I wrote:
> * Further to Aleksander's point about lack of test coverage for
> the to_char direction, I see from
> https://coverage.postgresql.org/src/backend/utils/adt/formatting.c.gcov.html
> that almost none of the existing roman-number code paths are covered
> today. While it's not strictly within the charter of this patch
> to improve that, maybe we should try to do so --- at the very
> least it'd raise formatting.c's coverage score a few notches.
For the archives' sake: I created a patch and a separate discussion
thread [1] for that.
regards, tom lane
[1] https://www.postgresql.org/message-id/flat/2956175.1725831136@sss.pgh.pa.us