I notice that when retrieving the 1 bean in a bidirectional 1:M relationship using findByPrimaryKey, all the PKs from the M (many) side are fetched.

Using Load Groups or some other technique, is it possible to defer fetching the foreign keys for the built-in findByPrimaryKey method?

My concern is, the foreign key result set could be very large for some relationships. Just prefetching the PKs (which could be composite keys) could present a performance problem.

TIA