1528775 - Disable automatic scroll handoff from subframes to enclosing scrollframes on desktop touch devices

Status: VERIFIED FIXED

(Core :: Panning and Zooming, defect, P3)

Description

[Cristian Fogel [:cfogel]]

Attachment: scrollCoast.gif

Affected versions

• 65.0.1, 66.0b8, 67.0a1 (2019-02-18)

Affected platforms

• Windows 10;

Steps to reproduce
TouchScreen device needed; zoom in until the scroll bar is enabled in the page/window;

1. Access https://jsbin.com/cokemav/quiet
2. Tap on the screen and swipe down(scroll coasting), a couple of times in the text-area;

Expected result

• When reaching the bottom of the text-area the swipe-scroll stops;

Actual result

• When reaching the bottom of the text-area the page starts scrolling down;

Regression range

• last good: 2016-01-30
• first bad: 2016-01-31
• pushlog URL: https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=54eea211e234217c0faa8c05bf4de9fd3005f5c2&tochange=941033a51983ddec2d99aa9f868a54c0196a4075

Additional notes

• Chrome and Edge don't have this issue;
• attached recording with the issue;
• turning the layout.css.scroll-anchoring.enabled pref off or on has no change in the behaviour.

Comment 1

[Cristian Fogel [:cfogel]]

Updated flags + info.

Comment 2

[Botond Ballo [:botond]]

(In reply to Cristi Fogel [:cfogel] from comment #0)

Actual result

• When reaching the bottom of the text-area the page starts scrolling down;

This is APZ's "immediate handoff" behaviour, and is intentional.

We did disable immediate handoff for Android in bug 1230552. We could do so for desktop as well, but I would like some UX input before doing so.

Comment 3

[Neha Kochar [:neha]]

Bulk change for all regression bugs with status-firefox67 as 'fix-optional' to be marked 'affected' for status-firefox68.

Comment 4

[Botond Ballo [:botond]]

Andreas, could you route this request for UX feedback to an appropriate person? Thanks!

To summarize the request:

• Given a page like https://output.jsbin.com/cokemav/quiet, where the page is scrollable, and there is also a scrollable div on the page
• If you touch-scroll the scrollable div on a desktop touch device, and you reach the end, should scrolling:
  (a) transition immediately to the page (current behaviour); or
  (b) only transition to the page on the next touch gesture (behaviour prior to 2016)?

Comment 5

[Andreas Bovens [:abovens]]

Is there a reason why desktop is behaving differently from Android in this respect? From my limited testing, it seems like (b) is the current behaviour on Chrome/Fenix/Fennec on Android, and on Chrome on Windows. My suggestion would be to align as it does feel a bit strange indeed, especially given the low scroll friction on Windows.

Comment 6

[Botond Ballo [:botond]]

(In reply to Andreas Bovens [:abovens] from comment #5)

Is there a reason why desktop is behaving differently from Android in this respect?

Some back story:

• The code that governs this behaviour was originally written for B2G, where we needed the "immediate handoff" behaviour due to the way B2G's hiding URL bar worked.
• Later that code was ported to all platforms, so all platforms inherited this behaviour.
• In bug 1230552, we changed the behaviour on Android in response to UX feedback, but left other platforms (B2G and desktop) alone.
• Then B2G was cancelled, leaving desktop the only platform with the "immediate handoff" behaviour.

From my limited testing, it seems like (b) is the current behaviour on Chrome/Fenix/Fennec on Android, and on Chrome on Windows. My suggestion would be to align as it does feel a bit strange indeed, especially given the low scroll friction on Windows.

That sounds reasonable to me.

Comment 7

[Botond Ballo [:botond]]

Attachment: Bug 1528775 - Disable immediate handoff on all platforms. r=tnikkel

Comment 8

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/2c35163a7686
Disable immediate handoff on all platforms. r=tnikkel

Comment 9

[Natalia Csoregi [:nataliaCs]]

https://hg.mozilla.org/mozilla-central/rev/2c35163a7686

Comment 10

[Ryan VanderMeulen [:RyanVM]]

Is this something we should consider for backport to Beta for Fx69 or can this ride Fx70 to release?

Comment 11

[Botond Ballo [:botond]]

As the previous behaviour has been in place for a long time, I think we can let this change ride the trains.

Comment 12

[Cristian Fogel [:cfogel]]

Fix verified with 71.0a1 (2019-10-04) and 70.0b12.