ome zarr chunks by ArneDefauw · Pull Request #1092 · scverse/spatialdata

ArneDefauw · 2026-03-12T08:18:57Z

Fix for #1090.

Note that unit tests still fail for ome-zarr ==0.14.0, due to #1091

codecov · 2026-03-12T08:55:13Z

Codecov Report

❌ Patch coverage is 85.45455% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.93%. Comparing base (6a3eef7) to head (385dd2e).
⚠️ Report is 26 commits behind head on main.

Files with missing lines	Patch %	Lines
src/spatialdata/_io/io_raster.py	85.18%	8 Missing ⚠️

Additional details and impacted files

@@ Coverage Diff @@## main #1092 +/- ##
==========================================
- Coverage 91.96% 91.93% -0.04% 
==========================================
Files 51 51 Lines 7729 7772 +43 ==========================================
+ Hits 7108 7145 +37 - Misses 621 627 +6

Files with missing lines	Coverage Δ
src/spatialdata/__init__.py	`95.65% <100.00%> (-0.51%)`	⬇️
src/spatialdata/_io/io_raster.py	`92.09% <85.18%> (-1.81%)`	⬇️

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

ArneDefauw · 2026-03-12T09:28:25Z

also added fix for #1091.

On linux, unit tests are failing, I will try to reproduce it

ArneDefauw · 2026-03-12T10:18:58Z

also added fix for #1091.
On linux, unit tests are failing, I will try to reproduce it

Used S0 instead of s0, for pyramid scale 0, therefore unit tests were failing on linux

LucaMarconato

Thanks for the PR! I would make some changes as described.

LucaMarconato · 2026-03-19T21:40:28Z

+def _prepare_single_scale_storage_options(
+ storage_options: JSONDict | list[JSONDict] | None,
+) -> JSONDict | list[JSONDict] | None:
+ if storage_options is None:
+ return None
+ if isinstance(storage_options, dict):
+ prepared = dict(storage_options)
+ if "chunks" in prepared:
+ prepared["chunks"] = _normalize_explicit_chunks(prepared["chunks"])
+ return prepared
+ return [dict(options) for options in storage_options]


This function behaves like _prepare_multiscale_storage_options(), without normalizing the list of storage options case. Can we remove it and just use _prepare_multiscale_storage_options() (after renaming it to _prepare_storage_options()?

I unified the two functions in 20696b5. Happy to hear what you think (in case we need two, we can revert, but I think we can proceed with one function).

LucaMarconato · 2026-03-19T21:41:11Z

+ if isinstance(value, str | bytes):
+ return False


Why do we need this? Please either remove or document.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:41:18Z

+ if isinstance(value, str | bytes):
+ return False


Same for this check.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:50:34Z

+def _is_regular_dask_chunk_grid(chunk_grid: Sequence[Sequence[int]]) -> bool:
+ # Match Dask's private _check_regular_chunks() logic without depending on its internal API.
+ for axis_chunks in chunk_grid:
+ if len(axis_chunks) <= 1:
+ continue
+ if len(set(axis_chunks[:-1])) > 1:
+ return False
+ if axis_chunks[-1] > axis_chunks[0]:
+ return False
+ return True


I'd add a docstring with examples (or examples in-line with the code) to show what fails and what not.
I would add the following:
triggers the continue in the first if:
[(4,)]
[()]
triggers the first return False
[(4, 4, 3, 4)]
triggers the second return False
[(4, 4, 4, 5)]
exits with the last return True
[(4, 4, 4, 4)], succeeds, all chunks equal
[(4, 4, 4, 1)], succeeds, final chunk is < of the initial one

Also: I would add all the examples above in a test, for the function _is_regular_dask_chunk_grid().

Added docstring and tests here: 2450bd4

LucaMarconato · 2026-03-19T22:04:20Z

+def test_write_irregular_dask_chunks_without_explicit_storage_options(tmp_path: Path) -> None:
+ data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
+ image = Image2DModel.parse(data, dims=("c", "y", "x"))
+ sdata = SpatialData(images={"image": image})
+
+ sdata.write(tmp_path / "data.zarr")


I would find it more natural if this test was failing. Now that chunks = raster_data.chunks has been removed it, writing doesn't fail, but it ignores the chunks in the data. I think a natural behavior is that if storage option specifies chunks, these are used, otherwise the ones from the data (and if the `chunks from the data are irregular and no storage options are specified, an error would be raised).

I now implemented this in 20696b5 changing the test so that it expects to fail.

LucaMarconato · 2026-03-19T22:16:19Z

I'll go ahead and implement a fix for the code review points.

LucaMarconato · 2026-03-19T22:18:17Z

the commit 930922c closes PointsModel.parse() fails with AttributeError: 'DataFrame' object has no attribute 'attrs' #1093, I included it here since I was encountering that when running some debug scripts.

…ied in storage options

LucaMarconato · 2026-03-19T22:29:10Z

I implemented the changes mentioned in the code review. Please let me know if you agree with the changes. If yes I'll merge and work on a release.

ArneDefauw · 2026-03-20T09:24:46Z

Hi @LucaMarconato , thanks for having a look!

My idea was to stop auto-filling storage_options["chunks"] with raster_data.data, i.e. here https://github.com/ArneDefauw/spatialdata/blob/2450bd437914004cc940279482f2dc60293e0575/src/spatialdata/_io/io_raster.py#L375

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

I still gaurded for irregular chunks, because spatialdata._io.io_raster import write_image exposes storage_options as a parameter.

I think there are valid arguments for both implementations.

If we go for auto-filling storage_options (your changes), then I would change the ValueError raised when doing e.g.:


import tempfile
from pathlib import Path
import dask.array as da
from numpy.random import default_rng
from spatialdata import SpatialData
from spatialdata.models import Image2DModel
tmpdir = Path(tempfile.mkdtemp())
RNG = default_rng(0)
data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
image = Image2DModel.parse(data, dims=("c", "y", "x"))
sdata = SpatialData(images={"image": image})
sdata.write(tmpdir / "data.zarr", overwrite=True)

from

ValueError: storage_options['chunks'] must be a Zarr chunk shape or a regular Dask chunk grid. Irregular Dask chunk grids must be rechunked before writing or omitted.

To specify how user can avoid it, e.g. change it to:

ValueError: storage_options["chunks"] must resolve to a Zarr chunk shape or a regular Dask chunk grid. The current raster has irregular Dask chunks, which cannot be written to Zarr. To fix this, rechunk before writing, for example by passing regular chunks=... to Image2DModel.parse(...) / Labels2DModel.parse(...).

ArneDefauw · 2026-03-20T10:22:52Z

Added small update to guard against storage_options["chunks"]="" and storage_options["chunks"]=b"auto" . Bit of an edge case, and the reason why this if isinstance(value, str | bytes) got introduced in

def _is_flat_int_sequence(value: object) -> TypeGuard[Sequence[int]]:
if isinstance(value, str | bytes):
return False
if not isinstance(value, Sequence):
return False
return all(isinstance(v, int) for v in value)

Also updated the ValueErorr to be more user friendly, see previous comment

LucaMarconato · 2026-03-20T13:49:28Z

Thanks for the comments and for the changes.

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

That's a good point! I restored the old implementation of _prepare_storage_options() that didn't depend on the data (while keeping the unified function instead of one for single-scale and one for multi-scale).

Ready to merge!

ArneDefauw · 2026-03-12T08:18:57Z

Fix for #1090.

Note that unit tests still fail for ome-zarr ==0.14.0, due to #1091

codecov · 2026-03-12T08:55:13Z

Codecov Report

❌ Patch coverage is 85.45455% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.93%. Comparing base (6a3eef7) to head (385dd2e).
⚠️ Report is 26 commits behind head on main.

Files with missing lines	Patch %	Lines
src/spatialdata/_io/io_raster.py	85.18%	8 Missing ⚠️

Additional details and impacted files

@@ Coverage Diff @@## main #1092 +/- ##
==========================================
- Coverage 91.96% 91.93% -0.04% 
==========================================
Files 51 51 Lines 7729 7772 +43 ==========================================
+ Hits 7108 7145 +37 - Misses 621 627 +6

Files with missing lines	Coverage Δ
src/spatialdata/__init__.py	`95.65% <100.00%> (-0.51%)`	⬇️
src/spatialdata/_io/io_raster.py	`92.09% <85.18%> (-1.81%)`	⬇️

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

ArneDefauw · 2026-03-12T09:28:25Z

also added fix for #1091.

On linux, unit tests are failing, I will try to reproduce it

ArneDefauw · 2026-03-12T10:18:58Z

also added fix for #1091.
On linux, unit tests are failing, I will try to reproduce it

Used S0 instead of s0, for pyramid scale 0, therefore unit tests were failing on linux

LucaMarconato

Thanks for the PR! I would make some changes as described.

LucaMarconato · 2026-03-19T21:40:28Z

+def _prepare_single_scale_storage_options(
+ storage_options: JSONDict | list[JSONDict] | None,
+) -> JSONDict | list[JSONDict] | None:
+ if storage_options is None:
+ return None
+ if isinstance(storage_options, dict):
+ prepared = dict(storage_options)
+ if "chunks" in prepared:
+ prepared["chunks"] = _normalize_explicit_chunks(prepared["chunks"])
+ return prepared
+ return [dict(options) for options in storage_options]


This function behaves like _prepare_multiscale_storage_options(), without normalizing the list of storage options case. Can we remove it and just use _prepare_multiscale_storage_options() (after renaming it to _prepare_storage_options()?

