The open-sourcing of Terracotta DSO has been one of the most exciting recent events in the Java world for me. With the help of their engineers I looked into making Terracotta work seamlessly with RIFE and more specifically RIFE's native Java continuations and its dynamic template system. This integration will soon be widely available, so stay tuned!
As I looked closer at the approaches of Terracotta, I realized that they come very close to what I've been striving for with RIFE:
- no serialization,
- reliance on Java language keywords to drive the core functionality,
- ease of configuration and
- non-intrusiveness
... and a lot of this is achieved through byte-code rewriting. I also was very excited to see that Terracotta took the open-source model seriously and wasn't just dumping their product on the world without properly following up. Public documentation is collaboratively being written for open-source usage and support through IRC and email has been top-notch. You can feel that everyone involved is very passionate and genuinely motivated to improve their product through the feedback of a larger community.
Native Java continuations fail-over and clustering has been one of the trickiest parts for RIFE users to date and I want to be part of developing the de-facto solution for POJO-based clustering. That's why I'm joining Terracotta as of April 2nd 2007.
Of course I'm still continuing to work on RIFE, to maintain solutions for Uwyn customers, and to talk about various Java-related subjects at conferences ... including Open Terracotta!
