IRCloggy #git 2012-01-07

Provider of IRC logs since 2005.

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

	2012-01-07
	← beatak_m left	00:00
	← churp left	00:00
sitaram	ShadeHawk: if I had to convince a PHB to move to a DVCS (any DVCS), section 1 of hginit.com wins over the git parable; though I'd have someone turn it into a presentation :-)	00:00
	← rendar left	00:01
	← stringoO left	00:01
ShadeHawk	sitaram: doesn't Pro Git include that?	00:02
	sitaram: anywa, we have now the new source: "Version Control by Example" by Eric Sink (with examples in Subversion, Mercurial, Git and their own new enterprisy DVCS Veracity)	00:03
	sitaram: and VCBE has nice chapter discussing merits and demerits of DVCS	00:04
sitaram	wow	00:04
	veracity... rings a bell	00:04
	← nixmaniack left	00:05
sitaram	oh	00:05
	these are the guys who claim "formal rename" as a plus point... I remember now	00:05
	← alexgl left	00:06
	← linusoleander left	00:06
	← BonQuisha left	00:06
	→ Morphous_ joined	00:06
sitaram	ShadeHawk: https://plus.google.com/115609618223925128756/posts/PDPdXTxAvZk	00:06
	→ stringoO joined	00:06
	→ BonQuisha joined	00:07
	← johnanderson left	00:07
	→ johnanderson joined	00:07
	← jaequery left	00:08
ShadeHawk	"Mercurial only needs enough letters to disambiguate, and there are no other commands that start with st." <-- I wonder if it would be worth to add this feature to Git (in addition to existing help.autocorrect (!!!))	00:08
	← flavius left	00:09
mugwump	ShadeHawk: tab completion is better I think	00:09
	→ linusoleander joined	00:09
	→ jaequery joined	00:09
sitaram	ShadeHawk: could also mess with many people's aliases, no?	00:10
ShadeHawk	mugwump: yep, though only bash and zsh are supported (the completion stuff from contrib covers both, but IIRC there is also native zsh one)	00:12
Mikachu	the one distributed with zsh is much better, yes	00:12
ShadeHawk	mugwump: cmd.exe is not supported, I don't think ;-)	00:12
	← Akufen left	00:13
	mugwump quickly ports git-completion.bash to 4DOS	00:13
	→ delight joined	00:13
ShadeHawk	Mikachu: in what sense is zsh completion from zsh better than one from contrib area in git.git? More commands supported, more up-to-date, or better context sensitivity (e.g. competing refs where applicable)?	00:13
jahid	is there any easy tutorial about making a git that will be accessed over the internet?	00:13
	← Bartzy left	00:13
Mikachu	ShadeHawk: the last one	00:13
	← mandric left	00:14
cmn	jahid: try out !gitolite	00:14
ShadeHawk	jahid: what do you mean by "making the git"?	00:14
gitinfo	jahid: gitolite allows you to host any number of repositories with extensive access control options for any number of users, using just one system account: http://github.com/sitaramc/gitolite	00:14
ShadeHawk	!gitinfo	00:14
gitinfo	I am an IRC bot which responds to certain keywords to provide helpful(?) information to humans. Please see http://jk.gs/git/bot for more information about how to use me.	00:14
jahid	ShadeHawk, making a git repo	00:14
	→ niloc132_ joined	00:14
	← penguinbait left	00:14
ShadeHawk	jahid: "git init"	00:15
sitaram	--bare	00:15
	(at the very least)	00:15
ShadeHawk	sitaram: I think that "enough letters to disambiguate" would trigger as last resort, and only for commands	00:16
	← niloc132_ left	00:16
jahid	ShadeHawk, I tried all these, and finally did a git bare remote	00:16
	← ramsey left	00:16
sitaram	ShadeHawk: I guess yeah, but we'd have to learn to stop occasionally mis-spelling our own aliases in case they get mistaken for a git command :)	00:16
jahid	but didnt work when tried to access over ssh	00:16
sitaram	jahid: !doesntwork	00:17
gitinfo	jahid: What happened? Did it fall asleep on your couch? Eat your homework? Destroy your hard drive? Is it making faces at you? Does it want more money? Is it sleeping with your best friend? Please be specific!	00:17
	sitaram loves* that one*	00:17
	← brian_g left	00:18
jahid	sitaram, the remote guy can access my comp using ssh, but when he tried to do git clone, he gets message like "remote comp hanged ...."	00:18
	→ signal-9 joined	00:19
	← spiette left	00:19
ShadeHawk	does help.autocorrect work for aliases? help.shortcuts would work the same way ;-/	00:19
	← z0n3d left	00:19
sitaram	jahid: "can access" means he can get a shell?	00:19
jahid	sitaram, yes	00:20
	← iocor left	00:20
ShadeHawk	jahid: please tell us how exactly he tried to clone (whole command), and how he checked that SSH acces work	00:20
	← linusoleander left	00:20
ShadeHawk	e.g.	00:20
sitaram	jahid: if so, a "git clone" or whatever should also work, except for possible network speed/disconnect issues that may break the connection	00:20
ShadeHawk	$ ssh git.example.com hostname	00:20
jahid	ShadeHawk, git clone abc@my_IP:/opt/git/mygit.git	00:20
Mikachu	note also that foo:path is relative to homedir, while ssh:// is relative to /	00:20
FauxFaux	Mikachu: ssh://foo@bar:port/~/lol/ works.	00:21
jahid	ShadeHawk, he checked ssh works by logging in with "ssh abc@my_ip"	00:21
Mikachu	sure, just mean by default, if you don't type a leading /	00:21
	→ brian_g joined	00:21
Mikachu	jahid: can you git clone /opt/git/mygit.git on that machine?	00:21
jahid	yes	00:21
sitaram	jahid: but after logging in, does he have permissions to read /opt/git/mygit.git?	00:21
jahid	yes	00:21
sitaram	jahid: is it a large repo?	00:22
Mikachu	is the git upload-pack stuff in $PATH for noninteractive shells?	00:22
	or whatever it is	00:22
jahid	was my cloning right? "git clone abc@my_ip:/opt/git/my_git.git"	00:22
ShadeHawk	jahid: check non-interactive shell, like "ssh abc@my_IP git --version"	00:22
jahid	for ssh do i have to use antying like ssh?	00:22
	→ linusoleander joined	00:22
sitaram	remove the ":"	00:23
ShadeHawk	jahid: git must be installed on server (at least git-upload-pack and git-receive-pack must be somewhere you know)	00:23
Mikachu	sitaram: err?	00:23
jahid	ShadeHawk, no :o	00:23
	i just installed git-core	00:24
sitaram	Mikachu: he had 'git clone abc@IP:/path'; I said that colon needs to go. Is that not right (/me goes off to double check)	00:24
Mikachu	no, the : should be there	00:24
ShadeHawk	sitaram, jahid: "url = [email@hidden.address] works for me	00:24
	← dankest left	00:24
sitaram	Mikachu: oops;you're right. I claim it's too early in the morning and I haven't woken up yet :)	00:25
jahid	ShadeHawk, when i do "apt-get" for "git-upload-pack" and "git_receive-pack", it says unable to locate package	00:25
	← lonewulftp left	00:25
ShadeHawk	sitaram: "git gui blame" and "git blame -w -C -C" are wonderfull tools	00:25
sitaram	jahid: does he get any output at all before the "hangup"?	00:26
ShadeHawk	sitaram: and code movement _tracking_ would not work at all	00:26
sitaram	ShadeHawk: what does "-w" do?	00:26
jahid	sitaram, "remote computer hanged unexpectedly"	00:27
ShadeHawk	jahid: try "git clone --verbose ...", and perhaps also GIT_TRACE=some-file	00:27
	← baburdick left	00:27
	→ penguinbait joined	00:27
	← penguinbait left	00:27
	→ penguinbait joined	00:27
cmn	git-upload-pack and git-receive-pack are part of the normal git package in any reasonable package	00:27
	→ dv310p3r joined	00:27
	→ baburdick joined	00:27
ShadeHawk	sitaram: "-w" is passed to diff, and means ignore whitespace (very useful for finding code movement, which is usually accompanied by reindent)	00:27
	← grekko left	00:28
	← brian_g left	00:28
	→ lonewulftp joined	00:28
ShadeHawk	cmn: I meant here that minimally you can install those two locally, and it would work if you configure it right	00:28
	← monokrome left	00:28
cmn	agree, the comment was to jahid searching for those packages	00:28
	← dv310p3r left	00:29
	→ dv310p3r joined	00:30
lonewulftp	...	00:30
ShadeHawk	sitaram: about https://plus.google.com/115609618223925128756/posts/PDPdXTxAvZk -- I am not sure (from what I have read in VCBE) that how Veracity implements locking is correct and useful	00:31
sitaram	ShadeHawk: I suspected as much	00:31
	ShadeHawk: inconsistent with "D"VCS fundamentally	00:31
	ShadeHawk: http://sitaramc.github.com/tips/blame-detection-and-C-levels.html <-- your comments appreciated (bit old though; I wrote it long ago)	00:31
	← neurodrone left	00:32
ShadeHawk	sitaram: anyway for locking support in DVCS, blocking on lock must be build in or doable via hooks	00:32
	← koo4 left	00:33
	→ monokrome joined	00:33
ShadeHawk	And now for something completly different -- I wonder if it would be useful to have gits:// protocol, i.e. git:// over SSL / TLS	00:34
	← k299 left	00:34
sitaram	ShadeHawk: I can do that on the server side to block a push but by then the manual effort of the change has already been made on the client. You will need a client that queries and honors advisory locks	00:34
	ShadeHawk: wouldn't socat or such already be able to do that if you wanted to?	00:35
ShadeHawk	sitaram: pre-commit hook, or built-in into commit with ability to break locks	00:35
	← macRover left	00:36
	← Shusshu left	00:36
	→ zsw joined	00:36
	← toabctl left	00:37
zsw	Is there a way to pull a remote branch without changing from master beforehand?	00:37
sitaram	fetch yes, but 'pull' -- which requires a merge -- no	00:37
	→ mloskot joined	00:38
	← mloskot left	00:38
	→ mloskot joined	00:38
ShadeHawk	or rebase, but lets not go into that ;-)	00:38
sitaram	you can even force your local copy to be the same as the remote while still being on master but if you have independant changes that need to be merged, you need to checkout the branch and do the merge	00:38
	zsw: this is because a merge potentially requires manual intervention, so it needs a working tree, so it can only be on the checked out branch	00:39
zsw	sitaram: ah yes, thank you.	00:39
	→ Goplat joined	00:40
	← v0n left	00:40
	← d0k left	00:40
ShadeHawk	sitaram: use sockat to establisg secure tunnel, or something? I was thinking about using gits:// in URL, which would be much simpler (though I wonder if socat can be used to write remote helper for gits://)	00:42
	← monokrome left	00:42
	← TheShagg left	00:43
	← drizzd left	00:44
	→ fourcolors joined	00:44
sitaram	ShadeHawk: true. I was just saying that conceptually it should not be hard and should not require protocol fiddling	00:44
	→ reklipz joined	00:44
	← DarkAR left	00:44
	← ipalaus_ left	00:45
	← AaronMT left	00:45
sitaram	ShadeHawk: the big problem with SSL is server side certs. Sshd host key is more "personal" -- so you accept the "is this key ok to add to known hosts?" thing much more confidently than you would for an SSL "cert unknown" prompt	00:45
ShadeHawk	and on server side you would have to fiddle with git-daemon and socat or something yourself	00:46
reklipz	Is it possible to have git ignore the autocrlf setting for certain files? For instance, MS-DOS batch scripts require CRLF line endings. If these files change with any frequency, using the autocrlf feature makes this a nightmare (especially if they are committed with files that the autocrlf feature should apply to).	00:46
ShadeHawk	reklipz: man gitattributes	00:46
gitinfo	reklipz: the gitattributes manpage is available at http://jk.gs/gitattributes.html	00:46
reklipz	ShadeHawk: Perfect; thank you!	00:47
	← reklipz left	00:48
sitaram	uggh... "branch the mercurial way, by cloning repositories". Just as SVN conflates branch with directory, Hg seems to consistently conflate branch with clone	00:48
	SethRobertson: ^	00:48
	← warthog9 left	00:49
SethRobertson	Remember, I was only recommending the website as form over function.	00:49
	Mercurial branches are really weird. Not normal	00:49
	→ monokrome joined	00:50
	→ boombatower joined	00:50
sitaram	SethRobertson: oh yeah, I have ranted about Hg branches before. Not to mention their crazy "bookmarks" and the thrashing over how they should behave	00:50
ShadeHawk	sitaram: ad http://sitaramc.github.com/tips/blame-detection-and-C-levels.html -- I think that "git gui blame" uses "-w -C -C" for code movement detection	00:50
	sitaram: and it needs <title> too	00:50
sitaram	I thikn they all need titles :-) I have to look into that later... thanks	00:51
	ShadeHawk: if you're sure about the "-w" I'll just add that in; I already said it does "-C -C"	00:52
ShadeHawk	sitaram: with "bookmark branches" now transferrable in modern Mercurial, clone to branch in Mercurial is no longer true (thoigh hgbook still starts from that as supposedly simpler workflow)	00:52
mugwump	you need to branch by cloning with mercurial, otherwise you can't delete revisions you didn't mean to make	00:52
ShadeHawk	sitaram: best to check in "git gui" code, but as it ignores reindent I think it does use "-w"	00:52
	→ warthog9 joined	00:53
	→ uu joined	00:53
	← setmeaway left	00:53
	← boombatower left	00:54
ShadeHawk	anyway so "bookmark branches" are transferrable... but they have global namespace, I think to avoid remote-tracking branches ides and remote name to remotr-tracking name ref mapping	00:54
	^ in Mercurial	00:54
	→ setmeaway joined	00:55
	→ ehsan joined	00:56
	← hxts left	00:56
sitaram	I always felt Hg had a lot of "we won't admit we got it wrong in the first place" issues	00:57
	← baburdick left	00:57
sitaram	in respect of bookmarks, branches, and how git does it versus how Hg does it	00:57
	→ baburdick joined	00:58
	← ehsan left	00:58
	← jonshea left	00:58
	← Agos left	00:58
	← FernandoBasso left	00:59
	← pantsman left	01:01
	← fourcolors left	01:02
ShadeHawk	sitaram: tags in Mercurial, ehhh...	01:03
	← ScottE left	01:04
	← dreiss left	01:04
	← Chaser left	01:04
ShadeHawk	anyway, having git-serve (gitweb + git-daemon + smart HTTP) would be nice; Perl HTTP::Server::Simple perhaps (where Mercurial uses Python's one whatever it is)	01:04
	?	01:04
	BTW I wonder if gitweb would have to be modified to support multi-line headers (signed commits and merges of signed tags)	01:07
jast	so, not git-instaweb?	01:07
	→ ehsan joined	01:08
	← apok left	01:08
jast	I think it should be possible to extend instaweb to do smart HTTP, though I don't care nearly enough about that feature to be the one to do it ;)	01:08
ShadeHawk	jast: git-instaweb is only about 1 i.e gitweb of of 3: gitweb, git-daemon, git-http-backend (aka "smart" HTTP)	01:08
jast	adding git-daemon is trivial	01:09
ShadeHawk	hmmm... good idea (it could be e.g. 'git instaweb --serve')	01:11
	git-daemon standalone is not enough to be feature-equivalent with hg-serve, unless git-instaweb would enable pushing via git:// in git-daemon	01:11
	← jaequery left	01:11
	→ neurodrone joined	01:12
jast	which it could, but it's not that useful to support anonymous push	01:12
	← kenperkins left	01:14
	← Rogi_ left	01:15
ShadeHawk	sitaram: a bit of self promotion wrt. Git vs Mercurial comparison: http://stackoverflow.com/questions/1598759/git-and-mercurial-compare-and-contrast/1599930#1599930 (note that it is from Aug 2011; things change)	01:15
	jast: hg-serve has unathenthicated push too, isn't it?	01:15
	← jahid left	01:15
ShadeHawk	unauthenthicated and anonymous	01:15
sitaram	ShadeHawk: looking...	01:16
	ShadeHawk: I'll check it out when I come back; have to go away now... weekend chores :-(	01:16
	← ldlework left	01:17
	ShadeHawk wonders about localtime difference with sitaram	01:17
	← kenichi left	01:19
	← ehsan left	01:20
jast	ShadeHawk: I have no idea... but it shouldn't. ;)	01:22
	→ JustMozzy joined	01:22
JustMozzy	hi guys. is there anyway to undo a checkout? :s	01:23
jast	ShadeHawk: hg serve does the same thing by default as git-daemon	01:23
	and you can optionally enable anonymous push	01:23
ShadeHawk	ah	01:24
JustMozzy	I think it is not possible... but maybe someone here knows better? I accidentally called checkout on a file which was not committed yet :s is there a way to undo that?	01:24
SethRobertson	JustMozzy: !lost	01:27
gitinfo	JustMozzy: As long as you have `git commit`ed your changes (or even `git add`ed them), your changes will not be lost for over two weeks unless you work really hard at it. There are two places where "lost" changes can be hiding. They might be in the reflog (`git log -g`) or they might be in lost&found (`git fsck --unreachable`). Type "!dangling" and "!reflog" into IRC for more info.	01:27
JustMozzy	yupp, then they are lost... oh well... at least it was only one file	01:28
	← baburdick left	01:28
SethRobertson	If you had not committed or added the file, it is toast. Read !best_practices for information about avoiding this (and the dangers of checkout in file mode)	01:28
gitinfo	There is no one right answer for git best practices, but a consensus from #git is available at https://gist.github.com/1540906	01:28
	→ brian_g joined	01:28
	← spyro left	01:29
	→ spyro joined	01:29
EugeneKay	That is a wall of a page.	01:30
	→ baburdick joined	01:30
SethRobertson	That is a bicycle of a comment	01:30
	→ cannonball joined	01:30
	← millerti left	01:31
	→ boombatower joined	01:32
	→ jonshea joined	01:33
	← boombatower left	01:33
	→ MarcWeber joined	01:33
MarcWeber	Is there a way to reset a commit to the contents of another commit without taking history?	01:33
	I'd like to squash many commits - but squashing would cause many conflicts	01:34
	← ericmuyser left	01:34
	fr0stbyte → ericmuyser	01:34
	→ ericm joined	01:34
kevlarman	MarcWeber: um, what?	01:34
	← zsw left	01:34
	← ObseLeTe left	01:34
SethRobertson	MarcWeber: I think you want this: !branch_replace	01:34
gitinfo	MarcWeber: To replace one branch with the contents of another: git clean -dfx; git checkout $destination; git reset --hard $source; git reset --soft ORIG_HEAD; git add -A .; git commit -m "Rewrite $destination with $source"; git merge -s ours $source	01:34
SethRobertson	source doesn't have to be a branch, it could be a random SHA	01:35
	← spyro left	01:36
MarcWeber	git add -A may add random files.	01:36
SethRobertson	Not after the other commands	01:36
MarcWeber	Then I can create a diff and apply it - I don't have binaries.	01:36
	← penguinbait left	01:36
	→ diegoviola joined	01:36
	→ Sir_Rai joined	01:36
