xroche wrote: > I don't want to be intentionally incompatible with MSVC, but it seems hard to > be compatible. Probably we could work out the rule with enough effort; > there's a limited number of things that could reasonably be checked. I'm okay > with approximating it as "cases where the callee calls a destructor for an > argument", since that's what should actually matter.
[This MSVC example](https://godbolt.org/z/nM6E53jPs) shows that checking whether the called function destroys an argument gives the opposite index order from MSVC for both `CopyOnly` and `DtorOnly` :worried: Do we want that simple rule anyway, or should we investigate a closer match to MSVC's evaluation order? Matching strictly MSVC behavior will be tough I think but :shrug: https://github.com/llvm/llvm-project/pull/215479 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
