> Anyway, i dont know if this is a bug in AbstractStatement or a bad
> usage we do.

Good question. I suspect this is not cleanly specified. JDBC specification
uses

int rowCount = stmt.executeUpdate();

or

boolean hasRs = stmt.execute();
if (!hasRs) {
int rowCount = stmt.getUpdateCount();
}

but never

stmt.executeUpdate();
int rowCount = stmt.getUpdateCount();

So, the current implementation is my personal interpretation of the
specification. Can anybody say anything from the experience with other JDBC
drivers/databases?

Roman