I think there is clearly need for clarification in the spec if there's so much room for (mis)interpretation. I am in favour of inheritance, but it's not altogether straightforward.
Say you have a feed F containing entries a, b, c: F = [a, b, c] and authorship is expressed as auth(a, x) The way I read inheritance as it stands was: auth(F, x) => auth(a, x), auth(b, x), auth(c, x) But the idea of authorship of the collection might screw this up, only allowing: auth(F, x) => auth([a, b, c], x) or are you actually asserting both? i.e. auth(F, x) => auth([a, b, c], x), auth(a, x), auth(b, x), auth(c, x) Things get even worse when individual entries have authorship too - which of the following is it? : auth(F, x), auth(c, y) => auth(a, x), auth(b, x), auth(c, x) or auth(F, x), auth(c, y) => auth(a, x), auth(b, x), auth(c, x), auth(c, y) or auth(F, x), auth(c, y) => auth([a, b, c], x), auth(c, y) or the combination of the last of these with each of the first two..? I think if we are to make use of inheritance then we need to differentiate clearly between what is meant by authorship (or whatever) of the feed as an entity in its own right and authorship of the feed as a series of entries, i.e. a, b, c... != [a, b, c...]. I think inheritance would be worthwhile as I think it would make it more likely that attribution data would be available to consumers. Cheers, Danny. -- http://dannyayers.com
