What exactly are Standard Ebook's "advanced epubs"?

[ownedbycats]

For those not already know, Standard Ebooks is a site that takes public domain works from sources like Project Gutenburg and makes them nicely formatted. Check them out if you're into that kind of thing.

With that out of the way:

All their books have these options (and their descriptions):

Compatible epub - All devices and apps except Kindles and Kobos.

azw3 - Kindle devices and apps. Also download the Kindle cover thumbnail to see the cover in your Kindle’s library.

kepub - Kobo devices and apps.

Advanced epub - An advanced format not yet fully compatible with most ereaders.

Does anybody know exactly what the "advanced epub" is? this page mentions that a Chrome extension called Readium uses them, but the Readium page doesn't really give any indication of what it offers other than viewing ePubs in Chrome.

[maddz]

I suspect it may be ePub3 rather than the more usual ePub2.

My experience with ePub3 is that the file size is at least 10 times bigger, and it has more in common with a PDF than an ebook. (This was after purchasing the second part of book 2 of The Barrow by Mark Smylie on Drivethru.) After kicking up a fuss, Mark re-exported as ePub2 for those of us with eInk readers. It was fine reading it on my iPad, but a right pain on the Kobo.

[DNSB]

Originally Posted by maddz

I suspect it may be ePub3 rather than the more usual ePub2.

My experience with ePub3 is that the file size is at least 10 times bigger, and it has more in common with a PDF than an ebook. (This was after purchasing the second part of book 2 of The Barrow by Mark Smylie on Drivethru.) After kicking up a fuss, Mark re-exported as ePub2 for those of us with eInk readers. It was fine reading it on my iPad, but a right pain on the Kobo.

Very likely a fixed layout epub3 which is, as you said, more like a PDF than a reflowable ebook format.

[ownedbycats]

I think Calibre's ebook-viewer also has trouble with fixed-layout ePubs, though I've not experienced this for myself.

It's a shame there isn't already a standardized ISO format out there that handles fixed layouts consistently. Maybe someone will develop one someday.

[DNSB]

Please note the FLO is not what most epub3 ebooks are produced in. Other than specialty books such as illustrated children's books, FLO is not all that useful.

In the case of Standard ebooks, the couple of "advanced" epubs I downloaded were reflowable epub3 ebooks with attention paid to accessibility information.

As for an ISO standard? We already have PDF and FLO epub3.

[PeterT]

I know there have been other threads discussing issues with their content. I'll search later on

[ownedbycats]

I saw a mention somewhere about weird bugginess with their kepubs.

Originally Posted by DNSB

As for an ISO standard? We already have PDF and FLO epub3.

I was actually referring to the PDF format. It passed into ISO standard some years ago so no royalties needed.

[JSWolf]

Originally Posted by ownedbycats

For those not already know, Standard Ebooks is a site that takes public domain works from sources like Project Gutenburg and makes them nicely formatted. Check them out if you're into that kind of thing.

With that out of the way:

Does anybody know exactly what the "advanced epub" is?

Advanced ePub is rubbish. It's not needed at all. It's just going to confuse a lot of people. What Advanced ePub is is ePub3 with no backward comparability. There's no need for it since the content is the same as the Comparable ePub. There is nothing advanced about it.

As for well formatted. Not at all.They have huge paragraph spaces and the section breaks are a full line going across the entire page. Then we have the text left justified. And the cover is not programmed to be displayed. So the first page is a titlepage. They look awful because they are awful. They really need to fix the formatting.

OK, I figured out why the formatting is so bad. It's because they have error(s) in the CSS and in that case. ADE ignores the CSS. This is why there formatting is lousy. They need to make the CSS actually valid. RMSDK (ADE) is the most used for displaying ePub and if they are going to botch the CSS, they may as well shut down until they learn CSS.

[JSWolf]

Originally Posted by ownedbycats

I saw a mention somewhere about weird bugginess with their kepubs.

That could be because their CSS has error(s). Garbage in. Garbage out.

Given that these eBooks do not have complex formatting, the CSS is hugely over complicated and that is what causes the problems. I could toss out the entire CSS and create a new CSS that's a lot neater and more simple. Even the OPF is a mess. It's no wonder the KePub has problems.

[DNSB]

To disagree with Jon, at one time I downloaded half a dozen ebooks in all four formats (3 epub and 1 azw3). John W. Campbell and H. Beam Piper from the Science Fiction section.

Contrary to Jon's statement, none of the stylesheets has issues that would have ADE disregarding the entire style though they did have issues with epub types being used ( a[epub|type~="noteref"] as an example). The W3C validator wasn't happy with those entries or the adobe proprietary items but nothing that was a show stopper. When tested with ADE, those entries were treated as no ops which is the way they should be handled. The only difference in the kepub variant was the addition of the kobo spans. Erratically the logo and titlepage image files would be missing from one of the formats.

All three epub formats also had an onix.xml file which was correctly structured and compliant though I haven't seen any programs that make use of the ONIX metadata standard.

