On Mon October 18 2004 09:56, Charles Lindsey wrote:
Ergo there is no necessity for <id-right>s to be compared
case-insensitively
No, in the first place it is recommended to be a domain name,
which is always case-insensitive. In addition, it has always been
a domain name (and remains so under the current full Standard).
Moreover, there are various circumstances (explained at length)
where case of the RHS may be changed, and only a case-
insensitive comparison will work correctly under those (not
unusual) circumstances.
and the change introduced by RFC 2822 is a welcome
simplification of the protocol.
It is not a simplification; it complicates matters since it is
no longer possible to determine whether or not the RHS is
a domain name, and if it is not, there is no indication of
whether or not it is case-sensitive.
RFC 2822 certainly does not state, imply, or suggest that
the RHS should be treated as case-sensitive. So there is
nothing in that RFC to support your unconventional opinion
that case-sensitive comparisons should be used for identifier
domains. Indeed, it defines id-right as dot-atom-text
(dot-atoms are used in case-insensitive domain names),
domain literals (IPv4 literals are numeric, so case-sensitivity
is not an issue; however IPv6 literals may have hexadecimal
digits, which are case-insensitive), or as traditional RHS
domain names (case-insensitive as always) or domain
literals with CFWS). The sole difference between the
dot-atom construct used in a domain and the dot-atom-text
specified for id-right is the optional presence of CFWS in the
former, and in fact RFC 2822 explicitly states that that
difference in CFWS is the difference between its definitions
of msg-id and angle-addr. Absence of CFWS does not imply
case-sensitivity.