W3C

Publishing Maintenance Working Group Telco

30 October 2025

Attendees

Present
Avneesh Singh, Brady Duga, Fredrik Fischer, George Kerscher, Gregorio Pellegrino, Hadrien Gardeur, Ivan Herman, Jen Strickland, Daniel Kimberg, Masakazu Kitahara, Matthew Atkinson, Matt Garrish, Shinya Takami, Susan Neuhaus, Toshiaki Koike, Wendy Reid
Regrets
-
Chair
Susan Neuhaus, Wendy Reid
Scribe
Brady Duga

Meeting minutes

APA and PMWG Joint Meeting

Wendy Reid: Goal is a joint meeting with APA. Typically done at TPAC, but that had scheduling issues

Gregorio Pellegrino: I have an agenda proposal, I would like extended desc to happen now for schedule reasons

Matthew Atkinson: We have discussed in TAG and APA, I have some ideas

Extended Descriptions

<Avneesh Singh> https://github.com/w3c/epub-specs/wiki/Standardizing-Extended-Descriptions:-User-Stories,-Testing-Results,-and-Current-Limitations/

Gregorio Pellegrino: Brief intro - for descriptions that go beyond alt text (formula, etc)
… anyone can see them, not just for screen readers
… we are missing ways to control them
… there are a few ways to do it, but they all have issues in epub
… for instance summary details breaks with pagination
… aria details is more interesting, and works ok when it is all in the same file
… But we need something that puts the description in one file and the item it describes elsewhere
… but have a link between them
… We have footnote and endnote metadata for links
… 1. How to identify a link goes to an extended desc
… 2. how can programmatic relationships be maintained cross file
… 3. How can we know something is an extended desc when we open a file

Matthew Atkinson: We have some possible solutions, I will give an overview
… these come from APA, with some ideas from Jeffrey Yasskin (TAG)
… some sounds a little like the issue with long desc, and why it is now gone (misuse, no support)
… We need to be aware of those risks
… ARIA details is still very low support
… Details and summary come to mind at first, but may not be right
… Putting the metadata on the link breaks down on an inter doc link, but maybe `rel` is better for cross document
… that might preserve the data

<Matthew Atkinson> Patching in content - discussed as an enhancement of a Declarative Partial Updates CG: <https://github.com/WICG/declarative-partial-updates/blob/main/patching-explainer.md#potential-enhancement---patch-contents-from-url>

Matthew Atkinson: There is also a proposal to patch in content
… which might be interesting to lazy patch in content from other sources
… APA had the same concerns as reviewed here (misuse, support)
… We had a different way of viewing the problem. One problem with navigating away is coming back
… easy to do in JS, but is this a problem with the platform that you need to be able to come to where you were as a general problem for the web
… One last thing is to get ARIA and TAG feedback
… And keep APA up to date

Avneesh Singh: Thank you, good overview
… main problem is detecting the extended desc via machine
… We really need the semantics, not a specific mechanism for showing it

Charles LaPierre: What is the APA view on extending DPUB ARIA roles for the extended description?

Wendy Reid: We still have the RS UA problem. Dynamic insertion of content doesn't really work in a RS due to pagination issues
… we have a similar issue with footnotes, we solve that with links and popups
… Goal for extended desc is to have a similar implementation
… i.e. popups

Wendy Reid: today we have trouble with footnotes as there are multiple implementations

Matthew Atkinson: Charles LaPierre, you suggested extending DPUB. Seems logical, and there is prior art
… but should discuss with ARIA if this can be a solution for a wider problem
… If you are sometimes talking same doc and sometime different, we should avoid two mechanisms
… So `rel` still sounds interesting
… regarding pagination, why not show as popup? But how do I know this is an extended desc to show in the popup
… The document you have covers a lot of use cases, but we might want some that are from outside epub to make sure we have a general solution

Matt Garrish: That is the concern. And long desc has a lot of baggage, so we would like to avoid that
… and we do want something that is more broadly applicable
… This will probably take more discussion/input and why we are socializing it now
… Is there a broader audience for this?

Avneesh Singh: Extended desc can have broader appeal
… In epub, for footnote we have aria role on link, and container has its own role
… The container role might have an aria role, not dpub:aria, which makes it useful across the web
… The other thing to highlight - whenever we discuss this, we are warned away from the long desc debate
… We don't want long desc, we just want semantics
… how can we avoid that comparison?

