IRCloggy #git 2009-11-29

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.

2009-11-29

← parigaudi left00:00
← madewokherd left00:02
Eraldo when I do "git branch -a" I get: http://pastebin.org/5826600:03
before making that bare repo I only had master and eeeeraldo00:03
why is there now 4 branches00:03
→ Paraselene__ joined00:03
← Paraselene_ left00:03
Paraselene__ → Paraselene_00:03
→ madewokherd joined00:05
lack Eraldo: Well, that output shows one local branch 'master' and 2 remote branches, remotes/origin/eeeeraldo and remotes/origin/master, plus the fact that the 'origin/master' is the 'HEAD' of that other repository.00:07
→ tfarina joined00:07
lack Eraldo: Everything prefixed with 'remotes/origin/' are REMOTE TRACKING branches, and they represent what is on that bare repo. They are only updated when you do a fetch or push operation.00:08
→ univate joined00:10
← icwiener left00:13
Eraldo lack: how can I remove eeeeraldo branch from my bare repo?00:14
"git branch -d eeeeraldo" ?00:15
selckin --git-dir=repo.dir00:15
← bcardarella left00:15
← Rayne left00:15
→ bcardarella joined00:16
lack Eraldo: To delete a remote branch: git push origin :eeeeraldo00:17
← DavidKlein left00:20
→ EricInBNE joined00:21
hipe_ → hipe00:21
lack Eraldo: I recommend reading a tutorial on how git works to learn more about remotes and such...00:22
← hakunin left00:23
→ hakunin joined00:23
→ seanmccann joined00:23
Eraldo lack: yup... that is planned... sorry ..it is my organisatonal system that I put under that repo I am talking about atm that is where I schedule my todos :)00:23
← Sho_ left00:24
lack Heh, that sounds backwards... using a tool to note that you should learn to use that tool... ;)00:25
→ yarddog joined00:26
yarddog → Guest6167200:26
← LiamH left00:27
← v0n left00:27
← adam12 left00:28
→ thenetduck joined00:32
Mage_Dude lack: Hehe Like leaving a note in your car that says 'Learn to drive."00:39
→ jelmer joined00:40
→ SRabbelier joined00:41
→ trochala joined00:42
→ Wraithan joined00:43
← raz left00:53
→ Dextorion joined00:57
← seanmccann left00:58
← intripoon left01:03
→ tvw joined01:03
← squentin left01:04
→ imajes joined01:05
← hakunin left01:06
← psoo left01:07
→ hakunin joined01:07
← tvw left01:09
← cubuanic left01:14
straszheim anybody know if the windows git requires a shell?01:15
if i write custom git scripts, i'm wondering if they'll be usable at all by windows folks01:16
looking to use plumbing to store metadata in specially-named branches... has to work on winnnows. :/01:17
→ jtdowney joined01:19
→ sshc joined01:20
← corni left01:26
← aspotashev left01:27
→ tek joined01:30
← doener left01:30
alus straszheim: I use git on the windows command line01:35
straszheim: so if you're writing something for use on all platforms, may I suggest a cross platform language like C or Python01:35
Glenjamin something interpreted like python, ruby, perl is probably a better fit than C01:36
straszheim alus: yeah, i'm good with all three01:36
er python sh C01:36
← pavelz left01:36
→ pavelz joined01:36
straszheim would need a python package for git that has good bindings to plumbing and is not too tough to redistribute01:37
alus Glenjamin: git is written in C, so it's sort of a given that C would work. requiring python might require an install of python on Windows01:37
straszheim we require python to build anyway, so that's not an issue01:38
alus ok01:38
Glenjamin yes, but python is better suited to being a cross platform alternative to bash scripts01:38
and having c "work" is misleading, cos you'd have to compile binaries for each platform01:39
straszheim true... thing is the only git i'd seen on windows comes with bash anyway01:39
i know that half of git was shell scripts at one point, was wondering if that is still the case, in which case the shell is there already anyhow01:39
← hakunin left01:40
← tbuehlmann left01:41
Glenjamin the start menu shortcut on windows opens up "git bash"01:41
but git is also on the normal windows command line path01:41
straszheim ah01:42
→ blakemizerany joined01:42
← Alagar1 left01:42
straszheim yeah, hm, it's in the same dir as bash.exe :)01:43
→ ph^ joined01:43
→ hakunin joined01:43
← ph^ left01:45
→ aspotashev|eeepc joined01:46
sitaram straszheim: bash exists on windows (msysgit at least); I rely on it now for gitolite install01:47
← jtdowney left01:47
← fualo left01:48
straszheim sitaram, yeah i'm seeing a few things that are still shell scripts on this windows install I have here, git-submodule for instance01:48
sitaram the only thing it doesn't seem to do (so far as the changes I had to make to gitolite's easy install is concerned) is that [[ foo =~ bar ]] (where bar is a regex not a glob) doesnt work01:48
straszheim is afraid he's going to find out that these are the things that are broken01:48
sitaram straszheim: however, all else being equal, I'd use perl. Guaranteed to be there, and guaranteed to work without special cases like above01:49
(although I agree that's not everyon'e cup of tea!)01:49
s/n'e/ne's01:49
straszheim I'm fine with perl, but that'd be an additional dependency01:49
→ ph^ joined01:49
sitaram no it wouldnt01:49
you have git, you have perl01:50
guaranteed :)01:50
msysgit has it for sure01:50
straszheim so I do :)01:50
sitaram straszheim: I find, on Windows, that relying on msysgit works out pretty well. Single installer (unlike cygwin) easy to distribute, etc.01:52
straszheim submodules just worked on this windows box, so i'm guessing that means libexec\git-core\git-submodule just ran which is a bash script. okay so this turns out not be so hard.01:52
sitaram: that sounds reasonable, thanks.01:52
← thenetduck left01:54
sitaram straszheim: not at all; as far as these two languages (bash and perl) are concerned... as I said, I rely on them for gitolite and have not been disappointed yet01:54
← hakunin left01:54
→ vbgunz joined01:57
→ fualo joined01:58
← fualo left01:58
→ jackylee joined01:59
← jlilly left02:01
← ludde left02:04
→ bartman joined02:05
← Keiya left02:06
→ Keiya joined02:06
← brainbox1100 left02:08
← aziz left02:08
← Keiya left02:11
→ Keiya joined02:11
→ hipe_ joined02:13
← hipe left02:13
← ab3 left02:15
→ Keiya_ joined02:17
← orafu left02:19
→ orafu joined02:19
→ dreiss joined02:20
→ VVSiz_ joined02:25
→ Alagar joined02:27
← bcardarella left02:29
← DamianZ left02:30
→ tillberg joined02:31
← ph^ left02:33
← Keiya left02:34
← tchan left02:37
→ tchan joined02:38
← Alagar left02:39
← Pupeno left02:39
← Dextorion left02:40
← jschoolcraft left02:41
→ bcardarella joined02:41
← VVSiz left02:43
← dreiss left02:46
→ EugenMayer joined02:47
→ xy joined02:49
→ Pupeno joined02:49
→ Adman65 joined02:50
→ jaysern joined02:50
← blakemizerany left02:50
→ Keiya joined02:51
Adman65 I have a remote branch and my local master branch. I've rebased the master branch by squashing a whole bunch of commits. I've now done more work on my local branch with more commits. I want to push back to the remote, but I can't because of non-fast-forward. What is the proper course of action?02:51
← jelmer left02:51
→ hipe joined02:53
straszheim merge the remote, then push?02:53
← hipe_ left02:53
Adman65 can you merge from a remote branch without checking it out locally ?02:56
straszheim sure02:57
→ jlilly joined02:57
Thell general question:: what would your approach be for the following situation? We are adding a tracing utility to our source tree (in the form of function guards) that can be enabled/disabled by individual devs. They are debug build only, and each dev may have their own desired groupings for which ones are on and which aren't so we wouldn't want these activated in files that would end up getting02:58
Adman65 hmm then im having a problem with the syntax02:58
Thell committed. One option is a special file extion that is ignored.02:58
straszheim Thell you got cut off at 'getting'02:58
Thell :) Another is to have a repo that houses some of the more likely grouping and have that be a submodule02:59
then they could make adjustments in there without worrying about their custom groups showing up in the main repo03:00
and if they thought it was worth sharing it could be submitted to that repo...03:00
← tek left03:01
Adman65 straszheim: how would you merge from the remote branch?03:02
Thell Adman65: git checkout $local_branch && git merge $remote_repo_name/$branch03:02
although I found I really like git merge --log --no-ff $remote_repo/$branch03:02
← tfarina left03:03
← cortana left03:05
Adman65 hmm, remote repo heroku has branch master. git merge heroku/master says doesn't point to a commit03:05
← Keiya left03:06
sitaram Adman65: did you "git fetch heroku"?03:09
Adman65 no03:09
← sergiodj_ left03:11
← Keiya_ left03:12
← botanicus left03:13
→ ilteris joined03:13
→ hakunin joined03:18
→ cytrinox_ joined03:20
← kpreid_ left03:20
→ kpreid_ joined03:21
EugenMayer sitaram: when I use gitolite to initialize repo and I check it out, I can't checkin, I always get that the remote branch has no master03:22
sitaram: in addition, what conditions are needed that "@my_repos = repo1" ..that thos branches are craeted in the gitolite repo03:23
s/check it out/clone it03:23
sitaram EugenMayer: first push has to be "git push origin master" since the remote is empty, and has no matching heads. This is nothing to do with gitolite :)03:25
or at least "git push origin somebranch"03:25
didnt understand your second question03:25
EugenMayer sitaram: ah! thanks..03:25
sitaram: wait, I prepare a pastebing03:25
→ jepler joined03:26
EugenMayer http://paste.debian.net/5269403:26
according to the docs, the gitolite-admin repo should not have a branch called "my_repos" ( which it does not have )03:27
*does not have03:27
← __iron left03:27
sitaram EugenMayer: read from "here's how to use that in practice" in doc/503:27
s/that/this03:28
EugenMayer I have read the docs several times. E.g. I don't not even find what "RW+" makes in difference to "RW"03:28
sitaram look in conf/example.conf. (The config file is actually documented there; was too cumbersome to do it as narrative text)03:30
EugenMayer you maybe should tell others that. its always confusing having docs on 2 differet places. Looking into03:31
sitaram EugenMayer: the admin doc does say to see the example conf for details on what goes in that file03:32
EugenMayer well, I have seriuosly read it several times and did not stumble uppon. Even if it is there, you might of think place it better :)03:33
sitaram EugenMayer: no03:34
EugenMayer: conf/example.conf is the right place for it03:34
EugenMayer I mean, place the note better, that its documented there03:34
pcc1 is it possible to obtain a list consisting of the intersection of commits in the dag bounded by commit c1 and commits in the dag bounded by commit c2?03:34
sitaram EugenMayer: I'd expect users, after install, to read the admin doc. Section is called "adding users and repos" and right there it says to look in conf/example.conf03:35
EugenMayer and don't forget, that example.conf is not in the gitolite-admin repo..so maybe people never find it03:35
sitaram EugenMayer: if you ran easy install you have a clone... come on. I'm not a baby sitter :)03:35
EugenMayer sitaram: well, sure you are not.03:36
← cytrinox left03:36
cytrinox_ → cytrinox03:36
EugenMayer but having the conf in the clone of YOUR git repo03:36
having it on the SERVER where it has the example docs under ~/gitoline-admin and under .gitoline/conf03:36
← rudi_s left03:37
EugenMayer there are several places, and the most obvious one, the one the admin always looks at, there it is not present. Well its up to you ( its just a user / admin experience )03:37
sitaram EugenMayer: dont know what you mean. As I said, if you ran "src/gl-easy-install", right there in that same directory is "conf/example.conf"03:37
→ Wolfman2000 joined03:39
sitaram EugenMayer: why would I clutter the admin clone with what amounts to documentation? the admin clone is just that -- a clone! There could be any number of them... I have one on each machine I *use* so that I can make changes from anywhere. You'd have a better case arguing that conf/example.conf should be doc/example.cond, frankly :)03:39
ss/cond/conf03:40
(damn!)03:40
