Seen on Windows XP with NB 55 060929 and JDK 1.5.0_07:

I see inconsistencies in how different ways of importing schemas behave, in
terms of what the user sees in the graphical tree view and what is generated in
the WSDL source. It would be good for the HIE folks to look at this sometime.

In all cases, start with a new WSDL documented created using the WSDL wizard
with default values. Have a few .xsd schema files in the same project as the
WSDL document.

Adding a referenced schema from root document or Types node:

1. Select root node or Types node, Add -> Import Schema. 
==> Note that a default prefix is specified.
2. Choose the schema and click OK.
3. Click the new Referenced Schema element. 
==> Note that in the properties pane, the prefix is "null".
4. Look at the source.
==> Note that the prefix is added to the Definition element.
5. Delete the referenced schema.
==> Note that the prefix definition is NOT removed from the Definition element.
==> If you delete and add the same schema multiple times, the Definition element
contains multiple prefix definitions for the same schema namespace.

Add an import from Referenced Schemas element:
1. Select Referenced Schemas element. Add -> Import.
==> Note that no default prefix is specified.
2. Choose a schema, specify a prefix, and click OK.
==> Note that in the properties pane, the prefix is as specified.
3. Look at the source.
==> Note that the prefix is added to the schema element.
4. Delete the added schema.
==> Note that the prefix is not deleted from the schema element.

==> If you mix the two:
1. Add a schema via the referenced schema with a prefix (prefix added to schema
element in source).
2. Delete the referenced schema.
3. Add the same schema again via the Types or root node. Specify a default prefix.
==> The new prefix is added to the Definition element, but in the properties
pane for the referenced schema, the Prefix property refers to the name on the
Schema element, not the prefix the user just specified.