proposal for file: uri standard

David Faure dfaure at
Mon Mar 29 16:22:27 EEST 2004

On Monday 29 March 2004 14:59, Thomas Leonard wrote:
> On Mon, Mar 29, 2004 at 02:35:33PM +0200, Alexander Larsson wrote:
> > On Mon, 2004-03-29 at 14:16, Waldo Bastian wrote:
> [...]
> > > Apart from that I have the suspicion that the hostname part is
> > > under-specified. E.g. what does an empty hostname mean, should
> > > "localhost" be handled differently, should the hostname be a FQDN?
> > 
> > Yeah.
> > 
> > I'm not actually aware of any app that uses the hostname part, but the
> > XDnD spec says its an error to not pass it [1]. The most basic use of
> > the hostname would be to check if it was the same as the current
> > machine, to do that it would be nice if the hostname was of some
> > canonical form, and its evident that localhost and empty hostname needs
> > to be handled specially (not with a strcmp).
> Should we even allow "localhost"? An empty hostname is already considered
> an error (because DnD between machines wouldn't work), and localhost is
> just as bad for this.

But uris are not only used for DnD!
If you pass a URI to an application on the command line, empty hostname or localhost is fine.

If you meant disallowing localhost in the XDnD spec, OK, but
not in the more generic file: uri standard here proposed.

David Faure, faure at, sponsored by Trolltech to work on KDE,
Konqueror (, and KOffice (

More information about the xdg mailing list