IRCloggy #git 2010-02-26

Logs Search ←Prev date Next date→ Channels Documentation

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

2010-02-26

→ Theravadan joined00:01
rtyler spearce: you has a javascript bug :(00:01
spearce: when clicking "side-by-side" diffs in a review: Class$u2c: Array index 381 out of range00:01
→ bcardarella joined00:01
spearce frell00:02
another?00:02
→ hipe joined00:02
spearce those are nearly impossible to track down without the exact commit/file to debug00:02
← hipe__ left00:02
→ maattd joined00:03
rtyler spearce: GWT pwnt?00:07
→ christophsturm joined00:08
→ hipe_ joined00:08
← hipe left00:08
→ cloudhead joined00:09
→ Paraselene__ joined00:09
← christophsturm left00:12
kinection anybody here know how to get a <redirector> in ant build.xml to not only createemptyfiles but to create directories?00:13
→ nessundorma_ joined00:13
spearce maybe try #ant, or #abuseofxmlandotherthingsthatarebracketlike ?00:13
b636bc7ca johnw / cbreak: Can I not do this when the upstream is a bare repo?00:13
cbreak you should be able to reset in bares00:14
never tried00:14
rebase -i won't work there00:14
but worst case, you can just rebase locally and then force push00:14
← nessundorma_ left00:14
cbreak (that will remove commits on the repository you push to, all commits that were made between you pulling before the change and pushing)00:14
b636bc7ca There've been no commits in between - when he pushed this, I shut the gitd down until I could fix it00:15
So I cloned the bare repo, did my rebase -i, and everything looks good. But when I say "git push origin", it tells me "Everything up-to-date"00:16
← hipe_ left00:17
← nessundorma left00:17
→ hipe joined00:17
→ saccade_ joined00:18
b636bc7ca ah, forgot to "git rebase --continue"00:18
→ jmcantrell joined00:20
→ hipe_ joined00:20
floppyears hi guys00:20
← hipe left00:20
floppyears I'm using git add -i to stage only parts of a file00:20
how can I specify parts of a chunk to stage?00:20
kinection ah crap, i totally asked my question in wrong channel, sorry.00:21
floppyears never mind, git answered my question00:22
cbreak floppyears: there should be a mode where you can edit the hunk00:22
floppyears cbreak: yeah, I used the inline help and it answered my question :)00:23
← hipe_ left00:23
→ hipe_ joined00:24
b636bc7ca cbreak / johnw: Perfect, that solved everything. All my guys have done a reset --hard and we appear to be solid. Thank you both!00:25
cbreak great :)00:26
→ cehteh joined00:31
→ jrmuizel joined00:32
rtyler spearce: could I bug you to come over into #hudson?00:37
→ scarabx joined00:38
→ melter_ joined00:39
→ corto joined00:39
→ seamushc joined00:40
seamushc I often hear people say that git is no good at binary files. I havent really seen this in my own experience. Does any one know the real scoop?00:41
spearce its fine with binary files.00:42
corto hi there, i'm starting to use git for my kernel patching, if I need to apply many patches, I see I could create new branches for them, and then merge them together later, but i wonder how rebase or submodules would help...?00:42
spearce its not fine with like 8 GiB binary files.00:42
a lot of the file handling mallocs the whole thing into memory at once00:42
seamushc Ahh, I see00:44
So, its really fine with binaries, as long as they will fit happily into ram?\00:45
rtyler seamushc: IMO it's no worse than other VCSes with binary files, provided they're reasonably sized00:47
rtyler moved a lot of binary (flash) files from SVN to Git a few years ago00:47
bremner and they don't change very much00:50
spearce yea, gits fine, so long as it fits in memory.00:51
actually, i think git is a bit better about binary files than other VCSes are, since it assumes by default everything is binary and shouldn't be munged with keywords or what have you00:51
bremner I recently made a repo with 500MB of binary files in it, (6-12MB each) and found it painful to work with00:52
→ Kelly_zhou joined00:52
bremner as in, it took ages to clone.00:53
cbreak no surprise there00:54
bremner yeah, I just thought I'd semi-quantify the discussion00:54
→ catalis joined00:56
← seamushc left00:58
← jrmuizel left01:00
→ jrmuizel joined01:01
→ brettgoulder joined01:03
→ ryan[WIN] joined01:04
← jrmuizel left01:05
→ jrmuizel joined01:05
← melter_ left01:05
→ lightcap joined01:07
← lightcap left01:07
→ rettub joined01:11
← jrmuizel left01:11
Mike||gone → Mike_lifeguard01:12
→ loincloth joined01:15
→ jrmuizel joined01:16
→ jinzhu joined01:20
→ threeve joined01:23
→ justin-george joined01:25
→ jackdempsey_ joined01:25
← Paraselene__ left01:26
← Mike_lifeguard left01:26
→ justinlilly joined01:27
→ ajpiano joined01:29
Ilari Ugh. I had no idea about how much code writing just low level parts of transport can take: total code: ~190kB. Of that, ~10kB is high-level code dealing with sending requests, ~80kB is dealing with connection itself and ~100kB is about authentication... :-/01:29
... And that's just for client end...01:32
← jrmuizel left01:34
→ ben_alman joined01:34
← bcardarella left01:34
→ bcardarella joined01:35
→ Jeena_ joined01:35
Jeena_ → Jeena01:38
→ jrmuizel joined01:38
← bcardarella left01:39
← ajpiano left01:42
→ ajpiano joined01:45
→ pinnacle joined01:49
→ dutchie joined01:58
← dutchie left01:58
→ dutchie joined01:58
→ imajes joined01:58
→ hursttx joined02:04
← hursttx left02:05
→ bgerber joined02:05
← bgerber left02:06
← corto left02:06
→ consolers joined02:11
→ Frlncr joined02:14
← Frlncr left02:16
WebDragon Ilari: I feel your pain, mang.02:19
forunately all I have to do this week is turn www.w-m-t.com/ into www.w-m-t.com/index_2010.php (still in progress but about 70-80% complete)02:20
→ tg joined02:22
→ dreiss joined02:26
→ VVSiz_ joined02:28
→ bgerber joined02:29
← loincloth left02:31
→ Bass10 joined02:31
→ fr0sty_ joined02:33
→ bcardarella joined02:35
→ loincloth joined02:37
← jackdempsey_ left02:41
← loincloth left02:42
→ loincloth joined02:42
→ hanasaki joined02:47
← Theravadan left02:48
← jrmuizel left02:49
FFForever can I rename a branch called remotes/origin/category to just category?02:50
Ilari FFForever: No. You presumably want 'git checkout -b category origin/category' (to create local branch out of it).02:51
FFForever Ilari, its not a local branch?02:51
the fact I cloned everything does not make it local?02:52
Ilari FFForever: Its remote tracking branch, i.e. mirror of branch from another repository.02:52
FFForever ahhh02:52
hanasaki what is a good linux gui for GIT?02:53
Ilari FFForever: Clone creates local branch for one of the branches in remote repo.02:53
FFForever ahhh02:53
Ilari FFForever: Remote tracking branches are mainly updated by fetch and push.02:54
hanasaki FFForever: so if its distributed... what is the remote repo?02:54
← bcardarella left02:54
FFForever its something on github I am playing with :)02:55
→ zehrique joined02:55
← pinnacle left02:55
hanasaki FFForever: how would someone setup their own?02:55
FFForever dunno02:56
hanasaki my developers will have most of what they need local but maybe not the whole history however we need to have a corp lifetime repos w/ full history and security02:56
→ jaysern joined02:56
FFForever hanasaki, they have a self hosted github version02:57
← hipe_ left02:58
→ hipe joined03:00
→ a joined03:00
hanasaki can't ya just front end it with apache like svn does03:00
?03:00
a hello everybody03:01
→ pinnacle joined03:01
→ mattcodes joined03:02
a i would like to save new copies of a single file i am working on03:03
FFForever hanasaki, dunno, tbh I have my own problems and researching something like that for someone else is not high on my list03:03
a is this possible with git03:03
?03:03
→ johnw joined03:03
→ Guest47746 joined03:04
a is it possible to have revision control on a single file?03:04
I would like to do it without setting up a complicated server03:04
kpreid_ a: git needs no servers03:04
← Guest47746 left03:05
Thumper_ a: just make a repo with that one file in it, track one file or lots of files whatever you need03:05
a ok ill hav a go at it03:05
ill probably be back03:05
thanks very much03:05
← catalis left03:07
← hipe left03:11
→ hipe joined03:11
→ khmarbaise joined03:11
hanasaki what's a good url for git and security.... ie: readonly code , id/pass needed to even view etc...03:12
FFForever hanasaki, you could checkout gitosis03:13
hanasaki thanks03:14
FFForever and you really should get a consultant to answer all of your questions03:14
Ilari, any reason why git is written in python and not c?03:15
→ scissors joined03:17
hanasaki FFForever: if I need a consultant I will stay on svn03:17
← scissors left03:17
FFForever hanasaki, we are not a replacement for google.com03:17
consolers dead c03:17
FFForever or well I am not :)03:17
consolers, c is not dead03:17
isn't python written in c (well the runtime compiler)03:18
consolers it is in asia03:18
asia minor03:18
Ilari FFForever: There's very little Python code in Git.03:18
FFForever Ilari, I thought git was written in python03:18
→ jargan joined03:19
consolers i havent used any python parts of git ever03:19
→ schambers joined03:19
consolers i think its in contrib03:19
→ Azeroth-Work joined03:20
← pinnacle left03:20
→ cytrinox_ joined03:21
→ defectiv joined03:21
consolers find -type f | wc -l => 3032, find -type f -name \*.py | wc -l => 303:21
defectiv is there a simple way to diff two branches?03:21
← brettgoulder left03:21
→ greg8si joined03:22
defectiv is there any way to diff two remote branches?03:22
consolers oh theres also contrib/gitview, which i'm seeing for the first time03:23
FFForever I stand corrected... from wikipedia: C, Bourne Shell, Perl[2]03:23
→ brettgoulder joined03:23
cytrinox_ → cytrinox03:24
→ pantsman joined03:24
→ hipe_ joined03:27
← hipe left03:27
← hipe_ left03:29
→ hipe joined03:29
← hanasaki left03:29
anjen Hi. I jsut made a silly mistake while commiting my repo and some how the result was insted of jsut doing a simple commit, it deleted dozens of files fomr the commit....anyway...since I didn't push it, I'd just as soon pull the current commit from github.03:31
My question is syntax03:31
git pull master origin?03:31
→ munichlinux joined03:33
→ hipe_ joined03:35
jhelwig anjen: You just want to start over, and use whatever state GitHub has?03:36
← hipe left03:36
anjen jhelwig, yeah, i think that's the most prudent way to go03:36
→ tk joined03:37
anjen but it's been months and motnhs since I had to do anything *from github...forgot how.03:37
jhelwig anjen: git reset --hard origin/master <--- This will blow away any uncommitted changes you have, and make your current branch, and working files match whatever state the master branch on your origin remote have.03:37
anjen jhelwig, thank you.03:38
→ SpookyET joined03:40
← hipe_ left03:42
→ hipe joined03:42
← fr0sty_ left03:46
→ SanManuel joined03:50
← munichlinux left03:50
← threeve left03:51
→ Ze_M joined03:51
Ze_M how can i get the git release as compared to a svn releaes?03:52
release*03:52
→ RobertLaptop_ joined03:53
anjen jhelwig, just wanted to exxperess my thanks again. had a hiccup during deployment that took our site down for about an hour, then the commit problem happend while I was online with the hosting company...had me a bit frazzled...anyway thanks so much.03:55
→ threeve joined03:55
→ munichlinux joined03:56
jhelwig anjen: No worries. You're welcome. :-)03:56
Ilari Ze_M: 'git describe'? I can't properly parse the question...03:57
Ze_M Ilari: how do i do for git as i used for svn "svn info" that would give the svn revsion03:58
of git doesnt have revision tags?03:58
ToxicFrog What does 'svn info' do?03:58
jhelwig Ze_M: Git does have revisions, but they're not numeric like SVN.03:59
Ilari Ze_M: Well, 'git describe' tries to describe current version in terms of last annotated tag.03:59
jhelwig Ze_M: Do this, and you'll see: git log --pretty=format:'%H' -103:59
← hipe left03:59
Ze_M jhelwig: so how are they?03:59
ToxicFrog They're SHA1s.03:59
Ilari Ze_M: Or 'git rev-parse HEAD' to see raw hexadecimal ID of last commit.03:59
→ hipe joined03:59
Ze_M ToxicFrog: i used svn revision numbers to set release in rpms anmings04:00
so now with git what can i use?04:00
ToxicFrog Each commit has a unique 40-digit hex ID.04:00
Hmm. Number of commits since the beginning? First N digits of commit id? Release number, where each release increments it?04:00
← a left04:02
Ze_M ToxicFrog: here is an example of a rpm naming: foo-1.2-0.svn12345.104:02
so now with git how can i do if there isnt a release04:03
Morasique Ze_M: you can take the first n characters of the sha hash; 7 is common04:03
ToxicFrog So what's wrong with foo-1.2-0.git309d72.1 ?04:03
Ze_M mean revision04:03
ToxicFrog: and how do i get that git revsion number?04:03
Ilari Ze_M: 'git describe' to get somewhat pretty-printed one?04:04
alduin Do you tag your versions? If so, isn't the version number and the svn revision redundant?04:04
→ stephans joined04:04
Ze_M ]$ git describe04:04
fatal: cannot describe 'd1cf9027b2c735a9a48830a40cbae824a8572d44'04:04
ToxicFrog Ze_M: git log --pretty=oneline -n104:05
Or, if you want "number of commits in this branch since the beginning", you can try something like:04:05
Ilari Ze_M: You don't have annoated tags to act as base. Maybe try 'git describe --tags' (consider all tags).04:05
Ze_M d1cf9027b2c735a9a48830a40cbae824a8572d44 Fixed typo in Phonon::EffectWidget implementation04:05
lines 1-1/1 (END)04:05
ToxicFrog git log --pretty=oneline | wc -l04:05
→ SpookyET_ joined04:06
ToxicFrog This has the advantage (?) that unlike the sha1, it's guaranteed to increase as you make more commits, barring history rewrites.04:06
Ze_M ]$ git describe --tags04:06
fatal: cannot describe 'd1cf9027b2c735a9a48830a40cbae824a8572d44'04:06
Morasique but has the disadvantage of not necessarily being unique04:06
→ DrNick joined04:06
ToxicFrog Morasique: quite.04:06
Ze_M ToxicFrog: so now i finally get a revision number04:07
i only name rpms when i use the svn source code04:07
so if i use the gir source code i also need to tag it with the git revision number04:07
ToxicFrog: pretty hard to get the git revision number....04:08
Ilari Ze_M: If you don't have anything to compare the version to, best one can do is to take some prefix of 'git rev-parse HEAD' (the commit ID).04:08
→ catalis joined04:08
Ze_M with svn, when doiung "svn up" it aleays shows the revsion in the end of the svn update04:08
Ilari For example: 'git rev-parse master' yields '72f06252494ed9f38fe7b7295eca1437869bdb7c' in one of my repos. That is the ID of newest version in master of that repo.04:09
Morasique Ze_M: git doesn't do global revision numbers; every commit has an id, and at any time your HEAD can be pointing at one of those commits04:09
