Re: associating constant feature sets wi

At 16:33 20/01/99 +0100, Johan Hjelm wrote:
[...]
>In a work that is being done in conjunction with the WAP Forum, myself and
>Hidetaka Ohto of Panasonic are currently developing exactly that. We should
>be able to make the protocol available for review here when the spec is in
>a decent state. 

>It turns out that HTTP is a rather good fit for this (if you use extension
>headers), but departs a lot from the protocol-independent direction the
>conneg group has been taking, since several of our ideas are using generic
>HTTP 1.1 functions to accomplish the process. 

Is this helpful?  It sounds to me as if a purely HTTP-based solution would
not be applicable to, say, e-mail environments.

IMO, one of the greatest benefits of pursuing a protocol-independent
approach is that the capability semantics are to some extent unified across
different protocols.  This in turn means that gatewaying between protocols
is easier to achieve without loss or distortion of information.

Can the *semantics* developed in this group be mapped exactly by the HTTP
extension syntax that you are proposing?

#g

------------
Graham Klyne
([email protected])