mirror of
https://github.com/scylladb/scylladb.git
synced 2026-04-29 20:57:00 +00:00
It turns out that Cassandra handles a restriction like `(c2) = (1)` just like `c2 = 1`, and is not limited like multi-column restrictions. In particular, this query works despite missing "c1", and may also use an index if c2 is indexed. But currently in Scylla, `(c2) = (1)` is handled like a multi-column restriction, so complains if c2 is not the first clustering key column, and cannot use an index. This patch adds several tests demonstrating this difference between Scylla and Cassandra (#13250). The xfailing tests pass on Cassandra but fail on Scylla. Refs #13250 Signed-off-by: Nadav Har'El <nyh@scylladb.com> Closes #13252