This reworks the language around "$vocabulary" to capture the original intent in normative language. In particular, it explicitly addresses vocabularies omitted from "$vocabulary". As noted in #1294, this is based on extensive recent discussion and does not add any actual new requirements. It does use RFC 2119 language in places where it had not previously been used, though.
The new text emphasizes the concepts of required or optional vocabularies more than the true and false boolean values that implement them, and adds some explanation of the utility of optional vocabularies.
It also adds a clarification around identically-named vocabulary and non-vocabulary keywords, including that the standard keywords MUST be considered vocabulary keywords and subject to "$vocabulary" control even if the implementation does not support custom vocabularies at all.
Fixes #1294 . Also includes the two-word #1290 change as it just read strangely without it given the reworked section headers. Either way it will be a merge conflict anyway so I'll sort that out as needed.
This reworks the language around "$vocabulary" to capture the original intent in normative language. In particular, it explicitly addresses vocabularies omitted from "$vocabulary". As noted in #1294, this is based on extensive recent discussion and does not add any actual new requirements. It does use RFC 2119 language in places where it had not previously been used, though.
The new text emphasizes the concepts of required or optional vocabularies more than the true and false boolean values that implement them, and adds some explanation of the utility of optional vocabularies.
It also adds a clarification around identically-named vocabulary and non-vocabulary keywords, including that the standard keywords MUST be considered vocabulary keywords and subject to "$vocabulary" control even if the implementation does not support custom vocabularies at all.
Fixes #1294 . Also includes the two-word #1290 change as it just read strangely without it given the reworked section headers. Either way it will be a merge conflict anyway so I'll sort that out as needed.