IRCloggy #git 2008-04-18

Logs Search ←Prev date Next date→ Channels Documentation

Provider of IRC logs since 2005.
WARNING: As Freenode became unjoinable and lost all warnings in topics, we cannot log channels on Freenode anymore.

2008-04-18

fujin hm, would 'git push origin master' push the changes to the master branch to remote origin no matter what branch I'm on?00:01
loops Lethalman, git checkout HEAD -- <files> should work00:01
fujin changes on00:01
jbunster Is it possible to add the old id of a commit to the log when converting with hg-fast-export ?00:01
Lethalman loops, done reset00:02
schacon_ left00:02
Lethalman but i was only a stupid00:02
becase i had to rebase instead00:02
loops fujin, by default it will push to the branch of the same name in the remote00:04
rickbradley left00:04
loops fujin, you can change this with a config option though.00:04
jbunster, you may have to ask that question on the mailing list.00:05
spearce joined00:05
loops fujin, so yes.. git push origin master means push my master to the origin remote, regardless of which branch i have currently checked out.00:06
QuercusMax joined00:06
QuercusMax left00:07
hector left00:08
kirindav_ left00:08
dsaxena left00:08
kirindav_ joined00:09
hamburglu joined00:10
Lethalman left00:11
kirindav_ left00:13
hamburglu Text hier eingeben...merhabalar00:15
flaguy joined00:16
mithro joined00:16
tmm1 how do i reset a specific file back to HEAD ?00:19
kirindav_ joined00:20
faheem_ joined00:21
tmm1 git checkout file00:22
hamburglu left00:22
ben_h left00:23
kirindav_ left00:23
ben_h joined00:23
faheem left00:27
zedr0n left00:29
tjafk1 joined00:31
myrizio left00:32
kirindav_ joined00:35
tjafk2 left00:43
phoenix24 left00:43
pombreda left00:44
kipras left00:45
kirindav_ left00:45
kirindav_ joined00:46
warlock_mza left00:48
jburd left00:51
jburd joined00:52
zed_DX joined00:54
robinr left00:57
stick joined00:58
thesaint4444 joined01:02
kirindav_ left01:04
phoenix24 joined01:04
thesaint4444 hi guys, I am trying to get my head around a rails deployment with git. I have checked my development app into git and created a git clone. I copied this to my server to create an external git repo. I have modifed my development app, added and commited my files. How do I update my external git repo? thanks.01:06
shoe thesaint4444: use capistrano01:08
KragenSitaker ideally log into it and use git pull01:08
oh, yeah, something rails-specific for deployment instead of git might be good too :)01:08
i haven't tried using git for rails01:09
thesaint4444 shoe: I am using capistrano ....01:09
shoe: but surely I have to keep my git repo in sync also....01:09
shoe: I think capistrano takes its source from the git repo on the server.01:10
shoe then cap will do a git pull from whatever address you specify in the cap script for :repository01:10
thesaint4444 shoe: so if I make changes to my local dev source I have to keep the external git repo up to date....01:10
johnw thesaint4444: Here's what I do when I want to start using a remote repo: commit to your local repo, rsync it over, then clone your remote repo to a local copy and start using that. This way you can just push/pull from now and things will work as you expect.01:11
KragenSitaker there is one surprising thing about doing that kind of thing though (independent of capistrano)01:12
shoe thesaint4444: oh, you're asking how to push to the external repo?01:12
KragenSitaker you don't want to "git push" to a repository that has an associate work area01:12
thesaint4444 johnw: actually that is the way I have it set up... everything is working just I am not sure how to push to the external repo....01:13
KragenSitaker associated work area01:13
shoe (independent of deployment)01:13
johnw thesaint4444: git push01:13
i'm assuming you pulled using the ssh:// or git+ssh:// protocol01:13
shoe thesaint4444: ^^ but you first might need "git remote add place-you-cloned-to"01:14
johnw shoe: if he did a clone from the remote that should already have happened01:14
shoe johnw: but I think he cloned the external one.01:14
johnw what does external mean in this case?01:15
shoe meaning, not the development repo, but the repo shared between developers and deployment targets.01:15
Mactaylor joined01:16
johnw I understood his question to involve two repo's: his local one, and the one he's setting up on the remote server01:16
if he copies from local to remote, then clones from remote, he should be all setup to push/pull01:16
thesaint4444 johnw; yes... pulled using ssh://, have already added git remote add.... sorted - works... thanks!01:17
shoe right, but I don't think he cloned from remote, just went back to hacking in the "original"01:17
johnw thesaint4444: you didn't need git remote add01:17
Mactaylor left01:17
johnw shoe: ah, I mentioned to him that he needs to work in the remote clone, and put "original" away after he copies it01:17
igorgue joined01:17
thesaint4444 johnw; I have already done that during my setup...01:17
igorgue_ joined01:18
igorgue_ left01:18
shoe ah good, that's an easy way01:18
thesaint4444 johnw: to add the repo as a remote git repository...01:18
johnw thesaint4444: that is done for you by the fact that you cloned01:18
thesaint4444 johnw: yeah? ok, will look at that... - will it cause any issues?01:19
johnw nope01:20
I use it all the time01:20
thesaint4444 cool...01:20
agib left01:20
johnw just type "git-remote" after you "git-clone" and you should see the "origin" already setup for you in the list01:20
asonge left01:20
johnw git branch -r will show all the remote branches that you can now checkout locally01:20
jburd left01:20
thesaint4444 johnw: great, a lot of stuff to cover! i have been playing with the capistrano stuff at the same time - very usefull...01:21
igorgue another rails user.... :(, 21signals should pass a check to git developers :)01:22
thesaint4444 igorgue: lol...01:22
igorgue: git is very nice.01:22
igorgue yep pretty cool01:24
ben_h left01:24
anttih- left01:25
jbunster left01:25
dysinger left01:25
pkufranky left01:25
Snadder left01:25
Lemurnomicon left01:25
tokkee left01:25
telmich left01:25
LoCusF left01:25
djpig left01:25
robtaylor left01:25
ska-fan left01:25
pjz left01:25
mxey left01:25
kynde left01:25
thommey left01:25
mqt left01:25
floppyears left01:25
ndim left01:25
kmap left01:25
jp_tix left01:25
ekim|home left01:25
jacobat left01:25
nipuL left01:25
fonseca left01:25
peper left01:25
normalperson left01:25
Cap_J_L_Picard left01:25
AlexanderS left01:25
GrumpyOldMan left01:25
eMBee left01:25
floppyea1s joined01:25
jacobat_ joined01:25
GrumpyOl1Man joined01:25
kynde_ joined01:25
tokkee joined01:25
fonseca joined01:25
jp_tix joined01:25
mqt joined01:25
robtaylor joined01:25
ska-fan joined01:25
ndim joined01:25
pjz_ joined01:25
pkufranky joined01:25
mxey joined01:25
djpig joined01:25
eMBee joined01:25
telmich joined01:25
thommey joined01:25
normalperson joined01:25
kmap joined01:25
AlexanderS joined01:25
LoCusF joined01:25
nipuL joined01:25
adante joined01:25
peper joined01:25
ben_h joined01:25
ekim|home joined01:25
floppyea1sfloppyears01:25
Snadder joined01:25
igorgue good bye everyone, hello everyone01:26
jerbear left01:27
johan-s left01:27
pygi left01:27
Tv left01:27
clarity_ left01:27
m3y3ring left01:27
tcoppi left01:27
AlexB left01:27
jmspeex left01:27
jeremyb left01:27
offby1 left01:27
ft left01:27
vlt left01:27
gram left01:27
zeenix left01:27
Aeternus_ left01:27
albertito left01:27
Tali left01:27
cods left01:27
Alex left01:27
com4 left01:27
jast left01:27
joevandyk left01:27
mtkd left01:27
theMuss left01:27
ehamberg left01:27
spleenk left01:27
blackdae1on left01:27
ar0x left01:27
dermoth left01:27
[dmp] left01:27
mountie left01:27
Aeternus joined01:27
blackdaemon joined01:27
spleenk joined01:27
gram joined01:27
cods joined01:27
Tali joined01:27
ehamberg joined01:27
zeenix joined01:27
albertito joined01:27
m3y3ring` joined01:27
AlexB joined01:27
theMuss joined01:27
jast joined01:27
ft joined01:27
ar0x joined01:27
com4 joined01:27
[dmp] joined01:27
vlt joined01:27
jmspeex joined01:27
dermoth joined01:27
jerbear joined01:27
clarity_ joined01:27
mountie joined01:27
tcoppi joined01:27
pygi joined01:27
Tv joined01:28
thesaint4444 KragenSitaker: hi, just reading your comments... what do you mean about not pushing to the associated work area? thanks.01:31
jeremyb joined01:32
Alex joined01:32
KragenSitaker thesaint4444: only push to "bare repositories"01:32
pull the rest of the time01:32
ggeecko joined01:33
thesaint4444 KragenSitaker: and a bare repository is?01:34
johnw one with no associated working tree, whose config file reads "bare = true"01:35
kirindav_ joined01:35
thesaint4444 johnw: I pretty much followed the scott chacon screencast to get started.......01:35
Modius joined01:35
johnw if you rsync'd over your working tree + .git, you can just rm -fr the working set, move project/.git to project.git, and edit project.git/config to set bare = true01:35
thesaint4444 KragenSitaker: aahh... that makes sence... thanks.01:35
KragenSitaker because push doesn't update the working tree, or the index, and so you can get into a kind of bad state01:36
vbgunz joined01:36
spearce johnw: for what its worth, the config having bare = true isn't usually necessary. we auto-sense bare trees most of the time. :-)01:37
well, auto-sense is based on it ending in ".git".01:37
KragenSitaker you mean, having something before the .git?01:37
spearce call it "project.blargh" and you may want bare = true in your .git/config.01:37
KragenSitaker: yea, its based on it being "foo.git" and not just ".git".01:37
KragenSitaker here's masukomi's recommendation for how to do this: http://weblog.masukomi.org/2008/03/11/sharing-a-public-git-repo-over-http-flow-chart01:38
thesaint4444 can anyone recommend a good document to read in relation to this stuff? my main thrust is using git as the scm for rails apps which I will deploy via capistrano.01:38
KragenSitaker although i'm not expert enough to know if she got something wrong01:38
well, on how to create a bare repo01:38
bartman left01:39
johnw spearce: nice to know01:39
destruct1 joined01:40
johnw thesaint4444: finding a good document for you will depend on how well you want to understand Git, or if you just want to use it from the highest level possible01:40
KragenSitaker after my first few hours using git, i wrote a document on the things i wished i'd known when i started01:42
johnw KragenSitaker: can I read it?01:42
KragenSitaker yeah, maybe you can tell me what i got wrong01:43
http://pobox.com/~kragen/tmp/git-learnings01:43
johnw please send it to me01:43
reading...01:43
KragenSitaker before it infects any other newbies01:43
johnw if you git push and then git commit, it shouldn't throw away your changes, it just means that your HEAD is now on a different branch, right?01:44
