pdf-association
Ambiguity of term : "Actionably points"
In 14289-2; 8.8 the sentence:
Where content actionably points to other content, such as links occurring in the context of captions, tables of contents, cross references, hyperlinks, or other links within a document, such content shall use one or both of the following mechanisms: — point to the other content by means of a link annotation using a Structure Destination (see ISO 32000-2:2020, 12.3.2.3); — use the Ref entry (see ISO 32000-2:2020, Table 355) on the source structure element to reference the target structure element.
Does the presence of a link annotation and/or Ref entry imply the actionability? Or is this about semantics, and therefore, the check is supposed to be human?
IMO, "actionably" here means simply that the PDF includes features that allow end users to change their view in the document. In itself, this requirement does not, IMHO, imply a requirement that the target of the "actionable" thing is the correct (author-intended) target.
As a matter of best practice it is certainly important that links point to the author-intended target, but PDF/UA-2 doesn't explicitly require this (IIRC). As mis-directed links affect all users, not just users with disabilities, perhaps this is not an accessibility question but more like a quality question, which is outside our scope...?
I remember some of the later discussions we had in WG 9 about this: it was about the action-ness vs. passiveness of the cross-reference. And I clearly remember robust discussions over this sentence 😊!
Just having some span of text that passively mentions another section in the doc (for example) as a cross-reference is not what it is intended to cover. And whether the link target is correct or not (as far as a human is concerned) is orthogonal, as @DuffJohnson mentions. The author needs to explicitly make/define the cross-reference as active/actionable for the user via a PDF feature, such as a Link annot. It is in these actionable cases, the target needs to be into the logical structure (such as an SD) rather than other types of destinations. A Ref by itself does not make anything actionable/active (as it is only semantic), whereas the Link annotation very much does.
I am now trying to think if there are any other tricky methods that 32K defines that might make something explicitly actionable/active (e.g., a URL in a Rich Text string? via JS? ??) and thus why we didn't explicitly call out Link annots... as I cannot recall that part of the discussion.
