> I would expect messages with policy POL1 to be scanned only through > SpamdClient; but what really happens is that both Spamassassin and > SpamdClient are called, and their scores added (looks like global > setting is applied). Looks like a bug to me.