As in, I wish that it actually accomplished all the things we want it to. Right now it is useful for making some computation more efficient, but that is about it. In its current state, it lacks the ability and configurability to do more. WASI ended up being a mediocre shim and a promise that has yet to ship. Components still aren't there yet either. Memory management doesn't exist (grow, but no shrink) and threading is half-baked.

Now, this has been a harsh review of WASM, which otherwise has accomplished some applaudable goals, but it cannot be overstated how much better the computing landscape would be if WASM had these missing features today. It really does have the potential to be a single, performant, multiplatform, flexible compilation target. We could have operating systems that embed a WASM engine in the kernel and execute everything there, benefitting from everything that WASM does well (eg. sandboxing) in addition to removing overhead and traditionally problematic integration points. It could be a platform that spans from the OS to the web seamlessly. That is a future I would like to live in.

My hope is that one day WASM will be real.