test and fix for issue #33: "ghost" type parameter in field

[web-online]

I punted on the "primitive" bit since it seems the ResolvedPrimitiveType wasn't intended for object primitives like Integer.class

[cowtowncoder]

Thanks! I'll merge and have a look; sounds safe enough, although I wish I understood why the ghost type is introduced in this case.

Also: correct, Integer is not a primitive type. It's wrapper type, but from classmate perspective that's just a non-interface non-primitive type, and nothing fundamentally special. Although it might not be a bad idea to have some means to check for wrappers as they do have special relationship with actual primitive "buddy" types...

[cowtowncoder]

Ok, I generalized the fix slightly, since it seemed that same should apply to interface types as well.
And after change, noticed one of unit tests failed; looking deeper realized test had been producing (and happily accepting...) ghost types, so had to fix the test instead.

Now: this will be fine for 1.3.3 (which I'll release shortly), but I have a nagging feeling that this actually points to a more general problem. It is possible that type binding given to leaf types might be incorrect, and so although this fix will remove ghost types for non-generic types, I wonder if there wouldn't be cases of partially resolved types that would exhibit either ghosts or even possible wrong ordering.

Anyway... for now things work so we'll cross the bridge (wrt possible add-on work) when get there.

[web-online]

Nice, thanks! I wasn't sure how far to take it, but that generalization makes sense to me. I hadn't realized it before you said it, but I had a similar nagging feeling about leaf types. I agree that this is the best solution for now.