type_path of constructors for inner classes

Alex Buckley alex.buckley at oracle.com
Tue Oct 15 14:36:21 PDT 2013

On 10/15/2013 2:20 PM, Joel Borggrén-Franck wrote:
> I am assuming you mean "lets not change anything to the encoding of
> the receiver parameter" by your comment to leave it out of this.

I meant "Let's leave the receiver parameter out of the business of 
emitting TYPE_USE annotations on the simple type name indicated by the 
ctor declaration."

> I'm not sure what the consensus is but I think it looks like you both
> think it is a good idea to add the location to the inner class
> constructor. As an extra justification, from my standpoint everything
> gets much simpler if type annotations on Inner has the same location
> encoding (one level of INNER_TYPE in the example of class Outer {
> class Inner {}} ) in all cases where it can occur.

Exactly. I'd expect a METHOD_RETURN [location=INNER_TYPE] and a 
METHOD_RECEIVER and that's it.


More information about the type-annotations-dev mailing list