[xquery-talk] functions in XQuery and XPath Data Model 3.0

[Michael Kay]

On 25/02/2012 12:14, James Fuller wrote:
> as http://www.w3.org/TR/xpath-datamodel-30/#other-items states the
> xpath data model has matured and needs to add 'other types' ... which
> have no xml serializations
>
> following on from this statement, I am wondering if the XQuery WG has
> considered what could be returned in the following scenario:
>
> let $a := fn:position#0
> return
> fn:string($a)
Actually fn:position#0 isn't allowed - we don't allow function items to 
be formed from functions that access the dynamic context, because 
there's too much scope for confusion about which dynamic context is used.

But if you substituted, say, root#1, we allow you to get the name of the 
function using function-name(root#1) and to get its arity using 
function-arity(root#1). That seems better to me that doing it by 
casting, which seems to rely on the purely accidental fact that all the 
properties of a function have distinct types.

Michael Kay
Saxonica
>
> it maybe nice to return the name of the function "fn:position" as a
> string ... conversely,
>
> let $a := fn:position#0
> return
> fn:QName($a)
>
> would return the qname of the function.
>
> its probably a stretch to extend this to function arity eg.
>
> let $a := fn:position#0
> return
> xs:integer($a)
>
> which would return a value of 0.
>
> it may also be useful to expand the description of what functions are
> in the data model via fully describing the term 'callable' here
> http://www.w3.org/TR/xpath-datamodel-30/#other-items.
>
> James Fuller
> _______________________________________________
> talk at x-query.com
> http://x-query.com/mailman/listinfo/talk
>