Unoptimized WASM output

[guest271314]

Here's the source code I used

import { readSync } from "node:fs";
const buffer = new Uint8Array(32);
readSync(0, buffer,  0, buffer.length );
process.stdout.write(buffer);

Here's the compilation command

SCRIPTC_NO_CACHE=1 SCRIPTC_CC=zigcc SCRIPTC_TARGET=wasm32-wasi bun x scriptc build hello.ts --keep-c --optimization release -o hello.wasm >/dev/null

Now, that spits out a 1.8 MiB WASM binary.

However, if I then do wasm2wat then wat2wasm starting with that same hello.wasm file, I get a 565.3 KiB WAT file and then a 53.1 KiB WASM file.

wabt-1.0.41/bin/wasm2wat hello.wasm -o hello.wat
wabt-1.0.41/bin/wat2wasm hello.wat -o hello1.wasm

Using wasm-opt directly on the hello.wasm file I get down to 46.9 KiB

wasm-opt hello.wasm -Os --strip-debug --vacuum -o hello2.wasm

Two questions arise for me:

1. How to determine if QuickJS NG is embedded in the resulting WASM binary?
2. How to optimize the WASM binary output so that is roughly matches the output of wasm-opt?