Techdot Internet Relay Chat Resources, Logs, and Records.

<albacker>    if someone does a checkout(he gets foo.txt, bar.txt), removes everything, adds extra stuff (foobar.txt), commits, and than pushes, will only foobar.txt be online?
<cbreak>    online?
<cbreak>    if someone commits a deletion and some new files, the new commit will not have the deleted stuff but the new stuff
<cbreak>    obviously…
<albacker>    cbreak, “online”
<albacker>    thanks
<cbreak>    online has no meaning in git
<albacker>    i know.
<m1sc>    albacker: but it stays in the odb
<albacker>    what is the odb?
<cbreak>    history
<albacker>    cool
<albacker>    cbreak, should i say “master branch” instead of “online” ?
<cbreak>    no
<cbreak>    say what you mean, not random words
<albacker>    well obviously you can understand that i’m a git newbie… i hope.
<cbreak>    man gitglossary
<cbreak>    it explains many of the words used when talking about git
*    NapDragon is now known as Spyro
<sitaram>    yeah but that’s alphabetical; I think http://sitaramc.github.com/concepts/0-terminology.html may be better for rank newbies
<sitaram>    albacker: ^
<albacker>    thanks sitaram
*    michael_mbp_ is now known as michael_mbp
*    Anxi80_ is now known as Anxi80
<sitaram>    you’re welcome!
*    khmarbaise_ is now known as khmarbaise
<sitaram>    I remember first reading gitglossary and getting confused even more.  Like learning to speak English from a dictionary
<jast>    man/faq re-enabled
<amgarchIn9>    “git log HEAD..FETCH_HEAD| grep commit| wc -l” says there are 7 new commits. How do I advance working copy one-by-one?
<thiago>    amgarchIn9: do you really want to do it one by one? Why?
<amgarchIn9>    I have to record incremental changes in darcs
<thiago>    git merge each commit, in order
<amgarchIn9>    git branch -a doesnt show the remote branch. Am I confusing something? I did git fetch already
<thiago>    how did you run git fetch?
<amgarchIn9>    git fetch alias master, where alias is username@host:/path
<thiago>    that doesn’t save to alias/master
<thiago>    that only saves in FETCH_HEAD
<amgarchIn9>    will “git merge commit-id” do it?
<thiago>    amgarchIn9: do what?
<amgarchIn9>    when I do gitk FETCH_HEAD everything is linear, only master lable is at position -7
<thiago>    amgarchIn9: ok
<amgarchIn9>    do the merge, I learned git merge expects a *branch*, not a commit
<thiago>    git merge expects a “committish”
<thiago>    anything that can be resolved to a commit
<thiago>    a branch can be resolved to a commit
<amgarchIn9>    is it reasonable to advance master one by one? Is it doable?
<thiago>    a git user would never do that
<thiago>    but it’s doable
<amgarchIn9>    if I git checkout FETCH_HEAD~7 will it advance master label?
<thiago>    no
<thiago>    it will check out that commit
<amgarchIn9>    so the best way is to advance one-by-one is to do “git merge commit-id” 7 times?
<thiago>    yes
<thiago>    if you want something to happen 7 times, do it 7 times
<amgarchIn9>    ok, thanks. Tedious but works.
*    ChanServ gives voice to romahagen
<romahagen>    test
<romahagen>    hi guys. I would be very thankful if you help me to clarify the following issue. I’ve googled a lot but with no luck. The problem is as follwing. I have my git so called super project. there are several submodules. the project is hosted at github. i use it for syncronization of my home and office computer. the modules that I have in the super-project are some read-only git repositories for me. now, I want to add some
<romahagen>    changes to a submodule and push it somehow to github. so the time I do pull at the office it will pull the submodule ADD add my home changes ON TOP of it. how that can be done, if at all? thanks!
<romahagen>    ADD=AND*
<SethRobertson>    cd submodule; git checkout master; echo change> file; git commit -a -m change; git push; cd ..; git add submodule; git commit -m “change in submodule”
<romahagen>    SethRobertson: note, I’m not able to push into submodules, they are readonly to me.
<SethRobertson>    You can explore gitslave as a possibly simpler (but non-standard) method.  It is better for some workflows and for your home directory workcase, probably not worse
<SethRobertson>    romahagen: Then make your own github copies of the repos and repoint the superproject to get the URLs from there
<jast>    if you can’t push into the submodules, the best you could do is to create a clone of the submodule somewhere and get the other people to use your clone as a submodule instead
<romahagen>    jast, SethRobertson hm.. yes, that’s the simplest and the ugliest solutions.. that one I had also in mind, but I thought git can solve it in direct way without creating additional repos.
<Marquel>    morning.
<Marquel>    after solving a conflict, do i need to tell git i did or does git realize on its own, the conflict is resolved?
<SethRobertson>    romahagen: You could pull directly from the home/office repo and avoid the server
<tango_>    Marquel: git add the files to tell git you did it
<romahagen>    SethRobertson: unfortunately this is not possible as I cannot reach home from office.
<romahagen>    SethRobertson: wouldn’t be this a nice feature to have? As I imagine a lot of people maintain _personal_ patches to the third-party code they use.
<SethRobertson>    What feature do you think git is missing?
<SethRobertson>    If you want to pull everything into the same repo, use git subtree merge.
<SethRobertson>    If you want to pull changes directly, feel free (your inability to do so is not git’s fault)
<SethRobertson>    If you want to store the changes elsewhere, feel free.
<SethRobertson>    If you want to push the changes to the master repo, feel free (again the permission problem is not git’s fault)
<romahagen>    SethRobertson: you are of course right. sorry, I’ve studied git for two days only. I have the following use case in mind. there is a submodule in the project. the submodule repository is read-only to the developers of the project. one of them performs a change in a submodule. he works in his private repo. now he wants to push changes to some public repo from which the other developers would be able to pull. yes, it’s
<romahagen>    possible, that they pull directly from him, but pushing changes to the “central” public repo will make them explicitly visible for all.
<SethRobertson>    The local master is pretty common.  That’s why you can fork repos in github
<romahagen>    SethRobertson: so the common practice in this case is to fork and push changes to forked submodules?
<SethRobertson>    yes
<SethRobertson>    Or if the real master or local master is not on github, just create a new repo and add a remote and push to it
<romahagen>    SethRobertson: ok, thank you very much. I just thought it would be “nicer” not to have this level of indirection. but if all the others like it then I’ll also like it thanks again!
*    EugeneSleep is now known as EugeneKay
*    Spyro is now known as Spyraway
<bn_>    hi, I’m using v1.7.4.4 on Ubuntu and recently upgrade the distro from v8.10 to v9.04.  After the upgrade I noticed “git commit –interactive” no longer seemed to work complaining that it couldn’t find the Error.pm PERL module (even though I appear to have several copies of it scattered around, some even under the paths searched).  I did a “sudo cp /usr/share/perl/5.10/CPANPLUS/Error.pm /usr/share/perl/5.10/.” and that resolved that error but I no
<FauxFaux>    a) You got cut off, b) why did you upgrade to a two and a half year old version of Ubuntu?
<bn_>    FauxFaux: a)  where did it cut off at?  b) because I was on my to v10 but you have to go through v9.04 first
<bn_>    FauxFaux: b)  s/my/my way/
<SethRobertson>    and that resolved that error but I now
<SethRobertson>    Why not keep upgrading and *then* fix  any remaining problems?
<bn_>    FauxFaux:  Everything worked fine at v8.10 but someone suggested I upgrade to 10 so I could get PHP v5.3.2
<bn_>    here’s the rest of it:
<bn_>    [2/3]  … but I now seem to get “Can’t locate object method “throw” via package “Git” at /home/brian/share/perl/5.10.0/Git.pm line 321.”  Does anyone know how to resolve this?  I’ve tried re-building & re-installing git (thinking the upgrade may have overwritten git (or changed the statically linked PERL libs)) but it didn’t seem to make a difference.  I also tried to check the git FAQ but …
<bn_>    [3/3] …it appears https://git.wiki.kernel.org/index.php/GitFaq is down?  Google isn’t helping much either   Help!
<bn_>    SethRobertson: I’m not too keen on continuing to upgrade.  In fact, the original problem that gave the impetus to upgrade has now been resolved by other means (by using a PHP v5.2.X compatible syntax).  Also, upgrading would mean I will not be able to debug/work on our legacy prod env, so it may be best if I stay at v9.04 (at least for now until I get/setup another machine and have more time)
<FauxFaux>    9.04 is out of security support…
<SethRobertson>    Your problem is using Error.pm from an older perl release I fear
<SethRobertson>    Install a more recent Error package from cpan
<bn_>    SethRobertson:  I’ve heard that PERL also has multiple modules called Error.pm (see 2nd comment on http://stackoverflow.com/questions/3303641/why-do-i-get-this-module-error-cant-locate-error-pm-in-inc).  I’m assuming git needs the one in core or from CPAN?
<SethRobertson>    Error.pm isn’t in core for my, but I am running perl 5.12.3
<SethRobertson>    My Error.pm (defining throw) is from SHLOMIF Version 0.17016
<bn_>    SethRobertson:  SHLOMIF?
