RE: [codelist] notes and open questions in requirements

We agreed in today’s teleconference that resolution was needed on some of the requirements that included notes and/or open questions. These may or may not require changes to the requirement, but need to be reviewed to make that determination. I have echoed the affected requirements below:

R14 Metadata can have a particular simple type

Metadata columns can have a defined data type. The W3C XML Schema simple types are the default set, but a different datatype library can be specified (through the use of an identifying URI).

Facets can be defined to restrict the data type (e.g. length, minimum/maximum, or pattern restrictions).

Note: there is an option question around whether it should be possible to define the “type” of complex data (XML fragments) in some sense.

HGM> Not a version 1.0 requirement

R20 A code list can be derived from existing code lists by removing all common columns before aggregating the remaining columns

It should also be possible to remove common keys as well as common columns. A column cannot be removed unless all keys that it is part of are removed, and there aggregate must contain at least one key.

Note: support for keys as well is columns is not yet implemented.

HGM> Sensible requirement, which probably needs to be accommodated in the metadata even if the function is not implemented in 1.0

R21 A derived code list can be required to contain a source code list as a row-wise subset

Note: there is an open question about whether it should be possible to specify that only keys are compared, or that only particular keys are compared.

HGM> Probably better to allow manual selection prior to integration - implementation of the integration service

R22 A derived code list can be required to contain a source code list as a column-wise subset

It should be possible to specify the subset via keys as well as via columns.

Note: support for keys as well is columns is not yet implemented.

HGM> Non-critical

R30 Each code list version has a unique identifier, different to the version-independent identifier for the code list

The code list definition contains the version number (or string) as well as the unique identifier (a URI).

Note: there is an open question around whether metadata-only code lists should be required to have explicit versions, especially for code lists where it is unlikely that the columns and keys will change over time.

HGM> Unlikely not= never, so need the option for specifiying versions

R33 Sets of code list versions can be represented

It must be possible to specify a “configuration” of versions of code lists that together form a coherent set for some purpose.

Note: there is open question around whether it should it be possible to include a version-independent reference to a code list, instead of a reference to a particular version of a code list.

HGM> Essential requirement for reconstructing previous configurations of codelists, eg for contractual reasons.

This post is intended to prompt discussion (using the list) on these items.

Regards,

Jim

=================

Jim Harris

Senior Court Technology Associate

National Center for State Courts

300 Newport Avenue

Williamsburg, VA 23185-4147

757-259-1804 tel

757-564-2048 fax

407-620-2335 cell

jharris@ncsc.dni.us