Autopack

Directory structure

• pyproject.toml: project metadata and abstract dependencies used to build the project and install development environment.
• conda-lock.yml: locked dependencies for conda, generated from pyproject.toml. Enables reproducible builds and development environments.
• tasks.py: automation tasks for building, testing, and releasing the project. See invoke for more information.

Development activities

Want to make a change? Terrific!

All steps involving terminal commands below assume that you are in the root directory of the project, i.e. that you have run cd <somewhere-on-your-computer>/autopack.

Getting started (setting up the development environment)

The project uses Conda to manage the development environment and its dependencies. Additionally, the conda-lock and invoke packages need to be installed in the base environment. Luckily, there's a script that will do all of this for you, and you should only need to do it once. Just follow the steps below:

1. Run bin/setup-global-environment.bat to setup your global Conda/Python environment.
2. In a new terminal (Powershell or whatever you prefer), cd to the root directory for this project and run invoke env to create the Conda development environment. This will create a new Conda environment called autopack and install all dependencies.
3. Code away, and see below for instructions on more specific tasks.

All instructions below assume you have an installed development environment.

Running the tests

1. Run invoke test.

or...

1. Activate the autopack environment by running conda activate autopack.
2. Run pytest with your preferred options.

Starting the application (for development)

1. Run invoke app.

Building the application (for distribution)

This will be done automatically by the CI/CD pipeline, but if you want to do it locally, follow these steps:

1. Run invoke build and wait. The zipped build will be placed in the dist directory.

Adding another package/dependency to the project

1. Activate the autopack environment by running conda activate autopack.
2. Doodle around with conda install <package> (or pip install <package> if needed) until you get what you want.
3. Add the package to the pyproject.toml file under the [project.dependencies] or [project.optional-dependencies] section. See the file for when to use which section.
4. Run invoke lock to update the conda-lock.yml file with the new dependency.
   • If you get an error from conda-lock, try removing the conda-lock.yml file and running invoke lock again.
5. Commit the changes to pyproject.toml and conda-lock.yml to source control. This will make sure that everyone on the project can install the exact same dependencies, and give us traceability for when things break.