Thomas Roessler wrote:
Is that a P&C specific concern or one that should better be covered in WebIDL?
It's a P&C concern, not related to WebIDL. I think WebIDL kinda assumes a layer of abstraction over ECMAScript - this is not the case with P&C which knows nothing about scripting languages, and should not have made any assumptions about what language it would be implemented in (again, my mistake was that I was writing a reference implementation in Java while I was spec'ing P&C, so it all made sense at the time:)). The problem is that the spec algorithms are written in pseudo code in a manner that tries to be generic... but fails a bit because it uses concrete concepts from ECMAScript/Java where it shouldn't. In most cases, it's ok, but as shown with null, it can lead to ambiguities.
