IRCloggy #git 2008-02-23

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.

2008-02-23

→ krh_ joined00:01
← Yuuhi left00:07
→ dsaxena joined00:08
← krh left00:12
→ dsaxena_ joined00:12
← psst left00:14
→ dsaxena__ joined00:15
← a-priori left00:23
→ ironfroggy joined00:24
← dsaxena left00:27
← dsaxena_ left00:30
← floppyears left00:31
← krh_ left00:31
→ dsaxena_ joined00:32
← philc left00:35
→ dsaxena joined00:35
→ philc joined00:36
→ drizzd_ joined00:39
→ Cerebon joined00:39
← toxx_ left00:40
← ebel left00:46
→ cmarcelo joined00:47
← dsaxena__ left00:47
← dsaxena_ left00:50
← drizzd left00:50
← robinr left00:55
→ lukass joined00:58
← leethal left00:58
→ tarbo joined00:59
→ a-priori joined00:59
← Thumper_ left01:00
← tarbo_ left01:10
← timocratic left01:12
← jmhodges left01:14
→ jmhodges joined01:16
← a-priori left01:16
← jmhodges left01:17
← kelvie_ left01:17
→ kelvie_ joined01:17
→ FunkeeMonk joined01:19
← aroben left01:19
→ madewokherd joined01:24
→ dsaxena_ joined01:27
→ dsaxena__ joined01:31
← glommer left01:32
← KirinDave left01:34
→ rxKaffee joined01:38
← lukass left01:38
cehteh shrug01:38
← dsaxena left01:41
← \ask_ left01:41
→ \ask_ joined01:41
→ Rtfum-aF joined01:42
← drizzd_ left01:42
Rtfum-aF a-atwood jaalto unreal Tali01:42
← Rtfum-aF left01:42
← dsaxena_ left01:44
→ sbohrer joined01:44
← \ask_ left01:44
→ nex3 joined01:47
→ a-priori joined01:47
nex3 Suppose I have a public repository01:48
→ SuttoL joined01:48
nex3 Someone clones it, makes some changes, and pushes them to their own public repo01:48
But they didn't rebase before pushing, so these changes are disorganized01:48
What's the proper way to pull them into my repo?01:49
Suck it up and deal with the disorganization? Rebase it myself and have incompatible commits?01:49
→ aunes joined01:49
← dsaxena__ left01:50
← sgrimm left01:50
→ sgrimm joined02:05
← devogon left02:05
→ demerphq joined02:06
tpope nex3: what do you mean by "incompatible commits"?02:08
← a-priori left02:08
tpope well, I think I see what you mean02:08
they will only be incompatible to people who have used this other guy's repo02:09
probably just the other guy02:09
so, serves him right?02:09
it's nothing he can't recover from02:09
the gentlemanly thing to do would be to suck it up02:10
merge and be done with it02:10
→ axiom joined02:18
axiom → majoh02:18
cehteh robtaylor: there?02:19
← demq left02:20
→ a-priori joined02:21
← philc left02:22
→ philc joined02:23
← a-priori left02:26
← tvachon left02:27
→ mithro joined02:29
← eikonos left02:30
← philc left02:34
→ philc joined02:35
→ Thumper_ joined02:35
→ bugfux joined02:36
bugfux how can I apply a patch and make it form the commit also02:37
so that the commit is from the person who created the patch and not me etc02:37
nex3 bugfux: git am02:37
bugfux oh nex3 whadup man02:37
nex3 Hello :)02:38
bugfux nex3: thats what i was looking for, thanks02:39
nex3 No problem02:39
→ sbeyer joined02:57
→ jbailey joined02:59
→ loglibrarian joined02:59
→ aroben joined03:04
← loglibrarian left03:05
← hadronzoo left03:07
→ Fullmoon joined03:09
← Fullmoon left03:10
← bugfux left03:14
→ \ask_ joined03:24
→ trav_15 joined03:26
← philc left03:27
→ philc joined03:28
← pete__c left03:29
→ pete__c joined03:29
← pedro__ left03:31
Strogg Hrmm.. how do you get git ls-files to show only the new files that have been added?03:35
hrmm.. maybe I'll just parse the output of git status.. hrmm03:35
→ nkallen joined03:38
← kukks left03:41
← madewokherd left03:52
sbeyer Strogg: you could look in the source of tools like qgit, that show new files green, deleted files red, changed files black, for example03:53
← nkallen left03:55
← aroben left03:59
cehteh huh sbeyer :)04:05
Strogg I just wanted to fix up my mess.. I did a git add, and more stuff got added than I wanted in there..04:16
→ BuGless joined04:16
BuGless Say I have, in one repository, two disconnected branches: A--B--C and D--E--F04:17
→ adante_ joined04:18
cehteh Strogg: git reset HEAD -- filename to un-add it04:18
BuGless And I'd like to (re)connect the two branches as: A--B--C--D--E--F04:18
cehteh or use git gui :)04:18
BuGless because D actually is a newer revision of C; how can this be accomplished?04:19
Strogg hrmm nice. I was using git rm --cached04:19
cehteh works too iirc04:19
BuGless: you could try if merge would just do a fast forward .. easy pie then ..04:20
if not rebase ... or even rebase --interactive04:20
.. good night04:21
← RyanTM left04:21
BuGless cehteh: I can try those, of course; just wondering if that won't result in the changes being stacked and duplicated04:21
← philc left04:23
→ philc joined04:24
→ notahat joined04:24
← artmi left04:28
← BuGless left04:28
→ ericflo joined04:30
← ericflo left04:30
← philc left04:32
