http://issues.apache.org/SpamAssassin/show_bug.cgi?id=3109





------- Additional Comments From [EMAIL PROTECTED]  2006-04-02 01:12 -------
(In reply to comment #27)
> undefined score/required?  I really don't like that one, to be honest. The
> X-Spam-Status is mainly a UI but also an API, and I think it's better to
> "degrade gracefully" for apps that won't know about short-circuiting, by
> providing them *something* relatively sane in the areas they're looking
> at, instead of breaking their expectations like that.

Even though it's the most accurate, I don't really like it either -- that's me
being used to safety critical systems where we never ever rely on a value that's
supposed to be there being there.

I'd much prefer that whatever causes the shortcicuit to have fired a rule that
will be certain to cause a final score that exceeds the threshold in the
appropriate direction, and we use that final score.


> option (d), same as (c) but with a higher value for clarity:
> 
> X-Spam-Level: ******************************************
> X-Spam-Status: Yes, score=100.0 required=5.0 
> tests=BAYES_50,HTML_IMAGE_ONLY_28,
>          HTML_MESSAGE,MIME_HTML_ONLY,SC_TEST shortcircuit=spam
>          autolearn=no version=3.2.0-r372567
> 
> 
> me, I like (d), with (c) as a second-best.

I'd settle for (d) but with a (default) score outside the range of typical
scores.  The score should be configurable though, as I believe it is now, so
that people using multiple thresholds for tag-accept/reject can still do that.



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

Reply via email to