IRCloggy #git 2010-06-26

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.

2010-06-26

← timyhac left00:00
← Paraselene_ left00:04
← vincent left00:06
→ chuck joined00:07
← corni left00:09
→ Paraselene_ joined00:10
ThiefMaster how would i create a branch which is not based on the current HEAD but a certain commit and contains all the commits following the base?00:11
← SlickT10 left00:12
ThiefMaster i.e. i basically want to undo 'git branch -d somemergedbranch'00:12
← andresj left00:12
← carrus85 left00:12
→ eletuchy joined00:17
→ jlebar joined00:19
jlebar Can I rebase a branch on top of another branch? Does that even make sense?00:19
← justin-george left00:19
ToxicFrog ThiefMaster: 'git branch somemergedbranch startpoint'00:19
← Alagar left00:21
← adamv left00:21
ThiefMaster ToxicFrog: already got that far, how to merge the following commits without git considering them a merge?00:21
← hakunin left00:23
← engla|swe left00:24
← ignatp left00:24
← dreiss left00:27
→ Elfix joined00:31
→ consolers joined00:33
← mjf left00:35
ThiefMaster nvm, rebase did the job00:38
← beatak left00:40
→ chrislerum joined00:41
← mastro left00:41
→ mastro joined00:45
← jlebar left00:46
→ jrmuizel joined00:50
→ robinbowes joined00:52
← mastro left00:53
← lispy left00:53
→ MicahElliott joined00:54
← Adman65 left00:55
← jrmuizel left00:56
→ Textmode joined00:57
→ mastro joined01:05
← mw____ left01:08
sjuxax I'm trying to do a subtree merge here according to http://www.kernel.org/pub/software/scm/git/docs/howto/using-merge-subtree.html. I'm on git read-tree; every time I try to execute the file, if the file already exists, I get an error that says untracked file would be overwritten by merge, and if the file doesn't exist, I get an error that says invalid path. What am I doing wrong?01:08
→ sh1m joined01:09
→ rayners joined01:11
← sh1mmer left01:12
sh1m → sh1mmer01:12
→ drdr joined01:14
→ Vampire0_ joined01:14
← chrislerum left01:15
→ fractal_heart joined01:15
← mastro left01:17
→ thomas_b joined01:17
→ redondos joined01:18
→ mastro joined01:18
→ chrislerum joined01:22
→ _VVSiz_ joined01:27
← name left01:30
→ skmidry_ joined01:30
← airborn left01:33
← Joeconyers left01:34
← Tuller left01:39
→ skmidry joined01:44
← skmidry_ left01:44
← eletuchy left01:46
→ yaotti_aw joined01:48
← skmidry left01:49
← yaotti_aw left01:50
← ChUrP1 left01:50
← drdr left01:51
→ jtibu joined01:56
← aliceinwire left01:57
← sjuxax left01:57
→ Elfe_ joined01:58
→ skmidry joined02:01
→ alsuren joined02:04
← jtibu left02:05
← MicahElliott left02:08
→ yaotti_aw joined02:12
← yaotti_aw left02:12
→ yaotti_aw joined02:12
← yaotti_aw left02:13
→ yaotti_aw joined02:13
← yaotti_aw left02:19
→ cytrinox_ joined02:20
→ yaotti_aw joined02:21
← yaotti_aw left02:21
→ yaotti_aw joined02:21
yaotti_aw → yaotti02:22
byang hi02:22
how can I add a new file part by part to the index?02:23
seem add -p <new file> does not work, it report a 'No change' ...02:23
cytrinox_ → cytrinox02:24
gebi git gui (to get an overview about your changes)02:24
→ nairb774 joined02:25
← alsuren left02:25
byang Now, I just add a new empty file to the index first and then add -p the real file. :)02:26
but the way is really ugly02:26
gebi: git gui?02:26
→ jrmuizel joined02:27
← zenazn left02:28
→ Elfe joined02:28
gebi seems strange that add -p doesn't pick up changes, git gui shows you the whole situation about your changes02:28
DrNick new files only have one hunk02:31
← Elfe_ left02:31
DrNick so interactively choosing whether or not you want to add that hunk is a bit stupid02:31
byang DrNick: yes, new file only has one trunk, but I shold add it splittly. I think that is what the 'e' option in -p does.02:33
← chrislerum left02:35
← hobodave left02:36
← yaotti left02:36
→ yaotti_aw joined02:37
yaotti_aw → yaotti02:37
← fractal_heart left02:39
← jrmuizel left02:39
← madewokherd left02:40
→ zenazn joined02:41
→ maronnax joined02:42
→ madewokherd joined02:44
→ Philwelch joined02:49
← spearce left02:51
→ timj joined02:52
← zenazn left02:52
→ MicahElliott joined02:54
← timj left02:57
→ candybar joined03:02
← masterkorp left03:02
← maronnax left03:05
→ hakunin joined03:05
→ timj joined03:09
→ maronnax joined03:09
→ Bass10 joined03:10
→ justin-george joined03:10
→ OFICINA7 joined03:12
← justin-george left03:15
→ masterkorp joined03:21
→ jrmuizel joined03:23
→ coderdad joined03:24
→ ige joined03:27
→ krawek joined03:29
← Eaven left03:34
← krawek left03:34
→ krawek joined03:35
← krawek left03:38
→ krawek joined03:39
yaotti → yaotti_aw03:39
→ harinath joined03:41
← Bass10 left03:43
← coderdad left03:58
→ kpreid_ joined04:01
← consolers left04:02
→ anathematic joined04:03
← krawek left04:04
→ sagsousuke joined04:08
← zimbres left04:11
→ zimbres joined04:12
→ ignatp joined04:12
→ nareshov joined04:15
← scarabx left04:17
→ krawek joined04:18
→ dreiss joined04:18
← zimbres left04:21
→ phantomcircuit joined04:34
← anathematic left04:37
← jrmuizel left04:39
→ markabur joined04:42
← dreiss left04:47
← harinath left04:47
→ skmidry_ joined04:48
← nareshov left04:49
← skmidry left04:50
← MicahElliott left04:50
→ skmidry__ joined04:51
← skmidry_ left04:53
→ jrmuizel joined04:54
maronnax I have repositories for libraryFoo and for programBar and want to include libraryFoo in programBar, particularly as its own self-contained branch of the original repo, however I can't seem to figure this out. Is there a best way to do this?04:58
→ harinath joined05:01
← jrmuizel left05:02
→ nareshov joined05:04
← phantomcircuit left05:06
→ jrmuizel joined05:06
← jrmuizel left05:06
→ kar8nga joined05:11
selckin sounds like a bad idea05:17
→ phantomcircuit joined05:18
→ VK7HSE joined05:20
← sagsousuke left05:21
← ignatp left05:24
→ lucid joined05:25
← kar8nga left05:30
→ MicahElliott joined05:34
→ orafu joined05:42
→ disappearedng joined05:46
→ keyvan joined05:47
→ marcial5 joined05:57
→ mindworx joined05:59
← mindworx left06:00
→ SplinterOfChaos joined06:00
SplinterOfChaos Is there a way to completely replace one branch with another?06:00
→ lhz joined06:00
selckin just delete it?06:00
mmattice rename them both?06:01
SplinterOfChaos I need to replace the master branch. Can i really delete or rename that?06:01
selckin sure06:01
→ mithro joined06:01
AAA_awright Delete the source, then rename, and push... I think that will still cause problems with pulling it's functionally the same as a rebase06:02
er, delete the target, if you can't rename over it06:02
hmm I forget how I did it now06:02
marcial5 Hello to all! I was using github as my remote repo and now i want to use a private account with unfuddle. what i would like to do is migrate my existing repo there instead of just committing the last state of my repo(wich is what i've just done). Can i change the settings and the key of my repo and make it point to unfudddle so i can push the entire repo? Thanks in advance!06:03
SplinterOfChaos Wow, it worked!06:03
← programble left06:03
selckin marcial5: just add it as a remote, and push your branches to it06:04
marcial5 and the keys?06:04
thy've changed06:04
selckin why do they mather,06:04
marcial5 i'll tell you later if they do (not sure at all)06:05
thnks06:05
So i just clone like this lines...06:06
[remote "origin"] url = [email@hidden.address] fetch = +refs/heads/*:refs/remotes/origin/*06:07
to make more like [remote "my new remoto"] url = [email@hidden.address]06:07
and what about .... ? fetch = +refs/heads/*:refs/remotes/origin/* do i leave it so?06:08
← adamm left06:09
→ icefox joined06:15
← madewokherd left06:18
marcial5 it worked!! didn't need to be afraid! smooth as silk!06:21
now... if i want to make the origin this new repo so i can forget the 'old' one at github?06:22
← Philwelch left06:22
selckin i would just make a clone of the new one to be sure its all good (i'm paranodi that way)06:23
marcial5 gitk only shows the refs at remotes/origin and those are the old ones!!06:23
that'll make06:23
← icefox left06:26
→ coppro joined06:27
→ Titosemi joined06:27
marcial5 what i mean is that although i've just pushed the entire master branch to unfuddle, on the .git/refs/remotes dir, it only has a folder with origin, and not (as i would expect) another one to unfuddle. mmmm...suspicious06:27
selckin git remote add fuddle git+ssh://bla && git push fuddle master06:29
should do it probably06:29
marcial5 thats what i just done! i added manually the remote by editing06:31
i did push fiddle06:31
and everything is there06:31
→ segher joined06:31
← phantomcircuit left06:31
marcial5 but no ref to ufeddleng on local .git/refs/remotes dir06:31
selckin then just make a new clone from fiddle imo06:31
marcial5 oooooook i think that will do06:32
thnks06:32
→ DavidKlein joined06:32
← DavidKlein left06:32
→ smorg joined06:32
marcial5 lets leave the hacking to hackers06:32
← masterkorp left06:33
← kenneth_reitz left06:38
→ drizzd joined06:41
→ tvw joined06:42
→ icefox joined06:43
→ Vadtec joined06:45
← icefox left06:45
Vadtec im trying to get the cia.py script to run on my remote repo (bare), but it doesnt appear that post-commit is getting called, i am running git-daemon in verbose mode, and connecting via ssh:// and git:// as well, how can i see if post-commit is being called or not?06:47
or, how can i see any output from post-commit?06:47
the cia.py for cia.vc that is06:47
→ frank_ joined06:51
← bambule left06:51
drizzd Vadtec: is it executable?06:51
← mithro left06:52
→ hobodave joined06:52
Vadtec drizzd: yes, everything is +x06:52
drizzd umh, post-commit only gets called locally, of course06:54
Vadtec drizzd: ok, does post-receive get called on the remote repo? (I assume yes)06:55
→ Jaxan joined06:55
drizzd Vadtec: yes, but the ciabot.py documentation suggests using the "update" hook06:55
are you using the one in contrib/ciabot?06:56
→ disappearedng_ joined06:56
→ poseid joined06:56
Vadtec drizzd: http://git.kernel.org/?p=git/git.git;a=blob_plain;f=contrib/ciabot/ciabot.py;hb=refs/heads/master06:56
thats the one im using06:56
drizzd ok, that's the one I'm looking at too06:57
Vadtec OH06:57
