss23: the 'git-reset' manpage can be found at http://git.or.cz/man/git-reset [automatic message]	06:45
frogonwheels	ss23: and if you use git gui you can just click the 'ammend last commit' radio butotn :)	06:46
	→ albel727 joined	06:46
sitaram	theoros: no they don't. gitolite started life as "gitosis-lite". I renamed it in about 4-5 days after announcing it. And then pretty soon it became "gitosis-NOT-lite" but I dare not rename it again now ;-)	06:46
theoros	blahdeblah: what command did you type?	06:46
ss23	Thanks theoros	06:46
blahdeblah	theoros: git diff origin/master...<commitid of last merge with origin/master>	06:46
theoros	ss23: as frogonwheels said, you could just git rm the file and then git commit --amend	06:46
frogonwheels	sitaram: hey at least it wasn't gitosisng	06:46
theoros	blahdeblah: you've done them the wrong way around	06:47
sitaram	theoros: so think of "lite" as being a "suffix" of some kind and that will help you remember it came later	06:47
theoros	blahdeblah: A...B is $(git merge-base A B) B	06:47
	blahdeblah tries again	06:47
sitaram	frogonwheels: oh I hate that ng stuff. I think of NTFS and NT and so on, for some reason (even though the N does not mean the same)	06:47
frogonwheels	theoros: or I was thinking anyway. I theoretically know how to use --amend	06:47
theoros	blahdeblah: you're looking for $(git merge-base B A) A, so it's B...A	06:47
	blahdeblah needs to understand merge-base better	06:48
theoros	frogonwheels: i use it a lot, i just also tend to use git reset HEAD^ <file>	06:48
blahdeblah	Thanks theoros - that's giving me the right stuff	06:48
	→ Dreamer3 joined	06:48
frogonwheels	theoros: I tend to commit the change and then rebase -i and 'bugfix' it	06:48
theoros	ss23: oh uh, git rm might not be what you want, depending on how you want "remove" to be interpreted.	06:48
chadh	sitaram: I thought it was like JJ from the TV show Good Times used to say: Git-O-LIIIITE!	06:48
	← albel727_2 left	06:48
theoros	frogonwheels: do you know about --autosquash?	06:48
	→ Tianon joined	06:48
	← Tianon left	06:48
	→ Tianon joined	06:48
sitaram	chadh: I'm probably not familiar with that one... :)	06:49
	← saysjonathan left	06:49
chadh	sitaram: :10 into this: http://www.youtube.com/watch?v=v19PpD5uqL0	06:49
	→ saysjonathan joined	06:49
	← sambarino left	06:49
sitaram	chadh: thanks; I'll bookmark it (youtube doesnt work at work)	06:49
	→ Utkarsh joined	06:50
theoros	sitaram: i try to do that to remember, but.. it just doesn't stick	06:50
chadh	sitaram: aww. It was a catch phrase of the character.	06:50
theoros	sitaram: maybe now that i've had you yourself tell me, it might stick	06:50
	→ sambarino joined	06:50
chadh	sitaram: btw, I just swapped out those \h's, and I am in business, so that was definitely the problem	06:50
sitaram	theoros: ah well; don't worry if it doesn't...	06:51
	chadh: with what? [ \t] is the correct substitute	06:51
theoros	if i ever get around to actually using gitolite, that will definitely do it.	06:52
sitaram	chadh: brace, space, backslash, t, brace-close	06:52
	theoros: :)	06:52
	← gretch left	06:52
chadh	sitaram: can you use \t in [] ? that's essentially what I used, though	06:52
theoros	blahdeblah: get it working?	06:52
chadh	sitaram: I used brace space tab brace-close	06:52
blahdeblah	blahdeblah: yep - beautiful. Thanks for your help.	06:53
	← giallu left	06:53
blahdeblah	s/blahdeblah/theoros/	06:53
theoros	blahdeblah: good stuff. it's worth noting that diff has its own interpretation of the A...B syntax	06:53
	git rev-parse will give you a different meaning	06:53
blahdeblah	I just need to understand the merge-base thingy a bit more.	06:54
theoros	which is, iirc, A B ^$(git merge-base A B)	06:54
	← saysjonathan left	06:54
	← eventualbuddha left	06:54
	← hyperair left	06:54
theoros	blahdeblah: i think of it as, merge-base A B is, pick either A or B, start following parents until you find a commit which is an ancestor of B	06:55
blahdeblah	ok	06:55
sitaram	chadh: yes you can use \t; that sequence is actually used in several examples in man perlre (on my version anyway)	06:55
	→ saysjonathan joined	06:55
theoros	(or A)	06:55
	← Eaven left	06:55
blahdeblah	theoros: Just RTFM-ing now	06:55
theoros	"Specifying revisions" in rev-parse --help will probably be useful	06:56
	← xiangfu left	06:56
theoros	and/or "specifying ranges", i forget	06:56
	→ hyperair joined	06:56
	→ eventualbuddha joined	06:57
	← xiong left	06:58
	→ Athox joined	06:59
	← blahdeblah left	06:59
	← Anti-X left	06:59
	← saysjonathan left	06:59
	→ saysjonathan joined	07:00
	← mention left	07:02
	← everyman_ left	07:04
	← saysjonathan left	07:04
	→ saysjonathan joined	07:05
	← albel727 left	07:05
	← spatulasnout left	07:05
	→ albel727 joined	07:05
	← SimonP86 left	07:05
	→ spatulasnout joined	07:06
	→ srcerer_ joined	07:06
	← abetaha left	07:07
	← v left	07:07
	← Utkarsh left	07:08
	→ v joined	07:08
	← srcerer left	07:08
	← nchaimov left	07:09
	← MattDiPasquale left	07:09
	→ mmoya joined	07:09
	← hyperair left	07:09
	← saysjonathan left	07:10
	→ levifig joined	07:10
	→ saysjonathan joined	07:11
	← goshawk left	07:11
	← mmoya left	07:12
	→ khmarbaise joined	07:13
	→ gretch joined	07:14
	← saysjonathan left	07:16
	→ saysjonathan joined	07:16
	→ mastroDani joined	07:17
	← khmarbaise left	07:17
	→ mtkd joined	07:17
	→ Adaptee joined	07:18
	→ opalka joined	07:19
	← dgtlmoon left	07:19
	→ cbreak_work joined	07:21
	← saysjonathan left	07:21
	← theoros left	07:21
	→ saysjonathan joined	07:21
	← lightcap left	07:22
	→ QinGW joined	07:23
	→ p3rg2z joined	07:23
	← saysjonathan left	07:26
	→ lucasvo joined	07:26
dquestions	how do i set my current repo to be the head of the master branch	07:26
	(there is something wrong with my current master head	07:26
	→ saysjonathan joined	07:26
dquestions	and if i try to merge it messes things up	07:26
	because the problem is with non source code files	07:27
	← bradleyayers left	07:27
	→ guardian joined	07:27
	← amerine left	07:28
	→ nchaimov joined	07:29
	← hasrb left	07:30
	→ giallu joined	07:31
	← saysjonathan left	07:31
	→ saysjonathan joined	07:32
	← MikhX left	07:32
	→ mmoya joined	07:33
	→ saschpe joined	07:34
	← JasonWoof left	07:34
	→ JasonWoof joined	07:34
	← JasonWoof left	07:34
	→ JasonWoof joined	07:34
	← saysjonathan left	07:36
	→ saysjonathan joined	07:37
	← urkud left	07:37
	← mmoya left	07:39
	→ acs` joined	07:39
	← acs` left	07:40
	← acs left	07:40
	→ mishok13 joined	07:40
	→ ph^ joined	07:40
	→ acs joined	07:41
	← saysjonathan left	07:41
	← ponto_ left	07:41
	→ mmoya joined	07:42
	→ sivy joined	07:42
	→ saysjonathan joined	07:42
	→ rraasch joined	07:43
	← guardian left	07:43
	← Fullmoon left	07:44
digitalfiz	when i create a tag does that stay the current tag till i set another one?	07:44
	→ Fullmoon joined	07:44
	← Bombe left	07:44
	→ Bombe joined	07:44
rohanpm	digitalfiz: I don't think there is any concept of a "current tag"	07:44
	← Fullmoon left	07:45
digitalfiz	ok i can just tag the commits after?	07:45
	← phantomcircuit_ left	07:45
	← saysjonathan left	07:46
rohanpm	I don't get what you mean	07:46
	you can make a tag pointing to a commit	07:46
	← ss23 left	07:46
rohanpm	once you've done that, the tag shouldn't change	07:46
digitalfiz	ok im sorry im just trying to grasp how this all works	07:47
	→ saysjonathan joined	07:47
digitalfiz	i think i understand	07:47
	→ hyperair joined	07:47
	← CrypTom left	07:48
	→ galderz joined	07:48
	← saysjonathan left	07:52
	→ saysjonathan joined	07:52
	→ dc5ala joined	07:53
	← jarrod_ left	07:54
	→ lightcap_ joined	07:56
	← saysjonathan left	07:56
	← lightcap_ left	07:56
	→ saysjonathan joined	07:57
	← Tianon left	07:58
	← joel_falcou_lab left	07:59
	← p3rg2z left	07:59
	→ xiangfu joined	07:59
	← mmc left	08:00
	→ joel_falcou joined	08:01
	→ MikhX joined	08:01
	→ napster_123 joined	08:02
	→ bentob0x joined	08:02
	← saysjonathan left	08:02
	joel_falcou → joel_falcou_lab	08:02
	→ saysjonathan joined	08:02
	→ dgtlmoon joined	08:03
	→ ExtraSpice joined	08:05
	→ khmarbaise joined	08:05
	→ mrtazz joined	08:06
	← saysjonathan left	08:07
	→ nevyn^_ joined	08:07
	→ saysjonathan joined	08:07
	← shiba_yu36 left	08:08
	← dreiss left	08:10
	← saysjonathan left	08:12
	→ saysjonathan joined	08:12
	← fujin left	08:14
	← dquestions left	08:16
	← saysjonathan left	08:17
	→ dquestions joined	08:17
	→ saysjonathan joined	08:17
	→ numbah joined	08:18
	← Mage left	08:19
	→ skoop joined	08:19
	→ cilly joined	08:21
	← saysjonathan left	08:22
	→ saysjonathan joined	08:22
	← lucasvo left	08:23
	→ VadtecWork joined	08:24
	← bentob0x left	08:24
	VadtecWork → Guest61475	08:24
	→ pantsman joined	08:26
	← saysjonathan left	08:27
	→ jpnurmi joined	08:27
	← hyperair left	08:27
	→ saysjonathan joined	08:27
	ChanServ set mode: +v	08:28
	← dgtlmoon left	08:28
	→ ochosi joined	08:28
	→ psoo joined	08:30
	← saysjonathan left	08:32
	→ saysjonathan joined	08:32
divVerent	I have a tricky problem...	08:32
	← gretch left	08:33
divVerent	my project is a fork from onerepo/branch_A	08:33
	onerepo/trunk is unrelated development line... MOSTLY	08:33
	but, I want to merge changes from onerepo/trunk that affect directory X	08:33
	→ gretch joined	08:33
divVerent	basically, I want to do merges that are restricted to a subdirectory... how to?	08:33
	(or actually, to three subdirectories)	08:34
	→ dibber joined	08:34
	→ tewecske joined	08:34
divVerent	I suppose it can be somehow done with nasty filter-branch magic... but then git svn cannot fetch updates to it	08:34
cbreak_work	git can't merge half histories	08:34
divVerent	sure, but is there a simple way to get what I want?	08:34
	filter-branching the repo down to the three subdirs, and merges should work fine	08:35
cbreak_work	create alternate history with filter branch and merge that	08:35
	← khmarbaise left	08:35
divVerent	but then how can I fetch updates from the upstream repo?	08:35
	or do I have to run the ages long filter-branch every time?	08:35
cbreak_work	of course	08:35
divVerent	there is no way to do that incrementally?	08:35
cbreak_work	there's nothing to increment uppon	08:36
	← beasty left	08:36
	→ beasty joined	08:36
cbreak_work	you could do some cherry picking	08:36
divVerent	no, often commits affect lots of places	08:36
	hm... one idea:	08:36
	assume I have a branch "master"	08:36
	can I make a branch "master-strippeddown" that is master stripped down to the three directories?	08:36
	and incrementally update that?	08:36
	← saysjonathan left	08:37
cbreak_work	as I said	08:37
	→ davep joined	08:37
cbreak_work	there's nothing to increment uppon	08:37
divVerent	there is	08:37
cbreak_work	it's completely independent history	08:37
	no common commit	08:37
divVerent	sure	08:37
cbreak_work	but as I also said	08:37
davep	im having https issues. can git function without it? it doesnt appear to be trying to use the proxy i set.	08:37
divVerent	assume I know the commit hash the tip of master-strippeddown is based on	08:37
cbreak_work	you could cherry pick	08:37
divVerent	I can track that in an extra file	08:37
cbreak_work	davep: git doesn't need network connections	08:37
	→ saysjonathan joined	08:37
	→ dytrivedi joined	08:37
divVerent	how would cherry pick help me?	08:38
davep	okay, assume i want to get a local copy of a remote repository over a network :P	08:38
divVerent	IIRC it cannot be restricted to paths	08:38
cbreak_work	sure	08:38
	→ guardian joined	08:38
cbreak_work	add -n, just commit the stuff you want	08:38
	davep: use a different protocol	08:38
divVerent	problem is, cherry pick will usually fail	08:38
	as commits will change files I do not want	08:38
davep	i changed the https in the url to http, but it fails	08:38
	i try a git pull and it reports a TLS error	08:39
cbreak_work	just ignore the failures	08:39
	← Keruspe left	08:39
davep	it ended up with an empty folder	08:39
cbreak_work	only commit the stuff you want and reset the rest away	08:39
divVerent	then I commit merge conflict markers :P	08:39
davep	:(	08:39
divVerent	hm... possibly doable with a complex ls-files and grep magic	08:39
cbreak_work	why?	08:39
divVerent	to reset all other files	08:39
cbreak_work	just only git add that directory	08:39
	→ Keruspe joined	08:40
divVerent	OH... cherry-pick does not add	08:40
	good	08:40
	← iravid left	08:40
divVerent	(when -n)	08:40
cbreak_work	davep: I meant "use a different protocol that git supports:	08:40
	like git, ssh, file"	08:40
	→ iravid joined	08:40
	← dytrivedi left	08:40
cbreak_work	divVerent: it puts it to the index, but you can easily reset the index away	08:41
	and then add the stuff you want again	08:41
davep	cbreak: git:// seems to have worked. thanks!	08:41
divVerent	right	08:41
	← acs left	08:41
divVerent	requires the work tree to be perfectly clean, so I'll have to use a separate local repo for thsi	08:41
	but yes, that sounds better	08:41
cbreak_work	why?	08:41
	are you someone who lives and works in dirt? :D	08:42
divVerent	sure :P	08:42
cbreak_work	a clean repo is to be expected...	08:42
divVerent	.o files for example	08:42
	← saysjonathan left	08:42
divVerent	most ops work fine in a nonclean repo :P	08:42
	and if not, they tell me	08:42
	→ Kenjin joined	08:42
divVerent	but, making that separate repo for these merging tasks sounds useful anyway	08:42
	→ saysjonathan joined	08:42
	← dc5ala left	08:42
divVerent	so I can keep state information somewhere	08:43
	so it knows what is the last "picked" commit	08:43
	→ gebi joined	08:43
	← gebi left	08:43
	→ gebi joined	08:43
	→ acs joined	08:43
cbreak_work	untracked files are not "dirty"	08:43
	they are just ... there	08:43
	and I seriously hope you don't track object files	08:44
divVerent	no, not tracking them	08:44
	but I mean	08:44
	→ lhunath joined	08:44
divVerent	git add -A or such stuff may add these	08:44
	or .swp files from vim :P	08:44
lhunath	how do I list all files in a directory from another branch?	08:44
	(without checking that branch out)	08:45
	→ ignatp joined	08:45
divVerent	that's why I would prefer cherry-pick, then reset all in wrong directory"	08:45
	instead of "cherry-pick, reset all, add these dirs"	08:45
	→ hyperair joined	08:45
	→ engla joined	08:46
	← engla left	08:46
	→ engla joined	08:46
	→ dc5ala joined	08:46
	← saysjonathan left	08:47
cbreak_work	lhunath: git show somebranch:somepath/	08:48
	→ saysjonathan joined	08:48
lhunath	cheers.	08:48
	← Kenjin left	08:51
	← saysjonathan left	08:52
	← yhager left	08:52
	← psankar left	08:53
	→ saysjonathan joined	08:53
	→ jetienne joined	08:56
	← digitalfiz left	08:56
jetienne	q. i got a submodule in my repo, and i want to change the version in it. aka set the submobule to a given tag	08:56
	← saysjonathan left	08:57
	→ saysjonathan joined	08:58
avar	cd submodule && git checkout tag	08:59
	→ amh345_ joined	08:59
	→ digitalfiz joined	09:00
	→ spaceonline joined	09:00
jetienne	avar: thanks	09:01
	← iravid left	09:02
	← saysjonathan left	09:02
	→ iravid joined	09:02
	← amh345 left	09:03
	amh345_ → amh345	09:03
	→ saysjonathan joined	09:03
	→ dr_lepper joined	09:04
	→ elmob joined	09:05
	← dquestions left	09:05
	→ dquestions joined	09:06
	← tchan left	09:06
	← skoop left	09:07
	→ tchan joined	09:07
	← saysjonathan left	09:08
	→ cordoval_ joined	09:08
	→ saysjonathan joined	09:09
	→ Bruce_Wayne joined	09:09
	← cordoval_ left	09:09
	← shorg left	09:10
	→ bpo joined	09:10
	→ lucasvo joined	09:11
bpo	I'm sure this is a frequent question, but a couple of searches didn't turn up the answer for me. What is the best way to distribute git hook scripts with a repository?	09:12
cbreak_work	inside a repo with an install script maybe. or separately.	09:12
	← saysjonathan left	09:13
jast	git doesn't ever do it automatically for security reasons	09:13
	→ saysjonathan joined	09:14
bpo	makes sense. I'll just put a symlink in our makefile. thanks.	09:14
	← _Vi left	09:15
	← sh1mmer left	09:16
	← pantsman left	09:16
	← saysjonathan left	09:18
	→ saysjonathan joined	09:19
	← ivank left	09:19
	→ sh1mmer joined	09:22
	→ dpino joined	09:23
	← saysjonathan left	09:23
	→ saysjonathan joined	09:24
	→ Tianon joined	09:24
	← Tianon left	09:24
	→ Tianon joined	09:24
	→ shruggar joined	09:26
	← lhunath left	09:26
	→ Schmallon joined	09:27
shruggar	I accidentally typed "git bisect bad" when I meant "git bisect good". How do I reverse my decision?	09:27
	← mmoya left	09:27
	← hyperair left	09:27
	← saysjonathan left	09:28
	→ saysjonathan joined	09:29
	← Schmallon left	09:31
	→ thehcdreamer joined	09:32
	← thiago left	09:32
engla	shruggar: maybe you can just tap in a new desicion if you use the <rev> argument to bisect? I never tried. git bisect good <the revision>	09:32
charon	shruggar: iirc you have to use log&replay	09:33
shruggar	tried that, it complains "<yadayada> was both good and bad"	09:33
charon	(and edit out that bad command of course)	09:33
thehcdreamer	Hi, I made a couple of commits on a branch. Is it possible to transfer these commits to a new branch? one solution I thought is to checkout on a new branch so that I have the latest commits, then go back to the original branch and do a reset. I wonder if there is another solution?	09:33
	← tildeequals left	09:34
	← saysjonathan left	09:34
	→ saysjonathan joined	09:34
	→ dels joined	09:34
	→ tvw joined	09:35
engla	thehcdreamer: it's simple to "duplicate" branches. git checkout -b <newname> and you have a new branch. then you can use rebase -i or similar tools to on each/any of them	09:35
jast	thehcdreamer: can't think of a shorter one	09:36
bob2	'git branch xxx HEAD~n ; git rebase master ; git co master ; git rset --hard HEAD~n' or so	09:36
	→ tildeequals joined	09:36
thehcdreamer	engla, jast, thanks	09:36
	→ tatsuya joined	09:37
	→ koltroll joined	09:37
	→ SvenDowideit joined	09:38
	← SvenDowideit_ left	09:38
	→ _Vi joined	09:38
	← saysjonathan left	09:39
	→ saysjonathan joined	09:39
	→ mklappstuhl joined	09:41
	→ parasti joined	09:41
	← dels left	09:42
Rickardo1	git commit -a --amend <<--- Does this undu the last commit?	09:42
jast	it changes the last commit	09:42
Rickardo1	jast: Define change	09:42
jast	it uses the current state of files and updates the last commit with it	09:42
Rickardo1	jast: ok, ty	09:43
	← BiggFREE left	09:43
	← saysjonathan left	09:43
Juri_Agarin	It's also a good way to fix a lacking commit message of the most recent commit.. :-)	09:44
bob2	or if you forgot to commit a file'	09:44
	→ saysjonathan joined	09:44
	← stliu left	09:48
	← OOPMan left	09:49
	← saysjonathan left	09:49
	→ saysjonathan joined	09:50
	← xiangfu left	09:52
	← mklappstuhl left	09:53
dunpeal	hi	09:53
	What's the recommended way to define shell aliases from the command line?	09:54
	I tried: git config --global alias.foo !"bar"	09:54
	and got: bash: !"git: event not found	09:54
	← saysjonathan left	09:55
jast	dunpeal: escape the '!'	09:55
	→ saysjonathan joined	09:55
dunpeal	jast: that results in "foo = !git bar && git baz"	09:56
jast	that's what you want, isn't it?	09:56
dunpeal	jast: I'd kind of like it to be `foo = !"git bar && git baz"`	09:57
	→ ramlev joined	09:57
dunpeal	if at all possible :)	09:57
jast	why?	09:57
	anyway, just put escaped quotes in there	09:57
dunpeal	jast: more redable	09:57
	*readable	09:57
jast	git config [...] \!"\"bar \""	09:57
	→ lanthan_afh joined	09:57
jast	or git config [...] '!"bar"'	09:57
	← shruggar left	09:58
jast	but personally I'd expect the quotes to mess things up	09:58
	nah, works	09:59
dunpeal	How do I make jast hm?	09:59
	err	09:59
	the first syntax doesn't seem to work?	09:59
jast	that's not a very detailed error description, you know :)	10:00
	← saysjonathan left	10:00
	→ saysjonathan joined	10:00
dunpeal	jast: $ git config --global alias.foo \!"\"bar \"" #=> foo = !\"bar \"	10:00
	← alnewkirk left	10:00
dunpeal	I want !"bar"	10:01
	← _sh3 left	10:01
jast	huh. interesting.	10:01
	well, try the other one	10:01
dunpeal	jast: same thing :/	10:01
jast	exactly the same result?	10:01
dunpeal	yes	10:02
jast	hmm	10:02
	← Athox left	10:03
jast	probably git is doing the escaping	10:03
	← JKL__ left	10:03
	→ OOPMan joined	10:04
	← saysjonathan left	10:05
dunpeal	jast: yeah, don't worry about it	10:06
	it works, which is what's important.	10:06
	→ saysjonathan joined	10:06
	← jspiros left	10:07
	→ Anti-X joined	10:07
	→ ayanes joined	10:07
	← ayanes left	10:07
	→ ayanes joined	10:07
	→ little_owl1 joined	10:09
dunpeal	jast: OK, another question:	10:09
	← saysjonathan left	10:10
	← bauruine left	10:10
dunpeal	I have a simple single-commit merge scenario with a graph that looks like this: http://dpaste.com/276768/	10:10
	→ shruggar joined	10:11
	→ curtana joined	10:11
dunpeal	On 3, what's the name of 2?	10:11
	→ saysjonathan joined	10:11
jast	I have no idea which direction the history goes in that picture :)	10:11
dunpeal	right	10:11
cbreak_work	what do you mean?	10:12
	call it child commit I guess	10:12
dunpeal	(parent)	10:12
	(it's RTL	10:12
	→ bluenovember joined	10:12
dunpeal	Basically, I need some symbolic name for all the commits that were merged into master.	10:12
	← QinGW left	10:12
jast	HEAD^1, HEAD^2 and so on	10:12
	→ Murr_ joined	10:12
dunpeal	jast: HEAD^1 can sometimes be 1	10:12
	I need only the branch commits.	10:13