MarcWeber	clean -d, right	01:36
Sir_Rai	Hi	01:37
	← brian_g left	01:37
kevlarman	SethRobertson: that's a bit of a nuclear option isn't it?	01:37
	→ Prasad joined	01:37
	← mkg7 left	01:38
SethRobertson	kevlarman: perhaps, but it appears to answer the problem description	01:40
	← dv310p3r left	01:40
signal-9	greetings all. I am trying to push to my github pages account. I have not got the first push to work. When I input the command 'git push origin gh-pages' it tells me the repository is not found. Do I have to create a new repository called gh-pages and push to that?	01:41
SethRobertson	`git remote` says what?	01:41
signal-9	SethRobertson: it outputs 'origin'	01:42
SethRobertson	Can you !repro this for us?	01:42
gitinfo	please paste (e.g. using http://git.pastebin.com/) a transcript of your terminal session in which the problem occurred so that we can troubleshoot more easily.	01:42
	← robotmay left	01:43
signal-9	gitinfo: here is the paste http://pastebin.com/hfe6RTg1	01:44
	← RaptorX left	01:44
	→ dreiss joined	01:45
SethRobertson	How about `git remote show origin` Are the URLs correct?	01:46
signal-9	SethRobertson: I am not farmiliar with !repro can you tell me how to do this	01:46
gitinfo	SethRobertson: please paste (e.g. using http://git.pastebin.com/) a transcript of your terminal session in which the problem occurred so that we can troubleshoot more easily.	01:46
SethRobertson	you did it	01:46
jameslord	hi	01:47
SethRobertson	signal-9: !gitinfo	01:47
gitinfo	signal-9: I am an IRC bot which responds to certain keywords to provide helpful(?) information to humans. Please see http://jk.gs/git/bot for more information about how to use me.	01:47
jameslord	how to push all local branches to a bare remote repo?	01:47
	http://paste.ubuntu.com/795586/	01:47
cbreak	git push bareremote --all	01:48
signal-9	SethRobertson: here is the paste for git show	01:48
cbreak	surprise	01:48
signal-9	http://pastebin.com/Ws73ZMDf	01:48
jameslord	seems only one branch is pushed to remote	01:48
	→ spyro joined	01:48
	→ kylef joined	01:48
cbreak	jameslord: because you only push one.	01:48
	← Prasad left	01:48
cbreak	are you that surprised that git does what you tell it to?	01:48
kylef	Is it possible to use "git checkout-index" with submodules?	01:49
cbreak	what for?	01:49
	← thiago left	01:49
SethRobertson	signal-9: that is all messed up. Can you paste your .git/config file?	01:49
	← friskd left	01:49
	← delight left	01:50
signal-9	SethRobertson: http://pastebin.com/LdSM1Mgv	01:50
	→ apok joined	01:51
jameslord	thank you	01:51
SethRobertson	signal-9: your URL is messed up. You need to git remoet set-url origin URL	01:51
	→ flip_digits joined	01:51
SethRobertson	THe URL is given to you in your github page. [email@hidden.address]	01:52
jameslord	my networks is not very good, so connection out when using git is usual, if a git process is interrupted , can it start from the point it is interrupted, or it is a transaction, can only fail as a whole	01:53
SethRobertson	no restarts	01:53
	If you want to do something like that, you need to create a git-bundle type architecture and transfer using a restarting protocol and program	01:53
	Not recommended	01:53
	Well, recommended. Just not simple	01:54
jameslord	which protocol is recommended? ssh is ok?	01:54
SethRobertson	ssh has the highest recommendation, but does not restart (without some kind of bundle+rsync thing)	01:54
	http* restarts more by default, but is harder to use for most things. In any case, creating a bundle architecture would allow restarting more easily	01:55
jameslord	oh, if failed, all should begin from zero?	01:55
SethRobertson	clones restart from zero. pulls/pushes restart from the previous pull/push	01:55
jameslord	if that is a quite large project, for example i git clone ssh:// android, if failed, i should start from zero?	01:56
SethRobertson	yes	01:56
jameslord	that is huge amount of code:(	01:57
	← wilmoore left	01:57
jameslord	but if i use http , it need not to start from zero?	01:57
SethRobertson	You could clone it somewher elese with good network, then create a tarball from the clone, and then use a restarting protocol (rsync+ssh or http) to download the tarball	01:57
	→ boombatower joined	01:58
jameslord	the problem is if i want to use ssh + rsync, the git server should support rsync, right?	01:58
	← j416 left	01:58
	← baburdick left	01:58
jameslord	not all remote repo support resyc, anyhow	01:58
sitaram	ShadeHawk: it's now 7:29am here in India	01:59
karstensrage	is there a way to have something committed in my local repo pushed to one remote but not to another	01:59
jameslord	10:00 am here:)	01:59
cbreak	jameslord: just get a non-sucky network	01:59
	← Araxia_ left	01:59
SethRobertson	jameslord: None of this stuff can be done with a standard git server. It is all stuff you must do with a shell account	01:59
	→ baburdick joined	02:00
SethRobertson	karstensrage: Yes, you do not push the branches with that commit to the remote you do not want that commit on	02:00
	jameslord: So for instance you could fire up an amazon instance, clone your thing, then rsync (with restart) to the amazon instance. Shut down the instance nad use it directly from then on	02:00
	→ dcarper1 joined	02:01
signal-9	SethRobertson: much respect http://jsudlow.github.com/	02:01
	→ mxweas joined	02:02
karstensrage	SethRobertson, hmm thats an interesting though, how would that be set up usually? like if i have a working branch that is for my own development and my own repo and then another branch that is for github only, what would that workflow look like?	02:02
SethRobertson	signal-9: nice	02:02
karstensrage	s/though/thought	02:02
	→ txomon\|home joined	02:02
sitaram	ShadeHawk: I had seen that before; read it all through too (I think shortly after you first posted it)	02:02
SethRobertson	karstensrage: Is this like a configuration file thing? If so then see ideas referenced in !config_files	02:03
	← ASIRINeiluj left	02:03
SethRobertson	!configfiles	02:03
gitinfo	the recommended way is to change your application so that you can (perhaps optionally) store site-specific configuration/data in separate files, then keep those files out of version control. Still, see https://gist.github.com/1423106 for ways people have worked around the problem.	02:03
jameslord	SethRobertson now i am using DH host:)	02:07
	→ nadavoid joined	02:08
	← amigojapan left	02:08
	→ macabre joined	02:10
	→ cccaldas joined	02:11
	→ BiggFREE joined	02:12
	← steffo left	02:15
karstensrage	SethRobertson, this is actual code, unit test .java files	02:15
	← [M]ax left	02:17
	→ kevwil_ipad joined	02:17
	→ [M]ax joined	02:18
	→ ttuttle joined	02:19
ttuttle	hi guys	02:19
	I have my repo set up with four branches:	02:20
	upstream, feature-upstream, local, and feature-local.	02:20
	I have developed a feature in feature-upstream, on top of upstrea.	02:20
	I would like to port those commits over to feature-local, on top of local.	02:20
	No matter what I feed to rebase, it always does something wrong (usually mangling feature-upstream).	02:21
	← ISF_ left	02:21
SethRobertson	ttuttle: Do you want to copy or move?	02:21
ttuttle	copy	02:21
	...oh.	02:22
cbreak	git rebase --onto local upstream feature-upstream	02:22
SethRobertson	Are there other changes in feature or upstream which you do NOT want on local?	02:22
	← CollinPDX left	02:22
ttuttle	SethRobertson: I want to copy only upstream..feature-	02:22
	er	02:22
	upstream..feature-upstream	02:22
	→ CollinPDX joined	02:23
	← cccaldas left	02:23
cbreak	merge that into the branch you want and restore it if you want.	02:23
SethRobertson	ttuttle: I recommend `git cherry-pick` then, though a rebase of a copy of feature-upstream might work as well.	02:23
ttuttle	I dont want to merge.	02:23
cbreak	you want	02:23
	it'll be ff	02:23
ttuttle	?	02:24
	if it weren't for reflog I'd've thrown this machine out the window by now	02:24
cbreak	just do what I told you	02:25
	or if you want, make a temp copy of the branch	02:25
	and rebase that	02:25
	then you don't have to restore it after the merge but can delete it	02:25
ttuttle	"just do what I told you" is how one ends up not understanding git	02:25
	I want to use rebase, and I want to know how to get it to do this right.	02:25
cbreak	read the man page if you want to understand it	02:25
ttuttle	I know enough about merge to know that I don't want it.	02:26
cbreak	you want it	02:26
	obviously you don't know it enough :)	02:26
	as I said above, it'll be fast forward	02:26
	rebase --onto the branch you want it to be based on	02:27
	then merge into that branch	02:27
ttuttle	okay, so, hold on a sec	02:27
cbreak	you can go the manual-labor route and cherry-pick the whole thing	02:28
	← ShadeHawk left	02:29
	← baburdick left	02:29
ttuttle	shouldn't this be a single rebase?	02:29
cbreak	it is...	02:29
	+ a merge	02:29
ttuttle	why do I need the merge?	02:29
	I don't get that.	02:29
cbreak	rebase changes a branch by basing it on top of something	02:29
	← matsebc left	02:29
ttuttle	right	02:29
cbreak	you want the result to end up in a different branch	02:29
	so you merge the result into that different branch	02:30
ttuttle	right, but Idon't want to modify the original either	02:30
	← kerrick left	02:30
cbreak	that's why you make a temp. copy of the branch	02:30
ttuttle	maybe rebase can't actually do this	02:30
cbreak	or restore it after the merge	02:30
ttuttle	oh	02:30
	ugh.	02:30
cbreak	as I wrote above	02:30
	→ baburdick joined	02:31
	← kevwil_ipad left	02:31
cbreak	you know, cherry-pick does work... and it does support range. you're probably able to get a very similar result	02:32
	← tystr left	02:32
ttuttle	what yeah	02:32
	I er	02:32
	er	02:32
	yeah	02:32
	cherry-pick should do it	02:32
	i just thought rebase could, too	02:32
	I am fuzzy on what exactly is going wrong	02:32
cbreak	it can	02:32
ttuttle	no, I mean without any extra steps	02:32
cbreak	as I described above :)	02:32
	rebase changes one branch only	02:32
SethRobertson	cherry-pick uses rebase internally, according to my imagination anyway	02:33
ttuttle	SethRobertson: it's the other way around for me	02:33
cbreak	I read it's the exact opposite	02:33
	rebase uses git apply or git cherry-pick	02:33
	(-i uses the latter)	02:33
ttuttle	okay, holy fuck, I got it to do what I want.	02:34
	git checkout feature-local	02:34
	git reset --hard feature-upstream	02:34
	git rebase --onto local upstream	02:34
	(apologies if that is isomorphic to what you just suggested, cbreak)	02:35
SethRobertson	This provides a command equivalence between the two for your sort of problem. Note that the cherry-pick range problem was eliminated	02:35
	http://stackoverflow.com/questions/1994463/how-to-cherry-pick-a-range-of-commits-and-merge-into-another-branch	02:35
cbreak	that's more or less what I told you, just a different order :)	02:35
ttuttle	okay :)	02:35
	yeah, the merge thing didn't make sense to me	02:35
cbreak	and potentially you lose some commits that way	02:35
ttuttle	I am allergic to merge, because it leaves merge commits, which I don't want	02:35
	this way? how?	02:35
cbreak	because you didn't merge	02:35
ttuttle	blink	02:35
	I don't understand why that matters?	02:36
	it's still cherry-picking all of them under the hood, no?	02:36
cbreak	you nuked changes from feature-local that aren't in local	02:36
ttuttle	oh, that's fine	02:36
	I didn't want them	02:36
	the point here was "I want feature-local to be upstream..feature-upstream, but based on local"	02:36
	→ dorisabayon joined	02:36
ttuttle	anyway, thanks for the help!	02:37
cbreak	normally, you'd have lots of feature branches, not one :)	02:37
ttuttle	feature-upstream is where I developed it	02:37
	cbreak has a few new ones every week	02:37
ttuttle	feature-local is just where I want it	02:37
	oh, yeah	02:37
	I mean, I have a few others	02:37
	but most are developed against local.	02:37
	→ disappea_ joined	02:38
	← apok left	02:38
cbreak	ttuttle: you should take a look at man gitglossary	02:39
gitinfo	ttuttle: the gitglossary manpage is available at http://jk.gs/gitglossary.html	02:39
cbreak	the fast-forward section	02:39
	← Fandekasp left	02:40
	← cannonball left	02:40
	← kukks left	02:41
	← oriba left	02:42
	→ kerrick joined	02:42
	→ Vampire0 joined	02:44
	← nadavoid left	02:46
	← Vampire0_ left	02:46
	← BiggFREE left	02:48
	← ttuttle left	02:49
	← boombatower left	02:50
	← eletuchy left	02:51
	← mdel left	02:52
	← baburdick left	02:53
	← brainbox1100 left	02:56
	← _Vi left	02:59
	→ [RHQme]stringo0 joined	02:59
	→ _Vi joined	03:00
	← stringoO left	03:02
	→ boombatower joined	03:02
	← checkers left	03:07
	→ ISF_ joined	03:07
	→ riyonuk joined	03:08
	→ mjessome joined	03:08
	→ xerora joined	03:09
riyonuk	Okay, so I'm having some trouble understand some things. Git stores settings on a per-project basis, right? In the .git folder, of whatever folder you just cloned. So where are the "git --global settings" stored? /usr/bin?	03:09
bremner	$HOME	03:11
	$HOME/.gitconfig	03:11
	riyonuk: see man git-config ENVIRONMENT	03:12
gitinfo	riyonuk: the git-config manpage is available at http://jk.gs/git-config.html	03:12
	← _Vi left	03:14
	→ igotnolegs joined	03:14
	← crisp left	03:15
riyonuk	bremner: Ahh, thanks. Just did a "cat .gitconfig" and saw all my settings. Thanks:)	03:15
	← linusoleander left	03:15
	→ _Vi joined	03:15
riyonuk	So git is pretty much built for one person to be using that machine, right?	03:15
	Like, it would be hell if multiple people used my laptop, as they'd have to re-configure their git settings?	03:16
bremner	riyonuk: huh?	03:16
	would all of your users share the same home directory?	03:16
	→ simplechat joined	03:16
riyonuk	Like, if my friend wanted to write some code, using my macbook, how would he do that? So he too could push to github, under his name though.	03:16
	← simplechat left	03:17
	→ simplechat joined	03:17
	← metcalfc left	03:17
	← mjessome left	03:17
	→ dotnull joined	03:17
bremner	I still see what the problem is. If you have settings to be shared between users, put them in the system config. If not, put them in the users home directory.	03:18
	s/see/don't see/	03:18
	← Guest85118 left	03:18
riyonuk	Yea, that would work.	03:20
	← milk left	03:22
	→ milk joined	03:22
	→ dys` joined	03:24
	→ ObseLeTe joined	03:24
	→ koo4 joined	03:25
	← kerrick left	03:26
	← dys left	03:26
	→ elisa87 joined	03:26
	← mloskot left	03:29
	← fayimora left	03:30
	→ truc joined	03:31
	← ObseLeTe left	03:31
	← truc left	03:31
	→ fayimora joined	03:31
	→ curtp joined	03:32
	← elisa87 left	03:32
curtp	my master (remote) branch is stuck on a commit. I can see the changes in the remote branch, but I can't figure out how to pull it local. Any pointers?	03:32
	→ apok joined	03:33
curtp	it is on an old commit from what I can tell. there are a few more commits that came in after it, and I'd like to be on the most recent commit	03:33
	← detaer left	03:34
	← curtp left	03:35
	← Sigma left	03:36
	← boombatower left	03:37
	→ truc12 joined	03:37
	← hyperair left	03:38
	→ keller joined	03:39
	← milk left	03:39
	→ milk joined	03:40
	← milk left	03:40
	← disappea_ left	03:41
	→ milk joined	03:42
	← flaguy48 left	03:48
	→ dr0id joined	03:50
	← orafu left	03:53
	→ orafu joined	03:53
	→ angelsl joined	04:00
xiong	I have this idea, which I admit is quite vague. I'd like to know if anything in the git community reflects it. It has to do with multiple related branches that remain in defined relationships to one another.	04:05
	← dorkmafia left	04:05
	← riyonuk left	04:06
xiong	For example, on one branch there might be a set of files in foo/; and in a second branch, the same files in bar/baz/. I'd like to say that when any such file is changed on either branch, switching to the other branch will immediately find the change made in the right place.	04:07
	→ tshauck joined	04:07
	← truc12 left	04:08
	→ dirkle joined	04:08
	→ tellone joined	04:09
	→ gitnewbie joined	04:16
	← tellone left	04:16
	gitinfo set mode: +v	04:17
gitnewbie	My remote master branch seems to be stuck on an old commit. When I try to pull the latest commit pushed by someone else, it isn't pulled down. git status shows the diff and I can see their changes, but for some reason when I pull i get the "already up to date" message.	04:17
	→ beautifulmind joined	04:17
gitnewbie	how can move the master branch to point to the most recent commit?	04:18
SethRobertson	gitnewbie: Do you want to save anything which you have done?	04:18
gitnewbie	I don't have any local changes	04:19
	if that is what you mean	04:19
SethRobertson	If you are quite sure of that, `git checkout master; git reset --hard @{u}` will zap the master branch and have it point to what your upstream is pointing to	04:19
	← Samual left	04:21
gitnewbie	SethRobertson: I tried your command and got the same "already up-to-date" message.	04:21
SethRobertson	There were two commands	04:21
	← mxweas left	04:22
SethRobertson	Neither should have produced that message	04:22
	Did you run a third command?	04:22
gitnewbie	the git reset --hard command is what caused it	04:22
	git pull	04:22
	sorry... that is the one that produced the message	04:22
SethRobertson	You should be up to date at this point, yes. What makes you think you are not?	04:22
gitnewbie	I looked for the change, but it isn't there.	04:22
	Oh.... when I do a git status, it shows me a diff that displays the changed file	04:23
	sorry..... git show	04:23
SethRobertson	git show what?	04:23
	no arguments?	04:24
	tizzo-afk → tizzo	04:24
gitnewbie	correct, no arguments	04:24
	← tshauck left	04:24
	← dreiss left	04:24
	→ tellone joined	04:24
gitnewbie	the diff shows the commit sha that was displayed when I did the reset --hard. then it goes into a diff display that shows the new code that is out there	04:25
	→ gavin_huang joined	04:25
gitnewbie	I had the person who did the new commit send me the output of 'git reflog' and it shows a different commit sha as the first entry	04:26
