Deployment of new URI schemes (was Re: request for a new URL scheme)

John C. Daub writes:
 > On Wed, 10 Apr 1996, Reed Wade wrote:
 > 
 > > This is not necessarily a bad idea but I think you'll
 > > have a very hard time getting this adopted. And if
 > > adopted, implemented by browser writers.
 > 
 > yes.  the actual implimentation into the URL standard wouldn't be
 > that hard to do.  it'd be getting all the software authors to
 > support it and release new versions of their software just for
 > this thing.

This is a major impediment to the deployment of any new URL or URN
scheme.

Browsers should allow users to specify proxies that handle specific
URI schemes, possibly new schemes.  Netscape, for one, still assumes
that any URI with a scheme it does not know about is a relative
URL, and it has no way to specify proxies to handle specific new
schemes.   Mosaic for X has fixed these problems, partly at my
insistence.

The most that most browsers let you do now is to pass ALL URIs on
to a proxy - to get over firewalls.  The proxy could be designed
to handle new schemes, but if the wrong URI is given to it in the
first place, what can be done?

More generally, browsers should allow users to specify any mapping
from patterns of URIs to handlers.  The handlers could be either
internal to the browser, some external program, a proxy, a CCI
process, a Java program, etc.  I strongly urge browser implementors
to add any of these URI extension capabilities - pick whatever is
easiest.

Daniel LaLiberte (liberte@ncsa.uiuc.edu)
National Center for Supercomputing Applications
http://union.ncsa.uiuc.edu/~liberte/

Received on Wednesday, 10 April 1996 18:03:08 UTC