This is embarrassing ...
I thought I would finish my app on time, I was wrong 😅.
In fact, 3 weeks ago, I stumbled across the hackathon announcement and decided to give it a go.
I collected some design ideas, shared my mocks on Twitter, and optimistically started developing...
Despite not finishing the app, I would like to share with you some learnings while you can find the code here.
But what to build?
After prolonged reflection, I settled on a goals tracker app. I know... there is a lot of them, so why another one?
I would say for 4 reasons :
I always feel behind when I use the current solutions: there is always something to do and I don't see my progress
I can't see the high picture with the current apps: I can track goals, I can track habits, but I don't see the link
I want an app that combines Pomodoro + goals tracker + actions/habits tracker
I want to own my data completely
The first step: Design
I love designing UIs. I am not that skilled, but I love it.
As a result, I spent ~20h on Adobe XD making the app mocks 🤦🏾♀️. I collected inspirations on Dribble, picked a color (red for energy), a font - "Karla", etc...
You can find some of the results below. Let me know what you think ;)
Second Step: Pick a stack
Along with UI design, I love learning new things. So for the challenge, I decided to go for the following tech stack:
- Typescript + Eslint
- Formik => for forms validations + creation
- Chakra => for React components
- Prisma => for handling databases
- NextAuth => for authentication
- and React-Select for input selects
Third Step: The development + challenges
I wouldn't want to use a slow web app. So, perf was a priority for me. Because of that, and early optimizations, I stumbled upon a series of issues :
1. Preact != React
My code would work locally but not in production. I tried many things but couldn't find the culprit.
2. Responsiveness and server-side rendering
Media queries don't work on the server. As a result, I was shipping some useless code for mobiles.
- Estimate if the user had a mobile by looking at the user agent in getServerSideProps
- Extensively use Next.js dynamic imports to only ship code when needed
3. Picking a database
I hesitated a lot between Firebase, Prisma, and Faunadb and listed the following pros and cons :
Pros: I already used it, it handles authorization well + It is free + I don't need a BE
Cons: The billing is not clear + I will learn nothing new 🙈
Pros: It looks really efficient + It handles authorization well + It has a free tier + I learn something new
Cons: The billing is not clear too + I couldn't use multi authentication in the free tier
Pros: It integrates well with Next.js + very friendly documentation + I can use it with NextAuth + I learn something new
Cons: I need to set up my own DB + I need to handle the authorization
I ended up choosing Prisma and I set up a Postgres instance with AWS :).
I would either :
- render too many components on the server,
- or use dynamic imports aggressively (which creates too many chunks)
- or not use dynamic imports enough and have big chunks sizes
Resolution This is still in progress. But, by using a combination of console coverage + Lighthouse, I was able to identify some culprits.
4. Encrypt user data (still searching)
At this point, I hesitate between using IndexDB (and store the data on the user desktop/phone) or use encryption on Node.js.
5. NextAuth not working in production Vercel (still looking for a fix😥)
What is next?
I am super excited about this app and I will be actively building it. If you are interested or have any idea, ping me on Twitter 💪🏾❤️
Thanks #Hashnode + #Vercel for organizing this great competition.
App => goalswin.vercel.app
Github => goalswin.vercel.app
Twitter => twitter.com/joyancefa