GREAT ⭑ BEAR

Food delivery sample application metarepo (PostgreSQL / GraphQL / React).

Repositories

• React + Apollo Frontend: https://github.com/sastraxi/great-bear-frontend
• Hasura GraphQL Engine backend: https://github.com/sastraxi/great-bear-hasura
• PostGraphile backend: https://github.com/sastraxi/great-bear-postgraphile

The docs in each repository will guide you through setup.

Tools used

• VSCode + Plugins
• GraphQL Playground
• pgsh

---

Work-in-progress presentation

• https://docs.google.com/presentation/d/11cpgG0O2LA16ewefyFsgJNcP6aQKR8K9pOURGPAU36s/edit?usp=sharing

Outline

• attempt at a fair comparison between Hasura and PostGraphile
  • why these 2 technologies (why graphql)?
  • why is a food delivery website the best app to build?
    • event triggers ("microservices")
    • subscriptions ("push notifications")
  • methodology?
    • stay as close to technology-specific best practices
• what are you measuring?
  • nloc (sql, ts)
• keep best practices wrt security, frontend tech
  • cookies
  • react hooks (not everything yet)
  • styled-components - keeps css near the js
• postgresql
  • denormalized database - don't prematurely optimise
  • camel_case vs snakeCase
  • auditability
  • security, speed, etc. etc. its a good database mmmmkay?
• aim for the absolute bare minimum of backend boilerplate
• event system
  • hasura: built-in, console helps hugely with debugging
  • postgraphile: pg-pubsub, capture + subscribe to events ourselves
• developer experience
  • hasura: docker not shutting down cleanly, no js extensibility
  • postgraphile: lots of setup, lots of power. well thought-out APIs.
• differences in approach
  • hasura: external "appliance", column security + public schema
  • postgraphile: external or library, RLS, different schemas (no col. security)
• takeaways
  • order of work: hasura, then postgraphile

Navigation idea

• first section is the "article", second section is technical detail
• after first section, link to next repo ("next page in the article")

What's left to do

• build a test harness to run performance / integration tests
• we can use https://www.postgresql.org/docs/11/pgstatstatements.html to get queries / stats
• compare some generated SQL in the presentation