SethRobertson	If `git show` with no arguments show the diff, the diff is in your repository	04:29
gitnewbie	my local repository or the remote repository?	04:30
SethRobertson	local	04:30
	→ eletuchy joined	04:30
SethRobertson	The same repo which run `git show` has the commit	04:30
gitnewbie	ok.... checking again	04:32
	← tellone left	04:34
gitnewbie	SethRobertson: you're right... it is there! Thanks for helping me out	04:34
SethRobertson	sure	04:35
gitnewbie	so.... can you give me a quick explanation of the git reset --hard @{u} command?	04:35
	or, is there a link I can go read up on it?	04:35
SethRobertson	`man git-reset` @{u} is the shortcut for the remote tracking branch for your upstream	04:37
gitinfo	the git-reset manpage is available at http://jk.gs/git-reset.html	04:37
gitnewbie	thanks	04:37
	← gitnewbie left	04:39
	← nardev left	04:40
	→ Moult joined	04:43
Moult	i've read here and there that gitosis is deprecated in favour or gitolite. i thought i'd better pop in here and check. is this true? anything else i should know?	04:44
	← neurodrone left	04:45
	→ alexim joined	04:45
SethRobertson	!gitosis	04:47
gitinfo	gitosis is no longer maintained and supported by the author; we usually recommend gitolite instead which has much better documentation and more features: http://github.com/sitaramc/gitolite	04:47
SethRobertson	Other things you should know? Plenty. Perhaps generically read !best_practices	04:47
gitinfo	There is no one right answer for git best practices, but a consensus from #git is available at https://gist.github.com/1540906	04:47
	← unreal left	04:48
	← berserkr left	04:49
	→ amigojapan joined	04:51
	→ amigojapan_ joined	04:52
Moult	SethRobertson: thanks. that's a very nice document as well for best practices :)	04:52
	→ nardev joined	04:52
	→ disappea_ joined	04:53
	→ wilmoore joined	04:54
	→ nadavoid joined	04:57
	→ Samual joined	05:02
	← dotnull left	05:02
	← lonewulf` left	05:10
	← beautifulmind left	05:11
	← guampa left	05:11
	← cilly left	05:11
	← edude03 left	05:12
	← nadavoid left	05:12
	← uu left	05:13
	→ cilly joined	05:13
	→ lonewulf` joined	05:14
	→ Chib joined	05:14
	→ unreal joined	05:16
	← jwmann left	05:17
	→ dreiss joined	05:18
	← binjured left	05:21
	← muhoo left	05:26
	← alexim left	05:26
	← Orbitrix left	05:27
tgunr	I'm having a little trouble with an ignore pattern. I have several paths like "( Xcode Projects )/MartenEngineLibFFI/( Tester )/( build )/" that i would like to ignore if "( build )" is in the path. I have tried several variations of `*/$\ build\ $' but can't cant to get it to work.	05:28
	→ Orbitrix joined	05:29
kevlarman	tgunr: no need to escape (), but you do need to quote it if it has spaces	05:31
tgunr	backslash?	05:32
	← agile left	05:32
kevlarman	no, quote	05:33
	← dcarper1 left	05:34
	→ Platehead joined	05:34
kevlarman	actually \ might work too	05:34
	← BonQuisha left	05:34
	← Platehead left	05:34
tgunr	\ doesn't work	05:34
	→ BonQuisha joined	05:34
kevlarman	more likely your * doesn't work	05:35
	if you want to ignore all folders named "( build )", that should be your pattern	05:36
	or better yet, get rid of the horrible names	05:36
	→ linusoleander joined	05:36
tgunr	can't change the names unfortunately	05:36
	→ Fandekasp joined	05:37
	→ secoif joined	05:37
	← secoif left	05:37
tgunr	duh, that worked, was trying too hard, thanks kevlarman	05:38
	→ secoif joined	05:38
tgunr	BTW< kevlar is one of my favorite materials, saved my butt once :)	05:39
kevlarman	normally i'd say rtfm	05:39
	but this particular one is a bit convoluted	05:39
	← dorisabayon left	05:40
tgunr	yeah, i hate his naming conventions	05:40
	→ nadavoid joined	05:42
	→ bobdobbs joined	05:43
tgunr	I was thrown off by this also "A pattern containing shell globbing characters, such as an asterisk (), is expanded as a shell glob pattern." presumed the need for leading	05:44
bobdobbs	I've got a dir with a bunch of subdirs in it. All the subdirs contain cloned git projects. Mostly javascript libraries and extensions and whatnot	05:44
	Whats the most economical way to update all of them.	05:45
	Instead of, say, entering each dir one by one and issueing 'git pull'.	05:45
tgunr	might take a look at gitslave or gitsubtree	05:46
bobdobbs	cool. will check them out.	05:46
	thanks	05:46
	← eletuchy left	05:46
	→ eletuchy joined	05:47
	← nadavoid left	05:47
	← eletuchy left	05:53
shuffle2	why does git sometimes automatically create a merge commit when i'm on master and `git pull`, even if i have no changes?	05:55
CareBear\	shuffle2 : your master has nothing to do with origin/master	05:57
	shuffle2 : being on master is not magical in any way	05:57
	shuffle2 : master is just like any other local branch	05:57
kevlarman	shuffle2: you probably do have changes	05:58
shuffle2	no there are no changes	05:58
CareBear\	shuffle2 : and a merge commit is created when your master can not be fast-forwarded to fetched commits	05:58
kevlarman	shuffle2: if you haven't done anything since that pull, it's pretty easy to figure out what they were	05:58
CareBear\	look at the commit graphs and individual commits	05:58
kevlarman	^or that	05:59
shuffle2	i mean i can do git pull...<time passes, people commit to origin/master>...i pull again, and a merge commit is generated	05:59
	← JustMozzy left	05:59
shuffle2	no local changes	05:59
Goplat	shuffle2: probably your own "master" branch has committed changes that aren't on the remote and thus aren't in remote/origin/master	05:59
CareBear\	now you're speculating	05:59
	look at what you have	06:00
shuffle2	lol	06:00
CareBear\	the actual commits you have, that is	06:00
shuffle2	Goplat: no	06:00
CareBear\	pastebin git log --graph	06:00
shuffle2	i've already reset to HEAD~1 to fix it	06:00
CareBear\	look in git reflog	06:01
	and run git log --graph on the merge commit	06:01
shuffle2	http://pastie.org/private/18ego2u4v26lnypcppdi1w	06:02
	the git reflog says	06:03
	ded1263 HEAD@{3}: pull : Merge made by recursive.	06:03
	git show ded126345449b83133a048e340331279e5578992 lists no files	06:04
CareBear\	are you running windows by any chance?	06:05
	→ edude03 joined	06:05
shuffle2	yes, i have core.autocrlf=true globally	06:05
CareBear\	that could play in	06:05
shuffle2	i have a hunch that it's because people do things like this sometimes: http://code.google.com/p/dolphin-emu/source/detail?r=f9f9bb98700492f78558afa195f8be511c73f636	06:05
kevlarman	shuffle2: autocrlf=true is really dangerous	06:06
shuffle2	uh huh	06:06
	i don't think it's related to this	06:06
kevlarman	shuffle2: you would probably notice in your diffs if it was	06:06
	shuffle2: but i'd still avoid it	06:06
shuffle2	i think it's because people sometimes merge the wrong way somehow (like i just linked)	06:07
CareBear\	"I tried to rebase but it didn't work"	06:07
shuffle2	they merge master into feature branch, and then make the resulting branch master	06:07
CareBear\	ok	06:07
	so they rewrite master	06:07
shuffle2	honestly dunno how they do it :P	06:07
CareBear\	that is equivalent to you having local changes	06:07
	=> you can not fast-forward	06:08
shuffle2	would that be allowed to be pushed without -f?	06:08
CareBear\	shrug	06:08
shuffle2	i need some type of punishment for that behavior :p	06:08
CareBear\	better to not let it into the repository in the first place	06:08
shuffle2	moderating gcode-hosted projects isn't very nice	06:09
	→ rathma joined	06:09
CareBear\	well, have fun with that :9	06:09
shuffle2	i mean, it's hard to do afaik	06:09
CareBear\	I would never host at google	06:09
	so you're talking to the wrong person	06:10
	but maybe they can provide a Gerrit for you	06:10
	(for everyone)	06:10
	← btanaka left	06:10
	→ Goplat` joined	06:11
	← Goplat left	06:11
shuffle2	wow gerrit sounds magical	06:12
	← dzonder left	06:12
	← avium left	06:12
	← _W_ left	06:13
	← nardev left	06:13
	← rurufufuss left	06:13
	← Gitzilla left	06:13
	← heinrich5991 left	06:13
	→ jweinberg joined	06:13
	→ didlix joined	06:13
	→ baburdick joined	06:14
	→ spacebat joined	06:14
	← Marquel left	06:18
	← jayne left	06:19
	→ heinrich5991 joined	06:19
	← koo4 left	06:20
	→ koo4 joined	06:20
	← _Vi left	06:24
	→ Gitzilla joined	06:25
	→ GutenTag joined	06:26
	→ nardev joined	06:26
	→ _W_ joined	06:26
	← amigojapan_ left	06:26
	→ rurufufuss joined	06:26
GutenTag	I just did a 'git add .' -- and there are about 18K files -- should I expect this to take a very long time? I've been waiting about 5 minutes for the operation to complete and it's still going. I'm under cygwin and on an 'older' computer.	06:27
CareBear\	yes, it will take time	06:29
	GutenTag : especially if they are binary files	06:29
SethRobertson	Or larger than free memory	06:30
CareBear\	or you are on windows	06:30
	or your hardware is old	06:30
	oh wait, you already said .. :p	06:30
GutenTag	LoL	06:30
	Ok, I'll be patient then. :)	06:30
	what is git doing when it add's a file?	06:31
	← _W_ left	06:31
SethRobertson	checksums, compressing, copying	06:31
GutenTag	Ah.	06:32
	Oh, will "copies" of the files end up in the .git directory?	06:32
CareBear\	yes	06:32
GutenTag	Aha. OK -- makes sense. Thanks.	06:32
	← Carmivore left	06:33
	→ Carmivore joined	06:34
	→ _W_ joined	06:34
	← brokenshakles left	06:34
GutenTag	So, my program is dependent on a bunch of external libraries (these 18K plus files, including their .jar files, documentation, etc..). I'm creating a separate repository for them to share with other team members from our actual source code. Is this good/common practice?	06:34
CareBear\	no	06:35
	:)	06:35
	well, common I don't know	06:35
	but git isn't so good at that	06:35
	→ _Vi joined	06:35
CareBear\	IMO you should only have your own code in your repo	06:35
	and manage dependencies with a package manager	06:36
	except Java doesn't have one :(	06:36
GutenTag	So, that's why I at least created a SEPARATE repo -- does that make the most sense?	06:36
	→ AmirBehzad joined	06:37
SethRobertson	GutenTag: That is pretty common. git-submodule or gitslave is often used for that. If it is your repo, then gitslave is less annoying, if non-standard	06:37
	← joshsmith left	06:37
	→ brokenshakles joined	06:38
SethRobertson	You might want to split it up even further to prevent massive repos	06:38
	Another option is to use git-annex for those large binary files	06:39
GutenTag	SethRobertson: Thanks. I'm not familar with either; I'll google. Oh crap, it's worse than I thought, too. My lib directory is about 450M and about 28K files. Does that constitute a massive repo?	06:40
	← dreiss left	06:40
GutenTag	We are using bitbucket at the moment, if that matters.	06:40
SethRobertson	I have a 1.7G one which I am not proud of	06:40
	I had to put special config options in to make it work	06:41
	quickly	06:41
CareBear\	GutenTag : binary megabytes are bad	06:41
	:)	06:41
	→ caseymcg joined	06:42
	← _Vi left	06:42
GutenTag	Should we not use git at all then to share our libraries? It still seems like one of the best options.	06:42
CareBear\	I think something else would be better	06:43
GutenTag	my add . finally finished, now my commit is taking it's time. Can't wait to push. :)	06:43
SethRobertson	It is a tough issue. It is there and convenient, but not really designed for the purpose	06:43
GutenTag	I see.	06:43
CareBear\	except it really isn't convenient	06:43
	with so much/binary data	06:43
	→ _Vi joined	06:43
GutenTag	It seems like it should be convenient enough, once every gets their initial copies. I don't /think/ it will grow that fast. (Of course, those are probably 'famous last words')	06:44
	→ bindaasomatic joined	06:44
	rphillips → Guest66160	06:45
CareBear\	I would use a real package manager	06:45
GutenTag	I'm pushing now. I'll see how long it takes to push.	06:45
CareBear\	du -s .git	06:45
	maybe -hs	06:46
GutenTag	CareBear\: Ok, I don't know about package manager, but it sounds like something I should look into, regardless, to make an informed decision.	06:46
	It said it had 14509 objects when I sent my push command.	06:46
	Do you have a recommendation for (Java) package managers? I'm not sure where to start.	06:46
CareBear\	oh you're on windows yeah, no, no package manager in sight there	06:46
	the only instance so far I've seen it done halfway sanely is Gentoo Linux	06:47
	but unfortunately the world of java developers insist on shipping all dependencies in every package	06:47
	so the opposite of the scenario with a successful package manager	06:48
	← _Vi left	06:48
CareBear\	..of course because there exist no useful deployment tools (package managers) for java	06:48
	small glitch in the portability	06:48
	sorry	06:48
	I really think java sucks in so many ways :p	06:48
	I will try to shut up now	06:48
	← adamm left	06:50
	→ dreiss joined	06:51
	← apok left	06:52
	← Goplat` left	06:53
	→ avium joined	06:53
	Guest66160 → rphillips	07:00
	→ macRover joined	07:00
	→ Blaster joined	07:02
Blaster	Hey, I just deleted a whole directory from my working folder, and now when I type 'git status' a whole ton of deletions show up. Normally I'd type git rm <path> to delete those files from the repo, but since theres hundreds of deletions, is there a command I can execute to delete the whole directory from the index?	07:03
CareBear\	git rm -r ?	07:03
	→ arvind_khadri joined	07:04
Blaster	perfect thanks	07:04
	← CollinPDX left	07:04
	→ jayne joined	07:06
	→ _Vi joined	07:09
	→ Spockz` joined	07:10
	← fermianyon left	07:12
	→ CollinPDX joined	07:12
Blaster	I have a bunch of files whos names went funny inside Git... They are now like %%E6^E66^E6665588%%results_detail_map.tpl.php	07:15
	anyone know why?	07:15
	← amigojapan left	07:15
	→ mdel joined	07:16
	← Fandekasp left	07:18
	← drev1 left	07:22
	→ Marquel joined	07:22
	→ ipalaus_ joined	07:25
	← koo4 left	07:28
	→ OOPMan joined	07:29
	→ delight joined	07:29
	→ ObseLeTe joined	07:32
	← Vortex35 left	07:32
	→ Vortex34 joined	07:33
Moult	possible to ask for gitweb help in here?	07:34
	← caseymcg left	07:34
	← brokenshakles left	07:35
	→ caseymcg joined	07:35
	← dhruvasagar left	07:36
	← diegoviola left	07:39
	→ fairuz_ joined	07:40
	← [RHQme]stringo0 left	07:41
	→ ThomasLocke joined	07:46
	← madewokherd left	07:46
	→ __iron joined	07:46
	→ ASIRINeiluj joined	07:47
	→ umren joined	07:47
	→ marthinal joined	07:48
	→ brokenshakles joined	07:50
	← _iron left	07:50
	← surfdue left	07:51
	→ surfdue joined	07:52
	← thebignoise left	07:52
	← dreiss left	07:52
	← OOPMan left	07:54
	← ObseLeTe left	07:57
	→ ripero joined	08:00
	→ koo4 joined	08:02
	← jason237 left	08:02
	← Chillance left	08:03
ripero	I'm having troubles with the setup of mirroring through gitolite, is anybody familiar with it?	08:04
	→ break57382 joined	08:05
	→ Dave^\| joined	08:06
	→ bemson joined	08:07
bemson	anyone know how to point to another repo in github? I see folders with green arrows in the file browser, and I'd like to do that instead of copying all the files into my repo	08:07
sitaram	ripero: !doesntwork	08:07
gitinfo	ripero: What happened? Did it fall asleep on your couch? Eat your homework? Destroy your hard drive? Is it making faces at you? Does it want more money? Is it sleeping with your best friend? Please be specific!	08:07
	→ psakrii joined	08:07
	← bemson left	08:11
	← brokenshakles left	08:11
	→ giallu joined	08:11
	→ Prasad joined	08:12
	← delight left	08:13
	← igotnolegs left	08:14
	← caseymcg left	08:17
	← lonewulf` left	08:18
fayimora	hey gys i just pushed something to my git repo but i wanna add more files to that commit..how do i do it?	08:20
	← giallu left	08:20
	tizzo → tizzo-afk	08:20
sitaram	fayimora: are you the only person using the repo you pushed to? if there's any chance someone else may have pulled your last commit, this might not be a good idea	08:21
fayimora	no its just me	08:21
	none is using it	08:21
	→ toabctl joined	08:22
fayimora	how do you merge 2 commits into one?	08:22
sitaram	ok then 'git add new files; git commit --amend; git push -f'	08:22
	← psakrii left	08:22
fayimora	cause i think i actually pushed to batches and it was supposed to be one	08:22
	*2 not to	08:22
	→ delight joined	08:23
sitaram	merging two commits into one: 'git rebase -i HEAD~3' then in the editor change the second 'pick' to 'squash' and save the file	08:23
ripero	sitaram: are "server keys" just regular keys (generated through ssh-keygen) available to the the gitolite user? imho, the manual is very clear about how to add the keys from other servers, but not on the meaning of "server keys" ( http://sitaramc.github.com/gitolite/mirrsetup.html )	08:23
sitaram	then in the next editor window that comes up edit the combined commit message how you like	08:23
	and save	08:23
	fayimora: ^ all that was for you	08:23
fayimora	sitaram: yeah.. when i ran the command, i got my last 5 commits	08:24
sitaram	ripero: they're not different from ssh angle; they're different in terms of how they get added to the authkeys file	08:25
fayimora	they all have pick, do i just change the last one to squash?	08:25
sitaram	ripero: the normal user keys get added with a 'command=' of /path/to/gl-auth-command	08:25
	fayimora: wait	08:25
fayimora	i have this -> pick 03cf60d Rename PostController to PostsController	08:25
	2 pick 98be1f5 Use RESTful routing	08:25
	3 pick 4d2af5e Fix resourceful behaviour of PostsController and associa ted views	08:25
	4 pick 31399c7 added comment functionality	08:25
	5 pick b3b46fb added comment functionality	08:25
sitaram	ripero: the server keys have a command= of /path/to/gl-mirror-shell	08:26
fayimora	sorry for the multiple post..nnot my doing.. i want to merge the last 2 on that list	08:26
sitaram	fayimora: please use a pastebin for anything more than 2-3 lines	08:26
	← Moult left	08:26
sitaram	fayimora: and I did say "wait"; not "paste" ;-)	08:26
fayimora	sorry bout that	08:26
sitaram	fayimora: np; my message crossed yours :-)	08:27
fayimora	so should i still wait?	08:27
	→ rendar joined	08:27
