Changes to Client Catalog schema

Changes for version 1.0.7

• bring DataFormatTypes and ServiceTypes enumerations up to date with the code

Changes for version 1.0.2

Changes to enumerated types:

• "serviceType" enumerations: deprecated "NetcdfServer"; added "NetcdfSubset".
• "dataFormatTypes" enumerations: added "McIDAS-AREA" and "text/html".
• "variableNameVocabulary" enumerations: deprecated "GRIB"; added "GRIB-1" and "GRIB-2"
  

Changes to elements and attributes:

• "catalog" element: deprecated the "base" attribute (not working in code, should follow XLink spec).
• "variable" element: added the "vocabulary_id" attribute.
• "property" element: require both "name" and "value" attributes.
  

Changes for version 1.0.1

• The version attribute is "1.0.1" .
• The schema can be found here. However, you should tell your parser to resolve to a local copy of it, both for performance and to decouple with Unidata's web server.

Catalog base attribute

The optional attribute base can be added to the catalog element to define a base URL. All relative URLs in the document are interpreted against base URL. This follows the XML Base specification, which in turn references the specification of URIs.

DatasetType type

The dataset definition is now made in the DatasetType type, and catalogRef and datasetScan extend from it and are in the same substitution group. This means that you can use a catalogRef and datasetScan wherever a dataset is allowed. This simplifies the schema slightly. There are a few important effects of this:

1. catalogRef now is-a dataset, and can have any and all metadata that a dataset can have.
2. catalogRef must have a name attribute, since that is required on a dataset. The name is ignored, in favor of the xlink:title attribute. This is the only (slightly) backward incompatible change, and only affects validation.

metadata element

1. process other namespaces with "lax" attribute, meaning validate if a schema can be found

timeCoverageType type

1. some technical corrections to simplify the schema, no semantic changes.

variables element

1. vocabulary attribute is now optional on a variables element
2. vocabulary_name attribute is now optional on a variable element
3. the variable element is now mixed content, meaning you can add text content, which is interpreted as a description of the variable.

Enumeration changes

All enumeration values should be tested case-insensitive.

1. collectionTypes: add ForecastModelRuns
2. dataTypes: add Point, Radial
3. dataFormatTypes: add NEXRAD2, NIDS, GINI
4. dateEnumTypes: add metadataCreated

Other requests

• Allow a variable element to map to more than one vocabulary (Luca)
• Add last update time of the catalog, for caching. Could use HTTP headers. (Benno, LEAD).
• Extended inheritance across catalog references (Michael Burek)

Comments to THREDDS mailgroup