Meeting minutes
<Daniel Weck> +present
ISO Update
Ivan Herman: There isn't much to say
… when we went to CR and gave the documents to Makoto, we did what was needed
… so far we have received no comments. Maybe no news is good news?
… we will have to decide what to do editorial once everything is ready
… There will be an ISO meeting coming up, Makoto will be there and will focus on EPUB and get feedback
… Someone also made a script to turn respec into an ISO formatted document
… unclear if that is good enough, but maybe
… also Phil Archer did it by hand for another group, took about a week
… We still have an offer from Yongsan (from Korea) to do it for us
… that is all we have on ISO
Wendy Reid: Questions?
… Just to disrupt - there was a question on moving things around
… taking full bleed earlier, which might influence chartering
Ivan Herman: Yes, but this is what I want to avoid. We don't want to have technical discussions during the charter discussion
… All we want for the charter is the headlines
Hadrien Gardeur: I think we can stay abstract, but ok
Ivan Herman: You are very optimistic
Susan Neuhaus: Maybe it would be better to move it forward then?
Ivan Herman: I have no opinion and leave it to the chairs
New Charter for EPUB 3.5
<Wendy Reid> presentation slides
Ivan Herman: Where we are - there is a charter proposal we have had for a while
… Most of the stuff up top is admin
<Wendy Reid> charter proposal
Ivan Herman: I will skip this first bit in yellow
… Annotations is basically the same as before. ISO is the same, minor updates about 3.4 going to ISO
… we need to update some timings, but that is all admin
… Then latest version is also admin, 3.5, a11y is still 1.0, audio book maintenance, etc
… timeline is pure fiction. We should consider if we want to make it more realistic
… We can do that later or via github later
… in my experience the timeline is a joke, no one actually checks it later
… we need to review the other groups with whom we have a liason
… Marisa DeMeglio, I will need a small PR with your group info
… I don't think we need some of these, but there is no harm
… external is DAISY, ISO, Readium, etc, same as before
Hadrien Gardeur: Was audio WG on the list? They may be working on TTS
Ivan Herman: Again?!
Hadrien Gardeur: They are considering bringing back web speech API, including TTS. So their next few years of work is quite relevant
Ivan Herman: Can you add a PR for that?
Hadrien Gardeur: Yes
Ivan Herman: Ok we will wait on Marisa DeMeglio and Hadrien Gardeur to fill in those PRs
… the charter is quite complete. We sent in a first draft for review and received no comments
… We still need privacy and TAG
… we may need to push on privacy
… TAG seemed ok to proceed without showing up in person
… Next it goes to a smaller group, then AC for final review
… We will now review each item in the scope. This is a draft so we can make any changes we want
… There have been some issues that came up recently that we wanted to defer to next version
… What we should do is regroup the items by 1 or 2 general titles
… we should agree on that today so we can get to review by Dec
… I will go through the scope items, and we need to approve the text. We should NOT discuss the technical issues
… Should we go one by one, or should we do a full overview, or go 1 by 1?
Laurent Le Meur: Let's get the overview
Ivan Herman: Ok. So first is media overlay. There were some issues about updating the overlays feature, and another issue that said we should update it
… I think our previous discussion said we shouldn't adopt WebVTT, but if we don't say anything about it then we won't be able to adopt it
Marisa DeMeglio: I don't want to close the door to it, but my impression from the room was that we wouldn't get it in 3.5
Ivan Herman: Ok, but if we say nothing it will be hard to do
… next is full bleed
Hadrien Gardeur: Light novels is mentioned
Brady Duga: Is this more than full-bleed? Light novels have interesting pagination questions, like full spreads
Hadrien Gardeur: do we need to talk about different layouts?
Hadrien Gardeur: The page-fitting part should cover that
Ivan Herman: a11y for FL and roll
… Question - is this normative?
… if it is non-normative, then we can add it whenever, but if there is a possibility that it is normative we should keep it
Wendy Reid: When this was suggested it was planned to be non-normative, but given the discussion yesterday, there are some possible normative changes
Ivan Herman: So we should leave it?
Wendy Reid: Yes
Gregorio Pellegrino: I agree. in the last 3 years we did a lot if research into FL a11y
… this may be become part of a11y 1.3
Gautier Chomel_: Part of what happened so reading systems are displaying reflowable on the fly
Ivan Herman: This one is important for the other documents, since we can only change them for one of these issues (privacy, security, i18n)
… Ok, this one is interesting - EPUBCFI to rec track? Should it be on the list?
… we will return to it
Ivan Herman: Metadata - we may need to do something with it.
… some reading systems ignore it, some use it all
… We should have something on the scope
… I tried to be diplomatic
Hadrien Gardeur: This is tricky because people care a lot, or not at all
… We have to make space for different groups. We haven't worked on metadata in years
… It is worth looking at what people are adding. There are a lot of extensions out there, document it, understand why they are extending, etc
… we need some community outreach, so that is a further question - how will we do that?
