Merge branch 'main' into enable-type-checking

.github/workflows/ci-manifest.yml

@@ -23,4 +23,4 @@ concurrency:
 jobs:
   manifest:
     name: "check-manifest"
-    uses: scitools/workflows/.github/workflows/ci-manifest.yml@2024.05.0
+    uses: scitools/workflows/.github/workflows/ci-manifest.yml@2024.06.0

.github/workflows/refresh-lockfiles.yml

@@ -14,5 +14,5 @@ on:
 
 jobs:
   refresh_lockfiles:
-    uses: scitools/workflows/.github/workflows/refresh-lockfiles.yml@2024.05.0
+    uses: scitools/workflows/.github/workflows/refresh-lockfiles.yml@2024.06.0
     secrets: inherit

.pre-commit-config.yaml

@@ -29,7 +29,7 @@ repos:
     -   id: no-commit-to-branch
 
 -   repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: "v0.4.4"
+    rev: "v0.4.7"
     hooks:
     -   id: ruff
         types: [file, python]
@@ -38,7 +38,7 @@ repos:
         types: [file, python]
 
 -   repo: https://github.com/codespell-project/codespell
-    rev: "v2.2.6"
+    rev: "v2.3.0"
     hooks:
     -   id: codespell
         types_or: [asciidoc, python, markdown, rst]

docs/src/conf.py

@@ -21,12 +21,14 @@
 
 import datetime
 from importlib.metadata import version as get_version
+from inspect import getsource
 import ntpath
 import os
 from pathlib import Path
 import re
 from subprocess import run
 import sys
+from tempfile import gettempdir
 from urllib.parse import quote
 import warnings
 
@@ -409,6 +411,26 @@ def _dotv(version):
 # -- sphinx-gallery config ----------------------------------------------------
 # See https://sphinx-gallery.github.io/stable/configuration.html
 
+
+def reset_modules(gallery_conf, fname):
+    """Force re-registering of nc-time-axis with matplotlib for each example.
+
+    Required for sphinx-gallery>=0.11.0.
+    """
+    from sys import modules
+
+    _ = modules.pop("nc_time_axis", None)
+
+
+# https://sphinx-gallery.github.io/dev/configuration.html#importing-callables
+reset_modules_dir = Path(gettempdir()) / reset_modules.__name__
+reset_modules_dir.mkdir(exist_ok=True)
+(reset_modules_dir / f"{reset_modules.__name__}.py").write_text(
+    getsource(reset_modules)
+)
+sys.path.insert(0, str(reset_modules_dir))
+
+
 sphinx_gallery_conf = {
     # path to your example scripts
     "examples_dirs": ["../gallery_code"],
@@ -420,11 +442,7 @@ def _dotv(version):
     "ignore_pattern": r"__init__\.py",
     # force gallery building, unless overridden (see src/Makefile)
     "plot_gallery": "'True'",
-    # force re-registering of nc-time-axis with matplotlib for each example,
-    # required for sphinx-gallery>=0.11.0
-    "reset_modules": (
-        lambda gallery_conf, fname: sys.modules.pop("nc_time_axis", None),
-    ),
+    "reset_modules": f"{reset_modules.__name__}.{reset_modules.__name__}",
 }
 
 # -----------------------------------------------------------------------------

docs/src/copyright.rst

@@ -27,10 +27,10 @@ are licensed under the UK's Open Government Licence:
 
     (C) British Crown Copyright |copyright_years|
 
-    You may use and re-use the information featured on this website (not including logos) free of
+    You may use and reuse the information featured on this website (not including logos) free of
     charge in any format or medium, under the terms of the
     `Open Government Licence <https://www.nationalarchives.gov.uk/doc/open-government-licence>`_.
     We encourage users to establish hypertext links to this website.
 
-    Any email enquiries regarding the use and re-use of this information resource should be
+    Any email enquiries regarding the use and reuse of this information resource should be
     sent to: [email protected].

docs/src/developers_guide/contributing_benchmarks.rst

@@ -7,8 +7,9 @@ Benchmarking
 Iris includes architecture for benchmarking performance and other metrics of
 interest. This is done using the `Airspeed Velocity`_ (ASV) package.
 
-Full detail on the setup and how to run or write benchmarks is in
-`benchmarks/README.md`_ in the Iris repository.
+
+.. note:: Full detail on the setup and how to run or write benchmarks is in
+          `benchmarks/README.md`_ in the Iris repository.
 
