The problem was solved by upgrading to v2.2.0a4. > -----Original Message----- > From: Vincenzo Gianferrari Pini > [mailto:[EMAIL PROTECTED] > Sent: Friday, June 13, 2003 3:05 AM > To: James Users List > Subject: RE: Problem with implementing Redirect in 2.2.0a3 > > > Chris, > > It could be another thing. > > I'm using a modified version of your antispam bayesian analyser > in my production server since 2-3 months now, and I'm too > redirecting to [EMAIL PROTECTED], using v2.2.0a4. > > The redirected message goes through all the antispam analysis > and, unless a check either on the sender being (i) either the > postmaster, or (ii) on the sender being local, or (iii) on the > recipient being [EMAIL PROTECTED] is done, the message goes to an endless loop. > > I have to do check (ii) after the bayesian analysis but before > switching to the spam processor (as I consider messages coming > from inside as not being spam but I want to allow a "rebuild spam > corpus"), plus check (iii) right before the analysis: all this > precisely to overcome the problem you exposed. So I think that > the behaviour is as expected. > > Now, why the previous Redirect mailet seemed to behave > differently? I may have introduced an unwanted change: I've been > very careful in avoiding that when I rewrote the mailet, but you > never know; I'll do another check "by eyes". But I think that the > fix to bug #18028 (see > http://nagoya.apache.org/bugzilla/show_bug.cgi?id=18028) is what > changed the behaviour: using the previous Redirect mailet, when > inline was "all" or "body", the redirected message had a fake > body and so the bayesian analyser was not considering it as spam; > now it contains the spam body, and you get the loop. > > Let me know your opinion about all this. > > Vincenzo > > P.S. > > 1) You do a Redirect to [EMAIL PROTECTED] with inline=all (and implied > attachment=none), I do a Forward (inline=unaltered and > attachment=none). Which one should we do, or better, what do they expect? > > 2) I'm a very happy user of the bayesian analyser - it is super! > I did extensive changes to it, mainly extending it to Paul > Grahams' Better Bayesian Filtering > (http://paulgraham.com/better.html) and fixing two small bugs > that had a *big* impact on the behaviour. Why don't we get in > touch (offlist) and exchange our fixes. We could also try to > build something to offer to the James distribution. > > Bye again, > > Vincenzo > > > -----Original Message----- > > From: Noel J. Bergman [mailto:[EMAIL PROTECTED] > > Sent: giovedì 12 giugno 2003 20.01 > > To: James Users List; [EMAIL PROTECTED] > > Subject: RE: Problem with implementing Redirect in 2.2.0a3 > > > > > > Chris, > > > > > However, this redirected message now reenters the mail > server, and goes > > > through all the processors again, matches as SPAM again, > > renters the SPAM > > > processor, and the cycle continues. > > > > That was fixed in v2.2.0a4. > > > > See: > > http://cvs.apache.org/viewcvs/jakarta-james/src/java/org/apache/ja > > mes/transp > > ort/mailets/AbstractRedirect.java.diff?r1=1.3&r2=1.4&diff_format=h > > > > --- Noel > > > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: [EMAIL PROTECTED] > > For additional commands, e-mail: [EMAIL PROTECTED] > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > >
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]