First of huge props to the students, this story is about. They managed to go from zero-web-development experience to a working Realtime-React app. If anything in this story sounds like it’s offending ―I i’m sorry, because actually I’m pretty proud that you guys managed to survive this hell with me. Starting from zero and being exposed to the uncertainty of a young startup the same time is definitely a bad starting point.Background
Technical University Darmstadt forces computer science students to do a non-payed practical project for a real company before they are technically allowed to write their bachelor thesis. As I’m currently working at a small startup in the need of developers and I had to do this internship myself two years ago I thought it would actually be fun to lead a small team of developers (4 students in their 5th semester) to do a little side project for us ― so we applied and got accepted.
I’ve basically been working in web-development since my first semester at university (student job to raise money for my living) and even before I had some courses at school (basics ofphp, html/css), so I somehow assumed that in the 5th Semester, people should know at least a bit about web-development(html/css).
After the first meeting with the group I realized I was wrong… and in retro perspective I should have known ―since I was at the same university:). TU-Darmstadt is proud to be very theoretical, so you can manage to get your B.Sc. basically without programming a single “ hello world ” (there are a few exceptions where you have to do programming in group work ― a bit of racket, java and bluespec/verilog <20/180cp ― but reading a good book will bring you through the paper exercises in the exam).
https://twitter.com/iamdevloper/status/748201714440900610 Getting started
So now we have a super nice, motivated group which had to work 250h per person. What to do in 1000h? Our plan was to get a running prototype of an authenticated real-time sensor-charting application with a public and private rest api.
We choose node.js + react as a stack beforehand, and gave all the other decision over to the team, as I thought it would be nice if they could choose something they are interested in and not requiring “ complicated ” stuff like global state management. They decided to go with create-react-app + firebase ― which i thought and still think is/was a good choice. Basically it provides anything we need(auth/soft-realtime via subscriptions/hosting/node-admin lib for building the custom rest api).The sadtruth
Choices: The variety of choices is overwhelming and if you have never worked with it before you end up with an awkward Frankenstein code with libraries which you think have “ enough ” stars on github.
Browser Madness: Imagine you’re new to web-development and suddenly you have to learn dom/vDom html css js… and then when you finally have enough knowledge to build something which works for you ― it most definitely isn’t working for you mac-safari hipster coding buddy and you have to start learning about browser support and polyfills.Lessons I havelearned
My takeaway : Everything is awesome ― just decide on your subset of awesomeness and stick to it.
For future projects, I’ll probably force the team to go with my subset of awesomeness , to make decisions easier and not overwhelming themby the variety of choices.
If I had created a base project including e.g.: moment.js, react-router@v3, mobx, eslint@airbnb, firebase-nest, mobx-firebase, jest… a lot of problems probably would never have happened.
Also I’d probably force a specific git-flow and nag the team if it doesn’t stick to it.
Another, kind of annoying, thing I noticed is, that people tend to google for results even if there are awesome docs/apis and end up with awkward/outdated implementations from SO or private blogs/forums. Next time I promised myself to ask “why” every time a notice such a thing, so that the students also learn to properly read API docs and end up being better developers.