Default values handling

The DM document specifies that elements and attributes use the [schema
normalized value] when constructing nodes from the PSVI. This means that
default values specified in the schema are used to synthetized attributes
that are not present in the original infoset. This is clearly needed when
performing queries on the data model. However, the DM does not seem to keep
track of the [schema specified] property, which means that attributes that
are absent in the original infoset will be copied and serialized as if they
were present. Is this a proper behavior when dealing with documents that
rely heavily on schema-specified default values (e.g. an XML Schema
document)? When using a possible "preserve" validation mode previously
mentioned by Don Chamberlin, copying and serializing default values is in
fact redundant, because the information is already present in the original
type (which is preserved).

Best regards,

Antoine Mensch