@JelteF
These changes throw error if the values are out of bounds , which breaks regression test named test_all_types , IIUC it tests the bounds of all data type in duckdb.

...
not ok 9     - test_all_types                             21 ms
....
1..34
# 1 of 34 tests failed.

postgres=# SELECT * FROM duckdb.query($$ select '4714-11-23 (BC)'::date as date $$);
ERROR:  (PGDuckDB/Duckdb_ExecCustomScan_Cpp) Out of Range Error: The value should be between min and max value (4714-11-24 (BC) <-> 5874897-12-31)
postgres=# SELECT * FROM duckdb.query($$ select '4714-11-24 (BC)'::date as date $$);
     date      
---------------
 4714-11-24 BC
(1 row)
postgres=# SELECT * FROM duckdb.query($$ select '5874897-12-31'::date as date $$);
     date      
---------------
 5874897-12-31
(1 row)

postgres=# SELECT * FROM duckdb.query($$ select '5874898-01-01'::date as date $$);
ERROR:  (PGDuckDB/Duckdb_ExecCustomScan_Cpp) Out of Range Error: The value should be between min and max value (4714-11-24 (BC) <-> 5874897-12-31)

If we dont to throw an error we can use the boundary values for out of range values but we should throw a elog(NOTICE that we are doing this or user could get confused)

Also if you have some tips on debugging across C/C++ boundary using gdb (VS code extension seems to not want to work), that would be really appreciated , was trying to walk throught the query lifecycle with pg_duckdb ?

Thanks for the contribution! Some thoughts:

1. We need the same thing for timestamp and timestamptz, not just for date. This could be done in a separate PR if you think that's easier.
2. We should probably also protect conversion the other way around for timestamp and timestamptz, for date it seems like the supported dates in DuckDB are a strict-superset of the supported dates in Postgres.
3. Throwing an error seems the correct thing to do. That's what both DuckDB and postgres do on out-of-bound dates. For the test_all_types test, Just add the necessary types to the exclude list, and then indeed do what you propose: add separate tests to test the bounds correctly.

Also, do you know if this also fixes this issue? #629

If so we should include that in the tests.

Also, do you know if this also fixes this issue? #629

If so we should include that in the tests.

With my changes it throws out of bounds Error

postgres=# select * from duckdb.query($$ select 'infinity'::timestamp, 'infinity'::date $$);
ERROR:  (PGDuckDB/Duckdb_ExecCustomScan_Cpp) Out of Range Error: The value should be between min and max value (4714-11-24 (BC) <-> 5874897-12-31)

With my changes it throws out of bounds Error

Hmm, I guess that's better than returning something wrong. But it would be good if we actually converted infinity/-infinity correctly for dates and timestamps, because both databases support it.

Thanks for the work. This looks almost good, left some mostly cosmetic feedback. CI is also failing due to a compiler error/warning.

Had to use PG_INT64* Macros for timestamp as specific macros used before were introduces in specific majors only. This will also avoid ugly pg version checks...