sitaram	fayimora: not sure how you got last 5 if you did 'rebase -i HEAD~3'. please delete all the lines in the editor -- every single line -- then save and exit first	08:27
	fayimora: then try again with HEAD~2 instead of HEAD~3 (my mistake on that)	08:28
fayimora	yaay i got the last 2 now..so i change the last one to squash right?	08:29
sitaram	yes	08:29
	→ ripero1 joined	08:30
	← ripero left	08:30
sitaram	then save it. You will get a new edit window that contains the combined commit message; edit that how you like and save that too. You're done.	08:30
	fayimora: note that this is a long winded way of squashing top two commits but more generic and preserves both commit messages for editing.	08:31
fayimora	sitaram: cool…thanks a lot	08:31
	→ uu joined	08:31
	→ whitman joined	08:31
	gitinfo set mode: +v	08:32
ripero1	sorry, I got disconnected at 09:26:23, is there a log of the room somewhere?	08:32
sitaram	/topic	08:32
	hmm sorry; wait	08:33
	http://tinyurl.com/gitlog seems to work but is not in the /topic; no idea why	08:33
ripero1	sitaram: thx	08:34
	sitaram: the difference in the authorized_keys is clear	08:34
	→ Fandekasp joined	08:35
	→ khmarbaise_ joined	08:35
	← khmarbaise left	08:35
	khmarbaise_ → khmarbaise	08:35
ripero1	sitaram: my point was that it's not that clear that where you have to put the private key	08:35
sitaram	guess?	08:36
	I'll be back in 10 min	08:37
ripero1	~gitoliteuser/.ssh/	08:37
	→ dorisabayon joined	08:40
	→ elena1986 joined	08:40
	← Fandekasp left	08:40
	→ ObseLeTe joined	08:41
	← lausi left	08:42
	→ lausi joined	08:43
	← slide left	08:43
sitaram	yup; assuming that's your gitolite server userid	08:45
	→ ph^ joined	08:45
sitaram	(typically 'git' or perhaps 'gitolite' if you used an RPM or DEB install, but could be anything really)	08:45
	ripero1: ^	08:46
	← yshh left	08:46
	← [M]ax left	08:46
ripero1	sitaram: yep	08:47
	→ pidus joined	08:48
	→ guns joined	08:48
	← arvind_khadri left	08:49
ripero1	sitaram: just that "ssh" + "server" has a tendency to make one think in /etc/ssh (at least that was my first thought, which I of course discarded)	08:49
	→ [M]ax joined	08:49
ripero1	sitaram: I would kindly suggest adding a few words to the manual explanation, in the like of "Generate ssh keys on each machine + for the gitolite server user"	08:51
	→ kevwil joined	08:51
ripero1	sitaram: and thank you very much for your help	08:52
sitaram	you're welcome!	08:53
	→ OOPMan joined	08:53
	← delight left	08:54
sitaram	I used to jump at every complaint and change my docs, till I realised that everyone has a different perspective and English != perl	08:54
	not precise enough	08:54
	so now I will wait for more than one complaint before changing something ;-)	08:54
ripero1	sitaram: ok :)	08:54
sitaram	since I'm here a lot of the time and on email, it should not be an issue	08:55
	← ObseLeTe left	08:55
	← macRover left	08:55
ripero1	sitaram: that's really helpful, indeed	08:55
IcePic	ripero1: I can somewhat see sitarams point here also, gitolite is simply riding on top of ssh, so teaching ssh in all its gory details may not be optimal in those docs either. Getting some familiarity with ssh itself may sometimes be in order.	08:56
	→ ManDay joined	08:56
sitaram	well, one of the reasons I wrote gitolite was because the author of gitosis would not even reply to emails. I have no intention of becoming like that :)	08:56
ManDay	I'd like to find all objects which correspond to files '*.xyz', how would I do that?	08:56
IcePic	find . -name "*.xyz" ?	08:56
	← rendar left	08:57
ManDay	Uhm, aren't objects stored as hashes?	08:57
	→ rendar joined	08:58
	← koo4 left	08:59
	← wilmoore left	08:59
ripero1	sitaram: besides, you do know that gl-tool in Debian installs is only available through /usr/share/gitolite/gl-tool (not in any exec default path), don't you?	08:59
sitaram	ManDay: it's complex; you'll have to look at every leaf commit (commit not reachable from elsewhere) and run 'git log -- filename' on it with some format options to print the SHA, then from those SHAs you have to "git ls-tree"	08:59
	ripero1: sorry, I'm not really familiar with Debian	09:00
	→ wilmoore joined	09:00
ManDay	sitaram: Are you sure of that? Shouldn't there be some plumbing command that makes it easier?	09:01
	← EyesIsServer left	09:02
sitaram	ManDay: maybe; it's the wrong time of day for most of the real experts so you get me :)	09:02
	← OOPMan left	09:02
ManDay	Haha ;D	09:02
	Thank you sitaram :)	09:03
	→ koo4 joined	09:04
sitaram	np	09:04
	→ drizzd joined	09:06
ripero1	sitaram: not that it's not a right place for the script, just wanted to point out that gl-setup is easily availabre from shell, but gl-tool will easily return a "command-not-found" error	09:06
	→ EyesIsServer joined	09:07
sitaram	ripero1: it's not the right place for the script if you ask me	09:08
	→ lonewulf` joined	09:08
sitaram	ManDay: what's the reason you need this? (Could be an X-Y problem you know...)	09:10
	→ glcrazy joined	09:11
ripero1	sitaram: I am not really familiar with right places and Debian packaging, probably rhonda (the Debian maintainer) can help	09:11
tango_	ManDay: the hashes of the objects are the hashes of the content, and they are not really associated with any particular file _name_	09:12
	ManDay: so the only way to get which objects have had a given filename in their history is by ls-treeing all possible trees	09:12
sitaram	ManDay: for example, if you were wanting to remove a sensitive file from the repo, and that is why you're asking, there are other answers... (git filter-branch)	09:13
	← kevwil left	09:17
	← LongBeach left	09:18
ManDay	tango_: sitaram: Thanks, yes, I know filter-branch. I was rather curious about listing them. My problem is that I did not properly .gitignore some files and I wonder where and when they were involved.	09:21
	← lapistano left	09:23
	→ lapistano joined	09:23
sitaram	ManDay: should be sufficient to find them in all visible refs then; no need to go digging in reflog, lost-and-found, etc.	09:24
	→ whyking joined	09:26
	→ FernandoBasso joined	09:26
	← break57382 left	09:26
	← Prasad left	09:30
	→ sattu94 joined	09:32
	← Silowyi left	09:33
sitaram	ManDay: hah!	09:33
	got it	09:33
	→ cli joined	09:35
sitaram	git log --raw --abbrev=40 --format=%n --all -- path/to/file \| egrep -o '[a-f0-9]{40}' \| sort -u	09:37
	← cli left	09:37
	← ripero1 left	09:38
	→ __cli joined	09:39
	→ stodan joined	09:39
ManDay	sitaram: I think this only tracks the history of a file.	09:40
	It doesn't allow globbing all files of ending *.xyz in the history	09:40
sitaram	ManDay: oh I forgot you wanted a glob	09:40
	hmm	09:40
	← Chib left	09:40
sitaram	seems to work if you put the glob in quotes	09:41
	← _Vi left	09:42
ManDay	Well yeah, if those files are in the current working tree	09:44
	but lets say there HAD been a file called a.xzy which I deleted 10 commits ago	09:45
	← khmarbaise left	09:45
	→ aruetten joined	09:45
	← glcrazy left	09:45
	← aruetten left	09:45
sitaram	I dont think it needs to be in the work tree	09:46
	nope; works fine	09:46
ManDay	nope	09:47
	I'm 95% certain that there had been days when I had abx.xyz commited, but when I do what you said with '*.xyz' nothing comes up	09:48
sitaram	then it's not in any ref that's reachable now; maybe went into reflog or even beyond	09:48
	→ cccaldas joined	09:49
ManDay	What do you mean it went into reflog?	09:49
sitaram	see the "--all" in my command? that means all branches/tags/remote-refs right?	09:50
ManDay	Yes	09:51
sitaram	now do you know what 'git commit --amend' does?	09:51
ManDay	Yes	09:52
sitaram	right. Now what happens to the commit that used to be at HEAD?	09:52
ManDay	what?	09:52
sitaram	it's not reachable by "--all" anymore	09:52
	→ break57382 joined	09:52
ManDay	I thought it were discarded	09:52
	Besides, what does that have to do with reflog?	09:52
sitaram	well yeah, but you can get it back. Limited offer, 30 days only	09:53
	operators are standing by. Call now	09:53
	→ Sigma joined	09:53
sitaram	damn, wrong window ;-)	09:53
ManDay	Wait, whom am I calling? I WANT THAT LIMITED OFFER!	09:54
sitaram	1-800-git-log-g :-)	09:56
	jokes apart, "git log -g" shows you the reflog. See the last section of http://sitaramc.github.com/gcs/index.html for some details on reflog but it's ideal to read the whole thing to understand it	09:56
	meanwhile, try this	09:56
	the command I gave you? add "-g" right after the "log"	09:56
	and see if your files show up	09:56
	← cilly left	10:02
	→ gusnan joined	10:03
	→ arvind_khadri joined	10:03
	← arvind_khadri left	10:03
	→ arvind_khadri joined	10:03
	← edude03 left	10:04
ManDay	My files?	10:05
	Ok	10:06
	Look	10:06
	How does the reflog work? Apparently, it can't be derived from the trees, can it?!	10:06
	← gavin_huang left	10:06
CareBear\	it can not	10:07
	it is a simple append only log of commits that are created	10:07
sitaram	ManDay: did they show up? those *.xyz files?	10:07
ManDay	CareBear\: So it's created and maintained separately of the rest of git?	10:09
CareBear\	yes	10:10
ManDay	sitaram: Show up where?	10:10
	Thanks CareBear\	10:10
	← elena1986 left	10:10
	→ unmanbearpig joined	10:11
	← __cli left	10:11
	← arvind_khadri left	10:11
	← cccaldas left	10:11
	→ sebrock joined	10:11
sitaram	in this command: git log -g --raw --abbrev=40 --format=%n --all -- "*.xyz"	10:11
	→ nixmaniack joined	10:12
ManDay	sitaram: No.	10:12
sitaram	then you dont have them in your repo	10:12
	← nixmaniack left	10:14
	→ arvind_khadri joined	10:14
	← arvind_khadri left	10:14
	→ arvind_khadri joined	10:14
ManDay	Hm, you are perhaps right... I just checked on of the commits where I thought I had them in, nothing there.	10:15
	there are all sorts of 95% ...	10:15
	→ __cli joined	10:15
	← __cli left	10:16
	→ __cli joined	10:17
ManDay	No	10:18
	Here, I found a commit in which is a file abc.xyz	10:18
	sitaram.	10:18
	I told you, that command does not account for files in history. It only shows the history of a file currently in the WD	10:19
	→ Bartzy joined	10:20
sitaram	ManDay: no. If yo u don't want to listen thats ok; but I have tested it with files not in the WD	10:20
ManDay	git log --raw -g --abbrev=40 --format=%n --all -- '*.cgns'	10:23
	And I have tested it, too.	10:23
	And for me it doesn't work.	10:23
	But if you don't want to listen, that's ok. Let's just both not listen to eachother.	10:23
	→ flavius joined	10:24
sitaram	ManDay: complete transcript of a test; empty directory, from scratch: http://dpaste.org/P3amk/	10:24
ManDay	That's most likely because you got commits dedicated to those files.	10:25
	I can give you a log from me, too	10:26
sitaram	what do you mean "commits dedicated to those files"?	10:26
	a commit in which only that file is committed? That doesnt make a difference to git	10:27
	← dvide left	10:28
	→ schnatta joined	10:29
jameslord	i need a mailinglist to post my question:(	10:29
	give me one	10:29
	← disappea_ left	10:30
	← xerora left	10:32
	→ churp joined	10:32
	← guns left	10:32
	→ airborn joined	10:32
	→ disappea_ joined	10:32
sitaram	ManDay: in case that's what you meant, try this: http://dpaste.org/IbVLm/	10:33
ManDay	sitaram: http://pastebin.com/TQAtNx0S	10:33
sitaram	temp copy of my gitolite repo, blow away everything, and see what the commands get you	10:33
	← Tommy[D] left	10:33
ManDay	try for yourself	10:33
	→ Akufen joined	10:34
	→ giallu joined	10:35
	→ Tommy[D] joined	10:36
sitaram	take out the --all if you use -g	10:37
	they're not mutually compatible	10:37
	I probably didn't say that clearly, sorry	10:38
ManDay	Still nothing.	10:38
	→ thebignoise joined	10:38
sitaram	I got the output	10:38
	http://dpaste.org/n07kH/	10:39
	→ krz joined	10:39
ManDay	goto onto origin/2d-base and try from there	10:39
	git log --raw -g --abbrev=40 --format=%n -- '*.cgns'	10:39
	gives me nothing on 2d-base	10:39
	→ tomislater joined	10:40
sitaram	view http://dpaste.org/7zhT6/raw/	10:40
	whats your git version?	10:40
ManDay	git version 1.7.3.4	10:41
sitaram	ok... is this OSX or Windows?	10:41
ManDay	neither?!	10:41
sitaram	then I dont know. My git is much newer but I dont see this as being recent	10:41
ManDay	whats your v?	10:41
sitaram	1.7.7.4	10:42
	→ tewecske joined	10:42
sitaram	wait I have a much older one at work lemme log on and try	10:42
	1.7.4.4 damn... not much older	10:42
ManDay	Wait... 7.4 isn't "much older" than 7.7 but 7.3 is?	10:43
	Logaritmic scale where needs be, eh	10:43
	:P	10:43
sitaram	damn	10:44
	it doesnt work on 1.7.4.4	10:44
ManDay	Why "damn"? That's good	10:44
	Now we know that it was a bug that must have been fixed.	10:44
	At least that's a plasuible explanation	10:44
sitaram	ManDay: yeah true... /me tries to look for release notes	10:45
ManDay	Thank you sitaram :)	10:46
sitaram	found it -- 1.7.5	10:47
	← arvind_khadri left	10:47
	→ shruggar joined	10:47
ManDay	saying?	10:47
sitaram	oh well. Sorry for being a little short. I did not think this was somehow "new" and thought I'd done it before, so until we really tested it, I never thought it could be a version difference	10:47
	wait lemme paste	10:48
	→ jceb joined	10:48
sitaram	line 71 of http://dpaste.org/ZwUP5/ (1.7.5 relnotes)	10:48
ManDay	ah	10:49
	← ipalaus_ left	10:50
	← linusoleander left	10:51
	→ Boris` joined	10:55
	← Boris` left	10:55
	→ glcrazy joined	10:55
jameslord	i git clone from a remote repo, then made changes to it on pc A , but have not push it to serer yet, then i copy a local repo to another machine B, after that i push push the change to server on PC B, then i come back to pc a, it still in a not push stage, should i push for a second time?	10:56
	but this will generate unnecessary log, and in fact push nothing to server	10:57
	what should i do?	10:57
cmn	you don't need to do anything, the change is in the server already	10:59
	do a fetch if you want to update the repo's idea of what's in the remtoe	10:59
	→ apok joined	11:01
jameslord	how to know if the local repo is the same as the remote repo?	11:01
aro	why is my commit to my repo hanging when i do 'git add .'	11:02
	← giallu left	11:03
ManDay	jameslord: by running git fetch	11:04
jameslord	git fetch will tell me if local repo is the same as remote repo	11:04
	？	11:04
	→ giallu joined	11:05
ManDay	？	11:05
cmn	git fetch will update your repo's idea of what's in the remote	11:05
aro	i dunno whats going on, it just isnt doing anything	11:05
cmn	aro: are those files very large?	11:05
ManDay	cmn: Why would it matter?	11:05
cmn	jameslord: if you don't want to download the new commits, ls-remote will tell you the state of the branches on the remote	11:06
	ManDay: IIRC git add is the one responsible for creating the blobs	11:06
	and large files have never been git's forte	11:06
	so if the file is large (esp. compared to the RAM), git add will take a long time	11:07
jameslord	there is no ls option when use git remote	11:07
	→ faber joined	11:07
cmn	ls-remote	11:07
ManDay	cmn: Does git add even care about the file's contents?	11:07
	→ cesc joined	11:07
cmn	yes	11:07
	it needs to create the blob	11:07
ManDay	Ah, yes	11:08
	← churp left	11:08
	← FernandoBasso left	11:11
	← fayimora left	11:14
	→ fayimora joined	11:14
jameslord	hi	11:15
	← jameslord left	11:15
	→ jameslord joined	11:15
	→ iocor joined	11:15
	→ dangerousdave joined	11:15
	← glcrazy left	11:17
jameslord	hi	11:18
cmn	!hi	11:18
gitinfo	[!welcome] Welcome to this channel. Go ahead and ask your question. Very few people are responding to greetings, but will respond if a question is asked.	11:18
jameslord	http://paste.ubuntu.com/795870/	11:19
	what this means	11:19
	→ FernandoBasso joined	11:19
	→ Cromulent joined	11:19
jameslord	two sha is the same:(	11:19
cbreak	so?	11:20
	→ glcrazy joined	11:20
cbreak	just means that master is "checked out"	11:20
	→ cyfi joined	11:21
cmn	it's a list of branches	11:21
	and in this case, master is the default branch for that repository	11:22
	→ khmarbaise joined	11:23
	→ canton7 joined	11:23
	← ASIRINeiluj left	11:23
	→ Textmode joined	11:25
	→ brian_g joined	11:26
	→ Cs-hard joined	11:27
	← giallu left	11:27
	← Cs-hard left	11:27
	→ giallu joined	11:28
	→ psoo joined	11:28
	← glcrazy left	11:29
	→ robotmay joined	11:33
	→ nixmaniack joined	11:36
	→ j416 joined	11:37
	→ Fandekasp joined	11:37
	← marthinal left	11:40
	→ ganesh_ariveguru joined	11:40
	← dorisabayon left	11:41
	← Cromulent left	11:44
	→ glcrazy joined	11:45
	← glcrazy left	11:46
	→ glcrazy joined	11:47
	← kermit left	11:47
	← fairuz_ left	11:47
	→ kermit joined	11:48
	← glcrazy left	11:51
	→ bigkm joined	11:51
	→ glcrazy joined	11:52
	← fayimora left	11:54
	← toabctl left	11:56
	← glcrazy left	11:57
	→ glcrazy_ joined	11:58
	→ Dave^\|\| joined	11:59
	→ yibe joined	12:00
	← Dave^\| left	12:02
	→ ngulden joined	12:04
	→ nax joined	12:08
	← nax left	12:08
	→ glcrazy__ joined	12:08
	→ d0k joined	12:10
	← glcrazy_ left	12:11
EugeneKay	Good morning, gits!	12:16
	← glcrazy__ left	12:16
sitaram	!hi	12:17
gitinfo	[!welcome] Welcome to this channel. Go ahead and ask your question. Very few people are responding to greetings, but will respond if a question is asked.	12:17
sitaram	couldn't resist ;-)	12:17
EugeneKay	At leasst somebody is sociable.	12:17
sitaram	easy when the bot is doing all the work ;-)	12:17
ManDay	I will respond to a greeting!	12:18
	Hello and welcome EugeneKay !	12:18
canton7	i'm impressed that the bot picked up on ! hi... good work!	12:19
EugeneKay	I think that means jast implemented the aliases I asked about.	12:19
canton7	ah, it has aliases	12:20
	haha, nice	12:20
EugeneKay	!triggers	12:20
gitinfo	See http://jk.gs/git/bot/trigger.php	12:20
EugeneKay	Yup. Nifty.	12:20
sitaram	yeah I think hi and hello were the first aliase (to welcome)	12:21
	aliases*	12:21
	→ LongBeach joined	12:21
ManDay	!pizza	12:21
gitinfo	DOUGH -- 200 g saw dust, 800 g flour, 590 millichopins water, 40 g yeast, 1 pinch of salt, 1 tbsp olive oil, 1 tbsp camphor. SAUCE -- 1 can minced tomatoes, 118 millichopins petroleum ether, 50 g laburnum (dried), 50 g repulsion gel, salt/pepper. TOPPINGS -- fish-shaped sediment, 100 g zombie flesh, 20 g rhubarb, 200 g cheese. Bake at 130 °Rø for 103.97 Sothic nanocycles. Enjoy!	12:21
