[Idrs-dev] EJB Integration...

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

Hello Everyone,

I am revisiting the idea of integrating EJBs into the IDRS.  I have
greatly simplified the process by rellying on the current substructures
instead of redesigning it.  I have posted a new EJB Integration paper on
the site.  http://idrs.sourceforge.net/dev/papers/ejb/index.html is the
link, for your convinience here is the text:

------------------------------------------------------------------------
IDRS EJB Integration
The purpose of this paper is to describe a process for integrating EJB's
into the IDRS.  The goal of integrating EJBs into the IDRS is for the
IDRS to become a front end to an existing application.  In order to have
the smoothest integration and usage process, the IDRS' substructure
should be utilized.  The IDRS sub-structure is based on JDBC.  This
substructure currently handles resource managment and presentation.
This substructure can be utlized in the following ways:
      * Utilize the current Connection pooling framwork
      * Utlizie the current Object and Report Pooling
      * Allow for EJBs to be used from any context where a datbase would
        be used, including the <head> and <body> sections
      * Allow the IDRS to handle as much of the connection and
        resolution process as possible.
These feature can all be used without changing the fundamentals of the
IDRS.  There will be two purposes for using EJBs in the IDRS, either to
report data or handle data entry.  In either case, it can be assumed
that Session Beans, be it statefull or stateless, will be used as the
primary interface to the EJB server. =20

In the case of reporting, there needs to be a way to use the IDRS'
presentation tags and structure to display the data.  There are several
ways data is brought back from an EJB.  The three most common seem to be
either a single bean, a Collection of beans or a RowSet.  The RowSet is
easiest to handle, becaus eit implements ResultSet.  The single EJB or
the Collection of beans can be handled by a special implementation of
the ResultSet class that can use reflection to map bean properties to
fields in a database table.  By handling the beans in a way that is
consitent with a JDBC ResultSet, we will not need to change the internal
structure of the IDRS.

For data entry/input, IDRS DataObjects can be used as a go-between from
the IDRS application and the EJB server.  DataObjects can populate a
skeleton object for passing to a session bean.  Since transactions are
managed by the container, there is no need for the IDRS to handle the
managment. =20

The last piece is to handle connections to the EJB container.  This can
be accomplished with two special implementations of the
java.sql.Connection  interface.  One for none-jndi based connections and
one for jndi based connections.  These Connection objects will store the
Context refrence and be responsible for instancing EJBs.  This will
allow the standard pooling substructure to be utilized.
---------------------------------------------------------------------------=
----

Any thoughts would be greatly appeciated.  Happy hacking

marc

--=20
Marc Boorshtein <mbo...@ya...>