MatterAI - Frontier LLM Intelligence at 70% Less Cost

D Anti-Patterns Overview

D, despite being a powerful and versatile language, has several common anti-patterns that can lead to performance issues, maintainability problems, and bugs. Here are the most important anti-patterns to avoid when writing D code.

Excessive Use of Runtime Type Information (RTTI)

// Anti-pattern: Excessive use of RTTI
void processObject(Object obj) {
    if (auto stringObj = cast(String)obj) {
        // Handle string
        writeln("String: ", stringObj);
    } else if (auto intObj = cast(Integer)obj) {
        // Handle integer
        writeln("Integer: ", intObj.value);
    } else if (auto arrayObj = cast(Array)obj) {
        // Handle array
        writeln("Array with ", arrayObj.length, " elements");
    }
    // And so on for many types...
}

// Better approach: Use polymorphism or templates
interface Processable {
    void process();
}

class String : Processable {
    string value;
    this(string value) { this.value = value; }
    void process() { writeln("String: ", value); }
}

class Integer : Processable {
    int value;
    this(int value) { this.value = value; }
    void process() { writeln("Integer: ", value); }
}

void processObject(Processable obj) {
    obj.process();
}

// Or using templates
void processObject(T)(T obj) {
    static if (is(T == string)) {
        writeln("String: ", obj);
    } else static if (is(T : int)) {
        writeln("Integer: ", obj);
    } else static if (isArray!T) {
        writeln("Array with ", obj.length, " elements");
    } else {
        static assert(false, "Unsupported type: " ~ T.stringof);
    }
}

Avoid excessive use of runtime type information (RTTI) with cast operations. This approach is slow, error-prone, and leads to brittle code. Instead, use polymorphism, interfaces, or templates to handle different types in a more type-safe and efficient manner.

Not Using @safe, @trusted, and @system Properly

// Anti-pattern: Not using safety attributes
void* dangerousFunction(int* ptr) {
    void* result = ptr;  // Unsafe pointer conversion
    result += 10;       // Unsafe pointer arithmetic
    return result;
}

// Better approach: Use safety attributes
@system void* unsafeOperation(int* ptr) {
    void* result = ptr;  // Unsafe pointer conversion
    result += 10;       // Unsafe pointer arithmetic
    return result;
}

@trusted void* wrappedUnsafeOperation(int* ptr) {
    // Call unsafe code but ensure it's used safely
    return unsafeOperation(ptr);
}

@safe void safeFunction() {
    // Only calls @safe or @trusted functions
    int[5] arr = [1, 2, 3, 4, 5];
    auto result = wrappedUnsafeOperation(&arr[0]);
    // Use result safely...
}

Use D’s safety attributes (@safe, @trusted, and @system) to clearly indicate the safety level of your code. Mark unsafe code as @system, wrap it with carefully reviewed @trusted functions, and aim to make most of your codebase @safe. This helps prevent memory safety issues and makes your code more robust.

Excessive Use of Shared Mutable State

// Anti-pattern: Excessive shared mutable state
int globalCounter = 0;

void incrementCounter() {
    globalCounter++;
}

void processData() {
    // Multiple threads might access globalCounter
    incrementCounter();
    // More processing...
}

// Better approach: Use immutability and message passing
immutable int CONSTANT_VALUE = 42;

// Or use thread-local storage
__gshared int globalCounter = 0;  // Explicitly mark as global shared

// Or use proper synchronization
import core.sync.mutex : Mutex;

class ThreadSafeCounter {
    private int _value = 0;
    private Mutex _mutex;
    
    this() {
        _mutex = new Mutex();
    }
    
    void increment() {
        synchronized(_mutex) {
            _value++;
        }
    }
    
    int value() {
        synchronized(_mutex) {
            return _value;
        }
    }
}

Avoid excessive use of shared mutable state. Shared mutable state makes code harder to reason about and can lead to race conditions in concurrent programs. Instead, prefer immutability, message passing, or proper synchronization mechanisms when shared state is necessary.

