Selection.selectRanges selects contents of a non-editable element

[Piotrek Koszuliński]

1. Go to placeholder sample.
2. Select placeholder.
3. Click bold button.
4. Click "strong" in elements path

• Expected: placeholder should be selected with fake selection.
• Actual: selection is anchored inside widget element.

Elements path uses the selection#selectElement method which does nothing more than selectRanges, so the issue is inside selectRanges or in browsers (less likely).

---

It also concerns IE8, but even more often. In #10887 we needed to disable two tests:

• ​http://ckeditor4.t/dt/core/style/editor.html?name=test-apply-inline-style-on-non-editable-inline-element---at-non-editable-inline-boundary
• ​http://ckeditor4.t/dt/core/style/editor.html?name=test-remove-inline-style-from-non-editable-inline-element---at-non-editable-inline-boundary

[Marek Lewandowski]

Commited to ​t/11042 dev and t/11042 in tests.

The only remark is that with that change, clicking parent elements ( to nested widget ) seems not to respond / be broken.

[Olek Nowodziński]

I rebased dev branch and pushed several (minor) improvements there.

Although the code is just alright, there are several issues in tests:

• Tests should be a part of dt since they touch a very general problem in core.
• Tests should have user-friendly names, unlike "test1".
• Empty <meta name="cktester-plugins" content=""> in head.
• id="outer_p" and class="foo" are obsolete and not utilized.
• Tcs are a little bit obscure:

  <p id="outer_p">aa <strong><span class="foo"><b><i contenteditable="false">bb</i></b></span></strong> cc</p>
  

  Why not:

  <p>aa <strong><i contenteditable="false">bb</i></strong> cc</p>
  

  It's clean, isn't it ? :)
• There are no tests for false positives, i.e.:

  <p>aa <strong>oops there's text inside<i contenteditable="false">bb</i></strong> cc</p>
  

  Perhaps there are other cases on which that fix may fail?

• Error messages should be inversed:

  assert.isTrue( !!sel.isFake, 'Produced selection is not fake' );
  
  to
  
  assert.isTrue( !!sel.isFake, 'Produced selection is fake' );
  

[Marek Lewandowski]

Correct :) html might be a little bit cleaner, because of moving to seleciton.fake, i changed [i][b] tags to strongs/em to avoid config change in selection.fake tests.

in addition: assert message was changed to

assert.isTrue( !!sel.isFake, 'Produced selection should be fake' );

since it's raised when selection is not faked. I did not add extra test cases, since other cases were already handled correctly when i began this ticket.

Tests updated in t/11042.

[Piotrek Koszuliński]

I pushed couple of commits which:

1. simplified tests,
2. added false positive test,
3. shortened dev code.

And I squashed other commits which didn't contain any interesting history.

Also, I checked regressions mentioned in this ticket's description, but those tests were still broken so I created #11242.

[Piotrek Koszuliński]

Fixed on master with ​git:eaa4592 on dev and 631db74 on tests.