Piston Developers 🍿🥤🐸
@pistondeveloper.bsky.social
A modular game engine written in Rust https://piston.rs
Research branch: https://advancedresearch.github.io
Discord (Piston): https://discord.gg/TkDnS9x
Discord (AdvancedResearch): https://discord.gg/JkrhJJRBR2
Research branch: https://advancedresearch.github.io
Discord (Piston): https://discord.gg/TkDnS9x
Discord (AdvancedResearch): https://discord.gg/JkrhJJRBR2
Pinned
#StandWithUkraine 🇺🇦 (since Feb 27, 2022)
Sometimes I talk to people that tell me they tried to understand Path Semantics over the years, coming back to it regularly, but still struggling to "get it".
I tell them that the foundation of Path Semantics is Logic at the hardest level. There is no harder level above this level.
I tell them that the foundation of Path Semantics is Logic at the hardest level. There is no harder level above this level.
December 11, 2025 at 6:36 AM
Sometimes I talk to people that tell me they tried to understand Path Semantics over the years, coming back to it regularly, but still struggling to "get it".
I tell them that the foundation of Path Semantics is Logic at the hardest level. There is no harder level above this level.
I tell them that the foundation of Path Semantics is Logic at the hardest level. There is no harder level above this level.
Frame graphs are very similar to command buffers, except that frame graphs can abstract. They can refer to a command list and inject it as if it is was encoded sequentially.
A frame graph is a graph, precisely because each command list can be thought of as a node. Edges are references.
A frame graph is a graph, precisely because each command list can be thought of as a node. Edges are references.
December 11, 2025 at 1:14 AM
Frame graphs are very similar to command buffers, except that frame graphs can abstract. They can refer to a command list and inject it as if it is was encoded sequentially.
A frame graph is a graph, precisely because each command list can be thought of as a node. Edges are references.
A frame graph is a graph, precisely because each command list can be thought of as a node. Edges are references.
Yesterday I watched this movie: www.imdb.com/title/tt2956...
It's a movie about Nazis that are trialed and executed. Also about some of the people that made it happen.
My father likes these kind of movies: Reality based stories. He doesn't like fantasy.
It's a movie about Nazis that are trialed and executed. Also about some of the people that made it happen.
My father likes these kind of movies: Reality based stories. He doesn't like fantasy.
Nuremberg (2025) ⭐ 7.6 | Biography, Drama, History
2h 28m | PG-13
www.imdb.com
December 9, 2025 at 5:22 PM
Yesterday I watched this movie: www.imdb.com/title/tt2956...
It's a movie about Nazis that are trialed and executed. Also about some of the people that made it happen.
My father likes these kind of movies: Reality based stories. He doesn't like fantasy.
It's a movie about Nazis that are trialed and executed. Also about some of the people that made it happen.
My father likes these kind of movies: Reality based stories. He doesn't like fantasy.
Reposted by Piston Developers 🍿🥤🐸
I agree. The Senate needs a badass like Crockett. Period.
December 9, 2025 at 5:09 PM
I agree. The Senate needs a badass like Crockett. Period.
Reposted by Piston Developers 🍿🥤🐸
So. Rep Crockett announced her TX campaign for US Senate.
Good.
Voters get a choice in the primary. They can research the candidates, find their positions on issues and choose.
She's getting ultra-gerrymandered out of her Rep seat, so why not take the shot.
It's a primary. Not a coliseum fight.
Good.
Voters get a choice in the primary. They can research the candidates, find their positions on issues and choose.
She's getting ultra-gerrymandered out of her Rep seat, so why not take the shot.
It's a primary. Not a coliseum fight.
December 9, 2025 at 4:48 PM
So. Rep Crockett announced her TX campaign for US Senate.
Good.
Voters get a choice in the primary. They can research the candidates, find their positions on issues and choose.
She's getting ultra-gerrymandered out of her Rep seat, so why not take the shot.
It's a primary. Not a coliseum fight.
Good.
Voters get a choice in the primary. They can research the candidates, find their positions on issues and choose.
She's getting ultra-gerrymandered out of her Rep seat, so why not take the shot.
It's a primary. Not a coliseum fight.
Reposted by Piston Developers 🍿🥤🐸
🤔 AI is an acronym for "Asinine Intelligence". Or do I have it wrong? 🤔 I can never remember it correctly.
