From: noer AT cygnus DOT com (Geoffrey Noer)
Subject: mount() semantics
21 Sep 1998 19:18:32 -0700
Message-ID: <19980921191324.55118.cygnus.cygwin32.developers@cygnus.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
To: cygwin32-developers AT cygnus DOT com

Consider:   mount Q:/baz /foo/bar

Under Unix, /foo/bar would have to be a valid directory.  Under
Cygwin32, it does not.  Even worse, /foo doesn't even have to exist so
if you cd into /foo/bar, and then try to cd .., an error results.
I'm pretty convinced something needs to change.

I am tempted to suggest requiring that /foo/bar exist as of b20,
modifying mount() in winsup accordingly.  That would have the
advantage of making mount()'s semantics much closer to Unix.  But it
would also have the disadvantage of requiring physical directory mount
points which could be a bummer to have to cope with.

If you all think that would be too drastic, we could at least check
for the parent directory's existance (although that's a bit trickier)...

Thoughts?

-- 
Geoffrey Noer
noer AT cygnus DOT com