JDBC Interview Questions And Answers _ 5

[ganeshbala]

DB2 Universal claims to support JDBC 2.0, But I can only get JDBC 1.0 functionality. What can I do?
DB2 Universal defaults to the 1.0 driver. You have to run a special program to enable the 2.0 driver and JDK support. For detailed information, see Setting the Environment in Building Java Applets and Applications. The page includes instructions for most supported platforms.

How do I disallow NULL values in a table?
Null capability is a column integrity constraint, normally applied at table creation time. Note that some databases won't allow the constraint to be applied after table creation. Most databases allow a default value for the column as well. The following SQL statement displays the NOT NULL constraint:
CREATE TABLE CoffeeTable (
Type VARCHAR(25) NOT NULL,
Pounds INTEGER NOT NULL,
Price NUMERIC(5, 2) NOT NULL

)

How to get a field's value with ResultSet.getxxx when it is a NULL? I have tried to execute a typical SQL statement:
select * from T-name where (clause);
But an error gets thrown because there are some NULL fields in the table.
You should not get an error/exception just because of null values in various columns. This sounds like a driver specific problem and you should first check the original and any chained exceptions to determine if another problem exists. In general, one may retrieve one of three values for a column that is null, depending on the data type. For methods that return objects, null will be returned; for numeric ( get Byte(), getShort(), getInt(), getLong(), getFloat(), and getDouble() ) zero will be returned; for getBoolean() false will be returned. To find out if the value was actually NULL, use ResultSet.wasNull() before invoking another getXXX method.

How do I insert/update records with some of the columns having NULL value?
Use either of the following PreparedStatement methods:
public void setNull(int parameterIndex, int sqlType) throws SQLException
public void setNull(int paramIndex, int sqlType, String typeName) throws SQLException
These methods assume that the columns are nullable. In this case, you can also just omit the columns in an INSERT statement; they will be automatically assigned null values.

Is there a way to find the primary key(s) for an Access Database table? Sun's JDBC-ODBC driver does not implement the getPrimaryKeys() method for the DatabaseMetaData Objects.
// Use meta.getIndexInfo() will
//get you the PK index. Once
// you know the index, retrieve its column name

DatabaseMetaData meta = con.getMetaData();

String key_colname = null;

// get the primary key information
rset = meta.getIndexInfo(null,null, table_name, true,true);
while( rset.next())
{
String idx = rset.getString(6);
if( idx != null)
{
//Note: index "PrimaryKey" is Access DB specific
// other db server has diff. index syntax.
if( idx.equalsIgnoreCase("PrimaryKey"))
{
key_colname = rset.getString(9);
setPrimaryKey( key_colname );
}
}
}

Why can't Tomcat find my Oracle JDBC drivers in classes111.zip?
TOMCAT 4.0.1 on NT4 throws the following exception when I try to connect to Oracle DB from JSP.
javax.servlet.ServletException : oracle.jdbc.driver.OracleDriver
java.lang.ClassNotFoundException: oracle:jdbc:driver:OracleDriver
But, the Oracle JDBC driver ZIP file (classes111.zip)is available in the system classpath.
Copied the Oracle Driver class file (classes111.zip) in %TOMCAT_Home - Home%\lib directory and renamed it to classess111.jar.
Able to connect to Oracle DB from TOMCAT 4.01 via Oracle JDBC-Thin Driver.

I have an application that queries a database and retrieves the results into a JTable. This is the code in the model that seems to be taken forever to execute, especially for a large result set:
while ( myRs.next() ) {
Vector newRow =new Vector();

for ( int i=1;i<=numOfCols;i++ )
{
newRow.addElement(myRs.getObject(i));
}
allRows.addElement(newRow);
}
fireTableChanged(null);

newRow stores each row of the resultset and allRows stores all the rows.

Are the vectors here the problem?
Is there another way of dealing with the result set that could execute faster?
java.util.Vector is largely thread safe, which means that there is a greater overhead in calling addElement() as it is a synchronized method. If your result set is very large, and threading is not an issue, you could use one of the thread-unsafe collections in Java 2 to save some time. java.util.ArrayList is the likeliest candidate here.
Do not use a DefaultTableModel as it loads all of your data into memory at once, which will obviously cause a large overhead - instead, use an AbstractTableModel and provide an implementation which only loads data on demand, i.e. when (if) the user scrolls down through the table.

