[BUG] Class<?> inference is confused about capture<?> · Issue #2256 · projectlombok/lombok

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement . We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Describe the bug
Title sounds a bit weird, but please bear with me.

Lombok will infer Object#getClass as Class<capture of ? extends Object> instead of Class<? extends Object> or something similar. If I then use a Class#getSuperclass , I get Class<capture of ? super capture of ? extends Object rather than Class<capture of ? super Object> or such.

To Reproduce

val obj = new Object(); // doesn't matter, any type works
var type = obj.getClass();
var super_type = type.getSuperclass(); // this causes a compile error, however IntelliJ isn't bothered
Main.java:8: error: incompatible types: Class<CAP#1> cannot be converted to Class<? super Object>
        var super_type = type.getSuperclass(); // this causes a compile error, however IntelliJ isn't bothered
  where CAP#1,CAP#2 are fresh type-variables:
    CAP#1 extends Object super: CAP#2 from capture of ? super CAP#2
    CAP#2 extends Object from capture of ?
Expected behavior

var super_type will be inferred to at the very least Class<?>.
Version info (please complete the following information):
Lombok version; 1.18.10
Platform: javac 1.8.0_222
          The problem here is that the signature of getSuperclass() of Class<T> returns Class<? super T>. Which makes sense. The type of the superclass of T has an upper bound of T.
However, if I have a variable of a certain type T, the value can be any subclass of T. So the type of getClass() is ? extends T. Now the lower bound is T.
We would need Class<? super (? extends T)>. That doesn't exist. The only thing that we could generate would be Class<?>.
For the resolution we depend on javac to give us the correct type. We're not sure if at that point in time we have the right information to fix this problem. We cannot replace all ? super X results, we should only do that it X is itself a wildcard (? or ? extends X, ? super X would be fine).