* Paul Hoffman <[EMAIL PROTECTED]> [2005-08-22 01:00]: > The crux of the question is: what happens when an extension > that does not specify the scope appears at the feed level?
Let me step back to look at the larger issue for a moment. That issue is inheritance. atom:author is the only precedent for it in Atom. It works because each entry ALWAYS has at least one author; the override mechanism is simple. Additionally, all Atom processors are aware of it by definition. But atom:author is not a model to emulate. It works only because it falls within a narrowly defined set of circumstances. The WG wisely avoided attempting to specify inheritance for atom:contributor. An entry MAY have no contributors at all, and this would require complications to express if feed-level atom:contributor elements inherited to entries. My point of view is that the default interpretation should avoid forcing extensions to specify complicated mechanisms that the WG avoided introducing when specifying atom:contributor, when an extension element’s semantics are the same as those of atom:contributor. And with that, getting back to your question, the answer seems pretty clear: it depends on whether the extension element is more like atom:contributor, ie defines a property which an entry may or may not have, or more like atom:author, ie defines a property that every entry inevitably has. But because that is a matter of interpretation, I would strongly prefer to say that if the extension does not specify a meaning for an element at the feed level, then the meaning is undefined. On the other, related point, the same principle of avoiding the necessity of complicated override mechanisms is why I say that aggregators should assume that unknown extension elements (which therefore have *unknown* as opposed to *undefined* semantics) pertain only to the feed, not to its entries. Regards, -- Aristotle Pagaltzis // <http://plasmasturm.org/>
