CWG2797 Meaning of "corresponds" for rewritten operator candidates

[jensmaurer]

There are special rules around rewritten operator candidates that make the following example well-formed, because the example appears in existing code bases:

  struct B {
    bool operator==(const B&);
  };
  struct C : B {
    C();
    C(B);
    bool operator!=(const B&);
  };
  bool c1 = B() == C();    // should be OK

Due to unrelated changes, the current wording makes this example ill-formed (because the two member operators have different object parameter types). This was not intended.

While fixing this, should the special rules be extended to also cover member operators with explicit object parameters? It seems unlikely that a new feature such as explicit object parameters would appear in a code base that also uses rather baroque combinations of comparison operators.

See CWG2797 for more details.

[erichkeane]

EWG discussed this during the Monday AM session in Tokyo. The following Poll was taken:

CWG2797 should be fixed with CWG2804, with a paper.

SF	F	N	A	SA
1	3	4	1	0

Result: Not Consensus

CWG2797- the special rules should be extended to also cover member operators with explicit object parameters:

SF	F	N	A	SA
1	0	4	1	1

Result: Not Consensus either way

Sending back to CWG: EWG has no consensus to extend to explicit object params.