← SpookyET left04:10
Ilari Git commit IDs are globally valid.04:11
Ze_M Ilari: i just need a way to tag the packages04:11
with svn i use the svn revision04:12
→ magn3ts joined04:12
Ze_M so isnt there a simple git command to get the git revision number?04:12
ToxicFrog <Morasique> Ze_M: git doesn't do global revision numbers; every commit has an id, and at any time your HEAD can be pointing at one of those commits04:12
Ilari Ze_M: 'git rev-parse <branch>'?04:12
Ze_M: Sows ID of newest revision on <branch>.04:13
*shows04:13
Ze_M branch?04:14
i think im using trunk04:14
what a confusion....04:14
ToxicFrog You mean master?04:14
Ilari Ze_M: And 'master' is just ordinary branch. It isn't something special like trunk is in SVN...04:14
← SpookyET_ left04:15
ToxicFrog Git doesn't have a single globally incrementing revision number like SVN; it doesn't have an incrementing revision number at all. Each commit gets a single unique ID based on its contents, and that ID is part of the commit, not part of the repo04:15
Ze_M ToxicFrog: so how can i do to name the rpms i do from git repository?04:16
ToxicFrog We've already suggested a few ways.04:16
Ze_M but im confused04:16
ToxicFrog You can take the first N digits of the commit ID, for starters; that's not guaranteed to increment, but it is guaranteed to be unique.04:16
Ze_M i liked the one you said: git log --pretty=oneline | wc -l04:16
→ SpookyET joined04:16
Ze_M ToxicFrog: in wich showed a number 206604:16
ToxicFrog That gives you the number of commits in the current branch04:17
Morasique that's just how many commits are in the history though04:17
Ze_M so this number will increasing with the commits in git tree?04:17
Morasique if you rebase it could go down, and if you branch you could have totally different commits with the same number04:17
→ hipe_ joined04:17
ToxicFrog For a given branch, yes04:17
But if you, say, are on branch A04:17
And it says 10004:17
You switch to branch B, and make some commits, and switch back to A04:17
It'll still say 100, because those commits aren't in A's history.04:17
And of course if you rewrite history all bets are off.04:18
Ze_M why would i change to other branch?04:18
ToxicFrog o.O04:18
→ hyperair joined04:18
Ilari Ze_M: This isn't SVN. :-)04:18
Morasique i continue to get the impression you don't use version control for much :)04:18
← hipe left04:18
ToxicFrog If you really need a single steadily increasingly number, you could always write a post-commit hook that increments and stores it somewhere04:19
Morasique if you only use one branch ever, and you never rebase or otherwise mess with commits, then i suppose the line count thing is ok04:19
Ze_M im usied to checkout svn trees and use the revision number to tag the rpms04:19
Morasique that's weird though04:19
Ze_M what a confusion04:19
let me give an exmaple04:19
Morasique you're trying really hard to apply an svn concept to git04:19
ToxicFrog Ze_M: git has fast, cheap, and easy branching, and thus most people using git use branches a lot - for example, one branch for each feature you're working on04:19
Morasique it's be like if somebody sat in #svn for 20 minutes asking how to rebase a branch04:19
ToxicFrog When you finish a feature, you merge it back into master04:19
Ze_M im packaging phonon, but now phonon as a git tree and not svn04:20
ToxicFrog The branch ceases to exist as a seperate entity and its changes are incorporated into the main branch (what would be "trunk" on svn)04:20
← ajpiano left04:20
→ anjen joined04:21
Morasique Ze_M: is it connected through git-svn? or is it an actual separate git tree04:21
Ze_M a git tree04:21
← johnw left04:22
Morasique mmk. git-svn has a way to get the latest svn revision number, but naturally standard git doesn't04:22
→ hipe joined04:22
Ze_M git clone git://gitorious.org/phonon/phonon.git04:22
ToxicFrog Ze_M: ok, question. Is it important that the release number be an _increasing_ number?04:23
Or is it enough that it just be unique?04:23
Ze_M ToxicFrog: increasing04:23
that increases with the commits04:23
Ilari Ze_M: Generating "sane" revision numbers is easier if you have tags for upstream releases. For example, for that 72f0... revision, one can get that it is 384th revision in its history. But knowing tag for JPC-RR-r8.1 lets one to say its 33 commits ahead of JPC-RR-r8.1, yielding 'JPC-RR-8.1.git33' which is much saner revision number.04:24
← hipe_ left04:24
Ze_M Ilari: your confusing me04:24
Morasique Ze_M: instead of counting from the beginning of the repo, you can count from the last time a tag was made04:25
assuming you ever make tags, which it sounds like you don't04:25
Ze_M Morasique: i dont what??04:26
→ douilspee joined04:26
Morasique tag commits04:26
← anjen left04:26
Morasique same concept as svn tags04:26
← SanManuel left04:27
Ze_M i do commits in some projects, whats the doubt?04:27
Morasique: but what that have to do to what i asked?04:27
← Bass10 left04:27
ToxicFrog Ze_M: you tag each version.04:28
For release numbers, you then count 'commits since last tag'04:28
Morasique it's another way of labeling revisions that's somewhat better than a straight incrementing number04:28
Ze_M ToxicFrog: i tag a rpm with the svn revision number04:30
Morasique it seems like we keep going in circles04:30
it sounds like you should just use git rev-list HEAD | wc -l04:30
SpookyET Has anyone got a link to a comprehensive subtree tutorial?04:30
Morasique it has deficiencies, but not any that seem to be a problem for you04:30
→ jackdempsey_ joined04:30
Ze_M if the svn revision is 20034 the rpm will be rpm-3.1-o.svn20034.104:31
Morasique we get the concept04:31
Ze_M so i just need a git revision to be able to correctly tag rpms that are using git sources04:31
← zehrique left04:32
Morasique am i on ignore or something?04:32
Ze_M Morasique: your confusing me more :)04:32
SpookyET I'm looking for in-depth information than git read-tree --prefix...04:32
Morasique Ze_M: "git rev-list HEAD | wc -l"04:32
Ilari Ze_M: Make tag for 3.1 (or whatever), use git describe and process the output a bit.04:32
Morasique that will give you a number. use that number04:32
Ze_M Morasique: that number is increased with git commits?04:33
Morasique to that particular branch, yes. it's the number of commits in the current branch04:33
Ze_M so it can be called as the git revision?04:34
i need to explain what was donein the %changelog04:34
Morasique i don't think i'd be comfortable saying that on the record, but if you're hell-bent on a monotonically-increasing number based on the number of commits in a branch, that's probably the best way04:35
Ze_M ok04:35
thanks04:35
→ pfote joined04:35
Ilari Its only best way if you don't have tags!04:35
Morasique he doesn't have tags04:35
and trying to talk about tags caused massive confusion04:35
← cloudhead left04:35
Ilari And if you don't have tags, then you get problems trying to form version numbers anyway.04:36
Ze_M Ilari: i know all that04:36
but if using from a svn repo i need to include the svn release in naming04:36
Ilari Ze_M: Like, is this version after or before 1.2?04:36
Ze_M it have to do with correclty tagging and naming a rpm04:37
→ tchan joined04:37
Ilari Ze_M: So should it be named '1.1-g633' or '1.2-g7'? Or what?04:37
→ pheaver joined04:37
Ze_M god04:38
if i dont use a source from a svn i dont need to include the svn revision number in the rpm naming04:38
Ilari Ze_M: Say you package r20034. How you know it is to be called 3.1 and not 3.0?04:39
Ze_M that is the version of the source app04:39
lets take the phonon example04:40
phnon version is 4.3.804:40
so now that i use git source it will be tagged as phonon-4.3.8-0.git2066.104:40
→ fahadsadah joined04:41
Ze_M when version 4.3.8 will be officially release than ill use the official source code and will tag rpm as phonon-4.3.8-104:42
4.3.8-0.git2066.1 is smaller than 4.3.8-104:42
understand now te naming and versioning process of a rpm?04:43
Ilari Ze_M: I have seen some git-versioned packages. IIRC, those used commit date, not revision count.04:44
Ze_M but ToxicFrog already pointed that way that you said04:44
this is the corect way to tag rpms04:44
since date doesnt say from where the source is from04:44
this way you can see that is from the git source tree and its revision number04:45
its far more accurate04:45
Ilari Ze_M: Those revision count numbers are NOT unique even within repo or even within a branch!04:47
Ze_M im not going to discuss the naming and versioning process04:47
← Ze_M left04:47
Ilari Well, that discussion was going nowhere. :-/04:48
→ madewokherd joined04:49
← brettgoulder left04:51
→ mdgrech joined04:52
→ disappearedng joined04:52
→ sh1mmer joined04:54
→ brettgoulder joined04:58
← mdgrech left04:59
← schambers left04:59
→ robinr joined05:02
← disappearedng left05:03
→ disappearedng_ joined05:03
disappearedng_ → disappearedng05:03
→ aresnick joined05:10
← tchan left05:10
← jackdempsey_ left05:14
→ roop joined05:15
→ Axius joined05:17
← defectiv left05:18
→ offby1` joined05:19
← sh1mmer left05:21
→ rchavik joined05:21
← brettgoulder left05:21
→ Arafangion joined05:23
→ tchan joined05:24
→ AkumaStreak joined05:25
→ sh1mmer joined05:26
→ csgeek joined05:26
← justin-george left05:26
→ carrus85 joined05:27
→ numberGrey joined05:28
→ jrmuizel joined05:28
→ AAA_awright joined05:28
← disappearedng left05:29
→ locklace joined05:31
→ zjason``` joined05:32
→ tewk joined05:33
← rettub left05:34
← hyperair left05:35
→ harv joined05:36
→ trodrigues joined05:36
← aresnick left05:36
→ shad joined05:37
→ rettub joined05:37
→ defectiv joined05:38
→ Mage_Dude joined05:40
→ neuro_damage joined05:41
neuro_damage would it be ok to move a .git repo on a server from one server to another without issue?05:42
→ jjore joined05:42
jhelwig neuro_damage: There's nothing in the repo itself to prevent this.05:43
→ disappearedng joined05:44
→ hyperair joined05:44
← pheaver left05:45
→ disappearedng_ joined05:45
← defectiv left05:47
neuro_damage jhelwig: ok very cool :)05:48
→ ArdRigh joined05:50
→ lightcap joined05:52
← AAA_awright left05:52
← tchan left05:52
→ AAA_awright joined05:53
→ Mage joined05:53
← maattd left05:53
→ defectiv joined05:54
→ dytrivedi joined05:57
← disappearedng_ left05:57
→ amerine joined05:57
← Mage_Dude left06:01
← defectiv left06:03
← consolers left06:03
→ tchan joined06:04
→ Khisanth joined06:06
→ cedricv joined06:06
→ Mage_Dude joined06:07
← Mage_Dude left06:09
→ Garen joined06:10
← Axius left06:11
← jmcantrell left06:19
← threeve left06:19
← hyperair left06:22
← SpookyET left06:23
→ SpookyET joined06:23
→ brettgoulder joined06:26
→ disappearedng_ joined06:27
← disappearedng left06:29
← brettgoulder left06:32
← munichlinux left06:33
→ brettgoulder joined06:34
→ JackWinter2 joined06:35
→ bambule joined06:39
← scarabx left06:39
→ disappearedng joined06:43
→ TDJACR joined06:43
← cehteh left06:45
← sh1mmer left06:48
→ spb joined06:49
→ mssm joined06:49
→ cehteh joined06:49
← mattcodes left06:50
← mithro left06:51
← brettgoulder left06:51
→ sh1mmer joined06:51
→ brettgoulder joined06:54
→ guide joined06:54
→ plediii_ joined06:57
→ gmarquezWRS joined06:57
→ jrmuizel_ joined06:58
← jrmuizel left06:58
jrmuizel_ → jrmuizel06:58
→ VAAGMAB joined06:58
→ dpino joined06:59
→ hyperair joined07:00
→ ribasushi joined07:01
← catalis left07:04
→ mattcodes joined07:05
→ mithro joined07:07
← numberGrey left07:07
← bambule left07:08
→ Ramosa joined07:10
← lightcap left07:10
→ SunilThaha joined07:11
→ defectiv joined07:12
← amerine left07:12
→ amerine joined07:13
→ ph^_ joined07:13
→ miro_dietiker joined07:13
← defectiv left07:14
→ johnw joined07:16
miro_dietiker hii can't find any short help for a simple merge problem: someone pushed to my master and i've a merge conflict with my local changes. can't git pull (error: Entry 'xxx' not uptodate. Cannot merge.)07:16
how do i successfully merge this?07:16
← AAA_awright left07:17
→ lightcap joined07:17
johnw do a fetch, then you can manage the merge manually07:18
you'll need to do something about the notuptodate entry07:18
← lightcap left07:19
→ posix4e joined07:19
→ lightcap joined07:19
miro_dietiker johnw - what actually will fetch do with the file?07:20
← posix4e left07:22
→ priidu joined07:22
← SunilThaha left07:22
johnw fetch just brings all the changes down locally07:23
so you can run "git merge" yourself07:23
instead of doing a pull07:23
i'm not sure exactly how to fix your situation, I always forget until I need to do it07:23
miro_dietiker johnw -- so it won't overwrite my file but only update the local git storage07:24
← VVSiz_ left07:24
johnw yeah07:24
miro_dietiker :-)07:24
→ VVSiz joined07:24
→ giallu joined07:26
miro_dietiker hmm.. doesn't do anything (seems it's fetched already)07:26
so i don't know how to handle the merge command07:26
← magn3ts left07:27
miro_dietiker aah .... git fetched HEAD into FETCH_HEAD07:27
→ mjf joined07:29
→ tbf joined07:32
miro_dietiker funny. that was a single line i've changed where i see no conflict when diff'ing ... created diff, undo changes, pulled, applied changes, pushed ...07:32
strange thing this worked07:32
thx johnw07:33
← mjf left07:34
johnw sure07:36
→ jackdempsey joined07:36
→ steffkes joined07:38
← amerine left07:38
→ unijunction joined07:39
→ maxthoursie joined07:40
← lightcap left07:40
→ darwin_ joined07:41
← mssm left07:41
→ lightcap joined07:43
→ cao joined07:44
→ thiago_home joined07:45
→ ntoll joined07:45
→ mtkd joined07:46
← SpookyET left07:47
← tchan left07:47
→ zubin71 joined07:48
zubin71 is it possible to make a patch file using git?07:48
→ psoo joined07:48
zubin71 how could i do that?07:48
→ Longinus00 joined07:49
ArjenL 'git diff $commit1 $commit2 > change.patch' ?07:50
→ munichlinux joined07:50
zubin71 ArjenL, thnkx07:50
Longinus00 so how do you deal with files that git won't reset?07:50
→ tchan joined07:51
Longinus00 ahh, autocrlf was being crazy07:53
→ ludde joined07:53
→ ixti joined07:53
← zubin71 left07:54
→ Paraselene__ joined07:57
→ trochala joined07:57
← Longinus00 left07:57
→ loops joined07:59
← cao left08:02
← justinlilly left08:03
→ malesca joined08:05
→ cao joined08:07
← malesca left08:07
← jrmuizel left08:08
→ db- joined08:11
→ AAA_awright joined08:12
→ jokaro joined08:12
→ wolog joined08:13
→ davidw joined08:13
← lightcap left08:13
davidw → Guest600508:13
→ psankar joined08:15
→ tmske joined08:16
db- consider the following. A branch created from foo called bar. Now I make changes in bar and foo which are on purpose and don't get merged. Now I have to fix a bug in foo, so I created foo_bug1234 branch, in that branch I did several commits to fix that bug. But how can I merge only the changes from foo_bug1234 into bar without also getting the changes from foo?08:19
