New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 808685 link

Starred by 1 user

Issue metadata

Status: Assigned
Owner:
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 3
Type: Bug



Sign in to add a comment

Align Input Invalidation for TTI and FMP

Project Member Reported by dproy@chromium.org, Feb 2 2018

Issue description

Currently, input invalidation for TTI is done on the browser side, while input invalidation for TTI is done on the renderer side. This leads to can lead to subtle bugs like  Issue 808536 . Once TTI definition is fully settled, we should unify the input invalidation story to prevent these in the future.

Additionally, if a TTI is invalidated by input today but the input happens before FMP, then FMP detector does not supply an FMP value and the TTI recording status is recorded as FMP not reached even though FMP was not reached because of an input event. Although technically this is working as intended, we should have a separate UMA to have better visibility into how often user input is the root cause of an aborted TTI. 
 

Comment 1 by maxlg@chromium.org, May 16 2018

Ping, any update here?
Owner: dproy@chromium.org
Status: Assigned (was: Available)
dproy@ I'm assigning this to you, you seem like a good owner for this?
Sure, thanks! 

Looks like I made a typo in comment 0. Should read: "Currently, input invalidation for TTI is done on the browser side, while input invalidation for **FMP** is done on the renderer side."
Deep, do you have update here?

Comment 5 by dproy@chromium.org, Jan 17 (6 days ago)

It's still a good idea to do this, but it's low priority. Keeping this as P3.

Sign in to add a comment