IRCloggy #git 2008-11-05

Logs Search ←Prev date Next date→ Channels Documentation

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

2008-11-05

→ josh^ joined00:00
← bdiego left00:01
spearce not implemented. git-shell is like a 10 line c program, with all of those lines going to the paranoid check for the 3 commands it lets you invoke00:01
← archis left00:02
derRichard i see00:04
i think i'll use openssh's chroot00:04
jast note that git needs to be available inside the chroot00:05
→ kennethlove_ joined00:05
derRichard i know :)00:05
← kennethlove_ left00:06
spearce derRichard: most people use gitosis for access control over ssh00:07
derRichard is gitosis stable?00:07
hmm gitosis looks good :)00:08
← hoban left00:08
derRichard thanks for the hint!00:08
← LiamH left00:09
→ urbanmonk joined00:13
← naeu left00:14
← mw|out left00:20
→ paltman_ joined00:21
→ flaguy joined00:21
← kennethlove left00:21
→ Fullmoon joined00:26
← ciaran_ left00:28
→ Ian00 joined00:32
→ paltman joined00:34
← grahal left00:37
→ stouset joined00:38
← Niamor left00:39
→ Niamor joined00:39
← hidek left00:47
→ Fullmoon_ joined00:49
← digitalhobbit left00:50
→ EmilMedve joined00:51
← paltman_ left00:51
← threeve left00:51
← davidfetter left00:54
← pygi left00:55
→ ciaran joined00:58
→ adamb90 joined01:00
← Fullmoon left01:05
← Pixar- left01:05
← patmaddox left01:06
← robinr left01:08
← bobesponja left01:09
← ahupp|fb left01:11
→ hidek joined01:13
→ Flathead joined01:14
← dreiss left01:19
← thannoy left01:22
→ gretchen joined01:22
← paltman left01:25
→ ahupp|fb joined01:26
← ceej left01:27
← loiseau left01:29
→ loiseau joined01:29
→ pjhyett joined01:30
→ ben_h joined01:30
→ craiger_ joined01:34
← dcolish left01:35
craiger_ → ThaDon_scorned01:35
← \ask- left01:35
← ThaDon_scorned left01:36
← ashleyw_ left01:44
→ ashleyw joined01:45
← ashleyw left01:46
→ ahupp|fb_ joined01:50
← joshdavey left01:54
← loiseau left01:58
← mandric left02:01
← ahupp|fb left02:02
→ mandric joined02:02
→ bobmcw joined02:03
← mandric left02:03
→ sdboyer-laptop joined02:06
← piclez_ left02:06
→ ekontsevoy joined02:09
→ electro joined02:09
→ paltman joined02:09
ekontsevoy The man page for git-diff mentions git-diff-tree, git-diff-files commands that I don't have on OSX. What are equivalets of those commands using git -params notation? Or should I install something else to get them?02:09
electro does this room provide support for github and git bash?02:09
← davidcramer left02:10
DrNick ekontsevoy: you have them, they're required for git to function. they may not be in your path, though02:11
→ dreiss joined02:13
dsturnbull ekontsevoy: probably in /opt/local/bin02:14
ekontsevoy dsturnbull: nope. I installed git via Macports and all I have is pretty much just "git" plus git-upload-* and git-receive-*02:16
DrNick: I'll do find / -name "git*" in a sec....02:16
→ joydivider joined02:16
← kanru left02:17
← ahupp|fb_ left02:18
← joydivider_ left02:18
→ johnw joined02:19
→ eno__ joined02:21
← joydivider left02:22
← parasti left02:26
→ piclez_ joined02:28
→ mandric joined02:29
→ threeve joined02:29
← freqmod_qu left02:29
→ freqmod_qu joined02:29
→ imyousuf joined02:31
← eno left02:32
→ davidfetter joined02:32
← ekontsevoy left02:33
→ random8r joined02:35
→ GNUix joined02:38
→ paulp joined02:40
← Zimm left02:41
← bobmcw left02:43
stouset → stouset|away02:46
→ Ademan joined02:49
← Ian00 left02:49
← iabervon left02:51
← spearce left02:52
→ perezd88 joined02:53
← sdboyer-laptop left02:54
← rtomayko left02:56
← eddyp left02:59
← sgrimm left03:00
→ popol joined03:01
→ lcantey joined03:02
← random8r left03:03
electro Hi.03:07
→ ahupp|fb joined03:10
← flaguy left03:13
← david_koontz left03:14
← carllerche left03:15
← mandric left03:17
→ mandric joined03:18
→ scrooloose joined03:19
→ kanru joined03:21
← Chicago left03:21
→ popol_ joined03:23
← popol left03:24
→ haesbaert joined03:26
← w00t_ left03:39
→ bo5 joined03:41
← bo5 left03:42
Hobbes → Guest7176103:46
→ Hobbes`` joined03:46
← jas4711 left03:48
→ carllerche joined03:48
→ jas4711 joined03:49
← perezd88 left03:51
← rmh3093 left03:52
→ rmh3093 joined03:52
haesbaert hello there how do I delete a remote branch ?03:57
I'm trying git-branch -d origin/multi-msnp03:58
multi-msnp is the name of the remote branch.03:58
← Guest71761 left03:59
← elmex left04:00
→ elmex joined04:02
→ jackdempsey_ joined04:04
← ahupp|fb left04:06
→ nipuL joined04:09
→ deskin joined04:11
← nipuL left04:15
→ FunkeeMonk joined04:19
← davidfetter left04:20
→ cedricv joined04:22
← offby1 left04:31
→ iabervon joined04:32
→ mtodd_ joined04:33
→ offby1 joined04:40
← mtodd left04:40
→ Ian00 joined04:41
→ muthu joined04:42
ToxicFrog haesbaert: git push <remote> :<branch>04:49
Eg, git push origin :multi-msnp04:49
bdrewery hey while you're on it, how can I push to a new remote branch?04:50
→ jony joined04:51
bdrewery guess just git push origin newbranch04:51
electro im trying to push for the first time and i am getting Bad File Number, I was able to git the repo and download files for my working copy but am having trouble pushing my changees. can anyone help a newb get started, im a little bit familar with repo's in genreal but not too much.04:52
← paltman left04:52
bdrewery heh04:53
← jackdempsey_ left04:56
← kukks left04:57
→ mbuf joined04:58
mbuf how can I create a single patch for all changes made from first revision?04:58
stouset|away → stouset04:58
offby1 mbuf: "git diff $(git rev-list | tail -1)", I think05:01
mbuf offby1: how do i know what is the current revision number it is in?05:03
offby1 dunno what you mean05:03
what is "it"?05:03
mbuf offby1: I made a change, created a patch; but change was incorrect; to fix it, i created a branch, made the change, and merged back;05:04
offby1 going to bed now; sorry05:04
mbuf offby1: and creating a new patch using git format-patch -1 -p, generates only the later patch, not including the first one05:05
offby1: ok, later05:05
electro bdrewery, im trying out gethub for the first time.05:05
so i know the actual repo is in place, and i pulled down a file modified it, but im havin' trouble comitting it.05:06
← jony left05:06
→ SjB joined05:07
bdrewery I dunno, I'm new to git too ;)05:07
← nadim left05:08
→ nadim joined05:08
→ tjafk2 joined05:10
electro im getting somewhere i got a new error.05:11
← bartman left05:11
→ mtodd joined05:14
← piclez_ left05:17
← tjafk1 left05:18
← mtodd_ left05:20
← dhruvakm left05:23
← beilabs left05:23
← threeve left05:23
← p4tux left05:28
→ p4tux joined05:28
← p4tux left05:29
→ jackdempsey_ joined05:30
← electro left05:33
← johnw left05:36
← loincloth left05:37
→ ahupp|fb joined05:39
→ jackdempsey__ joined05:40
← Ian00 left05:43
→ mtodd_ joined05:44
← Jacolyte left05:44
→ priidu joined05:44
→ loincloth joined05:45
→ scientes joined05:49
← niki left05:50
← jackdempsey_ left05:50
← mtodd left05:51
← Fullmoon_ left05:55
→ SRabbelier joined05:56
← jackdempsey__ left05:59
→ sgrimm joined06:01
→ Jacolyte joined06:01
← loincloth left06:05
→ loincloth joined06:06
← mandric left06:07
→ reithi joined06:08
mbuf is there any good book on git and git workflows?06:10
→ mikem joined06:12
← selckin left06:13
mikem hi, ran into a problem. i have a remote branch, which follows an upstream SVN tree. On this branch I have made some changes. I would now like to rebase, which I have successfully done on my local machine. However, now I try to push this new state of this branch back to the remote machine, but I get a fast-forward error, even with `git push --force`. What's the correct way to achieve this?06:14
I think it's the same issue described here: http://kerneltrap.org/mailarchive/git/2007/11/18/42572906:15
← stouset left06:16
← muthu left06:17
← cedricv left06:17
← siprbaum left06:17
→ stouset joined06:17
→ mczepiel joined06:20
← pjhyett left06:24
← stouset left06:25
dsturnbull i'm trying to figure out how to clone a git repo and allow it to svn dcommit.. the git repo was originally pulled down from svn06:27
→ litage joined06:27
← markelikalderon left06:28
→ ceej joined06:29
dsturnbull the problem is the git clone doesn't seem to have enough info to push back06:30
should i just make the git clone as a git-svn clone?06:30
← GNUix left06:30
→ muthu joined06:33
← releod left06:33
→ fhobia joined06:34
← madewokherd left06:35
← ahupp|fb left06:36
→ robinr joined06:37
→ ahupp|fb joined06:40
← k776 left06:43
mczepiel dsturnbull: I think you'd need to pull from the cloned respository back into the one that was git svn cloned, and then git svn dcommit from there06:45
dsturnbull yeah, i tried that, but the pull said "no common commits exist"06:46
→ takayaW joined06:46
mczepiel dsturnbull: odd, I know that works...or should06:46
→ rtomayko joined06:49
→ brettryan joined06:49
brettryan Hello guys, I've got a situation where I have a git repository, now I want the parent directory contained in that git repo, how would I go about doing this without loosing history?06:51
← fhobia left06:51
brettryan say I have foo/bar, where bar is managed by git, but now I want foo to also be managed by git.06:51
← coshx left06:53
← bronson left06:53
brettryan One solution would be to create a folder in bar - called `bar' the original `bar' to the new folder, then copy the contents of the parent folder in.06:53
← SjB left06:54
→ Mario joined06:55
← ahupp|fb left06:55
→ bartman joined06:58
← robinr left07:05
← priidu left07:10
← sgrimm left07:11
← stephenos left07:12
→ priidu joined07:13
← chrizz left07:15
← ben_h left07:18
← litage left07:20
→ fhobia joined07:20
← rtomayko left07:20
→ Marmouset joined07:22
→ sstangl joined07:25
→ mtodd joined07:29
→ a_l_e joined07:30
← dmq left07:32
→ dmq joined07:32
← mtodd_ left07:36
← rmh3093 left07:36
← ceej left07:37
→ ciskje joined07:39
← ciskje left07:40
→ ciskje joined07:40
← ciskje left07:41
→ ciskje joined07:41
→ Jaap joined07:45
Jaap Hello07:45
→ reel_ joined07:46
Jaap Bit of a newbie question here, but docs for rm, add, status, or reset didn't appear to have what I look for.07:46
I've made some fairly invasive changes in my working directory.07:46
git status lists over 40 deleted files.07:47
Not easily globbable.07:47
How can I tell Git that it needs to rm all those files from the index?07:47
With files added it's always easy, git add . will add everything, which does no harm if a file was already added.07:48
In the case of an rm however, I can't do a global wildcard.07:48
Basically, I want to tell Git, make my index what the working directory is like.\07:48
I'll have another look at git reset07:48
jdq git commit . maybe?07:48
→ bentob0x joined07:49
jdq that will commit any files that are on the index... so deleted files will be removed and changed files will be commited... unversioned files remain untouched07:49
Jaap My deleted files aren't marked as such yet in the index.07:50
Basically, I went behind Git's back, and deleted files by hand.07:50