Ivan Herman: There are a number of terminology issues that have come up
… Romain has raised a number of issues, lots of discussions with Matt Garrish
… CG is specially mentioned as having a closer relationship than other groups
… that is all, and here is the proposed text that will go in the charter
Hadrien Gardeur: We just started to work on the comics/manga TF. Panel by panel was raised twice already
… I mentioned there are things that could be used already
… region based nav is one
… Some of these things could require normative work. I don't know if we want it on the charter, but it has been raised by some big pubs already
Ivan Herman: So review will not like a lot of items, so if we can reword an existing one to include it that would be good
… so item 3?
Hadrien Gardeur: I think 1 and 3 could cover it
… not sure if it is explicitly covered
Ivan Herman: We just need to be able to point at an item to say "this is what covers this"
… strictly speaking, what you say is not a11y it is nav. 3 is a11y, so not the same
Hadrien Gardeur: Amazon has something like this, but it isn't great
… so the industry wants something standard
Ivan Herman: Maybe I could rephrase 3?
Susan Neuhaus: What about 8? Since it came from the CG
Ivan Herman: Yeah, but it would be good to have it explicit since it already came up
… if the CG comes up with some incubated issues, they need to create a final report and submit it to WG for implementation
… I don't know if you want to go through the trouble for this item
Wendy Reid: Pause - if you are here for the CSS intl meeting, you are in the wrong place
Various on zoom: Oops! Bye
Hadrien Gardeur: Our focus right now is best practices, I would prefer not to write a final report.
… And another one to point out, with 3.4 we have roll, but there is a remaining question over multi-chapter/story
… Typically you roll for each story, then start over for the next story
… We don't have a way to specify such breaks today
… it's not nav, more like rendering behavior
Ivan Herman: In spite of what I said, maybe add 1 more item - "further evolution of roll"
Wendy Reid: I don't think it would be unexpected
Ivan Herman: [adding as item 4]
Shinya Takami: Regarding #3 - will this include a11y for manga?
Ivan Herman: It is FL and roll, so it fits here already
… manga isn't part of the standard, so we don't mention it explicitly
Marisa DeMeglio: Like I mentioned yesterday, in Sync Media CG people are mentioning the cases that won't be solved by synced media
… like braille renditions. Would this fit in the publishing world? Maybe the CG?
… #9 is very broad, maybe it is covered there?
Ivan Herman: Yes, this is broad very intentionally
Marisa DeMeglio: I don't want to make this list longer
… I would like to redirect people to the publishing CG
Ivan Herman: We could add the sync media CG to #9
<Marisa DeMeglio> Accessible Synchronized Multimedia for Publishing CG
Ivan Herman: I would prefer to list explicitly in the charter
George Kerscher: I am hearing in the sync media a lot about sign language video. Is that mentioned here?
Marisa DeMeglio: #1? [reads the item]
… maybe we should add "possibly by extending SMIL or WebVTT"
George Kerscher: Did I hear properly that FL and the posibility of turning it into flowing is also in scope?
Wendy Reid: We have item #3 [reads it aloud] which covers it
George Kerscher: When people are reviewing it do they understand full bleed stuff?
Ivan Herman: There are two things - some people have no idea what EPUB even is, then there are some who do understand. We write for the second group
Charles LaPierre: I am worried about #9. It seems like a get out of jail free card. Was it in the previous charter?
Ivan Herman: No. We had the a11y task force, but not the CG. But it wasn't abused by that group
… we will see. Maybe the AC will have us remove it?
Charles LaPierre: What about the a11y and comics CGs?
Ivan Herman: Those are TFs under the CG
Susan Neuhaus: I agree with Charles LaPierre, but one thing that people complain about is our slowness, so maybe it is a good idea to keep it so we can get to things faster
Ivan Herman: We will leave it for now
Ivan Herman: I don't remember why this came up?
… wasn't it something other than annotations?
Wendy Reid: I raised in a chairs call that another reading system was implementing it
Ivan Herman: I am neutral, but we need to decide now if we will take it up
Laurent Le Meur: EPUBCFI is in the same place as multiple renditions
… no one wants to work on it now
Ivan Herman: a few years ago we looked at it, and it needs a lot of work (rewrite or extend)
… it is a lot of work
Hadrien Gardeur: We used to be able to use it internal, but that is gone now.
… it has been used, plenty of reading systems use it, but it isn't very standard
… so it is non-interoperable
… So it is tricky. I don't see much reason or momentum or it
Ivan Herman: Note that the way I put it in it is cautious
Brady Duga: It is a lot of work, if we do it we need a champion
Ivan Herman: we need someone to say it's necessary for their business
… we need this to be standardized
… I just want them to use it, if I'm going to spend time writing it
… if there's an actual need for it, we should consider it, but we need a champion
… do we want to leave room in the chatter for it
… but I want a champion first
George Kerscher: Is calibre a w3c member?