I unified the two functions in 20696b5. Happy to hear what you think (in case we need two, we can revert, but I think we can proceed with one function).

LucaMarconato · 2026-03-19T21:41:11Z

+ if isinstance(value, str | bytes):
+ return False


Why do we need this? Please either remove or document.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:41:18Z

+ if isinstance(value, str | bytes):
+ return False


Same for this check.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:50:34Z

+def _is_regular_dask_chunk_grid(chunk_grid: Sequence[Sequence[int]]) -> bool:
+ # Match Dask's private _check_regular_chunks() logic without depending on its internal API.
+ for axis_chunks in chunk_grid:
+ if len(axis_chunks) <= 1:
+ continue
+ if len(set(axis_chunks[:-1])) > 1:
+ return False
+ if axis_chunks[-1] > axis_chunks[0]:
+ return False
+ return True


I'd add a docstring with examples (or examples in-line with the code) to show what fails and what not.
I would add the following:
triggers the continue in the first if:
[(4,)]
[()]
triggers the first return False
[(4, 4, 3, 4)]
triggers the second return False
[(4, 4, 4, 5)]
exits with the last return True
[(4, 4, 4, 4)], succeeds, all chunks equal
[(4, 4, 4, 1)], succeeds, final chunk is < of the initial one

Also: I would add all the examples above in a test, for the function _is_regular_dask_chunk_grid().

Added docstring and tests here: 2450bd4

LucaMarconato · 2026-03-19T22:04:20Z

+def test_write_irregular_dask_chunks_without_explicit_storage_options(tmp_path: Path) -> None:
+ data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
+ image = Image2DModel.parse(data, dims=("c", "y", "x"))
+ sdata = SpatialData(images={"image": image})
+
+ sdata.write(tmp_path / "data.zarr")


I would find it more natural if this test was failing. Now that chunks = raster_data.chunks has been removed it, writing doesn't fail, but it ignores the chunks in the data. I think a natural behavior is that if storage option specifies chunks, these are used, otherwise the ones from the data (and if the `chunks from the data are irregular and no storage options are specified, an error would be raised).

I now implemented this in 20696b5 changing the test so that it expects to fail.

LucaMarconato · 2026-03-19T22:16:19Z

I'll go ahead and implement a fix for the code review points.

LucaMarconato · 2026-03-19T22:18:17Z

the commit 930922c closes PointsModel.parse() fails with AttributeError: 'DataFrame' object has no attribute 'attrs' #1093, I included it here since I was encountering that when running some debug scripts.

…ied in storage options

LucaMarconato · 2026-03-19T22:29:10Z

I implemented the changes mentioned in the code review. Please let me know if you agree with the changes. If yes I'll merge and work on a release.

ArneDefauw · 2026-03-20T09:24:46Z

Hi @LucaMarconato , thanks for having a look!

My idea was to stop auto-filling storage_options["chunks"] with raster_data.data, i.e. here https://github.com/ArneDefauw/spatialdata/blob/2450bd437914004cc940279482f2dc60293e0575/src/spatialdata/_io/io_raster.py#L375

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

I still gaurded for irregular chunks, because spatialdata._io.io_raster import write_image exposes storage_options as a parameter.

I think there are valid arguments for both implementations.

If we go for auto-filling storage_options (your changes), then I would change the ValueError raised when doing e.g.:


import tempfile
from pathlib import Path
import dask.array as da
from numpy.random import default_rng
from spatialdata import SpatialData
from spatialdata.models import Image2DModel
tmpdir = Path(tempfile.mkdtemp())
RNG = default_rng(0)
data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
image = Image2DModel.parse(data, dims=("c", "y", "x"))
sdata = SpatialData(images={"image": image})
sdata.write(tmpdir / "data.zarr", overwrite=True)

from

ValueError: storage_options['chunks'] must be a Zarr chunk shape or a regular Dask chunk grid. Irregular Dask chunk grids must be rechunked before writing or omitted.

To specify how user can avoid it, e.g. change it to:

ValueError: storage_options["chunks"] must resolve to a Zarr chunk shape or a regular Dask chunk grid. The current raster has irregular Dask chunks, which cannot be written to Zarr. To fix this, rechunk before writing, for example by passing regular chunks=... to Image2DModel.parse(...) / Labels2DModel.parse(...).

ArneDefauw · 2026-03-20T10:22:52Z

Added small update to guard against storage_options["chunks"]="" and storage_options["chunks"]=b"auto" . Bit of an edge case, and the reason why this if isinstance(value, str | bytes) got introduced in

def _is_flat_int_sequence(value: object) -> TypeGuard[Sequence[int]]:
if isinstance(value, str | bytes):
return False
if not isinstance(value, Sequence):
return False
return all(isinstance(v, int) for v in value)

Also updated the ValueErorr to be more user friendly, see previous comment

LucaMarconato · 2026-03-20T13:49:28Z

Thanks for the comments and for the changes.

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

That's a good point! I restored the old implementation of _prepare_storage_options() that didn't depend on the data (while keeping the unified function instead of one for single-scale and one for multi-scale).

Ready to merge!

ArneDefauw · 2026-03-12T08:18:57Z

Fix for #1090.

Note that unit tests still fail for ome-zarr ==0.14.0, due to #1091

codecov · 2026-03-12T08:55:13Z

Codecov Report

❌ Patch coverage is 85.45455% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.93%. Comparing base (6a3eef7) to head (385dd2e).
⚠️ Report is 26 commits behind head on main.

Files with missing lines	Patch %	Lines
src/spatialdata/_io/io_raster.py	85.18%	8 Missing ⚠️

Additional details and impacted files

@@ Coverage Diff @@## main #1092 +/- ##
==========================================
- Coverage 91.96% 91.93% -0.04% 
==========================================
Files 51 51 Lines 7729 7772 +43 ==========================================
+ Hits 7108 7145 +37 - Misses 621 627 +6

Files with missing lines	Coverage Δ
src/spatialdata/__init__.py	`95.65% <100.00%> (-0.51%)`	⬇️
src/spatialdata/_io/io_raster.py	`92.09% <85.18%> (-1.81%)`	⬇️

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

ArneDefauw · 2026-03-12T09:28:25Z

also added fix for #1091.

On linux, unit tests are failing, I will try to reproduce it

ArneDefauw · 2026-03-12T10:18:58Z

also added fix for #1091.
On linux, unit tests are failing, I will try to reproduce it

Used S0 instead of s0, for pyramid scale 0, therefore unit tests were failing on linux

LucaMarconato

Thanks for the PR! I would make some changes as described.

LucaMarconato · 2026-03-19T21:40:28Z

+def _prepare_single_scale_storage_options(
+ storage_options: JSONDict | list[JSONDict] | None,
+) -> JSONDict | list[JSONDict] | None:
+ if storage_options is None:
+ return None
+ if isinstance(storage_options, dict):
+ prepared = dict(storage_options)
+ if "chunks" in prepared:
+ prepared["chunks"] = _normalize_explicit_chunks(prepared["chunks"])
+ return prepared
+ return [dict(options) for options in storage_options]


This function behaves like _prepare_multiscale_storage_options(), without normalizing the list of storage options case. Can we remove it and just use _prepare_multiscale_storage_options() (after renaming it to _prepare_storage_options()?

I unified the two functions in 20696b5. Happy to hear what you think (in case we need two, we can revert, but I think we can proceed with one function).

LucaMarconato · 2026-03-19T21:41:11Z

+ if isinstance(value, str | bytes):
+ return False


Why do we need this? Please either remove or document.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:41:18Z

+ if isinstance(value, str | bytes):
+ return False


Same for this check.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:50:34Z

+def _is_regular_dask_chunk_grid(chunk_grid: Sequence[Sequence[int]]) -> bool:
+ # Match Dask's private _check_regular_chunks() logic without depending on its internal API.
+ for axis_chunks in chunk_grid:
+ if len(axis_chunks) <= 1:
+ continue
+ if len(set(axis_chunks[:-1])) > 1:
+ return False
+ if axis_chunks[-1] > axis_chunks[0]:
+ return False
+ return True


I'd add a docstring with examples (or examples in-line with the code) to show what fails and what not.
I would add the following:
triggers the continue in the first if:
[(4,)]
[()]
triggers the first return False
[(4, 4, 3, 4)]
triggers the second return False
[(4, 4, 4, 5)]
exits with the last return True
[(4, 4, 4, 4)], succeeds, all chunks equal
[(4, 4, 4, 1)], succeeds, final chunk is < of the initial one

Also: I would add all the examples above in a test, for the function _is_regular_dask_chunk_grid().

Added docstring and tests here: 2450bd4

LucaMarconato · 2026-03-19T22:04:20Z

+def test_write_irregular_dask_chunks_without_explicit_storage_options(tmp_path: Path) -> None:
+ data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
+ image = Image2DModel.parse(data, dims=("c", "y", "x"))
+ sdata = SpatialData(images={"image": image})
+
+ sdata.write(tmp_path / "data.zarr")


I would find it more natural if this test was failing. Now that chunks = raster_data.chunks has been removed it, writing doesn't fail, but it ignores the chunks in the data. I think a natural behavior is that if storage option specifies chunks, these are used, otherwise the ones from the data (and if the `chunks from the data are irregular and no storage options are specified, an error would be raised).

I now implemented this in 20696b5 changing the test so that it expects to fail.

