The part that made it much more feasible is that WASM's "new" GC ended up working nicely with Julia's GC, so WebAssemblyCompiler.jl was able to hook into it. With that, a lot more codes suddenly work because array allocations are fine. It means that in this state if someone took it seriously I think a pretty good version could be had in like 6 months, but it would take a real effort to make that something actually part of the "normal" maintained compiler stack.