HiveMind 1.2 progress
Howard Lewis Ship <hlship <at> gmail.com>
2006-02-06 02:51:54 GMT
Well, there hasn't been much, we've been focusing on 1.1.1.
I'm not satisified with my attempt to convert to Maven2 (in my branch).
But the experiment is useful, I'm going to take another crack at it,
on the trunk. I'm going to remerge 1.1.1 changes into the trunk, then
convert to Maven2 (but leave all the source folder locations as is,
for the meantime).
On the other hand, I'm getting into crunch mode with a client, so I
won't be able to do a lot with HiveMind for the next few weeks.
I do have a number of ideas I want to pursue for 1.2:
- <module> attribute to control the default builder factory
- A streamlined, smarter injection factory
- <interceptor-sets> ... a way to apply a set of interceptors to many
services (potentially, across many modules)
- Some kind of negotiation between the service extention point,
service lifecycle model, and service implementation builder to handle
negoation on the lifecycle model (i.e., to allow it to be determined
via an annotation on the implementation class), and to handle the
intracacies of event notification support for non-singleton models.
In addition, I want to start introducing an alternate approach to
creating services, one that invokes Java code to build the service
implementation. This may be based on annoations and/or naming
conventions. I see this ultimately as a way to reduce the amount of
XML in the system, make HiveMind more refactoring friendly, and
improve startup time for complex environments like Tapestry.