→ JonathanT joined08:19
db- i obviously don't want to find out every commit by hand and cherrypick those do I?08:20
← jaysern left08:20
← Jeena left08:22
johnw db-: rebase --onto can be used to "multi-cherry-pick"08:23
db- sounds dangerous. foo and bar are remote08:25
→ mrchrisadams joined08:27
→ mastro joined08:28
← thiago_home left08:28
→ mastroDani joined08:28
← mastroDani left08:28
→ aatanasov joined08:31
← jackdempsey left08:31
← AAA_awright left08:35
→ AAA_awright joined08:35
← wolog left08:35
→ wolog joined08:36
wereHamster johnw: pull = fetch+merge08:36
→ ttvd joined08:36
johnw ya08:36
→ ycyuxin joined08:39
← dreiss left08:40
→ ToXedVirus joined08:43
→ bibinou joined08:44
→ Jeena joined08:46
→ galderz joined08:46
← roop left08:48
→ gebi joined08:55
→ tvw joined08:57
← bibinou left08:58
→ mugwump joined09:00
← mrchrisadams left09:03
← greg8si left09:06
← wolog left09:10
← johnw left09:11
→ shruggar joined09:11
→ froschi joined09:11
→ wolog joined09:13
→ jelmer joined09:16
← hyperair left09:18
← Jeena left09:20
→ gma joined09:23
← gma left09:23
→ hyperair joined09:24
→ tatsuya joined09:25
→ christophsturm joined09:26
→ Justme` joined09:27
← ycyuxin left09:32
→ frogonwheels_ joined09:33
← Justme` left09:37
→ petrux joined09:38
→ mason-l joined09:38
→ tobiassjosten joined09:42
← Mage left09:43
→ themill joined09:44
→ patthoyts joined09:45
← rchavik left09:47
→ mrchrisadams joined09:54
→ bluenovember joined09:55
← ArdRigh left09:58
→ Juice10 joined10:00
→ roop joined10:00
Saur Is there an easier way than "git cat-file -p $sha | sed -e '0,/^$/d'" to get at the commit message for a specific object $sha?10:00
→ merlin83 joined10:01
← merlin83 left10:01
→ merlin83 joined10:01
← cehteh left10:03
→ cehteh joined10:03
sitaram git log -1 --format='%s%n%n%b' $sha10:04
ArjenL Saur: Something like this?: git log -1 --format="Subject:%s%nBody:%b" $sha10:04
sitaram almost the same thing10:04
← roop left10:04
sitaram (I get an extra newline though...)10:04
→ jasonn1234 joined10:04
← Kelly_zhou left10:07
Saur Thanks for the suggestions. However, git log does not seem to be able to regenerate the original commit message (it "unbreaks" multi-line subjects for one) so I think I'll stick with git cat-file.10:10
→ solydzajs joined10:11
ArjenL Ah yes, the convention is a one-line subject10:13
→ xanonus joined10:13
Saur Yeah, and since this is part of a commit message validator in the pre-receive hook, I want it to be exactly what the user committed10:14
→ [reaVer] joined10:15
← [reaVer] left10:15
→ [reaVer] joined10:15
→ d0ugal joined10:16
Guest6005 → davidw10:19
← davidw left10:19
→ davidw joined10:19
← davidw left10:20
→ DavidKlein joined10:20
[reaVer] → reaVer10:21
← aatanasov left10:21
→ remi joined10:22
→ aatanasov joined10:23
→ orafu joined10:26
→ IRSeekBot10 joined10:27
→ project2501a joined10:29
→ solydzajs_ joined10:29
← solydzajs_ left10:29
→ MetaCosm_ joined10:29
← solydzajs left10:31
← aatanasov left10:34
→ viniciusluiz joined10:36
→ loinclot_ joined10:38
← loincloth left10:38
← dpino left10:43
→ MetaCosm joined10:44
→ piranha joined10:46
piranha hi guys. Is it possible to run arbitrary command when push fails? (for example, when I tried to do a non-fast-forward push)10:46
→ aatanasov joined10:46
← froschi left10:47
→ d0k joined10:51
ludde warning: too many files (created: 169 deleted: 1577), skipping inexact rename detection10:51
← aatanasov left10:52
ludde how do I get git to NOT skip the rename detection?10:52
ArjenL ludde: configure 'diff.renameLimit' to a suffciently high number10:54
Grum piranha: what problem are you trying to solve?10:55
ludde thanks10:55
piranha Grum: I want a notification when a push is rejected. :-) It lasts for few seconds sometimes and I often don't remember to check if push was successful10:56
BearPerson shouldn't that affect the exit code of the push command?10:57
Grum BearPerson: yeah it should, so you could just wrap it10:57
BearPerson git push origin branch || aplay /usr/local/share/notify/push_failed_tadaa.ogg10:57
piranha hmmm10:58
ok, thanks, I'll try to go this way10:59
→ Wazzzaaa joined10:59
→ dpino joined10:59
← jinzhu left11:00
→ NetHawk joined11:02
→ gitte joined11:03
Wazzzaaa Hey, I want to exclude all class files to commit11:03
A common situation imho11:03
I added *.class to .gitignore. But when I do "git commit -a" I still see all class files11:03
Do I miss something ?11:04
→ cyberglitch joined11:06
→ loincloth joined11:06
← loinclot_ left11:07
Wazzzaaa even when I commit the .gitignore first...11:07
shruggar Wazzzaaa, what do you mean by "see"?11:07
Wazzzaaa The class files are showed as 'new file'11:07
shruggar dit you "git add" them?11:08
→ aatanasov joined11:08
Wazzzaaa uhmmm, yes11:08
shruggar .gitignore only ignores files you haven't specifically told it to add11:08
Wazzzaaa I see them in the text editor which opens after "git commit -a"11:08
→ RdeWilde joined11:09
Wazzzaaa CAn you tell me how to un-add them?11:09
→ marc_kdab joined11:09
→ j416 joined11:09
Wazzzaaa I just did: git add . (I thought it should ignore the *.class files).11:09
shruggar git add . should ignore what's already in .gitignore, yes11:10
→ simplechat joined11:10
shruggar so you made .gitignore, then did git add ., and now "git status" shows them as being new?11:10
j416 Wazzzaaa: but only if those files were not already tracked.11:11
Wazzzaaa yeah, they aren't yet committed. But I think I did the "git add .", before making the .gitignire11:11
.gitignore*11:11
j416 "git status" will tell you how to undo.11:12
Wazzzaaa ok, thnx11:12
← viniciusluiz left11:12
→ viniciusluiz joined11:13
shruggar .gitignore says "I do not intend to track these files, so don't show them to me when I ask which untracked things exist", git add says "track these files", and doesn't care what you've ignored, because it is tracking them, not listing them11:13
Wazzzaaa DO I need to unstage it ?11:13
j416 Wazzzaaa: yes11:13
Wazzzaaa ok :)11:13
j416 staged files will go into the next commit11:13
ignored files are not staged in the first place11:13
→ kukks joined11:13
Wazzzaaa thnx for the help11:14
I really like git :D11:14
so much better than CVS11:14
shruggar Wazzzaaa, that's because it's awesome11:14
→ jds joined11:14
Wazzzaaa agress *11:14
Wazzzaaa s/s/e11:14
← NetHawk left11:15
j416 heh11:15
shruggar actually, "awesome" is too strong a word.. "actually does what it should" sounds about right11:15
which is pretty awesome11:15
j416 hmm11:15
it does what I want it to do, more or less. Which makes it good. :)11:15
← mithro left11:16
→ lemonchicken joined11:17
← aatanasov left11:17
→ tschundeee joined11:19
→ mssm joined11:20
← piranha left11:20
← hipe left11:22
jargan → jast11:27
← munichlinux left11:28
→ bentob0x joined11:32
→ froschi joined11:34
→ steveire joined11:37
← steveire left11:37
→ steveire joined11:37
← bentob0x left11:38
→ bentob0x joined11:38
→ parasti joined11:40
→ viola joined11:44
← hyperair left11:45
→ csaba joined11:46
→ Sigma joined11:46
csaba in svn I cannot define arbitrary keywords, I must use the 4 defined ones... can I define my own custom keywords in git?11:47
← brettgoulder left11:47
csaba For example, I define xxx to have the value zzz, and then all instances of $xxx$ in the file get replaced with zzz11:47
587 users online and no one knows? :)11:48
or is it lunch break....11:48
→ TMinus36 joined11:49
Grum csaba: most people dont abuse git to do that stuff but simply have a autoconf process dealing with it11:49
zomg csaba: 2 minutes is hardly enough time for all 587 of us to see your question :P11:49
csaba the thing is I use a version number which needs to be defined globally... right now, when the src is checked out, we run a script which replaces all instances of @@version@@ with some predefined version11:50
but it would be awesome if we could define this somewhere as a keyword, and let git handle it for us11:50
→ NetHawk joined11:52
Saur csaba: See "Keyword Expansion" in http://progit.org/book/ch7-2.html11:52
TMinus36 Hi all. I'm a bit new to git and from reading the documentation I've become a bit turned around. Just wanted to confirm I'm doing this right. So I have this local repo of files I want to add to my remote repo "Test.git" on my server. Am I correct into thinking that the following sequence is correct, once I setup a "git remote yada" ? 'git add . | git commit -a | git push' ? Everything goes without but nothing is showing up in the directory11:52
goes without *error11:53
csaba ok thanks11:54
parasti TMinus36: sounds like faq non-bare11:54
TMinus36: if the bot would wake up... hang on11:54
TMinus36: http://git.wiki.kernel.org/index.php/GitFaq#Why_won.27t_I_see_changes_in_the_remote_repo_after_.22git_push.22.3F11:55
TMinus36 Ok I'll look it over. Thank you. :)11:56
← cyberglitch left11:57
→ sis joined12:01
→ yunosh joined12:01
yunosh hi. how do i correctly apply a commit patch created by git show manually? git apply only applies the patch like the patch command does, and git am requires a full email header12:02
→ mcdebugger joined12:05
→ mtkd_ joined12:05
shruggar is it possible to default pull from one remote, but default push to another?12:06
← mtkd left12:06
→ giorgian joined12:08
steffkes hm, how to set an chmod on windows :?12:08
← patthoyts left12:08
giorgian hi. I have a git repository with the sites I manage; it's way too big, and git status takes ages. how can I split it into several git repositories (one per site) without loosing history?12:09
j416 giorgian: man git-filterbranch12:10
hm12:10
giorgian: man git-filter-branch12:10
Gitbot giorgian: the git-filter-branch manpage can be found at http://git.or.cz/man/git-filter-branch12:10
j416 there we go :)12:10
this should be able to do what you want.12:10
giorgian j416: thanks12:11
← priidu left12:11
j416 giorgian: it may feel complex, but it can do a lot.12:11
→ mbroeker joined12:12
→ infogulch joined12:15
→ inffcs00 joined12:15
← inffcs00 left12:15
← remi left12:16
→ ksn joined12:16
steffkes anybody an idea? developing on a windows machine, testing on an debian sys .. of course, need some chmods .. how to set them on windows? the debian machine gitr-repo is read-only12:19
→ krynnotaur joined12:20
shruggar steffkes, can you rephase your question? Specifically, what you mean by "chmods", because I can think of three things that could be a misuse of the word to describe12:20
→ twmi joined12:21
steffkes shruggar, of course .. i mean chmod like "chmod 775 foo" on unix-systems :) as needed for write-permission for user in other groups then the owner.12:21
← twmi left12:21
→ twmi joined12:21
steffkes so, if the system needs write-access on the cache folder .. i have to change the chmod. on unix/debian no problem. but how to set an chmod under windows (in git)? git-update-index --chmod knows only execute-flag +-x12:22
better to understand now shruggar ?12:22
→ f0rkit joined12:22
shruggar steffkes, are you saying that you think git on debian is setting the mode to 775 when you check out, and you would like that to happen one windows? Or are you saying git currently does _not_ set the mode (other than +x) on any files, and you want it to do that?12:24
→ rrichardsr3 joined12:24
← infogulch left12:25
→ infogulch joined12:25
→ kkoehne joined12:26
rrichardsr3 i completely forgot about this - if you remove a remote branch, did git ever remove that reference from your local repo when doing a git pull? I dont remember always having to remove that remote reference locally. Its gone when i re-clone the repo however.12:26
steffkes shruggar, other try: if you modify the chmod on debian, then git see's this change and you can commit. on windows, of course you don't have an chmod. but on the final debian system, i'll need (on check out, yes) an correct chmod value12:27
so .. the easiest way is .. change chmod on debian sys, commit to the repo, and all is fine. but like said, the debian system in this project is read only. so the question was, if & if yes, how - can i set an chmod in the git-index .. or somewhere else in the repo ;o12:28
Wazzzaaa steffkes: you mean that when you commit on the windows machine. your permissions (chmod) are f00ked?12:28
→ fool__ joined12:28
← fool__ left12:28
→ fool__ joined12:28
steffkes Wazzzaaa, on windows yes .. f00ked .. hm, windows itself has no chmod-system .. so .. cannot set the right chmod on windows12:28
seem's to be a strange question for the git people ;D12:30
Wazzzaaa I'm relatively new to git. But I know you CAN set permissions on Windows.12:30
steffkes Wazzzaaa, permissions in case of .. write-protected / hidden? that's not what i mean12:30
Wazzzaaa ah, you want executable12:31
steffkes Wazzzaaa, no .. this is possbile with git update-index --chmod=+x ..12:31
Wazzzaaa, i want so set an good old chmod .. something like 775 for an folder12:31
← infogulch left12:32
Wazzzaaa what about chmod=ug+rwx ?12:33
steffkes --chmod=(+|-)x12:33
Set the execute permissions on the updated files.12:33
that is the only possible option .. believing the manual12:33
Wazzzaaa ah12:34
db- how to work with permanent parallel branches. any best practice?12:34
steffkes i cannot set an full chmod this way .. only possible to change the execute flag12:34
db-, best pratice for what?12:34
Wazzzaaa well, as I said, I'm relatively new.12:34
steffkes *practice12:34
Wazzzaaa good luck ;)12:34
db- working with permanent parallel branches12:35
e.g. never centralize in one master12:35
but share patches12:35
steffkes what not use merge?12:35
db- ?!12:39
steffkes perhaps i don't get your question .. :~12:39
also not an advanced git user too *g12:39
db- the short answer is "you can't merge downwards and upwards along another branch"12:42
→ NSchermer joined12:43
NSchermer is it possible to also rebase the first commit in a branch?12:45
← jokaro left12:48
→ webchick joined12:49
→ Axius joined12:49
charon NSchermer: you mean, use 'rebase -i' to edit a root commit? (otherwise you have to define "first" more precisely)12:50
NSchermer charon: yes. the commit has no parent12:50
charon NSchermer: currently not possible, you have to manually edit it and then rebase the rest on top of it12:50
i.e. try something like12:51
git checkout $root_sha1; git tag oldroot; <hack>; git commit --amend; git tag newroot; git checkout master; git rebase --onto newroot oldroot12:51
← Wazzzaaa left12:51
← Axius left12:54
→ keex joined12:54
← xanonus left12:54
→ infogulch joined12:58
← keex left12:59
→ whitman joined12:59
→ nanoo joined13:00
→ fool_ joined13:00
← fool_ left13:00
→ fool_ joined13:00
→ cantoma joined13:00
← infogulch left13:01
→ infogulch joined13:01
→ voxpelli joined13:01
← miro_dietiker left13:02
