At 05:59 PM 3/17/00 +0000, Miles Sabin wrote:
Actually, looking over 14.1 suggests that, for http at least,
the best way to deal with the sub-sub-type problem would be via
an additional parameter.
I know that this option has been raised before and rejected,
but on the face of it it seems like being the best way of
integrating with conneg.
Another approach to integrate would be to pick up the basic framework of
TCN (RFC 2295), namely an "Accept-Features" header, but use the conneg
syntax in place of the TCN-style feature expression. The same conneg
expressions could also be used by the server to describe variants.
A slightly different approach, in which the data source makes the final
determination, might be modelled on a proposal being considered for
Internet fax <draft-fax-content-negotiation-01.txt> (this updated version
due to hit the ID directories any time now).
#g
------------
Graham Klyne
(GK(_at_)ACM(_dot_)ORG)