Closed Bug 935680 Opened 6 years ago Closed 21 days ago
[meta] New Firefox responsiveness benchmark
We need a test dedicated to measuring and reporting Firefox responsiveness (jank) during common user operations.
This is for chrome operations, not necessarily while browsing a web page? I want to make sure we don't waste energy on bug 938644 or this. Can you spell out what measurements you would like to see which the google web latency benchmark doesn't cover? Would you find value in adding the web latency benchmark to talos given the fact that there is a need for this bug?
Right now, chrome and web page are pretty much the same as they both lag and jank together (and each affecting the other) on main thread event queue. The exceptions are OMT actions (future scroll, some animations, video, etc). So as long as our chrome runs in the same thread as the web page, and our architecture doesn't change (e.g. priority queues with different priorities for chrome/content - which I think are not on the table right now), then it should cover chrome well enough as well. And if it works right now, then maybe we should first try to land bug 938644, and only if it doesn't satisfy us, we'll try to expand it or add our own test.
Would E10s change matters? (chrome/content separation)
(In reply to Emanuel Hoogeveen [:ehoogeveen] from comment #3) > Would E10s change matters? (chrome/content separation) Probably, but I'd imagine that the tests could be adapted to also measure chrome UI.
Clearing needinfo, we're not going to get to this in Q1 '15 The intention here was to create a test that scripts some "typical" user actions and measure browser responsiveness. The measurement would be very similar to what google's latency benchmark does, but the test would be doing both UI interactions & loading content pages
Flags: needinfo?(dolske) → needinfo?(vladan.bugzilla)
Status: NEW → RESOLVED
Closed: 21 days ago
Resolution: --- → INVALID
You need to log in before you can comment on or make changes to this bug.