On 12/2/07, Norman Maurer <[EMAIL PROTECTED]> wrote:
Wel my idea why i want to move it into a subproject was that maybe some developer whould find it very usefull to just use a framework for getting a bayesian filter integrate with their software. I think most of developer don't need a full mailserver like james. Thats only my point of view..
ok. so factor it out that it starts standing on its own feet while being integrated with Server at the same time. if it starts to grow out of the Server, let's shift into a subproject. that would be a more evolutionary process. what is the size of the codebase we are talking about?
And still think that most work on the bayesian code should be still done by Vinzenco and Danny cause both of them work on that code and have the background knowing of the alcorithm etc.. I just whould help to move it as the first step ..
sounds like the last step before the first step to me. from what I have seen about the Bayesian code (not that much, actually) is that it looks very tight coupled with the Server. It would be very good if it would be more re-usable. There are use cases I can think of reusing it. Example: One Bayesian data set for every account. Multiple Bayesian filters in sequence. So I am not against having it as a subproject. I am against having it as a subproject now. Bernd Bernd --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
