Re: [Gm2] complex types

[Gaius Mulley]

Andreas Fischlin <address@hidden> writes:

> Hi Gaius,
>
> I am simply using CMath to implement whatever MathLib0 etc. I wish to
> implement. That should not break any legacy code. I have of course a legacy
> implementation of MathLib0 that uses portably algorithms if you wish to have a
> legacy implementation in order to have not the least significat bit being of
> max precision. However, I generally use CMath, since it is to the least
> significant bit precise, even if you use it as implementation of a REAL or 
> even
> SHORTREAL definition of the procedure. RETURNing the long real value exploits
> the assignment compatibility, e.g.
>
>     PROCEDURE Sin(x: REAL): REAL;
>     BEGIN
>       RETURN CMath.sin(x)
>     END Sin;
>
>  and that's about all there is.
>
> Regards,
> Andreas

Hi Andreas,

ahh, yes - sorry I missed that it was a definition module for CMath
(rather than MathLib0) - my mistake,

regards,
Gaius