Re: Capability terminology

[Graham Klyne <GK@xxxxxxxxxxxxxx>]

At 14:05 12/11/98 -0800, Dan Wing wrote:
>On Thu, 12 Nov 1998, Graham Klyne wrote:
>
>> >"Exchange" is a two-way transaction.  I "exchange" US dollars for
>> >a foreign currency (two-way transaction).  I "exchange" my efforts
>> >at work for my paycheck.  My body "exchanges" oxygen for carbon
>> >dioxide.
>> 
>> True.  I used this form of definition because the term "capability
>> exchange" has also been used in the fax context which is not a two-way
>> communication.
>
>With fax there isn't a symmetric capabilities exchange but there is a
>capabilities exchange between the transmitting and receiving terminals.
>
>However, the existing proposals for capabilities aquisition for email
>such as LDAP, DSN/MDN messages, etc., do not cause the sender to
>provide any of its capability information at all.

I do not see the sender disclosing any of its capabilities to the recipient
in G3 fax.  In eifax, the existence of certain MDN request extensions might
give some clues, but I still see no two-way transaction here.  I think that
is "capability identification".  But it is also called "capability
exchange" by some, hence my definition.

Anyway, do you wish to propose an alternative definition for "capability
exchange"?

And a question to ALL:  are these definitions helpful, and where should
they appear?


>Perhaps this is something we should address:  a sender knows what file
>types it could render a source document in: image/tiff, image/jpeg,
>application/pdf, etc., and maybe it would be useful for the sender to
>query the recipient with details on the recipient's ability to handle each
>individual MIME type.  I am sure in practice that this would degrade to
>http's "Accept: */*" pretty quickly, though....

I'm sure you're right.  One of my goals in the early 'conneg' work was to
devise a system that did not impose constraints on the mechanism used.
Hence the emphasis on symmetry (of expression).

Individual mechanism may, of course, be asymmetric.  It is my view, which I
tried to capture in the conneg -requirements- draft, that the appropriate
form of mechanism is dependent on the protocol/application with which it is
used.  So mechanisms for web browsing would be very different to mechanisms
for fax, etc.

#g

------------
Graham Klyne
(GK@xxxxxxx)