https://bz.apache.org/SpamAssassin/show_bug.cgi?id=7735
Bill Cole <billc...@apache.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |billc...@apache.org
--- Comment #69 from Bill Cole <billc...@apache.org> ---
(In reply to Henrik Krohns from comment #62)
> Ok so all pending commits are now done.
>
> This bug is still open for:
>
> - Discussion about what to do with "unrun" rules
> - Documentation
>
> Please vote or discuss:
>
> - Keep logic as is, unrun rules may prevent many metas from running
> - Revert to old logic, unrun rules are evaluated as not hitting (0)
>
> I probably lean a bit to the old logic. It doesn't require much changes to
> code, just some tweaking to last do_meta_rules/finish_meta_rules call.
I lean very strongly to the old logic because:
1. It is certain that some meta rules in the wild rely on it by design and will
have their performance changed if we use the new logic. The change won't
necessarily be obvious to people who have deployed such rules.
2. I don't think we understand the changed logic well enough to get it right in
a rush to a major release.
3. It limits the impact of disabling costly rules, because 'unrun' subrules
don't create a tree of "don't run" dependencies.
4. The logic is more obvious to users.
--
You are receiving this mail because:
You are the assignee for the bug.
