[Bug] DataFusion UPDATE and DELETE can affect more rows than LIMIT allows #1396

Search before asking

• I searched in the issues and found nothing similar.

Please describe the bug 🐞

DataFusion SQLContext drops ORDER BY and LIMIT when selecting rows for UPDATE/DELETE, affecting both Copy-on-Write and Data Evolution tables.

For example, run through SQLContext with the paimon catalog:

CREATE SCHEMA paimon.dml_limit_repro;
CREATE TABLE paimon.dml_limit_repro.t (id INT);
INSERT INTO paimon.dml_limit_repro.t VALUES (1), (2), (3);
DELETE FROM paimon.dml_limit_repro.t ORDER BY id LIMIT 1;
SELECT * FROM paimon.dml_limit_repro.t ORDER BY id;

Expected: only id=1 is deleted. Actual: all three rows are deleted. UPDATE has the same issue;

Solution

Preserve ORDER BY and LIMIT in the internal selection queries.

Are you willing to submit a PR?

• I'm willing to submit a PR!