ruggeri/koa-starter-app

Scripts

• eslint: watches src and runs eslint.
• serve: uses nodemon to watch dist/backend and run dist/backend/app.js.
• start:dev: uses concurrently to run ts:watch and serve.
• ts:watch, ts:watch:backend, ts:watch:frontend: watches all/backend/frontend directories and recompiles typescript code.

Packages

• koa
  • koa-bodyparser and koa-router.
  • Use @hapi/joi for validations.
• next
  • All next content lives in src/frontend, it is served via next which is served via koa.
  • I compile the frontend typescript => javascript with esnext style imports. Then next's server does the rest of the transpilation.
• typeorm
  • Needs reflect-metadata.
  • Also using their migrations support.

Dev Packages

• Yarn stuff:
  • concurrently for running simultaneous jobs.
  • nodemon for restarting servers.
  • wait-on for waiting for something to get built.
• eslint
  • I have a number of plugins (especially around typescript).
  • I use prettier for formatting.
• typescript
  • I also have ts-node, which could be handy sometime.

TODO

• I want to look into GraphQL and Relay.
  • https://github.com/graphql/graphql-js
    • You make the schema, and you list the resolvers.
  • https://github.com/graphql/express-graphql
    • Pretty sure this just creates an endpoint to receive graphql queries.
    • But you must have already made the schema.
  • https://github.com/apollographql/apollo-server
    • https://www.apollographql.com/docs/
    • https://www.apollographql.com/docs/apollo-server/
    • https://www.apollographql.com/docs/tutorial/introduction
  • Prisma?
  • https://typegraphql.ml/
    • Integrates with TypeORM.
    • Basically: it looks like one has to do a lot of work writing resolvers.
    • But is it easier? I mean, you still have to write the resolver code, right?
  • https://github.com/graphql/dataloader
  • https://graphql-code-generator.com/
    • Here is another approach.
• Randoms
  • Testing? Jest?
  • Ramda? (I haven't needed that yet.)
  • Rxjs?
  • Authorization.