On Mon, 13 Feb 2006, Tony Finch wrote:
> Exim's rewriting code doesn't understand Resent- headers, which is
> a bit of a problem for widen_domains and rewrite_headers. Also,
> verify = header_syntax doesn't check them for syntax errors and
> verify = header_sender doesn't verify the addresses they contain.
>
> I'm not sure what is the right thing to do: should Resent- headers be
> treated as synonyms for their normal counterparts, or should they have
> distinct header type codes, or when Resent- headers are present should
> they get the normal codes with other codes for the normal headers?
> The rewrite flags also need extending to cover Resent- header, but they
> use different letter codes from the header types.
The treatment of Resent- headers has always been a problem area, mainly
because I don't understand them. A search of all the ChangeLogs shows
various bits of faffing around over the years...
I think Exim did once do the header_sender thing on them, but seems to
have changed.
IIRC, I eventually took a stand on this statement from RFC 2822: "Resent
fields are strictly informational." You can argue that that means the
MTA should ignore them. However, Exim doesn't quite. See section 43.5.
Hmm. It claims that it does rewrite them...
--
Philip Hazel University of Cambridge Computing Service
Get the Exim 4 book: http://www.uit.co.uk/exim-book
