pandas-dev/pandas · error · ValueError

Unable to avoid copy while creating an array as requested.

Error message

Unable to avoid copy while creating an array as requested.

What it means

Raised by DatetimeLikeArrayMixin.__array__ when a caller requests dtype=object together with copy=False. Converting a packed datetime/timedelta/period array to object dtype fundamentally requires materialising Python objects (a copy), so a no-copy request is impossible and pandas surfaces the conflict as ValueError rather than silently ignoring the flag.

Source

Thrown at pandas/core/arrays/datetimelike.py:349

        -------
        ndarray[str]
        """
        raise AbstractMethodError(self)

    def _formatter(self, boxed: bool = False) -> Callable[[object], str]:
        # TODO: Remove Datetime & DatetimeTZ formatters.
        return "'{}'".format

    # ----------------------------------------------------------------
    # Array-Like / EA-Interface Methods

    def __array__(
        self, dtype: NpDtype | None = None, copy: bool | None = None
    ) -> np.ndarray:
        # used for Timedelta/DatetimeArray, overwritten by PeriodArray
        if is_object_dtype(dtype):
            if copy is False:
                raise ValueError(
                    "Unable to avoid copy while creating an array as requested."
                )
            return np.array(list(self), dtype=object)

        if copy is True:
            return np.array(self._ndarray, dtype=dtype)

        result = self._ndarray
        if self._readonly:
            result = result.view()
            result.flags.writeable = False
        return result

    @overload
    def __getitem__(self, key: ScalarIndexer) -> DTScalarOrNaT: ...

    @overload
    def __getitem__(

View on GitHub (pinned to 71959b8cb9)

Solutions

1. Allow the copy: drop copy=False, or pass copy=True when you need object dtype.
2. Use arr.to_numpy(dtype=object) (lets pandas choose copy semantics) or arr.astype(object).
3. If you truly need zero-copy, keep the native int64/datetime64 dtype instead of converting to object.

Example fix

// before
import numpy as np
arr = pd.date_range('2020', periods=3)._data
np.asarray(arr, dtype=object, copy=False)  # ValueError

// after
np.asarray(arr, dtype=object)  # copy permitted

import numpy as np
def to_object_array(arr):
    try:
        return np.asarray(arr, dtype=object, copy=False)
    except ValueError:
        return np.asarray(arr, dtype=object)

from typing import Any

def allows_nocopy_object(arr: Any) -> bool:
    # object-dtype materialisation always copies; never zero-copy
    return False

try:
    np.asarray(arr, dtype=object, copy=False)
except ValueError as e:
    if 'Unable to avoid copy' in str(e):
        np.asarray(arr, dtype=object)
    else:
        raise

When it happens

Trigger: np.asarray(datetime_array, dtype=object, copy=False); np.array(arr, dtype=object, copy=False); or any code path that calls __array__ with both object dtype and a false copy flag. NumPy's copy=False (or the older np.array(..., copy=False)) contract triggers this.

Common situations: Downstream libraries (e.g. dask, xarray, numba interop) that pass copy=False for memory efficiency, or hand-written np.asarray(..., copy=False) calls. Also surfaces with numpy>=2.0 where __array__ gained the copy kwarg.

Related errors

• {left_base!r} is {right_base!r}
• Unable to avoid copy while creating an array as requested.
• Casting to unit-less dtype 'datetime64' is not supported. Pa
• Passing in 'datetime64' dtype with no precision is not allow
• Unable to avoid copy while creating an array as requested.

AI-assisted analysis of pandas-dev/pandas@71959b8cb9 (2026-08-07). Data as JSON: /api/errors/f7bee161e4c3b9b3. Report an issue: GitHub.