W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > July to September 1998

techniques document: comments

From: Nir Dagan <nir.dagan@econ.upf.es>
Date: Sun, 27 Sep 1998 04:26:19 GMT
Message-Id: <199809261935.VAA09219@darwin.upf.es>
To: w3c-wai-gl@w3.org
I have some comments on the techniques document

1. Tables of Structural vs. presentational 
------------------------------------------

First, I think that there is also a third kind of elements/attributes.
These are those who give the user-agent some sort of processing 
instructions.

E.g., BDO indicates whether a browser should respect the intristic 
directionality of letters. 

Another example is the type atribute in anchors and LINK. The MIME type 
of a linked resource is nither presentational or sturctural. 

The same hold with PRE (how to treat white space).

STYLE and SCRIPT just indicate to send the content to the style/script 
engine. (for some reason SCRIPT is classified as structural and STYLE 
as presentational; a script can easily write a stylesheet dynamically)

Replaced elements like IMG, OBJECT, APPLET, IFRAME, FRAME are nither 
structural or presentational. (for some reason OBJECT is classified 
as structural but IFRAME as presentational, althoguh they are  
very similar.)

FRAMESET is correctly classified as presentational since it indicates 
how to divide the screen, but FRAME is a replaced element like OBJECT.

SUP and SUB are presentational, not structural.

SPAN and DIV are author defined, and as such they convey 
no structural meaning to the user agents. Their attributes may
convey structure (e.g., lang)

Second, it would be better to state attributes together with their 
elements: type in UL is different than type in LINK.


2. Automatic loading of alternative documents 
---------------------------------------------

Shouldn't it be done via HTTP headers? Seems more economical
for both server and client than the LINK element.

3. Lists Technique C.2.7
-------------------------

With all due respect to screen readers, nesting unordered lists is 
part of pre-HTML2.0. I think the guidelines here violate the balance
between author and user-agent responsibilities.

4. Concening style sheet for bullets
------------------------------------

I think it would be better to have one class "new", and 
avoid bullet styles to LI as it may cause cascading problems 
with ordered and unorderd lists.
 
instead of: 

<STYLE type="text/css" ...>
   .newtxt { font-weight: bold;
             color: red;
             background-color: yellow }
   .newbullet { list-style : url(yellow.gif) }
</STYLE>
...
<UL>
   <LI class="newbullet">Roth IRA <SPAN class="newtext">New</SPAN></LI>
   <LI> 401(k)</LI>
</UL>

I suggest:

<STYLE type="text/css" ...>
   LI.new SPAN { font-weight: bold;
                color: red;
                background-color: yellow }
   UL LI.new { list-style : url(yellow.gif) }
</STYLE>
...
<UL>
   <LI class="new">Roth IRA <SPAN>New</SPAN></LI>
   <LI> 401(k)</LI>
</UL>


Image maps
----------

Due to the fact that Explorer 4 (sort of) supports images with OBJECT,
but does not support OBJECT as a link or an image map, I would recommend:

1. Use IMG, MAP and AREA for image maps.
2. write alt to your AREA and IMG elements.
3. write redundant links for HTML2.0 browsers.
4. when old and buggy browsers like Explorer 4, will
   not be widely used, read the HTML4.0 spec. and use OBJECT.

This is shorter and more useful, I think.
 
Drop cap example
----------------
In the technique document we have:

Example. 
<STYLE ...>
      .dropcap { font-size : 120%; font-family : Helvetica } 
</STYLE>
...
<P><SPAN class="dropcap">O</SPAN>nce upon a time...

Note. As of the writing of this document, the CSS 
pseudo-element ':first-letter', which allows authors
to refer to the first letter of a chunk of text, is not 
widely supported. 

end of quotation.

First of all this is not drop cap, as you have to float the 
first letter to get it to "drop". But floating is not (widely?) 
supported by Internet Explorer 3.

Second, this is a hack. It is identical (up to the font-family thing), 
at least from the point of view of accesibility, to:

<P><BIG>O</BIG>nce upon a time...
 
which is HTML4.0 strict. With all due respect, SPAN has 
no more structural meaning than BIG. So I can't see why this example 
is any good.

Replacing inline typographical elements one by one with SPAN is *not*
separating structure from presentation.

Third, First-letter is supported by some browsers, notably Opera3.5b9.


Regards,

Nir Dagan                            
Assistant Professor of Economics      
Universidad Pompeu Fabra
Barcelona (Spain)

website: http://www.nirdagan.com
email: nir.dagan@econ.upf.es

"There is nothing quite so practical as a good theory." 
-- A. Einstein
Received on Saturday, 26 September 1998 15:25:55 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:46:58 GMT