Return type for the []= operator is being ignored

[DartBot]

This issue was originally filed by [email protected]

---

Hello,

It appears that the declared return type and the corresponding
returned value is being ignored, for the []= operator.

For example, if I have:

main() {
 Fred fred = new Fred();
 int x = 7, y = 23;
 String result = (fred[x] = y);
 print("Got result $result");
}

class Fred {
 String operator []=(int x, int y) {
   print("Inside []= operator code, x = $x, y = $y");
   return "${x} + ${y} = ${x + y}";
 }
}

Using the standalone vm, this program outputs:

Inside []= operator code, x = 7, y = 23
Got result 23

Using dartboard (which I understand is using dartc), it behaves the
same. (and if I enable type checks, it throws a type error when it
tries to assign the result (which expects a String, but which is
actually an int)).

No matter what return type I define for the operator, and no matter
what I actually return with the return statement, the result of:

String result = (fred[x] = y);

Is always y.

So, it's behaving as if it's an ordinary assignment statement, which
is defined in dart to evaluate the RHS value, essentially.

However, for the []= operator, it seems to be perfectly valid to
overload it with an arbitrary return type, and set it to return a
value, which apparently never gets used.

The spec doesn't appear to define anything about this, I had originally
assumed this might be an issue with the spec not describing this
operator has having some special interpretation, but apparently not.

Jason

[ghost]

Set owner to @sgmitrovic.
Added Accepted label.

[DartBot]

This comment was originally written by [email protected]

---

To update, it appears that the spec will likely evolve to reflect the current behavior, in which case no change would be needed for the vm or dartc. Stay tuned.

---

Removed Area-Language label.

[ghost]

Added WontFix label.