Use the chainConfig init param for
ActionServlet to specify a comma delimited list of
chain config files.
This will override the default value of
chainConfig, so unless you are
overriding the default request processing chain, you
must include the default chain config file in the list.
chainConfig
org/apache/struts/chain/chain-config.xml,
/WEB-INF/custom-chain-config.xml
For another method of loading chain config files, see Load a Catalog From a Web Application in the Commons Chain Cookbook.
Since Struts 1.1, the framework supports multiple application modules. All applications have at least one root, or default, module. Like the root directory in a file system, the default application has no name. (Or is named with an empty string, depending your viewpoint.) Developing an application with only a default module is no different from how applications were developed under Struts 1.0. Since Struts 1.1, you can add additional modules to your application, each of which can have their own configuration files, messages resources, and so forth. Each module is developed in the same way as the default module. Applications that were developed as a single module can added to a multiple module application, and modules can promoted to a standalone application without change. For more about configuring your application to support multiple modules, see Configuring Applications in the User Guide.
But to answer the question =:0), a modular application is an application that uses more than one module. Module-relative means that the URI starts at the module level, rather than at the context level, or the absolute-URL level.
The Struts Examples application is a modular application that was assembled from several applications that were created independently.
The framework grew in the telling and, as it evolved, some of the names drifted.
The good thing about a nightly build, is that everything becomes available to the community as soon as it is written. The bad thing about a nightly build is that things like class names get locked down early and then become difficult to change.
http://www.mail-archive.com/struts-user@jakarta.apache.org/msg19281.html
http://www.mail-archive.com/struts-user@jakarta.apache.org/msg19338.html
http://www.mail-archive.com/struts-user@jakarta.apache.org/msg20833.html
Originally, the rationale as that making ActionForm a class takes advantage of the single inheritance restriction of Java to it makes it more difficult for people to do things that they should not do. At the time, EJBs were becoming popular, and most developers were trying to combine EJB remoting with ActionForms, and the result was not pretty.
Since then, most developers use different approaches to data persistent, and most developers now have a good understanding of why we want to separate the model from the view. Accordingly, we do plan to introduce more interfaces into the framework. It's just a matter of when.
Meanwhile, DynaActionForms relieve developers of maintaining simple ActionForms. For near zero maintenance, try LazyActionForm and Hubert Rabago's FormDef.
The utilities that the framework uses (Commons-BeanUtils
since 1.1) require that ActionForm properties follow
the JavaBean patterns for mutators and accessors
(get*,set*,is*). Since the framework uses the
Introspection API
with the ActionForms, some containers may require that all
the JavaBean patterns be followed, including
declaring
"implements Serializable"
for each subclass. The safest thing is to review the
JavaBean
specification
and follow all the prescribed patterns.
Since Struts 1.1, you can also use DynaActionForms and mapped-backed forms, which are not true JavaBeans. For more see ActionForm classes in the User Guide and Using Hashmaps with ActionForms in this FAQ.
Yes. There are several ways that you can use other beans or hashmaps with ActionForms.
ActionForms (a.k.a. "form beans") are really just JavaBeans (with a few special methods) that the framework creates and puts into session or request scope for you. There is nothing preventing you from using other beans, or including them in your form beans. Let's look at some examples.
Collections as properties
Suppose that you need to display a pulldown list of
available colors on an input form in your application. You
can include a string-valued
colorSelected
property in your
ActionForm
to represent the user's
selection and a
colorOptions
property implemented as a
Collection
(of strings) to store the available color choices.
Assuming that you have defined the getters
and setters for the
colorSelected
and
colorOptions
properties
in your
orderEntryForm
form bean, you can render the
pulldown list using:
<html:select property="colorSelected">
<html:options property="colorOptions"
name="orderEntryForm"/>
</html:select>
</source>
<p>
The list will be populated using the strings in the
<code>colorOptions</code>
collection of the
<code>orderEntryForm</code>
and the value that the user selects
will go into the
<code>colorSelected</code>
property that gets posted to the subsequent
<code>Action</code>
. Note that we are assuming here that the
<code>colorOptions</code>
property of the
<code>orderEntryForm</code>
has already been set.
</p>
<p>
See
<a href="#prepopulate">How can I prepopulate a form?</a>
for instructions on how to set
form bean properties before rendering edit forms that
expect properties to be pre-set.
