WWDC 2019 Platform State of the Union


  • Less code, looks crisp
  • Declarative, compositional
  • Animations built-in, and interruptable
  • Shareable UI code across platforms
  • UI to data bindings
  • SwiftUI tutorials

Xcode 11

  • Live preview (code left, preview right) that allows multiple simulator types in a scroll view
  • Edit UI code with popups that edit the code
  • Swift Packages supported, top-level in workspace
  • Mini map (hold Cmd while scrolling to see Mark’s)
  • See code changes before/after using left-gutter
  • Test plans to combine all test configurations into a single place, into a single command - capture screenshots, localization details, device combinations, works great with Xcode Server and parallel testing
  • Device conditions – set various conditions, network throughput, thermal states, etc
  • Aggregated metrics from prod apps
  • TestFlight screenshots in App Store Connect


  • iPadOS app migrated to macOS app almost automatically


  • Push notifications can go directly to Watch app, without going through iOS
  • Account creation with text fields and Sign in with Apple
  • Streaming Audio APIs (previously had background audio for local files)
  • Extended Runtime API to give apps to stay running after user lowers wrist
  • App Store with direct downloads to watch

iOS 13

  • Cards as default full-screen presentation. Dismissible with a single downward swipe.
  • Peek and pop is more concise
  • SF Symbols, 1500, expressive of fonts, scales with dynamic type font, adjustable weights like fonts, searchable in Xcode and standalone Symbols app
  • UI Lifecycle split out from App Delegate into separate Scene Delegate in order to handle multiple windows
  • Accessibility with full voice control
  • Sign in with Apple, build-in 2FA, cross-platform, anti-fraud

Machine Learning

  • Text recognition in docs
  • Word embeddings available
  • On-device speech w/10 languages
  • Speech saliency (pitch, etc)
  • CoreML models in app can be updated with user-specific data. Ie. Personalized CoreML model.
  • Add to Siri, no longer needs user to speak the phrase