Wendy Reid: No. We could ask them to join as IEs
Various: Calibre or Colibrio
Marisa DeMeglio: George Kerscher asked about Colibrio
George Kerscher: Lars always asks about this
… can we invite him as an invited experted?
Ivan Herman: He should not come to you, he should come to the WG
… he has not approached us
… he did send a mail to the WG to say he implemented it
George Kerscher: What is the proper way for him to proceed? Come to you?
Ivan Herman: He should apply for IE and add a note that says he wants to standardize it
… should we leave it as is? And then remove it when we go to AC?
Wendy Reid: I would like to reach out to some people and see if we can find a champion
Brady Duga: maybe we should do some outreach and see if people want it?
Brady Duga: Can we mark it at risk?
Ivan Herman: No such notion in charter
Brady Duga: I'm concerned if people see it now, are happy, then we pull it, it's a bit rude
Ivan Herman: Now we are discussing the charter and do changes, that's ok, we need to be final before we send it to the AC for final approval
Ivan Herman: Now we are discussing the charter, we need a decision within 2 or 3 weeks
Marisa DeMeglio: Is there something else than CFI?
Laurent Le Meur: we have some that work
Marisa DeMeglio: Is there any loss?
Laurent Le Meur: CFI has a left and right part. For annotations we don't need the left
Marisa DeMeglio: So which is the left part? The path in the CFI?
Ivan Herman: OPF is unknown to the annotations
Marisa DeMeglio: Having a way to point in is murky
… so annotations will exist without CFI
Hadrien Gardeur: So CFI is a bit of an internal thing. It is fine used that way
… There are other means of referencing in. Some are not epub specific. So they can move between epub and the web
… The thing about epubcfi is it is all in one. The other solutions are more specific
… the people that love CFI love it because it does everything
… I don't see a signal that CFI is anything other than internals
Brady Duga: I know you said you don't care Marisa DeMeglio, but I will comment, EPUB CFI is really neat because I can say "this is the 4th thing", whereas annotations you don't do that, we had to say what couldn't be referenced
Hadrien Gardeur: it's also fragile, if I have two OPFs or files, and CFI breaks in the interchange, but breaks when I have two different EPUBs in the reading systems, at the end of the day, text fragments works
… nothing else for reflow that works across publications, CFI supports it, but we don't need it
Marisa DeMeglio: I mean that we don't need it, but informative
George Kerscher: Do I contact Lars?
Ivan Herman: We should talk to him. We need commitment that someone is willing to do the work in the next 2 or 3 weeks
… Could be Lars, could be someone else
Susan Neuhaus: Hadrien Gardeur has a good point, is this even under our purview?
Hadrien Gardeur: So even if we have a champion, we still shouldn't do it
Wendy Reid: Just because it can fail doesn't mean it will fail
Marisa DeMeglio: and if we are sharing internally then it could work
Hadrien Gardeur: It might not even work in a different system
Laurent Le Meur: and don't forget test suite
Brady Duga: The use case of one Kobo user sharing to another Kobo user, it's internal, can use what they want, it's only when it goes outside of the system it matters, you can make an EPUB CFI that works, but you need to be prepared if it doesn't
Laurent Le Meur: in the EPUB CFI you need to put the information you need, which come from text fragments, the answer is text fragments
… don't care what it is, I just need the text fragments
… it's possible I can share a CFI with someone else, but I can't assume it will work
… everything is a fancy wrapper for text fragments
Charles LaPierre: clarification - why would a company want to use CFI?
<Gregorio Pellegrino> Ace by DAISY uses CFIs :)
Charles LaPierre: is it simpler or faster?
Wendy Reid: good question - let us discuss over coffee
Ivan Herman: what should I do
… ?
Wendy Reid: Discuss after break
Charter Pt 2
Ivan Herman: Overall the feeling is that CFI should be passed over, I'll remove the entry
… the argument that it is a tool mostly for internal reading system function, and has nothing to do with our standards or the annotations interchange
Wendy Reid: Should we say something about why we aren't standardizing CFI, at least in annotations because it could be a reasonable question
Ivan Herman: we should still remove this from the charter, but then have a note in the annotations about why it isn't useful for the purpose of the standards
Brady Duga: I disagree, but we don't need to discuss it today, I don't think we need to explain our decision in the annotations text
Ivan Herman: we could discuss EPUB metadata, do we have enough here for the charter?
Hadrien Gardeur: serialization is in the spec
Brady Duga: reevaluation can have the connotation that we are planning to drop it
Marisa DeMeglio: review?
