Hello Rob,

Tuesday, August 17, 2004, 10:28:19 PM, you wrote:

RB> On Tuesday 17 August 2004 7:27 am, Robert Menschel wrote:

>> I agree with Loren's statement about the version difference. You have
>> Bayes enabled (per autolearn=no), but no BAYES results. It's possible you
>> lost your Bayes database.
RB> It did get borked from a nasty uncaught message. I got rid of the two bayes
RB> files in ~/.spamassassin about a week ago, before upgrading. Tonight, I
RB> uninstalled all the packages ran rpm -qa |grep spamassassin searching for
RB> more, then I deleted ~/.spamassassin. And reinstalled SA 3.00. We shall see
RB> if that fixes it. But not a lot of mail tonight.

Good luck.

>> I also don't see any network tests in your results. I find network tests
>> make the difference in 1/3 to 2/3 of my spam.
RB> I have no flipping idea what you are talking about.

Network tests are those which query network servers to determine whether
they might think the current email is spam(ish). See
the links under "Make use of other anti-spam projects" at
http://wiki.apache.org/spamassassin/FrontPage 

>> Header-0 and Header-1 should have kicked two of the three you posted over
>> your threshold.

RB> I am having another problem when RDJ attempts to ---lint the
RB> rulesets, I get a ton of errors: 

RB> Lint output: warning: description for SARE_RECV_IP_212164 is over 50 chars
RB> warning: description for SARE_MSGID_EMPTY is over 50 chars
RB> warning: description for RM_hm_ShortMsgid12 is over 50 chars
RB> warning: description for SARE_MULT_HEAD_LC is over 50 chars
RB> warning: description for RM_hm_EmtyMsgid is over 50 chars
RB> warning: description for SARE_TOCC_CONS6s is over 50 chars
RB> warning: description for SARE_RECV_IP_080178 is over 50 chars
RB> warning: description for SARE_FROM_NUM_8DIG is over 50 chars
RB> warning: description for T_RATWARE_ERROR_04 is over 50 chars
RB> warning: description for SARE_RECV_SUSP_3 is over 50 chars

RB> What do I do with this? Why the problems?

All of those but T_RATWARE_ERROR_04 are caused by long descriptions in
the (current) HTML and HEADER rule sets. I've already corrected those in
my development copies, and hope to publish the corrections soon. Jesse is
working on Ratware, so that problem should go away soon also.

(I realize "soon" isn't soon enough for those who have already migrated
to the not-yet-in-production 3.0.0.)

Chris T -- is it possible to enhance RDJ so it examines the --lint
output, and maybe have a switch that allows processing to continue if the
only output are warnings?

Bob Menschel



Reply via email to