Navigational Workflows Unleashed in WWF/ASP.NET 3.5






4.97/5 (40 votes)
Case-study on the internals of a Navigational Workflow engine for a fictional dating website called “World Wide Dating”.
Contents
- Introduction
- Navigational Logic and Rules
- Setting Up the Workflow Runtime
- Navigational Workflow Architecture
- Creating and Starting a New Navigational Workflow
- The Previous and Next Events: How Navigation Works
- Synchronization, Self-Healing, and the Browser Back/Forward Buttons
- Error Pages
- Persistence and Long Running Workflows
- Skipping Between Pages
- Running the Code
- History
Introduction
What is a Navigational Workflow? In a nutshell, a Navigational Workflow is a method of describing navigational logic for an application as well as the rules that ensure that logic is adhered to. Any application that must follow some sort of process flow requires a Navigational Workflow.
This article is a case-study on the internals of a Navigational Workflow engine for a fictional dating website called "World Wide Dating". The host application is written in ASP.NET, and the Navigational Workflow is written using Windows Workflow Foundation.
Navigational Logic and Rules
World Wide Dating has six main pages:
- Default.aspx (the start page)
- Basics.aspx
- Appearance.aspx
- Lifestyle.aspx
- Interests.aspx
- Complete.aspx (the last page)
The page that first appears when the application starts is default.aspx. From here, the user can create, rehydrate, and delete Navigational Workflows. The "natural flow" of the workflow is:
In addition to the above, the workflow allows jumping from one State to another if, and only if, the user has visited that State already using the "natural flow" defined above. To give a concrete example, imagine having navigated the following scenario: default => basics => appearance => lifestyle. The user is now in the lifestyle State. From here, they can follow the workflow "naturally" and go to the interests State or back to the appearance State. In addition, the user can choose to skip to the basics State. The user cannot, however, skip to the complete State, because that State has not yet been visited. If the user attempts to prematurely skip to a State, then they are redirected to an error page (more on error pages later).
Setting Up the Workflow Runtime
Because this is an ASP.NET web application, we must load the ManualWorkflowSchedulerService
into the workflow runtime.
If we do not, then workflows will execute asynchronously and the host application might return before the workflow is done executing.
This is set up inside the web.config:
<add type=
"System.Workflow.Runtime.Hosting.ManualWorkflowSchedulerService,
System.Workflow.Runtime,
Version=3.0.0.0,
Culture=neutral,
PublicKeyToken=31bf3856ad364e35" />
There is considerable overhead that gets introduced in creating and starting the workflow runtime with each request. To avoid this, we can load the workflow runtime just once during application startup, and shut it down when the application ends. This is done in Global.asax.cs:
protected void Application_Start(object sender, EventArgs e)
{
// create an instance of workflow runtime,
// loading settings from web.config
WorkflowRuntime workflowRuntime =
new WorkflowRuntime(
WorkflowManager.WorkflowRuntimeKey);
// add the external data exchange service to the
// runtime to allow for local services
ExternalDataExchangeService exchangeService =
new ExternalDataExchangeService(
WorkflowManager.LocalServicesKey);
workflowRuntime.AddService(exchangeService);
// start the workflow runtime
workflowRuntime.StartRuntime();
// save the runtime for use by the entire application
Application[WorkflowManager.WorkflowRuntimeKey] = workflowRuntime;
}
protected void Application_End(object sender, EventArgs e)
{
// obtain reference to workflow runtime
WorkflowRuntime workflowRuntime =
Application[WorkflowManager.WorkflowRuntimeKey]
as WorkflowRuntime;
// stop the runtime
if (workflowRuntime != null)
{
workflowRuntime.StopRuntime();
workflowRuntime.Dispose();
}
}
Navigational Workflow Architecture
Our Navigational Workflow is built using the State-Machine workflow template. Each StateActivity
in the Navigational
Workflow maps to a corresponding ASPX page. This mapping is inside the web.config:
<elements>
<element stateName="InitialState" pageName="default.aspx" weight="0" />
<element stateName="BasicsState" pageName="basics.aspx" weight="10" />
<element stateName="AppearanceState" pageName="appearance.aspx" weight="20" />
<element stateName="LifestyleState" pageName="lifestyle.aspx" weight="30" />
<element stateName="InterestsState" pageName="interests.aspx" weight="40" />
<element stateName="CompleteState" pageName="complete.aspx" weight="50" />
</elements>
The structure of the application is represented in the following diagram:
The ASP.NET host application uses the Workflow Manager as a wrapper to communicate with the Navigational Workflow. All requests into the Navigational Workflow are delegated through the Workflow Manager.
