Skip to content

Latest commit

 

History

History
39 lines (27 loc) · 2.03 KB

ARCHITECTURE.md

File metadata and controls

39 lines (27 loc) · 2.03 KB
Raw

Architecture

This document explains the high-level structure of the project as well as the reasoning behind some architectural decisions.

No backend

First, it is important to point out that this app has no backend. The entire thing builds to a completely static web bundle. This is made possible by utilizing SQLite and sql.js. The raw data is bundled into a single SQLite database file (schema) which is pushed to the browser. This means that all data queries are done completely in the browser!

This approach has a number of advantages:

  1. Hosting is free! We can use GitHub Pages to host the app completely for free.
  2. Simplifies hosting and deployment. Deploying a static website is very easy.
  3. Fewer moving parts. There's no need to worry about data conversion between the frontend and backend, for example.
  4. Speed. sql.js is very fast, making searches nearly instant.

...and disadvantages:

  1. Much larger bundle size. We gain about 2.5mb by bundling sql.js and a grades database with about 5 years of data. This puts the total bundle size at aorund 5mb (before compression). This is not insignificant and we have to be concious about adding more data.
  2. Inherently limits the scope of the project since we have to be careful abot how much data we add.

Monorepo

The project is organized as a monorepo containing the client and db subprojects. This separation is not strictly necessary, but keeping all database-related code in one place helps when making database-related changes.

We make use of NPM workspaces to simplify development in the monorepo. Take a look at the scripts in the root package.json to get an idea of how that works.

  • db is packaged as @utd-grades/db
  • client is packaged as @utd-grades/client and depends on @utd-grades/db

TODO: explain more here, walk through the flow of a search, for example