jast	I have no idea what you're talking about now :)	10:13
	← hohoho left	10:13
dunpeal	To generalize (and hopefully finally make myself clear :)	10:13
	→ yt_ror joined	10:13
	→ chris_n` joined	10:13
	→ hohoho joined	10:13
dunpeal	A branch foo was merged to master. How do I get a list of foo commits that were merged?	10:13
	← iravid left	10:14
jast	git rev-list $(git merge-base mergepoint^1 mergepoint^2)..mergepoint	10:14
	if there is a simpler solution, I can't think of it right now	10:14
cbreak_work	dunpeal: HEAD^2 and it's parents	10:15
divVerent	jast: I think he wants git rev-list megrepoint^1..mergepoint	10:15
	← OOPMan left	10:15
divVerent	which would JUST be the commits merged from one of the two sides	10:15
jast	oh yeah, that's simpler :)	10:15
	→ iravid joined	10:15
	← chris_n left	10:15
	→ akamike joined	10:16
	← shruggar left	10:17
	← tg left	10:17
	→ dgtlmoon joined	10:17
	← saysjonathan left	10:17
	→ tg joined	10:17
	→ skoop joined	10:18
	← skoop left	10:18
	→ _sh3 joined	10:18
	→ saysjonathan joined	10:18
	→ ToreadorVampire joined	10:19
	→ dr_win joined	10:19
	→ skoop joined	10:20
	← nchaimov left	10:21
	→ airborn joined	10:21
	← saysjonathan left	10:22
	→ Kenjin joined	10:23
	← mishok13 left	10:23
	→ saysjonathan joined	10:24
	→ drizzd joined	10:24
	← Murr_ left	10:25
	→ mishok13 joined	10:25
	→ nadim joined	10:26
	→ Schmallon joined	10:27
	← drizzd_ left	10:27
	← saysjonathan left	10:28
	→ saysjonathan joined	10:29
	→ shruggar joined	10:29
	← sec^nd left	10:30
	→ JonasF joined	10:31
	← galderz left	10:33
	→ galderz joined	10:33
	← saysjonathan left	10:34
	→ sec^nd joined	10:34
	→ saysjonathan joined	10:34
	← elmob left	10:36
	← dunpeal left	10:37
	← tvw left	10:37
	→ aep joined	10:38
aep	hey, can git somehow convert all timestamps to _my_ timezone in git log?	10:38
	← saysjonathan left	10:39
	→ thierryp joined	10:39
	→ tampakrap joined	10:39
	→ mmc joined	10:39
	← Kenjin left	10:39
cbreak_work	aep: man git-log	10:40
jast	aep: the 'git-log' manpage can be found at http://git.or.cz/man/git-log [automatic message]	10:40
cbreak_work	check out --date=local	10:40
	→ saysjonathan joined	10:40
	→ tvw joined	10:41
	← steph021 left	10:42
	← thehcdreamer left	10:42
	← dquestions left	10:43
	← shruggar left	10:45
	← saysjonathan left	10:45
	→ dunpeal joined	10:46
	→ saysjonathan joined	10:46
	← sh1mmer left	10:46
	← cilly left	10:48
	→ cillybabe joined	10:48
	← skoop left	10:48
	← saysjonathan left	10:50
	← iravid left	10:51
	→ saysjonathan joined	10:51
ochosi	hi, i'm looking for an easy way to put git diffs in a webpage(preferably php), i don't need a real web-frontend or anything. any recommendations for that?	10:51
	→ iravid joined	10:52
FauxFaux	ochosi: gitweb or gerrit or just fork git?	10:52
	← dunpeal left	10:53
	→ dunpeal joined	10:53
	→ lorenzosu joined	10:53
ochosi	hm, gitweb seems to be pretty sophisticated, i'll give gerrit a try, thanks FauxFaux	10:54
ToreadorVampire	Ohai - I have some commits into my repos that were made by a developer who seems they don't know how to use git. Their commits are marked with author "unknown <somesillyusername@.(none)". These commits are published to the git repository and three developers have cloned these commits into their own repos'. Say I wanted to "fix" the author name of those commits - should I bother (this is rewriting history across multiple repositori	10:54
	es isn't it, thus a headache?) or is there a way I can do it and then propagate those author-name changes in some sane fashion?	10:54
lorenzosu	Pretty newb question: When using autotools, is it best practice to include all the generated files in a push or not?	10:54
jast	ToreadorVampire: not without manual intervention by everyone who has the history	10:54
ToreadorVampire	jast: Yeah, that's what I thought ... I'll just chastise the guy who failed to configure their author details and make sure they do it in the future ...	10:55
	← speakman left	10:55
jast	lorenzosu: personally I keep automatically generated files out of my repositories, or I put them in a separate branch. I've seen people do it differently, though.	10:55
	← saysjonathan left	10:55
	→ khmarbaise joined	10:55
	ChanServ set mode: +v	10:55
	→ nchaimov joined	10:55
acs	test	10:56
	I downloaded the sources of emacs, I changed the code, and I want now to revert the sources exactly like in the moment they were when I downloaded. How can I do it ?	10:56
ToreadorVampire	A test? I never had time to study!	10:56
acs	i have voice now :)	10:56
jast	acs: did you commit anything?	10:56
curtana	acs, all the sources? or a particular file	10:56
	→ saysjonathan joined	10:56
acs	test of voice	10:56
	a few files	10:56
	i changed a few files, and did nothing else	10:57
	← ToreadorVampire left	10:57
jast	git reset --hard destroys all uncommitted changes	10:57
lorenzosu	jast, I see.. I guess having only the ones required by autoreconf would be enough	10:57
acs	jast: thanks	10:57
	→ steph021 joined	10:57
curtana	'git checkout HEAD file1 file2' will just check out those files	10:57
	→ shruggar joined	10:57
	→ sh1mmer joined	10:57
acs	curtana: what does checkout do concretely ?	10:57
jast	lorenzosu: at any rate, including those files in source releases is very helpful for people, hence the idea of keeping the tree including auto-generated files in a separate branch	10:58
ochosi	FauxFaux: even gerrit is really complex and feature-rich, anything more crude?	10:58
	← EdwardIII left	10:58
jast	ochosi: just viewing diffs should take about a minute to implement in PHP :)	10:58
	← dunpeal left	10:58
lorenzosu	jast, interesting.. how would I go about setting that up?	10:58
curtana	acs, several things depending on the arguments. it switches branches, or replaces files in the working tree with their counterparts in the index (by default) or, if you name a commit-ish object, in that commit (e.g., HEAD here)	10:58
ochosi	jast: true, i just wanted to check whether i could save me those minutes ;)	10:58
acs	curtana: i see.	10:59
jast	ochosi: it probably takes less time to write it than to wait for answers ;)	10:59
	lorenzosu: I don't have a canned solution	10:59
lorenzosu	jast, ok :)	10:59
	← jetienne left	11:00
	→ dunpeal joined	11:00
ochosi	jast: well, i'm not very good in php and very new to git, so maybe not	11:00
	→ jelrikvh joined	11:00
	← gretch left	11:00
	← saysjonathan left	11:00
	→ saysjonathan joined	11:01
	→ hyperair joined	11:01
	→ Tricia joined	11:03
	→ dquestions joined	11:03
	Tricia → Guest91296	11:03
	→ shiba_yu36 joined	11:05
	→ gusnan joined	11:05
	← saysjonathan left	11:06
	→ schambers joined	11:06
	→ saysjonathan joined	11:07
	→ chrischambers joined	11:07
jast	ochosi: <?php $repo = '/path/to/repo'; if (preg_match('/[0-9a-f]{4,40}/i', $_GET['id'])) { passthru("git --git-dir=$repo show $_GET[id]"); exit; } echo "<html><body><ul>" foreach (`git --git-dir=$repo --rev-list --pretty=oneline HEAD` as $r) { $r = trim($r); list($rev, $title) = explode(' ', $r, 2); echo "<li><a href=\"$_SERVER[PHP_SELF]?id=$rev\">$rev: $title"; } echo "</ul></body></html>";	11:07
	(untested)	11:07
	okay, almost. the regexp needs to be anchored: /^[0-9a-f]{4,40}$/	11:08
	→ Kenjin joined	11:09
ochosi	jast: wow, thanks a bunch!	11:09
jim	stuff like that is why a good templating system is needed	11:10
	← Schmallon left	11:10
ochosi	jast: ya know, i'm just a poor scientist in the humanities and i'm starting to track my latex-changes via git	11:10
jast	stuff like this doesn't use templating systems by definition	11:10
	otherwise it wouldn't be a proper dirty hack	11:10
	← saysjonathan left	11:11
	← shruggar left	11:12
	→ saysjonathan joined	11:12
	← davep left	11:13
	→ urkud joined	11:13
	← Kenjin left	11:13
	← khmarbaise left	11:14
	← amason left	11:16
	→ itamarjp joined	11:16
	→ Kenjin joined	11:16
	← saysjonathan left	11:17
	→ amason joined	11:17
	← zorzar__away left	11:17
	→ saysjonathan joined	11:17
	→ d0k joined	11:20
	← jelrikvh left	11:21
	→ jspiros joined	11:22
	→ alberto56 joined	11:22
	← saysjonathan left	11:22
	→ saysjonathan joined	11:23
JasonWoof	I've got this in my post-update hook: if git diff --quiet origin/master..master	11:23
	but it doesn't seem to have fired then I pushed _to_ this repo	11:24
	is "master" updated after post-update?	11:24
	→ shruggar joined	11:24
JasonWoof	how do I detect if I've now got commits that I've not pushed upstream yet?	11:24
	← Kenjin left	11:25
	→ goshawk joined	11:27
	→ yunosh joined	11:27
	← saysjonathan left	11:28
yunosh	how do i correctly drop a submodule? removing the submodule's directory, or removing the entry in .gitmodules? or both?	11:28
	→ saysjonathan joined	11:28
albel727	yunosh: both.	11:29
yunosh	thx	11:29
	→ tizbac joined	11:29
	→ raptor360 joined	11:29
	→ Schmallon joined	11:29
raptor360	Hi all, I'm using git-svn and when doing a git svn rebase I get the following error: Could not unmemoize function `lookup_svn_merge', because it was not memoized to begin with at /usr/lib/git-core/git-svn line 3178	11:30
	→ zorzar__away joined	11:31
	← saysjonathan left	11:32
	← coldboot left	11:33
	→ bauruine joined	11:33
	← shruggar left	11:33
	→ coldboot joined	11:33
	→ Yuuhi joined	11:33
	→ saysjonathan joined	11:33
albel727	yunosh: you will also need to remove submodule entries in .git/config, though	11:33
yunosh	k	11:34
	yunosh wonders why there is no "git submodule remove" command	11:34
	albel727 too	11:34
	← hohoho left	11:35
albel727	JasonWoof: first off, are you sure you have that "origin/master" in the repo you push to? if it is a bare repo, then I find it highly unlikely. secondly, do you really need git diff? as man githooks can tell you, all modified refs are passed as arguments to the hook. you can just see, if "refs/heads/master" is in there.	11:36
jast	albel727: the 'githooks' manpage can be found at http://git.or.cz/man/githooks [automatic message]	11:36
albel727	wow. jast have set up gitbot =)	11:37
cbreak_work	albel727: if he does a git push --mirror, origin/master can appear there	11:37
	← jspiros left	11:37
	← saysjonathan left	11:37
albel727	cbreak_work: yeah I do realize that. but it's highly unlikely, that he wants difference between master and origin/master in that case.	11:38
cbreak_work	yeah...	11:38
jast	the update hook sounds like a better fit	11:38
	→ saysjonathan joined	11:39
albel727	very probably. ref name + "is/will be" hashes really sound good.	11:39
JasonWoof	albel727: it is bare, and didn't have an origin/master at first, but I did: git config remote.origin.fetch '+refs/heads/:refs/remotes/origin/'	11:40
	maybe that was not a good idea	11:41
albel727	depends on where you did executed that git config, and what did you want to achieve by it.	11:42
	← bauruine left	11:42
	→ bauruine joined	11:42
albel727	JasonWoof: ^	11:43
	← saysjonathan left	11:43
JasonWoof	albel727: very cool that updated refs are passed to post-update	11:43
	→ saysjonathan joined	11:44
raptor360	hi all again, removing the .git/svn/.caches seemd to fix it.	11:44
	→ Kenjin joined	11:45
albel727	raptor360: Thank you. Now I'll be able to help the next person with the same problem. =)	11:45
	→ shruggar joined	11:46
JasonWoof	albel727: I'm trying to set up a sort of proxy repo. I have about 20 clones of a repo (and regularly make more) and it seems silly keep waiting for gitorious on every clone and pull.	11:46
raptor360	albel727: you're welcome.	11:46
	→ marcob joined	11:46
	→ thiago joined	11:47
JasonWoof	also, I almost never push to gitorious from any computer besides this one, so I though I'd have one master/proxy clone (bare) that I set everything to push to, and set a hook on it to automatically push from there to gitorious	11:47
	I thought (guessed) that I should do some kind of check before pushing, so I was trying (as shown above) to check if I actually had anything new before pushing	11:48
	it works fine without this check	11:48
	sounds like I should instead check for "master" in the hook args	11:48
	← saysjonathan left	11:49
	→ saysjonathan joined	11:49
	→ bitkiller joined	11:50
	← masonkramer left	11:51
	← ramlev left	11:52
	← raptor360 left	11:52
albel727	JasonWoof: well, as long, as you understand, that everything breaks down the moment you push from any computer besides this one, I see no problem.	11:53
	← saysjonathan left	11:54
Arafangion	I'm effectively doing something similar - albiet with users who have no idea about git. This "breakdown" can be ignored and assumed never to occur if every person only ever pushes to their own branch.	11:54
	← thiago left	11:54
	← kpreid left	11:54
albel727	Arafangion: hmm, yeah, that kind of workflow solves it.	11:55
	→ saysjonathan joined	11:55
	← napster_123 left	11:55
	← shruggar left	11:55
	→ yotaff joined	11:55
	← schambers left	11:56
albel727	(how did you end up with "users who have no idea about git", anyway? =D)	11:56
Arafangion	albel727: Contractors of contractors.	11:56
	albel727: Or more simply, contractors.	11:56
	→ khmarbaise joined	11:57
albel727	grand-contractors, contractors of second order, complex contractors... I can see math ideas realizing in life, now =)	11:58
Arafangion	albel727: It's the kind of fun that happens when two companies cooperate to write a program... But the actual hiring of the programmers are left to HR guys.	11:59
	albel727: With no input whatsoever from the programmers.	11:59
	→ zombor joined	11:59
	← saysjonathan left	11:59
albel727	meh. they didn't even run qualification tests there?	11:59
	→ saysjonathan joined	12:00
	→ nevyn^__ joined	12:00
	← nevyn^__ left	12:00
	← Kenjin left	12:01
Arafangion	albel727: Some people satisfy all the tests, regardless.	12:01
	← zombor left	12:02
albel727	bad tests, then. =/	12:02
	← boombatower left	12:02
albel727	(anyway, tests or not... "no input whatsoever from the programmers"? insane =/)	12:03
jim	it's interesting that contractors are named for what you do in your body when thinking about hiring them...	12:03
Arafangion	albel727: I'm annoyed... I am probably overreacting.	12:03
	jim: What?	12:03
	← nevyn^_ left	12:03
jim	contract	12:03
albel727	Arafangion: I don't think that you are overreacting.	12:05
Arafangion	jim: Many things contract, mucles being one of them.	12:06
	→ saysjona1han joined	12:06
Arafangion	jim: It's named after the contract - the type that you sign.	12:06
	← saysjonathan left	12:06
	→ nevyn^_ joined	12:06
	← nevyn^_ left	12:06
cbreak_work	people who have no idea about git can get an idea about git	12:06
	easily	12:06
	by reading the man pages, and git-for-computer-scientists	12:07
Arafangion	cbreak_work: He was struggling with subversion.	12:07
cbreak_work	that's normal.	12:07
	I do that too...	12:07
Arafangion	cbreak_work: Did you struggle because it wasn't a lock-edit-commit cycle?	12:08
	→ shruggar joined	12:08
cbreak_work	no, because it doesn't do what it says it would be supposed to do	12:08
Arafangion	cbreak_work: And the fact that other people might be working on the same code?	12:08
	← hyperair left	12:08
Arafangion	cbreak_work: He'd have been right at home with VSS.	12:08
cbreak_work	and because I can't even easily branch off some feature for local development	12:08
	← khmarbaise left	12:09
	→ hyperair joined	12:09
	→ xiangfu joined	12:09
	← saysjona1han left	12:10
	→ khmarbaise joined	12:10
	→ shamrock joined	12:10
	→ nevyn^_ joined	12:11
	← alberto56 left	12:11
	→ saysjonathan joined	12:11
Arafangion	cbreak_work: Why would you want to branch? </sarcasm>	12:12
	→ artefon joined	12:12
cbreak_work	people always forget the start tags.	12:12
albel727	Arafangion: well, yeah, working with underqualified people means that you need to do the parts of work they cannot themselves, but hey... http://ecomodder.com/forum/member-cfg83-albums-shtuff-04-picture2486-you-gotta-do-what-you-gotta-do.jpg	12:14
	→ sftp joined	12:14
jim	start tags? what are those?	12:14
	← saysjonathan left	12:16
	← khmarbaise left	12:16
jast	it's an SGML/XML thing.	12:16
	→ saysjonathan joined	12:16
	→ vitor-br joined	12:17
albel727	jast: <sarcasm> so you need people to use proper sgml syntax in their speech. ok. </sarcasm> =D	12:18
Arafangion	albel727: Next cbreak_work is going to complain that you left out the doctype declaration.	12:19
albel727	(proper english would be of use, too, but that's too much for me =D)	12:19
	haha	12:19
cbreak_work	nah. this is transitional	12:19
albel727	lol	12:19
	← nevyn^_ left	12:20
	→ jspiros joined	12:21
	← saysjonathan left	12:21
	Arafangion can't believe that he's winding through custom error handling because he doesn't want to have his stack unwound automatically.	12:22
cbreak_work	!?!	12:22
	→ saysjonathan joined	12:22
cbreak_work	you working with morrons?	12:22
Arafangion	Well, in this case, I want to be able to optionally either raise a real error (ie, in my unit tests) - or pop up a warning message and try again later.	12:23
	← shruggar left	12:23
albel727	which language?	12:24
Arafangion	C#	12:24
	As much as I hate its main platform (Windows), and loathe the lack of proper RAII, it's a nice language.	12:25
	→ Ryback_ joined	12:25
	← Ryback_ left	12:25
	→ Ryback_ joined	12:25
albel727	So, you kinda want sometimes to throw an exception, and sometimes you want to just print a message, and go on?	12:26
	← saysjonathan left	12:26
cbreak_work	why not just do it? :)	12:26
Arafangion	Yes. Basically, sometimes I want to acknowledge that there's an error, and see if it fixes up later on. (ie, it's not really an /exceptional/ exception), although sometimes it could well be and I need to alert the user to it.	12:27
	→ HG` joined	12:27
Arafangion	But if I'm unit testing... I'd rather it simply abort, and print a stack trace as early and as noisily as possible.	12:27
cbreak_work	just catch it? :)	12:27
	→ saysjonathan joined	12:27
Arafangion	_where_?	12:27
cbreak_work	where ever you want	12:27
Arafangion	I'm using a ton of databinding and deferred execution (thanks to Linq)	12:28
	→ akamike_ joined	12:28
Arafangion	And I don't want to have the "display the exception to the user" logic scattered throughout the application either.	12:28
	→ schambers joined	12:28
cbreak_work	you could put it into the exception class :D	12:28
	when it's instanciated, it ... man, that's evil	12:29
Arafangion	Glad you realise. :) Mind you, I do excactly that when I'm doing logging... But that's completely different.	12:29
	← orafu left	12:29
	→ orafu joined	12:29
Arafangion	This is more for things such as "The http host is down".	12:29
	→ khmarbaise joined	12:30
albel727	Arafangion: I'd say, that needs just a single helper method/class, which will take a exception object, and either throw it, or output a message, depending on a value of a single static variable...	12:30
Arafangion	albel727: Exactly what I'm doing. :)	12:30
	← akamike left	12:30
cbreak_work	you could probably do something with template magic	12:30
	or macro magic	12:30
Arafangion	albel727: I have two such classes, actually, sharing a common interface. Unit tests use one, and the main program uses another. :)	12:30
	cbreak_work: Macro magic is evil, though - templates are much nicer. Alas, this is C#, not C++.	12:31
	cillybabe → cilly	12:31
	albel727 was preparing to be killed for horrible good-coding-style violation, but now, seeing that someone actually uses the same approach... =D	12:31
	→ skoop_ joined	12:31
Arafangion	albel727: Reduce the stench by using pure-DI. :)	12:31
	← saysjonathan left	12:32
	→ metasyntactic joined	12:32
	← akamike_ left	12:32
	→ saysjonathan joined	12:33
Arafangion	It's interesting that this ability to either throw an exception, or continue from that point on, is almost what SEH exceptions allow.	12:34
	/almost/.	12:34
	→ Unknown[NF] joined	12:34
Arafangion	(The way SEH exceptions allowed it was too evil, though)	12:34
	→ shruggar joined	12:35
	← kunwon1 left	12:35
albel727	never worked with SEH directly =/ how do they allow it?	12:36
	← saysjonathan left	12:37
	→ saysjonathan joined	12:38
	→ ehabkost joined	12:38
	→ akamike joined	12:38
	→ chrisred joined	12:39
Arafangion	albel727: Easily - handle the exception as a first-class exception, with the stack preserved, and then jump back.	12:39
	albel727: Usually with disasterious results b ecause teh code assumes that never happens. :)	12:39
albel727	Arafangion: I see. Well, Lisp "exception" handling allows doing that (stably) and more... why won't you convert to it? =D	12:40
	→ kpreid joined	12:40
Arafangion	albel727: So I've implemented yet another part of lisp, then.	12:41
	← ExtraSpice left	12:41
albel727	Arafangion: Yet another? What were the other parts of Lisp, you had to implement, just because of your poor development language choice? =D	12:43
	← saysjonathan left	12:43
	← kanru left	12:43
	→ saysjonathan joined	12:43
Arafangion	albel727: Type conversions. :(	12:46
	← urkud left	12:47
	→ ExtraSpice joined	12:47
	→ urkud joined	12:47
	← saysjonathan left	12:48
	← shruggar left	12:48
	→ saysjonathan joined	12:48
	→ tydeas joined	12:49
Arafangion	albel727: Recently the framework has linq... Which is almost one of the great parts of Haskell.	12:49
albel727	Arafangion: yeah, and even more recently it get PLINQ, which is no less useful to me.	12:50
