[Ecls-list] [PATCH] Improve floating-point exception handling.

[Robert Dodier]

On Tue, Jun 2, 2009 at 8:53 AM, Juan Jose Garcia-Ripoll
<juanjose.garciaripoll at googlemail.com> wrote:

> - One or more NaNs?

Well, there are many NaN values. I think ECL should just preserve all the bits.

> - If NaNs are not always equal, not even to themselves, then we can
> probably print them using #. followed by a constant name or some
> function.

It might be useful to print the bits in hexadecimal.
Actually such a displayed form could be readable as well,
since if the bits are specified you can reconstruct the same NaN.

Alternatively perhaps just print #<NAN> or whatever, and make
the bits available via another function, maybe one just for NaN's,
or maybe extend INTEGER-DECODE-FLOAT.

> - Same would probably apply to +-Inf.

Perhaps it's possible to make +/- inf readable. That would be useful.

All the best

Robert Dodier