Shu-yu Guo

Yes, this change in early error behavior is the most worrisome behavior change. Our guiding intuition was little (if any) code depends on having early errors. Is that misguided? Would...

Also, this proposal is primarily motivated by implementations where cold load times matter, like the browser. I despise the phrase "normative optional", but it'd be nice to have the ecosystem...

We do have an opportunity here to perhaps not support sloppy mode, in that we can leverage build tooling to fail to compile existing sloppy code until they're fixed to...

@bakkot Ooh, interesting data point. Shows you how fake of a web dev I am.

Deriving it from the node name is fine. The intention was all things that need to be distinguished by some kind of tag have the type indicator. That the `Asserted*`...

@kannanvijayan It's all fine by me to keep that kind of distinction via Node subclassing. Two choices: 1. We can make another common ancestor for the non-real AST nodes to...

@littledan Yeah, that's what we had in mind.

I see how our README's statement can be misleading. I'll try to further clarify: it is not that the WebAssembly format itself is slower, but that given that WebAssembly is...

Forgot to reference the issue # in the commit. ed7acff639aaac9a67429b572c66ba05923419d9 is the attempt to clarify.

@ljharb Yeah, versioning will be very contentious. Please save the good arguments for committee!