Yep, makes sense, LGTM thanks! Test case could probably be reduced a bit further (the templates don't seem relevant, I think all you need is a struct type where constant emission produces an LLVM type different from the type that we'd naturally produce -- a union might be a more reliable way to create that) and folded into another CodeGen test, but I'm happy with this either way.
http://reviews.llvm.org/D5353 _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
