Merge branch 'main' into api_stability

.github/workflows/ci-additional.yaml

@@ -123,7 +123,7 @@ jobs:
           python -m mypy --install-types --non-interactive --cobertura-xml-report mypy_report
 
       - name: Upload mypy coverage to Codecov
-        uses: codecov/codecov-action@v5.0.7
+        uses: codecov/codecov-action@v5.1.1
         with:
           file: mypy_report/cobertura.xml
           flags: mypy
@@ -174,7 +174,7 @@ jobs:
           python -m mypy --install-types --non-interactive --cobertura-xml-report mypy_report
 
       - name: Upload mypy coverage to Codecov
-        uses: codecov/codecov-action@v5.0.7
+        uses: codecov/codecov-action@v5.1.1
         with:
           file: mypy_report/cobertura.xml
           flags: mypy-min
@@ -230,7 +230,7 @@ jobs:
           python -m pyright xarray/
 
       - name: Upload pyright coverage to Codecov
-        uses: codecov/codecov-action@v5.0.7
+        uses: codecov/codecov-action@v5.1.1
         with:
           file: pyright_report/cobertura.xml
           flags: pyright
@@ -286,7 +286,7 @@ jobs:
           python -m pyright xarray/
 
       - name: Upload pyright coverage to Codecov
-        uses: codecov/codecov-action@v5.0.7
+        uses: codecov/codecov-action@v5.1.1
         with:
           file: pyright_report/cobertura.xml
           flags: pyright39

.github/workflows/ci.yaml

@@ -159,7 +159,9 @@ jobs:
           path: pytest.xml
 
       - name: Upload code coverage to Codecov
-        uses: codecov/[email protected]
+        uses: codecov/[email protected]
+        env:
+          CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}
         with:
           file: ./coverage.xml
           flags: unittests

.github/workflows/pypi-release.yaml

@@ -88,7 +88,7 @@ jobs:
           path: dist
       - name: Publish package to TestPyPI
         if: github.event_name == 'push'
-        uses: pypa/[email protected].2
+        uses: pypa/[email protected].3
         with:
           repository_url: https://test.pypi.org/legacy/
           verbose: true
@@ -110,6 +110,6 @@ jobs:
           name: releases
           path: dist
       - name: Publish package to PyPI
-        uses: pypa/[email protected].2
+        uses: pypa/[email protected].3
         with:
           verbose: true

.github/workflows/upstream-dev-ci.yaml

@@ -140,7 +140,7 @@ jobs:
         run: |
           python -m mypy --install-types --non-interactive --cobertura-xml-report mypy_report
       - name: Upload mypy coverage to Codecov
-        uses: codecov/codecov-action@v5.0.7
+        uses: codecov/codecov-action@v5.1.1
         with:
           file: mypy_report/cobertura.xml
           flags: mypy

asv_bench/benchmarks/__init__.py

@@ -48,7 +48,7 @@ def randn(shape, frac_nan=None, chunks=None, seed=0):
 
 def randint(low, high=None, size=None, frac_minus=None, seed=0):
     rng = np.random.default_rng(seed)
-    x = rng.randint(low, high, size)
+    x = rng.integers(low, high, size)
     if frac_minus is not None:
         inds = rng.choice(range(x.size), int(x.size * frac_minus))
         x.flat[inds] = -1

doc/whats-new.rst

@@ -14,9 +14,9 @@ What's New
 
     np.random.seed(123456)
 
-.. _whats-new.2024.11.1:
+.. _whats-new.2024.12.0:
 
-v.2024.11.1 (unreleased)
+v.2024.12.0 (unreleased)
 ------------------------
 
 New Features
@@ -28,18 +28,28 @@ New Features
 
 Breaking changes
 ~~~~~~~~~~~~~~~~
-
+- Methods including ``dropna``, ``rank``, ``idxmax``, ``idxmin`` require
+  non-dimension arguments to be passed as keyword arguments. The previous
+  behavior, which allowed ``.idxmax('foo', 'all')`` was too easily confused with
+  ``'all'`` being a dimension. The updated equivalent is ``.idxmax('foo',
+  how='all')``. The previous behavior was deprecated in v2023.10.0.
+  By `Maximilian Roos <https://github.com/max-sixty>`_.
 
 Deprecations
 ~~~~~~~~~~~~
-
+- Finalize deprecation of ``closed`` parameters of :py:func:`cftime_range` and
+  :py:func:`date_range` (:pull:`9882`).
+  By `Kai Mühlbauer <https://github.com/kmuehlbauer>`_.
 
