public interface CachedRowSet extends RowSet, Joinable

The interface that all standard implementations of CachedRowSet must implement.

The reference implementation of the CachedRowSet interface provided by Oracle Corporation is a standard implementation. Developers may use this implementation just as it is, they may extend it, or they may choose to write their own implementations of this interface.

A CachedRowSet object is a container for rows of data that caches its rows in memory, which makes it possible to operate without always being connected to its data source. Further, it is a JavaBeans component and is scrollable, updatable, and serializable. A CachedRowSet object typically contains rows from a result set, but it can also contain rows from any file with a tabular format, such as a spread sheet. The reference implementation supports getting data only from a ResultSet object, but developers can extend the SyncProvider implementations to provide access to other tabular data sources.

An application can modify the data in a CachedRowSet object, and those modifications can then be propagated back to the source of the data.

A CachedRowSet object is a disconnected rowset, which means that it makes use of a connection to its data source only briefly. It connects to its data source while it is reading data to populate itself with rows and again while it is propagating changes back to its underlying data source. The rest of the time, a CachedRowSet object is disconnected, including while its data is being modified. Being disconnected makes a RowSet object much leaner and therefore much easier to pass to another component. For example, a disconnected RowSet object can be serialized and passed over the wire to a thin client such as a personal digital assistant (PDA).

1.0 Creating a CachedRowSet Object

The following line of code uses the default constructor for CachedRowSet supplied in the reference implementation (RI) to create a default CachedRowSet object.

     CachedRowSetImpl crs = new CachedRowSetImpl();
 

This new CachedRowSet object will have its properties set to the default properties of a BaseRowSet object, and, in addition, it will have an RIOptimisticProvider object as its synchronization provider. RIOptimisticProvider, one of two SyncProvider implementations included in the RI, is the default provider that the SyncFactory singleton will supply when no synchronization provider is specified.

A SyncProvider object provides a CachedRowSet object with a reader (a RowSetReader object) for reading data from a data source to populate itself with data. A reader can be implemented to read data from a ResultSet object or from a file with a tabular format. A SyncProvider object also provides a writer (a RowSetWriter object) for synchronizing any modifications to the CachedRowSet object's data made while it was disconnected with the data in the underlying data source.

A writer can be implemented to exercise various degrees of care in checking for conflicts and in avoiding them. (A conflict occurs when a value in the data source has been changed after the rowset populated itself with that value.) The RIOptimisticProvider implementation assumes there will be few or no conflicts and therefore sets no locks. It updates the data source with values from the CachedRowSet object only if there are no conflicts. Other writers can be implemented so that they always write modified data to the data source, which can be accomplished either by not checking for conflicts or, on the other end of the spectrum, by setting locks sufficient to prevent data in the data source from being changed. Still other writer implementations can be somewhere in between.

A CachedRowSet object may use any SyncProvider implementation that has been registered with the SyncFactory singleton. An application can find out which SyncProvider implementations have been registered by calling the following line of code.

      java.util.Enumeration providers = SyncFactory.getRegisteredProviders();
 

There are two ways for a CachedRowSet object to specify which SyncProvider object it will use.

• Supplying the name of the implementation to the constructor
  The following line of code creates the CachedRowSet object crs2 that is initialized with default values except that its SyncProvider object is the one specified.

            CachedRowSetImpl crs2 = new CachedRowSetImpl(
                                   "com.fred.providers.HighAvailabilityProvider");
       

• Setting the SyncProvider using the CachedRowSet method setSyncProvider
  The following line of code resets the SyncProvider object for crs, the CachedRowSet object created with the default constructor.

             crs.setSyncProvider("com.fred.providers.HighAvailabilityProvider");
        

See the comments for SyncFactory and SyncProvider for more details.

