[MIT-Scheme-devel] SYMBOL<? is not a total order consistent with EQ?

[Taylor R Campbell]

(eq? 'a (string->uninterned-symbol "a"))
;Value: #f

(symbol<? 'a (string->uninterned-symbol "a"))
;Value: #f

(symbol<? (string->uninterned-symbol "a") 'a)
;Value: #f

This makes cref confused when it tries to handle macro-generated names
of top-level definitions.  What to do?  Say all interned symbols
precede all uninterned symbols, or break name ties by putting interned
symbols first, perhaps?