flutter_startup_metrics 0.1.3

Measure Flutter app startup time and cold start in production: time to initial display (TTID), time to full display (TTFD), and a phase breakdown. Works in release builds.

0.1.3

• Drop a README caveat that described a limitation of our iOS test rig as if it were a property of the package. Apps in the field launch without a debugger and are unaffected; the constraint is documented in CONTRIBUTING, where it is relevant to anyone reproducing the measurements.

0.1.2

• Verified the iOS implementation on physical hardware for the first time (iPhone 8, iOS 16.7.12, release build): cold launch detected, phases contiguous, hostStartup correctly absent on iOS, and no launch falsely excluded as prewarmed. The README no longer says the iOS path is unverified.
• Added example/run_ios_device.sh for reproducing on-device runs, and documented in CONTRIBUTING why such runs cannot produce a trustworthy magnitude: iOS kills a developer-signed app when the debugger detaches, so the launch has to be measured with one attached. That is a test-rig constraint and does not affect apps in the field.

No library code changes.

0.1.1

• Shorten the package description to pub.dev's 180-character limit, which was costing 10 pub points. No code changes.

0.1.0

Initial release.

• Cold-start measurement from OS process start to Flutter's first rasterized frame, using FramePhase.rasterFinishWallTime to bridge the engine's monotonic clock to wall time.
• Contiguous phase breakdown; phases sum exactly to the total.
• Android splits host startup at the first Activity.onCreate via a ContentProvider that needs no setup in the host app.
• Opt-in time to full display via reportFullyDisplayed(), with a 30-second deadline so an abandoned call resolves rather than hangs.
• Launches that cannot be measured honestly — prewarmed, background-started, implausibly long, or with an out-of-order timeline — are dropped with a reason rather than clamped.