2.0 KiB
This is a Django-based web application that has an API, but primarily functions with traditional Django views with HTML templates to display data that mostly constitutes "scrobbled" items. The app started as a way to track a user's watched videos via a Jellyfin server, but has since grown to keep track of a number of media types: music tracks, tasks, videos, web pages, food, life events, sports events, podcasts, video games, board games, beers, brick (lego) sets, puzzles, books and geolocations.
The project is written in Python and prefers to use "fat" models where logical methods are contained in either instance methods on instatiated data models, or classmethods on the Django model class itself. When logic grows too complex, helper functions should be pulled out into utils.py files and the model instance ro class method should call the utility function.
Be sure to check pyproject.toml for project defaults. Specifically for black and isort expectations.
Imports in python files should always be top level if possible.
All tasks live in the PROJECT.org file and include an org ID that is a uuid to make them unique.
In local development, environment variables for various sensitive values live in a .envrc file
The .envrc file can be loaded into a shell environment to allow access to most third party services
Care should be taken when using .envrc that we do not spam services we use in production with requests
USING SESH COMMIT
This repository should use the sesh command to capture agentic session notes in git notes. You can
read the help next by running sesh by itself, but make sure that we use capture any reasoning
or prompts used for the work.
When asking to "commit" we should always assume we're using sesh commit and capturing session notes.
Release commits ([release] Bump to version X.Y) are handled by the user. Never auto-commit release steps; only prepare the changes (e.g. the PROJECT.org DONE/version entry and pyproject.toml bump) and leave them for the user to commit.