My journey through different languages.23 Mar 2018
I’ve been programming since close to six years now and over that time, I learned and tried to learn a lot of languages. This is a brief history of the languages which I stuck with over time.
At that time I was working heavily with the react+redux architecture and typescript prevented a whole class of bugs. Things like not having to step through deeply nested component trees to see where the wrong prop was passed down saved a huge amount of development time. It also invariably made me organize my projects better by putting some thought into the API’s exposed. I finally understood how amazing strongly typed dynamic languages were.
I wanted the same experience in the backend and I naturally turned to Go because it seemed vastly more approachable than Rust. I had been following the Rust project since before it hit 1.0. I read up, did some tutorials but never a non-trivial project, Plus the Rust web development story wasn’t the best at that time. So I started building web services in Go, but quickly hit a wall with tooling. There wasn’t an idiomatic way to manage packages and to my knowledge, the community is still split between different tools. Plus, I personally felt the way to implement interfaces was not good. There was no way to know if an interface was implemented for a struct just by looking at the method implementations. You have to know what methods are declared in the interface. After feeling like I was fighting the languages, I just wasn’t interested anymore. Sure Go has a really shallow learning curve, but it just isn’t that ergonomic a language.
I was still looking at Rust from the sidelines, but as a part of my grad school, I took an OS class which heavily used C and assembly. Here I really learned about managing memory and how awesome C is. It’s a tiny language which can be learned in the same or even lesser time as compared to Go. But it is far more flexible. Sure it has its many many warts, but to me, C will always hold a special place since it makes reasoning out what your code is doing so easy (Plus it was my first experience with systems programming). You look at a piece of code and you know what it does.
That’s when I really started looking at Rust seriously, since it is just as low level as C (for my use cases) plus, it has a ton of modern features. I’ve tried picking up C++ many times, but it was very difficult to pick up. The lack of a default build system and having a no package manager meant I had no idea how to get started on a non-trivial C++ project. Rust with cargo made getting started insanely simple and approachable to newcomers.
I am taking a distributed computing systems class and had to build a simulation of a distributed system using vector clocks and Rust seemed to be the perfect candidate for such a project. And even though I was new to Rust, I had a heavily multi-threaded project with a lot of inter-thread communication up and running within a few hours! It was an amazing feeling when I the first time I ran the program, it ran perfectly. Without a single issue. And this was with niceties like config files and logging to different backends. As far as writing my first multi-threaded program went, it couldn’t have been simpler.
The idea that if it compiles, it runs, is awesome and I plan on really diving into the deep end with Rust. I’m looking at the redox-os project lately to get myself acquainted with the codebase to start contributing small things to it. Here’s to many more years of building systems.