Explain what a data race is in the context of multithreaded development.

Data race is a condition where multiple threads or processes simultaneously access the same memory area (variable, object), and at least one of them performs a write, without any synchronization of access.

The consequences of a data race are unpredictable and can include:

• Incorrect variable values.
• Application crashes.
• Other hard-to-detect errors.

Example of a data race in Swift:

// Unprotected counter
var counter = 0

// Function that increments the counter in a thread
func incrementCounter() {
    for _ in 0..<1000 {
        counter += 1 // Here occurs a data race
    }
}

// Creating and starting multiple threads
let queue = DispatchQueue.global()
queue.async {
    incrementCounter()
}

queue.async {
    incrementCounter()
}

// As a result, counter may not be equal to 2000

Ways to prevent data races:

• Locks: NSLock, os_unfair_lock.
• Mutexes: Used with pthread_mutex.
• Semaphores: DispatchSemaphore.
• Concurrent queues with barriers: DispatchQueue(label: "myQueue", attributes: .concurrent) with .barrier flag during writing.
• Synchronized (Objective-C): @synchronized.
• Atomic operations: Using atomic types like AtomicBool.
• Data structures safe for concurrent access: For example, NSCache or specialized libraries.