 Continuous Integration
 ----------------------

docs/src/further_topics/dask_best_practices/dask_bags_and_greed.rst

@@ -7,7 +7,7 @@ Here is a journey that demonstrates:
 
 * How to apply dask.bags to an existing script
 * The equal importance of optimisation of non-parallel parts of a script
-* Protection against multiple softwares trying to manage parallelism
+* Protection against multiple software trying to manage parallelism
   simultaneously
 
 

docs/src/further_topics/netcdf_io.rst

@@ -66,7 +66,7 @@ creation of the :data:`iris.fileformats.netcdf.loader.CHUNK_CONTROL` class.
 Custom Chunking: Set
 ^^^^^^^^^^^^^^^^^^^^
 
-There are three context manangers within :data:`~iris.fileformats.netcdf.loader.CHUNK_CONTROL`. The most basic is
+There are three context managers within :data:`~iris.fileformats.netcdf.loader.CHUNK_CONTROL`. The most basic is
 :meth:`~iris.fileformats.netcdf.loader.ChunkControl.set`. This allows you to specify the chunksize for each dimension,
 and to specify a ``var_name`` specifically to change.
 

docs/src/further_topics/ugrid/partner_packages.rst

@@ -93,7 +93,7 @@ Applications
 * Regrid unstructured to structured.
 * Regrid with dask integration, computing in parallel and maintaining data
   laziness.
-* | Save a prepared regridder for re-use in subsequent runs.
+* | Save a prepared regridder for reuse in subsequent runs.
   | Regridders can even be re-used on sources with different masks - a
     significant efficiency gain.
 

docs/src/further_topics/which_regridder_to_use.rst

@@ -341,7 +341,7 @@ of ``kg m-2`` as an area weighted sum. With ``mdtol=0`` this will consistently
 underestimate this total and with ``mdtol=1`` will consistently overestimate. This can
 be somewhat mitigated with a choice of ``mdtol=0.5``, but you should still be aware of
 potential inaccuracies. It should be noted that this choice of ``mdtol`` is highly
-context dependent and there wil likely be occasions where a choice of ``mdtol=0`` or
+context dependent and there will likely be occasions where a choice of ``mdtol=0`` or
 ``mdtol=1`` is more suitable. The important thing is to *know your data, know what*
 *you're doing with your data and know how your regridder fits in this process*.
 

docs/src/whatsnew/3.6.rst

@@ -15,11 +15,11 @@ This document explains the changes made to Iris for this release
 
     We're so excited about our recent support for **delayed saving of lazy data
     to netCDF** (:pull:`5191`) that we're celebrating this important step change
-    in behavour with its very own dedicated release 🥳
+    in behaviour with its very own dedicated release 🥳
 
     By using ``iris.save(..., compute=False)`` you can now save to multiple NetCDF files
     in parallel.  See the new ``compute`` keyword in :func:`iris.fileformats.netcdf.save`.
-    This can share and re-use any common (lazy) result computations, and it makes much
+    This can share and reuse any common (lazy) result computations, and it makes much
     better use of resources during any file-system waiting (i.e., it can use such periods
     to progress the *other* saves).
 

docs/src/whatsnew/latest.rst

@@ -55,14 +55,14 @@ This document explains the changes made to Iris for this release
 🚀 Performance Enhancements
 ===========================
 
-#. N/A
-
 #. `@bouweandela`_ added the option to specify the Dask chunks of the target
    array in :func:`iris.util.broadcast_to_shape`. (:pull:`5620`)
 
 #. `@schlunma`_ allowed :func:`iris.analysis.cartography.area_weights` to
    return dask arrays with arbitrary chunks. (:pull:`5658`)
 
+#. `@bouweandela`_ made :meth:`iris.cube.Cube.rolling_window` work with lazy
+   data. (:pull:`5795`)
 
 🔥 Deprecations
 ===============
@@ -114,15 +114,17 @@ This document explains the changes made to Iris for this release
 #. `@pp-mo`_ reworked benchmark peak-memory measurement to use the
    `tracemalloc <https://docs.python.org/3.12/library/tracemalloc.html>`_
    package.
-   (:pull: `5948`)
+   (:pull:`5948`)
 
 #. `@pp-mo`_ added a benchmark 'trialrun' sub-command, to quickly test
-   benchmarks during development. (:pull: `5957`)
+   benchmarks during development. (:pull:`5957`)
 
 #. `@pp-mo`_ moved several memory-measurement benchmarks from 'on-demand' to
-   the standard set, in hopes that use of 'tracemalloc' (:pull: `5948`) makes
+   the standard set, in hopes that use of 'tracemalloc' (:pull:`5948`) makes
    the results consistent enough to monitor for performance changes.
-   (:pull: `5959`)
+   (:pull:`5959`)
+
+#. `@rcomer`_ made some :meth:`~iris.cube.Cube.slices_over` tests go faster (:pull:`5973`)
 
 #. `@bouweandela`_ enabled mypy checks for type hints. (:pull:`5956`)
 

lib/iris/analysis/calculus.py

@@ -760,7 +760,7 @@ def spatial_vectors_with_phenom_name(i_cube, j_cube, k_cube=None):
     The cube standard names must match one of the combinations in
     :data:`DIRECTIONAL_NAMES`.
 
