Aleksander Alekseev <aleksander(at)timescale(dot)com> writes:
> Due to DST and also changes in local laws, there could be gaps in
> local time [1].
Yup.
> 1. Should this behavior be documented in the 9.9.4. AT TIME ZONE
> section or maybe it's documented elsewhere and I just missed it?
https://www.postgresql.org/docs/current/datetime-invalid-input.html
> 2. Is it possible to detect an impossible timestamptz's for users who
> wants stricter semantics? If there is a way I think it's worth
> documenting as well.
Maybe convert back and see if you get an identical result?
regards, tom lane