Nick Ing-Simmons wrote:
Standards wise that is probably true.
But pragmatically whatever generates the message-id isn't
likely to change how it cases the strings it builds.
Bear in mind that an identifier may have passed through some
gateway, which may have made such a change, identifiers may be
obtained from diverse sources (cid URIs, mid URIs, mailto
URIs, from Content-Type or Content-Disposition parameters,
from Message-ID, References, In-Reply-To, Original-Message-ID,
Resent-Message-ID, Supersedes, or Delivery-Report-Content-Original
header fields, etc.), several of which may involve encoding
of various types (RFC 2231, RFC 2369).
So if case differs message was presumably generated by
at last a different instance of the thing - so it is a different
message.
Not necessarily. One identifier may have been obtained from a
URI via HTTP, having had RFC 2369 encoding reversed; another
may have been obtained from a parameter (having had RFC 2231
encoding reversed) in a field of a message that may have
passed through one or more gateways. Or transcribed from
a spoken conversation (over the telephone for example), or
from written material.