Not Using UFCS (Uniform Function Call Syntax)

// Anti-pattern: Not using UFCS
string convertToUpper(string input) {
    return toUpper(input);  // Traditional function call
}

// Better approach: Embrace UFCS
string convertToUpper(string input) {
    return input.toUpper();  // Using UFCS
}

// This allows for chaining operations
string processString(string input) {
    return input
        .strip()
        .toUpper()
        .replace("OLD", "NEW");
}

Embrace Uniform Function Call Syntax (UFCS), which allows free functions to be called using method syntax. This makes your code more readable, especially when chaining operations, and helps create a more consistent API style.

Not Using Range-Based Algorithms

// Anti-pattern: Manual iteration and transformation
int[] doubleValues(int[] arr) {
    int[] result = new int[arr.length];
    foreach (i, v; arr) {
        result[i] = v * 2;
    }
    return result;
}

// Better approach: Use range-based algorithms
import std.algorithm : map;
import std.array : array;

int[] doubleValues(int[] arr) {
    return arr.map!(a => a * 2).array;
}

// Even more complex operations become cleaner
import std.algorithm : filter, map, sum;

int sumOfEvenDoubles(int[] arr) {
    return arr
        .filter!(a => a % 2 == 0)  // Keep only even numbers
        .map!(a => a * 2)          // Double each value
        .sum;                      // Sum the results
}

Use D’s powerful range-based algorithms from std.algorithm and related modules instead of manual iteration and transformation. Range-based algorithms are more expressive, less error-prone, and often more efficient than manual loops.

Excessive Use of Dynamic Arrays

// Anti-pattern: Excessive use of dynamic arrays
void processLargeData() {
    int[] data = new int[1_000_000];  // Allocates on GC heap
    
    // Fill and process data
    foreach (i; 0..data.length) {
        data[i] = i;
    }
    
    // More processing...
}

// Better approach: Use stack arrays or custom allocators when appropriate
void processSmallData() {
    int[100] data;  // Stack allocation, no GC
    
    // Fill and process data
    foreach (i; 0..data.length) {
        data[i] = i;
    }
    
    // More processing...
}

// For larger data, consider custom allocators
import std.experimental.allocator;
import std.experimental.allocator.mallocator;

void processLargeDataWithCustomAllocator() {
    // Use Mallocator instead of GC
    auto data = Mallocator.instance.makeArray!int(1_000_000);
    scope(exit) Mallocator.instance.dispose(data);
    
    // Fill and process data
    foreach (i; 0..data.length) {
        data[i] = i;
    }
    
    // More processing...
}

Be mindful of excessive use of dynamic arrays, which allocate on the garbage-collected heap. For small, fixed-size arrays, use stack arrays. For large arrays or performance-critical code, consider using custom allocators from std.experimental.allocator to have more control over memory management.

Not Using Contract Programming

// Anti-pattern: Manual parameter checking
int divide(int a, int b) {
    if (b == 0) {
        throw new Exception("Division by zero");
    }
    return a / b;
}

// Better approach: Use contracts
int divide(int a, int b)
in {
    assert(b != 0, "Division by zero");
}
do {
    return a / b;
}

// More complex example with pre and post conditions
int[] sort(int[] arr)
in {
    assert(arr !is null, "Array cannot be null");
}
out (result) {
    // Verify the result is sorted
    for (size_t i = 1; i < result.length; i++) {
        assert(result[i-1] <= result[i], "Array not sorted");
    }
}
do {
    import std.algorithm : sort;
    return arr.dup.sort().array;
}

Use D’s contract programming features (in and out blocks) to specify preconditions and postconditions for functions. Contracts make your code more robust by clearly documenting and enforcing expectations about inputs and outputs.

