Re: external representations William D Clinger (24 Jun 2005 04:25 UTC)
Re: external representations Bradley Lucier (24 Jun 2005 22:36 UTC)
Re: external representations Bradley Lucier (27 Jun 2005 20:57 UTC)

Re: external representations Bradley Lucier 24 Jun 2005 22:36 UTC

On Jun 23, 2005, at 11:25 PM, William D Clinger wrote:

> Interesting.  Can you tell me of any hardware that actually
> does this?

Offhand, no.

This is why I believe, however, that NaNs that can be used as inputs
to or can be results of floating-point processing should have
external representations.  If you want to define other objects that
cannot arise as outputs or inputs to floating-point operations that
otherwise have the form of a NaN (maximum exponent, nonzero
mantissa), then I don't think these things should be called NaNs,
precisely, and I don't think that they need external representations.

Brad