> > > [12.5] IDN MUST NOT return illegal code points in
> responses, SHOULD
> > > reject queries with illegal codepoints. (one request to
> add; one request
> > > to remove)
> >
> > Undefined here is "illegal code point"; which ones
> are illegal?
> > Should talk about malformed CEFs, not so much about
> "illegal" code
> > points. Both UTF-8 and UTF-16 can be "malformed". Still not
> > fully defined though.
>
> The 'illegal' codepoints here does not refer to invalid or
> malformed UTF-8. It
> refers to codepoints which we feel should not be part of or be used in
> hostname, for example, punctation.
>
> What this statement say is that irregardless what illegal codepoint we
> decided, the protocol should not make sure restriction on the wire.
Aha. But then please rephrase (avoiding the word 'illegal'):
IDN MUST NOT return non-IDN characters (to be defined) in
responses, and SHOULD reject queries containing non-IDN characters.
I still see a problem with this, in that the UCS is open, and
new characters will be added to it for quite some time.
/kent k