W3C

Publishing Maintenance Working Group Telco

02 July 2026

Attendees

Present
Abe Jellinek, are, Dale Rogers, Brady Duga, Gautier Chomel, George Kerscher, Grigorily Manucharian, Gregorio Pellegrino, Hadrien Gardeur, Ivan Herman, Masakazu Kitahara, Matt Garrish, Shinya Takami, Susan Neuhaus, Toshiaki Koike, Wendy Reid
Regrets
Avneesh Singh
Chair
Susan Neuhaus, Wendy Reid
Scribe
Brady Duga, Wendy Reid

Meeting minutes

Wendy Reid: As we are going through issues we see there are a bunch in propose closing

Propose Closing - https://github.com/w3c/epub-specs/issues?q=is%3Aissue%20state%3Aopen%20label%3Apropose-closing

Wendy Reid: so we would like to clean those up
… we will work through from oldest to newest

w3c/epub-specs#2691

Wendy Reid: first is extended description in EPUB 3, from Gregorio Pellegrino
… there are some suggestions to ARIA, but I don't see any changes we need in epub 3
… where are we?

Ivan Herman: There were 2 choices
… the epub reading systems tag is misleading, it is a11y
… techniques
… And that would unblock CR
… ARIA doesn't depend on us, we can just suggest to them
… I propose to remove the epub tags and mark it deferred and we are done

Matt Garrish: It is a proposal to ATA so it doesn't really fit
… we have a doc with the proposal, and we have identified the issues, but we don't know where it will go
… I don't know if we need to defer, since we have no idea when we will talk to APA
… so just remove the tags

Wendy Reid: We will probably meet before tpac like we did last year
… I am happy to remove the extra spec tags, and maybe in the future move this to them

w3c/epub-specs#2716

Wendy Reid: next is one about content requirements for HTML

Brady Duga: I can't see any reason to keep it open

w3c/epub-specs#2719

Wendy Reid: Great! Next is ARIA structural semantics
… we had some caution in 3.4, is it already addressed?

Matt Garrish: This is a request to go over html elements with semantics, but this seems outside our scope
… there isn't a clear request here
… But how ARIA applies to HTML doesn't make sense for EPUB. There are better resources available
… so there isn't much we can do

sueneu2: This is a bigger question. The poster is frustrated with how the spec relates to topology
… This is a bigger question about how we inform the community about what they should do

Wendy Reid: In this case it is hard, since we don't control the technologies in question

<Gregorio Pellegrino> +1 to Matt

Matt Garrish: This really belongs to ARIA and HTML, it isn't our domain to get in here. It is up to HTML
… I don't see it fitting in here. Entirely outside our domain.
… and ARIA and HTML folks will gety annoyed with us

Dale Rogers: It is not part of the spec, but referring them to an explainer? It is like a puzzle and just tell them how it works
… Maybe just point at an explainer or note to help people

sueneu2: Thinking on the kindest and most helpful thing, is it just to close, or is it to add the resources Matt mentioned and leave it open for while?

George Kerscher: I say we add the DAISY KB links and close it

Matt Garrish: Where are we putting them?

George Kerscher: Just the bug

w3c/epub-specs#2765

Wendy Reid: From Gautier Chomel_ about moving annotations issues
… is this done?

Gautier Chomel_: Not completely
… some issues are still in the CG. we just need someone with permissions to do it
… there is a link to the old list in the issue
… for all of them, we need to transfer them

Ivan Herman: This should have been done a year ago, but doing it right now makes me uneasy
… we should review those to see if they still apply

Gautier Chomel_: Mostly they are addressed, but I would like to have everything in the same place
… since it is where the annotation discussion is happening
… Today to see what we have fixed I need to keep 2 repos open, I would prefer even the closed ones to get moved

Ivan Herman: I don't seem to have the permissions on the CG side, and you don't have the rights on the spec side
… I will give you rights on the spec side so you can do it

Ivan Herman: But this is less urgent since it isn't under CR at the moment

sueneu2: Once it is in the repo, should we assign them to someone to make sure they are addressed? Some process

Ivan Herman: I think Gautier Chomel_ should move them and review them

Wendy Reid" Ok, Gautier Chomel_?

Gautier Chomel_: yes

Wendy Reid: And when you are done with that, please close this issue

w3c/epub-specs#2946

Ivan Herman: I don't understand. In the image it shows blue text on black, but when I tested it it seems fine
… I think we can just close it

w3c/epub-specs#3000

Ivan Herman: Who is responsible for what? That is the question here
… I put in the proposal to close it, but no one has agreed

<Hadrien Gardeur> +1 for closing

<Shinya Takami> +1

<sueneu2> +1 for closing

Ivan Herman: Ok, I will close it
… should we take up 2719 now or later?
… if later I will remove the epub spec tag

Matt Garrish: I am just writing some stuff up for it. It will be closed in about 10 minutes

Annotations

Wendy Reid: We have open issues and PRs
… is there anything we wanted to discuss
… ?

