Possible issue with validation of elem offset

[MI3Guy]

For the wasm-3.0 branch, the validation for the elem section states that the offset expression must be valid with type [i32]. I would have expected this to be a type matching the index type of the table.

Some of the test cases are generating i64 constants in this position when converted to binary using the interpreter, and based on the spec of the text format, I believe that is correct.

[bvisness]

You're right, that seems to be a mistake. The 𝐶.tables[𝑥] = limits 𝑡 is also wrong; it should have an address type, which should then be used with the offset.

[MI3Guy]

I also just noticed that there's a similar issue for the data section. I think that should be using the index type of the associated memory.

[rossberg]

Indeed, please see #1877.