Can't import `zarr<3` with latest `numcodecs==0.16.0` #2963

ryan-williams · 2025-04-07T15:48:11Z

Zarr version

v2.18.5

Numcodecs version

v0.16.0

Python Version

3.11 (but probably all)

Operating System

Ubuntu and Mac (but probably all)

Installation

pip install 'zarr<3' numcodecs

Description

In zarr<3, zarr/util.py imports numcodecs.blosc.cbuffer_sizes (e.g. here in 2.18.5), which was removed in numcodecs 0.16.0 (released 40mins ago at time of writing)

This reference was removed in #2182, prior to 3.0.0, but causes zarr 2.x to fail to import when latest numcodecs is installed; GHA example:

  File "/home/runner/work/TileDB-SOMA/TileDB-SOMA/venv-soma/lib/python3.11/site-packages/zarr/convenience.py", line 8, in <module>
    from zarr._storage.store import data_root, meta_root, assert_zarr_v3_api_available
  File "/home/runner/work/TileDB-SOMA/TileDB-SOMA/venv-soma/lib/python3.11/site-packages/zarr/_storage/store.py", line 10, in <module>
    from zarr.meta import Metadata2, Metadata3
  File "/home/runner/work/TileDB-SOMA/TileDB-SOMA/venv-soma/lib/python3.11/site-packages/zarr/meta.py", line 10, in <module>
    from zarr.util import json_dumps, json_loads
  File "/home/runner/work/TileDB-SOMA/TileDB-SOMA/venv-soma/lib/python3.11/site-packages/zarr/util.py", line 34, in <module>
    from numcodecs.blosc import cbuffer_sizes, cbuffer_metainfo
ImportError: cannot import name 'cbuffer_sizes' from 'numcodecs.blosc' (/home/runner/work/TileDB-SOMA/TileDB-SOMA/venv-soma/lib/python3.11/site-packages/numcodecs/blosc.cpython-311-x86_64-linux-gnu.so)
Error: Process completed with exit code 1.

Not sure it's worth a 2.18.6 to pin numcodecs<0.16 (or remove the offending references), just flagging the issue.

Steps to reproduce

python -m venv .venv
. .venv/bin/activate
pip install 'zarr<3' numcodecs
python -c 'import zarr'  # ❌ ImportError: cannot import name 'cbuffer_sizes' from 'numcodecs.blosc' (/Users/ryan/c/single-cell-data/tiledb-soma/tmp/.venv/lib/python3.11/site-packages/numcodecs/blosc.cpython-311-darwin.so)

or:

docker run --rm --entrypoint bash python:3.11 -c 'pip install "zarr<3" numcodecs && python -c "import zarr"'
# ❌ ImportError: cannot import name 'cbuffer_sizes' from 'numcodecs.blosc' (/usr/local/lib/python3.11/site-packages/numcodecs/blosc.cpython-311-x86_64-linux-gnu.so)

Additional output

No response

The text was updated successfully, but these errors were encountered:

cf. zarr-developers/zarr-python#2963

dstansby · 2025-04-07T17:08:41Z

Thanks for the quick report, and sorry about this. I think the best thing to do is pin numcodecs here, I'll do that now.

cspades · 2025-04-07T17:29:46Z

Also hit this error, which affects NeMo:

/__w/_tool/Python/3.12.9/x64/lib/python3.12/site-packages/zarr/meta.py:10: in <module>
    from zarr.util import json_dumps, json_loads
/__w/_tool/Python/3.12.9/x64/lib/python3.12/site-packages/zarr/util.py:34: in <module>
    from numcodecs.blosc import cbuffer_sizes, cbuffer_metainfo
E   ImportError: cannot import name 'cbuffer_sizes' from 'numcodecs.blosc' (/__w/_tool/Python/3.12.9/x64/lib/python3.12/site-packages/numcodecs/blosc.cpython-312-x86_64-linux-gnu.so)

Following for the fix, thank you!

* `setup.py`: remove deprecated `License ::` from `classifiers` [sc-65131] * pin `numcodecs<0.16` cf. zarr-developers/zarr-python#2963

…#3918) * `setup.py`: remove deprecated `License ::` from `classifiers` [sc-65131] * pin `numcodecs<0.16` cf. zarr-developers/zarr-python#2963 Co-authored-by: Ryan Williams <ryan.williams@tiledb.com>

dstansby · 2025-04-08T13:26:31Z

This should now be fixed with a new 2.18.6 release - would be great if someone could confirm before we close this.

tcompa · 2025-04-08T13:41:19Z

Hi there

This should now be fixed with a new 2.18.6 release - would be great if someone could confirm before we close this.

