"A binary encoding of JS ASTs, maybe (arithmetic coders can achieve good compression; see Ben Livshits' JSZap from MSR, and earlier work by Michael Franz and Christian Stork at UCI). But this too is just a gleam in my eye, not on TC39's radar."
And WebAssembly is indeed a compressed AST encoding, not a stack bytecode. Shhh, don't tell anyone. You can still call it bytecode if you like.
With an AST representation, one can do interesting things. Everyone could have their own custom code formatting rules, and it wouldn't have to affect anyone else. Comments would have to be included in the AST for this to work sanely.
This might even go as far as languages, though for this to work well, the code would have to cover an intersection of the functionality in all the languages "rendered" to.
(Haven't watched the presentation yet. I'm in public without headphones.)
"A binary encoding of JS ASTs, maybe (arithmetic coders can achieve good compression; see Ben Livshits' JSZap from MSR, and earlier work by Michael Franz and Christian Stork at UCI). But this too is just a gleam in my eye, not on TC39's radar."
And WebAssembly is indeed a compressed AST encoding, not a stack bytecode. Shhh, don't tell anyone. You can still call it bytecode if you like.