Wolfman2000 Evening. I am right now thankful that I didn't make my latest commit, because I would have risked losing everything. My last good commit seems to be... [master 1090455]. I'm hoping that's a unique identifier. How can I reset to that commit?03:40
EugenMayer Wolfman2000: its hardly impossible to lose anything in git03:40
Wolfman2000 EugenMayer: That may be true, but...well, I panic easily.03:40
EugenMayer until you play arround with unstaging or reverting03:41
Wolfman2000 is still getting used to the administrative stuff.03:41
← stamina_ left03:41
Wolfman2000 As a reminder, what's the proper way to remove files in "Changes to be committed" but not actually remove them?03:42
EugenMayer sitaram: http://paste.debian.net/5269503:42
why are "groups" allowed there?03:42
Wolfman2000 tried git rm --cached -r . by mistake...almost removed the entire branch03:42
EugenMayer groups are then handled like repos..as they are only list either03:42
Wolfman2000: http://progit.org/book/ch2-4.html03:43
Wolfman2000 EugenMayer: Hmm...haven't sen this book yet.03:43
Wolfman2000 takes a look03:43
EugenMayer Wolfman2000: and before http://progit.org/book/ch2-2.html03:43
sitaram EugenMayer: again, I didnt understand the question "why are groups allowed there?"03:43
EugenMayer this pretty explains what is what Wolfman200003:44
sitaram: repo [one or more repos and/or repo groups]03:44
"one ore mote repos and or -->GROUPs<--03:44
repo @repos @admin03:44
RW = Someuser03:44
← MrGando left03:44
→ pdusen joined03:45
EugenMayer whould somehow create a repo for each user in admin, or?03:45
sitaram: I really don't want to waste your time, just trying to understand the configuration fully03:47
sitaram: in the 5-delegation docs "As you can see, for each repo group you want to delegate authority over, there's a branch with the same name in the gitolite-admin repo."03:47
you have that. But now branches have been created for me yet.03:48
Wolfman2000 EugenMayer: Definitely a nice, easy to understand book. I'll keep this in mind when I run into issues. Thanks again.03:57
EugenMayer yw03:57
→ cubuanic joined04:01
sitaram EugenMayer: sorry lost net connection for about 20 min04:07
← univate left04:07
sitaram EugenMayer: a group is like a macro -- gitolite does not distinguish between a list of users and a list of repos at the time a list is defined; it's how and where you use it that determines that04:07
→ alvarezp joined04:08
EugenMayer exactly...but that documentation just confuses people..04:08
anyway, I still can't see the branch and I also can't find and hints neither in the example.conf nor in the 5-delegation04:08
sitaram EugenMayer: what is confusing about "a group name is simply expanded to whatever names are on the right hand side when it is actually used"?04:09
→ nevyn_ joined04:09
→ tjaway joined04:10
EugenMayer sitaram: no need to discuss this. You expect people to place a list of repos there. Yes, people could use the lists they ment as groups here. And yes, it would work. And no this makes no sense. And no, this should then not be in the docs.04:10
← tjafk left04:12
sitaram EugenMayer: maybe I'm being more stupid today than I normally am, or maybe I have always been stupid and never knew, but I cannot understand what you are complaining about. If the delegation feature isn't working, let's dig into that. If you're talking about something more fundamental (like syntax) say so. I'm confused04:13
EugenMayer ok lets forget about the syntax. You are right. One thing at one time04:13
→ Ifrs joined04:13
EugenMayer I can't see any branches apear. And iam not aware why04:13
I have read the docs and maybe missed something.04:13
According to the docs those branches should apear automatically, but there is only one example, which is not generic enaugh to understand for me :04:14
oh no, I just found it :/04:15
sitaram EugenMayer: where does it say those branches appear automatically? if it does, my bad... meanwhile, read lines 78 onwards of doc/504:15
EugenMayer "As you can see, for each repo group you want to delegate authority over, there's a branch with the same name in the gitolite-admin repo."04:15
← tillberg left04:16
sitaram EugenMayer: in the config, yes...04:16
← cubuanic left04:16
EugenMayer right now its me who is stupid.04:16
sitaram not (yet) in the repo; that comes a couple lines later04:16
→ gbacon joined04:17
sitaram EugenMayer: I am starting to believe I wrote *too* much documentation. Someone did say that to me once, and I laughed it off. Now I'm not sure.04:17
EugenMayer is the conventing to <branchname> == <configfilename> ?04:17
sitaram: I seriusoly tried, but maybe iam still overtired..04:17
sitaram Maybe Tv was right after all... just put the software out, minimal documentation, and tell people "hop on to #git and hope for the best" as his README says04:18
sitaram sighs deeply04:18
→ thenetduck joined04:18
EugenMayer I think its pretty good you made the docu04:18
I rather tend to not used not doc`ed software04:19
sitaram not if it more detailed than people expect and they wont read the *whole* thing and come back with question on para 2 that are answered in para 404:19
just gets tiring04:19
← kukks left04:19
EugenMayer sorry.04:20
and, you won't believe, I have read this section 2 times.04:20
and do you know, what small detailed I missed?04:20
http://paste.debian.net/5269804:20
look at this.04:21
I thought, this are 2 REPO access defintions,not a single one04:21
so that way http://paste.debian.net/5269904:21
+ I simply did not recognise ( why ever ) this04:21
RW branch = list syntax04:22
eventhought its mentioned here : (R|RW|RW+) [zero or more refexes] = [one or more users]04:22
but as I did not cover the branch-docs yet, as I don't need branch protection yet, I could make the whole picture.04:22
← nevyn left04:23
← hakunin left04:25
← pdusen left04:26
← imajes left04:27
→ JoeConyers joined04:27
→ pdusen joined04:32
← xy left04:45
Ifrs what is git04:48
EugenMayer http://www.google.de/search?hl=de&q=what%20is%20git&meta=04:48
Ifrs do i need version controll for my windows office work04:50
EugenMayer http://www.google.de/search?hl=de&q=+do+i+need+version+controll+for+my+windows+office+work&btnG=Suche&meta=&aq=f&oq=04:51
Ifrs hm, they say git suxx04:51
guess its not that good04:51
← Ifrs left04:51
EugenMayer Anti-troll mission complete04:52
straszheim nice wrok04:53
work04:53
→ ProperNoun joined04:54
→ therrg joined04:54
→ pdusen_ joined04:55
← ankit9 left04:55
EugenMayer hey! git does save file-attributes?05:01
selckin only +x05:01
EugenMayer ah, ok. But this is anway great.05:02
esp. +x05:02
← pdusen left05:07
→ hakunin joined05:17
→ cytrinox_ joined05:20
← john268 left05:21
→ johnnyicon joined05:22
→ ff joined05:22
johnnyicon Anyone willing to lend me a hand with setting up Gitosis?05:23
← agib left05:23
sitaram what problem did you encounter so far?05:26
johnnyicon I can't clone a repository that I've created on my remote server05:26
I think it might have to do with the naming conventions of my user names05:26
i used [email@hidden.address] as my membername. Similarly, I created a RSA key using that same email05:27
When i tried to clone the repo I set up, I got the following error: ERROR:gitosis.serve.main:Repositroy read access denied05:28
fata: The remote end hung up expectedly05:28
I noticed in the example.conf, the @ symbol is used to refer to gitosis groups....05:28
so I guess I can't use [email@hidden.address] on the members line... am I right?05:29
sitaram I suggest not using @; user "foo" in conf file maps to keydir/foo.pub -- keep it simple05:29
johnnyicon ahh k so the member names map to the keys05:29
sitaram no05:30
to the keyfile name05:30
johnnyicon sorry, yea that's what i meant :)05:30
sitaram the comment inside the key (at the end of the long string) is irrelevant for all but the admin user05:30
johnnyicon i was thinking it had something to do with comment when I used ssh-keygen -C05:30
:) hehe05:30
k ill give it another shot05:31
← krynnotaur left05:32
← cytrinox left05:35
cytrinox_ → cytrinox05:35
EugenMayer sitaram: well you remember asking me the same? :)05:38
← hipe left05:39
→ hipe joined05:39
← bcardarella left05:47
→ john268 joined05:52
johnnyicon Still there Sitaram?05:54
EugenMayer is there a kind of "rebase dryrung" where I could see if errors would be accurr?05:54
johnnyicon I changed the filename to just john.pub and updated the conf file. Pushed my changes to my server. same error :(05:54
← qrush left05:55
→ ff_ joined05:57
← ff left05:57
EugenMayer how is the user named in the config file johnnyicon05:57
johnnyicon just john05:57
EugenMayer please post your repo difintion to a pastebin05:57
delete private data ofc05:57
also, go to the place, on the server, where your reporsitories reside05:58
please go in the gitolite-admin/hook folder05:58
and give me the output of "ls -la |grep update"05:59
johnnyicon k05:59
do you mean gitosis-admin?06:00
EugenMayer yes06:00
is there a partical reason you use gitosis and not gitolite?06:02
johnnyicon i only knew of gitosis06:02
EugenMayer well you should read this page :06:03
http://github.com/sitaramc/gitolite06:03
I don't think there is any reason not to switch to gitolite. gitolite is based on gitosis. There are some really cool features, which got added ( branch restrictions )06:04
johnnyicon ah k, yea i guess im behind06:04
is the set up relatively the same?06:04
EugenMayer well, for me, it was06:04
download gitolite onto my client06:04
cd src06:04
gl-easy-install and follow the stpes06:05
johnnyicon k cool06:05
EugenMayer be aware that you need a non-root user on the new system which is NOT used for normal maintainance. You will not be able to login to a normal shell with this user anymore06:05
johnnyicon am i going to have uninstall gitosis first?06:05
EugenMayer there is an upgrade-docu also, if you like to migrate06:06
http://github.com/sitaramc/gitolite/blob/master/doc/1-migrate.mkd06:06
johnnyicon ah sweet06:06
k... well that user requirement is going to be an issue06:06
my webhost only gives me one ssh user06:06
and i cannot create my own users :(06:06
EugenMayer well ask sitaram if there is way arround06:07
johnnyicon k06:07
EugenMayer the only way I see would be using key-auth for gitolite06:07
sitaram johnnyicon: it is not correct to say that you cannot logon to that user anymore06:07
EugenMayer and password auth for maintainance06:07
sitaram EugenMayer: please... that is not true06:07
johnnyicon so I have nothing to worry about then?06:08
sitaram johnnyicon: do you have password access for emergencies? (not strictly needed, but asking anyway)06:08
← plediii left06:08
johnnyicon do you mean using the passwd command?06:08
sitaram no06:09
johnnyicon sorry, what do you mean then? password access to...06:09
sitaram johnnyicon: let me rephrase. Can you log onto your "one ssh user" from anywhere by supplying a password?06:09
johnnyicon yes :)06:09
← pdusen_ left06:10
sitaram ok here's what you do. 3 simple steps if your workstation is linux06:10
johnnyicon yea im on a mac06:10
sitaram rats06:10
never mind we'll make it work :)06:10
how do you currently log on to the "one ssh user" account?06:10
johnnyicon k :D06:10
ssh06:10
→ kanru joined06:11
johnnyicon username@hostname06:11
sitaram does it prompt for a password?06:11
johnnyicon yea06:11
← sephr left06:11
johnnyicon f.... i think gitosis messed shit up06:11
i can't login anymore06:12
sitaram aah06:12
johnnyicon PTY allocation request failed on channel 006:12
ERROR:gitosis.serve.main:Need SSH_ORIGINAL_COMMAND in environment.06:12
luckily i still have an open SSH session06:12
EugenMayer no comment.06:12
johnnyicon I can remove gitosis... I don't know what to remove though06:12
sitaram johnnyicon: even if you didn't you'd be fine. All youd have to do is temply rename ~/.ssh to something else and log in, it would ask for password06:12
johnnyicon ah k06:13
sitaram johnnyicon: its people who dont even *have* password based login, or have disabled it on server (you *can* tell ssh not to support password logins you know) they might have a problem at this point06:13
johnnyicon ahh06:14
glad im not one of them06:14
sitaram yeah; is why I asked that first question :)06:14
johnnyicon :)06:14
EugenMayer well this is a way to late. Both gitosis and gitolite overtake the shell06:15
