[Bug 6527] mkrules erroneously omits nopublish rules from masscheck

[bu...@bugzilla.spamassassin.org]

https://bz.apache.org/SpamAssassin/show_bug.cgi?id=6527

John Hardin <jh...@impsec.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jhardin@impsec.org

--- Comment #10 from John Hardin <jh...@impsec.org> ---
(In reply to Justin Mason from comment #7)
> (In reply to comment #6)
> > 5 - There are rules like T_SURBL_MULTI2 that are in mkettler's sandbox.  There
> > is currently no reason I can see for why these are being published.  This is
> > NOT an error in his sandbox.
> 
> If I recall correctly, T_ rules *can* be dragged into updates if they are
> required as dependencies of a published meta rule.

This zombie thread is perhaps a better place to make this comment than a new
bug...

I just noticed that if a T_ rule *does* get published, it gets published
*without* an explicit score - so if the rule definition in the sandbox has a
score limit that's less than one, the T_ rule starts adding a full point to the
message's score when it hits, when it would add fewer points if it were
properly published.

I see two possible approaches to fix:

(1) If a T_ rule *does* get published, and the score limit is < 1, an explicit
score for it should also be published at that score limit.

(2) T_ rules should always be published with an explicit "informational" score
(0.0001) so that they still can be used in metas but they don't materially
affect the overall score.


Should this be a new, separate bug?

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