BUG. RC2 still creates invalid epub file

i saw something about epub in the change log. But it doesn’t seemto catch everything.
Just tested with compiling your tutorial and got the following result:

+++
Validating using EPUB version 3.0.1 rules.

epubcheck completed

ERROR(OPF-027): tutorial.epub/OPS/contents.xhtml(27,60): Undefined property: ‘reader-start-page’.

WARNING(HTM-025): tutorial.epub/OPS/body1.xhtml(378,143): Non-registered URI scheme type found in href.

FATAL(RSC-016): tutorial.epub/OPS/body6.xhtml(48,83): Fatal Error while parsing file ‘Attribute “class” was already specified for element “li”.’.

ERROR(RSC-005): tutorial.epub/OPS/body6.xhtml(-1,-1): Error while parsing file ‘Attribute “class” was already specified for element “li”.’.

Check finished with errors
+++

Checked using Epub Validator at ebookit.com/tools/bp/Bo/eBo … -validator.

ERROR(OPF-027): ./books/Bo/databases/eBookIt/temp_uploads/1567217291.epub/OPS/contents.xhtml(63,60): Undefined property: ‘reader-start-page’.

WARNING(HTM-025): ./books/Bo/databases/eBookIt/temp_uploads/1567217291.epub/OPS/body12.xhtml(18,144): Non-registered URI scheme type found in href.

About these lines:

OPF-027:

  • Start of Content
  • (HTM-025):

    This sentence has a footnote attached.

    The first problem: In order to reference a “type,” the epub MUST have this type defined somewhere. In this case, it’s likely defined by Apple’s preamble code, and I have typed about this elsewhere.

    The second problem: this is new. Scrivcmt:// is NOT a registered scheme; that path may be valid only on my computer (1 in 36 to the 28th power or so that someone else with thousands of similarly named files has a filename just like it), and the default browser has NO CLUE where that file is.

    In fact, if I were to distribute this epub to someone else not using my computer, how would the epub grab it? When I click the hyperlink in the reader, it tries to open it in a web browser, which tries to open Scrivener.

    This one is only a warning. Still, it makes no sense to use “scrivcmt://” as a hyperlink in an epub.

    Is this also validating the tutorial or something else?

    I’ve used two different validators, One is validator.idpf.org, the other one is something of the garden variety. They gave both the same results.

    I can’t comment my finds, just show them hoping the devs understand what’s wrong.

    Pity is,four or five versions ago, Scrivener was able to produce valid epubs. And mobi wihtout hyroglyphs in foreign language words.

    I tried to reproduce your results, so I compiled the tutorial. The website validator I used (which checks against v3.0.1) uses epubcheck, but it changes the name (because it’s stored in a temp file with a number).

    Interestingly, for epub v3.2, the error is no longer reported by epubcheck. I’m guessing they changed the standard so undeclared references are simply ignored by the epub software (which I think was/is a default anyway; none of the software I have for reading them every complains about those errors).

    This is essentially the same as the “ibooks” error we were getting a bit back (which, by the way, was mentioned in this thread: [url]https://forum.literatureandlatte.com/t/epub-validation/43031/1]). It’s just a different tag that’s causing issues; the property needs to be defined before it can be used.

    Hre we go again.
    And worse than before. Compiling the tutorial with BETA 23, now I got SIX error messages.

    Then I tried compiling a project of my own. It’s invalid too.