Having had my say above, in playing with epubs over the years, I have seen some things that demonstrate how bad the Adobe RMSDK page count algorithm is. The attached book shows this.
It has five chapters. Each chapter has a title followed by fifty paragraphs. Each paragraph has almost exactly the same number of letters in them (the paragraphs are numbered so the first nine have one less letter). The first chapter uses the "Lorem ipsum" text for each paragraph. The others use a copy of this that has been passed through the ScrambleEbook plugin in calibre. Each paragraph has a class and these are all the same lengths. This means that, apart from the chapter title, each chapter has exactly the same number of characters in it.
How to use this:
- Copy the book directly to the device. Do not add this to calibre and then send it to the device.
- Before opening this book on your device, turn on the Adobe Numbers and make sure that both the chapter and book page numbers are shown.
- Open it and page through it.
- Look at the total pages for each chapter.
- Watch the page numbers as you go through it.
Chapter 4 didn't work as well as I hoped. But, the rest demonstrate the problems inherent in the Adobe RMSDK page numbering algorithm. It's easy to game it. But, it's probably the simplest and most universal method we have for comparing the number of "pages" in ebooks. But, it is
not as standard.
If you put another copy directly on the device and change the name to end in ".kepub.epub", you can see how the kepub renderer treats it. Don't worry about the lack of spans as you don't really need them except to remember locations.