[R6RS] How SRFI 35 enables communication protocols via conditions

[Michael Sperber]

"Manuel Serrano" <Manuel.Serrano at sophia.inria.fr> writes:

>> R5RS already has aggregate types in the forms of pairs, vectors, and
>> procedures.  What mechanism a Scheme system uses for representing
>> conditions should be an implementation issue, not a specification
>> issue.
> No I disagree. Pushing your argument to its limit we could decide to ban
> everything but lambda! 

I didn't argue for banning anything.

> Exceptions have something to do with records because they need type
> predicate and accessors, exactly what records provide.

Sure---and that's exactly what the SRFI 35 reference implementation
does:

(define-record-type condition-type :condition-type
  (really-make-condition-type name supertype fields all-fields)
  condition-type?
  (name condition-type-name)
  (supertype condition-type-supertype)
  (fields condition-type-fields)
  (all-fields condition-type-all-fields))

-- 
Cheers =8-} Mike
Friede, V?lkerverst?ndigung und ?berhaupt blabla