EPUBCheck bug?

[Ric Wright]

We have encountered two files where our RS had some trouble rendering the first page.  Turned out that the files declared themselves pre-paginated but the first page (nav in both cases) did NOT declare the size of the HTML page size via the meta tag.  Subsequent pages render correctly as they DO declare the size.  Navigating back to the TOC renders correctly as we use the existing viewport info.

I believe this is a violation of section 2.5.4 of the spec.  Two questions:

• Shouldn’t EPUBCheck flag the missing size as an error?
• What should the RS do in this case?  There is no easy way for us to know what the size should be as it is not declared in the OPF.  I suppose if we find it missing we could the sniff the next page to find out what size it is, but that’s starting to get convoluted.

The attached file is one of the two causing the problem.  It also happens to declare 

<meta property="rendition:flow">scroll</meta>

But I don’t think that is relevant.

Thanks

Ric

[Daniel Weck]

Note that there was a similar issue with the Media Overlays fixed-layout test, so we fixed the content (missing head>viewport):

https://github.com/mgylling/epub-testsuite/commit/e786dfb27931904c72c683299b1db41e155b1bb9

Also note that EPUB 3.0.1 introduces a new rendition:viewport OPF metadata property, which may help Reading Systems to prepare the page layout:

https://epub-revision.googlecode.com/svn/trunk/build/301/spec/epub-publications.html#fxl-property-viewport

Regards, Daniel

[Daniel Weck]

On Tuesday, March 18, 2014 12:43:56 PM UTC, Ric Wright wrote:

<meta property="rendition:flow">scroll</meta>

This is not a legal EPUB 3.0.1 value:

http://www.idpf.org/epub/301/spec/epub-publications.html#flow

[Matt Garrish]

And despite the “3.0.1” version of epubcheck that doesn’t mean it supports EPUB “3.0.1” yet. Unfortunate coincidence.

 

I would expect these kinds of tests to be in the next version of epubcheck now that fxl is a core part of the specifications.

 

Not sure how to answer the what to do when not present question. I wouldn’t go sniffing at other pages, since there’s no guarantee of accuracy. I’d either set some default rendering size in the absence, or be really cruel and obvious and default back to reflowable so the author gets the idea that they’ve done something wrong. But I’m sadistic at heart...

 

Matt

--
You received this message because you are subscribed to the Google Groups "epub-testsuite-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to epub-testsuite-di...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Ric Wright]

Matt,

Understood.  I guess I should have said “missing/forthcoming feature”…  :-)

It IS a big messy area, handling malformed or incomplete documents.  They are bound to exist and users and publishers simply don’t/won’t accept that the RS doesn’t intuitively handle such problems.  I don’t like sniffing either.  And going to reflowable is messy too.  Another option is to simply skip the page and go to the next, but that too is not  too nice.  Sigh.

Ric

[Matt Garrish]

Right, and hopefully once epubcheck is updated it will mitigate the problem as the epubs won’t register as valid, although I realize you’re still saddled with having to account for people who don’t validate or distribute through channels that require it.

 

I agree there’s no one perfect answer, but I prefer to be obvious to the author that they’re doing something very wrong. If you give them a default canvas it makes the bug more subtle. A blank page might be appropriate, since technically by not setting the dimensions you can only assume 0 for them.

 

Matt

[Kathy Alley]

Thank you, I’ll correct in my file. 

Kathy