Repository navigation
Some numcodecs.zarr3 codecs fail with 3.0.4+ #2900
Description
Zarr version
3.0.4+
Numcodecs version
0.15.0
Python Version
3.12
Operating System
Linux
Installation
See reproducer
Description
The ZFPY and Pcodec codecs (and perhaps more) fail since 3.0.4 because they expect input arrays to be contiguous
Steps to reproduce
#!/usr/bin/env -S uv run
# /// script
# requires-python = ">=3.12"
# dependencies = [
# # "zarr==3.0.2,<3.0.3", # WORKS
# "zarr==3.0.4", # FAILS
# "numcodecs==0.15.0",
# "zfpy==1.0.1",
# "pcodec==0.3.2",
# ]
# ///
import numpy as np
import zarr
from numcodecs.zarr3 import ZFPY, PCodec
for serializer in [
ZFPY(mode = 4, tolerance=0.01),
PCodec(level = 8, mode_spec="auto")
]:
array = zarr.create_array(
store=zarr.storage.MemoryStore(),
shape=[2, 2],
chunks=[2, 1],
dtype=np.float32,
serializer=serializer,
)
array[...] = np.array([[0, 1],[2, 3]])Additional output
No response
Activity
d-v-b commentedon Mar 10, 2025
thanks for this report. I see 2 problems here:
mickyals commentedon Apr 3, 2025
Zarr Version
3.0.6
Numcodecs Version
0.15.1
Operating System
Linux
Description
The same issue as described previously -
ValueError: an array with contiguous memory is requiredoccurs due to PCodec expectation of contiguous arrays in memory. The result is corruption of the compressed arrays as shown below. Are there any updates regarding this bug?dstansby commentedon Apr 3, 2025
Yes, this issue bisects to 8b59a38. cc @dcherian who made that commit, do you have any insight into what a fix could be here?
d-v-b commentedon Apr 3, 2025
@dcherian any guesses what could be going on here?
dcherian commentedon Apr 4, 2025
Sorry for not responding. We are now better about not copying, so we are tripping up numcodecs
arr.copyfor that branch.terraputix commentedon Aug 3, 2025
I submitted a fix for this issue here.