[richfaces-issues] [JBoss JIRA] Commented: (RF-10940) Client side validation refers to non existing javascript function

Thu Aug 25 01:18:18 EDT 2011

    [ https://issues.jboss.org/browse/RF-10940?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12623909#comment-12623909 ] 

Brian Leathem commented on RF-10940:
------------------------------------

I took a close look at this, and found that indeed the a4j framework is being smart about not including duplicate javascript function definitions.  The problem is that when the part of the page is re-rendered, the reference to the original javascript method is lost.

As a workaround (which may not fit your use case), you can wrap all the ajax validation components in an a4j:outputPanel, as in:
{code}

<!DOCTYPE html>
<html xmlns="http://www.w3.org/1999/xhtml"
      xmlns:ui="http://java.sun.com/jsf/facelets"
      xmlns:h="http://java.sun.com/jsf/html"
      xmlns:f="http://java.sun.com/jsf/core"
      xmlns:a="http://richfaces.org/a4j"
      xmlns:rich-in="http://richfaces.org/input"
      xmlns:rich-v="http://richfaces.org/csv">

<f:view>
    <h:head>
        <title>Calendar Validation Test</title>
    </h:head>

    <h:body>
        <h:form>
            <h:outputLabel value="Start date" for="#{startDate}" />
            <a:outputPanel id="datePanel">
                <rich-in:calendar id="startDate" value="#{calendarTestBean.startDate}">
                    <rich-v:validator />
                </rich-in:calendar>

                <h:outputLabel value="End date" for="#{endDate}" />

                <rich-in:calendar id="endDate" value="#{calendarTestBean.endDate}">
                    <rich-v:validator />
                </rich-in:calendar>
            </a:outputPanel>

            <a:commandButton value="Rerender end date" render="endDatePanel" />
        </h:form>
    </h:body>
</f:view>
</html>
{code}

This way, the components are aware of the shared javascript methods on each page re-render.

To *actually* fix the problem, I believe we have to disable the logic that prevents duplicate (logically) javascript functions from being defined.  This seems like a backwards step, so I'd like to investigate if it makes sense before proceeding.

> Client side validation refers to non existing javascript function
> -----------------------------------------------------------------
>
>                 Key: RF-10940
>                 URL: https://issues.jboss.org/browse/RF-10940
>             Project: RichFaces
>          Issue Type: Bug
>      Security Level: Public(Everyone can see) 
>          Components: component-validators
>    Affects Versions: 4.0.0.Final
>            Reporter: Niels Soeffers
>            Assignee: Brian Leathem
>            Priority: Critical
>              Labels: tran
>             Fix For: 4.1.0.Milestone2
>
>
> Client Side Javascript functions are called on the component tags (for example onChange attribute) and defined at the bottom of the rendered page. We encountered a case where the function called in the onchange attribute is not defined at the bottom of the page.
> Our domain layer contains two different classes RG and SRG (they do not inherit from each other). However they both embed an embeddable class E with a field "name"
> In the GUI we have a page with a single form and two regions. One for adding a RG object and one for adding a SRG object. Both regions contain a h:inputText for the "name" field. So the first inputText is bound to {{RG.embedded.name}} and the second on {{SRG.embedded.name}}. Both inputTexts have a rich:validator tag as a child component.
> When the page is initially rendered both h:inputText have the same javascript function referred to in their onchange attribute (RGName_3Av). 
> Which is not wrong because the validation (@NotNull on the "name" field in the Embedded class) is the same for both value bindings. It is a little bit confusing to see a method with a name of RGName_3Av on a component that is actually bound to SRG and so we would expect the method to call SRGName_3Av, but it's certainly not wrong and I even find it a very nice optimization.
> However when we click on an a:commandButton in the second region and rerender this second region, then suddenly the javascript function in the onchange attribute of the second inputText is changed and the name refers to the SRG component (SRGName_3Av), but this javascript method is never defined, because it was optimized when we initially rendered the complete page.
> Hope you can reproduce the problem with the above description and fix the problem or provide a workaround?
> Kind Regards,
> Niels

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira