There are two separate issues:
1) Does a registration allow the 'charset' parameter or not.
2) Does an actual entity have the 'charset' parameter or not.
It is not totally clear to me which one of these you are talking above.
1)
In my opinion, it is highly preferable that all registrations
allow the 'charset' parameter, to avoid a patchwork. The draft
should contain some justification for this.
RFC 3023 already provides a registration template which introduces the
charset parameter. But the recommendation to introduce the charset parameter
is a SHOULD rather than a MUST. People read RFC 3023 but they intentionally
dropped the charset.
I studied existing +xml media types registered at IANA.
Here is the result.
1. IETF tree
With the exception of application/cnrp+xml, all +xml media types
have the charset parameter.
1) beep+xml [RFC3080]
This keeps the charset parameter.
2) cnrp+xml [RFC3367]
This omits the charset parameter. Not restricted to UTF-8.
3) cpl+xml [RFCXXXX]
http://www.ietf.org/internet-drafts/draft-ietf-iptel-cpl-08.txt
This keeps the charset parameter.
4) pidf+xml [RFC-ietf-impp-cpim-pidf-08.txt]
http://www.ietf.org/internet-drafts/draft-ietf-impp-cpim-pidf-08.txt
This keeps the charset parameter.
5) reginfo+xml [RFC-ietf-sipping-reg-event-00.txt]
http://www.ietf.org/internet-drafts/draft-ietf-sipping-reg-event-00.txt
This keeps the charset parameter.
6) watcherinfo+xml [RFC-ietf-simple-winfo-format-04.txt]
http://www.ietf.org/internet-drafts/draft-ietf-simple-winfo-format-04.txt
This keeps the charset parameter.
7) xhtml+xml [RFC3236]
http://www.rfc-editor.org/rfc/rfc3236.txt
This keeps the charset parameter.
2. Vendor tree
Only two media types have the charset parameter. Since media types in
the vendor tree do not always have accompanying documents, we do not
know if there are good reasons to omit the charset.
1) vnd.criticaltools.wbs+xml [Spiller]
http://www.iana.org/assignments/media-types/application/vnd.criticaltools.wbs+xml
This omits the charset parameter.
The details of this structure are proprietary to Critical Tools, Inc.
2) vnd.irepository.package+xml [Knowles]
http://www.iana.org/assignments/media-types/application/vnd.irepository.package+xml
This omits the charset parameter.
Use of this MIME type is limited to users of Lucidoc and associated
document management tools published by iRepository.net, Inc.
3) vnd.liberty-request+xml [McDowell]
http://www.iana.org/assignments/media-types/application/vnd.liberty-request+xml
This omits the charset parameter.
4) vnd.llamagraphics.life-balance.exchange+xml [White]
http://www.iana.org/assignments/media-types/application/vnd.llamagraphics.life-balance.exchange+xml
This keeps the charset parameter.
5) vnd.mozilla.xul+xml [McDaniel]
http://www.iana.org/assignments/media-types/application/vnd.mozilla.xul+xml
This keeps the charset parameter.
6) vnd.pwg-xhtml-print+xml [Wright]
http://www.iana.org/assignments/media-types/application/vnd.pwg-xhtml-print+xml
This omits the charset parameter.
7) vnd.wv.csp+xml [Ingimundarson]
http://www.iana.org/assignments/media-types/application/vnd.wv.csp+xml
This omits the charset parameter.
They wrote "No parameters are required - covered by client-server capability
negotiation."
8) vnd.wv.csp+wbxml [Salmi]
http://www.iana.org/assignments/media-types/application/vnd.wv.csp+wbxml
This omits the charset parameter.
They wrote "No parameters are required since WV capability negotiation
covers this."
9) vnd.wv.ssp+xml [Ingimundarson]
http://www.iana.org/assignments/media-types/application/vnd.wv.ssp+xml
This omits the charset parameter.
As for whether the actual entity should come with a 'charset'
parameter, we should also have a discussion of the various
issues in the draft.
By the way, "The Standard Hex Format" uses UTF-8 only.
http://www.ietf.org/internet-drafts/draft-strombergson-shf-00.txt
I have not found 'UTF-8' anywhere in this document, so I'm not
sure where you saw this restriction.
I am afraid that I made a mistake. We have discussed this issue in
the mailing list, but nothing is written down yet.
Another one is
http://www.ietf.org/internet-drafts/draft-sbml-media-type-02.txt.
This was approved by the IESG yesterday, so I guess it's too late
to try to change it. In practice, I very much hope that no implementation
will reject sbml data that comes with a redundant charset=utf-8.
Now we have two exceptions in the IETF tree.
Practically, what can we do?
Cheers,
--
MURATA Makoto <murata(_at_)hokkaido(_dot_)email(_dot_)ne(_dot_)jp>