Cheat_sheet__env_and_kernels.txt

#
# Virtual environments
# A walled garden into which you download packages needed to run your code
# Used to freeze versions and make your code future proof
#

https://conda.io/projects/conda/en/latest/user-guide/tasks/manage-environments.html

Windows: Use anaconda prompt
Mac: Use termail or anaconda prompt

Navigate to the folder containing the environment file. 
$ conda env create -f environment.yml

Or create a new environement from scratch:

$ conda create --name <myenv>
$ conda create --name <myenv> scipy 		# creates with a specified package downloaded
$ conda create -n <myenv> scipy=0.17.3 		# creates with a specified version of a package downloaded
$ conda create --name <myenv> python=2.7 	# creates with a specific version of python
$ conda create --name <myclone> --clone <myenv> # create a clone of an existing env
$ conda env create -f environment.yml

Creating an environment means we have downloaded the packages. 

$ conda activate <myenv>
Means we are now inside that set of downloaded packages

Other useful commands:
$ conda env list
$ conda list
$ conda remove --name <myenv> --all

To add packages on the fly, you can pip or conda install them. 

$ pip install <PackageName>
$ conda install <PackageName>

There is a bug in the interaction between pip and conda. 
Once you have started using pip to install packages in your environment, do not go back to use conda.
There are more packages registered on pip than conda.

If you have multiple versions of python on the go, then use the "python -m" prefix. eg:
$ python -m pip install <PackageName>

#
# Kernels
# A kernel provisions ipython to the jupyter notebook and allows you to do interactive programming. 
#

Your env must contain ipykernal:
Activate your environment and install ipykernel...
$ python -m ipykernel install --user --name <myenv>

Can launch jupyter lab from within that environment. 
If you do that, then the active environement should be the defult. 
You can check by opening a jupyter notebook and pasting this. 
$ import os
$ print(os.environ['CONDA_DEFAULT_ENV'])

Explore what kernels are generally available across your system using prompt
$ jupyter kernelspec list
Add your active environment to the generally available list
$ python -m ipykernel install --name {MACHINE_READABLE_NAME} --display-name "{DISPLAY_NAME}" --user
Remove environment form the generally available list
$ jupyter kernelspec uninstall {MACHINE_READABLE_NAME} 

-- {MACHINE_READABLE_NAME} is a placeholder for the machine project name. This name can only contain ASCII letters and numbers and these separators: dash, underscore, and period.
-- {PROJECT-NAME} is a placeholder for the human-readable project name. This name may contain spaces and is shown on the Launcher page.

Method from this website, more notes found there...
https://docs.posit.co/ide/server-pro/user/2022.12.0+353.pro20/jupyter-lab/guide/jupyter-kernel-management.html

Previously nb_conda_kernel was used:
$ conda install nb_conda_kernel
https://stackoverflow.com/questions/67202874/what-is-nb-conda-kernels-equivalent-for-python-3-9
Google or ChatGPT may return this as a recommendation, but it does not work with Python 3.9 (yet)

There are other approaches...
https://towardsdatascience.com/get-your-conda-environment-to-show-in-jupyter-notebooks-the-easy-way-17010b76e874
It can be a rabbit hole.