→ philc joined04:33
→ harinath joined04:33
← adante left04:34
adante_ → adante04:34
← notahat left04:34
→ doener joined04:37
→ tarbo_ joined04:43
← tarbo left04:54
← doener_ left04:55
Strogg wonders why he didn't put some stuff in his .gitignore before..04:57
→ fhobia joined04:57
← sbohrer left05:09
← GyrosGeier left05:12
← philc left05:14
→ philc joined05:15
← fhobia left05:15
→ ReinH_ joined05:21
← philc left05:24
→ philc joined05:25
→ ericflo joined05:27
← ericflo left05:28
← ReinH left05:36
← ReinH_ left05:39
← Ingmar left05:50
→ Ingmar joined05:51
→ cousin_it joined05:54
← adante left05:54
→ tarbo joined05:57
→ ReinH joined06:00
← tarbo_ left06:09
→ devogon joined06:13
← philc left06:24
← jbailey left06:25
→ philc joined06:25
← bobesponja left06:30
→ harinat1 joined06:35
← harinath left06:35
← mithro left06:35
harinat1 → harinath06:35
← trav_15 left06:36
← philc left06:43
→ adante joined06:43
→ janm joined06:43
→ philc joined06:43
← harinath left06:57
→ nkallen joined06:58
→ fhobia joined07:05
→ glguy joined07:10
← philc left07:11
glguy How do I do the equiv of git-gc on old version of git, say the one in debian stable07:11
cwillu glguy, git-repack I think is most of what you want; there's bugs from old versions though07:12
→ philc joined07:12
glguy I installed the latest git in my home directory for things like git-gc07:16
but people pushing will end up using the older git07:16
is that a concern?07:16
or should it be for me :)07:17
→ Eludias joined07:19
cwillu latest git-gc will cause grief for older clients I think, I believe there's an option to turn off the use of offsets (?) in the packs, not sure what it is exactly though07:19
glguy do bare repositories work in git with git-init-db?07:25
rxKaffee isn't git-init-db integrated into regular git-init functionality now? or am I talking crazy business and thinking of something else07:26
glguy I'm stuck with an old version of git on this remote machine I don't control07:26
when not logged into a shell07:26
cwillu init-db == init07:26
but old debian wouldn't have init07:26
thiago_home why don't you just upgrade?07:27
cwillu it's debian. upgrading is evil :p07:27
glguy the person that could do that isn't me and isn't available tonight :)07:27
rxKaffee you already did partial upgrade into ~/... why not do a full upgrade of git into ~/?07:27
perhaps would need to change path in git-shell so users pushing would use "new" tools07:28
← jerbear left07:28
glguy you can't use non-standard paths if sshd isn't set up to allow that07:28
when pushing07:28
(that I know of)07:28
I think you need PermitUserEnvironment or something07:28
thiago_home no, it isn't evil07:29
grab git from git, type: make install07:29
that's it, no need for root permissions07:29
glguy I did that, it's installed in my home directory07:29
but I don't know how to use that version when pushing for all other users by default07:29
ironfroggy without options its implied that $GIT_WORK_TREE=$GIT_DIR/.. right?07:29
thiago_home same here07:29
glguy as not root07:29
cwillu he'd need to do that for the remote user's home I'd presume?07:29
glguy without some exploit I don't have07:30
thiago_home glguy: if you don't have root, you do it for you only.07:30
cwillu glguy, have they patched vmsplice on that box yet? :p07:30
glguy thiago_home: right :)07:30
that's why I'm bothering you poor fellows with trying to work around this07:30
thiago_home work around what?07:30
glguy things like using bare repos in old versions of git07:31
thiago_home use a newer version of git07:31
glguy that's what I ended up doing07:31
thiago_home problem solved07:31
glguy I was just concerned that the rest of the users07:31
using old git07:31
might have had issues07:32
cwillu abbott and costello comes to mind for some reason :p07:32
glguy so I asked07:32
thiago_home the format should be compatible07:32
rxKaffee glguy: they on same physical server as your "new" git binarys?07:32
thiago_home incompatibilities are listed in the man pages07:32
→ hadronzoo joined07:32
glguy rxKaffee: do you know how to tell a "git-push" launched on my local machine to use the git-* stuff in a custom bin directory on the remote machine?07:33
without .ssh/environment07:33
cwillu glguy, could try going via git-daemon07:33
rxKaffee glguy: do you control the account on the remote machine?07:33
← careo left07:33
glguy rxKaffee: I control my account07:33
thiago_home git push --receive-pack=/path/to/remote/receive-pack ..07:33
glguy ok, cool07:33
rxKaffee glguy: do the other people pushing, push into your account?07:33
glguy rxKaffee: no07:33
rxKaffee then I don't understand how their git version affects you07:34
glguy rxKaffee: If they are using old git to accses a publicish repo on the server07:34
