Attribute: epub:type="endnote" confusion

[typopaul]

Hello at all! I hope to be on the right place to share this question. If not, please guide me to where this questions belongs to. Thank you in advance.

From a professional EPUB production service company I got the information that Apple did not allow the usage of the attribute epub:type="endnote" in the past. But as far as I know this attribute is valid with the EPUB specs.

We asked Apple about the current state of this and they refer to the Kobo Specs and a similar issue and say to not use epub:type="endnote", since it is not declared as an error by the EPUB validation check.

But:

• The DAISY Accessible Publishing Knowledge Base does mention the attribute in examples: http://kb.daisy.org/publishing/docs/html/epub-type.html
• The Autorhing Guide of the IDPF from 2019 also lists it in the mapping of ARIA roles: https://idpf.github.io/epub-guides/epub-aria-authoring

So now I am confused: Apple says, it is not an error in the EPUB check, but it still should not be used? Is the EPUB check not up to date? Or which resource is wrong here?

[TzviyaSiegman]

I can't speak for Apple. You will have to raise any Apple-specific issues with them, but I can try to explain some details:

I am guessing that this comes from the update to DPUB-ARIA, in which role="doc-endnote" was deprecated. See https://www.w3.org/TR/dpub-aria-1.1/#doc-endnote for an explanation about how best to handle it. Since Ace by DAISY looks for parity between roles and epub:type, I suspect it's been incorporated as a warning in there.

[typopaul]

@TzviyaSiegman Thank you for your reply and hints.

In Ace 1.2.7 epub:type="endnote" does not throw an error, warning or even an information about the usage of this attribute.

In the DPUB ARIA 1.1 it’s said epub:type="endnote" is deprecated and listitem should be used instead. I tried that now and the result is the same as above: no errors, warning or infos.

I think there should be a warning added in the Ace check when epub:type="endnote" is used, since it is classified as deprecated in the DPUB WAI-ARIA Module 1.1.

Or am I thinking the wrong way?

[danielweck]

Hello @typopaul , there is a known bug in DAISY Ace related to the endnote semantic role and other deprecated features. We're working on it. Some background information: At the time of writing, DPUB ARIA 1.1 is a W3C Working Draft (not a final Recommendation). Tools like DAISY Ace and Deque Axe (which Ace utilises under the hood) are making pragmatic design decisions in order to help content creators navigate the evolving landscape of accessibility specifications (WCAG etc.). The DAISY Ace project tracks the EPUB Accessibility standard (version 1.1 is being finalised, see https://www.w3.org/TR/epub-a11y-11/#changes-older ). There are technical as well as UX (user experience) considerations, in the sense that verifying compliance with older established standards may be necessary in some cases, but it may also be sensible to target the latest "revisions" of relevant specifications (i.e. non-final / near-final normative requirements) assuming a high-enough level of confidence that these will not be amended significantly until their official launch. At this point in time, neither the DAISY Ace CLI (command line interface) nor the DAISY App GUI (graphical user interface) support validation "profiles" or configurable assertion levels. I personally think that this is acceptable, as I believe the DAISY Ace tools should remain as simple as possible (i.e. input -> output with minimal configuration). In that light, it makes sense targeting the latest in-flux specifications (assuming a low risk of major fluctuation), as - for all intents and purposes - they more usefully reflect the needs of the publishing industry at this moment in time. In this transition phase some accessibility rules are incorrectly reported in DAISY Ace (silence = no false negatives, no false positive), due to how Deque Axe is integrated. We are working on a fix as we speak. Feel free to report Ace-related issues in https://github.com/daisy/ace/issues

[mattgarrish]

I'll add to this discussion that epub:type is only for reading system use. Using the attribute doesn't have an impact on assistive technologies, so issues with ARIA roles don't cross-translate to epub:type. Since ARIA role use has tended to be much smaller than epub:type use, deprecating the semantic will just add warnings to existing content for an issue that really isn't an issue in most cases.

Are Apple/Kobo rejecting endnote semantics because they want you to use footnote to work with their internal pop-up footnote feature or because there's a real issue they need to avoid? It would help to know more about why it's not allowed before deprecating it everywhere.

For more background, the doc-endnote role is deprecated because of a technicality on how ARIA wants roles inherited -- one that they only discovered after we published the 1.0 module. You can use that role and it will work as expected with AT (it's recognized as a list item), but they didn't want to have the core ARIA specification depend on a module to satisfy role requirements (in this case, that the list role could also be satisfied by doc-endnote or doc-biblioentry, not just by listitem).

The roles are not deprecated because there are real-world issues we need to avoid, in other words. That's why it's not critical to get the 1.1 update out as fast as possible (but it is hopefully going to CR soon).