コンテンツにスキップ
Experience Quality Coding

Chapter 4: Handling I/O: Extending the Block Universe Model

Introduction: Theoretical Consistency for Interactions

The previous chapters established the Timeline<'a> type, grounded in the Block Universe model (Chapter 0 and 2), and introduced fundamental operations like TL.at, TL.define (Chapter 2), and crucially, TL.map (Chapter 3). We saw how TL.map allows us to create new timelines derived from existing ones and how these operations build a conceptual Dependency Graph.

This chapter extends the Block Universe philosophy to another area often considered problematic in pure functional programming: Input/Output (I/O). Operations like reading user input, writing to files, or printing to the console inherently involve interaction with an external world.

Just as the internal state changes (_last) of Timeline<'a> were reframed as simulating observation within the Block Universe, I/O operations can also be integrated consistently. When viewed within the Block Universe, the side effects of I/O are simply part of the description of the universe’s immutable state transitioning between different time coordinates. Therefore, to maintain theoretical consistency, all I/O interactions should also be brought within the Timeline framework. Wrapping I/O actions using Timeline objects and operations like TL.map allows us to manage them declaratively within the same dependency graph, ensuring the integrity of our “mini Block Universe” simulation even when interacting with the external world.

4.1 I/O Functions: The Standard View vs. Block Universe Perspective

Section titled “4.1 I/O Functions: The Standard View vs. Block Universe Perspective”

Consider standard I/O functions like printfn (F#) or console.log (JavaScript). Conventional functional programming identifies these as fundamentally impure due to their side effects. They modify the state of an external entity (the console display) and don’t typically return meaningful values based solely on their inputs.

However, let’s re-examine this through the lens of the Block Universe model. In this view, everything is part of the immutable block. The state of the console display before a printfn call (at time t1) and the state after the call (at time t2) are simply two different, fixed states within the overall Block Universe. The printfn function itself simply describes the relationship or transition rule between these two immutable states.

The challenge isn’t that I/O is fundamentally incompatible with an immutable universe, but rather how to represent and manage these state transitions within our Timeline-based programming model. The core principle is straightforward: all I/O operations must also be managed by Timeline.

4.2 Wrapping Output Actions: The “Hello, World!” of Timeline I/O

Section titled “4.2 Wrapping Output Actions: The “Hello, World!” of Timeline I/O”

The most fundamental example is standard output. By using the TL.map function introduced in the previous chapter (Chapter 3), we can easily wrap an I/O action like F#‘s printfn. This involves creating a Timeline whose updates trigger the desired I/O side effect.

Note Header

For convenience in the following examples, we’ll use a simple helper function log that wraps printfn to handle generic types. (Assuming isNull is a globally available helper as discussed in Unit 6, Chapter 0).

F# Logo 
// Helper function for generic logging
// Adhering to F# style guide for function definition
let log<'a> : 'a -> unit =
    fun a -> printfn "%A" a
Note Footer

Now, let’s wrap this log action using Timeline and TL.map.

Example 1: Integer Logging Timeline

F# Logo 
// Assume Timeline factory, Now value, and TL module functions (TL.map, TL.define) are accessible
// No 'open Timeline' or 'open Timeline.TL'


// Create a Timeline initialized with an integer value
let logIntTimeline : Timeline<int> = Timeline 5 // Explicit type


// Connect the log function to the timeline using TL.map.
// TL.map returns a new timeline, but here we only care about the side effect.
// The dependency (logIntTimeline -> reaction) is established internally.
logIntTimeline
|> TL.map log // Apply the 'log' function whenever the timeline updates; Explicit TL.map
|> ignore   // We ignore the resulting timeline (often Timeline<unit>)


// Output: 5 (printed immediately upon map application due to initial value)

In this first example, logIntTimeline is created with an initial value of 5. The TL.map log operation immediately applies the log function to this initial value, causing 5 to be printed. Furthermore, a dependency is created: whenever logIntTimeline is updated via TL.define, the log function will be called again with the new value.

Example 2: String Logging Timeline (with Null Handling)

F# Logo 
// Assume 'isNull' is a globally available helper as established in Unit 6, Chapter 0.