ok, i see what i read wrong06:57
it says either post-commit *or* post-update06:58
→ doener joined06:58
Vadtec i didnt realize it matters on remote repos06:58
im migrating from svn to git, so....yeah, that06:58
drizzd it doesn't say post-update06:59
Vadtec # This script is meant to be run either in a post-commit hook or in an06:59
# update hook.06:59
language is unclear06:59
drizzd but I'm not sure why it would need the update hook06:59
Vadtec you just said post-commit isnt called on remote repos07:00
drizzd you're right, that's probably suggesting that it's ok to run it in the post-update hook07:00
Vadtec: there is the "update" and the "post-update" hook07:00
← disappearedng left07:00
Vadtec which is better, post-update?07:00
← keyvan left07:01
drizzd Vadtec: feels more intuitive to me, since you only need to run it after you successfully pushed07:01
← Jaxan left07:01
Vadtec is giving it a try as we speak07:01
drizzd that would also be more in line with the behavior of post-commit locally07:02
Vadtec well, i just got some output...let me fix a typo and see what i get07:04
← poseid left07:04
→ poseid joined07:06
← poseid left07:07
Vadtec drizzd: many thanks, that fixed the issue07:07
now to get rid of that tiny url stuff...07:08
→ project2501a joined07:08
← disappearedng_ left07:09
Vadtec drizzd: thanks again07:15
← Vadtec left07:15
← OFICINA7 left07:15
→ timyhac joined07:21
← ige left07:21
yaotti_aw → yaotti07:21
→ gmarselis__ joined07:22
→ disappearedng_ joined07:23
← MicahElliott left07:24
→ thiago_home joined07:26
← project2501a left07:26
→ sjohnson joined07:27
← VK7HSE left07:29
→ giskard joined07:29
→ masterkorp joined07:31
→ rox` joined07:32
→ VK7HSE joined07:32
← maronnax left07:34
← giskard left07:36
→ radioman-lt joined07:37
→ raichoo joined07:38
→ bodom joined07:39
bodom Hi! I have multiple projects and each of them got it's git repository. Now I want a folder (i.e. /lib) to be shared between some repositories. Is there a way to do that in git?07:41
← rayners left07:42
← krawek left07:43
→ aliceinwire joined07:45
cbreak bodom: maybe submodules do what you want07:46
but they have their limitations07:46
bodom: man git-submodules07:46
or was it man git-submodule?07:46
Gitbot cbreak: the git-submodule manpage can be found at http://git.or.cz/man/git-submodule07:46
bodom cbreak: ty07:47
→ priidu joined07:47
cbreak just a warning: submodules are evil07:47
strong binding07:47
so no automatic updates07:47
→ Ratler joined07:49
mmattice which can be good07:49
← Ratler left07:49
→ Ratler joined07:50
DrNick I don't understand why you'd want it any other way07:51
← gmarselis__ left07:51
→ icwiener joined07:51
DrNick if the projects are so strongly coupled that one has to be contained within the other, then why wouldn't you want a specific version pinned07:52
→ jayne joined07:52
coppro if they are maintained in parallel07:53
drizzd DrNick: yes, but even if a new version is added, your submodule will not get updated. It's always like having a dirty working copy.07:53
SplinterOfChaos DrNick: I dunno, updates can mean optimizations, design changes, etc. If i change the interface to a library i write, i want to ensure i make this change in any file using that library.07:53
Then again, is it really /that/ hard to just update the submodules before working?07:54
DrNick drizzd: well, yeah. you'd need to explicitly update the submodule, and then run your regression test suite in the containing repository to make sure it still works07:54
drizzd DrNick: yes, but I already told git that I want a new version when I did pull/checkout07:54
but now I have to remember to run submodule update every time07:54
SplinterOfChaos drizzd: Write a script?07:55
→ werdan7 joined07:55
drizzd SplinterOfChaos: how is a script going to be any help?07:55
→ amitprakash joined07:55
amitprakash how do i look at what changed in the last git commit07:56
SplinterOfChaos drizzd: You won't have to remember.Write a generic script to start working, have it include updating the submodules.07:56
drizzd I want to use git, not a wrapper for each git command07:56
I don't use git once a day and then work, I use git all the time07:57
→ unreal joined07:57
SplinterOfChaos The script would only need be run once, not every time you use git.07:57
drizzd it has to be run every time I use rebase, merge, pull, checkout, or any other command that can change the submodule version07:58
coppro could you not write hooks to do that?07:58
drizzd amitprakash: git log -p -108:01
amitprakash drizzd, yeah thanks .. figured it out :D08:01
→ project2501a joined08:01
→ gmarselis__ joined08:03
← project2501a left08:07
cbreak amitprakash: man git-show, man git-log08:07
Gitbot amitprakash: the git-show manpage can be found at http://git.or.cz/man/git-show08:07
→ mithro joined08:07
→ alsuren joined08:08
→ project2501a joined08:08
← gmarselis__ left08:09
→ gmarselis__ joined08:09
→ issyl0 joined08:10
issyl0 Hello. I'm trying to get gitweb working but, even following the install instructions it doesn't seem to want to work. It seems like a great thing to use compared to just a big repository full of files, especially as it's web based with commit history and everything. Can anyone help? Running Linux, Debian on my server, Ubuntu on my laptop.08:12
→ pallaert joined08:13
← project2501a left08:13
← gmarselis__ left08:14
← issyl0 left08:17
→ consolers joined08:17
← henkboom left08:20
→ henkboom joined08:21
→ psoo joined08:24
→ pantsman joined08:27
← unreal left08:27
→ unreal joined08:29
← mithro left08:31
→ MarkBao joined08:36
→ darwin_ joined08:37
→ fmardini joined08:37
coppro → Cu08:37
← consolers left08:44
→ eletuchy joined08:45
→ Acorn joined08:46
← aliceinwire left08:46
→ project2501a joined08:52
→ oriba joined08:53
→ g3d joined08:55
→ resmo joined08:57
→ Ambiguity joined08:58
→ aliceinwire joined08:58
← gebi left08:58
← icwiener left08:58
← harinath left08:59
← markabur left09:00
→ alip joined09:03
→ oriba_ joined09:05
→ rgr joined09:08
← oriba left09:08
→ harinath joined09:15
→ bababfds joined09:17
→ fr0sty joined09:19
← harinath left09:20
→ simplechat joined09:25
← hobodave left09:26
← darwin_ left09:26
→ darwin_ joined09:28
→ Falc joined09:30
→ engla|swe joined09:30
→ ziro` joined09:31
→ elijahbal joined09:33
elijahbal hello09:33
I would like to change the initial commit in a git repository.09:34
I have a source tree already modified from the original, and I would like to put the original as the first commit.09:34
Do you know how to do that ?09:34
← eletuchy left09:34
elijahbal Thank you very much09:34
→ harinath joined09:35
← henkboom left09:35
selckin rebase -i ?09:36
elijahbal thanks09:37
→ aziz joined09:37
doener won't work for root commits09:37
selckin :-(09:38
→ henkboom joined09:38
doener elijahbal: I didn't quite understand what you want to do... You modified something and now you want the pre-modification state to become a root commit? Did you previously commit that "original" state?09:39
→ AlexP joined09:40
elijahbal no09:40
I initiated the git repository with a modified source tree.09:41
doener Well, you'll need some way to get hold of the original state, git can't create that out of thin air09:41
elijahbal Ok09:41
np09:41
doener maybe you have a tarball of it?09:41
→ [Po]lentino joined09:42
→ squi joined09:44
squi i heard before that git clone is not a good idea? how should i do an initial checkout, if not using clone?09:44
jast squi: whoever told you that sounds a bit confused. clone is meant to be the main tool for that.09:44
squi ok09:45
maybe i misunderstood or something09:45
now09:45
how can i tell git not to change line endings?09:46
jast hard to tell in retrospect :)09:46
squi or beter yet09:46
how can i tell it to completely and utterly ignore whitespace changes? i am so sick of getting so many conflicts due to that09:46
selckin git won't change line endings unless you told it too (like default broken windows installer do)09:46
squi ok09:46
selckin that's a diffrend problem09:46
→ d0k joined09:46
squi i mean09:46
i jsut did a fresh clone09:47
and i already got hundreds of modifications09:47
i dont like that.09:47
jast are you using msysgit on windows?09:47
squi yes09:47
jast git config --global --unset core.autocrlf should do the trick09:47
note that if the repository uses LF line endings, you'll probably want to use an editor that uses those, too09:47
→ apsysoev joined09:48
jast or, alternatively, enable safecrlf along with autcrlf09:48
that tends to make the conversion much less intrusive09:48
squi jakob i dont even know where the problem is... we all use the same windows IDE (vs 2010)09:48
jast in general, if directly after a clone everything is listed as modified, it means that CRLF endings are used in the repository09:48
and autocrlf is meant for repositories in which LF is used09:49
squi i don't understand at all how i can have hundreds of modifications just upon checkout09:49
← alsuren left09:49
jast so that sounds like if you use the command I provided above, you'll be fine09:49
squi ok thanks ill try it09:49
i just gave up my local changes because i didnt even know what i changed anymore09:49
and it seemt too tedious to find them in hundreds of conflicts09:50
jast yeah, of course09:50
had you turned off autocrlf before giving up, everything would probably be fine now :)09:50
ah well, what's done is done09:51
squi still playing around with it09:52
ok this is odd09:52
i did the change that you recommended09:53
but i still have modified files upon fresh clone09:53
elijahbal doener: yes exactly09:53
I have a tarball of the original source tree, and I want to make it the first commit.09:53
jast squi: just to make sure that the command worked, what does "git config core.autocrlf" say?09:54
← henkboom left09:54
aliceinwire → durarara09:55
squi haha09:56
it says true09:56
ok let me try again09:56
yep09:56
your command doesn't change it :D09:56
now thats odd09:57
→ henkboom joined09:57
jast squi: how about "git config --global core.autocrlf"?09:59
doener elijahbal: ok, then simply: git add -A .; git commit -m tmp; rm -rf09:59
