Weird arithmetic

[Pascal Gallois]

In the long run, however, you can easily cause yourself and your colleagues
a lot of grief by ignoring the distinction between decimal and binary
arithmetic.

This is more an issue with excess precision and certainly x86 related. And
it also has to do with GCC insofar as even if you cast the value of
an expression involving excess precision to double which should get rid of
excess bits, it still goes wrong (against the standard (6.3.1.8 and
footnote 52)).

A possible solution would also be to use long double: long double m =
200.45L; The trailing L causes proper initialization of extended
precision bits.

- Pascal