I like the idea where this is going but am not 100% sold. Not have used the Digester myself (but just brwosing thru the examples), here is my question.
1. So how many commons libraries that we will need?
2. How does it handle configurable xml elements? For example, each cache loader would require different elements and types. Do you need to build the rule on the fly? If that is the case, do we leave it to the Cache loader provider to build it?
using Apache Digester for this usecase is not worth it. The only place in Jboss codebase where Digester is used, is in a tomcat (or more so JBossWeb) contribution by a user,to parse the context.xml under WEB-INF.
Look at the XB work Alexey has done. He has the pieces needed for parsing xml into ur object model.
Just a suggestion.
The only reason I picked Digester was that it was an XML-> Object model tool that I have used extensively in the past and know fairly well. I'm not particularly loyal to it and anything else that does the job (preferably with fewer dependency jars) would work.
I'll look into JBossXB.