-----------------------
- The date and time types now check their initial values when the type
is created, not when the data is marshalled.
- The date and time types are now parsed and returned as tuples (for
multi-element types) or scalars (for single element types) in UTC and thus
can represent the entire range of SOAP dates.
- If an element doesn't have a type but has a name with a namespace, the
name is tried as the type.
- Untyped compound types with more than one element and all the elements
the same name are turned into an array when parsing.
- When parsing a structType, elements with the same name are placed in a
list instead of saving just the last one. _getItemsAsList can be used to
get an element of a structure as a list, whether there was one or many
occurances of the item.
- Added schemaNamespace, schemaNamespaceURI, and namespaceStyle
configuration options. namespaceStyle takes one of 1999, 2000, or 2001,
and sets typesNamespace, typesNamespaceURI, schemaNamespace, and
schemaNamespaceURI.
- Normalized the type class names, replacing Compound with compoundType,
Struct with structType, Header with headerType, Body with bodyType, Array
with arrayType, TypedArray with typedArrayType, Fault with faultType, and
urType with anyType.
- Attributes now appear on an element itself instead of the element's
parent. For elements parsed to builtin python types, the attributes are
stored in a dictionary keyed by the element's python id. The dictionary
is in the Context object, can be returned from parseSOAP*, and can be
returned from method calls if the returnAllAttrs configuration option
is set.
- isinstance is used to check for a class, so classes can be subtyped.
- An encoding of None can be specified to not include encoding information.
- Problems with the SOAPProxy URL are now reported when the SOAPProxy
instance is created instead of when the first method call is made.
- The Binary, Boolean and DateTime types have been removed in favor of
binaryType, booleanType, and dateTimeType.