-    This routine is designed to identify the vector quantites which each
+    This routine is designed to identify the vector quantities which each
     of the cubes provided represent and return a list of their 3d
     spatial dimension names and associated phenomenon.
     For example, given a cube of "u wind" and "v wind" the return value

lib/iris/cube.py

@@ -4555,12 +4555,6 @@ def rolling_window(self, coord, aggregator, window, **kwargs):
         -------
         :class:`iris.cube.Cube`.
 
-        Notes
-        -----
-        .. note::
-
-            This operation does not yet have support for lazy evaluation.
-
         Examples
         --------
             >>> import iris, iris.analysis
@@ -4664,7 +4658,7 @@ def rolling_window(self, coord, aggregator, window, **kwargs):
         # this will add an extra dimension to the data at dimension + 1 which
         # represents the rolled window (i.e. will have a length of window)
         rolling_window_data = iris.util.rolling_window(
-            self.data, window=window, axis=dimension
+            self.core_data(), window=window, axis=dimension
         )
 
         # now update all of the coordinates to reflect the aggregation
@@ -4683,7 +4677,7 @@ def rolling_window(self, coord, aggregator, window, **kwargs):
                     "coordinate." % coord_.name()
                 )
 
-            new_bounds = iris.util.rolling_window(coord_.points, window)
+            new_bounds = iris.util.rolling_window(coord_.core_points(), window)
 
             if np.issubdtype(new_bounds.dtype, np.str_):
                 # Handle case where the AuxCoord contains string. The points
@@ -4729,9 +4723,12 @@ def rolling_window(self, coord, aggregator, window, **kwargs):
                 kwargs["weights"] = iris.util.broadcast_to_shape(
                     weights, rolling_window_data.shape, (dimension + 1,)
                 )
-        data_result = aggregator.aggregate(
-            rolling_window_data, axis=dimension + 1, **kwargs
-        )
+
+        if aggregator.lazy_func is not None and self.has_lazy_data():
+            agg_method = aggregator.lazy_aggregate
+        else:
+            agg_method = aggregator.aggregate
+        data_result = agg_method(rolling_window_data, axis=dimension + 1, **kwargs)
         result = aggregator.post_process(new_cube, data_result, [coord], **kwargs)
         return result
 

lib/iris/fileformats/netcdf/saver.py

@@ -459,7 +459,7 @@ def write(
         cube : :class:`iris.cube.Cube`
             A :class:`iris.cube.Cube` to be saved to a netCDF file.
         local_keys : iterable of str, optional
-            An interable of cube attribute keys. Any cube attributes with
+            An iterable of cube attribute keys. Any cube attributes with
             matching keys will become attributes on the data variable rather
             than global attributes.
 
@@ -2442,7 +2442,7 @@ def save(
         Underlying netCDF file format, one of 'NETCDF4', 'NETCDF4_CLASSIC',
         'NETCDF3_CLASSIC' or 'NETCDF3_64BIT'. Default is 'NETCDF4' format.
     local_keys : iterable of str, optional
-        An interable of cube attribute keys. Any cube attributes with
+        An iterable of cube attribute keys. Any cube attributes with
         matching keys will become attributes on the data variable rather
         than global attributes.
 

lib/iris/plot.py

@@ -43,6 +43,23 @@
 PlotDefn = collections.namedtuple("PlotDefn", ("coords", "transpose"))
 
 
+class _GeoAxesPatched(cartopy.mpl.geoaxes.GeoAxes):
+    # TODO: see cartopy#2390
+    #  Remove this once the bug is addressed in a Cartopy release.
+    def _draw_preprocess(self, renderer):
+        super()._draw_preprocess(renderer)
+
+        for artist in self.artists:
+            if hasattr(artist, "_draw_gridliner"):
+                # Note this is only necessary since Cartopy v0.23, but is not
+                #  wasteful for earlier versions as _draw_gridliner() includes
+                #  a check for whether a draw is necessary.
+                artist._draw_gridliner(renderer=renderer)
+
+
+cartopy.mpl.geoaxes.GeoAxes = _GeoAxesPatched
+
+
 def _get_plot_defn_custom_coords_picked(cube, coords, mode, ndims=2):
     def names(coords):
         result = []

lib/iris/tests/unit/cube/test_Cube.py

@@ -877,7 +877,7 @@ def setUp(self):
         self.cell_measure = CellMeasure([0, 1, 2, 0, 1, 2], long_name="bar")
         self.multi_dim_cube.add_cell_measure(self.cell_measure, 1)
 
