Michael Kay wrote:

>> wondering what benefit to this other then some syntatic human 
>> orientated sugar ?
>>   
>
>
> Do you need any other?
>  
>
ok, you got me....selfishly I was thinking 'all my perl scripts wouldnt 
work, if it had to consume a new definition of well formed xml'

There is still much adoption to occur with XML; lets not contemplate 
scratching the 'itch' of changing or adding to XML until we have 
unlimited storage, network bandwith, and processing power!

>> u know extending the XML spec to do this would have a pretty serious 
>> impact on other things such as XSLT, XPATH, etc...so its highly 
>> unlikely!
>>   
>
>
> It would have no impact at all on XSLT or XPath, since it only changes 
> the
> lexical representation of XML and not the data model.
>  
>
no doubt saxon would be the first to elegantly implement such a change, 
though I am still laughing at  the 'it *only* changes the lexical 
representation'....bit and how much work I would have to do and fix my 
perl/bash scripts.

I would weakly argue that lexical representation and data model are 
always interelated the moment the come into existence; if not only for 
the indirect inferences that programmers make (and depend upon) betwixt 
the two (as codified in my rubbish perl scripts).

As a sidenote; there is a 'chicken and egg' of what comes first with 
such things, as we are trying to do something useful in computing we 
tend to find a bit of one or the other already in/formally defined in 
existence e.g. the lexical represtentation or data model. In the case of 
XML its probable that it was a relatively clean slate (though heavily 
informed by experiences with SGML, html, lisp etc...etc...) and the spec 
itself defines good seperation between the two; though I would bet large 
sums against such an event occuring (going to ladbrokes.com now).

-- Jim Fuller