EPUB 3 Working Group FXL A11y Task Force Telco — Minutes

Date: 2021-04-20

See also the Agenda and the IRC Log


Present: Wendy Reid, Matthew Chan, Laura Brady, Deborah Kaplan, Rachel Osolen, Brady Duga, Charles LaPierre, Gregorio Pellegrino



Chair: Wendy Reid

Scribe(s): Matthew Chan


Wendy Reid: we had homework from last week, just going through the list
… review of Daisy KB

Ken Jones: I spent a few hours looking through Daisy KB FXL documentation
… spotted a few things that I thought were out of date/that I thought were wrong
… I’ve done a write up on DropBox paper
… this is like an index page, with comments
… I’ll just go through them now
… each entry is a link to the KB itself
… (feel free to add to my comments)
… KB0 is a naming consistency issue
… KB1 seems to be written from particular point of view (“are commonly” vs “may be”)
… I would add paragraph about benefits of fxl format
… reference to “where epub consists of primarily FXL…” is that referring to something deprecated?

Brady Duga: even without multiple renditions you can put pre-paginated/fixed layout pages in a reflow epub
… no real support for this in the wild
… by the way, are we taking over the Daisy KB? Or is this just recommendations?

Wendy Reid: we’re making recommendations, and mgarrish was here last week and in agreement

Ken Jones: feel free to add to my comments clarifying these sorts of points
… re. WCAG AA conformance, add resize text
… KB2 more references to fxl within reflow epub
… and found a broken link in the KB2 document
… recommend moving a summary paragraph up to the top of the document
… KB3 re. images in spine, this is not really something i’ve seen

Gregorio Pellegrino: i’ve seen it in manga created from PDFs
… this was in early days of FXL

Ken Jones: so we might talk about this in best practices
… it seems that whoever wrote this thought of manga as the main use of FXL

Gregorio Pellegrino: You can do it, but you have to have a fallback (an XHTML)

Wendy Reid: I feel like we should just steer people away from this

Ken Jones: maybe make it more clear at top of document that it is not a good way to do it, and is largely unsupported
… KB4 found a typo
… clarification that since you can embed raster img in SVG, there are some limits to scalability
… KB5 is WCAG conformance table
… I would add note that nature of FXL makes it impossible to achieve WCAG AA or higher
… and recommend, again, loosening language re. image-based FXL to cover uses other than manga-type publications
… clarifications re. the fact that in FXLs contrast and font choice can be issues
… document suggests that in some FXLs allow users to change appearance of text. This seems uncommon.

Brady Duga: Some RS may allow users to customize line spacing
… there’s no rule against it, but nobody does it

Ken Jones: that might surprise some authors of FXL, who might not expect that in their publications
… KB6 seems to suggest that the problem is finding a place to put the image descriptions, but often image descriptions are not visual?

Gregorio Pellegrino: I think that is in reference to long descriptions, not specifically alt-text

Ken Jones: well then it could simply be a link out to a different page? I still think the language in the KB could be updated
… continuing, animations have long been supported in FXL
… has anyone seen FXLs with extended image descriptions? How was it done?

Wendy Reid: i’ve seen it as an interactive element
… tapping an arrow under the image would open an overlay
… but only in highly interactive content
… same way definitions and videos were implemented (tap to reveal pop-up)

Ken Jones: Q re. Daisy KB generally. How to describe images that are spreads, or montages (i.e. combinations of smaller images)
… i’ve opened a github issue about this

Ken Jones: https://github.com/daisy/kb/issues/35

Ken Jones: i think summarizing that issue in the KB would be useful
… KB7, no suggested changes

Ken Jones: https://paper.dropbox.com/doc/Suggestions-for-additional-DAISY-KB-articles–BJOEvLh5aCOg9bXd3Rhz3YGzAg-MhUnnl1PcYVEMBVmeaC6A

Ken Jones: i’ve got suggestions for where there are currently gaps
… for additional articles
… one for structure (e.g. TOC)
… one for Tables
… one for interaction
… maybe one for comparison between FXL and PDFs?

