I just read it.
The problem is that it uses character sets other than ASCII and
Unicode. This is a serious burden, since it essentially requires
that every application in the world maintain character set
conversion tables.
If this were a new application protocol, I'd agree with you. But this
burden already exists for MIME mail readers, so by using existing MIME
charsets we're not imposing an additional implementation burden beyond
the need to parse the new parameter sytnax.
At some point we might get consensus to deprecate the use of
everything but Unicode in email. But I suspect it will be a long time
before this happens.
In the meantime, we'd only be increasing implementation complexity to
have one standard for charsets in displayable text, and another
standard for charsets in parameters.
Besides, we'd never get consensus to impose such a restriction.
Actually, it's worse than that; it uses charsets
which are even more disgusting than character sets.
The MIME notion of 'charset' was right (and is still right) for
text-based mail which is intended for display. 'charset' is probably
not right for the representation of filenames, but the differences are
so subtle that trying to explain them to most people only increases
the confusion.
Keith