The toolkit component is still under development and lives in the dom-scheduler repo [1] for now. But we already have a demo app running [2] to start experimenting. We're facing 2 issues to reliably test the scrolling performance: 1. simulating a reproducible scrolling gesture (checked with QA marionette developers and between the marionette overhead and the APZ heuristics scrolling "flings" are apparently far from consistent in marionette tests) 2. detecting checkerboarding / missing content during scrolling Looks like orangutan [3] might help with 1, which would give us a way to get reliable yet manual tests (already an improvement!) And we still need to figure out 2 if we want something fully automated/raptorable :) [1] https://github.com/etiennesegonzac/dom-scheduler [2] https://github.com/etiennesegonzac/dom-scheduler/tree/master/demo-app [3] https://github.com/wlach/orangutan

Firefox OS is not being worked on