// Create a Timeline for strings, initialized with null
let logStringTimeline : Timeline<string> = Timeline null // Explicit type


// Connect the log function, adding logic to handle null
let logNonNullString : string -> unit = // Define the function passed to map
    fun value ->
        if not (isNull value) then // Adhering to if/then/else style
            log value // Log only if the value is not null
        else
            () // Explicitly do nothing if null, or just omit else for implicit unit
        // No else branch explicitly needed if we only act on non-null;
        // 'if condition then expr' implicitly returns unit if condition is false.
        // For clarity, we can keep the empty 'else ()'.


logStringTimeline
|> TL.map logNonNullString // Explicit TL.map
|> ignore // Setup the reaction, ignore the resulting Timeline<unit>


// Output: (nothing printed initially because the value is null)

Here, logStringTimeline starts with null. The TL.map operation sets up a reaction that only calls log if the incoming value is not null. We can trigger the log later using TL.define:

F# Logo 
// Later, define a new value onto the timeline
logStringTimeline |> TL.define Now "Hello" // Explicit TL.define
// Output: Hello (printed when define is called)


logStringTimeline |> TL.define Now "Timeline I/O!"
// Output: Timeline I/O! (printed when define is called)


logStringTimeline |> TL.define Now null // Define null again
// Output: (nothing printed)

These examples constitute the “Hello, World!” for using Timeline to manage I/O. They demonstrate the fundamental pattern: wrap the I/O action within a function and use TL.map to apply that function whenever a source Timeline updates. Trigger the action by updating the source Timeline using TL.define.

(Why No Built-in log Timeline? User Choice!) You might wonder why a standard logTimeline isn’t included directly in the Timeline library. The reason lies in flexibility. Users might need different logging formats, destinations, or null-handling logic. Providing the core tools (Timeline factory, TL.map, TL.define) allows users to easily implement the exact I/O behavior they need by passing the appropriate function to TL.map.

4.3 Linking Timelines to Monitor for Debugging

Section titled “4.3 Linking Timelines to Monitor for Debugging”

Once you’ve created a useful I/O timeline like logStringTimeline, it becomes a powerful tool for debugging. The TL.link function provides a convenient way to connect two timelines.

The TL.link function simply propagates updates from a source timeline (timelineA) to a target timeline (timelineB). Internally, as per Timeline.fs, it registers a direct dependency and performs an initial sync.

Note Header

TL.link sets up a dependency so that any value defined on timelineA is subsequently defined on timelineB. It also typically propagates the initial value immediately.

// Signature (from Timeline.fs, TL module):
// val link<'a> : Timeline<'a> -> Timeline<'a> -> unit

(Note: The original text mentioned “Internally, it likely uses TL.map”. While link could be conceptually built with map, the provided Timeline.fs shows a direct implementation using DependencyCore.registerDependency and an initial TL.define for sync. The text here is adjusted to reflect a more general statement about its effect, not presuming its exact internal implementation in this explanatory chapter unless crucial.)

Note Footer

Consider a scenario where you have an arbitrary timeline in your application, timelineA, and you want to monitor its value changes using logStringTimeline:

F# Logo 
// An arbitrary timeline in your application (ensure type matches target)
let timelineA : Timeline<string> = Timeline null // Start with null; explicit type


// Assume logStringTimeline is already set up as before (prints non-null strings)
// let logStringTimeline : Timeline<string> = ...


// Simply link timelineA to your logging timeline!
timelineA |> TL.link logStringTimeline // Explicit TL.link
// Output: (nothing printed initially as timelineA is null and logStringTimeline's map handles null)

With this single line, any update to timelineA will now be automatically propagated to logStringTimeline, which in turn will print the value (if not null, based on logStringTimeline’s setup).

F# Logo 
// Now, whenever timelineA is updated...
timelineA |> TL.define Now "linked"
// Output: linked (propagated to logStringTimeline and printed)


timelineA |> TL.define Now "message!"
// Output: message! (propagated and printed)


timelineA |> TL.define Now null
// Output: (nothing printed)

Since all dynamic values in this system can be represented by Timelines, TL.link provides a simple yet powerful way to observe the state of any part of your application by connecting it to a pre-configured logging or display Timeline.

4.4 Handling Asynchronous Input: HTTP Request Example

Section titled “4.4 Handling Asynchronous Input: HTTP Request Example”

The previous examples showed how to send data to an I/O timeline (output). However, I/O often involves receiving data from the external world, such as the response to an HTTP request (input). We can handle this by using TL.map to trigger the asynchronous operation and TL.define within the async workflow to feed the result back into a Timeline.

Let’s sketch a simple example for making an HTTP GET request.

1. Define Timelines for Request and Response:

We need one timeline to trigger the request (e.g., by defining the URL) and another to receive the response content.