Not Using Proper Error Handling

// Anti-pattern: Poor error handling
void processFile(string filename) {
    auto file = File(filename, "r");  // Might throw if file doesn't exist
    auto content = file.readln();     // Might throw on I/O error
    // Process content...
}

// Better approach: Use proper error handling
import std.exception : collectException;

bool processFile(string filename, out string errorMessage) {
    try {
        auto file = File(filename, "r");
        auto content = file.readln();
        // Process content...
        return true;
    } catch (Exception e) {
        errorMessage = e.msg;
        return false;
    }
}

// Alternative: Use Nullable or std.typecons.Tuple for return values
import std.typecons : Tuple, tuple;
import std.typecons : Nullable;

Tuple!(bool, string) processFile(string filename) {
    auto e = collectException!Exception({
        auto file = File(filename, "r");
        auto content = file.readln();
        // Process content...
    }());
    
    if (e) {
        return tuple(false, e.msg);
    }
    return tuple(true, "");
}

Implement proper error handling in your code. D uses exceptions for error handling, but you should consider using return values or std.typecons types like Nullable or Tuple to represent success/failure states when appropriate, especially for operations that might reasonably fail.

Not Using Templates Effectively

// Anti-pattern: Duplicated code for different types
int findMaxInt(int[] arr) {
    if (arr.length == 0) throw new Exception("Empty array");
    int max = arr[0];
    foreach (v; arr[1..$]) {
        if (v > max) max = v;
    }
    return max;
}

double findMaxDouble(double[] arr) {
    if (arr.length == 0) throw new Exception("Empty array");
    double max = arr[0];
    foreach (v; arr[1..$]) {
        if (v > max) max = v;
    }
    return max;
}

// Better approach: Use templates
T findMax(T)(T[] arr) {
    if (arr.length == 0) throw new Exception("Empty array");
    T max = arr[0];
    foreach (v; arr[1..$]) {
        if (v > max) max = v;
    }
    return max;
}

// Even better: Use constraints and std.algorithm
import std.algorithm : max, reduce;
import std.traits : isOrderingComparable;

T findMax(T)(T[] arr)
if (isOrderingComparable!T) {
    if (arr.length == 0) throw new Exception("Empty array");
    return arr.reduce!max;
}

Use templates effectively to create generic code that works with multiple types. Add constraints using static if and template constraints to ensure type safety and provide clear error messages when templates are misused.

Not Using Proper Memory Management

// Anti-pattern: Memory leaks with manual memory management
class Resource {
    private void* _data;
    
    this() {
        _data = malloc(1024);  // Allocate memory
    }
    
    // No destructor to free memory
}

// Better approach: Implement proper cleanup
class Resource {
    private void* _data;
    
    this() {
        _data = malloc(1024);  // Allocate memory
    }
    
    ~this() {
        if (_data) {
            free(_data);  // Free memory
            _data = null;
        }
    }
}

// Even better: Use RAII with scope(exit)
void processResource() {
    void* data = malloc(1024);
    scope(exit) free(data);  // Will be called when function exits
    
    // Use data...
    // No need to manually call free at each return point
}

Implement proper memory management, especially when using manual memory allocation. Use destructors, the scope statement, or RAII (Resource Acquisition Is Initialization) patterns to ensure resources are properly cleaned up.

Not Using Immutability

// Anti-pattern: Excessive mutability
struct Point {
    int x;
    int y;
}

void processPoint(Point p) {
    // p can be modified, even if that's not intended
    p.x += 10;  // Oops, unintended modification
}

// Better approach: Use immutability
void processPoint(const Point p) {
    // p cannot be modified
    // p.x += 10;  // Compilation error
}

// For class references
void processObject(const Object obj) {
    // obj's fields cannot be modified
}

// For deeply immutable data
void processData(immutable int[] data) {
    // data cannot be modified at all
}

Use immutability (const, immutable) to make your code more robust and easier to reason about. Mark function parameters as const when they shouldn’t be modified, and use immutable for data that should never change.

Not Using Proper Modules

// Anti-pattern: Everything in one module
// myapp.d
module myapp;

// Hundreds of classes, functions, and variables all in one file

// Better approach: Organize code into modules
// myapp/package.d
module myapp;

public import myapp.core;
public import myapp.utils;
public import myapp.models;

// myapp/core.d
module myapp.core;

// Core functionality

// myapp/utils.d
module myapp.utils;

// Utility functions

// myapp/models.d
module myapp.models;

// Data models

Organize your code into proper modules instead of putting everything in one file. This improves maintainability, compilation times, and allows for better encapsulation of implementation details.