← fool__ left13:03
→ remi joined13:03
→ brettgoulder joined13:03
← rrichardsr3 left13:04
steffkes hm perhaps i should try to patch it? *think*13:05
← f0rkit left13:05
→ trtzrdb joined13:06
steffkes ah damn .. working on dir's sucks. isn't recorded by git .. argh13:10
← IRSeekBot10 left13:10
→ jas4711_ joined13:13
← RdeWilde left13:13
→ One-Man-Bucket joined13:13
One-Man-Bucket hello13:13
NSchermer charon: that worked, thanks13:14
One-Man-Bucket i want to compare an arbitrary commit and the version before13:14
so i do "git diff b4ff32^ b4ff32"13:14
is there a shorthand for that?13:14
teuf git show b4ff3213:14
Ilari One-Man-Bucket: Or 'git diff b4ff32^!'.13:15
One-Man-Bucket ah ok13:16
thanks, takes a while to get used to git =)13:16
← jas4711_ left13:16
teuf what does ! mean in this case?13:17
→ aatanasov joined13:17
→ jas4711 joined13:17
jas4711 → jas4711_13:18
jas4711_ → jas471113:18
Ilari Commit minus its parents.13:19
foo^! means include foo and exclude parents of foo.13:19
← brettgoulder left13:20
→ Bass10 joined13:20
→ f0rkit joined13:21
← psankar left13:22
→ auslander joined13:22
← auslander left13:22
→ auslander joined13:22
→ Pupeno joined13:23
← NSchermer left13:26
← loincloth left13:26
→ brettgoulder joined13:26
→ loincloth joined13:27
← kukks left13:28
yunosh shouldn't git format-patch use the email address from user.email when building the From: header?13:28
→ priidu joined13:28
→ kukks joined13:28
Ilari yunosh: AFAIK, it uses commit author.13:28
← jas4711 left13:29
yunosh okay. then next question: if i change user.mail since the last commit, should the commit author address be updated when running git commit --amend?13:30
Rhonda Only the commiter.13:30
For the author you need the --author switch13:30
yunosh thanks13:31
Rhonda There is a small distinction between those two, sometimes not completely obvious, even though their naming should suggest thinking the right direction. :)13:31
→ IRSeekBot joined13:31
← j416 left13:32
→ jas4711 joined13:32
yunosh cool, that did it13:33
→ patthoyts joined13:39
← Ramosa left13:40
→ radarek joined13:43
→ drizzd joined13:46
→ tildeequals joined13:47
← mssm left13:48
← brettgoulder left13:51
← radarek left13:52
→ scarabx joined13:53
← krynnotaur left13:54
→ tstclair joined13:56
→ Yuuhi joined13:57
→ brettgoulder joined13:57
→ aziz joined14:00
← dytrivedi left14:02
← ben_alman left14:02
→ mssm joined14:03
← dpino left14:05
→ DavidKlein^ joined14:07
← DavidKlein left14:08
← steffkes left14:08
→ radarek joined14:10
← ph^_ left14:11
→ ph^ joined14:12
→ ph^_ joined14:12
→ Ryback_ joined14:14
→ crankharder joined14:15
→ loinclot_ joined14:15
→ ultraviolet joined14:15
→ Tila joined14:15
Tila hello14:15
jacobat Hello14:15
crankharder so git-svn is pretty good at following standard branching/tagging conventions - but what if the people before me didn't do that? What are my options for converting the repo?14:16
I only really want to get trunk and 2-3 of the branches14:16
Tila i git-stash save some changes, then stash saved other changes, and i would like to apply both now, but stash apply doesn't like a dirty dir :(14:16
jacobat crankharder: I believe you can tell git-svn where branches are stored - or you could just opt for only getting trunk14:17
Tila how can i get out of it?14:17
jacobat Tila: commit the first set of changes then the second set of changes?14:17
← loincloth left14:17
Tila hum, the changes are not complete14:17
it would be a dirty commit14:17
jacobat Tila: I don't mind dirty commits, git is distributed after all14:18
Tila: If you want you can always squash the changesets together later on if you want14:18
→ threeve joined14:18
← mssm left14:18
Tila local doesn't mean invisible14:18
how to squash?14:19
jacobat Tila: 'git rebase -i' is one way14:19
shruggar I just tried it out, seems like a bug in stash, to me14:20
git reset HEAD^14:20
that will basically discard the last commit, but keep the changes it made14:20
Tila shruggar: but stash-apply will refuse to make modifications to a dirty dir14:21
parasti maybe stash is just playing safe, because "what Git doesn't know about doesn't exist" (and thus can't be automerged)14:22
→ ben_alman joined14:22
← brettgoulder left14:23
→ magcius joined14:23
shruggar parasti, it should be like switching branches with a dirty work tree: does the right thing unless the set of changed files has overlaps14:23
Tila jacobat: i didn't fully understand the rebase -i help14:23
shruggar Tila, the order is: "stash apply", "commit", "stash apply", "reset"14:23
jacobat Tila: http://www.gitcasts.com/posts/rebasing14:24
← f0rkit left14:24
parasti shruggar: don't know if it matters, but I think the problem is basically that you're applying untracked->untracked state, whereas when switching branches it's untracked->tracked14:24
→ reprore joined14:25
→ mst__ joined14:25
shruggar parasti, I'll admit I don't know the internals of what checkout does to determine if it's safe, but it's definitely something which can be calculated without creating any additional objects14:26
← Arafangion left14:27
Tila ok, nice, thanks :)14:27
jacobat Tila: :)14:28
shruggar parasti: ie: "do the files listed via "git stash show" overlap with those shown by "git diff"?"14:28
→ ajpi joined14:28
← trtzrdb left14:28
→ consolers joined14:29
parasti I pretty much avoid git stash exactly for this reason... it doesn't really fit into the workflow that the rest of Git encourages; there's never a reason not to commit and fixup later14:30
← locklace left14:31
Grum parasti: what about if committing would break a build14:31
← scarabx left14:31
Grum parasti: great reason to stash if you have to make another quick change14:31
→ locklace joined14:31
parasti Grum: but who cares? it's my local repo, I'm only breaking my own build which I won't even be compiling14:32
Grum there's never a reason not to commit and fixup later <-- just stating there are plenty :)14:32
parasti for me there aren't, that's what I meant14:33
consolers there might be one situation with git-stash, where you can merge/pick thats useful, but i didnt keep notes14:33
→ FND joined14:34
Grum well, someone asking you to look/fix/test something is usually a good indicator you want to stash anything you're working on :)14:34
consolers (this isnt the official git-workflows, i never found that to fit my style)14:34
Grum mm there is an official one?14:34
consolers the one you hear 100 times on the ml14:34
parasti Grum: commit, branch, reset is what I'd do -- save myself a headache14:34
→ TeckniX joined14:34
← jasonn1234 left14:35
consolers the fact that it made it into git to support someones use case is official enough!14:35
Grum stash, branch, checkout, stash pop gives the same result14:35
FND hi - is there a pre-push hook? I want to run py.test locally before connecting to the remote repo, but neither pre-receive nor update seem to be local (according to the man pages)14:35
Grum FND: often you'd run those in post commit hooks14:36
and a pre-push hook you can easily make using the shell14:37
← tildeequals left14:37
→ malumalu joined14:37
FND Grum: I guess I could do post-commt - but my personal workflow would benefit from pre-push (many commits, few pushes)14:37
← radarek left14:37
FND as for using the shell, I did consider using an alias, but that seems a bit overkill/brittle14:37
→ jasonn1234 joined14:37
Grum you could do it on the remote end?14:38
→ mssm joined14:38
→ niko joined14:38
→ Mezner joined14:38
FND Grum: I could do, and live with, all those options - I'd just prefer a pre-push hook - if that's not available, that's fine14:39
Grum FND: seems someone already suggested a patch for this: http://kerneltrap.org/mailarchive/git/2008/8/19/299640414:39
FND thanks, I'll browse that discussion14:40
→ ergates joined14:41
Grum FND: they are stating that having them clientside wouldn't be trustworthy and you 'could as well run: py.test && git push'14:41
→ jelmer__ joined14:42
FND Grum: yes, I expected that, and agree - my problem is that I generally forget to run py.test...14:42
→ f0rkit joined14:42
→ kang joined14:42
FND so similarly, I would forget to run $push_alias_which_runs_py.test14:42
Grum yeah they state that as well14:43
ergates afternoon all: I'm migrating to git from cvs, and am struggling through how-tos on gitosis: I'm wondering about a line in gitosis.conf that can be added - [group myteam]14:43
Grum you could git-alias it and train your fingers on something like git send :D14:43
ergates: it might be wise to setup gitolite instead of gitosis as the later is rather unsupported14:44
and less feature complete, developed, maintained than gitolite14:44
ergates oh, bother!14:44
FND Grum: but what about non-Python projects then? (sorry to be a pain - just playing advocatus diaboli)14:44
ergates Grum, thanks!14:44
Grum FND: i always do make test; before i push :)14:44
FND Grum: clearly your brain is superior to mine14:45
← kang left14:45
← jelmer left14:45
FND but make test is a good point14:45
as it would abstract away the language-specifics14:45
plus it would punish repos without a test target14:45
Grum well its a habit of mine to test it i guess14:45
FND adds send alias to ~/.gitconfig14:46
Grum i sortof agree a pre-push would seem useful as well14:46
as we also have pre-commit14:46
mbroeker i am currently hosting a few projects via webdav. i can grant write access for all projects or none. but now, i want to give write permissions to one user in one project. what solution do you suggest?14:46
Grum and post-receive is basicly pre-push on the remote side14:46
→ cannonball joined14:46
FND Grum: but that's presuming the remote can run the tests, is more than a data store (think GitHub)14:47
Grum mbroeker: i'd reckon that webdav should be able to handle permissions on subdir basis, if it cant, then gitolite could but thats not webdav :(14:47
FND: yeah that is indeed a fat problem :)14:47
FND I think the alias will do - thank you for suggesting that (implicitly)14:47
Grum FND: i'd bring it up on the mailinglist again if you really want it, and try and solve the unsolved problems from the last try14:48
FND I might raise the issue again if the alias doesn't suffice14:48
→ sockmonk joined14:48
ergates Grum, no debian package for gitolite though?14:48
mbroeker Grum, ok, i will read up the apache2 documentation :) thx so far.14:48
Grum mbroeker: its not so much a git restriction, git doesnt really do anything with permissions, maybe someone already made something to easily control a webdav env but i wouldnt know about it :(14:49
ergates: could be, i've just installed it (it ends up being some perl in a user's homedir in the end)14:49
no real reason to pack it14:50
(other than putting restrictions on which user and configuration)14:50
teukka ergates: not yet: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=55081714:50
mbroeker my projects are small, so the overhead of webdav doesn't really matter for me and my friends. it's easy to maintain and that's what i want.14:50
ergates thanks14:51
← loinclot_ left14:51
→ jschoolcraft joined14:53
jschoolcraft is it possible to cherry pick between two git-svn remote tracking branches?14:54
→ radarek joined14:55
parasti jschoolcraft: cherry-pick to a local branch and dcommit14:55
Grum one of them has to be local to be able to append commits to it14:55
jschoolcraft there both local branches14:55
charon jschoolcraft: sure, but make sure you remove the git-svn-id lines from the cherry-picked messages, otherwise confusion will ensue.14:55
ajpi → ajpiano14:58
jschoolcraft is there a better way than cherry pick if I have a 4 or 5 commits i the last 20 or so I want to bring over? basically i want to move a feature to stage14:58
charon rebase -i with a temp branch14:58
← jelmer__ left14:59
← radarek left15:00
→ LiamH joined15:01
← sh1mmer left15:02
→ radarek joined15:04
→ stephhh joined15:05
stephhh hi all15:05
← csaba left15:06
← tmske left15:06
← tschundeee left15:06
→ [reaVer] joined15:06
← [reaVer] left15:06
→ [reaVer] joined15:06
→ mvanholstyn joined15:06
← reaVer left15:07
patthoyts Has anyone attempted to look into 'fromcvs' and fix it to import tags? All the other cvs to git conversion tools are either non-incremental or brVS vendor branches.15:07
fromcvs seems to make a perfect job except there are _no_ tags imported!15:07
fr0sty jschoolcraft: perhaps start another branch and rebase -i to move the commits.15:07
patthoyts 'or break on CVS vendor branches' it should have said.15:09
→ hugoxrosa joined15:09
← stephhh left15:11
← kkoehne left15:15
→ sagsousuke joined15:15
jschoolcraft fr0sty: yeah, i'm trying that route, here's my dilemma. my changes are in master, i need to merge some to staging. so co master; co -b feature, rebase -i first_sha_of_feature_commit, git co staging, git merge feature?15:18
→ brizly joined15:18
← mst__ left15:20
← consolers left15:22
→ drubies joined15:22
→ corto joined15:24
corto hi there, i'm learning git... i tried to merge one remote with my local branch and got a huge list of conflicts, decied to 'git reset' the whole thing, but seems like some files are left behind that i can't get rid of (as seen with git status)15:25
← malumalu left15:25
patthoyts git reset just does the index, use git reset --hard HEAD to put the working tree back to HEAD as well.15:26
EvanR-work git reset --hard is how you avoid15:26
i mean abort15:26
corto patthoyts, EvanR-work, that's what I did, in fact i tried --soft and --mixed as well after, but i still have a few files left from the merged remote (checkout -f different branches doesnt help either)15:27
→ dscataglini_ joined15:28
dscataglini_ Hi there, I would like to report a serious bug15:28
can anybody help me with the procedure15:29
→ Haarg joined15:29
→ loincloth joined15:29
patthoyts corto: you could try git --dry-run clean and see if that will clear out the spare files.15:29
corto patthoyts, sure, gimme a sec15:30
patthoyts reset --hard should clean up a bad merge though.15:30
→ shoe joined15:30
charon dscataglini_: write to [email@hidden.address] with an exact description of the problem. bonus points for a shell script that reproduces the issue.15:30
patthoyts 'clean' can clear out things you might want to keep.15:30
corto patthoyts, ok... yea reset --hard did make the status list very short... hmmm, clean says it would and would not remove some files...15:31
patthoyts, would remove files actually, but would not remove directories...15:31
shoe hi, I've got a little problem - git gui starts up and hangs immediately with a frozen dialog window and a 'wish' process pegging the CPU.15:31
← jschoolcraft left15:32
shoe killing and restarting doesn't help15:32
corto patthoyts, ok thanks15:32
