On Mon, Jun 07, 2004 at 08:18:38PM +0200, Santiago Vila wrote: > On Mon, 7 Jun 2004, Andreas Metzler wrote: > > > I've enabled tagging xbl in spamassassin more than a month ago [...] > > SpamAssassin is not a reliable way to tag messages, because it > considers IPs in *all* Received: headers, not just the single one that > has to be considered (the one that contains the real IP from which the > spammer sends his crap). Umm... I think you're wrong there. There is code to guess which IPs are meant to be trusted and not. If it consistently fails to do this, you either need to fix your configuration or it's a bug. (Try the trusted_networks and internal_networks options) SpamAssassin may check more headers than are necessary, i.e. it will check forged headers added by spammers, but that can only improve it's hit rate and reliability, not decrease it. > Try using a more sophisticated script in perl or some such. More sophisticated than SpamAssassin? In perl? (I'd like to see you try to write a more sophisticated perl script than SpamAssassin...) I tried hard not to jump into this flame war again... We've heard it all too many times before. The problem is essentially that not all developers have the same preferences regarding spam filtering, and thus everyone should simply take it upon themselves to do their own filtering. Also, this is the only way to ensure that debian-admin is not responsible for any false positives. (Sure they could simply tag messages but that's no more useful than filtering on your own.) -- Duncan Findlay  Please don't actually file a bug unless it's present in the version of spamassassin in experimental.
Description: Digital signature