Thanks to everyone for all the good feedback we've gotten already,  
keep it coming!

On May 16, 2007, at 11:23 AM, Stuart Brown wrote:

> I share some of the doubts expressed by Peter et al; the proposed  
> system also risks creating (in the eyes of novices and tag abusers,  
> at least), an alternative typing system... Imagine a document with  
> lots of Q&As in it.  A novice might think Aha! these are formatted  
> differently, so I will insert a <rendition xml:id="q-para"> and  
> <rendition xml:id="a-para">, and then go and mark up all his Q&As  
> with <p rend="#q-para">, etc.  Innocuous, except that they may not  
> understand that no claim is being technically made about the q- 
> paras other than that they are styled in a certain way.  In other  
> words, we're getting dangerously close to word-processing, where  
> formatting becomes the *pre-eminent* method of object type  
> distinguishing.
> So, completely to the contrary of the propositions, I think we  
> should *discourage* elements in the text from pointing directly to  
> <rendition>s.


The rend attribute would continue to be used for describing the  
*source* document.  As such, I don't think the danger you envision  
really exists.  In P4 users can already incorrectly use <p rend="q- 
para"> and <p rend="a-para"> instead of semantically correct  
descriptive markup.  The pointer to <rendition> provides, among other  
things, a mechanism for being more explicit about the appearance of  
the *source* document.

> Secondly, I am opposed to the restriction of the content of  
> <rendition> to any code from any vocabulary.  When encoding  
> manuscripts, there are a number of smudges, blobs, lines, etc,  
> which are worth noting, but which may not be exactly describable  
> within any particular display environment; additionally it forces a  
> supposedly platform-neutral description into making display  
> judgments.  I would prefer a structure of <rendition> that allowed  
> for suggested display codes, but also a prosaic description.

The content of <rendition> would not be restricted under the proposed  
change.  The Guidelines already state that the <rendition> element  
"may be used to hold an description of the default rendition to be  
associated with the specified element, expressed in running prose, or  
in some more formal language such as CSS."  The use of a formal  
language is suggested as an option, not mandated.