(in a working tree on a remote repo)01:44
thesaint4444 KragenSitaker: cool, thanks... I will look and learn...01:45
johnw KragenSitaker: also, git-commint --amend only lets you edit HEAD; if you want to edit further back, you have to git-rebase -i01:45
thesaint4444 johnw: I want to understand it say 3 hours more than the highest level possible (just enough so I can use it without screwing up...)01:46
johnw KragenSitaker: and isn't git-add --patch similar to what you mean by git-hunk-commit?01:46
xjrn left01:46
johnw thesaint4444: that's a fair request, let me look at my collection of docs01:46
thesaint4444 johnw: many thanks....01:47
johnw I think this one might suit you: http://www.ibm.com/developerworks/linux/library/l-git/01:48
KragenSitaker: I'm using git 1.5.5 with a 4.4 Gb repository btw, maybe there were issues with 1.4?01:48
pombreda joined01:49
KragenSitaker johnw: yeah, maybe i should clarify that about --amend01:50
slightly worse than darcs in that regard i guess, but not much01:50
i haven't tried git-add --patch01:51
timocratic left01:51
destruct left01:51
KragenSitaker git-add --patch doesn't seem to be documented01:51
thesaint4444 johnw: thanks... I pretty much understand all of that. Some of the links off there are good. cheers.01:52
johnw KragenSitaker: I wonder if you'd like this: http://raphael.slinckx.net/files/git-darcs-record01:52
KragenSitaker the name is promising :)01:53
johnw it's basically a wrapper around git-add for new files, and git-add --patch for modified files, but it merges it all together into a recording session, ala darcs01:53
i'm also working on a port of darcsum.el to Git, in case you're an Emacs user01:54
KragenSitaker johnw: how many files are there in your 4GB repo? the problem I've been having is that whenever I try to commit more than a few hundred thousand new files at a time, things like git add and git commit become very slow01:54
yes, i use emacs a lot01:54
johnw KragenSitaker: let me check my HEAD commit tree size01:54
only 9285 files01:55
bartman joined01:55
thesaint4444 when you run git init-db in the root of your development directory then you add the files and commit this creates a local git repo, yes?01:55
KragenSitaker thesaint4444: yes01:55
Fullmoon left01:55
KragenSitaker although isn't it more fashionable to do git init rather than git init-db?01:55
johnw thesaint4444: it gets created the moment you say git-init01:55
thesaint4444 KragenSitaker: the object of which is to what? Do you continue doing your development in this 'area' continuously adding files and commiting?01:56
johnw KragenSitaker: perhaps I can enlist you as a beta tester?01:56
KragenSitaker johnw: very likely. i never used darcsum though, so I don't know its UI, so I might be a little bit of a pain ;)01:57
thesaint4444 KragenSitaker: with the possible intention of finally creating an external git repo.....01:57
KragenSitaker thesaint4444: yes, that's what you do. the benefits of this:01:57
- if you screw something up, you can recover pretty easily by looking at a previous version.01:57
- backups are dead easy.01:58
johnw KragenSitaker: the basic methodology of darcsum is to display your current changeset like a pcl-cvs buffer, but you can press RET to show all the hunk in a file, and 'm' to add/remove hunks from the index cache, and then 'c' to commit the bolded hunks/files01:58
KragenSitaker johnw: that sounds pretty awesome01:58
Fullmoon joined01:58
johnw in darcss there is no index cache, so the marking was done on an internal Lisp buffer, but with Git I want to use the index cache interactively01:58
KragenSitaker that makes sense01:58
johnw in a way, it's like a visual index editor01:59
hector joined01:59
KragenSitaker thesaint4444: apparently advanced users also use branches in git to easily suspend and resume unfinished development tasks01:59
thesaint4444 KragenSitaker: ok. thanks.. so every commit is effectively a snapshot of a previous version available for use if necessary contained within the external repo....01:59
johnw plus, and this is the really cool part, I want you to be able to dynamically change the Base/Current commits/working tree by which your changeset was determined01:59
thesaint4444 KragenSitaker: yes, so I have seen....01:59
KragenSitaker when they get interrupted with something higher priority, they can save their current work, work on something fresh, commit, and go back to what they were doing. but i haven't tried this.01:59
johnw automatically created branches as necessary if you commit changes against an earlier Base, auto-stashing your working tree if you change Current to an earlier commit, etc.01:59
KragenSitaker johnw: i don't understand the really cool part, but I think that's OK :)02:00
johnw sigh, I'm having a lot of trouble articulating it02:00
I think it will require a screencast for the coolness to leak out of my brain02:00
thesaint4444: every commit is a copy of your source tree at a specific point in time02:01
thesaint4444 johnw: ok..... got it.02:01
johnw thesaint4444: with insane internal linking to prevent duplicate content within the repository02:01
garrin joined02:01
KragenSitaker johnw: what was it that aiken said about stealing ideas?02:01
johnw a "patch" is the distance between two commits02:01
thesaint4444 johnw: ok....02:02
KragenSitaker i think it was something like "don't worry about people stealing your ideas. if your ideas are any good, you'll have to cram them down their throats"02:02
johnw haha, that's awesome02:03
well, this idea I have I've never seen in any VCS interface, so maybe it has earned its obfuscated nature02:03
aroben left02:04
johnw though gitk could easily approach what I'm thinking, if it were also an editor/commiting tool02:04
maybe I just need to play with gitk more02:04
spearce use git gui. :)02:04
johnw err, yes, that's what I just ran02:04
phoenix24 left02:04
vb123 left02:05
johnw hmm.. the scrollbar control is rather broken on OS X (1.5.5, Tk 8.5.2)02:05
i grab it and drag, but it disconnects from the cursor and no scrolling happens02:05
phoenix24 joined02:06
thesaint4444 so if you create a git clone: git clone --bare myapp/.git myapp.git is this linked to the initial 'development area' created with git init-db ?02:08
johnw thesaint4444: I think you're going about this the very long way around02:08
thesaint4444 johnw: id be most happy to hear the short way!02:09
johnw $ rsync -av .git/ server:/tmp/foo.git/ ; cd /tmp ; git clone ssh://server/tmp/foo.git02:09
that's all you need to setup a remote repository, and to start using it right away02:09
threeve joined02:09
johnw your remote is now at server:/tmp/foo.git, and your local working tree is in /tmp/foo02:09
just tar up your original working tree and put it in your archives, then move the two repos to where they should exist from now on02:10
done and done02:10
thesaint4444 johnw: thats it? - when you say move the 2 repos to where they should exist from now on what do you mean?02:11
johnw go to the server and run: mv /tmp/foo.git /srv/git/foo.git, or wherever you plan to server it from02:11
serve it from02:11
locally, type mv /tmp/foo ~/src/foo, or wherever you plan on working in it from02:11
your local working tree will already have a master branch linked to origin/master, so you can push/pull immediately02:12
if you had another created, do: git-checkout -b branchname origin/branchname, then you can work on it locally and push/pull to the remote branch02:13
thesaint4444 johnw: thinking....02:13
johnw: thats it? seems pretty easy!02:15
johnw most of git is, when you use it right02:15
GyrosGeier joined02:15
thesaint4444 johnw: thats the best explination I have seen so far! thanks.02:15
kirindav_ left02:16
johnw sure thing02:16
i'm going to post an in depth study of GIt from the bottom up this weekend, you might like it02:16
but maybe more info than you care to know02:17
thesaint4444 johnw; where will I be able to find it?02:17
johnw ftp://ftp.newartisans.com/pub/Git%20from%20the%20bottom%20up.pdf02:17
that's the final draft, let me know if it needs improvement02:17
peeja joined02:21
thesaint4444 johnw: looks very good.... many thanks, just had a quick scan. I will read it in the morning. Is there an email address on there?02:22
krawek indeed, thanks johnw =)02:24
johnw please send any comments to [email@hidden.address] I am most interesting in making it clear for those who wish to know how Git does what it does02:25
KragenSitaker awesome02:25
imyousuf joined02:27
floppyears hi guys02:28
quick question02:28
I'm using git-svn02:28
Ramune 42!02:28
johnw I think you meant to say, "I'm using git-svn?"02:28
floppyears can I add a submodule, but when I commit my work back to the svn repo, not have that submodule added ?02:29
Eridius using submodules adds a .gitsubmodule file IIRC02:29
floppyears ok02:29
would I add the git submodule on my master git-svn branch or in the topic branch and then merge my topic branch back to master02:30
ffeynman joined02:33
codeFiend left02:33
ffeynman guys, I'm trying to access a git repo over ssh like this: git clone ssh://git@myserver.org:2400/var/git/project.git and it's failing... I'm getting: ssh: myserver.org:2400: Name or service not known any ideas what I'm doing wrong?02:35
zeom joined02:35
johnw can you ssh to that server with that port?02:38
ffeynman johnw: yep02:38
Snadder left02:38
ffeynman i connect etc and get this: fatal: What do you think I am? A shell?02:38
which is good...02:38
johnw i assume you're using ssh -l git -p 2400 myserver.org?02:39
ffeynman yep02:39
johnw what version of Git?02:39
ffeynman i can execute that command just fine from my laptop... but not from my server02:39
$ git --version02:39
git version 1.5.2.502:39
johnw well, your command looks OK to me, so I'm not sure at this point02:40
ffeynman on my laptop I have this: $ git --version02:40
git version 1.5.4.502:40
xenoterracide left02:40
ffeynman weird :)02:40
spearce yea, port number support on ssh was added in 1.5.4 i think.02:40
ffeynman oh02:41
i'm on ubuntu... any idea how to grab latest one?02:41
johnw apt-get install git-core?02:41
ffeynman yep... and thats the one it gave me02:41
Sho_ left02:42
xenoterracide joined02:42
xenoterracide anyone use bzr fast-export?02:42
johnw ffeynman: i build from git's repo, so I'm not sure how to help you with Ubuntu02:43
ffeynman ok02:43
finally it works! I grabbed this version: http://www.backports.org/debian/pool/main/g/git-core/git-core_1.5.4.2-1~bpo40+2_amd64.deb02:44
thx02:44
tswicegood joined02:44
ffeynman it was the old version issue... ubuntu's repo has old version02:44
xenoterracide I'm just looking for basic syntax of how to use fast-export to export bzr to git02:44
brosner joined02:46
lu_zero xenoterracide there is a way not involving taylor to import from bzr to git?02:46