oops...10:00
→ nadim joined10:00
squi still true jast10:00
ah10:01
doener git add -A .; git commit -m tmp; rm -rf *; tar xf /path/to/tarball; git add -A .; git commit --amend; git read-tree -u --reset HEAD@{1}; git reset10:01
→ adymitruk joined10:01
doener elijahbal: like that ;-)10:01
squi jast simply add a "fale" then it works :)10:01
--unset probably sets it to the default value, which funnily is true10:01
→ DavidKlein joined10:01
jast hmm, --unset worked for me10:01
← DavidKlein left10:02
jast I guess they patched the default value right in the code or something10:02
doener elijahbal: assuming that you didn't commit anything yet10:02
jast but he said he already had his own history, didn't he?10:03
doener elijahbal: basically that monster makes a commit of your current state so you can restore it later, then you put in the initial state, commit that, replacing the previous commit, and restore the modified state10:03
jast: he said "initiated the repo with a modified source tree". I took that as just "git init", but yeah, you might be right there10:04
elijahbal jast: yes I have my own historx10:04
durarara → aliceinwire10:05
jast so, which convoluted solution would you like... A or B?10:05
elijahbal something like : git init . in the initial source tree.10:05
jast git init doesn't actually commit anything10:05
aliceinwire → durarara10:05
doener elijahbal: did you ever run "git commit" in that repo?10:05
→ parasti joined10:06
→ oriba__ joined10:06
elijahbal git init ; git add * ; git commit -a -m "first commit" ; cp -rf ../current_source_tree/* . ; git add * ; git commit -a -m "revision"10:06
it makes the trick, but is it the correct way ?10:06
squi hum10:06
jast if you had only one commit, it should be fine10:07
elijahbal No I had many commits10:07
→ kipras joined10:07
squi jast checking out again, it still has modified files and the autocrl property has also been restored10:07
elijahbal That's the problem, I want to keep the history.10:07
squi i think it was never that hard to change a config value10:07
jast that's, uh, weird...10:07
squi indeed it is10:07
jast elijahbal: well, solution A or B? they are both a bit convoluted, so you don't need to know anything about them to decide ;)10:07
squi aha10:07
i think i know what it si10:08
apparently every config value has a local and global setting and i didnt set it gloibally10:08
jast oh well10:08
squi k finally worked10:09
← oriba_ left10:09
→ chrislerum joined10:10
doener elijahbal: any merge commits in your history?10:12
→ Titosemi_ joined10:15
← harinath left10:15
squi how can i checkout a specific version of a repository?10:15
→ Chris64 joined10:16
Chris64 hi all10:16
elijahbal not doener. It's only a linear tree10:16
without branches.10:17
Chris64 is it possible to combine the usage of TortoiseGIT and git commandline tools?10:17
← Titosemi left10:17
Titosemi_ → Titosemi10:17
← oriba__ left10:18
doener elijahbal: which git version?10:18
elijahbal last version10:18
→ tizbac joined10:20
→ gebi joined10:21
doener git checkout --orphan new_history; git rm -rf .; git clean -xdf; tar xf /path/to/original.tar; git add -A .; git commit -m "Initial commit"; git rebase --onto HEAD --root master10:21
← Mage left10:22
doener that should do... but make sure you have everything committed10:22
(you might of course want to do that in a copy of the repo...)10:22
→ harinath joined10:32
→ Zephyrus joined10:32
← gebi left10:32
→ bentob0x joined10:33
← adymitruk left10:34
← squi left10:38
→ malumalu joined10:38
→ mbroeker joined10:39
→ ige joined10:39
→ nuoHep joined10:39
← frank_ left10:42
← henkboom left10:43
→ henkboom joined10:46
→ happypolla joined10:46
→ frank_ joined10:47
← unreal left10:50
→ unreal joined10:52
← VK7HSE left10:55
← raichoo left10:55
→ raichoo joined10:57
← psoo left10:58
→ HG` joined11:00
← engla|swe left11:02
← henkboom left11:03
→ henkboom joined11:04
← amitprakash left11:04
→ codeshepherd joined11:05
← HG` left11:06
← codeshepherd left11:09
→ froschi joined11:12
← raichoo left11:13
→ codeshepherd joined11:14
→ name joined11:15
→ ChUrP joined11:15
← bentob0x left11:15
→ EddieRingle joined11:17
← EddieRingle left11:18
→ keyvan joined11:21
→ locklace joined11:26
→ Tommy[D]_ joined11:29
← project2501a left11:36
← codeshepherd left11:38
← ziro` left11:38
← durarara left11:44
→ stamina joined11:44
→ DavidKlein joined11:46
→ Theaxiom joined11:48
Theaxiom I have a file that is tracked by git, it is webroot/index.php and when I add webroot/index.php to .gitignore or .git/info/exclude it still shows changes in git diff11:49
can anyone tell me what I am doing wrong?11:49
→ rkreis joined11:51
rkreis hi there11:51
is it bad when i push commits to another git repository and then use git svn dcommit? it rewrites the commits11:52
Theaxiom I figured it out11:54
git rm --cached webroot/index.php11:54
→ codeshepherd joined11:56
← happypolla left11:58
→ babafds joined11:59
→ project2501a joined11:59
← bababfds left11:59
drizzd rkreis: yes, publishing via both git and git-svn doesn't really work12:00
rkreis i see12:00
i tried something like that, and it ended up being a huge mess12:01
drizzd the only useful thing you can do is to set up a git mirror for the svn repo12:01
but you can't push to that directly12:01
rkreis so, to clarify, i can't have two git repos pulling/pushing from each other and have one of them update/commit to or from a svn repo?12:01
drizzd rkreis: that's correct12:02
← project2501a left12:02
rkreis so in that case, changes can only flow between the two git repos after they've been commited to the svn repo?12:03
→ hakunin_ joined12:03
← hakunin left12:03
drizzd svn basically forces you to use a centralized server, the svn server12:03
rkreis that's a shame12:04
thanks for clarifying12:04
so the best idea would be to only use git-svn and never pull or push between git repos in that case?12:05
→ tar- joined12:08
Chris64 why is this impossible?12:08
rkreis git svn dcommit rewrites my commits, and when i published them before, they essentially get duplicated12:08
selckin can share betwean repos till you commit to svn, then start a new branch based on svn :/12:09
rkreis i think git svn rebase makes it even worse12:09
jast git svn dcommit implicitly uses git svn rebase12:10
rkreis i see12:10
← paja left12:10
jast the problem is that if you don't linearize history, things get ambiguous with versions of svn that don't support merge tickets12:11
→ engla|swe joined12:11
jast and git-svn's support for merge tickets is limited AFAIK12:11
→ jstemmer joined12:11
rkreis i'll probably drop git-svn12:11
jast git-svn is really only useful if you only use it for communicating with the subversion repository12:12
also if you are extremely careful about trading changes in other ways12:12
rkreis can i merge svn commits into a published git repo?12:12
jast sure12:12
rkreis do you mind if i tell you about the whole scenario?12:13
jast not at all12:13
rkreis that's nice12:13
jast I might leave in a couple of minutes, but I'm sure someone else can pick up the ball12:13
rkreis we have three git repos, they push and pull changes between each other12:13
→ Jaxan joined12:14
rkreis but we also need an svn repository that's somehow kept in sync with it12:14
at the moment, one of thosse git repositories uses git-svn dcommit and rebase, but that doesn't work12:14
→ hyperboreean joined12:14
← nadim left12:15
← thiago_home left12:15
→ nadim joined12:15
rkreis my idea is pulling the svn commits into the git repositories, so i keep all the history there, and just commit the git working tree to svn every week or so12:15
does that sound horrible?12:15
→ thiago_home joined12:15
→ psil joined12:16
rkreis i wouldn't want all the git repos to use git-svn, what i really want is an svn mirror where you can also commit things that somehow find their way back into git12:16
→ dsop joined12:16
→ tghw_ joined12:16
rkreis git-svnserver would be great, but it doesn't exist?12:16
→ pielgrzym joined12:16
jast no, it doesn't12:16
currently there is work in progress for a different svn interface for git12:17
as part of google's summer of code12:17
rkreis that sounds interesting, will it allow this?12:17
jast I don't know, to be honest12:17
it's not in a state where you could actually do anything with it yet12:18
rkreis will git svn fetch and git merge import changes from svn without duplicating commits or rewriting git repos?12:18
→ corni joined12:18
← phlowx left12:18
jast one thing you could do, anyway, is to use some completely different mechanism to push the current state of things to subversion every now and then12:18
one yucky but sometimes used solution is to make one clone of the git repository a svn working copy as well, and add new files/commit to svn every now and then12:19
← navap left12:19
→ navap_ joined12:19
rkreis yes, that's what i meant by "just commit the git working tree to svn every week or so12:19
jast which would mean that the SVN repo wouldn't have the exact history as the git repo does, but the same content12:19
yeah12:19
→ Grum joined12:19
rkreis i will do just that12:19
jast if that's enough for you, at this point it's probably the easiest solution12:19
rkreis at least the whole history is in the git repos12:19
jast and there should be many ways to get a one-off patch from svn into git12:19
→ unixtippse joined12:20
← tar- left12:20
← froschi left12:20
← alip left12:20
← orafu left12:20
← Jordan_U left12:20
← stepnem left12:20
← [[sroracle]] left12:20
← Theaxiom left12:20
rkreis is git svn fetch followed by git merge sane in this case?12:20
→ LotR joined12:20
→ xou joined12:20
→ codejunky joined12:21
→ defn_ joined12:23
→ Blackguard joined12:23
jast personally I'd fetch and then cherry-pick individual commits12:24
merge will probably do insane things12:24
anyway, leaving now.12:24
→ airborn joined12:25
→ FoxFurry1 joined12:25
→ jksz_ joined12:26
→ project2501a joined12:26
→ tar- joined12:26
→ froschi joined12:26
→ alip joined12:26
→ orafu joined12:26
→ Jordan_U joined12:26
→ stepnem joined12:26
→ [[sroracle]] joined12:26
