EventHandler::updateSelectionForMouseDownDispatchingSelectStart should not use an orphaned selection

[Ryosuke Niwa]

Fix what appears to be the root/primary cause of the bug 221786 and add a test.

[Ryosuke Niwa]

Created attachment 420410 [details]
Patch

[Ryosuke Niwa]

Created attachment 420414 [details]
Fixed release builds

[Wenson Hsieh]

Comment on attachment 420414 [details]
Fixed release builds

View in context: https://bugs.webkit.org/attachment.cgi?id=420414&action=review

r=me (assuming EWS is green)

> LayoutTests/editing/selection/click-selection-with-selectstart-node-removal.html:32
> +    const boundingRect = editor.getBoundingClientRect();
> +    const x = boundingRect.x + 10;
> +    const y = boundingRect.y + 10;
> +    UIHelper.activateAndWaitForInputSessionAt(x, y);

Nit - can we just `UIHelper.activateElement(editor);` here?

[Ryosuke Niwa]

(In reply to Wenson Hsieh from comment #3)
> Comment on attachment 420414 [details]
> Fixed release builds
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=420414&action=review
> 
> r=me (assuming EWS is green)
> 
> > LayoutTests/editing/selection/click-selection-with-selectstart-node-removal.html:32
> > +    const boundingRect = editor.getBoundingClientRect();
> > +    const x = boundingRect.x + 10;
> > +    const y = boundingRect.y + 10;
> > +    UIHelper.activateAndWaitForInputSessionAt(x, y);
> 
> Nit - can we just `UIHelper.activateElement(editor);` here?

Hm... maybe? Does WebKitTestRunner knows to gracefully dismiss the keyword if it starts coming up?

[Wenson Hsieh]

(In reply to Ryosuke Niwa from comment #4)
> (In reply to Wenson Hsieh from comment #3)
> > Comment on attachment 420414 [details]
> > Fixed release builds
> > 
> > View in context:
> > https://bugs.webkit.org/attachment.cgi?id=420414&action=review
> > 
> > r=me (assuming EWS is green)
> > 
> > > LayoutTests/editing/selection/click-selection-with-selectstart-node-removal.html:32
> > > +    const boundingRect = editor.getBoundingClientRect();
> > > +    const x = boundingRect.x + 10;
> > > +    const y = boundingRect.y + 10;
> > > +    UIHelper.activateAndWaitForInputSessionAt(x, y);
> > 
> > Nit - can we just `UIHelper.activateElement(editor);` here?
> 
> Hm... maybe? Does WebKitTestRunner knows to gracefully dismiss the keyword
> if it starts coming up?

It should, yes. Though to be totally safe, you can do something like:

    await UIHelper.activateElementAndWaitForInputSession(editor);

...instead.

[Ryosuke Niwa]

Created attachment 420420 [details]
Patch for landing

[Ryosuke Niwa]

Comment on attachment 420420 [details]
Patch for landing

Wait for EWS

[Ryosuke Niwa]

Comment on attachment 420420 [details]
Patch for landing

Clearing flags on attachment: 420420

Committed r272928: <https://commits.webkit.org/r272928>

[Ryosuke Niwa]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/74405933>