I actually don't think it currently makes sense to alter the existing implementation. We need something different (intermingled Decl, Stmt and later other nodes), and we don't need some of the information the other ParentMap provides.
We might be able to pull of a common abstraction, but that abstraction would need to depend on DynTypedNode. I also wouldn't think having this code here is a deal breaker, as it is an implementation detail to the hasAncestor functionality, and thus will not taint any interfaces. http://llvm-reviews.chandlerc.com/D36 _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
