You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@superset.apache.org by ru...@apache.org on 2026/08/25 01:01:31 UTC

(superset) 01/01: chore: remove obsolete pandas/SQLAlchemy version compat shim

This is an automated email from the ASF dual-hosted git repository.

rusackas pushed a commit to branch chore/remove-pandas-sqlalchemy-compat-shim
in repository https://gitbox.apache.org/repos/asf/superset.git

commit 19c09ed7f17fa81cd16a66f23e0ef291b942b433
Author: Superset Dev <de...@superset.apache.org>
AuthorDate: Mon Aug 24 18:01:10 2026 -0700

    chore: remove obsolete pandas/SQLAlchemy version compat shim
    
    superset/utils/pandas_sqlalchemy_compat.py existed to work around
    pandas >= 2.2 silently rejecting SQLAlchemy 1.4 engines, and was
    already a no-op once SQLAlchemy >= 2.0 was installed. Now that
    pyproject.toml hard-requires sqlalchemy>=2.0.52 (#42803, discussion
    #40273), the condition it patches around can no longer occur, exactly
    as its own docstring called out as the removal trigger.
---
 superset/__init__.py                               | 10 ---
 superset/utils/pandas_sqlalchemy_compat.py         | 80 ----------------------
 .../utils/pandas_sqlalchemy_compat_test.py         | 67 ------------------
 3 files changed, 157 deletions(-)

diff --git a/superset/__init__.py b/superset/__init__.py
index 1cecdeed034..e3ab53c46b3 100644
--- a/superset/__init__.py
+++ b/superset/__init__.py
@@ -28,16 +28,6 @@ from werkzeug.local import LocalProxy
 # form.
 flask_appbuilder.Model.__allow_unmapped__ = True
 
