Quoting Mark Birbeck <mark.birbeck@x-port.net>:
> <img> is a bit like <a> in that sometimes an author wants to insert an image
> in a document, and it just *is* an image (or a link, in the case of <a>). In
> other words, it's not a list item that is represented by an image, or a
> heading that is represented by an image, it's just an image. (If it was a
> heading, for example, then <h src="..." etc.> would be a better choice for
> the author.)

Sometimes I want to insert a <video> in my document, or a <audio> fragment, or
some <radio>, or an <animation> or a <three-dee-image> etc.

Instead of inventing a lot of elements the HTML WG once thought of 
<object>. The
rationale was also that <object> was eventually going to replace <img> 
from what
I've heard. (That was also the reason <img> was out of the draft and <object>
has been tested everywhere for interop.)

Now you're arguing that sometimes specific elements are good?

For linking, <a href=""> can just be <span href="">. After all, an <a> element
without "href" specified has similar semantics to a <span>. Why wouldn't the
opposite be true? And I believe the opposite is true following the current
wording...

For embedding random objects, why can't <object data="">/<object src=""> be
replaced with <div src=""> or <span src=""> (block level replaced element
versus inline level replaced element). Same thing, imho.


-- 
Anne van Kesteren
<http://annevankesteren.nl/ >