12:56:45 <RRSAgent> RRSAgent has joined #epub-fxl
12:56:45 <RRSAgent> logging to https://www.w3.org/2021/06/22-epub-fxl-irc
12:56:47 <Zakim> RRSAgent, make logs Public
12:56:48 <Zakim> please title this meeting ("meeting: ..."), wendyreid
12:57:04 <wendyreid> meeting: Fixed Layout Accessibility TF Telco June 22, 2021
12:57:07 <wendyreid> chair: wendyreid
12:57:17 <wendyreid> date: 06-21-2021
12:58:13 <Ken_Jones> Ken_Jones has joined #epub-fxl
13:00:01 <MattChan> MattChan has joined #epub-fxl
13:00:30 <gpellegrino> gpellegrino has joined #epub-fxl
13:00:44 <gpellegrino> present+
13:00:47 <Rachel_Osolen> Rachel_Osolen has joined #epub-fxl
13:00:58 <MattChan> present+
13:00:59 <avneeshsingh> avneeshsingh has joined #epub-fxl
13:01:00 <laurab> present+
13:01:36 <avneeshsingh> present+
13:01:51 <MattChan> scribe+
13:02:27 <CharlesL> CharlesL has joined #epub-fxl
13:02:33 <CharlesL> present+
13:02:33 <gpellegrino> q+
13:02:37 <MattChan> wendyreid: are there any updates on the projects people have taken on?
13:02:41 <wendyreid> ack gpellegrino
13:02:45 <MattChan> ... Rachel_Osolen has sent info on a11y comics
13:02:50 <mgarrish> mgarrish has joined #epub-fxl
13:03:20 <MattChan> gpellegrino: i went through parts of the epub spec and found some interesting things that might help create some fxl epubs
13:03:31 <duga> duga has joined #epub-fxl
13:04:03 <duga> present+
13:04:07 <MattChan> ... one is the element collection inside the manifest that allows you to group parts of elements in metadata for describing content inside the epub
13:04:19 <MattChan> ... maybe this could allow us to group FXL content with non-FXL content
13:04:39 <MattChan> ... the other is when we have 2-pages with a single image across the spread
13:04:54 <MattChan> ... for FXL you can set the property page-spread-center
13:05:14 <MattChan> ... this sets the single content across both pages
13:05:45 <MattChan> ... so authors can instead use a single page of HTML content with this property set
13:06:09 <MattChan> wendyreid: page-spread-center is partially supported
13:06:21 <gpellegrino> https://www.w3.org/TR/epub-33/#page-spread
13:06:26 <Hadrien> Hadrien has joined #epub-fxl
13:06:47 <MattChan> ... but it isn't necessarily used for spreads across two pages, because a lot of RSes don't respond well when FXLs contain pages with different viewport sizes
13:06:50 <gpellegrino> https://www.w3.org/TR/epub-33/#sec-collection-elem
13:07:14 <MattChan> ... not universal behaviour, but that is a possible roadblock to using that technique
13:07:42 <MattChan> Hadrien: on collection, this was used to group together a number of resources and say something about them
13:08:09 <MattChan> ... e.g. in education space, to have sub-units of a publication that could be distributed together, or have a semantic meaning together
13:08:34 <MattChan> ... i.e. introducing the ability to create logical groups at the spine level
13:08:38 <gpellegrino> «rendition:page-spread-center property indicates to override the synthetic spread mode and render a single viewport positioned at the center of the screen.»
13:08:55 <MattChan> ... but since it was mostly an EDUPUB thing, it is not highly implemented
13:09:16 <CharlesL> q+
13:09:18 <MattChan> wendyreid: we also talked about the alt-text section of the documentation
13:09:21 <wendyreid> ack CharlesL
13:10:13 <avneeshsingh> q+
13:10:14 <MattChan> CharlesL: we did a branch of the doc, and started adding some basics in there. Once we are ready for group review we'll do a PR
13:10:17 <wendyreid> ack avneeshsingh
13:11:29 <CharlesL> q+
13:11:37 <MattChan> avneeshsingh: administrative concern: is the alt-text we are talking about specific to epub? If not, then we normally have a CG do this sort of work if it isn't specific to a particular spec
13:11:43 <wendyreid> ack CharlesL
13:11:45 <MattChan> wendyreid: this is specific to only FXL epub
13:12:48 <MattChan> CharlesL: we're thinking of having a sample FXL with examples of how someone might write alt-text for it, but in the resources section we just have some generic alt-text guides for now
13:12:50 <Ken_Jones> q+
13:12:54 <wendyreid> ack Ken_Jones
13:13:57 <MattChan> Ken_Jones: re: the 2-page spreads and Rachel_Osolen's a11y comic book email, a number of months ago I opened an issue in a DAISY repo to ask about describing 2-page spreads, and also spreads with photo montages
13:14:29 <Ken_Jones> https://github.com/daisy/kb/issues/35
13:14:34 <MattChan> ... in those instances, could we tile the areas of the spread, so that we can attach alt-text to those tiles?
