0 Replies Latest reply on Jun 15, 2009 10:54 PM by Arbi Sookazian

    Factory annotation and order of events in SFSB

    Arbi Sookazian Master

      I have a conversation-scoped SFSB backing bean with the following business methods:

      public void applyData(){

      @Factory(value="totalUndefined", scope=ScopeType.EVENT)
      public Long getTotalUndefined(){
      Long totalUndefined = (Long)entityManager.createQuery("select count(*) "+
      " from RecoveredEquipmentManagement rem "+
      " where rem.recoveryType = ( SELECT lv.listValueId "+
      " FROM ListValue lv, List l "+
      " WHERE l.listId = lv.list.listId "+
      " and l.listName = :recoveryType"+
      " and lv.listValue = :undefined)")
      .setParameter("recoveryType", RECOVERY_TYPE)                                                  
      .setParameter("undefined", UNDEFINED)
      return totalUndefined;

      The problem is that the following xhtml fragment is not being refreshed in time (in my test case, total number is expected to decrement by one):

      <a4j:outputPanel ajaxRendered="true"> 
            <h:panelGrid columns="2"
                   rendered="#{totalUndefined gt 0 and recoveredEquipmentManagementList.getRowCount() > 0}">
           <s:graphicImage value="/img/warning_red_lg.jpg"/>
            <h:outputText value="There are #{totalUndefined} undefined item number(s)."/>

      The @Factory method is executing before the applyData action-handler method for my a4j:commandButton.  Thus, the decremented value is not displayed in the JSF page.

      How can I reverse the order of execution while still using @Factory pattern?  Alternatively, I could probably just simply call the public method directly and remove the @Factory annotation.  Ideally, I only wan that method to be executed once per HTTP rep/resp cycle (that's why it's EVENT scoped).

      Also, I noticed in the debugger that the @Factory method is executing twice (at least in some cases) and there is a JPQL query in that method.  How best to handle that to avoid duplicate SQL executions in the same HTTP req/resp cycle?  Perhaps I can make the local variable an instance variable instead and check for null?