`differenceBy`

difference, comparing by a computed key instead of value equality.

Iterable<A> differenceBy<A, B>(B Function(A a) f, Iterable<A> iterable1, Iterable<A> iterable2) FxAsyncIterable<A> differenceByAsync<A, B>(FutureOr<B> Function(A a) f, FxAsyncIterable<A> iterable1, FxAsyncIterable<A> iterable2) Fx<T> Fx<T>.differenceBy<B>(B Function(T a) f, Iterable<T> other) // chain: values of the chain not keyed in other

Lecture

differenceBy(f, iterable1, iterable2) follows the exact same argument-order rule as difference: the result comes from iterable2, keeping only the elements whose f-key does not appear among iterable1's f-keys (with duplicates in the result collapsed). The f function is applied to both iterables, so — unlike differenceBy in some languages — both arguments must share the same element type A; only the derived key type B is what's being compared.

This is the shape you reach for when the "exclusion list" and the "source list" are full records rather than bare values: a list of blocked user objects and a list of user objects, matched by id; or an unavailable-products list and a product catalog, matched by sku.

difference itself is just differenceBy((a) => a, iterable1, iterable2). There's no chain method for either — call the data-first function directly. On the async side, the concurrency marker applies to iterable2, same as differenceAsync.

Demo 1 · Basics

Demo 2 · Async, with concurrency

Try it yourself

Exercise: use differenceBy (keyed on 'sku') to find products that are still available.

Related: difference — the value-equality version · intersectionBy — keep by a shared computed key instead · uniqBy — dedupe a single iterable by key · compress — filter by a parallel boolean mask