Consider this case:
class A {}
class B<T extends A, E extends T> {
B<?, A> b;
B<?, ? extends A> b2;
}
As I understand type bounds, in this case effective upper bounds of both T and E is class A. So the question: why javac doesn't accept class A as argument in declaration of field b, but accepts wildcard ? extends A in declaration of field b2?
With the following classes:
class A {}
class C extends A {}
class B<T extends A, E extends T> {}
Think of it like this:
E extends T extends A
With B<?,A> then T -> ? and E -> AA extends ? extends A
Where the ? could be any subclass of A, let's say C.A extends C extends A is clearly invalid.
So that's why it's a compile error.
Eclipse 4.9.0 compiler disagreed with javac 8u and Intellij and did not emit a compile error for the generic arguments in B<?,A>. I assume this is a bug in the Eclipse compiler but I have not consulted the JLS to confirm this.
class B<T extends A, E extends T> {
B<?, A> b; // <-- Eclipse does NOT emit a compile error
B<?, ? extends A> b2;
}
This assumed bug has been reported here.
Your declaration is incorrect. You are missing the point of 'wildcard'-s. They are used for unknown property declaration. For B<?, E> b the E must extend the T and the T must extend the A but you said that the first generic type is ? which is unknown! So you said that an unknown parameter must be extend T and so A. It is incorrect.
Relationship you created is like E -> T -> A. After that you declared as a 1. generic type ? -> E -> T -> A and 2. generic type as A -> E -> T -> ?. So ? must extends ? and A must extends E vs.. It is confusing and unknown for compiler...
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With