The current org.jboss.cdi.tck.tests.extensions.lifecycle.processBeanAttributes.specialization.VetoTest also verifies that a ProcessBeanAttributes does not get fired for AnnotatedTypes which are likely to end up as disabled Bean.

There are 2 problems with this behaviour as outlined in CDI-581
a.) the whole approach is not deterministic as it has an implicit chicken-egg problem
b.) the spec wording "if the class is an enabled bean,.." is simply wrong. There is no 'enabled bean' at this early stage!
This might have been a copy&paste error while moving the wording from ProcessBean (which has no veto and thus is deterministic) to ProcessBeanAttributes (which has a veto and can also amend other enable-relevant information).

Probably it's also a mixture with ProcessAnnotatedType? It should rather test if AnnotatedTypes which got vetoed (via annotation and also via PAT#veto()) does not result in a ProcessBeanAttributes.