-# pandas >= 2.2 advertises a minimum SQLAlchemy of 2.0 and silently ignores
-# older installations, breaking DataFrame.to_sql / read_sql with SQLAlchemy
-# 1.4 engines. Its SQL layer still works with 1.4, so restore support until
-# Superset itself requires SQLAlchemy >= 2. Must run before any pandas SQL IO.
-from superset.utils.pandas_sqlalchemy_compat import (  # noqa: E402
-    restore_pandas_sqlalchemy_support,
-)
-
-restore_pandas_sqlalchemy_support()
-
 from superset.app import create_app  # noqa: E402, F401
 from superset.extensions import (  # noqa: E402
     appbuilder,  # noqa: F401
diff --git a/superset/utils/pandas_sqlalchemy_compat.py b/superset/utils/pandas_sqlalchemy_compat.py
deleted file mode 100644
index 3f3291418a7..00000000000
--- a/superset/utils/pandas_sqlalchemy_compat.py
+++ /dev/null
@@ -1,80 +0,0 @@
-# Licensed to the Apache Software Foundation (ASF) under one
-# or more contributor license agreements.  See the NOTICE file
-# distributed with this work for additional information
-# regarding copyright ownership.  The ASF licenses this file
-# to you under the Apache License, Version 2.0 (the
-# "License"); you may not use this file except in compliance
-# with the License.  You may obtain a copy of the License at
-#
-#   http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing,
-# software distributed under the License is distributed on an
-# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
-# KIND, either express or implied.  See the License for the
-# specific language governing permissions and limitations
-# under the License.
-"""Compatibility shim letting pandas >= 2.2 use SQLAlchemy 1.4 engines.
-
-pandas 2.2 raised its advertised minimum SQLAlchemy version to 2.0 as a
-support-policy change. When an older SQLAlchemy is installed, pandas does not
-fail loudly: ``pandas.io.sql`` silently pretends SQLAlchemy is absent, treats
-Engine/Connection arguments as raw DBAPI connections, and falls back to its
-sqlite-only code path, breaking every ``DataFrame.to_sql`` / ``read_sql``
-call site (dataset uploads, example data loading, annotation queries, filter
-values).
-
-The pandas SQL layer itself still works with SQLAlchemy 1.4 because it only
-uses the API subset common to SQLAlchemy 1.4 and 2.x. Lowering the advertised
-minimum back to the pandas 2.1 value restores the working behavior.
-
-This module is obsolete once Superset requires SQLAlchemy >= 2; at that point
-the patch becomes a no-op and the module (and its call site in
-``superset/__init__.py``) can be deleted.
-"""
-
-import logging
-
-import sqlalchemy
-from packaging.version import Version
-
-logger = logging.getLogger(__name__)
-
-# The last pandas release line to support SQLAlchemy 1.4 (pandas 2.1)
-# required at least this version.
-_SQLALCHEMY_MINIMUM = "1.4.16"
-
-
-def restore_pandas_sqlalchemy_support() -> None:
-    """Lower pandas' advertised SQLAlchemy minimum so 1.4 engines work.
-
-    Only applies when the installed SQLAlchemy predates 2.0 and pandas
-    advertises a 2.x minimum; in every other combination this is a no-op.
-    Safe to call multiple times.
-    """
-    if Version(sqlalchemy.__version__) >= Version("2.0.0"):
-        # pandas supports SQLAlchemy 2.x natively; nothing to patch.
-        return
-
-    try:
-        from pandas.compat import _optional
-    except ImportError:
-        # The private module moved in a newer pandas; SQL IO with a pre-2.0
-        # SQLAlchemy will misbehave, so make the situation diagnosable.
-        logger.warning(
-            "Could not adjust pandas' minimum SQLAlchemy version; "
-            "DataFrame.to_sql/read_sql may not accept SQLAlchemy %s engines",
-            sqlalchemy.__version__,
-        )
-        return
-
-    advertised = _optional.VERSIONS.get("sqlalchemy")
-    if advertised and Version(advertised) > Version(_SQLALCHEMY_MINIMUM):
-        _optional.VERSIONS["sqlalchemy"] = _SQLALCHEMY_MINIMUM
-        logger.debug(
-            "Lowered pandas' minimum SQLAlchemy version from %s to %s so "
-            "pandas SQL IO keeps working with the installed SQLAlchemy %s",
-            advertised,
-            _SQLALCHEMY_MINIMUM,
-            sqlalchemy.__version__,
-        )
diff --git a/tests/unit_tests/utils/pandas_sqlalchemy_compat_test.py b/tests/unit_tests/utils/pandas_sqlalchemy_compat_test.py
deleted file mode 100644
index 980f3123546..00000000000
--- a/tests/unit_tests/utils/pandas_sqlalchemy_compat_test.py
+++ /dev/null
@@ -1,67 +0,0 @@
-# Licensed to the Apache Software Foundation (ASF) under one
-# or more contributor license agreements.  See the NOTICE file
-# distributed with this work for additional information
-# regarding copyright ownership.  The ASF licenses this file
-# to you under the Apache License, Version 2.0 (the
-# "License"); you may not use this file except in compliance
-# with the License.  You may obtain a copy of the License at
-#
-#   http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing,
-# software distributed under the License is distributed on an
-# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
-# KIND, either express or implied.  See the License for the
-# specific language governing permissions and limitations
-# under the License.
-import pandas as pd
-from sqlalchemy import create_engine, types
-
-from superset.utils.pandas_sqlalchemy_compat import (
-    restore_pandas_sqlalchemy_support,
-)
-
-
-def test_to_sql_accepts_sqlalchemy_engine_and_dtypes() -> None:
-    """
-    ``DataFrame.to_sql`` must accept a SQLAlchemy engine plus SQLAlchemy
-    ``dtype`` objects regardless of the installed pandas/SQLAlchemy combo.
-
-    This is the exact call shape used by dataset uploads
-    (``BaseEngineSpec.df_to_sql``), example data loading, and the test data
-    loaders; it breaks when pandas silently rejects the installed SQLAlchemy
-    as too old (pandas >= 2.2 with SQLAlchemy 1.x) and no compat shim is
-    applied.
-    """
-    restore_pandas_sqlalchemy_support()
-
-    engine = create_engine("sqlite://")
-    df = pd.DataFrame(
-        {
-            "name": ["a", "b"],
-            "num": [1, 2],
-            "ds": pd.to_datetime(["2021-01-01", "2021-01-02"]),
-        }
-    )
-    df.to_sql(
-        "birth_names",
-        engine,
-        index=False,
-        dtype={"ds": types.DateTime(), "name": types.String(255)},
-        method="multi",
-        chunksize=100,
-    )
-    df.to_sql("birth_names", engine, index=False, if_exists="replace")
-
-    result = pd.read_sql_query("SELECT name, num FROM birth_names", engine)
-    assert result["name"].tolist() == ["a", "b"]
-    assert result["num"].tolist() == [1, 2]
-
-
-def test_restore_pandas_sqlalchemy_support_is_idempotent() -> None:
-    from pandas.compat import _optional
-
-    restore_pandas_sqlalchemy_support()
-    first = _optional.VERSIONS.get("sqlalchemy")
-    restore_pandas_sqlalchemy_support()
-    assert _optional.VERSIONS.get("sqlalchemy") == first