How does one get column names for rows returned in a ResultSet?
ResultSet rs = ...
...
ResultSetMetaData rsmd = rs.getMetaData();
int numCols = rsmd.getColumnCount();

for (int i = 1; i <= numCols; i++)
{
System.out.println("[" + i + "]" +
rsmd.getColumnName(i) + " {" +
rsmd.getColumnTypeName(i) + "}");
}

What are the considerations for deciding on transaction boundaries?
Transaction processing should always deal with more than one statement and a transaction is often described as a Logical Unit of Work ( LUW ). The rationale for transactions is that you want to know definitively that all or none of the LUW completed successfully. Note that this automatically gives you restart capability. Typically, there are two conditions under which you would want to use transactions:
* Multiple statements involving a single file - An example would be inserting all of a group of rows or all price updates for a given date. You want all of these to take effect at the same time; inserting or changing some subset is not acceptable.
* Multiple statements involving multiple files - The classic example is transferring money from one account to another or double entry accounting; you don't want the debit to succeed and the credit to fail because money or important records will be lost. Another example is a master/detail relationship, where, say, the master contains a total column. If the entire LUW, writing the detail row and updating the master row, is not completed successfully, you A) want to know that the transaction was unsuccessful and B) that a portion of the transaction was not lost or dangling.
Therefore, determining what completes the transaction or LUW should be the deciding factor for transaction boundaries.

How can I determine where a given table is referenced via foreign keys?
DatabaseMetaData.getExportedKeys() returns a ResultSet with data similar to that returned by DatabaseMetaData.getImportedKeys(), except that the information relates to other tables that reference the given table as a foreign key container.

How can I get information about foreign keys used in a table?
DatabaseMetaData.getImportedKeys() returns a ResultSet with data about foreign key columns, tables, sequence and update and delete rules.

Can I use JDBC to execute non-standard features that my DBMS provides?
The answer is a qualified yes. As discussed under SQL Conformance: "One way the JDBC API deals with this problem is to allow any query string to be passed through to an underlying DBMS driver. This means that an application is free to use as much SQL functionality as desired, but it runs the risk of receiving an error on some DBMSs. In fact, an application query may be something other than SQL, or it may be a specialized derivative of SQL designed for specific DBMSs (for document or image queries, for example)."
Clearly this means either giving up portability or checking the DBMS currently used before invoking specific operations.

What is DML?
DML is an abbreviation for Data Manipulation Language. This portion of the SQL standard is concerned with manipulating the data in a database as opposed to the structure of a database. The core verbs for DML are SELECT, INSERT, DELETE, UPDATE, COMMIT and ROLLBACK.

What is the significance of DataBaseMetaData.tableIndexStatistics? How to obtain and use it?
To answer the second question first, the tableIndexStatistic constant in the TYPE column will identify one of the rows in the ResultSet returned when DatabaseMetaData.getIndexInfo() is invoked. If you analyze the wordy API, a tableIndexStatistic row will contain the number of rows in the table in the CARDINALITY column and the number of pages used for the table in the PAGES column.

What types of DataSource objects are specified in the Optional Package?
* Basic - Provides a standard Connection object.
* Pooled - Provides a Connection pool and returns a Connection that is controlled by the pool.
* Distributed - Provides a Connection that can participate in distributed transactions ( more than one DBMS is involved). It is anticipated, but not enforced, that a distributed DataSource will also provide pooling.

However, there are no standard methods available in the DataSource class to determine if one has obtained a pooled and/or distributed Connection.

What is a JDBC 2.0 DataSource?
The DataSource class was introduced in the JDBC 2.0 Optional Package as an easier, more generic means of obtaining a Connection. The actual driver providing services is defined to the DataSource outside the application ( Of course, a production quality app can and should provide this information outside the app anyway, usually with properties files or ResourceBundles ). The documentation expresses the view that DataSource will replace the common DriverManager method.

