Re: permissions

From: Alex Riesen <raa.lkml@gmail.com>
Date: Jun 9, 2010, 12:56 UTC
Message-ID: <AANLkTik1FqUkgLYLiz1frA_G0Lws8B4tf7c6Kp3dXcCO@mail.gmail.com>
In-Reply-To: <4C0F6CF0.3020603@wpursell.net>

On Wed, Jun 9, 2010 at 12:29, William Pursell <bill.pursell@gmail.com> wrote:

> I do care quite a lot actually.  My primary goal
> was to minimize the changes, and it seemed that
> is_git_directory() was the right place to make
> the change with minimal impact.

While smaller patches are preferred, they are not the goal, per say.

> Perhaps the following patch would be more to your liking:

Looks like a lot of effort just to get a little more information in an infrequent (and frankly, obscure) failure case. How about just use errno after failed is_git_directory? It seems to make sense even when the function calls down to validate_headref. You may even reset errno to 0 (this will also make obvious the expectation of valid errno from is_git_directory).