Re: build failure (alpha-current): floating point exception 8

[Alan Barrett <apb%cequrux.com@localhost>]

On Tue, 03 Mar 2009, Michael L. Hitch wrote:
> $ grep HUGE_VAL include/math.h
> /* 7.12#3 HUGE_VAL, HUGELF, HUGE_VALL */
> #define HUGE_VAL        __infinity.__val

You can't reliably answer a question about standards by checking the
implementation.

The ISO C99 standard permits, but does not require, HUGE_VAL to be
a positive infinity.  Annex F, which is about IEC 60559 (or "IEEE")
floating point arithmetic, does require HUGE_VAL to be an infinity,
but support for that kind of arithmetic is optional.

--apb (Alan Barrett)