feat(adbc): add Oracle, Db2, GizmoSQL, and Quack driver support

Makefile

@@ -346,10 +346,22 @@ bench-gate:                                         ## Run performance regressio
 DRIVERS ?= duckdb aiosqlite
 bench-gate-adapters:                                ## Compare selected adapter benchmarks with committed baselines
 	@mkdir -p /tmp/sqlspec-bench
-	@for driver in ${DRIVERS}; do \
-		uv run python tools/scripts/bench.py --driver "$${driver}" --extended --json-output "/tmp/sqlspec-bench/$${driver}.json"; \
-		uv run python tools/scripts/bench_compare.py "tools/perf_baselines/$${driver}.json" "/tmp/sqlspec-bench/$${driver}.json" --threshold "$${BENCHMARK_THRESHOLD:-30}"; \
-	done
+	@failed_drivers=""; \
+	for driver in ${DRIVERS}; do \
+		baseline="tools/perf_baselines/$${driver}.json"; \
+		run_args=$$(uv run --no-sync python tools/scripts/bench_compare.py "$${baseline}" --print-run-args); \
+		if ! uv run --no-sync python tools/scripts/bench.py --driver "$${driver}" --extended $${run_args} --json-output "/tmp/sqlspec-bench/$${driver}.json"; then \
+			failed_drivers="$${failed_drivers} $${driver}"; \
+			continue; \
+		fi; \
+		if ! uv run --no-sync python tools/scripts/bench_compare.py "$${baseline}" "/tmp/sqlspec-bench/$${driver}.json" --mode "$${BENCHMARK_MODE:-ratio}" --threshold "$${BENCHMARK_THRESHOLD:-30}"; then \
+			failed_drivers="$${failed_drivers} $${driver}"; \
+		fi; \
+	done; \
+	if [ -n "$${failed_drivers}" ]; then \
+		echo "Adapter benchmark gate failed for:$${failed_drivers}"; \
+		exit 1; \
+	fi
 
 .PHONY: bench-subsystems
 bench-subsystems:                                   ## Run subsystem micro-benchmarks

