WebBook Format

[Apparition B5]

It appears that there is a new open source eBook format being worked on, by the name of WebBook. Here's the announcement. The WebBook website itself can be found here.

From what I can gather, Daniel Glazman is a former Mozilla engineer and JavaScript programmer, and programmer of the BlueGriffon EPUB editor, that has become fed up with EPUB and wants to create a new, more open eBook format that more closely resembles a website in design.

Anyone know any more about it? Thoughts?

[j.p.s]

The name seems a bit scary, but it looks very promising. It sounds like it supposed to be epub done right. Making the spec a github project is very encouraging.

[Apparition B5]

The name references that the spec uses current web technologies and the eBooks can be natively read in a web browser. It won't be app-ified like EPUB 4.

Yeah, I liked what I read. I hope it goes somewhere, because it's pretty obvious that EPUB has serious issues between the killing off of EPUB 3.1 and the direction of EPUB 4.

[kovidgoyal]

It is basically EPUB 3.0.1 with the additional requirement that unzipping it into a folder would make it possible for a browser to read the book from the unzipped folder. However, that requirement is pointless since no one will actually read books like that -- there is no way to paginate and no way to search the entire book for a word or phrase when reading like that.

In addition, it basically punts on the question of metadata and has no provision for specifying a book cover.

Things I like about it -- it gets rid of useless bits of the EPUB spec like container.xml and the OPF file and epubcheck.

[DNSB]

Originally Posted by kovidgoyal

Things I like about it -- it gets rid of useless bits of the EPUB spec like container.xml and the OPF file and epubcheck.

Minor nit: epubcheck is not part of any epub spec. It does check that an epub follows the epub spec to the limits of it's programmers.

[kovidgoyal]

Originally Posted by DNSB

Minor nit: epubcheck is not part of any epub spec. It does check that an epub follows the epub spec to the limits of it's programmers.

True but the epub spec being such a ghastly mess is what motivated epubcheck in the first place. The vast majority of checking it does (XML schema checks) have been obviated by the move to HTML 5 where parsing is made robust instead of requiring HTML to be perfect. In general, the modern trend is to make parsers robust rather than serialization formats overly nit-picky.

[Sarmat89]

Looks like a nightmare to me. So instead of a clear document structure, we are supposed to fish the hyperlinks from [tt]index.html[/tt], or, alternatively, parsing each file to access [tt]rel[/tt]s.
Also, no trace of semantic metadata required for any modern book.
It is hard to invent a format crappier than EPUB, but those guys has made it.

[Sarmat89]

Originally Posted by kovidgoyal

The vast majority of checking it does (XML schema checks) have been obviated by the move to HTML 5 where parsing is made robust

How is HTML5 "more robust", and how does it remove the necessity of schema syntax checking?

[kovidgoyal]

Originally Posted by Sarmat89

How is HTML5 "more robust", and how does it remove the necessity of schema syntax checking?

In HTML 5 the parsing algorithm is fully specified and robust against bad markup, i.e. all spec compliant parser produce the same output for the same bad markup. So it doesn't matter if there are small mistakes in the markup, since they result in the same DOM in all user agents.

This argument about strict syntax vs. robust parsing has been had endlessly through all the ages that computers have existed. In the web world, robust parsing has won, the EPUB standards committee just didn't get the memo.

[DNSB]

Originally Posted by kovidgoyal

In HTML 5 the parsing algorithm is fully specified and robust against bad markup, i.e. all spec compliant parser produce the same output for the same bad markup. So it doesn't matter if there are small mistakes in the markup, since they result in the same DOM in all user agents.

This argument about strict syntax vs. robust parsing has been had endlessly through all the ages that computers have existed. In the web world, robust parsing has won, the EPUB standards committee just didn't get the memo.

The same parsing algorithm that has people concerned about it's ability to parse in linear time? The adoption agency algorithm is a horrible example of good intentions with it's nested loops that can go quadratic. Going by some of your posted code, you are aware of this issue.

[kovidgoyal]

Originally Posted by DNSB

The same parsing algorithm that has people concerned about it's ability to parse in linear time? The adoption agency algorithm is a horrible example of good intentions with it's nested loops that can go quadratic. Going by some of your posted code, you are aware of this issue.

The tradeoff of increased parsing time for greater robustness is both perfectly natural and perfectly justified. And I fixed my issues with HTML 5 parsing speed by writing https://github.com/kovidgoyal/html5-parser

[gravity_]

That's great news! Seems legit. Hopefully, it will work better than some others already existing out there.

[sherman]

What struck me as odd is using a zip file as a container, and explicitly denying the use of compression. Why? Text is highly compressible after all.

[Apparition B5]

Originally Posted by sherman

What struck me as odd is using a zip file as a container, and explicitly denying the use of compression. Why? Text is highly compressible after all.

IIRC, the WebBook spec doesn't explicitly deny compression. It discourages compression and states that if you do compress you can only use the deflate algorithm. In other words, you can compress but you can't use the bzip2, LZMA, or XZ algorithms.

[sherman]

Originally Posted by Celestial

IIRC, the WebBook spec doesn't explicitly deny compression. It discourages compression and states that if you do compress you can only use the deflate algorithm. In other words, you can compress but you can't use the bzip2, LZMA, or XZ algorithms.

Gahh... I need to check my reading comprehension Granted, the way it is worded is not the greatest. I think a single comma could have made it a lot clearer.

Upon rereading, you are, of course, correct. My bad.