On Fri October 29 2004 11:08, Keith Moore wrote:
Structured fields (as far as RFC 2822's definition of structured
vs. unstructured fields is concerned) are primarily used by
message handling programs (not only clients, but e.g. IMAP
servers, MTAs, Sieve filtering language (RFC 3028), etc.).
Fields intended to be used only by humans are generally
unstructured (again, in the RFC 2822 sense), e.g. Subject,
Comments, Content-Description.
Yes, this _tends_ to be the case, because fields designed to be parsed
by computer programs need to be structured, while fields designed
primarily for human consumption have less need to be structured.
However it's not a rule, it's just the way things usually work.
In practice it's often the case that we want a field to be both
human readable and usable by computer programs. I think that would
have been true for Archived-At even if we had started out with
the understanding that it was originally intended for human use.
We can distinguish general "computer programs" from those
that specifically handle RFC [2]822 messages. While it certainly
seem to be true that it's desirable for the URI conveyed to be
usable to *some* computer programs (viz. URI handlers, a.k.a.
browsers), it is not necessary for it to be part of some structure
which is handled by RFC [2]822 message-handling programs.
I.e. the URI can be conveyed in an unstructured field as far as
RFCs [2]822 are concerned, while still being a URI.
Since there is no defined role for message-handling programs,
the field body plays no role in message protocols. If the URI
is used by humans for pasting into other programs (browsers),
then as far as messages and massage-handling programs are
concerned, it plays no role in any message-related protocol.
It is a convention, rather than a protocol. As a convention,
it doesn't matter whether the URI is conveyed in an unstructured
field, as a comment in some specific structured field, as body
content, embedded in some existing field, etc., so long as the
particular convention is understood by interested parties.