DuckLake supports schemas inside other schemas. Nested schemas require DuckLake format version 1.1.
Examples
Create each parent schema before creating its children, and use the full schema path to access tables:
ATTACH 'ducklake:nested.ducklake' AS my_ducklake;
CREATE SCHEMA my_ducklake.sales;
CREATE SCHEMA my_ducklake.sales.europe;
CREATE TABLE my_ducklake.sales.europe.orders (order_id INTEGER, amount DECIMAL(10, 2));
INSERT INTO my_ducklake.sales.europe.orders VALUES (1, 25.00);
SELECT * FROM my_ducklake.sales.europe.orders;
Views and macros can also be created in nested schemas.
A quoted name containing a dot, such as my_ducklake."sales.europe", names a single top-level schema and is distinct from my_ducklake.sales.europe.
Use duckdb_schemas() to list schemas and their parents:
SELECT schema_name, parent_schema
FROM duckdb_schemas()
WHERE database_name = 'my_ducklake'
ORDER BY schema_name;
SHOW ALL TABLES includes nested tables, but its schema column contains only the final component of the schema name.
USE and SHOW TABLES FROM do not accept paths containing multiple schema components.
Use fully qualified object names instead.
Functions that take a schema argument accept a dotted path:
CALL ducklake_flush_inlined_data('my_ducklake', schema_name => 'sales.europe');
CALL my_ducklake.set_option('parquet_compression', 'zstd', schema => 'sales.europe');
Schema options apply to tables directly in that schema and are not inherited by child schemas. A nested schema's default directory uses its UUID under the DuckLake data path. The directory layout does not mirror the schema hierarchy.
To drop a schema containing tables, views, macros or child schemas, use CASCADE:
DROP SCHEMA my_ducklake.sales CASCADE;
Dropped schemas remain accessible through time travel while the relevant snapshots and data are retained.