13:14:54 <MattChan> ... as a way of locating the alt-text within the spread, and for reading order purposes
13:15:58 <MattChan> ... one way I was thinking of doing that was to have an invisible object that we could attach alt-text to, that we could then drop into the reading order
13:16:24 <mgarrish> q+
13:16:35 <wendyreid> ack mgarrish
13:16:56 <MattChan> ... this would present an obstacle to any technique that would transform a visual view to a textual view
13:17:31 <MattChan> mgarrish: not sure about invisible objects because that assumes the viewer can't see the image at all
13:18:07 <MattChan> wendyreid: we did previously discuss the idea of using mappable regions for this purpose
13:18:34 <CharlesL> q+
13:18:38 <wendyreid> ack CharlesL
13:18:40 <MattChan> mgarrish: right, better to try to build off existing technologies, and if that doesn't work, then we can take it back to a wider group
13:19:08 <MattChan> CharlesL: so in the case of a comic book with speech bubbles, sometimes in a small scripted font, for low-vision users this isn't good
13:19:19 <MattChan> ... and zooming in results in pixelation
13:20:00 <MattChan> ... the idea of zoning, plus being able to toggle user-defined fonts, font sizes, etc.
13:20:19 <MattChan> ... is that something that we can tackle in this group, or does it need to be run up to a higher level of W3C?
13:20:23 <duga> q+
13:20:30 <wendyreid> ack duga
13:20:38 <MattChan> wendyreid: zoning is something we might already have existing web tech for
13:21:06 <MattChan> duga: we've basically implemented the text features you mentioned, but we ran into limitations: available size, overlapping
13:21:26 <MattChan> ... but they're surmountable problems. We did end up implementing
13:21:41 <MattChan> s/available size/available space
13:22:14 <Ken_Jones> https://paper.dropbox.com/doc/Visual-to-Textual-Explainer--BNIYen9D3PBflNmBs6yax4JJAg-K3nAwKn2vlVqRpFyZ9KHN
13:22:42 <MattChan> wendyreid: Ken wrote an explainer about the idea of potentially turning FXL into reflow via a process
13:23:19 <MattChan> ... this bleeds into changing epub to do something new, so we might want to bring this to the wider W3C
13:24:01 <MattChan> ... what we might potentially need is a way to signal to RS that a single FXL contains both FXL and reflow modes, and that user should have the ability to switch between
13:24:11 <duga> q+
13:24:15 <wendyreid> ack duga
13:24:21 <MattChan> ... but we still have to think about what to do about images, audio, etc. in the reflow
13:25:16 <MattChan> duga: the problem i'm having trouble seeing the solution to is how to merge pages, e.g. where information like a paragraph is split across a page
13:25:20 <Hadrien> q+
13:25:26 <wendyreid> ack Hadrien
13:26:17 <MattChan> Hadrien: if we look at what has been tried in the past, there was the idea that regions across a spread could be mapped to IDs. With regard to how it looks, it depends on whether it is authored or not
13:26:33 <MattChan> ... if it is authored, both could look good. We just have to figure out the mapping question
13:26:41 <MattChan> ... of course, it is a lot of extra work
13:27:07 <Ken_Jones> q+
13:27:10 <wendyreid> ack Ken_Jones
13:27:51 <Hadrien> q+
13:27:55 <MattChan> Ken_Jones: i put in a section on page breaking, in which i proposed that we say when a page break is wanted in the reflow document
13:28:06 <MattChan> ... otherwise, the pages would run sequentially
13:28:37 <wendyreid> ack Hadrien
13:28:49 <MattChan> ... also, a lot of FXLs are designed to break cleanly across pages, so I'd have to see a problem example to know what to do
13:29:57 <MattChan> Hadrien: one question we need to ask is whether we'd be happy enough to be able to indicate a reflow equivalent, maybe on a spread-by-spread basis
13:30:14 <MattChan> ... not to get too far back into multiple-renditions
13:30:28 <MattChan> ... maybe we leave the idea of mapping individual fragments of a document for a later time
13:30:52 <MattChan> wendyreid: and I don't want to fixate on this one issue and have it block our progress
13:31:32 <MattChan> ... if we can get it working to where we can get a reflowable version of an FXL, then I would consider that iteration 1 complete
13:32:14 <MattChan> ... would love it if everyone could read over Ken_Jones's document, and think about the problems presented, and come up with some solutions for them
13:32:22 <avneeshsingh> Conceptually I very much like the idea of converting fixed layout to reflowable.