patthoyts man git-clean shows there is a -d option to include dirs.15:32
Gitbot patthoyts: the git-clean manpage can be found at http://git.or.cz/man/git-clean15:32
corto patthoyts, using git clean -d --dry-run looked good, so i replaced --dry-run with -f and it's allright now!15:32
patthoyts cool15:32
dscataglini_ charon: Thanks, btw, I can't do the script because I don't understand why or how can it possibly happen. the problem is that I can see some commits in gitk that don't show up in git log at all. When I try to check out the "invisible" commit I am successful, (I in that branch, with the right comment) but the file don't change15:33
shoe gitk has no problem viewing this repo15:33
→ hyperair joined15:33
dscataglini_ charon: The only place where I can see the changeset or that that commit exists is in gitk15:34
→ jelmer joined15:34
dscataglini_ or gitx15:34
→ jrmuizel joined15:34
shoe and git gui works fine in other repos.15:34
→ corni joined15:34
charon dscataglini_: note that if you rewrite commits, and then update (f5, not ctrl-f5 = reload), gitk will show both pre- and post-rewrite copies of the commit15:34
corto I realize now the merge i attempted may not be straight forward... i'm trying to merge my kernel branch with a linux-kernel patch whose remote probably include a tracked kernel, if that makes any sense (i'm still very noob here)...15:35
→ hebz0rl joined15:35
→ sumangurung joined15:35
shoe ok, I can avoid the frozen dialog because I did 'git gc', but the main git gui window still freezes.15:37
dscataglini_ charon: I'm not sure what you mean by rewrite commits. btw. pre- post-rewrite you mean the follows and preceds?15:38
← aatanasov left15:40
charon dscataglini_: rewriting includes 'git commit --amend', and any use of 'git rebase' or 'git filter-branch'15:40
shoe ah.. n/m. found it with strace -f. git gui was choking on 300 MB of untracked files.15:41
→ michael_scheibe joined15:41
← NetHawk left15:41
dscataglini_ charon: thanks, btw, do you have any clue on why git log wouldn't show anybody touching the file on the day of that commit? or why checking out doesn't show the code in the changeset? this has happened to my team at least a couple of dozen time15:44
← patthoyts left15:44
charon dscataglini_: no; however (no offense) my current feeling is that there is a misunderstanding, not a bug15:45
← jds left15:45
→ NetHawk joined15:45
charon so it might help if you just explained (in here or on the list) what exactly you are doing15:45
for example, what arguments you're giving to git-log15:45
dscataglini_ charon: if the commit hash is 123 and I check out that hash, I should see the code for that commit. Isn't it right?15:46
← giorgian left15:46
charon yes15:47
← gitte left15:47
← brizly left15:48
← sagsousuke left15:48
dscataglini_ charon: if a commit 123 touches many fies including file "foo" on x-mas, I should be able to do git log foo and see that commit on x-mas. Given that I am on the same branch of that commit15:49
→ AAABeef joined15:50
charon no15:51
dscataglini_ ok15:51
charon well, "maybe"15:51
if merges were involved, history simplification is more complicated than that15:51
if no merges were involved, then yes15:52
dscataglini_ ok, so I can understand not showing in the git log15:52
the thing that I don't understand is checking out the code and not seeing any code changes15:52
btw, it's retroactive15:52
charon retroactive?15:53
dscataglini_ the code is there for a week or 2 and then it disappears15:53
Pieter 6==-15:53
dscataglini_ meaning it disappears from git log15:53
it disappear from the codebase15:53
and checking out the commit hash doesn't show anything15:53
charon due to git's storage model that's about as likely as me getting hit by lightning right now (and i'm inside a building...)15:54
dscataglini_ today it's the first time that I was able to find the commit by searching for the comment15:54
← jrmuizel left15:54
dscataglini_ exactly15:54
→ jrmuizel joined15:54
dscataglini_ I am pretty sure that it's somebody in the team trying to do some -interactive -append or rebase stuff15:55
→ carlosleon joined15:55
dscataglini_ but I can't tell15:55
carlosleon Hi there! O/15:55
charon dscataglini_: what does 'git show --name-status $commit' say? does 'git show --name-status $commit:$file' show the right contents?15:55
← jrmuizel left15:55
charon dscataglini_: and if someone is running around rewriting history, anything is possible15:55
dscataglini_ commit show $commit shows the right comment and m files list15:57
→ tedoc2000 joined15:57
carlosleon I'm having an issue with "git pull" and have no clue about what to do. May somebody help this poor guy? (:15:57
dscataglini_ charon: nobody should be doing that, because most of them don't even create local branches most of the times15:58
→ alley_cat joined15:58
dscataglini_ charon: typical workflow is git co -b branch commit then check out master and do git merge mybranch15:58
← michael_scheibe left15:59
→ rostedt joined15:59
dscataglini_ btw, git show --name-status $commit:$file doesn't show the right content15:59
charon: git show --name-status $commit:$file doesn't show the right content (forgot to address you)16:00
rostedt I have an interesting question. Is there a way to update a branch on a bare git repo with one of its remote branches?16:00
carlosleon I'm doing git pull but for some weird reason, my friend's commits are not being retrieved :-/ am I using the incorrect command?16:00
shruggar rostedt, define "update"16:01
← mtkd_ left16:01
rostedt shruggar: let me explain what I'm doing16:01
I have several git repos for various projects that I work on that all relate to the linux kernel16:01
to save space, I download a bare repo from Linus's repo16:01
and have all other repos a clone of that one16:02
crankharder a bunch of the places I've read for converting from svn -> git mention creating a intermediary reposiotory -- and then git clone --bare repo.tmp repo.git -- I did that and it's not a working checkout --- e.g. I couldn't push it up to github, so why do I need to do that?16:02
← mcdebugger left16:02
rostedt I originally did not use a bare repo, but just downloaded Linus's tree as is16:02
charon dscataglini_: well, if 'git show $commit:$file' doesn't show the right content, then that's not git's fault... it means the commit contains something else than what you think it does16:02
rostedt but I never compile in that tree, so I figured I'd just make it bare16:02
charon (the --name-status was a cut&paste mistake)16:02
→ brizly joined16:03
rostedt but now when I update my other repos it only uses what the Linus bare repo was at the time of its clone16:03
charon dscataglini_: one potential way of figuring out what's going on might be to dig through 'git reflog' and find the various incarnations of the commit, if any16:03
rostedt I want to update the bare repo's "branch" that is pulled by other repos, to the origin/master that it has16:03
← drizzd left16:04
rostedt basically, I would like to do a git reset --merge origin/master16:04
but that can't be done without a work tree16:04
shruggar rostedt, why the --merge?16:04
→ Blackshark joined16:04
rostedt well, it seems that all the other repos pull from its local branches16:04
I want to update the local branch to the remote branch16:05
dscataglini_ charon: I hope I am not taking too much of your time16:05
→ drizzd joined16:05
rostedt I guess I can do a hack, and do a git branch update origin/master16:05
and then the other repos will get that update and use it16:05
→ jaysern joined16:06
← carlosleon left16:06
dscataglini_ charon: If I am not. Why would gitx show the right changeset? but git show does not?16:06
← One-Man-Bucket left16:07
shruggar rostedt, have you tried doing exactly what you're doing now, but without saying --merge?16:07
rostedt git reset doesn't work on a bare repo16:08
charon dscataglini_: that is a very good question16:08
rostedt I just did what I wanted in a hacky way16:08
→ struberg joined16:09
→ TAsn joined16:09
rostedt in the bare repo linus.git, I did: git fetch; git branch update origin/master16:09
→ mcdebugger joined16:09
charon dscataglini_: btw, since gitx means OS X ... have you checked whether you're running into any of the known limitations on OS X, mostly case-insensitivity and unicode normalization issues?16:09
← alley_cat left16:09
← jaysern left16:09
dscataglini_ No I'll check16:09
TAsn hey, I accidentally did a local commit, how do I uncommit it? (I want my local to be the same as the remote repo, and I don't want to push this commit)16:09
rostedt in one of my cloned repos I did: git remote update; git merge origin/update; git push origin local:origin/master16:09
dscataglini_ charon: no, I'll check16:09
rostedt now the bare linus.git tree master branch is updated to the origin/master :-p16:10
dscataglini_ charon: I wasn't aware of those issues16:10
rostedt would be nice to do that directly in the bare repo16:10
→ juan_arandaalvar joined16:10
→ roop joined16:11
dscataglini_ charon: thanks for your help16:11
charon dscataglini_: case insensitivity bites you if you have different casings of the same filename in your repo. unicode normalization shouldn't be a problem from GUIs, but it's possible to 'git add $file' and then immediately have it appear deleted if $file contains unnormalized characters16:11
EvanR-work i have a question about submodules. we just added a submodule to our repo, we pulled changes (.gitmodules) of the superproject, and git submodule init / updated. now were ok. but what if the submodule has updates, we want to pull them. how does the superproject know which submodule commit its on?16:11
← whitman left16:11
EvanR-work will that be recorded in .gitmodules?16:12
← psoo left16:12
dscataglini_ charon: I don't have different casing of the same filename. So that wouldn't be the problem. Is it possible that some of the guys are running an old version of git and that might cause it?16:13
→ bcardarella joined16:14
→ jlogsdon joined16:15
EvanR-work ah git submodule status16:15
← FND left16:15
EvanR-work does this just read the HEAD of the submodule?16:15
← Blackshark left16:17
← jelmer left16:17
→ Blackshark joined16:17
← roop left16:17
← [reaVer] left16:18
→ Richardigel joined16:18
Richardigel How, how can I compare whether my working tree and origin/master are identical?16:19
(minus ignored files)16:19
→ reaVer joined16:19
← reaVer left16:19
→ reaVer joined16:19
← jlogsdon left16:19
→ malumalu joined16:19
→ jlogsdon joined16:19
EvanR-work git diff origin/master16:19
well, that compares origin/master and the index16:20
actually it doesnt16:20
git diff origin/master16:20
← douilspee left16:20
Richardigel i just tried this, and it doesn't find an untracked file that i've added.16:20
EvanR-work if you added it, its in the index16:21
Richardigel i haven't16:21
→ bblfish joined16:21
← giallu left16:21
EvanR-work you cant add it without it being in the index16:21
← simplechat left16:21
→ alley_cat joined16:21
Richardigel i haven't added it.16:21
EvanR-work then its untracked16:21
Richardigel exactlz.16:21
EvanR-work so wont count16:21
Richardigel so how do i compare the working tree, including untracked files, to origin/master?16:21
→ douilspee joined16:21
rostedt shruggar: someone just told me this is what I need in my bare repo: git config remote.origin.fetch '+refs/heads/*:refs/heads/*'16:21
EvanR-work if you git add it, then git diff origin/master..HEAD16:22
bblfish anyone interested in making git a lot easier to log into, using a distributed web of trust?16:22
EvanR-work i think16:22
Richardigel of course i could git stash and then compare stash@{0} with origin/master16:22
mbroeker Grum, it's working with AuthGroupFile and Directory Entries for projects that have to be secured. nice and easy. thanks for your help16:22
Richardigel EvanR: ok, how do i compare without adding?16:22
EvanR-work and also not showing all untracked files?16:23
Richardigel EvanR: i do want to see untracked files.16:23
EvanR-work you can add and reset, adding isnt permanent16:23
Richardigel EvanR: i just dont care about ignored files, as specified by .gitignore.16:23
EvanR: it's like i said: i want to compare the result of git stash with origin/master. possibly without actually stashing.16:24
EvanR-work in that case, git stash and compare the stash object with origin/master16:25
:\16:25
Richardigel EvanR: that's clever, but it has the downside of actually stashing.16:25
→ webus joined16:25
EvanR-work well git only acts on objects16:25
untracked files arent in any object16:25
← rostedt left16:25
Richardigel EvanR: a side-effect-free comparison would be nice. the background is that i want to verify in one command whether all my things have been pushed. all my things is: everything in my folder.16:26
EvanR-work if its untracked, it hasnt been pushed16:26
Richardigel EvanR: well, yes. that's what i want to know: am i up to date? has everything been added, committed, pushed?16:26
in one command :)16:26
Saur git status should tell you that...16:27
EvanR-work make a script that reads git status and if it sees untracked (not ignore) files, prints YES or NO16:27
→ c|oneman joined16:27
Richardigel ahh.16:28
shruggar Richardigel, git diff -q origin/master && echo "YES" || echo "NO" ?16:28
err, stop, reverse that :)16:28
→ metellius joined16:28
EvanR-work whats -q16:30
← alley_cat left16:30
EvanR-work quiet16:30
man page doesnt say -q works ;)16:30
so coudnt find it16:30
← radarek left16:30
shruggar ah, sorry, it's --quiet16:31
EvanR-work but it wont work if you forgot to add files16:31
which i do all the time16:31
shruggar I did /-q in the man page, saw the entry I was looking for, didn't bother checking to see it had actually found the exact string I was looking for :)16:31
fr0sty jschoolcraft: (if you are still in the house) That would work. You could also look into doing 'git rebase -i --onto <staging> <first_sha1>'16:31
shruggar EvanR-work, when you forget to add files, would you like them to show up as "different" or "not different" ?16:32
Richardigel shruggar: i get the idea … i didn't know that git status compares with origin/master16:32
fr0sty jschoolcraft: That would eliminate a merge commit and the possible introduction of other changes from master that are not already reflected in staging.16:32
EvanR-work Richardigel: well, origin/whateverbranchyoureon16:32
or default remote for that branch16:32
shruggar: i just need to look more carefully at git status ;)16:33
shruggar EvanR-work, what? Okay, unless I'm very mistaken, "git status" does not do anything with other branches, only HEAD and the index16:34
EvanR-work HEAD does not mean master16:34
→ g0bl1n joined16:35
