Re-org of vocabularies and mechanisms

[mattgarrish]

This PR attempts to address the following vocabulary re-use issues:

• it moves the vocabulary association mechanisms section to an appendix and generalizes it to not be so specific to the package document
• the default vocabularies are defined in appropriate attribute to which they apply - the section in the appendix only describes their purpose
• it moves the epub:type attribute definition to an appendix because it is not strictly an HTML extension but used by HTML, SVG and media overlays. It is also generalized. (I tried to find a way to incorporate this under the vocabularies, but it just doesn't logically fit there.)

It also addresses some peripheral issues:

• it integrates the fixed layout re-org into this PR as it is also part of normalizing the vocabularies
• it adds the structure vocab to the vocabularies (this may be controversial, but it is getting to be a strange outlier now)
• it deletes the index - which, in addition to being unnecessary, also helps avoid too much bloat from adding the structure vocab

Fixes #1377
Fixes #1386
Fixes #1392

---

Preview | Diff

[iherman]

I very much like it. I also note that it covers #1387 insofar as it makes it clear that the EPUB version of prefix MUST be used with the epub CURIE prefix and not by itself, thereby avoiding the clash with the RDFa usage of a similar term.

Huge +1 from me, thus...

Minor points:

• the media prefix is also used in media overlay documents, and not only in the Package document. (ref to the new §C.1.5.). Maybe worth adding there (although it is mentioned in the separate media overlay part)
• The various schemas seem all to be hosted in the epubcheck repository. We may have to discuss this; I believe it would be better to store those (authoritative) schemas as part of the group's specs repository and serve them via a w3c.github.io URL.
• What is the role, if any, of the EPUB-SSV reference now? It looks to me as if C.8 has superseded the separate document. If that is the case, that reference should be removed. If this is not the case, we have to have a clear separation between the two, which definition is authoritative, etc.

[iherman]

Oops, my minor comment

the media prefix is also used in media overlay documents, and not only in the Package document. (ref to the new §C.1.5.). Maybe worth adding there (although it is mentioned in the separate media overlay part)

doesn't hold. It is used in the package document for media overlay, not in the smil file. Withdrawn... :-)

[mattgarrish]

The various schemas seem all to be hosted in the epubcheck repository. We may have to discuss this; I believe it would be better to store those (authoritative) schemas as part of the group's specs repository and serve them via a w3c.github.io URL.

Right, we had this discussion in 3.1 or 3.2. We used to host them with the specifications, but that meant this group had to maintain them separately from epubcheck, even though epubcheck is where the schema development is primarily undertaken. The schemas are only part of the checks required to validate epubs, so they're only ever incomplete at best when pulled out of a real implementation scenario.

Plus it also raised questions about the point of creating static schemas. Once authored, any fixes would be implemented in epubcheck, making the versions pointed to by the specification out of date.

[mattgarrish]

What is the role, if any, of the EPUB-SSV reference now?

I'd imagine none. I took the one out of the epub:type section, but forgot to hunt for others. I'll switch them to internal links now.

[mattgarrish]

I made a few minor clarifications/generalizations going back over the changes, but am good now with integrating this.