Bug Report

• I would like to work on a fix!

Current behavior

Re: https://stackoverflow.com/questions/66609657/optional-chaining-function-prototype-apply-was-called-on-undefined-which-is-a

There's a bug in Chrome when invoking a bare optional call (not a method on an object) when there is a spread arguments and a following argument:

Input Code
REPL (Chome throws, Firefox/Safari are fine)

const fn = null;
const n = null;
const o = {};

// Fine
fn?.(...[]);
o?.toString(...[], 1);
o?.toString?.(...[], 1);
n?.method?.(...[], 1);

// Throws (must be any argument that follows a ...spread)
fn?.(...[], 1); 
fn?.(...[], ...[]);
o.method?.(...[], 1);
// This one is most surprising!
n?.method(...[], 1);

Expected behavior

None of these should throw in Chrome

Babel Configuration (babel.config.js, .babelrc, package.json#babel, cli command, .eslintrc)

• Filename: babel.config.js

{
  "presets": [
    [
      "@babel/preset-env",
      {
        "targets": "Chrome > 80",
        "bugfixes": true
      }
    ]
  ]
}

Possible Solution

We need to add a new bugfix transform that detects these ...foo spread arguments inside an optional chain, and forces the transform.

Additional context

Edits by @JLHwung

• Added link to v8 issue for reference.