IRCloggy #git 2010-09-17

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.

2010-09-17

albel727_ Saturn2888: wait a sec, are you trying to setup gitosis?00:00
Saturn2888 yeah.00:00
although it's already setup00:00
albel727_ Saturn2888: why not gitolite?00:00
Saturn2888 bc I have Dreamhost and this is what was given to me by one of my team member00:01
members*00:01
→ Morasique joined00:01
albel727_ ok then, I need you to pastie.org me what you're seeing00:02
Saturn2888 We just want something to upload text files and images for a game interface. I need to get this working before we do it. Lemme do that00:02
→ alex__c2022 joined00:02
← Morasique left00:02
Saturn2888 albel727_: http://pastebin.org/90326300:03
← chronos left00:03
← kenichi left00:03
← Paraselene_ left00:04
← Chillance left00:05
← chrisshattuck left00:05
← nileshtrivedi left00:06
Saturn2888 albel727_: Seems I needed git add . right? So I make changes, add, commit, then push.00:06
albel727_ Saturn2888: ok. "because there are directories inside of it I can see" - I guess you're looking inside gitosis-admin.git on server, right? I need to see content of it too.00:07
Saturn2888: git add - yep, it's needed, of course00:07
Saturn2888: but that's not your problem00:08
Saturn2888: you use git commit -a00:08
Saturn2888: you don't need git add with -a option00:08
Saturn2888 albel727_: http://pastebin.org/90338400:08
albel727_ Saturn2888: it does "add" automatically00:09
Saturn2888 albel727_: well it didn't add it then for whatever reason00:09
→ Paraselene_ joined00:09
albel727_ Saturn2888: and you didn't need. because -a option of commit does it for you.00:10
Saturn2888 hmm, lemme try again with -a and see if it does it this time00:10
→ HarryS joined00:11
Saturn2888 albel727_: I have a feeling git commit -a isn't working because: http://pastebin.org/90361600:11
← dreiss left00:11
→ flazz joined00:12
← engla left00:13
albel727_ Saturn2888: oh crap, I mixed git add -A and git commit -a, again =(00:13
← nachinius left00:13
albel727_ Saturn2888: yeah, then, "git add ." is your way00:13
→ sjohnson joined00:13
sjohnson Q: i did "git stash pop", a merge conflict happened, then left me to my own00:13
Saturn2888 albel727_: last thing then since I think I've got this working. When I put the pub key on the server, it says http://pastebin.org/90371700:13
sjohnson it didnt say that it successfully popped off the stack, and said one file conflicted. if you resolve the 1 conflict, how do you know if the "un-stashing" was otherwise clean?00:14
does it only stop at 1 conflict, then are you susposed to run some command to continue where it left off?00:14
or is it only saying 1 conflict in "git status" because there really is only 1 conflict for the whole stash00:15
← mastro left00:16
← HarryS left00:16
albel727_ Saturn2888: ok, now I'm at loss. I'm no gitosis expert, sorry =(.00:17
Saturn2888 albel727_: I got that error after uploading the public key for this machine meaning, it no longer asks me for a password "good", but the key isn't working00:17
sjohnson "Applying the state can fail with conflicts; in this case, it is not removed from the stash list. You need to resolve the conflicts by hand and call git stash drop manually afterwards."00:18
← DrNick left00:18
← jaeckel left00:18
→ jaeckel joined00:18
← disappearedng_ left00:19
albel727_ Saturn2888: hmm, maybe you didn't allow yourself read/write access in gitosis config, before uploading the key?00:19
Saturn2888 strange, I dunno. Doesn't this show the right permissions? http://pastebin.org/90400100:20
sjohnson i what is the fastest way to git checkout everything? git -reset hard or soft?00:20
Saturn2888 albel727_: maybe it should be [email@hidden.address] instead of just example.com00:21
← jailbox left00:21
→ jailbox joined00:22
albel727_ Saturn2888: yeah, it might be the reason. but in this case, it probably should be [email@hidden.address] i.e username == git.00:23
Saturn2888 albel727_: maybe it has to do with my git config on the client? I possibly setup the keys wrong because I didn't do git config --global user.name correctly00:23
albel727_ Saturn2888: but what actually bothers me, is your gitosis.conf00:23
→ Heimidal joined00:23
Saturn2888 albel727_: what's wrong with it?00:23
→ DrNick joined00:23
albel727_ Saturn2888: are you properly written in it as the one with write access to gitosis-admin.git repo?00:23
Saturn2888 I should have.00:24
maybe I should look at more examples00:24
← marc_kdab_ left00:25
← rbarraud left00:25
albel727_ Saturn2888: did you look at it?00:25
Saturn2888 ah yeah, they're shown as user@domain00:25
← sjohnson left00:25
Saturn2888 albel727_: but in this example, the admin group has gitosis-admin and web.andrewdunn; the latter matching what I had http://blog.agdunn.net/?p=27700:26
← aliasre left00:27
albel727_ Saturn2888: so, in your file, there is gitosis-admin group, which has "writable = gitosis-admin", and "members = you"?00:27
→ mastro joined00:27
Saturn2888 nope00:27
albel727_ Saturn2888: so it's just "writable = somerepo"?00:28
Saturn2888 http://pastebin.org/903384 <- that one00:28
albel727_: writable is the repo. members are the ones with access (I'm assuming the names of the .pub files)00:28
albel727_ Saturn2888: so, are you Main, Grubber, or Colissio?00:29
→ xendo joined00:29
→ ForceBlast joined00:29
Saturn2888 Grubber00:30
and the .pub file is Grubber.octen.pub00:30
← aah left00:31
← ExtraSpice left00:31
→ a9b joined00:32
← stockholm left00:32
← phb left00:32
← blzn left00:33
← d0k left00:34
→ disappearedng_ joined00:34
← disappearedng_ left00:34
→ disappearedng_ joined00:34
albel727_ Saturn2888: ok, what was the command, that you cloned gitosis-admin repo?00:35
...with?00:35
Saturn2888 git clone [email@hidden.address]00:35
oh the admin, gitosis-admin.git instead of ThemeWorks.git00:36
albel727_ Saturn2888: btw, cloning ThemeWorks.git worked?00:36
→ stockholm joined00:37
← Shurakai_ left00:37
→ rubyruy joined00:38
albel727_ Saturn2888: ok, try this00:38
git clone [email@hidden.address]00:38
Saturn2888: are you still there? =)00:40
Saturn2888 yes00:40
← airborn left00:40
Saturn2888 albel727_: worked00:41
albel727_ Saturn2888: yippeee! =)00:41
Saturn2888: that's only natural, now that I think of it =) and how about ThemeWorks.git?00:41
Saturn2888 worked00:42
pretty neat00:42
albel727_ Saturn2888: great =)00:42
Saturn2888 no password needed either00:42
→ artefon joined00:42
← metoikos left00:42
albel727_ Saturn2888: well, you didn't bother yourself with all those public keys for nothing, did you? =)00:43
Saturn2888 :P00:43
I dunno what I should put in the file yet though. I'll find out in a minute. I happened to put a bunch of entries in there hoping one would catch00:43
albel727_ Saturn2888: you mean gitosis.conf? it was perfectly ok the way you showed me before.00:44
← ForceBlast left00:44
Saturn2888 albel727_: oh ok. So I wonder why it works now00:45
question, "Perhaps you should specify a branch such as 'master'.", how do I do that?00:45
albel727_ Saturn2888: the thing is, when you provided server with your public key, the special gitosis handling kicked in, which allows to skip full path to git repos (omitting repositories\ in url)00:45
Saturn2888: you probably making a push, yeah?00:45
Saturn2888: git push origin master00:45
Saturn2888 ooh00:45
/that's/ what that was for00:46
→ ricky joined00:46
← MrSamuel left00:47
← kanru left00:47
cirwin if a "merge commit" has more than one parent — what do you call a commit with more than one child?00:47
albel727_ cirwin: a commit, which is referred as a parent by multiple other commits00:48
← merlin83 left00:48
albel727_ cirwin: a typical example is a forked history00:48
cirwin: why?00:48
cirwin hrm, I was hoping for something more succinct00:48
just writing some documentation00:48
I find that a visual graph containing refs and "fork commits" allows you to see really easily which branches are doing what00:49
← skrul left00:49
→ simplechat joined00:50
albel727_ cirwin: well, there's nothing more to it. the commit itself doesn't even "know", that it is being referred by several others. (it is written in commit internaly, what parents does it have, but not what children)00:50
cirwin well, if there's no accepted name00:51
I'll go with forked commits00:51
I can provide a definition, so I don't really care what the commit itself thinks about this ;)00:51
→ hakunin joined00:51
← SimonP86 left00:52
albel727_ cirwin: well, why invent some new concepts, if git itself doesn't need them to work? =) are you by any chance looking for some command, which allows to draw history graph, or given commit tells what his children are?00:52
→ SimonP86 joined00:52
cirwin albel727_: I have just built a tool to do this, having not found one00:52
and I would now like to write the documentation so that everyone else can understand00:53
albel727_ cirwin: you mean for getting children?00:53
cirwin yes — obviously it will be limited by how much of the log you want to fit in memory at a time00:53
albel727_ cirwin: hmm00:54
Saturn2888 thanks so much albel727_! :) Things seem to be working, I think. We'll see when the team members try00:54
albel727_ Saturn2888: no problem. good luck in your work =)00:54
→ cenuij joined00:55
← cenuij left00:55
→ cenuij joined00:55
cirwin I tend to concentrate on the fork commits, maybe I do git upside down00:55
→ ehabkost joined00:56
albel727_ cirwin: hmm, strange though it might feel to me realising it, but I think your approach might be more natural.00:57
← intripoon left00:58
cirwin interesting :)00:58
albel727_ cirwin: on the other hand, during git work, people are usually notice not when they've forked, but when they're doing merge, so... =)00:59
← tchan left00:59
cirwin interestingly enough, merging needs to know the fork commits too :)00:59
it just does it for you00:59
amason morn all, run in to an issue with a commit to my local working (non-bare) repository when i run git commit; error: unable to create temporary sha1 filename : Invalid argument01:00
→ tchan joined01:00
amason what would cause this problem ?01:00
albel727_ cirwin: well, if we call forking commits - the one with many children, as opposed to merging commit - the one with many parents, then users need to know how to produce the latter, not the former =P01:00
amason the error messages don't give any indication of what is actually wrong.01:01
albel727_ amason: unix, windows?01:01
amason well ...linux01:01
cirwin heh01:01
it's too confusing01:02
albel727_ amason: file permissions on .git subdirectory are ok?01:02
amason well it's almost a brand new clone so they should be, will check01:02
albel727_ amason: (and all containing files/directories ofc)01:02
← mquin left01:02
albel727_ amason: really strange, indeed01:02
amason ok no you're right. some of the objects are owned/group'd to me but are read only01:03
any ideas how / why that would occur ?01:04
[-r--r--r-- amason amason ] 36b320cbccd02a600b509b846d7d6c9eed4a0e01:04
umask is 000201:04
→ Yuuhi` joined01:05
albel727_ amason: hmm, I'm not sure. do you work with this repo over ssh or git-daemon?01:05
← disappearedng_ left01:05
amason albel727_: no actually the remote (bare) repository is located on the same file system. /git/somerepo.git01:06
← Yuuhi left01:06
← Saturn2888 left01:06
amason i do run a git fetch every now and then to see what has changed.01:06
i usually do a git fetch and then git merge origin/branchname01:07
albel727_ amason: ok, let me guess.01:07
amason but i was running into similar issues pushing. so i re-cloned the repo01:07
albel727_ amason: you've created a bare repo, and then cloned repo01:07
albel727_: locally01:07
amason yep01:08
thats correct01:08
is that not a good idea ?01:08
albel727_ amason: I'm a bit rusty on unix, but don't file permissions are keeped in inode? so if you're hardlinking, then permissions are shared between all instances?01:09
← stliu left01:10
amason hmm i see what your saying.01:10
albel727_ amason: I'm asking that, because when doing local clone with git, it plays smart and hardlinks objects from cloned repo.01:10
→ mquin joined01:10
albel727_ amason: you can get rid of that in two ways01:10
amason: 1) clone, specifying an option forbidding linking 2) clone via ssh or any other "remote" protocol01:11
let me see01:11
man git-clone01:11
Gitbot albel727_: the git-clone manpage can be found at http://git.or.cz/man/git-clone01:11
amason ok awesome, i will try that.01:11
→ hebz0rl__ joined01:11
← Bucciarati left01:12
albel727_ amason: ok (the option is --no-hardlinks) =)01:12
← hebz0rl__ left01:12
jast or clone using a file:// URL01:12
albel727_ exactly01:12
amason oh ok01:13
← tildeequals left01:13
amason so why does git have this behaviour then ? surely if it's saving space but people can't commit to the repositories thats going to cause problems?01:14
← hebz0rl_ left01:14
jast I don't know what the problem is. object files are not supposed to be anything but read-only anywya.01:15
albel727_ amason: well, I guess, supposed use-case was when you're having several working clones of the same project on your hard-disk. =)01:15
← alvaro_ left01:15
albel727_ amason: so you can save your space01:15
→ jalvarado_ joined01:15
← jalvarado_ left01:15
offby1 amason: my hunch is: the code required to fix the problem is significant; whereas the effort on your end to avoid the problem is minor.01:15
amason ah ok, but when there are multiple users on the same box it causes issues ?01:15
offby1 amason: with multiple users on the same box, I'd consider gitolite over ssh, even though that seems like overkill.01:16
If ever someone wants to work on some _other_ box, you're all set.01:16
← SimonP86 left01:17
offby1 disk space is cheap; don't tie yourself in knots trying to save it.01:17
jast I don't see any problem even if they are different users01:17
albel727_ amason: well, probably. but I'm bothered more with what jast have said. Objects are really supposed to be read-only. If there are problems with permissions on directories, then maybe, but otherwise we probably didn't solve your issue yet =(.01:18
← jalvarado left01:19
jast and directories are obviously not hardlinked from the source repository :)01:19
← ehabkost left01:20
albel727_ jast: yep, that's what I have in mind =/01:20
offby1 there was a time, long ago, when you _could_ make hard links to directories.01:20
at least, root could.01:20
I think it led to the ensual of hilarity.01:20
albel727_ amason: ok, I recommend you to check, that all files have correct user/group ownership. since ssh probably works under git/nobody or something, which might be the problem.01:22
jast I would do a few more in depth checks, but I'm too tired to deal with this right now. sorry.01:22
instead I'll just sneak away, hoping that I manage to reach my bed without falling asleep on the way. bye. :)01:23
← spearce left01:23
→ scottj joined01:25
scottj if I've forked on github and committed and now I want to pull changes from original that I forked from, do I pull or rebase?01:25
albel727_ jast: bye =)01:25
amason heh ok thanks i will try and change the perms01:25
and then re-clone using the suggested method01:25
albel727_ amason: sorry, I really don't know the problem =/ hope this helps.01:26
amason well you've helped just by giving me some more information about how git works01:26
← xendo left01:26
amason so if nothing else it is valuable and i thank you for your time and effort01:26
albel727_ amason: that's only natural =) no problem.01:27
← Bombe left01:28
→ Bombe joined01:29
← dngor_ left01:31
← zombor left01:31
← ed_ruder left01:31
→ zombor joined01:31
→ dngor joined01:32
→ hyperair joined01:35
→ cbeust joined01:37
← pom left01:37
← brntbeer left01:38
→ Pacman_Home joined01:38
→ bphogan joined01:38
← alex__c2022 left01:39
→ kanru joined01:41
← ricky left01:41
→ rbarraud joined01:42
→ chronos joined01:45
← __name__ left01:45
← linusoleander left01:48
→ stringo0 joined01:48
→ tungtn joined01:50
→ Titosemi_ joined01:51
← Titosemi_ left01:51
← Titosemi left01:53
← dr_win left01:57
→ dr_win joined01:58
← cirwin left02:00
← mquin left02:01
← dr_win left02:02
→ jaql joined02:03
← hyperair left02:03
← bcardarella left02:03
→ robgleeson joined02:06
offby1 scottj: that's up to you :-|02:09
scottj: depends on whether you dislike merge commits.02:09
→ mquin joined02:09
← Elfe left02:11
→ Elfe joined02:14
→ nadavoid joined02:18
← kukks left02:20
→ alester joined02:20
→ tty1 joined02:21
← programble left02:21
→ xiangfu joined02:22
→ devanjan joined02:23
← zenom left02:23
← bandu left02:24
← stockholm left02:27
← ryec left02:28
→ User_007 joined02:28
User_007 Hey guys, i have a doubt: can use git to download (and update) just the last linux kernel?02:29
← stringo0 left02:30
→ stockholm joined02:32
User_007 How can i use git in order to download and upgrade just the last linux kernel? Is there a way to do that?02:32
→ bsmntbombdood joined02:32
amason User_007: you can clone the kernel repository02:38
the compiling / upgrading will be up to you as git doesn't do that02:38
but it will allow you mirror the kernel repository02:39
User_007 amason, can i download just the last stable and update it?02:40
amason when you say last stable, what exactly do you mean ?02:41
the tagged 2.6.3502:41
?02:41
User_007 yes02:41
→ stliu joined02:41
← stliu left02:41
→ stliu joined02:41
amason what do you want to update it to ? or do you mean make local modifications ?02:42
→ merlin83 joined02:42
← merlin83 left02:42
→ merlin83 joined02:42
User_007 like, if it changes from 2.6.35 to 2.6.36, can i do just a update and get just the changes02:42
→ jtaby joined02:42
amason yeh you could do that. you would probably have to merge that branch though02:43
but yes02:43
← Heimidal left02:43
→ jtaby_ joined02:44
User_007 amason, where can i find how to do that?02:44
← jsquared left02:44
→ jsquared joined02:45
amason well first git clone (repo) where repo is the kernel repository02:45
guessing kernel.org will have that information02:45
User_007 amason, won't it download all the sources? (since 2.6.0)02:45
→ Heimidal joined02:45
amason User_007: not entirely sure to be honest.i haven't looked at how their repo has been set up02:47
i imagine some of those are submodules02:47
kernel.org does provide patches for each change02:47
minor revision02:48
← nairb774 left02:48
amason which you could apply manually if you want.02:48
if your not indending on writing your own additions to the kernel that could be an option.02:48
← jaql left02:50
← brainbox1100 left02:50
→ nairb774 joined02:50
User_007 amason, i just want to give a command and get the new source.... i got used to do it with svn with some apps02:50
amason yeh git works a bit differently to subversion. you may be able to do something similar with the kernel repo, it really depends how it's been set up02:52
→ lru joined02:52
← tchan left02:52
→ timj_ joined02:52
amason you could track linus's master branch, i imagine that would be very similar but as to how much you going to have to download initially i really don't know02:52
really depends on how the repository has been set up02:52
lru I want to track a SVN repo that happens to have externals in it... so I'm also using git-submodule... this leaves me with a top level repo that contains a submodule commit that I never want to git svn dcommit.... help?02:53
→ stringo0 joined02:53
amason User_007: so you will need to look at this for your self. Once you have the repo cloned you should be fine02:53
User_007 ok amason, ty a lot.. i will get a look on it02:53
← batok left02:53
amason np02:54
← stliu left02:54
→ stliu joined02:54
← stliu left02:54
→ stliu joined02:54
← ph^ left02:54
← timj__ left02:56
→ tchan joined02:56
← User_007 left02:57
← Heimidal left02:58
→ Thell_ joined03:01
← orafu left03:02
→ orafu joined03:02
→ chrisshattuck joined03:02
← joeytwiddle left03:05
→ SethRobertson1 joined03:06
→ sshc_ joined03:06
← merlin83 left03:07
← artefon left03:09
→ merlin83 joined03:13
← merlin83 left03:13
→ merlin83 joined03:13
→ comptekki joined03:14
comptekki if i have a repo on github and have a machine at work that i use to push work to03:15
→ sshc joined03:15
comptekki after cloning to home laptop how do I then push the laptop changes up03:15
offby1 "git push", probably.03:16
SethRobertson1 Normally when you clone, the master branch is automatically set up as a tracking branch03:16
Short answer, yes, `git push`03:16
comptekki can i add my laptop as a valid system to do push since it gives me an error03:16
offby1 if that goes to the wrong box, then add the other box as a remote with "git remote add"03:16
comptekki: you can. It'd help to see a transcript, though.03:17
← stringo0 left03:17
→ alex__c2022 joined03:17
← sshc_ left03:17
→ zombor_ joined03:17
comptekki git remote add - ah yes thx03:18
← zombor_ left03:18
← scottj left03:18
← flazz left03:19
← comptekki left03:19
← zombor left03:20