mikem Jaap: how about `git add .` ?07:50
Jaap That won't update the deleted files.07:50
← Marmouset left07:51
mikem then try `git rm <files>`07:51
← vasandgvd left07:51
jdq they are shown as deleted in in git status but not commited as those if you do git commit .?07:51
Jaap I know that git rm <files> will work, but the glob is entirely non trivial.07:52
Alternatively, I could pipe the output of git status into git rm.07:52
I found a simple way, 30 mouse-clicks in git-gui did the trick.07:53
mbuf guilt patch series is specific to a branch?07:53
Jaap Not really the proper way, but oh well.07:53
jdq git ls-files --deleted07:53
Jaap Thanks guys07:53
mikem Jaap: Yeah, output of `git status` isn't exactly sed-friendly07:53
:)07:53
Jaap Thanks jdq!07:53
mikem jdq: neat :D07:53
jdq but git commit -a is what makes the index like the working copy... (despite from untracked files)07:56
← careo left07:57
← takayaW left07:58
→ jony joined08:01
→ lucsky joined08:02
lucsky 'morning08:02
← felipec left08:03
← carllerche left08:05
→ quoin- joined08:06
→ koke joined08:08
← quoin left08:10
→ nud joined08:11
→ tritz joined08:11
→ nadim_ joined08:18
→ loiseau joined08:18
nadim_ given a 'Merge with strategy recursive failed' when I do a git pull. How can I say take the one from origin and be done?08:19
← Ademan left08:24
→ charon joined08:24
→ ph^ joined08:28
→ drizzd joined08:31
← Mario left08:32
→ ankit9 joined08:35
← sdboyer|ecrire left08:39
← adamb90 left08:40
→ mediogre joined08:42
← CodeOfficer left08:45
← drizzd_ left08:46
→ siprbaum joined08:48
→ naeu joined08:53
→ sciz joined08:54
→ cilly joined08:58
→ muthu_ joined09:01
← muthu_ left09:01
→ lamont` joined09:02
← muthu left09:02
→ Shilin3 joined09:07
← lamont left09:08
lamont` → lamont09:09
← iulian left09:12
← mczepiel left09:13
→ Pupeno-M joined09:13
← dreiss left09:14
→ felipec joined09:15
→ cradek_ joined09:16
← cradek left09:16
→ mithro joined09:16
→ powerfox joined09:21
→ Mario joined09:21
powerfox Is there any git-perforce tool similar to git-svn? Google knows only import perforce repo to git tool, not the thing I need :(09:21
Oh, there is git-p4. Is it safe to use?09:22
← cilly left09:26
thiago_home yes09:27
I used it for well over a year09:27
nud I guess it is, Qt people used it for a long time09:28
thiago_home we based our import on it09:28
← mikem left09:29
powerfox thiago_home: thanks. Now I feel save :)09:30
nud: thanks09:31
← mtodd left09:31
→ mtodd joined09:32
← boto left09:32
← hpa left09:32
← jriga_ left09:37
← archis_ left09:39
← tango_ left09:41
→ shodges joined09:41
← shodges left09:41
→ LotR joined09:47
→ hpa joined09:48
→ zumbrunn joined09:49
samjam When git diff is applied to a commit where an empty file was deleted, the resulting patch looks something like this (below); which the patch program rejects, saying "patch: **** Only garbage was found in the patch input."09:50
diff --git a/source3/include/dcerpc.h b/source3/include/dcerpc.h09:50
new file mode 10064409:50
index 0000000..e69de2909:50
It's hardly git's fault that it spits out a patch that patch is too dumb to interpret09:50
Does anyone have experience on the issue; is it worth optionally generating a patch like this?09:51
diff -Nru 1/here 2/here09:51
--- 1/here 2008-11-05 09:43:55.000000000 +000009:51
+++ 2/here 2008-11-05 09:43:58.000000000 +000009:51
@@ -0,0 +1 @@09:51
+09:51
diff -Nru 1/here 2/here09:51
--- 1/here 2008-11-05 09:37:23.000000000 +000009:51
+++ 2/here 1970-01-01 01:00:00.000000000 +010009:51
@@ -1 +0,0 @@09:51
-09:51
which makes the file non-empty and then lets it be deleted?09:51
charon points at where it says 'pastebin' in the topic09:52
→ swhitt joined09:53
samjam wonders how much is enough for the pastebin09:53
← Jaap left09:54
charon dunno. may want to search the list archives btw, i remember linus commenting on that 1970-01-01 magic timestamp a month or two ago09:55
samjam thanks.09:56
→ OlofS joined10:04
→ Guest51976 joined10:05
← bartman left10:07
← jony left10:09
→ jony joined10:09
powerfox Can someone please give me some git-perforce assistance? I did git-p4 sync //depo, so I have a full log, but no git objects...10:11
→ pantsman joined10:14
← kobani left10:15
← Jacolyte left10:19
imarcusthis is there something like git-svn that works the other way?10:22
cehteh doesnt git-svn work in both ways?10:22
← sciz left10:23
→ cilly joined10:23
→ hurikhan|Work joined10:25
← FunkeeMonk left10:27
samjam I think it does; there are two git svn tools, I think one is git-svnimport or something which is not two-way and seems to be out of favour these days10:27
→ stouset joined10:27
→ tango_ joined10:27
samjam git-svn dcommit will commit to svn10:28
→ LuboC joined10:28
← stouset left10:28
← Guest51976 left10:31
thiago_home powerfox: git p4 clone //depot/project/main@all10:32
← mithro left10:33
thiago_home add --detect-branches and drop "main" if you have proper branch descriptions (p4 branches)10:33
powerfox thiago_home: neither git-p4 clone "--detect-branches //all_in1/simplelib@all" or "git-p4 clone //all_in1/simplelib/main@all" didn't help. In all cases I get "Could not detect main branch. No checkout/master branch created." In second also "No changes to import!"10:37
all_in1 is a client name10:37
client shows10:38
"//simplelib/... //all_in1/simplelib/..." simplelib is a depo.10:38
← cilly left10:42
→ clairvy joined10:45
imarcusthis cehteh: i meant, something that lets svn users commit to a git repository.10:46
cehteh isnt that what git-svn does?10:47
doener I think he means something like git-cvsserver, but for svn10:49
→ parasti joined10:49
cehteh git storage backend for svn :)10:51
samjam So it seems that the only problem is when an empty file is removed for which there is no 1-step patch equivalent, git handles the case when a non-empty file is removed10:55
→ seezer joined10:55
← seezer left10:55
→ seezer joined10:55
← felipec left10:56
→ cedricv joined10:58
← tritz left10:59
← jamesm_ left11:00
← naeu left11:01
→ stouset joined11:02
→ Dark_Shikari joined11:02
→ bschindler joined11:03
→ naeu joined11:03
Dark_Shikari bah, I forgot how to go back and edit a past commit that wasn't the previous one11:03
whats the command again?11:03
Mikachu man11:05
Dark_Shikari I already read through it, none of its obvious11:05
much like all git commands11:05
Mikachu are you thinking of git rebase -i?11:05
← tango_ left11:06
Dark_Shikari seems so11:06
ok, now that I'm done with that, I have the following issue11:07
I have changes A, B, and C.11:07
I pushed change A, then change B, and then I made C locally.11:07
Change A had a typo, so I corrected it and want to force an update when I push again, to correct history.11:08
But I don't want to push C.11:08
I've corrected things locally, now I have to push the changes--without pushing C.11:08
→ xjjk joined11:08
Dark_Shikari what's the easiest way to do this?11:08
other than just rolling back A, B, and C, both locally and repo-side, and redoing them all11:08
→ Sho_ joined11:11
Dark_Shikari hmm, nvm, got it. I just killed commit C11:11
← imyousuf left11:12
← Dark_Shikari left11:12
bschindler Hi, I removed some binary data from git because it started to take too much space. But after removing it, the .git dir is obviously still very large. Is there a way to remove those two commits (adding binary data, removing it) so the .git dir will again be small?11:16
← gretchen left11:17
← CelticSoul left11:18
charon git rebase -i if your history is linear, git filter-branch otherwise11:18
then you'll have to git reflog expire and git gc to get rid of the associated objects11:18
make a backup first, all of those are sort of destructive11:18
bschindler I've got a remote ^^11:18
← clairvy left11:19
→ therrg joined11:19
← OlofS left11:20
← not-xjjk left11:21
→ earcar joined11:21
→ lorandi joined11:21
← therrg left11:22
→ muthu_ joined11:26
→ harinath joined11:27
cbrake_away → cbrake11:32
← muthu_ left11:33
← earcar left11:33
← mdalessio left11:35
nadim_ given a 'Merge with strategy recursive failed' when I do a git pull. How can I say take the one from origin and be done?11:38
vmiklos git reset --hard origin/master?11:39
doener heh :-)11:39
bschindler charon: I don't quite see how to do the rebase. I've got an almost linear history. But git-rebase -I gives me a noop - so I don't seem to understand exactly what I'm supposed to do :)11:39
← jony left11:40
← ankit9 left11:40
doener bschindler: you need to give the parent of the first commit that you want to edit/drop/reorder as upstream to rebase -i11:40
bschindler hmm okay...11:41
← popol_ left11:41
bschindler doener: now something bad just happened. I just checked out an earlier commit to save some of the files somewhere else, but now HEAD is on this early commit - is it not possible to get back to head?11:41
doener git rebase -i <upstream> <branch> operates on the commits <upstream>..<branch>, and <branch> defaults to the current one11:41
if you used "git checkout <commit>" you're on a detached HEAD now (git branch says "no branch")11:42
just check out your branch again11:42
→ tango_ joined11:42
bschindler ah11:42
→ shenie joined11:42
doener if you had used "git reset <commit>" you're branch would have been affected, and then you could use the reflog to find the other commit to reset back to it, JFYI11:43
→ boto joined11:44
← boto left11:44
← urbanmonk left11:45
→ jnareb joined11:51
→ grahal joined11:52
← fhobia left11:53
← shenie left11:56
→ reel__ joined11:57
jnareb Pieter: I have thought about adding the reverse of /whois, namely resolving real names (as they appear in commit messages) to IRC nicks to GitBot... till I realized that GitBot is written in Ruby, programming language which I have no experience with11:57
→ Ryback_ joined11:57
→ tritz joined11:58
→ mdalessio joined12:03
Pieter jnareb: ah :( If you have a design or so in mind on how to do it, if you send it to me I can create it12:03
← stouset left12:05
jnareb Pieter: how about "!nick A U Thor" trigger?12:06
Pieter jnareb: should be doable ($nick though).. where should it get the info?12:07
bschindler one question. Now when I'm done with rebasing (deleting two commits) - what happens when I do a git pull? I want those commits to be removed from my primary remote and then in the end on the other checkouts on other machines12:10
jnareb Pieter: from /whois realname; I don't know what library for IRC interaction it uses. either scan list of users every hour for example, checking nicknames which are not in database, and taking care to not flood the server, or check on join...12:10
← Flathead left12:12
jnareb bschindler: I don't know what you want to do exactly, but it might require pushing (or pulling) with the --force option or equivalent... unless you did change in unpublished part of history (generally changing published history is frowned upon, because it causes trouble for your downloaders/forkees)12:14
← reel_ left12:15
