ADF's af:showDetailItem tag is used as a child to parent tags such as the af:panelAccordion and af:panelTabbed. JDeveloper's online documentation states the following about the af:showDetailItem tag:
The showDetailItem component is used inside of a panelAccordion or panelTabbed component to contain a group of children. It is identified visually by the text attribute value and lays out its children. Note the difference between "disclosed" and "rendered": if "rendered" is false, it means that this the accordion header bar or tab link and its corresponding contents are not available at all to the user, whereas if "disclosed" is false, it means that the contents of the item are not currently visible, but may be made visible by the user since the accordion header bar or tab link are still visible.
The lifecycle (including validation) is not run for any components in a showDetailItem which is not disclosed. The lifecycle is only run on the showDetailItem(s) which is disclosed.
I've never been a fan of the property "disclosed", why not just call it "open"? At least developers then don't have to deal with double negatives like disclosed="false". Regardless the last paragraph highlights an interesting point that the contents of the af:showDetailItem are not processed by the JSF lifecycle if the showDetailItem is currently closed (disclosed="false"). That's desired behaviour, particularly if you have a page with multiple af:showDetailItem tags that in turn query from the business service layer, potentially kicking off a large range of ADF BC queries. Ideally you don't want the queries to fire if their relating af:showDetailItem tag is currently closed.
This feature can be demonstrated via a simple example. Note the following BasicShowDetailItem.jspx page constructed under JDev 11.1.1.4.0:
<?xml version='1.0' encoding='UTF-8'?>
<jsp:root xmlns:jsp="http://java.sun.com/JSP/Page" version="2.1" xmlns:f="http://java.sun.com/jsf/core"
xmlns:h="http://java.sun.com/jsf/html" xmlns:af="http://xmlns.oracle.com/adf/faces/rich">
<jsp:directive.page contentType="text/html;charset=UTF-8"/>
<f:view>
<af:document id="d1">
<af:form id="f1">
<af:panelAccordion id="pa1">
<af:showDetailItem text="Alpha" disclosed="true" id="sdi1">
<af:panelFormLayout id="pfl1">
<af:inputText label="Alpha Value" value="#{basicBean.alphaValue}" id="it1"/>
<af:commandButton text="Submit1" id="cb1"/>
</af:panelFormLayout>
</af:showDetailItem>
<af:showDetailItem text="Beta" disclosed="false" id="sdi2">
<af:inputText label="Beta Value" value="#{basicBean.betaValue}" id="it2"/>
<af:commandButton text="Submit2" id="cb2"/>
</af:showDetailItem>
</af:panelAccordion>
</af:form>
</af:document>
</f:view>
</jsp:root>
This is backed by the following simple requestScope POJO bean entitled BasicBean.java:
package test.view;
import oracle.adf.share.logging.ADFLogger;
public class BasicBean {
public static ADFLogger logger = ADFLogger.createADFLogger(BasicBean.class);
private String alphaValue = "alpha";
private String betaValue = "beta";
public void setAlphaValue(String alphaValue) {
logger.info("setAlphaValue called(" + (alphaValue == null ? "<null>" : alphaValue) + ")");
this.alphaValue = alphaValue;
}
public String getAlphaValue() {
logger.info("getAlphaValue called(" + (alphaValue == null ? "<null>" : alphaValue) + ")");
return alphaValue;
}
public void setBetaValue(String betaValue) {
logger.info("setBetaValue called(" + (betaValue == null ? "<null>" : betaValue) + ")");
this.betaValue = betaValue;
}
public String getBetaValue() {
logger.info("getBetaValue called(" + (betaValue == null ? "<null>" : betaValue) + ")");
return betaValue;
}
}
(More information on the ADFLogger and enabling it can be found in Duncan Mill's recent 4 part
blog).
