IRCloggy #git 2009-12-28

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.

2009-12-28

← JackTheRabbit left00:00
← gma left00:01
← Paraselene_ left00:01
← jlilly left00:04
← reprore left00:04
→ ereslibre_uni joined00:05
→ reprore joined00:05
← Nalwin left00:06
← hipe left00:07
← maxter left00:07
→ ferdy joined00:09
← ereslibre_laptop left00:09
→ JackTheRabbit joined00:10
→ Paraselene_ joined00:10
← Jak_o_Shadows left00:10
ereslibre_uni → ereslibre00:11
→ Nalwin joined00:12
← JackTheRabbit left00:12
← Paraselene_ left00:15
← Nalwin left00:17
→ Nalwin joined00:17
→ sanity joined00:19
← Nalwin left00:19
→ Twisted_Mentat joined00:21
→ Nalwin joined00:22
→ otay joined00:22
← x|ixti|x left00:22
← ph^_ left00:24
← teukka left00:31
← malumalu left00:31
→ davido joined00:32
→ jumoit joined00:33
← kpreid_ left00:34
← imajes left00:35
← jumoit left00:36
← otay left00:37
→ ngw_ joined00:37
→ ph^_ joined00:39
← ph^_ left00:40
→ ph^_ joined00:40
→ sgh_ joined00:41
← TDJACR left00:41
← Chillance left00:42
→ jmcantrell joined00:46
← giallu left00:47
← pantsman left00:48
kampasky → pasky00:51
→ datechman joined00:51
← loincloth left00:51
psypointer → Melanchemiker00:52
← exygen left00:52
→ lightcap joined00:52
← jmcantrell left00:53
← sgh left00:54
ngw_ → ngw_|away00:55
→ loincloth joined00:55
← eno left00:56
→ TDJACR joined00:57
← aleray left00:57
mentor_ → mentor00:57
ngw_|away → ngw_00:57
← ph^_ left00:57
→ eno joined00:58
← r11t left00:58
← jessep left00:59
→ thiago_h1me joined01:00
thiago_h1me → thiago01:00
← thiago_home left01:01
→ fcuk112 joined01:09
→ schmukkk joined01:09
→ corni_ joined01:11
← tbf left01:11
← tbuehlmann left01:14
→ Piot joined01:16
→ tjc joined01:17
← tjc left01:18
← fcuk112 left01:19
← d0k left01:19
← Piot left01:20
→ Piot joined01:20
← tfarina left01:21
← Piot left01:21
→ cjs joined01:25
← Fissure left01:25
→ jmspeex joined01:27
← corni left01:27
← nitrate left01:30
← j416 left01:31
→ nadavoid joined01:31
→ kpreid_ joined01:32
← _Vi left01:34
← project2501a left01:35
→ _Vi joined01:36
→ FelipeMcMont joined01:39
← jackylee left01:39
← FelipeMcMont left01:39
→ vvsh joined01:44
Melanchemiker → Wuerstlstand01:48
← sanity left01:48
→ sh1mmer joined01:49
→ loinclot_ joined01:49
← mentor left01:51
← loincloth left01:55
← warthog9 left01:59
→ j416 joined02:01
→ s1kx- joined02:01
→ JackTheRabbit joined02:02
→ bugfux joined02:02
← sgh_ left02:03
← corni_ left02:07
→ simgislab joined02:07
→ RaceCondition joined02:08
← RaceCondition left02:08
← JackTheRabbit left02:11
→ madewokherd joined02:12
← S1kx left02:18
← bugfux left02:18
← lightcap left02:22
← Jippi___ left02:26
→ VVSiz_ joined02:27
← davido left02:31
→ dreiss_ joined02:32
ngw_ → ngw_|away02:32
ngw_|away → ngw_02:32
→ munichlinux joined02:33
→ Guest52613 joined02:33
← thiago left02:34
← zackattack left02:36
→ remi joined02:36
→ upul joined02:36
→ yaotti joined02:37
← Xiol left02:38
← spearce left02:43
← VVSiz left02:45
→ agib joined02:45
→ project2501a joined02:48
← Guest52613 left02:50
→ Guest52613 joined02:50
ngw_ → ngw_|away02:51
ngw_|away → ngw_02:53
← Rolenun left02:53
← nadavoid left02:54
← Guest52613 left02:55
→ davido joined02:55
→ moconnor joined02:59
← godsflaw left03:00
→ spearce joined03:00
→ godsflaw joined03:01
← ereslibre left03:01
→ nadavoid joined03:02
→ eletuchy joined03:05
← vvsh left03:05
← spearce left03:11
← yaotti left03:13
→ munichpython joined03:14
← nadavoid left03:14
← munichpython left03:14
→ tedoc2000 joined03:14
→ lightcap joined03:14
← lightcap left03:17
← remi left03:18
→ rchavik joined03:19
← project2501a left03:19
→ cytrinox_ joined03:20
← moconnor left03:21
← cytrinox left03:22
cytrinox_ → cytrinox03:22
← munichlinux left03:22
← schmukkk left03:23
→ schmukkk joined03:24
← tekky left03:27
← mengu_ left03:31
→ ProperNoun joined03:35
← kukks left03:36
← ngw_ left03:36
→ ngw joined03:37
→ chalkbag_ joined03:41
← chalkbag left03:41
→ kazzy__ joined03:43
← LiamH left03:44
← schmukkk left03:44
← kazzy_ left03:48
→ aresnick_ joined03:49
→ dys`` joined03:52
← parasti left03:58
← bentob0x left04:01
→ jepler_ joined04:01
→ hyperair joined04:01
→ eno_ joined04:03
← eno left04:04
← krynnotaur left04:05
← sh1mmer left04:05
→ hipe joined04:07
← dys` left04:07
→ krynnotaur joined04:08
→ moconnor joined04:11
← ProperNoun left04:13
← hipe left04:13
← jepler left04:15
← tedoc2000 left04:15
rchavik hi.. is it possible to undo a push? http://pastie.org/75840404:15
→ lark joined04:15
lark git hooks confuse me, why do they run on the client?04:16
DrNick not all of them do04:16
lark oh really?04:16
my scenerio is repository and production are on the same server04:16
when i push to the repository i want production automatically updated04:16
so i figured creating the post-update hook in the repository hooks would trigger that04:17
but apparently that only triggers on my local machine04:17
which is useless :-/04:17
← ngw left04:19
DrNick lark: man githooks04:20
Gitbot lark: the githooks manpage can be found at http://git.or.cz/man/githooks04:20
← babilen left04:20
lark yeah i read that DrNick04:20
ben_h lark: you probably want post-receive04:21
lark right ben_h i did switch to that04:21
→ ProperNoun joined04:21
ben_h i use that to run commands on the remote end, works well04:21
lark so if i have in my bare repo hooks post-receive with some bash commands i know work04:21
when i push from a remote machine to that bare repo it should execute post-receive?04:21
because that is not the case for me right now04:21
shit would be so cash if i could figure out these hooks04:22
→ JoeConyers joined04:23
→ babilen joined04:24
→ sh1mmer joined04:27
→ bcardarella joined04:28
lark urgh it appears that the hook should be running04:28
if i do a push04:28
← chalkbag_ left04:30
← cloudhead left04:30
→ schmukkk joined04:33
lark oh snap it is running04:36
just put a debug line in04:36
← simgislab left04:36
→ hipe joined04:37
→ Joe_Conyers joined04:41
eno_ → eno04:41
← moconnor left04:42
→ moconnor joined04:45
← Joe_Conyers left04:48
→ fcuk112 joined04:49
lark ah its permissions04:50
error: cannot open .git/FETCH_HEAD: Permission denied04:50
thats a bingo04:50
i was testing my bash script using root04:50
→ Joe_Conyers joined04:50
← Joe_Conyers left04:50
→ xxthink joined04:52
xxthink How to change the port of ssh ?04:53
← datechman left04:53
xxthink I use the following command04:53
git remote add origin git@localhost:example.git04:53
but my ssh port is 222204:53
How to use the git remote add command?04:54
→ jollyroger_ joined04:55
← schmukkk left04:56
jollyroger_ guys, i have a bad problem with git: my laptop froze during a "git push origin master"04:56
← JoeConyers left04:56
jollyroger_ now i cant push to github anymore: http://pastie.org/75843204:57
i guess the problem is this: fatal: object 39c13b0a4c3f549d52fb5dd510cb199116b607ee is corrupted04:57
hwo can i fix this?04:57
→ tekky joined04:58
→ schmukkk joined05:00
← moconnor left05:05
← schmukkk left05:05
j416 jollyroger_: find that object in your backup05:05
and restore a non-broken version of it05:05
it's inside .git/objects/39/05:06
file name is c13b0a4c3f549d52fb5dd510cb199116b607ee05:06
← PuffTheMagic left05:07
lark ben_h: got it working, i'm a nub who didn't realize there were permission errors05:07
ben_h :)05:07
j416 jollyroger_: if the object is packed into a pack, I'm not sure how you can replace it05:07
perhaps you need to unpack your repo, replace, and re-pack05:07
I am not sure.05:07
should be an easier way.05:07
lark okay now git works perfect, <305:08
jollyroger_ j416, there is only one file .git/objects/39/c13b0a4c3f549d52fb5dd510cb199116b607ee in my git-directory05:08
j416 ok, then it's probably packed05:09
hmm05:09
jollyroger_ and it's binary rubbish, so i cant edit it05:09
j416 it's compressed05:09
you shouldn't edit it05:09
xlistener since it was a push is the problem on the client or server?05:11
jollyroger_ xlistener, i assume it's on the client: http://pastie.org/758432j41605:12
j416, maybe I could undo my last commit - do my editing, then re-commit it?05:12
the broken file is from 30 minutes ago05:12
could this do the trick?05:12
j416 well, you need to get that broken object fixed somehow05:12
i don't think so05:12
git won't re-write it if it exists05:12
you need to find a way to replace it05:12
if it weren't packed, it would have been easy05:13
(just drop the file in there)05:13
but I'm not sure how to do it if it's packed05:13
→ r11t joined05:13
← __iron left05:13
jollyroger_ hmm, ok, apparently this problem is FAR beyond my git knowlegde05:14
F**K05:14
← Davey left05:15
j416 jollyroger_: you have a backup right?05:15
don't lose that05:15
xlistener git reset --soft HEAD~1 should put your branch before the bad commit & leave the changes in your working tree05:15
jollyroger_ yeah, i got multiple backups...:-)05:15
xlistener and your pastie didn't work 404 not found05:15
← Bass10 left05:15
j416 xlistener: that won't fix his broken object, though05:16
will it?05:16
→ datechman joined05:16
← fcuk112 left05:16
jollyroger_ xlistener, ah, sorry: http://pastie.org/75843205:16
xlistener git gc should delete it right?05:16
j416 no it won't05:16
git gc won't work05:16
meanburrito920_ jollyroger_: I'd reset your head to an earlier version, pull from github, then re-apply your changes and push05:16
j416 hmm05:17
an easy way would be to make a new clone from github, make patches from your new changes and apply them on top of that05:17
but there should be an even easier way05:17
hang on05:17
jollyroger_ j416, yes, that's probably the easiest way for me. but for the sake of arguments (and since i have multiple backups) i could meanburrito920_ / xlistener way05:18
xlistener gc deletes unreferenced commits right? or am i thinking of something else?05:18
meanburrito920_ xlistener: you are correct, but the commit is still referenced05:19
j416 jollyroger_: run this:05:19
git ls-tree -r HEAD | grep c13b0a4c3f549d52fb5dd510cb199116b607ee05:19
xlistener not if the branch is reset to an earlier point05:19
j416 do you get a file name?05:19
← bcardarella left05:19
j416 try to find what file c13b0a4c3f549d52fb5dd510cb199116b607ee is05:19
then you might be able to re-write the object05:19
if the file is not from the latest commit, you need to loop through previous commits05:20
you can also do: git show c13b0a4c3f549d52fb5dd510cb199116b607ee05:20
in your backup05:20
to show the contents, to guess what file it is05:20
jollyroger_ j416, there it is: 100644 blob 39c13b0a4c3f549d52fb5dd510cb199116b607ee config/deploy.rb05:21
j416 ok!05:21
try this in your broken repo then: git hash-object -w config/deploy.rb05:21
does that work?05:21
→ rachman joined05:21
jollyroger_ git hash-object -w config/deploy.rb gives me:05:21
39c13b0a4c3f549d52fb5dd510cb199116b607ee05:21
j416 it should return that same sha105:21
yeah05:21
and it should have written that file05:21
try: git show c13b0a4c3f549d52fb5dd510cb199116b607ee05:22
oops05:22
git show 39c13b0a4c3f549d52fb5dd510cb199116b607ee05:22
sorry :)05:22
(prepend 39 to the other sha1s i mentioned, of course)05:22
i just pasted what was in clipboard.05:22
my bad.05:22
if "git show 39c13b0a4c3f549d52fb5dd510cb199116b607ee" works, it's probably fixed05:23
jollyroger_ no problem, ok, wait a sec, I'll do it again for the sha05:23
j416 try "git fsck" next05:23
oh05:25
hash-object won't write if it's already there05:25
hmm05:25
jollyroger_ j416, ok, a git hash-object -w config/deploy.rb doesnt replace the file - so I'll just delete it05:25
j416 it's not the file that's the problem05:25
it's the object05:25
the file is ok05:26
jollyroger_ ah, it worked!!05:26
j416 wow05:26
what did you do?05:26
ah, you resetted to previous commit?05:27
jollyroger_ j416, 1.) rm .git/objects/39/c13b0a4c3f549d52fb5dd510cb199116b607ee then 2.) git hash-object -w config/deploy.rb05:27
j416 oh05:27
of course05:27
i thought it was packed!05:27
