Parallel Collection Work

List.parallel splits pure collection transformations across worker threads.

Pure Callbacks

The callback is pure, which prevents shared mutable state and I/O inside the parallel work.

items.parallel(
    pure func(chunk): chunk.map(pure func(x): x * 2)
)

List.parallel()

parallel hands the callback a ParallelList view with parallel-safe map, filter, and filter_map.

items.parallel(
    pure func(chunk): chunk.map(pure func(x): x * x)
)

When It Helps

Use it for expensive per-element work, not tiny transforms where scheduling costs dominate.

pure func expensive(n: Int) -> Int:
	n * n + n

Collecting Results

parallel returns an ordinary List after the scoped work completes.

results: List[Int] = transform(items)

List.concurrent allows impure callbacks and bounds active tasks. It returns List[Result[T, ConcurrencyError]] in input order, not completion order. List.parallel requires pure callbacks and returns ordinary values.

import:
	channel: sleep_for
	units: milliseconds


func render_page(name: String) -> String:
	sleep_for(milliseconds(10))
	"rendered ${name}"


func render_all(names: List[String]) -> List[Result[String, ConcurrencyError]]:
	names.concurrent(4, func(name: String): render_page(name))

A Shared Deadline

concurrent_with_timeout applies one whole-operation deadline. Inspect every result rather than reporting a partial batch as successful.

import:
	channel: sleep_for
	units: milliseconds, seconds


func fetch_metric(id: Int) -> Int:
	sleep_for(milliseconds(10))
	id * 10


func fetch_all(ids: List[Int]) -> List[Result[Int, ConcurrencyError]]:
	ids.concurrent_with_timeout(4, seconds(5), func(id: Int):
		fetch_metric(id)
	)

Example

parallel-chunks.brp

pure func expensive(n: Int) -> Int:
	n * n + n


pure func transform(items: List[Int]) -> List[Int]:
	items.parallel(pure func(chunk: ParallelList[Int]):
		chunk.map(pure func(n): expensive(n))
	)


pure func sum(items: List[Int]) -> Int:
	var total: Int = 0
	for item in items:
		total += item
	total


func main(args: List[String]) -> Void:
	print(sum(transform([1, 2, 3, 4]))) -- prints: 40

Try it

terminal

blorp run parallel-chunks.brp