Hi there. (this message contains personal comments and does not represent an official response from the CSS WG)
I have read the recent Selectors API Level 2 draft [1] and have a few important comments to make: 1. I don't like the idea of refNodes. I think having the APIs specified at Element level makes it confusing. I would recommend applying the NodeSelector interface to NodeList instead. If queryScopedSelector() and queryScopedSelectorAll() are applied to an Element or a NodeList, the corresponding element(s) are the refNodes of the query. Same comment for matchesSelector(). 2. I am extremely puzzled by the parsing model of scoped selectors. In particular, I think that the :scope pseudo-class introduces things that go far beyond scoping. Let's consider the selector ":scope+p". Clearly, it's _not_ scoped since it queries element that are outside of the subtree the context element is the root of. Furthermore, these elements can be queried without scopes, and I don't see why this is needed at all!!! I would recommend dropping the pseudo-class :scope and make a simpler model where a fictional :scope pseudo-class and a descendant combinator are prepended to all selectors passed as the argument of the corresponding APIs. I don't like the idea that implementors will have to check if the first sequence of simple selectors in a selector contains or does not contain a given pseudo-class to prepend something to the context. This is clearly the kind of things I think we should avoid in Selectors in general. 3. the section about :scope does not include error handling. What happens if multiple :scope are present? 4. what's the specificity of that pseudo? Since it's proposed as a regular and non-fictional pseudo, web authors _can_ use it in regular stylesheets, even if it's meaningless outside of a scoped stylesheet. What's the behaviour in that case? What's the specificity? [1] http://www.w3.org/TR/selectors-api2/ </Daniel> -- W3C CSS WG, Co-chair