tydeas	Hello i have made some modifications in some files. Now i want to move this modifications to a branch.	12:51
albel727	(not part of Haskell, but still... =D)	12:51
	(and yes, I like Haskell, too =D)	12:51
FauxFaux	tydeas: stash, checkout otherbranch, stash pop.	12:51
tydeas	FauxFaux: thanks	12:51
	albel727 is ashamed, since someone else actually helps here, while he is blabbing about C# =/	12:52
	← normanrichards left	12:52
	→ OOPMan joined	12:52
	← khmarbaise left	12:53
cbreak_work	modifications are moved to a branch by committing them	12:53
	→ tlonim joined	12:53
	← saysjonathan left	12:54
Arafangion	If you want to write them to a new branch, merely 'checkout -b newbranch HEAD' followed by a commit would be sufficient.	12:54
	→ saysjonathan joined	12:54
	← chrischambers left	12:55
	← leroux left	12:55
	← iravid left	12:55
	← shiba_yu36 left	12:55
	← bauruine left	12:55
	← _Vi left	12:55
	→ iravid joined	12:55
	← gusnan left	12:56
	← devanjan left	12:56
	→ chrischambers joined	12:56
	← kpreid left	12:56
	→ kpreid joined	12:58
	← galderz left	12:58
	← saysjonathan left	12:58
	→ shiba_yu36 joined	12:59
	→ saysjonathan joined	12:59
	← kenneth_reitz left	13:00
	→ squentin joined	13:01
	→ shruggar joined	13:01
	→ snearch joined	13:01
	skoop_ → skoop	13:02
	← Amorphous left	13:03
	→ bauruine joined	13:03
	← saysjonathan left	13:04
	→ saysjonathan joined	13:05
	→ Chaser joined	13:06
	→ rolfb joined	13:06
	→ alberto56 joined	13:08
	← saysjonathan left	13:09
	→ saysjonathan joined	13:10
	← bauruine left	13:10
	→ cirwin joined	13:10
	→ bauruine joined	13:10
	→ hohoho joined	13:12
	← skoop left	13:12
	→ bentob0x joined	13:12
	→ zomgbie joined	13:13
	→ mabrand joined	13:13
	← lamont left	13:13
	← saysjonathan left	13:14
	→ lamont joined	13:15
	→ saysjonathan joined	13:15
	→ pfrog joined	13:16
	← chris_n` left	13:17
pfrog	how would I sync up a local and remote branch? git reset ?	13:17
zomg	pull from the remote to get changes, push to the remote to send changes	13:17
	← q0rban\|afk left	13:17
zomg	reset can be used if you want to discard all changes and just go directly to the state the remote is in	13:17
tango_	unless you mean that you pushed to a non-bare remote repo and the workdir on the remote repo was not updated, in which case it IS git reset	13:17
	(on the remote repo)	13:18
pfrog	ok maybe what I'm asking instead is how do I get my local branch to track a remote one	13:18
	→ benishor joined	13:18
	→ Amorphous joined	13:18
pfrog	but the local one exists already	13:18
	ChanServ set mode: +v	13:19
	→ madsy joined	13:19
	← madsy left	13:19
	→ madsy joined	13:19
benishor	Hello all. Is there a way to make git remember webdav password?	13:19
cbreak_work	git doesn't care about things like passwords	13:19
	don't think it can do that	13:19
	← saysjonathan left	13:19
benishor	I'm using tortoisegit in order to push to a webdav remote repository	13:20
	it is constantly prompting for password	13:20
	and it's pretty tiresome	13:20
	→ q0rban joined	13:20
benishor	Is there any way to fix that?	13:20
	→ saysjonathan joined	13:20
cbreak_work	at least with ssh, git delegates things like authentication to the client	13:21
benishor	I just fixed it. I'm using http://user:pass@server/repo	13:22
	embeded url authentication	13:23
	works wonders \o/	13:23
	→ ntoll joined	13:23
	→ Fullma joined	13:24
	← saysjonathan left	13:25
	→ PerlPilot joined	13:26
pfrog	nm I figured it out	13:26
	→ saysjonathan joined	13:26
	→ itamarjp1 joined	13:26
	← itamarjp left	13:26
	→ jaql joined	13:27
	← foocraft left	13:28
	← PerlJam left	13:29
	→ foocraft joined	13:29
	← amh345 left	13:29
	← saysjonathan left	13:31
	→ cannonball joined	13:31
	← cannonball left	13:31
	→ cannonball joined	13:31
	→ saysjonathan joined	13:31
	← bauruine left	13:32
	→ IslandUsurper joined	13:32
	← HG` left	13:33
	→ napster_123 joined	13:34
	← urkud left	13:35
	← saysjonathan left	13:35
	→ saysjonathan joined	13:36
	← ntoll left	13:37
	→ warlock_mza joined	13:39
	→ masonkramer joined	13:40
	← saysjonathan left	13:41
	→ saysjonathan joined	13:41
	← zomgbie left	13:44
	← saysjonathan left	13:46
	→ MattDiPasquale joined	13:46
	→ saysjonathan joined	13:46
	← richard28530 left	13:47
	← Guest61475 left	13:48
	→ Guest61475 joined	13:48
	Guest61475 → VadtecWork	13:48
	→ elmob joined	13:51
	← saysjonathan left	13:51
	→ saysjonathan joined	13:52
	← Bruce_Wayne left	13:52
	→ mbroeker joined	13:55
	← saysjonathan left	13:56
	→ saysjonathan joined	13:57
	← sonnym left	13:59
	← artefon left	14:00
	→ rindolf joined	14:01
	← justin-george left	14:01
rindolf	Hi all, how do I get a git log -p on the entire repository? All branches.	14:01
	← saysjonathan left	14:02
	engla → engla-on-the-wal	14:03
	→ saysjonathan joined	14:03
cbreak_work	rindolf: by passing --all...	14:03
	it's in the man page	14:03
rindolf	cbreak_work: OK, thanks.	14:03
	→ nevyn^_ joined	14:04
	→ khmarbaise joined	14:05
	← khmarbaise left	14:07
	← saysjonathan left	14:07
	→ saysjonathan joined	14:08
	→ intripoon joined	14:08
	engla-on-the-wal → engla	14:09
	→ malfy_ joined	14:10
	← malfy_ left	14:10
	→ malfy_ joined	14:10
	← malfy_ left	14:10
	← benishor left	14:10
	← shiba_yu36 left	14:10
	→ wuj joined	14:11
	→ SjB joined	14:11
	← cilly left	14:11
	→ cilly joined	14:11
	← saysjonathan left	14:12
	← epid left	14:13
	→ saysjonathan joined	14:13
	← alberto56 left	14:14
molokoi	hello.	14:15
	← acs left	14:17
	← rchavik left	14:17
	→ malfy_ joined	14:17
	← malfy_ left	14:17
molokoi	I keep track of central repository in origin/master thank to fetch. then, git checkout origin/master	14:17
	← saysjonathan left	14:18
molokoi	first of all, at git branch i get a *(no branch). Could I have a real branch instead ?	14:18
cbreak_work	sure	14:18
	make one	14:18
	or do a git checkout of one that exists	14:18
	→ felipe joined	14:19
cbreak_work	remote tracking branches do NOT exist locally	14:19
	you can create one with git checkout -t origin/master	14:19
	→ saysjonathan joined	14:19
molokoi	nice.	14:19
	then, can I merge only one file from master into origin/master ?	14:19
cbreak_work	that makes no sense, no	14:19
molokoi	ok.	14:19
cbreak_work	git's history is atomic, commit based	14:20
	a commit can not be split	14:20
	← kpreid left	14:20
molokoi	I guess I should commit my local configuration into a local branch, then merge with that one ?	14:20
	→ Bisaram`office joined	14:20
cbreak_work	no	14:21
	configurations should not be part of a repo	14:21
molokoi	oh ;-)	14:21
	← foocraft left	14:21
	← Bisaram left	14:21
molokoi	thanks for you answers.	14:21
	← Adaptee left	14:22
	→ fr0sty joined	14:23
	← saysjonathan left	14:23
	→ TeckniX joined	14:24
	→ saysjonathan joined	14:24
cbreak_work	no problem	14:24
	← lorenzosu left	14:25
	→ JKL__ joined	14:26
	→ edude03 joined	14:26
	→ coderdad joined	14:26
	← roop_ left	14:27
	← iravid left	14:29
	→ iravid joined	14:29
bnovc	msysgit is so buggy :(	14:29
	← rindolf left	14:29
	← Rickardo1 left	14:30
	→ kylehayes joined	14:30
	→ gemmes joined	14:30
bnovc	so normally at work we only do one level of branching off of master for features but someone has a rare case and wants a branch off his branch. While that's the right git way to do it, it doesn't really work with our versioning scheme and some tools, so it seems like it would be equally acceptable for him to mimic his branch (create another from the same point and cherry-pick all his changes).... can anyone see a problem w/doing that?	14:31
	→ kpreid joined	14:32
	→ mxmdrone joined	14:32
	← rolfb left	14:33
	→ rolando joined	14:33
	← saysjonathan left	14:33
	→ nonno joined	14:34
	→ saysjonathan joined	14:34
cbreak_work	bnovc: if c branches of b	14:36
	and b branches of a	14:36
	→ EricInBNE joined	14:36
cbreak_work	then b and c share history, before they branched off	14:37
	it belongs to both	14:37
	both have branched of a at the same time :)	14:37
	nothing wrong with just doing the normal thing	14:37
bnovc	are you saying that c and b both branch from the same commit? c isn't really even a branch of b at that point though	14:38
	← saysjonathan left	14:38
bnovc	the problem is that we have versioning and tools setup that assume one level off of master	14:38
patrikf	bnovc: well you said you wanted C to branch off B.	14:38
	→ saysjonathan joined	14:39
	→ osanaisho joined	14:39
	→ sonnym joined	14:40
	← frogonwheels left	14:42
cbreak_work	bnovc: branching is transitive	14:43
	if c branches of b, and b branches of a, then c also branches of a. at the same commit b does. and it then has the same commits as b does, for some time.	14:43
	it's like a tree	14:43
	you can't say "ZOMG, This leaf is not connected to the stem!"	14:43
	← saysjonathan left	14:43
	→ mxmdrone_ joined	14:44
bnovc	ya, I understand that	14:44
	metasyntactic → kunwon1	14:44
bnovc	I was trying to find a free drawing program to show what I meant but oh well	14:44
	→ saysjonathan joined	14:44
patrikf	there's some fancy AJAX stuff out there, but I can't remember it's url	14:44
	← EricInBNE left	14:45
	← SvenDowideit left	14:45
	→ ooooPsss joined	14:45
bnovc	http://dev.bnovc.com/tmp/branches.txt	14:45
	like that	14:45
cbreak_work	bnovc: http://www.dabbleboard.com/	14:46
	→ ereslibre joined	14:46
	← ereslibre left	14:46
	→ ereslibre joined	14:46
cbreak_work	bnovc: it'd be stupid	14:46
patrikf	bnovc: cherry-picking X on its parent commit results in X.	14:46
cbreak_work	why do you not want e and e' to be the same?	14:46
patrikf	i.e. it's the same commit	14:46
	← mxmdrone left	14:47
	→ zombor joined	14:47
bnovc	patrikf: well I was designating that it is in two places	14:47
	which I'm pretty sure it would be	14:47
	→ mfacenet joined	14:47
	← mfacenet left	14:47
	→ mfacenet joined	14:47
bnovc	I mean this seems hacky but I didn't see what problems it would cause	14:48
cbreak_work	bnovc: it's pointless	14:48
	why do you care that e and e' is the same?	14:48
bnovc	because we version based on your branch position	14:48
	sec	14:48
	← saysjonathan left	14:49
cbreak_work	but do you check if that commit is in an other branch?	14:49
fr0sty	bnovc: what does that even mean?	14:49
bnovc	http://dev.bnovc.com/tmp/feature_branches.png	14:49
fr0sty	the name of your branch? e.g. branch = 1.5.4 ?	14:49
bnovc	like that	14:49
	(those are tags)	14:49
	where red is master	14:50
	→ saysjonathan joined	14:50
fr0sty	what is the point of all the tagging?	14:50
	← airborn left	14:50
cbreak_work	man, that sounds bad...	14:50
fr0sty	are you tagging each commit?	14:50
patrikf	bnovc: so what's wrong if right now, the new branch still has "oldbranch" as it's most recent tag, and the next tag is "newbranch"?	14:50
bnovc	if someone has a release off of a feature branch, we want to know what fixes it included from master easily	14:50
	fr0sty: no	14:50
	fr0sty: I just was drawing boxes there are releases	14:51
patrikf	why would you release off a feature branch?!	14:51
	→ madewokherd joined	14:51
bnovc	they're not releases to customers...they're releases to SQ	14:51
	to ensure stability before going back into master	14:51
cbreak_work	ever heard of git merge-base?	14:52
bnovc	patrikf: I'm not sure what you meant by your previous question	14:52
	← PerlPilot left	14:52
patrikf	bnovc: and "git describe" is not enough as a version number for qa?	14:52
	→ PerlJam joined	14:52
bnovc	cbreak_work: how does it that affect this?	14:53
cbreak_work	you don't need to make sure that e and e' are not the same	14:54
bnovc	I don't mean that they're actually different. I mean that e' is a cherry-pick of e onto the bottom branch	14:54
cbreak_work	because merge-base will always find the branch point	14:54
	why?	14:54
	well, as soon as you rebase one of the two branches, they will get rewritten anyway	14:54
	but... what's the point wasting time with cherry picking?	14:55
	just branch off what ever branch you want!	14:55
bnovc	because in the top situation, the branch that 'h' is on is two levels out, so it doesn't have a good signifier for its version number	14:55
	← saysjonathan left	14:55
	→ saysjonathan joined	14:55
bnovc	patrikf: I didn't know that existed, but that's interesting..... but if I had two levels of branches, it still wouldn't show what features were incorporated in the change that way	14:56
cbreak_work	bnovc: no	14:56
	bnovc: branches do not have levels	14:56
bnovc	patrikf: it would show the last tag on master	14:56
	→ alberto56 joined	14:56
cbreak_work	they are all the same level	14:56
	there's no difference between master and h	14:57
	as far as git is concerned, you could just rename them	14:57
bnovc	there is in process though	14:57
	master is significant, so its a count off of master	14:57
	→ rayners joined	14:57
cbreak_work	h is branched of master.	14:57
	and g too	14:58
bnovc	h is branched off of f	14:58
	sec let me refresh my diagram	14:58
cbreak_work	no, g is	14:58
fr0sty	bnovc: f is the point where h and g diverge	14:58
	one is not 'branched' off the other.	14:58
	bnovc: branches don't have 'starting points'	14:59
bnovc	ok refresh the .txt	14:59
fr0sty	talking about where a branch was 'created' is non-sensical in git's world.	14:59
bnovc	my point is that when I tag a release on "branch C", our scheme can't indicate that it has changes from "branch B" also	14:59
	unless you guys have a better suggestion for versioning which I'm totally open to	15:00
fr0sty	bnovc: why is that important?	15:00
	← sambarino left	15:00
bnovc	because SQ wouldn't know that they're testing additional features that were added in B	15:00
	← saysjonathan left	15:00
bnovc	or supposed to be	15:00
	→ saysjonathan joined	15:01
fr0sty	but why is that important? the features 'from b' (not really from there, but I'll grant it for now) are required for C anyway.	15:01
	and they will be tested when they test B as well.	15:01
	I'm still failing to understand what the problem is beyond 'we have an inflexible tagging scheme'	15:02
bnovc	ya, that is the problem, basically	15:02
	→ Vampire0 joined	15:02
	← Vampire0 left	15:02
	→ Vampire0 joined	15:02
fr0sty	btw, still tagging the way you do should work fine.	15:02
	the fact that branches have common commits should not cause anyone greif.	15:02
	→ Ruudjah joined	15:03
	← rolando left	15:03
	→ HG` joined	15:03
bnovc	I think it matters that the tag includes it though. In http://dev.bnovc.com/tmp/feature_branches.png, if I create featurename-1.0.1-1, then SQ knows that it should include all fixes from 1.0.1	15:03
	but in featurename-1.0.0-1, they realize it doesnt	15:03
fr0sty	bnovc: is SQ incapable of using git log --graph?	15:04
	← freimatz left	15:04
fr0sty	or gitk	15:04
bnovc	our SQ has no knowledge of git	15:04
	they don't have checkouts/etc.	15:04
	we have a lot of non-technical SQ	15:04
	(not my decision and not defending that, but its how it is)	15:04
fr0sty	your branching strategy is the least of your worries it seems...	15:05
bnovc	well a lot of the tests they perform really don't require technical knowledge	15:05
	we have some technical SQ too but it needs to be flexible for both	15:05
fr0sty	still, it would be trivial to write a little wrapper on 'git log --graph --decorate master..tag-to-test' to show them what is in play	15:05
	← tydeas left	15:05
	← saysjonathan left	15:05
	→ gusnan joined	15:05
	← Guest60181 left	15:06
bnovc	I think that would be overwhelming to them to see a lot of things they don't understand	15:06
	that doesn't correlate to the testing plan	15:06
	I was considering that before though	15:06
fr0sty	then just let them test things twice	15:06
	never hint that branches share history	15:06
	and get on with your work.	15:06
	→ saysjonathan joined	15:06
bnovc	well, we don't want them reporting issues for every branch they test	15:06
	thats a lot of duplication	15:06
fr0sty	if they are neve looking at the history it doesn't matter how it is organized	15:06
bnovc	and we don't want them not reporting regressions that they think are duplicates	15:06
	(and I'm really not trying to defend this versioning system but trying to see your thoughts on how to improve it)	15:07
	it seemed like the issue of redoing the versioning system could be circumvented by just having a 2nd branch with cherry-picks from the other one, leaving it one level away	15:07
	but I couldn't think of what issues that may cause	15:07
cbreak_work	just tell them that abc-1 and def-1 are the same.	15:07
fr0sty	bnovc: if you have a dependent branch don't send it to SQ until the original branch has been tested.	15:08
	→ mtkd_ joined	15:08
fr0sty	or set the branching point of B and C (from your txt) as a testing point and then test B and C after that.	15:08
	this is where merge-base comes in later.	15:08
	→ SvenDowideit joined	15:09
	← Aaaarg left	15:09
fr0sty	if SQ can test the code up to each fork in the road then testing the branch-heads will not result in duplicate bugs.	15:09
bnovc	hm...that seems true	15:09
fr0sty	another alternative would be to merge your feature branches into an SQ branch and tag that for periodic testing.	15:09
	that way they will test batches of changes and you can figure out the regressions yourself.	15:10
	→ boombatower joined	15:10
	← boombatower left	15:10
	→ boombatower joined	15:10
	→ Razz joined	15:11
bnovc	ok but just in the case of one branch... if it was from 1.0.0 and they test it, then anther release on that branch without merging up to master, how do they know whether to re-report the bugs thye find	15:11
	← saysjonathan left	15:11
	→ saysjonathan joined	15:11
	→ zomgbie joined	15:12
Razz	Does anyone know the current status of a Git library (pref. written in C).	15:12
	← mtkd left	15:12
fr0sty	bnovc: do you have an issue tracker?	15:12
bnovc	yes	15:12
fr0sty	just leave unresolved bugs/tickets 'open'	15:12
	← bluenovember left	15:13
	← mmc left	15:13
fr0sty	they can skip anything that wasn't marked as 'resolved'	15:13
	← Tianon left	15:13
	→ babusri joined	15:13
	→ nullvariable joined	15:13
[exa]	Razz: I'm kindof afraid that there's no such thing coming. Git is too simple to actually have a library, and commandline frontends/porcelains are a little bit hard to simulate in simple C	15:13
bnovc	fr0sty: but that only works if they're only ever testing feature branch releases that are after the last master they test	15:14
	← mlq left	15:14
	→ theflow joined	15:14
	→ mlq joined	15:14
	← mlq left	15:14
	→ mlq joined	15:14
bnovc	fr0sty: because if the issue is resolved on master but the feature hasn't pulled it in, it will look like a new bug to them	15:14
	→ alin-gym joined	15:14
fr0sty	Razz: what are you trying to accomplish?	15:14
[exa]	Razz: people usually take some of git sources and build them into their projects. Nice example of such thing is cgit, imho worthy reading	15:14
fr0sty	bnovc: then pull master in before testing.	15:14
	bnovc: how big is your company?	15:15
bnovc	I'm not sure that thats always a reasonable requirement	15:15
fr0sty	# of devs # of SQ folk	15:15
Razz	[exa]: thx, I'll lookinto that.	15:15
bnovc	fr0sty: I don't know... hundreds of each. My department is (sort of) new...there are maybe 50 devs here and I don't know how many SQ	15:15
Razz	I'm trying to build a versioning file system (as an assignment)	15:15
	fr0sty: ^	15:16
	← saysjonathan left	15:16
fr0sty	Razz: most front ends just wrap the command line using the --porcelain options	15:16
bnovc	jgit is an API to git	15:16
selckin	copy-on-write: done?	15:16
	→ yairgo joined	15:16
Razz	fr0sty: I would prefer not using the cmdline but I'll keep it as an option	15:17
	→ saysjonathan joined	15:17
fr0sty	bnovc: unless you want SQ to go insane they should probably have an idea of what issues are supposed to be fixed by each revision they test (and which are still open).	15:17
bnovc	fr0sty: hence our versioning system! :)	15:18
	→ Aaaarg joined	15:18
fr0sty	Razz: man system you can invoke all the git goodness from C :-)	15:18
	→ xiong joined	15:19
Razz	fr0sty: right, but that's not a very pretty solution ;-)	15:19
	→ tstclair_ joined	15:19
	← tstclair left	15:19
fr0sty	or fork+exec	15:19
	this is a class project. simple wins over pretty.	15:20
	and wrapping the command line will be simpler than hooking into the source	15:20
	← napster_123 left	15:20
Razz	not sure, they want a nice design and some real research done into versioning fs's	15:20
	← unabonger left	15:21
Razz	also, they require the use of FUSE, but that should (AFAIK) not interfere with still using the cmdline	15:21
	← saysjonathan left	15:21
	← OOPMan left	15:22
bnovc	fr0sty: so it seems like I'm back to our vesrioning system and one level....and it really doesn't seem like cherry-picking and having duplicate changes on a 2nd branch causes problems except that its ugly to have duplication	15:22
	→ saysjonathan joined	15:22
	← mfacenet left	15:22
doener	Razz: the only lib I know of is libgit2, but I have no idea about its current status -- http://repo.or.cz/w/libgit2.git	15:23
	← osanaisho left	15:24
Razz	doener: thx	15:24
	← Mpenz left	15:25
	← opalka left	15:25
fr0sty	doener: that was defunct-ish last I heard.	15:26
	→ mr789 joined	15:26
	→ Mpenz joined	15:26
	← saysjonathan left	15:26
	← Knirch left	15:26
	→ mmc joined	15:27
	→ mfacenet joined	15:27
	→ saysjonathan joined	15:28
