Datasources are configured to bind a pool to the JNDI namespace. This is done at the MBean level. I will create a datasource forum shortly so we can put these questions there. The snippet you want to configure with 3.0 is a bit obscure in it's declaration, for now you need to look at the hsqldb example (with distribution) and look for the "driver,url,username, passwd" combo xml snippet in there.
The datasource is deployed dynamically through the deploy directory of your installation (so you can dynamically add/remove them). Check the default installation for a working example. Your datasource will then be available under that name space and you can configure your beans to use that datasource.
Thanks for answering Marc. I think that I'm understanding, but I have a couple more questions (I didn't see a datasource forum yet).
Are some elements of the JNDI namespace application specific, rather than global across jboss?
Am I supposed to deploy an .ear specific datasource from inside the application context? An mbean inside an application? (That doesn't sound right to me.)
Should I provide the authentication information in the JNDI context when I look up the datasource? I assume that this would take advantage of the per-username/password pooling that the pool manager offers. This *does* feel right to me, as it would provide the application scoped authentication that I'm looking for. I just can't find any information on how to do this.
Thanks, I'm having a great time using Jboss,