docs/changelog.rst

@@ -9,6 +9,113 @@ important operational fixes.
 Recent Updates
 ==============
 
+v0.67.0 - Enterprise ADBC drivers, Spanner graph and hybrid search, hot-path execution, and Python 3.11+
+--------------------------------------------------------------------------------------------------------
+
+**Added:**
+
+* ``sqlspec.adapters.adbc`` adds pure-Go GizmoData ADBC driver support for
+  Oracle (``adbc-driver-oracle``), IBM Db2 (``adbc-driver-db2``), native
+  GizmoSQL (``adbc-driver-gizmosql`` with automatic preference over
+  ``adbc-driver-flightsql``, ``gizmosql://`` URIs, and
+  ``gizmosql_auth_type="external"`` browser OAuth/SSO), and DuckDB Quack
+  (``adbc-driver-quack`` with ``quack://`` / ``quack+tls://`` URIs and
+  ``adbc.quack.token`` bearer authentication), including dialect resolution,
+  parameter binding (``:1`` positional colon for Oracle, ``?`` qmark for Db2),
+  savepoints, SQLSTATE/vendor exception mapping, and default
+  ``enable_lowercase_column_names=True`` normalization for Oracle and Db2.
+  (`#839 <https://github.com/litestar-org/sqlspec/pull/839>`_)
+* ``ADBCStore`` (Litestar sessions), ``AdbcEventQueueStore`` (event channels),
+  ``AdbcADKStore``, and ``AdbcADKMemoryStore`` (Google ADK sessions and memory)
+  add Oracle (``oracle``) and IBM Db2 (``db2``) backend support with idempotent
+  DDL guards (PL/SQL ``SQLCODE`` blocks for Oracle and ``SYSCAT`` catalog probes
+  for Db2) and dialect-specific timestamps, types, and search predicates.
+  (`#839 <https://github.com/litestar-org/sqlspec/pull/839>`_)
+* Spanner and Spangres SQLGlot dialects support ``CREATE [OR REPLACE] PROPERTY
+  GRAPH [IF NOT EXISTS]`` and ``DROP PROPERTY GRAPH [IF EXISTS]`` DDL (with
+  ``NODE TABLES``, ``EDGE TABLES``, ``SOURCE KEY``, ``DESTINATION KEY``,
+  ``LABEL``, and ``PROPERTIES`` clauses), ``GRAPH_TABLE`` queries with ``MATCH``
+  path patterns and ``COLUMNS`` projections via ``graph_table()``, ``CREATE
+  VECTOR INDEX [IF NOT EXISTS] ... STORING (...)``, ``DROP VECTOR INDEX [IF
+  EXISTS]``, ``DROP SEARCH INDEX [IF EXISTS]``, ``HIDDEN`` generated column
+  constraints (``TOKENLIST AS (TOKENIZE_FULLTEXT(...)) HIDDEN``), and
+  ``approx_cosine_distance(..., options=...)``, ``snippet()``, and
+  ``score_ngrams()`` expression helpers.
+  (`#834 <https://github.com/litestar-org/sqlspec/pull/834>`_)
+* ``SpannerSyncADKMemoryStore`` and ``SpannerAsyncADKMemoryStore`` add vector
+  embedding columns (``ARRAY<FLOAT32>``) with ``CREATE VECTOR INDEX`` DDL,
+  optional ``CREATE PROPERTY GRAPH`` DDL overlays, batched ``UNNEST``
+  deduplication on ``insert_memory_entries()``, and 4-mode ``search_entries()``
+  supporting Hybrid Reciprocal Rank Fusion (RRF) combining
+  ``APPROX_COSINE_DISTANCE`` and ``SEARCH``/``SCORE``, vector similarity search,
+  full-text ``SCORE`` ranking, and case-insensitive substring fallback.
+  (`#834 <https://github.com/litestar-org/sqlspec/pull/834>`_)
+* Added ``SpannerSyncADKArtifactStore`` and ``SpannerAsyncADKArtifactStore``
+  implementations of ``BaseSyncADKArtifactStore`` and
+  ``BaseAsyncADKArtifactStore`` with composite primary keys
+  ``(app_name, user_id, session_id, filename, version)``, user-scoped
+  ``session_id`` normalization, and optional ``ROW DELETION POLICY`` TTL
+  cleanup.
+  (`#834 <https://github.com/litestar-org/sqlspec/pull/834>`_)
+* ``SpannerSyncEventQueueStore`` and ``SpannerAsyncEventQueueStore`` add
+  snapshot-read candidate polling (``provide_session(transaction=False)``),
+  automatic ``Aborted`` retry wrapping via ``run_in_transaction()``, native
+  ``ROW DELETION POLICY`` / ``TTL INTERVAL`` retention (``cleanup_on_ack=False``),
+  optional ``shard_id`` hash-shard prefixing (``shard_count > 1`` on
+  ``EventsConfig``), covering queue indexes (``STORING`` on GoogleSQL,
+  ``INCLUDE`` on ``spangres``), full ``spangres`` DDL generation, and
+  ``database.update_ddl()`` execution for ``create_table()`` and
+  ``drop_table()``.
+  (`#834 <https://github.com/litestar-org/sqlspec/pull/834>`_)
+* Added Python 3.15 support across package metadata, CI test matrices, and
+  ``cibuildwheel`` MyPyC wheel builds.
+  (`#835 <https://github.com/litestar-org/sqlspec/pull/835>`_)
+
+**Changed:**
+
+* Upgraded the minimum supported Python version to 3.11+ (``requires-python =
+  ">=3.11, <4.0"``), removed the ``tomli`` fallback in favor of stdlib
+  ``tomllib``, and modernized ``enum.StrEnum``, ``datetime.UTC``, builtin
+  ``TimeoutError``, and stdlib ``typing`` usage across the codebase.
+  (`#835 <https://github.com/litestar-org/sqlspec/pull/835>`_)
+* Standardized driver session lifecycles, statement compilation caching, SQLGlot
+  builder immutability, migration version tracking, ``driver_features``
+  type-system coverage, PEP 562 lazy package exports, and MyPyC compilation
+  across all 20 database adapters (``core``, ``type_converter``, ``pool``, and
+  ``_typing`` modules).
+  (`#837 <https://github.com/litestar-org/sqlspec/pull/837>`_)
+* Unified DuckDB row materialization across ``execute()``, ``select()``, and
+  ``fetch*`` methods to consistently return DuckDB-native ``fetchall()`` Python
+  types on both first and cached executions, while ``select_to_arrow()``
+  continues to expose raw Arrow tables.
+  (`#838 <https://github.com/litestar-org/sqlspec/pull/838>`_)
+* Reduced per-statement and per-session execution overhead with per-config shared
+  statement caches (``StatementCacheHolder``), zero-copy cached SQLGlot AST
+  sharing with copy-on-mutation via ``ensure_owned_expression()``, fast-path
+  execution for parameterless and scalar-positional calls, fast-path execution
+  when statement observers are active, lazy ``SQLResult.statement``
+  materialization on cached ``SELECT`` hits, single-pass ``execute_many()``
+  parameter preparation, and idle-queue/clean-transaction fast paths in
+  ``AiosqliteConnectionPool``.
+  (`#838 <https://github.com/litestar-org/sqlspec/pull/838>`_)
+
+**Fixed:**
+
+* Hardened parameter conversion, Arrow batch streaming (``adbc_ingest`` stream
+  forwarding and ``mssql_python`` batch streaming), transaction and savepoint
+  rollback recovery, ``SlotPooledStatement`` cloning on ``SQLResult.statement``,
+  ``sqlglot >= 30.22.0`` single-table ``MappingSchema`` column qualification,
+  dialect transform isolation (``Spanner``, ``Spangres``, ``DB2``, ``PGVector``,
+  and ``ParadeDB``), self-referential table dependency ordering and composite
+  foreign key introspection, Windows ``file:///D:/...`` storage URI
+  normalization, and exception mapping across core, storage, migrations,
+  extensions, and database adapters.
+  (`#837 <https://github.com/litestar-org/sqlspec/pull/837>`_)
+* Prevented unparsed or ``COMMAND``/``UNKNOWN`` fallback compilations from
+  populating the driver statement cache, and cached parsed target-table metadata
+  on DuckDB Arrow bulk ``execute_many()`` inserts.
+  (`#838 <https://github.com/litestar-org/sqlspec/pull/838>`_)
+
 v0.66.0 - Native AsyncIO Spanner adapter, extension stores, and sync/async parity
 ---------------------------------------------------------------------------------
 

