Optimisation by storing json in jsonb columns instead of text columns

[hylkevds]

Currently FROST stores all JSON content in text columns, meaning that to filter on it, PostgreSQL has to parse the text to json each time. This can be improved by using jsonb columns.

[hylkevds]

Interestingly, some queries are slower on JSONB columns than on text columns. It seems that this is because DISTINCT is slower on JSONB columns than on TEXT columns. If I take the JSONB column out of the result set of such a query, by doing a $select=name,id, then the result is a lot faster:

Flat Things; Top; Filter DSTime, 1DepId, ResultCode;                  3773
Flat Things; Top; Filter DSTime, 1DepId, ResultCode; Select name,id   1446

This makes sense, since comparing text is easy. Comparing json trees is a lot more involved.