rich text diffs: using plaintext diffs + editing actions

[garyfeng]

To implement and test the following scheme:

• to track the editing action, e.g., "bold" + extendedInfo = {sPos, ePos}
• plus changes in the innerText, in which case will be no change.

This is different from doing diffs of innerHTML contents. Which is better?