Pete Resnick wrote:
And the diff from -05
Thanks. Stuff I stumbled over while looking at the diff:
---- 1: <obs-domain-list>
RFC 821 uses the syntax:
| <a-d-l> ::= <at-domain> | <at-domain> "," <a-d-l>
| <at-domain> ::= "@" <domain>
RFC 2821 and 2821bis agree with that idea: There are no
"null" <at-domain>s. Why do 2822 and 2822upd keep the
RFC 822 ASCII-art commas in source routes, when that was
never allowed in (2)821(bis) ?
---- 2: Routes SHOULD be ignored
| When interpreting addresses, the route portion SHOULD
| be ignored.
That statement in isolation *after* the syntax is odd,
why not add it to the prose before the ABNF ? Proposal:
= There are four primary differences in addressing. First,
= mailbox addresses were allowed to have a route portion
= before the addr-spec when enclosed in "<" and ">". The
= route is simply a comma-separated list of domain names,
= each preceded by "@", and the list terminated by a colon.
+ When interpreting addresses, the route portion SHOULD be
+ ignored.
+
= Second, CFWS were allowed between the period-separated
= [...rest as is...]
At the moment 2821bis still says "SHOULD be ignored", with
"MUST be stripped down" only for error notifications. It
would be nice if 2821bis and 2822upd say "MUST be ignored",
but of course that's mainly a problem in 2821 and 2821bis.
---- 3: <obs-group-list>
That's a kind of <obs-mbox-list> without <mailbox>, but
at least one comma. Anything else is covered elsewhere.
Rather obscure, how about renaming it, <obs-empty-list> ?
You could use <obs-empty-list> three times, where you
have it as <obs-group-list>, and to simplify <obs-bcc>:
= obs-bcc = "Bcc" *WSP ":"
- address-list / (*([CFWS] ",") [CFWS]) CRLF
+ address-list / CFWS / obs-empty-list CRLF
Ditto in <obs-resent-bcc>. The case "only CFWS, no comma"
needs an explicit CFWS, the same idea as in <group-list>:
| group-list = mailbox-list / CFWS / obs-group-list
---- 4: obs-dtext
RFC 821 uses the syntax (for domain literals):
| <element> ::= <name> | "#" <number> | "[" <dotnum> "]"
| <dotnum> ::= <snum> "." <snum> "." <snum> "." <snum>
In essence RFC 2821 and 2821bis agree with that idea, no
control characters in domain literals. Why do 2822 and
2822upd keep the RFC 822 <obs-NO-WS-CTL> in <obs-dtext>,
when it was never allowed in (2)821(bis) ? Did the DRUMS
Charter a decade ago force 2822 to keep RFC 822 errors ?
IMO you can limit <obs-dtext> to <quoted-pair>. Nobody
produced cases where "MUST accept obs-NO-WS-CTL in domain
literals" is REQUIRED (as specified in RFC 2119 1 and 6).
---- 5: obs-phrase
- period character is currently used in many messages in
- the display-name portion of addresses
+ period character is sometimes erroneously used in
+ messages in the display-name portion of addresses
Most users have no problem to use a <quoted-string> where
it is required, <atom> simply cannot permit trailing dots.
---- Last Call now ?
Frank