I agree that being unable to shadow a keyword would be a wart if keywords were really identifiers, but in Schemes that support them they are not.  In Chicken or Guile, for example, you cannot write (let ((foo: 32) (+ 32 foo: 45)) => 109, and it would be really bad if you could, because what if instead of + you used some function with keywords instead?  (Note: In Guile you must incant (set-reader! keywords suffix) to make this work.)

I am willing to say that RnRS was insufficiently forward-looking by not to allow space for identifier-like things that don't behave as identifiers, even if it doesn't define their exact behavior.  #:foo is most certainly not an identifier, but only Chicken, Guile, Racket, and Kawa support it as keyword syntax, and in Chez it means "uninterned symbol" as in CL.

I think allowing mixing is a great mistake.  I don't know of any language, Lisp or not, that permits such a thing, even when the syntax makes clear when formal parameters appear in the call and when they don't.  I am indifferent between Options 3 and 4, but prefer 2 to either.

John Cowan          http://vrici.lojban.org/~cowan        xxxxxx@ccil.org
Original line from The Warrior's Apprentice by Lois McMaster Bujold:
"Only on Barrayar would pulling a loaded needler start a stampede toward one."
English-to-Russian-to-English mangling thereof: "Only on Barrayar you risk to
lose support instead of finding it when you threat with the charged weapon."