bschindler jnareb: Well, I understand that. I'm the only one checking out from this repository, so I can live with this12:15
→ ben_h joined12:15
→ bobmcw joined12:18
bschindler Okay, I have rebased now, but I'm not sure what's going on now: I got a Could not apply 178957f... Merge branches 'master' and 'viewer'. git status shows me 'not currently on any branch, nothing to commit'. What should I do now?12:25
jnareb bschindler: are you sure that you are not in the middle of conflicted rebase? e.g. forgot "git rebase --continue"?12:26
bschindler jnareb: this could very well be, but how can I resolve the conflicts?12:26
jnareb bschindler: if you are not, you cannot rebase on detached HEAD; you must first create a new branch ("git checkout -b <newbranch>")12:26
bschindler I mean, I don't see what's causing a conflict12:26
jnareb bschindler: git status, or git diff, or git ls-files -u12:27
bschindler jnareb: I rebased on master, not on detached head12:27
jnareb bschindler: and resolving conflict is editing conflicted files, "git add <file>" after resolving conflict, and then "git rebase --continue" (or "git rebase --skip" if you have removed all changes...)12:28
bschindler well I did a git status and didn't show any issues. I just continued and it seemd to have worked12:29
← cout left12:31
← bobmcw left12:32
→ cout joined12:32
→ releod joined12:34
bschindler jnareb: hmm... git gc did not really remove a lot. My pack file is still about 300mb's large. Is there a way to find out why?12:35
jnareb I don't know... note that git-gc still keeps objects pinned by reflog (up to reflog expire time)... either expiring reflogs (and losing undo info), or trying to pack righter with larger window and delta length size (see VCSCompare blog entry on this) could help (force repack, because otherwise git would keep old deltas etc.)12:38
← mdalessio left12:43
→ boto joined12:44
→ tvw joined12:45
dwmw2_gone → dwmw212:45
cradek_ → cradek12:47
→ mbuf` joined12:48
← pantsman left12:48
bschindler jnareb: in the reflog, I've got the checkout I did earlier which could cause this. I tried a git reflog expire e4d3e80..., but it told me that it points nowhere...12:49
← mbuf left12:50
← EmilMedve left12:50
jnareb bschindler: "git reflog expire --expire-unreachable=now --verbose <branch name>" then "git gc --prune" (and perhaps "git prune")12:51
bschindler: perhaps git reflog expire --dry-run first12:51
→ markelikalderon joined12:55
bschindler jnareb: that won't work either :(... I'm just thinking about creating a new tree and applying all the commits onto that new tree... is this kinda possible in git?12:56
→ cilly joined12:56
bschindler there aren't a lot of commits, so this should be pretty quick12:57
← LuboC left13:01
jnareb or just delete reflog, and check with "git fsck" if the commits you deleted are really unreachanble...13:02
← markelikalderon left13:04
← koke left13:04
→ markelikalderon joined13:04
jnareb checking with "git fsck" if commit is really unreachable is a good idea anyway to check if you have truly deleted the commits you want to expunge from reflog (or just try "git reflog show <refname>")13:05
dwmw2 → dwmw2_gone13:06
→ LuboC joined13:06
← LuboC left13:07
→ markelik_ joined13:08
jnareb Pieter: about GitBot $nick service: take note that one person can use a few different nicknames, like Petr Baudis with pasky, kampasky and gitsky13:08
← CIA-2 left13:10
doener says shadehawk ;-)13:11
→ oto joined13:11
← oto left13:11
jnareb true13:12
→ CIA-2 joined13:15
jnareb usually there is a reason or season, like (from what I guess) Petr using pasky and kampasky on different machines, and gitsky for GitTogether13:15
or me using jnareb and ShadeHawk in different IRC clients13:15
doener I didn't meant to criticize that, just wondered why you took pasky as an example ;-)13:18
jnareb well, that what came to mind...13:18
← cilly left13:20
← deskin left13:20
→ LiamH joined13:23
← markelikalderon left13:23
jnareb See you later13:23
← jnareb left13:23
← cedricv left13:24
→ stouset joined13:24
← markelik_ left13:27
→ jriga joined13:28
← Mario left13:35
→ Beket joined13:38
← mbuf` left13:39
← ben_h left13:39
← reel__ left13:39
← mtodd left13:39
← lamont left13:39
← nadim_ left13:39
← nud left13:39
← reithi left13:39
← SRabbelier left13:39
← tjafk2 left13:39
← lcantey left13:39
← sugi left13:39
← awarde left13:39
← ft left13:39
← jackdempsey left13:39
← Hallin left13:39
← vbgunz__ left13:39
← ruphy left13:39
← mpr left13:39
← dirker left13:39
← dkr left13:39
← precision left13:39
← masterdriverz left13:39
← Thumper_ left13:39
← Jc2k left13:39
← broonie left13:39
← GarethAdams left13:39
← RandalSchwartz left13:39
← Illocution left13:39
← AlexanderS left13:39
← resmo left13:39
← redondos left13:39
← cardioid left13:39
← jmspeex left13:39
← jpeeler left13:39
← mika left13:39
→ vbgunz__ joined13:39
→ mbuf` joined13:40
→ ben_h joined13:40
→ reel__ joined13:40
→ mtodd joined13:40
→ lamont joined13:40
→ nadim_ joined13:40
→ nud joined13:40
→ reithi joined13:40
→ SRabbelier joined13:40
→ tjafk2 joined13:40
→ sugi joined13:40
→ awarde joined13:40
→ ft joined13:40
→ jackdempsey joined13:40
→ Hallin joined13:40
→ ruphy joined13:40
→ mpr joined13:40
→ redondos joined13:40
→ jpeeler joined13:40
→ mika joined13:40
→ GarethAdams joined13:40
→ jmspeex joined13:40
→ RandalSchwartz joined13:40
→ resmo joined13:40
→ precision joined13:40
→ dkr joined13:40
→ Thumper_ joined13:40
→ dirker joined13:40
→ Illocution joined13:40
→ masterdriverz joined13:40
→ cardioid joined13:40
→ AlexanderS joined13:40
→ broonie joined13:40
→ Jc2k joined13:40
→ _zack joined13:40
← mbuf` left13:41
← broonie left13:41
→ broonie joined13:41
← ruphy left13:41
_zack if not, is there any deep reason for that?13:41
← Illocution left13:42
→ Illocution joined13:42
→ jriga_ joined13:43
← jriga left13:43
→ qrush joined13:49
← ben_h left13:51
→ ben_h joined13:51
parasti _zack: if not what? it seems your question went missing13:55
_zack ah, trying again13:55
is it possible to restrict the DenyNonFastForward setting to a specific branch (pattern) ?13:55
→ kukks joined13:55
_zack parasti: better now?13:55
parasti much better13:56
→ name joined13:56
→ rdavila joined13:56
← rdavila left13:57
← a-atwood_ left13:58
→ a-atwood joined13:58
→ mw__ joined14:02
→ coshx joined14:06
← stuffcorpse left14:07
→ stuffcorpse joined14:07
→ muthu joined14:11
→ juanjoc joined14:12
← Beket left14:15
→ EmilMedve joined14:15
→ bdiego joined14:16
→ bartman joined14:16
→ SjB joined14:20
← nud left14:21
→ clairvy joined14:24
→ litage joined14:24
→ nud_ joined14:25
→ selckin joined14:25
→ threeve joined14:26
← EmilMedve left14:26
← vmiklos left14:30
→ vmiklos joined14:30
→ koke joined14:30
→ harinath1 joined14:32
→ bronson joined14:32
← tchan left14:36
→ tchan joined14:37
← telmich left14:37
← stouset left14:39
← clairvy left14:39
madduck _zack: no, not possible, unless you write a hook.14:40
the setting is black-white14:40
a receive hook should be able to do that though14:40
litage is it possible to determine the latest commit to affect a specific line in a specific file?14:40
Mikachu litage: git blame file14:40
← krh left14:40
madduck litage: pickaxe14:40
_zack madduck: thanks14:40
litage thanks guys, reading manpages now..14:40
madduck litage: git blame is better\14:40
(for this)14:40
doener pickaxe is long deprecated and just an alias for blame nowadays14:40
→ EmilMedve joined14:41
→ krh joined14:41
madduck i didn't know that. :)14:41
cbreak git log file?14:41
→ clairvy joined14:41
madduck git log is awesome too14:41
Mikachu i think pickaxe in gitweb does git log -S, fwiw14:41
→ Yuuhi joined14:41
doener cbreak: that's not for a specific line though. At least not easily14:41
cbreak hmm... it works if the file only has one line...14:43
admitedly a special case.14:43
doener heh14:43
← harinath left14:44
← clairvy left14:44
harinath1 → harinath14:44
→ d0k joined14:45
→ clairvy joined14:46
← scientes left14:50
→ scientes joined14:54
→ jelmer joined14:54
→ willb joined14:54
→ kmap joined14:58
← ben_h left14:58
→ gercokees joined14:58
← reithi left14:59
→ rdavila joined15:00
← _zack left15:01
→ bobesponja joined15:02
← loincloth left15:02
← Sho_ left15:02
qrush git log --pretty=format:%H origin..HEAD15:05
that's not working for my current repo...any reasons why?15:05
fatal: ambiguous argument 'origin..HEAD': unknown revision or path not in the working tree.15:05
Use '--' to separate paths from revisions15:05
this worked fine on Ubuntu but not on osx15:05
shd try: git remote add origin REPOPATH15:08
→ ben_h joined15:08
qrush fatal: remote origin already exists.15:09
shd does origin/master work?15:09
doener origin == origin/HEAD, IIRC you get that only from git clone15:09
just use the "real" remote tracking branch instead, eg. origin/master15:10
qrush so git remote add origin/master REPOPATH worked15:10
but that doesn't help the log command any15:10
→ stephenos joined15:10
qrush or are you saying i have to push to origin/master now15:11
doener adding a remote called origin/master is, uhm, "unusual"15:11
qrush is very confused.15:11
doener see, your remote is/was called "origin", right?15:11
qrush yes15:11
doener then you have remote tracking branches like origin/master, origin/foo etc.15:11
and when "git clone" created that remote, you also get origin/HEAD15:12
← powerfox left15:12
doener and _if_ origin/HEAD exists, you can just use "origin" instead of "origin/HEAD"15:12
normally, when using just "origin", you mean the remote, but "git log" expects a committish (commit, branch, remote tracking branch, whatever)15:13
qrush So what is that log command trying to do then?15:13
doener and then "origin" is expanded to "origin/HEAD"15:13
in your case, you probably don't have origin/HEAD, so "origin" can't be found15:13
you should just use the actual remote tracking branch you're interested in instead15:14
git log origin/master..15:14
← bronson left15:14
doener or git log origin/foo.. if you're interesting in origin/foo and not in origin/master15:14
→ ltbarcly joined15:14
shd qrush: you should delete the remote name "origin/master" first.. it makes git syntax ambiguous15:14
qrush i already deleted it.15:15
ltbarcly how can I revert a file to a previous version?15:16
I have tried for about an hour, all I can get it to do is move the entire repository to a previous commit15:16
cbreak you can git checkout it15:16
qrush I'm still confused about what to do about this log command15:16
ltbarcly I tried that every which way15:16
git checkout hash:filename15:16
git checkout filename hash15:16
git checkout hash filename15:16
qrush git log --pretty=format:%H origin..HEAD < what is that trying to do15:16
ltbarcly I can't find an example anywhere15:16
shd ltbarcly: git checkout OLDCOMMIT -- filename ; git commit ...15:17
cbreak git checkout hash -- file15:17
ltbarcly thaks!15:17
cbreak you need to git add it too15:17