sitaram johnnyicon: look even though I'm the author of gitolite, if you've already installed gitosis and it serves your needs, dont worry about upgrading...06:15
EugenMayer and - for my security standards - key-auth login is pretty default. So you gonna lock out users as far as I understand the issue06:15
johnnyicon well it doesnt seem to be working :(06:16
sitaram EugenMayer: not if you have 2 keys06:16
EugenMayer I _had_ 2 keys.06:16
and iam not new to "ssh key auth".06:16
call it a debian lenny issue, but ssh pretty does not care much06:16
where you place this "command" to. It simply uses it for both keys, eventhough one of the keys does not have it.06:17
sitaram EugenMayer: you had only one key, and you're the reason I put in (1) a refuse to run as root check and (2) an extra warning saying "is this your only key"06:17
EugenMayer I had and have 2 keys :)06:17
I created a second key for root for the installation process before I started06:17
anyway, its time for johnnyicon06:18
sitaram 11:45:23 < EugenMayer> well this is a way to late. Both gitosis and gitolite overtake the shell06:25
EugenMayer: explain ^ please?06:25
→ plediii joined06:28
EugenMayer sitaram: when I started installing gitlote, I had a root shell, key auth only on my VM06:29
before I started to install gitolite, I created a new pair, so gitolite can login with the root user without using my current "default" key06:30
sitaram and the key was called EUgenemeyer.pub not id_rsa.pub06:30
EugenMayer no, my key was not named like this. Its named how the server is named.06:31
and the new gitolite key was named EugenMayer.pub06:31
after install gitolite, I could not login with my old key anymore. Looking at the servers autheticed_keys06:31
there was a specail command for the second key...which acutally is planed only to be exectuted when the first key is used. In my case, it always gots executed. The only false-positive I can think of here is06:32
that my ssh-agent had the key EugenMayer.pub in use. BUT the server asked for the passphrase of the other, older key, so I suppose this is not the false-positive06:33
sitaram EugenMayer: http://colabti.org/irclogger/irclogger_log/git?date=2009-11-26#l103506:34
EugenMayer yeah _pub_ keys06:35
in my _workstation_ .ssh06:35
why should I keep pub keys there?06:35
this was only one pub key, because that's exactly the one I created for gitolite some minutes before06:36
→ dtrott joined06:37
sitaram 11:46:21 < EugenMayer> I _had_ 2 keys.06:39
12:06:18 < EugenMayer> this was only one pub key, because that's exactly the one I created for gitolite some06:39
← Bass10 left06:39
sitaram please make up your mind06:39
EugenMayer ?06:41
← vbgunz left06:41
sitaram EugenMayer: I am trying to tell you that gitolite install does not, normally, screw you shell access. Indeed the easy install script itself *uses* the shell access you have. But normal people have their existing shell access using (say) "id_rsa.pub", and then run easy install with 3rd arg as (say) "eugenmeyer".06:44
EugenMayer: your *existing* shell access used the same key as the 3rd argument in the easy install06:44
EugenMayer this seems not to work, that's the point06:44
sitaram EugenMayer: after you did this, I have now put in a warning -- you might remember this06:44
EugenMayer it not about the workstation06:44
its about the servers ~/.ssh/authenticated_file06:45
if you use the _same_ user gitolite used for installing, looing in with a other key, you file ( as far as I understand )06:45
using password auth OR some other user will work without a problem06:46
sitaram EugenMayer: I do this every day because I do other stuff on that server so please do not tell me it doesn't work06:46
← altrux left06:46
EugenMayer ok fine06:46
sitaram EugenMayer: http://github.com/sitaramc/gitolite/blob/pu/doc/6-ssh-troubleshooting.mkd explains how very clearly06:46
it gives nice examples too, IIRC06:47
read from "explanation" onward06:47
EugenMayer: and "why two keys on client"06:48
EugenMayer sitaram: please simply forget about this. Iam not trolling and I don't bother you any longer.06:48
sitaram EugenMayer: thank you06:49
← Adlai left06:58
← johnnyicon left06:59
→ plediii_ joined07:00
sitaram EugenMayer: I also request that you not mislead people by saying "You will not be able to login to a normal shell with this user anymore". That is when I stepped in today with johnnyicon07:00
"with this key" would be accurate, not "with this user"07:01
EugenMayer (07:07:33) Eugen Mayer: the only way I see would be using key-auth for gitolite07:02
(07:07:42) Eugen Mayer: and password auth for maintainance07:02
→ bousquet joined07:03
← juan_arandaalvar left07:04
sitaram EugenMayer: not true; I use key-auth for both shell access and gitolite. Two different keys of course07:05
EugenMayer and it is true for me. So lets gonna fight? ..07:06
sitaram (I dont even remember the password now, actually... hmm, could be a problem if I ever lose the priv keys but I have physical access to that box anyway)07:06
← plediii_ left07:06
sitaram EugenMayer: I dont fight. Sorry. If it didnt work for you, I can help you make it work, but fighting is pointless.07:07
EugenMayer Well iam not going to fight and I don't even need help, as its woking.07:07
← hyperair left07:10
→ hyperair joined07:11
→ JasonWoof joined07:11
→ anno^da joined07:13
← plediii left07:13
→ aspotashev joined07:14
→ Sigma joined07:17
← yann left07:18
← aspotashev|eeepc left07:20
→ juan_arandaalvar joined07:20
← juan_arandaalvar left07:20
← jackylee left07:23
→ tedoc2000 joined07:30
← jaysern left07:35
← ff_ left07:38
← Sigma left07:38
→ bambule joined07:44
← dtrott left07:47
→ robinsmidsrod joined07:48
→ plediii joined07:49
→ mbroeker joined07:52
← cloudhead left07:53
← tedoc2000 left07:53
← EugenMayer left07:58
← therrg left08:02
→ Nathaniel1 joined08:03
Nathaniel1 → lhvwb08:03
← lhvwb left08:04
→ jaysern joined08:04
VVSiz_ → VVSiz08:06
→ mdel joined08:07
← hipe left08:09
→ hipe joined08:09
← hipe left08:12
→ hipe_ joined08:12
← Twisted_Mentat left08:14
→ NET||abuse joined08:15
→ _ponto joined08:16
→ sis joined08:20
DrNick2 → DrNick08:21
→ irc2samus joined08:29
← ponto_ left08:32
← mdel left08:33
→ corni joined08:43
→ DavidKlein joined08:45
→ vbgunz joined08:48
← DavidKlein left08:50
→ DavidKlein joined08:50
← hipe_ left08:57
→ hipe joined08:57
→ Axius joined08:58
← Axius left09:04
→ cubuanic joined09:08
→ Axius joined09:12
→ PN joined09:13
→ g3d joined09:13
→ node joined09:18
← thenetduck left09:18
← Axius left09:21
→ tbuehlmann joined09:21
← r11t_ left09:25
← ProperNoun left09:25
← aresnick left09:26
→ khmarbaise joined09:28
→ Axius joined09:29
→ tPl0ch joined09:33
tPl0ch how can I make file permissions being recognized by git version 1.6.3.3?09:34
node hi, I know i hae asked this question before, but I forgot how to do it and I can't find it on google. When I update my git repo, it tells me this. I wanna go ahead and actually delete these files from the repo. http://pastie.org/71899609:34
← tbuehlmann left09:34
→ tbuehlmann joined09:35
node I just need to make they files delete from the repository09:35
tPl0ch node: well, have you committed your change?09:36
node that's what prints out when i commit my changes09:37
tPl0ch and pushed it?09:37
node I can't push before I commit09:37
wouldn't make since09:37
i thought it was something like commit -u or something that did it09:38
or something09:38
oh I think it was "git add . -u09:39
"git add . -u" that did it09:39
tPl0ch node: I never had to do that to delete files from the repo09:39
node tPl0ch: really? ... git always asks me to do crap to delete a file09:39
i'm running a mac though09:39
vs linux09:40
tPl0ch git rm file, git commit -m "deleted files", git push09:40
node (kind of wish that wasn't the cast though)09:40
tPl0ch: oh well ya, but what if you have 50 files you deleted from the repo09:40
it would take forever to type out each 50 file you know what I mean?09:40
tPl0ch node, it was 2 files in your pastebin09:41
node tPl0ch: ya, your correct, I just needed to know because I have had a ton of deleted files before, like a whole directory09:41
of files09:42
thanks for the help though tPl0ch :)09:43
→ maleknet joined09:45
node → thenetduck09:45
← hakunin left09:49
→ hakunin joined09:50
→ aspotashev_ joined09:51
← aspotashev left09:51
→ ixti joined09:53
← jaysern left09:55
→ imajes joined09:55
← Axius left09:57
← irc2samus left09:59
→ yhager_ joined10:00
← yhager_ left10:02
→ yhager_ joined10:02
← pheaver left10:02
← vbgunz left10:03
← JasonWoof left10:03
→ cga joined10:05
cga hi all,10:05
up_the_irons, ping10:05
→ froschi joined10:06
cga up_the_irons, i'm following your gitosis guide but i have some problems that i don't understand since i'm new to both git and gitosis10:06
up_the_irons, can you help?10:06
→ voker57 joined10:06
RurouniJones cga - Standard practice on IRC is just ask the questin10:07
cga hi RurouniJones , ok =)10:07
RurouniJones If anyone can help they will. Asking a specific person if they can help is a sure-fire way of not getting a response10:07
→ Grahack joined10:07
cga RurouniJones, i asked him since he says so on the guide10:07
but glad if anyone can help10:08
here's the question:10:08
→ pantsman joined10:09
cga this is the guide: http://scie.nti.st/2007/11/14/hosting-git-repositories-the-easy-and-secure-way and when i do "git clone git@glenys:gitosis-admin.git" i get this error: http://pastebin.com/d517691010:09
→ voker57_ joined10:11
cga i guess it's beacuse i need to initialize a .git repo before, is it?10:12
→ node joined10:14
Grum if you followed that guide then the username is not cga10:14
← tetha left10:14
→ tetha joined10:15
Grum unless ofcourse you didnt follow the guide :P10:15
cga Grum, my bad, i used gitosis user now, this is the error: http://pastebin.com/d60c3aac510:16
Grum there is also no gitosis user in that story10:17
trying random things will not make it work i'm afraid :D, what user did you create?10:17
cga Grum, but my sistem uses gitosis and not git for that10:17
Grum ok, so you *didn't* follow the guide10:17
pantsman cag, you said: when i do "git clone git@glenys:gitosis-admin.git"10:18
you clearly didn't do that, according to either of your pastebins :)10:18
Grum cga: tell me what you did please10:18
cga Grum, what's the difference in installing gitosis with apt-get or compiling? it's only that i have to adapt the guide to my system. isn't it?10:18
pantsman err, cga*10:18
Grum erm no really10:18
→ radarek joined10:18
Grum apt-getting it sets it up as well10:18
in its own way10:19
← omerj left10:19
Grum following the guide will get you nowhere after that10:19
→ omerj joined10:19
← omerj left10:19
Grum also, if you read the guide, you should have seen it is actually python, there is not much compiling going on10:19
→ omerj joined10:19
→ pheaver joined10:20
cga Grum, i guess that as long as the gitosis user has a /home/gitosis and permission it should be teh same. no?10:20
Grum no?10:20
cga uhm10:20
Grum what did you do exactly? :(10:20
cga ok10:21
Grum because you didn't follow the guide at all, because you wouldnt even need the very first command10:21
cga apt-get install git-core git-doc gitosis, this creates the user gitosis and install in /srv/gitosis , then i up;paded the id_pub and imported it as gitosis user with gitosis-init < /tmp/id_rsa.pub10:22
← thenetduck left10:22
cga then i try that remote command that fails.10:22
Grum you cant run init twice, i assume the apt-getting already inits10:23
else it would leave a broken product10:23
well, i'd say, find the manual for the gitosis package someone made10:23
you need to know what parts of the installation it already did10:23
(and yes, the person who made gitosis has nothing to do with that)10:23
cga of course10:24
anyway it initialize twice as the guide says10:24
← Pupeno left10:24
cga with one command10:24
checking the README for that package10:24
→ Pupeno joined10:24
→ v0n joined10:25
← voker57 left10:25
Grum the installation is not much more than: make key; get gitosis; install gitosis in to a dir; init with your key; presto10:25
oh, i forgot; tinker in gitosis.conf :P10:25
cga this is what it creates after init: gitosis .gitosis.conf repositories .ssh10:25
not .ssh of course10:26
