11 Replies Latest reply on Jul 14, 2007 8:04 AM by Vicky Kak

    DataSource/ConnectionFactory programmatic deployment

    Weston M. Price Master

      Ok, so I have been working on


      And I am starting to think about things (please note the date and the time because it doesn't happen very often).

      Basically the structure, using XB, is there in the connector module on HEAD, the skeleton of the DsDeployer is also there, based upon the existing RARDeployer. XB ugliness aside, I am starting to wonder if this is really the way we want to go.

      Currently, the XSLT deployer while strange, does a fairly good job in dynamically generating the MBeans to bind a ConnectionManager, Pool etc, etc to an 'OldRarDeployment' (stupid name, but there it is). Again, while this conceptually seems bizarre, it's working, manageable and seems to fit the model we currently employ in mapping a datasource file to our RAR set. This is largely a bi-product of the way we allow datasources/connection factories to be deployed and configured. A non-gui approach as it were.

      The programmatic new stuff, bascially is going to be doing the same thing just with a known meta-model. Use of XB in this case is a bit challenging, since we are not using a predefined XSD for our DS deployments, but more than that, the programmatic process is just going to be doing the SAME thing that the XSLT does now.

      I have been looking at the JCA POJO prototype, and it would seem to make more sense to sort of jump headfirst into generating deployment info that the MC/POJO JCA can understand, rather than trying to back-port something into the 4.x codebase. We know we need MBeans in 4.x anyway, why reinvent the wheel in this regard?

      Any thoughts on this please let me know...I just get this sinking feeling that I am doing something wrong, or not understanding the scope of what this new *-ds.xml mechanism is supposed to provide.