jollyroger_ i just pushed successfully to github05:27
j416 :)05:27
then yeah05:27
that'll work fine05:27
jollyroger_ j416, thanks a ton, seriously05:27
→ loincloth joined05:28
→ project2501a joined05:28
j416 oh05:28
→ bcardarella joined05:28
j416 it seems git uses file timestamps if there are multiple objects with the same name05:28
so next time, if the corrupted object is in the pack05:29
→ munichlinux joined05:29
j416 maybe, just maybe, you can just copy the non-corrupted object into the objects dir and repack05:29
haven't tried though05:29
jollyroger_ I'll keep that in mind...:-)05:29
← reprore left05:30
→ ealden joined05:30
jollyroger_ so, it's half past 6 in the morning and i am going to bed - thanks for the help guys....05:30
← jollyroger_ left05:31
← ealden left05:39
→ loinclo__ joined05:40
← rchavik left05:40
→ t0rc joined05:41
← loinclot_ left05:43
→ sunil joined05:43
→ exygen joined05:47
→ jfxberns joined05:50
→ lightcap joined05:50
j416 whoa05:52
new things05:52
git-notes, interesting05:52
j416 just discovered05:52
→ tedoc2000 joined05:53
← lightcap left05:54
meanburrito920_ git-notes?05:54
meanburrito920_ heads to google05:54
← loincloth left05:54
meanburrito920_ that must be in a new version of git; i found the man page online, but its not in 1.6.305:56
← bcardarella left05:58
j416 meanburrito920_: 1.6.606:00
meanburrito920_: http://www.kernel.org/pub/software/scm/git/docs/RelNotes-1.6.6.txt06:00
meanburrito920_ ah06:00
j416 not entirely sure what I can use it for though06:00
hm06:00
← aziz left06:00
meanburrito920_ hm, 1.6.6 isnt in the ubuntu repos yet...06:02
← tekky left06:03
→ impulze_ joined06:04
← aresnick_ left06:04
→ impulze__ joined06:04
j416 meanburrito920_: well, it was released not even a week ago06:04
december 2306:04
→ loincloth joined06:06
→ tekky joined06:10
→ covidiu joined06:10
← loinclo__ left06:10
j416 I'm still trying to think of a real-world use-case for git-notes06:13
covidiu Hello. Full disclosure: I'm a git newbie. I want to host a central git repository and allow other developers to push into it. How do I do that without allowing them to do something destructive (like login via SSH and rm the whole thing). Just point me in the right direction, I can read manuals. Thanks.06:13
j416 hm06:13
covidiu: git has a restricted shell you can use06:13
also check gitolite, a project for permission management06:13
covidiu: man git-shell06:14
Gitbot covidiu: the git-shell manpage can be found at http://git.or.cz/man/git-shell06:14
covidiu Alright, thanks a lot.06:14
j416 covidiu: http://github.com/sitaramc/gitolite06:14
I have used neither, I just do solo-development.06:14
but gitolite is good they say.06:15
→ dirk joined06:15
→ schmukkk joined06:18
→ gma joined06:22
← s0ber left06:23
← ProperNoun left06:25
→ ProperNoun joined06:26
← sh1mmer left06:27
docgnome → dkerschner06:30
dkerschner → docgnome06:31
← lark left06:36
← tedoc2000 left06:39
← t0rc left06:43
→ jceb joined06:43
← xxthink left06:44
→ rudi_s joined06:45
← traviscline left06:46
← gma left06:47
→ traviscline joined06:47
← schmukkk left06:48
← traviscline left06:49
← covidiu left06:52
← EricInBNE left06:53
← j416 left06:54
→ lhunath joined06:54
lhunath is it possible to set up git such that git push will push to multiple remotes instead of just one?06:55
← jjore left06:56
gitster yes.06:56
rachman → rchavik06:56
→ amerine joined06:57
lhunath do reveal how or where I can find out how.06:57
Ilari lhunath: Just put multiple urls for remote?06:57
← exygen left06:58
gitster We should document that you can have more than one remote.<name>.pushurl for the same <name>, I guess.06:59
← rudi_s_ left07:01
lhunath ah. neat.07:03
probably; the documentation uses singular; should probably indicate that its a 1..n07:04
← hyperair left07:04
→ hyperair joined07:04
← hyperair left07:04
← dreiss_ left07:05
gitster The doc was written back in the day when we allowed only one. When Linus added the push-to-multiple-places nobody bothered to update it.07:05
→ hyperair joined07:07
lhunath though how does that work for pull? One remote can have different masters, then07:09
gitster that is why I said you can have more than one PUSHurl ;-)07:09
lhunath guess it does a git pull or each, and there's no remote branch for each?07:09
gitster The multi-push is about publishing your work to more than one places.07:10
lhunath indeed.07:10
gitster and that is what you originally asked about, is it not?07:11
lhunath it is. but when I think of it; a multi-pull makes sense, too, then.07:11
and I recon that's exactly what it does when I use multiple url's instead of multiple pushUrls.07:12
→ maxter joined07:12
gitster it doesn't to me at all.07:12
lhunath if I have three places I publish my commits to; and either of those gets new commits from someone else; I want to pull those in, as well.07:12
so basically, what I want would be like having threee remotes, group em into one and use that group to pull and push.07:13
gitster I used the word "publish" in a very strict sense. It is for _you_ to _show_ your work to others. You are talking about something completetely unrelated---you are talking about a shared repository where others would also push into.07:14
In that context, multi-push doesn't make sense, either, let alone multi-pull.07:14
lhunath why is that?07:14
can you not merge with each URL?07:15
Ilari lhunath: Well, you can have multiple remotes and then do 'git remote update' to fetch them all.07:16
lhunath right; but then I still haven't merged them into master. I recon that's asking a bit too much as each of those merges might potentially fail to complete successfuly.07:17
gitster Think what it means to have more than one _shared_ repository, in a project that works on a shared single goal.07:17
In a "publishing" context, having more than one makes sense---it will let others to fetch from any one of them, the closest one (or the one that happens to be up), and guarantee that they will get the latest.07:18
It gives you the benefit of distribution and robustness.07:18
Ilari lhunath: And also, such merge could wind to be octopus merge. If one manages to make one, they look cool in gitk, but that's about it.07:18
gitster It doesn't make sense to distribute _shared_ repository that way at all.07:19
→ sis joined07:19
gitster Your contributors may start from the tip of one repository, work on it, and push back, without first pulling from the other one.07:19
lhunath indeed.07:20
gitster They cannot push to the other one before merging from the other one, which means they have to pull from both (or "all" if you have more than two).07:20
lhunath I fail to see why that is any different than push/pulling from different people directly.07:20
gitster That is not giving them robustness or fallback.07:20
← davido left07:20
gitster Exactly. What you are failing to see is that then that is not even a central shared repository anymore. There is no central place for people to meet.07:21
→ jjore joined07:21
lhunath true. but I have no need for a central shared repository. isn't that the strength of GIT?07:21
every once in a while I'll fetch all sources, merge and push to all three to bring them up to date.07:22
in my particular case; I would like to publish my code to github, codaset and gitorious; each of those places may have different contributers.07:23
gitster You can abuse a tool in any way you want. A larger project _can_ have more than one shared repositories, and have different _sets_ of people working on different parts of project. Each subset of people would have _a_ shared repository for their part of the system (subsystem repository) to exchange their work. Anything more complex is simply something you would do "because you can", not "because it makes sense." Think about it.07:23
← patcito left07:24
→ drizzd joined07:24
gitster In other words, you _can_ arrange to cause anarchy. It doesn't necessarily mean it would be beneficial to developers and users. They wouldn't know which one is the most recent and authoritative (unless of course you arrange the repositories in a certain way to say "this one is to work on this area" and let that arrangement known).07:25
lhunath it doesn't make sense if you considder all contributers to be under a common authority; but if they are part of a social coding site, it changes.07:26
gitster And once you start _partitioning_ the system and designate different repositories for different _purposes_, then it wouldn't make sense for you (or anybody else) to push the same thing to all these different repositories with different purposes anymore.07:27
← nighthwk1 left07:28
lhunath I find it makes plenty of sense without partitioning, too. people from github will just pull and push to github; people from codaset will push/pull to codaset. every now and then I'll sync them up and tag releases which will be shared amoungst them.07:28
the key being that each of these shared repositories is the meeting point for a different group of people.07:29
truthfully; it would be more optimal if they used a common meeting point; one central shared repository. but that'snot how these social coding sites work.07:30
gitster For what purpose do you partition your contributors into these different sites? And don't tell me that "some have accounts on only one service". I am asking what benefit the resulting code and history would get by allowing such an arrangement.07:30
If one person who has only account on github got your tip and worked on one thing, while another one gets the same from gitorious and made an overlapping work, how would they coordinate their work?07:31
You'll make yourself the bottleneck that way.07:31
→ xxthink joined07:32
xxthink questions on the git init usage07:32
lhunath that is true.07:32
gitster Now, I think the common workflow at least at github is not "shared repository everybody pushes into", but is more like "you publish, others fork and push into theirs not into yours, and you pull".07:32
xxthink I have create a git repo with the totorisGit init07:33
gitster So as long as you do not allow people to push into _your_ repository, what you say makes sort of sense, but then we are talking about "publishing".07:33
lhunath correct. but I'm not completely certain the others work the same way.07:34
xxthink But only the .git directory is created07:34
I do develop and commit many times07:34
Now I want to move this repo to linux07:35
but when I use totorisGIT to clone the repo in windows07:35
It always tell me "it is not a git repo"07:35
does not appear to be a git repository07:36
But if I create a new repository in linux with the command07:36
git --bare init07:36
I can clone this repository in windows07:37
← jceb left07:37
xxthink How can I change my .git repository to a TRUE repository?07:37
zomg Define "TRUE" repository07:39
xxthink zomg: then how to transfer my log in .git to this new repository?07:39
→ patcito joined07:40
gitster I don't know how well Tortoise works (or how it works in general), but you (1) created what Tortoise thinks is a git repository on your Windows box and committed things there; (2) created an unrelated bare repository on a different box running Linux using "git --bare init", right?07:41
If so, the logical next step would be (3) push from the Tortoise created repository to that bare repository on the Linux box.07:42
Am I making any sense?07:42
xxthink gitster: yes, you are right07:44
I will try step (3)07:44
→ loinclot_ joined07:44
← drizzd left07:44
gitster Let me remind you that I don't know Tortoise at all, so regrettably I wouldn't be of any help, but hopefully you will find how to push from it to other places.07:45
xxthink gitster: one problem07:45
how can I push to this new repository07:45
gitster Sorry, I don't do Windows.07:46
xxthink ok07:46
let me try07:46
← patcito left07:46
← Fullmoon left07:49
← loincloth left07:59