btw02:46
is there a way to import from monotone?02:46
xenoterracide lu_zero: hmm? there is a way? or is that a question02:47
Ramune lu_zero: try tailor http://progetti.arstecnica.it/tailor02:48
lu_zero there is a question02:48
Ramune lu_zero: never tried it myself, so... good luck02:48
lu_zero Ramune ^^;02:48
xenoterracide fast-export has non-existant docs02:49
lu_zero: pmerge-ing :P02:51
lu_zero merging what?02:52
xenoterracide whoops :P02:52
I was looking at the tailor thng02:52
xenoterracide should read more carefully02:53
a-priori joined02:55
xenoterracide lu_zero: oh and yes fast-export is a bzr plugin that should allow exporting to git02:55
I haven't figured out how to use it though02:55
igorgue left02:56
Fullmoon left02:56
spearce xenoterracide: `mkdir foo; cd foo; git init; cd ../bzr; bzr fast-export | (cd ../foo; git fast-import)` ?02:59
csc` joined02:59
tswicegood hey guys - is there a line threshold for copy-and-paste to be detected?02:59
spearce might need to be at least 2 lines to avoid copying "\t}\n" all over the place.03:00
xenoterracide spearce: bzr: ERROR: unknown command "fast-export"... I have the 'fast-import' plugin installed and fast-export is a sub of that... any ideas?03:01
spearce no clue. i only briefly looked at it03:01
Fullmoon joined03:02
xenoterracide spearce: k03:03
tswicegood spearce: seems to be three actually...03:03
lu_zero xenoterracide old version of bzr?03:03
xenoterracide lu_zero: 1.303:04
err... 1.103:04
guess I forgot to run that update a few min AGO03:05
igorgue joined03:05
phoenix24 left03:06
KragenSitaker left03:09
KragenSitaker joined03:12
david_koontz left03:15
wagle hrmm.. why does --shared mean completely different things for git-init and git-clone?03:18
spearce hysterical rasins03:18
queso left03:18
spearce gah, i can't spell today03:19
yea, its one of the warts in the ui03:19
panagos joined03:19
halorgium wonders how to use git pickaxe03:20
dermoth Hi there. I'm trying to use git-svnimport and no matter what uri/path scheme I use all I get is a bunch of "Unrecognized path" errors for each rev.03:20
panagos how do i do an initial empty commit? i tried { git init; git commit -a; }; it didn't work03:20
tswicegood left03:21
wagle why do you want that?03:21
panagos wagle: for completeness :)03:22
Ramune left03:22
wagle you cant ever commit "nothing"..03:23
panagos wagle: i could do that with cogito though :|03:23
RandalSchwartz you didn't *create* the emptyness03:24
wagle i dunno what it mean after trying it myself, then thinking about it03:24
RandalSchwartz that was god's work. :)03:24
you can only commit what you change. :)03:24
wagle means03:24
panagos so how did cogito create empty initial commits?03:25
wagle dunno03:25
RandalSchwartz and since cogito is dead, doesn't matter.03:25
it probably did the primitive calls, bypassing the checks03:25
wagle a commit is a set of changes03:27
asdx joined03:29
iabervon cogito was before anything other than the primitive calls...03:33
spearce wagle: no, a commit is not a set of changes. its a tree snapshot and an optional message.03:33
iabervon Well, and references to zero or more parents.03:36
a-priori left03:36
iabervon Incidentally, is there a good reason we have two commit object parsers in pretty.c, rather than just using parse_commit_header() whenever we need to parse a commit object?03:38
pergesu joined03:40
dermoth Geez, I finally managed to do an svnimport. Looks as though the path component doesn't work, so instead I used the full path for -T, -t and -b03:40
now I gotta be patient as it doesn't look like it's gonna be finished by tomorow morning :)03:43
spearce iabervon: two? both in pretty.c ?03:43
papasi finally figured out how to use git to version control my home dir03:44
iabervon Yeah, parse_commit_header() and pp_header(). pp_header() is older and less clever, I think, but I don't think it's any more efficient.03:44
dermoth papasi, version control your home dir?03:45
spearce iabervon: then probably no good reason other than that nobody has cleaned this up yet.03:45
dermoth I hope you exclude .*03:45
papasi dermoth: something like this http://kitenet.net/~joey/svnhome/03:45
yes ii need to vc all the dotfiles03:46
especially .emacs03:46
iabervon Okay. I'll probably do that sometime soon.03:46
papasi it's non obvious because i cannot use clone03:46
dermoth I use svn yo version-control my config files on my servers, but it's on a separate dir...03:47
iabervon Your .emacs isn't a symlink to working/dotfiles/.emacs?03:47
ilogger2 joined03:49
GyrosGeier joined03:49
pete__c joined03:49
timonator joined03:49
pasky joined03:49
vmiklos joined03:49
CIA-2 joined03:49
Amorphous joined03:49
ThunderChicken joined03:49
sverrej joined03:49
fearoffish joined03:49
Modius joined03:49
mitsuhiko joined03:49
jarod joined03:49
iabervon for i in *; if [ -f $i ]; then ln -s working/dotfiles/$i ~/.$i; fi03:50
IRSeekBot joined03:50
papasi is that a common practice?03:50
iabervon I don't know. I've been doing it for ages, and I know other people version control their dotfiles, but I haven't asked exactly how.03:51
papasi the only down side is that you need to keep the symlink in sync when you discover a new tool03:51
Hobbes joined03:51
offby1 joined03:52
papasi a new tool that creates its own dotfile03:52
iabervon For a lot of them, you want different ones on each machine, so you probably don't want to pick them up automatically anyway.03:53
papasi good point03:53
fred joined03:53
vlt joined03:53
telmich joined03:53
Cap_J_L_Picard joined03:53
pasky left03:53
pete__c left03:53
vmiklos left03:53
timonator left03:53
GyrosGeier left03:53
reuss joined03:53
GyrosGeier joined03:53
pete__c joined03:53
pasky joined03:54
iabervon Looks like I've been doing it since Aug 9 2002.03:54
vmiklos joined03:54
iabervon But making symlinks didn't come until Feb 16, 2004.03:55
papasi i almost lost my svn repo (twice). so i finally take it up and convert everything to git03:55
i think the only thiing i miss is checking out a submodule in svn - the git way is too complicated03:56
iabervon Yeah, I nearly lost my CVS repo sometime between 2005 and 2008.03:56
Git's trying to do several tricky and clever things with submodules, and it's not quite easy yet.03:58
timonator joined03:58
papasi without easy access to sub module, i might need to end up like this guy03:58
http://git.kitenet.net/03:58
creating tons of repo for little things03:59
offby1 ah, Joey!03:59
my inspiration.03:59
papasi it's another list of thing that i _dont_ want to manage03:59
see, using svn i just have one repo with a directory containing everything.04:00
iabervon: so you think git's submodule support will change in the future to make our lives easier?04:00
aLone joined04:00
iabervon Actually, I think that issue will be solved by url.*.insteadOf.04:01
jbunster joined04:01
iabervon You have a directory that contains all your little repos, and configure "url.<path>.insteadOf" to ":"04:02
papasi is this technique currently usable?04:03
iabervon Then you can clone ":something".04:03
Except for clone, which is what you'd really want it for.04:04
When builtin-clone lands (1.5.6, I expect), it'll actually be useable.04:04
kirindav_ joined04:07
iabervon One major point of git's submodules is that the supermodule specifies the particular commits of each submodule that go together, so you can make sure that the projects stay in sync wrt API changes and such. It's therefore no good for just collecting unrelated stuff, but the per-repository maintenance isn't a big deal for git anyway.04:08
jburd joined04:10
papasi yeah but the manual submodule maintenance is a big deal :)04:11
kirindav_ left04:12
iabervon Hmm? Just don't use them like that.04:13
papasi so for now the only feasible way is like what joel did?04:14
iabervon Yeah, but that also works really well.04:15
papasi create tons of repos and remember the the exact path to clone a module?04:15
ec I'm getting a weird error when trying to get-rebase -i HEAD~1004:15
fatal: ref HEAD is not a symbolic ref -- Could not get HEAD04:15
iabervon Well, create tons of repos, and put them somewhere you can get a directory listing.04:16
ec: does "git show HEAD" work?04:17
ec iabervon: Actaully, some other stuff was just weird, so I just removed the whole dir and re-checked out from upstream (I just pushed anwyay)04:17
spearce ec: you are on a detached HEAD.04:17
ec iabervon: thanks anyway lol04:17
spearce: huh?04:17
spearce either upgrade git, or checkout a branch to start the rebase from.04:18
ec upgrade git?04:18
I think I'm on the latest... wha?04:18
spearce are you on 1.5.5? i thought rebase -i was only recently fixed to run on detached heads, but that the fix was included in 1.5.5.04:19
ec Nah, sorry04:19
robinr joined04:19
ec 1.5.3.704:19
I guess that's the latest macports has )-:04:19
What's a detacahed head, anyway?04:19
ec thinks that sounds bad d-:04:19
spearce you current state is just a commit, its not a branch.04:20
this can happen if you do say `git checkout deadbeef...` or `git checkout v1.0` (where v1.0 is a tag)04:20
johnw MacPorts has 1.5.4.504:20
ec hmm04:20
spearce or if you are in the middle of an existing rebase04:20
johnw (for Leopard, that is)04:20
ec johnw: sudo port upgrade git-core #=> ---> Activating git-core 1.5.3.7_0+doc+svn; Error: Activating git-core 1.5.3.7_0 failed: Image error: git-core @1.5.3.7_0+doc+svn is already active.04:21
iabervon Or if you check out something like "origin", which I think is the most common.04:21
johnw ec: what does port info git-core say? I see: git-core 1.5.4.5, devel/git-core ...04:22
a-priori joined04:23
jbunster 0xdeadbeef :D04:23
ec Woah.04:24
This project sure has one CONVOLUTED gitk graph.04:24
I am so putting a screenshot of this on flickr.04:24
robinr left04:24
ec http://flickr.com/photos/elliottcable/2422686718/04:25
jbunster So, when uniting history using grafts and filter-branch ... any ideal way to mark the join point?04:25
ec Hahaha04:25
spearce ec: http://www.spearce.org/2007/07/difficult-gitk-graphs.html04:26
wagle declares spearce the weiner!04:27
asdx_ joined04:27
a-priori left04:27
ec spearce: lmao wow.04:28
iabervon Yeah, that's not convoluted, that's just repetitive.04:28
wagle though i've seen graphs witrh lots of crosses04:28
a-priori joined04:30
kirindav_ joined04:30
iabervon That graph always makes me think of weaving.04:32
ec Uh04:33
I have a commit, my most recent commit, that removes a single file.04:33
I have a commit, two commits before that, that works with files relating to the need to remove that file.04:33
I have a merge-commit between those two.04:33
Every time I try to git-rebase -i HEAD~10, so I can squash HEAD into HEAD^^, I get weird merge errors about random files from the 'middle' merge-commit04:34
