Re: [I18ndir] [art] New Version Notification for draft-bray-unichars-06.txt

[Tim Bray <[e-mail]>]

The unichars draft is immensely less ambitious than modern-network-unicode,
which is covering a lot of complex subjects. Unichars covers only the issue
of code points, what the issues are with them, and subsetting them.  It
would be useful as a reference point in many currently active Internet
drafts, which tend to ignore this issue.

Turning modern-network-unicode into an IETF consensus document will be a
long complicated task, with a lot of difficult issues. For example:


   1. At the moment, I don’t believe in the distinction between “1d" and
   "2d”.  Because I don’t think that controls like \n and \r shouldn be given
   semantic significance in modern protocols, we have things like JSON and
   CBOR for packaging.  So this issue needs a lot more work.
   2. I think it should be a best practice that protocol designs SHOULD NOT
   go anywhere near normalization because so many text fields are used as
   unique identifiers and thus hash keys.


Neither of those positions are very strongly held, but it is clear that a
general best-Unicode-practices spec is a big task.

In the meantime, unichars explains one poorly-understood problem and
provides handy references for sane solutions to it, and has a chance of
being completed quickly.

I just don’t think a discussion of unichars vs modern-network-unicode makes
sense. They’re not trying to do the same things.

On Oct 1, 2023 at 10:56:34 AM, Carsten Bormann <[e-mail]> wrote:

> On 2023-10-01, at 19:44, Tim Bray <[e-mail]> wrote:
>
>
> I repeat: To inform protocol designers that it’s insufficient to say
> “Unicode characters” for their text fields. And, having explained why,
> offer reasonable alternative character repertoires.
>
>
> That is approximately the reason why I wrote modern-network-unicode.
> Please explain how draft-bray-unichars is more useful for this purpose.
> It doesn’t even distinguish 1d from 2d applications.
> That is not the level of advice I saw that people needed when I decided to
> write modern-network-unicode.
>
> Grüße, Carsten
>
>