Re: How to merge required operations under a common name
On Wed, 27 Jan 2016 11:45:29 +0000, David Lester wrote:
> I even think you could use the computable reals; but that would
> compromise the comparison operations.
It would violate the requirement that each type be a finite set, in
order to have a well-defined interval hull. For example, MPFR would
have to be defined as a family of precision-indexed types and not as
a single type.
I fully agree with your main point however: 1788-2015 is not bound
to 754-2008 in any way; 754-2008 is just one of the families of number
formats that can be used in a conforming manner.
Michel.
---Sent: 2016-01-27 13:39:08 UTC