if you want :)15:17
→ leavengood joined15:17
→ FunkeeMonk joined15:18
ltbarcly that worked well, thanks so much15:18
phrixos Suppose I have two branches, 'a' and 'b', and 'a' is a parent of 'b'15:18
qrush ok...it's <since>...<until>15:18
phrixos is there are a way of finding out how many commits there are between a and b?15:18
ltbarcly that should be in the docs somewhere!15:19
phrixos diff or diff-tree don't suggest anything interesting15:19
→ rmh3093 joined15:19
phrixos i suppose i could log and then wc -l15:20
leavengood phrixos: I was thinking along those lines15:21
phrixos i guess that'll have to do it15:23
qrush so what this log command is trying to do is similar to phrixos' request15:24
it wants to list the commits that aren't pushed yet15:24
so origin..HEAD15:24
→ spearce joined15:24
qrush this DEFINITELY worked on my copy of the repo on Ubuntu.15:24
I really don't get why this isn't working on my copy of the repo in OSX.15:24
← scientes left15:25
→ paltman joined15:25
← clairvy left15:25
shd qrush: note: a..b is not the same as a...b15:26
← CIA-2 left15:26
qrush Well, neither work.15:26
http://pastie.org/29804315:27
It's part of this series of commands.15:27
→ LuboC joined15:27
→ aroben joined15:27
← eternaleye left15:27
← derRichard left15:28
← spearce left15:28
→ sdboyer joined15:28
leavengood I built git from its repo last night and git submodules is not working on another project (which I am new on)15:29
→ CIA-2 joined15:30
leavengood which branch should I build from for a stable git?15:30
maint?15:30
← EmilMedve left15:31
→ gma joined15:31
← ben_h left15:31
gma is it a bad idea to copy a repo to another host then run "git remote add srchost ssh://srchost.lan/path/to/repo" ?15:31
I'm finding I can't push/pull between them, and get "error: remote 'refs/heads/master' is not a strict subset of local ref 'refs/heads/master'" when I push back to srchost...15:32
← mw__ left15:32
← bschindler left15:32
→ bschindler joined15:33
→ diskussions_D joined15:34
→ aroben_ joined15:36
→ stouset joined15:37
← quoin- left15:38
→ davidfetter joined15:39
→ girishr joined15:40
→ careo joined15:41
qrush well, i solved it by blowing away my repo, and cloning again15:41
sucks but now what i wanted works15:41
→ Araneidae joined15:41
Araneidae How do I see what the origin of my repository is? (I think I have more than one...)15:42
→ incidence joined15:42
Araneidae Ah: .git/config -- but I don't understand what it means to have two or more remotes!15:43
incidence Hey, does Git have "versioning" files like SVN has "$Id" "$Author" or something?15:44
PerlJam incidence: why do you ask?15:44
→ bobmcw joined15:44
diskussions_D Doh.. I just did a git merge --ff branch_name when I meant to do a git merge --no-ff branch_name is there a way to revert that change somehow?15:44
PerlJam incidence: are you going to perpetuate CVS's mistakes?15:44
incidence PerlJam, I am just getting used to Git. :)15:44
diskussions_D something like git reset ORIG_HEAD or something? perhaps git reset @{1}15:44
→ jerbear joined15:44
← nadim_ left15:45
← gercokees left15:46
Araneidae Is there an explanation somewhere of how multiple remotes work?15:48
I'm trying to add the stable point releases to my kernel git, and I don't understand what I'm doing at all!15:48
← jas4711 left15:48
← jerbear left15:49
→ rubydiamond joined15:49
→ jas4711 joined15:49
Araneidae I can see that I have a remote called stable25 ... and I guess I want to add the corresponding .26 ... but I'm flying blind here.15:49
`git add remote stable26 git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-2.6.26.y.git` will presumably do the trick ... but what the hell is it doing?!15:50
parasti Araneidae: cp .git/config .git/config.orig; git add remote ...; diff -u .git/config .git/config.orig15:50
though it's actually git remote add15:51
Araneidae Oh, yes15:51
Well, looking at config it's kind of easy to see what's happening there ... but what about the commit structure?15:52
I mean, how does tag v2.6.26.x relate to what I've already got.15:52
And how do I safely update my existing remote25 -- do I use `git fetch` for that?15:52
→ icwiener joined15:53
← ahhsm left15:53
Araneidae I see: adding the remote *only* updated .git/config, is that right?15:54
offby1 yep15:54
→ spearce joined15:55
← aroben left15:55
offby1 if you "fetch" from that remote, I think you'll get some new objects in your repository, and some new refs under refs/remotes but it won't hose anything. I think :)15:55
might wanna try it out on a clone to be sure15:55
Araneidae That's what I'm hoping ...15:55
Ya15:55
I gather that cloning a git repository with cp is dangerous15:56
→ ahhsm joined15:57
sstangl not at all15:57
→ EmilMedve joined15:57
sstangl git only persists in the .git directory; you can move the entire hierarchy around at whim.15:58
Araneidae mv is one thing - but cp can break hard links15:58
offby1 sure, but so what?16:01
worst that'll happen is that the clone will use as much disk space as the original. Big whoop16:01
→ sgrimm joined16:01
offby1 I "clone" with rsync (which is just a faster version of cp) occasionally.16:01
→ loincloth joined16:01
offby1 Note that the result isn't strictly speaking a clone at all16:01
it's just another repository that has the same contents16:01
← loincloth left16:01
Araneidae Anyhow, `git remote add stable26 ...; git fetch stable26` seems to have worked ok.16:01
offby1 sure16:01
Araneidae Wish I didn't get all the *-rc? tags -- I have to delete them every time!16:02
→ jony joined16:04
→ vbgunz_ joined16:05
→ loincloth joined16:06
→ cedricv joined16:07
← lorandi left16:08
→ lorandi joined16:08
← CIA-2 left16:09
→ Beket joined16:10
offby1 why?16:10
are they stealing your good silver?16:10
Araneidae Problem is I can't see the interesting tags for all the noise16:11
→ CIA-2 joined16:11
← girishr left16:16
brettryan If I've added something to .gitignore that I previously forgot, how do I tell git to ignore current changes16:17
← SjB left16:17
brettryan and is it possible to rewrite the history to contain no reference to those files?16:17
← vbgunz__ left16:18
← bobmcw left16:18
→ G_SabinoMullane joined16:18
litage does git support nested branches?16:20
← koke left16:20
→ mw joined16:20
← mw left16:20
→ mw joined16:21
← mw left16:21
→ mw joined16:21
→ insulina joined16:22
insulina hello, how do i say git to use nano insted of vim16:22
?16:22
jast litage, i think you don't understand how branches work. in a way, yes, it does, but that's just because branches are just like sticky labels that refer to a specific commit in the overall graph of commits.16:22
insulina, for example, change the environment variable GIT_EDITOR to /usr/bin/nano (or whatever)16:22
alternatively you can set core.editor in ~/.gitconfig16:23
brettryan, .gitignore only ever applies to untracked files, if that's what you were asking.16:23
offby1 Araneidae: ah, gotcha16:23
jast brettryan, as for the other question, it's possible to filter a repository, changing all commits in certain ways. so yes, it can be done, but it changes the IDs of many commits and thus makes merging with unfiltered copies of the repository impossible.16:24
azuk insulina, git also uses generic VISUAL and EDITOR environment variables which may be set to vim in your system16:24
insulina, so you might be better off setting either of them instead of GIT_EDITOR assuming you want to use nano elsewhere as well and not just with git16:25
insulina ok tx16:26
brettryan jast: thankyou, :) I discovered git rm --cached $file will remove previously tracked files to no longer be tracked16:26
though the history will still remain16:27
jast that's right16:27
brettryan that's not so bad, but there's about 1GB of previously tracked content that doesn't need to be there ;)16:27
jast as i said, if you don't mind making the repo unmergable with other copies of it, you can filter all previous stuff16:28
→ SjB joined16:28
ahhsm apparently I modified some files on my master branch, then checked out a new branch, are those modified files automatically added to the index when the new branch is created?16:28
brettryan jast: How would I go about performing that?16:28
ahhsm because when I commit, it seems to want to commit those files and I don't want it to16:29
→ cory_ joined16:29
jast brettryan, basically it works along the lines of git filter-branch --index-filter 'git rm -f --ignore-unmatch file1 file2 ...' branch1 branch2 ...16:29
oh16:30
git rm --cached -f, in fact16:30
(and the rest)16:30
← icwiener left16:30
brettryan no problems jast, thankyou very much for your help16:30
→ Pistos joined16:30
brettryan It's no biggie if I can't get the content out, I'll play with a new test repo to try it out :)16:31
jast after the filtering it will be a bit of a challenge to actually get rid of the superfluous objects, in fact16:31
because git is very careful about removing stuff16:31
brettryan ah right, I might just leave it alone, lol16:31
Pistos I know about mergeoptions: I've been able to put --no-ff in there. But what I really want is --no-ff on a literal merge, but normal (allow ff) on a pull. Is this possible?16:31
jast Pistos, perhaps something like alias.xmerge = merge --no-ff?16:32
→ jesselucas joined16:32
← doener left16:32
Pistos jast: I _guess_ that's a workaround...16:32
→ doener joined16:32
jast i don't think it's currently possible to configure merge to act differently when used by pull16:32
Pistos jast: My whole thang is that I like having a marker in the git log when I do literal branch merges, but whenever I am pulling from remotes (which is fairly often), it is too much litter.16:33
→ markelikalderon joined16:33
Pistos "Merge branch 'master' of [email@hidden.address]16:33
jast: I will try your alias workaround for a while, see how it turns out.16:34
← charon left16:34
← leavengood left16:34
← nud_ left16:36
→ aroben joined16:36
→ kumbayo joined16:36
→ p4tux joined16:37
← Beket left16:39
← davidfetter left16:40
→ bobmcw joined16:44
← reel__ left16:45
← qrush left16:45
Modius_ → Modius16:47
→ digitalhobbit joined16:48
← tritz left16:50
→ aroben__ joined16:50
→ chrizz joined16:51
← aroben_ left16:52
jesselucas I'm just switching from SVN to git. I use RapidSVN, and my question is with SVN I can view the log history and then get a specific revision. How do I do this with git?16:54
cbreak you can use git log and then git checkout16:54
Pieter :)16:57
← diskussions_D left16:58
← jriga_ left16:58
→ aroben_ joined16:58
← aroben left16:58
← aroben__ left16:58
aroben_ → aroben16:58
jesselucas cbreak; Do I use the SHA1 ID to tell git checkout which commit?16:59
← coshx left17:00
Pistos jesselucas: Yep.17:00
bremner jesselucas: or the first few characters anyway17:01
→ CodeOfficer joined17:01
→ insulina_ joined17:02
← insulina_ left17:03
← jony left17:04
→ hidek_ joined17:10
