I could imagine that the collation order and/or other characterset related things can happen if one connection is configured differently from the other. For example, if Latin-1 data is inserted, the database needs to be told in some way that the data should be interpreted as Latin-1.

In absence of the relevant statements and/or the relevant data and error causes though, this is all idle speculation.

Thanks for the answers. My query is selecting records on Oracle 11G database, based on some attributes and timestamp. I know that it's very strange, infact also comparing SQL developer and TOAD oracle client they result (same query) different output. I don't know how this difference could happen, but at the moment i would like to try changing connection string from DBI to Java one if exist (i would like to set into perl script a new database connection like sql developer client do). Thanks. Regards, Fabrizio

Maybe you can show us the SQL command(s) and examples of where/how the results differ between Java and TOAD and Perl.

Note that both, ODBC and JDBC have ways of tracing their commands so maybe you can find a difference there.

Are you using DBD::ODBC or DBD::Oracle to connect to your database?