→ gnufied joined03:24
→ zombor joined03:25
→ eletuchy_ joined03:26
← steph021 left03:27
→ mithro joined03:27
← bphogan left03:27
← jtaby left03:29
← eletuchy left03:29
→ hoggarth joined03:30
→ Raimondi_ joined03:32
← Raimondi left03:32
→ bphogan joined03:33
SethRobertson1 I have a meta-git script we use extensively and I am wondering if it would be useful to the broader community enough to clean up for submission (and where the best place to submit would be).03:33
The script creates a group of related repositories all of which are concurrently developed on and on which all git operations should operate, so when you branch, each repository in the project is branched in turn. git-submodules went a different way where you were referring to a particular remote repository.03:33
The meta-script combines (and occassionally processes for special commands) the git output to make everything clearer and we have hacked the email notification script so we don't get spammed with mail from each repo during tag operations.03:33
However, some modules may be in multiple projects, so it doesn't necessarily make sense to try to make one gigantic respository out of them. Partial checkouts might in theory be another way to accomplish this.03:33
Doing it this way does have downsides: you can have partial success and failure, programs like `gitk` will not show the global systems' history only a specific repository's, and things can get a little squirrelly if different branches/tags have different slave modules associated with them and you checkout back and forth.03:33
However, we have not had any significant problems.03:33
So, is this interesting and where should it be announced once it is ready (it is solid now and has been used for 2+ years, but perhaps is not as documented as one might hope for a public release). It can be downloaded at http://www.baka.org/projects/gits.git !03:33
→ FYLGood joined03:34
offby1 dunno where it should be announced; it indeed sounds interesting.03:35
It sounds vaguely like what Google uses for their android stuff03:35
I'd post to the mailing list.03:35
SethRobertson1 ok03:35
← urlicht left03:39
← eletuchy_ left03:39
→ eletuchy joined03:40
← bphogan left03:43
← hoggarth left03:45
→ peteonrails joined03:45
peteonrails if i have an app that has config files but i have changed those config files to suit my local dev env .. how can i keep those files in the repo but not get overridden when i pull or push ?03:46
← errordeveloper left03:46
→ nakkor joined03:47
← alex__c2022 left03:47
SethRobertson1 Putting an intermediate branch between your workspace and the tracking branch, I imagine you could set something up with git-rerere. However, it probably would be easier to stash before and pop after in some kind of wrapper script.03:47
→ bphogan joined03:48
→ errordeveloper joined03:49
peteonrails if i add a file to .gitignore it takes it out of the repo right?03:49
and when i pull it would get overridden ?03:49
← webus left03:51
SethRobertson1 It does not appear that .gitignore overrides a committed file. Mods are still reflected in `git status`.03:51
peteonrails SethRobertson1: so say we have a file config.xml thats crucial to the app... i have tweaked it so it works locally.. if i gitignore that file does that mean my local copy will always stay the same and not affect the shared repo?03:52
SethRobertson1 I just did a quick test. `echo README >> .gitignore; echo foo >> README; git status`03:53
For a committed file README the .gitignore had to effect that I noticed. For an uncommitted file it worked as expected03:53
peteonrails sorry still dont understand03:54
← nakkor left03:54
SethRobertson1 Run these commands (after I type them out and test them)03:55
peteonrails ok03:55
SethRobertson1 mkdir foo; cd foo; git init; echo foo > README; git add README; echo README > .gitignore; echo random >> .gitignore; git add .gitignore; git commit -m "Initial file"03:56
This represents a repository you have cloned from upstream. We are ignoring that it is not actually attached anywhere.03:57
Now let us make a local configuration file change.03:57
peteonrails ok that makes sense03:57
SethRobertson1 echo "Local change" > README; git status03:57
peteonrails yup03:57
so even tho it was added to .gitignore when i do git status its coming up as modified03:58
→ Heimidal joined03:59
peteonrails SethRobertson1: right? so if i was to do a git add . ; git commit "test"; then my local chages to that file will go to the repo?04:01
← SethRobertson1 left04:02
→ SethRobertson1 joined04:02
SethRobertson1 That was odd04:02
peteonrails SethRobertson1: timed out apparently04:03
SethRobertson1 How could I have a 272 second timeout when I was talking a few seconds before?04:03
peteonrails who knows04:03
SethRobertson1 git reset --hard; echo "Program file" > test.c; git add test.c; git commit -m "Program files"04:04
This sets up a non-config file for us.04:04
peteonrails yup04:04
← zombor left04:05
← _sh3 left04:05
peteonrails but wait, we created a readme file... added that to .gitignore and then i did a git status04:05
and it comes up as modified04:05
→ zombor joined04:05
peteonrails so when i did that commit then it would have pushed those changes (potentially config ones) to the repo?04:05
SethRobertson1 Yes, thus .gitignore is not the droid you are looking for04:05
peteonrails SethRobertson1: dam04:05
back to the drawing board, any ideas?04:06
SethRobertson1 echo "Local change 2" > README; echo "Program change" >> test.c; git status04:06
→ scocos joined04:06
SethRobertson1 At this point we see the "config" change and the real change, the program change.04:06
git add test.c; git stash --keep-index; git status; git commit -m "Program change"04:06
peteonrails yeah a stash is one way04:06
but i was wondering if there is a better option04:06
SethRobertson1 We specifically add `test.c` to the index, then stash everything else (i.e. the config change), then commit the index04:06
← Thell_ left04:06
SethRobertson1 `git stash pop` will bring back the config change04:07
→ ed_ruder joined04:07
SethRobertson1 The only other solution I can think of is to have a translucent filesystem or symlink tree or something where you do your testing. That testing will see your hand-crafted config file. However, you do you git operations in the real copy of the tree04:08
scocos hi, is there anyone online now who knows a decent amount about submodules?04:08
peteonrails SethRobertson1 right... hmm04:08
→ Adaptee joined04:08
peteonrails is there a way to git stash all modified files?04:08
→ webus joined04:08
SethRobertson1 Yes, that is what it does by default. What I was hoping for was a way to stash specifically named files. I didn't see that.04:09
scocos: I know a very little. Mostly enough to know they didn't work for me.04:09
scocos I can't figure out why whenever I submodule update, git needs to take my submodules off their current branch? If I'm on master, and I pull, it does a fetch/merge. if I submodule update --merge, I still seem to end up off master04:09
SethRobertson1 Sorry, perhaps someone else.04:09
scocos Also, whenever I make changes in a submodule, commit, push, and go to the supermodule, I git add the changed subdirs, push, and would think I'm then in sync with origin but it seems I get stuck in a permanent state where I can keep updating and merging new changes.04:11
SethRobertson1 peteonrails: You can probably use a second branch to solve your problem, however, then in order to commit you would have to `commit, checkout, merge, push, checkout`04:12
← scarabx left04:13
→ dreiss joined04:14
peteonrails SethRobertson1: ok thanks for the help04:15
appreciate it04:15
← pedrobelo left04:16
→ tildeequals joined04:16
← jamesmacleod left04:16
→ cytrinox_ joined04:20
→ pedrobelo joined04:20
→ mofaph joined04:20
→ roop joined04:21
← pedrobelo left04:22
← SunilThaha left04:23
← cytrinox left04:24
cytrinox_ → cytrinox04:24
→ pedrobelo joined04:24
← pedrobelo left04:24
→ pedrobelo joined04:24
← daed_ left04:25
← nadavoid left04:27
← Yuffster left04:28
← zombor left04:31
→ radioman-lt joined04:33
← TheEmpath left04:33
← peteonrails left04:33
→ rubyruy_ joined04:34
← rubyruy left04:37
→ spearce joined04:39
→ amason_ joined04:39
← amason left04:39
← emmanueloga left04:39
← spearce left04:42
← scocos left04:43
← SethRobertson1 left04:44
→ SethRobertson joined04:45
→ slonopotamus joined04:47
→ skippyish joined04:49
→ ajpiano joined04:53
← welly left04:55
← mofaph left04:55
→ mofaph joined04:57
← stockholm left04:57
→ Radar joined04:57
← Khisanth left04:58
→ ricky joined04:59
→ freimatz joined05:01
← Fangorn_ left05:01
← stringoO left05:01
→ carrus85 joined05:01
→ Khisanth joined05:01
→ Fangorn_ joined05:02
→ brntbeer joined05:05
← brntbeer left05:05
→ stockholm joined05:05
← tchan left05:07
← ricky left05:07
← gnufied left05:09
← devanjan left05:10
→ discordant joined05:11
→ tchan joined05:12
→ justin-george joined05:12
← chronos left05:13
← avinashhm left05:15
→ WarpCore joined05:15
← Fangorn_ left05:15
WarpCore Is there something for git like Trac for SVN?05:15
← Eridius left05:18
SethRobertson git integrates decently well with most bug tracking software. It is usually pretty easy if there is an email interface.05:22
→ posciak joined05:22
WarpCore I guess I will write my own bugtracker05:22
..05:22
.05:22
slonopotamus why?05:23
WarpCore I want a bugtracker that has revisions and other good shit.05:23
SethRobertson We process via git hooks on the central repository and if the commit message matches a regex send the email off in the right format to the embedded bug number. We have done this for RT and bugzilla. There are plenty of such integration hook scripts lying around05:23
→ aspotashev joined05:24
slonopotamus WarpCore: "a bugtracker that has revisions" -- what's that?05:24
WarpCore Basically storing patches and other stuff in git.05:25
slonopotamus maybe you just want a maillist and git am? :)05:26
← phantomcircuit left05:26
SethRobertson ticgit does this05:27
you can use git with trac instead of subversion05:28
→ ricky joined05:28
→ Fangorn_ joined05:28
← Fangorn_ left05:28
→ urkud joined05:29
→ jtaby joined05:29
→ giallu joined05:29
→ doener joined05:31
← hobodave left05:32
← webus left05:33
← carrus85 left05:33
← giallu left05:33
WarpCore I have used Git with Trac, but it breaks with a drop of a hat.05:35
→ anjen joined05:37
amason_ WarpCore: what are your requirements . i'm writing bugtracking software that is based on git05:38
i.e it stores the bug information in a nother branch05:38
with seperate history05:38
so none of your files from the master repo are actually in the branch with the bugtracking05:39
and it can be pushed just like any other branch05:39
→ hyperair joined05:39
amason_ there are a couple around but they all use use just a file or if they do use a branch they have been inactive for some time05:40
and the code is ......interesing =)05:40
copy paste etc..05:40
← aspotashev left05:40
SethRobertson How is this different from ticgit?05:41
→ aspotashev joined05:41
→ psankar joined05:41
← psankar left05:41
→ psankar joined05:41
← FYLGood left05:42
→ Athox joined05:42
amason_ SethRobertson: i tried ticgit but there was some reason we decided not to use it, i'll look at it again.05:42
→ rbarraud_ joined05:42
← rbarraud left05:42
← Anti-X left05:43
amason_ SethRobertson: does it require ruby gems or something ?05:43
← Eaven left05:43
amason_ something that wasn't packaged ?05:43
SethRobertson It does use ruby I think. I've never used it.05:44
← slonopotamus left05:45
amason_ i will have to look again. some of them were quite good05:46
→ ph^ joined05:46
amason_ i think both the ruby ones were good05:46
just they required me / the sysadmin / whoever to install a bunch of stuff that wasn't packaged and i couldn't work out how to package it05:46
it hadn't really been written for that use case i don't think05:46
which is fair enough, but would't be something we could deploy in our org05:47
← ajpiano left05:48
← sivy left05:48
← jtaby left05:48
→ sivy_ joined05:49
← lru left05:50
→ Tuomas joined05:53
← sivy_ left05:54
← alester left05:55
→ frogonwheels joined05:55
→ apsysoev joined05:56
→ phb joined05:58
← ed_ruder left06:00
← madewokherd left06:02
← AoiKage left06:03
→ ntoll joined06:04
← shennyg left06:04
→ madewokherd joined06:04
→ avinashhm joined06:04
→ ajpiano joined06:05
← anjen left06:05
← ajpiano left06:06
→ ajpiano joined06:06
→ sambarino joined06:07
→ brainbox1100 joined06:11
← brainbox1100 left06:13
→ brainbox1100 joined06:13
→ opalka joined06:14
→ jtaby joined06:15
← zenazn left06:16
→ cbreak_work joined06:16
→ hobodave joined06:18
← rgr left06:19
→ rgr joined06:20
→ guardian joined06:22
→ airborn joined06:22
← stockholm left06:24
→ giallu joined06:25
→ kenneth_reitz joined06:27
← kenneth_reitz left06:27
→ kenneth_reitz joined06:28
← Mage left06:32
← kenneth_reitz left06:32
← bphogan left06:34
← giallu left06:35
→ giallu joined06:35
→ stockholm joined06:38
← krawek left06:39
→ relachs joined06:39
→ bphogan joined06:39
→ Torsten joined06:39
← anli__ left06:40
← sambarino left06:40
→ sambarino joined06:42
→ thiago_home joined06:42
← jrmuizel left06:42
→ radarek joined06:42
← WarpCore left06:42
← stockholm left06:42
← jarh left06:43
→ eban joined06:44
← jtaby left06:44
rgr can I abort an interactive rebase that failed?06:46
aha "abort"06:46
← guardian left06:47
→ sambarino1 joined06:47
→ Alagar joined06:47
→ devanjan joined06:48
→ gnufied joined06:48
← mofaph left06:49
rgr its not obvious how I push to my remote repo after rebasing as the remote repo has all the discarded commits.06:51
← sambarino left06:51
← sambarino1 left06:51
→ anli__ joined06:52
cbreak_work the key word is force06:52
thiago_home rgr: use the force06:52
← eletuchy left06:52
→ typemore joined06:52
→ alkemann joined06:52
typemore I know how to use git to diff two commits06:52
← alkemann left06:52
typemore how do I make it into something I can feed to patch?06:52
thiago_home remember that forcing will not help pullers, though. They need to reset and re-do any changes they may have done, on their own.06:52
there's nothing to help them there.06:52
typemore: git format-patch06:53
→ metoikos joined06:53
typemore whoa06:53
that is too cool06:53
→ eletuchy_ joined06:54
← eletuchy_ left06:55
→ eletuchy joined06:55
← anli__ left06:57
← Heimidal left06:57
mcnellis I'm making branches when I complete a release. e.g. just did git branch version-1.0.5 now that I'm done with all these tasks. Now I want to push this release to production i.e. I think I want to checkout this specific release06:57
this specific branch*06:58
it pulls from a central repo though generally06:58
cbreak_work why do you make a branch when you are done?06:59
I would make a branch before starting06:59
← eletuchy left06:59
cbreak_work makes much more sense to me06:59
← Alagar left07:00
← Radar left07:00
→ skoop joined07:02
← aspotashev left07:02
mcnellis Yeah i Just kinda realized that heh07:03
I'm a noob to git07:03
so there is a learning curve to me using branches07:03
only the 2nd one I've made. Anyway so on my local machine I did git branch version-1.0.5 and then git push remote_site version-1.0.507:04
cbreak_work what I do after branches are completed is to merge them into master07:04
mcnellis where remote_site is like my central repo (Using unfuddle to host it)07:04
cbreak_work if they were important, I tag them07:04
if not, they vanish07:04
and if a branch dies without getting merged, I tag it and delete the branch07:05
mcnellis so when you start a new version you make a branch, work on the branch and then how do you merge it into master?07:05
cbreak_work I make branches for each new feature, not for versions07:05
but you can do what ever you want07:05
mcnellis I'd like to keep it in versions, but do I merge a branch into master and then would I just push to the master as normal?07:06
← Wicked left07:06
→ kenneth_reitz joined07:08
→ anli__ joined07:08
→ voxpelli joined07:10
← Amorphous left07:11
amason_ mcnellis: why not just make tags for versions07:13
?07:13
that way you aren't locked into having a branch stick around.07:13
mind you there isnt much difference really07:14
← reflectivedev left07:14
← anli__ left07:14
mcnellis heh I don't know about tags in git07:15
anyawy how do you merge the changes from one branch to another?07:15
amason_ git merge07:15
mcnellis i.e. from my vesion branch to my master07:15
amason_ so if you're on master you do git merge branchname07:16
pretty simple and addictive :)07:16
amason_ loves git07:16
→ insomnia1alt joined07:17
→ tvw joined07:17
← brainbox1100 left07:17
← fdv left07:17
← krosenvold left07:18
mcnellis yeah I'm loving it as I learn more things!07:18
Thanks for properly explaining me this07:18
what if you're head is at the branch07:18
and you want to merge it into master07:18
or should you change your head to master first?07:18
← kenneth_reitz left07:18
← Ferris left07:19
Bombe mcnellis, merge always merges the branch you specify into the branch you’re currently on.07:19
typemore is there way in git to say: I didn't mean to name my last commit "git sucks", I meant to name it "git is better than svn"07:19
→ groovehunter joined07:19
Bombe mcnellis, so depending on whether you want the current branch to contain the other or the other branch to contain your current branch you need to change branches (or not). :)07:20
akheron typemore: git commit --amend, but this rewrites history, so don't use if you've already pushed07:20
→ sis joined07:20
mcnellis ok Bombe thanks. How do you change branches again?07:20
git checkout branch ?07:20
typemore akheron: nice; tahnks07:21
← insomniaSalt left07:21
insomnia1alt → insomniaSalt07:21
Bombe mcnellis, “git checkout branch”, yes.07:21
mcnellis awesome it really is very simple07:21
I resurrected a deleted file with some help of someone in here the other day07:21
was awesome :)07:21
Bombe With great power comes great complexity. :)07:21
mcnellis lol!07:22
→ nevyn^_ joined07:23
← _Vi left07:25
→ joeytwiddle joined07:25
← urkud left07:25
→ cptG joined07:25
← joeytwiddle left07:26
→ xendo joined07:26
→ joeytwiddle joined07:27
← giallu left07:27
→ Amorphous joined07:27
→ giallu joined07:28
← relachs left07:30
jotik^^ → jotik07:30
← psankar left07:31
← thiago_home left07:31
← giallu left07:31
→ gzy joined07:32
TML "git merge -s theirs upstream" says "Could not find merge strategy 'theirs'." - http://www.kernel.org/pub/software/scm/git/docs/git-merge.html tells me this should exist. What am I doing wrong?07:32
→ giallu joined07:32
← giallu left07:32
→ giallu joined07:32
Raimondi_ → Raimondi07:32
gzy hi, github q if you don't mind, i'm pushing tags annotated with messages, i can see them with git show tag_name, how can i view them in github, i can see the tag menu but where are the msgs viewable?07:33
pigeon hmm, git reset --hard gets rid of stashes too?07:33
cptG hello! i'm using gitolite and want to delete a repo from the server permanently - how can this be done?07:33
→ psankar joined07:33
← typemore left07:34
TML cptG: "rm -rfv"?07:35
← caseyw left07:35
→ the_groove joined07:35
cptG TML: is this safe? i was wondering since the doku says not to do anything manually on the server, so i thought there might be some kind of gitolite-specific way07:35
← the_groove left07:35
→ guardian joined07:36
→ caseyw joined07:36
TML cptG: *shrug* The documentation is telling you not to do anything on the server because you might make things get in a state where gitolite doesn't know what to do. If you're nuking it, there'll be nothing for gitolite to be confused about.07:37
← gzy left07:38
sis hello, I am trying to use git filter-branch with rm --cached, but it fails http://pastebin.com/e81EpKjT07:38
→ _frijole joined07:39
← dreiss left07:39
← madewokherd left07:39
→ psoo joined07:39
_frijole I have two banches, I want to merge the two. It should just be 'git merge branchname' right? I keeps saying everything is up to date...07:39
albel727_ sis: maybe you meant "git rm", not "rm"?07:40
TML sis: 'git rm' instead of 'rm'07:40
cbreak_work sis: read the error message07:40
also, if you use tree filter, use rm, if you want to use git rm, use an index filter07:40
sis omg, thx07:40
cbreak_work index filters are faster, so I would recommend that one07:40
sis thx alot!07:41
→ _Vi joined07:42
→ RichGuk joined07:42
→ fr0ggie joined07:43
_frijole I branched my code, have since made 16 commits to the brach 'v2' how do I merge the two? v2 and master?07:43
fr0ggie whats the 'right' way in git to say "at this path, i want git://... pulled? where that's always a remote path07:43
for example i want to embed libusb into a tree when checked out but latest sources. so that a checkout can build a complete freestanding system07:43
TML _frijole: git co master; git merge v207:43
selckin fr0ggie: make a branch for it?07:43
fr0ggie selckin: and will it stay automatically up-to-date?07:44
← groovehunter left07:44
selckin fr0ggie: oh missunderstood the question, you might want submodules07:44
fr0ggie: it's a way to version across diffrend git repos07:44
mostly07:44
_frijole TML: it just says 'everything up to date' ?07:44
selckin fr0ggie: but people always think it's like svn externals and get into trouble07:45
fr0ggie im putting together a small linux image for netbooting some devices07:45
at top: git submodule add git://... hw/bus/usb/libusb1.007:46
← amason_ left07:46
→ amason joined07:48
fr0ggie selckin: right?07:48
selckin i've never seriously used them, just toyed with them and read about them on the ml, so probaly07:49
fr0ggie ahh.. not automatic. that can be done in cmakefile so its ok07:49
→ tobiassjosten joined07:49
→ sambarino joined07:49
fr0ggie thanks07:49
TML _frijole: That means you've already merged them07:49
→ froschi joined07:50
TML _frijole: 'git log' should show you that07:50
← Slix` left07:51
→ pdelgallego joined07:51
← gnufied left07:52
← cptG left07:54
← vdv left07:56
→ MikhX joined07:58
→ oriba joined07:59
→ jrmuizel joined07:59
→ kpreid joined08:00
→ phb_ joined08:01
← phb left08:02
phb_ → phb08:02
→ Wicked joined08:04
← Wicked left08:04
← justin-george left08:06
→ Wicked joined08:10
→ khmarbaise joined08:10
→ fedesilva joined08:11
← theoros left08:11
← simplechat left08:11
← ntoll left08:12
→ ntoll joined08:12
→ nroot7 joined08:13
← mithro left08:13
nroot7 I just installed gitosis on a fresh ec2 cluster. I added a new repo in config file but when I try to push I get this error. "gitosis.serve.main:Repository read access denied"08:14
→ eletuchy joined08:14
nroot7 also there is no new directory for the repo I created under /home/git/repositories08:15
→ rraasch joined08:15
cbreak_work did you create the repository?08:18
→ jatt joined08:18
← phb left08:19
nroot7 I added the entry in the gitosis.config file08:20
but when I try to push to this repo, it fails with the above error08:20
cbreak_work as I said08:20
did you even create the repo?08:20
gitosis is unrelated with git08:20
as far as I know, all it does is restricting access08:21
doener gitosis creates the repo on demand08:21
→ aslakhellesoy joined08:21
nroot7 the gitosis redme says if I push changes to config file, it will create a repository08:21
cbreak_work intersting08:22
aslakhellesoy Does anyone know how to map svn committer names to different git committer names when importing a repo with git-svn?08:22
albel727_ nroot7: ok, did you reclone gitosis-admin.git repository, after pushing your public key to server?08:22
→ bauruine joined08:22
→ AoiKage joined08:23
nroot7 I didnt reclone anything08:23
← chrisshattuck left08:23
→ maguey__ joined08:23
cbreak_work aslakhellesoy: man git-svn08:23
Gitbot aslakhellesoy: the git-svn manpage can be found at http://git.or.cz/man/git-svn08:23
→ rolfb joined08:23
cbreak_work there;s --authors-file and --authors-prog08:23
albel727_ nroot7: ok, I assume you firstly added, committed and pushed your public key, and then tried to add repo to config, right?08:23
← sambarino left08:24
nroot7 albel727: My public key is in the keydir and I am able to read/commit gitosis-admin.git repo too08:25
← jrmuizel left08:25
← maguey_ left08:26
maguey__ → maguey_08:26
← _frijole left08:26
aslakhellesoy cbreak_work: thanks!08:26
→ notze joined08:26
albel727_ nroot7: ok, we seem to misunderstand each other, so lets try suspected solution rightaway. what was your url for cloning gitosis-admin.git repo for the first time?08:26
→ marc_kdab_ joined08:27
← stamina_ left08:27
nroot7 albel727: git clone git@<long-ec2-url>:gitosis-admin.git08:28
← koo5 left08:28
→ zoyd joined08:28
albel727_ nroot7: ok, I see, looks like that's not it. so "git push" inside gitosis-admin works ok?08:29
→ koo5 joined08:29
nroot7 albel727: yes push to gitosis-admin works fine08:29
zoyd hi, can a 'git diff' produced file be applied using the 'patch' program? like a normal 'diff -u'?08:29
→ stamina_ joined08:29
→ stamina_3 joined08:29
→ stamina_4 joined08:29
→ stamina_5 joined08:30
→ AAA5BQ5 joined08:30
→ WAAFBXI joined08:30
zoyd BOT!08:30
albel727_ nroot7: ok then, what did you add to gitosis.conf, concerning new repo?08:30
← stamina_2 left08:30
← Mage_Dude left08:30
← stamina__ left08:30
cbreak_work zoyd: take a look at git format-patch08:30
→ workmad3 joined08:31
← stamina_1 left08:31
zoyd cbreak_work: ok, thanks. there seems to be a bot infestation around here. later08:31
← zoyd left08:31
nroot7 albel727: http://pastebin.com/kLESzyUZ this is my conf file08:31
→ stamina__ joined08:32
nroot7 albel727: strange thing is I am not even getting any debug info, after adding that flag.08:32
→ mtkd_ joined08:32
albel727_ nroot7: never used that flag before, maybe it writes messages to some inner log file. ok then, how is your public key file named?08:34
→ Mage_Dude joined08:35
albel727_ nroot7: more precisely, does it end with ".pub"? and since you list your members as "e-mail", then does key file name have "@", etc?08:35
nroot7 albel727: in keydir folder right ? its <email>.pub08:35
albel727: as git push to one repo works, I think key is fine08:36
albel727: though I think gitosis didnt register the changes08:36
albel727_ nroot7: ok, do you have server access, so we can check on repo/keys presence directly?08:37
nroot7 albel727: yes I have08:37
albel727: keys are present but only one repo is there, gitosis-admin one08:38
sis i still have problems with removing some files track http://pastebin.com/ikuXLEYa08:38
nroot7 albel727: can I somehow make gitosis re-read the conf file ?08:39
→ sambarino joined08:39
albel727_ nroot7: hmm, nothing better comes to my mind, than adding an empty line to .conf and committing =/08:40
nroot7: looks like a have to give up at that point. I'm no expert in gitosis =(08:41
s/a have/I have08:41
nroot7 albel727: Thanks :-)08:41
→ ExtraSpice joined08:41
albel727_ nroot7: no problem.08:42
← sambarino left08:44
→ GarethAdams|Work joined08:45
→ tatsuya joined08:45
→ Dave^| joined08:46
→ sixteneighty joined08:49
← MikhX left08:49
→ MikhX joined08:50
← marc_kdab_ left08:51
→ cilly joined08:52
fr0ggie am i doing this wrong: git svn clone url -- i'd rather only grab latest revision08:52
→ sambarino joined08:53
→ marc_kdab_ joined08:54
← MikhX left08:55
← maguey_ left08:55
GarethAdams|Work fr0ggie: git deals in entire repositories08:55
albel727_ fr0ggie: git svn clone -r N svn://some/repo/branch/some-branch08:56
→ steph021 joined08:56
GarethAdams|Work fr0ggie: you can export a snapshot at a particular commit, but then you don't have a git repository for you to feed anything back into08:56
albel727_ fr0ggie: where -r N specifies revision, since which you want to get all the commits08:57
← sambarino left08:57
albel727_ fr0ggie: (N - is the number of svn revision, you need to figure out yourself, what is the latest revision in your repo)08:57
→ maguey_ joined08:58
← maguey_ left08:58
GarethAdams|Work oh sorry, I didn't read the question properly - ignore everything I said08:59
→ maguey_ joined08:59
→ mabrand joined08:59
albel727_ GarethAdams|Work: happens to me all the time =)09:00
zz_rubydiamond → rubydiamond09:00
→ aziz joined09:00
→ curtana joined09:02
sis I am trying to remove some files track but they are stucked, here are my commands http://pastebin.com/ikuXLEYa am I missing something?09:02
← stamina_4 left09:03
→ Shurakai joined09:03
← stamina__ left09:04
← stamina_3 left09:05
← stamina_5 left09:05
albel727_ sis: hmm, you filtered branch, all right, but did you change tags (namely v0.91 and v0.92), so that they now refer to newly filtered commits?09:05
← stamina_ left09:06
← notze left09:06
sis albel727_: not yet,, do I have to go through all of them separately like git tag v0.01 HEAD~39 etc?09:07
cbreak_work the tags will point to the same thing they pointed before filtering09:07
sis: man git-filter-branch09:07
Gitbot sis: the git-filter-branch manpage can be found at http://git.or.cz/man/git-filter-branch09:07
→ Schmallon joined09:07
→ cillybabe joined09:09
← masonkramer left09:09
← cilly left09:09
→ masonkramer joined09:09
albel727_ sis: not sure about manual tags retargeting. I'm investigating it.09:09
sis cbreak_work: Do I have some misunderstood?09:09
cillybabe → cilly09:09
cbreak_work I think you need to do filter-branch with tags too09:09
sis hmm, thx, working on that09:10
→ bluenovember joined09:10
cbreak_work you can just list them together with the branches you rewrite09:10
→ cpf_ joined09:10
albel727_ sis: try using "--tag-name-filter cat -- --all" when calling git-filter-branch.09:11
← nroot7 left09:11
cbreak_work no09:11
I don't think the tag names are the problem09:11
cpf_ I'm having an unpack error while doing git push on a git:// (non-authenticated) daemon...09:11
cbreak_work you hacked git:// to allow pushing?09:11
→ Asone joined09:11
← Asone left09:12
cpf_ cbreak_work, Not hard: Adding receivepack = true to the config.09:12
→ malumalu joined09:12
→ royalty joined09:13
albel727_ cbreak_work: this is a dirty hack to have tags with same names, but pointing to modified objects. see --tag-name-filter option description and "Checklist for Shrinking a Repository" section in man git-filter-branch.09:13
Gitbot cbreak_work: the git-filter-branch manpage can be found at http://git.or.cz/man/git-filter-branch09:13
cbreak_work you could just use the original filter-branch09:14
albel727_ cbreak_work: meaning?09:14
cbreak_work just specify the tags you want to rewrite in addition to the branches09:15
→ phb joined09:15
← steph021 left09:17
→ Hyra joined09:18
albel727_ cbreak_work: are you sure it is possible? somehow I doubt, that rewriting tags can be default behavior. and the man page only mentions specifying branch names...09:19
→ gnufied joined09:20
sis do you mean git filter-branch --subdirectory-filter foodir -- --all what I need?09:20
sis is saving the whole .git dir before breaking it :)09:20
→ thierryp joined09:21
sis I see 1/76... instead of 1/39, new hope :)09:22
albel727_ sis: probably not. subdirectory filter LEAVES only the specified directory, while I get the idea, that you want to REMOVE the directory and leave everything else...09:22
sis i only ment -- --all09:22
→ awilkins joined09:23
→ urkud joined09:23
→ Titosemi joined09:23
awilkins Is there any software I can exploit to rapidly illustrate branch revision graphs?09:23
→ vdv joined09:23
albel727_ sis: hmm, --index-filter with --all... it might work, sure.09:24
awilkins: you mean, besides gitk?09:24
← skippyish left09:24
albel727_ awilkins: or "git log --graph"?09:24
awilkins: or better "git log --graph --decorate --all"?09:25
→ linusoleander joined09:25
← sixteneighty left09:25
← a9b left09:26
albel727_ awilkins: or "git log --graph --oneline --decorate --all"...09:26
awilkins albel727_, More like "how can I demonstrate hypothetical graphs to management"09:26
albel727_, Thanks for the suggestiosn09:26
→ g0bl1n joined09:26
← phb left09:27
→ phb joined09:27
← maguey_ left09:27
albel727_ awilkins: tutorial needs? maybe try to introduce them to progit.org book? otherwise, no good graph drawing software comes to my mind right now, sorry. =/09:28
← cenuij left09:30
→ parasti joined09:31
sis albel727_: with -- --all it works except some tags where it says WARNING: You said to rewrite tagged commits, but not the corresponding tag. WARNING: Perhaps use '--tag-name-filter cat' to rewrite the tag.09:32
Hyra does anyone happen to have a link to how to get git working with Komodo IDE (osx) .. google is not my friend today.09:32
sis what is the difference between tagged commits and the corresponding tag?09:33
→ rbarraud__ joined09:33
cbreak_work sis: what is the difference between a file and a symlink to that file?09:34
← eletuchy left09:34
sis cbreak_work: the tagged commits are like the symlinks?09:34
cbreak_work no, like files09:35
← cilly left09:35
→ cillybabe joined09:35
cbreak_work tags are like labels you put on them, like symlinks to commits09:35
albel727_ sis: http://progit.org/book/ch2-10.html for future reference. for the time being, I recommend you to go with --tag-name-filter, like git suggests.09:36
sis so the corresponding tags are like symlinks, then why do I need to rewrite them as well?09:36
cbreak_work because you create a new commit if you filter-branch09:36
the tags still point to the original09:37
you might know how commits are identified09:37
if you think a bit about it, it is clear :)09:37
sis hmmmm I see now! thank you !!!09:37
albel727_ sis: thing is, changing commits is not only about changing their contents. it's also about changing their (internal to git) names too.09:37
← amason left09:37
cbreak_work their sha1 is derived from their content09:37
← rbarraud_ left09:37
sis albel727_: thank you too!09:37
cbreak_work so, commits are effectively immutable09:38
albel727_ sis: no problem =)09:38
→ amason joined09:38
sis cbreak_work, albel727_: thank you very much! you've helped me a lot!09:39
← xiangfu left09:39
cbreak_work no prob09:39
→ priidu joined09:42
→ thm joined09:49
→ blaines joined09:50
← vdv left09:51
thm if I am tracking multiple remote branches, can I arrange it in such way that a single pull merges them all in their respective local branches (at least if that's a ff op)?09:51
→ priidu_ joined09:53
← priidu left09:54
→ brian_g joined09:55
→ jonkri joined09:55
→ cilly joined09:55
← cilly left09:55
← merlin83 left09:55
→ cilly joined09:55
jonkri is it correct that /usr/share/git-core/templates/description is -rw-r----- 1 root root? i'm getting an error "fatal: cannot copy /usr/share/git-core/templates/description to /var/diaspora/devise/.git/description" when running git as a regular user09:56
← nevyn^_ left09:57
← hyperair left09:57
→ dr_win joined09:57
→ sambarino joined09:58
→ HarryS joined09:58
albel727_ thm: well, you can always write a script, which will for every branch A check, that "git-merge-base A origin/A" equals to "git-rev-parse A", and then do "git-update-ref A origin/A", or something...09:58
← cillybabe left09:59
→ flazz joined09:59
jonkri i don't know why the permissions would be wrong - i installed git through apt-get10:00
albel727_ jonkri: strange. it also has o+r permission on my machine.10:00
→ frakturfreak joined10:00
albel727_ jonkri: I guess you'll be safe to change it manually, though. description is not a script, anyway...10:01
jonkri albel727_: both branches, hooks and info in the same directory has o+rx10:01
hmm, it simply contains "Unnamed repository; edit this file to name it for gitweb."10:01
albel727_ jonkri: maybe a bug with the package? you may file it somewhere at your leisure =)10:02
jonkri: yeah, that's what it usually contains10:02
jonkri i'll try purge and reinstall git, if that doesn't work i'll change the permission10:02
← HarryS left10:02
albel727_ albel727_: and nobody usually bothers to change it in their repos =)10:02
jonkri :P10:02
albel727_ jonkri: whoops. =) ^10:02
albel727_ has a multiple personality disorder. please don't mind him =)10:03
jonkri ah, it seems that there's a bug, missed that earlier. thanks http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=59572810:04
→ willmarshall joined10:04
← aziz left10:05
willmarshall So - I have two existing git repositories: and I'd like to make one a folder inside another10:05
← xendo left10:05
willmarshall Originally a separate project, but now a crucial component for the bigger project10:05
Is there an elegant way to do this?10:05
← sis left10:05
albel727_ thinks that willmarshall is in for a lot of fun...10:05
albel727_ willmarshall: there are usually two (or three) alternatives for doing that10:06
willmarshall Go on...10:06
This is a very simple example - I could just copy and paste - I am just wondering about *good* ways to do this, if I had lots of project history and old commits10:07
albel727_ willmarshall: you have to concider, do you really need all the history of smaller project in bigger project?10:07
willmarshall albel727_: Nope! I'm just curious if it can be done10:07
And done well10:08
albel727_ willmarshall: does anyone still work on smaller project, so that you'll need to periodically propagate changes to parent project?10:08
willmarshall albel727_: I work on both of them - but the smaller project is not intrinsically attached to the bigger10:08
It's a crucial component and must always be packaged within it, but can be used standalone10:09
Sadly this environment has no support for dependencies or anything cool like that10:09
← cilly left10:09
→ cillybabe joined10:09
albel727_ willmarshall: ok, I see. The three alternatives are "git submodule", "git subtree" and, you guessed it, just copying smaller project over and forgetting about it.10:09
→ vu3rdd joined10:10
albel727_ willmarshall: git submodule is usually considered too complicated to work and manage for majority of users, so it never have gained popularity.10:10
willmarshall: useful reading is, of course man git-submodule, and http://progit.org/book/ch6-6.html10:11
Gitbot willmarshall: the git-submodule manpage can be found at http://git.or.cz/man/git-submodule10:11
cillybabe → cilly10:12
willmarshall Interesting10:12
Does look rather gnarly10:12
albel727_ willmarshall: well, once you master it and use on regular basis - nothing complicated really (and git guys are constantly working on making it more accessible to mere mortals).10:13
willmarshall: but it might be an overkill, indeed.10:13
→ trevor_morse joined10:13
willmarshall The link you gave me is definitely understandable, but a tad fiddly10:13
Overkill for this simple case, but good to know about10:13
albel727_ willmarshall: well, then there is the subtree merge strategy: http://www.kernel.org/pub/software/scm/git/docs/howto/using-merge-subtree.html10:13
willmarshall heh10:13
"Then, you e-mail that guy and yell at him."10:13
→ jtauber joined10:14
← Torsten left10:15
← jtauber left10:15
← urkud left10:16
← frakturfreak left10:16
→ frakturfreak joined10:16
→ intripoon joined10:17
→ altamic joined10:17
altamic I've somehow lost my billing hours data for the last month, all I have is a git repo with my commits. I wonder if I could build a simple script with some git tricks to extract my commits grouped by day and find the time interval between the first and the last of the day. The problem is I am by no mean a git expert, where to look to start? git-log? Anything already out there?10:19
← jatt left10:20
FauxFaux git log --format="%ad" is all you need.10:21
altamic thx let's try it out10:21
← mstsirkin left10:21
FauxFaux The shell to work out the first and last time on a given day is hard, though. ¬_¬10:21
← jonesy_ left10:22
altamic great10:22
It's what I was looking for10:22
→ dels joined10:22
albel727_ git log --format="%ad %cn" | grep "<yourname>"10:23
← flazz left10:23
← hipe left10:24
altamic that's fantastic10:24
:)10:24
thank you albel727_10:24
albel727_ altamic: no problem =)10:24
← workmad3 left10:24
→ hipe joined10:25
← rbarraud__ left10:25
← cilly left10:26
→ cillybabe joined10:26
cillybabe → cilly10:26
→ workmad3 joined10:27
← hipe left10:29
← Schmallon left10:33
← dr_win left10:33
← _Vi left10:34
← bphogan left10:35
→ hipe joined10:37
← trevor_morse left10:38
← linusoleander left10:41
← gnufied left10:45
← opalka left10:47
→ opalka joined10:47
→ HarryS joined10:47
→ cenuij joined10:48
← cenuij left10:48
→ cenuij joined10:48
← HarryS left10:49
→ HarryS joined10:50
→ kpreid_ joined10:50
← stliu left10:51
← workmad3 left10:52
← opalka left10:52
→ opalka joined10:53
← Hyra left10:53
← kpreid left10:54
kpreid_ → kpreid10:54
HarryS → Guest2228810:56
← Guest22288 left10:57
→ HarryS joined10:58
→ gnufied joined10:59
→ Textmode joined10:59
→ krosenvold joined11:00
→ typemore joined11:01
typemore is there an open source implementaaion of github?11:01
→ shiba_yu36 joined11:02
← HarryS left11:02
frakturfreak typemore: you may try gitorious.org11:03
← krosenvold left11:05
← kanru left11:05
← kpreid left11:05
typemore frakturfreak: is gitorious open source? I don't see a download link -- I want something like github, but behind a private firewall.11:05
frakturfreak Ah download11:06
typemore http://www.gitorious.org/about11:06
gitorious itself looks open source11:06
frakturfreak gitolite11:06
→ jetienne joined11:06
frakturfreak if you don’t want to pay for github.fi11:06
typemore well; i want control11:06
jetienne q. i would like to stage a given modification in a large file. this file contains several of them... how can i do that ?11:07
← royalty left11:07
→ boombatower joined11:07
← boombatower left11:07
→ boombatower joined11:07
frakturfreak http://github.com/sitaramc/gitolite/11:07
typemore: There you go11:07
← gnufied left11:07
typemore wait ....11:07
why can't i download gitorious? girotious itself looks open source11:07
http://gitorious.org/gitorious11:08
is what I wnat :-)11:08
→ gnufied joined11:08
albel727_ typemore: you may also take a look at cgit, if you want something simpler (not sure about it capabilities, though)11:10
