When I say "Zero Knowledge," I'm not talking about ZK Proofs or anything like that, I'm talking about my experience level with ANY of the tech used at the start of this project. 😂
This is the second post in a series on how I implemented Super Bowl Squares on Solana.
In this post, I give a bit of context on the project and discuss the technology decisions I made.
Context
Every year, when the Super Bowl rolls around, my friends and I play a game called Super Bowl Squares, which I described in the previous post in this series.
This year, right after the AFC and NFC championship games ended, exactly two weeks before the Super Bowl, I had an idea. I wanted to put the Super Bowl wager game called Squares on a blockchain. I floated the idea in the Discord. When I woke up to some encouragement that Monday morning, I made my mind up to actually do it.
Setting aside the insane 13 day deadline I just signed myself up for, there were a few obstacles I faced -- like, for one, I had absolutely zero experience with any of the tech necessary to pull this off.
While I am tutorial-familiar with Solidity and web3 in general, I had never written and deployed a smart contract or dApp from start to finish.
I knew this wouldn't work on Ethereum due to high gas fees. I am also not a huge fan of the L2/L3 approach to scalability (which I'll not get into) so even my very small amount of Solidity familiarity isn't going to help us here.
I'd need to pick another blockchain to deploy on.
I narrowed my choices to Solana and Sui, two of the fastest, cheapest, and most popular L1s. But they both have their pros and cons. For example:
Solana is immensely popular right now; Sui not as much.
Yet Solana's smart contracts are written in daunting Rust, which looks like z̶̪͉̅͗a̷͚͛l̸͙̝̀g̷͈͆̉o̶̤͉͊ ̶͉̾̔t̵͉͛ḛ̶̢̓͝x̶͚͚́͠ť̷̑͜ to me; Sui uses the beautiful Move language.
Luckily, while I was doing research, I found this article by @kklas_.
Medium
It was a godsend for me in the moment.
While the article clearly puts Move on a pedestal over Rust/Anchor, it did an amazing job of explaining the programming model and limitations of Rust/Anchor-based smart contract development. I was thinking, "damn, if only Solana supported Move!" 1
With my new-found baseline understanding of Solana's account-based programming model, I bit the bullet. Everyone I know has Phantom installed. No one I know has a Sui wallet installed.
Initial approaches
After making my mind up to go with Solana, I figured I'd get the contract deployed and tested first, then try to figure out the front-end situation.
While looking around at resources on the net, I noticed a project called Seahorse. Seahorse is a compiler that lets you write Solana programs in Python.
I was really trying to avoid writing any Rust.
Look, I'm a Ruby fan boy. I've been in love with Ruby's expressiveness and readability for almost two decades now. It's the most beautiful language ever created and second place is not even close.
So my first thought was to use Seahorse.
Python in general is very, very similar to Ruby, it's just not quite as good as Ruby.
I thought that since I'm familiar with Python, writing the contract in Python would be way easier than trying to learn both Rust and Solana.
However, I quickly ran into roadblocks and lots of questions. The docs aren't as fleshed out, there are not many code examples, and my inexperienced eyes couldn't distinguish whether I had Seahorse or Anchor/Solana problems.
I also poked around with Solang and experienced similar issues.
I decided to go with Rust and started the Hello, Anchor tutorial on Solana Playground.
Solana Playground
Solana Playground is really neat. I didn't have to worry about setting up my local environment until I was more comfortable with what I was doing.
Every time you build - deploy - update IDL, your Typescript types are updated and automatically imported in your client and test files.
There's even a graphical method invocation tab for testing your program's instructions without writing any code.
I highly recommend using the Playground to familiarize yourself with the development process if you're just getting started with Solana development.
The Playground also has built in tutorials. The code is loaded directly in the IDE and you can build and deploy as you would your own project.
I did some tutorials on Solana Playground and learned about program-derived addresses (PDAs). Once I started to grasp the concept of PDAs, I was able to start mentally building a data model and visualize the state transitions.
Rust, Anchor, and what front-end?
After I had written a rough draft of a contract with a workable data model, I decided to start researching how I was going to build the front-end.
I went to GitHub and tried to find some example Solana projects with a frontend. Pretty much every single app that I could find was built with React.
Guess what? I also had zero React experience, but without any other option2, I didn't have much of a choice here.
I found this scaffold project called create-solana-dapp and generated a Next.js project styled with Tailwind and DaisyUI components.
I have some strong opinions about Next.js and React. I really, really wanted to abandon the front-end and but I decided to stick with it.
Experience with chosen stack
If you're keeping track at home, I had 13 days to build an app with:
- Zero smart contracts ever deployed
- Zero understanding of Solana's programming model
- Zero experience with Rust
- Zero experience with Anchor
- Minimal experience with Typescript
- Zero experience with React
- Zero experience with Next.js
- Zero experience with DaisyUI
- Minimal Tailwind experience
ChatGPT as a learning tool
Before I dive into implementation details in the next post, I just wanted to say this absolutely would not have been possible without ChatGPT. I'd still be on page 52 of The Book. ChatGPT as a learning tool is still severely underrated and undervalued and I will think that until every single person on Earth agrees with me.
Does it make mistakes? Yes. Does it provide less-than-optimal solutions sometimes? Yes. Does it sometimes lead you down a trail of inexplicable red herrings? Again, yes.
BUT SO DO HUMANS.
Think about it, when you are pairing with a coworker, that person does not know every single thing there is to know without ever making a mistake. They could possibly lead you astray and you know it. So, what do you do? You stay vigilant and keep an open mind for mistakes.
It's a learning journey together.
Treat ChatGPT as you would your co-worker and good things will happen, I promise.
Conclusion
The next post in this series will be a technical post about the implementation that I came up with.
I honestly have no idea if it was a good approach or bad approach. I'm looking forward to more folks getting some eyes on the code and hopefully hearing some feedback!
-
I'm not really sure what happened with the project, but there [exists an announcement] from like 4 years ago (https://medium.com/solana-labs/solana-now-supports-libras-move-vm-f5fabaf3033c) from Toly that says Solana now supports Move. However, the GitHub project has since been archived and I couldn't find any code examples. ↩
-
I later found the Anchor/Solana/Svelte (ASS) stack. Also had no experience with Svelte, but Lord, it looks so much better. I will probably try Svelte next time. ↩
Top comments (0)