From: J.B. Moreno (planb@newsreaders.com)
Date: Thu Mar 06 2003 - 13:07:47 CST
On 3/6/03 1:19 PM, Russ Allbery at <rra@stanford.edu> wrote:
> J B Moreno <planb@newsreaders.com> writes:
>
>> The majority of messages sent with non-ASCII content in headers, are
>> sent with "raw" byte values, not 2047 encoded. That's a fact, that's
>> part of "news", although it may not be part of 1036. We are /supposed/
>> to be telling people that /want/ to write news software what to expect
>> -- if we tell them that what they should expect is ASCII, then we are
>> LYING to them.
>
> By that argument, if we tell them to expect UTF-8, we're also lying to
> them. That argument says that the standard should tell people to expect
> random unlabelled 8-bit character sets.
Which is exactly what we *do* say:
Encoding by other means is not compliant with this standard.
Nevertheless, encoding using other character sets (with no indication
of which one beyond the user's ability to guess based upon other
clues in the article, or custom within the newsgroup) has been in use
in some hierarchies, and such usage may be expected to continue for
some period after the introduction of this standard.
We are telling the users and programmers to switch, and warning programmers
that such will not happen immediately, that they will, for a time, have to
continue to cope with random charsets.
As I've said before we can tell the programmers and users to do one of three
things: tell them to do exactly what they are doing now (100% chance that
they'll take this and run with it), tell them to use 7 bits only (evidence
indicates this will be rejected), tell them 8 bits is OK, but use UTF8
because it offers X benefits.
We aren't lying to them, we are telling them to make a change. Anything but
telling them to do whatever they like is going to meet resistance, how much
resistance depends upon what they are told to do.
-- J.B. Moreno