Apple have released an ePub guideline - Recommendations - they call it. Thankfully just one and a half pages on preparing ePubs for the iPad. Mostly it addresses their particular set of device limitations.
The iPad is a game changer, insofar as the rendering and presentation quality is excellent. The paper page metaphor is too strong for my taste but for a paged reader, the CSS implementation far ahead of ADE. This means their is no need to create crappy CSS for a particular device; and the XHTML and CSS is more future ready.
We have a standard test-book which has to be stylistically dumbed down for the quirky display properties and bugs in ADE and those devices that inherit ADE. Things are move lively on iPad. The sample screen grab here shows images sized by CSS and floated left.The recommendations are mostly common sense and straight-forward.There are a number of screen-grabs of our Trade test book included showing the presentation of various structures.
Adobe Page Template
This is a very good recommendation. The Adobe Page Template XPGT sucks and if any part of XSL:FO gets into e-pub officially it is time to move on to a new format. This took me back to the anti-CSS rant in the orginal ADE "Best Practices" guide:
"CSS specification is evolving beyond CSS2.1 in the shape of multiple CSS3 modules. Several of CSS3 modules aim to the solve the same problems as page template: adding headers and footers, multi-column layout and dynamic styling. All of these modules are still being designed and cannot be relied upon. Since CSS group chose to ignore a lot of previous work in this area (e.g. inventing its own expression syntax for media queries instead of using XPath, designing its own incompatible replacement of page master instead of building on top of XSL:FO), the amount of the implementation work that would be required to achieve parity even with Digital Editions 1.0 is huge. This does not preclude implementing CSS3 syntax for these features if they become stable and gain some acceptance, but it makes too expensive for us to experiment with them at this time."
The anti-CSS-3 rant is in italics. This argument has been around on the Internet for years. Anyway CSS-3 is emerging as a significant and excellent way to manipulate paged media. XSL:FO is arcane, opaque and very difficult to create and maintain. (Note: I am not particularly trying to start an argument here, just show why some thing happen and others don't.) We did five years of XSL:FO programming and when CSS-3 got to a stage of sufficient maturity grabbed it like a life-raft.
Images that have any unused or transparent areas should be PNG format with transparency
This creates a mismatch with ADE as it does not support transparency with PNG but would have to be regarded as a generally good recommendation.
To ensure proper viewing of images in content, use the HTML img tag instead of wrapping images in svg:img.
Probably OK, but they should support images in svg as that is a requirement. Haven't tested this one yet.
No shocks here except they don't support embedded fonts, which is a real shame because it is required more and more. So how well do they handle extended Unicodes and symbols - we are yet to find out. Better than ADE hopefully. (In ADE we have prepared Unicode table fonts made using the SIL based Gentium font so we can quickly embed a small sub-set of glyphs covering the widest range of languages).
Here is a new one, and a good one. In fact we are going to implement this in our standard epubs immediately.
To prevent text from being clipped by the bounds of the content area, insert soft hyphens into long words and especially into linked text and headings.
Apple encourage linking to sections - counter to ADE "best practices". iPad will definitely not have the resource constraints of ADE devices so more extensive header linking is possible and encouraged. Also the use of a page map. The rational is educational content and page referencing. It is unclear whether the intention is to map this to the same page numbers as a hard copy.
Interestingly the Apple recommendations "seems" to mandate the use of the <guide> structure which is optional in the OPF specification. This will be interesting as most ePubs don't worry about using the optional guide and use the spine for reading and access order and the NCX for TOC presentation and user navigation. Since I know the test book illustrated doesn't have a <guide> section, and plays and presents well, it must be assumed this remains optional.