	← snearch left	15:28
	→ Bass10 joined	15:28
	→ chrislerum joined	15:28
	→ Bass2 joined	15:29
	→ drl joined	15:29
shruggar	from the look of things mailing-list side, I'd say that things are (glacially, and far from universally) being written with libization in mind	15:29
	→ fowlduck joined	15:29
	← Bass10 left	15:30
	← dquestions left	15:30
	← Bass2 left	15:30
	→ Bruce_Wayne joined	15:30
	→ j0shua joined	15:30
	← noktoborus_ left	15:30
	→ Bass10 joined	15:30
	→ dquestions joined	15:31
	→ mkramer joined	15:31
	→ Bass2 joined	15:31
	← Bass10 left	15:31
	← saysjonathan left	15:32
	→ lsls01 joined	15:32
	→ napster_123 joined	15:32
	← albel727 left	15:32
	← little_owl1 left	15:32
	→ saysjonathan joined	15:33
	→ mkramer1 joined	15:33
	→ pigdude joined	15:34
	→ simosx joined	15:34
	← burhan left	15:34
bnovc	fr0sty: I would love for you to convince me there's a better method, but I'm not seeing it, given our setup	15:34
doener	fr0sty: at least it took part in GSoC and had new commits just 2 weeks ago	15:35
	← mkramer left	15:35
	→ normanrichards joined	15:35
fr0sty	bnovc: in both cases you are going to be testing commits twice	15:35
	which is what you wanted to avoid.	15:35
	or so I thought.	15:36
	← mkramer1 left	15:36
	← saysjonathan left	15:37
bnovc	well that part is unavoidable...the thing is that I can still designate where it came off of master easily with this verisoning scheme and 1 level of branching	15:37
	fr0sty: all of that aside, do you see any problems with doing the cherry-picking?	15:37
	beside it seeming a bit ugly	15:37
	→ noktoborus_ joined	15:37
fr0sty	it's unnecessary	15:37
	← nevyn^_ left	15:38
fr0sty	whether the commits are cherry-picked or not, just use the same naming scheme and save yourself the trouble.	15:38
	→ OOPMan joined	15:38
	→ saysjonathan joined	15:38
bnovc	hmm	15:38
	→ nevyn^_ joined	15:38
fr0sty	in both cases the two branches will have 'duplicate' commits (from one point of view), but this way the repository only has one copy of the changes	15:38
	this will also make your merging a little simpler.	15:38
doener	cherry-picking means duplication and indirection (when looking for the "original" commit)	15:39
bnovc	I guess that's true that maybe I could just version the same	15:39
fr0sty	you also lose the ancestry of changes.	15:39
	→ t0rc joined	15:39
bnovc	since applying this system to a second branch still won't represent the duplication	15:39
	→ gh34 joined	15:40
doener	and since you seem to be talking about testing: Testing a cherry-picked commit is not the same as testing the original commit	15:40
	(unless they happen to be tree-same)	15:40
bnovc	it is if its cherry-picked to the same parent	15:40
	ya	15:40
	← _sh3 left	15:41
	← ExtraSpice left	15:41
	→ alester joined	15:42
	← kpreid left	15:42
	→ albel727 joined	15:42
	← fr0sty__ left	15:42
	← mr789 left	15:42
	← saysjonathan left	15:43
	→ fr0sty__ joined	15:43
	→ epid joined	15:43
	→ saysjonathan joined	15:43
	← napster_123 left	15:44
	IslandUsurper → IslandUsurperAFK	15:44
	← Schmallon left	15:45
	→ smuf joined	15:46
	← dc5ala left	15:46
	← warlock_mza left	15:47
	→ khmarbaise joined	15:47
	← saysjonathan left	15:48
	→ saysjonathan joined	15:48
	→ alex__c2022 joined	15:48
	→ Schmallon joined	15:48
	→ thm joined	15:49
	← nevyn^_ left	15:49
	← Anti-X left	15:50
	← sec^nd left	15:50
	→ beatak joined	15:51
	→ sec^nd joined	15:52
	→ jrmuizel joined	15:52
	← lsls01 left	15:52
	→ chrisshattuck joined	15:53
	→ Adaptee joined	15:53
	→ ExtraSpice joined	15:53
	← saysjonathan left	15:53
	→ saysjonathan joined	15:53
	← thierryp left	15:54
	→ thiago joined	15:54
	thiago → thiago_home	15:54
	← yt_ror left	15:55
	→ Anti-X joined	15:56
	← khmarbaise left	15:56
	→ kpreid joined	15:57
	← saysjonathan left	15:58
	← dr_lepper left	15:58
	→ eto joined	15:58
	→ khmarbaise joined	15:58
	→ saysjonathan joined	15:59
	→ yt_ror joined	15:59
	→ gentooer joined	15:59
	→ Bruce_Wayne_ joined	15:59
	→ montylounge joined	16:00
	← gemmes left	16:00
	→ jmil joined	16:00
	→ jmil_ joined	16:01
	→ NET\|\|abuse joined	16:01
NET\|\|abuse	hi there folks. i was talking with a friend who's a web designer, he outsources developer work and project manages. He's a bit untechy so hadn't a clue about svn/bzr/git options.	16:01
	IslandUsurperAFK → IslandUsurper	16:01
	→ kukks joined	16:02
	← edude03 left	16:02
bnovc	is there a follow-up to that	16:02
NET\|\|abuse	i said i would think about and look into a way to setup a central multiproject server so he can have a hub where he version controls his projects, and allows remote dev's to work on his projects, while maintaining control of the overall system.	16:02
	→ yhager joined	16:03
	← tvw left	16:03
cbreak_work	git + gitolite sounds suitable.	16:03
NET\|\|abuse	i'm thining of various vcs web ui front ends, and i'm trying to think what he'd be suited to .	16:03
	gitolite.. hmm	16:03
	← saysjonathan left	16:04
imMute	NET\|\|abuse: what about github?	16:04
	→ icwiener joined	16:04
NET\|\|abuse	the other question was what to give him, could he handle git.	16:04
	← sraue left	16:04
	→ saysjonathan joined	16:04
NET\|\|abuse	imMute, hmm, there's a paid version isn't there.. how much is it?	16:04
	← jmil left	16:04
	← gentooer left	16:05
	→ linusoleander joined	16:05
	→ tj joined	16:05
NET\|\|abuse	is there a windows gui like tortoise for git?	16:05
cbreak_work	unfortunately	16:06
NET\|\|abuse	i thought i'd tried tortoisegit and it was buggy as all hell	16:06
cbreak_work	it's called tortoise	16:06
	and it's terrible	16:06
NET\|\|abuse	hmm,	16:06
	would love to promote adoption of the cli,, but not sure he'd accept and commit to that	16:06
	← jmil_ left	16:07
	← saschpe left	16:08
	→ sraue joined	16:08
	← saysjonathan left	16:08
	→ saysjonathan joined	16:09
	← flaguy48 left	16:11
	← hobodave_ left	16:11
theflow	Hi, I'm trying to understand the release management of git.git and if it makes sense to use something like that for a webapp. It seems a bit heavy for that, is there something in between what git.git is doing and everyone pushing directly to master?	16:12
imMute	TortoiseGIT is buggy as shite, but its still pretty new. TortoiseSVN is quite good, and I've had success teaching it to people enough to use it for VC	16:13
	← thiago_home left	16:13
SethRobertson	theflow: some people like http://nvie.com/posts/a-successful-git-branching-model It has an associated tool	16:13
	→ opalka joined	16:13
	← saysjonathan left	16:13
imMute	NET\|\|abuse: also, github's "small" plan is 10 private repos, and 5 private collaborators, and thats $12/mo	16:13
	→ thiago_home joined	16:13
	→ Dave^\| joined	16:14
	→ saysjonathan joined	16:14
SethRobertson	theflow: gitflow	16:14
NET\|\|abuse	imMute, i'm wondering which of git or svn is the best choice for this guy. he's really not the best at grasping more technical stuff..	16:15
theflow	SethRobertson: yeah, I looked at that, but it requires that you do everything in a topic branch, as far as I understand it	16:15
imMute	svn is probably going to be easier to grasp, although I say that because I used svn for much longer than git.	16:15
NET\|\|abuse	imMute, also, with his outsourcing, i'm doubting many of the php dev's in india he deals with would be setup for git use. they're svn as far as I can see.	16:15
imMute	well then svn is probably the better bet	16:15
	← noktoborus_ left	16:15
NET\|\|abuse	imMute, i've used svn for 5 years myslef, toyed with git and bzr on and off only last 2 years	16:16
SethRobertson	You mean if he switches to git, we get to infect the sub-continent with git?	16:16
NET\|\|abuse	SethRobertson, hehe yeh,	16:16
imMute	SethRobertson: lets not.	16:16
NET\|\|abuse	he deals with about 6 projects in a month, :)	16:16
	so he's likely to impact the working habits of about 8 indian devs	16:16
SethRobertson	NET\|\|abuse I do have to say that it is far easier to use git vs svn when talking about outsourced development and developers on a different continent	16:17
	← linusoleander left	16:17
SethRobertson	The central repository starts to break down seriously	16:17
NET\|\|abuse	SethRobertson, i do feel like there's a lot to argue for git..	16:17
	← xiangfu left	16:17
imMute	true, git is probably a better choice for dealing with devs across the world, but if they're already using svn and havnt touched git...	16:17
NET\|\|abuse	but tortoise not being ready, the imcumbent use of svn, kind of are a barrier	16:18
	→ amh345 joined	16:18
imMute	could certainly use some kind of git-svn bridge..	16:18
	← aah left	16:18
imMute	I dunno if there's anything that would present a git repository as an svn repo though...	16:18
NET\|\|abuse	yeh, not sure.	16:19
	→ psankar joined	16:19
	← psankar left	16:19
	→ psankar joined	16:19
SethRobertson	The git wiki talks about gitsafe as a (non-free) git graphical interface	16:19
	← saysjonathan left	16:19
NET\|\|abuse	how's trac as a git front end?	16:19
	→ pbeckingham joined	16:19
	→ saysjonathan joined	16:19
	→ Guest52674 joined	16:19
	← cbreak_work left	16:20
imMute	NET\|\|abuse: AFAIU, trac is more of a thing to combine VC, issue tracking, a wiki, and stuff all under one roof. it's hardly a client side git client	16:20
NET\|\|abuse	imMute, oh of course, i know.	16:20
	imMute, part of this would be to have a web ui with the bug tracking, per project tickets and per project wiki, so that he can become more organised in his own project management..	16:21
	the guys was a mural painter and graphic designer by trade before adopting the web industry :) He's not a trained pm, or particularly together guy, but he's a good sales guy :)	16:21
	And i just kind of want to help him.	16:22
	→ ntoll joined	16:22
	q0rban → q0rban\|lunch	16:24
	← saysjonathan left	16:24
	→ saysjonathan joined	16:25
	→ Tianon joined	16:26
	← Tianon left	16:26
	→ Tianon joined	16:26
	→ flaguy48 joined	16:26
	← Dave^\| left	16:26
	→ kostja_osipov joined	16:26
imMute	NET\|\|abuse: then trac is probably going to be a good suit	16:27
	→ amerine joined	16:27
	← Guest52674 left	16:28
	q0rban\|lunch → q0rban	16:28
	→ _sh3 joined	16:29
	← saysjonathan left	16:29
	→ Dave^\| joined	16:29
	→ jmil joined	16:30
	→ saysjonathan joined	16:30
	← kunwon1 left	16:30
	→ Heimidal joined	16:31
	← monokrome left	16:31
	→ reuteras joined	16:31
	→ ooooPsss_ joined	16:32
	← ntoll left	16:32
	→ agile joined	16:32
	← ooooPsss left	16:32
	ooooPsss_ → ooooPsss	16:32
	→ kunwon1 joined	16:33
	→ cylence joined	16:34
	← saysjonathan left	16:34
	→ saysjonathan joined	16:35
	→ urkud joined	16:36
	← khmarbaise left	16:37
	→ SimonP86 joined	16:37
	→ saschpe joined	16:37
	← giallu left	16:37
	← chrisshattuck left	16:38
	← ooooPsss left	16:38
	← Adaptee left	16:38
	→ zerium1 joined	16:38
	→ djh_ joined	16:38
	← djh_ left	16:38
	→ kanru joined	16:39
zerium1	having some difficulty figuring out how to delete a remote branch, git branch -r shows me a bunch of stuff outside of "origin" how would I go about deleting those?	16:39
	ie: there is a just a "trunk" (imported from SVN)	16:40
jast	delete it in the remote repository or delete your 'local remote branch'?	16:40
	← saysjonathan left	16:40
zerium1	in the remote	16:40
	ie git push origin :branchname	16:40
	works for anything under origin...	16:41
	→ saysjonathan joined	16:41
zerium1	but how would I do something outside of origin?	16:41
	→ ooooPsss joined	16:41
tango_	git push otherremote :branchname ?	16:41
jast	well, 'trunk' doesn't exist in any upstream git repository, does it? :)	16:41
zerium1	maybe you are right... however it shows up when i do a "git branch -r"	16:41
jast	well, you could just remove it from there :)	16:42
	if you no longer want to work on the SVN trunk, that is :)	16:42
zerium1	yeah i don't	16:42
	so that is only a "git svn" thing?	16:42
	leftover from the import i suppose?	16:42
	← ereslibre left	16:42
jast	yep, should be	16:42
zerium1	ok	16:42
	i should have just checked out the repo to a new directory	16:43
	→ thierryp joined	16:43
zerium1	so do i just "git svn -d branchname" ?	16:43
jast	no, git svn doesn't really give you any tools for that	16:44
	if a simple 'git branch -r -d trunk' doesn't work, try 'git update-ref -d refs/remotes/trunk' (not quite sure that's the name used internally, but I think so)	16:44
zerium1	yep that worked	16:44
	← koltroll left	16:44
jast	right :)	16:44
	← kanru left	16:45
zerium1	thanks	16:45
	→ codeslinger joined	16:45
	← saysjonathan left	16:45
	→ punkoff joined	16:45
	→ saysjonathan joined	16:46
codeslinger	I tried to do 'git commit foo.php' but the log message template it asks me to edit sounds like it is going to commit all other files that have changes as well. How do I just commit a few files that I am done with?	16:46
punkoff	How do I find closest unmergen branch for a given commit-ish?	16:46
	*unmerged	16:46
FauxFaux	codeslinger: That sounds right.	16:46
	← Tianon left	16:46
offby1	I'd just eyeball "gitk --all"	16:46
punkoff	e.g. if someone makes commits A,B,C in branch 'feature', it should be 'feature'	16:46
	and if 'feature' gets merged into master, it becomes master	16:47
codeslinger	I am a newbee for git, but have used cvs and svn heavily before	16:47
	← jmil left	16:47
codeslinger	I am now using git 1.5.5.6	16:47
zerium1	codeslinger: best $12 you'll spend is on the peepcode git tutorial	16:47
	→ hobodave joined	16:48
offby1	But I can buy two cheesesteaks for $12	16:48
jast	codeslinger: git commit -o <files>. and update your version of git. that one is practically ancient.	16:48
	→ ereslibre joined	16:48
	→ djh_ joined	16:48
	→ khmarbaise joined	16:48
	← djh_ left	16:48
	← ereslibre left	16:48
	→ ereslibre joined	16:48
codeslinger	jast: interesting. Are there specific arguments besides it being ancient that I could use to convince my admin to upgrade?	16:48
zerium1	offby1: your arteries will thank you for not though... ;)	16:48
codeslinger	And is there a risk to the repo by updating git.	16:49
	← sh1mmer left	16:49
	← ooooPsss left	16:49
	← khmarbaise left	16:49
jast	codeslinger: perhaps that that old version was substantially harder to use than the current ones. and no, git is pretty good about being backward-compatible.	16:49
	→ djh_ joined	16:49
	← zerium1 left	16:49
thm	there's no way to get a remote log other than fetching stuff to a local repo?	16:49
	← djh_ left	16:50
codeslinger	if I could qualify those statements I would convince them, i.e. what makes it easier for the newer version	16:50
jast	thm: that's correct	16:50
	← saysjonathan left	16:50
jast	codeslinger: there are ~149 KB worth of release notes since then that you're free to work your way through ;)	16:51
codeslinger	jast: lol	16:51
	→ saysjonathan joined	16:51
codeslinger	thanks	16:51
	→ tvw joined	16:51
jast	truth be told, it's hard for me to recall which things we were missing back then	16:51
	← Vampire0 left	16:51
jast	I only remember that a fair deal of the "hey, this makes things lots easier" moments were in 1.5.6 or later	16:52
	but to summarize from the bigger releases...	16:52
	← psoo left	16:52
	← fowlduck left	16:54
	→ LiamH joined	16:55
	→ Adaptee joined	16:55
	← saysjonathan left	16:55
	→ saysjonathan joined	16:56
	→ ooooPsss joined	16:56
jast	git init automatically initializes repositories properly on e.g. windows. you can view history in graph form. several segfaults fixed. better memory usage when decompressing deltas. tons of messages made more understandable. (cont'd)	16:56
	→ Vampire0 joined	16:56
	→ steph021_mac joined	16:57
	← steph021_mac left	16:57
	→ steph021_mac joined	16:57
FauxFaux	Surely I'm being dumb and there's a short version of $(git rev-parse HEAD).	16:58
codeslinger	jast: not using windows in 2010 :)	16:58
	→ TorstenB joined	16:58
codeslinger	this is the year of os-x and linux	16:58
	→ shiba_yu36 joined	16:58
FauxFaux	Hahahahha, very funny.	16:58
codeslinger	or really of iOS and android :)	16:59
bnovc	FauxFaux: what're you trying to do? you can probably just use HEAD	16:59
	→ Yuffster joined	16:59
	→ aspotashev joined	16:59
	→ DrNick2 joined	17:00
FauxFaux	bnovc: It's going into grafts (or any other 3rd party system needing the full id).	17:00
	← DrNick left	17:00
	← saysjonathan left	17:00
	→ saysjonathan joined	17:01
	→ rgr joined	17:02
	← shiba_yu36 left	17:02
bnovc	ah, well, then how would you get something shorteR?	17:02
	→ khmarbaise joined	17:02
	← khmarbaise left	17:02
bnovc	I mean you could cat .git/refs/heads/<branch> but thats not really 'shorter'	17:03
	→ koltroll joined	17:03
	srcerer_ → srcerer	17:03
	→ abetaha joined	17:04
jast	diffs and some other things are pagered by default on terminals. more robust packfiles. temporary data for am/rebase can no longer accidentally be added to a repository. per-line staging of changes. better at dealing with corrupt objects. branch --contains/--merged. easy mirroring. stash with index. various rebase -i fixes. (cont'd)	17:04
	← _sh3 left	17:05
	← cilly left	17:05
	← saysjonathan left	17:05
	← mabrand left	17:06
	← alex__c2022 left	17:06
	→ saysjonathan joined	17:06
	← eletuchy left	17:07
	← alberto56 left	17:09
	← koltroll left	17:09
	← normanrichards left	17:09
	← saysjonathan left	17:11
	→ kipras joined	17:11
	→ saysjonathan joined	17:11
fr0sty	jast: many thanks for your Gitbot homage.	17:11
	(I would have thanked you earlier but this is the firest I have seen of the non-automated you)	17:11
	faq non-bare	17:12
jast	fr0sty: Pushing to non-bare repositories is discouraged: please see https://git.wiki.kernel.org/index.php/GitFaq#non-bare [automatic message]	17:12
	fr0sty is disproportionately happy about that.	17:12
	→ nakkor joined	17:13
	→ letas joined	17:13
	← akamike left	17:13
	← Yuffster left	17:13
	→ Yuffster joined	17:14
	→ alberto56 joined	17:14
	→ aah joined	17:14
letas	hey I was wondering what kind of good stuff you have defined on your gitconfig, like shortcuts or colors	17:14
	→ _sh3 joined	17:14
	← punkoff left	17:15
	← saysjonathan left	17:16
	← amh345 left	17:16
fr0sty	color.ui=auto	17:16
	→ Tianon joined	17:16
	← Tianon left	17:16
	→ Tianon joined	17:16
nakkor	letas: I have an alias for log: lg = log --graph --pretty=format:'%Cred%h%Creset -%C(yellow)%d%Creset %s %Cgreen(%cr)%Creset' --abbrev-commit --date=relative	17:16
	← schambers left	17:17
	← Bruce_Wayne_ left	17:17
letas	nakkor nice!	17:17
	→ cilly joined	17:18
	→ amh345 joined	17:18
	← mastroDani left	17:18
	← ereslibre left	17:18
	← simosx left	17:18
	→ sebastorama joined	17:20
jast	easier syntax for setting up branch tracking. easier syntax for getting stages of files. option to prevent remote people from deleting branches via push. prevent adding submodule contents to supermodule. hard limit of 2 GB objects removed. checks for the common problematic case of pushing to a non-bare repository. automatic threaded packing. password prompts for user@ HTTP URLs. all kinds of performance optimizations. sparse checkout. HTTP digest auth. easier syntax f	17:20
	← chrislerum left	17:20
	← Bruce_Wayne left	17:21
	→ jmil joined	17:21
nakkor	is there a way to check how far ahead/behind a branch is from its remote without doing the full fetch?	17:22
	→ simosx joined	17:22
jast	hmm, the release notes seem to be missing a few things.	17:22
	← cylence left	17:22
	← smuf left	17:23
jast	nakkor: you can only list the current top commits of all refs of a remote, and that doesn't give you complete ahead/behind information	17:23
	← alberto56 left	17:23
jast	to fully determine 'aheadness'/'behindness' you have to match both sides' histories against each other, and for that you need the full history of both sides anyway	17:23
patrikf	nakkor: do you mean "fetch" or do you mean "pull"	17:24
sec^nd	hi	17:24
nakkor	patrikf: no because a pull would merge in the remotes changes	17:24
sec^nd	I have git repos and I'm having a problem pushing to them	17:24
patrikf	nakkor: there's also "git remote update", which is essentially a fetch for all branches and all remotes	17:24
	nakkor: so you don't even want fetch? why not?	17:24
sec^nd	I want to be able to work on one repo (clone) and push changes back and forth between reepos	17:25
	I have a local and remote computer, but I work on boht	17:25
nakkor	patrikf: mostly just curious, I would love to be able to see what I am missing on the server before I pull it down	17:25
sec^nd	*both	17:25
nakkor	epecially if the remote has a big changeset	17:25
	← lucasvo left	17:25
patrikf	nakkor: so you're all about traffic concerns	17:25
nakkor	patrikf: yeah	17:25
	→ d2dchat joined	17:26
sec^nd	I'm getting push errors	17:26
jast	sec^nd: push is problematic if both sides have an associated working tree. use pull from the other side instead. details at http://git.or.cz/gitwiki/GitFaq#non-bare	17:26
	→ lucasvo joined	17:26
patrikf	nakkor: well, you could do a git log on the remote side	17:26
nakkor	I was just wondering (and hoping) that there was a utility to see how many changesets are going to be fetched prior to actually transmitting the data	17:26
patrikf	nakkor: either through ssh, or through e.g. gitweb	17:26
	→ kanru joined	17:27
patrikf	nakkor: apart from that there's not much you can do	17:27
	← urkud left	17:27
nakkor	patrikf: cool, that is a strait forward answer I can go with :)	17:27