	→ Shusshu joined	12:21
	← preflex left	12:21
canton7	glad that's still around :)	12:21
sitaram	yeah that one's nice too, though I like ! doesntwork best	12:22
ManDay	you mean	12:22
	!work	12:22
gitinfo	Sorry to hear it doesn't work. What did you want it to do? What happened instead? Did you get any error messages?	12:22
ManDay	:P	12:22
sitaram	!doesntwork	12:22
gitinfo	What happened? Did it fall asleep on your couch? Eat your homework? Destroy your hard drive? Is it making faces at you? Does it want more money? Is it sleeping with your best friend? Please be specific!	12:22
	→ preflex joined	12:22
ManDay	Oh :)	12:22
	→ snearch joined	12:24
sitaram	whoever wrote that deserves a pizza	12:24
	or beer	12:24
	hmm.... why is this text called "fortune"? !fortune	12:28
gitinfo	I'm sorry that you are stupid, ugly, and not using git. Still, some misguided souls think that something other than git might be better for them. Who knows, it might even be true for you. In any case, we don't care. Good-bye.	12:28
	sitaram doesn't get the connection	12:28
	← apok left	12:29
cmn	seeking fortune somewhere else maybe?	12:30
IcePic	"fortune" is the old name for the "give me a randomized text from a file with mostly witty sayings" if that is a hint.	12:31
cmn	old name? fortune still exists	12:31
canton7	incidentally, having proper fortunes in gitinfo would be awesome for those slow days	12:32
	IcePic, which in turn comes from fortune cookies? Or is it the other way around?	12:32
cbreak	fortune cookies are just a specific means to deliver fortunes	12:32
sitaram	IcePic: yeah but this one doesn't change	12:33
cbreak	there are others, like overpriced "premium" phone numbers or archaic unix services	12:33
EugeneKay	I'd like to see a !vend tied to http://itvends.com	12:33
cbreak	!blend?	12:34
IcePic	canton7: the first would be right, I gather	12:34
	fortunes as commit messages if the user doesnt supply one. =)	12:34
	"Man who kisses blowfish gets free lip piercing"	12:35
cbreak	that sounds... evil.	12:35
canton7	fortune \| cowsay is probably my favourite linux tool combination	12:36
	→ linusoleander joined	12:37
	→ vdv joined	12:38
	← canton7 left	12:39
	← flavius left	12:39
	→ nikuyoshi joined	12:41
	→ munichlinux joined	12:42
Mikachu	fortune \| cowsay \| figlet	12:43
	← schnatta left	12:43
	← linusoleander left	12:43
cbreak	OS X has a command line tool "say". It's great. :)	12:43
	you can make the computer of your little sister talk to her via ssh :)	12:44
	... not that I would ever do that...	12:44
EugeneKay	git config commity.empty=true\|false\|fortune	12:45
	-y	12:45
	→ QaDeS joined	12:45
	→ Cromulent joined	12:46
j416	lol	12:46
	→ schnatta joined	12:49
	← wilmoore left	12:49
	← Cromulent left	12:51
	→ sudoquai joined	12:51
matjas	cbreak: https://github.com/mathiasbynens/dotfiles/blob/master/.aliases#L88-91 not that I would ever use any of those…	12:51
	→ Jackneill joined	12:51
	← Jackneill left	12:52
	→ Jackneill joined	12:52
	→ ripero joined	12:53
	← break57382 left	12:53
cbreak	naturally... :)	12:54
	← ripero left	12:54
	→ ripero joined	12:54
	← ripero left	12:57
	→ spaceonline joined	12:58
	→ Chib joined	13:02
	← dangerousdave left	13:03
	← ganesh_ariveguru left	13:05
	→ NiklasFiekas joined	13:06
EugeneKay	matjas - one alias I use a bunch is cdc='cd && clear'	13:07
	→ ganesh_ariveguru joined	13:07
	→ Spockz joined	13:11
	← Spockz left	13:11
	→ ikpL joined	13:11
ikpL	Hello Everyone, I want to learn about git	13:11
	what is the best book for git	13:11
cmn	!book	13:12
gitinfo	'Pro Git' is a complete book about git that is available online for free (there's a print version, too): http://progit.org/book/	13:12
ikpL	great thanx gitinfo	13:12
	→ break57382 joined	13:13
cbreak	the bots! they steal our jobs!	13:14
	→ Spockz joined	13:14
	← vdv left	13:15
NiklasFiekas	:D	13:15
ikpL	is gitinfo a bot ?	13:15
FauxFaux	We should lynch that jsat guy.	13:15
NiklasFiekas	yep. and cmn made him say that with "!book"	13:15
cmn	!gitinfo	13:15
gitinfo	I am an IRC bot which responds to certain keywords to provide helpful(?) information to humans. Please see http://jk.gs/git/bot for more information about how to use me.	13:15
	← ganesh_ariveguru left	13:16
	← gusnan left	13:16
	→ ganesh_ariveguru joined	13:17
	← ganesh_ariveguru left	13:17
ikpL	Also I am a noob to IRC, I didn't know that	13:19
_ikke_	Bots are quite common on IRC	13:19
wereHamster	you're yourself one, _ikke_, admit it!	13:19
cbreak	basic infrastructure :)	13:20
_ikke_	wereHamster: Bots are quite common on IRC	13:20
ikpL	bots useful	13:20
	← psoo left	13:21
_ikke_	wereHamster: Calling me a bot is an insult to my intelligence and sophistication	13:21
wereHamster	we have a smart bot here, he could even pass the turing test :)	13:22
_ikke_	How may I help you	13:23
	← sebrock left	13:24
cbreak	I am not a robot, I am a unicorn	13:24
EugeneKay	ikpL - if you want an intro to Version Control Systems in general, I found "Version Control by Example" to be a good text. It's free at http://www.ericsink.com/vcbe	13:26
	← schnatta left	13:26
j416	matjas: scary 'undopush' you have there :)	13:26
	→ fayimora joined	13:27
	← ikpL left	13:29
	→ nicoulaj joined	13:31
	← lonewulf` left	13:32
	→ LeMike joined	13:37
	gitinfo set mode: +v	13:38
	→ selevt joined	13:39
LeMike	hi there. i like to use git as a backup-system. so I have a external hd which shall be the "master" and a usb stick containing the last 100 revisions. first: how do i clone only the last 100 revs?	13:39
EugeneKay	!backup	13:39
gitinfo	Taking a backup of a git repository is always a good idea before doing something unusual (well, it is an excellent idea as a matter of course). Usually, the best way to TACTICALLY back up a git repo is `git clone --mirror`. However, some unusual maintenance might require `tar cf repo-backup.tar repodir`. Testing in a clone is also an excellent idea.	13:39
EugeneKay	You really can't clone "the last X revisions" because any revision is dependent upon all parent revisions.	13:39
	(and that trigger isn't what i thought it was.... hrm)	13:40
Mikachu	you can, git clone --depth=99	13:40
cmn	to grab only the last 100 revisions, --depth 99 should work, almost	13:40
	but that kinda depends on what you consider "last"	13:40
	--depth will grab that amount on each branch	13:40
	and you can't tell a priori what the last 100 revs are by date	13:41
EugeneKay	That's a shallow clone, innit? With different SHAs	13:41
cmn	it's a shallow clone, but you get the same shas	13:41
	the earliest commit is a lie, though	13:41
EugeneKay	Ah, it doesn't do push/pull.	13:42
cmn	it's put there with a graft or a replacement or some such	13:42
EugeneKay	Cuz the chain of SHAs is broken	13:42
cmn	you kinda can, but it's not always obvious when you can push	13:42
	fetching should work, no probls	13:42
	well, mostly	13:42
cbreak	LeMike: go look at bup or a real backup tool	13:43
EugeneKay	But anyway. Git is not a backup tool.	13:43
cbreak	git on its own does not support proper history pruning	13:43
	things like --depth are inadequate replacements since they prevent you from validating repo integrity	13:43
	also, git has quite some problems with big binary files, which bup tries to solve	13:44
Mikachu	shallow clones predate grafts and replacements, iirc	13:44
	→ ASIRINeiluj joined	13:44
	→ neurodrone joined	13:45
LeMike	it's just small files non-binary. at weekends I like to pull / sync the stick on the hd. that's possible via sync / pull, too?	13:46
	→ Cromulent joined	13:46
cbreak	don't bother with history pruning then :)	13:46
EugeneKay	Sure, but your repo will continuously grow.	13:46
LeMike	that's okay. it wont grow that big	13:47
_ikke_	CLI version of github issues: https://github.com/vesln/issues	13:48
LeMike	i will try that on two pc's now. thanks so far ;) btw: it was svn all the time and I have my history since ages ;D but now I fell in love with git.	13:49
	← uu left	13:49
EugeneKay	_ikke_ - "Requires: node.js"	13:50
	← shruggar left	13:51
_ikke_	EugeneKay: Saw that afterwards	13:51
EugeneKay	I'm not really sure wtf the architecture of that thing is	13:52
	But if it requires node.js to make an API call against Github..... urdoinitrong	13:52
	That's the sort of thing thats implementable in bash. Or perl.	13:52
_ikke_	EugeneKay: Yup	13:53
	or python	13:53
	or any other language	13:53
EugeneKay	I know there's a bash library for JSON, which is what the github v3 api is.	13:53
Mikachu	network programmin in shell script, no thanks	13:54
	← Octalot left	13:54
EugeneKay	Mikachu - it's not even network programming, it's doable with curl	13:54
Mikachu	how is that not network? :)	13:54
EugeneKay	It's REST, not TCP stack.	13:54
cmn	hm, if shallow clones predate grafts and replacements, then I've no idea how the hell they're lying to the DAG	13:55
	→ gusnan joined	13:55
	→ DarkAR joined	13:56
EugeneKay	Poke about in git.git ;-)	13:56
Mikachu	hm, maybe they don't predate grafts, but they have explicit support	13:56
	→ eijk joined	13:56
Mikachu	Documentation/technical/shallow.txt	13:56
	→ k299 joined	13:57
cbreak	javascript?	13:57
	→ gavin_huang joined	13:57
	← k299 left	13:57
cmn	it's what the cool kids are using nowadays instead of bash, apparently	13:57
Mikachu	"Programs are written in JavaScript, using event-driven, asynchronous I/O to minimize overhead" says wikipedia about node.js	13:57
	→ ipalaus_ joined	13:57
Mikachu	which seems like an oxymoron, but whatever ;)	13:57
	→ lonewulf` joined	13:58
wereHamster	Mikachu: why oxymoron?	13:58
	→ k299 joined	13:58
Mikachu	javascript + minimize overhead	13:58
_ikke_	Mikachu: Language != implementation	13:58
cmn	the interpreters are pretty good nowadays, v8 (and thus node.js) has JIT	13:58
	← k299 left	13:58
Mikachu	afaik, it's still not possible to make more advanced programs in current javascript than you could run on a real computer 15 years ago?	13:59
wereHamster	Mikachu: define 'advanced'	13:59
	→ k299 joined	13:59
Dynom	git man	13:59
_ikke_	man git	13:59
gitinfo	the git manpage is available at http://jk.gs/git.html	13:59
cmn	for advanced in what way? you can do pretty man anything with js	13:59
	→ erichynds joined	13:59
Dynom	thanks gitinfo	13:59
cbreak	cmn: mostly malware and fluff	14:00
	thanks to the dumbest idea of web 2.0: downloading and executing code via untrusted networks	14:00
	and even more, from untrusted sources	14:00
Mikachu	"dumbest idea of web 2.0", now moving from oxymorons to tautologies :P	14:00
cbreak	and even more, even trusted sources embed untrusted sources for Advertisement	14:00
cmn	you mean you don't trust my website that promises you can win an iPad 3 if you just click on that little linik?	14:01
	link*	14:01
cbreak	java script reliance is even worse than abusing an inherent stateless pull protocol like http for interactive applications...	14:01
wereHamster	cbreak: so I take you download html pages with netcat and view them with ed, right?	14:01
cbreak	wereHamster: no. I download html pages and view them in a web browser...	14:02
	I don't just execute the code embedded in those pages	14:02
EugeneKay	Use node.js and MongoDB for web scale	14:02
cbreak	why bother?	14:02
EugeneKay	Because it's eb scale	14:03
cbreak	Firefox can scale web pages without that	14:03
EugeneKay	http://youtu.be/b2F-DItXtZs	14:03
	cbreak can't understand belgian	14:03
cbreak	flemish?	14:03
	or french I guess	14:03
	← giallu left	14:03
jast	everybody needs more web scale	14:03
	even the web isn't web scale enough	14:04
	if only they used node.js for the web, it would all be different	14:04
_ikke_	/dev/null db is webscale	14:04
jast	oh yes	14:05
	I store all of my data in it	14:05
_ikke_	I put my back-ups in there	14:05
	→ linusoleander joined	14:05
	→ crisp joined	14:06
cmn	the graphs are more interesting with the /dev/urandom db	14:08
	← linusoleander left	14:09
EugeneKay	Backup with 'dd if=/dev/sda1 of=/dev/null bs=1M'	14:09
	The null device is unlimited in size, and very fast, so your backups will be there instantly.	14:09
jast	cool, I gotta try that	14:09
cbreak	you should use /dev/random	14:10
	higher chance for getting the data back out	14:10
_ikke_	Who wants to get his data back out?	14:10
jast	hmm, 48 MB/s	14:10
cbreak	that's kind of the point of backups	14:10
jast	that's not actually very fast	14:10
	← __cli left	14:10
EugeneKay	On a serious note, while I understand the rationale behind projects like Node.js(asynchronous I/O+callbacks to handle more clients simultaneously by forking off the I/O blocking), I can't for the life of me figure out why in the turd they used JS.	14:11
	I mean, wtf?	14:11
jast	because JS is web scale	14:11
	the goal was to appeal to web developers, I guess	14:11
	because any serious developer would have realized straight away that Node.js is nothing special at all ;)	14:11
EugeneKay	'Web developers' are usually..... what's the word I'm after.	14:11
cmn	js represents a low barrier of entry	14:11
	they're excitable	14:12
jast	EugeneKay: "not developers"? :}	14:12
	two words, though	14:12
IcePic	isnt js one of the more portable languages there is right now?	14:12
jast	it's not particularly more portable than other languages	14:12
EugeneKay	In the same sense that C is portable	14:12
cmn	no, C runs on more machines	14:12
jast	the main thing about JS is that it has no standard library	14:12
	it's easy to port something without a standard library	14:12
cbreak	take a look at Lua	14:13
jast	by default, all you've got is the language, with no way to do anything like I/O	14:13
cbreak	it's actually well designed, extremely easy to embed, and fast	14:13
jast	I don't like lua, for reasons I'm not completely sure about	14:13
cbreak	maybe one of the fastest scripting languages out there	14:14
	(thanks to LuaJIT)	14:14
	→ mandric joined	14:14
matjas	j416: re:undopush, yes, very scary… but i sometimes need it. is there a better way to do it?	14:14
Mikachu	EugeneKay: in episode 2 it is revealed the author actually thinks php is good	14:14
jast	I have fairly specific requirements of a new language that I'd be willing to learn	14:14
EugeneKay	"PHP is a hell of a drug."	14:15
j416	matjas: well the problem is, what if your remote is not 'origin' and you want to undopush a branch that is not 'master'?	14:15
jast	one of them is: static typing with type inference (and optional type annotations)	14:15
j416	matjas: if you're not careful, you'll force-push something else to the master branch of origin	14:15
EugeneKay	I rather like PHP, as a website language. It's a lot more webfriendly than perl.	14:15
jast	I do quick one-off stuff in PHP	14:15
matjas	j416: true	14:15
j416	:)	14:16
	← swombat left	14:16
jast	such as the web interface for the bot :)	14:16
	→ tshauck joined	14:16
jast	but all of the backend stuff is actually done by the bot	14:16
Mikachu	every time i look up anything in php, it looks like they made it that specific way only to make it easy to get wrong and produce exploits	14:16
j416	jast: what language is the bot in?	14:16
jast	via a control connection	14:16
	perl	14:16
cbreak	jast: Lua has strong typing, but it types values and not variables	14:16
j416	perl	14:16
cbreak	which is... interesting	14:16
bremner	cbreak: sure, but it means type-inference is pretty much impossible	14:17
jast	I've got my own plans for typ inference	14:17
cbreak	it's runtime	14:17
	(and manual)	14:17
jast	+e	14:17
bremner	that's not inference, that's just observation ;)	14:17
jast	unfortunately they're part of a rather big project and I'm not really making any progress on it	14:17
	← stamina left	14:17
cbreak	well, my main language is C++	14:17
	probably one of the most strongest typed languages there is :)	14:18
	(and type names can span dozens of lines too!)	14:18
jast	so, my ideal language has static/strong typing and type inference	14:18
	and extensible syntax	14:18
cbreak	-> C++? :)	14:18
jast	C++ doesn't have type inference	14:18
cbreak	sure does	14:18
jast	example?	14:18
cbreak	I can create types from other types	14:19
	auto result = doStuff(input)	14:19
jast	type inference means I can write code without telling the compiler which type a variable is, and it automatically determines the type	14:19
	auto?	14:19
	← neurodrone left	14:19
cbreak	yes	14:19
jast	is that like C++1x or something?	14:20
cbreak	template<typename T> void swap(T const & a, T const & b)	14:20
	yes	14:20
jast	well templates OTOH are an abomination :}	14:20
cbreak	generic programming is older though	14:20
	they are the essence of type inference	14:20
	you can infer the weirdest things	14:21
	→ hyperair joined	14:21
jast	C++ is too full of pitfalls (and abyssal ones at that) for my taste	14:21
cbreak	hey	14:21
	at most 50% of C++ is pitfalls	14:21
	and half of those is from C	14:21
jast	plus I hate the STL with a passion	14:22
cbreak	you don't have to use it :)	14:22
jast	and there isn't really any reasonable alternative to it	14:22
cbreak	unless you write code for sun	14:22
jast	(don't say boost)	14:22
cbreak	Qt	14:22
jast	yeah, Qt is an alternative	14:22
cbreak	boost :D	14:22
jast	but it's kind of heavy	14:22
	Qt is the main reason I'd consider using C++ at all	14:22
cbreak	well	14:22
	compare Qt to other things	14:23
jast	yeah, sure	14:23
	← sattu94 left	14:23
jast	I liked the compilers that allowed me to link only those parts of a library that were actually needed by the program	14:24
	borland's newer pascal compilers, for instance	14:24
	← rurufufuss left	14:24
cbreak	That's not as easy as it sounds	14:25
	← DarkAR left	14:25
	← NiklasFiekas left	14:25
	← ngulden left	14:25
	← tewecske left	14:25
	← dr0id left	14:25
	← spyro left	14:25
	← warthog9 left	14:25
	← moonlightcheese left	14:25
	← frogonwheels left	14:25
	← shachaf left	14:25
	← basepi left	14:25
	← dhoss left	14:25
	← munro left	14:25
	← borodon left	14:25