December 9, 2025 at 3:11 PM
🤔 AI is an acronym for "Asinine Intelligence". Or do I have it wrong? 🤔 I can never remember it correctly.
I know that Rust game developers in general are not the smartest people in the game industry.
However, compared to AI startups, Rust game developers look like geniuses.
I can't think of any worse financial strategies than I see in 99% of all AI startups.
You know what AI means, right?
However, compared to AI startups, Rust game developers look like geniuses.
I can't think of any worse financial strategies than I see in 99% of all AI startups.
You know what AI means, right?
December 9, 2025 at 3:04 PM
I know that Rust game developers in general are not the smartest people in the game industry.
However, compared to AI startups, Rust game developers look like geniuses.
I can't think of any worse financial strategies than I see in 99% of all AI startups.
You know what AI means, right?
However, compared to AI startups, Rust game developers look like geniuses.
I can't think of any worse financial strategies than I see in 99% of all AI startups.
You know what AI means, right?
Came across this on the Internet:
"I have developed C++ professionally for over a decade. For small things I will use Python if run time performance doesn't matter or Rust if run time performance matters."
Rust isn't going to replace C++ overnight. It will start to slowly creep in, piece by piece.
"I have developed C++ professionally for over a decade. For small things I will use Python if run time performance doesn't matter or Rust if run time performance matters."
Rust isn't going to replace C++ overnight. It will start to slowly creep in, piece by piece.
December 9, 2025 at 1:33 PM
Came across this on the Internet:
"I have developed C++ professionally for over a decade. For small things I will use Python if run time performance doesn't matter or Rust if run time performance matters."
Rust isn't going to replace C++ overnight. It will start to slowly creep in, piece by piece.
"I have developed C++ professionally for over a decade. For small things I will use Python if run time performance doesn't matter or Rust if run time performance matters."
Rust isn't going to replace C++ overnight. It will start to slowly creep in, piece by piece.
Turbine v0.1 is released!
This update removes the old code and reexports 3 libraries for content production: Turbine-Process3D, Turbine-Reactive and Turbine-Scene3D.
crates.io/crates/turbine
This update removes the old code and reexports 3 libraries for content production: Turbine-Process3D, Turbine-Reactive and Turbine-Scene3D.
crates.io/crates/turbine
crates.io: Rust Package Registry
crates.io
December 8, 2025 at 1:22 PM
Turbine v0.1 is released!
This update removes the old code and reexports 3 libraries for content production: Turbine-Process3D, Turbine-Reactive and Turbine-Scene3D.
crates.io/crates/turbine
This update removes the old code and reexports 3 libraries for content production: Turbine-Process3D, Turbine-Reactive and Turbine-Scene3D.
crates.io/crates/turbine
Turbine-Scene3D v0.42 is published!
This update moves the OpenGL backend to a separate library: `turbine_scene3d-opengl`.
This makes it easier to create new backends for Turbine-Scene3D, e.g. WGPU.
crates.io/crates/turbi...
This update moves the OpenGL backend to a separate library: `turbine_scene3d-opengl`.
This makes it easier to create new backends for Turbine-Scene3D, e.g. WGPU.
crates.io/crates/turbi...
crates.io: Rust Package Registry
crates.io
December 8, 2025 at 12:38 PM
Turbine-Scene3D v0.42 is published!
This update moves the OpenGL backend to a separate library: `turbine_scene3d-opengl`.
This makes it easier to create new backends for Turbine-Scene3D, e.g. WGPU.
crates.io/crates/turbi...
This update moves the OpenGL backend to a separate library: `turbine_scene3d-opengl`.
This makes it easier to create new backends for Turbine-Scene3D, e.g. WGPU.
crates.io/crates/turbi...
I opened a new issue for Turbine library design: github.com/PistonDevelo...
The plan is to do something similar to the Piston framework, but more for content production.
The plan is to do something similar to the Piston framework, but more for content production.
Turbine library design · Issue #179 · PistonDevelopers/turbine
The idea is to put three libraries under one: Scene3D Process3D Reactive These libraries will be backend-agnostic, similar to the Piston framework. I plan to add Turbine as one of the libraries und...
github.com
December 8, 2025 at 8:44 AM
I opened a new issue for Turbine library design: github.com/PistonDevelo...
