Templates in XQuery

[Brett Zamir]

  On 9/5/2010 5:22 AM, Martin Probst wrote:
>> There really is a practical and philosophical problem with 2 "competing" XML
>> transformation languages (and yes I 'dare' say xquery is a transformation
>> language).
> I wouldn't agree. To me, XQuery is an XML aggregation language. I
> never do any significant XML transformation work using XQuery.

Outside of the important omission of templates or the handful of 
features from XSL which have no equivalents in XQuery, what 
transformation feature of XSL cannot be expressed in XQuery and more 
succinctly so?

For coders like myself who like to use light-weight programming tools 
without auto-complete on, I find XSL very cumbersome; there's nothing 
wrong with the processing, it's the syntax.  With XQuery, you don't have 
to type namespace prefixes like "xsl:" everywhere for example, as you're 
working with a different syntax. And it's much clearer to read XQuery 
where you can see at a glance what part is output XML and what part is 
processing language. I think this clear visual distinction explains the 
popularity of say templating languages like Smarty in PHP over strictly 
XML-based ones, despite some of the latter's admitted inherent advantages.

Discussions about the "right tool for the job" pertain more to existing 
implementations, rather than language design (and even for 
implementations I see this phrase often over-applied). And as far as the 
consequences for such a proposed change on implementations, while it 
could be handy for the likes of XML databases retooling themselves to 
allow complete transformations via XSL-style templating, by suggesting 
it as an optional feature, I'd be more hoping existing XSL 
implementations could allow swapping of their parsing layer (if it can 
indeed be that simple), giving XQuery syntax as an option, rather than 
forcing existing XQuery implementations to start supporting the 
push-based processing of templates).

This is not to say there is not a valid basis of a best practice 
discussion weighing the value of relying on transformational constructs 
like templates vs. targeted constructs like for...in (or for-each) in 
different contexts, as these are indeed different "tools" suited for 
different situations, but XSL already offers both types, and I see no 
reason why XQuery could not, as a syntax, do the same.

best wishes,
Brett