1914560 - Failing WPTs in /pointerevents/coalesced_events_attributes.https.html?mouse

Status: RESOLVED FIXED

(Core :: DOM: UI Events & Focus Handling, defect)

Description

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

See https://wpt.fyi/results/pointerevents/coalesced_events_attributes.https.html%3Fmouse.

Comment 1

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Attachment: Bug 1914560 - Make synthesized events have coalesced events if they are required at least one r=smaug!,#dom-core

Currently, we set coalesced events only when eMouseMove or eTouchMove
is received in a remote process. Therefore, synthesized events for tests and
ePointerMove event which is converted from non-eMouseMove event won't have
coalesced events.

This patch makes the synthesizer methods set the coalesced event as the event
itself to emulate exactly same data at handling user input. Additionally,
make PointerEventHandler::DispatchPointerFromMouseOrTouch guarantees it
even in the parent process for the consistency.

Comment 2

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Attachment: Bug 1914560 - Make `PointerEvent::GetCoalescedEvents()` ensure one event if trusted `ePointerMove` r=smaug!,#dom-core

Currently, we set coalesced events only when eMouseMove or eTouchMove
is received in a remote process. Therefore, synthesized events for tests and
ePointerMove event which is converted from non-eMouseMove event won't have
coalesced events.

This patch makes PointerEvent::GetCoalescedEvents() ensure at least one
coalesced event if it's a trust event, a pointermove event and its widget
event does not have coalesced events.

Ideally, we should fix this at synthesizer and PointerEventHandler to make
consistent data between user initiated events and synthesized events for tests
to check the behavior exactly. However, it requires much more complicated
patch and C++ default handlers shouldn't refer the coalesced events.
Therefore, this patch touches the path in PointerEvent::GetCoalescedEvents().

Comment 3

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/c9eea28a07a6
Make `PointerEvent::GetCoalescedEvents()` ensure one event if trusted `ePointerMove` r=smaug

Comment 4

[Atila Butkovits]

Backed out for causing failures at coalesced_events_attributes_under_load.https.

Backout link: https://hg.mozilla.org/integration/autoland/rev/43cf0e81627ee5f53e061f2d7ecfadb4fa06b2f6

Push with failures: https://treeherder.mozilla.org/jobs?repo=autoland&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel&revision=c9eea28a07a66b5985df0eeba9b3abfcd2c3e77b&selectedTaskRun=TMUQ8RdUTWWievixsFDCzw.0

Failure log: https://treeherder.mozilla.org/logviewer?job_id=472044302&repo=autoland&lineNumber=9900

Comment 5

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Hmm, bad timing. The test was updated recently. I'll retry to land it tomorrow after it comes m-c.

Comment 6

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

The patch makes PointerEvent.getCoalescedEvents() slower since synthesized events have not been initialized correctly before applying the patch. Additionally, the timeout itself is reported in bug 1915342. So, the slowness with this patch just increased the frequency. I'll make the manifest allow timeout if it's mac-opt build.

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/bc0138b53517
Make `PointerEvent::GetCoalescedEvents()` ensure one event if trusted `ePointerMove` r=smaug

Comment 8

[Serban Stanca [:SerbanS]]

https://hg.mozilla.org/mozilla-central/rev/bc0138b53517