← gnufied left11:10
→ zorzar joined11:11
→ sis joined11:11
albel727_ jetienne: "git add -p"? "git citool", choose file, right click on change, stage line/hunk11:13
← dels left11:13
sis how could I point to project's main dir? I have tried with ./*.patch with filter-branch and now it is removing all *.patch not only in main dir11:14
← zorzar_ left11:14
albel727_ jetienne: in general, see man git-add for -p and --interactive options.11:14
Gitbot jetienne: the git-add manpage can be found at http://git.or.cz/man/git-add11:14
jetienne albel727_: thanks a lot!11:14
sis * in all subdirs not only in main dir11:15
← cilly left11:16
→ cillybabe joined11:16
→ gnufied joined11:16
jetienne albel727_: any diff between "git gui" and "gi citool" ?11:16
cillybabe → cilly11:16
→ vladiator joined11:17
typemore gitorious looks much prettier than cgit11:17
imho11:17
← boombatower left11:17
← radioman-lt left11:17
albel727_ jetienne: well, git citool IS a git gui which is run in commit only mode. there are certainly more commands available in "general" git gui mode, than in "commit" mode. but if you're commiting, you don't need them most of the time...11:18
jetienne albel727_: ok11:19
albel727_ jetienne: also git gui will frequently pop up you with some annoying committing-unrelated stuff, like running garbage collection, etc11:19
jetienne: you may want to take a look at gitk, too.11:20
jetienne albel727_: will do11:20
← gnufied left11:21
← cwillu_at_work left11:22
albel727_ jetienne: ok. make no mistake, it's a built-in-git utility, just like git gui. greatly helps to visualise and manipulate history/branches.11:23
← tvw left11:23
cbreak_work gitx ftw11:23
→ Quit joined11:23
albel727_ cbreak_work: mac os mustdie!!!!!111 =)11:24
← Quit left11:24
cbreak_work unix ft11:25
WIN11:25
albel727_ vigorously nods in agreement.11:25
cbreak_work and Mac OS is the only unix that is actually still in use11:26
altamic IMHO Mac Os has the best UI around11:27
avar I think cron(1) has the best UI around11:27
albel727_ cbreak_work: well, I'm not so sure about that. Last time I checked, Mandriva, Ubuntu and Gentoo are used pretty much everywhere.11:27
won't argue about UI11:27
albel727_ won't argue about UI, because he doesn't want to start another holywar =)11:28
cbreak_work albel727_: those are not unix11:29
those are just linux distributions11:29
→ masterkorp joined11:29
altamic avar: cron has a compact syntax and very useful11:29
← typemore left11:29
cbreak_work and as everyone should know Linux Is Not Unix11:29
albel727_ cbreak_work: and Linux Is Not UniX. yeah yeah.11:29
jast bleh. osx is a combination of a darwin kernel and (mostly) bsd userspace11:29
if osx is unix, presumably the BSDs are unix too11:29
cbreak_work yes11:30
albel727_ cbreak_work: (which is a big fat lie =P)11:30
jast and I do think the BSDs are "still in use"11:30
by which standard is linux not a unix, anyway?11:30
cbreak_work I think novel controls the trademark11:30
they have some test suit for which you pay to be tested11:31
and then you can be certified11:31
→ cwillu_at_work joined11:31
albel727_ cbreak_work: oh, so now we're playing along legal definitions, I see... =)11:31
jast the open group owns the trademark11:31
→ pjp joined11:31
cbreak_work hmm... then what did novel own? they got into trouble with sco over something11:31
bmalee Novell own the copyright to (some?) Unix source code, I believe.11:32
→ gnufied joined11:32
selckin so i'm inside the editor of commit --amend, and want to abort the commit, since i just noticed the head wasn't the one i wanted to amend, any trick to cancel the commit from the editor? or shall i just reflog & reset back?11:33
← roop left11:33
albel727_ retreats back into shadows, having accomplished his evil holywarmongering deeds.11:33
← avinashhm left11:33
avar selckin: empty the buffer and save it, it'll refuse to commit with an empty message11:33
pjp hi, I'm using git version 1.7.2.2, and it has a weird problem that, when I checkout branches, the changes from the current branch get merged with the new branch, is there a way to stop it?11:33
avar or send a sigint to the git commit --amend process11:33
selckin avar: thanks11:33
→ avinashhm joined11:33
avar pjp: you mean the changes from your working directory that aren't checked in yet?11:34
→ _Vi joined11:34
pjp avar: yep, the changes not yet commited,11:34
avar well, then commit or stash those and then switch branches11:34
cbreak_work pjp: user error then :)11:34
jast define "merged"11:35
cbreak_work the working directory does not belong to a branch11:35
jast it's normal for changes to be carried over if the changed file is identical in both branches11:35
← blaines left11:35
pjp avar: but, it never used to happen earlier, isn't there any setting to stop it, without me having to commit changes or stash them?11:35
jast in other cases, the checkout should abort11:35
cbreak_work pjp: git checkout will not merge any branches11:36
avar pjp: Yes it did happen earlier, git has always done that.11:36
cbreak_work it will only check out the commit that it is requsted to check out11:36
→ Fullmoon joined11:36
avar pjp: do you want to save those changes?11:36
← cenuij left11:36
cbreak_work and merge it with the working directory changes that you have performed11:36
pjp cbreak_work: not merge branches, but take the local changes to the new branch where you want to switch11:37
← bauruine left11:37
pjp avar: may be, but I don't want to commit just yet,11:37
just want to switch branches,11:37
→ linusoleander joined11:37
cbreak_work pjp: git always did this11:37
jast pjp: what would you want to happen instead?11:38
cbreak_work there's nothing else it can do really, appart from not switching branches11:38
← psankar left11:38
cbreak_work since local changes don't belong to a branch11:38
→ Schmallon joined11:38
pjp jast: I want the changes in a branch should remain there, and unless I manually merge or commit them, git should not do it11:38
cbreak_work you can commit them into a temporary commit or stash11:39
← schlortoo left11:39
cbreak_work pjp: apparently you didn't understand what I said before11:39
local changes do not belong to a branch11:39
they belong to the working directory11:39
a branch contains only commits11:39
jast pjp: where is "there"?11:39
pjp jast: in a branch,11:40
→ bauruine joined11:40
ptor pjp: If switching branches imply changing files, it can't just keep (uncommited) changes to said files. It would prevent the switch.11:40
jast pjp: but changes are not in a branch. they are in the working tree.11:40
pjp cbreak_work: oh,11:40
→ tjubaluba joined11:40
albel727_ pjp: ok then. let's see it from the point of git. you have local changes, which you don't want committed on the branch, and don't want them to remain in working directory. where should git preserve them?11:40
cbreak_work my tip: commit it.11:40
that's the only way to put stuff into a branch.11:41
you can reset the commit away later, amend it, or what ever.11:41
pjp albel727_: how does git manages when I commit them?11:41
← vladiator left11:41
jast stuff that is committed will stay on that branch when you switch11:41
→ bphogan joined11:41
→ vladiator joined11:41
pjp cbreak_work: yeah, but everytime you commit, it also asks you for a message that's sometimes is really annoying,11:41
jast stuff that isn't... won't, if git can carry over the changes without messing anything up :)11:41
pjp: git commit -m 'dummy message here'11:42
avar git commit --allow-empty-message11:42
cbreak_work pjp: git commit -m "WIP: bugfixes"11:42
one or two words11:42
albel727_ pjp: git stores them into repository, and makes branch reference to refer to them, so next time you checkout this branch, git will know what to bring out.11:42
j0ran pjp: have you tried "git stash"?11:42
cbreak_work if your changes are not worth that much, you can just nuke them away :)11:43
pjp jast: avar: I know that, but I was just wondering if I could configure git to not do it,11:43
cbreak_work I don't like to use stash, I always forget stuff in there11:43
avar pjp: you can alias 'git commit' to 'git commit --allow-empty-message' in your shell11:43
cbreak_work avar: I doubt that11:43
ptor just does 'git stash list' now and then. But it's only useful if stash'ed with a message really.11:43
pjp avar: hmmn....11:44
→ trochala joined11:44
cbreak_work shell aliases would only work for whole git11:44
jast cbreak_work: depends on the shell :)11:44
cbreak_work and git aliases can not change predefined commands11:44
avar what jast said :)11:44
jast you can always use a function that parses the args11:44
in fact I have one such function11:44
it... discourages me from using git pull in a specific repository that I have11:44
→ kpreid joined11:44
avar jast: tasty, got a link to it?11:44
← ntoll left11:44
avar you can also just do a string-replacement on the string your shell is about to accept, some allow that11:45
(e.g. eshell does)11:45
→ oriba_ joined11:45
← yatiohi left11:45
← oriba left11:45
jast avar: http://git.pastebin.com/ARJMDb6911:47
nothing special, really11:48
← Fullmoon left11:49
→ schlortoo joined11:49
→ merlin83 joined11:49
← merlin83 left11:49
→ merlin83 joined11:49
avar ah11:50
jast I have a similar script that runs on every 'cd' and displays todo entries for that directory if applicable11:51
← thierryp left11:52
jast based on devtodo11:52
← sonnym left11:53
→ Fullmoon joined11:55
avar I do that by calling a function from PS111:55
← cpf_ left11:57
→ cpf_ joined11:58
→ thierryp joined11:59
→ ods15 joined12:00
→ rwagner joined12:00
ods15 Hi.. if i copied a ".git" from somewhere, without copying the checkout, what would be the right method to cause git to checkout all the files again?12:00
← rwagner left12:01
cbreak_work ods15: git checkout? :)12:01
ods15 ok, wasn't sure :) didn't want to try it before being sure12:02
← ThiefMaster left12:02
ods15 yup, works great :) thanks!12:03
→ lourense joined12:04
← ods15 left12:04
rgr what facilities exist to hook into certain git functions? I could do it at the elisp "magit" level but guess it might be possible to get git to call certain scripts when it commits (for example). Ideas or existing examples?12:05
← Athox left12:06
reto_ rgr: there are hooks12:06
rgr: and you can create your own comamnds and use git functionality in there12:06
← schlortoo left12:06
reto_ rgr: name your scripts 'git-<anyname>' and add them to your path, and then you can do git <anyname>12:07
rgr that i dont want to do. I want to use the native git functions (well, a third party utility (magit) does.12:07
SethRobertson pjp are you around?12:07
pjp SethRobertson: yep12:08
→ schlortoo joined12:08
SethRobertson You wanted to ignore certain config file changes and commit everything else?12:08
→ Bucciarati joined12:09
pjp SethRobertson: no, when I git checkout branch, I want git to not take along the local changes to the new branch12:09
← mtkd_ left12:09
→ ntoll joined12:09
pjp SethRobertson: irrespective of whether they are committed or not12:09
fr0sty__ rgr: What are you trying to accomplish?12:10
rgr I want git to call certain scripts when it does certain things.12:10
SethRobertson So you want an implicit `git reset --hard` I guess12:10
→ johne joined12:10
→ Anti-X joined12:10
rgr githooks is it12:10
fr0sty__ rgr: man githooks12:10
Gitbot rgr: the githooks manpage can be found at http://git.or.cz/man/githooks12:10
rgr ;)12:10
SethRobertson pjp: Well, sounds like you can use githooks :-) Make a post-checkout hook to do the `git reset --hard`. Sounds dangerous to me, but what the heck12:12
fr0sty__ pjp: just add 'git reset --hard' to your post-checkout script.12:12
SethRobertson: beat me to it...12:12
pjp SethRobertson: fr0sty oh..let me try,12:14
← danger` left12:14
pjp SethRobertson: fr0sty : the changes don't get lost right?12:14
→ becomingGuru joined12:14
SethRobertson Oh yes, they do12:15
Any commited change is safe, but uncommited changes...not so much12:15
You are back to stashing for that case, and there does not seem to be a pre-checkout hook12:15
pjp SethRobertson: weird12:17
→ chronos joined12:17
→ hebz0rl joined12:18
fr0sty__ pjp: where do you want the changes to go when you switch branches?12:19
→ SimonP86 joined12:19
pjp fr0sty: no where, just stay in the branch,12:20
ptor pjp: a branch is what is commited. If not commited, can't stay in the branch.12:20
SethRobertson Though with a pre-checkout hook we could stash to a branch name and in a post-checkout hook we could pop.12:21
ptor pjp: you can have many branches, but only a single working tree. changes to that working tree have to go somewhere when you switch branches.12:21
← hebz0rl left12:21
fr0sty__ pjp: Why are you so concerned about tracking untracked changes? Why not just commit them?12:21
pjp ptor: strange, if it can be done for committed change, why not for uncommitted ones12:21
← aresnick left12:22
→ __name__ joined12:22
Paraselene_ → chrischambers12:22
→ hebz0rl joined12:22
→ sonnym joined12:22
ptor pjp: Because of what I said: There's only a single working tree (that's the files you see, what you have checked out). What's under version control is what you have commited.12:23
pjp fr0sty: well, I could want to switch between the branches quite often,12:23
fr0sty__ if you don't care about your changes enough to commit them why do you care what else happens to them?12:23
cbreak_work pjp: go read git-for-computer-scientists12:23
pjp cbreak_work: yeah, I saw that12:24
SethRobertson pjp: just write a pjp-git-checkout wrapper script which does the stash, checkout, stash pop.12:24
cbreak_work pjp: read it again :)12:24
pjp fr0sty: well, it's not like I don't care about the changes, just that I don't want to commit them just yet12:24
cbreak_work you obviously didn't understand it12:24
branches only consist out of commits12:25
ptor pjp: The commited changes are in blobs under the top-level .git/ directory. Only that part can possibly be kept when you switch branches. Outside of .git is your working tree, and that's just normal files. If they are modified compared to what's in the commited files (the stuff under .git/) the changes must either be abandonded, commited, or stashed before switching to another branch. The two latter ones stuff the changes into .git/ somewhere.12:25
cbreak_work there's nothing else12:25
→ zombor joined12:25
cbreak_work the only way to make something belong to a branch is to commit it12:25
← hebz0rl left12:25
pjp cbreak_work: weird12:26
cbreak_work how so? :)12:26
→ hebz0rl joined12:26
ptor pjp: Just stash the changes, there's nothing to it. git checkout master; work work work; git stash; git checkout other-branch; work work commit; git checkout master; git stash pop; continue work12:26
→ urkud joined12:26
← fr0sty left12:27
fr0sty__ → fr0sty12:27
pjp ptor: hmmn..guess, that's the only option12:27
→ fr0sty_ joined12:27
cbreak_work you could just make more working directories12:28
there's a script, git-new-workdir12:28
pjp cbreak_work: hmmn...12:28
cbreak_work but I don't know how stable it is12:28
fr0sty pjp: you can also make a 'work in progress' commits.12:28
cbreak_work I would recommend making commits12:28
it's the cleanest approach12:28
pjp cbreak_work: hmmn...ok12:29
fr0sty +1 for committing.12:29
← froschi left12:30
cbreak_work this is not subversion12:30
ptor yep, then there's an opportunity to learn git commit --amend (to continue the staged work)12:30
cbreak_work where commiting, branching and tagging is dangerous and expensive12:30
→ jaql joined12:30
ptor nods enthusiastically12:31
→ roop joined12:32
→ aresnick joined12:32
← kpreid left12:33
→ thoolihan joined12:35
→ IslandUsurper joined12:35
awilkins is working on a SVN-managed project, with a guy who does not grok VCS at all12:37
awilkins Hell on earth12:37
He totally wrecks the ability to use a decent VCS in tandem with SVN by doing things like deleting the whole tree and replacing it with a new one *sob*.12:38
ptor force the guy to provide all updates via patches.. that'll teach him to appreciate VCS.12:39
__name__ My deepest sympathy12:39
→ d0k joined12:39
awilkins Hah, wish I could ; he's the repo admin12:40
doener awilkins: I once worked with someone who "branched" by creating a new svn repo, copying files from his working copy into a working copy for that new repo, and finally committing the new repo into the original one, somewhere in the branches/ directory12:40
cbreak_work awilkins: just use git blame and blame him for every bug he commited12:40
(by adding a whole tree)12:40
awilkins Hehehehe12:41
Wouldn't change his bug count muh12:41
← zombor left12:41
→ maguey_ joined12:42
← linusoleander left12:46
→ ehabkost joined12:49
FauxFaux Is there a plausibly easy way to estimate the size of the tree at a revision? i.e. anywhere near what would be returned by git checkout X && git clean -dfx && rm -rf .git && du -sh .12:51
cbreak_work FauxFaux: looked at the tree?12:52
→ notze joined12:53
→ LiamH joined12:53
FauxFaux I just noticed git ls-tree has the object size, but that's compressed; it doesn't store the original size, right?12:53
cbreak_work don't know12:54
← vladiator left12:55
frerich FauxFaux: If your repository mostly contains text files, you could check the number of insert and number of deleted lines shown by 'git diff --shortstat <commit>' and then say that each line is approx. 60 bytes.12:55
cbreak_work you could decompress it12:55
FauxFaux ls-tree | cut -d' ' -f3 | cat-file -s --batch seems to be as close as you're going to get.12:55
frerich Then you would knew how much larger (or bigger) the checkout is, roughly, than your current checkout12:55
FauxFaux frerich: Ooh, nice idea.12:55
← pjp left12:55
frerich Just a crazy idea, it's just an estimate :)12:56
FauxFaux Gah, can't batch and -s at the same time.12:56
← maguey_ left12:56
← devanjan left12:57
FauxFaux git ls-tree -r HEAD | awk '{print $3}' | git cat-file --batch | wc -c # looks about right. Thanks. ¬_¬12:57
→ Rearden joined12:57
albel727_ FauxFaux: -l option of git-ls-tree gives me real sizes for objects12:57
← Fullmoon left12:57
rubydiamond → zz_rubydiamond12:58
FauxFaux ...so it does. That file must be a lot smaller than I thought it was. Summing in shell is harder than the above, though. =p12:59
zz_rubydiamond → rubydiamond12:59
→ kanru joined12:59
→ jshsu joined13:01
→ bentob0x joined13:01
→ denysonique joined13:02
→ Cap_J_L_Picard joined13:03
← Rearden left13:03
→ Fullmoon joined13:03
← rraasch left13:04
SethRobertson git ls-tree -r HEAD | awk '{print $3;}' | xargs -n1 git cat-file -s | awk '{sum += $1;} END { print sum;}'13:04
→ Rearden joined13:06
→ voob_of_doom joined13:06
← Rearden left13:07
→ Rearden joined13:07
thiago git archive is also very good for that13:07
→ alex__c2022 joined13:08
cbreak_work it might be the best solution.13:08
→ tvw joined13:08
← chronos left13:08
← Rearden left13:09
SethRobertson Moves into approximation land due to tar/zip overhead13:09
→ kjy112 joined13:09
→ Rearden joined13:11
← notze left13:11
← Rearden left13:11
← jonkri left13:12
→ Rearden joined13:13
dwmw2_gone → dwmw213:14
← xxoxx left13:15
← Rearden left13:16
→ hyperair joined13:19
← _Vi left13:19
→ kenneth_reitz joined13:19
← kenneth_reitz left13:20
→ kenneth_reitz joined13:21
→ royalty joined13:21
→ xxoxx joined13:21
→ beatak joined13:21
← royalty left13:22
→ nevyn^_ joined13:23
→ notze joined13:23
fedesilva i am doing git add -i make a partial commit. I choose patch and it shows me the first hunk ( is that the word? ) but it chooses too many changes ( lines ),13:23
is there a way to select less lines.13:23
?13:23
→ _Vi joined13:23
cbreak_work press s for subdivide13:23
or e for edit13:23
or h for help13:23
→ zombor joined13:23
fedesilva cbreak_work: ok, checking out s.13:24
cbreak_work: s works for me. thanks.13:25
somehow I missed it when reading the output of (?)13:26
→ DJPcCopath joined13:26
DJPcCopath hello13:26
→ pielgrzym joined13:26
pielgrzym hi :)13:26
→ formale joined13:26
pielgrzym if I have a submodule - what happens when I just change it's files? it becomes a no-branch code?13:27
← DJPcCopath left13:27
← thierryp left13:27
← aresnick left13:27
→ wjames6 joined13:27
wjames6 hi, im trying to find docs on how to add a new branch to a remote. thanks for help in advance.13:28
SethRobertson pielgrzym: Do you want the submodules to be directly managed like the rest of the repo? Commits/tags/branches descend into them? pull update everything?13:28
pielgrzym the thing is: I have some code reusable components which I import as submodule. it's very inconvenient to edit them, since they make sense only within a project and only there I can test them. So each time I want to change a submodule I have to open a separate tmux session, edit files of the submodule git repo, push it and pull it back in bigger project to see if my changes are ok13:28
bremner wjames6: man git-push13:28
