I just want to make sure I understand the problem you're encountering.
I'm particularly concerned with multipart/mixed wrappers around
multipart/digest - clients not specifically designed to deal with
them (notice I am not singling out any specific large corporation
here) could (and, reportedly, do) create some serious havoc with
these messages. (I kinda like the way Roger Fajman reported LISTSERV
handling it, by forcing the TOC to a message/, though. Might be
interesting to set up a test list that did that.)
Are you focused on clients that do not understand multipart/digests, or
are these clients more fundamentally broken? Specifically, do these
same clients handle multipart/mixed inside of multipart/mixed?
If they handle mixed inside of mixed, but not digest inside of mixed,
then they are broken because that is not compliant. I realize that's
not particularly satisfying and certainly doesn't do anything for the
problem you need to solve. I'm just trying to understand the problem
space.
What's probably happening is they are treating the entire content-type
of multipart/digest as unknown and using the MIME default for unknown
content-types of application/octet-stream to handle it. What they are
overlooking is that there is a separate conformance requirement for
unknown multipart sub-types, that being treat it as multipart/mixed.
Again, this doesn't help you because you have a user community to
support and you have to meet their needs. However, I think it is useful
to at least understand the problem.
Thanks,
Jim