To assist understanding what's happening we'll also include our own PhaseListener logs such that we can see the JSF lifecycle in action:
public class PhaseListener implements javax.faces.event.PhaseListener {
public static ADFLogger logger = ADFLogger.createADFLogger(PhaseListener.class);
public void beforePhase(PhaseEvent phaseEvent) {
logger.info(phaseEvent.getPhaseId().toString());
}
public void afterPhase(PhaseEvent phaseEvent) {
logger.info(phaseEvent.getPhaseId().toString());
}
public PhaseId getPhaseId() {
return PhaseId.ANY_PHASE;
}
}
At runtime when the page is first rendered we see:
Note that the first af:showDetailItem is disclosed (open) and the second af:showDetailItem is closed. Also note the first af:showDetailItem is showing the alpha value from our requestScope bean, but the beta value is hiding in the 2nd closed af:showDetailItem.
At this stage the logger class gives us an insight into how the requestScope Basicbean has been used. In the log window we can see:
<PhaseListener> <beforePhase> RESTORE_VIEW 1
<PhaseListener> <afterPhase> RESTORE_VIEW 1
<PhaseListener> <beforePhase> RESTORE_VIEW 1
<PhaseListener> <afterPhase> RESTORE_VIEW 1
<PhaseListener> <beforePhase> RENDER_RESPONSE 6
<BasicBean> <getAlphaValue> getAlphaValue called(alpha)
<BasicBean> <getAlphaValue> getAlphaValue called(alpha)
<PhaseListener> <afterPhase> RENDER_RESPONSE 6
We can see the getter accessors for alphaValue have been called twice during the JSF render response phase to render the page. (Why twice? Essentially the JSF engine doesn't guarantee to call a getter once in a request-response cycle. JSF may use a getter for its own purposes such as checking if the value submitted with the request has changed, in order to fire a ValueChangeListener. Google abounds with further discussions on this and the JSF lifecylce including the following by
BalusC).
As expected note that the getter for Beta was not called, as it's not displayed. To take the example one step further, if we hit the submit button available in first af:showDetailItem, the log output still shows no mention of the beta accessors, only calls to getAlphaValue():
<PhaseListener> <beforePhase> RESTORE_VIEW 1
<PhaseListener> <afterPhase> RESTORE_VIEW 1
<PhaseListener> <beforePhase> APPLY_REQUEST_VALUES 2
<PhaseListener> <afterPhase> APPLY_REQUEST_VALUES 2
<PhaseListener> <beforePhase> PROCESS_VALIDATIONS 3
<BasicBean> <getAlphaValue> getAlphaValue called(alpha)
<PhaseListener> <afterPhase> PROCESS_VALIDATIONS 3
<PhaseListener> <beforePhase> UPDATE_MODEL_VALUES 4
<PhaseListener> <afterPhase> UPDATE_MODEL_VALUES 4
<PhaseListener> <beforePhase> INVOKE_APPLICATION 5
<PhaseListener> <afterPhase> INVOKE_APPLICATION 5
<PhaseListener> <beforePhase> RENDER_RESPONSE 6
<BasicBean> <getAlphaValue> getAlphaValue called(alpha)
<BasicBean> <getAlphaValue> getAlphaValue called(alpha)
<PhaseListener> <afterPhase> RENDER_RESPONSE 6
For the purpose of demonstration, if we change the alpha value and resubmit the logs show:
<PhaseListener> <beforePhase> RESTORE_VIEW 1
<PhaseListener> <afterPhase> RESTORE_VIEW 1
<PhaseListener> <beforePhase> APPLY_REQUEST_VALUES 2
<PhaseListener> <afterPhase> APPLY_REQUEST_VALUES 2
<PhaseListener> <beforePhase> PROCESS_VALIDATIONS 3
<BasicBean> <getAlphaValue> getAlphaValue called(alpha)
<PhaseListener> <afterPhase> PROCESS_VALIDATIONS 3
<PhaseListener> <beforePhase> UPDATE_MODEL_VALUES 4
<BasicBean> <setAlphaValue> setAlphaValue called(alpha2)
<PhaseListener> <afterPhase> UPDATE_MODEL_VALUES 4
<PhaseListener> <beforePhase> INVOKE_APPLICATION 5
<PhaseListener> <afterPhase> INVOKE_APPLICATION 5
<PhaseListener> <beforePhase> RENDER_RESPONSE 6
<BasicBean> <getAlphaValue> getAlphaValue called(alpha2)
<BasicBean> <getAlphaValue> getAlphaValue called(alpha2)
<PhaseListener> <afterPhase> RENDER_RESPONSE 6
In this case we see the additional setAlphaValue() call, and multiple getAlphaValue() calls, but no get/setBetaValue() calls. With this we can conclude that indeed the second af:showDetailItem is suppressing the lifecycle of its children.
