Meeting minutes
Review of PR for restructuring FXL techniques.
Avneesh Singh: Wendy created a PR for the new structure of the FXL techniques; feedback is welcome.
<Avneesh Singh> w3c/
<Avneesh Singh> https://
Wendy Reid: Changes match the EPUB accessibility techniques doc as much as possible. No dramatic changes.
… any existing content was put into the right place.
… maybe it could be placed differently.
… I added some placeholders where it makes sense. I may have missed some.
… some sections may not make as much sense now with this restructuring.
… reading order may be one area that needs to be discussed.
Charles LaPierre: Was looking at this, halfway down, WCAG techniques for FXL, three headings; they are empty. Is that what you're looking for?
… ensure meaningful order, should be a heading?
Wendy Reid: Yes
Charles LaPierre: Might be some formatting issues.
Gregorio Pellegrino: I didn't have time to look, but thanks Wendy for rearranging. Leave the PR open until the next call and give some more time for proper review.
Charles LaPierre: Just wanted to mention when you first sent out the preview links there was an issue with GitHub's SSL certificate and it wasn't working, so review wasn't possible, but it seems to be working now.
Avneesh Singh: We can keep it open for another week.
Review of PR: Add aural rendering and TTS sections
Matt Garrish: This came from Rick Johnson, mentioning we'd lost mention of TTS from the core spec.
… we'd created a TTS note, since things like CSS Speech and PLS and SSML lexicons weren't supported and wouldn't make it through the rec process for 3.3.
… we have nothing stated in the core spec. Rick asked to create an aural rendering section, that TTS was important even if we didn't have specific requirements.
… this PR restructures the media overlays section to include this.
… I mention the original intentions, the TTS note, what technologies were out there; there's discussion about CSS Speech, there might be more work out there.
… at this point we don't have guidance aside from content structure and alternative text.
… mostly informative.
… my question is about the RS side: what should we say about how RSs should implement TTS?
… optional feature with recommended features based on the DAISY work: pause, play, restart.
… pointing to other implementations, but create a distinction from media overlays since it is different.
… reading systems don't need to become assistive technology.
… would like some feedback on this.
… want to make sure people are paying attention to this.
Avneesh Singh: I was also wondering about the same questions; are people ok with so much detail?
… do people in mainstream standards appreciate such precise guidance?
Matt Garrish: It's an optional feature; these are recommendations.
… maybe later on we can be more direct and require some, but not for now.
Avneesh Singh: The complete section is optional.
Gregorio Pellegrino: TTS is underspecified. I think we've discussed having guidance on TTS, a note or something like that.
… Hadrien from EDRLab was working on this.
… guidance on who has to implement TTS, and how.
… do I have to read the alt text, MathML, do I need to include the semantics?
… for developers who are unaware, it's new; the expectations for end users are unclear.
… given all this, and we don't have consensus, I'm wondering if this section becomes the starting point for this.
… two issues: this might be too much for the EPUB spec, and if we want this to be from the community, we need broader consensus.
Charles LaPierre: Do we make it clear what we mean by TTS?
… I think a lot of folks confuse it with assistive technology.
… having a screen reader vs. a reading system self-voicing.
… does that need to be made clearer?
Avneesh Singh: Matt added something.
Matt Garrish: There's a new introduction. I haven't gone into implementation on the RS side; there's no information on technologies right now for RS.
… there's limitations on that; there's no universal way to TTS HTML content. Will people agree to do it the same way when they have existing implementations?
… have tried to limit it to basic features and what they would be, not the implementation.
Avneesh Singh: Too low level.
… maybe Matt can have a look to make it clearer, higher-level.
… we'll get the feedback from the community.
Wendy Reid: I will add my comments. One challenge is the confusion thing that Charles mentioned: TTS vs. screen readers vs. audio/media overlays, etc.
… this is not just an EPUB problem; browser content platforms could have a TTS implementation.
… news websites: listen to audio version, etc., but we don't know who should specify.
Matt Garrish: That's the big challenge. I understand the desire for information; we just aren't at a place to get recommendations on authoring or implementation because there's no information or standards out there.
… what should TTS be able to do? We can cover basic functionality. I don't know how much value it has, but if we want to spell some stuff out, maybe this is the way to do it.
