On Feb 12, 2010, at 4:50 AM, Neil Booth wrote: >> +// CHECK: b: 1 >> +// CHECK: b: 2 >> + > > Chris - forgetting Clang's implementation, I cannot agree with > the comment and rationale here. The reason that no include guard > optimization is valid is because the header has a non-empty #else. > The status of the macro on initial inclusion is irrelevant.
I agree with you, but: a) we don't currently see the #else because of the current implementation of the MI optimziation. b) this is not a case that matters in practice, fighting to catch it is not worth it. -Chris _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
