Re: [PATCH 3/5] Clean up work-tree handling

[Johannes Schindelin <Johannes.Schindelin@gmx.de>]

Hi,

On Fri, 27 Jul 2007, Junio C Hamano wrote:

> Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:
> 
> > But there is a fundamental question I have to ask: Is there any reason why 
> >
> > 	$ git --git-dir=/some/where/else.git bla
> >
> > should pretend that the repo is bare if core.bare == 1?  I mean, we are 
> > implicitely setting the work tree to the cwd, no?
> 
> I have two repositories at the primary k.org machine.
> 
>  - /home/junio/git.git --- this is with a worktree so that I can
>    build and test on a FC machine (my primary development
>    machine at home is a Debian).
> 
>  - /pub/scm/git/git.git/ --- this is a bare repository that is
>    mirrored out to git://git.kernel.org/ and friends.
> 
> And I usually am in the former.  From time to time, I do this:
> 
>  $ GIT_DIR=/pub/scm/git/git.git/ git fsck
>  $ GIT_DIR=/pub/scm/git/git.git/ git repack
> 
> because I am old fashioned, but I would expect these to be
> equivalent to the above:
> 
>  $ git --git-dir=/pub/scm/git/git.git/ fsck
>  $ git --git-dir=/pub/scm/git/git.git/ repack
> 
> I do not think these imply that the repository is with worktree.

But in your use cases it does not matter, since neither fsck nor repack 
need a worktree.

Here is one of _my_ scenarios: I want to track a directory I have no write 
access to (or better put: I should have no write access to).  So I created 
a bare repository, and when I add new files I use the "GIT_DIR=... git..." 
mantra.

But you're right, I could just set core.worktree=... and unset core.bare.

I can change that easily enough.  Other comments on the series?

Ciao,
Dscho