Ilari xxthink: Using ssh:// url probably.08:02
← lhunath left08:02
→ ph^_ joined08:03
→ patcito joined08:04
floppyears do you guys use git to keep track of documents and personal files?08:06
if so do you have any tips or suggestions?08:06
gitster "don't panic"? ;-)08:07
← hyperair left08:08
← patcito left08:08
→ hyperair joined08:08
← datechman left08:10
→ zackattack joined08:10
→ datechman joined08:10
agib is there anyway to specify directories with gist.github.com?08:12
← mbernstein left08:12
← xlistener left08:13
← hobodave left08:15
← zackattack left08:16
← sis left08:17
← hyperair left08:26
→ hyperair joined08:26
→ droden joined08:26
→ galderz joined08:27
→ Axius joined08:27
← amerine left08:29
← QshelTier left08:29
→ ixti joined08:30
→ __iron joined08:30
→ mithro joined08:31
→ SpookyET joined08:31
SpookyET Hi. Is it possible to undo a git checkout?08:32
→ patcito joined08:32
jtbandes SpookyET: git checkout master? (no promises, I don't know git very well)08:32
→ incredible joined08:33
→ schmukkk joined08:35
SpookyET jtbandes: i check out and overwrote my files08:36
← tekky left08:36
← patcito left08:38
← loinclot_ left08:38
→ loincloth joined08:38
← hyperair left08:39
→ hyperair joined08:39
→ kumbayo joined08:40
→ patcito joined08:43
wereHamster git doesn't overwrite files unless there is a way to recover those, or if you specify --force or --hard08:44
SpookyET: if you checked out a branch and it overwrote your files, you can check out the old branch to get the files back08:45
I think 'git checkout -' does the right thing (tm)08:46
← patcito left08:46
← aussiegeek left08:49
→ razvand joined08:50
→ kmartiin joined08:51
→ tekky joined08:52
→ j416 joined08:53
← Twisted_Mentat left08:53
→ s0ber joined08:55
← s0ber left08:55
→ Fissure joined08:57
← schmukkk left08:57
→ maleknet joined09:05
→ solofight joined09:06
→ Alfonzo joined09:07
← solofight left09:07
→ solofight joined09:08
← datechman left09:09
→ tha joined09:10
→ martiin joined09:10
→ yann joined09:10
← kmartiin left09:10
→ aleray joined09:10
→ patcito joined09:10
← Axius left09:12
← tha left09:13
→ fcuk112 joined09:13
← maleknet left09:14
xxthink Ilari: when I use git clone ssh://url09:14
← patcito left09:14
xxthink I can clone the repos09:14
But on other pc, I can't clone09:15
it says the remote repo is not a git repo09:15
→ kmartiin joined09:16
→ JackTheRabbit joined09:16
Ilari xxthink: What's the exact error message?09:16
→ martink joined09:17
← martiin left09:17
xxthink the remote doesn't appear to be a git repository09:17
← kmartiin left09:17
Ilari xxthink: Huh? I don't think git can print error like that.09:18
xxthink test_repo doesn't appear to be a git repository09:18
← meanburrito920_ left09:18
xxthink test_repo is my repository on my server09:19
solofight how do i stop the rebase which is already running ?09:19
doener xxthink: is "url" using a path relative to $HOME? e.g. ssh://host/~/test_repo?09:19
solofight: rebase --abort09:19
Ilari xxthink: Exact error message would show what exactly it is trying to access.09:19
→ bambule joined09:20
Ilari xxthink: E.g. "fatal: '/nonexistent' does not appear to be a git repository"09:20
xxthink fatal: '/work/test' does not appear to be a git repository09:20
← krynnotaur left09:21
← agib left09:21
solofight doener++09:21
Ilari xxthink: Note the '/' in start, it means that path is relative to filesystem root.09:22
→ rachman joined09:23
Ilari xxthink: Just try 'ls /work/test' on remote side. Will likely complain.09:23
xxthink Ilari: thank you09:23
the path is not correct09:23
I think09:23
Ilari xxthink: Maybe 'ssh://user@host/~/work/test'?09:24
solofight doener, how do i perform a rebase on last five commits ?09:24
git rebase -i master isnt doing that09:24
it asks me for a rebase message alone09:24
wereHamster rebase -i HEAD~509:24
solofight not able to select which to squash and which to leave as seperate commit09:25
thank you09:25
→ loinclot_ joined09:26
→ jumoit joined09:27
xxthink Ilari: yes, it work09:31
thank you very much for your help09:31
→ Chillance joined09:31
← martink left09:31
→ kmartiin joined09:31
← loincloth left09:32
← madewokherd left09:35
→ tha joined09:36
← rchavik left09:37
← kmartiin left09:39
→ kmartiin joined09:39
→ v0n joined09:41
solofight i have 6 commits and i want to make them 2,09:43
→ patcito joined09:43
solofight ex:1,2,3,4,5,609:43
i want to09:43
s 1, s2, s3 , s4, pick5, s609:43
← kazzy__ left09:44
solofight will s1,2,3,4,6 become a commit09:44
→ Axius joined09:44
solofight and 5 alone be a seperate commit ?09:44
or how do i do that ?09:44
← kmartiin left09:44
→ kmartiin joined09:44
Jaxoreth solofight: man git-rebase09:45
Gitbot solofight: the git-rebase manpage can be found at http://git.or.cz/man/git-rebase09:45
← patcito left09:46
solofight thank you09:46
Jaxoreth git rebase -i HEAD~609:46
← xxthink left09:46
Jaxoreth pick 1, s2, s3, s4, s6, pick 509:47
→ ngw joined09:48
rachman → rchavik09:48
→ martiin joined09:50
← kmartiin left09:50
→ intripoon joined09:52
VVSiz_ → VVSiz09:55
← JackTheRabbit left09:55
Ilari Hmm... Why hook output is sent on stderr instead of sideband #2? Breaks hook output if transport can't carry stderr.09:56
→ tibra joined09:57
← SpookyET left09:58
→ patcito joined09:59
← patcito left10:01
→ meanburrito920_ joined10:01
→ sgh joined10:01
← bambule left10:02
→ s0ber joined10:03
ngw → ngw|away10:03
Ilari Apparently because there's no sideband muxing in downstream direction when pushing...10:04
← Axius left10:06
Ilari And looks like hook output doesn't work in smart http push either...10:07
→ Geralt joined10:09
solofight Jaxoreth, how do i cut and paste in this sad vi editor /10:10
?10:10
can i use emacs as my default editor ?10:10
← AkumaStreak left10:10
Ilari solofight: 'export VISUAL=emacs'?10:11
→ [Po]lentino joined10:11
solofight Ilari, where i need to type that ?10:11
in config?10:11
Ilari solofight: Bash.10:11
← martiin left10:11
Ilari solofight: (or to bash initialization files)10:12
→ Paraselene_ joined10:12
← j416 left10:13
→ mathieu joined10:14
mathieu Hi, I see there are some patches floating around for pull/merge that does fast-forward only, but is there anything already included in a git release?10:15
→ marc_m joined10:15
marc_m → marc_kdab10:16
→ pantsman joined10:16
→ khmarbaise joined10:17
← sgh left10:18
Ilari mathieu: AFAIK, in v1.6.610:18
mathieu Ilari, that would be great -- it makes some workflows more robust10:19
← jhushan left10:19
→ jhushan joined10:19
← e9 left10:19
mathieu it's indeed mentioned in the changelog10:20
→ cedricv joined10:20
← cedricv left10:21
→ cedricv joined10:22
→ patcito joined10:22
← mathieu left10:23
← r11t left10:24
← _Vi left10:25
→ mbroeker joined10:25
← patcito left10:26
← aleray left10:28
← mbroeker left10:33
← solofight left10:33
→ gebi joined10:35
← pantsman left10:36
→ mathieu joined10:37
→ HanzZ joined10:37
HanzZ hi, how to commit to new branch?10:38
meanburrito920_ switch to the branch using 'git branch <branchname>'10:40
and then commit like normal10:40
git commit10:40
← laj_ left10:41
HanzZ thanks10:42
→ mbroeker joined10:46
Ilari HanzZ: 'git checkout -b <newbranchname>' and then 'git commit'.10:49
→ laj joined10:49
→ pantsman joined10:53
→ patcito joined10:56
← patcito left10:58
ngw|away → ngw11:01
← cjs left11:03
→ cjs joined11:04
→ hISLImr]R joined11:04
← cbreak left11:04
← cjs left11:04
← ankit9 left11:04
hISLImr]R DCC SEND "поÉпоɦʞɔпɟ¦ÊžÐ¿Ð¾É¦ÊžÉ”пɟɔпɟ" 0 0 011:04
← hISLImr]R left11:04
→ cbreak joined11:04
→ ankit9 joined11:05
→ cjs joined11:05
[Po]lentino → [Po]lentino|away11:05
← Weasel[DK] left11:05
→ JackTheRabbit joined11:05
→ Weasel[DK] joined11:05
→ tbuehlmann joined11:09
→ malumalu joined11:10
← pantsman left11:12
→ d0k joined11:13
→ jlilly joined11:13
← laj left11:13
→ loincloth joined11:16
→ patcito joined11:18
→ sgh joined11:19
→ sitaram joined11:19
← patcito left11:21
← sgh left11:21
→ Wazzzaaa joined11:22
← loinclot_ left11:22
ngw → ngw|away11:22
← upul left11:23
← khmarbaise left11:25
→ ereslibre joined11:27
← Geralt left11:31
→ Geralt joined11:32
← Tommy[D] left11:32
→ pantsman joined11:33
ngw|away → ngw11:36
← Lisimba left11:36
→ patcito joined11:42
[Po]lentino|away → [Po]lentino11:43
← patcito left11:45
← dirk left11:47
← pantsman left11:49
→ jelmer joined11:51
→ iwein joined11:54
→ brizly joined11:57
← fcuk112 left11:57
→ ehabkost joined11:58
→ JackTheRabbt joined11:58
← JackTheRabbit left11:59
→ jelmer__ joined12:00
← jelmer left12:00
→ loinclot_ joined12:02
→ Sky[x] joined12:03
→ ehabkost_ joined12:03
→ ph^__ joined12:04
→ dondani joined12:04
dondani when I add a remote to my git project, I have to specify the full path to my remote: git remote add origin [email@hidden.address]12:04
→ ph^___ joined12:04
dondani is there a way to omit the '/home/repouser/repositories' part?12:04
→ aresnick joined12:04
wereHamster if you use 'user@host:path/repo.git' it will use the path relative to your HOME12:05
→ parasti joined12:05
dondani good to know, thx12:05
wereHamster also, it looks like you are using gitosis/gitolite, is that right?12:05
dondani indefero, driving me crazy12:06
wereHamster don't know about that one12:06
dondani is there a way to omit that last 'path/', so that /home/repouser/repositories is the default path for all repos?12:06
→ MacGyverNL joined12:06
dondani it's a php app, kind of like google code12:07
zomg ah talk about convenience :P Added an alias for git stash && git svn rebase && git stash pop12:07
wereHamster gitosis/gitolite do that by default (if you clone git@host:project.git they look in $HOME/repositories/project.git_12:07
dondani will have to look into gitosis12:07
does gitosis have a web ui?12:07
wereHamster I'd expect indefero do the same, but I've never used it12:07
dondani: use gitolite, nto gitosis!12:07
no12:08
not built into it, but you can use gitweb/cgit12:08
dondani don't know the difference between the two...12:08
ok12:08
i'm on a shared hosting, I don't know if I'll run into much trouble configuring it12:09
thanks anyway12:09
← mathieu left12:09
wereHamster dondani: can you set up a new user on that host?12:09
dondani yup12:10
wereHamster then you shold be fine. gitolite only requires that and ssh access to the server12:10
dondani cool12:10
← ehabkost left12:13
cjs is there a way to merge two commits into one?12:13
wereHamster such operation is usually called 'to squash two commits', and that can be done with rebase -i12:13
cjs Hm, I can squash only interactively?12:14
→ vbgunz__ joined12:15
wereHamster how else would git know which commits to squash?12:15
vbgunz__ → vbgunz12:15
cjs Just specify them on the command line.12:15
I dunno, interactive feels too much like talking to a person. I don't like that feeling. :-)12:15
wereHamster I guess that's possible with plumbing commands, but I'd stronly recommend rebase -i12:16
→ WinterWeaver joined12:16
cjs Yeah, I'm not going to have a go at plumbing right now. I've found the bit in the manpage; I'll give it a go. Thanks.12:16
→ DavidKlein joined12:17
← ph^_ left12:18
← loincloth left12:18
← ph^ left12:18
← jlilly left12:21
→ khmarbaise joined12:22
→ loincloth joined12:23
← dondani left12:23
→ sde_antrix1 joined12:25
iwein I'm fighting with getting a combo of git-svn and git-submodule to play nice. Doesn't seem that was meant to work...12:26
I'd be interested to see some examples if you have them12:26
→ Tommy[D] joined12:26
← Wazzzaaa left12:27
wereHamster iwein: you can't directly reference a svn repo as a git submodule, but that's about the only restriction I'm aware of12:28
iwein whereHamster, ah, so I should be able to reference a git repo as a submodule and commit the content back to my svn repo?12:28
wereHamster first you have to create a git clone of the svn repo (using git-svn), then you can reference that git repo as a submodule in another git prepo12:29
iwein I'm trying to reference a git repo into a clone of a svn repo (so the other way around)12:31
