list_reduce can use column values from another row

What happens?

Using list_reduce with a constant initial value and a column reference inside the lambda can return the wrong result when the lists have different lengths. The lambda can read the column value from another row. With a filtered input, it can also fail with an INTERNAL Error.

To Reproduce

SELECT v, list_reduce(l, lambda acc, x: acc + x + v, 0) AS result
FROM (VALUES ([1], 10), ([1, 2], 30)) t(l, v);

Actual result:

v   result
10  11
30  43

The second result should be 63: (0 + 1 + 30) + 2 + 30. Instead, the last step uses 10 from the first row.

Adding a filter can also produce an INTERNAL Error:

SELECT v, list_reduce(l, lambda acc, x: acc + x + v, 0) AS result
FROM (VALUES ([], 0), ([1], 10), ([1, 2], 30)) t(l, v)
WHERE v > 0;

INTERNAL Error: DataChunk::SetChildCardinality - vector has size 2 but expected size 1 - and cannot change vector size because it is not a flat or constant vector

OS:

Ubuntu 22.04.5 LTS, x86_64

DuckDB Version:

v2.1.0-dev84749 (2bbfe6f236)

DuckDB Client:

CLI

Hardware:

No response

Full Name:

Ilya Mishurov

Affiliation:

Postgres Professional

Did you include all relevant configuration (e.g., CPU architecture, Linux distribution) to reproduce the issue?

• Yes, I have

Did you include all code required to reproduce the issue?

• Yes, I have

Did you include all relevant data sets for reproducing the issue?

Yes