I'm switching a database connection pool from XA to simple JDBC. In the log, I was not seeing the pooled object being reused, though the number of objects in the pool was correct. Digging through the pool code to compare XA to JDBC caused me to pull out most of my already thin hair! I'm amazed at how different the implementation is for these two, and I don't understand the reason why. For example, I finally discovered that JDBC creates a new wrapper every time it hands out a connection, while XA does not. That's why JDBC gives out a new object every time, making it impossible to see what connections are getting reused where. Does anyone understand why such different approaches were taken? Another example is that closing a JDBC connection causes firing of a close event, while closing of an XA connection invokes the releaseObject method.

What is the future for the database pooling code? Is it soon to be discarded in favor of the connector equivalents? If it is going to persist, I'll volunteer to assist in making some mods. I have one I promised Toby I'd make way back in the summer, and I see more needed.