docs/conf.py

@@ -65,22 +65,22 @@
     "auto_pytabs.sphinx_ext",
     "sphinx_copybutton",
     "sphinx_click",
-    "click_extra.sphinx",
     "sphinx_design",
     "sphinx_paramlinks",
     "sphinxcontrib.mermaid",
     "numpydoc",
     "sphinx_iconify",
     "tools.sphinx_ext.playground",
 ]
+myst_enable_extensions = ["alert", "colon_fence"]
 intersphinx_mapping = {
     "python": ("https://docs.python.org/3", None),
     "litestar": ("https://docs.litestar.dev/latest/", None),
     "click": ("https://click.palletsprojects.com/en/stable/", None),
     "anyio": ("https://anyio.readthedocs.io/en/stable/", None),
     "multidict": ("https://multidict.aio-libs.org/en/stable/", None),
     "cryptography": ("https://cryptography.io/en/latest/", None),
-    "pydantic": ("https://docs.pydantic.dev/latest/", None),
+    "pydantic": ("https://pydantic.dev/docs/validation/latest/", None),
     "sanic": ("https://sanic.readthedocs.io/en/latest/", None),
     "flask": ("https://flask.palletsprojects.com/en/stable/", None),
     "fastapi": ("https://fastapi.tiangolo.com/", None),

docs/extensions/adk/adapters.rst

@@ -15,13 +15,14 @@ Use async adapters for best performance with ADK runners:
 - **MySQL/MariaDB**: ``aiomysql``, ``asyncmy``
 - **SQLite**: ``aiosqlite`` (development and single-process)
 - **Oracle**: ``oracledb``
+- **IBM Db2**: ``db2`` (native ``ibm_db`` sync and async modes)
 - **DuckDB**: ``duckdb`` (analytics; reduced-scope for ADK)
-- **ADBC**: ``adbc`` (Arrow-native portability; reduced-scope for ADK)
+- **ADBC**: ``adbc`` (Arrow-native portability across PostgreSQL, SQLite, DuckDB, BigQuery, Snowflake, Oracle, and IBM Db2)
 - **Spanner**: ``spanner`` (Google Cloud, globally distributed; supports both async and sync modes)
 - **SQL Server**: ``mssql_python`` (native Arrow and BulkCopy), ``pymssql`` (FreeTDS), ``arrow_odbc`` (Arrow over Microsoft ODBC Driver 18)
 
-Sync adapters (``psycopg`` sync mode, ``spanner`` sync mode, ``sqlite``, ``mysqlconnector``,
-``pymysql``, ``mssql_python``, ``pymssql``, ``arrow_odbc``) work but require wrapping with ``anyio`` for async
+Sync adapters (``psycopg`` sync mode, ``spanner`` sync mode, ``db2`` sync mode, ``sqlite``, ``mysqlconnector``,
+``pymysql``, ``mssql_python``, ``pymssql``, ``arrow_odbc``, ``adbc``) work with ADK and use ``anyio`` worker-thread dispatch for async
 ADK runners.
 
