在存储库中冲洗

Question

In most of the code I see (Spring), it seems no one calls entityManager.flush() from a repository. Is there a reason for that ?

Answer 1

Yes, there is a reason. By default, flush() is called automatically before the transaction is committed, or before executing a query whose result might depend on the not-flushed-yet modifications. So an explicit flush() is almost never needed.

And it's a good thing to flush as late as possible, because it avoids executing queries in case the transaction is rolled back.

Answer 2

Usually, if your JPA code is correctly enclosed in a transaction, flush is done automatically. If you keep a look at logs during code execution you can see:

• update or create operation are not write to the database until:
  • the end of the session (thus end of transaction); thus batching the SQL commands
  • a 'find' operation is executed (because it must be done on a coherent data set by the DB)
• SQL commands can be reorder to be more efficient

Thus, flush command is useful if you want explicit 'flush'.

Animal animal = new Animal();
animal.setName(“Pluto”);
entityManager.persist(animal); /* pluto is saved here */

Owner owner = new Owner();
owner.setName(“Mickey”);
animal.setOwner(owner);
entityManager.persist(owner); /* mickey is saved here */

owner.setName(“Minnie”);
entityManager.merge(owner);
animal.setName(“Mickey”);
entityManager.merge(animal);

/* pluto and mickey are updated here, just before the find query */
Query q = entityManager.createQuery(“FROM Animal a”);