One small quibble/request...
In section VII, please exclude any use of this (or any other) encoding
in "trace" fields (i.e. Received and Return-path), even in comments made
parts of those fields. This is the transport boundary, and can cause
nothing but trouble, with little or no benefit that I can discern.
And a suggestion or three...
Wrt ISO 8859-n/Latin-m character sets, this proposal would be enhanced
by some guidance as to which one to use if there is a choice (there
usually is) in the spirit of "don't encode ASCII" at the end of section
VII. There has been good practical experience with the "use the lowest
value of N that will do the job" rule that the Kermit folks use, unless an
intended side-effect of this proposal is to require that all mail readers
support all ISO 8859- variants.
Incidentally, unless we wish to be accused of "single-digit-bias",
the ISO 8859 group is, I think, up to -11 at least. Might be wise to do a
little block-reserving, e.g.,
0 Future extension for 10646
1-50 (probably enough) ISO 8859-n, 1 <= n <= 50
51 ... (future extension).
Question...
I have an extremely vague recollection of languages whose written forms
use question-markers at the beginning and end of questions, effectively
quoting them. I don't see that this causes any problems for the proposal,
but it might make the advantages of leading "?" as a marker a bit less.
Has anyone looked at this?
Editorial...
Insofar as ASCII can be said to be a subset of ISO8859-1, it is equally
a subset of all other ISO 8859-n character sets, present and future.
Otherwise, I'm quite happy with this personally and think it is a very
clever bit of work, arriving in the nick of time.
--john