I just say that it should be registered, for example, as
charset=iso-10646-sanskrit-japanese-utf2
or
charset=iso-10646-hindi-chinese-utf2
What's wrong with that?
What are the defining documents for these two different character sets?
If someone really want to use 10646 with MIME, the document should
appear as RFC on how to profile it.
If someone really want to use ASCII with MIME, the document should
appear as RFC on how to profile it. Or, the MIME RFC might be able to
refer the charset ASCII in RFC1345.
Or, don't want to use exclamation marks and know it will be displayed
as nothing other than exclamation marks?
My understanding is that 10646 claims to define only one.
No, it explicitly allows subsetting. So, the easiest way to profile
DIS 10646-1.2 is to remove Han and Devanagari (and some other, maybe).
Yes, DIS 10646-1.2 have provided only one set of code points to Han or
to Devanagari, that's why we need further profiling to get correct
rendering of characters.
On Han unification, see section 25 of the DIS.
There, differently shaped characters are list and assigned the same
codes.
Thus, as the section 25 of the DIS clearly demonstrates, you can't
display characters with unified code without further profiling
information.
Just as multiple instances of ISO 646 should be given different "charset"
names, it is necessary to give different "charset" names, if you are
applying 10646 to languages of unified characters.
Masataka Ohta