Re: [xsl] Specification of a transform.

[davep <davep@xxxxxxxxxxxxx>]

On 10/09/13 09:09, Wolfgang Laun wrote:
> There is a substantial subset among the "expressive power" between
> XPath and the language of Rule Based Systems, especially those that
> are embedded in an object-oriented environment. "Constraint
> programming" would be the common denominator. Thus, an XSLT template
> works quite similar to a rule (although rules appear to be easier to
> write for data that is organized like a relational DB than that in a
> deep hierarchy).


Which might suggest trying a rules language within a template - possibly
building on Kens idea. Template for 'where', rules for constraints,
checking etc.

> Thus, rules can express the data analysis side of a transformation
> very well. One system I know very well (Drools) also features a
> technique for implementing a Domain Specific Language of your design,
> with a translator to the native rule language.
>
> Production Rule Systems let you also specify actions, to be executed
> for each match on the data, and this would normally be used to define
> the synthesis part, but since this is procedural it does not map so
> well to the functional approach, especially when you look at the way a
> "pull" transformation is designed.

Again suggesting a mix of ideas.

Thanks Wolfgang.








regards

--
Dave Pawson
XSLT XSL-FO FAQ.
http://www.dpawson.co.uk