Keith Moore wrote:
>> > >HTTP is an application by itself. TCP/IP is not.
> > >IPP is trying to layer one application on top of another.
Actually, I've long been of the opinion that while the lines are blurred
somewhat that current HTTP is really more a combination of Session and
Presentation layers - it is used by web browsers to impelement interactions
between web browsers and web servers, but it is clear that it is being
adopted for other purposes as well.
The request/response components of HTTP form brief 'sessions' - associations
between application layer entities carried over a transport. Note that HTTP
persistent connections do not even all carry requests from the same
application endpoints (as in the case of a proxy combining requests from
multiple clients to an origin server).
MIME is really a form of presentation layer - it provides the metadata that
enables the applications to communicate the form of the data being
exchanged.
Only the cache control features of HTTP are really application functions.
It is my hope that the HTTP-NG effort can clarify the distinctions between
the session-like, presentation-like, and application-like features, despite
the fact that the Internet protocol family has historically behaved as
though it was anathema to call anything above TCP anything but an
application.
> > In a sense you are right, but in effect IPP is sending MIME
> > packages over HTTP. This has been our stragegy for more than a year.
> > Is MIME over SMTP considered a separate application that needs
> > it's own scheme? It wasn't last time I checked.
> The belief in this case is that IPP is different enough from
> ordinary HTTP traffic that the two should be distinguished.
But that doesn't change the fact that the IPP group has been quite carefull
not to alter or extend HTTP in order to do so - they have in fact layered
thier work.
--
Scott Lawrence Consulting Engineer <lawrence at agranat.com>
Agranat Systems, Inc. Embedded Web Technology http://www.agranat.com/