<SethRobertson>    The module author
<SethRobertson>    http://search.cpan.org/~shlomif/
*    Spyraway is now known as Spyro
<bn_>    SethRobertson:  It looks like that was the one I needed to install.  Once I did “sudo cpan Error” (and read everything the autoconfigure stuff did) it worked!  Thanks for guiding me in the right direction
*    ChanServ gives voice to qFox
<qFox>    is there a way to remove a branch on the server?
<SethRobertson>    git push –delete
<SethRobertson>    The server must allow it
<qFox>    does github?
<qFox>    if you have write xs ofc..
<SethRobertson>    yes
*    ChanServ gives voice to pixelmuerto
<pixelmuerto>    hi
<pixelmuerto>    I make a script for save the git remotes in .gitremotes file http://j.mp/gitRemote
<qFox>    SethRobertson, thanks
<systemclient>    is there some way to export a “diff –color-words” for someone who does not use the terminal at all?
<systemclient>    Like a PDF with colors or so?
<canton7>    systemclient, pipe it to ansi2html?
<systemclient>    canton7: that sounds like a plan
<CannedCorn>    is it possible to reference a commit in a issue comment
<CannedCorn>    i know it’s possible to reference an issue in a commit
<CannedCorn>    but can you do the reverse?
<CannedCorn>    or how would i amend my commit message to have reference the issue
*    ChanServ gives voice to cxreg
<systemclient>    canton7: Works like a charm
<datadevil>    hi. I’m trying to figure out the best way of working with the following scenario: we have several branches for projects at the same time, which at some time need to merge into master at some point. Within those projects we’d like to share subdirectories like lib and common already, but not have to merge all the other code into eachothers projects yet, in order to be able to merge them into master at different times and not having the other p
<datadevil>    I’ve looked at subtrees, but not sure if that is the best solution yet
<cbreak>    submodules
<thiago>    invert the hierarchy
<thiago>    make the libs and common as submodules of the rest
<datadevil>    mm..reading about submodules again then. Some things in pro-git made me afraid of submodules
<datadevil>    I don’t get the revert the hierarchy btw..can you explain?
*    Spyro is now known as NapDragon
<sorin>    Why doesn’t git diff –no-ext-diff –no-index <(curl “foo”) <(curl “bar”) work? error: /dev/fd/18: unsupported file type
<sorin>    regular diff supports that with complaining.
<monokrome>    Then use regular diff? :/
<thiago>    sorin: no one said git diff supported anything but regular files
<r3b00tx>    hi, i have made some changes to few files but now i dont want them and revert back to last commit. how can i do this. plz help
<j416>    r3b00tx: git reset –hard kthx
<possibilities>    i’m working on a new project where i have to change some files for development but i don’t want to include them when i do pull requests. is there some way to have a feature branch based on a local branch but based on master when pushed? or some other way git can help me?
<CareBear\>    possibilities : ignore your local files?
<r3b00tx>    j416: thanks btw one more doubt, whats the actual diff between git reset –hard HEAD and just issuing git reset –hard
<j416>    r3b00tx: no difference.
<possibilities>    i’m changing files in the repo
<r3b00tx>    j416: great, thanks buddy
<j416>    r3b00tx: HEAD is implied.
<r3b00tx>    j416: ye ye got it
<CareBear\>    possibilities : why? what are the files?
<possibilities>    one, for instance, is a Gemfile… it needs to be changed in master also but i can’t wait for that to happen before i start creating pull requests
*    NapDragon is now known as Spyro
<knittl>    hi. can i use multiple merge-strategy options?
<knittl>    i’d like to do a subtree merge with renormalization
<knittl>    ignore-space-change would be enough
<behelit>    is it possible to, per command, or w/o commit, exclude files from git-archive?
<bremner>    man gitattributes
<jast>    *automatic message* the ‘gitattributes’ manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/gitattributes.html
*    tj is now known as Guest49585
<VladGh>    does anyone have a logwatch configuration for varnish?
<VladGh>    sorry
<CareBear\>       iiiiiiiiiiiiiii
<CareBear\>    oh – sorry!
<milki>    i/o error again?
<empity>    anyone knows a good way to centralize the management of many projects?
<empity>    all the projects for example that work with some revision control systems
<empity>    which might be not always git
<empity>    for example zc.buildout+mr.developer might do what I need, but only for python projects
<txxt>    how do i “connect” an existing folder with lots of code, to an existing (and empty) repo ?
<cmn>    txxt: git add -A .; git commit -m “initial commit”; git push user@somewhere:/path/to/repo.git ? or what king of connection are you looking for?
<cmn>    s/king/kind/
<txxt>    cmn: make that folder the repo and import the existing code to the repo
<cmn>    then what I wrote is what you want
<txxt>    cmn: no new to git new repo , or something before starting to do add’s ?
<txxt>    so i need only to git push git@myrepo:repo and can add the folder(s) later?
<cmn>    you need to have at least one commit in order to push
<txxt>    ah!
<txxt>    that i didn’t know
<txxt>    tks cmn ;
<txxt>   
<cmn>    what were you trying to push then?
<txxt>    nothing
<txxt>    sorry, starting with git recently
<txxt>    from svn
<txxt>    still no adapted
<cmn>    !book
<jast>    *automatic message* ‘Pro Git’ is a complete book about git that is available online for free: http://progit.org/book/
<cmn>    that answers almost any question about what you should do if you’re starting out
<cmn>    and many more besides
<txxt>    cmn: thanks for the recommendation. was thinking about get that book soon
<txxt>    cmn: it seems i cannot “add” anything ==>  fatal: Not a git repository (or any parent up to mount parent /Users)
<cmn>    you need to create the repository then
<txxt>    that is what i was wondering before
<txxt>    if no need to create empty repo or something
<cmn>    yes, before you can do anything on a repo, you need to initialise it
<cmn>    if you already know where you’re going to push, clone that repo
<cmn>    otherwise run git init
<peterhil`>    Hi! I can’t understand why I can’t get my new branch to show up on Github.
<txxt>    the repo is empty
<txxt>    so git init i think is what i need. thanks cmn
<peterhil`>    Doing git push or git push origin just says everything is up-to-date after a while.
<peterhil`>    Should I use ‘git remote set-branches –add mybranch’ to get it show up?
<txxt>    cmn: sorry, last question, already added, now i was trying to push, but ==> Perhaps you should specify a branch such as ‘master’.
<cmn>    that will happen if there are no matching branches
<cmn>    do what it says
<txxt>    cmn: yes, i wanted. but don’t hknow hot to sepecify a branch yet
<cmn>    man git push
<jast>    *automatic message* the ‘git-push’ manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-push.html
<cmn>    and where it tells you to put the branch name, put a branch name
<txxt>    cmn: jast : not found
<cmn>    or maybe that manpage only speaks about refspecs, put it tere
<txxt>    cmn: tks. checking the man
<cmn>    you have man pages locally, though
<ShadeHawk>    Can anyone (qFox maybe) remove announcement about “Git User’s Survey 2011″ from #git channel topic?  The survey has been closed since 3 October 2011.
*    sorin is now known as Guest18706
<rslifka>    hey guys, i have a pull request against my repo, i’d like to check out the pull request locally and run tests, etc. before merging
<rslifka>    How do I get a handle to it?
<rslifka>    https://github.com/rslifka/elasticity/pull/1
<shirt>    should i name git branches with-dash or with_underscore ? or mixedCase?
<canton7>    shirt, entirely your call
<rslifka>    i.e. how do i check out the pull request?
<bob2>    rslifka, not #github
<bob2>    rslifka, you can check out the upstream branch and ‘git cherry-pick -x’ each of the revids in the pull request I guess
<osxorgate>    hi guys i have an odd problem, git can ‘see’ a file that was deleted when i look at status, but it cannot see the file when i want to remove it from the repo: http://pastebin.com/qVwaK5W2
<shirt>    canton7: what is the recommended style?
<rslifka>    bob2 – thanks Bob, I sometimes get confused where git stops and github begins
<canton7>    cherry-picking here is entirely unnecessary
<canton7>    git remote add wouter https://github.com/wouter/elasticity.git; git fetch wouter; git checkout -b wouter_master wouter/master
<canton7>    (although there is a way to do the same without adding wouter as a remote, but it escapes me)
<rslifka>    canton7: awesome man, thank you!
<bob2>    ah, didn’t notice they were contiguous
<canton7>    bob2, commits in github pull requests are always a branch
<bob2>    canton7, ahhh, thank you
<canton7>    it just looks at where the submitted branch diverges from the branch it should be merged into, and sticks those commits into the pull request
<canton7>    osxorgate, sorry you got lost in the noise. the filename’s got a space in it. git ls-files isn’t quoting the filename or escaping the space
<osxorgate>    canton7: when i run “git ls-files -d” it does have “” around the filename though
<osxorgate>    mybe the problem is the \264 ?
<canton7>    well your pastebin shows that git is trying to delete a file with a name that ends just where the space is… although there is an opening quote, it’s true
<canton7>    try a straight git rm “JSONs/Boni…..txt” ?
<osxorgate>    ok i will try
<osxorgate>    same error, pathspec did not match any files
<osxorgate>    im not too experienced in bash, but putting quotes around a filename should take care of spaces and such, right?
