Copy [[PluralCategories]] in Intl.PluralRules.prototype.resolvedOptions?

[anba]

Follow-up from tc39/proposal-intl-plural-rules#28 (comment)

Seems like I've misinterpreted the SpiderMonkey implementation, because SpiderMonkey definitely returns a copy of the [[PluralCategories]] array in Intl.PluralRules.prototype.resolvedOptions(): https://searchfox.org/mozilla-central/rev/6e96a3f1e44e286ddae5fdafab737709741d237a/js/src/builtin/intl/PluralRules.js#235-240

[littledan]

I thought we settled on not copying the array. Sorry if I didn't document this clearly in bug threads, but reviewing that thread, it looks like we settled on not copying there. Do we want to reconsider this decision? @caridy @zbraniecki @thejoshwolfe

[zbraniecki]

I don't have a strong opinion here.

[anba]

I thought we settled on not copying the array.

Well, I thought SM didn't copy the array (tc39/proposal-intl-plural-rules#28 (comment)), but actually SM does the copy the array and it did so since day one. That's why I wanted to discuss this issue again. And always returning a new array is consistent with how resolvedOptions() works in general, that is it always returns a new object.

[littledan]

I'd recommend not copying, per the resolution in tc39/proposal-intl-plural-rules#28 , and we can pursue a specification change separately if you want to make the case for it. If there are no test262 tests that verify not copying, then it'd be great if someone added one.

[littledan]

We discussed this in the April 2018 Intl call and the conclusion was to return a new array each time.

[littledan]

Drafted spec text for this change at #233 . Does anyone want to create test262 tests to verify?