On Sun, 29 Jun 2003 15:47:48 -0400, Jason M. Kikta <kiktajm@muohio.edu> 
wrote:

>
> I think you misunderstood what I was saying (or I didn't present it well, 
> which is more likely).
>
> It is important to break backwards compatibility in this case, because of 
> an existing bug. If IE won't try to render it, it will move on to the 
> fallback. The beauty of <object> is that it allows for multiple nested 
> "fallbacks" if the browser can't handle the top level object. So you can 
> do this (really rough example, tabs are for clarity):
>
> <object data="test.png">
> 	<object data="test.jpg">
> 		<img src="test.gif" alt="Test Picture" />
> 	</object>
> </object>
>
> In this situation, the rendering engine will try to load the PNG first. 
> If it doesn't understand/know what it is, it move on to the JPEG. If it 
> still can't load it, doesn't understand the file format, or doesn't know 
> what <object> is, it will load the GIF (with the ALT text as a further 
> fallback).
>
> The problem, like I said, isn't browsers that don't understand <object>, 
> because they will move on to the backup. The problem is idiot browsers 
> like IE, that can't render it properly but think that they can. Switching 
> to <obj> would solve this problem, since IE would go to the <img> tag, 
> and you would still have valid XHTML 2.
>
> Jason
>
>

Sorry for posting twice, but I thought it would be worth mentioning that 
your example would not actually be valid XHTML2, not unless you included an 
additional module that defined the <img> element for use in the document.  
You can do that, of course, but to do so to fix a bug in a user-agent when 
you could simply use older technology is wasteful.

-- 
J.