Susan Neuhaus: one of the use cases that came from the community, people who manage their own local libraries
Marisa DeMeglio: want to be able to add their own metadata about the state, likes, preferences
Ivan Herman: we may not need that because we need only a few use cases for chartering
Wendy Reid: do we need to say that we are developing new use cases? and new technical capabilities?
Hadrien Gardeur: we always take for granted that there are reading systems that have special use cases like specialized libraries, people we don't really talk to right now. We may not need to change anything for chartering, however
Ivan Herman: Is Matt OK with 7 as it is?
Matt Garrish: yes, there are clarifications needed around the spine
Romain Deltour: this might also include changing references to normative standards to better fit into the web system, and look into important parts of how the document is written
Ivan Herman: I'm not sure how to add that to item 7
Wendy Reid: maybe alignment with the web platform will be viewed favorably
Matt Garrish: should it say "and" alignment or "to" align?
Brady Duga: if we say "to" align, what if we clarify something that doesn't relate to the web platform, does that mean we can't clarify anything that isn't part of the web platform? I don't want that
Laurent Le Meur: Do we want WebVTT?
Ivan Herman: we don't want to make it impossible to use VTT, to leave it open in the charter
Charles LaPierre: The adoption of EPUB is higher education, many are reliant on PDFs, which they like because they can make a separate PDF from a page or a chapter. Should we consider something similar for EPUB where it can be chunked?
Wendy Reid: This use case is like what we talked about yesterday, modifying content. Including alternate format providers. DSO? is one of the most important secondary users of the file format. do we need to understand more of their use?
Hadrien Gardeur: The most common use case is creating a preview, technically it is possible to chunk an EPUB but not as easy as with a PDF
… the problem is that you might inadvertently remove HTML that is needed for structure or style. This is true for HTML in general. We can do it, but we run this risk, and could be worse for screen readers
Matt Garrish: We tried distributable objects, and it didn't go anywhere. This feels a little like a retread of the same issue
<Avneesh Singh> Chunking is a good topic for incubation
Gautier Chomel: It might look simple to chunk and epub, but we might lose accessiblity.
… if you want to do it the right way it isn't so easy
Wendy Reid: this falls into number eight, CG incubation. Charles can open an issue
Avneesh Singh: We are trying to bring in the PDF functions, we should look into the reasons and the use case. This is good for incubation
Ivan Herman: I will rework this section and set up a PR with the working group
Brady Duga: do we want to be more explicit about accessibility on this list?
Marisa DeMeglio: everything we do must pass that
Wendy Reid: we are more explicit in the motivation and background section
Ivan Herman: plus we publish the accessibility standard separately.
Full Bleed Images
Wendy Reid: we have one open issue with full bleed images, and the primary use case is light novels, but its similar to mixed formats, covers, some books are using fixed layout intead
Shinya Takami: the situation in Japan is partially fixed layout indicated in the spine. Best practice is no CSS on fixed layout, since that always means one image per page in the spine.
… this topic should be discussed globally.
Hadrien Gardeur: almost every reflowable epub needs this, at least for the cover
… today comic creators put an image into an HTML file, and then in a reflowable system. They inject styles, margins, multi columns. Which can have a negative impact on how the content is displayed.
… some reading systems don't have the ability to show a full page image and this leads to a bad experience. This can be an accessibility issue if an image can't be zoomed into.
… there are a lot of reasons to do this. The Japanese approach is rooted in 3.0.1. The problem is that most RS have different rendering systems for FL and Reflow. That's why we see little adoption outside Japan. In a reflow EPUB, the user has options like fonts and columns. The options for fixed layout can be completely different. From a UX
perspective this opens a lot of questions.
… it is both a technical issue and a UX issue. This is why it is not used outside of Japan
… layout overrides goes well beyond this issue. With those, we could have fixed layout that are not images. So it might be overkill. Perhaps we could solve this problem without going nuclear.
Wendy Reid: I agree, and as long as the cover is identified most reading systems will attempt to display it as fixed. I've notice that most fixed layout readers depend on the viewport dimensions not changing. It might be shrinking or cropping.
Susan Neuhaus: depending on the kind of book, all the images could be a different size. We need to make it clear that you can't rely on a standard viewport.
… there is no where in the spec that requires the same viewport on every page of an FL
Hadrien Gardeur: the viewport problem will be discussed in the Manga group. There are many extensions for providing the viewport. It is not uncommon for comics to have multiple viewports. It makes it harder for some systems to show a full spread. And this is one of the problems about relying on fxl. What do we need to give reading systems for this?
… so they know when something is an image, and which image it is, and then they are more likely to get something out of it that is effective
Susan Neuhaus: My customers do make books of poetry, it's very important that the poetry be displayed with the line breaks etc
Hadrien Gardeur: most of the editors find reflow for poetry unacceptable, I do FXL, but sometimes they demand images even
… I want to see that addressed somehow
Brady Duga: there are two usecases with full bleed, here's the cover, here's a full bleed image inside reflowable text. Are we going to tackle both of these? Are we going to figure out how to identify a full bleed image in the spine.