2.0 Retrieving Data from a CachedRowSet Object

Data is retrieved from a CachedRowSet object by using the getter methods inherited from the ResultSet interface. The following examples, in which crs is a CachedRowSet object, demonstrate how to iterate through the rows, retrieving the column values in each row. The first example uses the version of the getter methods that take a column number; the second example uses the version that takes a column name. Column numbers are generally used when the RowSet object's command is of the form SELECT * FROM TABLENAME; column names are most commonly used when the command specifies columns by name.

    while (crs.next()) {
        String name = crs.getString(1);
        int id = crs.getInt(2);
        Clob comment = crs.getClob(3);
        short dept = crs.getShort(4);
        System.out.println(name + "  " + id + "  " + comment + "  " + dept);
    }
 

    while (crs.next()) {
        String name = crs.getString("NAME");
        int id = crs.getInt("ID");
        Clob comment = crs.getClob("COM");
        short dept = crs.getShort("DEPT");
        System.out.println(name + "  " + id + "  " + comment + "  " + dept);
    }
 

2.1 Retrieving RowSetMetaData

An application can get information about the columns in a CachedRowSet object by calling ResultSetMetaData and RowSetMetaData methods on a RowSetMetaData object. The following code fragment, in which crs is a CachedRowSet object, illustrates the process. The first line creates a RowSetMetaData object with information about the columns in crs. The method getMetaData, inherited from the ResultSet interface, returns a ResultSetMetaData object, which is cast to a RowSetMetaData object before being assigned to the variable rsmd. The second line finds out how many columns jrs has, and the third line gets the JDBC type of values stored in the second column of jrs.

     RowSetMetaData rsmd = (RowSetMetaData)crs.getMetaData();
     int count = rsmd.getColumnCount();
     int type = rsmd.getColumnType(2);
 

The RowSetMetaData interface differs from the ResultSetMetaData interface in two ways.

• It includes setter methods: A RowSet object uses these methods internally when it is populated with data from a different ResultSet object.
• It contains fewer getter methods: Some ResultSetMetaData methods to not apply to a RowSet object. For example, methods retrieving whether a column value is writable or read only do not apply because all of a RowSet object's columns will be writable or read only, depending on whether the rowset is updatable or not.

NOTE: In order to return a RowSetMetaData object, implementations must override the getMetaData() method defined in java.sql.ResultSet and return a RowSetMetaData object.

3.0 Updating a CachedRowSet Object

Updating a CachedRowSet object is similar to updating a ResultSet object, but because the rowset is not connected to its data source while it is being updated, it must take an additional step to effect changes in its underlying data source. After calling the method updateRow or insertRow, a CachedRowSet object must also call the method acceptChanges to have updates written to the data source. The following example, in which the cursor is on a row in the CachedRowSet object crs, shows the code required to update two column values in the current row and also update the RowSet object's underlying data source.

     crs.updateShort(3, 58);
     crs.updateInt(4, 150000);
     crs.updateRow();
     crs.acceptChanges();
 

The next example demonstrates moving to the insert row, building a new row on the insert row, inserting it into the rowset, and then calling the method acceptChanges to add the new row to the underlying data source. Note that as with the getter methods, the updater methods may take either a column index or a column name to designate the column being acted upon.

     crs.moveToInsertRow();
     crs.updateString("Name", "Shakespeare");
     crs.updateInt("ID", 10098347);
     crs.updateShort("Age", 58);
     crs.updateInt("Sal", 150000);
     crs.insertRow();
     crs.moveToCurrentRow();
     crs.acceptChanges();
 

NOTE: Where the insertRow() method inserts the contents of a CachedRowSet object's insert row is implementation-defined. The reference implementation for the CachedRowSet interface inserts a new row immediately following the current row, but it could be implemented to insert new rows in any number of other places.

Another thing to note about these examples is how they use the method acceptChanges. It is this method that propagates changes in a CachedRowSet object back to the underlying data source, calling on the RowSet object's writer internally to write changes to the data source. To do this, the writer has to incur the expense of establishing a connection with that data source. The preceding two code fragments call the method acceptChanges immediately after calling updateRow or insertRow. However, when there are multiple rows being changed, it is more efficient to call acceptChanges after all calls to updateRow and insertRow have been made. If acceptChanges is called only once, only one connection needs to be established.

4.0 Updating the Underlying Data Source

When the method acceptChanges is executed, the CachedRowSet object's writer, a RowSetWriterImpl object, is called behind the scenes to write the changes made to the rowset to the underlying data source. The writer is implemented to make a connection to the data source and write updates to it.

A writer is made available through an implementation of the SyncProvider interface, as discussed in section 1, "Creating a CachedRowSet Object." The default reference implementation provider, RIOptimisticProvider, has its writer implemented to use an optimistic concurrency control mechanism. That is, it maintains no locks in the underlying database while the rowset is disconnected from the database and simply checks to see if there are any conflicts before writing data to the data source. If there are any conflicts, it does not write anything to the data source.

