Thomas Lively

cc @ashleynh, who has worked on the internal representation of data segments in support of multiple memory. @dcodeIO's idea to reuse that class utility sounds like a good goal to...

It would be cool to have a global analysis to provide detailed information on where every read (of any kind) flows to. Then the global type optimization and various other...

No, not that I know of.

Being read-only is a property of the data, not individual loads, so I don't think intrinsics make much sense. What would it mean to have a read-only read with a...

It would make sense to add a `getGreatestLowerBound` function. We'll need it as part of LUB calculation if we ever support function subtyping, too.

By "benchmark" I mean measuring time to build and canonicalize the types after parsing the type section, with the goal of understanding how the type system will affect module validation...

Although it is linear, isorecursive canonicalization still does significantly more work than nominal "canonicalization," even when using a single group, so it is worth measuring. I agree that I would...

Initial measurements with a single large recursion group look good: https://github.com/WebAssembly/gc/pull/243#issuecomment-1027321008

I'm also curious about why we decided to disallow spurious wakeups. https://github.com/WebAssembly/design/pull/1019#discussion_r112070603 suggests it is because JS doesn't allow them, but I don't know why that is.

@rianhunter I also depend on the absence of spurious wakes in the code that initializes memory when threads are enabled: https://github.com/llvm/llvm-project/blob/adcd02683856c30ba6f349279509acecd90063df/lld/wasm/Writer.cpp#L728-L768. That's not user-level code, but it would be expressible...