Support Matrix¶

Compatibility and feature support for sqlalchemy-cubrid releases.

Version Compatibility¶

SQLAlchemy Version	Status	Notes
2.0.x	✅ Supported	Minimum required version
2.1.x	✅ Supported	Latest tested
≥ 2.2	❌ Not supported	Code uses private SA internals (see below)
< 2.0	❌ Not supported	SA 1.x API removed

Why <2.2? The dialect accesses private SQLAlchemy APIs that may change without notice:

Private API	Location	Usage
`select._limit_clause`	`compiler.py:81`	LIMIT clause compilation
`select._offset_clause`	`compiler.py:82`	OFFSET clause compilation
`select._for_update_arg`	`compiler.py:71`	FOR UPDATE clause
`coercions._is_literal`	`compiler.py:144`	Literal value detection
`BindParameter._with_binary_element_type`	`compiler.py:150-151`	Binary parameter handling

These will be migrated to public APIs when SA 2.2 is released.

Driver	Install	URL Scheme	Status
CUBRID-Python (CCI)	`pip install "sqlalchemy-cubrid[cubrid]"`	`cubrid://`	✅ Supported
pycubrid (Pure Python)	`pip install "sqlalchemy-cubrid[pycubrid]"`	`cubrid+pycubrid://`	✅ Supported

Feature	Status	Notes
`create_engine()`	✅	Both `cubrid://` and `cubrid+pycubrid://` schemes
SQL compilation	✅	SELECT, INSERT, UPDATE, DELETE, JOIN, subqueries
DDL compilation	✅	CREATE TABLE, ALTER, DROP, AUTO_INCREMENT, COMMENT
Type system	✅	All CUBRID types mapped (see below)
Schema reflection	✅	Tables, columns, PKs, FKs, indexes, unique constraints, comments
Transaction management	✅	commit, rollback, savepoint (no RELEASE SAVEPOINT)
Connection pooling	✅	SA pool with `pool_pre_ping`, disconnect detection
Statement caching	✅	`supports_statement_cache = True`

Feature	Status	Notes
Auto-discovery	✅	`alembic.ddl` entry point
Schema migrations	✅	CREATE, ALTER, DROP
Autogenerate	✅	Including collection types (SET, MULTISET, SEQUENCE)
Transactional DDL	❌	CUBRID auto-commits DDL

Feature	Status	Notes
`ON DUPLICATE KEY UPDATE`	✅	Via `sqlalchemy_cubrid.insert()`
`MERGE` statement	✅	Via `sqlalchemy_cubrid.merge()`
`REPLACE INTO`	✅	Via `sqlalchemy_cubrid.replace()`
`GROUP_CONCAT`	✅
`TRUNCATE TABLE`	✅	Autocommit detected
`FOR UPDATE`	✅	Including `OF` clause
Recursive CTE	✅	`WITH RECURSIVE` (CUBRID 11.x+)
Window functions	✅	ROW_NUMBER, RANK, LAG, LEAD, etc.
Index hints	✅	Via SA `with_hint()` / `suffix_with()`

Feature	Status	Notes
JSON type	❌	CUBRID has JSON since 10.2, but not yet mapped in dialect
Native Enum	❌	CUBRID lacks ENUM — use VARCHAR + CHECK constraint
Interval type	❌	Not supported by CUBRID
RETURNING clause	❌	`INSERT/UPDATE/DELETE ... RETURNING` not supported
BOOLEAN	⚠️	Mapped to SMALLINT (0/1) — no native boolean
Sequences	❌	CUBRID uses AUTO_INCREMENT only
CHECK constraint reflection	❌	`get_check_constraints()` returns empty list
Multi-schema	❌	CUBRID has single-schema model
RELEASE SAVEPOINT	❌	No-op (CUBRID doesn't support it)
Lateral joins	❌	CUBRID lacks LATERAL subquery support
Full-text search	❌	No MATCH … AGAINST syntax
Async DBAPI	❌	CUBRID drivers have no async support