Bug 209025 – REGRESSION (r257739): [ iOS ] fast/events/autoscroll-in-iframe.html is still flaky failing

• Component: Tools / Tests
• Status: RESOLVED CONFIGURATION CHANGED

[Jacob Uphoff]

fast/events/autoscroll-in-iframe.html

This test has been flaky failing for all of iOS since around r257740 

Started off failing more consistently but now is failing more flakily.

History:

https://results.webkit.org/?suite=layout-tests&test=fast%2Fevents%2Fautoscroll-in-iframe.html&platform=ios

Diff:

--- /Volumes/Data/slave/ios-simulator-13-debug-tests-wk2/build/layout-test-results/fast/events/autoscroll-in-iframe-expected.txt
+++ /Volumes/Data/slave/ios-simulator-13-debug-tests-wk2/build/layout-test-results/fast/events/autoscroll-in-iframe-actual.txt
@@ -1,3 +1,4 @@
 
+FAILED: Clicking in the iframe scrolled the page (window.scrollTops is 0)
 PASSED: selecting in the iframe did not scroll the page.

[Radar WebKit Bug Importer]

<rdar://problem/60392238>

[Jacob Uphoff]

Set expectations here: https://trac.webkit.org/changeset/258370/webkit

[Alexey Proskuryakov]

There was an attempt to fix these in r208520, which improved the situation. But it's still failing around half of the time.

[Alexey Proskuryakov]

I meant to say, in bug 208520.

[chris fleizach]

(In reply to Alexey Proskuryakov from comment #4)
> I meant to say, in bug 208520.

if we really think  r257740  is the cause of this, it's likely the reason is that accessibility is being enabled on iOS unexpectedly. that is what I'll look for

[chris fleizach]

(In reply to chris fleizach from comment #5)
> (In reply to Alexey Proskuryakov from comment #4)
> > I meant to say, in bug 208520.
> 
> if we really think  r257740  is the cause of this, it's likely the reason is
> that accessibility is being enabled on iOS unexpectedly. that is what I'll
> look for

Must have been

https://bugs.webkit.org/show_bug.cgi?id=208434

[chris fleizach]

So there was clearly something enabling accessibility previously. That was clearly fixed. Looking through original patch, this is the only thing that does something different.

-    return textMarkerRangeFromVisiblePositions(self.axBackingObject->axObjectCache(), startPosition, endPosition);
+    auto* backingObject = self.updateObjectBackingStore;
+    if (!backingObject)
+        return nil;
+
+    return textMarkerRangeFromVisiblePositions(backingObject->axObjectCache(), startPosition, endPosition);

by calling self.updateObjectBackingStore

[Andres Gonzalez]

(In reply to chris fleizach from comment #7)
> So there was clearly something enabling accessibility previously. That was
> clearly fixed. Looking through original patch, this is the only thing that
> does something different.
> 
> -    return
> textMarkerRangeFromVisiblePositions(self.axBackingObject->axObjectCache(),
> startPosition, endPosition);
> +    auto* backingObject = self.updateObjectBackingStore;
> +    if (!backingObject)
> +        return nil;
> +
> +    return
> textMarkerRangeFromVisiblePositions(backingObject->axObjectCache(),
> startPosition, endPosition);
> 
> by calling self.updateObjectBackingStore

that was fix in patch for bug:

https://bugs.webkit.org/show_bug.cgi?id=208648

so that cannot be the cause if this is still failing.