Open dauwhe opened 8 years ago
There are two separate issues here, could you divide things a little differently?
First one is that you don't want to use "links". Second one is that you don't want to separate the spine from the manifest.
I'd like to propose a solution for that, which would be compatible with your proposed example:
Basically, this would avoid defining manifest differently (it's still a collection), and would also extend to any other collection whenever it's useful.
A full collection uses:
While a compact collection is limited to link objects.
While it does make authoring easier, it'll make parsing those files a little more complex (since you'll have to detect if you have a full collection or a compact collection) but it's probably an acceptable compromise.
My proposal would still separate the spine from the manifest btw, but it makes the syntax more compact and closer from what you're proposing.
In the proposal, a rendition object (see issue #13 ) contains a links array which contains objects describing the files, unless they are not spine items, in which case they're one level deeper in a manifest object. Might a simpler structure contain the same information?
One could determine which files are spine items by file extension and/or attribute. Again, in the common cases HTML files are spine items, and everything else isn't.
Advantages:
Disadvantages: