create table t1(
  ts BIG INT,
  dur BIG INT,
  a BIG INT,
  PRIMARY KEY (ts)
) without rowid;

create table t2(
  ts BIG INT,
  dur BIG INT,
  b BIG INT,
  part BIG INT,
  PRIMARY KEY (part, ts)
) without rowid;

-- Insert some rows into t2 which are in part 0 and 1 but before t1's rows.
INSERT INTO t2(ts, dur, part, b)
VALUES
(0, 25, 0, 111),
(50, 50, 0, 222),
(0, 40, 1, 333);

-- Then insert some rows into t1 in part 1, 3, 4 and 5.
INSERT INTO t1(ts, dur, a)
VALUES
(100, 400, 111),
(500, 50, 222),
(600, 100, 333),
(900, 100, 444);

-- Insert a row into t2 which should be split up by t1's first row.
INSERT INTO t2(ts, dur, part, b) VALUES (50, 200, 1, 444);

-- Insert a row into t2 should should be completely covered by t1's first row.
INSERT INTO t2(ts, dur, part, b) VALUES (300, 100, 1, 555);

-- Insert a row into t2 which should span between t1's first and second rows.
INSERT INTO t2(ts, dur, part, b) VALUES (400, 250, 1, 666);

-- Insert a row into t2 in partition 2.
INSERT INTO t2(ts, dur, part, b) VALUES (100, 1000, 2, 777);

-- Insert a row into t2 before t1's first row in partition 4.
INSERT INTO t2(ts, dur, part, b) VALUES (50, 50, 4, 888);

-- Insert a row into t2 which perfectly matches the second row in partition 4.
INSERT INTO t2(ts, dur, part, b) VALUES (500, 50, 4, 999);

-- Insert a row into t2 which intersects the first row of partition 5.
INSERT INTO t2(ts, dur, part, b) VALUES (50, 75, 5, 1111);

-- Insert a row into t2 which intersects the third row of partition 5.
INSERT INTO t2(ts, dur, part, b) VALUES (525, 75, 5, 2222);

-- Insert a row into t2 which misses everything in partition 6.
INSERT INTO t2(ts, dur, part, b) VALUES (0, 100, 6, 2222);

create virtual table sp using span_left_join(t1, t2 PARTITIONED part);

select * from sp;