</p>
<p>
<em>Independent DTO</em>
An
<code>Action</code>
that retrieves a list of open orders (as an
<code>ArrayList</code>
of
<code>Order</code>
objects) can use a DTO independently of any
form bean to transfer search results to the view. First,
the Action's
<code>execute</code>
method performs the search and puts the DTO into the
request:
</p>
<source>
ArrayList results =
businessObject.executeSearch(searchParameters);
request.setAttribute("searchResults",results);
</source>
<p>
Then the view can iterate through the results using the
"searchResults" request key to
reference the DTO:
`</p>
<source>
<logic:iterate id="order" name="searchResults"
type="com.foo.bar.Order">
<tr><td><bean:write name="order"
property="orderNumber"/><td>
<td>..other properties...</td></tr>
</logic:iterate>
See also: Map-Backed ActionForms (since Struts 1.1)
http://www.mail-archive.com/struts-user@jakarta.apache.org/msg24504.html
http://www.mail-archive.com/struts-user@jakarta.apache.org/msg22949.html
The short answer to this question is: No, you are not limited to JavaServer Pages.
The longer answer is that you can use any type of presentation technology which can be returned by a web server or Java container. The list includes but is not limited to:
Some people even mix and match apparently unrelated technologies, like PHP, into the same web application.
ActionForms are added to a servlet scope (session or request) as beans. What this means is that, for certain functionality to be available, your ActionForms will have to follow a few simple rules.
First, your ActionForm bean must have a zero-arguments constructor. This is required because the framework must be able to dynamically create new instances of your form bean class, while knowing only the class name. This is not an onerous restriction, however, because the framework will also populate your form bean's properties (from the request parameters) for you.
Second, the fields of your form bean are made available to the framework by supplying public getter and setter methods that follow the naming design patterns described in the JavaBeans Specification. For most users, that means using the following idiom for each of your form bean's properties:
private {type} fieldName;
public {type} getFieldName() {
return (this.fieldName);
}
public void setFieldName({type} fieldName) {
this.fieldName = fieldName;
}
NOTE - you MUST obey the capitalization conventions shown above for your ActionForm properties to be recognized. The property name in this example is "fieldName", and that must also be the name of the input field that corresponds to this property. A bean property may have a "getter" method and a "setter" method (in a form bean, it is typical to have both) whose name starts with "get" or "set", followed by the property name with the first character capitalized. (For boolean properties, it is also legal to use "is" instead of "get" as the prefix for the getter method.)
Advanced JavaBeans users will know that you can tell the
system
you want to use different names for the getter and setter
methods, by
using a
java.beans.BeanInfo
class associated with your form
bean. Normally, however, it is much more convenient to
follow the
standard conventions.
WARNING - developers might be tempted to use one of the following techniques, but any of them will cause your property not to be recognized by the JavaBeans introspection facilities, and therefore cause your applications to misbehave:
getFoo()
method for your
getter, but a
setBar()
method for your setter, Java
will not recognize these methods as referring to the
same property.
Instead, the language will think you have a read-only
property named
"foo" and a write-only property named "bar".
setStartDate(java.util.Date date)
method and a
setStartDate(String date)
method in the same class, and
the compiled code would know which method to call
based on the
parameter type being passed. However, doing this for
form bean
properties will prevent Java from recognizing that you
have a
"startDate" property at all.
There are other rules to follow if you want other features of your form beans to be exposed, especially in terms of indexed attributes and mapped attributes. Specific rules are covered in detail in other areas of the documentation, in particular, Indexed Properties, Mapped Properties, and Indexed Tags.
For a complete explanation of what a JavaBean is, and everything it can do, see the JavaBeans Specification (version 1.01).
This is an interesting question. As a newbie, it is a good
practice to create a new
ActionForm
for each action
sequence. You can use
DynaActionForm
s to help reduce
the effort required, or use the code generation facilities
of your IDE.
Some issues to keep in mind regarding reuse of form beans are as follows:
struts-config.xml
for the same
ActionForm
subclass can help (especially if you
store your form beans in session scope).
Alternatively,
storing form beans in request scope can avoid
unexpected
interactions (as well as reduce the memory footprint
of your
application, because no server-side objects will need
to be
saved in between requests.
false
value.
ActionForm
s,
Action
s,
etc.) prior to being able to put together a smooth
workflow
environment using a single form bean.
As you get more comfortable, there are a few shortcuts you
can
take in order to reuse your
ActionForm
beans. Most of
these shortcuts depend on how you have chosen to implement
your
Action
/
ActionForm
combinations.
