BUG: Use Python pickle protocol version 4 for np.save (numpy#26388)

* BUG: Use default Python pickle protocol version rather than outdated protocol 3 * Update default pickle of np.save to 4 * Adds document for default pickle protocol 4
charris · May 16, 2024 · 892af5a · 892af5a
1 parent c2f0f17
commit 892af5a
Show file tree

Hide file tree

Showing 3 changed files with 5 additions and 2 deletions.
diff --git a/doc/release/upcoming_changes/26388.performance.rst b/doc/release/upcoming_changes/26388.performance.rst
@@ -0,0 +1,3 @@
+ * `numpy.save` now uses pickle protocol version 4 for saving arrays with
+   object dtype, which allows for pickle objects larger than 4GB and improves
+   saving speed by about 5% for large arrays.
diff --git a/numpy/lib/_npyio_impl.py b/numpy/lib/_npyio_impl.py
@@ -520,7 +520,7 @@ def save(file, arr, allow_pickle=True, fix_imports=True):
         arbitrary code) and portability (pickled objects may not be loadable
         on different Python installations, for example if the stored objects
         require libraries that are not available, and not all pickled data is
-        compatible between Python 2 and Python 3).
+        compatible between different versions of Python).
         Default: True
     fix_imports : bool, optional
         Only useful in forcing objects in object arrays on Python 3 to be

diff --git a/numpy/lib/format.py b/numpy/lib/format.py
@@ -741,7 +741,7 @@ def write_array(fp, array, version=None, allow_pickle=True, pickle_kwargs=None):
                                  "when allow_pickle=False")
         if pickle_kwargs is None:
             pickle_kwargs = {}
-        pickle.dump(array, fp, protocol=3, **pickle_kwargs)
+        pickle.dump(array, fp, protocol=4, **pickle_kwargs)
     elif array.flags.f_contiguous and not array.flags.c_contiguous:
         if isfileobj(fp):
             array.T.tofile(fp)