Grum what is in .ssh/authorized_keys?10:26
should be your pubkey with some 'command=......' prepended10:26
cga ok done10:27
→ Axius joined10:27
Grum done what? :(10:28
cga it was a permission error on sshd config :P i didn't include gitosis in ssh group which is the only one allowed to ssh10:28
→ jschoolc_ joined10:28
← giallu left10:28
Grum k, now that you've installed gitosis, upgrade to gitolite ;)10:28
cga thanks for helping so far10:28
arf10:28
what's gitolite?10:28
← aspotashev_ left10:28
Grum gitolite is actually documented, sane, simple, debuggable ;)10:29
cga cool10:29
Grum oh and more advanced10:29
cga gonna check it10:29
Grum and the author lurks this channel :p10:29
http://github.com/sitaramc/gitolite/10:29
cga is this: http://github.com/sitaramc/gitolite ?10:29
cool10:29
node → thenetduck10:30
← voker57_ left10:30
→ Dextorion joined10:31
→ d3xter joined10:32
→ icwiener joined10:32
← mike8901 left10:32
Grum cga: its really made after all the frustrations we've seen here in the channel :) gitosis works fine if it works but its really made for someone who got tired of doing it by hand10:33
cga Grum, i see. i'm going to try it10:34
Grum you can harras sitaram if you end up having problems with it ;)10:34
cga ok10:35
Grum sitaram: nice easy-install script but you might want to link back to the documentation you made when you give an error; ea line 187: die "pubkey access didn't work; please set it up using 'ssh-copy-id' or something"10:37
sitaram Grum: sadly, most people dont want to read docs. So any "die" is followed up by "run $0 without any arguments for help and tips"10:39
Grum ah! good =D10:39
sitaram Grum: and when they do *that*, they get (among other nuggets): Pre-requisites:10:40
- you must already have pubkey based access to user@host. If you currently10:40
only have password access, use "ssh-copy-id" or something equivalent (or10:40
copy the key manually). Somehow (doesn't matter how), get to the point10:40
Grum from the looks you are already being *really* nice to people :P10:40
sitaram where you can type "ssh user@host" and get a command line.10:40
(sorry for long paste; I thought it'd all be one-line, duh!)10:40
Grum: well "die" referred to the program not the user ;-) <LOL>10:41
Grum aaaaaw; (10:41
sitaram cga: despite all that, don't convert just for the sake of converting if gitosis is working fine for you and you don't really need the extra features in gitolite10:42
→ ff_ joined10:42
cga sitaram, honestly i just started with git and chose gitosis to automate things i don't really know and to have ssh enalbed repo managed. so i can use gitolite if it's so better.10:43
sitaram, my repo wn't be even pulbic or big. it's for my static generated blog. just to have something to push instead of using tarballs and scp10:44
sitaram cga: only one user? why do you need gitosis or gitolite? they're only useful if you have >1 user who needs access10:46
cga sitaram, that's what i as going to say: in the fuiture i'll host my code on my server. they'll be multi user and public. that's hwy i wanted to start hosting with something automgic10:47
i prefer to start this kind of things thining ahead and learn them with something smaller that no one sees or use atm10:48
sitaram cga: good point10:49
cga =)10:49
→ squentin joined10:49
← tPl0ch left10:50
→ Axius_ joined10:52
pcc1 is there a way of finding the most recent commit that modified a file X, that respects merges, i.e. if the file was modified in both branches then the algorithm should return the merge commit10:52
← Axius left10:53
cga ok see you later, shower time. thanks so far.10:57
→ galderz joined10:59
→ jas4711_ joined10:59
← galderz left10:59
sitaram pcc1: even if it was modified in both branches, if it was not modified *in* the merge itself (like maybe conflict resolution) I'm not sure if there is a way to show you the merge commit10:59
→ dsch04 joined11:02
→ ludde joined11:06
← thenetduck left11:12
← ff_ left11:18
← _ponto left11:23
→ Sigma joined11:25
pcc1 sitaram: thanks. this is an interesting problem. I'll need to think about it more11:26
→ intripoon joined11:27
sitaram pcc1: the git log manpage has a "-c" option; not sure if it does what you want though; try it out...11:28
→ c0ll3c70r joined11:28
→ Pupeno_ joined11:29
← c0ll3c70r left11:29
→ Axius joined11:29
← Pupeno left11:30
Pupeno_ → Pupeno11:30
← misiu_mp left11:31
→ spaceonline joined11:32
pcc1 sitaram: thanks, that may be (part of) the solution. it is not clear which column of the "git diff-tree -p -c" output corresponds to which parent. is it always in the order of the parent list given by "git show"?11:33
sitaram pcc1: no idea; sorry... I just looked at the man page looking for "merge" and saw that :)11:35
→ Laibsch joined11:36
← ia left11:36
Laibsch Hi11:36
→ ia joined11:37
← Axius_ left11:39
← cga left11:41
→ razvand joined11:43
→ admiral0 joined11:45
admiral0 hi i am having trouble with git svn11:46
git svn clone https://kmess.svn.sourceforge.net/svnroot/kmess/trunk/kmess -r 569911:47
it does nothing but make me a .git dir11:47
has anybody had this type of problems?11:48
← ixti left11:50
← spaceonline left11:51
→ fcuk112 joined11:53
← Axius left11:55
→ Pupeno_ joined12:00
← cytrinox left12:03
→ botanicus joined12:06
← bousquet left12:06
← pheaver left12:06
→ jfkw joined12:14
← hipe left12:15
→ hipe joined12:15
← Pupeno left12:16
Pupeno_ → Pupeno12:16
→ cytrinox joined12:19
→ cortana joined12:24
→ __iron joined12:24
charon admiral0: git svn clone is to clone entire histories, and there is no history of /trunk/kmess to be cloned in r5699 (as that commit only touches /branches/kmess/kmess-2.0.x).12:27
admiral0: if you do care about the history, use 'git svn clone' without revision limiters and then later check out the appropriate revision. if you don't, simply use svn clone as there would not be a point in using git12:27
s/svn clone/svn checkout/12:27
← Guest61672 left12:29
← khmarbaise left12:29
admiral0 thanks charon :)12:32
→ ab3 joined12:37
→ omerj_ joined12:38
← ia left12:39
← mbroeker left12:40
straszheim here's one: I'm in the plumbing, I've pulled down a blob from a remote via fetch-pack, this blob is a config file. I wantget some configured values out and then take further action. looks like you have to write the blob to a /tmp/file and then git config -f /tmp/file... is there a 'best practice' for this?12:41
→ inffcs00 joined12:43
charon (i thought remote transports weren't supposed to give you anything not pointed to by a ref?)12:44
→ ia joined12:44
straszheim hm?12:45
ls-remote: list of refs, fetch-pack --depth=1: commits trees and blobs12:46
← omerj left12:46
→ ph^ joined12:46
straszheim that wasn't clear12:46
i get the list of remote refs with ls-remote and fetch-pack them, then i look at their trees for a certain file12:46
→ solistic joined12:47
straszheim so i've got a blob that contains [mystuff] key = value, etc.... but no obvious place to put it12:47
straszheim wishes git-config would have a 'read from stdin' option12:48
jast might I suggest mktemp12:48
straszheim that'll work12:48
jast well, git config files aren't meant to be in the database, so nobody thought of that kind of option :)12:48
solistic Hello, I just can't figure out how to rebase on newly created remote branches with "git svn". What ever I do, the new remote branche does not show up with 'git brach -a'. Any hints?12:49
jast solistic: you might have to do a git svn fetch first12:49
straszheim jast: yeah, i follow you there12:49
→ Adlai joined12:50
solistic jast: thx, that seems to work, I only tried "git svn rebase"12:50
← Sigma left12:51
jast git svn rebase only fetches new stuff for the currently checked out branch iirc12:51
straszheim hmm, .gitmodules is an exception, those go in the database... but those live in the checkout12:52
jast yeah12:52
solistic jast: so should "git rebase new-svn-branch" && "git svn dcommit" commit my changes on "new-svn-branch" ?12:54
← cubuanic left12:54
jast solistic: I'm not sure. If, after the rebase, "git log" shows the new branch name in the git-svn-id line of the topmost log entry, then definitely yes.12:55
← Adlai left12:56
→ Adlai joined12:56
jast and I think it's likely to be that way, but I wouldn't bet my sanity on it.12:56
then again I don't have any sanity anyway.12:56
→ khmarbaise joined12:57
← maleknet left12:59
solistic jast: the git-svn-id of the topmost log entry refers to the new branch, so I'll give it a try :)12:59
→ ankit9 joined13:01
← EricInBNE left13:02
→ wvd joined13:04
→ Yuuhi joined13:08
straszheim am I asking for trouble if I make a subdir of $GIT_DIR for my own use?13:10
(one with a bizarre name that almost certainly won't collide with any normal git usage)13:10
→ Sigma joined13:12
→ reaVer joined13:14
← Sigma left13:15
← wvd left13:17
→ LukasLT10 joined13:18
← modsognir left13:18
← Laibsch left13:19
→ Sigma joined13:21
LukasLT10 sitaram: Hi, you recently helped me to set up a gitolite server13:21
sitaram LukasLT10: hi13:21
LukasLT10 sitaram: however i set up my ssh key with passphrase, but now i'd to have a passwordless login13:22
and i'm afraid to break thongs down again :(13:22
sitaram LukasLT10: passphrase != password :) Can I assume you mean the passphrase protecting your private key?13:23
LukasLT10 I guess i have to regenrate shh keys, but how to upload the to server?13:23
sitaram no no no13:23
no13:23
:)13:23
LukasLT10: Can I assume you mean the passphrase protecting your private key?13:24
LukasLT10 yes13:24
but the one i created originaly is 25 chars lenght, and a bit too long to be retyped each time13:24
→ Paraselene__ joined13:25
sitaram you have 2 options: (1) use a program like ssh-agent or keychain to manage this for you (2) use ssh-keygen's "-p" option to remove the passphrase on your existing key13:25
LukasLT10: there is *absolutely* no need to regenerate keys for this13:25
← Adlai left13:25
sitaram LukasLT10: passphrase protection is completely local to the workstation; no need to involve the server in any way either13:26
jjuran LukasLT10: ssh-agent comes highly recommended.13:26
→ sergiodj_ joined13:26
→ Adlai` joined13:26
sitaram LukasLT10: a bit pf time spent exploring those alternatives will give you years of security13:26
LukasLT10: I type my passhrase (23 characters long by the way, just 2 short of yours) only once per login13:26
linux, using "keychain", does all this magically13:27
yhager_ → yhager13:27
sitaram (keychain internally uses ssh-agent by the way)13:29
LukasLT10 i'll tha a look at them :)13:30
i'll take a look at them :)13:30
sitaram LukasLT10: back to gitolite, if you do ever need to upload a new set of keys to the server, it's easy. Just add a new key, put it in keydir/ as "lukas@key2.pub" (keep the part before the "@" the same as your existing gitolite username), add it, commit, push. Done13:31
jjuran sitaram: Mine's over 100 characters. :-)13:31
sitaram jjuran: you rock!13:31
LukasLT10 by the way, is there a way to block some files from being added by add . ?13:31
sitaram jjuran: I, however, type like a rock :( so if I had it that long I'd be logging on for 20 minutes :)13:31
→ Rayne joined13:31
jjuran LukasLT10: use .gitignore13:32
LukasLT10 like eclipese o project settings files?13:32
sitaram nods13:32
→ psoo joined13:32
→ Stravy joined13:32
Rayne does gitosis (i hope that i am in the right channel) assigns commits to the virtual users or does gitosis "forget" about the commiters?13:33
jjuran I just got git to compile with Metrowerks C 2.4.1 without major fuss.13:33
wereHamster Rayne: gitosis can't influence the commiter. The commiter is set when you commit in your local repository13:34
jjuran The problem is that it won't initialize an aggregate with values determined at runtime. The trick is to use the C++ translator for those functions.13:34
Adlai` → Adlai13:34
Rayne wereHamster, good news :)13:34
jjuran sitaram: ^^^13:35
sitaram jjuran: ?? (about the Metrowerks thing?)13:37
jjuran yes13:37
→ Tommy[D]_ joined13:37
← solistic left13:37
sitaram jjuran: oh... however, my knowledge of C is largely historical, like Al Gore's chances of becoming president :)13:37
jjuran The point being my patches for git are now maintainable and worth publishing.13:39
sitaram jjuran: aah -- sorry :) didnt think of that!13:41
