W3C home > Mailing lists > Public > wai-xtech@w3.org > August 2007

Re: Alternate Additional Attribute Set for a Single Quote Element

From: Robert Burns <rob@robburns.com>
Date: Thu, 2 Aug 2007 15:15:49 -0500
Message-Id: <474C9B4D-8D51-4EFA-9390-DA011DA54AB0@robburns.com>
Cc: wai-xtech@w3.org
To: "Dave Pawson" <dave.pawson@gmail.com>


On Aug 2, 2007, at 9:11 AM, Dave Pawson wrote:

> On 02/08/07, Robert Burns <rob@robburns.com> wrote:
>
> dp
>>> I differentatiate between markup, explicit to contained content
>>> (Gregory's
>>> proposal) and the relationship between markup (xpath expressions)
>>> that you propose.
>>
>> I see. I'm not really proposing xpath expressions. This would simply
>> involve a DOM attribute and CSS pseudo-class selectors
>
> Which IMHO is an algorithmic solution rather than markup.
> I'm very much in favour of Gregory's plain markup solution.

It is most definitely a markup solution. It allows authors to focus  
on semantics of using a quotation element when they are directly  
quoting someone else. Then the quotation can involve any of the  
various content model markup. The author doesn't have to perform the  
redundant step of adding a @type attribute value that simply  
describes the markup already performed. In the case of code and  
blockcode in XHTML2 an attribute is necessary to distincguish them.  
So XHTML2 adds the @layout attribute. In a way parallel to Q/ 
BLOCKQUOTE authors could just use CODE and set the @layout attribute  
to indicate the whitespace (especially the line-breaks) provide the  
structure for the element. However, with Q/BLOCKQUOTE no attribute is  
necessary. It is readily apparent from the contents of the Q element  
that it contains structural markup.

>>> An algorithm would suppose that was the intent, is that correct?
>>
>> Right. The algorithm would use introspection on the quotation element
>> to determine it's current state. The author would determine the value
>> of the DOM attribute by what content model the author employed within
>> the quotation element.
> IMHO it's not a content model it's a structural one.
> If the author gets the structure wrong, your algorithm will fail.

Could you provide an example of how an author would get the structure  
wrong? I don't see how that could happen. If the author intended to  
not include the enclosing <p> elements, that might be considered a  
mistake. However in terms of rendering it would be readily apparent  
and it could not be considered a markup mistake (only a mistake of  
the authors intentions). This is in contrast to an author markuping up:

<q type='inline' ><p>Many authors use improper markup using divs as  
their only block level division.</p></q>
The above is block level and would be returned from the DOM API that  
way and made available as a CSS class selector.

<q type='block' ><em>Many</em> authors use improper markup using divs  
as their only block level division.</q>
The above is strictly-inline level and would be returned from the DOM  
API that way and made available as a CSS class selector.

Where now we have markup mistake in the sense that the type attribute  
contradicts the contents of the element.

>
> rb.
>> I agree (as an author). I would much prefer to provide quotation
>> marks through CSS. However, this proposal (which is separate from the
>> content model algorithm proposal) is to deal with authors and UAs
>> (Internet Explorer) that prefer to handle quotations with quotation
>> marks.
>
> dp.
> I'd rather the spec drive implementations than the other way round.

Me too. I see this second proposal (adding the @needsmarks boolean  
attribute) as rescuing spec initiated proposals from an  
implementation junta (i.e, IE trying to redefine how a quotation  
should work)..

Take care,
Rob
Received on Thursday, 2 August 2007 20:16:37 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 13:15:43 GMT