Some browsers do not support <?xml version="1.1"?> tag

[kimtuck]

I have some books which contain xml tags specified like this:

<?xml version="1.1"?>

At the present, I'm getting around this by patching the code to replace "1.1" with "1.0". I think eventually we'll deal with this at the time we acquire the books from the publishers, either by asking them to package them with strictly "1.0", or by replacing the version number before we process the books and put them online.

So, I'm not sure whether you want to patch readium to make this change, or leave it as an issue for content providers. If you're leaving it to content providers, you may want to trap the error and log it to the console, so that users will better understand what is going on. I think that, without some sort of patch, the code simply fails without any diagnostics.

[da70]

We ran into this issue too, specifically with Firefox, which was not loading certain books.
We ran the EpubCheck validator on our collection, which returned this error for the books in question (file info snipped):
ERROR(HTM-001): ... : Any publication resource that is an XML-based media type must be a valid XML 1.0 document. XML version found: 1.1.

In our case the books were EPUB version 2. According to OPF 2.0 spec: "An XML Document is a complete and valid XML document as defined in XML 1.0 (Fourth Edition) (http://www.w3.org/TR/2006/REC-xml-20060816/)."

We didn't have the option of going back to the vendor to have the EPUBs corrected, so we did it ourselves. After clearing this error the books loaded.
I've asked the department that does the outsourcing for the books that going forward they require vendor delivered EPUBs to pass EpubCheck without errors before we declare them acceptable.

When I was debugging, I did see console errors that hinted at the problem. Shortly before all console messages stopped, Firefox reported "XML declaration not well-formed" for what I believe was the OPF file. I agree Readium could enhance this error message. In fact, it seemed to fail to process the exception. The final console error was "TypeError: e is undefined" in a file I can't identify because I am looking at a screenshot of the console output, but it is definitely a readium file. Note that we are using a Sep 2, 2015 version of master, which is far behind HEAD now.

[kimtuck]

What I ended up doing was patching the zipReader -- at the point where the file contents have been extracted from the zip file, replace any 1.1 declarations with 1.0 declarations. We will be asking our publishers to send us files with the declaration as 1.0; but a number of my test files were using 1.1 and this was the quickest way to get past this particular issue. If you're interested, I can post the code that I modified.

[da70]

Yes I would like to see the code, just to learn more about Readium. We don't actually need it, though, because we fixed the declarations in our EPUBs. Maybe this is an option for you as well? That way you won't need to deal with patches that would be overwritten or even possibly invalidated when updating Readium from upstream.
It turned out to be very easy. First unzip, then edit, then re-package. Here's what I wrote in our internal wiki about re-packaging:

---

Zip 3.0 by Info-ZIP is available on Mac OS X on the command-line.

Following instructions from Epub Format Construction Guide - HXA7241 - 2007:

    zip -X0 EpubGuide-hxa7241.epub mimetype
    zip -Xur9D EpubGuide-hxa7241.epub *

The flags being used (see http://www.info-zip.org/mans/zip.html):

• -9: maximum compression
  

• -D | --no-dir-entries: do not create entries in the zip archive for directories.
  

• -u | --update: replace (update) an existing entry in the zip archive if it had been modified more recently.
  

• -r | --recurse-paths: travel the directory structure recursively.
  

• -X | --no-extra: do not save extra file attributes.
  

---

We tested the correctness of this method in various ways:

• Round-trip zipping and unzipping an exploded EPUB from our website and checking that files/directories are binary identical.
  

• Opening the resulting EPUB file in Adobe Digital Editions and testing page turning, font size change, TOC, bookmarking, and full-screen mode.
  

• Validating EPUB file using [epubcheck](https://github.com/IDPF/epubcheck).  No errors were introduced by the zipping process.
  

So that's an option you might want to consider.
Please do feel free to post the code though, as I would be curious to see it.

[danielweck]

Thank you for this bug report. I am able to reproduce on Windows 10 with Edge, Internet Explorer, Firefox (only Chrome works with version="1.1" in package.opf).