sec^nd	jast: what is the best way to sync between repos ?	17:27
jast	sec^nd: didn't I just say that? git pull	17:27
	or use a bare repository as a go-between if you really want to push	17:27
sec^nd	jast: I created a repo and git cloned it, and then I tried push	17:27
	jast: I want to be able to push it	17:28
	jast: how do I create a bare repo ?	17:28
jast	what's the big advantage of push over pull?	17:28
	git init --bare, git clone --bare	17:28
patrikf	jast: publicizing your own repo, for example.	17:28
sec^nd	jast: I dont have to log into the other server and enter my pass	17:28
patrikf	jast: or simple workflows involving 2 to 3 people	17:28
fr0sty	jast: you can't merge. it's a feature...	17:28
jast	patrikf: yes, but when you publicize the target is usually bare already	17:28
sec^nd	patrikf: that is what i'm trying todo	17:28
	← ayanes left	17:29
sec^nd	I keep having to log in and enter my password on a server I dont completly strust	17:29
	*trust	17:29
patrikf	sec^nd: yeah, you should only push to bare repos, and then pull from them.	17:29
sec^nd	kk	17:29
jast	sec^nd: well, you said you were working on both, so I assumed you were logged in on both anyway :)	17:29
	← montylounge left	17:29
letas	wow jast I just read what I think you have in your .gitconfig and I have no clue how to set most of them - Google mode on!	17:30
jast	letas: uh, what?	17:30
nakkor	jast: so you are looking at his config huh... kind of private stuff... ;)	17:30
	← lucasvo left	17:31
nakkor	x that, I mean you letas	17:31
	→ jarrod_ joined	17:31
	→ psoo joined	17:31
	→ brntbeer joined	17:31
jast	here is my .gitconfig file: http://git.pastebin.com/BFAu9YRi	17:32
	← brosner left	17:32
	→ ereslibre joined	17:33
	← jmil left	17:33
	→ unabonger joined	17:34
	DrNick2 → DrNick	17:34
	→ dreiss joined	17:35
kostja_osipov	hi, i'm trying to add a commit trigger that sends an email when a person commits changes to their local repository.	17:35
	As far as I can see .git/hooks files are not versioned.	17:35
	I wanted to add the trigger to the master branch to make sure the trigger is fired in all clones	17:36
	what's the proper way to accomplish that with git?	17:36
	where do I put my trigger so that's it's versioned and whoever clones a repo gets it?	17:36
	← theflow left	17:36
kostja_osipov	if you know that it's documented anywhere, just give me a hint, and i'll google it up.	17:37
helo	i did a bad cherry-pick that has now left a bunch of untracked files that are preventing me from checking out the branch i was cherry-picking from	17:38
	is there an easy way to blow those all away?	17:38
	→ justin-george joined	17:38
jast	kostja_osipov: active hooks are never automatically distributed to other repositories for security reasons. you can include a script in your repository that will set up the hooks for people who decide to run it.	17:38
fr0sty	helo: git reset --hard HEAD	17:38
jast	fr0sty: for untracked files? :)	17:39
mmattice	git status -s \| grep '^??' \| awk '{print $2}' \| xargs rm	17:39
jast	helo: "git clean" removes all untracked files	17:39
	← ph^ left	17:39
	← chrisred left	17:39
helo	hmmm.... i'm not sure that i want to do that	17:39
fr0sty	jast: good point, but git clean is a dangerous beast...	17:39
jast	I know :)	17:39
	→ sh1mmer joined	17:39
kostja_osipov	jast: thanks for your answer, but i think this is just nasty. both bk and bzr allow that	17:40
helo	i think i'll just keep rerunning git checkout <branch> and manually removing the files it complalins about	17:40
	→ warbee joined	17:40
jast	helo: git checkout -f to force overwriting files that are in the way	17:40
fr0sty	there is always checkout -f	17:40
	← dreiss left	17:40
kostja_osipov	i think someone is trying to be too smart and make a decision for me.	17:40
fr0sty	gah, beat me to it. 2nd place is the first loser...	17:40
helo	yeah, thats the ticket	17:40
jast	kostja_osipov: you're free to believe that. I personally believe that this design decision is not about you, it's about everyone you work with.	17:40
	← xiong left	17:41
bremner	kostja_osipov: you might be interested in gitolite, which can support global hooks	17:41
	→ kenichi joined	17:41
	← ekontsev1y left	17:41
kostja_osipov	jast: exactly. i need to make the decision for everyone i work wiht, and i have an authority to make it. but i don't have the tool support.	17:41
	lame	17:41
fr0sty	kostja_osipov: and to protect them from feeling like 'someone is trying to be too smart and make a decision for me.'	17:41
kostja_osipov	fr0sty: have you heard about BZR_NO_TRIGGERS switch?	17:41
fr0sty	kostja_osipov: yes you do, add it to a hooks directory and tell everyone to link that directory to .git/hooks/	17:42
kostja_osipov	which anyone can set in their shell to disable these hoosk?	17:42
fr0sty	if you have authority, use it.	17:42
jast	if you feel that strongly about it, why not use bk or bzr instead? this is one thing that I'm almost positive will never change in git.	17:42
kostja_osipov	i could just as well request everyone to manually run a diff and send an email before commit	17:42
	which i have an authority to do	17:42
fr0sty	and if your coworkers are disinclined to listen to you then you are hosed either way.	17:42
kostja_osipov	but this is not automation, which i am asking for	17:42
	← flaguy48 left	17:43
kostja_osipov	jast: right, i am personally strongly considering it. my 3d day with git and looks like someone is trying to be too smart.	17:43
fr0sty	kostja_osipov: you have been told how to automate this.	17:44
	it is not hard	17:44
kostja_osipov	always imposing some workflow on me, not helping me have it the way i want it	17:44
	fr0sty thanks a bunch	17:44
	← goshawk left	17:44
bmalee	git imposes a workflow? :/	17:44
kostja_osipov	thanks for telling me there is no other way, saves time	17:44
	→ keith_p joined	17:44
fr0sty	kostja_osipov: you can use whatever workflow you want	17:44
	← keith_p left	17:44
bremner	never argue with people who want to go use software you don't have to support	17:44
	→ matjas joined	17:45
	← bpo left	17:45
fr0sty	bremner: good point.	17:45
matjas	I `git clone`’d a repo on GitHub to which I don’t have write access and made some commits. Later, I forked the repo. How can I tell git to push my commits to MY fork and not to the original origin?	17:46
kostja_osipov	bremner: the issue is only that there is no perfect revision control system	17:46
	unfortunately	17:46
	in 2010	17:46
	fr0sty thinks there needs to be a 'git dealbrakers' webpage somewhere.	17:46
jast	there is never anything perfect, whether in 2010, 2020 or 5690	17:46
bremner	kostja_osipov: like I said, gitolite adds in that particular feature, but people will always be able to edit their local hooks	17:47
	← yairgo left	17:47
jast	not least because perfection is inherently subjective	17:47
	bremner: how does it do that? gitolite has no client-side part	17:47
	and I don't git letting itself getting swindled into fetching hooks from who knows where	17:47
	*see	17:47
	← guardian left	17:47
bremner	jast: hmm. good question. maybe it is default hooks in remote repos created by gitolite	17:48
jast	much more likely	17:48
bremner	which also avoids the local editing issue	17:49
kostja_osipov	jast: btw, if one is so concerned about security with hooks, one could just always fail to automerge them when pulling from upstream	17:49
jast	uh, what?	17:49
kostja_osipov	that would force users to look at the hook before letting it work in the local tree	17:49
jast	so users would be forced to look at the hook during clone?	17:49
kostja_osipov	yes.	17:50
jast	I don't see that mamking things more convenient	17:50
kostja_osipov	wait.	17:50
	not during clone	17:50
	during pull	17:50
[exa]	:]	17:50
jast	well, so during clone we should allow repositories to run arbitrary commands on people's machines?	17:50
	such as rm -rf ~	17:50
	← petran left	17:50
bremner	finally, a clean homedir	17:50
	← chrischambers left	17:50
jast	well, I guess it's up to personal preference :)	17:50
	I like my homedir messy	17:51
	← kanru left	17:51
matjas	I `git clone`’d a repo on GitHub to which I don’t have write access and made some commits. Later, I forked the repo. How can I tell git to push my commits to MY fork and not to the original origin?	17:51
[exa]	matjas: you need to mofify origin remote information. see git remote manpage	17:51
	like git-remote	17:51
sitaram	bremner: kostja_osipov wants client side hooks. Gitolite does nothing to the client and cannot. Which is good :) regardless of what kostja_osipov thinks	17:52
matjas	[exa] git remote, thanks	17:52
kostja_osipov	jast: ok. one other option would be to write a git plugin and ask all who wants these hooks to be run to install it.	17:52
[exa]	not sure how github handles forks, but setting 'origin' to your new push url should just work	17:53
kostja_osipov	that would be safe from security pov, don't y0	17:53
	ou think?	17:53
jast	hmm, I find no source that tells me how to get bzr to copy hooks from the server	17:53
sitaram	kostja_osipov: the correct place for hooks is on the receive side of a push. The local repo on the developers own machine is just that -- a local repo. He can do what he wants, but when he touches the central server, you "hook" him	17:53
	→ nevyn^_ joined	17:53
kostja_osipov	jast: you can use plugins in bzr	17:53
jast	yes, but you have to install the plugins yourself, don't you?	17:53
	→ goshawk joined	17:53
kostja_osipov	jast: right	17:53
	→ saysjonathan joined	17:53
kostja_osipov	can i do the same with git?	17:53
	→ Chillance joined	17:54
jast	it's not too difficult to install a git script that wraps around the real git binary and adds other operations on top	17:54
	I do that myself for a bunch of things	17:54
kostja_osipov	jast: see mysql_plugins project for a bzr plugin that runs local hooks	17:54
	→ koltroll joined	17:54
SethRobertson	So do I, and it even automatically installs git hooks for me!	17:54
kostja_osipov	ok, i see. messy though	17:54
	thanks!	17:54
	→ chrisshattuck joined	17:54
	→ petercoulton joined	17:55
	→ yairgo joined	17:55
jast	there were some efforts for adding global hooks quite a while ago, but they never really made it into anything	17:55
bremner	kostja_osipov: note that people could still call "the real git" to circumvent your script/plugin	17:55
kostja_osipov	bremner:	17:56
	that would be fine	17:56
	← sh1mmer left	17:56
kostja_osipov	i'm looking for automation, so that users of the project don't have to run a bunch of messy commands to participate	17:56
[exa]	kostja_osipov: I still wonder what practical purpose would that have, can you point me to some?	17:56
kostja_osipov	and the project workflow is around commits.	17:56
	← nonno left	17:56
kostja_osipov	one does a commit, a mail is sent to the list, this mail can be used for code review	17:57
jast	[exa]: well, in some teams developers are not supposed to be able to do experimental work without everyone seeing it ;o)	17:57
	← dpino left	17:57
[exa]	jast: oh I'm starting to see.	17:57
kostja_osipov	jast: you can disable the hook if you're shy and don't want others to see it.	17:58
jast	so instead of using something like gerrit and keeping local commits local, people will tend to stop committing experimental stuff, just like in subversion	17:58
kostja_osipov	but the deefault is that everyone sees what everyone else does	17:58
jast	yes, technically you can	17:58
	← elmob left	17:58
jast	but that means you're making it harder for people with good workflow to do the right thing... I do see what you're after, though.	17:58
kostja_osipov	jast: i can change the worflow so that a mail is sent only when htere is a push to a branch	17:58
[exa]	kostja_osipov: there's a way better way to do that, just make a script that keeps an eye on your HEAD and notifies everyone on changes. Anything else is absolutely ineffective	17:58
kostja_osipov	but every new branch we create will require manual setup anyway	17:59
jast	kostja_osipov: well, you can process pushes centrally on the server :)	17:59
	← epid left	17:59
kostja_osipov	exa: we have dozens of branches, and they are created and dropped easily.	17:59
	it's not about the head	17:59
	← tlonim left	17:59
	→ adamm joined	17:59
jast	[exa]: so people will get mails for amends, rebases, resets, switched branches, etc., too? ;)	17:59
kostja_osipov	yep	17:59
	not bad in a midsize project	17:59
bremner	that should be fun	17:59
kostja_osipov	you can filter all this	18:00
[exa]	that forces people to create a mail folder that they totally ignore	18:00
	← madewokherd left	18:00
fr0sty	bremner: just wait until the rebasing starts...	18:00
kostja_osipov	exa: if done wrong, yes	18:00
[exa]	kostja_osipov: much better way is to ask (or order, if you can) people to push all of their stuff often, and make a nice visualisation	18:01
jast	well, I have mentioned my thoughts about this. ultimately it's not our decision how kostja_osipov sets up his team workflows, and we don't really know whether it's not going to work wonderfully. I consider my work here done. :}	18:01
[exa]	eye candy makes wonders	18:01
bremner	kostja_osipov: I think working with pushes rather than commits is a better plan.	18:01
jast	kostja_osipov: anyway, for reviews, you might want to look at gerrit. it's a git review tool.	18:01
	→ sh1mmer joined	18:01
kostja_osipov	exa: for an example of this kind of workflow, please take a look at postgres project	18:02
	they do send a mail on every commit	18:02
	← albel727 left	18:02
	→ _Vi joined	18:03
	→ albel727 joined	18:04
fr0sty	"The project has decided, more-or-less, to use Git like it was CVS"	18:04
	from the postgres git-migration page ^	18:04
helo	better than using CVS like it was git	18:04
jast	postgresql is a really special project as far as workflows go	18:04
	← psoo left	18:04
fr0sty	and that is one email per mainline, authoritative, central repository commit.	18:05
[exa]	:D	18:05
fr0sty	not one email for every checkpoint, amend and rebase step.	18:05
	→ joel_falcou joined	18:05
fr0sty	just finished products.	18:05
	← rgr left	18:05
	→ bpo joined	18:06
	helo tries to imagine a world where git is taught in first- or second-year CS programs	18:06
	→ cylence joined	18:07
	→ rgr joined	18:07
	← nevyn^_ left	18:07
	→ tlonim joined	18:07
	→ goldbergus joined	18:09
bremner	helo: I planned to, but ran out of time.	18:10
	← bpo left	18:10
	→ bpo joined	18:11
bremner	so I think that world is not as far away as you think	18:11
	→ p3rg2z joined	18:11
	← NET\|\|abuse left	18:12
	← Schmallon left	18:12
	← mmc left	18:13
helo	seems like it would reinforce a lot of good development considerations	18:13
	← albel727 left	18:14
	← itamarjp1 left	18:16
	← chrisshattuck left	18:16
	← tizbac left	18:16
	→ tizbac_ joined	18:16
	→ itamarjp joined	18:16
	← codeslinger left	18:16
	← letas left	18:19
	→ chrisshattuck joined	18:19
	← itamarjp left	18:21
	→ dreiss joined	18:21
sgronblo	fr0sty: but wasn't that just to get CVS-devs slowly used to git?	18:23
	← tatsuya left	18:23
sitaram	bremner: you and Mpenz should get together; he's also a prof :)	18:23
	shake the world of CS with git!	18:24
fr0sty	sgronblo: exactly. it is nowhere close to a 'best practice' approach.	18:24
	→ __name__ joined	18:24
	← __name__ left	18:24
	→ __name__ joined	18:24
sgronblo	fr0sty: yeah but after a while when all the oldies get used to it, don't you think they'll change that?	18:25
fr0sty	without a doublt.	18:25
	doubt, even.	18:25
	from the article: "It's possible, even likely, that eventually the PostgreSQL project will move towards the "normal Git workflow" where branches and merges would be used for feature work. But it definitely won't happen this year. "	18:26
	→ brosner joined	18:26
sgronblo	I was using git like a dummy for at least a year too before really starting to learn how you can use it	18:27
	→ albel727 joined	18:27
kostja_osipov	fr0sty, jast: i'm reading up more. It seems what I am really trying to achieve, which is better peer collaboration with code reviews, is done in git using Signed-off mechanism.	18:27
jast	signoffs don't really have any fixed meaning	18:27
	← tizbac_ left	18:28
	→ troyt joined	18:28
jast	in git they are used by contributors to certify that the code is legally clean	18:28
kostja_osipov	jast: basically I am looking for a controlled way to push into the master branch.	18:28
	→ ttelford joined	18:28
	→ theoros joined	18:28
	← theoros left	18:28
	→ theoros joined	18:28
kostja_osipov	total publicity of commits is one way to do it.	18:28
sgronblo	kostja_osipov: gitolite with branch level permissions?	18:28
	← NotreDev left	18:28
jast	kostja_osipov: that sounds like a job for an update or pre-receive hook on the server	18:28
	← yotaff left	18:28
	→ tizbac_ joined	18:28
	← p3rg2z left	18:28
kostja_osipov	jast: it's a hell to maintain.	18:29
jast	hence my suggestion of gerrit	18:29
	← goldbergus left	18:29
kostja_osipov	i want a flexible configuration when a team or group or any assignee can be responsible for a given patch	18:29
	git is too centralized.	18:29
sgronblo	lol?	18:29
jast	http://code.google.com/p/gerrit/wiki/Background	18:29
	→ Elfe joined	18:29
	← koltroll left	18:29
sgronblo	git is a dvcs = DISTRIBUTED version control system	18:30
kostja_osipov	Google developed Mondrian, a Perforce based code review tool to facilitate peer-review of changes prior to submission to the central code repository	18:30
	sgronblo: you can read the above	18:30
	central code repository	18:30
	in git there is also central code repository.	18:30
jast	not necessarily	18:30
zombor	no there isn't	18:30
sgronblo	nope	18:30
kostja_osipov	not necessarily, but in any real life project there is.	18:30
zombor	only by assignment, not by definition	18:30
sgronblo	by convention there usually is	18:30
jast	you can have several repositories, each of which is deemed authoritative for a different part of the project, for example	18:30
zombor	kostja_osipov: there doesn't ahve to be	18:31
	linux doesn't use any central repository	18:31
kostja_osipov	I bloody know i can have many repository	18:31
	ies	18:31
zombor	the "central repsository" is Linus's repo	18:31
	← Elfe_ left	18:31
kostja_osipov	zombor: exactly	18:31
	← goshawk left	18:31
	→ acs joined	18:31
kostja_osipov	most projects have the mainline	18:31
zombor	but it's not a central repository	18:31
	by definition	18:31
	it's his repo	18:31
kostja_osipov	and git seems to have been built around a single person controlling the mainline	18:31
zombor	not at all	18:31
	← matjas left	18:31
sgronblo	yeah thats not how i see it	18:32
PerlJam	kostja_osipov: I don't see how you can come to that conclusion	18:32
zombor	at our company , many employees control the mainline	18:32
jast	me neither	18:32
sgronblo	its just one possible workflow	18:32
	ChanServ set mode: +v	18:32
zombor	we all have push access to the "central repository"	18:32
acs	test	18:32
PerlJam	kostja_osipov: or perhaps you mean "multiple single people, each controlling their own mainline"	18:32
sitaram	kostja_osipov: it's an administrative decision in git, ont a technical one	18:32
PerlJam	?	18:32
sitaram	not*	18:32
sgronblo	anyway, if git is centralized by your definition, I don't think you'll find any "de-centralized" VCS anywhere	18:32
zombor	heh	18:33
	← hyperair left	18:33
	→ goldbergus joined	18:33
	← mtkd_ left	18:33
kostja_osipov	sgronblo: it's not centralized, just read the exact link I got: http://code.google.com/p/gerrit/wiki/Background it's saying the same i am trying to say	18:34
zombor	i wouldt even know what "de-centraled" would mean :)	18:34
fr0sty	kostja_osipov: if you want people to review each other's code it is a simple as 'git fetch <url> <refspec>; pand then eek at FETCH_HEAD'	18:34
sgronblo	kostja_osipov: that's a link to gerrit, not git	18:34
zombor	kostja_osipov: it says "assumed"	18:34
fr0sty	*and then peek at	18:34
zombor	thats a workflow decision, not a technical one	18:34
jast	gerrit enhances one particular workflow you can use with git	18:35
	→ phantomcircuit joined	18:35
jast	in any case, I'd say that with more than, say, three contributors, it gets combinatorially infeasible to not have a single mainline	18:36
	git allows you not to have one, but I'm not sure you'll still be sane a few months down the road	18:37
kostja_osipov	hm... damn, if I use stuff like gerrit there is no post-moderation	18:37
jast	post-moderation?	18:37
kostja_osipov	and in most cases a post-push OK is just as fine..	18:37
	jast: some patches do need to be formally approved before a push to the mainline	18:37
	zombor doesnt see the problem	18:37
kostja_osipov	others just can be pushed without an approval, but an email needs to go out to the engineer so that he can see what was pushed.	18:37
	← phantomcircuit left	18:38
jast	as far as I know, gerrit allows you to have things pushed to a side branch and have them merged to mainline when they get approved	18:38
	but I haven't used it myself	18:38
	← flupke_ left	18:38
sitaram	jast: that is how it works; the "side" is spelled "for/" or something that's all	18:40
	like "commits-for/master"	18:40
	← abetaha left	18:40
	→ alvaro_o joined	18:40
