efriedma-quic wrote: Looking a bit more, I'm not sure I buy the premise that we don't need to mangle lambdas written in unevaluated contexts. Consider the following related example:
``` template<class Tp> using simd_vector = Tp; template<class VecT> using simd_vector_underlying_type_t = decltype([]<class Tp>(simd_vector<Tp>) { return 3; }); template<class VecT> auto temp() { return [](simd_vector_underlying_type_t<VecT> x) { return x(10); }; } int call() { return temp<simd_vector<char>>()({}); } ``` Both gcc and MSVC seem perfectly happy with this. https://github.com/llvm/llvm-project/pull/124572 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits