Tuple member operator

[davidweichiang]

In #40 we are trying to implement type inference, but run into trouble with

define f x = x.1

because there's no such thing as "a tuple with at least one member."

Possible solutions:

• Go back to binary &-tuples
• Change elimination form to let (x, _, _) = e in e
• Don't let global functions be polymorphic over the size of a tuple

[ccshan]

Yeah, the complication of making .1 polymorphic between projecting the first of 3 and projecting the first of 5 is best dealt with separately later

[davidweichiang]

• Add something crazy like codata Foo = <Bar: Bool, Baz: Nat> and e.Bar. Consuming e.Bar consumes all of e.