W3C home > Mailing lists > Public > public-html@w3.org > May 2007

Re: 'role' should be property

From: Murray Maloney <murray@muzmo.com>
Date: Wed, 23 May 2007 10:40:34 -0400
Message-Id: <>
To: public-html@w3.org

At 08:59 AM 5/23/2007 -0500, Dimitri Glazkov wrote:

>No disrespect to you personally intended. Only the idea. If you're
>offended, I apologize.
>It's not about some software spending some cycles. It's about access
>barrier. Per the original idea, in order to write something like
>Operator or Technorati's Microformats Search, I'll have to also write
>a CSS parser and the DOM kidoodle that comes with the necessity to
>resolve rules and apply propertie. Which quickly puts software
>development like this out of reach for anybody but a browser vendor.
>And you're mixing meaning into styling. How much worse could it get?
>If I don't get it, educate me.


I think that you may be misunderstanding what is intended.

First of all, it would help if you thought about the 'role' attribute in 
much the same
way that you might think about the 'class' attribute. That is, the value of 
'class' is a key
by which CSS can make discriminating selections and thereby differentiate 
style for
a given element or set of elements. Similarly, it is being proposed that 
the value of 'role'
is a key which may be used by [software agents to be named later] to make 
selections and thereby differentiate meaning for a given element or set of 

Others have pointed out that it is both theoretically and practically 
possible to assert
such keys without resorting to the use of either the 'class' or 'role' 
attributes. Still others
have pointed out that such techniques can prove cumbersome both for authors 
for software developers.

The expedient solutions would seem to be the ones that markup language 
have been employing for two decades now. To wit: bind semantics to an 
element type,
a data type, or an attribute value. The problem is somewhat complicated by 
the need
to be able to express the semantic out-of-band -- that is, not in the DTD 
or Schema.

HTML's existing profile mechanism and the proposed 'role' attribute offer 
means to
relate a key to an out-of-band definition of the meaning of the key.

Unfortunately, neither offer much in the way of a schema for such 
definitions, unless
one considers RDF/XML a suitable format.

I think that more thought and effort should be put into how to encode 
semantic relations
into HTML documents and how to write definitions using HTML.


Received on Wednesday, 23 May 2007 14:41:28 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:15:21 UTC