schemas will be fragments of protocols

Usage Scenario 6 (Open and uniform transfer of data between
applications, including databases) says to me that schemas will be
fragments of protocols.  A protocol defines not one "message type" but
several, and says things about how they relate.  So a protocol language
addresses a superset of the issues addressed by a schema language.  Or
maybe not: one of the kinds of data you'd like to express in a schema is
"a reference to an object that obeys protocol Y".

One could imagine the usage playing out like this: there is a schema for
each of several "message types", and a protocol definition (in some
other language) that references the several schemas (which have some
data types in common, leading to further structure of specification). 
This could work, but seems a bit clunky.  A nicer vision is one where
there is simply a protocol definition, in a language that includes the
schema language as a fragment.

I'm not sure whether this rises to the level of a requirement, or a
"goal", or what.  But I'd like to see the issues of writing protocols
explicitly addressed.

Cheers,
Mike

Received on Thursday, 4 March 1999 12:55:51 UTC