-
Mikaël Salson authored
See #3097 Discussed with @magiraud and @RyanHerb: The UTF-8 is now properly handled when processing JSON. An example of that is that runs and sets don't need safe_encoding. However there is still a small issue (that pre-existed): in anon mode, the three first bytes are recovered which may not correspond to a valid character.
0dd7e228