William D Clinger wrote: > I can't imagine a rationale for requiring (b), however.
The only one I can think of is that I want to compare IEEE singles (doubles) obtained from a byte vector with numbers obtained by read and I know my implementation implements inexact real numbers with sufficient precision to express all IEEE singles (doubles). But your comment prompts a wider question: what problem is x|p trying to solve? In other news: The current draft also implies that 1.1f0 should be interpreted as 1.1f0|53. This seems bizarre. Regardless of the meaning of |p, perhaps the default values of p should be f exponents and 53 for e and l exponents. I am not sure what the default should be for s exponents. Furthermore, suppose I have a Scheme with that uses both IEEE single-precision and IEEE double-precision formats for real numbers. Suppose I write x|24. If x is a normalized number, the implementation "should" represent it using a single-precision format. If x is a denormalized number, the representation "should" represent it using a double-precision format (because denormalized single-precision numbers have fewer than 24 bits of precision). Regards, Alan _______________________________________________ r6rs-discuss mailing list [email protected] http://lists.r6rs.org/cgi-bin/mailman/listinfo/r6rs-discuss