<Susan Neuhaus> ack: matalk

Matthew Atkinson: I'll have a rough go about differentiating from long desc
… the point you make about identifying content that is currently visible is important
… it seems like the big issue with long desc was that it wasn't rendered
… So making that clear should help. And socializing this is very important to get all feedback.
… Might be a harder sell in the epub sense, since it is in a different file
… So there really should be an affordance for all users that this is available
… So my question is, what should be next steps?
… Maybe another call with APA?

Janina Sajka: In terms of selling programmatic access (the idea of that), it might be useful. If we don't make it discoverable then perhaps it was underspecified. That might be a way to sell it

George Kerscher: So the link is always visible. On the web details is great, but has trouble in EPUB
… in either case, it is always visible to all users, but we need to be clear that the semantics are important and need to be available
… the understanding that extended desc is there on the link and target is important

Avneesh Singh: Next steps - we need to do one more pass on the document. Need to clarify we are looking for semantics
… we will also leverage html to make things programmatically available as part of sales pitch

<Charles LaPierre> +1 for a meeting with APA after TPAC.

Avneesh Singh: Then maybe some of us can join an APA call after TPAC
… Is a wiki page ok, or do we need to be more formal

Jen Strickland: I see how the semantics is a lovely pairing, but it doesn't work well with epub
… Does epub ever have a popup? It seems like the semantics needs to be clarified for that

<Matthew Atkinson> Next steps: turn the wiki page into an Explainer (<https://www.w3.org/TR/explainer-explainer/>) or write one as a markdown file in revision control.

<Jen Strickland> The HTML popover API is the thing I was thinking of.

Matthew Atkinson: The proposed next step (an explainer) is a good way to list all the rejected approaches (better than a spec)
… so using an explainer format will help with review since they are familiar with the format
… and I would be willing to help review
… then TAG could give some good feedback

Updates from APA

Matthew Atkinson: First pronunciation

<Matthew Atkinson> https://www.w3.org/WAI/about/groups/task-forces/spoken-presentation/

Matthew Atkinson: Need to make sure TTS will properly pronounce things
… lots of use cases are around education
… we have explored some tech approaches
… typically this has leaned on SSML, etc and we haven't had much success
… there has also been a lot of discussion around CSS speech, some of that might be easier to implement
… but we are really diving into real use cases first
… Some tech answers, there may be some low hanging fruit in AT [I think I missed a detail here]
… At some point we will share the use cases to get feedback

Hadrien Gardeur: Pronunciation has a lot of solutions, but none are implemented
… another issue is SSML is just really poorly implemented
… so even if we had the pronunciation, we have no way to use it
… so we have a risk of coming to an unimplemented solution

<Hadrien Gardeur> https://kb.daisy.org/publishing/docs/text-to-speech/index.html

<Susan Neuhaus> Brady Duga: on a previous project we worked on tts for audio book content. We gave up on

<Susan Neuhaus> …anyway for the the author to provide pronunciation except having the author provide us with a sound file

<Susan Neuhaus> …of them pronouncing it

Matt Garrish: this is a good time to bring it up, big problem in need of solution

Susan Neuhaus: I do a lot of mixed language books, specifying language causes a voice change (jarring)
… so I hope bilingual is on the list

Shinya Takami: Japanese text has ruby for pronunciation, but is also used for other things
… we don't have a spec for which should be pronounced
… sometimes both, sometimes base, sometimes ruby

symbols

Matthew Atkinson: Next, symbols. You may have seen them as AAC symbols
… often used to help understand text
… There is also Bliss symbolic languages with is entirely symbolic

<Matthew Atkinson> https://www.w3.org/WAI/about/groups/task-forces/adapt/

Matthew Atkinson: there is a dictionary, which is constantly changing (it's a language)
… You can't just add to web pages as there a number of symbols
… we are working on a way to specify the symbols, then drop in the user preferred symbol set
… We are currently looking at ruby to implement this, though that is a little controversial
… main question is, is ruby supported in EPUB?

<Susan Neuhaus> Brady Duga: There is ruby support in reading systems that support asian languages

George Kerscher: Is there a symbol for extended desc?
… seems perfectly logical. Publishers want a visual element for an extended desc

Shinya Takami: RS in Japan all support ruby, as do major players launched in Japan

Janina Sajka: I am willing to bet that extended desc will have such a symbol, assuming we get AAC fully supported

Minutes manually created (not a transcript), formatted by scribe.perl version 248 (Mon Oct 27 20:04:16 2025 UTC).