From 8c213f0c5e8402b65b19a47ec8bc560933b0fd6f Mon Sep 17 00:00:00 2001
From: Bouwe Andela <b.andela@esciencecenter.nl>
Date: Fri, 25 Oct 2024 11:49:06 +0200
Subject: [PATCH] Improve handling of masks in concatenate (#6187)

* Improve handling of masks

* Add whatsnew
---
 docs/src/whatsnew/latest.rst                    |  2 +-
 lib/iris/_concatenate.py                        |  2 +-
 lib/iris/tests/unit/concatenate/test_hashing.py | 17 +++++++++++++++++
 3 files changed, 19 insertions(+), 2 deletions(-)

diff --git a/docs/src/whatsnew/latest.rst b/docs/src/whatsnew/latest.rst
index 1ef00079e3..73cc419c5e 100644
--- a/docs/src/whatsnew/latest.rst
+++ b/docs/src/whatsnew/latest.rst
@@ -82,7 +82,7 @@ This document explains the changes made to Iris for this release
    may happen when there are very many or large auxiliary coordinates, derived
    coordinates, cell measures, or ancillary variables to be checked that span
    the concatenation axis. This issue can be avoided by disabling the
-   problematic check. (:pull:`5926`)
+   problematic check. (:pull:`5926` and :pull:`6187`)
 
 🔥 Deprecations
 ===============
diff --git a/lib/iris/_concatenate.py b/lib/iris/_concatenate.py
index 841cecfd53..ac9e699790 100644
--- a/lib/iris/_concatenate.py
+++ b/lib/iris/_concatenate.py
@@ -310,7 +310,7 @@ def _hash_ndarray(a: np.ndarray) -> np.ndarray:
 
     # Hash the bytes representing the array data.
     hash.update(b"data=")
-    if isinstance(a, np.ma.MaskedArray):
+    if np.ma.is_masked(a):
         # Hash only the unmasked data
         hash.update(a.compressed().tobytes())
         # Hash the mask
diff --git a/lib/iris/tests/unit/concatenate/test_hashing.py b/lib/iris/tests/unit/concatenate/test_hashing.py
index 7a56be1db8..24062a2af3 100644
--- a/lib/iris/tests/unit/concatenate/test_hashing.py
+++ b/lib/iris/tests/unit/concatenate/test_hashing.py
@@ -21,6 +21,8 @@
         (np.array([np.nan, 1.0]), np.array([np.nan, 1.0]), True),
         (np.ma.array([1, 2], mask=[0, 1]), np.ma.array([1, 2], mask=[0, 1]), True),
         (np.ma.array([1, 2], mask=[0, 1]), np.ma.array([1, 2], mask=[0, 0]), False),
+        (np.ma.array([1, 2], mask=[1, 1]), np.ma.array([1, 2], mask=[1, 1]), True),
+        (np.ma.array([1, 2], mask=[0, 0]), np.ma.array([1, 2], mask=[0, 0]), True),
         (da.arange(6).reshape((2, 3)), da.arange(6, chunks=1).reshape((2, 3)), True),
         (da.arange(20, chunks=1), da.arange(20, chunks=2), True),
         (
@@ -33,6 +35,21 @@
             da.ma.masked_array([1, 3], mask=[0, 1]),
             True,
         ),
+        (
+            np.arange(2),
+            da.ma.masked_array(np.arange(2), mask=[0, 0]),
+            True,
+        ),
+        (
+            np.arange(2),
+            da.ma.masked_array(np.arange(2), mask=[0, 1]),
+            False,
+        ),
+        (
+            da.ma.masked_array(np.arange(10), mask=np.zeros(10)),
+            da.ma.masked_array(np.arange(10), mask=np.ma.nomask),
+            True,
+        ),
         (
             np.ma.array([1, 2], mask=[0, 1]),
             np.ma.array([1, 3], mask=[0, 1], fill_value=10),