Performance pass: timeline, media, and Compose hot paths #85

Open
opened 2026-07-11 21:43:34 +00:00 by starsetbyte · 0 comments
Owner

Goal

Run a focused performance pass across the timeline and media surfaces, consolidating the existing performance plans instead of starting from unmeasured cleanup.

Scope

  • Recomposition and layout work during steady scrolling.
  • Image decode/cache and graphics-layer hotspots.
  • Paging/Room mapping and unnecessary list churn.
  • Baseline profile and macrobenchmark feasibility.

Acceptance criteria

  • Capture a before/after baseline on a named device/emulator with repeatable scenarios.
  • Use profiler or Macrobenchmark evidence to identify the highest-impact changes.
  • Fix only measured regressions/hotspots, with no broad behavior changes hidden in the pass.
  • Add or update performance regression coverage where practical and record remaining device-only work.
## Goal Run a focused performance pass across the timeline and media surfaces, consolidating the existing performance plans instead of starting from unmeasured cleanup. ## Scope - Recomposition and layout work during steady scrolling. - Image decode/cache and graphics-layer hotspots. - Paging/Room mapping and unnecessary list churn. - Baseline profile and macrobenchmark feasibility. ## Acceptance criteria - Capture a before/after baseline on a named device/emulator with repeatable scenarios. - Use profiler or Macrobenchmark evidence to identify the highest-impact changes. - Fix only measured regressions/hotspots, with no broad behavior changes hidden in the pass. - Add or update performance regression coverage where practical and record remaining device-only work.
Sign in to join this conversation.
No description provided.