Documentation of gfortran.

[Paul Brook]

On Thursday 13 January 2005 19:12, Toon Moene wrote:
> 1. -d8 means the same as -i8 -r8 (to ensure the Standard behaviour that
>     a default integer is the same size as a default real).
>
> 2. On compilers that support 128-bit floating point numbers, -d8 often
>     means: If -r8 is specified, do *not* promote DOUBLE PRECISION to
>     128-bit floating point.  This is useful on processors on which
>     128-bit floating point arithmetic is expensive, and there are only
>     a few DOUBLE PRECISION variables in the code.
>
> Note that neither of the three options works, at the moment, so we can
> just document what we want (and put a note in that it doesn't work at
> the moment).
>
> My preference is for option 2.

I agree.
I also think we should add standard -fsomething versions of these options.

Paul