[macOS/Safari] Test "EditorView selection / syncs the DOM selection with the editor selection"

[FrankReh]

Issue might be better labeled, inconsistent return of getSelection().

Here is an old w3c/editing task force issue that seems to be indicating the same thing: w3c/editing#156

In the CM6 unit test in question, the call to window.getSelection() returns the text node rather than the div node. The test passes for Chrome and Firefox on my macOS system.

The unit test actually tests seven different selections and by commenting out the rest so just one at a time is tried, one can see four are returning the text node itself rather than the expected text node's parent. Those four are shown with an extra arrow comment here.

test(0, abc.parentNode, 0); // <-- fails Safari
test(1, abc, 1);
test(2, abc, 2);
test(3, abc.parentNode, 1); // <-- fails Safari
test(4, cm.contentDOM.childNodes[1], 0);
test(5, def.parentNode, 0); // <-- fails Safari
test(6, def, 1);
test(8, def.parentNode, 1); // <-- fails Safari

In fact, if window.getSelection() is called immediately after the test's requireFocus() call to setup things, the difference can also be seen.

[marijnh]

I'd say I wrote the test in a too restrictive way and it should use something similar to isEquivalentPosition in view/src/dom.js to test this.

[marijnh]

Does attached patch make the test pass?

[FrankReh]

Yes it does. Nice.