Fail to compute pipeline saying Timevector containing nulls

[jonatas]

Relevant system information:

• OS: Mac OS - (M1 2020 processor)
• PostgreSQL version: PostgreSQL 14.4 on x86_64-apple-darwin21.5.0, compiled by Apple clang version 13.1.6 (clang-1316.0.21.2.5), 64-bit
• TimescaleDB Toolkit version: │1.9.0-dev
• Installation method: "source"

Describe the bug
I'm trying to benchmark and compare how the lambdas are performing against some language-based processing code. I got an error when I introduce a few thousand records.

To Reproduce
Here is a small script that can create a scenario to test:

CREATE TABLE "measurements" ("device_id" integer not null, "val" decimal not null, "ts" timestamp not null);
SELECT create_hypertable('measurements', 'ts', chunk_time_interval => INTERVAL '1 day');

-- Generating some data
INSERT INTO measurements (ts, device_id, val)
  SELECT ts, device_id, random()*80
 FROM generate_series(TIMESTAMP '2022-01-01 00:00:00',
        TIMESTAMP '2022-02-01 00:00:00', INTERVAL '1 second') AS g1(ts),
      generate_series(1, 3) AS g2(device_id);

set search_path to toolkit_experimental, public;


-- Validating if sort is generating some null value
WITH a as (
  SELECT device_id, (timevector(ts, val) -> sort() -> unnest()).*
  FROM "measurements"
  GROUP BY device_id)
select a.* FROM a WHERE a.time is null or a.value is null;

-- Trying only to compute delta
SELECT device_id, timevector(ts, val) -> sort() -> delta()
FROM "measurements"
GROUP BY device_id ;

-- Limiting scope to make it easy to compute 
SELECT timevector(ts, val) -> sort() -> delta() -> abs() -> sum() as volatility
FROM "measurements"
WHERE device_id = 2
LIMIT 10;

Expected behavior
I was expecting to compute deltas as I have no null values in my table either timevector.

Actual behavior
The following SQL:

SELECT device_id, timevector(ts, val) -> sort() -> delta()
FROM "measurements"
GROUP BY device_id ;

returns an error saying

ERROR:  Unable to compute deltas over timevector containing nulls
CONTEXT:  extension/src/time_vector/pipeline/delta.rs:35:9

I have tested several scenarios trying to identify where is the problem, and it seems related to managing a large number of records.

Changing the generate_series to 2 minutes interval between records will generate 66963 records and it works fine on my computer but If I move it to 1 minute it will generate 130k records and then the error pops up. I tried to also increase the number of devices but from 3 to 4 devices it already fails.

ERROR:  Unable to compute stats aggregate over timevector containing nulls
CONTEXT:  extension/src/time_vector/pipeline/aggregation.rs:112:9