Did you actually know that there are no peer / object classes for the
ID_TABLE used by the IDBroker of Torque?

Do you want to know the reason why? You can't build them! Try it. If
you do, you will notice that there is a clash between the constant for
the TABLE_NAME column (which is called IdTablePeer.TABLE_NAME) and the
TABLE_NAME constant which contains the Name of the table (IdTable).

This quite largeish patch does two things (besides being quite intrusive
and not much tested):

- It prefixes all the columns from the ID_TABLE with "ID_TABLE_". So TABLE_NAME
  becomes ID_TABLE_TABLE_NAME and NEXT_ID becomes TABLE_NAME_NEXT_ID

  Yes, you must rebuild your database scheme after rebuilding your SQL/OM
  with this patch applied. 

- It adds (besides reworking the build process and removing a few bugs)
  torque generated peer classes and a mapper to the torque.jar. These are
  in org.apache.torque.oid.om. If you actually want to _use_ the Peer, 
  you must make sure that it registers correctly with your database map
  (it has database "default" hard coded) so you won't get any clashes
  (Torque itself patches up some sort of "virtual aka hacked" table map
  for the ID_TABLE of each database, which then in turn contains the 
  right database name instead of being hardcoded on "default". This wants
  more testing!

This patch definitely is 3.2 material. Please test. It applies only if
you have applied the java-columname patch before it (else you will get 
rejects in TorqueInstance).