Shu-yu Guo

I would like to push back on this, because using De Bruijn indices is extra verification and inches us closer to language bifurcation. I want the binary representation to be...

@Yoric The problem I foresee is extra verification. If indices are used in the tree structure itself, the bijection to source text is lost. In addition to verifying that the...

> I'm not sure I follow. In which case would it require gensymming new names? Ah, I believe I misunderstood your scheme. The declarations are staying as-is, so the original...

I think full streaming compilation is the realistic goal, see issue at https://github.com/binast/ecmascript-binary-ast/issues/12 I think streaming interpretation, given the deferred function-at-a-time error model we're going with, will end up slowing...

@Yoric did a great job separating out the @isiahmeadows' language feature ideas from other suggestions. One of the highest, if not *the* highest constraints in this design space is to...

I believe this is the right way forward too. IMO the biggest gain here is that it makes adoption a lot more practical. Old codebases have an option to get...

> Entire function, including nested functions? If so, we're game over, but I would be a bit surprised by it. Good clarification. I meant **not** including nested functions. > What's...

I don't understand 4. How do you know there will be no deferred `SyntaxError`s ahead of time if you're streaming the interpretation of the top level?

We discussed in a backchannel some more and have agreed on that @Yoric's 1-4 make sense. I've been using "streaming interpretation" to mean a fine-grained streaming execution. To avoid confusing...

@bakkot is right that `const` can't be merged with `var`. In the interest of making incremental progress, I recommend we punt on Annex B.3.3 until things are more mature. I...