‘conda update -all’ yields different results on different machines:
Old HP Laptop/Win10 ca 2007, fresh install of miniconda and a few packages, and update-all a few days ago.
Versions:
pandas, numpy, numba: (‘1.4.1’, ‘1.20.3’, ‘0.54.1’)
Python 3.9.12
ipython 8.2.0
conda 4.12.0
2014 MacBook Air, older miniconda install, same packages, update -all today
Versions:
pandas, numpy, numba: (‘1.3.4’, ‘1.21.5’, ‘0.55.1’)
Python 3.7.13
ipython 7.20.0
conda 4.12.0
I am not an expert but why is one update quite a bit behind the other? Is it because of Mac vs PC? Would a install and reinstall help?
Advertisement
Answer
According to the conda docs, the result of running conda update python depends on which Python version is installed in the current environment. Python will be updated to the latest micro version of that version, but not to a later version.
For example, on your MacBook, you evidently ran the command in a Python 3.7.x environment, so it was updated to 3.7.13, which is the latest micro version of Python 3.7 (see Wikipedia). To get another version instead, the recommended way is to set up a new environment. When creating an environment, you can specify the Python version you want:
conda create --name my_environment python=3.9
The other packages are of course depending on the Python installation, so for example pandas was updated to version 1.3.4 on your MacBook, because I suppose this is the latest version that does not require Python 3.8+. But this dependency management can get very tricky, which is why setting up a new environment is recommended.