A few questions regarding AdminApp

erik.jogi · **Posted:** Mon Jan 19, 2004 6:34 pm

Hello,

I am reading the AdminApp sources (0.9.1). First of all - a brilliant piece of work - very easy to follow. Thanks Gavin for this excellent sample of Hibernate+WebWork2.

A few questions that remained a bit unclear for me:

User and Role have their IDs defined as java.lang.Long. However in the HBM there is this:

Code:

      <id name="id" column="USER_ID"
         unsaved-value="0">
         <!-- primitive int, so use 
            unsaved-value="0". -->
         <generator class="increment"/>
      </id>

Shouldn't the 'unsaved-value' be 'null' in this case?

What is actually the recommended way: have IDs as primitives or objects? What are the implications of either approach?

Also regarding caching:

In Role.hbm.xml therw is:

Code:

      <!-- there are few roles, and they are  
         not updated so cache them -->
      <cache usage="read-only"/>

If you define this 'read-only' but nevertheless write some objects (as is the case with roles) what happens? No problems if it is not a cluster?

Also in hibernate.cfg.xml there is:

Code:

      <class-cache 
         class="org.hibernate.admin.model.Role" 
         region="Role" 
         usage="read-write"/>

I couldn't find anything in the docs about this. Isn't the 'read-write' conflicting with the Role.hbm.xml?

best regards,
erik

christian · **Posted:** Tue Jan 20, 2004 12:16 pm

The identifer unsaved value is indeed wrong, thanks for spotting this. It has no effect in the current application (no detached objects), luckily. There is actually not much difference between a primitive and an object identifier type, just use whatever you like.

The cache setting in hibernate.cfg.xml overrides the setting in the mapping file, I added a comment. The read-only strategy works fine as long as you only _save_ objects, but never _update_ them.