One part of the test suite distributed with the sample implementation is the following:

(set! syms2 (list->set eq-comparator '(e f)))

;; syms2 is now {e, f}

(test-eq 2 (set-size syms2))

(test-assert (set-contains? syms2 'e))

(test-assert (set-contains? syms2 'f))

(list->set! syms2 '(a b))

(test-eq 4 (set-size syms2))

The last test looks meaningless and non-portable to me. In a purely functional implementation (no side effects) of SRFI 113, list->set! won't change syms2.

I think, the test should read:

(set! syms2 (list->set! syms2 '(a b)))

(test-eq 4 (set-size syms2))

Marc

P.S.: The reason for my recent comments on SRFI 113 is that I am currently in the process of writing a purely functional implementation of it so that I can evaluate whether we really need another SRFI for immutable sets or whether SRFI 113 suffices.

(BTW: Is there already a draft for a SRFI for maps that builds upon SRFI 113?)