I'd like to avoid them having problems07:34
with the repo I'd created07:34
using new git07:34
that's about it07:34
rxKaffee oooooh, shared reposwithout gitosis /me dosn't trust g+w :P07:35
thiago_home glguy: if they're using an old, incompatible version, they'll have to upgrade as well.07:35
glguy: if they come here and ask for help, that's what we'll tell them.07:36
glguy thiago_home: I was just worried that the old version of git-recieve-pack on the server07:36
that they'd use hwen sshing in07:36
would be an issue07:36
sounds like everything is fine07:36
we can all move on :-D07:36
thiago_home there should be no incompatibilities, unless you explicitly enable them07:36
← philc left07:38
→ philc joined07:39
← janm left07:48
→ HG` joined07:52
nex3 Is there an easy way to get a list of commits in one repo that aren't in another, in a git-log sort of format?07:53
thiago_home no07:53
you have to do that per branch07:53
nex3 "per branch"?07:54
→ dsaxena joined07:55
nex3 Do you mean I can get such a comparison between branches?07:55
Because that would be fine07:55
thiago_home yes, that's what I mean07:56
nex3 How would I do that?07:56
ReinH eh?07:56
the branch can be a remote07:56
git log ..origin/master (for instance)07:57
which is probably close to what you want07:57
nex3 Okay, thanks07:57
Ilari nex3: 'git log A..origin/A', 'git log origin/A..A', 'git log A...origin/A'... First is in remote but not local, second is the reverse, and third is only in one.07:57
nex3 Okay, great, thanks a lot07:58
→ janm joined07:58
← chrizz left07:59
ReinH Ilari: didn't know about ... :)07:59
→ pitra joined07:59
← pitra left08:00
Ilari ReinH: ... is the symmetric difference operator.08:03
→ robinr joined08:03
ReinH Ilari: I'm familiar with it in the context of Ranges, where it's exclusive08:04
ie 1...4 => 1,2,308:04
but that's it08:04
good to know :D08:04
thiago_home remember that a branch isn't a single commit, but a list of them08:05
so the .. and ... operators are differences08:05
ReinH right08:08
well, a branch is a ref to a head :p08:08
the head's graph forms the commit history08:08
</stickler>08:08
→ chrizz joined08:10
→ tcoppi_ joined08:17
← tcoppi left08:18
tcoppi_ → tcoppi08:20
← nkallen left08:24
→ z3om joined08:25
→ pietia joined08:28
← dsaxena left08:33
→ pergesu joined08:37
← zeom left08:42
z3om → zeom08:42
← cmarcelo left08:44
← Sho_ left08:44
wagle_ → wagle08:46
← bdowning_ left08:50
→ bdowning joined08:51
→ mahmoud2 joined08:55
← philc left09:14
→ adulteratedjedi joined09:14
→ philc joined09:14
← pietia left09:17
→ pietia joined09:18
← glguy left09:33
← janm left09:36
← rxKaffee left09:44
→ sverrej joined09:47
← da5id left09:49
← \ask_ left09:50
→ \ask_ joined09:54
→ Arme[N] joined09:54
→ leethal joined09:55
← \ask_ left10:00
→ d0k joined10:01
→ trochala joined10:08
→ lukass joined10:08
→ da5id joined10:14
← trochala left10:16
→ harinath joined10:29
→ dv_ joined10:31
dv_ hello10:31
I'm trying to use the git-diff-files functionality to transmit some changes to another pc10:32
I did git-diff-files -p >my_patch10:32
but now I wonder what to use for applying this patch on the other pc. patch -p 1 <./my_patch ?10:32
Mikachu you probably just want 'git diff'10:32
and yeah, patch -p1 is the default patcehs git outputs10:33
dv_ (invoked in the project root dir)10:33
whats the difference between git diff and git-diff-files -p ?10:33
Mikachu i don't know, git diff-files is more of an internal command though10:33
← pietia left10:37
Ilari dv_: Maybe use 'git format-patch'? Or if the another PC has copy of the repository, 'git bundle'? And git produced diffs should be appiled using 'git apply', as it understands constructs like binary patching and file moves...10:40
dv_: But patch should understand them if they don't use any advanced features of git diff format.10:41
→ janm joined10:41
dv_ Ilari, thanks, apply and bundle were unknown to me10:41
but the diff worked here nicely, because no advanced stuff was required10:42
but in future I'll use these two10:42
← aunes left10:43
Ilari dv_: Bundle actually copies the commits themselves into the file it makes. Format-patch just exports description of the changes...10:43
→ aunes joined10:43
dv_ Ilari, so suppose I did some commits on my laptop,10:43
and in addition made changes to some files, added some files etc. but did NOT commit these changes10:44
Bundle copies all this in a way for me to merge these local commits and changes with the repo I have on my pc?10:44
Mikachu no, you have to commit first10:44
you can uncommit it after if you want10:44
Ilari dv_: And if the desktop runs SSH and has git installed, pushing over SSH could be much nicer.10:46
dv_ what, git push ssh://login@ip is enough?10:48
