Re: Ignore Order while validating XSD

On 2009-01-26, at 05:32:31, Michael Kay wrote:

>> Not being an XSD expert, maybe my question is "why can't the
>> content ordering possibilities be specified as a regex-like  
>> pattern?".

> One of the criticisms of DTDs, which XSD attempted to address, was  
> that DTDs
> were written in a non-XML-like syntax which made it difficult for
> applications to process the information (and difficult in many cases  
> for
> authors to remember the syntax). The thinking was that by using an  
> XML-based
> syntax, the data would be much more accessible; and some argued that
> verboseness doesn't matter because authors would be working with  
> specialized
> tools rather than editing at the source level.

I see. I believe that syntax integrity is good. However, my thinking  
is that xs:pattern regexen are part of vocabulary, and elements,  
including user-defined element types, are now supported data types.  
Therefore a pattern can express element ordering possibilities using a  
subset of regex notation. So this mythical 'collection' model group  
would subsume 'all', 'choice', and 'sequence' by having the ordering  
specified in a 'pattern' attribute provided it could contain itself.

<xsd:collection pattern="(b|c|x|xsd:collection){0,}, a{2}, (a|b|c|x) 
{0,}">
   …
</xsd:collection>

This would result in a simplification of both the specification and  
implementation and probably be more cost efficient in the long term.


> There is continued interested in defining a less verbose syntax.  
> However, the XML Schema WG is very stretched for resources.


If I had the money, I'd send you all millions. XSD is a truly worthy  
endeavor.


Philip Aker
echo astwta@l...@nl | tr a-z@. p-za-o.@

Democracy: Two wolves and a sheep voting on lunch.