	← ShooterMG left	14:25
	← djh left	14:25
	← Sinjo left	14:25
	← CodeBlock left	14:25
	← phaedrix left	14:25
	← coopsh left	14:25
	← tabakhase left	14:25
	← xoebus left	14:25
	← trodrigues left	14:25
	← cYmen left	14:25
	← Nilbus left	14:25
	← project2501a left	14:25
	← Zygo left	14:25
	← htoothrot left	14:25
	← aow left	14:25
	← mikejw left	14:25
	← Lisimba left	14:25
	← sobel left	14:25
	← tpope left	14:25
	← sente left	14:25
	→ dhoss joined	14:25
cmn	gcc + ld can do that, or should be able to	14:25
	→ aspotashev joined	14:25
cmn	you tell gcc to keep blocks separate, and tell ld to discard unused code	14:25
	→ DarkAR joined	14:25
	→ ngulden joined	14:25
	→ tewecske joined	14:25
	→ dr0id joined	14:25
	→ spyro joined	14:25
	→ moonlightcheese joined	14:25
	→ frogonwheels joined	14:25
	→ shachaf joined	14:25
	→ munro joined	14:25
	→ borodon joined	14:25
	→ ShooterMG joined	14:25
	→ djh joined	14:25
	→ Sinjo joined	14:25
	→ mikejw joined	14:25
	→ coopsh joined	14:25
	→ tabakhase joined	14:25
	→ xoebus joined	14:25
	→ trodrigues joined	14:25
	→ cYmen joined	14:25
	→ project2501a joined	14:25
	→ Nilbus joined	14:25
	→ Zygo joined	14:25
	→ htoothrot joined	14:25
	→ tpope joined	14:25
	→ aow joined	14:25
	→ Lisimba joined	14:25
	→ sobel joined	14:25
	→ sente joined	14:25
	holmes.freenode.net set mode: +vv	14:25
	→ NiklasFiekas joined	14:25
IcePic	isnt it the default when linking against static libs?	14:25
	→ warthog9 joined	14:25
	← dr0id left	14:25
	← shachaf left	14:25
	→ phaedrix joined	14:25
	→ glcrazy joined	14:25
	← sudoquai left	14:25
	→ shachaf joined	14:26
	→ basepi joined	14:26
IcePic	libfoo.a will contain a lot of .o files in a lump, then when you link, you pick out the ones needed to make your app work	14:26
	→ dr0id joined	14:26
jast	manually, right?	14:26
cmn	libfoo.a needs to have several sections, though	14:26
jast	I've never seen any project do that	14:26
cmn	I think we do this in libgit2 when you build the static library	14:27
	← glcrazy left	14:27
	← htoothrot left	14:27
	→ sudoquai joined	14:28
	← frogonwheels left	14:28
	→ flaguy48 joined	14:28
cbreak	because C++ is an unmanaged language	14:29
	you can do the weirdest crap with it	14:29
	→ frogonwheels joined	14:29
cmn	I can't find it in the Makefile now, but we did play around with those options	14:29
cbreak	inside a library, you can do things like jump to code directly	14:29
	without going over exported symbols	14:30
	← munichlinux left	14:30
cbreak	or sometimes even without doing PC relative jumps (which makes the code not position independent)	14:30
	→ htoothrot joined	14:30
cbreak	non-position-independent code is faster, but it doesn't play well with ASLR and similar tech	14:30
	and doesn't work in dynamic libs	14:31
	but appart from that, such jumps also make it extremely hard to find and eliminate dead code	14:31
	→ heoa joined	14:31
	→ glcrazy joined	14:31
heoa	sorry stupid question -- what was the debug/verbose mode in git?	14:32
	→ CodeBlock joined	14:33
cbreak	GIT_TRACE?	14:33
	well, or just -vv	14:33
heoa	cbreak: thank you.	14:34
cbreak	(they do different things)	14:35
	← iocor left	14:36
	→ dorisabayon joined	14:37
	→ ramlev joined	14:39
	← ramlev left	14:39
	← flip_digits left	14:40
	← Fandekasp left	14:41
	→ ramlev joined	14:42
	→ weeb1e joined	14:42
jast	.random	14:42
gitinfo	[]	14:42
jast	well, that is random	14:42
cbreak	password generator:	14:43
	dd if=/dev/random bs=512 count=8 \| openssl base64	14:43
	→ amigojapan joined	14:44
cmn	I wonder how many DBs would accept such a long password	14:45
	→ giallu joined	14:45
cmn	well, the long password only happens on /dev/urandom, my dd just gives up after a bit	14:45
	when I point it to /dev/random	14:45
	← ramlev left	14:45
	→ ramlev joined	14:45
cbreak	that's the linux implementation of /dev/random	14:46
	← ramlev left	14:46
	→ ramlev joined	14:46
cbreak	unlike the one in OS X, it blocks on lack-of-entropy	14:46
	← xxthink left	14:47
_ikke_	if you don't want to be blocked, use urandom	14:47
jast	.random	14:48
gitinfo	<cbreak> password generator: <cbreak> dd if=/dev/random bs=512 count=8 \| openssl base64[2]	14:48
IcePic	watch out when using /dev/random on linux, since it may give you short reads.	14:48
	instead of blocking	14:48
cbreak	cmn: asking wether a DB supports such passwords is the wrong thing to ask	14:50
	cmn: unless you want to be a screwup like Sony or Stratfor and store Unhashed Passwords or Credit Card Numbers + Verification Codes :)	14:50
IcePic	cbreak: the fines from the PCI-DSS authorities should be insanely severe if you screw up like that	14:51
cbreak	well...	14:52
	we'll see :)	14:52
	I've not heard of stratfor in a few days	14:52
	wonder if they still exist	14:52
	← brian_g left	14:52
	→ EdwardIII joined	14:52
EdwardIII	hey	14:52
cbreak	but it was even on the local (Swiss) radio news that they got compromised and lost CC Numbers	14:52
	← CodeBlock left	14:53
	→ CodeBlock joined	14:53
cbreak	so any regulative organisation should know :)	14:53
cmn	good point, the DB should only store the hash	14:53
jast	all right, guys, feel free to add random factoids or nonsense to the .random db, using .random add foobar	14:53
EdwardIII	i'm using git svn. i've received a message to let me know upstream has made some changes, now i don't want to go ahead and rebase without a backup in case i break something	14:53
cbreak	.random add heads (chosen by fair coin flip)	14:53
gitinfo	cbreak: okay.	14:53
jast	EdwardIII: if your stuff is all committed, we can restore it even if the rebase goes awry	14:53
EdwardIII	if i do git diff remotes/trunk # i don't seem to see any changes	14:53
cbreak	.random add tails (chosen by fair coin flip)	14:54
gitinfo	cbreak: okay.	14:54
cbreak	EdwardIII: you have to do git svn fetch first	14:54
EdwardIII	jast: right - would i do that by running git reflog, figuring out which was the last commit before the rebase then doing a git reset hard?	14:54
cbreak	also, git svn rebase would still have the git reflog safety net	14:54
jast	EdwardIII: that's one way to do it, yeah	14:54
EdwardIII	cbreak: do git svn fetch,then git diff?	14:54
cmn	.random add 4 (chosen by fair die throw)	14:54
gitinfo	cmn: okay.	14:54
EdwardIII	not worth making a backup branch first?	14:54
cbreak	EdwardIII: sure.	14:54
	sure.	14:55
	will only cost you 4kb or so	14:55
jast	EdwardIII: you can make one if you like. it doesn't cost anything, right? :)	14:55
cbreak	(depending on your HD Cluster Size)	14:55
EdwardIII	heh	14:55
	or maybe a tag would be more sensible	14:55
cbreak	tags == branches	14:55
	in this respect	14:55
EdwardIII	right	14:55
jast	since it's temporary anyway, it doesn't matter	14:55
EdwardIII	well, it's probably a good time for me to tag now anyway (haven't made any tags yet)	14:56
	← Cromulent left	14:57
	→ Cromulent joined	14:57
	← amigojapan left	14:57
	EdwardIII git fetches	14:57
EdwardIII	* git svn fetch	14:57
	man, any git/svn stuff always seems to take an age heh	14:57
jast	that's because svn is slow :}	14:58
EdwardIII	heh	14:58
	→ glcrazy_ joined	14:58
jast	especially if you use its http transport	14:58
	← giallu left	14:58
	→ Goplat joined	14:58
EdwardIII	i do	14:59
jast	almost everyone does	14:59
EdwardIII	i mostly use git svn for wordpress plugins i want to mess with	14:59
cbreak	git-svn isn't like normal svn that doesn't care about integrity	15:00
	or history	15:00
	and just fetches current state	15:00
	git-svn imports the whole history, so it uses way more data/interaction	15:00
EdwardIII	ah ha	15:00
	← glcrazy left	15:01
jast	yeah, normally svn just fetches updates to all files. git-svn fetches the complete history (at least the parts it doesn't have yet)	15:01
EdwardIII	git has so many, many features	15:01
cbreak	svn server admins love people who do git svn clones :)	15:01
EdwardIII	sarcasm?	15:02
jast	nonono	15:02
EdwardIII	heh	15:02
jast	never :)	15:02
cmn	all that memory and processing power was just gonna sit there unused, we're helping	15:02
EdwardIII	hah	15:02
	despite git's complexity it does seem to do simple stuff better than svn though - ignoring files, for example	15:03
	← ngulden left	15:03
EdwardIII	that always seemed like a big rigmarole with svn	15:03
cbreak	one of the groups I work with has an SVN Commit EMail hook to the whole group	15:03
	they also love when ever I git svn dcommit a bunch of new things...	15:03
jast	yeah, I remember when I was in a similar situation	15:04
cbreak	especially since I write good (-> Detailed and descriptive) commit messages	15:04
jast	nobody ever complained, though	15:04
cbreak	and I commit nicely and atomically	15:04
Mikachu	"stop working so much >_>"	15:04
cbreak	oh, I had someone say that he liked it	15:04
	→ marthinal joined	15:04
cbreak	he'd actually see what changed	15:04
	← Cromulent left	15:06
	← AmirBehzad left	15:08
	→ linusoleander joined	15:08
	← ramlev left	15:08
	← nicoulaj left	15:09
	→ berserkr joined	15:11
EdwardIII	whaddyaknow, upstream implemented the same thing i implemented downstream	15:11
	if i'd thought he would want it i'd have sent a patch...	15:11
LeMike	this is great -.- remote refusing to update checked out branch. i have a new clone and new pull, just changed one file. how could that be?	15:12
	← drizzd left	15:13
cmn	because the reason is the checked-out branch	15:13
	doesn't matter what actually changed	15:13
cbreak	LeMike: don't push to non-bare repositories	15:13
LeMike	yee. what does that "non-bare" mean?	15:14
cbreak	LeMike: man gitglossary	15:15
gitinfo	LeMike: the gitglossary manpage is available at http://jk.gs/gitglossary.html	15:15
	LeMike: an explanation of bare and non-bare repositories can be found here: http://bare-vs-nonbare.gitrecipes.de/	15:15
	→ munichlinux joined	15:16
	→ mxweas joined	15:17
	→ ramlev joined	15:17
	← ramlev left	15:17
	← mxweas left	15:18
	→ ramlev joined	15:18
	← Bartzy left	15:19
	← ramlev left	15:19
	→ ramlev joined	15:19
	← glcrazy_ left	15:19
LeMike	thanks cbreak. can I solve this by rebase to the .git folder or is a complete new clone / init needed?	15:20
	→ matsebc joined	15:20
EdwardIII	hrm what do you guys use for conflicts? 3 way vimdiff?	15:20
cbreak	rebase? nothing to do with that	15:20
	you have to change the server repo	15:20
	(the one you want to push to)	15:20
	or push to a branch that is not checked out in the current server repo	15:20
	EdwardIII: TextWrangler, a normal programmers text editor	15:21
	EdwardIII: but take a look at man git-merge-tool	15:21
LeMike	i ment something like: git remote add origin file:///.../.git	15:21
EdwardIII	cbreak: macalicious?	15:21
cbreak	or was that man git-mergetool?	15:21
gitinfo	the git-mergetool manpage is available at http://jk.gs/git-mergetool.html	15:21
cbreak	LeMike: no	15:21
	LeMike: that's not going to do anything	15:22
LeMike	damn. ok i read again.	15:22
cmn	LeMike: it's not the path that makes bar or non-bare	15:22
LeMike	it's up from the init	15:22
	→ mloskot joined	15:23
	← mloskot left	15:23
	→ mloskot joined	15:23
	← jameslord left	15:27
	→ glcrazy_ joined	15:27
	→ joshsmith joined	15:27
	← glcrazy_ left	15:29
	→ vpopov joined	15:30
	→ glcrazy_ joined	15:30
	← Chib left	15:32
	→ neurodrone joined	15:32
	→ jameslord joined	15:32
	→ Yuuhi joined	15:32
	tizzo-afk → tizzo	15:34
	← glcrazy_ left	15:35
	→ yshh joined	15:36
	→ glcrazy_ joined	15:36
	← Spockz` left	15:37
	← tomislater left	15:38
	← ramlev left	15:38
	→ ramlev joined	15:39
	→ psoo joined	15:41
	→ delight joined	15:43
	← bigkm left	15:45
	→ glcrazy__ joined	15:47
	← glcrazy_ left	15:47
	→ glcrazy_ joined	15:49
	→ cantoma joined	15:49
	← glcrazy__ left	15:52
	→ sebrock joined	15:57
	← munichlinux left	15:57
	→ glcrazy__ joined	15:59
	← glcrazy__ left	15:59
	← glcrazy_ left	16:01
	→ brian_g joined	16:02
	→ zxiest joined	16:05
zxiest	Hello :-)	16:05
	I would like to understand a little thing about git...	16:05
	→ eijk_ joined	16:05
	← eijk left	16:05
	→ rgr joined	16:05
zxiest	Let's say I have the master branch, I branch and checkout B1, do some changes. Then, I switch back to master. When I branch and checkout B2, I get B1's files. Is there a way to branch from master?	16:06
rgr	how do I use git difftool to compare current version of a file and one X commits back?	16:06
sitaram	rgr: git difftool oldsha -- file # should work I think	16:07
jast	zxiest: if you never committed your changes, they are "floated" to other branches you check out (unless that's not possible, in which case checkout refuses to switch branches)	16:07
	→ jwmann joined	16:07
rgr	I dont want to use sha . just filename	16:08
	← psoo left	16:08
zxiest	jast i did commit to B1 .. but I would like to branch from master	16:08
rgr	or ^ notation	16:08
jast	rgr: then say HEAD~X instead of the SHA	16:08
	if you want it to use a version X commits back, you have to tell it somehow, obviously	16:08
	it can't read X from your mind	16:08
rgr	so "git difftool file file~3 "?	16:08
sitaram	no	16:08
	← gavin_huang left	16:09
sitaram	git difftool HEAD~3 -- file	16:09
rgr	obviously. I never mentioned it reading my mind.	16:09
	ah yes	16:09
	thanks	16:09
jast	zxiest: if you branched from master, the resulting branch should have the same files as master	16:09
	← dr0id left	16:10
sitaram	zxiest: if you branched from master to B1, then committed on B1, then checkout master, then branch B2 from there, it should branch from master	16:10
	→ mkg7 joined	16:10
rgr	Is there a general consensus on best difftool/diff browser for linux based osen? p4merge is pretty nice.	16:10
bremner	kdiff3 is ok for a gui tool	16:11
	→ jameslord1 joined	16:11
sitaram	rgr: general consensus on subjective stuff is hard; I love kdiff3	16:11
rgr	thanks	16:11
bremner	right. so sitaram and I have a consensus ;)	16:11
sitaram	indeed ;-)	16:11
zxiest	jast and sitaram thanks .. :-) .. I'll try again.. I must have done something wrong	16:11
	→ binjured joined	16:11
	← binjured left	16:11
jast	I like kdiff3 too (though I'm not a big fan of having to load the KDE libs for something like that)	16:11
bremner	jast: you can get kdiff3 with only QT stuff	16:12
sitaram	yeah that is a pain; but it's a very nice tool -- lots of keyboard controls	16:12
rgr	well, Im aware that there will be the usual "my choice" but most semi sane people can generally point you awre from their faorite to one they know is more generally used or better for beginners ;)	16:12
jast	question is, can I get it without building it myself? :)	16:12
	← rgr left	16:12
bremner	jast: on debian, yes	16:12
	← sebrock left	16:12
jast	oh, neat	16:12
	← jameslord left	16:12
jast	thanks for pointing that out	16:12
	→ sebrock joined	16:13
bremner	a pleasure. ;)	16:13
sitaram	bremner: does debian carry 2 kdiff3 packages then?	16:13
	tizzo → tizzo-afk	16:13
jast	yes	16:13
	kdiff3 and kdiff3-qt	16:13
bremner	just like evince and evince-gtk	16:14
sitaram	interesting...	16:14
	← gusnan left	16:14
jast	now if only dpkg weren't so slow reading its database :\	16:14
	→ jeffisabelle joined	16:15
	← mandric left	16:16
	← tshauck left	16:17
	← crisp left	16:18
	→ soulcake joined	16:21
	→ gavin_huang joined	16:22
	← thomas_b left	16:26
	← stodan left	16:30
	← jeffisabelle left	16:32
	→ sp3ctr joined	16:32
	← Sigma left	16:32
	→ p3rror joined	16:34
	← brian_g left	16:37
	← aspotashev left	16:38
	← nikuyoshi left	16:38
	← krz left	16:38
	→ krz joined	16:38
	← krz left	16:38
	→ krz joined	16:38
	→ Destos joined	16:39
	→ nikuyoshi joined	16:41
	→ fermianyon joined	16:42
	→ Cromulent joined	16:42
	→ aspotashev joined	16:43
	← FernandoBasso left	16:43
	→ notola joined	16:43
	→ brian_g joined	16:43
	← mloskot left	16:44
	→ arvind_khadri joined	16:45
	← arvind_khadri left	16:45
	→ arvind_khadri joined	16:45
	← notola left	16:45
	→ Sigma joined	16:46
	→ mandric joined	16:49
	→ millerti joined	16:49
	→ mloskot joined	16:50
	← nikuyoshi left	16:52
	← mloskot left	16:55
	← Carmivore left	16:56
	→ valtih joined	16:56
	gitinfo set mode: +v	16:56
	→ Carmivore joined	16:56
	← DrHouse\|TARDIS left	16:57
valtih	How do I commit to another branch? If I checkout then all my current changes are lost.	16:57
	← delight left	16:57
EdwardIII	hrm sorry for this uberdaft question - in a 3 way merge, i get 'remote' and i get 'local' - but what's 'base'?	16:58
Mikachu	valtih: if you 'git checkout otherbranch', then your changes are carried across if possible, otherwise nothing happens	16:59
	valtih: if that happens, the safest way is to commit your changes, switch, cherrypick, then reset the first branch again	16:59
jast	EdwardIII: the most recent point after which the history of remote and local diverges	16:59
	i.e. the last time they were merged together	17:00
EdwardIII	hrm	17:02
	← Araxia left	17:02
	→ diegoviola joined	17:03
	→ dreiss joined	17:04
cbreak	valtih: switch to an other branch and commit.	17:05
	← matthiaskrgr left	17:05
cbreak	EdwardIII: man git-merge-base	17:05
gitinfo	EdwardIII: the git-merge-base manpage is available at http://jk.gs/git-merge-base.html	17:05
EdwardIII	thanks!	17:06
	← vpopov left	17:06
	→ Araxia joined	17:07
	← simplechat left	17:07
	← aspotashev left	17:08
	← millerti left	17:09
	→ madewokherd joined	17:10
	← ISF_ left	17:10
	→ canton7 joined	17:10
	← brian_g left	17:12
	← break57382 left	17:12
	→ gusnan joined	17:13
	→ hwit joined	17:15
	→ vpopov joined	17:15
	← soulcake left	17:15
	→ soulcake joined	17:15
	→ akosikeno joined	17:16
	→ stringoO joined	17:18
	→ brian_g joined	17:19
	← DarkAR left	17:19
	← ericm left	17:20
	→ ericm joined	17:20
	← fisted left	17:23
	→ flavius joined	17:24
LeMike	how to "sync" two bare repos? major is a week old without any push and minor is up to date with lots of pushes	17:24
	→ antix joined	17:24
cmn	push or fetch	17:26
	all repos are equal (mostly)	17:26
	→ winferno joined	17:27
cmn	but if you have one repo that hardly gets updated, are you sure you need that repo?	17:27
	→ kerrick joined	17:27
	← canton7 left	17:28
	← heoa left	17:29
	→ adamm joined	17:29
