Re: MS-FO-LC1-017: cast untypedAtomic to other type with arithmetic and compare ops.


> Why don't we cast untypedAtomic to other type instead of directly to
> xs:double and use xs:double only as fallback for arithmetic and
> compare?
> This would mean that an expression such as xs:untypedAtomic(1)+1
> results in an integer(2) instead of a double(2.0e0) as it does now.
The problem is that it would also mean that xs:untypedAtomic('2.5')+1
resulted in a dynamic error, since the untyped atomic value '2.5'
isn't a legal lexical representation of xs:integer. I think that this
would be disconcerting.



Jeni Tennison

Received on Tuesday, 1 July 2003 04:53:20 UTC