Bruce Lilly wrote:
It's also noteworthy that use of domain literals is strongly
discouraged, accompanied by SCREAMING ALL CAPS in 822
Yes, it's funny. But the idea is clear, today we would use a
SHOULD NOT + a statement about the terminology + a normative
reference to 2119.
OTOH you and Keith think that TLD .invalid is an obscure idea,
if that's true it justifies to ignore "SHOULD NOT" for any(_at_)[]
But there's the 2821 ABNF that doesn't permit that.
The RfC 2821 ABNF also doesn't permit localhost, tv, and other
FQDN host names without a dot. No problem for any(_at_)[], because
that's the idea of this construct. Or at least not worse than
any(_at_)thing(_dot_)invalid until users and UAs know these conventions.
Your "updates RfC" section could be a new record.
[...]
I'm not sure what you mean; making the field optional will
update 822 and 2822.
Sorry, I confused this with your chapter 3 (scope) and its
nine (?) affected RfCs not counting "etc.". It's also a pain
for mail2news gateways, for message/news, for the processing
of cancel messages in net news, "etc.".
Bye, Frank