pielgrzym SethRobertson: ^13:28
Gitbot wjames6: the git-push manpage can be found at http://git.or.cz/man/git-push13:28
pielgrzym SethRobertson: I wonder if I could create a working branch in a nested submodule13:29
SethRobertson: and then push the changes back to my global submodule's master branch13:29
SethRobertson pielgrzym: I have a gits wrapper which makes "slave modules". When you perform the wrapper operations, they are performed on the super-project and all of the slave projects.13:29
pielgrzym SethRobertson: I'm not sure I made it clear ;) to speak simply: I want to edit submodule's files in a parent git repo and then push only the changes back to submodule's repo :)13:30
fr0sty pielgrzym: you can checkout a branch in a submodule, edit, commit, push and then update the superproject.13:30
cd submodule; git checkout <branch>; edits...; git commit; git push <remote> <refspec>13:30
→ scarabx joined13:30
pielgrzym SethRobertson: I made a cool alias: ac which stands for 'add component' and it spawns immediately a submodule from a given path :)13:30
fr0sty: ah! great!13:31
fr0sty: and then I should just git submodule update?13:31
SethRobertson: could you show me the wrapper?13:31
← notze left13:31
fr0sty pielgrzym: yep.13:32
→ zenom joined13:32
SethRobertson pielgrzym: Sure http://www.baka.org/projects/gits.git13:32
→ TeckniX joined13:33
SethRobertson The script creates a group of related repositories all of which are concurrently developed on and on which all git operations should operate, so when you branch, each repository in the project is branched in turn. git-submodules went a different way where you were referring to a particular remote repository.13:33
And likewise, when you commit or push, each respository is pushed in turn.13:34
wjames6 bremner: ok ive reread push again and the best i find is that it wants a matching branch. but if its a new branch that doesnt exist in the remote yet i dont find any clear instruction for pushing that.13:34
ptor interesting.13:34
(the gits thing13:34
← mabrand left13:35
SethRobertson I've been toying with more broadly announcing it13:35
→ emmanueloga joined13:35
→ jonesy_ joined13:35
bremner wjames6: push remotename new-branch:new-branch , for example13:35
wjames6 ah github docs say just add the new remote name13:35
as if it already exists13:35
thanks bremner13:36
ptor I've been thinking forth and back about how to move from a unified CVS repo with tons of modules that are only semi-linked, to a Git model where I would prefer to both let each module be its own git repo, but also as a part of something bigger.13:36
bremner wjames6: no problem13:36
← urkud left13:36
SethRobertson ptor: That is *exactly* why I wrote gits. We had that exact same issue.13:36
ptor Been looking at submodules as well as git-subtree.13:36
wjames6 ptor: superproject and submodules13:37
→ aresnick joined13:37
← thm left13:37
→ chrisshattuck joined13:37
ptor SethRobertson: Right, I'll definitely look at this. Submodules doesn't fit what I want to do.13:37
wjames6 SethRobertson: link for gits?13:37
SethRobertson submodules don't work well if you are doing concurrent development and have full control over all of the slaves13:37
http://www.baka.org/projects/gits.git13:37
wjames6 any readme?13:38
SethRobertson Sure13:38
Inside :-)13:38
It sounds like there is enough interest I should explore putting it on sourceforge or something13:38
rubydiamond → zz_rubydiamond13:38
wjames6 or github even13:39
pielgrzym SethRobertson: cool thing :)13:39
← aresnick left13:39
pielgrzym SethRobertson: thanks for the repo :)13:39
ptor SethRobertson: Looks lovely. I'll definitely try this out.13:39
SethRobertson I'd like the mailing lists and bug tracking of sourceforge. I didn't notice that github had that.13:39
→ Ryback_ joined13:39
← Ryback_ left13:39
→ Ryback_ joined13:39
wjames6 ill jump into your convo since it is interesting to me if you dont mind. we migrated a cms with 50 some odd parts which were all in a cvs13:40
← ExtraSpice left13:40
wjames6 we're now on git with superproject and submodules.13:40
→ trtzrdb joined13:41
→ copelco joined13:41
wjames6 because the system is componentized there are many different configurations one might do. submodules has actually made configuration much easier13:41
SethRobertson It didn't seem very convenient to do tagging, branching, and checkout on all repositories at once13:41
zz_rubydiamond → rubydiamond13:42
SethRobertson With the wrapper, it is `gits checkout -b foo` instead of `git checkout -b foo`13:42
ptor SethRobertson: It looks like I can just keep adding (eh, attaching) slave modules (repos) to the superproject, right?13:42
wjames6 for a developer I may only want to work on core modules and a few add ons, so thats what you put in your superproject.13:42
→ ProperNoun joined13:42
→ aresnick joined13:42
SethRobertson You can do that with gits as well13:42
Though mostly people don't bother except for a few huge slave repos.13:43
← chrisshattuck left13:43
← Fullmoon left13:43
ptor wjames6: What's "superproject" in the submodules context? Something new, or what's already in submodules?13:43
wjames6 its the thing that holds the submodules13:44
from my understanding its really just a repo with submodules13:44
rubydiamond → zz_rubydiamond13:45
← aresnick left13:45
→ artefon joined13:45
→ ddv joined13:45
→ xiangfu joined13:45
wjames6 one of the inconveniences of git is that changes in two repos are hard to ref. so we have multiple core components of we move a file between the two thats kinda lost as a relation.13:45
→ daedra joined13:46
→ thierryp joined13:46
ptor wjames6: Right. I see. But that's the part of (the current) submodules that doesn't fit with how I want to work. (There seems to be some patch sets floating around on the mailing list which may move submodules to something more useful (for me), but for now either git-subtree or this 'gits' seems the best option. Particularly the latter, but must read up on it a bit more)13:46
← lourense left13:47
ptor (the above was about the superproject/submodules question)13:47
wjames6 we came into git thinking submodules would be a huge hassle13:47
→ aresnick joined13:47
→ nadavoid joined13:48
← thierryp left13:48
SethRobertson Yes, it has significant drawbacks when trying to visualize the complete set of changes across all projects. `gits exec gitk` to run gitk on each in turn technically works, but when you have a dozen slaves it is not quite so easy to understand13:48
wjames6 the lost cross module relations is a bummer. and the other thing that is tricky if figuring out how to best manage all contributors submodule forks, since each user needs to fork every part13:48
→ thierryp joined13:48
SethRobertson That is where gits shines13:49
wjames6 can gits do any meta tracking of related commmits? so maybe you run a dual commit on two submodules and it would tract the two keys as a relation?13:49
← scarabx left13:49
← daedra left13:49
← alex__c2022 left13:50
→ Fullmoon joined13:50
wjames6 gitolite with its branch permissions may greatly simplify our secondary issue. so that instead of forking each module we just have development branches in each for each developer.13:50
SethRobertson No, we have not attempted to modify the commit message or something like that to put a nounce in there for that sort of thing13:50
← gnufied left13:51
→ daedra joined13:51
wjames6 i was thinking more like gits commit submodule submodule -- and that would track the keys by running the git commit on each itself and then put that into some record.13:51
SethRobertson Each developer is on a fork different from their local checkout? That seems like...overkill13:51
→ notze joined13:51
wjames6 how do you pull their commits if they dont have a remote?13:51
→ chrisshattuck joined13:52
→ hugoxrosa joined13:52
SethRobertson I suppose it depends on the dev style you are using. I am in a setting where I trust the devs so they can commit into the main shared central repo13:53
→ alvaro_ joined13:53
wjames6 ah well then youre using a cvs model instead of a git model :)13:53
SethRobertson And even if I didn't, I could reach into their local checkouts and see them directly13:53
wjames6 as a project lead i want linus fascist like control over what goes into the deployment/release branch13:54
thats if you have access to their puters13:54
masterkorp can anyone help me : https://bbs.archlinux.org/viewtopic.php?pid=827317#p82731713:54
ptor I must admit I still don't get it about the forking.. what's "fork" in this context?13:54
SethRobertson In the github context.13:54
ptor hm. Ok. But why would it be necessary in order to be able to pull changes? Sorry for being thick.13:55
wjames6 ptor you understand the git concept is each developer is an island - repos are distributed, there isnt a central repository like in cvs.13:55
SethRobertson Each user has their own github account all of which are using the same project name. They upload changes to github and the fascist pulls them13:55
nevyn^_ → nevyn^13:55
SethRobertson Most users at home would/count not allow people on the internet to pull from their home systems' checkouts13:56
wjames6 its a fundamental difference and really worth understanding because almost all git features and confusion stems from this difference13:56
ptor Just so that devs can push to individual server repos and their commits can be pulled from there?13:56
Oh, SethRobertson beat me to it. Right.13:56
← thierryp left13:56
→ spearce joined13:56
→ herdingcat joined13:56
SethRobertson BTW, that model would work just fine with gits. Obviously it cannot automate configuring github :-)13:56
ptor So, the forking model is just a workaround, an intermediate, to allow pulling. IC.13:57
→ thierryp joined13:57
← herdingcat left13:57
→ engled joined13:57
reto_ you can add as many contributor as you like13:57
ptor just pulls between everything! When and if the system's are online or VPN'ed etc.13:58
reto_ and use github like you would use a central svn repo13:58
wjames6 sort of. i dont think ive ever seen suggestion that a local development environment is also a remote.13:58
but yes the remote if for distribution. at least thats my sense of it13:58
ptor okidoki. It's all clear.13:59
Bombe I do have a local development environment that is a remote.13:59
It’s only used locally, though. :)13:59
If my coworkers here in the office would also use Git I’d have three repositories on my local machine.13:59
wjames6 we've done some work using a shared remote. it can get a little more confusing when people push conflicts, but its not terrible. i think its cleaner when each is on his own branch.14:00
ptor wjames6: However, an alternative would be that your single central server has branches for each developer, and they push to it. gitolite could (as far as I understand) introduce access controls on that.14:00
← zenom left14:00
← tungtn left14:00
SethRobertson I often push changes not ready for everyone onto my local dev installs on other test systems14:00
← notze left14:00
← altamic left14:00
wjames6 ptor: yah, i commented as much a few mins ago, that is looking like a much better solution than forking all over the place14:01
SethRobertson It is a fine style if you have somenoe who dedicates their lives to managing the central repo. I have better things to do myself.14:01
Trust but verify14:01
→ shennyg joined14:01
→ sivy joined14:01
wjames6 ive heard others say that before. i dont have time to unwind crappy commits. so the git model works well for me.14:01
← jetienne left14:01
→ zenom joined14:02
SethRobertson The only problem with per-user branches is that how would you support multiple release branches? (e.g. 1.0 + master)14:02
wjames6 i find merges are trivial so im finding it a net gain to have to merge work. i have to review everything anyway.14:02
SethRobertson Everytime you branched you would create an additional branch for each developer?14:02
wjames6 oh so if we have multiple development branches and now a dev needs to work on those multiple - thats a great question.14:03
← freimatz left14:03
→ phb_ joined14:03
→ ch077179 joined14:03
→ simplechat joined14:03
← simplechat left14:03
→ simplechat joined14:03
SethRobertson I review commits but my dev team is quite good and the number of times I have to reject is <1%14:03
→ dmlloyd_road joined14:03
ptor SethRobertson: Also, in a company setting it's often difficult to dedicate a maintainer to do the pulling and merging. What if he's on holidays, or on mission or whatever. Things stop. On the other hand, when all developers push to a central repo you inevitably end up with a mix of features (thus new bugs) and bugfixes on the tip of the master branch, so master is never really stable. And if you dedicate a branch to being 'stable' it's always14:04
obsolete.14:04
SethRobertson Well, I do that by managing the bug tracking system and controlling the milestone on the tickes14:05
milestone controls the branch they work on and whether they are allowed to even work on the ticket at all14:05
← phb left14:05
phb_ → phb14:05
SethRobertson It all depends on the dev style, skill of devs, and attention to detail of the devs14:06
← Fullmoon left14:06
← voxpelli left14:06
→ rayners joined14:06
ptor In any case, whatever way it's done I find it safer and better to do all this in Git than in CVS. E.g. cherry-picking a bugfix back to a maintenance branch, everything related to branching and merging etc.14:07
wjames6 it does. but i do find the review and merging process so trivial that its not a burden to keep each developer on an island. its the initial forking thats still a bit of a bitch.14:07
← ch077179 left14:08
→ ch077179 joined14:08
← ponzao left14:08
← ch077179 left14:08
ptor (Just following the git mailing list and reading the patch sets and their style has improved my own thinking about how changes should be introduced.)14:08
ptor will do some testing with 'gits' in the weekend.14:09
wjames6 well thanks for the interesting convo ptor and SethRobertson its an ongoing concern. gotta get back to doing stuff they pay me for.14:10
→ disappearedng_ joined14:10
ptor So do I :-)14:10
← disappearedng left14:11
← nadavoid left14:11
← EricInBNE left14:13
← avinashhm left14:13
→ avinashhm joined14:14
→ jtaby joined14:14
← aslakhellesoy left14:15
→ StrongOrder joined14:15
← __name__ left14:17
→ FYLGood joined14:17
← FYLGood left14:17
→ kergoth_ joined14:18
kergoth_ git-svn runs log with --pretty=medium (as opposed to diff-tree directly), but its regex doesn't account for decorations, so setting log.decorate in git-config breaks dcommit. just needs to pass --no-decorate, that fixes it14:18
← phb left14:18
← wjames6 left14:19
→ phb joined14:19
→ herdingcat joined14:20
← herdingcat left14:21
→ herdingcat joined14:21
→ boombatower joined14:22
← boombatower left14:22
→ boombatower joined14:22
← spearce left14:22
→ phb_ joined14:23
← phb_ left14:23
→ krosenvold joined14:23
→ phb_ joined14:23
← phb left14:23
phb_ → phb14:23
← rolfb left14:24
→ vector_xyz joined14:25
← _Vi left14:25
vector_xyz if i remove a few files from my folder what do i type to recognize changes14:25
i tried git add .14:25
but i guess that doesnt work with removed/renamed files :z14:25
cbreak_work vector_xyz: man git-add14:25
Gitbot vector_xyz: the git-add manpage can be found at http://git.or.cz/man/git-add14:25
cbreak_work there are a few flags14:25
I always forget which, but -A might work14:25
→ spearce joined14:26
vector_xyz git add -u .14:26
SethRobertson git commit -a -m "Deleted file"14:26
vector_xyz ah14:26
→ _Vi joined14:26
→ jstemmer joined14:27
cbreak_work no need for .14:27
← spearce left14:27
→ spearce joined14:27
→ pigdude_ joined14:28
→ beni- joined14:28
← tjubaluba left14:28
← ajpiano left14:28
← pigdude_ left14:29
→ phb_ joined14:29
→ aziz joined14:29
ptor 'git rm file' will both remove the file and prepare for 'commit -m "Delete file"' in one go.14:29
→ izzie321 joined14:29
← daedra left14:29
← malumalu left14:29
← izzie321 left14:29
ptor well, delete and stage, to put it simpler.14:30
← artefon left14:31
→ malumalu joined14:31
← phb left14:31
phb_ → phb14:31
→ FabianB joined14:31
→ nadavoid joined14:31
← metoikos left14:31
→ skippyish joined14:32
→ krawek joined14:32
SethRobertson James/Alex: I am planning on moving gits development over to sourceforge (under the name gitslave, but keeping the same command name).14:32
→ izzie321 joined14:33
← FabianB_ left14:33
SethRobertson I could wipe away the git history and start from scratch, translate your tanis.org, counterstorm.com, and trustedcs.com email addresses to something else, or leave it alone.14:33
Any opinions?14:33
← _Vi left14:34
→ sftp joined14:34
→ disappearedng__ joined14:34
SethRobertson Ah, wrong window. Sorry14:34
← digitalfiz left14:35
→ stringoO joined14:35
→ EricInBNE joined14:35
→ digitalfiz joined14:36
← digitalfiz left14:36
← izzie321 left14:36
ptor In any case, feel free to mention it here when you've moved to sourceforge :-)14:36
→ izzie321 joined14:36
← disappearedng_ left14:36
SethRobertson I was recommended to send email to the git mailing list last night. Do you or anyone else have any opinions?14:37
vector_xyz ptor what if i removed files14:37
from my dir14:37
in recycle bin14:37
and not only 114:37
but like 2514:37
is there like a - check for any kind of update and add it14:37
SethRobertson git status14:37
vector_xyz if i added 2, deleted 5, modded 2014:37
← formale left14:37
vector_xyz git status only gives me status ?14:37
← oriba_ left14:37
vector_xyz or updates, deletes, addds14:37
← bamccaig left14:37
SethRobertson Yes14:37
← ayanes left14:38
← skippyish left14:38
→ eli joined14:38
SethRobertson Then you `git add .; git commit -a -m "All of my changes"` and all of the adds, delete, and modifications will have been made.14:39
→ skippyish joined14:39
← hebz0rl left14:39
← xxoxx left14:40
← skippyish left14:40
→ _Vi joined14:40
→ skippyish joined14:41
← phb left14:41
→ tjubaluba joined14:41
→ zombor_ joined14:41
← zombor left14:44
vector_xyz thx :]14:44
→ alex__c2022 joined14:45
→ Yuffster joined14:46
← sambarino left14:46
← stringoO left14:46
→ gnufied joined14:46
← opalka left14:46
→ sambarino joined14:47
→ ch077179 joined14:47
← ch077179 left14:47
→ ch077179 joined14:47
← rayners left14:47
← ch077179 left14:48
→ urs_ joined14:48
← urs_ left14:48
→ ch077179 joined14:48
← ch077179 left14:48
→ urs_ joined14:48
← urs_ left14:49
→ hohoho joined14:49
→ ch077179 joined14:49
→ rayners joined14:49
← ch077179 left14:49
← rayners left14:49
ptor vector_xyz: Another way: If you just rm'ed a bunch of files, you can do: git rm `git status --porcelain | grep '^ D' | cut -d" " -f3`14:49
Or whatever incarnation of script-fu you prefer..14:50
cbreak_work or just git add -A as I said half an hour ago :)14:50
← kanru left14:50
ptor Indeed :-)14:50
← _Vi left14:50
← vu3rdd left14:50
ptor Knowing about status --porcelain as a general tool can be useful though.14:51
→ rayners joined14:52
← sambarino left14:53
→ k4336 joined14:53
→ giorgian joined14:54
giorgian hi. how do I find out which commit deleted one row on a file?14:55
→ sambarino joined14:55
← skippyish left14:55
← roop left14:55
→ skippyish joined14:55
← skippyish left14:57
SethRobertson I use gitk, you can also `git log -s'content of row'`14:57
← jtaby left14:58
→ jtaby joined14:59
→ metoikos joined14:59
← Exodist left14:59
giorgian SethRobertson: thanks15:00
ptor SethRobertson: Didn't see your question until now - yes, for me at least, your posting on the mailing list would be great.15:00
← metoikos left15:00
→ kanru joined15:01
→ exodist joined15:01
← frerich left15:01
→ frerich joined15:01
→ metoikos joined15:01
← radarek left15:02
→ tj joined15:02
→ warbee joined15:02
tj → Guest6089815:03
→ akahn joined15:03
Guest60898 → tjholowaychuk15:03
→ yotaff joined15:03
akahn Is there a way to set git to use -v by default for commit?15:03
yotaff alias ?15:04
→ pigdude_ joined15:04
yotaff civ = "commit -v"15:04
← gnufied left15:04
← LRN left15:04
→ Heimidal joined15:06
k4336 anyone have insight/suggestions for setting up GIT server replication? I'm looking into gitolite, gitosis, and possibly Gerrit2; tbh I have very little experience w GIT so far and just kinda looking for suggestions/things to look into15:06
← hakunin left15:07
→ loinclot_ joined15:07
→ hakunin joined15:07
→ roop joined15:07
SethRobertson What do you mean by "git server replication"?15:09
← frogonwheels left15:09
← simplechat left15:10
→ archcoder joined15:10
← jtaby left15:11
→ justin-george joined15:11
→ xxoxx joined15:11
akahn yotaff: okay, i'll use that approach. thanks.15:11
← akahn left15:11
← Belna left15:11
→ t0rc joined15:12
→ Torsten joined15:13
← jsquared left15:13
→ jsquared joined15:14
← chrisshattuck left15:17
← xiangfu left15:18
← hobodave left15:19
→ reflectivedev joined15:21
← metoikos left15:21
→ morganick joined15:22
← cbreak_work left15:22
→ altamic joined15:24
