Please don't take me wrong. I thinnk MIME is a great thing,
especially for those who have MIME capable mailers to read AND send
MIME messages.
It is also of value to sites that don't have support for it, given its emphasis
on presentations that attempt to impact the non-MIME installed base as little
as possible. (This is one of classic PEM's major failures, in my opinion.)
This approach is far from perfect, but it works a lot better than any other
scheme that has been proposed.
As a liaison to the user community in the companies I deal with who do
not have the MIME-intelligent mailers widely deployed/used, I see
requiring MIME as an added obstacle for deploying security services to
e-mail. I do realize that for communities (such as yours and
Amanda's) who already have MIME, the situation is reversed.
I'm sorry, but I just don't buy it. Part of the MIME work has been to provide
means to integrate capabilites into existing mail systems in a as seamless
and automatic a fashion as possible. A considerable amount of effort has been
invested on this front, with the result that it is possible to add MIME
capabilities to an enormous number of existing mail systems fairly easily.
The PEM community has not invested anything approaching this when it comes to
integrating PEM into existing messaging environments. As such, I see MIME as a
major facilitator of adding PEM services to existing systems, and this applies
even when we're talking about systems that don't support MIME now. (PEM may
also facilitate getting MIME services deployed in the long run, but we're a
long long way from that point currently.)
As for those without MIME, MIME/PEM is not appreciably harder to implement
than
classic PEM when there's no MIME engine around. As a matter of fact I'd say
it
is considerably easier, given the way things are labelled in each scheme.
I have to disagree but if you provide me the tools to do this easily,
I'd reconsider.
The software I'd use to do this would be MetaMail, which was developed at
Bellcore (thumper.bellcore.com is the FTP site). In particular, there are
standalone tools provided that will take apart a multipart structure for you
and do just about anything you want with it.
As for assembling multipart objects, I'm sorry, but it is so easy to do this
yourself that I don't see any point to leveraging off of someone else's code to
do it.
I completely agree. I tried to implement classic PEM within my MIME mailer,
but
it was just too complex and never worked well enough.
Again, I have to disagree here too. We have classical-PEM integrated
with MIME. We use what Jeff Schiller had also an ID for some time
ago, basically using Content-Type and Metamail. It works fine without
the need to separate PEM body parts.
Well, you must be a better programmer and implementor than I am. I could never
get this stuff to work properly. There were just too many problems in too
many different areas to deal with.
Ned