I use Anaconda3 GUI to instantiate a vanilla python kernel 3.8 virtual environment. I then use Anaconda3 GUI to “install” jupyter-notebook in said environment. I then use the Anaconda3 GUI to start a shell in this environment. I then execute > jupyter-notebook at the shell’s command line. The notebook starts just fine. I then use notebook to instantiate a ipynb file with any listed kernel (presuming that the kernel in this environment must necessarily be on this list) and not one works.
It isn’t clear to me that this rather linear, uncomplicated process could possibly fail. Each gesture is exactly predicated on the last. How did I step outside the rational execution envelope? And how do I step back in?
I should add that the CFO of a small python-based asset trading company and the CTO of a medical informatics company have advised me NOT to use anaconda as it is unpredictable. My personal experience is that it does require rip-out-and-reinstall protocol every few months. Does Jupyter also advise against using Anaconda as a virtual environment and package management tool? Such a conclusion would be consistent with my experience.