Something is still not working for me:

$ python --version
Python 3.11.11

$ pip install "zarr<3" numcodecs
[...]

$ pip freeze
asciitree==0.3.3
fasteners==0.19
numcodecs==0.16.0
numpy==2.2.4
typing_extensions==4.13.1
zarr==2.18.6

At a first look, I think that this is because #2965 modified the docs entry for numcodecs (https://github.com/zarr-developers/zarr-python/blob/support/v2/pyproject.toml#L45), rather than the actual line (https://github.com/zarr-developers/zarr-python/blob/support/v2/pyproject.toml#L16).

dstansby · 2025-04-08T15:12:31Z

😱 thanks for checking - working on a working fix...

gastoneb · 2025-04-08T20:52:48Z

Experiencing the same issue even with older zarr versions. Workaround until a fix is available is to explicitly require numcodecs==0.15.0 in your deps, since it will always try to install 0.16 otherwise.

dstansby · 2025-04-09T10:49:37Z

Right, I'm pretty sure this is fixed now - just did a fresh install, and with zarr 2.18.7 it pulls numcodecs 0.15.1. Could someone else double check? Sorry again for the disruption.

schenker · 2025-04-09T11:48:31Z

With zarr 2.18.7 it works as expected, thanks!

`zarr==2.18.7` removes the need for it, cf.: - zarr-developers/zarr-python#2963 - #3917

`zarr==2.18.7` removes the need for it, cf.: - zarr-developers/zarr-python#2963 - #3917 Co-authored-by: Ryan Williams <ryan.williams@tiledb.com>

This was fixed in the latest Zarr 2.18.7 release, but for some reason it doesn't get installed in GitHub actions: zarr-developers/zarr-python#2963 PiperOrigin-RevId: 750643134

This was fixed in the latest Zarr 2.18.7 release, but for some reason it doesn't get installed in GitHub actions: zarr-developers/zarr-python#2963 PiperOrigin-RevId: 750778532