→ frakturfreak joined12:26
→ Morasique joined12:26
→ [bobrik] joined12:26
→ chrowe joined12:26
→ teknotus joined12:26
→ ivan joined12:26
→ _iksik joined12:26
→ SRabbelier joined12:26
→ Ahri joined12:26
→ rvsjoen joined12:26
→ mstormo joined12:26
→ electrogeek joined12:26
→ trochala joined12:26
→ hachi joined12:26
→ ruskie joined12:26
→ xrogaan joined12:26
→ odin_ joined12:26
← stepnem left12:26
← tar- left12:26
→ neutrino joined12:26
rkreis thanks a lot, jast and the others12:26
← d0k left12:27
← mstormo left12:27
← apsysoev left12:27
→ stepnem joined12:28
→ eMBee joined12:30
→ mstormo joined12:32
→ d0k joined12:33
→ bavbat joined12:33
→ Tabmow joined12:33
→ poseid joined12:34
← aziz left12:37
→ aless joined12:39
← disappearedng_ left12:41
← mastro left12:42
neutrino → tar-12:42
← poseid left12:42
← engla|swe left12:43
→ disappearedng joined12:44
← darwin_ left12:44
→ dr_win joined12:45
← rox` left12:46
← dr_win left12:47
→ dr_win joined12:47
← masterkorp left12:51
→ rox` joined12:51
→ mastro joined12:54
← Falc left12:56
← codeshepherd left13:00
← keyvan left13:00
→ raichoo joined13:02
→ OSInet joined13:02
← ige left13:04
← [bobrik] left13:04
→ codeshepherd joined13:05
OSInet hello. After some commits to a local git repos and a git rebase interactive I find myself with a broken repos (git status ="fatal: bad object HEAD", git branch="error: refs/heads/master does not point to a valid object!"). Tried git --fsck to no avail. None of the solutions I could find online (including on StackOverflow) appeared to fix anything. Is my repos doomed, or could someone suggest a link to some repair procedure ?13:05
→ jrmuizel joined13:05
OSInet (in case that can help, I have (read) the git book)13:07
for more symptoms: there are just 5 objects in the repos for 5 actual files remaining in the working directory and at least 5 commits. I can restore individual files from backup if needed.13:11
→ mw joined13:11
← mw left13:11
→ mw joined13:11
→ mw2 joined13:12
→ poseid joined13:12
wereHamster OSInet: maybe HEAD is corrupted, but the reflog may contain valid commits (possible the one just before you did the rebase)13:14
← Jaxan left13:18
OSInet wereHamster: thx for looking into it "reflog" == logs/refs/* ?13:18
Chris64 is rake part of the git project?13:18
wereHamster OSInet: man git-reflog13:18
→ aless_ joined13:19
OSInet checking13:19
hmm, no: "fatal: bad object HEAD"13:20
← aless left13:21
wereHamster when you run 'git reflog'?13:21
OSInet yes13:21
← Textmode left13:21
OSInet FWIW, I there is a sensible log in both logs/refs/HEAD and logs/refs//heads/master . they don't match, though13:22
wereHamster well, it's not unusual that they don't match13:22
← simplechat left13:22
→ hohoho joined13:23
OSInet however, they reference commits numbers which have no matching files in the repos, which I do not understand13:24
one clue (maybe) is that there is a COMMIT_EDITMSG.swp in the repos, which could mean a commit was interrupted13:26
wereHamster 'no matching files'?13:26
← Paraselene_ left13:26
→ cilly joined13:27
→ Paraselene_ joined13:27
→ saidi joined13:28
OSInet there are only five files in the objects/?? directories and i was under the (misled ?) impression that every commit it was supposed to match an object in these.13:28
saidi how can i create a snapshot ?13:29
→ aziz joined13:29
Chris64 a what?13:30
FauxFaux A commit?13:32
saidi snapshot from a commit ?13:33
get an archive of a git repository13:36
→ jfkw joined13:36
← pielgrzym left13:36
OSInet saidi: why not just do a "tar czf .git" ?13:37
← raichoo left13:37
OSInet saidi: oops "tar czf snap.tgz .git" ?13:38
saidi OSInet: what if i want to get an archive from a commit ?13:39
wereHamster saidi: like, man git-archive ?13:40
OSInet: yes, that's true, every commit will create at least one new object inside the objects directory (usually three or more)13:40
→ engla|swe joined13:41
wereHamster OSInet: but keep in mind that objects can be packed in a packfile (objects/pack/..)13:41
saidi wereHamster: yes, thanks13:41
OSInet wereHamster: objects/pack is an empty directory13:41
← codeshepherd left13:42
→ vu3rdd joined13:45
→ paja joined13:46
← tizbac left13:48
→ wakiki_ joined13:48
→ sagsousuke joined13:48
wakiki_ → wakiki13:50
← engla|swe left13:50
→ sdivp joined13:51
→ giskard joined13:51
← harinath left13:51
OSInet wereHamster: I take it this means I must assume everything is lost and I'd better accept losing my history and reinitializing a repos from the current version of the files ?13:51
sdivp hi, is there any free private git host ?13:52
→ Titosemi_ joined13:54
jettero sdivp: you mean like linux? openssh?13:54
sdivp jettero: like that13:55
→ scarabx joined13:55
← Titosemi left13:56
Titosemi_ → Titosemi13:56
← giskard left13:57
jettero sdivp: I clearly don't understand the problem...13:57
jast sdivp: you mean, you want someone else to host private repositories for you for free?13:58
I'm not aware of any such service13:58
jettero that's not likely to happen (if that's the case); but you shouldn't need it anyway. If your needs are private, just ssh to eachother's computers. Problem solved.13:58
← Zephyrus left13:59
sdivp ok thanks13:59
→ tedoc2000 joined14:00
← hohoho left14:01
→ multigrain joined14:02
OSInet I have restored all the deleted files under objects/?? and things seem to be a bit better14:03
← babafds left14:03
OSInet git fsck --full now says dangling tree (some commitid)14:03
→ hohoho joined14:03
sdivp jettero: what about http://indefero.net14:04
→ khmarbaise joined14:04
→ consolers joined14:05
OSInet and git fsck --full --unreachable sais unreachable tree (the same commitid)14:05
FauxFaux Dangling is normal.14:06
→ harinath joined14:06
jettero sdivp: please not hilight me for opinions about random websites I'm not going to load up14:07
← sdivp left14:08
OSInet FauxFaux: meaning I should just ignore it, or something more I have to do ? All commands seem to work again: I was able to commit the pending changes, git status and gitk no longer complain14:09
← saidi left14:09
FauxFaux As the recovering the repo guide says, yes, dangling are normal and ignorable.14:09
→ babafds joined14:10
OSInet FauxFaux: "recovering the repo guide" ? do you have an URL for this ? is it http://www.kernel.org/pub/software/scm/git/docs/user-manual.html#recovering-from-repository-corruption ?14:10
FauxFaux Yes.14:11
OSInet great. Thanks a lot wereHamster and FauxFaux for your help14:11
← rox` left14:11
OSInet rushes to do another backup before losing data again14:12
← consolers left14:13
wereHamster jettero: indefero is a web-based repo management software. He wanted your opinion about it. But you don't seem to be aware of its existence.14:13
→ LRN joined14:14
Chris64 do you know if there are placeholders in files for git like in svn?14:15
→ raichoo joined14:15
jettero wereHamster: I get that he wanted my opinion on it, but I already gave it. Just use ssh14:15
Chris64 like last commiter, commit date and so on14:15
wereHamster Chris64: yes, but their use is discouraged14:16
Chris64 discouraged?14:16
why? :|14:16
are they not necessary?14:16
wereHamster actually, git itself only supports $id$ which expands to the blob id. But you can make your own smudge/clean filters if you want14:16
no, they are not14:17
← SplinterOfChaos left14:17
Chris64 thank you, i'll search for it :)14:17
→ rox` joined14:17
wereHamster Chris64: look in the git faq14:17
Chris64 okay14:17
← henkboom left14:17
wereHamster but really, you should not put these informations into the files that you check into git14:18
Chris64 mh, okay14:20
→ henkboom joined14:20
→ RobertLaptop joined14:21
→ aspotashev joined14:23
← poseid left14:23
← OSInet left14:27
← sagsousuke left14:30
→ poseid joined14:31
→ rajeshsr joined14:31
→ sagsousuke joined14:31
← sagsousuke left14:34
→ sagsousuke joined14:34
← _ikke_ left14:36
→ Stummi joined14:37
→ pellera joined14:41
→ eletuchy joined14:42
← RobertLaptop left14:44
→ giskard joined14:46
← harinath left14:46
← poseid left14:47
← skmidry__ left14:53
← unreal left14:54
→ unreal joined14:55
→ poseid joined14:59
← airborn left14:59
→ MetaCosm joined15:00
→ harinath joined15:03
← cilly left15:04
← sagsousuke left15:05
→ phlowx joined15:05
→ skmidry__ joined15:05
→ Zephyrus joined15:06
← skmidry__ left15:06
→ sixteneighty joined15:10
→ tizbac joined15:12
← eletuchy left15:12
← giskard left15:12
→ eletuchy joined15:13
Blackguard → Blackguard|AFK15:16
Acorn is there an easy way to replace all the usernames in a git repo that match a certain name with another name?15:16
rkreis mh, you'd rewrite all those commits i guess15:18
thiago_home yes15:18
jast Acorn: assuming you already know that that will change pretty much all commit IDs, head on over to http://git.or.cz/man/git-filter-branch and see the --env-filter option and the example for removing all commits by a certain author15:18
thiago_home using git-filter-branch15:18
rkreis hi jast :)15:18
Acorn it's only a small repo that was started recently without about 20 commits from two different people15:19
← tedoc2000 left15:19
→ OFICINA7 joined15:19
Acorn hmm, found this on SO git-filter-branch --env-filter "export GIT_AUTHOR_NAME='New name'; export GIT_AUTHOR_EMAIL='New email'" HEAD15:20
how does it know which person's details to replace?15:20
jast that's where the example I mentioned comes in15:21
the one involving the McBribe fellow15:22
yaotti → yaotti_aw15:22
← MarkBao left15:23
→ skmidry__ joined15:23
→ tedoc2000 joined15:24
← tedoc2000 left15:25
→ jrguitar21 joined15:25
Acorn jast, thanks, does this look right? http://paste.pocoo.org/show/230265/15:25
← jrguitar21 left15:26
→ nsussman joined15:27
Acorn and can I just run that command and then push?15:28
→ jezdez joined15:29
← nsussman left15:29
→ phlowx_ joined15:29
Ilari Acorn: Note that rewrites history, so push will fail unless you override safety checks.15:29
jezdez how can I get a list of all committers of a repo?15:29
Ilari Acorn: And it will be very nasty to others using the repo.15:29
← wakiki left15:30
Ilari jezdez: Log with suitable format, sort and uniq?15:30
