rjmccall added inline comments.
================ Comment at: lib/CodeGen/CGExprAgg.cpp:1002 + return EmitFinalDestCopy( + E->getType(), CGF.MakeNaturalAlignAddrLValue(Select, E->getType())); +} ---------------- Is there something in Sema actually validating that the comparison types is trivially copyable? Because EmitFinalDestCopy does not work on non-trivial types. Also, are we certain that we're allowed to do a copy from an actual global variable here instead of potentially a constexpr evaluation of the variable reference? And if we are doing a copy, are we registering ODR-uses of all the variables in Sema? I don't think you should worry about trying to generate a select between the "actual" comparison result values. At least not yet. https://reviews.llvm.org/D45476 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
