RE: SVG 1.2 Comment: Detailed last call comments (all chapters)

On Wed, 3 Nov 2004, Doug Schepers wrote:
> 
> It's in the SVG Charter to "meet the needs of the user community." The 
> vast majority of SVG users, as you will observe from the topics on 
> svg-developers@yahoogroups.com, are using it to developing applications.

It is unfortunately that the W3C is not fulfilling the needs of the 
community in providing accessible, device-independent languages for user 
interface.


> It seems that you don't want people to use SVG for anything but 
> graphics. I'm sorry, but it's a bit late for that.

People use HTML for graphics and layout, and there is demand for even more 
graphical things to be added to HTML, but you don't see the XHTML working 
group adding more <font>-like elements to HTML.


> You're misrepresenting (or misunderstanding) what I said. SVG should be the
> presentation layer of the user interface, not the whole user interface.

I have nothing against SVG being used to present applications, so long as 
the language sent over the wire is a standard UI language that can also be 
implemented without SVG, e.g. for blind users using speech-based browsers. 
As, for instance, XForms can.


> | XHTML2+XForms, for example, is capable of being fully usable in such 
> | diverse environments, and in fact is so usable by default 
> 
> No more so than SVG. Are you seriously claiming that XHTML is somehow 
> more able to be leveraged for speech UAs?

Are you seriously claiming that it isn't?!

SVG:
   <rect x="10" y="10" width="100" height="100" onclick="doit()"/>
   ...
   <text x="10" y="10">Test</text>

XHTML1 (or HTML4, for that matter) [1]:
   <button onclick="doit()">Test</button>

In a speech-driven UA, the XHTML version can _trivially_ be implemented by 
saying something along the lines of "Button: Test" and then saying "Test" 
will cause the voice-recognition software to automatically invoke the 
doit() function (by invoking a 'click' event on the button element).

The SVG version, however, gives the speech-driven UA no clue whatsoever 
that the button is a button, that the text is associated with the button, 
etc. There are, in fact, strictly no semantics involved.

This isn't hypothetical. There are speech-based browsers for HTML. I have 
even heard of dynamic braille-based HTML browsers.


[1] Similar things can be done in XHTML2+XForms, but I'm more familiar 
with XHTML1 so I'll use that for now.


-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Thursday, 4 November 2004 03:30:28 UTC