Reading progression is stuck in an infinite loop

[jccr]

This issue is a Bug

Expected Behaviour

To be able to flip pages in the reader as normal

Observed behaviour

Reading progression is stuck in an infinite loop

Steps to reproduce

1. Load the specified EPUB file with the instructed modifications into the reader
2. Flip through the pages
3. Notice that it only navigates through the first two spine items repeatedly

Test file(s)

1. Download the Accessible EPUB 3 sample from here: http://idpf.github.io/epub3-samples/samples.html
2. Uncompress it and modify the OPF file's spine list to this:

<spine>
        <itemref idref="cover" linear="no"/>
        <itemref idref="spi_ad"/>
        <itemref idref="id-id2442754"/>
        <itemref idref="spi_ad"/>
        <itemref idref="htmltoc" linear="yes"/>
        <itemref idref="id-id2632344"/>
        <itemref idref="id-id2604743"/>
        <itemref idref="id-id2629773"/>
        <itemref idref="id-id2620395"/>
        <itemref idref="id-id2638681"/>
        <itemref idref="id-id2611884"/>
        <itemref idref="id-id2627310"/>
        <itemref idref="id-id2635343"/>
        <itemref idref="id-id2622654"/>
        <itemref idref="id-id2624850"/>
        <itemref idref="id-id2640702"/>
        <itemref idref="id-id2641220"/>
        <itemref idref="id-id2642385"/>
        <itemref idref="id-id2644001"/>
        <itemref idref="id-id2644238"/>
        <itemref idref="id-id2645594"/>
        <itemref idref="id-id2645682"/>
        <itemref idref="id-id2645862"/>
    </spine>

Product

• Readium Chrome extension
  • latest official version available from the Chrome Web Store
  • Version of the Chrome browser: 53
  • OS and version: OSX El Capitan

Additional information

The OPF spec specifies that: "a specific spine item (from the perspective of its id attribute value in manifest) must not appear more than once in spine."
iBooks doesn't have this issue, it skips the duplicate spine item in the spine

[jccr]

What can we do about this? Do the same as iBooks?

[rkwright]

@jccr Does the modified EPUB pass epubcheck? It should be flagged as an error.

[jccr]

I believe it would cause an error in epubcheck.

On Wed, Sep 21, 2016, 8:19 AM Ric Wright notifications@github.com wrote:

@jccr https://github.com/JCCR Does the modified EPUB pass epubcheck? It
should be flagged as an error.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#568 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AE5RbFwg3aTEmvCIVcKA71opB-FYi_LDks5qsUsVgaJpZM4KCSUj
.

[rkwright]

@jccr can you copy your modified file to here:
https://drive.google.com/open?id=0BzaNaBNAB6Fja2E3RkZPZVZ5Mnc

[jccr]

@rkwright Done

[rkwright]

@jccr the file has many problems. Not just a duplicated spine item
epubcheck_errors.txt
However, when I get the "real" accessible epub file and make the change to the spine there is no real problem - except of course, we do show the duplicated spine item twice. I don't think we need to do anything right now. If we build Readium-2 we can fix it then if it seems worth it.

BTW, epubcheck doesn't report any errors from the duplicated item. I'll report that to the epubcheck folks.

[jccr]

@rkwright
Thanks! Yea I accidentally sent you a modified copy of accessible epub that I had on my drive dated from 2012. I should have modified the latest version..

[jccr]

@rkwright
Just to be thorough about this:

If you modify the spine like so:

<spine>
        <itemref idref="cover" linear="no"/>
        <itemref idref="spi_ad"/>
        <itemref idref="spi_ad"/>
        <itemref idref="id-id2442754"/>
        <itemref idref="htmltoc" linear="yes"/>

You will see the spine item twice when paging through.

But if you have any other spine item in between the original and the duplicate you then get the infinite reading progression loop I was talking about:

<spine>
        <itemref idref="cover" linear="no"/>
        <itemref idref="spi_ad"/>
        <itemref idref="id-id2442754"/>
        <itemref idref="spi_ad"/>
        <itemref idref="htmltoc" linear="yes"/>

[danielweck]

thanks for the clarification Juan, that's indeed a "bug".