This page describes what is supported in DuckDB and DuckLake in relation to DuckDB standalone (i.e., :memory: or DuckDB file modes). We can make a distinction between:

• What is currently not supported by the DuckLake specification. These are features that are supported by DuckDB when using DuckDB's native database format but will not work with a DuckLake backend since the specification does not support them.

• What is currently not supported by the ducklake DuckDB extension. These are features that are supported by the DuckLake specification but are not (yet) implemented in the DuckDB extension.

Unsupported by the DuckLake Specification

Within this group, we are going to make a distinction between what is not supported now but is likely to be supported in the future and what is not supported and is unlikely to be supported.

Likely to Be Supported in the Future

The following features are currently not supported in DuckLake but are likely to be supported in future versions:

• User defined types.

• Fixed-size arrays, i.e., ARRAY type

• ENUM type

• CHECK constraints (not to be confused with Primary or Foreign Key constraints)

• Generated columns

Unlikely to Be Supported in the Future

The following features are currently not supported in DuckLake and are unlikely to be supported in future versions:

• Indexes

• Primary key or enforced unique constraints and foreign key constraints are unlikely to be supported as these are prohibitively expensive to enforce in data lake setups. We may consider supporting unenforced primary keys, similar to BigQuery's implementation.

• Upserting is only supported via the MERGE INTO syntax since primary keys are not supported in DuckLake.

• Sequences

• VARINT type

• BITSTRING type

• UNION type

Unsupported by the DuckDB Extension

• DROP TABLE ... CASCADE is not supported. DROP SCHEMA ... CASCADE can drop a schema and its contents.

• VARIANT columns nested inside another SQL type, such as STRUCT(v VARIANT) or VARIANT[], can be inlined but cannot be written to Parquet or flushed.

• ALTER TABLE ... ADD COLUMN only supports literal defaults. Expression defaults are supported in CREATE TABLE and ALTER TABLE ... SET DEFAULT:

  -- This is allowed
  CREATE TABLE t1 (id INTEGER, d DATE DEFAULT '2025-08-08');
  
  -- This is also allowed
  CREATE TABLE t2 (id INTEGER, d DATE DEFAULT now());
  

See schema evolution for changing defaults and nested fields.