Has the team explored Porffor?

[saschanaz]

Porffor is an active research project (thus not stable) that compiles JS to WASM. It seems that if this works out, the TS compiler would get free performance boost by just compiling to WASM.

Arguably it cannot be used immediately as the Test262 score is still low, but just bringing it up to make sure it's marked considered.

[RyanCavanaugh]

We're aware of it and it's a very cool concept.

"What if we wrote in a subset of JS that could be AOT'd very efficiently?" has been on the table for a while (including looking at Static Hermes) but both projects are not there yet in terms of stability/correctness, and the thing you really need for performance is shared memory concurrency over rich data structures, which implies some new thing that has a feature that isn't in JavaScript. So at that point you're authoring code in a brand new language that isn't JS and can't run in a JS environment, which is a sort of worst-of-both-worlds situation where you have JS's quirks/limitations, lack of built-in native expressitivity (e.g. should a param be a pointer or copy), and can't run natively in-browser.