@bkmartinjr

commit 777dcfc Author: Bruce Martin <bruce.martin@tiledb.com> Date: Thu Apr 24 16:10:09 2025 -0700 skip redundant metadata writes (#4014) commit 7448877 Author: XanthosXanthopoulos <38084549+XanthosXanthopoulos@users.noreply.github.com> Date: Thu Apr 24 08:37:46 2025 -0700 Fix error introduced by renaming lambda (#4012) commit 435a522 Author: XanthosXanthopoulos <38084549+XanthosXanthopoulos@users.noreply.github.com> Date: Thu Apr 24 08:22:28 2025 -0700 [c++] Optimize `nnz` computation (#3898) * count cell on overlapping region only * Use count aggregate for slow nnz * Fix iteration * Add unit test * Update libtiledbsoma/src/soma/soma_array.cc Co-authored-by: John Kerl <kerl.john.r@gmail.com> * Add tracking issue * Fix memory leak in unit test * Do not early fall back on `nnz_slow` in partial overlap/consolidated fragment * Extract common logic to lambda function * Apply PRs 3958 and 4001 * Permutate relevant fragment list based on non emprty domain sorting * Add randomized overlap nnz test * Fix spelling --------- Co-authored-by: John Kerl <kerl.john.r@gmail.com> commit af82ab8 Author: John Kerl <kerl.john.r@gmail.com> Date: Wed Apr 23 13:25:20 2025 -0400 [python] Improve null-index handling in Arrow dictionaries (#3992) * Unit-test cases * Minimal change for segv-avoidance * iterating * finalize unit-test cases; next: simplify and eludicate * neaten * prep for code-dedupe * more prep for code-deduping * continuing * complete dedupe * add some comment material * add bool-valued tests * space/time tradeoff * revert previous * Apply suggestions from code review Co-authored-by: nguyenv <vivian@tiledb.com> * code-review feedback, and fix an error from another suggestion --------- Co-authored-by: nguyenv <vivian@tiledb.com> commit 07e9281 Author: John Kerl <kerl.john.r@gmail.com> Date: Wed Apr 23 09:02:05 2025 -0400 apis/python/example-scripts/ingest_h5ads.py (#4006) commit 13de2c7 Author: John Kerl <kerl.john.r@gmail.com> Date: Tue Apr 22 16:16:29 2025 -0400 [c++] Trivial method-order rearrange (#3996) * [c++] Trivial method-order rearrange * one more spelling uniformization commit 58baebb Author: Bruce Martin <bruce.martin@tiledb.com> Date: Tue Apr 22 08:39:24 2025 -0700 fix bad test (#4002) commit 63d8ae1 Author: John Kerl <kerl.john.r@gmail.com> Date: Mon Apr 21 17:38:53 2025 -0400 [c++] Fix error in raise-if-slow path (#3999) * [c++] Fix error in raise-if-slow path * Workaround spurious CI failure commit 1326a02 Author: nguyenv <vivian@tiledb.com> Date: Wed Apr 16 15:48:38 2025 -0500 [python] Have `SOMAFileHandle` encapsulate both `VFS` and `VFSFilebuf` (#3986) * [python] Hold `SOMAContext` smart pointer in `SOMAVFS` * [wip] Refactor SOMAVFS * [wip] Rename to FileHandle and check if open * [wip] Open in constructor * Remove uri member commit 1d3d18d Author: Bruce Martin <bruce.martin@tiledb.com> Date: Tue Apr 15 15:42:10 2025 -0700 h5py cleanup race fix (#3984) * h5py cleanup race fix * revert bug commit 5268209 Author: Ryan Williams <ryan.williams@tiledb.com> Date: Tue Apr 15 10:37:25 2025 -0400 [python] `SOMAObject.open`: rm unused `clib_type` kwarg (#3954) * `SOMAObject.open: rm unused `clib_type` kwarg * fix comment typos cf. #2820 * fix error-msg typo cf. #3847 commit 7b9684b Author: John Kerl <kerl.john.r@gmail.com> Date: Tue Apr 15 09:08:01 2025 -0400 [python] Test `get_experiment_shapes` for underpopulated experiments (#3966) commit 615561a Author: nguyenv <vivian@tiledb.com> Date: Mon Apr 14 17:21:30 2025 -0500 [python][c++] Correctly manage lifetime of `SOMAVFS` with `SOMAVFSFilebuf` (#3976) * [python] Correctly manage lifetime of `SOMAVFS` with `SOMAVFSFilebuf` * Update apis/python/tests/test_basic_anndata_io.py Co-authored-by: John Kerl <kerl.john.r@gmail.com> * collect vfs --------- Co-authored-by: John Kerl <kerl.john.r@gmail.com> commit a8f5e1a Author: Ryan Williams <ryan.williams@tiledb.com> Date: Mon Apr 14 14:18:17 2025 -0400 [python] Harmonize/GC daily-GHA-failure issue-templates (#3973) * update `python-dockers.yml` failure issue template cf. JasonEtco/create-an-issue#86 (comment) * Harmonize/GC daily-GHA-failure issue-templates commit a996f79 Author: Ryan Williams <ryan.williams@tiledb.com> Date: Mon Apr 14 14:18:05 2025 -0400 [python] Remove `typing.