Problem with arduino maths

AWOL:
It allows you to compile software unchanged, but gives lower precision results.

Why not just use

#define double float

or something of that sort? (if you really are that desperate for it to compile)