I am concerned about the convenience features that the SRFI 99 syntactic
layer adds to SRFI 9.  For the sake of being able not to specify the
conventional names of the constructor, predicate, accessors, and mutators,
it becomes impossible to implement the syntactic layer in syntax-rules
alone (and therefore in fully portable R5RS), as SRFI-9 was.  I consider
this too high a price for too little gain.

I have no objection to, and indeed I favor, the innovations of (name
parent) to indicate single inheritance and #f to indicate no constructor
or no predicate.

In addition, the fact that field specs are always lists in SRFI 9 means
that one can add optional non-list arguments just before the field specs
with new meanings, like the record-printer argument I have proposed at
http://tinyurl.com/ltqmnu .

--
John Cowan   xxxxxx@ccil.org  http://www.ccil.org/~cowan
Most languages are dramatically underdescribed, and at least one is
dramatically overdescribed.  Still other languages are simultaneously
overdescribed and underdescribed.  Welsh pertains to the third category.
        --Alan King