JSON in PLOMID — overview

Mental model#

flowchart LR
    Col["JSONB column"] --> Op["operators: navigate + predicate"]
    Op --> Fn["functions: build + transform + aggregate"]
    Fn --> Tbl["JSON_TABLE: rows"]
    Tbl --> Join["JOIN / GROUP BY / window"]

flowchart LR
    Col["JSONB column"] --> Op["operators: navigate + predicate"]
    Op --> Fn["functions: build + transform + aggregate"]
    Fn --> Tbl["JSON_TABLE: rows"]
    Tbl --> Join["JOIN / GROUP BY / window"]

JSON values flow through ordinary SQL: filter (@>), project (->>), aggregate (json_agg), join on extracted keys, window over time. Missing field → SQL NULL (see NULL).

Quick taste#

CREATE TABLE events (id BIGINT PRIMARY KEY, payload JSONB);
INSERT INTO events VALUES (1, '{"user":{"name":"ada"},"ok":true}');
SELECT payload->'user'->>'name' AS name FROM events WHERE payload @> '{"ok": true}';
SELECT json_agg(payload ORDER BY id) FROM events;

For future VECTOR / GRAPH planners#

Vector embeddings and graph traversals are not in v0.1.0. JSON arrays can hold embedding-like data today ([0.1, 0.2] + jsonb_array_elements), but there are no distance operators or indexes. See Future data models for the explicit firewall.

Next: Operators → Functions → JSON_TABLE → Query JSON guide