sgronblo	so it's like branch-level permissions with an e-mail notification system?	18:41
	← mxmdrone_ left	18:42
	← shruggar left	18:42
jast	no. it's a review system.	18:42
	see it in action here: https://review.source.android.com//#q,status:open,n,z	18:42
	← MikhX left	18:43
sgronblo	ok so it also has a web app for reviewing, accepting and commenting on patches?	18:43
	← jds left	18:43
sitaram	yes	18:44
	→ abetaha joined	18:44
sitaram	the whole workflow is web based, if I understand right	18:44
	← alvaro_o left	18:45
	→ alvaro_o joined	18:45
	← werdan7 left	18:46
	→ mmc joined	18:48
	← wuj left	18:50
	← t0rc left	18:52
	← jrmuizel left	18:53
	→ jrmuizel joined	18:53
	→ MikhX joined	18:53
	← aspotashev left	18:53
	→ davidfetter_vmw joined	18:53
	→ lhz joined	18:54
	→ pantsman joined	18:55
	← yt_ror left	18:55
	← nullvariable left	18:58
	← goldbergus left	19:01
	→ bronson joined	19:01
	← bitkiller left	19:02
	← nakkor left	19:02
	→ bitkiller joined	19:03
	← curtana left	19:04
	→ goldbergus joined	19:06
	→ burhan joined	19:06
	→ mxmdrone joined	19:06
	→ alberto56 joined	19:08
	← mxmdrone left	19:08
	→ matjas joined	19:09
	→ ph^ joined	19:09
	← ereslibre left	19:16
	← doener left	19:16
	→ flaguy48 joined	19:18
	← rgr left	19:20
	← psankar left	19:21
	→ nakkor joined	19:21
	→ sako joined	19:22
sako	how do i compare the HEAD version of a file to an older commit?	19:23
	im trying git diff HEAD commitSHA file/name	19:23
IslandUsurper	git dif <commit>..HEAD -- file/name	19:23
sako	thanks	19:24
IslandUsurper	though...I guess I'm surprised your version didn't work	19:24
	→ xtagon joined	19:24
sako	i tried that and its comparing to /dev/null	19:25
	---- or ++++ are giving me /dev/null?/?	19:25
	something is not right :/	19:25
	← nakkor left	19:25
thiago_home	did the file exist in both commits?	19:25
sako	yea	19:26
	i found it via gitk	19:26
IslandUsurper	named the same in both?	19:26
sako	ya, hasnt changed hmm	19:26
	ok nvm	19:27
	it was named different	19:27
	sorry im helping another developer... just double checked	19:27
	how would i handle the name change?	19:27
	→ normanrichards joined	19:27
albel727	sako: not sure, but try "git diff HEAD:<filename1> <commitSHA>:<filename2>"	19:27
sako	file/name1..file/name2?	19:27
	oh i see	19:27
	→ scompt joined	19:28
	→ hobodave_ joined	19:28
IslandUsurper	if that doesn't work, try -M	19:29
	→ fedesilva joined	19:29
albel727	sako: or rather, the other way round =) i.e <commitSHA>:.. first, HEAD:... - second.	19:29
	→ foocraft joined	19:30
sako	that worked :)	19:30
	thanks alot	19:30
	← hobodave left	19:30
albel727	sako: yw =)	19:31
scompt	how do I create a copy of repository B inside of repository A?	19:31
	→ psoo joined	19:31
albel727	scompt: man git-submodule	19:31
jast	scompt: the 'git-submodule' manpage can be found at http://git.or.cz/man/git-submodule [automatic message]	19:31
scompt	I don't want to use clone because then the remote is still repo B	19:31
	I don't want to use submodules because then the remote is still repo B, right?	19:31
	repo B is a 'skeleton' of a project that I want to make a part of repo A	19:32
	→ fedesilva_ joined	19:32
scompt	something 'svn export repoB' and then adding all of the exported files to repo A	19:32
	← hobodave_ left	19:33
	→ hobodave joined	19:33
albel727	uhm, either I don't get your "remote is still repo B" line, or I don't see, what can be your problem with that. if you want just a copy, then make a copy.	19:33
	→ shennyg joined	19:34
scompt	albel727: I think I just want to make a copy, is there a way to do that with git? or should I clone it and then delete the .git?	19:34
	← ricky left	19:35
	← fedesilva left	19:35
	fedesilva_ → fedesilva	19:35
albel727	scompt: well, git doesn't have a very convenient export unfortunately. you can either make a shallow clone (git clone --depth=1 <url>), or use "git archive --remote=<url> master", to get a tar file, which you'll then need to untar (there's an option for zip, iirc).	19:36
scompt	albel727: k, thanks, I'll give that a shot	19:37
albel727	scompt: "git archive --remote=<url> --output=<dest.tar> master", to be precise. "git archive --format=zip --remote=<url> --output=<dest.zip> master", for zip format.	19:38
	→ Rickardo1 joined	19:38
albel727	you're welcome.	19:38
	→ napster_123 joined	19:38
	→ letas joined	19:38
	← napster_123 left	19:38
	← scompt left	19:39
	→ nullvariable joined	19:40
	← tlonim left	19:42
	← psoo left	19:44
	→ EvanR-work joined	19:45
EvanR-work	is there a way to delete commits from the history of your branch?	19:46
	→ andyburke joined	19:46
	← marcob left	19:46
albel727	EvanR-work: you mean, besides git rebase?	19:46
EvanR-work	how would you use rebase to accomplish this?	19:46
	← AAA_awright left	19:46
	← ph^ left	19:47
albel727	EvanR-work: to delete a commit from branch: "git rebase --onto <commit>~1 <commit> <branch>"	19:47
	→ ph^ joined	19:47
EvanR-work	so far only came up with branch at the delete point and cherry pick the rest one by one	19:47
	ok, checking that comman	19:48
albel727	EvanR-work: you need that on a big scale? why not git rebase --interactive, then?	19:48
EvanR-work	interactive lets you squash	19:48
	but also delete?	19:48
albel727	EvanR-work: ...and delete altogether	19:48
	yep	19:48
EvanR-work	ah	19:48
albel727	just remove a line with commit	19:48
EvanR-work	great	19:49
albel727	EvanR-work: but remember, that rebasing this way is dangerous. you not only remove a commit, you remove whatever changes it did introduce.	19:49
	rayners → rayners_afk	19:49
albel727	EvanR-work: if you want to remove a commit, but retain the changes it introduces, squash is the way. filter-branch is the another way.	19:50
	→ Mannequin1 joined	19:50
	→ gshank joined	19:50
EvanR-work	what exactly would be the command?	19:50
albel727	EvanR-work: for?	19:50
EvanR-work	interactive	19:50
	im not rebasing onto anything	19:51
	← Mannequin left	19:51
albel727	EvanR-work: basically, just "gir rebase --interactive <oldest commit you want to delete>~1". and then remove the lines for commits you want to delete.	19:51
	EvanR-work: that works for current branch, ofc	19:52
EvanR-work	ok	19:52
	whats ~1 ?	19:52
albel727	s/gir/git/	19:52
	EvanR-work: one commit back in history.	19:52
SethRobertson	EvanR-work: man git-parse	19:52
EvanR-work	thats not ^1 ?	19:52
SethRobertson	jast-bot?	19:53
EvanR-work	man git-rev-parse	19:53
jast	EvanR-work: the 'git-rev-parse' manpage can be found at http://git.or.cz/man/git-rev-parse [automatic message]	19:53
SethRobertson	Ah yes, thanks	19:53
parasti	commit^n syntax specifies parents, commit~n syntax specifies ancestors	19:53
albel727	EvanR-work: well, in this case it coincides with ~1, but generally - no. for instance, ^2 is not the same as ~2. ^2 - is the second parent. ~2 is the second older commit. the man says the rest.	19:54
EvanR-work	oh so its time based rather than parent based?	19:54
albel727	EvanR-work: yeah	19:54
EvanR-work	that sounds bad	19:54
parasti	no	19:54
albel727	?	19:54
EvanR-work	well ill tell him this command and then	19:55
	understand it	19:55
albel727	EvanR-work: ah. no, not in the sense you imagine.	19:55
	EvanR-work: what I meant to say, is that ~2 is the first parent of the first parent.	19:55
	← ph^ left	19:55
jast	whoops, small bug in the script. fixed.	19:55
albel727	EvanR-work: no time involved whatsoever.	19:55
	→ ph^ joined	19:55
EvanR-work	oh ok	19:56
	← spaceonline left	19:56
EvanR-work	^ lets you specify which parent	19:56
	if theres more than one	19:56
albel727	EvanR-work: yeah.	19:56
SethRobertson	True only for merges, and your intuition on which parent is which may be wrong in the case of fast-forwarding	19:56
albel727	since when we have several parents after fast-forwarding? =)	19:58
	→ carutsu joined	19:58
albel727	a single parent doesn't require much intuition, does it? =)	19:58
EvanR-work	its possible to fast forward and have multiparents	19:58
	since the latest commit could be a merge	19:58
albel727	haha. you got me.	19:59
	← petercoulton left	19:59
SethRobertson	Due to fast-forward merges you can have situations where you have multiple branches (A and B, eg) and while you "know" that you branched B from A and later merged it back in, git believes that the both branches A&B were on the human-knowledge-A branch of development, or vis-versa, both A&B were on the human-knowledge-B branch of development. It is very annoying	20:00
	→ petercoulton joined	20:00
SethRobertson	I have a recipe which will recreate this annoying situation. If you disable ff-merges it will never happen.	20:01
	← saschpe left	20:01
SethRobertson	I wish I could disable fast-forward manually initiated merges (e.g. always have -no-ff on the `git merge` command). Obviously I could create an alias, but...	20:02
	→ pedrobelo joined	20:03
albel727	SethRobertson: hmm, there might be a config key for that... let me look.	20:03
SethRobertson	I'd be happy to learn that. I've looked (well, searched for "fast")	20:03
	← ph^ left	20:04
	← nullvariable left	20:05
	← fedesilva left	20:05
	← _Vi left	20:05
	→ ph^ joined	20:05
	→ _Vi joined	20:06
	→ nullvariable joined	20:06
albel727	SethRobertson: well, at least there is "branch.<name>.mergeoptions" which allows to specify default options for merges into specified branch. let me look some more, to see if there is a global setting...	20:07
mnemoc	hi, I have 4 repositories imported from CVS making A A/B A/C A/D which I turned into a single repository using git-filter-branch and then pulling. how can I now reorder this super repository by date?	20:08
	→ bhowmisu joined	20:09
	← babusri left	20:09
SethRobertson	`gitk --date-order`?	20:09
mnemoc	SethRobertson: in the history	20:09
	→ Evious joined	20:10
SethRobertson	That shows you your history in date order	20:10
	← d2dchat left	20:10
SethRobertson	Perhaps I am missing something	20:10
mnemoc	yes, but i don't want to be shown in order, i want it to _be_ in order :)	20:10
	i don't see something like --date-order in git-filter-branch help	20:11
	→ darkredandyellow joined	20:12
albel727	mnemoc: first I want you to confirm, that gitk --date-order, indeed shows commits in the desired order.	20:12
	← bhowmisu left	20:12
	→ jelmer joined	20:12
mnemoc	ow... i got the merge as single commit :'(	20:13
albel727	SethRobertson: I'm afraid, that there are no global settings for git pull/merge. sorry =/	20:13
	jelmer → Guest67633	20:13
SethRobertson	albel727: Thanks for looking	20:13
	→ WebDragon joined	20:13
	← ph^ left	20:14
	→ detrate` joined	20:15
detrate`	can I undo a merge?	20:15
	→ ph^ joined	20:15
SethRobertson	detrate`: have you pushed?	20:15
	← brntbeer left	20:15
	→ Eaven joined	20:15
detrate`	no	20:15
	← brosner left	20:16
	→ nakkor joined	20:16
	← ph^ left	20:16
SethRobertson	detrate`: Then yes. Step one, take a backup of your repo (e.g. git clone or whatever). Step two: `git reset ORIG_HEAD`. See `man git-reflog` if you have done merges since then	20:16
jast	detrate`: the 'git-reflog' manpage can be found at http://git.or.cz/man/git-reflog [automatic message]	20:16
albel727	mnemoc: huh? sorry I don't get your last remark about single commit. what's the problem with having a merge as a single commit? (and how do you imagine a merge consisting from several commits?)	20:17
	→ ph^ joined	20:17
cirwin	SethRobertson: why do you advise taking a backup at that point? (just curious)	20:17
albel727	s/consisting from/consisting of/	20:17
	← bentob0x left	20:17
mnemoc	albel727: there is zero conflict, I need it as I would cherry-pick every commit from the other respositories	20:17
detrate`	okey thank you SethRobertson and jast	20:18
mnemoc	albel727: instead of a "merge"	20:18
SethRobertson	cirwin: I always recommend newbees to take backups when running `git reset`	20:18
	→ schambers joined	20:18
cirwin	fair enough	20:18
SethRobertson	and filter-branch	20:19
albel727	mnemoc: hmm, that was my another guess. let me think...	20:19
SethRobertson	isn't that a job for rebase?	20:20
	← MattDiPasquale left	20:20
SethRobertson	Unless you need the donor branch to be unmolested	20:20
mnemoc	SethRobertson: to me?	20:20
SethRobertson	and/or albel727	20:21
albel727	SethRobertson: hmm, that is actually a good idea! if rebase understands inserting commit lines, that is.	20:22
mnemoc	the original repository was made of 4 "virtual modules" ... I want the resulting git to reflect the work that was done at the same time in the different modules	20:22
	the interim git repositories are irrelevant, all with get trashed	20:22
albel727	ok, I'm testing the git rebase approach. give me some time....	20:23
SethRobertson	I wonder if sending it through `git-fast-export` and `git-fast-import` would do something useful	20:23
	Perhaps with a sort in the pipe	20:24
	→ khmarbaise joined	20:24
	← TorstenB left	20:24
mnemoc	SethRobertson: that was my initial plan... but failed to get them working	20:25
albel727	SethRobertson: I don't think so. unless you know an easy way to take 4 fast-streams and reorder commits in them according to date order.	20:25
mnemoc	it can be done in two steps	20:25
	first taking all commit from the other repositories, and later reorder the world by commit date	20:25
SethRobertson	He already has a unified repo, no? He just needs to `git fast-export` and then break it up into commit objects, sort, and then reassemble.	20:25
	← andyburke left	20:26
mnemoc	no, I de-unified it	20:26
	because I got "merges"	20:26
	← hobodave left	20:26
	→ MattDiPasquale joined	20:26
	→ ceez joined	20:26
SethRobertson	Personally I just left my CVS repos as separate git repos and have a meta-repo-manager script, but it does seem like this would be a useful feature.	20:27
	→ hobodave joined	20:27
albel727	mnemoc: ok, a question. is what you have now a single merge (or maybe 3 of them) on top of your history, that unifies 4 development lines?	20:28
mnemoc	had that until git reset HEAD^3; git reset --hard	20:28
albel727	I see.	20:29
	← mmc left	20:29
	→ andyburke joined	20:29
mnemoc	so I have 4 repositories, and the content of the 3 sub repositories moved into a subdirectory	20:29
	← khmarbaise left	20:29
mnemoc	A/B/.git -> A/B/{.git,B/}	20:29
	→ johnw joined	20:29
	← johnw left	20:30
	→ rryouumaa joined	20:30
albel727	mnemoc: do files in those subrepos, reside in respective subfolders (e.g A/B/readme.txt)?	20:30
	→ Torsten joined	20:30
mnemoc	SethRobertson: tried keeping submodules, but the very first time a coworker used it got a file from A/B commited into A messing up the world	20:30
	→ ereslibre joined	20:31
mnemoc	albel727: yes, they are ready to "collapse" into a single repo	20:31
albel727	mnemoc: and they were ready to do so from the very start of history? or did you just do git mv at the last moment?	20:32
mnemoc	albel727: since day one different cvs submodules, no conflict at all	20:32
	→ chrischambers joined	20:32
	Guest91296 → Whtiger	20:33
	→ khmarbaise joined	20:33
	→ ereslibre_laptop joined	20:33
mnemoc	A has no reference to B,C or D in it's history. all hidden by cvs	20:33
	← ereslibre left	20:33
SethRobertson	mnemoc: I don't use submodules, I have a different script. However, if you can get a unified repo (I cannot) that is the superior solution	20:33
	ereslibre_laptop → ereslibre	20:33
	→ galderz joined	20:33
albel727	SethRobertson: ok, after asking my questions, I conclude, that your approach might indeed work. so, do you have an easy way to sort commits by date in a fast-import stream?	20:34
	→ hasrb joined	20:35
mnemoc	git fast-export ..master \| wc -l	20:35
	3	20:35
	→ rolando joined	20:35
mnemoc	^---- snif snif snif	20:35
	← Evious left	20:35
albel727	SethRobertson: alternatively, I suggest using git rebase --interactive with output, generated by git log --oneline	20:35
	→ yt_ror joined	20:36
mnemoc	albel727: that's indeed a nice idea after I get the repositories "collapsed" into one (without merges)	20:36
albel727	"git log --oneline --date-order", to be precise. and we need to get each line prefixed with "pick "	20:36
mnemoc	sed s,^,pick ,	20:37
SethRobertson	albel727/mnemoc: I was playing with multi-repos and git-fast-*port and now understand the scope of the problem. It isn't just that these need to be date ordered, you actually need to synthesize a combined tree on each commit.	20:37
mnemoc	sed "s,^,pick ,"	20:37
	→ simosx_ joined	20:37
	← yt_ror left	20:37
	→ Evious joined	20:37
	← sonnym left	20:38
	← gusnan left	20:38
albel727	mnemoc: thing is, you actually need to run that rebase of mine on a repo with merges, and hopefully, it will provide a single branch without them. but that might take one hell of a time.	20:38
	← simosx left	20:38
mnemoc	uhm	20:39
	i'll try adding a sub repositories as remotes and cherry-pick in a loop.... maybe i'm lucky	20:39
SethRobertson	Do you have multiple branches?	20:40
mnemoc	no, only one relevant	20:40
	i can tag manually after finishing the dance	20:41
SethRobertson	That is very good for you	20:41
mnemoc	it's good to have poor development practices sometime :p	20:41
	→ johnw joined	20:42
johnw	sitaram: ping	20:42
	→ programble joined	20:43
	→ dl9pf_ joined	20:43
	← dl9pf left	20:43
ScottO_	what is the procedure for updating a feature branch to the master?	20:44
helo	i used a series of 'git mv ... ...' to move some files around, and then stashed the changes, and applied... now git status shows all of the destination files as 'new', instead of renamed from their previous location. will this be corrected when i actually commit?	20:44
albel727	SethRobertson: mnemoc: well, actually we can also use a combination of git filter-branch --tree-filter, with "git checkout -M 'older revision on another branch'" and "git-rev-list -1 --before GIT_AUTHOR_DATE". or some presumably faster alternative of --index-filter and git-read-tree -M...	20:44
ScottO_	helo: i think it does that when you add them before the commit	20:44
	→ stringoO joined	20:45
ScottO_	helo: when it notices that the file has same file has been removed from one location and added to another	20:45
	← sh1mmer left	20:45
SethRobertson	albel727: mnemoc: If someone gets a script/recipe working, they should post it to the git mailing list	20:45
albel727	SethRobertson: understood.	20:46
	→ dharrigan joined	20:46
	← dharrigan left	20:46
	→ dharrigan joined	20:46
SethRobertson	ScottO_: it depends. you can `git rebase master` or `git merge master` depending how you feel. Other people just like to leave it stale and resolve the problems on merge-back	20:46
ScottO_	well the branch has some changes that arent in master	20:47
	should merging master into the feature branch be good enough?	20:47
	← goldbergus left	20:47
ScottO_	SethRobertson: ok thanks, i guess ill merge master back into the feature branch	20:47
SethRobertson	ScottO_: Yes. As long as you `git checkout feature` you can `git rebase master` or `git merge master` without problem	20:47
	→ slide joined	20:48
SethRobertson	I'd probably rebase, but that is just me.	20:48
ScottO_	rebase will make the commits on the branch get lost right?	20:48
mnemoc	albel727: your elegant proposal is far beyond my skills to implement :(	20:48
	ScottO_: rebase puts your commits on top of the other	20:48
ScottO_	ah so rebase grabs master and then reapplies the commits on top of that	20:49
mnemoc	yes	20:49
	simosx_ → simosx	20:49
ScottO_	cool tat might be better in the long run	20:49
	acs loves git grep <options> file/s	20:49
	→ petran joined	20:50
	→ FabianB__ joined	20:50
	← rolando left	20:51
	← FabianB_ left	20:51
sitaram	johnw: pong; but it's 2:20am, and my night-bird days are long gone. If it's a quick question, fine	20:51
	mnemoc loves git's power to mangle history :)	20:51
ScottO_	is there any automated way to get tag numbers isnerted into every file as a comment?	20:51
SethRobertson	ScottO_ What is a "tag number"?	20:51
ScottO_	like for version numbers	20:51
	which i assume is usualy done with tags	20:52
	→ imMute joined	20:52
SethRobertson	ScottO_: man gitattributes and look for $Id$	20:52
jast	SethRobertson: the 'gitattributes' manpage can be found at http://git.or.cz/man/gitattributes [automatic message]	20:52
	→ goldbergus joined	20:53
SethRobertson	Also smudge and clean	20:53
	Generally people don't bother with git.	20:53
	→ kylehayes__ joined	20:53
ScottO_	why not	20:54
SethRobertson	Because git's SHA checksum of every file makes file identification pretty easy.	20:54
	→ dark joined	20:54
	← chrisshattuck left	20:54
	→ peterjmag joined	20:54
ScottO_	right now i have no clue which commit i last updated the production server to	20:55
	i could find out with some effort, but might be an opportunity to change the way things are done to make it easier	20:55
dark	I was in a feature branch, and I needed to work on another branch. I did git stash, git checkout anotherbranch, and then did the work. I then did git stash pop to go back, but it applied my stash on top of anotherbranch, not the branch I were when I did git stash (I thought it would first do git checkout ..)	20:55
	how to revert the stash, then go to the another branch and apply it?..	20:55
	← kylehayes left	20:56
	kylehayes__ → kylehayes	20:56
	kylehayes → 30BAAGZB3	20:56
