[v3-3-test] Review and update constraint dependencies (#70652)

airflow-core/pyproject.toml

@@ -143,7 +143,7 @@ dependencies = [
     "rich-argparse>=1.0.0",
     "rich>=13.6.0",
     "setproctitle>=1.3.3",
-    "sqlalchemy[asyncio]>=2.0.48",
+    "sqlalchemy[asyncio]>=2.0.50",
     "svcs>=25.1.0",
     "tabulate>=0.9.0",
     "tenacity>=8.3.0",

providers/databricks/README.rst

@@ -57,7 +57,7 @@ PIP package                                 Version required
 ``apache-airflow-providers-common-compat``  ``>=1.13.0``
 ``apache-airflow-providers-common-sql``     ``>=1.32.0``
 ``requests``                                ``>=2.32.0,<3``
-``databricks-sql-connector``                ``>=4.0.0``
+``databricks-sql-connector``                ``>=4.4.0``
 ``aiohttp``                                 ``>=3.14.0,<4``
 ``mergedeep``                               ``>=1.3.4``
 ``pandas``                                  ``>=2.1.2; python_version < "3.13"``

providers/databricks/docs/index.rst

@@ -105,7 +105,7 @@ PIP package                                 Version required
 ``apache-airflow-providers-common-compat``  ``>=1.13.0``
 ``apache-airflow-providers-common-sql``     ``>=1.32.0``
 ``requests``                                ``>=2.32.0,<3``
-``databricks-sql-connector``                ``>=4.0.0``
+``databricks-sql-connector``                ``>=4.4.0``
 ``aiohttp``                                 ``>=3.14.0,<4``
 ``mergedeep``                               ``>=1.3.4``
 ``pandas``                                  ``>=2.1.2; python_version < "3.13"``

providers/databricks/pyproject.toml

@@ -63,7 +63,7 @@ dependencies = [
     "apache-airflow-providers-common-compat>=1.13.0",
     "apache-airflow-providers-common-sql>=1.32.0",
     "requests>=2.32.0,<3",
-    "databricks-sql-connector>=4.0.0", # use >=4.2.7 for pandas 3.0.0. See https://github.com/databricks/databricks-sql-python/pull/768
+    "databricks-sql-connector>=4.4.0",
     "aiohttp>=3.14.0, <4",
     "mergedeep>=1.3.4",
     'pandas>=2.1.2; python_version <"3.13"',

scripts/in_container/run_generate_constraints.py

@@ -392,21 +392,13 @@ def generate_constraints_pypi_providers(config_params: ConfigParams) -> None:
     # * pyarrow>=22.0.0 on Python 3.14 — older pyarrow releases have no prebuilt wheels for
     #   Python 3.14 and uv falls back to building from source, which fails. pyarrow 22.0.0 is
     #   the first release shipping cp314 wheels.
-    # * pymysql<1.2 — pymysql 1.2.0 changed Connection.ping() to require `reconnect` as a
-    #   positional arg, which breaks SQLAlchemy's AsyncAdapt_aiomysql_connection.ping() (it
-    #   has no default for `reconnect`). The released apache-airflow-providers-mysql on PyPI
-    #   does not yet carry this cap, so we mirror it here so PyPI constraints stay installable
-    #   until the SQLAlchemy fix is released. Tracked upstream at
-    #   https://github.com/sqlalchemy/sqlalchemy/issues/13306
-    # * databricks-sql-connector>=4.0.0 - added to keep databricks-sql-connector from downgrading
-    #   because of https://github.com/apache/thrift/pull/3584 - which shipped thrift 0.24.0. Older
-    #   versions of databricks-sql-connector do not have the thrift<=0.23.0 limitation, so the
-    #   resolver preferred the latest thrift over the latest connector and downgraded the connector.
-    #   This is tracked in https://github.com/databricks/databricks-sql-python/issues/859
+    # * gremlinpython>=3.8.0 — older versions of gremlinpython have more relaxed dependencies and
+    #   resolver might choose to downgrade gremlinpython to an older version,
+    #   which in turn might downgrade tinkerpop provider. Having gremlinpython>=3.8.0 ensures
+    #   that the resolver will not downgrade the provider.
     additional_constraints_for_highest_resolution: list[str] = [
         "pyarrow>=22.0.0; python_version >= '3.14'",
-        "pymysql>=1.0.3,<1.2",
-        "databricks-sql-connector>=4.0.0",
+        "gremlinpython>=3.8.0",
     ]
 
     result = run_command(