Max Graey

I don't know yet. Plz leave this open

Good point. Need to think more. Perhaps `save` can accept optional argument. Or we can use stack-based approach for save / restore

Flow is pretty linear stuff. It can go into deep inner scope, but I don't think we need tree or DAG for handling this

Basically problem with trilling comma. This will work: ```ts export type PriceLookup = ( tokenAddress: string, block: i32, currentPool: string // removed last comma ) => string | null; ```

AS supports trailing commas even in cases where [TS fails](https://github.com/AssemblyScript/assemblyscript/blob/main/tests/parser/trailing-commas.ts#L31). We just accidentally miss this case.

> but doesn't seem very different from anything else? Generally it is different because unlike scalar operations simd operations do not have the full range of commands for usize /...

If someone wants to use usize for Simd they will just have two implementations: ```ts if (ASC_TARGET == 2) { // Wasm64 c = i64x2.add(a, b) ... } else if...

> Or they could just do v128.add(a, b) As I said above it can lead to many problems. I also forgot to mention replace_lane and extract_lane, which can lead to...

In principle, yes. I would give up the generic options at all for simd, but that would be significant breaking change. For beginning, I would like just get rid of...

I would leave everything as it is for now, except for usize / isize in simd. Because if we remove generics for atomics and simd at all it will definitely...