So are we finally going in for #3? If yes, we must drop v1.1 plug-ins before we release GEP 2.1.0 as most of them may not be working as expected!
Hi, The JAXB refactoring of the GEP 2.1.x code is almost complete for the 2.0.x and 2.1.x versions of the Geronimo servers. Most major functions are now working and we are much better positioned to handle future schema changes in a more timely manner. Traditionally, the GEP has supported 3 to 4 versions of the Geronimo server (primarily to provide a migration/upgrade path), and we had originally planned on supporting v1.1, v2.0.x, v2.1.x. However, since we are almost 2 months behind the release of the v2.1 Geronimo server I would like to discuss some possible alternatives for supporting the v1.1 Geronimo server in this release of the GEP:
#1. Proceed with the JAXB refactoring work for the v1.1 code (obviously the most expensive in terms of time and testing required)
#2. Leave the v1.1 support in the current EMF implementation (i.e., the JAXB and EMF implementations would co-exist)
#3. Remove support altogether for v1.1 in this release of the GEP -- support only the v2.0 and v2.1 Geronimo servers (the least expensive in terms of time and testing required)
I'm now of the opinion that we should pursue alternative #3 and remove v1.1 support entirely. My primary rationale is that the the old 2.0 release of the GEP can still be used to provide v1.1 server support, and still provides a migration path from v1.1 to v2.0. It's true that we would lose the v1.1 to v2.1 migration path, but this is mitigated somewhat since the support in the GEP for the v2.0 and v2.1 versions of the server is almost identical. Equally important is that we could then focus entirely on fixing the few remaining JIRAs and augmenting our JUnit testcases, and release the GEP 2.1 quicker (i.e., in the next week or 10 days). Thoughts ??