Guidance on Correct Environment Setup & IDE Configuration for Isaac Sim Source Build

[ayaancdry]

Hello Isaac Lab Team,

First, thank you for your work on this incredible project. I've successfully managed to get training scripts running with the latest feature/isaacsim_5_0 branch and a source build of Isaac Sim. However, I am very confused about the intended developer environment and workflow, and I'm concerned my current working state might not be correct or robust for future development.

My Setup:

• OS: Ubuntu 22.04
• GPU: RTX 5070 Ti (requiring the nightly PyTorch build, which is installed and working)
• Installation: Followed the official docs to build Isaac Sim from source, then installed the feature/isaacsim_5_0 branch of Isaac Lab. Everything is located in a parent directory, like ~/project_isaac/, with IsaacSim and IsaacLab as siblings inside. The symbolic link _isaac_sim has been created correctly.
• Current Status: The core functionality works. I can successfully run ./isaaclab.sh -p scripts/reinforcement_learning/skrl/train.py --task Isaac-Ant-v0.

My Question and Point of Confusion:

My main issue is with the environment setup scripts. When I run source ../IsaacSim/_build/linux-x86_64/release/setup_conda_env.sh or source ../IsaacSim/_build/linux-x86_64/release/setup_python_env.sh, nothing seems to happen. No new activatable environment appears in conda env list, no venv folder is created, and my shell prompt does not change.

This has led me to work without an explicitly "activated" environment, which feels fragile. I would be grateful for clarification on the following points:

1. Purpose of Setup Scripts: What is the intended purpose of the setup_conda_env.sh and setup_python_env.sh scripts? Are they supposed to create a persistent environment that developers should activate manually in new sessions, and if so, how?

2. Recommended Workflow: What is the official, recommended workflow for developing within Isaac Lab? Should we always rely on the ./isaaclab.sh -p launcher to run our scripts, or is there a way to "enter" the correct development environment first so that commands like python and pip are correctly pathed?

3. Potential Future Issues: Will my current approach—working without an explicitly activated environment—cause issues later when I need to install additional packages with pip or work on more complex, multi-terminal ROS 2 projects that need to interface with the correct Python interpreter?

4. IDE Configuration: For IDE integration, specifically with VS Code, what is the definitive path to the Python interpreter that we should configure in settings.json? Is it the one located at .../_build/linux-x86_64/release/kit/python/bin/python3, or is there another intended method?

[RandomOakForest]

Thank you for posting this question. I'll move it to our Discussions section for follow up. Here is a summary that you may find helpful. Thank you for your interest in Isaac Lab.

Isaac Sim & Isaac Lab Developer Environment: Clarification and Best Practices

1. Purpose of setup_conda_env.sh and setup_python_env.sh

• setup_conda_env.sh and setup_python_env.sh are not designed to create new Conda or Python virtual environments.
  • Instead, these scripts set up environment variables (like PYTHONPATH, LD_LIBRARY_PATH, etc.) so that the Isaac Sim Python environment and its extensions are correctly configured for the current shell session¹.
  • Sourcing these scripts does not create a new environment visible in conda env list or generate a venv folder. They simply update your current shell's environment variables to point to the correct libraries and binaries.
• Persistent Activation:
  • These scripts are meant to be sourced in each new shell session where you want to use Isaac Sim directly, but they do not provide a persistent, activatable environment like a Conda or venv environment would¹.

2. Recommended Workflow for Isaac Lab Development

• Preferred Approach:
  • The official workflow is to use the ./isaaclab.sh launcher, which automatically sets up the correct environment and invokes the right Python interpreter (bundled with Isaac Sim or from your active virtual environment)².
  • When you run ./isaaclab.sh -p ..., it ensures all necessary environment variables are set, and the correct Python is used, regardless of whether you are in a Conda/venv environment or not.
• Conda Environment (Optional):
  • If you prefer a more standard Python development workflow, you can create a Conda environment as follows.
  • Once activated, you can use python and pip as usual, and the environment will be correctly configured for Isaac Lab development².

./isaaclab.sh --conda
# or with a custom name
./isaaclab.sh --conda my_env

Then activate it with:

conda activate env_isaaclab  # or your custom name

• Direct Python Usage:
  • If you are inside an activated Conda/venv environment, you can run scripts with python ... directly, and the environment will be set up correctly, provided you have installed the necessary dependencies².

3. Potential Issues with Not Using an Activated Environment

• Package Management:
  • Working without an explicitly activated environment (Conda or venv) can lead to:
    • Difficulty managing dependencies (risk of polluting system Python or conflicting packages).
    • Problems when installing additional packages with pip, as they may not be installed in the correct Python environment.
    • Challenges in multi-terminal workflows (e.g., ROS 2 projects), where each terminal must have the correct environment variables and Python interpreter²¹.
• Best Practice:
  • For robust development, especially with complex or multi-terminal projects, it is strongly recommended to use a Conda or venv environment and activate it before working. This ensures all dependencies are isolated and reproducible².

4. IDE (VS Code) Configuration

• Recommended Python Interpreter Path:
  • The Isaac Lab documentation recommends setting the Python interpreter in VS Code to the python.sh script provided by Isaac Sim, via the symbolic link:

"${workspaceFolder}/_isaac_sim/python.sh"

This ensures that all environment variables and paths are set up correctly when running or debugging code from VS Code³.

• Alternative (Conda/venv):
  • If you are using a Conda or venv environment, you can select the Python interpreter from that environment in VS Code (e.g., via the command palette: Python: Select Interpreter)³.

5. Key Takeaways

• setup_conda_env.sh and setup_python_env.sh only set environment variables; they do not create or activate environments.
• Use ./isaaclab.sh -p for most workflows, or create and activate a Conda/venv environment for more robust, reproducible development.
• For IDEs like VS Code, set the interpreter to the Isaac Sim python.sh script or your Conda/venv Python as appropriate.
• Avoid working in a "bare" shell without an activated environment for anything beyond simple, single-script workflows.

By following these practices, your development environment will be robust, reproducible, and ready for advanced workflows, including multi-terminal and ROS 2 integration²³¹.

Footnotes

1. https://docs.isaacsim.omniverse.nvidia.com/latest/python_scripting/manual_standalone_python.html ↩ ↩² ↩³ ↩⁴

2. https://isaac-sim.github.io/IsaacLab/main/source/setup/installation/binaries_installation.html ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

3. https://isaac-sim.github.io/IsaacLab/main/source/overview/developer-guide/vs_code.html ↩ ↩² ↩³