<canton7>    i’m not too hot on bash either, but i would have assumed so. unless that backslash is mucking things up like you suggested
<osxorgate>    i can get it to stage a remove when i use ‘ instead of \264, but then it is still listed as \264
<canton7>    git add -A is another approach
<osxorgate>    ill have to check what that does, seeing that smiley
<canton7>    add -A stages all adds, as well as all rms
<osxorgate>    well that did the trick. thanks for the help canton7
<osxorgate>    so -A is like the simple way of adding and removing whatever happened? I have this folder where files are generated and removed from time to time so i’d like to just bulk add/rm in one go
<canton7>    yeah that’s the one
<canton7>    add . adds new and stages changes; add -u stages changed only; add -A stages changed, adds new and stages removed
<osxorgate>    cool , sounds just like what i need
<cbreak>    add . doesn’t add new and stages
<cbreak>    it adds the current folder
<canton7>    sorry, unclear. it adds new and changed in the current folder?
<cbreak>    it adds every file that is still in the current folder
<osxorgate>    hm back to school for me then
<osxorgate>    i was under the impression that “git add .” adds and stages all files in all subfolders also
<cbreak>    only if they are part of the current directory…
<canton7>    all of them work just in the current directory and subfolders, though
<cbreak>    by default
<NickASloan>    we have a repo on our development server that is like the shared, main development copy, and I have my own working copy on the server. I have set up ssh to use my key, so pulling and committing to my repo works just fine
<NickASloan>    but when I try to pull the main copy, it’s using my creds, and that copy was originally pulled by another developer
<NickASloan>    how do I make this work?
<sh1mmer>    Sorry for the stupid question, but if I pull from a remote and I am up to date why is diff giving me a diffs back?
<sh1mmer>    https://gist.github.com/1274420
*    spearce_ is now known as spearce
<kevlarman>    sh1mmer: git pull won’t touch your local changes
<sh1mmer>    sure
<sh1mmer>    but I don’t think I’ve touched those files
<EugeneKay>    canton7 – git fetch <remotespec>, I believe
<canton7>    EugeneKay, nice. And just use FETCH_HEAD?
<EugeneKay>    I do believe
<canton7>    EugeneKay, cheers! :0
<canton7>    *
<ries>    hello, I am propely overseeing something, but this isn’t working : git filter-branch –tree-filter ‘changeAll.sh’ –all
<ries>    git comes back with Usage: git filter-branch……..
<ries>    did I miss something?
<ries>    I.. I had to add HEAD
<muhoo>    i have a question.
<muhoo>    i have a merge conflict
<muhoo>    i attempt to resolve the conflict by editing the file, and committing it
<muhoo>    but it won’t let me commit it because i’m “in the middle of a conflicted merge”
<muhoo>    wtf
<muhoo>     cannot do a partial commit during a merge.
<muhoo>    so i have huge list of unpushed commits to svn from git
<muhoo>    i don’t want to push all that ccrap. just a clean patch, is all i want
<bob2>    you forgot to git add
<muhoo>    git add what?
<muhoo>    what file?
<muhoo>    the one i merged? oh, hmm. lessee
<bob2>    the file you fixed the merge conflict in
<bob2>    git status
<muhoo>    ah, thanks
<bob2>    (it did tell you that as part of the HI I HIT A CONFLICT message, I believe)
<muhoo>    it said “you forgot to git add”, but it didn’t say WHAT to add, but i got that now, thanks
<muhoo>    this is the next hurdle: http://pastie.org/2668516
<muhoo>    i want to squash EVERYTHING between line 1 and line 32, thgat whole branch
<muhoo>    i tried git rebase -i but it seems to always not show that first commit
<bob2>    what?
<bob2>    why?
<muhoo>    because thos commit messages are stuopid
<muhoo>    those were just thrash on private branches
<muhoo>    experimenting, testing etc
<bob2>    meh
<muhoo>    i do NOT want to push all that garbage up
<bob2>    who cares then?
<muhoo>    just one nice clean patch
<bob2>    suspect you’re mixing things up
<muhoo>    the customer cares
<bob2>    the merge commit is “just one nice clean patch”
<muhoo>    ah, ok, cool.
<muhoo>    but doesn’t git svn dcommit push ALL the interim commits?
<muhoo>    the man page says it does, it flattens it out
<muhoo>    if this were just git, i wouldn’t worry, but git-svn says it tries to undo what i’ve just done
<bob2>    “I’m using git-svn” <- relevant
<muhoo>    cool
<bob2>    don’t know, only used git-svn long enough to migrate coworkers
<muhoo>    this is one of those things, the git  users don’t care about svn enough to knw about this, and the svn users don’t care enogh about git to know abotu it either
<muhoo>    (damn lag losing carachters)
<bob2>    well, you can use rebase to squash the whole thing if you want to
<muhoo>    i tried
<muhoo>    git rebase doesn’t show any commits i want to keep, it only shows the ones to squash, then it says that i can’t squash them all.
<muhoo>    http://pastie.org/2668531
<muhoo>    (that’s with using  the commit for rebase as the last one before the trees diverged
<muhoo>    ah, wait, i got it
<muhoo>    bob2: thanks, it looks like it worked
<ries>    Gents, how do ai remove a complete directory from my git repo? it contained copyrighted stuff I need to remove
<nb>    ries, look at man git filter-branch
<jast>    ries: *automatic message* the ‘git-filter-branch’ manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-filter-branch.html
<ries>    nb: that link doesn’t work, but I know what’s in there…
<ries>    I was reading that you need to do more then just run a rm -rf ‘my directory’ to get it really removed
<nb>    ries, sorry, i don’t remember the exact syntax, but git filter-branch will do it
<nb>    and then you’ll have to do git push –force (and have receive.denyFastforwards set to false)
<ries>    I made a little bash script that does the rm work for me
<ries>    then I run : git filter-branch –tree-filter ‘changeAll.sh’ HEAD –all
<nb>    ries, oh ok
<nb>    nice
<ries>    nb: when I do for example git checkout dbf5293867534ea5536ffa8af11eb17a471970bc
<ries>    then I d see the directory back I wanted to have removed..
*    Spyro is now known as NapDragon
*    miguet is now known as robotek`
*    robotek` is now known as group
*    group is now known as miguet
<Cowmoo>    I committed something I shouldn’t have, how do I ‘delete’ the commit and take working tree back to previous one?
<kevlarman>    Cowmoo: have you pushed the change yet?
<Cowmoo>    no, there is nowhere to push to
<Cowmoo>    git revert HEAD maybe?
<kevlarman>    that will generate a commit that undoes the change
<Cowmoo>    hm ok, not exactly what I want
<kevlarman>    git reset HEAD^ will remove the commit and leave your working tree intact
<kevlarman>    git reset –hard HEAD^ will remove the commit and make your working tree match HEAD^
<Cowmoo>    is HEAD^ same as HEAD~1?
*    cytrinox_ is now known as cytrinox
<JakeyChan>    can git show the difference between two branch ?
<bob2>    sure
<bob2>    exactly the way you’d expect – git diff whatever…whocares
<JakeyChan>    thans !
<JakeyChan>    thanks
<JakeyChan>    super easy
*    JakeyChan_ is now known as JakeyChan
<JakeyChan>    how to select merge tool ? git mergetool  tool_name ??
<JakeyChan>    git mergetool –tool=tool
<JakeyChan>    aha, I found it
<sitaram>    bob2: you used 3 dots in your “whatever…whocares”.  Clearly you don’t (care) but git will.
<sitaram>    JakeyChan: 3 dots is not what you want, you want 2 dots
<bob2>    oops
<bob2>    sorry, i mix them up
<sitaram>   
<JakeyChan>    it’s ok
<JakeyChan>    I merged
<JakeyChan>    but now it has conflict
<JakeyChan>    I need good tool, do you have any suggestion ?
<JakeyChan>    for merge tool ?
<EugeneKay>    git shutupanddowhatiwantyoustupidcli
<bob2>    http://stackoverflow.com/questions/5256249/git-diff-doesnt-show-enough/5257065#5257065 <- relevant
*    aidenhong_ is now known as aidenhong
<lorfds>    i am working with another person, and they pushed two commits to github that i want to delete completely
<lorfds>    i also have some local commits on my machine, that i then want to push to github
<lorfds>    how do i do this
<shachaf>    lorfds: Well, you *could* just push –force. Be careful, though — I’m not sure if that’s what you want.
<lorfds>    i want to remove the last two commits
<lorfds>    so how do i force those to go away?
<lorfds>    you mean if i force push, it will copy my working tree to gitthub
<lorfds>    but will that destroy all the history logged in github
<JakeyChan>    hi
<JakeyChan>    I want to remove the file that generated when merge, but it is too much do you have any quick method to delete them ?
<slonopotamus_>    JakeyChan: git clean?
<JakeyChan>    let me try it
<JakeyChan>    cool!
<JakeyChan>    git clean -f
<JakeyChan>    but use this command careful, it will delete all the file that not be tracked
*    ChanServ gives voice to ShawnMcCool
<ShawnMcCool>    my understanding of git on github is that i can fork a project, such as a development framework.  then i can write my site using that project while still being able to pull down updates to the framework directly into my project.  my understanding is that github has created the term ‘fork’ and that it’s actually more or less just branching.  i’ve done some google searching but haven’t really found a resource that makes me understand it.  what is
<ShawnMcCool>    the correct terminology for this (so that i can use it to search) and does anyone know a good resource to use to find out best practices for this?
<voot545>    is core.filemode purely a local variable, or does it ever get transferred along with a repository push/pull ?
