Only cache hashes on *lazy* arrays.

Author: pp-mo

lib/iris/tests/unit/coords/test_AuxCoord.py

@@ -775,23 +775,29 @@ def test_nanbounds_eq_self(self):
         co1 = AuxCoord([15.0, 25.0], bounds=[[14.0, 16.0], [24.0, np.nan]])
         assert co1 == co1
 
-    def test_lazy_compares_via_hash(self):
+    @pytest.mark.parametrize("bothlazy", [True, False], ids=["bothlazy", "onelazy"])
+    def test_lazy_compares_via_hash(self, bothlazy):
         def lazify(coord):
             coord.bounds = coord.lazy_bounds()
 
         co1 = AuxCoord([15.0, 25.0])
         co2 = AuxCoord([15.0, 25.001])
+
         co1.points = co1.lazy_points()
-        co2.points = co2.lazy_points()
+        if bothlazy:
+            co2.points = co2.lazy_points()
         assert co1.has_lazy_points()
-        assert co2.has_lazy_points()
+        assert co2.has_lazy_points() == bothlazy
+
         assert not hasattr(co1.core_points(), "_iris_array_hash")
-        assert not hasattr(co2.core_points(), "_iris_array_hash")
+        if bothlazy:
+            assert not hasattr(co2.core_points(), "_iris_array_hash")
 
         eq = co1 == co2
         assert not eq
 
         assert co1.has_lazy_points()
-        assert co2.has_lazy_points()
         assert hasattr(co1.core_points(), "_iris_array_hash")
-        assert hasattr(co2.core_points(), "_iris_array_hash")
+        if bothlazy:
+            assert co2.has_lazy_points()
+            assert hasattr(co2.core_points(), "_iris_array_hash")

lib/iris/util.py

@@ -511,12 +511,16 @@ def normalise_array(array):
         if is_lazy_data(array1) or is_lazy_data(array2):
             # Compare lazy arrays by hashing, and cache the hashes...
             def array_hash(array):
-                if not hasattr(array, "_iris_array_hash"):
+                if hasattr(array, "_iris_array_hash"):
+                    hash = array._iris_array_hash
+                else:
                     from iris._concatenate import _hash_array
 
                     hash = _hash_array(array)
-                    array._iris_array_hash = hash
-                return array._iris_array_hash
+                    if isinstance(array, da.Array):
+                        # Can't save hashes on a numpy array, but CAN on a Dask array
+                        array._iris_array_hash = hash
+                return hash
 
             eq = array_hash(array1) == array_hash(array2)
             # # Use a separate map and reduce operation to avoid running out of memory.