Axial
Repository F# setup
open System
open System.IO
open System.Threading
open System.Threading.Tasks
open Axial
open Axial.Layers
open Axial.Console
open Axial.FileSystem
open Axial.Hosting
open Axial.Hosting.Browser
open Axial.Hosting.Node
open Axial.PlatformService
open Axial.State
open Axial.Telemetry
open Axial.Telemetry.JavaScript

Transforming Values

Pure operators transform one pulled value without introducing another effect:

> (FlowStream.fromSeq [ 1..20 ] : FlowStream<int>)
- |> FlowStream.filter (fun value -> value % 2 = 0)
- |> FlowStream.map (fun value -> value * 10)
- |> FlowStream.skip 2
- |> FlowStream.take 3
- |> FlowStream.runCollect
- |> Flow.run ();;
val it: Exit<int list,Never> = Success [60; 80; 100]

FlowStream.choose combines filtering and mapping. FlowStream.indexed, FlowStream.scan, and FlowStream.distinctUntilChangedBy retain only the state needed for the next result. FlowStream.takeWhile and FlowStream.skipWhile stop or change behavior according to the first matching value.

Effectful transformations

FlowStream.mapFlow runs one Flow for each value and emits its result:

> (FlowStream.fromSeq [ "a"; "b" ] : FlowStream<string>)
- |> FlowStream.mapFlow (fun letter -> Flow.succeed (letter.ToUpperInvariant()))
- |> FlowStream.runCollect
- |> Flow.run ();;
val it: Exit<string list,Never> = Success ["A"; "B"]

FlowStream.tapFlow runs an effect but preserves the original value. The example uses Flow.delay so printing happens when the stream pulls the value, not when the pipeline is constructed:

> let observed = ResizeArray<int>() in
- (FlowStream.fromSeq [ 1; 2 ] : FlowStream<int>)
- |> FlowStream.tapFlow (fun number ->
-     Flow.delay (fun () ->
-         observed.Add number
-         Flow.succeed ()))
- |> FlowStream.map (fun number -> number * 10)
- |> FlowStream.runCollect
- |> Flow.map (fun emitted -> List.ofSeq observed, emitted)
- |> Flow.run ();;
val observed: List<int> = seq [1; 2]
val it: Exit<Tuple<int list,int list>,Never> = Success ([1; 2], [10; 20])

Both operators remain sequential. They do not pull the next value until the current mapping has completed and the consumer asks for another result. Use Batching and parallelism when mappings should overlap.

Errors and cancellation

FlowStream.mapError changes only the typed failure channel. Defects and interruption remain structurally distinct. When an operator fails, downstream receives that cause and no further upstream values are pulled.