Rachel Osolen: re. descriptions for spreads
… some resources recommend that the entire image description appear all on the first half, with the second half simply containing text referring back to the first half
… a different way is to split the description over the composite parts of the spread (works well for children’s books)
… for montages and complex images, recommendation is to provide a general layout of the image, and then drill down and describe each part
… and keep in mind that not all parts of such an image need to be describe in the same level of detail

Charles LaPierre: alt-text is great, but that is only available via AT
… how do we describe these images to people with, for example, cognitive disabilities?
… we have to think about not just screen-readers
… also, when alt-text gets very long, that’s a problem for editing
… no support for semantic markup within alt-text
… that’s why extended desc are good
… the other best practice should probably be to make the image itself clickable (e.g. if you have no space to add a click here link)
… that would then bring up HTML rendered text equivalent of the image

Brady Duga: re. splitting images into pieces for describing complex images
… is this discussed in the KB?

Ken Jones: its not in the KB, but its in the daisy github issue (above)

Brady Duga: about alt-text, its not necessarily true that only screen readers access info n alt-text or ARIA label
… we’ve extracted ARIA labels to do other things, e.g. getting definitions, highlighting words
… we could recommend that RS do other things with this sort of data that is already available

Wendy Reid: we don’t need to get too deep into alt-text, this is fairly well understood already
… but extended desc has an UI component which may be more of an issue for FXL

Ken Jones: so, would that sort of desc be something authored?

Charles LaPierre: i thought so, yes. And then it would be up to the RS how to access that information
… but the idea of the RS adding something on top of the book is like what we are doing with Image Share
… but that requires plug-ins

Wendy Reid: this is great
… and because this is hosted in github we can probably directly help with the edits as well

Charles LaPierre: +1 great job Ken!

Wendy Reid: what kind of content uses FXL? Common use-cases for FXL.
… talked to Rachel_O about this

Rachel Osolen: started working on this. I’m waiting on BNC. Wanted to reach out to eBOUND.
… I have a list of types of books which commonly use FXL
… hoping to get actual numbers

Brady Duga: its a hard question. Are we asking who sorts of books are MADE using FXL? Or what FXL books are actually read?
… if read, then its going to be mostly manga
… might not be worth asking 2nd question

Rachel Osolen: in terms of what is being made, then in Canada its probably going to be comics, children’s books, maybe textbooks, and anything else image heavy
… would be nice to get data from distributors

Brady Duga: yes, sales would be very useful

Charles LaPierre: we’re working on a set of guidelines for libraries and bookstores to show the a11y metadata in books
… i’m wondering if we should also recommend that they give an additional warning where the book is FXL

Ken Jones: i wonder whether its worth mentioning the interactive aspect of FXL?
… its under used, not many publishers use it
… some educational publishers want this sort of interactivity

Deborah Kaplan: i don’t see anything in Schema metadata specifically about FXL
… should we recommend putting something about FXL in a free-text section of the Schema?
… if the Schema vocabs are not set in stone, maybe recommend that something be added to identify FXL?

Charles LaPierre: yes, we can add values to vocab
… but what I was referring to was using what is already in the metadata to identify FXL
… re. interactive. Has anyone seen comics where the reader zooms into the publication, and allows you to browse panel by panel?

Wendy Reid: yes, comixology (Amazon), kobo allows you to zoom, Google Play has bubble detection

Brady Duga: so we present the entire page, and then increase the size of the text while making it appear still part of the publication
… useful for low vision, but there is a limit to how big you can make the text
… also allows you to read the text in order separate from the images, uses region navigation internally to do this

Ken Jones: I read the EUAA article by laurab
… so should not have a single sensory way of reading the book
… rather than forcing aspects of a11y in FXL, which is difficult, can we consider a way of reading an FXL that depends on just the text in the document?

Wendy Reid: we talked about this last week, i think, with multiple renditions
… there’s this idea that we shouldn’t have separate documents

Ken Jones: i would want it to be a single publication

Wendy Reid: i think we should focus on making sure that whatever is going on behind the FXL IS that textual equivalent
… in terms of what we say in our best practices, etc. I think that’s what we should be aiming for

Brady Duga: about detecting when a book is FXL based on metadata, we kind of do that with our alert to people that certain content is better on large screens
… but also, just knowing that something is FXL doesn’t necessarily tell you how accessible the book itself is
… don’t want to encourage throwing out the good with the bad

Wendy Reid: yes, reflowable books can be really poor from a11y point of view too

Brady Duga: even from RS side, knowing whether a book is FXL or not doesn’t say anything about accessibility

Ken Jones: i think we can use Schema accessModes to describe things

Wendy Reid: I’ll have an outline of our progress for next week
… will get in touch with business group re. getting FXL sample content that we can share
… hopefully BNC can help us with data about FXLs
… thank you everyone, and thank you Ken!
… bye everyone!