- From: Shelby Moore <shelby@coolpage.com>
- Date: Thu, 19 Dec 2002 02:53:25 -0600
- To: Ray Whitmer <rayw@netscape.com>
- Cc: www-dom@w3.org
I wanted to elaborate on why one-to-one correspondence between the presentation state and markup state (objects (DOM) and their style) is the only useful way to ask questions about the presentation state, when the goal is to change the presentation state without reducing markup state to presentation state abstraction. And this same logic would apply to any transformed state derived from markup state. Say you have markup objects, Foo, with style FooStyle. Now say you ask some question about presentation state, which does NOT include Foo. To modify the presentation state based on the answer, then you would need a new model of changing the presentation state which is not markup. Thus markup is no longer the only way to create a presentation state. Thus markup becomes the union of what was markup and the new model of creating presentation state. Thus markup abstraction is reduced to presentation abstraction. Now say you ask some question about presentation state, which DOES include Foo and FooStyle. You can then change Foo and FooStyle to change the presentation state based on the answer. Thus markup abstraction remains markup. A counter logic might be that the presentation layer is optional. That is fine with me, especially if you trust that people won't use it in all cases where they could reasonably accomplish same with markup only. Thus, you must expose the one-to-one correspondences between the markup state and the transformed derived state, to give the greatest opportunity for markup to be used exclusively for changing the presentation state. The only other logic is do not expose transformed states. But as I pointed out, then it is impossible to enable scripting on the presentation state, which is obviously in high demand. -Shelby Moore
Received on Thursday, 19 December 2002 03:53:12 UTC