Re: ISSUE-41/ACTION-97 decentralized-extensibility

Julian Reschke On 09-10-06 11.38:

> Maciej Stachowiak wrote:
>>> I agree that adding something else should be avoided. One way to avoid 
>>> it would be to align namspaceURI/localName more between text/html and 
>>> application/xhtml+xml.
>> namespaceURI and localName are aligned already, in the current HTML5 
>> draft. What's different is the parser behavior. HTML parsing behavior 
>> can't be identical to XML, within compatibility constraints. It's an 
>> open question how much closer it could get. 
> 
> Yes.

The Microsoft proposal is to allow namespaces only on <body> and 
children of <body>. This to avoid any negative effects of 
namespaces used "in the wild" today.

But if one introduced e.g. a <root> element for declaring of 
namespaces, then there would be no such effects whatsoever. Also, 
this would permit some default behavior for the root element to be 
specified, and thus hopefully easier deal with DOM issues?

For prefixed namespaces, one could do:
   <root xmlns:svg='http://www.w3.org/2000/svg'>

For default namespace one could do:
   <root xmlns='http://www.w3.org/2000/svg'>

For well known namespaces one could offer a shorter syntax. E.g. a 
short syntax for prefixed names could be:
   <root xmlns:svg="[svg]" >

Short syntax for default namespace could use (predefined) CURIEs:
   <root xmlns="[svg]" >
   <root [svg] >

The word 'root' is known for many from e.g. CSS. And I think 
having a root element will allow authors to give more attention to 
the namespace _URI_, which really is the key.

Thoughts?
-- 
leif halvard silli

Received on Tuesday, 6 October 2009 12:33:13 UTC