EricWF marked an inline comment as done. EricWF added a comment. In https://reviews.llvm.org/D37035#849574, @rsmith wrote:
> I don't like the model of conditionally rebuilding the default initializer / > default argument if it contains one of these builtins; it seems more > heavy-handed than necessary. I'm also not convinced that we need to store the > computed value in the AST. > > As an alternative, `CodeGen` and the expression evaluator can track a current > call context (the innermost call expression or list initialization that is > not within a default argument / default member initializer) consisting of the > enclosing `FunctionDecl` and `SourceLocation`, and when the value of one of > these builtins is needed, pass that context to the AST node to ask what value > it has in that context. This will require a bit more work for the static > analyzer, but should be fairly simple elsewhere and will avoid the extra AST > nodes for the desugared form. That sounds good to me. Could you suggest where said context should correctly live? ================ Comment at: docs/LanguageExtensions.rst:2118 +point is the location of the caller. When the builtins appear as part of a +NSDMI the invocation point is the location of the constructor or +aggregate initialization used to create the object. Otherwise the invocation ---------------- rsmith wrote: > What is an NSDMI? I think you mean "default member initializer". "Non-static default member initializer". I'll change it to your suggestion. ================ Comment at: include/clang/AST/Expr.h:3838 + /// function. + const char *getBuiltinStr() const LLVM_READONLY; + ---------------- majnemer wrote: > StringRef? Sure. Why not. I'm never sure how to handle string literals. https://reviews.llvm.org/D37035 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
