Development

IDE

Visual Studio Code is recommended as the IDE for developing the package. You can install the recommended extensions by opening the workspace in the IDE. Useful commands are also provided as tasks in the .vscode/tasks.json file.

Though, all IDEs that support Python development can be used, and those useful commands can be copied from the .vscode/tasks.json file to the IDE of your choice.

Poetry

Poetry is used for managing the package dependencies, virtual environment, building and publishing the package. It is recommended to install Poetry globally on your system, instead of in the same virtual environment that installs the package. The recommended version is 1.8.2, though newer versions should work.

Using your own choice of virtual environment manager is also possible, but the steps to install for GPU support might be different.

Development Environment

To set up the development environment, follow these steps:

Install Python (3.10 is recommended, but 3.10 - 3.11 should work)
Install Poetry
Clone the repository and checkout a non-main branch
Select the Python version to use with Poetry. For example, to use Python 3.10:
```
poetry env use path/to/python3.10
```
Install the package in editable mode
```
poetry install
```
Select the Python interpreter in the IDE to the virtual environment created by Poetry
Upgrade the dependencies if you want to develop with NVIDIA GPU

Activate the virtual environment created by Poetry if it is not activated:
```
poetry shell
```
Then, run the following command:
```
pip install -U torch==2.1.1 torchaudio==2.1.1 --index-url https://download.pytorch.org/whl/cu121
```
For CUDA version 11.*, you can change the cu121 to cu118. So the command will be:
```
pip install -U torch==2.1.1 torchaudio==2.1.1 --index-url https://download.pytorch.org/whl/cu118
```
Good to go!

Testing

To run the tests, run the following command:

poetry run pytest

If the poetry environment is activated, you can just run pytest straight away:

pytest