I've heard about this before, but I dont understand how this is supposed to work10:49
.. wait.10:49
git push ssh://login@ip/path/to/git/repo ?10:49
Ilari dv_: As test, what does 'ssh login@ip git --version' print. It should only say 'git version <something>'...10:49
→ chris2 joined10:50
Ilari dv_: Asking for password should be OK, but pretty much nothing else is...10:52
← fhobia left10:55
Ilari dv_: And if you have repository you often push to or fetch from, you should register a remote for it (git remote).10:56
jengelh well, git push ssh://xxx branchname10:58
(as I don't want to push all)10:58
← aunes left10:58
Ilari And watch out if the remote repo has working copy associated. Pushing to branch that is checked out will screw the repo...10:59
dv_ k11:01
Ilari It is possible to push to references that arent branches. Usually if one needs to push to repo with working tree (called non-bare), the stuff is pushed somewhere under refs/remotes11:02
→ philc_ joined11:13
← philc left11:15
metze → metze_away11:24
→ d0k_ joined11:26
→ psst joined11:32
← d0k left11:35
→ Yuuhi joined11:35
→ harinat1 joined11:37
← da5id left11:37
→ priidu_ joined11:45
→ da5id joined11:50
← HG` left11:51
← harinath left11:55
← devogon left11:55
→ somnium_ joined12:03
← philc_ left12:04
→ philc joined12:04
← leethal left12:04
→ dv____ joined12:04
← dsop left12:06
← sverrej left12:18
harinat1 → harinath12:20
← dv_ left12:21
cwillu where can I find docs on ignoring files in git?12:25
→ leethal joined12:26
context cwillu: man gitignore12:28
cwillu ya, just found it; didn't google very well :p12:28
thanks12:29
← kanru left12:31
→ pdherbemont_ joined12:33
→ Sho_ joined12:34
→ bdiego joined12:41
← philc left12:58
→ philc joined12:59
→ destruct_ joined13:10
→ tjafk1 joined13:13
← destruct left13:20
← cwillu left13:21
halorgium can i filter the branches/tags pulled in by git-svn?13:23
i'm importing the capistrano svn repo, but the branches dir has got the rails ones in there too :(13:24
→ kanru joined13:30
destruct_ → destruct13:33
→ HG` joined13:39
→ pietia joined13:41
← bdiego left13:45
← panagos left13:48
→ drizzd joined13:49
→ kukks joined13:55
→ ruphy-lapdog joined14:01
→ sverrej joined14:06
← dooglus left14:11
← ruphy left14:18
→ rofrol joined14:19
← leethal left14:21
← da5id left14:26
← pdherbemont_ left14:28
ruphy-lapdog → ruphy14:28
→ drizzd_ joined14:35
robtaylor cehteh: am now14:36
→ erikvee1 joined14:39
← drizzd left14:45
→ da5id joined14:46
→ madewokherd joined14:50
← erikvee1 left15:17
← SuttoL left15:19
← philc left15:23
→ philc joined15:23
← rofrol left15:25
→ harinat1 joined15:28
← harinath left15:29
harinat1 → harinath15:29
psst how good is git at versioning binary files?15:30
Ilari psst: Depends on nature of those binary files.15:32
psst: The files should be at most moderately sized, not compressed and not encrypted.15:32
→ lu_zero joined15:33
→ pikpus joined15:37
← pikpus left15:37
Ilari psst: In general, the file data should change only relatively little between versions. Compressed/encrypted files tend to change completely each time.15:39
psst makes sense15:40
thank you15:41
→ a-priori joined15:41
Ilari psst: Internally git makes no distiction for text and binary files, but it is still aimed for source code.15:41
psst So it would work better on C source than on LaTeX documents, even though both are ascii?15:42
jengelh latex is also text :p15:42
psst but it isn't "source code"15:42
jengelh just use git, ok? :)15:42
Ilari psst: Isn't latex turing complete? And in sense, it is source code..15:43
jengelh like html15:44
psst (the binaries I mentioned earlier are PDFs, PNGs, and some NeoOffice/MSOffice/iWorks docs)15:44
Ilari psst: at least PDFs and PNGs are compresed, altough it would be possible to decompress the PNGs.15:45
psst Ilari: ok - true. I wondered for a moment because I just watched Linus on Google Video saying how git could track a function being moved around files15:45
s/around/between/15:46
← a-priori left15:46
← sbeyer left15:47
→ trav_15 joined15:48
somnium_ is someone here using git-cvsimport often?:15:50
somnium_ → dsop15:50
dsop the problem is that i got duplicated commits when reimporting a commit i made from my git repository15:51
i do a initial git-cvsimport than changes on master, after that i commit the change back into cvs using git-cvsexportcommit and then doing git-cvsimport again15:52
→ sean joined15:52
dsop and i then i have the same commit with two different sha1 on my master, the one i did myself and the one that was reimported from the cvs.15:53
sean is there an article somewhere that talks about branching/management methods with git?15:53
dsop i use the -k switch to avoid keywords..15:53
→ sbeyer joined15:56
← FunkeeMonk left15:56
