I’ve started building a wordle-analyzer. I got nerdsniped, and now I’m implementing the game (already did a game and a cli implementation using the abstracted interface) and a solver.
The idea is to provide:
game - a wordle game that can be used to implement the game in many frontends, a cli version and exists already
solver - something that can solve the game
wordlist - any wordlist people might want.
bench - if you want to know how good your solver is
I’m providing built-in versions, but anyone could implement the traits.
I currently have two solver implementations;
stupid - literally tries words at random
naive - checks which letters matched and reuse them, require letters that are included but not matched in other words, then use the most common word that matches
The naive solver can actually solve the game in less then 10 steps most of the time. Mathematically, the optimum is about 3,4 steps. There are two amazing 3blue1brown videos going into details, and my eventual goal is implementing solvers making use of that math.
I’ve been using generics and traits like never before for that project. Solver? It’s a trait. Game? It’s a trait. Word lists? It’s a trait.
And all my structs have generics <'wl, WL> so that I only need to have the word list once to save resources. You get a little crazy from the lifetime errors but it’s fun.
Besides that, my homeserver got janky this month, and today I started migrating it to proxmox. It’s hard, because I need to do a lot more thinking than just docker go brr now.
Not very stable at all, but the cli game (wordlec) is playable with the responses of whether you hit a letter or not, the naive solver and the stupid solver work too.
However, I expect that the API will change a lot before v0.1. I will release it when I deem it somewhat stable. Happy to hear you like it.
I’ve started building a wordle-analyzer. I got nerdsniped, and now I’m implementing the game (already did a game and a cli implementation using the abstracted interface) and a solver.
The idea is to provide:
I’m providing built-in versions, but anyone could implement the traits.
I currently have two solver implementations;
The naive solver can actually solve the game in less then 10 steps most of the time. Mathematically, the optimum is about 3,4 steps. There are two amazing 3blue1brown videos going into details, and my eventual goal is implementing solvers making use of that math.
I’ve been using generics and traits like never before for that project. Solver? It’s a trait. Game? It’s a trait. Word lists? It’s a trait.
And all my structs have generics
<'wl, WL>so that I only need to have the word list once to save resources. You get a little crazy from the lifetime errors but it’s fun.Besides that, my homeserver got janky this month, and today I started migrating it to proxmox. It’s hard, because I need to do a lot more thinking than just docker go brr now.
In case you want to check my wordle-analyzer out: https://git.cscherr.de/PlexSheep/wordle-analyzer I need to update the readme before publishing.
That sounds like fun! Wow. How stable is it at the moment?
Not very stable at all, but the cli game (wordlec) is playable with the responses of whether you hit a letter or not, the naive solver and the stupid solver work too.
However, I expect that the API will change a lot before v0.1. I will release it when I deem it somewhat stable. Happy to hear you like it.