On Tue, Dec 10, 2013 at 3:42 PM, Richard Smith <[email protected]> wrote: > Maybe also remove the old (and now unreachable) check from > `CheckFunctionDeclaration`?
It's still reachable for ABIs where static and non-static member function have the same calling convention though. > This slightly degrades our diagnostic in this case: we get (one error + one > note) per overridden function, rather than one error + (one note per > overridden function), but I can live with that. We could also go the other way actually. We could suppress the error from CheckOverridingFunctionAttributes if the new function is static and rely on the existing check in CheckFunctionDeclaration to throw an error. This means we'd be adding a method to overridden_methods even when the CCs don't match, but I guess that doesn't matter since the function is static and we error out anyway. - Hans > http://llvm-reviews.chandlerc.com/D2375 _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
