Online Ebookmaker should validate generated files, NOT submitted files #4
Description
It is not true that either HTML5 validation or XHTML validation is sufficient for acceptance by PG. HTML5 validation is a best practice. What we require in practice is that files produced by Ebookmaker validate (i.e. EPUB2, EPUB3, and HTML5.) That's what the Ebookmaker validation is for. It runs the nu validator on generated HTML5 files, and epubcheck on EPUB2 and EPUB3 files. The HTML inside EPUB is XHTML.
Ebookmaker itself doesn't care whether the submitted files validate or not. Having them validate is good policy, but it's a SHOULD , not a MUST.
the XHTML validator uses DTD validation which is not able to verify compliance with requirements such as having the right number of cells in a table. the nu validator is based on javascript and does that validation. Unfortunately it has bugs not present in the older validators; HTML5 is a living standard and files that validate today may not validate in the future.
Because Ebookmaker has to create files that work in both EPUB and HTML5, Ebookmaker will reject some valid HTML5 and CSS3 code. This is a complex situation which can be difficult to explain, so it's best to emphasize for users of Online Ebookmaker that the important thing is to ensure that Ebookmaker generates valid files.
The code that purports to test whether a submitted file is HTML5 or XHTML is flawed and buggy, but is also unneeded and should be removed.