>> I noticed some traffic about UTF8 as a connection character set a
>> while back, so is this a known problem? And if so:
>>
>> - Is it in the client or the server?
>> - Has it been fixed in 2.0.1?
>
> Just to be sure that this was considered - starting with FB 2.x the
> UTF8 character is now 4-byte long, while UNICODE_FSS - only 3. In FB
> 1.x UTF8 was simple alias for UNICODE_FSS and was 3 bytes long. Can it
> be an issue here?
>
> Roman
>
> P.S. I have tried "select (extract(year from current_timestamp) || '')
> as bla from rdb$database" using Jaybird with our own implementation of
> the wire protocol - everything works fine.

The "group by" is important here (and the connect charset is important just indirectly). Try these
SELECT ((EXTRACT(YEAR FROM CURRENT_DATE) || _UTF8'')) FROM rdb$database;
SELECT ((EXTRACT(YEAR FROM CURRENT_DATE) || _UTF8'')) FROM rdb$database GROUP BY 1;
SELECT ((EXTRACT(YEAR FROM CURRENT_DATE) || _WIN1250'')) FROM rdb$database GROUP BY 1;

Ivan