Do timestamps get implicitly casted to another timezone?

[Abdullahsab3]

While I was looking into #10308 , I noticed in this query:

select 
    date_bin(interval '1 hour', time at time zone 'Europe/Brussels') 
from 
    raw_data 
where 
    time >= '2021-03-27T22:00:00.000Z' 
and 
    time <= '2021-03-29T00:00:00.000Z'
-- exclude the daylight savings hour
and time not between '2021-03-28T02:00:00Z' and '2021-03-28T03:00:00Z'
group by date_bin(interval '1 hour', time at time zone 'Europe/Brussels') 

that the timestamps in the where clause seem to be interpreted as a timestamp with 'Europe/Brussels' timezone, which means that the Z seems to be ignored. The and time not between '2021-03-28T02:00:00Z' and '2021-03-28T03:00:00Z' clause, for instance, refers to the daylights saving hour in local time, while it should be referring to the hour between 2021-03-28T02:00:00 UTC and 2021-03-28 03:00:00 UTC.

Maybe I misunderstood something there 🤔

[alamb]

I think the timestamps may be interpreted as "None" which then has special rules for casting

You can see the actual types using the arrow_typeof function

> select '2021-03-27T22:00:00.000Z'::timestamp;
+----------------------------------+
| Utf8("2021-03-27T22:00:00.000Z") |
+----------------------------------+
| 2021-03-27T22:00:00              |
+----------------------------------+
1 row(s) fetched.
Elapsed 0.011 seconds.

> select arrow_typeof('2021-03-27T22:00:00.000Z'::timestamp);
+------------------------------------------------+
| arrow_typeof(Utf8("2021-03-27T22:00:00.000Z")) |
+------------------------------------------------+
| Timestamp(Nanosecond, None)                    |
+------------------------------------------------+
1 row(s) fetched.
Elapsed 0.002 seconds.

[alamb]

Filed #10344 to try and document the current behavior