DEV Community

andreas wagner
andreas wagner

Posted on

The Mechanics of Good Object “immutable proxy that animate data”

#oop #java #proxy #immutable

The Good objects should always be immutable proxies that animate data, rather than mapping it. — That is the central thesis of the blog posts by Yegor Bugayenko and Mihai A. RODEGBFR.

This holds true even when they represent real-world entities that change frequently, such as a mutable “Title” of a document object.

Meme

1. The Problem: State vs. Identity

The fundamental problem in conventional object-oriented programming is the mixing of identity and state.

The Identity of an object is its uniqueness (e.g., its memory address or its unique ID).

The State is the values of its internal fields.

When we make an object mutable, we allow the internal state to change while the object’s identity remains constant. This leads to confusion and potentially inconsistent behavior, as the “essence” of the object changes without a new object being created.

A typical object consists of identity, state and behavior. An object is mutable if its internal encapsulated information (the state) can be changed after creation.

Requirement:

If a property (e.g., the title of a document) is frequently changed, it should not be part of the object’s internal state.

Example 1: The Mutable (Bad) Object

public class Document {

    private int id;  
    private String title;

    public Document(int id, String title) {  
        this.id = id;  
        this.title = title;  
    }

    // Changes the object's internal state  
    public void setTitle(String newTitle) {  
        this.title = newTitle;  
    }

    // ... Getters and toString()  
}

// Usage: The same object (identity) with new state

Document doc = new Document(50, "Old Title");  

// State changes, identity remains
doc.setTitle("New Title");
Enter fullscreen mode Exit fullscreen mode

This object is bad because it retains its internal identity (the memory address) even though its state changes.

2. The Immutable Value (Value Object)

In conventional immutable design, a new object is created with every change. The state becomes the identity.

Requirement:

If a document is very large, it is inefficient to clone the entire data structure with every small change (like a title update).

Example 2: The Immutable Value Object (Inefficient for Frequent Changes)

public class Document {

    private final int id;  
    private final String title; // Part of the internal state

    public Document(int id, String title) {  
        this.id = id;  
        this.title = title;  
    }

    // Returns a new object on change (Wither)  
    public Document withTitle(String newTitle) {  
        return new Document(this.id, newTitle);  
    }

    // ... printer methods like id() or title() 
    // and equals/hashCode (based on id AND title)  
}


// Usage: Two distinct objects (identities)  
Document first = new Document(50, "Title A");  

// Creates a new object
Document second = first.withTitle("Title B");
Enter fullscreen mode Exit fullscreen mode

3. The Solution: Immutable Proxy Object

The solution lies in recognizing that the title is not part of the object state, but part of the real-world entity (e.g., a file or a database row) that the object represents.

A good immutable object should only encapsulate its identity internally (e.g., the ID or storage location) and provide behavior to manipulate the external, mutable world. The object acts as a Proxy.

Example 3: The Immutable Proxy Object (The Recommended Approach)

The object itself is immutable (only the ID is final), but its methods manipulate an external, mutable resource (simulated here by ExternalStorage).

public class Document {  

    private final ExternalStorage externalStorage = new ExternalStorage();

    // The ONLY internal state is the identity to the real world  
    private final int id;

    public Document(int id) {  
        this.id = id;  
    }

    // BEHAVIOR: Reads the title from the external world  
    public String title() {  
        return externalStorage.readTitle(this.id);  
    }

    // BEHAVIOR: Writes the title to the external world  
    public void title(String text) {  
        // Manipulates the external entity, not the object itself  
        externalStorage.writeTitle(this.id, text);  
    }

    // equals() and hashCode() 
    // are based ONLY on the ID (the identity)

    @Override  
    public boolean equals(Object doc) {  
        return doc instanceof Document  
            && Document.class.cast(doc).id == this.id;  
    }  
}

// Simulates the external, mutable storage (database, file, etc.)  
public class ExternalStorage {

    private static final Map<Integer, String> storage = new HashMap<>(0);

    public static String readTitle(int id) {  
        return storage.getOrDefault(id, "Title not found");  
    }

    public static void writeTitle(int id, String title) {  
        storage.put(id, title);  
    }  
}

// Usage: The object (identity) remains constant,
// only the behavior changes the external world  

Document doc = new Document(50);  

doc.title("Title A"); // External storage is changed  

doc.title("Title B"); // External storage is changed 

// doc is still the SAME object
Enter fullscreen mode Exit fullscreen mode

3.1. Horizontal Decomposition with Interfaces and Decorator Pattern

“A good object always works by contracts” and their technical aspects like (Caching, Logging, UI notification) can be wrapped (horizontally) around the interfaces as Decorators.

