[PATCH] D104500: [clang] Apply P1825 as Defect Report from C++11 up to C++20.

[Matheus Izvekov via Phabricator via cfe-commits]

mizvekov added a comment.

In D104500#2866079 <https://reviews.llvm.org/D104500#2866079>, @rsmith wrote:

> I wonder if we can restore the extension behavior for the types that want it, while still using the C++98 rules for the rest of the cases? Perhaps we could use the new C++11-to-C++20 logic even in C++98 if the return type declares a move constructor?

Well if we want to keep supporting it, then my plan would be:

It comes back, but it's not exactly the same "C++98 implicit moves" that you used to know, it now has a few cybernetic enhancements:

- As far as the rules of what id-expressions are considered move eligible, we keep the same ones as C++20 has. The old one used to apply only to the subset where we can do copy elision.
- We jerry rig the first overload resolution, by iterating the construction steps and aborting it in case we pick a constructor or conversion operator which does not take an rvalue reference.

I think this way would be a balance where we don't add back too much code, but at the same time we keep compatibility with pure C++98 programs. I think this is in line with the idea of best effort.

Though if I may suggest, if we ever hope to sunset C++98, then to stop providing these C++11 extensions would be a good first step...


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D104500/new/

https://reviews.llvm.org/D104500