dark	simply git stash, git checkout previousbranch, git stash pop?	20:56
	→ aaron01 joined	20:56
peterjmag	Hello all! I'm looking to set up colored git output on a coworker's OSX 10.5 machine. I have it on mine, but I believe it was automatically installed by the "repo" script when I was messing around with the android source code (following instructions at http://source.android.com/source/git-repo.html ). How can I set up colored git output manually? Rather, what does repo do to git to enable colored output in Terminal.app?	20:56
SethRobertson	ScottO_: People often have the server with a .git directory on it which you can directly query to find the current location	20:56
ScottO_	yeah maybe i need to have git installed on the roduction server	20:57
	but my git repo isnt accessible to the production server	20:57
aaron01	How do I update refs/remotes/origin/master. I tried to fetch origin master, but the ref isn't being updated?	20:57
cbreak	peterjmag: man git-config	20:58
jast	peterjmag: the 'git-config' manpage can be found at http://git.or.cz/man/git-config [automatic message]	20:58
SethRobertson	peterjmag: man git-config Search for color.ui and friends	20:58
jast	peterjmag: the 'git-config' manpage can be found at http://git.or.cz/man/git-config [automatic message]	20:58
cbreak	there are a few color variables	20:58
	also, take a look at gitx	20:58
SethRobertson	ScottO_: You can always blow the .git directory over there when you blow everything else	20:58
ScottO_	true	20:59
mnemoc	git log --oneline --date-order doesn't work :(	20:59
	← zomgbie left	20:59
peterjmag	cbreak: SethRobertson: ah, thanks! I was looking in the wrong config file, but now I can see the "[color] ui = auto" line	20:59
SethRobertson	ScottO_: Or just `git describe > .checkout-description` and send that one file	20:59
mnemoc	it seems to use commit date instead of the actual date	20:59
dark	actually git checkout previousbranch alone did the job (maybe because I just added an untracked file, and thus git stash was unnecessary?)	20:59
	but anyway i would like to know how to undo git stash pop..	21:00
	← goldbergus left	21:00
	→ basicxman joined	21:00
	← galderz left	21:01
	→ Aleks_Minton joined	21:02
	← Guest67633 left	21:02
	← Aleks_Minton left	21:02
	→ gusnan joined	21:05
	→ goldbergus joined	21:05
albel727	mnemoc: --yeah, --date-order probably uses commit timestamp. you can make the commit date to coincide with patch date by git filter-branch --environment-filter and something like "set GIT_COMMITTER_DATE=GIT_AUTHOR_DATE" for the actual filter.	21:06
mnemoc	oh, nice trick	21:07
dark	another question: i have two branches, stable and devel. i needed to add debug something on stable and wanted to port the result to devel. i commited, then got a patch with git format-patch, then deleted the commit with git reset --hard HEAD^. applied it on devel, then merged devel on stable. is this procedure reasonable?	21:07
	s/add debug/debug/	21:08
	→ drizzd_ joined	21:08
fr0sty	dark could have stashed the change and changed branches, git stash pop and then committed	21:09
	you could have cherry-picked the commit from stable to devel and reset stable's head	21:10
dark	but the stable server lives on a different account o.o after merge into stable, i needed to go to that account and do a git pull	21:10
fr0sty	you could have created a teporary branch , commited, and done a rebase --onto devel HEAD~	21:11
dark	the stable server does not hold devel branch at all	21:11
	← tvw left	21:12
aaron01	I'm trying to do the following: /opt/foo is a clone of a feature branch in /opt/main. I'm trying to check if there are changes in the master branch in /opt/main that are not contained in the feature branch in /opt/foo. So something like: git rev-list HEAD..refs/remotes/origin/master. What I'm not sure of is how to update the rev-list for refs/remotes/origin/master within /opt/foo. If that makes any sense...	21:12
	→ mmc joined	21:12
	← drizzd left	21:12
fr0sty	dark: where were you doing your work?	21:12
dark	stable	21:12
fr0sty	what server	21:13
	what user	21:13
dark	user ts, on a different computer	21:13
	in order to transmit some commit to devel, i sent a patch.. what i could do is to have the devel branch at the stable server, then do a git pull from the devel machine	21:13
fr0sty	if you are not on the same machine, and you won't add additional remotes and fetch other branches then passing around patches is pretty much your only option.	21:14
dark	but it seemed actually more awkward, not less	21:14
cirwin	aaron01: you want to fetch from /opt/main (I think)	21:14
	← Unknown[NF] left	21:14
	→ alpheus joined	21:14
	← peterjmag left	21:15
fr0sty	if you have the devel branch on your stable server so that you can commit to the devel branch on that machine you can push/pull to/from another machine without needing all this patch business.	21:15
ScottO_	any reason why this would be giving me all of the history instead of just since the commit?	21:15
	git log --name-only 41b117fbe4c8b00b464153c215116cb34205a076	21:15
albel727	mnemoc: anyway, what do you need this log for? you can achieve a log totally ordered by date in some other way, eg something like git log --format="%at %h %s" \| sort	21:15
	→ aisipos joined	21:15
bnovc	whats the most effective way to find the last tag on the branch that I'm on	21:15
slide	I have a repo where I need to add my own project files in its directory structure but Id like to keep my files seperate, is there any way to do that?	21:15
fr0sty	Scott0: what were you expecting to see?	21:15
aaron01	cirwin: doing a: git fetch origin master only updates FETCH_HEAD (with the correct sha1), and doing: fetch --update-head-ok origin master, doesn't update the ref either.	21:15
ScottO_	i swear id id this the other day correctly, but the history for that disappeared	21:15
	← yunosh left	21:15
fr0sty	bnovc: man git-describe	21:16
jast	bnovc: the 'git-describe' manpage can be found at http://git.or.cz/man/git-describe [automatic message]	21:16
ScottO_	fr0sty: list of files changed since commit	21:16
bnovc	fr0sty: but that gives other stuff too and doesn't have a switch just for the tag (or maybe it does? I'll re-read)	21:16
ScottO_	fr0sty: just the filenames themselves	21:16
fr0sty	slide: do you want them tracked or not?	21:16
	→ chrisshattuck joined	21:16
fr0sty	ScottO_: you need commitid.. if you want 'since' behavior	21:16
mnemoc	albel727: if i can fix the order once, it will just work everytime in future, without any wrapping. so it's a kind of "investment"	21:17
bnovc	ya, it seems to always suffix it	21:17
	I mean I could awk that part off but it seems nasty	21:17
ScottO_	fr0sty: commitid?	21:17
bnovc	or I guess sed because the tag could have a dash	21:17
slide	fr0sty, the other repo? I guess I could track the original repo, but how would I then import changes from it where my files wouldnt be?	21:17
fr0sty	Scotto_: do you want a list of files from each commit or just the endpoints.	21:17
ScottO_	endpoints?	21:17
albel727	mnemoc: this --environment-filter "fix" will disappear upon next filter-branch or cherry-pick anyway, so you might as well not bother yet.	21:18
	← __name__ left	21:18
bnovc	ah, --abbrev=0	21:18
fr0sty	ScottO_: commit and HEAD or every commit in between.	21:18
dark	but, the question was more like, is this reasonable, to expect all new code to come from devel, and make sure all commits on stable are just merges? if the code actually came when working on stable, one could make an exception and merge stable onto devel	21:18
ScottO_	i just need a list of all the fiels that had a modification and it could span multiple commits, but i dont want them to be seperated by commit	21:18
dark	but i don't know if the history would become too weird	21:18
fr0sty	dark: better to create feature branches off of stable, merge them to devel and then merge devel back into stable.	21:19
	less hacky.	21:19
cirwin	aaron01: I was under the impression that fetch gave you more than just an updated FETCH_HEAD	21:19
mnemoc	albel727: true, i'll have to do it again after getting rid of cvs	21:19
fr0sty	slide: I don't understand what you are trying to do.	21:19
dark	hmm. and them, transmit them to the devel machine using git fetch?	21:19
	then*	21:19
	→ notjohn joined	21:20
ScottO_	fr0sty: so i guess the short answer is every file modification going from commit to HEAD	21:20
albel727	mnemoc: well, just remember, that filter-branch allows specifying several filters simultaneously, so if you wont forget to add that --environment-filter to whatever else you're filtering for, everything will be ok.	21:21
	→ zomgbie joined	21:21
	→ WebDragon\|laptop joined	21:21
ScottO_	i swore it was git log --name-only but that spits out every file modification since the very first commit	21:21
	rather than starting at the commit provided	21:22
	← WebDragon left	21:22
	WebDragon\|laptop → WebDragon	21:22
fr0sty	ScottO_: you need the two dots	21:22
	are you thinking about git-whatchanged?	21:22
ScottO_	did that, it returns the right span of data, but groups them by commit	21:23
	maybe..	21:23
aaron01	cirwin: hmm. looks like explicitly specifying the ref is what was wrong. Simple 'git fetch' updated properly.	21:24
cirwin	interesting	21:25
ScottO_	fr0sty: but apprntly you cant use the commit hash with whatchanged	21:25
dark	aaron01, don't you need to have the ref on .git/config?	21:25
ScottO_	n/m	21:25
slide	fr0sty, I have a git repo that is owned and controlled by other people. My project files are intertwined in it. If I create my own repo with the original repo files and my new stuff can I then sync the original repo files without harming/removing my own?	21:25
	→ amackera_away joined	21:25
	amackera_away → amackera	21:25
ScottO_	but it still returns the same data as git log --name-only	21:25
fr0sty	slide: define 'sync'	21:25
slide	fr0sty, pull down their changes in my repo	21:26
dark	here [remote ".."] enables me to use git fetch ..	21:26
albel727	mnemoc: I'm afraid I lied. filter-branch will preserve dates, unless instructed specifically. so, it's only cherry-pick that might be a problem.	21:26
	→ blackxored joined	21:26
blackxored	I always forget this	21:26
	how to setup an existing branch to track a remote branch	21:26
fr0sty	ScottO_: git diff --name-only <commit>	21:27
ScottO_	fr0sty: maybe im thinking of git diff	21:27
	lol	21:27
	← shennyg left	21:27
fr0sty	slide	21:27
	→ rgr joined	21:27
ScottO_	:D thanks though	21:27
cbreak	Blackguard: man git-branch, search for -u	21:27
jast	Blackguard: the 'git-branch' manpage can be found at http://git.or.cz/man/git-branch [automatic message]	21:27
cbreak	sorry, blackxored	21:27
fr0sty	slide: just create a private branch and do you work there ane then merge upstream changes in, or rebase your changes on top of theirs.	21:27
rgr	just need to double check. I have just made substantial changes to my code base. I have not checked it in. I forgot to branch before hand. Do I need to stash before making the new branch and then restore the stash after changing to the new branch or can I simply branch and then check in? Whats best procedure?	21:28
esc	whats the current trend in Terminology, Index, Staging Area, Cache or just Stage?	21:28
	← boombatower left	21:28
	← _Vi left	21:28
	→ shennyg joined	21:28
blackxored	anyone?	21:29
fr0sty	cirwin: aaron01: 'git fetch' or 'git fetch <remotename>' will update remote tracking branches, but 'git fetch <remote> <refspec>' will only update FETCH_HEAD	21:29
cbreak	rgr: just create it	21:29
amackera	hey all, any ideas about a git svn error: "Unable to determine upstream SVN information from working tree history"	21:29
cbreak	blackxored: just learn to read	21:29
blackxored	cbreak, --set-upstream doesn't work	21:29
fr0sty	esc: all of them, depending on who you talk to.	21:29
cbreak	sure it does.	21:29
blackxored	i want to track a remote branch from a local branch	21:29
fr0sty	'index' is still the canonical name as far as I know.	21:29
blackxored	cbreak, you pass <remote> <local> to it right ;)	21:29
	??/	21:29
cbreak	no	21:30
blackxored	or local remote	21:30
esc	fr0sty: yeah, thats what i use, but i find stage has a much nicer ring to it.	21:30
cbreak	just read the damn docs...	21:30
	before complaining	21:30
blackxored	Branch origin/master set up to track local branch master.	21:30
	i'm not complaining	21:30
esc	fr0sty: but then is it: "in the stage" or "on the stage"?	21:30
	← darkredandyellow left	21:30
blackxored	and you need to be a little more kind with people ;) i just don't remember	21:30
aaron01	fr0sty: thanks for the clarification	21:30
cbreak	just switch the args.	21:31
fr0sty	esc: 'in the index' ;-)	21:31
	→ darkredandyellow joined	21:31
cbreak	git branchs first argument is always the branch name of the branch you want to change.	21:31
esc	fr0sty: yeah i use it too	21:31
blackxored	ambigous object name origin/master	21:31
	by typing	21:31
	git branch --set-upstream master origin/master	21:32
	← gebi left	21:32
	→ Eridius joined	21:32
fr0sty	blackxored: if you actually received "Branch origin/master set up to track local branch master." before it is because you created a local branch called origin/master.	21:32
	→ MetaCosm joined	21:32
fr0sty	git branch -d refs/heads/origin/master	21:33
blackxored	fr0sty, I don't have origin/master local branch dude	21:33
cbreak	sure you do	21:33
	you just created one above	21:33
rgr	just need to double check. I have just made substantial changes to my code base. I have not checked it in. I forgot to branch before hand. Do I need to stash before making the new branch and then restore the stash after changing to the new branch or can I simply branch and then check in? Whats best procedure?	21:33
cbreak	type git branch to list it :)	21:33
blackxored	cbreak, oh yeah ;)	21:33
fr0sty	blackxored: 'git branch'	21:33
blackxored	cbreak, was on the config ;)	21:33
cbreak	rgr: just learn to read, I already answered above	21:33
rgr	cbreak: your help would be much more appreciated without the snide comments. Thanks anyway.	21:33
	← rgr left	21:33
fr0sty	rgr: just create it.	21:34
	and remember: git is very reluctant to destroy your work	21:34
cbreak	I help to help people, not to make them worship me.	21:34
fr0sty	and will only do so if you ask nicely.	21:34
SethRobertson	cbreak: Your help would be much more useful if people would listen	21:34
blackxored	I did it already, thanks so much guys, I've always blown the param order ;)	21:34
cbreak	SethRobertson: maybe I should try all caps?	21:34
blackxored	no no caps plz ;)	21:34
	→ s0ber_ joined	21:35
	← Mannequin1 left	21:35
	→ Mannequin joined	21:36
acs	what is the diff between checkout and clone ?	21:36
Mannequin	hi. I'd like to deploy to a remote server using git.	21:36
acs	or between checkout and branch ?	21:36
Mannequin	Problem: remote server doesn't have git installed, nor seems possible to install it. (remote server is just a shared hosting server)	21:36
	← manoot left	21:36
cbreak	acs: they do completely different things	21:36
esc	acs: checkout is for local versions or branches, clone for an entire git repository	21:36
cbreak	what do you think is NOT the difference?	21:37
mnemoc	clone does a checkout too :)	21:37
acs	esc: and between checkout & branch ?	21:37
Mannequin	I wonder: could it be possible to mount a remote server folder and use that folder as it is a local folder?	21:37
cbreak	Mannequin: git's not a deployment tool. Consider using hooks to start some deployment script	21:37
Mannequin	(and thus, be able to work with git)	21:37
acs	mnemoc: hmmm...	21:37
	← s0ber left	21:37
esc	acs: in a branch context, checkout will switch branches, and branch will create them	21:37
	→ manoot joined	21:37
	s0ber_ → s0ber	21:37
fr0sty	Mannequin: git is not a deployment tool	21:37
esc	mnemoc: indeed :)	21:37
Cacheaway	Mannequin: possible, linux has various network-based mounting options, i am pretty sure there's an ftpmount, and probably an ssh mount	21:38
	Mannequin: why not just use rsync though?	21:38
cbreak	branch can also destroy, change or rename branches	21:38
acs	esc: ah, checkout moves a pointer on that graph... and branch does not	21:38
cbreak	no	21:38
fr0sty	Mannequin: man git-archive which can get you a tarball of a particular revision.	21:38
jast	Mannequin: the 'git-archive' manpage can be found at http://git.or.cz/man/git-archive [automatic message]	21:38
cbreak	checkout does not move pointers in graphs	21:38
	← dunpeal left	21:38
cbreak	in it's normal mode it just changes HEAD	21:38
Cacheaway	Mannequin: do you have shell access to the host? you would need that anyway	21:39
Mannequin	cbreak, fr0sty: yes, I've read -not that much- about git not being a deployment tool. But I don't know/understand the downside of deploying with git. It seems to make deployment so easy...	21:39
	(note: I'll keep the .git/ folder outside DocumentRoot)	21:39
	Cacheaway: yes, I've shell access to the host	21:39
fr0sty	Mannequin: #1 for web deployment is that git does not track r/w permissions.	21:40
	but there are many others...	21:40
	← petercoulton left	21:40
cbreak	git doesn't track any permission besides executability	21:40
esc	would it be reasonable to claim that checkout switches branches?	21:40
Cacheaway	Mannequin: then can you download and build git there yourself?	21:40
acs	it just changes HEAD : this is what esc wants to say "checkout will switch branches" ?	21:40
cbreak	in normal mode it does, indeed	21:40
	you can also detach your head with it	21:41
	but it has a few more special modes	21:41
	IslandUsurper draws a finger across his throat menacingly.	21:41
	mnemoc doesn't like to have the head detached	21:41
esc	yes, and restore files in your working tree that have been modified	21:41
	← schambers left	21:41
cbreak	not in normal mode	21:41
	→ fowlduck joined	21:41
Mannequin	Cacheaway: I've read about that, it looked a bit overkilling (and mounting a remote folder on my local machine seems easier...), but will look into it again.	21:41
acs	esc: ya	21:42
	I try it now:	21:42
	git checkout origin/gtk-tabs	21:42
	(in emacs)	21:42
esc	acs: trying out commands is always best	21:42
	→ compnerd joined	21:42
IslandUsurper	err	21:42
esc	acs: but checking out a remote-trackin-branch will not be so helpful :(	21:43
acs	indeed, it switched to another branch, and destroyed the original master	21:43
IslandUsurper	reading what commands actually do is better	21:43
Cacheaway	Mannequin: one thing people seem to not get about unix systems is that there is no universal install that requires root permission. you should be able to build and "install" everything inside your home directory	21:43
esc	acs: the master is still there, its not destroyed	21:43
Cacheaway	unless your host has it mounted noexec. . .	21:43
	← HG` left	21:43
Mannequin	Cacheaway: thanks for the tip	21:43
acs	If you want to create a new branch from this checkout, you may do so	21:43
	(now or later) by using -b with the checkout command again. Example:	21:43
	git checkout -b <new_branch_name>	21:43
	HEAD is now at dcb86ef... Tab keys are symbols. Tab switch is in elisp. Save/restore point in tabs also.	21:43
esc	acs: use git checkout master to "switch" back	21:44
	acs: and use a pastebin for pasting, please :)	21:44
acs	(this is displayed after git checkout origin/gtk-tabs	21:44
Cacheaway	Mannequin: if there is an autoconf you probably want to start out with something like ./configure --prefix=~/	21:44
acs	ok	21:44
	LOL	21:45
esc	acs: what you did was checkout a remote-tracking-branch, you will not be able to commit there	21:45
acs	git grep tabs src/* reports lots of matches now	21:45
esc	acs: don't use emacs :) sorry.	21:46
acs	esc: hm ?	21:46
	← zombor left	21:46
	→ jonesy joined	21:46
	← d0k left	21:46
	→ linusoleander joined	21:46
acs	emacs is the greatest editor of all the times	21:46
slide	How do I pove a repo to my own server? do I just change the remote and push it back to the new bare git repo their?	21:46
esc	acs: try this one: http://www.gitready.com/beginner/2009/03/09/remote-tracking-branches.html	21:46
jonesy	is there a way to clone the contents of a git repo into the $CWD? In other words, I don't want it to create a directory for me.	21:47
	← joel_falcou left	21:47
slide	s/pove/move	21:48
	→ Athox joined	21:48
	← jonesy left	21:49
	→ Murr_ joined	21:49
	← manoot left	21:49
SethRobertson	slide: Sure, that is one way	21:49
Mannequin	cbreak, fr0zty: will research a bit more about using (or not) git for deployment. Besides the permission issues, it's too tempting to run a website directly from a git repository...	21:49
slide	is that the normal way? heh	21:49
	→ manoot joined	21:49
Mannequin	1) it makes it easier to backport changes done directly in the production website (like images or other non-database content added by clients)	21:49
	2) you could do a `git checkout maintenance_mode` to switch to, well, a branch purposed to maintenance	21:49
	3) running `git status` would help to spot added/modified (hacked?) files	21:49
SethRobertson	slide: Another is to do a `git clone --bare --mirror oldupstream` and then repoint the remote	21:49
	← goldbergus left	21:49