Example 3.1: The Immutable Proxy Object (Core Structure)


import java.util.Map;
import java.util.HashMap;

// 1. Vertical Interface (Domain behavior)
public interface Document {
    String title();
    void title(String text);
}


// 2. Interface for the technical layer (Persistence behavior)
public interface DocumentStorage {
    String readTitle(int id);
    void writeTitle(int id, String title);
}


// 3. The Immutable Proxy Object (Encapsulates only the ID)
public class DefaultDocument implements Document {

    private final int id;
    private final DocumentStorage externalStorage;

    public DefaultDocument(int id, DocumentStorage storage) {  
        this.id = id;  
        this.externalStorage = storage;
    }

    @Override
    public String title() {  
        return this.externalStorage.readTitle(this.id);  
    }

    @Override
    public void title(String text) {  
        this.externalStorage.writeTitle(this.id, text);  
    }

    // equals() and hashCode() 
    // are based ONLY on the ID (the identity)

    @Override  
    public boolean equals(Object doc) {  
        return doc instanceof DefaultDocument &&
                     MyDocument.class.cast(doc).id == this.id;  
    }  
}


// 4. The base implementation of the storage
public class SimpleExternalStorage implements DocumentStorage {

    private final Map<Integer, String> storage = new HashMap<>(0);  

    @Override
    public String readTitle(int id) {  
        System.out.println("-> DB: Reading title from the database.");
        return storage.getOrDefault(id, "Title not found");  
    }  

    @Override
    public void writeTitle(int id, String title) {  
        System.out.println("-> DB: Writing title to the database.");
        storage.put(id, title);  
    }  
}
Enter fullscreen mode Exit fullscreen mode

3.1.1. Horizontal Decorator: Caching for Persistence (DB Aspect)

The CachedDocumentStorage wraps the base storage (SimpleExternalStorage) and adds the caching logic.

// CachedDocumentStorage.java
public class CachedDocumentStorage implements DocumentStorage {

    private final DocumentStorage origin; // The wrapped storage
    private final Map<Integer, String> cache = new HashMap<>(0);

    public CachedDocumentStorage(DocumentStorage origin) {
        this.origin = origin;
    }

    @Override
    public String readTitle(int id) {
        if (cache.containsKey(id)) {
            System.out.println("-> CACHE: Title retrieved from cache.");
            return cache.get(id);
        }
        // Only delegate to original logic on a cache miss
        String title = this.origin.readTitle(id);
        cache.put(id, title);
        return title;
    }

    @Override
    public void writeTitle(int id, String title) {
        this.origin.writeTitle(id, title);
        cache.put(id, title); // Update the cache
    }
}
Enter fullscreen mode Exit fullscreen mode

3.1.2. Horizontal Decorator: UI Behavior (View Aspect)

a) Observability (Event Triggering):


public class ObservedDocument implements Document {

    private final Document origin; // The wrapped domain object

    public ObservedDocument(Document origin) {
        this.origin = origin;
    }

    @Override
    public String title() {
        // Pure read operations are delegated directly
        return this.origin.title();
    }

    @Override
    public void title(String text) {

        // First, execute the technical logic (UI notification event)
        System.out.println("-> EVENT: Notifying all listeners about title update.");

        // Then, delegate the domain logic (persistence update)
        this.origin.title(text);
    }
}
Enter fullscreen mode Exit fullscreen mode

b) Presentation (Real UI Interaction):

The PresentedDocument is the final decorator that injects the simulated (Fake)UIComponent.