shruggar right, but since when does HEAD point to origin/master? Unless you've just got a really weird set-up16:35
→ RdeWilde joined16:36
PerlJam shruggar: there are multiple HEADs you know16:36
EvanR-work i didnt say HEAD was origin/master16:36
or if git status was talking about HEAD16:36
bblfish hi, where could I make an architectural suggestion for authentication improvement in git?16:36
PerlJam That's part of the beauty of git ... it lets you do weird and wonderful things :)16:36
EvanR-work git status comprares your index with HEAD, which might be master or might be something else16:36
dscataglini_ charon: I made one more discovery. One commit changed 2 files. git show --name-status $hash:$file1 shows the wrong code. git show --name-status $hash:$file2 shows the right code16:38
shruggar PerlJam, there are multiple HEADs, or multiple heads? Because I thought there was one HEAD, and multiple heads16:38
dscataglini_ charon: also git whatchanged shows the right diffs16:38
EvanR-work shruggar: one HEAD which is a reference to whatever you want, doesnt have to be latest commit on master. but earlier i was talking about latest commit on some branch16:39
→ tech13 joined16:39
PerlJam shruggar: well, at any given moment, there is one HEAD. But it all depends on which side of the looking glass you're on.16:39
→ Zifer joined16:39
← juan_arandaalvar left16:39
→ jaysern joined16:40
← corto left16:40
← nanoo left16:40
→ codeswing joined16:41
→ alley_cat joined16:41
fr0sty bblfish: [email@hidden.address]16:43
← wolog left16:43
→ jrmuizel joined16:43
← f0rkit left16:43
fr0sty bblfish: you could try it out on us as well, but there are only a few git devs here.16:43
bblfish ah ok16:44
→ lucsky joined16:44
bblfish we are working on a very simple distributed authentication protocol that uses SSL just as it was meant to be16:44
http://esw.w3.org/topic/foaf+ssl16:44
this would make it very easy for people to allow access to git by simple rules such as: allow all developers on this project edit access16:45
where the group could be defined in a foaf file16:45
tech13 attempting to checkout a copy of Zend Framework following a blog entry's how-to. Having an issue at the step for handling their incubator. http://pastie.org/844366 for the commands, a link to the blog and .git/config used. Suggestions on what to check?16:45
bblfish I am sure there is something one could do with git+foaf+ssl that would be very nice :-)16:46
→ tildeequals joined16:46
← tobiassjosten left16:50
EvanR-work i am confused about tags16:52
are tag objects distributed with the repo?16:53
i cant seem to 'commit' and tag objects16:53
any*16:53
→ floppyears joined16:55
← Tila left16:55
EvanR-work ah, git push origin --tags16:57
→ coolcat joined16:57
→ amerine joined16:57
→ pheaver joined16:58
→ Hackykid joined17:00
→ silug joined17:01
→ corto joined17:02
corto hi guys, i'm wondering if my distro is responsible for this or now... has gitk changed name at some point? i can't seem to locate it17:02
→ ericpugh joined17:03
← shoe left17:03
→ shoe joined17:03
→ Zephyrus joined17:03
← Zephyrus left17:03
→ Zephyrus joined17:03
← alley_cat left17:04
ericpugh Hi all, I am using git-svn on OSX, and git svn --version reports svn at 1.4.4, but svn --version reports 1.6. where does git find it's version of SVN it is useing?17:04
→ lightcap joined17:04
davr corto: still gitk as far as I know. it may be a separate install in your distro than from the main git package17:04
corto davr, ok, i also found gitview... is this related?17:05
davr ericpugh: it's built into git-svn I think? I have the same issue on windows with msysgit17:05
shruggar ericpugh, probably the version of the SVN perl module17:05
ericpugh davr I was afarid of that.. I ahven't been able to locate the SVN perl module to try and update it.17:05
davr corto: I think gitview is a different app with similar functions17:05
ericpugh I am using MacPorts to do the install17:05
← gebi left17:06
← douilspee left17:06
davr ericpugh: it's not a huge issue, SVN 1.4.4 can talk to SVN 1.6 servers, it's just a lot slower than accessing the SVN repo directly via file access17:06
corto davr, ok, thanks, i'll try to sort this out... i seem to have the man pages for gitk, but not the executable... =/ i'll try to get help from the distro support17:06
← marc_kdab left17:07
davr ericpugh: I ended up doing git-svn on the linux server that hosted the SVN repo, and the pulling from there to my windows client17:07
ericpugh davr, unfortunantly my svn server rejects my commits because the client isn't 1.5 or later. I suppose I could do that approach of using an intermediate server.17:07
I was hoping it was fixable via makeing a symbolic link to my 1.6 svn executable!17:08
thanks for the help though.17:08
← lightcap left17:09
→ mtkd joined17:10
← mtkd left17:10
→ Hackykid_ joined17:10
← bblfish left17:11
← g0bl1n left17:11
← Hackykid left17:12
→ douilspee joined17:12
→ keex joined17:13
keex hello17:13
→ lightcap joined17:13
corto i'm trying to understand the concept of tracking... if i do 'git checkout --track -b myBranch remotes/someGit/master' ... will myBranch be built upon the latest changes in "someGit/master" (ie will it rebase automatically) or will it point to the commit at the moment of this checkout?17:14
← ericpugh left17:14
keex is it possible to migrate a git repository to svn ? I have done all my work using git (which has saved me looooots of time and nerves) but the place i did my thesis for only makes use of SVN. I have a SVN account but havent checked it out yet...17:15
→ pingwin joined17:15
→ alley_cat joined17:15
shruggar all --track does is set up some config variables so that "git push" and "git pull" default to pulling/pushing from the tracked branch17:15
→ psankar joined17:15
keex my work is done, and I'd like to give them all revisions or commits17:16
← jrmuizel left17:16
→ DirtyColombian joined17:16
→ jrmuizel joined17:16
pingwin is there a way for me to work in a branch, push files (stage them) without having to commit and have that message... change branches... work, come back to the branch work some more and THEN commit the changes?17:16
shruggar keex, is your history linear?17:16
DirtyColombian hey folks, whats the best way to apply a stash to a dirty working tree?17:16
keex shruggar, most of it17:16
ToxicFrog pingwin: git stash?17:16
← jrmuizel left17:17
pingwin ToxicFrog: k, i'll take a look at it17:17
DirtyColombian git stash apply/pop respond with : Cannot apply to a dirty working tree, please stage your changes17:17
shruggar pingwin, are you sure you're using the word "push" correctly?17:17
← froschi left17:18
← TAsn left17:19
← christophsturm left17:19
→ robinbowes joined17:20
→ ericpugh joined17:21
→ kipras joined17:21
tsb I've made some changes. I'd like to save these in some other branch, the git pull the updates into master. How'd I do that? I'm guessing git branch is involved but..17:21
pingwin shruggar: no.. I do not want to push to my remote branch, stash looks like it's exactly what I wanted17:21
← reprore left17:22
DirtyColombian for my case: what about git commit, git stash pop, git reset HEAD^ --soft17:22
← d0ugal left17:22
← pingwin left17:23
shruggar pingwin, also useful might be the knowledge that if you don't push, history is not set in stone. That is, you can commit, switch to another branch, work there for a while, come back, commit more, and then clean up your commits so they are cleaner when you are ready to push17:23
← JonathanT left17:23
← alley_cat left17:23
shruggar pingwin, ie: "commit" means "make sure I can get back to this state" not "save this state forever so that everyone can see it"17:24
→ mtkd joined17:24
→ kumbayo joined17:25
→ mizipzor joined17:25
ajpiano → mikes_family17:25
→ jrmuizel joined17:25
mizipzor in my project i have two branches; "branch_a" and "branch_b"... i have some local changes i planned to commit in branch_b, but im in branch_a... git wont let me switch due to the local changes... what the best way to commit local changes to another branch?17:26
RandalSchwartz "git checkout branch_b" doesn't work?17:26
how old is your git?17:26
← mastro left17:26
← corni left17:26
shruggar mizipzor, stash, checkout, then stash pop17:27
→ mklappstuhl joined17:27
→ corni joined17:27
mikes_family → ajpiano17:27
mizipzor shruggar, thx17:27
shruggar mizipzor, alternatively, commit to branch_a, checkout branch_b, cherry-pick branch_a, checkout branch_a, then reset --hard HEAD^17:27
RandalSchwartz and if you're making changes, why aren't you on a topic branch? :)17:27
mizipzor shash seems a simplier approach17:27
RandalSchwartz, honest misstake :)17:27
RandalSchwartz no - I mean a new branch just for those changes17:28
← corni left17:28
RandalSchwartz branches are *very* cheap17:28
→ corni joined17:28
mizipzor i thought i had created a new branch, but i hadnt17:28
RandalSchwartz git checkout -b newbranch didn't work either?17:28
what did it say?17:29
modern git can switch branches with a dirty tre17:29
→ corni_ joined17:29
RandalSchwartz since that's a common mistake17:29
← corni left17:29
mizipzor hmm... didnt try the -b flag17:29
← corni_ left17:29
RandalSchwartz how old is your git17:29
← lightcap left17:30
→ corni joined17:30
mizipzor 1.6.5.1.1367.gcd4817:30
RandalSchwartz that's not too old.17:30
so the "git checkout -b" should work17:30
mizipzor the stash approach seemed to do exactly what i wanted though17:31
← themill left17:31
← tildeequals left17:31
mizipzor the branch i wanted to commit the stuff to was already created... would -b have worked anyway? seems to create a new branch at a set "start_point"... which i assume means my dirty tree?17:32
← tatsuya left17:32
shruggar mizipzor, you can create with -b a "topic branch", meaning "a branch just for the purpose of what I'm working on right now"17:32
mizipzor, then when you're done, you can merge it into the branch you actually want the changes to be in17:33
mizipzor thats what i usually do... but i create my branches with -t17:33
← shruggar left17:34
→ tetha joined17:34
RandalSchwartz -t ?17:34
add -b to -t17:35
they're orthogonal17:35
mizipzor the way ive understood it, -t makes so that the branch remembers from which branch it was created from... to make it easy to pull in new changes or something... havent really made use of that yet, i just merge "normally" if there is such a thing...17:37
→ maxter joined17:37
tech13 attempting to checkout a copy of Zend Framework following a blog entry's how-to. Having an issue at the step for handling their incubator. http://pastie.org/844366 for the commands, a link to the blog and .git/config used. Suggestions on what to check?17:39
← webchick left17:40
davr tech13: it's saying "svn-incubator" is not a valid branch, and instead trying to interpret it as a file17:41
← petrux left17:41
davr tech13: you can do 'git branch -a' to see all available branches. my guess is that it should be "svn/incubator" instead, but that's just a guess17:41
→ thiago_home joined17:42
tech13 davr, doesn't list any variation of incubator.17:42
← corni left17:42
tech13 davr, the fetch doesn't have any output, so I'm guessing something about the remote portion of the config is wrong, but not sure why or what to tweak about it.17:43
davr yeah normally 'git svn fetch' should give you piles and piles of output and take ages17:43
it's weird, I just copied & pasted your .git/config into a new empty git repo, ran your git svn fetch command as listed, and i'ts doing a bunch of stuff now for me17:45
→ stkrzysiak joined17:45
← DirtyColombian left17:45
→ seanmccann joined17:46
→ sh1mmer joined17:46
tech13 davr, weird. able to start clean and the git incubator works. Haven't tried getting the standard svn portion yet though. Guess I'll change the order of the steps and run with it.17:47
mizipzor im working on an svn repos... after "git svn fetch", the output of "git svn log" shows the latest revision, but "git log" does not... how do i merge the svn changes into my own tree? and isnt fetch supposed to do that?17:47
→ psoo joined17:48
davr mizipzor: no, fetch just loads new data from remote locations into your local repository, it doesn't actually do anything to your local branches17:48
uau 'git svn fetch' should fetch the changes into the git-svn branch17:48
so something like 'git log git-svn' should show them17:48
mizipzor wierd... i only have a master branch (havent changed anything locally, just keeping it up to date)17:48
uau if you're on another branch then that branch will not change and 'git log' will of course not show anything different17:48
the 'git-svn' branch is classified as a 'remote' branch17:49
so 'git branch' only shows it if you use -r (or -a)17:49
mizipzor git branch -a shows master along with all tags/branches on the svn repos under remotes/17:49
← priidu left17:50
mizipzor hmm... should i create a git-svn branch and run git svn fetch again?17:50
uau if you configured git-svn to fetch multiple svn "branches" separately then the branch naming may differ17:50
→ jakob joined17:51
mizipzor i guess remotes/trunk is the one i want... so i need to merge that one into my master17:51
seems to have done the trick :)17:52
← Zephyrus left17:52
→ wolog joined17:53
→ marc_kdab joined17:54
← Richardigel left17:57
← jakob left17:57
→ godsflaw joined17:57
→ jakob joined17:57
← tchan left17:57
← sh1mmer left17:57
→ defectiv joined17:58
defectiv to diff master and staging do i just do git diff master staging?17:58
→ yann joined17:59
← tstclair left18:00
uau defectiv: if "staging" is a branch name then yes (why did you ask here instead of simply testing it?)18:00
→ tchan joined18:00
defectiv diff origin/staging..origin/master18:00
i did test it. but i had no idea whether what it was doing was what i wanted.18:00
thx18:01
uau 'git diff branch1 branch2' is the same as 'git diff branch1..branch2'18:01
→ sh1mmer joined18:03
→ justin-george joined18:03
→ rgr joined18:04
→ segher joined18:04
← jrmuizel left18:06
→ kpreid__ joined18:08
kpreid__ → kpreid_18:08
← tedoc2000 left18:09
→ dreiss joined18:09
→ g0bl1n joined18:10
← mtkd left18:10
→ tedoc2000 joined18:14
← Juice10 left18:14
← tedoc2000 left18:15
→ dcampano joined18:15
→ ntoll_ joined18:16
dcampano if i want to revert a file back to an old commit, i thought you could do "git checkout SHA file", but it doesn't seem to work18:16
am i doing that wrong?18:16
fr0sty what error are you getting?18:17
→ tedoc2000 joined18:17
dcampano well, when i show the status it says there are changes, but git diff doesn't show any changes and file is still the same18:18
← ntoll left18:19
ntoll_ → ntoll18:19
← mrchrisadams left18:20
fr0sty are you sure there is a difference between HEAD and <SHA>? 'git diff SHA..HEAD file'18:21