Wendy Reid: we've had discussions about RS ability to convert FXL to reflowable so it can be customized, but we can still see the original image/styling.
Susan Neuhaus: spine structure could get very strange with a full bleed image within a Reflowable
Hadrien Gardeur: we have full bleed images, but maybe relying on the viewport we could find another way, like using just height. We could rethink how we fit things in general. Fitting isn't always the right option
Samuel Petit: yesterday we talked about comic accessiblity via audio, but there are other use cases like dyslexia, that can interfere with the author's artistic expression. In the EAA, the publisher can just say, well you will modify my content so I don't need to do anything. Perhaps we can get a layout that displays the original page, and then another
way to display the text that is more accessiblity. There is a tension between the artistic expression and things like guided navigation
Wendy Reid: we talked a lot about this in the FXL Accessibility Task Force, the most helpful is to have the content, not flattened like in an image, and I think the same is true for the comics accessibility, knowing the panels and the reading order of the page. With full bleed images we want to be able to have context, and not be thrown out of the
reading order. This a good thing to keep talking about. We could add a property like "full bleed image" similar to "cover" though that won't solve everything
Hadrien Gardeur: to show readers what they want we have to inject things like columns and margins, it would be very difficult in the middle of an HTML resource to suddenly display something full screen. I don't see how we would achieve that. Maybe we need another type of view, you can tap on an image and view it in the viewer. It can handle ARIA and all that.
Wendy Reid: we have two high level full bleed image cases, one is the full bleed image that is a unique resource, and one case that is a full bleed image that is part of a document
Laurent Le Meur: the reading system should know that the spine item is an image, and then it can do whatever it wants. I don't see any other requirement that an image is identified in the spine
Brady Duga: maybe we should have a different term than full bleed. Maybe full bleed is misleading. This is basically inexpensive fixed layout.
Marisa DeMeglio: when we say the reading system should know something is an image, are we talking about an image within an HTML
Laurent Le Meur: I was referring to the requirement without specifying the solution. another requirement should express what the reading system must do. Especially about the user affordances.
Susan Neuhaus: We have a situation in middle grade readers, students beginning to read fluently but don't have the stamina for long novels,
Laurent Le Meur: reflow text with full page images to rest
… clients will accept the current state, even an image with a white border, but they don't like the awkward break in text with large white space, image, then more text
Hadrien Gardeur: what we've seen with layout overrides is when we change things like this and it doesn't get adopted, it causes problems, so I think it should stay as HTML so we don't break things.
… so then how do we provide the hint that it is an image. One option is to include the image as a fallback. This is common in comics. The other way is a a property or new value that conveys that information like we do now for MathML and javascript. The semantics would be slightly different because it contains content. The advantage to the first image is that we don't have to change the spec
… the first is a misuse of fallbacks, but people will have done this for a long time.
… in Japan they use extensions to indicate how the resource should be fitted
gpelligrino: this feature may be used outside novels like for guides and cookbooks. In many cases with a full bleed image there is also a caption. How will handle that?
Hadrien Gardeur: if we have html in the spine, you could put the caption there.
Hadrien Gardeur: we should use HTML in the spine, with an indication that it is actually an image, and the image as a fallback
Laurent Le Meur: using an images as a fallback is and advantage to an RS that decides to display the image, especially if you use a URL for the image, because the path is there
Marisa DeMeglio: can the RS parse the HTML and see that there is nothing but an image?
Hadrien Gardeur: most of the RS don't do this live, I think hinting will help, if I know what the image is, I will have information for displaying the image. It is kind of a balancing act, you wouldn't skip the html, but in preprocessing you would get hints
Marisa DeMeglio: so when you pre process the HTML, you get all the information you need for an image?
Hadrien Gardeur: that's why this is a good solution because it gives them the information they need without pre processing. It could still be misused, of course
Wendy Reid: I know a reading system with two pre processors, and if it detects text it in the first few pages it goes to one renderer, if not it becomes mostly a slideshow. I worry that people will use this as a loophole, and default to images at the expense of accessibility
Susan Neuhaus: I agree we should have some kind of signal, but having an image fallback only could be abused
<Charles LaPierre> +1 Wendy's point
Brady Duga: I know content exists in Japan with fallback images and specialized readers. Are we certain that there isn't other content with image fallbacks? Is this the only example of this technique?
Wendy Reid: I think since reading systems don't support this, most publishers have just picked reflowable or fixed layout
Hadrien Gardeur: if people want to misuse things we can't do much to stop them. I think if we only tell people that something is an image without know which image, it isn't enough for the reading system to do its job
… whenever we start providing an image, we don't trust it, and end up ignoring it. We can't trick people into accessibility even with good intentions
Marisa DeMeglio: if we change anything about fallbacks will it affect accessibility checkers like ACE?
