I really don't think that MIME has to be more specific about the
interpretation of the right-to-left languages than the ISO character set
The reason is same as why we shouldn't have full "ISO-2022" as a MIME
charset, that is, interoperability.
ISO character set specification is not intended to be interoperable
without a lot of prior negotiation. National variant characters of
ISO 646 and Han unification of ISO 10646 are such examples.
People using ISO-8859- will most certainly exchange bi-directional
text. So, why not assure interoperability?
Just as the case with 7 bit 8859, I don't think I'm adding something
new. It's just a clarification of the defacto standard and the
agreement at Houston meeting.
My take on this is that the next MIME RFC should simply say that
the effect of escape sequences (or any format effectors other
than a CR LF pair) are undefined for the charsets it defines.
For charsets that require bi-directionality, define them in
a separate document.
That's effectively equivalent to refering RFC1556 and should be OK.