If we replace guard with assert, the compiler will crash during testing.

▌ Whoops, the AssemblyScript compiler has crashed during compile :-(
▌
▌ Here is the stack trace hinting at the problem, perhaps it's useful?
▌
▌ AssertionError: assertion failed
▌ at Z.assert (/Users/caicongcong/dev/assemblyscript/std/portable/index.js:216:11)
▌ at getExpressionId (/Users/caicongcong/dev/assemblyscript/src/module.ts:3022:3)
▌ at isConstZero (/Users/caicongcong/dev/assemblyscript/src/module.ts:3063:7)
▌ at t.operandsTostack (/Users/caicongcong/dev/assemblyscript/src/compiler.ts:6767:14)
▌ at t.makeCallDirect (/Users/caicongcong/dev/assemblyscript/src/compiler.ts:6910:24)
▌ at t.ensureConstructor (/Users/caicongcong/dev/assemblyscript/src/compiler.ts:8865:18)
▌ at t.compileNewExpression (/Users/caicongcong/dev/assemblyscript/src/compiler.ts:8758:21)
▌ at t.compileExpression (/Users/caicongcong/dev/assemblyscript/src/compiler.ts:3432:21)
▌ at t.compileVariableStatement (/Users/caicongcong/dev/assemblyscript/src/compiler.ts:3011:27)
▌ at t.compileStatement (/Users/caicongcong/dev/assemblyscript/src/compiler.ts:2241:21)
▌ at t.compileStatements (/Users/caicongcong/dev/assemblyscript/src/compiler.ts:2284:23)
▌ at t.compileFunctionBody (/Users/caicongcong/dev/assemblyscript/src/compiler.ts:1733:20)
▌ at t.compileFunction (/Users/caicongcong/dev/assemblyscript/src/compiler.ts:1612:17)
▌ at t.makeCallDirect (/Users/caicongcong/dev/assemblyscript/src/compiler.ts:6841:15)
▌ at t.compileCallDirect (/Users/caicongcong/dev/assemblyscript/src/compiler.ts:6364:17)
▌
▌ If you see where the error is, feel free to send us a pull request. If not,
▌ please let us know: https://github.com/AssemblyScript/assemblyscript/issues

some background: I am trying to revert the relationship between AS and binaryen. it will create a new c++ program integrated binaryen in native as static lib and WASM runtime. Then it will load AS bootstrapped WASM to compile source code.
The benefit is we can reuse the lots of WASM infra in binaryen to improve the performance and code size.
But it is harmful to browser compatibility...

Regarding that compiler crash, maybe we should find out where a null pointer is being passed, and see if that's intended? If it is intended, this change makes more sense.

Regarding your testing with the AS compiler embedded within a C++ executable...I don't think you can upstream that, but it might be interesting to open-source if possible. It would definitely be cool to see how necessary the JS glue in binaryen.js is to AssemblyScript.

it might be interesting to open-source if possible

open source is in progress. it should be finished in this year

Regarding that compiler crash, maybe we should find out where a null pointer is being passed, and see if that's intended? If it is intended, this change makes more sense.

It generated by module.runExpression. I would fix it in caller side.