<toralf>    I added “git remote add remote_stable git://git.kernel.org/…/table/linux-stable.git” to my local clone of Linus’ repo – does a “git pull” now automatically fetch+pull from both remote repos ?
<deanc[work]>    git checkout –track -b <local branch> <remote>/<tracked branch>
<deanc[work]>    Is it possible to track a local branch also?
<deanc[work]>    so:
<deanc[work]>    git checkout –track -b <local branch> <local tracked branch>
<Jckf>    Can I force git to use IPv4 when pulling from a repo?
<cbreak-work>    Jckf: give it an ipv4 ip address?
<rue>    (Or an alias to an ipv4)
<Jckf>    I’d prefer using the remote’s FQDN, but okay
<Jckf>    (the FQDN has AAAA records)
<cbreak-work>    well, then both should work
<Jckf>    Not the case for Assembla. They have a different fingerprint for their IPv6 SSH server
<Jckf>    So the same address has two fingerprints
<cbreak-work>    how pecular.
<Jckf>    Indeed. Someone should give them a slap about that
<cbreak-work>    you could use a different hostname for it
<cbreak-work>    like adding a dot (.) at the end
<cbreak-work>    maybe ssh will add a new key entry for it
<Jckf>    I won’t know when my machine decides to use IPv4 or 6 though
<cbreak-work>    true.
<cbreak-work>    I wonder who decides that
<cbreak-work>    it must be something inside git
<Jckf>    Or something the people in #debian would know =p
<Jckf>    I’ll go with using the IP address though. Easiest way out it seems
<cbreak-work>    Jckf: you use ssh, right?
<Jckf>    Yes
<cbreak-work>    so you could set up an alias in the config file
<cbreak-work>    (and use that instead of the raw ip)
<Jckf>    The repository’s config?
<Jckf>    No
<Jckf>    Sorry, do not know which file you’re talking about
<semarj>    Hey, I have a local repo on my machine, and I would like to push that up to github or google code.  I’ve searched for this, but must be using the wrong terms.  Can anyone tell me what command I’m looking for, or a tutorial?
<Jckf>    semarj: git remote add origin g...@github.com:caius/foo.git
<Jckf>    That’s from here http://caiustheory.com/adding-a-remote-to-existing-git-repo
<Jckf>    Which was the first hit on Google
<Jckf>    Please deposit 5 dollars for the next Jckf Google proxy search
<semarj>    well, I must not have used the right search terms, because all my results were about converting from svn to git or other things
<Jckf>    I searched for “git add origin” without quotes
<semarj>    well I had no notion of this add origin thing you speak of.
<semarj>    Thanks so much by the way
<Jckf>    No problem
<semarj>    and sorry for the preposition ending sentence
<Jckf>    Haha
<djh>    Morning.
<djh>    We have a repository. It has 89k comits, 89k annotated tags, 6k branches.
<FauxFaux>    Cool story bro.
<djh>    Doing a git clone on it is particularly slow. Where should we start looking?
<wereHamster>    djh: an annotated tag for each commit?
<djh>    FauxFaux: :p ;o)
<wereHamster>    that’s insane
<djh>    wereHamster: Okay. The reason we did it was for mapping from an old version control system. Is there a better way?
<FauxFaux>    git repack -adf the repo.
<wereHamster>    djh: notes
<wereHamster>    djh: http://progit.org/2010/08/25/notes.html
<djh>    wereHamster: Thanks. How do we get hold of the sha for a given note?
<wereHamster>    notes are attached to a commit
<djh>    FauxFaux: Thanks. Is that substantially different from git gc?
<wereHamster>    or you write the sha into the note itself
<djh>    wereHamster: Forgive my misunderstanding: I have a tag FOO.CBL_15.231 right now. So I can git co FOO.CBL_15.231 or do a diff on it, etc. How would I do the same with a note?
<crashanddie>    djh, I had roughly the same issue, but with 23k .tar.gz files that I automagically imported into git. We tried working with it for a while, but it became a very painful process very quickly. We decided to leave the repository on the server for future reference (and all devs usually clone it once), but work from a “fresh” repository (last version from previous system imported as initial release), or a few dozen most important r
<crashanddie>    eleases.
<djh>    crashanddie: This is a bit more practical, though – these are individual code files.
<djh>    Correction – these are just tags, not annotated tags.
<wereHamster>    for the clone process it doesn’t matter. They are refs, and there’s a lot of them.
<djh>    wereHamster: Okay – so whether it’s anontated or not, it’s going to be similar for clone.
<djh>    Correction again: ;o) These had -m supplied, so they’re annotated. They don’t need to be, though.
<djh>    I wonder if it’s worth de-messaging them.
<wereHamster>    no
<djh>    wereHamster: Okay – thanks.
<djh>    Nearly finished running a git repack -adf on my clone.
<djh>    And we’ll test again from that.
<djh>    wereHamster: Is it possible that the 6k branches are a bigger problem than the tags?
<wereHamster>    djh: no. both tags and branches are refs.
<djh>    wereHamster: Okie doke.
<djh>    wereHamster: I got all excited that “lightweight” tags would solve my problem (
<djh>    wereHamster: I’ll see if I can find a way to reduce the number of tags.
<wereHamster>    djh: can’t you have one file where you put the mapping?
<djh>    wereHamster: Could do. Was trying to avoid it so everything can be pure git.
<djh>    wereHamster: e.g. git diff RAOPEN.CBL_15.1712..RAOPEN.CBL_15.1899
<wereHamster>    ah, I see
<wereHamster>    you could unpack the tag map after clone though
<djh>    wereHamster: The clone isn’t our biggest problem, even.
<djh>    wereHamster: tab-completion becomes non-functional, git log takes a hell of a time, and pushing is more painful than … I’ll stop that analogy before it gets too bad (pun intended)
<wereHamster>    ah, well, there you go. I don’t see any solutino to your problem
<djh>    wereHamster: Would we have this problem with *just* 89k commits, and only a few branches and <100 tags ?
<djh>    wereHamster: In other words, is 89k tags and 6k branches *likely* to be the problem?
<djh>    Certainly for tab-completion, I would imagine.
<wereHamster>    djh: no
<wereHamster>    djh: commits don’t matter
<wereHamster>    only refs do (branches, tags)
<djh>    wereHamster: Thank you. I’d have your hamsterlings for giving that answer!
*    Sico is now known as SicoWork
<evildmp>    i did “git branch video” instead of “git checkout video” – how do I destroy this silly local branch that i have lumbered myself with?
<deanc[work]>    git branch -d <branch>
<deanc[work]>    deletes a branch
<masterkorp>    evildmp: -D if don’t have it merged
<evildmp>    thank you both
<wereHamster>    masterkorp: by definition, if he just did ‘git branch video’ then the branch is already merged
<masterkorp>    wereHamster: yeah, just giving the other possible future cases info :p
<evildmp>    if I checkout a remote branch (checkout origin/<branch> it puts my local git into a detached HEAD state – whereas what I want is that remote branch in a local branch
<evildmp>    should i create it locally, then check out the remote one?
<ngulden>    evildmp: You can create a tracked local branch for the remote one: git branch –track <local name> <remote branch>
<evildmp>    thanks ngulden
<djh>    wereHamster: One further question: On a clone of that repo I was talking about, is it possible to remove all tags and branches, perhaps?
<wereHamster>    djh: git tag/branch -d
<djh>    wereHamster: That deletes *all* branches and tags?
<djh>    wereHamster: (Any danger it might delete master too?)
<endra>    Hey
<fishguy>    hi
<endra>    How are you
*    ChanServ gives voice to lnx-hatt
<lnx-hatt>    I’m using filter-branch with subdirectory-filter, but only some of my merge commits are preserved while other merges are flattened. any ideas what can cause this?
<evildmp>    i would like my local repo to point at both my github version of it and someone else’s, so I can switch between them for testing – I have added a [remote] section to git config, with the url = <his git repo>
<evildmp>    is that the correct way to do it?
<nevyn>    evildmp: usualy you’d do it with git remote add his url
<evildmp>    ah
<g0to>    hello
<g0to>    I’m trying to set up a git repository on my server but have lots of doubts
<evildmp>    nevyn: git remote add -f –tags stefan <url> – now how do I switch to the “stefan” version?
<g0to>    I want to store git projects under /var/www/git, but that means I need root access to add or modify any file under
<deavid>    g0to: no, just change the owner/group of that folder
<bremner>    or use symbolic links
*    TheDaniel0108 is now known as Danile0018
<deavid>    for example, add a “git” user, and do “chown git /var/www/git -R”
*    Danile0018 is now known as Daniel0108
<deavid>    when doing a “git archive” command, i can tell GIT to place an extra file into the zip or tar, containing the project version as in “git describe”?
<g0to>    deavid: I don’t know if this will work when I decide to use gitweb on that directory
<g0to>    because Apache (www-data user) need to own served directory
<bremner>    g0to: no, it  just needs read access
<bremner>    g0to: a typical trick is to make everything group readable and add www-data to group git
<g0to>    bremner: good! I’m gonna play with that workaround
<bremner>    deavid: I don’t think there are hooks for git-archive. Of course you can just use tar to add a file. Maybe an alias/script to archive then add
<deavid>    bremner: and if i wanted to use zip instead? has zip any similar command as tar to append a file?
<bremner>    I’m pretty sure it does
<deavid>    ok
<nevyn>    evildmp: git remote update
<nevyn>    then git checkout stefan/master
<ndim>    deavid: For the tarballs, IIRC git archive encodes some git information into the tarball in some way.
