Discuss source, catalog, and collapse

[jojenki]

The idea is that we add a new field to the metadata that is "source", which indicates a different copy of the same device for the same person. For example, if someone were wearing two Fitbits, the format of the data would be the same and the streams could be aggregated, but, to distinguish them, you would need an extra piece of metadata that dictated which "source" a data point came from.

The "catalog" is a way to get the set of sources. Assuming the "source" is added to the metadata and that the "data path" is extended to drill down into pieces of data, this can be achieved by the following path: "/omh/v1/[id]/[version]/data/metadata/source".

The problem with this solution is that each index in the resulting array is its own data point, whereas the "catalog" should not return duplicates. This can be solved by adding a "collapse" parameter that will prune out duplicates, ala "distinct" in SQL. This can be a more generic parameter when people want to know only the set of some values rather than every point's value.