Finer DCCD/DCCM control

I am using DCCD/DCCM to do greylisting which in itself is very effective 
eliminating spam without a lot resource overhead. In the process of 
doing the greylisting DCCM will return the message checksums. So while 
searching the web I found someone who set their dcc_conf DCCM_REJECT_AT 
to some value (I use 100) and then passed the -aIGNORE switch to DCCM.

The usefulness of these settings is that I can greylist messages as 
usual but don't have to use dccifd or dccproc in Spamassassin to check 
DCC. I just wrote a simple rule to check the X-DCC header added by the 
DCCM milter and assigned my desired score. This procedure all works 
flawlessly, but I am finding that there is some inefficiency to it. What 
I really want is to use the -aREJECT for checksums say >10000 (to pick a 
number) but for checksums between 100 and 10000 I still want the "bulk" 
setting to appear in the X-DCC header and the -aIGNORE behavior 
regarding the message disposition.

This finer control would allow me to reject obvious Spam at the MTA 
level avoiding all the overhead associated with Spamassassin. Yet for 
less obvious, potential spam messages I would still have a "bulk" rating 
in the X-DCC header of the message that could then be used in the full 
Spamassassin scan performed later. I guess I am suggesting two 
thresholds ... one used with -aIGNORE and another used with -aREJECT.

Short of modifying the DCCM code is there any way to get this behavior 
to occur now? If not, is this such an esoteric use of DCC that 
implementing such a feature is not worth the effort or for that matter 
is it even possible to get this behavior from DCCM? Are there any other 
suggestions?

-- 
Paul (ganci@nurdog.com)