Creating and Starting a New Navigational Workflow
The host application makes the following call to start a new workflow:
_workflowManager.StartNewWorkflow();
The StartNewWorkflow
method is located in WorkflowManager.cs, and is responsible for creating
and starting a new Navigational Workflow instance:
public string StartNewWorkflow()
{
// create and start the navigational workflow instance
WorkflowInstance instance =
_workflowRuntime.CreateWorkflow(typeof(NavigationWorkflow));
instance.Start();
// execute the workflow synchronously on the current thread
RunWorkflow(instance.InstanceId);
// put the workflow instance id in session for this user
this.CurrentWorkflowInstanceId = instance.InstanceId;
// execute the workflow on the current thread and return the page to go to
return GetPageToGoTo(instance.InstanceId);
}
From here, control transfers to the Navigational Workflow instance defined in NavigationalWorkflow.cs.
The InitialState
is registered as the starting state of the workflow, and is the point of entry into the workflow:
The first Activity inside the InitialState
Activity is of type StateInitializationActivity
. The Activities
within a StateInitializationActivity
execute immediately upon entering a StateActivity
.
Double-clicking on the initializeInitialState
Activity opens up the dialog below:
The first Activity that executes is a CodeActivity
called codeInitializeInitialState
, and it executes the code below:
private void codeInitializeInitialState_ExecuteCode(object sender, EventArgs e)
{
_heaviestWeight = -1; // default setting for heaviest weight
}
The _heaviestWeight
variable is a marker that keeps track of how far along in the Navigational Workflow a particular user got.
This is used in determining whether or not the user can skip to a given State (and in essence, whether or not the user can skip
to a given ASPX page). This marker will be discussed in more detail later in the section Skipping Between Pages.
We will defer discussion of the remaining code in the InitialState
and move on to explore the role Events play in navigation.
The Previous and Next Events: How Navigation Works
Introduction
The heart of the Navigational Workflow is the Previous and Next events. The host application raises these events via the Workflow Manager
to obtain the previous and next pages to go to. Every StateActivity
in the Navigational Workflow listens for the Previous
and Next events. It is these events that drive transitions between different StateActivity
s. Once the workflow transitions
into a StateActivity
, the corresponding ASPX page of the current StateActivity
is queried
from the web.config and passed to the host application.
Explanation
Let us imagine that the current State of the Navigational Workflow is the LifeStyle State:
According to the web.config, the ASPX page that maps to this State is lifestyle.aspx (that is the page the user should be on when in this State), and the weight of the Lifestyle State is 30:
<element stateName="LifestyleState" pageName="lifestyle.aspx" weight="30" />
The LifeStyle State listens for two events. These events are Previous
and Next
,
and are defined in INavigationService.cs:
event EventHandler<ExternalDataEventArgs> Previous;
event EventHandler<ExternalDataEventArgs> Next;
Let us examine how to transition to the previous page starting from the host application and work our way back to the Navigational Workflow.
To navigate to the previous page, the host application makes the following call into the WorkflowManager
:
// navigate to the previous page
string pageToGoTo = _workflowManager.Previous();
_workflowManager.ManagedRedirect(pageToGoTo);
The above code first calls the Previous
method on the WorkflowManager
to obtain the page to go to.
