...
- Avoid default name and version (for each template/node/data type, etc.. name and version have to be mandatory)
- Java code to manual check for each table if name and version are already used
Conclusion
- Move to Hibernate with EclipseLink tables is not possible due the presence of ElementCollection not used properly.
- The presence of longblob types used to store whole objects is definitely not so maintainable. The fix for those issues could be done in both Eclipse-Link and Hibernate. Before to apply the fix it needs to check the max length of varchar fieldsis an issues, it is readable only by Java language or by any other language compatible with Java binary code. I am not sure if a db-migrator can handle that.
- Document databases store all information for a given object in a single instance in the database, and every stored object can be different from every other. So, using MongoDB/Cassandra will solve all issues.
JpaRepositoryandMongoRepositoryextendCrudRepositoryand expose the capabilities of the underlying persistence technology in addition to the rather generic persistence technology-agnostic interfaces like e.g.CrudRepository.
...