Hello,
You definitely need to have Python 3.9 for the moment. We hope to support Python 3.11 in the next couple of months, but right now, we are experience a problem with a 3rd party library.
The typical way to deal with multiple Python version, and also the recommended way to install packages, is to use Python virtual environments (env for short). A Python environment is nothing like a VM or a container. It is instead a directory containing everything that Python needs to run (Python binaries, all the necessary modules, etc.). Installing Python modules or tools can be made within an environment. This avoids polluting your global Python and system directories. When you want to start from fresh, or get rid of a library, just delete the environment folder. Likewise, you can have multiple env folders (one folder per env) to install multiple versions of a given tool/SDK/module and switch from one env to another. Envs are isolated from one-another and from the system directories.
A popular tool to manage environments is called "venv". It is Python-native, but it has been seen causing some hard-to-troubleshoot issues, due to incompatible binary libraries (openssl for instance).
For security and complexity-management reasons, RADKit has rather strict dependencies on some libraries.
We like to use Anaconda environments (miniconda to be specific, which is a light version of Anaconda) as a reference point. Conda (the short name for Anaconda, miniconda etc.) is also much more forgiving for new users.
You can explore miniconda from here:
https://docs.conda.io/en/latest/miniconda.html
Also, our team loves Linux over which most of our build pipeline is developed. We support Windows, Mac and Linux. The main (only) hardship with Linux are the many flavors; some distros are more modern than others.
Assuming you have a Linux distribution that support miniconda, and assuming you are ok with miniconda, just install it
If Anaconda / miniconda or environments in general are completely new to you, please familiarize yourself as it is a good investment. Here is an intro video:
https://www.youtube.com/watch?v=lBXDOY6ZtBg
Just ignore the part that says to "stop using pip"; you can also use pip inside an environment and you will have to do that in practice; the world is not perfect.
Then download the RADKit wheels .tar.gz for Linux (as of today, that would be cisco_radkit_1.4.9_pip_linux.tgz in the download area).
Then follow the instructions from here which should be exactly what you need to get going on Linux + conda + pip :
https://radkit.cisco.com/docs/pages/start_pip_wheels.html#start-pip-wheels
specifically:
# Install miniconda - folow their instructions
# create a directory to unpack the RADKit packages
mkdir radkit_wheels
cd radkit_wheels
# move cisco_radkit_1.4.9_pip_linux.tgz in this directory
# Now unpack the tgz archive. The wheels are in it
tar xzf cisco_radkit_1.4.9_pip_linux.tgz
# create an environment using conda
conda create -n radkit-1.4.9 python=3.9
# activate the environment
conda activate radkit-1.4.9
# from here onward, anything you pip install will be deployed in this environment
# so lets' install RADKit in it
# pip install -f . means install the wheels found in . which means "the current directory"
# pip will prefer the local file, and will download the rest from the internet
pip install -f . cisco_radkit_service cisco_radkit_client
# You can now delete the radkit_wheels directory.
cd ..
rm -rf radkit_wheels
As long as your shell prompt shows it, your env is active.
From there, use RADKit normally (radkit-client, radkit-network-console, radkit-service, radkit-service-gui, ...), code (import the modules etc.)
If you want to exit the environment (e.g. if you need to access your global python install):
conda deactivate
The RADKit CLI will not be accessible anymore (it won't be in your shell or python path any longer).
At any time, you can get back to the env an use RADKit again:
conda activate radkit-1.4.9
Since pip has its own will and can be difficult, we recommend a fresh install in a fresh environment (you can create as many envs as your disk can take -- that is usually "a lot").
I hope this helps!
- Fred from the RADKit team
