Some problems with (X)HTML 5 specification

[rdeltour]

From [email protected] on May 30, 2013 13:15:26

VERSION:
3.0

STEPS TO REPRODUCE:

1. check with epubcheck the attached file (it's the printable version of the "autism" page of wikipedia, converted to epub with Calibre)
2. read the errors ^^

RESULT: epubcheck reports the errors:
a. attribute "class" not allowed here [in "html" tags]
b. element "script" missing required attribute "type"
c. value of attribute "dir" is invalid ["auto"]
d. attribute "srcset" not allowed here [in "img" tags]
e. attribute "role" not allowed here [in "div" tags]

EXPECTED: None of these errors should be displayed. In (X)HTML 5:
a. "class" attribute is allowed in any tag
b. attribute "type" is no longer required for "script" tags
c. attribute "dir" can have value "auto"
d. "srcset" is a proposed html extension for images: http://www.w3.org/html/wg/drafts/srcset/w3c-srcset/ e. "role" attribute seems to be reintroduced in HTML 5: http://www.w3.org/TR/role-attribute/

Attachment: autism.epub

Original issue: http://code.google.com/p/epubcheck/issues/detail?id=274

[rdeltour]

From [email protected] on June 24, 2013 19:16:09

You generated an EPUB 2 containing XHTML 1.1 content using Calibre, so all the errors are valid.

You need to convert the content to EPUB3/XHTML5 if you want it to validate as such.

[rdeltour]

From [email protected] on June 25, 2013 00:13:42

Right. In this case epubcheck, I think, should alert about the inconsistency between the epub version and the doctype of XHTML files, since it's preferable that you'll convert the epub 2 to an epub 3 instead of convert all xhtml 5 documents to a previous xhtml specification.

In theory this is another issue, you can close this bug and open another one if you want :)

[rdeltour]

From [email protected] on June 25, 2013 05:19:58

But no doctype was specified in these files that I found, so epubcheck is doing exactly what it should for an epub declared as 2.0 in the OPF.

If the html5 doctype had been specified, then I agree it would be useful to at least report an unexpected doctype.

[rdeltour]

From [email protected] on July 07, 2013 04:23:30

Well, the fact the doctype is removed is a calibre bug. Anyway I think epubcheck should alert you the doctype is missing, since html 5 specs requires it: http://www.whatwg.org/specs/web-apps/current-work/multipage/syntax.html#the-doctype

[mattgarrish]

The doctype requirement only applies to the HTML serialization in HTML5:

XML documents may contain a DOCTYPE if desired, but this is not required to conform to this specification. This specification does not define a public or system identifier, nor provide a format DTD.
http://www.w3.org/TR/html5/the-xhtml-syntax.html#writing-xhtml-documents

And technically there is a set of obsolete DTDs that can be used with HTML5 as defined in that doctype section, one of which is the XHTML 1.1 doctype epub 2 uses.

XHTML 1.1, on the other hand, does require a doctype:

There MUST be a DOCTYPE declaration in the document prior to the root element.
http://www.w3.org/TR/xhtml11/conformance.html#s_conform

But I don't think it's invalid to use the HTML5 doctype in an XHTML 1.1 file from what I see. The XHTML 1.1 specification doesn't require any specific doctype, only that one be present.

So, long story short, epubcheck should have emitted a warning that no doctype was present for the epub 2 content documents, but had doctypes been inserted I still think it would have resulted in confusion since the doctype isn't used for validation.

[rdeltour]

Closing as invalid. The part about EPUBCheck checking the XHTML 1.1 doctype declaration can be tracked in #982.