After having abandoned my work on Bayesian analysis late last year I am now working on this again, and have been for a few weeks now. I resumed work on this because a response to a question I asked on a forum here led me to the mixdist R package which has now enabled me to model kernel density estimates for a naive Bayes classifier. For more on using a kernel density estimate in a naive Bayes classifier see the link in the answer here.
I expect that it will be a few weeks yet before all the coding and testing of this is complete.