Romain Deltour: I don't think it makes a difference for ACE, but it makes a difference what the reading system does with the file.
Romain Deltour: I think it is the case where the effective accessibility depends on how it is processed in the reading system
Gautier Chomel: we want accessibility to be the principal experience, so the implementation will need some guidance not to use the fallback
Wendy Reid: very clearly we have to talk a lot more about this. This has been really productive. We have to do some research and fact finding about captions and other supplemental parts of an image. We won't officially be working on this until the new charter
Charles LaPierre: is there a situation where you would want a fallback, but not a full bleed? So it fits in the viewport that you already have?
Brady Duga: if we're using an image as a fallback to mean that "this is a full bleed page" it might over ride other uses of an image fallback that isn't intended to be a full bleed.
Charles LaPierre: that's why I think another way to identify a full bleed image would avoid this
Wendy Reid: I could see a use case as a fall back to a 3d image
Hadrien Gardeur: I could see this as having multiple uses, and maybe we think of it as just looking for a better way to display that image because in many ways it is now broken. That's why having a property is helpful, and better than having to get it all from the HTML
… there is good information from the image and good information from the HTML
Wendy Reid: we have more to do. We do need to see if there is a backwards compatibility
[LUNCH BREAK - 1h15m]
Update on Accessibility
Wendy Reid: a11y and FXL.
… prepaginated / roll, we have options
Epub a11y document?
Avneesh Singh: Ongoing. WG impact us, metadata, MO. metadata for AI a11y content generated much broader topic
… Fixed layout will have a big impact on a11y. techniques doc for fixed layout for a11y, we are already working on it by Gregorio, Wendy & Charles
… Manga considerations. DAISY members are using Media Overlays. can we get to a stage where Manga can be accessible.
… as issues come up in the WG we will take up a11y concerns.
Hadrien Gardeur: one cat. of fixed layout where you have HTML where it doesn't cross the spread. another is Manga comics where there is a story telling involved. where you may want to highlight a specific area of the spread.
… you may need specific mappings. News papers and magazines mapping to reflow potentially
… fxl content can be in a few different categories. Real text is easy, but what about the others. how could they be addressed can be a challenge.
Wendy Reid: doc. today what is feasible. you won't be able to meet the current standards, but you will move the needle on a11y.
… I need to do an editorial pass to add roll. that it exists and how to include it. next step future looking work. little more to get into the specific ways it can be used. There may be similar needs. like panel detection and order processing.
… you could present them in a specific order. Yesterday in the manga how to identify speakers, speech bubble vs. background vs sound effects, or side bar text like Jokes.
… what is the hope with presenting it? what is a fxl layout, and what does the reader want to hear in a read aloud scenario.
Gregorio Pellegrino: current best practices and fixed laoyout a11y. we decided to look at HTML + CSS currently haven't trakcked SVG, images with spine. its not a complete work needs work.
Dale Rogers: as producer of comics myself - I agree various levels in comics, what is a11y / legal point of view. what is required vs. nice to have.
… translation vs adaptation. similar to a movie is similar by different than the novel.
Hadrien Gardeur: how do you chunk it. some content in different resources. challenge for a11y.
… resource level would need a structure on top of it. reflowable wouldn't get in the way as much as prepaginated.
… when dealing with a1y
Samuel Petit: audio experimentation we switch from visual modality to text to sound. audio version we want to fall on the side of a11y. legally we want to be compliant. what is an adaptation that falls on the side of a11y vs just an adaptation. Want to define some guidelines
… I enjoyed yesterdays demos. is this just an adaptation or a requirement for a11y.
… 1:1 equivalent with text hasn't yet been met.
Wendy Reid: what can we do as a format to allow for what is possible. authorial intent and what the content is meant to be. we want to keep the spirit of the content, and this can be challenging.
… tools we have now can't really support this yet. how to chunk it and allow for layering.
… its all about choice and how can we ensure everyone can get the content in a format that works for their needs.
Hadrien Gardeur: the emotion that was conveyed in yesterdays demo was done through voice inflections etc. its difficult to do this in an accessible way and some choices of what the audience may want. What do we want to focus on / how far do we go. We may not need a description of the scene as some readers just needs to have the text spoken. but again not all will need just that.
Avneesh Singh: we are a Publishing Maintenance WG, making things accessible. we started making some guidelines for helping tools like Indesign, but thought we didn't want to be too prescriptive. production processes are evolving via AI. maybe Manga can be part of the Maintenance WG. work we are doing with the APA for semantics. interlaced text. real production process / best practices may be part of the CG.
Shinya Takami: we want to discuss the gap between the demo and the current EPUB specs. WG/CG we need to have the people to discuss the problem. wondering mandatory metadata is on the table for the a11y specs.
