system and desktop VFS merged

Sean Middleditch elanthis at
Mon Mar 28 03:52:08 EEST 2005

On Sun, 2005-03-27 at 13:09 -0800, John Davidorff Pell wrote:
> On 27 Mar 2005, at 11:59, Sean Middleditch wrote:
> > If people are going to implement system-level hacks to access D-VFS, I
> > really do think the absolute best bet is to make open() recognize a  
> > URI
> > and start talking to the D-VFS daemon instead of the kernel.  Better
> > performance, better compatibility with other D-VFS apps, and easier to
> > port to other systems.  There are some problems with such an approach,
> > but they're things I'm quite sure the Free Software community can
> > tackle.  You're all a bright bunch.  ;-)
> I don't think that Rosen meant any such thing. No system-level hacks,  
> no messing with open(). What I understood he meant was that the  
> shares, or those that are easily mountable (no need to do any special  
> trickery to mount, for example, an sftp connection), are mounted  
> under /vfs. This means that D-VFS aware apps would continue to use  
> the D-VFS library to communicate with the daemon, but it now allows  
> non-D-VFS aware apps to access the shares as well, though their mount  
> points. If a mount point is not available, then I'll live. :-)

The problem is, none of the shares are easy mountable.

The few remote file systems commonly supported by Linux, BSD, etc have
absolutely no way to cleanly integrate with D-VFS.  There is no safe,
secure way to hand them a password for authentication.  There is no good
way to handle user sessions.  There is no good way to provide cookies or
other special tokens that some protocols need.  The list goes on.  If
the access doesn't happen through the daemon, it won't be capable of
working with D-VFS properly.  The only way to get legacy apps to work
with D-VFS is to get them to actually use D-VFS.  That requires either a
libc hack or a kernel hack (like FUSE).

And AGAIN - nothing stops people from providing a FUSE module for D-VFS.
I SUPPORT THAT!  D-VFS just won't depend on FUSE.

IN any event, these discussions are still pointless.  There is ZERO code
for D-VFS at this point in time.  I'm still trying to get a firm idea on
what application developers need out of an API, because the API is the
most important part - implementation details should be 100% hidden
behind the API, so whether or not something like FUSE *is* used is
absolutely irrelevant right now.

> JP
> --
> God is dead, now the war shall never end.
> _______________________________________________
> xdg mailing list
> xdg at
Sean Middleditch <elanthis at>

More information about the xdg mailing list