Script 'mail_helper' called by obssrc Hello community, here is the log from the commit of package python-zarr for openSUSE:Factory checked in at 2026-10-18 19:26:07 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Comparing /work/SRC/openSUSE:Factory/python-zarr (Old) and /work/SRC/openSUSE:Factory/.python-zarr.new.3698736 (New) ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Package is "python-zarr" Sun Oct 18 19:26:07 2026 rev:22 rq:1385968 version:3.4.1 Changes: -------- --- /work/SRC/openSUSE:Factory/python-zarr/python-zarr.changes 2026-05-20 15:25:32.908678988 +0200 +++ /work/SRC/openSUSE:Factory/.python-zarr.new.3698736/python-zarr.changes 2026-10-18 19:26:48.890520767 +0200 @@ -1,0 +2,75 @@ +Sun Oct 18 13:05:22 UTC 2026 - Ben Greiner <[email protected]> + +- Skip big memory tests on 32bit + +------------------------------------------------------------------- +Thu Oct 15 04:51:43 UTC 2026 - Steve Kowalik <[email protected]> + +- Update to 3.4.1: + * perf: cut the fill_value check on the chunk write path + * fix(dtype): accept any byte order for byte-order-irrelevant v2 dtype + names + * fix(ci): run zarr-indexing tests on the matrix interpreter + * fix(sharding): reject variable-size codecs in index_codecs + * fix(dtype): fix nested v2 round trips and detect unsupported field + conversions + * fix(dtype): resolve struct fields with the registry that resolves the + struct + * fix(store): run every LocalStore filesystem call off the event loop + * fix(storage): yield only direct children from FsspecStore.list_dir + * fix(dtype): preserve scaled generic time units and normalize the us alias + * fix(array): reject unknown keys in ArrayConfig.from_dict + * fix(indexing): avoid full-grid allocations for sparse selections + * fix: byte-order bugs and little-endian test assumptions on big-endian + hosts + * fix: honor storage_options in save and save_group + * fix: minor correctness and hygiene fixes from the sync-pipeline audit + * fix: reject malformed chunk keys in DefaultChunkKeyEncoding + * fix: make consolidated metadata nesting independent of persisted key + order + * fix: preserve non-JSON-serializable storage options in _make_async + * fix: accept numpy integers as chunk sizes + * Add unified JSON metadata validation via msgspec + * fix: accept universal-pathlib UPath as a StoreLike value + * feat: name packages that provide a codec zarr cannot find + * Fix infinite loop in _guess_num_chunks_per_axis_shard for 0-dimensional + arrays + * BUG: Fix empty Group nodes lacking a "consolidated_metadata" field + * fix(zarr-metadata): model stored metadata more closely + * fix: allow writing to 0-dimensional arrays with sharding + * perf: cache default ArraySpec for regular chunk grids + * refactor(sharding): store chunks_per_shard explicitly in _ShardIndex + * Optimize partial shard reads + * Widen ChunksLike type alias + * prune old string dtype class + * fix: compare metadata objects + * fix(FsspecStore): close owned async filesystem on store.close() + * fix: make xarray downstream tests work + * chore: indexing test cleanup + * Modify custom dtype example to use only public API + * fix: copy attributes passed from the user to avoid mutation + * perf: cache lexicographic chunk coords in sharding codec + * deprecate more enums + * Control sharding codec read coalescing with ArrayConfig and runtime + config options + * feat![zarr-metadata]: extensive top-level exports + * Remove struct unstable warning + * Fix save/load for scalar arrays + * fix: skip object store prefix markers in list_dir + * fix: byte-order handling for structured dtypes in the bytes codec + * Fix/memorystore buffer aliasing + * feat(zarr-metadata): Zarr metadata model layer + * fix: let open_like create arrays by default + * fix: keep FusedCodecPipeline compute off the event-loop thread + * fix(ArraySpec): proper and robust equality semantics for ArraySpec by + checking fill_value for byte-identicality + * perf(indexing): speed up sorted 1-D coordinate selections + * fix: FusedCodecPipeline falls back to async path for sharded arrays with + async-only inner codecs + * fix: fused pipeline falls back for partial-mixin codecs without sync + partial methods + * fix: ManagedMemoryStore sync methods must respect the path prefix + * fix: minor correctness and hygiene fixes from the sync-pipeline audit + * fix: reject malformed chunk keys in DefaultChunkKeyEncoding + +------------------------------------------------------------------- Old: ---- zarr-3.2.1.tar.gz New: ---- zarr-3.4.1.tar.gz ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Other differences: ------------------ ++++++ python-zarr.spec ++++++ --- /var/tmp/diff_new_pack.xBvQbh/_old 2026-10-18 19:26:49.637552036 +0200 +++ /var/tmp/diff_new_pack.xBvQbh/_new 2026-10-18 19:26:49.639552120 +0200 @@ -22,9 +22,8 @@ %bcond_with libalternatives %endif %{?sle15_python_module_pythons} -%global skip_python311 1 Name: python-zarr -Version: 3.2.1 +Version: 3.4.1 Release: 0 Summary: An implementation of chunked, compressed, N-dimensional arrays for Python License: MIT @@ -41,10 +40,11 @@ Requires: python >= 3.12 Requires: python-donfig >= 0.8 Requires: python-google-crc32c >= 1.5 -Requires: python-numcodecs >= 0.14 +Requires: python-msgspec >= 0.19 +Requires: python-numcodecs >= 0.16 Requires: python-numpy >= 2.0 Requires: python-packaging >= 22 -Requires: python-typing_extensions >= 4.13 +Requires: python-typing_extensions >= 4.14 Suggests: python-dbm Suggests: python-ipytree BuildArch: noarch @@ -60,7 +60,8 @@ BuildRequires: %{python_module donfig >= 0.8} BuildRequires: %{python_module google-crc32c >= 1.5} BuildRequires: %{python_module hypothesis} -BuildRequires: %{python_module numcodecs >= 0.14} +BuildRequires: %{python_module msgspec >= 0.19} +BuildRequires: %{python_module numcodecs >= 0.16} BuildRequires: %{python_module numpy >= 2.0} BuildRequires: %{python_module numpydoc} BuildRequires: %{python_module pytest-asyncio} @@ -68,7 +69,7 @@ BuildRequires: %{python_module pytest-xdist} BuildRequires: %{python_module pytest} BuildRequires: %{python_module tomlkit} -BuildRequires: %{python_module typing_extensions >= 4.13} +BuildRequires: %{python_module typing_extensions >= 4.14} # /SECTION %python_subpackages @@ -90,19 +91,18 @@ rm -v tests/test_examples.py # https://pytest-xdist.readthedocs.io/en/stable/known-limitations.html donttestparallel="test_docstring_consistent_parameters" -# avoid broken tests in s390x, gh#zarr-developers/zarr-python#1375 -%if "%_arch" == "s390x" -donttest+=" or test_hexdigest or test_nbytes_stored" -donttest+=" or test_array_1d or test_array_1d_fill_value or test_array_1d_selections" -donttest+=" or test_array_2d or test_array_2d_edge_case or test_array_order" -donttest+=" or test_resize_2d or test_append_2d or test_append_2d_axis" -donttest+=" or test_np_ufuncs or test_iter or test_islice or test_non_cont" -donttest+=" or test_read_nitems_less_than_blocksize_from_multiple_chunks" -donttest+=" or test_read_from_all_blocks" -donttest+=" or test_format_compatibility" -%endif -%pytest -n auto -k "not ($donttestparallel $donttest)" -%pytest -k "$donttestparallel" +# 32-bit failures in tests: not able to allocate that much memory +if [ $(getconf LONG_BIT) -eq 32 ]; then + donttest+=" or (test_e2e and (test_read_array or test_write_array) and batched)" + donttest+=" or test_sparse_selections_on_arrays_with_many_chunks" + donttest+=" or test_sparse_indexer_projection_reconstructs_coordinates" + donttest+=" or test_coordinate_indexer_many_chunks" +fi +# these take long +donttest+=" or (test_indexing and test_sharded_morton and memory)" +donttest+=" or test_basic_indexing_complex_rectilinear" +%pytest -n auto -k "not ($donttestparallel $donttest)" --ignore src/zarr +%pytest -k "$donttestparallel" --ignore src/zarr %pre %python_libalternatives_reset_alternative zarr ++++++ zarr-3.2.1.tar.gz -> zarr-3.4.1.tar.gz ++++++ ++++ 64753 lines of diff (skipped)
