gcc/kcc difference

[Martin v. Loewis]

> But then, there's Koenig lookup, and a contorted reading of its
> definition may lead one to believe that definitions of Unipolar in the
> global namespace should be considered because `int' (the type of the
> first argument of the function) is in the global namespace (is it?).

No. 3.4.2/2:

# If T is a fundamental type, its associated sets of namespaces and
# classes are both empty.

Only user-defined things have associated namespaces.

Regards,
Martin