→ jlilly joined12:32
iwein I did git svn clone A, then git submodule add B A/B, then git submodule update and some changes.12:32
finally I can't get git svn dcommit to work anymore.12:32
← babilen left12:32
wereHamster I don't think that will work well, because you can't commit that back to svn12:32
→ thiago_home joined12:32
iwein yup, that's my problem :)12:33
→ babilen joined12:33
← LRN left12:33
wereHamster I mean, if you commit that back, how will svn clients know that there is a submodule, and mor importantly, know how to clone the git repository?12:33
→ LRN joined12:34
iwein in fact for my use case that's not really a problem. but I can see that the tooling would need more information in order to decide what information I want to ignore.12:34
s/tooling/git-svn/12:34
→ onigiri_ joined12:35
iwein basically the only thing I want is to sync the head of a git repo with a sub directory in an svn repo (two ways)12:35
wereHamster you don't need submodules for that ...12:36
→ Lisimba joined12:36
iwein it does seem like overkill..12:36
→ vdidel_ joined12:38
← loinclot_ left12:39
cjs Gah. I'm lost in a maze of twisty little commits, all alike.12:39
← v0n left12:39
← vdidel_ left12:39
→ v0n joined12:39
iwein the main problem I'm having is that I don't know how to prune of the parent directories of the svn repo before syncing it with my git repo, I thought git-svn and git-submodule would give me that and some other niceties, but it's been a time trap so far.12:39
wereHamster iwein: from what I understand you want a git-svn clone of a subdirectory of g svn project, right?12:40
→ cloudhead joined12:41
iwein whereHamster, I think so, but I'm not 100% certain on the terminology. I need to commit the cloned subdir back to a separate git repo12:42
in fact it's already there12:42
I'd use svn:externals if this was svn12:43
wereHamster let's start with what you have and what you are trying ti achieve. Obviously you have an existing svn repository with a project in it.12:44
thiago_home there's no equivalent to svn:externals in git-svn12:44
the Git equivalent is similar, but not exactly equal12:44
and it won't work with git-svn12:44
→ d2dchat joined12:44
iwein yes, let's call that svn repo A. then I also have an existing git repo B. I want B to be in a subdir of A and commit it's contents to both A and B.12:45
somewhat like this: http://panthersoftware.com/articles/view/3/svn-s-svn-externals-to-git-s-submodule-for-rails-plugins12:46
thiago_home you can have a Git repo inside a SVN repo12:46
wereHamster iwein: in such setup, where code is both in a svn repo and git repo, you have to decide which repository is canonical. I strongly recommend making the svn repo canonical12:46
thiago_home if such dir is part of the larger SVN repo, then the git checkout will have .svn dirs12:46
you will want to ignore them12:47
that's all12:47
iwein ok, so I ignore .svn in git and .git in svn?12:47
and when I have a change, I do git commit, git push, svn ci.12:48
I still don't get it do I...12:49
→ trochala joined12:49
thiago_home that's correct12:49
except if the svn commit fails, then you need to svn up12:49
in which case your git commit will have been wrong12:49
→ Bombe joined12:50
Bombe Good morning.12:50
iwein sure, but you can dance around for a bit and then it will be ok12:50
Bombe Oh, did I arrive at the right time for the daily naked dancing?12:50
iwein the problem is now that there are other developers that need to check this out. It would be nice if they also committed their changes to both git and svn.12:50
@Bombe, that will be at 1 GMT be patient.12:51
Bombe Ah, okay.12:51
wereHamster iwein: you need to decide which repository is the main one (and make the other one a mirror fo the first one)12:51
everything else (commiting to both git and svn repos) just asks for trouble12:51
iwein so changes flow in one direction only?12:52
that makes sense12:52
but it also makes me the sync monkey :(12:52
wereHamster as said earlier, I strongly recommend that you commit to the svn repo, and make the git repo a git-svn clone12:52
thiago_home iwein: you can hire a cronjob to do the syncing12:52
its time is much less expensive than yours12:53
:-)12:53
→ bigmack83 joined12:53
iwein heh, I'll see if I can find new and inventive ways to screw this up. wereHamster, thiago_home: hanks for the input!12:56
→ patcito joined12:56
← aresnick left12:56
← patcito left12:58
← mithro left12:59
→ davido joined13:00
← kumbayo left13:03
→ aziz joined13:03
← kpreid_ left13:05
→ remi joined13:05
→ loinclot_ joined13:05
→ patcito joined13:06
← tibra left13:06
→ timj joined13:07
← patcito left13:10
← jelmer__ left13:10
→ beni- joined13:13
tbuehlmann I'm writing a short thesis about version control systems and could need some pictures of branching, committing changes and things. do you have any recommendations for those?13:13
thiago_home pictures as in images?13:13
cbreak http://eagain.net/articles/git-for-computer-scientists/13:14
you can make them yourself with graphviz though13:15
tbuehlmann pictures as in images, yes :)13:15
→ GNUtoo joined13:15
cbreak (this is a git alias): graphviz = "!f() { echo 'digraph git {\n node [shape="record"];' ; git log --pretty='format: \"%h\"[label=\"{%ci|%s}\"]' \"$@\" | sed 's/|\\([^}]\\{0,32\\}\\)[^}]*/|\\1/g' ; git log --pretty='format: %h -> { %p }' \"$@\" | sed 's/[0-9a-f][0-9a-f]*/\"&\"/g'; echo '}'; }; f"13:15
it's not perfect though, but it works okish :)13:15
GNUtoo hi, is there a way to push only certain commits without using the following tricks:13:15
tbuehlmann ah, thanks so far :)13:16
cbreak is the trick to make a temporary branch? :)13:16
thiago_home GNUtoo: when you push, you push a commit and all its parents to a given branch13:16
GNUtoo: there's no way around it13:16
GNUtoo ok thanks13:16
so I'll avoid creating the commit and patch back after13:16
yes there are several tricks:13:17
cbreak better:13:17
make a branch for yourself13:17
GNUtoo rebase -i13:17
cbreak and only cherry-pick or merge to the master (the one you want to push)13:17
what you want to push13:17
GNUtoo ok I wanted to avoid that but at the end I'll do that13:17
cbreak you can branch off with git checkout -b mybranch13:17
and then rebase -i the master13:17
GNUtoo thanks a lot13:18
cbreak well, it's sort of clear that you can not push anything without pushing the ancestors :)13:18
(because the ancestors are part of the history, and history can never change in git)13:19
← loincloth left13:20
→ patcito joined13:20
← patcito left13:23
GNUtoo ok thanks a lot13:23
← mm_work left13:24
← cedricv left13:24
← rudi_s left13:25
→ Axius joined13:25
← malumalu left13:26
← Axius left13:26
→ malumalu joined13:26
← JackTheRabbt left13:26
[Po]lentino → [Po]lentino|away13:27
← hummingkid left13:30
→ Axius joined13:31
← ixti left13:32
← Axius left13:32
→ Axius joined13:33
→ giallu joined13:33
→ bentob0x joined13:34
→ patcito joined13:34
→ loincloth joined13:35
← Axius left13:35
→ Piot joined13:36
← patcito left13:37
→ Pupeno joined13:37
← loinclot_ left13:39
← munichlinux left13:40
→ tibra joined13:41
→ munichlinux joined13:41
→ cedricv joined13:41
→ kukks joined13:41
← cedricv left13:42
← munichlinux left13:42
→ ayaz joined13:43
→ yhager joined13:44
→ cedricv joined13:44
→ khinester joined13:44
khinester hello13:44
→ ereslibre_laptop joined13:45
Piot hi13:46
→ ereslibre_desk joined13:46
← ereslibre left13:46
← yhager_ left13:48
← beni- left13:48
→ beni- joined13:48
→ imajes joined13:49
← orafu left13:53
→ orafu joined13:53
→ tvw joined13:54
khinester i get an error: Untracked working tree file when i try to checkout a devel branch13:55
git status currently states 'nothing to commit ...'13:55
wereHamster delete or commit the file13:56
← Nalwin left13:56
khinester i did update the .gitignore not to include these files13:56
← cedricv left13:57
cbreak did you delete them?13:57
if you didn't they are still tracked13:57
git doesn't ever ignore tracked files13:57
→ patcito joined13:57
→ mm_work joined13:58
khinester ok, so i'll git rm tools/compile/* where tools/compile is the files i don't want to track13:58
wereHamster and then commit13:59
← patcito left13:59
→ gnutoo_ joined14:00
→ mbernstein joined14:00
SethX hello hello14:00
i'm still stuck on my detatched head issue14:01
i had the guy redownload the entire repo14:01
← GNUtoo left14:01
SethX and its back to a detatched head14:01
khinester i get this tools/smarty/compile/%%0F^0FB^0... ' did not match any files14:01
SethX with no loca branch for the detatched head14:01
teuf what does he do after cloning it?14:01
SethX nothing14:01
just swear i guess14:01
→ JackTheRabbit joined14:02
cbreak does the remote have a detached head?14:02
→ Nalwin joined14:02
cbreak (is that even possible?)14:02
SethX i wouldnt know how...14:02
khinester so i am unable to checkout the branch14:02
Piot fsck?14:02
SethX lemme check that out14:02
← Nalwin left14:03
→ Nalwin joined14:04
Piot git fsck --full14:04
→ EricInBNE joined14:04
← Nalwin left14:04
→ corni joined14:04
SethX Piot: that to me§?14:04
→ Aldwuin joined14:05
Piot just wanted to know if you tried it :)14:05
→ sis joined14:07
← yhager left14:07
→ yhager joined14:07
SethX srv:/home/gitjail/repositories/aurora.git# git branch14:07
* master14:07
the remote is on master14:07
← Aldwuin left14:07
Wuerstlstand → Wurschtlprater14:07
→ munichlinux joined14:08
cbreak and if you clone, you aren't on master?14:08
→ Aldwuin joined14:08
SethX no14:09
i just cloned14:09
and git branch outputs "no branch"14:09
→ keystr0k joined14:09
→ jelmer joined14:10
→ segher_ joined14:11
parasti SethX: it must have printed a message when it failed to create the branch14:11
SethX: it = git clone14:12
← agile left14:12
SethX if i do, its long gone, since it takes some 4 hours to clone it over a hsdpa connection14:12
→ DeFirence joined14:12
SethX anyhow ill look into it some more14:12
DeFirence hi everyone, git.exe format-patch -o "C:\Documents and Settings\Windows XP\My Documents\Visual Studio 2008\Projects\SmartIrc4Net_02_Sept_09\smartirc4net" 8b3d4fa0132baf86286909e6f890906a3198544714:12
fatal: Not a git repository (or any of the parent directories): .git14:12
SethX perhaps ill find some useful info14:12
→ munichpython joined14:13
SethX thanks in the emanwhile14:13
DeFirence im using TortoiseGit to run the command, but i dont understand why i cant format a patch...14:13
← hyperair left14:14
← beni- left14:14
→ hyperair joined14:14
← segher left14:14
segher_ → segher14:14
→ altrux joined14:17
cbreak DeFirence: you have to be in a repository14:18
SethX: if you can find out which hash the remote branch head is on, you could compare to the local14:18
DeFirence so TortoiseGit isnt passing the corrent working dir?14:18
cbreak no idea what that tortoise thing is14:18
ngw hi *14:18
cbreak but did you try to cd to the right place?14:18
ngw how do I reset a single file to my HEAD ?14:19
cbreak you can use git checkout HEAD -- filename I think14:19
← thiago_home left14:19
ngw oh will try14:20
cbreak: thanks14:20
→ loinclot_ joined14:22
→ Bass10 joined14:23
← loincloth left14:23
→ keex joined14:23
← Piot left14:24
← bigmack83 left14:24
← remi left14:24
← HanzZ left14:25
→ OtherJakeSays joined14:25
→ bigmack83 joined14:25
← loinclot_ left14:26
→ loincloth joined14:27
← keex left14:27
← JakeSays left14:28
→ raboof joined14:29
← munichlinux left14:29
→ maioO joined14:29
← partisan left14:29
→ nighthwk1 joined14:32
← Geralt left14:32
maioO hi. I merged few branches then I pushed it into remote repo and now I want to 'unmerge' one of that branches. how should I do that?14:32
← d2dchat left14:33
→ tweened joined14:35
→ Fullmoon joined14:39
← khinester left14:41
charon maioO: unmerging is hard without rewriting history. what one usually does is: 'git revert -m 1 $the_merge' to "unmerge", and then rebase the branch on top of the revert to restore sanity14:41