← jrmuizel left15:30
← EricInBNE left15:30
→ Jaxan joined15:30
→ MarkBao joined15:31
→ dreiss joined15:32
→ madewokherd joined15:33
← phlowx left15:33
Acorn Ilari, how do I make the change then? It's just me and one other person using the repo15:33
Ilari Acorn: The other person needs to either discard or rebase their changes (NOT merge).15:34
jezdez Ilari: thanks and d'oh :)15:34
Ilari: fwiw, `git shortlog -s | sort -r` seems like what I needed15:35
Acorn Ilari, so, we make sure that we both have the same version of the repo, I then run that command, and then I push with safety checks off, and deletes his copy and reclones or "rebases"?15:35
→ zimbres joined15:36
Ilari Acorn: fetch + reset can be used to overwrite local changes.15:36
← zimbres left15:36
→ zimbres joined15:37
← rox` left15:38
← tizbac left15:39
→ fantasticulous joined15:39
← fantasticulous left15:39
Chris64 will git always merge per line?15:41
jast Acorn: yes, that's one way that should work. reclone or fetch+reset are the easiest/quickest ways.15:41
Chris64 or also in a line?15:41
or is it not useful?15:42
→ m0 joined15:42
jast Chris64: if both sides of the merge change the same line (or lines that are very close together), git treats that as a conflict that needs to be fixed manually when you do the merge15:42
← khmarbaise left15:42
Chris64 because when writing latex code sometimes replacing a whole line is not useful ^^15:42
Acorn jast, great, I'll do that when he's next online so I don't cause any lost work15:42
Chris64 okay15:42
→ rox` joined15:43
tghw_ → tghw15:46
→ FunctorSalad joined15:47
← jezdez left15:47
FunctorSalad git newb question: I suddenly get this error with my /etc repository: >>fatal: pathspec '"X11/ratmenu/debian.anwendungen.b\374ro.menu"' did not match any files<<15:48
→ plediii_ joined15:48
FunctorSalad I don't care about that file, how can force erasure or whatever of it?15:48
→ khmarbaise joined15:48
thiago_home git rm15:49
FunctorSalad (the error breaks the whole commit)15:49
→ sylr joined15:49
← sylr left15:49
FunctorSalad oh even if it doesn't exist? (haven't checked)15:49
thiago_home if the file doesn't exist, git rm --cached15:50
← project2501a left15:50
FunctorSalad is the nonascii character at fault?15:53
← poseid left15:53
FunctorSalad (git itself can presumably handle it, but maybe it got scrambled by something else)15:53
→ yann joined15:54
→ jrmuizel joined15:56
→ programble joined15:58
→ psoo joined16:00
← henkboom left16:02
← alip left16:02
→ alip joined16:03
← jrmuizel left16:04
→ apsysoev joined16:05
→ henkboom joined16:05
→ Chillance joined16:07
→ jrmuizel joined16:12
← rox` left16:13
← raichoo left16:15
→ shruggar joined16:17
← harinath left16:17
→ rox` joined16:18
→ icefox joined16:19
← sixteneighty left16:19
← jrmuizel left16:27
offby1 my money is indeed on the non-asii character16:27
Chris64 which one?16:29
← malumalu left16:30
→ lresende joined16:30
thiago_home that \37416:30
← henkboom left16:31
→ harinath joined16:32
→ jrmuizel joined16:33
→ henkboom joined16:34
→ bmac_ joined16:43
← bmac left16:45
→ Yuffster joined16:45
← FunctorSalad left16:45
→ MicahElliott joined16:48
→ patrikf joined16:48
← vu3rdd left16:52
← g3d left16:52
← henkboom left16:52
← MicahElliott left16:53
→ henkboom joined16:55
→ engla|swe joined16:56
← dreiss left16:58
→ alester joined17:00
← henkboom left17:01
→ henkboom joined17:02
→ raichoo joined17:06
← raichoo left17:09
→ tom][v joined17:10
→ warlock_mza joined17:12
← fmardini left17:15
→ ExtraSpice joined17:16
← shruggar left17:19
← mw left17:20
→ snearch joined17:20
← Jaxan left17:20
← jrmuizel left17:24
← mw2 left17:25
← snearch left17:28
→ jrmuizel joined17:29
→ Titosemi_ joined17:32
→ RobertLaptop joined17:33
→ snearch joined17:34
← Titosemi left17:34
Titosemi_ → Titosemi17:34
← snearch left17:35
→ snearch joined17:35
→ Leemp joined17:39
→ Tuller joined17:40
→ slashbeast joined17:40
→ marc_kdab joined17:42
← rado1 left17:43
→ Bennid joined17:44
→ iruediger joined17:46
← Chillance left17:48
→ tizbac joined17:50
← tom][v left17:50
→ Transformer joined17:52
← Transformer left17:53
← eletuchy left17:56
→ eletuchy joined17:56
← alester left17:58
← froschi left18:00
→ mindworx joined18:00
← zimbres left18:00
← chrislerum left18:02
→ spearce joined18:03
→ hobodave joined18:04
→ froschi joined18:04
← harinath left18:08
→ mattikus joined18:09
← psoo left18:11
→ Vadtec joined18:14
→ gebi joined18:14
Vadtec drizzd: are you there by chance?18:14
← mindworx left18:14
→ mindworx joined18:15
→ zimbres joined18:16
← tar- left18:16
→ neutrino joined18:16
neutrino → tar-18:16
→ tom][v joined18:17
← eletuchy left18:19
← jrmuizel left18:20
→ Jaxan joined18:22
Vadtec guess not...well, in any case, last night i *finally* got a simple remote git repo setup, and to go along with it, I made a howto/tutorial/list of steps detailing what I did to get it to work, the results of which can be found here: http://wiki.vadtec.net/wiki/Git:RemoteRepo18:22
feel free to pass it along to users trying to do the same thing, because google failed me when i tried to find such a simple example of how to setup a remote git repo18:23
← priidu left18:23
Vadtec enjoy18:23
→ harinath joined18:23
← Vadtec left18:23
← tar- left18:23
→ neutrino joined18:23
neutrino → tar-18:23
→ jmcantrell joined18:25
→ MicahElliott joined18:25
← OFICINA7 left18:26
→ eletuchy joined18:26
wereHamster ugh, advising users to push through git:// ..? bad idea18:27
← henkboom left18:27
← jmcantrell left18:27
← snearch left18:28
← pellera left18:29
wereHamster wow, that guide is wrong on so many things..18:30
→ henkboom joined18:31
→ icwiener joined18:32
← Chris64 left18:33
← mastro left18:34
→ archis joined18:34
← marc_kdab left18:34
← mattikus left18:37
→ Alagar1 joined18:38
← froschi left18:39
→ giskard joined18:40
← giskard left18:41
chuck → charlie18:41
→ apsysoev1 joined18:41
← Alagar1 left18:42
→ Alagar1 joined18:42
charlie → Guest2460318:42
← Jordan_U left18:42
← apsysoev left18:43
→ mastro joined18:44
← Tuller left18:45
← Alagar1 left18:46
→ dl9pf_ joined18:46
→ Alagar joined18:46
← dl9pf left18:47
← Alagar left18:47
← archis left18:49
→ brosner joined18:55
← brosner left18:55
→ brosner joined18:56
← brosner left18:56
→ Alagar joined18:56
→ Tuller joined19:00
→ whaley joined19:00
← bavbat left19:00
→ brosner joined19:01
← brosner left19:01
→ brosner joined19:03
← mindworx left19:03
← ph^ left19:06
→ ph^ joined19:06
→ jailbox joined19:09
→ Oxf13 joined19:10
→ ziro` joined19:10
Oxf13 Can anybody think of a reliable way to figure out what first remote branch from master a given local branch is tracking? Lets say a local branch is tracking an upstream branch of a branch (of a branch). I need to reliably discover what first level branch it eventually tracks back to.19:11
→ stringo0 joined19:11
thiago_home forget it19:11
a branch, once created, is indistinguishable from the branch it branched from19:11
the only information saved is the "tracking" branch and that's only for local ones19:12
Oxf13 yet you can track the ancestry via the commits, you can kind of see it with git list-branch19:12
thiago_home you can't access that info in remote branches19:12
I recommend you rethink what you're trying to accomplish19:12
patrikf Oxf13: so when the two branches diverge, how would you know which one was there first?19:12
Oxf13 in my situation, we pre-create "top level" branches for our users19:13
they can then branch from those top level to do specific work19:13
I need a client tool to be able to discover which "top level" branch their work is the ancestor of19:13
or the child of I should say19:13
thiago_home encode it in the branch name19:13
patrikf Oxf13: so every user has an own branch?19:13
thiago_home or ask them to write it down19:13
← RobertLaptop left19:14
Oxf13 patrikf: they can.19:14
thiago_home or simply don't care19:14
Oxf13 patrikf: every module will get branches such as "F-13" and "F-14"19:14
thiago_home why do you need to know which branch they branched off from?19:14
Oxf13 the user has the option to further branch that for specific work19:14
patrikf Oxf13: sounds like a workflow not well fitted to git19:14
Oxf13 thiago_home: We have to make some runtime decisions based on what upstream branch the code is from, when it is sent off to build.19:14
thiago_home Oxf13: ask the developers to write it down and save it in your build tool19:15
→ mindworx joined19:15
Oxf13 encoding it in the branch name would work, but it would somewhat suck.19:15
thiago_home: that's not going to work for 9K repos and 2K developers19:15
patrikf Oxf13: many tools that layer over git commit a dotfile to track additional information19:16
Oxf13 yeah, I wanted to avoid junk in the module.19:16
thiago_home Oxf13: then you need to write a tool to help them19:16
Oxf13 I could do that, but didn't want to.19:16
i was just hoping there was a heuristic you could use to walk the commit history and discover the common top level branch.19:17
patrikf Oxf13: commit, yes, branch, no19:17
→ Vadtec joined19:17
Oxf13 well the commit lives in a branch no?19:17
patrikf Oxf13: no19:18
Oxf13 git show-branch is awfully close to showing me what I need19:18
thiago_home the commit lives in all branches19:18
patrikf Oxf13: a branch references a commit... but not the other way round19:18
thiago_home the best you can do is figure out which branch is closer to that commit19:18
but that's a risky business if your top-level branches are somehat close to each other19:18
Oxf13 what if we make a specific commit that marks it as a branch, or a specific tag?19:19
can you track back to a specific tag?19:19
→ ChUrP1 joined19:19
thiago_home I didn't understand19:19
you commit, then mark it by creating a branch on it19:19
that's possible19:19
patrikf Oxf13: what are you *actually* trying to do? why do you have branches with strange names such as F-13?19:19
Oxf13 so lets say when you make the branch, you then tag it with a specific tag, like F-13-Start19:19
thiago_home so what?19:19
ok19:20
