Its always good to make decision based on data.
I’m interested in Rust as I predict it will overtake C++ in popularity within the next 1-2 years if the trends in the above graph from Stackoverflow Trends continues. Rust is growing in popularity exponentially and C++ seems to have a linear decline.
The reasons for this, I think, are many.
Rust has a built in package manager which also handles compilation and makes the developer experience easy. Easy always trumps difficult, every time2. This is a huge boon to productivity and also a huge breath of fresh air after C++. It took me about 6 weeks to get started with C++ (finding a package manager amongst many, trying to understand CMake). I think what is happening is young1, C++ developers are trying Rust and then being won over by this after struggling through C++.
Current problems with Rust¶
No native linear algebra libraries or Scientific/High-Performance Computing (HPC). It does have bindings for OpenBLAS a C++ lib but its apparently a nightmare to get working cross platform. This holds it back from deep learning and general data science which both rely heavily on matrix operations.
But its still early days and I think these issues will be address in time.
Here are some materials I’ve been using:
- The Rust Lang Book video series is quite handy to get up to speed on Rust quickly. Its basically going over the official The Rust Programming Language Book which takes you through all the features of Rust in a systematic way. I find someone going over the material (at double speed) seems to be easier than reading the material directly. But its not enough by itself. Need to apply what I’ve learning quickly or else it doesn’t sink in.
- There is also an interactive quiz based leaning aid that I just discovered while writing this. I have never been into puzzels that much but I know if you don’t try to apply what you have just learnt
- Learn Bevy Engine 0.10 Beginner Tutorial Series Bevy is an ECS game system. ECS are all the range in game development because they are fast because they take advantage of CPU caching. This tutorial series takes you from zero to build a basic interactive game and give you a template for your own project. Although I’m not familiar will all of Rust’s syntax I find learning by doing the best strategy. I’ve just finished the first 10 episodes which is all there is at this time. Time to make my first game with this template.
Some projects I’m thinking of starting: - An Educational Game - Asher is learning to read at the moment and I think a game to help him match letters to sounds might help. Repetition is the key. The idea is some and I’m going to build it off the back of the Bevy tutorial I’ve been going through. - Chess engine - or now I’m thinking, rather a general purpose discreet, game engine. Create an interface to some set of rules which generate moves for a given position (and also indicate a winning move/position) it should be able to learn via self-play. I guess replicating AlphaZero would be the first step? - Cave Boy, a kids, open-world, adventure game. I’ve started prototyping in Python. I’m thinking something like Don’t Starve (but less scary), or Stardew valley (with less detail) with a Captain Underpants sense of humour (i.e. potty). I’m thinking to aim it for 4-8 year olds, full touch screen and very simple mechanics. Everything is visually displayed at all times (no hidden menus or state i.e. inventory). TODO: break this out into a full blog post (i will as I get more into it). I have some concept art already! - Solid Pod Server - all in one identity/public+private content server with built in authorization (OAuth+). The hard part of Solid, because it was very boring, was trying to understand the standards which are verbose and incomplete. It also uses OAuth + OpenID Connect at the core of its authorization protocol which in itself was the hardest part, because its complicated and I found it hard to find a straight forward explanation.