SAX Parser and Attribute Normalization

A real XML question (HEY!!):

I've read the XML 1.0 specification 
(actually, Bray's annotated version, thanks Tim!) 
and understand the rules for attribute normalization 
using an XML Processor (or I hope I do... lotsa rules), 
but when using a SAX parser, should the parser 
strip trailing white space in an attribute 
value?  Empirical testing says no.

I'm not a good skater either. Sigh...

len