LucaMarconato · 2026-03-19T22:16:19Z

I'll go ahead and implement a fix for the code review points.

LucaMarconato · 2026-03-19T22:18:17Z

the commit 930922c closes PointsModel.parse() fails with AttributeError: 'DataFrame' object has no attribute 'attrs' #1093, I included it here since I was encountering that when running some debug scripts.

…ied in storage options

LucaMarconato · 2026-03-19T22:29:10Z

I implemented the changes mentioned in the code review. Please let me know if you agree with the changes. If yes I'll merge and work on a release.

ArneDefauw · 2026-03-20T09:24:46Z

Hi @LucaMarconato , thanks for having a look!

My idea was to stop auto-filling storage_options["chunks"] with raster_data.data, i.e. here https://github.com/ArneDefauw/spatialdata/blob/2450bd437914004cc940279482f2dc60293e0575/src/spatialdata/_io/io_raster.py#L375

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

I still gaurded for irregular chunks, because spatialdata._io.io_raster import write_image exposes storage_options as a parameter.

I think there are valid arguments for both implementations.

If we go for auto-filling storage_options (your changes), then I would change the ValueError raised when doing e.g.:


import tempfile
from pathlib import Path
import dask.array as da
from numpy.random import default_rng
from spatialdata import SpatialData
from spatialdata.models import Image2DModel
tmpdir = Path(tempfile.mkdtemp())
RNG = default_rng(0)
data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
image = Image2DModel.parse(data, dims=("c", "y", "x"))
sdata = SpatialData(images={"image": image})
sdata.write(tmpdir / "data.zarr", overwrite=True)

from

ValueError: storage_options['chunks'] must be a Zarr chunk shape or a regular Dask chunk grid. Irregular Dask chunk grids must be rechunked before writing or omitted.

To specify how user can avoid it, e.g. change it to:

ValueError: storage_options["chunks"] must resolve to a Zarr chunk shape or a regular Dask chunk grid. The current raster has irregular Dask chunks, which cannot be written to Zarr. To fix this, rechunk before writing, for example by passing regular chunks=... to Image2DModel.parse(...) / Labels2DModel.parse(...).

ArneDefauw · 2026-03-20T10:22:52Z

Added small update to guard against storage_options["chunks"]="" and storage_options["chunks"]=b"auto" . Bit of an edge case, and the reason why this if isinstance(value, str | bytes) got introduced in

def _is_flat_int_sequence(value: object) -> TypeGuard[Sequence[int]]:
if isinstance(value, str | bytes):
return False
if not isinstance(value, Sequence):
return False
return all(isinstance(v, int) for v in value)

Also updated the ValueErorr to be more user friendly, see previous comment

LucaMarconato · 2026-03-20T13:49:28Z

Thanks for the comments and for the changes.

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

That's a good point! I restored the old implementation of _prepare_storage_options() that didn't depend on the data (while keeping the unified function instead of one for single-scale and one for multi-scale).

Ready to merge!

ArneDefauw · 2026-03-12T08:18:57Z

Fix for #1090.

Note that unit tests still fail for ome-zarr ==0.14.0, due to #1091

codecov · 2026-03-12T08:55:13Z

Codecov Report

❌ Patch coverage is 85.45455% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.93%. Comparing base (6a3eef7) to head (385dd2e).
⚠️ Report is 26 commits behind head on main.

Files with missing lines	Patch %	Lines
src/spatialdata/_io/io_raster.py	85.18%	8 Missing ⚠️

Additional details and impacted files

@@ Coverage Diff @@## main #1092 +/- ##
==========================================
- Coverage 91.96% 91.93% -0.04% 
==========================================
Files 51 51 Lines 7729 7772 +43 ==========================================
+ Hits 7108 7145 +37 - Misses 621 627 +6

Files with missing lines	Coverage Δ
src/spatialdata/__init__.py	`95.65% <100.00%> (-0.51%)`	⬇️
src/spatialdata/_io/io_raster.py	`92.09% <85.18%> (-1.81%)`	⬇️

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

ArneDefauw · 2026-03-12T09:28:25Z

also added fix for #1091.

On linux, unit tests are failing, I will try to reproduce it

ArneDefauw · 2026-03-12T10:18:58Z

also added fix for #1091.
On linux, unit tests are failing, I will try to reproduce it

Used S0 instead of s0, for pyramid scale 0, therefore unit tests were failing on linux

LucaMarconato

Thanks for the PR! I would make some changes as described.

LucaMarconato · 2026-03-19T21:40:28Z

+def _prepare_single_scale_storage_options(
+ storage_options: JSONDict | list[JSONDict] | None,
+) -> JSONDict | list[JSONDict] | None:
+ if storage_options is None:
+ return None
+ if isinstance(storage_options, dict):
+ prepared = dict(storage_options)
+ if "chunks" in prepared:
+ prepared["chunks"] = _normalize_explicit_chunks(prepared["chunks"])
+ return prepared
+ return [dict(options) for options in storage_options]


This function behaves like _prepare_multiscale_storage_options(), without normalizing the list of storage options case. Can we remove it and just use _prepare_multiscale_storage_options() (after renaming it to _prepare_storage_options()?

I unified the two functions in 20696b5. Happy to hear what you think (in case we need two, we can revert, but I think we can proceed with one function).

LucaMarconato · 2026-03-19T21:41:11Z

+ if isinstance(value, str | bytes):
+ return False


Why do we need this? Please either remove or document.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:41:18Z

+ if isinstance(value, str | bytes):
+ return False


Same for this check.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:50:34Z

+def _is_regular_dask_chunk_grid(chunk_grid: Sequence[Sequence[int]]) -> bool:
+ # Match Dask's private _check_regular_chunks() logic without depending on its internal API.
+ for axis_chunks in chunk_grid:
+ if len(axis_chunks) <= 1:
+ continue
+ if len(set(axis_chunks[:-1])) > 1:
+ return False
+ if axis_chunks[-1] > axis_chunks[0]:
+ return False
+ return True


I'd add a docstring with examples (or examples in-line with the code) to show what fails and what not.
I would add the following:
triggers the continue in the first if:
[(4,)]
[()]
triggers the first return False
[(4, 4, 3, 4)]
triggers the second return False
[(4, 4, 4, 5)]
exits with the last return True
[(4, 4, 4, 4)], succeeds, all chunks equal
[(4, 4, 4, 1)], succeeds, final chunk is < of the initial one

Also: I would add all the examples above in a test, for the function _is_regular_dask_chunk_grid().

Added docstring and tests here: 2450bd4

LucaMarconato · 2026-03-19T22:04:20Z

+def test_write_irregular_dask_chunks_without_explicit_storage_options(tmp_path: Path) -> None:
+ data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
+ image = Image2DModel.parse(data, dims=("c", "y", "x"))
+ sdata = SpatialData(images={"image": image})
+
+ sdata.write(tmp_path / "data.zarr")


I would find it more natural if this test was failing. Now that chunks = raster_data.chunks has been removed it, writing doesn't fail, but it ignores the chunks in the data. I think a natural behavior is that if storage option specifies chunks, these are used, otherwise the ones from the data (and if the `chunks from the data are irregular and no storage options are specified, an error would be raised).

I now implemented this in 20696b5 changing the test so that it expects to fail.

LucaMarconato · 2026-03-19T22:16:19Z

I'll go ahead and implement a fix for the code review points.

LucaMarconato · 2026-03-19T22:18:17Z

the commit 930922c closes PointsModel.parse() fails with AttributeError: 'DataFrame' object has no attribute 'attrs' #1093, I included it here since I was encountering that when running some debug scripts.

…ied in storage options

LucaMarconato · 2026-03-19T22:29:10Z

I implemented the changes mentioned in the code review. Please let me know if you agree with the changes. If yes I'll merge and work on a release.

ArneDefauw · 2026-03-20T09:24:46Z

Hi @LucaMarconato , thanks for having a look!

My idea was to stop auto-filling storage_options["chunks"] with raster_data.data, i.e. here https://github.com/ArneDefauw/spatialdata/blob/2450bd437914004cc940279482f2dc60293e0575/src/spatialdata/_io/io_raster.py#L375

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

I still gaurded for irregular chunks, because spatialdata._io.io_raster import write_image exposes storage_options as a parameter.

I think there are valid arguments for both implementations.

If we go for auto-filling storage_options (your changes), then I would change the ValueError raised when doing e.g.:


import tempfile
from pathlib import Path
import dask.array as da
from numpy.random import default_rng
from spatialdata import SpatialData
from spatialdata.models import Image2DModel
tmpdir = Path(tempfile.mkdtemp())
RNG = default_rng(0)
data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
image = Image2DModel.parse(data, dims=("c", "y", "x"))
sdata = SpatialData(images={"image": image})
sdata.write(tmpdir / "data.zarr", overwrite=True)

from

ValueError: storage_options['chunks'] must be a Zarr chunk shape or a regular Dask chunk grid. Irregular Dask chunk grids must be rechunked before writing or omitted.

To specify how user can avoid it, e.g. change it to:

ValueError: storage_options["chunks"] must resolve to a Zarr chunk shape or a regular Dask chunk grid. The current raster has irregular Dask chunks, which cannot be written to Zarr. To fix this, rechunk before writing, for example by passing regular chunks=... to Image2DModel.parse(...) / Labels2DModel.parse(...).

