In one of the Java classes I am reviewing I see the following code

private oracle.sql.CLOB getCLOB() {

oracle.sql.CLOB xmlDocument = null;

CallableStatement cstmt = null;

ResultSet resultSet = null;

Connection connection = null;

try {

connection = Persistence.getConnection();

cstmt = connection.prepareCall("{call pkg.proc(?,?)}");

cstmt.registerOutParameter(1, OracleTypes.CURSOR);

cstmt.setString(2, id);

cstmt.execute();

resultSet = (ResultSet)cstmt.getObject(1);

if (resultSet.next()) {

xmlDocument = ((OracleResultSet) resultSet).getCLOB(1);

}

} finally {

Persistence.closeAll(resultSet, cstmt, connection);

}

return xmlDocument;

}

The oracle.sql.CLOB that is returned by getCLOB() is read in another method:

private void anotherMethod() {

...

oracle.sql.CLOB xmlDocument = getCLOB();

clobLength = xmlDocument.length();

chunkSize = xmlDocument.getChunkSize();

textBuffer = new char[chunkSize];

for (int position = 1; position <= clobLength; position += chunkSize) {

charsRead = xmlDocument.getChars(position, chunkSize, textBuffer);

outputBufferedWriter.write(textBuffer, 0, charsRead);

}

...

}

I am new to this project and the folks here say this code is working. I don't understand how we can read a CLOB (which, in my understanding, is a reference) after the underlying database connection is closed. What am I missing?

EDIT: Another point to note is that this code is running in an app server. Persistence.getConnection() gets the connection from a data source (most probably with a connection pool). I wonder if the database connection is used after it is returned to the connection pool.

EDIT2: Using the connection after it was returned to the pool might not be the cause. The app server is Oracle's Glassfish server Websphere and I am hoping they would guard against such usage.

解决方案

JDBC driver prefetches LOBs selected into a result set. Read API can use prefetch buffers

without connection. Buffer size specified by oracle.jdbc.defaultLobPrefetchSize parameter with default 4000.