Griffon26 interesting, so you rebase on the commit before the merge, right?14:43
→ ArjenL joined14:43
→ cannonball joined14:43
→ Turias joined14:46
cbreak http://whygitisbetterthanx.com/ interesting14:46
maioO: rewriting history of remotes is bad14:47
if you really want to do it, make sure no one pulls from it14:47
→ spearce joined14:47
→ LiamH joined14:49
← munichpython left14:49
→ aleray joined14:50
→ sgh joined14:51
→ Boriskr joined14:51
→ laj joined14:52
← EricInBNE left14:53
→ EricInBNE joined14:53
→ j416 joined14:54
← JackTheRabbit left14:54
murr4y by chacon of course :)14:55
← loincloth left14:56
→ bry joined14:57
→ karbassi joined14:57
→ JackTheRabbit joined14:59
→ coolcat joined14:59
← ereslibre_laptop left15:01
→ ereslibre_uni joined15:02
← ereslibre_desk left15:04
← keystr0k left15:05
→ godsflaw_ joined15:05
→ agile joined15:06
→ keystr0k joined15:08
← maxter left15:11
← khmarbaise left15:11
← [Po]lentino|away left15:17
→ [Po]lentino joined15:17
[Po]lentino → [Po]lentino|away15:17
← sgh left15:19
← godsflaw left15:19
← ereslibre_uni left15:19
← bigmack83 left15:21
phako is there someting like git add -pu which allows me not to add but to drop (parts of) patches?15:23
ngw → ngw|away15:24
→ pantsman joined15:24
ngw|away → ngw15:25
sitaram didnt the newer gits have a reverse option for this? (or am I confusing with something else...?)15:25
anyway, not in one of the released ones I think15:26
← jelmer left15:26
→ hummingkid joined15:26
cbreak can't you git add -i?15:26
→ dejans joined15:26
phako that can only revert stuff already added15:27
→ nadavoid joined15:27
phako I want to go through all the changes and revert some of them15:27
to HEAD15:28
ngw → ngw|away15:28
ngw|away → ngw15:28
dejans hey peeps. how can I revamp a file, with all its histroy? it was deleted months ago and I would like it back.15:28
→ jelmer joined15:28
→ ereslibre_uni joined15:28
→ loincloth joined15:29
← Fullmoon left15:29
→ Geralt joined15:30
← WinterWeaver left15:32
← tweened left15:34
← dejans left15:35
cbreak just reading a comparison between mercurial and git15:36
apparently git needs "frequent manual “repacks” of its metadata. Without these, performance degrades, while space usage grows rapidly. "15:37
strange that I didn't notice that15:37
→ drogue joined15:37
← raboof left15:37
DrNick cbreak: that's because they're automatic15:38
pantsman using mercurial requires frequent reading of hg-vs-git comparisons to convince yourself you've made the right choice15:38
with made-up statistics like "git is faster but hg's speed is enough for 99.9% of users"15:39
bremner presumably hanging out in #hg also helps15:40
cbreak the comparison I just read said "git is faster... but only on linux, on windows hg is faster!"15:40
bremner that could be true?15:40
cbreak no idea15:40
bremner yeah, neither of us cares, which is arguable an issue with git :)15:40
→ munichlinux joined15:41
bremner s/arguable/arguably/15:41
← ereslibre_uni left15:42
→ ereslibre_uni joined15:42
ngw → ngw|away15:42
ngw|away → ngw15:42
← aleray left15:43
← [Po]lentino|away left15:44
→ [Po]lentino joined15:44
[Po]lentino → [Po]lentino|away15:44
→ rettub joined15:45
→ schierbeck joined15:47
→ WinterWeaver joined15:48
→ agib joined15:48
← rettub left15:48
← JackTheRabbit left15:49
→ rettub joined15:49
cbreak "Some Git fans claim Git is now as easy to learn as Bazaar or Mercurial. With respect, we strongly disagree. For example, read and compare the help for the log command across the various tools: 4 pages for bzr log, one page for svn log, 26 pages for git log, and one page for hg log."15:49
interesting :)15:49
POWER!15:50
hyperair what a fail comparison15:50
Griffon26 where does that come from?15:50
← rettub left15:50
→ _Vi joined15:51
teuf they forgot that git is 3 letters to type while hg is just 2 letters!15:51
so much faster and easier15:51
→ rettub joined15:51
Griffon26 I use text-to-speech you insensitive clod!15:51
← rettub left15:52
hyperair git log is as easy to use as bzr log, but can do that much more, hence the extra pages15:52
Griffon26 damnit, ruined it15:52
cbreak: where did you read that?15:52
→ rettub joined15:52
cbreak http://doc.bazaar.canonical.com/migration/en/why-switch-to-bazaar.html15:52
→ tweened joined15:53
cbreak the first few lines of the comparison are approximately: I like bazaar more, so I'll write about it, but you better try for yourself15:53
← rettub left15:53
← tibra left15:54
→ rettub joined15:54
→ bcardarella joined15:54
→ aspotashev_ joined15:54
cbreak I guess it's good to have competition15:55
← rettub left15:55
→ tedoc2000 joined15:55
→ rettub joined15:56
← jelmer left15:56
← rettub left15:57
← nadavoid left15:57
→ rettub joined15:57
← ngw left15:58
→ mason-l joined15:59
← rettub left16:00
→ patcito joined16:00
→ sumangurung joined16:00
← DeFirence left16:00
← bentob0x left16:01
← patcito left16:02
→ Scorchin joined16:03
→ rettub_ joined16:04
ArjenL I always like the phrase "With respect, ...", which most of the time isn't16:04
Scorchin What's the general "best practice" when using multiple external libraries in a git project? E.g. I have a /libs folder where I have cloned libraries, should I leave these as-is so I can regularly do pulls on them to keep things up-to-date or just do clean exports of the codebase?16:05
← marc_kdab left16:05
→ bentob0x joined16:05
← rettub_ left16:05
→ rettub_ joined16:06
← rettub_ left16:07
← aspotashev left16:08
→ rettub_ joined16:08
← Geralt left16:09
→ jelmer joined16:09
→ JackTheRabbit joined16:12
→ dys``` joined16:12
→ patcito joined16:12
→ zuez joined16:12
→ Xaositect joined16:13
← _Vi left16:14
Wurschtlprater → jemand16:15
jemand → Jodlbjorn16:16
← patcito left16:17
→ nadavoid joined16:21
← mason-l left16:21
← rchavik left16:22
← dys`` left16:24
→ mason-l joined16:25
→ patcito joined16:26
charon Scorchin: submodules or subtree merge. discarding history is rarely a good idea16:26
Scorchin thanks charon16:26
will using submodules prevent me from making local changes then?16:27
or just confine those changes to that specific submodule?16:27
charon it will confine them16:27
i.e. you cannot make commits that touch several repos, and a submodule is a bit like a repo of its own16:27
Scorchin brilliants, that's exactly what I want16:28
*brilliant16:28
← giallu left16:29
→ khmarbaise joined16:30
← tedoc2000 left16:31
← patcito left16:32
→ tedoc2000 joined16:34
GyrosGeier hm16:34
somehow filter-branch does something weird16:34
it deletes refs/heads/master and then complains about HEAD pointing nowhere16:34
← jelmer left16:37
→ onigiri__ joined16:38
← Lisimba left16:41
→ remi joined16:41
← onigiri_ left16:42
→ partisan joined16:42
← galderz left16:43
← j416 left16:43
← TonyBurn_ left16:48
→ TonyBurn joined16:48
← [Po]lentino|away left16:48
← partisan left16:50
→ patcito joined16:50
← stamina_ left16:52
→ partisan joined16:52
← patcito left16:52
← remi left16:53
→ hebz0rl joined16:53
→ trs21219 joined16:55
→ r11t joined16:58
→ catalis joined16:59
→ irc2samus joined17:00
← irc2samus left17:01
← lemonchicken left17:01
← DavidKlein left17:02
→ patcito joined17:06
→ warthog9 joined17:07
← patcito left17:08
← JackTheRabbit left17:10
trs21219 can anyone help with this? http://stackoverflow.com/questions/1970459/best-workflow-with-git-github17:10
→ lemonchicken joined17:12
→ JackTheRabbit joined17:12
charon GyrosGeier: what's your full gfb command line?17:13
→ hobodave joined17:13
→ irc2samus joined17:13
dys``` → dys17:14
→ chuck joined17:15
← trs21219 left17:15
→ bambule joined17:15
GyrosGeier charon, I think I found the problem17:15
chuck Is it possible to reattribute commits that have already been made?17:15
I forgot to set my email before making them17:15
GyrosGeier a --commit-filter should do more than just set variables, shouldn't it?17:16
← EricInBNE left17:17
→ sgh joined17:17
← bentob0x left17:17
GyrosGeier hm17:17
an empty --commit-filter has the same effect17:18
removes the entire branch17:18
← bambule left17:18
lack chuck: If you haven't pushed them yet, yes.17:19
chuck lack, I've already pushed them to github17:19
→ _Vi joined17:19
→ tweened1 joined17:19
lack chuck: In that case, it *can* be done, but is probably not a good idea.17:20
← tweened left17:20
chuck why not?17:20
lack chuck: Have you read the 'man git-rebase' section about RECOVERING FROM UPSTREAM REBASE?17:21
Gitbot chuck: the git-rebase manpage can be found at http://git.or.cz/man/git-rebase17:21
lack This would constitute rewriting history, and rewriting history can be a bad thing if other people have pulled your code.17:21
If you don't care or are fairly sure no one has pulled your code yet, you can do it.17:21
But just be aware of the consequences.17:21
chuck yeah, other people have pulled the code17:22
does that mean it's pretty much impossible? ;-)17:22
(or a very bad idea, whatever)17:22
→ Lisimba joined17:23
← tweened1 left17:24
lack Impossible: no. Bad idea: yes. I'm serious about reading the RECOVERING FROM UPSTREAM REBASE section of the rebase manpage. It explains what the consequences are to anyone who has based their own work on what you are thinking of rewriting.17:25
← tbuehlmann left17:30
→ aresnick joined17:30
→ d0ugal joined17:31
→ MetaCosm joined17:33
jast chuck: basically it means that those people get the old version of the history AND the new version of the history, and pull will attempt to combine the two. it will probably succeed since the changes are exactly the same, but everyone who does the pull will end up having that part of the history twice.17:35
← mbroeker left17:36
jast (once with new attributions, once with the old ones)17:36
cbreak -> ugly.17:37
charon GyrosGeier: well, as the manpage says, it must do something along the lines of git-commit-tree (but can of course do more)17:38
if you just want to set variables, use --env-filter17:38
jepler_ Does this ring a bell for anyone? In my gitweb there are links like <a class="list" href="/git?p=aethertool.git;a=summary"> but when I click them in my web browser the ; becomes %3b (as shown in the address bar) and the request doesn't work.17:38
← cytrinox left17:39
→ cytrinox joined17:39
lack jast: I wonder if we could make 'git pull' a little smarter in that respect... Check the remote branch before the fetch, check the remote branch after the fetch, and DO NOT MERGE if that update is a non-fast-forward merge?17:40
charon lack: but people would just go "why doesn't it work! i'll try again!" and suddenly it would merge17:40
jepler_ → jepler17:40
→ fcuk112 joined17:41
→ bdiego joined17:41
jast lack: I could hack up a patch to add a command line switch for that17:42
jepler ugh, somehow mod_rewrite is tangled up in this as well17:42
jast jepler: yeah, it's apache madness, I'm afraid. I haven't quite figured out how to fix it yet. try adding [NE] to the rewrite rules...17:43
but that probably won't help17:43
jepler jast: thanks, that's exactly it17:43
jast oh, great17:44
→ tbuehlmann joined17:44
lack charon: True dat. But at least *one* level of check is better than zero...17:44
jast ah, yeah, the really big problem is different17:44
lack Some people may see it...17:44
jast lack: well, git pull is sort of meant for getting real merges. :)17:44
(mod_rewrite really gets hairy once you rewrite a URL with doubly-escaped things in it, or somesuch. I can't quite remember... thankfully, I might add)17:45
→ spaceonline joined17:47
jast lack: in fact, --ff-only already exists, it's just not documented17:48
lack Oh? Neat!17:48
jast I'll go and fix that :)17:48
lack jast: What version?17:49
jast lack: 1.6.617:50
lack Ah, I haven't upgraded that far yet. Thanks ;)17:50
→ patcito joined17:50
jast oh, and it's documented, too17:50
I just haven't actually installed 1.6.6 yet :)17:50
I build git from source from the master branch, usually17:51
doing that right now17:51