The simplest way to prepopulate a form is to have an
Action
whose sole purpose is to populate an
ActionForm
and forward
to the servlet or JSP to render that form back to the
client. A separate
Action
would then be use to process the submitted form fields,
by declaring an instance of the same form bean name.
The
MailReader
application, part of the Struts Applications
subrpoject, illustrates this design pattern nicely. Note
the following
definitions from the
struts-config.xml
file:
...
...
<-- Registration form bean -->
...
...
<-- Edit user registration -->
Note the following features of this approach:
/editRegistration
and
/saveRegistration
actions use the same form bean.
/editRegistration
action is entered, the
framework will have pre-created an empty form bean
instance, and passed it to
the
execute()
method. The setup action is free to
preconfigure the values that will be displayed when
the form is
rendered, simply by setting the corresponding form
bean properties.
ActionForm
that points
at the page which will display this form. If you are
using the
Struts JSP tag library, the
action
attribute on your
/saveRegistration
in order for the form to be submitted to the
processing action.
/editRegistration
) turns off
validation on the form that is being set up. You will
normally want
to include this attribute in the configuration of your
setup actions,
because you are not planning to actually process the
results -- you
simply want to take advantage of the fact that the
framework will precreate
a form bean instance of the correct class for you.
/saveRegistration
), on the other
hand, leaves out the
validate
attribute, which defaults
to
true
. This tells the framework to perform the validations
associated with this form bean before invoking the
processing action
at all. If any validation errors have occurred, the
framework will forward
back to your input page (technically, it forwards back
to an
ActionForward
named "registration" in this case, because
the example webapp uses the
inputForward
attribute in the
struts-config.xml
for more information)
instead of calling your processing action.
Yes. If your
Action
does not need any data and it does not need to make any
data available to the view or controller component that it
forwards to, it doesn't need
a form. A good example of an
Action
with no
ActionForm
is the
LogoffAction
in the Struts MailReader application:
This action needs no data other than the user's session,
which it can get from the
Request
, and it doesn't need to prepare any view elements for
display,
so it does not need a form.
However, you cannot use the
First off, there's an even newer Validator rule called
validwhen
,
which is almost certainly what you want to use, since it
is much easier and
more powerful. It will be available in the first release
after 1.1 ships.
The example shown below could be coded with validwhen as:
This is an excellent question. Let's step back a second and think about a typical mid to large size application. If we start from the back end and work toward the view we have:
1) Database: Most modern databases are going to validate for required fields, duplicate records, security constraints, etc.
2) Business Logic: Here you are going to check for valid data relationships and things that make sense for the particular problem you are triing to solve.
... This is where the framework comes into the picture, by now the system should be pretty well bulletproof. What we are going to do is make validation friendlier and informative. Rember it is OK to have duplicate validations...
3)
ActionErrors validate(ActionMapping map,
HttpServletRequest req)
is where you can do your validation and feed back to the
view,
information required to correct any errors.
validate
is run after
the form has been
reset
and after the
ActionForm
properties have been set from corresponding view based
input. Also remember you
can turn validation off with
validate="false"
in the
action
mapping in the
struts-config.xml
. This is done
by returning an
ActionErrors
collection with messages from your
ApplicationResources.properties
file.
Here you have access to the request so you can see what
kinds of action is
being requested to fine tune your validations. The input
attribute of the
struts-config.xml
action
allows you to send
validation errors to a particular jsp / html / tile page.
4) You can have the system perform low level validations
and client side
feedback using a
ValidatorForm
or its derivatives. This will
generate javascript and give instant feedback to the user
for simple data entry
errors. You code your validations in the
validator-rules.xml
file. A working knowledge of
regular
expressions
is necessary to use this feature effectively. For more
information, see
User Guide
The simplest way is to have two actions. The first one has the job of setting the form data, i.e. a blank registration screen. The second action in our writes the registration data to the database. The framework would take care of invoking the validation and returning the user to the correct screen if validation was not complete.
The EditRegistration action in the Struts MailReader application illustrates this:
<action path="/editRegistration"
type="org.apache.struts.webapp.example.EditRegistrationAction"
attribute="registrationForm"
scope="request"
validate="false">
<forward name="success path="/registration.jsp"/>
</action>
When the /editRegistration action is invoked, a
registrationForm is created and added to the request,
but its validate method is not called. The default value
of the
validate
attribute is
true
, so if you do not want an action to trigger form
validation, you need to remember
to add this attribute and set it to
false
.
