rollup()

Combine multiple timevectors into a single timevector for re-aggregation

Early access 1.3.0

Combine multiple pre-constructed timevectors into a single timevector. This is useful for re-aggregating timevectors that were created separately, such as when grouping by different dimensions or combining data from multiple sources.

Samples

Combine timevectors from different groups

Aggregate per-device timevectors into a single combined timevector:

CREATE TABLE samples(time TIMESTAMPTZ, batch INT, data DOUBLE PRECISION);

-- Create timevectors per batch
CREATE VIEW batch_series AS
    SELECT batch, timevector(time, data) AS batch_series
    FROM samples
    GROUP BY batch;

-- Combine all batches into one timevector
SELECT rollup(batch_series)
FROM batch_series;

Re-aggregate across partitions

Combine timevectors from different time ranges:

CREATE VIEW daily_series AS
    SELECT
        date_trunc('day', time) AS day,
        timevector(time, temperature) AS series
    FROM sensor_data
    GROUP BY 1;

-- Get weekly timevector from daily timevectors
SELECT
    date_trunc('week', day) AS week,
    rollup(series) AS weekly_series
FROM daily_series
GROUP BY 1;

Arguments

The syntax is:

rollup(
    series Timevector
) RETURNS Timevector

Name	Type	Default	Required	Description
series	Timevector	-	✔	Previously constructed timevector objects to combine

Returns

Column	Type	Description
rollup	Timevector	A timevector combining all the underlying series