parsing dates

[Michael Kay]

> 
> I think that's not quite correct. Typeswitch case clauses 
> expect a SequenceType, which includes ItemType. So the tests 
> can be any node type or any schema type (see 
> http://www.w3.org/TR/xquery/#id-typeswitch and 
> http://www.w3.org/TR/xquery/#doc-xquery-SequenceType).
> 
> Here's an example: note that the order of the 'case' clauses 
> is important. Since xs:ID is derived from xs:string, it must 
> be tested before any xs:string test case.
> 
> xquery version "1.0-ml";
> 
> let $i := "test"
> return
>    typeswitch ($i)
>     case xs:integer return "integer"
>     case xs:ID return "id"
>     case xs:string return "string"
>     default return error((), "UNEXPECTED", $i)
> 

But we aren't starting with an xs:date or xs:gYearMonth; we are starting
with a string that is capable of being converted to an xs:date or
xs:gYearMonth. So "castable as" is indeed the right solution.

Except that matches() would be more flexible, because it can also test for
formats that don't happen to correspond to any known schema-defined type.

>Is that because xs:integer is considered an AtomicType?

xs:integer is most definitely an AtomicType, there's no "considered to be"
about it.

Regards,

Michael Kay
http://www.saxonica.com/
http://twitter.com/michaelhkay