RE: Representation of floating point values.

>Doesn't sound useful enough to force the backward
>incompatability with 1.0.

Changing the mapping from largest/smallest value to Infinity is a
similar change, is not it?

>This is much like the situation with comparing dateTime values, one
with
>a timezone and one without.  

Not sure it is a similar situation.  There is no silent mapping and no
loss of information in this case.  

Oh, well.  I guess those who care about large numbers will just find a
way to bypass this schema limitation.

Daniel;