Perhaps this has been discussed before, but I was wondering if it might make sense in many instances for SA not to invoke non-local agents if the local spam score is already above the spam threshold? Thus, if this new configuration dependent mode was enabled, SA would only check RBL, razor, pyzor, what have you *if* and *only if* the current score (derived from local sources, and non-local sources consulted so far) is below the spam threshold.
Given that local checks are probably better than 90% effective, then a conditional non-local mode might reduce non-local queries by perhaps 90% or more? ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