 Bug fixes
 ~~~~~~~~~
 - Fix type annotations for ``get_axis_num``. (:issue:`9822`, :pull:`9827`).
   By `Bruce Merry <https://github.com/bmerry>`_.
 - Fix unintended load on datasets when calling :py:meth:`DataArray.plot.scatter` (:pull:`9818`).
   By `Jimmy Westling <https://github.com/illviljan>`_.
+- Fix interpolation when non-numeric coordinate variables are present (:issue:`8099`, :issue:`9839`).
+  By `Deepak Cherian <https://github.com/dcherian>`_.
+
 
 Documentation
 ~~~~~~~~~~~~~

xarray/backends/zarr.py

@@ -1135,9 +1135,7 @@ def set_variables(self, variables, check_encoding_set, writer, unlimited_dims=No
                 else:
                     encoded_attrs[DIMENSION_KEY] = dims
 
-                encoding["exists_ok" if _zarr_v3() else "overwrite"] = (
-                    True if self._mode == "w" else False
-                )
+                encoding["overwrite"] = True if self._mode == "w" else False
 
                 zarr_array = self._create_new_array(
                     name=name,

xarray/coding/cftime_offsets.py

@@ -62,15 +62,13 @@
 )
 from xarray.core.common import _contains_datetime_like_objects, is_np_datetime_like
 from xarray.core.pdcompat import (
-    NoDefault,
     count_not_none,
     nanosecond_precision_timestamp,
-    no_default,
 )
 from xarray.core.utils import attempt_import, emit_user_level_warning
 
 if TYPE_CHECKING:
-    from xarray.core.types import InclusiveOptions, Self, SideOptions, TypeAlias
+    from xarray.core.types import InclusiveOptions, Self, TypeAlias
 
 
 DayOption: TypeAlias = Literal["start", "end"]
@@ -943,51 +941,14 @@ def _generate_range(start, end, periods, offset):
             current = next_date
 
 
-def _translate_closed_to_inclusive(closed):
-    """Follows code added in pandas #43504."""
-    emit_user_level_warning(
-        "Following pandas, the `closed` parameter is deprecated in "
-        "favor of the `inclusive` parameter, and will be removed in "
-        "a future version of xarray.",
-        FutureWarning,
-    )
-    if closed is None:
-        inclusive = "both"
-    elif closed in ("left", "right"):
-        inclusive = closed
-    else:
-        raise ValueError(
-            f"Argument `closed` must be either 'left', 'right', or None. "
-            f"Got {closed!r}."
-        )
-    return inclusive
-
-
-def _infer_inclusive(
-    closed: NoDefault | SideOptions, inclusive: InclusiveOptions | None
-) -> InclusiveOptions:
-    """Follows code added in pandas #43504."""
-    if closed is not no_default and inclusive is not None:
-        raise ValueError(
-            "Following pandas, deprecated argument `closed` cannot be "
-            "passed if argument `inclusive` is not None."
-        )
-    if closed is not no_default:
-        return _translate_closed_to_inclusive(closed)
-    if inclusive is None:
-        return "both"
-    return inclusive
-
-
 def cftime_range(
     start=None,
     end=None,
     periods=None,
     freq=None,
     normalize=False,
     name=None,
-    closed: NoDefault | SideOptions = no_default,
-    inclusive: None | InclusiveOptions = None,
+    inclusive: InclusiveOptions = "both",
     calendar="standard",
 ) -> CFTimeIndex:
     """Return a fixed frequency CFTimeIndex.
@@ -1006,16 +967,7 @@ def cftime_range(
         Normalize start/end dates to midnight before generating date range.
     name : str, default: None
         Name of the resulting index
-    closed : {None, "left", "right"}, default: "NO_DEFAULT"
-        Make the interval closed with respect to the given frequency to the
-        "left", "right", or both sides (None).
-
-        .. deprecated:: 2023.02.0
-            Following pandas, the ``closed`` parameter is deprecated in favor
-            of the ``inclusive`` parameter, and will be removed in a future
-            version of xarray.
-
-    inclusive : {None, "both", "neither", "left", "right"}, default None
+    inclusive : {"both", "neither", "left", "right"}, default "both"
         Include boundaries; whether to set each bound as closed or open.
 
         .. versionadded:: 2023.02.0
@@ -1193,8 +1145,6 @@ def cftime_range(
         offset = to_offset(freq)
         dates = np.array(list(_generate_range(start, end, periods, offset)))
 
-    inclusive = _infer_inclusive(closed, inclusive)
-
     if inclusive == "neither":
         left_closed = False
         right_closed = False
@@ -1229,8 +1179,7 @@ def date_range(
     tz=None,
     normalize=False,
     name=None,
-    closed: NoDefault | SideOptions = no_default,
-    inclusive: None | InclusiveOptions = None,
+    inclusive: InclusiveOptions = "both",
     calendar="standard",
     use_cftime=None,
 ):
@@ -1257,20 +1206,10 @@ def date_range(
         Normalize start/end dates to midnight before generating date range.
     name : str, default: None
         Name of the resulting index
-    closed : {None, "left", "right"}, default: "NO_DEFAULT"
-        Make the interval closed with respect to the given frequency to the
-        "left", "right", or both sides (None).
-
-        .. deprecated:: 2023.02.0
-            Following pandas, the `closed` parameter is deprecated in favor
-            of the `inclusive` parameter, and will be removed in a future
-            version of xarray.
-
-    inclusive : {None, "both", "neither", "left", "right"}, default: None
+    inclusive : {"both", "neither", "left", "right"}, default: "both"
         Include boundaries; whether to set each bound as closed or open.
 