… I'm open to alternatives.
Avneesh Singh: Traditionally, did we provide this kind of guidance for Media Overlays?
… stop and start from the same position.
… did we give that specificity?
Matt Garrish: I can't remember off the top of my head.
… I can take a look and try and match it up.
… some of the same stuff for MO.
Avneesh Singh: That would be helpful.
… keeping it high level would be nice. If we want to have a proper guidance document, reading systems guidelines like from DAISY, that's huge detail; we need to think about a different document for engineering level.
Matt Garrish: I'll check it and try to make it match up, make sure we aren't adding in features.
Avneesh Singh: Good discussion, didn't expect the details; we spotted some important details.
Matt Garrish: I was hoping for this feedback.
Review of extended descriptions explainer:
<Gautier Chomel_> w3c/
Gautier Chomel_: On extended descriptions, I think I addressed all of the comments in my PR. As it is, I think we can start a new review.
… [shares link to document]
<Gautier Chomel_> https://
Avneesh Singh: It was a readme; I was able to read it.
Gautier Chomel_: No questions on my side. Image links are broken and I will fix that. I'm waiting for feedback.
Avneesh Singh: Please share your comments.
Charles LaPierre: The link you've shared, it's going to a GitHub page; the explainer is in the readme?
Gautier Chomel_: The second link is the content as an HTML page.
Charles LaPierre: It's the same as the first link?
… and the image links work.
Gautier Chomel_: They work on the readme and not the page, probably a path issue.
Charles LaPierre: Just looking for feedback on this? Is there an issue to collect it?
Gautier Chomel_: You can add issues in an issue I'll create in the tracker.
Avneesh Singh: If we're ok with the current state, we can share it with APA, at least the chairs.
… Matthew mentioned he'd like to help; or should we wait?
Jonas Lillqvist: I'm a bit confused. Has anything changed since we discussed it last week?
… Gautier, has it been updated?
… I read the version in the agenda link, and I didn't see any changes.
Gautier Chomel_: The changes were in a PR; I just merged it, and it's changed now. You can see a diff in the closed PR.
Jonas Lillqvist: Ok, it's hard to review that during the meeting.
… which link?
<Gautier Chomel_> https://
Gautier Chomel_: I think we need one more week to review before pushing to APA.
Avneesh Singh: We'll wait for a week and review. In early Feb I'll ask APA chairs to review and give feedback.
Gregorio Pellegrino: Editorial question: do we have time to move this to the HTML format?
… we started from a model suggested by APA.
Avneesh Singh: HTML or Markdown?
Gregorio Pellegrino: I thought it was HTML with ReSpec.
Wendy Reid: You can use Markdown with ReSpec.
Avneesh Singh: How was the original structured?
Gautier Chomel_: I don't remember. I think we kept with Markdown to keep it simple.
… to get the feedback, pushed as readme, the first version is HTML, so we moved from HTML to Markdown. I can go back and check.
Avneesh Singh: I think we moved to Markdown but can't remember why.
Gautier Chomel_: Mainly to ease the editorial process.
… since it's just an explainer.
… people can figure out what we want to do; people can take action.
… we have explainers in the epub-spec repo.
Wendy Reid: [explains ReSpec in Markdown]
Avneesh Singh: Maybe we move it to HTML then.
Gautier Chomel_: The W3C explainer template is in Markdown.
Matt Garrish: I thought it was just going to live as a page in the repo; does it need to be ReSpec?
… if it's just an explainer doc, does it need ReSpec?
… no requirement for it.
… lots of other explainers and their samples are in Markdown in repos.
Avneesh Singh: The ReSpec page looks more formal as well, gives an impression of a specification, but this is an explainer. The image can matter.
Matt Garrish: There was a template we were given; we should just use that.
Wendy Reid: ReSpec thing might be my fault; I wasn't sure if this was going to be published or not.
Avneesh Singh: There's a lot of discussion; where should this proposal go?
Jonas Lillqvist: I had a look at the PR. It seems like it still doesn't mention the possibility of aria-details having a direct reference to the image description instead of referencing it through a link,
… which would be important, I think, to mention.
… there are techniques both for EPUB and the web where you would not put the extended description on a separate page and link to it.
