Bug 194601 – REGRESSION: [ iOS ] Layout Test editing/input/ios/rtl-keyboard-input-on-focus.html is a Timeout

• Component: Tools / Tests
• Status: RESOLVED FIXED

[Truitt Savell]

The following layout test is Timing out on iOS

editing/input/ios/rtl-keyboard-input-on-focus.html

Probable cause:

This test began timing out constantly around r241308. Unable to reproduce locally. 

Flakiness Dashboard:

https://webkit-test-results.webkit.org/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=editing%2Finput%2Fios%2Frtl-keyboard-input-on-focus.html

[Wenson Hsieh]

*** Bug 194650 has been marked as a duplicate of this bug. ***

[Wenson Hsieh]

Failure appears to be in this range: https://trac.webkit.org/log/webkit/?stop_rev=241285&rev=241317&limit=1000

[Wenson Hsieh]

(In reply to Wenson Hsieh from comment #2)
> Failure appears to be in this range:
> https://trac.webkit.org/log/webkit/?stop_rev=241285&rev=241317&limit=1000

A large number of tests in editing/selection/ios (which are run only on Apple-internal bots) also began failing at around this range; these might be related.

[Tim Horton]

SURELY it's https://trac.webkit.org/changeset/241311/webkit

[Wenson Hsieh]

(In reply to Tim Horton from comment #4)
> SURELY it's https://trac.webkit.org/changeset/241311/webkit

My thoughts too :/

That being said, I haven't dug into it quite yet to see why.

[Radar WebKit Bug Importer]

<rdar://problem/48080316>

[Wenson Hsieh]

Created attachment 362478 [details]
Patch

[Wenson Hsieh]

> cleanUpAfterActivityStateChangeDueToBecomingFirstResponder

...definitely taking better suggestions :|

[Wenson Hsieh]

Created attachment 362531 [details]
Patch

[Tim Horton]

Comment on attachment 362531 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=362531&action=review

> Source/WebKit/WebProcess/WebPage/RemoteLayerTree/RemoteLayerTreeDrawingArea.mm:514
> +    for (const auto& callbackID : callbackIDs)
> +        m_webPage.send(Messages::WebPageProxy::VoidCallback(callbackID));

These are going to beat the transaction, which seems bad (maybe not in this case, but for people expecting the callbacks to mean what they mean in the non-UI-side world).

[Wenson Hsieh]

(In reply to Tim Horton from comment #10)
> Comment on attachment 362531 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=362531&action=review
> 
> > Source/WebKit/WebProcess/WebPage/RemoteLayerTree/RemoteLayerTreeDrawingArea.mm:514
> > +    for (const auto& callbackID : callbackIDs)
> > +        m_webPage.send(Messages::WebPageProxy::VoidCallback(callbackID));
> 
> These are going to beat the transaction, which seems bad (maybe not in this
> case, but for people expecting the callbacks to mean what they mean in the
> non-UI-side world).

Fair point! As we discussed on IRC, this is OK for now, but might need to be adjusted in the future for UI-side compositing on macOS.

I'll leave a FIXME for this.

[Wenson Hsieh]

Created attachment 362548 [details]
Patch for landing

[WebKit Commit Bot]

Comment on attachment 362548 [details]
Patch for landing

Clearing flags on attachment: 362548

Committed r241846: <https://trac.webkit.org/changeset/241846>