→ ch077179 joined15:24
→ alester joined15:24
← becomingGuru left15:24
→ artefon joined15:24
alester Your branch is ahead of 'origin/master' by 39 commits.15:24
What is the magic to find out what those 39 commits are?15:24
_rane git log master..origin/master15:25
or other way around15:25
can never remember15:25
→ chrisshattuck joined15:25
k4336 SethRobertson: i'm really new to this stuff so pardon me if any of my terminology is wrong here, but bascailly i'm trying to set up a git repository on one server, and have another server in another location mirror any changes that happened on the first server15:25
← Pupeno[work] left15:25
← enherit left15:25
← shiba_yu36 left15:26
alester Thanks, _rane15:26
k4336: OK, go on.15:26
→ _Vi joined15:26
→ Eaven joined15:26
→ skippyish joined15:27
→ nakkor joined15:27
SethRobertson k44336: seems like a post-commit or post-receive hook is what you want. Instead of that, I put in a cron job to pull every hour to a disaster recovery secondary repo.15:27
→ jtaby joined15:28
→ anoldhacker joined15:28
k4336 Ok, that sounds like a good idea.15:28
← Adaptee left15:29
→ Adaptee joined15:30
pigdude_ I'm having trouble pulling a branch...I have never had this issue before. it is not master: http://paste.pocoo.org/show/263229/15:30
ideas?15:30
→ Lgb joined15:30
k4336 My boss was saying to try to hook the Push event with gitosis, so that when ppl here push the other server would automatically pull the changes15:31
anoldhacker I'm trying to get my head around submodules. In my particular case, we have a number of projects which include various (shared) submodules. It seems to me that whenever I change a branch, that I need to accompany it with a submodule update. (The code is required to be present for testing.)15:31
IslandUsurper → IslandUsurperAFK15:32
← izzie321 left15:32
SethRobertson anoldhacker: if you are doing concurrent development on the slave repositories, I have a meta-git wrapper which might be right for you15:32
→ Ramblurr joined15:33
→ jefferai joined15:33
anoldhacker SethRobertson: I figured things would go that way. Thanks!15:33
SethRobertson It does not use submodules in any way.15:33
anoldhacker pigdude_: Are you in a tracking branch? Pull needs to be told where to pull from. If you are not tracking, it cannot tell.15:33
k4336 he wants to set it up so that the guys in the office sync their changes with their local server, but they'd commit their changes to our server here15:34
SethRobertson anoldhacker: I will be moving it to sourceforge very soon, but currently it is at http://www.baka.org/projects/gits.git15:34
← tjubaluba left15:35
anoldhacker SethRobertson: Thkx!15:35
→ fedesilva_ joined15:35
SethRobertson Basically you have a superproject and one or more slave projects. If you want to branch, you can `gits checkout -b newbranch` and the git operation will execute on all repositories.15:35
exodist → Exodist15:35
← skoop left15:36
← justin-george left15:37
← fedesilva left15:37
fedesilva_ → fedesilva15:37
mmattice → temptme15:38
→ becomingGuru joined15:39
esc git scripting question, how do i get name of current branch?15:39
← errordeveloper left15:39
jast git symbolic-ref HEAD15:39
← guardian left15:39
esc jast: thank15:39
s15:39
temptme → mmattice15:40
← sifi left15:40
← skippyish left15:40
jast you can also parse the output of "git name-rev HEAD" instead if that's easier; it contains the shortened branch name15:41
← cilly left15:41
← Titosemi left15:41
→ mefesto joined15:41
← alex__c2022 left15:42
← becomingGuru left15:42
→ cilly joined15:43
mefesto if i cherry-pick a commit from branch (new-stuff) into master, then later merge new-stuff into master, will i get merge conflicts? it's just one commit that is fairly self contained15:43
→ skrul joined15:43
← morganick left15:44
← GarethAdams|Work left15:45
← airborn left15:46
← hyperair left15:46
← khmarbaise left15:46
← robgleeson left15:47
→ hyperair joined15:48
→ eblake joined15:48
← eblake left15:49
→ HG` joined15:49
← jtaby left15:49
nakkor mefesto: you probably wont get any merge conflicts unless you change some lines on either commit, but what you will have is two seperate commits recording the same chnage15:49
if you can, merge is preferred15:50
anoldhacker mefesto: Git keeps up with the commits already present in a branch when merging is done. The only way you might run into an issue is if you do a merge --squash.15:50
← giallu left15:50
anoldhacker mefesto: I've not played with it but a merge --rebase would likely "disappear" the cherry-picked commit--which is a good thing.15:50
sitaram k4336: ping15:51
mefesto ok thanks nakkor, anoldhacker15:51
sitaram k4336: gitolite has mirroring support. I myself have not started using it in prod due to non-technical reasons, but I know someone who is and is very happy with it15:52
→ skippyish joined15:52
nakkor mefesto: just make sure not to rebase a branch that is live, upstream rebased is a *bad* thing15:53
→ khmarbaise joined15:53
pigdude_ anoldhacker: I thought I was...15:53
anoldhacker: I guess I am not15:53
anoldhacker: I never have trouble pushing this branch to origin but this is the first time I've had to pull changes15:54
→ gnufied joined15:54
mefesto nakkor: what i was going to do is when this new-stuff branch is finished, merge it into master then delete new-stuff15:54
nakkor ok, as long as new-stuff hasn't been made available upstream. If it is, then other ppl will have issues if you do a rebase and then push that change upstream again15:56
mefesto just happened that a commit went into new-stuff that really should be in master so i cherry-picked it so i could get a build out.15:56
← gnufied left15:56
← hakunin left15:57
→ gnufied joined15:57
← skippyish left15:57
SethRobertson If I have many many tags on a particular revision, is there any convenient way to get the list of all such tags? `gitk` truncates the list at the screen width. `git describe` only shows one15:58
→ skippyish joined15:58
← skippyish left15:59
← aresnick left15:59
jast SethRobertson: something involving git log --decorate, probably15:59
→ skippyish joined15:59
nakkor SethRobertson: git tag16:00
→ mdel joined16:00
nakkor that will list all your tags line by line16:00
and you can pipe it to a file if you need to16:00
← skippyish left16:00
SethRobertson But will not tell me where they belong. `git log --decorate` appears to be good for me16:01
→ skippyish joined16:01
nakkor ahh, sorry, didn't know what you were looking for16:01
anoldhacker mefesto: Then you really don't want a merge-rebase. :P16:02
jast git show --decorate --pretty=oneline <commit>16:02
→ jrmuizel joined16:02
← Heimidal left16:02
→ akahn joined16:03
→ kenichi joined16:03
← jrmuizel left16:04
← skippyish left16:05
← giorgian left16:05
mefesto i thought i had it straight but now im confused :)16:05
master branch hasn't changed much since the new branch was created16:05
all dev is in the new branch atm16:05
→ skippyish joined16:06
mefesto master is treated as our release branch so it's pretty stable16:06
SethRobertson So?16:06
→ cannonball joined16:06
mefesto so there was one commit that was cherry picked into master16:06
SethRobertson Great16:07
git will do the right thing16:07
pigdude_ anoldhacker: so how would I style that pull call?16:07
mefesto so all that is going to happen is that once new branch is done... master will merge it in16:07
pigdude_ git pull ... ?16:07
mefesto from that point new branch is basically dead and will be deleted16:08
anoldhacker pigdude_: Is it your intent for this to be a tracking branch?16:08
mefesto i just wasnt sure if the cherry pick would cause a conflict when that merge happened16:08
i should be good. nakkor, anoldhacker thanks again for the help.16:09
→ hobodave joined16:10
pigdude_ anoldhacker: I just want to pull origin release-0.1 to local release-0.116:10
← beni- left16:10
anoldhacker pigdude_: So when there are more updates to origin release-0.1, you will want to pull them as well, right?16:11
← t0rc left16:12
→ morganick joined16:12
pigdude_ anoldhacker: likely. I'd like to know the manual way to do it too though16:13
← kanru left16:13
→ ProperN[out] joined16:13
anoldhacker Oh, so you want to be able to fish with flies AND worms?16:14
← ProperN[out] left16:14
→ ProperN[out] joined16:14
→ Pupeno[work] joined16:14
pigdude_ anoldhacker: and spinners :^)16:14
← TeckniX left16:15
anoldhacker Okay. The "real" fix is to edit .git/config to make the branch track. You can just cargo cult the entry for master onto your release-0.1 entry.16:15
→ TeckniX joined16:15
→ kumbayo joined16:15
pigdude_ anoldhacker: `git pull origin release-0.1` seemed to work. is that correct?16:15
anoldhacker yep. I was just double-checking the manual.16:16
pigdude_ thanks, now I know :^)16:16
← rayners left16:17
← ProperNoun left16:17
ProperN[out] → ProperNoun16:17
→ radarek joined16:17
← thierryp left16:18
← tty1 left16:18
← tstclair left16:18
→ irc2samus joined16:18
← akahn left16:19
→ dreiss joined16:20
→ twomashi joined16:21
→ rayners joined16:22
← Chaser left16:23
→ Chaser joined16:24
PerlJam spearce: ping16:24
← herdingcat left16:25
→ ajpiano joined16:25
→ TheEmpath joined16:25
k4336 sitaram - my boss actually said he would really like to use gitolite because it offers security at the branch level16:26
← _Vi left16:26
← anoldhacker left16:26
→ jrmuizel joined16:26
← Torsten left16:27
→ t0rc joined16:27
← psoo left16:28
spearce PerlJam: pong16:28
k4336: but only for writes. gerrit can do reads too. :-)16:29
← RichGuk left16:29
PerlJam spearce: you did .git-completion.bash, right?16:29
spearce yea16:29
→ tty1 joined16:30
PerlJam spearce++ for that :) but I wonder if it's possible to tweak GIT_PS1_SHOWUNTRACKEDFILES somewhat? currently it shows % even when git status says working directory clean. I believe that's because of the ignored files.16:32
→ kukks joined16:32
spearce that i didn't do, its someone else's work. but they probably missed a flag to tell it to consider the standard ignore files16:32
→ kpreid joined16:32
PerlJam ah16:33
← awilkins left16:33
jast it actually does use --exclude-standard16:33
IslandUsurperAFK → IslandUsurper16:33
→ psoo joined16:33
PerlJam spearce: do you have any feeling one way or the other about making it pay attention to or ignore .gitignore?16:33
→ _Vi joined16:34
jast it already does that16:34
← kpreid left16:34
jast (pay attention to it)16:34
← intripoon left16:34
→ kpreid joined16:34
PerlJam Hrm.16:34
← thoolihan left16:35
→ mbroeker joined16:35
← jrmuizel left16:36
← bphogan left16:36
← kpreid left16:36
k4336 spearce that's a pretty cool feature16:36
← psoo left16:36
PerlJam Perhaps something is screwed up on my local machine.16:37
← mkramer left16:38
→ urkud joined16:39
→ herdingcat joined16:40
EvanR-work i fetched from someone with a million branches16:40
he almost certainly didnt want them to be public, and its cluttering up my gitk16:40
can i removed them or not show them?16:40
→ LRN joined16:41
← boombatower left16:42
→ tty2 joined16:42
PerlJam doh! my .bashrc was broken.16:43
spearce: sorry for the noise16:43
→ mkramer joined16:43
← mkramer left16:43
→ enherit joined16:43
→ aah joined16:43
← tty1 left16:45
→ urkud_ joined16:46
← urkud left16:47
→ Heimidal joined16:49
← imMute left16:49
← tvw left16:50
→ Nickeeh joined16:51
→ andym joined16:51
→ kpreid joined16:52
andym i'm trying to merge a pull request from github, whenever i do git merge offbyone/fafdbe1 i get 'offbyone/fafdbe1' does not point to a commit16:52
is there a way to tell what commit's are available, and hence what i need to use in git merge16:52
SethRobertson Have you fetched the information yet?16:52
→ izzie321 joined16:53
← izzie321 left16:53
SethRobertson `git fetch offbyone` I think, then you can look at what has changed16:53
andym yeah i did: git fetch offbyone16:53
→ akahn joined16:53
fr0sty andym: 'git merge fafdbe1'16:53
→ izzie321 joined16:53
andym ah ha!16:53
→ jrmuizel joined16:53
andym thanks fr0sty, SethRobertson16:53
← goddard left16:54
fr0sty np16:54
SethRobertson To see what has changed, `git log ^offbyone HEAD` and/or `git diff HEAD...offbyone`16:54
→ linusoleander joined16:54
→ montylounge joined16:55
→ Chillance joined16:56
← jrmuizel left16:56
← bbatsell left16:56
← tty2 left16:56
→ efarrer joined16:56
← izzie321 left16:57
→ tty1 joined16:57
← Nickeeh left16:57
→ jrmuizel joined16:59
→ tizbac joined16:59
→ imMute joined17:00
← naimon left17:01
→ dr_win joined17:01
← gnufied left17:01
→ jalvarado joined17:02
→ Radar joined17:03
→ psoo joined17:03
← Radar left17:04
→ zombor joined17:04
← zombor_ left17:04
← artefon left17:04
← d1b left17:05
→ d1b joined17:07
← brian_g left17:09
→ tjholowaychu joined17:09
← tjholowaychuk left17:09
← t0rc left17:10
→ comptekki joined17:11
comptekki i asked the ? yesterday about working on one project from 2 boxes17:12
← johne left17:12
→ _frijole joined17:12
comptekki would it work to to the push from main box then do a fork on the laptop then do work on laptop then do a pull and merge when i got back in to the office17:12
can this be done on the same github account?17:13
← herdingcat left17:13
vector_xyz does anyone know of any good git hosting sites that also has calendar, tickets, also perhaps FTP Deployment stuff... (must not be free) commercial is fine17:13
comptekki I trying to think how to maintain same repo from 2 boxes17:14
fr0sty comptekki: one github account and push/pull to the same repository.17:14
→ altamic_ joined17:14
← GodEater left17:15
comptekki someone mentioned git add remote - would I do that from the laptop then do a pull (diff than clone??)17:15
→ GodEater joined17:16
← GodEater left17:16
→ GodEater joined17:16
← altamic left17:17
altamic_ → altamic17:17
fr0sty close creates a repository, fetch updates it.17:17
→ hakunin joined17:18
← Yuffster left17:18
← akahn left17:18
vector_xyz can someone recommend something better then Assembla17:19
← twomashi left17:20
→ skyl joined17:20
skyl I have a couple of files that are deleted but not `add`ed; what's the recommended way to just put them back?17:21
← roop left17:21
← LongBeach left17:21
← jtaby_ left17:21
skyl hrm, use "git checkout -- <file>..." to discard changes in working directory .. might be a good clue17:21
→ LongBeach joined17:22
← dr_win left17:22
skyl yay, ~thanks17:23
← linusoleander left17:27
← jrmuizel left17:29
→ gnufied joined17:29
→ tjubaluba joined17:29
← jefferai left17:29
← Ramblurr left17:29
← tjubaluba left17:32
→ brainbox1100 joined17:32
→ loinclo__ joined17:32
← jshsu left17:32
→ singalong joined17:32
→ jshsu joined17:33
← jshsu left17:33
← curtana left17:33
← bluenovember left17:34
← loinclot_ left17:34
→ bambule joined17:34
→ gzmask joined17:35
→ csanz joined17:35
← avinashhm left17:35
→ batok joined17:35
→ robgleeson joined17:35
← singalong left17:35
← GodEater left17:35
← sambarino left17:35
→ avinashhm joined17:36
EvanR-work what are tags for?17:36
→ sambarino joined17:36
EvanR-work we arent using any at all17:36
gzmask hello folks. got a rookie qestion: I follow the gitosis R EADME and I got these: git clone [email@hidden.address] into gitosis-admin...17:36
[email@hidden.address] password:17:36
fatal: 'gitosis-admin.git' does not appear to be a git repository17:36
fatal: The remote end hung up unexpectedly17:36
← bambule left17:37
→ frakturfreak_ joined17:37
gzmask and I changed the command to git clone [email@hidden.address] then it works, but after I git push the server file doens't got modified at all17:37
→ pantsman joined17:37
→ bambule joined17:38
→ GodEater joined17:38
← GodEater left17:38
→ GodEater joined17:38
← frakturfreak left17:39
frakturfreak_ → frakturfreak17:39
→ vu3rdd joined17:40
dwmw2 → dwmw2_gone17:41
→ sh1mmer joined17:41
→ kipras joined17:41
→ ProperN[out] joined17:41
← ProperN[out] left17:41
→ ProperN[out] joined17:41
← _frijole left17:42
← Yuuhi` left17:42
→ nicoulaj joined17:42
→ MikhX joined17:42
← hyperair left17:43
← andym left17:43
→ aresnick joined17:44
→ andym joined17:45
← ProperNoun left17:45
ProperN[out] → ProperNoun17:45
→ Ramblurr joined17:46
← Ramblurr left17:46
→ Ramblurr joined17:46
→ jefferai joined17:46
→ guardian joined17:47
← Rhonda left17:47
← enherit left17:47
→ tstclair joined17:48
IslandUsurper EvanR-work, usually they mark a specific, published version of the code, which is available as a package somewhere else. but, like branches, they only have whatever meaning you give them17:48
→ cytrinox_ joined17:48
→ jaloga joined17:50
→ Rhonda joined17:50
→ lightcap joined17:50
← cytrinox left17:50
cytrinox_ → cytrinox17:50
← yotaff left17:51
← jaloga left17:52
→ mxi joined17:52
← madsy left17:52
← Schmallon left17:53
→ zombor_ joined17:53
→ madsy joined17:53
← madsy left17:53
→ madsy joined17:53
→ linusoleander joined17:53
← mxi left17:53
← zombor left17:54
← ljungk left17:54
← Rhonda left17:54
→ ljungk joined17:54
→ Rhonda joined17:54
bremner gzmask: if you are just getting gitosis working, please consider using gitolite instead. gitosis is no longer supported by the author (or anyone else, really)17:55
→ mkramer joined17:55
mkramer god damnit, this just keeps coming up for me17:55
I want to rebase and except all conflicts from upstream17:55
gzmask arggg.... thanks bremner17:56
← kpreid left17:56
← skippyish left17:56
mkramer accept*17:57
rather different meaning there17:57
NfNitLoop "accept all conflicts"?17:57
→ zombor joined17:57
← tatsuya left17:57
→ skippyish joined17:57
→ corni joined17:58
← BlindOracle left17:58
← zombor_ left17:58
← cpf_ left17:58
← archcoder left17:58
→ JoseLobato joined17:59
doener mkramer: -X ours?17:59
→ hyperair joined17:59
doener mkramer: ah, heh, only in 1.7.3-rc018:00
← loinclo__ left18:00
← Rhonda left18:02
→ Rhonda joined18:02
← comptekki left18:02
← caseyw left18:02
→ zombor_ joined18:03
← zombor left18:04
→ akahn joined18:04
→ kpreid joined18:05
← sambarino left18:05
← Determinist left18:05
SethRobertson mkramer, if it keeps happening, git-rerere might be your ticket18:07
→ sambarino joined18:07
→ Yuuhi joined18:07
→ jrmuizel joined18:08
← xxoxx left18:08
→ loinclot_ joined18:10
← apsysoev left18:11
← madsy left18:12
← warbee left18:12
← aresnick left18:12
← timj_ left18:14
← loinclot_ left18:14
← posciak left18:14
→ loinclot_ joined18:14
zombor_ → zombor18:16
← akahn left18:17
→ caseyw joined18:17
tjholowaychu → tjholowaychuk18:20
→ menace joined18:22
→ aboudreault joined18:24
aboudreault hi18:24
← EricInBNE left18:24
aboudreault I set my git client to use a SSH key, which doesn't have a passphrase. Am I supposed to be able to used it? git still request the passphrase18:24
→ drizzd_ joined18:25
albel727_ aboudreault: unix, win?18:26
← bentob0x left18:27
aboudreault aboudreault, unix.18:27
albel727_, unix.18:27
SethRobertson aboudreault: If you have keys with passphrases ssh will require you to give up on those first18:27
you should be able to hit return and get to the passphraseless key18:27
← ajpiano left18:27
aboudreault I think the problem may be that its not a "rsa" but a dsa, could it be?18:28
SethRobertson or have a separate keyfile and somehow get git's ssh to look there18:28
no18:28
→ sixteneighty joined18:28
SethRobertson Assuming it works at all, of course18:28
Ilari I use passphraseless DSA keys with no trouble. SSH won't prompt for passphrase.18:28
← drizzd left18:28
aboudreault I'm using my key for debian package and my mail.. so I assume it works.18:28
→ askhader joined18:28
← HG` left18:28
aboudreault ok18:29
Ilari And the passphrase prompt comes from SSH, not git.18:29
askhader So if I use 'git push origin master' to push from my host to a repository, how do I then update the master to this 'pushed' version?18:29
→ bentob0x joined18:29
SethRobertson aboudreault: Do you have multiple private keys in your file?18:29
aboudreault no only one18:29
SethRobertson askhader: It will automatically be updated18:29
selckin aboudreault: sdoes ssh githost ask a pass?18:29
askhader SethRobertson: it isn't...18:30
loinclot_ → loincloth_18:30
SethRobertson Is the master a bare repo or a checked out repo?18:30
askhader bare.18:30
Just my own little repo.18:30
← JoseLobato left18:30
aboudreault selckin, yes, If I hit enter without typing anything... look ok.. but permission denied.18:30
← csanz left18:31
aboudreault It seems the problem is from the server side18:31
→ giallu joined18:31
← hipe left18:31
← zombor left18:31
SethRobertson If you type in a passphrase does it work?18:31
Ilari aboudreault: Is it passPHRASE or passWORD?18:31
aboudreault passPHRASE yes18:31
askhader SethRobertson: git push reports that everything is up to date but when I go to the actual master host and view the files I edited, no changes were made. Do I need to issue a command on the master to apply the changes or something?18:31
aboudreault Entering passphrase for key xxx18:31
Ilari aboudreault: Maybe SSH is trying some wrong key that does happen to have passphrase set?18:32
