gcc/kcc difference

[Joe Buck]

> Of course, it seems that Stroustrup's example is incorrect: In 'g(2)',
> g is not a dependent name, so it is bound at the point of definition,
> not at the point of instantiation.

Stroustrup describes this exactly as you do, so if there was any mistake
it was in my explanation of his example: he says that g is bound at point
of definition (even though a "better" g is declared later)