On Wed, Apr 14, 2010 at 7:06 AM, Jonathan Nieder <jrnieder@gmail.com> wrote:
quoted text
> So here are some ideas for future work (not necessarily in order).
>
>  - Teach remaining commands that need to search for a repository to
>   use RUN_SETUP_GENTLY, with appropriate exceptions where needed
>   (for --no-index, for example).

 - Two patches "worktree setup: calculate prefix.." and "index-pack:
trust the prefix" may form a separate series. That would unblock
RUN_SETUP_GENTLY series.

quoted text
>  - Introduce unset_git_directory and the RUN_UNSETUP option (yes,
>   this needs a better name).  Teach commands that work without a
>   git directory to use it (this should fix the init poisoned by
>   parent repository and aliased init problems).

 - Two patches that move enter_repo() to setup.c and fix it up can go
first. They should have no impact on system.

quoted text
>
>  - Teach git_config() to ignore the repository-specific configuration
>   if have_run_setup is true but have_repository is false.
>
>  - Teach git_attr() to ignore .git/info/attributes if have_run_setup
>   is true but have_repository is false.
>
>  - Teach git_config() to optionally die if have_run_setup is not true
>   and the setup_git_dir* to optionally die if have_run_setup is true.
>   test-lib.sh would enable this option.
>

 - Introduce startup_info->prefix. I think that's a good change, but
it's independent to this series. The reason is, prefix can change
after setup_git_dir is called (i.e. setup_work_tree()).
 - The fix for "git cmd -h" can make a separate series (4 patches)
 - Finally the soft guard patch and assorted fixups to avoid warnings.

I'll reorder my tree and see if it looks good. Patches of this 9-patch
series will be untouched.
-- 
Duy
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html