https://issues.dlang.org/show_bug.cgi?id=14353

Shammah Chancellor <shammah.chancellor@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |shammah.chancellor@gmail.co
                   |                            |m

--

https://issues.dlang.org/show_bug.cgi?id=14353

ag0aep6g@gmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ag0aep6g@gmail.com

--- Comment #1 from ag0aep6g@gmail.com ---
I think this is invalid. `t.add` and `s.add` are parentheses-less calls. So the error about T.add not being callable "using argument types ()" is correct. To get a delegate, add '&':

struct S {
    int i;
    T t;

    auto add(int a) {
        t.i = a + i;
        return &t.add; /* ! */
    }
}

struct T {
    int i;
    int add(int a) {
        return i + a;
    }
}

int main() {
    S s;
    s.i = s.t.i = 1;
    auto dg1 = &s.add; /* ! */
    auto dg2 = dg1(34);
    return dg2(7);
}

--

https://issues.dlang.org/show_bug.cgi?id=14353

--- Comment #2 from Shammah Chancellor <shammah.chancellor@gmail.com> ---
Definitely possible.  I will file an SDC bug if this turns out to be the case.

--

https://issues.dlang.org/show_bug.cgi?id=14353

--- Comment #3 from Shammah Chancellor <shammah.chancellor@gmail.com> ---
I'm not so sure this is desirable.  What if there add as a ref return.  What does &t.foo give then?

--

https://issues.dlang.org/show_bug.cgi?id=14353

--- Comment #4 from Ketmar Dark <ketmar@ketmar.no-ip.org> ---
i believe that it will be a bug in the D code, as there is no variable to ref. at least it should be a bug. ;-)

--

https://issues.dlang.org/show_bug.cgi?id=14353

--- Comment #5 from ag0aep6g@gmail.com ---
(In reply to Shammah Chancellor from comment #3)
> I'm not so sure this is desirable.  What if there add as a ref return.  What does &t.foo give then?

You mean something like the following?
----
struct T
{
    int i = 42;
    ref int foo() {return i;}
}
void main()
{
    T t;
    auto dg = &t.foo;
    auto i = &t.foo();
}
----
Here dg is a delegate of the method foo and i is a pointer to the ref-returned field i.

The spec says this [1]:
"In most places, getter property functions are called immediately. One
exceptional case is the address operator."
"Even if the given operand is a property function, the address operator returns
the address of the property function rather than the address of its return
value."
Now foo is no @property, but the spec says that "[if] a function call does not
take any arguments syntactically, it is callable without parenthesis, like a
getter property functions." So it's reasonable that the same rules apply.

[1] http://dlang.org/function.html#property-functions

--

https://issues.dlang.org/show_bug.cgi?id=14353

Kenji Hara <k.hara.pg@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |INVALID

--- Comment #6 from Kenji Hara <k.hara.pg@gmail.com> ---
(In reply to ag0aep6g from comment #1)
> I think this is invalid. `t.add` and `s.add` are parentheses-less calls. So the error about T.add not being callable "using argument types ()" is correct. To get a delegate, add '&':

That's right. When a function symbol identifier without address operator appears in expressions, it will be evaluated to a parenthesis-less function call.

Therefore, it's definitely a bug in SDC.

--

https://issues.dlang.org/show_bug.cgi?id=14353

Andrei Alexandrescu <andrei@erdani.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Version|unspecified                 |D2

--