task-list

Prerequisite Software

• pnpm: https://pnpm.io/installation#on-posix-systems
• docker: https://www.docker.com/

Start (Development)

Run these:

• cd task-list-backend && pnpm i && cd ../task-list-frontend && pnpm i && cd ..
• docker compose up
• go to http://localhost:3001 on your browser

Start (Production build)

• Update the backend's .env to use the production string for DATABASE_URL
• Run the following from the root folder:
  • docker build -f task-list-backend/Dockerfile-prod ./task-list-backend -t task-list:latest
  • docker run --env-file=./task-list-backend/.env -p "3000:3000" task-list:latest
  • docker build -f task-list-frontend/Dockerfile-prod ./task-list-frontend -t task-list-frontend:latest
  • docker run --env-file=./task-list-frontend/.env -p "3001:3001" task-list-frontend:latest
• Go to your browser at localhost:3000

Notes:

• I've commited the .env file to simplify running the application locally

Commands

Backend

To sync and create migration (if required) after updating schema.prisma

docker exec -it task-list-backend-1 sh
db:migrate-dev

To reset DB and rerun all migrations (deletes all data)

docker exec -it task-list-backend-1 sh
db:reset

Design Decisions

• Uses Prisma as the ORM
  • Personal preference. It allows for type-safe queries, easy management of the DB schema via the schema.prisma file
  • Improvement: It (still) doesn't support polymorphic relations prisma/prisma#1644
• Uses the standard Service pattern
  • Improvement: The Service-Repository pattern would be preferred, where there would also be a repository layer to manage and isolate the data access logic for the data layer.
• Validation
  • Validation is done using the class-validator package which has a innate integration with NestJS via the @Body() decorator
  • Improvement: validation should also be done in the service layer, as typically a service method may be called by another service method, or via another source other than the REST endpoint, e.g. a Kafka consumer
• Data separation layer between controller and service
  • To separate the controller from the service, and provide standardized API into the service regardless of the caller - e.g. the (REST) controller, a kafka consumer or another service
  • If a repository layer is used, there would also be the same separation of layers (i.e. different types in and out of the layer)