Jim Mathies [:jimm] Reporter
	Description • 8 years ago

We have a couple bugs out there related to apz regression in tp5o_scroll.

Bug 1179735 - [e10s] tp5o_scroll regression on windows compared to non-e10s

This bug is where initial diagnosis took place that pointed to apz as being responsible for a regression in tp5o_scroll. This bug also points out the some of the regression occurs in scroll operations that do not use apz scrolling.

Bug 1253860 - Add a lightweight mechanism to update APZ scroll offset without a full repaint

Current work taking place to address the non-apz scroll issues. Also bug 1253860, comment 9 discusses how apz invalidates this test to some degree in terms of what it represents when apz is turned on.

Independent of all this, e10s appears to have regressed tp5o_scroll even when apz is turned off. 

https://wiki.mozilla.org/Electrolysis/Release_Criteria#Scrolling

We need to figure out how serious this regression is and if it matters if we turn apz on.

	Jim Mathies [:jimm] Reporter
	Comment 1 • 8 years ago

with apz off I'm seeing improvements in tp5o_scroll between non-e10s and e10s.

https://treeherder.mozilla.org/#/jobs?repo=try&author=jmathies@mozilla.com&selectedJob=18169960

The apz regressions are gone as well.