I was playing with D's function pointers and immutability and ran into
some trouble. It's probably because function pointers are treated just
like data pointers when it comes to immutability. To mend this, function
pointers should be treated as values. It may sound odd but makes sense --
you can't possibly dereference a function pointer, so we're dealing only
with addresses themselves and these are values.
One of the results of the amendment would be allowing implicit conversion
from immutable/const to mutable and vice versa. Note that
immutable(void function()) ifp;
void function() fp = ifp; // Error: cannot implicitly convert
expression (fp) of type immutable(void function()) to void function()
is exactly the same as
immutable int ii;
int i = ii;
so should compile.
I was also thinking that an expression taking a function's address (&fun)
should be an immutable function pointer. Not sure about this one, though.
BTW, a quick bugzilla search shows that I'm not the only one who wants
function pointers as values.
http://d.puremagic.com/issues/show_bug.cgi?id=3650
Tomek
