rfcs VS typedlua

I would add that Equality saturation/E-graphs has become quite a hot topic recently, since their POPL21 paper, with workshops dedicated to applications of e-graphs. They have even recently been added to Cranelift as an IR for optimizations.

I think with Cranelift's investment into an e-graph based optimizer (https://github.com/bytecodealliance/rfcs/blob/main/accepted/cranelift-egraph.md) they are well positioned to have quite competitive performance as a backend.

They're also actively working in this area, for example the recently added equality saturation framework and the pattern matching DSL it builds on.

There's an RFC here: https://github.com/bytecodealliance/rfcs/pull/28 and Saúl Cabrera, the person who is leading this effort and implementing the compiler tier, has a work-in-progress draft PR here: https://github.com/bytecodealliance/wasmtime/pull/4907

it's based off MIR, does it have something to do with https://mlir.llvm.org/ ?

for typed lua, there is another effort https://github.com/teal-language/tl in addition to the mentioned typescript approach: https://github.com/andremm/typedlua

Gradual typing. This is what TypeScript is: you have some typed values and untyped values, and want to verify and possibly optimize the typed code while permitting the untyped code to compile. Besides JavaScript/TypeScript, you'll find gradual typing on pretty much any other popular untyped language including Python, Lua, and Racket. Mojo is also going to have gradual typing.