1506949 - Consider using idle queue for RefreshDriver ticks after first contentful paint but before load

Status: RESOLVED FIXED

(Core :: Layout, enhancement, P3)

Description

[Olli Pettay [:smaug][[e-mail]]]

Some work was happening in 
https://bugzilla.mozilla.org/show_bug.cgi?id=1503339

Comment 1

[Olli Pettay [:smaug][[e-mail]]]

And if we do this, perhaps we could remove nglayout.initialpaint.delay

Comment 2

[Olli Pettay [:smaug][[e-mail]]]

Attachment: slower_load_vsync_3.diff

Comment 3

[Olli Pettay [:smaug][[e-mail]]]

Attachment: slower_load_vsync_4.diff

This is quite similar to bug 1339135.

This is also related to bug 1517850, which tries to ensure we get first contentful paint sooner, but after that we should slow down painting rate.

I don't have yet too good numbers, but at least in performance profiles we end up spending less time reflowing/painting the page, which should lead to faster page loads.

Comment 4

[Andreas Farre [:farre]]

Comment on attachment 9037810 [details] [diff] [review]
slower_load_vsync_4.diff

Review of attachment 9037810 [details] [diff] [review]:
-----------------------------------------------------------------

Discussed a bit about if prefs would be helpful or not on irc. I'm ok without.

Comment 5

[Olli Pettay [:smaug][[e-mail]]]

Attachment: slower_load_vsync_5.diff

rebasing. Won't land before 67
https://treeherder.mozilla.org/#/jobs?repo=try&revision=6865ab32d4e79bf591a16160686897935568a133

Comment 6

[Olli Pettay [:smaug][[e-mail]]]

Attachment: slower_load_vsync_5.diff

This might even build.

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/840dbcc8d159
use idle queue for RefreshDriver tick betweek fcp and load, r=farre

Comment 8

[Stefan Hindli [:stefan_hindli]]

https://hg.mozilla.org/mozilla-central/rev/840dbcc8d159