[Mimedefang] SA 2.51?

[David F. Skoll]

On Fri, 21 Mar 2003, Robert Covell wrote:

> Not sure if everyone is on the SA mail list so here is the link for a patch
> to make SA 2.51 work.  I have SA 2.51 working with 2.30 just fine with the
> patch applied below.

Hmm... this reinforces my belief that the Bayesian code is not
production-ready.  The BayesStore and Bayesian modules are extremely
complex and fragile-looking; I expect the SA team to be chasing
locking issues for weeks or months.

Does anyone on this list have a feel for how practical it would be to
get the Bayes module to use an SQL database rather than Berkeley DB?
Feasible?  Totally whacky idea?  A proper database with transactions
and row-level locks might ease most of the locking issues.

--
David.