SethRobertson	Mannequin: http://toroid.org/ams/git-website-howto if you did not see it before	21:50
slide	SethRobertson, do that on the server right?	21:50
Mannequin	SethRobertson: thanks	21:50
SethRobertson	slide: new server, yes	21:50
	However, I agree that git is not a deployment tool	21:50
	← Anti-X left	21:51
	← blackxored left	21:52
	← manoot left	21:52
	← nullvariable left	21:52
	→ manoot joined	21:53
	← jaql left	21:53
acs	esc: I read it. I do not understand too much. What's the diff between a local branch and a remote branch ?	21:54
	→ mrFrog joined	21:55
	← mikeric left	21:55
	→ schambers joined	21:56
	← Dave^\| left	21:57
	← manoot left	21:57
	→ manoot joined	21:58
esc	acs: i try to explain, remote-tracking-branches are pointers in your DAG that point to the same commits as those pointers in your remote repository, for example "origin"	21:58
	← chrisshattuck left	21:58
	← schambers left	21:58
	→ p3rg2z joined	21:58
	→ frogonwheels joined	21:59
	← WebDragon left	21:59
	→ aspotashev joined	21:59
esc	acs: so if master points to commit 1234abcd in the remote repository "origin", and you clone that repository, origin/master will point to 1234abcd in the clone	21:59
	← IslandUsurper left	21:59
	← cilly left	21:59
	← matjas left	22:00
esc	acs: is that clear?	22:00
acs	esc: i am pondering ...	22:00
slide	SethRobertson, when i try to clone the new repo on the server im getting "does not appear to be a git repository"	22:01
	→ d2dchat joined	22:01
mrFrog	Can someone explain something to me? I'm confused with how deleting files works. If I start a new branch and delete/move files and then switch back to the master branch, shouldn't those files still be there?	22:01
SethRobertson	slide: On the new server you are supposed to clone (bare/mirror) from the old server	22:01
cirwin	mrFrog: if they were committed, yes, otherwise no	22:02
acs	esc: by cloning, a remote repository lies on my own computer, even if it is "remote", isn't it ?	22:02
SethRobertson	slide: The old mirror obviously still has to be working for this technique to work.	22:02
slide	i ran, "git clone --bare --mirror theoriginalsource newdir"	22:02
mrFrog	I though I had. I'll check again though and try it.	22:02
	rayners_afk → rayners	22:02
esc	acs: depends	22:02
	→ anathematic joined	22:02
	→ joel_falcou joined	22:02
	← mfacenet left	22:02
	← p3rg2z left	22:03
SethRobertson	slide: and theoriginalsource is the same as the `url shows in your original clone by `git remote -v show`?	22:03
esc	acs: if you have a repository in a directory locally, and you clone that into another directory, then you "remote" is effectively local	22:03
	← sundsted left	22:03
slide	SethRobertson, i used the public repo url that i would normally use with git clone	22:03
acs	esc: after I make a clone after a remote repo, and within that remote repo arrive other commits from the others, how can I update my clone, such that my "remote" repo to be syncronized with the remote repo ?	22:04
esc	acs: if you clone a repository over the net, the remote repository resides on the server from which you cloned it	22:04
SethRobertson	slide: but...is it exactly the same as the `git remote -v show`?	22:04
slide	SethRobertson, nope, but it did pull down all their code	22:04
acs	esc: I see...	22:04
SethRobertson	slide: you are confusing me. I thought you said it failed	22:04
slide	SethRobertson, oh no the server part worked fine. but when I try to clone that on my local computer it fails	22:04
esc	there are basically 3 commands you need to interact with remote repositories	22:05
acs	esc: however, when I clone a repo over the net, like emacs.git, it brings to me about 350MB.	22:05
esc	acs: fetch and pull -- and push	22:05
	acs: yeah... its emacs	22:05
SethRobertson	slide: what command failed?	22:05
esc	acs: you get the emacs source code, with all its history	22:05
mrFrog	ciwin: It still doesn't do what I'm thinking it should. If I commit my master branch and then do 'git checkout -b testbranch' and then remove a folder in the project directory and then do 'git checkout master', the files aren't there anymore.	22:06
acs	fetch brings latest commits, and pull does the same ?	22:06
slide	"git clone urlfornewrepo" for the new repo	22:06
esc	acs: pull does some more magic	22:06
	← cannonball left	22:06
acs	probably push "pushes" my patches remotely (if I have access)	22:06
SethRobertson	slide: Is that url something like: ssh://hostname/path/to/repo?	22:06
esc	if the branch you are on, and this is where it gets confusing, is setup to track a remote branch, pull will do a fetch and a merge	22:06
acs	esc: what does pull in plus than fetch ?	22:06
	← Torsten left	22:06
	→ noahcampbell joined	22:07
noahcampbell	When I get this message: Untracked working tree file 'packaging/rundeck.spec' would be overwritten by merge.	22:07
acs	esc: i see.	22:07
slide	yes	22:07
noahcampbell	when I try a checkout...the interwebs says to do a git co -f	22:07
mrFrog	cirwin: It still doesn't do what I'm thinking it should. If I commit my master branch and then do 'git checkout -b testbranch' and then remove a folder in the project directory and then do 'git checkout master', the files aren't there anymore.	22:07
esc	acs: thats a bit tricky, unfortunately	22:07
noahcampbell	Is there something more elegant, then simply blowing away the file?	22:08
SethRobertson	slide: Then what does `ssh hostname ls /path/to/repo` print?	22:08
noahcampbell	Can I merge the two versions?	22:08
	← mlq left	22:08
cbreak	noahcampbell: delete it, rename it...	22:08
cirwin	mrFrog: bizarre — does git whatchanged show you that you committed them on master correctly?	22:08
cbreak	noahcampbell: or start tracking it	22:08
	→ quentusrex joined	22:08
fr0sty	mrFrog: that is because 'testbranch' and 'master' are at the same commit	22:08
quentusrex	How do I get git to stop pushing a particular branch.	22:08
fr0sty	git checkout shouldn't change anything in that case.	22:08
quentusrex	I no longer want that branch pushed to the central repo	22:08
esc	and push will push stuff, however by default it will push all local branches, which have a branch on the remote sind with the same name	22:08
cirwin	noahcampbell: not easily, unless you know what the parent was	22:08
fr0sty	noahcampbell: move the file aside.	22:09
esc	acs: you can configure that behaviour with push.default	22:09
noahcampbell	okay...thanks	22:09
slide	SethRobertson, i think i failed to add the repo to my list of allowed repos, (i havnt messed with this server code in a while) let me see....	22:09
mrFrog	fr0sty: then what would I do to switch back and have my files back?	22:09
fr0sty	mrFrog: git status will tell you.	22:09
esc	acs: but the rule of thumb is that fetch just updates remote-tracking-branches, and pull will also do a merge	22:09
fr0sty	(presuming the directory was tracked by git.	22:10
esc	acs: but sometimes i can't get fetch to do what i want :(	22:10
	← ph^ left	22:11
mrFrog	git status just says 'deleted: filename.ext' (3 lines)	22:11
acs	esc: ya, I am too beginner	22:12
	← manoot left	22:12
acs	esc: I need to learn it, otherwise I cannot use emacs as I wish ...	22:13
	→ manoot joined	22:13
	→ jmil joined	22:13
	← jarrod_ left	22:13
	← dquestions left	22:14
mrFrog	fr0sty: what should I do to keep track of moved/deleted files so they come back when I switch branches?	22:14
fr0sty	esc: what is your problem?	22:14
	use 'git mv' or 'git rm', commit the change, then switch branches.	22:14
	→ skalnik joined	22:14
esc	sometimes git fetch will not update my remote-tracking-branches	22:14
fr0sty	esc using what command?	22:14
mrFrog	so I have to use git to move the files or remove them? It doesn't just figure that out?	22:14
acs	esc: have you seen these videos : http://edgyu.excess.org/git-tutorial/2008-07-09/	22:15
esc	fr0sty: git fetch $REMOTE	22:15
	← manoot left	22:15
	← Murr_ left	22:15
	→ dquestions joined	22:15
esc	fr0sty: git remote update will work	22:15
	→ manoot joined	22:15
esc	fr0sty: its whenever i am in a branch not set up to track anything	22:15
	fr0sty: i think	22:15
fr0sty	esc which remotes are you giving?	22:16
skalnik	Is there anyway to refer to commits from the initial commit? Like master# is the original commit on master and master#1 is the second and so on?	22:16
fr0sty	and are you running 'git remote update' or 'git remote update <remotename>' ?	22:16
esc	i run plain 'git remote update'	22:16
fr0sty	skalnik: no. Git's graph is not bi-directional.	22:16
	esc and which remote is not updating?	22:16
	← yairgo left	22:16
esc	fr0sty: no with 'git remote update' everything is fine	22:17
	but a plain 'git fetch $REMOTE' will fail	22:17
fr0sty	how many remotes do you have.	22:17
	?	22:17
esc	one sec	22:17
fr0sty	I was asking about git fetch, sorry I wasn'	22:17
	t clear	22:17
skalnik	fr0sty: Alright thanks. Is there any easy way to walk commits from the initial commit to master?	22:17
esc	i should create a cannonical example	22:17
fr0sty	skalnik: what are you trying to do?	22:18
	← rayners left	22:18
	q0rban → q0rban\|afk	22:18
skalnik	fr0sty: I'm trying to do a code demo where each commit is another step in the demo	22:18
	← opalka left	22:18
	→ sonnym joined	22:18
fr0sty	git log --oneline --reverse	22:18
skalnik	Sweet, that's a lot of help :)	22:19
	→ stanmancan joined	22:20
esc	fr0sty: i seem to be unable to recreate this now, i will pay more attention next time it happens	22:21
	← mrFrog left	22:22
stanmancan	Sorry for a lame question, I'm new with Git and in a bit of a crunch so I don't have time to look into it myself... If I have changes done locally and want to pull the latest version, git tells me I need to stash or commit. I don't want to commit, so is there a way to stash my changes, pull the latest ones from the repo, and then re-apply my old un-commited changes?	22:22
cirwin	man git-stash	22:22
jast	cirwin: the 'git-stash' manpage can be found at http://git.or.cz/man/git-stash [automatic message]	22:22
stanmancan	Missed my first sentence I take it? :(	22:23
cirwin	if you'd clicked the link you would already know the answer	22:23
	← simosx left	22:23
stanmancan	It's funny how some IRC channels like to be friendly and useful, and others like to be cirwin	22:24
mnemoc	git log --date-order still lists them in reverse commit order even after setting the commiter env equal to author env :(	22:24
bronson	stanmancan, you're asking someone to take type to type the git-stash manpage into irc?	22:24
	that would take more of your time too.	22:25
crab	stanmancan: stash. pull. stash pop.	22:25
	← notjohn left	22:25
cirwin	bronson: it's too late, he already made me type more than I needed to to explain	22:25
	← MikhX left	22:25
bronson	cirwin, I know, it's weird.	22:25
	→ MikhX joined	22:26
	← gh34 left	22:26
stanmancan	Actualyl crab answered my question in four words.	22:26
	← manoot left	22:27
cirwin	mine would have been two if it had worked :)	22:27
	you win some you lose some	22:27
fr0sty	esc: 'git fetch' with no remote argument will only work if tracking is set up (man git-fetch will explain). 'git fetch <remotename>' should do pretty much what 'git remote update <name>' would do.	22:27
jast	esc: the 'git-fetch' manpage can be found at http://git.or.cz/man/git-fetch [automatic message]	22:27
	→ Guest106371 joined	22:27
	← madsy left	22:27
	→ _Vi joined	22:28
	← ereslibre left	22:28
	→ manoot joined	22:28
esc	fr0sty: yeah i reckon so, i just remember being baffled a couple of times, cause it didn't :)	22:28
	→ mikeric joined	22:29
	→ psoo joined	22:29
stanmancan	"file.ext: needs merge \| cannot apply to a dirty working tree, pelase stage your changes"	22:30
fr0sty	stanmancan: what did you do?	22:30
	← pantsman left	22:30
stanmancan	git stash, git pull, git stash pop	22:30
fr0sty	stanmancan: you need to resolve the conflicts from 'git pull' first	22:30
stanmancan	fr0sty: and I go about that .... how?	22:31
fr0sty	I have no way of knowing what conflicts your merge produced.	22:31
	git status will tell you what files need to be dealt with.	22:32
	→ kar8nga joined	22:32
fr0sty	those files will have >>>>>> ======= <<<<<< conflict markers in them.	22:32
	If you hadn't figured it out yet, "file.ext" is one of those files.	22:32
mnemoc	how can I pass an ordered file to git rebase -i? `git log "--pretty=format:%at %h %s" \| sort -n \| sed -e 's,^[^ ]\+,pick,'`	22:33
	← lhz left	22:33
	→ Mage joined	22:33
cbreak	mnemoc: evil!	22:34
	tried copy&paste?	22:34
fr0sty	mnemoc: cut and paste into the editor	22:34
mnemoc	ouch	22:34
fr0sty	what were you expecting?	22:34
	← Eridius left	22:34
	← MattDiPasquale left	22:34
mnemoc	cbreak: yes, evil :)	22:34
	← MikhX left	22:34
	→ zwnj joined	22:35
mnemoc	fr0sty: an argument to pass a file instead of opening $EDITOR :)	22:35
	← Yuffster left	22:35
stanmancan	fr0sty: is it possible to just deelete the conflicting file from the stash?	22:35
fr0sty	mnemoc: GIT_EDITOR=my_evil_plan.sh	22:35
mnemoc	oh	22:35
	fr0sty: thanks :)	22:36
	← pigdude left	22:36
fr0sty	stanmancan: this has nothing to do with the stash. you need to complete the pull. (or did you do that already?)	22:36
stanmancan	fr0sty: the pull went fine, it's not happy when i try to stash pop	22:36
	← khmarbaise left	22:37
	← manoot left	22:37
fr0sty	stanmancan: what does 'git status' tell you.	22:37
stanmancan	Unmerged Paths: both modified: file.ext	22:38
	→ manoot joined	22:38
stanmancan	i don't need the stashed version of file.ext though	22:38
	← rryouumaa left	22:38
fr0sty	that means your pull did not got 'fine'	22:39
	← stringoO left	22:39
	→ airborn joined	22:39
fr0sty	I will repeat, since you are short on time: your current problem has nothing to do with the stash and everything to do with your failed merge.	22:39
	→ Eridius joined	22:39
	→ pantsman joined	22:41
	← pantsman left	22:41
	→ pantsman joined	22:41
	← 30BAAGZB3 left	22:41
	← ehabkost left	22:41
	→ kylehayes joined	22:42
	← alin-gym left	22:42
mnemoc	GIT_EDITOR=../evil_editor.sh git rebase -i 3dc2588c4	22:42
	Successfully rebased and updated refs/heads/master.	22:42
	:D	22:42
	\o/	22:42
	→ MikhX joined	22:44
	← tildeequals left	22:44
fr0sty	mnemoc: congratulations on whatever it is you just managed to do.	22:44
	← kylehayes left	22:45
	→ alin-gym joined	22:45
	cirwin loves fr0sty's EDITOR= git rebase trick	22:45
	← amason left	22:46
	→ khmarbaise joined	22:46
stanmancan	thanks for the help fr0sty	22:47
fr0sty	did you figure it out?	22:47
stanmancan	i just ended up deleting the messedup files and restoring from a backup LD	22:47
	→ Slix` joined	22:47
stanmancan	all looks good now	22:47
	well, good considering	22:47
	→ amason joined	22:47
	← anathematic left	22:47
	→ kylehayes joined	22:48
fr0sty	that is one way to do it...	22:48
	→ anathematic joined	22:48
cbreak	if you have a git repo, just restore them from the history	22:48
stanmancan	Yeah... I need to go through git a bit more thoroughly and really figure it out but I'm _swamped_	22:48
	my manager decided to disappear the last 3 days right as we're going live with a bunch of trials	22:49
	so now i'm the one scrambling to get our servers up, bugs fixed, trials running ect.	22:49
cbreak	lucky you	22:49
stanmancan	:(	22:49
cbreak	managers just stand in the way anyway	22:49
stanmancan	yeah, but it's also a second set of hands	22:49
cbreak	a set with two left hands? :)	22:50
stanmancan	and he's the one that's developed 80% of the application so	22:50
	→ stringoO joined	22:50
stanmancan	:D, well put	22:50
fr0sty	cbreak: what do you have against left hands?	22:52
cbreak	in general? nothing. If they are attached to a right arm it gets anoying though.	22:53
	← khmarbaise left	22:53
	← kylehayes left	22:54
fr0sty	fair enough...	22:54
	→ kylehayes joined	22:54
	→ airborn_ joined	22:54
mnemoc	fr0sty: turn 4 cvs virtual modules into a single git repo with all individual commits ordered by date	22:55
	→ MattDiPasquale joined	22:55
	→ simplecoder joined	22:56
SethRobertson	fr0sty/mnemoc: and more specifically developed on the same branch (as I understand it)	22:56
mnemoc	yes, master-only	22:56
	but it's good enough for my needs :)	22:57
	← flazz left	22:57
SethRobertson	I meant appearing as if they were developed on the same branch, as opposed to what sub-tree merge does	22:57
	mnemoc nods	22:57
fr0sty	so you merged the four and then rebase -i to make it all linear by commit date, eh?	22:58
	← airborn left	22:58
	airborn_ → airborn	22:58
	← dgtlmoon left	22:58
mnemoc	fr0sty: yes, but without a "merge" in the history	22:58
fr0sty	hence: 'linear'	22:58
mnemoc	right	22:58
	→ airborn_ joined	22:58
mnemoc	which i doubt can be done on any other VCS	22:59
	→ joel__ joined	22:59
	← airborn_ left	22:59
fr0sty	well, not in anything approaching an automated fashion.	22:59
	← manoot left	22:59
	← joel_falcou left	22:59
	→ manoot joined	23:00
fr0sty	you could probably pull patches and timestamps, mangle the paths, sort them and apply them, but that is for people with more time than sense.	23:00
	← zwnj left	23:00
mnemoc	=)	23:00
fr0sty	that doesn't count as 'using' a VCS though...	23:00
	← coderdad left	23:02
	→ guardian joined	23:02
	q0rban\|afk → q0rban	23:03
	→ mlq joined	23:04
	← mlq left	23:04
	→ mlq joined	23:04
helo	i used git fsck to find a stash i accidentally blew away... how do i recreate a stash from the sha?	23:05
	just cherry-pick, reset HEAD^, and stash it?	23:05
fr0sty	helo: git stash apply <sha> doesn't work?	23:06
	q0rban → q0rban\|afk	23:06
	→ ricky joined	23:06
fr0sty	hmm, docs don't seem to mention that working...	23:07
helo	ahh, i'd never referred to a stash via sha... thanks :/	23:07
fr0sty	can't hurt to try though.	23:07
	→ wuj joined	23:07
	← warbee left	23:08
fr0sty	otherwise a stash is just another commit, so your cherry-pick+reset+stash would work.	23:08
	← ricky left	23:08
helo	apparently must be cherry-picked as a merge	23:09
	stash apply <sha> worked fine, thanks :)	23:09
fr0sty	good to know, that might warrant a mention in the docs...	23:09
	→ ricky joined	23:10
	← MikhX left	23:10
	→ AdmiralSausage joined	23:10
	← kipras left	23:10
	← jrmuizel left	23:10
	→ MikhX joined	23:11
	← ceez left	23:12
	← j0shua left	23:12
	← burhan left	23:15
	→ shiba_yu36 joined	23:15
	← stringoO left	23:16
	← dquestions left	23:16
	→ anathema_ joined	23:16
	→ dquestions joined	23:16
	→ beatak_tksh joined	23:17
	← beatak_tksh left	23:17
	← anathematic left	23:17
	→ mrFrog joined	23:18
	← sraue left	23:19
	→ sraue joined	23:19
	← beatak left	23:20
mrFrog	so I have git installed on my local computer as well as a remote server. How can I push the repo to the remote server?	23:23
cbreak	with git push...	23:24
	← guardian left	23:25
cbreak	create a bare repo at the destination, add the destination as remote in the source	23:25
	then push	23:25
	← tizbac_ left	23:25
	← simplecoder left	23:26
	← albel727 left	23:26
mrFrog	ok so how do I add a url as remote?	23:26
cbreak	you add the repository as remote	23:27
	← Ryback_ left	23:27
cbreak	with, surprise surprise, git remote add	23:27
	read the man page	23:27
bremner	mrFrog: have you considered man gittutorial ?	23:27
jast	mrFrog: the 'gittutorial' manpage can be found at http://git.or.cz/man/gittutorial [automatic message]	23:27
	→ simplechat joined	23:28
	→ brosner joined	23:29
	← mrFrog left	23:30
	← kylehayes left	23:30
	→ Khisanth joined	23:32
	→ xiong joined	23:36
	→ zorganae joined	23:37
	→ guardian joined	23:38
	→ dgtlmoon joined	23:39
	← TeckniX left	23:41
	→ guille_ joined	23:42
guille_	hi	23:42
	← thiago_home left	23:44
fr0sty	hi	23:44
	→ constantc joined	23:46
fr0sty	...	23:46
	→ ignatp_ joined	23:47
	→ project2501a joined	23:47
guille_	I changed with a tree-filter the .gitmodules to point to a different url but github isn't linking the external repositories correctly. Any ideas?	23:47
	→ zombor joined	23:48
	← project2501a left	23:48
guille_	(is there any other place where the submodules info gets stored?)	23:48
	→ schambers joined	23:48
coldboot	Is there a git command to remove a file from the index and working tree completely?	23:49
	It seems you need to "git reset; git checkout; git reset; git checkout" for some files with status 'UU', 'DU', and others.	23:49
	← pantsman left	23:50
	← constantc left	23:50
SethRobertson	guille_: see man git-submodule .gitmodules and .git/config have info	23:50
jast	SethRobertson: the 'git-submodule' manpage can be found at http://git.or.cz/man/git-submodule [automatic message]	23:50
SethRobertson	guille: However, I would like to take the opportunity to promote gitslave as an alternative to git-submodules. It can be better (or worse) depending on your desired workflow.	23:51
	← masonkramer left	23:51
	→ masonkramer joined	23:51
	→ eletuchy joined	23:51
SethRobertson	It also of course tracks submodule information in the internal git tree	23:52
	→ Tuller joined	23:53
	→ jarrod_ joined	23:55
guille_	SethRobertson: I'll check it out. Now trying doing a git submodule sync after .gitmodules have been sed'ed.	23:55
	thanks	23:55
	← kar8nga left	23:56
fr0sty	guille_: did you force a push on your filtered branches?	23:57
	← manoot left	23:58
	← AdmiralSausage left	23:58
	→ manoot joined	23:58

Logs Search ←Prev date Next date→

Channels Documentation