ArneDefauw · 2026-03-20T10:22:52Z

Added small update to guard against storage_options["chunks"]="" and storage_options["chunks"]=b"auto" . Bit of an edge case, and the reason why this if isinstance(value, str | bytes) got introduced in

def _is_flat_int_sequence(value: object) -> TypeGuard[Sequence[int]]:
if isinstance(value, str | bytes):
return False
if not isinstance(value, Sequence):
return False
return all(isinstance(v, int) for v in value)

Also updated the ValueErorr to be more user friendly, see previous comment

LucaMarconato · 2026-03-20T13:49:28Z

Thanks for the comments and for the changes.

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

That's a good point! I restored the old implementation of _prepare_storage_options() that didn't depend on the data (while keeping the unified function instead of one for single-scale and one for multi-scale).

Ready to merge!

ArneDefauw · 2026-03-12T08:18:57Z

Fix for #1090.

Note that unit tests still fail for ome-zarr ==0.14.0, due to #1091

codecov · 2026-03-12T08:55:13Z

Codecov Report

❌ Patch coverage is 85.45455% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.93%. Comparing base (6a3eef7) to head (385dd2e).
⚠️ Report is 26 commits behind head on main.

Files with missing lines	Patch %	Lines
src/spatialdata/_io/io_raster.py	85.18%	8 Missing ⚠️

Additional details and impacted files

@@ Coverage Diff @@## main #1092 +/- ##
==========================================
- Coverage 91.96% 91.93% -0.04% 
==========================================
Files 51 51 Lines 7729 7772 +43 ==========================================
+ Hits 7108 7145 +37 - Misses 621 627 +6

Files with missing lines	Coverage Δ
src/spatialdata/__init__.py	`95.65% <100.00%> (-0.51%)`	⬇️
src/spatialdata/_io/io_raster.py	`92.09% <85.18%> (-1.81%)`	⬇️

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

ArneDefauw · 2026-03-12T09:28:25Z

also added fix for #1091.

On linux, unit tests are failing, I will try to reproduce it

ArneDefauw · 2026-03-12T10:18:58Z

also added fix for #1091.
On linux, unit tests are failing, I will try to reproduce it

Used S0 instead of s0, for pyramid scale 0, therefore unit tests were failing on linux

LucaMarconato

Thanks for the PR! I would make some changes as described.

LucaMarconato · 2026-03-19T21:40:28Z

+def _prepare_single_scale_storage_options(
+ storage_options: JSONDict | list[JSONDict] | None,
+) -> JSONDict | list[JSONDict] | None:
+ if storage_options is None:
+ return None
+ if isinstance(storage_options, dict):
+ prepared = dict(storage_options)
+ if "chunks" in prepared:
+ prepared["chunks"] = _normalize_explicit_chunks(prepared["chunks"])
+ return prepared
+ return [dict(options) for options in storage_options]


This function behaves like _prepare_multiscale_storage_options(), without normalizing the list of storage options case. Can we remove it and just use _prepare_multiscale_storage_options() (after renaming it to _prepare_storage_options()?

I unified the two functions in 20696b5. Happy to hear what you think (in case we need two, we can revert, but I think we can proceed with one function).

LucaMarconato · 2026-03-19T21:41:11Z

+ if isinstance(value, str | bytes):
+ return False


Why do we need this? Please either remove or document.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:41:18Z

+ if isinstance(value, str | bytes):
+ return False


Same for this check.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:50:34Z

+def _is_regular_dask_chunk_grid(chunk_grid: Sequence[Sequence[int]]) -> bool:
+ # Match Dask's private _check_regular_chunks() logic without depending on its internal API.
+ for axis_chunks in chunk_grid:
+ if len(axis_chunks) <= 1:
+ continue
+ if len(set(axis_chunks[:-1])) > 1:
+ return False
+ if axis_chunks[-1] > axis_chunks[0]:
+ return False
+ return True


I'd add a docstring with examples (or examples in-line with the code) to show what fails and what not.
I would add the following:
triggers the continue in the first if:
[(4,)]
[()]
triggers the first return False
[(4, 4, 3, 4)]
triggers the second return False
[(4, 4, 4, 5)]
exits with the last return True
[(4, 4, 4, 4)], succeeds, all chunks equal
[(4, 4, 4, 1)], succeeds, final chunk is < of the initial one

Also: I would add all the examples above in a test, for the function _is_regular_dask_chunk_grid().

Added docstring and tests here: 2450bd4

LucaMarconato · 2026-03-19T22:04:20Z

+def test_write_irregular_dask_chunks_without_explicit_storage_options(tmp_path: Path) -> None:
+ data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
+ image = Image2DModel.parse(data, dims=("c", "y", "x"))
+ sdata = SpatialData(images={"image": image})
+
+ sdata.write(tmp_path / "data.zarr")


I would find it more natural if this test was failing. Now that chunks = raster_data.chunks has been removed it, writing doesn't fail, but it ignores the chunks in the data. I think a natural behavior is that if storage option specifies chunks, these are used, otherwise the ones from the data (and if the `chunks from the data are irregular and no storage options are specified, an error would be raised).

I now implemented this in 20696b5 changing the test so that it expects to fail.

LucaMarconato · 2026-03-19T22:16:19Z

I'll go ahead and implement a fix for the code review points.

LucaMarconato · 2026-03-19T22:18:17Z

the commit 930922c closes PointsModel.parse() fails with AttributeError: 'DataFrame' object has no attribute 'attrs' #1093, I included it here since I was encountering that when running some debug scripts.

…ied in storage options

LucaMarconato · 2026-03-19T22:29:10Z

I implemented the changes mentioned in the code review. Please let me know if you agree with the changes. If yes I'll merge and work on a release.

ArneDefauw · 2026-03-20T09:24:46Z

Hi @LucaMarconato , thanks for having a look!

My idea was to stop auto-filling storage_options["chunks"] with raster_data.data, i.e. here https://github.com/ArneDefauw/spatialdata/blob/2450bd437914004cc940279482f2dc60293e0575/src/spatialdata/_io/io_raster.py#L375

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

I still gaurded for irregular chunks, because spatialdata._io.io_raster import write_image exposes storage_options as a parameter.

I think there are valid arguments for both implementations.

If we go for auto-filling storage_options (your changes), then I would change the ValueError raised when doing e.g.:


import tempfile
from pathlib import Path
import dask.array as da
from numpy.random import default_rng
from spatialdata import SpatialData
from spatialdata.models import Image2DModel
tmpdir = Path(tempfile.mkdtemp())
RNG = default_rng(0)
data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
image = Image2DModel.parse(data, dims=("c", "y", "x"))
sdata = SpatialData(images={"image": image})
sdata.write(tmpdir / "data.zarr", overwrite=True)

from

ValueError: storage_options['chunks'] must be a Zarr chunk shape or a regular Dask chunk grid. Irregular Dask chunk grids must be rechunked before writing or omitted.

To specify how user can avoid it, e.g. change it to:

ValueError: storage_options["chunks"] must resolve to a Zarr chunk shape or a regular Dask chunk grid. The current raster has irregular Dask chunks, which cannot be written to Zarr. To fix this, rechunk before writing, for example by passing regular chunks=... to Image2DModel.parse(...) / Labels2DModel.parse(...).

ArneDefauw · 2026-03-20T10:22:52Z

Added small update to guard against storage_options["chunks"]="" and storage_options["chunks"]=b"auto" . Bit of an edge case, and the reason why this if isinstance(value, str | bytes) got introduced in

def _is_flat_int_sequence(value: object) -> TypeGuard[Sequence[int]]:
if isinstance(value, str | bytes):
return False
if not isinstance(value, Sequence):
return False
return all(isinstance(v, int) for v in value)

Also updated the ValueErorr to be more user friendly, see previous comment

LucaMarconato · 2026-03-20T13:49:28Z

Thanks for the comments and for the changes.

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

That's a good point! I restored the old implementation of _prepare_storage_options() that didn't depend on the data (while keeping the unified function instead of one for single-scale and one for multi-scale).

Ready to merge!

ArneDefauw · 2026-03-12T08:18:57Z

Fix for #1090.

Note that unit tests still fail for ome-zarr ==0.14.0, due to #1091

codecov · 2026-03-12T08:55:13Z

Codecov Report

❌ Patch coverage is 85.45455% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.93%. Comparing base (6a3eef7) to head (385dd2e).
⚠️ Report is 26 commits behind head on main.

Files with missing lines	Patch %	Lines
src/spatialdata/_io/io_raster.py	85.18%	8 Missing ⚠️

Additional details and impacted files

@@ Coverage Diff @@## main #1092 +/- ##
==========================================
- Coverage 91.96% 91.93% -0.04% 
==========================================
Files 51 51 Lines 7729 7772 +43 ==========================================
+ Hits 7108 7145 +37 - Misses 621 627 +6

Files with missing lines	Coverage Δ
src/spatialdata/__init__.py	`95.65% <100.00%> (-0.51%)`	⬇️
src/spatialdata/_io/io_raster.py	`92.09% <85.18%> (-1.81%)`	⬇️

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

ArneDefauw · 2026-03-12T09:28:25Z

also added fix for #1091.

On linux, unit tests are failing, I will try to reproduce it

ArneDefauw · 2026-03-12T10:18:58Z

also added fix for #1091.
On linux, unit tests are failing, I will try to reproduce it

Used S0 instead of s0, for pyramid scale 0, therefore unit tests were failing on linux

LucaMarconato

Thanks for the PR! I would make some changes as described.

