A performance bottleneck?

Paul Withers wrote an very interesting article about the difference between a passthrough UIComponents and the corresponding XPages elements. This means the use of <br> instead a <xp:br>, or a <div> instead of a <xp:div>. A while ago I have tested if this would affect the performance, and as far as I know it makes no difference.

Here is the code I used to test the performance:

<?xml version="1.0" encoding="UTF-8"?>
 <xp:view xmlns:xp="http://www.ibm.com/xsp/core">
    <h1>Name of the Element</h1>
    <xp:repeat id="repeatFull" rows="9999999">
       <xp:this.value><![CDATA[#{javascript:
          var arr:java.util.Vector = new java.util.Vector();
          for( var i=0; i<10;i++){
             arr.add( i );
          }
          var start = java.lang.System.currentTimeMillis();

          arr}]]>
       </xp:this.value>

       <xp:repeat id="repeat1" rows="9999999">
          <xp:this.value><![CDATA[#{javascript:
             var arr:java.util.Vector = new java.util.Vector();
             for( var i=0; i<100000;i++){
                arr.add( i );
             }
             var start = java.lang.System.currentTimeMillis();
             arr}]]>
          </xp:this.value>

          <!-- HERE COMES THE REPEATING ELEMENT //-->

       </xp:repeat>
       <xp:label id="labelTimer">
          <xp:this.value>
             <![CDATA[#{javascript:
                var end = java.lang.System.currentTimeMillis();
                "Total " + (end - start) + " ms."
             }]]>
          </xp:this.value>
       </xp:label>
    </xp:repeat>
    <xp:br />
    <xp:button value="Refresh" id="buttonRefresh">
       <xp:eventHandler event="onclick" submit="true"
          refreshMode="partial" refreshId="repeatFull">
       </xp:eventHandler>
   </xp:button>

</xp:view>

• Test with a <xp:div>

• Open the page

• Partial Refresh

• Test with a <div>

• Opening the page

• Partial Refresh

[Similar results with the other tags]

As you can see there is nothing to worry about. Yes, a UIPassthroughComponent is faster then the corresponding XPages UIComponent. But keep in mind that these test are for 100.000 elements at once. If you have an XPage with 100.000 components on it, you will have other problems (f.e. memory usage etc.).