This is an addendum to the proposal for nested functions
that allows the 'address' of a nested function
to be taken in certain circumstances without forming a closure.

Recall that a nested function is a sort of 'member' of a
function, then:

void func() {
 void a(int){ .. }
 void b(int){ .. }
 void c(int){ .. }
 ...
 void (func::* f)(int)  = &func::b;
 ...
 (func.*f)(4);
}

This is the exact syntax that would be used if 'func' had been
a class. We will simplify this to:

void func() {
 void a(int){ .. }
 void b(int){ .. }
 void c(int){ .. }
 ...
 void (func::* f)(int)  = b;
 ...
 (*f)(4);
}

That is, the 'address' of a nested function can be taken
as a 'pointer to member'.

When a pointer to member is 'executed', it is bound to the current
instance of its environment, as if an ordinary call had been
made.

A function is only viewed as a 'class' or 'namespace' within
the function, so that it is impossible to declare a pointer
to member type for a nested function outside the function
in which it is nested, that is, there is no way to pass
such a pointer to member out of the function.

Thus this extension to the original proposal is both statically
and dynamically secure.

[Preview: closures are taken as follows:

 void (closure*)(int) = &b; // or just plain 'b'

and these can be passed out of the function]

--
        JOHN (MAX) SKALLER,         INTERNET:maxtal@suphys.physics.su.oz.au
 Maxtal Pty Ltd,      CSERVE:10236.1703
        6 MacKay St ASHFIELD,     Mem: SA IT/9/22,SC22/WG21
        NSW 2131, AUSTRALIA