Re: Last Call: UTF-16, an encoding of ISO 10646 to Informational

Date: Mon Aug 16 1999 - 10:49:31 EDT

On Sun, Aug 15, 1999 at 08:55:42AM -0700, Patrik Fältström wrote:
> --On sön 15 aug 1999 17.43 +0200 wrote:
> > I have a problem with UTF-16 being a MIME charset, as I do not believe
> > it qualifies, UTF-16 covers over at least 2 encodings:
> > 1. big-endian encoding
> > 2. little-endian encoding.
> >
> > To make UTF-16 a MIME charset you need to qualify which encoding
> > form (big/litte endian) that you are using.
> Just checking: you say that the registration of UTF-16LE and UTF-16BE as
> the two registered names to be used in MIME is wrong?

(It meant that I had not read the draft, olny listened into the
email discussion - blush). The UTF-16LE and UTF-16BE names take care
of the problem I mentioned.

However, after having a quick scan thru the draft, I find a
number of problems, such as misleadig statements (saying that
UTF-8 is variable width and indicating that UTF-16 is not variable width
is misleading), possible technical errors (saying that UTF-16
can be represented in 20 bit is possibly an error, as UTF-16 can
encode 17 planes - plane 0 to 16 - I am not sure of this one, but
this is my recollection), some wording that could be improved, and
some emphasis on a vendor industry standard, where I think we should
promote International Standards instead (we should rather call the
standard UCS than UNICODE). There may be other inconsitences
on terminology - if UNICODE is the referrence then that terminology
should be used, currently ISO terminology is used a number of
places, where UNICODE terminology is actually different.
Wearing my hat as liaison officer from ISO/IEC JTC1/SC2 to IETF
I would recommend that terminology aligned with ISO terminology be used
and that international standards be given preference over vendor
industry standards.

Kind regards
Keld Simonsen

> I.e. from my point of view, Paul tries to register three different names
> which can be used in MIME:
> UTF-16
> UTF-16LE
> UTF-16BE
> I need to, as area director, to know wether it is wrong or right to do this
> registration. Note that the author point out that UTF-8 is the encoding to
> use in Internet applications, and that these names are to be used only when
> absolutely necessary. As of today, there is no way what so ever to send
> UTF-16 (any variant) in electronic mail. So, I personally see this as a
> continuation of the draft which Keld wrote, which include all names which
> _can_ be recognized. I.e. registration of names like this doesn't say that
> the name should be used -- or that the encoding should be preferred before
> UTF-8 etc.
> If someone of you which oppose this paper want a different introduction
> which makes this point clearer, please submit text and I will be happy to
> discuss this with Paul.
> Also, I happen to know that Paul is on vacation which means that we can not
> hear all details of why the text is as it is until he is back.
> Patrik

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:51 EDT