Message11926

My shot at this is now in the repo at: https://hg.python.org/jython/rev/78b574d6a6f7

It passes tests on Java 7 cleanly. On Java 9 I see 12 test failures but we had those before. I believe I have eliminated the warnings that stem from PyReflectedFunction.

Also, I claim the first overt use of a MethodHandle in Jython. I *think* I had a good reason to use it :)

I remain a little dubious about the logic of PyJavaType.init, as it seems to me that the addition of handleSuperMethodArgCollisions as an afterthought sits awkwardly. What it does (and I've extended) is necessary, but maybe the whole of our field and method processing should work that way? For example, I think it would not deal as it is with a private class implementing List directly because handleSuperMethodArgCollisions would not be called for it. Fortunately, in all cases I've come across, the implementation extends AbstactList, for which handleSuperMethodArgCollisions will have been called. Similarly Set, Map, etc., but I think we could find/make a failing example. However, that could be an enhancement after 2.7.2 .