RETURN and FINISH

RETURN projects expressions into an Arrow result table. A variable keeps its name; give computed expressions an explicit AS alias to make result schemas stable. Otherwise scalar expressions receive names such as column_1. A comparison of a property that has different non-numeric types uses each stored type, not the unified text column.

A comparison of two properties in RETURN, SELECT, HAVING, or ORDER BY uses each fragment’s stored type, just as WHERE does.

MATCH (p:Person)
RETURN p.name AS name, p.age + 1 AS next_age
ORDER BY name;

RETURN ALL preserves duplicate rows and is the default. RETURN DISTINCT removes duplicate projected rows. RETURN * expands available bindings, including whole elements where present. An INSERT that declares a node or edge variable exposes that variable to RETURN *. In a later NEXT statement it expands the incoming working table, including after FILTER, ORDER BY, or LIMIT.

FOR n IN [1, 1, 2] RETURN DISTINCT n ORDER BY n;

Use FINISH to complete a pipeline without returning its working bindings. Write statements with no explicit result have an implicit FINISH; the CLI reports their affected-element count and commit sequence.

MATCH (p:Person {name: 'Alice'}) SET p.age = 31 FINISH;

Inside an explicit transaction, result rows describe the pending transaction snapshot. They do not acknowledge durability until COMMIT succeeds. See result handling.

Arithmetic and concatenation on a mixed property use each source fragment’s stored type. A fragment that cannot apply the operator contributes null. Wrappers such as COALESCE can supply a fallback for that null.

DISTINCT rejects values that combine incompatible stored families, such as numbers and text. The check follows scalar aliases, including those expanded by RETURN *. Integers and floats remain comparable, and nulls do not add a family.