… DAISY provides ACE but we can't put in the conformance tag since we can't yet meet WCAG. in Japan if we use ACE maybe accessibilitySummary must be requried since we can't put in the conformsTo metadata.
Hadrien Gardeur: how can we gradually do more and more. what you demo'ed what is possible you can do it with pre-recorded audio. full spreads, its technically possible today. if you want to do this but want to zoom into a panel or a panel + Text with special font/spacing this is where we have a gap. divide this in a journey. what is step 1, 2,3. generate on the fly vs. pre-recorded looks different.
Wendy Reid: next steps - obvious we have this in the next charter. an outreach component, production processes especially with the use of AI. semantic gaps how can we make it clearer, panel / reading order. separating out layers of content / background, foreground, aux content.
… getting more Use cases, for FXL, role layout, manga, gaps in spec currently.
… then right some documentation to support all of this.
Avneesh Singh: in charter a11y use cases for FXL then we can make the decisions.
Wendy Reid: anything else with a11y?
Issue Clean Up
Wendy Reid: we did not go though the issues (70 open issues) :(
… #3097
<Romain Deltour> +1 on leaving it deferred
Ivan Herman: some with PRs SVG / Media conformance PR is ready to be merged.
Matt Garrish: Media table stuff is done, but there are some outstanding issues with core media types
… if the resources are not used in the rendering are they even part of the core resources. leave to terminology review
Ivan Herman: 3101 we can probably close within a week with proposed closing flagging.
<Matt Garrish> https:/
Matt Garrish: pintched to zoom are not deferred yet.
w3c/epub-specs#3100
Matt Garrish: we took out a MUST so we are re-inserting it back in.
Ivan Herman: this branch has a conflict.
Matt Garrish: I can do it later. once we are done.I will fix the conflict if its still a problem.
w3c/epub-specs#3101
Wendy Reid: no opposition. ok to merge
<Avneesh Singh> I need to switch to another meeting now. Thanks
Ivan Herman: has been accepted by Brady / Ivan for over a week. Matt agrees its ok to merge
Wendy Reid: Ok to merge
w3c/epub-specs#3044
seuneu: specialized libraries from 2021
Hadrien Gardeur: they produce a11y comics but didn't think they could do epubs. pane by panel with audio text on visual.
… presentation from Norway / Sweden (MTM / Nora) they had to re-invent the wheel to do what they want.
Wendy Reid: pulled from CG repo.this is an EPUB 3 issue that is differed to EPUB 3.5
Ivan Herman: I put in some tags
w3c/epub-specs#3043
Hadrien Gardeur: range as source of text. current spec allows you to use anything but only allows currently for IDs the spec is under specified.
Marisa DeMeglio: we could allow for it and it probably will work.
Ivan Herman: so this is for the future
Wendy Reid: yes
w3c/epub-specs#3042
Wendy Reid: get citation from George (Reading System)
… annotations use case
George Kerscher: metadata issue as well
Ivan Herman: charter we refer to citations so we can deferred as part of annotations?
Laurent Le Meur: RS could export citations separate from annotations. its related but export of annotations for citations.
Ivan Herman: there is multiple citation export types.
George Kerscher: is it possible to do a partial export of a citation with the expectation the user will flesh out what is missing?
Ivan Herman: citations tech. fields is different than other types of content
Wendy Reid: this would be a reading system feature they would build, if a RS wants to export citations what is missing from the EPUB spec to make that feature possible.
Ivan Herman: not really anything in the package document can hold the information needed.
… number of metadata fields may be needed to create a specific citation which we already support
George Kerscher: blind user who put this forward is struggling because they can't get the page # he is on in order to get the citation. If we can get the title, page # is on and that would be a good start. min few types of data to get the citation. give them something to work with.
Wendy Reid: what is not in the spec, the page list is there so technically the RS should be able to get the data
George Kerscher: if this is a published book there is a req. that page #'s are present.
seuneu: any metadata in the package EPUBCheck will flag them if they add their own metadata.
Ivan Herman: you can link to a metadata file and you can put a link from the Package document to that metadata file.
Marisa DeMeglio: citation styles: apa,Chicago, etc. do have equivalents for websites so that would be a good start. some sort of locator. is whats needed? to say get to the chapter heading for example.
George Kerscher: scholarly is asked to get the Pring page number and they have to go to the PDF version to find that page # when asked when doing citations for their course.
Wendy Reid: publishers should have this already in the EPUB if they don't all users would have this same problem, not just blind.
Ivan Herman: the citation checkers are looking for an arabic # for page #'s. URL is now being accepted.
Brady Duga: George we have Page # through page list, and the title / author
Wendy Reid: IV and 4 are different we have this. what do we need in the spec. for this?
George Kerscher: I think we are pretty good with that. we need to go yell at the publishers / reading systems.
