Murray S. Kucherawy wrote:
Michael Thomas wrote:
What's the big deal about having cfws after every tag/value? That makes
each element completely symmetric and thus easier to parse.
As Tony mentioned, the current ABNF already allows that; there's CFWS
just about everywhere between tokens.
Why? It used to be last. I never lost sleep over it one way or the
other. Making the order irrelevant definitely makes the parsing
easier though.
Since it was Tony that advocated for that change, I'll let him defend
it... ;)
My first guess though would be that each result is contained between
";" characters, so you could easily skip forward looking for one you
want without having to consume entire blocks of the header.
The concern I have with making ws the delimiter is that in real life, is
that
some things in the mail stream will sometimes rewrap lines, and they
sometimes
do it at non-ws boundaries. I really wish that we had made dkim immune to
fws insertions by keeping nowsp for headers, but illegal for the body.
making
auth-res immune to the same manglers seems like a prudent thing, imo.
Mike, gnu mailman's python smtp library being a prime culprit
_______________________________________________
NOTE WELL: This list operates according to http://mipassoc.org/dkim/ietf-list-rules.html