… also, I feel that the document is not explicitly about EPUB publications only, but the technique mentioned is mostly for EPUB. On the web you'd probably use details, or some other element on the same page as the image.
… it doesn't make sense to create a separate page for the descriptions.
… something that we touched upon in the TIES meeting earlier this week was that one possible use case in reading systems for this semantic markup would be to have functionality for hiding or skipping extended descriptions in the main text flow,
… because the option of having extended descriptions in the main text flow isn't mentioned, that is a gap.
Charles LaPierre: We referenced skippability and escapability; visually you may also want to do the same thing, linked or within the document, to be able to hide those and not interact. It would impact TTS or MO interactions.
… that use case is important to show because it's hard without these semantics.
Avneesh Singh: The document is created to emphasize the need for these semantics.
… as to the other comment about aria-details pointing to the same page:
… how does it highlight the need?
Charles LaPierre: I think it does.
… it's the same thing; it's just another option that publishers want to use, to have the extended description by the image, in something like an aside. But you need to be able to do the same processing on the same page or a different document.
… we don't want to prevent the publisher or force them into one way of doing things.
… we want the same functionality and features for all use cases.
… we don't need to limit it if it's on the page or not.
Matt Garrish: This is sort of the challenge when we go to ARIA. They may only take the extended description on the main page as an ARIA property; they may not like the linking.
… depends on what they see as a viable approach.
Charles LaPierre: Let's not give them an excuse to dismiss the idea.
Matt Garrish: Ultimately we're presenting the EPUB need, not the web, but we'll see.
Avneesh Singh: All of us are alluding towards allowing aria-details pointing within the document as well as outside.
Matt Garrish: Gives them something they might understand, friendlier to their way of thinking.
Gautier Chomel_: I'm not sure I caught everything. I understand that we should not speak about external files?
Avneesh Singh: I don't think we're removing that approach; we are including the same-file approach.
Charles LaPierre: Adding to, not removing.
Gautier Chomel_: The sample pattern has an extended description file; need to update.
… I'll open an issue with what I understand and people can contribute to that.
… easier for me to understand that.
Avneesh Singh: We'll add to the issue.
… anything else?
DPUB ARIA roles proposed/discussed by us in past
Avneesh Singh: Discussion of DPUB-ARIA roles of the past: we've not taken actions for that.
… are there other DPUB-ARIA roles that we need to do something about?
<Avneesh Singh> w3c/
<Avneesh Singh> https://
Gregorio Pellegrino: One was about questions and answers.
… since we worked with educational publishers, we had some troubles identifying interactive exercises in publications.
… it was raised in the past; there was some pushback.
… no way to identify an item that is an interactive exercise, might be problematic.
Matt Garrish: That's going back 10 years. It was something we brought up when we did the initial DPUB-ARIA work; it was shot down.
… didn't see the need for it.
… not specific enough.
… if we're going to make a foray into education, it's more than EPUB.
… we shouldn't create stuff in the DPUB space that isn't exclusive to EPUB.
… there is lots of stuff on the web in the education space outside of EPUB; I don't know how we get there without looking at an educational module.
… or support from educational publishers more generally.
… or backing from industry.
… it looks like we're just trying to add things. I see a lot of problems; it's not something you'd associate with book structure.
… we can try again, but I have reservations.
Wendy Reid: I have the same reservations. What is the point of semantics like this?
… what do we want that ARIA semantics doesn't already have? Is this a metadata play?
… if you are to have doc-qna, what does that mean?
Gregorio Pellegrino: I agree it is not something related to digital publications, but maybe we can raise it to the broader ARIA group and get publisher support.
… what is the goal? In digital publications, exercises are noted graphically, easy to identify visually, not as easy to find with AT.
Gregorio Pellegrino: The use case is to make the content discoverable, but maybe forms or another element could serve this purpose.
Avneesh Singh: We need to consider how we can persuade ARIA; we need to have strong explanations.
… we need to do some homework before we go to ARIA.
Charles LaPierre: I was going to agree with Gregorio Pellegrino.
Avneesh Singh: I'm worried about these going simultaneously. We have extended descriptions, and interactivity roles, but we need to weigh it.
Another use case would be to print out a list of questions, possibly :)
Avneesh Singh: Thanks everyone; please give your feedback on the open PRs.
<muntxi> Thank you for the interesting discussions!