On Mon, 28 Jun 2004, jfcm wrote:
There are two things you may want to modify with an OPES:
- the content - and this is protocol transparent
- the addresses. Like changing a name from ascii to chinese.
Agreed. In other words, there is message data (payload/content/etc)
and meta-data (headers/trailers/etc). And, as we have already
established while working on OCP Core, the definition of each and the
boundary between the two are application- and adaptation-specific.
What's content for some may be metadata for others.
I am no specialist of SMTP either, but I do not see how you can
change the address and not resend the mail in a way or another. This
is not an OPES since the modified data would not come back.
Not sure what you mean. My understanding is that an SMTP callout
service can, in theory, change a From: or To: address and everything
will work just fine as long as the corresponding OPES processor (SMTP
server) is capable of handling such a change. Thus, this is still
OPES.
There is a political question of whether this WG can document such
adaptations (they smell like moving towards URI resolution that IAB
does not like), but that's a different question.
So, apparently you are right, this can only be work on the content.
Things like From/To MIME headers are probably common among all
MIME-based mail protocols. Things like transport connections and
authentication are probably different among those protocols. The
question is whether the common subset is large and autonomous enough to
be adapted without regard to uncommon protocol-specific things.
Alex.