LucaMarconato · 2026-03-19T21:40:28Z

+def _prepare_single_scale_storage_options(
+ storage_options: JSONDict | list[JSONDict] | None,
+) -> JSONDict | list[JSONDict] | None:
+ if storage_options is None:
+ return None
+ if isinstance(storage_options, dict):
+ prepared = dict(storage_options)
+ if "chunks" in prepared:
+ prepared["chunks"] = _normalize_explicit_chunks(prepared["chunks"])
+ return prepared
+ return [dict(options) for options in storage_options]


This function behaves like _prepare_multiscale_storage_options(), without normalizing the list of storage options case. Can we remove it and just use _prepare_multiscale_storage_options() (after renaming it to _prepare_storage_options()?

I unified the two functions in 20696b5. Happy to hear what you think (in case we need two, we can revert, but I think we can proceed with one function).

LucaMarconato · 2026-03-19T21:41:11Z

+ if isinstance(value, str | bytes):
+ return False


Why do we need this? Please either remove or document.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:41:18Z

+ if isinstance(value, str | bytes):
+ return False


Same for this check.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:50:34Z

+def _is_regular_dask_chunk_grid(chunk_grid: Sequence[Sequence[int]]) -> bool:
+ # Match Dask's private _check_regular_chunks() logic without depending on its internal API.
+ for axis_chunks in chunk_grid:
+ if len(axis_chunks) <= 1:
+ continue
+ if len(set(axis_chunks[:-1])) > 1:
+ return False
+ if axis_chunks[-1] > axis_chunks[0]:
+ return False
+ return True


I'd add a docstring with examples (or examples in-line with the code) to show what fails and what not.
I would add the following:
triggers the continue in the first if:
[(4,)]
[()]
triggers the first return False
[(4, 4, 3, 4)]
triggers the second return False
[(4, 4, 4, 5)]
exits with the last return True
[(4, 4, 4, 4)], succeeds, all chunks equal
[(4, 4, 4, 1)], succeeds, final chunk is < of the initial one

Also: I would add all the examples above in a test, for the function _is_regular_dask_chunk_grid().

Added docstring and tests here: 2450bd4

LucaMarconato · 2026-03-19T22:04:20Z

+def test_write_irregular_dask_chunks_without_explicit_storage_options(tmp_path: Path) -> None:
+ data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
+ image = Image2DModel.parse(data, dims=("c", "y", "x"))
+ sdata = SpatialData(images={"image": image})
+
+ sdata.write(tmp_path / "data.zarr")


I would find it more natural if this test was failing. Now that chunks = raster_data.chunks has been removed it, writing doesn't fail, but it ignores the chunks in the data. I think a natural behavior is that if storage option specifies chunks, these are used, otherwise the ones from the data (and if the `chunks from the data are irregular and no storage options are specified, an error would be raised).

I now implemented this in 20696b5 changing the test so that it expects to fail.

LucaMarconato · 2026-03-19T22:16:19Z

I'll go ahead and implement a fix for the code review points.

LucaMarconato · 2026-03-19T22:18:17Z

the commit 930922c closes PointsModel.parse() fails with AttributeError: 'DataFrame' object has no attribute 'attrs' #1093, I included it here since I was encountering that when running some debug scripts.

…ied in storage options

LucaMarconato · 2026-03-19T22:29:10Z

I implemented the changes mentioned in the code review. Please let me know if you agree with the changes. If yes I'll merge and work on a release.

ArneDefauw · 2026-03-20T09:24:46Z

Hi @LucaMarconato , thanks for having a look!

My idea was to stop auto-filling storage_options["chunks"] with raster_data.data, i.e. here https://github.com/ArneDefauw/spatialdata/blob/2450bd437914004cc940279482f2dc60293e0575/src/spatialdata/_io/io_raster.py#L375

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

I still gaurded for irregular chunks, because spatialdata._io.io_raster import write_image exposes storage_options as a parameter.

I think there are valid arguments for both implementations.

If we go for auto-filling storage_options (your changes), then I would change the ValueError raised when doing e.g.:


import tempfile
from pathlib import Path
import dask.array as da
from numpy.random import default_rng
from spatialdata import SpatialData
from spatialdata.models import Image2DModel
tmpdir = Path(tempfile.mkdtemp())
RNG = default_rng(0)
data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
image = Image2DModel.parse(data, dims=("c", "y", "x"))
sdata = SpatialData(images={"image": image})
sdata.write(tmpdir / "data.zarr", overwrite=True)

from

ValueError: storage_options['chunks'] must be a Zarr chunk shape or a regular Dask chunk grid. Irregular Dask chunk grids must be rechunked before writing or omitted.

To specify how user can avoid it, e.g. change it to:

ValueError: storage_options["chunks"] must resolve to a Zarr chunk shape or a regular Dask chunk grid. The current raster has irregular Dask chunks, which cannot be written to Zarr. To fix this, rechunk before writing, for example by passing regular chunks=... to Image2DModel.parse(...) / Labels2DModel.parse(...).

ArneDefauw · 2026-03-20T10:22:52Z

Added small update to guard against storage_options["chunks"]="" and storage_options["chunks"]=b"auto" . Bit of an edge case, and the reason why this if isinstance(value, str | bytes) got introduced in

def _is_flat_int_sequence(value: object) -> TypeGuard[Sequence[int]]:
if isinstance(value, str | bytes):
return False
if not isinstance(value, Sequence):
return False
return all(isinstance(v, int) for v in value)

Also updated the ValueErorr to be more user friendly, see previous comment

LucaMarconato · 2026-03-20T13:49:28Z

Thanks for the comments and for the changes.

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

That's a good point! I restored the old implementation of _prepare_storage_options() that didn't depend on the data (while keeping the unified function instead of one for single-scale and one for multi-scale).

Ready to merge!

ArneDefauw · 2026-03-12T08:18:57Z

Fix for #1090.

Note that unit tests still fail for ome-zarr ==0.14.0, due to #1091

codecov · 2026-03-12T08:55:13Z

Codecov Report

❌ Patch coverage is 85.45455% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.93%. Comparing base (6a3eef7) to head (385dd2e).
⚠️ Report is 26 commits behind head on main.

Files with missing lines	Patch %	Lines
src/spatialdata/_io/io_raster.py	85.18%	8 Missing ⚠️

Additional details and impacted files

@@ Coverage Diff @@## main #1092 +/- ##
==========================================
- Coverage 91.96% 91.93% -0.04% 
==========================================
Files 51 51 Lines 7729 7772 +43 ==========================================
+ Hits 7108 7145 +37 - Misses 621 627 +6

Files with missing lines	Coverage Δ
src/spatialdata/__init__.py	`95.65% <100.00%> (-0.51%)`	⬇️
src/spatialdata/_io/io_raster.py	`92.09% <85.18%> (-1.81%)`	⬇️

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

ArneDefauw · 2026-03-12T09:28:25Z

also added fix for #1091.

On linux, unit tests are failing, I will try to reproduce it

ArneDefauw · 2026-03-12T10:18:58Z

also added fix for #1091.
On linux, unit tests are failing, I will try to reproduce it

Used S0 instead of s0, for pyramid scale 0, therefore unit tests were failing on linux

LucaMarconato

Thanks for the PR! I would make some changes as described.

LucaMarconato · 2026-03-19T21:40:28Z

+def _prepare_single_scale_storage_options(
+ storage_options: JSONDict | list[JSONDict] | None,
+) -> JSONDict | list[JSONDict] | None:
+ if storage_options is None:
+ return None
+ if isinstance(storage_options, dict):
+ prepared = dict(storage_options)
+ if "chunks" in prepared:
+ prepared["chunks"] = _normalize_explicit_chunks(prepared["chunks"])
+ return prepared
+ return [dict(options) for options in storage_options]


This function behaves like _prepare_multiscale_storage_options(), without normalizing the list of storage options case. Can we remove it and just use _prepare_multiscale_storage_options() (after renaming it to _prepare_storage_options()?

I unified the two functions in 20696b5. Happy to hear what you think (in case we need two, we can revert, but I think we can proceed with one function).

LucaMarconato · 2026-03-19T21:41:11Z

+ if isinstance(value, str | bytes):
+ return False


Why do we need this? Please either remove or document.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:41:18Z

+ if isinstance(value, str | bytes):
+ return False


Same for this check.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:50:34Z

+def _is_regular_dask_chunk_grid(chunk_grid: Sequence[Sequence[int]]) -> bool:
+ # Match Dask's private _check_regular_chunks() logic without depending on its internal API.
+ for axis_chunks in chunk_grid:
+ if len(axis_chunks) <= 1:
+ continue
+ if len(set(axis_chunks[:-1])) > 1:
+ return False
+ if axis_chunks[-1] > axis_chunks[0]:
+ return False
+ return True


I'd add a docstring with examples (or examples in-line with the code) to show what fails and what not.
I would add the following:
triggers the continue in the first if:
[(4,)]
[()]
triggers the first return False
[(4, 4, 3, 4)]
triggers the second return False
[(4, 4, 4, 5)]
exits with the last return True
[(4, 4, 4, 4)], succeeds, all chunks equal
[(4, 4, 4, 1)], succeeds, final chunk is < of the initial one

Also: I would add all the examples above in a test, for the function _is_regular_dask_chunk_grid().

