Proposed investigation: move filter calculation to WASM

[KaliedaRik]

Proposal
Currently, all SC calculation happens in the JS main thread, which has clear (and annoying) disadvantages and dangers for the web page hosting an SC canvas. Thus there is a clear need to move the most compute-intensive calculations out of the main thread.

This has been attempted before: prior to SC v8.4.0 filter calculations were handled in a Web Worker. That approach however led to an Async Hell and a Cascade of Promises so, for v8.4.0, the entire approach/functionality associated with filters in SC was rewritten from scratch - including the abandonment of the Web Worker (which ironically led to faster canvas execution).

WebAssembly is a (relatively) new language which can run in the browser:

WebAssembly is a type of code that can be run in modern web browsers — it is a low-level assembly-like language with a compact binary format that runs with near-native performance and provides languages such as C/C++, C# and Rust with a compilation target so that they can run on the web. It is also designed to run alongside JavaScript, allowing both to work together.

It makes sense to investigate whether we could port the SC filters code over to WebAssembly, with the aim of making the filter functionality and output significantly faster.

For this investigation, I propose we look at porting the code over to the Zig language, which supports building for WebAssembly out of the box.

• Why use Zig? The language's developers claim it is a rewrite of C, but with many of the dangers of C mitigated or removed.
• Why not use Rust? Rust is a reasonable alternative, but if I'm going to learn a new language for this investigation, my personal preference is to give Zig a go.

Risks
There are no risks involved with this investigation. Anything that leads to moving high cost code off the JS main thread, without damaging the library's current speed, has to be a win!

Also, if Zig doesn't work out, there's other languages we can try.

Way forward

• Check out the sleibrock/zigtoys GitHub for inspiration about learning about Zig and Wasm
• daneelsan also has a minimal-zig-wasm-canvas demo up on GitHub that may offer pointers on how to get this approach to work.