← trav_15 left16:03
← wildfire left16:04
← philc left16:07
Ilari dsop: You should probably rebase your topic branches to be up to date. Rebasing should notice the duplicate and skip the commit.16:07
← lukass left16:08
→ philc joined16:08
← Pistahh left16:08
→ Pistahh joined16:10
dsop Ilari: okay thanks, i'll try that16:13
← cousin_it left16:13
→ cousin_it joined16:15
→ toxx_ joined16:24
→ Morphous joined16:25
→ leethal joined16:28
→ artmi joined16:30
→ dvorak joined16:30
dvorak so I just started using git, and I used git-svn to import an existing project from svk. looking back, I now have git-svn-id in all of my old commit messages. any way to remove that?16:31
← lamont left16:32
vmiklos you don't plan to use git svn dcommit in the future?16:32
dvorak no16:32
vmiklos git filter-branch16:32
dvorak and I want to publish the repo, and I don't want the old urls in there, since it's for a private work repo16:32
← Amorphous left16:33
Morphous → Amorphous16:33
← ruphy left16:40
vmiklos git filter-branch --commit-filter 'sed "/^git-svn-id:/d" |git commit-tree "$@"' should do the trick16:41
← sverrej left16:42
→ ruphy joined16:43
dvorak hrm, that gives me the usage info16:44
ah, it needs a branch name to work on16:44
Pieter give it head16:45
dvorak I told it 'master'16:45
sbeyer hihi: file ....pack -> Quake I or II world or extension16:45
dvorak which seems to have worked16:45
vmiklos maybe you have a shightly older filter-branch :)16:45
→ lamont joined16:46
dvorak 1.5.3.716:46
vmiklos yup. i have 1.5.4. maybe that's because it was not necessary to specify HEAD here16:46
← priidu_ left16:47
→ priidu joined16:49
← orospakr left16:50
← priidu left16:50
→ orospakr joined16:50
sbeyer is there a way to do something like merge.no-commit = true in the .gitconfig? I mean, that a merge always behaves like having --no-commit?16:57
Ilari sbeyer: At least you can alias some command to behave like merge --no-commit...16:58
→ a-priori joined16:59
← philc left17:00
→ philc joined17:01
→ ebel joined17:02
→ Simoyd joined17:09
Simoyd where do I obtain git-svn?17:09
I can't find it for the life of me17:09
Ilari Simoyd: It comes with git tarball... In package managers, it is often 'git-svn'.17:10
Simoyd oh darn... I'm running msysgit and it doesn't appear to be included with that17:10
← leethal left17:11
→ StatMan joined17:14
Simoyd what does "hack on Git" mean (I'm not sure if more context if required for that question or not)17:14
StatMan → seba__17:14
seba__ hi17:15
i used hg-fast-export.sh to import from mercurial, but now I can't see the files17:15
and when I git status it says all files were deleted17:15
vmiklos git checkout -f? :)17:16
seba__ thx man :)17:17
i thopught the tool would do it :)17:17
btw hg-fast-export is indeed the fastest tool i found around for hg -> git17:18
Ilari seba__: It is, because it uses git-fast-import for importing the repo, and it is fastest method for importing...17:18
cehteh hum i have a problem with submodules17:21
← moh left17:21
cehteh git clone git://git.pipapo.org/webgit webgit_test ...17:22
remote: error: remote: unable to find 2026669cd...17:22
202666 is the gitlink for the submodule17:22
← PerlJam left17:25
→ PerlPilot joined17:25
robtaylor cehteh: hey, you pinged before?17:25
cehteh yeah ... but i have to fix this problem now first17:26
robtaylor cehteh: kk17:26
robtaylor wonders if any git guys are at FOSDEM17:26
cehteh robtaylor: wondering if you have an idea how to hook in custom diff printing17:26
$ git ls-tree 02579f592ff6724692f301f1eb1801b6642cadc8 | grep commit17:29
160000 commit 2026669cd415653cd221a7e1aa1a0162d2baa596 git17:29
.. thats my submodule commit .. wtf does it call it 'misssing object' its obliviously a missing object since it is not a commit in this repository but a commit from another one17:30
robtaylor cehteh: sure, use GIT_EXTERNAL_DIFF17:30
cehteh robtaylor: libgit without forking17:30
robtaylor cehteh: oh, in that case, no ..17:31
doener robtaylor: IIRC someone gives a git talk for debian devs or sth. like that17:31
cehteh heh ... quote:17:32
[18:28] <hermanr> Grievances from FOSDEM2008: It's too much going on at the same time, the ventilation is inadequate and the wireless is struggleing.17:32
robtaylor doener: yeah, 'packaging with git (or $DSCM)'17:32
robtaylor has wireless =)17:32
cehteh will be at LAC next week in colonge17:32
doener will be at his desk, cleaning up &§$§$%§$ PHP code17:33
cehteh heh17:33
robtaylor doener: :(17:33
cehteh any ideas about my submodule problem .. i really dont know whats going on there17:34
robtaylor cehteh: submodules scare me17:34
cehteh: they're heavily broken at the moment, imho17:35
cehteh looks like that17:35
Simoyd $ git-svn17:37