there, all new.17:52
← munichlinux left17:55
lack I run the latest gentoo ebuild, which has 1.6.6 but I haven't updated in a couple weeks. Probably time.17:56
Now if only I can get all my co-17:57
workers to update their Ubuntu VMs...17:57
← MetaCosm left17:57
→ JoeConyers joined17:57
→ MetaCosm joined17:59
lack jast: Hmmm... Just reading git-pull.sh in my git.git clone: Doesn't look like it's the same thing, maybe.17:59
Plus I can't find the documentation for it...17:59
→ fcuk112_ joined18:00
lack Ah, there's the docs.18:00
Looks like it's close to what I wanted to see...18:00
→ munichlinux joined18:01
← hebz0rl left18:01
OtherJakeSays → JakeSays18:01
← ProperNoun left18:03
→ d2dchat joined18:03
← JackTheRabbit left18:04
→ ProperNoun joined18:05
→ WALoeIII joined18:05
← jfxberns left18:06
→ jceb joined18:06
→ gilesgoatboy joined18:07
← d2dchat left18:08
← ia left18:09
→ d2dchat joined18:09
→ ngw joined18:09
← ngw left18:09
→ ixti joined18:10
← chuck left18:13
← Pupeno left18:13
← ereslibre_uni left18:14
→ ereslibre_uni joined18:16
← Tommy[D] left18:16
← MetaCosm left18:16
→ ereslibre joined18:17
→ MetaCosm joined18:17
← gilesgoatboy left18:18
→ madewokherd joined18:19
← sgh left18:20
→ anno^da joined18:23
→ ajonat joined18:23
← vbgunz left18:24
→ kifkif joined18:24
→ ia joined18:25
→ aconbere joined18:25
kifkif hi , i come from mercurial, is there an equivalent to mercurial rm --after ?18:25
→ nickoe_ joined18:25
aconbere is there a good way to get a diff from my current head, to the point at wich my branch was created?18:25
I guess this is the same as asking18:25
nickoe_ How do you clone a specefic revision/branch/"wahtever you call it in git hope you understand"?18:25
aconbere is there a good way to get the id of when the branch was created18:25
nickoe_ say this: http://openocd.git.sourceforge.net/git/gitweb.cgi?p=openocd/openocd;a=tag;h=f74fdc5e4afc94b65cba4b4c357f0e67cc9d185a18:25
wereHamster nickoe_: if you clone the whole repository I'm sure you also get that revision18:29
aconbere nickoe_: you can't clone anything but the full repository, but you can just combine that and checkout18:29
→ ambuli joined18:29
aconbere git clone <uri>; git checkout <branch>18:30
ambuli hi all, i am trying git clone url. it just hanges in Initialized empty Git repository in ....18:30
any clue where could be the problem ??18:30
→ icwiener joined18:31
← jceb left18:31
wereHamster ambuli: how long did you wait, which url did you use to clone (http/git/ssh?)18:31
← MetaCosm left18:32
ambuli i use like this git clone git@hostname is that the correct one. ??18:32
wereHamster git@hostname:path/to/repo.git .. right?18:33
nickoe_ aconbere, how do I locate the branch name?18:33
← doener left18:33
wereHamster nickoe_: git branch18:33
→ MetaCosm joined18:33
wereHamster nickoe_: er, you can checkout directly the tag (git checkout f74fdc)18:33
ambuli wereHamster, yes.. it is something like git clone git@hostname:ttkp/module118:34
→ jpieper joined18:34
ambuli wereHamster, it is not even displaying what is going on ????18:35
← TonyBurn left18:36
rettub_ → rettub18:36
→ doener joined18:37
wereHamster ambuli: you could try setting GIT_TRACE18:38
← irc2samus left18:38
← catalis left18:40
← project2501a left18:46
→ vbgunz joined18:48
← aspotashev_ left18:48
→ ebneter joined18:49
nickoe_ wereHamster, thank you18:49
← ayaz left18:50
→ Anjen1 joined18:50
→ munichpython joined18:51
← munichlinux left18:51
munichpython → munichlinux18:51
← loincloth left18:51
→ wolf2k_ubuntu joined18:53
→ jessep joined18:54
← gebi left18:54
→ datechman joined18:54
wolf2k_ubuntu Afternoon. In a stroke of...stupidity I guess, I somehow caused a dangling commit. The graph here shows. http://github.com/wolfman2000/pumpproedits/network I am unable to git push the svg branch due to this. I'm fine with deleting commit 7d4009d...if I can figure out how.18:55
wereHamster that's no dangling commit, it's still referenced by the 'svg' branch18:56
Anjen1 wolf2k_ubuntu, what's the message commit is splitting back?18:56
wereHamster if you don't care about that commit, why don't you delete the branch?18:56
Anjen1 *spitting18:56
wolf2k_ubuntu wereHamster: ...right. I already did the merge. nevermind18:57
ambuli wereHamster, do i have to give it as parametetr ??18:57
wereHamster ambuli: GIT_TRACE=1 git clone ...18:57
ambuli ok18:57
→ schmukkk joined18:57
→ Sigma joined18:58
← babilen left18:58
→ JoseLobato joined18:58
→ babilen joined18:58
→ dreiss_ joined18:59
← JoseLobato left18:59
→ solofight joined18:59
← solofight left19:00
→ solofight joined19:01
→ TonyBurn joined19:03
← Alfonzo left19:05
← solofight left19:07
← ajonat left19:07
← Anjen1 left19:09
→ defectiv joined19:10
Ilari spearce: Smart HTTP supports pushing, right? And based on my reading of the backend and hook code, hook messages would not work.19:11
→ project2501a joined19:12
← meanburrito920_ left19:14
← pantsman left19:15
→ Axius joined19:17
← Boriskr left19:17
← LiamH left19:17
→ jelmer joined19:17
→ m104 joined19:18
→ LiamH joined19:21
← LiamH left19:21
→ LiamH joined19:21
← datechman left19:24
← aconbere left19:24
spearce Ilari: all true19:25
← defectiv left19:27
← nickoe_ left19:27
→ zackattack joined19:29
Ilari spearce: I guess getting it to work would require protocol extension that sent hook messages in sideband...19:29
← vbgunz left19:29
→ SpookyET joined19:31
Ilari spearce: If it was also supported in "persistent connection" (not just "RPC") mode, it would fix hook messages for git:// and related protocols (right now ssh:// and files:// are the only ones where hook messages work).19:32
→ icwiener_ joined19:33
Ilari *file://19:33
spearce well, really you should just implement it in the git protocol stack just like we have sideband on upload-pack19:33
i actually had a patch series for this many many months ago19:33
there was a conflict with something else peff was doing at the same time in receive-pack (enhanced status report i think) so i held back waiting for his work to stablize, and then i just never finished.19:34
also, at the time, pushing over ssh:// and file:// were the only sane ways to push as git:// had no authentication.19:34
so hooks firing on remote side echoing back by stderr wasn't important at the time19:34
← icwiener left19:34
→ vinse joined19:35
→ ehabkost joined19:36
Ilari The days where ssh:// and file:// are the only protocols that can run hooks and can sanely push are gone.19:36
← r11t left19:37
→ r11t joined19:37
→ NfNitLoop joined19:39
NfNitLoop I'm using git-svn. I've done `git svn branch my-branch` and created my own branch, but it doesn't appear to have committed any of my git revisions into it.19:39
I assume I need to `git svn dcommit` to do so. Will dcommit know to use my-branch?19:40
or do I need to specify it every time?19:40
If it does know, how can I see what branch each of my local branches will commit to?19:40
spearce Ilari: yes, that's true, so we should implement sideband in the push-pack/receive-pack protocol19:40
← jessep left19:41
parasti NfNitLoop: you need to rebase your local commits on top of the new svn branch19:41
NfNitLoop: "git svn dcommit -n" to see where they will go19:42
NfNitLoop they should be based on it already. It created the branch from the last svn version in my local branch.19:42
thanks.19:42
Ilari spearce: IIRC, receive-pack uplink is sidebanded, but the downlink isn't.19:43
parasti NfNitLoop: "git svn branch" only creates the branch and fetches it (usually as remotes/$branchname), AFAIK19:43
NfNitLoop ah, yeah, it still wants to commit to trunk. I guess that means I always have to specify where I'm dcommitting to?19:43
Ilari NfNitLoop: Git-svn infers that from history.19:43
parasti NfNitLoop: no, that's not even possible as far as I know19:44
→ karbassi_ joined19:44
spearce Ilari: nope, uplink isn't sideband19:44
→ JackW joined19:44
← Axius left19:45
Ilari spearce: Ah, it only uses sideband functions in RPC mode.19:45
← karbassi_ left19:45
NfNitLoop Oh, hmm. So I can't 1) checkout from trunk. 2) Create a branch in git. 3) Push that branch to svn. Instead I have to: 3) fetch that branch 4) rebase my local branch onto it 5) and then commit that branch?19:45
← godsflaw_ left19:46
→ karbassi_ joined19:46
Ilari NfNitLoop: See if that git svn branch created any new remote tracking branches (git branch -r)?19:46
spearce Ilari: its not even sideband in RPC mode, sideband uses a 5 byte pkt-line header, RPC mode uses only 4 bytes19:46
← karbassi_ left19:47
NfNitLoop Ilari: it did. I'm a bit of a git n00b though and don't quite know how to work with remote tracking branches.19:47
I'll read up on that.19:47
→ karbassi_ joined19:47
Ilari NfNitLoop: Rebase your changes to be on top of that and then git svn might figure out proper branch to commit to.19:47
← karbassi left19:48
← MetaCosm left19:49
NfNitLoop Ilari: aha. Thanks!19:49
Ilari spearce: There are few send_sideband() calls in send-pack. And those appear to use sideband number -1 (which gets sent as sideband #255).19:49
spearce sideband < 0 means don't include the 5th byte that has the sideband number on it19:50
Ilari Ah yeah.19:50
← WinterWeaver left19:51
NfNitLoop Ilari: Worked like a charm. Thanks! (@parasti too.) :)19:51
← ehabkost_ left19:53
← zackattack left19:53
→ amerine joined19:56
Ilari Also, that project to create much faster asciidoc -> manpage converter was abandoned?19:58
← SpookyET left20:01
← jtbandes left20:01
→ loincloth joined20:02
→ loinclot_ joined20:02
→ jessep joined20:03
← yaotti_aw left20:03
→ JackTheRabbit joined20:03
→ TeMochkiN joined20:05
loinclot_ → loincloth_20:06
→ warlock_mza joined20:07
← ehabkost left20:09
← TeMochkiN left20:14
← LiamH left20:15
← unreal left20:16
→ unreal joined20:16
← loincloth left20:19
← ereslibre_uni left20:20
loincloth_ → loincloth20:20
→ LiamH joined20:25
→ Axius joined20:26
→ drizzd joined20:26
→ greboides joined20:28
→ jceb joined20:28
← erikvold left20:28
greboides hello, i have made some modifications on this branch but i want to add these changes on another branch to commit it, how can i do such thing cleanly?20:30
wereHamster greboides: either merge the branch into the other branch or cherry-pick the commits20:32
→ d0k_ joined20:32
greboides wereHamster: these changes were not committed yet20:32
nor added20:32
wereHamster then simply checkout the othe rbranch and commit there20:32
greboides wereHamster: but it complains that i cant switch branchs20:33
wereHamster in that case, you can either use stash/checkout/stash pop or checkout -m20:33
greboides wereHamster: if i stash will the modified file still be modified on the other branch?20:34
wereHamster no20:34
stash makes your working tree clean20:34
greboides is there any way to do it like this?20:34
wereHamster that's the whole point of git-stash20:35
like what?20:35
greboides i want to change branches and keep the changes20:35
wereHamster I'll repeat myself, but I'll try anyway: use stash/checkout/stash pop or checkout -m20:35
← icwiener_ left20:36
→ johnw joined20:36
← jessep left20:36
greboides wereHamster: can i stash pop on the other branch?20:37
johnw sure20:37
wereHamster if you follow what I suggested, you will be on the other branch when you stash pop20:37
but maybe I wasn't clear enough about that...20:37
greboides ok this is what i need i think20:38
wereHamster $ git stash && git checkout the-other-branch && git stash pop;20:38
greboides wereHamster: thank you, gonna solve some conflicts now20:39
→ datechman joined20:39
→ sgh joined20:39
→ jessep joined20:40
gnutoo_ → GNUtoo20:40
→ hakunin joined20:42
→ catalis joined20:42
→ xlistener joined20:43
osfameron gitosis is asking me for a password when I try to 'git clone'20:44
← Axius left20:44
osfameron I've updated gitosis.conf and keydir/user@hostname, committed and pushed...20:44
what am I missing? ;-)20:44
