</source>
        </p>
	<h5>
	Database Import:
	</h5>
	<p>
	Use the very efficient <a href="components.html#streamingdataset">StreamingDataSet</a> to load your XML dataset when 
	working with forward only database operations like UPDATE, INSERT, REFRESH. 
	</p>

      </answer>
    </faq>    
    
    <faq id="flatxmlvalidation">
      <question>
      How to enable flat XML dataset validation?
      </question>
      <answer>
      <p>
      Flat XML validation is disabled by default even if you are using a DTD. Following sample demonstrate how to load a flat XML dataset with DTD validation enabled:
<source>
    FlatXmlProducer producer = new FlatXmlProducer(
            new InputSource("dataset.xml"));
    producer.setValidating(true);
    IDataSet dataSet = new CachedDataSet(producer);
</source>
</p>

      </answer>
    </faq>  

    <faq id="performance">
      <question>
      How to improve the performance of my DbUnit tests?
      </question>
      <answer>
      <p>
      It is normal that testing with a real database is slower than testing with <a href="http://www.mockobjects.com/">MockObjects</a>. Here are few tricks that will help to speedup your DbUnit tests.
      </p>
      <h5>
      1. Reuse the same connection thorough your test suite
      </h5>
      <p>
      Creating a new DbUnit connection every time has a cost. The overhead is much more than just creating a new JDBC connection. DbUnit need to fetches tables' metadata to determine columns data types. This information is cached in the DbUnit connection. So this is highly recommended to reuse the same DbUnit connection thorough your test suite; more you have tables greater are the benefits. 
      </p>
      <h5>
      2. Specify the database schema name 
      </h5>
      <p>
      If your database server supports multiple schemas, like Oracle, you should always specify the schema name you want to use when 
      creating the DbUnit connection. DbUnit can potentially fetch the metadata of all tables it have access to. 
      This include tables from other schemas if you are using a <b>god JDBC connection</b>. So in this situation, 
      specifying a schema name can dramatically improve DbUnit performance. 
      </p>
      <h5>
      3. Test with little data
      </h5>
      <p>
      Unit testing require relatively little data. So try to <a href="bestpractices.html#smalldatasets">keep your setup datasets as small as possible</a>. There is no necessity to reset the entire database content at the beginning of every test. Try to use only the data you need for a particular test case.  
      </p>
      <h5>
      4. Setup stale data once for the entire test suite
      </h5>
      <p>
      If most of your tests are using the same read-only data, you should consider initializing this data once for an entire test class or test suite. 
      </p>
      <h5>
      5. Enable the batched statement feature
      </h5>
      <p>
      The <a href="properties.html#batchstatement">batched statements</a> feature is disabled by default because there are many JDBC drivers incompatible with it. This is recommended to enable this feature if your driver supports it. The performance gain may not be very significant when testing with <a href="bestpractices.html#smalldatasets">small datasets</a>.
</p>

      </answer>
    </faq>  
    
    <faq id="tableseq">
      <question>
        How to automatically orders tables according their foreign keys?
      </question>
      <answer>
        DbUnit operations insert and update tables' rows in the same order they are found in your 
        dataset and delete are done in reverse order. You must order your tables and rows 
        appropriately in your datasets to prevent foreign keys constraint violation.
        <br/>
	Since version 2.0, the <a href="components.html#filtereddataset">DatabaseSequenceFilter</a> can now be used to automatically determine the tables order using foreign/exported keys information. 
        <br/>
	The following sample demonstrate how to use this class to export a flat XML dataset:
<source>
IDatabaseConnection conn = new DatabaseConnection(jdbcConn);

ITableFilter filter = new DatabaseTableFilter(conn);
IDataSet dataset = new FilteredDataSet(filter, 
       conn.createDataSet());

FlatXmlDataSet.write(dataset, new File(fileName));
</source>
      </answer>
    </faq>    
    <faq id="doctypeexport">
      <question>
	How to add the DOCTYPE declaration when writing a flat XML dataset?
      </question>
      <answer>
      <p>
	Use the setDocType() method of the FlatXmlWriter class like this:
<source>
    FlatXmlWriter datasetWriter = new FlatXmlWriter( 
            new FileOutputStream("dataset.xml")); 
    datasetWriter.setDocType("dataset.dtd"); 
    datasetWriter.write(connection.createDataSet());
</source>
    This can also be done with the <a href="anttask.html">DbUnit Ant task</a>.
</p>    
      </answer>
    </faq> 

    <faq id="columnfilter">
      <question>
      How to exclude some table columns at runtime?
      </question>
      <answer>
<p>
      The FilteredTableMetaData class, introduced in DbUnit 2.1, can be used in combination with the IColumnFilter interface to decide the inclusion or exclusion of table columns at runtime.
<source>
    FilteredTableMetaData metaData = new FilteredTableMetaData(
        originalTable.getTableMetaData(), new MyColumnFilter());
    ITable filteredTable = new CompositeTable(metaData, originalTable);
</source>
</p>
<p>
You can use your own IColumnFilter implementation or use the DefaultColumnFilter class provided by DbUnit. DefaultColumnFilter supports wildcards. This class also offers some convenience methods, includedColumnsTable() and excludedColumnsTable(), to ease creation of column filtered table.
<br/>
The following sample demonstrates the usage of DefaultColumnFilter to exclude all columns prefixed with "PK" or suffixed by "TIME". 
</p>
<source>
    DefaultColumnFilter columnFilter = new DefaultColumnFilter();
    columnFilter.excludeColumn("PK*");
    columnFilter.excludeColumn("*TIME");