{Dict,List,Optional,Set,Tuple,Type,Union}` references (#3956) * rm `Union`s * rm `Tuple`s * rm more `Union`s * rm `List`s * rm `Dict`s * rm `Optional`s * rm `Type`s * rm `Set` commit 64e8e5e Author: XanthosXanthopoulos <38084549+XanthosXanthopoulos@users.noreply.github.com> Date: Mon Apr 14 09:51:14 2025 -0700 [python] Increase unit-test coverage for various metadata value-types (#3972) * Add more unicode sequences * Improve error messaging * Fix macos build issue * Use `bool_` instead of `bool` * Update apis/python/src/tiledbsoma/_tdb_handles.py Co-authored-by: John Kerl <kerl.john.r@gmail.com> --------- Co-authored-by: John Kerl <kerl.john.r@gmail.com> commit 8fce365 Author: Bruce Martin <bruce.martin@tiledb.com> Date: Mon Apr 14 06:57:18 2025 -0700 [python] Fix mishandling of join-ID column (#3970) * fix mishandling of joind column * pr fb commit a6ad758 Author: nguyenv <vivian@tiledb.com> Date: Fri Apr 11 19:17:03 2025 -0500 [python] Ensure `ndarray` byteorder is little-endian on matrix writes (#3963) commit 78df60e Author: John Kerl <kerl.john.r@gmail.com> Date: Fri Apr 11 18:09:46 2025 -0400 [python] Fix a typo on #3949 (#3965) commit 0ccd574 Author: John Kerl <kerl.john.r@gmail.com> Date: Fri Apr 11 16:28:40 2025 -0400 [python] Refactor `io/shaping.py`; `get_experiment_shapes` (#3962) * [python] Refactor `io/shaping.py` * code-review feedback * [python] Implement `tiledbsoma.io.get_experiment_shapes` (#3949) * [python] `get_experiment_shapes` * code-review feedback; rebase coming [skip ci] commit 7d57cc4 Author: nguyenv <vivian@tiledb.com> Date: Thu Apr 10 20:56:14 2025 -0500 [python] Ensure schema is explicitly passed to `pa.Table.from*` converters (#3957) commit c4198c3 Author: John Kerl <kerl.john.r@gmail.com> Date: Thu Apr 10 17:33:47 2025 -0400 [python] Fix a spurious exception-class issue (#3958) * fix an issue raised by ryan [wip] * put py::arg in the right place -- thanks google commit b319af6 Author: Ryan Williams <ryan.williams@tiledb.com> Date: Thu Apr 10 11:06:10 2025 -0400 [python] Remove `pre-commit/mirrors-prettier` hook (#3952) * Use `rbubley/mirrors-prettier` @ 3.5.3 `pre-commit/mirrors-prettier` was archived, this appears to be a reasonably-maintained and -starred fork, which also only updates vs. full (non-alpha) prettier` releases * rm `prettier` pre-commit hook commit 8b8183b Author: Ryan Williams <ryan.williams@tiledb.com> Date: Wed Apr 9 15:27:24 2025 -0400 python-uv.dockerfile (#3948) commit d96e81c Author: Ryan Williams <ryan.williams@tiledb.com> Date: Wed Apr 9 15:22:14 2025 -0400 rm numcodecs<0.16 pin (#3946) `zarr==2.18.7` removes the need for it, cf.: - zarr-developers/zarr-python#2963 - #3917 commit f0f1cc5 Author: John Kerl <kerl.john.r@gmail.com> Date: Wed Apr 9 13:04:32 2025 -0400 [python] Extend previous-version tests for experiment query (#3945) commit e0b594c Author: John Kerl <kerl.john.r@gmail.com> Date: Wed Apr 9 10:56:13 2025 -0400 [python] Previous-versions testing for enum getter/extender (#3940) * [python] Extend previous-versions testing * extender commit 847d85f Author: John Kerl <kerl.john.r@gmail.com> Date: Wed Apr 9 10:29:42 2025 -0400 [python] Improve UX for `extend_enumeration_values` with non-Arrow types (#3939) commit fc5d2c1 Author: John Kerl <kerl.john.r@gmail.com> Date: Wed Apr 9 08:46:12 2025 -0400 [c++] Follow-up to #3873 and #3928 (#3937) commit d65892d Author: Bruce Martin <bruce.martin@tiledb.com> Date: Tue Apr 8 15:13:55 2025 -0700 fix valgrind errors in unit tests (#3935) commit 4746ae3 Author: XanthosXanthopoulos <38084549+XanthosXanthopoulos@users.noreply.github.com> Date: Tue Apr 8 13:10:08 2025 -0700 Replace std::unique_ptr with raw pointer for attribute arrow schema (#3928) commit 10b2549 Author: Bruce Martin <bruce.martin@tiledb.com> Date: Tue Apr 8 12:23:53 2025 -0700 [python] Ingest performance improvements (#3865) * first cut at ingest tuning * mp optional param * fix tests, add load from S3 * reshape fix * lint * add multiprocessing config warning * performance tuning * pass context * lint * add test * interim checkpoint on enum cleanup * add test for subset * add new extend enum code (commented out) * migrate to new enum API * bump typeguard version to work around typeguard#504 * improve doctrings * more cleanup * loosen existance checks when from_anndata/from_h5ad are run in schema_only mode * cleanup and additional error checking * more docstring cleanup * more docstring cleanup * more docstrings * pr fb * pr fb * more pr fb * increase test code coverage * additional code coverage * lint * PR fb * PR fb * more PR fb * lint * remove extraneous line break in docstring commit 92fc2b8 Author: nguyenv <vivian@tiledb.com> Date: Tue Apr 8 13:21:43 2025 -0500 [python] Return `datetime` `domain`/`maxdomain` as `pa.TimestampScalar` (#3877) - The domain/maxdomain was return as `np.datetime64`. Give that Arrow is used as the default type for all return values, this is now corrected to `pa.TimestampScalar` - Allow the passed-in type of the domain to be `int` and `pa.TimestampScalar` in addition to `np.datetime64` commit 5d7e70b Author: Paul Hoffman <mojaveazure@users.noreply.github.com> Date: Tue Apr 8 12:51:01 2025 -0400 [r] Add support for reading v5 assays from an axis query (#3008) * [r] Add support for reading v5 assays from an axis query Seurat v5 adds support for ragged arrays, where not every `X` layer has exactly the same cells and features. To handle this, ragged `X` layers need to be re-indexed and re-shaped on ingestion to resize down to only the data present Modified SOMA methods: - `SOMAExperimentAxisQuery$to_seurat()` and `SOMAExperimentAxisQuery$to_seurat_assay()`: now read in as v5 assays New SOMA methods: - `SOMAExperimentAxisQuery$private$.to_seurat_assay_v5()`: helper method to read in ragged and non-ragged arrays into a v5 assay; note this method only handles expression layers, all other assay-level information is handled by parent `$to_seurat_assay()` to share code with v3 assay outgestion Requires #2523 and #3007 [SC-52261](https://app.shortcut.com/tiledb-inc/story/52261/) * Use hint helpers * Fix shape, regex uses * Use hint helper functions * Use updated column names * Apply suggestions from code review Co-authored-by: John Kerl <kerl.john.r@gmail.com> * Finish incomplete suggestions * more codereview feedback --------- Co-authored-by: John Kerl <kerl.john.r@gmail.com> commit 0b46462 Author: Bruce Martin <bruce.martin@tiledb.com> Date: Tue Apr 8 07:10:32 2025 -0700 [python] Fix partition readers' naïve assumption that `nnz` is fast (#3915) * add work-around for single-fragment nnz * add raise_if_slow to nnz, and use in partitioning readers * add raise_if_slow to nnz, and use in partitioning readers * revert likely buggy nnz optimization * improve fallback partitioning for ExperimentAxisQuery * pin numcodecs * pr fb * pr fb * PR fb commit 9d2ab41 Author: John Kerl <kerl.john.r@gmail.com> Date: Tue Apr 8 09:35:16 2025 -0400 Include 1.16.1 datasets in soma-experiment-versions tests (#3912) commit 5d386d4 Author: Ryan Williams <ryan.williams@tiledb.com> Date: Mon Apr 7 19:00:44 2025 -0400 work around #3917 in RTD build (#3922) commit 2f4241b Author: John Kerl <kerl.john.r@gmail.com> Date: Mon Apr 7 17:51:06 2025 -0400 [python] Protect against an AnnData/SciPy incompatibility (#3919) commit 8644785 Author: John Kerl <kerl.john.r@gmail.com> Date: Mon Apr 7 17:17:20 2025 -0400 [python] Use `typeguard==4.4.2` (#3921) commit ad5fae4 Author: Ryan Williams <ryan.williams@tiledb.com> Date: Mon Apr 7 13:47:01 2025 -0400 `setup.py`: remove deprecated `License ::` from `classifiers` (#3916) * `setup.py`: remove deprecated `License ::` from `classifiers` [sc-65131] * pin `numcodecs<0.16` cf. zarr-developers/zarr-python#2963 commit cdc0d67 Author: Julia Dark <24235303+jp-dark@users.noreply.github.com> Date: Mon Apr 7 11:57:44 2025 -0400 [python] Directly read Visium table instead of filtering through AnnData (#3835) * Directly read data from 10x HDF5 instead of opening it with AnnData. * (bug fix) Add `platform_config` to spatial array creation. commit 37480ff Author: XanthosXanthopoulos <38084549+XanthosXanthopoulos@users.noreply.github.com> Date: Fri Apr 4 17:01:36 2025 -0700 Replace deprecated `dump` function (#3913) commit f9450ba Author: Ryan Williams <ryan.williams@tiledb.com> Date: Fri Apr 4 12:03:40 2025 -0400 Publish ARM Linux wheels (#3909) * `python-packaging`: rm `continue-on-error: true` from Test PyPI publish That was masking failures to upload to test.pypi.org (due to 10GB space quota being full), e.g. https://github.com/single-cell-data/TileDB-SOMA/actions/runs/13913129586/job/38931629454 * Publish manylinux aarch64 wheels * Audit/Factor `cibuildwheel` env vars * rm unused `cc`/`cxx` smoke-test matrix vars * add 3.9 smoke-test jobs * rm unused `CIBW_BEFORE_BUILD` hook commit 1c6a057 Author: XanthosXanthopoulos <38084549+XanthosXanthopoulos@users.noreply.github.com> Date: Fri Apr 4 06:10:52 2025 -0700 [c++] Fix Arrow memory leaks in unit tests (#3873) * Accept const reference of unique pointers * Release allocated arrow objects * Fix deallocation of arrow buffers, return raw pointer for all non parent arrow objects * Release allocated memrory in unit tests * Pass raw pointers to `set_array_data` commit d77edea Author: John Kerl <kerl.john.r@gmail.com> Date: Thu Apr 3 13:28:54 2025 -0400 [c++] Show errant value in not-array-not-group error messages (#3897) * [c++] Show errant value in not-array-not-group error messages * fix dangling WIP bits * code-review feedback commit fb12025 Author: John Kerl <kerl.john.r@gmail.com> Date: Thu Apr 3 09:41:51 2025 -0400 [ci] Remove ticket-cutter for no-longer-nightly job (#3900) commit 485d4d8 Author: Ryan Williams <ryan.williams@tiledb.com> Date: Wed Apr 2 23:19:46 2025 -0400 `python-ci-packaging.yml`: only run on `push` events to `main` (#3895) Otherwise it runs on any branch push that touches relevant `paths`, which is not necessary (given the `paths` are already tested for `pull_request` events) commit 6c08b42 Author: John Kerl <kerl.john.r@gmail.com> Date: Wed Apr 2 19:55:07 2025 -0400 [c++/python] Expose enumeration-extender (#3815) * wip * Constraints testing against old pandas/numpy versions * Improve structuring * Merge in #3845 * fix for older numpy * resolve #3815 (comment) [skip ci] * ordered flag * #3815 (comment) * resolve #3815 (comment) * more on #3815 (comment) * temp/wip [skip ci] * RAII-related code-review feedback * resolve #3815 (comment) * null-checking * iterating on code-review feedback * code-review feedback * add a helpful comment * try to be even more clear * check nan bits * Advice from #3888 thanks to Xanthos * Apply suggestions from code review Co-authored-by: nguyenv <vivian@tiledb.com> * neatens for code review * Apply suggestions from code review Co-authored-by: Julia Dark <24235303+jp-dark@users.noreply.github.com> * fix lint induced by code-review suggestions * const-refs for column name -- code-review feedback [skip ci] * some static_cast callsites [skip ci] * docblocks per code-review feedback [skip ci] * docblocks per code-review feedback * try-catch solution; ScopedExecutor coming next * ScopedExecutor --------- Co-authored-by: nguyenv <vivian@tiledb.com> Co-authored-by: Julia Dark <24235303+jp-dark@users.noreply.github.com> commit d95ba39 Author: Ryan Williams <ryan.williams@tiledb.com> Date: Wed Apr 2 11:28:05 2025 -0400 [ci/python] `python-dockers.yml`: improve issue template (#3894) * Improve `python-dockers` issue template * gha wiki link update [skip ci] commit 4a3cfc6 Author: Ryan Williams <ryan.williams@tiledb.com> Date: Wed Apr 2 11:21:57 2025 -0400 [python] Pin `build-system.requires`, ARM Linux Docker-builds to `cmake<4` (#3891) * pin cmake<4 in setuptools build * pip constraint `cmake<4` in ARM Linux docker builds * `python-dockers.yml`: `fail-fast: false` * `python-dockers.yml`: improve job names commit c89fed9 Author: John Kerl <kerl.john.r@gmail.com> Date: Tue Apr 1 15:57:59 2025 -0400 [ci] Run remote-storage workflow on all commits (#3885) commit 5e52122 Author: John Kerl <kerl.john.r@gmail.com> Date: Tue Apr 1 15:12:10 2025 -0400 [c++/python] Revert 3764 for global-order writes (#3886) commit ef96798 Author: XanthosXanthopoulos <38084549+XanthosXanthopoulos@users.noreply.github.com> Date: Tue Apr 1 09:16:23 2025 -0700 [r] Use exposed logger in C++ bindings (#3868) * Roll back to used `spdl::` [skip ci] * Replace spdl::* with tdbs::* commit 945d195 Author: Ryan Williams <ryan.williams@tiledb.com> Date: Tue Apr 1 10:23:01 2025 -0400 `python-ci-full.yml`: fix `include` rules (#3881) @bkmartinjr noticed our ubuntu GHAs had `cc`/`cxx` set to `clang`: sc-65188 commit 0aae54d Author: Bruce Martin <bruce.martin@tiledb.com> Date: Mon Mar 31 18:01:48 2025 -0700 [python] Ingest from S3 performance (#3872) * h5ad read from s3 performance * add simplicity * increase cache size * more macro data refinement * cleanup * darn clang-format * remove dead code * py39 lint * fix pre py 3.12 typeguard error * log python packages installed * pr fb * PR fb * refactor for clarity commit ee8aab4 Author: nguyenv <vivian@tiledb.com> Date: Mon Mar 31 13:33:16 2025 -0500 [python/c++] Allow GOW multi-submit for single commit (#3764) Co-authored-by: John Kerl <kerl.john.r@gmail.com> commit 88317fb Author: Julia Dark <24235303+jp-dark@users.noreply.github.com> Date: Thu Mar 27 14:40:32 2025 -0400 [python] Prepare `io.shaping` for spatial additions (#3856) * Update docstrings to reflect which elements inside the experiment are being accessed. * Refactor `_treewalk` to make adding more elements easier. --------- Co-authored-by: John Kerl <kerl.john.r@gmail.com> commit 5331558 Author: nguyenv <vivian@tiledb.com> Date: Thu Mar 27 13:06:52 2025 -0500 [c++] Correctly handle NaN category values (#3845) Add ManagedQuery::_find_enum_match which uses bitwise comparison. Like std::find, the method returns an iterator that points to the match value in the container; if not found, returns an iterator to the end of the container. commit 88c1664 Author: John Kerl <kerl.john.r@gmail.com> Date: Thu Mar 27 10:54:42 2025 -0400 [python] Fix a venial typo in a unit-test case (#3862)

