John Rose skrev: > It would let you emulate the super(...) call in a language which needs > to make a subclass of some Java class with a non-trivial constructor, > without forcing the language to generate bytecodes for that subclass's > constructor call. When I think a little bit more about this, if you want to call a constructor through a method handle, you will have to use the uninitialized referenced (which is a reference created with new but no constructor has executed on it yet) as an argument to the invokevirtual call. The bytecode verifier will prohibit this. Both in a plain method and in a constructor.
The bytecode verifier also prohibits calling a constructor on an already initialized reference. With a MethodHandle to a constructor it would be possible violate this restriction. I am not sure this is a good idea. //Fredrik _______________________________________________ mlvm-dev mailing list mlvm-dev@openjdk.java.net http://mail.openjdk.java.net/mailman/listinfo/mlvm-dev
