Description of the issue
After installing from source I get
ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject
pip reveals that pandas 2.1.0 and numpy 2.1.0 have been installed.
According to the changelog, pandas <2.2.2 was not fully compatible with numpy >=2.0 which can lead to the error above. After installing pandas==2.2.2 the error is gone.
Interestingly, pandas 2.2.2 (and numpy 2.1.0) is used when installing from pypi. However, the dev requirements installed above are not the reason for the different versions as I also tried without these.
In a later installation attempt, pandas 2.2.2 has been used end everything is fine. I guess, the pandas version 2.1.0 from my cache had been used before (which meets our requirement >=2.1).
Steps to Reproduce
- Install from source as described here (branch: production)
- Start python, type e.g.
from open_mastr.soap_api.download import MaStRAPI
Ideas of solution
To avoid incompatible versions, I propose to set min. pandas version to 2.2.2.
This is in accordance to our min. python version (v3.9) given in the toml.
Workflow checklist
Description of the issue
After installing from source I get
ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObjectpip reveals that pandas 2.1.0 and numpy 2.1.0 have been installed.
According to the changelog, pandas <2.2.2 was not fully compatible with numpy >=2.0 which can lead to the error above. After installing pandas==2.2.2 the error is gone.
Interestingly, pandas 2.2.2 (and numpy 2.1.0) is used when installing from pypi. However, the dev requirements installed above are not the reason for the different versions as I also tried without these.
In a later installation attempt, pandas 2.2.2 has been used end everything is fine. I guess, the pandas version 2.1.0 from my cache had been used before (which meets our requirement >=2.1).
Steps to Reproduce
from open_mastr.soap_api.download import MaStRAPIIdeas of solution
To avoid incompatible versions, I propose to set min. pandas version to 2.2.2.
This is in accordance to our min. python version (v3.9) given in the toml.
Workflow checklist