[HTML Survey] Adopt some way to signify that an EPUB uses HTML instead of XHTML

[reidmore-online]

Section

https://www.w3.org/TR/epub-34/#sec-package-elem

Describe the problem

Several responses to the survey have made reference to or requested a way to differentiate EPUB files with HTML from those using XHTML. Different methods have been proposed, including:

• Separate version number (i.e. EPUB4)
• Separate file extension (i.e. .epub4)
• Metadata to identify use of HTML

Describe the fix or new feature you propose

(Chair hat off) I understand the desire to differentiate the books in some way, but I don't think this change rises to the level of a complete version number change like EPUB4, without us bundling in many more features and changes that would make it worth such an advancement.

I do think having some way to easily identify these books is compelling, and we should discuss different ways to do it. Some options off the top of my head:

• <package version="3.4"...>
• <meta property="rendition:serialization"...>html</meta>

[mattgarrish]

For reference, we had some discussion of this previously on the mailing list that @iherman has documented here: #2715 (comment)

If we restrict the version change to only html epubs (3.0 = xhtml, 3.4 = html) it would actually work as a deterrent for older reading systems that can't handle html from breaking on the new content.

Otherwise, we're back in the 3.1 boat where you can't be sure a reading system will open a higher version epub, even if it's xhtml it could display.

[bduga]

Do we need another mechanism for this? We already have mime type for individual spine items, we could just require that they actually match the content document, enforced via epubcheck. And that has the added benefit that older reading systems may complain when given a package that references HTML documents, and looking for the presence of HTML in the spine is just as easy as looking for an attribute. That is, it is does not require inspecting content documents.

[iherman]

This was discussed during the pmwg meeting on 10 July 2025.

View the transcript

Signify that an EPUB uses HTML - w3c/epub-specs#2761

wendyreid: about adoption on HTML there were already suggestions. Major version EPUB4, change the extension, check the OPF ...

gpellegrino: I'm in favor of a different version number in the OPF. It is the only way for RS to announce something to the end user. If not, the RS will not announce anything before breaking.

duga: I hear about old RS, what they do or not. But how big is their market share? Who are they? What would they do in these scenarios (media-types are usually wrong)?
… before we decide, we have to understand. A list of RS, and some way to test.

CharlesL: perfect example where we can create test files with different options and check RSs.

DaleRogers: as a web designer and teacher, I said to may students open a tag, close a tag. RS are smart enough to forgive unclosed tags. Most browsers are smart enough to ajust. I'm wondering if EPUB RS can be smart enough.

https://www.mobileread.com/forums/showthread.php?p=4523848#post4523848

wendyreid: I'm also curious about RS reactions on HTML.
… Suzan opened a thread on Mobileread.
… if we use the EPUB version in the OPF, EPUB creator will use a static EPUB 3.4 even if they don't use HTML.

Hadrien: a different media-type is a solution, the EPUB version in the OPF is not properly processed by RS.
… with all the things we are discussing, a version number may be needed, but usually we register a media-type and extension.

duga: non well-formed documents are not possible in XML.

ivan: HTML is know to be less strict.
… I'm worried that we are running ahead. I hope we'll see answers in the survey. Will today's RS are easily transformable into HTML-processors? If none of the big RS around (Readium, Apple, Kobo, Amazon ...) all way we won't do that until we are forced, and publishers say we won't do that because RS don't process that, then we'll be stuck.

LaurentLM: We need to talk about distributors too
… they can block this if they say no
… if its not distributed, you won't find people to make it

LaurentLM: Some distributors are retailers, but many are intermediaries between publishers and retailers
… some transform EPUB, like Amazon, but others don't, or only make partial changes, they use other checking tools like EPUBCheck or others

Hadrien: distributors mostly rely on EPUBCheck and ACE. They don't check content too much, but they process publications to generate samples, watermark content, inject marketing material ... Most processes rely on XML.

DaleRogers: if we create a version of the spec that distributors and RS do not handle, are we advancing the medium.?

gpellegrino: Apple has its own validator. Do the chair have plans to reach such big players?

---