Re: UTF-8 revision

[Ned Freed <Ned.Freed@INNOSOFT.COM>]

> Registering UNICODE-1-1-UTF-8 is much better as it doesn't cause
> compatibility problems with MIME readers.  The long ugly name is also good
> since we *really* want to discourage its use.

> I don't like "charset-edition" as defined in RFC 1922.  In order for it to
> function interoperably with changing character sets, it would require a
> reset of MIME to proposed standard so that all MIME MUAs could be required
> to support it.  I think that's a horrible idea.

Specifically, MIME says that a charset defines a mapping from octets to
characters. The minute you use something like charset-edition to distinguish
between two versions of Unicode with different code points it becomes part of
what's necessary to determine the right octet to character mapping, since
without it a given octet could map to two or more characters. Having to change
a core piece of MIME like this would necessarily require a reset to proposed.

> Now a "charset-subset" parameter would be quite useful down the road as
> characters are added.  Clients have the problem that the installed fonts
> may not have all the characters in the latest 10646/Unicode.  A
> "charset-subset" advisory parameter (e.g., "amend5" subset only uses the
> subset of 10646 range defined in 10646 + amendments 1-5) could be useful.
> But it wouldn't be necessary for interoperability.

Right, because no ambiguities develop in the mapping that charset defines.

				Ned