<ndim>    Not as a tarred file, though.
<bremner>    ndim: really? I’m curious about that.
<bremner>    ah, it’s in the man page
<deavid>    i’m interested in getting version information *after* the tar or zip has been uncompressed
<rhalff>    hi, I have a very complex problem
<rhalff>    i need to revert my file to the last commit state…
<rhalff>    how to do this ?
<ndim>    deavid: I use automake  and “make dist” for those kinds of tarballs
<ndim>    rhalff: git checkout filename?
<rhalff>    ah I tried fetch and pull
<rhalff>    ndim, thanks for that vital information, I tried reset also before
<rhalff>    which I’m not sure what that did
<evildmp>    nevyn: well – i now have a remote called “stefan”, pointing at his repo on github, which contains a branch master: https://github.com/stefanfoulis/django-widgetry
<evildmp>    but git checkout stefan/master reports that “pathspec ‘stefan/master’ did not match any file(s) known to git”
<nevyn>    hrm
<nevyn>    did you run git remote update first?
<evildmp>    yes
<nevyn>    does git branch -a show stefan’s branches?
<evildmp>    I don’t think so – it only shows them at “origin”
<evildmp>    which is mine
<evildmp>    neither does git branch -r, but maybe that is expected
<evildmp>    git remote shows origin and stefan
<ndim>    Have you fetched?
<ndim>    git fetch -v –all?
<evildmp>    ndim: yes
<deavid>    what is the reverse of “git describe”? i mean, something like: git which-commit-is $(git describe “a1234″ ) -> “a1234…”
<evildmp>    it says (and said previously after git remote update): Fetching stefan
<deavid>    something that returns the commit-id from a tag, branch name, or any output from “git describe”
<b_jonas>    deavid: just take the hexadecimal number after the last “-g” from the output of git describe
<deavid>    ok, don’t mind
<deavid>    i was doing a mistake
<deavid>    i was getting an error on git log $DESCRIBED_COMMIT , but because it included something different
<fortysixandtwo>    Hi! Newbie question. I have two branches master and develop. On my server I have to seperate folders, one named develop and one production. Is it enough to simply checkout “develop” branch in my develop folder and run git pull to just update from the develop branch? And vica versa for master/production
<thiago>    fortysixandtwo: no
<thiago>    fortysixandtwo: you need to add the remote for the other repository
<fortysixandtwo>    Ah, that i have
<fortysixandtwo>    when i type git branch on server, it says master and develop
<thiago>    so you have two repositories in the server, each with two branches?
<fortysixandtwo>    yes
<fortysixandtwo>    Should I do it any different?
<thiago>    not really
<fortysixandtwo>    Ok, great.
<fortysixandtwo>    Thanks
<thiago>    it just depends on whether you wanted that and won’t get confused by having develop/master, develop/develop, master/master and master/develop
<fortysixandtwo>    There is only two branches, but the repository is cloned in two seperate folders.
<thiago>    so there are four branches
<fortysixandtwo>    But they both track the same origin branch?
<fortysixandtwo>    folder “production” tracks origin master, and origin develop, and the same for folder “develop” ?
<fortysixandtwo>    ah, sorry
<fortysixandtwo>    My mistake
<fortysixandtwo>    I see what you tried to explain to me..
*    ChanServ gives voice to freddy
<freddy>    Hi all
<freddy>    is there a place for asking some question to the git community ? It’s about thinking to some workflow, i’d like to read the opinion from other git users, but i couldn’t find some forums or whatever to ask my question
<tuxx->    ask it here, and wait for reply
<tuxx->    lots of git users here that can probably help you
<tuxx->    win 15]
<freddy>    ok, thank you for your answer so i try to organize my ideas to ask the good question :p
<ksinkar>    hi guys
<ksinkar>    i am guys
<evildmp>    ksinkar: well, everyone’s welcome here
<ksinkar>    evildmp: pardon for that typo
<ksinkar>    I am using git 1.6.0.4 and i am unable to checkout to a particular branch
<thiago>    ksinkar: why not?
<ksinkar>    thiago: i am unable to figure that out, when i do git branch -r i can see that branch in the list of branches displayed
<ksinkar>    but I am not able to checkout to that branch
<thiago>    ksinkar: why aren’t you able?
<ksinkar>    thiago: this is the erro that i get
<ksinkar>    error: pathspec ‘takeover-intellecash’ did not match any file(s) known to git.
<thiago>    ksinkar: so you typed: git checkout takeover-intellecash ?
<ksinkar>    yes
<ksinkar>    thiago: yes
<thiago>    ksinkar: what did git branch -r list?
<ksinkar>    thiago: it listed all possible branches
<thiago>    what did it say about takeover-intellecash ?
<freddy>    ksinkar, with some prefix ?
<freddy>    sorry, thiago
<thiago>    ksinkar: my guess is you forgot the prefix that git branch -r showed
<ksinkar>    origin/takeover-intellecash
<thiago>    ksinkar: you forgot the “origin/” part
<thiago>    also note that you cannot check out a remote branch
<thiago>    if you try that, you’ll check out the commit that the remote branch points to. You will be in no branch.
<freddy>    ksinkar, if you checkout a remote branch this way, that’s typically for inspecting it
<freddy>    ksinkar, you will need to create a local branch which can track your remote branch, in order to add some work in it
<masterkorp>    how can i merge only 1 commit from on brach?
<thiago>    masterkorp: is it the first commit on that branch from the point you’re at?
<masterkorp>    which happens to be the first in history line =
<FauxFaux>    masterkorp: cherry-pick
<thiago>    masterkorp: git merge thatcommithere
<arvind_khadri>    Hi, what actually happens when I do a checkout -b ? Will it affect the remote?
<thiago>    arvind_khadri: no, it will not
<Silex>    if I have a repo that wasn’t pushed but has origin correctly set, then I copy that repo and push it from somewhere else, but in the meantime the original repo adds more commits, what will happen when I do git pull on the original repo?
<arvind_khadri>    thiago, oh ok
<masterkorp>    thiago: i would do that from the master bracnh (which is the branch i want to merge the commit too ) ?
<Silex>    it’ll just be aware that origin/master points at commit X in history?
<thiago>    masterkorp: yes
<masterkorp>    thanks
<thiago>    Silex: sorry, I got lost with three repos
<thiago>    Silex: give them names A (the origin), B and C
<thiago>    Silex: then try again
<Silex>    ok
<Silex>    A is origin and is empty atm. B is the repo where you do the original import of stuffs and is ready to be pushed to A but the push didn’t happen yet. Repo C is a manual copy (cp -r) of B to some other computers. One week passes, repo C is pushed to A. Then repo B does some commits, and eventually does a “git pull origin master”
<thiago>    Silex: then let’s stop at “A is empty”
<thiago>    Silex: why are you cloning empty repositories?
<Silex>    I’m not!
<thiago>    Silex: then try again without “A is empty”
<Silex>    A: git init –bare B: git add somefiels, git remote add origin someurl
<Silex>    where someurl points at A
<thiago>    huh?
<thiago>    git init –bare B?
<Silex>    A: git init –bare
<thiago>    oh, ok
<Silex>    B: git add somefiles; git remote add origine A_URL
<solofight>    people is there a git UI tool which i can use in windows, linux and mac os
<solofight>    ?
<thiago>    ok, go on
<solofight>    apart from smart git
<Silex>    B: git commit && modify_files && git commit etc (no push yet)
<solofight>    any good tool which is more intuitive
<thiago>    Silex: ok, so far so good
<Silex>    C: cp -r B C; cd C; git push origin master
<Silex>    B: modify_files && git commit && modify_files && git pull
<Silex>    what happens to B?
<thiago>    Silex: nothing. It works properly
<Silex>    it’ll just set origin/master to the correct commit?
<thiago>    Silex: yes
<Silex>    alright. Thanks!
<Tak>    partway through a merge, can I reset a file to its just-merged state?
<Silex>    solofight: maybe emacs with magit
<Tak>    as in: git merge blah; somefile has conflicts; resolve conflicts in somefile; test; realize conflicts in somefile were resolved incorrectly
<Silex>    solofight: or vim with fugitive
<Silex>    solofight: the portability precondition reduces your choices a lot
<ksinkar>    freddy, thiago: thanx for your help. i created a new branch locally and then pulled the remote branch into it
*    evildmp_ is now known as evildmp
<Tak>    aha @ checkout –merge
*    ChanServ gives voice to dirker
<dirker>    Hi all, is there a way to setup a transparent git proxy that caches repositories locally when a user executes git-receive-pack?
<cbreak-work>    dirker: I don’t think so
<cmn>    dirker: you can use gitolite to backup or replicate repos
<cbreak-work>    unless you manage to hack the server, the client, or ssh
<JeffJohnson>    howdy
<cmn>    or you can do fetch on a cronjob
*    ChanServ gives voice to freddy
<JeffJohnson>    i create branches for work on some features, after this is rebase the branch on to the master… After this I cant use the rebased branch anymore to add more commits and rebase again with the master, the historys seem to have diverged. Is there a way to still work with the branch and rebase commits on the master(without using git push –force)?
<cbreak-work>    rebase rewrites history
<cbreak-work>    either don’t rebase, or force push the new, rewritten history
<PerlJam>    the former is best.
