webdriver: Dispatch embedder TouchEventType::Move#41793
webdriver: Dispatch embedder TouchEventType::Move#41793yezhizhen wants to merge 3 commits intoservo:mainfrom
TouchEventType::Move#41793Conversation
TouchEventType::MoveTouchEventType::Move properly
Signed-off-by: Euclid Ye <[email protected]>
Signed-off-by: Euclid Ye <[email protected]>
12ec98a to
6744b10
Compare
|
🔨 Triggering try run (#20848500909) for Linux (WPT) |
TouchEventType::Move properly TouchEventType::Move
|
Test results for linux-wpt from try job (#20848500909): Flaky unexpected result (38)
Stable unexpected results that are known to be intermittent (21)
|
|
✨ Try run (#20848500909) succeeded. |
|
Emm this isn't right. This PR has improvements and no regression at all, but some of these TIMEOUT are not plausible. |
| point, | ||
| )); | ||
| if last { | ||
| self.send_blocking_input_event_to_embedder(input_event); |
There was a problem hiding this comment.
Ok. The TIMEOUT is a more general problem: TouchMove is not guaranteed to be sent to constellation, and we really shouldn't block above, due to below:
servo/components/compositing/webview_renderer.rs
Lines 454 to 455 in f0233c3
Signed-off-by: Euclid Ye <[email protected]>
|
🔨 Triggering try run (#20851194916) for Linux (WPT) |
|
Test results for linux-wpt from try job (#20851194916): Flaky unexpected result (38)
Stable unexpected results that are known to be intermittent (26)
Stable unexpected results (57)
|
|
|
|
I will open a new PR: And I still want to keep a reference to this obsolete version. |
Fixes: #41725
Fixes: #41620
Testing: New passing in
/dom/events/non-cancelable-when-passive/non-passive-touchmove-event-listener-on-body.html/dom/events/non-cancelable-when-passive/non-passive-touchmove-event-listener-on-div.html/dom/events/non-cancelable-when-passive/non-passive-touchmove-event-listener-on-document.html/dom/events/non-cancelable-when-passive/non-passive-touchmove-event-listener-on-root.htmlCrash Explanation: It is not CRASH, but actually no response.
These tests never succeeded, mostly due to missing pointer events implementation in Servo. This is unlike #41067, where we actually panic. This is another proof that we got it right this time, as we never touchmove without active point now:
servo/components/compositing/touch.rs
Line 436 in 271c859