Added docstring and tests here: 2450bd4

LucaMarconato · 2026-03-19T22:04:20Z

+def test_write_irregular_dask_chunks_without_explicit_storage_options(tmp_path: Path) -> None:
+ data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
+ image = Image2DModel.parse(data, dims=("c", "y", "x"))
+ sdata = SpatialData(images={"image": image})
+
+ sdata.write(tmp_path / "data.zarr")


I would find it more natural if this test was failing. Now that chunks = raster_data.chunks has been removed it, writing doesn't fail, but it ignores the chunks in the data. I think a natural behavior is that if storage option specifies chunks, these are used, otherwise the ones from the data (and if the `chunks from the data are irregular and no storage options are specified, an error would be raised).

I now implemented this in 20696b5 changing the test so that it expects to fail.

LucaMarconato · 2026-03-19T22:16:19Z

I'll go ahead and implement a fix for the code review points.

LucaMarconato · 2026-03-19T22:18:17Z

the commit 930922c closes PointsModel.parse() fails with AttributeError: 'DataFrame' object has no attribute 'attrs' #1093, I included it here since I was encountering that when running some debug scripts.

…ied in storage options

LucaMarconato · 2026-03-19T22:29:10Z

I implemented the changes mentioned in the code review. Please let me know if you agree with the changes. If yes I'll merge and work on a release.

ArneDefauw · 2026-03-20T09:24:46Z

Hi @LucaMarconato , thanks for having a look!

My idea was to stop auto-filling storage_options["chunks"] with raster_data.data, i.e. here https://github.com/ArneDefauw/spatialdata/blob/2450bd437914004cc940279482f2dc60293e0575/src/spatialdata/_io/io_raster.py#L375

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

I still gaurded for irregular chunks, because spatialdata._io.io_raster import write_image exposes storage_options as a parameter.

I think there are valid arguments for both implementations.

If we go for auto-filling storage_options (your changes), then I would change the ValueError raised when doing e.g.:


import tempfile
from pathlib import Path
import dask.array as da
from numpy.random import default_rng
from spatialdata import SpatialData
from spatialdata.models import Image2DModel
tmpdir = Path(tempfile.mkdtemp())
RNG = default_rng(0)
data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
image = Image2DModel.parse(data, dims=("c", "y", "x"))
sdata = SpatialData(images={"image": image})
sdata.write(tmpdir / "data.zarr", overwrite=True)

from

ValueError: storage_options['chunks'] must be a Zarr chunk shape or a regular Dask chunk grid. Irregular Dask chunk grids must be rechunked before writing or omitted.

To specify how user can avoid it, e.g. change it to:

ValueError: storage_options["chunks"] must resolve to a Zarr chunk shape or a regular Dask chunk grid. The current raster has irregular Dask chunks, which cannot be written to Zarr. To fix this, rechunk before writing, for example by passing regular chunks=... to Image2DModel.parse(...) / Labels2DModel.parse(...).

ArneDefauw · 2026-03-20T10:22:52Z

Added small update to guard against storage_options["chunks"]="" and storage_options["chunks"]=b"auto" . Bit of an edge case, and the reason why this if isinstance(value, str | bytes) got introduced in

def _is_flat_int_sequence(value: object) -> TypeGuard[Sequence[int]]:
if isinstance(value, str | bytes):
return False
if not isinstance(value, Sequence):
return False
return all(isinstance(v, int) for v in value)

Also updated the ValueErorr to be more user friendly, see previous comment

LucaMarconato · 2026-03-20T13:49:28Z

Thanks for the comments and for the changes.

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

That's a good point! I restored the old implementation of _prepare_storage_options() that didn't depend on the data (while keeping the unified function instead of one for single-scale and one for multi-scale).

Ready to merge!

ArneDefauw · 2026-03-12T08:18:57Z

Fix for #1090.

Note that unit tests still fail for ome-zarr ==0.14.0, due to #1091

codecov · 2026-03-12T08:55:13Z

Codecov Report

❌ Patch coverage is 85.45455% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.93%. Comparing base (6a3eef7) to head (385dd2e).
⚠️ Report is 26 commits behind head on main.

Files with missing lines	Patch %	Lines
src/spatialdata/_io/io_raster.py	85.18%	8 Missing ⚠️

Additional details and impacted files

@@ Coverage Diff @@## main #1092 +/- ##
==========================================
- Coverage 91.96% 91.93% -0.04% 
==========================================
Files 51 51 Lines 7729 7772 +43 ==========================================
+ Hits 7108 7145 +37 - Misses 621 627 +6

Files with missing lines	Coverage Δ
src/spatialdata/__init__.py	`95.65% <100.00%> (-0.51%)`	⬇️
src/spatialdata/_io/io_raster.py	`92.09% <85.18%> (-1.81%)`	⬇️

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

ArneDefauw · 2026-03-12T09:28:25Z

also added fix for #1091.

On linux, unit tests are failing, I will try to reproduce it

ArneDefauw · 2026-03-12T10:18:58Z

also added fix for #1091.
On linux, unit tests are failing, I will try to reproduce it

Used S0 instead of s0, for pyramid scale 0, therefore unit tests were failing on linux

LucaMarconato

Thanks for the PR! I would make some changes as described.

LucaMarconato · 2026-03-19T21:40:28Z

+def _prepare_single_scale_storage_options(
+ storage_options: JSONDict | list[JSONDict] | None,
+) -> JSONDict | list[JSONDict] | None:
+ if storage_options is None:
+ return None
+ if isinstance(storage_options, dict):
+ prepared = dict(storage_options)
+ if "chunks" in prepared:
+ prepared["chunks"] = _normalize_explicit_chunks(prepared["chunks"])
+ return prepared
+ return [dict(options) for options in storage_options]


This function behaves like _prepare_multiscale_storage_options(), without normalizing the list of storage options case. Can we remove it and just use _prepare_multiscale_storage_options() (after renaming it to _prepare_storage_options()?

I unified the two functions in 20696b5. Happy to hear what you think (in case we need two, we can revert, but I think we can proceed with one function).

LucaMarconato · 2026-03-19T21:41:11Z

+ if isinstance(value, str | bytes):
+ return False


Why do we need this? Please either remove or document.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:41:18Z

+ if isinstance(value, str | bytes):
+ return False


Same for this check.

Removed here: 2450bd4

LucaMarconato · 2026-03-19T21:50:34Z

+def _is_regular_dask_chunk_grid(chunk_grid: Sequence[Sequence[int]]) -> bool:
+ # Match Dask's private _check_regular_chunks() logic without depending on its internal API.
+ for axis_chunks in chunk_grid:
+ if len(axis_chunks) <= 1:
+ continue
+ if len(set(axis_chunks[:-1])) > 1:
+ return False
+ if axis_chunks[-1] > axis_chunks[0]:
+ return False
+ return True


I'd add a docstring with examples (or examples in-line with the code) to show what fails and what not.
I would add the following:
triggers the continue in the first if:
[(4,)]
[()]
triggers the first return False
[(4, 4, 3, 4)]
triggers the second return False
[(4, 4, 4, 5)]
exits with the last return True
[(4, 4, 4, 4)], succeeds, all chunks equal
[(4, 4, 4, 1)], succeeds, final chunk is < of the initial one

Also: I would add all the examples above in a test, for the function _is_regular_dask_chunk_grid().

Added docstring and tests here: 2450bd4

LucaMarconato · 2026-03-19T22:04:20Z

+def test_write_irregular_dask_chunks_without_explicit_storage_options(tmp_path: Path) -> None:
+ data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
+ image = Image2DModel.parse(data, dims=("c", "y", "x"))
+ sdata = SpatialData(images={"image": image})
+
+ sdata.write(tmp_path / "data.zarr")


I would find it more natural if this test was failing. Now that chunks = raster_data.chunks has been removed it, writing doesn't fail, but it ignores the chunks in the data. I think a natural behavior is that if storage option specifies chunks, these are used, otherwise the ones from the data (and if the `chunks from the data are irregular and no storage options are specified, an error would be raised).

I now implemented this in 20696b5 changing the test so that it expects to fail.

LucaMarconato · 2026-03-19T22:16:19Z

I'll go ahead and implement a fix for the code review points.

LucaMarconato · 2026-03-19T22:18:17Z

the commit 930922c closes PointsModel.parse() fails with AttributeError: 'DataFrame' object has no attribute 'attrs' #1093, I included it here since I was encountering that when running some debug scripts.

…ied in storage options

LucaMarconato · 2026-03-19T22:29:10Z

I implemented the changes mentioned in the code review. Please let me know if you agree with the changes. If yes I'll merge and work on a release.

ArneDefauw · 2026-03-20T09:24:46Z

Hi @LucaMarconato , thanks for having a look!

My idea was to stop auto-filling storage_options["chunks"] with raster_data.data, i.e. here https://github.com/ArneDefauw/spatialdata/blob/2450bd437914004cc940279482f2dc60293e0575/src/spatialdata/_io/io_raster.py#L375

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

I still gaurded for irregular chunks, because spatialdata._io.io_raster import write_image exposes storage_options as a parameter.

I think there are valid arguments for both implementations.

If we go for auto-filling storage_options (your changes), then I would change the ValueError raised when doing e.g.:


import tempfile
from pathlib import Path
import dask.array as da
from numpy.random import default_rng
from spatialdata import SpatialData
from spatialdata.models import Image2DModel
tmpdir = Path(tempfile.mkdtemp())
RNG = default_rng(0)
data = da.from_array(RNG.random((3, 800, 1000)), chunks=((3,), (300, 200, 300), (512, 488)))
image = Image2DModel.parse(data, dims=("c", "y", "x"))
sdata = SpatialData(images={"image": image})
sdata.write(tmpdir / "data.zarr", overwrite=True)

from

ValueError: storage_options['chunks'] must be a Zarr chunk shape or a regular Dask chunk grid. Irregular Dask chunk grids must be rechunked before writing or omitted.

