[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Scheme-reports] Formal Comment: R7RS 'eqv?' cannot be used for reliable memoization
> I have changed "conforming to" to "implemented in the style of",
> which I think eliminates that problem.
I am happy with this change. Thanks.
I believe eqv? should be equivalent to operational equivalence for numbers. Given that, I believe that the definition is theoretically flawed, since one might be able to imagine an implementation in which inexacts are not "implemented in the style of" IEEE numbers and in which two numbers can be = without being operationally equivalent. However, in practice, this change gives all likely implementations (including those using MPFR) permission to make eqv? equivalent to operational equivalence for numbers.
Furthermore, I expect that no important implementation will implement inexacts that are not "implemented in the style of" IEEE, so we will be able to revisit this (at length!) in the R8RS without breaking anything that was not already broken.
Scheme-reports mailing list