From: Per Abrahamsen (abraham@dina.kvl.dk)
Date: Tue Oct 22 2002 - 05:53:18 CDT
Russ Allbery <rra@stanford.edu> writes:
> Charles Lindsey <chl@clw.cs.man.ac.uk> writes:
>
>> But the problem with that approach is that people will undoubtedly
>> implement, and make work, *something*, but it will not be UTF-8. It will
>> be "guess the charset", or "we use this code in this hierarchy".
>
> If that's the only approach that people are willing to implement, then
> obviously UTF-8 is the wrong choice. If UTF-8 is the right approach,
> people will be willing to implement it.
There are two parties here, the news administrators and the client
authors[1]. These two do not communicate, yet depend entirely on each
other.
- Clients tend to display 8bit characters using the local charset.
This is what happens by default, when you don't add code to do
anything else.
- Administrators will create groups using an encdoing that work for
the target audience, with the existing clients. This means using the
most common character set among the target audience.
- Clients who wish to display these groups correctly, then add ad-hoc
tables mapping hierachies or servers to character sets.
This is not a question of what people are "willing to implement", but
what happens when there is no communication between developers and
users of technology, as is the case between client authors
(developers) and news administrators (users).
The solution is inferior to both parties, news administrators lose the
part of the target audience who have the wrong default charset, and
client authors will have to constantly update the build in maps to
keep relevant.
The only way to break this spiral of inadequacy would be to define
some standard. It can only be done by a single dominant player (that
is, Microsoft), or a de-jure standard body (that is, USEFOR).
Unfortunately, Microsoft doesn't care, and USEFOR consists of Usenet
kooks, to happy to discuss to ever agree on anything. So it is not
going to happen.
Footnotes:
[1] The server is luckily mostly transparent.