The next line is a wrapper around Response.Redirect
to transfer the user to the previous page. We will now explore
the Previous
method of the WorkflowManager
:
public string Previous()
{
/* 180 */ // prepare external data exchange arguments
/* 181 */ ExternalDataEventArgs args =
new ExternalDataEventArgs(this.CurrentWorkflowInstanceId);
/* 183 */ // make local service call into Navigation Workflow
/* 184 */ _navigationService.OnPrevious(args);
/* 186 */ // execute the workflow on the current thread and return the page to go to
/* 187 */return GetPageToGoTo(args.InstanceId);
}
Line 181 prepares the ExternalDataEventArgs
to send to the Navigational Workflow. Line 184 raises the Previous
event inside the LifestyleState
in the Navigational Workflow; however, since we are using
the System.Workflow.Runtime.Hosting.ManualWorkflowSchedulerService
to execute our workflows,
the Navigational Workflow will not receive the event until ManualWorkflowSchedulerService
explicitly
runs the workflow instance. Let's take a look at line 187:
private string GetPageToGoTo(Guid instanceId)
{
/* 399 */ // execute the workflow on the current asp.net thread
/* 400 */ RunWorkflow(instanceId);
/* 402 */ // return where the user should go
/* 403 */ return _pageToGoTo;
}
Line 400 makes a call to RunWorkflow
, which is where the ManualWorkflowSchedulerService
gets called
to execute the current Navigational Workflow instance:
private bool RunWorkflow(Guid instanceId)
{
ManualWorkflowSchedulerService scheduler = GetWorkflowSchedulerService();
return scheduler.RunWorkflow(instanceId);
}
Now that control has transferred to the Navigational Workflow, let us examine how the Lifestyle State is set up to listen
for the Previous
event, and what happens when it receives it:
The eventLifestyleStatePrevious EventDrivenActivity
contains the code that listens for the Previous
event.
Double-clicking eventLifestyleStatePrevious
opens the following:
The HandleExternalEventActivity
named handleLifestylePrevious
is listening for the Previous
event.
Examining the handleLifestylePrevious
properties illustrate how to set up the event listening:
After the Previous
event is received, the SetStateActivity
executes and moves the Navigational Workflow into the Appearance State:
Upon entering the Appearance State, the StateInitializationActivity
named initializeAppearanceState
executes.
Double-clicking it opens up the designer for the Activity:
There is a single CallExternalMethodActivity
. To figure out what exactly is happening, we need to investigate the properties
of the CallExternalMethodActivity
:
The CallExternalMethodActivity
calls the OnPageToGoTo
Local Service method, passing the PageToGoToEventArgs
into it.
The OnPageToGoTo
method raises the PageToGoToReceived
event which the WorkflowManager
is listening for.
Before the Local Service method executes, the CallExternalMethodActivity
executes the InitializeOutgoingMessage
method:
private void InitializeOutgoingMessage(object sender, EventArgs e)
{
/* 90 */ // initialize the outgoing args to send to the host application
/* 91 */ _pageToGoToEventArgs =
new PageToGoToEventArgs(this.WorkflowInstanceId,
GetPageToGoTo());
/* 92 */
/* 93 */ // each state has a weight associated with it. keep
// track of the heaviest weight (logically the
/* 94 */ // furthest state in the navigation process)
// so we can skip there later
/* 95 */ if (_heaviestWeight < NavigationSection.Current.
Elements[this.CurrentStateName].Weight)
/* 96 */ {
/* 97 */ _heaviestWeight = NavigationSection.Current.
Elements[this.CurrentStateName].Weight;
/* 98 */ }
}
The GetPageToGoTo
method on line 91 retrieves the corresponding page associated with this StateActivity
from the web.config:
private string GetPageToGoTo()
{
// query the page to go to from the navigation section
string pageToGoTo = NavigationSection.Current.
