While building a lightweight data model for a settings screen, I needed to decide how to store user preferences.

Choose the appropriate type—struct or class—to represent the model while ensuring safe data handling.

I explained that structs are value types copied on assignment, providing immutability and thread‑safety, whereas classes are reference types that support inheritance and shared mutable state. I chose a struct because the settings data is simple, immutable after creation, and does not require inheritance.

The app behaved predictably, with no unexpected side‑effects from shared references, and memory usage stayed low.

In a networking layer I received JSON that might omit certain fields.

Safely unwrap the optional values before using them to avoid force‑unwrap crashes.

I used a guard let to unwrap the optional URL string, returning early with an error if nil. This ensured the subsequent code operated on a non‑optional value.

The app no longer crashed on malformed responses; errors were gracefully handled and logged.

The app displayed a feed of articles with thumbnail images fetched from a remote server.

Create a smooth scrolling table view where each cell shows the article title and loads its image without blocking the UI.

I subclassed UITableViewCell, added an UIImageView, and used URLSession data tasks inside a dedicated ImageLoader singleton that caches images. In cellForRowAt I called the loader’s fetch method, which returned the cached image or started a download. I ensured the cell’s image view was reset in prepareForReuse and used a weak reference to avoid race conditions when cells were reused.

Scrolling remained fluid, images appeared progressively, and network usage was reduced thanks to caching.

I was refactoring a news app that performed multiple dependent API calls (fetch token, then fetch articles).

Replace the callback‑heavy URLSession code with a declarative Combine pipeline that handles errors and cancellation.

I created a NetworkService that returned AnyPublisher<Data, URLError>. Using flatMap I chained the token request publisher to the articles request publisher, applied decode operators, and used receive(on: DispatchQueue.main) to update the UI. I also stored the cancellables in a Set to manage lifecycle and used .catch to provide fallback data.

The code became more readable, error handling was centralized, and UI updates responded instantly to data changes. Unit tests could now mock publishers easily.