The reader/writer facility provided by the SyncProvider class is pluggable, allowing for the customization of data retrieval and updating. If a different concurrency control mechanism is desired, a different implementation of SyncProvider can be plugged in using the method setSyncProvider.

In order to use the optimistic concurrency control routine, the RIOptimisticProvider maintains both its current value and its original value (the value it had immediately preceding the current value). Note that if no changes have been made to the data in a RowSet object, its current values and its original values are the same, both being the values with which the RowSet object was initially populated. However, once any values in the RowSet object have been changed, the current values and the original values will be different, though at this stage, the original values are still the initial values. With any subsequent changes to data in a RowSet object, its original values and current values will still differ, but its original values will be the values that were previously the current values.

Keeping track of original values allows the writer to compare the RowSet object's original value with the value in the database. If the values in the database differ from the RowSet object's original values, which means that the values in the database have been changed, there is a conflict. Whether a writer checks for conflicts, what degree of checking it does, and how it handles conflicts all depend on how it is implemented.

5.0 Registering and Notifying Listeners

Being JavaBeans components, all rowsets participate in the JavaBeans event model, inheriting methods for registering listeners and notifying them of changes from the BaseRowSet class. A listener for a CachedRowSet object is a component that wants to be notified whenever there is a change in the rowset. For example, if a CachedRowSet object contains the results of a query and those results are being displayed in, say, a table and a bar graph, the table and bar graph could be registered as listeners with the rowset so that they can update themselves to reflect changes. To become listeners, the table and bar graph classes must implement the RowSetListener interface. Then they can be added to the CachedRowSet object's list of listeners, as is illustrated in the following lines of code.

    crs.addRowSetListener(table);
    crs.addRowSetListener(barGraph);
 

Each CachedRowSet method that moves the cursor or changes data also notifies registered listeners of the changes, so table and barGraph will be notified when there is a change in crs.

6.0 Passing Data to Thin Clients

One of the main reasons to use a CachedRowSet object is to pass data between different components of an application. Because it is serializable, a CachedRowSet object can be used, for example, to send the result of a query executed by an enterprise JavaBeans component running in a server environment over a network to a client running in a web browser.

While a CachedRowSet object is disconnected, it can be much leaner than a ResultSet object with the same data. As a result, it can be especially suitable for sending data to a thin client such as a PDA, where it would be inappropriate to use a JDBC driver due to resource limitations or security considerations. Thus, a CachedRowSet object provides a means to "get rows in" without the need to implement the full JDBC API.

7.0 Scrolling and Updating

A second major use for CachedRowSet objects is to provide scrolling and updating for ResultSet objects that do not provide these capabilities themselves. In other words, a CachedRowSet object can be used to augment the capabilities of a JDBC technology-enabled driver (hereafter called a "JDBC driver") when the DBMS does not provide full support for scrolling and updating. To achieve the effect of making a non-scrollable and read-only ResultSet object scrollable and updatable, a programmer simply needs to create a CachedRowSet object populated with that ResultSet object's data. This is demonstrated in the following code fragment, where stmt is a Statement object.

    ResultSet rs = stmt.executeQuery("SELECT * FROM EMPLOYEES");
    CachedRowSetImpl crs = new CachedRowSetImpl();
    crs.populate(rs);
 

The object crs now contains the data from the table EMPLOYEES, just as the object rs does. The difference is that the cursor for crs can be moved forward, backward, or to a particular row even if the cursor for rs can move only forward. In addition, crs is updatable even if rs is not because by default, a CachedRowSet object is both scrollable and updatable.

In summary, a CachedRowSet object can be thought of as simply a disconnected set of rows that are being cached outside of a data source. Being thin and serializable, it can easily be sent across a wire, and it is well suited to sending data to a thin client. However, a CachedRowSet object does have a limitation: It is limited in size by the amount of data it can store in memory at one time.

8.0 Getting Universal Data Access

Another advantage of the CachedRowSet class is that it makes it possible to retrieve and store data from sources other than a relational database. The reader for a rowset can be implemented to read and populate its rowset with data from any tabular data source, including a spreadsheet or flat file. Because both a CachedRowSet object and its metadata can be created from scratch, a component that acts as a factory for rowsets can use this capability to create a rowset containing data from non-SQL data sources. Nevertheless, it is expected that most of the time, CachedRowSet objects will contain data that was fetched from an SQL database using the JDBC API.