jesselucas I set git config --global core.editor "mate -w" . To change my editor to textmate, but how do I change the git log viewer to textmate?17:11
← SjB left17:12
jast you mean the pager?17:12
core.pager :)17:12
jesselucas thanks jast!17:12
← CIA-2 left17:12
jast if you really want to spawn a gui editor every time something gets paged17:12
Pistos Heh.17:12
jesselucas ahh, good thought. I guess nano it is17:13
jast seriously, why not something like less or most?17:13
i mean, it's not like you need to edit the logs etc.17:14
jesselucas jast: I'll definitely try them out. I'm just ignorant on what is appropriate, not really a sys admin. :)17:15
jast less is the default, actually (if it's installed, which it is in most environments)17:15
→ CIA-3 joined17:15
← careo left17:18
← bentob0x left17:18
← insulina left17:19
→ SjB joined17:20
→ EiNZTEiN joined17:21
← SjB left17:21
← naeu left17:21
→ SjB joined17:21
← tvw left17:21
→ careo joined17:23
→ jony joined17:23
→ piclez_ joined17:26
→ deskin joined17:27
→ tritz joined17:28
→ carllerche joined17:31
dotsintacks msys-git's sh.exe is crashing on me all of the sudden17:32
sh.exe.stackdump says STATUS_ACCESS_VIOLATION17:32
← sgrimm left17:32
dotsintacks no function names tho17:32
has anyone on windows ever run into this?17:33
← bschindler left17:34
← Pistos left17:37
← a-atwood left17:40
→ a-atwood joined17:41
→ ruphy joined17:43
← ciskje left17:43
→ zeeeee joined17:44
zeeeee hi, i ran git svn clone on machine A and git clone of A onto machine B. normally when working on A everything is fine - i can git svn rebase and git commit changes - but i committed some changes at B and pulled to A, and now git svn rebase shows me conflicts. is this due to my not having rebased and instead implicitly merged when doing the pull? how should i fix this?17:46
→ iulian joined17:47
→ naeu joined17:48
deskin zeeeee: do the base remote refs have the same hash? I've had git-svn mess things up, in that two clones don't necessarily translate svn info to the same commit objects17:48
← justeco left17:49
→ koke joined17:49
deskin that, and git svn doesn't like merge commits17:49
← mediogre left17:49
zeeeee deskin, i don't know what that means... i'm new to git. what are base remote refs? how do i check if they have the same hash? (fwiw i don't think that machine B is aware of the svn repo, because git svn rebase never worked on it)17:51
deskin oh, I misread17:51
thought B was a clone via git-svn too; never mind that point17:52
zeeeee deskin, oh17:52
deskin I'd guess that it's due to the merge commits; you could try -p flag to rebase, to preserve merges17:53
in general, it's best to avoid merge commits when working with git-svn17:53
← ciaran left17:55
zeeeee deskin, right, i suspected that the merge commit was the problem. i just don't know how to fix this now. you say i should use git rebase -i -p? here's a screenshot of my situation: http://img128.imageshack.us/my.php?image=githorizonticagg4.png17:56
deskin, that merge at the top is the problem (it's what i committed at machine B and pulled to machine A). this is gitk running on A. i don't know how to put everything into a single line.17:56
← naeu left17:58
→ ciaran joined17:58
→ naeu joined17:58
← ciaran left17:58
deskin so the right branch of the merge was the commits you did on B?17:58
→ digitalhobbit_ joined17:59
← naeu left17:59
→ naeu joined17:59
← deskin left18:01
→ deskin joined18:01
→ dreiss joined18:02
→ ciaran joined18:03
→ Weaselweb joined18:04
→ helo joined18:04
helo is there a good project management tool that integrates with git well, ala trac+svn?18:04
→ woeye joined18:05
→ Daenyth joined18:05
bremner helo: did you investigate trac+git ?18:05
Daenyth Is there a way I can partially revert a commit? A few dozen files were changed and I want to undo the changes to 2 or 318:05
→ taruti joined18:05
← digitalhobbit left18:05
jast well, you can git revert -n and then mess with git add -i18:06
zeeeee deskin, no, the left branch is18:06
taruti Is there something like "git submodule update" that would use a branch instead of checking out invidual commits?18:06
wereHamster Daenyth: git-checkout the files and commit the new versions18:06
Daenyth you can checkout individual files?18:07
oh ok18:07
wereHamster Daenyth: git checkout HEAD^ -- path/to/files18:07
taruti (basically foreach submodule do (cd $submodule && git pull))18:07
Daenyth wereHamster: snazzy, thanks18:07
← rmh3093 left18:07
← neuro_damage left18:07
tritz Daenyth: see the examples in 'git reset --help': Undo a commit and redo18:07
helo bremner: yeah, it appears the plugins aren't very robust18:07
wereHamster Daenyth: only use git-reset if you haven't published the history already.18:08
Daenyth someone else made "syntax cleanup" changes and broke some of my code18:08
already pushed it -.-18:09
→ ahupp|fb joined18:09
Daenyth can I git checkout goodcommit -- whole_directory/ ?18:09
jast yes18:09
Daenyth mmk18:09
context how do you checkout a new branch with no parent18:09
jast that won't automatically remove files that didn't exist in that version of the directory, so look out for that18:09
zeeeee oh...whoa....wait...look at where they branch off: http://img61.imageshack.us/my.php?image=githorizontica1yu0.png18:09
deskin zeeeee: I'd try git svn rebase -p; if that doesn't work, then I'd make a new branch pointing to master; reset --hard to the last commit from B; and rebase --onto the last commit from A (just before the merge commit), using 'master' as the upstream18:09
jast context, it's a bit complicated because it's extremely discouraged. :)18:10
context jast: yeah i know18:10
Daenyth jast: thanks18:10
zeeeee deskin, why did git reproduce the changes in both branches??18:10
→ acid2 joined18:10
context friend wants to im gonna try talking him out of it18:10
acid2 I'm doing a rebase and have some merge problems. Is there anyway to resolve all conflicts with the "new" part of the conflict?18:10
zeeeee i actually can't tell right now which branch was machine A and which branch was machine B. the change at the very top was the one i made at machine B.18:11
deskin zeeeee: did you clone B before dcommitting? dcommit rewrites your commits, so they'll have the same result but different sha1s18:11
zeeeee deskin, i never dcommitted18:12
← MarkMenard_ left18:12
jast context, fwiw, it works like this: git symbolic-ref HEAD refs/heads/newbranch && rm .git/index; ...; git commit18:12
→ rmh3093 joined18:13
← muthu left18:14
deskin I don't know what's going on there; perhaps it's due to rebase failing?18:14
→ coshx joined18:16
bremner helo: the only other thing I know about is ticgit, but that is pretty different from trac18:17
→ ciaran_ joined18:17
← lorandi left18:18
→ jordibunster joined18:19
← bobmcw left18:19
jordibunster Any obvious reason why git cvsimport would return "invalid argument" when a cvs checkout works?18:19
← jesselucas left18:21
deskin zeeeee: at any rate, I'd reset to before you merged, and instead rebase the commits from B onto the commits from A; that way you'll have linear history18:22
git pull --rebase might be helpful18:22
→ vintrepid joined18:23
→ ciaran__ joined18:24
← Tv left18:25
← fdr- left18:26
deskin zeeeee: I gotta go, hope some of that helps18:29
← deskin left18:29
bdrewery jast: why is creating a new branch without a parent discouraged? It seems to work fine, fsck doesnt complain. I'm just wanting to track 1 or 2 notes files in the small branch (not pushing it remote)18:31
→ ThaDon joined18:32
← cedricv left18:32
→ niki joined18:33
bdrewery I would think that if I ever decide it was stupid I could just remove the branch ref and has fsck clean out the entire tree18:34
← ThaDon left18:36
← Pupeno-M left18:36
zeeeee can anybody explain how to do what deskin just suggested? that's exactly what i had set out to figure out from the start.18:37
→ sgrimm joined18:38
→ mtodd__ joined18:39
← woeye left18:39
← ciaran_ left18:39
jast bdrewery, it tends to be a confusing alternative to just creating a separate repository. it does work well enough, apart from that.18:42
← loiseau left18:43
bdrewery k18:43
← ciaran left18:44
← coshx left18:45
→ _graham_ joined18:45
← mtodd left18:46
ciaran__ → ciaran18:47
→ jesselucas joined18:51
← taruti left18:52
← jordibunster left18:52
← tango_ left18:53
← ahupp|fb left18:53
← ciaran left18:54
→ ruthd joined18:54
ruthd anyone here?18:55
bremner ruthd: nope, nobody here but us chickens18:55
ruthd =)18:56
hallo chickens18:56
i come in supplication18:56
→ joshdavey joined18:56
ruthd so I am trying to switch branches18:56
in the current branch i just applied a patch which added some files and modified others18:57
now i want to switch to another branch without said modifications (git checkout {branch}, is that right?)18:57
bremner ruthd: right18:57
ruthd only it's all like, fatal: Entry '{filename}' is not uptodate. Cannot merge.18:57
→ supercali joined18:58
ruthd and I don't want to merge; I just want to switch18:58
and this whole branch thing is very confusing.18:58
bremner did you commit your changes ?18:58
← jony left18:58
ruthd no, do I have to do that before switching?18:58
thiago_home yes18:58
git can't switch to another branch in this particular case18:59
bremner ruthd: unless you use e.g. git stash18:59
thiago_home from your current branch to the other one, the file {filename} was modified18:59
but you have local changes in it18:59
so git refuses to switch to the other branch so that you won't lose your changes18:59
ruthd ok18:59
so if i commit them and then switch, the other branch won't have the committed changes, will it?18:59
thiago_home correct18:59
it won't contain them19:00
ruthd how about the added files?19:00
thiago_home when you switch branches, you check out a clean copy of the other branch19:00
unmodified19:00
ruthd okay.19:00
wsuff there is a git uktil script called git-switch will stash your local changes then change if u don't want to commit them yet stash is an alternative19:00
thiago_home so, are you asking instead "I want to commit these changes to branch B, but I started doing my work in branch A?"19:00
→ jony joined19:00
← stouset left19:01
→ CelticSoul joined19:02
→ jriga joined19:03
mw → mw|food19:03
→ Sigma joined19:07
zeeeee hi, would any git gurus here be able to help me out with this question? http://marc.info/?l=git&m=122591215617019&w=219:10
→ Sonderblade joined19:11
→ fhobia joined19:12
→ deskin joined19:13
context zeeeee: this is why i avoid pull .19:13
zeeeee: pull = fetch & merge ...19:13
zeeeee context, ah. so fetch only, then rebase?19:13
context yes.19:13
zeeeee context, but in my case, is there any way to fix things?19:13
← gma left19:13
context you can reset the branch19:14
← SjB left19:14
→ tango_ joined19:14
zeeeee context, so...i should reset master... to something before the pull?19:15
context yes19:16
← rubydiamond left19:20
← rdavila left19:20
→ coshx joined19:22
→ rubydiamond joined19:24
→ shenie joined19:24