13:32:23 <Ken_Jones> q+
13:32:32 <MattChan> ... and while you are doing so, please keep in mind that the webbier the better (i.e. reusing existing tech)
13:32:32 <wendyreid> ack Ken_Jones
13:32:42 <MattChan> Ken_Jones: shall we make a list of existing issues?
13:33:45 <MattChan> wendyreid: your document has "advice required" tags, so those are a good start.
13:34:43 <MattChan> ... But yes, 1) what to do about images, children's books where images are most of the content, 2) other media content (audio, video), maybe via media queries?
13:35:16 <MattChan> Hadrien: media queries is also problematic because it might be something that the RS doesn't have access to
13:35:37 <MattChan> ... might be better to rely upon things that we already know are implemented in epub
13:36:01 <MattChan> wendyreid: another problem is how to deal with text styling/alternate style sheets
13:36:16 <gpellegrino> q+
13:36:21 <gpellegrino> q+
13:36:22 <MattChan> Ken_Jones: is there documentation on alternate style sheets?
13:36:29 <MattChan> wendyreid: yes, its part of the HTML spec
13:36:30 <wendyreid> ack gpellegrino
13:37:05 <MattChan> gpellegrino: they are similar to media queries, and you have keywords to tell UA to use the alternate style sheet
13:37:18 <wendyreid> https://developer.mozilla.org/en-US/docs/Web/CSS/Alternative_style_sheets
13:37:50 <MattChan> ... there are other keywords, e.g. to render the document in alternative colors, or with a magnifier, greyscale, black and white
13:38:07 <MattChan> ... we'd have to check which of those keywords are supported or not. It is not universal.
13:38:33 <mgarrish> q+
13:38:49 <wendyreid> ack mgarrish
13:38:53 <MattChan> Ken_Jones: so this would be chosen by the user before they open the FXL?
13:39:41 <duga> q+
13:39:50 <MattChan> mgarrish: there are 3 ways to get alternate style sheets. 1) Just have additional style sheets, 2) the media query type way, and 3) the epub specific alternate styles
13:40:11 <wendyreid> ack duga
13:40:50 <MattChan> duga: the epub specific alternate styles are rarely implemented. I would discourage.
13:41:07 <CharlesL> q+
13:41:29 <MattChan> ... and I think it would be fine for users to switch style sheets mid-reading. Whether or not that works well is really down to the RS
13:41:30 <gpellegrino> CSS media query: https://developer.mozilla.org/en-US/docs/Web/CSS/Media_Queries/Using_media_queries
13:41:35 <wendyreid> ack CharlesL
13:41:50 <duga> q+
13:41:54 <MattChan> CharlesL: couldn't there be js hooked up to a button that changes the style sheet?
13:41:56 <wendyreid> ack duga
13:42:14 <Hadrien> q+
13:42:30 <wendyreid> ack Hadrien
13:42:32 <MattChan> duga: not sure that gets you what you want, because you want the RS to do a bunch of things differently, like enable pagination etc.