Elements[this.CurrentStateName].PageName;
return pageToGoTo;
}
When the Local Service method OnPageToGoTo
executes, the Workflow Manager is listening for it and executes
the following code upon receiving the event:
private void _navigationService_PageToGoToReceived(object sender,
PageToGoToEventArgs e)
{
// get the page to go to
_pageToGoTo = e.PageToGoTo;
}
The above code inside the Navigational Workflow executes when the RunWorkflow
method is called.
If we go back and take a look at the GetPageToGoTo
method:
private string GetPageToGoTo(Guid instanceId)
{
/* 399 */ // execute the workflow on the current asp.net thread
/* 400 */ RunWorkflow(instanceId);
/* 402 */ // return where the user should go
/* 403 */ return _pageToGoTo;
}
By the time line 403 executes, the _pageToGoTo
variable will already have been populated with the page to go to.
The host application receives this page, and then redirects accordingly.
Synchronization, Self-Healing, and the Browser Back/Forward Buttons
Browser Back/Forward Buttons
In the past, the browser's back/forward buttons have caused many issues with synchronization. The issue is that most browsers cache web pages so that when you click the browser's back/forward buttons, a cached copy of the web page is retrieved. Because the page is cached, there is no postback to the server to update the Navigational Workflow and to perform server side validation. In the past, the most simple solution was to disable the browser's back/forward buttons. This section proposes a solution to the problems presented using the browser back/forward buttons.
The first thing that must be done is caching must be disabled. With caching disabled, the browser has no choice but to postback to the server to retrieve the web page whenever the browser's back/forward buttons are clicked. There is a performance hit associated with this in that every request must be sent to the server; however, this is the only way to force a postback to the server to occur when the user clicks the browser's back/forward buttons. The code below is in the MasterPage of the host application, and disables caching across the website:
protected void Page_Load(object sender, EventArgs e)
{
// disable page caching
Response.Expires = 0;
Response.Cache.SetNoStore();
Response.AppendHeader("Pragma", "no-cache");
}
The code above will make sure that the page postbacks to the server whenever the user clicks the browser's back/forward buttons.
It has been tested and verified to work in both Internet Explorer 7 and Firefox. Safari 3.1, however, requires one extra step.
To get the page to postback to the server when the browser's back/forward button is clicked using the Safari browser, the following
JavaScript method is attached to the onunload
event of the HTML body in the Master page:
<body onunload="javascript:safariBackForwardBrowserButtonFix();">
There is nothing fancy about the JavaScript method that is called. Its method body is empty, but nevertheless, this is what is required to force the Safari browser to postback to the server when its back/forward buttons are clicked.
But, how do we know that the page is really posting back to the server when clicking the browser's back/forward buttons? One way to test for this is to use breakpoints. However, there is also a control on every page that is called: DebuggingPane.ascx. One piece of information this control displays is the current datetime timestamp. If the timestamp changes to the current time when you click the browser's back/forward buttons, then you can be sure a postback to the server occurred.
