95e13941-faef-4709-a2da-0473495e7d0b
commented
4 years ago Just to make things more complex, I just discovered
http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2019/p1825r0.html https://github.com/cplusplus/draft/issues/3002
The clang (lack of) implementation is documented at https://clang.llvm.org/cxx_status.html#cxx11 inside the c++11 summary since it's considered a DR so applies back to c++11 even though it's a c++20 thing. gcc seems to be tracking implementing it at https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91427 .
Which means there is likely to be another round of this '-in-c++11' as '-in-c++17' madness. And will probably also affect pessimizing-move (note https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89780).
Extended Description
In "D11615 Fix redundant move warning" the redundant-move diagnostic was limited due to clang at the time not implementing CWG1579. Later with "D21619 [Sema] Implement C++14's DR1579: Prefer moving id-expression out of functions" this was implemented, but the redundant-move diagnostic was not updated. In the meantime gcc has implemented a redundant-move diagnostic which does support CWG1579, so the diagnostics are confusingly different without need. Clang's current redundant-move diagnostic would be better named redundant-move-in-c++11.
This is somewhat complicated by the introduction in "D43322 Diagnose cases of "return x" that should be "return std::move(x)" for efficiency" of return-std-move-in-c++11 which is helpful if older toolchains need to be supported. It would be nice if the redundant-move diagnostic were split into redundant-move-in-c++11 and redundant-move since return-std-move-in-c++11 and a full CWG1579 redundant-move are in conflict (by definition).