Hi,
When during a query against a DB2 database a statement or lock timeout
occurs, an SQLException is thrown, as expected. However the exception
is not thrown from Statement.executeQuery but rather from
ResultSet.next(), when the cursor moves to the record that could not
be retrieved any more. This behavior seems a bit strange, but JDBC is
not very specific about where such an exception could occur and when
not. DB2 behaves like this and other RDBMS might as well.
Now the problem is that OJB's RsIterator silently swallows all
exceptions thrown from ResultSet.next(). Thus the application receives
a truncated query result and has no indication that something went
awry.
Is there a specific reason not to throw a PersistenceBrokerException
that wraps SQLExceptions from ResultSet.next() in RsIterator? We
patched the code as shown below and everything seemed to work ok:
public synchronized boolean hasNext()
{
try
{
if (!isHasCalledCheck())
{
setHasCalledCheck(true);
setHasNext(getRsAndStmt().m_rs.next());
if (!getHasNext())
{
autoReleaseDbResources();
}
}
}
catch (Exception ex)
{
setHasNext(false);
autoReleaseDbResources();
logger.warn("Calling ResultSet.next() failed", ex);
if(ex instanceof ResourceClosedException)
{
throw (ResourceClosedException)ex;
}
else
{
throw new PersistenceBrokerException("Calling
ResultSet.next() failed", ex);
}
}
if (logger.isDebugEnabled())
logger.debug("hasNext() -> " + getHasNext());
return getHasNext();
}
Gerhard
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