// Simulates a simple UI component that can display changes
public class FakeUIComponent {

    public void displayTitleUpdate(String newTitle) {
        System.out.println("-> FAKE UI: Updating title display to: " + newTitle);
    }
}

// The final decorator that controls presentation
public class PresentedDocument implements Document {

    private final Document origin;
    private final FakeUIComponent uiComponent;

    public PresentedDocument(Document origin, FakeUIComponent uiComponent) {
        this.origin = origin;
        this.uiComponent = uiComponent;
    }

    @Override
    public String title() {
        // First, read the title (goes through all decorators back to storage)
        String currentTitle = this.origin.title();
        this.uiComponent.displayTitleUpdate(currentTitle); // Display title
        return currentTitle;
    }

    @Override
    public void title(String text) {
        // First, delegate the change to the Observable layer (triggers event, writes to DB/Cache)
        this.origin.title(text);
        // After the change, update the UI immediately
        this.uiComponent.displayTitleUpdate(text);
    }
}
Enter fullscreen mode Exit fullscreen mode

3.2. Type-Safe Usage (Combination)

The client combines the aspects using interfaces.

// 1. The base storage (DB connection)
DocumentStorage dbStorage = new SimpleExternalStorage(); 

// 2. Horizontally add Caching
DocumentStorage cachedStorage = new CachedDocumentStorage(dbStorage);

// 3. The immutable Proxy Object uses the cache
Document documentProxy = new DefaultDocument(50, cachedStorage); 

// 4. Horizontally add UI behavior (Observability)
Document observedDocument = new ObservedDocument(documentProxy);

// 5. Add the final presentation layer (with UI component)
FakeUIComponent ui = new FakeUIComponent();
Document finalDocument = new PresentedDocument(observedDocument, ui);

System.out.println("\n--- First Call: Writes to DB, Cache, and fires UI Event to animate data ---");
finalDocument.title("New, cached Title"); 

System.out.println("\n--- Second Call: Reads from Cache (no DB access) and updates UI ---");
finalDocument.title();
Enter fullscreen mode Exit fullscreen mode

4. Conclusion (Object ⇒ Data Animator)

A good object should be a proxy for the effective animation of data.

The data itself (in memory, in a file, in a database) is dead and mutable.

The object is alive and immutable, using its identity (ID) as a key to access and manipulate this external, mutable data.

Diagram: Simplified Flow of Data Animation and Behavior Delegation

Diagram

Explanation of the individual statements and components:

Object = Proxy

  • Analogy: An object can be treated like a proxy.
  • Meaning: A proxy in a virtual environment representative for an object in the real world environment.

This statement suggests that in some contexts, one does not work directly with the actual object, but through a proxy that intercepts and modifies operations. This is a common design pattern in software development to control, validate, or log behavior.

State ∧ Behavior ∈ Object

  • Analogy: State and Behavior are components (∈) of an object.
  • Meaning: This is a fundamental definition of object-oriented programming.

An object bundles data (its state) and the methods that operate on that data (it’s behavior).

State ≠ Behavior

  • Analogy: State and Behavior are not the same thing.
  • Meaning: The state of an object describes its properties or its current values.

Behaviordescribes the actions the object can perform. Behavior can change the state, but the two are conceptually different things.

State = Data

  • Analogy: The state of an object is equal to its data.
  • Meaning: The state is represented by the values stored in the object’s fields or attributes.

In this sense, the state is simply the collection of data that the object contains at a specific point in time.

Behavior = Animation

  • Analogy: Behavior is equal to animation.
  • Meaning: This is a more specific interpretation of the term “behavior” in a visual or graphical context.

Particularly in the field of computer graphics and data visualization, the behavior of an object is often described by its animation — how it changes, moves, or reacts over time.

The above examples include “Behavioral Animation”, where characters move based on rules, or animations in user interfaces.

Object ⇒ Data Animator

  • Analogy: An object implies a Data Animator.
  • Meaning: This statement summarizes the previous points and interprets them in the context of data visualization.

If an Object has a State (Data) and a Behavior (Animation), then an object is essentially a Data Animator.

The Data (State) defines what should be visualized. The Animation (Behavior) ensures that the representation of the data changes or moves.

A good object is an immutable animator of mutable data. Even though it is immutable and data is mutable, it is alive and data is dead in the scope of the object’s living environment. — (Yegor Bugayenko)

5. References

🙏 Please feel free to comment 💘 or like if you found this helpful.

Top comments (0)