F# Logo 
open System.Net.Http // Required for HttpClient


// Timeline to trigger the request with a URL
let httpRequestUrlTimeline : Timeline<string> = Timeline null // Explicit type


// Timeline to receive the response content (or error message)
let httpResponseTimeline : Timeline<string> = Timeline null // Explicit type

2. Set up the I/O Handler (Async Request):

This uses TL.map on the request timeline. When a non-null URL is defined, the function passed to TL.map triggers an asynchronous HTTP GET request. When the request completes (successfully or with an error), it calls TL.define on the response timeline.

F# Logo 
// Use a single HttpClient instance for efficiency (simplified example)
let httpClient : HttpClient = new HttpClient() // Explicit type


// Define the function that will be passed to TL.map
let httpRequestFn : string -> unit = // Explicit function type
    fun url ->
        if not (isNull url) then // Using global isNull
            // Asynchronously perform the HTTP GET request
            async {
                try
                    // Perform the async operations
                    let! response = httpClient.GetAsync(url) |> Async.AwaitTask
                    response.EnsureSuccessStatusCode() |> ignore // Throw exception on HTTP error
                    let! content = response.Content.ReadAsStringAsync() |> Async.AwaitTask


                    // Define the successful response content onto the response timeline
                    httpResponseTimeline |> TL.define Now content // Explicit TL.define


                with ex ->
                    // Define an error message onto the response timeline
                    let errorMsg : string = sprintf "HTTP Request Failed: %s - %s" url ex.Message // Explicit type
                    httpResponseTimeline |> TL.define Now errorMsg // Explicit TL.define
            }
            |> Async.StartImmediate // Start the async workflow without waiting
        else
            () // Do nothing if URL is null


// Set up the reaction on the request timeline using map
httpRequestUrlTimeline
|> TL.map httpRequestFn // Explicit TL.map
|> ignore // Ignore the Timeline<unit> result of map (since httpRequestFn returns unit)

3. Connect the Response Timeline (e.g., to Logging):

Now, we can react to the results arriving on httpResponseTimeline just like any other timeline, for example, by linking it to our logger (assuming logStringTimeline is set up).

F# Logo 
// Assume logStringTimeline is set up as before
// let logStringTimeline : Timeline<string> = ...


// Link the response timeline to the logging timeline
// httpResponseTimeline |> TL.link logStringTimeline // Example, assuming logStringTimeline is defined

(Example link commented out as logStringTimeline setup is assumed from prior examples, not redefined here).

4. Trigger the Request:

Defining a URL onto httpRequestUrlTimeline starts the process.

F# Logo 
// Trigger the HTTP request by defining a URL
httpRequestUrlTimeline |> TL.define Now "https://www.google.com"


// Console Output (from logStringTimeline, after request completes, example):
// Hello (if logStringTimeline was linked and google.com returned "Hello")
// or:
// HTTP Request Failed: https://www.google.com - <error details>

(The console output example here is more generic, as the actual output depends on what logStringTimeline is linked to and how it processes the data from httpResponseTimeline.)

In this pattern:

  1. Defining a value on httpRequestUrlTimeline triggers the function inside TL.map.
  2. This function starts an asynchronous I/O operation.
  3. When the operation completes, its result (or error) is fed back into the reactive system by calling TL.define on httpResponseTimeline.
  4. Other parts of the system (like a logger connected via TL.link) react declaratively to updates on httpResponseTimeline.

4.5 Benefits: System-Wide Consistency

Section titled “4.5 Benefits: System-Wide Consistency”

By wrapping I/O side effects (both output and input) within Timelines and using standard operations like TL.map, TL.define, and TL.link to manage them, I/O becomes a well-behaved participant in the FRP dependency graph introduced earlier (Chapter 3).

This approach ensures that:

  1. Consistency: All dynamic values and actions, including interactions with the outside world, are managed through the same Timeline propagation mechanism.
  2. Declarative Control: Dependencies and reactions (including I/O triggers and responses) are defined declaratively using functions like TL.map and TL.link, making the system flow easier to reason about.
  3. Model Integrity: The “mini Block Universe” simulation remains internally consistent. External interactions don’t happen as unmanaged side effects but are channeled through Timelines, respecting the dependency graph.

This practice of representing I/O via Timelines is fundamental to building robust, large-scale applications using this FRP philosophy, ensuring the entire system adheres to the principles derived from the Block Universe model.