Synchronization and Self-Healing
Synchronization ensures that the web page the user is on is the web page that they should be on. The following pages require
synchronization (note that the pages in the list below are represented by StateActivity
s in the Navigational Workflow,
the distinction is that Error pages are not represented by StateActivity
s):
- Basics.aspx
- Appearance.aspx
- Lifestyle.aspx
- Interests.aspx
- Complete.aspx
The pages above all call the SynchronizeWorkflow
method inside of their Page Load event. The SynchronizeWorkflow
method
is located inside the Workflow Manager class:
public void SynchronizeWorkflow()
{
// check if there is a workflow in session
if (this.CurrentWorkflowInstanceId == Guid.Empty)
{
// redirect to the start page
ManagedRedirect(GetStartPage());
}
// the page to go to if the workflow is not synchronized
string pageToGoTo = string.Empty;
try
{
// raise event to ask the workflow if it is synchronized
IncomingSynchronizeEventArgs args =
new IncomingSynchronizeEventArgs(
this.CurrentWorkflowInstanceId, GetCurrentPageName());
_navigationService.OnIncomingSynchronize(args);
// execute the workflow synchronously on the current thread
RunWorkflow(this.CurrentWorkflowInstanceId);
// return if the workflow is synchronized
if (_isSynchronized)
{
return;
}
// get the name of the state that the user should be in
StateMachineWorkflowInstance instance =
new StateMachineWorkflowInstance(
_workflowRuntime, this.CurrentWorkflowInstanceId);
instance.SetState(_stateToGoTo);
// get the name of the page that the user should be on
pageToGoTo = Rehydrate(this.CurrentWorkflowInstanceId);
}
catch (Exception ex)
{
// could not synchronize the workflow
ManagedRedirect(GetStartPage());
}
// if the page the user should be on is not
// the same as the page that they currently
// are on then redirect the user to the page they should be on
if (!pageToGoTo.Equals(GetCurrentPageName()))
{
ManagedRedirect(pageToGoTo);
}
}
The SynchronizeWorkflow
method asks the Navigational Workflow if it is in sync by raising the IncomingSynchronize
event.
The Navigational Workflow listens for this event, and returns its response by raising the OutgoingSynchronizeReceived
event.
The synchronization code checks to see if any type of URL hacking has occurred. By URL hacking, I am referring to modifications made
to the World Wide Dating website URL by the user and not through the Navigational Workflow code. If the URL has been modified by the user,
then the synchronization code checks to see if the URL the user is trying to navigate to is an allowable page for them to be on.
If it is allowed, then the Workflow Manager will self-heal the Navigational Workflow and transition it to the proper StateActivity
.
If it is not, then the Workflow Manager will redirect the user back to the page they were on before they attempted to hack the URL.
It is worth noting that clicking the browser's back/forward button is a way of hacking the URL; the reason being is that by clicking
these buttons, events are not raised inside the Navigational Workflow that indicates it needs to transition to another State.
Therefore, self-healing also occurs when using the browser back/forward buttons.
If we double-click the eventIncomingSynchronize EventDrivenActivity
inside the Navigational Workflow, the below appears in the designer:
The HandleExternalEventActivity
called handleIncomingSynchronize
listens for the IncomingSynchronize
event.
The CallExternalEventActivity
called callOnOutgoingSynchronize
has the following properties:
The first thing that happens here is the InitializeOutgoingSynchronizeMessageEventArgs
method executes:
private void InitializeOutgoingSynchronizeMessageEventArgs(object sender, EventArgs e)
{
NavigationElement expectedElement =
NavigationSection.Current.Elements.
GetElementByPageName(GetPageToGoTo());
NavigationElement currentElement =
NavigationSection.Current.Elements.
GetElementByPageName(_incomingSynchronizeEventArgs.CurrentPage);
// initialize the outgoing synchronize event args
_outgoingSynchronizeEventArgs =
new OutgoingSynchronizeEventArgs(this.WorkflowInstanceId);
// check if the workflow is synchronized by comparing the
// expected page to the current page the user is on
_outgoingSynchronizeEventArgs.IsSynchronized =
expectedElement.PageName.Equals(currentElement.PageName);
// workflow is synchronized
if (_outgoingSynchronizeEventArgs.IsSynchronized)
{
return;
}
/* the below checks are in reference to if the
user tried to manipulate the URL by hand */
// if the weight of the page the user is on is less than or equal
// to the heaviest weight, then the workflow is able to "self-heal" and
// transition itself to the state it should be in.the host application
// will make the transition, but the StateToGoTo below tells it which is
// the right state to transition to.