Does the database server have to be running Java or have Java support in order for my remote JDBC client app to access the database?
The answer should always be no. The two critical requirements are LAN/internet connectivity and an appropriate JDBC driver. Connectivity is usually via TCP/IP, but other communication protocols are possible. Unspoken, but assumed here is that the DBMS has been started to listen on a communications port. It is the JDBC driver's job to convert the SQL statements and JDBC calls to the DBMS' native protocol. From the server's point of view, it's just another data request coming into the port, the programming language used to send the data is irrelevant at that point.

Which Java and java.sql data types map to my specific database types?
JDBC is, of necessity, reliant on the driver and underlying DBMS. These do not always adhere to standards as closely as we would like, including differing names for standard Java types. To deal with this, first, there are a number of tables available in the JDK JDBC documentation dealing with types.

Are the code examples from the JDBC API Tutorial and Reference, Second Edition available online?
Yes.

When an SQL select statement doesn't return any rows, is an SQLException thrown?
No. If you want to throw an exception, you could wrap your SQL related code in a custom class and throw something like ObjectNotFoundException when the returned ResultSet is empty.

Why should I consider optimistic versus pessimistic approaches to database updates?
In a modern database, possibly the two most important issues are data integrity and concurrency ( multiple users have access to and can update the data ). Either approach can be appropriate, depending on the application, but it is important to be aware of possible consequences to avoid being blindsided.
A pessimistic approach, with locks, is usually seen as good for data integrity, although it can be bad for concurrency, especially the longer a lock is held. In particular, it guarantees against 'lost updates' - defined as an update performed by one process between the time of access and update by another process, which overwrites the interim update. However, other users are blocked from updating the data and possibly reading it as well if the read access also tries to acquire a lock. A notorious problem can arise when a user accesses data for update and then doesn't act on it for a period of time. Another situation that occurred with one of my clients is that a batch ( non-interactive ) process may need to update data while an interactive user has an update lock on the same data. In that case, data integrity goes out the window and, depending on how the application is written, more problems may be introduced. ( No, we did not write the interactive update program and yes, we had recovery procedures in place. )
An optimstic approach can alleviate lock concurrency problems, but requires more code and care for integrity. The "optimistic" definition usually says that expectations of update clashes are rare, but I view them as normal occurrances in a heavily used database. The basics are that any changes between time of access and time of update must be detected and taken into account. This is often done by comparing timestamps, but one must be sure that the timestamp is always changed for an update and, of course, that the table contains a timestamp column. A more involved, but more complete method involves saving the original columns and using them in the 'Where' clause of the Update statement. If the update fails, the data has changed and the latest data should be reaccessed.

What is optimistic concurrency?
An optimistic approach dispenses with locks ( except during the actual update ) and usually involves comparison of timestamps, or generations of data to ensure that data hasn't changed between access and update times. It's generally explained that the term optimistic is used because the expectation is that a clash between multiple updates to the same data will seldom occur.

What is pessimistic concurrency?
With a pessimistic approach, locks are used to ensure that no users, other than the one who holds the lock, can update data. It's generally explained that the term pessimistic is used because the expectation is that many users will try to update the same data, so one is pessimistic that an update will be able to complete properly. Locks may be acquired, depending on the DBMS vendor, automatically via the selected Isolation Level. Some vendors also implement 'Select... for Update', which explicitly acquires a lock.

Can I get information about a ResultSet's associated Statement and Connection in a method without having or adding specific arguments for the Statement and Connection?
Yes. Use ResultSet.getStatement(). From the resulting Statement you can use Statement.getConnection().

How can I tell if my JDBC driver normalizes java.sql.Date and java.sql.Time objects?
To actually determine the values, the objects must be converted to a java.util.Date and examined. See What does normalization mean for java.sql.Date and java.sql.Time? for the definition of normalization. Notice that even a debugger will not show whether these objects have been normalized, since the getXXX methods in java.sql.Date for time elements and in java.sql.Time for date elements throw an exception.
So, while a java.sql.Date may show 2001-07-26, it's normalized only if the java.util.Date value is:
Thu Jul 26 00:00:00 EDT 2001
and while a java.sql.Time may show 14:01:00, it's normalized only if the java.util.Date value is:
Thu Jan 01 14:01:00 EST 1970