[Tex2002ans]

Originally Posted by KevinH

There is for epub3 metadata.

See the "source-of: pagination" metadata properties in the epub3 docs.[...]

... doubt Amazon's ingestion takes that part of the EPUB3 specs into account, but great to know.

I was recently catching up to a few DAISY Youtube videos, and they were raving about how important RPNs are. I'll have to see if I can find the exact talks/location + what metadata they recommended.

Side Note: They actually had a fantastic 3-part webinar on tackling proper alt text: "The Art and Science of Describing Images".

Originally Posted by jhowell

In my testing the conversation to KFX (Enhanced Typesetting) will reject all of the page number information if even a single one is wrong. Only Arabic and Roman numerals are allowed. For example “5”, not “Page 5”. Targets styled as display:none also won’t work.

Fantastic information to know.

I was wondering a lot of that since the 2019: "Fractional Page Numbering" thread.

I believe the EPUB specs themselves don't specify anything, but in Amazon's docs, they currently insist on:

• Roman Numbers
  
  • i, ii, iii
• Arabic Numbers
  
  • 1, 2, 3

And I'm still scratching my head on newspapers/technical-documentation and stuff like that (where you may have Page A-1).

I've even seen some Physics/Maths papers where they use Greek letters.

Side Note: And keeping up on the bleeding edge of LibreOffice stuff, there's even these types of numbering:

• Russian
• Serbian
• Old Hungarian
• Hebrew
• [...]

There's dozens of numbering schemes used around the world.

Probably why the EPUB specs themselves don't mandate anything.

Originally Posted by Oxford-eBooks

Adding the RPNs (pageList) entries to the NCX is fine, we have our own tool that's been putting together ePubs for about 13 years now and it'll take a spreadsheet of page refs either hand or machne generated. I even wrote a little plugin for InDesign a while back that finds and tags the first word on each page in the XHTML for that purpose.

I'd be interested in seeing some of that stuff.

Or how you're tackling the generation of RPNs.

I've written about various methods over the years, but haven't found anything solid.

Luckily, I haven't had to do any of this RPN stuff... but listening to DAISY talks, they tend to rank RPNs extremely high on the Accessibility scale.

Would be good to get faster and more solid RPN workflows for the future.

Side Note: The biggest pain point for me is still marking the exact locations. And dealing with edge cases like:

• Frontmatter
  
  • copyright page may be shifted to end of EPUB.
    
    • So i, ii, iv, 1...200, iii
• Endnotes/Footnotes
  
  • In ebooks, these things get shifted around.
    
    • Endnotes -> footnotes at bottom of HTML file.
  • Chapter 1's text ends on page 100, but the endnotes begin on page 600. RPNs will go:
    
    • 100...600...101?
  • What happens when Chapter 2 ends?
    
    • 149...600...150?
• Multi-page footnotes that carry on to other pages
  
  • Does the footnote exist "only where the little supescript # appears"? Or would it be "all pages where the footnote text exists"?
    
  • If I recall correctly, I believe DAISY recommends where the # appears... But I've seen a 7-page beast of a footnote!
• Newspaper-like articles split into chunks across many pages.
  
  • Article 1 is on Page 1, 3, 7
    
  • Article 2 is on Page 1, 2, 5
    
  • Article 3 is on Page 1, 4, 5, 7

(And if I'm remembering the specs correctly, you can't have "duplicate" RPNs...)