rsmith accepted this revision. rsmith added a comment. This revision is now accepted and ready to land.
Seems fine to me, once you add the test for always_destroy + no_destroy. ================ Comment at: clang/include/clang/Basic/LangOptions.def:311 +LANGOPT(RegisterStaticDestructors, 1, 1, "Register C++ static destructors") + ---------------- erik.pilkington wrote: > rsmith wrote: > > Should be a `BENIGN_LANGOPT` because it doesn't affect AST construction, > > only the generated code. > It does affect AST construction though, we don't mark a static VarDecl's > type's destructor referenced in this mode, or check it's access (because we > aren't actually using it). Do you think this is the wrong choice of semantics? Ah, no, good point. This is right, then. (And I think avoiding checking the destructor is also appropriate, so that a private or deleted destructor can be used to "remind" people to use the attribute.) https://reviews.llvm.org/D50994 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