jjuran: so, posting it to the ML?13:41
s/it/them13:41
← JEEBcz left13:45
jjuran sitaram: yup, after I get outgoing mail configured.13:45
→ bauruine joined13:46
← anno^da left13:47
sitaram yeah that's always a pain for me, since I use gmail :) I know there are some scripts to help and the documentation even has a section on it; have to read...13:47
→ parasti joined13:47
→ remi joined13:48
→ aziz joined13:50
→ mbroeker joined13:51
→ j416 joined13:51
← cytrinox left13:52
→ JEEB joined13:53
→ iron joined13:54
→ cytrinox joined13:54
cbreak gmail doesn't support imap?13:56
siks it does13:57
→ quizme joined13:57
quizme how do you erase the last commit completely ?13:57
siks settings -> forwarding and pop/imap13:58
(to cbreak, not quizme)13:58
cbreak quizme: git reset. but only do that if you know what you are doing and din't push yet13:58
j416 quizme: you have to remove all references to the commit (that is, in branch(es) and the reflog, tags etc.), then you can do 'git gc' and 'git prune' to remove it completely.13:59
if that is in fact what you want..13:59
quizme there is only the master branch13:59
j416 no, there is your reflog as well.14:00
Glenjamin is that re-flog or ref-log ?14:01
j416 reflog14:01
jast its a log of refs, if that answers your question :)14:02
*it's14:02
jjuran Sometimes one flogging isn't enough :-)14:02
j416 log of refs14:02
→ gazra joined14:02
jast oh... actually, a reflog is the log for one ref, not several.14:02
quizme: anyway, point is, by using something like reset the commit will disappear from the history, but git will keep the object file(s) around for several weeks in case you want to recover the data14:03
→ vdidel_ joined14:03
← hipe left14:04
quizme jast: how do i get rid of the object files14:06
prune ?14:06
← gazra left14:07
jast that's where it gets complicated, as people have already hinted at. you'll have to clean at least two reflogs in the general case before you can repack and prune.14:09
← __iron left14:09
→ spaceonline joined14:12
← ab3 left14:16
jast for example: git reflog --expire=0 HEAD refs/heads/yourbranch; git repack -Ad; git prune14:16
→ hgimenez joined14:16
Rayne does gitosis support disc/traffic quota or is the traffic countable by iptables for the different ssh keys ("users")?14:16
quizme thanks14:17
jast Rayne: I think the answer is: no to both14:17
quizme a bit too complicated for my blood14:17
Rayne github uses something that they developed (only the git backend) for themself (or customers which want pay for github.fi)?14:19
jast probably, yes14:20
← remi left14:20
→ cenuij- joined14:21
← Rayne left14:22
← maattd left14:23
→ maattd joined14:23
→ LiamH joined14:30
→ Sho_ joined14:30
→ laj_ joined14:31
← cenuij- left14:32
← j416 left14:32
← Pupeno left14:35
→ yarddog joined14:37
← cenuij left14:37
yarddog → Guest3889714:37
← mbroeker left14:38
→ mithro joined14:38
← Natch| left14:38
→ Natch| joined14:39
→ rudi_s joined14:51
→ munichlinux joined14:53
sitaram (he left, but) github has a custom ssh that hooks into a MySQL db instead of reading authkeys...14:56
← mithro left14:57
→ cenuij- joined14:59
→ munichpython joined14:59
← munichlinux left15:00
munichpython → munichlinux15:00
← spaceonline left15:00
← Tommy[D]_ left15:01
→ lhz joined15:07
← radarek left15:08
→ krunk- joined15:09
→ reprore joined15:13
→ Bass10 joined15:14
→ will_h joined15:19
→ cga joined15:20
→ kumbayo joined15:21
→ mbroeker joined15:22
LukasLT10 might be someone uses Eclipse's GIT plugin?15:25
cga hi all, i get [remote rejected] master -> master (hook declined) on git push origin master. the set up is as follow: remote server repo is /var/nginx/nanoc3/cga.cx/.git , i did git init in that dir , then git add . and git commit ; afteward i created dir ~/cga.cx and git init plus git remote add origin cga@glenys:/var/nginx/nanoc3/cga.cx/.git and git pull origin master. edited some file , git commit -m "test" + git push origin master , where do i fail? thanks15:26
→ spaceonline joined15:27
← quizme left15:31
jast cga: you've got a hook script installed in the remote repository that aborts the push15:31
→ aspotashev joined15:32
cga jast, oh i see. i guess that was created with git init. how do i know which is it?15:32
← Grahack left15:33
jast cga: probably the pre-receive or upload hook. if you don't need them, simply make sure they're not executable.15:34
cga jast, ok i'll have a look at them15:35
jast err, update hook15:35
cga ok15:35
gotta go now15:36
thanks15:36
LukasLT10 sitaram: is it possible to re-enable simple user+password login, since eclipse's git plugin does not accept passwordess way :(15:37
→ munichpython joined15:38
← munichlinux left15:38
munichpython → munichlinux15:38
← kpreid_ left15:39
→ kpreid_ joined15:39
← parasti left15:41
→ Determinist joined15:45
← kanru left15:46
wereHamster cga: did git push not print any additional information why it was declined?15:47
← luke` left15:48
→ dirk2 joined15:48
← bauruine left15:48
→ Chase_ joined15:49
Chase_ → gbit6915:49
gbit69 → gbit15:49
gbit hey guys...just getting started with Git and trying to figure out the best setup for myself15:49
→ djszapi joined15:53
djszapi http://pastie.org/71920515:53
How can I fix it ? git reset --mixed or how ?15:53
→ Kissaki joined15:55
← flaguy48 left15:56
Kissaki what's the best way for a simple conversion of a bzr repo to a git repo? there's only one branch, not tags...15:58
PN → ProperNoun15:58
→ rudi_s_ joined15:58
djszapi http://pastie.org/71920715:59
Ooo.15:59
if someone could help me, it would be nice :(15:59
Kissaki Also, I'd only like to harmonize my commit-names (they changed through time, 3 values). Either while converting anyway or afterwards, if possible.15:59
→ yann joined15:59
me22 djszapi: iirc a reset --hard will get you out of the intermediate merge state16:00
djszapi me22: Will I lost then my last git stash operation ?16:00
So I did a git stash before this pulling.16:00
I'd like to preserve that.16:00
me22 IIRC stash commits a branch, so I doubt it, but I'm not sure16:01
→ Pupeno joined16:01
djszapi what do you doubt ?16:01
sitaram LukasLT10: not with gitolite -- nor even gitosis actually. They depend on ssh auth keys to distinguish one user from another16:02
→ bx2 joined16:02
sitaram LukasLT10: eclipse's git plugin cannot be so stupid; I know I have colleagues who are using it -- I can check with them tomorrow16:02
djszapi the basic question is whether git reset --hard delete the stash state.16:02
the stash list16:02
cbreak if you want to meddle with the stash16:03
use git stash16:03
djszapi ?16:03
not sure I understand16:03
cbreak git reset does not care about the stash16:04
djszapi so can I do a git reset --hard without worrying ?16:04
Will my stash content be preserved ?16:04
me22 djszapi: I just haven't used git in anger, so I'm uncertain about most things16:05
djszapi doener, Ilari: where are you ? :)16:05
→ davey joined16:06
→ flaguy48 joined16:10
straszheim more and more boost people showing up on #git, very good, very good16:10
→ hugoxrosa joined16:11
→ amirite joined16:11
jjuran Yes, git reset --hard doesn't affect the stash16:11
me22 yay; intuition worked again16:11
jjuran djszapi: ^^^16:13
LukasLT10 sitaram: yey :) i found out a way to make it work16:13
← munichlinux left16:13
sitaram ok16:13
I was sure it would work -- would be meaningless otherwise (but I dont use Eclipse so would have had to ask my team tomorrow)16:14
← rudi_s left16:15
← spaceonline left16:17
← gbit left16:18
straszheim here's one: can i get the config from a remote w/o pulling the whole repo down?16:19
→ brennino joined16:19
← brennino left16:20
→ hurug joined16:21
← davey left16:21
→ davey_ joined16:21
hurug hi, how can I restore a file of a previous commit in git?16:22
→ EugenMayer joined16:22
EugenMayer Hello. How would I determine if a rebase / merge actually will need to merge files, regardless if this will be resolved or and in a conflict. Is the "make a upgrade branch and test it" the only approach given here? Actually iam trying to determine, if a merge / rebase can be 100% safely done by a bash script ( safe == no merges on file-base at all )16:22
→ doener joined16:23
← LukasLT10 left16:23
→ LukasLT10 joined16:24
LukasLT10 but the way it handles projects is quite strange and not very comfortable :(16:24
wereHamster EugenMayer: just do the merge/rebase and see if any files are left unmerged or if there were rebase conflicts..16:25
EugenMayer and if, reset?16:25
wereHamster yes16:25
← hurug left16:26
EugenMayer well, maybe I should start my question the other way.16:26
How can I check, if the current repo has changes, his origin does not have16:26
← hugoxrosa left16:26
EugenMayer git status I guess. With this informaiton16:26
selckin EugenMayer: never used this i'm just reading random man pages, git-merge-tree might not modify the state?16:26
EugenMayer iam reading the git-merge manpage16:27
← davey_ left16:29
→ anno^da joined16:31
← psoo left16:32
← khmarbaise left16:33
→ ntoll joined16:36
unixtippse EugenMayer: git diff master...origin/master <-- From my shell history.16:38
→ jayne joined16:38
unixtippse EugenMayer: You need to fetch (not pull) origin/master before diffing.16:39
EugenMayer well this is cool , as it generates the merge16:39
→ roop joined16:40
← roop left16:40
← Dextorion left16:40
EugenMayer thank you16:41
unixtippse EugenMayer: NP, I'm happy to help with the little git that I know. ;-)16:41
→ cilly joined16:41
EugenMayer wel iam extending braid to be able to send out emails of summaries if it could not smoothly update the nested repos16:42
→ remi joined16:42
EugenMayer so like: "braid check --update --nomerge --summary-email=you@here.tld --email_verbosity=error16:43
unixtippse I'm afraid I haven't had any contact with braid so far.16:44
EugenMayer well I think its rarely used16:44
← Adlai left16:46
→ vbgunz joined16:47
EugenMayer unixtippse: is git fetch origin master16:48
doing the same as git remote update, only just for the master branch?16:49
unixtippse I must say I haven't had any contact with "git remote update" either. Is that shorthand for fetching all tracked remote branches? :-)16:49
→ ixti joined16:49
→ tedoc2000 joined16:51
EugenMayer http://paste.debian.net/5273216:51
that's in the man, but I can call, if this does the same :/ Could someone clarify?16:52
→ BlindOracle joined16:54
EugenMayer is it possible to do a git diff master...origin/master and actually make a default std out, without using the editor? I know that > file works, but how to stdout default?16:55
cbreak git diff doesn't use an editor here16:55
selckin export GIT_PAGER=cat or some such16:55
cbreak or | cat16:56
if you don't want a pager16:56
selckin or like >&116:57
→ ph^_ joined16:57
selckin or git --no-pager!16:57
ok back to work16:57
EugenMayer http://linux.die.net/man/1/git-diff16:58
can't find your --no-pager switch here ( its not working either)16:58
unixtippse Works for me.16:58
selckin git --no-pager diff, man git16:58
EugenMayer ah! it's a git switch. Well I still have a lot to learn :/17:01
git diff --no-pager .. does not work17:01
git --no-pager diff .. does work17:01
thank you!17:01
← djszapi left17:01
→ tedoc2000_ joined17:02
← cilly left17:03
unixtippse wonders whether git diff master...foo/master is in fact the preferred way for reviewing remote changes, or if there is anything else.17:03
← tedoc2000_ left17:04
offby1 seems reasonable tome17:04
→ cilly joined17:04
offby1 I do the opposite to review my _own_ changes before pushing -- git diff foo..17:04
EugenMayer well that command only seems to check incoming changes, not outgoing17:05
→ Aikawa_ joined17:06
offby1 really?17:06
works for me17:06
strictly speaking, I do "git diff trunk..", where "trunk" is a git-svn remote17:07
EugenMayer I have incoming commits from the origin, but I also have local changes ( I did not yet add them with diff add to the stage )17:08
diff does not show me the diff of that file17:08
unixtippse They will only be displayed when they have been commited.17:08
→ futoase joined17:08
EugenMayer git --no-pager diff origin/master does the job for me. Yeah, I undestand17:09
→ khmarbaise joined17:09
unixtippse Yes, because that way you're diffing against your staging area. (Or your working copy? Hmmm...)17:09