← viola left18:22
← seanmccann left18:22
parasti dcampano: git diff --cached should show changes if the files were different18:22
→ seanmccann joined18:22
parasti dcampano: that's because git checkout $rev $file will set both the work tree _and_ index to the new state18:23
dcampano parasti: --cached does show the changes18:23
parasti dcampano: so plain git diff, which compares work tree with index, will show no changes18:23
→ eletuchy joined18:23
→ mw joined18:23
← godsflaw left18:25
→ froschi joined18:25
dcampano parasti: stupid user error, i was refreshing our prod site looking for the file to be different instead of dev18:26
← rgr left18:26
→ rgr joined18:26
→ bindjp joined18:27
← ericpugh left18:28
← bluenovember left18:28
dcampano parasti: thanks for the tip on --cached18:28
bindjp should I use core.autocrlf = true or core.autocrlf = input and what is the difference? The documenation was hard to understand18:28
← tedoc2000 left18:29
→ parigaudi joined18:31
parigaudi hi18:32
i'm still quite a n00b to git18:32
and have a prob with the following:18:32
i made different changes on 2 different machines. now i pushed one to origin, but have merge problems on the second.18:33
what's the correct way of dealing with this?18:34
is there a way to have both files (local and origin, with a suffix) side to side and merge them by hand?18:35
say somefile.cpp.local and somefile.cpp.origin or so?18:35
← galderz left18:35
→ Deep6 joined18:36
Ilari parigaudi: You got merge conflicts from git merge or git pull?18:36
→ zodiak joined18:36
Deep6 guys is it possible to have git only work on content of files as opposed to datestamps etc18:36
parigaudi pull18:36
Ilari parigaudi: It should leave conflict markers to files...18:36
parigaudi: Also, there's 'git mergetool'.18:36
Deep6: Err... Git doesn't store file dates.18:37
← psoo left18:37
parigaudi ilari: you mean: conflict markers _inside_ the files?18:37
Ilari parigaudi: Yes.18:37
parigaudi eek18:37
is there no way to have instead two separate files, one local and one origin?18:38
Ilari parigaudi: 'git status' lists files that have merge conflicts.18:38
→ xperience joined18:38
Ilari parigaudi: Some merge tools (git mergetool) might support that mode...18:38
→ SpookyET joined18:39
← Sigma left18:39
parigaudi ok thx ilari, i'll try that18:39
→ radarek joined18:40
→ kylefox joined18:41
crankharder how would a pre-commit 'verification' hook work in git's distributed environment, that is -- say I require commit messages to match a certain regex or it fails, how do I ensure that this applies to everyone18:41
← psankar left18:41
crankharder if I add the hook in the main repo that everyone pushes to, they're not going to have the hook in their local repo18:41
← jasonn1234 left18:42
kylefox Hey everyone. Newb question here. How can I revert my entire branch back to a specific commit? We tried adding a new feature (a series of commits in master) but it's totally not working, and we want to go back to a specific checkout.18:42
Ilari crankharder: update hook in "central" repo that rejects pushes. There is no way to ensure that everyone runs some hook, and for good reasons.18:43
→ jasonn1234 joined18:44
Ilari kylefox: 'git reset --merge <commit>'. But that can be problematic if those commits have been pushed...18:44
kylefox yeah, the commits have already been pushed :/18:45
Ilari kylefox: Another way would be 'git stash', 'git read-tree <commit>', 'git commit', 'git reset --hard'.18:45
→ webchick joined18:47
→ hipe joined18:48
→ gebi joined18:49
← drubies left18:52
→ smotchkkiss joined18:52
bindjp what is the diff: core.autocrlf = true or input ? The documentation on this hard to understand. My actionscript changes are marking the whole page as changed.18:52
← corto left18:53
← hebz0rl left18:53
Ilari bindjp: true does the translation on both checkout and add, input does it only at add time.18:53
bindjp Ilari: thank you very much for clearing that up. So true may be my better option for me and my team?18:57
cbreak work with proper line endings from the start18:58
and don't auto muck around with them18:58
as long as you're consistent, it should work with both LF and CRLF18:58
→ [reaVer] joined19:00
← [reaVer] left19:00
→ [reaVer] joined19:00
→ tildeequals joined19:00
← seanmccann left19:00
← reaVer left19:02
bindjp cbreak: I inherited a project that had some strange ones in only some of the files19:02
→ chewbranca joined19:02
[reaVer] → reaVer19:02
chewbranca I'm trying to figure out a good deploy strategy. I've got a framework that will be powering a number of applications that only differ in config files and static files such as css and images. Should I just make a fork of the framework for each of the apps I'm building or is there a better way to do it?19:04
bindjp cbreak: I am sure they would be consistent and so the teams if we were doing it from scratch because we all use LF on textmate19:05
→ godsflaw joined19:06
RandalSchwartz assemble your app with your configs using puppet19:07
put the whole config tree of puppet into a separate repo19:07
← kylefox left19:09
chewbranca hmm... haven't played with puppet, I'll have to look into that, although I'm a little hesitant to store database configs and api keys in github even if it is a private repo19:09
→ tedoc2000 joined19:09
→ tstclair joined19:09
→ dj_tjerk joined19:10
RandalSchwartz no - that's why it should be a separate repo19:10
put your core code public19:10
and your configs locally19:10
chewbranca the configs have to be accessible by several people19:11
even with having a separate repo for configs, it has to be online somewhere19:11
RandalSchwartz sure, but you don't need github :)19:11
chewbranca true19:11
RandalSchwartz buy a $10/month VPS, put gitosis on it19:11
control your own ssh keys19:12
← tk left19:12
→ tk joined19:12
chewbranca I've got gitosis on my home server, and could possibly just put it on the server hosting all these apps19:12
→ joachim_ joined19:12
RandalSchwartz so, there ya go. :)19:13
→ johnw joined19:13
← magcius left19:18
chewbranca cool, that looks like it might actually work pretty well. I'm using capistrano for deploys, but from my limited researching in the last few minutes it looks like puppet and capistrano can play together pretty well, dividing up tasks, isolate capistrano for deploying and puppet for system state/configuration19:18
thanks for pointing out puppet19:18
← maxter left19:19
→ tobiassjosten joined19:20
→ jmcantrell joined19:20
← tildeequals left19:21
← bindjp left19:21
→ bindjp joined19:21
→ _iron joined19:23
← tedoc2000 left19:25
← mklappstuhl left19:26
→ brodyberg joined19:27
← tvw left19:28
→ Chillance joined19:29
→ mvanholstyn_ joined19:29
→ tedoc2000 joined19:30
→ SubMatrix joined19:30
→ spearce joined19:30
SubMatrix i am getting the following error: ssh: Could not resolve hostname C: hostname nor servname provided, or not known. I believe it has to do with cygwin not allowing me to reference my C drive directly. How do I get around this?19:31
← mvanholstyn left19:32
mvanholstyn_ → mvanholstyn19:32
→ themill joined19:34
Ilari SubMatrix: I think drives are mounted under Cygwin (Unix filesystem has only single root).19:34
SubMatrix Ilari: meaning what?19:35
← ergates left19:35
SubMatrix Ilari: i am just trying to do a git pull19:36
Ilari SubMatrix: I think that it has to be specified as /cygdrive/c/...19:36
SubMatrix how would i modify my git pull to account for such?19:37
→ Richardigel joined19:38
→ nanoo joined19:38
← Richardigel left19:40
Ilari SubMatrix: Change the URL (via command line or via .git/config)19:42
→ dys joined19:45
Deep6 Ilari, sorry just getting back from lunch now19:46
← bcardarella left19:46
Deep6 Ilari, I'm kind of using git for a questionable purpose basically I am backing up all my network equipment onto a tftp server via an snmp writenet script19:46
→ jaql joined19:47
Deep6 every night,so everytime the file is replaced, it's a "change" that needs to be committed to git19:47
I am wondering if I can "dumb" down git so it deosn't care about time stamps etc/md5 sum's (pardon my ignorance if I am way off base)19:48
→ vinse joined19:48
← tbf left19:48
→ alreadygone joined19:48
alreadygone hi, when I do: git commit I get this: fatal: object ca9c99ff4e022b9c783dc92e5d84312a272466c8 is corrupted19:49
Ilari Deep6: The only things that are stored about each file per revision: name, Regular/Executable/Symlink and contents.19:49
alreadygone what to do? electricity goes out a lot, and that's what happened19:49
PerlJam alreadygone: get better power19:49
rchady buy a UPS19:50
Ilari alreadygone: 'mv .git/objects/ca/9c99ff4e022b9c783dc92e5d84312a272466c8 .git/objects/ca9c99ff4e022b9c783dc92e5d84312a272466c8.bak' and then run 'git fsck --full'.19:50
Deep6 ok, so if the file doesn't change at all then it won't need a re-commit?19:50
alreadygone thanks Ilari19:50
Ilari Deep6: Yes.19:50
Deep6 ok, well that's good news....with one exception19:51
is there a way for git to ignore a certian pattern within a file?19:51
alreadygone Ilari, http://www.pastie.org/84466119:52
Ilari Deep6: Well, what is committed doesn't have to bear resembelance to file on disk, but doing that is often bad idea.19:52
← codeswing left19:52
← mw left19:52
← unijunction left19:52
→ mw joined19:52
← mw left19:52
→ mw joined19:52
Deep6 because unfortunately everytime the snmp script gets the file, it creates a timestamp within the config file in the form of "Last Configuration Change" and NVRAM config last updated19:52
alreadygone Ilari, kindly reload it, I edited the post...19:53
← keex left19:54
Ilari alreadygone: 'git reset --soft master@{1}'.19:54
alreadygone: But if you just commit after that, presumably new commit will be combined with previous one...19:54
→ BlindOracle joined19:55
alreadygone kindly reload the page again Ilari :)19:55
← g0bl1n left19:55
← mvanholstyn left19:56
Ilari alreadygone: Does 'git status' complain about HEAD or something?19:56
→ eletuchy_ joined19:57
→ mvanholstyn joined19:57
alreadygone Ilari I just tried git commit now and it worked!19:57
← eletuchy_ left19:57
← eletuchy left19:57
Ilari alreadygone: You might want to use 'git show HEAD' to show whats in latest commit and then split it using interactive rebase...19:58
→ eletuchy joined19:58
alreadygone ok19:58
Deep6 ilari any chance I can make git that "dumb"?19:58
Ilari Deep6: Using hash-object followed by update-index one can insert anything one wants as some file contents...19:59
→ doener joined19:59
Deep6 hrm, I don't think I need to add something, I just need Git to ignore certain patterns within the files19:59
← eletuchy left20:00
← smotchkkiss left20:00
Ilari Deep6: That would require doing some really advanced things..20:01
Deep6: Can you generate the file as you want it to look in commit?20:02
→ HG` joined20:02
Deep6 well the file is a cisco router config, that is essentially written via tftp20:02
Ilari Deep6: E.g. if there are lines that shouldn't be in the commit to remove them.20:02
→ struberg1 joined20:02
← struberg left20:03
Deep6 hrmm, I'm wondering if that invalidates the "control" process20:03
this is one of those audit things20:03
Ilari Deep6: If you want git to commit something else than what you have on disk, first you need to generate the version that will be committed.20:03
→ eletuchy joined20:03
Deep6 and I need to generate it outside the git tracked repository correct?20:04
and have the final version end up within the repositoru20:04
er repository20:05
Ilari Deep6: Well, you can use the last committed state as reference when generating the new version.20:05
← alreadygone left20:05
Deep6 well the file comes in automatically via a tftp transfer20:05
Ilari Deep6: The main inputs you have when generating new version is last committed version and version received via TFTP.20:07
→ _Vi joined20:07
Deep6 Ilari correct.20:07
Ilari Deep6: From those, make the file look like you want it to be in repository, add and commit.20:08
Deep6: (to get the current version: 'git cat-file blob HEAD:<name-of-file>' (dumps to stdout).20:08
→ unijunction joined20:08
← tedoc2000 left20:08
Ilari Deep6: And presumably that needs to occur unattended, so be very careful.20:10
→ Nies joined20:10
→ jackdempsey joined20:11
← jaql left20:11
Deep6 Ilari, well I think if I can prune out the text I don't want before it ends up in the git directory then for the first one I'll have a differential that I'll need to commit, but thereafter I should only have to commit when I've actually made a confiugration change20:12
← bentob0x left20:12
Ilari Deep6: You could also compare what you would commit before committing and not do the commit if new tree and tree of HEAD (you need to build new tree anyway, since commit SHOULD NOT be used unattended).20:13
If those trees have the same ID that is.20:14
Deep6 For my purposes is it a problem if the commit occurs unattended?20:14
I'm not managing source code, just router configurations20:14
Ilari Deep6: "Commit" there refers to to 'git commit' command.20:15
Deep6 I was thinking until I get a more controlled process, it's not a huge deal if I end up having to auto commit and the "Last Configuration stuff" ends up being the difference in each file for months on end20:15
Ilari, I'm aware of what you were suggesting20:16
Ilari Deep6: The way to do commits unattended is to first make index look like you want it, then build tree out of it (write-tree), then commit the tree (commit-tree) and finally update HEAD (update-ref).20:16
Deep6 hrmm.... I'm going to take that as a note20:17
Ilari Deep6: 'git commit' is designed for interactive use. Those commands are the low-level commands to manipulate repository, and are designed to be scripted.20:18
← webus left20:19
→ agib joined20:21
→ struberg joined20:22
← struberg1 left20:22
← struberg left20:23
→ Novelocrat joined20:28
→ d0ugal joined20:30
← gebi left20:32
← kumbayo left20:34
Deep6 Ilari thanks!20:35
silug could someone please verify that i'm not telling anyone anything too stupid/dangerous in this blog post i just made? http://bit.ly/aTlPu420:39
vinse silug: you should remove the part about running with knives20:41
crab silug: seems ok.20:41
silug crab!@#$!20:41
crab :)20:41
→ Plouj joined20:42
Plouj hi20:43
howcome I can't use git merge/pull on bare repositories?20:43
→ JoeConyers|away joined20:45
← JoeConyers|away left20:45
→ joeconyers joined20:46
→ Fern joined20:47
← Fern left20:47
→ juan_arandaalvar joined20:48
← juan_arandaalvar left20:48
← Haarg left20:49
→ Alagar joined20:50
crab because merge (and thus pull) needs a working directory.20:50
otherwise there wouldn't be any way to handle conflicts.20:50
