Oct 27, 2010 kl. 6:53 AM skrev Gabor Greif:

> What about a test of a final struct with
> a baseclass that defines a virtual function
> but the derived class does not override it?
> That should also be devirtualized.

Yeah, that's next on my list.

> Also, did you think about local classes (in
> functions) or classes of TU-scope (i.e.
> in a provably non-included .cpp file or
> anonymous namespace? I know these are much
> more hairy, but IDEs might prove such things.
> 

Nope, not yet. I don't think there's a way in clang right now to "get all the 
classes in the TU that derives from class X" but it might be worth adding.

- Anders


_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to