Tuesday, August 29, 2006 4:27 PM
So I have this collection of code that is a complete unit. It provides translation services from one language to another language. It does need some configuration data. It is currently setup with the assumption that it will exist in a MicroKernel (mK) architecture. The total number of components that it adds to a given mK is about 8. I am fascinated with the concept that this could be deployed as a Facility, Castle's parlance. However, I am not sure if the best way to reuse this code is as a Facility or to just load up all of the components seperatly. I hate writing all of the code/config to load these up and manage changes for each of my apps.
I am also in love with the concept of deploying this code into a seperate kernel and then placing a proxy in the main app mK. Is this sane? It makes it a bit easier to share Facilities between kernels (ie NHibernateIntegration ).
I could also instanitate a completely seperate mK and then hide every thing behind a singleton, but that just makes testing harder. And where would I put the config data? What about duplicate facilities?
Hmmm....