I have been able to use Drools 4.07 in a 3.3 RCP application and face similar
situation - new developers will want to create knowledge support systems for
other domains.
I created a plugin that contains ALL the Drools 4.07 binaries and
dependencies, and I deployed that plugin as a jar file plugin in my
development system and in my PDE target. I did buddy registration just to
be safe, though not sure it has had impact. I do NOT maintain the project
that created this monstrous plugin in my workspace.
I have objects that are declared in some core plugins, and they do not refer
in any fashion to Drools. These objects are extended for the specific
domain.
http://www.nabble.com/file/p20871943/Core.jpg
For the specific domain, I create a new Drools resource, and in the java
source I put the extended object that is relevant to this domain. I also
put all the Drools firing code here (though the firing code will be similar
for other domains). The extended objects are extending objects from other
plugins. But the specific beans that are consumed by the inference engine
are in the same plugin as the rules that fire.
http://www.nabble.com/file/p20871943/Drools.jpg
This works. It took me about six months. The key was to not create a
normal plugin for this portion, but rather to create a Drools project.
My main RCP application is in its own plugin, and depends on these.
Hope this is helpful.
- Mike
--
View this message in context:
http://www.nabble.com/Embedding-Drools-in-Eclipse-tp20814423p20871943.html
Sent from the drools - user mailing list archive at
Nabble.com.