9.0 Setting Properties

All rowsets maintain a set of properties, which will usually be set using a tool. The number and kinds of properties a rowset has will vary, depending on what the rowset does and how it gets its data. For example, rowsets that get their data from a ResultSet object need to set the properties that are required for making a database connection. If a rowset uses the DriverManager facility to make a connection, it needs to set a property for the JDBC URL that identifies the appropriate driver, and it needs to set the properties that give the user name and password. If, on the other hand, the rowset uses a DataSource object to make the connection, which is the preferred method, it does not need to set the property for the JDBC URL. Instead, it needs to set properties for the logical name of the data source, for the user name, and for the password.

NOTE: In order to use a DataSource object for making a connection, the DataSource object must have been registered with a naming service that uses the Java Naming and Directory Interface (JNDI) API. This registration is usually done by a person acting in the capacity of a system administrator.

In order to be able to populate itself with data from a database, a rowset needs to set a command property. This property is a query that is a PreparedStatement object, which allows the query to have parameter placeholders that are set at run time, as opposed to design time. To set these placeholder parameters with values, a rowset provides setter methods for setting values of each data type, similar to the setter methods provided by the PreparedStatement interface.

The following code fragment illustrates how the CachedRowSet object crs might have its command property set. Note that if a tool is used to set properties, this is the code that the tool would use.

    crs.setCommand("SELECT FIRST_NAME, LAST_NAME, ADDRESS FROM CUSTOMERS " +
                   "WHERE CREDIT_LIMIT > ? AND REGION = ?");
  

The values that will be used to set the command's placeholder parameters are contained in the RowSet object's params field, which is a Vector object. The CachedRowSet class provides a set of setter methods for setting the elements in its params field. The following code fragment demonstrates setting the two parameters in the query from the previous example.

    crs.setInt(1, 5000);
    crs.setString(2, "West");
 

The params field now contains two elements, each of which is an array two elements long. The first element is the parameter number; the second is the value to be set. In this case, the first element of params is 1, 5000, and the second element is 2, "West". When an application calls the method execute, it will in turn call on this RowSet object's reader, which will in turn invoke its readData method. As part of its implementation, readData will get the values in params and use them to set the command's placeholder parameters. The following code fragment gives an idea of how the reader does this, after obtaining the Connection object con.

    PreparedStatement pstmt = con.prepareStatement(crs.getCommand());
    reader.decodeParams();
    // decodeParams figures out which setter methods to use and does something
    // like the following:
    //    for (i = 0; i < params.length; i++) {
    //        pstmt.setObject(i + 1, params[i]);
    //    }
 

At this point, the command for crs is the query "SELECT FIRST_NAME, LAST_NAME, ADDRESS FROM CUSTOMERS WHERE CREDIT_LIMIT > 5000 AND REGION = "West". After the readData method executes this command with the following line of code, it will have the data from rs with which to populate crs.

     ResultSet rs = pstmt.executeQuery();
 

The preceding code fragments give an idea of what goes on behind the scenes; they would not appear in an application, which would not invoke methods like readData and decodeParams. In contrast, the following code fragment shows what an application might do. It sets the rowset's command, sets the command's parameters, and executes the command. Simply by calling the execute method, crs populates itself with the requested data from the table CUSTOMERS.

    crs.setCommand("SELECT FIRST_NAME, LAST_NAME, ADDRESS FROM CUSTOMERS" +
                   "WHERE CREDIT_LIMIT > ? AND REGION = ?");
    crs.setInt(1, 5000);
    crs.setString(2, "West");
    crs.execute();
 

10.0 Paging Data

Because a CachedRowSet object stores data in memory, the amount of data that it can contain at any one time is determined by the amount of memory available. To get around this limitation, a CachedRowSet object can retrieve data from a ResultSet object in chunks of data, called pages. To take advantage of this mechanism, an application sets the number of rows to be included in a page using the method setPageSize. In other words, if the page size is set to five, a chunk of five rows of data will be fetched from the data source at one time. An application can also optionally set the maximum number of rows that may be fetched at one time. If the maximum number of rows is set to zero, or no maximum number of rows is set, there is no limit to the number of rows that may be fetched at a time.

