Hi.

Say I have these entities:


In other words, an A "has" many B's and a B "has" many C's.

Can I do this?



I was expecting to, but it seems I'm wrong. It would mean I wouldn't have to write a boring EJBQL queries or do iterations just to get the C's for an A.

If you can't, how do you usually handle those cases in your apps? If I write a method in A to retrieve the C's, it would mean I have to load and iterate the entire graph of B's first, which doesn't sound right. Retrieving them through queries in session beans is clumsy and plus you lost the OO approach advantages (like being able to write #{myA.cs} wherever I need those in the presentation pages).

TIA

What OO approach?

What you want is a total non OO approach.

Use simple HQL:
select distinct c from A a, B b, C c where c.b = b and b.a = a and a.id = ?

Say a User has many Accounts, each of which has many Mailboxes, each of which has many Folders, each of which has many Messages.

I supposed I could have something like this on User:



and I was kinda surprised that I can't, because it would have been much more convenient than writing queries in backing beans (or iterating the entire graph programatically). But I guess it was just wishful thinking.

... try using named Queries



Hibernate should take care of all joining and this is still an easy to use feature.