The stylesheets were not what I would consider elegant but they did work and were not overly complex. The three stylesheets clocked in at 130 lines, 20 lines and 105 lines.

[robin_reala]

There are some misconceptions in this thread. Hopefully I can clear them up.

First off, SE’s “Advanced epub” is just the default codebase we work on, zipped up. It’s a modern epub3, with no attempt to be bug-compatible with readers beyond meeting the standard spec. When work is finished on the production, it’s zipped up and made available for download.

The “Compatible epub” is the advanced epub, but with some extra processing to make it work better in real life ereaders. The main thing that’s added is a compatibility.css file that has style fixes for common platforms (for example to work around bugs with hanging punctuation in Apple Books). We also render all SVG lineart and MathML formulae to PNGs to get past common missing functionality.

The Kobo kepub has Kobo spans and the .kepub extension applied to trigger the native Kobo renderer. We also apply some of the compatibility fixes, but notably not the MathML prerendering as Kobo’s native MathML renderer is capable.

Finally, the Kindle version is the Kindle version. Not much we can do there.

JSWolf: obviously I don’t want any rendering problems in our epubs. Just out of interest I ran our stylesheets through the CSS Validator to see if I could replicate what you’re seeing, and it did flag some errors. On inspection though, the files themselves don’t contain the problems reported, so it’s a bug with the validator(!)? In future, if you find any more issues and have time, please feel free to PM me, or file them directly at https://github.com/standardebooks/tools/issues/new . Thanks!

[DNSB]

For an example of one issue, in the kepub version of John W. Campbell's The Black Star Passes, there are two references to logo.png and one to cover.png in the html but neither image is included in the kepub.epub container. They are in the compatible epub version (the advanced version uses .svg images).

[robin_reala]

I’ve just built and extracted the kepub for that book and logo.png is in there and referenced correctly. cover.png isn’t, but cover.jpg is and is referenced from the manifest with a cover-image properties.

Potentially you’ve got an older version? We do find and fix bugs, and the corpus is republished when that happens.

[JSWolf]

Originally Posted by DNSB

To disagree with Jon, at one time I downloaded half a dozen ebooks in all four formats (3 epub and 1 azw3). John W. Campbell and H. Beam Piper from the Science Fiction section.

Contrary to Jon's statement, none of the stylesheets has issues that would have ADE disregarding the entire style though they did have issues with epub types being used ( a[epub|type~="noteref"] as an example). The W3C validator wasn't happy with those entries or the adobe proprietary items but nothing that was a show stopper. When tested with ADE, those entries were treated as no ops which is the way they should be handled. The only difference in the kepub variant was the addition of the kobo spans. Erratically the logo and titlepage image files would be missing from one of the formats.

All three epub formats also had an onix.xml file which was correctly structured and compliant though I haven't seen any programs that make use of the ONIX metadata standard.

The stylesheets were not what I would consider elegant but they did work and were not overly complex. The three stylesheets clocked in at 130 lines, 20 lines and 105 lines.

They have changed things since the last time looked at the site before this thread was started. Have a look at the style sheet for McTeague by Frank Norris. The compatible ePub isn't compatible. The reason they are saying it's not compatible with Kobo is because the CSS has errors and isn't compatible with RMSDK.

[JSWolf]

Originally Posted by robin_reala

There are some misconceptions in this thread. Hopefully I can clear them up.

First off, SE’s “Advanced epub” is just the default codebase we work on, zipped up. It’s a modern epub3, with no attempt to be bug-compatible with readers beyond meeting the standard spec. When work is finished on the production, it’s zipped up and made available for download.

The “Compatible epub” is the advanced epub, but with some extra processing to make it work better in real life ereaders. The main thing that’s added is a compatibility.css file that has style fixes for common platforms (for example to work around bugs with hanging punctuation in Apple Books). We also render all SVG lineart and MathML formulae to PNGs to get past common missing functionality.

The Kobo kepub has Kobo spans and the .kepub extension applied to trigger the native Kobo renderer. We also apply some of the compatibility fixes, but notably not the MathML prerendering as Kobo’s native MathML renderer is capable.

Finally, the Kindle version is the Kindle version. Not much we can do there.

JSWolf: obviously I don’t want any rendering problems in our epubs. Just out of interest I ran our stylesheets through the CSS Validator to see if I could replicate what you’re seeing, and it did flag some errors. On inspection though, the files themselves don’t contain the problems reported, so it’s a bug with the validator(!)? In future, if you find any more issues and have time, please feel free to PM me, or file them directly at https://github.com/standardebooks/tools/issues/new . Thanks!

Install ADE 2.0.1 and use it to view McTeague by Frank Norris and you'll see that the entire CSS is ignored. The CSS is way to complicated for what it needs to be. The eBook does not have complex formatting. So the CSS should be rather simple. But it's not. It's a mess. If you like, I can fix it up and show you how simple the CSS really should be. Also, the OPF is rather a mess (IMHO). If you want the maximum comparability, the CSS and OPF need to be edited. The advanced ePub can be an ePub 3 version with whatever you feel is needed. But the compatible ePub should be ePub 2.