862276 - [meta] reduce costs from descendants of simple dynamic style changes (e.g., changing 'transform' in style attribute)

Status: RESOLVED FIXED

(Core :: CSS Parsing and Computation, defect, P3)

Description

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

A number of the performance issue in Firefox OS related to style code are related to the cost of relatively simple dynamic style changes (such as changing 'transform' in a style attribute) that's the result of assigning new style contexts to the element's descendants.

Bug 779968 did some previous work to improve this case.

Two ideas on further improvements here are in bug 804975 and bug 862274.  I need to figure out which, if any, of these ideas are practical and how to move forward.

Comment 1

[Benoit Girard (:BenWa)]

I listed a few problems in bug 939460 where style flushes are preventing us from having smooth animations. Ideally this bug will be sufficient to close all of these. Given this I think this should be a harder blocker for 1.3.

Comment 2

[Milan Sreckovic [:milan] (needinfo for best results)]

This is indeed critical.

Comment 3

[Milan Sreckovic [:milan] (needinfo for best results)]

David, this came up in the rendering meeting, we talked about it a bit.  We think it's a blocker for 1.3 (Dec 9) - can you help us figure out how to handle this?

Comment 4

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I think bug 931668 (already on dependency list) should make a big dent here, and heycam is working on that.

Comment 5

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

(And by "make a big dent here", I mean I think it will be sufficient to address the problems we're seeing.)

Comment 6

[Milan Sreckovic [:milan] (needinfo for best results)]

Thanks for the quick response, and I agree, that's a very big dent :).  I will roll this back to 1.3?, as we would want to get the results from bug 931668, and hopefully we won't need the rest for 1.3.

Comment 7

[Mike Lee [:mlee]]

Minusing from 1.3 per Milan's comment 6. Renom if necessary after bug 931668 lands.

(In reply to Milan Sreckovic [:milan] from comment #6)
> Thanks for the quick response, and I agree, that's a very big dent :).  I
> will roll this back to 1.3?, as we would want to get the results from bug
> 931668, and hopefully we won't need the rest for 1.3.

Comment 8

[Mike Lee [:mlee]]

FxOS Perf P1 as this has a direct negative impact on Frame Uniformity.

Comment 9

[Kevin Grandon :kgrandon]

We need this in order to make some of the new FxOS homescreen animations perform well.

Comment 10

[Geo Mealer [:geo] -- This account is inactive after 2015-07-07]

We shouldn't block on the meta--please nom the dependent bugs as necessary.

Comment 11

[Kevin Grandon :kgrandon]

I'll nom as appropriate. Thanks!

Comment 12

[Kevin Grandon :kgrandon]

This is super nice to have, but I'm guessing this won't make 2.0.

Comment 13

[Benoit Girard (:BenWa)]

If bug 862274 is a dupe of bug 931668 then there should only be bug 804975 left. We should try to wrap it up ASAP since this bug is causing significant delays all over FFOS.

Comment 14

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Bug 1180120 could also help here.  Bug 1180118 is less directly related to comment 0 but could also help.

Comment 15

[Sylvestre Ledru [:Sylvestre]]

Moving to p3 because no activity for at least 1 year(s).
See https://github.com/mozilla/bug-handling/blob/master/policy/triage-bugzilla.md#how-do-you-triage for more information

Comment 16

[Sylvestre Ledru [:Sylvestre]]

Moving to p3 because no activity for at least 1 year(s).
See https://github.com/mozilla/bug-handling/blob/master/policy/triage-bugzilla.md#how-do-you-triage for more information

Comment 17

[Gregory Pappas [:gregp]]

Inactive metabug