<canton7>    (pushing rewritten history has some bad consequences if you’re working with other people, though)
<cbreak-work>    just declare all feature branches private
<JeffJohnson>    thats the reason why i dont want to do this:=)
<canton7>    so, merge?
<JeffJohnson>    cbreak-work: declare private?
<cbreak-work>    JeffJohnson: correct!
<JeffJohnson>    cbreak-work: how? or do you mean simple name the branch private?:)
<cbreak-work>    JeffJohnson: just tell all your coworkers “If you clone this branch, be aware that I WILL rewrite it, so don’t commit on it, and don’t base your own stuff on it, unless you know what you’re doing”
<PerlJam>    JeffJohnson: why so much rebase?
<FauxFaux>    Rebase is good and awesome.
<JeffJohnson>    PerlJam: to merge multiple commits and edit commit messages
<bremner>    so are sawzalls
<bremner>    but you can cut off your foot with both
<cbreak-work>    JeffJohnson: if you want to rewrite history, live with the consequences
<JeffJohnson>    cbreak-work: mmh k, If I thing about it git push –force shouldnt be a problem, the branch is read-only for others and If i would make an force push other who cloned the repository should automatically have the new correct version of the branch after a git-pull, right?
<cbreak-work>    (and be careful to not kill your father)
<JeffJohnson>    :E
<cbreak-work>    time travel is dangerous
<rane_>    JeffJohnson: wrong
<JeffJohnson>    rane_: :/
<rane_>    unless git has changed
<rane_>    they have to pull with rebase
<cbreak-work>    JeffJohnson: they have to either reset or pull –rebase
<canton7>    JeffJohnson, there are examples of what will happen if you push –force and they pull on man git-rebase
<jast>    JeffJohnson: *automatic message* the ‘git-rebase’ manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-rebase.html
<cbreak-work>    depending on if they commit or not
<JeffJohnson>    mmh yes
<dirker>    cmn: the gitolite replication is “instant”?
<PerlJam>    JeffJohnson: I use rebase all the time for the very reasons you cited, but I never have any problems sharing because I only rewrite my local history.
<JeffJohnson>    ok, simply solution seems to be create a new branch with another name after the rebase
<cmn>    dirker: IIRC you can set it up to be instant, yes, the docs will say how
<JeffJohnson>    PerlJam: u dont push your branch to the server?
<cbreak-work>    JeffJohnson: why do other people have to access the branch?
<dirker>    cmn: ok thanks, I will check it out
<cbreak-work>    I just declare my branches private
<PerlJam>    JeffJohnson: I do push my branch.  I just don’t rebase stuff I’ve already pushed.
<cbreak-work>    so everyone who uses them knows
<JeffJohnson>    cbreak-work: they dont have to, but I want to have an backup for the branch and need to clone and work on it on other machines
<solofight>    Silex, emacs, vim are becoming base installers and git is depending on them here. Any other way to eliminate that ?
<solofight>    any stand alone installations ?
<cbreak-work>    JeffJohnson: doesn’t sound like a problem
<JeffJohnson>    PerlJam: after you rebased your branch with the master you will create a new branch and delete the old?
<PerlJam>    JeffJohnson: no.
<cbreak-work>    as long as all who use it know they have to pull –rebase
<solofight>    Silex, or any package which bundles and delivers these as a single executable which less computer literate users can utilize to install ?
<cbreak-work>    why delete a branch? that sounds dumb
<solofight>    Silex, sorry for late reply
<JeffJohnson>    PerlJam: than whats the trick to still work on the branch?
<PerlJam>    cbreak-work: that’s what svn does
<cmn>    solofight: what platform are you on?
<PerlJam>    JeffJohnson: see that’s the thing .. there is no trick.
<cbreak-work>    PerlJam: I rest my case
<JeffJohnson>    PerlJam: after i do this i have a diverged history, and another rebase on to the master from this branch have weird results
<solofight>    cmn, work in both windows and linux regularly and rarely on mac
<PerlJam>    JeffJohnson: just don’t rewrite history you’ve already shared.
<solofight>    cmn, in windows itself work both in xp and win 7
<JeffJohnson>    PerlJam: how do you apply the master version to the branch?
<Silex>    solofight: what? I don’t understand your question
<PerlJam>    JeffJohnson: or, if you really must share “incomplete” work, do as cbreak-work said.
<cmn>    solofight: I mean where you have your problem; on windows, you need to get the mingw/msys environment, there isn’t really a way around it
<JeffJohnson>    PerlJam: or didnt you update the branch to the master revision?
<cbreak-work>    what I do is not for sharing
<cbreak-work>    it’s for backup
<cbreak-work>    we have the master branch, and private/feature branches
<cbreak-work>    only the person who makes the feature branch is supposed to work with it
<cbreak-work>    so they are essentially private
<cbreak-work>    you have to force push to the server if you want to update them
<JeffJohnson>    PerlJam: how do you keep your branch after the rebase uptodate with the master?
<PerlJam>    apply the master version to the branch” for one thing.  I use merge.
<PerlJam>    oops
<solofight>    Silex, you said we can use magit and fugitive right, these are for emacs and vim. my question is do we have a standalone whichreally not depending on other packages like vim or emacs
<cmn>    solofight: on any Linux distro, the git or git-core packages should provide the base packages
<PerlJam>    JeffJohnson: I don’t use rebase to “apply the master version to the branch” for one thing.  I use merge.
<cbreak-work>    JeffJohnson: just rebase onto master again and again
<JeffJohnson>    PerlJam: yes ok
<cbreak-work>    until you’re ready to integrate it
<Silex>    solofight: no
<cbreak-work>    then fast forward merge it into master (or non-fast-forward if you like)
<solofight>    cmn, the problem is different UI in different environment. so was wondering is there a UI tool which can work on all 3 OS
<cbreak-work>    after it’s integrated, I usually delete it
<cbreak-work>    because then it’s no longer useful
<Silex>    solofight: gitk?
<Silex>    or git gui
<JeffJohnson>    cbreak-work: after rebasing my branch to the master, I also want to have all other master commits in the branch, that seems to be the problem
<cmn>    solofight: as Silex says, gitk should run on almost anything (though I’m not sure about Windows) and kdiff3 should be portable to any platform because of Qt
<Silex>    it runs on windows
<cmn>    good
<PerlJam>    JeffJohnson: Why don’t you use merge?
<JeffJohnson>    PerlJam: when? To update the branch?
<cbreak-work>    JeffJohnson: nah
<cbreak-work>    that’s no problem
<JeffJohnson>    its my problem:P
<cbreak-work>    after rebasing onto master, you have all master commits
<PerlJam>    JeffJohnson: for anything appaerently 
<cbreak-work>    after rebasing onto master, all of master is your branch’s ancestor
<cbreak-work>    so you have all its commits too
<JeffJohnson>    cbreak-work: mmh ok, I missed that, than i have no problem:)
<JeffJohnson>    thx
*    NapDragon is now known as Spyro
<freddy>    so, sorry for my poor skills in english :
<iivvoo>    a colleague (web designer) has messed up my branch, doing some strange merges and reverts
<iivvoo>    I know commit xyz is ok. can I git reset xyz to fix things?
<solofight>    thank you Silex cmn
<lnx-hatt>    iivvoo: git reset –hard xyz
<FauxFaux>    iivvoo: Sure.  You might want to tag the old stuff first so it’s easier to get back if yuo want it.
<freddy>    In my company, we are working on 4-5 differents websites which are made of many modules. We are in charge of these modules, and each module is managed by maven. We have some maven projects wich represent the differents website and set up the dependencies for each website. (SiteA needs moduleA in its version 2.3, moduleB in its version 3.1, etc)
*    iivvoo seems to finally grasp git (and reset)
<freddy>    For the moment, everything is versionned with subversion, in a single repository which contains many subdirectories, one per module. The real problem that we have with this architecture is to manage different versions of each module for each website on which we can deploy them. For exemple, WebsiteA needs the version 2.3 of moduleA, but WebsiteB needs the version 3.4 …
<freddy>    It’s a bit complicated to make them consistant. As a git user, i’m trying to think about a workflow that would allow us to “think different” and manage all our constraints in the best way.
<freddy>    (sorry again for the english…)
<iivvoo>    and everything will work with other remote tracking branches if I push things back?
<rue>    You may stop apologising now
<canton7>    freddy, that, for once, actually sounds like the exact scenario for which git submodules were developed
<canton7>    freddy, if each submodule is its own git repo, with whatever branches/tags you need to sort out the versioning
<canton7>    freddy, then each project can require, as git submodules, whatever versions of whatever submodules it needs
<freddy>    canton7, yes, i’ve heard about submodules. So I bought the book “Pro Git” in order to understand well how to use them, but i didn’t understand the way for keeping the “super repo” up to date
<freddy>    therefore, i was thinking that was perhaps the wrong way to do
<canton7>    what exactly do you mean by “up to date”?
<freddy>    canton7, “aware” of its children evolutions
<canton7>    freddy, that isn’t done automatically, but you can update all submodules with a single command
<iivvoo>    hmm, I need to pull before I can push, and pulling is redoing the fauly commits again
<freddy>    so each developper have to share his work in submodules, then afterthat, to share the updated super-repo ?
