VODataService

VODataService is a XML Schema that extends the core VOResource schema to describe data collections and services that provide access to data. It forms the basis for extension schemas that describe support for standard IVOA protocols.

Internal Working Draft Document

VODataService: a VOResource Schema Extension for Describing Collections and Services
in progress

Internal Working Draft Schema

VODataService-v1.1wd3.xsd

Working Draft Schema

VODataService-v1.0.xsd

Features

VODataService defines the follow extensions of the generic Resource type:

• DataCollection: describes a collection of data which can include a mix of images, catalogs, spectra and any other type of data. A collection description may include a description of coverage on the sky and in time and frequency, and it may include a description of the component tabular data.
• DataService: This is an extension of the Service Resource type that represents a service that provides access to data. One can describe the data's coverage on the sky and in time and frequency and its association with facilities and instruments.
• CatalogService: This is an extension of the DataService type to describe a service that provides access to data, and specifically interacts with or returns information in a tabular form. In addition to the DataService metadata, one may also describe the strucuture of the tables.
• STCStandard: This resource extension is used to define standard coordinate systems using STC.

It also defines an extension of the generic Interface type:

• ParamHTTP: this describes a service interface in the form an HTTP Get URL and comprised of a base URL and a set of arguments made up of key=value parameters. In this extension, it is possible to describe the input parameters supported by the service.

Some samples

Version 1.1 (proposed)

• collection.xml
• catalog.xml
• catalogservice.xml
• stc.xml

Version 1.0 (currently in use by IVOA registries)

• collection.xml
• catalog.xml
• stc.xml

Proposed Changes

These changes follows on discussion of changes from InterOpSep2007ResReg and InterOpMay2008ResReg and is recorded in part below under "Previous Discussion of Proposed Changes" below. This proposed version attempts to incorporate feedback on the last proposal. Here is a summary of the changes:

1. The descriptions of tables is done via the tableset element
   
   The tableset element aggregates a set of logically related tables with no implied relationship to database catalogs or schemas. It responds to several issues regarding the hierarchical organization of the table descriptions:
   • • "catalog" is an overloaded term, with different connotations in astronomy and database technology; tableset replaces the catalog element in attempt to shed connotations from both domains. In contrast, the term "table set" has a direct connection to the jargon behind the development of TAP and is more directly appropos.
     • To make searching more straight forward, it is considered important that all table elements are addressable via a single simple XPath location (e.g. tableset/table).
     • The introduction of tableset does not address the issue of backward compatibility with VODataService v1.0 in which the CatalogService resource type lists tables at one level higher--i.e. there is no tableset level. This proposal takes the position that the table set organization is more important than this backward compatibility. Consequently, it would be helpful to users to migrate uses of v1.0 to v1.1 as broadly and quickly as possible.

2. tableset/name and tableset/table/qualifiedName elements are required and must be unique.
   
   NVO portal applications discovered that it is important to have unique ways of identifying tables.

3. tableset/table/qualifiedName should include database catalog and/or schema qualifiers as applicable.
   
   This allows the name to be plugged directly into an ADQL query without further construction from separately tagged schema and catalog names. A dot-delimiting syntax for the name would allow for parsing out the catalog and schema if is desired, though in most cases, this will not be necessary.

4. the TableService is dropped.
   
   This proposal posits that this is unnecessarily redundant with CatalogService. There are currently no TableService resources registered, suggesting there is little need to distinguish between a service with coverage one without (noting that coverage is optional).

5. the function and join elements are dropped from the last proposal
   
   
   • • There was the prudent suggestion (by DougTody and supported by others) that these need further prototyping before we lock them into a schema that would be difficult to revise later. (McGlynn noted a similar join feature in HEASARC services that proved not to be useful in practice.) Instead extension mechanisms will provide a way to add these in the future.

6. Table description extension mechanisms: xsi:type and extended types.
   
   Richer descriptions of tables will be enabled in two ways:
   • • an xsi:type attribute can added to <tableset> or <table> to allow additional elements to be added. (Note you can pretty much do this anywhere in a VOResource document.)
     • the <dataType> elements feature optional attributes, extendedType and extendedSchema to indicate a column or parameter has an extended type. When present, the element value is the type to assume if the extended type is not recognized or supported.

7. the <table> element's attribute role has been changed to type; added documentation for the recognized value "view".

Previous Discussion of Proposed Changes

Proposed Changes

These changes were proposed in Cambridge and were motivated by recent development work on the Simple Spectral Access (SSA) Protocol. The presentation here reflects the feedback provided during the Registry Working Group Session in Cambridge.

1. Add an optional <testArgs> to the ParamHTTP Interface type
   • Definition: a formatted set of arguments that, when applied to the access URL, can be used to test the service.
   • this mechanism is useful regardless of whether the service complies with a standard or not.
   • the format is a string such that when concatonated after the access URL, it produces a legal URL.
   • the arguments should be chosen such that the resulting query returns a legal, non-empty result.
2. Add an optional <rights> element to the DataService Resource type. DROPPED
   • This proposal has been dropped. When proposed, this was thought to be missing from DataService; however, it is included in the definition of the more general Service Resource type.
3. Add an optional <dataset> element to the DataService and DataCollection Resource types. DROPPED
   • provides a listing of Dataset identifiers that are available for use as input to the service
   • This proposal is being dropped as it does not address the original intended use of dataset IDs within SSA.

This change was motivated by recent work on TAP and VOSI.

1. Add extra elements to describe use of tables and databases.
   • Extra types are RelationalSchema, Function, RelationalJoin.
   • Two possible schema: one with mandatory nesting table in schema in catalog; one with the structure partially flattened.
   • XML schema with deeper nesting
   • XML schema partially flattened
   • The flatter schema allows all resources that validated against VODataService 1.0 to validate against VODataService 1.1.