You of all people should understand some of the compromises made in
order to deal with the existing infrastructure and old versions of
applications that took less than elegant approaches to some of these
problems. Sure, in a perfect world, and under configuration control, I
But the fact is that there are several hundred thousand copies of our
product out there that I need to be able to communicate with and those
versions use the attachment name to indicate which attachment encodes
our custom information. I need to be compatible with those copies in
an environment where I don't have control over the gateway.
But the one doesn't follow from the other! You started this discussion by
saying that standardization of a header that can carry file name information
is urgently needed because you need to have this information in order to make
your applications behave properly. Excuse me, but this header is every bit as
new (in fact its considerably newer) as MIME, and adding support to convert
this header into whatever legacy format you're talking about is not materially
different from converting MIME content type information into the legacy format.
I just don't see any difference in the impact on the installed base between
these two mechanisms. It doesn't matter whether the installed base is one
system or a trillion -- a new mechanism is still a new mechanism.