Bug 12773 – Compiler implicitly converts delegate into function when taking an address of a method

This is extremely nasty. Take a look: ----- alias Func = void function(); class C { static void foo() { } void bar() { } } void main() { { // Func func = C.foo; // disallowed, since this is a function call } { Func func = &C.foo; // ok, the proper syntax usage. func(); // ok } { Func func = &C.bar; // oops, we forgot to mark 'bar' as static!! func(); // access violation (requires 'this') } } ----- The current semantics force us to be extremely careful when using function pointers. We should ban the implicit conversion of &foo to a function pointer if this is really a non-static class method. People can use .funcPtr or some other equivalent instead.

I suspect you meant to suggest removing the implicit conversion of &C.bar (a delegate) to a function pointer. I would agree with this, and am extremely surprised this is allowed in the first place. Perhaps it was an attempt to implement implicit conversion from a function pointer to a delegate? (which does make sense to be possible)

I believe this behaviour is by design. There is no implicit conversion, because &C.bar is not a delegate, it is a function pointer. The following code works as expected: alias Func = void function(); alias Del = void delegate(); class C { string str = "bar"; static void foo() { } void bar() { writeln(str); } } void main() { { // Func func = C.foo; // disallowed, since this is a function call } { Func func = &C.foo; // ok, the proper syntax usage. func(); // ok } { Func func = &C.bar; // correct C c = new C; Del del = &c.bar; // correct Del del2; del2.funcptr = func; del2.ptr = cast(void*)c; del2(); // works, output is "bar" } }

An old favorite. *** This issue has been marked as a duplicate of issue 3720 ***