http://issues.apache.org/SpamAssassin/show_bug.cgi?id=4947





------- Additional Comments From [EMAIL PROTECTED]  2006-06-07 22:36 -------
Unfortunately, this is invalid logic in software development.  You don't just
throw out the baby with the bathwater and start over.  The point to new version
is to keep compatibility with older versions (including older rulesets). 
Whether or not you make the older compatibility is there by default is another
matter.  To drastically change the rules in such a way as to score an email as
4.5 and then score it .9 with it in a completely different version makes no
sense at all.  Basically, you're saying that 2 years ago it would have nearly
been spam, but today it isn't.  That makes no sense.

If anything, the same email should have scored at least as much, if not even
higher.  Not lower.  For example...

0.2 HTML_50_60             BODY: Message is 50% to 60% HTML (2.63)

This one isn't even in 3.1.3.  Why?  The email still has 50% to 60% html.  And
this information is still useful (even if it the score had been reduced to 0.0).
 Why is it missing?

By changing it this drastically, you have now fractured the Internet and made
this tool unreliable.  Meaning, if some installations are using 2.63, then an
email could be marked as spam and junked.  Where, in other installations with
3.1.3, it would be allowed through.  This is not a good situation.  Versions of
software such as this should remain consistent so that anyone using it to test
can get an accurate picture of problems in a given email.

Right now, this has really made spamassassin a useless out-of-the-box utility
for judging the quality of an email and now forces each user to create their own
personal rulesets instead of using the supplied ones.  If you're going to go
this route, you might as well just delete all the rulesets and force everyone to
configure it with their own personal rules.



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

Reply via email to