@jensmaurer Is this PR still appropriate after the recent round of papers that touch encodings?

I think we still want this.

OK! (I'll need a rebase then.)

Also, @tahonermann, @cor3ntin, could you kindly take a look?

The term "valid" is often use to describe a properly encoded sequence, or a sequence of byte that can be cleanly decoded with a given encoding and applying that terms to the encoding itself is confusing.
I'd suggest a different term, such as "conforming encoding", "compliant encoding".

And... all encodings describe in the standard should be conforming, tautologically.
So... I question we do need a definition at all.

All literal and execution encodings shall [bullet list of requirements - which is correct as modified]

I would keep The ordinary and wide literal encodings are valid encodings, but are otherwise \impldef{ordinary and wide literal encodings} in lib-intro.tex, as we have been trying to not mention them in the core wording too much. ie, in lib-intro

The sets of additional elements (if any) are locale-specific. The encodings of the execution character sets are locale-specific and implementation-defined.

I agree with Corentin's review; I found the proposed wording with regard to valid encodings confusing. I do like the idea of naming the conditions an encoding must meet to be acceptable as a literal or execution encoding though. Perhaps "basis encoding"? The intent is to define a set of minimum requirements that can be extended; I think "basis" captures that intent. The wording might then look something like:

A basis encoding is a character encoding that satisfies the following conditions
...
The ordinary and wide literal encodings are \impldef{ordinary and wide literal encodings} basis encodings\iref{lex.charset}.
...
Those encodings are basis encodings\iref{lex.charset}.
...
The encodings of the execution character sets are locale-specific basis encodings\iref{lex.charset}.

Editorial meeting: the objections to the overly generic term "valid" are well taken, and we should look for a better term. The overall goal of factoring out the core notion in question is valuable, but we need to find the right way to express it.

Ideas welcome!

@tkoeppe,

Ideas welcome!

The prior comment included a suggestion.

Ah yes! Hm, "basis" is only one letter away from "basic", which may have too much historic ballast in this domain? But yes, variations on that approach occurred to us, too. E.g. "admissable encoding".

If we need to keep a terminology, "admissible", "conforming", "supported", or variations on that theme do seem like good solutions.