Good points...I'll try to clarify:
There may not be the ability to change the layout, but there are "layout" considerations when developing desktop software. If you are building a desktop application and drag and drop form fields (a convenient example, I'll admit) their tab order is in the order in which they were dragged on to the "form" or "stage" or whatever-the-thing-is-called-in-your-situation. So, what do we do? We make the layout more linear using the tabIndex property. The linear, logical order is the equivalent to layout - it mimics to a certain extent the logical visual layout that is apparent on through visual grouping, proximity and similarity of style.
If tab index was all we were worried about, this discussion would be over. It goes way beyond that. As you indicated, this example was a bit contrived - even in the web, a document stripped of all supporting files still maintains tab index. The same cannot be said for the implied importance a certain color scheme gives to an element. If we could 'notify' the screen readers that some specific DIV was 'important', 'ranked higher' or had 'new content', we have provided a usable substitute to color alone.
I'm not following what you mean here (seriously!) - are you saying that semantic structure doesn't exist in desktop applications? I'm not sure how that applies here... Can you clarify?
I was attempting to draw the parallel with web documents stripped of all styling methods and desktop apps stripped of layout context (if that were possible). If you have ever used Spy++ (an MS tool designed to show handle references), you would see that even the simplest applications are made up of many, nested objects that don't necessarily contribute to the semantic meaning of the application (and in some cases contradict it). For example, a simple drop down list (like our SELECT tag) has it's own window handle in the OS. Is it a window to the user??? Absolutely not. Is it a window in the context of the application??? No. Does this (on it's own) effectively convey what this control is for?? No. Not very semantic (by definition, anyway). Yet desktop screen readers are fine with this type of hierarchy. So to bring this example to the web (and to use my own contrived example), let's say I have an dv / iframe / object floating (visually) somewhere at the bottom of my web app that is delivering a custom channel of information that's updated every minute or so, and it briefly flashes when it updates. What would be the harm in allowing me to 'register' this tag with the screen reader as the 'Stock Quote Feed' to give it some context???? And also, when it updates, I want to the let the reader know via a raised event. I think the WHATWG was/is considering something like this, but I'm surprised it has taken this long.
I don't think the gap is as big as you think it might, to be honest.
Maybe not for web content sites that are document-centric to begin with, but the web application world is not so easy to 'flatten'. But that's a topic for another day.... HTH, Mike ******************************************************************* List Guidelines: http://webstandardsgroup.org/mail/guidelines.cfm Unsubscribe: http://webstandardsgroup.org/join/unsubscribe.cfm Help: [EMAIL PROTECTED] *******************************************************************