/my-go-app

Primary LanguageGo

my-go-app

Test (Local)

make run-test

Run (Docker)

docker-compose build docker-compose up API Document: http://localhost/swagger/index.html

Deployment (Read-only)

  • Deployment on Heroku
    • Commit and push codes on https://github.com/tonyltf/my-go-app
    • Login Heroku Dashboard, create/select a project
    • In Deploy section, select GitHub and connect to the repo
    • Run Deploy Branch under Manual deploy

API Document: https://my-go-app.herokuapp.com/swagger/index.html

Background

Stack

  • Go-Chi is the suggested framework
  • SQLite was planned to use originally because it is easy to setup for local hosting. But then deployed on Heroku and found it is not supported
  • Postgres is the selected database because I want to enable transforming different source into the same schema so it will be easier for querying the price / average price
  • Didn't plan to use NoSQL because the data insertion looks not very frequently

Enhancement

Business

  • Better handling of exchange rate not found, I think it is kind of business/product requirement because of a few concerns
    • data in the current database does not reflect the complete history of exchange rate
    • if there is another third party service supports complete rate history by specifying the backdate timestamp, those API maybe costly so I would suggest adding a fallback mechanism to lookup and store when needed
  • API to get price with timestammp - if a future time is provided, we only shows the last price. Should have better handling for this

System

  • App structure - I am new to it so didn't organized the folder well
  • Services - Split the cron job and web server into 2 different services
  • Configure file - Better usage , i.e. for different environment like .env in JavaScript projects
  • Monitoring - if the currenct third party API is down or any error happens, i.e. rate-limit reached
  • Error handling - Should have centralized error logging / handling so we can do something like alert / notification
  • Logging - better loggin mechanism like fluentd
  • Data pipeline - more decoupled data pipeline like ELT instead of directly dumping into the database
  • Some testing is missing - factory pattern is not tested