All,
Oven in the "JSON[B] arrays are second-class citizens" thread [1] I made the observation that the only way to get a PostgreSQL array from a JSON array is via the "elements->cast->array_agg" chain. For JSON arrays that are homogeneous in nature the capability to go "directly" from JSON to json[], text[], bigint[], etc... seems like it would have some value.
A couple, so far, of community members have chimed in. Maybe this ends up on a ToDo somewhere but for the moment I figured I'd float the idea on its own thread since the other one is really about a different (though somewhat related) topic.
In the spirit of the json_populate_record like functions something with the following signature seems usable:
as_array(anyarray, jsonb) : anyarray [2]
so that actual calls look like:
SELECT as_array(null::text[], '["a", "b", "c"]'::jsonb)
For better or worse while every table gets a corresponding composite type explicitly created types cannot be treated as row types in this situation
SELECT jsonb_populate_record(null::text[], '["a", "b", "c"]'::jsonb)
> ERROR: first argument of jsonb_populate_record must be a row type
Loosening the restriction and allowing jsonb_populate_record to fulfill the role of the above described "as_array" function would be something to consider, but likely not possible or worth any effort in trying to make happen.
David J.
[2] ROTQ[3] - whose idea was it to put the type first and the data second?
[3] Rhetorical Off-Topic Question