Closed ngzhian closed 3 years ago
Btw, I'll be on vacation for the rest of the month, so expect some delay in code reviews.
I wonder if it wouldn't be cleaner (in both spec and interpreter) if Simd was treated as a third kind of value type (perhaps introducing a new name for the union of the two).
Hm, not too sure if what the end result will look like if we do that. I will file a TODO on myself to think about this a bit more and maybe do some experiments, thanks for the suggestion! Merging as it is for now.
Quite a number of changes from upstream because ref-types was merged. Most of the fixes have to do with num_type v.s. value_type, wrapping up V128 into Num, or pattern matching a Num V128.
Moved some code from Eval_numeric into Values to avoid circular imports:
For the spec text, I put SIMD 128 sections after the reference types section, e.g. in validation.