Re: [I18ndir] draft-bray-unichars-01

[Carsten Bormann <[e-mail]>]

On 2023-08-29, at 21:10, Asmus Freytag <[e-mail]> wrote:
> 
> useful controls

Indeed.
Interesting value assignment.
“Helpful” would be so much better, but see below.

RFC 7464 (JSON Sequences) is one example for systematically using a C0 character, U+001E (RS), that is not allowed in JSON.
Our own RFCs themselves use U+000C (FF) for a specific purpose (pagination), until RFC 8650 killed pagination in text RFCs.

The ones singled out in the document to be useful mostly aren’t!
OK, some people use U+0009 (HT) for something they think is useful.
This, together with U+000D, should no longer be used except when required for backward compatibility.

Good practice:

A new protocol that needs 1D strings (no line structure) should not use C0 characters at all, unless specific characters are selected for specific purposes such as in RFC 7464.
A new protocol that needs 2D strings (sequences of lines) should use U+000A (LF) as a line terminator.
(U+001E RS and U+000C FF are examples for using C0 characters to create a 3D structure, but we usually do this in a different way now.)

But this is all about actual advice to protocol designers, not the other point I brought up.

(C0 usage no longer has anything to do with i18n, so I added back [e-mail])

Grüße, Carsten