wagle the world needs a 3d "flight simulator" viewer for the gitk graph..04:34
a-priori left04:34
ec wagle: agreed d-:04:34
wagle: or a 'roller-coaster simulator' instead!04:34
johnw spearce: wow04:35
ec Anyway, does anybody know how I can merge those two commits?04:36
iabervon I'd do "git reset --hard HEAD~2", remove the file, "git commit --squash", and redo the merge.04:37
Unless the merge was a mess of conflicts.04:37
johnw --squash?04:38
iabervon Er, --amend.04:38
ec is --squash different from --amend?04:38
ah.04:38
iabervon Yes, --squash isn't really a commit option. :p04:38
The commit you're trying to merge into is on the "first parent" side of the merge, right?04:39
jbunster left04:39
ec Yep04:39
Can I check if there is a merge 'going on'?04:40
i.e. I just did a merge, and then manually fixed a conflict.04:40
How would i know, if I came back at a random point of time now, that this was in the middle of a 'merge' commit?04:40
iabervon "git diff HEAD" would look really weird.04:41
wagle git-status tells you, i think04:41
ec Okay. What a convoluted little moment, but it's all cleared.04:42
xenoterracide http://trak3r.blogspot.com/2008/04/git-for-idiots-and-java-developers.html04:42
lol love the title04:42
ec iabervon: So, why was the reversion necessary? Isn't this all what git-rebase -i is exactly for?04:42
iabervon rebasing past merges is not the easiest thing.04:43
priidu__ joined04:45
iabervon It's somewhat unclear as to whether you want to rebase the commits added to the history by the merge on top of the base or what.04:45
xenoterracide http://vim.pastey.net/85948 git import who should I be bugging the bzr repo owners? git? or the bzr-fast-export maintainer04:45
ec iabervon: nah, I don't want to deal with the merge at all - it was just an incidental child of the situation. If I hadn't merged in between the two, it would have been a simple git-commit --amend04:47
iabervon: but since something happened between the two, I would have needed to do an interactive rebase. was just wondering why that wasn't working.04:47
iabervon It's that rebase doesn't know what to do with merges, because it's tricky.04:48
ec ah, so it's a special case with a merge in the middle.04:48
sweet, thanks...04:48
iabervon Yeah. We'll figure it out in the fall sometime and probably make it nice.04:49
cartman joined04:53
shit_burger joined04:55
phoenix24 joined04:58
warthog9 joined04:59
jburd left05:01
priidu__ left05:03
stick joined05:03
cartman left05:05
tjafk2 left05:06
priidu__ joined05:06
Ramune joined05:10
pombreda joined05:17
krawek left05:17
cousin_it joined05:19
stick left05:20
stick joined05:20
codeFiend joined05:21
stick left05:22
ckm joined05:28
ckm left05:30
st1ck joined05:32
st1ck left05:35
stick joined05:35
quik_ joined05:37
com4 I'm trying to apply a patch i made with git-format-patch using git-am. The patch I'm trying to apply adds some new files, but git-am tells me "these files don't exist" Is there a way I can tell git-am to just make the files if they don't exist?05:40
johan-s joined05:41
stick left05:44
stick joined05:45
eternaleye joined05:45
diacoti joined05:53
GarethAdams joined05:54
eetfunk joined05:55
eetfunk_ joined05:57
eetfunk left05:57
phoenix24 left05:58
eetfunk_ left05:58
phoenix24_ joined05:58
eetfunk joined05:59
eetfunk left06:02
diacoti left06:08
eetfunk joined06:10
bousquet joined06:10
eetfunk left06:11
namenlos joined06:13
berto- joined06:24
berto- i ran a git-clone on a repository and now want to make it bare. is there any way to make it bare after the fact?06:24
Pieter git config core.bare true06:26
kirindav_ left06:26
Pieter and then just take the .git directory06:26
saintdev joined06:26
berto- Pieter: cool. So I did: "git config core.bare true"; rm -rf *06:28
Pieter: ah, I can now simply rename .git to <whatever> and it no longer needs to be a hidden dir. thanks, by the way!06:29
_zack joined06:30
johan-s left06:30
dsal joined06:34
anttih- joined06:35
garrin joined06:36
jburd joined06:38
panagos joined06:38
asdx_ left06:43
kixxx joined06:46
dsal left06:52
gitster might be a better idea to also set core.bare = true in .git/config (now project.git/config).06:56
GarethAdams left06:57
devogon joined06:58
deavid joined06:58
phoenix24 joined06:58
phoenix24_ left06:58
johnw joined06:59
xjrn joined07:10
cypher23 joined07:11
aLone left07:11
FunkeeMonk joined07:12
johnw left07:12
aLone joined07:16
mtkd joined07:18
johan-s joined07:21
xjrn left07:25
_zack left07:26
xjrn joined07:26
_zack joined07:27
charon joined07:30
quik_ left07:31
FunnyMan3595 joined07:31
FunnyMan3595 While trying to push: fatal: Unable to create temporary file: Permission denied07:32
... say what?07:32
ChrisMDP joined07:32
FunnyMan3595 Plenty of free space.07:32
ChrisMDP left07:33
bousquet left07:33
FunnyMan3595 Best guess is that there's an old temporary file somewhere that didn't get collected. Any idea where that'd be?07:33
leethal joined07:34
nud joined07:35
codeFiend left07:41
GarethAdams joined07:43
kixxx left07:43
kipras joined07:53
shaftyy joined07:55
rayvans joined07:55
phoenix24 left07:58
phoenix24 joined07:59
sverrej left08:00
_zack left08:03
tjafk joined08:04
zedr0n joined08:05
ssb is "git-repack -a -d -f" a subset of "git-gc --aggressive" ?08:09
ChrisMDP joined08:12
FunnyMan3595 wanders off, noting that git-update-index appears to have mysteriously fixed it.08:13
FunnyMan3595 left08:13
imyousuf joined08:14
ssb looks so, right.08:15
ChrisMDP left08:19
yann left08:22
_zack joined08:23
offby1` joined08:30
rubydiamond joined08:31
kristoffer joined08:34
sverrej joined08:43
offby1 left08:44
cedricv joined08:54
phoenix24 left08:59
zedr0n left09:00
mithro joined09:02
Tv joined09:03
pergesu if I delete a branch, can I "undelete" it?09:05
Pieter yes09:05
if you're lucky you can find the ref in `git reflog`09:05
pergesu and git co a new branch, using that sha?09:06
phoenix24 joined09:06
Pieter yes09:06
cedricv hey all! how could I do the equivalent of "git log -p" only for a given author ?09:08
pergesu how can I find it in the reflog? is there an option I can use that will show the branch? I can't find one09:08
Pieter cedricv: git log -p --author=author09:09
pergesu: there might be something that says "switched to <branchname>", search for that09:09
cedricv Pieter: ooh stupid me.. how could i missed that in the man :p sorry09:10
ebel joined09:10
pergesu Pieter: ah, I see it now. "Reflog message: checkout: moving from mytest to master"09:10
cool, thank you09:10
[tla] joined09:12
leethal left09:13
kixxx joined09:14
johan-s what should I tell people wanting to run git on windows? msysgit?09:15
Pieter probably, unless they run cygwin already09:16
Ramune iirc, cygwin's installer has msysgit in there already.09:17
Pieter that seems unlikely, as msysgit ships with it's own version of perl etc09:17
[tla] "git status" shows currently staged files (I think?). I want to commit a subset of those files so I run "git commit -o <path1> <path2> <pathN>" but git complains "error: pathspec '<truncated-path>' did not match any file(s) known to git.". This is not a valid use for commit?09:18
Pieter what does -o do?09:20
[tla] According to the manpage "--only Make a commit only from the paths specified on the command line" - it can be omitted as it is default.09:21
rubydiamond left09:22
rubydiamond joined09:22
fhobia joined09:22
[tla] Ah, I think I need to use "git commit -- -o <path1> ... etc"09:23
The -- seems important09:23
Pieter that specifies paths.. you can't puth -o after the --, but you probably shouldn't use -o at all09:24
[tla] Is that the right thing to do anyway? I have a load of modifications but I want to bundle them into groups with common commit messages. So I want to commit file1, file3, file5 in one commit then file2, file4 in the next?09:25
Pieter sure09:26
rayvans left09:27
[tla] Pieter: ty 4 help :)09:31
BabelO joined09:32
kixxx left09:33
pergesu can you show the content of a commit, or does it only make sense to diff two commits?09:35
Pieter pergesu: with git log -p, you can see the commit diffed against its parent(s)09:35
namenlos left09:48
kristoffer left09:52
adante left09:55
proxie_ joined09:58
deavid left09:59
proxie__ joined10:01
mithro left10:06
berto- left10:06
phoenix24_ joined10:06
phoenix24 left10:07
chris2 joined10:12
proxie_ left10:15
Sho_ joined10:22
adante joined10:24
mtkd if I've deleted a lot of files, is there a way I can quickly remove them all from the index?10:27
(like using git add . to add all changed files)10:27
Pieter git add -u perhaps10:28
jengelh call git rm instead of rm :p10:30
frodo joined10:32
Husio joined10:34
jburdrabbibulla10:36
Tv left10:37
mtkd jengelh, git rm removes all files though10:39
context uhh no10:39
it removes the files you tell it to10:39
tpope what use would a command be that deleted everyting?10:39
fhobia left10:40
Pieter space efficiency ;)10:40
swombat joined10:46
swombat hmm10:46
"git revert HEAD application.rb" << doesn't work. why not?10:46
d0k joined10:47
Mikachu at least pretend you read the man page first10:47
context git checkout ?10:48
swombat i read the cheat sheet :-)10:48
aah10:48
git checkout. ok, thank you :-)10:48
swombat apologises for utter n00bness on this one.10:48
swombat will pretend he read the man page next time :-)10:48
leethal joined10:52
rabbibulla left10:56
phoenix24 joined11:07
phoenix24_ left11:07
_FrnchFrgg_ joined11:07
nicholas_ joined11:08
johan-s left11:08
Beket joined11:10
bentob0x joined11:14
mithro joined11:14
garrin left11:16
deavid joined11:26
imyousuf left11:41
Rasi joined11:43
Rasi hi11:43
just a quick question, because google is somewhat unfriendly to me11:43
how can i simply clone a branch of some git repo?11:43
jengelh you always clone it all11:44
Rasi hmmm11:44
Mikachu you can use init + fetch to get only one branch, if it's a very large repo / low bandwidth11:44
Rasi this is a small one11:45
jengelh fetch fetches repos too, does not it11:45
Mikachu if you're confused about remote branches, see git branch -r11:45
Rasi but when i clone the git i only get the main branhc11:45
Mikachu and git checkout -b foo origin/foo11:45
jengelh Rasi: you get it all11:45
Mikachu jengelh: if you say git fetch origin foo, it will only fetch the data for the foo branch11:46
Rasi jengelh: i checked.. there is only the main branch.. and its easy to check, because its very small :)11:46
jengelh git branch -a11:46
Mikachu feels ignored11:46
jengelh pats Mikachu11:46
Mikachu purrs11:46
shaftyy left11:48
aLone left11:48
warlock_mza joined11:50
frodo left12:03
QuercusMax joined12:04
Beket left12:05
mcella joined12:06
phoenix24 left12:07
sverrej left12:11
Ademan_ joined12:13
sverrej joined12:14
markkalderon joined12:21
a-priori joined12:24
deavid left12:24
QuercusMax left12:27
lcapitulino joined12:37
