Deprecate partial read/writes in v2 (#2844)

* Deprecated partial read/writes in v2

* Remove unused imports

* Filter warnings on more partial read/write tests

Author: dstansby

docs/release.rst

@@ -14,9 +14,19 @@ Release notes
     # re-indented so that it does not show up in the notes.
 
 .. note::
-   Zarr-Python 2.18.* is expected be the final release in the 2.* series. Work on Zarr-Python 3.0 is underway.
-   See `GH1777 <https://github.com/zarr-developers/zarr-python/issues/1777>`_ for more details on the upcoming
-   3.0 release.
+   Zarr-Python 2.* is in support mode now, and no new features will be added.
+
+
+Unreleased
+----------
+
+Deprecations
+~~~~~~~~~~~~
+
+* Deprecated support for ``partial_decompress`` when creating an array.
+  This functionality is no longer supported in ``numcodecs``, and will be removed
+  in ``zarr-python`` 2.19.0.
+  By :user:`David Stansby <dstansby>`
 
 .. _release_2.18.4:
 
@@ -40,9 +50,6 @@ Maintenance
   the Delta filter (see https://github.com/zarr-developers/numcodecs/issues/653 for more information).
   By :user:`David Stansby <dstansby>` (:issue:`2544`).
 
-Deprecations
-~~~~~~~~~~~~
-
 .. _release_2.18.3:
 
 2.18.3

zarr/core.py

@@ -6,6 +6,7 @@
 import re
 from functools import reduce
 from typing import Any
+import warnings
 
 import numpy as np
 from numcodecs.compat import ensure_bytes
@@ -90,13 +91,6 @@ class Array:
         If True (default), user attributes will be cached for attribute read
         operations. If False, user attributes are reloaded from the store prior
         to all attribute read operations.
-    partial_decompress : bool, optional
-        If True and while the chunk_store is a FSStore and the compression used
-        is Blosc, when getting data from the array chunks will be partially
-        read and decompressed when possible.
-
-        .. versionadded:: 2.7
-
     write_empty_chunks : bool, optional
         If True, all chunks will be stored regardless of their contents. If
         False (default), each chunk is compared to the array's fill value prior
@@ -124,7 +118,7 @@ def __init__(
         synchronizer=None,
         cache_metadata=True,
         cache_attrs=True,
-        partial_decompress=False,
+        partial_decompress=None,
         write_empty_chunks=True,
         zarr_version=None,
         meta_array=None,
@@ -154,6 +148,13 @@ def __init__(
         self._synchronizer = synchronizer
         self._cache_metadata = cache_metadata
         self._is_view = False
+        if partial_decompress is not None:
+            warnings.warn(
+                "Support for partial decompression is no longer supported in numcodecs. "
+                "Support for partial decompression will be removed in a future version of zarr-python v2.",
+                DeprecationWarning,
+                stacklevel=1,
+            )
         self._partial_decompress = partial_decompress
         self._write_empty_chunks = write_empty_chunks
         if meta_array is not None:

zarr/creation.py

@@ -466,7 +466,7 @@ def open_array(
     object_codec=None,
     chunk_store=None,
     storage_options=None,
-    partial_decompress=False,
+    partial_decompress=None,
     write_empty_chunks=True,
     *,
     zarr_version=None,
@@ -522,10 +522,6 @@ def open_array(
     storage_options : dict
         If using an fsspec URL to create the store, these will be passed to
         the backend implementation. Ignored otherwise.
-    partial_decompress : bool, optional
-        If True and while the chunk_store is a FSStore and the compression used
-        is Blosc, when getting data from the array chunks will be partially
-        read and decompressed when possible.
     write_empty_chunks : bool, optional
         If True (default), all chunks will be stored regardless of their
         contents. If False, each chunk is compared to the array's fill value

zarr/tests/test_core.py

@@ -2,6 +2,7 @@
 import sys
 import pickle
 import shutil
+
 from typing import Any, Literal, Optional, Tuple, Union, Sequence
 import unittest
 from itertools import zip_longest
@@ -84,6 +85,11 @@
 
 # noinspection PyMethodMayBeStatic
 
+pytestmark = [
+    pytest.mark.filterwarnings("ignore:Call to deprecated function .* \_cbuffer\_sizes.*"),
+    pytest.mark.filterwarnings("ignore:Call to deprecated function .* \_cbuffer\_metainfo.*"),
+]
+
 
 class TestArray:
     version = 2
@@ -94,7 +100,7 @@ class TestArray:
     dimension_separator: Optional[DIMENSION_SEPARATOR] = None
     cache_metadata = True
     cache_attrs = True
-    partial_decompress: bool = False
+    partial_decompress: bool | None = None
     write_empty_chunks = True
     read_only = False
     storage_transformers: Tuple[Any, ...] = ()
@@ -2481,6 +2487,9 @@ def expected(self):
 
 
 @pytest.mark.skipif(have_fsspec is False, reason="needs fsspec")
+@pytest.mark.filterwarnings(
+    "ignore:.*Support for partial decompression will be removed in a future version.*"
+)
 class TestArrayWithFSStorePartialRead(TestArray):
     compressor = Blosc(blocksize=256)
     partial_decompress = True
@@ -2547,6 +2556,9 @@ def expected(self):
 
 
 @pytest.mark.skipif(have_fsspec is False, reason="needs fsspec")
+@pytest.mark.filterwarnings(
+    "ignore:.*Support for partial decompression will be removed in a future version.*"
+)
 class TestArrayWithFSStoreNestedPartialRead(TestArrayWithFSStore):
     compressor = Blosc()
     dimension_separator = "/"
@@ -3020,6 +3032,9 @@ def expected(self):
 
 @pytest.mark.skipif(have_fsspec is False, reason="needs fsspec")
 @pytest.mark.skipif(not v3_api_available, reason="V3 is disabled")
+@pytest.mark.filterwarnings(
+    "ignore:.*Support for partial decompression will be removed in a future version.*"
+)
 class TestArrayWithFSStoreV3PartialRead(TestArrayWithFSStoreV3):
     partial_decompress = True
 
@@ -3038,6 +3053,9 @@ def expected(self):
 @pytest.mark.skipif(have_fsspec is False, reason="needs fsspec")
 @pytest.mark.skipif(not v3_api_available, reason="V3 is disabled")
 @pytest.mark.skipif(not v3_sharding_available, reason="sharding is disabled")
+@pytest.mark.filterwarnings(
+    "ignore:.*Support for partial decompression will be removed in a future version.*"
+)
 class TestArrayWithFSStoreV3PartialReadUncompressedSharded(TestArrayWithFSStoreV3):
     partial_decompress = True
     compressor = None