Re: What's the problem? "Reuse of 1998 XHTML namespace is potentially misleading/wrong"

Hi Robert,

On Feb 12, 2009, at 11:12 , Robert J Burns wrote:
> On Feb 12, 2009, at 3:15 AM, Robin Berjon wrote:
>> I really don't understand your point? Are you saying that elements  
>> with different semantics with the same fully qualified name are  
>> always bad? So long as they can be distinguished by context, it  
>> really doesn't matter. The only thing it breaks is DTD validation  
>> and no one sane would ever care about that.
>
> I'm not sure I disagree with what you're saying in terms of reusing  
> the same name in different contexts. However, there's no reason to  
> expect that two separate elements with the same name will only be  
> used in different contexts. For example, imagine that XHTML2 adds an  
> element 'small' that means the text is smaller than the surrounding  
> text (visually only). However, HTML5 introduces a 'small' element  
> where the element means specific legal disadvantages or caveats.  
> Imagine another meaning of 'small' element is as a small thought not  
> directly relevant to the main topic of the document. I'm not sure  
> how context gets us out of the name collision problem. Unless you  
> mean that a heuristic reading by an intelligent  human can always  
> discern what all the text in a document is about, but then we could  
> just name every element in the XHTML namespace 'small' from the root  
> element on down the list. So if a new semantic is worth introducing,  
> I think it is worth introducing a new name for that semantic instead  
> of overloading existing names. If its not worth introducing a new  
> name it probably is not worth introducing the new semantic.

To be fair in the case of 'small' I think the name is poorly chosen  
anyway, but for other reasons. It should be 'smallprint' (even when  
the style sheet renders it big on screen media) or something similar.  
I reckon 'proviso' might be a good pick.

For certain the case of two elements carrying different semantics that  
may appear in the same context is undesirable (see 'input' for  
precedent). But I was thinking *within a single language*, if you  
recall I was pushing back on your claim that HTML5 conflicted with  
itself. The case of two languages sharing the same namespace is  
another issue with a totally different entertainment value.

>> What matters is if a v1 implementation cannot do anything useful  
>> with a v2 document (and worse, vice-versa) or if an implementation  
>> cannot usefully distinguish between two incompatible languages  
>> sharing the same language.
>
> Which can be completely avoided by avoiding name collisions. With no  
> name collisions, its possible to allow an implementation to  
> distinguish among every semantics in the namespace even if they come  
> from two or ten different WGs

No, I'm afraid that to ensure compatible versioning you need a fair  
bit more than avoiding name collisions :)

-- 
Robin Berjon - http://berjon.com/
     Feel like hiring me? Go to http://robineko.com/

Received on Thursday, 12 February 2009 10:48:58 UTC