RE: Representation of floating point values.

It seems to me (I am not speaking on the behalf of the XQuery WG) that
only allowing mapping it to Infinity may be not sufficient.

There is an explicit way to construct IEEE754's Infinity.   It is not
the same as having a too large number - and many applications using
schema data types may want to know that this has happened.

For this reason, numerical operations in XQuery allow for all three
options - mapping to a largest value, to Infinity and raising an error.
Same for underflow.

>From my personal experience in dealing with a variety of scientific data
formats not having an option to detect a non-representable value is not
particularly user-friendly.

Is there a way to phrase the Schema in a way as to allow the "hosting"
application to choose how to deal with overflow/underflow?

Daniel;


-----Original Message-----
From: Dave Peterson [mailto:davep@iit.edu] 
Sent: Monday, March 07, 2005 4:56 PM
To: Ashok Malhotra; Michael Kay; Daniel Engovatov;
www-xml-schema-comments@w3.org
Cc: w3c-xml-query-wg@w3.org
Subject: RE: Representation of floating point values.

I always thought that Schema 1.0 would produce an infinity, but you've
convinced me we didn't write it that way.  Probably by accident.  I
suspect that if Query has druthers, Schema would consider fixing it,
at least in 1.1.
-- 
Dave Peterson
SGMLWorks!

davep@iit.edu