On 13.11.21 06:40, Thomas Munro wrote:
Yes, it is a valid codepoint, but unfortunately PostgreSQL can't support it because it sometimes deals in null terminated string, even though internally it does track string data and length separately. We have to do that to use libc facilities like strcoll_r(), and probably many other things.
By the way, I think the server-side issues around this are fixable. The real problem in my mind is how to present such values in client APIs such as libpq.
