Regarding the subject: Maybe we can change it to "JPA schema update not working if reserved keywords are used as entity class names" as The initial creation works fine but an update triggered on top of that breaks.

In Spring Data JPA the additional update is not tied to the very same bootstrap but basically one test case causing Hibernate to be bootstrapped and causing the initial schema creation. Now a second test case re-uses the in-memory database instance that already has a schema but bootstraps a new EntityManagerFactory on top of it with the hibernate.hbm2ddl.auto set to update. This is a quite widely used scenario and I'd argue it's a valid one.

Anyway, thanks for taking care of this. Much appreciated!