Happy Bevybirthday
Posted 2024-08-13 05:19:42 ‐ 5 min read
Happy Bevybirthday!
This post is an answer to bevy's birthday post, now that I am an active part of the ecosystem, I can't miss out!
A bit about me
I’ve been involved in mobile development for ~10 years, most of the time in the gaming industry, and since discovering Rust in 2015 and bevy 4 years ago, I’m convinced this enbody the future of game development I want to be part of.
Now, yes, bevy is not mature yet, I’m maintaining a curated list of bevy projects in production, and while I’m very excited to see it grow, and wish them success with all my heart, the « execution level » is (mostly) not yet comparable to the existing giants of the gamedev ecosystem.
But that’s to be expected: most projects are made with limited resources, Rust is young in gamedev, and bevy is still a "baby".
To the future!
Now why do I believe this "baby" bevy is the future ?
I can’t spare you the rust-y « reliability and performance » reasons, but I’ll try to give more under-appreciated reasons to believe in Rust and bevy:
- Rust’s « open source by default » mentality helps with sharing resources to make the whole ecosystem better, which bevy embraces with its modularity.
- Rust’s ability to be tested or spun off into smaller projects helps with consolidating parts of any projects, which is a godsend when investigating a bug, or teach a feature.
Why do I care about bugs and teaching ?
More about me...
I’m glad you asked! I believe so much in Rust (and bevy) that I did quit my unity gamedev job earlier this year to hop on the bevy train.
Special thanks to the fellow tech enthusiasts who listened to my few talks about bevy ecosystem, you empowered me.
And what a turn of events! I’m now a freelance open source developer, with my first 1 year mission to maintain Rapier and its surrounding ecosystem.
To the community
So of course, I’m excited about physics engines, and seeing avian make such a fulgurant entry is both humbling and a great sign for production projects. Congrats Jondolf !
Physics engines is not the only area where high quality libraries are growing: UI (bevy birthday post has a great list), tooling, rendering…
Now, these are mostly « tech-oriented » projects, and my wish for this next bevy year is to get more non-tech projects: Let bevy reach the end user!
To the team
🌹: Bevy is genuinely delightful to have multiple programmers
🌹: But for artists... eh not so much
It’s a bit too early for most non-tech to feel welcome into bevy, and not only artists: every question asked is at risk of being answered by « it depends »:
- How do I contribute assets ?
- How to collaborate on a bevy project ?
- Which platform can we target ?
- How can I get funded ?
- Do I need to know Rust ?
It’s a proof of solid foundation to be able to provide multiple answers to a given question, but it’s clearly overwhelming, and unfit to quick assesment of the engine.
Where you can build a game « naïvely » in Unity or Godot, without knowing exactly where you’re going, it’s not advisable to do that in bevy if you’re not a veteran (or at all).
So my wish for next bevy year, is to improve the « naïve » proofness of bevy:
- reduce surprises: add tests, documentation, listen to and act on user experiences
- ensuring tools work and help each other: bevy mod system dump is one of my personal favorite, that I wish all projects would include. Mockersf initiative of screenshot comparison (Pixel Eagle) is another great example of tooling which most projects would benefit from.
- Crates compatibility: avian collider creation ; nalgebra compatiblity with glam...
- Everything developer experience, which must now trickle down (up ?) to project-team experience, in order to make bevy take its flight.
- Let's not settle only for Developer eXperience (DX), let's popularize the term Team eXperience (TX),
What now?
I've already started this goal of improving the ecosystem, but I have to admit the most difficult part is identifying the lowest possible hanging fruit where I'll be able to help, without crunching.
I'm proud of having led the effort for a CI test for conflicting systems ordering ambiguities, and I'd love more visual tooling around bevy development. I'd love to take a closer look at the source location in change detection info from aevyrie, and Blenvy from kaosat.