This seems actually in the last month or so to have been a very big shift. Definitely a good idea to include some of these things in the non-spam corpus, possibly an even better idea to drop the CTYPE_JUST_HTML rule altogether. It used to be a pretty good one, but it seems like it may no longer be so. On the other hand, it might be a good idea to keep it, to encourage people to be better netizens by including a text/plain alternative.
I have been working on some planning-stage stuff for a rsync-based corpus/nonspam.log submission system. I should be able to announce details to beta users sometime soon. It won't be able to deal with huge volumes, but that's what mass-check is for. And the -z flag to rsync. C On Mon, 2002-02-04 at 12:41, Donald Greer wrote: > Folks, > The current scoring for HTML_Only mail may be just a little high. > I've recieved reports that some newsletters (which are html-only) are > being rejected as spam. Specifically I allow my users to signup to news > letters from "cluebie.com" (see "http://austintx.cluebie.com"; if you > want to signup and get a sample... sorry I don't subscribe myself :^) > and those messages have no text/plain part to them. > Perhaps the test corpus might include one or two of these newsletters > as non-spam examples. If the html-only test hadn't been set so high > (4.33) then I think this message might have made it under the radar. > For now, I've whitelisted the clubie.com domain, so that will solve > the problem for _that_ newsletter. > Don _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