Can't locate SVN/Core.pm in @INC (@INC contains: /usr/lib/perl5/site_perl/5.8 /u17:37
sr/lib/perl5/5.8/cygwin /usr/lib/perl5/5.8 /usr/lib/perl5/site_perl/5.8/cygwin /17:37
usr/lib/perl5/vendor_perl/5.8/cygwin /usr/lib/perl5/vendor_perl/5.8 .) at /usr/b17:37
in/git-svn line 29.17:37
what am I doing wrong?17:37
doener you need the svn perl bindings17:37
← sean left17:37
Illocution is there a way to continue a git-svn clone operation that halted halfway through?17:38
doener git svn fetch -- that should just continue fetching17:38
clone is init + fetch17:38
Illocution hmm it's throwing an error17:39
never mind, got it, duped line in .git/config for some reason17:39
hooray, git-svn rocks my world17:39
Simoyd ok perl bindings installed for svn. now I'm getting this (sorry for spamming):17:40
$ git-svn17:40
24 [main] perl 1632 C:\cygwin\bin\perl.exe: *** fatal error - unable to rem17:40
ap C:\cygwin\lib\perl5\site_perl\5.8\cygwin\auto\SVN\_Core\_Core.dll to same add17:40
ress as parent(0xB60000) != 0xCE000017:40
10 [main] perl 2296 fork: child 1632 - died waiting for dll loading, errno17:40
1117:40
5156893 [main] perl 4172 C:\cygwin\bin\perl.exe: *** fatal error - unable to rem17:40
ap C:\cygwin\bin\cygdb-4.2.dll to same address as parent(0xC10000) != 0xED000017:40
any ideas?17:40
→ hacim joined17:41
doener Simoyd: there's http://git.pastebin.com/17:41
that would probably even keep it readable ;-)17:41
Simoyd right17:41
my bad17:41
tbh that's how it's showing up in the window anyways17:41
hacim i want to do a git-merge --squash on the changes that I have in my current master, i've only done git-merge --squash when I was in a different branch17:42
Simoyd http://git.pastebin.com/m1cd78740 < any ideas?17:42
hacim is there a way to squash commit in master, or do I need to create a branch just to do that?17:43
doener Simoyd: http://m2ward.blogspot.com/2006/10/ruby-on-cygwin-error-unable-to-remap_04.html17:44
← Eludias left17:44
doener google, first hit here ;-)17:44
hacim: the last few commits?17:45
hacim: git reset HEAD~5; git commit -- that should create a commit that's the result of squashing the last 5 (or 4?) commits17:46
Simoyd apparanylt my google skills suck :(17:46
doener replacing those commits, of course17:46
Simoyd: :-)17:46
← cousin_it left17:46
→ bobesponja joined17:47
doener Simoyd: I used: cygwin "unable to remap", and because most hits were from 2004, I then added "2008" to it.17:47
→ cousin_it joined17:47
doener Simoyd: And that was pretty successful :-)17:47
Simoyd =D17:48
thanks it appears to have worked17:48
robtaylor cehteh: for doing your own builtin diff, you'd have to hack up run_diff_cmd in diff.c17:48
cehteh robtaylor: yes well i could use the existing diff, but i thinking about adding a print_fn to diff_options17:49
it not the diff whats bad, its the way that its hardcoded output is17:50
← sbeyer left17:55
Simoyd ok I give up again...18:00
Internal error: REPORT request failed on '/svnroot/octgn/!svn/vcc/default': Can't find a temporary directory: Error string not specified yet at /usr/bin/git-svn line 381318:00
that happens when I try to svn-clone18:01
err18:01
git-svn clone18:01
hacim doener: thanks18:02
doener: i went too far back... how do I move forward?18:03
Ilari hacim: See 'git log -g', and then git reset to suitable entry and retry?18:07
doener Simoyd: that sounds like an error on the svn repo side18:10
Simoyd fuck18:10
I mean darn18:10
doener too late, it's captured in the logs now ;-)18:10
Simoyd hides18:10
→ knyttr joined18:10
hacim Ilari: heh, problem is I dont understand what git log -g is telling me to be able to know which one to switch to18:12
knyttr Hello, I've set up new git repository by git-init. I added new files by 'git add files'. Then I did git-commit and set the initial commit. But after making first changes and doing git-commit again, it says: error: Couldn't set refs/heads/master, fatal: cannot update HEAD ref. Google says nothing about it..18:12
doener hacim: it tells you the recent commits that HEAD was pointing to18:12
← chris2 left18:13
Ilari hacim: HEAD@{0} is probably the commit, HEAD@{1} is the reset?18:13
← HG` left18:13
hacim got it, thanks18:13
→ HG` joined18:13
drizzd_ Tv: I believe you told me a few days ago to point PATH to ~/src/git in order to test git. However, git.c changes PATH to whatever prefix was configured to in the build. So it's important to also set GIT_EXEC_DIR.18:15
Just thought you'd like to know. I've spent some time until I finally figured that out.18:15
← jkp left18:20
Tv drizzd_: hey it was just a quick suggestion18:20
pdherbemont Tv: any chance to have a look at my gitosis patches?18:21
Tv pdherbemont: sort of busy for a while18:21
drizzd_ I'm sorry if that sounded like an accusation. I didn't mean it like that.18:21