What's happens if we open the 2nd af:showDetailItem, which closes the 1st af:showDetailItem:
In the log window we see:
<PhaseListener> <beforePhase> RESTORE_VIEW 1
<PhaseListener> <afterPhase> RESTORE_VIEW 1
<PhaseListener> <beforePhase> APPLY_REQUEST_VALUES 2
<PhaseListener> <afterPhase> APPLY_REQUEST_VALUES 2
<PhaseListener> <beforePhase> PROCESS_VALIDATIONS 3
<BasicBean> <getAlphaValue> getAlphaValue called(alpha)
<PhaseListener> <afterPhase> PROCESS_VALIDATIONS 3
<PhaseListener> <beforePhase> UPDATE_MODEL_VALUES 4
<BasicBean> <setAlphaValue> setAlphaValue called(alpha2)
<PhaseListener> <afterPhase> UPDATE_MODEL_VALUES 4
<PhaseListener> <beforePhase> INVOKE_APPLICATION 5
<PhaseListener> <afterPhase> INVOKE_APPLICATION 5
<PhaseListener> <beforePhase> RENDER_RESPONSE 6
<BasicBean> <getBetaValue> getBetaValue called(beta)
<BasicBean> <getBetaValue> getBetaValue called(beta)
<PhaseListener> <afterPhase> RENDER_RESPONSE 6
In this we can see:
1) A single get and set of alpha value – why? – because the af:showDetailItem button still issues a submit to the midtier. At the point in time the second af:showDetailItem's button is pressed, alpha is still showing, and its changes need to be communicated back to the midtier. The getAlphaValue() is to test if the values changed to fire the ValueChangeListener, and the setAlphaValue() is a call to write the new value submitted to the midtier.
An observant reader might pick up the fact that in this log the getAlphaValue call returns a value of alpha rather than alpha2. Surely in the step prior to this one we had already set the value to alpha2? (in fact you can see this in the log output) The answer being this bean has been set at requestScope, not sessionScope, so the state of the internal values are not being copied across requests (which is a useful learning exercise with regards to bean scope but beyond the scope (no pun intended) of this blog post).
2) Two separate calls to getBetaValue() – as the second af:showDetailItem opens, similar to the original retrieval of the alpha value, the JSF lifecycle now calls the getter twice.
If we now press the submit button in the second af:showDetailItem we see the following in the logs:
<PhaseListener> <beforePhase> RESTORE_VIEW 1
<PhaseListener> <afterPhase> RESTORE_VIEW 1
<PhaseListener> <beforePhase> APPLY_REQUEST_VALUES 2
<PhaseListener> <afterPhase> APPLY_REQUEST_VALUES 2
<PhaseListener> <beforePhase> PROCESS_VALIDATIONS 3
<BasicBean> <getBetaValue> getBetaValue called(beta)
<PhaseListener> <afterPhase> PROCESS_VALIDATIONS 3
<PhaseListener> <beforePhase> UPDATE_MODEL_VALUES 4
<PhaseListener> <afterPhase> UPDATE_MODEL_VALUES 4
<PhaseListener> <beforePhase> INVOKE_APPLICATION 5
<PhaseListener> <afterPhase> INVOKE_APPLICATION 5
<PhaseListener> <beforePhase> RENDER_RESPONSE 6
<BasicBean> <getBetaValue> getBetaValue called(beta)
<BasicBean> <getBetaValue> getBetaValue called(beta)
<PhaseListener> <afterPhase> RENDER_RESPONSE 6
As previous when we pressed the submit button in the first af:showDetailItem, the log output and the calls to getBetaValue match the frequency and location of getAlphaValue. Again now that the first af:showDetailItem is fully closed, we see no JSF lifecycle on the get/setAlphaValue() methods.
