https://issues.dlang.org/show_bug.cgi?id=24024

kinke <kinke@gmx.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |kinke@gmx.net

--- Comment #1 from kinke <kinke@gmx.net> ---
This is the behavior I would expect - a class-ref `this` is special and not to be mutated. What if `foo()` sets it to null?

--

https://issues.dlang.org/show_bug.cgi?id=24024

--- Comment #2 from Walter Bright <bugzilla@digitalmars.com> ---
Looking into the code, the reason the match fails is because `this` for a class is considered only an rvalue, not an lvalue.

--

https://issues.dlang.org/show_bug.cgi?id=24024

Walter Bright <bugzilla@digitalmars.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           See Also|                            |https://issues.dlang.org/sh
                   |                            |ow_bug.cgi?id=13116

--- Comment #3 from Walter Bright <bugzilla@digitalmars.com> ---
https://issues.dlang.org/show_bug.cgi?id=13116 made class `this` not an lvalue.

--

https://issues.dlang.org/show_bug.cgi?id=24024

FeepingCreature <default_357-line@yahoo.de> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |default_357-line@yahoo.de

--- Comment #4 from FeepingCreature <default_357-line@yahoo.de> ---
I mean, in the extreme, if this was allowed, shouldn't this happen?

class C
{
    void mem()
    {
        foo(this);
    }
}

void foo(ref C c) { c = null; }

C c = new C;
c.mem;
assert(c is null);

And what about if `c` is genuinely an rvalue, such as `C createC();
createC.mem;`?

--

https://issues.dlang.org/show_bug.cgi?id=24024

--- Comment #5 from Walter Bright <bugzilla@digitalmars.com> ---
(In reply to FeepingCreature from comment #4)
> I mean, in the extreme, if this was allowed, shouldn't this happen?

It isn't any different from:

 class C
 {
     static void mem(C c1)
     {
         foo(c1);
     }
 }

 void foo(ref C c2) { c2 = null; }

 C c3 = new C;
 C.mem(c3); // c1 is a copy of c3
 assert(c3 is null); // c1 is set to null, not c3


> And what about if `c` is genuinely an rvalue, such as `C createC();
> createC.mem;`?

Rvalues cannot be passed by ref.

13116 created a special case, and like special cases, it works inconsistently with the rest of the semantics.

--

https://issues.dlang.org/show_bug.cgi?id=24024

--- Comment #6 from FeepingCreature <default_357-line@yahoo.de> ---
I guess I'm just allergic to ref-passing variables that you cannot see declared anywhere. In your example, C c1 is explicitly declared; in the original example, it's implicitly declared by the method call. IMO, implicitly declared variables should always be rvalues, but I guess that's a lost cause.

--

https://issues.dlang.org/show_bug.cgi?id=24024

--- Comment #7 from Walter Bright <bugzilla@digitalmars.com> ---
Consider UFCS and the ability to do a.foo() and foo(a). When `this` behaves
differently than a parameter, it generates a corner case with surprise
behavior.

Early on I spent much care to make the address of a nested function (i.e. a delegate) indistinguishable from the address of a member function. This symmetry and consistency turned out to be a big win.

Avoiding special cases as much as possible should pay off for us.

--

https://issues.dlang.org/show_bug.cgi?id=24024

Adam D. Ruppe <destructionator@gmail.com> changed:

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

--

https://issues.dlang.org/show_bug.cgi?id=24024

Dlang Bot <dlang-bot@dlang.rocks> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |pull

--- Comment #8 from Dlang Bot <dlang-bot@dlang.rocks> ---
@WalterBright created dlang/dmd pull request #15389 "fix Issue 24024 - cannot pass class this to ref class" fixing this issue:

- fix Issue 24024 - cannot pass class this to ref class

https://github.com/dlang/dmd/pull/15389

--

https://issues.dlang.org/show_bug.cgi?id=24024

Dlang Bot <dlang-bot@dlang.rocks> changed:

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

--- Comment #9 from Dlang Bot <dlang-bot@dlang.rocks> ---
dlang/dmd pull request #15389 "fix Issue 24024 - cannot pass class this to ref class" was merged into master:

- 7d9679081f79e94793a13829e3adb54809e2da4e by Walter Bright:
  fix Issue 24024 - cannot pass class this to ref class

https://github.com/dlang/dmd/pull/15389

--