aboudreault Permission denied (publickey,gssapi-with-mic).18:32
SethRobertson askhader: how do you view the files on the upstream?18:32
→ mischosch joined18:32
askhader SethRobertson: Let me explain the setup I have.. It's rather non-conventional. The files exist on a server. In the top level directory of the files I issued a git --init, then I cloned the directory on a host where I work on the project.18:32
SethRobertson: So I guess to answer your question, with a file edito.18:33
editor18:33
tango_ askhader: so you are pshing to a non-bare editor?18:33
SethRobertson So, the upstream is NOT a bare repo18:33
→ stringoO joined18:33
tango_ non-bare REPO, sorry18:33
askhader sorry, define bare?18:33
_rane angry bare18:33
tango_ askhader: bare = without a working copy18:34
_rane or tree? or both?18:34
tango_ = not checkoued out18:34
askhader hm. I'm not sure.18:34
albel727_ askhader: man git-init, see option --bare18:34
Gitbot askhader: the git-init manpage can be found at http://git.or.cz/man/git-init18:34
→ timj joined18:34
← timj left18:34
→ hipe joined18:34
SethRobertson https://git.wiki.kernel.org/index.php/GitFaq#Why_won.27t_I_see_changes_in_the_remote_repo_after_.22git_push.22.3F18:35
← hipe left18:35
albel727_ askhader: oh wait, this man is totally unhelpful.18:35
← tildeequals left18:35
askhader albel727_: =/18:36
→ akahn joined18:36
← IslandUsurper left18:36
→ frakturfreak_ joined18:36
← priidu_ left18:37
fr0sty askhader: generally you would have a separate bare repository that you would push to and then you would pull from that same repository on the server.18:37
faq non-bare18:37
Gitbot fr0sty: Pushing to non-bare repositories is discouraged. See https://git.wiki.kernel.org/index.php/GitFaq#non-bare18:37
→ IslandUsurper joined18:38
← montylounge left18:38
← frakturfreak left18:38
frakturfreak_ → frakturfreak18:38
← IslandUsurper left18:39
→ frakturfreak_ joined18:39
← frakturfreak_ left18:40
→ ajpiano joined18:40
mkramer SethRobertson: doener: thanks guys18:40
→ aspotashev joined18:40
→ IslandUsurper joined18:42
askhader I changed the repository to bare and pushed changes, and still they do not appear in the master18:42
tango_ they do not appear in the checkout directory because that one is not updated by pushes18:43
fr0sty askhader: and changing a non-bare repository to a bare repository is not what I suggested.18:43
askhader I just read the FAQ18:43
→ bphogan joined18:43
fr0sty you need a 3rd repo: local, remote-bare remote-checked-out18:43
askhader ffs18:43
EvanR-work the more repos the merrier18:44
tango_ lol18:44
askhader fr0sty: Can you direct me to documentation that outlines this process?18:44
tango_ askhader: the other way to do this is to push to remote branches which are NOT checked out18:44
→ p3rg2z joined18:44
→ aresnick joined18:45
← robgleeson left18:45
askhader And if the changes don't go to the master, to where do they get sent?18:45
← corni left18:45
EvanR-work IslandUsurper: ah i see18:45
→ xperience joined18:45
EvanR-work so like v1.456 could be a tag18:45
← kpreid left18:46
← nevyn^ left18:46
← vu3rdd left18:46
fr0sty askhader: no, I can't.18:46
← _Vi left18:46
fr0sty askhader: the changes you pushed to the master branch on your non-bare remote were pushed there.18:46
→ timj joined18:47
fr0sty 'git status' in that directory will show you that the working copy is not in sync with 'HEAD'18:47
you need 'git reset --hard' to be able to see the changes in the repository you pushed to.18:47
→ ziggles joined18:48
→ theoros joined18:49
→ planet8 joined18:50
→ _frijole joined18:50
← rayners left18:50
← SimonP86 left18:51
← ziggles left18:51
planet8 I am used to working with mercurial and have just started using Git. I have initialized an empty repo on a remote backup machine. I have cloned the repo to my local machine and have added and commited some files.. when I want to push the files I can't, I get the "updating the current branch in a non-bare repository" what does this mean? What's the problem?18:52
→ comptekki joined18:52
EvanR-work planet8: dont push to non bare repos18:52
fr0sty planet8: faq non-bare18:52
Gitbot planet8: Pushing to non-bare repositories is discouraged. See https://git.wiki.kernel.org/index.php/GitFaq#non-bare18:52
comptekki i created a test github rep and pushed the test file18:52
EvanR-work planet8: remove your remote backup, and clone --bare from your local version18:53
comptekki on the desktop in then did a clone of it and edited a file18:53
EvanR-work now you can push18:53
comptekki I added the rsa key from the desktop on github18:53
in then do a push of the change and it gives me a error: cannot access ursll .... git repo return code 2218:54
fatal: git-http-push failed18:54
planet8 EvanR-work, sorry if I don't understand, but I need to push to the remote machine.18:54
comptekki what am i doing wrong?18:54
EvanR-work planet8: you need to make the remote machines repo bare18:54
→ programble joined18:54
tango_ so remove its clothes18:55
EvanR-work hot18:55
planet8 what's bare repo?18:55
EvanR-work a repo with no working tree18:55
→ BlindOracle joined18:55
EvanR-work just history18:55
← reflectivedev left18:56
comptekki planet8: take a look at http://www-cs-students.stanford.edu/~blynn/gitmagic/ch03.html18:56
EvanR-work see the FAQ18:56
fr0sty comptekki: you shouldn't be pushing to an http://... url18:56
your remote should be defined as a git://... url18:56
comptekki fr0sty: ah yes - i need to change a config somewhere to ssh18:56
EvanR-work or ssh18:56
knittl git brancc --set-upstream origin/next opti18:56
EvanR-work git:// has no authentication18:56
knittl works exactly the wrong way one would expect18:57
fr0sty yeah, ssh18:57
sorry.18:57
knittl also git branch --set-upstream origin/next (without local branch)18:57
comptekki how come i don't need to do anything special on the laptop (init the first test repo)18:57
→ Rh0nda joined18:57
← Rh0nda left18:57
fr0sty what does 'git remote show origin' tell you on the two machines?18:57
← jsquared left18:57
fr0sty I'm guessing the url is different, comptekki.18:57
planet8 EvanR-work, comming from hg this doesn't make sense at all.. must look into it18:58
comptekki just a sec18:58
sitaram spearce: so you guys have completed the read-restrictions support in gerrit huh? nice... I guess this is because you have your own git stack right? (you dont use the normal git.git)18:58
← jrmuizel left18:58
EvanR-work planet8: i dont know how hg works, mind explaining?18:58
i heard its similar to git18:58
→ zombor joined18:58
spearce sitaram: yes, exactly18:59
planet8 EvanR-work, found this http://www.gilesthomas.com/?p=37918:59
EvanR-work planet8: well we posted several real faqs about it18:59
sitaram spearce: do you use jgit? or really your (gerrit's) own?18:59
comptekki on the desktop it has18:59
local branch git pull19:00
local ref git push19:00
spearce sitaram: jgit. we taught jgit how to do branch level controls by allowing the embedding application to implement a ref filter interface that jgit uses to decide which refs to offer to the current client19:00
comptekki whereas the laptop only has local ref git push19:00
← bphogan left19:01
fr0sty comptekki: what is the 'Fetch URL:' and 'Push URL:' ?19:01
also, what is 'git --version?'19:01
→ Fullmoon joined19:01
sitaram spearce: is that fundamentally different from a hypothetical "pre-upload" hook?19:01
comptekki ah one says git: (laptop) and the desktop says http:19:01
do i just do a git config something right to set this?19:02
spearce sitaram: for upload, no, its the same idea. for receive-pack its a bit more complicated.19:02
PerlJam comptekki: or edit .git/config by hand19:02
← tjholowaychuk left19:02
sitaram spearce: for receive-pack you already have "update" hook in git.git... wouldn't it be similar in jgit?19:02
EvanR-work planet8: this blog is mostly right. use a bare repo for public shared stuff that is *pushed* to19:03
→ tildeequals joined19:03
EvanR-work such a thing does not need working tree files19:03
comptekki ok - odd that both ~/.gitconfig on laptop and desktop say the same but one has git:// and the other http://19:03
EvanR-work do all your work in non bare repos like the one you normally work in19:03
spearce sitaram: we do. but when a subset of refs are offered to a client, you need to prove during the push that a commit or a tree pushed by the client does not reference an object they cannot see. and to be safe, you also need to consider delta bases that aren't sent in the thin pack.19:03
→ tj joined19:03
comptekki is the info in the repo data19:04
spearce sitaram: the current hook system doesn't support this validation.19:04
← Fullmoon left19:04
EvanR-work and in that case use git pull19:04
tj → Guest9346119:04
spearce sitaram: not that it cannot be added, we added it to jgit, it could be added to native git too19:04
Guest93461 → tjholowaychuk19:04
→ Shurakai_ joined19:04
sitaram spearce: couldnt they have obtained those objects from elsewhere (like my boss, with higher privs, giving it to me so I can complete my tasks)?19:04
EvanR-work planet8: the comment at the bottom is basically what i suggested, recreate your 'backup repo' by doing git clone --bare on the one you were working in19:05
← copelco left19:05
spearce sitaram: yes, but then their client should have sent a copy to the server under the assumption that since you can't see it, the server doesn't have it.19:05
→ bphogan joined19:05
← tobiassjosten left19:05
spearce sitaram: its only the evil client that has a SHA-1 from somewhere, but can't see it, so tries to see it by creating a new commit or tree that points at it and sends a partial pack to the server. a server offering only certain refs might happily accept this partial pack, because it actually has the missing objects.19:06
planet8 EvanR-work, this is gonna take some getting used to :) Thanks!19:06
sitaram spearce: and in that case (if the client sends it) it is allowed?19:06
spearce right19:06
sitaram spearce: got it19:06
spearce if the client sends us a copy, they can obviously see it, they have a copy19:06
← sixteneighty left19:06
← efarrer left19:07
← Shurakai left19:08
→ copelco joined19:08
sitaram spearce: while I've got you chatting, you have your own internal sshd also -- does it use a database and have you measured how many keys it can take and still perform ok?19:09
spearce yes it uses a database, and we can take a ton of keys and still be ok because each user has their own username.19:10
so we can very quickly narrow our key search to just 2-5 keys almost instantly19:10
comptekki thx all that workd - i added [remote "origin"] url = [email@hidden.address]19:11
and it pushed ok19:11
:)19:11
EvanR-work also you could have done19:11
← kumbayo left19:11
EvanR-work git remote add origin git@github:name/repo.git19:11
comptekki EvanR-work: thx19:13
← raz left19:14
comptekki i deleted it out of my ~/gitconfig file and tried the command line and it says it already exists19:14
where is it being put?19:14
EvanR-work git remote lists them19:14
git remote rm origin will remove it19:14
sitaram spearce: "each user has their own username", meaning gerrit URLs are not ssh://<some-common-id-like-git>@server but ssh://myname@server?19:14
spearce yes19:14
comptekki i did git remote and all it says is "origin"19:15
EvanR-work comptekki: probably in .git/refs/remotes19:15
comptekki: right isnt that the one you said you deleted?19:15
git remote rm origin will really delete it19:15
spearce sitaram: review.source.android.com has 1058 keys on right now, it doesn't care how many are in the table.19:15
→ kpreid joined19:15
comptekki i deleted from a file ~/.gitconfig on os x19:15
→ blaines joined19:15
EvanR-work oh19:16
← cehteh left19:16
EvanR-work dont use that for repo-specific settings19:16
sitaram spearce: but those users cannot be real users on the actual unix box; they're "virtual" users that exist only in the gerrit instance?19:16
comptekki there is no ~/.git folder19:16
← p3rg2z left19:16
spearce sitaram: correct19:16
EvanR-work .git in your repo19:16
not the home dir19:16
comptekki ah yes19:16
doy :(19:16
EvanR-work origin is likely different for different repos19:16
comptekki so git must get it from ~/.gitconfig then stick in for future use19:17
EvanR-work put global settings in there19:17
not repo specific stuff like remotes and branches19:17
← gnufied left19:18
→ gp5st joined19:18
sitaram spearce: got it; thanks...!19:18
comptekki i see there is a repo/.git/config that has [remote "origin"] section19:19
gp5st i'm sorry. i'd search for this but i don't know what to search for. i have a copy of a repo that's accessible via http. how do i update it so that the most recent updates are available via http?19:19
zz_rubydiamond → rubydiamond19:19
EvanR-work comptekki: right19:19
← altamic left19:19
← akahn left19:19
→ brett_h joined19:19
brett_h does anyone know why cherry-pick -x would "Append the note only for cherry picks without conflicts."19:20
→ Mage joined19:20
→ cehteh joined19:21
← andym left19:21
← stringoO left19:21
← loincloth_ left19:22
→ eletuchy joined19:22
← morganick left19:25
← MikhX left19:25
→ trevor_morse joined19:27
→ JoseLobato joined19:29
→ frakturfreak_ joined19:33
← redbaritone left19:33
← aspotashev left19:33
← albel727_ left19:34
← frakturfreak left19:34
frakturfreak_ → frakturfreak19:34
← urkud_ left19:35
← sh1mmer left19:35
← hugoxrosa left19:36
→ q0rban joined19:36
← trevor_morse left19:37
comptekki i added a line to a file and pushed it up19:38
i then for a test tried to do a push from the laptop and got an error - expected19:38
planet8 EvanR-work, Mercurial is so much different in this regard, I just don't think I am going to get used to bare vs non-bare dirs. It is upstructing my work habit.19:38
comptekki i then try to do a pull and it says it has a merge conflict because I added a line on the laptop and committed it19:38
automatic merge failed19:39
fix conflicts -19:39
how do I fix this conflict?19:39
EvanR-work planet8: well, gits ideal work habit is to not have central repos19:39
← lightcap left19:39
← guardian left19:39
EvanR-work people just pull from who they want changes from19:40
but barerepos are there for times when you want a public repo only used for pulling and no work19:40
storage only19:40
planet8 EvanR-work, Mercurial has that too, but also makes it possible to push, but at the same time it doesn't restrict it to "storage only".19:41
EvanR-work it is possible to force a normal git repo to accept pushes and automatically update its working tree19:41
→ spaceonline joined19:41
_frijole I deleted a file in one branch, and didn't in the other. now I can't merge. and it won't let me checkout the other branch to delete it there... WTH do I do?19:41
planet8 EvanR-work, how is that done?19:41
EvanR-work there is an update hook which you can use to run git checkout master for instance19:42
i mean19:42
git reset --hard19:42
fr0sty comptekki: http://www.kernel.org/pub/software/scm/git/docs/user-manual.html#resolving-a-merge19:42
EvanR-work any reason why you would want this?19:42
_frijole fro0sty: is that for me?19:42
fr0sty: that link for me?19:43
fr0sty friskd: might work for you too. Do you want to keep the file or throw it away?19:43
planet8 EvanR-work, normally the pull only strategy works fine and we use it, but I have a remote machine that I use as a private backup, and with mercurial I just push to it from time to time.19:43
→ TravisCarden joined19:44
← aboudreault left19:44
EvanR-work i have something similar on my machine19:44
planet8 EvanR-work, I tried the same approach with Git and ran into the problem.19:44
EvanR-work i have a repo i use for work, and i have a bare repo i push to regularly19:44
people only know about my bare repo, they dont know the difference19:44
i dont push stuff to it i dont want them to get19:44
→ disappearedng_ joined19:45
EvanR-work i also push to the 'central' bare repo19:45
fr0sty planet8: you have trouble because you insist oh having a working tree on your backup copy.19:45
EvanR-work right, a backup repo does not need to be non bare19:45
makes no sense19:45
planet8 Fr0sty, ahh, yes that makes sense19:45
← masterkorp left19:45
TravisCarden Newbie question: I'm using git to manage my software project, but I'm using a couple of git projects as part of it. I want to be able to update those sub-projects through git, but I also want to track them as part of the main project. Can anyone give me any direction?19:46
planet8 I am just used to beeing able to "see" the files in the backup as well19:46
← gp5st left19:46
EvanR-work use gitk19:46
← disappearedng__ left19:46
→ rayners joined19:46
→ opalka joined19:47
→ frakturfreak_ joined19:47
← frakturfreak left19:47
planet8 EvanR-work, Fr0sty, thanks, now I understand fully.19:47
frakturfreak_ → frakturfreak19:47
EvanR-work or a git webpage19:47
← kipras left19:47
EvanR-work the reason is that a working tree may have changes to it that will be either overwritten or unable to be merged correctly from the pushing end19:49
'i wouldnt change the files though' well, you dont need to have files at all19:50
planet8 EvanR-work, it makes perfect sense, a wise precaution. I wonder how Mercurial views this problem, but as I remember a merge problem presents itself.19:50
← voob_of_doom left19:50
← Tommy[D] left19:53
← kenneth_reitz left19:54
← tstclair left19:55
← muep left19:55
← unrael left19:55
→ robgleeson joined19:55
→ unrael joined19:55
← q0rban left19:55
→ snearch joined19:56
chrisshattuck What's the GIT equivalent to svn status -u ?19:58
EvanR-work what does it do?19:58
chrisshattuck Check19:58
EvanR-work check what19:58
chrisshattuck (oops, 1 sec ;) 019:58
It checks to see what has changed in the repository and shows you what files have been changed.19:59
EvanR-work git status19:59
do you mean what has changed on the remote repo?19:59
fr0sty chrisshattuck: changed in reference to what?19:59
chrisshattuck Right, the remote repo, or the only repo in SVN19:59
fr0sty: In reference to the local checkout (in svn)20:00
fr0sty working copy vs. remote branch?20:00
SethRobertson You can `git fetch` and then `git log ^HEAD origin`20:00
to see what the remote changes are20:00
← hyperair left20:00
→ Alagar joined20:00
← dreiss left20:01
← up_the_irons left20:01
chrisshattuck SethRobertson: Would git fetch update the local repository?20:01
EvanR-work yes but not your working tree20:01
← comptekki left20:01
EvanR-work or your local branches20:01
fr0sty chrisshattuck: 'git fetch remote' will update the 'remote tracking branches' but you never work on those.20:01
chrisshattuck Ah, okay, interesting.20:02
Still wrapping my mind around git awesomeness20:02
fr0sty It is not possible to 'diff' against a remote repository, only your fetched state of that remote repository.20:02
← tjholowaychuk left20:02
fr0sty and if you didn't do git remote update or git fetch <remote> you will be diffing against stale state.20:02
→ Eridius joined20:03
→ muep joined20:03
chrisshattuck Cool, good to know, thanks guys!20:04
→ tjholowaychuk joined20:04
← imMute left20:05
← copelco left20:05
← SethRobertson left20:05
→ SethRobertson joined20:05
→ guardian joined20:08
→ up_the_irons joined20:08
EvanR-work svn and cvs seem so bizarre from what i heard20:09
→ dr_win joined20:09
EvanR-work would be weird to use and learn them i think20:09
a branch is apparently another copy of the repo20:10
SethRobertson That is svn for you.20:10
← guardian left20:11
← planet8 left20:12
← ntoll left20:12
← ajpiano left20:12
← rayners left20:13
← Adys left20:13
← _frijole left20:14
← Alagar left20:15
← snearch left20:15
← dna left20:16
← drubies left20:16
← CoinOpeBoy left20:16
→ tbsaunde joined20:16
← TravisCarden left20:17
tbsaunde hey, quick question with git 1.5 on rhel how do I convinnce git clone to use the existing foo/ instead of failing since it already exists?20:17
→ imMute joined20:17
← imMute left20:17
→ imMute joined20:17
→ bousquet joined20:17
EvanR-work the man page says you cant20:18
→ Adys joined20:18
EvanR-work delete or move foo20:18
→ dreiss joined20:18
tbsaunde EvanR-work: hmm, I don't have permissions for that :(20:18
armenb_ b20:18
tbsaunde EvanR-work: missed that piece of the man page20:18
← sonnym left20:18
EvanR-work tbsaunde: it lists no 'force' option20:19
and you have permission to write to the dir?20:19
tbsaunde EvanR-work: true, did that get changed?20:19
EvanR-work what?20:20
tbsaunde EvanR-work: I have permision to write in foo/ but not the dir foo is in20:20
EvanR-work get whoever has permission to do this20:20
tbsaunde EvanR-work: as in is there still no force option in current 1.7.2 instead of 1.5.520:20
fr0sty tbsaunde: don't clone. in foo: git init; git remote add origin <url>; git fetch origin;20:20
EvanR-work you do not have proper security clearance to manage the repo :)20:20
tbsaunde EvanR-work: I realized there is a better way anyway20:21
EvanR-work i have 1.7.2, no force20:21