LukasLT10 chmod for hooks should be -rwx------ ?17:09
← tedoc2000 left17:09
← futoase left17:09
← vital left17:13
EugenMayer LukasLT10: for gitolite / gitosis? yes17:13
unixtippse: my one is against wc17:13
wereHamster LukasLT10: they need to be executable by git, that's all17:13
← Bass10 left17:13
← khmarbaise left17:13
→ bugxek joined17:13
EugenMayer unixtippse: yours might be staging or commited17:14
LukasLT10 strange i can't get them fire :(17:14
→ Bass10 joined17:14
LukasLT10 post-update is executed after push?17:15
← jlilly left17:15
EugenMayer The user you connect with must be able to execute them, is git is running by this user17:15
← Bass10 left17:16
→ Bass10 joined17:16
→ didi joined17:18
LukasLT10 supid me, exec git-update-server-info exided before my code should be executed :D17:18
→ voker57 joined17:18
→ Pupeno_ joined17:18
→ spaceonline joined17:19
← Determinist left17:19
didi I want to get just the contrib/emacs from git.kernel.org/pub/scm/git/git.git. Is it possible? Or, is it possible to get just a directory from a repository?17:20
← bugxek left17:21
Wolfman2000 is starting to understand the appeal of branches. :)17:21
← Aikawa left17:22
Wolfman2000 Alright, I'm not a total conversion case, but I'm close.17:22
EugenMayer Wolfman2000: read the book :) 3.4, 3.5 and 3.617:22
me22 EugenMayer: which book?17:22
Wolfman2000 EugenMayer: I did read those sections.17:22
EugenMayer Did you like the way its descibed?17:22
me22: http://progit.org/book/17:23
doener didi: with history? No. As of now, you can only clone repos as a whole17:23
Wolfman2000 Keeps it simple. So far, I won't need to use much of what's in 3.4-617:23
3.2 and 3.3 keep it simple enough for what I'm doing right now17:23
didi doener: And just the HEAD files?17:23
me22 thanks, EugenMayer17:23
LukasLT10 If i want to work with more than 1 repo i same script, i have to unset GIT_PATH before switching to next repo?17:24
doener didi: the kernel.org gitwed allows to get snapshots of arbitrary tree objects17:24
didi: http://git.kernel.org/?p=git/git.git;a=tree;f=contrib/emacs;h=dcd71f6e77ec89d788595d8fd5e0a104b3fb2e8a;hb=66abce05dd5b9da9c889034781dc3de38b6e231b17:24
Gitbot [git dcd71f6e7]: http://tinyurl.com/y9cd46f [tree]17:24
didi doener: I see. Thank you.17:25
EugenMayer How would make a diff of all : not updated, staged, commited changes to a specific branch? -- does not-updated <-> branch , --cached does staged <-> branch .. how to combine?=17:26
→ lresende joined17:26
← Bass10 left17:27
→ Bass10 joined17:27
→ virginia joined17:27
→ voker57_ joined17:28
← spaceonline left17:29
doener EugenMayer: hm? "git diff" ("--" makes no difference) is working tree vs. index17:29
EugenMayer git --no-pager diff HEAD17:29
doener EugenMayer: "git diff HEAD" is working tree vs. HEAD17:29
EugenMayer this is what I was looking for17:29
ah ;-)17:29
← madewokherd left17:30
EugenMayer just for pure interest, how would I also add the "not tracked" files to it?17:30
doener EugenMayer: that's not a combination of "git diff" and "git diff --cached" though17:30
← voker57 left17:30
doener EugenMayer: git add -N <pathspecs>17:30
EugenMayer: -N means "intent to add", the paths get a special entry in the index that makes "diff" show them as new files17:31
← kumbayo left17:31
EugenMayer ah, ok. Well git is huge :)17:31
cga does "git add ." looks recursively?17:31
← didi left17:31
EugenMayer compared to my SVN knowledge, its huge. And I love it17:31
cga: yes if you do git add .17:31
→ Axius joined17:32
cga mmmm ok , that is strange then. it adds the files but if i commit then i don't find the files on the server.17:32
even if it says it has committed17:32
doener cga: committing is local17:32
EugenMayer did you check with git "status"?17:32
selckin cga: faq bare17:33
selckin pokes Gitbot17:33
doener selckin: faq non-bare17:33
Gitbot selckin: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare17:33
cga nope EugenMayer , thanks selckin17:33
→ jaysern joined17:33
doener selckin: but as I read it, cga only ran "git commit", not "git push"17:33
selckin possible17:34
doener (though of course, he'd probably run into the non-bare push thing next)17:34
→ d0k joined17:34
← Pupeno left17:35
cga doener, also push. my bad not saying it clear17:35
Pupeno_ → Pupeno17:35
EugenMayer doener: how are changes called, when they are submitted? "snapshotted" ?17:36
→ madewokherd joined17:38
doener EugenMayer: committed?17:38
EugenMayer Well yeah, they are commited. That actually means, they are in a snapshot : http://progit.org/book/ch2-2.html17:39
just thinking about this picture17:39
doener EugenMayer: well, if you run "git add foo" to 'stage' foo, git already takes a snapshot of foo's contents17:40
EugenMayer well git add file does not stage?17:40
← Bass10 left17:40
EugenMayer or no, it does ofc17:40
doener EugenMayer: the "snapshot" term is usually just used to say "git stores snapshots, not deltas"17:40
EugenMayer ok but with --amend I could modifiy this snapshot I guess17:41
→ Bass10 joined17:41
EugenMayer I undestand.17:41
doener EugenMayer: no. Objects are immutable. "git commit --amend" creates new objects17:41
→ parasti joined17:41
EugenMayer doener: just trying to understand more deeply.17:41
Ah, ok17:41
so amend actually just change the commit. Every commit has number(all?) of snapshots of files (changed?)17:42
doener when you run "git add foo" (and foo is a file), git reads the contents of "foo" and puts them into an object (the object's name depends on its contents --> content addressable storage)17:42
jast not quite. amend creates a new commit that replaces the one you had before.17:43
doener and it creates/updates an entry in the index, that says "foo is a file, its contents are in this object" (roughly)17:43
EugenMayer so the other is removed ( plus the objects where snapshotted for it?)17:43
jast not immediately. git keeps old objects around for a while for safekeeping.17:43
doener EugenMayer: did you read Git From The Bottom Up?17:43
EugenMayer ah ok. But in general, they are not in direct use anymore17:43
jast yep17:44
doener EugenMayer: that does a good job of explaining the inner workings17:44
EugenMayer doener: well the book till chapter 517:44
doener: well, it's a lot of informations. Reading, understanding and actually then "recognising" is a long process for me17:44
doener EugenMayer: if you just do "git commit --amend" to change the commit message. You get _one_ new object, a new commit. The trees and blobs will still be the same17:45
EugenMayer using CVS / SVN for so a long time, I really have problems with change the way of thinking :)17:45
doener: well, this is if I did not stage any new files / changes before amend <->old commit, right. otherwise I can change the commit message + the commit is changed ( objects stored )17:46
doener the commit object is _never_ changed. It's immutable17:46
cga ok, i did git commit on server after git status showed me the changes. but still can't see them live. what else do i fail?17:46
selckin cga: did you read faq non-bare ?17:47
Gitbot cga: Pushing to non-bare repositories is discouraged. See http://git.or.cz/gitwiki/GitFaq#non-bare17:47
EugenMayer doener: well yeah, more precise, the objects IN the commit are exchanged17:47
doener EugenMayer: if you added modified files to the index prior to "commit --amend", then the "git add" part created new blobs. And the "commit" will call write-tree, which will create new tree objects. And the commit object will reference one of those new trees17:47
→ spaceonline joined17:47
doener EugenMayer: a commit object only references zero or more parents commits + 1 tree object17:48
EugenMayer Ah17:48
→ jlilly joined17:48
cga selckin, yes i read that. how do i set up something that get istant changes when i push from local to remote? i'm the only one working on it. is it git bare?17:48
EugenMayer cga: http://progit.org/book/ch4-4.html17:49
cga thanks17:49
← Wraithan left17:49
doener EugenMayer: Git From The Bottom Up explains that quite nicely, using the plumbing commands. http://ftp.newartisans.com/pub/git.from.bottom.up.pdf17:49
EugenMayer cga: http://progit.org/book/ch2-5.html17:49
cga thanks17:50
EugenMayer doener: thanks! I tread that one up on time. Stoping wasting your time then17:50
← hakunin left17:50
→ hakunin joined17:50
→ Wraithan joined17:50
doener EugenMayer: it's quite short, regarding the current topic, the first 20 pages are enough17:51
EugenMayer: actually, even just the first 15. The other 5 are already about rebasing17:52
← Bass10 left17:54
→ Bass10 joined17:54
→ misiu_mp joined17:54
→ __Joker joined17:55
← kpreid_ left17:56
← Paraselene_ left18:00
Paraselene__ → Paraselene_18:00
→ Killer_X joined18:01
→ cloudhead joined18:01
← Bass10 left18:01
→ Bass10 joined18:01
→ tedoc2000_ joined18:02
Killer_X hello, wonder if someone can help me. im rather new to git18:02
→ maattd_ joined18:03
← parasti left18:03
Killer_X on os x, when i do "git rm somefile" and the git commit, git complains that "somefile" is not there...18:03
then i do "touch somefile" and retry commit git is happy18:04
wereHamster Killer_X: what does 'git status' say just after you run git rm?18:04
doener Killer_X: uhm, a plain "git commit" complains?18:04
wereHamster my guess it's because your filesystem is case insensitive18:04
Killer_X git commit -m "jkhkdsfhsdf"18:04
# (use "git reset HEAD <file>..." to unstage)18:05
#18:05
# renamed: includes/tasker_object.class.php -> includes/classes/TObject.class.php18:05
#18:05
← Axius left18:05
doener wereHamster: hm, would the case insensitivity matter for "git commit"? I'd expect that to be a problem for "git rm", but not for commit18:06
Killer_X: do you use any hooks?18:06
Killer_X doener: thanks!18:07
that was it :)18:07
doener Killer_X: a broken hook?18:07
Killer_X i had a php syntax check hooked in18:07
doener ah18:07
Killer_X so it wanted to check the missing files18:07
doener Killer_X: http://git.pastebin.com/m1ee7d36018:08
← asksol left18:09
Killer_X hey, thanks18:09
→ Axius joined18:09
→ jkp joined18:09
doener Killer_X: the --diff-filter limits it to Added, Copied, Modified and Renamed files18:10
→ adamm joined18:10
doener Killer_X: and the loop also takes care to only check the version you're going to commit, not whatever happens to be in the working tree18:10
← spaceonline left18:12
→ vital joined18:12
← jkp left18:13
Kissaki what's the best way for a simple conversion of a bzr repo to a git repo? there's only one branch, not tags...18:13
→ tedo__ joined18:13
← maattd left18:18
maattd_ → maattd18:18
← Pupeno left18:19
← tedoc2000_ left18:21
← Axius left18:25
← tedo__ left18:25
→ Pupeno joined18:26
→ Axius joined18:28
← hyperair left18:29
← voker57_ left18:30
→ voker57_ joined18:30
→ drizzd joined18:32
← __Joker left18:35
← icwiener left18:36
→ smuf joined18:37
Fissure Kissaki: bzr fast-export/git fast-import18:37
← Axius left18:38
Kissaki Fissure: is that included with the binary releases?18:39
Fissure for git, it's built-in18:39
← cilly left18:39
Fissure bzr, dunno... separate package on ubuntu at least18:39
i think it's a separately-maintained plugin18:40
→ Pupeno_ joined18:40
→ mike8901 joined18:41
← mike8901 left18:41
→ fridim__ joined18:44
→ Adlai joined18:47
← drizzd_ left18:47
Kissaki Fissure: hm, ok thx18:48
→ icwiener joined18:48
→ kpreid_ joined18:48
← BlindOracle left18:52
← pjackson left18:54
← voker57_ left18:54
← Analias left18:54
← Pupeno left18:56
Pupeno_ → Pupeno18:56
→ cilly joined18:57
cga can you tell me if this post-update hook goes on remote or local please? http://git.or.cz/gitwiki/GitFaq#Whywon.27tIseechangesintheremoterepoafter.22gitpush.22.3F18:57
→ Axius joined18:58
doener cga: different question: How would a hook in repo A change the working tree in repo B?18:58
→ talal joined18:58
doener cga: considering that repo B might be on a different box to which you have no access except for "git fetch/push"18:59
cga doener, heh i guess that is a good question, but i'm totally new to git and getting nut since a few hours to get a repo working XD18:59
no offense, i just don't understand it19:00
← virginia left19:00
doener cga: well, repo A == local, repo B == remote. You obviously want to change the working tree of repo B19:00
