On Thu, 2004-01-15 at 00:49 +0000, Philip Hazel wrote:
> On Wed, 14 Jan 2004, Nigel Metheringham wrote:
>
> > You do need to watch this since routers are a per recipient item, and
> > acls may also be per recipient items, so the header set you are carrying
> > will also become per-recipient before it ever gets to the transports.
> > This could make life very interesting indeed :-
>
> Thanks for highlighting another perhaps non-obvious feature. Or perhaps
> "feature".
Bear in mind that if you implement immediate header mangling you may
also break existing configurations which rely on header changes not
being effectively immediately, such as...
warn condition = ${if !def:h_Message-ID: {1}}
message = Message-ID: <E$message_id@$primary_hostname>
< ... other checks ... >
accept hosts = +relay_hosts
< ... more stringent tests, including ... >
deny condition = ${if !def:h_Message-ID: {1}}
message = RFC2822 says you SHOULD have a Message-ID.\n\
Most messages without it are spam, so your mail has been rejected.
Not that this means I think you shouldn't _do_ it -- just that it's
something that needs documenting clearly in the release notes.
--
dwmw2
