Re: XML namespaces on the Web - proposal restated

Liam Quin wrote:
> On Thu, Nov 19, 2009 at 11:00:19AM -0500, Sam Ruby wrote:
>> A more relevant question would be: what is Liam's intent?  Having talked  
>> to him at TPAC, I gather that he would like to see an XML' (read as: XML  
>> prime) which differs as little as possible from the current XML  
>> recommendation but is somewhat more suitable for a (possible niche) set  
>> of use cases that he doesn't perceive HTML5 satisfying.
> 
> Sam, thanks for steering the ship back onto course.
> 
>> If I understand Liam's intent correctly, I believe that such entails a  
>> lot of work: a lot of specification, a lot of advocacy, a lot of coding,  
>> and lot of testing, etc.
> 
> I hope not. I'll restate my original goals -- it might be that
> they don't align with the "distributed extensibility" topic.
> Certainly I did not intend to open the door to changing XML to allow
> silent error recovery as part of this proposal.
> 
> 1. Make it possible to serve XHTML as text/html in such a way that
>    HTML and XML processors end up with the same interpretation of 
>    the document as HTML documents, even in the presence of "svg" and
>    "math" elements.

It already is possible to serve a subset of XHTML documents as text/html 
in such a way that HTML and XML processors end up with essentially the 
sam interpretation of both documents, even in the presence of "svg" and 
"math" elements.

Such a subset does not, in general, include inline script elements. 
There are quite a few other differences too.  If your goal is to not 
have it be a subset, then suffice it to say that I'm skeptical.  Very 
skeptical.

>    I have proposed "Unobtrusive Namespaces" as a mechanism
>    for XML processors outside Webbrowsers, and "Imaginary Namespaces" as
>    a mechanism to describe how HTML 5 Web browsers actually work.
>    If accepted, there would be no code change for a Web browser to
>    support Imaginary Namespaces.  There would, however, be a
>    namespace definition file that would be part of the HTML 5 spec
>    (I hope as an external file referenced by the spec, though).

So, unless I'm misunderstanding, your goal with #1 is to increase the 
subset by reducing the need for xmlns "talismans", and offsetting this 
by a requirement for a side file.

> 2. Make "namespace mashups" and user-defined namespaces possible without
>    the need for, or with a reduced need for, the existing XML
>    namespace syntax.  This is bring the XML "Unobtrusive Namespaces"
>    proposal into Web browsers, and is an attempt to address
>    disrtibuted extensibility in HTML/Web user agents.
>    This part would indeed require coding and testing, although
>    "a lot" is subjective.
> 
> I hope this is clearer.  I'll see what I can do with regard to
> a prototype of unobtrusive namespaces (and anyone who can help
> with that feel free to contact me, I'm pretty maxed out for
> the next 2 weeks, despite officially being on vacation...)

 From my perspective, this is a set of modest improvements (that may or 
may not be doable or worthwhile: I'm not making value judgments here), 
that fall well short of the stated goal of "Make it possible to serve 
XHTML as text/html in such a way that HTML and XML processors end up 
with the same interpretation of the document as HTML documents".

> Liam

- Sam Ruby

Received on Thursday, 19 November 2009 17:04:37 UTC