How to install RASA 3.X without Docker or enabling Windows Long Path Support, to install its Tensorflow dependency, that has risk of file corruption?

Question

RASA 3.X is a chatbot development framework that has Tensorflow as its training dependency. Pip installing RASA using Windows or VS Code Command Prompt leads to error in installation of Tensorflow, which is further elaborated in attempt to download Tensorflow separately. The error displayed is:

ERROR: Could not install packages due to an OSError: [Errno 2] No such file or directory: 'C:\\Users\\DC\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.10_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python310\\site-packages\\tensorflow\\include\\external\\com_github_grpc_grpc\\src\\core\\ext\\filters\\client_channel\\lb_policy\\grpclb\\client_load_reporting_filter.h'

Although there is a similar problem solved on StackOverflow, but there is no approach listed for installing RASA or its dependencies without enabling Long Path Support that is risky due to file corruption so unsuitable for sensitive production environments.

Firstly, I tried searching for solutions for the Tensorflow download issue and came across the solution: TensorFlow 2.8.0 installing error using pip and poetry with python 3.10.4 as well as an unanswered question on that still asks for long path enabling: TensorFlow install error, Windows LongPath support not enabled

Next, I was suggested to pull Docker of RASA directly, however I was not allowed to install Docker by changing the BIOS settings on my workplace system due to security policies, thus I had to find alternative.

Finally, I came across that suggested creating a Conda environment to download RASA 3.X with Tensorflow dependency successfully, which I will detail in the solution.

Answer 1

Download Anaconda for creating a Python Environment for RASA:

1. Download Anaconda from: https://www.anaconda.com/download
2. Install the exe file, with pre-ticked checkboxes, only unchecking the last two checkboxes once installed
3. Go to System Properties > Environment Variables > Path > Edit > New
4. Paste the path of the scripts folder that is inside the folder where you installed Conda
5. Close all the windows and it will be saved itself

Create a RASA Environment (remember to change VS-Code CLI from Powershell to Command Prompt)

1. create --name rasa-env python=3.8 activate rasa-env
2. pip install rasa
3. rasa init
4. conda env export --file environment.yml
5. To recreate the environment when collaborating: conda env create -n rasa-env -f environment.yml