Tv pdherbemont: nothing gets lost but i just don't have the brain power available in the short term future18:21
pdherbemont Tv: oh. ok. Do you have any ETA for your return?18:22
Tv pdherbemont: i'm supposed to deliver one prototype by end of next week18:22
← Simoyd left18:22
Tv pdherbemont: hopefully things will ease off after that18:22
pdherbemont oki, cool. So no need to prepare a fork and such ;)18:23
jengelh gitk output is funny with --no-merges :)18:26
→ kevwil joined18:27
← hadronzoo left18:31
→ hadronzoo joined18:31
← artmi left18:37
← a-priori left18:38
→ artmi joined18:42
Tv pdherbemont: replied18:43
← pietia left18:43
→ nkallen joined18:48
← hadronzoo left18:52
→ hadronzoo joined18:53
← hadronzoo left18:53
← philc left18:58
← captnmark left18:59
→ philc joined18:59
→ captnmark joined19:00
← Helios left19:01
→ Helios_ joined19:01
← kevwil left19:01
← knyttr left19:03
→ loops_ joined19:07
← loops left19:08
← loops_ left19:08
→ loops joined19:08
→ a-priori joined19:09
→ kahmalo joined19:12
kahmalo Is the structure of .git/rr-cache documented somewhere? Not in <http://www.kernel.org/pub/software/scm/git/docs/repository-layout.html>.19:12
Also, is there some convenient way to get a resolved conflict recorded there when I've already committed the merge but not pushed it?19:14
I resorted to undoing the merge with git reset --hard, redoing it, checking out the conflicted files from the previous resolution, and committing again.19:15
← a-priori left19:16
Mikachu resolved commits will be recorded automatically when rerere is enabled19:16
kahmalo Yes but it was not yet enabled when I committed.19:16
pdherbemont Tv: thanks, me too :)19:16
kahmalo hmh, perhaps git commit --amend would have sufficed?19:17
Mikachu where would it get the pre-resolution image from then?19:18
→ StatMan joined19:18
kahmalo I don't know. Where does it get it now?19:18
← StatMan left19:19
kahmalo In fact git commit --amend does not suffice.19:20
It created an empty .git/rr-cache/MERGE_RR file though, what is the purpose of this?19:20
→ hadronzoo joined19:21
← hadronzoo left19:22
← philc left19:24
→ philc joined19:24
→ GyrosGeier joined19:28
← loops left19:29
→ docelic joined19:29
→ loops joined19:29
kahmalo With git-new-workdir and git gc, I got plenty of warnings about missing commits in the reflog of HEAD. Because .git/logs/HEAD is in each working directory and gc does not see it.19:30
So now I created ~/src/elinks.git/logs/HEAD as a directory and made ~/src/elinks-0.13/.git/logs/HEAD a symlink to ../../../elinks.git/logs/HEAD/elinks-0.13. Does that seem a good idea?19:31
Likewise for the other workdirs.19:31
← seba__ left19:31
vmiklos wonders why having multiple workdirs is a good idea19:32
kahmalo To reduce compilation times19:32
and I am using ccache as well19:32
vmiklos ccache does the same here alone :)19:32
kahmalo also autoconf times19:33
→ pflanze joined19:33
vmiklos autoconf itself isn't that slow, other autotools are :)19:33
kahmalo elinks is using autoconf and aclocal but not the rest of automake.19:34
pflanze Hello. Is one of the gitweb authors here? I'm considering implementing some things.19:34
jengelh you can even implement something when they are not here ;-)19:36
vmiklos or just ask on the mailing list19:38
→ nex3` joined19:41
← dash__ left19:42
→ tongueroo joined19:42
→ dsaxena joined19:43
← tongueroo left19:44
← docelic left19:45
← arun left19:49
← lamont left19:55
→ lamont joined19:56
← nex3 left19:56
→ arun joined19:58
→ priidu_ joined19:59
→ larus joined20:01
← larus left20:02
← harinath left20:04
→ mithro joined20:05
→ pietia joined20:07
← nkallen left20:18
← dv____ left20:21
← pietia left20:25
→ \ask_ joined20:33
← mithro left20:34
→ gebi_ joined20:40
→ SuttoL joined20:43
← loops left20:46
→ technoweeni joined20:48
technoweeni → technoweenie20:49
← gebi left20:52
← \ask_ left20:55
→ nkallen joined20:56
→ moh joined20:56
→ pietia joined20:59
→ loops joined21:00
← philc left21:01
← SuttoL left21:01
→ philc joined21:02
nex3` → nex321:03
→ a-priori joined21:08
← BartVB left21:11
← gebi_ left21:17
← a-priori left21:20
→ a-priori joined21:22
← Eridius left21:23
→ gebi joined21:37
← philc left21:43
→ philc joined21:44
→ Eridius joined21:48
→ tarbo_ joined21:50
Helios_ → Helios21:54
→ pedro__ joined21:55
← dsaxena left21:58
← tarbo left22:01
← a-priori left22:02
→ a-priori joined22:05
← pergesu left22:06
→ aunes joined22:08
→ baffled joined22:08
→ ebel_ joined22:09
baffled Could someone explain exactly I need to provide to get the post-receive-email scrip working?22:09
So far I've added hooks.mailinglist and hooks.prefix but now I'm getting some about a description not being found.22:11
← csc` left22:13
← ebel left22:15