if (currentElement.Weight <= _heaviestWeight)
{
_outgoingSynchronizeEventArgs.StateToGoTo = currentElement.StateName;
}
// otherwise the user is trying to skip to a page further
// in the workflow than they can currently go, so set the
// state of the workflow to be what we expect it should be
else
{
_outgoingSynchronizeEventArgs.StateToGoTo = expectedElement.StateName;
}
}
The above code populates the _outgoingSynchronizeEventArgs
data member with the information needed for the Workflow Manager
to synchronize the Navigational Workflow. The _outgoingSynchronizeEventArgs
data member maps
to the OutgoingSynchronizeMessageEventArgs
property. This property gets passed into the OnOutgoingSynchronize
Local Service method,
which when called raises the OutgoingSynchronizeReceived
event that the Workflow Manager is listening for. All of this happens from
the SynchronizeWorkflow
method inside the WorkflowManager
.
Error Pages
So far, we have established that every ASPX web page (that is part of the natural workflow, i.e.,~ not an error page) maps to a unique
StateActivity
within the Navigational Workflow, and that synchronization is used to ensure that we are on the correct page. Inevitably though,
an error will occur sometimes in our applications, and our Navigational Workflow needs to be prepared to handle them.
When an error occurs, the host application determines what kind of error it is, and then raises the Error event inside the Navigational Workflow, and indicates to it what type of error has occurred. The Navigational Workflow then returns the error page to navigate to. The following code is an example of how the host application raises the Error event inside the Navigational Workflow and indicates to it that the type of error is "Unknown":
string pageToGoTo =
_workflowManager.Error(DateSiteNavigation.ErrorType.Unknown);
_workflowManager.ManagedRedirect(pageToGoTo);
The Workflow Manager executes the following code to raise the Error event inside the Navigational Workflow to obtain the error page to navigate to:
public string Error(ErrorType errorType)
{
// prepare external data exchange arguments
ErrorEventArgs args =
new ErrorEventArgs(this.CurrentWorkflowInstanceId, errorType);
// make local service call into Navigational Workflow
_navigationService.OnError(args);
// execute the workflow on the current thread and
// return the page to go to
return GetPageToGoTo(args.InstanceId);
}
The first concept to recognize when dealing with error pages is that an error page is always a valid page to be on regardless of what State the Navigational Workflow is in. The Error event is defined in the Local Service as:
event EventHandler<ErrorEventArgs> Error;
To quote Popeye The Sailor Man, "I am what I am and that's all that I am." Every page in our workflow, whether it be a navigational page or an error page, knows about itself. It does not know anything about the other pages, but it knows what its responsibilities are. That said, an error page knows it is an error page, and it understands the concept that it is always a valid page to be on in terms of navigation. Therefore, all error pages make sure to not call the synchronize workflow method.
Double-clicking the EventDrivenActivity
called errorEvent
inside of the Navigational Workflow opens the following in the designer:
The first Activity is a HandleExternalEventActivity
called handleError
. If we investigate its properties,
we will see that it is configured to listen for the Error event, and that it binds the argument e
to the Navigational Workflow
property ErrorEventArgs
:
After the Error event is raised, the CallExternalMethodActivity
called callErrorOnPageToGoTo
executes.