aLone joined12:40
a-priori_ joined12:44
a-priori__ joined12:48
FunkeeMonk left12:51
deavid joined12:53
nud left12:54
maddox joined12:58
rubydiamond left12:58
nud joined12:58
aLone left12:59
a-priori left13:00
chrisb joined13:00
a-priori joined13:02
queso joined13:02
a-priori_ left13:02
a-priori_ joined13:04
jlilly_ joined13:06
shoe joined13:06
a-priori___ joined13:07
meyering joined13:07
a-priori__ left13:08
ironfroggy joined13:08
Ademan_ left13:08
Ademan_ joined13:11
meyering left13:11
meyering joined13:12
a-priori__ joined13:13
chrisb left13:16
chris2 left13:17
a-priori____ joined13:18
beeble joined13:19
a-priori left13:20
chris2 joined13:21
Sigurd joined13:21
a-priori_ left13:22
chris2 left13:22
Sigurd left13:22
a-priori joined13:22
chris2 joined13:23
namenlos joined13:26
a-priori_ joined13:27
a-priori___ left13:28
aroben joined13:29
Aeternus joined13:29
a-priori___ joined13:32
a-priori__ left13:33
Aeternus left13:34
Aeternus_ joined13:34
threeve joined13:37
myrizio joined13:39
vbgunz joined13:39
a-priori____ left13:40
a-priori__ joined13:40
a-priori left13:42
proxie__ left13:44
blastura joined13:46
a-priori_ left13:47
krawek joined13:47
a-priori__ left13:47
cannonball joined13:50
zlin joined13:50
a-priori joined13:51
deavid left13:53
a-priori___ left13:55
bartman joined13:56
peeja_ joined13:59
chrisb joined14:02
jtoy joined14:04
jtoy how can i search the repo for old code that used to be used?14:06
leethal left14:11
flazz joined14:11
Pieter You'll have to be less abstract for an answer to that14:12
[tla] Is there a way to "git archive" the current branch including the staged files that are not committed? The use case is while developing I want to cut a tar and try it out on the destination server without having to commit on the development server.14:12
Pieter git commit -m "temp" && git archive && git reset HEAD^14:13
Tv joined14:13
mdp79 joined14:13
jtoy haha, I just laughed very hard Pieter14:13
[tla] trundles off to read what that does ;)14:13
mdp79 sorry, new to git, is there a way that I can list all the files in my current working git repo and the changes to each file, I use git-diff but it only gives me back the changes I've made. Basically i just want a list of the files and whose modified each one and when.14:15
[tla] so HEAD^ is the commit before the last one? Hmm, commit, archive, roll back :)14:15
jtoy Pieter: I am trying to find a string X that appeard anywhere14:15
I tried git log -S'X' but that only shows the commit?14:16
[tla] mdp79: git blame?14:16
xjrn left14:16
Pieter jtoy: git log -p -S?14:16
jtoy nice!14:17
a-priori_ joined14:17
QuercusMax joined14:18
jtoy what does the -p ? I dont have man pages14:18
mdp79 tla: yeah but I just want a list of each file and who modified it, not a line by line rundown for a specific file, basically the tree with each file and who touched it14:19
Ok, basically I'm looking to list files the way github.com does, file name - date modified- and author14:20
I'm sure it's easy, but that I'm just missing it14:20
chris2 left14:21
Pieter jtoy: it gives a patch with the log14:22
Tv mdp79: you would be wrong in assuming that; even github disables that display on larger subdirs, for performance reasons14:23
mdp79 TV, hah, ok so they're doing it file by file. wow14:23
Tv: thanks for the clearing this up for me, I was thinking it was just a one liner14:24
kumbayo joined14:25
a-priori__ joined14:26
FunkeeMonk joined14:28
priidu__ left14:30
priidu__ joined14:31
a-priori left14:32
priidu__ left14:33
johnreilly joined14:33
priidu__ joined14:34
EmilMedve joined14:37
a-priori joined14:38
a-priori_ left14:42
is_null joined14:45
a-priori_ joined14:45
jtoy is there a way to not paginate teh results from git log -s ?14:46
tpope git --no-pager log -s14:47
a-priori___ joined14:49
mdp79 left14:50
lydgate joined14:54
agib joined14:54
yec joined14:55
tommorris joined14:56
kennethkalmer joined14:59
a-priori__ left14:59
namenlos left15:01
a-priori left15:01
namenlos joined15:01
namenlos left15:01
namenlos joined15:02
kewark joined15:03
dbtid joined15:04
pdherbemont joined15:05
dbtid hello; i've misplaced my notes :/ -- i want to update my database to the latest 2.6.2515:05
i tried git fetch v2.6.2515:05
git didn't like that15:05
a-priori_ left15:06
Pieter git fetch; git merge 1.6.2515:06
or rebase, if you want to da that15:06
dbtid how does git fetch know to pull the latest?15:06
Pieter it fetches everything on the remote15:07
dbtid i see15:07
Pieter so also the tags15:07
a-priori joined15:08
charon_ joined15:09
bobesponja joined15:10
charon left15:10
charon_charon15:10
dbtid git diff says there's no difference between one of my files and the one in the database, but it refuses to do the merge15:11
what did i do wrong?15:11
kelvie_ dbtid: I just find it does that sometimes... just add it15:12
dbtid thanks15:12
jpeeler joined15:12
dbtid says it would be overwritten by merge15:12
kelvie_ though it's likely a bug.. if you can reproduce it (and if your code is releasable :o) I'd post it to the list15:12
oh that's different15:12
dbtid Entry 'X' would be overwritten by merge. Cannot merge.15:13
queso how do I specify a cherry pick from another remote?15:13
Pieter queso: first fetch the remote repository, then cherry-pick the sha15:13
dbtid: perhaps you don't track that file, while the remote does?15:13
dbtid since they're the same, how do i tell git to overwrite it anyway?15:14
queso so it figures out the commit it wants based on all the local sha's, eh?15:14
panagos left15:14
a-priori_ joined15:14
pgssgp joined15:15
Pieter queso: you can only cherry-pick local commits, so you have to fetch them first15:16
queso do I need to "git fetch" or "git fetch remote"?15:16
Pieter git fetch remote, if the remote isn't specified as origin15:16
queso yeah15:17
kennethkalmer left15:18
dbtid Pieter: how can i force the copy that is in the database over the file that's in my directory?15:19
git diff says they're the same.15:19
a-priori__ joined15:19
krawek left15:19
Pieter dbtid: if you're not tracking it currently, just delete it on your side15:20
and then merge15:20
dbtid ok, how do i know if i'm tracking it?15:20
i don't recall it as being one i care about15:21
rudi joined15:21
Pieter dbtid: what does git status path/to/file say?15:21
dbtid says i modified it...15:22
Pieter then you'll have to commit that first15:22
or reset it15:22
dbtid why doesn't git diff tell me it's different?15:23
i did15:23
Pieter because you added it?15:23
jengelh git diff --cached?15:23
Pieter try git diff --cached15:23
dbtid git diff p/t/f ?15:23
Rasi left15:23
jengelh ptf?15:23
dbtid lots of changes15:23
if i commit, i'm only committing to my local database, right?15:23
Pieter yes15:24
jengelh you are always only committing to local15:24
rudi left15:24
Pieter oh, that's what you mean by database.. I was wondering about that15:24
mithro[tp]mithro15:24
dbtid sorry15:24
Pieter you should probably use the term 'repository' or so15:24
dbtid ok15:24
someone had used the term 'object database'15:24
so i shortened it; sorry15:24
jengelh repo is the short name15:25
Pieter :) the object database is only part of your repository15:25
dbtid ah15:25
tvachon joined15:25
dbtid git merge v2.6.25 happy chunking away now15:25
rayvans joined15:25
a-priori____ joined15:26
dbtid worked fine15:26
thanks15:26
markkalderon left15:28
a-priori left15:28
vbabiy-laptop joined15:28
a-priori___ left15:29
a-priori joined15:30
leethal joined15:31
queso is there a way to cherry pick only part of a commit?15:33
jengelh you only get the full cherry. otherwise, uh, do something like lemon? :D15:34
Pieter git cherry-pick -n15:34
and then just commit a part15:34
a-priori_ left15:35
tvachon_ joined15:36
tvachon left15:36
kumbayo left15:40
[tp]mithrotp_mithro15:41
tp_mithromithro15:41
mithro left15:41
mithro joined15:42
mithrotp_mithro15:44
tp_mithromithro15:44
mithro left15:44
mithro joined15:45
a-priori__ left15:45
mithrotp_mithro15:46
a-priori____ left15:46
tp_mithromithro15:46
mxNet joined15:46
mithrotp_mithro15:46
a-priori_ joined15:47
ruphy joined15:49
mxNet hi guys. i am a little new, been working with basic svn features before. question about branching. so i create a branch, make changes in there.. then switch to master, and make changes there. but the files on filesystem have my other branch changes in them. i am trying to understand - how am i supposed to jump between branches if the my files keep all the changes together?15:51
beeble commit them15:52
a-priori__ joined15:52
dbtid left15:52
cedricv or stash them (if you want to keep your changes but commit later)15:52
jtoy left15:52
mxNet beeble: cedricv: so if i commit changes in branch B, then just switch to branch A - i will see my files the wya they were before changes in B?15:53
cedricv yes15:53
beeble jes15:53
yes15:53
mxNet thanks!15:54
Ilari mxNet: Not carrying over the uncommitted changes would be destructive.15:54
jtoy joined15:55
mxNet Ilari: you mean, if I neither stash nor commit changes in branch B - branch A will be affected too? but once I do either of those - branch A will go back to normal?15:56
Ilari: this is a little confusing, since i only have one set of files to work with, and switching branches doesn't actually switch the set of files in my working dir15:57
kipras left15:57
mxNet so when i switch branch, i don't get what's happening15:57
threeve mxNet: the files *will* change if the branches are different15:58
Ilari mxNet: Consider what would happen if it just overwrote all files on checkout?15:58
a-priori___ joined15:58
beeble branch A is never abnormal, your changes from the WIP get merged into the other branch if you switch15:58
Pieter anyone ever created a hook that compiles latex documents and puts them somewhere after an update?15:59
Ilari mxNet: And the 'carry-over' behaviour does not affect either branch. It affects working tree and staging area only.15:59
janm joined15:59
Morphous joined16:01
mxNet threeve: beeble: Ilari: i have myFile.txt in master branch. I created a branch "test" , switched to it, and added //foo to myFile.txt. I did nothing else. Then I switch back to master branch - and //foo should not be in the file anymore? is that the case?16:01
threeve mxNet: no, you first need to commit the changes to myFile.txt. They are not in the branch until you commit, only in the working tree16:02
beeble only if you commited it of put it on the stash16:02
mxNet threeve: and working tree is shared among branches, right?16:02
there's only one working tree in other words16:03
threeve mxNet: right, there is only one working tree16:03
Ilari mxNet: Yes, staging area and working tree are shared.16:03
mxNet adding stuff to index is shared too, got it16:03
threeve that is updated as you switch between branches, do pulls, merge, rebase, commit, etc16:03
