[Bug] Numba version interface has changed

DEUCE1957 commented 1 year ago

Bug report checklist

[X] Searched the issues page for similar reports Related Issue: https://github.com/e2nIEE/pandapower/issues/2041
[X] Read the relevant sections of the documentation
[X] Browse the tutorials and tests for usefull code snippets and examples of use
[X] Reproduced the issue after updating with pip install --upgrade pandapower (or git pull)
[X] Tried basic troubleshooting (if a bug/error) like restarting the interpreter and checking the pythonpath

Reproducible Example

In the current version of the PandaPower codebase, the version of Numba is checked in auxiliary.py by importing:

from numba._version import version_version as numba_version

However, in the latest versions of Numba (checked with Numba 0.58.1), the contents of numba._version are:

import json

version_json = '''
{
 "date": "2023-10-16T15:33:43+0200",
 "dirty": false,
 "error": null,
 "full-revisionid": "d4460feb8c91213e7b89f97b632d19e34a776cd3",
 "version": "0.58.1"
}
'''  # END VERSION_JSON

def get_versions():
    return json.loads(version_json)

So the new way to check the version would be to import:

from numba._version import get_versions

Then inside the function '_check_if_numba_is_installed':

def _check_if_numba_is_installed(use_numba:bool):
    try:
        # Get the numba version from the JSON description
        if version.parse(numba._version.get_versions()["version"]) < version.parse("0.2.5"):
            logger.warning('Warning: numba version too old -> Upgrade to a version > 0.25.\n' +
                           numba_warning_str)
            use_numba= False
    except:
        logger.warning(numba_warning_str)
        use_numba= False
    return use_numba

Where I have renamed the local variable "numba" to "use_numba" to prevent a name conflict with the numba module.

The downside of this solution is that it may break again if numba._version changes again, perhaps remove the try-except block or add the normal error message to the numba_warning_str.

Alternatively, we could just the package attribute:

if version.parse(numba.__version__) < version.parse("0.2.5"):
     ...

Issue Description and Traceback

If you comment out the try-except keywords in the "_check_if_numba_is_installed" function (auxiliary._check_if_numba_is_installed) then we get the following error:

Import Error: cannot import name 'version_version' from 'numba._version'

Otherwise a warning message is shown:

numba cannot be imported and numba functions are disabled. Probably the execution is slow. Please install numba to gain a massive speedup. (or if you prefer slow execution, set the flag=False to avoid this warning)

Expected Behavior

I expect no warning message nor error, since I have Numba installed and the version is greater than 0.2.5.

Installed Versions

Python: 3.11.3 Numba: 0.58.1 Pandapower: 2.13.1

Label

[x] Relevant labels are selected

rbolgaryn commented 1 year ago

Hi @DEUCE1957 ,

thank you for the issue!

Can you please double-check if the issue is occurring on the develop branch?

BackMountainDevil commented 10 months ago

python 3.10.11 with pandapower-2.13.1. numba warning version(0.57.0, 0.58.0, 0.58.1). numba-0.56.0 is fine. test code is here

BDonnot commented 8 months ago

Hello,

I am still getting this error:

numba cannot be imported and numba functions are disabled. Probably the execution is slow. Please install numba to gain a massive speedup. (or if you prefer slow execution, set the flag numba=False to avoid this warning!)

With the latest version of numba (0.59.1) and pandapower (2.13.1) on ubuntu 20.04.

I am pretty sure I got similar error messages for every version of numba (>= 0.58) and pandapower. Only way for me to silence these issues is either not to use numba (which is much slower) or downgrade to numba < 0.57 (which does not work on python >= 3.11)

e2nIEE / pandapower