Fwd: Re: A common VFS and a Common conf-system [Part II]

[Robert Wittams]

Look, it will be perfectly possible for you to write a backend for D-VFS 
that uses fuse for sshfs, dav, or any other protocol. You can do the 
same for windows shell extensions, and OS X's in kernel fses.

That is a lot of work. And you already know it is going to suck. 
Traditional posix apis make a lot of ancient assumptions that make very 
little sense. Have you railed against the Posix AIO apis in the same way?

The fact that you can waste a lot of time bridging D-VFS to provably 
suboptimal backends does not mean we need to posixly cripple the api.

Wrt to the namespace issue:
The fact is that people *already use URis*. KDE does. Gnome does. Open 
Office does. Mozilla does.

If you could go back in time and redesign uris, it would have been nice 
to make them compatible with unix filenames : as you say, 
/remote/protocol/server/path/to/resource.

This didn't happen. URIs have won. It is *far* more realistic to adapt 
unix utilities to accept both uris and unix filenames than to remove 
uris from the face of the earth.