cga indeed19:01
doener cga: so which repo should get the hook? The one you don't want to change, or the other one to which you might have no direct access? The hook obviously has access to the repo it lives in...19:01
cga ok so i need the hook on B19:01
doener right ;-)19:02
cga thanks =)19:02
doener besides the above 'example' (that one should be able to think through without knowing git), the post-update hook is only executed when a repo is pushed into. So that hook needs to go into the receiving repo19:03
ankit9 → ankit9|zzz19:04
→ Axius_ joined19:06
← Axius left19:06
← dirk2 left19:07
cga doener, everything works now. i only had the problem in pushing from A to B in non bare repo19:08
thanks a lot19:08
→ dreiss joined19:09
cga you might wonder why i want such a behaviour. well it's a personal repo of nanoc3 generated blog. i want the commits to go live as soon as i push.19:09
doener cga: ikiwiki supports "update on push" behaviour, but in a different way19:10
cga: the repo you push to is bare, and a hook in it goes to a non-bare clone and updates its working tree19:10
cga: IMHO, that's somewhat cleaner19:10
cga doener, heh that's nice but i don't use ikiwiki =) i need a blog not a wiki. (if i get it right that's a wiki)19:11
→ j416 joined19:12
← jlilly left19:12
→ bauruine joined19:12
← john268 left19:15
← cilly left19:16
← razvand left19:17
← misiu_mp left19:17
→ hakunin_ joined19:20
← hakunin left19:21
→ juan_arandaalvar joined19:21
← juan_arandaalvar left19:21
← paltman left19:24
→ ab3 joined19:25
→ maattd_ joined19:25
← talal left19:26
→ engla joined19:29
← reprore left19:30
bremner ikiwiki can also do blogs fwiw19:30
→ reprore joined19:30
← bauruine left19:30
← anno^da left19:31
→ john268 joined19:33
← maattd left19:34
maattd_ → maattd19:34
← will_h left19:34
→ radarek joined19:36
→ bauruine joined19:38
← Axius_ left19:38
← reprore left19:41
← Bass10 left19:44
← Adlai left19:44
→ Bass10 joined19:44
→ Adlai joined19:44
← Killer_X left19:45
→ dutchie joined19:51
← Bass10 left19:51
← Sigma left19:51
→ Bass10 joined19:51
→ jack_ joined19:52
jack_ hi19:52
i am a git newbie, sorry, can anybody tell me :who to git diff one file with last two commit ? thanks19:52
cortana git diff HEAD^^ file19:52
jack_ ok19:53
thanks cortana, thanks you very much :)19:53
cortana HEAD^^ identifies the 2nd last commit19:53
dutchie I accidentally made a commit on a detached HEAD, how can I apply it to master? git commit -C <id> doesn't work19:53
cortana you can stick ^ on the end of a branch name and it'll work from there, too19:53
dutchie: cherry pick it?19:53
jack_ very appreciate your help, cortana19:53
cortana you're welcome19:53
dutchie great, thanks cortana19:54
→ razvand joined19:54
→ Tommy[D]_ joined19:54
j416 dutchie: note the sha1, check out master, git merge <sha1>19:54
or, if you have done no changes on master and just want it to resemble what you have in your detached HEAD, you can simply do: git checkout master; git reset <sha1>19:55
→ Twisted_Mentat joined19:55
j416 cherry-pick works also, of course, but only for single commits19:55
→ reprore_ joined19:59
dutchie j416: cherry-pick did the trick, as it was only the one, but thanks20:00
j416 :)20:01
cherry-pick is a little overkill though20:01
but sure.20:01
← jack_ left20:02
← smuf left20:02
→ piem joined20:03
piem howdy. is there anything else than tailor + svn to switch from bzr to git?20:04
→ bernie_ joined20:04
← india7145 left20:05
← cga left20:06
→ india7145 joined20:08
← dreiss left20:16
Fissure piem: one-time transfer of history from bzr to git?20:16
use bzr fast-export plugin and git fast-import20:17
← Paraselene_ left20:19
dazman → Guest6215420:21
← omerj_ left20:22
piem Fissure: yep, got that reading your words in the backlog20:23
← engla left20:23
piem Fissure: but I get a "fatal: Corrupt mode: M 040000 - doc"20:23
Fissure strange that question got asked twice in such a short time20:24
look for an option in bzr fast-export to not output directory entries20:24
→ dreiss joined20:25
piem Fissure: indeed :-) can't find such an option, and it doesn't seem git fast-import has an ignore one20:26
but i can use grep -v ...20:26
→ __Joker joined20:26
→ _science_ joined20:27
Fissure make sure that doesn't corrupt and file data...20:27
_science_ How can one setup a folder /project/frameworks where frameworks should hold all the git remote that one has added ?20:28
piem Fissure: it still fails, complaining about "Path ... not in branch"20:29
Fissure piem: perhaps you removed an entry you should not have?20:29
go bother the bzr-fast-export person for writing a broken stream :P20:30
← mbroeker left20:30
← bx2 left20:30
→ bx2 joined20:31
→ blakemizerany joined20:31
→ developernotes joined20:34
← developernotes left20:34
← dutchie left20:35
← bx2 left20:35
← Eraldo left20:36
→ bx2 joined20:37
→ omerj joined20:37
→ Determinist joined20:39
→ dutchie_ joined20:39
← dutchie_ left20:39
→ TML joined20:41
← radarek left20:42
dsop ha finally finished a patch to clone just wone branch20:43
← NetHawk left20:44
→ Hawk416 joined20:44
← TML left20:46
→ TML joined20:46
→ aresnick joined20:48
← blakemizerany left20:50
→ tedo__ joined20:52
← tedo__ left20:52
→ radarek joined20:53
wereHamster _science_: ?20:54
← razvand left20:58
→ jmspeex joined20:59
← bernie_ left21:00
→ bernie_ joined21:00
→ khmarbaise joined21:01
→ Paraselene_ joined21:02
piem Fissure: https://bugs.launchpad.net/bzr-fastimport/+bug/427188 talks about this issue21:02
Fissure: and the patch at http://launchpadlibrarian.net/31687430/fix-plain-dir-rename works for me (tm)21:04
bernie_ → bernie21:04
Fissure that's good21:04
→ blakemizerany joined21:04
← froschi left21:04
→ threeve joined21:05
piem Fissure: now, you mentionned "one-time transfer", what's wrong with this approach if i want to keep the old bzr branch up to date for a while?21:05
Fissure i suppose it will work for incremental updates if you use a marks file, but it would probably be a headache to make things bidirectional as anything more than a stopgap21:06
← plediii left21:07
→ plediii joined21:07
piem i see. git is soooo much faster i don't really want to bother with bzr anymore! :-] but it'd still be nice to my users to keep the bzr branch up-to-date for some time21:07
AAA_awright Can you git-fast-export the revisions made in Git out to bzr?21:10
piem: http://bazaar-vcs.org/BzrFastImport21:11
→ BlindOracle joined21:13
Fissure AAA_awright: yes, you can21:14
i've never really used fast-import as anything other than an archival format and for a few test repos because i was curious21:15
AAA_awright That is all I could think of, I have searched for a bi-directional bzr<->git program but can't find one21:15
← aresnick left21:16
Fissure AAA_awright: http://github.com/pieter/git-bzr https://launchpad.net/bzr-git neither of these works?21:16
← corni left21:19
AAA_awright Fissure: Hmm.... I think I was looking for a git program, I think I rejected that for some reason, I'll check again21:20
← LukasLT10 left21:20
AAA_awright Probably because it's Python21:20
Hate, hate, hate21:20
jast ahh, a kindred spirit21:21
← intripoon left21:22
jast just wait until I release my own programming language (work in progress)... nobody but me will know it, but at least I can stop hating all competition because I'll stop needing it :)21:22
→ plediii_ joined21:23
→ ereslibre joined21:24
→ krynnotaur joined21:26
← khmarbaise left21:28
peper hello21:29
can i git clone to a remote dir via ssh?21:29
xenoterracide is there any way I could use git to detect duplicates between 2 directories (neither of which are currently under git)21:30
← omerj left21:30
straszheim sure21:31
→ omerj joined21:31
xenoterracide and what would be the best way to go about doing that. note the filenames between the 2 directories are not consistant although some of the files may be identical21:32
straszheim well you can import both directories and diff them... then there'd be some scripting i spose, as diff will give you the opposite of what you want. maybe there's the Perfect Flag in there somewhere.21:33
http://stackoverflow.com/questions/224687/git-find-duplicate-blobs-files-in-this-tree21:34
← Rolenun left21:35
Fissure xenoterracide: i'd just use diff on the sorted output of sha1sum for those two directories21:35
xenoterracide Fissure: a diff of a sha1sum alone won't be useful because everything will be different as the filenames are different21:37
Fissure xenoterracide: actually, you might want to do something clever with uniq21:37
can there be duplicates within the directories?21:38
xenoterracide no21:38
each directory is unique21:38
however there may be duplicates between the directories21:39
Fissure sha1sum dir1/* dir2/* | cut -f1 -d' ' |sort | uniq -d | while read sha1; do sha1sum dir1/* dir2/* | grep $sha; done21:39
something like that :)21:40
← plediii left21:40
plediii_ → plediii21:40
Fissure er, s/while read sha1/while read sha/21:40
gotta use the same variable name :)21:40
→ Rolenun joined21:41
remi is there a page where I can go and see that "commits names" I can use in "git diff" (like HEAD, etc...) ?21:41
Fissure obviously you could pregenerate the sha1 lists21:41
remi * see all "commit names"21:41
Fissure remi: you mean the different naming conventions?21:41
man git-rev-parse21:41
Gitbot Fissure: the git-rev-parse manpage can be found at http://git.or.cz/man/git-rev-parse21:41
remi Fissure: excellent, thank you!21:42
xenoterracide Fissure: and what does that find me? the duplicates?21:43
Fissure xenoterracide: it should :)21:43
totally untested, of course21:43
← __Joker left21:43
→ colossus joined21:44
← d0k left21:44
← colossus left21:44
xenoterracide Fissure: well a quick check of one of the results say it does. thanks. is there any way to get sha1sum to ignore subdirectories?21:45
Fissure xenoterracide: it doesn't recurse into subdirectories here21:47
and wildcards are expanded by the shell anyway21:47
xenoterracide no it doesn't here it just complains about them. generates a lot fo noise. I suppose I could just filter the noise. I think it's stderr so some redirection should work21:48
Fissure yes, it is21:48
→ Sigma joined21:48
Fissure or use `find dir1 -type f -depth 0` (or depth 1? i don't remember) instead of shell wildcards21:48
← maattd left21:48
xenoterracide Fissure: k, thanks you've been a real help21:50
← bx2 left21:50
← cloudhead left21:50
→ modsognir joined21:52
← plediii left21:54
← dreiss left21:54
→ aresnick joined21:54
→ cloudhead joined21:56
← bgerber left21:56
doener Fissure: -maxdepth 1, I guess? -depth doesn't take an argument, but makes find process the directory's contents before the directory itself21:58
Fissure doener: certainly; was just guessing as to the syntax because i knew find had to have that option even if i didn't know what it was called21:59
→ icwiener_ joined21:59
→ bgerber joined22:01
→ Rayne joined22:03
← j416 left22:03
→ Pupeno_ joined22:05
Rayne if i use git on my local system where does it store its information? in project/.git as far as i know. but what do i have to do, if i fork my project to test a new feature but want to use another branch which sends its data to the origin project/.git directory?22:05
like master and testing branch22:06
cbreak just make a branch22:06
Rayne but how do i get my files for working with them?22:06
cbreak checkout -b testbranch master22:06
then you're on the testbranch22:06
and can do what you want22:06
if you want to go back, just git checkout master22:06
← reprore_ left22:07
cbreak the files in the working directory will be replaced on switching22:07
if you want an additional working directory, just git clone the original22:07
Rayne ah22:07
cbreak but remember: do NOT push to repositories with working directories22:07
unless you know what you are doing22:07
Rayne cool, it works :)22:08
but does that not take much time with big projects to switch the branch?22:09
← fcuk112 left22:09
← tjaway left22:10
→ tjaway joined22:10
Rayne i read that if i clone a repo i will get the whole repo and its whole history and old files. is it possible to fetch only the current ones for branch xyz?22:10
(i read that some people use git for managing their documents and this sounds neat)22:10
← ph^ left22:10