donspaulding joined16:03
mxNet threeve: i see16:03
tvachon_tvachon16:04
mxNet thank you, i think it's clear now16:04
janm left16:05
a-priori_ left16:05
donspaulding Hi all, I'm new to git, have used svn, and I'm feeling some pain trying to map the old way I did things in svn to git.16:05
beeble just forget them ;)16:06
donspaulding http://git.or.cz/course/svn.html has been a good starting point, but it tells me not to push to a remote that I've cloned elsewhere16:06
a-priori left16:06
beeble you should only push into bare repos16:07
donspaulding what's a "bare repo"?16:07
beeble its a preository without a working tree16:07
codeFiend joined16:07
donspaulding so like svn export?16:07
Ilari donspaulding: Exactly the opposite?16:08
codeFiend left16:08
donspaulding right, just realized that16:08
a-priori joined16:09
namenlos left16:09
janm joined16:09
blastura left16:11
m3thos joined16:11
a-priori__ left16:13
gavinstark joined16:14
m3thos hi there, is there a channel for git-svn16:15
I'm using git-svn and having some difficulties16:15
Mikachu no, but you can ask here16:16
Amorphous left16:17
bobesponja left16:19
lamont joined16:19
charon left16:20
mxNet beeble: you said "you should only push into bare repos"... what do you do with non-bare ones?16:20
beeble: i have repo on github and another local one16:21
beeble that on github is a bare16:21
mxNet beeble: ah , wait, got it16:21
beeble in non bare you should just pull and never push16:21
as in pull from other repos not push into non bare16:22
mxNet beeble: but I can push from my local non-bare into github's bare. i see.16:22
beeble yes16:22
mxNet great16:23
Modius left16:23
Modius joined16:24
ruphy left16:24
a-priori___ left16:24
pombreda left16:25
pombreda joined16:26
sbeyer left16:26
nud left16:26
sbeyer joined16:26
Ilari mxNet: Things non-bare repos are used for: making new commits, doing merges, rebasing commits. And of course you can copy commits from other accessable repositories.16:26
jtoy left16:27
jamesjb joined16:28
kynde joined16:28
axiom joined16:28
mxNet Ilari: i see16:28
axiommajoh16:28
ferdy joined16:28
mxNet Ilari: similar to thinking of non-bare as client+server, and bare - just server16:28
ndim joined16:28
Ilari mxNet: Actually, non-bare as client.16:29
mxNet Ilari: client+server. since you can push/pull from each other16:29
Ilari: and it's practiced a lot16:29
schacon_ joined16:30
Ilari mxNet: Yes, you can push into non-bare repo, but it is best avoided if you don't understand the implications. The same goes for offering non-bare repos for fetch/pull.16:30
leethal left16:32
mxNet Ilari: wait, but what about linus speech and the whole distributed approach, the whole "pulling from anyone you like" and "network of trust". github is just a hub for those not on a network. It seems quite natural to exchange between non-bares both ways.16:32
donspaulding if a bare repo is just git's metadata, and that's what I push to from my laptop, and pull from on my server, how do the files get transferred?16:32
Mikachu joined16:32
dsaxena joined16:32
Tv donspaulding: um, that question seems to be based on flawed premises16:33
donspaulding: try this: http://eagain.net/articles/git-for-computer-scientists/16:33
leethal joined16:33
donspaulding will I be able to read that? I'm not a computer scientist :-P16:34
Eridius joined16:35
mario_ joined16:35
Ilari mxNet: Fetching from non-bare repos can be bad because you should have buffer to fix mistakes. Pushing to them is bad because of what happens if you hit HEAD:16:35
a-priori left16:35
mxNet i see16:35
mario_pygi16:36
Ilari mxNet: The git standard workflow is everybody to have two repos, non-bare one for creating commits and bare one for publishing them.16:36
vbabiy-laptop_ joined16:40
halfline joined16:40
pasky donspaulding: bare repo is not just git's metadata, but git's data )16:40
only the working tree is missing16:40
GarethAdams left16:43
siyb joined16:43
vbabiy-laptop left16:43
ndim left16:43
ndim joined16:43
donspaulding right, it's everything but my actual code, right? we're just arguing definitions then, anything git's storing that isn't my code, I call "metadata"16:43
siyb how do i remove a tag from a specific commit, i tried: git tag -d sometag HASH but somehow it does not work as expected16:44
Ilari donspaulding: Plus staging area is missing.16:44
schacon_ left16:44
Ilari siyb: 'git tag -d sometag'?16:44
johannes joined16:44
siyb Ilari: it tells me that the tag i want to delete can not be found16:44
adante_ joined16:45
Mikachu then you're done16:45
siyb although it is defiantly there16:45
Mikachu you tried to delete two tags, "sometag" and "HASH", one succeeded and the other did not16:45
is my guess16:45
jaalto`` joined16:45
siyb ah ok16:46
thanks16:46
eternaleye left16:46
ndim left16:47
ndim joined16:47
_jarod joined16:47
ricky joined16:47
adante left16:47
adante_adante16:47
jarod left16:49
_jarodjarod16:49
rphillip1 joined16:51
ndim left16:52
ndim joined16:52
siyb left16:57
warlock_mza left16:59
shaftyy joined17:01
Sigurd joined17:04
_zack left17:05
cedricv left17:06
FunkeeMonk left17:07
ebel left17:07
Six_oneone joined17:13
_graham_ joined17:16
booyah joined17:16
pombreda left17:16
booyah is anyone here using git to store dotfiles, and if so, what does your directory structure look like. Do you have one subfolder for the repo and another for a working copy of the files?17:18
or can you have one location for both?17:18
timocratic joined17:18
tpope well, you could run git init directly in your home dir, then use a monster .gitignore17:19
you could have a directory for your config files and then symlink into it17:19
vbabiy-laptop_ left17:19
roylee17 joined17:21
broonie booyah: google for etckeeper, IIRC.17:21
booyah I'm inclined to have a directory for the config files, say $HOME/config, and create symbolic links. Can I create the repo in $HOME/config AND use the files in the folder directly, or do I need to checkout a workign copy.17:21
oh broonie, I'll do that now, thanks17:21
krunk- joined17:22
tpope booyah: a repo has a working directory by default17:22
you don't need to do anything special17:22
etckeeper seems neat17:23
Plouj left17:23
roylee17 Is there a canned git command or option to revert "a single file" of a commit?17:24
angasule_ joined17:24
broonie roylee17: git checkout HEAD <file>?17:25
amfwrk joined17:25
bartman left17:26
eikonos joined17:26
roylee17 broonie: checkout <ref> <file> would lost the changes commited after that <ref>17:27
broonie So you want to undo the changes in some specific older commit rather than to revert to a specific version?17:30
roylee17 I currently do this by 'git revert commit_A' to revert whole commit and 'git checkout HEAD <files_I_don't_want_to_revert>'17:30
yes17:30
bartman joined17:30
roylee17 I wonder if there exists a canned command or option to do the same thing?17:30
_zack joined17:32
aroben left17:36
QuercusMax left17:37
QuercusMax joined17:37
pombreda joined17:37
booyah left17:41
_graham_ left17:44
_graham_ joined17:44
glommer joined17:46
aroben joined17:48
codeFiend joined17:51
Illocution joined17:52
Modius_ joined17:54
spearce joined17:56
codeFiend left17:56
_graham_ left17:56
adante left17:56
Eridius left17:56
sbeyer left17:56
Modius left17:56
tp_mithro left17:56
gavinstark left17:56
spearcegit_spearce17:56
yecstudent_yec17:57
sbeyer joined17:57
blackdaemon joined17:58
zeom joined17:59
many joined17:59
Illocution left17:59
tp_mithro joined17:59
adante joined18:00
coopsh joined18:00
Illocution joined18:00
tp_mithromithro18:01
mithro left18:01
mithro joined18:01
mithrotp_mithro18:02
gavinstark joined18:02
codeFiend joined18:02
Beket joined18:03
schacon_ joined18:03
pgssgp left18:04
Eridius joined18:05
amfwrk I did a git-svn clone ... a long time ago and now I want to strip all the svn data from it, in other words, I want my repo to look like it's been a git repo from the start. Is there a good tutorial or documentation on this?18:07
_graham_ joined18:09
Modius joined18:10
redondos joined18:10
Six_oneone left18:10
Beket left18:13
Ilari amfwrk: Basically, you need to rewrite history to omit those git-svn-id lines, fix authors/committers and perhaps fix merges.18:13
redondos_ joined18:13
Ilari amfwrk: Command to use is 'filter-branch' and its various submodes (relating to commits, since there is no need to modify trees)...18:14
_FrnchFrgg_ left18:14
redondos left18:15
Modius left18:15
redondos_redondos18:15
zlin left18:15
Ademan_ left18:16
amfwrk Ilari: That makes sense, I guess a git-svn clone --no-metadata will be easier18:17
git_spearce left18:17
amfwrk Ilari: Or is rewriting history a better approach?18:17
jerbear joined18:17
Pieter there's an exapmel exactly describing what you want in the git-filter-branch manpage18:17
Ilari amfwrk: But just --no-metadata won't fix other stuff besides those git-svn-id lines...18:18
Tv left18:19
zlin joined18:19
schacon_ left18:21
glommer left18:21
tvachonosaf_tvachon18:21
Sho_ left18:21
Sho_ joined18:22
Ilari amfwrk: To fix up merges, use grafts file to mark them. Then using filter-branch will 'burn them in'.18:22
Tv joined18:22
drizzd joined18:22
Pieter burn!18:22
jengelh roof is on fire.18:23
amfwrk Ilari: Thanks for the advice!18:23
Ilari Monitor has bunred in. :-)18:23
jengelh redpun...18:23
Ilari amfwrk: Most difficult is fixing up authors/committers (assuming you don't have proper mailmap file.18:24
charon joined18:25
pombredaeclipse-pombreda18:25
Ilari amfwrk: Fixing up one author wouldn't be that difficult, but fixing multiple ones is...18:25
Pieter a couple of authors also isn't that tricky18:26
just pipe through a few sed regexes ;)18:26
angasule_angasule18:29
bentob0x left18:30
warthog9 left18:30
modoc I have a conflict merge that I have fixed in the file. How do I resolve the conflict?18:40
Mikachu git add18:42
drizzd modoc: git add file18:42
charon left18:42
charon joined18:43
devogon left18:43
modoc thanks18:44
yann joined18:45
ivanz joined18:45
ivanz Hi. How can I diff a single file between 2 branches ?18:45
charon_ joined18:45
devogon joined18:45
iabervon ivanz: git diff branch1 branch2 path?18:46
charon left18:46
charon_charon18:46
offby1`offby118:46
jengelh diff br1..b2 path18:46
ivanz oh oh I was setting . for current branch18:47
:P18:47
git diff . other-branch path was what I was trying18:47
jengelh diff br1..HEAD path18:47
ivanz how can I say "current branch" ?18:47
HEAD?18:47
ok18:47
Ilari ivanz: for current branch as one of them, you can do 'git diff ..b2 -- path' or 'git diff b2.. -- path' (depending which way you want the diff):18:48
s/:/./218:48
ivanz thank you all18:48
Ilari First is from HEAD to b2, second is from b2 to HEAD.18:48
aroben left18:49
iabervon You might also want "git diff b2 -- path" which uses the current branch commit you're preparing instead of the last committed state on it.18:50