<canton7>    yeah. it’s a slight pain, granted, but it means that the supermodule can guarantee exactly what commit (and therefore version) each of its submodules is at
<iivvoo>    I guess I need rebase
<freddy>    canton7, ok i see… What if a developper decide to checkout a branch in a submodule and update the supermodule after that ? It will refer to the branch head of the submodule ? Can’t it be confusing ?
<canton7>    freddy, the supermodule only tracks the commit that the submodule is at. if the submodule was checked out at af23d45, and the developed checkes out master (of which the tip is af23d45), the supermodule won’t notice any difference
<freddy>    canton7, yes i understand. Thank you for your answer. It’s a little complicated but I’ll continue to investigate that way
<freddy>    canton7, so you advise me to make one super-repo for each website ?
<canton7>    freddy, without knowing more that’s what I’d suggest, yeah
<freddy>    canton7, ok, thanks a lot
<canton7>    freddy, git submodules *are* more confusion that svn externals. But with that added complexity, you get more power at the end of the day. People often don’t need that power, which is why submodules get the label of being overly-confusing, but it sounds like you do
<freddy>    canton7, ok I see. Does a supermodule (which is made of many submodules) have its own history ?
<canton7>    freddy, sure. a supermodule is just a normal repo. the key point is that in addition to tracking files and folders, it can also track submodules
<iivvoo>    this is harder than I thought
<iivvoo>    I’m working on a remote tracking branch, revering locally doesn’t fix anything
<freddy>    canton7, so an update in a submodule adds an entry to the supermodule history ?
<iivvoo>    simplest solution would probably to checkout the known-good commit, branch from it and develop from there
<iivvoo>    but it would be cleaner to somehow reset the original branch
<canton7>    freddy, an update in a submodule will show up as an unstaged change in the supmodule, just as if you’d changed the contents of a file. you can then choose to add and commit that change, whereupon the supermodule will update its record of what commit the submodule should be at
<freddy>    canton7, wow sounds great, I love this idea
<canton7>    freddy, it does make sense, doens’t it? P
<canton7>    *
<canton7>    yes, it’s a pain if you need to change the submodule, as you also need to change the supermodule… but you get that added level of control, which lets you decide “actually no, i don’t want this particular supermodule to point to my new commit in the submodule”
<freddy>    canton7, yes, the concept is pleasant, i’m not sure i will be able to use it the right way, but i will try
<canton7>    freddy, it’s worth setting up a few test repos and experimenting before trying to do anything with production code
<freddy>    canton7, yes, that is what i’m going to do. Thanks again for your explanation
<canton7>    freddy ok cool, no worries
<canton7>    it’s worth re-reading the pro git stuff on submodules
<freddy>    oui
<canton7>    but give us a shout if you hit any big problems. submodules are a pretty common question
<freddy>    yes*
<iivvoo>    anyone? Please? How can I get my reverted branch in sync with the remote branch?
<iivvoo>    eh.. reset branch
<freddy>    ok, i won’t hesitate, thank you
<canton7>    iivvoo, did you revert or reset?
<iivvoo>    cantoma, reset
<iivvoo>    reset –hard <good-commit>
<iivvoo>    but now I’m behind the remote branch and need to pull to get back in sync, basically undoing the reset
<canton7>    iivvoo, then you’ve got two choices. you can either push –force, which will be rewriting public history, and everyone else tracking the repo will have to reset that branch. or you can revert instead of reset (revert creates a commit which undoes the changes from antoher commit)
<iivvoo>    revert sounds like what I need then
<canton7>    reset yoruself back to where you were before the first commit (reset –hard origin/master, for example), then do the revert
<canton7>    *before the first RESET, even
<iivvoo>    reset + revert?
<iivvoo>    I don’t really understand it – to what should I reset
<canton7>    the reset is to undo the first reset you did, so it’s as if you never did it. then the revert is to do effectively what the reset did, but in a more history-friendly way
<iivvoo>    right.. but pulling will also undo the reset
<canton7>    true in this case, yes
<canton7>    good point
<canton7>    so just pull
<canton7>    then revert
<iivvoo>    ok, makes sense
<iivvoo>    thanks
<sitaram>    dirker: instead of caching a repo locally, turn it around.  Tell the user to push to the local (server) and in it you place a couple of lines as ‘post-receive’ that does ‘git push –mirror …’ to your second server.  Done.  (This is how gitolite mirroring works but you don’t need all of gitolite if this is all you want to do)
*    ontologiae is now known as Montaigne
<dirker>    sitaram: thanks, so basically all repositories in the master have a post-receive hook that immediately pushes to the mirrors
*    Montaigne is now known as ontologiae
*    ChanServ gives voice to ontologiae
<ontologiae>    hello everyone
<ontologiae>    sorry fo the traditionnal beginner question, i just would like to create a central repo on a server. It’s for a empty repo
<ontologiae>    and all the tuorial i try failed
<iivvoo>    canton7 things seem to be ok again, many thanks for the help
<m1sc>    ontologiae: git init –bare ?
<Silex>    a collaborator has to push a big repository (700Mb) over ssh, and it failed. We don’t have the error message so no clue… is this a known problem?
<Silex>    is all we can do to ask to retry and give us the error message?
<ontologiae>    m1sc : i made a git init –bare firstly on the server, i made a git config file for xinetd (9418 port), and when i’m trying to clone it don’t find it : “does not appear to be a git repository”
<canton7>    iivvoo, ok sweet!
<canton7>    Silex, stick GIT_TRACK=1 in front of the command
<cmn>    isn’t it GIT_TRACE?
<canton7>    yeah, typo
<canton7>    sorry
<FauxFaux>    Keys are right next to each other.
<canton7>    my fingers have a mind of their own, and just go for a nearby word either than nearby keys
<canton7>    *rather than
<canton7>    ^^ as demostrated
<lnx-hatt>    heh
<FauxFaux>    Haha.
*    FauxFaux does that, would like to see a good description of what’s actually going on inside my haed.
<canton7>    agreed! I’m tempted to learn Cokemak, and see if re-training my fingers solves the problem
<canton7>    *Colemak. Argh this is getting annoying
*    FauxFaux is using colemak right now, on an unfamiliar keyboard, however, as my nice keyboard died this morning and I don’t keep spares in the office.
<canton7>   
<Silex>    canton7: cool, thanks for the tip
<cmn>    Silex: if all else fails, you could ask him to send you a bundle over FTP or something else that’s resumable
<bgy>    hi
<bgy>    How could remove a file from a previous commit?
<FauxFaux>    Fix it then git commit –amend.
<shruggar>    bgy: describe your situation in more detail, please
<bgy>    I created a branch, mobile_device, I commited some changes, some files, I’ve a total of 3 commits, in the first commit I added a file Version123, and in my last commit I added a file Version456 I want to rebase all thoses commit before rebasing master, and I’d like to remove the Version123 file which I don’t want to be in my repository
<sitaram>    dirker: something like that.  If you’re asking about *gitolite* mirroring it’s a lot more complex than that but at the bottom of it all is still that same basic idea
<ontologiae>    and gitolite just remove gl-easy-install
<bgy>    shruggar: FauxFaux: well I got rebase + edit and amend my commit worked well, thanks anyway
<ontologiae>    “sudo git-daemon –base-path=/home/git –verbose
<ontologiae>    fatal: unable to allocate any listen sockets on port 9418″
<ontologiae>    is it normal ?
*    wyang is now known as Guest87267
<FauxFaux>    ontologiae: a) Why the hell are you running that as root?  b) netstat -an | grep 9418
<ontologiae>    FauxFaux: ok it’s a mistake, i discover sudo sv stop git-daemon works, but i dunno what is it
<bremner>    sigh. It’s a foolishness with the Debian/Ubuntu git-daemon-run package
<bremner>    it uses runit, which is a replacement for the standard init system
<wereHamster>    why don’t they use upstart (in ubuntu)?
<bremner>    did I mention I thought using runit for one package was foolish?
<drizztbsd>    runit is not a replacement
<bremner>    then wtf does it edit /etc/inittab for?
*    lnx-hatt is now known as lnx
*    bremner is still slightly scarred from his runit experience
<kaimi>    is there a way to push to a git repo via https/gitweb like with hg and hgweb?
<bremner>    gitweb no, https, yes.
<bremner>    but I don’t know how well it works
<kaimi>    bremner, where can i find documentation for this?
<bremner>    kaimi: http://progit.org/2010/03/04/smart-http.html
<kaimi>    thanks
<Kaapa>    hey there
<Kaapa>    when we do a git tag, does it automatically “pushes” that info to remotes?
*    tj is now known as Guest26405
<Kaapa>    cause after tagging, I do a git push and get a “everything up to date”
<milki>    no it doesnt
<Kaapa>    how do I push it?
<milki>    well
<milki>    it could
<milki>    –tags?
<Kaapa>    ah!
<Kaapa>    perfect, thanks
<milki>    np
<thiago>    Kaapa: depends on where the tag is
<thiago>    Kaapa: tags are pushed along with the commits that the tag points to
<thiago>    Kaapa: if you tagged a commit that the remote already has, the tag won’t be pushed
<Kaapa>    thiago: yeah, guess it was the case
<thiago>    Kaapa: if you tagged a commit that is outside the branches you’re pushing, it won’t go either
<Kaapa>    cause I just sent 2 tags now
<leeping>    Hi there – is it a good or bad idea to set up a git repository in my home directory to keep track of a large number of files – not specifically for a software package, but just a collection of scripts and files that I’d like to keep synchronized?
