32-bit and smaller integers

All 32-bit and smaller integers are represented on Deno side as plain JS numbers. Passing in floating point numbers as integers does not cause a TypeScript type warning but will cause a runtime error. It will also cause the symbol function to deopt.

8 and 16 bit numbers are internally converted from the V8 32-bit integer class using Rust's as u8 / as u16 conversion, eg.

let u8_value = v8::Local::<v8::Uint32>::try_from(arg) // `v8::Uint32` is the V8's internal 32-bit unsigned integer class
  .map_err(|_| type_error("Invalid FFI u8 type, expected unsigned integer"))?
  .value() as u8; // the `.value()` method (`Value()` in C++ API) returns a `u32`

and in V8 Fast API calls using dedicated CPU calls to copy only the used bytes from the return value, zeroing or sign-extending the rest.

Fast API support

All 32-bit and smaller integers are Fast API compliant as both parameters and return values.