any chance we can get a `vararg` that means one or more rather than zero or more?

# language-proposals

[groostav]

any chance we can get a vararg that means one or more rather than zero or more?

[ianbrandt]

I filed an issue for this: https://youtrack.jetbrains.com/issue/KT-57266/One-or-more-varags.

I believe it would pair nicely with non-empty collections (https://youtrack.jetbrains.com/issue/KT-21217/Non-empty-collections) for robust, compile-time checked domain modeling of one-or-more scenarios.

[ephemient]

perhaps this should depend on what happens in https://youtrack.jetbrains.com/issue/KT-43871/Collection-literals. using arrays to represent varargs is always going to require runtime checking of one-or-more. if you could instead write

fun f(args: NonEmptyList<T>)
f([1, 2, 3])

I think it would be a better option

[ianbrandt]

Maybe varargs+ could be represented by a new NonEmptyArray type, but I agree that collection literals would be related and should be considered.

[groostav]

how far can a contract get us here?

[ephemient]

vararg+ wants to constrain callers to provide at least one argument
contracts do not constraint callers. they declare certain guarantees about the callee's behavior to the caller