→ stouset joined19:28
→ the_undefined_ joined19:29
the_undefined_ does anybody here have some hints when setting up a git user with git-shell on handling permisssions?19:30
← dreiss left19:30
the_undefined_ I currently have the idea of having one git user with full access to all repositories. And then create a new user for every project that needs access control19:31
← joshdavey left19:31
the_undefined_ I could make the projects folder belong to the group of the project user19:31
and then simply select how gets access by copying the right SSH keys to the user account belonging to the project19:32
so I have 1 git user that can access all projects (as he owns them)19:32
and 1 user per project that I use to do ACL with by simply only adding the ssh keys of people I want to work on a certain repository in19:32
→ _eric joined19:33
the_undefined_ So ... my question is: Does this kind of setup make sense or is it a stupid approach? : )19:33
← Strogg left19:33
→ Strogg joined19:34
→ bobmcw joined19:35
→ patmaddox joined19:35
← rubydiamond left19:36
coshx the_undefined_: you might want to check out gitosis, which pretty much does exactly what you're thinking of, with very simple configs (and doesn't require you to add user accounts for all the users, but rather just copy their public keys into a keydir)19:36
→ ashleyw joined19:37
← tcoppi left19:38
→ larspind joined19:39
the_undefined_ coshx: I've seen gitosis, I just wondered if I really need an additional tool + python dependency or if I can get away with this simple hack as well :)19:39
coshx the_undefined_: if you're using the remote repository as a central repository, then it's only a python dep on the remote server, and the clients see it as a normal git repository19:40
the_undefined_: it makes it dirt simple to add repositories and users and set permissions, and is also more secure than adding system users19:40
the_undefined_ coshx: you think git-shell is not secure?19:41
coshx the_undefined_: I've never used git-shell, but it's probably secure. i just didn't get the feeling it does what you want :)19:41
→ dancor joined19:42
the_undefined_ coshx: Well lets say I wouldn't have more then 15-20 projects19:42
coshx: I actually want to do a screencast on how to roll your own git server for small teams to collaborate19:43
coshx: And I'm hesitant recommending people to use gitosis since there is no documentation yet and not everybody is going to want python on there boxes19:43
* their19:44
reuss what's the easiest way of preventing a user to be able to push back to origin, even if he has write access to the files. I presume something in a hook, but what to put in there? i wan't to deny every push, no matter what.19:44
coshx the_undefined_: personally, i'd rather only have 1 git user (who can only run gitosis-serve) on the system, and manage who can login as git (but only to run git commands) through gitosis, than to have to create system accounts and make sure the shell is git-shell (and also keep up-to-date on any possible security advisories for that shell)19:44
dancor is there an easy way to change the commit message for a commit that is the only commit in a branch?19:44
coshx the_undefined_: if you have a screencast for setting up gitosis, then you'll have the best documentation :)19:45
dancor normally i reset --soft to the previous commit19:45
the_undefined_ coshx: : )19:45
coshx: are you using gitosis?19:45
coshx the_undefined_: yeah, and it's working great. Because the config file is in its own git repo, I just make changes to it locally and push those changes, and voila, I can now use a new remote repo19:46
→ bentob0x joined19:46
coshx the_undefined_: so you don't even have to login to the remote server to add users or projects19:46
→ w00t joined19:47
the_undefined_ coshx: well my idea could be automated as well ..., but I get the point : )19:47
← vintrepid left19:47
→ foca joined19:48
the_undefined_ coshx: I just don't like that the gitosis repo hasn't seen activity for 2 month I guess19:48
bobesponja the_undefined_: gitorious has it automated19:49
← helo left19:49
the_undefined_ bobesponja: are they using the multi-user approach?19:49
bobesponja the_undefined_: multiple user approach?19:49
coshx dancor: check out git-commit --amend19:49
the_undefined_ bobesponja: have you followed the conversation?19:50
← eno__ left19:50
→ nud joined19:50
bobesponja the_undefined_: you can have various committers per projects with gitorious19:50
the_undefined_: and each user can be added from a web interface19:51
dancor sweetness19:51
the_undefined_ bobesponja: I want to do a screencast teaching people how to roll their own central git server so small teams can collaborate19:51
bobesponja by the owner of the project19:51
the_undefined_ bobesponja: In terms of recommending a service out there I'll probably send folks towards github as I like and use them myself19:51
bobesponja the_undefined_: you can try it here http://gitorious.org19:51
→ SjB joined19:52
bobesponja the_undefined_: gitorious can be installed on your own server, it's not just a service19:52
→ vintrepid joined19:52
the_undefined_ bobesponja: Oh, I did not know that19:52
bobesponja: thanks19:52
→ tvw joined19:52
the_undefined_ bobesponja: What is it written in?19:52
bobesponja the_undefined_: rails19:53
→ joshdavey joined19:53
→ johnw joined19:53
the_undefined_ bobesponja: ouch ...19:53
bobesponja the_undefined_: what teck are you looking for?19:53
← Cacheaway left19:53
the_undefined_ bobesponja: Ok maybe I should have mentioned that, the screencast is for people who have limited server administration skills, setting up a rails stack is not an option : )19:53
→ tcoppi joined19:54
bobesponja the_undefined_: gem install rails is not too hard19:54
→ Mario_ joined19:54
→ pygi joined19:54
← Mario_ left19:55
the_undefined_ bobesponja: -bash: gem: command not found19:55
; )19:55
bobesponja the_undefined_: gitosis requieres bash too, and so do git19:55
ah19:55
there's an exec19:56
the_undefined_ bobesponja: I might do a screencast on more fancy setups as well19:56
← larspind left19:56
the_undefined_ bobesponja: But right now I'm looking for the least involved solution offering some basic ACL19:56
bobesponja go for gitosis then, but it's not entirely automated, gitorious might be more difficult to install but then automate things for new collaborators19:57
the_undefined_ bobesponja: If people need to have a GUI and click & point administration but are unable to set it up - I guess they really should use a hostet solution as I'd be scared to see them run any complicated stack on their servers :)19:57
bobesponja: are you involved with the project itself?19:58
bobesponja I used to19:58
← bartman left19:58
the_undefined_ bobesponja: so what is the current status? Is it being maintained ?19:58
jesselucas just curious. What does "resolving deltas" mean?19:59
bobesponja jesselucas: resolving diffs19:59
pygi the_undefined_, bobesponja is silly, dont listen to him19:59
:)19:59
← EmilMedve left19:59
jast resolving deltas is related to how git compresses data in its pack files19:59
the_undefined_ pygi: so which git acl solution have you written? :P19:59
bobesponja the_undefined_: it is maintained but not extremely active19:59
jast when you run git gc, it looks for very similar files and expresses one in terms of the other, saving lots of space. to read those files, you need to resolve the delta information.20:00
bobesponja the_undefined_: he's just a troll, /ignore pygi =)20:00
pygi the_undefined_, none20:00
but I can haz powerz over bobesponja !20:01
the_undefined_ bobesponja: gitorious actually looks nice20:01
bobesponja the_undefined_: gitosis is the simplest so go for it, you will need a unix server though, not sure if that enters in your requierment for simplicity :)20:01
pygi bobesponja, how's the writing going?20:02
bobesponja I finished chapter 3, waiting for the "project manager" approval then we might finally sign a contract20:02
the_undefined_ bobesponja: EC2 has a bunch of them laying around ... ; )20:02
→ dreiss joined20:02
the_undefined_ bobesponja: you writing a git book?20:03
pygi bobesponja, what was chap3 once again?20:03
I forgot ..20:03
the_undefined_, yup, he is...20:03
→ bartman joined20:03
← paulp left20:03
bobesponja the_undefined_: still need to sign the contract with the editor20:03
pygi bobesponja, serving and stuff, plus git push?20:03
bobesponja pygi: yeppers20:03
jesselucas i accidentally pulled a branch into our master branch - how could I undo that20:03
the_undefined_ bobesponja: can I ask what publisher? I'm in contact with one that seeks an author for a git book but I don't want to do it20:03
pygi bobesponja, nice, wanna forward? :)20:04
the_undefined_, it's apress20:04
jesselucas do i pull from the master again?20:04
the_undefined_ jesselucas: not done anything else in the meantime? git reset ORIG_HEAD --hard should work20:04
bobesponja pygi: if you still want to write a book about git, you might join in20:04
← loincloth left20:04
the_undefined_ jesselucas: or to be on the save side: 'git log' find the sha1 before the merge, then git reset <sha1> --hard20:05
pygi bobesponja, what do you mean?20:05
bobesponja join me in20:05
the_undefined_ apress is good20:05
or so I think20:05
pygi bobesponja, you mean join with you?20:05
how would that be possible now?20:05
And wouldn't Frank eat you if you suggested that?20:05
→ qrush joined20:05
bobesponja the contract still hasn't been re-written I think20:06
pygi bobesponja, I know xD20:06
bobesponja pygi: or pick another subject :p20:06
→ ahupp|fb joined20:06
pygi bobesponja, no ...20:06
you know why we wanted to go for Git20:06
bobesponja why? :)20:07
this OT, pm me20:07
pygi bobesponja, are you on jabber?20:07
→ ciskje joined20:07
jesselucas the_undefined: Thank you! On the same subject is there a way to clone a repo and get all of it's branches, not just the master?20:08
or to pull all the branches?20:08
bobesponja jesselucas: git branch -r20:08
they are all there20:08
jesselucas bobesponja: But locally I have to create and pull them individually?20:09
bobesponja jesselucas: yep20:10
→ k776 joined20:11
← Sonderblade left20:12
mw|food → mw20:12
← pygi left20:14
→ pygi joined20:15
→ jriga_ joined20:18
← w00t left20:19
← shenie left20:21
← jriga left20:27
→ paulp joined20:30
→ \ask- joined20:33
→ EmilMedve joined20:33
→ naeu_ joined20:37
← naeu left20:38
← Daenyth left20:38
→ Pupeno-M joined20:41
→ robinr joined20:41
← hurikhan|Work left20:42
← Sigma left20:45
ruthd how do i diff between two branches?20:45
nud git diff ^branch1 branch2 ?20:46
ruthd what does the ^ do?20:47
← markelikalderon left20:48
← SjB left20:49
acid2 any git-svn gurus around? I accidently ctrl-c'ed a git svn dcommit while it was in progress, and I'm now kind of stuck and can't dcommit any more20:51