leethal left18:50
fonseca joined18:51
warlock_mza joined18:51
warthog9 joined18:54
a-priori joined18:55
dayo joined19:04
ceez joined19:04
flazz what is the command to put a set of diffs in a mbox?19:06
shd flazz: git format-patch ?19:06
flazz yep19:06
thanks19:06
student_yec left19:07
chrisb left19:09
strangy joined19:11
phoenix24 joined19:13
zedr0n joined19:17
jbunster joined19:18
git_spearce joined19:19
dayo what's so good about git compared to svn?19:19
threeve left19:22
Ilari dayo: 1) Proper merge support, 2) Ability to fix unpublished history, 3) Ability to publish previously unpublished history, 4) ability to have local branches, 5) Way faster, etc...19:23
gregf joined19:24
KragenSitaker dayo: it's decentralized19:26
so it keeps working when your server is down, and nobody can change published history without being detected19:28
also it has working branching and merging support, which is pretty much a necessity for decentralized work19:28
the decentralized aspect also means that everybody gets the benefits of using the source control system, not just an anointed set of "committers", and it's easy to accept changes from third parties19:29
kelvie_ dayo: a bit one is that anyone can submit a patch and have it attributed to them19:30
there's a committer and an author for each commit19:30
s/bit/big19:30
so the committer doesn't take credit for the author's work19:30
Ilari dayo: Also, if you are not 'committer', you can easily do whole series of patches instead of just single diff with SCM tools...19:31
threeve_ joined19:31
m3thos left19:31
janm left19:35
jburd joined19:37
priidu__ left19:38
priidu_ joined19:40
bavardage joined19:44
bavardage how do I restore a file from the git repo19:44
I have forgotten19:44
git ??? /path/to/file19:44
drizzd bavardage: which version?19:45
vbabiy-laptop joined19:45
jburd left19:45
bavardage well I have a file which I have screwed up19:45
and at the last commit19:45
it was ok19:45
I want to get that file back to how it was19:45
Ilari bavardage: 'git checkout HEAD -- path/to/file'.19:45
bavardage thanks19:45
I leave the -- in?19:46
Ilari bavardage: It works in most cases without, but the -- is needed in some cases.19:47
drizzd I believe it's not necessary in this case. But it doesn't hurt either.19:47
jburd joined19:47
tp_mithro left19:48
bavardage right this is wierd19:49
sverrej left19:49
bavardage the file is not updating to the new version19:49
hmm19:49
in gitk19:49
I choose the file19:49
and select newversion19:50
and it looks how I want it19:50
but the checking out remains at the old version19:50
zedr0n_ joined19:50
redondos left19:50
dayo Ilari, kelvie_, KragenSitaker : thanks19:51
redondos joined19:52
hunger joined19:52
ivanz left19:54
ivan1 joined19:54
ivan1ivanz19:54
tibbe joined19:55
threeve_ left19:55
tibbe left19:56
jerbear Tv: when i create a new project in gitosis, the permissions for the repo dir are not permissive enough for them to show up in gitweb19:56
Tv jerbear: yup, and here's the catch19:57
priidu_ left19:57
Tv jerbear: 1) user git has no way to chgrp things to www-data, and shouldn't have 2) some people (me included) aren't happy fully trusting gitosis to make that call, anyway19:57
bavardage so why does the file look fine in gitk19:58
but doesn't pull the correct file19:58
jerbear Tv: all i had to do was allow read by all (a+r, 755)19:58
Tv jerbear: oh eww19:58
hector joined19:59
Tv jerbear: you're making assumptions that are in no way universal19:59
jerbear Tv: you have a point there19:59
Tv jerbear: oh and some people might also choose to run gitweb as user git, too19:59
jerbear: in fact, i'd do that if web servers didn't suck19:59
ReinH Hi folks :)19:59
mxNet anybody here does github + lighthouse?20:00
ReinH How do I create a remote branch? ie git push origin master:new-branch20:00
jerbear Tv: np.. that makes sense20:00
ReinH when new-branch doesn't exist20:00
Tv mxNet: you and just about every other RoR fanboy ;)20:00
ReinH: with new enough git, that should work20:00
ReinH mxNet: I bet #github does, but feel free to ask here of course20:00
Tv ReinH: older ones want you to say refs/heads/new-branch20:00
ReinH Tv: newer than this morning? ;)20:00
Ilari ReinH: That command seems reasonable (except that naming local and remote branches differently causes some minor issues.20:01
zedr0n left20:01
mxNet Tv: funny you should mention, i'm the scripteka guy, working mainly in php at this point. : )20:01
ReinH I could just create a local branch for it actually20:01
yeah, why fight it20:01
ty20:01
mxNet ReinH: thakns for pointer20:01
ReinH mxNet: np :)20:01
mxNet Tv: but definitely planning next project in ror.20:02
:)20:02
Ilari ReinH: Those 'minor issues' are push not pushing it by default.20:02
ReinH Ilari: I'm going with the "normal" workflow, ie creating the branch locally then pushing it20:02
ivanz left20:03
threeve joined20:04
strangy left20:07
tommorris left20:10
aggieben_ joined20:13
phoenix24 left20:13
ec I made some MAJOR changes in a branch.20:14
I want to do a git-merge, but instead of -s ours, I need something like -s theirs20:15
but it doesn't exist.20:15
ivanz joined20:15
ivanz how can I import an exported commit?20:15
want to transfer a single commit over from my laptop and purge the working copy there as it is very old and not needed any more20:15
I exported the last commit with git-gui20:15
Actually I used the gitk "write commit to file"20:16
seancribbs joined20:16
Ilari ec: Create new branch starting from the head of another branch, merge master into it, and then merge the temp branch to master and delete the temp branch?20:17
ctennis joined20:18
seancribbs left20:18
ceez left20:20
zedr0n_ left20:20
Ilari ec: Or do merge -s ours' from that branch, and then merge from master(?) branch.20:21
ivanz meh I will just use patch20:21
ec Hahahaha20:21
My git-status on this commit is 225 lines long!20:22
Mikachu in other words, you changed ~220 files?20:22
git_spearce left20:22
Ilari ec: 'git-status'? It prints about status of working tree or staging area, not about commits...20:22
ec Mikachu: yep20:23
Ilari: yep20:23
bavardage left20:24
Ilari ec: Maybe changing 220 files in single commit (unless most are some autogenerated stuff) is likely bit too much?20:26
ec Ilari: of course20:26
sgrimm joined20:26
ec Ilari: but I was doing a major change across the whole codebase - re-arranging everything, and all the interdependencies had to be updated at once. There was pretty much no way to split it up into multiple commits.20:26
Ilari: So... yay, monster commit of doom! >,<20:26
gavinstark left20:26
cypher23 left20:26
drewish joined20:28
ag joined20:29
gregf left20:29
drewish still getting used to git... i accidentally committed a file deletion. how do i roll that back? to make it more interesting i'm not sure exactly when it got deleted20:30
threeve left20:30
Ilari drewish: You can only restore the file in versions that haven't yet been published...20:30
agib left20:31
drewish Ilari: i'm not sure i understand the published terminology20:31
tommorris joined20:31
strangy joined20:32
phoenix24 joined20:32
drewish git it doesn't keep a copy of the file kicking around in .attic some place?20:32
Ilari drewish: 'git log --diff-filter=D -- path/to/file' to find what commit deleted it.20:33
drewish: No, the problem is that if others have seen the commit deleting the file, you can't just remove that commit without creating a mess.20:34
leethal joined20:35
ctennis left20:35
jbunster drewish: you committed a file deletion? You can revert that commit, but it'll also revert whatever else you did on the commit.20:35
drewish looking at it i think it might have been an earlier merge20:36
jbunster Mmmm.20:36
drewish it's a small enough file that i can just recreate it20:36
but i was hoping to figure out the right way to restore it20:36
Ilari drewish: You could just fetch it from history and commit recreation...20:38
BabelO left20:39
BabelO joined20:39
flazz i have a ton of untracked files. how can do i get rid of them. everything i want is committed, nothing is in index20:39
drewish pardon one more ignorant question, if i've got the commit id when it was initially created how would i checkout that one file20:39
Mikachu you want the commit id when it was deleted20:40
threeve joined20:40
Mikachu well, it doesn't matter if the file never changed of course20:40
Ilari drewish: Maybe better would be last commit containing it.20:40
drewish: Which can be obtained from commit deleting it using parent operator (<commitid>^).20:41
QuercusMax left20:42
drewish Ilari: the git log --diff-filter=D didn't return anything so i wasn't sure when it was deleted20:42
and i don't think it changed since it was first committed20:42
Ilari drewish: Then the first version containg it is probably OK. 'git checkout <commit> -- path/to/file'. The file should appear in 'git status' output.20:43
drewish Ilari: cool! thanks for your help20:45
Ilari drewish: And those files are not saved in '.attic'. Git is not file-based but snapshot-based.20:46
bobesponja joined20:47
agib joined20:47
drewish Ilari: right, i'd just misunderstood and thought you were saying that once a delete is committed the file's gone... that just seemed totally crazy to me.. because it would be ;)20:47
Mikachu isn't it a little bit weird that git-log didn't show the delete commit?20:47
hm or maybe it was detected as a copy/rename?20:48
Ilari Mikachu: But rename detection wasn't asked.20:48
Mikachu he could have diff.renames enabled, but yeah that seems unlikely :)20:48
flazz i tried to merge, it didn't work, i tried a reset --hard, seemed to work but now git status shows me a bunch of untracked and modified stuff20:48
what gives?20:48
Ilari drewish: Actually, I was thinking that you wanted to undo the deletion completely. If the commit that deleted it was pushed, then that can't be done...20:49
drewish: If it was pushed, the only thing doable is committing recreation of the file.20:50
david_koontz joined20:53
johan-s joined20:55
lcapitulino left21:00
[tla] left21:02
d0k left21:03
warlock_mza left21:06
rphillip1rphillips21:07
leethal left21:08
warlock_mza joined21:08
flazz left21:09
eternaleye joined21:09
flazz joined21:09
flazz if i'm merging from a branch into master and i want master to be just like the branch, is there a merge strategy to do this? i see that there is a strategy called ours, but no theirs. any suggestions?21:13
ironfroggy left21:15
Ilari flazz: Create temp branch staring from the other branch, merge master into it using ours, checkout master, merge temp branch into it using normal merge, delete the temp branch?21:15
flazz i'll try it21:16
vbabiy-laptop left21:16
