This repository acts as initial boiler plate for setting up a Miniconda driven Python development environment.
- Download a Python 10 variant of Miniconda from this link.
- Open the downloaded .exe file.
- Follow the installation instructions:
- Screen one: "Next"
- Screen two: "Agree"
- Screen three: "Next"
- Screen four: "Next"
- Screen five: Select the checkbox to add Miniconda to the PATH and click "install"
- Screen six: "Next"
- Screen seven: "Install"
- Open a bash terminal (Git Bash or other).
- Ensure your current working directory is in the repository root folder.
- Execute the command
. setup-env.sh
You will now have an activated virtual environment in your terminal. To deactivate your environment use the command conda deactivate
To remove the environment entirely use conda env remove --name dev_env
Why are you using source activate dev_env and not conda activate dev_env in your setup.sh script?
I got this error - CommandNotFoundError: Your shell has not been properly configured to use 'conda activate'. If using 'conda activate' from a batch script, change your invocation to 'CALL conda.bat activate'.. I felt simply using source as I have done in the past was a touch cleaner and acts a non intrusive subsitute.
Why did you remove/change some of the package versions?
Several sub dependency conflicts prevented the environment build, e.g.
Package pathlib2 conflicts for:
sqlalchemy=1.4.39 -> importlib-metadata -> pathlib2
jupyterlab=3.6.1 -> ipython -> pathlib2
The Conda depedency resolver, in its default configuration did not demonstrate the agility on Windows to solve the conflicts.
I also slimlined the environment to essentials for ease of debugging, we can add more dependencies incrementally as needed. To do so, install the package without a version and use conda list to get the package version that satistifies Conda before adding it to env.yml.
The original goal of this template repository was to encapsulate the entire setup into a single script. From the installation of Miniconda into the Windows operating system to the eventual activation of a development-ready virtual environment. This ambition was causing problems, particularly during the installation of Miniconda. Attempts at scripting an approach to perform this task are shown below.
curl -O https://repo.anaconda.com/miniconda/Miniconda-latest-Windows-x86_64.exe
mkdir %UserProfile%\Miniconda3
start /w "" %~dp0Miniconda-latest-Windows-x86_64.exe /InstallationType=JustMe /RegisterPython=0 /S /D=%UserProfile%\Miniconda3
curl --output Miniconda-latest-Windows-x86_64.exe --url https://repo.anaconda.com/miniconda/Miniconda-latest-Windows-x86_64.exe
mkdir "C:\Program Files\Miniconda3"
Start-Process Miniconda-latest-Windows-x86_64.exe -Wait -ArgumentList @('/S', '/InstallationType=JustMe', '/RegisterPython=0', "/D=C:\Program Files\Miniconda3")