XHTML annotation with MathML descendant content valid or invalid?

[mattgarrish]

Another oddity in the valid tests is the one for MathML inside an XHTML annotation.

Inside the test we still have this comment:

<x:p>This annotation cannot contain math (sch test)</x:p>

as this was an error until last year. It is still flagged as an error by validator.nu.

The MathML spec isn't too specific about the element's use:

Alternate representation annotations are most often used to provide renderings for an expression, or to provide an equivalent representation in another markup language.

While XHTML is another markup language, using presentation mathml again inside it seems to defeat the purpose of it being an alternative to presentation mathml.

I'm wondering if we should at least open this against the validator schemas and see if we can harmonize which interpretation is the right one?

[murata2makoto]

Are you saying that this should be an issue against EPUB 3 rather than epubcheck? I would agree.

[mattgarrish]

Are you saying that this should be an issue against EPUB 3 rather than epubcheck?

Might come down to that.

It looks like an addition to the html schemas. The schema in the MathML spec doesn't restrict the content model for annotations, but the one in the validator github repo adds restricted cases for xhtml, svg and mathml annotations.

Seems like a common-sense addition and we're supposed to follow html validation unless we have a specific override.

[rdeltour]

@mattgarrish any new thoughts on that? I'm trying to see if this can be tackled and resolved in v5.0.0.

[rdeltour]

I'm moving this out of the v5.0.0 milestone, as it looks like the specs are not definitive about this, and it is low priority. Thoughts still welcome.