Unfortunately, the majority of MIME dispatchers fall into this category. So
much so that they cannot be ignored. You can complain about their brokenness
all you want but it isn't accomplishing anything.
Then if someone wants to dispatch iotp, and have it recognised as
something better than just xml, he will upgrade to a better dispatcher.
But he is only 1 in 1,000,000 of the worldwide users of email, and so the
rest of them can carry on using the old dispatchers for many years without
problem.
That illustrates a general principle with regard to user agents. If there
is some new functionality on offer, the marketplace will surely provide
products to support it. Those that require the new functionality will
happily buy the new products. Those who don't need it won't bother.
Again, this is entirely at odds with reality. For one thing, upgrade decisions
are vastly more complex than this. So what happens is that someone who tags
their materal with a generic XML tag and a parameter identifying the specific
variant effectively loses the ability to dispatch to their handler on a
majority of platforms. This leaves them with two choices: Use a specific tag
and get immediate gratification or wait for an upgrade that might possibly be
real for some small subset of their customers, someday, maybe, hopefully. The
answer is, of course, obvious.
We've seen numerous cases of this play out in MIME already, as well as in
numerous other areas of standardization.
You can afford to take decisions that require people to use the proper
user agents.
Sorry, but there is overwhelming evidence that this is simply not correct.
What you can't do, OTOH, is to take decisions that break the
underlying transport mechanisms - those indeed takes many years to filter
through.
Nobody is proposing any such thing.
Ned