Re: [Dev-C++] double and float ??

VC++ doesn't have an long double data type. A long double variable will be
8 byte large, just as a double. In short, there will be no problem trying
to send a long double parameter to a variadic function. Both sender and
receiver will assume 8 bytes storage size.

It is only code that relies on extra precision and/or range of a long
double that will have problems. For example a numeric algorithm using a
numerically unstable method of solving an equation might diverge with
double numbers while it always have enough precision to converge on a
solution with true long doubles.

/Per W

On Fri, 24 Feb 2006, Daniel K. O. wrote:

> Per Westermark escreveu:
> > %Lf specifies long double (when supported)
> >
>
> Which the MS runtime doesn't seem to support... from my experiences (I
> don't have the MS compiler, I just asked somebody else to compile some
> code and send me the program output and some assembly dumps), VC++
> "demotes" long doubles to plain doubles in printf.
>
> But now that you mentioned %Lf, I would like to confirm with anyone here
> who has VC++; what does happen when trying to printf a long double on
> VC++? For me, it looks like VC++ tries a "magic" conversion in variadic
> functions, but if it is true, it may break legit code (when trying to
> va_arg a long double where a plain double was put).
>
>
>
> ---
> Daniel K. O.
>
>