Zarr version

2.14.2

Numcodecs version

0.11

Python Version

3.10.9

Operating System

Mac OSX

Installation

pip into miniconda

Description

import zarr

Open store

z1 = zarr.open('data/example.zarr', mode='w', shape=(10000, 10000), chunks=(1000, 1000), dtype='i4')
z1[0,0] =41

Confirm value has been written

print("z1:" + str(z1[0,0]))

Open same store

z2 = zarr.open('data/example.zarr', mode='w', shape=(10000, 10000), chunks=(1000, 1000), dtype='i4')

Value is zero

print("z2:" + str(z2[0,0]))

Original value is also zero

print("z1:" + str(z1[0,0]))

python zarrMulti.py
z1:41
z2:0
z1:0

pip show zarr

Name: zarr
Version: 2.14.2
Summary: An implementation of chunked, compressed, N-dimensional arrays for Python

Steps to reproduce

Run script provided

Additional output

python zarrMulti.py
z1:41
z2:0
z1:0

This shows the value of the store that is written, and then loading the zarr store reflects that the data was either not stored, or the data is being overwritten.