So in conclusion, if the af:showDetailItem is closed, and not in the processing of being closed, then its children will not be activated.
Okay, but what's up with af:showDetailItems and af:regions?Now that we know the default behaviour of the af:showDetailItem, let's extend the example to show where the behaviour changes.
Within JDev 11g we can make use of af:regions to call ADF bounded task flows. As example we may have the following page entitled ShowDetailItemWithRegion.jspx:
<?xml version='1.0' encoding='UTF-8'?>
<jsp:root xmlns:jsp="http://java.sun.com/JSP/Page" version="2.1" xmlns:f="http://java.sun.com/jsf/core"
xmlns:h="http://java.sun.com/jsf/html" xmlns:af="http://xmlns.oracle.com/adf/faces/rich">
<jsp:directive.page contentType="text/html;charset=UTF-8"/>
<f:view>
<af:document id="d1">
<af:form id="f1">
<af:panelAccordion id="pa1">
<af:showDetailItem text="Alpha" disclosed="true" id="sdi1">
<af:panelFormLayout id="pfl1">
<af:region value="#{bindings.AlphaTaskFlow1.regionModel}" id="r1"/>
<af:commandButton text="Submit1" id="cb1"/>
</af:panelFormLayout>
</af:showDetailItem>
<af:showDetailItem text="Beta" disclosed="false" id="sdi2">
<af:region value="#{bindings.BetaTaskFlow1.regionModel}" id="r2"/>
<af:commandButton text="Submit2" id="cb2"/>
</af:showDetailItem>
</af:panelAccordion>
</af:form>
</af:document>
</f:view>
</jsp:root>
Note the embedded regions within each af:showDetailItem. The setup of the rest of the page is the same, with the first af:showDetailItem being disclosed (open) and the closed when the page first renders.
The task flows themselves are very simple. As example AlphaTaskFlow contains one fragment AlphaFragment which is the default activity:
The AlphaFragment.jsff includes the following code:
<?xml version='1.0' encoding='UTF-8'?>
<jsp:root xmlns:jsp="http://java.sun.com/JSP/Page" version="2.1" xmlns:af="http://xmlns.oracle.com/adf/faces/rich">
<af:inputText label="Alpha value" value="#{backingBeanScope.alphaBean.alphaValue}" id="it1"/>
</jsp:root>
Of which references a backingBean scoped bean for the task flow named AlphaBean:
package test.view;
import oracle.adf.share.logging.ADFLogger;
public class AlphaBean1 {
public static ADFLogger logger = ADFLogger.createADFLogger(AlphaBean.class);
private String alphaValue = "alpha1";
public void setAlphaValue(String alphaValue) {
logger.info("setAlphaValue called(" + (alphaValue == null ? "<null>" : alphaValue) + ")");
this.alphaValue = alphaValue;
}
public String getAlphaValue() {
logger.info("getAlphaValue called(" + (alphaValue == null ? "<null>" : alphaValue) + ")");
return alphaValue;
}
public void taskFlowInit() {
logger.info("Task flow initialized");
}
public void taskFlowFinalizer() {
logger.info("Task flow finalized");
}
}
This bean carries the alphaValue plus the associated getters and setters. The only addition here is the taskFlowInit() and taskFlowFinalizer() methods which we'll use in the task flow to log when the task flow is started and stopped:
In terms of the 2nd task flow BetaTaskFlow, it's exactly the same as AlphaTaskFlow except it calls the beta equivalent. As such the BetaFragment.jsff:
<?xml version='1.0' encoding='UTF-8'?>
<jsp:root xmlns:jsp="http://java.sun.com/JSP/Page" version="2.1" xmlns:af="http://xmlns.oracle.com/adf/faces/rich">
<af:inputText label="Beta value" value="#{backingBeanScope.betaBean.betaValue}" id="it1"/>
</jsp:root>
The backingBeanScope BetaBean:
package test.view;
import oracle.adf.share.logging.ADFLogger;
public class BetaBean1 {
public static ADFLogger logger = ADFLogger.createADFLogger(BetaBean.class);
private String betaValue = "beta1";
public void setBetaValue(String betaValue) {
logger.info("setBetaValue called(" + (betaValue == null ? "<null>" : betaValue) + ")");
this.betaValue = betaValue;
}
public String getBetaValue() {
logger.info("getBetaValue called(" + (betaValue == null ? "<null>" : betaValue) + ")");
return betaValue;
}
public void taskFlowInit() {
logger.info("Task flow initialized");
}
public void taskFlowFinalizer() {
logger.info("Task flow finalized");
}
}
And the BetaTaskFlow initializer and finalizers set:
With the moving parts done, let's see what happens at runtime.
