Comment by martin-t

As someone in a similar position (also ~3 years doing gamedev in Rust but only free time + open source), I feel very similarly.

There's low hanging fruit i've been complaining about for years where Rust is protecting us from ourselves - orphan rules, global state, ... Look, we're adults, we can make decisions with tredeoffs.

Compile times are a tougher one, I understand that Rust does analysis that is more complex than many langs and i feel ungrateful to people who spend their free time improving Rust. But also i don't think the complexity justifies all of it. Make dynamic linking easier, reduce how much needs to be recompiled, compile generics to dynamic dispatch in debug builds, etc. - there's gotta be a ton of stuff that can investigated.

ECS just plain sucks. People use it because what they want at first is some way to have relationships between entities. References/pointers are the obvious solution in most langs but in rust, they're obviously out. The second option is Vec and indices but that falls apart as soon as you start removing entities. The next step up the ladder of complexity should be generational arenas but for some reason people immediately reach for the big guns - ECS. And it infests their game with two things that make gamedev a slog - dynamic typing and boilerplate.

Boilerplate is obvious to anyone who has done gamedev the "obvious" way before. What could be projectile.shooter.score += 1 is multiple lines which (depending on your particular choice of ECS) usually involve generics. You shift your focus from tweaking your game logic and tuning the experience to typing out boiletplate.

Dynamic typing means entities are no longer structs with fields where you can understand how they relate to each other at a glance but instead any component can be anywhere, entities are no longer real, refactoring always causes silent bugs.

However, by far the biggest issue is the community's handling of criticism.

There are practically no experienced gamedevs coming to Rust from other langs so there's nobody to give Rustaceans a reality check. Rust gamedevs are almost always writing their first game (or, yes, engine). And there's nothing wrong with that, i was writing my first game at one point too. But their attitude is that they chose Rust because they heard it's the best and they got invested in the language because it's hard(er) to learn and now with all this investment if they hear rust or their particular favorite engine might not be that great, it feels like wasted effort so they get emotional and defensive.

I've personally chatted with over half a dozen other gamedevs who came to rust with years of experience under their belt and a common pattern is that they avoid the rust (gamedev) community because they're beat down by the negativity heaped upon them every time they try to discuss the negatives. It doesn't matter they take every effort to offer constructive criticism, it becomes a social instead of technical topic.

I came to Rust because i care about code correctness and, well, quality (of tooling, docs, testing). And Rust delivers there on a lot of that. But i also wanted to write games of a larger scale than can be done in one person. My hope was that there'd be other people with the same values who wanna build cool games together. Instead there's a low single digit number of serious open source projects and a bunch of small one man games and a a whole lot of loud people who seem to think gamedev is about hyping up an engine like it's a sports team.

Myself, I apparently chose the wrong engines for my games in both cases. Not because they're bad technically. In fact, having 5 years of gamedev experience before Rust, i think my choices are better from a technical perspective but there's just not the critical mass to build a serious open source game around them.