W3C home > Mailing lists > Public > www-html@w3.org > July 2003

The embedding attribute collection

From: J. King <mtknight@dark-phantasy.com>
Date: Mon, 14 Jul 2003 23:58:36 -0400
To: www-html@w3.org
Message-ID: <oprsb2nyqmk4suho@mail.dark-phantasy.com>

Forgive me if I have missed a glaring piece of information in the latest 
working draft of XHTML 2.0 or if I misunderstand the importance of 
semantics in the continued work on XHTML, but I have been giving this some 
consideration and am at a bit of a loss.

While discussing the importance of <object>, Masayasu Ishikawa pointed me 
to the embedding attribute collection module [1], which he said would be 
used more often to insert simple raster images than <object>.

Though the potential is quite interesting, I don't see how most uses for 
these attributes, especially those presented by the examples in the working 
draft, can possibly be semantically proper, or in some cases even 
practical.

The first example, an image marked up as a paragraph, seems to be poor 
form.  A paragraph implies text, and unless the image in question were a 
graphical representation of text (such as text in a writing system Unicode 
does not yet include, a fictitious language, or radically styled text that 
CSS cannot reproduce), This would be a gross misuse of markup in a semantic 
sense.  As the example is obviously not a representations of text, the 
draft seems to be promoting a poor use of the mechanism.

In the case of the second example, a table, the use of an image is more 
likely than not unnecessary and a waste of effort, though completely 
proper.  Given the capabilities of CSS2 and CSS3 and the fact that 
alternate content is provided in the example, an image, considering how 
much easier markup is to edit, would be a waste of time and bandwidth.

The examples, in short, should be reworked to better promote semantic 
correctness.

In most cases, I get the impression that no element such as <p> would be 
semantically correct, and using the generic elements (<span> and <div>) 
would cause the document to dramatically lose clarity.  Add to this the 
fact that <img> and <object> are typically styled as inline-blocks whereas 
other elements are not, and you run into quite a mess, especially when you 
take into account conventions (and necessities) of CSS styling of blocks 
and inlines.

As I stated above, there are some cases where the attributes would be 
useful (mostly as graphical representations of stylised text), but 
relatively few.  One case that comes to mind would be <hr>, where the 
element would be a graphic without need of a textual alternative, but 
that's about all I can think of.

The collection does have its uses, but I get the impression that <object> 
will be used far more than Mr. Ishikawa believes, simply because it's much 
more practical--and logical--to most people, including myself.

[1] http://www.w3.org/TR/xhtml2/mod-attribute- 
collections.html#col_Embedding

-- 
J. King
mtknight@dark-phantasy.com
http://dark-phantasy.com/
http://snap.dark-phantasy.com/
Received on Monday, 14 July 2003 23:58:39 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 27 March 2012 18:15:56 GMT