After properties have been set, the CachedRowSet object must be populated with data using either the method populate or the method execute. The following lines of code demonstrate using the method populate. Note that this version of the method takes two parameters, a ResultSet handle and the row in the ResultSet object from which to start retrieving rows.

     CachedRowSet crs = new CachedRowSetImpl();
     crs.setMaxRows(20);
     crs.setPageSize(4);
     crs.populate(rsHandle, 10);
 

When this code runs, crs will be populated with four rows from rsHandle starting with the tenth row.

The next code fragment shows populating a CachedRowSet object using the method execute, which may or may not take a Connection object as a parameter. This code passes execute the Connection object conHandle.

Note that there are two differences between the following code fragment and the previous one. First, the method setMaxRows is not called, so there is no limit set for the number of rows that crs may contain. (Remember that crs always has the overriding limit of how much data it can store in memory.) The second difference is that the you cannot pass the method execute the number of the row in the ResultSet object from which to start retrieving rows. This method always starts with the first row.

     CachedRowSet crs = new CachedRowSetImpl();
     crs.setPageSize(5);
     crs.execute(conHandle);
 

After this code has run, crs will contain five rows of data from the ResultSet object produced by the command for crs. The writer for crs will use conHandle to connect to the data source and execute the command for crs. An application is then able to operate on the data in crs in the same way that it would operate on data in any other CachedRowSet object.

To access the next page (chunk of data), an application calls the method nextPage. This method creates a new CachedRowSet object and fills it with the next page of data. For example, assume that the CachedRowSet object's command returns a ResultSet object rs with 1000 rows of data. If the page size has been set to 100, the first call to the method nextPage will create a CachedRowSet object containing the first 100 rows of rs. After doing what it needs to do with the data in these first 100 rows, the application can again call the method nextPage to create another CachedRowSet object with the second 100 rows from rs. The data from the first CachedRowSet object will no longer be in memory because it is replaced with the data from the second CachedRowSet object. After the tenth call to the method nextPage, the tenth CachedRowSet object will contain the last 100 rows of data from rs, which are stored in memory. At any given time, the data from only one CachedRowSet object is stored in memory.

The method nextPage returns true as long as the current page is not the last page of rows and false when there are no more pages. It can therefore be used in a while loop to retrieve all of the pages, as is demonstrated in the following lines of code.

     CachedRowSet crs = CachedRowSetImpl();
     crs.setPageSize(100);
     crs.execute(conHandle);

     while(crs.nextPage()) {
         while(crs.next()) {
             . . . // operate on chunks (of 100 rows each) in crs,
                   // row by row
         }
     }
 

After this code fragment has been run, the application will have traversed all 1000 rows, but it will have had no more than 100 rows in memory at a time.

The CachedRowSet interface also defines the method previousPage. Just as the method nextPage is analogous to the ResultSet method next, the method previousPage is analogous to the ResultSet method previous. Similar to the method nextPage, previousPage creates a CachedRowSet object containing the number of rows set as the page size. So, for instance, the method previousPage could be used in a while loop at the end of the preceding code fragment to navigate back through the pages from the last page to the first page. The method previousPage is also similar to nextPage in that it can be used in a while loop, except that it returns true as long as there is another page preceding it and false when there are no more pages ahead of it.

By positioning the cursor after the last row for each page, as is done in the following code fragment, the method previous navigates from the last row to the first row in each page. The code could also have left the cursor before the first row on each page and then used the method next in a while loop to navigate each page from the first row to the last row.

The following code fragment assumes a continuation from the previous code fragment, meaning that the cursor for the tenth CachedRowSet object is on the last row. The code moves the cursor to after the last row so that the first call to the method previous will put the cursor back on the last row. After going through all of the rows in the last page (the CachedRowSet object crs), the code then enters the while loop to get to the ninth page, go through the rows backwards, go to the eighth page, go through the rows backwards, and so on to the first row of the first page.

     crs.afterLast();
     while(crs.previous())  {
         . . . // navigate through the rows, last to first
     {
     while(crs.previousPage())  {
         crs.afterLast();
         while(crs.previous())  {
             . . . // go from the last row to the first row of each page
         }
     }