    FilteredTableMetaData metaData = new FilteredTableMetaData(
            originalTable.getTableMetaData(), columnFilter);
</source>
<p>
Same than above but using the excludedColumnsTable() convenience method.
<source>
    ITable filteredTable = DefaultColumnFilter.excludedColumnsTable(
            originalTable, new String[]{"PK*", "*TIME"});
</source>
</p>
<p>
See also <a href="howto.html#compareignorecolumns">Ignoring some columns in comparison</a>.
</p>
      </answer>
    </faq> 
        
    <faq id="usertypeidentity">
      <question>
      How to use InsertIdentityOperation with user defined types?
      </question>
      <answer>
      <p>
      The IColumnFilter interface is now used by InsertIdentityOperation to detect identity columns. The default implementation assumes that type name of identity columns end with "identity". If you are using user defined types that does not follow this assumption you can now provide your own implementation via the <a href="properties.html#identitycolumnfilter">MS SQL identity column filter property</a>.
<source>
    IDatabaseConnection connection = new DatabaseConnection(jdbcConnection);
    connection.getConfig().setProperty(
        "http://www.dbunit.org/properties/mssql/identityColumnFilter",
        new MyIndentityFilter());
</source>
</p>    
      </answer>
    </faq>  
    
    <faq id="customprimarykeys">
      <question>
      How to customize primary keys detection?
      </question>
      <answer>
      <p>
      The IColumnFilter interface can also be used to determine which columns are primary keys instead of using DatabaseMetaData.getPrimaryKeys().  This can be useful if your primary keys are not explicitly defined in your database model.
<source>
    IDatabaseConnection connection = new DatabaseConnection(jdbcConnection);
    connection.getConfig().setProperty(
        "http://www.dbunit.org/properties/primaryKeyFilter",
        new MyPrimaryKeyFilter());
</source>
</p>    
      </answer>
    </faq>       

  </part>  
  <part id="use">
    <title>Errors</title>

    <faq id="AmbiguousTableNameException">
      <question>
     Why I get a "AmbiguousTableNameException"?
      </question>
      <answer>
      <p>This error occurs when no schema is specified and that DbUnit detect 
        that it is getting columns information from multiple tables having the 
        same name and located in different schemas.</p>
      <p>You can solve this problem in three different ways:
      <ol>
        <li>Provide the schema name when creating the database connection. <b>Note that for Oracle you must specify the schema name in uppercase.</b></li>
        <li>Ensure that the connection is restricted to access only one schema.</li>
        <li>Enable the <a href="properties.html#qualifiedtablenames">qualified table names</a> feature.</li>
      </ol>
      </p>
      </answer>
    </faq>     

    <faq id="fkconstraintviolation">
      <question>
        Why I get a "Foreign keys constraint violation" error?
      </question>
      <answer>
        See <a href="#tableseq">How to automatically orders table according their foreign keys?</a>
      </answer>
    </faq>
    
    <faq id="typenotrecognized">
      <question>
       Why I get a "data type not recognized" warning?
      </question>
      <answer>
      <p>
        By default, DbUnit only support standard JDBC data types. You will get this warning message if you are using vendor specific data types. 
      </p>
      <p>
Read how to <a href="#typefactory"> replace the default data type factory</a> and how to <a href="properties.html#typewarning">disable this warning message</a>.
      </p>

      </answer>
    </faq>    
    
    <faq id="clonedconnection">
      <question>
       Why I get a "Can't start a cloned connection" 
        exception when I use InsertIdentityOperation?
      </question>
      <answer>
        <p>If you are using the Microsoft driver (i.e. <a class="code">com.microsoft.jdbc.sqlserver.SQLServerDriver</a>), 
        you'll need to use the <a class="code">SelectMethod=cursor</a> parameter 
        in the JDBC connection string (as outlined by <a href="http://forum.java.sun.com/thread.jsp?forum=48&amp;thread=184797">this 
        JDC thread</a>). Your database Url would look something like the following: 
      <source>jdbc:microsoft:sqlserver://localhost:1433;DatabaseName=mydb;SelectMethod=cursor</source>
      </p>
      </answer>
    </faq>    
    
    <faq id="UnsatisfiedLinkError">
      <question>
      Why I get an "UnsatisfiedLinkError" 
        with the DB2 driver?
      </question>
      <answer>
       <p>DbUnit uses JDBC 2.0 features (batch updates). By default, DB2 installs 
        the JDBC 1.0 driver. You have to install the JDBC 2.0 driver in order 
        for DbUnit to work or you will get an <a class="code">UnsatisfiedLinkError</a> 
        from the DB2 JDBC 1.0 driver.</p>
      <p>The steps for installing the DB2 JDBC 2.0 driver are covered in the DB2 
        documentation.</p>
      </answer>
    </faq>    
    
    
  </part>
</faqs>