[SCL] Obvious SCL spec ideas for extensions

[Tanel Tammet]

Hi Chris,

Wanted to point out that:

Although I have proposed to have a separate
 ECL language in the SCL document, we could
alternatively proceed by putting seqvars and the
two kinds of predicates into the SCL modules,
as we originally thought.

This would be possible only if you manage to
come up with a really easily understandable MT
and abstract syntax for these two things, building
directly on SCL core.

I have no intuition if that can be done (initially
suspected that it cannot, hence suggested a separate
ECL language).

Simpler things could be done, like this:

For example, a natural extension idea for core SCL
is to allow people to write tuples of OTHER types
than Atom_i and Term_i. These tuples could be,
like, say, Man_1 and Father_2, ie ordinary
TFOL predicates. In order to bring them in
we would simply need to extend the abstract
syntax and concrete syntaxes (the latter must
then contain special markup). There is even
no need to modify the current quantifiers. The
Man_1 and Father_2 type of tuples would
simply be disjoint from predicates as constants.
This is simple to do and will give us traditional
FOL, but does not catch your idea of arbitrary overlap.
For users, it would be a tad confusing though:
some people would use "tuple predicates"
while others would use "constant predicates",
with no good reason as for why use one or
another: this is a bit worrisome. We'd have
to give good guidelines for usage, but still
cannot completely avoid the possibility of confusion.

Any ideas?

Regards,
           Tanel