(it's for a previously working gitosis instance, so it's not the FAQ about AllowUsers, for example. At least don't think)20:45
→ sis_ joined20:46
greboides osfameron: do you have a ssh-agent running,knowing about your key and active for your working shell(i guess you need it)?20:47
osfameron greboides: nope, never figured out ssh-agent20:47
but that's not a pre-requisite for gitosis is it?20:47
given that I've got it workign from various ohter places...20:47
greboides osfameron: its a pre-requisite for ssh which gitosis uses for auth20:48
osfameron greboides: um?20:48
I've never set up ssh-agent anywhere20:48
→ smotchkkiss joined20:48
osfameron and I use ssh20:48
unless you mean that it's running transparently in the background20:48
→ Cesario joined20:49
greboides osfameron: i use keychain to manage it for me transparently in the background20:49
osfameron right. but I don't20:49
greboides osfameron: i think you need to set it up but i can be wrong i guess20:49
osfameron I've never had to use it before20:49
→ Twisted_Mentat joined20:50
osfameron and I don't want to complicate my setup to add something that isn't related (though I appreciate I *should* learn ssh-agent at some point...)20:50
← sis left20:50
charon osfameron: are you sure your gitosis hook works? i've seen server moves break it because the admin forgot to copy with permissions (losing the x bit), and the initial setup will work without the hook too20:51
osfameron charon: I cloned from a working user/ssh-key recently (last week iirc)20:52
and I've not moved server, and can't think of any other massive changes like that off the top of my head20:52
(as I'm not a very good sysadmin, I tend to avoid such...)20:52
← jessep left20:53
→ calin joined20:54
← d0k left20:54
d0k_ → d0k20:54
← fcuk112 left20:54
osfameron scp's over the entire repo to the server for now...20:54
greboides osfameron: is your user on the member's list of the group u are trying to clone from?(silly i know but i already forgot it once)20:56
not user, key i mean20:57
← mm_work left20:57
→ gmarquezWRS joined20:58
gmarquezWRS anyone familiar with gitolite (I see sitaram not online) ?20:59
I can 'git clone' from users who have an account on a Git Server using password-less ssh keys, but users without accounts can't clone anything even though I have user listed in gitolite.conf and key in keydir =/21:00
charon osfameron: another common mistake is to specify the wrong user. usually the gitosis user is 'git' or 'gitosis'. but i'm really stabbing in the dark...21:01
← incredible left21:03
gmarquezWRS I think I installed gitolite correctly, but the ACLs aren't being enforced.21:04
→ kumbayo joined21:04
gmarquezWRS What can I check to verify ACL are enforced or how can i enforce them?21:04
← d0ugal left21:05
Fissure most of the magic is in ~git/.ssh/authorized_keys21:05
← sde_antrix1 left21:06
→ sde_antrix1 joined21:06
jast gmarquezWRS: what happens if you ssh to the server using the gitolite account and one of the gitolite ssh keys?21:06
Fissure it uses the command option to force-run the gitolite server program when the user connects21:06
gmarquezWRS Yea as normal user I can ssh username@gitserver without login password, which means ssh authorized_keys is correct21:06
but its because I have an account on the server too21:07
Fissure gmarquezWRS: that's not using gitolite at all, though21:07
you need to look at the one for the git account that gitolite is installed under21:07
gmarquezWRS gitolite and gitosis is supposed to also allow for users without an account to use git from the gitolite.conf file no?21:07
Fissure right21:07
but they need an account (usually 'git') to run under21:08
i.e. a real system account21:08
gmarquezWRS hmm okay21:08
kifkif How can I merge only certain commits between branches ?21:09
_rane git cherry-pick21:10
← hyperair left21:10
→ hyperair joined21:12
kifkif seriously21:12
← brizly left21:15
gmarquezWRS OP gmarquezWRS, ?21:15
← pasky left21:15
← spearce left21:16
← anno^da left21:16
jast kifkif: seriously what?21:17
→ spearce joined21:18
← jceb left21:19
← razvand left21:19
Jodlbjorn → HektorTheseus21:22
→ Axius joined21:22
→ anjen1 joined21:22
Ilari gmarquezWRS: Try 'ssh <user and host you pass to git> git --version'?21:22
← Axius left21:23
kifkif jast, How can I merge only certain commits between branches ?21:23
Is that possible ?21:23
(I'm a noob as you can see)21:23
← bdiego left21:23
Ilari kifkif: 'git merge <commit>' would grab all parents of that commit too...21:23
jast kifkif: you can copy individual patches from another branch by using git cherry-pick, as _rane said. that's not technically a merge but it gets you the changes from the commit of your choice.21:24
→ defectiv joined21:24
kifkif jast, haha I though "cherry-pick" was a joke21:24
thanks21:25
teuf or if there are many of them you want to cherry-pick, you can create a temporary branch, remove the unwanted commits with git rebase -i, and then merge tha tbranch21:25
jast no, we tend to be actually helpful here ;)21:25
→ Jippi___ joined21:26
kifkif teuf, this is a really good advice21:26
jast, ;-)21:26
HektorTheseus → Kurvendiskussion21:26
→ pasky joined21:27
← tommd left21:27
→ Jippi_ joined21:27
→ schacon joined21:28
← LRN left21:29
← munichlinux left21:30
→ godsflaw joined21:30
← godsflaw left21:33
← khmarbaise left21:37
← wolf2k_ubuntu left21:38
← __iron left21:38
→ numberGrey joined21:38
→ khmarbaise joined21:38
→ __iron joined21:38
← calin left21:39
← sde_antrix1 left21:39
→ tommd joined21:40
← Jippi___ left21:42
→ khmarbaise_ joined21:43
← schierbeck left21:43
→ fcuk112 joined21:44
← defectiv left21:44
← lemonchicken left21:47
← ambuli left21:49
→ flaguy48 joined21:49
→ Wazzzaaa joined21:50
Wazzzaaa which git programme is recommended for windows?21:50
→ makeworkeasy joined21:50
jast for command line stuff, msysgit21:50
Wazzzaaa ok21:51
jast as a GUI you might like tortoisegit (I don't know how good it is, I just know that other people use it)21:51
← tekky left21:51
gmarquezWRS jast: do you know if you can setup multiple paths with gitolite or gitosis?21:52
Wazzzaaa I'll first try msysgit21:52
jast gmarquezWRS: multiple paths for what?21:53
Fissure gmarquezWRS: if you want to store things in separate places on the filesystem, you can always symlink them21:54
← madewokherd left21:54
→ godsflaw joined21:55
← LiamH left21:57
← warlock_mza left21:58
← anjen1 left21:59
← khmarbaise left21:59
khmarbaise_ → khmarbaise21:59
← godsflaw left22:02
→ ambuli joined22:02
→ eletuchy_ joined22:03
← eletuchy left22:03
← JackTheRabbit left22:04
→ Pupeno_ joined22:04
← Sigma left22:05
karbassi_ → karbassi22:05
Kurvendiskussion → yy22:06
← MadCoder left22:06
yy → c22:06
→ MadCoder joined22:06
c → Guest3840522:06
→ danigm joined22:07
→ lemonchicken joined22:08
danigm hi all, I'm trying to run gitweb with lighttpd but it doesn't work, that is my lighttpd config file http://pastebin.com/d61bda12c anyone knows anything about that?22:08
→ JackTheRabbit joined22:08
→ schierbeck joined22:09
Guest38405 → psypointer22:09
→ tfarina joined22:10
Fissure doesn't work... how?22:11
gitweb is just a normal cgi script22:11
→ LiamH joined22:11
danigm the server get me the perl file22:13
mugwump did you enable mod_cgi?22:14
lighttpd can be a bit special at times22:14
danigm in server.modules i put mod_cgi22:14
mugwump yeah. well, try #lighttpd I guess22:14
I've been too lazy to follow the herd to nginx22:15
danigm mugwump: I'm trying there22:16
Arrowmaster i thought nginx was mostly used for static content22:16
mugwump it suppoorts fastcgi22:17
wereHamster Arrowmaster: or as a proxy22:18
mugwump damn o key has a stutter22:18
← catalis left22:19
→ t0rc joined22:20
← r11t left22:20
← JackW left22:20
→ r11t joined22:21
→ meanburrito920_ joined22:21
→ erikvold joined22:23
→ gebi joined22:23
← yann left22:24
← danigm left22:24
← LiamH left22:25
← jelmer left22:25
→ nullvariable1 joined22:29
← nullvariable left22:29
nullvariable1 → nullvariable22:29
← erikvold left22:29
← altrux left22:30
← Wazzzaaa left22:30
← tommd left22:31
← tfarina left22:32
→ tfarina joined22:32
→ EricInBNE joined22:34
→ Tureba joined22:34
← r11t left22:35
→ r11t joined22:36
→ [sr] joined22:37
[sr] howdy people22:37
→ univate joined22:37
[sr] i'm a very very newbie guy to git22:37
after set the repo on the server22:38
jast we all were once :)22:38
[sr] i was able to clone it on other machine22:38
jast: true :)22:38
then i added some files, and runned:22:38
git add .22:38
git commit -m "aa"22:39
git push22:39
and i got:22:39
jast hanging out in this channel seems to magically turn people into experts, though22:39
[sr] To git://192.168.1.206/project/myproject.git22:39
! [remote rejected] master -> master (n/a (unpacker error))22:39
22:39
jast you can't normally push using the git:// protocol22:39
most of the time push is used with the ssh:// protocol22:39
[sr] jast: that's what i expect :) svn is so easy for me, and git is a really a pain for me...22:39
hum...22:40
jast git:// is for anonymous access only22:40
and while it's possible to enable anonymous push, you typically don't want that :)22:40
[sr] well then i seem to have much more work here than i thinked22:40
→ godsflaw joined22:40
Ilari In fact, reaching that point requires enabling anonymous push... :-)22:40
[sr] about security there's no problem, 'cause this will be only internal22:40
but i want to use auth anyway22:40
← r11t left22:40
Ilari [sr]: There are other error messages that would probably give the true cause.22:41
→ thiago_home joined22:41
→ Jak_o_Shadows joined22:41
→ r11t joined22:41
Ilari Ah, but because its git://, those other errors do not appear.22:42
[sr]: Best guess: Permissions are wrong (it tries to access that repository using whatever user git-daemon is running as.22:42
[sr] it seems to be make sence that error with what you guys say...22:42
→ AkumaStreak joined22:43
[sr] let me do a chmod 777 and test :)22:43
Ilari [sr]: FAIL.22:43
← keystr0k left22:43
Ilari [sr]: Does 192.168.1.206 run sshd?22:43
[sr] yap22:43
Ilari [sr]: Then push over that: '<user>@192.168.1.206:path/relative/to/home/directory'.22:44
[sr] bindo, permissoes :D22:44
i mean, bingo22:44
ok to push using auth, let me test that...22:45
but anyway, i can push using git://22:45
Ilari [sr]: Another is 'ssh://<user>@192.168.1.206/path/relative/to/filesystm/root'.22:46
← greboides left22:47
← d0k left22:47
→ giallu joined22:47
→ r11t_ joined22:47
[sr] let me...22:47
← intripoon left22:47
→ d0k joined22:48
Ilari What git-daemon2 likes about trying to push over git://: 'fatal: remote error: W access for jpcrr DENIED to anonymous'. :-)22:49
→ chrizz joined22:50
jast git-daemon2?22:50
Arrowmaster git-daemon2??22:50
[sr] hum22:50
i'm trying other thing22:50
Ilari The daemon for gits:// protocol (it can also deal with git://). I should put that somewhere.22:50
[sr] dont kill me for this :)22:50
with tortoiseGIT22:51
Arrowmaster gits://?22:51
jast ah, I see. was that the thing using the authentication scheme you talked about some time ago?22:51
[sr] i'm seeing how to specify the auth...22:51
Ilari jast: Yup.22:51
gits:// -> Git using TLS.22:51
jast fun. once you make it available somewhere, I think it would be great if you announced it on the git list.22:51
Ilari And if something in that error message reminds of gitolite.... :-)22:52
← project2501a left22:53
← sumangurung left22:53
[sr] how would be my push command using ssh22:53
Ilari [sr]: What is the right user on remote side?22:53
[sr] "jorge"22:53
← Jak_o_Shadows left22:54
← v0n left22:54
Ilari [sr]: And what is path to that repository. Either relative to filesystem root, or jorge's home?22:54
jast Ilari: nah, no similarity at all.22:54
Ilari: in fact I think that's actually not a good thing. "W" is a bit cryptic.22:54
[sr] Ilari: /home/apache/project/myproject.git22:55
← khmarbaise left22:55
Ilari [sr]: 'ssh://jorge@192.168.1.206/home/apache/project/myproject.git'.22:55