here's what happens. I've checked .git/refs/remotes/git-svn and it points to the same commit as what made it to subversion20:51
when I run git svn dcommit, I get:http://pastie.org/30823820:52
and if I try to do git svn rebase: http://pastie.org/30823920:53
← priidu left20:53
johnw acid2: I use it a lot, let me see if I can help you20:54
→ felipec joined20:54
tritz nud: it's probably only 'git diff branch1 branch2', isn't it?20:54
acid2 johnw: thanks a bunch :)20:54
johnw have you done the rm -fr it suggested?20:54
also, try "git rebase --abort"20:55
tritz ruthd: branch1^ would pick the parent commit of branch120:55
acid2 i have before, but I forgot the what that cause - I'll try20:55
I've done git rebase --abort as well, with no luck20:55
one sec, I'll do them and give you some more output20:55
http://pastie.org/30824320:56
i'm doing the svn rebase, because presumably, it should just say "svn up to date" or whatever it normally says20:56
then I can push my remaining commits out20:56
→ loincloth joined20:56
tritz ruthd: see 'git rev-parse --help', section SPECIFYING REVISIONS20:57
← bdiego left21:01
← joshdavey left21:02
johnw aha21:03
→ shenie joined21:03
johnw well, it seems some of your local commits conflict with what's under version control now21:03
try "git diff trunk" to see what the conflicts may be21:04
stouset → stouset|away21:04
johnw also, do "git log", and see if the log looks right (your commits on top of trunk)21:04
and also "git log -p trunk.."21:04
i think you may need to actually resolve these conflicts to get further21:04
stouset|away → stouset21:05
→ hobodave joined21:07
hobodave hey guys21:07
doener acid2: are you on your branch, or on a detached HEAD (no branch)? git branch tells that21:07
hobodave how can I clone to an existing directory?21:08
→ cilly joined21:08
doener hobodave: git init && git remote add -f origin git://url/repo && git checkout -b master origin/master21:08
hobodave thanks doener21:09
← Pupeno-M left21:09
→ Tv joined21:10
→ Pupeno-M joined21:10
acid2 doener: (no branch)21:12
sorry just got back in, I'll try that stuff now then johnw21:12
there can't be a conflict, I'm the only one who's commit to this svn branch21:12
doener acid2: ok, are you still in the middle of a rebase?21:12
if so, abort that21:12
then check again if you're brack on your branch, and if not, check that branch out21:12
acid2 ok21:13
ok, so I'm back on master now21:13
doener then use "git reflog show <your-branch>" and look for the commit at which you have been before the dcommit you canceled21:13
once you've found it, use "git reset --hard <that-commit>" to move your branch back there21:14
acid2 "eaa1352... master@{9}: commit: Fix bug with searchs that cause a redirect" I think21:14
that's the HEAD of svn atm, if that's what you mean21:14
doener no, I meant the local commit that you had21:14
hobodave ok here's a weird one21:14
doener you apparently had a few local-only commits.21:14
acid2 yea, after that commit21:14
doener basically, you should fine the latest one you made before you did the dcommit21:14
acid2 (2 more)21:14
doener so the reset takes you back to before the dcommit21:15
hobodave I'm using OS X (case insensitive file system) and I have a file named Growler.js, it's causing file not found errors when I deploy to unix. I've tried git mv Growler.js growler.js, but it throws an error in OS X, because those are identical. How can I rename it and commit this change?21:15
← deskin left21:15
doener acid2: then use "git svn fetch" and then "gitk --left-right <your-branch>...<the-svn-branch>"21:15
acid2 doener: http://pastie.org/308257 is from gitref log21:16
← Weaselweb left21:16
doener acid2: hm, and HEAD@{5} and earlier?21:16
acid2 http://pastie.org/308258 ?21:17
doener ok, yeah, it master@{6}, so "git reset --hard master@{6}"21:17
acid2 ok21:17
doener (sorry, HEAD@{5} came from my finger-memory)21:17
acid2 heh21:18
ok, so i'm back at master@{6}21:18
doener ok, now the gitk from above. You should see some "duplicated" commits, those that already got dcommitted but not yet replaced in your branch, right?21:18
acid2 what's <the-svn-branch>?21:18
doener in your case, seems like it is remotes/git-svn21:19
acid2 right21:19
doener (or just git-svn, if that is not ambiguous)21:19
acid2 yea, I see those duplicated commits21:19
doener ok, then check whether the commits are equal on both sides (checking just the latest ones is fine)21:20
→ unstable joined21:20
doener ie. "git diff <latest-duplicated-local-commit> <latest-duplicated-svn-commit>"21:20
acid2 yup, they are21:20
doener ok21:20
acid2 eyeballing the diffs of each commit in gitk shows them the same anyway21:21
doener then try to just "git svn rebase" now. That _should_ work.21:21
acid2 ok21:21
nope, conflicts again :(21:22
doener Possibly, it tells that there are conflicts, and you end up with an empty "git diff", then just tell it to "rebase --skip" that commit (only if the diff is empty!)21:22
acid2 hold on, gitk master points to my top commits still though21:22
that doesn't seem right?21:22
doener yeah, master gets updated only when the rebase is finished21:22
acid2 oh, ok21:22
doener that's why the master reflog was relatively clean, the HEAD reflog gets pretty messy with large rebases :-)21:22
acid2 :)21:23
doener so, what does a plain "git diff" say?21:23
acid2 empty21:23
so i'm safe to skip?21:23
doener ok, and "git status"?21:23
acid2 no changes21:23
doener ok, and the messages on the screen probably tell you that it was trying to apply one of the duplicated commits, right?21:24
(IIRC the commit message is hidden in there somewhere)21:24
→ abbe joined21:24
acid2 yea, I think that's what it says21:24
doener Applying: Improve previewing release in add release21:25
the latest line like that21:25
then it is fine to say --skip, yep21:25
because the commit already exists in the svn upstream21:25
acid2 yup, exactly that :)21:25
so I skip that, and now I get more conflicts, heh, but again an empty git diff21:26
← G_SabinoMullane left21:26
doener ok. That complicated last part was to make sure that it's just rebase complaining about skip-able commits21:26
if you want to, you can abort the rebase and do: git rebase --onto git-svn <last-local-duplicated-commit> master21:26
← unstable left21:27
doener that takes the commits in the range <last-local-duplicated>..master, replays them on top of git-svn and then updates master to point to the result21:27
the range only includes the commits that come after the last duplicated one, so you won't see any of those conflicts21:28
the difference between that and "git svn rebase" is that "git svn rebase" rebases all "local" commits, and can't automatically tell whether some are duplicated21:28
so "git svn rebase" does "git rebase --onto git-svn git-svn master"21:29
acid2 aborting the rebase first?21:29
← Ryback_ left21:29
doener yeah. Either --skip a bunch of times, or --abort and use the other rebase command21:30
acid2 ok, i'll give it a shot...21:30
← d0k left21:30
acid2 heh, git rebase --abort "no rebase in progress"21:30
→ __name__ joined21:30
doener ah, then the second --skip was already the final one21:30
← name left21:30
doener ie. you only had 2 duplicated commits, and you're all set now21:31
→ rdavila joined21:31
doener for just 1 or 2 duplicated commits, the --skip'ing is just a small nuisance. For 200 duplicated commits, it's a big PITA and then the other rebase command comes in handy21:31
acid2 cool, so I can just git svn rebase again from here?21:31
← mtodd__ left21:31
acid2 or dcommit?21:32
doener I'd "git svn rebase" first, just to make sure21:32
→ mtodd joined21:32
acid2 ok21:32
ARGH, conflicts :(21:32
doener hu?21:32
acid2 let me paste what i've done, see if I've slipped up21:32
__name__ → name21:33
← shenie left21:34
← bobmcw left21:36
← FunkeeMonk left21:37
← juanjoc left21:37
← jesselucas left21:37
acid2 ok, lol I tried again, and now doing a dcommit doesn't error or anything21:38
but it just doesn't use my commits21:38
doener paste of what you've done, please? I'm a bit confused right now...21:39
acid2 because my commits are now not in any branch...21:39
yea, you and me both :P21:39
← lucsky left21:39
acid2 http://pastie.org/30826521:39
→ prhlava joined21:39
acid2 checks lost and found21:39
doener acid2: master@{10} was too far back. You wanted master@{7} there. Which is the latest duplicated commit? The "Improve previewing..." one?21:41
← EmilMedve left21:41
doener git log -1 --pretty=format:%s git-svn21:41
what does that say?21:42
acid2 doener: I tried redoing what we had again, this time it was pointing to @{10} it seemed21:42
← robinr left21:42
acid2 "Fix bug with searchs that cause a redirect"21:42
i now have master and git-svn at that commit (great)21:43
→ peeja joined21:43
doener http://git.pastebin.com/m17413b2321:43
→ mulicheng joined21:44
acid2 but my 2 other commits are now lost. giggle does show them with shas, so I'm trying to rebase them back in21:45
but it causes conflicts, oddly21:45
doener nah, the sequence I pasted should already return them to master21:45
← zeeeee left21:45
acid2 yea, but i was obviously too noob, and broke something lol21:45
if I can bring these 2 local commits back onto master, I should be ok doing a dcommit again though, I think21:46
→ jerbear joined21:46
doener Just do those four commands. Master should then have all your commits, none duplicated and "git svn rebase" should report that it is uptodate21:46
no matter what you did in the meantime21:46
→ Zimm joined21:46
acid2 hmm, ok21:47
even now that I've broken master though?21:47
doener the reset _fixes_ master, really...21:47
acid2 lol, sorry about this :/21:47
right21:47
→ jesselucas joined21:47
acid2 ok, so master is at the first duplicated local commit - which is where I want it, right?21:48
← jas4711 left21:48
doener After executing the commands I pasted?21:49
→ jas4711 joined21:49
acid2 i'm just trying to get back to where we were at the start, I think this is after what was reset master@{6}21:49
← dotsintacks left21:50
doener what's wrong with the four commands I pasted? They restore your commits, drop the duplicates and have your master up-to-date. It doesn't matter at all where master is when you start executing those four commands21:50
The abbreviated sha1 hashes I've used there make them work independent of the current state of master21:52
acid2 ok, here we go, back on track then :) http://pastie.org/30827921:53
git diff is not empty though now21:53
← the_undefined_ left21:53
doener ok, "git log -1 --pretty=format:%s git-svn" probably shows something else now, too?21:54
acid2 it shows "Fix bug with searchs that cause a redirect" which is svn head21:54
doener ok, then I guess someone else committed to the svn repo in the meantime and you just got normal conflicts now21:55
is that possible?21:55
where "in the meantime" means "since your last successful 'git svn rebase'"21:56
