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

[Gabriel Dos Reis]

On Tue, Jun 2, 2009 at 2:34 AM, Juan Jose Garcia-Ripoll
<juanjose.garciaripoll at googlemail.com> wrote:
> On Mon, Jun 1, 2009 at 6:21 PM, Gabriel Dos Reis
> <gdr at integrable-solutions.net> wrote:
>> As a user, I would like to know what impact this change would have
>> for ECL-based OpenAxiom.  OpenAxiom is quite happy that it does
>> not have to go through whoops to get NaNs or infinities as
>> directly provided by the hosting system.
>
> AFAIK, CommonLisp mandates that all floating points are valid numbers.
> NaNs and infinities are not contemplated. So far, ECL has not
> supported them. There is no way to print them or read them, and the
> fact that in some architectures they have been generated (due to not
> trapping the appropriate exceptions), has been accidental. Even
> without these patches, on my Mac OS X, division by zero produces
> exceptions.

I fully understand what the CL spec says, but then it was then written
ages ago before the widespread availability of IEEE-754 semantics.

>
> Now, I understand that for some purposes it might be desirable to have
> no-numbers and infinities. However, one should then ponder how to do
> this in the least disruptive way, and as an optional feature.

Yes, it would be useful for people to get a way choose IEEE-754 over
whatever CL thinks is best for them.