Re: Unmangling XML and the XML Core WG (was: XML namespaces on the Web)

On Wed, 18 Nov 2009 23:02:49 +0100, John Cowan <cowan@ccil.org> wrote:

> Liam Quin scripsit:
>
>> I'm with John that it'd be better not to call it XML Error Recovery,
>> but, e.g. "Web Browser and RSS Reader Recovery For Sad And Pathetic
>> Content Vaguely Like XML" :-)
>
> Or even "HOWTO Force Any Old Octet Sequence Into Well-Formed But Possibly
> Stupid XML".

Just a note: XML5 doesn't try to turn a sequence of octets into XML, it  
tries to turn a sequence of octets into a DOM. The difference is that the  
latter doesn't necessarily need to make sure that e.g. local names match  
NCName.

(The HTML5 parser doesn't make sure that local names match NCName, but  
there's  
http://www.whatwg.org/specs/web-apps/current-work/multipage/the-end.html#coercing-an-html-dom-into-an-infoset  
which will among other things make sure local names match NCName for APIs  
that need it.)


>> yes, a separate spec talking about a strategy for building a
>> XML-compatible DOM out of not-quite-XML input would be fine, I think.
>
> The XML Core WG discussed the question today.  What follows is  
> unofficial,
> just what I think I heard there.
>
> I urged the WG to take up the question of writing down such a strategy,
> on the grounds that people want it, someone should do it (just once), and
> the Core WG has the necessary expertise.  Our staff contact said that in
> his view a rechartering would be required to perform such work, as it is
> independent of the XML language as such, XML being merely the output of
> the process.  I agreed, but further urged that we seek such rechartering.
> The question will be talked about further at later Core WG meetings.
>
> I agree with Shelley that the HTML WG shouldn't hold its breath on
> this one.
>


-- 
Simon Pieters
Opera Software

Received on Thursday, 19 November 2009 07:24:38 UTC