	→ eletuchy joined	17:31
	← stringoO left	17:31
	→ mjessome joined	17:31
	← winferno left	17:32
	← tewecske left	17:33
	→ tewecske joined	17:33
	← tewecske left	17:34
	→ folsto joined	17:34
LeMike	yee its a child-father-grand-backup	17:34
	→ nikuyoshi joined	17:35
SethRobertson	I have a weird question. You have outstanding commits, you do a `git fetch` (without recording the refs or the number of outstanding commits first) and now you want to discover if the history of @{u} (or I guess any other branch from upstream) was rewritten, how can you do so?	17:35
	Programmaticly of course	17:35
LeMike	thanks cmn! push did it.	17:36
	→ tewecske joined	17:36
pasky	SethRobertson: git reflog origin/blabla?	17:36
cmn	I don't think git keeps reflogs for remote branches	17:37
	→ Fandekasp joined	17:38
	← BonQuisha left	17:38
wereHamster	cmn: git keeps reflogs of all refs. Hence reflogs	17:38
cmn	though I wonder if core.logallrefupdates would cause git to do so even for fetch	17:38
wereHamster	otherwise it wold be named 'localbranchlogs' or something like that	17:39
	← nikuyoshi left	17:39
	→ lacrymology joined	17:39
	→ BonQuisha joined	17:39
	→ Silowyi joined	17:40
cmn	true, but the manpage says "Reflog is a mechanism to record when the tip of branches are updated"	17:40
	→ stringoO joined	17:40
	← dorisabayon left	17:41
wereHamster	cmn: and I guess you have no local repository where you could have run 'git reflog origin/master' to check for yourself?	17:41
	→ iocor joined	17:41
	→ Pieter_ joined	17:42
SethRobertson	Reflog message: fetch: forced-update	17:43
cmn	origin/master doesn't show anything here, but the branches that get updates do show it	17:43
	I think the bare repo not recording pushes threw me off about what gets logged	17:43
wereHamster	cmn: bare repos don't have reflogs by default	17:44
cmn	right, and I mixed up remotes and remote branches	17:45
	→ mishok13 joined	17:46
SethRobertson	It isn't necessarily trivial to see the list of all reflog entries for origin/* which were updated since the last fetch	17:47
	But still, this provides enough hints for my purposes	17:48
	thanks	17:49
	← mjessome left	17:49
	→ flip_digits joined	17:50
	← gusnan left	17:52
	← unreal left	17:53
	→ stratwine joined	17:54
	← mkg7 left	17:54
	→ shruggar joined	17:54
	→ aspotashev joined	17:55
	← sp3ctr left	17:55
	→ sp3ctr joined	17:56
	← adamm left	17:56
	→ adamm joined	17:56
	→ __cli joined	17:57
	→ unreal joined	17:57
	← mandric left	18:00
	→ rmillerx joined	18:01
	← eletuchy left	18:03
	→ edude03 joined	18:03
	→ kukks joined	18:05
	→ najam joined	18:08
	→ mxweas joined	18:09
	← najam left	18:11
	→ o_ joined	18:12
	← Fandekasp left	18:12
	← brian_g left	18:12
	← soulcake left	18:14
	← akosikeno left	18:15
	← stringoO left	18:17
	← thedeeno left	18:24
	→ Octalot joined	18:26
	→ brian_g joined	18:28
	→ v0n joined	18:29
	→ Fooker22 joined	18:29
	← Fooker22 left	18:29
	← mxweas left	18:30
	→ FernandoBasso joined	18:32
	→ gusnan joined	18:32
	→ diegovio1a joined	18:32
	diegoviola → Guest24769	18:32
	diegovio1a → diegoviola	18:33
	← Guest24769 left	18:33
	→ ceej_ joined	18:36
	← EricInBNE left	18:37
	← unreal left	18:38
	→ unreal joined	18:39
	→ AmirBehzad joined	18:40
	→ wilmoore joined	18:42
	← cesc left	18:42
	→ mbj joined	18:42
	→ VladGh joined	18:43
mbj	Is it a design limitation git archive cannot do an archive based on commit-id when using --remote=$myremoterepo ?	18:43
	Only symbolic names (refs) seem to be supported.	18:43
wereHamster	mbj: same limitation as in git fetch	18:43
	same reason for the limitation...	18:44
mbj	wereHamster: Will consult the git fetch limitation reason documentation. Thx.	18:44
wereHamster	I don't think it's documented.	18:44
mbj	So I have to ask about the reasons :D	18:45
	→ dangerousdave joined	18:45
wereHamster	so you can't fetch arbitrary commits	18:45
mbj	I have a use case where I need to archive arbitary commits (btw the workspace state at a specific arbitrary commit)	18:46
	Im not into git internals terminology, currently.	18:47
wereHamster	mbj: fetch, then create the archive	18:47
	← rphillips left	18:48
mbj	When I read the sourcecode I just realized I have to flip the remote arg in builtin/upload-archive.c	18:48
	← Samual left	18:49
	→ Samual joined	18:49
mbj	I'd like to avoid the creation of a local repository to fetch and archive. One step would be nicer.	18:49
SethRobertson	git was recently updated so that remote repos will refuse to create arbitrary archives, but instead only archives that have tags or branches pointing at them	18:49
mbj	SethRobertson: why?	18:50
	Maybe I doing something dump here :D	18:50
SethRobertson	Security. There may be commits that you don't want someone to access or are old or something	18:50
cmn	ssh into the machine where the repo is and run archive there	18:51
SethRobertson	You can change where a tag points to, but you cannot change the SHA that was published in print three years ago that someone tries to get	18:51
mbj	cmn: would be possible, but this does not work with restricted shells like gitolite or git-shell.	18:52
	tizzo-afk → tizzo	18:52
mbj	Maybe this behaviour should be configurable.	18:52
SethRobertson	Then clone (once) and archive (many times) from that clone	18:52
mbj	Or compile my own version. Thx for all your answers, will take your input into consideration!	18:54
	→ rphillips joined	18:54
	← rphillips left	18:54
	→ rphillips joined	18:54
	→ legnaleurc joined	18:54
	→ cesc joined	18:55
	← lacrymology left	18:57
	→ lacrymology joined	18:59
	← legnaleurc left	18:59
	→ legnaleurc joined	19:00
	← angelsl left	19:00
	← fayimora left	19:00
	← pidus left	19:00
	→ fayimora joined	19:00
	← kogent left	19:00
xiong	I'm certain this is trivial but it's new to me. I've been using git for development of my own projects with success. Now I'm trying to help a new guy learn; and I'm encouraging him to use git and github to share his work.	19:03
	→ wulffeld joined	19:04
	← aro left	19:04
xiong	So I have a local clone of his repo, for convenience; but I'm not sure whether to pull, or fetch, or what. I want to see all branches; but I don't much want to create a branch of my own. I'm just looking to see "over his shoulder".	19:05
wereHamster	git fetch	19:05
	ChanServ set mode: +v	19:05
wereHamster	then you can look at origin/??	19:05
	← snearch left	19:05
	← Octalot left	19:05
xiong	Yes. But still there is a local 'master' branch -- which doesn't correspond to any useful concept.	19:06
wereHamster	HEAD needs to point to something.	19:06
	and if you don't care about the working tree, create a bare clone	19:07
xiong	I should like my HEAD to point to my student's HEAD.	19:07
wereHamster	clone --mirror, if you want	19:07
	← Textmode left	19:08
xiong	Ah. No, not bare; the purpose of cloning is so I can poke around in his tree and find mistakes or discern intent.	19:08
	→ avandenhoven joined	19:08
	gitinfo set mode: +v	19:09
	ceej_ → ceej	19:09
	← linusoleander left	19:09
wereHamster	then make a clone, fetch, checkout whatever you want, inspect the tree.	19:09
	ignore the local branch master. Or use reset --hard instead of checkout.	19:10
	← rphillips left	19:11
xiong	Yes, that's a rational statement: Just ignore it.	19:11
	→ break57382 joined	19:11
	→ jason237 joined	19:12
xiong	I have been doing detached checkouts on origin/work and origin/master; which is what I want.	19:12
wereHamster	xiong: here's an idea: git checkout HEAD^{}; git branch -D master;	19:12
	← nixmaniack left	19:13
	← break57382 left	19:13
	→ break57382 joined	19:13
	← disappea_ left	19:13
xiong	Well, after thinking about it, I'm just going to shuffle my local 'master' down to the root of the whole tree to get it out of the way; and not worry about it.	19:14
	→ mxweas joined	19:14
wereHamster	'down to the root of the whole tree'? what does that mean?	19:14
	→ mloskot joined	19:14
	→ LimeBlast joined	19:15
xiong	The root of the commit tree. The ancestor of all commits.	19:15
	'Initial commit.'	19:15
wereHamster	what would that solve?	19:15
	→ disappe__ joined	19:15
xiong	It would get it out of my way, that's all.	19:15
shruggar	define "in your way"	19:16
	→ soulcake joined	19:16
wereHamster	xiong: I told you how to delete master. Y U no use that solution?	19:16
wulffeld	Hey, I've got a problem where git archive is crashing: https://gist.github.com/1574600 - is there a way I can test git-upload-archive on the remote?	19:17
xiong	wereHamster, Yes well, our messages are crossing, likely because we're both doing other things -- I'm studying up on what you're saying.	19:18
	→ tellone joined	19:18
	← folsto left	19:19
xiong	shruggar, Ideally, I should see no branches at all except what my student has put there. Ideally, I should see what he sees, no more, no less. Except when I decide to change the rules and deliberately create a 'tutor' branch and do work for him. Usually, not. Let him work.	19:19
	← jceb left	19:19
cmn	wulffeld: it's not crashing, it's giving you an error message	19:19
	← o_ left	19:20
wulffeld	cmn: Okay, well, I'm none the wiser unfortunately. The ref is in the repos and I can push/pull fine.	19:20
	→ legnaleurc_ joined	19:21
cmn	what you're giving it is not a ref, it's a commit	19:21
	← legnaleurc left	19:21
	← flip_digits left	19:21
cmn	try with the branch name	19:22
xiong	Oh. Really! Well thank you wereHamster, the solution was much easier than I thought. I've been able to delete local 'master' without any fanfare.	19:22
	→ fl0w joined	19:22
wulffeld	cmn: I see. This is part of a deployment procedure that first does "git ls-remote triad:/scm/myrepos.git HEAD" that used to work fine. If I use master:site it's the same error msg.	19:23
	→ Octalot joined	19:23
cmn	it won't the the exact same one... does master exist?	19:24
	what does HEAD:site do?	19:24
	← NiklasFiekas left	19:24
	→ legnaleurc joined	19:24
	→ dorkmafia joined	19:25
	→ codebeaker_ joined	19:25
	← mxweas left	19:25
	← legnaleurc_ left	19:25
wulffeld	cmn: http://pastebin.com/DCkbUcyL	19:26
	→ ensonic joined	19:26
	ChanServ set mode: +v	19:26
cmn	oh, bugger, turns out it's actually the syntax that's wrong	19:27
	→ SegFaultAX joined	19:27
wulffeld	cmn: Oh?	19:27
cmn	it might work locally, but upload-archive doesn't seems to take magic pathspecs	19:27
	→ eijk__ joined	19:27
wulffeld	cmn: Worked fine when I moved the repos to another server.	19:28
	→ linusoleander joined	19:28
cmn	also with remote archive?	19:28
	← eijk_ left	19:28
cmn	the manpage says to give the ref as one arg and the path as a different one	19:28
	→ brainbox1100 joined	19:28
wulffeld	cmn: Yep. Like a charm.	19:28
ensonic	hi, where does git store the Follows/Precedes info for tags, I still have some tags in my converted repo where the tag follows itself	19:29
cmn	there is no such data stored	19:29
	not explicitly, anyway, you get it by examining the DAG and where tags point	19:29
wulffeld	cmn: Deprecated syntax maybe? The one that works is 1.7.2.5 - the one that barfs is 1.7.8.1 (or 2)	19:29
	← Zyrkon left	19:29
ensonic	cmn, hmm, mybe a bug in gitk then ..	19:30
	→ nzee joined	19:30
cmn	wulffeld: shouldn't be, HEAD:site should be a valid way to get a treeish	19:31
	might be a regression	19:31
	ensonic: perhaps, but are you expecting? you're probably inspecting a commit, no?	19:32
	← kermit left	19:32
	→ kermit joined	19:32
wulffeld	cmn: I'll fiddle around with an older version and see if that works. Thanks a bunch for your help!	19:32
cmn	I'd consider it correct if it said a commit follows a tag that points to it	19:33
ensonic	cmn, I have a couple of tags, REL_0_1, REL_0_2, ... gitk shows them as hyperlinks and for most tags 0_2 follows 0_1 and preceeds 0_3, but for a few 0_X follows 0_X	19:34
	cmn, this is when clicking on the commit where the tag points to	19:35
	← fl0w left	19:36
cmn	here clicking on a tagged commit shows a follows of the one before, so apparently you're expecting the right behaviour	19:37
	← VladGh left	19:37
	← dorkmafia left	19:37
cmn	it seems to behave a bit erratically	19:38
	so yeah, it looks like a gitk bug	19:39
	← cesc left	19:40
wulffeld	cmn: Yup, 1.7.6.5 works for me so something changed.	19:41
	→ schnatta joined	19:42
ensonic	cmn, you see the wrong beaviour sometimes too?	19:43
cmn	the fields are missing in some cases	19:43
	or blank, rather, even though for later and earlier tags it's fine	19:44
	wulffeld: sounds like a regression	19:44
ensonic	cmn, I was just wondering about this, as I have converted the repo and I could still fix it as the repo is not yet shared, but if this is not explicitely stored in the repo, I can push the repo	19:44
cmn	that information can only be inferred, git doesn't know or care what follows what in the tags	19:44
	the only following git has is in the history	19:44
ensonic	great gitk does not support --version and neither shows it in the about box :/	19:45
wulffeld	cmn: I just got a mail from a guy who has the same problem. He was in here earlier and someone told him it was changed for security reasons: "git archive/fetch only supports fetching or archiving for commits referenced as tags or branches. And this all for security reasons.".	19:46
	← SamB left	19:47
cmn	yes, but it's still referenced by a branch, if you say HEAD:site	19:47
	→ crp joined	19:47
cmn	it was probably hardened a bit so it takes a commit instead of a tree-ish	19:47
SethRobertson	1.7.8.1 "git archive" mistakenly allowed remote clients to ask for commits	19:48
	that are not at the tip of any ref.	19:48
	→ mandric joined	19:48
cmn	but HEAD:site /is/ at the tip of a ref	19:49
cbreak	isn't that part of the spec?	19:49
	← brian_g left	19:49
SethRobertson	I was just clarifying what I said earlier, not arguing against HEAD:site	19:49
cmn	ok	19:50
	→ SamB joined	19:50
	← aspotashev left	19:50
jast	IIRC git archive has allowed downloading arbitrary trees if upload-archive was enabled at all	19:50
	← brainbox1100 left	19:51
	→ dvide joined	19:52
	→ aspotashev joined	19:52
	← thebignoise left	19:52
cmn	so the fix was probably to just take the whole arg and try to parse that as a ref	19:53
	← soulcake left	19:53
	← Destos left	19:53
	→ brainbox1100 joined	19:54
	→ DaveIngram joined	19:54
	→ nixmaniack joined	19:55
	→ Zyrkon joined	19:56
	← lonewulftp left	19:56
	→ NiklasFiekas joined	19:56
	← nixmaniack left	19:56
	→ nixmaniack joined	19:58
	← mandric left	19:59
	→ dorkmafia joined	19:59
	→ churp joined	19:59
	← fermianyon left	20:01
	→ crp_ joined	20:01
	→ xerora joined	20:02
	← AmirBehzad left	20:02
	← crp_ left	20:03
SethRobertson	I bet that is not true. That would have been written differently	20:04
	← crp left	20:05
	← ipalaus_ left	20:05
	← mishok13 left	20:05
	← NiklasFiekas left	20:06
	→ ipalaus_ joined	20:13
	tizzo → tizzo-afk	20:14
	→ lacrymol1gy joined	20:16
	← lacrymology left	20:17
	← codebeaker_ left	20:17
	← eighty4 left	20:17
	← dreiss left	20:18
	← Octalot left	20:18
	→ Octalot joined	20:19
	→ codebeaker_ joined	20:19
	← krz left	20:21
	← Cromulent left	20:22
	→ snearch joined	20:24
	→ mjessome joined	20:24
	← codebeaker_ left	20:25
	← lacrymol1gy left	20:26
	← SegFaultAX left	20:27
	← bgerber left	20:29
	→ bgerber joined	20:29
	← stratwine left	20:29
	← Dynom left	20:30
	→ Dynom joined	20:30
	← Dynom left	20:30
	← faber left	20:33
	← aspotashev left	20:34
cmn	well, from the message "HEAD:site" ref doesn't exist, that's what it points to	20:36
	← matsebc left	20:37
	← v0n left	20:37
	← gavin_huang left	20:37
	← yshh left	20:37
	→ yshh joined	20:38
	→ friskd joined	20:38
	← friskd left	20:39
	← kanru left	20:39
	→ kanru joined	20:40
	→ friskd joined	20:42
	← kermit left	20:42
	← yshh left	20:42
	→ kermit joined	20:42
	→ brian_g joined	20:42
	→ Leemp2 joined	20:45
	← marthinal left	20:45
	→ NiklasFiekas joined	20:46
	← Leemp2 left	20:47
	→ igotnolegs joined	20:48
	← khmarbaise left	20:49
	← BonQuisha left	20:51
	→ Cromulent joined	20:54
	← LimeBlast left	20:56
	← mjessome left	20:56
	→ prince_igor76 joined	20:59
	← break57382 left	21:00
	← snearch left	21:02
	← umren left	21:02
	→ matsebc joined	21:03
	← prince_igor76 left	21:03
	→ rgk joined	21:04
	→ metcalfc joined	21:05
	→ crisp joined	21:05
rgk	i'm trying to configure gitweb on a gentoo box and besides copying the folder instaweb creates in the webroot, can't get it to work, im using lighttpd	21:06
	← rmillerx left	21:08
	→ rmillerx joined	21:08
	→ nowhere_man joined	21:12
	→ rphillips joined	21:12
	← kephas left	21:13
	← neurodrone left	21:13
	→ Xeross joined	21:14
	→ BonQuisha joined	21:14
	→ psoo joined	21:14
	→ ldlework joined	21:14
	← frogonwheels left	21:15
	← Dulak left	21:16
	← Sir_Rai left	21:16
	→ codeshah joined	21:18
	← Spockz left	21:20
	← wilmoore left	21:25
	← avandenhoven left	21:26
	← brian_g left	21:26
	→ wilmoore joined	21:26
	← siprbaum left	21:28
	→ siprbaum joined	21:28
	← Cromulent left	21:30
	→ delight joined	21:30
	← rhl left	21:31
	tizzo-afk → tizzo	21:32
	← TomyLobo left	21:33
	→ TomyLobo2 joined	21:33
	TomyLobo2 → TomyLobo	21:33
	← ManDay left	21:36
	← valtih left	21:36
	tizzo → tizzo-afk	21:38
bremner	rgk: I run several gitweb installs on lighttpd	21:39
	rgk: iirc, there are even howto-s	21:40
	rgk: did you enable CGI?	21:40
rgk	bremner: thats what im looking for	21:40
	yeah	21:40
	it works when i copy the instaweb install	21:40
	but all the guides tell me to link to the /usr/share/gitweb	21:41
	so it stays updated	21:41
	← fir31ight left	21:42
	→ tomislater joined	21:42
bremner	I don't know what that is. It surely must depend on the OS/distro	21:42
	here is /etc/lighttpd/conf-available/20-gitweb http://paste.debian.net/151348/	21:42
	→ Spockz joined	21:43
bremner	that is on debian.	21:43
rgk	yea that looks different then what i was doing	21:43
	thanks ill try it	21:43
	bremner: why the $ at the end of the $HTTP["host"] value	21:45
	← whitman left	21:46
	← phaedrix left	21:46
bremner	it's a regular expression	21:47
	that marks the end of the string	21:47
	→ nicoulaj joined	21:48
	← iocor left	21:50
