I'm inclined to think that there are relatively simple solutions here,
but none so simple as deferring the problem. There is ABSOLUTELY NO
REASON that a followup-to-MIME document can't clarify this issue. And
there's a strong reason for doing it, which is the desire to get MIME
out the door. The relative importance of 8-bit file names versus
standardization of MIME is very small.
In particular, if this has to get resolved IN the base MIME document,
I'm extremely prejudiced in favor of the "all file names are always
7-bit ASCII" solution because it is simple & clear & backward-compatible
with the MIME software that's out there. This doesn't necessarily mean
that it's the best solution. But any other solution will either delay
MIME or go into a separate document. I vote for giving ourselves time
to think it through outside the context of the main MIME document, which
can remain restricted to ASCII file names until some document supersedes
that part of it.
Imagine, for example, that this issue hadn't arisen until the NEXT
meeting of the IETF. Then we'd be solving it in a completely separate
document out of necessity, and that would be fine. -- Nathaniel