Oxf13 patrikf: because we're moving to get to manage package sources for Fedora.19:20
patrikf: and we branch the package modules for each release19:20
→ patmaddox joined19:20
thiago_home are the branches ever merged to each other?19:20
Oxf13 thiago_home: rarely19:20
thiago_home ever?19:20
even once?19:20
patrikf Oxf13: and why do you need to track back?19:20
Oxf13 patrikf: because our build tool decides how to autofill some of our package info based on what branch/release you're building for19:21
→ jensn joined19:21
thiago_home if they are never merged, then you can find some exclusive commits19:21
Oxf13 it also decides where to send the build request based on what branch/release you're working on19:21
thiago_home that is, you can find a particular commit in the history of a branch, but not a particular other commit19:21
Oxf13 thiago_home: does merging one branch to another also merge tags?19:21
thiago_home your question doesn't make sense. Tags don't belong to a branch.19:21
they belong to the repository19:21
the only thing I can think of is simply marking well-known commits that exist only in a particular branch19:22
you can tag those marker commits19:22
Oxf13 "well-known" commit could be the tag19:22
← ChUrP left19:22
thiago_home but you must never merge, because then you'd have two or more markers in a same branch19:22
patrikf Oxf13: I think your life would be easier if that kind of version information lived in a separate file, and users could freely name their branches19:22
Oxf13 that's unfortunate.19:23
→ GEEK00 joined19:23
Oxf13 I guess I'm having a hard time wrapping my head around "commit belongs to the repo not the branch" when there are commits that I can see in the history of a branch, that I can't see in the history of another branch19:24
DrNick there's nothing stopping that commit from appear in another branch, too19:25
patrikf Oxf13: you also have earlier commits that both branches contain19:25
thiago_home Oxf13: one commit points to one or more parents19:25
Oxf13: a branch is nothing more than a pointer to a commit19:25
Oxf13 DrNick: sure, if somebody pulls that or cherry-picks that commit19:25
thiago_home if you have two commits A and B, then there are three options:19:25
DrNick no, not cherry-picking19:25
patrikf Oxf13: also, if your purpose is just some automatic guessing (with possible correction by the user), then of course you can just check which of the F-* branches is the closest19:25
thiago_home A is a descendant of B, B is a descendant of A, or A and B are disjoint19:26
Oxf13 DrNick: or you merge19:26
patrikf: how do I discover which of the F-* is closest?19:26
thiago_home patrikf: you can count the number of commits between the F-* base and the particular branch19:26
by definition, the lowest number is the closest19:27
← brosner left19:27
thiago_home but that isn't necessarily what you expect, if you merge19:27
← name left19:27
Oxf13 damn. I guess I will have to go with a .junk file19:28
patrikf Oxf13: don't you already have a meta-information file in fedora? e.g. like debian's debian/changelog19:28
Oxf13 no19:28
there is a .spec file19:28
patrikf that sounds like the place for such information19:29
Oxf13 but they are designed so that they can be exactly the same across multiple branches19:29
thiago_home I don't see why such information needs to be saved19:29
Oxf13 so that you can merge if you're keeping all branches the same.19:29
→ alester joined19:29
patrikf Oxf13: you could also make it an option for your build tool19:29
Oxf13 thiago_home: because when I clone a branch of a particular package, I need to be able to easily build it without having to guess and research.19:29
→ shruggar joined19:29
Oxf13 patrikf: asking the users to remember what target to build for won't work. They'll get it wrong, constantly.19:30
this has to work very simply. "fedpkg build" should do the right thing.19:30
thiago_home I don't get it19:30
Oxf13 fedpkg is our helper program19:30
thiago_home no, I just don't get what the issue is19:31
sounds like you branch from master when you release the distro (or branch the distro)19:31
Oxf13 so lets say you've got a check out of "yum" and it's on a local branch by the name of "CoolBranch"19:31
thiago_home you do that in all repositories. Fine.19:31
so?19:31
Oxf13 CoolBranch is an upstream branch from F-1319:31
DrNick if a use does a scratch build of an arbitrary branch, it needs to be sent off to the right builder for the distro version that branch is targetting19:31
Oxf13 so the build helper needs to be able to tel lwhen you say "fedpkg build" that the build is supposed to go to our F-13 target19:32
thiago_home so a user clones a repository, branches off a particular branch, then says "now build this' ?19:32
Oxf13 and that certain macros in our spec files resolve out to the F-13 version of them.19:32
thiago_home: that is one work flow yes.19:32
thiago_home sounds like that info needs to be saved in the files, not in Git19:32
Oxf13 otherwise most people just build from the pre-created F-?? branches.19:32
thiago_home what happens if they take the thing out of Git?19:33
← mastro left19:33
Oxf13 thiago_home: then it doesn't get built in our buildsystem.19:33
these aren't upstream repos19:33
thiago_home so the users push a branch and it gets built?19:33
Oxf13 they are repos we create to track the spec files and aptches.19:34
thiago_home: they push, then request a build of a given hash.19:34
if that hash isn't found in our repos, the build fails.19:34
thiago_home you said that the discovery would be just a hint and users would be able to override, right?19:34
Oxf13 the discovery would be what the tool uses to discover. A user might be able to override that with specific options to the tool19:34
but this is the edge case, done by people who know what they are doing.19:35
thiago_home I still think this needs to be saved in the files themselves19:35
Oxf13 (and the buildsystem may yet refuse to do it)19:35
thiago_home sounds a lot easier19:35
Oxf13 easier yes, but still dirty feeling19:35
thiago_home the only difficulty is to modify the control file at branch-distro time, but not that much more difficult than branching in all repositories19:35
it has a clean feeling for me19:35
Oxf13 it's something added to the repo that doesn't actually go in the package.19:36
which I was trying to avoid.19:36
thiago_home you probably won't add .gitignore files either19:36
DrNick Oxf13: why can't the target be encoded in the branch name?19:36
Oxf13 DrNick: people can name local branches however they want19:37
DrNick or do you just want it to work with arbitrary ... right19:37
thiago_home local branches are local19:37
Oxf13 DrNick: the difficulty would be tracking back to what base upstream branch the local branch (of a branch?) came from19:37
thiago_home when they push to the server, that's different19:37
Oxf13 thiago_home: but at build request time, all I really have is the local branch name to work with.19:37
thiago_home no19:37
Oxf13 or whatever remote branch that local branch is tracking19:37
according to the local repo19:37
thiago_home right19:38
but the local branch name is useless anywhere else but local19:38
the only info that is everywhere is the SHA-119:38
← thomas_b left19:38
thiago_home will *all* packages have *at least* one commit in *every* single release?19:39
Oxf13 we can guarantee that if we have to19:39
all work starts in origin/master19:39
that's our development branch19:39
thiago_home yup19:39
Oxf13 when we shift into release mode to stabalize, we branch from origin/master and create origin/F-1419:39
thiago_home if you don't ensure at least one commit, then two branches could be actually equal19:39
Oxf13 if we have to, we can make a bullshit commit to that newly created branch19:40
DrNick you'd have to make the bullshit commit anyway for your .target files19:40
Oxf13 that would guarantee that every top level branch has at least one commit19:40
→ thomas_b joined19:40
Oxf13 DrNick: I'm trying to do away with having .target files19:40
patrikf Oxf13: you said the F-* branches could be merged sometime?19:40
thiago_home you mean "one commit that is not on master"19:41
Oxf13 thiago_home: we do want to allow users to create their own remote branches, they would be branches of a top level F-?? branch, or of origin/master19:41
patrikf: yes, because we design our spec files so that they work without change when building from one release to another, it is possible for somebody to make a change on origin/master, then merge to the other branches19:41
or make a change on F-13 and merge it to F-12/F-1119:41
thiago_home right19:42
patrikf Oxf13: usually you cherry-pick in such situations19:42
thiago_home the "git way" would be that, if you don't have to, you don't create a commit19:42
so you could have exactly the same SHA-1 for F-12, F-13 and F-1419:42
Oxf13 thiago_home: right, that's one of the reasons why we're moving to git, to make this kind of thing easier19:42
and cleaner19:42
do the work once, merge it where appropriate.19:42
thiago_home but then it's impossible to tell which one was branched from19:42
because there's no distinction. They are all the same.19:43
← tom][v left19:44
thiago_home so your options are:19:45
Oxf13 oh right, the .foo junk won't work19:45
thiago_home - find the branch that is closest, without ensuring at least one commit19:45
Oxf13 because you'd try and merge it and then you'd screw up the contents.19:45
thiago_home - find the branch that is closest, ensuring at least one commit19:45
← GEEK00 left19:45
Oxf13 unless we can setup a merge rule that never tries to merge changes to .buildmap19:45
or .branch or whatever I call it19:45
thiago_home I don't think it would conflict19:46
→ lispy joined19:46
Oxf13 if on origin/master the .branch says "rawhide", you branch for F-13 and change it to say "F-13", commits happen for a while, then you merge origin/master over to F-13, what happens to the contents of .buildmap ?19:46
er, .branch19:46
I suppose if it wasn't changed on master after it was changed on F-13 that might be OK19:47
