Re: [PATCH 3/3] completion: Prevents GIT_PS1_DIRTYSTATE from breaking when CWD is .git

[Ted Pavlic <ted@tedpavlic.com>]

>> This patch uses "git rev-parse --is-inside-work-tree" to determine
>> whether a "--work-tree=.." should be added to the "git diff".
>
> Why ".."?  What prevents you from "cd .git/refs/heads"?

It was my impression that --work-tree=.. was relative to the GIT_DIR. My 
tests seem to confirm that. Within the git work tree I do:

	echo "test">>COPYING

then

	cd .git/refs/heads
	git --work=tree=.. diff

and that produces a diff of COPYING showing the new line.

("man git" confused me about how WORK_TREE was supposed to work)

> Your "is-inside-work-tree" might be a good change, but if you were to
> spend a letter to notify the users, "Warning: You are inside GIT_DIR! This
> is something unusual, proceed with caution." is a lot more important
> notice to give them than "You seem to have unstaged changes" notice.

Ok, so if "is-inside-git-dir" then send a special "!" flag (and no 
dirtyState flags)? Does that seem reasonable?

Additionally, is it a good idea to echo the branch name when inside the 
git dir? That is, what does it "mean" to be on "master" when you're 
inside .git?

> You have at least three possible states:
>   * You are not in git repository at all;
>   * You are somewhere in $GIT_DIR, perhaps in a bare repository, perhaps a
>     repository with a work tree.
>   * You are inside a work tree.

It seems like (psuedocode)...

	if git rev-parse --is-inside-git-dir; then
		use '!' flag to indicate caution

	elif git rev-parse --is-inside-work-tree; then
		proceed as before (with '*' and '+' flags)

	else
		do nothing

I think that handles those cases. No?

Thanks --
--Ted

-- 
Ted Pavlic <ted@tedpavlic.com>

   Please visit my ALS association page:
         http://web.alsa.org/goto/tedpavlic
   My family appreciates your support in the fight to defeat ALS.