Rayne (in this case it only want the latest versions on my system, but the complete repo for instance on my external hdd)22:11
← icwiener left22:11
← g3d left22:13
→ Gitzilla joined22:14
→ j416 joined22:15
d3xter Rayne: stores the complete history of every "git clone", or for example rebase wouldnt be possible anymore22:16
or fast-forward merge22:16
→ soupeee joined22:17
Rayne d3xter, but i am able to delete old stuff which i am really sure that i will never need it again?22:18
doener Fissure, xenoterracide: eval $(echo join '<(find '{x,y}' -maxdepth 1 -type f -print0 | xargs -0r sha1sum | sort)')22:18
→ bugxek joined22:18
doener Fissure, xenoterracide: x and y are the directories to look into22:18
madewokherd if you delete a branch it'll go away "eventually"22:18
doener Fissure, xenoterracide: for large directories, that's quite a bit faster than the loop that executes sha1sum over and over again22:19
madewokherd and if you delete a repo, it'll go away now22:19
soupeee nobody@upix /var/git $ git init22:19
/var/git/.git: Permission denied22:19
offby1 soupeee: are you trying to ask a question?22:19
soupeee yes22:20
offby1 soupeee: what might that question be?22:20
offby1 whistles innocently22:20
cbreak Rayne: deleting stuff out of git is hard22:20
it requires rewriting history22:20
d3xter Rayne: first of all, git doesnt save the complete file, which has been modified, you can only delete the blob/tree objects22:20
cbreak (not hard as in that it's hard to do, because it's not, but as in that it's a pain to manage)22:20
d3xter Rayne: but afterwards, git will complain about missing objects22:20
offby1 guesses the question is "What does that error message mean, and what do I do about it?"22:20
d3xter Rayne: you could copy all of the files from your working directory and then create a new repo22:20
soupeee why do I get permission denied when I try to use git-init?22:21
doener d3xter: "git doesnt save the complete file, which has been modified" -- hm? How do you mean that?22:21
d3xter soupeee: you should to that in /var/git as root22:21
Fissure doener: never seen the join command before, and redirection from a subshell is not something i would immediately think of22:21
← ab3 left22:21
offby1 soupeee: because, as the error message says, you don't have permission to write to the directory named /var/git.22:21
Fissure doener: can we come up with a shorter one using perl? :)22:21
soupeee I can't become root either, perimission denied22:21
offby1 soupeee: this is probably because you're logged in as "nobody", who traditionally has no permission to do pretty much anything.22:21
Rayne d3xter, that is a simple but good idea, thanks for that hint22:21
doener Fissure: I learned about join(1) from some blog post (I'm pretty sure it was some database blog ;-))22:21
← Pupeno left22:22
Pupeno_ → Pupeno22:22
offby1 soupeee: so put the new repository in some directory to which you _do_ have write access.22:22
d3xter doener: i mean, git saves every change which has been commited in a blob object22:22
→ ijuma_ joined22:22
doener Fissure: and the <() redirection was mentioned by charon (I think) in here lately, so I had to put them into action :-)22:22
d3xter doener: it doesnt save an old version of a file in ".git"22:22
Fissure doener: i also said it would make sense to store the sha1sum output in a temporary file22:22
doener d3xter: and the blob object holds the whole file contents22:22
soupeee been following gentoo-wiki/wiki/git that's what it says to do22:22
doener Fissure: sorry, must have missed that22:22
offby1 soupeee: either you're screwing something up, or it's inaccurate.22:23
d3xter doener: no, it just contains the changes22:23
madewokherd Rayne: it's best to have one repo for each "project"; if you have unrelated things in the same repo it'll just get annoying (inability to delete them is only one example)22:23
d3xter doener: i mean, it contains the patch, which has been applied to the branch :)22:23
soupeee also I started the session as root then I was dropped into the nobody shell22:23
Fissure d3xter: git does not store patches, it stores complete trees22:23
d3xter and this blob item is encrypted22:23
← hgimenez left22:23
offby1 soupeee: I get the feeling you don't understand the fundamentals of Unix.22:23
doener d3xter: no. Git objects always contain a the full contents. Deltas are only created when things get packed22:23
d3xter: and the objects aren't encrypted. Just gzipped22:24
d3xter Fissure: doener: thanks for the clarification :)22:24
→ giallu joined22:24
doener whenever git needs a patch, it creates it on the fly22:24
Rayne madewokherd, thanks for that hint22:24
soupeee which fundamental in particular?22:25
← giallu left22:25
offby1 soupeee: I also get the feeling that your gentoo documentation is assuming that there's already a directory named "/var/git", and that that directory is owned by "nobody". You can confirm (or deny!) this by typing "ls -ld /var/git", and looking for a user name in the output.22:25
→ giallu joined22:25
← EugenMayer left22:25
offby1 (actually, you'll probably see _two_ user names; you can ignore the second one.)22:25
← Adlai left22:26
Fissure well, second isn't a *user* name...22:26
→ hipe joined22:26
→ Adlai joined22:26
offby1 Fissure: yeah, yeah. I take some liberties so as not to confuse the n00b :)22:26
too many details can be counter-productive.22:26
← Sigma left22:26
→ erlehmann joined22:27
Fissure fair enough22:27
offby1 wonders where soupeee went off to22:28
soupeee chown root:nobody /var/git, same problem22:29
→ dreiss joined22:29
doener soupeee: and the group has write permission?22:31
soupeee just a sec22:31
offby1 soupeee: I'd really appreciate it if you'd run "ls -ld /var/git", and paste the output into your IRC client. It'd make it easier for me to help you.22:31
doener: might be a good idea for just one of us to help; too many cooks is confusing22:31
← doener left22:32
← adamm left22:33
soupeee upix soupeee # ls -ld /var/git22:33
drwxr-xr-x 3 root nobody 1024 Nov 29 13:27 /var/git22:33
offby1 ok, that explains everything.22:34
← blakemizerany left22:34
offby1 (or, almost everything; it doesn't explain why the gentoo docs are broken)22:34
soupeee tell me more22:35
offby1 anyway, as root, I suggest you type "chmod 775 /var/git"22:35
I'm not absolutely certain that that's the right thing to do; it's possible that the right thing is instead "chown nobody /var/git"22:35
both will at least get you past your current roadblock, but one might have some sort of subtle problem later. Unfortunately I can't predict which is which, because (frankly) I'm too lazy to read that gentoo doc.22:36
→ loincloth joined22:36
soupeee nobody@upix /var/git $ git init22:36
Initialized empty Git repository in /var/git/.git/22:36
OK, now were gettin somewhere22:36
offby1 I have to suggestions for you in the long term: 1) become familiar with Unix permissions; they're pretty easy; and 2) quit using probably-broken documentation from gentoo, and instead use proper documentation from the Git project itself.22:36
s/to/two/22:36
→ jelmer joined22:37
← radarek left22:39
→ dhoss joined22:40
→ petteri joined22:41
← ixti left22:42
soupeee ok thanks does this mean I dont need to be nobody?22:45
← aresnick left22:45
→ secoif joined22:46
← india7145 left22:46
→ kukks joined22:47
→ india7145 joined22:47
offby1 I'm not sure.22:48
my guess is you need to be nobody until you start the daemon, and then you can resume being "yourself".22:49
← bambule left22:49
← ludde left22:49
offby1 soupeee: also keep in mind that it _might_ be easier all around for you to use one of the free git hosting sites, like repo.or.cz or github, rather than running your own server.22:50
→ adamm joined22:50
← bugxek left22:50
soupeee so I didn't need to emerge git even?22:52
→ davido joined22:53
soupeee all I wanted to do originally was to update fsck which wont check the fs on my netboot unless the ac cord is plugged in22:53
offby1 I would guess you don't, since you seem to already have a git binary. But "emerge" is a very gentoo-specific topic, and I don't understand it at all well.22:53
offby1 stares blankly22:53
offby1 what on EARTH does that have to do with git?!22:53
offby1 isn't sure he actually wants to know22:54
Rayne kthxbye22:55
← Rayne left22:55
dominikh mhm. when using git-svn, is it save to clone the git repo, commit to that, merge the clone back to the "main" git-svn repo and then svn dcommit? (if not using any branches)22:56
soupeee according to the developer of fsck I need to use his git repository to fix this.22:56
offby1 soupeee: I think you're waaay over your head.22:57
I know this isn't what you want to hear, but: I recommend that you use a stable boring distro, like Debian or Ubuntu, rather than gentoo.22:58
AAA_awright Gentoo users are in over their head by definition I thought22:58
I would know, I love Gentoo :)22:58
Wolfman2000 AAA_awright: ...I wish you luck.22:58
offby1 dominikh: yes, although it's pretty easy to find yourself having committed the "rebasing a public branch" sin.22:58
dominikh offby1: mind elaborating a bit?22:59
offby1 dominikh: just be careful, and get in the habit of doing "git --dry-run dcommit" a lot, to see which branch you're going to dcommit to before you actually do it22:59
offby1 gestures upwards22:59
dominikh oki22:59
offby1 dominikh: I still haven't come up with a completely satisfactory way to work on the same project on multiple boxes. Last thing I tried was passing patches around with "git format-patch" and "git am -3"; that seemed to work OK23:00
dominikh well. the svn repo only got one branch, and I will stick to one git branch (suits the workflow). I hope that makes it less dangerous :)23:00
AAA_awright soupeee: Trusting that compiling fsck by hand is what you are supposed to do, "emerge git" and in your home directory or something "git clone $url"23:00
offby1 dominikh: as long as you understand "git for computer scientists" in your very bones, you'll be OK23:00
AAA_awright You shouldn't need to git init an existing project23:00
offby1 AAA_awright: OK, he's yours now :)23:00
AAA_awright noes23:01
:)23:01
Wolfman2000 dominikh: You will learn to at least appreciate branches.23:01
dominikh Wolfman2000: I do appreciate branches, I use them in my projects. but I simply don't need them for this one ;)23:01
← remi left23:01
← _science_ left23:01
Wolfman2000 dominikh: I assumed so myself. I proved myself wrong.23:01
dominikh Wolfman2000: well, believe me ;) -- I have been working on this (git-)svn repo for months now23:02
Wolfman2000 *shrugs* whatever you say23:02
dominikh I had to resort to git-stash only once23:02
selckin more multitasking!23:02
dominikh :)23:03
← sis left23:03
← fridim__ left23:04
← ereslibre left23:12
← Stravy left23:12
← lhz left23:13
→ MrGando joined23:14
MrGando Hello guys, I'm trying to set up git to work with a small team of web developers... the thing is that we are used to work on a development box, and then when we hit a milestone we move that to a production server, what would be the best way to achieve this setup ?23:15
→ eyl joined23:15
← jaysern left23:17
→ admiral0_ joined23:19
← soupeee left23:19
Rolenun MrGando: http://progit.org/book/ch5-1.html23:21
MrGando Thank you very much23:21
→ drry joined23:25
offby1 dominikh: "resort to"? git stash is _fun_!23:25
you should look for excuses to use it!23:26
offby1 glances around nervously.23:26
dominikh yeaah.. :p23:27
← dbatbold left23:31
→ siks joined23:31
← admiral0 left23:33
→ hyperair joined23:36
← inffcs00 left23:36
→ CarlFK joined23:39
← admiral0_ left23:40
← ntoll left23:42
← icwiener_ left23:45
Rolenun MrGando: The main page: http://progit.org/book/ may also be a good reference if you are just learning git23:46
CarlFK if I have a symlink in my dirtree, what will happen if I check in both the file and the link?23:48
selckin it will do the right thing23:49
CarlFK yay23:50
this had me worried: git status = new file: static/css/login_form.css23:50
selckin a symlink is a file23:50
CarlFK fair enough23:51
← pantsman left23:55
← vdidel_ left23:55
→ kstarr joined23:58
→ skammer joined23:58
Rolenun tackling submodules today.....yikes :D23:59
→ blakemizerany joined23:59
→ methanol joined23:59

Logs Search ←Prev date Next date→ Channels Documentation