<Hadrien Gardeur> Laurent is currently in Malaysia, won't be back before July 10th

Annotation order - w3c/epub-specs#3030

Wendy Reid: Laurent opened an interesting issues
… this is about order of annotations

Wendy Reid: Why can't we have access to the spine?

Hadrien Gardeur: Usually to order, you need something to order on (percentage, page number, etc)
… also you may have lost access to the content but still have the annotations
… also could have services that deal only with the annotations

Ivan Herman: I believe it is good to have
… we could just make the set a list and sort it
… but we should probably add metadata to say what the order is
… I might want to use size, or creation date, etc
… we can add a property and some reasonable values for it

Hadrien Gardeur: So I don't think that is how we should do it
… We need to give the software the ability to order as it wants
… There should be data in the annotations to sort it how it sees fit
… so we need this information on a per annotation basis

Wendy Reid: It is an important feature
… I do think it is a new field like playorder

<Hadrien Gardeur> FYI this is how we'll handle this in OPDS: https://drafts.opds.io/opds-progression-1.0.html#examples

Wendy Reid: we need some way to communicate this

Brady Duga: Yes, if only we had some sort of a locator with ordering built into it (hahaha) (CFI)
… they are inherently sortable without the original source
… it's hard with a locator, it's so complex people don't implement
… if we stick ordering into locators, we need to use CFI
… the easier solution is playorder or whatever we call it
… the problem is that they have orders that can be determined, we are over-constrained
… knowing which annotation is which, the playorder could just be a hint, it might be wrong
… once you put the annotations into the book

George Kerscher: Lars reminded me he has implemented CFI as open source
… just advertising that here

Ivan Herman: I was going to say the same thing
… if it was added we would have to porperly spec it
… but I don't know where this idea of adding a property comes from
… if I import it is good information to know that it is ordered and how

Wendy Reid: We need some sort of information about the location, since just knowing the file isn't enough to know where the annotation is
… e.g. 3 annotations in chapter 3 we won't know the order
… we would need a more precise locator

Hadrien Gardeur: Some locators are sortable. There are several in PDF.
… in pre-paginated if you have the index and the total number you can sort
… that was category 1
… category 2 is roll publications, you calculate the height of a publication by the tile height and number of tiles, and determine from there
… in epub we calculate positions
… we calculalte the total number of positions, and we know the current position
… so we know the overall location
… cfi wouldn't help with roll puibs

Brady Duga: We did discuss this at some point, the conclusion was, sometimes you won't know the ordering of annotations, you may be able to guess
… you have to go through where in the HTML these are pointing to and sort
… that may not be done until the chapter is opened
… you'll have a rough sort by file referenced, people might be ok with that
… "lazy" evauliation of resources

Ivan Herman: I am now horrified. Because the order of html and human perception may not be the same
… They could be entirely different

<Hadrien Gardeur> +1 for what Ivan Herman just said, that's why it's calculated at rendering time in Readium

Ivan Herman: so I think we should keep away from it for now

Dale Rogers: From web design, I know the DOM has a purpose. Everything is in it's own order. But if we annotate something in the DOM, does the order of the document determine the location, or do we have a seperate positioning for the annotation?
… So CSS can change the apparent location of something
… it sounds like a complicated can of worms

Ivan Herman: Yes, things can be moved via CSS, so the perception order can very different

Hadrien Gardeur: The question is how deep do you want to go
… I have to discuss with reading systems who are using pseudo-CFI, or XPointer, etc
… So this is really just contextual information, but you can't use it to actually place the annotations
… if we want to go deeper we will be here for 5 years

Shinya Takami: We can push to RS
… in the list of annotations to sort by date is easy for RS
… If RS want to sort by location, since it is hard to maintain, should be up to RS implementations

Wendy Reid: I understand the technical challenge, but we have to consider the user need
… Even just to get pretty close
… The problem disappears with the book content

<Gregorio Pellegrino> For accessibility we have A reading order...

George Kerscher: We have the a11y requirement to maintain a reading order
… I realize CSS allows the order to change, but I hope we have something we can rely on
… I think on export the RS should export the reading order
… The RS already needs to maintain this

Dale Rogers: This seems to be on the RS side
… We have talked about how authors style things and the RS just tosses it
… so the order could just be the html reading order
… so maybe we just say it is reading order and leave it up to the RS to figure it out

Wendy Reid: We can't generate an annotation without the book source
… so we have the data at time of creation what the order is

Minutes manually created (not a transcript), formatted by scribe.perl version 244 (Thu Feb 27 01:23:09 2025 UTC).