13:42:56 <MattChan> Hadrien: also any js could be blocked by the RS for a number of different reasons. We're not on the web, after all (and even if we were on the web)
13:43:20 <MattChan> ... no matter which technique we use, there is also the problem of completeness
13:43:25 <avneeshsingh> sorry, need to switch to another call
13:44:06 <MattChan> ... the other challenge is that if we want user to be able to switch between these, then it becomes a UX problem
13:44:27 <MattChan> ... better if this feature is known to the RS, and then the RS could help you select it
13:45:57 <Hadrien> q+
13:46:06 <wendyreid> ack Hadrien
13:46:17 <MattChan> wendyreid: for next call, everyone please read Ken_Jones's explainer doc and try to come up with some solutions. Don't have to have complete solutions. Just suggestions is okay
13:46:41 <MattChan> Hadrien: have we explored the possibility of solutions based on multiple OPFs?
13:46:55 <MattChan> ... especially if we're setting aside the issue of how to handle fragments of content
13:47:18 <mgarrish> q+
13:48:08 <gpellegrino> q+
13:48:11 <MattChan> wendyreid: no. The biggest change we've considered is adding another meta property to say that "this is a visual to textual epub". With maybe something to signal multiple style sheets. We want to stay away from multiple-renditions and multiple packages
13:48:12 <wendyreid> ack mgarrish
13:48:27 <MattChan> Hadrien: fallbacks?
13:48:34 <duga> q+
13:48:41 <MattChan> wendyreid: not really, not sure about level of implementation
13:48:50 <wendyreid> ack gpellegrino
13:49:29 <Hadrien> q+
13:49:36 <MattChan> gpellegrino: about fallbacks, what would trigger those fallbacks? Normally the trigger is automatic, i.e. when the RS can't do something the epub wants it to do
13:49:42 <wendyreid> ack Hadrien
13:50:29 <MattChan> Hadrien: so media type is one thing you can fallback on, but also the idea that you could move between resources if different languages, or even if one resource is FXL and other is reflowable
13:51:09 <MattChan> ... i.e. the core idea of multiple renditions, but without the implementation cost of having multiple packages
13:51:24 <wendyreid> ack duga
13:52:04 <Hadrien> q+
13:52:06 <MattChan> duga: i don't think fallbacks are off the table. We should continue to consider them. But my big concern is potentially conflicting fallback chains
13:52:13 <wendyreid> ack Hadrien
13:52:53 <MattChan> Hadrien: and maybe fallback is not the right term in this case. In Readium we have the concept of alternates, and you can say that "this alternate is available in... different media type, resolution, etc."
13:53:11 <MattChan> ... and what you trying to do is maybe define concept of alternates at the spine level
13:53:26 <MattChan> ... potentially giving us the benefits of multiple renditions without the complications
13:53:53 <MattChan> wendyreid: main obstacle to fallbacks is that they are under implemented right now, but maybe this gives new reason for RS to enable fallbacks
13:54:13 <gpellegrino> q+
13:54:27 <MattChan> ... is it worth building a tiny sample of this textual to visual technique?
13:54:49 <MattChan> Ken_Jones: happy to do that? Shall I prioritize this before I write the next section of the doc?
13:54:55 <wendyreid> ack gpellegrino
13:55:04 <MattChan> wendyreid: yes, it doesn't need to be a huge example, even 1 page
13:55:48 <MattChan> gpellegrino: along with the example, maybe we could have a mock-up of what it would behave like (e.g. maybe running in a browser or something)
13:56:14 <MattChan> Ken_Jones: i was going to just have two different documents, to show what the textual version would look like, and what the visual version would look like
13:56:41 <MattChan> ... i've also been invited to prepare a short video/presentation on a11y FXL
13:57:06 <MattChan> ... should I just give them an overview of the topics we've discussed?
13:57:36 <MattChan> wendyreid: the idea behind the videos is just to communicate to the community what we've been doing behind the scenes
13:58:01 <MattChan> ... so the best practices document, the idea of visual to textual. It should be very high level.
13:58:21 <MattChan> ... maybe you can just show us a bullet point list of what you're planning to include
13:58:48 <CharlesL> CharlesL has left #epub-fxl
13:59:07 <duga> duga has joined #epub-fxl
14:00:31 <laurab> laurab has left #epub-fxl
16:03:14 <laurent__> laurent__ has joined #epub-fxl
16:16:41 <Zakim> Zakim has left #epub-fxl
16:25:52 <duga> duga has joined #epub-fxl