Ned Freed writes:
The IETF is free to pursue topics and to develop specifications as it
sees fit. It most definitely is not tied to the work of any other group.
However, the general policy of IETF operations is to pay attention to work
done by other groups, particularly by other standards groups, and to use
as much of it as is reasonable. In other words, we can change or reject
other work, but only if we have a very, very good reason. We definitely
do not simply ignore such other work. We need to have an alternative
that we deem superior.
This is a reasonable position to take and I support it. However, it is not
directly relevant to character set issues because it isn't possible to simply
enbrace the work done by other groups. The problem in a nutshell is that
there's too much work done by too many groups with entirely different agendas,
resulting in conflicting, competing, and overlapping standards.
Well, I do not think that the effort is out of reach.
In RFC1345 a vast number of charsets are specified and registred
for MIME use, and they can be handled uniformly according to
a model which is not that voluminous, RFC 1345 prose is about 10 pages.
The Unicode people has done a similar effort.
The IETF must therefore pick and choose between these efforts; the alternative
of embracing everything violates numerous principles and requirements:
simplicity, elegance, lack of ambiguity.
There is an IETF specification in RFC1345, which demonstrates that
embracing all of the character set world is doable, it can be done
with about 10 pages of prose (which is not outrageous compared to other
IETF specifications) and the stuff is specified without ambiguity.
Elegance is a matter of taste, so I won't comment on that.
What Ned is looking for has already been accomplished within IETF.
I hope that this clarifies the question of IETF use of outside work by
other standards bodies.
I don't think it does. It is nice when there's nice clean unambiguous work
done by a small number of other groups, but character sets don't fit this
model.
RFC 1345 only draws on a few sources for its information, about 10
sources is used, and only 2 sources are very significant, namely ISO
with its ISO 2375 register and ISO 10646, and IBM. I estimate these
2 sources count for 90 % of the information in RFC1345. The remaining
charsets are very similar to species from the main two sources.
Keld