Comments on: The shape of EPUB to come, #1 http://www.teleread.org/blog/2008/04/09/the-shape-of-epub-to-come-1-for-our-more-technically-inclined-readers/ News & views on e-books, libraries, publishing and related topics Tue, 07 Oct 2008 01:07:12 +0000 http://wordpress.org/?v=2.6.2 By: Aaron S. Miller, CTO of BookGlutton http://www.teleread.org/blog/2008/04/09/the-shape-of-epub-to-come-1-for-our-more-technically-inclined-readers/#comment-760185 Aaron S. Miller, CTO of BookGlutton Fri, 11 Apr 2008 02:15:20 +0000 http://www.teleread.org/blog/2008/04/09/the-shape-of-epub-to-come-1-for-our-more-technically-inclined-readers/#comment-760185 I like the possibilities of RDF but it seems heavy to me right now. My own experience with parsing it for practical reasons has not been good or fun. From a user agent perspective, it's not lightweight enough, IMO. So you would propose a way to link to an RDF contained within the OPS? And within that would be information about highlights, bookmarks, ids, or whatever? My problem comes from having a purely web-based perspective, I think. I want epub to be something that browsers can run with without much overhead, and right now, many browsers don't know what to do with RDF, and if you want to have them parse it, you're talking about lag times. It would be nice for people, for example, to be able to have a list of hyperlinks on their blog or home page, each of which links into an OPS structure, and perhaps describes passages or collections of passages. That would be more in keeping with the state of the web right now. Re: multiple editions -- this comes down to UUIDs or ISBNs, and back to URIs, right? The spec states that each package should have a unique identifier and a specified type for that identifier, so whatever it may be, each edition of a book should have a new one. I really like the fact that the spec for OPS allows for alternate formats of a book to exist within the package itself, such as PDF/HTML/XHTML versions and partitioned/non-partitioned or graphical/non-graphical versions all existing in one package. But I do think different editions should be separate packages. I like the possibilities of RDF but it seems heavy to me right now. My own experience with parsing it for practical reasons has not been good or fun. From a user agent perspective, it’s not lightweight enough, IMO.

So you would propose a way to link to an RDF contained within the OPS? And within that would be information about highlights, bookmarks, ids, or whatever?

My problem comes from having a purely web-based perspective, I think. I want epub to be something that browsers can run with without much overhead, and right now, many browsers don’t know what to do with RDF, and if you want to have them parse it, you’re talking about lag times.

It would be nice for people, for example, to be able to have a list of hyperlinks on their blog or home page, each of which links into an OPS structure, and perhaps describes passages or collections of passages. That would be more in keeping with the state of the web right now.

Re: multiple editions — this comes down to UUIDs or ISBNs, and back to URIs, right? The spec states that each package should have a unique identifier and a specified type for that identifier, so whatever it may be, each edition of a book should have a new one.

I really like the fact that the spec for OPS allows for alternate formats of a book to exist within the package itself, such as PDF/HTML/XHTML versions and partitioned/non-partitioned or graphical/non-graphical versions all existing in one package. But I do think different editions should be separate packages.

]]> By: Hadrien http://www.teleread.org/blog/2008/04/09/the-shape-of-epub-to-come-1-for-our-more-technically-inclined-readers/#comment-760107 Hadrien Thu, 10 Apr 2008 23:16:47 +0000 http://www.teleread.org/blog/2008/04/09/the-shape-of-epub-to-come-1-for-our-more-technically-inclined-readers/#comment-760107 Well honestly I think that in this case, a URI is mostly an identifier. All the things that you listed could be represented as triples. You can't strictly rely on a URI approach for linking: in some cases, you'll have to use the DublinCore (author, title etc..) describing the resource to find and launch the right download, in another case you might use the resource that you provided using dcterms:hasFormat etc... You could list all these attributes somehow directly in an URI but it's much cleaner to use something like RDF for this. The real challenge of linking is the existence of multiple editions of a book. If you know which edition you're linking too, it's pretty easy to use DOM elements for example to identify everything. Unfortunately, most of the time, you won't have this sort of information. Well honestly I think that in this case, a URI is mostly an identifier. All the things that you listed could be represented as triples. You can’t strictly rely on a URI approach for linking: in some cases, you’ll have to use the DublinCore (author, title etc..) describing the resource to find and launch the right download, in another case you might use the resource that you provided using dcterms:hasFormat etc…
You could list all these attributes somehow directly in an URI but it’s much cleaner to use something like RDF for this.
The real challenge of linking is the existence of multiple editions of a book. If you know which edition you’re linking too, it’s pretty easy to use DOM elements for example to identify everything. Unfortunately, most of the time, you won’t have this sort of information.

]]> By: Aaron S. Miller, CTO of BookGlutton, a Web-based community of readers http://www.teleread.org/blog/2008/04/09/the-shape-of-epub-to-come-1-for-our-more-technically-inclined-readers/#comment-760093 Aaron S. Miller, CTO of BookGlutton, a Web-based community of readers Thu, 10 Apr 2008 22:23:26 +0000 http://www.teleread.org/blog/2008/04/09/the-shape-of-epub-to-come-1-for-our-more-technically-inclined-readers/#comment-760093 Hadrien, I'm very glad to see a discussion of linking! I've pondered the issue myself here, and you bring up interesting additional concerns. While RDF is great at describing relationships and metadata, when it comes to linking, I think the URI approach is the most important one for the IDPF to address. Beyond identifying the resource itself, a URI should be able to: -identify any DOM element in the publication, WITHOUT relying on a unique ID -identify any range of elements or characters in the document -identify any COLLECTION of ranges of elements or characters in the publication -identify any point in the document (corresponding to a unicode point) -identify a fallback mechanism if the resource has changed or been removed: SUBSTITUTE, SEARCH, or REDIRECT Hadrien,

I’m very glad to see a discussion of linking! I’ve pondered the issue myself here, and you bring up interesting additional concerns. While RDF is great at describing relationships and metadata, when it comes to linking, I think the URI approach is the most important one for the IDPF to address. Beyond identifying the resource itself, a URI should be able to:

-identify any DOM element in the publication, WITHOUT relying on a unique ID
-identify any range of elements or characters in the document
-identify any COLLECTION of ranges of elements or characters in the publication
-identify any point in the document (corresponding to a unicode point)
-identify a fallback mechanism if the resource has changed or been removed: SUBSTITUTE, SEARCH, or REDIRECT

]]> By: Robert Nagle http://www.teleread.org/blog/2008/04/09/the-shape-of-epub-to-come-1-for-our-more-technically-inclined-readers/#comment-759609 Robert Nagle Thu, 10 Apr 2008 06:37:33 +0000 http://www.teleread.org/blog/2008/04/09/the-shape-of-epub-to-come-1-for-our-more-technically-inclined-readers/#comment-759609 So the URL would need to indicate the fragment of the text corresponding to the link. The ISBN could be the local copy which could be checked against before trying to find it externally. Is it possible to use a hyperlink now which could open up an ebook at a specific spot? (with digital edition or mobipocket?) So the URL would need to indicate the fragment of the text corresponding to the link. The ISBN could be the local copy which could be checked against before trying to find it externally. Is it possible to use a hyperlink now which could open up an ebook at a specific spot? (with digital edition or mobipocket?)

]]>