flazz how can i put the checked out master in sync with the HEAD? i have many untracked files that don't let me checkout other branches21:17
rm -rf; checkout ?21:17
Mikachu you don't care about these files?21:17
rm -rf actually won't let you checkout either :)21:18
checkout -f will21:18
flazz no they are cruft from a failed merge21:19
kirindav_ joined21:29
SEJeff joined21:30
SEJeff From an already checked out git repo, you can do git-checkout -b newbranch vSomeTag21:31
How can you check out a tag from the start instead of trunk?21:31
"git checkout git:/repository -b newbranchname vTagName" doesn't work21:32
phoenix24 left21:32
SEJeff And the man pages don't help21:32
Ilari SEJeff: What does 'from start instead of trunk' mean?21:33
Thumper_ joined21:33
phoenix24 joined21:34
SEJeff Ilari, Well I'm trying to pull down the Linux kernel source from linus's git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git but can not for the life of me figure out how to only pull down tag v2.6.20 from that repo21:35
Ilari, Instead of cloning the repo and then checking out the v2.6.20 tag21:35
Ilari SEJeff: Init repo and fetch that tag?21:35
QuercusMax joined21:37
johnreilly left21:37
QuercusMax left21:37
Ilari SEJeff: OTOH, cloning the entiere thing wouldn't be that much larger transfer than just grabbing the one tag...21:38
QuercusMax joined21:38
SEJeff Ilari, Right, but it just seems wrong when you only want 2.6.20 and nothing else21:39
QuercusMax left21:40
Ilari SEJeff: 'git fetch gti://... refs/tags/v2.6.20:refs/tags/v2.6.20' if you really want that...21:40
SEJeff: And that needs 'git init' before that.21:40
SEJeff Ilari, Thanks, the format of that in the man pages was messing me up. I'll try that21:42
Ilari SEJeff: That should be git://...21:43
Mikachu if you only want 2.6.20, download the tarball ;)21:44
SEJeff Ilari, Right, but that didn't seem to work. is the format: repository/refs/tags/v2.6.20:refs/tags/v2.6.20? That didn't seem to work21:44
Sho_ left21:44
SEJeff Mikachu, Yeah, but in git preferably21:44
Ilari SEJeff: There should be space before that first 'refs'.21:44
Sho_ joined21:45
Mikachu and no /21:45
Sho_ left21:46
SEJeff Mikachu, Trying to convince a coworker git is easier than manually using patch and he can't even checkout 2.6.20. He is using a tarball and cursed when a patch had rejects21:47
Sho_ joined21:47
TuxCM joined21:48
SEJeff git-fetch git://... refs/tags/v2.6.20:refs/tags/v2.6.20 2.6.20 and git-fetch git://... refs/tags/v2.6.20 2.6.20 don't seem to work21:49
vbgunz left21:50
SEJeff The manpage says shorthand notation of git://repo tagname localdirectoryname work, but it says couln't find remote ref21:50
Ilari SEJeff: 'git fetch git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git refs/tags/v2.6.20:refs/tags/v2.6.20'.21:50
SEJeff fatal: Couldn't find remote ref 2.6.2021:50
krunk- left21:51
SEJeff git-fetch doesn't work quite like git-clone. Thats been my problem all along. Thanks for pointing out my ignorance21:51
ec Woah, weird shit going on.21:51
Ilari SEJeff: The command I gave above is WORKSFORME.21:51
ec http://pastie.org/18323121:52
"fatal: just how do you expect me to merge 0 trees?"21:52
"warning: You appear to be on a branch yet to be born."21:52
SEJeff Yeah I'd been trying git init; git fetch git://repo tag foldername like you would do with git clone. Thankyou Ilari21:52
ec "error: src refspec master does not match any."21:52
none of those make any sense, lmao21:52
eikonos left21:53
Ilari ec: The commands don't make much sense to me... :-)21:53
ec Ilari: http://github.com/elliottcable/dotfiles/tree/master/home/.profile#L15821:54
Ilari: sorry, they're engraved by force of habit by now )-:21:54
Ilari ec: You need to make commit before you can push it.21:54
ivanz left21:54
ec holy crap, I totally forgot to commit. thanks >,>21:54
ec feels retarded21:54
Tux32 joined21:55
RandalSchwartz you must be a guy - fear of commitment.21:55
ec RandalSchwartz: lol21:56
nbrunson joined21:56
eternaleye left22:02
kirindav_ left22:05
nbrunson left22:09
leethal joined22:14
_zack left22:17
threeve left22:22
a-priori left22:24
slyphon joined22:24
slyphon anyone using gitosis?22:24
i'm having a hell of a time trying to allow people access22:25
eclipse-pombredapombreda22:25
slyphon pokes Tv and hopes he's around (and not busy)22:25
Tv i'm here but it's late22:26
slyphon it seems gitosis is recognizing the username and the repo, but denying access, is there a way to just open up permissions on all repositories to all users with registered keys?22:28
Tv well @all is all known users22:29
there's no "all repos" syntax, on purpose22:29
Mislav joined22:29
slyphon nods22:29
slyphon it's just odd b/c *i* have access to everything, but i also have my key outside of gitosis in the authorized_keys file22:30
so that's not a good test22:30
Tv: can i do [group @all] ?22:31
and give a 'writeable = foo/bar'22:31
Ilari slyphon: If you have such key, how you can say anything of authentication working? And what evidence you have about it 'regcognizing the username and the repo'?22:33
Tv slyphon: [group foo] members=@all22:33
slyphon Ilari: because i'm logged in as a different user22:33
Ilari: and i turned on debug logging22:33
DEBUG:gitosis.group.getMembership:found 'trotter' in 'systems'22:33
phoenix24 left22:34
Tv slyphon: oh, you're one of those people who can't spell22:35
slyphon: hint: dict writeable22:36
slyphon OMFG22:36
slyphon facepalms loud enough for Tv to hear it22:36
offby1 left22:36
slyphon Tv: thank you22:36
Tv: uhm, perhaps i can submit a patch that will stab the user in the face if they make that mistake22:37
Tv slyphon: something like if not config.get('writable') and config.get('writeable'): log.warning("you're stoopid and shuld lern to spel")22:38
slyphon yes!22:38
Tv sry i mean "your stoopid"22:38
Mikachu there's already this22:38
mxNet left22:38
Mikachu % git commit -amend22:38
error: did you mean `--amend` (with two dashes ?)22:38
slyphon sys.kick_in_the_nuts(force='hard')22:38
Tv git commit --amen22:38
slyphon :D22:38
Tv that makes it final22:38
can't amend it after that22:38
slyphon can i get an --amen?22:38
shaftyyy joined22:39
Tv ok off to read a book (again)22:39
slyphon Tv: thanks!22:39
jbunster left22:42
ckoehler joined22:43
ckoehler just a conceptual question:22:43
tracking trunk from svn in master in git, and some branch "slider" in svn in branch "slider_git" in git22:44
if i want to merge those in svn, i can just merge slider_git into master, and then git svn dcommit?22:44
the last inside master, of course22:44
context no22:44
Ilari ckoehler: It won't result true merge in SVN (because such thing does not exist, at least in <=1.4.X).22:45
context git-svn cant keep track of what is suppose to go where when you merge the two trees like that22:45
slyphon left22:45
context since dcommit cant be told WHAT remote branch to push to22:45
Mislav left22:46
ckoehler ah okay22:47
so i have to merge in svn, rebase master, and then i can trash slider_git?22:47
Ilari Actually, it will likely do approximately the right thing (the right thing can't be done), if the merge isn't fast-forward (either because it can't be, or because you force it so).22:47
ckoehler: Also dcommit slider_git first.22:48
context you can cherry-pick and dcommit22:48
ckoehler Ilari: yeah, make sure all the stuff from slider_git is in the svn branch22:49
understood :)22:49
don't really understand what you mean with approximately the right thing22:49
bododo joined22:49
ckoehler i guess i don't know when it doesn't do a fast forward merge22:49
context, never used cherry pick22:50
Ilari ckoehler: It won't be marked as merge in SVN side. Actually, doing 'git merge --squash' would probabl be better.22:50
Eridius_ joined22:50
ckoehler will look at the man page22:50
Ilari ckoehler: As merge --squash will do merge like SVN does it...22:50
david_koontz left22:51
ckoehler Ilari: so git merge --squash acts like svn, where the merge is just there in the working copy until you commit it?22:52
so git merge --squash is git merge without the actual commit?22:52
that's what i understand the man page to mean22:52
Ilari ckoehler: Actually, if merge complates without conflicts, 'git merge --squash' will commit too..22:53
ckoehler: There's seperate --no-commit.22:53
ckoehler ah okay....22:53
hm22:53
confused now :)22:53
Ilari ckoehler: Git-svn is confusing and limited.... :-/22:55
flazz left22:55
ckoehler :) it's that or just svn22:55
just found out that the svn merge to the slider branch was actually an import, not a copy22:55
so it doesn't really matter if i properly merge it back or just merge in git and dcommit22:56
Ilari ckoehler: Because SVN can't represent merges?22:56
ckoehler but nice informative discussion, thanks for that22:56
no, user error :)22:56
svn import instead of svn cp22:56
ah well22:56
not a big branch22:56
shaftyy left22:57
bododo "Better Git it in your soul" (dixit Charles Mingus)22:58
now I know where Linus got the word22:58
RandalSchwartz "git er done!" - larry the cable guy22:58
Eridius_ stabs RandalSchwartz22:58
Eridius left22:58
bododo :p22:58
Eridius_Eridius22:58
Eridius Larry the Cable Guy is the antithesis of all that is good and holy in this world22:58
jpeeler left22:59
bododo RandalSchwartz: you're fine papy?22:59
Ilari 'Goddamn Idiotic Tridge'? :->22:59
a-priori joined23:00
bododo Git It there23:01
ckoehler well thanks for the help guys! this is a very nice channel, thanks for that23:01
a-priori left23:01
cannonball left23:04
ckoehler left23:04
bododo cannonball !! what a nickname!23:05
Mikachu a hell of a lot better than bododo, that's for sure!23:08
jarod :P23:08
bododo yeah! :(23:08
i should have choosen something like... hum..23:08
hey! choose me one23:08
jarod /nick hum23:09
bododoThelonious23:10
Thelonious what about this one?23:10
EmilMedve left23:11
hectorstudent_yec23:11
Theloniousbododo23:14
dsaxena left23:14
spearce joined23:17
anttih- left23:18
leethal left23:26
alkh joined23:34
bousquet joined23:37
devogon left23:40
thiago joined23:40
BabelO left23:44
charon is there a specific rule that lets me use branch/merge "the git way" with git-svn without any strange things happening?23:44
tpope no23:45
charon damn23:45
Ilari charon: Because SVN is too limited.23:45
charon i currently cherry-pick back and forth and it's getting _very_ messy23:45
rayvansaway23:45
SEJeff left23:45
charon well, i'd be fine if there was a way to have git branches off an svn branch, then later merge them back to the svn branch, such that in svn it looks as a single commit that came from nowhere23:46
but there are warnings all over the git svn docs23:46
don't merge, and all that.23:46
donspaulding left23:46
tpope you can merge --squash if you want23:47
shaftyyy left23:54
deavid joined23:57

Logs Search ←Prev date Next date→ Channels Documentation