SpookyET Does anyone use TextMate as a terminal editor? What are the conditions under which it refuses to exit. For example, git commit gets stuck.20:50
→ Haarg joined20:50
← Alagar left20:51
→ aresnick joined20:51
→ Alagar joined20:52
← HG` left20:52
← Alagar left20:52
→ Alagar joined20:52
→ montylounge joined20:53
→ tvw joined20:55
← Alagar left20:56
→ Alagar joined20:56
← parigaudi left20:58
← ntoll left20:58
→ Tureba joined21:00
→ bugfux joined21:00
→ corni joined21:00
→ nanooo joined21:01
← d0ugal left21:02
→ justinlilly joined21:02
→ krawek joined21:03
← krawek left21:03
→ krawek joined21:03
← rgr left21:04
← tech13 left21:04
← nanoo left21:04
→ patthoyts joined21:05
← krawek left21:09
→ multi_io joined21:10
→ schambers joined21:10
→ rgr joined21:12
← bindjp left21:13
→ dmsuperman joined21:14
dmsuperman Does git use the gnu diff tool?21:14
← yunosh left21:14
Novelocrat dmsuperman: no21:14
it has its own built-in diff enginer21:14
engine*21:14
dmsuperman Can I use it as a standalone diff tool?21:14
Outside of a git repository that is21:14
doener yep21:15
→ graywh joined21:15
→ yunosh joined21:16
→ bindjp joined21:16
dmsuperman Sweet hopefully it's a little better21:16
Damn it does the same thing21:16
← bindjp left21:16
graywh i doubt this is possible, but i want to extract 2 branches (one often merging the other) into a new repo21:17
dmsuperman Perhaps this is just something I can't really do with an automatic diff tool, but take a look at this diff (git diff does the same thing)21:17
http://paste.cjohnson.me/25121:17
← ryan[WIN] left21:17
doener graywh: two branches? That's trivial, just fetch them21:18
graywh i want to wipe out the other branches21:18
Novelocrat git branch -D21:18
git gc21:18
dmsuperman Notice how it shows the listen method as being removed, and then re-added entirely (but in a seemingly random spot in the diff output)21:18
graywh dividing this project into 2 projects21:18
Novelocrat or, git filter-branch21:18
dmsuperman listenAnswerForms is an even better example21:18
← auslander left21:19
dmsuperman Is that just something diffs everywhere demonstrate?21:19
doener dmsuperman: indentation differs, try -b or -w21:19
→ magcius joined21:19
→ bindjp joined21:20
parasti graywh: you don't have what you don't fetch, there's no need to wipe anything out21:20
→ krawek joined21:20
dmsuperman doener: That's it!21:20
doener: Thanks, these weird looking diffs have been bugging me for weeks21:21
Code reviews become so over complicated heh21:21
← dscataglini_ left21:22
→ radarcg joined21:23
graywh i want to split the history into two separate projects21:23
radarcg so I'm a noob with git, mind if I ask a noob question?21:23
← viniciusluiz left21:23
graywh but i don't want to lose the merges between two branches21:23
← montylounge left21:24
Novelocrat radarcg: just ask your question, don't debase yourself21:25
→ minaguib joined21:25
← dmsuperman left21:26
← unijunction left21:26
minaguib If I have a git commit hash, how can I tell all branches it's under ? Following the tree in gitx/gitk/git graph is near impossible21:26
RandalSchwartz with git branch21:27
← voxpelli left21:27
RandalSchwartz see --merged21:27
git branch --merged SHA21:27
minaguib RandalSchwartz: Thanks. Nice to bump into you after so many years :)21:28
radarcg uh, ok. well, I'm trying to use the git plugin for eclipse, and it seems performing commits only commits to the local repository. To actually commit the changes to the remote repository, I need to use a 'Push' command. That seems, well, a little odd. Am I understanding this correctly?21:28
Novelocrat That's exactly how it's supposed to work21:28
minaguib radarcg: Not odd at all. That's how it works. Nothing specific to eclipse.21:28
Novelocrat you are understanding what's happening correctly21:28
← joachim_ left21:29
Novelocrat The only operations that talk across the network are fetch (which pull calls) and push21:29
← remi left21:29
← graywh left21:30
radarcg ok. Is it just a limitation then of the plugin which seems to fail to remember where (url, credentials, etc) to push the changes?21:30
RandalSchwartz "commit locally, push globally" :)21:30
radarcg additionally, when I do re-indentify where to push the changes, all but the initial commit requires me to select "force update", otherwise the push is rejected. That too seems odd.21:32
→ janhouse joined21:36
→ mrchrisadams joined21:37
→ unijunction joined21:40
→ tschundeee joined21:45
← radarcg left21:46
← tobiassjosten left21:46
← Mezner left21:47
← yann left21:48
→ hecticjeff joined21:48
← hecticjeff left21:48
← RdeWilde left21:48
← khmarbaise left21:49
← rgr left21:54
← Zifer left21:56
← eletuchy left21:57
→ rgr joined21:59
← brodyberg left22:00
→ eletuchy joined22:01
→ bcardarella joined22:02
← d0k left22:03
→ nanoo joined22:03
→ graywh joined22:03
graywh after doing "git symbolic-ref HEAD refs/heads/foobar", the PWD in my zsh $PS1 is wrong (using %~)22:05
← nanooo left22:05
graywh instead of ~/path/to/repo, it's ~_git_refs_cache_pwd22:07
← graywh left22:10
← xperience left22:12
← ph^_ left22:12
→ mike31 joined22:13
→ dajero joined22:14
mike31 what is a portable way to search for a program in the search path. gmake has pathsearch macro and bsdmake exists. maybe which as alternative?22:14
I'm asking because git's makefile is rather portable and complex22:15
→ sis_ joined22:15
← sis_ left22:16
← sis left22:16
drizzd mike31: hmm, try running the command maybe?22:18
→ tildeequals joined22:18
→ joachim_ joined22:18
← froschi left22:19
← schambers left22:19
← joachim_ left22:19
→ joachim_ joined22:20
→ ProperNoun joined22:23
→ g0bl1n joined22:23
← tvw left22:23
← threeve left22:24
← hugoxrosa left22:25
→ candybar joined22:25
← tstclair left22:26
mike31 drizzd: sadly too expensive to do so as a check when calling make multiple times for example22:26
← Chillance left22:28
→ EricInBNE joined22:28
← ben_alman left22:29
← mvanholstyn left22:29
→ urbanmonk joined22:30
← urbanmonk left22:31
← justinlilly left22:31
→ urbanmonk joined22:31
→ Zifer joined22:31
← coolcat left22:31
← Zifer left22:31
→ bambule joined22:32
davr 'which' should work on *nix, mac, and msys22:33
→ coolcat joined22:33
mike31 davr: and for extended search: PATH=$PATH:additonal_pathes which COMMAND22:34
will try22:34
thiago_home the problem is that the output of which isn't standard22:35
→ timcharper joined22:36
mike31 thiago_home: isn't it always the path to the executable?22:36
thiago_home no22:36
Novelocrat how about 'type'?22:37
thiago_home actually, I've just tested 3 different Unix and it just printed the path name22:37
so which is better22:37
→ mjf joined22:37
thiago_home type is definitely not portable22:37
mike31 thiago_home: which ones and why did you think it's not in the 1st place?22:38
thiago_home I must have confused with type22:38
I tested AIX 6, HP-UXi 11.23 and Solaris 1022:38
→ revelnick joined22:39
← radarek left22:39
mike31 wow, cool22:39
I know it works on Darwin and Linux22:39
not sure about FreeBSD and the other demons22:39
→ jmut joined22:41
← jmut left22:41
→ flaguy48 joined22:43
← malumalu left22:43
→ brettgoulder joined22:44
→ Mike||gone joined22:44
→ Chillance joined22:45
→ PuffTheMagic_ joined22:45
→ sanguisdex joined22:47
← brizly left22:47
→ robin joined22:48
sanguisdex svn covert here, I just ran git rm <file> but I need to restore the file git checkout <file> is not returning any thing. how can I get it back?22:49
robin → Guest160722:49
← robinr left22:49
parasti sanguisdex: git status tells you22:50
Novelocrat sanguisdex: 'git reset <file>; git checkout <file>'22:50
Ilari sanguisdex: 'git checkout HEAD -- <file>'.22:50
Novelocrat with git rm, you staged its removal22:50
→ shennyg joined22:51
sanguisdex Novelocrat: parasti Ilari, cheers thanks22:51
Mike||gone → Mike_lifeguard22:52
→ __iron joined22:52
FFForever error: You have local changes to 'index.php'; cannot switch branches., how can I disregard the changes?22:53
Mike_lifeguard FFForever: do you want to nuke them totally? Otherwise, try git stash22:53
Bombe FFForever, you can "git checkout -f other-branch" or "git reset --hard" before changing branches.22:54
Or git stash if you want to keep the changes, yes.22:54
parasti or, git checkout -m ...22:54
FFForever I don't care about the changes22:54
shennyg so I have a goofy setup and also a no-branch problem... and I am not sure how to solve it22:54
Bombe FFForever, then I'd recommend git-checkout -f branch.22:54
shennyg I am trying to do a git-svn rebase back to the svn repo22:55
← nanoo left22:55
FFForever thanks :)22:55
shennyg and it had a conflict22:55
← _iron left22:56
← bindjp left22:56
Bombe shennyg, c22:56
→ bindjp joined22:56
Bombe shennyg, conflicts happen. Fix them. :)22:56
mike31 ok, got it working. the $(shell sh -c '') call is for porability, isn't it?22:57
jayne is there a list of large projects using git?22:57
mike31 I've been inspired by git's makefile, see uname_S definition22:58
→ killerchicken_ joined22:58
drizzd jayne: on git-scm.com and wikipedia22:58
jayne yeah, I saw the list on git.or.cz.22:58
jayne checks wikipedia22:58
killerchicken_ There was a tool showing you the status of your local branches (whether they're merged upstream, and where to). I think it was written in Ruby? I forgot the name though, and google doesn't give me anything. Any hints?22:58
← jmcantrell left22:59
drizzd killerchicken_: topgit? (random guess, probably not even ruby)22:59
← amerine left22:59
killerchicken_ hm, I think that wasn't it from the description23:00
→ jelmer joined23:01
← eletuchy left23:01
shennyg Bombe, I don't think I am fixing them properly... I edit the file then add then continue rebase and it says the same conflict.23:01
→ eletuchy joined23:01
← dajero left23:02
shennyg don't I just edit the file locally and delete the indicator lines and resave?23:02
Bombe shennyg, that can actually happen, yes. It depends on what kind of changes you’re rebasing your changes to.23:02
← Tureba left23:02
mike31 thanks guys!23:03
← mike31 left23:03
← brettgoulder left23:03
shennyg how would I get out of that loop?23:03
is there a good rebase conflict tutorial?.23:03
← Guest1607 left23:03
← sockmonk left23:03
Bombe shennyg, there can not be one because a conflict is highly depending on your situation.23:03
shennyg, you need to fix the changes for as long as it takes.23:04
shennyg ok23:04
Bombe At least I’m not aware of any way around that.23:04
shennyg I will keep on chugging23:04
thanks23:04
killerchicken_ maybe it wasn't ruby...23:04
Bombe Have fun. :)23:04
← mcdebugger left23:04
Novelocrat shennyg: there is a tool called 'rerere' that can repeat conflict resolutions23:04
shennyg thx23:05
→ brettgoulder joined23:05
→ p0w3r3d joined23:06
← TeckniX left23:07
→ scarabx joined23:07
← brettgoulder left23:08
← scarabx left23:08
← eletuchy left23:08
→ mcdebugger joined23:08
← jackdempsey left23:09
Mike_lifeguard Once you have a conflicted file, is there a tool that can go in and accept all the local/mine changes? I've been doing this by hand, and it sucks23:10
← SubMatrix left23:11
← ixti left23:12
→ Adjective joined23:13
killerchicken_ drizzd: hah23:13
drizzd: I was thinking about http://git-wt-commit.rubyforge.org/23:13
← urbanmonk left23:13
← agib left23:14
← loincloth left23:15
← joeconyers left23:15
← johnw left23:15
← ProperNoun left23:15
drizzd Mike_lifeguard: I think git checkout -2 does that23:17
→ shruggar joined23:17
→ WebDragon joined23:17
→ remi joined23:19
shruggar anyone got a page comparing benefits of submodules vs subtree merge?23:19
→ EricInBNE_ joined23:20
← EricInBNE_ left23:20
→ joeconyers joined23:21
FFForever is there anyway I can run a script on my server everytime something is pushed or send a notification out?23:22
→ d0ugal joined23:22
← bindjp left23:22
Novelocrat FFForever: post-receive hook23:22
have a look at foo.git/hooks/post-receive.sample23:22
← cannonball left23:24
→ jmcantrell joined23:24
drizzd shruggar: well, I think the most important difference is that in case of subtree merge you cannot easily merge back changes to the original subtree's repo23:24
shruggar: and submodules are cloned separately, so you don't have to clone them23:24
FFForever where do I get foo.git?23:25
drizzd shruggar: but submodules are a little dumb23:25
Novelocrat that was a placeholder, for whatever repository you're interested in notifications from23:25
← LiamH left23:26
FFForever ahhhh cool23:26
shruggar submodules being dumb was my recollection last I attempted them :/23:26
→ daaku joined23:26
← dcampano left23:26
→ kokx joined23:26
→ scarabx joined23:28
shruggar really I suppose I could solve this by writing a decent makefile and pointing it at another directory.. I'm not likely to ever want to go back a version or have project-specific patches23:29
→ ntoll joined23:30
drizzd shruggar: well, if you only ever make changes in the subtree repo and merge them back23:30
instead of modifying it in the superrepo directly23:30
then there should be no problems23:30
shruggar hm, yeah, and I guess a pre-commit hook could prevent accidental changes23:31
← sumangurung left23:31
killerchicken_ does anyone here use git-wtf?23:32
← pantsman left23:37
← tschundeee left23:37
→ eletuchy joined23:38
→ dcampano joined23:39
← sanguisdex left23:39
zodiak without wanting to start a recursive problem; wtf is git-wtf ?23:41
← Novelocrat left23:41
← imajes left23:41
← ntoll left23:41
daaku anyone know of a way to do something along the lines of: echo "my commit msg" | git svn set-tree master -e --23:42
(which doesn't work)23:42
wereHamster zodiak: doesn't the git-wtf (home/web)-page describe what it is?23:42
→ Textmode joined23:43
zodiak if I could get to any page that had the 'wtf' in it, well, I could check - however the firewall admins have labelled that as 'potentially bad' :)23:43
→ imajes joined23:44
← scarabx left23:44
wereHamster zodiak: your admins have blocked access to this website: http://git-wt-commit.rubyforge.org/ ?23:44
→ scarabx joined23:45
← yunosh left23:45
zodiak nope.. jst anything that has potential 'non work related' words and 'wtf' the made the list :)23:45
← Ryback_ left23:46
Textmode zodiak: proxy? :P23:46
daaku thanks ssh -D23:47
doener daaku: GIT_EDITOR='bash -c "cp <(echo my commit msg) $1"' git svn set-tree master -e -- # might do, only tried with "git commit"23:47
zodiak Textmode fair idea .. I was wondering what killerchicken was talking about23:47
daaku doener: ah, i shall try it23:48
← joachim_ left23:49
← mbroeker left23:56
webchick → webchick|afk23:56
← g0bl1n left23:57
→ sagsousuke joined23:58
← mrchrisadams left23:58
← sagsousuke left23:59
→ urbanmonk joined23:59

Logs Search ←Prev date Next date→ Channels Documentation