Resource schema enumeration #8
Description
There are cases where enumerating values in the resource schema is cumbersome and where it doesn't accurately reflect how data is being brought in to the datapackage.
An example might be commuting-and-residence where place of residence and work are column whose value enumeration should come from our default town list. Calculating dimension_groups is tricky because we don't want to have to enumerate the full list.
A few possible solutions:
- Modify the syntax of dimension groups to take a list of objects with an optional count.
- Modify the syntax of resource schema parsing to leverage the foreign key properties.
- Modify the testing plugin to bootstrap values from the local value and use those to override defaults in the plugin.
Of these three, 2 seems like the most reasonable. It is compliant w/ the Frictionless paradigm. It extends our tool in a way that brings it closer to the schema specs vs. distancing it. It doesn't introduce a breaking change to the exposed fixtures. The process of handling look ups will be abstracted away in the implementation and the end result will be fixtures that return the same information as is currently provided (list of combinations in the case of dimension_groups).