Hadrien Gardeur: this is a universal issue. I don't know if shouting is going to do any good. because using a string is a bad idea. its not implemented / UX issues.
… maybe look back at what we have done
Charles LaPierre: In addition to students, I recently had an author on the Benetech board who did a book, it took her over 2 week to find all of the citations of all the different digital books
Hadrien Gardeur: nightmare to go through things
… lots of time to get assistance, but needed physical books and help to get it
Wendy Reid: we don't need to make a change, I propose closing this issue. we can open a new issue for doing page #'s better but that is separate.
Charles LaPierre: talk to citation bodies, more and more are digital first which never had page #'s
Ivan Herman: books, scientific papers, etc publishing PDF.
Wendy Reid: good to close this.
w3c/epub-specs#2877
Ivan Herman: we need to raise this with the W3C,
Matt Garrish: they say its fixed. need to double check.
Ivan Herman: we can close this for us.
George Kerscher: I will add a comment with latest release of FIDO it can take an EPUB / PDF and look at them side by side can add page#'s in EPUB.
Wendy Reid: those are all the unlabelled issues.
AOB
Call times
Marisa DeMeglio: call time is not West coast friendly 6AM :)
Wendy Reid: propose some times that could work.
Seth Delackner: I can usually not meet due to very late at night for me.
… I find Slack are very useful.
… asynchronous Slack could be very useful.
Wendy Reid: chairs will take this off-line. I like slack, and is not archived, so we would need some scafolding to record decisions.
Ivan Herman: there was alternative meetings china / west coast then another for east / EU. but miss some information so its hard and hasn't worked in the past when attempted.
Educational publishers' issues
Gregorio Pellegrino: educational publishers EAA, most embraced EPUB for reflowable textbooks most provide a closed platform reflow / fixed and you switch between the two. need something for educational publishers, interop for vendors for producing the book. how to manage the fixed vs. reflowable versions. maybe a charter topic.
… being able to fill in exercises in EPUBs.like a paper book and saved either in or outside the EPUB. we dont have a standard way to export the notes.
Ivan Herman: web you would have JS and stor the data in the browsers db. all this works in EPUB if it accepts scripts. UseCase for allowing scripts in EPUB it is doable but needs development. not usre if publishers will do the work
Hadrien Gardeur: most of the time the LMS backend knows the student and would normally do this. this seems out of scope for EPUB.
… I don't see how EPUB is a good fit for this. do it on the Web. we don't have any info, who the student is storing form data wouldn't solve the problem.
George Kerscher: we have been working with MS. we can go to a MathML expression and copy it out. maybe an exercise worksheet in epub that you can copy it and paste it into word.something more specific to export page to Word/GoogleDocs.
Wendy Reid: is the desire for this in EPUB we want to do this outside of an LMS sense that LMS my school uses is not accessible and I like my own RS that is fully accessible. is that the ask here?
Gregorio Pellegrino: in southern EU we only have digital books with exercises printed as text inside the book.
Wendy Reid: in the EPUB could link to the page on the web for this
seuneu: concerns safety security privacy issues may be an issue. we have no one from Southern Hemisphere.
… and missing big publishers here.
Ivan Herman: we have tried.they don't have tech. people to participate in the group.
Gautier Chomel: breakout in TPAC Thorium Reader showed this was possible
Ivan Herman: english language / membership fees
Matt Garrish: I wouldn't dismiss this outright. Norton folks asked Matt directly. Matt said to raise this in the CG.
… there is still a need. scriptable components could do this.
Ivan Herman: CSS has gone though a number of innovations for this
Dale Rogers: from Higher ed / teaching online we ran into this. templates, worksheets was in an LMS and after the class ends all this is now closed off. zip file that the instructor could download or something for the student can download. I am thinking this type of thing could be better left to the publisher to support the book.
Wendy Reid: I don't want to dismiss outright. but Security / Privacy most of their concerns sending reading progress to servers, but we never opened any ways to do this in the spec. there are regions without LMS and even if there is students may not be able to use the LMS. what can't the spec do now?
Gregorio Pellegrino: fixed to reflowabe / JS could do it we don't have a declarative way. story/retrieve that data.
Ivan Herman: Should we put Out of scope in the charter for (Multiple Renditions)
Brady Duga: for the MR you brought up Gregorio what would we need to put in the spec?
Gregorio Pellegrino: you can't put the content in the saeme container.
Brady Duga: do we need to put something in the spec?
Gregorio Pellegrino: standardization or a best practice since this is not being implemented the same way
<seuneu> +1 Wendy Reid
Hadrien Gardeur: we did vote on it in 3.4 current limitation you can't have something outside the spine in a different layout.
… need another reading order content document. new toc / mapping.
… question can this be achieved without duplicating everything.
… which adds complexity and hackey using CFIs for internal references.
https:/
<Daniel Weck> well done everyone!!!