Hello.
At Mon, 15 Jun 2026 15:54:24 +0900, Tatsuya Kawata <kawatatatsuya0913(at)gmail(dot)com> wrote in
> While looking at the lock-related code, I noticed that pg_stat_lock
> is the only statistics view whose timing column (wait_time) uses
> bigint. Every other statistics view uses double precision for
> measured-time columns. I do not see a reason for pg_stat_lock to
> differ.
It seems to me that this was intentional. As described in the
documentation, since wait_time is only accumulated for waits longer
than deadlock_timeout, sub-millisecond precision was probably not
considered particularly useful.
Regards,
--
Kyotaro Horiguchi
NTT Open Source Software Center