To specify how user can avoid it, e.g. change it to:

ValueError: storage_options["chunks"] must resolve to a Zarr chunk shape or a regular Dask chunk grid. The current raster has irregular Dask chunks, which cannot be written to Zarr. To fix this, rechunk before writing, for example by passing regular chunks=... to Image2DModel.parse(...) / Labels2DModel.parse(...).

ArneDefauw · 2026-03-20T10:22:52Z

Added small update to guard against storage_options["chunks"]="" and storage_options["chunks"]=b"auto" . Bit of an edge case, and the reason why this if isinstance(value, str | bytes) got introduced in

def _is_flat_int_sequence(value: object) -> TypeGuard[Sequence[int]]:
if isinstance(value, str | bytes):
return False
if not isinstance(value, Sequence):
return False
return all(isinstance(v, int) for v in value)

Also updated the ValueErorr to be more user friendly, see previous comment

LucaMarconato · 2026-03-20T13:49:28Z

Thanks for the comments and for the changes.

This way we let ome_zarr handling the rechunking of irregular chunks, and users of SpatialData do not need to worry about rechunking before writing to a zarr store using .write_element or .write. (ome_zarr writes a warning log message if rechunking is necessary for irregular chunks)

That's a good point! I restored the old implementation of _prepare_storage_options() that didn't depend on the data (while keeping the unified function instead of one for single-scale and one for multi-scale).

Ready to merge!

ArneDefauw added 2 commits March 12, 2026 09:17

ome zarr chunks
e9ea783

set scale factors to emtpy list + fix unit tests
c4e8608

mypy
da9eef3

ArneDefauw marked this pull request as ready for review March 12, 2026 09:04

lowercase to fix unit test linux
1c060b3

bump ome zarr in pyproject toml
85148d7

LucaMarconato mentioned this pull request Mar 19, 2026
SpatialData.write() fails for spatialdata.datasets.blobs() with ome-zarr==0.14.0 (TypeError: Expected an iterable of integers) #1090
Closed

LucaMarconato requested changes Mar 19, 2026
View reviewed changes

dask accessor is now always loaded
930922c

LucaMarconato added 2 commits March 19, 2026 23:19

deduplicate storage option util; use chunks from data when not specif…
20696b5
…ied in storage options

simplify, document and test the chunk helper functions
2450bd4

LucaMarconato mentioned this pull request Mar 19, 2026
Restrict ome-zarr version range in meta.yaml conda-forge/spatialdata-feedstock#33
Merged
5 tasks

guard against storage_options["chunks"]="" + Change ValueError
3dd6121

LucaMarconato added 2 commits March 20, 2026 14:47

remove data argument from _prepare_storage_options()
8c85587

remove data argument from _prepare_storage_options()
385dd2e

LucaMarconato approved these changes Mar 20, 2026
View reviewed changes

LucaMarconato merged commit 78f75ef into scverse:mainMar 20, 2026
9 checks passed

LucaMarconato added the release-fixed label Mar 23, 2026

Artur-man mentioned this pull request Apr 10, 2026
Unable to write to data.zarr file due to ome-zarr (0.14.0)BIMSBbioinfo/dummy-spatialdata#1
Open

lucas-diedrich mentioned this pull request Apr 22, 2026
Writing spatialdata object fails in v0.7 #1108
Closed

namsaraeva mentioned this pull request Apr 23, 2026
New release of spatialdata #1109
Closed

ArneDefauw added 2 commits March 12, 2026 09:17

ome zarr chunks
e9ea783

set scale factors to emtpy list + fix unit tests
c4e8608

mypy
da9eef3

ArneDefauw marked this pull request as ready for review March 12, 2026 09:04

lowercase to fix unit test linux
1c060b3

bump ome zarr in pyproject toml
85148d7

LucaMarconato mentioned this pull request Mar 19, 2026
SpatialData.write() fails for spatialdata.datasets.blobs() with ome-zarr==0.14.0 (TypeError: Expected an iterable of integers) #1090
Closed

LucaMarconato requested changes Mar 19, 2026
View reviewed changes

dask accessor is now always loaded
930922c

LucaMarconato added 2 commits March 19, 2026 23:19

deduplicate storage option util; use chunks from data when not specif…
20696b5
…ied in storage options

simplify, document and test the chunk helper functions
2450bd4

LucaMarconato mentioned this pull request Mar 19, 2026
Restrict ome-zarr version range in meta.yaml conda-forge/spatialdata-feedstock#33
Merged
5 tasks

guard against storage_options["chunks"]="" + Change ValueError
3dd6121

LucaMarconato added 2 commits March 20, 2026 14:47

remove data argument from _prepare_storage_options()
8c85587

remove data argument from _prepare_storage_options()
385dd2e

LucaMarconato approved these changes Mar 20, 2026
View reviewed changes

LucaMarconato merged commit 78f75ef into scverse:mainMar 20, 2026
9 checks passed

LucaMarconato added the release-fixed label Mar 23, 2026

Artur-man mentioned this pull request Apr 10, 2026
Unable to write to data.zarr file due to ome-zarr (0.14.0)BIMSBbioinfo/dummy-spatialdata#1
Open

lucas-diedrich mentioned this pull request Apr 22, 2026
Writing spatialdata object fails in v0.7 #1108
Closed

namsaraeva mentioned this pull request Apr 23, 2026
New release of spatialdata #1109
Closed

ArneDefauw added 2 commits March 12, 2026 09:17

ome zarr chunks
e9ea783

set scale factors to emtpy list + fix unit tests
c4e8608

mypy
da9eef3

ArneDefauw marked this pull request as ready for review March 12, 2026 09:04

lowercase to fix unit test linux
1c060b3

bump ome zarr in pyproject toml
85148d7

LucaMarconato mentioned this pull request Mar 19, 2026
SpatialData.write() fails for spatialdata.datasets.blobs() with ome-zarr==0.14.0 (TypeError: Expected an iterable of integers) #1090
Closed

LucaMarconato requested changes Mar 19, 2026
View reviewed changes

dask accessor is now always loaded
930922c

LucaMarconato added 2 commits March 19, 2026 23:19

deduplicate storage option util; use chunks from data when not specif…
20696b5
…ied in storage options

simplify, document and test the chunk helper functions
2450bd4

LucaMarconato mentioned this pull request Mar 19, 2026
Restrict ome-zarr version range in meta.yaml conda-forge/spatialdata-feedstock#33
Merged
5 tasks

guard against storage_options["chunks"]="" + Change ValueError
3dd6121

LucaMarconato added 2 commits March 20, 2026 14:47

remove data argument from _prepare_storage_options()
8c85587

remove data argument from _prepare_storage_options()
385dd2e

LucaMarconato approved these changes Mar 20, 2026
View reviewed changes

LucaMarconato merged commit 78f75ef into scverse:mainMar 20, 2026
9 checks passed

LucaMarconato added the release-fixed label Mar 23, 2026

Artur-man mentioned this pull request Apr 10, 2026
Unable to write to data.zarr file due to ome-zarr (0.14.0)BIMSBbioinfo/dummy-spatialdata#1
Open

lucas-diedrich mentioned this pull request Apr 22, 2026
Writing spatialdata object fails in v0.7 #1108
Closed

namsaraeva mentioned this pull request Apr 23, 2026
New release of spatialdata #1109
Closed

ArneDefauw added 2 commits March 12, 2026 09:17

ome zarr chunks
e9ea783

set scale factors to emtpy list + fix unit tests
c4e8608

mypy
da9eef3

ArneDefauw marked this pull request as ready for review March 12, 2026 09:04

lowercase to fix unit test linux
1c060b3

bump ome zarr in pyproject toml
85148d7

LucaMarconato mentioned this pull request Mar 19, 2026
SpatialData.write() fails for spatialdata.datasets.blobs() with ome-zarr==0.14.0 (TypeError: Expected an iterable of integers) #1090
Closed

LucaMarconato requested changes Mar 19, 2026
View reviewed changes

dask accessor is now always loaded
930922c

LucaMarconato added 2 commits March 19, 2026 23:19

deduplicate storage option util; use chunks from data when not specif…
20696b5
…ied in storage options

simplify, document and test the chunk helper functions
2450bd4

LucaMarconato mentioned this pull request Mar 19, 2026
Restrict ome-zarr version range in meta.yaml conda-forge/spatialdata-feedstock#33
Merged
5 tasks

guard against storage_options["chunks"]="" + Change ValueError
3dd6121

LucaMarconato added 2 commits March 20, 2026 14:47

remove data argument from _prepare_storage_options()
8c85587

remove data argument from _prepare_storage_options()
385dd2e

LucaMarconato approved these changes Mar 20, 2026
View reviewed changes

LucaMarconato merged commit 78f75ef into scverse:mainMar 20, 2026
9 checks passed

LucaMarconato added the release-fixed label Mar 23, 2026

Artur-man mentioned this pull request Apr 10, 2026
Unable to write to data.zarr file due to ome-zarr (0.14.0)BIMSBbioinfo/dummy-spatialdata#1
Open

lucas-diedrich mentioned this pull request Apr 22, 2026
Writing spatialdata object fails in v0.7 #1108
Closed

namsaraeva mentioned this pull request Apr 23, 2026
New release of spatialdata #1109
Closed

ArneDefauw added 2 commits March 12, 2026 09:17

ome zarr chunks
e9ea783

set scale factors to emtpy list + fix unit tests
c4e8608

mypy
da9eef3

ArneDefauw marked this pull request as ready for review March 12, 2026 09:04

lowercase to fix unit test linux
1c060b3

bump ome zarr in pyproject toml
85148d7

LucaMarconato mentioned this pull request Mar 19, 2026
SpatialData.write() fails for spatialdata.datasets.blobs() with ome-zarr==0.14.0 (TypeError: Expected an iterable of integers) #1090
Closed

LucaMarconato requested changes Mar 19, 2026
View reviewed changes

dask accessor is now always loaded
930922c

LucaMarconato added 2 commits March 19, 2026 23:19

deduplicate storage option util; use chunks from data when not specif…
20696b5
…ied in storage options

simplify, document and test the chunk helper functions
2450bd4

LucaMarconato mentioned this pull request Mar 19, 2026
Restrict ome-zarr version range in meta.yaml conda-forge/spatialdata-feedstock#33
Merged
5 tasks

guard against storage_options["chunks"]="" + Change ValueError
3dd6121

LucaMarconato added 2 commits March 20, 2026 14:47

remove data argument from _prepare_storage_options()
8c85587

remove data argument from _prepare_storage_options()
385dd2e

LucaMarconato approved these changes Mar 20, 2026
View reviewed changes

LucaMarconato merged commit 78f75ef into scverse:mainMar 20, 2026
9 checks passed

LucaMarconato added the release-fixed label Mar 23, 2026

Artur-man mentioned this pull request Apr 10, 2026
Unable to write to data.zarr file due to ome-zarr (0.14.0)BIMSBbioinfo/dummy-spatialdata#1
Open

lucas-diedrich mentioned this pull request Apr 22, 2026
Writing spatialdata object fails in v0.7 #1108
Closed

namsaraeva mentioned this pull request Apr 23, 2026
New release of spatialdata #1109
Closed

ArneDefauw added 2 commits March 12, 2026 09:17

ome zarr chunks
e9ea783

set scale factors to emtpy list + fix unit tests
c4e8608

mypy
da9eef3

ArneDefauw marked this pull request as ready for review March 12, 2026 09:04

lowercase to fix unit test linux
1c060b3

bump ome zarr in pyproject toml
85148d7

LucaMarconato mentioned this pull request Mar 19, 2026
SpatialData.write() fails for spatialdata.datasets.blobs() with ome-zarr==0.14.0 (TypeError: Expected an iterable of integers) #1090
Closed

LucaMarconato requested changes Mar 19, 2026
View reviewed changes

dask accessor is now always loaded
930922c

LucaMarconato added 2 commits March 19, 2026 23:19

deduplicate storage option util; use chunks from data when not specif…
20696b5
…ied in storage options

simplify, document and test the chunk helper functions
2450bd4

LucaMarconato mentioned this pull request Mar 19, 2026
Restrict ome-zarr version range in meta.yaml conda-forge/spatialdata-feedstock#33
Merged
5 tasks

guard against storage_options["chunks"]="" + Change ValueError
3dd6121

LucaMarconato added 2 commits March 20, 2026 14:47

remove data argument from _prepare_storage_options()
8c85587

remove data argument from _prepare_storage_options()
385dd2e

LucaMarconato approved these changes Mar 20, 2026
View reviewed changes

LucaMarconato merged commit 78f75ef into scverse:mainMar 20, 2026
9 checks passed

LucaMarconato added the release-fixed label Mar 23, 2026

Artur-man mentioned this pull request Apr 10, 2026
Unable to write to data.zarr file due to ome-zarr (0.14.0)BIMSBbioinfo/dummy-spatialdata#1
Open

lucas-diedrich mentioned this pull request Apr 22, 2026
Writing spatialdata object fails in v0.7 #1108
Closed

namsaraeva mentioned this pull request Apr 23, 2026
New release of spatialdata #1109
Closed

ArneDefauw added 2 commits March 12, 2026 09:17

ome zarr chunks
e9ea783

set scale factors to emtpy list + fix unit tests
c4e8608

mypy
da9eef3

ArneDefauw marked this pull request as ready for review March 12, 2026 09:04

lowercase to fix unit test linux
1c060b3

bump ome zarr in pyproject toml
85148d7

LucaMarconato mentioned this pull request Mar 19, 2026
SpatialData.write() fails for spatialdata.datasets.blobs() with ome-zarr==0.14.0 (TypeError: Expected an iterable of integers) #1090
Closed

LucaMarconato requested changes Mar 19, 2026
View reviewed changes

dask accessor is now always loaded
930922c

LucaMarconato added 2 commits March 19, 2026 23:19

deduplicate storage option util; use chunks from data when not specif…
20696b5
…ied in storage options

simplify, document and test the chunk helper functions
2450bd4

LucaMarconato mentioned this pull request Mar 19, 2026
Restrict ome-zarr version range in meta.yaml conda-forge/spatialdata-feedstock#33
Merged
5 tasks

guard against storage_options["chunks"]="" + Change ValueError
3dd6121

LucaMarconato added 2 commits March 20, 2026 14:47

remove data argument from _prepare_storage_options()
8c85587

remove data argument from _prepare_storage_options()
385dd2e

LucaMarconato approved these changes Mar 20, 2026
View reviewed changes

LucaMarconato merged commit 78f75ef into scverse:mainMar 20, 2026
9 checks passed

LucaMarconato added the release-fixed label Mar 23, 2026

Artur-man mentioned this pull request Apr 10, 2026
Unable to write to data.zarr file due to ome-zarr (0.14.0)BIMSBbioinfo/dummy-spatialdata#1
Open

lucas-diedrich mentioned this pull request Apr 22, 2026
Writing spatialdata object fails in v0.7 #1108
Closed

namsaraeva mentioned this pull request Apr 23, 2026
New release of spatialdata #1109
Closed

ArneDefauw added 2 commits March 12, 2026 09:17

ome zarr chunks
e9ea783

set scale factors to emtpy list + fix unit tests
c4e8608

mypy
da9eef3

ArneDefauw marked this pull request as ready for review March 12, 2026 09:04

lowercase to fix unit test linux
1c060b3

bump ome zarr in pyproject toml
85148d7

LucaMarconato mentioned this pull request Mar 19, 2026
SpatialData.write() fails for spatialdata.datasets.blobs() with ome-zarr==0.14.0 (TypeError: Expected an iterable of integers) #1090
Closed

LucaMarconato requested changes Mar 19, 2026
View reviewed changes

dask accessor is now always loaded
930922c

LucaMarconato added 2 commits March 19, 2026 23:19

deduplicate storage option util; use chunks from data when not specif…
20696b5
…ied in storage options

simplify, document and test the chunk helper functions
2450bd4

LucaMarconato mentioned this pull request Mar 19, 2026
Restrict ome-zarr version range in meta.yaml conda-forge/spatialdata-feedstock#33
Merged
5 tasks

guard against storage_options["chunks"]="" + Change ValueError
3dd6121

LucaMarconato added 2 commits March 20, 2026 14:47

remove data argument from _prepare_storage_options()
8c85587

remove data argument from _prepare_storage_options()
385dd2e

LucaMarconato approved these changes Mar 20, 2026
View reviewed changes

LucaMarconato merged commit 78f75ef into scverse:mainMar 20, 2026
9 checks passed

LucaMarconato added the release-fixed label Mar 23, 2026

Artur-man mentioned this pull request Apr 10, 2026
Unable to write to data.zarr file due to ome-zarr (0.14.0)BIMSBbioinfo/dummy-spatialdata#1
Open

lucas-diedrich mentioned this pull request Apr 22, 2026
Writing spatialdata object fails in v0.7 #1108
Closed

namsaraeva mentioned this pull request Apr 23, 2026
New release of spatialdata #1109
Closed

Uh oh!

Conversation

ArneDefauw commented Mar 12, 2026

Uh oh!

codecovBot commented Mar 12, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

ArneDefauw commented Mar 12, 2026

Uh oh!

ArneDefauw commented Mar 12, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

LucaMarconato left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

LucaMarconato commented Mar 19, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

LucaMarconato commented Mar 19, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

LucaMarconato commented Mar 19, 2026

Uh oh!

ArneDefauw commented Mar 20, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ArneDefauw commented Mar 20, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

LucaMarconato commented Mar 20, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

ArneDefauw commented Mar 12, 2026

Uh oh!

codecovBot commented Mar 12, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

ArneDefauw commented Mar 12, 2026

Uh oh!

ArneDefauw commented Mar 12, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

LucaMarconato left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

codecovBot commented Mar 12, 2026•
edited
Loading

ArneDefauw commented Mar 12, 2026•
edited
Loading

LucaMarconato commented Mar 19, 2026•
edited
Loading

LucaMarconato commented Mar 19, 2026•
edited
Loading

ArneDefauw commented Mar 20, 2026•
edited
Loading

ArneDefauw commented Mar 20, 2026•
edited
Loading

codecovBot commented Mar 12, 2026•
edited
Loading

ArneDefauw commented Mar 12, 2026•
edited
Loading

LucaMarconato commented Mar 19, 2026•
edited
Loading

LucaMarconato commented Mar 19, 2026•
edited
Loading

ArneDefauw commented Mar 20, 2026•
edited
Loading

ArneDefauw commented Mar 20, 2026•
edited
Loading