(http://nowplaying13.herokuapp.com)
Me and a few of my friends use WhatsApp almost exclusively for sharing Spotify links. I decided a replacement for that would be fun to build: a messaging site where all you can do is share the song you're listening to, and comment on other people's shared songs. Surprisingly, Spotify's own app doesn't have that capability.
- "As a user I need to be able to log in with my Spotify account."
- "As a user I want to see songs that others have shared recently."
- "As a user I want to be able to always see the song I'm currently listening to on Spotify."
- "As a user I want a simple Share button to post my current song for other users to see."
- Trello board (https://github.com/ryanmonro/nowplaying/blob/master/docs/trello.jpeg)
- Wireframes (https://github.com/ryanmonro/nowplaying/blob/master/docs/wireframes.JPG)
- ER Diagram (https://github.com/ryanmonro/nowplaying/blob/master/docs/er.JPG)
In a way, I started building this app four times. After deciding it was what I wanted to build, I first began building a Rails app with a React front-end via react-rails front-end. After realising react-rails's implementation of React was not what I'd hoped for, I reverted to just building a Rails app. After getting to the working-database-and-routes stage I decided "I made a full stack JS app" was something I really wanted to achieve by the end of the week, and built the backend in Express with a rudimentary front end that accessed the API via AJAX. Then I threw that front end out the window and built a React front-end. The React development happened inside the client
folder, and Express served the routes in the following manner:
/ (statically serving /client/build)
/api/* (handled by express)
/login (handled by express
This allowed me to use the React development server in development, and then build it to the build folder so it could be served by my Express server.
- It took a while to understand how it worked and why it didn't work when it didn't, but in the end I loved how React uses state, and reusable stateful components really simplified the app's structure.
- I really wanted to build something with this since I started the course, so this was immensely satisfying. The plethora of NPM packages a few keystrokes away came in handy too.
- The Spotify API was something I wanted to incorporate, as it's an app that I use on a daily basis. Once the authentication was sorted out, it was very easy to fetch user and song data from my front-end.
- React in the front, Express in the back, on one server! That was the dream. I wanted to avoid running two servers just because I knew it was possible.
- Databases and user authentication in Node. That stuff is easy in Rails. It was completely untested waters for me in Node.
- I had the bright idea of using sqlite3 as a database because 'it seemed simple'. Unfortunately, you can't deploy it to Heroku.
- Chrome's caching letting React get away with handling the routes I wanted Express to handle.
- Getting used to the jQuery DOM manipulation and then having to forget about all that and do things the React way.
- Authorisation with the Spotify API. It was tricky enough trying to understand that whole complicated secret handshake, let alone implement it.
- Troubleshooting React without knowing exactly how it works. They were hard issues to google because my situation was so specific.
- If you try to put more than five Spotify players on a page, they won't all work. Each one has to maintain a TCP connection to the Spotify API, and it turns out HTTP actually has a limit of the number of open connections a page can have. That was the end of my dream of having a feed full of Spotify players.
- Waiting for things to happen asynchronously, and making sure they still happened in the right way.
- Testing. I should have gotten that involved earlier. It's called test driven development, not test-as-an-afterthought driven development.
- I really wanted to build something in Node and React. In hindsight, it would have been better to choose the right tool for the job.
- This was possibly too much new tech to take on in go. In the end I built an app that doesn't do much, that I had to learn a lot of new things to build. There would be a lot more features if I'd stuck with Rails.
- On the other hand, half-building the app in Rails first, and then with jQuery and Ajax second, and then finally in React, really taught me the differences between working in all those technologies.
- let users comment on posts. It was a big part of my idea from the start, and the database supports it, but I didn't have time to make it work
- let users follow other users. The default view could be a feed of 'shared tracks from those who I'm following', with another view for 'last ten tracks shared by all users'
- User view: see their profile, recently shared tracks
- allow sharing from sources other than Spotify
- explore artists/albums/genres via the Spotify API - there's a lot of data out there