On Sun, Jul 24, 2011 at 4:43 PM, Anne van Kesteren <annevk@opera.com> wrote: > I think Selectors need to be defined in terms of the DOM. The DOM is the > tree model Selectors work upon in all implementations, and it is the tree > model the languages we care about are defined in. And if we want Selectors > to work for other tree models we could just state they have to be equivalent > to the DOM. That would work. But hand-waving about the tree model Selectors > work against should stop I think. There are several instances of Selectors being used on non-DOM structures. For example, WebVTT, or <http://jsonselect.org/>. There's no reason to exclude things like this. What's the benefit of adopting the DOM as the data model? The current element-tree is almost the same thing; an element is just a thing with a type, an id, a bag of classes, a bag of attribute key/value pairs, and an arbitrary set of pseudo-classes. That's basically just the DOM notion of an element, minus anything irrelevant for Selectors. ~TJReceived on Monday, 25 July 2011 00:14:20 UTC
This archive was generated by hypermail 2.4.0 : Monday, 23 January 2023 02:14:02 UTC