Let's take a look at its properties:
Notice that the properties are almost identical to the those of the CallExternalMethodActivity
discussed in the section:
The Previous and Next Events: How Navigation Works. The only difference is, the MethodInvoking
property
calls the method InitializeErrorMessage
before the Local Service method fires. This method is responsible for initializing
the PageToGoToEventArgs
with the name of the ASPX error page to go to:
private void InitializeErrorMessage(object sender, EventArgs e)
{
/* 109 */ // initialize the outgoing args to send to the host application
/* 110 */ _pageToGoToEventArgs =
new PageToGoToEventArgs(
this.WorkflowInstanceId, GetErrorPageToGoTo());
}
The GetErrorPageToGoTo
method on line 110 gets the name of the ASPX error page to go to:
private string GetErrorPageToGoTo()
{
// query the page to go to from the navigation section
string errorName = _errorEventArgs.ErrorType.GetErrorName();
string pageToGoTo = NavigationSection.Current.Errors[errorName].PageName;
return pageToGoTo;
}
This method obtains the name of the error from the ErrorEventArgs
. From there, it proceeds to retrieve the error page name
from the web.config using NavigationSection
. The web.config section that contains the error name to error page mapping looks like:
<errors>
<error errorName="generic" pageName="genericError.aspx" />
<error errorName="unknown" pageName="unknownError.aspx" />
</errors>
Persistence and Long Running Workflows
Most real world scenarios have a requirement that a user be able to start a new workflow, stop, and then continue at some later
point in time. That later point in time could be hours, days, or even months later. We have already discussed
Creating and Starting a New Navigational Workflow;
now, we will cover how to rehydrate a navigational workflow from the point the user left off at. How to setup the WWF Persistence
Service database tables will be covered in the Running the Code section. There is one additional
table that is used to store dating profile data such as eye color, hair color, height, etc. This table is called the Profile table.
It is also the table that stores the unique workflow instance ID for each user that creates a dating record. That same workflow instance ID
maps to a record in the out-of-the-box WWF database table called InstanceState. The WWF table called InstanceState is where
Windows Workflow Foundation stores data about workflows that have been persisted. It stores information such as the current
StateActivity
the workflow is in, private data members, etc.
The host application rehydrates workflows by calling the Rehydrate
method of the Workflow Manager.
The Rehydrate
method takes the ID of the workflow instance to be rehydrated as its parameter.
Below is the implementation of the Rehydrate
method:
public string Rehydrate(Guid instanceId)
{
try
{
// prepare external data exchange arguments
ExternalDataEventArgs args =
new ExternalDataEventArgs(instanceId);
// make local service call into Navigation Workflow
// to retrieve step user left off at
_navigationService.OnRehydrated(args);
// put the workflow instance id in session
this.CurrentWorkflowInstanceId = instanceId;
// execute the workflow on the current
// thread and return the page to go to
return GetPageToGoTo(args.InstanceId);
}
catch (Exception ex)
{
// the given workflow instance could not be rehydrated
return GetStartPage();
}
}
The Navigational Workflow listens for this event. Double-clicking the eventRehydrated EventDrivenActivity
opens up the below in the designer:
The HandleExternalEventActivity
called handleRehydrated
listens for the Rehydrated
event.
Once received, the CallExternalMethodActivity callRehydrated
executes. If you look at the properties of this Activity, you will notice
it is exactly the same as in the section The Previous and Next Events: How Navigation Works:
The reason it is the same has to do with what happens when you rehydrate a workflow instance in WWF: When you rehydrate a workflow instance,
WWF retrieves the workflow instance from the database for the duration required to process the request made to it. Part of the information that
is retrieved is the current StateActivity
of the workflow. Our Navigational Workflow uses this to obtain the ASPX page from
the web.config that maps to the current StateActivity
. It is this page that is returned to the Workflow Manager
and then to the host application. After the instance finishes processing the request, WWF will save it back into the persistence store.
Skipping Between Pages
Since every web page is associated with a unique StateActivity
, we are actually skipping between StateActivity
s in the Navigational
Workflow to achieve skipping between web pages. Most real-world scenarios require a user to fill out a web form following some predefined navigational order.
Once the user has gone through the predefined steps, they should (in certain circumstances) be allowed to skip back and forth between pages they have already
been on to edit information. To allow this, our navigational workflow listens to the following events:
event EventHandler<ExternalDataEventArgs> SkipToBasics;
event EventHandler<ExternalDataEventArgs> SkipToAppearance;
event EventHandler<ExternalDataEventArgs> SkipToLifestyle;
event EventHandler<ExternalDataEventArgs> SkipToInterests;
event EventHandler<ExternalDataEventArgs> SkipToComplete;
The EventDrivenActivity
s that encapsulate these events are defined at the workflow level:
Double-clicking on the EventDrivenActivity eventSkipToLifestyle
opens up the following in the designer:
Once the HandlExternalEventActivity
called handleSkipToLifestyle
receives the SkipToLifestyle
event,
control transfers to the IfElseActivity
. The IfElseBranchActivity
on the left uses a code condition to check
and see if skipping to the LifestyleState
is a valid navigational move. This is where the _heaviestWeight
variable
comes into play. Each StateActivity
has a weight that is associated with it, and as the navigational workflow transitions
to States with heavier weights, it stores that weight in the _heaviestWeight
variable.
