| 01:50 | <sirisian> | I'm probably searching for the wrong keywords here. If you can use a record with Map and Set, why can't it be used as a key in a record? #{ #{ a: 1 }: 2 }? |
| 02:23 | <Ashley Claymore> | Related: https://github.com/tc39/proposal-record-tuple/issues/203 |
| 02:26 | <Ashley Claymore> | The main spec complexity is that the spec would also need to define an ordering for R&T because their keys are sorted |
| 02:27 | <Ashley Claymore> | I also imagine it would add considerable implementation complexity to allow more types as keys, relative to "only strings". |
| 02:31 | <Ashley Claymore> | It did come up in conversations, as it kind of provides a kind of immutable set. But overall did not seem motivated enough to include in the proposal, which was trying to be maximally minimal |
| 03:48 | <sirisian> | yeah, not having a kind of sort is why there aren't sets or unordered tuples it seems. Just realized both ways would solve what I was trying. I was basically trying to use records as keys to make a set. |