DrNick custom merge driver that always keeps the version from the destination branch19:47
thiago_home yes, that workflow wouldn't conflict19:47
but in my opinion the merge is in the wrong direction19:47
→ mmw joined19:48
thiago_home the .branch file would contain "F-13" today, in the master branch19:48
once F-13 branches and the work starts on F-14, the .branch files are all updated to say "F-14"19:49
Oxf13 no19:49
thiago_home yes19:49
Oxf13 we've been down that road19:49
things go weird around branch time19:49
thiago_home anyway, once a fix is done in the F-13 branch, it's merged up to master19:49
Oxf13 I've found it better if the map never changes on origin/master19:49
thiago_home no conflict in the file19:49
Oxf13 and only gets updated once we branch19:49
thiago_home I'm telling you what works for me with version files19:50
Oxf13 thiago_home: a lot of times the work on the F-13 branch is never seen on origin/master19:50
because the F-13 branch may be package version 3.2 and have 3.2 bugfixes, but origin/master moved on to 3.3 or even version 419:50
→ Chris64 joined19:50
Oxf13 so the changes on F-13 aren't appropriate for origin/master19:50
thiago_home assumes that any work done for a release X is valid for release Y provided Y > X19:50
thiago_home at least that's how it works for doing code in Git19:51
Oxf13 not at all19:51
massive rewrites happen19:51
thiago_home in the newer branches19:51
Oxf13 a bugfix for code in version 3.2 may not even exist for version 419:52
thiago_home but then the merge will conflict, requiring a manual fix19:52
Oxf13 the merge would "conflict" massivly19:52
thiago_home but at least you know that there was a change that may require your attention19:52
I know, it happens19:52
Oxf13 it's completely inappropriate to merge something like that19:52
thiago_home but for the normal use-case of code not getting rewritten at every release, it works just fine19:52
Oxf13 also, this isn't "code" in the normal sense.19:53
this is a rpm spec file, and maybe some .patch files19:53
→ chrislerum joined19:53
thiago_home actually, .spec files do look like code19:53
Oxf13 and another file that lists the active source tarball19:53
→ little_owl joined19:53
Oxf13 changes get made on origin/master that are generally pretty dramatic, new versions, changes to how the spec looks, changelog and version entries for rebuilds for new compilers, things like that19:54
← slashbeast left19:54
thiago_home yup19:54
Oxf13 changes on the branches are less dramatic, but diverge from origin/master19:54
thiago_home yup19:54
but you do want those minor changes to show up in the next version, right?19:54
Oxf13 and are often irrelevant to origin/master19:54
no19:54
thiago_home so there's a rewrite in every single release?19:54
Oxf13 the changes aren't what you're thinking19:54
or even if so.19:55
lets say you've got a minor bug in the 3.2 version of your source in a particular function19:55
so you fix it in 3.2.19:55
little_owl Is there some intro for a mercurial user, who is tempted to try git?19:55
Oxf13 the bug doesn't exist in 3.3, the function doesn't even exist in 419:56
so why would you try to merge this change into 3.3 and 4?19:56
thiago_home Oxf13: because if the bug was in 3.2, it was in 3.3 as well19:56
3.2 and 3.3 were branched off the same source19:56
same "master"19:56
Oxf13 no it isn't19:56
thiago_home yes it is19:57
Oxf13 ...19:57
thiago_home there's master, it branches off to 3.219:57
then some time later, master is branched off to 3.319:57
Oxf13 sorry I meant "no it isn't" to "the bug was in 3.2 it's in 3.3"19:57
thiago_home if there's a bug in 3.2, there's a good likelihood that it's there in later versions too19:57
→ hyperair joined19:57
Oxf13 "a good likelihood"19:57
except for when it is obviously not19:57
because say that function got re-written between 3.2 and 3.319:57
thiago_home it's not if the bug is a regression in 3.2.1 compared to 3.2.019:58
or in the case of rewrites19:58
but, like I said, rewrites are not the common case19:58
Oxf13 the way our distro works, the more common case is to have "rewrites" between the branches19:58
thiago_home and the only other case I can think of not wanting a fix in a later version is of temporary hacks19:58
I know, I understand that19:58
Oxf13 (for that matter, rewrites happen on the branches too)19:58
thiago_home but that's not the common case for long-lived code19:58
Oxf13 thiago_home: upstream packages change fast19:58
we pull in those new versions frequently19:59
we have 9K+ packages19:59
thiago_home anyway, my recommendation is to find the closest branch and not enforce a commit19:59
Oxf13 many many of those are still fast moving projects.19:59
thiago_home if a package wasn't touched between two Fedora versions, then there's an ambiguity19:59
Oxf13 thiago_home: ok, how does one find the closest branch?19:59
thiago_home but I'd say it's something you can live with19:59
git rev-list <basebranch>..<sha1> | wc -l19:59
compare the count20:00
the lowest strictly-positive number is the one you probably want20:00
the 0 case can happen if basebranch == sha120:01
← little_owl left20:01
thiago_home if you use tags (especially annotated tags), both the branch name and the commit count are shown in git describe's output20:02
← yann left20:02
Oxf13 our tags are typically in the form of <name>-<version>-<release> for the given package.20:03
thiago_home assuming you never or rarely merge, this will probably work quite well20:03
Oxf13 which may or may not have any information about the specific Fedora release.20:03
it could jsut be "glibc-2.3.2-2.5"20:03
thiago_home you can ask describe to describe only relative to a specific set of tags20:03
→ huehnts joined20:04
← huehnts left20:08
← hohoho left20:08
→ offby1` joined20:11
→ dreiss joined20:14
← gebi left20:14
→ jmspeex joined20:15
→ gebi joined20:16
Vadtec greetings all, it came to my attention that the guide i linked earlier had some nasty issues, thanks to wereHamster, i believe they have been sorted, as such, i have updated my guide and welcome any comments, http://wiki.vadtec.net/wiki/Git:RemoteRepo20:22
← tar- left20:22
→ neutrino joined20:23
neutrino → tar-20:23
← hyperair left20:24
→ OFICINA7 joined20:25
← gebi left20:26
→ froschi joined20:27
→ ilogger2 joined20:28
shruggar Vadtec, first note I have from reading the first paragraph: your guide seems to be talking about setting up a remote shared repository, but the title and opening lines seem to imply that all this additional set-up is required merely for creating a remote git repos20:31
I'd add mention that the guide is for a remote repos to be used by multiple users20:31
→ raichoo joined20:32
shruggar (and I wouldn't call the requirement of multiple users to be a "simple" setup)20:32
→ gebi joined20:33
→ jrmuizel joined20:34
Vadtec shruggar: hmmm, yes, it could be used by multiple users, in this case, its only me using this particular repo atm, which is reflected in the wording20:35
→ ilogger2 joined20:35
Vadtec ill reword it to reflect the fact that multiple users can use it as well20:35
im curious though, how is this not a simple setup for multiple users? you add the repos and their ssh keys and they can commit, can it get any more simple than that?20:36
shruggar if only one user is using it, a group shouldn't be necessary. nor should git-daemon (though I've only been skimming, it seems you have ssh access anyway)20:36
→ marcial5_ joined20:37
Vadtec the group is only for the shell user that is being created20:37
as in the user:group20:37
shruggar Vadtec: I'm just hesitant to call something a "simple guide to setting up a git remote", when it goes through all these extra steps which aren't needed for the most basic set-up20:38
→ Weems joined20:38
Vadtec as for git-daemon, i track this project with Ohloh, which doesnt provide an ssh key, so i need read-only access to allow it to obtain updates20:38
marcial5_ → marcial520:39
Vadtec is it the git-daemon section that you think makes it "not a simple" setup?20:39
shruggar the most basic set up for a git remote in my mind is: 1) ensure you have ssh access to the remote location. 2) ssh in to that location 3) from the remote, create a new directory called something.git 4) from within that directory, do git init --bare 5) from your local repository, do "git remote add someremote username@remote-host:path/to/directory.git 6) type "git push someremote somebranch"20:40
Vadtec, anything that requires an extra group, and extra user, any access controls, is not a "simple guide to setting up a remote". It might be a simple guide to setting up a git repository which allows for read-only access from Ohloh, but it's far too complex to be a simple guide for merely setting up a remote20:42
so, my only issue is with the name. Other than that, it seems thorough and well-phrased20:42
Vadtec it really doesnt have a name, but i can see your point of view, ill work some wording changes into it as i have the time20:43
AAA_awright Oxf13: The thing to avoid getting caught up in is a branch is just a pointer to a single commit, it doesn't describe an ancestry in any fashion. A branch is a tag that can advance when a new commit is made, pretty much.20:43
Vadtec but right now, its time to bbq20:44
shruggar this is the part I disagree with: "How to setup a simple remote git repo without needing gitosis or other such things."20:44
Vadtec shruggar: fwiw, i just threw those steps together mostly for my own sake last night, but i had intended to re-work it into a proper guide anyways, its just not my style20:45
ok, so ill remove simple and replace it with moderate20:45
or something of the sort20:45
Oxf13 AAA_awright: yeah, I kinda get that, yet still there are tools which show ancestry of branches.20:46
→ jmcantrell joined20:46
Vadtec shruggar: refresh the page, that should be satisfactory20:46
→ jasonn1234 joined20:48
→ henkboom joined20:48
← henkboom left20:53
shruggar Vadtec, works for me. Just trying to avoid scaring off people who might switch to git if not for the giant fears of "omg wtf why is it so complicated?"20:53
