When writing on a parser with normal infix syntax, I found that treating tuples as reifications of values convenient. However, implementing this in a systematic way is very inefficient it seems, as all calls must be passed through call-with-values. Would having this extension risk inefficient Scheme implementations?

Roughly, a normal infix expression f(x_1, ..., x_k) corresponds to Scheme syntax (f x_1 ... x_k), where (x_1, ..., x_k) is a data type satisfying a singleton reduction rule (x) = x. If writing (f (values x_1 ... x_k)) the result would be the same as (call-with-values (lambda () (values x_1 ... x_k)) f), so call-with-values would not be necessary here. If returning (values x_1 ... x_k) in a position where the standard now says the result is unspecified, one would get a singleton reduced tuple (x_1, ..., x_k).

Hans

_______________________________________________
Scheme-reports mailing list
Scheme-reports@scheme-reports.org
http://lists.scheme-reports.org/cgi-bin/mailman/listinfo/scheme-reports