[Scl] Re: Report on Common Logic

[John F. Sowa]

Bill,

Actually, I was combining several issues in one email note,
so the separate points might not be clear.  Three issues:

  1. The SCL document needs to be expanded with more examples
     and made easier to read -- for everybody.  Chris agrees
     on that point.

  2. There are political issues in getting approvals from
     ISO, W3C, etc.  Those are separate issues, and I should
     not have mixed them up with the discussion of seqvars.

  3. And finally I agree with you on the value of sequence
     variables.  I would like to see them in the CL language.

My concern right now is a technical issue:

  1. I want to have an ontology for sequences, which can be used
     for list processing as in Lisp and Prolog.  That same ontology
     could also be used to support KIF-like sequence variables,
     but with the ability to have full quantification over the
     sequences -- and with no need to distinguish sequence
     variables from any other kind of variables.  The syntax
     for using them could be the same as in the old KIF, but
     with some new possibilities for list processing.

  2. The Z language has a pure FOL semantics without any built-in
     ontology in the core, but it does have some optional ontologies
     in the "Z workbench", which includes sets, bags, sequences,
     and integers.  I consider the Z workbench to be a valuable
     supplement to the CL core, and I believe that it would be
     very useful for most, if not all applications of CL and
     languages defined in terms of CL.  Since Z and the Z workbench
     are already ISO standards, we could just include a reference
     to the Z workbench and recommend it as an optional supplement
     to CL.

  3. If we have an optional ontology for sequences as first-class
     entities, we can make that ontology a prerequisite for any
     use of sequence variables in the sense of KIF.  People who had
     been using KIF sequence variables could continue to use them,
     but they would also have new options for list processing.

My question for Pat and Chris is whether we can or should use
this approach to handle sequence variables.

John