----- Original Message -----
From: "Robinow, David" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Thursday, January 04, 2001 9:29 AM
Subject: RE: values.h / 2
> > Is it possible, to include a regular values.h in cygwin?
> I suggest you use DBL_MIN and DBL_MAX from float.h
>
> I'm not familiar with the various standards
> controlling this sort of thing, but I'm guessing
> that values.h can be considered obsolete
>
As others have said, obsolete since the advent of C89.
> for cygwin float.h defines DBL_MIN and DBL_MAX
>
> My Irix has both float.h and values.h and
> DBL_MIN = MINDOUBLE
>
> but my SunOS 5.6 has (I find this strange)
> DBL_MIN = 2/225073858072014E-308 and
> MINDOUBLE = 4.94065645841246544e-324
Several versions of Irix omitted support for gradual underflow; thus there
would be no non-zero number less than DBL_MIN. SunOS presumably defaults to
supporting gradual underflow; thus the smallest non-zero would be DBL_MIN *
DBL_EPSILON. In either case, DBL_MIN would be the smallest number
represented without loss of precision. I'm not too thrilled at the
possibilities opened for revisiting these behaviors by setting precision
mode and abrupt underflow on IA32 processors.
--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple
