Fix evaluation order with object spread

[Zzzen]

Q	A
Fixed Issues?
Patch: Bug Fix?	Yes
Major: Breaking Change?
Minor: New Feature?
Tests Added + Pass?	Yes
Documentation PR Link
Any Dependency Changes?
License	MIT

TLDR: currently, the code below is not transpiled correctly. REPL

var k = { a: 1, b: 2 };
var o = { a: 3, ...k, b: k.a++ };

expected o.a to be 1.

Long story: I found weird __assign js code generated by tsc. Out of curiosity, I checked the relevant transformer code and feed the test to babel 💥. This fix is migrated from TypeScript. Original Issue

[nicolo-ribaudo]

Can we limit this deoptimization to when path.isPure() is false for some property after the spread?

[nicolo-ribaudo]

This change seems unnecessary.

[Zzzen]

Yeah, scope.isPure is not working as intended. What's the difference between scope.isPure and path.isPure?

[Zzzen]

😵 My bad , scope.isPure is working.

[nicolo-ribaudo]

Thanks!

I have left a comment for another bug, that we can fix in a separate PR.

[nicolo-ribaudo]

I would expect this output to be

var output = { ...pureA, get foo() {}, get bar() {}, ...pureB, ...pureC, ...impureFunc(), ...pureD, pureD }

var output = _objectSpread(
  _objectSpread(
    {},
    pureA,
    { get foo() {}, get bar() {} },
    pureB,
    {},
    pureC,
    {}
  ),
  impureFunc(),
  {},
  pureD,
  { pureD }
);

I think that it's a bug in the isPure implementation, that only handles isClassMethod instead of isMethod (that also catches object methods/accessors).

[Zzzen]

🙆‍♂️ I'm gonna give it a try.

[babel-bot]

Build successful! You can test your changes in the REPL here: https://babeljs.io/repl/build/21217/

[JLHwung]

Thanks.