fr0sty tbsaunde: Alternatively: git clone <url> (in another directory) and move .git/ into foo.20:21
EvanR-work do what fr0sty said20:21
← hohoho left20:21
→ Tommy[D] joined20:21
tbsaunde EvanR-work: I've always sort of wished there was a use foo/ option20:21
fr0sty: thanks!20:22
fr0sty np20:22
→ apsysoev joined20:22
EvanR-work tbsaunde: there is a 'use dir'20:22
tbsaunde EvanR-work: as in if I'm board some time I may consider making a patch, I can't see a reason its a bad option20:22
EvanR-work git clone foo bar20:22
tbsaunde EvanR-work: yeah, but atleast with 1.5 if you use foo . you get already exists20:23
← magcius left20:23
EvanR-work . ??20:23
that sounds insane20:23
normally when you clone, you are not already in the target dir20:23
fr0sty EvanR-work: if you are cloning into a non-empty directory a failed clone will leave a mess that is hard to clean up20:23
and it might clobber existing files as well...20:24
EvanR-work makes sense20:24
← dmlloyd_road left20:25
tbsaunde fr0sty: sure, and rm can to the point isn't it should be a default, but from time to time it would be useful20:25
← robgleeson left20:26
→ urkud_ joined20:28
EvanR-work get permission to mess with important stuff20:29
fr0sty maybe you could clone into a blank directory...20:29
EvanR-work there should be support for people with half-way permission20:29
→ tstclair joined20:29
→ lightcap joined20:29
EvanR-work fr0sty: right, just he cant put it where he wants20:29
s/should be/shouldnt be/20:29
← tty1 left20:30
tbsaunde fr0sty: EvanR-work yeah, there was a better place for it in this case, but imho a --force-use-the-existing-dir would be nice20:31
→ thiago_home joined20:31
EvanR-work which does what20:31
removes the contents first20:31
tbsaunde EvanR-work: if foo/ exists git clone clones into it20:32
EvanR-work erasing everything that was there?20:32
clone by default checks out a working tree20:32
tbsaunde EvanR-work: I gues you it might erase certian things, so you might need two options one for overwrite and a default of don't over write20:33
EvanR-work: true, the checkout thing does make the attempt to keep stuff hard but there's no reason not to support it for empty dirs is there?20:33
→ CoinOpeBoy joined20:33
EvanR-work if its empty anyway, the feature you want is accomplished by prepending rmdir &&20:34
← aresnick left20:34
← ljungk left20:35
EvanR-work combining simpler solutions into a big one dynamically rather than building the same feature into everything is a nice philsoophy20:35
→ ljungk joined20:35
← nkuttler left20:36
tbsaunde EvanR-work: fair point though I think you need rm -r dir even if its empty20:36
EvanR-work no, thats the point of rmdir20:36
→ jord joined20:36
EvanR-work to specifically delete a dir only if its empty20:36
← krosenvold left20:38
→ krosenvold joined20:38
← g0bl1n left20:38
tbsaunde EvanR-work: oops, I saw rm dir not rmdir :p20:38
EvanR-work: any fair enough , thanks20:39
← tbsaunde left20:39
← rpg left20:39
→ kenneth_reitz joined20:41
← TeckniX left20:42
← jord left20:42
← kenneth_reitz left20:43
→ kenneth_reitz joined20:43
→ yota joined20:43
→ Schmallon joined20:44
→ cbrown_ joined20:44
→ ProperN[out] joined20:45
← ProperN[out] left20:46
→ ProperN[out] joined20:46
→ rayners joined20:46
← eletuchy left20:47
→ aresnick joined20:48
← ProperNoun left20:48
ProperN[out] → ProperNoun20:48
← RSchulz2 left20:49
← jonesy_ left20:50
cbrown_ Hi all, am having a problem with a repo where 'git fsck' as well as other cmds report 'fatal: index file smaller than expected'20:50
← khmarbaise left20:51
patrikf cbrown_: try rm .git/index && git reset (will not modify your working tree)20:53
cbrown_ equivalent to git reser --mixed?20:54
patrikf yes, --mixed is the default20:55
← jaql left20:55
→ Slix`` joined20:55
cbrown_ thks. any idea WHY it could get into that bad state?20:55
→ gnufied joined20:55
← sis left20:56
→ RSchulz2 joined20:56
patrikf git bug, filesystem corruption...20:57
IslandUsurper gremlins...20:57
cbrown_ evil spirits in the clutch housing20:57
← jstemmer left20:59
cbrown_ that fixed 'er thks patrikf21:00
← RSchulz2 left21:00
← kogent left21:03
→ Quevin joined21:04
→ RSchulz2 joined21:04
← aresnick left21:05
← dreiss left21:05
→ reflectivedev joined21:06
← IslandUsurper left21:06
← aah left21:07
→ disturb joined21:08
← masonkramer left21:09
disturb hi folks21:09
→ magcius joined21:09
→ masonkramer joined21:09
← ch077179 left21:09
disturb I'm experimenting with subtree merge strategy and can't find a way to trip the original repo's history from the merged one21:10
... or rather the other way around21:10
mischosch hi there,21:11
I try to change my svn setting to git and got the most things up running. But one thing I miss and didnt understand how to handle with it are my svn:external definitions.21:11
1. I used svn:externals to include other frameworks or libraries in a vendors folder. I got these cases covered with the git submodules, no problem.21:11
2. But I used svn:externals to fill up a working copy with checkout specific files. Imagine a web project with public folder, and subfolders for css and js files. Inside these folders there was a "projects" folder with svn:externals for this specific project. All this project specific files are stored in a separate repository with complex structure for storing all the project related files, like:21:11
project121:11
- js21:11
- css21:11
- img21:11
project221:11
- js21:11
- css21:11
- img21:11
So, while getting into git I didn't find any solution, that would help me to get my setting working nice with git. Submodules do not work for this case, it would need many different reps to build submodules from, because i cannot fetch any specific file path of some repository.21:11
Any ideas, how to find a solution, for keeping main project AND project related files under version control with git? (I thought of symlinks and git ignore definitions perhaps, but perhaps there is a better way)?21:11
← gnufied left21:12
→ aresnick joined21:13
→ Guest888855 joined21:15
← vinisterx left21:15
→ vinisterx joined21:16
→ jrmuizel joined21:17
→ sonnym joined21:19
→ ericdc1 joined21:20
← ericdc left21:20
→ dreiss joined21:21
← JoseLobato left21:22
→ aspotashev joined21:23
← mquin left21:24
→ mquin joined21:24
→ lru joined21:26
lru is it possible to override a git command with a local version during development? Say for example, I want 'git svn' to use a local git-svn script.21:26
EvanR-work how about shell alias21:27
fr0sty set PATH appropriately?21:27
export PATH=$HOME/testbin/:$PATH or something21:27
→ MikhX joined21:28
← pigdude_ left21:28
← LiamH left21:29
→ _Vi joined21:30
fr0sty mischosch: make project1, project2, etc. into their own repositories and include them as submodules?21:30
→ josefrichter_ joined21:30
SethRobertson mischosch, I have something which might help21:31
lru fr0sty: I thought that git always looks in its own git-core directory21:31
fr0sty disturb: man git filter-branch21:31
mischosch yes, that would be possible, but project structure was public->js->projects->project1->js files / public->js->projects->project2->css files21:32
SethRobertson If you have a superproject with a bunch of slave repositories under them where you do active development on all slaves and manage all slave repositiroies in the same way (branch, checkout, tag all at the same time), I have a package which will let you do this21:32
I'm not quite sure if that is you case or not21:32
fr0sty lru: what version of git are you using?21:32
mischosch (public->css->projects->project2->css files sorry c/p fail)21:33
← mefesto left21:33
lru fr0sty: 1.7.2.321:33
fr0sty I'm not aware of a git-core directory21:33
→ _frijole joined21:34
lru that's ok... I think I found a workaround21:34
SethRobertson mischosch: is js css and image repositories of their own or small bits from a larger repository? Could they be made their own repositories?21:34
lru thanks21:34
mischosch SethRobertson: slave repositories is new to me, is it a special feature of git or like a plugin you wrote?21:35
josefrichter_ guys, I need to "undo" my last commit (I made nonsene commit coz I was in other project's directory). so I want to return to the state right before the commit, with all the uncommited changes in index, etc. I think it is "git reset --soft" but want to double check its correct...21:35
SethRobertson A wrapper that I wrote.21:35
EvanR-work should images be in their own repo?21:35
fr0sty no problem, I guess... :-)21:35
EvanR-work a submodule maybe?21:35
SethRobertson http://gitslave.sourceforge.net21:35
rubydiamond → zz_rubydiamond21:35
SethRobertson josefrichter, did you push yet?21:35
mischosch no, css/js/img files of different projects are al part/bits of one huge repository21:35
SethRobertson The only thing I can think of mischosch is to have the huge repo somewhere like ../huge and populate project 1 and 2 with symlinks.21:36
josefrichter_ SethRobertson: no. just made commit I didnt want to :-) nothing else21:36
EvanR-work SethRobertson: a submodule?21:37
SethRobertson git reset HEAD^21:37
EvanR-work, what are you referring to when you say a submodule21:37
← kjy112 left21:37
EvanR-work SethRobertson: git submodule21:37
→ stringoO joined21:38
EvanR-work josefrichter_: the default is --mixed, not --soft, and is better for going 'right before the commit'21:38
disturb fr0sty: I tried git filter-branch --subdirectory-filter path/to/subtree and it works but then git subtree -s merge subtree-name master rewrites all the subtree's files to the root21:38
SethRobertson That is similar to gitslave but doesn't work well if you want to branch, checkout, tag, etc all repositories at once, and develop all repositories simultaneously by one person21:38
EvanR-work SethRobertson: and a symlink is?21:38
fr0sty josefrichter_: git reset --mixed HEAD^21:39
SethRobertson Windows or Linux, EvanR-work?21:39
mischosch SethRobertson: yes, that was my idea, too, but symlinks just feel a little bit like a workaround ;)21:39
EvanR-work i know what a symlink is, question was, a symlink allows all that?21:39
josefrichter_ and whats the difference between mixed and soft, please?21:39
fr0sty josefrichter_: man git reset21:39
josefrichter_: man git-reset21:39
Gitbot josefrichter_: the git-reset manpage can be found at http://git.or.cz/man/git-reset21:39
SethRobertson EvanR-work, my gitslave git wrapper allows all of that. It is an alternative to submodules depending on the development style you need21:40
EvanR-work right both sound better than a symlink21:40
git submodule is a glorified symlink21:40
SethRobertson Neither works if you want a partial checkout of a huge project21:40
EvanR-work whats a partial checkout?21:40
mischosch ;)21:41
← stringoO left21:41
mischosch like only one folder21:41
EvanR-work checking out will not checkout any of the submodules21:41
SethRobertson Something that mischosch wants21:41
← menace left21:42
josefrichter_ fr0sty: I see the docs but not sure I understand it correctly. so if I do --mixed, then git status will show me the uncommited changes, while after --soft it won't ?21:42
EvanR-work you can specifically decide which submodules need to be synchronized with the main repo21:42
josefrichter_: their difference is in how the index is affecte21:42
both dont mess with the working tree, --hard does21:42
SethRobertson EvanR-work: right, if you are not doing active development on all modules at the same time and want to fix a particular upstream revision, submodules is what you want. If you are, gitslave is what you want.21:43
mischosch SethRobertson: I will look for Gitslave and read through the descriptions21:43
EvanR-work well submodules can be all checked out at once21:43
josefrichter_ ok, will try --mixed :-)21:43
thank you21:44
SethRobertson Look somewhere around -> http://gitslave.sourceforge.net21:44
By checkout, I mean switch branches or whatever21:44
← dotCOMmie left21:45
mischosch but as far as I understand, gitslave will not be the solution to fetch from a huge repository in different css/js folders.21:45
→ fedesilva_ joined21:45
SethRobertson No, sorry.21:45
← fedesilva_ left21:45
EvanR-work im wondering if its smart/popular to put all the images in another repo or simply not track them21:46
or a submodule21:46
mischosch seems to be the best way to reorganize my structure to be contained in one folder that can be fetched from a project specific rep21:46
→ stringoO joined21:47
mischosch images are one thing, project specfic js and css files are really good to have version controlled I think21:47
EvanR-work i consider js to be source code so its under control21:47
css files probably also, though they are dependent on images sometimes21:48
chrisshattuck What's the nice little command that shows a visual representation of branches. Something like gitx...21:48
EvanR-work gitk21:48
chrisshattuck Gah, tahnks EvanR-work21:48
EvanR-work and also, gitk --all21:48
mischosch: so not tracking images is sane?21:49
← fedesilva left21:49
EvanR-work they are contributing a lot of disk space to our repo despite not being very large by themselves21:50
josefrichter_ if I forgot to put some files to .gitignore and put them there later, do I also need to use git-rm to stop tracking them?21:50
→ Tuller joined21:50
EvanR-work if you dont want to track them, git rm --cached21:50
← ph^ left21:50
← cilly left21:51
EvanR-work warning, other people who pull will have the file deleted21:51
← psoo left21:51
mischosch EvanR-work: everybody how he wants, for webdesign img files (related to css) are really small, I dont see any problem here, but I know Designer managing psd files with version control,21:51
josefrichter_ thanks evan21:51
← stringoO left21:52
mischosch EvanR-work: js & css files are more important to have under version controll for me21:52
→ ph^ joined21:53
← ph^ left21:53
→ dotCOMmie joined21:53
← jrmuizel left21:53
← xperience left21:54
→ ph^ joined21:54
mischosch so it seems that git users are used to have many repositories ;) … thanks EvanR-work, SethRobertson & fr0sty21:55
← doener left21:55
mischosch your input was really helpful21:55
→ adymitruk joined21:57
adymitruk what is the url for the admin repo in gitolite21:57
not locally21:57
← Adaptee left21:57
← pantsman left21:58
adymitruk tried git clone git://myip/admin21:58
anyone here?21:58
→ Adaptee joined21:59
adymitruk I also tried git clone myid@gitoliteserverip:gitolite-admin.git22:00
no go22:00
← bambule left22:02
adymitruk anyone here?22:02
← adymitruk left22:02
← dr_win left22:02
← radarek left22:03
fr0sty looks up... users:667...22:03
→ adymitruk joined22:03
→ aah joined22:04
fr0sty probably: gitolite-admin/ (no git)22:04
but I don't use it, so I can't vouch for that information...22:05
→ jackdempsey joined22:05
← urbanmonk left22:06
Nugget Can you spot the IT guy? http://www.deansproperty.com.au/Home/Profiles No cheating!22:06
sitaram adymitruk: which install method did you use?22:06
adymitruk: git:// has nothing to do with gitolite -- gitolite works over ssh22:07
← jackdempsey left22:07
← j0ran left22:08
sitaram adymitruk: and it wont be "myid"; it'll likely be "git" or "gitolite": git@server:gitolite-admin22:08
← sivy left22:08
sitaram adymitruk: for easy install (from client) type, try "gitolite:gitolite-admin"22:08
→ ajpiano joined22:10
→ sivy joined22:10
→ fedesilva joined22:12
adymitruk I did it using the client install22:12
I did have a local clone on one machine of the gitolite-admin repo22:12
← rayners left22:13
adymitruk I tried to clone the admin repo on another connection (same user and key)22:13
but I have no idea what the url is on a different machine22:13
do I just reinstall?22:13
→ j0ran joined22:14
→ sh1mmer joined22:14
← shennyg left22:15
← aziz left22:15
← fedesilva left22:15
← mmattice left22:16
← bhah left22:17
← dquestions left22:17
← ptor left22:17
← nate_h left22:17
← mshaw left22:17
→ mshaw joined22:18
← yota left22:18
→ mmattice joined22:18
→ dquestions joined22:18
→ shennyg joined22:18
→ nate_h joined22:19
← engled left22:19
→ ptor joined22:19
← zombor left22:24
← jotik left22:24
← Cap_J_L_Picard left22:25
← sh1mmer left22:26
→ sh1m joined22:26
→ zombor joined22:28
← Guest888855 left22:28
← tizbac left22:28
← ehabkost left22:29
← SRabbelier left22:30
← PerlJam left22:30
← murr4y left22:30
← parasti left22:30
→ jotik joined22:30
→ murr4y joined22:30
→ PerlJam joined22:31
adymitruk gl-easy-install fails on the cloning of the admin repo22:32
I'm using msysgit and hitting an ubuntu server22:32
← ph^ left22:33
← zombor left22:33
← nevyn left22:34
← sh1m left22:34
→ sh1mmer joined22:35
→ nevyn joined22:35
← sh1mmer left22:36
← Eridius left22:36
← _Vi left22:36
← Tommy[D] left22:36
← gbacon left22:36
← askhader left22:36
← LeLutin left22:36
→ LeLutin joined22:37
← PerlJam left22:37
← Exodist left22:37
← sitaram left22:37
← Reticenti left22:37
← agile left22:37
→ exodist_ joined22:37
→ zombor joined22:37
→ sitaram joined22:37
← sitaram left22:37
→ sitaram joined22:37
→ PerlJam joined22:37
→ Reticenti joined22:38
→ Eridius joined22:38
→ askhader joined22:38
→ gbacon joined22:38
→ Tommy[D] joined22:38
→ agile joined22:38
→ _Vi joined22:38
← zombor left22:39
← nairb774 left22:39
→ dna joined22:40
→ fedesilva joined22:43
← irc2samus left22:44
← ymasory left22:44
← mbroeker left22:45
← opalka left22:48
→ hipe joined22:52
← hipe left22:53
← Yuuhi left22:53
← cannonball left22:55
← frakturfreak left22:55
→ tjubaluba joined22:56
→ etank joined22:56
→ comptekki joined22:56
→ frakturfreak joined22:57
comptekki how come when i run this on snow leopard i get the error trying git bash auto completion22:57
$ . git-completion.bash22:57
-bash: git-completion.bash: line 123: syntax error near unexpected token `done'22:57
-bash: git-completion.bash: line 123: ` done < <(git config -z --get-regexp '^(svn-remote\..*\.url|bash\.showupstream)$' 2>/dev/null | tr '\0\n' '\n ')'22:57
what is this error from22:57
snow leo have an old bash?22:57
→ scrts joined22:58
thiago_home are you sure you're on bash?22:59
comptekki SHELL=/bin/bash23:00
thiago_home that's a weird syntax, though23:00
< <(.....)23:00
comptekki i found a site that talked that it was correct but i didn't see a fix23:01
← skyl left23:01
→ ericflo joined23:02
comptekki ah - gotta catch a bus thx...23:02
← comptekki left23:02
ericflo Dumb question: I have the sha for a merge. How do I get a diff of the changes that were made in that merge?23:03
← joeytwiddle left23:04
thiago_home relative to which parent?23:04
ericflo The...first one?23:05
← tjubaluba left23:05
thiago_home the first parent is $SHA^ or $SHA~ (same thing)23:05
ericflo In concrete terms, I merged a topic branch into master, and I'd like to see the equivalent diff23:05
thiago_home so you want: git diff $SHA~ $SHA23:06
→ madsy joined23:06
→ madewokherd joined23:06
thiago_home or git diff $SHA~..$SHA23:06
or the shorthand: git diff $SHA^!23:06
ericflo thiago_home: Nice, I'll try that.23:06
thiago_home: That worked great, thanks!23:07
thiago_home you can get the other diff by: git diff $SHA^2..$SHA23:07
→ psoo joined23:09
← malumalu left23:09
→ zombor joined23:09
← zombor left23:10
→ robgleeson joined23:11
→ zombor joined23:11
← Zenopus left23:11
→ Zenopus joined23:11
← madeofcode left23:12
→ madeofcode joined23:12
← lightcap left23:12
← zombor left23:13
← nakkor left23:13
eseidelAFK → eseidel23:15
eseidel → eseidel223:15
→ Nickeeh joined23:16
→ fedesilva_ joined23:17
← aspotashev left23:17
← kenneth_reitz left23:17
← fedesilva left23:19
fedesilva_ → fedesilva23:19
→ zenazn joined23:19
→ ymasory joined23:19
← psoo left23:24
→ scarabx joined23:27
→ cpf_ joined23:29
→ zombor joined23:31
→ jonesy_ joined23:31
→ felipellrocha joined23:32
→ lightcap joined23:32
→ priidu joined23:33
← Ryback_ left23:35
← zombor left23:37
→ LeLutin_ joined23:37
← LeLutin left23:40
← nadavoid left23:42
← marc_kdab_ left23:42
→ kanru joined23:42
← chrisshattuck left23:46
← bousquet left23:47
← hakunin left23:47
← cwillu_at_work left23:48
← MikhX left23:48
→ MikhX joined23:49
← sonnym left23:49
→ cwillu_at_work joined23:49
→ hohoho joined23:49
→ EvanR joined23:51
← Dave^| left23:51
→ sonnym joined23:52
← Adaptee left23:54
← MikhX left23:54
→ flowbee joined23:55
flowbee setting up top level get repo. any point to having the tag/branch/trunk mentality we had with subversion23:55
or should i just put my app code in root git repo?23:55
← ericflo left23:55
→ michaeldiamond joined23:55
← JEEB left23:56
EvanR flowbee: all repos should probably be identical as far as the main branches go23:58
Eridius flowbee: tag/branch/trunk in svn is a hack to work around the lack of first-class support for tags/branches23:58
→ zombor joined23:59
← priidu left23:59
EvanR the concept of 'root git repo' is largely arbitrary23:59
→ eletuchy joined23:59
EvanR its probably the one with the least branches23:59
master devel bugfixes etc23:59

Logs Search ←Prev date Next date→ Channels Documentation