The plan is to do something similar to the Piston framework, but more for content production.
The plan is to do something similar to the Piston framework, but more for content production.
One kind of game I would like to see, is a Chess game where you sit in the park playing against strangers.
The Chess pieces should have physics, so the wind might topple them over occasionally, or you can flip the table when you lose.
AI opponents get older the better they are.
The Chess pieces should have physics, so the wind might topple them over occasionally, or you can flip the table when you lose.
AI opponents get older the better they are.
December 8, 2025 at 12:37 AM
One kind of game I would like to see, is a Chess game where you sit in the park playing against strangers.
The Chess pieces should have physics, so the wind might topple them over occasionally, or you can flip the table when you lose.
AI opponents get older the better they are.
The Chess pieces should have physics, so the wind might topple them over occasionally, or you can flip the table when you lose.
AI opponents get older the better they are.
Maybe the first library we can integrate with WGPU is Turbine-Scene3D. It's a Scene graph library and I feel it's easier to use than working with WGPU or OpenGL directly.
We could use two Cargo features, one for OpenGL and one for WGPU. Or, we can split them up into backend crates.
We could use two Cargo features, one for OpenGL and one for WGPU. Or, we can split them up into backend crates.
December 7, 2025 at 11:40 PM
Maybe the first library we can integrate with WGPU is Turbine-Scene3D. It's a Scene graph library and I feel it's easier to use than working with WGPU or OpenGL directly.
We could use two Cargo features, one for OpenGL and one for WGPU. Or, we can split them up into backend crates.
We could use two Cargo features, one for OpenGL and one for WGPU. Or, we can split them up into backend crates.
GlutinWindow v0.73 is released!
This update took 2 days of work, where I had to first work on WinitWindow to fill out features from GlutinWindow, such that I could later upgrade GlutinWindow using source from WinitWindow.
Phew! 😮💨
We have to patch it with more fixes too.
crates.io/crates/pisto...
This update took 2 days of work, where I had to first work on WinitWindow to fill out features from GlutinWindow, such that I could later upgrade GlutinWindow using source from WinitWindow.
Phew! 😮💨
We have to patch it with more fixes too.
crates.io/crates/pisto...
crates.io: Rust Package Registry
crates.io
December 7, 2025 at 11:18 PM
GlutinWindow v0.73 is released!
This update took 2 days of work, where I had to first work on WinitWindow to fill out features from GlutinWindow, such that I could later upgrade GlutinWindow using source from WinitWindow.
Phew! 😮💨
We have to patch it with more fixes too.
crates.io/crates/pisto...
This update took 2 days of work, where I had to first work on WinitWindow to fill out features from GlutinWindow, such that I could later upgrade GlutinWindow using source from WinitWindow.
Phew! 😮💨
We have to patch it with more fixes too.
crates.io/crates/pisto...
Reposted by Piston Developers 🍿🥤🐸
see share of games that use unity or unreal
both use c++
both use c++
December 7, 2025 at 6:20 PM
see share of games that use unity or unreal
both use c++
both use c++
WinitWindow v0.20 is released!
This update brings the Winit backend for Piston up to speed with the other Window backends.
crates.io/crates/pisto...
This update brings the Winit backend for Piston up to speed with the other Window backends.
crates.io/crates/pisto...
crates.io: Rust Package Registry
crates.io
December 7, 2025 at 5:43 PM
WinitWindow v0.20 is released!
This update brings the Winit backend for Piston up to speed with the other Window backends.
crates.io/crates/pisto...
This update brings the Winit backend for Piston up to speed with the other Window backends.
crates.io/crates/pisto...
Here's my draft for configuring key code usage in the Winit backend for Piston:
![/// Settings for whether to ignore modifiers and use standard keyboard layouts instead.
///
/// This does not affect `piston::input::TextEvent`.
///
/// Piston uses the same key codes as in SDL2.
/// The problem is that without knowing the keyboard layout,
/// there is no coherent way of generating key codes.
///
/// This option choose different tradeoffs depending on need.
#[derive(Copy, Clone, PartialEq, Eq, Debug)]
pub enum KeyboardIgnoreModifiers {
/// Keep the key codes that are affected by modifiers.
///
/// This is a good default for most applications.
/// However, make sure to display understandable information to the user.