According to the web.config, the Lifestyle
State has a weight of 30:
<element stateName="LifestyleState" pageName="lifestyle.aspx" weight="30" />
Skipping to the Lifestyle state is valid if 30 is less than or equal to the heaviest known weight.
The CodeCondition
method that determines if skipping to the LifeStyle State is valid is defined as:
private void CanSkipToLifestyleState(object sender, ConditionalEventArgs e)
{
int lifestyleWeight =
NavigationSection.Current.Elements[LifestyleState.Name].Weight;
e.Result = lifestyleWeight <= _heaviestWeight;
}
If the Code Condition evaluates to true, then the Navigational Workflow transitions to the Lifestyle State;
otherwise, the IfElseBranchActivity
on the right executes. Examining the properties
of the CallExternalMethodActivity callSkipToLifestyleOnPageToGoTo
reveals that the Navigational Workflow is preparing to return
an error page as the page to go to. This illustrates an example of how to redirect the host application to an error page from the Navigational Workflow.
The example in the section Error Pages illustrated how to redirect to an error page from the host application.
Below are the properties of callSkipToLifestyleOnPageToGoTo
:
Investigating the InitializeCannotSkipToStateErrorMessage
reveals which error page the Navigational Workflow is going to return:
private void InitializeCannotSkipToStateErrorMessage(object sender, EventArgs e)
{
// get the error page to go to
string pageToGoTo =
NavigationSection.Current.
Errors[ErrorType.Generic.GetErrorName()].PageName;
// initialize the outgoing args to send to the host application
_pageToGoToEventArgs =
new PageToGoToEventArgs(this.WorkflowInstanceId, pageToGoTo);
}
The GenericError
is defined in the web.config as:
<error errorName="generic" pageName="genericError.aspx" />
What this means is that the user will get redirected to genericError.aspx whenever they attempt to skip to a web page they
have not yet been on. The "natural flow" of the Navigational Workflow is to transition to and from StateActivity
s using
the Previous
and Next
events. The skip events are used as an example to simulate one type of real-world scenario.
Running the Code
World Wide Dating requires Microsoft SQL Server 2005 (any edition). The required SQL scripts are in a folder called "database setup procedure" inside the DateSite project. The setup procedure is as follows:
- Create a database called Dating
- Run Profile_Schema.sql
- Run SqlPersistenceService_Schema.sql (WWF script)
- Run SqlPersistenceService_Logic.sql (WWF script)
Alternatively, the composite.sql file contains all the necessary SQL scripts bunched into one file. After the database has been setup, open up the DateSite Solution and run the DateSite project.
Note About Different SQL Server 2005 Versions
Many people have posted messages regarding the following error:
navigationService = _workflowRuntime.GetService() as NavigationService System.NullReferenceException: Object reference not set to an instance of an object
The most common cause of this error is an incorrect connection string. World Wide Dating was coded using SQL Server 2005 Developer Edition, and assumes the Dating database is located on the server named localhost. If you are getting this error, you may need to modify your database connection string to point to the proper database server.
Happy Dating!
History
- 21/10/2008: Updated Running the Code section.
- 17/08/2008: Correction made to the Navigational Logic and Rules section.
- 22/05/2008: Updated article diagrams.
- 08/05/2008: Formatted code for proper printing.
- 29/04/2008: Initial release.