When we run the base page we see:
From the log output we see:
<PhaseListener> <beforePhase> RESTORE_VIEW 1
<PhaseListener> <afterPhase> RESTORE_VIEW 1
<PhaseListener> <beforePhase> RESTORE_VIEW 1
<PhaseListener> <afterPhase> RESTORE_VIEW 1
<PhaseListener> <beforePhase> RENDER_RESPONSE 6
<AlphaBean> <taskFlowInit> Task flow initialized
<BetaBean> <taskFlowInit> Task flow initialized
<AlphaBean> <getAlphaValue> getAlphaValue called(alpha1)
<AlphaBean> <getAlphaValue> getAlphaValue called(alpha1)
<PhaseListener> <afterPhase> RENDER_RESPONSE 6
In context of we saw in the previous example, this is an interesting result. While we see that only getAlphaValue() has been called similar to our previous example that didn't use regions, what we can also see in the RENDER_RESPONSE phase unexpectedly that the initializer for *both* task flows have been called. We expected the task flow initializer for the AlphaTaskFlow to be called, but the framework has decided to start the BetaTaskFlow as well. Another observation though is even though the BetaTaskFlow was started, somehow the framework didn't call getBetaValue?
An assumption you could make here is the framework is priming the BetaTaskFlow and calling the task flow initializer, but not actually running the task flow. We can disapprove this fact by extending the BetaTaskFlow to include a new Method Call as the task flow activity:
...where the Method Call simply calls a new method in the BetaBean1:
public void logBegin() {
logger.info("Task flow beginning");
}
If we re-run our application the current log output is shown when the page opens:
<PhaseListener> <beforePhase> RESTORE_VIEW 1
<PhaseListener> <afterPhase> RESTORE_VIEW 1
<PhaseListener> <beforePhase> RESTORE_VIEW 1
<PhaseListener> <afterPhase> RESTORE_VIEW 1
<PhaseListener> <beforePhase> RENDER_RESPONSE 6
<AlphaBean> <taskFlowInit> Task flow initialized
<BetaBean> <taskFlowInit> Task flow initialized
<BetaBea> <logBegin> Task flow beginning
<AlphaBean> <getAlphaValue> getAlphaValue called(alpha1)
<AlphaBean> <getAlphaValue> getAlphaValue called(alpha1)
<PhaseListener> <afterPhase> RENDER_RESPONSE 6
This proves that the activities within the BetaTaskFlow are actually being called. However the ADF engine seems to stop short at processing the BetaFragment, the effective viewable part of the task flow.
The conclusion we can draw here is even though you think you're hiding the BetaTaskFlow, and the af:showDetailItem documentation says it wont process the lifecyle of it's children, for af:regions using task flows, this is not the case, it is in fact processing them (up to a point). The implication of this is (at least some) unnecessary processing will occur even if the user never looks at the contents of the closed af:showDetailItem.
In the
next post in this series, still using JSPX pages and JDev 11.1.1.4.0, we'll look at how we can programmatically control the activation of the hidden region to stop unnecessary processing.
The final
post in the series will look at what options are available to us under JDev 11.1.2 using Facelets.
Sample ApplicationA sample application containing solutions for part 1 of this series is available
here.
ThanksThis post was inspired by Oracle's Steve Davelaar who highlighted the new region processing in JDev 11.1.2. The introduction of the new 11.1.2 feature led me to explore the default behaviour under 11.1.1.4.0 without the new feature.