///
/// If you experience user problems among gamers,
/// then you might consider allowing other options in your game engine.
/// Some gamers might be used to how stuff works in other traditional game engines
/// and struggle understanding this configuration, depending on how you use keyboard layout.
None,
/// Assume the user's keyboard layout is standard English ABC.
///
/// In some non-English speaking countries, this might be more user friendly for some gamers.
///
/// This might sound counter-intuitive at first, so here is the reason:
///
/// Gamers can customize their keyboard layout without needing to understand scan codes.
/// When gamers want physically accuracy with good default options,
/// they can simply use standard English ABC.
///
/// In other cases, this option displays understandable information for game instructions.
/// This information makes it easier for users to correct the problem themselves.
///
/// Most gaming consoles use standard controllers.
/// Typically, the only device that might be problematic for users is the keyboard.
/// Instead of solving this problem in your game engine, let users do it in the OS.
///
/// This option gives more control to users and is also better for user data privacy.
//…](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:olippfg4rukwzrkxsdwpd5bc/bafkreidi7xogqbyxxuxi4p5qz34glev7z5yr5eddfwgrgjddnl374gdkra@jpeg)
December 7, 2025 at 11:33 AM
Here's my draft for configuring key code usage in the Winit backend for Piston:
There is no way to translate keyboard events coherently from key codes to scan codes and vice versa.
Piston follows SDL2's approach. This is because when you give instructions to the user to press e.g. "P", this should be visually coherent with the keyboard device.
Modifiers shouldn't matter.
Piston follows SDL2's approach. This is because when you give instructions to the user to press e.g. "P", this should be visually coherent with the keyboard device.
Modifiers shouldn't matter.
December 7, 2025 at 2:25 AM
There is no way to translate keyboard events coherently from key codes to scan codes and vice versa.
Piston follows SDL2's approach. This is because when you give instructions to the user to press e.g. "P", this should be visually coherent with the keyboard device.
Modifiers shouldn't matter.
Piston follows SDL2's approach. This is because when you give instructions to the user to press e.g. "P", this should be visually coherent with the keyboard device.
Modifiers shouldn't matter.
Reposted by Piston Developers 🍿🥤🐸
Game development is a process that spans multiple disciplines and paradigms.
There is no single paradigm or way of thinking that is the "final answer" of design.
Productivity depends as much on the people participating in the process, as the tools and language biases involved.
There is no single paradigm or way of thinking that is the "final answer" of design.
Productivity depends as much on the people participating in the process, as the tools and language biases involved.
December 6, 2025 at 12:04 PM
Game development is a process that spans multiple disciplines and paradigms.
There is no single paradigm or way of thinking that is the "final answer" of design.
Productivity depends as much on the people participating in the process, as the tools and language biases involved.
There is no single paradigm or way of thinking that is the "final answer" of design.
Productivity depends as much on the people participating in the process, as the tools and language biases involved.
PistonWindow v0.133 is released!
This update migrates to Winit and WGPU. Now you can turn on a "batteries" feature and get an awesome set of libraries for gamedev. Bevy ECS, Image formats, Physics, Audio, 3D assets, Scripting (Dyon), AI behavior trees ... we got it.
crates.io/crates/pisto...
This update migrates to Winit and WGPU. Now you can turn on a "batteries" feature and get an awesome set of libraries for gamedev. Bevy ECS, Image formats, Physics, Audio, 3D assets, Scripting (Dyon), AI behavior trees ... we got it.
crates.io/crates/pisto...
crates.io: Rust Package Registry
crates.io
December 5, 2025 at 8:28 PM
PistonWindow v0.133 is released!
This update migrates to Winit and WGPU. Now you can turn on a "batteries" feature and get an awesome set of libraries for gamedev. Bevy ECS, Image formats, Physics, Audio, 3D assets, Scripting (Dyon), AI behavior trees ... we got it.
crates.io/crates/pisto...
This update migrates to Winit and WGPU. Now you can turn on a "batteries" feature and get an awesome set of libraries for gamedev. Bevy ECS, Image formats, Physics, Audio, 3D assets, Scripting (Dyon), AI behavior trees ... we got it.
crates.io/crates/pisto...
