Re: [Mingw-users] expm1()

[Danny S. <dan...@cl...>]

Greg Chicares wrote:
> On 2005-01-27 03:29 AM, Danny Smith wrote:
>> BTW, do we need long double versions of the M_* defines to ensure enough
>> precision?
>
> Gosh, you're right--that's a real problem.
>
> We want DECIMAL_DIG digits in a long double context. That means
> 21 digits according to C99 5.2.4.2.2/13:
>
<snip>
>
>> Will you update your patch on the Sourceforge patch tracker?
>
> I will. First, though, let's try to solve the M_LN2 problem.
> I'd hate to hardcode a constant value in expm1() now, when what
> we really want is to use that macro, but with a value that's
> not overstated.
>

If we really want exact 64-bit representation we could use float hex format, but
IIRC 2.95 had a bug with hex format (S Moshier  published a patch to fix the gcc
bug, and I think I put it into some of the later 2.95.3-x distros)..  The hex
float numbers could be generated using a union of shorts and long double.

I could use the Cephes  ldtoa code to get decimal values as well, and check
against published values
In any event, I think we should keep the OPENX defines and add new ones, like
_M_EL, M_LOG2EL, etc.
Danny