product is deprecated as of NumPy 1.25.0 #996
Description
Most of the warnings raised during tests are about product being deprecated as of Numpy 1.25.
Whithin climada, they are raised here:
| warning | module | lineno |
|---|---|---|
product is deprecated as of NumPy 1.25.0, and will be removed in NumPy 2.0. Please use prod instead. |
climada.hazard.io | 1149 |
product is deprecated as of NumPy 1.25.0, and will be removed in NumPy 2.0. Please use prod instead. |
climada.hazard.io | 1169 |
product is deprecated as of NumPy 1.25.0, and will be removed in NumPy 2.0. Please use prod instead. |
climada.hazard.centroids.centr | 1003 |
product is deprecated as of NumPy 1.25.0, and will be removed in NumPy 2.0. Please use prod instead. |
climada.hazard.centroids.centr | 1004 |
product is deprecated as of NumPy 1.25.0, and will be removed in NumPy 2.0. Please use prod instead. |
climada.hazard.centroids.centr | 1022 |
product is deprecated as of NumPy 1.25.0, and will be removed in NumPy 2.0. Please use prod instead. |
climada.util.hdf5_handler | 68 |
product is deprecated as of NumPy 1.25.0, and will be removed in NumPy 2.0. Please use prod instead. |
climada.hazard.centroids.centr | 1006 |
product is deprecated as of NumPy 1.25.0, and will be removed in NumPy 2.0. Please use prod instead. |
climada.hazard.centroids.centr | 1007 |
The cause of this is h5py being restricted to <=3.8, because eccodes is pinned to 2.27.
It happens wherever np.array(obj) is called on a obj that is a h5py Store or Dataset.
The solution is to unpin eccodes, as requested in PR #947
Alternatively, all lines with np.array(obj) could be replaced by obj[:] (https://github.com/CLIMADA-project/climada_python/tree/feature/avoid_product_in_h5py)