Remove superfluous bogus BR elements

[sjbufton]

• IE versions below 11 pad empty elements too much if a <br data-mce-bogus="1"> element is automatically added; the empty element has a height of two lines, not one.
• Nested empty elements, e.g. <div class="a"><div class="b"></div></div> were transformed into <div class="a"><div class="b"><br data-mce-bogus="1"></div><br data-mce-bogus="1"></div>; the final <br ...> is unnecessary, and again pads the display too much - on all browsers. The extra parameter "bogusElements" to Node.isEmpty() allows elements marked as "data-mce-bogus" to also count as non-empty, and is used when setting the editor content to ensure that only the innermost empty element is padded in this way.

[spocke]

Do you have a bug case where this is happening? Steps to reproduce? Haven't seen a lot of reports so a bit curious about this patch.

[sjbufton]

It's fairly easy to reproduce: use the 'development' manual test with IE 9 or 10 (11 in the appropriate mode works the same, and makes it more visible in the Developer Tools), and use the JavaScript console to set the content for point 1:

tinymce.editors[0].setContent('<p></p><p>a</p><h1></h1><h1>b</h1><div></div><div>c</div>');

Then use the DOM Explorer (IE11 Developer Tools) to highlight the blocks: the empty elements are significantly taller than the elements with content.

For point two, use IE11 in Edge mode, Firefox, Chrome etc., modify the 'development' page to include extended_valid_elements: 'div[!class]' and set the content as above to <div class="a"><div class="b"></div></div><div class="a"><div class="b">c</div></div>; use the developer tools to compare the heights of the empty and non-empty divs.

[spocke]

Yes, managed to reproduce it. Need to get a CLA signed though in order for me to pull this in. See on the commits that you use a somewhat anonymous email.

[sjbufton]

I thought I had the e-mail address fixed, but it has only set the committer (I rebased my original commits). Sorry. Can you use the committer address for the CLA?

[fyrkant]

I am closing this pull request due to old age and merge conflicts. If you still think this is relevant and something that should be merged, fix the merge conflicts and leave a comment and I can open it up again. Thank you for the contribution!