Weird ruleqa numbers for dnswl

[Da...@chaosreigns.com]

http://ruleqa.spamassassin.org/?daterev=20110122&rule=%2FDNSWL

  MSECS    SPAM%     HAM%     S/O    RANK   SCORE  NAME   WHO/AGE
      0   0.0111  17.8052   0.001    0.88    0.00  RCVD_IN_DNSWL_HI  
      0   0.0294  12.1266   0.002    0.81    0.00  RCVD_IN_DNSWL_MED  
      0   0.0235   6.4141   0.004    0.80    0.00  RCVD_IN_DNSWL_LOW  
      0   0.1320  20.2586   0.006    0.70   (n/a)  __RCVD_IN_DNSWL  
      0   0.3296  10.2721   0.031    0.58    0.00  RCVD_IN_DNSWL_NONE  

I believe __RCVD_IN_DNSWL should include the rest of these rules?  So
assuming HI, MED, LOW, and NONE have the right numbers, __RCVD_IN_DNSWL
should be .3936% of spam (not 0.1320%), and 46.6180% of ham (not 20.2586%)?

Re: Weird ruleqa numbers for dnswl

[Da...@chaosreigns.com]

On 01/23, Darxus@chaosreigns.com wrote:
> I believe __RCVD_IN_DNSWL should include the rest of these rules?  So
> assuming HI, MED, LOW, and NONE have the right numbers, __RCVD_IN_DNSWL
> should be .3936% of spam (not 0.1320%), and 46.6180% of ham (not 20.2586%)?

That could be my fault.  I'm not actually using the normal dnswl rules
- I'm doing dnswl tests at postfix, inserting a header based on the
results, and then I have my own dnswl SA rules that use those headers.  It
should have exactly the same results, except I have no __RCVD_IN_DNSWL.

So for my own purposes, I can just sum up HI/MED/LOW/NONE instead of using
__RCVD_IN_DNSWL.  And stuff should otherwise be the same.  Does anybody
else care?

-- 
"When we remember we are all mad, the mysteries of life disappear and
life stands explained." - Mark Twain
http://www.ChaosReigns.com