Not Using Proper Documentation

// Anti-pattern: Poor or no documentation
int calculate(int a, int b) {
    return a * b + 42;
}

// Better approach: Use proper documentation comments
/**
 * Calculates a special value based on two inputs.
 *
 * The calculation multiplies the inputs and adds a magic number.
 *
 * Params:
 *     a = First input value
 *     b = Second input value
 *
 * Returns: The calculated result
 *
 * Throws: Nothing
 *
 * Examples:
 * ---
 * assert(calculate(2, 3) == 48);
 * ---
 */
int calculate(int a, int b) {
    return a * b + 42;
}

Use proper documentation comments for your code. D supports DDoc, a documentation generator that processes specially formatted comments to create documentation. Include information about parameters, return values, exceptions, and examples to make your code more accessible to others.

Not Using Unit Tests

// Anti-pattern: No unit tests
int add(int a, int b) {
    return a + b;
}

// Better approach: Include unit tests
int add(int a, int b) {
    return a + b;
}

unittest {
    assert(add(2, 3) == 5);
    assert(add(-1, 1) == 0);
    assert(add(0, 0) == 0);
    
    // Test edge cases
    import std.math : isClose;
    assert(isClose(add(int.max, 1), int.min));  // Overflow
}

Include unit tests in your code using D’s built-in unittest blocks. Unit tests help ensure your code works correctly and continues to work as you make changes. They also serve as executable documentation showing how your code is intended to be used.

Not Using Properties

// Anti-pattern: Direct field access
class Person {
    string name;  // Public field
    int age;      // Public field
}

// Better approach: Use properties
class Person {
    private string _name;
    private int _age;
    
    @property string name() const { return _name; }
    @property void name(string value) { _name = value; }
    
    @property int age() const { return _age; }
    @property void age(int value) {
        if (value < 0) throw new Exception("Age cannot be negative");
        _age = value;
    }
}

// Usage remains simple
auto person = new Person();
person.name = "John";  // Calls the setter
auto name = person.name;  // Calls the getter

Use properties (getters and setters) instead of public fields for class and struct members that need validation or might change implementation in the future. Properties allow you to encapsulate the internal representation while providing a simple interface.

Not Using Proper Concurrency Patterns

// Anti-pattern: Unsafe shared data access
shared int counter = 0;

void incrementCounter() {
    import core.thread : Thread;
    
    auto threads = new Thread[10];
    foreach (i; 0..10) {
        threads[i] = new Thread({
            foreach (j; 0..1000) {
                counter++;  // Race condition!
            }
        });
        threads[i].start();
    }
    
    foreach (t; threads) {
        t.join();
    }
}

// Better approach: Use proper synchronization
import core.atomic : atomicOp;

shared int counter = 0;

void incrementCounter() {
    import core.thread : Thread;
    
    auto threads = new Thread[10];
    foreach (i; 0..10) {
        threads[i] = new Thread({
            foreach (j; 0..1000) {
                atomicOp!"+="(counter, 1);  // Atomic operation
            }
        });
        threads[i].start();
    }
    
    foreach (t; threads) {
        t.join();
    }
}

// Even better: Use message passing
import std.concurrency;

void counterThread() {
    int counter = 0;
    
    while (true) {
        receive(
            (string msg) {
                if (msg == "increment") {
                    counter++;
                } else if (msg == "get") {
                    send(ownerTid, counter);
                } else if (msg == "quit") {
                    break;
                }
            }
        );
    }
}

void useCounter() {
    auto tid = spawn(&counterThread);
    
    foreach (i; 0..1000) {
        send(tid, "increment");
    }
    
    send(tid, "get");
    auto count = receiveOnly!int();
    writeln("Counter: ", count);
    
    send(tid, "quit");
}

Use proper concurrency patterns when writing multithreaded code. Prefer message passing using std.concurrency over shared mutable state. When shared state is necessary, use proper synchronization mechanisms like atomic operations, mutexes, or reader-writer locks.

Getting Started

AI Coding Agents

Axon Models

AI Code Reviews

Integrations

D