In my limited understand, SMTP is a store-and-forward protocol. Will
this affect your proposal? I imagine it would not be pleasing to
click on something you expect to be interactive and have it respond 4
hours later. Maybe I should go read your draft more thoroughly --
you probably already thought of this.
It is certainly possible an interactive reply from a distant mail server
could take 4 hours so liberal a time to live were allowed. I don't see
that as being common, however. I want to say that 10 - 15 years ago
something like this would have been more than the norm than the
exception. Today, however, many servers on the application layer of the
transmission stack, whether they be web, email, or other, typically sit
on high bandwidth networks connected to other high bandwidth network.
In my case specifically I am currently in Afghanistan, which is perhaps
an isolated communication point for communications that live mostly in
the United States. Dispite the probably delays associated with my
isolation and distance my response to and from other email servers
appears instantaneous. If I misstype the user name portion of an email
address and send the email I will have a reply from the distant server
about no such address almost instantly. This is dispite being limited
to a 128kbit connection. If in my extreme condition bandwidth is not an
issue then I fail to see how such an issue could be common enough to
provide any real service problems.
Further to the point the dynamic responses detailed in the specification
would represent fragments of a document and not a full communication, so
large amounts of bandwidth would only be consumed at the initiation and
not in the interaction. This should be equivocal to the bandwidth
demands of a XMLHttpResponse on the web.
Does this answer your question?