There is a special scenario with bridge methods where Weld will fail to correctly generate the subclass (e.g. to actually decide to intercept a bridge method).

This can be simulated by taking our existing InterceptorBridgeMethodTest and adding the following method:

    @Test
    public void testSpecialBaseAsInterface(SpecialBase specialBase) {
        // SpecialBase extends Base<String>
        reset();
        specialBase.invokeA("foo");
        verify(SpecialParent.class);
        reset();
    }

More details can be seen in Quarkus issue - https://github.com/quarkusio/quarkus/issues/14696
And the solution will have to be very similar as well - https://github.com/quarkusio/quarkus/pull/14767

In a nutshell, we'll have to try and detect this situation where the compiler generates a bridge without an implementation in the same class and then sign it up for interception.
The detection will always be best-effort, but it's still better than nothing.