	← nicoulaj left	21:51
	→ dv310p3r joined	21:53
	← psoo left	21:53
iheffner	cmn: unless you are trying to git-archive from a repository that has a HEAD (under refs/heads/ or as a file named "HEAD" at the root of the repository), "HEAD" is meaningless to git-archive.	21:53
	sorry; not cmn.... wulffeld ^^	21:54
	→ brian_g joined	21:54
	→ fir31ight joined	21:55
cmn	it does understand HEAD over file://	21:58
	just tried over ssh (gitolite) and git archive --remote=origin HEAD does work, w/o any HEAD under refs/heads	22:00
rgk	bremner: thanks forgot about that, this is what im stuck with right now which is more then i can say from before http://git.madgizmo.org/	22:01
	it won't pickup the /etc/conf.d/gitweb.conf though	22:01
	← sebrock left	22:01
iheffner	SethRobertson: looking at git commit ee27ca4a7818, it's not about being at the tip; it's about being reachable	22:02
	→ caseymcg joined	22:03
	← naee left	22:04
	← linusoleander left	22:04
iheffner	wulffeld: that commit also states "This also restricts some legitimate requests, too: ... 2. Sub-trees of reachable commits, like: git archive --remote=$url v1.7.7:Documentation	22:05
	→ riyonuk joined	22:06
iheffner	and cmn, "Local requests continue to use get_sha1, and are not restricted at all."	22:06
cbreak	alternative --remote=$url v1.7.7 -- Documentation	22:06
cmn	yes, I know local requests work	22:06
	but that's not at all what we were talking about	22:06
	← bindaasomatic left	22:06
iheffner	cmn: no, I know. This just explains why local requests still work.	22:07
cmn	right, but archive --remote and a local archive execute two different pieces of code	22:08
	the remote part is handled by upload-archive	22:09
	← matsebc left	22:09
	→ matsebc joined	22:11
	← baccenfutter left	22:11
	→ iocor joined	22:12
	← lonewulf` left	22:12
	← LongBeach left	22:13
	→ ipalaus__ joined	22:15
	← ipalaus_ left	22:15
	→ lonewulf` joined	22:15
	→ pantsman joined	22:15
	← pantsman left	22:15
	→ pantsman joined	22:15
	→ eletuchy joined	22:17
	→ canton7 joined	22:18
	← ThomasLocke left	22:18
	← arvind_khadri left	22:20
	→ Cromulent joined	22:23
	← ensonic left	22:23
	→ avandenhoven joined	22:25
	→ xsaidx joined	22:25
	← xsaidx left	22:26
	→ kosaidpo joined	22:26
	← kosaidpo left	22:27
	→ macRover joined	22:29
	→ neurodrone joined	22:29
	← wulffeld left	22:29
	← avandenhoven left	22:29
	← xerora left	22:30
	→ avandenhoven joined	22:32
	→ xerora joined	22:32
	→ mjessome joined	22:32
	→ arvind_khadri joined	22:33
	→ btanaka joined	22:34
	→ fisted joined	22:35
	→ EricInBNE joined	22:35
	← avandenhoven left	22:36
	→ steffo joined	22:36
	← Jackneill left	22:37
	← rendar left	22:39
	← caseymcg left	22:41
	← diegoviola left	22:42
	→ DarkAR joined	22:43
	← Spockz left	22:43
	← mjessome left	22:44
	→ frogonwheels joined	22:45
	→ eletuchy_ joined	22:46
	→ Araxia_ joined	22:47
tgunr	I accidently checking a huge binary into my repo a few commits back as I just discovered (490MB). How can I prune it completely out of the repo? I don't want a clone to have to deal with this huge file.	22:47
FauxFaux	tgunr: man git filter-branch	22:47
gitinfo	tgunr: the git-filter-branch manpage is available at http://jk.gs/git-filter-branch.html	22:47
tgunr	thanks, i couldn't recall which tool to use.	22:48
Mikachu	http://help.github.com/removing-sensitive-data/	22:49
	might be helpful too, huge or sensitive should be same deal :)	22:49
	← tomislater left	22:49
	← eletuchy left	22:50
	← Araxia left	22:50
	Araxia_ → Araxia	22:50
	→ Goplat` joined	22:52
	← Goplat left	22:52
	← sitaram left	22:54
tgunr	cc	22:54
	← Sigma left	22:54
	→ sitaram joined	22:56
	← sitaram left	22:56
	→ sitaram joined	22:56
	← macRover left	22:58
	→ aspiers joined	23:03
	← txomon\|home left	23:05
aspiers	hi all, is there a single git CLI command to get a directory listing similar to what is displayed here: https://github.com/git/git - in other words, filename, age, and commit message?	23:05
	(and possibly the abbreviated commit SHA1 too)	23:06
cmn	no	23:06
aspiers	I can do it with a combination of git ls-tree and git log -1	23:06
	but that's ugly, and doesn't scale to large directories	23:06
Mikachu	git instaweb ;)	23:07
aspiers	Mikachu: yeah, I want it on STDOUT though :)	23:07
	wow, that's pretty surprising if it doesn't exist	23:07
	← brian_g left	23:07
cmn	git instaweb & curl localhost:8080/whatever	23:07
	it's not surprising at all	23:08
	that information isn't stored in git	23:08
	you need to infer it, and it's an expensive operation	23:08
aspiers	git log -1 is expensive?	23:08
Mikachu	for each file, you need to find the last file where it was changed	23:08
cmn	no, running log for each path is expensive	23:08
	← kermit left	23:08
Mikachu	last commit, rather	23:09
cmn	for each path you need to run git log --filter-diff=M and grab the first commit it shows	23:09
Mikachu	well, you don't strictly need to, but it's the most straightforward way	23:09
cmn	sure, you can cache	23:09
aspiers	I've heard this kind of argument before, from the author of git-annex	23:09
Mikachu	it's not an argument, it's a fact	23:09
aspiers	Mikachu: you misunderstand me	23:10
cmn	argument?	23:10
aspiers	let me finish	23:10
cmn	it's the way the git data is modeled	23:10
Mikachu	you're the one who stopped	23:10
	:)	23:10
aspiers	I'm not talking about that	23:10
	I'm composing the next sentence :)	23:10
Mikachu	nobody who wanted it wanted enough to write it, obviously	23:11
cbreak	most people on IRC can type really fast	23:11
aspiers	the argument is: "this is a polynomial-time problem, so in the worst cases, it can be extremely expensive and end up hogging CPU and/or RAM, therefore we shouldn't implement it at all"	23:11
cbreak	we're impatient	23:11
aspiers	which I strongly disagree with, because in many cases, it is very useful functionality which a modern system is easily capable of handling	23:12
cmn	it's implemented in several places, just not in git-core	23:12
aspiers	and if it starts grabbing CPU/RAM, then you just kill it, or set limits.conf	23:12
cmn	caring about indivial files isnt' very gittish	23:12
aspiers	the majority of repos are nowhere near big enough for it to be an issue	23:12
Mikachu	git ls-files \| while read file; do git log -n 1 --pretty="Filename: $file, commit: %h, date: %ad" -- $file; done \| cat	23:13
	that's how slow the naive way is (stole it from stackoverflow)	23:13
	← dirkle left	23:13
cbreak	what does he want?	23:14
	← zxiest left	23:14
cmn	why is there a pipe to cat at the end?	23:14
Mikachu	cmn: to not open less 132 times	23:14
aspiers	Mikachu: I assume you mean http://stackoverflow.com/questions/223678/git-which-commit-has-this-blob which I had already read	23:14
Mikachu	that is incorrect	23:14
cmn	oh, right, because git would try to open the pager	23:15
Mikachu	http://stackoverflow.com/questions/5183199/how-print-last-commit-info-for-every-file-in-a-git-repository	23:15
	well, it's correct that you assumed it, but it's not what i meant	23:15
aspiers	haha :)	23:15
	→ kermit joined	23:15
Mikachu	fwiw, that command has been running for 2 minutes on git.git now	23:15
aspiers	ahah, I couldn't find that SO question	23:15
	but my quick hack was almost identical to that	23:16
	ah, no	23:16
	that's not what I asked for	23:17
	I only want the info for a single directory	23:17
Mikachu	that's conceptually the same thing	23:17
cmn	either ls-files has flags to do that or use ls-tree \| cut	23:17
aspiers	it does not have the same algorithmic complexity, except in the worst case where your repo only has one directory	23:17
cmn	it has the same complexity, it might not have the same n	23:18
aspiers	if you restart your git.git run on a single directory and time it, I think you will prove my point :)	23:18
Mikachu	git ls-tree HEAD --name-only instead of ls-files	23:18
aspiers	cmn: complexity depends entirely on your definition of n	23:18
	→ dfr\|mac joined	23:18
	← canton7 left	23:18
Mikachu	it's obviously the same complexity	23:18
	that's like saying truncating filenames before the letter f is a lower complexity, / isn't special in git ;)	23:19
cmn	n is the amount of files you want to look the info for	23:19
aspiers	no, n could be the number of files in the repo	23:19
cmn	and it runs faster because it has less work to do	23:19
	how would that definition of n make sense?	23:19
Mikachu	i'm not going to argue about it	23:19
cbreak	Mikachu: there are much more efficient ways to implement that	23:19
Mikachu	there are a bunch of files in t from 2006, those are really slow to list	23:20
aspiers	n could also be the number of commits in the repo	23:20
	my point is that a single-dimensional complexity analysis is not very helpful here	23:20
cbreak	algorithm O(n) in number of commits and O(n) in number of files	23:21
aspiers	because the performance is critically dependent on multiple characteristics of the repo and the directory in question	23:21
	no one metric can yield a valuable predictor of run-time	23:21
	cbreak: that's O(m.n) :-)	23:22
Mikachu	well, for 2147 files and 25318 commits, it is Pretty Slow	23:22
cbreak	Mikachu: because your algorithm is O(numcommits*numfiles)	23:23
aspiers	right, but for a very new directory containing 10 files, in a repo with 9629483619 commits, it might be pretty fast	23:23
	← delight left	23:23
cmn	or very slow	23:23
aspiers	cmn: how so?	23:24
Mikachu	sure, but that has nothing to do with algorithmic complexity	23:24
cmn	if one file was only ever touched in the first commit	23:24
Mikachu	just some of the numbers in your algorithm end up smaller if you pick only newer files	23:24
	but it's always the same algorithm	23:24
aspiers	Mikachu: yeah, I'm trying to encourage this discussion back to the real world ;-)	23:24
Mikachu	( git ls-files \| while read file; do; git log -n 1 -- $file; done; ) 612.71s user 28.95s system 98% cpu 10:52.65 total	23:25
aspiers	in which this functionality would be useful in a large number of the git repos out there in the world	23:25
	← ipalaus__ left	23:25
j416	git ls-files \| while read file; do git log -n 1 --date=iso --format="%Cblue%h%Creset %C(yellow)%an%Creset %Cgreen%ad%Creset $file %C(cyan)[%s]%Creset" -- $file; done	23:25
Mikachu	i don't think the colors make it faster	23:25
j416	it's hard to get good alignment :)	23:25
aspiers	lol	23:25
j416	lol	23:25
	← nzee left	23:25
j416	speed isn't an issue here	23:25
cmn	not even the "go faster stripes"?	23:25
aspiers	%Cred would definitely make it faster	23:25
j416	small repo	23:25
	haha	23:25
aspiers	j416: that's my point exactly	23:26
j416	just thought I'd share it.	23:26
cbreak	Mikachu: your algorithm is worst case numberOfCommits*numberOfFiles, which is very slow. But you can make an algorithm that is in that case just numberOfCommits	23:26
aspiers	I have < 5000 files	23:26
Mikachu	yeah	23:26
aspiers	and 426 commits	23:26
cbreak	it's still slow	23:26
Mikachu	the worst case is presumably numberOfCommitsToOldestFileInSet or so :)	23:26
	ie, for the best algorithm	23:26
aspiers	and the directory in question only has 23 files	23:26
cbreak	about as slow as running git log	23:26
Mikachu	but that might use a _lot_ of memory	23:26
cbreak	all through the end	23:26
	Mikachu: na	23:27
	you just have to remember the commit each file is changed in	23:27
	so a list as long as the number of files in the newest commit is enough	23:27
	← DaveIngram left	23:27
cbreak	(since you want to find out when those files are modified last)	23:27
Mikachu	okay, yeah, only a couple of megabytes for linux.git probably	23:28
j416	that looks better aligned: git ls-files \| while read file; do git log -n 1 --date=iso --format="%Cblue%h%Creset %Cgreen%ad%Creset %C(yellow)%an%Creset $file %C(cyan)[%s]%Creset" -- $file; done	23:28
Mikachu	you probably want git --no-pager log	23:28
	→ Destos joined	23:28
j416	(me?)	23:30
	← sp3ctr left	23:30
	← wilmoore left	23:30
Mikachu	yeah	23:30
j416	why?	23:30
Mikachu	it might open the pager for each file otherwise	23:30
j416	seems not	23:30
	I set GIT_PAGER=less first, no difference	23:30
	hm	23:30
Mikachu	your less probably has the --exit-if-one-screen option set	23:30
	git has some options hardcoded if LESS_OPTIONS is not set, iirc	23:31
j416	does it not work for you?	23:31
aspiers	j416: pager issues for me here	23:31
j416	hm ok	23:31
	well I was using the same thing, only colours added, as was suggested on some link above	23:32
	j416 shrugs	23:32
aspiers	fixed by --no-pager	23:32
j416	ok :)	23:32
	on which git?	23:32
	git --no-pager log?	23:32
Mikachu	yes	23:32
aspiers	yup. 1.7.8.206.g71406	23:32
j416	k	23:32
	→ mloskot_dogwalk joined	23:32
j416	now to figure out what to name the alias for this	23:33
cmn	git pretend-im-svn	23:33
	← mloskot left	23:34
j416	haha	23:35
Mikachu	git pis	23:35
aspiers	j416: looks pretty nice but I prefer --date=relative with it coming after the other fields	23:35
j416	whatever you like :)	23:35
aspiers	actually I just changed my mind again	23:35
	← baburdick left	23:36
aspiers	--date=relative is better for web pages which can actually do proper alignment and have more than 80 columns :)	23:36
	→ jab416171\|Cloud joined	23:36
aspiers	--date=iso is sortable too	23:36
j416	"git log --stat" on git.git is significantly faster than that thing, maybe it would be easy to parse that for files against git ls-files	23:36
	I wonder how nicely one could do it in sh	23:36
	→ baburdick joined	23:36
aspiers	j416: that's a good idea	23:37
Mikachu	https://git.wiki.kernel.org/articles/e/x/a/ExampleScripts_6d49.html	23:37
	ran in 3.4 seconds for the whole git repo :)	23:38
aspiers	was that already there?	23:38
Mikachu	yes	23:38
j416	Mikachu: heh, yeah, that's the thought	23:38
	cool :)	23:38
	→ drev1 joined	23:39
j416	whoa fast	23:39
	< 2 s here	23:39
	→ brian_g joined	23:39
j416	formatting isn't very nice, though	23:39
Mikachu	yeah, could use some options	23:40
	← Cromulent left	23:40
j416	but, I don't have much need for something like this anyway, so I won't waste my time doing it. :)	23:41
Mikachu	incidentally in git.git it seems to take exactly the same time without the -r both in / and in some subdirs	23:41
cbreak	if you want to write it efficiently	23:42
	you should analyze the output of git log --name-only	23:42
j416	Mikachu: I'd assume it's git log that takes time to run, not git ls-tree	23:42
Mikachu	j416: yeah, but the theory was that it would be faster if it didn't need to traverse the full history	23:43
j416	mhm	23:43
	ok	23:43
cbreak	something like git log --name-only \| (remember the last sha, and fill a map data structure with path->hash associations, without overwriting)	23:44
	→ mloskot__ joined	23:44
	→ storrgie joined	23:44
cbreak	if you want you can prepopulate the map and only fill the paths that exist in the newest commit	23:44
	and abort when all are filled	23:44
Mikachu	cbreak: that's what the script does	23:44
cbreak	you sure? It looks like noise to me.	23:45
	← Goplat` left	23:45
	→ Chib joined	23:45
	→ the_cybersphinx joined	23:46
Mikachu	it seems to not be perfect, some files get printed with -1 for me	23:46
	oh well	23:46
aspiers	haha, it uses 'break' instead of 'last'	23:48
	Bareword "break" not allowed while "strict subs" in use	23:48
	← mloskot_dogwalk left	23:48
aspiers	perfect example of why 'use strict' should never ever be omitted	23:48
	← unmanbearpig left	23:48
Mikachu	i have no idea what you just said	23:48
	← cybersphinx left	23:48
Mikachu	break doesn't break loops in perl?	23:49
aspiers	nope	23:49
Mikachu	ah	23:49
	that would be too easy?	23:49
aspiers	last	23:49
Mikachu	so break is just a noop?	23:49
aspiers	and next, not continue	23:49
	no, break is a compile error	23:49
	→ fermianyon joined	23:49
aspiers	unless you break the golden rule	23:49
Mikachu	but the default mode in perl is to ignore compile errors?	23:49
aspiers	yes	23:49
	and I'm not going to defend that	23:49
Mikachu	okay	23:50
	i'm just asking as a perl noob	23:50
	incidentally, vim syntax hilights break;	23:50
aspiers	like a keyword?	23:50
Mikachu	yeah, now it's faster in a subdir	23:50
	yes	23:50
	exactly the same color as last;	23:50
aspiers	oh dear	23:51
	← Morphous_ left	23:51
Mikachu	it's probably not a perfect script	23:51
	given what you said, and it not finding commits for all files	23:51
j416	cool	23:51
aspiers	it's working for me	23:51
Mikachu	well, near the top of git.git, it has a bunch of -1	23:52
aspiers	oh	23:52
Mikachu	which is what it inits the attribution to	23:52
aspiers	it mixes numeric with string	23:52
	doh	23:52
	should init to "" not -1	23:52
Mikachu	doesn't really matter, it didn't find the commit	23:52
	it should find commits for every file, or the file wouldn't be there :)	23:52
aspiers	in fact why does it even init?	23:52
	oh I see	23:53
	← mloskot__ left	23:53
	→ mloskot joined	23:53
	← mloskot left	23:53
	→ mloskot joined	23:53
	← hwit left	23:53
	← DarkAR left	23:53
aspiers	why is it not finding commits for you?	23:54
Mikachu	how would i know? :)	23:54
	← legnaleurc left	23:54
aspiers	maybe a parsing bug?	23:54
	or you think that git log isn't finding them?	23:54
Mikachu	when i just run git log, it seems to find them	23:54
	→ dreiss joined	23:54
cbreak	maybe there are files that are never changed.	23:55
	does adding count as change?	23:55
aspiers	cbreak: they would still be in the root commit	23:55
cbreak	probably the case.	23:55
Mikachu	they're probably not in the root commit	23:56
	it's the release notes :)	23:56
aspiers	hehe	23:56
	which file exactly?	23:56
cbreak	maybe no one wrote them.	23:56
aspiers	I could try on my git.git repo	23:56
cbreak	or they got in via an evil merge	23:56
	evil merges are ... not usually diffed	23:56
	(just like normal merges)	23:56
Mikachu	almost all files in Documentation/RelNotes/	23:57
	→ legnaleurc joined	23:57
Mikachu	understandably a bunch of stuff in git-gui/	23:57
aspiers	ok	23:57
	→ ipalaus_ joined	23:57
	→ jdav` joined	23:58
	← d0k left	23:59

Logs Search ←Prev date Next date→

Channels Documentation