iOS and Swift Blog

20 min read

Swift Set: When contains(...) => true Misleads You and Hides a Broken Collection

One changed ID can leave two equal objects in a Swift Set while contains still succeeds. Follow the broken index, investigate hash seeds, and choose stable identity.

Read more →

19 min read

Priority Inversion in Swift: Why DispatchSemaphore Is a Poor Lock

How a medium-priority thread delays a high-priority one through a lock held by a low-priority thread, what happened on Mars Pathfinder, and how to choose between DispatchSemaphore, Mutex, and actor in Swift 6.

Read more →

18 min read

Stop Rebuilding Loading, Success, and Error UI in Every iOS App

Configurable defaults, replaceable state views, and a shared presentation layer for SwiftUI and UIKit.

Read more →

10 min read

Static vs Dynamic Framework Dispatching (Part 3 of 3): Debug, Release, @rpath, and the debug dylib

Diagnose frameworks that load from Xcode but fail in a standalone app. Separate runtime search paths, framework embedding, and Xcode's debug dylib layout.

Read more →

11 min read

Static vs Dynamic Framework Dispatching (Part 2 of 3): SPM vs Xcode frameworks

Inspect static and dynamic libraries built with Swift Package Manager and Xcode. Follow symbols, dependencies, call sites, and fixups in real Mach-O binaries.

Read more →

18 min read

How I Built a Subtitle Pipeline for 106 TV Episodes with Whisper, Groq, and Python

How I built a Python subtitle pipeline that uses Whisper and Groq for transcription, reuses existing subtitles when possible, and falls back across five translation services when it cannot.

Read more →

10 min read

Does Claude Haiku 4.5 Actually Think? Measuring Extended Thinking From the Terminal

A reply in a comment thread said Haiku has no thinking mode, and that this is why it is faster than Sonnet and Opus. So I went and measured it. A reproducible terminal experiment that needs no API key: thinking token counts, a multiplication task with a deterministic answer, why a language model fails arithmetic that a $2 calculator gets right, and an uncomfortable surprise about tool use, where the model reported running a bash command that was never allowed to run.

Read more →