Last week I found three "bugs" related to the coding of characters / unicode (Py 3). Bugs, that are making impossible to write safe code when manipulating text/strings as Python is supposed to do. Safe code == not broken, nothing to do with a "regression". jmf