MWUAHAHAHAHAHAHA! (evil laughter)
![# Batteries
bevy = { version = "0.17.3", optional = true, default-features = false }
camera_controllers = { version = "0.35.2", optional = true }
dyon = { version = "0.50.1", optional = true }
image = {version = "0.25.9", optional = true }
kira = { version = "0.11.0", optional = true }
nalgebra = { version = "0.34.1", optional = true }
vecmath = { version = "1.0.0", optional = true }
rapier2d = { version = "0.31.0", optional = true }
rapier3d = { version = "0.31.0", optional = true }
[features]
default = ["batteries"]
batteries = [
"bevy",
"camera_controllers",
"dyon",
"image",
"kira",
"nalgebra",
"rapier2d",
"rapier3d",
"vecmath",
]](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:olippfg4rukwzrkxsdwpd5bc/bafkreicjtb5leygama335qwalycyfyo6jriqh2dlgz3tyhix4xhuiufkka@jpeg)
December 5, 2025 at 2:02 PM
MWUAHAHAHAHAHAHA! (evil laughter)
The PR that fixes the memory leak in WGPU has been merged: github.com/gfx-rs/wgpu/...
I could wait for the next release, but I'll try see if can execute commands directly. Regardless of whether we'll go with RenderBundle or not, we need to optimize the design of the WGPU Graphics2D backend.
I could wait for the next release, but I'll try see if can execute commands directly. Regardless of whether we'll go with RenderBundle or not, we need to optimize the design of the WGPU Graphics2D backend.
[wgpu] Properly drop `CoreRenderBundle`s by andyleiserson · Pull Request #8661 · gfx-rs/wgpu
Possible fix for #8656.
Testing
Untested.
Squash or Rebase? Squash
Checklist
Run cargo fmt.
Run taplo format.
Run cargo clippy --tests. If applicable, add:
--target wasm32-unknown-unknown
...
github.com
December 5, 2025 at 10:59 AM
The PR that fixes the memory leak in WGPU has been merged: github.com/gfx-rs/wgpu/...
I could wait for the next release, but I'll try see if can execute commands directly. Regardless of whether we'll go with RenderBundle or not, we need to optimize the design of the WGPU Graphics2D backend.
I could wait for the next release, but I'll try see if can execute commands directly. Regardless of whether we'll go with RenderBundle or not, we need to optimize the design of the WGPU Graphics2D backend.
It's only been a few hours since I reported the memory leak in GPU, but we now have a PR (already!) by Andy Leiserson github.com/gfx-rs/wgpu/...
I tested it on the example code I gave. It fixes the memory leak.
Wow.
I knew WGPU got a lot more attention that Gfx, but still... wow. That was quick!
I tested it on the example code I gave. It fixes the memory leak.
Wow.
I knew WGPU got a lot more attention that Gfx, but still... wow. That was quick!
[wgpu] Properly drop `CoreRenderBundle`s by andyleiserson · Pull Request #8661 · gfx-rs/wgpu
Possible fix for #8656.
Testing
Untested.
Squash or Rebase? Squash
Checklist
Run cargo fmt.
Run taplo format.
Run cargo clippy --tests. If applicable, add:
--target wasm32-unknown-unknown
...
github.com
December 4, 2025 at 11:28 PM
It's only been a few hours since I reported the memory leak in GPU, but we now have a PR (already!) by Andy Leiserson github.com/gfx-rs/wgpu/...
I tested it on the example code I gave. It fixes the memory leak.
Wow.
I knew WGPU got a lot more attention that Gfx, but still... wow. That was quick!
I tested it on the example code I gave. It fixes the memory leak.
Wow.
I knew WGPU got a lot more attention that Gfx, but still... wow. That was quick!
I'm not familiar with WGPU yet, so in the process of upgrading the WGPU Graphics2D backend for Piston to use it in PistonWindow, I'm learning how it works.
It's going to be some iterations on the new PistonWindow design. There might also be some bugs, performance issues and missing features.
It's going to be some iterations on the new PistonWindow design. There might also be some bugs, performance issues and missing features.
December 4, 2025 at 9:48 PM
I'm not familiar with WGPU yet, so in the process of upgrading the WGPU Graphics2D backend for Piston to use it in PistonWindow, I'm learning how it works.
It's going to be some iterations on the new PistonWindow design. There might also be some bugs, performance issues and missing features.
It's going to be some iterations on the new PistonWindow design. There might also be some bugs, performance issues and missing features.