[sr]: SSHd runs on port 22, right?22:55
[sr] yes default, but tell me22:55
mugwump Ilari, a while ago the idea of signed pushes was also mooted22:56
[sr] the git command: git push ssh://jorge@192.168.1.206/home/apache/project/myproject.git ?22:56
Ilari [sr]: Looks right.22:56
mugwump: Signed pushes are very nasty for client side.22:56
mugwump you mean by that, you have to keep entering your passphrase?22:57
Ilari mugwump: Plus have tools to sign the stuff on push. And entering passphrase repeatedly is quite tiresome (as I have noticed).22:58
← fcuk112_ left22:58
mugwump well, you can use an agent like gnupg-agent or pageant22:58
← spearce left22:59
mugwump it does have some benefits thouogh22:59
→ spearce joined22:59
Ilari Yeah. Those same ones would actually work with the gits:// client I have (I protected the actual keypair with my main GPG key).23:00
mugwump mmm, they're probably complementary approaches23:00
[sr] Ilari: before that, i have two files on the git server, that were already pushed by me, i'm trying to download them on another place, everything's ok23:01
my question, why are not the files in /home/apache/project/myproject.git ?23:01
where they live? :)23:01
wereHamster [sr]: is the myproject.git repository bare (did you create it with git init --bare)?23:02
[sr] no, just git init myproject.git23:03
wereHamster [sr]: faq non-bare23:03
Gitbot [sr]: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare23:03
Fissure [sr]: then they're in a .git directory inside that23:03
← r11t_ left23:03
[sr] that files exists, true23:04
before going to the faq, let me reask23:04
starting frrom the beguinning23:04
← Tureba left23:05
[sr] --reuseaddr --syslog --export-all --enable=receiv23:05
e-pack --verbose --base-path=/home/apache/gitprojects23:05
Ilari [sr]: Since you figured out ssh, you don't need that --enable=receive-pack option anymore.23:05
[sr] where shouls the bare repo be created, git init --bare /home/apache/gitprojects ?23:06
→ asterisco joined23:06
asterisco sorry my net ... cabum!23:07
Ilari [sr]: Presumably '/home/apache/gitprojects/fooproject' (since git repositories are per project).23:07
asterisco i'm [sr]23:07
← r11t left23:07
asterisco so, is the base is /home/apache/gitprojects, i have to init the bare repo there, right?23:08
i didn't read that part23:08
Ilari asterisco: If you have project named 'foo', you probably should create repo for it to '/home/apache/gitprojects/foo'.23:09
← ixti left23:09
wereHamster asterisco: you still have to create a directory per project (mkdir /home/apache/gitprojecs/myproject.git)23:09
but you type 'git init --bare' inside it instead of 'git init'23:09
Ilari Hmm... Why there isn't shell primitive to create and switch directories in one go? It would be useful in some cases.23:10
← GNUtoo left23:10
asterisco i have a confusion here23:10
for the base dir, i have to init that directory with --bare, correct?23:10
ex: git init --bare /home/apache/gitprojects23:10
Ilari asterisco: No.23:10
wereHamster no23:10
→ Th0ma5 joined23:11
← JackTheRabbit left23:11
asterisco them, what for the --bare creation?23:11
them=then23:12
→ justin-george joined23:12
wereHamster asterisco: at some point in the past you created the directory /home/apache/project/myproject.git and ran 'git init' inside it, right?23:12
← [sr] left23:12
asterisco wereHamster: yap, but now i deleted all dirs and i'm starting again23:12
asterisco → [sr]23:12
Ilari [sr]: Server side only hopefully?23:13
wereHamster now you do everything exactly the same, except you run 'git init --bare' instead of just 'git init'23:13
[sr] Ilari: yap, it's a clean server for now23:13
← smotchkkiss left23:13
→ zackattack joined23:14
[sr] ah... for every project, i run: git init --bare /home/apache/gitprojects/project1.git23:14
git init --bare /home/apache/gitprojects/project2.git23:14
and so on... correct?23:14
Ilari [sr]: Right.23:14
[sr]: (assuming that also creates the directories).23:14
[sr] Ilari: hum... i've read on 2/3 places on the web that was only need git init "directory.git"23:15
← giallu left23:15
[sr] the extension project.git23:15
is mandatory?23:15
Ilari [sr]: The .git extension is just convention.23:15
Arrowmaster but its a good convention to follow23:15
[sr] yap, i like it :)23:15
Ilari [sr]: Wheither it creates bare or non-bare repo is determined by presence or absence of --bare flag.23:16
[sr] Ilari: got that!23:16
i have one problem now..23:16
no problem, dumbness mine :)23:17
now gitweb show's "project.git"23:17
great :)23:17
now i can clone using git://192.168.1.206/project.git23:17
one question (dont hit me)23:17
→ giallu joined23:17
[sr] projects ALWAYS have to be started on the server, or can they be initialized remotly?23:18
→ kpreid_ joined23:18
NfNitLoop gah, I did 'git svn init' to specify my SVN branches directory and now 'git svn fetch' is re-fetching everything. >.<23:18
Ilari [sr]: Remote initialization would require something like gitolite.23:19
[sr] ok i'll skip that!23:19
→ eletuchy joined23:19
← sis_ left23:20
[sr] there's something strange with this tortoiseGIT but i'll check it later23:22
mugwump NfNitLoop: it's a bit tricky, you have to make sure your refs are arranged as the fetch is expecting23:23
then it should rebuild the metadata database23:23
[sr] hum, Ilari, wereHamster, there's something here that i'm missing also23:24
steps, created my bare project23:25
cloned it locally23:25
remote i mean23:25
commit, and then push, but git this:23:25
$ git push ssh://jorge@192.168.1.206/home/apache/gitprojects/myproject.git No refs in common and none specified; doing nothing. Perhaps you should specify a branch such as 'master'. Everything up-to-date23:25
Arrowmaster add master to the end of the command23:26
Ilari [sr]: 'git push origin master' (push branch 'master' to 'origin').23:26
← johnw left23:26
Ilari [sr]: git clone sets up name 'origin' to refer to place the repo was cloned from.23:26
→ altrux joined23:27
[sr] i believe i understood... origin = from where were cloned | master = from where is going to be pushed23:28
Ilari [sr]: No. 'master' is name of branch.23:28
[sr]: 'git branch' lists known (local) branches in current repository.23:28
→ robinr joined23:28
[sr] yap, it shows, cool :)23:29
Ilari [sr]: The default is to push only what exists on both sides, and since the remote is empty, this is nothing.23:29
← Twisted_Mentat left23:30
[sr] i believe i've learned more today!23:31
← drizzd left23:31
[sr] hey one thing, the gitweb interface can't show the euro symbol23:31
from an text file that contains it23:31
Ilari [sr]: Incorrect character sets?23:32
← schierbeck left23:32
[sr] hum... it's FF, looking at the source it is with utf823:33
ok the file was created using ANSI, may be that the problem23:33
→ krawek joined23:33
Ilari [sr]: Maybe gitweb is assuming the file is UTF-8?23:33
← eletuchy_ left23:33
[sr] let me test with a file created using utf823:34
← Adlai left23:34
[sr] ok that's the problem23:35
a text file created using utf8 encoding works fine23:35
← _Vi left23:35
→ corni_ joined23:35
← corni_ left23:36
[sr] i'm almost a master in git :)23:36
i should NEVER delete the .git directory inside each project!23:37
or i'll kill it...23:37
Ilari [sr]: Well, there's only one non-obsolete character set: UTF-8 encoded Unicode.23:37
→ _Vi joined23:37
→ erikvold joined23:37
Ilari [sr]: Also be very careful with recursive commands (best to use 'git ls-files' to get list of files).23:38
← erikvold left23:38
[sr] let me ask one opinion, i'm going to use git for some projects that are .php files23:39
and for some that are .net projects23:39
do you see any inconvinient for the .net projects?23:39
Ilari [sr]: I don't know about VS project files...23:40
[sr] Ilari: they are normal text files23:40
with some images (if added), some .dll's)23:40
Ilari [sr]: And also, VS can not be configured to save Unix format. Thus, keep 'core.autocrlf' enabled.23:40
[sr] but those will work as binary files23:40
→ adamm joined23:41
Ilari [sr]: If CRLFs can work their way into repository, expect problems.23:41
← Chillance left23:41
Ilari [sr]: But it should be enabled by default on msysgit. To check, do 'git config --global core.autocrlf'.23:42
[sr] didn't returned any info23:42
anyway, i'm going to use tortoiseGIT, since this is windows, i believe it'll work ok...23:43
→ neunon joined23:43
zuez If I want to run an arbitrary shell script after a commit gets pushed to a remote server, what hook should I be paying closer attention to?23:43
wereHamster zuez: post-receive or post-update23:44
Ilari [sr]: To set it: 'git config --global core.autocrlf true'.23:44
zuez wereHamster: Ah, okay.23:44
[sr] thank you23:44
Ilari zuez: And nitpick: It can be multiple commits. :-)23:44
zuez Ilari: Yeah. I don't want the shell script fired off for each individual commit. Typically a user will make several commits locally, then push them to a remote server.23:45
Illari: I just want the trigger to go off when the git push happens.23:45
[sr] Ilari: implications if i only do anonymous push's ? besides that i won't know who did it23:45
→ RaceCondition joined23:45
Ilari [sr]: Anybody who can connect to that service can push too.23:46
RaceCondition where could I find a static binary of git?23:46
[sr] Ilari: hum ok, no problem, that's internal service, wont be exposed to the web23:46
Ilari RaceCondition: Nobody has built one because git also has critical shell parts.23:47
psypointer → FatboyPsy23:47
RaceCondition Ilari: ahh, of course, how stupid of me, i forgot23:47
← _Vi left23:48
RaceCondition but can git be built and installed and used as non-root?23:48
Ilari RaceCondition: Of course.23:48
RaceCondition OK23:48
← Fissure left23:49
Ilari RaceCondition: I do so to get versions newer than what's in distro repo (I also have compile of pu branch version).23:49
[sr] Ilari: tell me, a textfile, created by windows, in this case with notepad++, gitweb always show in the end of its content: \ No newline at end of file23:49
should i worry?23:49
RaceCondition Ilari: OK, great23:49
Ilari [sr]: That means the file ends with something else than line change.23:50
[sr] got it..23:50
Ilari [sr]: Usually not a problem.23:50
[sr] before i put here my .net projects, i'll use the smaller one for a while for tests..23:50
to tune something that may need to be tunned..23:51
→ \ask joined23:51
Ilari [sr]: The projects need to be huge (or contain large files) before you hit scalability limits. But its good to keep it small at beginning (to learn about workflows and that stuff).23:53
→ Geralt joined23:53
[sr] Ilari: the total size may be 100mb each one, but for contaning forms, arround 150/20023:54
Ilari [sr]: The 'large files' refers to individual files.23:54
[sr] undertood, i believe i won't have nothing larger then 10mb per file23:54
what is large for git, 100mb file?23:55
← kumbayo left23:55
Ilari [sr]: 100MB is already quite large.23:55
[sr] thanks, that gives me the ideia of the size i need23:55
Ilari The approximate current scalabilty limits (be careful before exeeding): 10MB per file, 100k files per project (Linux, on windows, maybe 10k), 100M revisions.23:56
[sr] hum23:57
doesn't linux kernel pass that limit?23:57
it has so many projects created23:57
cbreak with what command can I find out what a hash is?23:57
Ilari [sr]: Actually, that 'files per project' refers to number of files in one commit.23:58
wereHamster cbreak: what a hash is? What do you mean with that?23:58
bremner cbreak: git show $SHA1 ?23:58
[sr] Ilari: ah per commit23:58
ok them :)23:58
cbreak fatal: bad object 9cfcd715f0263113efe0946bfec752974716df3e, looks bad?23:58
wereHamster cbreak: try git fsck --full23:58
Ilari [sr]: The reason why the limit it higher on Linux is that Linux has better VFS system.23:58
← cannonball left23:59
← Geralt left23:59

Logs Search ←Prev date Next date→ Channels Documentation