Connection charset not being used anymore on fields with charset NONE [CORE6240] #6484
Comments
|
Commented by: @asfernandes The change was introduced as a side effect of CORE5846 and I think it was a correct change. The connection charset applies to literals (not this case) and in the last phase, when converting data to be sent to client. So when you use "field collate ...", the connection charset is not yet applied. Note also the previous behavior was causing problems, see: isql -ch utf8 db.fdb select * from test where without_charset collate unicode_ci_ai = '123'; Statement failed, SQLSTATE = 22001 |
Modified by: @pcisarstatus: Resolved [ 5 ] => Closed [ 6 ] |
Submitted by: Rudi Feijó (rudi.feijo_multidadosti.com.br)
On firebird 3.0.3, if I connected to the db specifying a charset like WIN1252, fields with charset NONE would accept a collation of the charset specified in the connection.
This behavior is missing in 3.0.5, if I try to specify a collation in a query, it will throw a "collation x for charset y is not defined", regardless of the connection charset.
CREATE TABLE TEST (
WITHOUT_CHARSET VARCHAR(10)
);
query : SELECT * FROM TEST WHERE without_charset collate WIN_PTBR CONTAINING('abc')
on firebird 3.0.3, without specifiying connection charset : FAIL (COLLATION WIN_PTBR for CHARACTER SET NONE is not defined)
on firebird 3.0.3, specifiying charset WIN1252 on connection : SUCCESS
on firebird 3.0.5, without specifiying connection charset : FAIL (COLLATION WIN_PTBR for CHARACTER SET NONE is not defined)
on firebird 3.0.5, specifiying charset WIN1252 on connection : FAIL (COLLATION WIN_PTBR for CHARACTER SET NONE is not defined)
The text was updated successfully, but these errors were encountered: