> The SRFI 225 idea of using a type predicate to support polymorphism can
> only reliably support either alists or plists but not both.

(pair? (car list)) could identify an alist, since plists rarely if ever
have pairs as keys. (Usually plists have symbol keys so that eq? can be
used, though it almost certainly makes sense to support keys that are
eqv? or even just equal? without being eq?. However, pair keys are a
pretty extreme case.)

I do agree that the above is non-ideal, since the other procedures in
the SRFI would happily create plists with pair keys if told to do so.

The case of the empty list is also problematic. Is it an alist or a plist?