ebel_ → ebel22:15
→ csc` joined22:17
drizzd_ baffled: try adding a file named 'description' to your repository root directory.22:24
baffled drizzd, should it have anything in it or just a zero length file?22:25
drizzd_ baffled: whatever you put in there will be prefixed to the generated email subject22:26
It's supposed to be a general project description. Maybe just a name would be good.22:26
Only the first line will be used, so if you want a more detailed description (for gitweb, for example), just use the remaining lines like you would for a commit message.22:26
← ruphy left22:27
baffled Okay I'll try that thank you.22:27
→ ruphy joined22:29
→ hadronzoo joined22:30
← HG` left22:31
← ruphy left22:34
← d0k_ left22:34
→ ruphy joined22:34
→ drizzd joined22:35
baffled Send it to my home address.22:36
→ jeffwheeler joined22:37
jeffwheeler If I use git rm filename, is there anyway to undo that by bringing the file back from the previous commit? Currently, I've been committing the deletion and then just reverting it, but I suspect there's a better way.22:38
← baffled left22:38
kahmalo you could use git checkout HEAD filename I think22:39
Eridius you'd probably have to git reset the filename first, no?22:39
jeffwheeler Hmm, I'm not sure; let me try.22:39
← tmokros left22:40
Mikachu i would recommend git checkout HEAD^ -- filename22:40
jeffwheeler (what's funny is that I just deleted the same directory three times in a row, and then each time had to go through the same process)22:40
← ruphy left22:40
Mikachu jeffwheeler: maybe you wanted git rm --cached filename ?22:40
to remove from git but leave the files22:40
jeffwheeler Mikachu: no I actually _did_ want to delete the directory, just not the one I did ;)22:41
Mikachu does the famous read-only-the-last-three-lines maneuver22:41
Mikachu oh i see, there were only 3 lines to read :)22:41
jeffwheeler checkout from HEAD worked well22:42
thanks, everybody :)22:42
Mikachu so you didn't commit yet then22:42
jeffwheeler I had been committing just to get rid of them, but I hadn't committed on accident.22:42
I just kept deleting the wrong files and wanting to bring them back .... silly me.22:43
doener Mikachu: he noticed his error before the commit, but then comitted to be able to revert22:43
← philc left22:43
Mikachu i was thrown off by "from the previous commit"22:44
ah, now i see22:44
literally 'git revert'22:44
→ philc joined22:44
doener yeah22:44
guess I implicitly parsed "previous" as "most recent" there22:44
jeffwheeler oh, ha; git is intelligent in recognizing moved files now22:45
Mikachu since a while i think :)22:45
thiago_home wants it to detect copied files too22:45
Mikachu does that since a while too22:45
← drizzd_ left22:45
jeffwheeler oh; I don't do much beyond simple commits and pushes22:45
← a-priori left22:46
Mikachu maybe you want to git config --global diff.renames copy22:46
it will make diff -C the default iiuc22:46
thiago_home Mikachu: not very well22:46
Mikachu it depends on how much you change the file in the same commit of course22:47
thiago_home no change22:47
doener thiago_home: you might need to pass -C -C if you didn't modify the source in the same commit22:47
by default, git only considers changed files, because that's a lot cheaper22:47
thiago_home interesting...22:47
thiago_home thought that intact files would be easier to find because they have the same SHA-122:48
Mikachu that sounds really illogical22:48
jeffwheeler I had the same thoughts about renamed files -- they'd be easy to recognize because of the SHA-1.22:48
doener well, there's more than just the SHA-1, it also tries to detect copies that _did_ change the file22:48
you get something like "copy, similarity 95%"22:49
Mikachu i would think the exactly similar files would be trivialer to detect22:49
doener and git can then also give you a diff of what you changed in the copy22:49
jeffwheeler funky; anyways, git now tells me I copied a file that no longer exists (a blank one, __init__.py) to make this new __init__.py in another place; hehe22:50
thiago_home always forgets the --follow option22:50
doener maybe unchanged copy detection might be a welcome addition, never checked the code for that22:50
cehteh mhm .. now i borked my submodule problem repo completely :P22:50
← technoweenie left22:55
← drizzd left23:10
→ adante_ joined23:15
→ sbeyer joined23:22
← mahmoud2 left23:23
→ dsaxena joined23:23
← philc left23:24
→ philc joined23:24
← nkallen left23:27
← adante left23:31
adante_ → adante23:31
→ nkallen joined23:36
→ ebel_ joined23:39
→ timocratic_ joined23:46
→ leethal joined23:47
→ Psiclonik joined23:49
← priidu_ left23:49
→ jerbear joined23:52
← ebel left23:56

Logs Search ←Prev date Next date→ Channels Documentation