The basic idea is a series of actions with next, back, cancel and finish actions with a common bean. Using a LookupDispatchAction is reccomended as it fits the design pattern well and can be internationalized easily. Since the bean is shared, each choice made will add data to the wizards base of information. A sample of struts-config.xml follows:
<form-beans>
<form-bean name="MyWizard"
type="forms.MyWizard" />
</form-beans>
<!-- the first screen of the wizard (next action only
available) -->
<!-- no validation, since the finish action is not
available -->
<actions>
<action path="/mywizard1"
type="actions.MyWizard"
name="MyWizard"
validate="false"
input="/WEB-INF/jsp/mywizard1.jsp">
<forward name="next"
path="/WEB-INF/jsp/mywizard2.jsp" />
<forward name="cancel"
path="/WEB-INF/jsp/mywizardcancel.jsp" />
</action>
<!-- the second screen of the wizard (back, next and
finish) -->
<!-- since finish action is available, bean should
validated, note
validation should not necessarily validate if back action
requested, you
might delay validation or do conditional validation -->
<action path="/mywizard2"
type="actions.MyWizard"
name="MyWizard"
validate="true"
input="/WEB-INF/jsp/mywizard2.jsp">
<forward name="back"
path="/WEB-INF/jsp/mywizard1.jsp" />
<forward name="next"
path="/WEB-INF/jsp/mywizard3.jsp" />
<forward name="finish"
path="/WEB-INF/jsp/mywizarddone.jsp" />
<forward name="cancel"
path="/WEB-INF/jsp/mywizardcancel.jsp" />
</action>
<!-- the last screen of the wizard (back, finish and
cancel only) -->
<action path="/mywizard3"
type="actions.MyWizard"
name="MyWizard"
validate="true"
input="/WEB-INF/jsp/mywizard3.jsp">
<forward name="back"
path="/WEB-INF/jsp/mywizard2.jsp" />
<forward name="finish"
path="/WEB-INF/jsp/mywizarddone.jsp" />
<forward name="cancel"
path="/WEB-INF/jsp/mywizardcancel.jsp" />
</action>
The pieces of the wizard are as follows:
forms.MyWizard.java - the form bean holding the information required
actions.MyWizard.java - the actions of the wizard, note the use of LookupDispatchAction allows for one action class with several methods. All the real work will be done in the 'finish' method.
mywizard[x].jsp - the data collection jsp's
mywizarddone.jsp - the 'success' page
mywizardcancel.jsp - the 'cancel' page
Chaining actions can be done by simply using the proper mapping in your forward entries in the struts-config.xml file. Assume you had the following two classes:
/* com/AAction.java */
...
public class AAction extends Action
{
public ActionForward
execute(ActionMapping mapping,
ActionForm form,
HttpServletRequest request,
HttpServletResponse response) throws
Exception
{
// Do something
return mapping.findForward("success");
}
}
</source>
<source>
/* com/BAction.java */
...
public class BAction extends Action
{
public ActionForward
execute(ActionMapping mapping,
ActionForm form,
HttpServletRequest request,
HttpServletResponse response) throws
Exception
{
// Do something else
return mapping.findForward("success");
}
}
</source>
<p>
Then you can chain together these two actions with the
Struts
configuration as shown in the following excerpt:
</p>
<source>
...
<action-mappings type="org.apache.struts.action.ActionMapping">
<action path="/A"
type="com.AAction"
validate="false">
<forward name="success" path="/B.do" />
</action>
<action path="/B"
type="com.BAction"
scope="session"
validate="false">
<forward name="success" path="/result.jsp" />
</action>
</action-mappings>
...
Here we are assuming you are using a suffix-based (
.do
) servlet
mapping, which is recommended since module support
requires it. When you
send your browser to the web application and name the
action
A.do
(i.e.
http://localhost:8080/app/A.do
) it will
execute
AAction.execute()
, which will then forward to the
"success" mapping.
This causes the execution of
BAction.execute()
since the
.do
suffix.
Of course it is also possible to chain actions programmatically, but the power and ease of being able to "reroute" your web application's structure using the XML configuration file is much easier to maintain.
As a rule, chaining Actions is not recommended. If your business classes are properly factored, you should be able to call whatever methods you need from any Action, without splicing them together into a cybernetic Rube Goldberg device.
If you must chain Actions, be aware of the following: calling the second Action from the first Action has the same effect as calling the second Action from scratch. If both of your Actions change the properties of a formbean, the changes made by the first Action will be lost because the framework calls the reset() method on the formbean when the second Action is called.