RE: Casting/constructors should accept the empty sequence

> Scary, but I agree with Michael (Rys).
>     Jim
> 
> At 12:26 AM 6/1/2004 Tuesday, Michael Rys wrote:
> 
> >I don't like xs:int?().
> >
> >
> >I think using
> >
> >Expr cast as T
> >Expr cast as T?
> >
> >And T(Expr) as a synonym for the later quite acceptable and 
> do not see a
> >reason to change.

I have concerns about making this change. If a function is declared to
accept an argument declared as "$i as xs:integer", then I don't think
calling it with xs:integer(foo) should ever fail with a type error. Our
normal convention is that when a type name is used with no occurrence
indicator, it means "exactly one".

Michael Kay 

Received on Wednesday, 9 June 2004 05:16:48 UTC