more acceptance of git by the easily-scared = easier time convincing co-workers to use git :)20:54
→ henkboom joined20:55
→ Mage joined20:55
→ offby1 joined20:57
patrikf Vadtec: I don't think having a separate git user for ssh access makes any sense21:02
→ brosner joined21:02
← brosner left21:02
Vadtec sits down after lighting the grill21:02
patrikf Vadtec: maybe for git-daemon as a privilege separation measure, but otherwise just give every developer his own login21:02
Vadtec patrikf: its a priv separation issue21:03
and it makes it easier to keep track of all the repos, imo21:03
patrikf Vadtec: how is it a privilege separation issue?21:03
Vadtec for 1, the git user has no password, so unless someone steals an ssh key, it cant be used as an attack vector21:04
patrikf Vadtec: well, I'm not telling you how to do it, I just think it's way overcomplicated and not elegant21:04
Ilari Seperate git user would make fs permissions easier...21:04
Vadtec and for 2, by using git shell, if it is compromised, the damage is minimal21:04
→ ilogger2 joined21:05
Vadtec it increases greatly21:06
because they have more access to more things21:06
patrikf how so?21:06
Vadtec git-shell only allows 3 commands21:06
a regular bash shell allows gcc, cat, etc21:06
shruggar Vadtec, you can restrict commands usable by an ssh key21:06
Vadtec if there is a local root sploit and they get a regular shell login21:07
they can get at the box21:07
sure you can21:07
→ t0rc joined21:07
Vadtec but this is simpler than having to edit the ssh rules21:07
patrikf Vadtec: I don't know if we're talking about the same thing. Even with a separate git user, I take it you would still have another user able to log in via SSH in order to administrate the box?21:07
Vadtec the only account on this vps that has a password is my account21:08
everything else uses ssh keys21:08
patrikf: yes, there are separate logins to administer the box21:08
shruggar Vadtec, I don't see how "make a new user and group" is simpler than "add a line to .ssh/authorized_keys", especially considering that the "add a line" part is a step in /both/ procedures ;)21:08
Vadtec but id love to see anyone get the ssh keys that would allow them to login21:08
ok, its this simple21:09
i want to have all remote repos on that box housed under ONE location21:09
i chose to do that via a dedicated git user21:09
patrikf /srv/git?21:09
Vadtec: you get more simplicity and more fine-grained access control (should you ever need it) if you give each developer a regular account, instead of sharing one21:10
Chris64 do you really need ssh access?21:10
Vadtec but then if you need to have multiple developers commiting to each others repos, you have to add their ssh keys to everyones login21:10
which is a major security risk21:11
patrikf Vadtec: nah21:11
shruggar patrikf, unless he uses gitolite ;)21:11
patrikf Vadtec: that's why unix has groups...21:11
Chris64 what is gitolite?21:11
patrikf (and git has --shared=group)21:11
→ Titosemi_ joined21:11
Vadtec yeah well, dont bash me because im new to git, ive only been using it for a few weeks now21:11
that wont change my mind about how i setup my repo21:12
Ilari Chris64: Gitolite performs authorization of access into git repositories.21:12
shruggar gitolite is a way of solving all the various problems which people have when they want to do something more-complicated than "ssh in to my own account and push into a git repos which is owned only by me"21:12
Chris64 interesting :)21:13
Vadtec if you all think my guide is so bad, feel free to edit it and clean it up21:13
Titosemi_ → Titosemi21:13
→ brosner joined21:13
← brosner left21:13
Vadtec it really is a moot point regardless21:13
Chris64 do you know how github works? because they won't have ssh access for everyone or?21:14
Ilari Chris64: Usually gitolite is run on top of sshd.21:14
patrikf Chris64: they do, but you can only execute git receive/upload-pack21:15
Chris64 mh, okay21:15
→ ilogger2 joined21:16
Ilari But sshd isn't the only option (well, practically it is)...21:18
→ ratc joined21:19
→ Tuller joined21:20
→ airborn joined21:21
→ hyperair joined21:21
patrikf Ilari: indeed, there's WebDAV ;-)21:21
Ilari patrikf: Gitolite can't run on top of WebDAV.21:21
patrikf Ilari: oh, thought you were talking about git in general21:22
→ Jaxan joined21:23
→ _ikke_ joined21:24
shruggar anyone know where I can look for an example of determining whether or not the work-tree and/or index are dirty in C? (as in, within the git code)21:29
→ JStout joined21:29
← Jaxan left21:30
→ MetaCosm joined21:30
→ Jaxan joined21:31
→ SRabbelier joined21:31
→ Paraselene_ joined21:32
← Jaxan left21:33
→ Jaxan joined21:33
→ jasonn1234 joined21:38
→ brosner joined21:38
← SRabbelier left21:39
→ consolers joined21:42
← hyperair left21:43
← Jaxan left21:43
← jasonn1234 left21:45
→ ChUrP joined21:45
→ SRabbelier joined21:47
← MetaCosm left21:50
→ MetaCosm joined21:50
wereHamster shruggar: see which git command you would use, then see which cmd_ method that command is using, then call that method :P21:53
→ mithro joined21:54
shruggar yeah, that's where I've been digging21:54
→ ciupicri joined21:55
→ jasonn1234 joined21:55
ciupicri how can I still commit a file from an ignored directory?.gitignore has "/SOURCES" and I want to commit (repo_root)/SOURCES/django-threaded-multihost-no-ez_setup.patch21:56
→ jmcantrell joined21:58
shruggar ciupicri, "git add the/file", then commit21:59
→ iruediger joined22:00
shruggar ciupicri, possibly add -f to that "git add"22:00
→ mw joined22:00
← Tuller left22:01
← MetaCosm left22:02
ciupicri shruggar, let me try the -f22:02
shruggar, yep, it worked. Thanks! I guess they don't say "Use the force Luke!" for nothing :-)22:03
→ froschi joined22:03
→ mw2 joined22:03
shruggar ah, that explains why I couldn't find what I'm looking for - I was running the wrong command22:04
diff-index I wanted, not diff-files22:04
→ Tuller joined22:07
← jasonn1234 left22:07
→ skmidry_ joined22:10
← mw left22:12
← froschi left22:13
→ bobrik joined22:14
→ jasonn1234 joined22:14
← jasonn1234 left22:15
← mw2 left22:16
← consolers left22:18
→ henkboom joined22:18
→ mindworx joined22:23
→ aliceinwire joined22:25
→ chrislerum joined22:30
→ harinath joined22:32
→ adamm joined22:33
← adamm left22:33
→ adamm joined22:33
← mindworx left22:33
← ratc left22:35
→ ratc joined22:35
← chrislerum left22:36
→ bagoor joined22:38
bagoor git-daemon supports encryption ?22:38
→ eletuchy_ joined22:39
→ mindworx joined22:39
← harinath left22:44
← iruediger left22:44
kokx bagoor: afaik, not, i recommend using ssh for push access ;)22:45
← mindworx left22:45
Ilari bagoor: Nope, git-daemon does not support authentication nor encryption.22:46
→ coppro joined22:47
← jmcantrell left22:47
→ EricInBNE joined22:47
→ drizzd joined22:47
bagoor What's the best method for providing GIT without SSH Account ? Gitosis ?22:50
Ilari bagoor: Gitolite.22:50
bagoor Ilari, is it better than Gitosis ?22:52
→ mindworx joined22:52
Ilari bagoor: Yup.22:53
← drizzd left22:54
→ maaadbob joined22:55
maaadbob How do I make my local 'master' branch from git.kernel.org updated to the latest version without having to delete and re-download everything ?22:57
→ harinath joined22:57
wereHamster maaadbob: git checkout master; git pull22:58
maaadbob From there I can now clone that download and carry on as usual ?22:59
..branch that download...22:59
→ giskard joined22:59
→ iruediger joined23:00
wereHamster you already cloned a repo, right?23:00
maaadbob Yes, but deleted it. I should just 'branch newfile', 'checkout newfile' ?23:01
wereHamster deleted what/23:02
the whole clone?23:02
maaadbob I downloaded the master, then made a branch of it. It is the latter I deleted with 'git branch -D xxxx'23:02
wereHamster and now you want to do what..?23:03
maaadbob Create a new branch and work on that. I should be able to do that with the commands branch and checkout.23:04
wereHamster maaadbob: git branch the-new-branch master; git checkout the-new-branch23:04
maaadbob Yes, I understand now.23:05
Done that. Git report 'M drivers/staging/comedi/drivers/amcc_s5933.h'/23:06
which is the file I updated when I had my old branch. Is the 'M' important ?23:07
wereHamster it tells you that the file is modified in your working tree23:08
→ skmidry__ joined23:08
maaadbob I see. Must have missed that in the docs.23:09
Thanks for your help. Submitted my first ever kernel patch recently. It was nothing more than 'code cleanup', but it feels good to be able to contribute.23:10
→ jensn joined23:10
← jensn left23:10
← skmidry_ left23:12
← maaadbob left23:12
→ plediii_ joined23:14
→ aless joined23:15
← giskard left23:15
→ brizly1 joined23:17
→ starcoder|Svr joined23:24
→ giskard joined23:24
→ mw joined23:24
→ mw2 joined23:29
← giskard left23:29
→ homiziado joined23:32
← mindworx left23:33
→ Transformer joined23:33
← Transformer left23:34
→ ft joined23:37
← mw left23:37
→ jasonn1234 joined23:40
← mw2 left23:41
← eletuchy_ left23:43
← ciupicri left23:44
→ little_owl joined23:44
→ javaJake joined23:45
javaJake The subversion repository I'm cloning has a nice trunk/branch/tag layout. However, there is another extra folder with other projects that don't have such nice layouts, and are basically trunks of other projects. How do I add these as git branches in git-svn?23:46
← Tuller left23:46
→ mindworx joined23:46
javaJake So it looks like this: nice-folder/{trunk,branches,tags} , not-so-nice/[project-name]/23:46
I'd like to add all the not-so-nice/* folders as more branches, but of different projects.23:47
Should each be a different git repo?23:47
wereHamster if those are different projects, then you probably should import them into separate git repos23:47
javaJake OK23:47
Sounds reasonable, actually23:47
wereHamster one git repo = one project23:47
javaJake Yep, yep, makes sense. I've used svn for far too long. ;)23:48
← airborn left23:51
javaJake Thanks for the advice.23:51
← javaJake left23:51
→ airborn joined23:51

Logs Search ←Prev date Next date→ Channels Documentation