[PATCH] D44883: [Sema] Extend -Wself-assign with -Wself-assign-overloaded to warn on overloaded self-assignment (classes)

[Roman Lebedev via Phabricator via cfe-commits]

lebedev.ri added a comment.

Ok, once we understand why "field" diag is separate, and why since initial implementation it only warned for "builtin" operators, i'll work on the code further.

In https://reviews.llvm.org/D44883#1048485, @rjmccall wrote:

> That's an interesting question!  You're absolutely right, I do think we should warn about trivial C++ assignments as part of the builtin group.  If a C program includes a self-assignment that happens to be of struct type, we'll warn about that, and it seems to be that we shouldn't lose the warning just because that code is compiled as C++ instead.
>
> As for your example, I think it should be based on whether the assignment operator selected is trivial.  If it's non-trivial, even if it's defaulted, we should warn about it as a "user-defined" operator, since ultimately it does involve calling such an operator.


To summarize:
https://godbolt.org/g/gwDASe - `A`, `B` and `C` case should be treated as built-in, and `D`, `E` and `F` as user-defined.


Repository:
  rC Clang

https://reviews.llvm.org/D44883