ryan-williams added the bug Potential issues with the zarr-python library label Apr 7, 2025

ryan-williams added a commit to single-cell-data/TileDB-SOMA that referenced this issue Apr 7, 2025

pin numcodecs<0.16

bdb59d5

cf. zarr-developers/zarr-python#2963

ryan-williams mentioned this issue Apr 7, 2025

[Bug] spatialdata==0.3.0 pins zarr<3 which is incompatible with latest numcodecs (0.16.0) single-cell-data/TileDB-SOMA#3917

Closed

ryan-williams added a commit to single-cell-data/TileDB-SOMA that referenced this issue Apr 7, 2025

pin numcodecs<0.16

f077879

cf. zarr-developers/zarr-python#2963

leehart mentioned this issue Apr 7, 2025

Broken import of numcodecs malariagen/malariagen-data-python#753

Closed

dstansby mentioned this issue Apr 7, 2025

Pin numcodecs to < 0.16 to fix v2 #2965

Merged

jakirkham mentioned this issue Apr 7, 2025

cannot import name 'cbuffer_sizes' from 'numcodecs.blosc' zarr-developers/numcodecs#721

Open

lorenzocerrone mentioned this issue Apr 8, 2025

Addition of Yaml in input lorenzocerrone/fractal-hcs-drogon-converters#1

Merged

mdbenito mentioned this issue Apr 8, 2025

zarr < 3 and numcodecs aai-institute/pyDVL#667

Closed

This was referenced Apr 8, 2025

Drop support for python 3.10 fractal-analytics-platform/fractal-tasks-core#932

Merged

Remove numcodecs pin from dev-1.5 branch fractal-analytics-platform/fractal-tasks-core#935

Open

Jan-Willem added a commit to casangi/xradio that referenced this issue Apr 8, 2025

