Spread transformation requires Symbol.iterator now

[clshortfuse]

v7 Regression

Potential Commit/PR that introduced the regression

4da3f3b

Describe the regression

Transformation of Spread now requires Symbol

Input Code

const divElements = [...document.getElementsByTagName('div')];

Babel Configuration (.babelrc, package.json, cli command)

// from webpack:

module: {
  rules: [{
    test: /\.js$/,
    exclude: /(node_modules|bower_components)/,
    use: {
      loader: 'babel-loader',
      options: {
        presets: [
          ['@babel/preset-env'],
        ],
      },
    },
  }
}

Expected behavior/code

Before this commit, all you needed was a polyfill for Array.from(). Now Symbol.iterator is needed as well. This breaks IE11.

Environment

• Babel version(s): @babel/preset-env: 7.2.3,
• Node/npm version: 8.11.3 / 6.5.0
• OS: OSX 10.14.2
• Monorepo: ??
• How you are using Babel: webpack

If, for some reason we still need Symbol.iterator, perhaps we perform the isIterable check AFTER the prototype check for [object Arguments]. That way, requiring Symbol.iterator would be a last-ditch effort.

[babel-bot]

Hey @clshortfuse! We really appreciate you taking the time to report an issue. The collaborators
on this project attempt to help as many people as possible, but we're a limited number of volunteers,
so it's possible this won't be addressed swiftly.

If you need any help, or just have general Babel or JavaScript questions, we have a vibrant Slack
community that typically always has someone willing to help. You can sign-up here
for an invite.

[nicolo-ribaudo]

Thank you for the bug report! We are tracking it at #7597, and there is a PR at #8947.

[clshortfuse]

@nicolo-ribaudo I'm not sure this is the same issue, or perhaps I'm just being too specific. #7597 targets Symbol entirely, which would also remove caveats for Array destructuring and Array For Ofas listed in https://babeljs.io/docs/en/caveats . That would fall more into a "nice-to-have" category, while this is an actual regression for Spread.

This is just talking about the regression in Spread specifically. If anything, I would say PR #8947 would fix this issue as well as #7597.

[nicolo-ribaudo]

I can leave this open if you prefer, just in case someone else has the same problem to avoid another issue.

[ljharb]

That PR only solved it for strings, not yet for arrays, or Maps or Sets (all builtins should work even without the presence of Symbol.iterator)

[nicolo-ribaudo]

Yeah sorry, I'm opening the new PR now.

[nicolo-ribaudo]

I'm leaving this open until the example in the original post works.

[nicolo-ribaudo]

The original example works when using the allowArrayLike option, or the upcoming (Babel 7.13.0) arrayLikeIsIterable assumption.