acid2 mmm, don't think so. the conflicts are between my own local commits21:56
(ie, work i've done)21:56
doener hm... Rebase was trying to apply "Improve previewing...", which is the first commit that you had after the "Fix bug with searchs..." one21:58
acid2 yup21:58
doener "git diff eaa1352 git-svn" -- empty?21:58
→ paltman_ joined21:59
acid2 no, that does show someone elses svn work, but I thought that was before my work21:59
evidently not21:59
→ EmilMedve joined21:59
doener acid2: that was before your work. It's in the commit in svn, but not yet in the local commit you had.21:59
← paltman_ left21:59
← elmex left22:00
doener acid2: do the differences you see "match" the conflicts you're seeing?22:00
acid2 i see22:00
doener acid2: if so, then yes, the conflicts you're seeing are valid and because someone else committed to the svn repo in the meantime22:00
← swhitt left22:00
← rdavila left22:00
doener so just resolve them as "usual", then "git add" and "git rebase --continue"22:01
acid2 yup, they seem like normal conflicts then :) sorry, first time i've actually had to deal with conflicts, so i'm a bit slow here22:02
→ rdavila joined22:02
→ markelikalderon joined22:03
doener no problem, just glad we finally figured out what's up :-)22:03
→ thannoy joined22:03
acid2 i owe you many a bear ;)22:03
ok, all conflicts resolved, now I can just dcommit?22:06
(and that conflict resolving seemed much nicer than subversion, heh)22:06
doener the last rebase was the "manual" one from me, right? Then "git svn rebase" first, just to make sure all svn revisions are really fetched now22:07
(Yes, I'm really really careful when dealing with git-svn)22:07
acid2 yea, and then I resolving each conflict and did rebase --continue22:07
← cilly left22:08
acid2 my god it's working!22:08
\o/22:08
doener :-)22:08
acid2 makes a note to cause much pain on his colleague for causing the conflict >:)22:09
doener for the record, the next time you kill a dcommit (or it just dies for some reason), the steps are: abort any rebase that might be in progress, checkout your branch, use the reflog to find your latest commit (top-most entry you recognize in the branch reflog), reset --hard to that commit, then svn rebase22:10
acid2 rgr that, saved this log to text for good measure22:11
you get my pm?22:11
← EmilMedve left22:12
← acid2 left22:14
← paltman left22:14
← LiamH left22:16
← stouset left22:18
peeja is it possible to filter lines, say through grep, before they hit git-diff, git-add, etc? sort of like a line-item .gitignore?22:18
→ warlock_mza joined22:24
PerlJam peeja: there's usually an interactive version if that's what you're after.22:24
peeja: git add -i # for instance.22:25
peeja PerlJam: yeah, i know about that. I'm looking for something automatic22:25
so I can, say, tag a line with NOGIT or something and have it ignored22:25
PerlJam give me an example with diff.22:26
peeja for lines which I only want in my working tree22:26
→ ciaran joined22:27
PerlJam or even, what's a use case here?22:27
It sounds like you want git to track files, but selectively ignore them for a while.22:28
peeja so, in ruby22:28
http://www.pastie.org/30833322:28
I add line 3, but I don't want to ever commit it22:28
but I want to commit lines around it22:28
PerlJam ah.22:28
That sounds crazy :)22:29
peeja so I want git to filter it out before adding it22:29
I know :)22:29
PerlJam I take it this is for passwords and such?22:29
peeja crazy-awesome though, right ;)22:29
PerlJam (or other sensitive info)22:29
peeja not necessarily22:29
← threeve left22:29
peeja logging, like here, would be a good case too22:29
debugging22:29
wereHamster peeja: if logging_enabled then log a_variable end22:29
peeja or adding comments which you didn't want others to see22:29
wereHamster: right, but the whole point is that I don't want other people to have it22:30
it should never get committed22:30
wereHamster I'm afraid git can't do that on it's own22:30
peeja dang22:30
any idea how I might build something that would?22:30
PerlJam carefully.22:30
peeja I couldn't find a suitable hook22:30
doener a clean filter might do22:30
wereHamster a shell script that replaces the f22:30
peeja PerlJam: heh22:30
doener: what's that?22:31
wereHamster a shell script that strips the NOGIT lines from the files, commits the change and then restores the files22:31
← qrush left22:31
doener peeja: look for "filter" in man gitattributes22:31
Gitbot peeja: the gitattributes manpage can be found at http://git.or.cz/man/gitattributes22:31
doener peeja: I've never tried such a thing though, might kill your kitten22:31
peeja oh, yes! sweet, that might work22:31
or break horribly, but we shall see22:32
→ deskin joined22:32
PerlJam I simultaneously like and dislike people for the same reason--they're always coming up with crazy stuff22:32
doener heh :-)22:33
← a_l_e left22:34
Pieter I just dislike everybody22:34
← jerbear left22:38
→ ciaran_ joined22:41
← doener left22:42
← rmh3093 left22:42
jesselucas when I init my submodules it isn't setting them up to have the master branch checked out at first. It just says * no branch. Not all of them are doing this though. Is there a way have them all default to the master branch?22:42
← name left22:45
→ fynn joined22:46
fynn hia22:46
how can I get a Git log in XML format?22:46
← prhlava left22:47
Pieter you'll have to do that yourself22:47
for example by using the --format=pretty: option22:47
fynn Pieter: :/22:47
Pieter :) usually if you want to export data from git, you use a null-separator22:48
fynn I'm just used to the comfort of 'svn log --xml' :)22:48
thanks anyways though.22:48
← fynn left22:49
→ flyingsymbols joined22:51
flyingsymbols got a quick 'how-to' question for git22:51
i'm using git to manage an installer creation script22:51
where I want to throw in several config files, then run the script to create a customer-specific installer22:52
← ph^ left22:52
flyingsymbols what i'm interested in is whether there's a command that will reset the repository status back to what it was before22:52
Pieter before what?22:53
flyingsymbols (before the config files are added/modified)22:53
Pieter to the repository, or created by a build script or so?22:53
that is, are they committed already?22:53
flyingsymbols no, they won't be committed22:53
and some of the files won't even be tracked22:53
Pieter git clean -f to remove untracked files22:53
git reset --hard to undo changes to tracked files22:53
flyingsymbols git reset --hard without a refspec doesn't remove a commit then?22:54
Pieter no22:54
flyingsymbols excellent22:54
← deskin left22:55
flyingsymbols thanks Pieter22:55
that's exactly it22:55
i'm liking git, but still a lot to wrap my head around22:55
← krh left22:58
→ radarek joined22:59
cbrake → cbrake_away23:00
→ tagx_ joined23:01
tagx_ how do i ignore all files in a directory but not the directory itself?23:01
_eric git doesn't store directories23:02
you need to have at least one file in it23:02
← ciaran left23:02
tagx_ ah, hmm23:02
_eric (an easy one is an empty .gitignore file)23:02
tagx_ that doesn't appear to work23:03
_eric git add directory/to/.gitignore23:04
→ deskin joined23:04
tagx_ but then how do i ignore all other files the directory?23:04
bdrewery can you add the directory to ../.gitignore?23:05
Arrowmaster add * to the .gitignore in that directory23:05
tagx_ hmm, the .gitignore is now ignoring itself23:08
→ quoin joined23:09
← Pupeno-M left23:09
← tagx_ left23:09
← cory_ left23:09
parasti add it, as said before23:10
Pieter hellowwwwwww23:11
parasti ...I don't remember how to unignore parts in irssi :P23:11
← vintrepid left23:11
→ zeeeee joined23:14
→ qrush joined23:14
→ SjB joined23:16
pasky you can always just edit the config23:16
zeeeee hi, i have a git svn repo, and i'd like to generate some patches to send back (only for certain files). how might i do this? for instance, i just want to send patches for file a. my 3 commits are: modified a & b, modified b, modified a.23:16
i looked at git format-patch, but that seems to generate a patch for all changes23:17
→ tagx_ joined23:17
Pieter try git format-patch -123:17
tagx_ basically, i want to add an empty directory to a repository23:17
← _graham_ left23:17
Pieter to generetae the last patch23:17
format-patch takes many of the parameters that git log takes23:18
zeeeee Pieter, but i don't want just the last patch, right?23:18
Pieter, like in my example above23:18
Pieter zeeeee: you can't create patches for specific files, I think23:18
→ Mario_ joined23:18
zeeeee Pieter, what about git diff?23:18
git diff -p23:18
Pieter I dunno, try it out23:18
zeeeee yeah, i guess that's why i wanted to come here to ask first though23:18
← pygi left23:19
Pieter I always create commits that are logical units, so no need to have a patch for only a single file23:19
zeeeee Pieter, so are mine23:19
Pieter or, if I do, I split the commit in two23:19
← floppyears left23:19
zeeeee ok, so it seems that the way to do this is: git diff -p remotes/trunk file1 file2 ...23:21
← ciaran_ left23:22
← willb left23:22
Pieter zeeeee: format-patch takes path specifiers too23:22
zeeeee however, will this spell problems for me further down the line, when they merge in my changes into svn? will a subsequent git svn rebase throw conflicts at me?23:22
_eric what happens if someone does a commit in the middle of a rebase (with merge conflicts)?23:23
← kumbayo left23:23
→ ph^ joined23:24
← tritz left23:24
← naeu_ left23:26
Pieter zeeeee: if you're lucky, no23:26
it'll detect the two changes are the same23:26
drizzd _eric: you have to commit if you want to resolve the merge23:27
Pieter zeeeee: but if they change the commit, or add new commits after it that change the same code, the rebase will complain23:27
← boto left23:27
← tagx_ left23:29
← Yuuhi left23:31
← thannoy left23:31
← loincloth left23:37
→ ciaran joined23:37
← peeja left23:37
← flyingsymbols left23:37
→ flyingsymbols joined23:38
← ciskje left23:38
→ ardrigh joined23:40
← SRabbelier left23:40
← radarek left23:41
← ph^ left23:43
→ tagx_ joined23:44
→ madewokherd joined23:44
tagx_ how do i add an empty directory to a git repository?23:44
Pieter you can't23:45
if you really need to, you can add a file in it to track it, like a .gitignore23:46
deskin tagx_: rumour has it that such a thing can exist, if e.g. it's imported by git-svn23:46
→ Ryback_ joined23:47
← ashleyw left23:48
→ ashleyw joined23:48
→ Fullmoon joined23:49
→ WALoeIII joined23:52
← ashleyw left23:52
→ ashleyw joined23:52
← bentob0x left23:52
→ foca_ joined23:54
Pieter you can also do something like23:54
echo -e "040000 4b825dc642cb6eb9a060e54bf8d69288fbee4904\tnew_dir" | git update-index --index-info23:54
to add an empty dir to git23:54
but I'd say that isn't recommended ;)23:54
← jony left23:55
→ vanjuggler joined23:55
→ jstash joined23:56
→ ben_h joined23:57
← nud left23:58

Logs Search ←Prev date Next date→ Channels Documentation