-        self.mock_agg = mock.Mock(spec=Aggregator)
+        self.mock_agg = mock.Mock(spec=Aggregator, lazy_func=None)
         self.mock_agg.aggregate = mock.Mock(return_value=np.empty([4]))
         self.mock_agg.post_process = mock.Mock(side_effect=lambda x, y, z: x)
 
@@ -919,6 +919,21 @@ def test_kwargs(self):
         )
         self.assertMaskedArrayEqual(expected_result, res_cube.data)
 
+    def test_lazy(self):
+        window = 2
+        self.cube.data = da.ma.masked_array(
+            self.cube.data, mask=([True, False, False, False, True, False])
+        )
+        res_cube = self.cube.rolling_window("val", iris.analysis.MEAN, window, mdtol=0)
+        self.assertTrue(self.cube.has_lazy_data())
+        self.assertTrue(res_cube.has_lazy_data())
+        expected_result = ma.array(
+            [-99.0, 1.5, 2.5, -99.0, -99.0],
+            mask=[True, False, False, True, True],
+            dtype=np.float64,
+        )
+        self.assertMaskedArrayEqual(expected_result, res_cube.data)
+
     def test_ancillary_variables_and_cell_measures_kept(self):
         res_cube = self.multi_dim_cube.rolling_window("val", self.mock_agg, 3)
         self.assertEqual(res_cube.ancillary_variables(), [self.ancillary_variable])
@@ -1036,10 +1051,10 @@ def test_all_permutations(self):
 @tests.skip_data
 class Test_slices_over(tests.IrisTest):
     def setUp(self):
-        self.cube = stock.realistic_4d()
+        self.cube = stock.realistic_4d()[:, :7, :10, :10]
         # Define expected iterators for 1D and 2D test cases.
         self.exp_iter_1d = range(len(self.cube.coord("model_level_number").points))
-        self.exp_iter_2d = np.ndindex(6, 70, 1, 1)
+        self.exp_iter_2d = np.ndindex(6, 7, 1, 1)
         # Define maximum number of interactions for particularly long
         # (and so time-consuming) iterators.
         self.long_iterator_max = 5
@@ -1075,7 +1090,7 @@ def test_1d_slice_nonexistent_dimension_given(self):
             _ = self.cube.slices_over(self.cube.ndim + 1)
 
     def test_2d_slice_coord_given(self):
-        # Slicing over these two dimensions returns 420 2D cubes, so only check
+        # Slicing over these two dimensions returns 42 2D cubes, so only check
         # cubes up to `self.long_iterator_max` to keep test runtime sensible.
         res = self.cube.slices_over(
             [self.cube.coord("time"), self.cube.coord("model_level_number")]
@@ -1094,7 +1109,7 @@ def test_2d_slice_nonexistent_coord_given(self):
             )
 
     def test_2d_slice_coord_name_given(self):
-        # Slicing over these two dimensions returns 420 2D cubes, so only check
+        # Slicing over these two dimensions returns 42 2D cubes, so only check
         # cubes up to `self.long_iterator_max` to keep test runtime sensible.
         res = self.cube.slices_over(["time", "model_level_number"])
         for ct in range(self.long_iterator_max):
@@ -1109,7 +1124,7 @@ def test_2d_slice_nonexistent_coord_name_given(self):
             _ = self.cube.slices_over(["time", "wibble"])
 
     def test_2d_slice_dimension_given(self):
-        # Slicing over these two dimensions returns 420 2D cubes, so only check
+        # Slicing over these two dimensions returns 42 2D cubes, so only check
         # cubes up to `self.long_iterator_max` to keep test runtime sensible.
         res = self.cube.slices_over([0, 1])
         for ct in range(self.long_iterator_max):
@@ -1135,11 +1150,11 @@ def test_2d_slice_nonexistent_dimension_given(self):
             _ = self.cube.slices_over([0, self.cube.ndim + 1])
 
     def test_multidim_slice_coord_given(self):
-        # Slicing over surface altitude returns 100x100 2D cubes, so only check
+        # Slicing over surface altitude returns 10x10 2D cubes, so only check
         # cubes up to `self.long_iterator_max` to keep test runtime sensible.
         res = self.cube.slices_over("surface_altitude")
         # Define special ndindex iterator for the different dims sliced over.
-        nditer = np.ndindex(1, 1, 100, 100)
+        nditer = np.ndindex(1, 1, 10, 10)
         for ct in range(self.long_iterator_max):
             indices = list(next(nditer))
             # Replace the dimensions not iterated over with spanning slices.