From: Beman Dawes (bdawes_at_[hidden])
Date: 2002-09-20 11:04:49
At 04:23 PM 9/19/2002, Thomas Witt wrote:
>Hello
>
>I propose the following changes to the filesystem path interface in
>order to solve the problem of rooted/absolute paths.
After my first reply to your posting, I've been thinking more about the
problem you brought up of teasing apart a path in a portable way, and
related issues like query functions.
Right now the functions that return a portion of the path are leaf(), and
its complement, branch().
You have suggested adding root(). It seems to me the complement might also
be useful. Something like relative(), which returns a path with any root
elements removed.
I'm still not set in my mind how to handle the components of roots in the
many non-POSIX operating systems which separate the concept of
drive/device/volume/share/host/filesystem from the concept of
root-directory. I think the library may need the ability to extract those
elements as separate entities, and possibly their complements.
--Beman