On Sat, 2007-12-15 at 12:50 -0800, Dean S. Messing wrote: > : I am having a problem with spamassassin registering a false positive. I > : have tried to whitelist (unjunk) a particular individual whose name? > : always seems to be junked. They are a good friend and I don't want to > : keep losing their mail. I am using Evolution 2.10.3 with spamassassin > : 3.2.3. > : > : If someone knows how to solve this problem I will stop here. > : > : If I can get a SA gui from the Fedora Repo that doesn't interfere with > : Evolution, I will try that. I prefer a gui for this job because it's > : not something I will have to do often. > > I use GUIs for very little and I don't use evolution at all so this > may be of no help. But I suggest you look in > ~/.spamassassin/user_prefs to see if it contains a line like > > whitelist_from your_friend@your_friends_domain > > On my system the presence of such a line starts my SPAM score out at > -100 for that incoming message. Mail must pass a threshold of 4.5 to > be tagged as spam so the mail must contain a wopping lot of indicators > for this to happen. (It never has, so far.) Here's the Spam-Status > line from a message from a Whitelisted person on my system: > > X-Spam-Status: No, score=-101.9 required=4.5 tests=AWL,BAYES_00,NO_RELAYS, > USER_IN_WHITELIST autolearn=disabled version=3.2.3 > > Note the presence of USER_IN_WHITELIST. Does yours have that? > > It my case the Bayesian test pulled the score below -100, because it > judged it to be non-spam from other factors. But even a Bayes score > of "100% spam" only adds 4.3 points to the total (on my system). > (Being entirely HTML adds another 4.5.) > > : Can someone suggest a spamassassin gui that I can download from the > : Fedora 7 repo and that will fix this problem? The spamassassin site > : recommends several front ends but I couldn't find any of them on the > : Fedora add programs list (pirut). > > It's easy to edit your user_prefs file and insert that one line under > the comment. > Evolution does not uuse the options of user_prefs except I think the baysian database. -- ======================================================================= The 357.73 Theory: Auditors always reject expense accounts with a bottom line divisible by 5. ======================================================================= Aaron Konstam telephone: (210) 656-0355 e-mail: akonstam@xxxxxxxxxxxxx
