Rethink definition / term relationship and naming

[scottaohara]

The relationship for a term/definition is defined in the spec by creating an aria-labelledby or nesting relationship with the definition to the term.

If I recall correctly, the naming idea was so that a definition could be associated with its term. but looking at this now as i'm going through the roles / how they can be named, i'm not really sure why we didn't use aria-details instead?

Also, I thought it odd that at term could get name from author, but i realize that might be because of these previously being considered standins for dt and dd elements, but that was also rethought when working on aria 1.2.

[scottaohara]

looking at the simplified explainer to move aria annotations into the aria spec proper, it already states that

A term should be linked to its defintion using aria-details rather than aria-labelledby.

it seems that part of my issue is already answered, in that this part of the spec should change.

now the question remains if term/definition should allow for aria-label/labelledby at all. i lean towards 'no' / they should probably be author prohibited.

[scottaohara]

updated spec text should talk about how a details relationship between these elements is necessary only in situations where the association might not be clear. otherwise, it is not a requirement for authors to make the association.

otherwise, these roles should be changed to be name prohibited.

[spectranaut]

Discussed in today's ARIA WG meeting: https://www.w3.org/2023/11/30-aria-minutes.html#t07