<leeping>    for instance, because I have access to many computers and want to keep a uniform set of commands, I’d like to have a git repository on my workstation and simply type “git pull” on the workstation to update all of my scripts
<leeping>    I am also curious, if I set up two git repositories, one in my home directory and one in a subdirectory, does this lead to strange conflicts?  is this ill-advised or does it not matter?
<m1sc>    leeping: you shouldn’t nest repos. a lot of people keep their dotfiles/scripts/.. in git repos, but probably it’s better not to make your home dir a repo.
<nkuttler>    leeping: i just have a ~/dev/system/ that contains my stuff. run make install, done. different hosts need different files
<yrlnry>    nesting repos has always worked okay for me.  Whats the concern?
<m1sc>    yrlnry: you ignored the subrepos?
<yrlnry>    I think Git ignored them automatically.
<yrlnry>    It seemed to me that if you had a dir/.git directory in your working tree, Git would just ignore dir/ completely, and not look in there for changes.
<leeping>    yrlnry, interesting .. so if I have $HOME/.git and $HOME/sub_dir/.git, the git repository in $HOME will ignore the changes in $sub_dir?
<yrlnry>    I think that’s what happens.  That’s what has seemed to happen for me.
<leeping>    Sorry my computer crashed .. if you guys answered my question, can you paste it? if not, no worries.
<leeping>    Ideally, I could have two git repositories, one in $HOME and one in $HOME/sub_dir, and I would be able to control their files independently.  But I don’t know if this is a reasonable or logical thing to ask for
<bremner>    leeping: there are lots of setups for managing dotfiles using symlinks or detached work trees.
<cbreak>    leeping: you can have as many git repositories as you wan
<cbreak>    twant
<EugeneKay>    Almost all of them end in pain and suffering
<shirt>    leeping: btw, channel logs in topic
<cbreak>    and unless you bind them together with mechanics such as submodules, they are completely independent
<mrtazz>    leeping: I usually just manage specific subfolders (.zsh or .vim) in git and symlink the toplevel rc files, which works quite good for me
<mrtazz>    leeping: of course it’s a bit more to set up than just `git pull`
<leeping>    mrtazz, I see what you mean. that is very helpful.
<leeping>    shirt, thanks!
*    ries_ is now known as ries
*    Praise- is now known as Praise
<j4cbo>    is there an option to gitk that will make it show *all* commits, even those not on any branch?
<j4cbo>    gitk –all shows all branches, but not everything
<FauxFaux>    j4cbo: gka() { gitk –all $(git log -g –format=”%h” -50) “$@”; }; gka
<davegb3>    can i change the date of the last commit – commit –amend and rebase -i don’t seem to have any affect on the date.  I’d like the date to be “now”.
<FauxFaux>    j4cbo: Or with git fsck –unreachable or whatever the flag is.
*    |davegb3| is now known as davegb3
<EugeneKay>    davegb3 – revert back, commit again?
<EugeneKay>    Er, reset*
<davegb3>    EugeneKay: ah, ok.  do you happen to know which form of reset I’d need?
<EugeneKay>     soft
<davegb3>    super, thank you!
<EugeneKay>    Though if you have changed files, I believe it’d need to be –mixed
<EugeneKay>    Consult the man page
<davegb3>    nope, everything is committed, it’s just the date that’s wrong (I’ve squashed a lot of commits into one, and I’d like to the date to be now, rather than the date of the first commit)
<davegb3>    I’m reading the man page now – many thanks.
<EugeneKay>    Well, there you go then.
<davegb3>    EugeneKay: that worked like a charm. And now i understand what reset –soft is for   thanks again.
<EugeneKay>    Works good for stashing-without-using-stash
<cbreak>    stop spamming Liquid-Silence
<canton7>    missed him
<ningia>    hi all
<canton7>    hi
<EugeneKay>    Hola
*    verto is now known as wyrd
<LeoJweda>    hi, i know what origin in “git push origin master” is, but what’s master? is it the branch on my local repo that’s being pushed or is that the target branch in the remote repo?
<kevlarman>    LeoJweda: both
<kevlarman>    sort of
<LeoJweda>    kevlarman: well, I’m starting a new project and I’m planning to use this model for my branching: http://nvie.com/posts/a-successful-git-branching-model/
<LeoJweda>    kevlarman: basically, you develop in a “develop” branch then merge with master when you’re ready to deploy
<kevlarman>    if master isn’t a tracking branch, then it’s used as both the local and remote branch name, otherwise it’ll use branch.master.merge instead
<LeoJweda>    so I’m ready for my first push, how do I push to develop (not master)?
<kevlarman>    git push -u origin develop
<kevlarman>    (-u turns it into a tracking branch so you won’t have to specify the branch name/remote explicitly later on)
<LeoJweda>    kevlarman: that’s what I thought What about the local repo?
<kevlarman>    what about it?
<LeoJweda>    kevlarman: so I’ll basically be pushing from my local master to the remote develop?
<kevlarman>    no
<kevlarman>    from your local develop to remote
<LeoJweda>    kevlarman: so it switches my local branch to develop too?
<kevlarman>    no
<LeoJweda>    kevlarman: well, right now all i did was an init and an add ., am i on master locally now?
<kevlarman>    yes
<canton7>    kevlarman, git push doesn’t use any tracking information by default, see https://gist.github.com/1053846
<canton7>    (even if tracking info exists)
<LeoJweda>    kevlarman: so what happens when I do git push origin develop since I don’t have a local develop? and is there a way to switch my local branch to develop too (if that’s what i need to do)
<canton7>    LeoJweda, you might like github.com/nvie/gitflow
<kevlarman>    LeoJweda: it won’t let you
<kevlarman>    you’ll want to create the branch locally first
<rue>    (Activity Monitor sampler – no need to even GDB!)
<rue>    Sorry, mischan.
<canton7>    wrong window?
<canton7>    heh
<LeoJweda>    canton7: Thank you so much!
<LeoJweda>    kevlarman: thanks for answering my questions, i now know what the parameters for git push are
<canton7>    LeoJweda, it’s no replacement to knowing what’s going on under the hood, but it makes life easier by reducing typing
<LeoJweda>    canton7: I understand the article completely, i’m just having trouble with the initial setup. i’ll give gitflow a shot
<kevlarman>    annoying thing about git is that easy stuff seems harder than it should be at first
<kevlarman>    awesome thing about git is that stuff that seems like it should be hard is as easy as the easy stuff once you know how it actually works
*    ries_ is now known as ries
<Baribal>    Hi. I used the instructions from Git Pro (http://progit.org/book/ch4-2.html) to create a blank ssh-accessible repository on a server. I also created a local repository and committed some code. I thought that now I could simply ‘git push ssh://username@host/~/my/project/directory.git/’, but this is what I get: http://paste2.org/p/1701301      I’m very n00b when it comes to git, but isn’t master the branch to which everything gets committed
<Baribal>    as long as I don’t branch? Or does the error message refer to a local branch? In which case, shouldn’t it default to the branch I’m on (which, as I haven’t checkout’ed anything, should be master, I guess)? What am I missing?
<sadmac>    Baribal: I think you always have to specify a target branch for push.
<sadmac>    Baribal: git push ssh://… master
<jakobm>    Baribal: there is no matching branch (i.e. a branch with the same name as your currently checked out) on the remote, you have to explicitly specify it as suggested in the error message
<SamB>    Baribal: try “git branch” to see what you’re actually on
*    naee is now known as eean
<Baribal>    Yup, it’s master, but it seems I really had to specify it explicitly. Thanks.
<canton7>    Baribal, https://gist.github.com/1053846 might be interesting
<Baribal>    jakobm, so… master gets created on the first commit, was that the problem?
<jakobm>    in this case, yes, the gist canton7 linked explains the different default-pushing cases
<canton7>    (and the confusing default, which definitely isn’t what people expect)
*    dwmw2_gone is now known as dwmw2
*    _psychic__ is now known as _psychic_
*    d0k_ is now known as d0k
<arnpro>    after I set up git serverand try to clone the gitosis-admin.git repo it keeps asking or a password, but I never enabled 1,in fact I did adduser with –disable-password param
<kevlarman>    arnpro: 1) gitosis is deprecated, 2) that means your key isn’t set up properly
<m1sc>    arnpro: !gitosis
<jast>    arnpro: *automatic message* gitosis is no longer maintained and supported by the author; we usually recommend gitolite instead which has much better documentation and more features: http://github.com/sitaramc/gitolite
<arnpro>    oooo, that’s why then… how can I get rid of it?
<arnpro>    heh I have tried so many things to get my git server to work, without success
<arnpro>    kevlarman: when Im making my client rsa key, when trying to scp it, it keeps asking for a password I’m using this: scp ~/.ssh/git.pub git@hostname:git.pub
<kevlarman>    arnpro: that’s not how you do it
<arnpro>    kevlarman: may you tell me how? I’m afraid I might be following the wrong tutorials…
<kevlarman>    arnpro: you add the key to your gitosis-admin
<kevlarman>    or gl-admin or w/e
<kevlarman>    (don’t use gitosis)
<arnpro>    I removed gitosis
<arnpro>    which key are you referring to? my server’s or my client’s?
<kevlarman>    your client’s
<arnpro>    ok, I read you firstneed to generate it, then scp it to the server, correct?