>>> [log in to unmask] 2004-11-29 14:50:15 >>>
> . We don't want to publish schemas that don't validate or let you
> instances properly (unless we can document that it is strictly an
> bug). I hope other people might have some insight into this.
It's definitely a XMLSpy bug. See the second paragraph following
the table at http://www.w3.org/TR/xmlschema-1/#element-element. If
<xsd:element> is a child of <xsd:schema> it declares a global element,
no matter the type.
Some other observations:
The binding of prefix "ns1" to the xml namespace
http://www.w3.org/XML/1998/namespace must be removed; see erratum NE05
<mads> needs an ID attribute; there is currently no way to identify
the authority record, as opposed to a name or titleInfo inside the
Since related records are authoritative, <related> should have
optional XLink attributes to link to another MADS record in lieu of
duplicating the content.
The "descriptor" type should support the MODS language attribute
group (lang, xml:lang, transliteration, script), so that e.g. <variant
type="translation" lang="fre">... is valid.
If the content of <address> were defined with <xsd:all> instead of
<xsd:sequence>, the native order of address parts could be preserved
(e.g. street-city-state-postcode in the US,
postcode-prefecture-city-street in Japanese). Also, <address> doesn't
need more than one <street> subelement.