KO
|
EN
gitlite — search
Search
#javascript
#python
#hacktoberfest
#react
#ai
#typescript
#llm
#go
#golang
#android
#machine-learning
#rust
#deep-learning
#linux
frpong
★ 13
Open GitHub ↗
No description available.
Download README (.md)
Explore Similar Repositories
VEDA-InfiniteRave
:
No description available.
vuetypes
:
Spec for Types in Vue to enhance Vue template editing experience
auto-ui
:
凹凸 UI 组件
ulauncher-remmina
:
🖥 A Ulauncher extension for quick access to Remmina profiles
100-million-requests-aiohttp
:
Creating enormous tasks using aiohttp(asyncio) without memory error.
// repository documentation
Was this content helpful?
★ 0
(0 ratings)
Select Rating:
★
★
★
★
★
Submit Feedback
Recent Feedback
×
Download README
Do you want to download the
README.md
file for
frpong
?
Download (.md)
[](https://travis-ci.org/dakom/frpong) # [Live Demo](https://dakom.github.io/frpong) ---- FRPong is a proof-of-concept for architecting a game on the web using a combo of technologies. Specifically, a mixture of Purescript, SodiumFRP, Rust/Wasm, Typescript, WebGL/WebAudio/WebWorkers, etc. ## Controls * Serve Ball: **Space** * Move Paddle: **up/w** or **down/s** # Tech Specs ## Game (via WebWorker) * Purescript + SodiumFRP + Rust + Typescript glue * Main logic is all Purescript and SodiumFRP * Solves collision, out-of-bounds checking, and interpenetration adjustments by using time and history * Physics is done by changing _functions_ to derive motion, not per-tick numerical accumulation * Rust/WebAssembly is mostly just a proof of concept and used for math helpers ## IO (entrypoint) * Raw WebGL renderer * WebAudio synth * Passes input controller and tick updates * Typescript * CRT shader effect taken from [libretro](https://github.com/libretro/glsl-shaders/blob/master/crt/shaders/crt-pi.glsl) (only looks good on largish window size) # Future ideas * With upcoming offscreenCanvas support, might separate rendering into its own thread too * Could use a renderer built for functional pipelines, like [pure3d](https://github.com/dakom/pure3d) * Train AI against human data (would require hundreds of hours of playing...) * Library idea - automatically adapt all the libretro shaders for easy swapping # Misc Notes * Experimental and purposefully over-engineered - though I dunno, this also kinda feels like a cool starting point? * The multi-threading is probably pointless and wasteful in this case, but it doesn't really hurt afaik * Different webpack modes ("dev", "build" and "bundle") - as well as travis setup for running tests * It might be worth exploring state serialization via flatbuffers and then using that to transfer across all boundaries (even to the wasm layer)