         .. versionadded:: 2023.02.0
-
     calendar : str, default: "standard"
         Calendar type for the datetimes.
     use_cftime : boolean, optional
@@ -1294,8 +1233,6 @@ def date_range(
     if tz is not None:
         use_cftime = False
 
-    inclusive = _infer_inclusive(closed, inclusive)
-
     if _is_standard_calendar(calendar) and use_cftime is not True:
         try:
             return pd.date_range(

xarray/core/dask_array_ops.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import math
+from functools import partial
 
 from xarray.core import dtypes, nputils
 
@@ -75,6 +76,47 @@ def least_squares(lhs, rhs, rcond=None, skipna=False):
     return coeffs, residuals
 
 
+def _fill_with_last_one(a, b):
+    import numpy as np
+
+    # cumreduction apply the push func over all the blocks first so,
+    # the only missing part is filling the missing values using the
+    # last data of the previous chunk
+    return np.where(np.isnan(b), a, b)
+
+
+def _dtype_push(a, axis, dtype=None):
+    from xarray.core.duck_array_ops import _push
+
+    # Not sure why the blelloch algorithm force to receive a dtype
+    return _push(a, axis=axis)
+
+
+def _reset_cumsum(a, axis, dtype=None):
+    import numpy as np
+
+    cumsum = np.cumsum(a, axis=axis)
+    reset_points = np.maximum.accumulate(np.where(a == 0, cumsum, 0), axis=axis)
+    return cumsum - reset_points
+
+
+def _last_reset_cumsum(a, axis, keepdims=None):
+    import numpy as np
+
+    # Take the last cumulative sum taking into account the reset
+    # This is useful for blelloch method
+    return np.take(_reset_cumsum(a, axis=axis), axis=axis, indices=[-1])
+
+
+def _combine_reset_cumsum(a, b, axis):
+    import numpy as np
+
+    # It is going to sum the previous result until the first
+    # non nan value
+    bitmask = np.cumprod(b != 0, axis=axis)
+    return np.where(bitmask, b + a, b)
+
+
 def push(array, n, axis, method="blelloch"):
     """
     Dask-aware bottleneck.push
@@ -91,16 +133,6 @@ def push(array, n, axis, method="blelloch"):
     # TODO: Replace all this function
     #  once https://github.com/pydata/xarray/issues/9229 being implemented
 
-    def _fill_with_last_one(a, b):
-        # cumreduction apply the push func over all the blocks first so,
-        # the only missing part is filling the missing values using the
-        # last data of the previous chunk
-        return np.where(np.isnan(b), a, b)
-
-    def _dtype_push(a, axis, dtype=None):
-        # Not sure why the blelloch algorithm force to receive a dtype
-        return _push(a, axis=axis)
-
     pushed_array = da.reductions.cumreduction(
         func=_dtype_push,
         binop=_fill_with_last_one,
@@ -113,26 +145,9 @@ def _dtype_push(a, axis, dtype=None):
     )
 
     if n is not None and 0 < n < array.shape[axis] - 1:
-
-        def _reset_cumsum(a, axis, dtype=None):
-            cumsum = np.cumsum(a, axis=axis)
-            reset_points = np.maximum.accumulate(np.where(a == 0, cumsum, 0), axis=axis)
-            return cumsum - reset_points
-
-        def _last_reset_cumsum(a, axis, keepdims=None):
-            # Take the last cumulative sum taking into account the reset
-            # This is useful for blelloch method
-            return np.take(_reset_cumsum(a, axis=axis), axis=axis, indices=[-1])
-
-        def _combine_reset_cumsum(a, b):
-            # It is going to sum the previous result until the first
-            # non nan value
-            bitmask = np.cumprod(b != 0, axis=axis)
-            return np.where(bitmask, b + a, b)
-
         valid_positions = da.reductions.cumreduction(
             func=_reset_cumsum,
-            binop=_combine_reset_cumsum,
+            binop=partial(_combine_reset_cumsum, axis=axis),
             ident=0,
             x=da.isnan(array, dtype=int),
             axis=axis,