Add numcodecs<0.16 to fix zarr-developers/zarr-python#2963

779b80b

dstansby closed this as completed Apr 9, 2025

ryan-williams added a commit to single-cell-data/TileDB-SOMA that referenced this issue Apr 9, 2025

rm numcodecs<0.16 pin

831d0b1

`zarr==2.18.7` removes the need for it, cf.: - zarr-developers/zarr-python#2963 - #3917

ryan-williams added a commit to single-cell-data/TileDB-SOMA that referenced this issue Apr 9, 2025

rm numcodecs<0.16 pin

0423451

`zarr==2.18.7` removes the need for it, cf.: - zarr-developers/zarr-python#2963 - #3917

ryan-williams mentioned this issue Apr 9, 2025

[python] Remove numcodecs<0.16 pin single-cell-data/TileDB-SOMA#3946

Merged

ryan-williams added a commit to single-cell-data/TileDB-SOMA that referenced this issue Apr 9, 2025

rm numcodecs<0.16 pin (#3946)

d96e81c

`zarr==2.18.7` removes the need for it, cf.: - zarr-developers/zarr-python#2963 - #3917

github-actions bot pushed a commit to single-cell-data/TileDB-SOMA that referenced this issue Apr 9, 2025

rm numcodecs<0.16 pin (#3946)

8f1cd7b

`zarr==2.18.7` removes the need for it, cf.: - zarr-developers/zarr-python#2963 - #3917

johnkerl pushed a commit to single-cell-data/TileDB-SOMA that referenced this issue Apr 9, 2025

rm numcodecs<0.16 pin (#3946) (#3950)

d0eb581

`zarr==2.18.7` removes the need for it, cf.: - zarr-developers/zarr-python#2963 - #3917 Co-authored-by: Ryan Williams <ryan.williams@tiledb.com>

This was referenced Apr 10, 2025

Use Numcodecs pre-0.16.0 with Zarr 2 conda-forge/conda-forge-repodata-patches-feedstock#1009

Merged

Update to 2.18.5 conda-forge/zarr-feedstock#127

Merged

jdeschamps mentioned this issue Apr 14, 2025

Added a Prediction Widget CAREamics/careamics-napari#22

Merged

This was referenced Apr 23, 2025

Pin numcodecs to <0.16 for unit tests google-research/weatherbenchX#56

Closed

Pin numcodecs to <0.16 for unit tests google-research/weatherbench2#258

Merged

ways mentioned this issue Apr 28, 2025

Add test for MultiEncDecPredictor metno/bris-inference#125

Closed

This was referenced May 1, 2025

Monthly issue metrics report #3030

Open

Monthly issue metrics report sanketverma1704/zarr-python#10

Open

jonasutz mentioned this issue May 6, 2025

Bump zarr to ver 2.18.6 to fix issue with numcodecs 0.16.0 DeepMicroscopy/Exact#134

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Can't import `zarr<3` with latest `numcodecs==0.16.0` #2963

Can't import `zarr<3` with latest `numcodecs==0.16.0` #2963

ryan-williams commented Apr 7, 2025 •

edited

Loading

dstansby commented Apr 7, 2025

cspades commented Apr 7, 2025

dstansby commented Apr 8, 2025

tcompa commented Apr 8, 2025

dstansby commented Apr 8, 2025

gastoneb commented Apr 8, 2025

dstansby commented Apr 9, 2025

schenker commented Apr 9, 2025

Can't import zarr<3 with latest numcodecs==0.16.0 #2963

Can't import zarr<3 with latest numcodecs==0.16.0 #2963

Comments

ryan-williams commented Apr 7, 2025 • edited Loading

Zarr version

Numcodecs version

Python Version

Operating System

Installation

Description

Steps to reproduce

Additional output

dstansby commented Apr 7, 2025

cspades commented Apr 7, 2025

dstansby commented Apr 8, 2025

tcompa commented Apr 8, 2025

dstansby commented Apr 8, 2025

gastoneb commented Apr 8, 2025

dstansby commented Apr 9, 2025

schenker commented Apr 9, 2025

Can't import `zarr<3` with latest `numcodecs==0.16.0` #2963

Can't import `zarr<3` with latest `numcodecs==0.16.0` #2963

ryan-williams commented Apr 7, 2025 •

edited

Loading