Skip to content

Create Your Own Environment

Self-managed

Use this setup if you are following the tutorial on your own x86-64 Linux computer or server.

You need:

  • Linux.
  • x86-64 architecture.
  • Conda or Miniforge.
  • Git.

If Conda is not installed, install Miniforge before continuing.

1. Clone the tutorial repository

Clone the repository into the standard tutorial location:

git clone https://github.com/floability/tutorial-portable-by-design.git ~/tutorial
cd ~/tutorial

The repository contains the tutorial instructions, scripts, and examples. Run the remaining setup commands from the repository root.

2. Create the tutorial environment

Run:

conda create -y -n tutorial-env \
  -c conda-forge \
  --strict-channel-priority \
  python=3.11 \
  pip \
  git \
  jupyterlab \
  ndcctools=7.17.1 \
  floability=0.3.0 \
  cloudpickle=3.1.2 \
  numpy=2.4.3 \
  pandas \
  matplotlib \
  onnxruntime=1.28.0 \
  pillow=12.1.1

Activate it:

conda activate tutorial-env

Install Sciunit inside the active environment:

python -m pip install --upgrade sciunit2

The RAG-Lite exercise intentionally uses a separate environment containing its application dependencies. Keeping those packages out of tutorial-env lets you demonstrate that Sciunit can repeat a captured execution without them being installed in the active environment.

Whenever you open a new terminal and return to the tutorial, activate it again:

conda activate tutorial-env

3. Check the installation

Run:

python --version
vine_worker --version
sciunit --version
python -c "from sciunit2.command.exec_ import ExecCommand; print('Sciunit: OK')"
python -c "import ndcctools.taskvine; print('TaskVine: OK')"
python -c "import cloudpickle, matplotlib, numpy, onnxruntime, pandas; from PIL import Image; print('Scientific examples: OK')"
floability --help >/dev/null && echo "Floability: OK"
python -m pip check

You should finish with:

Sciunit: OK
TaskVine: OK
Scientific examples: OK
Floability: OK
No broken requirements found.

4. Understand the environment layout

The tutorial-env environment contains the same participant-facing package set as the live AWS environment. It supports the TaskVine, matrix, MobileNet, Sciunit, and Floability exercises. The RAG-Lite Sciunit exercise creates a separate raglite-sciunit-env for its application dependencies, then returns to tutorial-env to demonstrate dependency-free replay.

During later exercises, Floability may create additional software environments for individual backpacks. Those environments are separate from this tutorial tools environment and may be cached for reuse.

The self-managed exercises use directly launched TaskVine workers. A Slurm installation is not required.

You are ready

Continue to TaskVine Quickstart.