last Re: testing for string and number in XSLT 2

>If the data is untyped then ($x castable as xs:string) will always be true.
>If the string consists of digits then ($x castable as xs:integer) will also
>be true, as your experiments show.
>
thx for being patient with me Michael (et al).

synthesizing data from 3 specs (e.g. no mention of 'castable of' or
'instance of' in XSLT 2.0 spec) plus having a dodgy installation of
oXygen which was showing cached' results that was apparantly from prev
transforms .. made me doubt everything (yes trust but verify...should
been running in parallel from commandline).

not such a good thing when trying to understand some slightly subtle (to
me) behavior of untyped data.

I still think having something like a is-number or is-string (and
something to return the type of something) is valid, if only to avoid
getting a type errors...though convinced that this can be a custom
function rather then added to core....its good to know that current XSLT
2.0 has thinking in depth behind its concepts and assumptions.

cheers, Jim Fuller