 Each Adapter Provides

docs/extensions/adk/backends.rst

@@ -83,6 +83,11 @@ The table below classifies every backend by its ADK support level.
      - Full
      - Full
      - Oracle Database driver.
+   * - db2
+     - Supported
+     - Full
+     - Basic
+     - IBM Db2 (LUW); ``POSSTR`` memory search.
    * - duckdb
      - Reduced-scope
      - Full
@@ -92,7 +97,7 @@ The table below classifies every backend by its ADK support level.
      - Reduced-scope
      - Full
      - Basic
-     - Portability layer; native adapters provide optimized search.
+     - Arrow Database Connectivity across PostgreSQL, SQLite, DuckDB, BigQuery, Snowflake, Oracle, and IBM Db2.
    * - mssql_python
      - Supported
      - Full
@@ -172,9 +177,8 @@ PostgreSQL backends (asyncpg, psycopg, psqlpy) provide the fullest feature set:
 - Full-text search via ``tsvector`` for memory entries.
 - ``UPSERT`` and ``RETURNING`` clauses for atomic operations.
 - ``append_event_and_update_state()`` executes as a single transaction.
-- ``fillfactor``, ``autovacuum_vacuum_scale_factor``, and
-  ``autovacuum_analyze_scale_factor`` tune the generated tables. The existing
-  ``fillfactor=80`` default is preserved.
+- ``fillfactor`` (default ``80``), ``autovacuum_vacuum_scale_factor``, and
+  ``autovacuum_analyze_scale_factor`` tune the generated tables.
 
 **Recommended for production deployments.**
 
@@ -290,25 +294,27 @@ DuckDB provides session and event storage but has limitations:
 ADBC
 ----
 
-ADBC (Arrow Database Connectivity) provides a driver-agnostic interface:
+ADBC (Arrow Database Connectivity) provides a portable Arrow-native store
+(:class:`~sqlspec.adapters.adbc.adk.AdbcADKStore` and
+:class:`~sqlspec.adapters.adbc.adk.AdbcADKMemoryStore`):
 
-- Works with any ADBC-compatible driver (PostgreSQL, SQLite, DuckDB, etc.).
+- Dialect-aware DDL, parameter binding, and pagination for PostgreSQL, SQLite,
+  DuckDB, BigQuery, Snowflake, Oracle (``:1, :2`` placeholders and PL/SQL
+  idempotent DDL), and IBM Db2 (``SYSCAT`` catalog guards and ``POSSTR`` memory
+  search).
 - Arrow-native data transfer for high-throughput event ingestion.
-- Backend capabilities depend on the underlying database driver.
-- Memory search uses the portable baseline path; choose a native adapter for
-  backend-specific FTS, retention, and storage tuning.
+- Portable substring memory search across all supported ADBC dialects; choose a
+  dedicated native adapter when you need backend-specific full-text search
+  indexes or retention policies.
 
 SQL Server (mssql-python and pymssql)
 -------------------------------------
 
 ``mssql_python`` and ``pymssql`` provide SQL Server-backed ADK storage:
 
 - Session and event storage use SQL Server tables with ``DATETIME2(6)`` and
-  ``NVARCHAR(MAX)`` JSON payload columns. Microsoft's
-  `type mapping reference <https://learn.microsoft.com/en-us/sql/connect/python/mssql-python/data-type-mappings>`_
-  lists native ``JSON`` as unsupported, even when the server supports the type.
-  The existing ``native_json=True`` override remains available for deployments
-  that have independently verified driver support.
+  ``NVARCHAR(MAX)`` JSON payload columns. Pass ``native_json=True`` when your
+  SQL Server deployment and driver support native ``JSON`` columns.
 - ``append_event_and_update_state()`` commits the session update, event row, and
   scoped state in one transaction.
 - Memory search is ``LIKE``-based across memory entries.